Pankow Cherry Dnapl Book 1996

DENSE
CHLORINATED
SOLVENTS
and other
DNAPLs in Groundwater:
History, Behavior, and
Remediation
James F. Pankow
John A. Cherry
WATERLOO
PRESS
Pankow, James F., Cherry, John A.

Dense Chlorinated Solvents and other DNAPLs in Groundwater
Includes index and bibliographical references.
Editorial/production supervision: James F. Pankow

Interior design/production: ETPIHarrison
Cover design: Jonathan Pennell
1996 by Waterloo Press.

P.O. Box 91399
Portland, Oregon 97291-1399
Address orders correspondence to:
Waterloo Educational Services
2 Taggart Court, Unit 4
Guelph, Ontario
Canada NIH 6H8
All rights reserved. No part of this book may be
reproduced, in any form or by any means,
without permission in writing from the publisher.
Printed in the United States of America

10 9
Library of Congress Catalog Card Number: 95-61690
Dedication
This book is dedicated to the following individuals who helped establish the research path
that has led to the present understanding of dense chlorinated solvents in groundwater
described here.
The late Richard M. Spann, formerly Manager of Environmental Programs for IBM
in Poughkeepsie, New York, who in 1982 focused our research interests on the behavior
of chlorinated solvents in groundwater by insisting on a scientific basis for remediation
decision making at the industrial sites in his care.
Dr. Friedrich Schwille, formerly Chief Groundwater Hydrologist at the Federal
Institute of Hydrology in Koblenz, Germany (now retired), whose laboratory experimentation in the late 1970s and early 1980s provided the scientific foundation for our
subsequent research, and who provided advice and encouragement for our research efforts.
Dr. Rolf Bemegger, formerly Corporate Vice-President of the Ciba-Geigy Corporation in Ardsley, New York (now retired), who in 1987, through encouragement and
initial funding, fostered the concept of a major research program focused on the behavior of dense chlorinated solvents performed by a consortium of university groups and
supported by a consortium of corporations and funding agencies.
For their influences, we are truly grateful.
John A. Cherry
Waterloo, Ontario
June 1995
James F. Pankow
Portland, Oregon
June 1995
Stanley Feenstra
Waterloo, Ontario
June 1995
Contents
PREFACE
1
xv
DENSE CHLORINATED SOLVENTS IN GROUNDWATER: BACKGROUND

AND HISTORY OF THE PROBLEM
1.1
1.2
1.3
1.4
1.5
1.6
1.7
Backgrou nd
1.1.1 General
1.1.2 DNAPLs and LNAPLs
Chlorinated Solvent Usage
Source Types for Solvents in Groundwater
The Physical, Chemical, and Biological Properties of Chlorinated Solvent
Compounds
Drinking Water Limits
Typical Chlorinated Solvent Plumes in Groundwater
A History of How We Became Aware that Dense Chlorinated Solvents Can
Pose Serious Threats to Groundwater Systems
1.7.1 General
1.7.1.1 Basis of interest
1.7.1.2 The history has two stages
1.7.2 Timelines in the History
1.7.3 The First Stage: Discovering that Chlorinated Solvents are in Groundwater
1.7.3.1 The 1950s to the mid-1960s: the early years
1.7.3.2 The mid-1960s to 1972: finding toxic organic compounds in
some drinking water supplies
1.7.3.3 1972 to 1975: finding the first chlorinated VOCs in drinking
water: the trihalomethanes (THMs)
1.7.3.4 1976 to 1979: widespread testing for THMs in drinking water
reveals the presence of other VOCs: coming to the conclusion
that many important aquifers are contaminated with
chlorinated solvents
1.7.3.5 1950 to 1951: the evidence that chlorinated solvents are
widespread groundwater contaminants becomes overwhelming
1
3
3
6
7
11
13
13
15
16
16
16
16
17
17
17
18
20
21
23
v
vi
Contents
1.7.4 The Second Stage: Discovering the DNAPL Nature of Liquid Chlorinated
Solvents
1.7.4.1 1982 to the late 1980s
1.7.4.2 The late 1980s to the present
1.8
1.9
2.3
2.4
2.5
2.6
2.7
2.8
2.9
28
References
46
Introduction
Physical and Chemical Principles
2.2.1 Immiscible-Phase Physics
2.2.2 Phase Transfer Processes
DNAPLs in the Vadose Zone
2.3.1 The Non-Aqueous Phase
2.3.2 The Vapor Phase
DNAPLs in the Groundwater Zone
2.4.1 Granular Aquifers
2.4.2 Fractured Non-Porous Media
2.4.3 Fractured Porous Media
2.4.4 Layered Aquifer-Aquitard Systems
Plumes from DNAPLs
2.5.1 Dispersion, Diffusion, and Density
2.5.2 Transformations
2.5.3 Mass Considerations
2.5.4 Ultimate Fate
Implications for Site Characterization and Monitoring
Implications for Site Remediation
Conclusions
References
MECHANICS AND MATHEMATICS OF THE MOVEMENT OF DENSE

NON-AQUEOUS PHASE LIQUIDS (DNAPlS) IN POROUS MEDIA
3.1
3.2
3.3
28
Conclusions
CONCEPTUAL MODELS FOR THE BEHAVIOR OF DENSE NON-AQUEOUS

PHASE LIQUIDS (DNAPlS) IN THE SUBSURFACE
2.1
2.2
26
26
Introduction
53
54
57
57
59
61
61
63
64
64
67
70
71
75
75
78
80
82
84
85
85
88
89
3.2.2 Interfacial Tension, Wettability and Capillarity

3.2.3 Dependence of Fluid Content on Capillary Pressure
3.2.4 Hysteresis and Residual DNAPL
3.2.6 Summary
Distribution of DNAPL in a Pool Above an Aquitard
90
91
91
92
93
95
99
99
3.3.1 General
99
Capillary Pressure and Fluid Contents

3.2.1 General
Contents
3.3.2 Distribution of Capillary Pressure
100
3.3.3 Vertical Distribution of DNAPL in a Pool
102
3.3.4 DNAPL in a Monitoring Well

Darcy's Law and Relative Permeability
3.4.1 General
3.4.2 The Darcy Equations
3.4.3 Relative Permeabi Iities
The Effect of Groundwater Flow on DNAPL Migration
3.5.1 General
3.5.2 Forces Acting on a DNAPL in Flowing Groundwater
3.5.3 Static DNAPL in a Sloping Bed
3.5.4 Incipient Motion of DNAPL Fingers (Ganglia)
104
105
105
105
106
108
108
109
110
111
3.6
Distribution of DNAPL in Unsteady, Horizontal Flows
112
112
3.7
3.6.1 General
3.6.2 The Differential Equations for Two-Fluid Flow
3.6.3 The Buckley-Leverett Solution
3.6.4 Lateral Migration of DNAPL
3.6.5 Displacement of DNAPL by a Linear Water Flood
3.6.6 Displacement of DNAPL by Radial Water Flood
Stability of Two-Fluid Displacements
3.8
3.9
Conclusions
References
3.4
3.5
NUMERICAL SIMULATION OF THE MIGRATION OF DENSE NON-AQUEOUS

PHASE LIQUIDS (DNAPLS) IN POROUS MEDIA
4.1
4.2
4.3
4.4
4.5
4.6
vii
Introduction
The Modeling Process
Model Usage
Example
Conclusions
References
EXPERIMENTAL STUDIES OF THE MOVEMENT OF CHLORINATED SOLVENT

COMPOUNDS AND OTHER DNAPLS IN THE VADOSE, CAPILLARY, AND
GROUNDWATER ZONES
5.1
5.2
5.3
5.4
5.5
Introduction
DNAPL Movement in the Vadose Zone
DNAPL Movement in the Capillary Zone
DNAPL Movement in the Groundwater Zone
5.4.1 General
5.4.2 The Role of Aquifer Heterogeneity in DNAPL Movement
DNAPL Retention Capacity of Porous Media
113
114
116
119
121
123
125
126
129
129
130
137
139
142
143
145
146
147
150
154
154
155
158
viii
Contents
5.6
5.7
5.8
Solvent Migration in Fractured Media

Studies Using Microscopic Models of Porous Media
Experimental Releases of Chlorinated Solvents in the Field
5.8.1 General
5.8.2 Vadose Zone Releases
5.8.3 Groundwater Zone Releases
5.9 Chlorinated Solvent Vapor Transport in the Vadose Zone
5.10 Chlorinated Solvent Dissolution in the Groundwater Zone
5.11 Conclusions
5.12 References
159
160
166
166
166
169
170
173
VAPOR MIGRATION IN THE VADOSE ZONE
179
6.1
6.2
6.3.1 General
6.3.2 Diffusion
6.3.3 Advection
6.4 Aqueous Transport
6.5 Interactions with the Atmosphere
6.6 Interactions with Groundwater
6.7 Soil-Gas Surveys
6.7.1 General
6.7.2 Sampling
6.7.3 Contaminant Detection
6.8 A Pair of Field Experiments
6.9 Conclusions
6.10 References
180
182
182
182
185
186
186
186
187
187
188
189
192
192
193
194
194
194
194
195
197
198
DISSOLUTION OF DENSE NON-AQUEOUS PHASE LIQUIDS (DNAPLS) IN

THE SUBSURFACE
203
6.3
7.1
7.2
7.3
Introduction
Phase Partitioning
6.2.1 General
6.2.2 Vaporization
6.2.3 Gaseous-Aqueous Partitioning
6.2.4 Aqueous-Solid Partitioning
6.2.5 Gaseous-Solid Partitioning
6.2.6 Retardation Factor
Gaseous Transport
Introduction
Solubility from Pure DNAPLs and from DNAPL Mixtures
Dissolution Processes
7.3.1 Theory and Conceptual Models
7.3.1.1 General
175
175
204
204
208
208
208
Contents
7.3.1.2 Stagnant film model
7.3.1.3 Penetration model
7.3.1.4 Random surface renewal model
7.3.1.5 Dimensionless numbers
7.3.2 Laboratory Studies
7.4
7.5
7.6
8
7.3.3 Modeling Studies

7.3.4 Persistence of DNAPL Zones
7.3.5 Field Observations of DNAPL Dissolution
Multi-Component DNAPLs
7.4.1 Theory and Laboratory Studies
7.4.2 Estimates of DNAPL Source Depletion
Conclusions
References
SORPTION OF DISSOLVED CHLORINATED SOLVENTS TO AQUIFER

MATERIALS
8.1
8.2
8.3
8.4
8.5
8.6
8.7
8.8
Introduction
Effects of Sorption on Transport
Sorption Mechanisms: Effects of Sorbate and Sorbent on Soil/Water Partitioning
8.3.1 Definitions
8.3.2 Sorption to Soil Organic Carbon
8.3.3 Sorption Directly to Mineral Surfaces
Methods for Measuring Sorption
8.4.1 General
8.4.2 Small-Scale Methods
8.4.2.1 Batch methods
8.4.2.2 Column methods
8.4.3 Large-Scale Methods
Non-Ideal Sorption Behavior in a Saturated System
8.5.1 Non-Ideal Breakthrough Curves
8.5.2 Pore-Scale Mechanisms Causing Non-Ideal Behavior
8.5.3 Contribution of Spatial Variability in the Field to Non-Ideal Sorption
Behavior
Effects on Sorption Caused by Properties of the Aqueous Phase
8.6.1 General
8.6.2 Effects of Natural Aqueous Geochemistry
8.6.3 Effects of Co-Solvents, Dissolved Organic Matter, Suspended Colloids, and
Surfactants
8.6.3.1 Co-solvents
8.6.3.2 Dissolved organic matter (DOM) and suspended colloidal material
8.6.3.3 Surfactants
Sorption to Unsaturated Soil
Conclusions
ix
210
211
212
212
214
215
220
222
223
223
226
228
229
233
234
235
239
239
243
247
248
248
248
248
249
250
250
250
252
253
254
254
254
255
255
255
256
257
258
Contents
8.9
9
260
CHEMICAL AND MICROBIOLOGICAL TRANSFORMATIONS AND

DEGRADATION OF CHLORINATED SOLVENT COMPOUNDS
267
9.1
9.2
Introduction
268
Chemical Reactions
269
9.2.1 Overview
9.2.2 Substitution Reactions
9.2.2.1 Hydrolysis and dehydrohalogenation
9.2.2.2 Nucleophilic substitution by reduced sulfur
9.2.3 Reduction Reactions
9.2.4 Implications for Field Studies and for Remediation
Microbiology of the Subsurface
9.3.1 General
9.3.2 Ecological Considerations
269
270
270
272
273
276
278
278
278
9.3.3 General Biotransformation Processes
280
Biotransformation of Haloaliphatic Compounds

9.4.1 General
9.4.2 Aerobic Biotransformations
9.4.3 Anaerobic Biotransformations
9.4.3.1 General
9.4.3.2 Haloalkanes
9.4.3.3 Haloalkenes
283
283
283
286
286
286
288
290
9.4.5 Field Studies

9.4.6 Bioremediation
Conclusions
References
299
300
303
304
9.3
9.4
9.5
9.6
10
References
THE EFFECTS OF CHLORINATED SOLVENTS ON THE PERMEABILITY OF

ClAYS
10.1
10.2
10.4
10.5
Introd uction
313
Mechanisms Controlling the Swelling of Clays

10.2.1 General
314
315
315
10.2.2 Interlayer Mechanisms

10.2.3 Interparticle Mechanisms
Relevant Properties of Organic Chemicals
Laboratory Investigations of Clay/Organic Interactions
10.5.1 X-Ray Diffraction Studies of Intracrystalline Swelling
10.5.2 Laboratory Permeability Tests
10.5.2.1 General
315
316
318
318
318
319
319
xi
Contents
10.5.2.2 The influence of permeameter type on laboratory

determination of permeability
320
10.5.2.3 Hydraulic gradients used in permeability tests
10.6
10.7
10.8
11
12
10.5.2.4
10.5.2.5
Field Studies
10.5.2.7
10.5.2.6
Conclusions
References
322
Permeability tests on "dry" clays/organic mixtures

Permeability tests of water-wet clays and NAPLs
Permeability tests of clays and water-miscible organic fluids
Permeability tests using aqueous solutions of NAPL chemicals
323
323
330
328
328
331
333
PHYSICS GOVERNING THE MIGRATION OF DENSE NON-AQUEOUS PHASE

LIQUIDS (DNAPLS) IN FRACTURED MEDIA
337
11.1
11.2
11.3
11.4
11.5
11.6
11.7
Introduction
DNAPL Entry into Fractures
Rates of DNAPL Migration Through Fractures
Residual and Pooled DNAPL
DNAPL Migration Through Fracture Networks
Height of DNAPL in a Well
Conclusions
337
338
343
347
349
351
352
11.8
References
352
THE EFFECTS OF MOLECULAR DIFFUSION ON DNAPL BEHAVIOR IN

FRACTURED POROUS MEDIA
12.1
12.2
12.3
12.4
12.5
Introduction
DNAPL Persistence in Fractures
12.2.1 Approach
12.2.2 Contaminant Mass Storage Capacities
12.2.3 Rates of DNAPL Loss from Single Fractures
12.2.4 Rates of DNAPL Loss from 3-D Fracture Networks
12.2.5 Limitations
Implications for DNAPL Disappearance from Fractures
12.3.1 Interpretation of Monitoring Well Data
12.3.2 Flow, Immobilization, and Invasion Distances
12.3.3 Source Zone Definition and Delineation
12.3.4 Data Needs for Site Characterization
12.3.5 Remediation
Dissolved-Phase Transport in Fractured Porous Media
12.4.1 Example 1: Solute Transport Through a Vertically Fractured
Clay Aquitard
12.4.2 Example 2: Solute Transport in Fractured Porous Rock
Conclusions
355
356
361
361
364
365
371
375
378
378
379
380
381
381
382
382
386
387
xii
12.6
13
Contents
References
390
DIAGNOSIS AND ASSESSMENT OF DNAPL SITES
395
13.1
397
398
398
403
404
405
405
406
409
409
410
412
413
413
413
413
418
421
421
422
423
425
425
426
427
429
429
430
432
432
432
435
437
437
438
440
13.2
13.3
13.4
13.5
Introduction
General Site Assessment
13.2.1 Types of Chemicals That Are DNAPLs
13.2.2 Uses for Chemicals That Are DNAPLs
13.2.3 Assessment of Site Operations
13.2.4 Assessment of DNAPL Penetration
13.2.4.1 Influencing factors
13.2.4.2 DNAPL penetration in the vadose zone
13.2.4.3 Effects of DNAPL vaporization
13.2.4.4 DNAPL penetration in the groundwater zone
13.2.4.5 DNAPL penetration in fractured media
13.2.4.6 Reliability of estimates
Determining the Presence of DNAPL Based on Field Observations
13.3.1 Observations During Drilling
13.3.1.1 Visual evidence
13.3.1.2 Field testing
13.3.2 Observations in Wells
13.3.3 Short Circuiting of DNAPL During Drilling and Sampling
13.3.3.1 Short circuiting pathways
13.3.3.2 Investigations using the "outside-in" approach
13.3.3.3 Precautions during drilling and sampling
13.3.4 Alternatives to Drilling and Wells
13.3.4.1 Test pits
13.3.4.2 Drive-point investigations
13.3.4.3 Geophysics
Soil Gas Surveys
13.4.1 Description
13.4.2 Factors Influencing Vapor Concentrations
Chemical Analysis of Soil and Rock
13.5.1 Soil Samples
13.5.1.1 Sample collection
13.5.1.2 Sample analysis
13.5.2 Partitioning Calculations
13.5.2.1 General
13.5.2.2 Method description
13.5.2.3 Example calculation: contamination with TCE
13.5.2.4 Example calculation: contamination with PCBs
in hydrocarbons
13.5.2.5 Reliability
13.5.3 Direct Determination of DNAPL
442
443
443
Contents
13.6
13.7
13.8
13.5.4 Sampling and Analysis of Fractured Media

13.5.4.1 DNAPL in fractures
13.5.4.2 DNAPL components in the matrix
Character of Dissolved Plumes
13.6.1 General
13.6.2 Spatial Distribution of Dissolved Plumes
13.6.3 Magnitude of Dissolved Concentrations
444
444
446
449
449
449
454
13.6.4 Persistence and Temporal Variability of Dissolved Concentrations

13.6.5 Response of Dissolved Concentrations During Aquifer Pump-and-Treat
13.6.6 Case History-Chlorinated Solvent Plume, Dayton, New Jersey
13.6.6.1 Background
13.6.6.2 Extent of groundwater contamination
13.6.6.3 Remedial measures
13.6.6.4 Regeneration of the plume
Conclusions
References
459
460
462
462
462
464
465
14 CONCEPTS FOR THE REMEDIATION OF SITES CONTAMINATED WITH

DENSE NON-AQUEOUS PHASE LIQUIDS (DNAPLS)
14.1
14.2
14.3
14.4
14.5
14.7
14.8
14.9
xiii
Introduction
Characteristics of DNAPL Sites
Remediation Goals
14.3.1 Plume Containment
14.3.2 Aquifer Restoration
14.3.3 Source-Zone Containment/Plume Removal
Low Permeability Enclosures
Pump-and-Treat
Source Zone Restoration
14.7.1 Source Zone Restoration in the Vadose Zone
14.7.2 Source Zone Restoration in the Groundwater Zone
14.7.2.1 General
14.7.2.2 Pumping of pools
14.7 .2.4 Mass-destruction technologies
14.7.2.5 Prospects for source zone restoration
Conclusions
References
475
476
478
482
482
482
483
484
489
494
494
496
496
496
501
502
503
505
APPENDIX
507
INDEX
513
Preface
A decade ago, little was known concerning the behavior of dense chlorinated solvents
and other dense non-aqueous phase liquids (DNAPLS) in the subsurface. Even less was
known about what options and prospects existed for the remediation of sites contaminated with DNAPLs. Research activity in this area grew rapidly in the late 1980s, and has
accelerated in the current decade. The first major research program in North America to
focus on dense organic solvents was the University Consortium Solvents-In-Groundwater
Research Program. This program was initiated in 1987, became well-established in 1988,
and continues today. The consortium is made up of four principal collaborating universities: two in Canada (University of Waterloo and Queens University), and two in the
United States (Colorado State University and the Oregon Graduate Institute). Several
other universities and government organizations have been involved as collaborators
over the years. These include SUNY/Buffalo, the University of Toronto, the University
of Western Ontario, Environment Canada (National Water Research Institute), the U.S.
Geological Survey (Water Resources Division), and Mitre Corporation. Most recently,
the University of Stuttgart and the Danish Geological Survey have become collaborating
organizations.
Financial support for the Solvents-In-Groundwater Research Program has come
from The Boeing Company, Ciba-Geigy Corporation, Dow Chemical Canada/USA, Eastman Kodak Co., General Electric Co., Laidlaw Environmental Systems Ltd., Mitre Corporation, The Natural Sciences and Engineering Research Council of Canada, the Ontario
University Research Incentive Fund, PPG Industries Corporation, and United Technologies Corporation. Infrastructure support has been provided by the Waterloo Centre for
Groundwater Research (an Ontario Centre of Excellence) at the University of Waterloo
and Canadian Forces Base Borden. The administrative framework for the corporate and
government funding for the program is distinctive. Each member company and government agency makes an equal, annual contribution to the research program. The funds are
pooled, and then allocated to the researchers based on an annual plan determined by the
university consortium members in consultation with the sponsors. The goal has been to
provide a maximum of scientific and operational flexibility with a minimum of bureaucracy. The research program has many components. As components are completed, the
results are presented in M.S. and Ph.D. theses, and published as peer-reviewed journal
articles, conference proceedings, and symposia. Many of these publications are cited in
the chapters in this book. Each year, new projects within the program are begun.
xv
xvi
Preface
The goals of the research program are to acquire and integrate information regarding DNAPLs in the subsurface, and the associated groundwater- and vadose-zone
contamination. A range of laboratory experiments, field experiments, and mathematical
modeling studies have been used in this work. The research is being conducted by a
group of faculty members, research associates, technicians, and graduate students in science and engineering, a group whose size varies from 50 to 70 people. Large, controlled,
field experiments in which liquid chlorinated solvents have been introduced into a sandy
aquifer on a military base (Canadian Forces Base Borden) and into large tanks packed
with layered porous media have been used as reality checks on the mathematical models,
and for performance assessments of selected technologies for remediating contaminated
groundwater and vadose zone systems.
Much of the information presented in this book has been developed in the context
of the above research program. Although our understanding of the behavior of dense
chlorinated solvents and other DNAPLs in groundwater systems is continuing to advance,
the body of scientific knowledge on many aspects of this subject has reached sufficient
maturity to warrant a synthesis at this time. The intention of this book is to provide that
synthesis. We thank the corporate sponsors and government agencies that have made this
work possible, and we thank the authors whose commitment and attention to detail have
made this book a reality.
John A. Cherry
Waterloo, Ontario
June 1995
James F. Pankow
Portland, Oregon
June 1995
Sponsors of the University Consortium Solvents in Groundwater Research Program
The Boeing Company, Seattle, WA, 1991-present;
Ciba-Geigy Corporation, Ardsley, NY, 1988-present;
Dow Chemical Company, Midland, MI, 1988-1991;
Eastman Kodak Company, Rochester, NY, 1990-present;
General Electric Company, Hartford, CT, 1988-present;
Laidlaw Environmental Services, Burlington, ON, 1992-1995;
Mitre Corporation, 1993-present;
Natural Sciences and Engineering Research Council of Canada, 1988-1994;
The Ontario University Research Incentive Fund, 1988-1994;
PPG Industries Corporation, Pittsburgh, PA, 1994-present; and
United Technologies Corporation, 1994-present.
1
Dense Chlorinated Solvents in
Groundwater: Background
and History of the Problem
James F. Pankow 1, Stan Feenstra2 ,3 , John A. Cherry",
and M. Cathryn Ryan3
1 Department
of Environmental Science and Engineering

Oregon Graduate Institute
P.O. Box 91000
Portland, Oregon
U.S.A. 97291-1000
2Applied Groundwater Research, Ltd.

The Pentagon Building, Suite 207
2550 Argentia Road
Mississauga, Ontario
Canada L5N 5R1
3Waterloo Centre for Groundwater Research
University of Waterloo
Waterloo, Ontario
Canada N2L 3G1
Page
1.1
1.2
1.3
1.4
1.5
1.6
1.7
Background
1.1.1 General
1.1.2 DNAPLs and LNAPLs
Chlorinated Solvent Usage
The Physical, Chemical, and Biological Properties of Chlorinated Solvent
Compounds
A History of How We Became Aware that Dense Chlorinated Solvents Can
Pose Serious Threats to Groundwater Systems
1.7.1 General
3
3
6
7
11
13
13
15
16
16
16
16
1.8
1.9
Dense Chlorinated Solvents: Background and History of Problem

1.7.3 The First Stage: Discovering that Chlorinated Solvents are in Groundwater
1.7.3.1 The 1950s to the mid-1960s: the early years
1.7.3.2 The mid-1960s to 1972: finding toxic organic compounds in
some drinking water supplies
1.7.3.3 1972 to 1975: finding the first chlorinated VOCs in drinking
water: the trihalomethanes (THMs)
1.7.3.4 1976 to 1979: widespread testing for THMs in drinking water
reveals the presence of other VOCs: coming to the conclusion
that many important aquifers are contaminated with
chlorinated solvents
1.7.3.5 1980 to 1981: the evidence that chlorinated solvents are
widespread groundwater contaminants becomes overwhelming
1.7.4 The Second Stage: Discovering the DNAPL Nature of Liquid Chlorinated
Solvents
1.7.4.11982tothelate1980s
Conclusions
References
Chap. 1
17
17
17
18
20
21
23
26
26
28
28
46
ABSTRACT
Chlorinated organic solvents like trichloroethylene are common groundwater contaminants. Contamination levels are usually in the MgfL to mg/L range (dissolved). The U.S.
production is in the hundreds of millions of kg per year for most of the chlorinated
solvents. When found in the subsurface as organic liquids, these compounds are denser
than water, and are then commonly referred to as "dense non-aqueous phase liquids"
(DNAPLs). If spilled into the subsurface in sufficient quantities. a DNAPL has the capacity to penetrate below the water table where it can provide a long term source of
contamination. Other reasons why chlorinated solvents are common groundwater contaminants are that they exhibit: I) low liquid viscosities (are able to move easily into
the subsurface); 2) low interfacial tensions with water (are able to enter into water-wet
fractures relatively easily); 3) high volatilities (are able as gases to diffuse rapidly downwards into the unsaturated zone); 4) low absolute solubilities (are difficult to remove
from the groundwater zone); 5) high solubilities relative to drinking water limits (are
able to cause significant contamination even when small amounts dissolve); 6) low partitioning to soils (are not retarded by aquifer materials); and 7) low degradabilities. At
many sites of contamination by chlorinated solvents, the amounts that are present in
dissolved form in the groundwater are relatively small. often no more than the equivalent of a few tens of drums liquid solvent. Since spill volumes have in most cases been
much larger than this, significant DNAPL solvent still exists in the subsurface at most of
these sites.
The history of the problem of chlorinated solvents in groundwater is examined in
this chapter in the hopes that doing so will help scientists and engineers avoid other environmental problems in the future. The physical and chemical properties of the chlorinated
Sec. 1.1
Background
solvents should have led scientists and engineers to anticipate that these compounds would
be prone to causing groundwater contamination. Unfortunately, their common presence
in important aquifers was only discovered by accident during the mid-to-late 1970s as
a result of an interest in the trihalomethane compounds (e.g., chloroform) that can be
formed when drinking water is chlorinated. Late 1981 can be viewed as the point in time
at which the problem of dissolved chlorinated solvents in groundwater became widely
recognized. However, the fact that chlorinated solvents also exist in the subsurface in
DNAPL form at many important sites did not become generally recognized until the
mid-to-late 1980s. By this point in time, many remediation efforts had already failed
because they had not taken into account the large, persistent DNAPL sources present in
the subsurface at those sites.
1.1 BACKGROUND
1.1.1 General
Chlorinated solvents were first produced in Germany in the nineteenth century; production in the United States began around 1906. Widespread use of chlorinated solvents
in manufacturing industries began during World War II, and increased markedly during
the next three decades. Contamination of groundwater by these compounds went largely
unrecognized until the late 1970s. The denser-than-water nature of liquid chlorinated solvents has led to their becoming known as "dense non-aqueous phase liquids" (DNAPLs,
see Tables 1.1 and 1.2).
Like petroleum products (e.g., gasoline and fuel oil), chlorinated solvents are produced and usually used in an organic liquid form (i.e., as "neat", "free product" liquids),
but are generally found in water supplies in dissolved form at f..LglL (ppb) to mg/L (ppm)
concentrations. Unlike petroleum products, however, the chlorinated solvents are generally: 1) not noticed by taste or odor at typical groundwater contamination levels; and
2) not stopped at the water table when released in DNAPL form to the subsurface in
significant quantities. Once in an aquifer, DNAPL source material is usually very difficult
to find. In contrast, when petroleum products are spilled in significant volumes, being
light NAPLs (i.e., being LNAPLs), they "float" on the water table where they can be
detected easily (e.g., see McKee et al., 1972). Historically, since chlorinated solvents in
DNAPL form were found only rarely in wells, recognition of the problems posed by
these compounds in groundwater had to await the development and general availability
of analytical methods which allowed the determination of dissolved organic contaminants
in water at the f..LglL to mg/L level.
It is now apparent that of the many industrial organic chemicals identified to date
in groundwater, the common chlorinated organic solvents are the most ubiquitous. In a
survey in New Jersey of 1,070 wells, trichloroethylene (TCE) was found in 58% of the
wells, carbon tetrachloride in 65%, and tetrachloroethylene (PCE) in 43% (Page, 1981).
Because of the DNAPL nature of many of the corresponding contamination source zones,
the chlorinated solvents pose great challenges in groundwater remediation (Plumb and
Pitchford, 1985; Abelson, 1990).
TABLE 1.1
Chap. 1
Organic Compounds That Are DNAPLs.
Compound class
Compound
Chlorinated solvents
Tetrachloroethylene
Trichloroethylene
I,I,I-Trichloroethane
Dichloromethane
Chloroform
Carbon tetrachloride
1,2-Dichloroethane
Chlorobenzene
1,2-Dichlorobenzene
1,3-Dichlorobenzene
I, I,2-Trichloroethane
1,2,4-Trichlorobenzene
Other halogenated
organics
Benzyl chloride
Bromobenzene
Bromochloromethane
Bromodichloromethane
Bromoform
4-Bromophenyl phenyl ether
bis(2-Chloroethyl)ether
2-Chloroethyl vinyl ether
bis(2-Chloroisopropyl)ether
I-Chloro-I-nitropropane
4-Chlorophenyl phenyl ether
Dibromochloromethane
I,I-Dichloroethane
1,2-Difluorotetrachloroethane
1-Iodopropane
Hexachlorobutadiene
Pentachloroethane
I, I ,2,2-Tetrabromoethane
1,1,2,2-Tetrachloroethane
1,2,3-Trichloropropane
I, I,2-Trichloro-I ,2,2-trifluoroethane
The first significant recognition in the research community of the potential DNAPL
character of chlorinated solvents in groundwater is attributable to Friedrich Schwille.
As a result of two case studies in West Germany during the late 1960s and early 1970s
involving chlorinated solvents, Schwille developed conceptual models and conducted
physical model studies on the behavior of these compounds at the Federal Institute of
Hydrology in Koblenz from 1977 to 1984. Schwille examined how and when chlorinated
solvents could penetrate into unsaturated and saturated sands, and into fractured media,
and how liquid solvents could be dissolved by flowing groundwater. Unfortunately, this
research remained largely unknown in North America and even in Europe until the late
1980s. At that time, an understanding of the behavior of chlorinated solvents in the
Sec. 1.1
Background
TABLE 1.1
(continued)
Substituted aromatics,
phtbalates, and
miscellaneous
organics
PCB mixtures
Pesticides
Chloroanilines
Chlorotoluenes
Nitrotoluenes
Nitrobenzene
Benzyl butyl phtbalate
Di-n-butyl phtbalate
Diethyl phthalate
a-Anisidine
Phenyl etber
Tri-a-cresol phosphate
Aroclor 1221
Aroclor 1232
Aroclor 1242
Aroclor 1248
Aroclor 1254
Chlordane
Chloropicrin
1,2-Dibromo-3-chloropropane
1,2-Dichloropropane
1,2-Dichloropropylene
Dichlorvos
Disulfoton
Etbion
Ethylene dibromide
Malatbion
Paratbion
TABLE 1.2 The Most Common Chlorinated Solvents With Some of Their Chemical Synonyms and
Commerical Product Names (Merck Index, 1989).
Chemical name
Synonyms
Perchloroethylene
PeE, tetrachloroethylene, etbylene tetrachloride, Nema, Tetracap, Tetropil,

Perclene, Ankilostin, Didakene, PerSec
Trichloroethylene
TCE, ethinyl trichloride, Tri-Clene, Trielene, Trilene, Trichloran, Trichloren,

Algylen, Trimar, Trline, Tri, Trethylene, Westrosol, Chlorylen, Gemalgene,
Germalgene
I, I,I-Trichloroethane
TCA, metbyl chloroform, Chlorotbene, Solvent 111, TRI-ETHANE
Dichloromethane
DCM, metbylene chloride, methylene dichloride, metbylene bichloride
subsurface was being developed independently, largely without knowledge of Schwille's

pioneering studies t.
tSchwille was also involved in early research into tbe behavior of LNAPLs in the subsurface
(Schwille, 1967).
Chap. 1
Schwille (1981) was the first English language publication to present the concept
of heavier-than-water solvents sinking below the water table. Feenstra (1984), Cherry
(I 984a, 1984b), Villaume (1985), and Feenstra (1986) also provided descriptions of the
general concepts governing DNAPL behavior in the subsurface. A summary report on
Schwille's impressive accomplishments was published in German (Schwille, 1984a), then
in English translation (Schwille, 1988). More recent discussions have been presented by
Feenstra and Cherry (1988), Hunt et al. (1988a, 1988b), Mercer and Cohen (1990), and
Cohen and Mercer (1993).
This chapter provides background information on chlorinated solvent usage, and
examines the history of the two-staged discovery of the extent and causes of groundwater
contamination by these compounds. The first stage of the discovery involved the simple
recognition of the prevalence of chlorinated solvents as groundwater contaminants; the
second stage involved the recognition of the common presence of these compounds as
difficult-to-remediate DNAPL source zones in the subsurface at many sites of contamination. This history is presented so as to place the following chapters in context in the
field of contaminant hydrogeology, and also to permit an understanding how and when
we came to only belatedly appreciate the nature and magnitude of this problem. Indeed,
an analysis of how we came to appreciate the chlorinated solvents problem will permit
us to critically evaluate the way in which science has progressed in the groundwater
community.
1.1.2 DNAPls and lNAPls

Historically, the term "NAPL" was coined in 1981 during studies of a hazardous waste
landfill (the "S-Area Landfill") in Niagara Falls, New York. At this site, a black, denserthan-water, "immiscible" organic liquid was discovered in the soil and bedrock underlying
the site. The liquid was found to be a complex mixture of halogenated benzenes, chlorinated solvents, and other halogenated hydrocarbons. Engineers, scientists, and attorneys
dealing with the site selected the term "NAPL" to differentiate this liquid from other
contaminated materials, and from the groundwater (Kolmer, 1991), though we would
now refer to it more specifically as a "DNAPL".
DNAPLs have sometimes also been referred to as "dense immiscible liquids",
and LNAPLs have sometimes been referred to as "light immiscible liquids". However,
as with the use of "non-aqueous" in the terms DNAPL and LNAPL, the descriptor
"immiscible" is misleading. Indeed, an organic liquid can exist as a stable, separate phase
in equilibrium with water only after its dissolved concentration in the water has reached
the saturation limit for that liquid. Thus, no organic liquid is ever totally immiscible with
water. We also note that there will always be some water dissolved in any "non-aqueous"
phase.
Although DNAPL solvents are now recognized as a major groundwater problem,
only a few published descriptions of actual field cases are readily available at this time.
We have, however, reviewed unpublished data pertaining to a large number of industrial
and waste disposal sites across North America where DNAPL solvents are the key cause
of groundwater contamination. These include many sites at which data has been collected
Sec. 1.3
as a result of Superfund (i.e., CERCLA) and RCRA legislation, though most of this
voluminous information lies buried in the files of regulatory agencies.
1.2 CHLORINATED SOLVENT USAGE
In hindsight, it is not surprising that widespread groundwater contamination by chlorinated solvents has occurred. Very large quantities of these chemicals have been, and
continue to be produced, transported, used, and disposed of by many sectors of our modem society. Production of chlorinated solvents began in the U.S. in 1906 with carbon
tetrachloride, which remained the main solvent used in North America for the first half
of the century. Production of TCE and PCE began in 1923. Carbon tetrachloride was
commonly used for dry cleaning in the 1930s, having replaced gasoline. TCE and PCE
became the most commonly used solvents in the 1960s when the post-World War II
manufacturing economy expanded greatly. Increased chlorinated solvent use paralleled
economic growth into the 1970s. Human health and environmental concerns, first raised
in the 1970s, have since caused a decline in the production of PCE and TCE to fifth and
sixth rank in U.S. production (Figure 1.1). Table 1.3 shows the 1986 U.S. production
of selected chlorinated organic solvents. Production quantities range from hundreds of
millions to billions of kg per year.
A very wide range of industries use chlorinated solvents in large quantities. The
principal current uses of the four most common chlorinated solvents are given in Table 1.4, and the principal uses of several less common chlorinated solvents are given
in Table 1.5. As a result of their widespread use, the release of chlorinated solvents to
the subsurface has by no means been limited to the "smoke stack" industries that have
traditionally been associated with pollution. Indeed, the electronic, instrument manufacturing, and aerospace industries, all generally regarded as relatively "clean" with respect
to pollution, have used very large quantities of chlorinated solvents, and have caused
numerous spills. Also, a range of small businesses in virtually every community use
chlorinated solvents, and have been responsible for problematic spills. These include dry
cleaning, machine, photographic processing, and printing shops. The principal uses of
other organic chemicals which are DNAPLs in pure form are given in Tables 1.6 to 1.9.
An overview of the industrial and commercial activities which have produced or used
DNAPL chemicals is given in Table 1.10.
1.3 SOURCE TYPES FOR SOLVENTS IN GROUNDWATER

Table 1.11 summarizes the different ways in which chlorinated organic solvents have been
released to the subsurface environment. Because chlorinated solvents are now recognized
as being among the most problematic groundwater contaminants in industrialized countries, some major solvent users are switching to other chemicals where possible, and/or
are implementing much more careful solvent-handling procedures. For example, many
underground solvent storage tanks are being replaced with above-ground tanks which can
be more effectively inspected for leaks. However, usage has been and will continue to be
widespread because the common chlorinated solvents have exceptionally good properties
400
a 300
...:
s 200
~
100
...---h...
~'---
/'
i...
8...
400
.i'300
.~ 200
r---2100
~ r-,
;;;;
<,
1..:.=...1
--.
Ii...
400
a 300
...:
.2oo
;;;;
2 100
----
Ii...
----
~: /
.--l
1----""'"
TeA
.....
.:>
Figure 1.1 U.S. production of common chlorinated solvents during the period 19601990. Data are from the Halogenated Solvents Industry Alliance (1987) and the U.S.
International Trade Commission (1991).
Chap. 1
Sec. 1.4
Physical, Chemical, and Biological Properties

TABLE 1.3
1986 U.S. Production of Chlorinated Organic Solvents.
Compound
Abbreviation
(10 6 lb.)
(106 kg)
Drums"
1,2-Dichloroethane"
1,1,1-Trichloroethane"
Carbon tetrachloride"
Methylene chloride"
Chloroform"
Tetrachloroethylene"
Trichloroethylene"
1,2-DCA
l,l,l-TCA
CTET
DCM
TCM
PCE
TCE
12,940
648
627
561
422
405
165
5,871
294
284
255
191
184
75
22,900,000
1,100,000
890,000
960,000
640,000
560,000
260,000
"American Chemical Society (1986)

bHalogenated Solvents Industry Alliance
cBased on 200 L, or approximately 55 US gallons per drum
TABLE 1.4 Principal Current Uses of the Major Chlorinated Solvent

Products. (Data from Halogenated Solvents Industry Alliance, 1987.)
Application
1,1,I-TCA
Adhesives
Aerosols
Chemical intermediate
Cold cleaning
Dry cleaning/textile production
Electronics
Metal cleaning/degreasing
Paint removal/stripping
Uretbane foam
Miscellaneous
9
II
Percentage use
PCE
DCM
TCE
29
29
20
56
6
44
8
9
27
10
8
II
85
1
I, 1/ . TCA - 1,1,1-Trichloroethane
DCM - Methylene chloride

PCE - Tetrachloroetbylene
TCE - Trichloroethylene
for their intended uses. Thus, it seems likely that releases of chlorinated solvents to the
subsurface (and therefore to groundwater zones) will continue to occur at some rate in
nearly all urban or suburban areas in the industrialized world.
1.4 THE PHYSICAL, CHEMICAL, AND BIOLOGICAL PROPERTIES OF
CHLORINATED SOLVENT COMPOUNDS
In addition to the facts that: 1) large quantities of DNAPL solvents have been produced
and used (Table 1.3); and that 2) these solvents have been able to enter the subsurface
environment due to a variety of different source types (Table 1.11), it is important to
recognize that the physical, chemical, and biological properties of the chlorinated solvents have made these chemicals particularly likely causes of groundwater contamination
10
Chap. 1
TABLE 1.5 Principal Uses of Other Common Chlorinated Organic Compounds Which Are DNAPLs
or DNAPL Components.
Uses
Chemical
Chloroform
Solvent for fats, oils, rubber, waxes, resins; manufacture of rubber
Solvent in oils, fats, laquers, varnishes, waxes, resins; dry cleaning; seed
fumigant; extraction of seed oils
1,2-Dichloroethane
Solvent for fats, oils, waxes, gums, resins, rubber; seed fumigant, manufacture
of acetyl cellulose; manufacture of vinyl chloride
Chlorobenzene
Solvent for paints; manufacture of phenol, aniline, DDT, dyes; heat transfer
liquid
1,2-Dichlorobenzene
Solvents for waxes, gums, resins, tars, rubber, oils, asphalts; insecticide for
termites; degreasing of leather and wool; manufacture of dyes; heat transfer
medium
1,3-Dichlorobenzene
Fumigant and insecticide
I, I,2-Trichloroethane
Solvent for fats, waxes, resins
Dielectric fluids; manufacture of chemicals, dyes; heat transfer medium
(Pankow and Cherry, 1988). These properties and their implications are summarized in
Table 1.12.
As pointed out in Tables 1.1 and 1.5-1.9, organic compounds besides the chlorinated solvents are classifiable as DNAPLs. These compounds include the halogenated
benzenes, the polychlorinated biphenyls (PCBs), some pesticides, coal tar, and creosote.
These chemicals have not been found in groundwater nearly as frequently as have the
chlorinated solvents. There are several reasons for this. Firstly, compared to the chlorinated solvents, these other chemicals have not been used in as many industrial applications, or in quantities that are nearly as large. Secondly, they do not possess a combination
of physical and chemical properties that impart a high propensity for widespread groundwater contamination. For example, some of these other DNAPL chemicals are much less
soluble than the chlorinated solvents, and so they are much less mobile because they
partition strongly to soil materials. As another example, important soluble components
of creosote and coal tar are more degradable in the groundwater zone than are most
chlorinated solvents.
1.5 DRINKING WATER LIMITS
The drinking water limits (or "maximum concentration limits", MCLs) for several of
the chlorinated solvents have been set at very low levels, e.g., 5 JLgfL for TCE (EPA,
1987a). The justification for these standards has been that compounds like TCE,
chloroform, and carbon tetrachloride are suspected carcinogens. There is also evidence
that excessive exposure to carbon tetrachloride and PCE may result in kidney and
liver damage. Table 1.13 illustrates how the solubilities of the chlorinated solvents are
all many orders of magnitude higher than their corresponding drinking water limits.
11
Sec. 1.5
TABLE 1.6
Principal Uses of Less Common Halogenated Industrial Chemicals Which Are DNAPLs.
Chemical
Uses
Benzyl chloride
Manufacture of benzyl compounds, perfumes, pharmaceuticals,

dyes, tannins, resins
Bromobenzene
Organic synthesis; solvent; additive in motor oils
Bromochloromethane
Used in fire extinguishers; organic synthesis
Solvent for waxes, oils, fats, resins; used in fire extinguishers;

flame retardant; organic synthesis
Bromoform
Mineral separations
4-Bromophenyl phenyl ether
Research chemical
bis(2-chloroethyl) ether
Cleaning textiles; soil fumigant; Processing of fats, waxes,

greases; solvent in paints, varnishes, lacquers; manufacture of
lubricating oils, finish removers, cleaners
2-chloroethyl vinyl ether
Manufacture of anesthetics, sedatives, cellulose ethers
bis(2-chloroisopropyl) ether
Manufacture of dyes, resins, pharmaceuticals; solvent for fats,

waxes, greases; used in textile manufacturing; paint and spot
removers
l-Chloro-J-nitropropane
Fungicide
4-Chlorophenyl phenyl ether
Research chemical
Used in fire extinguishers; aerosol propellants; refrigerants;

organic synthesis
1,1-Dichloroethane
Used in paint, varnish and finish removers; manufacture of

vinyl chloride and I,I,I-TCA; rubber cement; solvent for
plastics, oils and fats; insecticide and fumigant
1,2-Difluorotetrachloroethane
Used in organic synthesis
1-Iodopropane
Used in organic synthesis
Hexachlorobutadiene
Solvent for elastomers, rubbers; manufacture of fluorinated

lubricants and rubber compounds; heat transfer medium;
transfomer and hydraulic fluid; fumigant for grapes
Pentachloroethane
Solvent for oil and grease in metal cleaning
1,1,2,2-Tetrabromoethane
Mineral separation
I, I,2,2-Tetrachloroethane
Solvent for chlorinated rubber; manufacture of insecticides,

bleach, paint/varnish/polish removers; metal cleaning;
manufacture of I,I-DCE; solvent for oils and fats, pesticides
Solvent; paint and varnish remover
I, I,2-Trichloro-I ,2,2-trifluoroethane
Solvent used in electronics and precision equipment manufacture; used in fire extinguishers; dry cleaning; blowing agent
The chlorinated solvents are thus characterized by high solubilitylMCL ratios (see
also item 6 in Table 1.12). Therefore, although the chlorinated solvents have low absolute solubilities, the contamination coming off of a pool of chlorinated solvent DNAPL
will typically be greatly in excess of its MCL: even relatively small quantities of
these materials in the subsurface can result in large-scale groundwater contamination
problems.
12
TABLE 1.7
Chap. 1
Principal Uses of Substituted Aromatic Compounds and Phthalates Which Are DNAPLs.
Chemical
Uses
Chloroanilines
Manufacture of dyes, pharmaceuticals and agricultural chemicals
Chlorotoluenes
Manufacture of dyes; organic synthesis
Nitrotoluenes
Manufacture of dyes, nitrobenzoic acids, toluidenes; organic synthesis
Nitrobenzene
Solvent for cellulose ethers; used in metal and shoes polishes; manufacture of
aniline, benzidine, quinoline, azobenzene, drugs, photographic chemicals
Benzyl butyl phthalate
Plasticizer for polyvinyl chloride (PVC); additive in polyvinyl actetate

emulsions, ethylene glycol, ethyl cellulose
Di-n-buytl phthalate
Plasticizer; organic synthesis; insect repellant
Diethyl phthalate
Plasticizer; denaturant for ethanol; ingredient in insecticide sprays, explosives;

used in dyes, perfumes; solvent for nitrocellulose and cellulose acetate
o-Anisidine
Manufacture of aZO dyes
Phenyl ether
Heat transfer medium; perfuming soaps; resins for laminated electrical

insulation; organic synthesis
Tri-o-cresol phosphate
Plasticizer in lacquers, varnishes, PVC, polystrene, nitrocellulose;

waterproofing; heat transfer medium; hydraulic fluids; fire retardant for plastics;
additive to high pressure lubricants
TABLE 1.8
Principal Uses of PCB Aroclors Which Are DNAPLs.
PCB mixture
Uses
Aroclor 1221
Capacitors, gas-transmission turbines, hydraulic fluids, adhesives, plasticizer in rubbers
Aroclor 1232
Hydraulic fluids, adhesives, plasticizer in rubbers
Aroclor 1242
Transformers, hydraulic fluids, adhesives, plasticizer in rubbers, heat transfer medium,

wax extender, carbonless reproducing paper
Aroclor 1248
Vacuum pumps, gas-transmission turbines, hydraulic fluids, plasticizer in resins,

adhesives, plasticizer in rubbers
Aroclor 1254
Capacitors, transformers, vacuum pumps, hydraulic fluids, plasticizer in resins,

adhesives, plasticizer in rubbers, wax extender, dedusting agent, extender in
pesticides/inksllubricants/cutting oils
1.6 TYPICAL CHLORINATED SOLVENT PLUMES IN GROUNDWATER
Mackay and Cherry (1989) have calculated that the dissolved solvents in several problematic plumes measuring up to several kilometers in length represent the equivalent of
only 0.5 to 70 drums of liquid solvent. Our unpublished assessments of more than 30
other major chlorinated solvent plumes at Superfund sites show that this range for plume
mass is typical of large plumes. These amounts represent small fractions of the quantities of solvent that presumably entered the groundwater zone at these sites. Therefore,
significant DNAPL source zones presumably: I) still exist at these sites; 2) will be feeding
their respective plumes long into the future.
Sec. 1.6
TABLE 1.9
13
Principal Uses of Pesticide Products Which Are DNAPLs.
Pesticide product
Uses
Chlordane
Insecticide, acaricide, termite control
Chloropicrin
Seed and grain fumigant
1,2-Dibromo-3 -chloropropane
Soil fumigant, nematocide
l,2-Dichloropropane
Soil fumigant, nematocide; manufacture of PCE; solvent
l,3-Dichloropropylene
Soil fumigant, nematocide
Dichlorvos
Insecticide and fumigant
Disulfoton
Insecticide for tobacco and vegetables
Ethion
Ethylene dibromide
Insecticide, acaricide
Grain and fruit fumigant; anti-knock additive in gasoline
Malathion
Parathion
Insecticide
Insecticide, acaricide
TABLE 1.10 Industrial and Commercial Activities Which Produced or Used

DNAPL Chemicals.
Chemical production and distribution
Manufacturing uses
Solvents
General metal fabrication
Solvents (recycled)
Tool and die making
Dyes and paints
Aircraft production
Electronics/electrical fabrication
Adhesives
Pharmaceuticals
Automobiles and parts production
Plastics and polymers
Printing
Pesticides
Coal tar distillates
Weapons production
Photographic work
Creosote
Wood preserving
Transformer and capacitor production

Foundry operations and castings
Service uses
Commercial and retail dry cleaning
Pesticide application
Waste products
Coal and oil gasification by-products
Blast furnace coking by-products
Food processing (decaffeinated coffee)
Compared to the number of DNAPL sites, there are many more sites where
LNAPLs (e.g., gasoline, fuel oil, and aviation fuel) have been spilled or leaked into
groundwater systems. However, large plumes of dissolved chemicals (e.g., benzene,
toluene, ethyl benzene, and the xylenes) from LNAPL source zones are rare in comparison
14
Chap. 1
TABLE 1.11 Source Types for Chlorinated Solvent Contamination of

Groundwater.
1. Leaking underground or above-ground storage tanks.

Leaking drum storage areas.
Leaking, buried chemical distribution pipelines.
Spillage at chemical loading and off-loading facilities.
Spillage during highway accidents and train derailments.
Intentional disposal into the subsurface in various ways, including:
a. domestic septic tile fields (due to the use of solvent-containing
septic tank cleaning fluids);
b. municipal landfills;
c. chemical waste disposal landfills;
d. settling ponds and lagoons;
e. landfarming of contaminated sludges and solids; and
f. injection well disposal of either used liquid solvent, or contaminated liquids.
2.
3.
4.
5.
6.
TABLE 1.13 Pure Compound Solubilities at ~ 20 DC for Selected

Chlorinated Organic Solvents and Corresponding Maximum Concentration
Limits (MCLs) for Drinking Water Set By EPA (1987).
1,2-Dichloroethane
I, 1,1-Trichloroethane
Methylene chloride
Chloroform
Tetrachloroethylene
Trichloroethylene
Solubility
(mglL)
MCL (EPA)
(mglL)
8,690
720
785
20,000
8,200
200
1,100
0.005
0.2
0.005
o.oi0.1"
0.005
0.005
SolubilitylMCL
ratio
2
4
2
2
8
4
2
x
x
x
x
x
x
x
106
103
105
106
104
104
105
"New York State Department of Environmental Conservation Guidelines for

Groundwater
to plumes of dissolved chlorinated solvents. The fact that LNAPLs do not penetrate much
below the water table, and the fact that the most mobile chemicals from LNAPLs are
prone to microbial degradation apparently account for the large difference in groundwater
impacts of chlorinated solvents and petroleum products.
Spills of chlorinated solvents in aquifers are, in general, exceptionally difficult to
clean up. At some sites, the plumes have reached water supply wells at concentrations
that are much above the MCLs for drinking water, thereby threatening public health
and valuable groundwater resources. Although some steps must be taken in such circumstances, given the typically long term nature of chlorinated solvent sites, it may
sometimes be more cost effective to simply prevent further release by isolating and/or
controlling the source zone rather than attempting a complete remediation. At other sites,
Sec. 1.6
15
TABLE 1.12 Physical, Chemical, and Biological Properties of the Chlorinated Solvent Compounds
Which Have Helped Lead to Extensive Groundwater Contamination by this Compound Class.
1. The high volatilities of the chlorinated solvents led to a false sense of security regarding how these
chemicals must be handled. Historically, it was believed that chlorinated solvent released to the unsaturated zone would easily volatilize to the atmosphere. Thus, when poured on dry ground, although a
chlorinated solvent may appear to be lost entirely to the atmosphere, some will be transported into the
subsurface by gaseous diffusion, by infiltration of contaminated water, and as a moving DNAPL phase.
And, once contamination reaches the saturated zone (including direct releases to the saturated zone),
high volatility is of little assistance in removing the solvents: transport across the capillary fringe can
be exceedingly slow (McCarthy and Johnson, 1992).
2. The high densities of the chlorinated solvents (1.2 to 1.7 g/crn') relative to that of water (I g/cm')
mean that if a sufficient volume of a typical chlorinated solvent is spilled, then liquid solvent may be
able to penetrate the water table. In the saturated zone, the unstable nature of the solvent flow causes
the solvent to form thin "fingers" which can lead to the collection of large amounts of solvent in one
or more "pools" on top of less permeable layers. Since a pool presents a very low cross section to
on-coming groundwater flow, absolute removal rates of dissolved solvent from the pool will usually be
very low (Johnson and Pankow, 1992).
3. The relatively low viscosities of the chlorinated solvents allow relatively rapid downward movement in
the subsurface. Chlorinated solvent mobility in the subsurface increases with increasing density/viscosity
ratios (Cohen and Mercer, 1993).
4. The low interfacial tension between a liquid chlorinated solvent phase and water allows a liquid
chlorinated solvent to enter easily into small fractures and pore spaces, facilitating deep penetration
into the subsurface. Low interfacial tension also contributes to the low retention capacities of soils for
chlorinated solvents.
5. The low absolute solubilities of the chlorinated solvents (typically on the order of hundreds of mg/L)
mean that when a significant quantity of such a compound is spilled on the ground surface, liquid
solvent will be able to migrate as a DNAPL phase in the subsurface, potentially accumulating as one
or more pools on the tops of low permeability layers. The low solubility will then permit such pools to
persist for decades to centuries (Johnson and Pankow, 1992).
6. The high relative solubilities of the chlorinated solvents mean that a solvent spill can cause groundwater
contamination at levels which are high relative to concentrations which appear harmful to human health.
7. The low partitioning to soil materials exhibited by the chlorinated solvents means that soil and rock
materials will bind these compounds only weakly. This applies to both the unsaturated and saturated
zones. Thus, sorption to soils will not significantly retard the movement of a chlorinated solvent, and
zones of contamination can grow essentially as quickly as the groundwater can move.
8. The low degradabilities of the chlorinated solvent compounds, either by biological means, or by
abiotic-chemical reactions, mean that subsurface lifetimes of these chemicals can be very long.
the plumes terminate at hydrologic boundaries such as rivers or lakes. In such cases, the
mass flux is transferred from the groundwater zone to surface waters where, because of
dilution and volatilization, usually little or no degradation of the water quality occurs.
Large remediation efforts may not always be warranted at sites where the plumes have
reached surface water bodies: unlike many other well-known industrial contaminants such
as PCBs and some pesticides, the chlorinated solvents do not accumulate in ecological
food chains.
16
Chap. 1
1.7 A HISTORY OF HOW WE BECAME AWARE THAT DENSE

CHLORINATED SOLVENTS CAN POSE SERIOUS THREATS TO
GROUNDWATER SYSTEMS
1.7.1 General
Widespread understanding in government, industry, and the professional groundwater
community of the exceptional propensity for chlorinated solvents to cause severe and
persistent contamination of groundwater was not achieved until most of today's problems
had already been created. It is now apparent that if the groundwater research community
had been quicker to recognize the importance of chlorinated solvents, many crucial
studies of solvent behavior in the subsurface could have been carried out decades earlier,
even though advanced methods for chemical analyses were not yet available. (All of
Schwille's work was conducted without such methods.) And, if awareness of and concern
for solvent hazards to groundwater had developed earlier, then practices for storage,
handling, transportation, and disposal of solvents could also have been improved more
quickly so that much of today's contamination of groundwater by chlorinated solvent
compounds could have been averted.
With the benefit of hindsight, it is apparent that the late awareness of the hazards
posed by chlorinated solvents to groundwater represents a failure of several segments of
modem society: governmental regulation, industrial practice, and environmental research
in the professional groundwater community. This failure makes it important to analyze
the history according to which our society came to realize that chlorinated solvents
must be handled with exceptional care. Such an analysis may: 1) provide insight that
will reduce the possibility for other environmental problems: and 2) result in research
and government regulations that prevent future problems in all compartments of the
environment.
We also note that the history according to which the threats posed by chlorinated
solvents to groundwater became recognized is a matter of current legal interest to the
site owners, insurance companies, local governments, lawyers, and consulting scientists
and engineers that are involved in disputes over matters of liability at contaminated
sites. For example, at a site where agreement has been reached as to who caused the
contamination problem, there is oftentimes considerable disagreement regarding when
and if the responsible party should have been aware of the threats posed by chlorinated
solvents.
As noted in Section 1.1.1, the discovery of the chlorinated solvents problem occurred in
two stages: I) the simple recognition of their prevalence as groundwater contaminants;
and 2) the recognition of the common presence of these compounds as difficult-toremediate, subsurface DNAPL source zones. Neither of these discoveries was anticipated
by the North American groundwater community.
Sec. 1.7
Becoming Aware of the Problem: A History
17

The evolution of knowledge regarding the threats posed by chlorinated solvents to groundwater has been reconstructed in Table 1.14 as four parallel timelines detailing events in
four different areas: Analytical Developments, Societal and Anecdotal Events, Literature
Reports, and Federal (U.S.) Government Actions.
Analytical Developments (column a) is a crucially important category for the history in Table 1.14 since it may have been unreasonable to have expected industrial or
consulting parties to analyze groundwater for chlorinated solvents before routine analytical methods were available. For example, since analytical methods were not readily
available in the early 1960s to detect chlorinated solvents at a level like 5 jLgfL, then
that situation makes it difficult to argue that a site owner should have been analyzing
groundwater samples at that time to ensure that contamination at that type of level had
not occurred.
Societal and Anecdotal Events (column b) is a significant category for the history
because it is important to document the evolution of environmental issues and perceptions
during the last 40 or so years.
Literature Reports (column c) is an important component of the history because
many reports of important developments regarding chlorinated solvents first appeared in
the scientific and engineering literature. Once those publications began to appear, then
one can argue that after some reasonable length of time, responsible parties should have
begun to take appropriate precautions to prevent further releases of contamination.
Federal (U.S.) Government Actions (column d) is included as a category in the
history because there are many examples of how a step taken by the federal government
was either an important regulatory climax of other events in the history, and/or an action that initiated additional developments in the other categories. For example, the Safe
Drinking Water Act of 1974 was the culmination of events in the other three categories,
and then itself caused further developments in the other categories. In the next section, a
reference to one of the elements in one of the timelines will be followed by a bold letter
which will indicate the column in which that element appears. For the sake of brevity,
not all of the elements in Table 1.14 will be discussed; the reader can examine the table
to fill in some of the details.
1.7.3 The First Stage: Discovering that Chlorinated Solvents are

in Groundwater
1.7.3.1 The 1950s to mid-1960s: the early years
Prior to the 1960s, the measurement of organic contamination in water was largely
limited to a few indicator parameters such as biological oxygen demand (BOD) and
bacterial concentration counts. It was generally assumed that groundwater quality was
largely unaffected by man. The subsurface was believed to be particularly capable of
absorbing "organic" contaminants: organic contaminants were thought at that time to
include mainly putrescible waste and sewage, which, indeed, are substantially renovated
18
Chap. 1
by natural processes in the subsurface. In one of the early textbooks on groundwater,

Thomas (1951) stated that organic wastes are "the easiest to eliminate {through the]
natural processes of separation, filtration, dilution, oxidation, and chemical reaction",
and that if "powerful influences of protection or purification were not at work in nature,
it would be almost impossible to find unpolluted {ground]water anywhere." Thus, in
the nascent field of contaminant hydrogeology, "organic" contaminants (including those
which could be detected by taste or smell) were receiving very little attention. The
contaminant that received the most attention in the early decades of the field was salt,
mostly from seawater intrusion.
Table 1.14 begins in 1953 (a) when the prominent journal Analytical Chemistry
published an article entitled "Mass Spectrometer Determination of Volatile Contaminants
in Water" by Melpolder et al. (1953). This article described a primitive purge and trap
method that involved the use of mass spectrometry to measure fLglL levels of petroleum
products in water. The authors stated: "The method may be extended to volatile compounds other than hydrocarbons, such as industrial solvents and cleaning fluids". On the
basis of this publication and others (e.g., Seto and Schultze, 1956 (a); Montgomery and
Conlon, 1967 (a)), it can be argued that if an analyst had set out with determination and
good laboratory resources to measure chlorinated solvents at tens of fLglL and higher in
groundwater, then suitable methods could probably have been derived from publications
available in the 1950s and 1960s.
As evidence in support of the above argument, we note that during the 1950s to
1960s time frame, chemists had undertaken to measure trace levels of certain organic
contaminants in water in which they were especially interested. For example, alkyl benzene sulfonate (ABS) detergents were being measured at this time in surface waters at
the ~ 10 fLglL level (Rosen et al., 1956 (a)). (Interest in the ABS detergents was due to
their increasingly ubiquitous presence as foaming agents in waste waters and in certain
receiving waters.) A concerted effort was also made in the late 1950s to measure fLglL
levels of certain organic pesticides like DDT and aldrin (Middleton and Lichtenberg,
1960 (a); see also the history provided by Rosen, 1976), and methods for the determination of selected pesticides were published in the water works literature in the 1960s
(Goodenkauf and Erdie, 1964 (a); Gutenmann and Lisk, 1964 (a)).
We can conclude that while anthropogenic organic compounds in general were not
of widespread interest during this time period, when there was a will to measure specific
organic compounds in water, then analytical methods suitable for use at the fLglL level
could be developed. And, although the chlorinated solvents were being produced in
quantities that were similar to those of the pesticides and the foaming, ABS detergents,
the inability to detect the chlorinated solvents by taste, smell, or sight caused them to be
simply overlooked as groundwater contaminants until much later in time.
1.7.3.2 The Mid-1960s to 1972: finding toxic organic

compounds in some drinking water supplies
On March 23, 1967 (b), Dr. Leslie L. Glasgow, Chairman of the Louisiana Stream
Control Commission, wrote to the Commissioner of the U.S. Federal Water Pollution
Sec. 1.7
19
Control Administration t requesting technical assistance in determining the extent to which

organic chemicals were affecting the quality of New Orleans drinking water. The reason for the request was that there were many complaints of "chemical" and "oily"
flavors in the Mississippi River-derived water. These complaints resulted in a concerted
effort to identify and measure trace organics in water using gas chromatography coupled
with mass spectrometry (GCIMS). Most of this work was carried out at the Southeast
Water Research Laboratory in Athens, Georgia in the late 1960s (d) (Keith, 1974)+.
EPA, formed in October of 1970 (d) by President Nixon, continued the study. Using
an analytical method using adsorption on carbon followed by solvent extraction with
chloroform, the study was completed in April, 1972 (d), confirming the presence of
a wide variety of "extractable" organic compounds in New Orleans drinking water.
EPA (1972) concluded that the problem was due to industrial wastes discharged to
the river, and that chlorination of the drinking water may also have been playing a
role. As a direct result of these findings, Congress enacted the Clean Water Act (CWA)
on October 18 of 1972 (d) to regulate waste discharges into all navigable waters of
the U.S. The CWA did not deal in a significant way with groundwater, and ironically,
had an adverse effect on groundwater because it directed waste discharges from surface waters to the land, where they no doubt degraded groundwater quality in some
locations.
Since the method that EPA (1972) (d) used to analyze the New Orleans samples
used recovery with chloroform, no other volatile, chlorinated solvents that might have
been present could be easily determined at that time in the face of the overwhelming
amount of that chlorinated solvent. Therefore, there was as yet little indication that
volatile organic compounds (VOCs) might be a problem in drinking waters obtained
from surface water supplies, and certainly no indication to the hydrogeology, industrial,
or regulatory communities that chlorinated solvent VOCs might be prevalent groundwater
contaminants. Indeed, the Manufacturing Chemists' Association was stating in "Chemical
Safety Data Sheets" for TCE, PCE, and 1,1,1-TCA distributed as late as 1968 (b) that for
disposal, these solvents"... may be poured on dry sand, earth, or ashes . . . and allowed to
evaporate into the atmosphere." The disposal method suggested in a similar data sheet for
TCE dated 1971 (b) and distributed by a large manufacturer stated that one should "Bury
[the TCE] away from water supply or allow solvent to evaporate to atmosphere at a safe
distance from inhabited buildings?", Despite the widespread ignorance of the effects that
tThe Federal Water Pollution Control Administration was one of the federal agencies that formed the
basis of, and was superceded by the EPA.
'By 1970, computerized GCIMS equipment had become commercially available, greatly increasing the
speed of organic pollutant analysis (Keith, 1974).
Prior, related comments on this subject were made in the Handbook of Vapor Degreasing (American
Society for Testing and Materials (ASTM), 1962) This handbook advised that for the disposal of spent solvents,
"If there are no local regulations forbidding it, the sludge may be poured on dry ground at a safe distance from
buildings and allowed to evaporate. If the solvent is free-flowing and can soak into the ground before the solvent
evaporates, it may be poured into shallow containers to permit the solvent to evaporate before dumping". We
note in this context that in the U.S. in the 1960s, no local regulations of which we are aware forbade ground
disposal of solvent sludge.
20
Chap. 1
such chlorinated solvent disposal practices would have on groundwater quality, as seen
by various entries in column b in Table 1.14, concern regarding the safety of groundwater
supplies was nevertheless beginning to grow in the late 1960s and early 1970s.
1.7.3.3 1972 to 1975: finding the first chlorinated VOCs in

drinking water; the trihalomethanes (THMs)
With interest increasing during the early 1970s in the matter of organic compounds in
drinking water, Rook (1972) (a) adapted a method from the field of beverage flavor
analysis to determine VOCs in water. In this method, a large (10 L) sample was heated
to 60C, and the "headspace" gas above the liquid was directed over silica gel to concentrate the VOCs. After concentration, the VOCs were determined by gas chromatography
(GC). In the same article, Rook (1972) (c) went on to discuss the fact that analyses
at a Rotterdam drinking water plant indicated that trihalomethane (TUM) compounds
(e.g., chloroform) were somehow being formed during the chlorination of the drinking
water. Then, in the summer of 1974, Rook (1974) (c) made the connection that it was
reactions between aqueous chlorine and natural humic materials that were responsible
for the formation of the haloforms.
In December of 1974, using an approach for VOC analysis that was similar to
that devised by Mieure and Dietrich (1973) (a), Bellar and Lichtenberg (1974) (a) developed the current form of the now very-well-known purge and trap method. Purge and
trap was used by Bellar et al. (1974) (c) in conjunction with GC/MS in a study which
revealed that haloform compounds were also present in U.S. drinking waters. Consequently, in December of 1974 (d), EPA began the National Organics Reconnaissance
Study (NORS) which examined surface water and groundwater supplies used to provide
drinking water for 80 U.S. cities. The water samples were analyzed for several THMs,
1,2-dichloroethylene, and carbon tetrachloride (Symons et al., 1975). In May of 1975 (d),
results of the NORS project revealed the essentially ubiquitous presence of haloforms
in chlorinated drinking water in the U.S.
During the same time period that the issue of haloforms in drinking water was
being investigated in Europe and in the U.S., the matter of organic contaminants in New
Orleans drinking water came to a climax when EPA (1974) (d) reported to the public in
November of 1974 that at least 66 trace organic chemicals (including four chlorinated
solvents: carbon tetrachloride, 1,1,1-TCA, TCE, and PCE) had been identified in Mississippi River water. Almost immediately thereafter, the Safe Drinking Water Act was
signed into law on December 16, 1974 (d) by President Ford. This act dealt primarily
with the protection of surface waters, though it also brought the first Federal action to
protect groundwater resources inasmuch as it required the States to regulate all direct
injections of wastes to the subsurface.
We see then that by the end of 1975, concern about organic contaminants in water was focussed primarily on: a) the new issue of contamination of drinking water by
THMs as produced during chlorination; and b) problems with contamination of surface water supplies, primarily by industrial pollutants. Concern about contamination of
Sec. 1.7
21
groundwater by organic chemicals, however, was as yet very limited. Indeed, in a Chemical Safety Data Sheet that was dated 1973 (b) and distributed over the next several
years, the same large chemical manufacturer of chlorinated solvents discussed in the
previous section stated for the disposal of waste methylene chloride that "In some cases,
it can be transported to an area where it can be placed on the ground and allowed to
evaporate'il Thus, since direct pouring of solvents on the ground was not ruled out
in this recommendation, the likely impact on groundwater of chlorinated solvents had
not yet been clearly recognized. And, although some specialists were beginning to fear
that groundwater supplies were in danger (e.g., see the opinions of Geraghty and Miller
published in June of 1973 (b) as well as the discussion in ES&T in December of 1973
(c) of a national policy on subsurface management), there was as yet no clear evidence
that groundwater contamination was actually already a problem.
1.7.3.4 1976 to 1979: widespread testing for THMs in drinking

water reveals the presence of other VOCs: coming
to the conclusion that many important aquifers are
contaminated with chlorinated solvents
As noted above, the NORS project demonstrated that chloroform and other THMs were
very common in U.S. drinking waters, and that the chlorination process was responsible
for that presence. However, NORS revealed relatively little about VOCs other than the
THMs. Therefore, in order to: I) investigate the relationships between the nature of a
water supply and the resulting degree of THM formation upon chlorination; and to 2)
expand the list of VOCs of interest beyond the THMs, in March of 1976 (d), EPA initiated the National Organics Monitoring Study (NOMS). In NOMS, 113 different public
drinking water supplies representing various types of raw water sources and drinking
water treatment processes were selected for examination. Of these supplies, 21 depended
primarily on groundwater. Several important chlorinated solvent compounds were included as target analytes. The detection limits by the purge and trap method used (with
GCIMS) ranged from 0.01 to 0.3 JLg/L.
The first published NOMS results appeared in 1977 (c) in a book chapter by Brass
et al. (1977). For the "Phase II" portion of the NOMS sampling (May to July of 1976),
Brass et al. (1977) noted that TCE was found in 28 of the 113 drinking water systems.
There was no discussion of the implications of this EPA finding either in Brass et al.
(1977), or in the final EPA report for NOMS (EPA, 1977b) even though some of these
systems were supplied by groundwater. This meant that it would be necessary for other
studies of THMs in drinking water to rediscover the presence of chlorinated solvents in
important groundwater supplies. Fortunately, the publicity surrounding the THM problem
generated by the study of Bellar et al. (1974), by NORS, and by NOMS led directly to a
large number of local studies of THMs in drinking water. The fact that the chlorinated
tEven in 1976, in their new edition of Handbook of Vapor Degreasing, ASTM (1976) used wording
regarding spent solvent disposal which was identical to that given in the preceding footnote regarding the 1962
edition.
22
Chap. 1
solvents are VOCs just as are the THMs meant that chlorinated solvent compounds would
be found in many of these local studies.
Facilitating the local studies of THMs in drinking water was the important fact
that by 1977, purge and trap had become a tested and proven method for determining
VOCs at the low ppb level. Also, during this same general time period, rapid improvements were being made in the application of GC/MS in environmental organic analysis
(Heller et al., 1975 (a)). In addition, micro-solvent extraction with ~5 mL of pentane
and ~35 mL of water (Henderson et al., 1976, (a became available as an alternative
to purge and trap with GC/MS for trace levels of chlorinated compounds in water samples. The main advantage of micro-solvent extraction in analyses for aqueous chlorinated
compounds is that the extract can be analyzed on a GC fitted with an electron capture
detector (ECD): a GCIECD allows the sensitive and rather specific detection of many
chlorinated compounds at very low levels, without requiring the complexity and cost of
a GC/MS.
The ready availability of suitable analytical methods for VOCs together with
the motivation to look for TOMs in local water supplies quickly led to a series of reports of local groundwater contamination by chlorinated solvents. As examples, consider
the following occurrences. In 1976 (b), New York State revealed that aquifers on Long
Island were severely contaminated with TCE that had been used by many homeowners
as a septic tank cleaner. In 1977 (b), 16 private wells in Gray, Maine were found to be
contaminated with various chlorinated solvents. In 1978 (b), New York State officials
reported that a search for THMs in drinking water had led to the discovery of a well
field that was highly contaminated with TCE: the well field was closed. In 1978 (b), officials of the Nassau-Suffolk (Long Island, New York) Regional Planning Board reported
widespread contamination by several chlorinated solvents: up to 375 fJglL of PCE, up
to 310 fJglL of I,l,l-TCA, and up to 300 fJg/L of TCE were found in 36 wells; the
wells were quickly closed (Koppleman, 1978). In May of 1978 (b), 80% of the drinking
water of Bedford, Massachusetts was found to be contaminated with TCE at levels as
high as 500 fJglL. Towards the end of the 1970s, hundreds of water supply wells had
been closed in the U.S. due to contamination by chlorinated solvent compounds. The increasing evidence of contamination of groundwater was reviewed during this time frame
by Pojasek (1977) (c).
As the evidence mounted that many groundwater supplies were contaminated
with chlorinated solvents, some local reconnaissance studies were begun which had the
goal of determining the extent of the contamination. For example, in 1978 (c), New Jersey
began testing 500 public and private wells located near industrial sites for 50 organic
compounds; a number of these compounds were chlorinated solvents. This reconnaissance
for chlorinated solvents in groundwater was very analogous to EPA's use of NORS
to determine the extent of UJ.e nation's THM problem in chlorinated drinking water
supplies.
Drinking water regulations for various VOCs began to be promulgated in the U.S.
in the late 1970s. New York State (1978) (b) published standards which included limits
for three chlorinated solvent compounds; these are the first drinking water limits for
chlorinated solvents of which we are aware. On November 20, 1979 (d) EPA's final
Sec. 1.7
23
THM regulations stated that total THMs in drinking water should not exceed 100 {LglL,
foreshadowing the regulations that would later be put in place for chlorinated solvents
like TCE (see 1987 (dj), In 1979 (d), EPA also promulgated their list of "priority pollutants" (see Keith and Telliard, 1979) which were to be determined in waste water
discharges. The priority pollutant list includes all of the common chlorinated solvents,
and has become a commonly-used list of target analytes in many environmental water
analyses. In this same context, we note that EPA's 600 series methods for VOCs were
also formally issued in 1979 (a). Since these methods were designed to be sensitive at
the {LglL level, additional advances in analytical chemistry had virtually no further effect
on the history of interest here, and so entries in column a in Table 1.14 end at this point
in time.
We see then that by the end of 1979, a reasonable amount of hard evidence had
accumulated indicating that groundwater contamination by solvent chemicals might be
a significant problem, especially in densely populated and/or industrialized areas. However, as has been typical in the discovery of other environmental problems (e.g., stratospheric ozone depletion by chlorofluorocarbons), the general recognition of the problem of solvents in groundwater was not to take place until the weight of the evidence
became overwhelming. Thus, the same major producer of chlorinated solvents mentioned above for 1973 was still saying in 1979 (b) that, for the disposal of methylene
chloride, one could: "Send solvent to a reclaimer. In some cases, small amounts can
be transported to an area where it can be placed on the ground and allowed to evaporate safely if local, state, and federal regulations permit". t There is also evidence
that throughout the 1970s, chlorinated solvents were being disposed of at U.S. military
bases by pouring them on the ground (New York Times, November 2, 1980; Schroeder,
1985; Schaumburg, 1990). Thus, since: 1) disposal on the ground was still viewed by
many as a viable disposal option; and 2) groundwater protection was as yet largely
not provided at either the state or federal level, it would not be until sometime in the
1980s that the dangers posed by chlorinated solvents to groundwater would receive
widespread recognition in government, industry, and in the professional groundwater
community.
1.7.3.5 1980 to 1981: The evidence that chlorinated solvents
are widespread groundwater contaminants becomes

overwhelming
The period 1980 to 1981 was characterized by a virtual explosion of reports which
indicated that many important groundwater supplies were contaminated by chlorinated
solvents. As an important example, the January 1980 (c) issue of the widely-read journal
Environmental Science and Technology gave a cover-story account of the facts that: 1)
groundwater contamination (including contamination by chlorinated organic compounds)
had become a matter of very serious concern; and that 2) tight federal and state rules
tWe note that relative to the 1973 recommendations. the qualifier concerning "local. state, and federal
regulations" was now included, and so apparently the company's concern about chemicals such as methylene
chloride was increasing.
24
Chap. 1
requiring groundwater monitoring and protection would soon be coming. In the same
month (b), public health officials for the State of California closed 39 TCE-contaminated
public wells that had been providing water to 400,000 people in 13 cities in the San
Gabriel Valley. 1980 (b) was also the year of discovery of the now well-known, large
scale contamination of groundwater in California's Santa Clara "Silicon" Valley. By May
of 1980 (b), agencies in 18 states had measured numerous VOCs, most commonly TCE,
in 2894 wells (see CEQ, 1981).
As of early 1980, although reasons to be concerned with groundwater contamination
by chlorinated solvents were accumulating at a rapid pace, other than the rules regarding
the injections of waste imposed by the Safe Drinking Water Act (see above), there were
very few regulatory protections in place for groundwater resources. Indeed, in May of
1980 (b) in a special, red-paged "Regulatory Alert", ES&T (1980) reported that: "Approximately 50% of the [nation's] population depends on groundwater for its drinking water
supply. Yet, unlike surface water or air, there is no specific federal statute to protect it
from pollution." In the same month, the Environment Reporter (l980a) (b) stated that
few states bad acted as yet to declare authority to protect groundwater supplies, and
that most did not even recognize the magnitude of the problem. In August of 1980
(b), Environment Reporter (l980b) stated that existing federal regulations were still
being viewed by some as so weak that new laws would be needed to provide effective
groundwater protection.
In terms of EPA's role in the evolving circumstances, it was in May of 1980 (d)
that EPA Administrator Costle, speaking on NBC's Meet the Press (see Josephson, 1980),
identified groundwater protection as a principle EPA priority. In the same month (d), in
an effort to deal with the deficiencies in groundwater legislation that were discussed in
the preceding paragraph, EPA promulgated the RCRA regulations that required large scale
groundwater monitoring at disposal sites. By the mid-1980s, groundwater from tens of
thousands of monitoring wells at industrial sites had been analyzed for chlorinated solvent
compounds and other VOCs: in an early survey of data from 183 hazardous waste sites
in the U.S., four of the top fivel , and 10 of the top 20 most-frequently identified organic
chemical contaminants were chlorinated solvents (Plumb and Pitchford, 1985)t. In June
of 1980 (d), EPA (1980a) conducted a planning workshop on the groundwater contamination problem, and in November of 1980 (d), EPA (1980b) announced their resulting
National Ground Water Protection Strategy, with publication of the strategy to occur
in the Federal Register in January 1981. In December of 1980 (d), U.S. Congress passed
the EPA-administered Superfund (i.e., CERCLA) Program to deal with "uncontrolled"
hazardous waste sites.
Naturally, a great deal of publicity surrounded the events described in the preceding paragraph. It is not surprising therefore that in 1980, just one year after the
t The four were TCE, PCE, dichloromethane (DCM), and I, I-dichloroethane.
tMore recently, EPA's 1993 Superfund Chemical Analysis Result Report indicated that seven of the 20
most frequently detected organic contaminants at Superfund sites are chlorinated solvents. The Comprehensive
Environmental Response, Compensation and Liability Information System (CERCUS) lists over 32,000 sites
in the U.S., and EPA has estimated that over one third of these are contaminated with solvents (EPA, 1991).
Sec. 1.7
25
release of its prior version of its Chemical Safety Data Sheet on methylene chloride,
the same major chemical company mentioned above for 1973 and 1979 tightened considerably its recommendations for disposal of this chemical. In particular, in 1980 (b),
the data sheet on methylene chloride was revised to state: "... Dumping into sewers,
on the ground, or into any body of water is strongly discouraged, and may be illegal."
The arrival of January 1981 (b) brought a landmark report from the President's
Council on Environmental Quality (CEQ). Entitled Contamination of Groundwater by
Organic Chemicals, this report detailed the avalanche of reports on groundwater contamination in the U.S. Identifying chlorinated solvents as the most common type of
contaminants, CEQ (1981) noted that "These incidents involve discoveries - sometimes
completely accidental [e.g., looking for THMs, but finding other compounds as well] of high concentrations of synthetic organic chemicals in a community or private drinking
water well. It is now evident that hundreds of drinking water wells affecting millions of
people have been closed because of contamination by toxic organic chemicals . . . . No one
can say for sure how long the contamination has existed, ... ". "... the EPA regions report
serious contamination of drinking water wells in 34 states, and there are indications that
the true number is at least 40". The release of this study was immediately reported upon
in the February 1981 issue of ES&T (1981b) (c), and again in the July issue (c). With
respect to what was beginning to occur at this time on the remediation front, we note
that the January 1981 issue of ES&T (1981a) (c) reported as a news item that Recon
Systems, Inc. (Somerville, NJ) had built and was operating an air stripping tower to
remove 1,1,1-TCA from groundwater that was being pumped at 100,000 gpd from an
aquifer in the Northeast.
ES&T was not the only journal reporting with urgency during early 1981 on the
problem of chlorinated solvents in groundwater systems. For example, in April of 1981
in the Journal of the American Water Works Association, Petura (1981) (c) wrote an article entitled "Trichloroethylene and methyl chloroform [I,I,l-TCA] in groundwater: A
problem assessment". The article stated that groundwater pollution was a "problem of
immense concern", with TCE and 1,1,1-TCA having been found in groundwater in numerous states including Pennsylvania, Delaware, New Hampshire, and California. "Hundreds
of wells, many in totally unrelated aquifers" were described as being contaminated with
these two compounds. At about the same time, in a book entitled Quality of Groundwater, Giger and Schaffner (1981) (c) and Zoeteman et al. (1981) (c) documented the
presence of chlorinated hydrocarbons (including TCE, chloroform, and carbon tetrachloride) in groundwater at various sites in Switzerland and The Netherlands, respectively.
Zoeteman et al. (1981) found TCE in 67% of the wells tested, carbon tetrachloride
in 43%, and PCE in 19%. 1981 (c) also brought Schwille's first mainstream, English
language journal article on the behavior of NAPLs in subsurface systems (Schwille,
1981).
Perhaps the clearest indication of the widespread nature of groundwater contamination by chlorinated solvents in the Northeast appeared in December of 1981 (c) in
a study published in ES&T by Page (1981). Using data for 1076 wells in New Jersey, Page (1981) found that NJ groundwaters were at least as contaminated as
26
Chap. 1
NJ surface waters. A total of 17 halogenated solvent compounds were found in the

groundwaters. For TCE, of 669 wells sampled, 388 were found to be contaminated,
with 635 p.,glL being the highest concentration. For 1,1,1-TCA, of 1071 wells sampled,
835 were contaminated, with 608 p.,glL being the highest concentration. For PER, of
421 wells sampled, 179 were contaminated, with 91 p.,glL being the highest concentration.
By mid-1981, EPA was beginning to recognize the scale of the problem posed by
chlorinated solvents. In particular, on June 9, 1981 (d), in a presentation at a national
meeting of the American Water Works Association, EPA employee Herbert J. Brass
discussed results of the nationwide Community Water Supply Survey (CWSS) (Brass
et al., 1981). Based on samples from 106 surface water sources and 330 groundwater
sources, Brass et al. (1981) found clear evidence for the presence of compounds like
PCE, TCE, and 1,1,1-TCA in many groundwater supplies. These three compounds were
found in about 5% of the groundwater sources tested, with maximum concentrations of
30,650, and 210 p.,glL. Of the 330 groundwater systems tested, 50 (15%) were found to
be contaminated with some YOCs.
Based on the above evidence, we believe that it is clear that it is the end of 1981
that provides the milestone for the end of the period of general ignorance of the fact that
poor solvent handling practices and careless disposal can cause severe contamination of
groundwater. Nevertheless, we know that contamination continued to take place beyond
this date. For example, in their January 1981 (b) report, CEQ (1981) discussed the fact
that septic tank cleaning fluids (most of which contained TCE, benzene, and/or methylene
chloride) could still be purchased in most of the U.S. during 1981, with the 1979 usage
estimated at 400,000 gallons on Long Island alone. Also, some of the common chlorinated
solvents were still being used as "inactive" ingredients in pesticide formulations; these
"inactive" ingredients often comprised most of the mass of such products. Regulations
forbidding their use in this manner were not promulgated until the late 1980s (d) (EPA,
1987b).
What is notable about the discovery of the prevalence of dissolved-phase chlorinated solvents in groundwater is that it occurred almost entirely outside the realm of the
groundwater science and engineering community. Nearly all of the surveys that revealed
such contamination were conducted by state and federal agencies using samples pumped
from water supply wells. We also note that in the late I 970s through 1981, there was essentially no discussion of dissolved chlorinated solvent contamination of groundwater in
the major journals dedicated to groundwater science and engineering, e.g., Ground Water
and Water Resources Research; hydrology professionals were learning of the problem
from developments taking place outside of their field.
1.7.4 The Second Stage: Discovering the DNAPl Nature of

Liquid Chlorinated Solvents
1.7.4.1 1982 to the late 1980s
From 1982 onward, new reports of groundwater contamination by chlorinated solvents
continued to appear (see columns band c in Table 1.14). The issue of solvents in ground-
Sec. 1.7
27
water had by then evolved from the stage wherein the problem was being discovered
and understood, to a stage wherein the goal was to have contamination contained, remediated, and prevented. Thus, in 1982 (b), the San Francisco Bay Regional Water
Quality Control Board (RWQCB) launched a program to inventory underground storage systems in the industrialized areas under its jurisdiction. The resulting survey by
mail to 2644 companies in the key groundwater basins ultimately lead to the publication
of RWQCB's "2051" report (Industry Clean Water Task Force, 1985) which indicated
that a significant amount of the contamination in the valley was due to high-tech industrial operations. As of September of 1985, the RWQCB had listed 75 sites within their
jurisdiction at which groundwater contamination was: 1) under investigation; 2) in the
process of being defined; 3) being cleaned up; or 4) had been cleaned up. On the Federal
scene, in July of 1982, EPA (1982) (d) issued new rules governing hazardous wastes in
landfills, surface impoundments, and other repositories. In 1986 (d), the Hazardous and
Solid Waste Amendments to RCRA were passed mandating monitoring of the groundwater and vadose zones at many underground storage tank locations. On July 8, 1987
(d) a final drinking water standard value for TCE of 5 jJglL was published by EPA
(1987a).
Although they had become aware of the general presence of dissolved chlorinated solvents in groundwater by about 1981, for perhaps five to seven years afterward,
groundwater scientists and engineers in North America and in Europe remained almost
entirely unaware of two facts: 1) that DNAPL-phase solvent could be present in large
amounts in the subsurface at sites contaminated with chlorinated solvents; and 2) that
such a presence posed many difficult problems in remediating such sites. In the
early 1980s, the pioneering laboratory experiments of Schwille in these topic areas were
largely unknown to these scientists and engineers (most of the European groundwater community was also unaware of this work). Although Schwille had published two
English-language papers in the early 1980s (Schwille, 1981; Schwille, 1984b) (c), neither had been placed in the widely-read literature, and Schwilles (1984a) (c) summary
report on the behavior of chlorinated solvents as DNAPLs in subsurface systems had
been published in German. Thus, although the general presence of chlorinated solvents
in groundwater had become recognized, it was not yet realized that long term source
zones of DNAPL-phase chlorinated solvent were present below the water table at many
contamination sites. Without knowledge of Schwille's work, North American groundwater scientists and engineers began now to rediscover what Schwille had concluded a
decade earlier.
An understanding of how DNAPLs behave in the subsurface evolved in North
America within a few groups of groundwater scientists and engineers who had each
encountered DNAPLs at depth in the subsurface in the late 1970s. As was discussed
in Section 1.1.2, the term "NAPL" was first coined in 1981 (b) during studies of a
hazardous waste landfill in Niagara Falls, New York. At this site, a denser-than-water,
organic liquid was discovered when it was bailed from wells installed in the soil and
bedrock underlying the site. The liquid was a complex mixture of halogenated benzenes,
chlorinated solvents, and other halogenated hydrocarbons. Similar organic liquids were
discovered at about the same time at two other sites in the Niagara area, the Hyde Park
28
Chap. 1
Landfill and Love Canal, and at a handful of unrelated sites elsewhere in North America.
Thereafter, it soon became clear that NAPLs were an important source of the groundwater
contamination at these sites. Unfortunately, nearly all of the information regarding NAPL
and DNAPL behavior at those sites was the subject of litigation, usually in the context
of RCRA and Superfund issues, and so that information did not become available until
the late 1980s.
The development of conceptual models regarding how DNAPLs behave in the
subsurface began in North America with conference proceedings and publications of
limited circulation (e.g., Feenstra, 1984; Cherry, 1984a and 1984b; Miller, 1984) (c).
Although the number of mainstream, English-language publications reporting the presence of dissolved chlorinated solvents in groundwater had exploded in the 1980-1981
time frame, the first widely-read publications to discuss DNAPL behavior were those
of Villaume (1985) and Mackay et al. (1985) (c). Both of these publications contained schematic diagrams of plumes emanating from DNAPL source zones in the
subsurface, both as residual zones and as pools. However, neither of these publications emphasized the far-reaching implications of subsurface DNAPLs for groundwater
quality.
At the time that the papers discussed in the preceding paragraph were being published,
tremendous efforts and finances were being directed towards remediation efforts at countless Superfund and RCRA sites. Unfortunately, many of these efforts were largely unsuccessful. These efforts failed because in their design, the likely presence and nature of
DNAPL source zones had not been considered (Bredehoft, 1992).
Schwille's work finally received wide North American recognition when an English translation of his comprehensive monograph was published in 1988 (c) (Schwille,
1988). The first textbook to include the DNAPL conceptual model, at least in a cursory way, was Fetter (1988) (c). The first substantive textbook treatment of DNAPLs
appeared in Domenico and Schwartz (1990) (c). These and other publications brought
at least a general understanding of how DNAPL solvents behave in the subsurface into
full circulation within the groundwater science and engineering community. However,
even into the early 1990s, only a relatively small fraction of the groundwater and environmental science community was aware of how DNAPL solvents could behave in the
subsurface.
The first indication that EPA, the agency regulating the Superfund investigations,
recognized some aspects of the DNAPL issue consists of a cursory reference in their
document entitled Guidance on Remedial Actions for Contaminated Ground Water at
Superfund Sites (d) (EPA, 1988). Evidence that EPA recognized the full implications
of DNAPL behavior for Superfund investigations was not apparent until the publication of a Ground Water Issue Paper on DNAPLs (d) (Hurling and Weaver, 1991), and
their sponsorship of a two day DNAPL workshop at the Kerr Laboratory in Ada, Oklahoma in April, 1991 (d). A comprehensive monograph entitled DNAPL Site Investigation
Sec. 1.8
Conclusions
29
sponsored by EPA was published in 1993 (d) (Cohen and Mercer, 1993). 1993 also
brought the EPA report entitled "Evaluation of Likelihood of DNAPL Presence at NPL
Sites - National Results (EPA, 1993). Thus, although EPA had led the way in identifying
and determining trace organic contaminants in groundwater in the 1970s, official recognition by EPA of the important ways in which DNAPLs affect groundwater contamination
and remediation came more than a decade after the presence of dissolved chlorinated solvents had been generally recognized around 1981. We conclude from this that a strong
science program within EPA is still needed to support its official and regulatory actions.
1.8 CONCLUSIONS
Major production and use of dense chlorinated solvents began in the 1950s and 1960s.
Given the properties of these compounds, in hindsight, it is not surprising that these
compounds are now problematic groundwater contaminants. Simply put, it was the extensive and widespread industrial usage of these compounds coupled with some naive
attitudes about how they could be handled that led to most of today's problems with
these compounds.
Recognition of the importance of chlorinated solvents in groundwater arrived very
late in the research community. Indeed, after about 30 years of unwise practices, it was
in the late 1981 time frame that our industrialized society came to a wide recognition
of the environmental hazards posed by this class of chemicals. The special DNAPL
nature of chlorinated solvents in the subsurface became recognized in the late 1980s/early
1990s; other than Schwille's studies of the late 1970s and early 1980s, there had been
little substantial research directed specifically at the behavior of chlorinated solvents in
groundwater.
The ability of chlorinated solvents in DNAPL form to persist in the subsurface for
long periods of time, even in the face of aggressive remediation efforts, is the most vexing
aspect of groundwater contamination by such compounds. The length of time required
to dissolve a typical accumulation of solvent on top of an aquitard can be hundreds of
years under natural conditions; significant decreases in the time for complete dissolution
for such an accumulation will in general require aggressive treatment efforts (Johnson
and Pankow, 1992).
Much of the information presented in the chapters of this book is the result of
recent study. Some of the topics included are the subjects of intensive on-going research.
Our understanding of the movement and fate of chlorinated solvents in groundwater has
now advanced sufficiently that most of the major aspects of how chlorinated solvents
move in the subsurface after spills are reasonably well understood. In contrast, much less
is known about the behavior of chlorinated solvents in the subsurface when subjected
to remediation efforts. The remaining chapters in this book will seek to provide a clear
understanding of the pertinent properties of the chlorinated solvents so that contamination
assessment, control, and remediation (when possible) can take place at sites of interest
throughout the world.
30
TABLE 1.14
Timelines.
Chap. 1
A History of the Problem of Chlorinated Solvents in Groundwater in Four Parallel
a. Analytical
developments
b. Societal and
anecdotal events
1953: The journal Analytical

Chemistry publishes the article
"Mass Spectrometer Determination
of Volatile Contaminants in Water"
by Melpolder et al. (1953).
1956: The journal Analytical
Chemistry publishes the article
"Determination of Trichloroethylene,
Trichloroacetic Acid, and
Trichloroethenol in Urine" by Seto
and Schultze (1956).
1964: Methods for the determination
of various pesticides are published
in the water works literature by
Goodenkauf and Erdie (1964) and
Gutenmann and Lisk (1964).
1967: The journal Water Pollution
Control publishes the article "The
Detection of Chlorinated Solvents
in Sewage Sludge" by Montgomery
and Conlon (1967).
1967, March 23: Or. Leslie L. Glasgow (Chairman, Louisiana

Stream Control Commision) asks the Commissioner of the U.S.
Federal Water Pollution Control Administration for technical
assistance in determining the extent to which organic chemicals
are affecting water quality in the Mississippi River. There have
been many complaints that the drinking water in New Orleans
has "chemical" and "oily" flavors.
1967 and 1968: In Chemical Safety Data Sheets SO-14, SO-24,
SO-90, dated 1956, 1948, and 1965, respectively, but still
being distributed, the Manufacturing Chemists Assoc. states that
"Residue[s] [of TCE, PCE, and i,i,i-TCA] may he poured on
dry sand, earth, or ashes . . . and allowed to evaporate into the
atmosphere." The Chemical Hazards Bulletin of the American
Insurance Association (1967) contains similar wording.
1969: Charles F. Luce, chairman of the National Water
Commission, tells the 5th Annual Meeting of the Ground Water
Resources institute at Gov. Winthrop Rockefeller's Winrock Farm
(Morrilton, Ark.) that the quality of ground water is in grave
danger. Carl F. Klein, Asst. Secretary of Interior, reports that
"Most of the same types of pollution that affect surface water
will eventually affect groundwater with greater impact and longer
lasting harm."
Sec. 1.8
31
Conclusions
c. Literature
reports
d. Federal governmental
actions
1956: Rosen et al. (1956) report the

identification of anionic synthetic
detergents in surface waters.
1960: Middleton and Lichtenberg
(1960) report the presence of
pesticides like DDT and aldrin in
certain of the nation's rivers.
Late 1960s: Most of the work to identify and measure the organic
contaminants in the Mississippi River water is conducted at the
federal Southeast Water Research Laboratory in Athens, Georgia
(see Keith, 1974).
1970, October: EPA is established by President Nixon.

1971: A patent is given to Warner
Co. (Philadelphia, Penn.) for an
impermeable membrane to line the
base of a landfiII.
1971, February 22: William Ruckelshaus is sworn in as first

Administrator of EPA.
32
TABLE 1.14
Chap. 1
(continued)
a. Analytical
developments
b. Societal and
anecdotal events
1971: A large manufacturer states in a chemical data handling
sheet for TCE that one should "Bury [the TCE] away from water
supply or allow solvent to evaporate to atmosphere at a safe
distance from inhabited buildings."
1972, Fall: Rook (1972) adapts a

beverage analysis method for use
in determining VOCs in drinking
water. The headspace above 10 L
of 60C water is passed through
silica gel. Compounds collected on
the silica gel are analyzed by gas
chromatography with detection by
mass spectrometry (i.e.. by GC/MS).
This a forerunner of the P&T
method later developed by EPA.
1972, March: ES&T (1972) reports that: "After many years of

apathy and indifference. private citizens, industry, and government
are now advocating national policies for groundwater pollution
and conservation." Although groundwater has always been
considered insulated against contamination, ES&T notes that
groundwater contamination is increasing.
1972: The revised Chemical Hazards Bulletin of the American
Insurance Association (1972) contains the same wording given
above regarding disposal, but does add a qualifier regarding
complying with new federal regulations.
1973, June: ES&T (1973) reports that "Geraghty and Miller,
a firm of consulting groundwater specialists, is performing a
14 month study of groundwater contamination problems in
11 northeastern states which account for 26% of the nation' s
population. Project leader David Miller says, 'No one really
knows how severe the groundwater contamination problem is in
the Northeast and one of the goals of the study is to put this
problem in its proper perspective. ...
1973, November: Mieure and

Dietrich (1973) publish a method
for determining VOCs in which the
headspace over a sample of water is
passed through a trap (Tenax-GC).
The contents of the trap are
subsequently thermally desorbed to
a gas chromatography column for
analysis. This is a forerunner of the
P&T method developed later by
EPA.
1973: A major chemical company recommends for disposal that

users of methylene chloride can "Send solvent to a reclaimer.
In some cases, it can be transported to an area where it can be
placed on the ground and allowed to evaporate."
1974, July: State of Louisiana and City of New Orleans ask EPA
Region VI to undertake another analytical survey of the organic
chemicals present in the finished water of three drinking water
treatment plants in New Orleans.
Sec. 1.8
33
Conclusions
c. Literature
reports
1972, Fall: Rook (1972) reports

that haloform compounds are
somehow formed during the
chlorination of Rotterdam drinking
water.
1972, November: Kleopfer and

Fairless (1972) report the presence
of numerous, industrially-related
organic compounds in the municipal
water obtained for Evansville,
Indiana from the Ohio River. The
compounds that were found include
toluene, ethylbenzene, the xylenes,
and PCE.
actions
1972, April: EPA (1972) reports that there is a problem with

New Orleans drinking water that is drawn from the Mississippi
River. Using a solvent extraction based method, 46 specific
organic compounds are found in drinking water from three
drinking water treatment plants. (44 more compounds are found in
wastewaters discharged by 10 industrial plants to the Mississippi
just above New Orleans.) EPA (1972) concludes that the problem
is due to wastes from industries discharging to the river, and the
chlorination of the drinking water may also be playing a role.
1972, October: The Clean Water Act (CWA) is enacted over a
Nixon veto as a major amendment to the Federal Water Pollution
Control Act (Public Law 92-500). The CWA provides for the
regulation of the discharges into all navigable waters of the U.S.
There are two goals: I) "fishable" and "swimmable" waters by
1983; and 2) eventually, no discharges at all of pollutants into
waterways. The Act provides federal funding to states to set up
and implement water quality management programs. Although the
main focus of the Act is the protection of surface water quality,
some funding is provided to protect groundwater resources from
contamination due to the land application of municipal wastes.
1973, December: ES&T (1973)

reports that "Injection wells for
industrial and sewage waste waters
increase slowly in number as the
call goes out for a national policy

on underground space and wastes
management. "
1974, Summer: Rook (1974)

shows that haloform compounds
like chloroform can form during
chlorination of drinking water due
to reactions with natural humic
compounds present in raw drinking
water. (As a group, trihalomethanes
are referred to as "THMs".)
1974, July: EPA Region VI accepts the request of the State of

Louisiana and City of New Orleans for more help with their
drinking water problem, and begins a study using GCIMS and
other methods to determine organic compounds in selected
samples.
1974, November: EPA (1974) releases a preliminary report
on the New Orleans situation, listing 66 organic chemicals in
the finished water from the three New Orleans drinking water
treatment plants. Most concentrations range from O.OJ to 5 J,LglL.
Later, the total number of compounds identified is raised to 94.
1974, November 8: Concurrently with releasing the report on
New Orleans drinking water, EPA announces plans for a National
Organics Reconnaissance Survey (NORS) of water supplies for
many of the nations' cities (see Symons et al., 1975). Both raw
and finished water are to be studied to see how raw water quality
and chlorination affect the concentrations of THMs. NORS will
begin as soon as the Safe Drinking Water Act is signed.
34
TABLE 1.14
Chap. 1
(continued)
a. Analytical
developments
b. Societal and
anecdotal events
1974, December: Working for

EPA, Bellar and Lichtenberg
(1974) publish the basis of EPA's
purge and trap (P&T) procedure.
Gas is bubbled through a water
sample. the VOCs are stripped
from the water, adsorbed on a
trap containing adsorbents like
Tenax-GC and carbon, and the trap
is then thermally desorbed to a
gas chromatography column for
analysis.
1975, March: EPA scientists Heller

et al. (1975) advocate GC/MS as an
accurate and economical approach
for the determination of organic
pollutants in all media.
1976, January: Henderson et

al. (1976) publish a method
for determining trace levels of
chlorinated organic compounds
using micro solvent extraction
(flSE). A portion of the solvent
extract is injected onto a gas
chromatograph, and detection is
with an electron capture detector
(ECD).
1975, November: The Safe Drinking Water Committee of the

National Research Council holds its first public meeting. The
study will evaluate health effects due to contaminants in drinking
water. (See ES&T, 1976a.)
1976: New York State Department of Health (1980) first learns of

severe groundwater contamination on Long Island due to TCE
and other chemicals. TCE is being used extensively as a septic
tank cleaner by homeowners.
Sec. 1.8
35
Conclusions
c. Literature
reports
actions
1974, December: Bellar et al.

(1974) report that haloforms are
formed when drinking waters
are chlorinated in the U.S., just
as Rook (1972) has found in
The Netherlands. (As a group,
trihalomethanes are referred to as
"THMs".)
1974, December 16: Safe Drinking Water Act is signed into law
by Pres. Ford. Although originally expected to be vetoed, its
passage is due in large part to the storm of controversy that has
erupted over the New Orleans problem (see above). The Act
gives the EPA Administrator the power to take such action as is
needed to protect drinking water supplies. Also, to help protect
groundwater resources, the Act requires that EPA must develop
regulations governing how states control the disposal of wastes
by underground injection.
1974, December 18: The 80 cities selected for NORS are
identified. 16 cities depend on groundwater, 64 depend on
surface water. Of the 80 cities, 10 cities have been selected
for comprehensive study. Two of these ten, namely Miami and
Tucson, are selected as representatives of cities depending on
groundwater. The compounds to be quantitated in the samples
will be the THMs (e.g., chloroform), plus 1,2-dichloroethane
and carbon tetrachloride. Other VOCs like TCE and PER are
not included in the study because they cannot be determined
easily using the existing P&T method. The compound-dependent
detection limits ranged from 0.2 (l,2-dichloroethane) to 2 JLgfL
(carbon tetrachloride).
1975, January-April: Sampling for NORS is carried out.
1975, August: Dowty et al.

(1975) report the presence of
numerous industrial-waste and
chlorination-related organic
compounds in New Orleans drinking
water. GCIMS was used to identify
major aromatic and halogenated
aliphatic compounds. Approximately
60-70 compounds were commonly
found in the finished drinking water.
1975, May: EPA (1975) releases interim report to Congress

on results of the National Organics Reconnaissance Survey
(NORS) for halogenated organics. NORS confirms that THMs are

essentially ubiquitous in chlorinated drinking water, and that their
formation involves the chlorination process (see discussion in
Symons et al., 1975).
1976, March: In part because of the limited nature of the group

of compounds determined in NORS (i.e., primarily THMs), EPA
begins the National Organics Monitoring Study (NOMS) of 113
public water supplies representing various types of sources
and drinking water treatment processes (21 of these supplies
depend primarily on groundwater). The goal is to provide data
for possible establishment of additional maximum concentration
levels of organic compounds in drinking water. NOMS will
include as target analytes several important chlorinated solvent
compounds. Detection limits for various halogenated purgeables
vary from 0.01 to 0.3 JLgfL. (See Brass et al., 1977.)
1976, October 11: Toxic Substances Control Act is signed. The
Act establishes a regulatory framework for controlling the release
of toxic substances to the environment.
36
TABLE 1.14
Chap. 1
(continued)
a. Analytical
developments
b. Societal and
anecdotal events
1977: 16 private wells are closed in Gray, Maine due to

contamination with various chlorinated solvents, as well as certain
other compounds.
1977: In this year alone, in 58 square miles of densely populated
Nassau County (Long Island, New York), homeowners buy
67,500 gallons of septic tank cleaners (NRC, 1984). CEQ (1981)
notes that such fluids are typically used in a dose of I gallon to
dissolve sludge (i.e., grease) in a septic tank. The cleaners contain
80% aromatic and chlorinated solvents (NRC, 1984).
1978: In anticipation of the Feb. 9, 1978 EPA regulations, New

York State undertakes a statewide survey to evaluate the degree to
which THMs (and other selected compounds) are present in their
drinking waters. As described by Schreiber (1981), New York
State discovers definite evidence of groundwater contamination,
e.g., a well supply is found to be highly contaminated with TCE.
The well field is closed,
1978: Nassau-Suffolk (Long Island, New York) Regional
Planning Board finds that organic chemical contamination is
widespread in the upper glacial aquifer (Koppleman, 1978), 36
wells are closed, The closed wells contain up to 375 /l-glL of
PER, 310 /l-glL of I,l,l-TCA, and 300 /l-glL of TCE.
Sec. 1.8
37
Conclusions
c. Literature
reports
1977, April: Pojasek (1977) reports

that: "in the past 5-10 years, the
number of major incidents of
groundwater contamination have
increased dramatically. These
incidents involved chemical spills,
municipal landfills, wastewater
lagoon seepage, and industrial waste
burial. "
1977: In a book chapter that
precedes the final EPA report on
NOMS, Brass et al. (1977) indicate
that I,I,I-TCA, TCE, and carbon
tetrachloride, which are not formed
in the chlorination process, are
present in raw waters used as
drinking water. For the May to July,
1976 sampling, TCE was found
in 28 out of 113 drinking waters.
While this evidence for groundwater
contamination by solvent
compounds was obtained in NOMS,
the implications of this finding were
not discussed in this chapter, nor in
the final report (EPA, I977b). It
is only much later that Westrick
(1990) will note that 6 of the 21
groundwater-dependent supplies
were found to be contaminated
with TCE, and that such "industrial
chemicals can occur frequently in
water supplies using either surface
water or groundwater."
1978, February: Barrett (1978)
reports on the "EPA Consent
Decree that ultimately leads to
the promulgation of the "priority
pollutant" list. This list contains
many of the important chlorinated
solvents.
actions
1976, October: Resource Conservation and Recovery Act (RCRA)
is passed by Congress as an amendment to the Solid Waste
Disposal Act of 1965. The main purpose of RCRA is to address
the problem of how to safely dispose of the huge volumes of solid
and hazardous waste generated in the U.S. each year. However, it
is not until 1980 that regulations authorized under RCRA bring
about large scale groundwater monitoring at disposal sites, and
not until 1986 that amendments to RCRA mandate monitoring of
the groundwater and vadose zones at many underground storage
tank locations.
1977, February 9: EPA proposes regulations to limit certain
organic contaminants (including THMs) in drinking water because
there is a reason to believe that such compounds pose a serious
threat to health (Kimm, 1978).
1977, June: Interim EPA drinking water regulations go into effect
for 10 inorganics and 6 organic pesticides, but no other organic
compounds (ES&T, 1976b).
1977, December: Amendments to the Clean Water Act of 1977
give EPA's Environmental Monitoring and Support Laboratory in
Cincinnati (EMSLlCin) the responsibility for developing adequate
and sensitive methods for the determination of trace organic
compounds in municipal and industrial wastewater, including the
chlorinated solvents (EPA, 1977a).
1978, February 9: Under provisions of the Safe Drinking Water

Act, EPA (1978) publishes interim primary regulations for certain
organics in drinking water, including THMs. Total THMs not to
exceed 100 JLglL.
1978: EPA (1978) publishes results of the National Organics

Monitoring Survey (NOMS) for Halogenated Organics. NOMS
confirms the widespread occurrence of THMs in chlorinated
drinking water.
38
TABLE 1.14
Chap. 1
(continued)
a. Analytical
developments
b. Societal and
anecdotal events
1978, May: Four wells providing 80% of the drinking water to
Bedford, Mass. are found to be contaminated with dioxane at up
to 2100 j.LgfL and with TCE at up to 500 j.LglL. The discovery
was accidental: a resident engineer was testing the water in
preparation for a paper he was writing (Mass. Special Legislative
Commission on Water Supply, 1979).
1978: New York State publishes drinking water standards for
three chlorinated solvent compounds.
1979, April-May: GCIMS is

proposed in the literature as
the best all around method for
determination of organic compounds
in environmental samples (Keith and
Telliard, 1979; and Finnigan et al.,
1979).
1979, December: EPA (1979)
publishes a series of organic
analytical methods (including purge
and trap) that use GCIMS and other
forms of gas chromatography. The
methods have been developed for
use in determining toxic organic
pollutants in wastewaters as
mandated by the 1977 Amendments
to the Clean Water Act, and
specifically for the list of 114
organic "priority pollutants" arising
from the EPA Consent Decree of
1978.
1980, August: Mieure (1980)

reviews methods that can be used to
determine trace levels of chlorinated
solvents in water: headspace, purge
and trap, and the micro solvent
extraction (j.LSE) method proposed
by Henderson et al. (1976). j.LSE
needs to be used with a specific
detector such as an electron capture
detector (ECD). Detection limits for
numerous chlorinated solvents are:
tens of j.LglL by headspacefFID,
1979: The same major chemical company mentioned in 1973

above qualifies its prior recommendations for disposal of
methylene chloride. It now says: "Send solvent to a reclaimer. In
some cases, small amounts can be transported to an area where it
can be placed on the ground and allowed to evaporate safely if
local, state, and federal regulations permit."
1980, January: California public health officials close 39 public
wells providing water to 400,000 people in 13 cities in the San
Gabriel Valley, due to TCE contamination.
1980, March: By this point in time, agencies in 18 states have
measured VOCs, most commonly TCE, in water from 2894 wells
(see CEQ, 1981).
1980, May: In a special, red-paged "Regulatory Alert", ES&T
(1980) reports that: "Approximately 50% afthe [nation :s]
population depends on groundwater for its drinking water supply.
Yet, unlike surface water or air, there is no specific federal statute
to protect it from pollution. . . . Recently, EPA set as one of its
'highest priorities' the development of a National Groundwater
Protection Strategy."
1980, May: Few states have as yet acted to declare authority to
protect groundwater supplies, and most do not even recognize the
magnitude of the problem (Environ. Reporter, 1980a).
1980, August: Existing federal regulations are viewed by some
as still so weak that new laws will be needed to provide effective
groundwater protection (Environ. Reporter, 1980b). In this time
frame, the federal government promulgates new regulations
affecting: I) waste disposal sites under RCRA in May 1980; and
later 2) uncontrolled sites under CERCLA in December 1980; and
3) underground storage tank sites under amendments to RCRA in
1986.
Sec. 1.8
39
Conclusions
c. Literature
reports
1978, May: ES&T (1978) reports
that: "The New Jersey Dept. of Env.
Protection is testing 500 public and
private wells located near industrial
sites for 50 selected toxic and
cancer causing chemicals, including
organic compounds. "
actions
1978, June 7: The "EPA Consent Decree" is signed by EPA and

several environmentally-concerned plaintiffs (Natural Resources
Defense Council, Environmental Defense Fund, Businessmen for
the Public Interest, National Audubon Society, and Citizens for a
Better Environment). The Consent Decree starts the process by
which the EPA establishes their list of /29 priority pollutants; 114
of these pollutants are organic. The list, which includes nearly
all of the chlorinated solvent compounds, is to be the basis for
regulation of effluent discharges (see Keith and Telliard, 1979).
1979, July 30: EPA (1979) publishes the list of "priority
pollutants" in the Federal Register as mandated by the EPA
Consent Decree of 1978.
1980, January: In a cover story

on groundwater contamination, the
journal ES& T (1980) discusses
how ground waters can become
contaminated, how they can be
protected, and how tight federal and
state rules requiring monitoring and
protection will be coming.
1979, November 20: Under provisions of the Safe Drinking Water

Act, EPA releases final regulations to control total trihalomethanes
(THMs) in drinking water at 100 j.tgfL.
1980, February: Smith et al.

(1980) discuss a careful study of
variations in THM concentrations in
a chlorinated drinking water supply.
1980, May: EPA Administrator Costle identifies groundwater as a
principle EPA priority on NBC's Meet the Press (see Josephson,
1980).
1980, May: Regulations authorized under RCRA are promulgated
by EPA, and bring about large scale groundwater monitoring at
disposal sites.
1980, June: EPA (l980a) discusses in a planning workshop that
a landfill in Jackson Township that was licensed to accept sewage
sludge and septic wastes has caused extensive groundwater
contamination. Approximately 100 wells around the landfill are
closed. Analyses reveal concentrations of up to 3,000 j.tg/L of
methylene chloride, and 1,000 j.tgfL of TCE.
40
TABLE 1.14
Chap. 1
(continued)
a. Analytical
developments
b. Societal and
anecdotal events
hundredths of JLglL by headspacel

ECD, and about 1 JLglL by both
P&TIFID and P&T/GCIMS.
(RWQCB) begins to receive numerous reports of leaks to the
1980: San Francisco Bay Regional Water Quality Control Board

subsurface from local industrial chemical storage systems in
"Silicon Valley" (see: Industry Clean Water Task Force, 1985).
1980: In just one year, the same major chemical company

mentioned above for 1973 and 1979 completely changes its
recommendations for disposal of methylene chloride: " ...
Dumping into sewers, on the ground, or into any body of water is
strongly discouraged, and may be illegal."
1981, January: The Council on Environmental Quality (CEQ,
1981), located in the Executive Office of the U.S.President,
releases their landmark, comprehensive report entitled
Contamination of Groundwater by Organic Chemicals. The
document painstakingly compiles the avalanche of reports on
groundwater contamination in the U.S. Chlorinated solvents are
the most common type of contaminants. CEQ (1981) notes:
"These incidents involve discoveries - sometimes completely
accidental - of high concentrations of synthetic organic chemicals
in a community or private drinking water well. It is now evident
that hundreds of drinking water wells affecting millions of people
have been closed because of contamination by toxic organic
chemicals. . . . No one can say for sure how long the contamination
has existed, . the EPA regions report serious contamination of
drinking water wells in 34 states, and there are indications that
the true number is at least 40."
January, 1981: CEQ (1981) reports that septic tank cleaning
fluids, most of which contain TCE, benzene, or methylene
chloride, can still be purchased in most areas by homeowners.
CEQ (1981) notes that an estimated 400,000 gallons of such
septic tank cleaners were used by homeowners in 1979 on Long
Island alone. This subsurface release of solvents has been the
cause of extensive contamination.
1981: DNAPLs are encountered in the subsurface at several sites

in the Niagara Falls area (Kolmer, 1990).
Sec. 1.8
41
Conclusions
c. Literature
reports
actions
1980, November 24: EPA (l980b) announces a National Ground
Water Protection Strategy. The Strategy has three phases. Phase
I: data was gathered during the winter and spring of 1980.
Phase 11: two workshops were conducted in June, 1980 with 80
participants. Phase 1l1: the planning document is now released to
the public, with proposed strategy to be published in the Federal
Register in January 1981.
1980, December: U.S. Congress passes the Comprehensive
Environmental Response, Compensation, and Liability Act
(CERCLA), better known as "Superfund" to deal with the threats
caused by "uncontrolled" hazardous waste sites.
1981, Jauuary: ES&T (l98Ia)

reports that Recon Systems, Inc.
(Somerville, NJ) has built and
is operating an air stripping
tower to remove I,I,I-TCA from
groundwater that is being pumped at
100,000 gpd from an aquifer in the
Northeast.
1981, February: ES&T (l98Ib)
reports that studies sponsored by the
Council on Environmental Quality
(CEQ) have found "Significant
evidence of increased cancer risks
from drinking chlorinated drinking
water". The studies indicated
increased risk of rectal cancer
of 13-93% in groups that drank
chlorinated drinking water.
1981, April: Petura (1981) describes
groundwater pollution as a "problem
of immense concern" with TCE and
I,I,I-TCA having been found in
groundwater in numerous locations,
e.g., Penn., Del., New Hamp., and
Calif. "Hundreds of wells, many
in totally unrelated aquifers" are
described as being contaminated
with these two compounds.
1981, July: ES&T (l98Id) reports
again that CEQ (1981) has found
that U.S. groundwater is surprisingly
contaminated (see also Jan. 1981,
column b).
1981, April: ES&T (1981c) reports that: I) despite the fact that
she has no environmental experience, Anne Gorsuch has been
named EPA Administrator; and 2) that President Reagan had
been considering doing away with the Council on Environmental
Quality (CEQ), and is proposing to cut CEQ's budget by 70%.
Council members say that the budget cut will not permit them
to carry out the tasks that have been assigned to them by the
National Environmental Policy Act of 1979.
1981, June 9: In a presentation at the national meeting of the
American Water Works Association, EPA employee H. J. Brass
with co-authors presents results of the Community Water Supply
Survey (CWSS, Brass et al., 1981). The survey compares water
quality for surface water samples from 106 sources to ground
water samples from 330 sources. Clear evidence is found for the
presence PER, TCE, I, 1,1-TeA, and other chlorinated solvent
compounds in many groundwater supplies. These three
42
TABLE 1.14
Chap. 1
(continued)
a. Analytical
developments
b. Societal and
anecdotal events
1982: The San Francisco Bay area "RWQCB" launches a program

to inventory underground storage systems in industrialized areas.
RWQCB undertakes a large survey by mail to 2644 companies
in the key groundwater basins. Most of the sites are in the
Santa Clara ("Silicon") Valley. The survey ultimately leads
to the publication of RWQCB's "2051" report (see: Industry
Clean Water Task Force, 1985). A significant amount of the
contamination in the valley is due to high tech industrial
operations.
Sec. 1.8
43
Conclusions
c. Literature
reports
actions
1981: Giger and Schaffner (1981)

and Zoeteman et al. (1981)
document the presence of TCE,
chloroform, and carbon tetrachloride
in groundwater in Europe.
compounds are each found in about 5% of the groundwater

sources tested, with maximum concentrations of 30, 650, and
210 {LgfL. The large number of groundwater systems tested
(as compared to the 21 groundwater supplies tested in NOMS)
allowed some very serious groundwater contamination to be
found. Of the 330 groundwater systems tested, 50 were found to
be contaminated with some VOCs. (See also Westrick, 1990.)
1981: Schwille (1981) discusses

behavior of NAPLS in subsurface
systems (in English).
1981, December: Page (1981) notes
"contamination of public drinking
water supplies by toxic substances
poses a serious problem in the U.S.".
Data for 1076 wells in NJ indicate
that groundwaters in NJ are at least
as contaminated as surface waters.
17 halogenated solvent compounds
are found in the groundwater. For
TCE, of 669 wells sampled, 388
were found to be contaminated,
with 635 {LgfL being the highest
concentration. For I,I,I-TCA, of
1071 wells sampled, 835 were
contaminated, with 608 {Lg/L being
the highest concentration. For PER,
of 421 wells sampled, 179 were
contaminated, with 91 {LgfL being
the highest concentration.
1982, August: Dyksen and Hess
(1982) state "the infiltration
of organic compounds into

groundwater resources has been
occurring for decades. . .. organic
infiltration of groundwater
supplies is widespread", e.g.,
31 municipal wells within 13
different water systems have been
closed due to high levels of TCE.
Affected groundwater supplies
typically contain numerous organic
compounds; TCE is the most
common.
1983, November: ES&T (1983)
reports on the solvent injection/tracer
test occurring at the Borden Landfill.
1982, July: EPA (1982) issues new rules (taking effect in

January of 1983) governing hazardous wastes in landfills, surface
impoundments, and other repositories. The rules state, that liners
must be used to prevent leaching of contaminants from such
facilities, and that backup leachate collection must be provided.
44
TABLE 1.14
Chap. 1
(continued)
a. Analytical
developments
b. Societal and
anecdotal events
1985, September: San Francisco Bay area RWQCB lists 75 sites

within their jurisdiction at which groundwater contamination is:
I) under investigation; 2) in the process of being defined; 3)
being cleaned up; or 4) has been cleaned up (see: Industry Clean
Water Task Force, 1985).
Sec. 1.8
45
Conel usions
c. Literature
reports
actions
1983: Pye et al. (1983) observe:

"The contamination of groundwater
by synthetic organic chemicals
has come to light recently, mainly
because they are not routinely
monitored."
1984: Schwille's (1984a)
comprehensive monograph on
chlorinated solvent behavior in
subsurface systems appears, but in
German.
1984: Westrick et al. (1984)
report the findings of EPA's
Ground Water Supply Survey
(GWSS). GWSS concludes that
contamination of groundwater by
chlorinated solvent type compounds
is common and widespread. Of
500 randomly-selected community
groundwater supplies, 230 were
found to be contaminated. (See also
Westrick, 1990.)
1984: Several reports in conference
proceedings and publications of
limited circulation describe the
DNAPL nature of liquid chlorinated
solvents, and the dangerous
implications of that nature for
groundwater (Feenstra, 1984; Cherry
1984a and 1984b; Miller, 1984).
1985: The first widely-read
publications describing the DNAPL
nature of liquid chlorinated solvents
appear (Villaume, 1985; and Mackay
et aI., 1985).
1988: Schwillc's (1984a) monograph
(see above) appears in translation in
English as Schwille (1988).
1988: Fetter (1988) provides a
brief discussion of DNAPLs and
groundwater in his textbook on
applied hydrogeology.
1986: Hazardous and Solid Waste Amendments to RCRA are

passed. The amendments mandate monitoring of groundwater and
vadose zone at many underground storage tank locations.
1987, July 8: Final drinking water standard value for TCE is

published by EPA (1987) as 5 JLg/L.
1987: EPA (1987b) promulgates regulations prohibiting the used
of certain dense chlorinated solvents as "inactive ingredients" for
pesticide formulations.
1988: EPA provides a cursory reference to the behavior of
DNAPLs in the subsurface in their document Guidance on Remedial Actions for Contaminated Ground Water at Superfund Sites.
46
TABLE 1.14
Chap. 1
(continued)
a. Analytical
developments
b. Societal and
anecdotal events
ACKNOWLEDGEMENT
Funding support for this research was provided in part by the University Consortium
Solvents-in-Groundwater Research Program. Sponsors of the Program between 1988
and 1994 have included: The Boeing Company, Ciba-Geigy Corporation, Dow Chemical
CanadalUSA, Eastman Kodak Co., General Electric Co., Laidlaw Environmental Systems
Ltd., Mitre Corporation, The Natural Sciences and Engineering Research Council of
Canada, and the Ontario University Research Incentive Fund.
1.9 REFERENCES
Abelson, P. H. (1990) "Volatile contaminants of drinking water", Science, 247, 141.
American Insurance Association (1967) Chemical Hazards Bulletin, Engineering and Safety Dept.,
85 John St., New York, NY, October 1967.
American Insurance Association (1972) Chemical Hazards Bulletin, Revised, Engineering and
Safety Service, 85 John St., New York, NY, March 1972.
ASTM (1962) Committee 0-16 on Halogenated Organic Solvents, Handbook of Vapor Degreasing,
American Society for Testing and Materials, Special Publication No. 310. Philadelphia, Penn.
ASTM (1976) Committee 0-26 on Halogenated Organic Solvents, Handbook of Vapor Degreasing,
American Society for Testing and Materials, Special Publication No. 3 lOA. Philadelphia, Penn.
Barrett, B. R. (1978) "Controlling the entrance of toxic pollutants into U.S. waters", Environ. Sci.
Technol. 12, 154-162.
Bellar, T. A. and J. J. Lichtenberg (1974) "Determining volatile organics at the microgram per
litre level by gas chromatography", J. Am. Water Works Assoc. 66,739-744.
Bellar, T. A., J. J. Lichtenberg, and R. C. Kroner (1974) "The occurrence of organohalides in
chlorinated drinking water", J. Am. Water Works Assoc. 66, 703-706
Sec. 1.9
47
References
c. Literature
reports
actions
1990: Domenico and Schwartz

(1990) provide the first substantive
textbook treatment of DNAPLs and
groundwater.
1991: EPA publishes their GroundWater Issue Paper on DNAPLs
(Hurling and Weaver, 1991).
1993: EPA sponsors publication of a comprehensive monograph
on DNAPLs DNAPLSite Investigation (Cohen and Mercer, 1993).
1993: EPA (1993) publishes the final report Evaluation of
Likelihoodof DNAPL Presence at NPL Sites - NationalResults
Brass, H. J., M. A. Feige, T. Halloran, J. W. Mello, D. Munch, and R. F. Thomas (1977) "The National Organic Monitoring Survey: Samplings and analyses for purgeable organic compounds",
in Drinking Water Quality Enhancement through Source Protection, R. B. Pojasek (Ed.) Ann
Arbor Science, Ann Arbor, MI, pp. 393-416.
Brass, H. J., M. J. Meisner, and B. A. Kingsley (1981) "Community Water Supply Survey: Sampling
and analysis for purgeable organics and total organic carbon". Paper presented at the Annual
Meeting of the American Water Works Association, Water Quality Division, June 9, 1981.
Bredehoft, J. (1992) "Much contaminated groundwater can't be cleaned up", Ground Water, 30,
834-835.
CEQ (Council on Environmental Quality) (1981) Contamination ofGround Water by Toxic Organic
Chemicals, January, 1981, U.S. Government Printing Office, Washington, D.C., 84 pages.
Cherry, J. A. (1984a) "Contaminant migration in groundwater: Processes and problems". In: Proceedings: Second National Water Conference: The Fate of Toxics in Surface and Ground Water,
Philadelphia Academy of Science, January 24-25, 1984.
Cherry, J. A. (l984b) Groundwater Contamination, Mineralogical Association of Canada Short
Course in Environmental Geochemistry, Short Course Handbook Vol. 10, M.E. Fleet (Ed.),
May, 1984, London, Ontario.
Cohen, R. M. and J. W. Mercer (1993) DNAPL Site Evaluation, (EPA Project Officer: J. Matthews),
C. K. Smoley/CRC Press, Boca Raton, Florida.
Domenico, P. A. and F. W. Schwartz (1990) Physical and Chemical Hydrogeology, John Wiley &
Sons, New York.
Dowty, B. J., D. R. Carlisle, and J. L. Laseter (1975) "New Orleans drinking water sources tested
by gas chromatography-mass spectrometry", Environ. Sci. Technol. 9, 762-766.
Dyksen, J. E. and A. F. Hess, III (1982) "Alternatives for controlling organics in groundwater
supplies", J. Am. Water Works Assoc. 74, 394-403.
Environment Reporter (l980a) May 30, 1980, 142.
Environment Reporter (l980b) August I, 1980,477.
48
Chap. 1
EPA (1972) "Industrial pollution of the Lower Mississippi River in Louisiana", Region VI, Dallas,
Texas, Surveillance and Analysis Division.
EPA (1974) "Draft analytical report, New Orleans area water supply study", Region VI, Surveillance and Analysis Division, EPA Report 90611 0-74-002, November, 1974.
EPA (1975) "Preliminary assessment of suspected carcinogens in drinking water: Interim report to
Congress", and "Appendix to the interim report to Congress", June 1975.
EPA (l977a) 'The Clean Water Act showing changes made by the 1977 amendments", Federal
Register, 95-12, December, 1977.
EPA (l977b) 'The National Organic Monitoring Survey", Technical Support Division, Office of
Water Supply, U.S. Environmental Protection Agency.
EPA (1978) "Control of organic chemical contaminants in drinking water", Federal Register, 43,
No. 28, Feb. 9, 1978, pp. 5775 et seq., and "Identification of Conventional Pollutants", Federal
Register, 44, pp. 44501-44503.
EPA (1979) "Guidelines establishing test procedures for the analysis of pollutants: proposed regulations", Federal Register, 44, (233) 69464.
EPA (1980a) Planning Workshops to Develop Recommendations for a Ground Water Protection
Strategy, Appendices, Washington, D.C.
EPA (l980b) Proposed Ground Water Protection Strategy. Office of Drinking Water. 61 pp.
EPA (1982) Federal Register, 47, No. 143,32274.
EPA (l987a) Federal Register, 52, No. 130,25690-25691, July 8, 1987.
EPA (l987b) Federal Register, 52, No. 77, 13305-13309.
EPA (1988) Guidance on Remedial Actions for Contaminated Ground Water at Superfund Sites,
EPA/540/G-88/003, Office of Emergency and Remedial Response, U.S. EPA, Washington, D.C.,
December 1988.
EPA (1991) CERCUS Characterization Project - National Results, EPA 540-8-91/080.
EPA (1993) Evaluation of Likelihood of DNAPL Presence at NPL Sites - National Results, Final
Report, EPA/540/R-93-073.
ES&T (1972) "Groundwater pollution and conservation", Environ. Sci. Technol. 6,213-215.
ES&T (1973) "Currents", Environ. Sci. Technol. 7,485.
ES&T (1976a) "Currents", Environ. Sci. Technol. 10, II.
ES&T (1976b) "Currents", Environ. Sci. Technol. 10, 115.
ES&T (1978) "Currents", Environ. Sci. Technol. 12,503.
ES&T (1980) "Safeguards for groundwater", Environ. Sci. Techno!. 14,38-44.
ES&T (1981a) "Currents", Environ. Sci. Technol. 15, II.
ES&T (l98Ib) "Currents", Environ. Sci. Technol. 15, 131.
ES&T (l98Ic) "Currents", Environ. Sci. Technol. 15,373.
ES&T (198Id) "Currents", Environ. Sci. Technol. 15,727
ES&T (1983) "Subsurface organic contaminants" Environ. Sci. Technol. 17, 518A-521 A.
Feenstra, S. (1984) "Groundwater contamination by dense non-aqueous phase liquid (DNAPL)
chemicals", Geological Association of Canada, Annual Meeting, May 14-16, 1984, London,
Ontario.
Sec. 1.9
References
49
Feenstra, S. (1986) "Subsurface contamination from spills of dense non-aqueous phase liquid
(DNAPL) chemicals". In: Proceedings: Technical Seminar on Chemical Spills, Environment
Canada, February 5-7, 1986, Montreal, Quebec, p. 11-22.
Feenstra, S. and J. A. Cherry (1988) "Subsurface contamination by dense non-aqueous phase liquid
(DNAPL) chemicals". In: Proceedings: International Groundwater Symposium, International
Association of Hydrogeologists, May 1-4, 1988, Halifax, Nova Scotia.
Fetter, C. W. (1988) Applied Hydrogeology, 2nd Edition, Merrill Publishing Company, Columbus,
Ohio.
Finnigan, R. E., D. W. Hoyt, and D. E. Smith (1979) "Priority pollutants. II. Cost-effective analysis", Environ. Sci. Technol. 13, 534-541.
Giger, W. and C. Schaffner (1981) "Groundwater pollution by volatile organic chemicals". In
Quality of Groundwater, W. van Duijvenbooden, P. Glasbergen, and H. van Lelyveld (Eds.)
Elsevier, Amsterdam, pp. 511-522.
Goodenkauf, A. and J. Erdie (1964) "Identification of chlorinated hydrocarbon pesticides in river
water", J. Am. Water Works Assoc., 56, 600-606.
Gutenmann, W. H. and D. J. Lisk (1964) "Gas chromatographic methods for Silvex", J. Am. Water
"'lirksAssoc. 56, 189-190.
Halogenated Solvents Industry Alliance (1987) personal communication.
Heller, S. R., J. M. McGuire, and W. L. Budde (1975) "Trace organics by GCIMS", Environ. Sci.
Technol. 9, 210-212.
Henderson, J. E., G. R. Peyton, and W. H. Glaze (1976) "A convenient liquid/liquid extraction
method for the determination of halomethanes in water at the parts per billion level". In:
Identification and Analysis of Organic Pollutants in Water. L. H. Keith (Ed.) Ann Arbor Science,
Ann Arbor, Michigan, pp. 105-111.
Hunt, J. R., N. Sitar, and K. S. Udell (1988a) "Nonaqueous phase liquid transport and cleanup. I.
Analysis of mechanism", Water Resources Research, 24, 1247-1258.
Hunt, J. R., N. Sitar and K. S. Udell (1988b) "Nonaqueous phase liquid transport and cleanup. 2.
Experimental studies", Water Resources Research, 24, 1259-1269.
Hurling, S. G. and J. W. Weaver (1991) Dense nonaqueous phase liquids. U.S. EPA Ground Water
Issue Paper. EPA/540/4-91-002. 40 p.
Industry Clean Water Task Force (1985) Protecting the Valley's Water: A Factual Summary of
Actions Taken by Industry to Remove Solvents from Soil and Groundwater in the "Silicon Valley",
September, 1985, 8 pages.
Josephson, J. (1980) "Groundwater strategies" Environ. Sci. Technol. 14, 1030-1036.
Johnson, R. L. and J. F. Pankow (1992) "Dissolution of dense chlorinated solvents into groundwater.
2. Source functions for pools of solvents", Environ. Sci. Technol. 26, 896-901.
Keith, L. H. (1974) "Chemical characterization of industrial wastewaters by gas chromatographymass spectrometry", The Science of the Total Env., 3, 87-102.
Keith, L. H. and W. A. Telliard (1979) "Special Report: Priority pollutants. I. A perspective view",
Environ. Sci. Techno!' 13, 416-423.
Kirnm, V. J. (1978) "ES&T Guest Editorial: Control of organic contaminants in drinking water",
Environ. Sci. Technol. 12, 363.
Kleopfer, R. D. and B. J. Fairless (1972) "Characterization of organic components in a municipal
water supply" Environ. Sci. Techno!' 6, 1036-1037.
50
Chap. 1
Kolmer, J. (1991) "The history of the term "dense non-aqueous phase liquid". In: DNAPLs in
Porous and Fractured Media: A Short Course. Waterloo Centre for Groundwater Research,
October 7-10, 1991. Kitchener, Ontario.
Koppleman, L. E. (1978) "The Long Island Waste Treatment Management Plan", Nassau-Suffolk
Regional Planning Board, Hauppague, New York.
Mackay, D. M., P. V. Roberts, and J. A. Cherry (1985) 'Transport of organic contaminants in
groundwater", Environ. Sci. Technol., 19, 384-392.
Mackay, D. M. and J. A. Cherry (1989) "Groundwater contamination: Pump and treat remediation",
Environ. Sci. Technol. 23, 630-636.
Massachusetts Special Legislative Commission on Water Supply (1979) Chemical Contamination,
Boston.
McCarthy, K. A. and R. L. Johnson (1992) "Transport of volatile organic compounds across the
capillary fringe", Water Resources Research, 29, 1675-1683.
McKee, J. E., F. B. Laverty, and R. M. Hertel (1972) "Gasoline in groundwater", J. Water Poll.
Control Fed., 44, 293-302.
Melpolder, F. W., C. W. Warfield, and C. E. Headington (1953) "Mass spectrometer determination
of volatile contaminants in water", Anal. Chem., 25, 1453-1456.
Merck & Co., Inc. (1989) The Merck Index: An Encyclopedia of Chemicals, Drugs and Biologicals,
Eleventh Edition, Merck & Co. Inc., Rahway, New Jersey.
Mercer, J. W. and R. M. Cohen (1990) "A review of immiscible fluids in the subsurface: properties,
models, characterization and remediation", J. Contaminant Hydrology, 6, 107-163.
Middleton, F. M. and J. J. Lichtenberg (1960) "Measurements of organic contaminants in the
nation's rivers", Ind. Eng. Chem., 52, 99A-102A.
Mieure, J. P. and M. W. Dietrich (1973) "Determination of trace organics in air and water" J.
Chromatographic Science, 11, 559-570.
Mieure, J. P. (1980) "Determining volatile organics in water", Environ. Sci. Technol. 14, 930-935.
Miller, D. W. (1984) Groundwater Contamination: A Special Report, American Assoc. for Advancement of Science, Washington, D.C., 25 p.
Montgomery, H. A. C. and M. Conlon (1967) "The detection of chlorinated solvents in sewage
sludge",1. Water Poll. Control. Fed., 66, 190-192.
New York State (1978) Ground Water Classifications, Quality Standards, and Effluent Standards
and/or Limitations (Title 6, Official Compilation of Codes, Rules, and Regulations, Part 703),
Department of Environmental Conservation, Effective September I, 1978.
New York State (1980) Organic Chemicals and Drinking Water, Department of Health, Albany.
New York Times (1980, November 2) Dumped Plane Fuel at Lakehurst Stirs Water-Pollution Fear,
pages I and 26.
NRC (National Research Council) (1984) Groundwater Contamination, National Academy Press,
Washington, D.C., p. 124.
Page, G. W. (1981) "Comparison of groundwater and surface water for patterns and levels of
contamination by toxic substances", Environ. Sci. Technol. 15, 1475-1480.
Pankow, J. F. and J. A. Cherry (1988) Foreward to Dense Chlorinated Solvents in Porous and
Fractured Media: Model Experiments, Schwille, F. (1988), translated from the German by J. F.
Pankow, Lewis Publishers, Boca Raton, Florida, 146 pages.
Sec. 1.9
References
51
Petura, J. C. (1981) "Trichloroethylene and methyl chloroform in groundwater: A problem assessment", J. Am. Water Works Assoc. 73, 200-205.
Plumb, R. H. Jr. and A. M. Pitchford (1985) "Volatile organic scans: Implications for ground water
monitoring. In: Proceedings: Petroleum Hydrocarbons and Organic Chemicals in Groundwater,
National Water Well Association, Houston, November 13-15, 1985, pp. 207-222.
Pojasek, R. B. (1977) "How to protect drinking water sources", Environ. Sci. Techno/., 11, 343-347.
Pye, V. I., R. Patrick, and J. Quarles (1983) Groundwater Contamination in the United States,
Univ. Penn. Press, Philadelphia, p. 108.
Rook, J. J. (1972) "Production of potable water from a highly polluted river", J. Water Treatment
Exam. 21, 259-274.
Rook, J. J. (1974) "The formation of haloforms during chlorination of natural waters", J. Water
Treatment Exam. 23, 234-243.
Rosen, A. A., F. M. Middleton, and N. W. Taylor (1956) "Identification of anionic synthetic
detergents in foams and surface waters", J. Am. Water Works Assoc., 48, 1321-1330.
Rosen, A. A. (1976) 'The foundations of organic pollutant analysis", In: Identification and Analysis
of Organic Pollutants in Water, L. H. Keith (Ed.) Ann Arbor Science, Mich. pp. 3-14.
Schaumburg, F. D. (1990) "Banning trichloroethylene: Responsible reaction or overkill?", Environ.
Sci. Techno/., 24, 17-22.
Schreiber, J. (1981) 'The occurrence of trihalomethanes in public water supply systems of New
York State", J. Am. Water Works Assoc., 73, 154-159.
Schroeder, E. H. (1985) "Learning from the past, looking to the future. The Air Force and Hughes
Aircraft: Working together to protect groundwater", J. Freshwater, 1985.
Schwille, F. (1967) "Petroleum contamination of the subsoil - A hydrological problem". In: The
Joint Problems of the Soil and Water Industries. Proceedings of a symposium held at the Hotel
Metropole, Brighton, Jan. 18-20, 1967. Institute of Petroleum. Elsevier, London. pp. 23-54.
Schwille, F. (1981) "Groundwater pollution in porous media by fluids immiscible with water", The
Sci. of the Total Environment, 21, 173-185.
Schwille, F. (1984a) Leichtfluchtige Chlorkohlenwasserstoffe in porosen und kliiftigen Medien,
Besondere Mitteilungen zum Deutschen Gewasserkundlichen Jahrbuch, Nr. 46, Bundesanstalt
fur Gewasserkunde, Koblenz.
Schwille, F. (1984b) "Migration of organic fluids immiscible with water in the unsaturated zone".
In: Pollutants in Porous Media: The unsaturated zone between soil surface and groundwater,
Springer-Verlag, Germany.
Schwille, F. (1988) Dense Chlorinated Solvents in Porous and Fractured Media: Model Experiments, translated from the German by J. F. Pankow, Lewis Publishers, Boca Raton, Florida,
146 pages.
Seto, T. A. and M. O. Schultze (1956) "Determination of trichloroethylene, trichloroacetic acid,
and trichloroethanol in urine", Anal. Chem., 28, 1625-1629.
Smith, V. L., I. Cech, J. H. Brown, and G. F. Bogdan (1980) 'Temporal variations in trihalomethane
content of drinking water", Environ. Sci. Techno!' 14, 190-196.
Symons, J. M., T. A. Bellar, J. K. Carswell, J. CeMarco, K. L. Kropp, G. G. Robeck, D. R. Seeger,
C. J. Slocum, B. L. Smith, and A. A. Stevens (1975) "National Organics Reconnaissance Survey
for halogenated organics", J. Am. Water Works Assoc., 67, 634-647.
52
Chap. 1
Thomas, H. E. (1951) The Conservation of Groundwater: A Survey of the Present Groundwater

Situation in the United States, McGraw-Hill, New York.
United States International Trade Commission (1991) Synthetic Organic Chemicals: United States
Production and Sales, USITC Publication 2470, Washington, D.C.
Villaume, J. F. (1985) "Investigations at sites contaminated with dense non-aqueous phase liquids
(NAPLs)", Ground Water Monitoring Review, Spring, IV, 60-74.
Westrick, J. J., J. W. Mello, and R. F. Thomas (1984) "The groundwater supply survey", J. Am.
Water Works Assoc. 76, 52-59.
Westrick, J. J. (1990) "National Surveys of Volatile Organic Compounds in Groundwater and Surface Waters". In: Significance and Treatment of Volatile Organic Compounds in Water Supplies.
Ram, N. J., R. F. Christman, and K. P. Cantor (Eds.), Lewis Publishers, Boca Raton, Florida.
Zoeteman, B. C; E. de Greef, and F. J. J. Brinkman (1981) "Persistency of organic contaminants in groundwater, lessons from soil pollution incidents in The Netherlands". In: Quality
of Groundwater, W. van Duijvenbooden, P. Glasbergen, and H. van Lelyveld (Eds.). Elsevier,
Amsterdam, pp. 465-476.
2
Conceptual Models for the
Behavior of Dense
Non-Aqueous Phase Liquids
(DNAPLs) in the Subsurface
Stan Feenstra 1 ,2 , John A. Cherry", and Beth L. Parker?
1 Applied
Groundwater Research, Ltd.

2550 Argentia Road
Mississagua, Ontario
Canada L5N 5R1

Waterloo, Ontario
Canada N2L 3G1
Page
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
Introduction
Plumes from DNAPLs
2.5.4 Ultimate Fate
Implications for Site Characterization and Monitoring
Implications for Site Remediation
Conclusions
References
54
57
57
59
61
61
63
64
64
67
70
71
75
75
78
80
82
84
85
85
88
53
54
Conceptual Models, Behavior, Liquids in Subsurface
Chap. 2
ABSTRACT
Dense non-aqueous phase liquids (DNAPLs) behave differently in the subsurface than
do other contaminants. Their behavior normally results in complex DNAPL distributions
which are strongly influenced by geologic heterogeneities. This chapter qualitatively describes the factors that control subsurface DNAPL behavior and the plumes of vadoseand groundwater-zone contamination that can emanate from DNAPL source zones. Also
presented are conceptual models for DNAPL behavior in permeable porous media, fractured media with porous low-permeability matrix material, and fractured media with
non-porous matrix material. Most of the conceptual models apply to all DNAPLs, though
some are specific to halogenated organic solvents. The high densities, low interfacial tensions, and low viscosities of halogenated solvents can lead to deep DNAPL penetration
at many sites. DNAPL movement through small fractures in low permeability aquitards
is common.
In permeable porous media, much of the DNAPL contaminant mass that enters a
groundwater zone can remain as a persistent source zone of "residual" and/or "free-phase"
DNAPL for many years and decades after the release. In some types of fractured porous
media such as fractured clay, shale, and sandstone, DNAPL dissolution and subsequent
diffusive loss from the fractures into the surrounding porous matrix can permit DNAPL
disappearance from the fractures so that most of the contaminant ends up in the dissolved
and sorbed states in the porous matrix. In either case, without appropriate remedial action,
the source zone can contribute contaminants to a groundwater plume for decades to
centuries. Plumes originating from such source zones commonly travel large distances
and ultimately impact water-supply wells, or enter rivers or lakes. The conceptual models
discussed here constitute a paradigm for DNAPL behavior which provides a framework
for developing new strategies for site investigations and remediation as discussed in
subsequent chapters of this book.
2.1 INTRODUCTION
Conceptual models for the movement, distribution and fate of "dense non-aqueous phase
liquids" (DNAPLs) in the subsurface have evolved from a primitive state a decade ago,
to groups of models that are designed to consider many of the large variety of hydrogeological conditions occurring at field sites. The conceptual models are based on
laboratory studies, field experiments, mathematical model results, as well as observations that have been made at many chemical spill and waste disposal sites across North
America. The conceptual models and the many implications derived from them constitute
a paradigm for the behavior of DNAPLs in groundwater. This paradigm provides the
scientific basis for the monitoring, diagnosis, and remediation of sites where DNAPLs
are the primary cause of subsurface contamination. Prior to the development of the
DNAPL paradigm, these sites were being assessed using the same general paradigm
applied for many decades to all types of industrial sites where groundwater contamination occurs, with the implicit assumption that DNAPLs were not present and did not
cause unique problems. It was believed during that earlier period that groundwater con-
Sec. 2.1
Introduction
55
tamination was caused at industrial sites by discharges of aqueous waste, or because

of leaching of contaminated soil/waste situated above the water table (i.e., in the vadose zone). In contrast, in the DNAPL paradigm, much of the contaminant mass in the
groundwater can come from the dissolution of persistent DNAPL mass(es) situated below the water table. This chapter describes conceptual models for how DNAPLs can
behave in common hydrogeologic settings. The physical and chemical processes influencing subsurface DNAPL behavior as well as plume behavior are also described briefly;
detailed discussions of these behaviors and of the implications of the conceptual models
for site investigations, site diagnosis, and site remediation are presented in subsequent
chapters.
The movement and ultimate distribution of a DNAPL in the subsurface depends
on many factors, e.g., the type of geologic medium into which the DNAPL enters. Three
broad categories of geological media can be identified: l) porous media, also referred
to as granular media; 2) fractured porous media; and 3) fractured non-porous media.
In porous media, the water and contaminant movement can occur in the connected pore
space between the solid grains or particles. At many sites, the geologic media present
fit conveniently into these categories. However, in the most general context, geologic
media offer a continuum of properties ranging from ideal porous media such as coarsegrained sand to relatively non-porous fractured rock. Unconsolidated sand or gravel are
commonly found in porous-media aquifers. Unweathered silty or clayey strata devoid
of fractures are examples of porous-media aquitards.
In most saturated fractured porous media, the term "fractures" is used to refer
collectively to the open joints, fissures, and or faults that are present. Water flow through
a fractured porous medium occurs exclusively, or nearly so, through the fractures. The
flow of a DNAPL phase through this type of medium is also generally limited to the
fractures. The low-permeability porous medium between the fractures is known as the
matrix. Fractured silty or clayey deposits are common examples of fractured porous
media, as are many types of fractured sedimentary rocks such as fractured sandstone and
shale. The porosity of matrix material is in the range of a few percent for low porosity
sedimentary rock to 50 percent or more for some types of unconsolidated clayey deposits.
This porosity is sufficient to significantly affect the migration rates and distributions of
contaminants in these fractured porous media.
In fractured non-porous media, the matrix has both negligible permeability and negligible porosity. All significant groundwater flow and contaminant migration (including
DNAPL flow) takes place in the open space of the fracture network. Fractured crystalline rock such as granite and marble are common examples of fractured non-porous
rock. Fractured non-porous media differ from fractured porous media primarily by the
influence of the matrix porosity. In fractured porous media, the matrix has sufficient
interconnected porosity to allow significant diffusion of contaminants into and out of
the matrix. In fractured non-porous media, no appreciable contaminant migration occurs
within the matrix.
Figure 2.1 is a schematic representation of what can occur when DNAPLs and
LNAPLs (light non-aqueous phase liquids) are released from leaky storage tanks into a
granular porous medium. The DNAPL penetrates below the water table due to its density.
56
Chap. 2
Figure 2.1 Schematic illustration of a DNAPL and a LNAPL in a porous medium.

showing geologic and pore scales. A low-permeability clay layer deflects the DNAPL.
DNAPL dissolution causes a plume (from Mackay and Cherry, 1989).
Both the DNAPL and LNAPL can dissolve into water. These dissolved components
form plumes of contamination that travel in the direction of the groundwater flow. Any
partitioning of contaminants onto the aquifer solids will retard this movement relative
to the velocity of the groundwater flow.
Figure 2.2 represents a release of a DNAPL into a thin layer of permeable overburden overlying fractured porous bedrock. The DNAPL sinks through the overburden
into the fracture network. Once in the fractures, the DNAPL dissolves, causing a plume
of dissolved contamination in the fracture network as well as entry of dissolved contamination into the matrix. The movement and distribution of DNAPL here are both very
different from that presented in Figure 2.1 for a non-fractured porous medium.
The conceptual models presented in this chapter are generally relevant to all of
the important types of DNAPLs found in groundwater, including chlorinated solvents,
creosote, coal tar, and PCB oils. While the specific intent of this chapter is to discuss
the aspects of these models that are the most relevant to chlorinated solvents, there are
Sec. 2.2
57
Diffused
into and sorbed
onto rock matrix
Dissolved
Figure 2.2 Schematic illustration of a DNAPL in a fractured porous medium showing

geologic and pore scales. The DNAPL dissolves to create contamination in the fractures
and in the low-permeability rock matrix (from Mackay and Cherry, 1989).
important differences between the groundwater impacts of solvent DNAPLs relative to

other DNAPLs; some of these differences will be discussed.
2.2 PHYSICAL AND CHEMICAL PRINCIPLES
Although the subsurface behavior characteristics of DNAPLs differ considerably from

those of petroleum hydrocarbon LNAPLs, the basic physics and mathematics describing
the multiphase flow of DNAPLs and LNAPLs in the subsurface are the same. In the
vadose zone, consideration of DNAPL migration must account for the interactions between the DNAPL, the water, and the air. In the groundwater zone, interaction between
the DNAPL and water phases are important. For the purpose of this discussion, it is
convenient to consider three different but related aspects of DNAPL migration through
58
Chap. 2
geologic media: 1) conditions of DNAPL entry; 2) conditions for DNAPL flow; and
3) the ultimate static distribution of the DNAPL after the flow of DNAPL has ceased.
Before proceeding, we emphasize that the term DNAPL refers only to an actual nonaqueous phase so that "liquid DNAPL" would be redundant, and "dissolved DNAPL"
would be contradictory.
The entry of DNAPL into porous or fractured media is controlled primarily by
capillary phenomena arising from the facts that an interfacial tension is present between
two mutually immiscible fluids (e.g., air/DNAPL or water/DNAPL), and that the pores or
fracture openings are small. In addition, the wettability of the waterlDNAPLIsolid system
influences the conditions under which a DNAPL enters a given permeable geologic
medium. We note that most DNAPLs of interest will be non-wetting on geologic solids
with respect to water, but wetting with respect to air. This typical behavior is observed
in both laboratory and field experiments involving pure chlorinated solvents. However,
because the wettability of a given system is a function of the surface chemistry of the
solid geologic material, the groundwater composition, and the DNAPL composition, each
of which may vary greatly at a site or from site to site, complex wettability relationships
may need to be taken into consideration at actual field sites.
For a DNAPL which is non-wetting on the geologic solids with respect to water,
water will coat the soil grains and occupy the smaller pores and pore throats: the DNAPL
is restricted to the larger pore openings (see Figure 2.3). In the less common situation
where the DNAPL is the wetting phase with respect to water, the situation is reversed;
the DNAPL coats the soil grains and occupies the smaller pores. When the DNAPL is
the non-wetting phase, capillary forces oppose the entry of the DNAPL into wet geologic
media.
The forces driving subsurface DNAPL movement are a function of the DNAPL
density and the pressures resulting from its release into the subsurface. As is discussed
in detail in Chapter 3, for DNAPL movement to occur in wet media, these driving forces
a.
b.
Non-wetting
DNAPL
Non-wetting
DNAPL
Wetting
DNAPL
Figure 2.3 Pore-scale representation of non-wetting and wetting DNAPL residual in:
a) water-saturated sand; and b) a fracture.
Sec. 2.2
59
must overcome the capillary resistance. The DNAPL in the larger pore openings must
deform to pass through smaller pore throats to reach other pore openings. The pressure
required for this deformational movement is the entry pressure. The value of the entry
pressure is proportional to the interfacial tension between the DNAPL and the water, and
inversely proportional to the size of the pore throats. Therefore, the entry of a DNAPL into
a fine-grained porous medium or into a fractured medium having small apertures requires
high driving forces to overcome high entry pressures. Consequently, low permeability
strata can be barriers to DNAPL migration. If the waterlDNAPL interfacial tension is
reduced, for example as a result of surfactant compounds in the groundwater, the DNAPL
can enter fine-grained media more easily. If the DNAPL is wetting with respect to water,
capillary forces will enhance rather than resist DNAPL entry into fine-grained media;
the DNAPL in that case moves preferentially into smaller pore openings.
As is discussed in detail in Chapters 3 and 11, the rate of flow of a DNAPL
through a geologic medium depends on: 1) the density and viscosity of the DNAPL;
2) the pressure driving the DNAPL migration; 3) the intrinsic permeability of the
geologic medium; and 4) the degree of DNAPL saturation of the pore space in the
medium. More permeable media and higher DNAPL saturations will permit higher rates
of DNAPL flow. Higher density and lower viscosity fluids also permit higher flow rates.
Most pure chlorinated solvents are much denser than water, and have viscosities lower
than water, so that in a given geologic medium with conditions of high DNAPL saturation,
most liquid chlorinated solvents can migrate at rates comparable to, or faster than, those
of water.
Nearly all movement of subsurface DNAPL occurs within zones of continuous
(i.e., connected) DNAPL. In such zones, the DNAPL in different pore openings forms
an immiscible-phase continuum through the intervening pore throats. Once the release
of DNAPL into the subsurface ceases, the forces driving DNAPL movement eventually
dissipate and the DNAPL in the pore openings become disconnected to form a zone
of residual DNAPL. Very high hydraulic gradients are then required in groundwater to
induce movement of this residual DNAPL.
We have thus far only discussed the presence of contaminants and their movement as
immiscible liquids. However, chlorinated solvents and other chemicals of interest have
tendencies to partition among all of the phases in the subsurface (DNAPL, air, water, and
geologic solids). Therefore, contaminant molecules in the subsurface will tend to partition to all local phases, seeking an equilibrium distribution condition. At equilibrium,
there is a defined contaminant concentration in each phase of the system. A partition
coefficient describes the relative abundance of a chemical between two adjacent phases
in equilibrium.
In the context of the subsurface fate and behavior of chlorinated solvent DNAPLs,
there are four phase partitioning processes of particular interest: 1) dissolution of DN APL
into water (as has already been mentioned); 2) volatilization of dissolved chemicals from
water into air; 3) vaporization of DNAPL into air; and 4) sorption of dissolved chemicals
60
Chap. 2
from water to solid surfaces. In general, the chlorinated solvents have vapor pressures
which are relatively high, and aqueous solubilities which are low in absolute terms.
However, even though the aqueous solubilities of these compounds allow them to persist
as DNAPLs for very long times in many hydrogeologic situations, their solubilities are
still four to seven orders of magnitude greater than their current drinking water quality
standard values. These factors combine to provide DNAPL chlorinated solvents with
exceptional potential for causing significant groundwater contamination. (See also the
related discussion in Chapter 1.)
Under non-equilibrium, dynamic conditions, the relative importances of the various phase transfer mechanisms are determined not only by the corresponding partition
coefficients, but also by the kinetic rates at which the various interphase mass transfers
can occur. These case-dependent interphase mass transfer rates have important implications for DNAPL flow, the ultimate distance of DNAPL penetration, and the overall
contaminant mass distribution in the subsurface. Indeed, as DNAPL flows through the
unsaturated zone, mass transfer from the DNAPL to both the air and water will reduce
the DNAPL phase volume and will tend to reduce the rate of DNAPL flow and as well as
the ultimate depth of migration. Overall, this diminishes the bulk volume of the medium
containing the immiscible liquid. That is, it increases the overall retention capacity of
the medium for the DNAPL.
In the vadose zone, mass transfer to the soil air occurs directly from the DNAPL
phase as well as from the aqueous phase following dissolution. Due to the high volatilities
of the common chlorinated solvents, mass transfer from the DNAPL phase to the soil
gas is expected to be the most significant phase transfer mechanism for DNAPLs in
the vadose zone. The vaporization and vadose zone transport of chlorinated solvents is
facilitated by the very high diffusion coefficients of compounds in the gas phase. In some
circumstances, the vaporization of a DNAPL causes rapid depletion of the DNAPL phase
and significant vapor plumes can form rather quickly.
In the interface zone and below the water table, partitioning into the gas phase
is minimal to non-existent. Dissolution of DNAPL is then the most important phase
transfer process. Once in the aqueous phase, dissolved chemicals can partition to the
solids. The front of the groundwater plume emanating from a DNAPL zone will travel
in the dominant direction of groundwater flow. In situations where the DNAPL source
is removed or isolated so that the DNAPL no longer contributes dissolved contaminants
to the plume, the back end of the plume will separate from the former source zone
and travel downgradient. In all cases, part of the contaminant mass in the plume will be
dissolved, and the remainder will be sorbed (i.e., adsorbed and/or absorbed). Partitioning
in the plume prevents contaminants from being in the dissolved phase all the time, with
the result that the front of the plume as well the main body of the plume advances more
slowly than the mean groundwater velocity.
The effects of sorption on plume behavior depend strongly on the type of geologic
medium and the identities of the dissolved chemical constituents that emanate from the
DNAPL. Plumes from solvent DNAPLs are only slightly retarded by sorption because
their sorption affinities for nearly all types of geologic media are weak. In contrast, plumes
from PCB DNAPLs rarely travel far because of the strong sorption of dissolved PCBs.
Sec. 2.3
61
However, DNAPLs that contain PCBs commonly also contain other contaminants such
as chlorinated benzenes that are less sorbed, and therefore form much more extensive
plumes.
The influence of sorption is generally greater in fine-grained porous media and in
fractured porous media (e.g., unfractured and fractured clayey or silty deposits) than in
granular aquifers. In fractured porous media, sorption in the low-permeability matrix is
generally much more important than sorption in the fractures. As a plume in a fractured
porous medium advances, sorption in the matrix maintains concentration gradients that are
directed from the fractures to the matrix. This helps maintain the transfer of contaminant
mass to the matrix. Sorption processes are considered in further detail in Chapter 8.
2.3 DNAPLS IN THE VADOSE ZONE
Studies at industrial sites show that, with few exceptions, DNAPLs will tend to penetrate
through the vadose zone into the groundwater zone where much of the DNAPL mass
then accumulates to cause persistent contamination. Nevertheless, at some sites, e.g. those
where the water table is exceptionally deep or where the volume of DNAPL released to
the soil is very small, it is possible that the entire mass of DNAPL will reside in the
vadose zone (Figure 2.4).
For a known volume of DNAPL released into a particular granular geologic medium, the depth of penetration is influenced by the area over which the release occurs, the
rate of the release, and the nature of the geological layering in the system. The natures of
the pore-size distributions in the layers depend largely on the type and scale of geologic
heterogeneity in the system. Poulsen and Kueper (1992) performed two experimental
field releases of tetrachloroethylene (PCE), six liters each, into the vadose zone of a
slightly-stratified beach sand aquifer. One release was an instantaneous spill to 0.1 m 2
on bare ground. This release caused PCE to penetrate to a depth of 2.15 m. In the second
release, the six liters of PCE were dripped onto a soil surface area of I cm 2 during 100
minutes so as to simulate a very slow, point-source leak. The PCE from the second spill
penetrated through the vadose zone to the water table at a depth of 3.3 m, leaving a trail
of DNAPL residual all the way from ground surface to the water table. The drip release
penetrated deeper because the surface area of the input was much smaller. For both
spills, the DNAPL migration was strongly influenced by the small (millimeter) scale of
the stratifications in the sand. The DNAPL migrated preferentially along coarser-grained
layers, and did not enter fine-grained layers.
Laboratory column experiments using "homogeneous" sand indicate immisciblephase retention capacities for solvents in dry homogeneous sands in the range of 3 to
30 L'm', on average, or I to 10% of the pore space (SchwiIIe, 1988). For the moist
stratified vadose-zone sands considered in the field experiments of Poulsen and Kueper
(1992), retention capacities of the DNAPL-containing layers were in the range of 2 to 18%
of the pore space. These values are similar to those measured for petroleum hydrocarbons
in unsaturated sands (Wilson and Conrad, 1984). In assessing the movement and fate of
Chap. 2
62
:-i .: DNAPL
: .. '.'.< RESIDUAL
capillary fringe
.: . H:
z.: - ~Cu~i-~~ -~-~-~
......FLOW
( b)
". 'VAPOR' . .
.' '.'.:. " . t+ ... H t i

:-:-~~~~-:-~~~~-:~.--:,---..--;.-.-: -c- -s;
. . . PLUME..
(d)
--.
--
...-
.5......: -
/impervious cover
. . TRAPPED..
. . . VAPOR'.
'.
.'
'.
..
-:Water Tcible flUctUation ~one ~.

---,--- -:--. ----,-- -.- .-. ---:- --:- .-. -.-
-:-...L..;;;--:~
PLUME
Figure 2.4 Conceptual scenarios for a DNAPL in the vadose zone in granular geologic deposits:
a) homogeneous case - no vapor plume; b) heterogeneous case - no vapor plume; c) heterogeneous
case - vapor plume; and d) effect of impervious ground cover over vapor-releasing DNAPL source.
a DNAPL at a spill or disposal site, predictions of the depth of penetration based on

the DNAPL volume, the area of release, and estimates of the retention capacity of the
geologic medium beneath the spill can be useful. However, because of the influence of
site-specific geologic heterogeneities on DNAPL migration and thus on the distribution of
residual DNAPL, depth predictions are commonly so unreliable that they are misleading
or even useless.
In the experiments conducted by Poulsen and Kueper (1992), small stratifications
caused the downward-moving DNAPL to be distributed in a complex manner; they did not
serve as significant impediments to downward penetration. Impediment can nevertheless
occur where lower-permeability layers of sufficient size and permeability contrast exist in
the vadose zone. Figures 2.4.b and 2.4.c show that such layers can cause perched zones
of free-phase DNAPL in the vadose zone. DNAPL accumulation will not take place in
near-surface deposits when erosional or depositional gaps, fractures, root holes, worm
holes, and/or animal burrows exist through low permeability layers.
Because the retention capacities of most porous media are relatively small in absolute terms, and because of the channels that exist through many large clay and silt
layers, at many sites where only relatively "small" volumes of solvent were released to
the soil, we can conclude that DNAPL solvent has penetrated to considerable depths.
Sec. 2.3
63
There is controversy over what numerical quantities should be associated with a likelihood of penetration. As a general rule, we propose that at sites where the water table
exists at depths of a few tens of meters or less, and where there are no geologic strata
with exceptional capability for impeding DNAPL penetration, a solvent DNAPL release
of a few tens of liters at a single location should be considered capable of permitting
DNAPL to enter the groundwater zone. If the water table is much deeper, penetration of
the groundwater zone may require a release of a few hundred liters, i.e., a few drums of
solvent. However, repeated small releases at precisely the same location can also produce deep penetration. Indeed, we note that significant accumulation of DNAPL below
the water table can occur at sites as a result of storage and handling activities where
only small amounts of solvent are released at any single location, but the releases are
repeated.

In Figures 2A.a and 2A.b, the DNAPL in the vadose zone has a very low vapor pressure
and therefore causes no vapor plume. In these cases, a groundwater plume results only
from leaching of the residual DNAPL when water infiltrates through it; if infiltration
is prevented by an impervious surface cover such as parking-lot asphalt or a building,
little groundwater contamination can form. In Figure 2A.c, the DNAPL is volatile, and a
vapor plume emanates from the source zone. Indeed, as has been noted above, most liquid chlorinated solvents have high vapor pressures, and therefore where solvent DNAPL
exists in the vadose zone, a vapor plume will develop in the surrounding soil air. Infiltrating water that contacts the vapor will cause the development of a groundwater plume.
Infiltration through the residual zone will also cause groundwater contamination. Within
a period of weeks or months, a small mass of solvent DNAPL in the vadose zone can
produce a vapor plume in a sandy medium that has grown, by means of gas phase
molecular diffusion, to sizes extending to a few tens of meters. After many years or
decades, vapor can move downward through very thick vadose zones to depths of many
tens of meters or more. For high vapor pressure chemicals that also have large vapor
densities, density-driven advection in high permeability media can enhance downward
movement of vapor to the water table (Mendoza and Frind, I990a, 1990b).
The conceptual model for a high-vapor pressure DNAPL source zone covered with
an impervious barrier is given in Figure 2A.d. Even though no water infiltrates through
the DNAPL residual or through the vapor, a groundwater plume is caused by fluctuations
in the water table that trap vapor in the water table zone. Also, the diffusion of solvent
vapor to and across the capillary fringe contributes solvent mass to the groundwater zone.
In a field experiment in an unconfined sandy aquifer, Hughes et at. (1992) placed
a 0.5 cubic meter volume of sand mixed with trichloroethylene (TCE) at residual levels
in a one meter deep hole in the 3.5 m thick vadose zone. This DNAPL-laden sand was
removed by excavation after a month. Prior to the excavation, a vapor plume emanated
rapidly outwards IO m, and downwards to the water table. The experiment was repeated
half a year later. After three years, the vapor plumes caused by the temporarily-emplaced
TCE sources ultimately caused a thin (1-2 m), shallow plume, 40 m wide and more than
64
Chap. 2
130 m long. Maximum concentrations in the plume were in the thousands of f.LglL range.
The front of the groundwater plume was captured by pump-and-treat (Rivett, 1994).
Thin shallow plumes caused by vapor transport are difficult to locate when conventional approaches to aquifer monitoring are used. Conventional monitoring wells
generally have screened intervals of two to five meters, and draw most or all of their
water from depths below the water table. The behavior of vapor in the vadose zone is
discussed further in Chapter 6; implications for monitoring are discussed in Chapter 13.
2.4 DNAPLS IN THE GROUNDWATER ZONE

Much of the groundwater supply in North America and Europe comes from sand or
gravel aquifers, and many large plumes of chlorinated solvent contamination occur in
such aquifers. Some of these plumes extend for many kilometers from the industrial sites
where they originate. Almost without exception, these plumes result from the dissolution
of persistent DNAPL sources situated below the water table.
Figure 2.5.a shows the result of a release of DNAPL small enough to cause penetration below the water table, but insufficient for DNAPL to reach the bottom of the
aquifer. The distribution of the DNAPL below the water table is affected by the geologic
layering present. In granular aquifers, small horizontal zones of residual or free-phase
DNAPL need not be caused by particularly low permeability zones such as silt or clay.
A minor contrast in grain size distribution and hence permeability, as from a coarse sand
layer to a finer sand, causes variation in DNAPL entry pressure (Kueper et at. 1989; see
also Chapter 3). A DNAPL moving downwards through a coarse-grained material will
encounter a higher entry pressure when a finer-grained layer is contacted. This will cause
lateral spreading of the DNAPL. The DNAPL will accumulate on the finer-grained layer
while spreading laterally until it reaches the edge of the layer (Figure 2.5.c), or until the
height of free-product accumulation on the layer exceeds the entry pressure for the layer.
In the latter case, the DNAPL will pass through the layer and continue its downward
movement towards the bottom of the aquifer.
The cases shown in Figures 2.5.b and 2.5.c have much of the DNAPL mass
in the aquifer offset horizontally from the DNAPL release point. In these particular
cases, the DNAPL mass in the groundwater zone is situated upgradient or downgradient
from the DNAPL release point. Given the subtlety of geologic heterogeneity that can
cause a deflection of DNAPL in the saturated zone, as well as the complexity of the
spatial distribution of such geologic heterogeneity in most systems, DNAPL pathways
are generally unpredictable in the saturated zone (see Section 2.3.1 for similar comments about the vadose zone), even when considerable information on the stratigraphy
of the subsurface environment is available. Indeed, although site investigations normally
provide information on the macrostratigraphy, as has been noted above, the microstratigraphy can have an important or even dominant influence on DNAPL pathways.
The plumes of dissolved contamination shown in Figures 2.5.a and 2.5.b occupy
only the upper part of the aquifer because the residual and perched DNAPL exists only in
65
Sec. 2.4
(c)
.--
FLOW
AQUIFER
(b)
(d)
-~-
:'>:---:-. PLUME'
"
.:. '~:.:';.' _:.-.~ ~ . : .:.....::.:.. :.:-:;~.:.~.ij;H-fitM.
.'
.. _---_ ... ----------
----~-
-------~
--
\.~ .. : :.::/. ':., .'.<;(,ji/
,-
PLUMES
~.,
'-.
'-
-,
--
<,
"
\
/
...'il. ...
J
\
<,
<,
'-
Figure 2.5 Conceptual scenarios for a DNAPL in the groundwater zone in granular aquifers: a)
partial penetration; b) partial penetration with offset; c) full penetration with offset; and d) same
as part c, but at a later stage after DNAPL residual has disappeared due to dissolution in flowing
groundwater.
the upper portion of the aquifer. Vapor in the vadose zone contributes contaminant mass
only to the very shallow part of the plume. Locating the interface between the portion of
the plume contamination that originated in the vadose-zone with the portion originating
in the groundwater zone is difficult to impossible.
The volume of DNAPL released in the case shown in Figure 2.5.c is larger than
in Figures 2.5.a and 2.5.b, causing the DNAPL to penetrate to the bottom of the aquifer.
Therefore, in Figure 2.5.c the plume of dissolved contamination spans all depths in
the aquifer at some locations because DNAPL residual is present at all depths. An
accumulation of free-phase DNAPL occurs in a depression on the aquitard surface. This
accumulation or "pool" is a major feature of Figure 2.5.c. In field investigations of
sites where extensive solvent contamination exists, pools of free-product solvent are
found only rarely, even when their existence is not in doubt. Such accumulation zones
are rarely found because conventional site investigation techniques are not well-suited
for their detection, and because drilling and sampling the large number of boreholes
necessary to encounter small to medium size pools is usually not practical.
A considerable volume of free-phase DNAPL can occur in a relatively small depression on the aquitard surface. For example, a spill of 800 liters of DNAPL could be
66
Chap. 2
contained in a pool at the bottom of a sand aquifer, with the pool measuring only 3 m
x 3 m in area and 0.3 m thick. A grid of boreholes with exceptionally close spacing,
such as 50 m, would have only a low probability of detecting such a pool. Even a pool
occupying a much larger depression on the aquitard surface would have a low probability
of being detected in most site investigations because of small pool dimensions relative to
borehole spacing, and because of the difficulty in detecting thin DNAPL accumulations
during drilling and sampling.
The few data available on residual contents of chlorinated solvents in saturated
granular materials indicate values that are higher than those for unsaturated materials
(Schwille, 1988). The values for saturated soils range from 5 to 50 Llm 3 , or 2 to 15%
of the pore space). Although the presence of residual solvent DNAPL in the pore space
reduces the permeability of the aquifer only slightly, in a pool, the DNAPL can occupy
40 to 70% of the bulk pore space, will preferentially occupy the larger pores, and will
therefore cause the permeability in such zones to be reduced substantially: groundwater
flow through these zones will occur only very slowly.
As is discussed in detail in Chapter 7, even though flowing groundwater causes
dissolution of DNAPL solvents, chlorinated solvents in residual zones and in pools do
not rapidly disappear from a groundwater system. Consider, for example, a 3 m cube
of sandy aquifer materials that contains TCE at residual levels. Within the cube, the
residual level varies from 0 to 15% of the pore space; the average residual content will
be assumed to be 10 L'rrr'. This corresponds to about 270 liters, or 1.5 drums of TCE.
The linear groundwater velocity will be assumed to be 53 m/year, i.e., the rate that would
be found for a sandy aquifer with a hydraulic conductivity of 10- 2 cm/s, a porosity of 0.3,
and under the influence of a hydraulic gradient of 0.005. The groundwater exiting such
a variable residual zone would likely have dissolved concentrations ranging from low
levels up to the solubility limit. Even if the average TCE concentration in groundwater
exiting this residual zone was only 10% of the TCE solubility (110,000 IJ,g/L), and this
is higher than what is typically found at actual field sites, 25 or more years would need
to pass before the residual DNAPL was completely dissolved and thereby removed as a
source zone. Moreover, based on laboratory experiments (Schwille, 1988; Imhoff et al.,
1989), the actual time required for complete dissolution of this residual DNAPL can be
expected to be longer than 25 years because the dissolved concentrations resulting from
dissolution and consequently the mass rate of dissolution both tend to decline with time
as a residual DNAPL mass is depleted.
For a pool of DNAPL, one can expect that the time required for the complete dissolution by natural groundwater flushing of a given mass of DNAPL will be much longer
than for the complete dissolution of the same DNAPL mass distributed in residual form.
The low relative permeability to water in a pool zone limits the groundwater flux through
such a zone. Also, the DNAPL content is higher in a pool than in a residual zone. Based
on the results of a laboratory experiment by Schwille (1988) and the analysis by Johnson
and Pankow (1992), for a pool of TCE in a sandy aquifer in which the groundwater is
moving with a linear groundwater velocity of 275 m/year, then a 1.5 m long, 0.5 m wide,
and 0.2 m thick pool containing only 50 L of TCE would require at least 100 years to
be dissolved completely (see also Chapter 7). (We note here that this hypothetical pool
Sec. 2.4
67
is much smaller than pool sizes expected at many sites. Also, the assumed velocity of
275 mlyear is relatively high.) Such long source lifetimes are consistent with common
field observations which indicate that severe groundwater contamination from subsurface DNAPL sources continues to be generated at sites where the chemical release(s) or
disposal(s) occurred decades ago, and in some cases more than 50 years ago.
Figure 2.5.d pertains to the same case as Figure 2.5.c, but at a much later time when
the zones of residual DNAPL below the water table have disappeared due to groundwater
flushing. The zone of DNAPL in the vadose zone has disappeared due to vapor loss
and infiltration leaching. Only the perched and bottom pool zones below the water table
remain. The plume of dissolved contamination now derives only from these accumulation
zones, and therefore the plume at this stage is much different than the one generated at
the early stages when the DNAPL residual zones also contributed dissolved contaminants
to the plume. At many sites where plumes of solvent contamination are observed, the
DNAPL entered the aquifers long ago. At some sites, then, one can expect that longterm dissolution has caused disappearance of many of the residual zones, leaving only the
larger residual zones and the accumulation zones to cause continued plume generation.
Many DNAPLs are mixtures of several chemical compounds. For example, a
solvent DNAPL could be an equal mixture of TCE, PCE, and dichloromethane (DCM).
These three compounds have very different solubilities in water (DCM > TCE > PCE).
For this mixture, the high solubility of the DCM causes it to be dissolved first, followed
by the TCE, then followed by the PCE. The chemical composition of such a DNAPL
and the plume would change as this occurs. Related discussions for DNAPLs which are
mixtures can be found in Chapters 7 and 13.
As mentioned above, at many DNAPL sites, where a thin overburden covers fractured
bedrock, DNAPL has moved through the overburden and into the fractures (Figure 2.2).
At some sites, the DNAPL has moved through thick overburden into fractured bedrock.
DNAPLs like the chlorinated solvents that have high densities, low waterlDNAPL interfacial tensions, and low viscosities, have exceptional propensity to penetrate into
small fractures in bedrock and into fractures in other geologic media such as clayey
deposits.
The pattern of DNAPL movement and ultimate distribution in fractured geologic
media is controlled primarily by the orientation and interconnection of the fractures.
Figure 2.6.a shows fractured non-porous rock that has horizontal fractures along bedding
planes, or along stress relief planes. Vertical fractures connect the horizontal fractures.
DNAPL moves downward through the vertical fractures and laterally along the horizontal
fractures. As it moves along the horizontal fractures, the DNAPL can encounter more
vertical fractures that allow downward movement to continue. The horizontal fractures,
like all fractures, have irregular surfaces. Hence, fracture apertures vary along the length
and across the plane of the fracture. The horizontal fractures therefore provide small
but numerous depressions for free-phase DNAPL to exist as small, disconnected freephase accumulations. This situation is accentuated in cavernous or karstic rock where
68
~ DNAP~ ~ELEASE
(0) ~
(c)
,jL"~""'"
~--:-.~'~'.~.,:
I
I
'I
I
I
I
"J
"T
Chap. 2
-:-- -; ~OVERBUR?EN:
"T
I 1
1
:::I
-,
.,.
-/
(b)
Figure 2.6 Conceptual scenarios for a DNAPL in the groundwater zone in fractured non-porous
media: a) regular bedded rock with bedding plane fractures and connecting vertical fractures;
b) solution-channelled carbonate rock with free-phase DNAPL accumulations; c) layered rock with
contrasting fracture networks; and d) complex fracture network in a single rock type.
large amounts of DNAPL can accumulate in the depressions of major flow channels or
cavities (see Figure 2.6.b).
Bedrock typically has permeable zones separated vertically by less permeable zones
in which vertical fractures are much less frequent. Such a case is represented in Figure 2.5.c where a DNAPL exists in numerous fractures in the upper fractured zone.
From this zone, DNAPL has drained through sparse vertical fractures into the lower
fractured zone where it spreads along a different set of fractures. In site investigations,
detection of DNAPL in the more permeable zones could be feasible, whereas detection in
the intermediate zone where fractures are sparse is unlikely. In the deeper fractured zone,
fewer fractures contain DNAPL and therefore the dissolved concentrations observed in
monitoring wells downgradient from this DNAPL zone would be expected to be much
lower than those in the shallow plume, even though both aquifers have DNAPL in fractures. Interpretation of dissolved concentrations from monitoring wells in fractured rock
is typically very difficult.
Figure 2.6.d shows DNAPL penetration into a much less ordered fracture network,
i.e., of the type commonly observed in crystalline rock such as granite. DNAPL enters
fractures at the bottom of the overburden. As in the cases already considered, these
Sec. 2.4
69
fractures act as conduits to other fractures. Some of these fractures are dead ends for
DNAPL, either because they do not intersect other fractures, or because they intersect
fractures that are too small for DNAPL entry.
Schwille (1988) conducted laboratory experiments using rough-walled glass plates
to represent planar vertical fractures above and below the water table. From experiments
with 200 {Lm apertures, he estimated the retention capacity (residual retention content)
of such fractures to be less than 0.05 L'rrr'. For a fractured rock of moderate hydraulic
conductivity having a three-dimensional fracture network with fracture frequency of one
fracture per meter and fracture apertures of 200 {Lm, this represents a residual content,
expressed relative to the bulk rock volume, of approximately 0.15 L'm'. Although this is
large relative to the total fracture porosity, it is much smaller than the retention capacity
expected for granular aquifers which have a bulk granular porosity that is two to four
orders of magnitude larger than the fracture porosity of most bedrock. Therefore, for a
given volume of chlorinated solvent released to fractured rock, the solvent can spread to
occupy a volume of the subsurface domain that is much greater than in granular aquifers
or other porous geologic media. Indeed, in the previous example for a sand aquifer, an
800 liter pool of DNAPL was calculated to occupy a volume of aquifer measuring 3 m
x 3 m in area and 0.3 m in thickness. The same volume of DNAPL in fractured rock
with a bulk retention capacity in the fracture network of 0.15 Llm 3 would occupy a
zone measuring 130 m by 130 m with the same 0.3 m thickness. This could suggest that
DNAPL accumulation zones in fractured rock are much more readily detected because of
their size than those in granular aquifers. However, in fractured rock, it can be difficult
to detect free product or residual DNAPL even where boreholes penetrate such zones.
Often, so little free product drains into wells from fractures or is flushed to the surface
during drilling that its presence goes unnoticed.
The maximum depth to which a DNAPL penetrates in fracture networks depends
on several factors, including the fracture aperture, the number and type of fracture connections, the physical properties of the DNAPL, and the height of the column of continuous
DNAPL above the fractures at the front of the DNAPL zone. At some fractured-rock
sites, DNAPL is known to have penetrated to depths of several hundred meters and
more. At many fractured rock sites, investigations have not yet conclusively established
the maximum depth of DNAPL penetration. In some cases, the maximum depth is likely
greater than 1000 m. If the nature of the release and fracture network geometry are such
that a large column of continuous DNAPL forms, the propensity to travel through to great
depths is large, thereby causing deep penetration even where the fracture apertures are
very small. It is generally not possible to predict the maximum depth of DNAPL penetration at fractured rock sites even where there is exceptionally-detailed information on the
geology, groundwater flow, and fracture network properties because the fracture-specific
data for such predictions generally cannot be obtained from our limited investigative
methods. Attempts at measuring fracture apertures generally provides inaccurate values,
or the measured values reflect apertures altered due to the test methods. The maximum
depth of penetration must be inferred from monitoring of the dissolved contaminant distribution. Chapter 11 describes the physics of DNAPL behavior in fractured media in
further detail.
70
Chap. 2
In fractured porous media, the relatively large volume of voids in the matrix between the
fractures influences: 1) the migration of DNAPL and of the dissolved-plume; 2) the persistence of the DNAPL phase; and 3) the design of remediation efforts. As in fractured
non-porous media, groundwater in the fractures causes DNAPL to gradually dissolve.
However, as has been noted above, in fractured porous media the dissolved-phase mass
can also enter the water in the porous matrix, and can sorb to the matrix solids. For
the most soluble DNAPLs, such as the common chlorinated solvents, mass loss from
the fractures to the matrix can eventually cause complete disappearance of the DNAPL
from some or all fractures, particularly in situations where the thickness of DNAPL in
each fracture is small and the fracture spacing and matrix porosity are appreciable (see
Figure 2.7). Thus, dissolution combined with diffusion causes a change in the physical
state of the solvent mass, from the DNAPL phase to the dissolved and sorbed phases.
As has been discussed by Parker et al. (1994), this change of state was recognized only
recently. The implications of this disappearance and the rates at which it can take place
are subjects of current research. In fracture networks in very porous geologic deposits
such as unconsolidated silt or clay, the disappearance of solvent DNAPL may occur
within days or weeks or, in some types of porous sedimentary rocks such as sandstone
or shale, within months to several years. Chapter 12 provides more details regarding this
process.
In fractured non-porous media, dissolution of DNAPL solvent feeds contaminant
mass to the plume as the plume travels along the fracture network. However, once the
DNAPL source mass is totally dissolved, groundwater flushes the remaining dissolved
and sorbed contaminant mass out of the fractures in the former source zone relatively
rapidly. In contrast, in fractured porous media, because significant contaminant mass
has dissolved and diffused into the matrix, contaminant mass can persist in the former
DNAPL zone long after the DNAPL phase has disappeared. This mass will diffuse back
out of the matrix when clean water begins to flush through the fracture network.
As noted above, once DNAPL has entered fractures in rock with a non-porous matrix, the propensity for deep penetration is high because the height of free-phase DNAPL
accumulation in the fractures adds to the effective pool height. The DNAPL can then
move downward into smaller and smaller fractures as the effective pool height increases.
In fractured porous media, however, the transfer of DNAPL contaminant mass from fractures to the matrix diminishes the propensity for downward penetration because loss of
mass from the DNAPL diminishes the volume of DNAPL. Given certain fracture conditions, diffusion can also cause a disconnection of the DNAPL phase, further reducing
the height of the free-phase DNAPL and hence reducing downward DNAPL movement
(see Figure 2.8).
In fractured non-porous media, the total pore space in which contaminant mass
can reside is measured by the fracture porosity. In contrast, the total volume ultimately
available for contaminant mass in fractured porous media is the total of the fracture
porosity plus the matrix porosity. In situations where entry of DNAPL to the fracture
network continues (i.e., DNAPL release replenishes the DNAPL in the fractures), the
Sec. 2.4
71
FRACTURE NETWORK
IN A POROUS MEDIUM
TIME
early
___L
DNAPL
RESIDUAL
DNAPL
late
intermediate
--
~;;;
~
POOL~in~~~~~tJ=JT-H\
INSERT
FRACTURE APERTURE
Figure 2.7 Conceptualization of DNAPL persistence and distribution with DNAPL mass loss
due to diffusion in water-saturated fractures in a porous medium (such as a fractured clay till or
sedimentary rock) as a function of: a) time; and b) fracture size (aperture). Diffusion halos around
fractures containing DNAPL become large (more developed) with increases in time, as shown with
the concentration vs. distance plots for the three relative time periods in part a. At any particular time
tl, the DNAPL mass distribution will vary in different size fractures, with DNAPL disappearance
occurring most rapidly in the smallest fractures; residual or disconnected DNAPL will be present in
medium-sized fractures, and free DNAPL will be present in the largest fractures that are connected
to overlying DNAPL pools. DNAPL disappearance times (tn) increase with an increase with fracture
aperture.
maximum storage capacity is attained when all of the pore water in the matrix is at the
solubility limit.

There are several industrialized regions in the U.S., such as the Santa Clara ("Silicon")
Valley in California and parts of the Gulf Coast, where solvent production, usage, and/or
disposal have been large, and where the geology comprises layered sequences of uncon-
72
Chap. 2
DNAPL
1
WATER-SATURATED
MATRIX
FRACTURE
Figure 2.8 The effects of diffusive mass loss on DNAPL invasion distances and resultant height
of interconnected DNAPL in a single, water-saturated fracture in a porous medium: a) no diffusive
mass transfer and DNAPL pool height Hb; b) reduced DNAPL invasion distance due to diffusion
into the porous matrix adjacent to the fracture with a subsequent reduction in DNAPL mass and in
the interconnected pool height H~; and c) reduced invasion distance due to diffusive mass loss from
DNAPL in the fracture and disconnection of DNAPL phase during flow causing further reduction
in interconnected DNAPL heights H'Jy and H'h.
solidated sandy aquifers separated by clayey or silty aquitards. The thickness of aquifers
and aquitards is variable, generally from a few meters to several tens of meters. On a
regional scale, many of the strata are not extensive. However, at many sites, the layers
are continuous across the "source zone" of DNAPL. Immense financial and technical resources have been invested for the study and remediation of solvent sites in these layered
sedimentary environments.
Prior to the mid- to late-1980s, it was commonly thought that clayey or silty
aquitards situated beneath aquifers would prevent DNAPLs from penetrating below the
uppermost aquifer. However, field data from countless sites now show that many (if not
most) silty or clayey aquitards allow DNAPL to move through them, thereby causing
contamination of underlying aquifers, At some sites, solvent DNAPL has penetrated
through several aquitards to cause contamination much deeper than was believed possible
a few years ago,
Penetration of DNAPL through aquitards may occur through one or more of the
following types of vertical pathways: I) unsealed or improperly-sealed boreholes;
2) disposal wells; 3) stratigraphic windows; and 4) open fractures or faults, The first
three types of pathways have been recognized for many years. The origin and occurrence
of open fractures in aquitards is an emerging research topic. Deep, open fractures are
now known to be common in many clayey aquitards. The intuitive notion that fractures in
unconsolidated or semi-consolidated silty or clayey deposits are closed due to overburden
Sec. 2.4
73
stress is generally not valid, except perhaps at very great depths or in very plastic clayey
deposits.
An accumulation of DNAPL on top of a fractured aquitard allows several possible
outcomes. Figure 2.9.a shows a static DNAPL pool resting on the top of the fractured
aquitard with no entry of DNAPL into fractures because the fracture entry pressure is not
_ _ _ _ .5Z
_ _ _ _ -.5.2
_ _ _ _ ..52
____ .9
-
DNAPl RESIDUAL
_
_
-
Figure 2.9 Conceptual scenarios for a

DNAPL accumulation zone on a fractured
aquitard: a) height of free-phase DNAPL
insufficient for DNAPL entry into fractures;
b) only largest fractures allow DNAPL
entry; and c) DNAPL entry into all
fractures.
74
Chap. 2
achieved at the bottom of the pool. Figure 2.9.b shows the same situation, except that
a few of the fractures are larger, which allow DNAPL entry into them. In Figure 2.9.c,
all of the fractures are large enough for DNAPL entry, which is a less likely situation
because it is probable that entry into the largest fractures and subsequent DNAPL flow
in these fractures would limit build up of the pool heights, thereby preventing sufficient
height to develop for entry into the smaller fractures. Given that fracture networks in
aquitards are generally difficult to delineate, and that it is expected that not all fractures
contain DNAPL, locating those particular fractures that contain DNAPL is very difficult.
Figure 2.10 shows four scenarios in which DNAPL enters into, and in some cases
travels through, an aquitard. In Figure 2.1O.a, the fractures do not penetrate to the bottom
of the aquitard and therefore the aquitard protects the aquifer from actual DNAPL entry.
Figures 2.10.b and 2.IO.c show aquitards that have a few fractures that extend to the
bottom of the aquitard, providing pathways for DNAPL entry to the underlying aquifers.
In Figure 2.IO.d, the DNAPL pathway through the aquitard is a stratigraphic window
rather than fractures. At some sites, it is not possible to discern from the available field
DNAPL
CLAY
SAND
(b)
(d)
......
------_
_ce=.---:o-::::I
SAND
..
:::::'~".-.': :-~.:... :.~.".
-,
-
Figure 2.10 Conceptual scenarios for DNAPL movement through clayey aquitards: a) partial
penetration because open fractures do not extend to the bottom of the clay deposits; b) penetration
through a few deep fractures causes DNAPL accumulation in the underlying aquifer; c) penetration
through fractures in an aquitard between two sand aquifers; and d) a stratigraphic window in the
aquitard provides a pathway for DNAPL flow.
Sec. 2.5
Plumes from DNAPLs
75
data whether the cause of DNAPL contamination in an underlying aquifer is fractures or

such windows.
The capacity of fractured clayey or silty aquitards to retain DNAPL material is
small because the fracture apertures are small, generally less than 100 lim for fractures
below the water table. Even if DNAPL dissolution causes the matrix pore water to reach
the solubility limit, the mass retention capacity is relatively small, ranging from the liquid
equivalent of less than 5 to 10 Llm3 for the most soluble chlorinated solvent (DCM) down
to 0.05 to 0.5 Llm3 for PCE, which is a low-solubility solvent. These calculations take
into consideration what would be considered typical sorption properties of aquitard solids.
Therefore, this low mass retention capacity and the small thickness of many aquitards
reduces the chances that such a fractured aquitard will be depth-limiting with respect to
the contamination. In the underlying aquifer, however, geologic heterogeneity will permit
horizontal DNAPL accumulations (perched pools) at various depths; this provides much
larger retention than in the aquitard.
The actual maximum depth of DNAPL penetration in a given system will be
attained when the driving forces for DNAPL flow are balanced by the forces resisting
flow. This can take place anywhere within a layered heterogeneous system, depending on
the volume of DNAPL released, the nature of the DNAPL, and the various sequences of
geologic conditions encountered along the pathway. At sites where multiple aquifers and
aquitards exist, there is usually no reliable basis for predicting the maximum depth of
DNAPL penetration, even when release volume, geologic, geophysical, and stratigraphic
data are available. Direct monitoring of the presence of DNAPL, and/or inferences of
such a presence based on dissolved concentrations are necessary for conclusions about
the depth of penetration to be made.
As is discussed in detail in Chapter 10, there is debate in the literature about
whether DNAPL accumulations on clayey strata can cause vertical pathways through
these strata to open up due to chemical desiccation of the clay minerals. In particular,
the desiccation is believed by some to cause bulk shrinkage and thus formation of open
fractures. However, the mineralogical properties of natural clayey strata and the chemical
properties of the common chlorinated solvent DNAPLs make this chemical desiccation
scenario unlikely, except possibly in rare specific circumstances.
2.5 PLUMES FROM DNAPLS

Plumes of dissolved contamination at DNAPL sites exhibit various shapes and internal
concentration distributions that depend on the aquifer characteristics and the morphology
and extent of the subsurface DNAPL source zone. Figure 2.11 shows two hypothetical
cases for sand or gravel aquifers. One of these cases assumes that hydrodynamic dispersion causes considerable lateral spreading; the other assumes little spreading because of
weak dispersion.
The influence of dispersion on plumes from DNAPL source zones is no different
from its influence on plumes of other types of dissolved chemicals, whether they be
76

"... "
"...
'. " :
Chap. 2
"'\
STRONGDISPERSION '. '. :' : '. ~ FAN SHAPED PLUME
WEAK D1SPERSIONNARROW PLUME
Figure 2.11 Influence of hydrodynamic dispersion on plumes from a DNAPL source

zone in a sand or gravel aquifer with uniform steady flow. The plume is long and narrow
for the weak dispersion case, and fan-shaped for the strong dispersion case.
organic or inorganic. Studies of dispersion in numerous sand and gravel aquifers in

the 1980s showed that, almost without exception, dispersion is weak, particularly in
the directions transverse to groundwater flow. In the fan-shaped plume of Figure 2.11,
transverse and longitudinal dispersion causes considerable decline in concentration as
the plume spreads in the aquifer with groundwater flow. In the narrow plume, on the
other hand, rate of concentration decline with distance in the direction of flow is much
smaller.
The type of nearly uniform plume width found in Figure 2.11 is expected in aquifers
where the groundwater flow is relatively uniform, devoid of diverging or converging
flowpaths. Because dispersion is typically weak, many solvent plumes emanating from
DNAPL source zones exhibit high concentrations far from the source zone. And, since
the maximum contaminant levels (MCLs) specified in drinking water standards are very
low, concentrations far from the source can be much above these MCLs. Although plume
concentrations for chlorinated solvents and other DNAPLs are large relative to drinking
water standards, they are not large enough to cause the plumes to sink due to density
of the aqueous solution. Density is an important factor at DNAPL sites, but only with
respect to movement of the DNAPL and not the migration of plumes.
The shape and size of a source zone has a strong influence on plume dimensions.
Thus, although many solvent plumes in granular aquifers are long and narrow, many
others are wide because their DNAPL source zones are wide. The details of the distribution of DNAPL residual and accumulation zones below the water table are rarely known.
However, at most sites, it is expected that they are complex due to geologic heterogeneity
and variable DNAPL releases. Downgradient of such DNAPL source zones, weak dispersion allows the distribution of dissolved concentrations to be highly variable spatially,
particularly close to the source zone. Farther downgradient, dispersion causes variability
of the concentration profiles to diminish. Figure 2.12 shows a schematic example of this
dispersion effect. The actual distance at which the concentration profiles become less
irregular depends on site conditions.
Little is known about dispersion in fractured geologic media. Although there are
many large contaminant plumes in fractured-rock aquifers, none has been monitored in
Sec. 2.5
77
Plumes from DNAPLs
DNAPL
SOURCE ZONE
PLUME
DISTANCE
Figure 2.12 Influence of dispersion on saturated-zone concentration distribution in a

plume caused by a complex DNAPL source zone.
sufficient detail to provide much information on dispersion. The generalizations presented

above for granular media are probably not relevant to fractured media.
In low permeability zones, molecular diffusion in the aqueous phase governs the
migration of dissolved contaminants. Where DNAPL accumulations rest on unfractured
clayey or silty layers, diffusion of dissolved contaminants occurs into these layers (Figure 2.13). The rate of penetration of the front is slow for chlorinated solvents, less
than 1 m in the first decade, and decreases with time. Therefore, molecular diffusion is
generally not the cause of significant breakthrough of contaminants through aquitards.
Unfractured low-permeability aquitards of appreciable thickness can therefore provide
long-term protection of aquifers from contamination.
Many granular aquifers have numerous thin interbeds or lenses of silt or clay. In
these aquifers, advection (bulk groundwater flow) advances the plume front and governs
the flux of contaminant mass in the flow direction. At the same time, diffusion causes
loss of mass from the permeable zones to the low-permeability layers in the aquifer.
The low permeability of these zones combined with sorption to the layer solids can
temporarily immobilize contaminants. When a plume is advancing through the aquifer,
the low-permeability layers act as a sink for contaminant mass. However, at later time
when the DNAPL source zone (or parts of this zone) is depleted by dissolution or has
been isolated by engineered enclosures or hydraulic containment, complete flushing of
the plume from the aquifer is delayed because of diffusive release from the layers.
78
Chap. 2
TeE
mg/L
1000
,0
./'
-7
:f?/
I ~ ....
CLAY-
2
(m)
/~o
3 /
DNAPL
POOL
CLAY~ ~
AQUITARD
4
5
Figure 2.13 Diffusion profiles for dissolved TCE in an unfractured clay deposit beneath
a free-phase TCE accumulation. TCE solubility is 1100 mg/L; there is no sorption.
Loss of dissolved contaminants from a plume into low-permeability layers is shown

schematically in Figure 2.14. The concentration profiles for TCE in one of these layers
assuming a porosity of 0.3 and a retardation factor of 1.4 are represented as a function of
a dimensionless time T. The profiles approach equilibrium (i.e., T> 0.9) rapidly (months)
in thin beds 10 ern thickness) and slowly (decades and longer) in thick beds (c- 1 m).
Once the permeable zones in the aquifer are cleaned by pump-and-treat, or by natural
flushing, compared to the time required for contaminants to diffuse in, the time required
for diffusion out of the layers is longer.
As described in detail in Chapter 9 and also in Vogel et al. (1987), chlorinated solvents can be transformed microbiologically or abiotically to other organic compounds.
Anaerobic conditions, which will favor some of these transformations, are more prevalent in low-permeability zones. Therefore, in addition to providing for the temporary
storage of dissolved and sorbed solvents, low-permeability layers may act as sources
that slowly release both solvents and their transformation products to plumes. In fractured porous media, the matrix blocks between fractures can act in much the same
manner.
Nearly all of the compounds produced by degradation reactions are relatively stable,
and some are more hazardous in drinking water than the initial compounds. Figure 2.15
displays the transformation sequences for PCE and TCE. Of particular importance in
the transformation series is the production of vinyl chloride (VC), VC is a known carcinogen for which the U.S.EPA has specified the very low MCL in drinking water of
Plumes from DNAPLs
Sec. 2.5
.:
<...
. . .. JI--
-..': . =:
-
_01. . - - -
DNAPL
. :
,," .
PARAMETERS
clayey
-h
,
' ...
layer~ '. '.
~.
yO
'..
L
b
granular .a~u.ife~~
- -=c=:: ..
[":1
De =3 xl0- 6 POROSITY = 0.3

R=1.4
DIFFUSION TIMES: TCElyrs
diffusion in clayey layer

t
t
f i
t
t
~.
Co
Co
l/l
IU
T=0.05
0.3
-0.5
:0:
U
0.1
0.02
0,1
0.2
0.3
l-
0.5
0.4
2.6
4.3
7. 7
IU
1.7
10.2
17.1
30.7
43
256
427
768
0.9
Co
PROFI LES IN CLAY LAYER

l/l
b (m )
---
--
79
:J:
e
m
concenlration
Figure 2.14 Loss of contaminant mass from a plume by diffusion into a low-permeability
layer. Lower insert shows calculated concentration profiles for TeE in the layer at four
values of the dimensionless time T. Actual times for layers of various thickness are listed
in the table.
2 ~glL. VC is found in many plumes at DNAPL sites. These plumes are commonly
decades old, therefore it is evident that VC can be persistent, presumably due to continued
formation.
The formation of hazardous products such as VC complicates and enlarges the
contamination problem at chlorinated solvent sites. VC sorbs on aquifer and aquitard
solids less than does its parent compounds, and it is therefore more mobile in plumes.
Also, the complexity in the spatial and temporal distribution of contaminants at a site
caused by such transformations makes the task of site characterization and remedial investigations more difficult. This increases investigation costs and causes confusion in
data interpretation. The recognition in the scientific community that the common chlorinated solvents can transform in groundwater to hazardous compounds such as VC
did not come until the early 1980s; the conclusion that these transformations are common was not established until the late 1980s. The scientific, engineering, and regulatory
communities are still adjusting to the complications and challenges posed by these transformations.
80
Chap. 2
1,000
PCE
100
TiE
PCM
a TCA
12-~
A
"-r '')DCAT!v
10
VC
CM
DCM
CA
a - Abiotic transformation
Solvent Products:
PCE Perchloroethene
TCE Trichloroethene
TCA 1,1,1-Trichloroethane
12-DCA - 1,2-Dichoroethane
PCM Carbon Tetrachloride
TCM Chloroform
DCM Dichloromethane
Transformation Products:
12-DCE - 1,2-Dichloroethene
11-DCE - 1,1-Dichloroethene
VC Vinyl choride
11DCA - 1,2-Dichloroethane
CA Chloroethane
CM Chloromethane
Figure 2.15 Products of biotic and abiotic transformations of chlorinated solvents dissolved in groundwater and their relative mobilities expressed by their organic carbonwater partition coefficient (K oc ) .

At creosote and coal tar sites, the volume or mass of DNAPL situated in the vadose and
groundwater zones can be estimated with some degree of reliability using drilling records
and core analyses from DNAPL zones. At chlorinated solvent sites, however, residual
or free-phase DNAPL is much less frequently encountered in boreholes or wells below
the water table because of the effect of geologic variability and generally smaller release
volumes compared to creosote and coal tar sites. As a result, for chlorinated solvents,
estimates of total DNAPL mass in the groundwater zone and comparisons with the mass
in the vadose zone are unreliable; even order-of-magnitude estimates normally are very
uncertain. The uncertainty derives from difficulty in delineating the overall dimensions
of the source zone and the wide range in possible DNAPL saturations in pools and
residual zones due to microscale heterogeneities. Reliable conclusions that a DNAPL
actually exists below the water table at at given site must be based on several lines of
indirect evidence (see Chapter 13); these indirect methods do not provide DNAPL mass
estimates.
Sec. 2.5
81
Plumes from DNAPLs
At solvent DNAPL sites in granular aquifers where plumes are delineated using
extensive monitoring well networks, useful estimates of dissolved contaminant masses
in the plumes can be made. Table 2.1 provides six examples from Mackay and Cherry
(1989). These plumes occur in sand or gravel aquifers at Superfund or similar sites where
they have evolved over several decades. The plume mass for each plume is expressed in
terms of the equivalent number of 55 gallon drums of DNAPL, based on the dissolved
mass and the pure-phase DNAPL densities of the dominant solvents in the plumes.
Although the dissolved plumes are large, extending between 2 and 5 km downgradient
from their respective DNAPL source zones, the dissolved masses are small. These masses
range between 0.4 and 190 drums of equivalent DNAPL. When the quantities and periods
of DNAPL use and disposal at these sites as well as the persistence of the plumes are
TABLE 2.1 Solvent Plumes in Sand-Gravel Aquifers. (Reprinted with permission from MacKay and Cherry, 1989.)
..
...
.,....,
Site location and plume map
Presumed sources
Predominant DNAPL
contaminants
Plume volume
(liters)
Estimated chemical mass

dissolved in plume
(as equivalent DNAPL
volume in liters
or 55-gal drums)
chemical plant
Trichloroethylene
Tetrachloroethylene
5,700,000,000
15,000 (72 drums)
electronics plant
Trichloroethylene
I, 1, l-Trichloroethane
6,000,000,000
9800 (47 drums)
sewage infiltration
beds
Trichloroethene
Tetrachloroethylene
40,000,000,000
1500 (7 drums)
special waste
landfill
I A-Dioxane
Freon 113
electronics plant
trainyard, airport
Ocean City, New Jersey
Mountain View, California
Cape Cod, Massachusetts
Gloucester, Ontario
102,000,000
190 (0.9 drum)
Freon 113
I,I-Dichloroethylene
5,000,000,000
130 (0.6 drum)
1,1,1-Trichloroethane
Trichloroethylene
Dibromochloropropane
4,500,000,000
80 (0.4 drum)
San Jose, California
~
Denver, Colorado
5km
Flow
82
Chap. 2
considered, it appears highly likely that much larger amounts still remain in DNAPL
form in the source zones. Mass calculations for many other plumes in granular aquifers
in the U.S. provide results that are similar to those in Table 2.l.
At a given site, a number that may be of greater interest than plume mass is the
mass flux of dissolved contaminants. This flux can be estimated by multiplying the plume
concentrations by the groundwater fluxes through cross sections orthogonal to the mean
flow direction. The plume flux, expressed in equivalent drums per year, is a small fraction
of the total plume mass for the plumes in Table 2.1, and also for the many other plumes
that have been assessed in this manner. Plume fluxes range from less than 0.1 equivalent
drum per year to approximately 10 equivalent drums per year.
Near a DNAPL source zone, the plume flux equals the mass rate of dissolution of
the DNAPL. Farther downgradient, the plume flux should generally be similar, perhaps
with some decline due to transformations. Therefore, the ultimate impact of plumes
emanating from solvent DNAPL source zones can be evaluated in terms of impact of
relatively small annual mass fluxes to the receptor such as water-supply wells or surface
waters. In some cases, the fluxes present significant risk to human health and/or the
environment, and extensive remedial action is warranted. In other cases, the fluxes are
insignificant, and remedial action would provide little or no actual environmental risk
reduction.
2.5.4 Ultimate Fate

In preceding sections of this chapter, we have presented general conceptual models for
DNAPL sites that include subsurface source zones feeding dissolved contaminants to
plumes. The considerable masses of DNAPL that typically exist in the sources and the
small dissolved mass fluxes from the sources mean that such source zones will persist
for many decades or centuries. Therefore, in cases where remedial measures are not
imposed on the plume, the plume front will advance in the direction of groundwater
flow until the plume encounters a natural hydrologic boundary such as a river, lake,
or wetland, or until it enters the capture zone of a water supply well (Figure 2.16).
In some cases, the concentrations of dissolved contaminants leaving the source zone
may be sufficiently low to allow dispersion in the aquifer to diminish plume concentrations below regulatory MCLs prior to arrival of the plume at water supply wells
or at natural hydrologic boundaries. However, this is an unlikely scenario. Therefore,
it should be assumed that plume fronts (including unknown fronts) that have not yet
been observed at natural hydrologic boundaries are still advancing towards these boundaries.
Many plumes from solvent DNAPL source zones reached their natural hydrologic
boundaries long ago, and have therefore already attained their ultimate sizes. The contaminant flux may cross the interface between the groundwater zone and surface water
without alteration or attenuation. The ultimate fate of the contaminants then depends on
processes in the surface water systems. Some plumes enter surface waters by moving
up through organic-rich sediments at the bottoms of streams or lakes, or up through
organic-rich soils in wetlands. Little is known about the influence of such layers on
Sec. 2.5
83
Plumes from DNAPLs
plume chemistry or aquatic ecology. Scientific literature pertaining specifically to the

behavior and fate of solvents in surface water bodies is limited. However, it is known
that mixing as well as subsequent escape by volatilization to the atmosphere are normally rapid. Solvent concentrations above drinking water MCLs or even above very low
detection levels are rarely observed in streams or lakes downgradient of plume discharge
zones. Nevertheless, in Canada and in the northern part of the U.S., winter ice cover
can temporarily prevent escape to the atmosphere. Whether or not ice cover results in
significant adverse impacts on stream or lake ecology has not been determined.
Figure 2.16.a shows a plume that extends to a nearby river. In this case, the
plume travels in a local-scale groundwater flow system that is bounded laterally by
the river. At many sites, however, the deepest part of the DNAPL zone extends below
the bottom boundary of the groundwater flow for the local watershed. Flowpaths in such
a system can transport dissolved contamination towards natural hydrologic boundaries
that are situated at great distances from the site, often in a different watershed. For this
to occur in a system like that represented in Figure 2.16.a, the deep part of the plume
must travel beneath the river to the more distant boundary while the shallow part of the
plume discharges to the river (or to a local lake).
Many plumes are captured (and therefore contained) by water supply wells, thereby
causing contamination of the supplies (see Figure 2.16.b). If the wells are shut down,
plume containment ceases, and the plume begins to advance towards other boundaries.
Some plumes that cause contamination of supply wells are only partly captured by the
wells. The uncaptured part advances beyond the wells towards other wells or natural
boundaries.
-I
DiSSOLVED
CONTAMINATION
b.
--.~
--.
DNAPL
ZONE
~-
2!5
"-....
PLUME CAPTURl
INDRAWDOWN
CONE
-+
/.
::----'=------==
Figure 2.16 Two containment scenarios for a plume from a DNAPL source: a) plume
encounters a hydrologic boundary (river); and b) plume front is captured by a pumping
well.
84
Chap. 2
Monitoring required by the U.S. federal regulatory frameworks of Superfund and

RCRA has caused discovery of many thousands of solvent plumes in granular aquifers
and fractured bedrock. The average groundwater velocity in the horizontal direction
in sand or gravel is usually between 400 to 4,000 m1decade. The retardation of the
common solvents such as TCE, PCE, TCA, and DCM is generally small in such aquifers.
Therefore, plume fronts typically advance at rates close to the groundwater velocity,
although other chemical constituents of the plume may move more slowly. Plumes in
fractured rock, particularly those in relatively non-porous rock such as granite, or those
in rock with large fractures such as karstic limestone, can advance much more rapidly
than in granular aquifers. In large regional aquifers, particularly in southwestern U.S.,
the distances between recharge and discharge areas are tens to hundreds of kilometers.
Ultimately, solvent plumes in these aquifers can become immense if not contained by
extraction wells or other control systems.
2.6 IMPLICATIONS FOR SITE CHARACTERIZATION AND

MONITORING
A site investigation should provide the data necessary to determine whether the site
is a DNAPL site, and further, provide the data necessary to form technically-sound
decisions on remediation. Site characterization and monitoring at nearly all DNAPL sites
has been done using conventional methods previously developed and widely applied
to non-NAPL sites. The need to proceed rapidly with many site investigations in the
1980s, as driven by Superfund and RCRA regulations, contributed to this preference
for conventional methods: innovation requires time and flexibility. However, it is now
known that the task of delineating contamination at DNAPL sites presents unprecedented
difficulties to the scientific, engineering, and regulatory communities, and that these
difficulties can often be exacerbated when conventional methods are used. Drilling into
or through DNAPL zones poses hazards for DNAPL remobilization, and uncertainty
in the validity of sampling results.
Determination of the "nature and extent" of contamination is a demand inherent
in many regulations or guidelines for site investigations. Strictly speaking, however, this
is only very rarely accomplished at DNAPL solvent sites because DNAPL distributions
are so complex. The rarity of direct detections of residual or free-phase DNAPL in the
groundwater zone at the many intensely investigated DNAPL solvent sites is testimony
to this subsurface complexity.
Shelby-tube or split-spoon soil samples taken at selected depth levels, together
with permanent monitoring wells installed in clusters of two to four at different depths,
are common components of site investigations. For non-NAPL sites, these techniques are
often effective. For DNAPL sites, however, even after great expense has been invested in
the site investigation, they rarely provide the level of detail needed to provide a reliable
picture of the nature and extent of solvent DNAPL below the water table. In recent years,
new strategies and investigative technologies have been developed for DNAPL sites, and
their performances are being assessed at the present time.
Sec. 2.8
Conclusions
85
2.7 IMPLICATIONS FOR SITE REMEDIATION

In the 1980s, decisions were made at thousands of DNAPL sites across North America
to restore aquifers using groundwater extraction wells, above-ground water treatment,
then surface disposal of the treated water. This is the pump-and-treat method. At many
sites, the plumes are contained hydraulically in this approach, but the DNAPL source
zones persist with no apparent reduction in long-term concentrations in the source-zone.
The common futility of pump-and-treat for restoration of DNAPL source zones is now
well established.
In some hydrogeologic situations, pump-and-treat can restore the plume portion of
the aquifer to drinking water standards. However, for the restoration to be permanent, the
source zone must be isolated to prevent further contribution of dissolved contamination
to the plume. This can be done by positioning groundwater extraction wells within or
close to the source zone for hydraulic containment of the source. In order to prevent
plume re-growth, these wells must operate for as long as the DNAPL source zone exists,
which can be many decades, or probably longer at nearly all DNAPL sites.
An alternative to pump-and-treat for source zone containment is the construction of
an impervious enclosure around the entire source to prevent groundwater flow through the
source zone. Several technologies are now available for such enclosures in unconsolidated
deposits, some of which can be installed to depths of 50 to 100 m. Enclosures around
DNAPL source zones in most types of fractured bedrock are much more difficult and
expensive to construct because very closely-spaced boreholes must be used for injection
of sealants into the fractures.
Much effort in the field of groundwater remediation has been directed at cleaning
up surface soil and deeper parts of the vadose zone. This is commonly done in order
to prevent contaminants from leaching into the groundwater zone. However, at DNAPL
sites the contaminant mass flux to the groundwater zone from the vadose zone is typically
small relative to the flux contributed to the plume from the DNAPL situated below the
water table. Thus, DNAPL sites are distinct from non-NAPL sites in that remedial efforts
in the vadose zone generally contribute little or no reduction of risk resulting from the
groundwater contamination. At some DNAPL sites, however, removal or remediation of
surface soil is needed to reduce the risk at the surface to land users (e.g., homeowners), or to remove the potential hazards that would be caused by soil erosion or vapor
emission.
The recent recognition that pump-and-treat is generally not effective for remediating
DNAPL source zones has prompted intense research and development efforts aimed at
identifying and developing new technologies. Some of these technologies are directed at
source zone restoration, and others at plume containment. Chapter 14 describes some of
these new and emerging technologies and indicates their current development status.
2.8 CONCLUSIONS
The retention capacity of geologic media for DNAPLs in the vadose zone, particularly
chlorinated solvent DNAPLs, is typically small. Therefore, even small releases ofDNAPL
86
Chap. 2
commonly result in the movement of DNAPL all the way through the vadose zone and
into the groundwater zone. Repetitive small releases at the same location (as beneath a
leaky tank or distribution pipe) can cause penetration as deep or deeper than a single
sudden release of the same total spill volume. In some cases a water table of 50 or
100 m is not too deep to prevent DNAPL entry into the groundwater zone if solvent
leakage or disposal has occurred for many years. Lateral DNAPL movement from the
input location can also be large. We also note that low-permeability layers, either above
or below the water table, commonly contain fractures, root holes or other small openings
that allow DNAPL to move through aquitards into underlying aquifers. The DNAPL
retention capacity of most fractured rock is particularly small. This can result in very
deep DNAPL penetration, in some cases hundreds of meters or more.
Once the release of DNAPL into the subsurface ceases, subsurface movement of
DNAPL also ceases soon thereafter, perhaps within weeks or months at solvent sites.
The resulting immobile DNAPL then exists in the DNAPL source zone as "residual"
non-aqueous liquid and also possibly as "free-product" accumulations ponded on lower
permeability layers within aquifers, or on the tops of aquitards. The free-product DNAPL
will not become mobile again unless a release of more DNAPL causes further accumulation in the same zones, or unless there are changes in pressure in the surrounding water
phase due to groundwater pumping or injection.
Subsurface DNAPL can act as a persistent source for contaminant plumes in both
the vadose and groundwater zones. In fractured porous media, much or all of the DNAPL
in the fractures can dissolve and diffuse into the low-permeability material surrounding
the fractures. DNAPL source zones are often persistent because the majority of the entire
subsurface contaminant mass is usually still situated in the source zone, and the release
of contaminants from the source is slow, often diffusion controlled. Source depletion
mechanisms include diffusive loss to the vadose zone and diffusive loss to flowing
groundwater.
Plumes of groundwater contamination from chlorinated solvent source zones commonly extend great distances (kilometers or more) because: 1) these compounds exhibit
generally weak sorption affinity for most aquifer materials; 2) they usually undergo little
dilution; and 3) the contaminants degrade slowly or not at all. Plumes at DNAPL sites
have complex shapes. Multiple plumes can occur where shallow dissolved contaminant
migrates in one direction and deeper contamination in another direction, or where the
source zone has multiple DNAPL input locations. Plume size is often increased by lateral
spreading of vapor-phase contamination. Natural attenuation is usually not sufficient to
cause the fronts of chlorinated solvent plumes to stabilize at steady-state positions. Many
such plumes are ultimately destined to reach water supply wells. If no wells exist in
the path of a plume, hydrologic boundaries such as rivers or lakes, where dilution and
volatilization will occur, often tend to minimize the severity of the environmental impact.
The overall size and internal anatomy of a DNAPL source zone and the plume(s)
caused by such a source zone will depend on many factors. These factors include the
type of DNAPL, the volume of the release, the depth to the water table, and the general
geologic and topographic settings. What normally distinguishes the degree of complexity
of a DNAPL source zone from other contaminant sources is the considerable influence of
Sec. 2.8
Conclusions
87
small-scale variabilities, geologic structure, and/or bedding. The scale at which geological
variability affects DNAPL flow and distribution is much smaller than can be studied by
current field techniques. Thus, there is a critical need to develop and depend on conceptual
models which can guide decisions regarding the nature of this problem, as well as the
suitability and practicality of various options for corrective action.
The release of a given volume of DNAPL will produce much different results
from one hydrogeologic situation to the next because the retention capacity properties of
the subsurface will vary greatly from site to site and within sites. The determination of
feasible cleanup levels, time frames, and ultimate costs depend on estimates of the total
contaminant mass and the distribution of that mass in the subsurface. It is particularly
important to understand how much of the chlorinated solvent is present in slow release
zones, fast release zones, and/or present as free-product accumulations. Usually, these
estimates must be based on conceptual models: direct answers to such questions are
generally not attainable because of the excessive cost of field investigations at the relevant
scales. The examination and interpretation of site data requires the perceptual framework
of hypotheses represented by conceptual models. It is the responsibility of practicing
groundwater scientists and engineers to further refine these conceptual models by building
site-specific conceptual models for actual sites. The more tested our conceptual models
become through data acquisition and analysis, the more effective and less risky will be
our decisions regarding corrective actions.
This chapter has provided an overview of a broad range of geological conditions
and conceptual models so as to serve as the basis for organizing the subsequent chapters of this book, and to help the reader integrate the details presented in each of the
subsequent chapters into an overall paradigm for the behavior of DNAPLs in the subsurface. The methods that are available for the investigation of DNAPL sites as well as
the mathematical models that are available for representing subsurface DNAPL behavior
are improving rapidly. We expect that new information will permit further refinement of
the DNAPL paradigm.
ACKNOWLEDGEMENTS
The concepts and interpretations described in this chapter were developed during the
past ten years as a result of many discussions with our numerous colleagues in industry,
academia, consulting, and regulatory agencies; we offer these colleagues our sincere
thanks. We also acknowledge the pioneering research on chlorinated solvents conducted
by Dr. Friedrich Schwille, and thank him for the advice and encouragement he provided
to the solvent research group at the Waterloo Centre for Groundwater Research.
Solvents-in-Groundwater Research Program. Sponsors of the Program between 1988 and
1994 have included: The Boeing Company, Ciba-Geigy Corporation, Dow Chemical
Canada/USA, Eastman Kodak Co., General Electric Co., Laidlaw Environmental Systems
88
Chap. 2
2.9 REFERENCES
Hughes, B. M, R. W. Gillham, and C. A. Mendoza (1992) "Transport of trichloroethylene vapors in
the unsaturated zone: A field experiment". In: Proceedings: Conference on Subsurface Contamination by Immiscible Fluids, International Association of Hydrogeologists, Calgary, Alberta,
April 18-20, 1990.
Imhoff, P. T., P. R. Jaffe, and G. F. Pinder (1989) "Experimental investigation of the dissolution dynamics of chlorinated hydrocarbons in porous media". In: Proceedings: Symposium on
Processes Governing the Movement and Fate of Contaminants in the Subsurface Environment,
Stanford, California.
2. Source functions for pools of solvent", Environ. Sci. Technol., 26, 896-901.
Kueper, B. H. and D. B. McWhorter (1991) "The behavior of dense non-aqueous phase liquids in
fractured clay and rock", Ground Water, 29, 716-728.
Kueper, B. H., W. Abbot, and G. J. Farquhar (1989) "Experimental observations of multiphase
flow in heterogeneous porous media", J. Contaminant Hydrology,S, 83-95.
Mackay, D. M. and 1. A. Cherry (1989) "Groundwater contamination: Limits of pump-and-treat
remediation", Environ. Sci. Technol., 23, 630-636.
Mendoza, C. A. and E. O. Frind (1990a) "Advective-dispersive transport of dense organic vapors
in the unsaturated zone. I. Model development", Water Resources Research, 26, 379-387.
Mendoza, C. A. and E. O. Frind (I 990b ) Advective-dispersive transport of dense organic vapors
in the unsaturated zone. 2. Sensitivity analysis", Water Resources Research, 26, 388-398.
Parker, B. L., R. W. Gillham, and 1. A. Cherry (1994) "Diffusive disappearance of immiscible
phase organic liquids in fractured geologic media", Ground Water, 32, 805-820.
Poulsen, M. and B. H. Kueper (1992) "A field experiment to study the behavior of tetrachloroethylene in unsaturated porous media", Environ. Sci. Technol., 26, 889-895.
Rivett, M. O. (1995) "Soil-gas signatures from volatile chlorinated solvents: Borden field experiments", Ground Water, 33, 84-98.
Schwille, F. (1988) Dense Chlorinated Solvents in Porous and Fractured Media - Model Experiments, translated from the German by J. F. Pankow, Lewis Publishers, Boca Raton, Florida.
Vogel, T. M, C. S. Criddle, and P. L. McCarty (1987) "Transformations of halogenated aliphatic
compounds", Environ. Sci. Technol., 21, 722-736.
Wilson, J. L. and S. H. Conrad (1984) "Is physical displacement of residual hydrocarbons a realistic
possibility in aquifer restoration?", In: Proceedings: National Water Well Assoc. Conference on
Petroleum Hydrocarbons and Organic Chemicals in Groundwater, Houston, Texas, November
5-7, pp. 274-298.
3
Mechanics and Mathematics
of the Movement of Dense
(DNAPLs) in Porous Media
David B. McWhorter 1 and Bernard H. Kueper/
1 Department
of Chemical and Bioresource Engineering, Colorado State University,

Fort Collins, Colorado, 80523
2Department of Civil Engineering, Queen's University, Kingston, Ontario,

Canada, K7L 3N6.
Page
3.1
3.2
3.3
3.4
3.5
Introduction
3.2.1 General
3.2.3 Dependence of Fluid Content on Capillary Pressure
3.2.5 Algebraic Pc!Sw) Relationships
3.2.6 Summary
3.3.1 General
3.4.1 General
3.4.3 Relative Permeabilities
3.5.1 General
90
91
91
92
93
95
98
99
99
99
100
102
104
105
105
105
106
108
108
109
110
111
89
90
3.6
3.7
3.8
3.9
Mechanics / Mathematics of Movement of DNAPLs in Porous Media

3.6.1 General
3.6.3 The Buckley-Leverett Solution
3.6.4 Lateral Migration of DNAPL
Conclusions
References
Chap. 3
112
112
113
114
116
119
121
123
125
126
ABSTRACT
The fundamental principles governing multi-phase flow in porous media are presented.
Topics addressed include interfacial tension, capillary pressure, fluid contents, entry pressure, wettability, residual and pool formation, and relative permeability. The equations
governing the static distribution and movement of non-aqueous phase liquids (NAPLs)
in the subsurface are also outlined. Simple applications involving each of these topics is
provided, including the height of dense non-aqueous phase liquid (DNAPL) in a well,
the distribution of DNAPL in unsteady horizontal flow, the Buckley-Leverett solution,
and DNAPL displacement by waterflooding.
3.1 INTRODUCTION
Chlorinated solvents entering the subsurface from a spill, leak, or other release often do
so as constituents of a nonaqueous phase liquid (NAPL). When a significant fraction
of the NAPL is made up of one or more chlorinated solvents, the specific gravity will
be greater than unity. The term dense nonaqueous phase liquid (DNAPL) is then used
to describe the fluid. DNAPLs are often described as being "immiscible" with soil
gas and subsurface waters because they can retain their identity as a separate phase,
distinguishable from contiguous fluids by distinct interfaces. Actually, no fluid can ever
be totally immiscible with another, and the constituents of a DNAPL can partition into
the gaseous and aqueous phases, and will sometimes form extensive vapor and dissolved
plumes. Indeed, a primary motivation for the study of DNAPLs in aquifers is the fact
that they can function as enduring sources for soil-gas and aqueous-phase contamination.
A working knowledge of the way in which DNAPLs migrate and distribute themselves at equilibrium in geologic materials is an essential ingredient of informed decisions
concerning the characterization, diagnosis, and remediation of DNAPL sites. This chapter presents the physical principles regarding such knowledge. The connection between
theoretical concepts and practical considerations is demonstrated and emphasized by application of the principles to several specific types of situations that arise at DNAPL
sites. Of course, we must do this in a rather generic way, and the practitioner is likely
to find applications confounded by heterogeneity, lack of data, and other complicating
factors. Nevertheless, the physical principles embedded in our examples form reliable
guides for appropriate transfer of experience from one site to another.
Sec. 3.2
91
3.2 CAPILLARY PRESSURE AND FLUID CONTENTS

3.2.1 General
The invasion of groundwater by a DNAPL is a process in which previously-existing
pore waters are displaced by the invading DNAPL. Consider, for example, the pool of
DNAPL that has accumulated on the bedrock in Figure 3.1. The upper surface of the
pool is a macroscopic interface that corresponds to the interface separating water and
DNAPL in any ordinary vessel, say a beaker in the laboratory. However, in contrast to
the situation in a beaker, the water and DNAPL coexist in the pore spaces located below
the macroscopic surface of the DNAPL pool. In fact, water is continuous throughout the
thickness of the aquifer, including through the DNAPL pool.
Pore-scale interfaces separate the water and DNAPL at points below the macroscopic interface. Unlike the macroscopic interface, whose orientation is controlled by
gravity, the pore-scale interfaces are primarily affected by interfacial forces that exist
between the two fluids and among the fluids and the solid. Interfaces at the pore scale
may be randomly oriented with respect to the gravitational field. Furthermore, the relative
amounts of water and DNAPL in a representative volume of the porous medium is not
constant below the macroscopic interface. The relative amounts of water and DNAPL
will depend upon the difference in pressure between the two fluids.
Based on the above discussion, we see that the intuitively appealing notion of
separate water and DNAPL bodies that interact only on their macroscopic boundaries, as
in a beaker or pipe, is inappropriate. In the remainder of this section, we explore these
WATER
MACROSCOPIC INTERFACE
DNAPL
WATER
PORESCALE
INTERFACE
Figure 3.1 Macroscopic and pore-scale interfaces between DNAPL and water in a
porous medium.
92
Chap. 3
matters in more detail. While much of the foregoing discussion has been framed in the
specific context of a static DNAPL pool, the principles involved apply in both static and
dynamic situations.

The interface between two contiguous immiscible fluids experiences a tension that is
similar to that in a stretched membrane. This tension is called the interfacial tension
and arises because of unbalanced cohesional forces on molecules at the interface (Adam,
1968). The tension causes the interface between the two fluids to contract and form an
area that is as small as possible. Interfacial tension is usually expressed as a force per
unit length, but can also be thought of as energy per unit area. The tendency to contract
minimizes the excess free energy of the interface.
A force balance on a curved interface between two fluids (Corey, 1986) leads to
the conclusion that the pressures in the fluids on either side of the interface are not equal;
the difference being given by
!J.P = a
[~+~]
r, rz
(3.1)
where !J.P is the pressure difference across the interface, a is the interfacial tension, and
rj and ri are the principal radii of curvature of the interface. The fluid on the concave
side of the interface is at the higher pressure. If the interface forms a subsection of a
sphere, then r, = ri = r , and the more familiar
!J.P
= 2alr
(3.2)
is obtained. Interfacial tension is seen to be the property that permits two fluids to exist in contact at different pressures. Values of interfacial tension range from about 20
to 50 dynes/em for most water-DNAPL fluid pairs (Mercer and Cohen, 1990). However, DNAPLs encountered at field sites are often mixtures of organic liquids and have
undergone aging processes that may have altered the interfacial properties. The interfacial tension of field DNAPL-water systems should therefore be measured whenever
possible.
When two fluids are in contact with a solid, one usually has a greater affinity for
the solid than the other. The fluid with the greater affinity for the solid is said to be
the wetting fluid, the other being the non-wetting fluid. The wetting fluid preferentially
spreads over the solid. The relative affinity of the two fluids for the solid (wettability)
is manifested in a contact angle, which is the angle that the fluid-fluid interface makes
with the solid. Wettability depends upon the chemical properties of the two fluids, and
upon the composition of the solid surface. Anderson (1986) has reviewed the ways in
which wettability can be measured.
The concept of contact angle permits one to relate the pressure difference across an
interface not only to the interfacial tension and interface geometry as in Eq.(3.1), but also
to the geometry of the opening across which the interface is positioned. For example, the
pressure difference across an interface in a capillary tube with a circular cross-section of
Sec. 3.2
93
radius a is
~P
cos o
= 20"-a
(3.3)
where ex is the contact angle. This result follows from Eq.(3.2) upon recognition that the
radius of the interface is a/cos a.
Organic solvents are often observed to be the nonwetting fluid when present in
aquifers. For this reason, DNAPLs preferentially occupy the larger openings, while water
tends to coat the solids and occupy the smaller, more constricted portions of the pore
space. This pore-scale arrangement of fluids is illustrated in Figure 3.1 where the DNAPL
is found on the concave side of the pore-scale interfaces and is the fluid at the higher
pressure. The difference between the nonwetting-fluid (DNAPL) pressure and the wettingfluid (water) pressure is called the capillary pressure P;
(3.4)
where P n w is the pressure of the nonwetting fluid and P w is the pressure of the wetting
fluid. Throughout this chapter we will assume that water is wetting with respect to a
DNAPL.
The geometry of an interstitial pore space is obviously highly complex. Nevertheless, it is possible to conceive of a network of interstitial spaces connected by pore
throats of a smaller characteristic dimension. To the extent that Eq.(3.3) applies in such
a complex geometry, it predicts the threshold value of the capillary pressure that must
be exceeded for DNAPL to pass through a pore throat of radius a. Thus, DNAPL is
denied access to an interstitial pore until the capillary pressure exceeds the threshold
value associated with the largest throat already in contact with the DNAPL. Once entry has been achieved, the DNAPL moves into the pore. The water-DNAPL interfaces
then position themselves across regions of the pore space that support radii of curvature
consistent with the prevailing capillary pressure. As the capillary pressure is increased,
successively smaller pore throats are invaded by the DNAPL.
3.2.3
Dependence of Fluid Content on Capillary Pressure
The foregoing conceptualization of DNAPL invasion into a previously water-filled porous

medium suggests that there should exist a relationship between capillary pressure and
the relative amounts of DNAPL and water in a porous medium. Indeed, this is the
case, but the complex geometry of the pore space precludes direct computation of the
relationship: it must be measured for a given medium. The measured relationship is
known as the capillary pressure-saturation function or, sometimes, simply as the
capillary pressure curve. This macroscopic function is defined only for volume elements of a porous medium that are very large relative to the volume of individual
grains.
It is convenient to express the fluid contents in terms of the fraction of the
total interstitial void volume occupied by each fluid. Let V v be the void volume in a
94
Chap. 3
representative element of bulk porous medium. Then

(3.5)
(3.6)
Sw + Snw = I
(3.7)
where Vw and Vnw are the volumes of wetting and nonwetting fluids, respectively, and
Sw and Snw are the relative volumes of wetting and nonwetting fluids expressed as a
fraction of the pore volume. The relative volumes are referred to as saturations (e.g., the
"wetting-fluid saturation").
The capillary pressure-saturation function Pc(Sw) is often measured in a cell similar
to that shown in Figure 3.2. This cell was used by Kueper (1989) to measure the capillary
pressure curve for tetrachloroethylene (peE) and water in several sands. Such a test is
initiated with all of the pore space filled with water; the DNAPL pressure is then increased
incrementally. Each step-increase in DNAPL pressure causes increased invasion of the
pore space by DNAPL. The volume of DNAPL injected and/or the volume of water
displaced is measured and used to calculate the fluid content after equilibrium has been
~AlWUlWlTHVENTHOlE
WATER
PARAFUI WITH
AIR
/VENTHOlE
V
.k"" PCE RESERIIOIR
PCE
1EFI..OH TUBE
L====:==;l
ZElJ VAl.VE
TEFLON
k'TOPPlAlE
-V
WATER
WATER
QLASS
PCE
FUNNEL
0
o
SAMPI.
AlITT8l
DISC
VAl.VE
FmTfM
PlATE
~lHREADED
ROD
FL.EXIIl_
HOSE
Figure 3.2 Schematic of a capillary pressure-saturation cell. (Reprinted with permission

from Kueper, 1989.)
Sec. 3.2
95
established at each pressure level. The porous plate on which the porous medium rests
allows the displaced water to leave the cell, but prevents DNAPL from doing so. It is
usually constructed of porous ceramic. The water pressure is indicated by the water level
in the outflow burette, and the DNAPL pressure is calculated from the fluid levels in the
DNAPL reservoir. Therefore, the capillary pressure corresponding to each equilibrium
fluid content is known. More sophisticated devices employing pressure transducers and
utilizing prescribed increments of injected volume instead of increments of pressure are
also available.
Data obtained by the method described above are shown in Figure 3.3. These data
were utilized in a numerical modeling study by Kueper and Frind (1991 b). The capillary
pressure at a given water saturation was found to be greater in the sands with lower
hydraulic conductivity. This is consistent with Eq.(3.3) in that the sands with smaller
hydraulic conductivity are comprised of smaller pores and pore throats. Also, little or no
DNAPL was found to invade the samples until the capillary pressure reached a threshold
value determined by the largest pore throat (or throats) exposed to the DNAPL on the
exterior surface, and providing access to the interior of the sample. While some invasion
of DNAPL occurred at capillary pressures equal to or slightly above the threshold, it is
not likely that DNAPL became continuous throughout the sample until an even greater
capillary pressure was exerted. The capillary pressure at which the DNAPL becomes
continuous in the macroscopic sense and is capable of flowing through the material
is known as the entry pressure (Corey, 1986). For most materials, the entry pressure
corresponds to a water saturation in the range of 0.8 to 0.95.
Another characteristic capillary pressure that has found widespread use is the displacement pressure (P d ) . The displacement pressure is described in Figure 3.4, and is
the minimum capillary pressure required to initiate invasion of a water saturated porous
medium by a DNAPL. The value of P d is obtained from the capillary pressure curve by
drawing a tangent to the curve at the inflection point, and extrapolating that tangent to
Sw = 1. Often, there is little numerical difference between the entry pressure and the
displacement pressure.
Insofar as the displacement pressure must be exceeded in order that DNAPL invasion occur, it should not be surprising that the subsurface distribution of Pd plays
a central role in determining the paths of DNAPL migration (Kueper and Frind, 1991a,
1991b). Indeed, it is a key factor in the diversion of DNAPL around fine-grained lenses
and in the formation of DNAPL pools. These subjects are taken up in more detail below.
However, before doing so, we must first address the fact that the capillary pressuresaturation function is not unique, even for a particular porous medium and fluid pair.
The capillary pressure-saturation curves shown in Figure 3.3 were measured under conditions in which the water saturation decreased and the capillary pressure increased as
the result of invasion of DNAPL into previously water-saturated media. This phenomena
is known as drainage and corresponds to field situations in which the DNAPL advances
into ground water from some type of source. Once the source is exhausted, the DNAPL
96
Chap. 3
t001r------r-----r-------r-----,
'i:"
Ii
~
75
K-l.21E-02 om/o
K-O.1QE.03 om/o
K-7.38643 om/o
K-5.84E-03om/o
X K-5.81 E-43 om/o
.a
o
a:
:::>
K-a.44E.Q3 om/o
K-4.30E.Q3om/o
fI)
fI)
W 5
a:
a.
>-a:
a::
~
25
oo~---*"----:!,;::----~!':----~
25
50
75
WATER SATURATION
100
Figure 3.3 Measured capillary

pressure-saturation curves for a
perchloroethylene-water system. (Reprinted
with permission from Kueper and Frind,
1991b.)
will continue to migrate away from the source and will be replaced in that zone by water.
The situation is now one of increasing water saturation and decreasing capillary pressure,
referred to as wetting. Injection of water to displace DNAPL in remediation efforts is
another example of a circumstance in which the water saturation increases and capillary
pressure decreases.
Unfortunately, the Pe(Sw) relationship is not the same for decreasing Sw as for
increasing Sw. Figure 3.4 shows the type of hysteresis that is commonly observed. The
water saturation at any particular capillary pressure is less during the wetting process than
during drainage. Of particular importance is the fact that the maximum water saturation
(Sm) achievable during the wetting process is less than unity. The quantity (I - Sm) is
known as the residual DNAPL saturation. Residual DNAPL is comprised of blobs and
fingers (ganglia) of DNAPL that have been cutoff and disconnected from the continuous
DNAPL body by the invading water. It can be said that residual DNAPL is occluded by
water.
A body of DNAPL issuing from a source continues to sink following the elimination or exhaustion of the source. As water replaces the draining DNAPL, residual DNAPL
is formed on the trailing surface of the DNAPL body. In the absence of an obstructing
geologic feature, most of the DNAPL eventually will be converted into residual, and
DNAPL migration will cease. At the pore scale, the reinvasion of water and the conse-
Sec. 3.2
97
CAPILLARY PRESSURE CURVE
100%
MAIN ORAl AGE
MAIN WETTING
----:'---.. . .
DISPLACEMENT
PRESSURE
---""'I!:'----..:...--~~-100%
Figure 3.4
Sw
Wetting Phase Saturation
Snw
Nonwetting Phase Saturation
Swr
Residual Wetting Phase
Snwr
Residual Nonwetting Phase
Capillary pressure-saturation relationships showing hysteresis.
quent displacement of DNAPL occurs as a sequence of menisci advancement in pore

bulges by jumps and the consequent rupture and choke-off of menisci positioned across
pore necks. The result is a rupture of nonwetting-f1uid connections and the isolation of
fingers of non wetting fluid. The pore doublet model (e.g., Morel-Seytoux, 1969; Stegemeier, 1974) is a popular explanation for the entrapment of non wetting fluid by the
invasion of a wetting fluid. Chatzis and Dullien (1983) show that pore doublet models
must include a neck-bulge-neck type of pore geometry in order to explain the entrapment
of the non wetting fluid. According to Mohanty et al. (1987), the amount of nonwetting
fluid isolated from the connected fluid body depends upon the local pore geometry, pore
topology, fluid properties, the externally imposed pressure gradient, and the amount of
nonwetting fluid present prior to displacement.
Residual DNAPL does not move with the flowing ground water. Instead, it forms a
distributed, immobile source for dissolved-phase partitioning into the contiguous ground
98
Chap. 3
water. The solubilities of most organic solvents are orders of magnitude greater than the
concentrations at which there is concern for human health. Thus, since residual DNAPL
will initially be present virtually everywhere there was once continuous DNAPL, it will
be an instrumental feature in the creation of dissolved-phase groundwater plumes. Typical
values of residual DNAPL saturation in the field are presently unknown; in cases where
small-scale heterogeneities are present, they are a function of the scale of measurement.
In a field experiment conducted in a sand aquifer, Kueper et at. (1993) observed residual
DNAPL saturations below the water table that ranged from 0.01 to 0.15. In homogeneous
sands, Schwille (1988) reports laboratory values on the order of 0.02 to 0.15. In coarse
Ottawa sand, residual saturations of DNAPL in the range 0.15 to 0.4 were observed
(Anderson, 1988).
Once the DNAPL ceases to flow downward, there will remain a connected body
of DNAPL of height Pd / !:'>.pg. This is the thickness required to generate the minimum
capillary pressure for DNAPL penetration. Thus, it is expected that all DNAPL sites will
have some pools, even in very homogeneous media.
3.2.5 Algebraic Pc(Sw) Relationships

In virtually all calculations of two-phase phenomena in porous media, it is convenient
to use algebraic expressions for PASu,) as opposed to the experimental data themselves.
Many algebraic formulae have been proposed for this purpose. Among the most popular
is the Brooks-Corey (1966) equation:
(3.8)
where Pd is the displacement pressure for the medium, and A is a pore-size distribution
index. Another popular expression is the van Genuchten (1980) equation
r,
= po(S;l/m - 1)1-m
Pi>:
(3.9)
where Po is the characteristic capillary pressure and m is a fitting parameter determined

by the pore-size distribution. In both of these expressions, S, is a normalized wetting-fluid
saturation defined by
S _ Sw - Sr
e - Sm - Sr
(3.10)
We recall that Sm is unity when DNAPL is invading into a water-saturated porous

medium; it is less than unity when water displaces DNAPL because of the entrapment
of DNAPL.
The parameters Pd , A, and S, for the Brooks-Corey equation are evaluated by fitting
the equation to experimental data. The values of m (note that 0 < m < 1), Po, and S,
in the van Genuchten equation are similarly determined. The wetting-fluid saturation S,
is often called the residual saturation. It is approximately the wetting-fluid saturation at
which the Pc(Sw) curve appears to approach a vertical asymptotic value (see Figure 3.4).
It is not to be confused with the residual saturation for the DNAPL. Parameter values
Sec. 3.3
99
(particularly Pd and Po) will be different when determined from wetting data and drying
data. Other algebraic PC<Sw) relationships such as that obtained by Luckner et al. (1989)
account for "scanning curves" between the main drainage and main wetting limbs.
3.2.6 Summary
The replacement of groundwater in the pores of an aquifer by invading DNAPL never
becomes complete. Thus, water and DNAPL will coexist in the interstitial pore space
anywhere that DNAPL is present". Interfacial tension, a property of two adjacent fluid
phases, permits adjacent fluid phases to be under different pressures. The pressure difference across pore-scale interfaces is balanced by the interfacial forces acting tangent
to the curved interfaces. Because water is usually the wetting fluid in saturated subsurface systems, it preferentially occupies the smaller, more constricted regions of the pore
spaces. A DNAPL solvent, being the nonwetting fluid, resides and moves through the
larger regions of the pore network and is the fluid at the higher pressure. The dependence
of the pressure difference across the curved interface yields the macroscopic relationship
between capillary pressure and fluid content that is the capillary pressure-saturation function. This function enables one to quantitatively relate the relative volume of water and
DNAPL in a representative volume element of porous medium to the pressure difference
between the DNAPL and water in that element. The capillary pressure-saturation function has been shown to apply for both static and dynamic problems in many situations
of practical interest.
3.3 DISTRIBUTION OF DNAPL IN A POOL ABOVE AN AQUITARD

3.3.1 General
In Section 3.2.4 we found that there exist relationships between capillary pressure and
fluid saturations. Thus, given knowledge of the capillary pressure curve and the value
of capillary pressure, one can determine the water and DNAPL saturation values. How
one might know the capillary pressure in the first place has remained unexplained to this
point. In the most general case, the capillary pressure must be determined by solution of
the differential equations for flow of both water and DNAPL (see below).
The capillary-pressure distribution and, therefore, the fluid-content distribution, is
calculated most easily for a case when neither fluid is in motion. The downward migration
of DNAPL from a large source may be arrested by an aquitard. Upon encountering a
stratum that it is unable to penetrate, the DNAPL will accumulate and spread laterally
to form a pool. The accumulation of DNAPL will be accompanied by an increasing
capillary pressure. Eventually, the pool thickness may cease to increase with the system
then becoming static. Alternatively, the DNAPL may accumulate to the point where
penetration of the aquitard occurs. The distributions of capillary pressure and the fluid
tThe context here is that invading DNAPL can oever physically force out all of the water in the pore
space. Nevertheless, DNAPL can come to occupy 100% of the pore space when the percent of the pore space
occupied by the invading DNAPL is high enough that all of the remaining water can dissolve into the DNAPL.
100
Chap. 3
saturations in a static pool and the conditions for incipient penetration of the aquitard
may be determined as follows.
A pool of DNAPL with thickness T is shown perched on an aquitard in Figure 3.5. Such
a pool might be formed by a gradual thickening due to lateral migration of DNAPL
from a remote source. During formation of the pool, the DNAPL is advancing into a
previously water-saturated porous medium and the relevant capillary pressure-saturation
relation is the drainage curve shown in Figure 3.4. Once the source is exhausted, some
DNAPL drainage from the pool interior occurs as the pool continues to expand laterally.
The result is a recession (lowering) of the upper surface of the pool, accompanied by reinvasion of water and the formation of a halo of residual DNAPL. Because the recession
is a wetting process, the capillary pressure on the upper surface of the connected DNAPL
is approximately zero.
The water pressure increases with depth below the water table in the usual way,
namely
(3.11)
where Pw is the density of water, g is the gravitational constant, and Z is a vertical

coordinate which is measured positively upward. The constant C 1 can be evaluated from
the knowledge that P w = 0 on the water table at a known value of Z. Eq.(3.11) is the
familiar equation of hydrostatics and applies in the aquifer just as it does in a body of
standing surface water. It also applies in the portion of the aquifer where DNAPL is
Figure 3.5
DNAPL pool above an aquitard.
Sec. 3.3
101
present. For the variation in DNAPL pressure within the DNAPL pool we have
(3.12)
where Pnw is the DNAPL density (Pnw > Pw).
Eqs.(3.11) and (3.12) are substituted into the definition of capillary pressure
(Eq.(3.4 to obtain
P; = -t1pgZ
+C
(3.13)
where t1p is the positive quantity Pnw - Pw. The constant C is evaluated by noting that
P; = 0 at Z = T. Therefore, the distribution of capillary pressure within the DNAPL
pool is given by
Pc = !'!,.pg(T - Z)
(3.14)
Z2: T
Eqs.(3.11) and (3.12) are shown graphically in Figure 3.6. It is apparent from this figure
that the capillary pressure increases with depth in the pool because the DNAPL pressure
increases more rapidly with depth than does the pressure in the water. This is due,
of course, to the fact that Pnw > Pw. Note that the capillary pressure in the pool is
independent of the depth of the pool below the water table.
Clearly, the maximum capillary pressure occurs at the base of the DNAPL pool,
and is directly proportional to both the density difference and the pool thickness. That
the DNAPL has not penetrated the underlying aquitard and dissipated the pool is directly
attributable to the condition that the displacement pressure of the aquitard exceeds the
capillary pressure imposed by the pool. Incipient invasion of the aquitard is given by the
Ground Surface
Water
Waterand DNAPL
Pressure
Figure 3.6
Variation of fluid pressures in a DNAPL pool.
102
Chap. 3
condition
(3.15)
where Ps refers to the aquitard, not the medium in which the pool resides.
As we have already observed, Pd depends upon the size of the pore throats, as
does the hydraulic conductivity. A close relationship between Pd and hydraulic conductivity might therefore be anticipated. Leverett (1941) assumed that Pd as well as all
other capillary pressures are inversely proportional to the square root of the intrinsic
permeability and, therefore, of the hydraulic conductivity. Laliberte et al. (1968) showed
from theoretical considerations that P d should be inversely proportional to the square
root of the hydraulic conductivity. McWhorter and Nelson (1979) report a correlation of
Pd with hydraulic conductivity that can be expressed as
Pd=9.6
(J
(K)
-
-0.403
<t>
(Ja w
(3.16)
where <t> is the porosity, (J is the interfacial tension of the fluid pair of interest, and
is the interfacial tension of water against air. Other investigators have found similar
correlations (e.g., see Thomas et al., 1968). The hydraulic conductivity Kin Eq.(3.l6) is
commonly expressed in units of cm/s. Common units for the displacement pressure are
em of water equivalent.
By combining Eqs.(3.15) and (3.16), one obtains the simple formula
(Jaw
T= 9.6 (~) (~. ) (K)

<t>
f!,.p
-0.403
(3.17)
(Jaw
from which the pool thickness (ern) required to initiate DNAPL entry into a stratum with
hydraulic conductivity K (cm/s) can be computed. For example, a pool thickness of 106
em is required for entry of TeE into a stratum for which K = 0.001 cm/s and <t> = 0.3.
We have assumed interfacial tensions of 34 dynes/em and 65 dynes/em for TCE/water
and air/water, respectively. The value of f!,.p for TCE and water is 0.47 g/cm'.
It is clear from calculations with Eq.(3.17) that stratigraphic features with K
values which are only moderately low relative to the surrounding materials can halt and/or
redirect the downward migration of DNAPLs. Eq.(3. 17) also indicates that DNAPLs with
large specific gravities are less likely to form large pools than are DNAPLs with specific
gravities near unity, other factors being equal. Indeed, it is the authors' experience that
chlorinated solvents with their high densities form thick pools only rarely. (In contrast,
creosote and coal tars, which are DNAPLs with lower densities, often occur in pools of
substantial thickness.)

The variation of capillary pressure through the DNAPL pool, as described by Eq.(3.l4),
is readily translated into a vertical distribution of fluid saturations by use of the capillary
pressure-saturation function. For example, the capillary pressure given by Eq.(3.l4) can
Sec. 3.3
103
be substituted for capillary pressure in Eq.(3.8) or Eq.(3.9) to arrive at explicit expressions

for water saturation as a function of Z. Using Eq.(3.9), the result is
Z= T _ . [(Sill - Sr)-l/m _ l]l-m

f".,pg
(3.18)
Sm - Sr
Distributions of water and DNAPL for pool thicknesses of 0.5 m and 0.3 m were calculated using Eq.(3.18) and are shown in Figure 3.7. The parameter values used for these
calculations are Pol f".,pg = 0.2 m, m = 0.75, Sm = 0.9, and S; = 0.2.
It is apparent from Figure 3.7 that thick pools result in greater DNAPL saturations
than do thin pools. DNAPL saturations are similarly sensitive to the parameter Pol f".,pg.
The important parameter actually is Tis pg / Po. Significant DNAPL saturations exist
only in pools for which this ratio is greater than unity. For any given pool thickness,
the ratio is large when f".,p is large and/or Po is small. Small values of Po are associated
with porous media with large hydraulic conductivity (see Section 3.3.2). The ability of
a DNAPL to flow from a pool into a drain or a well will be seen below to be critically
dependent upon the DNAPL saturation in the pool. The greater the DNAPL saturation,
the more readily the DNAPL will move under an applied driving force. Thus, thick pools
of DNAPL with high specific gravity located in highly permeable material will be the
most amenable to direct recovery by drains or wells.
If investigations eventually establish the thickness and area of a given DNAPL pool,
it will be possible to estimate the volume of DNAPL in the pool. This is accomplished
sz
(m)
(m)
Water
Water
SZ
0.5
0.4
Water
and
0.2
Water
0.2
and
0.1
DNAPL 0.1
xx
0 0.2 0.4 0.6 0.8 1 X

Sw
Figure 3.7
.,....
0.4
0.3
0.3
DNAPL
'>(
0.5
Water
-- I
Water
x x X o 0.2
tPNAPLI
i
0.4 0.6 0.8
Sw
Distributions of water and DNAPL in pools of different thicknesses.
104
Chap. 3
by evaluating
Vnw=<I>
(l - Sw)d Z
(3.19)
with the use of Eq.(3.14) and either Eq.(3.8) or Eq.(3.9). Here, Vnw is the volume of
DNAPL per unit area. The integral in Eq.(3.19) evaluates the area labeled "DNAPL" in
the saturation distributions shown in Figure 3.7.
Upon construction of a monitoring well, a DNAPL may enter the well. If such entry
occurs, the DNAPL will seek a height such that the DNAPL pressure in the well is
equal to that in the adjacent aquifer at all elevations. The water-DNAPL interface in the
monitoring well will establish itself at an elevation equal to or greater than the upper
surface of the connected-phase pool, depending upon the capillary pressure at the top of
the pool.
The monitoring well shown in Figure 3.5 penetrates a pool that has experienced
a recession following the halt of DNAPL input. Under this circumstance, the capillary
pressure on the upper surface of the pool is approximately zero, as previously explained.
In this case, the water-DNAPL interface in the monitoring well is at the same elevation
as the upper surface of the connected-phase pool.
In the case of a pool that has undergone no recession, the capillary pressure will
equal the displacement pressure on the upper surface of the pool. Now, the DNAPL
pressure in the well at the elevation Z = T must exceed the water pressure in the adjacent
aquifer by Pd' This is possible only if the DNAPL in the well rises to an elevation that
is Pd / /";.pg above the top of the pool. Thus, the difference in elevation (/";.E) between the
DNAPL-water interface in the monitoring well and the top of the pool in the aquifer is
given by
(3.20)
with the top of the pool in the aquifer being at the lower elevation. It is as if the DNAPL
forms a negative capillary fringe.
Implicit in Eq.(3.20) is the assumption that the capillary pressure-saturation function
is such that the saturation of connected DNAPL is zero at a nonzero capillary pressure
(i.e., a Brooks-Corey function). In some cases the van Genuchten function, which considers DNAPL to be continuous at all P; as given in Eq.(3.9), may be more appropriate.
Examples include soils containing preferential pathways such as root and worm holes.
In a van Genuchten material, the DNAPLIwater interface in a monitoring well will be at
the elevation of the top of the DNAPL pool in the aquifer, with /";.E = O.
DNAPL will fill an observation well to a height dictated by the DNAPL pressure
in the aquifer only if DNAPL cannot escape the well bore at some point below the level
of DNAPL entry. For example, DNAPL entering a well bore from a lensor pool located
high in the screened section will have the opportunity to again flow out through the
screen at a lower level. Also, the length of the DNAPL column in the well is rarely
Sec. 3.4
105
related to the thickness of the DNAPL pool in the aquifer. Only if the well is terminated
at the base of the pool will the DNAPL thickness in the well correspond to the thickness
of the DNAPL pool in the aquifer. We can conclude that a great deal of caution should
be associated with the interpretation of DNAPL column heights in wells.
3.4 DARCY'S LAW AND RELATIVE PERMEABILITY
3.4.1 General
As in our previous considerations, our goal is to know the distribution of DNAPL saturations, but now under circumstances in which one or both of the fluids is in motion. For
static fluids, a balance of the forces due to pressure gradient and gravity in each fluid,
coupled with the function Pc(Sw), was all that was required. In the case of fluid motion,
the capillary pressure is no longer linearly related to the vertical space coordinate. The
distribution of capillary pressure must be calculated from equations that fully account for
fluid flow. Just as in a single-fluid system, Darcy's Law is the starting point for analyzing
multi-fluid flow. Darcy's Law is written twice for a two-fluid problem; once for each
fluid. Also, the concept of permeability must be generalized to account for the presence
of two fluids in the pore space.
The motion of individual fluid elements in porous media is extremely complex owing to
the highly irregular geometry of the pores in which motion can occur. The intractable
complexities associated with a pore-scale description of fluid flow are largely eliminated
by use of a macroscale relationship between the rate of flow and the driving forces on
the fluid. This is Darcy's Law. Variables such as density and pressure that refer to fluid
elements in traditional fluid mechanics now denote values averaged over a representative
volume of porous medium.
Darcy's Law is an empirical constitutative equation that forms the basis for practically all quantitative work with fluid motion in porous media. One consequence of
passing from the pore-scale to the macro scale view is the introduction of the permeability parameter. The permeability of a porous medium to a particular fluid strongly
depends upon the size, geometry and fluid content of the pores. Also, one cannot properly
speak of fluid velocity at the macroscale since the velocity of fluid elements is described
only in some average sense. It has become common to use the term "volume flux" or
"flux density" to denote the volume of fluid passing a macroscopic unit area of porous
medium per unit time.
Fluid motion in a porous medium is impelled by the gradient of fluid pressure
and the body force due to gravity. Motion is resisted by viscous shear which, in turn,
depends upon the rate of motion, the fluid viscosity, and the size, shape and tortuosity of
each opening through which the fluid passes. The latter are characterized by the intrinsic
permeability, k, of the porous medium when a single fluid is present. The coexistence of
another fluid in the pores reduces the area available for flow of either fluid and increases
the tortuosity of the flow path which fluid elements must traverse. These effects are
106
Mechanics! Mathematics of Movement of DNAPLs in Porous Media
Chap. 3
incorporated in the relative permeabilities to each fluid and are strong functions of the
fluid saturations.
We use the tensor notation that has become common in subsurface hydrology to
write Darcy's Law for the individual fluids as
(3.21)
(3.22)
where i and} are indices that extend over the three Cartesian coordinates. In Eqs.(3.21)
and (3.22), q is the volume flux, k is the intrinsic permeability, k; is the relative permeability, P is the pressure, p is the density, f.1 is the dynamic viscosity, and the subscripts
wand nw denote the wetting fluid (water) and nonwetting fluid (DNAPL), respectively.
By convention, the repeated index,}, implies a summation over} = 1, 2, and 3.
At first glance, Eqs.(3.21) and (3.22) appear to be uncoupled. Such is not the case,
however, because the relative permeabilities krw and krnw are strong functions of the
saturations. It was previously pointed out that fluid saturation is a function of capiIlary
pressure. Therefore, the relative permeabi1ities are functions of P; = Pnw - Pw, and
Eqs.(3.21) and (3.22) are strongly coupled. Explicit demonstrations of the importance
of that coupling are provided by the example applications discussed later in this chapter.
3.4.3 Relative Permeabilities
Typical relative permeability functions are shown in Figure 3.8. By definition, these
functions range from zero to unity. The relative permeability to the wetting phase (water)
is thought to be practically free of hysteresis. On the other hand, the saturations at which
k rnw is zero in a wetting or draining process are not the same because of entrapment of
nonwetting fluid during the wetting process. It is usual to assume that krnw is greater than
zero in a draining process for all Sw < 1. In reality there will be a threshold saturation
of nonwetting phase required to bring about an initial network of connected pore and
pore throats across the sample volume of interest. This threshold value of saturation at
which the nonwetting phase first becomes continuous corresponds to the entry pressure
discussed in Section 3.2.3. During wetting, the relative permeability to DNAPL becomes
zero at Sw = 1 - Sm. Also, it is worth noting that the relative permeability to water
during wetting does not become unity because of the presence of residual DNAPL.
While it is possible to measure the relative permeability functions in the laboratory
(e.g., Johnson, et al., 1959; Jones and Roszelle, 1978; Sigmund and McCaffery, 1979;
Honarpour and Mahmood, 1988), it is often found convenient to calculate them from
Pc(Sw) data. Burdine (1953) and Mualem (1976) developed models that connect relative
permeability to the capillary pressure-saturation function. These models can be used in
conjunction with Eqs.(3.8) and (3.9) to derive closed-form expressions for krw(Sw) and
krnw(Sw). Brooks and Corey (1966) used Eq.(3.8) in the model of Burdine (\ 953) to
Sec. 3.4
107
Snw
1.;.;;0""""'"
1.0
k mw (drying)
Figure 3.8
curves.
Typical relative permeability
derive
k rw --
S(2+3A)/A
(3.23)
and
2+).
krn w = (1 - Se)2(1 - SeT)
(3.24)
By evaluating the Mualem (1976) model with Eq.(3.9), van Genuchten (1980) derived
(3.25)
For the relative permeability of the non wetting phase, Parker et at. (1987) used the same
procedure to write
(3.26)
Still other forms can be obtained by using the Brooks-Corey equation (Eq.(3.8 in the
Mualem model, or the van Genuchten equation (Eq.(3.9 in the Burdine model.
As noted above, the parameters A, m, and S, are determined by fitting Eq.(3.8) or
Eq.(3.9) to Pc(Sw) data. Thus, the rather difficult task of measuring the relative permeability functions is avoided. Values for A typically range from about 0.5 for materials
with a large range of pore sizes (e.g., aggregated clay) to perhaps 4.0 for very uniform
materials. A value of A = 2 is typical for sandy aquifer material. The corresponding
range for m is approximately 0.35 to 0.80. From a theoretical point of view, A can be
any number greater than zero, but m must be between zero and unity.
108
Chap. 3
3.5 THE EFFECT OF GROUNDWATER FLOW ON DNAPL MIGRATION

3.5.1 General
The migration of DNAPL is affected by gravity, the pressure gradients in both the
DNAPL and water phases, and the distributions of capillary and hydraulic properties of
the porous medium. Owing to the density of a DNAPL, the tendency is for the migration
path to be generally downward. However, heterogeneity has a marked influence on the
direction of DNAPL migration. A random distribution of permeability and displacement
pressure will result in a highly erratic pattern of DNAPL flow (Kueper and Frind, 1991a
and 1991b); lateral spreading well beyond the extent ofthe source area has been a striking
feature observed by these workers.
Paths of DNAPL migration which are readily-definable are created by permeability
and displacement pressure contrasts of regular geometry. For example, downward-moving
DNAPL may encounter a dipping stratum that it is unable to penetrate, so that it moves
downdip along the contact. Also, DNAPL pooled on the subcrop of an aquifer with
dipping bedding planes will preferentially enter the layers with low displacement pressure.
The result is downdip DNAPL migration in a series of distinct layers separated by
intervals entirely free of DNAPL (Figure 3.9). Such distributions of DNAPL have been
observed in rock cores, and also in porous media by Poulsen and Kueper (1992) and
Kueper et at. (1993).
Calculating the migration path of a DNAPL, even along a I-D flow path, requires
the solution of the full set of governing equations. However, results with significant
Ground Surface
~ WaterTable
'V
Figure 3.9
Migration of DNAPL in a sloping stratum subcropping in a pool of DNAPL.
Sec. 3.5
109
practical utility are obtained by examining situations in which groundwater flow and
capillary forces cause migration of DNAPL to cease. Similar analyses by Arthur D.
Little, Inc. (1982), Villaume (1985) and Hunt et al. (1988) are summarized by Mercer
and Cohen (1990).
The forces acting on a DNAPL may sometimes combine to cause the DNAPL to cease
migrating. We have already encountered one such circumstance when neither the DNAPL
nor the groundwater was in motion. It was necessary for the DNAPL to accumulate on a
low-permeability stratum in order for that situation to develop. The force on a DNAPL
due to flowing groundwater is now considered. One of the conclusions will be that
groundwater flow alone can arrest the motion of DNAPL.
We begin by re-writing Eq.(3.22) for DNAPL flow along a I-D path in an isotropic
medium, obtaining
_
kk rnw
qnwi - - - f.1nw
nw
(dP
dZ)
-d-- + Pnwg -
d Xi
Xi
(3.27)
where Z is the vertical coordinate, measured positively upward. No flow of DNAPL

occurs when the forces in parentheses add to zero. Since the effect of groundwater flow
is somewhat difficult to discern from Eq.(3.27), it is useful to replace P nw with the
equivalent P; + Pw. Then, upon noting that Pw = Pwg(H-Z), one obtains
qnwi
=-
kkrnwPwg d (
H
f.1nw
dx,
r; + ~p)
+ -Z
Pwg
Pw
(3.28)
wherein H is the familiar hydraulic head for water.

The quantity in parentheses is given the symbol <t>:
<t> = H
~p
Pwg
Pw
+ _c +-Z
(3.29)
Eq.(3.29) defines a force potential (Hubbert, 1940), the constancy of which means that
no DNAPL will flow. The negative gradient of <t> is the net driving force on the DNAPL.
Situations in which <t> is constant are examined in the following sections.
The downdip migration of DNAPL along a sloping contact or in laminae associated

with bedding planes may be arrested by updip groundwater flow. The situation to be
considered is one in which DNAPL has entered a sloping layer that subcrops in pooled
DNAPL as shown in Figure 3.9. The DNAPL has migrated to a distance, D, at which
point further downward motion is arrested by updip groundwater flow. Our purpose is
to compute the hydraulic gradient required to achieve this result and to calculate D.
Once DNAPL motion has ceased, the requirement that <t> be constant translates to
kk rw
qw = f.1w
(dP
-- c
de
.)
~pgsma
(3.30)
110
Chap. 3
in which qw is the groundwater flux and is a constant because we assume the flow to be
steady. The spatial coordinate is measured positive downward along the bed with dip
angle a. Because the groundwater flux is constant, qw is the same within and downdip of
the DNAPL body. However, the hydraulic gradient is constant only in the zone downdip
of the DNAPL. Variable water saturation and, therefore, variable relative permeability to
water results in a spatially variable hydraulic gradient within the DNAPL body.
The groundwater flux, evaluated with the hydraulic gradient as measured downdip
of the DNAPL is
qw
kPwg I1H
= - KI1H
- - = ----11
11
fl,w
(3.31 )
where K is the hydraulic conductivity and I1H/I1 is the constant hydraulic gradient
downdip of the DNAPL. An equation for the migration distance D is obtained by combining Eqs.(3.30) and and (3.31), followed by the integration
D = _1_ jP,(Ol
Pwg
Pd
dpc
I1H k - l
11 rw
f;.p.
p;;;
sm a
(3.32)
The lower limit of integration is the displacement pressure of the layer in which the
DNAPL is present and is the capillary pressure at the downdip extent of the DNAPL
body. The capillary pressure at the source is Pe(O) and is determined by the thickness of
the source pool.
Eq.(3.32) makes physical sense only when
I1H
I1p.
11
Pw
- - > -sma
(3.33)
This inequality constitutes the condition that must be satisfied if the motion of DNAPL is
to be arrested by the updip flow of groundwater. The result for vertical flow is obtained
of course with sin a = I.
The calculation indicated in Eq.(3.32) is performed by expressing kr w as a function of P; and carrying out the integration numerically or graphically. Figure 3.10 is a
dimensionless graph resulting from evaluation of Eq.(3.32) using
rw
= (
P:
)-5
(3.34)
The use of Figure 3.10 can be demonstrated by a specific numerical example.

Suppose the DNAPL in question has a density of 1.2 g/cm ', and is pooled to a depth
of 1.0 m on the subcrop of a stratum with Pd = 0.15 m of water. From Eq.(3.14) it
is determined that the capillary pressure at the base of the pool and on the face of the
subcropping stratum is Pe(O) = 0.2 m of water. Thus the ordinate value in Figure 3.10
is Pe(O)/Pd = 1.33. We will assume that the stratum dips at an angle of 3 degrees and
that the updip hydraulic gradient is 0.011. Then,
I1H (
Pw
) = 0.011 (
1.0
) = 1.05
11
I1p sino:
0.2 x 0.0523
Sec. 3.5
111
Figure 3.10 gives

DD.pg sina = 0.615
Pd
from which D is determined to be about 9 m. Note that the minimum updip hydraulic
gradient that would arrest the DNAPL migration is 0.0105. This shows that an updip
gradient only slightly greater than the critical value is sufficient to greatly limit the extent
to which the DNAPL will migrate downdip.
1.6
~H
--;;-, ( ~
1.4
.1
Pw
. ) =1.05
P Sll1 a
1.2
__ ~H (Pw
~I
0.4
0.6
0.8
1.0
)=/.0,
/),psina
1.2
D/),pg sino
Pd
Figure 3.10 Dependence of depth of DNAPL penetration on the upward hydraulic
gradient and capillary pressure on top of bedrock.
For the circumstances assumed in the previous section, the migration of the DNAPL will
be promoted by both gravity and a capillary pressure gradient. Once the DNAPL source
is depleted, however, the gradient of capillary pressure is reversed and opposes DNAPL
flow. DNAPL motion may now cease, even if groundwater flow is downward along the
migration path. Again, the constancy of the force potential leads to a force balance on
a static finger (ganglion) of length D residing in a sloping fracture or thin stratum. The
112
Mechanics I Mathematics of Movement of DNAPLs in Porous Media
Chap. 3
force balance is expressed as

b.p sin o
PeeD)
sn
---------=0
Pw
PwgD
D
(3.35)
The quantity b.H is the difference in hydraulic head across the finger and is positive
when the groundwater flow is updip. It has been assumed that the capillary pressure is
zero on the trailing end of the DNAPL body. The capillary pressure on the leading end
is PeeD).
Fingers of DNAPL that have migrated to positions of static equilibrium can be remobilized by pumping groundwater. For example, suppose a 5 m long finger of DNAPL
(Pnw = 1.2 g/crrr') has come to rest in a thin stratum (a = 10 degrees) as a consequence
of encountering a medium with Pd = 0.2 m of water. Suppose, also, that the groundwater
is static. According to Eq.(3.35), the capillary pressure at the leading edge is 0.174 m
of water. If nearby pumping of groundwater then induces downdip flow of water in the
stratum, a corresponding increase in capillary pressure will occur on the leading end
of the finger due to the reduction of the water pressure. Incipient remobilization of the
DNAPL will occur when PAD) = 0.2 m of water (i.e., PAD) = P d ) . Eq.(3.35) predicts that the finger will be remobilized by any downdip gradient (b.HID) greater than
0.0053.
The DNAPL fingers considered above are large relative to the representative element of volume for the porous media in which they reside. Residual DNAPL, which is
entrapped at a much smaller scale, is very difficult to mobilize by hydrodynamic forces.
The ratio of capillary forces opposing mobilization to the viscous forces promoting the
movement of residual is called the capillary number N; where
kpwg sn
Nc = - - - (J'
b.e
(3.36)
Wilson and Conrad (1984) reported that residual phase can begin to be mobilized at
N, ~ 2 X 10- 5 and that practically all residual becomes mobilized at N; ~ 1.3 X 10- 3 .
These values can be used to conclude that the hydraulic gradients required to mobilize
significant quantities of residual DNAPL in most types of porous media are unachievabIy
large.
3.6 DISTRIBUTION OF DNAPl IN UNSTEADY, HORIZONTAL flOWS

3.6.1 General
The simultaneous, horizontal flow of water and DNAPL is discussed in this section. The
design and performance assessment of well and drain fields intended for the recovery
of pooled DNAPL is one application of the type of analysis presented here. However,
such applications are not yet fully developed and remain an area of active research.
Analyses of the type discussed in this section have also found use for the verification of
numerical models. Because the flows discussed in this section are unsteady, their analysis
Sec. 3.6
113
is substantially more complicated than for the problems of fluid statics and steady flow
discussed above. The capillary pressure-saturation function, relative permeabilities, and
Darcy's Law remain applicable, but now two additional equations are required. These
equations express the conservation of mass for water and DNAPL.
The mathematical basis for the calculation of the simultaneous flow of DNAPL and
water in an aquifer is completed by the introduction of the relevant conservation laws.
A conservation law for each fluid, together with the flux equations of Section 3.4, result
in two strongly coupled partial differential equations. Considering the relations Pc(Sw),
krw(Sw), krnw(Sw), and the identity Sw+ Snw = 1, the mathematical statement consists of
eight equations and eight unknowns. These must be solved simultaneously subject to the
appropriate boundary and initial conditions.
In the present development, both the DNAPL and water are considered to be
incompressible and the porous medium is regarded as rigid. Furthermore, it is assumed
that DNAPL solubility is slight such that component partitioning across phase boundaries
can be neglected. Under these conditions, the conservation laws are
aqwi _ -<P asw
aXi
at
1, 2, 3,
(3.37)
and
aqnwi
aXi
-- =
asnw
at
-<P--
1, 2, 3.
(3.38)
As before, the repeated index i implies a summation over all i.

The left sides of these equations represents the divergence of the fluid flux and
consists of three terms in each case. Explicit reference to the fluid fluxes is eliminated by
simply combining the conservation laws with the fluxes given by Eqs.(3.21) and (3.22).
For example, by combining Eqs.(3.21) and (3.37), one obtains
/ - [kijk r w
aXi
fLw
(apw
aXi
PWgj )
= <P
as w
at
(3.39)
An identical result with the subscript w replaced by nw is obtained from Eqs.(3.22) and
(3.38). These results are supplemented by the auxiliary relations P, = Pnw - Pw, Pc(Sw),
krw(SllJ, krnw(Sw), and Sw+ Snw = I.
Addition of Eqs.(3.17) and (3.18) yields
-a (qwi + qnwi)
Xi
=0
(3.40)
which requires that the sum qwi+ qnwi be space invariant. This can be thought of as the
conservation law for the combined fluid system. The quantity qw+ qllW is known as the
total volume flux, q., and is at most a function of time. Any time dependency in qt must
derive from the boundary conditions imposed on the solution domain.
114
Chap. 3
While perhaps not apparent upon casual inspection, all of the problems analyzed
so far in this chapter are special cases of the above general flow equations. Specialization
for horizontal, unsteady flow is accomplished by eliminating the terms involving gravity.
The general differential equations reduce to (Fokas and Yortsos, 1982; McWhorter and
Sunada, 1990):
(3.41)
wherein
f(Sw) =
1 + krnwf.-Lw
krwf.-Lnw
)-1
(3.42)
kkrnwf d P;
D(Sw) = - - - - - f.-Lnw as;
q(t) = qw(x, t)
(3.43)
+ qnw(x, t)
(3.44)
Eq.(3.41) is written with the water saturation as the dependent variable. It can be written,
also, with DNAPL saturation or capillary pressure as the dependent variable.
The total flux qt must be prescribed (known) or, alternatively, calculated from the
boundary conditions. In the latter case, the required formula is
0~)=
Pw(O) - Pw(L) - J~'(6~\1
Jo (t~) + t::
L
f)dPe
(3.45)
dx
where L is the length of the flow domain, Pw(O) is the water pressure at x = 0, Pw(L)
is the water pressure at x = L, and Pe(O) and PeeL) are the corresponding capillary pressures. The derivation of Eq.(3.45) proceeds by substituting Darcy's Law into Eq.(3.44)
and integrating the result over the flow domain (Morel-Seytoux and Khanji, 1974). Evaluation of the denominator in Eq.(3.45) requires knowledge of the saturation distribution
because both k.; and knw depend upon Sw. Thus, Eqs.(3.41) and (3.45) must be solved
simultaneously.
3.6.3 The Buckley-leverett Solution
The classical solution for unsteady, horizontal flow was presented by Buckley and Leverett (1942). These authors were interested in the calculation of oil displacement by
injection of water. Buckley and Leverett (1942) recognized that when the injection rate
is large, the second term in Eq.(3.41) is dominant. They neglected the second order term
and solved the resulting first-order, partial differential equation to obtain
x(Sw, t)
f'i
=-
<I>
qt(r)dr
rv;
=<I>
(3.46)
Sec. 3.6
115
where f is the function defined by Eq.(3.42), i' = df /d Sw, Vw is the volume of water
injected, and <f> is porosity. The quantity x(Sw, t) denotes the coordinate of saturation Sw
at time t.
The function f is typically S-shaped, so l' is not single valued. Eq.(3.46) thus
predicts two different saturations for the same location x. This physically untenable
result is circumvented by the Welge (1952) tangent construction in which f is replaced
by the two part function:
f =
(1 +
krnwflw)-l
krwflnw
(3.47a)
(3.47b)
where S, is the uniform initial saturation. Accordingly, all saturations between S, and
Sb propagate at the same velocity given by qt!(Sb)/<f>(Sb - Si)' This propagation is in
the form of a shock front as shown in Figure 3.11 as the sudden drop in Swat some
distance x.
WELGE
TANGENT"
o s,
_____________ L-
1.0
o~-------------
Distance, x
Figure 3.11 Buckley-Leverett profile.
116
Chap. 3
The Buckley-Leverett solution is the limiting condition achieved when the impelling forces due to gradients of capillary pressure are negligible relative to the overall
gradient of fluid pressure induced by fluid injection on the boundary. Morel-Seytoux
(1969) presented a thorough discussion of the Buckley-Leverett analysis and demonstrated its use for oil recovery predictions. However, rarely will conditions of interest to hydrogeologists dealing with DNAPL problems be similar to those in which the
Buckley-Leverett equation applies. Nevertheless, the Buckley-Leverett solution remains
important for historical perspective and as a limiting case.
3.6.4 lateral Migration of DNAPl

Eq.(3.41) is a highly nonlinear differential equation, and few exact solutions are available.
Fokas and Yortsos (1982) have presented an exact closed-form solution subject to constant
qt and uniform initial saturation. Their solution was enabled by restricting i(Sw) such
that!, = aD with D = (bS w + c)-2, and a, b, and c are arbitrary constants. Unfortunately,
these forms do not exhibit some of the important features of real porous media, and the
resulting solutions are of limited use for the study of DNAPL invasion in aquifers.
Exact integral solutions for arbitrary D(SuJ and i(Sw) were derived by McWhorter
and Sunada (1990) for I-D and also for radial flows. Their solution for I-D invasion of
DNAPL into water saturated porous media is
Vnw I
x(Sw, t) = -$Fnw(Sw)
(3.48)
where Vnw is the accumulated volume ofDNAPL in the porous medium,

and the fractional-flow function F nw is given by
{So
F
(5)
1_
ls"
nw w
F~w
= dFnwldS w,
(f3 - Sw)D(f3) df3

Fnw(f3) - in(f3)
(Sw - l)D as;
{SO
II
(3.49)
Fnw - in
where in = 1- f The saturation So is the water saturation on the inflow boundary (i.e.,
at x = 0) and is constant. However, the value of So depends upon the prescribed value
of A through the implicit relation
A2
= _<t>
2
[SO (5 - I)D dS
w
Fnw - in
(3.50)
As before, x is the coordinate of saturation Swat time t. Eq.(3.48) is exact if Vnw varies
with time according to
Vnw = 2At 1/2
(3.51)
where A is an arbitrary constant.

While the solution as expressed in Eq.(3.48) is quite simple, the computation of
Fnw from Eq.(3.49) is not a trivial matter. The computation proceeds by selecting a value
for So and evaluating the integrals on the right side using Fnw = 1 as a first iteration. The
Sec. 3.6
117
result is a second estimate for Fnw(Sw) which is again used on the right side to produce a
third estimate of the fractional-flow function. The process is continued until convergence
is achieved. The integration is accomplished by a suitable numerical procedure.
The solution outlined above is useful in the exploration of the factors affecting the
extent and rate of lateral migration of a DNAPL. Figure 3.12 shows the fluid distributions calculated using Eqs.(3.8), (3.23), and (3.24) (a Brooks-Corey porous medium) and
using Eqs.(3.9), (3.25), and (3.26) (a van Genuchten porous medium). It is evident that
prescribing So = 0.8 for van Genuchten porous media with different values of m results
in greatly different distributions of DNAPL. Penetration of DNAPL in a van Genuchten
medium with uniform pore size (large m) is not nearly as great as for a wide distribution
of pore size (small m). It is observed that the DNAPL tends to penetrate in narrow elongated distributions. The more blunt penetrations in Brooks-Corey media result because
the ratio krnwlkrw is much smaller at large Sw than in van Genuchten media.
The sensitivity of DNAPL penetration to the capillary-hydraulic properties is also
revealed by computation of the injection constant A as a function of m with So = constant.
The results of such a computation are shown in Figure 3.13, indicating that the rate of
DNAPL invasion ranges over an order of magnitude, depending upon the value of m. A
small value of m results in a large value of A, other factors being equal.
Because the invasion of DNAPL into groundwater occurs in elongated distributions,
the invasion of even small volumes can result in extensive spreading. Also, the rate of
penetration given by qnwlCPSnw is inordinately large owing to the small values of Snw
1.Or-----~------:7------,_-----___=:::::=O-------~
0.8
0.6
~
If)
0.4
0.2
00
0.1
0.2
xt- l12
0.3
Figure 3.12 Penetration of a non-wetting phase affected by the pore-size distribution, with Jlw/Jlnw
0.4
2.
118
Chap. 3
10..--------------------------"\
So =0.6
So =0.8
-I
10
c::
<l:
/0-2
0:2
Figure 3.13
0.4
0.6
0.8
An as affected by m and So, with J1.w/J1.nw
= 2.
that prevail. Both the rate and extent of penetration are greater in porous media with
features that cause krnw/k rw to be large near Sw = 1.
The remarkable sensitivity of DNAPL penetration to the capillary-hydraulic properties can be expected to result in highly complex, seemingly chaotic, saturation distributions in heterogeneous media. For example, consider a porous medium comprised of
thin strata, each with different permeability, characteristic capillary pressure, and poresize distribution. The penetration of DNAPL in the individual layers will differ greatly,
much more so than would be the case for the penetration of a wetting phase. As noted
above, DNAPL in thin laminae separated by intervals of DNAPL-free media have been
Sec. 3.6
119
observed in cores. The distribution of DNAPL determined by averaging the saturation

across many such layers will appear especially diffuse.

In the petroleum industry, the injection of water ("water flood") to force oil to recovery
wells is a common practice. A similar process utilizing subsurface drains for the recovery
of DNAPLs is described by Sale et at. (1988). One-dimensional displacement such as that
described by Sale et al. (1988) can be analyzed with a mathematical solution for water
invasion that closely parallels that given above for DNAPL invasion. Of particular interest
is the recovery efficiency as indicated by the DNAPL fraction in the total produced fluid,
and the cumulative volume of DNAPL recovered as a function of volume injected.
A solution of Eq.(3.4l) presented by McWhorter and Sunada (1990) that forms a
basis for analyzing the efficiency of a linear (1-D) water flood is
V (I - 1,.)
W
<t> I F'w(S w)
x (S w, t ) --
(3.52)
where Vw is the volume of water injected. As before, the function F;,JSw) is the derivative
of Fw(Sw) with respect to Sw where F w is now defined as
F _ qw/qt - Ii
wI-Ii
(3.53)
with f; = j(Si) and Si = the initial water saturation. The fractional flow function F w is
calculated by iteration using
S0
F (S ) = I _
w
w
l[SO
Fw(f3) - In(f3)
(Sw - Si)D
1,
Fw(f3) - In(f3)
Sill
(f3 - Sw)D(f3) df3
as;
(3.54)
with In = if - f; )/(1- f;).

We consider an injection plane located a constant distance L from a recovery plane.
The volume of DNAPL displaced across the recovery plane is
v., =
<I>
(3.55)
(Sw - Si)dx
Substituting Eq.(3.52) into Eq.(3.55) followed by an integration by parts eventually yields
v:nw
<t> (1 - Si) L -
(V)(I-f,.)
~
<t> L
_ _'
I _ S,
S
+ (w
- S F' - F
,) w
w)
(3.56)
in which it is understood that Sw is the water saturation at x = L. The volume of water

injected is related implicitly to Swat x = L by
(3.57)
120
Chap. 3
Finally, the fraction of DNAPL in the total produced fluid is obtained from a rearrangement of Eq.(3.53) so that
-qnw =
qt
(3.58)
(l - /;)(1 - F(Sw
Table 3.1 contains results of calculations obtained by the above procedure for the
performance of a water flood. The functions Fw(Sw) and F/(Sw) were calculated as
described in detail by McWhorter and Sunada (1990). The values of V u/ (<1> (l - Si) L)
in Table I are the injected water volumes expressed as a fraction of the pore volume.
The volume of DNAPL recovered as a fraction of the total volume initially present
is Vnw/ (<1> (l - si)L). Recovery of 100% of the initial DNAPL will never be possible
because of entrapment of some residual DNAPL by the invading water. The theoreticallyrecoverable volume per unit area is <1>(Sm - S, )L, where Sm < I because of the entrapped
residual DNAPL.
Performance of DNAPL Recovery by Water Flood. Input
m = 0.5 in Eqs.(3.9L (3.25), and
(3.26); fLw/ll nw = 2; residual DNAPL = 1- sm = O. 10.
TABLE 3.1
Data:
sm = 0.90; 5; = 0.75; 5, = 0.2;
511'
Fw(Sw)
F:(Sw)
Vw/<PL
V,,,,'/<P( I -S;)L
qnw/q,
0.750
0.753
0.764
0.778
0.792
0.814
0.848
0.876
0.R82
0
0.044
0.189
0.346
0.484
0.674
0.887
0.985
0.995
14.59
11.99
10.48
9.33
7.66
4.91
1.96
0.257
0
0.094
0.114
0.130
0.146
0.178
0.278
0.695
1.091
0
0.276
0.327
0.361
0.389
0.426
0.485
0.534
0.544
0.733
0.701
0.595
0.479
0.378
0.239
0.083
0.011
0.004
The data in Table 3.1 show that initially, the produced fluid contains about 73%
DNAPL. However, by the time about one-third pore volume of water has been injected,
the produced fluid contains less than 10% DNAPL. At one pore volume of injected
water, the produced fluid contains less than 0.5 percent DNAPL. About 54% of the total
DNAPL and about 90% of the theoretically-recoverable DNAPL have been recovered
by injection of one pore volume of water. The ~ 10% of the theoretically recoverable
DNAPL remaining is virtually unrecoverable because of the large injection volumes
required to displace it. In other words, the displacement efficiency falls to essentially
zero for injected volumes greater than one pore volume.
The performance of a water flood for the purposes of DNAPL recovery is rather
sensitive to the pore-size distribution parameter m in Eqs.(3.9), (3.25), and (3.26). The
sensitivity derives from the strong dependence of the relative-permeability ratio krnw/k rw
upon m. McWhorter and Sunada (1990) show that this ratio is about one order-ofmagnitude greater for m = 0.5 than for m = 0.8 in the saturation range of relevance for
Sec. 3.6
121
1.0
0.8
'" (I-S)l
--------m=0.5
m=0.6
---06
Vw
08
'.0
1.2
~
Figure 3.14
The sensitivity of water flood performance to pore-size distribution.
DNAPL recovery by water flood. Figure 3.14 depicts the substantial change in performance that is caused by increasing m from 0.5 to 0.6.

Injection of water through wells for the purpose of displacing DNAPL to recovery wells
is an alternative to the I-D water flood discussed above. Injection and recovery wells
are easier to install and operate than are the horizontal drains required to effect I-D
displacements. The expected recovery efficiency under various operating conditions can
be calculated as follows.
The water saturation at any radial distance, r, from an injection well is given by
(McWhorter and Sunada, 1990)
r(Sw, t)
Q t (l - f i ) I
rrb<l>
Fw(Sw)t
)1/2
(3.59)
in which Qt is the injection rate at the well, b is the thickness of the injection zone, and
the fractional flow function Fw(Sw) is defined by
F
_ Qw - fiQt
w - Qt(l - Ii)
(3.60)
The discharge of water across the lateral surface of a cylinder of radius r and height b
is Qw. The value of Qw ranges from Qt at the well to f;Qt at large distances from the
injection well. Thus, the fractional-flow function ranges from 0 to I.
122
Chap. 3
The fractional-flow function is calculated by iteration in
j s,
exp
Sm
exp
(-4rrh
Q/(l - J;)
jf3
I _
Q/(
h)
as; )
f3-4rrb
s,
D(f3)
s, F w(f3) - In(f3)
1,
df3
(3.61)
) df3
D(f3)
dS
F w (f3 ) - In(f3)
w
As before, In = (f - JJ/(l - Ji), f, and D are calculated from the capillary-hydraulic

properties as indicated in Eqs.(3.42) and (3.43).
The above equations represent an exact solution for the differential equation that
is Eq.(3.41) as modified for radial flow. The injection rate is a constant in this case. The
volume of DNAPL Vn w that is displaced across a surface of radius R is calculated from
V
nw
<P(l- Si)rrR 2b -
(1-+)(
J_'I
1- S,
Vw
<prrR2b
+ (w
-r
S. F'-F
,) w
w)
(3.62)
with
_ [(1 _ +)F' ]-1
Vw
<Prr R2b .II
w
(3.63)
Again we have expressed the volume of displaced DNAPL as a fraction of the initial
DNAPL volume, and the injected volume as a fraction of the pore volume. In this respect,
Eqs.(3.62) and (3.63) are identical to Eqs.(3.56) and (3.57) for a I-D water flood; the
numerical values of F w are different, of course, for the two cases.
The performances of two water floods conducted under identical conditions except for injection rate are represented in Figure 3.15. The dimensionless injection rates
indicated on the figure refer to
Qn
Q/(l - J;)/.Lw
(Sm - Si)kPo
(3.64)
where k is the intrinsic permeability and Po is a characteristic capiIlary pressure. The

parameter Qn has the physical significance of being the ratio of viscous forces to the
capillary forces. It has been observed that water flood performs better when this ratio is
large (i.e., large injection rates).
For a given aquifer, the injection rate that can be achieved is dictated by the
pressure difference between the injection and recovery wells and the well spacing. The
relationship among these factors is given to a rough approximation by
Zn kbk r w (S, ),6. P

Q/=-----/.Lw In(R/r w)
(3.65)
where k r w (Si) is the relative permeability to water at the initial water saturation and ,6.P
is the pressure difference. The magnitude of ,6.P, expressed as a height of water, will be
on the order of the thickness of the DNAPL layer. One can then adjust the spacing R to
achieve a target injection rate that will promote an efficient displacement process.
Sec. 3.7
123
1.0
0.8
Q n =/.0
Q n =0.25
0.6
/2
Figure 3.15 The performance of two radial water floods.

The above remarks include no consideration of costs. Small spacing will promote
efficient displacement, but will cost more than a large spacing. A computation of performance with different values of R and an expression of cost as a function of R will permit
the computation of a performance vs. cost curve. A decision on spacing that strikes a
reasonable balance between performance and cost is then possible.
3.7 STABILITY OF TWO-flUID DISPLACEMENTS

When a DNAPL penetrates the water table, water is displaced by a liquid which is denser
and often less viscous than water. Such displacements give rise to a gravity and/or viscosity induced instability we have referred to as "fingering", and results in a chaotic and
unpredictable propagation of the macroscopic interface separating the fluids of interest.
Figure 3.16 illustrates the fingering process for the displacement of a more viscous liquid by a less viscous liquid. As shown, the originally planar interface separating the two
liquids becomes increasingly distorted with travel distance. In general, the fingering process occurs whenever the displacement is characterized by an unfavorable density ratio
such as the downward displacement of a light liquid by a heavier liquid, or the upward
displacement of a heavy liquid by a lighter liquid, and in cases where the viscosity of
the displacing fluid is less than that of the displaced fluid. In cases where one of either
the density ratio or viscosity ratio is favorable and the other is not, a critical velocity
determines whether or not the displacement will be stable.
It is generally accepted that the fingering process is triggered by microscopic heterogeneities such as those encountered at the pore scale. Studies illustrate that the spacing
124
Chap. 3
...
Figure 3.16 Fingering in a Hele-Shaw cell. The initial planar interface originated at
the left side of the cell. Shown are four subsequent locations of the advancing interface.
(Reprinted with permission from Perkins and Johnston, 1969.)
of fingers evolving in homogeneous media is a function of the porous medium properties,

the fluid properties, and the velocity of the advancing interface. The fingering model, first
introduced by Chuoke et al. (1959), assumes that the medium through which the fluids
are moving is perfectly homogeneous. This may well be approximated in the laboratory,
but in the field such media rarely exist. Within heterogeneous media, macroscopically
planar interfaces will not exist between the fluids of concern. Stability analyses yielding
optimum wavelengths of finger initiation will be invalid since heterogeneous media do
not provide an infinite number of wavelengths from which to initiate fingering. Fingering
will simply be initiated in the most permeable regions which exhibit the least resistance
to the advancing front.
Exactly what degree of permeability variation is required to invalidate the homogeneous instability model is unclear, but experiments (Gupta et al., 1973; Schwille, 1988)
seem to indicate that it is very slight. If the porous medium of interest has macroscopic
heterogeneities of a length scale equal to or less than the scale of the critical wavelengths
required to initiate fingering, then no fingering by definition actually takes place. The
highly irregular fluid distributions that result are due to local permeability variations, and
should perhaps be termed channeling to distinguish them from similar fluid distributions
that can arise in theoretically homogeneous media.
It should be noted that situations may occur where gravity- and viscosity-induced
instabilities dominate over the influence of heterogeneity in determining the migration
pathways of an immiscible contaminant. One such situation involves the case of low or
negligible interfacial tension. Under such conditions, for a given set of fluid parameters
and flow conditions, it would be expected that there will be a threshold degree of heterogeneity characterized by the variance and structure of the permeability field below which
the fingering process dominates. Also, in situations where the fluid properties of interest
and the prevailing flow conditions bring about a fingering wavelength below the scale of
heterogeneity present, then the fingering process may be influential in determining the
fate of an immiscible contaminant. One such situation may arise in a stratified porous
Sec. 3.8
Conclusions
125
medium having individual beds of a much larger horizontal extent than vertical extent.
In such a case a great deal of lateral flow may occur above finer grained lenses, and this
flow process would be subject to viscosity-induced instability along the plane of bedding.
The fingering process is described further by Kueper and Frind (1988).
In this chapter we have assumed that interfacial tension forces are not negligible
and that the length scale of heterogeneity in the porous medium of interest is on the order
of the wavelengths of fingering. This allows use of the governing equations presented
without modification to account for possible fluid instabilities. In practice, it is likely that
the effects of heterogeneity will dominate the fate of DNAPL in field situations. When
performing laboratory experiments in homogeneous media, however, explicit account of
possible fluid instability should be taken.
3.8 CONCLUSIONS
The displacement of groundwater in the pores of an aquifer by an invading DNAPL

can never be accomplished completely by physical means alone: unless there is enough
DNAPL to allow all of the residual water to dissolve into that DNAPL, water and
DNAPL will coexist in the interstitial pore space where DNAPL is present. The volumes
of DNAPL and water present in the aquifer per unit of pore volume is characterized by
the empirical capillary-pressure curve for the medium. This relationship is of fundamental
importance to understanding and quantifying the movement and distribution of DNAPLs
in aquifer systems.
A threshold capillary pressure, termed the entry pressure, must be attained for a
DNAPL to become continuous and capable of flow through a porous medium that is
otherwise saturated with water. The entry pressure is highly variable in heterogeneous
aquifers, and results in seemingly-chaotic patterns of flow, as well as randomly-distributed
pools and trails of residual DNAPL. A pool will form on top of a textural discontinuity when the capillary pressure is insufficient to exceed the entry pressure of the next
material encountered. The thickness of a DNAPL pool required to initiate penetration of
an obstructing stratum is inversely proportional to the buoyant density of the DNAPL.
Fluids with densities near that of water are likely to form thick pools. The relative rarity
of thick pools of chlorinated solvents is explained by the fact that even thin pools of
these very dense liquids can result in significant capillary pressure.
Analyses of flow problems involving DNAPL are based on Darcy's Law written
for each fluid. The permeability to the individual fluids is a strong function of the fluid
contents, and therefore of the capillary pressures. Forces acting to cause DNAPL flow can
be expressed in terms of the familiar gradient hydraulic head in the aqueous phase, the
gradient of capillary pressure, and the buoyant density of the DNAPL. This approach for
expressing the driving forces on a DNAPL phase leads to a particularly simple and general
expression of the conditions under which upward hydraulic gradients in groundwater are
sufficient to halt the downward migration of DNAPL that would otherwise occur. The
same concept sets out the conditions under which a static DNAPL ganglion may be
mobilized by modifying the gradient in the groundwater.
126
Chap. 3
Calculation of unsteady flow of DNAPL can show the remarkable sensitivity of

such flow to the capillary-hydraulic properties of the porous medium. Even in the idealized case of a perfectly homogeneous medium, DNAPL can be expected to penetrate
in the form of narrow, elongated distributions in which the mean saturation of DNAPL
is small. The result is that even small volumes of DNAPL can be expected to result
in extensive spreading. Also, the small DNAPL saturation behind the advancing front
results in large DNAPL seepage velocities. This, in combination with a large density and
low viscosity, can cause rapid DNAPL migration.
ACKNOWLEDGEMENTS
3.9 REFERENCES
Adam, N. K. (1968) The Physics and Chemistry ofSurfaces, Dover Publications, N.Y., Unabridged
and corrected republication of the third (1941) edition.
Anderson, M. R. (1988) The Dissolution and Transport of Dense Nonaqueous Phase Liquids In
Saturated Porous Media, Ph.D. Thesis, Oregon Graduate Institute, Portland, Oregon.
Anderson, W. G. (1986) "Wettability literature survey. Wettability measurements', J. Petrol. Technol., 38, 1246-1262.
Arthur D. Little, Inc. (1982) Capillary Pressure and its Effect on Non-Aqueous Phase Liquid Migration and Containment. Report to Wald, Harkrader & Ross, Washington, D.C., prepared for
the Environmental Protection Agency by Arthur D. Little, Cambridge, Massachusetts, March.
Buckley, S. E. and M. C. Leverett (1942) "Mechanism of fluid displacement in sands", Trans.
AlME, 146, 107-116.
Burdine, N. T. (1953) "Relative permeability calculations from pore-size distribution data", Trans.
AlME, 198,71-77.
Brooks, R. H. and A. T. Corey (1966) "Properties of porous media affecting fluid flow", J. Irrig.
Drain. Div., Am. Soc. Civil Eng., 92, 61-88.
Chatzis, I. and F. A. Dullien (1983) "Dynamic immiscible displacement mechanisms in pore
doublets: theory vs. experiment", J. Colloid and Interface Sci., 91, 199-222.
Chuoke, R. L., P. van Meurs, and C. van der Poel (1959) "The instability of slow, immiscible,
viscous liquid-liquid displacements in permeable media", Trans. AIME, 216, 188-194.
Corey, A. T. (1986) Mechanics of Immiscible Fluids In Porous Media, Water Resources Publications, Littleton, Colorado, 259 p.
Fokas, A. S. and Y. C. Yortsos (1982) "On the exactly solvable equation occurring in two-phase
flow in porous media. SIAM J. Appl. Math., 42, 318-332.
Sec. 3.9
References
127
Gupta, S. P., J. E. Varnon, and R. A. Greenkorn (1973) "Viscous finger wavelength degeneration
in Hele-Shaw models", Water Resources Research, 9, 1039-1046.
Honarpour, M. and S. M. Mahmood (1988) "Relative permeability measurements: an overview",
J. Petrol. Technol., 40, 903-966.
Hubbert, M. K. (1940) 'The theory of groundwater motion", J. Geology, 48, 785-943.
Hunt, J. R., N. Sitar, and K. S. Udell (1988) "Non-aqueous phase liquid transport and cleanup 1.
Analysis of mechanisms", Water Resources Research, 24, 1247-1258.
Johnson, E. F., D. P. Bossler, and V. O. Nauman (1959) "Calculations of relative permeability
from displacement experiments", Trans. A/ME, 216, 370-372.
Jones, S. C. and W. O. Roszelle (1978) "Graphical techniques for determining relative permeability
from displacement experiments", Society of Petroleum Engineers, 6045, SPE-AIME Meeting,
New Orleans, Oct. 3-6.
Kueper, B. H. and E. O. Frind (1988) "An overview of immiscible fingering in porous media", J.
Contaminant Hydrology, 2, 95-110.
Kueper, B. H. (1989) The Behaviour of Dense, Non-Aqueous Phase Liquid Contaminants in Heterogeneous Porous Media, Ph.D. Thesis, University of Waterloo, Waterloo, Ontario.
Kueper, B. H. and E. O. Frind (199Ia) 'Two-phase flow in heterogeneous porous media,!. Model
development", Water Resources Research, 27, 1049-1057.
Kueper, B. H. and E. O. Frind (1991b) 'Two-phase flow in heterogeneous porous media, 2. Model
application", Water Resources Research, 27, 1059-1070.
Kueper, B. H., D. Redman, R. C. Starr, and S. Reitsma (1993) "A field experiment to study the
behaviour of tetrachloroethylene below the water table, Ground Water, 31, 756-766.
Laliberte, G. E., R. H. Brooks, and A. T. Corey (1968) "Permeability calculated from desaturation
data", J. /rrig. and Drainage Div., Am. Soc. Civil Eng., 94, Proc. Paper 5843, 57-71.
Leverett, M. C. (1941) "Capillary behavior in porous solids", Trans. A/ME, 142,152-169.
Luckner, L., M.Th. van Genuchten, and D. R. Nielsen (1989) "A consistent set of parametric models
for the two-phase flow of immiscible fluids in the subsurface", Water Resources Research, 25,
2187-2193.
Mualem, Y. (1976) "A new model for predicting the hydraulic conductivity of unsaturated porous
media", Water Resources Research, 12, 513-522.
McWhorter, D. B. and D. K. Sunada (1990) "Exact integral solutions for two-phase flow", Water
Resources Research, 26, 399-413.
McWhorter, D. B. and J. D. Nelson (1980) "Seepage in the partially saturated zone beneath tailings
impoundments", Mining Eng., April, 432-439.
Mercer, J. W. and Cohen, R. M. (1990) "A review of immiscible fluids in the subsurface: properties,
models, characterization and remediation", 1. Contaminant Hydrology, 6, 107-163.
Mohanty, K. K., H. T. Davis, and L. E. Scriven (1987) Physics of oil entrapment in water-wet
rock", Society of Petrol. Engineers, Reservoir Eng., 2, 113-128.
Morel-Seytoux, H. J. (1969) "Introduction to flow of immiscible liquids in porous media in flow
through porous media". Chapter II in Flow Through Porous Media, R. J. M. de Wiest (Ed.),
Academic Press, New York, New York, pp. 456-516.
Morel-Seytoux, H. J. and J. Khanji (1974) "Derivation of an equation of infiltration", Water Resources Research, 10, 795-800.
128
Chap. 3
Parker, J. c., R. J. Lenhard, and T. Kuppansamy (1987) "A parametric model for constitutive
properties governing multiphase flow in porous media", Water Resources Research, 23, 618624.
Perkins, T. K. and O. C. Johnston (1969) "A study of immiscible fingering in linear models", Soc.
Petrol. Eng. J., March, 39-46.
Poulsen, M. and Kueper, B. H. (1992) "A field experiment to study the behaviour of tetrachloroethylene in unsaturated porous media", Environ. Sci. Technol., 26, 889-895.
Sale, T., D. Steib, K. Piontek, and R. Kuhn (1988) "Recovery of wood-treating oils from an alluvial
aquifer using dual drainlines". In: Proceedings: Conference on Petroleum Hydrocarbons and
Organic Chemicals in Ground Water: Prevention, Detection and Restoration, National Water
Well Assoc./American Petroleum Institute, Nov. 9-11, Houston, Texas.
Schwille, F. (1988) Dense Chlorinated Solvents In Porous and Fractured Media - Model Experiments, translated by J. F. Pankow, Lewis Publishers Inc., Boca Raton, Florida.
Sigmund, P. M. and F. G. McCaffery (1979) "An improved unsteady-state procedure for determining the relative permeability characteristics of heterogeneous porous media", Soc. Petrol. Eng.,
Journal, 19, 15-28.
Stegmeir, G. L. (1974) "Relationship of trapped oil saturation to petrophysical properties of porous
media", Society of Petroleum Engineers, 4754, Improved Oil Recovery Symposium, Tulsa,
Oklahoma, April.
Thomas, L. K., D. L. Katz and M. R. Tek (1968) "Threshold pressure phenomena in porous media",
Soc. Petrol. Eng. J., 8, 174-184.
van Genuchten, M. Th. (1980) "A closed-form equation for predicting the hydraulic conductivity
of unsaturated soils", Soil Sci. Soc. Am. J., 44, 892-898.
Villaume, J. F. (1985) "Investigations at sites contaminated with dense nonaqueous phase liquids",
Ground Water Monitoring Review, 5, 60-75.
Welge, H. J. (1952) "A simplified method for computing oil recovery by gas or water drive",
Trans. AIME, 195, 91-98.
Wilson, J. L. and S. H. Conrad (1984) "Is physical displacement of residual hydrocarbons a realistic possibility in aquifer restoration?". In: Proceedings: NWAAlAPI Conference on Petroleum
Hydrocarbons and Organic Chemicals In Groundwater, Houston, Texas, National Water Well
Association, Worthington, Ohio, pp. 274-298.
4
Numerical Simulation of the
Migration of Dense
(DNAPLs) in Porous Media
Bernard H. Kueper 1 and Emil O. Frind-'
1 Department of Civil Engineering
Queen's University Kingston, Ontario, Canada K7L 3N6

University of Waterloo, Waterloo, Ontario, Canada N2L 3G1
Page
4.1
4.2
4.3
4.4
4.5
4.6
Introduction
Model Usage
Example
Conclusions
References
129
130
137
139
142
143
ABSTRACT
Numerical models provide an essential tool for decision making, assessment, and design
of remediation systems at hazardous waste sites. The steps of the modeling process,
from conceptual design to modeling implementation are described including examples
and a case history. The reliability and usefulness of model results are dependent upon
the degree of site characterization. In most cases, a single, site-specific prediction cannot
be obtained, but rather an envelope of possible behavior.
4.1 INTRODUCTION
Upon release at the ground surface, a dense non-aqueous phase liquid (DNAPL) will
migrate through the subsurface according to a variety of complex and inter-related processes. In both saturated and unsaturated deposits, a DNAPL can distribute itself in
"residual" form as disconnected blobs and ganglia, and also as larger accumulations
referred to as pools. Residual DNAPL is formed at the trailing edge of a migrating
DNAPL body due to pore-scale snap-off and trapping mechanisms, and is extremely difficult to displace by hydraulic means alone (see Chapter 3). DNAPL pools, on the other
129
130
Numerical Simulation of the Migration of DNAPLs
Chap. 4
hand, form on top of capillary barriers and are relatively easily mobilized through either
manipulation of hydraulic gradients, or by a lowering of the DNAPLIwater interfacial
tension. It is known that the migration pathways of DNAPL are governed primarily by
heterogeneity and porous media structure, with subtle variations in permeability capable
of causing significant degrees of lateral spreading, and migration patterns which are both
sparse and tortuous.
Below the water table, residual and pooled DNAPL will slowly dissolve into flowing groundwater, giving rise to evolving plumes of contaminated groundwater. The magnitude of the concentrations arising in these plumes is a function of the chemical composition of the DNAPL phase, as well as the degree of surface area available for interphase
mass transfer. It is known that residual and pooled DNAPL may not always be directly
accessible to flowing groundwater, in which case the dissolution process becomes diffusion limited. This results in an inherently slow mass transfer process, and is often the
cause of the well-known tailing phenomenon observed in remediation efforts. Above the
water table, residual and pooled DNAPL will lead to the evolution of a vapor plume,
which can in tum transfer contaminants directly to the water table, or to infiltrating water. In cases where DNAPL is trapped in layers of high water content, or in fine-grained
material, the transfer of components to the air phase may also be diffusion controlled.
The above processes can be described through numerical simulation. A numerical
model is a tool that allows integration of all the processes which are relevant in a
given situation, and allows analysis of the behavior of a system under a wide range of
conditions. Since field experimentation with hazardous immiscible liquids is prohibited
by law in many regulatory jurisdictions, numerical modeling may offer the only means
of gaining insight into field-scale processes and remedial actions at real sites. Some
parameters, such as field-scale dispersivities and mass transfer coefficients, are difficult
to measure directly and must be determined by fitting field data to models. Also, through
the use of a numerical model, a variety of "what if" scenarios and worst case conditions
can be examined for a particular system. Individual processes can be turned on and off to
determine their importance. Moreover, by means of Monte Carlo methods and similar
statistical approaches, the uncertainty associated with a given event can be estimated.
The understanding which can be gained from modeling can assist in anticipating
the consequences of a spill, and can help judge the prospect of being able to clean it up.
Modeling, by virtue of the integrated insight it provides, can be an important component
of the overall regulatory, environmental, economic, and political decision making process
at a particular site.
4.2 THE MODELING PROCESS
The modeling process begins by defining the modeling objectives. The specific objectives
to be met will be site specific and may seek to answer one or more of the following
questions:
What is the relative importance of the various processes and parameters?
What are the possible worst-case scenarios?
Sec. 4.2
131
What are the possible migration paths and travel times of contaminants?
What are the arrival curves of contaminant concentration at critical points under
various scenarios?
Will the contaminant be cleaned up in a reasonable time frame by natural processes
alone?
How effective will remediation efforts be in cleaning up the aquifer, and could
attempts at remediation actually worsen the extent of contamination?
Once the modeling objectives have been clearly identified, a conceptual model
must be formulated. A conceptual model is an idealization of the real system incorporating
the relevant processes and degree of coupling required between these processes. The
underlying assumptions for including certain processes and excluding others should be
carefully considered and the implications of these assumptions should be fully understood
by the model user. For example, if the tailing associated with a soil vacuum extraction
system is of interest, then the possible presence of low permeability lenses and DNAPL
in zones of high water content must be incorporated; conceptualizing such a site as a
homogeneous domain would be inappropriate.
The conceptual model underlying a given numerical model incorporates information
on the extent, configuration, variability, and properties of both the contaminant and the
hydrogeological units. This information is obtained from field investigations, and must be
of sufficient quality and quantity so that a valid conceptual model can be formulated. The
conceptual model determines what spatial dimensions are necessary to meet the modeling
objectives. Since all real sites are three-dimensional and a DNAPL can spread in three
dimensions, a 3D model will in almost all cases provide the most complete representation.
However, 3D models are expensive to operate, and it is therefore desirable to choose the
fewest dimensions that will satisfy the stated objectives. A model incorporating DNAPL
migration, for example, should always include the vertical dimension to account for the
action of gravity. Complex aquifers and a desire to study the influence of small-scale
heterogeneity will require a full 3D model. If the spill in question is small and the
aquifer is relatively homogeneous, a 2D axisymmetric model including the radial and
vertical dimensions may suffice. Similarly, to simulate the overall migration of a plume
of dissolved DNAPL in a thin confined aquifer, a 2D plane model may be adequate.
The time scale of the various processes of interest must also be considered in
the selection of an appropriate conceptual model. At many sites, downward DNAPL
migration and the formation of residual and pooled DNAPL may take place within days
or months, while the dissolution process may carryon for decades and possibly centuries.
In such situations it would be appropriate to decouple the migration of the immiscible
phase from the evolution of a dissolved plume, with a separate numerical model used to
describe each. The decoupling of processes will, in virtually all cases, greatly simplify
the modeling effort, leading to cost savings.
Once a conceptual model has been formulated, a mathematical model must be
created. At this step, physical processes are formulated in the form of a boundary value
problem which comprises the governing partial differential equations, relevant consti-
132
Chap. 4
tutive relationships, material properties, boundary conditions, and initial conditions. In

the most general case, the equations governing the migration of DNAPL, air, and water,
subject to mass transfer across phase boundaries, can be assembled from the following
mass balance equation for chemical component k in phase ex:
a
a
-(EaPaWk)
at
a
a
-(EaPaWk Va)) aXi
a
a
-Jk,i
aXi
= Ik
+ Ek
(4.1 )
where Ea is the fraction of bulk volume occupied by the a phase, Pa is the average
mass density of the a phase, wf is the mass fraction of component k in the a phase,
Va,i is the velocity of the a phase in the i -th direction, Jk,i is the non-advective flux
of k in the a phase in the i -th direction, If: represents the transfer of k due to phase
change and diffusion across phase boundaries, and Ef is the external supply of k to the ex
phase through biotic and abiotic transformations. It follows from Eq.(4.1) that the mass
fractions within a given phase must sum to unity, i.e.,
n
Lwr = I
(4.2)
k=1
where n is the total number of components comprising phase a. It also follows that the
fractional phase volumes sum to unity, i.e.,
N
LE
=1
(4.3)
a=1
where N is the total number of phases, in the most general case being four (air, water, DNAPL, and solid). The following expresses the conservation of mass of a given
component distributed amongst the phases:
N
LJf:=1
(4.4)
a=1
Eq.(4.l) requires that the phase velocities, Va,i, be defined. For saturated groundwater
flow these are usually obtained from Darcy's Law. Based on laboratory experiments
performed in the petroleum industry (Wyckoff and Botset, 1936; Leverett, 1938), it
has been suggested that Darcy's Law can be extended to DNAPL systems as well.
Others (e.g. Allen, 1985) have shown that Darcy's Law can be extended to multiphase
systems by considering a momentum balance for each phase of interest. Such studies
show Darcy's Law to be valid if inertial effects are negligible and there is no transfer of
momentum between phases. It appears that in most cases an extended form of Darcy's
Law is applicable to DNAPL systems in order to define the phase velocities required in
Eq.(4.1). Darcy's Law extended to muitiphase systems may be expressed as
(4.5)
Sec. 4.2
133
where Pa is the pressure of the ex phase, g is the gravitational constant, z the vertical
direction, and Kfj a second rank mobility tensor given by
k k
Ka.=~
!.La
t,j
(4.6)
where !.La is the dynamic viscosity of the ex phase, ki . j is a second rank intrinsic permeability tensor, and kr,a the relative permeability to the ex phase accounting for the
reduction in permeability due to the presence of other phases.
Equation (4.1) requires that the non-advective flux vector, lk,i' be defined. This
vector accounts for the molecular diffusion and mechanical dispersion of component k
within a given phase. It is often assumed that both of these are Fickian in nature in
which case the processes are additive leading to
a
k a a (Pawk)
k a a(Pawk)
lk i = -caTa,ijDo '
- caD,,; i]
aXj'
aXj
,
(4.7)
where Ta,ij is a tensor of phase tortuosity coefficients, D~,a is the free-solution molecular
diffusion coefficient of k in ex, and D~,~j a mechanical dispersion tensor of k in ex.
The proper form of the mechanical dispersion tensor for a given scale of interest, flow
conditions, and saturation distribution is not always obvious. For solute transport in
groundwater, various forms of the dispersion tensor have been proposed as discussed by
Bear (1972) and Gelhar (1986).
In order to complete the set of equations required to represent DNAPL migration in
the subsurface, the various phase pressures must be coupled through capillary pressure
relationships. For a general air-water-DNAPL system, the following capillary pressures
can be defined:
p~.W(Sa,
Sw)
Pa - Pw
(4.8)
p~.n(Sa,
Sn)
Pa - P;
(4.9)
P~~,W(Sn,
Sw)
P; - P11I
(4.10)
where the scripts a, nand ware taken to represent air, NAPL, and water, respectively.
The phase saturations, Sa, are expressed as a fraction of pore space and are related to
the phase volume fractions by Sa = caN where is the porosity of the medium.
The relationships in Eqs.(4.8)-( 4.10) are typically determined experimentally for
the media and fluids of interest. It is known that these relationships vary spatially at
real sites, often at the scale of centimeters (e.g. see Poulsen and Kueper, 1992). For the
case of two-phase flow, various workers have proposed empirical relationships allowing
a mathematical function to be fit to experimental data (e.g. Brooks and Corey, 1966;
van Genuchten, 1980). Leverett (1941) presented a relationship demonstrating that capillary pressure curves can be scaled as functions of hydraulic conductivity and interfacial
tension. The use of interfacial tensions to scale capillary pressure curves has also been
demonstrated by Schiegg (1984) and Parker et at. (1987).
Where three fluid phases are present, it has been suggested that two-phase capillary
pressure relationships can be used to approximate three-phase capillary pressures (e.g.
Leverett, 1941). These studies assume that the total liquid saturation given by ST = SN +
134
Chap. 4
Sw is a function of the air-DNAPL capillary pressure only, and that water saturation is a
function of DNAPL-water capillary pressure only. This assumes that water is wetting with
respect to DNAPL, and that DNAPL is wetting with respect to air. Although validated
under certain specific flow conditions (e.g. Lenhard and Parker, 1988), the use of twophase capillary pressure relationships to represent three-phase systems has not yet been
experimentally validated for a wide set of flow conditions. Since most three-phase flow
models available at the present time adopt this approximation, they should be used in
three-phase flow situations with caution until further experimental validation is obtained.
As with the capillary pressure relations in Eqs.(4.8)-( 4.10), the relative permeability
relationships in Eq.(4.6) are path-dependent functions of fluid saturation that vary spatially
at real sites. Because relative permeability curves are difficult to measure in practice, they
are often predicted from measured capillary pressure data (e.g. Brooks and Corey, 1966).
This is justified by recognizing that the distribution of pore and throat sizes which largely
determine the shape of a capillary pressure curve also largely determine the nature of
the relative permeability curve. In three-phase systems it has been demonstrated that
water-relative permeability and air-relative permeability are dependent mostly on water
and air saturation, respectively (e.g. see Leverett and Lewis, 1941). As a result, various
methods have been proposed to predict three-phase relative permeabilities from twophase data (e.g. Stone, 1970; Lenhard and Parker, 1987). As with capillary pressures,
these extensions have been validated for selected flow conditions.
In cases where the transfer of components across phase boundaries is of interest,
such as the vaporization of a DNAPL in the unsaturated zone, or DNAPL dissolution below the water table, it must be decided whether or not the mass transfer terms appearing
in Eq.(4.l) will be treated on the basis of local equilibrium. The local equilibrium assumption assumes that the transfer of components between phases occurs instantaneously
relative to the bulk phase flow rates. This implies that equilibrium mass fractions of
components are uniformly distributed within all phases at all times. Commonly adopted
equilibrium relations include Henry's Law for the partitioning of components between
water and air, linear sorption isotherms for partitioning of components between water and
solid phases, and Raoult's Law for partitioning from a multicomponent DNAPL phase
to another phase.
For many situations, it is known that the local equilibrium assumption is not valid,
in which case a kinetic (i.e., rate-limited) approach must be adopted. These situations
often occur when describing a particular process at the field scale; examples include the
diffusion of components out of low permeability zones leading to tailing in soil vacuum
extraction systems, and the dependence of solubilization on flow rate in a solubilizing
surfactant flood.
Once a mathematical formulation has been arrived at for a given problem of interest,
a solution technique must be used to solve the governing equations. Because the equations governing multiphase/multicomponent flow and transport are typically non-linear,
and since material properties vary spatially, exact analytical solutions are available only
for simplified cases. The majority of situations require that a numerical solution technique such as finite difference, finite element, or finite volume be employed. Numerical
techniques require that the problem domain be discretized into a grid, with the governing
Sec. 4.2
135
equations approximated at the cells or nodes of the grid by means of algebraic equations
which link the cells or nodes together. The unknowns of interest, such as DNAPL saturation or contaminant concentration, are then solved for at these node locations or cell
centroids. Typical solution methods are described in textbooks such as Huyakorn and
Pinder (1983).
In many situations of interest, an implicit, iterative solution scheme will be required
to treat the non-linear nature of the governing equations. Typical non-linearities include
the dependence of relative permeability and capillary pressure on fluid saturation, and
the dependence of mass transfer rates on chemical concentration. In some systems the
non-linearity of the governing equations may be weak, allowing a simple iterative or
time-lagging scheme to be adopted. In other cases, such as the migration of a DNAPL,
the governing equations can be highly non-linear, in which case a Newton-Raphson
approach is usually adopted. Given a set of non-linear algebraic equations, F(X) = 0,
the Newton Raphson approach leads to a linearized matrix equation of the form
aF(X)n
--ax =
ax
_F(X)n
(4.11 )
where ilf~~)" is a square matrix of Jacobian derivative terms, F(Xt is a vector of residual
values, n the iteration level, and ax a vector of incremental unknowns given by:
ax =
Xn + 1 _ X"
(4.12)
where X is a vector of discrete unknowns. The algorithm in Eq.(4.11) is repeated until

the residual vector has converged to within a prescribed tolerance.
The set of algebraic equations arising through the application of a numerical technique must be solved for each iteration and/or time step. The simplest matrix solution
techniques involve the use of Gaussian elimination. For large sets of equations, direct
solvers utilizing Gaussian elimination will be prohibitively slow and prone to machine
round-off errors. Computer storage may also be a problem. Iterative solvers have been
shown to be more efficient at solving large sets of equations, and offer an additional
advantage that only the bands of non-zero elements of the coefficient matrix need to
be stored in memory. Letniowski (1989) gives an excellent review of preconditioned
iterative methods for solving sparse matrix equations.
Once a numerical solution technique has been applied to discretize the governing
partial differential equations, a computational algorithm is created by transposing the
numerical technique into a computer code through the use of a chosen computer language.
The resulting computer code is often referred to as "the model", but it should be clear
from the preceding discussion that this is merely one stage of the overall modeling
process.
In order to apply a given computer code to a site-specific situation, the user must
decide upon an appropriate scale of nodal discretization. For example, if the individual
migration pathways of DNAPL through heterogeneous sands is of interest, this may
require discretization at the centimeter scale (e.g. Kueper and Frind, 1991b). In a threedimensional model this may lead to the use of hundreds of thousands of nodal points,
leading to an extremely CPU- and memory-intensive effort. In recent years there has
136
Chap. 4
been much attention given to the development of effective, large-scale parameters for
multiphase/multicomponent flow and transport (e.g. Mantoglou and Gelhar, 1987; Kueper
and McWhorter, 1992). These parameters would allow much coarser computational grids
to be employed, but the resulting simulations would provide only the average behavior
of the system. It is important to realize, however, that effective, large-scale parameters
currently exist only for certain simplified situations. The unfortunate implication is that
reliable field-scale numerical simulation of certain DNAPL processes can at the present
time only be carried out using small-scale nodal discretization.
Figure 4.1 presents a numerical simulation of trichloroethylene (TCE) migration
through a heterogeneous sand aquifer where small-scale variability of hydraulic conductivity has been accounted for (Gerhard, 1993). A total of 32,000 nodal points has
been incorporated into the solution domain, with horizontal and vertical node spacings
of 25 em and 12.5 em, respectively. A spatially correlated, random hydraulic conductivity field was generated using a log-normal distribution as input (mean In(K) = -4.6,
variance In(K) = 2.0). An exponential autocorrelation function was assumed with horizontal and vertical correlation lengths of 5.0 m and 0.5 m, respectively. As can be seen,
the TCE spreads laterally across various lower permeability horizons, with a significant
degree of lateral spreading at the upper elevations. Simulations presented by Kueper et
al. (1990) have demonstrated that the use of an equivalent homogeneous domain incorporating a large-scale hydraulic conductivity as might be obtained from a pump-test, does
not yield the proper infiltration rates and degree of lateral spreading for point releases of
this nature.
DISTANCE
em)
Figure 4.1 Trichloroethylene (TeE) migration through a heterogeneous sand aquifer from a point
release. (Reproduced with permission from Gerhard (1993).)
Sec. 4.2
137
Once a computer code has been created, it must be verified to confirm that the
governing equations have been solved correctly and that the code is free of errors. Verification is carried out by running test problems and comparing results to those obtained
independently with an exact method such as an analytical solution. Since analytical solutions are only available for idealized situations with simple boundary conditions, it
is in general extremely difficult to completely verify a multiphase/multicomponent flow
and transport model. A less stringent method of verification is to compare the results of
a model with those of a previous, independently created numerical model. While there
is still the possibility that both models are making the same errors, the exercise should
nevertheless be carried out. Under no circumstances should a numerical model be used
if it has not been at least partially verified against an analytical solution, or compared to
the results of a previously published model.
Even if a computer code has been verified to ensure that the chosen numerical
method has been properly implemented, it remains to be seen whether or not the governing equations upon which the model is based have been correctly formulated. To ensure
this, the computer code should be validated through comparison to a controlled laboratory or field experiment. Validation is carried out by independently measuring all model
input parameters and comparing the model output to that of the controlled experiment.
This provides a complete check of not only whether the appropriate governing equations
have been employed, but also whether adequate constitutive relationships and material
property definition has been incorporated. From a strict scientific point of view, however,
agreement of the model results with those from a given experiment simply implies that
the model is capable of simulating that particular experiment. According to Konikow and
Bredehoeft (1992), a computer model cannot be shown to be valid, it can only be shown
to be invalid. This is a particularly important distinction where complex processes are
concerned such as those associated with the behavior of a DNAPL in the subsurface.
We recommend the use of the term "model testing" to avoid the misleading notion of
unconditional validity and even infallibility.
Calibration involves fitting the model output to a set of observed data while adjusting the input. Once the input parameters have been adjusted to produce a satisfactory
fit, these are deemed to be the proper input parameters so that the model can be used for
predictive purposes. A pump test may be an opportunity to calibrate a groundwater flow
model, for example, so that it can be used for future prediction of the performance of a
full-scale pump-and-treat system. For complex situations where not all controlling input
parameters can be obtained by independent measurement, model calibration may be the
only practical way to obtain certain parameters such as dispersivity or a large-scale mass
transfer coefficient. In theory, only one parameter should be fitted to a given set of data
in order to obtain a unique solution. In practice, if different parameters affect the solution in distinctly different ways, more than one parameter can be fitted. The possibility
of non-uniqueness, however, where more than one combination of parameters give an
equally good fit, increases with the number of parameters fitted. It may be determined
during the model calibration stage that additional field work is necessary to constrain
certain input parameters in order to reduce the uncertainty associated with the model
output.
138
Chap. 4
4.3 MODEL USAGE
For the typical model user, the mathematical solution, numerical solution, computational
algorithm, and verification steps of creating a model will have been completed by the
model author. The task therefore simplifies to formulation of the conceptual model,
the choice of proper scale and dimensionality, and the obvious task of selecting an
appropriate computer code. Model testing will still be necessary to demonstrate that an
appropriate model has been selected, and model calibration will in most cases be needed
to determine uncertain parameters. It should be emphasized that the task of selecting an
appropriate computer code can only be carried out successfully if the model user fully
understands the mathematical formulation upon which the model is based. There are
numerous assumptions and simplifications that can be implemented at the mathematical
formulation stage, and one must carefully examine the final set of partial differential
equations which are solved to determine exactly what physics the model represents.
Once an appropriate model has been selected, tested, and calibrated, the user is
in the position to carry out a variety of tasks. For complex systems, the best way to
understand the interaction and relevance of various processes is to carry out a sensitivity
analysis. This involves varying key input parameters one at a time within a physically
realistic range and observing the response of the model. The parameter ranges are usually
known to the experienced hydrogeologist and in many cases may be narrowly defined.
A sensitivity analysis is an excellent means of educating those involved with a particular
site or class of contaminants, and is often the only "experimental" method available to
gain insight into specific processes.
Once the calibration of a numerical model has been completed, in many cases the
model can be used for design purposes. Examples may include selecting the number
and location of pumping wells in a pump-and-treat remediation scenario, determining
the hourly mass removal rate from a soil vacuum extraction system so that an above
ground treatment system can be designed, or selecting the spacing and number of wells
to use in a surfactant or alcohol flood designed to remove DNAPL. Whenever used for
design purposes, however, it must be kept in mind that all model output is subject to
uncertainty. A proper design exercise will involve producing a probable range of system
behavior, with the final design having enough flexibility to accommodate the full range
of possible system behavior.
In certain circumstances a numerical model can be run to examine a variety of
"what-if" situations. An example may be the loss of capture in a pump-and-treat system
upon failure of one well, or the rate at which a vapor plume will resume spreading upon
accidental shut-down of a soil vacuum extraction system. A model can also be utilized to
study the uncertainty associated with a given process. This is often done with a Monte
Carlo analysis, which is simply a sensitivity analysis in which parameter values are
selected at random from their respective ranges. The suite of model outputs can then be
analyzed and described in statistical terms to yield the most probable behavior of the
system, along with the uncertainty associated with that behavior.
Whenever a model is used for purposes of site specific prediction, it must be kept
in mind that multiphase/multicomponent flow and transport models are highly sensitive
Sec. 4.4
139
Example
to site-specific factors such as heterogeneities, the scale of discretization, and mass transfer parameters. In addition, it must be kept in mind that some controlling processes are
still imperfectly understood. The possibilities for full validation at real sites are therefore
limited. The prudent approach to numerical modeling at DNAPL sites is to utilize a
numerical model for education purposes, and to produce a range of possible system behavior rather than one individual prediction that is claimed to be a perfect representation
of site conditions.
4.4 EXAMPLE
To demonstrate the use of a DNAPL migration model, the two-phase flow model developed by Kueper and Frind (l991a,b) was utilized to examine the possible rates of
DNAPL leakage from a waste lagoon located at Ville Mercier, Quebec. These lagoons
were operational from 1968 through to 1972 and received a variety of oils, sludges and
other debris (Poulin, 1977). Because exact records were not kept, it is unknown at present
exactly how much DNAPL could have leaked out of the base of the lagoons, and at what
rates. Figure 4.2 presents a vertical cross section through the site illustrating the subsurface conditions. As can be seen, the unlined lagoon was situated in a sand and gravel unit
(K = 5 x 10- 3 cm/s) underlain by a discontinuous upper silty sand till (K = 1 x 10- 3
cm/s). This till is in turn underlain by a discontinuous basal till (K = I x 10- 4 cm/s)
situated above fractured bedrock. For the purposes of modeling, the lagoon was attributed
a total area of 8,400 nr'. Examination of aerial photographs taken during the operating
period of the lagoon suggests that this area is a minimum.
The specific objective of the modeling was to determine what possible rates of
DNAPL leakage could have occurred through the base of the lagoon during the first
two years of operation. The conceptual model assumed that DNAPL acted as a nonwetting liquid in the presence of water, and that DNAPL dissolution did not significantly
influence leakage rates over the four year period of operation. It was also noted that the
base of the lagoon was below the water table during the entire operating period. The
migration process of interest was therefore DNAPL displacing water in the absence of
mass transfer by dissolution. On the basis of site history, it was further conceptualized
that the level of DNAPL in the lagoon was maintained at a relatively constant level over
the four year period, but that a sludge layer built up with time. It was also decided that
spatial variability of geological properties within each of the major lithologic units would
not be influential in governing the overall rate of leakage since the size of the lagoon
was large relative to the scale of heterogeneity.
The mathematical model required to simulate DNAPL displacing water in the absence of mass transfer is based on the following partial differential equations, which can
be created by summing Eq.(4.1) over all components comprising a given phase:
~ (k;,)kr,a [apa _ Pag~])
ax;
/ha
ax)
ax)
aSa, ex =
at
W,
(4.13)
where the subscripts wand n denote water and DNAPL, respectively. The pair of
equations represented by Eq.(4.l3) are coupled through the capillary pressure relation
...
080MS
\ 01\3176\
<O<f)
SOUTH
....,
51f
....,
0..
~~~ ~
-0
I~
00
lJ..Ct::tn
f-u.-
0::(1)
'III
l.L.
~N
0..' I
......
II
f--LA:O~--l
(1m>
'j'<j'
U1a:::
,,~
_0
o~
(")
~I'r---
::g
000
l.L.Cl::(/l
IJ..
I I I
'11
u.-u.-
45
40
35
....,
NORTH
oo
o
o
I
u,
tL~1
II 10
30
..,'"~
25
-;
20
"
~
~
15
<;
ss:
10
0
-5
-10
-15
-20
100
200
400
300
MERCIER REMEDIATION PANEL
~
F-OOI
F-l10
R-07,S-09
PW-113
BOREHOLES BY CNFS (1990, ref.: 91-2)

BOREHOLES BY CNFS (1991 AND 1992, ref.: 92-4)
BOREHOLES BY UNIVERSITt LAVAL (1989, ret.: 91-5)
TREATMENT PLANT PUMPING WELL (1984,)
CD
f1LL
SiLT AND SILTY CLAY
SAND AND GRAVEL
600m
500
@
@
FIGURE 2-3
UPPER
TILL
BASAL
TILL
BEDROCK
TYPICAL NORTH-SOUTH
GEOLOGICAL
CROSS-SECTION (L 150)
Figure 4.2 Vertical cross-section illustrating major geological units beneath waste lagoons at Ville
Mercier, Quebec. (Reproduced with permission from Mercier (1993).)
Sec. 4.4
141
Example
in Eq.(4.10) and are subject to the constraint that the two phase saturations sum to
unity.
The finite difference model developed by Kueper and Frind (1991a) solves Eq.(4.13)
fully implicitly using full Newton Raphson iteration. A sparse matrix solver based on
the ORTHOMIN technique (Vinsorne, 1976) is employed to minimize CPU time and
computer storage requirements. The computational algorithm was created using the FORTRAN language, with all simulations carried out on an IBM 6000 Rise based workstation. As described in Kueper and Frind (1991a), the model was verified in one dimension
against an exact analytical solution by McWhorter and Sunada (1989), and was tested
through comparison with a controlled laboratory experiment (Kueper et al., 1989) involving the migration of tetrachloroethylene through a heterogeneous sand pack.
For the modeling task under consideration here, all model input parameters were
estimated from various previous studies completed at the site. These studies, as well as
further details of the numerical modeling, are found in Mercier (1993). Although large
amounts of data are available for many hydrological parameters, certain model input
parameters such as DNAPL density, interfacial tension, and viscosity were not readily
available. To account for this uncertainty, these properties were varied within expected
ranges. In addition, the properties of the sludge layer across the lagoon bottom were not
precisely defined, demanding that these also be varied over an expected range.
It was decided to adopt a two-dimensional vertical cross-section grid with vertical
and horizontal nodal spacings of 1.0 m and 2.0 m, respectively. This resulted in adequate
resolution of the geological features in Figure 4.2, and resulted in a total of 8,000 nodes
in a 40 m high by 400 m wide solution domain. To account for the uncertainty in various
input parameters, a total of 12 simulations were performed. Figure 4.3 presents both
~100000
............
Q)
75000
:::J
-0
>
UPPER BOUND
LOWER BOUND
50000
.....J
0....
<{
25000
--
200
--- --400
600
800
TIME (days)
Figure 4.3 Maximum and minimum DNAPL leakage rates through base of lagoon estimated using numerical simulation. (Reproduced with permission from Mercier (1993).)
142
Chap. 4
the maximum and minimum rates of leakage from this suite of simulations. The bottom
curve in this figure represents a DNAPL having a density of 1100 kg/m", a viscosity of
0.050 Pa-sec, an interfacial tension of 0.045 N/m, and a 0.50 m accumulated DNAPL
pool height. The upper curve represents a DNAPL having a density of 1300 kg/m",
a viscosity of 0.002 Pa-sec, an interfacial tension of 0.015 N/m, and an accumulated
DNAPL pool height of 0.50 m.
The overall conclusion of this modeling assignment was that the geological conditions beneath the lagoons at Mercier were such that significant volumes of DNAPL
likely escaped the lagoon during its operating period. Additional simulations (not shown),
however, have demonstrated that a silting of the lagoon bottom could have completely
arrested DNAPL leakage after an initial period. Even with such silting, the initial infiltration rates were found to be significant. Further details of the modeling are presented
in Mercier (1993).
4.5 CONCLUSIONS
Numerical modeling offers an efficient and cost-effective means of gaining insight into
the various processes which govern multiphase/multicomponent flow and transport in a
variety of geological environments. Used properly, modeling can be a useful component
of the design and decision making process at most sites. In regulatory jurisdictions
where it is prohibited by law to introduce contaminants into the subsurface for research
purposes, numerical modeling may be the only experimental technique available to study
and assess the behavior and clean-up of DNAPLs at the field scale.
The modeling process begins by clearly stating the modeling objectives. Following
this, an appropriate conceptual model must be formed for the problem at hand. The conceptual model will include identification of the relevant physical and chemical processes,
possible decoupling of time scales, a clear definition of the spatial scales and dimensionality required, and definition of relevant material and fluid properties. The conceptual
model then leads to formulation of a mathematical model, which includes the governing
partial differential equations, boundary conditions, and necessary constitutive relationships. Through the use of a numerical solution technique such as finite differences or
finite elements, the governing equations are solved and a computational algorithm is created. This computer code must in turn be verified to ensure that the numerical solution
technique has been properly implemented and tested to ensure that the governing partial
differential equations have been selected properly.
Once a numerical model has been chosen, the model user will likely calibrate the
model to a known set of data from a site in order that certain input parameters can be
constrained. The model can then be used to perform sensitivity and uncertainty analyses,
to assist in the design process, and to examine a variety of "what if" scenarios. While
much can be gained from the proper use of a numerical model, it must be remembered
that failing to pay attention to critical issues (e.g. the scale of discretization and the proper
choice of input parameters) can result in a completely misleading interpretation of site
conditions. This is particularly true for multiphase/multicomponent flow and: transport
modeling where the physical and chemical processes involved are an order of magni-
Sec. 4.6
References
143
tude more complex than those associated with traditional groundwater flow and solute
transport modeling.
ACKNOWLEDGEMENTS
4.6 REFERENCES
Allen, M. B. (1985) "Numerical modelling of multiphase flow in porous media", Advances in Water
Resources, 8, 162-187.
Bear, J. (1972) Dynamics of Fluids in Porous Media, Elsevier, New York.
Brooks, R. H. and A. T. Corey (1966) "Properties of porous media affecting fluid flow", Journal
Irrigation Drainage Division, American Society of Civil Engineers, 92 (IR2), 61-88.
Gelhar, L.W. (1986) "Stochastic subsurface hydrology form theory to applications", Water Resources Research, 22, 135S-145S.
Gerhard, J. (1993) B.A.Sc. Thesis, Queen's University, Kingston, Ontario, Canada.
Huyakorn, P. S. and G. F. Pinder (1983) Computational Methods in Subsurface Flow, Academic
Press.
Konikow, L. F. and Bredehoeft, 1. D. (1992) "Groundwater models cannot be validated", Advances
in Water Resources, 15, 75-83.
Kueper, B. H., W. Abbot, and G. Farquhar (1989) "Experimental observations of multiphase flow
in heterogeneous porous media", Journal of Contaminant Hydrology,S, 83-95.
Kueper, B. H. and E. O. Frind (199Ia) "Two-phase flow in heterogeneous porous media, I. Model
development", Water Resources Research, 27, 1049- 1057.
Kueper, B. H. and E. o. Frind (l99Ib) "Two-phase flow in heterogeneous porous media, 2. Model
application", Water Resources Research, 27, 1059- 1070.
Kueper, B. H., E. O. Frind, and D. B. McWhorter (1990) "Application of a numerical model and
laboratory parameter measurement to the movement of dense, immiscible phase liquids in a
heterogeneous sand aquifer", In: Proceedings: IAH Conference on Subsurface Contamination
by Immiscible Liquids, Calgary, Alberta, April 18-20.
Kueper, B. H. and D. B. McWhorter (1992) "The use of macroscopic percolation theory to construct
large-scale capillary pressure curves", Water Resources Research, 28, 2425-2436.
Lenhard, R. J. and 1. C. Parker (1987) "A model for hysteretic constitutive relations governing
multiphase flow. 2: Permeability-saturation relations", Water Resources Research, 23, 21972206.
Lenhard, R. J. and J. C. Parker (1988) "Experimental validation of the theory of extending twophase saturation-pressure relations to three-phase systems for monotonic drainage paths", Water
144
Chap. 4
Letniowski, F. W. (1989) "An overview of preconditioned iterative methods for sparse matrix equations", Research report CS-89-26, Faculty of Mathematics, University of Waterloo, Waterloo,
Ontario.
Leverett, M. C., (1938) "Flow of oil-water mixtures through unconsolidated sands", Trans. A./.M.E.,
132, 149-171.
Leverett, M. C. (1941) "Capillary behaviour in porous solids", Trans. Soc. Pet. Eng. A./.M.E., 142,
152-169.
Leverett, M. C. and W. B. Lewis (J 941) "Steady flow of gas-oil-water mixtures through unconsolidated sands", Trans. A.l.M.E., 142, 107.
Mantoglou, A. and L. W. Gelhar (1987) "Stochastic modelling of large-scale transient unsaturated
flow systems", Water Resources Research, 23, 37-46.
McWhorter, D. B. and D. K. Sunada (1990) "Exact integral solution for two-phase flow", Water
Mercier Remediation Panel (1993) "Evaluation of remedial measures for the subsurface contamination associated with the former mercier lagoons at Ville Mercier, Quebec", Final report
submitted to Laidlaw Environmental Services (Mercier) Ltd.
Parker, J.e., R. J. Lenhard and T. Kuppusamy (1987) "A parametric model for constitutive properties governing multiphase flow in porous media", Water Resources Research, 23, 618-624.
Poulin, M. (1977) "Groundwater contamination near a liquid waste lagoon, Ville Mercier, Quebec",
M.Sc. Thesis, University of Waterloo, Waterloo, Ontario, Canada.
Poulsen, M. and B. H. Kueper (1992) "A field experiment to study the behaviour of tetrachloroethylene in unsaturated porous media", Environmental Science and Technology, 26, 889-895.
Schiegg, H. O. (1984) "Considerations on water, oil and air in porous media", Wat. Sci. Tech., 17,
467-476.
Stone, H. L. (1970) "Probability model for estimating three-phase relative permeability", J. Pet.
Tech., 214-218.
van Genuchten, M. Th. (1980) "A closed form equation for predicting the hydraulic conductivity
of unsaturated soils", Soil Sci. Am. J., 44, 892-898.
Vinsome, P. K. W. (1976) "ORTHOMIN, an iterative method for solving sparse sets of simultaneous linear equations", SPE Symposium on Numerical Simulation of Reservoir Performance,
Society of Petroleum Engineers, Los Angeles, Feb. 19-20.
Wyckoff, R. D. and H. G. Botset (1936) Physics, 7, 325.
5
Experimental Studies of the
Movement of Chlorinated
Solvent Compounds and other
DNAPLs in the Vadose,
Capillary, and Groundwater
Zones
Richard L. Johnson1 and Bernard H. Kueper '
1 Department

Beaverton, OR 97006
2 Department
of Civil Engineering
Queen's University
Kingston, Ontario, Canada K7L 3N6
Page
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5.9
5.10
5.11
5.12
Introduction
5.4.1 General
DNAPL Retention Capacity of Porous Media
5.8.1 General
Chlorinated Solvent Vapor Transport in the Vadose Zone
Chlorinated Solvent Dissolution in the Groundwater Zone
Conclusions
References
146
147
150
154
154
155
158
159
160
166
166
166
169
170
173
175
175
145
146
Experimental Studies of DNAPL Movement
Chap. 5
ABSTRACT
The behavior of dense non-aqueous phase liquids (DNAPLs) in naturally heterogeneous

porous media is extremely complex and difficult to predict. Physical model experiments
have therefore played and continue to play an important role in understanding that complexity. They provide a means to incorporate all of the physics and chemistry which
are important for DNAPL behavior while also providing a great deal of experimental
control. Perhaps most importantly, physical model experiments can be conducted and
insights can be gained without an a priori understanding of all of the relevant processes.
Physical model experiments are therefore excellent for developing conceptual models of
how DNAPLs behave in the subsurface.
Physical model experiments can be used to examine all aspects of DNAPL behavior
in the subsurface, including DNAPL flow, dissolution, vaporization, and remediation. As
with numerical models, an important strength of physical models is that experimental
conditions can be reproduced. For example, the effects of different release conditions can
be studied in the same physical setting. In the same way, different remediation strategies
can be compared directly using a series of similar releases. The experiments can be
conducted using a variety of media and DNAPLs, and can be conducted under both
saturated and unsaturated conditions.
Perhaps the most important long term benefit of physical model experiments is in
the validation of numerical models. The experimental data reported here are of use in
evaluating the performance of the models which provide the basis for decisions regarding
the design and implementation of remediation systems.
5.1 INTRODUCTION
The pioneering laboratory experiments of Schwille (l984a, 1984b, 1988a,b) and others on the movement of dense chlorinated solvents in porous media has resulted in the
conceptualizations of dense chlorinated solvent behavior presented in Chapter 2. As discussed in Chapter 4, numerical modeling offers an alternative means of predicting the
behavior of chlorinated solvents and other dense non-aqueous phase liquids (DNAPLs) in
complex systems. However, although some modeling successes have been achieved, such
efforts are still limited by difficulties in characterizing the systems of interest in sufficient
detail that meaningful predictions can be obtained. Conducting field experiments with
DNAPLs in the subsurface may be a preferable means of investigation in some cases, but
the associated drawbacks usually include high costs, difficulties in reproducing experiments, and regulatory restrictions. Thus, physical model experiments remain important
for the development of an improved understanding of chlorinated solvent behavior in
both porous and fractured media.
The fundamentals of multi-phase flow and transport as well as the factors which
control chlorinated solvent movement in the subsurface are discussed in Chapters 2, 3,
and II. In this chapter, the microscopic and macroscopic behavior of chlorinated solvents
are discussed based on the results of various laboratory and field experiments. To date, the
Sec. 5.2
147
majority of experimental work has concerned flow through homogeneous porous media.
Since most chlorinated solvent spills occur in media where heterogeneity controls the
pathways of infiltration, the current trend is towards experiments on DNAPL behavior
in more complex, heterogeneous media.
For the purpose of this discussion, it is convenient to divide the subsurface into
three general domains: 1) the vadose or unsaturated zone; 2) the capillary zone; and
3) the saturated or groundwater zone. The vadose zone is defined here as that portion
of the soil Which is near or at residual water saturation. DNAPL migration in the vadose
zone concerns primarily the displacement of air by DNAPL. The capillary zone spans
the region over which the water content rises from the residual value to full saturation
of the pore space by water. It is in this region that three-phase flow can take place, with
DNAPL acting both as a wetting fluid displacing air and as a non-wetting fluid displacing
water. The groundwater zone is, of course, below the free water table, and most DNAPLs
there typically act as non-wetting fluids with respect to water, though exceptions to this
generalization do occur in practice.
As the above discussion indicates, the primary change that occurs in a porous
medium when moving downwards from ground surface is an increase in the water content. Moving from the vadose zone to the saturated zone represents a change from a
system in which two immiscible fluids (air and water) initially share the pore space, to a
system where water occupies the entire pore space. When a liquid chlorinated solvent is
introduced, the relationships between the fluids and the porous medium become considerably more complex. Thus, at a spill site, a portion of the unsaturated zone can contain
up to three fluids; below the water table, two fluids may be present.
5.2 DNAPL MOVEMENT IN THE VADOSE ZONE

The laboratory experiments conducted by Schwille (1988a) in homogeneous sand under conditions of a point source demonstrated that, within the vadose zone, chlorinated
solvent DNAPLs are well behaved. Tetrachloroethylene (PCE) was observed to move
downwards as a uniform front and drain to an apparently homogeneous residual distribution. In these experiments, liquid solvents were found to penetrate relatively quickly
due to their low viscosity and high density; there was little tendency for the solvent body
to spread laterally as it sank. Figure 5.1 shows the behavior observed in a typical small
spill of PCE in a column of fine-grained silica sand (K = 1.2 x 10- 4 m/s). Note that
the infiltration body remained very thin as it penetrated through the unsaturated portion
of the column (the zone above the open inverted triangle is at residual water saturation
and therefore represents the vadose zone).
Schwille (1988a) and others have observed that within the vadose zone, chlorinated solvents behave in ways that are similar to petroleum-derived organic liquids. The
behavior of the latter group has been studied in considerable detail (e.g., see: Van Dam,
1967; Dracos, 1978; Van der Waarden et al., 1971 and 1977; Dietz, 1967; and Eckberg
and Sunada, 1984). The experiments of Eckberg and Sunada (1984), in which oil was
spilled into a column of glass beads initially under drainage conditions, demonstrate the
Chap. 5
148
18
DENSE CHLORINATED SOLVENTS

v 10L
c
--_"":""_--,
I C2 CII.
h s=70cm
Figure 5.1 Infiltration of 10 L of tetrachloroethylene (PCE) into a fine-grained sand

(K = I to 2 x 10- 4 m/s). (Reprinted with permission from Schwille (1988a).)
AIR
WATER
AIR
WATER
20
40
60
80
100
20
40
FRACTION OF PORE SPACE
60
80
Figure 5.2 Distribution of water and oil before and after oil is spilled into a column of glass beads
under drainage conditions. (Adapted with permission from Eckberg and Sunada (1984).)
100
Sec. 5.2
149
type of three-phase interactions which can be expected between air, water, and solvent.
Figure 5.2 depicts the distribution of fluid phases before and after the addition of the oil.
Both air and water were displaced by the infiltrating liquid. Air was displaced by the
more-viscous, more-wetting NAPL. Water was mobilized in part because the capillary
pressure across the water/oil interface is less than that across the original air/water
interface (oil is wetting with respect to air on a water surface, and therefore preferentially
situates itself between the water and air phases). Some water is also mobilized because
of the downward pressure exerted by the oil body on the water phase.
In homogeneous media, infiltration through the unsaturated zone can occur in the
form of "fingers" caused an inherent instability in the advancing front. As discussed in
Kueper and Frind (1989), this instability can arise from an unfavorable density and/or
viscosity ratio between the displacing and displaced fluids. Schwilles (1988a) experiments with "sheet-like" spills of solvent show preferential penetration along such fingers
(Figure 5.3). It is important to note that a fingered front will, in general, not arise from
a point source release of solvent when the size of the release point is less than the
critical wavelengths required to initiate fingering. In addition to the density and viscosity ratios, the velocity of the displacement and permeability of the medium influence
the spacing of the fingers formed. Similar results have been obtained for water infiltrating into homogeneous, unsaturated sand. For example, Glass et al. (1989) showed
that the extent to which the wetting front was unstable could be predicted based on the
properties of the medium and the fluid. In general, as the permeability of the medium
decreases, the importance of preferential or "fingered" flow through the unsaturated zone
will increase.
Unsaturated zone experiments have demonstrated that when the medium is heterogeneous, DNAPL flow is controlled by those heterogeneities. Johnson (1990) conducted
an experiment in a large cylindrical column which contained a single layer of lower
permeability sand (K = 1.7 x 10- 4 m/s) within a medium sand (K = 3.8 x 10- 4 m/s,
Figure 5.4). The fine lens was characterized by a water content higher than the host
sand. A release at the surface of 100 mL of dyed peE moved downwards through the
medium sand as a relatively uniform front. When it reached the finer-grained layer,
it pooled and spread over nearly the entire cross-section of the column. The solvent
then fingered into the finer layer at a number of points; these fingers terminated within
the first few centimeters due to the limited volume of the spill. Although the permeabilities of the two sands differed by only a factor of two, that difference controlled the initial water distribution. The solvent spread laterally along the fine layer
because that layer did not contain enough air to allow the solvent to continue to imbibe and displace air. Since it was easier for the solvent to imbibe laterally and displace air rather than displace water vertically as a non-wetting fluid, lateral flow took
place. Had the fine-grained layer been primarily saturated with air, spontaneous imbibition of solvent into the layer would have taken place. Since permeability contrasts between layers of an order of magnitude or more are common at field sites, it
is likely that solvent flow in the unsaturated zone will be strongly affected by such
heterogeneities.
150
Chap. 5
Figure 5.3 Finger formation during a "sheet-like" infiltration of tetrachloroethylene

(peE) into unsaturated sand. (Reprinted with permission from Schwille (l988a).)
5.3 DNAPL MOVEMENT IN THE CAPILLARY ZONE

Chlorinated solvent behavior within the capillary zone is greatly complicated by the fact
that up to three fluid phases can be actively flowing at one time. Also, as discussed
above, the relative permeability to liquid chlorinated solvents is lower in regions of
higher water content. We note then that even within relatively homogeneous media,
Sec. 5.3
151
70
-4
K=3.8xlO
-------
~40
!2
~30
K=3.8xlO
-------
-4
.......---55cm
0.1
0.2
0.3
0.4
VOLUMETRIC
WATER
CONTENT
Figure 5.4 Horizontal excavations of a cylindrical physical model showing the infiltration of
trichloroethylene (TeE) into a medium sand (K = 3.8 x 10-4 m/s) containing a single lowerpermeability layer (K = 1.7 x 10-4 m/s).
spatial vanations in water content in the capillary zone can be large. Thus, in many
cases, water-content-induced permeability contrasts in the capillary zone are sufficient
to control the infiltration pathways of a liquid solvent.
Anderson (1988) conducted 3-D model experiments to examine solvent movement
through the capillary zone in fine sand. The dimensions of the model were 50 em long by
25 em wide by 30 em high. Under drainage conditions, spills of 75 mL of PCE caused
100 to 500 mL of water to be displaced out of the tank. This was in part the result
of a collapse of the capillary fringe brought about by the lower capillary pressures
that existed across the solvent/water interfaces as compared to the original air/water
interfaces. (Eckberg and Sunada (1984) observed an identical effect.) The release of water
was also affected by the fact that in regions of high water content, PCE must displace
water in order to undergo gravity-driven migration. Excavation of the tank following
the spill revealed that when the DNAPL reached the capillary zone, the infiltration front
152
Chap. 5
became unstable and fingers formed (Figure 5.5). The fingers had many geometries, some
being vertically very thin and laterally extensive. Others penetrated downward into the
saturated zone and were in places only millimeters in diameter. This irregular behavior
existed despite the fact that the tank had been carefully packed with very homogeneous
sand.
In another set of experiments, Anderson (1988) released peE into a 60 ern high
by 30 em wide by 2 em thick tank (Figure 5.6). The tank had again been packed with
a fine sand (K = 2 x 10- 4 cm/s) and existed under drainage conditions prior to the
l1li
(b) +17.4 em:.:::'
(a) +22.4 em.'.'.
iI~;;;::;: : .:. :.:-: :-: .: :.: .:.: . : . . : :.:.: .::.: : : .

.................................................................: .......................: ..:
.
..............................
.
........................................................
........................................................
" ':.',.. ': ..... ',.:.', ': .:: .. ': ..', .. ', ..... : .:: ..', ':.'::.', .. ',':. ':.',', .... ',.:.....
.......................................................
..................................
. . .. .. ....... .. ...... ... . . ....... .. . .. .... .. . ........ ...
.....................
(e)':~.I?:~.:~~.::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::/:::::
. (.~l.:~.~:.~:.~.~:~::~;;::::~:::::::::::::::::::::::::::::::':-:::':'-::.':::
:::.}::::::::::::::+:::::-:::::::':?:.:::)).:));}}:))) ))))1..(0%:;:)))));(:::::::::::::::-::
:-:.:::.:::1
..............................................................................
. ... .. . . .. .. ..........
...............
...
..
..
..
.........................................
.
...
.............................................
...................................
.
(e) +7.9 em .:.:.:.:.:.:::.:-:.:-:.:.:-:.:.:-:-:.:.::-:.:-:.:.:.:.:.:.:-:::.:-:.:.:.:. (t) +2.9 em -:-::..:-:-:.:-:-:.:.:.:-:.:::-:.:.:.:.::.::-:-:.:-:.:.:-:.:-:.:-:.:-:
..............................................................................................................
.......................................................
.......................
.
..
..............................................
.............
........................................
... .::::
::::::
::::
:::
::::
:.:
::::::
: .:::::::::::::::::::::::::::::
illlllll1
Figure 5.5
Distribution of peE at various elevations above the water table during
the excavation of a 75 mL release into a three dimensional sand tank. (Reprinted with
permission from Anderson (1988).)
Sec. 5.3
153
0L...-_....L-_--'-_--JL.......a_-+o 0.1 0.2 0.3
.....;::",
.....::::~~
SOIL MOISTURE CONTENT

(vol/vol)
Figure 5.6 a. Soil moisture profile prior to a tetrachloroethylene (PCE) release; b. Infiltration of PCE into a fine-grained sand showing the depression of the capillary zone
caused by the solvent. (Reprinted with permission from Anderson (1988).)
spill. The spill was made across the entire thickness so that the DNAPL movement could
be monitored through the transparent walls of the tank. As the solvent moved into the
capillary zone, it again appeared to cause a collapse of the capillary zone in the region
ahead of the main DNAPL body. This resulted in a "funneling" effect that focused the
DNAPL body as it approached the water table. This promoted the build-up of DNAPL
head such that, in this case, the entry pressure of the underlying water-saturated sands
could be overcome.
As a DNAPL moves into the tension-saturated portion of the capillary zone, it
must displace water if continued downward migration is to occur. Since it is generally a
non-wetting fluid with respect to water, there must be a sufficient build-up of capillary
pressure to allow the solvent to overcome the entry pressure of the nearly water-saturated
medium. In the case of imbibition conditions for which there is a sharp increase in water
content at the top of the capillary zone, some simultaneous lateral spreading of DNAPL
154
Chap. 5
may take place due to imbibition of air-filled pores. If sufficient solvent reaches the top
of the capillary zone, the zone can be breached and the solvent can continue downward
into the saturated zone. Within the capillary zone as well as in the underlying saturated
zone, the infiltration of DNAPL is accompanied by the displacement of water. This
process is unstable due to the fact that the displacement is by a less-viscous, moredense fluid. If the width of the solvent front exceeds the critical wavelength for finger
initiation, continued solvent infiltration will proceed in the form of fingers. (In cases
where the solvent viscosity is greater than that of water, a velocity threshold will exist
beyond which the displacement is unstable.)
Fingers form in homogeneous porous media within pore-scale zones of higher
permeability. Once fingers have been established, the DNAPL permeability within the
fingers is greatly increased such that the fingers act as conduits for subsequent DNAPL
movement. In heterogeneous porous media, solvent migration through the capillary zone
will proceed through the higher permeability regions, again giving the appearance of
fingers. In this case, however, these "channels" of solvent are simply due to the fact that
higher permeability migration pathways exist, and are independent of the fact that an
unstable displacement process may be taking place (Kueper and Frind, 1988).
5.4 DNAPL MOVEMENT IN THE GROUNDWATER ZONE

5.4.1 General
Interest in the behavior of solvents in the subsurface usually stems from concern over
the potential impact of the solvents on groundwater quality. In that context, the behavior
of solvents within the saturated zone is often of greatest interest. There are many case
histories in which spills of chlorinated solvents and other DNAPLs have penetrated all
three subsurface zones and resulted in wide-spread contamination (e.g., see Kueper et al.
1992). In many cases, however, the evidence of contamination is limited to the detection
of dissolved compounds. Direct observation of "free-product" solvents in the subsurface
is the exception rather than the rule.
The behavior of chlorinated solvents as they penetrate the saturated zone has been
followed visually in the large tank experiments of Schwille (1988a), and also in smaller
tank experiments at the Oregon Graduate Institute and at the University of Waterloo.
Two important conclusions have been drawn from these saturated-zone observations:
1) as in other zones, subtle permeability contrasts can have a controlling influence on
the movement of a DNAPL; and 2) typical groundwater flows do not have much effect
on the flow behavior of a DNAPL. The latter point is clearly demonstrated in Schwilles
(l988a) large tank experiments (see Figure 5.7) where under conditions of a relatively
high groundwater velocity (v > I m/day), there was essentially no distortion of the
penetrating solvent body.
Once having crossed the water table, the downward migration of a chlorinated
solvent may continue until the bottom of the aquifer is reached. In general, a barrier to
groundwater flow will also be a barrier to solvent migration. DNAPL solvent that reaches
an aquitard will tend to form a flat pool, and/or collect in local depressions on the top
Sec. 5.4
155
I j I I rr-+......
Mlm/sl,......
..,
2' h
~-~--------------
...
_ _ _ _ _ _ _ _ _ .liZ.. _ _
1=0.005
Figure 5.7 Infiltration pattern of an areal spill of tetrachloroethylene (PCE) into a

medium sand. (Reprinted with permission from Schwille (l988a).)
of the aquitard. In the case of a chlorinated solvent having even a moderate interfacial
tension with water, lateral flow and pooling will also occur above regions (e.g., lenses)
of only slightly lower permeability (Kueper and Frind, 199Ia,b).

As discussed above and also in Chapter 3, a DNAPL which is migrating below the water
table will spread laterally whenever a horizon of lower permeability is encountered. A
DNAPL will accumulate on top of that horizon until the entry pressure of the lower
permeability material is exceeded. As demonstrated by Kueper and Frind (1991 b), lateral
spreading can continue even after a lower permeability lens has been penetrated. This
is due to the fact that a higher relative permeability to the DNAPL exists above the
lens as compared to within. This follows from the fact that a saturation discontinuity
exists whenever a permeability contrast exists, leading to higher DNAPL saturations
in the more permeable material. The net result is that solvents will also be spreading
sideways if they are migrating vertically through horizontally-stratified porous media.
The numerical simulations presented in Chapter 4 suggest that a DNAPL solvent can
spread laterally several meters below a point source before even having migrated one or
two meters vertically.
Kueper et al. (1989) studied the influence of porous media heterogeneity on DNAPL
migration in an initially water-saturated, sand-packed cell. The experimental device and
the arrangement of the four types of sand used to pack the cell are illustrated in Figure 5.8.
The source area for the injection of PCE was located along the top boundary; lateral
outflow of water occurred at the side boundaries. The sand pack had a depth in the
third dimension of 0.6 em, which is significantly less than the wavelengths for fingering
that would be expected for this system. Drainage PCE-water capillary pressure curves
measured for each of the four sands are presented in Figure 5.9. As expected, the curves
for the lower permeability sands are shifted vertically upwards. A Brooks-Corey (Brooks
and Corey, 1966) capillary pressure function was fitted to each of the four measured
curves. The best fit Brooks-Corey parameters and the results of falling-head permeameter
testing are presented in Table 5.1.
156
Chap. 5
SOURCE
--
#70SILICA
#50SILICA
IIII #25 OTTAWA
LENGTH (ern)
10
#16 SILICA
Figure 5.8 Schematic drawing of the physical model used by Kueper et al. (1989) for
tetrachloroethylene (PCE) spills into saturated heterogeneous porous media. (Adapted
with permission from Kueper et al. (1989).)
Throughout the duration of the Kueper et al. (1989) experiment, the source area was
subjected to a constant head of 4 ern of PCE. Figure 5.10.a illustrates the fluid distribution
in the cell 126 seconds after initial PCE penetration. The PCE is seen to be pooling and
cascading off of the finer sand (#50 silica) just below the source area. The inability to
penetrate this lens is due to the fact that the capillary pressure immediately above the lens
is less than the 13.5 cm displacement pressure for the sand in the lens. Figure 5.10.b
illustrates the fluid distribution in the cell after 220 seconds. The PCE is continuing to
cascade off the finer sands (#50 and #70 silica) it has encountered, and is seen to have
just reached the #25 Ottawa sand lens. Figure 5.1O.c corresponds to the point in time at
which the PCE first penetrated the #25 sand lens. However, because the displacement
pressure of the #25 sand is higher than that of the # 16 sand, lateral flow occurred above
the #25 sand lens prior to penetration. This lateral flow and associated pooling allowed
the capillary pressure above the lens to build up and exceed the displacement pressure
of the lens. The build-up of capillary pressure is accompanied by an increasing DNAPL
saturation.
Figure 5.1O.d gives the status of the Kueper et al. (1989) experiment after 313
seconds. The PCE has now fully penetrated into the #25 Ottawa sand lens, but is also
continuing to build above that lens. This is an example of how a non-wetting fluid will
continue to pool and flow laterally above a finer-grained lens even after penetration of
that lens. It is clear that: 1) in a natural aquifer, significant degrees of lateral flow will be
157
Sec. 5.4
100
#70 SILICA
" #50 SILICA
() #250ITAWA
v #16 SILICA
o
80
60
40
20
20
40
60
80
100
% SATURATION
Figure 5.9 Measured capillary pressure

curves of the tetrachloroethylene-water
system for the four sands used by Kueper
et al. (1989). (Adapted with permission
from Kueper et al. (1989).)
TABLE 5.1 Best-fit Brooks-Corey parameters and the results of

permeameter test for the four sands used in the physical model of
Kueper et al. (1989).
Sand
#16
#25
#50
#70
Silica
Ottawa
Silica
Silica
Permeability
(m 2 )
5.0
2.1
5.3
8.2
x
x
x
x
10- 10
10- 10
10- 11
10- 12
Pel
(cm H 2O)
So
3.77
4.43
13.5
33.1
3.86
3.51
2.49
3.30
0.078
0.069
0.098
0.189
caused by permeability variations; and 2) vertical migration of a DNAPL in heterogeneous

materials will not occur without some associated degree of lateral spreading.
The observation that NAPLs are often not able to cross capillary zones has been
used to advantage by the oil pipeline industry. When an underground pipe is sited in finegrained media, the pipe is often installed within a bed of very coarse-grained materials.
As the result of the capillary contrast between those materials and the surrounding media,
oil which leaks from the pipeline will often be collected and contained exclusively within
the coarser-grained medium.
Within the saturated zone, DNAPL movement can be affected by heterogeneities
which are not observable even with careful inspection. A most striking example of this is
the microscopic investigation of flow through a bed of glass beads conducted by Schwille
158
Chap. 5
Figure 5.10 Distribution of tetrachloroethylene (PCE) in the experiments by Kueper et al. (1989).
a. 126 seconds after initiation of the release; b. 220 seconds after initiation of the release; c. 245
seconds after the initiation of the release, at the point in time at which the PCE first penetrates the
#25 Ottawa sand lens; and d. 313 seconds after initiation of the release. (Adapted with permission
from Kueper et al. (1989).)
(1988a). In real-world aquifers, where it is possible only to characterize large-scale aquifer

properties, the small-scale heterogeneities which may control DNAPL behavior will be
difficult if not impossible to detect. Schwilles (I988a) experiments showed that a permeability decrease of only a factor of four can be sufficient to completely halt the inflow of
solvent. Within many naturally-occurring porous media, much greater local permeability
contrasts are to be expected. The parameters governing infiltration will in this case be
the lateral extent, bedding angle, and continuity of the low permeability layers.
5.5 DNAPL RETENTION CAPACITY OF POROUS MEDIA
The ability of a porous medium to exhibit a retention capacity for an infiltrating spill is
important in determining the spatial distribution of the solvent within the medium. The
retention capacity of a medium for a solvent (R.,) is defined here as the average volume
of solvent which is retained per unit volume of the aquifer, regardless of whether cer-
Sec. 5.6
159
tain lenses within the overall region are accessible to DNAPL or not. R, differs from
the residual saturation for a solvent in that the latter refers to the fraction of the pore
space which is occupied by solvent. Also, the term residual saturation is typically reserved for regions of porous media through which actual solvent migration has taken
place. The R, concept is frequently used along with spill volume and cross-sectional
area to predict the depth of penetration of a spill. Since heterogeneities in real systems
can lead to uneven flow patterns, simple models using R, values derived from smallscale laboratory tests should always be used with considerable caution. In heterogeneous
media, flow may occur preferentially in a few channels, and result in penetration depths
which are much larger than would be expected based on laboratory-estimated R, values.
Mercer and Cohen (1990) have reviewed the R, values documented in the literature,
and report a range of 3 to 30 Llm 3 for unsaturated sandy porous materials and 5 to 50
Llm 3 for sandy materials below the water table. Kueper et at. (1993) report a value of
8.6 Llm 3 for PCE below the water table in sandy deposits, while Brewster (1993) reports
a value of 3.2 Llm 3 for PCE in the same aquifer. Poulsen and Kueper (1992) report R,
values ranging from 4.9 Llm 3 to 12.6 Llm 3 for point releases of PCE in unsaturated sandy
deposits. The study by Poulsen and Kueper (1992) demonstrated that both R, and the
ultimate depth of migration can be expected to be functions of the release rate at ground
surface. Wilson and Conrad (1984) provide an extensive summary of the mechanisms
leading to residual formation as well as the porous media and fluid factors which govern
residual values in typical sands and gravels.
5.6 SOLVENT MIGRATION IN FRACTURED MEDIA
As with porous media flow, the movement of solvents within fractured media is controlled by the density and viscosity of the solvent, as well as the capillary pressures
in the system of interest (see also Chapters 4, 11, and 12). Once again, Schwille's
(1988a,b) experiments represent pioneering work in this area. They involved spilling
small volumes of chlorinated solvents between closely-spaced (0.1 or 0.3 mm separation) glass plates. The opposing faces of the plates were either smooth, or roughened
by sandblasting. Releases were conducted into saturated as well as partially-saturated
fractures. In hydraulically-rough, unsaturated fractures, a substantial quantity of solvent was retained as droplets and films within both the unsaturated and saturated
zones (Figure 5.11). This retention of solvent was both increased and more widely dispersed when the width of the fracture was reduced from 0.3 mm to 0.1 mm. Schwille
(1988b) also conducted a series of experiments using a network of vertical fractures
mounted in a near-horizontal network. Most of the published data on these experiments were obtained with both DNAPL solvents and miscible fluids. These experiments with idealized fractures demonstrated that substantial amounts of a non-wetting
DNAPL fluid can be retained within fractures. This conclusion has important ramifications for the cleanup of fractures contaminated with chlorinated solvents (see also
Chapter 12).
160
Chap. 5
\.
_.. .__.~.:$_ . . ._l
H =24 em
~.
.,
Figure 5.11 Tetrachloroethylene (Pf'E) release into a rough, unsaturated "fracture" with
aperture 0.2 mm. (Reprinted with permission from Schwille (l988a).)
In naturally-occurring fracture systems, variations in fracture aperture will have

a significant influence on DNAPL movement. Tsang and Tsang (1987) found significant
spatial variation in the aperture dimension in sections of individual fractures. They
concluded that water flow in many fractures will occur in channels as opposed to
occurring uniformly across the fractures. It can be anticipated that solvent flow will
be even more heterogeneous than water flow, due to the strong influence of capillary
effects. It will generally not be possible to characterize a real fracture system in sufficient detail to allow deterministic predictions of solvent behavior within a fractured
medium. Kueper and McWhorter (1991) outline the conditions necessary for a DNAPL
to enter a water-saturated fracture, and demonstrate the influence of a spatially-variable
aperture.
5.7 STUDIES USING MICROSCOPIC MODELS OF POROUS MEDIA
Schwille (1988a) conducted a series of microscopic experiments to examine the behavior

of chlorinated solvents in porous media. The media in these experiments were glass beads
of two different size ranges. One or more layers of beads was placed between two glass
sheets, and the model was placed into an outer trough which was used to control water
Sec. 5.7
161
Figure 5.12 Distribution of peE after infiltration into a saturated glass bead micromodel (bead diameter ~ 1 mm). (Reprinted with permission from Schwille (1988).)
level and content. Experiments were conducted under both unsaturated and saturated
conditions. Figure 5.12 shows the results obtained when the infiltration of PCE took
place under saturated conditions. The PCE is the non-wetting fluid with respect to water;
the result is infiltration into the medium as small fingers.
Figure 5.13 shows the microscopic behavior of PCE observed by SchwiJIe (l988a)
when the PCE encountered a lower-permeability zone (i.e., smaller grains). Penetration of that zone did not occur initially because the entry pressure for the non-wetting
PCE is greater in those zones. As has been noted above in related contexts, once the
thickness of the overlying pool becomes sufficiently large, then the PCE could begin
to penetrate into the lower zone. When doing so, it once again moves in preferential
pathways or channels. As is seen in the figure, these pathways often have dimensions
similar to the grain size. In some cases, the channels are widely separated. As has been
observed in sand models, once a preferential pathway is formed, the relative permeability of that zone for the chlorinated solvent substantially greater than for the bulk of the
medium. Consequently, continued solvent flow may be limited largely to those channels, leading to a substantial increase in the depth of penetration into the groundwater
zone.
Wilson et al. (1990) carried out an extensive examination of NAPL movement
using microscopic models. Their physical models consisted of etched glass plates which
were fused together. Because of the construction technique, the "pore structure" within
162
Chap. 5
Figure 5.13 Retention of tetrachloroethylene peE at the interface between two glass
bead layers in a saturated glass-bead micro-model. (Reprinted with permission from
Schwille (1988a).)
these models (Figure 5.14) was more variable than in Schwille's (l988a) models. Figure 5.15 shows the movement of a NAPL (Soltrol) through an initially water-saturated
medium. As with Schwille's (l988a) experiments, infiltration occurred by the formation
of many channels of solvent. As a consequence, only a fraction of the pore space in
the medium actually contained NAPL. During the subsequent infiltration of additional
NAPL, the pattern shown in the figure remained constant, resulting in a state of residual
saturation for the water (the wetting phase). When the NAPL flow was stopped and water
flow was induced in the opposite direction, the water displaced a substantial fraction of
the NAPL (Figure 5.16). The result was a stable state of residual saturation for the NAPL
(the non-wetting phase).
In addition to the work discussed above, Wilson et at. (1990) conducted micromodel experiments which contained zones of larger diameter pores. During NAPL infiltration, because of their lower entry pressures, they observed that the NAPLs spread
more quickly in the coarser zones. In experiments where coarser-grained zones extended
across the length of the model, Wilson et al. (1990) report that the bulk of the NAPL
flow occurred in those zones, and that most of the surrounding, finer-grained zones were
bypassed. After the infiltration of NAPL had stopped, water flow in the opposite direction
was initiated at a water velocity of ~ 14 em/day. Little of the NAPL in the coarser zones
was removed, while a significant fraction of the NAPL in the finer zones was removed.
163
Sec. 5.7
'"
..
~l>-\
n:
""""
~~
:".I
~
X'
tJo
-,
.-'l<
".J.
"
~~
.-{.
.Pt
'"'i.
"
~
XJ
T'()
1;lS
r<
;.--
HR
.A'\~ ~
lr~
.....-
r-, .....
-.<
11
f
~
-<
.l'
>J
C'r"
~
~~
.7"-
~
o,j
..,
~~
..
~I
~
..,...,.
..uN.'
I-
KR I
. \. \'
Figure 5.14 Pore structure of one of the micro-models used by Wilson et al. (1990).
(Reprinted with permission from Wilson et al. (1990).)
This is due to the fact that, at this velocity, capillary forces dominated over viscous
forces and the water was preferentially drawn into the finer pores. At a velocity of rv 140
em/day, viscous forces played a much larger role, and a substantial fraction of the NAPL
was displaced from the coarser zones.
164
Figure 5.15 NAPL movement through an initially water-saturated medium in a microscale model. (Reprinted with permission from Wilson et al. (1990).)
Chap. 5
Sec. 5.7
165
as
._ ...._L...._
Figure 5.16 Non-wetting (NAPL) residual saturation resulting from water displacement
of the NAPL in Figure 5.15. (Reprinted with permission from Wilson et al. (1990).)
166
Chap. 5
5.8 EXPERIMENTAL RELEASES OF CHLORINATED SOLVENTS

IN THE FIELD
5.8.1 General
As has been discussed above, it is very difficult to characterize the movement of chlorinated solvents at actual spill sites. At the same time, it is very difficult to reproduce
naturally-occurring micro- and macro-scale heterogeneities in laboratory models. Thus,
well-controlled field experiments can be an invaluable means of confirming how chlorinated solvents behave in the "real world". To date, the best-characterized field experiments have been conducted at the Canadian Forces Base Borden site in Ontario, Canada
(Poulsen and Kueper, 1992; Greenhouse et al., 1993; Kueper et al., 1993). These experiments include liquid chlorinated solvent releases in both the vadose and groundwater
zones.
The Borden site has been described in a number of previous publications (MacFarlane et al., 1983; Mackay et al., 1986; Sudicky, 1986; and Brewster, 1993). Briefly, the
Borden site is an unconfined aquifer consisting of fine to medium grained beach sands.
The water table in the vicinity of the field experiments described here ranges areally from
ground surface to a depth of 3 to 4 m, depending upon local topography. The aquifer is
relatively homogeneous, with small and medium scale (millimeters to meters) variations
in permeability of up to three orders of magnitude. Bedding in the sand ranges from near
horizontal to steeply dipping. Groundwater flow velocity at the site is on the order of
10 ern/day, and the organic carbon content of the sand is quite low. Laboratory studies
show that in Borden sand, chlorinated solvents are non-wetting fluids with respect to
water.

Two releases of small volumes of PCE (6 liters) into the vadose zone of the Borden
aquifer have recently been described by Poulsen and Kueper (1992). These releases took
place at a location where the aquifer exhibits a total thickness of approximately 9 m,
with the water table ~3.5 m below ground surface. The first release, referred to as the
"ponded release", involved the rapid release of 6 L of PCE over a period of 90 seconds.
In this release, the solvent ponded before it was able to infiltrate fully. The second release,
referred to as the "drip release", was carried out at a site 4.5 m away from the ponded
release and involved the slow release of 6 L of PCE over a period of 100 minutes.
After allowing for re-distribution, excavation of each release was carried out in
small lifts. It was noted for both releases that PCE had migrated through individual sand
laminations ranging in thickness from a few millimeters to a few centimeters. PCE was
found to be distinctly present in certain laminations and not in others. It was noted that
angled bedding existed at the site, and that PCE had migrated parallel to the bedding
at all times. This provided evidence that the migration process was governed to a large
degree by capillary forces. The ponded release migrated to a total depth of 2.1 m, and
the drip release to a depth of 3.2 m.
Sec. 5.8
167
Figures 5.17 and 5.18 illustrate the overall volumes within which the PCE migration
pathways occurred. The greater depth achieved in the drip release is attributed to the fact
that relative to the ponded release, higher capillary pressures were maintained on average
throughout the release. The overall average bulk solvent retention capacity (R,) values
for the ponded and drip releases were 12.6 and 4.9 L'rrr', respectively. These values are
lower than most estimates derived from laboratory experiments. The low values are due to
the influence of heterogeneity and the fact that only certain lenses and laminations below
the release points were invaded by PCE. Detailed permeability measurements revealed
that the hydraulic conductivities of those invaded lenses and laminations were generally
less than a factor of 1.7 different from adjacent, PCE-free lenses and laminations.
Figure 5.19 presents a plot of depth of PCE migration vs. volume released based
on the bulk R, values measured for the ponded and drip releases. The plot indicates that
a slow release of 200 L of PCE into Borden-type sand could reach a depth of 100 m.
These results have serious implications for the many slow releases of chlorinated solvents
which have occurred during the past 50 years at industrial sites. It must be concluded
from these experiments that even small releases of chlorinated solvents have the potential
to reach even deep water tables.
Figure 5.17 Rapid ("ponded") infiltration of Pf'E into unsaturated Borden sand.
(Reprinted with permission from Poulsen and Kueper (1992).)
168
Chap. 5
Figure 5.18 "Drip-like" infiltration of tetrachloroethylene (peE) into Borden sand.

(Reprinted with permission from Poulsen and Kueper (1992).)
200 ....,..--------r----~~
/
/
/
/
/
""0 150
/
Q)
/
/
0...10 0
Vl
/
/
/
/
Q)
50
:::J
>
Instantaneous
Drip Release
/
/
/
o -+-----,-.--r-r--r---r-~~-----,~
25
50
75
100
Depth of Migration
125
(m)
Figure 5.19 Depth of migration vs.

volume spilled at ground surface using
cross-sectional area of migration and
overall bulk residual contents. (Reprinted
with permission from Poulsen and Kueper
(1992).)
Sec. 5.8
169

Two large-scale releases of PCE into the groundwater zone have been conducted at the
Borden site. The first of these was the release of 231 L of PCE at the water table into
a 3 m x 3 m x 3.4 m deep sheet-pile cell (Kueper et al., 1993). As in the vadose zone
releases, the PCE was dyed red with SUDAN IV to aid in visual location of the PCE
during excavation and coring.
During and after the 231 L release, the movement of the PCE was followed using
time-domain reflectometry (TDR). Following cessation of DNAPL movement, the upper
0.9 m of the cell was excavated in shallow lifts, and three continuous cores were taken
through the remaining thickness of the cell. During both the excavation and coring processes, small samples were taken and analyzed for PCE content. A histogram of PCE
saturation for all samples is presented in Figure 5.20. Note that the saturation distribution is highly variable, with a high frequency of low values. Laboratory measurements
presented in Kueper et at. (1993) suggest that all saturation values above 15% represent
pooled PCE, and that values less than that represent residual DNAPL. During excavation
it was noted that the PCE was found to be present in certain laminations and distinctly
absent from others, with what appeared to be subtle variations in texture controlling migration pathways. The PCE migrated preferentially through the relatively coarser-grained
Core #3
Core #1
Core #2
198.8
198.8
198.8
198.4
198.4
198.4
198.0
198.0
198.0
197.6
197.6
197.6
197.2
197.2
197.2
m 196.8
196.8
196.8
196.4
196.4
-E
c
0
ctI
>
(1)
1960
r=:==z
L.
0
0
196.4
~
196.0
196.0
0'>
ro
'" '"
OJ
ro
./>0
0
0
PCE Saturation
(% of pore space)
W
0
'"
0'>
ro
OJ
'"
./>0
W
0
'" '" '" '"
PCE Saturation
(% of pore space)
Ol
ro
OJ
ro
./>0
0
0
0
PCE Saturation
(% of pore space)
Figure 5.20 Tetrachloroethylene (peE) saturation profiles obtained from subsampling

of three vertical cores. (Reprinted with permission from Kueper et al. (1992).)
W
0
'"
170
Chap. 5
laminations, consistent with the fact that PCE is non-wetting with respect to water in
Borden sand. Lateral spreading of the PCE allowed the liquid PCE to reach the cell walls
at numerous locations.
In the second large-scale release of PCE at Borden, 770 L of PCE was spilled into a
9 m x 9 m x 3.3 m deep, sheet-pile cell near the site of the first release. The release took
place from a point source at the water table over a 70 hour period. The primary purpose
of this release was to assess the extent to which a variety of geophysical methods could
characterize the movement and final distribution of PCE in the aquifer. The methods
included ground-penetrating radar, neutron probe, TDR, gamma-gamma logging, and
electrical resistance logging. All of these techniques, with the possible exception of
gamma-gamma logging, were able to track the progress of the PCE into the groundwater
zone. They were able to confirm that the PCE front moved relatively quickly (i.e., in a
matter of hours) down through the medium. They also indicated that there was significant
lateral movement of the PCE due to the presence of subtle structures within the aquifer,
and that the PCE encountered the side walls of the cell at a depth of approximately I m. A
significant portion of the PCE became immobilized in pools on top of lower-permeability
zones and in slightly coarser lenses within the medium. As in the 231 L release, coring
confirmed that the PCE migration paths were governed by centimeter-scale variations in
hydraulic conductivity.
The success of the geophysical monitoring of PCE in the 770 L release was due
in large part to the availability of pre-spill data for the site. Thus, it can be concluded
that after-the-fact use of surface and borehole measurements to detect where solvents are
located at actual sites will likely be difficult due to the inherent variabilities in the geology
at most sites. On the other hand, the potential for using geophysical monitoring to follow
the progress of the remediation of subsurface chlorinated solvent contamination is quite
good: the initial "pre-cleanup" data for a site can be compared to changes observed as
the remediation progresses.
5.9 CHLORINATED SOLVENT VAPOR TRANSPORT
IN THE VADOSE ZONE
Residual solvents left behind in the vadose and groundwater zones can act as longterm sources of contamination in the subsurface. In the vadose zone, the solvents can
dissolve into infiltrating water, or they can volatilize into pore air. The volatilities of
most of the common chlorinated solvents are sufficiently high that vapor transport in
the vadose zone is important. Vapor transport can occur both by molecular diffusion and
by advection. Both of these processes have been examined using large-scale physical
models. Advection can arise due to density effects caused by the presence of the organic
vapors in the air, and due to external pressure gradients. Only density-driven advective
flow as might occur near source zones will be discussed here. Numerical modeling of
vapor transport is discussed in Chapter 6.
The diffusive transport of contaminants has been investigated in several large-scale
experiments. Hughes et at. (1992) observed the diffusion of trichloroethylene (TCE) from
a residual source in the vadose zone at the Borden field site. In their experiments, the
171
Chlorinated Solvent Vapor Transport in the Vadose Zone
Sec. 5.9
source consisted of a cylindrical zone (~1 m in diameter and 1 m high, and located
near ground surface) into which was placed sand containing 4% TCE by volume. The
depth to the water table was 3 to 4 m. The moisture content for most of the soil profile
was near field capacity (~1O% of the total soil volume). They observed that TCE vapors
moved more than 5 m in ~20 days. Numerical modeling of the experiments indicated
that this was consistent with the rates expected due to molecular diffusion.
MacPherson (1991) also conducted large-scale diffusion experiments in a medium
sand. In their work, a liquid mixture of organic contaminants was allowed to vaporize
and diffuse into the vadose zone from a simulated pool of organic liquid near the water table. Subsurface conditions were quite similar to those described by Hughes et at.
(1992). The MacPherson (1991) data show large differences in the diffusion rates of the
various compounds (Figure 5.21); the differences can be explained based on differences
in partitioning of the organics to the soil and to the pore water. The results obtained with
,,-... 2
ISOOCTANE
"-'
1
SOURC~
~ 0
~
~ 2
~
:>
<
FLUOROBENZENE
~
~
C/)
1--4
CUMENE
1
0
0
DISTANCE (m)
Figure 5.21 Diffusion of isooctane, cumene, and 1,2-dichlorobenzene in a large sand
model. (Adapted with permission from MacPherson (1991).)
10
172
Chap. 5
numerical models which focussed on the role of molecular diffusion compared well with
the experimental data.
Many chlorinated solvents have both sufficient volatility and molecular weight that
density-driven advective flow can be important in coarse media. This has been examined
by Schwille and Weber (1991) and Johnson et al. (1992). Johnson et at. (1992) conducted
experiments with TCE and with dichlorfluoromethane (Freon 22) in both medium sand
and pea gravel. For TCE, density-driven flow was found to be very important in the pea
gravel, but played only a minor role in the sand (Figure 5.22). For Freon 22, which has
a vapor density three times that of air, density-driven flow was found to be significant
even in the sand. Johnson et al. (1992) found that the numerical model developed by
Mendoza and Frind (1990a,b) supported their conclusions regarding the behavior of both
TCE and Freon 22.
Diffusive and advective transport can carry chlorinated solvent vapors towards the
capillary zone (and therefore the groundwater zone) over areally-extensive regions. Once
at the capillary zone, the vapors can be transported across the water table by infiltration,
water-table fluctuations, and/or diffusive/dispersion. Movement of TCE across the water
table has been examined using both physical and numerical models by McCarthy and
IllOUR
SAND
0.33 HOURS
PEAGRAVEL
:r
.. HOUR
ITI
:r
'~pI'
12HOUR
:r
oJ
C
48 HOL'R
:r
oJ
DISTANCE(M)
.,
Figure 5.22 Diffusion and density-driven vapor transport of trichloroethylene (TCE) in sand and
pea gravel. (Adapted with permission from Johnson et al. (1992).)
Sec. 5.10
Chlorinated Solvent Dissolution in the Groundwater Zone
173
Johnson (1993). They examined TCE movement from groundwater under conditions
of steady groundwater flow with both a stationary and a fluctuating water table. In a
sand with an hydraulic conductivity of '"'-'0.1 cmls and at a groundwater velocity of
0.1 mid, they concluded that mass transfer was controlled by vertical dispersion in the
groundwater zone, and that the dispersion was very small (on the same order as aqueous
diffusion). Nevertheless, given the potentially large areal extent of many vapor plumes,
and the very low drinking water limits for many chlorinated solvents, movement of
vapors into the groundwater zone may be a significant mechanism for the contamination
of shallow groundwater.
5.10 CHLORINATED SOLVENT DISSOLUTION
IN THE GROUNDWATER ZONE
In the groundwater zone, chlorinated solvents become immobilized as residual droplets

or ganglia, and as pools perched on top of lower-permeability zones. The effects of
groundwater flow and solubility limits cause residual and pooled solvent to dissolve at
significantly different rates. Dissolution from both residual and pools has been examined
experimentally. Van der Waarden et al. (1971), Fried et al. (1979), and Miller et at.
(1990) examined dissolution from residual solvent in porous media. These experiments
were carried out in I-D columns where the medium was first mixed with residual-levels
of the solvent, then packed into a column. Two unrealistic features of these experiments
were that: I) the residual was distributed very uniformly within the medium; 2) the I-D
geometry of the column forced the water to flow through the residual zone.
In order to avoid the limitations of a I-D column model as discussed above, Anderson et al. (1992a) conducted experiments in a 3-D physical model. Solvent was emplaced
by spilling PCE inside of a cylindrical, sand-containing column situated inside of a larger
tank of sand. Dissolution from this zone was allowed to begin when the cylinder was
pulled out and water was allowed to begin to flow around and through the residual
zone. The water flow through the residual zone was somewhat reduced (by 20%) due to
the presence of the residual solvent. The water that did flow through the residual zone
attained saturation concentrations quickly. At actual sites, residual zones may be relatively small and widely separated from one another. As a consequence, with increasing
distance from the source, the initially saturated concentrations from the residual zones
may become diluted significantly. This may be one reason why the concentrations of
chlorinated solvents in contaminated groundwater are often far below saturation, even
at sites where a pure "free-product" solvent is known to be present in the groundwater
zone (Anderson et al., 1992b).
As discussed above, experimental data suggest that at many sites, a significant
portion of the mass of the chlorinated solvent(s) in the groundwater zone is present in
pools. Unlike solvent that is suspended in a state of residual saturation, the solvent in long,
flat pools will have limited contact with the flowing groundwater. Consequently, the rate
of dissolution will be limited by vertical dispersion of the dissolved solvent up into the
flowing groundwater. Schwille (I 988a) studied this process experimentally. Johnson and
Pankow (1992) examined Schwille 's (1988a) data and concluded that because vertical
174
Chap. 5
mixing process is quite weak, the lifetimes of pools chlorinated solvents will usually
be measured in decades to centuries (Figure 5.23). They also concluded that increasing
the groundwater flow rate over the pools (e.g. by pumping) might be only marginally
successful at shortening pool lifetimes.
Rivett et at. (1991) conducted a large-scale dissolution experiment at the Borden
site. The source zone for the experiment consisted of a three-component mixture of solvents which was emplaced below the water table in a rectangular volume. The source
chemicals were mixed with sand at a value below residual saturation just before emplacement. Groundwater concentrations of the dissolved solvent chemicals were monitored in
a 3-D array of point samplers. Dispersion in the transverse directions was weak, and as
a result a long, thin groundwater plume developed. Once the multicomponent nature
of the source liquid was taken into consideration, the concentrations observed in the
heart of the plume were found to be near the values expected for saturation with that
source.
700
600
~
CZ)
500
<
~
400
---~
300
=s
~
E-t
200
10m
100
2m
1m
0
0
0.5
VELOCITY (m/d)
Figure 5.23 Dissolution time for TeE vs. groundwater velocity for pool lengths. (Reprinted
with permission from Johnson and Pankow (1992).)
Sec. 5.12
References
175
5.11 CONCLUSIONS
The physical model studies discussed in this chapter demonstrate the complexity of
chlorinated solvent movement in porous and fractured media. The results show that
very subtle changes in permeability can greatly affect the subsurface distribution of
chlorinated solvents, with differences in permeability of a factor of two or less capable
of completely redirecting solvent movement. Thus, the past movement of solvents at
industrial sites probably cannot be calculated in a deterministic manner. There can be
significant pooling on the tops of lower-permeability lenses in the saturated zone. This
type of distribution can substantially limit the rate at which dissolution will occur, leading
to very long lifetimes of chlorinated solvents in the subsurface (see also Chapter 7).
The complex nature of chlorinated solvent migration in the subsurface makes it
very difficult to predict the depth to which solvents will penetrate the subsurface at a
given site. Factors such as bedding angle, length scales of layers, release rate, permeability
contrasts, and water content all affect the retention capacity. Since all of these parameters
will usually not be known for a particular site of interest, it is clear that estimates of
depth of penetration will have a significant degree of uncertainty associated with them.
Physical model studies obviously comprise an important component of DNAPL
research. However, to date, most controlled experiments have been conducted using wellsorted sands and gravels. This is the case because of the relative ease of working with
these media. The result, though, is a relatively incomplete picture of solvent behavior.
Model studies in fine-grained and/or heterogeneous media, as well as in fractured rock,
should be viewed as important research activities for the next decade.
ACKNOWLEDGEMENTS
5.12 REFERENCES
Anderson, M. R. (1988) Ph.D. Thesis, The Dissolution and Transport of Dense Non-Aqueous Phase
Liquids in Saturated Porous Media, Oregon Graduate Institute, Portland, Oregon.
Anderson, M. A., R. L. Johnson, and J. F. Pankow (1992a) "Dissolution of dense chlorinated
solvents into groundwater: I. Dissolution from a well-defined source geometry", Ground Water,
30, 250-256.
Anderson, M. A., R. L. Johnson, and 1. F. Pankow (1992b) "Dissolution of dense chlorinated
solvents into groundwater. "3. Modeling contaminant plumes from fingers and pools of solvent",
Environ. Sci. Technol., 26, 901-908.
176
Chap. 5
Brewster, M. L. (1993) Observed migration of a controlled DNAPL release by ground penetrating

radar, M.Sc. Thesis, University of Waterloo, Waterloo, Ontario.
Brooks, R. H. and A. T. Corey (1966) "Properties of porous media affecting fluid flow", J. Irrig.
and Drainage Div., Am. Soc. Civil Eng., 92(IR2), 61-88.
Dietz, D. N. (1967) "Pollution of permeable strata by oil components". In: Joint Problems of the
Oil and Water Industries, P. Hepple (Ed.), Institute of Petroleum, London.
Dracos, T. (1978) "Theoretical considerations and practical implications on the infiltration of
hydrocarbons in aquifers". In: Proceedings: International Symposium on Ground Water Pollution
by Oil Hydrocarbons, Prague, Czechoslovakia, 127-137.
Eckberg, D. K. and D. K. Sunada (1984) "Nonsteady three-phase immiscible fluid distribution in
porous media", Water Resources Research, 20, 1891-1897.
Fried, J. J., P. Muntzer, and L. Zilliox (1979) "Ground-water pollution by transfer of oil hydrocarbons", Ground Water, 17, 586-592.
Glass, R. J., J. Y. Parlange, and T. S. Steenhuis (1989) "Wetting front instability. I. Theoretical
discussion and dimensional analysis", Water Resources Research, 25, 1187-1194.
Greenhouse, J., M. Brewster, G. Schneider, D. Redman, P. Annan, G. Olhoeft, J. Lucius, K. Sander
and A. Mazzella (1993) "Geophysics and solvents: The Borden experiment", The Leading Edge,
April, 261-267.
Hughes, B. M., R. W. Gillham, and C. A. Mendoza (1992) 'Transport of trichloroethylene vapours
in the unsaturated zone: A field experiment". In: Proceedings: Subsurface Contamination by
Immiscible Fluids, Calgary, Alberta, April 18-20, 1990. K. U. Weyer (Ed.) A. A Balkema,
Rotterdam. 588 pp. Order from A. A. Balkerna, Old Post Road, Brookfield, Vermont 05026.
Johnson, R. L. (1990) Unpublished data. Oregon Graduate Institute, Portland, Oregon.
Johnson, R. L. and 1. F. Pankow (1992) "Dissolution of dense chlorinated solvents into groundwater.
Johnson, R. L., K. A. McCarthy, M. Perrott, and C. A. Mendoza (1992) "Density-driven vapor transport: Physical and numerical modeling". In: Subsurface Contamination by Immiscible
Fluids, K. U. Weyer (Ed.), A. A Balkema, Rotterdam. 576 pp.
Kueper, B. H., W. Abbott, and G. Farquhar (1989) "Experimental observations of multiphase flow
in heterogeneous porous media", J. Contaminant Hydrology,S, 83-95.
Kueper, B. H. and E. 0 Frind (1988) "An overview of immiscible fingering in porous media", 1.
Kueper, B. H. and E. 0 Frind (1989) "An overview of immiscible fingering in porous media", J.
Contaminant Hydrology,S, 83-95.
Kueper, B. H. and E. O. Frind (l991a) "Two-phase flow in heterogeneous porous media, I. Model
Kueper, B. H. and E. O. Frind (l991b) 'Two-phase flow in heterogeneous porous media, 2. Model
application", Water Resources Research, 27, 1059-1070.
Kueper, B. H. and D. B. McWhorter (1991) "The behavior of dense non-aqueous phase liquids in
Kueper, B. H., C. S. Haase, and H. King (1992) "Leakage of DNAPL from waste impoundments constructed in fractured rock and clay: Theory and case history", Canadian Geotechnical
Journal, 29, 234-244.
Sec. 5.12
References
177
Kueper, B. H., D. Redman, R. C. Starr, S. Reitsma, and M. Mah (1993), "A field experiment to
study the behavior of tetrachloroethylene below the water table: Spatial distribution of residual
and pooled DNAPL" Ground Water, 31, 756-766.
Macfarlane, D. S. J. A. Cherry, R. W. Gillham, E. A. Sudicky (1983) "Migration of contaminants
in groundwater at a landfill: a case study. I. Groundwater flow and plume delineation", J.
Hydrology, 63, 1-29.
Mackay, D. M., D. L. Freyberg, P. V. Roberts, and J. A. Cherry (1986) "A natural gradient
experiment on solute transport in a sand aquifer", Water Resources Research, 22, 2017-2030.
MacPherson, J. R. (1991) Gas Phase Diffusion of Organic Compounds in Porous Media: Physical
and Numerical Modeling, Ph.D. Thesis, Oregon Graduate Institute, Portland, Oregon.
Mendoza C. A. and E. O. Frind (l990a) "Advective-dispersive transport of dense organic vapors
Mendoza C. A. and E. O. Frind (1990b) "Advective-dispersive transport of dense organic vapors
Miller, C. T., M. M. Poirier-McNeill, and A. S. Mayer (1990) "Dissolution of trapped nonaqueous
phase liquids: Mass transfer characteristics", Water Resources Research, 26, 2783-2986.
Mercer, J. W. and R. M. Cohen (1990) "A review of immiscible fluids in the subsurface: Properties,
models, characterization and remediation", 1. Contaminant Hydrology, 6, 107-163.
Poulsen, M. M. and B. H. Kueper (1992) "A field experiment to study the behavior of tetrachloroethylene in unsaturated porous media", Environ. Sci. Technol., 26, 889-895.
Rivett, M., S. Feenstra, and J. Cherry (1991) "Field experimental studies of a residual solvent source
emplaced in the groundwater zone". In: Proceedings: Petroleum Hydrocarbons and Organic
Chemicals in Ground Water Conference, 20-22 November, 1991, Houston, Texas, 283-299.
Schwille, F. (1984a) leichtfluchtige Chlorkohlenwasserstoffe in Porosen and Kluftigen MedienModelversuche, Besondere Mitteilungen zum Deutschen Gewasserkundlichen Jahrbuch, Nr.46,
Herausgegeben von der Bundesanstadt fur Gewasserkunde in Koblenz.
Schwille, F. (1984b) "Migration of organic fluids immiscible with water in the unsaturated zone".
In: Pollutants in Porous Media, Ecological Studies, 47, Springer-Verlag, 27-48.
Schwille, F. (I988a) Dense Chlorinated Solvents in Porous and Fractured Media: Model Experiments, translated from the German by J. F. Pankow, Lewis Publishers, Boca Raton, Florida.
Schwille, F. (l988b) "Fluid-mechanical aspects of the migration of chemicals in fractured media".
In: Groundwater Flow and Quality Monitoring, E. Custodio et al., (Eds.), D. Reidel Publishing
Co., 515-537.
Schwille, F. and D. Weber (1991) "Model experiments on gravity spreading of heavy organic
vapors in the zone of aeration", Schr. Angew. Geol. Karlsruhe, 12, 153-211.
Sudicky, E. A. (1986) "A natural gradient experiment on solute transport in a sand aquifer: Spatial
variability of hydraulic conductivity and its role in the dispersion process", Water Resources
Research, 22, 2069-2082.
Tsang, Y. W. and C. F. Tsang (1987) "Channel model of flow through fractured media", Water
Van Dam, J. (1967) "Migration of hydrocarbons in a water bearing stratum". In: Joint Problems
of the Oil and Water Industries, P. Hepple (Ed.), Institute of Petroleum, London, 55-95.
178
Chap. 5
Van der Waarden, M., A. L. A. M. Bridie, and W. M. Groenewoud (1971) 'Transport of mineral
oil components to groundwater. 1. Model experiments on the transfer of hydrocarbons from a
residual oil zone to trickling water", Water Research,S, 213-226.
Van der Waarden, M., W. M. Groenwoud, and A. L. A. M. Bridie (1977) 'Transport of mineral
oil components to groundwater. II. Influence of lime, clay, and organic soil components on the
rate of transfer", Water Research, 11, 359-365.
Wilson, 1. L. and S. H. Conrad (1984) "Is physical displacement of residual hydrocarbons a realistic
possibility in aquifer restoration?", In: Proceedings of the NWWAIAPI Conference on Petroleum
Hydrocarbons and Organic Chemicals in Ground Water-Prevention, Detection and Restoration,
National Water Well Association, November 5-7, 1984, Houston, Texas.
Wilson, J. L., S. H. Conrad, W. R. Mason, W. Peplinski, and E. Hagan (1990) Laboratory Investigation of Residual Liquid Organics, EPA/600/6-90/004, Cincinnati, Ohio.
6
Vapor Migration in the Vadose
Zone
Carl A. Mendoza" Richard L. Iohnson/, and Robert W. Gillham"
1 Department
of Geology
University of Alberta
Edmonton, Alberta
Canada,T6G 2E3
2 Department

r.o. Box 91000
Portland, Oregon
U.S.A. 97291-1000

Waterloo, Ontario
Canada N2L 3G1
Page
6.1
6.2
6.3
6.4
6.5
6.6
6.7
6.8
6.9
6.10
Introduction
Phase Partitioning
6.2.1 General
6.2.2 Vaporization
Gaseous Transport
6.3.1 General
6.3.2 Diffusion
6.3.3 Advection
Aqueous Transport
Interactions with the Atmosphere
Interactions with Groundwater
Soil-Gas Surveys
6.7.1 General
6.7.2 Sampling
A Pair of Field Experiments
Conclusions
References
180
182
182
182
185
186
186
186
187
187
188
189
192
192
193
194
194
194
194
195
197
198
179
180
Vapor Migration in the Vadose Zone
Chap. 6
ABSTRACT
When a dense non-aqueous phase liquid (DNAPL) is released to the unsaturated zone,
it will migrate downwards and capillary forces will immobilize some of the liquid in
the pore spaces as residual. Pools of DNAPL may also accumulate on lenses of lower
permeability and on top of the capillary zone. This DNAPL will then begin to vaporize into the gas phase and migrate by diffusion away from the source. Gaseous advection due to pressure or density gradients can also play a role in the overall transport process. As the vapors migrate, they will partition into the aqueous and solid
phases. These partitioning processes will tend to retard the rate of vapor migration.
Partitioning into soil water will also make contaminants readily available for aqueous-phase transport. Eventually, the contaminants will reach the capillary fringe and
dissolved DNAPL will be transported into groundwater by diffusion across the water
table, by infiltrating water, and as the result of water-table fluctuations. At the ground
surface, vapors can be lost to the atmosphere if no impermeable cover is present; alternatively, a cover will retain vapors and may lead to increased lateral migration.
Since vapor transport by diffusion is commonly a rapid process that acts in all directions away from a residual source, groundwater contamination may occur over a
wider area and much more quickly than if infiltration of either water or free-product
were the only processes transporting DNAPL-related contaminants into the saturated
zone. Groundwater contamination may even occur hydraulically up- or cross-gradient
from a residual source. Failure to recognize the role of vapor transport can therefore lead to serious misinterpretations of the groundwater contamination at a given
site.
6.1 INTRODUCTION
As a group, the dense chlorinated solvents are quite volatile, and can thus partition into
the gas phase and migrate great distances as vapors in the unsaturated zone. Because
these vapors can also partition into any soil moisture that is present, vapor transport processes have the potential to contaminate underlying groundwater resources. In this way,
even if the residual source is completely contained within the vadose zone, vapor transport can cause rapid and widespread groundwater contamination. A conceptual model
illustrating some of the processes that are responsible for vapor transport and subsequent
groundwater contamination is presented in Figure 6.1.
As has been discussed in prior chapters, the dense chlorinated solvents are dense
non-aqueous phase liquid (DNAPLs). As a DNAPL moves downwards in the unsaturated
zone, some of the liquid will be immobilized by capillary forces in the pore spaces as
residual. Pools of DNAPL may also form on the tops of lower permeability lenses.
In the vadose zone, a DNAPL solvent will vaporize into the gas phase and migrate
by gaseous diffusion in all directions away from the source. Gaseous advection due
to pressure or density gradients may also play a role in the transport process. As the
vapors migrate, the partitioning into the aqueous and solid phases that occurs will tend
to retard the rate of vapor migration. Eventually the contaminants will reach the top of
Sec. 6.1
181
Introduction
Infiltration
~
Surface Cover
::.---T-~---
Disso~n
into
Groundwater
..
Groundwater Flow
Figure 6.1 Conceptual model of vapor transport processes from a residual DNAPL
source in the unsaturated zone. (Reprinted with permission from Mendoza and Frind,
1990a.)
the capillary fringe. Contaminants in the dissolved phase can then be transported into
groundwater by diffusion across the water table, by infiltrating water, and as the result
of water-table fluctuations. At the ground surface, vapors may be lost to the atmosphere
if no impermeable cover is present. On the other hand, if a cover is present, it will retain
vapors and may lead to increased lateral migration.
Since vapor transport by diffusion is commonly a rapid process that acts in all
directions away from a residual source, groundwater contamination may occur over a
wider area, and much more quickly, than if infiltration of water and/or the chlorinated
solvent itself were the only mechanisms transporting contaminants from a release in the
unsaturated zone. Groundwater contamination may even occur hydraulically up- or crossgradient from the residual source. Thus, at many sites, failure to recognize the role of
vapor transport could lead to serious misinterpretations of the potential for groundwater
contamination. This chapter discusses all of the above aspects of passive vapor transport
in the vadose zone. Although this discussion is oriented towards the behavior of vapors
derived from dense chlorinated solvents, many of the same principles apply to the vapors
of other volatile organic compounds and mixtures (e.g., gasoline).
When dealing with flow and transport in several phases, as we are here, the notation
can become cumbersome. To simplify that notation, most parameters that apply to the
gas (air) phase have not been given the subscript a. All other parameters have w, l, or
s subscripts to show that they pertain to the aqueous (water), liquid (DNAPL), or solid
phases, respectively. In addition, it is implicitly assumed that mixtures of compounds
are being discussed, and so subscripts are not generally used to distinguish between
constituents.
182
Chap. 6
6.2 PHASE PARTITIONING

6.2.1 General
The partitioning of contaminants between the different phases in the unsaturated zone
plays an important role in vapor transport. It is, in fact, the process of partitioning
between the organic liquid phase and the gas phase that provides the source necessary
for vapor migration. One-dimensional screening models, such as that developed by Jury
et at. (1990), use phase partitioning concepts to indicate the likely fate of contaminants in
the unsaturated zone. For given site conditions, these models indicate whether a particular
organic constituent is more likely to: 1) be transported to groundwater in the gas phase
or the aqueous phase; 2) be lost to the atmosphere; or 3) remain more-or-Iess in place.
6.2.2 Vaporization
The extent to which a compound vaporizes is determined by both its vapor pressure
and the composition of the liquid from which the vapors are derived. For a liquid that
contains only one constituent, the equilibrium vapor concentration is given simply by the
pure compound vapor pressure, expressed in appropriate units. However, if the residual
liquid is a mixture of different constituents, the equilibrium vapor concentration of each
constituent can be approximated by Raoult's Law which states that
po
c=X t -
RT
(6.1)
where P" is the pure compound vapor pressure of a particular constituent at the temperature T (Kelvin) of interest, Xl is the mole fraction of the compound in the source
liquid, and R is the gas constant. The vapor concentration c is expressed in units of mols
per unit volume.
Since the vapor concentration of a given compound at the source depends on the
mole fraction of that compound in the liquid phase, the source vapor concentration will
change over time as the compound is depleted or enriched relative to other compounds
in the source liquid. If the different compounds in the original liquid-phase source are
in equal proportions, the more volatile constituents will preferentially vaporize and become depleted relative to the less volatile compounds. In general, then, the source vapor
composition over time will be a function not only of the values of P" for the various
constituents, but also of the values of Xl for those constituents.
Bloes et al. (1992) described a laboratory column experiment in which a mediumgrained sand was brought to residual saturation with an organic liquid consisting of
approximately equal mass fractions of benzene, trichloroethylene (TeE) and toluene. A
constant flow of air was then passed through the column. Figure 6.2 shows experimentallymeasured and model-simulated vapor concentrations in the effluent. Applicable experimental parameters and chemical properties are given in Table 6.1. The vapor concentrations for all three compounds that were observed initially are seen to be: 1) less than their
pure phase equilibrium concentrations; and 2) consistent with Raoult's Law. As the most
volatile compound (benzene) was depleted, the concentration of the next-most volatile
Sec. 6.2
183
Phase Partition ing
300
250
r-...
< 200
01
'-'"
c:
.s 150
+'
o
c:
L..
+'
~ 100
c:
o
o
oeGO> benzene
- _ . TCE
50
D IHlIHI
toluene
18
15
20
time
Figure 6.2 Breakthrough curves for vapor extraction of a residual mixture of organic
liquids in a one-dimensional laboratory column. Points are measured values; solid lines
are predicted values based on Raoult's Law. (Reprinted with permission from Bloes et
al.,1992.)
TABLE 6.1
Experimental Parameters for Column Study of Bloes et al. (1992).
Vapor concentrations
(mg/L)
Initial column
value
Measured initial
residual fractions
Pure
compound
Compound
Mass
Mole
Literature
value
Calculated
by Eq. (6.1)
Measured
benzene
0.351
0.432
379
164
158
TCE
0.349
0.255
497
127
114
toluene
0.300
0.313
117
37
43
compound (TCE) increased to a maximum, reflecting its enrichment in the remaining

liquid mixture. Similar behavior was observed for toluene as the TCE was depleted in
the source liquid. A comparison of the experimental data with the solid lines in Figure 6.2 indicates good agreement between the observed results and the predictions of a
184
Chap. 6
mathematical model based on Raoults Law. Analogous behavior is to be expected in the

vadose zone as vapors passively migrate away from a multicomponent residual source.
Kinetic mass-transfer effects do not appear to playa significant role in the vaporization of residual organic chemicals during passive migration. This statement is
supported by observations from vapor extraction experiments where air was constrained
to flow through porous media containing residual organic liquids (e.g., Johnson et al.,
1987; McClellan and Gillham, 1990; and Bloes et al., 1992). Because any kinetic partitioning limitations should be accentuated under forced advection, and because saturated
vapors were nevertheless observed in the presence of free-product during these studies,
we conclude that kinetic limitations are not likely to be important under passive migration
circumstances.
In each of the studies referred to above, it is probable that the source liquid was
in direct contact with the gas phase. Direct contact will occur in any organic liquid/soil
system, and in any organic liquid/water/soil system in which the organic liquid is nonwetting (of the soil phase) relative to water. However, in a system of the latter type in
which the organic liquid is the wetting phase (relative to water), such a liquid would be
restricted to the smaller pores and might have a film of water between it and the gas
phase. In such a case, it is conceivable that non-equilibrium (i.e., kinetic effects) related
to rate-limiting diffusion through the aqueous phase might be observed; to our knowledge
a system of this type has not been investigated.
The vapor pressures of the chlorinated solvents increase significantly with increasing temperature. Figure 6.3 illustrates this dependence on temperature for TCE. The
temperature dependence of the vapor pressure of a given compound is commonly calculated from the Antoine Equation whose regression coefficients are tabulated in standard
chemistry reference texts (e.g., Stephenson and Malanowski, 1987; Dean, 1985; and
Weast, 1987).
The strong dependence of vapor pressure on temperature has several consequences
for vapor transport. Firstly, the temperature will control the vapor concentrations in the
source area and thus the vapor concentrations throughout the unsaturated zone. Secondly,
ro~
~o
:J
:Il
::;
LI)
et
o 0
l}o
>
Stephenson and Malanowski
OJ
Dean
::::::;':":!:::::::::_-"
o
~+rrrrrnrn'-''-'''''''''''''''''''TTTTTTTTTTTTTTTTTTTTl
-5
15
Temperature (Celsius)
35
Figure 6.3 Temperature-dependent

relationships and literature data values for
the vapor pressure of trichloroethylene
(TeE). (Adapted with permission from
Mendoza et al.. 1990.)
Sec. 6.2
185
Phase Partitioning
if the porous medium is sufficiently coarse that density-induced advection is important,

increased temperature will increase the propensity for gas-phase advection by increasing
the source concentrations and thus the density gradients in the source area (see Section 6.3.3).
Vapor sampling in the field commonly indicates that the concentrations observed
within suspected chlorinated solvent source areas are not as high as might have been
expected based on equilibrium partitioning. The possible reasons for this may include
one or more of the following:
heterogeneity in both porous media properties and residual fluid distribution (see
Chapter 5);
uncertainty concerning the actual location of the residual source;

not accounting for mixtures of compounds;
poor sampler installation or sampling techniques;
recent gas-phase advection caused by vapor extraction systems; and
not correcting concentrations for temperature variations.

In the absence of forced advection, the assumption of equilibrium partitioning between
the gaseous and aqueous phase appears to be valid for chlorinated solvent compounds
in the unsaturated zone (Johnson et al., 1987). This equilibrium partitioning process is
represented by Henry's Law as
c
H=Cw
(6.2)
where c and C w are the concentrations in the gaseous and aqueous phases, respectively, in
units of mols per unit volume. The parameter H is then the dimensionless Henry's Law
constant expressed on a molar basis (i.e., mol/volume per mol/volume). For compounds
which are only slightly soluble in water, common means for estimating the Henry's Law
constant is (Mackay et al., 1979)
pOlS
H=--
RT
(6.3)
where S is the solubility of the compound in water in units of mols per volume.
As is indicated in Eq.(6.3), the Henry's Law constant for a given compound depends on both the compound's vapor pressure and its solubility. Because vapor pressures
generally increase strongly with temperature (see Figure 6.3), and solubilities increase
moderately with temperature, Henry's Law constants increase moderately with temperature. The reader should be aware that there are several alternate sets of units used for
H values. When the level in the gas phase is expressed in units of pressure, H is often reported in units of atm-rrr'zmol, or in units of kf'a-rrr'zrnol. Alternatively, some texts
186
Chap. 6
define H to be the dimensionless ratio of the gas concentration to the water concentration.
Thus, care must be taken in the use of H values that are reported in the literature.

Aqueous-solid partitioning at equilibrium in the unsaturated zone is commonly described
using a distribution coefficient Kd approach which assumes linear, instantaneous, and
reversible partitioning between the water and the soil phases. This description is identical to that often used in saturated groundwater transport. As is discussed in detail in
Chapter 8, for organic contaminants, K d is often predicted using
x,
= Kocfoc
(6.4)
where K oe is the water/organic carbon partitioning coefficient and foe is the fraction of
organic carbon in the soils. The level of uncertainty associated with using Eq.(6.4) to
predict Kd values is discussed in Chapter 8.

If the unsaturated zone is very dry, as can occur in soils immediately adjacent to the
ground surface. it may be possible for contaminants to partition from the gaseous phase to
the solid phase in a manner that is different than when liquid water is present. Although
the mechanisms are not currently well understood, the available evidence suggests that
this mode of partitioning can be strong as well as highly non-linear (Chiou and Shoup.
1985). This type of gaseous-solid partitioning can be important only when the relative
humidity in the unsaturated zone is less than about 90%. At humidities of 90% and
greater, the soil grains become completely covered with a very thin film of water. and
the sorption properties of the soil tend towards what is expected when macroscopic
amounts of water are present (Chiou and Shoup, 1985). Because completely dry soils
rarely exist to any significant depth in the unsaturated zone. this type of enhanced sorption
is generally not important for passive vapor migration.

The equilibrium phase partitioning concepts discussed above lead to a retardation factor
that can be defined as (Baehr, 1987)
R
PhKd
= I + -ew -I + -
ell H
eaH
(6.5)
where ea and ew are the volumetric soil-gas and soil-moisture contents, respectively, and
Ph is the bulk density of the porous medium. It should be noted that R is a strong function
of ew since the soil-gas and soil-moisture contents are related through the expression
e = ew + ea, where e is the porosity.
The terms that comprise R describe the relative masses of contaminant that occupy
each of the three phases at any particular point in space. The retardation factor also
Sec. 6.3
187
Gaseous Transport
describes the rate of vapor plume migration relative to a vapor plume that does not
partition to the aqueous or solid phases. This retardation factor is analogous to that used
in groundwater transport modeling, except that the pore water as well as the solids act
as immobile phases. The middle term on the right hand side of Eq.(6.5) represents
partitioning between the gaseous and aqueous phases while the last term represents
partitioning between the aqueous and solid phases. A thin film of soil moisture is assumed
to be present on all the soil grains so that the type of enhanced sorption by the solid
phase discussed in Section 6.2.5 is not important.
6.3 GASEOUS TRANSPORT
6.3.1 General
Although diffusion due to chemical concentration gradients is usually the primary transport mechanism in the unsaturated zone, vapors may also be transported by advection in
that zone. Passive vapor transport can thus be described using the advection-dispersion
equation
~
[()aDij~]
-()aVi~
=
aXi
aXj
aXi
()a
R ac
at
i,j=I,2,3
(6.6)
where i and j represent the three coordinate directions, and t is time. The average linear
gas-phase velocity (Vi) is determined from the Darcy Equation.
The terms in Eq.(6.6) are similar to those for contaminant transport by groundwater
in the saturated zone. The first term on the left represents transport by dispersion (including diffusion), while the second represents transport by bulk gas-phase advection. The
change in concentration over time is described by the term on the right. For simplicity,
no terms have been included for external sources or sinks.
The dispersion coefficient is a second rank tensor that has components describing
both mechanical dispersion and chemical diffusion so that
Di,
= aTlvloij + (aL
v.,
- aT)~
+ Deoij
i, j
1,2,3
(6.7)
where at. and aT are the longitudinal and transverse dispersivities respectively, De is the
effective gaseous diffusion coefficient, t and oij is the Kronecker delta (Bear, 1972).
Concentrations here are in units of mols per unit volume. Often, however, gaseous
concentrations are reported in a wide variety of units that include volume fraction units
(e.g., parts-per-million-by-volume (ppmV) and percent (%, pressure units (e.g., atm and
Pa), or mass concentration units (e.g., mglL and fLglL). Using the Ideal Gas Law, the
temperature, and the molecular weight of the contaminant of interest, conversions can
be made between these various units.
Eq.(6.6) has been written in three-dimensional Cartesian coordinates. Often, however, releases of volatile organic compounds to the subsurface can be confined within a
relatively small lateral area. In such cases, especially if diffusion is the primary transport
tThe effective diffusion coefficient is sometimes also referred to as D*.
188
Chap. 6
mechanism, a two-dimensional axisymmetric representation of the transport equation may

be valid because the vapors will then be migrating in all directions from a centralized
residual source. The axisymmetric formulation of Eq.(6.6) may be written as
i,j=I,2
(6.8)
where the coordinate axes are oriented in the radial and vertical directions and have their
origin at the center of the residual source. The radius from the center of the spill is also
denoted by r.
6.3.2 Diffusion
Transport by diffusion occurs in response to a chemical activity gradient. As written in
Eq.(6.6), it is assumed that the activity gradient can be represented by the concentration
gradient and that diffusion can be described by Fick's Laws. Ideally, Fick's Laws apply
only to scenarios involving isothermal transport in dilute binary systems where the compounds have similar molecular weights. Nevertheless, in practice, Fick's Laws appear to
be reasonable approximations under a wide range of conditions.
In very fine-grained materials and at low absolute pressures, interference with the
pore walls may limit the applicability of Fick' s Laws, and Knudsen diffusion may provide
a more appropriate description (Thorstenson and Pollock, 1989). The Knudsen diffusion
model is not, however, generally used.
The effective gas phase diffusion coefficient in Eq.(6.7) may be described by
De = rD
(6.9)
where r is the tortuosity of the porous medium and D is the free-air diffusion coefficient. The tortuosity, which lies between 0.0 and 1.0, accounts for the increased distance
that contaminant molecules must travel in a porous medium relative to the straight line
distance that is possible in a free gas phase.
Free-air diffusion coefficients are typically several orders of magnitude greater
than aqueous diffusion coefficients. For example, the free-air diffusion coefficient for
TCE and similar chlorinated solvents is slightly less than 10- 5 m2/s at 20C, while the
aqueous phase diffusion coefficient for TCE is on the order of 10- 9 m 2/s at 20C, This
large difference commonly leads to gas phase diffusion dominating over aqueous phase
diffusion as a transport mechanism in the unsaturated zone.
Free-air diffusion coefficients have been measured for a wide range of different
compounds and are tabulated in chemistry reference texts (e.g., Weast, 1987). Alternatively, they can be estimated using empirical relationships (e.g., see, Lyman et al.,
1984). These estimation equations are also useful for determining the dependence of D
on temperature. However, in general, D values increase only slightly with increasing
temperature. Between 10C and 20 DC, the values for D for the common chlorinated
solvents increase by an average of less than 5% (Mendoza and Frind, 1990b).
Sec. 6.3
189
Gaseous Transport
Several empirical estimates have been developed for the tortuosity of porous media.
The most widely used approximation is that usually attributed to Millington and Quirk
(1961)
(6.10)
where 8 is the bulk (effective) porosity of the medium. Similar relationships have been
developed by Currie (1970) and Grable and Siemer (1968), among others. These models
are applicable to sandy materials whose soil-moisture saturation (8 w/8) is less than about
70% (Weeks et al., 1982). At higher water saturations, a continuous air phase may cease to
exist and aqueous diffusion begins to playa relatively more important role in controlling
the diffusion process. Also, these tortuosity relations tend to breakdown in silty and
clayey materials due to the absence of a continuous air phase.
Eq.(6.1O) illustrates the strong dependence of the effective diffusion coefficient on
soil moisture content: as moisture content increases, De decreases dramatically. However,
even at relatively high moisture contents, De may still be several orders of magnitude
greater than effective aqueous diffusion coefficients.
Effective diffusion coefficients can be measured in the laboratory or in the field. In
the laboratory, they are measured by monitoring the migration of gases diffusing through
a column filled with soil with different moisture contents. The data analysis may be
either transient or steady-state. In the field, compounds are allowed to diffuse through
the medium of interest in its natural state (e.g., Kreamer et al., 1988). Numerical or
analytical models are then used to back-calculate the diffusion coefficients. Typically,
transient methods yield effective reactive diffusion coefficients which implicitly include
retardation factors, with the quantity that is being measured directly being De/R. A
separate measurement of R then allows calculation of De.
6.3.3 Advection
Because diffusion in the gas phase can be such an important transport process, it commonly dominates over advection. Nevertheless, advection can be important under some
circumstances because the low viscosity of gases causes them to be highly mobile even
when the pressure or density gradients are very low.
For coarse-grained materials, advective flow in the gas phase may be described
using the gas-phase continuity equation and the Darcy Equation. If density effects are
known to be negligible, a pressure-based formulation may be appropriate:
a [ kij ap]
ap
aXi k r - u. -aXj = 8a f Jat-
i,j=I,2,3
(6.11)
However, in some cases, density gradients may contribute significantly to vapor transport.
One form of the flow equation that includes density is the equivalent fresh-air head
formulation (Mendoza and Frind, 1990a):
-a
aXi
[ krkijPog (ah*
P - Po az )]
+ ----u.
aXj
Po
aXj
= 8aPog fJ -ah*
at
i,j=I,2,3
. (6.12)
190
Chap. 6
In the above equations, P is the pressure, f.t is the gas viscosity, f3 is the gas-phase
compressibility, kij is the intrinsic permeability of the medium, and k, is the relative
gas permeability. The relative permeability, which depends on the soil and its moisture
content, may be described by the Brooks and Corey (1964) relationships, or may be
measured in the laboratory. In the equivalent head formulation, P is the actual density
of the gas phase, Po is the density of pure air, and g is the gravitational constant. The
equivalent fresh-air head is defined by the equation
h*
P
=-+z
Pog
(6.13)
where z is the elevation above some datum. Since the most significant contribution to
variations in density is usually caused by changes in concentration, Eq.(6.12) and the
vapor transport equation (Eq.(6.6)) form a coupled pair of non-linear partial differential
equations. As with the advection-dispersion equation, axisymmetric formulations may
also be written for these equations.
In fine-grained materials, especially under the influence of a strong applied pressure
gradient, the Darcy Equation may no longer strictly apply. This is due to the Klinkenberg
effect where the assumption of no-slip along the pore walls is no longer valid and the
flow of gases will exceed that predicted by the Darcy Equation. However, for the passive
migration of gases in unsaturated, fine-grained media, this effect is rarely of consequence
because the driving forces that commonly arise are not sufficient to cause significant
advection.
As written, Eqs.(6.ll) and (6.12) require that the gas phase be essentially incompressible. Strictly speaking, gases do not satisfy this requirement since their compressibilities are on the order of \0-5 Pa- I , compared to 4.4 x \0-10 Pa- I for water. However,
Massmann (1989) has shown that the assumption of incompressibility is applicable for
pressure differentials up to about 20,000 Pa (0.2 atm). Pressure differences of this magnitude are not expected for passive vapor migration, and so the assumption of an essentially
incompressible gas phase is appropriate for such situations.
Pressure and density gradients that are sufficiently large to cause advective transport in the unsaturated zone may arise from several causes, including density gradients
due to the heavy nature of chlorinated solvent vapors, and pressure gradients due to:
1) barometric pressure variations; 2) vaporization from the liquid product; 3) landfill gas
generation; and 4) the movement of water in the unsaturated zone.
A common characteristic of many chlorinated solvents is that their saturated vapors
are much denser than air. This is because they have high molecular weights (commonly
three to four times greater than the mean molecular weight for air), and they are very
volatile (saturated vapor concentrations of up to several tens of percent of I atmosphere).
As a result, large density gradients may exist near a residual source if the vapor source
approaches saturation. These density gradients may cause downward advection with velocities on the order of meters per day (Mendoza and Frind, 1990b; Johnson et al.,
1992).
The effects of density-induced transport have been investigated in numerical modeling studies by several researchers, including Sleep and Sykes (1989), Falta et al. (1989),
Sec. 6.3
Gaseous Transport
191
Mendoza and Frind (1990a,b) and Mendoza and McAlary (1990). These studies show
that in high permeability environments, density-induced advection can be a significant
transport mechanism that equals or exceeds the effect of diffusion. The threshold permeability above which density-induced advection becomes important for the chlorinated
solvents appears to be on the order of 10- 11 m2 (equivalent to a hydraulic conductivity of 10- 2 cmls). The vapor source must be maintained at a high concentration in
order to maintain high density gradients. Also, the greater the thickness of the unsaturated zone, the greater the advective effects due to density gradients. Temperature can
have a strong effect on the density of the gas phase through its influence on the vapor
pressure.
The atmospheric pressure fluctuates according to global and regional weather
patterns. This means that the pressure along the upper boundary of the unsaturated zone
(i.e., the ground surface) is changing constantly. Although these pressure variations are
generally relatively small, they can exceed 5 kPa (0.05 atm). If the ground surface is open
to the atmosphere, these pressure variations are applied uniformly over the entire boundary and so act in the same manner over the whole domain, either retaining vapors (high
pressure zone) or enhancing vapor removal (low pressure zone). However, this effect
will only be significant if the depth to groundwater is very large. If the ground surface is
restricted by a cover, vapors will be pumped out of, or confined within, the area beneath
the cover as barometric pressure changes act along the edges of the cover. Massmann
and Farrier (1992) showed that the net effect of barometric pumping is to increase the
apparent dispersion of vapors in the unsaturated zone and that the advective effects of
barometric pumping beneath a cover can be significant. Thibodeaux et al. (1982), Nilson
and Lie (1990), and Nilson et al. (1991) considered the effects of barometric pumping
on contaminants located below a fractured impermeable cover. These analyses showed
that the transport of vapors through cracks could be significantly enhanced by barometric
pressure fluctuations.
The mixture of wastes placed in landfills commonly leads to microbial action
that produces large amounts of carbon dioxide and methane gas. In generating this gas,
pressure gradients may be established towards the exterior of the landfill. Although these
gradients are typically low (on the order of a few Palm), they can be large enough
to cause significant bulk movement of gas from the landfill. If volatile organic wastes
are also present in the landfill, their vapors will migrate along with this bulk gas flow
(Metcalfe and Farquhar, 1987).
Even in the absence of biologically-produced gases, vaporization from a residual
source can result in bulk gas migration by advection away from the source (Bird, 1956).
For chlorinated solvents, this aspect of vapor transport was considered by Mendoza and
Frind (1990b) who showed that while it may be of secondary importance in some cases,
transport of volatilized solvents is generally dominated by diffusion and other advective
effects.
Regional movement of the water table may also result in advection in the gas
phase. If the water table drops, air is drawn into the unsaturated zone to fill the newly desaturated pore space. Similarly, if the water table rises, gas is expelled to the atmosphere.
This movement of the gas phase will transport whatever contaminants are in the vapor
192
Chap. 6
phase. Weeks et al. (1982) document a regional case study where long-term declines in
water table elevation have drawn contaminants further into the unsaturated zone than
would have been expected based on diffusion alone.
6.4 AQUEOUS TRANSPORT

Contaminants can be transported in the aqueous phase in the unsaturated zone just as
they can be transported in the gas phase. In fact, some form of transport in the aqueous
phase is necessary for the creation of groundwater contamination from vapor plumes
since vapors themselves cannot enter the saturated zone. Transport in the aqueous phase
may be due to either advection or diffusion. If we consider equilibrium between the
gaseous and aqueous phases, Eq.(6.6) can be modified to include aqueous transport by
writing
i, j
1,2,3
(6.14)
Here V w is the velocity of water due to infiltration. The aqueous dispersion tensor (D wu )
may be written in a form analogous to Eq.(6.7).
By assuming equilibrium between the gaseous and aqueous phases, aqueous concentrations can be written in terms of gaseous concentrations using Henry's Law. Thus,
aqueous concentrations do not appear in Eq.(6.14). If equilibrium between the gaseous
and aqueous phases is not assumed, two equations need to be written, one for the concentration in each of the two phases. Because the physics of aqueous flow and transport
in the unsaturated zone is the topic of many texts (e.g., Hillel, 1980), it is not discussed
in detail here.
6.5 INTERACTIONS WITH THE ATMOSPHERE

The nature of the ground surface may have a great influence over the distribution and fate
of chlorinated solvent vapors in the unsaturated zone. If the ground surface is open to
the atmosphere, vapors will be transported by diffusion from the unsaturated zone to the
atmosphere: having essentially zero concentrations of these compounds, the atmosphere
acts as a sink. Kimball and Lemon (1971) found that wind passing across the surface of
a coarse porous medium resulted in greater losses of heptane vapors from shallow depths
than in the absence of the wind. If the ground surface is covered with a finer-grained
layer or with vegetation, atmospheric losses will be restricted due to the resistance
to flow and diffusion caused by such covers. The ground surface may be covered by
pavement or a building foundation which is impermeable to both gas flow and vapor
diffusion. Cracks or small open sections in such a cover may, however, allow the loss
of significant vapor mass to the atmosphere (Thibodeaux et al., 1982; Nilson et al.,
1991).
Sec. 6.6
Interactions with Groundwater
193
The character of the ground surface also affects the extent of lateral migration
in the unsaturated zone. If diffusive transport is the primary vapor transport mechanism,
an open ground surface results in less lateral transport than beneath a covered ground
surface. This is due to the mass losses through the open ground surface. However, if
density-induced transport is a significant transport mechanism, the opposite effect may
be observed due to higher gas-phase velocities in the absence of a cover (Mendoza and
Frind, I990b).
Obviously, the nature of the ground surface boundary also affects the infiltration
of water. An impermeable cover at the ground surface will prevent infiltration, whereas
an open ground surface will allow infiltratation. In the latter case, water can pass through
contaminated zones, including the residual source area, ultimately allowing contaminated
water to recharge the groundwater zone.
6.6 INTERACTIONS WITH GROUNDWATER
As the interface between the unsaturated and groundwater zones, the capillary fringe is
the boundary through which contaminants must move if contamination in the vadose zone
is to cause groundwater contamination. Conversely, contaminants may migrate from an
established plume in the groundwater zone into the unsaturated zone; this mode of transport is required for the successful mapping of groundwater plumes by vapor sampling.
Because the capillary fringe does not contain a connected gas phase, transport
of contaminants through this zone must occur in the aqueous phase. The three main
mechanisms for contaminants to migrate through the capillary fringe are diffusion and
dispersion, advection, and fluctuations in the elevation of the water table. The plume of
groundwater contamination caused by these processes is expected to be confined to the
regions of the saturated flow regime that are near the water table.
In the unsaturated zone, contaminants can rapidly reach the top of the capillary
fringe by vapor transport processes. At the top of the capillary fringe, the contaminants
will partition into the aqueous phase and an aqueous-phase concentration gradient will
be established into the saturated zone. Such a gradient will result in transport through the
capillary fringe by molecular diffusion or dispersion. However, McCarthy and Johnson
(1992) have shown that vertical dispersion within the capillary fringe is a weak process.
Moreover, aqueous-phase diffusion is typically weak due to the small magnitude of
aqueous-phase diffusion coefficients. Thus, advective transport due to infiltrating water
is likely to be a more significant mechanism for transporting contaminants downwards
through the capillary fringe.
Another important mechanism which can contribute to groundwater contamination
involves the upward and downward movements of the water table in response to changes
in regional infiltration. As noted above, if the water table drops, the unsaturated zone will
be enlarged and contaminant vapors will be drawn deeper into the domain. The vapors
will then be able to partition to the soil moisture in the newly exposed section of the
unsaturated zone. If the water table later rises, this contaminated soil moisture will be
incorporated into the groundwater flow system and a groundwater plume will be created.
194
Chap. 6
6.7 SOil-GAS SURVEYS

6.7.1 General
Soil-gas surveys provide a rapid and inexpensive method of detecting and mapping
vapor concentrations in the unsaturated zone. These surveys can be useful as preliminary
surveys to plan more costly operations such as coring or piezometer installation. Possible
sources of the vapors to be delineated include free-product liquid in the unsaturated
zone, vapor plumes, and aqueous plumes in the saturated zone underlying the site of
interest.
6.7.2 Sampling
In general, vapor sampling methods may be classified as being either active or passive.
Active sampling involves analyzing vapor samples drawn from a temporary or permanent
access tube; passive sampling involves the analysis of vapors collected by a sorbent
detector buried in the subsurface for some period of time. Active sampling appears to
be the preferred method of conducting soil-gas surveys. The reasons for this are that the
vapor concentration values can be obtained directly and rapidly. Consequently, a soil-gas
survey can easily be modified before it is completed. A disadvantage of active sampling
may be that the vapor distribution in the unsaturated zone can be disturbed by sample
withdrawal. Active soil-gas sampling is discussed in detail by Kerfoot (1988), Tillman
et al. (1989), and Hughes et al. (1990), among others.
The primary advantages of passive sampling are that it is less costly than active
sampling and that time-averaged results are obtained. Disadvantages include the fact
that longer periods of time (typically a few weeks) are required to obtain results, the
survey cannot be modified in-progress, and only relative concentrations are obtained.
With regard to the last point, the measured concentrations are not absolute concentrations
in the soil gas; rather, they give an indication of relative amounts of different chemicals
at a particular location and between locations from the same survey. Passive sampling
techniques have been discussed by Voorhees et al. (1984), Malley et al. (1985), and
Kerfoot and Mayer (1986).

In principle, with a properly-conducted soil-gas survey, it is straightforward to delineate a
vapor plume in the unsaturated zone. It is, however, not always as simple to determine
whether free product is also present. Interpretation problems generally arise because
saturated vapor concentrations are rarely observed, even if free product is present. The
primary reasons for this are outlined in Section 6.2.2. In these cases, good judgement,
experience, and alternative indicators must be used to interpret the data.
The idea of detection of a groundwater plume by vapor sampling in the unsaturated zone (see Figure 6.4) has become popular in the past few years, and a number of
successful demonstrations have been reported (e.g., Marrin and Thompson, 1987; Kerfoot, 1987; Marrin and Kerfoot, 1988; Bishop et al., 1990). However, because several
Sec. 6.8
195
A Pair of Field Experiments
gas
/ " 5 ample_-ll-:=,,/;'A_
partitioning
to gas- phase
--=1- contaminant
plume
I-1-f-lt- -
-tnuctuating
ransverse
""":7--. bl
t ro e
di
.
I wcrer
rspersron
,
:-..
..~
advect~
______ transport
Figure 6.4 Conceptual model for processes involved in groundwater plume detection
using soil-gas sampling techniques.
interrelated factors control mass transfer across the capillary fringe, it seems likely that,
at most, soil gas surveys can only provide qualitative information about groundwater
contamination. In particular, because dispersion transverse to flow is a weak process,
only plumes that are immediately adjacent to the water table will be detected by a soil
gas survey. In addition, the presence of any source of vapors (e.g., small undetected
pools or pockets of residual within the unsaturated zone) will greatly complicate the delineation of groundwater plumes. Active infiltration due to precipitation may make such
a delineation impossible.
6.8 A PAIR OF FIELD EXPERIMENTS
A pair of detailed field experiments involving the transport of TCE vapors at the Borden
field site were reported by Hughes (1991) and Hughes et al. (1992). The unsaturated zone
at the site consists of a medium-grained sand with moisture contents at field capacity
and organic carbon contents that are generally low (~0.02%). The upper section of
the unsaturated domain is heterogeneous, with increased organic carbon contents (see
Figure 6.5). The first experiment was conducted in the winter and had a covered ground
surface to limit vapor losses to the atmosphere. The second experiment was conducted in
the summer and had an open ground surface, except for immediately above the residual
source. Variations in temperature greatly affected the experimental results; temperature
profiles are shown in Figure 6.5.
A residual source of TCE was prepared by mixing liquid TCE with excavated native
sand at near residual saturation, and placing it in the ground. Vapor distributions were
obtained using active sampling techniques along an array of sampling probes extending
radially outwards from the source. Between the two experiments, vapor extraction was
used to reduce TCE vapor concentrations to less than 20 ppmV.
196
Chap. 6

Carbon
(%)
Moisture
2.5
0.0
(%)
Temperature
0.0
40.0
0.0
Temperature
30.0
0.0
30.0
0.0
0.0
0.0
0.0
1.0
1.0
1.0
1.0
2.0
2.0
2.0
2.0
3.0
3.0
3.0
3.0
4.0
4.0
4.0
4.0
#1
'-"
-!:
1i
Q)
Cl
Figure 6.5 Physical parameters measured at the Borden vapor transport site: a) soil
organic carbon content; b) soil moisture content; c) temperature distribution for the first
experiment. and d) temperature distribution for the second experiment. (Reprinted with
permission from Mendoza et al., 1990.)
Field observations of the vapor plumes for each experiment after 18 days are shown
in Figure 6.6. In both cases, the plumes migrated a considerable distance in a short period
of time. However, in the second experiment, the overall concentrations were higher and
the leading edge of the plume extended farther from the source zone.
Sensitivity analyses were performed using a numerical model (Mendoza et al.,
1990, 1992). The overall differences in concentration between the two experiments are
primarily attributable to the different temperature regimes in the two experiments: the
first experiment had lower vapor concentrations (by a third) due to lower temperatures in
the source area. In addition, the lower temperatures near the ground surface resulted in
a higher degree of partitioning to the aqueous phase, and so led to increased retardation.
The higher organic carbon contents near the ground surface, in particular a continuous
layer at about 0.8 m depth, were shown to exhibit high retardation factors which led to
slower transport in this region in both experiments, and tended to limit vapor losses to
(0)
.
: 'ttfT';~'\
J
.\:",'~"~~'
-0.5
-1.5
"0
-2.5
0
10.6
5.0
Radius
(m)
3. 5
,/
....., _.....: """':_....-':

5.0
Radius (rn )
(b)
-0.4
-1.4
E
.J:.
-2.4
a.
Q)
-3.4
10.0
Figure 6.6 Measured vapor concentrations after 18 days in the Borden vapor transport experiments:
a) covered experiment; b) uncovered experiment. Contours are logarithms of concentration expressed
in parts-per-million-by-volume (ppmV). (Reprinted with permission from Mendoza et al., 1992.)
Sec. 6.9
Conclusions
197
the atmosphere in the second experiment. This led to the similar plume shapes despite
the different ground surface boundary conditions.
The sensitivity analyses also indicated that density-induced advection played a
small role in the experiments, with the effect being greater in the second experiment where
the source concentrations were higher. The average permeability (1.2 x 10- 11 rrr') of the
sands is in the range where density-induced advection can begin to become a significant
transport process, providing that the vapor density is sufficiently high (Mendoza and
Frind, 1990b).
A thin, but laterally extensive, groundwater plume developed beneath the portion
of the vadose zone used in the two experiments (Rivett and Cherry, 1991). Although TCE
was used for other experiments at the site, it is felt that this contamination is primarily a
result of the vapor migration experiments. Annual water table fluctuations on the order
of I m and precipitation appear to be responsible for the development of this plume.
6.9 CONCLUSIONS
Vapor transport from residual or pooled chlorinated solvents in the vadose zone can have
a significant impact on groundwater quality. The primary reasons for this are that vapor
transport is often very rapid relative to aqueous transport, and that vapors tend to migrate
in all directions from the source. Thus, contamination from a spill of chlorinated solvents
may quickly reach the water table over a wide area and in a manner that is not controlled
by the hydraulic gradients in the saturated zone.
Vapor transport can be described by equations that are similar in form to groundwater transport equations. They are conceptually easy to understand and existing groundwater transport models may be modified to simulate vapor transport. The greatest difference between the behavior of contaminants in the two zones is that diffusion tends to
be the dominant transport process in the unsaturated zone, while advection is commonly
the primary transport mechanism in the saturated zone.
In high-permeability porous media, density-driven advective flow can play an important role in transport if the vapor concentrations are high. Initial velocities on the
order of meters per day are possible, and the result can be high gaseous and aqueous
concentrations in the vicinity of the water table. Advective transport due to pressure
gradients arising from barometric pressure fluctuations may also increase dispersion and
lead to a wider distribution of contaminants.
The rate of vapor transport is strongly dependent on the soil-moisture content of
the unsaturated zone because the effective diffusion coefficient and the retardation factor
are both functions of the soil-moisture content. Temperature variations in the subsurface
can affect vapor concentrations and migration rates through their influence on vapor
pressures and Henry's Law constants (Kerfoot, 1991).
The character of the ground surface often controls the degree of vapor losses
to the atmosphere and the degree of infiltration into the unsaturated zone. Losses to
the atmosphere remove contaminants that might otherwise reach the water table, while
infiltration can transport contaminants to the water table.
198
Chap. 6
While vapor transport mechanisms are responsible for distributing contaminant vapors throughout the unsaturated zone, aqueous-phase transport mechanisms cause groundwater contamination in the saturated zone. The primary aqueous-phase transport processes
involved are advection due to the movement of infiltrating water, diffusion across the
capillary fringe, and regional fluctuations in the elevation of the water table. These transport processes typically result in thin plumes of contaminated groundwater immediately
adjacent to the water table.
ACKNOWLEDGEMENTS
6.10 REFERENCES
Bear, J. (1972) Dynamics of Fluids in Porous Media, Elsevier, New York.
Baehr, A. L. (1987) "Selective transport of hydrocarbons in the unsaturated zone due to aqueous
and vapor phase partitioning", Water Resources Research, 23, 1926-1938.
Bird, R. B. (1956) "Theory of diffusion", in Advances Chem. Eng., T. B. Drew and J. W. Hoopes,
Jf. (Eds.), Academic, San Diego, 155-239.
Bishop, P. K., M. W. Burston, D. N. Lerner, and P. R. Eastwood (1990) "Soil gas surveying of
chlorinated solvents in relation to groundwater pollution studies", Quarterly J. Eng. Geol., 23,
255-265.
Bloes, M. B., K. M. Rathfelder, and D. M. Mackay (1992) "Laboratory studies of vapor extraction
for remediation of contaminated soil". In: Proceedings: Conference on Subsurface Contamination by Immiscible Fluids, K. U. Weyer (Ed.) International Association of Hydrogeologists,
Calgary, April 18-20. Order from A. A. Balkema, Old Post Road, Brookfield, Vermont 05036.
Brooks, R. H. and A. T. Corey (1964) "Hydraulic properties of porous media", Hydrology Papers,
3, Colorado State University, Fort Collins.
Chiou, C. T. and T. D. Shoup (1985) "Soil sorption of organic vapors and effects of humidity on
sorptive mechanism and capacity", Environ. Sci. Technol., 19, 1196-1200.
Currie, J. A. (1970) "Movement of gases in soil respiration", In: Sorption and Transport Processes
in Soils, SCI Monographs, 37, Rothamsted Experimental Station, Harpenden, England.
Dean, J. A. (Ed.) (1985) Lange's Handhook of Chemistry, 13th Edition, McGraw-Hill, New York.
Falta, R. W., I. Javandel, K. Pruess, and P. A. Witherspoon (1989) "Density-driven flow of gas in
the unsaturated zone due to the evaporation of volatile organic compounds", Water Resources
Research, 25, 2159-2169.
Grable, A. R. and E. G. Siemer (1968) "Effects of bulk density, aggregate size, and soil water
suction on oxygen diffusion, redox potentials, and elongation of com roots", Soil Sci. Soc. Amer.
Proc., 32, 180-186.
Sec. 6.10
References
199
Hillel, D. (1980) Fundamentals of Soil Physics, Academic Press, New York.

Hughes, B. M. (1991) Vapour Transport of Trichloroethylene in the Unsaturated Zone: A Field
Experiment, M.Sc. Thesis, Dept. Earth Sciences, University of Waterloo, Waterloo, Ontario.
Hughes, B. M., R. W. Gillham, and C. A. Mendoza (1992) 'Transport of trichloroethylene vapours
in the unsaturated zone: A field experiment". In: Proceedings: Conference on Subsurface Contamination by Immiscible Fluids, K. U. Weyer (Ed.), International Association of Hydrogeologists, Calgary, April 18-20. Order from A. A. Balkema, Old Post Road, Brookfield, Vermont
05036.
Hughes, B. M., R. D. McClellan, and R. W. Gillham (1990) "Application of soil-gas sampling
technology to studies of trichloroethylene vapour transport in the unsaturated zone", In: Groundwater Quality and Analysis at Hazardous Waste Sites, S. Lesage and R. E. Jackson (Eds.), Marcel
Dekker.
Johnson, R. L., K. A. McCarthy, M. Perrott, and C. A. Mendoza (1992) "Density-driven vapor transport: Physical and numerical modeling". In: Proceedings: Conference on Subsurface
Contamination by Immiscible Fluids, K. U. Weyer (Ed.), International Association of Hydrogeologists, Calgary, April 18-20. Order from A.A. Balkema, Old Post Road, Brookfield, Vermont
05036.
Johnson, R. L., C. D. Palmer, and J. F. Keely (1987) "Mass transfer of organics between soil, water
and vapor phases: Implications for monitoring, biodegradation and remediation". In: Proceedings: NWWAIAPI Conference on Petroleum Hydrocarbons and Organic Chemicals in Ground
Water: Prevention, Detection, Restoration, November 17-19, Houston, Texas.
Jury, W. A., D. Russo, G. Streile, and H. EI Abd (1990) "Evaluation of volatilization by organic
chemicals residing below the soil surface", Water Resources Research, 26, 13-20.
Kerfoot, H. B. (1987) "Shallow-probe soil-gas sampling for indication of ground-water contamination by chloroform", Int. J. Environ. Anal. Chem., 30, 167-181.
Kerfoot, H. B. (1988) "Is soil-gas analysis an effective means of tracking contaminant plumes in
ground water? What are the limitations of the technology currently employed?", Ground Water
Monitoring Review, Spring, 54-57.
Kerfoot, H. B. (1991) "Subsurface partitioning of volatile organic compounds: Effects of temperature and pore-water content", Ground Water, 29, 678-684.
Kerfoot, H. B. and C. L. Mayer (1986) "The use of industrial hygiene samplers for soil-gas
surveying", Ground Water Monitoring Review, Fall, 74-78.
Kimball, B. A. and E. R. Lemon (1971) "Air turbulence effects upon soil gas exchange", Soil Sci.
Soc. Amer. Proc., 35, 16-21.
Kreamer, D. K., E. P. Weeks, and G. M. Thompson (1988) "A field technique to measure the
toruosity and sorption-affected porosity for gaseous diffusion of materials in the unsaturated
zone with experimental results from near Barnwell, South Carolina", Water Resources Research,
24, 331-341.
Lyman, W. J., W. F. Reehl, and D. H. Rosenblatt (1984) Handbook of Chemical Property Estimation
Methods, McGraw-Hill, New York, New York.
Mackay, D., W. Y. Shiu, and R. P. Sutherland (1979) "Determination of air-water Henry's constants
for hydrophobic pollutants", Environ. Sci. Technol., 13, 333-337.
Malley, M. J., W. W. Bath, and L. H. Bongers (1985) "A case history: Surface static collection
and analysis of chlorinated hydrocarbons from contaminated ground water", In: Proceedings:
200
Chap. 6
NWWAIAPI Conference on Petroleum Hydrocarbons and Organic Chemicals in Ground Water:

Prevention, Detection, Restoration, November 13-15, Houston, Texas.
Marrin, D. L. and G. M. Thompson (1987) "Gaseous behavior of TeE overlying a contaminated
aquifer", Ground Water, 25, 21-27.
Marrin, D. L. and H. B. Kerfoot (1988) "Soil-gas surveying techniques", Environ. Sci. Techno/.,
22, 740-745.
Massmann, J. W. (1989) "Applying groundwater flow models in vapor extraction system design",
J. Environ. Eng., 115, 129-149.
Massmann, J. W. and D. F. Farrier (1992) "Effects of atmospheric pressures on gas transport in
the vadose zone", Water Resources Research, 28, 777-791.
McClellan, R. D. and R. W. Gillham (1990) "Vapour extraction of trichloroethylene under controlled field conditions". In: Proceedings: Conference on Subsurface Contamination by Immiscible Fluids, International Association of Hydrogeologists, Calgary, April 18-20.
Mendoza, C. A. and E. O. Frind (I 990a) "Advective-dispersive transport of dense organic vapors
Mendoza, C. A. and E. O. Frind (1990b) "Advective-dispersive transport of dense organic vapors
Mendoza, C. A. and T. A. McAlary (1990) "Modeling of groundwater contamination caused by
organic solvent vapors", Ground Water, 28, 199-206.
Mendoza, C. A., E. O. Frind and B. M. Hughes (1990) "Vapour transport of organic compounds:
Simulation of a field experiment and model validation", In: Calibration and Reliability in
Groundwater Modelling, K. Kovar (Ed.) International Association of Hydrological Sciences,
Publication 195, pp. 331-340.
Mendoza, C. A., B. M. Hughes, and E. O. Frind (1992) "Transport of trichloroethylene vapours in
the unsaturated zone: Numerical analysis of a field experiment". In: Proceedings: Conference on
Subsurface Contamination by Immiscible Fluids, K. U. Weyer (Ed.), International Association of
Hydrogeologists, Calgary, April 18-20. Order from A. A. Balkema, Old Post Road, Brookfield,
Vermont 05036.
Metcalfe, D. E. and G. 1. Farquhar (1987) "Modeling gas migration through unsaturated soils from
waste disposal sites", Water, Air, and Soil Pollution, 32, 247-259.
Millington, R. J. and J. M. Quirk (1961) "Permeability of porous solids", Trans. Faraday Soc., 57,
1200-1207.
Nilson, R. H. and K. H. Lie (1990) "Double-porosity modelling of oscillatory gas motion and
contaminant transport in a fractured medium", Int. J. Numerical and Analytical Methods in
Geomechanics, 14, 565-585.
Nilson, R. H., E. W. Peterson, K. H. Lie, N. R. Burkhard, and J. R. Hearst (1991) "Atmospheric
pumping: A mechanism causing vertical transport of contaminated gases through fractured
permeable media", 1. Geophysical Research, 96, 21,933-21,948.
Rivett, M. O. and J. A. Cherry (1991) "The effectiveness of soil gas surveys in delineation of
groundwater contamination: Controlled experiments at the Borden field site". In: Proceedings:
NWWAIAPI Conference on Petroleum Hydrocarbons and Organic Chemicals in Ground Water,
November 20-22, Houston, Texas.
Sec. 6.10
References
201
Sleep, B. E. and J. F. Sykes (1989) "Modeling of transport of volatile organics in variably saturated
media", Water Resources Research, 25, 81-92.
Stephenson, R. M. and S. Malanowski (1987) Handbook of the Thermodynamics of Organic Compounds, Elsevier, New York.
Thibodeaux, L. J., C. Springer, and L. M. Riley (1982) "Models of mechanisms for the vapor
phase emission of hazardous chemicals from landfills", J. Hazardous Materials, 7, 63-74.
Thorstenson, D. C. and D. W. Pollock (1989) "Gas transport in unsaturated porous media: The
adequacy of Fick's Law", Rev. Geophysics, 27, 61-78.
Tillman, N., K. Ranlet and T. J. Meyer (1989) "Soil gas surveys: Part II/Procedures," Pollution
Eng., 21, 79-84.
Voorhees, K. J., J. C. Hickey, and R. W. Klusman (1984) "Analysis of groundwater contamination
by a new surface static trapping/mass spectrometry technique", Anal. Chem., 56, 2602-2604.
Weast, R. C. (Ed.) (1987) Handbook of Chemistry and Physics, 68th Edition, CRC Press, Boca
Raton, Florida.
Weeks, E. P., D. E. Earp, and G. M. Thompson (1982) "Use of atmospheric fluorocarbons F-II
and F-12 to determine the diffusion parameters of the unsaturated zone in the southern high
plains of Texas", Water Resources Research, 18, 1365-1378.
7
Dissolution of Dense
Non-aqueous Phase Liquids
(DNAPLs) in the Subsurface
Stan Feenstra 1 and Nilson Guiguer'
1 Applied
Groundwater Research Ltd.

2550 Argentia Rd.
Canada L5N 5R1
and
Waterloo Centre for Groundwater Research,
Waterloo, Ontario
Canada, N2L 3G1
2Waterloo Hydrogeologic Inc.
200 Candlewood Cr.
Waterloo, Ontario
Canada N2L 5Y9
Page
7.1
7.2
7.3
7.4
7.5
7.6
Introduction
7.3.1.1 General
Conclusions
References
204
204
208
208
208
210
211
212
212
214
215
220
222
223
223
226
228
229
203
204
Dissolution of DNAPLs in the Subsurface
Chap. 7
ABSTRACT
The magnitude of the concentrations of dissolved contaminants in groundwater and the
length of time that associated DNAPL source zones will cause groundwater contamination
are determined by the rate of DNAPL dissolution. In porous media, under typical subsurface flow conditions, laboratory and computer modeling studies indicate that groundwater
that exits residual DNAPL zones does so at saturation with the DNAPL. This is due to
the intimate contact that exists between residual DNAPL and groundwater. In contrast,
studies show that dissolved concentrations derived from pools of DNAPL will typically
be considerably lower than saturation when the concentration is averaged over a vertical
distance of 0.5 or more above the pool. At field sites where DNAPL is present, saturated
dissolved concentrations are not frequently observed in groundwater. This is the result of
the variability in the spatial distribution of the DNAPL residual zones and pools, dilution
in monitoring wells with long intakes, and dispersion processes in the aquifer. For a
multi-component DNAPL, the effective solubility of a given compound will be given
by the product of the pure-phase solubility of that compound with the mole fraction of
the compound in the mixture. During dissolution of such a mixture, the compound(s)
having the highest effective solubility(ies) will be removed preferentially, and both the
DNAPL composition and the dissolved concentrations will change with time. At a site
contaminated with a multi-component DNAPL, an examination of changes in the dissolved concentrations with time may provide a method for determining the degree to
which the DNAPL source mass has been depleted by dissolution.
7.1 INTRODUCTION
The rate of dissolution of a dense non-aqueous phase liquid (DNAPL) from the source
zone at a contaminated site will influence the magnitude of the dissolved chemical concentrations in the groundwater as well as the length of time that the DNAPL zone
will persist. An understanding of DNAPL dissolution rates is therefore required when
evaluating existing sources of dissolved chemical plumes with respect to future plume
migration, as well as the effectiveness of pumping to remove dissolved plumes. This
chapter describes conceptual models for the dissolution of DNAPLs in the subsurface.
The models consider: I) the solubility of organic compounds in water; 2) the theory of
mass transfer from DNAPL phases into water; and 3) the dynamic dissolution behavior
of DNAPLs as observed in laboratory experiments. Understanding the implications of
these conceptual models will allow assessments of what concentration data for dissolved
contaminants at a site might indicate regarding: I) the possible presence of a DNAPL
phase in the subsurface at the site; 2) the degree of mass depletion of such a DNAPL
source by dissolution; and 3) the nature of the groundwater transport of chemicals away
from such a DNAPL source.
7.2 SOLUBILITY FROM PURE DNAPLS AND FROM DNAPL MIXTURES

The mass dissolution rate of a DNAPL in the subsurface is a function of the solubility(ies) of the compound(s) in the DNAPL, the groundwater velocity, the mass and
Sec. 7.2
205
distribution of the DNAPL material in the subsurface, the pore distribution of the
medium, the aqueous-phase diffusion coefficient(s) of the compound(s) in question, and
the effects of other chemical constituents in the system. A DNAPL can be comprised
of a single compound, or it can be a mixture. In their pure forms, chlorinated organic
solvents are liquids which exhibit a wide range of solubilities in water. These water solubilities are not very temperature dependent over the ambient groundwater temperature
range.
The available solubility data for chlorinated solvent compounds have been obtained
in studies conducted for a variety of purposes, over a span of several decades, using
reagents of varying purity, and using a variety of analytical techniques. The result is that
there is disagreement in the published solubility data for these compounds. For example,
the solubility values for trichloroethylene (TCE) reported in Horvath (1982) for 2025C vary from 700 to 1470 mg/L. For some other compounds, the range in reported
values is even larger. This situation has made it difficult for practicing hydrogeologists
to determine what value is best when assessing the dissolution processes for a given
chlorinated solvent compound. In general, it can be assumed that the most reliable values
are those that have been measured recently using good analytical methods (see Table 7.1
and Appendix I).
TABLE 7.1 Measured Solubilities (mg/L) of Four Chlorinated Solvent Compounds at
23 to 25 -c.
Compound
Broholm and Feenstra

(1995)
(23-24C)
Horvath (1982)
(25C)
Mackay and
Shiu (1981)
(25 DC)
Chloroform
Carbon Tetrachloride
I, I, I-Trichloroethane
Trichloroethy lene
8700
780
1250
1400
7920
793
1495
1100
7925
1000
720
1050
For an organic NAPL mixture, the aqueous-phase concentration of each component

that is in equilibrium with the mixture can be approximated using a solubility analog of
Raoult's Law for vapor pressure. This analog is
(7.1)
where C.wl,m is the aqueous solubility of component m from the mixture (also sometimes
referred to as the "effective solubility" from the mixture), X m is the mole fraction of
component m of interest in the NAPL mixture, and C~al is the solubility of the pure
compound.
Laboratory experimental studies (e.g., Banerjee, 1984, and Broholm and Feenstra,
1995) suggest that Eq.(7.1) is a reasonable approximation for mixtures of structurallysimilar hydrophobic organic liquids, such as mixtures of chlorinated solvents. Broholm and Feenstra (1995) conducted experiments with mixtures of chloroform, carbon
206
Chap. 7
tetrachloride, I, I, l-trichloroethane (l, I, 1-TCA), trichloroethylene (TCE), and tetrachloroethylene (PCE). They concluded that Eq.(7.l) works well when predicting the effective aqueous solubilities of binary mixtures of these compounds. An example of the
results obtained in that study is shown in Figure 7.1. Laboratory studies (e.g., Leinonen
and Mackay, 1973) have shown that use of Eq.(7.1) may cause some error with complex
mixtures of certain structurally-dissimilar compounds (e.g., alkanes dissolved in aromatics). However, the error is usually smaller than a factor of 2. Such deviations from ideal
behavior are small enough that they may usually be safely ignored in environmental
studies where there are often much larger uncertainties t.
1.0
<D
f/)
al
s:
0.8
0f/)
::l
0
<D
0.6
::l
CT
-c
<D
.S;
E
0.4
of
.s
0.2
...
0.0
0.0
0.2
0.4
0.6
0.8
1.0
peE Mole Fraction in the Organic Phase

Figure 7.1
PCE.
Solubility as a function of mole fraction for mixtures of l,l,l-TCA and
The DNAPLs encountered at waste disposal sites are commonly mixtures of chemicals that are liquids in their pure forms at ambient temperatures. However, DNAPLs
at waste sites also sometimes contain dissolved compounds that are solids in their
pure forms at ambient temperatures, e.g., some chlorinated benzenes, polychlorinated
biphenyls (PCBs), and polycyclic aromatic hydrocarbons (PAHs) (see Table 7.2). For such
a compound, its "sub-cooled" liquid solubility (i.e., also referred to as its "super-cooled"
liquid solubility) must be used as C~at in Eq.(7.I). For relatively rigid molecules of the
type that are of interest here, the sub-cooled liquid solubility (Csat.liq) at a given temperature (T, K) of interest can be estimated based on the solid-phase solubility (Csat.sol)
tExamples of exceptions to this rule include: I) an alcohol dissolved in a largely alkane (non-polar)
mixture (as occurs with alcohol additives in gasoline); 2) an alkane (non-polar) dissolved in one or more
alcohols; and 3) a chlorinated solvent dissolved in an alcohol.
Sec. 7.2
207
and the melting point (Tm , K) of the compound of interest according to (Shiu et al.,
1988)
Cwt,liq
= Csat,so! exp[6.8(Tm/T
- 1)]
(7.2)
Sub-cooled liquid solubilities can be considerably higher than the corresponding solidphase solubilities, and the difference increases as Tm increases (see Table 7.2).
A comparison of observed groundwater concentrations with calculated effective
solubility values can be carried out based on data from a Superfund site in Pennsylvania
where DNAPL has been found in numerous wells in a fractured sandstone aquifer (ERM,
1987). Two samples of DNAPL recovered from wells were analysed and found to contain
the same compounds, though at different concentrations (see Table 7.3). This type of
variability is likely to be common at many multi-component DNAPL sites. Because the
DNAPL composition varies at the Pennsylvania site, the dissolved concentrations in the
groundwater would be expected to vary to a similar degree. For each DNAPL sample,
the effective solubilities of the four principal components were calculated and are given
in Table 7.3 along with data for the groundwater. For groundwater collected close to the
DNAPL source, the dissolved concentrations are comparable to the calculated effective
solubilities, and follow the expected concentration trend of 1,2,3-trichloropropane >
xylenes > toluene> ethylbenzene.
As implied in the heading for Table 7.3, a significant fraction of the organic DNAPL
phase at the Pennsylvania site could not be identified or quantified. This type of situation
has been encountered at other sites. However, when estimating the mole fractions of
the compounds of interest in a given DNAPL sample, one requires an estimate of the
weight percent of the unidentified fraction as well as an estimate of the average molecular weight (MW) of the compounds making up the unidentified fraction. The type of
uncertainty which arises when making such assumptions is indicated in Table 7.4 for
one of the two DNAPL samples considered in Table 7.3. Increasing the estimate of the
average MW of the unidentified fraction causes the estimated effective solubilities for the
known contaminants to increase: for a given weight % of unidentified fraction, a higher
TABLE 7.2 Melting Points, Solid-Phase Solubilities at 25 DC, and Estimated Sub-Cooled
Liquid Solubilities of Selected Compounds in Water at 25C.
PAH Compounds
Melting
point
(0C)
p-Dichlorobenzene
Naphthalene
Fluorene
Anthracene
Benzo[a]pyrene
Biphenyl
2,2' ,5,5' -Tetrachlorobiphenyl
Decachlorobiphenyl
53
80
117
217
179
71
87
306
(Tm , K)
Solid-phase
solubility
(J1.glL, 25C)
Estimated
sub-cooled
liquid solub.
(J1.glL, 25 DC)
326
353
390
510
452
344
360
579
79,000
31,000
1,900
730
3.8
7,000
103
0.0012
150,000
110,000
16,000
92,000
130
20,000
423
0.73
Melting
point
208
Chap. 7
TABLE 7.3 Pure Compound Properties, Weight Percent Composition Values for
Two DNAPL Samples from a Superfund Site in Pennsylvania, Calculated Effective
Solubility Values for those Compositions, and Observed Dissolved Groundwater
Concentrations at the Site. (The average molecular weight (MVV) of the unidentified
portions of the two DNAPL samples was assumed to be 150 g/rnol.)
Property
1,2,3Trichloropropane
Pure Compound Solubility (mg/L)

Molecular Weight (g/mol)
1,900
146
Toluene
580
92
Weight %, DNAPL Sample I

Effective Solubility (mg/L)
23
400
4.2
35
Weight %, DNAPL Sample 2

Effective Solubility (mg/L)
73
1,400
Observed Groundwater Cone.

Lower Value (mg/L)
Upper Value (mg/L)
220
1,200
Ethyl
benzene
Xylenes
200
106
190
106
17
43
3.8
9.1
0.9
8.1
5.8
16
0.9
2.3
2.1
53
7.3
74
1.1
12
TABLE 7.4 Calculated Effective Solubilities at 25C for Components of DNAPL Sample 1 from
a Superfund Site in Pennsylvania for Different Assumed Molecular Weight (MVV) Values for the
Unidentified Fraction.
Effective solubility (mg/L)

C.~)af
Compound
Wt.%
MW
(g/mol)
Toluene
Xylenes
Ethyl Benzene
23
146
1900
4.2
17
3.8
92
106
106
580
200
190
(mg/L)
Assumed MW of unidentified fraction

MW
100
325
29
33
7.4
MW
200
MW
450
520
40
46
10
46
53
12
300
average MW results in a lower calculated mole fraction for the unidentified fraction, and
correspondingly higher mole fractions for the identified contaminants.
7.3 DISSOLUTION PROCESSES
7.3.1.1 General
In the subsurface, DNAPL will occur in residual zones, in layers above an underlying
aquitard, and in pools on top of an underlying aquitard. In residual zones, DNAPL is
present as immobile blobs and ganglia that may occupy 10% or less of the pore space.
Sec. 7.3
209
Groundwater flow occurs through the remaining pore space, although the presence of
the DNAPL does reduce the relative permeability of the medium to water. Dissolution
of the DNAPL occurs as groundwater flows through the residual zone. Within layers or
pools of free-phase DNAPL, the DNAPL may occupy as much as 50 to 70% of the pore
space and may substantially reduce groundwater flow through these zones. Dissolution
of the DNAPL then occurs predominantly along the tops, lateral margins of the layers
and pools. Flow may also occur beneath layers of DNAPL.
The rate of dissolution of DNAPL in the subsurface will depend on the effective
solubilities and diffusivities of the DNAPL components, the physical distribution of the
DNAPL in the porous or fractured medium (e.g., the contact area between the DNAPL
and the groundwater), and the rate of groundwater flow through and around the DNAPL.
These fundamental relationships are illustrated on a sub-pore scale and a pore scale in
Figure 7.2. At the interface between the DNAPL and the groundwater, the concentration
in the groundwater approaches the solubility or effective solubility of each of the components. Dissolved chemicals are transferred into the flowing groundwater by molecular
diffusion.
It is necessary to express the rate of dissolution mathematically in order to quantify
the dissolution process. The rate of mass transfer from a DNAPL to the water phase is
typically expressed as a function of a mass transfer coefficient, a driving force, and
the interfacial contact area between the two phases of interest so that
Rate of
Mass Transfer
Mass Transfer
Coefficent
(MIT)
(LIT)
Concentration
Difference
Contact
Area
(7.3)
The driving force for mass transfer is the concentration difference across the mass transfer
boundary layer, and is usually defined as the difference between the effective solubility
of the component and the dissolved concentration in the bulk water flowing past the
DISSOLUTION: PORE-SCALE
DISSOLUTION: SUB-PORE SCALE
Residual
DNAPL
Groundwater Flow
Diffusion
+ t +C
- - - - - - 1..
~
sat
DNAPL
Sand Grains
Figure 7.2
scale.
Schematic representation of dissolution processes on a sub-pore and pore
210
Chap. 7
DNAPL. The rate of mass transfer will determine the dissolved-phase concentrations
that will occur in the groundwater as well as the persistence of the DNAPL zone in the
subsurface. The mass transfer coefficient is a parameter determined largely by the water
flow conditions. Several models have been used in the field of chemical engineering to
describe mass transfer coefficients for the dissolution of an immiscible phase. The three
most common models are the stagnant film model, the film penetration model, and the
random surface renewal model (Cussler, 1984).
In the stagnant film model, all the resistance to mass transfer is assumed to reside in
a stagnant layer of thickness 8 adjacent to the interface, as shown in Figure 7.3. The
concentration at the interface is taken to be equal to Cat: it is assumed that the rate of
equilibration across the fluid-fluid interface is rapid, and that the slower rate of molecular
diffusion in the water phase coupled with advective flow determines the overall rate of
mass transfer (Levich, 1962; Turitto, 1975). The aqueous-phase concentration of the same
species in the bulk water is represented as CWo Burris and MacIntyre (1986) measured
the dissolution of a four-component liquid hydrocarbon layer overlying a stirred aqueous
phase. Their results suggest that chemical equilibrium likely existed at the interface in
that system.
For I-D diffusion, the flux of a species is given by Fick 's First Law as
ac
ax
(7.4)
N=-D-
where: N = flux of the species (units of M / L 2T); D = diffusion coefficient of species

A (units of L 2/T); and ac/ax = concentration gradient of the species in the water
near the interface (units of M L -3 / L). In the stagnant film model, it is assumed that the
concentration gradient is linear over the boundary layer thickness 8 as shown by the
dashed line in Figure 7.3 so that
(7.5)
ax
and the particular form of Equation (7.4) becomes
(7.6)
wate,
~..-
Stagnant
Film
'Co
()
DNAPL
~Ca
Figure 7.3 Stagnant film model for

dissolution mass transfer.
Sec. 7.3
211
where K; = D /8 is the mass transfer coefficient. Although the stagnant film model
assumes that the concentration decreases linearly over the distance 8, the concentration
dependence may in fact be more like that given by the solid line in Figure 7.3.
The stagnant film model predicts that the mass transfer coefficient is directly proportional to the diffusion coefficient. Hence, doubling the diffusion coefficient should
double the rate of mass transfer. This mass transfer coefficient also depends on variables
such as the fluid viscosity and local water velocity (both of which depend strongly on
the heterogeneity of the porous medium): the effects of these variables are lumped into
the film thickness 8.
In porous media, it is useful to define a mass transfer rate per volume of porous
medium N* by multiplying the mass flux by the ratio of DNAPL surface contact area
An to the unit volume of porous medium V so that
(7.7)
where A has dimensions of liT and is defined as the lumped mass transfer coefficient
such that A = KcAn/V. The use of such a lumped mass transfer coefficient in porous
media is a common practice (Miller et al., 1990; Sleep and Sykes, 1989; Zilliox et al.,
1978).
If a multi-component DNAPL is present, the mass transfer rate for each individual
component within the DNAPL mixture can be calculated according to
(7.8)
where Am is the lumped mass transfer coefficient for the component m and C w . m is the
concentration of component m in the bulk aqueous phase. Csa t . m is the effective solubility
of component m which can be calculated using Eq.(7.l).

This model, proposed by Higbie (1935), was the earliest attempt to find an alternative
for the stagnant film model. Higbie (1935) felt that it was unreasonable to assume that
a liquid film could remain static, and not mix at all with the adjacent turbulent bulk
liquid. Higbie (1935) pictured a small fluid element moving from the bulk phase (in this
case the water phase) to the interface where it remains for a time period t , after which
it returns to the bulk phase. During the time t, transient diffusion is assumed to take
place between the interface and the element so that
s, =
rv
V;(
(7.9)
Although it has been suggested that a square root relationship between K; and D may
be more reasonable than the direct proportionality suggested by the stagnant film model,
the characteristic penetration time t is not known for situations of practical interest.
212
Chap. 7
The random surface renewal model was proposed by Dankwerts (1951) who basically
adopted Higbie's (1935) approach, but instead of assuming that all the elements have the
same contact time t ; assigned a random distribution to t. The mass transfer coefficient
then becomes:
(7.10)
where s is the rate of surface renewal.

The three mass transfer models discussed above all relate K; for a compound
to its diffusion coefficient in water and to an empirical parameter (i.e., the stagnant
layer thickness, the contact time, or the rate of surface renewal) which in each case is
a function of the flow conditions and fluid properties. The mass transfer coefficient is
therefore usually treated as a fitting parameter in models used to examine the results of
lab or field experiments. This is particularly true in groundwater systems where one must
also contend with other complicating factors such as the distribution of residual DNAPL
in the pores, the effective contact area, and the degree of aquifer heterogeneity. Thus,
although such models provide the basis for the description of mass transfer coefficients, at
the present time there is no evidence that one particular model works better than another
when predicting DNAPL dissolution in groundwater. Since the stagnant film model is
the simplest of the available mass transfer models, it has been the model used in most
practical applications.
In the chemical engineering literature, experimental mass transfer data are seldom
reported directly. Rather, the results are expressed as correlations equations between
different dimensionless numbers. These numbers are useful because they allow the prediction of mass transfer coefficients under a wider range of conditions than would otherwise
be possible. The dimensionless numbers commonly used in mass transfer correlations
are given in Table 7.5 (Cussler, 1984; Wakao and Kaguei, 1982; Sherwood et al., 1975).
The Sherwood and Stanton numbers involve the mass transfer coefficient. The Schmidt
and Lewis numbers involve different kinds of diffusion. The Reynolds, Grashof, and
Peeler numbers involve flow considerations. The Damkohler number involves diffusion
and chemical reaction.
A key point about each of the numbers in Table 7.5 is that its exact definition implies
a specific physical system. For example, the characteristic length l in the Sherwood
number (Sh) could be the stagnant layer thickness for a liquid-to-liquid transfer, or the
average grain size of a porous medium (Powers et al., 1992; Miller et al., 1990). Some
correlations involving the mass transfer coefficient (expressed in terms of Sh) that have
been used in previous works involving NAPL dissolution are given in Table 7.6 (Powers
et al., 1991; Miller et al., 1990; Powers et al., 1992; Geller and Hunt, 1993). In that
table, Sh = Kcl/D everywhere except for the study of Miller et al. (1990) for which
Sh = Ad~/ D where A is the lumped mass transfer coefficient and d p is the mean diameter
of the sand grains.
Sec. 7.3
213
TABLE 7.5
Dimensionless Numbers Relating to Dissolution of DNAPLs in the Subsurface.
Group
Physical meaning
Used in
mass transfer rate
Usual dependent
variable
Sherwood number
Sh=
~l
dijfusion rate
Stanton number
St =
mass transfer rate
~c
flow rate
Schmidt number
Sc =
dijfusivity of momentum
dijfusivity of mass
15
Lewis number
Le =
dijfusivity of energy
dijfusivity of mass
T5
Occasional dependent
variable
Correlations of gas and

liquid data
Simultaneous heat and
mass transfer
Reynolds number
Re -l.Y.
- v
inertial forces
Grashof number
G _ 13 gl'1 p/ p
r-
or
viscous forces
v2
flow velocity
buoyancy forces
viscous forces
Peelet number
Pe=
Damkohler number
kl 2
w=V
Forced convection
momentum velocity
flow velocity
diffusion rate
reaction rate
diffusion rate
Free convection
Correlations of gas
or liquid data
Correlation involving
reactions
D = diffusion coefficient [L2jT]

g = acceleration due to gravity [LjT2]
= mass transfer coefficient [LjT]

= characteristic length [L]
v = fluid velocity [LjT]
K;
a = thermal diffusivity [L2/T]
= first-order reaction rate constant [ljT]

v = kinematic viscosity [L2jT]
I'1p/p
= fractional density change
The published correlations confirm that the mass transfer coefficient is a function of
the flow conditions and the diffusion coefficient. Usually, once a correlation equation is
fitted to laboratory data, the predictions of the mass transfer coefficient are very reliable
(Cussler, 1984). For example, Miller et al. (1990) and Powers et al. (1992) were able to
simulate their experimental results using correlations given in Table 7.6. However, this
214
TABLE 7.6

A Selection of Mass Transfer Correlations Previously Used in NAPL Dissolution Studies.
Dissolution of:
oil in porous media
Basic equation
Sh = 0.55
Characteristic length
+ 0.25P e 1.5
solid spheres
in fluidized beds
Sh = 1.09(IiS w ) - 1Pe0
solid spheres in
packed beds
flow around single
spheres
TCE in sand column
33
Sh = O.978Pe0 33(IiSw ) - O.33

(Pe > 10) or Sh = 2 (Pe < I)
Sh
37.2Reo. 656
toluene in sand
column
Ii
= porosity;
Chap. 7
= water
saturation; S;
Reference
sand grain diameter
Pfannkuch (1984)
mean sphere diameter
Wakao & Kaguei

(1982)
mean sphere diameter
Wilson & Geankoplis

(1966)
sphere diameter
Bowman et al. (1961)
mean sand grain

diameter
Powers et al. (1992)
mean sand grain

diameter
Miller et al. (1990)
NAPL saturation.
type of success can only be expected within the range of experimental conditions (flow,
compound type, etc.) used in the experiments underlying the correlations. To date, there
is no evidence that these correlations can be made even more general.
Under most circumstances, the difficulty in predicting mass transfer coefficients may
not be a large problem when evaluating DNAPL dissolution rates into groundwater.
Indeed, theoretical studies and the review of laboratory dissolution experiments by
Pfannkuch (1984) suggest that the applicable mass transfer coefficients may often be
sufficiently large to result in saturated aqueous concentrations in DNAPL zones at
typical residual contents and typical groundwater velocities. (At high groundwater velocities, several laboratory studies have found that mass transfer coefficients can become
dependent on groundwater velocity resulting in a reduction in the dissolved concentrations.)
Laboratory batch experiments indicate that for water in contact with a NAPL, the
dissolved concentrations often approach effective solubility values in minutes to hours,
depending on the NAPLIwater volume ratio (Shiu et al., 1988; Poulsen et al., 1992).
Laboratory column and tank experiments (Schwille 1988; Anderson et al., 1992; Imhoff
et al., 1994) suggest that for groundwater flow at up to I m/day (linear velocity) through
DNAPL zones at typical residual contents, saturated concentrations will be attained after
contact times of ~ 30 min. Under such conditions, DNAPL dissolution is not limited
by the rate of interphase mass transfer. (Theoretical and laboratory studies have shown,
however, that dissolution may become mass transfer rate limited at high groundwater
velocities ( > 1 m/day) as can be found in the immediate vicinity of pumping wells
(Pfannkuch, 1984; Geller and Hunt, 1993)).
Sec. 7.3
215
In situations where the DNAPL is present in layers and pools, theoretical and
laboratory experiments suggest that less-than-saturated dissolved concentrations are
often observed (Schwille, 1988; Geller and Hunt, 1993). Within a DNAPL layer or pool,
the DNAPL may occupy a large proportion of the pore space and substantially reduce the
groundwater flow through such a zone. In addition, horizontal layers or pools present a
very small cross-sectional area of DNAPL zone to horizontal groundwater flow. Thus,
dissolution of DNAPL from a layer on a low permeability lens or from a pool occurs
predominantly along the margins of the lens or pool. Consequently, the rate of dissolution
may be much lower than for a residual zone.
A laboratory experiment was performed by Schwille (1988) to illustrate the dissolution of a TCE pool measuring 1.5 m in length and 0.5 m in width (Figure 7.4).
The pool was formed in a tank containing medium-grained sand. Water flow through the
tank passed over the TCE pool at different velocities and the dissolved TCE concentrations were measured in the tank effluent. The results of the experiment are given in
Table 7.7. The dissolved concentrations were found to be well below saturation levels,
even when dissolved concentrations are averaged over a short vertical distance above the
pool.
VOLUME AVERAGED
DISSOLVED
CONCENTRATION
WATER_..........
FLOW
..
- - - - - 1 5 0 em -----t~
Figure 7.4
TCE pool experiment of Schwille (1988).
TABLE 7.7 Results of TeE Pool Dissolution Experiment

by Schwille (1988).
Linear velocity
Concentration
(mid)
(mg/L)
Percent saturation
2.3
1.1
2.3
4.5
6.8
67
6.1%
90
87
8.2%
7.9%
73
77
6.6%
7.0%

Over the past three decades, many models have been developed that describe some
aspect of multi-phase transport (see Abriola, 1989; Kueper and Frind, 1991; and Guiguer,
216
Chap. 7
1993). Several of these models address the dissolution of NAPLs in the saturated zone.
Whether it incorporates NAPL movement or not, such a model is usually known as a
compositional model, since at some point the model utilizes mass balance equations in
some form for each species m in each phase (soil, gas, water, or DNAPL). A species
can be defined as a specific chemical component that is present in one or more phases,
or can be considered as a group of such components if average characteristics can be
defined. In the case of dissolution of DNAPLs by percolating water, the mass balance
equation for the water phase can be written as:
a
at (Su" C w.m)
change in mass
in dissolved
phase
a
a(SWVi, Cw,m)
Xi
~
aXi
w.
(SwD;j aC m )
aXj
Nm
0
(7.11)
adjective
transport
dispersive
transport
dissolution
where: Vi = average pore water velocity; Sw = average fraction of pore volume occupied
by water; C w . m = concentration of component m in water; Dij = dispersion coefficient;
Nm = mass transfer of component m from DNAPL to water. In a multi-component
model, the number of equations required depends on the number of chemical components
of interest. A solution results not only in a quantification of the remaining residual
saturation and its composition, but also in the delineation of the contaminant plume of
each component at any point in space and time.
The mass transfer term N m is generally treated in one of two ways: 1) local equilibrium is assumed between the phases, i.e., the concentration of component m is always
equal to its effective solubility; or 2) as a first order reaction given by
(7.12)
Abriola and Pinder (1985a, 1985b), Baehr and Corapcioglu (1987), and Kaluarachchi and Parker (1990) present 2-D models that solve for the migration of NAPL
as an immiscible phase, while the dissolution process is taken into account using the
local equilibrium assumption. Guiguer (1993) presents a 3-D model for the migration
of NAPL where the dissolution process is considered as a first order reaction. Guiguer
(1993) used that model to examine the effects of heterogeneities on the evolution of
contaminant plumes emanating from residual zones. Powers et al. (1992) and Miller et
al. (1990) developed I-D models for a single component DNAPL trapped as immobile
residual. Both studies assumed first order mass transfer; the mass transfer coefficients
used were obtained as a function of dimensionless numbers determined in lab column
experiments. Brusseau (1992) developed a I-D solution for a single component NAPL.
That model utilized a two-domain concept where mass transfer was assumed to be essentially instantaneous for some portion of the NAPL zone and rate limited with first
order mass transfer for the remainder.
The mass transfer coefficient Am expresses the process of diffusive mass transfer
from the surfaces of residual blobs to free-flowing groundwater. This parameter depends on the characteristics of the component of interest (through its aqueous phase
Sec. 7.3
217
diffusion coefficient), and on the nature of the porous medium. However, as dissolution progresses, Am can change due to a changing geometry of the residual zone, e.g., a
lengthening of the diffusion pathways from residual trapped in deadend pores. Guiguer
and Frind (1994) incorporated this time-varying behavior by expressing Am in a general
way as a nonlinear function of the DNAPL saturation so that Am = A~ Sn fJ where the
exponent f3 is a lumped parameter that accounts for the residual distribution and the
soil characteristics (e.g., the relative abundance of deadend pore zones). A property of
the porous medium that would be expected to affect f3 is the pore size distribution of
the medium. As a practical matter, f3 must be considered to be a fitting or calibrating
parameter.
Analytical solutions for the dissolution of DNAPL residual in a I-D column were
developed by Hunt et al. (1988). They idealized the distribution of the residual DNAPL
as spherical droplets of uniform size and showed that DNAPL residuals can persist for
long periods releasing low concentrations because the mass transfer rate can decrease
with time. Hunt et al. (1988) also developed an analytical solution for dissolution of
DNAPL pools, considering that the dissolution occurred only along the edges of the
pool and that the location of the interface did not change with time. The vertical profile
of dissolved concentrations above the downgradient edge of a pool can then be expressed
as (Johnson and Pankow, 1992)
C(z)
= Csaterfc (
Z 05)
.
(7.13)
2 (DyL)
where z is the vertical distance above the pool, D; = (tv V + De is the dispersion coefficient in the vertical direction, (tv is the vertical dispersivity, v is the linear groundwater
velocity, and De is the effective molecular diffusion coefficient in the porous medium. An
example of the vertical concentration profile above a 25 m long pool in a sandy aquifer
is given in Figure 7.5. The dissolved concentrations are expressed as a percentage of the
saturated concentration. Saturated concentrations are found only very close to the top of
the pool.
1.0
Typical Sandy Aquifer

v - 0.5 mId
U v - 2.3E-4 m
De - 2.5E-10 m2/s
Concentration profile
above pool
~ 0.5
Qi
Groundwater
Flow
..
0.0
+-_--+---:====1:==--.
20
40
60
80
% Saturation
Figure 7.5 Example of dissolved concentration profile above a DNAPL pool.
100
218
Chap. 7
The distance that dissolved concentrations extend upward from a pool will be
greater for longer pool lengths, lower groundwater velocities, and larger coefficients
of vertical dispersion. Concentration profiles above a pool in a typical sandy aquifer for
various pool lengths are shown in Figure 7.6. In this case, the linear groundwater velocity
was taken to be 0.5 mid, and a v was taken to be 0.00023 m. For these conditions, at the
downgradient edge of the pool, the concentration profiles extend less than I m upward
from the pool, even for pool lengths as great as 100 m. This result illustrates that the
vertical concentration profile above a pool at the down gradient edge of the pool does not
depend strongly on pool length.
Concentration profiles above a pool (at the down gradient edge) in a sandy aquifer
are given in Figure 7.7 for a range of groundwater velocities. The pool length was
taken to be 25 m, and a v was assumed equal to 0.00023 m. At a groundwater velocity of 0.001 mid, the concentration profile extends more than 2 m upward from
the pool. However, at groundwater velocities 0.1 and I mid, the concentration profile extends less than 0.5 m upward from the pool; the water is swept away before the
contaminant can diffuse very far in the upward direction. These calculations suggest
that the vertical concentration profile above a pool will be very thin for groundwater
velocities in the range of 0.1 to 1 mid, i.e., for the range that is typical for sandy
aquifers.
Calculated concentration profiles above a pool (at the downgradient edge) in a
sandy aquifer for values of a v in the range 0.00023 to 0.1 m are given in Figure 7.8 for a
pool length of 25 m and a linear groundwater velocity of 0.5 mid. Measurements of a v for
1.0
0.9
0.8
g
"8a.
Pool Length
0.7
---5m
0.6
Q)
.,s
.0
0.5
Q)
.. 100m
g 0.4
C5 0.3
-,
0.2
0.1
<,
-,
------
0.0
---
-, -
t------
20
40
- -- 60
80
100
% Saturation
Figure 7.6 Dissolved concentration profiles at the down gradient edge of DNAPL pools
of varying length when the linear groundwater velocity = 0.5 m1day and a v = 0.00023 m.
Sec. 7.3
219
2.0
1.8
1.6
I
'0
..s
...
0
1.4
n,
1.2
.0
1.0
Linear Velocity (mid)
\
\
-.
o
c 0.8
- - - 0.001 mid
- - - - - 0.01 mid
<,
-. -.
1n
is 0.6
<,
"-
0.4
<,
~
~
~
0.2
- -
.;
-.....:.::...:
0.0
20
40
-- - 0.1 mid
-- 1 mid
<,
--- .>.
;;..,----=----~
60
80
100
% Saturation
Figure 7.7 Dissolved concentration profiles at the down gradient edge of a 25 m long
DNAPL pool for varying groundwater velocities when a v = 0.00023.
2.0
1.8
1.6
1.4
'8a.
1.2
VerticalDispersivity
- - - - 2.3E-4 m
Ql
>
.8
s'"
Q)
1.0
-----1.0E-3m
- - -- -- - - - 1.0E-2 m
0.8
i5 0.6
\
0.4
-,
<,
0.2 I~
0.0
-- - - - ---r-----1--
20
40
- 60
80
100
%Saluration
Figure 7.8 Dissolved concentration profiles at the downgradient edge of a 25 m long
DNAPL pool for varying a v values when groundwater velocity = 0.5 m1day.
sands derived from small-scale laboratory experiments suggest a value of ~0.00023 m

(Johnson and Pankow, 1992). Large-scale field experiments such as those at the Borden
test site (Rivett et al., 1994), the Twin Lakes site (Moltyaner and Wills, 1991) and at
Cape Cod (Garabedian et al., 1991) suggest a v values in the range 0.001 to 0.01 m.
Therefore, in any given situation, there will be considerable uncertainty in the value of
220
Chap. 7
au' Some of the vertical dispersion observed in large-scale tests is likely due to the effects
of macro-scale heterogeneities and temporal fluctuations in groundwater flow direction
that may not be relevant to the local process of dissolution from a pool. For the conditions
in Figure 7.8, when au = 0.01 m, the concentration profile at the downgradient edge
extends more than 2 m upward from the pool; when au = 0.00023 and 0.001 m, the
concentration profile extends about 0.5 m upward from the pool.
The distance that significant dissolved concentrations extend upward from a pool
into an aquifer will have a very large influence on the magnitude of the dissolved concentrations that are observed in monitoring wells in an aquifer. Wells with a short intake
interval (e.g., ~ 1.5 m) that are situated at the same elevation as a pool may exhibit
dissolved concentrations that are 10% or more of the saturated concentration. However, wells with a longer intake interval that reach the same elevation as the pool may
exhibit dissolved concentrations of less than 1% of saturation. Wells with an intake
interval that does not reach close to the elevation of a pool may yield no measurable dissolved concentrations at all. The implications of these considerations for the
diagnosis of the presence of DNAPL pools in the subsurface is discussed in Chapter 13.
Based on the results of laboratory and modeling studies of dissolution, it is possible to

make estimates of the possible longevity of DNAPL sources in residual and pool form
in the subsurface. For a residual source, consider a zone of residual DNAPL occupying
1 m3 of a sandy aquifer and containing 30 Llm 3 of dichlorornethane, TCE, or PCE.
Groundwater is assumed to flow through this zone at a linear velocity of 0.14 m/day,
reflecting a typical sandy aquifer with a hydraulic conductivity of 10- 2 cm/s, a hydraulic
gradient of 0.005, and a porosity of 0.3. If it is assumed that the dissolved concentration
in the residual zone is at saturation and that this level is maintained until the source
is completely dissolved, then the estimated minimum source-lives are: dichlorornethane,
46 days; TCE, 2 years; and PCE, 13 years.
The above estimates are absolute minima, and will be considerably longer for several reasons: I) the DNAPL within the residual zone may be heterogeneously distributed
such that saturated dissolved concentrations may not be produced throughout the 1 m3
residual zone; 2) the presence of DNAPL residual will reduce the permeability of the
residual zone to groundwater flow, and this will reduce the mass flux out of the residual
zone; and 3) saturated concentrations will not be maintained as the residual mass is depleted. Laboratory studies by Lamarche (1991) clearly illustrate (see Figure 7.9) that dissolved concentrations will decline as the mass of residual DNAPL is depleted. They also
illustrate that low but significant dissolved concentrations can persist for very long times.
Source-life calculations can also be performed for pools. For the TCE pool in the laboratory experiment of Schwille (1988), the pool was 1.5 m long, 0.5 m wide, and 0.2 m thick.
For the mass transfer rate that can be extrapolated from the Schwille (1988) data when
the groundwater velocity is 0.75 m/day (see Johnson and Pankow, 1992), the estimated
source-life for such a pool is 50 to 100 years. These calculations suggest that zones of
Sec. 7.3
221
10000
1000
'\
-. <.;
100
10
TeE (mg/L)
0.1
0.01
0.001
250
500
750
1,000
1,250
1,500
Pore Volumes
Figure 7.9
TeE concentration vs. time for a column experiment by Lamarche (1991).
residual DNAPL and especially pool DNAPL can persist in the subsurface and contribute
to groundwater contamination for decades to centuries. These estimates are consistent
with the observation of on-going groundwater contamination problems at waste disposal
facilities and chemical spill sites which have existed since the 1940s, 1950s, and 1960s.
As noted above, as residual DNAPL dissolves into groundwater, the volume of
DNAPL per volume of porous medium decreases, and the DNAPL interfacial area per
volume of porous media also decreases. This decreases the value of A. Powers et at. (1991)
and Geller and Hunt (1993) used correlations between the volume of DNAPL remaining
in the pores and the surface area in order to calibrate their lab column experiments. Very
simplistic geometrical shapes of the DNAPL blobs, such as spheres and cylinders, had to
be assumed. A more realistic idealization of the dynamic behavior of the mass transfer
coefficient inside a representative elementary volume (REV) is illustrated in Figure 7.10.
Figure 7.10
Dissolution processes at the pore scale.
222
Chap. 7
In any given case, residual DNAPL will be located in pores with different geometries.
When the dissolution process begins, some droplets will be in direct contact with the
flowing water, some will be in partial contact with the flowing water, and some will
be trapped in dead end pores with a stagnant layer of water separating them from the
flowing water. Mass transfer from the droplets in direct contact with the flowing water
will be dissolution-controlled (i.e., occur at saturation) and will proceed more quickly.
Mass transfer between the droplets in the dead end pores and the flowing water will
be diffusion-controlled and will depend on the thickness of the layer of stagnant water.
As the overall dissolution proceeds, the droplets in contact with the flowing water will
become completely dissolved, and the only droplets remaining will be those trapped in
the dead end pores. As the size of these droplets decreases, the thickness of the stagnant
water layer will increase, and the dissolution rate of the remaining droplets will decline
further. The concentrations of dissolved DNAPL in the groundwater leaving the DNAPL
zone may then become substantially lower than saturation, but remain non-zero for long
periods of time. Overall, it can be expected that as long as DNAPL residual is present,
some mass transfer will be dissolution-controlled and some will be diffusion-controlled.
As long as there is a diffusion component, the overall process will be naturally asymptotic.
At the same time, we note that the decrease in the volume of the trapped DNAPL will
open up some new flow paths, so that some blobs may become accessed by the flowing
groundwater.
Finally, we note that the presence of heterogeneity in a porous medium can slow
the dissolution rate of a DNAPL in that medium. Figure 7.11 illustrates a possible scenario. When the DNAPL moves downward through the saturated zone, it will first tend to
invade the larger pores. Depending on the forces involved and the pore interconnections,
it may also penetrate layers of lower permeability. Since more water will flow through the
higher permeability zones, the residual DNAPL in these zones will be dissolved first. As
less pore space is occupied by the DNAPL in these zones, the permeability will become
even larger, and more water will be diverted away from the lower permeability zones.
With less flow passing through the low permeability zones, the mass transfer rate there
will decrease, and the residual DNAPL there will persist longer. When dilution by clean
water passing through the system and into monitoring wells is considered, the overall
result can be aquifer concentrations which are substantially lower than the solubility limit.

Saturated dissolved concentrations are seldom observed in groundwater except in the
immediate vicinity of large DNAPL sources. In the past, concentrations below saturation have commonly been interpreted to mean that DNAPL is not present. However,
considerations such as those presented in the preceding sections indicate that such interpretations can be incorrect. Indeed, concentrations below saturation may be the result of:
1) reduced effective solubilities due to the DNAPL being multi-component in nature;
2) the heterogeneous distribution of the DNAPL as residuals, layers, or pools; 3) dispersion in the aquifer; and 4) dilution in monitoring wells having long intakes lengths.
As a "rule-of-thumb", the finding of dissolved concentrations that exceed 1 % of the
Sec. 7.4
223
contaminated zone
\
residual DNAPL zone

DNAPLpool
low permeability layer . - - ,,
Ir----==
flowlines "
mass transfer from
..r-...- DNAPL to water
-----
.,
(
t.
:'::';:::::;'::::::;:::;::':::;:::/;~
1,i1:. ::~::f;@::;;::;::~:tr~f
--Figure 7.11
----
.>.
'T"Y """77
2.2""<>
................
;"-:;:;t:~.4~~~~~ttifl:
Effects of lower permeability zones on dissolution.
effective solubility should probably be cause for serious consideration of the presence
of a DNAPL phase in the subsurface. Such a finding should stimulate the collection and
evaluation of other lines of evidence to help confirm or disprove the presence of DNAPL
in the subsurface at the site of interest.
7.4 MULTI-COMPONENT DNAPLS

The preferential removal of the most soluble components from synthetic mixtures of
aromatic hydrocarbons, mixtures of chlorobenzenes, diesel fuel, and crude oil has been
examined in laboratory studies by Mackay et al. (1991). The results indicate that for a
multi-component DNAPL, the compounds having the highest effective solubilities will
be removed most quickly. The preferential removal of the most soluble compounds will
change the composition of the DNAPL, thereby changing the effective solubilities of the
remaining compounds.
224
Chap. 7
Changes in the dissolved concentrations of constituent compounds can be predicted

based on the effective solubilities of the compounds by simulating the incremental depletion of the DNAPL phase of interest (Shiu et al., 1988; Mackay et al., 1991). This
concept is referred to here as the Effective Solubility Model (ESM). Laboratory experiments on the dissolution of organic mixtures (Shiu et al., 1988) found that the dissolved
concentration changes predicted using the ESM compared favorably with the concentration changes that were observed (see Figure 7.12). These experiments were conducted at
water flow rates low enough to yield saturated dissolved concentrations.
The results of a laboratory experiment conducted by Geller and Hunt (1993) can be
used to show that the ESM can also simulate the relative changes in dissolved concentrations in situations when water velocities are high and dissolution is limited by the rate
of mass transfer. In that experiment, a 50/50 mixture of benzene and toluene was used.
(Although this mixture was not a DNAPL, the dissolution processes for such a mixture
will be very similar to that expected for a DNAPL.) A known volume of the mixture
was injected into the center of a column packed with 40-45 mesh glass beads. Water
flowed through the column at a Darcy velocity of 10 mJday (linear velocity = 25 mJday).
At this high flow rate, the initial dissolved concentrations in the column effluent were
approximately 55 mg/L for benzene and 14 mg/L for toluene (see Figure 7.13). These
values are well below the effective solubility saturation values of 900 and 250 mg/L,
respectively. The low concentrations were due to mass transfer limitations resulting
from the high water flow rates and from the limited extent of the residual zone.
250
0
200
:::;0>
.s
150
to
.~
0
0
'E
Q)
g 100
0
o
0
0
50
o
o
0
0
5.000
10.000
15.000
20,000
Water/NAPL Volume Ratio
Figure 7.12 Results of dissolution experiment using a mixture of chlorobenzenes. Key:

chlorobenzene, open squares; 1,2,4-trichlorobenzene, open diamonds; 1,2,3,5-tetrachlorobenzene, open triangles. The data are fitted according to predictions of the Effective
Solubility Model (ESM). Data from Mackay et al. (1991).
Sec. 7.4
225
60
,'.
50
40
......... Benzene
mglL30
- - - - Toluene
20
10
O+---I-----+---+----+---=~
20
40
60
80
100
Pore Volumes
Figure 7.13 Changes in benzene and toluene concentrations as benzene is removed

preferentially from a benzene/toluene mixture. Data from Geller and Hunt (1989).
As the benzene/toluene mixture dissolved, the dissolved concentrations changed

as the more soluble benzene was removed preferentially. The results of the experiment
are shown in Figure 7.14 as a plot of the toluene/benzene ratio in the water with the
fraction of the source remaining in the column. The observed changes compare well
with the changes predicted using the ESM. This is an expected result, even for high
groundwater velocities. Indeed, all of the models used to predict mass transfer coefficients suggest that for a given set of water flow conditions, the mass transfer coefficients
for different chemical components should depend principally on the aqueous diffusivities
10.0
0/
0/
...
Q)
c:
Q)
0/
c: 1.0
Q)
-'r'
Q)
.,....
c:
Q)
:l
(5
r-
t-
<>
~
v
j..--;
0.1
1.00
0.90
0.80
0.70
0.60
0.50
0.40
0.30
0.20
Fraction Source Remaining
Figure 7.14 Change in toluene/benzene ratio in water during dissolution. Open diamonds give experimental results. The data are fitted according to predictions of the
Effective Solubility Model (ESM). Data from Geller and Hunt (1989).
226
Chap. 7
of the compounds of interest. Because the aqueous diffusivities of most organic chemicals
of interest span a relatively small range (see Table 7.8), the mass transfer coefficients
for the components of a multi-component DNAPL will remain similar one-to-another.
Consequently, for a given groundwater flow regime and DNAPL residual zone, the
relative rates of mass transfer of the components of a multi-component DNAPL, and
therefore the concentration ratios should depend primarily on the effective solubilities
of the components. The relative concentrations of dissolved constituents may therefore
be useful in evaluating the dissolution history of subsurface multi-component DNAPL
sources.
TABLE 7.8 Aqueous Diffusivity Values for Selected Organic Compounds. (Benzene data
from Bonolli and Witherspoon (1968); correlation data for other compounds from Wilke
and Chang (1955) as described in Perry's Chemical Engineers Handbook (1984).)
20C
Compound
Toluene
Xylenes
Benzene
Ethylbenzene
Chlorobenzene
Trichloroethylene
Perch1oroethylene
Dichloromethane
Chloroform
1, I, l-Trichloroethane
MW (g/mol)
92.1
106.2
78.1
Diffusion coefficient
cm 2 /s
Diffusion coefficient
cm 2/s
6.7 x 10- 6
6.2 x 10- 6
7.4 x 10- 6
9.0 x 10- 6
8.4 x 10- 6
1.0 x 10- 5
8.3 x 10- 6
x 10- 6
106.2
6.1
112.6
6.3 x
7.4 x
6.8 x
9.1 x
153.8
10- 6
10- 6
10- 6
10- 6
7.9 x 10- 6
7.1 x 10- 6
133.4
7.0 x 10- 6
131.4
165.8
84.9
119.4
8.5 x
1.0 x
9.3 x
1.2 x
1.1 x
9.5 x
9.4 x
10- 6
10- 5
10- 6
10- 5
10- 5
10- 6
10- 6

During the 1980s, groundwater purging operations were implemented at many waste
disposal and chemical spill sites. When these purging operations began, the objective of
most was aquifer restoration. Experience has now shown that clean-up of groundwater
contamination by organic chemicals is often not possible when the goal is background
conditions, or even satisfaction of drinking water criteria. At many sites, elevated dissolved chemical concentrations continue in the groundwater even after years of purging.
This is most frequently the case near suspected source areas. We now understand that
these cleanup failures have often been due to the presence of DNAPL sources, either in
residual zones, or in pools.
Groundwater purging operations conducted within or close to a DNAPL source
area will, of course, accelerate depletion of the source. However, the actual degree of
depletion usually cannot be determined because the mass of DNAPL in the source zone
is not known. Nevertheless, for a multi-component source, a consideration of the relative
Sec. 7.4
227
concentrations of the dissolved contaminants may provide a method for evaluating the
degree of depletion. The application of this method can be illustrated using the results
of monitoring conducted during a groundwater purging operation at an industrial plant
site in the eastern United States (Feenstra, 1992).
The site of interest is situated on the Atlantic Coastal Plain and is underlain by
a sand and gravel aquifer approximately 15 m thick. In 1977, a dissolved plume of
1,1,1-TCA and PCE was identified in the shallow aquifer. The principal source of the
contaminants is believed to have been the chemical transfer/storage area of the plant.
In 1978, a groundwater purging system was implemented and operated until late 1984.
During that time, up to 14 purge wells were operated throughout the plume area as well as
close to the suspected source area. With the exception of the wells immediately adjacent to
the source area, the dissolved 1,1,1-TCA and PCE concentrations in the groundwater were
reduced from as high as several thousand jLg/Lto below the clean-up criteria of 50 jLg/L;
in some wells, the levels were reduced to below detection limits. In purge wells close to
the source area, although the dissolved concentrations were reduced significantly during
purging, they remained relatively high (see Figure 7.15). Approximately 60% of the total
chemical mass removed by purging was removed from two wells close to the source.
The dissolved concentrations in the purge wells close to the source area were much
lower than saturation, but still relatively high, suggesting that DNAPL remains in the
subsurface in the source area. The dilution relative to saturation likely occurred due to
the heterogeneous distribution of the DNAPL material, and the fact that the purge wells
were screened over the full thickness of the aquifer. During the period of groundwater
purging, the PCEIl, 1,1-TCA ratio increased in the purge wells close to the suspected
source. This trend was not observed in purge wells far from the source. The change in
the PCEIl,1,I-TCA ratio is shown in Figure 7.16. Since 1,1,I-TCA is more soluble than
PCE (see Table 7.1), this increase is consistent its preferential dissolution from the source
DNAPL.
Based on the initial PCEIl,1,1-TCA ratio of ~ 0.5 in the groundwater, the initial
composition of the DNAPL is estimated to have been 88% PCE and 12% I,I,I-TCA.
Based on this value, the ESM can be used to predict subsequent changes in the PCEIl, 1,1TCA ratio as the DNAPL mass is depleted (see Figure 7.17). Although the PCEIl,I,ITCA ratio in the groundwater shows fluctuations (see Figure 7.16), it is clear that the
4,000
3,000
....J
~ 2,000
1,000
O+-+-+--+--+--+--+--+--+--+--+--=F-...:r
78-2
80-2
82-2
84-2
Figure 7.15 Dissolved concentrations

in purge well GW-32 close to suspected
source zone during pumping. Key:
l,l,l-TCA, open squares; PCE, solid
squares.
228
Chap. 7
10.0
.I
.Q
iii
a:
-c
o
1.0
t-;-
o
a...
.-.
f7T'
UJ
0.1
78-2
I
80-2
I I
82-2
I
84-2
Year
Figure 7.16 Change in PCEIl,I,I-TCA

ratio in water in purge well GW-32 close to
suspected source zone during pumping.
10.0
./
.Q
iii
./'
a:
(3
I-
.-""
1.0
,/
W
oa...
0.1
0.9
0.8
0.7
Fraction Source Remaining
0.6
Figure 7.17 Predicted change in

PCEIl,I,I-TCA ratio using the Effective
Solubility Model (ESM) for dissolution of a
DNAPL with an initial composition of 88%
PCE and 12% l,l,l-TCA.
ratio increased during the pumping from ~0.5 up to ~5. For such an increase, the ESM
implies that ~ 30% of the DNAPL phase was removed during the pumping.
7.5 CONCLUSIONS
The rate of dissolution of DNAPL to groundwater is commonly expressed as a function

of a mass transfer coefficient and the contact area between the DNAPL and groundwater.
A universal relationship between the mass transfer coefficient and groundwater velocity has not been established. Likewise, the typical dependencies of the dissolution rate
on the contact area and DNAPL residual content have not been established. However,
these uncertainties can be largely set aside by consideration of empirical observations
of DNAPL dissolution from laboratory experimental studies. These observations suggest that the groundwater exiting DNAPL residual zones acquires saturated dissolved
concentrations when the residual zones are at least several centimeters in length, and
groundwater velocities are typical of natural groundwater conditions ( < I m1day). For
groundwater passing over DNAPL pools, studies (e.g., Johnson and Pankow, 1992) indicate that concentrations that are considerably lower than saturation will be developed
for typical pool lengths and sampling intervals over which the groundwater to be analyzed is acquired. For most chlorinated solvents, the rate of dissolution of pools will
Sec. 7.6
References
229
be sufficiently slow that the DNAPL zones will cause significant contamination of the
groundwater for centuries or more (Johnson and Pankow, 1992).
Saturated dissolved concentrations are not frequently observed in groundwater except in the immediate vicinity of large DNAPL sources. In the past, concentrations below
saturation have commonly been interpreted to mean that DNAPL is not present at a given
site. In many cases, evidence of other types now indicate that such interpretations have
often been incorrect. Concentrations below saturation may be a result of: 1) a multicomponent rather than a single component DNAPL; 2) the heterogeneous distribution of
DNAPL as residuals and layers or pools; 3) dilution in monitoring wells having long
intakes lengths; and 4) dispersion in the aquifer. The importance of these various factors
will be site specific. In most hydrogeologic situations, the finding of dissolved concentrations exceeding 1% of the effective solubility should be cause for serious consideration
of the presence of DNAPL.
For multi-component DNAPLs, laboratory studies and field examples suggest that
the effective solubility of a given constituent compound can be estimated as the product
of the mole fraction of that compound in the DNAPL with its pure-phase solubility. The
chemicals having the highest effective solubilities will be dissolved preferentially from
the DNAPL. This will result in changes in the composition of the DNAPL, which in
turn will change the effective solubilities. For a given DNAPL composition, it may be
possible to relate the changes in the ratios of the dissolved concentrations derived from
the DNAPL to the degree of mass depletion of the DNAPL source.
ACKNOWLEDGEMENTS
7.6 REFERENCES
Abriola, L. M. and G. F. Pinder (l985a) "A multiphase approach to the modelling of porous media
contamination by organic compounds. 1. Equation development", Water Resources Research,
21, 11-18.
Abriola, L. M. and G. F. Pinder (1985b) "A multiphase approach to the modelling of porous media
contamination by organic compounds. 2. Numerical simulation", Water Resources Research, 21,
19-28.
Abriola, L. M. (1989) "Modeling multiphase migration of organic chemicals in groundwater systems - A review and assessment", Environ. Health Perspectives, 83, 117-143.
Anderson, M. R., R. L. Johnson, and J. F. Pankow (1992) "Dissolution of dense chlorinated solvents
into groundwater: 3. Modeling contaminant plumes from fingers and pools of solvent", Environ.
Sci. Techno!., 26, 901-908.
230
Chap. 7
Baehr, A. L. and M. Y. Corapcioglu (1987) "A compositional multiphase model for groundwater
contamination by petroleum products. 2. Numerical simulation", Water Resources Research, 23,
201-203.
Banerjee, S. (1984) "Solubility of organic mixtures in water", Environ. Sci. Technol., 18, 587-591.
Bonolli, L. and P. A. Witherspoon (1968) "Diffusion of aromatic and cycloparaffin compounds in
water at 20C to 60 GC", J. Phys. Chem., 72, 2532-2534.
Bowman, C. W., D. M. Ward, A. I. Johnson, and O. Trass (1961) "Mass transfer from fluid and
solid spheres", Canadian J. Chem. Eng., 39, 9-13.
Broholm, K., and S. Feenstra (1995) "Laboratory measurements of the aqueous solubility of mixtures of chlorinated solvents", Environmental Toxicology and Chemistry, 14,9-15.
Brusseau, M. L. (1992) "Rate-limited mass transfer and transport of organic solutes in porous
media that contain immobile immiscible organic liquid", Water Resources Research, 28, 33-45.
Burris, D. R. and W. G. MacIntyre (1986) "Solution of hydrocarbons in a hydrocarbon-water system
with changing phase composition due to evaporation", Environ. Sci. Technol., 20, 296-299.
Cussler, E. L. (1984) Diffusion, Mass Transfer in Fluid Systems, Cambridge University Press, New
York, New York.
Danckwerts, P. W. (1951) "Significance of liquid-film coefficients in gas absorption", Ind. Eng.
Chem., Process Design Development, 43, 1460-1467.
ERM (1987) Off-site Operable Unit Remedial Investigation Report - Volume l . Tyson's Site, Montgomery County, Pennsylvania. Submitted to U.S. EPA, Region III, Philadelphia on behalf of
Ciba-Geigy Corporation, Ardsley, New York. Prepared by Environmental Resources Management, Inc., Exton, Pennsylvania.
Feenstra, S. (1992) "Evaluation of multi-component DNAPL source by monitoring of dissolved
phase concentrations". In: Proceedings: international Con! on Subsurface Contamination by Immiscible Fluids, K. U. Weyer (Ed.), International Association of Hydrologists, Calgary, Alberta,
April 18-20, 1990. A. A. Balkema, Rotterdam, pp. 65-72.
Garabedian, S. P., D. R. LeBlanc, L. W. Gelhar, and M. A. Celia (1991) "Large-scale natural
gradient tracer test in sand and gravel, Cape Cod, Massachusetts. 2. Analysis of spatial moments
for a nonreactive tracer", Water Resources Research, 27, 911-924.
Geller, J. T. and 1. R. Hunt (1993) "Mass transfer for nonaqueous phase organic liquids in watersaturated porous media", Water Resources Research, 29, 833-846.
Guiguer, N. (1993) Dissolution and Mass Transfer Processes for Residual Organics in the Saturated
Groundwater Zone: Numerical Modelling. Ph.D. Thesis, Dept. of Earth Sciences, University of
Waterloo, Waterloo, Ontario.
Guiguer, N. and E. O. Frind (1994) "Dissolution and mass transfer processes for residual organics
in the saturated groundwater zone". In: Proceedings: International Symposium on Transport and
Reactive Processes in Aquifers, International Association for Hydraulic Research. April 11-15,
ETH Zurich.
Higbie, R. (1935) "The rate of absorption of a pure gas into a still liquid during short periods of
exposure", Trans. Amer. Inst. Chem. Eng., 31, 365-385.
Horvath, A. L. (1982) Halogenated Hydrocarbons: Solubility-Miscibility with Water, Marcel-Dekker, Inc., New York, New York.
Hunt, J. R., N. Sitar and K. S. Udell (1988) "Nonaqueous phase liquid transport and cleanup. I.
Analysis of mechanisms", Water Resources Research, 24, 1247-1259.
Sec. 7.6
References
231
Imhoff, P. T., P. R. Jaffe and G. F. Pinder (1994) "An experimental study of complete dissolution of
a nonaqueous phase liquid in saturated porous media", Water Resources Research, 30, 307-320.
Kaluarachchi, J. J. and J. C. Parker (1990) "Modeling multicomponent organic chemical transport
in three-fluid phase porous media", J. Contam. Hydrology,S, 349-374.
Kueper, B. H. and E. O. Frind (1991) "Two-phase flow in heterogeneous porous media. I. Model
Lamarche, P. (1991) Dissolution of Immiscible Organics in Porous Media, Ph.D. Thesis, Dept. of
Earth Sciences, University of Waterloo, Waterloo, Ontario.
Leinonen, P. J. and D. Mackay (1973) "The multicomponent solubility of hydrocarbons in water",
Canadian J. Chem. Eng., 51, 230-233.
Levich, V. G. (1962) Physicochemical Hydrodynamics, Prentice Hall, New Jersey.
Mackay, D. M., W. Y. Shiu (1981) "A critical review of Henry's Law constants for chemicals of
environmental interest", J. Phys. Chem. Ref Data, 10, 1175-1199.
Mackay, D. M., W. Y. Shiu, A. Maijanen and S. Feenstra (1991) "Dissolution of non-aqueous
phase liquids in groundwater", J. Contam. Hydrology, 8, 23-42.
Miller, C. T., M. M. Poirier-McNeill and A. S. Mayer (1990) "Dissolution of trapped nonaqueous
phase liquids: Mass transfer characteristics", Water Resources Research, 26, 2783-2796.
Moltyaner, G. L. and C. A. Wills (1991) "Local- and plume-scale dispersion in the Twin Lake 40and 260-m natural-gradient tracer tests", Water Resources Research, 27, 2007-2026.
Perry, R. H. (1984) Perry's Handbook of Chemical Engineering, Sixth Edition. McGraw-Hili, New
York, New York.
Pfannkuch, H. O. (1984) "Determination of the contaminant source strength from mass exchange
processes at the petroleum-groundwater interface in shallow aquifer systems". In: Proceedings:
Conference on Petroleum Hydrocarbons and Organic Chemicals in Groundwater, National Water Well Association/American Petroleum Institute, Nov. 5-7, Houston, Texas, pp. 111-129.
Powers, S. E., C. E. Loureiro, L. M. Abriola, and W. J. Weber, Jr. (1991) "Theoretical study of the
significance of nonequilibrium dissolution of nonaqueous phase liquids in subsurface systems",
Water Resources Research, 27, 463-477.
Powers, S. E., L. M. Abriola and W. J. Weber, Jr. (1992) "Development of phenomenological
models for NAPL dissolution processes". In: Proceedings: Subsurface Restoration Conference,
June 21-24, Dallas, Texas.
Poulsen, M. M., L. Lemon, and J. F. Barker (1992) "Dissolution of monoaromatic hydrocarbons
into groundwater from gasoline-oxygenate mixtures", Environ. Sci. Technol., 26, 2483-2489.
Rivett, M. 0., S. Feenstra, and J. A. Cherry (1994) "Comparison of Borden natural gradient
tracer tests". In: Proceedings: IAHRIAIRII Symposium on Transport and Reactive Processes in
Aquifers, Dracos, Th. & Stauffer (Eds.) A. A. Balkema, Rotterdam, pp. 283-288.
Schwille, F. (1988) Dense Chlorinated Solvents in Porous and Fractured Media - Model Experiments, Translated by J. F. Pankow, Lewis Publishers, Boca Raton, Florida.
Sherwood, T. K., R. L. Pigford, and C. R. Wilke (1975) Mass Transfer, McGraw-Hili, New York,
New York.
232
Chap. 7
Shiu, W. Y., A. Maijanen, A. L. Y. Ng, and D. Mackay (1988) "Preparation of aqueous solutions
of sparingly soluble organic substances: II. Multicomponent systems - Hydrocarbon mixtures
and petroleum products", Environ. Toxicol. Chem., 7, 125-137.
Sleep, B. E. and Sykes, J. F. (1989) "Modeling the transport of volatile organics in variably
saturated media", Water Resources Research, 25, 81-92.
Turitto, V. T. (1975) "Mass transfer in annulli under conditions of laminar flow", Chem. Eng. Sci.,
30, 503-509.
Wakao, N. and S. Kaguei (1982) Heat and Mass Transfer in Packed Beds, Gordon and Breach,
New York, New York.
Wilke, C. R. and P. Chang (1955) "Correlation of diffusion coefficients in dilute solutions", American Inst. Chem. Eng. J., 1, 264-270.
Wilson, E. 1. and C. J. Geankoplis (1966) "Liquid mass transfer at very low Reynolds numbers",
Ind. Eng. Chem. Fund., 23, 9-14.
Zilliox, L., P. Muntzer, and 1. 1. Menanteau (1978) "Probleme de l'Echange entre un Produit
Petrolier Immobile et l'Eau en Movement dans un Millieu Poreux", Revue de l'Institut Francais
du Petrole, XXVIII, 185-200.
8
Sorption of Dissolved
Chlorinated Solvents to
Aquifer Materials
Richelle M. Allen-King", Robert W. Gillham", and Douglas M. Mackay?
1 Department of Geology
Washington State University
Pullman, Washington, USA 99164-2812

Waterloo, Ontario, Canada N2L 3Gl
Page
8.1
8.2
8.3
8.4
8.5
8.6
8.7
8.8
8.9
Introduction
Sorption Mechanisms: Effects of Sorbate and Sorbent on Soil/Water Partitioning
8.3.1 Definitions
8.4.1 General
8.4.3 Large-Scale Methods
8.5.3 Contribution of Spatial Variability in the Field to Non-Ideal Sorption Behavior
8.6.1 General
8.6.3 Effects of Co-Solvents, Dissolved Organic Matter, Suspended Colloids, and
Surfactants
8.6.3.1 Co-solvents
8.6.3.2 Dissolved organic matter (DOM) and suspended colloidal material
8.6.3.3 Surfactants
Conclusions
References
234
235
239
239
243
247
248
248
248
248
249
250
250
250
252
253
254
254
254
255
255
255
256
257
258
260
233
234
Sorption of Dissolved Chlorinated Solvents to Aquifer Materials
Chap. 8
ABSTRACT
The sorption of dissolved chlorinated solvent compounds to aquifer materials reduces

the rate of contaminant transport in aquifers, increases the pumping required to flush
compounds out of an aquifer relative to non-sorbing compounds, and can affect transformation rates. The degree of sorption in a given circumstance depends on the chemical
properties of the given compound and also on the nature of the aquifer solids. Sorption
increases as hydrophobicity increases, as can be predicted based on the compound's water
solubility and its octanol/water partition coefficient. The role of the aquifer solids can be
predicted using the fraction of organic carbon (foe) of the solids. Such predictions do not
account for: 1) sorption to minerals (which can be important in low foe aquifers); 2) variations in the sorptivity of the organic carbon in the aquifer solids; or 3) non-equilibrium
or non-linear behavior. Variations in the natural geochemistry of groundwaters general1y
have a minimal effect on sorption. However, sorption can be decreased when surfactants,
co-solvents, and/or large concentrations of organic carbon are present in the groundwater.
8.1 INTRODUCTION
Sorption is the general term used to describe the uptake by soil of dissolved or gaseous
species without reference to a specific mechanism (Chiou, 1989). The movement of dissolved organic pol1utants in the subsurface is retarded by sorptive interactions. The diminished mobility caused by sorption has the obvious advantage that the rate of migration
of dissolved contaminants from a spill site is less than the velocity of the groundwater.
Thus, the volume of the porous medium contacted by a plume in a given period of time
is less for more highly sorbing compounds than for less sorbing species. In this manner,
sorption provides a natural mechanism which reduces the size of a contaminant plume.
At the same time, however, the retarded behavior of sorbing compounds can make them
difficult to remove from the subsurface. In particular, when sorbing pollutants are involved, more effort will be required to remediate a site using "pump and treat" methods
than would be required for a site contaminated with non-sorbing species (Mackay and
Cherry, 1989).
In addition to affecting migration rates, sorption can reduce the rates at which organic contaminants are biodegraded. Laboratory studies have shown that in some cases,
the portion of the compound that is sorbed to the solid phase is unavailable for biodegradation (Mihe1cic and Luthy, 1991; Ogram et al., 1985, Steen et al., 1980). Thus, sorption
can decrease the rate of biotransformation of an organic compound in soil by making a
portion of the compound at least temporarily unavailable. Recent laboratory experiments
have shown that biodegradation can be enhanced by adding a low concentration of a
surfactant which will promote desorption (Aronstein et al., 1991).
The degree of sorption in the subsurface depends on the physical and chemical
properties of the dissolved contaminant, the solids, and the aqueous solution. The sorption of different compounds may differ by orders of magnitude in the same geologic
material. Sorption can also vary significantly between sites for a single compound. In
sand and gravel aquifers, typically the degree of sorption is low compared to sorption
Sec. 8.2
235
in silt, clay, or organic-rich sediments. We also note that the degree of sorption between
different sand and gravel aquifers varies less than when sand and gravel aquifers are
compared with silty, clayey, or organic-rich sediments.
Although the properties of the compound and the solids exhibit the greatest influence on the degree of retardation, system-specific properties (i.e., pH, temperature, natural
dissolved organic matter (DOM) concentration, co-solvent concentration, surfactant concentration, and the soil water content) can also affect the degree of sorption. In some
cases, natural DOM can significantly reduce sorption; high concentrations of the contaminants themselves can also reduce sorption. The fact that surfactants and co-solvents in the
groundwater can significantly decrease sorption has stimulated much current research. In
some circumstances involving complex mixtures of dissolved contaminants, the degree of
sorption of a particular compound can be decreased by the presence of other contaminant
compounds though in cases involving relatively dilute levels of organic contaminants,
the effect will usually be small.
In general, solid-aqueous interactions are not completely understood, and comprise
an active area of research. However, a body of field and laboratory observations exists
from which discernable trends can be obtained. With the appropriate degree of caution,
predictions of sorption and retardation for dissolved organic solvents can be made from
correlations derived from these empirical findings, and from laboratory and field methods
developed to measure sorption or retardation directly. The goal of this chapter is to
describe the underlying mechanisms of sorption of chlorinated solvents as we understand
them, and to discuss the effects of various system properties on sorption.
8.2 EFFECTS OF SORPTION ON TRANSPORT

The Stanford-Waterloo natural-gradient field tracer test (Mackay et al., 1986a) provides quantitative information demonstrating the effect of sorption on the transport of
several dissolved chlorinated and brominated compounds under natural conditions. In
this experiment, measured masses of several contaminants, including carbon tetrachloride (CTET), tetrachloroethylene (PCE), 1,2-dichlorobenzene (DCB), hexachloroethane
(HCE), and bromoform (BROM) were dissolved together with two non-reactive tracers (chloride and bromide) in a known volume of water which was then injected into a
sandy, low fraction organic carbon (foe) aquifer near the "Borden Landfill" at Canadian
Forces Base Borden. The movements of these compounds and the non-reactive tracers
were monitored as a function of time.
Figure 8.1 shows a plan view of the vertically-averaged concentration of chloride at four different sampling times (1, 85, 462, and 647 days) during the experiment.
Chloride behaved conservatively, showing the transport of the injected plume of water.
Figure 8.2 shows a plan view of the distributions of chloride, CTET and PCE approximately 21 months (633 to 645 days) after injection. CTET and PCE both lag behind
chloride in terms of plume movement. Over the course of the experiment, the velocity
of the PCE center of mass was observed to be a factor of 2.7 to 5.9 slower than the
unretarded velocity of the chloride. The degree of decreased mobility of PCE, characterized by this factor of 2.7 to 5.9 in this particular example, is commonly referred to
236
Chap. 8
(rn)
a) chloride
60
50
40
462 days
30
20
10
-IO'--
-10
. , B5doys
1 day
..L...
-'-
10
---'-
--J.
----'
y (m)
Figure 8.1 Chloride ion concentration (vertically averaged) in Stanford-Waterloo injection experiment, 1, 85, 462, and 647 days after injection. Contour interval is 5 mglL.
(Reprinted with permission from Mackay et al., 1986.)
as the retardation factor. The retardation of CTET and PCE was due to sorption to
the aquifer materials (Roberts et al., 1986). For the various compounds injected, the
retardation factors were generally proportional to the degree of hydrophobicity as given
by the octanoVwater partitioning coefficient (K ow ) .
Breakthrough curves of concentration in one piezometer over time through the
experiment also show the effect of sorption on mobility (Figure 8.3). The chloride and
bromide were the first to arrive at a given point down gradient. They were followed by
Sec. 8.2
237
(rn)
60
50
40
30
CTET
20
633 days
peE
10
633 days
10
I
20
I
30
I
y(m)
Figure 8.2 Separation of chloride, carbon tetrachloride (CTET), and tetrachloroethylene

(PCE) plumes in Stanford-Waterloo injection experiment 21 months (633 or 647 days)
after injection. The contour interval for chloride is 5 mgIL and for CTET and PCE is
0.1 J.LgfL. (Reprinted with permission from Roberts et al., 1986.)
238
0.9
Chloride
0.8
0.7
0.6
Tetrachloroethylene
>
0.5
.....J
0.4
I-
Chap. 8
a:
0.3
0.2
0.1
0
0
200
400
TIME (DAYS)
Figure 8.3 Breakthrough curves for chloride, carbon tetrachloride (CTET), and tetrachloroethylene (PCE) in Stanford-Waterloo injection experiment in a piezometer located
5 m from the injection point. (Reprinted with permission from Roberts et al., 1986.)
the least hydrophobic (least sorbing) compounds, which were in turn followed by the
more hydrophobic (more sorbing) compounds.
In the Stanford-Waterloo experiment, the retardation factors of the compounds examined ranged from about 1.7 to 9.0. While significant, these retardation factors are relatively low; retardation factors which are orders of magnitude higher have been observed
for some compounds. Anderson and Pankow (1986) predicted that the retardation factors
for several very hydrophobic polychlorinated biphenyl (PCB) isomers in a certain silty
aquitard are extremely high, ranging from 270,000 to 1,400,000; compounds with such
high retardation factors, will be essentially immobile if no transporting organic matter or
particles are moving with the groundwater. At the same site, the retardation factors for the
smaller and less hydrophobic 1,2,4-trichlorobenzene and I,2,4,5-tetrachlorobenzene were
estimated to be 120 and 600, respectively. For the even less hydrophobic compound PCE,
the retardation factor in the same material would be expected to be about an order of magnitude less than for the chlorinated benzenes. Thus, depending on the compound, varying
degrees of hydrophobicity can cause many orders of magnitude difference in mobility.
As noted above, sorption for a given compound can vary between sites. At the
sandy site of the Stanford-Waterloo experiment, the retardation factor for PCE averaged
around 4.5 (Roberts et al., 1986). The retardation factor for PCE in sand and gravels near
the River Aare is about 5 (Schwarzenbach et al., 1983). In the sand and gravel aquifer at
the Otis Air Force Base on Cape Cod in Massachusetts, Barber et al. (1988) has found
Sec. 8.3
239
Sorption Mechanisms
that PCE exhibits a retardation factor of ~ 1 (no retardation). As with PCE, literature
values for the retardation factor for TCE at sandy sites are generally less than 10, often
between 1 and 2.5 (Mackay, 1990; Mehran et al., 1987; Wilson et al., 1981).
The Stanford-Waterloo injection experiment demonstrated some of the complexities
of interpreting the effects of sorption on transport in the field. Based on a simplistic model
of sorption in a homogeneous system, "ideal" breakthrough curves are symmetrical.
However, the breakthrough curves observed in the Stanford-Waterloo experiment were
not all symmetrical (Roberts et al., 1986). "Tailing" breakthrough curves such as those in
Figure 8.3 indicate "non-ideal" behavior. Asymmetrical breakthrough curves with long
tails are commonly observed at field sites and also in laboratory experiments. We note
that ideal behavior in a homogeneous material is characterized by a constant retardation
factor for a given compound. However, the retardation factor for each of the compounds
in the Stanford-Waterloo experiment was observed to increase during the course of the
experiment (Figure 8.4). Suggested reasons for the observed non-ideal behavior include
both pore-scale and field-scale rate-limited processes. In order to understand such nonideal behavior, it is important to understand the concepts and assumptions implicit in the
assumption of ideal behavior.
00
200
400
TIME, DAYS
600
Figure 8.4 Plume retardation factors over

time estimated from field synoptic sampling
results in Stanford-Waterloo injection
experiment. HCE = hexachloroethane;
DCB = dichlorobenzene; PCE =
tetrachloroethylene; BROM = bromoform;
CTET = carbon tetrachloride. (Reprinted
with permission from Roberts et al., 1986.)
8.3 SORPTION MECHANISMS: EFFECTS OF SORBATE AND SORBENT

ON SOIL/WATER PARTITIONING
8.3.1 Definitions
A plot of sorbed concentration vs. solution concentration at a given temperature IS a

sorption isotherm. Isotherms for various organic compounds on several natural subsurface samples have been obtained (e.g., Weber et al., 1992; Ball and Roberts, 1991a).
240
Chap. 8
For dilute solutions of chlorinated solvents sorbing to soils, sorption isotherms are
often linear with a zero intercept (e.g., Figure 8.5). For linear sorption, the partition coefficient, also known as the distribution coefficient, Kd expresses the ratio of
the equilibrium sorbed and solution phase concentrations, and equals the slope of the
isotherm, i.e.,
(8.1)
where C, and C w are the concentrations of the compound of interest in the solid phase
(mass of sorbate/mass of solid) and water phase (mass of solute/volume of water), respectively. For the linear case, the dimensions of K d are volume of water per mass of
solid, typically mUg.
For linear sorption in a saturated system, and when equilibrium between the solid
and solution phases occurs rapidly (and reversibly) with respect to flow velocity, the
retardation factor R is given by
Ph
= -VV = 1 + -K
e d
c
(8.2)
(Freeze and Cherry, 1979) where v and V c are the average linear velocities of the water
and the contaminant respectively, Ph is the bulk density of the solids, and ew is the
water-filled porosity. R is independent of Cll! when the isotherm is linear.
240
'0
(J)
-.s
Cl
Cl
.s
160
SINGLE SOLUTE ISOTHERMS:

o BENZENE (UPPER & RIGHT SCALES)
o 1,3 - DICHLOROBENZENE
6. 1,2,4 - TRICHLOROBENZENE
BINARY SOLUTE ISOTHERMS:
1,3 - DICHLOROBENZENE
1,2,4, - TRICHLOROBENZENE
...i
0
(J)
>lD
LlJ
~
80
I0..
~
0
0
60
40
20
EQUILIBRIUM CONCENTRATION,
(mglL)
80
c;
Figure 8.5 Examples of linear equilibrium isotherms for three hydrophobic compounds
and a moderate to high foe soil. (Reprinted with permission from Chiou et al., 1983.)
Sec. 8.3
241
Sorption Mechanisms
A Freundlich relationship, of the form

C, = KfC;jn
(8.3)
is often used to fit non-linear isotherms, as shown in Figure 8.6 for a sample of sand from
the Borden aquifer. The quantity I In describes the degree of deviation from linearity and
K f indicates the magnitude of sorption. Values of K f and 1In are usually obtained by
fitting sorption data. Non-linear isotherms are considered to be most important at high
concentrations when the sorption capacity of the solids is approached (e.g., Murphy et
al., 1990; Grathwol, 1990). Non-linear isotherms can often be approximated as linear at
low concentrations (Ball and Roberts, 1991a). However, extrapolation from low to high
concentrations assuming linear behavior can produce significant errors (Figure 8.6).
The retardation factor corresponding to a Freundlich sorption isotherm is given by
R
= 1+
Ph
~Kf c[ljn-l]
w
(8.4)
ew n
and is dependent on solution concentration. As seen in Figure 8.7, this type of behavior
has been found for sorption of peE to Borden sand. Non-linear isotherms suggest that
competition between organic compounds should affect sorption. When lin < 1 for each
1E6
1E5
...--.
C)
Observed
1E4
C>
::::J
.........
Freundlich
(J)1000
Linear
100
10
10
100
1000
1E4
1E5
C w [ug/L]
Figure 8.6
Example of non-linear equilibrium isotherm for tetrachloroethylene (PCE)
in low joe Borden sand from the site of the Stanford-Waterloo injection experiment.
The projection of Kd for approximately linear, low concentration range is shown. The
projection emphasizes the importance of the non-linearity at high concentrations.
242
Chap. 8
16
12
o
Observed
0:: 8
Freundlich
Linear
o
10
100
1000
c, [ug/L]
1E4
1E5
Figure 8.7 Retardation factor (R) versus concentration for tetrachloroethylene (PCE) in
Borden sand corresponding to the isotherm shown in Figure 8.6. The value of R depends
on the concentration for the non-linear isotherm. The projected linear retardation factor
obtained from low concentration data is given as a dashed line.
of the competing compounds, sorption of a particular solute will generally be decreased

when others are present at significant concentrations (McGinley et al., 1993). Although
Brusseau and Rao (1989) suggest that, in most cases, organic contaminant concentrations
in groundwater are sufficiently low that they are within the range of the linear portion
of the isotherm, sufficiently high concentrations may be present near chlorinated solvent
sources that isotherm non-linearity may be important.
An understanding of the mechanisms underlying sorption of chlorinated solvents to
aquifer materials is needed to allow reliable prediction of sorption in diverse subsurface
environments. Unfortunately, the mechanism(s) for the sorption of nonionic organic compounds are not yet completely defined nor agreed upon in the literature. In this chapter,
"adsorption" will be taken to refer to an interaction which occurs at the surface of a
solid; "absorption" will refer to the uptake of the sorbate within the volume (or a fraction
of the volume) of the sorbing phase. Based on various empirical observations and theoretical arguments, several researchers believe that sorption of nonionic organic compounds
to soils is dominated by an absorption mechanism (Chiou et al., 1983, 1988, 1990; Chiou
and Shoup, 1985; Gschwend and Wu, 1985). Others (e.g., Murphy et al., 1990; MacIntyre
and Smith, 1984; Ming1egrin and Gerstl, 1983) are proponents of adsorption mechanisms.
Actually, there seems to be evidence to conclude that there is a spectrum of cases in which
one or both mechanisms contribute (see Karickhoff, 1984; Hassett and Banwart, 1989).
Sec. 8.3
243
Sorption Mechanisms

In the "hydrophobic sorption model", it is absorptive partitioning between the dissolved phase and the organic fraction of the soil that is thought to be the mechanism for sorption (Chiou et al., 1983). (This model is analogous to the "like dissolves like" partitioning of a compound that can occur when a hydrophobic organic
compound partitions into a non-polar phase such as octanol or oil.) In this model,
nonionic organic compounds are viewed as being "dissolved" in the organic portion
of the soil solids. Figure 8.8 shows that for sorption by soils with foe > 0.001, the
Kd value of a given chlorinated aromatic compound increases with increasing foe.
Also, at constant foco for a group of different compounds a trend of increasing Kd
with decreasing solubility (increasing hydrophobicity) is demonstrated. These two observations have been commonly made for sorption of nonionic organic pollutants from
low concentration aqueous solutions, generally less than half of the water solubility
log Kpc
o 1,2,4,5 - TECB
(TECB)
0/
1,2,4 - TCB
6. 1,4 - DCB
CBL
log Kpc
"/
(TCB)
/
log Kpc
(DCB)
/
/
109K~
( LB)
/
/
.....
/
/
'"
Cl
.......
""u
00
06.6.
Cl
..Q
++
o/l~
...J~
-1
."'~O
loU a::: IX
~u.Ju..o::
~~~
<:::Eu..a:::
-3
Vl
I-a:
zw
wI-
!Zw
:;)
...J
Vl
~~ffi
llj",~
w'"
is
w
"'''-",
a,<
C2<~
>
Vl
t.i<
'"
:5
-2
-1
log foe
Figure 8.8 Kd values for some nonionic organic compounds as a function of the foe
of the solid phase. (Reprinted with permission from Schwarzenbach and Westall, 1981.)
244
Chap. 8
(Karickhoff, 1981; Karickhoff eta!., 1979; Chiou et al., 1979, 1983; Schwarzenbach and
Westall, 1981).
As discussed above, when sufficient solid organic carbon is present, K d is generally directly proportional to foe (Schwarzenbach and Westall, 1981). In such cases, it is
common to calculate the sorption coefficient normalized to the organic carbon content as
K oe = K d / foe. For a given compound, K ae values have been found to be approximately
constant (within a one order of magnitude range) for sorption to a wide variety of sediments and soils (Karickhoff, 1981; Schwarzenbach and Westall, 1981). These and related
studies suggest that sorption from dilute solutions is primarily a function of the solute's
solubility or hydrophobicity, with the sorptive nature of the organic carbon (gram for
gram) in the solid materials being remarkably constant from location to location (Chiou
et al., 1983). In the studies listed above and in others (e.g., Gerstl, 1990), K ae values
have been correlated with compound-specific properties such as solubility S (mg/L) and
K ow that can be found in the literature (e.g., see Montgomery and Welkom, 1990; Hansch
and Leo, 1979). The resulting correlation equations can be used for estimating K oe for
compounds based either upon S or K aw , with estimation of K d then possible given the
foe of the solids.
The form of the correlation equations available for estimating K ae from K aw is
log K ae = a log K ow
+b
(8.5)
the correlation equation based on solubility is

log K oe = clog S + d
(8.6)
where a, and b, as well as c and d are empirical, compound-class-dependent parameters

that are derived by regression of the available data. Some of the most commonly-used
regression coefficients from the literature were obtained by Karickhoff (1981, 1984),
Schwarzenbach and Westall (1981), and Chiou et al. (1983). The coefficients discussed
in those studies were derived using data for a variety of compounds and sediment/soil
samples.
Values of K aw and S vary over several orders of magnitude for organic compounds
of typical interest. This variation causes a similar range of variation in the corresponding
K d values for a soil material of interest. The K ow values for very hydrophobic chemicals,
such as some PCBs, are greater than about 104 . Moderately hydrophobic compounds,
such as the di- and trichlorobenzenes, have K ow values in the 103 _104 range. K ow values
for TCE, CTET and other chlorinated solvent compounds are in the 10-103 range.
Correlation equations like Eqs.(8.5) and (8.6) provide a quick and inexpensive
method for qualitatively predicting sorption. However, one does need to remember that
there are always limitations involved in using empirically-based equations in describing
naturally-variable environmental systems, and that these equations apply best for sorption
to soils and sediments when foe is larger than ~O.OOI (Schwarzenbach and Westall, 1981).
We also note that when using either Eq.(8.5) or Eq.(8.6), it is best to choose a form of the
regression equation that is based on data for compounds that are as similar as possible to
the compound(s) of interest (e.g., of the same compound class); some researchers have
suggested that within a homologous series of compounds, a K oe estimate will be within
Sec. 8.3
245
Sorption Mechanisms
a factor of 2 of the measured value (Schwarzenbach and Westall, 1981; Karickhoff,

1981).
Recommended values for a and b for Eq.(8.5) are given in Table 8.1 for halogenated
nonaromatic compounds (which includes almost all of the chlorinated solvents) and also
for halogenated aromatic compounds (e.g., p-dichlorobenzene); recommended values for
c and d for Eq.(8.6) are given in Table 8.2. As a result of the many sorption studies
that have been carried out involving common groundwater pollutants, desired K oc values
may often be found directly in the literature (Montgomery and Welkom, 1990), though
we note again that K oc values measured for a given compound in different soils can vary
by as much as an order of magnitude.
In addition to using Eqs.(8.5) and (8.6), K oc values can also be estimated based on
the value of the "molecular connectivity index", a parameter which is calculated from
the molecular structure of a compound. An equation incorporating this index was found
to adequately predict K oc values for a test group of diverse compounds (Sabljic, 1987).
However, Sabljic (1987) has shown that this index alone is no better at predicting the K oc
values of a large test group of compounds than are Eqs.(8.5) or (8.6). Other correlation
equations have been developed which include corrections for molecular structure. Again,
however, these are not necessarily more accurate than those based on measured K ow or
S values (Gerstl, 1990).
TABLE 8.1
Coefficients for Log KocLog Kow Correlation.
Compound class
r2
nonaromatic,
halogenated
compounds
0.827
-0.039
0.698
Gerst! (1990)
aromatic
0.722
0.417
0.855
Gerst! (1990)
0.69
0.22
0.99
Piwoni and Banerjee (1989)
Reference
halogenated
compounds
TCE, PCE,
DCB, and
benzene on low
foe solids
TABLE 8.2
Coefficients for Log KocLog 5 Correlation.
Compound class
r2
Reference
nonaromatic,
halogenated
compounds
-0.346
1.28
0.452
Gerstl (1990)
aromatic
halogenated
compounds
-0.475
1.318
0.839
Gerst! (1990)
246
Chap. 8
The Stanford-Waterloo natural-gradient field tracer experiment (Mackay et al.,

1986a) provided benchmark, quantitative bulk field measurements of R for several halogenated compounds for a low-carbon sandy aquifer. These values can be compared to
values predicted based on foe as measured for the Borden aquifer, although the foe value
of the aquifer (0.0002) is much lower than the generally accepted threshold value for
sorption dominance by organic carbon. Listed in Table 8.3 are the field-measured R values
along with the values predicted using three regression equations from the literature. The
relative degrees of retardation of the compounds in the aquifer agree with the predicted
sequence. However, the predicted R values are typically lower than the field values (Ball
and Roberts, 1991a; Ptacek and Gillham, 1992; Roberts et al., 1986; Curtis et al., 1986).
Generally, the predicted K d values are a factor of two-to-five times lower than would be
needed to obtain agreement between the measured and predicted R values. This example
emphasizes the "order of magnitude" accuracy of the correlation-based estimates of Kd.
Ball and Roberts (l991b) found that full sorption equilibrium was reached only
after very long contact times, much longer than the contact times used for standard K d
measurements. Therefore, there is a possibility that some of the Kd values in the literature do not reflect equilibrium conditions (Ball and Roberts, 1991c). As a result, existing
correlation equations may have a built-in bias to underestimate sorption. Another limitation may be related to the type and amount of organic carbon in the soil. We note that
most of the available correlation equations were derived from observations of sorption in
surficial soils, sediments, and sludges, with foe values greater than 0.001. Typical sand
TABLE 8.3 Measured (Field) and Calculated Retardation Coefficients for Three Chlorinated
Hydrocarbons in Borden Sand.
R
Calculated (b1
(0
Compound (a)
Log K o w
Measured
(c)
(d)
(e)
CTET
2.73-2.83
(g)
1.8-2.5
1.31-1.37
1.01
1.18-1.22
1.24-1.48
PCE
2.53-3.78
(h)
2.7-5.9
1.22-2.79
1.01-1.02
1.12-2.34
1.23-1.40
DCB
3.38-3.55
(g)
3.9-9.0
1.92-2.21
1.02
1.12-1.15
1.20-2.86
(a) CTET = carbon tetrachloride, PCE = tetrachloroethylene, DCB = 1,2-dichlorobenzene;

(b) linear retardation (Eq.(8.2ll with Pb = 1.81 g/cm-', Ow = 0.33,
correlations or source;
Joe =
0.0002, and K oc is from cited
(c) equation of Schwarzenbach and Westal1 (1981);

(d) equation of Chiou et al. (1983) and assuming that
Joe
= 0.5
OM (OM
(e) using equation of Gerst! (1990) for halogenated hydrocarbons;

(f) calculated from K oc values as cited in Montgomery and Welkom (1990);
(g) Montgomery and Welkom (1990); and

(h) as cited by Bal1 and Roberts (1992).
= organic
matter);
Sec. 8.3
247
Sorption Mechanisms
and gravel aquifers, like the examples in Table 8.4, have very low foe values. Low
foe values pose two problems for use of the correlation estimation methods. Firstly,
low foe values themselves are difficult to measure accurately (Ball and Roberts, 1991c).
Secondly, sorption may not be dominated by organic matter in systems of very low foe.
The literature on sorption in low foe materials is not as complete as that for high foe
soils. Murphy et al. (1990) found that the typical correlation equations do not accurately
predict K d values when foe is low. When K oe is back-calculated as K d / foc> Macintyre
et al. (1991) and Mackay et al. (1986b) have found poor correlation between K oe and
K ow values in low foe sandy materials, suggesting that sorption may not be dominated
by organic carbon in these media. Therefore, although mineral surfaces have been shown
to be unimportant in sorption of nonionic organic compounds in high foe soils, it has
been suggested that sorption to minerals may playa significant role in low foe systems.
TABLE 8.4
Organic Carbon Content (foel of Typical Sand Aquifers (after Mackay, 1990).
Site location
Gloucester, Ontario
Organic carbon content

0.001-0.006
Material
interstratified silts,
sands, gravels
Borden, Ontario
0.0002
medium to fine sand
Moffett Naval Air Station,

California
0.0011
coarse gravel
Otis Air Force Base,

Massachusetts
0.0001-0.0075
sand and gravel

It has been shown that organic compounds can sorb directly to mineral surfaces (Perlinger and Eisenreich, 1991; Perlinger et al., 1993). For nonionic compounds, such sorption is perhaps most pronounced in materials with substantial swelling clay content, and
in materials of low foe value (Karickhoff, 1984). The literature on the sorption of chlorinated solvent compounds to natural, clay-rich soils and sediments is limited. In a review
of the pertinent literature, Karickhoff (1984) suggested that at ratios of swelling clay
fraction to organic matter fraction of greater than approximately 25 to 60, adsorption to
mineral surfaces will dominate the sorption. In several studies, measured K d values for
chlorinated hydrocarbons in a low foe clay-rich material were found to be approximately
an order of magnitude greater than would have been predicted based just on the foe value
(Barone et al., 1992; Myrand et al., 1992; Allen-King et al., 1995; Johnson et al., 1989);
sorption to clay mineral surfaces was identified as a possible explanation for the enhanced
sorption. When Allen-King et al. (1995) reduced the foe value of their clay by baking,
the measured Kd values were correspondingly reduced. Recent research has suggested
that in some instances, the organic surface coatings on the clay (rather than direct mineral
surface adsorption) may be responsible for enhancing the sorption beyond what would
be expected based on the foe value alone (Murphy et al., 1990; Smith et al., 1990b).
248
Chap. 8
In low foe materials, such as the aquifer materials listed in Table 8.4, there is evidence to suggest that minerals other than clays may affect sorption (Piwoni and Banerjee,
1989; Stauffer and MacIntyre, 1986). In sand from the Borden aquifer (Ball and Roberts,
1991a; Mackay et al., 1986b; Curtis et al., 1986) and in sand from the sand and gravel
aquifer at Otis Air Force Base on Cape Cod in Massachusetts (Barber et al., 1992), measured sorption has been found to be greater than expected based just on the measured
value of foe. We also note that Ball and Roberts (l991a) found that for very long contact
times, PCE and 1,2,4,5-tetrachlorobenzene sorption coefficients were about an order of
magnitude greater than one would expect from K oe foe calculations, and that the sorption isotherms were non-linear over a wide concentration range. They hypothesized that
slow intraparticle diffusion was responsible for the long equilibration period. They found
that the presence of freshly-broken surfaces in a pulverized sample did not significantly
change the equilibrium Kd values compared to a natural sample, a result which they
interpreted to mean that surface coatings (not necessarily organic) were likely important
in this low foe material.
8.4 METHODS FOR MEASU RI NG SORPTION

8.4.1 General
The sorption of organic compounds to subsurface materials has been measured using
both small- and large-scale methods. In the laboratory, the commonly used small-scale
methods are batch experiments and column experiments. A field column method
has also recently been developed (Gillham et al., 1990). Large-scale methods include
natural-gradient and forced-gradient field tests.
In studies involving volatile compounds such as the halogenated solvents, care
must be taken during all steps of the measurement process to prevent losses. Materials
such as glass and stainless steel should be chosen so as to minimize sorption losses to the
measurement equipment (Lion et al., 1990). Control experiments should be carried out
to evaluate the importance of any volatilization and sorption losses. It should be noted
that Teflon may sorb some compounds enough to affect the results (Ball, 1989).

Batch methods can be used to measure a Kd value for a discrete sample under conditions
which maximize the contact between the solids and the solution. Typical short term batch
methods are described by Ptacek and Gillham (1992), Lion et al. (1990), and Curtis et al.
(1986). In these methods, a subsample of the solids is placed in a vial, the vial is filled with
an aqueous solution containing the compound(s) of interest, and the vial is mixed for a
set period of time. The equilibration should take place at a temperature similar to those in
the subsurface (K d values are temperature dependent). After equilibration, centrifugation
is used to separate the phases. The aqueous phase and/or the solid phase are sampled and
analyzed. For sampling and analysis of the aqueous phase alone, mass balance is used to
Sec. 8.4
249
determine the concentration on the solids. Vials are usually replicated and prepared over a
range of initial concentrations to allow the generation of an isotherm. If the equilibration
temperature differs from the subsurface temperature of interest, temperature corrections
on the measured partition coefficient can be made (Curtis et al., 1986).
Short term batch methods typically call for a 24 to 48 hr contact time (e.g., Brusseau
and Rao, 1989). It has been noted that the rate of uptake from the solution is very rapid
initially, but decreases significantly after the first few hours of contact. Equilibrium is
assumed following the contact period. However, as noted in the previous section, results
collected over the past few years with Borden sand cast doubt over the assumption that
equilibration is always reached during a 24 to 48 hr contact time. For example, the low
concentration batch Kd values found by Ball and Roberts (1991a) for very long contact
periods (several months) were more than half an order of magnitude greater than Ptacek
and Gillham's (1992) 24-hr batch values. On the other hand, long-term sorption studies
conducted with different compounds and sand from another site suggest that equilibrium
was attained within 24 hr (MacIntyre et al., 1991). It would seem that the assumption of
equilibrium after ~24 hr may not generally be appropriate. Methods for conducting batch
experiments may improve for the measurement of equilibrium sorption coefficients.
In batch methods, the formation of suspended rnicroparticles has been shown to
affect the determination of a sorption coefficient. The interference of microparticles can
be avoided for volatile compounds by measuring concentrations in the headspace instead
of the aqueous phase (Garbarini and Lion, 1985), or by varying the solids/water ratio
when a significant effect is suspected (Gschwend and Wu, 1985).

In column methods, the contact between the aqueous and solid phases is more like the
natural field conditions than it is in the batch experiments. Laboratory column methods
have been described by Ptacek and Gillham (1992), MacIntyre et al. (1991) and Brusseau
et al. (1991a, 1991b), among others. The methods generally include packing aquifer
material into a column, establishing a steady flow of water through the column, injecting
a small volume of an aqueous solution containing the compounds of interest as well as
a non-reactive tracer, then collecting samples of the effluent water over time.
An in situ field column method has been developed by Gillham et al. (1990) and
used by Ptacek and Gillham (1992) for measuring retardation of chlorinated solvents in
the sandy Borden aquifer. In this method, a natural column of aquifer material is isolated
from the surrounding material by a split spoon core barrel, which then serves as the
container for the column. Prior to installation, the core barrel is fitted with two screened
zones for injecting and withdrawing solutions. The natural geochemistry and stratigraphy
of the system are preserved in this method.
For all column methods, data analysis is achieved by fitting a transport model to
the breakthrough curves. The retardation factor and the distribution coefficient are obtained from the curve fitting exercise. However, these values may not be as accurate
as those obtained by batch methods because non-equilibrium and transport effects often
complicate the interpretation of column results.
250
Chap. 8
8.4.3 large-Scale Methods
Few large-scale natural-gradient tests such as the Stanford-Waterloo injection experiment

have been conducted due to the labor and cost associated with such undertakings. Nevertheless, these tests are the best-documented cases of how transport in a natural system is
affected by sorption. As such, this type of field experiment provides a benchmark against
which other methods for measuring partitioning can be compared. Forced-gradient tests
can be conducted over shorter periods of time than natural-gradient tests, and may therefore prove to be very useful for measuring large-scale effects of sorption on transport in
situations where accurate field information is essential.
In a forced-gradient test, a solution containing a tracer is injected in one well while
pumping occurs at another well. Because of the gradient induced by pumping, the travel
time between the two wells is less than it would be under natural conditions. Samples of
the effluent water from the pumping well and from piezometers along the flowpath are
collected. The forced-gradient test conducted by Mackay et al. (1994) involved injecting
dissolved chlorinated solvents together with conservative tracers at a site directly adjacent
to the site of the Stanford-Waterloo field experiment. Alternatively, at a contaminated site,
clean water can be injected, thus desorbing contaminants from the system (Whiffin and
Bahr, 1985; Bahr, 1989; Mackay and Thorbjarnson, 1990; Mackay et al., 1994). For either
method, the results are analyzed by fitting breakthrough curves with a numerical transport
model. If the data are sufficiently detailed, insight can be gained on the spatial variability
of the hydraulic conductivity, as well as the contaminant sorption capacity and kinetics.
8.5 NON-IDEAL SORPTION BEHAVIOR IN A SATURATED SYSTEM

Breakthrough curves generated from column and field experiments are defined as ideal
if they fit the advection-dispersion equation with a retardation term that assumes linear, reversible, instantaneous (i.e., equilibrium) sorption. A non-sorbing and a sorbing
tracer (tritium and calcium respectively) were found to demonstrate ideal behavior in the
laboratory column experiment represented in Figure 8.9.a. When normalized by the retardation factor, the breakthrough curves for the two tracers are coincident (Figure 8.9.b).
In the same experiment, diuron, a nonionic organic pesticide compound, was found to
exhibit non-ideal behavior. The breakthrough curve for diuron shows both early breakthrough as well as tailing compared to the ideal tracers in Figure 8.9.b. Similar non-ideal
breakthrough curves were observed at the field scale by Bahr (1989) with the compound
tetrahydrofuran for samples taken in a piezometer along the flowpath between the injection and purge wells of a forced-gradient field test. We also note that asymmetry and tailing were observed in the breakthrough curves of PCE and CTET in the Stanford-Waterloo
experiment (see also Section 8.2 and Figure 8.3). Non-ideal breakthrough curves, such as
those shown in Figures 8.3 and 8.9 have frequently been observed for nonionic organic
compounds in column experiments (Ptacek and Gillham, 1992; Lee et al., 1988, 1991;
Sec. 8.5
1.0
f-
i=
~
0::
0.8
0.6
fzw
e
e
f-
a
~
0.4
>
i= 0.2
f-
:3w
f-
0::
a
a
eA
3H 0
DIURON
4S
,0
a
~
~ ~ ~aa
ea
10
PORE VOLUME (p )
1.0
0.8
zw
0.6
z
o
0.4
f-
o
0
W
~
I-
0.2
:5
W
e::
I-
a~
Figure 8.9.a Breakthrough curves from

a saturated column experiment. 45Ca
and 3H20 demonstrate ideal behavior
of retarded and conservative tracers,
respectively.
_.
0
0
fO~
~~~
f-
251
H 2O
DIURON
45 Ca
fl
2
RELATIVE PORE VOLUME (p/R)
Figure 8.9.b Diuron does not behave

"ideally". (Reprinted with permission from
Nkedi-Kizza et al., 1989.)
Brusseau et al., 1989b, 1991a,; Nkedi-Kizza et al., 1989; Bouchard et al., 1988) and in
the field (Mackay et al., 1994; Mackay and Thorbjarnson, 1990; Bahr, 1989; Roberts et
al., 1986).
As has been noted above, non-ideal behavior yields early breakthrough and tailing relative to ideal behavior. Travel time for contaminant movement from a source to
a designated location such as a drinking-water well would be overpredicted by models assuming ideal behavior. Thus, the well would become contaminated more quickly
than expected. In remedial applications, the long tails (extended section of low concentration at long times) of breakthrough curves relative to the ideal case mean that the
amount of pumping required to remediate a site using the pump-and-treat method will
be significantly greater than predicted based on ideal behavior (Mackay and Cherry,
1989).
Non-ideal behavior has been attributed to both pore-scale effects and field-scale
effects. The reader is referred to Brusseau and Rao (1989) for an in-depth discussion of
pore-scale contributions to non-ideal sorption. The references cited below are primarily
those which have appeared subsequent to the review by Brusseau and Rao (1989), or
which are of particular relevance. Recent research has suggested that field-scale heterogeneity may be more important than pore-scale effects in producing non-ideal behavior
in the field.
252
Chap. 8

Non-ideal behavior due to pore-scale effects may be due to isotherm non-linearity, nonsingularity (i.e. hysteresis), and/or to non-equilibrium sorption. In general, the last of
these is considered to be the most important. Isotherm non-linearity will result in either
early breakthrough or tailing, depending upon the value of n in the Freundlich isotherm
(Eq.(8.3)). However, the simultaneous occurrence of both early breakthrough and tailing
as commonly observed, cannot be accounted for by isotherm non-linearity. Similarly,
hysteresis in an adsorption/desorption isotherm can cause tailing, but not early breakthrough. Therefore, the typically-observed non-ideal behavior is not generally attributed
to either isotherm non-linearity or hysteresis.
In recent field studies, apparently anomalous, long-term persistence of volatile chlorinated hydrocarbons has been attributed to kinetically slow desorption (Pignatello et al.,
1990, 1993; Pavlostathis and Jaglal, 1991). Non-equilibrium sorption is caused by the
combined effects of the flow characteristics, the nature of the sorbate, and the nature of
the sorbent (Valocchi, 1985). It has been observed that the nature of the sorbent has a
dominant effect on the degree of non-equilibrium observed. For the small chlorinated solvent molecules, the properties of the sorbate will exert only a secondary effect (Brusseau
and Rao, 1991).
Non-equilibrium sorption has been attributed to physical and/or chemical mechanisms. Physical non-equilibrium is conceptualized as rate-limited sorption due to a
resistance to mass transfer between two regions in the porous medium. For example, the
mass transfer resistance may be conceptualized as mass transfer by diffusion from "mobile water" in interconnected pores into regions of "immobile water", such as dead-end
pores. Chemical non-equilibrium has been characterized by two types of reaction sites,
"fast sites" and "slow sites". Typically, the fast sites are considered to equilibrate instantaneously; the slow sites are assumed to be rate-limited, and are often characterized using
a first order mass transfer rate. Nkedi-Kizza et al. (1984) have shown that under certain
conditions, physical and chemical non-equilibrium conceptual models are mathematically
equivalent.
Transport models incorporating non-equilibrium effects require more parameters
than models assuming ideal sorption. The additional parameters account for: 1) the rate
of the rate-limited step; 2) the volume ratio of the immobile and mobile zones; and/or
3) the distribution of sorption capacity between different types of sorption sites. Determination of the non-equilibrium parameters is generally achieved by fitting models to
experimentally-determined breakthrough curves (Brusseau et al., 1989b). However, because of the number of fitting parameters, there is a danger that non-unique solutions
may be obtained by this method. The parameters are not easily measured independently,
although methods have been developed to approach this goal (Karickhoff, 1980; Brusseau
et al., 1990; Ball and Roberts, 1991a,b). Because of the mathematical similarity between
physical and chemical non-equilibrium models, a good fit with one of the models does
not generally confirm the validity of the assumed mechanism.
It has been argued that results such as those presented in Figure 8.9 do not support
the mobile/immobile water concept as a cause of non-equilibrium effects (Nkedi-Kizza et
Sec. 8.5
253
al., 1989; Brusseau and Rao, 1989). Simply stated, although immobile domains of water
(e.g., dead-end pores) should affect non-sorbing tracers as well as sorbing tracers, effects
on non-sorbing tracers are not typically observed. In such cases, the immobile/mobile
water concept may still fit the observations if the immobile water is a small percentage of
the total, and the immobile zone contains a large percentage of the sorption sites. Proof
of such a mechanism would be difficult to obtain.
Two similar cases of a physical cause of non-equilibrium invoke slow sites accessed
either by slow diffusion into organic matter, or intraparticle diffusion into mineral or soil
grains. Intraorganic matter diffusion has been shown to be consistent with experimental
results in several studies (Ptacek and Gillham, 1992; Brusseau et al., 1991a, 1990, 1989a,
1989b; Lee et al., 1988, 1991; Nkedi-Kizza et al. 1989; Bouchard et al., 1988). The major
arguments supporting this mechanism as a rate-limiting step in non-equilibrium sorption
are: 1) asymmetrical breakthrough curves for sorbing organic compounds in the presence of symmetrical curves for ideal tracers; and 2) symmetrical breakthrough curves
for sorbing organic compounds and diminished retardation after the soil is treated to
reduce the organic matter content. Intraparticle diffusion in mineral grains is thought
by some researchers to control the rate of sorption in Borden sand (Ball and Roberts,
1991a, 1991b; Curtis and Roberts, 1986). When organic matter is distributed through
a soil particle, intraorganic matter diffusion becomes intraparticle diffusion (Wu and
Gschwend, 1986). For such diffusion-limited sorption, the sorption rate constant will
be inversely proportional to Kd . Both intraorganic matter and intraparticle diffusion are
consistent with the two-stage sorption behavior frequently observed in batch measurements. Finally, we note that although non-ideal behavior has been observed frequently
with sorbing organic compounds, there are some cases, generally with very low organic
carbon materials, in which non-equilibrium effects were negligible (MacIntyre et al.,
1991).
8.5.3 Contribution of Spatial Variability in the Field to Non-Ideal
Sorption Behavior
Figure 8.3 shows a typical asymmetric field breakthrough curve for PCE with a tail
of slowly-declining concentration. In a remedial application, the concentration at which
pumping can safely be discontinued (assuming no remaining source area containing liquid
solvent) would usually be very far down on the tail because drinking water standards
are often much lower than groundwater contamination levels. The amount of pumping
required will be much greater than would be predicted based upon ideal transport.
In the previous section, we considered how non-ideal sorption at the pore-scale
affects lab-scale test results. In some cases, non-ideal behavior at the field scale has been
ascribed to pore-scale processes (Bahr, 1989; Goltz and Roberts, 1988). Alternatively,
spatial variability in the hydraulic conductivity K and in Kd can result in apparently nonideal, field-scale asymmetry in breakthrough curves. Heterogeneity in a layered system
can be thought of as a special case of physical non-equilibrium. In this scenario, layers
of low hydraulic conductivity provide zones of slow uptake and release of solute due
to mass transfer resistance between layers. Sorption within layers and non-equilibrium
254
Chap. 8
effects within layers can also contribute to apparent non-ideal behavior in the field. K d
has been observed to vary significantly in space, even in what most would consider a
relatively homogeneous sandy aquifer (Mackay et al., 1986b and 1994).
Recent theoretical research (Burr, 1992) suggests that moderate heterogeneity, such
as that present in the relatively uniform sand at the Borden site, may be a dominant factor
in creating apparently non-ideal field-scale transport of a sorbing solute, and that porescale effects may be secondary. In this research, the increasing trend of R with time in
the natural-gradient field test (Figure 8.4) can be obtained by assuming spatially-variable
and negatively-correlated K and K d values.
8.6 EFFECTS ON SORPTION CAUSED BY PROPERTIES OF THE
AQUEOUS PHASE
8.6.1 General
Solution effects on sorption act primarily through effects on solubility. In general, mechanisms which increase solubility also decrease sorption of organic compounds, and vice
versa. Effects are generally most significant for very hydrophobic compounds since the
effect on the solubility of such compounds can be large. Usually, natural geochemical
variability has no significant effect on the sorption of organic compounds. Effects can be
observed when the concentrations of dissolved organic compounds are large, or in the
presence of surfactants or co-solvents.
System parameters which can affect sorption can be classified into two groups:
1) geochemical parameters (temperature, pH, and ionic strength); and 2) dissolved or
suspended/colloidal material parameters (e.g., anthropogenic organic compounds such
as the ones already under transport at the site, co-solvents, surfactants, natural humic
organic matter, and colloidal particles).
Variability in the inorganic geochemistry of groundwater does not generally have much
effect on the sorption of nonionic organic compounds. The solubility of a given compound may either increase or decrease with increasing temperature within the narrow,
environmentally-relevant range; the behavior is compound specific. As a result, depending upon the compound, sorption will decrease or increase with temperature, respectively.
For some chlorinated compounds (CTET, BROM, PCE, and DCB) Kd has often been
found to vary by less than 10% over the range of 9 DC to 23C (Curtis et al., 1986;
Ptacek and Gillham, 1992). In one of those studies, the greatest observed variation was
33% for DCB. Thus, K d variation due to temperature is probably small compared to
other uncertainties.
For the halogenated solvent compounds, the pH does not have much effect on
sorption. For example, no significant effect on TCE sorption on a silty clay material was
observed between pH 3 to 13 (Pavlostathis and Jaglal, 1991). For ionizable compounds
like pentachlorophenol (PCP) and other organic acids and bases, however, sorption can
be strongly influenced by pH (Lee et al., 1990, 1991; Jafvert et al., 1990). The degree
Sec. 8.6
255
of influence which pH has on sorption depends on the magnitude of the differences in

affinities of the ionized and neutral species for the solid phase, as well as the ability to
enter into ionic interactions such as ion pair formation and ion exchange (Johnson and
Westall, 1990).
Increased ionic strength decreases solubility of hydrophobic organic compounds,
and increases sorption, due to a "salting out" effect. The effect is not large (Karickhoff
et al., 1979). Pavlostathis and Jaglal (1991) observed little effect on TCE sorption over a
range of ionic strengths under 0.1 M (e.g. up to 5.9 gIL NaCl). When the ionic strength
exceeded 0.1 M, the degree of sorption was increased. Sorption of nonionic organic compounds from a high ionic strength solution approximately equivalent to seawater (NaCI
~ 0.5 M) is predicted to be only ~20% greater than from distilled water (Karickhoff,
1984).
In summary, variability in typical geochemical parameters over normal ranges results in little or no effect on the sorption of nonionic organic compounds. For ionizable
compounds, however, significant pH effects may occur.
8.6.3 Effects of Co-Solvents, Dissolved Organic Matter,

Suspended Colloids, and Surfaetants
8.6.3.1 Co-solvents
When an organic solvent that is at least somewhat miscible with water is mixed with
water, the properties of the water are changed. Methanol and acetone are solvents that
are "completely miscible" with water, i.e., miscible with water over all proportions. As
the fraction organic co-solvent increases, the solubility of a third organic compound will
increase and the sorption of that compound to soils will decrease. A shallow, negativelysloped log-linear relationship has been observed between sorption coefficient and fraction
organic co-solvent in solution (Nkedi-Kizza et al., 1985; Woodburn et al., 1986). The
degree of reduction in sorption depends on the identity of the co-solvent, and is apparently relatively independent of the specific properties of the sorbing solids. Co-solvents
have been shown to significantly increase mobility only at relatively high co-solvent
concentrations, greater than about 5% of the solution (Nkedi-Kizza et al., 1987). This
result suggests that except for close to source zones contaminated with mixed wastes including large quantities of co-solvents, co-solvent effects on plume mobility will usually
not be significant. In addition to completely miscible co-solvents, there are examples of
co-solvents which are not completely miscible with water. The effects of such partiallymiscible co-solvents on organic contaminant mobility are similar on a mass fraction basis
to those of the completely miscible solvents (Rao et al., 1990).
8.6.3.2 Dissolved organic matter (DOM) and suspended

colloidal material
Sorption to dissolved macromolecular organic matter as well as to suspended colloidal
matter will increase the mobility of an organic compound in the subsurface. This effect
is referred to as facilitated transport. Laboratory studies have shown that partitioning of
organic compounds from water to natural dissolved organic matter ("DOM", e.g., "humic
256
Chap. 8
acid" and "fulvic acid") is generally rapid, linear, reversible, and non-competitive (Chiou
et al., 1986). The magnitude of the effect of DOM on the transport of a compound of
interest depends on: 1) the distribution coefficient for the compound of interest to the
DOM; and 2) the concentration of the DOM (Magee et al., 1991).
Laboratory studies such as those by Chin et at. (1990), Spurlock and Biggar (1990),
and Abdul et at. (1990) suggest that the mobility of organic compounds can be significantly enhanced by DOM-facilitated transport, but only for very hydrophobic compounds
with log K ow values greater than 4 to 5 (Chin et al., 1990) and only when DOM concentrations are very high (e.g., 100 mg/L). Since the chlorinated solvents have relatively
low K ow values, the effect of DOM on chlorinated solvent transport will be negligible.
There is some evidence to show that natural DOM travels at greater velocity than
bulk water, probably due to exclusion from small pores (Enfield et al., 1989). Enhanced
mobility due to DOM may be made more significant in this special circumstance, but
the chlorinated solvents will remain largely unaffected.
Recently, some research has been done on the effects on contaminant mobility that
can be caused by non-filterable, colloidal inorganic microparticles (e.g., clays and iron
and manganese hydrous oxides) (Ryan and Gschwend, 1990). The conclusions of that
work are similar to the findings for DOM, i.e., the mobility of some compounds can
be enhanced due to sorption to mobile colloidal particles (Ryan and Gschwend, 1990).
However, as with the effects of DOM, the effects of colloidal particles on chlorinated
solvent transport will be negligible.
8.6.3.3 Surfactants
Surfactants are a broad class of compounds representing a wide variety of properties
and uses, e.g., as detergents and emulsifiers. The effects of surfactants on solubility
and sorption cause this group of compounds to be of interest in groundwater remediation research. In particular, at concentrations of surfactant above the "critical micelle
concentration" (CMC) for the surfactant of interest, aggregates (i.e., micelles) of the
surfactant molecules form. Hydrophobic organic compounds will readily partition into
the micelles. Thus, above the CMC, the apparent solubility of a hydrophobic organic
compound can be increased significantly (Kile and Chiou, 1989). Laboratory research
has recently shown that flushing of contaminated porous media with surfactant solutions
can facilitate the removal of organic compounds, particularly for very hydrophobic contaminants (Abdul and Gibson, 1991). The enhancement of mobility by surfactant addition
is currently a very active research area (e.g., Abdul et al., 1992).
At low concentrations, below the CMC, surfactants are dispersed throughout an
aqueous solution and are also concentrated at the air/water and water/solid interfaces.
The effect of the surfactant on organic compound behavior depends not only on the
concentration, but also on the type of surfactant: anionic, nonionic, or cationic. The solubilities of organic compounds have been shown to be increased by the presence of low
concentrations of nonionic surfactants (Kile and Chiou, 1989). (The increase in solubility
is small compared to the effect observed above the CMC.) The effect is believed to be
due to association of the organic compound with surfactant monomers. Aronstein et at.
Sec. 8.7
257
(1991) demonstrated in laboratory studies that the enhanced desorption of a hydrophobic organic compound caused by the addition of a nonionic surfactant also enhanced
the biodegradability of the compound. No enhanced desorption was observed with low
concentrations of anionic surfactant.
In contrast with anionic and non-ionic surfactants, since cationic surfactants can
sorb strongly to natural, negatively-charged solids (e.g., clays), they can enhance contaminant sorption by creating an organic coating on the solid phase (Boyd et al., 1988).
Indeed, cationic surfactant-coated clay tested in laboratory studies was found to be more
sorptive of organic compounds than the uncoated clay (Boyd et al., 1988). Laboratory research indicates that the mobility of dissolved organic compounds in an aquifer
may be decreased significantly when the subsurface is treated with a cationic surfactant
(Burris and Antworth, 1992). (There was no apparent decrease in the hydraulic conductivity.) Holsen et al. (1991) obtained a similar result with anionic surfactant-coated
clays at low pH. Research on the application of surfactants to limit plume mobility is in
progress.
8.7 SORPTION TO UNSATURATED SOIL
Sorption in unsaturated porous media depends on the degree of unsaturation of the material. At field capacity and wetter, the sorption will be essentially the same as if the
medium was saturated. However, in relatively dry soils, mechanisms apparently exist
which can increase the sorption coefficient above that for wet soils. One can differentiate
soils according to three water content regimes (Ong and Lion, 1991a): extremely dry
(i.e., oven dry), an intermediate dry range, and moderately dry to wet (approximately
field capacity and greater). The specific water contents for these regimes depend on the
properties of the porous medium, particularly the specific surface area.
In the moderately-dry to wet range, the mechanisms for sorption of hydrophobic
organic compounds to the solid phase are the same as those discussed for saturated
materials: all of the solid grains are covered with water. Partitioning between the aqueous
and gaseous phases is governed by Henry's Law (Ong and Lion, 1991a; Call, 1957
as cited in Ong and Lion, 1991b; Spencer and Cliath, 1970), and the K d value for
water/solid partitioning is the same as it is when none of the void space is filled with air.
Assuming linear and instantaneous partitioning, retardation can be described according
to the relationships presented in Chapter 6.
In the laboratory, sorption of gas phase hydrophobic organic compounds to oven
dry materials has been shown to occur directly to unhydrated mineral surfaces (Chiou
and Shoup, 1985). Sorption of TeE to solids under these circumstances can be four
orders of magnitude stronger than in the corresponding saturated system (Peterson et
al., 1988). This enhanced sorption is only of importance when there is less than one
molecular layer of water on the mineral surfaces (Spencer and Cliath, 1970; Ong and
Lion, 1991a, 1991b), a condition which exists only under extraordinarily dry conditions
(Chiou and Shoup, 1985; Rhue et al., 1989).
Ong and Lion (l991b) found that apparent K d values for TCE for air dried
soils at 68% relative humidity were about three times greater than K d values at field
258
Chap. 8
capacity for five different mineral solids. In this intermediate water content range, although there are one to five monolayers of water on the mineral surfaces, there is not
enough water for significant dissolution into the water layer to occur. Sorption in this
range of water content can be attributed to two possible mechanisms: 1) direct adsorption
on unhydrated surfaces which may be present due to uneven coverage of the surfaces
by water; and/or 2) accumulation of the compound at the air/water interface. The second
phenomenon is well known in the surface chemistry literature. At this low water content range, the air-water interface is very large relative to the volume of water. Thus,
accumulation at the interface may contribute significantly to sorption under these conditions.
At field capacity and greater, it has been shown that Henry's Law accurately describes partitioning between the gas and soil water phases (Ong and Lion, 1991a; and,
Call, 1957, as cited by Ong and Lion 1991b). The critical water content for adherence to Henry's Law is equivalent to five molecular layers, but the weight percent
varies depending on the surface area of the medium of interest. Ong and Lion (1991a)
found that the critical water content varied for five materials tested from less than 5%
for kaolinite and iron oxide, up to 15-25% for alumina, humics-coated alumina, and
montmorillinite. Shearer et at. (1973) observed no changes in the partitioning process
above a 4% water content in Gila silt loams (desert soils, ~ 18% clay, 0.6% organic
matter).
Although additional research is needed to determine the significance of enhanced
sorption to dry soils, the laboratory results discussed above suggest that such enhanced
sorption will only be important under very dry conditions. Ong and Lion (1991a) and
Smith et at. (1990a) speculate that soil in arid and semi-arid regions, and surface soils in
temperate regions during the summer season may be dry enough to warrant consideration
of the effects of this enhanced sorption of hydrophobic organic compounds.
8.8 CONCLUSIONS
Sorption can reduce the mobility of dissolved chlorinated solvents in groundwater. Sorption can also reduce the rate of biotransformation. The degree of sorption depends primarily on the identity of the compound of interest, the solids of interest, and the properties of
the aqueous solution. The degree of sorption, as indicated by the distribution coefficient
K d is related to the hydrophobicity of the compound of interest. Somewhat hydrophobic
chemicals such as TeE and CTET are sorbed to a much lower degree than are highlyhydrophobic compounds such as PCBs. When foe> 0.001, K d values can be predicted
based on tabulated K oe values and the foe value of the solids of interest. Since the resulting Kd is an empirically-based estimate, it is important to use these predictions with
the appropriate degree of caution.
If foe is measured accurately and is low 0.1%), K d is typically underestimated by
predictions of the above type. The relatively higher K d values found in low foe materials
may be due to several causes, including direct interactions with mineral surfaces (e.g.,
swelling clays) and a greater affinity of the sorbate for the sorbent organic carbon when
j;}e is low.
Sec. 8.8
Conclusions
259
Equilibrium linear or non-linear isotherms can be measured directly by batch methods with relatively small samples. Isotherm non-linearity can be significant in some
materials when solvent concentrations are relatively high. Column experiments give an
indication of the nature of non-equilibrium effects, and sorption coefficients can be deduced from the results of such experiments. The non-equilibrium effects serve to complicate the interpretation of sorption from the results. However, column experiments are
typically run with flow velocities much greater than one would find in nature.
Large-scale methods provide the most accurate representation of retardation at the
field-scale. Natural-gradient tests are viewed as research tools, and provide quantitative
information about actual behavior in the field. Forced-gradient tests may be a practical
method for determination of retardation, and they can provide information about nonideal behavior which can facilitate the design of a pump-and-treat extraction system.
Non-ideal behavior, characterized by early breakthrough and prolonged tailing relative to ideal behavior, has important implications for the manner in which contaminated
sites are treated. Early breakthrough suggests that the arrival time of a contaminant at a
point of interest, such as a drinking water well, will be earlier than predicted based on
the ideal model. Tailing has important implications for the operation of pump-and-treat
systems. Because the concentrations in a tail are usually significant relative to drinking
water standards, pumping may need to be continued for much longer than one would
predict based on an ideal desorption model.
Non-ideal behavior has been observed in laboratory and field studies. It has been
attributed to both pore-scale non-equilibrium sorption and to heterogeneities at the field
scale. Recent research has suggested that pore-scale effects may be secondary to fieldscale heterogeneities in causing non-ideal behavior at some sites.
Solution effects on sorption typically act through their effects on solubility. Factors
which increase solubility tend to decrease sorption. Variations in pH, temperature, and
ionic strength usually have only small effects on the extent of sorption. However, the
presence of high concentrations of co-solvents and surfactants can reduce sorption and
thus enhance mobility.
Except under very arid conditions, sorption in the unsaturated zone is controlled
by the same mechanisms as in the saturated zone, with gas/water equilibrium following
Henry's Law. Under very dry conditions, additional sorption mechanisms have been
observed to enhance sorption to the solids by up to three-to-four orders of magnitude
with respect to the same sorbent at field capacity or wetter. The water saturation at which
sorption is enhanced varies with the texture and surface area of the materials.
ACKNOWLEDGEMENTS
260
Chap. 8
8.9 REFERENCES
Abdul, A. S., T. L. Gibson, and D. N. Rai (1990) "Use of humic acid solution to remove organic
contaminants from hydrogeologic systems", Environ. Sci. Technol., 24, 328-333.
Abdul, A. S. and T. L. Gibson (1991) "Laboratory studies of surfactant-enhanced washing of
polychlorinated biphenyl from sandy material", Environ. Sci. Technol., 25, 665-671.
Abdul, A. S., T. L. Gibson, C. C. Ang, J. C. Smith, and R. E. Sobczynski (1992). "In situ surfactant
washing of polychlorinated biphenyls and oils from a contaminated site", Ground Water, 3,
219-231.
Allen-King, R. M., L. M. McKay, and M. R. Trudell (1995) "Organic-carbon dominated trichloroethene sorption in a clay-rich glacial deposit", in preparation.
Anderson, M. R. and J. F. Pankow (1986) "A case study of a chemical spill: Polychlorinated
biphenyls (PCBs) 3. PCB sorption and retardation in soil underlying the site", Water Resources
Research, 22, 1051-1057.
Aronstein, B. N., Y. M. Calvillo, and M. Alexander (1991) "Effect of surfactants at low concentrations on the desorption and biodegradation of sorbed aromatic compounds in soil", Environ.
Sci. Technol., 25, 1728-1731.
Bahr, J. (1989) "Analysis of nonequilibrium desorption of volatile organics during a field test of
aquifer decontamination", J. Contaminant Hydrology, 4, 205-222.
Ball, W. P. (1989) Ph.D. Thesis, Stanford University, Stanford, California.
Ball, W. P. and P. V. Roberts (199Ia) "Long-term sorption of halogenated organic chemicals by
aquifer material. I. Equilibrium", Environ. Sci. Technol., 25, 1223-1236.
Ball, W. P. and P. V. Roberts (199Ib) "Long-term sorption of halogenated organic chemicals by
aquifer material. 2. lntraparticle diffusion", Environ. Sci. Technol., 25, 1237-1249.
Ball, W. P. and P. V. Roberts (199Ic) "Diffusive rate limitations in the sorption of organic chemicals. In: Organic Substances and Sediments in Water, Vol 2., Processes and Analytical, R. A.
Baker (Ed.) Lewis Publishers, Boca Raton, Florida, pp. 273-310.
Ball, W. P. and P. V. Roberts (1992) "Comment on 'Long-term sorption of halogenated organic
chemicals by aquifer material. 1. Equilibrium' ", Environ. Sci. Technol., 26, 2301-2302.
Barber, L. B., E. M. Thurman, M. P. Schroeder, and D. R. LeBlanc (1988) "Long-term fate
of organic micropollutants in sewage-contaminated groundwater", Environ. Sci. Technol., 22,
205-211.
Barber, L. B., E. M. Thurman, and D. D. Runnells (1992) "Geochemical heterogeneity in a sand
and gravel aquifer: Effect of sediment mineralogy and particle size on the sorption of chlorobenzenes", J. Contaminant Hydrology, 9, 35-54.
Barone, F. S., R. K. Rowe, and R. M. Quigley (1992) "A laboratory estimation of diffusion and
adsorption coefficients for several volatile organics in a natural clayey soil", J. Contaminant
Hydrology, 10, 225-250.
Bouchard, D. c., A. L. Wood, M. L. Campbell, P. Nkedi-Kizza, and P. S. C. Rao (1988) "Sorption
nonequilibrium during solute transport", J. Contaminant Hydrology, 2, 209-223.
Boyd, S. A., M. M. Mortland, and C. T. Chiou (1988) "Sorption characteristics of organic compounds on hexadecyltrimethylammonium-smectite", Soil Sci. Soc. Am. J., 52, 652-657.
Brusseau, M. L. and P. S. C. Rao (1989) "Sorption nonideality during organic contaminant transport
in porous media", CRC Crit. Rev. Environ. Control, 16, 33-93.
Sec. 8.9
References
261
Brusseau, M. L. and P. S. C. Rao (1991) "Influence of sorbate structure on nonequilibrium sorption

of organic compounds", Environ. Sci. Techno!., 25, 1501-1506.
Brusseau, M. L., R. E. Jessup, and P. S. C. Rao (1989a) "Rate-limited sorption and nonequilibrium transport of organic chemicals in low organic carbon aquifer materials", Water Resources
Research,25, 1137-1145.
Brusseau, M. L., P. S. C. Rao, R. E. Jessup, and J. M. Davidson (1989b) "Flow interruption: a
method for investigating sorption nonequilibrium", J. Contaminant Hydrology, 4, 223-240.
Brusseau, M. L., R. E. Jessup, and P. S. C. Rao (1990) "Sorption kinetics of organic chemicals:
evaluation of gas-purge and miscible displacement techniques", Environ. Sci. Technol., 24,
727-735.
Brusseau, M. L., R. E. Jessup, and P. S. C. Rao (1991a) "Nonequilibrium sorption of organic
chemicals: elucidation of rate-limiting processes", Environ. Sci. Technol., 26, 134-142.
Brusseau, M. L., A. L. Wood, and P. S. C. Rao (1991b) "Influence of organic cosolvents on the
sorption kinetics of hydrophobic organic chemicals", Environ. Sci. Techno!., 25, 903-910.
Burr, D. (1992) M.Sc. Thesis, University of Waterloo, Waterloo, Ontario, Canada.
Burris, D. R. and C. P. Antworth (1992) "In situ modification of an aquifer material in flowthrough systems by a cationic surfactant to enhance retardation of organic contaminants", J.
Chin, Y.-P., W. J. Weber, Jr., and B. J. Eadie (1990) "Estimating the effects of dispersed organic
polymers on the sorption of contaminants by natural solids. 2. Sorption in the presence of humic
and other natural macromolecules", Environ. Sci. Techno!., 24, 837-842.
Chiou, C. T. (1989) "Theoretical considerations of the partitioning uptake of nonionic organic
compounds by soil organic matter", In: Reactions and Movement of Organic Chemicals in
Soils, B. L. Sawhney and K. Brown (Eds.), Soil Sci. Soc. America Special Publication No. 22,
pp. 1-29.
Chiou, C. T. D. E. Kile, and R. L. Malcolm (1988) "Sorption of vapors of some organic liquids
on soil humic acid and its relation to partioning of organic compounds in soil organic matter",
Environ. Sci. Techno/., 22, 298-303.
Chiou, C. T. and T. D. Shoup (1985) "Soil sorption of organic vapors and effects of humidity on
sorptive mechanisms and capacity", Environ. Sci. Techno!., 16, 1196-1200.
Chiou, C. T., L. 1. Peters, and V. H. Freed (1979) "A physical concept of soil-water equilibria for
nonionic organic compounds", Science, 206, 831-832.
Chiou, C. T., P. E. Porter, and D. W. Schmedding (1983) "Partition equilibria of nonionic organic
compounds between soil organic matter and water", Environ. Sci. Techno/., 17, 227-231.
Chiou, C. T., R. L. Malcom, T. I. Brinton, and D. E. Kile (1986) "Water solubility enhancement
of some organic pollutants and pesticides by dissolved humic and fulvic acids", Environ. Sci.
Techno/., 20, 502-508.
Chiou, C. T., J. F. Lee, and S. A. Boyd (1990) "The surface area of soil organic matter", Environ.
Sci. Techno/., 24, 1164-1166.
Curtis, G. P., P. V. Roberts, and M. Reinhard (1986) "A natural gradient experiment on solute
transport in a sand aquifer. 4. Sorption of organic solutes and its influence on mobility", Water
Enfield, C. G., G. Bengtsson, and R. Lindqvist (1989) "Influence of macromolecules on chemical
transport", Environ. Sci. Techno/., 23, 1278-1286.
262
Chap. 8
Freeze, R. A. and J. A. Cherry (1979) Groundwater, Prentice Hall, Englewood Cliffs, New Jersey.
Garbarini, D. R. and L. W. Lion (1985) "Evaluation of sorptive partitioning of nonionic pollutants
in closed systems by headspace analysis", Environ. Sci. Technol., 19, 1122-1128.
Gerst!, Z. (1990) "Estimation of organic chemical sorption by soils", J. Contaminant Hydrology,
6, 357-375.
Gillham, R. W., M. J. L. Robin, and C. J. Ptacek (1990) "A device for in situ determination of
geochemical transport parameters. 1. Retardation", Ground Water, 28, 666-672.
Goltz, M. N. and P. V. Roberts (1988) "Simulations of physical nonequilibrium solute transport
models: application to a large-scale field experiment", J. Contaminant Hydrology, 3, 37-63.
Grathwol, P. (1990) "Influence of organic matter from soils and sediments from various origins
on the sorption of some chlorinated aliphatic hydrocarbons: Implications on K oc correlations",
Gschwend, P. M. and S. Wu (1985) "On the constancy of sediment-water partition coefficients of
hydrophobic organic pollutants", Environ. Sci. Technol., 19, 90-96.
Hausch, C. and A. Leo (1979) Substituent Constants for Correlation Analysis in Chemistry and
Biology, John Wiley, New York, New York.
Hassett, J. J. and W. L. Banwart (1989) "The sorption of nonpolar organics by soils and sediments",
In Reactions and Movement of Organic Chemicals in Soils, B. L. Sawhney and K. Brown (Eds.)
Soil Sci. Soc. America Special Publication No. 22., pp. 21-43.
Holsen, T. M., E. R. Taylor, Y-C. Seo, and P. R. Anderson (1991) "Removal of sparingly soluble
organic chemicals from aqueous solutions with surfactant-coated ferrihydrite", Environ. Sci.
Technol., 25,1585-1589.
Jafvert, C. T., 1. C. Westall, E. Grieder, and R. P. Schwarzenbach (1990) "Distribution of hydrophobic ionogenic compounds between octanol and water: Organic acids", Environ. Sci. Technol.,
24, 1795-1803.
Johnson, C. A. and J. C. Westall (1990) "Effect of pH and KCI concentration on the octanol-water
distribution of methylanilines". Environ. Sci. Technol., 24,1869-1875.
Johnson, R. L., J. A. Cherry, and J. F. Pankow (1989) "Diffusive contaminant transport in natural
clay: a field example and implications for clay-lined waste disposal sites", Environ. Sci. Technol.,
23, 340-349.
Karickhoff, S. W., D. S. Brown, and T. A. Scott (1979) "Sorption of hydrophobic pollutants on
natural sediments", Water Research, 13, 241-248.
Karickhoff, S. W. (1980) "Sorption kinetics of hydrophobic pollutants in natural sediments". In:
Contaminants and Sediments, Volume 2, R. A. Baker (Ed.), Ann Arbor Science, Ann Arbor,
Michigan, 193-205.
Karickhoff, S. W. (1981) "Semi-empirical estimation of sorption of hydrophobic pollutants on
natural sediments and soils", Chemosphere, 10, 833-846.
Karickhoff, S. W. (1984) "Organic pollutant sorption in aquatic systems", J. Hydraulic Eng., 110,
707-735.
Kile, D. E. and C. T. Chiou (1989) "Water solubility enhancements of DDT and trichlorobenzene
by some surfactants below and above the critical micelle concentration", Environ. Sci. Technol.,
23, 832-836.
Sec. 8.9
References
263
Lee, L. S., P. S. C. Rao, M. L. Brusseau, and R. A. Ogwada (1988) "Nonequilibrium sorption

of organic contaminants during flow through columns of aquifer materials", Environ. Toxicol.
Chem., 7, 779-793.
Lee, L. S., P. S. C. Rao, P. Nkedi-Kizza, and J. J. Delfino (1990) "Influence of solvent and sorbent
characteristics on distribution of pentachlorophenol in octanol-water and soil-water systems",
Environ. Sci. Techno!., 24, 654-661.
Lee, L. S., P. S. C. Rao, and M. L. Brusseau (1991) "Nonequilibrium sorption and transport of
neutral and ionized chlorophenols", Environ. Sci. Techno!., 25, 722-729.
Lion, L. W., T. B. Stauffer, and W. G. MacIntyre (1990) "Sorption of hydrophobic compounds on
aquifer materials: analysis methods and the effect of organic carbon", J. Contaminant Hydrology,
5, 215-234.
MacIntyre, W. G., T. B. Stauffer, and C. P. Antworth (1991) "A comparison of sorption coefficients
determined by batch, column and box methods on a low organic carbon aquifer material",
Ground Water, 29, 908-913.
MacIntyre, W. G. and C. L. Smith (1984) "Comment on 'Partition equilibria of nonionic organic
compounds between soil organic matter and water' ". Environ. Sci. Technol., 18, 295.
Mackay, D. M., D. L. Freyberg, and P. V. Roberts (1986a) "A natural gradient experiment on solute
transport in a sand aquifer. I. Approach and overview of plume movement", Water Resources
Research, 22, 2017-2029.
Mackay, D. M., W. P. Ball, and M. G. Durant (1986b) "Variability of aquifer sorption properties
in a field experiment on groundwater transport of organic solutes: methods and preliminary
results", J. Contaminant Hydrology, 1, 119-132.
Mackay, D. M. (1990) "Characterization of the distribution and behavior of contaminants in the
subsurface", In: Ground Water and Soil Contamination Remediation: Toward Compatible Science, Policy, and Public Perception, National Academy Press, Washington, D.C., pp. 70-90.
Mackay, D. M. and K. W. Thorbjarnson (1990) Flushing of Organic Contaminants from a Ground
Water Plume at the Rocky Mountain Arsenal: Field and Laboratory Studies. UCLA School of
Public Health, Environmental Science and Engineering Program Technical Report No. 90-69,
Volume I (report) and Volume II (appendices).
Mackay, D. M., G. C. Bianchi-Mosquera, A. Kopania, H. Kianjah and K. Thorbjarnson (1994) "A
forced-gradient experiment on solute transport in the Borden aquifer. I. Experimental methods
and moment analyses of results", Water Resources Research, 30, 369-383.
Mackay, D. M. and J. A. Cherry (1989) "Groundwater contamination: Pump-and-treat remediation",
Environ. Sci. Techno!., 23, 630-636.
Magee, B. R., L. W. Lion, and A. T. Lemley (1991) "Transport of dissolved organic macromolecules and their effect on the transport of phenanthrene in porous media", Environ. Sci.
Techno!., 25, 323-331.
McGinley, L. E. Katz, and W. J. Weber, Jr. (1993) "A distributed reactivity model for sorption
by soils and sediments. 2. Multicomponent systems and competitive effects", Environ. Sci.
Technol., 27, 1524-1531.
Mehran, M. M., R. L. Olsen, and B. M. Rector (1987) "Distribution coefficient of trichloroethylene
in soil-water systems", Ground Water, 25, 275-282.
Miheicic, 1. R. and R. G. Luthy (1991) "Sorption and microbial degradation of naphthalene in
soil-water suspensions under denitrification conditions", Environ. Sci. Techno!., 25,169-177.
264
Chap. 8
Minglegrin, U. and Z. Gerst! (1983) "Reevaluation of partitioning as a mechanism of nonionic

chemicals adsorption in soils", J. Environ. Qual., 12, 1-11.
Montgomery, J. H. and L. M. Welkom (1990) Groundwater Chemicals Desk Reference Lewis
Publishers, Boca Raton, Florida.
Murphy, E. M., J. M. Zachara, and S. S. Smith (1990) "Influence of mineral-bound humic substances on the sorption of hydrophobic organic compounds", Environ. Sci. Technol., 24, 15071516.
Myrand, D., R. W. Gillham, E. A. Sudicky, S. F. O'Hannesin, and R. L. Johnson (1992) "Diffusion
of volatile organic compounds in natural clay deposits", J. Contaminant Hydrology, 10, 159-177.
Nkedi-Kizza, P., J. W. Biggar, H. M. Selim, M. Th. van Genuchten, P. J. Wierenga, J. M. Davidson,
and D. R. Nielson (1984) "On the equivalence of two conceptual models for describing ion
exchange during transport through an aggregated oxisol", Water Resources Research, 20, 11231130.
Nkedi-Kizza, P., P. S. C. Rao, and A. G. Hornsby (1985) "Influence of cosolvents on sorption of
hydrophobic organic chemicals by soils", Environ. Sci. Technol., 19, 975-979.
Nkedi-Kizza, P., P. S. C. Rao, and A. G. Hornsby (1987) "Influence of organic cosolvents on
leaching of hydrophobic organic chemicals through soils", Environ. Sci. Technol., 21, 11071111.
Nkedi-Kizza, P., M. L. Brusseau, P. S. C. Rao, and A. G. Hornsby (1989) "Nonequilibrium
sorption during displacement of hydrophobic organic chemicals and 45Ca through soil columns
with aqueous and mixed solvents", Environ. Sci. Technol., 23, 814-820.
Ogram, A. Y., R. E. Jessup, L. T. Ou, and P. S. C. Rao (1985) "Effects of sorption on biological
degradation rates of 2,4-dichlorophenoxyacetic acid in soils", Appl. Environ. Microbiol., 49,
582-587.
Ong, S. K. and L. W. Lion (l99lb) "Mechanisms for trichloroethylene vapor sorption onto soil
minerals", J. Environ. Qual., 20, 180-188.
Ong, S. K. and L. W. Lion (l991a) "Effects of soil properties and moisture on the sorption of
trichloroethylene vapor", Water Research, 25, 29-36.
Pavlostathis, S. G. and K. Jaglal (1991) "Desorptive behavior of trichloroethylene in contaminated
soil", Environ. Sci. Technol., 25, 274-279.
Perlinger, J. A. and S. J. Eisenreich (1991) "Sorption of alkylbenzenes to mineral oxides". In:
Organic Substances and Sediments in Water, Vol 2., Processes and Analytical, R. A. Baker
(Ed.), Lewis Publishers. Boca Raton, Florida, pp. 49-78.
Perlinger, J. A., S. J. Eisenreich, and P. D. Capel (1993) "Application of headspce analysis to the
study of sorption of hydrophobic organic chemicals to o -Al-O;", Environ. Sci. Technol., 27,
928-937.
Peterson, M. S., L. W. Lion, and C. A. Shoemaker (1988) "Influence of vapor-phase sorption
and diffusion on the fate of trichloroethylene in an unsaturated aquifer system", Environ. Sci.
Technol., 22, 571-578.
Pignatello, J. J., F. J. Ferrandino, and L. Q. Huang (1993) "Elution of aged and freshly added
herbicides from a soil", Environ. Sci. Technol., 27, 1563-1571.
Pignatello, J. J., C. R. Frink, P. A. Marin, and E. X. Droste (1990) "Field-observed ethylene
dibromide in an aquifer after two decades", J. Contaminant Hydrology, 5, 195-214.
Piwoni, M. D. and P. Banerjee (1989) "Sorption of volatile organic solvents from aqueous solution
onto subsurface solids", J. Contaminant Hydrology, 4, 163-179.
Sec. 8.9
References
265
Ptacek, C. J. and R. W. Gillham (1992) "Laboratory and field measurements of nonequilibrium

transport in the Borden Aquifer", J. Contaminant Hydrology, 10, 119-158.
Rao, P. S. C., L. S. Lee, and R. Pinal (1990) "Cosolvency and sorption of hydrophobic organic
chemicals", Environ. Sci. Technol., 24, 647-654.
Rhue, R. D., P. S. C. Rao, K. D. Pennell, and W. H. Reve (1989) "Competitive adsorption of
alkyl benzene and water vapors on predominantly mineral surfaces", Chemosphere, 18, 19711986.
Roberts, P. V., M. N. Goltz, and D. M. Mackay (1986) "A natural gradient experiment on solute
transport in a sand aquifer. 3. Retardation estimates and mass balances for organic solutes",
Water Resources Research, 22, 2046-2058.
Ryan, P. A. and P. Gschwend (\990) "Colloidal mobilization in two Atlantic Coastal Plain aquifers:
Field Studies", Water Resources Research, 26, 307-322.
Sabljic, A. (1987) "On the prediction of soil sorption coefficients of organic pollutants from molecular structure: Application of a molecular topology model", Environ. Sci. Technol., 21, 358-366.
Schwarzenbach, R. P., W. Giger, W. Hoehn, and J. K. Schneider (1983) "Behavior of organic compounds during infiltration of river water to groundwater. Field Studies", Environ. Sci. Technol.,
17,472-479.
Schwarzenbach, R. P. and J. Westall (1981) "Transport of nonpolar organic compounds from surface
water to groundwater. Laboratory sorption studies", Environ. Sci. Technol., 15, 1360-1366.
Shearer, R. C., J. Letey, W. J. Farmer, and A. Klute (1973) "Lindane diffusion in soil", Soil Sci.
Soc. Amer. Proc., 37, 189-193.
Smith, J. A., C. T. Chiou, 1. A. Krammer, and D. E. Kile (l990a) "Effect of soil moisture on
the sorption of trichloroethene vapor to vadose-zone soil at Picatinny Arsenal, New Jersey",
Smith, J. A., P. R. Jaffe, and C. T. Chiou (l990b) "Effect of ten Quaternary ammonium cations
on tetrachloromethane sorption to clay from water", Environ. Sci. Technol., 24, 1167-1172.
Spencer, W. F. and M. M. Cliath (1970) "Desorption of lindane from soil as related to vapor
density", Soil Sci. Soc. Amer. Proc., 34, 574-578.
Spurlock, F. C. and J. W. Biggar (1990) "Effect of naturally occurring soluble organic matter on
the adsorption and movement of simazine [2-chloro-4,6-bis(ethylamino)-s-triazine] in Hanford
sandy loam", Environ. Sci. Technol., 24, 736-741.
Stauffer, T. B. and W. G. MacIntyre (1986) "Sorption of low-polarity organic compounds on oxide
minerals and aquifer material", Environ. Toxieol. Chem., 5, 949-955.
Steen, W. c., D. F. Paris, and G. L. Baughman (1980) "Effects of sediment sorption on microbial
degradation of toxic substances", In: Contaminants and Sediments, Vall., R. A. Baker (Ed.),
Ann Arbor Science, Ann Arbor, Michigan, pp. 477-482.
Valocchi, A. J. (1985) "Validity of the local equilibrium assumption for modeling sorbing solute
transport through homogeneous soils", Water Resources Research, 21, 808-820.
Weber, W. J., Jr., P. M. McGiney, and L. E. Katz (1992) "A distributed reactivity model for
sorption by soils and sediments. I. Conceptual basis and equilibrium assessments", Environ.
Sci. Technol., 26, 1955-1962.
Whiffin, R. B. and J. M. Bahr (1985) "Assessment of purge well effectiveness for aquifer decontamination", In: Proceedings: Fourth National Symposium on Aquifer Restoration and Ground
Water Monitoring, May 23-25, 1984, NWWA, Worthington, Ohio, 75-81.
266
Chap. 8
Wilson, J. T., C. G. Enfield, W. J. Dunlop, R. L. Cosby, D. A. Foster, and L. B. Baskin (1981)

"Transport and fate of selected organic pollutants in a sandy soil", J. Environ. Qual., 10, 501506.
Woodburn, K. B., P. S. C. Rao, M. Fukui, and P. Nkedi-Kizza (1986) "Solvophobic approach
for predicting sorption of hydrophobic organic chemicals on synthetic sorbents and soils", J.
Wu, S. and P. M. Gschwend (1986) "Sorption kinetics of hydrophobic organic compounds to
natural sediments and soils", Environ. Sci. Technol., 20, 717-725.
9
Chemical and Microbiological
Transformations and
Degradation of Chlorinated
Solvent Compounds
Barbara J. Butler and James F. Barker
Waterloo Centre for Groundwater Research
University of Waterloo,
Waterloo, Ontario, Canada, N2L 3Gl
Page
9.1
9.2
9.3
9.4
9.5
9.6
Introduction
Chemical Reactions
9.2.1 Overview
9.3.1 General
9.4.1 General
9.4.3.1 General
9.4.3.2 Haloalkanes
9.4.3.3 Haloalkenes
9.4.5 Field Studies
Conclusions
References
268
269
269
270
270
272
273
276
278
278
278
280
283
283
283
286
286
286
288
290
299
300
303
304
267
268
Chemical/Microbiological Transformations and Degradation
Chap. 9
ABSTRACT
Chemical and microbial transformations of common haloaliphatic solvent compounds

provide considerable potential for natural attenuation and remediation in the subsurface.
Aerobically, abiotic hydrolysis rates are generally of little consequence, though there
are significant exceptions (e.g., 1,1, I-TCA). Anaerobically, on the other hand, reductive
dechlorination can occur under strongly reducing conditions, especially in the presence
of elemental iron. When sulfide is present, abiotic substitution of sulfide for aliphatic
halogens (especially bromine) can occur. Generally, then, it can be said that reducing
conditions hold the most potential for providing for abiotic transformations of solvent
aliphatics.
In nature, subsurface microbes are likely involved in most transformations, either indirectly by creating chemical conditions that are favorable for abiotic reactions,
or directly by conducting the transformations themselves. In general, compounds that
are more highly halogenated are more prone to being transformed by reductive dehalogenation. Compounds with few halogens, such as vinyl chloride, may be more easily
degraded under aerobic conditions. Nevertheless, the complete anaerobic degradation
of both trichloroethylene (TCE) and tetrachloroethylene (PCE) to ethylene has been
observed. Co-metabolic utilization by aerobic, methane oxidizers also provides some
potential for aerobic bioremediation of many chlorinated solvent compounds. Close to
DNAPL sources, inhibition of biotransformations due to high contaminant concentrations
is likely to be a problem.
9.1 INTRODUCTION
The behavior of the chlorinated solvent compounds in the field is the net result of
various processes operating in concert. These processes include advection, dispersion,
sorption, chemical transformation, biotransformation, and perhaps vaporization. Many
halocarbons do not usually react abiotically in groundwater environments. As a result,
biotransformation reactions are important for their natural removal from contaminated
systems. Indeed, in some situations, biotransformation of the original "parent" chlorinated
solvent compounds has been implicated as the mechanism by which other "daughter"
compounds come to pollute groundwaters (Parsons et al., 1984; Cline and Viste, 1985;
Lesage et al., 1990; Major et al., 1991). In the field, these occurrences may sometimes
serve a useful predictive purpose, because prudent selection of the parameters analyzed
in groundwater samples and sediment cores can allow one to characterize a contaminated
site not only in hydrogeological and geochemical terms but also in microbiological terms.
Together, this information may be used to predict behavior of the contaminant plume
and may also aid in selection of the most feasible remediation strategy.
This chapter will generally be restricted to a discussion of the haloaliphatic chlorinated solvents. We will begin with a discussion of the chemical reactions of halocarbons,
emphasizing the common chlorinated solvents. A short overview of the microbial ecology
of subsurface systems and the general processes of biotransformation and biodegradation
Sec. 9.2
Chemical Reactions
269
will follow. The remainder of the chapter will concern biotransformations of haloaliphatic
compounds.
9.2 CHEMICAL REACTIONS
9.2.1 Overview
Many halocarbons have the potential for chemical (abiotic) transformations and microbial transformations. The realization of this potential depends upon the halocarbon and
the groundwater environment. This section reviews the most likely chemical transformations in groundwaters. Many of the same reaction pathways are catalyzed by microbes.
In many, if not most cases, the microbial reaction rate will be faster than the abiotic rate
(e.g., Jafvert and Wolfe, 1987). In addition, other microbial mechanisms may produce
transformations of a contaminant when no abiotic reactions are likely.
Schwarzenbach et al. (1985) offer strong evidence of abiotic substitution reactions occurring in actual groundwater systems. Laboratory evidence for abiotic reductive
dehalogenation is compelling (Gillham and O'Hannesin, 1992; Barbash and Reinhard,
1989; Reinhard et al., 1990). Controlled field experiments support these findings (Acton, 1990), but conclusive field evidence is often elusive. The most significant transformation reactions have been reviewed by Baxter (1989) and Vogel et al. (1987). The
nomenclature of Vogel et al. (1987) is followed below. Four reaction types have been
recognized:
1. Substitution reactions include hydrolysis, for which water serves as the nucleophile, and substitution, for which other nucleophiles such as sulfide attack the
halocarbon. In either case, the nucleophile (OH- or S-) replaces the halogen (X-).
2. Dehydrohalogenation involves the elimination of HX and the formation of a C=C
double bond in place of a C-C single bond (i.e., an alkene forms at the expense of
an alkane).
3. Oxidation is a common reaction for organics, given the rather reduced state of
carbon in organic compounds. Since the presence of a halogen substituent in an
organic compound makes for a more oxidized compound than would be the case
in the absence of the halogen, the potential for oxidation (as by oxygen) decreases
as the extent of halogenation increases. Various oxidation reactions are possible.
Epoxidation may be the most significant, and it involves the addition of oxygen
across the C=C bond of an alkene like TCE. The halogenated epoxide is usually
reactive, producing halogenated aldehydes and acyl halides. Subsequent transformation products (e.g., halogenated acids and alcohols, CO 2 , and carbon monoxide)
are usually short-lived in aqueous environments. No abiotic oxidations of typical halocarbon solvents have been reported, although microbial oxidations are
found.
4. Reduction reactions are probably dominated by hydrogenolysis and dihalo-elimination. The former involves the breaking of a carbon-halogen bond with hydrogen then replacing the halogen. The latter involves the reductive removal of two
270
Chap. 9
halogens, typically producing a C=C double bond from a C-C single bond (i.e.,
producing an alkene from an alkane). Reductions are commonly microbially-mediated, but abiotic reductions have been observed.

Hydrolysis and dehydrohalogenation are probably the most thoroughly studied abiotic
transformations of the halocarbons (see Table 9.1). Early work by Dilling et al. (1975),
the review and compilation by Mabey and Mill (1978), the predictive methodology of
Lyman et al. (1982), the review of Vogel et al. (1987), and the recent experiments by
Jeffers et al. (1989) summarize the most significant information. Hydrolysis reactions are
usually first order (i.e., dependent on the concentration of halocarbon), but are usually
composed of a neutral reaction and a base-catalyzed reaction that are individually
first order. The latter is pH dependent and, when it is important, the overall hydrolysis
rate also depends on the concentration of the hydroxide ion. In some cases, the basecatalyzed reaction is faster than the neutral reaction even at neutral pH (e.g., as with
pentachloroethane and 1,I ,2,2-tetrachloroethane in Table 2 of Jeffers et al., 1989). Our
discussions here will focus on relative hydrolysis rates, with the total hydrolysis rate at
pH = 7 serving as the basis for comparison; the actual hydrolysis reaction mechanisms
will not be considered.
TABLE 9.1 Abiotic Hydrolysis (Substitution and Dehydrohalogenation) Half-Lives at 10 to 25C of
Halocarbons (Best Values).
Half-life
in years (ref)
Compound
Methanes
Chloromethane
Dichloromethane
Trichloromethane
Tetrachloromethane
Bromomethane
Dibromomethane
Tribromomethane
Bromochloromethane
Compound
Half-life
in years (ref)
Ethanes
no data
704 (1)
3500 (1), 1800 (3)
41 (3)
0.10 (1)
180 (1)
690 (1)
44 (1)
140 (1)
270 (1)
Chloroethane
I,I-Dichloroethane
1,2-Dichloroethane
1,1,I-Trichloroethane
I, 1,1,2-Tetrachloroethane
1,1,2,2-Tetrachloroethane
Pentachloroethane
Hexachloroethane
Bromoethane
1,2-Dibromoethane
0.12 (4)
61 (3)
72 (3)
1.1(3), 1.7 (5), 2.5 (6)
140 (3), 170 (5)
47(3), 380 (5)
0.3 (8), 0.4 (3), 0.8 (5)

0.01 (3, 5)
1.8 x 109 (3)
0.08 (1)
2.5 (7)
Ethylenes
1,1-Dichloroethylene
Trichloroethylene
Tetrachloroethylene
1,2-Dichloroethylene
1.2
0.9
0.7
2.1
x 108 (3)
(2), 1.3 x 106 (3)
(2), 1.3 x 106 (3)
x 1010 (3)
References: (1) Mabey and Mill (1978); (2) Dilling et al. (1975); (3) Jeffers et al. (1989); (4) Vogel et al.
(1987); (5) Mabey et al. (1983); (6) Vogel and McCarty (1986); (7) Weintraub et al. (1986); (8) Cooper et
al. (1987).
Sec. 9.2
271
Chemical Reactions
A considerable amount of laboratory data for homogeneous substitution reactions

(i.e., in aqueous solution, without solid surface catalysts) has been produced at elevated
temperatures of 40 to 200 DC; the slow rate of reaction at environmental temperatures

(0 to 30 DC) does not permit reasonable experiment times at ambient temperatures. The
rate constant k at an environmental temperature (T2) is usually extrapolated from the
measured rate at an elevated temperature (T]) using the Arrhenius equation,
(9.1)
where temperature (1) is in degrees Kelvin caC + 273.15), Ea is the activation energy
(typically about 80 to 120 kJ/mol),and R is the gas constant, 8.314 J/mol-K. For typical
activation energies, the reaction rate will decrease dramatically with decreasing temperature, about 3 to 4 fold for each 10 "C decrease. Before proceeding, we should also note
that the reaction mechanisms and products observed at elevated temperatures may not
always be representative of what occurs at ambient groundwater temperatures.
Hydrolysis will produce an alcohol from an alkane or alkene:
R-X
+ H2 0
---+
R-OH
+ H-X
(9.2)
where R is the part of the organic molecule unaffected by the reaction, -X is a halogen
substituent, and -OH is the alcohol functional group. Non-alcohol products have also been
found (Vogel et al., 1987; Jeffers et al., 1989). In some cases, these other products are further transformation products, but in many cases they are products of one or more competing dehydrohalogenation reactions. 1,2-dichloroethane has been found to yield some ethylene glycol (Jeffers et al., 1989), as has 1,2-dibromoethane (Vogel et al., 1987); ethylene
glycol represents a classic replacement of both halogen atoms in these molecules by OH.
Chloroethane has been found to yield chloroethylene, also known as vinyl chloride
(Jeffers et al., 1989), and 1,2-dibromoethane has been found to yield bromoethylene (see
Table 3 of Vogel et al., 1987), indicating that dehydrohalogenation can occur according to
CH 2X-CH2X ---+ CH 2
= CH 2X + HX,
= Cl
or Br.
(9.3)
Vogel et at. (1987, see their Table 3) indicate that such dehydrohalogenation reactions
are common. Jeffers et at. (1989) found that pentachloroethane yielded tetrachloroethylene (PCE). Roberts and Gschwend (1991) found the same reaction to be rapid, and
noted that the reaction of hexachloroethane to PCE could proceed, in large part, via
pentachloroethane.
Considerable discrepancies often arise when comparing older reviews and recent
experimental studies. For example, compared to the study of Dilling et at. (1975), Jeffers
et at. (1989) reported a much shorter ambient half-life (t1/2) for carbon tetrachloride (41
VS. 7000 years), and the opposite for TCE and PCE. Due to the experimental difficulties encountered by Dilling et at. (1975), the more recent data of Jeffers et at. (1989)
are usually accepted as correct. Where significant conflicts exist, the data from Dilling
272
Chemical! Microbiological Transformations and Degradation
Chap. 9
et al. (1975) do not appear in Table 9.1. The reader is cautioned to treat all half-lives as
approximate.
Bromomethanes and bromochloromethanes react significantly faster than the corresponding chloromethanes. The chloromethanes show an increasing half-life as chlorination increases from 1 to 3 chlorines, but carbon tetrachloride has the shortest half-life of
the chloromethanes; Jeffers et al. (1989) suggest some combination of steric and energetic
effects may be responsible for these non-uniform trends in reactivity.
Like the halomethanes, the haloethanes show a large range in reaction rates,
with the bromo- and dibromoethanes reacting significantly faster than their chlorinated
analogs. The reaction rate is faster when the chlorines are on the same carbon, although the trend with increasing overall chlorination is inconsistent. The very reactive (t1/2 < 3 years) haloethanes include chloroethane, bromoethane, 1,2-dibromoethane,
1,1,1-trichloroethane, 1,1,2,2-tetrachloroethane and pentachloroethane. Chloroethane, the
dichloroethanes, and 1,1, I-trichloroethane were found to produce alcohols (Jeffers et
al., 1989). However, Bouwer and McCarty (1983a) do report the abiotic production of
acetic acid from 1,1,l-trichloroethane. 1,1,2-trichloroethane, tetrachloroethanes, and pentachloroethane were found to dehydrohalogenate to 1,2-dichloroethylene, TCE and PCE,
respectively. For the chloroethanes, alcohols are probably the result of neutral hydrolysis,
while dehydrohalogenation appears to be linked to alkaline hydrolysis and becomes more
important as the degree of chlorination increases (Jeffers et al. 1989).
The chloroethylenes do not undergo significant hydrolysis reactions. Unfortunately,
no studies for vinyl chloride (chloroethylene) were found. Although this compound may
not be common as an original cause of contamination, its presence has been reported in
many dissolved chlorinated solvent plumes. Since its drinking water standard is among the
lowest for the halocarbons (see Table 1.13), the chemistry of formation and degradation
of vinyl chloride is of considerable interest.

Schwarzenbach et al. (1985) provided both field and laboratory evidence that the HSor H2S that can be present under anaerobic conditions can act as stronger nucleophiles
than water. Substitution reactions of the type
R-X
+ HS-
-+
R-S-H
+ X-
(9.4)
are therefore possible. Apparently, only haloalkanes are affected. In most groundwaters,
reduced sulfur is a product of microbial sulfate reduction, and so the indirect, if not
also the direct role of microbes is certain. Possible products include dialkyl sulfides,
mercaptans, thiols and dithiols. Roberts et al. (1992) reported that dihalomethanes reacting
with HS- produce mainly poly(thiomethylene) (HS-(CH2-S)n-H) along with a minor
amount of dithiomethane (CH 2(SHh).
Schwarzenbach et al. (1985), Barbash and Reinhard (1989), and others have demonstrated that reaction (9.4) can occur abiotically, but that its reaction rate can be enhanced by microbial catalysts. Abiotic half-lives at 25C for 1,2-dichloroethane and
1,2-dibromoethane were observed to be about 6 years and 40 to 70 days respectively
(Barbash and Reinhard, 1989). Kriegman-King and Reinhard (1992) demonstrated that
Sec. 9.2
Chemical Reactions
273
minerals increased the rate of abiotic transformation of carbon tetrachloride to CS z in

the presence of HS-; the transformation mechanism was not clear. Both Haag and Mill
(1988) and Roberts et al. (1992) note that polysulfides, which are expected when elemental sulfur is present under reduced conditions at pH 7 and higher, react much faster
with haloalkanes than HS-. Hydrolysis and sulfide substitution reactions can occur concurrently, as has been evidenced by the presence of both alcohols from hydrolysis and
sulfur-bearing hydrocarbons from sulfide substitution.
As noted above, two important reductive dechlorination reactions are hydrogenolysis

and dihaloelimination. The former involves the simple replacement of a halogen by a
hydrogen, and the latter involves the removal of two halogens with the formation of an
additional C=C double bond.
Abiotic reductive dechlorination reactions have been reported in a number of circumstances, These include:
1. Hexachloroethane and carbon tetrachloride were transformed to TCE and chloroform, respectively, in the presence of reductants such as HS-, L-cysteine, and
Fe2+, in the presence of zeolite, biotite, vermiculite, pyrite, and marcasite, and
in the presence of the Fe-, Mn-, organic matter-rich fraction of Borden aquifer
material (Reinhard et al., 1990);
2. Wang and Tan (1990) reported the reduction of TCE, bromoform, and carbon
tetrachloride to ethane, methane, and methane, respectively, during aPt-catalyzed
reaction between elemental magnesium and water, producing Hz; and
3. A number of biochemical systems were found to have this effect, including vitamin
B12 (Co), coenzyme F43D and hematin (Fe) (Gantzer and Wackett, 1991); and
4. Systems involving FeD have been used to reductively dehalogenate carbon tetrachloride, chloroform, TCE, and PCE, with half-lives of 2 to 4000 minutes (O'Hannesin
and Gillham, 1992).
A useful approach for evaluating the possibility of reductive dehalogenation was
presented by Vogel et al. (1987, see their Figure 6). Testing for a favorable free energy
change for the overall redox reaction can be easily carried out visually for various
combinations of oxidations and reductions. Higher halogenation produces higher relative
reduction potentials and the potential for release of more free energy during reduction.
Both hexachloroethane and 1,2-dibromoethane have higher electrochemical reduction
potentials than oxygen, and so are theoretically reducible under typical aerobic conditions.
Essentially all reductive dehalogenation reactions are energetically favored when coupled
with typical inorganic (e.g., Hz to H+, and Fe2+ to Fe(OH))) and biological (ferredoxin.g,
to ferredoxinj., NADH to NAD+) redox couples.
Following the method of Vogel et al. (1987), we have calculated the free energy
change per mol of electrons transferred for a single dechlorination step (hydrogenolysis),
274
Chap. 9
i.e., for the reaction

!RCI + !H+
+ e:
---+ !RH + !Cl-
(9.5)
The free energy yields were calculated at 25C, for water with a pH of 7.0, and at a
Cl: concentration of 10- 3 M. The results are presented in Figure 9.1. In all cases, there
is a favorable free energy yield, indicating that the dechlorination reaction is, in general,
thermodynamically favored under the assumed conditions.
The free energy yields for selected chlorinated halocarbons undergoing a single
dechlorination reduction were also calculated when coupled to each of the following
oxidations: Fe o to Fe 2+, Mn o to Mn 2+, S2- to So, and CH 4 (methane) to HC0 3. The
overall free energy yields per mol of electron transferred were calculated at 25 DC for
water at a pH of 7.0, with cr, Fe2+, Mn 2+, S2-, CH 4 , and HC0 3, each present at 10- 3
M. Figure 9.2 gives the free energy yields for the sequential dechlorination of tetrachloroethylene (PCE) to ethylene (EE), and for tetrachloroethane (TCEA) to trichloroethane
(TCA), and carbon tetrachloride (CT) to chloroform (CF). As seen in Figure 9.2, in
all cases, there is a free energy gain, indicating that each of these overall reactions is
thermodynamically favored for the specified conditions. Note that Feo oxidation produces
the most free energy. This is consistent with the success that has been observed with using
Feo oxidation to abiotically dechlorinate some solvents (Gillham and O'Hannesin, 1992).
In the sequence of PCE to ethylene, it is interesting to note that thermodynamically, the
least favored reaction is trans-I,2-dichloroethylene (12tDCE) to vinyl chloride (VC),
This would suggest that trans-l,2-dichloroethylene might be a more commonly-found
product than vinyl chloride.
The above analysis indicates that most chlorinated solvents are susceptible to reduction reactions. However, this potential is rarely met except under very reducing conditions, and usually not without mediation by bacteria. Pignatello (1986) reported that aerobic degradation of 1,2-dibromoethane occurred only with microbial mediation. Jafvert and
Wolfe (1987) found that hexachloroethane, 1,1,2,2-tetrachloroethane, 1,2-diiodoethane,
1,2-dibromoethane, but not 1,2-dichloroethane were rapidly degraded (within days) in
anoxic sediment microcosms. The reactions were faster in biologically active systems,
but were significant in inhibited systems. This indicates that the reactions are possible
abiotically, but that microbial activity can enhance the reaction rates. Products from hexachloroethane included PCE; ethylene was generated from dibromo- and diiodoethane.
Criddle et al. (1986) also reported loss of hexachloroethane from anaerobic microcosms
made from Borden sand. Reexamining their data reveals that sterile controls showed about
the same rate of mass loss as did non-sterile microcosms, implicating an abiotic reaction
(dihaloelimination), Similar abiotic reduction reactions in homogeneous and heterogeneous systems consisting of minerals and Borden aquifer material have recently been
observed and may be closely coupled to microbially-produced reductants such as ferrous
iron and sulfides, and perhaps mediated by humic acids in situ (Reinhard et al., 1990).
Acton (1990) noticed a slight increase in TCE along with significant losses of PCE after
about 21 days in anaerobic Borden sand microcosms, and in anaerobic field columns.
This could be a result of hydrogenolysis of PCE to TCE, again perhaps abiotically since
it proceeded in sterile controls and also in a field column treated with formaldehyde.
Sec. 9.2
275
Chemical Reactions
b. Sequence:
a. Sequence: HCA to EA
CT to M
60
20
40
80
o
CT to CF,-------.------r---,-_,
020406080
HCA to PCAr-----r----,----,-----.--,
PCA to 1112TCEA
CF to MC
PCA to 1122TCEA
1112TCEA to 111TCA
111TCA to 11DCA
MC to CM
11DCA to CA
CA to EA L------'~----L-__'__---'
o 20 40 60 80
CM to M
ENERGY YIELD (kJ/moll
20
40
60
ENERGY YIELD IkJ/mol1
c. Sequence: PCE to EE
PCE to TCE
20
TCE to 11DCE
TCE to 12cDCE
TCE to 12tDCE
11DCE to VC
12cDCE to VC
12tDCE to VC
VC to EE
20
40
60
80
ENERGY YIELD {kJ/moll
Figure 9.1 Free energy yield per mol e: for single dechlorinations of chloroethanes,
chloromethanes, and chloroethylenes, where aqueous species are 10- 3 M, pH =7 following the method of Vogel et al. (1987). a. Sequence for hexachloroethane (HCA) to ethane
(EA); PCA = pentachloroethane; 1112TCEA = 1,l,l,2-tetrachloroethane; 1122TCEA
= 1,1,2,2-tetrachloroethane; 111TCA = l,l,l-trichloroethane; llDCA = l,l,-dichloroethane; CA = chloroethane; EA = ethane. b. Sequence for carbon tetrachloride (CT) to
methane (M); CF = chloroform; MC = methylene chloride; CM = chloromethane; M =
methane. c. Sequence for tetrachloroethylene (PCE) to ethylene (EE); TCE = trichloroethylene; llDCE = l,l-dichloroethylene; 12cDCE = cis-I,2-dichloroethylene; 12tDCE =
trans-I,2-dichloroethylene; VC = vinyl chloride; EE = ethylene.
80
276
PCE to TC E r - -
50
100
150
Chap. 9
200
----,-+-----,-_---,-~_____,
TCE to 12tDCE
12tDCE to VC
VC to EE
1112TCEA to 111 TCA
CT to CF
L~~_--'----I----'_-----i>--1-~--'
50
100
150
200
ENERGY YIELD (kJ/mol)

....... Fe to Fe2 +
.......
Mn to Mn 2 +
-+- 52. to 5
......
M to HC03
Figure 9.2 Free energy yield per mol e- for sequential dechlorination of tetrachloroethylene (PCE) and for dechlorination of TCEA and CT. Aqueous species are 10- 3 M and
pH = 7. DCE and TCEA are 12tDCE and 1l12TCEA respectively, but there is little
difference for other isomers.

Only the nucleophilic substitution by reduced sulfur and reductive dehalogenation have
been demonstrated to be significant abiotic transformations in groundwaters. Attributing
changes in either the presence or absence of halocarbons or the concentrations of halocarbons to abiotic processes is usually difficult. Bouwer and McCarty (1983a) indicate
that 1,1,2-trichloroethane can be degraded by three pathways under extremely anaerobic
(methanogenic) conditions. One pathway involves an initial abiotic dehydrohalogenation to I, l-dichloroethylene. Thus, the presence of the latter compound in contaminated
groundwater could suggest that this abiotic reaction is significant. However, Jeffers et al.
(1989) did not detect this product from the hydrolysis of I, I, I-trichloroethane, though
it could have been present as a low yield product, or it could be favored only under very anaerobic conditions. Cline and Viste (1985) do report l,l-dichloroethylene
in groundwaters apparently contaminated by I, I, l-trichloroethane. This could support
the abiotic reaction model, although other sources of I, l-dichloroethylene are possible.
Another pathway leads biologically from trichloroethane through I, I-dichloroethane to
chloroethane which can then undergo abiotic hydrolysis to ethanol. I, I-dichloroethane
was also reported by Cline and Viste (1985) where I,l-dichloroethylene was found, suggesting that the pathway leading through ethanol could also be operative at this field site.
However, ethanol was not determined in that study.
Sec. 9.2
Chemical Reactions
277
One can argue that chlorinated compounds which have short (perhaps < 1 year)
hydrolysis half lives (see Table 9.1) should not be found at significant concentrations in
dissolved plumes distant from a source. This hypothesis should be tested against field
data. It implies that compounds such as chloroethane, 1,1,2,2-tetrachloroethane and pentachloroethane should not be persistent contaminants. Cline and Viste (1985) indicate
that 1,1,2,2-tetrachloroethane was found at the water table on the site of a solvent recovery operation in Wisconsin, but was neither found at depth, nor downgradient; other
halocarbons were found, some still at relatively high concentrations. Chloroethane was
found only at depth downgradient, and appears to be a possible product of trichloroethane
transformation (see above).
In a controlled field experiment in the Borden aquifer, a number of dissolved
haloaliphatic compounds at concentrations of 20 to 35 ~gfL were injected into the
groundwater. Hexachloroethane suffered about 70% mass removal within 85 d, and essentially complete removal within 330 d (Roberts et al., 1986). Subsequent laboratory
and field studies (Criddle et al., 1986; Reinhard et al., 1990; Acton, 1990) suggest that
an abiotic, reductive dechlorination reaction was likely responsible. Although the Borden aquifer segment used in this experiment was generally aerobic, zones of low and/or
undetected oxygen 1 mg/L) were commonly encountered. Such zones, as well as
anaerobic microsites, could have been active in this transformation. Higher-than-injected
PCE levels were noted in the early part of the field experiment when hexachloroethane
was suffering the most significant mass loss, perhaps supporting the transformation of
hexachloroethane to PCE. Laboratory experiments (e.g., Acton, 1990) have found that
peE is persistent.
Where dissolved plumes are evaluated for evidence of the fate of halocarbons,
two aspects of the water chemistry should be investigated. One is the redox condition.
Reductive dechlorination is restricted to anaerobic systems, and so reactions of this type
can only be invoked for anaerobic waters. Redox potential is not reliably determined
in groundwaters, but an interpretation of the concentrations of redox-sensitive parameters such as oxygen, nitrate/ammonia, sulfate/sulfite, methane, Fe(III)/Fe(II) etc. along
with the pH can provide an estimate of the redox level. The other aspect that should be
considered in plume evaluation is the possible presence of non-halogenated breakdown
products such as alcohols and acids. To substantiate that hydrolysis is occurring in a
given circumstance, the presence of these products should be established. Unfortunately,
these products are, in general, more easily biodegraded than their halogenated precursors,
and so can be difficult to detect even as they are being generated. Most field investigations to date have not sought these products and this type of evidence for hydrolysis of
halocarbons has not yet been gathered.
Because the environmental conditions that promote dehalogenation reactions (low
redox level, HS-, Fe2 + , etc.) are the result of microbial processes, clear delineation
between chemical and biological processes will be difficult in the field. Perhaps such a
distinction is only worthwhile when in situ remediation is considered. For example, if the
remediation is microbially based, the need to maintain a microbial reaction may require
considerable manipulation of the subsurface environment, perhaps involving injection of
reductants, nutrients, primary substrates, etc. This may be difficult. For remediation by
278
Chap. 9
abiotic means, adding elemental iron as in a permeable zone through which the groundwater flows appears to produce useful abiotic reductive dechlorination (O'Hannesin and
Gillham, 1992).
9.3 MICROBIOLOGY OF THE SUBSURFACE
9.3.1 General
This section discusses concepts critical to understanding how natural microbial populations exist and function, how they react when challenged with pollutants such as chlorinated solvents, and how we may exploit knowledge of their biodegradative potential to
help predict the behavior of contaminant plumes. An exhaustive review of the microbial
ecology of the subsurface and the biodegradative potential of subsurface microorganisms
is beyond the scope of this work; articles covering these areas are available (Poindexter,
1981; Large, 1983: McCarty et al., 1984; Alexander, 1985; Grady, 1985; Zeikus et al.,
1985; Ghiorse and Wilson, 1988; Morgan and Watkinson, 1989).
Early studies have shown a sharp decline in the microbial populations in soils with
increasing depth (e.g., Waksman, 1916). These findings coupled with the low levels of
available nutrients in groundwaters have led many to assume that microbial activity in the
subsurface is extremely limited and unimportant. However, our growing understanding
of the terrestrial subsurface, both as an important source of potable water and as a waste
repository (intended or unintended), have led to a re-consideration of it as a habitat for a
variety of microorganisms (McNabb and Dunlap, 1975). Indeed, it has become evident
that although microbial populations do decline with depth in the unsaturated zone, they
frequently increase again when the saturated zone is reached. Indigenous microorganisms
can be widely distributed in groundwaters, often in numbers comparable to some soils
and surface waters (Wilson et al., 1983; Hirsch and Rades-Rohkohl, 1983; Ghiorse and
Balkwill, 1985; Balkwill et al., 1989; Jones et al., 1989; and others).
To allow survival, activity, and growth of its population, a microbial habitat must
provide nutrients for the synthesis of cellular constituents and the generation of energy.
Thus, sources of carbon, nitrogen, phosphorus, sulfur and low levels of various elements
(Fe, Mg, Mn, etc.) must be available. Heterotrophic organisms require organic carbon
sources; autotrophic organisms can use carbon dioxide as a sole source of carbon. A group
of aerobic, Cj-utilizing bacteria, including the methanotrophs, occupy an intermediate
position. They are "specialized heterotrophs", and are able to use reduced, one-carbon
compounds as sole sources of carbon and energy.
Suitable electron donors (such as organic compounds) together with electron acceptors (such as O 2, nitrate, ferric iron, sulfate, C02, or certain types of simple organic
compounds) are required for the generation of energy. N, P, and S sources may be
organic or inorganic, with some microorganisms able to use either while others are
more restricted. Physical and chemical factors such as the E H (redox potential), pH, and
Sec. 9.3
279
temperature will also profoundly influence the size, type, and activity of the microbial
population. In shallow aquifers at least, the pH and temperature are commonly in ranges
conducive to microbial activity (Ghiorse and Wilson, 1988). However, key nutrients may
be depleted in the subsurface environment because they are removed at the surface by
soil microbiota during infiltration of water into the ground.
Even pristine groundwaters contain some dissolved organic carbon (DOC), leached
from sedimentary rocks and/or freshly transported to the subsurface in recharge water
(McNabb and Dunlap, 1975). DOC concentrations ranging from 0.1 to 10 mgIL are
common, although DOC levels can reach several tens of mgIL (Freeze and Cherry,
1979). The specific chemical composition of naturally-occurring organic matter is largely
unknown; much of it may resemble humus, and biodegrade slowly. Typical levels of DOC
are sufficient for the growth of oligotrophic microorganisms, that is, organisms adapted to
exploit habitats with low nutrient fluxes (Poindexter, 1981). McNabb and Dunlap (1975)
suggest that the quality, not the quantity, of available carbon in the subsurface may limit
microbial activity.
The release of organic contaminants into the subsurface at concentrations well in
excess of the normal level of organic matter may significantly alter the makeup of the
microbial population. Bacteria suited to higher nutrient levels (copiotrophs), facultative
oligotrophs that are able to adapt to higher nutrient concentrations, or allochthonous
("foreign") bacteria imported into the subsurface (e.g., along with the spilled pollutant),
may become dominant in an environment that is no longer oligotrophic. The original
oligotrophs may not be able to compete for the available nutrients. Obligately oligotrophic
microorganisms are inhibited by excess carbon, perhaps because of the accumulation of
toxic metabolites such as hydrogen peroxide, or because of inhibition of certain enzymes
(Kuznetsov et al., 1979).
Contaminant compounds, if present in high enough concentration in the environment, may impair or destroy microbial activity. Phelps et al. (1988) could not detect
microbial biomass in areas of a subsurface profile contaminated with very high levels
( 500 mg/L) of TCE, although active, TCE-degrading microorganisms were present in
surrounding, less contaminated areas. Belay and Daniels (1987) reported that halogenated
hydrocarbons, particularly brominated compounds, were toxic to the methanogens; sensitivity varied, depending on the bacterium and the compound tested, but dibromoethane,
the most toxic compound tested, completely inhibited all test bacteria at a concentration
of 1.3 mg/L. Despite the potential toxicity of many contaminants, evidence suggests
that the metabolic potential for contaminant degradation is often present in the subsurface environment. Ventullo and Larson (1985) found that bacterial populations in
pristine and contaminated groundwaters had the potential to metabolize several xenobiotic organic substrates, although at reduced rates compared with surface water systems.
An increase in substrate carbon as a result of a spill may cause a previously
carbon-limited microbial population to become limited by another nutrient such as N
or P. Oxygen-containing groundwaters may become anoxic as the available oxygen
is used to oxidize the added organic matter. As a result, the activity of strictly aerobic microorganisms will cease. Facultative anaerobes may continue activity, although
280
Chap. 9
metabolic pathways may change. Escherichia coli, for example, mineralizes glucose
to carbon dioxide and water aerobically, but in the absence of oxygen it can ferment
glucose to produce organic acids, ethanol, carbon dioxide, and hydrogen. Strict anaerobes will be active in anoxic groundwaters when the redox potential is sufficiently low.
However, these bacteria may be extremely sensitive to oxygen (Hungate, 1969), and
could be scarce in recently-anaerobic groundwaters. Some species can survive adverse
conditions as resistant endospores and will renew growth when environmental conditions are more conducive to their activity. Sulfate-reducing bacteria are quite oxygen
tolerant and seem relatively widespread, even in aerobic environments. They remain
inactive until the redox potential falls to -100 mV or less (Postgate, 1984). Jones et
al. (1989) found viable methanogens and sulfate-reducing bacteria throughout the depth
profile of Atlantic Coastal Plains sediments although the sediments were not predominantly anaerobic. It should be noted, however, that anaerobic microsites are easily established even in environments that are essentially aerobic (Mayfield and Aldworth,
1974).

An understanding of biologically-mediated transformations is critical to the prediction
of organic contaminant behavior in the subsurface. Parsons et al. (1984) emphasize this
conclusion in a description of the problems associated with PCE-contaminated water:
one may need to search for the point source of a contaminant that is different from
the main contaminant that is being found in the water, while concurrently preparing to
remove a daughter contaminant which groundwater microbiota may produce in the
future through metabolism of the main contaminant. The biotransformation potential of
a given groundwater environment is dependent upon the enzymatic systems operating
there. Considerable differences in this potential exist even within anaerobic environments, because different microbial types function under the various redox conditions,
and not all microorganisms can do all things. For example, in laboratory studies, the
anaerobic degradation of monochlorophenols and monochlorobenzoates was more often
successful in methanogenic enrichments than in denitrifying or sulfate-reducing enrichments (Genthner et al., 1989). Extremely reducing, methanogenic conditions favor many
important biotransformations including those of halogenated one- and two-carbon hydrocarbons and of halogenated aromatic hydrocarbons (Kobayashi and Rittmann, 1982;
Wilson and Wilson, 1985). Given the complete range of redox conditions in groundwaters, a wide range of biotransformations is theoretically possible; however, because
a specific flow system may have a small redox range and the diversity of its microbial population is finite, the actual biotransformation potential may be quite limited.
Biotransformation pathways can be complex and obscure, especially in field situations. Natural microbial communities can be highly interdependent, and the biotransformation of a given compound may require the concerted activity of a number of
community members. The efficient function of a methanogenic community, for example,
requires the obligate interaction of fermentative, acetogenic, and methanogenic bacteria
Sec. 9.3
281
(Large, 1983). With respect to the degradation of xenobiotic compounds, the biodegradative capacity of a community is often greater than that of any single component of that
community. A single microbial species possesses a limited amount of genetic information,
and relatively limited adaptive potential, although it may adapt through mutation. The
adaptive potential of a community is much greater because the genetic capability of each
community member is in effect at the disposal of the whole community. A community
has a better chance of possessing the means to detoxify a toxic contaminant. Members
of a microbial community may acquire new metabolic capabilities from others in the
community by gene transfer. For example, the transfer of genes for 3-chlorobenzoate
catabolism has been documented in a natural surface water environment (Fulthorpe and
Wyndham, 1989).
The study of microbial communities should aid the understanding of contaminant
biotransformations in the natural environment more than the study of pure cultures. Unfortunately, such study is exceedingly difficult. Consequently, much of our knowledge
is based on pure culture work. Another problem with an oligotrophic environment such
as the subsurface is that those types of nutrient concentrations do not support the production of sufficient biomass to be studied with many of our present biochemical and
physiological research tools (Poindexter, 1981).
Organic contaminants may serve as primary substrates for a microbial population, or may be cometabolized. Cometabolism is "the transformation of a non-growth
substrate in the obligate presence of a growth substrate or another transformable compound" (Dalton and Stirling, 1982). In primary metabolism the organic substrate is used
as a source of carbon and energy, and may be mineralized, or fermented to other organic products. A phenomenon sometimes observed when multiple primary substrates
are present is diauxic growth, a result of catabolite repression. When multiple substrates
are available, bacteria may use only the most readily-metabolized substrate, and production of the enzymes needed for the degradation of the other substrate(s) will be
repressed until the best substrate has been exhausted; the poorer substrates may then
persist. Schmidt and Alexander (1985) reported that a Pseudomonas strain metabolized
glucose and aniline diauxically (glucose preferred) at 300 I-tgfL each, although the compounds were metabolized simultaneously when present at 3 I-tgfL each. Biodegradation
of a given primary substrate (typically at mgfL concentrations) may also be affected by
other "secondary substrates" at I-tgfL levels. The outcome will depend on the primary
contaminant substrate of interest, the concentration of the second substrate, and the nature
of the microorganisms in the environment (Schmidt and Alexander, 1985). Pseudomonas
sp. (strain LP), for example, simultaneously metabolizes methylene chloride and acetate,
but degrades methylene chloride preferentially over acetate, whether as the primary or
as the secondary substrate, providing the methylene chloride concentration is below an
inhibitory level of 10 mgfL (LaPat-Polasko et al., 1984).
Many biotransformations of xenobiotic compounds result from cometabolic reactions. Because of a lack of specificity with respect to substrate binding, some enzymes
will act on compounds that are analogs of their natural substrates and will transform them
(Slater and Bull, 1982). The microorganism gratuitously metabolizes the compound in
question, but derives no benefit from doing so. Three important points follow from this
282
Chap. 9
concept of cometabolism. Firstly, in cometabolism, the energy and carbon requirements

of the microorganisms must be met by other, primary substrates. Secondly, because no
benefit is derived from the cometabolized substrate, growth of the microbial population
will not be correlated with transformation of that compound. In contrast, utilization of a
primary substrate is correlated with population growth. Thirdly, unless they serve as substrates for some further enzyme-catalyzed step, perhaps conducted by another member
of the microbial community, products of cometabolic reactions may accumulate in the
environment. Such a product may also be of environmental concern, and may be more
problematic than its parent: as discussed above, vinyl chloride accumulates from the
transformation of more highly chlorinated ethylenes under methanogenic conditions, and
is of greater toxicological concern than the parent compounds. Product compounds are
commonly less hydrophobic than substrates, and tend to migrate more rapidly in groundwaters because they undergo less sorptive retardation (see also Figure 2.6 in Chapter 2).
Cometabolism and secondary substrate utilization may be significant removal processes
in groundwaters, because low (LglL) levels of certain compounds are considered hazardous. Removal of more than 99% of certain chlorobenzenes and other aromatics was
observed when these were added at concentrations of less than 15 (LglL to a bioreactor
where acetate at 1 mg/L served as the primary substrate (McCarty et al., 1984).
In the natural environment, degradation of an organic compound may lower the
concentration of the compound to a threshold level below which degradation will not
occur (Alexander, 1985). McCarty et al. (1984) discuss the parameter Smin, which is the
minimum bulk concentration of rate-limiting substrate able to sustain a biofilm in steady
state. The concept applies equally to single cells. Thus, a substrate concentration exists
below which an organism cannot obtain sufficient energy to function. Smin values of 0.1
to 1.0 mg/L are typical for aerobic utilization of organic compounds such as acetate
under steady-state conditions in biofilms (McCarty et al., 1984), and may be even lower
for oligotrophs. However, Smin values may be higher for anaerobic systems than for
aerobic systems (McCarty et al., 1981). In situ, microorganisms use a range of organic
substrates, so the overall Smin value may be a sum for a number of compounds. The
threshold concept also applies to the transformation of compounds that are not primary
substrates, because enzyme induction and activity, as well as growth, are subject to a
minimum inducer/substrate concentration. However, if the microbial population is using
an unrestricted primary substrate, the threshold level for the xenobiotic compound may
be below the detection levels of current analytical techniques (Alexander, 1985).
It has sometimes been assumed that biotransformation rates are proportional to
concentration. Extrapolations are therefore commonly made from high laboratory concentrations to low environmental concentrations, though such extrapolations are not always valid. Boethling and Alexander (1979) found that degradation of diethylamine was
proportional to concentration over the range from 10 mg/L to 10 (LglL. Alexander (1985)
reported that in stream water, 80% or more of added 2,4-D was removed at initial concentrations of 200 (LglL and 22,000 (LglL. However, for the same time period, less than 10%
was removed for an initial concentration of 2 (LglL. Kinetic models describing the rate
of substrate transformation as a function of initial substrate concentration and bacterial
cell density have been developed (Simkins and Alexander, 1984; Alexander, 1985).
Sec. 9.4
283
The biotransformation of organic contaminants in the subsurface is a multifaceted

process. Wang et a!' (1984) reported an interesting situation where isopropyl N-phenylcarbamate (Il'C) was mineralized in lake water when present at 400 jLglL, but was cometabolized to organic products at 1000 jLglL. At the lower level, it was suggested that IPC
served as a primary substrate for oligotrophic growth, but that the higher level inhibited
oligotrophic activity and copiotrophs unable to function at the low substrate level then
cometabolized the compound. The example illustrates the dynamic nature of biotransformations in the natural environment. Substrate utilization patterns, the active segment of
the microbial population, and the makeup of the microbial population can shift depending upon conditions. Similarly, halogenated hydrocarbons may be recalcitrant, degraded
as a primary substrate, or cometabolized, depending upon the halocarbon involved, the
site-specific microbial populations, and the environmental conditions. Biotransformation
at the dilute, leading edge of a contaminant plume may be conducted by the natural
oligotrophic population, while activity within the more concentrated plume core may
involve a very different, adapted microbial population.
Retardation of contaminants by sorption may also be critical in facilitating biotransformation of plume fronts. Where contaminants are readily biotransformed, as was
the case in a study of a contaminant plume derived from creosote waste, biotransformation rates may be, in effect, instantaneous within the context of slow groundwater
flow (Wilson et al., 1985). The length of the adaptation period may be influenced by
the time required for the destruction of inhibitors (the compound of interest itself, or
some other), the concentration of the compound, presence of other substrates, rarity of
the microorganism capable of the transformation, time required for small populations to
become large enough to effect a detectable compound loss, competition for inorganic nutrients, the physicochemical conditions at the site, or the time required for the creation of
a new genotype (Wyndham, 1986; Wiggins et al., 1987; Wiggins and Alexander, 1988).
It should also be recognized that adaptation may fail to occur.
9.4 BIOTRANSFORMATION OF HAlOALlPHATIC COMPOUNDS
9.4.1 General
Because few of the enzyme systems involved in halocarbon degradation have been thoroughly investigated, the pathways of halocarbon biotransformation are not documented
completely. Although all of the abiotic reactions affecting halocarbons discussed above
might also be catalyzed by microorganisms, the current literature suggests that only a few
reaction types are common. Leisinger (1983) indicated that biotransformations of halocarbons were of three general types: nucleophilic substitution, oxidation, and reductive
dehalogenation.
The aerobic utilization of haloalkanes such as dichloromethane (Brunner et al., 1980;

Stucki et al., 1981; LaPat-Polasko et al., 1984) and 1,2-dichloroethane (Stucki et al.,
1983; Janssen et al., 1985) as primary growth substrates has been reported. Cometabo1ic
284
Chap. 9
reactions seem to be more commonplace, and such biotransformations of many haloalkanes and haloalkenes have been reported under a variety of environmental conditions.
With respect to haloalkanes as primary growth substrates, Stucki et al. (1981) reported that a dichloromethane-utilizing Hyphomicrobium species produced formaldehyde
by first catalyzing the nucleophilic displacement of both chloride atoms. Xanthobacter autotrophicus 0110 possesses two hydrolytic dehalogenases. One of these is plasmid-borne,
and is specific for short-chain haloalkanes; the second is specific for halogenated carboxylic acids (Janssen et al., 1985; Tardif et al., 1991). A pathway was proposed for the
growth substrate 1,2-dichloroethane which involved dehalogenation to 2-chloroethanol,
then oxidation by a nonspecific methanol dehydrogenase to 2-chloroacetaldehyde. The
chloroacetaldehyde was dehydrogenated to 2-chloroacetate, and the latter was dehalogenated by the haloalkanoic dehydrogenase to acetate which can be metabolized through
normal catabolic pathways (Janssen et al., 1985). Some halocarbons actually act as "suicide substrates", that is, they are metabolized to intermediates that impair cell function. For example, the toxicant bromoacetaldehyde is produced from 1,2-dibromoethane
(Janssen et al., 1987).
Aerobic cometabolism of some haloalkenes may occur via a number of catabolic
oxygenases. Examples include biotransformation by nonspecific activity of methane
monooxygenase (MMO), by propane monooxygenase, by oxygenases normally involved
in the catabolism of aromatic compounds, by ammonia monooxygenase, and by alkene
monooxygenases. One important feature of aerobic, cometabolic transformation of the
chloroethylenes is that vinyl chloride does not appear as a product (Fogel et al., 1986)
as it does anaerobically (see below).
Wilson and Wilson (1985) observed degradation of TCE to CO 2 in an unsaturated
soil column exposed to 0.6% natural gas in air, and suggested that the degradation resulted
from the action of methanotrophic microorganisms. Experiments with a methane-utilizing
mixed culture (Fogel et al., 1986) and a pure methanotroph culture (Little et al., 1988)
demonstrated halocarbon degradation to CO2 and other products. The range of substrates
included dichloroethylenes and vinyl chloride, as well as TCE. PCE was recalcitrant.
Further studies with methane-enriched soils indicated that some chloroalkanes were also
removed (Henson et al., 1988). Enrichment cultures derived from methane-stimulated
soil degraded dichloromethane, chloroform and 1,2-dichloroethane as well as TCE, cis1,2-dichloroethylene, and trans-l,2-dichloroethylene, but not PCE, carbon tetrachloride,
or 1,1, l-trichloroethane (Henson et al., 1989); the latter was apparently removed from
the methane-enriched soil (Henson et al., 1988).
In methanotrophic growth, MMO catalyzes the oxidation of methane to methanol,
but the enzyme has a broad substrate specificity (Large, 1983) and can oxidize and
dechlorinate halogenated methanes (Haber et al., 1983). Mixed function oxidase systems catalyze the epoxidation of chlorinated alkenes (Miller and Guengerich, 1982), and
methanotrophs are known to epoxidate ethylene (Hou et al., 1979). Thus, MMO-catalyzed
degradation of chloroalkenes is believed to occur via epoxidation and subsequent hydrolysis (Fogel et al., 1986; Little et al., 1988). For example, Little et at. (1988) suggested that
TCE epoxide forms and then spontaneously breaks down to dichloroacetate, glyoxylate,
and the C\ compounds carbon monoxide and formate. Glyoxylate and dichloroacetate
Sec. 9.4
285
were detectable products of TCE degradation in pure methanotroph cultures (Little et

al., 1988). Most methanotrophs are unable to utilize multi-carbon compounds, thus accumulation of these products is not unexpected. Carbon monoxide and formate could be
oxidized further. It is noteworthy that in mixed methane-utilizing cultures and soil column
systems, a larger fraction of the TCE carbon appears as C02 and less as water soluble
organic carbon than in pure methanotroph cultures. Compounds such as glyoxylate are
readily metabolized by heterotrophs in the mixed systems (Little et al., 1988).
The aerobic degradation of TCE to CO 2, chloride, and other nonvolatile products
has been reported for a number of aromatic-degrading bacteria (Nelson et al., 1987;
Nelson et al., 1988; Harker and Kim, 1990). Activity of the required enzyme is induced
by a compound such as toluene, and the enzyme fortuitously oxidizes TCE. TCE itself
cannot induce enzyme synthesis. TCE degradation is inhibited by the natural enzyme
substrate, and vice versa, as the two compounds compete for the same catalyst (Folsom
et al., 1990). The toluene dioxygenase of Pseudomonas putida Fl also degrades 1,1-, cis1,2- and trans-1,2-dichloroethylene, but not PCE, vinyl chloride, or ethylene (Wackett
and Gibson, 1988). P. cepacia G4 and P. mendocina possess toluene monooxygenase
activities which also transform TCE (Nelson et al., 1987; Shields et al., 1989; Winter
et al., 1989). An Alcaligenes eutrophus strain possesses both a chromosomal phenoldependent pathway and a plasmid-encoded 2,4-dichlorophenoxyacetic acid pathway to
degrade TCE (Harker and Kim, 1990). The routes of TCE degradation by these aromaticdegraders have not been completely delineated, but are assumed to be similar to that for
MMO, i.e., the formation of unstable oxoderivatives which decompose to nonchlorinated
compounds. Radio1abelled glyoxylate and formate were detected in a P. putida F1 culture
incubated with 14C_TCE (Wackett and Householder, 1989).
Degradation of TCE by aromatic-degraders has been described mostly for pure,
laboratory cultures. The exceptions to this are the initial work of Nelson et al. (1986) who
reported one positive result out of 43 samples, and a paper discussing pilot studies in both
surface bioreactors and in situ remediation (Nelson and Kinsella, 1990). However Nelson
et al. (1988) believe that microorganisms possessing the required oxygenase activities are
not uncommon in the natural environment. They suggest that TCE recalcitrance reflects
a lack of the appropriate inducing conditions. TCE may be aerobically biodegraded in
situ if an appropriate growth substrate is also present.
Under aerobic conditions, the autotrophic, nitrifying bacterium Nitrosomonas europaea can biotransform a number of di- and trihalogenated alkanes, vinyl chloride, the
three dichloroethylene isomers, cis-bromoethylene, and TCE (Vannelli et al., 1990); it
cannot degrade carbon tetrachloride, PCE, or trans-dibromoethylene. The cometabolic
reactions are catalyzed by ammonia monooxygenase. This activity may be useful in
bioremediation schemes because the bacteria grow on ammonia and carbon dioxide, and
do not require an organic carbon source.
It should be noted that halogenated hydrocarbons may produce deleterious effects
in the bacteria responsible for their biotransformation. Such effects have been reported
in aromatic-degraders (Wackett and Householder, 1989), methanotrophs (Alvarez-Cohen
and McCarty, 1991; Henry and Grbic-Galic, 1991; Oldenhuis et al., 1991), in a Xanthobacter strain that cometabolizes chloroalkenes via monooxygenase (Ensign et al.,
286
Chap. 9
1992), and in Nitrosomonas (Rasche et al., 1991). Usually, toxicity appears to result
from a nonspecific covalent binding of reactive degradation intermediates to cellular
proteins, including the enzyme responsible for the cometabolic reaction. Recovery of
laboratory cultures is possible under conditions allowing de novo synthesis of the affected biomolecules (Wackett and Householder, 1989; Oldenhuis et al., 1991; Rasche et
al., 1991).
9.4.3.1 General
Many chlorinated aliphatic compounds are reductively dechlorinated under anaerobic

conditions. This appears to be the sole route of PCE biodegradation, and TCE is also resistant to biodegradation in aerobic environments, except under the special circumstances
reviewed above. Tri- and tetrahalogenated methanes and ethanes are generally recalcitrant
under aerobic conditions, but are transformed anaerobically. The following paragraphs
discuss the biotransformation of haloalkanes and haloalkenes under the various anaerobic
redox environments defined earlier. It must be emphasized, as will become evident in
the discussion below, that the dehalogenation activity of mixed cultures under given redox conditions (e.g., nitrate-reducing conditions) does not imply that the microorganisms
utilizing that electron acceptor (e.g., those reducing nitrate) are the ones responsible for
the observed dehalogenation. A fermentative microorganism, for example, might be
active over the entire redox spectrum and could conceivably be a source of dehalogenating activity under aerobic, nitrate-reducing, sulfate-reducing or methanogenic conditions.
Pure culture studies are generally required to attribute dehalogenation unequivocally to
a certain type of microorganism utilizing a certain electron acceptor.
9.4.3.2 Haloalkanes
The transformation of carbon tetrachloride and brominated trihalomethanes has been

observed under denitrifying conditions in batch culture (Bouwer and McCarty, 1983b).
Reductive dechlorination of carbon tetrachloride to chloroform was detected, but carbon
tetrachloride was also mineralized to CO 2 and some carbon was assimilated by the
biomass. In anoxic biofilm columns under denitrifying, sulfate-reducing, or methanogenic conditions, reductive dehalogenation was the predominant mechanism of removal
of a number of halogenated aliphatic compounds, but 14C02 production from labelled
carbon tetrachloride and 1,2-dibromoethane was also noted (Bouwer and Wright, 1988).
Eg1i et al. (1988) suggested that acetogens in the mixed, denitrifying cultures could have
produced the observed CO 2. Acetobacterium woodii catalyzes reductive dechlorination,
and transforms carbon tetrachloride to CO 2, although the CO 2-producing reactions have
not been elucidated (Egli et al., 1988). Criddle et al. (1990a) have in fact isolated a
denitrifying Pseudomonas sp. that transforms carbon tetrachloride, forming CO 2 and an
unidentified water-soluble fraction as major products. These authors note, however, that
denitrifying microorganisms able to transform carbon tetrachloride seem rare, supporting
the suggestion of Eglis group.
Sec. 9.4
287
The denitrifying pseudomonad of Criddle et al. (1990a) possesses a unique carbon

tetrachloride transformation pattern, as little or no chloroform is produced. Competing
pathways for carbon tetrachloride transformation seem more usual, not only in mixed culture systems (Bouwer and Wright, 1988, for example), but also within single strains of
bacteria. The two major biotic pathways are hydrogenolysis (which produces chloroform
(and dichloromethane under sufficiently reducing conditions, and hydrolysis (which produces CO 2) (Egli et al., 1988; Criddle et al., 1990a, 1990b; Criddle and McCarty, 1991).
Criddle and McCarty (1991) have demonstrated these processes in an electrolytic model
system, concluding that they may occur simultaneously, and that the latter biotransformation could involve the hydrolytic reduction of carbon tetrachloride to formate or carbon
monoxide, followed by oxidation to carbon dioxide. A similar reduction then subsequent
oxidation, might also explain the production of acetic acid from 1,1,l-trichloroethane by a
fermentative Clostridium sp. (see below). In either instance, an additional electron donor
would be required for microbial metabolism because no extra electrons are released by
the proposed transformation sequence (Criddle and McCarty, 1991).
Escherichia coli has been found to reductively dechlorinate carbon tetrachloride
(Criddle et al., 1990b). In that study, the product distribution and the rate of biotransformation were both found to be electron acceptor-dependent. The most rapid biotransformation occurred under fermentative conditions, where most labelled carbon from 14C-carbon
tetrachloride was recovered as chloroform or as cell-bound 14C; almost no CO 2 was
formed. Under fumarate-respiring conditions, a larger portion of the label was associated
with CO 2 and the nonvolatile but soluble fraction; chloroform was still produced. Relatively little activity was observed when oxygen or nitrate were provided as the electron
acceptors (Criddle et al., 1990b).
A fermentative Clostridium sp. isolated from a methanogenic bioreactor known
to degrade PCE, TCE, 1,1,l-trichloroethane, and chloroethane was shown to reductively dechlorinate 1,1, l-trichloroethane to 1,I-dichloroethane; other products, including
acetic acid were also produced (Galli and McCarty, 1989a). It was proposed that 1,1,1trichloroethane may compete with natural electron acceptors (e.g., amino acids) present
in the growth medium (Galli and McCarty, 1989b). Comparison of the transformation
capabilities of the Clostridium sp. and the parent methanogenic mixed culture (Vogel
and McCarty, 1987) clearly showed that multiple strains were responsible for the activity observed in the mixture. The Clostridium accumulated Ll-dichloroethane from
1,1, l-trichloroethane, whereas the mixed culture carried the biotransformation further,
producing chloroethane and C02. The mixed culture also degraded PCE and TCE (Vogel
and McCarty, 1987), but the Clostridium did not (Galli and McCarty, 1989a).
Reductive dechlorination of halomethanes and haloethanes under sulfate-reducing
conditions has been observed (e.g., Bouwer and Wright, 1988). Egli et al. (1987) showed
that Desulfobacterium autotrophicum reduces carbon tetrachloride to chloroform and
dichloromethane, and 1,1, I-trichloroethane to 1,1-dichloroethane.
Ethane, ethylene and acetylene production from bromoethane, dibromo- or dichloroethane, and 1,2-dibromoethylene, respectively, have been observed in pure cultures of a
number of methanogens (Belay and Daniels, 1987). Combinations of dehydrohalogenation and reductive dehalogenation reactions were suggested as the pathways. Mikesell and
288
Chap. 9
Boyd (1990) suggested that the methanogen Methanosarcina sp. DCM may dechlorinate
chloromethanes to CO 2 and other products via methyltransferase, a corrinoid enzyme
that normally binds methanol. C02 formation from the chloromethane was presumed to
occur via the same pathway by which CO 2 is formed from methanol disproportionation.
In vitro, reductive dehalogenation of chloromethanes by both corrinoids and the
methanogen-specific cofactor F430 has been demonstrated with titanium (III) citrate or
dithiothreitol as the electron donor (Krone et al., 1989a,b; HoBiger et al., 1992a). In
Methanobacterium thermoautotrophicum, it is the factor F430-associated enzyme methy1coenzyme M reductase, not the corrinoid enzymes, that is able to catalyze reductive
dechlorination of 1,2-dichloroethane (Holliger et al., 1992b). Corrinoid enzymes may
still be important, however, since not all dechlorinating anaerobes possess F430 . Dechlorination catalyzed by cell extracts of Methanosarcina barkeri seems to involve both
corrinoids and factor F430 (Holliger et al., 1992a). Eg1i et al. (1988) suggest that the
dechlorinating ability of acetogens, too, may be correlated with possession of corrinoid
enzymes of the acetyl CoA pathway.
9.4.3.3 Haloalkenes
The anaerobic biotransformation of chloroethylenes has been demonstrated in methanogenic, fixed film bioreactors (Bouwer and McCarty, 1983a; Vogel and McCarty, 1985), in
anaerobic soils (Kloepfer et al., 1985), in muck from an aquifer recharge basin (Parsons
et al., 1984), in anaerobic sewage sludge (Fathepure and Boyd, 1988a), in subsurface
materials (Wilson et al., 1986; Barrio-Lage et al., 1987; Suflita et al., 1988; Major et
al., 1991), in mixed methanogenic enrichment cultures (Baek and Jaffe, 1989; Freedman and Gossett, 1989), in sulfate-reducing enrichments (Bagley and Gossett, 1990), in
groundwater-derived enrichments (Kastner, 1991), in pure cultures of methanogens (Fathepure et al., 1987; Fathepure and Boyd, 1988b), and in the sulfidogenic, dehalogenating
bacterium strain DCB-l (Fathepure et al., 1987).
Under highly-reducing conditions, the chloroethylenes are transformed mainly
through successive reductive dech1orinations from PCE to TCE to cis-1,2-dichloroethylene to vinyl chloride (Parsons et al., 1984; Parsons and Lage, 1985; Barrio-Lage
et al., 1986; Vogel and McCarty, 1985). The compound cis-1,2-dichloroethy1ene is the
predominant dichlorinated species (Parsons and Lage, 1985; Barrio-Lage et al., 1986;
Scholz-Muramatsu et al., 1990; Major et al., 1991) although Freedman and Gossett (1989)
detected predominantly the trans isomer in their enrichments. 1,l-dichloroethylene is the
least significant isomer, and may be the product of sulfide-mediated abiotic dechlorination
(Kastner, 1991). The non-chlorinated products CO 2 (Vogel and McCarty, 1985) and ethylene (Freedman and Gossett, 1989) have also been detected in mixed enrichment culture
studies. Subsidiary pathways also operate in mixed cultures. Barrio-Lage et al. (1986)
detected chloroethane production in microcosms spiked with cis-1,2-dichloroethylene.
Chloroethane was a product of TCE degradation in the mixed cultures studied by Baek
and Jaffe (1989). Their evidence suggested that fermentative bacteria were important
in the hydrogenation of the C=C double bond. More chloroethane was evolved when
a mixture of two cultures (one enriched for fermentative bacteria and the second for
Sec. 9.4
289
methanogenic bacteria) was used as the inoculum than when the inoculum was taken
only from the methanogenic enrichment.
Reductive dechlorination under methanogenic conditions is dependent upon the
availability of an electron donor such as acetate, methanol, or H2 , and frequently parallels
methane production (Fathepure et al., 1987; Freedman and Gossett, 1989). The ratio of
dechlorination to methane production varies with the compound being attacked. Strains of
Methanosarcina dechlorinated about 0.71 {Lmole of chloroform per millimol of methane
produced, but only about 0.05 {Lmole of PCE per mmole methane (Fathepure et al.,
1987; Fathepure and Boyd, 1988b; Mikesell and Boyd, 1990). Since highly chlorinated
compounds are most readily dechlorinated (Vogel et al., 1987), the rate-limiting step in
the conversion of PCE to ethylene by a methanogenic enrichment is the final step from
vinyl chloride to ethylene (Freedman and Gossett, 1989).
In some studies of dechlorination under methanogenic conditions, methanol has appeared to be more effective than acetate as a source of the reducing equivalents (BarrioLage et al., 1987; Fathepure et al., 1987; Freedman and Gossett, 1989), though it is not
clear why this is so. It could be significant that methanol metabolism by methanogens
involves additional steps not required to metabolize other methanogenic substrates, e.g.,
the step catalyzed by methyltransferase I, the corrinoid enzyme invoked by Mikesell and
Boyd (1990) in their scheme for dechlorination reactions (see above). Consistent with this
view, Fathepure et al. (1987) suggest the higher dechlorination rate by a Methanosarcina
sp. growing on methanol may be related to the reducing power generated. Methanol
metabolism produces six reducing equivalents, versus two reducing equivalents from acetate. Acetate-cleaving methanogens are nevertheless clearly capable of reductive dechlorination (Fathepure and Boyd, 1988a).
The inhibitor 2-bromoethanesulfonate (BES) blocks the final step in methanogenesis (see Zeikus et al., 1985), and affects PCE and TCE degradation as well as methane
production in mixed, methanogenic cultures (Fathepure and Boyd, 1988a; Freedman and
Gossett, 1989). This suggests that methanogens playa key role in observed reductive
dechlorinations. In contrast, Scholz-Muramatsu et al. (1990) found that if methanogenesis
was inhibited with BES in a mixed culture fed with benzoate and PCE, dechlorination of
PCE to cis-l,2-dichloroethylene continued as benzoate was fermented. They suggested
that during benzoate fermentation, a portion of the benzoate was reduced to cyclohexylcarboxylate while the rest was oxidized to acetate, with PCE accepting the electrons from
the oxidation. Thus, both benzoate itself and PCE served as electron acceptors.
Gossett and his group have adapted their methane-producing enrichments to high
PCE levels (up to 55 mg/L aqueous concentration, DiStefano et al., 1991). Under these
conditions, methanogenesis ceased, apparently inhibited by the PCE level, but virtually
complete conversion of PCE to ethylene occurred with little vinyl chloride accumulation. An electron balance showed that methanol consumed by the cultures was accounted
for by dechlorination (31%) and acetate production (69%). Evidence of a shift in the
makeup of the microbial population from methanol-utilizing methanogens to predominantly methanol-utilizing acetogens was obtained, suggesting that the latter were responsible for dechlorination. However, because the microbial population was mixed and interactive, direct proof is lacking and other explanations are possible (DiStefano et al., 1991).
290
Chap. 9
PCE dechlorination stoichiometrically related to sulfide production has been observed in sulfide-reducing enrichment cultures (Bagley and Gossett, 1990). Such cultures exhibited virtually no methane production in the presence of BES, but PCEdechlorinating ability was equal to that of equivalent cultures devoid of BES. This
evidence suggests that reductive dechlorination may have been brought about by sulfatereducing bacteria (Bagley and Gossett, 1990). The authors noted that the sulfate-reducing
enrichments were not nearly as effective in PCE transformation as their methanogenic
enrichments (Freedman and Gossett, 1989) operated under similar conditions.
Based on the above discussions, we can conclude that reductive dehalogenation of
halogenated, aliphatic hydrocarbons has been observed under anaerobic conditions. These
transformations are carried out by certain fermentative, acetogenic, nitrate-reducing,
sulfate-reducing and methanogenic bacteria. The reactions appear to be cometabolic in
nature, although assimilation of some carbon has been observed (Bouwer and McCarty,
1983a,b), and the strain DCB-l may gain energy from the dehalogenation (Dolfing and
Tiedje, 1987)).
The accumulated data suggest that the reductive dehalogenation is fortuitously
catalyzed by certain types of enzymes, with the halogenated compounds acting as alternate electron acceptors in place of the natural acceptor molecules (Egli et al., 1987;
1988; Freedman and Gossett, 1989; Galli and McCarty, 1989b; Mikesell and Boyd,
1990; Scholz- Muramatsu et al., 1990; Criddle et al., 1990a; Holliger et al., 1992a,b ).
Whether enzyme proficiency is type-specific and/or species-specific remains to be determined. This is an area of current research interest. Cofactors such as cytochromes,
NADH, or the unusual methanogenic coenzymes (e.g., F430 ) , could mediate electron
transfer, and it is not known whether intracellular or extracellular electron transfer reactions are responsible for reductive dehalogenations. An alternate view might be that
reductive dechlorination is driven by any of the enzymes in question, simply as a result
of nonspecific diversion of reducing power generated during normal catalytic activity
from the normal substrate-product reaction to the chlorinated compound. Then, reductive
dechlorination would be governed by the amount of enzyme protein(s) and cofactor(s)
available and the ease with which the chlorinated compound is dechlorinated under the
given redox environment. The fact that microorganisms have strain-specific capabilities
to reductively dechlorinate only certain compounds suggests that the first view may prove
to be correct, i.e., some enzymes are more adept at the process than others. Fathepure
and Boyd (l988a) observed that two species of Methanosarcina differed in their ability
to dechlorinate PCE, and some types of anaerobic bacteria may not transform halocarbons at all (Fathepure et al., 1987; Egli et al., 1988). Because of the importance of
phenomena such as interspecies hydrogen transfer and cross-feeding in anaerobic consortia, however, other community members are likely to indirectly influence reductive
dechlorination.
Most of the studies discussed in the previous sections were performed in the laboratory
under ideal reaction conditions. Flow was incorporated as a parameter in some studies;
Sec. 9.4
291
others were conducted under static conditions. The progression from pure, laboratory
cultures to enrichment cultures to microcosms using materials taken from the field is one
of increasing complexity, but also one of increasing reality in terms of what occurs in the
natural environment. The more "real" the laboratory study, the more difficulty encountered in pinpointing the operative chemical or biological mechanism(s) and influential
factors. Thus, although pure chemical/culture work is invaluable for defining mechanisms
and pathways, and also for testing hypotheses, we must continue to grapple with more
complex systems if we are to understand the abioticlbiotic transformation potentials of
natural environments.
Laboratory studies, particularly with pure cultures, are frequently conducted at temperatures optimal for activity of the microorganisms, often above 20C. However, most
groundwater systems in high latitudes are considerably cooler. Laboratory experiments
at lower temperatures are therefore more realistic, because not only are reaction rates
affected by temperature, but different bacterial strains are adapted to distinct temperature
ranges.
Information on the rates and products of biotransformations and the microbial population dynamics of natural aquifer communities under in situ conditions are needed.
However, many studies of reductive dechlorination have been conducted using microbial
enrichments derived from wastewater treatment systems, not from contaminated groundwaters. Comparison of sewage sludges from different sources have shown considerable
variation in their degradation capacity (Fathepure and Boyd, 1988a). Whether subsurface
environments will demonstrate the same heterogeneity with respect to biodegradative capability is not known, but since microbial numbers may be smaller in the subsurface than
in sewage sludges, one might expect that the probability of a certain microorganism's
presence will be lower in the subsurface than in sewage sludge. Few data are available
to support or rebut this contention, but Kastner's (1991) study of reductive dechlorination, which was conducted with groundwater-derived enrichment cultures, suggests
results from wastewater/sludge enrichments may not be directly applicable to all groundwater environments. Kastner (1991) demonstrated that reductive dechlorination of PCE
and TCE could be driven by the transition of aerobic, contaminated groundwater-derived
enrichments to anaerobic conditions (E H equal to between -150 and -210 mY). cis1,2-dichloroethylene was the major product, but no further dechlorination occurred and
no methane was detected. Repeated anaerobic subculture of the dechlorinating cultures
resulted in a gradual loss of dechlorinating capacity. Only aerobic subcultures stably
maintained dechlorination activity, indicating that the aerobic-anaerobic shift was necessary for biotransformation. Enrichments derived from pristine sites had no ability to
reductively dechlorinate under similar conditions. The observed cis-l,2-dichloroethylene
accumulation is comparable to what is observed in contaminated aquifers in Germany,
where cis-l,2-dichloroethylene is often the main contaminant, and only traces of vinyl
chloride and methane are observed (Kastner, 1991).
Table 9.2 summarizes the conditions under which biotransformation of a given
compound has been documented in laboratory studies. Key references are cited. Some
were chosen in part because they mention earlier work. The comments in Table 9.2 provide additional information. In studies classed as "anaerobic" for example, a distinction
TABLE 9.2
N
Summary of Laboratory Observations on Biotransformations.
Compound
Condition
Biotransformation
aerobic
methane-oxidizing
aromatic-oxidizing
Nl-lj-oxidizing
anaerobic
NO.3-reducing
no
yes
Janssen et at. (1985)

Dalton & Stirling (1982)
yes
Rasche et al. (1991)
yes
yes
LaPat-Polasko et al. (1984)

Henson et al. (1988); Stirling & Dalton (1982)
chloromethane
Reference
SO~- -reducing
C02-reducing
dichloromethane
bromomethane
aerobic
methane-oxidizing
aromatic-oxidizing
Nfh-oxidizing
anaerobic
NO.3-reducing
yes
Vannelli et at. (1990)
yes
Bouwer & McCarty (1983); Brunner et at. (1980)
SO~- -reducing
C02-reducing
yes
Gossett & Freedman (1989)
yes
Dalton & Stirling (1982)
yes
Rasche et at. (l990a)
yes
Vannelli et al. (1990)
aerobic
methane-oxidizing
aromatic-oxidizing
NH3-oxidizing
anaerobic
NO.3-reducing
SO~- -reducing
C02 -reducing
dibromomethane
aerobic
methane-oxi dizing
aromatic-oxidizing
Nllj-oxidizing
anaerobic
NO.3-reducing
SO~- -reducing
C02-reducing
Comments
pure culture
TABLE 9.2
continued
Compound
dibromochloromethane
bromodichloromethane
tribromomethane
(bromoform)
trichloromethane
(chloroform)
1.0
carbon
tetrachloride
Condition
Biotransformation
abiotic
aerobic
methane-oxidizing
aromatic-oxidizing
anaerobic
NO;- -reducing
SO~- -reducing
C02 -reducing
yes
no/no
abiotic
aerobic
methane-oxidizing
aromatic-oxidizing
anaerobic
NO;--reducing
SO~- -reducing
CO2-reducing
yes
yes/no
yes
Bouwer & McCarty (l983b); Bouwer & Wright (1988)
yes
yes
Bouwer & Wright (1988)

Bouwer & McCarty (l983a); Bouwer & Wright (1988)
abiotic
aerobic
methane-oxidizing
aromatic-oxidizing
anaerobic
NO;--reducing
SO~- -reducing
CO2-reducing
yes
yes
Bouwer & McCarty (l983a)

Tabak et al. (1981)
yes
Bouwer & McCarty (l983b); Bouwer & Wright (1988)

aerobic
methane-oxidizing
aromatic-oxidizing
Nl'Ij-oxidizing
anaerobic
NO;--reducing
Reference
Bouwer et al. (1981); Tabak et al. (1981)
yes
Bouwer & McCarty (l983b)
yes
yes
yes
yes/no
yes
Comments
Bouwer & McCarty (I 983a)

Tabak et al. (1981 )/Bouwer et al. (1981)
Tabak et al. (l981)1B0uwer et al. (1981)

Henson et al. (1988)
yes
yes
no

Galli & McCarty (l989a)
SO~- -reducing
CO2-reducing
no
yes

Bouwer & McCarty (l983a); Mikesell & Boyd (1990)
abiotic
aerobic
yes
yes
Reinhard et al. (1990)

Tabak et al. (1981)
fermentative
TABLE 9.2
N
continued
Compound
Condition
Biotransformation
methane-oxidizing
aromatic-oxidizing
Nl-lj-oxidizing
anaerobic
NO;- -reducing
no
Reference
Comments
1.0
""
SO~- -reducing
CO2-reducing
chloroethane
no
yes
yes/no

Criddle et al. (I 990b); Egli et al. (1988)
Criddle et al. (l990a)/Egli et al. (1988)
Egli et al. (l988)/Bouwer & Wright (1988)
2 species/-
no
Janssen et al. (1985)
pure culture
yes
Rasche et al. (I 990b)
yes
yes
yes
Belay & Daniels (1987)
yes
yes
Tabak et al. (1981)

yes
SO~- -reducing
CO2-reducing
yes
Vogel & McCarty (1987)
aerobic
methane-oxidizing
yes
yes

aerobic
methane-oxidizing
aromatic-oxidizing
Nl-lj-oxidizing
anaerobic
NO;- -reducing
yes & no/yes

yes
SO~- -reducing
CO2-reducing
bromoethane
aerobic
methane-oxidizing
aromatic-oxidizing
NH3-0xidizing
anaerobic
NO;- -reducing
SO~- -reducing
CO2-reducing
I,I-dichloroethane
1,2-dichloroethane
fermentative
-/autotroph
aerobic
methane-oxidizing
aromatic-oxidizing
Nl-Ij-oxidizing
anaerobic
NO;- -reducing
TABLE 9.2
continued
Compound
Condition
aromatic-oxidizing
Nl-ij-oxidizing
anaerobic
NO.3-reducing
SO;- -reducing
CO2-reducing
1,2-dibromoethane
1,I, l-trichloroethane
1,1,2-trichloroethane
1.0
U1
1,1,2,2-tetrachloroethane
aerobic
methane-oxidizing
aromatic-oxidizing
Nllj-oxidizing
anaerobic
NO.3-reducing
SO;- -reducing
CO2-reducing
Reference
Biotransformation
yes
yes
Bouwer & McCarty (l983a); Egli et al. (1987)
yes
Pignatello (1986)
yes
no
Bouwer & McCarty (l983b)
yes
Belay & Daniels (1987); Bouwer & Wright (1988)
yes
no
yes/no

Henson et al. (l988)!Henson et al. (1989)
abiotic
aerobic
methane-oxidizing
aromatic-oxidizing
NH3-oxidizing
anaerobic
NO.3-reducing
maybe/no
Bouwer & Wright (l988)lBouwer& McCarty (l983b)
SO;- -reducing
CO2-reducing
maybe/yes
yes
Bouwer & Wright (l988)/Egli et al. (1987)

Vogel & McCarty (1987); Bouwer & Wright (1988)
yes
Comments
pure culture
soil/enrichment
Janssen et al. (l985)!Castro & Belser (1990)

aerobic
methane-oxidizing
aromatic-oxidizing
Nl-Ij-oxidizing
anaerobic
NO.3-reducing
SO;- -reducing
CO2-reducing
no/yes
yes
yes
aerobic
methane-oxidizing
aromatic-oxidizing
no
pure culture
pure culture
TABLE 9.2
N
continued
Compound
I.C
a"
Condition
NH3-0xidizing
anaerobic
NO;- -reducing
SO~- -reducing
CO2-reducing
hexachloroethane
vinyl chloride
cis-I,2-DCE
yes
Rasche et at. (1991)
yes
abiotic
aerobic
methane-oxidizing
aromatic-oxidizing
anaerobic
NO;- -reducing
yes
yes
Reinhard et at. (1990)

Tabak et al. (1981)
yes
SO~- -reducing
CO2-reducing
yes
yes

aerobic
methane-oxidizing
propane-oxidizing
aromatic-oxidizing
Nl-lj-oxidizing
anaerobic
NO;- -reducing
yes
yes
yes
yes
Davis &
Fogel et
Wackett
Wackett
Vannelli
yes
yes
yes
yes
yes
yes
yes
Tabak et at. (1981)

Fogel et al. (1986)
Wackett et al. (1989)
Wackett & Gibson (1988)
Barrio-Lage et al. (1986)
SO~- -reducing
C02 -reducing
yes
Wilson et al. (1986)
aerobic
methane-oxidizing
yes
yes
Tabak et at. (1981)

Fogel et al. (1986)
SO~- -reducing
CO2-reducing
l,l-DCE
(gem-)
Reference
Biotransformation
aerobic
methane-oxidizing
propane-oxidizing
aromatic-oxidizing
Nl-lj-oxidizing
anaerobic
NO;- -reducing
no
Comments
Carpenter (1990)
al. (1986)
et al. (1989)
& Gibson (1988)
et al. (1990)
muck microcosm
TABLE 9.2
continued
Compound
Condition
Biotransformation
propane-oxidizing
aromatic-oxidizing
NlIj-oxidizing
anaerobic
N0 3-reducing
yes
yes
yes
yes

yes
yes
yes
yes
yes
yes
yes
Tabak et al. (1981)

Fogel et al. (1986)
yes
SO;- -reducing
CO2-reducing
trans-I,2-DCE
aerobic
methane-oxidizing
propane-oxidizing
aromatic-oxidizing
NH3-oxidizing
anaerobic
N03-reducing
SO;- -reducing
CO2-reducing
TCE
aerobic
methane-oxidizing
propane-oxidizing
aromatic-oxidizing
NlIj-oxidizing
anaerobic
N0 3-reducing
SO;- -reducing
CO2-reducing
PCE
N
<,t:J
"
aerobic
methane-oxidizing
propane-oxidizing
aromatic-oxidizing
NH3-0xidizing
anaerobic
N0 3-reducing
SO;- -reducing
CO2-reducing
Reference
yes/no
yes
yes
yes
yes
yes
Tabak et al. (1981 )/Bouwer et al. (1981)

Wilson & Wilson (1985); Little et al. (1988)
Wackett & Gibson (1988); Harker & Kim (1990)
yes
yes
Bagley & Gossett (1990)

Wilson et al. (1986); Freedman & Gossett (1989)
yes/no
no
no
no
no
yes
no
yes
yes
Tabak et al. (1981)1 Bouwer et al. (1981)

Fogel et al. (1986)
Scholz-Muramatsu et al. (1990)
Bagley & Gossett (1990)
Fathepure & Boyd (1988b); Freedman & Gossett (1989)
Comments
muck microcosm
muck microcosm
muck microcosm
fermentative
298
Chap. 9
is drawn between a case wherein the redox state was undefined (e.g., Barrio-Lage et al.,
1986), and one wherein the microbial activity was fermentative (e.g., Galli and McCarty,
1989a). The broad, aerobic biodegradative capacity recorded by Tabak et al. (1981) may
be somewhat optimistic, based on the results of others, while the inability of the Xanthobacter isolate of Janssen et al. (1985) to aerobically degrade certain chloromethanes
and chloroethanes does not mean isolates with broader substrate ranges will not be found
in the future. If its significance was observed under conditions conducive to biotransformation, abiotic transformation is also noted in Table 9.2.
A comparison of biotransformation rates of various halogenated hydrocarbons via
the different pathways discussed above would be most useful. However, because of
the wide range in experimental conditions and methods of reporting results, meaningful
comparisons are difficult. Table 9.3 summarizes some reported rates for PCE and TCE.
Most are derived from pure culture studies, and are likely to differ significantly from
field rates. However, unlike field-derived rate data, they do relate biotransformation to
the amount of microbial biomass present. In general, the rates of PCE transformation
by the anaerobic species are slower than the rates of TCE transformation by the aerobic
cultures. Strain DCB-l, the fastest dechlorinator of the anaerobes listed, was isolated
from an anaerobic consortium (Dolfing and Tiedje, 1987) which degraded PCE even
TABLE 9.3
Biotransformation of Chloroethenes.
Degradation rate
Culture, condition
(rz.mol/d-mg protein)
Reference
PCE degradation
strain DCB-I, anaerobic
Methanosarcina sp., anaerobic
M. mazei, anaerobic
0.00234
8.4 x 10- 4
4.8 x 10-4
Fathepure et al. (1987)

TCE degradation
mixed methanotrophs,
aerobic, + CH4
Methylosinus trichosporium
aerobic, + CH4
Pseudomonas putida FI
aerobic, toluene induced
P. cepacia G4
aerobic, phenol induced
Alcaligenes eutrophus
AEOI06, aerobic,
phenol induced
0.06
Fogel et at. (1986)
> 521
Tsien et at. (1989)
2.6 1
6.8
Wackett and Gibson

(1988)
Folsom et al. (1990)
0.3
Harker and Kim (1990)
Nitrosomonas europaea,
0.99 1
aerobic, + NH3
1converted from g cells dry weight to mg protein by assuming a cell is 55% protein
(dry weight)
lconverted from g cells wet weight to mg protein by assuming that I g cells wet weight
is equivalent to 0.29 g cells dry weight, and using the above conversion to protein
tlConversion factors are from Ingraham et al. (1983), p. 3.
Sec. 9.4
299
faster than the monoculture of DCB-1 (Fathepure et al., 1987). Of the TCE-transforming
cultures, the mixed methanotrophic culture of Fogel et at. (1986) was grown at 20C,
while the other heterotrophic aerobic rates were assayed at 26 to 30 "C. The latter
range is probably better for the microbes. One cannot directly compare the unknown
methanotrophic strain(s) of the mixed culture with the M. trichosporium strain studied
by Tsien et at. (1989), but it is likely that part of the vast difference in their reported TCE
transformation rates is attributable to the optimization of culture conditions by Tsien et
at. (1989). In addition, not all biomass in a methanotrophic mixture will be active in
TCE transformation. An expectation of a 100-fo1d improvement in biotransformation
rates through optimization of conditions for microbial activity in the field is unrealistic.
Smaller improvements should be feasible, however.
A series of dynamic column studies of biofilm-mediated degradation (McCarty
et al., 1981; Bouwer and McCarty, 1983a; Bouwer and McCarty, 1983b; McCarty et
al., 1984; Vogel and McCarty, 1985; Vogel and McCarty, 1987; Bouwer and Wright,
1988) has produced valuable information on transformation pathways and also on how
to approach the modelling of transformation processes (e.g., see Bouwer and Wright,
1988). These studies have usually employed low concentrations of halocarbons (less
than 100 /LglL) that are similar to levels in contaminated groundwaters. However, the
primary substrate concentrations (from 1 up to 250 mg/L) and thus the amount of biomass
may be higher than normally encountered in the subsurface. The modelling aspect allows
predictions about real situations to be made. For example, Bouwer and Wright (1988)
found that shorter retention times increased the adaptation period and decreased the extent
of compound removed. They suggested that in the field, biotransformations of pollutants
may be less favorable in the vicinity of injection and extraction wells where local groundwater velocities are relatively high. These authors also believe that laboratory-derived
kiK, ratios (maximum specific substrate utilization ratelhalf-velocity constant) may be
useful in estimating pollutant half-lives in groundwater. Estimated half-lives were shortest
under aerobic and methanogenic conditions, and significantly longer under denitrifying
and sulfate-reducing conditions (Bouwer and Wright, 1988). Extrapolations from their
model run indicated that biotransformation could be significant even at the very low
microbial concentrations typical of subsurface environments.
9.4.5 Field Studies
The major advantage of field studies is that there the observed behavior is real and
demonstrable. On the other hand, field studies of actual contamination are often impaired
by a poor definition of input history. Moreover, the control and precision of laboratory
experiments cannot be duplicated, even in extensively monitored field experiments. A
recently described in situ microcosm testing device which is usable in fairly permeable
subsurface materials offers an intriguing intermediate study approach (Gillham et al.,
1990).
In terms of expanding our knowledge base, the ideal field study is probably one
that is highly controlled, and supported by supplementary laboratory studies using microcosms and perhaps enrichment cultures as well. The design of field experiments is often
300
Chap. 9
based on laboratory studies. Once field experiments such as those involving injections
of contaminants and tracers are initiated, they can be monitored, but often not substantially altered. The collected field data are used to construct hypotheses concerning what
is happening in the subsurface. Supplementary laboratory studies allow such hypotheses
to be tested. Plausible explanations of unforseen field occurrences may be more readily obtained using the relatively simple, easily manipulated environment of a laboratory
microcosm. Hopefully, consistent information is obtained from the field and laboratory
studies.
The study at the Moffett Field Naval Air Station is an example where an integrated
field-and-laboratory-study approach was taken (Semprini et al., 1987; Roberts et al., 1989;
Roberts et al., 1990; Lanzarone and McCarty, 1990; Semprini et al., 1990). Similarly,
controlled studies have been conducted at the Borden Landfill (Base Borden, Ontario)
study site, which is probably the most-studied aquifer in the world (e.g., see J. Hydrol.
1983, Vol. 63, No. 1/2 and Water Resources Research, 1986, Vol. 22, No. 13).
The site studied by Major et al. (1991) is not the result of a controlled spill, but
through fortuitous circumstances it provides a good study area for PCE. For example,
PCE's less chlorinated daughter compounds were never stored on-site. Also, the plume
is well isolated and occupies a relatively small area. Extensive analyses conducted on
groundwater from the site, including a search for possible biotransformation products and
substrates for microbial activity, have allowed hypotheses regarding on-site biotransformations to be made. Studies of laboratory microcosms have confirmed the hypotheses
and have suggested directions for future experimental investigation regarding the in situ
fate of the contaminants. Comparisons of this site, where biologically-mediated reductive
dechlorination is occurring, with other ostensibly similar sites where the halocarbons are
recalcitrant, should prove enlightening.
Where subsurface environments do not favor the rapid, natural biotransformation of halocarbon contaminants, there is an increasing interest in creating more favorable conditions,
i.e., in conducting in situ bioremediation of dissolved chlorinated solvents. The discussion of haloaliphatic biotransformation presented in the preceding sections makes it clear
that difficulties will be encountered in real-world bioremediation of sites contaminated
by these compounds. For example, an attempted aerobic bioremediation process will not
completely degrade highly-chlorinated species such as PCE; an anaerobic process can
degrade PCE, but it will also allow the accumulation of less chlorinated daughter products such as vinyl chloride. Complete degradation of a mixture of hexachlorobenzene,
PCE and chloroform to water-soluble metabolic products and CO 2 has been obtained
in a laboratory-scale, two-stage anaerobic-aerobic biofilm reactor (Fathepure and Vogel,
1991). Thus, the sequential application of both aerobic and anaerobic conditions in an
aquifer might be ideal, but large scale manipulation of the redox potential in a contaminated aquifer is not easily carried out. Alternatively, if we consider a simplified
scenario of a zone contaminated with a chloroethylene mixture, one approach might
be to initially encourage anaerobic conditions in the aquifer (especially near the source
Sec. 9.4
301
zone), favoring reductive dechlorination. This is a "go-with-the-flow" approach: assuming that there is enough metabolizable carbon, N, P, etc. (as is the case in many landfill
leachates), and given enough time, anaerobic conditions will evolve naturally (e.g., see
Lesage et al., 1990). Eventually, accumulation of less chlorinated daughter products and
perhaps methane should become apparent downgradient, and significant dilution of the
plume with uncontaminated, oxygenated water may begin to occur. Aerobic metabolism
by methanotrophs could then be encouraged in this aerobic zone, perhaps by pumping
O 2 or H 202 into the plume to degrade the less-chlorinated compounds. Alternatively,
a downgradient pump-and-treat system could bring the less chlorinated contaminants to
the surface for treatment in aerobic bioreactors.
Exploitation of aromatic-degrading bacteria to cometabolize TCE and other compounds remains a remediation possibility, but a major difficulty is that induction of the
desired oxygenases requires the provision of toluene or phenol. In some instances, these
aromatics may comprise part of the contaminant plume, but it is difficult to conceive of
regulatory agencies approving purposeful spills of one priority pollutant to clean up a
second. Less harmful means of induction must be found if these microorganisms are to
be commonly used in situ. A TCE-degrading mutant strain not requiring induction has
been developed (Shields and Reagin, 1992). Pump-and-treat systems employing bioreactors of pre-induced, aromatic-degrading bacteria can also be envisioned. Folsom and
Chapman (1991) recently demonstrated TCE degradation by phenol-fed P. cepacia 04
in a recirculating bioreactor; they reported an average degradation rate of 0.7 g TCE per
day per gram of cell protein.
A limited amount of research has been directed toward genetically-engineered,
TCE-degrading bacteria (Winter et al., 1989; Zylstra et al., 1989). The genes for toluene
dioxygenase of P. putida have been successfully cloned into Escherichia coli (Zylstra
et al., 1989). Interestingly, TCE degradation rates were more sustained in E. coli than
in P. putida, suggesting that E. coli suffered less metabolic damage. Recombination of
the toluene dioxygenase activity with a more TCE-tolerant host strain may be a fruitful
strategy for optimizing TCE biodegradation (Zylstra et al., ]989).
In situ remediation schemes which stimulate methane-oxidizing microbial populations are probably the most feasible of the currently available approaches. Methane
occurs naturally and is not a "priority pollutant", so that its controlled introduction into
the subsurface should be relatively acceptable. Additionally, because of the remarkably
broad substrate specificity of MMO, a wide range of pollutants might be affected by this
approach. Cometabolic attack on haloalkanes, haloalkenes, ethers, cyclic hydrocarbons,
and aromatic hydrocarbons has been demonstrated.
A comprehensive investigation of the feasibility of stimulating native methaneoxidizing populations to degrade chlorinated compounds in situ has been conducted
(Roberts et al., ]989; Roberts et al., ]990; Semprini et al., ]990). Methane and oxygen
were introduced into a shallow, confined, sand and gravel aquifer contaminated with
TCE, dichloroethylene and vinyl chloride. Alternative pulsing of oxygen and methane
kept clogging of the injection well under control. The temporary cessations of methane
feed inherent in the pulse system actually improved halocarbon transformation; since
methane and the halocarbons compete for the same enzyme, methane pulsing may have
302
Chap. 9
reduced the periods of methane inhibition of halocarbon degradation without starving

the population of primary substrate for extended periods. It was found that in the field,
the microbial population actively transformed halocarbons for only a short time after the
methane feed was stopped. In the small, biostimulated zone of the aquifer, a steady state
was achieved after about 2 weeks. The extents of transformation were approximately 95%
for vinyl chloride, 85% for trans-l,2-dichloroethylene, 40% for cis-l,2-dichloroethylene,
and 20% for TCE (Roberts et al., 1989).
When water that is contaminated with a chlorinated solvent is pumped to the
surface, common treatments have included air stripping and/or adsorption of the contaminants on charcoal or activated carbon. An attractive alternative treatment method
is to process the pumped water in bioreactors, thereby avoiding either the release of
the pollutants to the atmosphere or the need to dispose of contaminated adsorbent materials.
Microbial consortia that are fed methane, propane, and other organic substrates
have been used to remove TCE and dichloroethylenes from water in laboratory-scale
bioreactors. In one study, both propane and a propane-methane mixture were found
to be superior to methane alone as primary substrate, possibly because the consortium
was able to use propane more efficiently than methane, or because propane does not
compete as effectively as methane for the TCE-transforming enzymes (Phelps et al.,
1990). Using a methane-fed bioreactor with a 50 min residence time, Strandberg et al.
(1989) observed removal of more than 50% of the TCE and more than 90% of the cis1,2-dichloroethylene from a synthetic groundwater containing 1 mg/L of each compound.
Additional degradation of the TCE was obtained with liquid recycle in the bioreactor, but
an upper limit for TCE degradation of 90 to 95% was encountered. Although a satisfactory
explanation has not been advanced, other groups have apparently also observed this result
(Strandberg et al., 1989). Enrichments capable of dealing with relatively high halocarbon
levels can be developed. The bioreactors of Phelps et al. (1990) were fed 20 mg/L TCE,
and earlier work with batch cultures produced consortia capable of degrading more than
99% of TCE added at 50 mg/L (Fliermans et al., 1988). Interestingly, the early batch
cultures were not methanotrophic enrichments. Although the TCE-degrading consortia
were stimulated by methane, they could not use it as a sole carbon and energy source
(Fliermans et al., 1988).
It has been found that TCE degradation is not unduly inhibited by the presence
of dichloroethylene (Strandberg et al., 1989), and priority pollutants were not detected
among the bioreactor products (Strandberg et al., 1989; Phelps et al., 1990). These are
advantages that may not be obtainable in anaerobic bioremediation schemes because all
studies of anaerobic chloroethylene dehalogenation to date have clearly shown sequential transformation, not parallel transformation. The less chlorinated compounds (e.g.,
vinyl chloride) accumulate while more highly chlorinated compounds are dechlorinated,
and significant dechlorination of these daughter products does not begin until the parent has disappeared (e.g., see Major et al., 1991). Unfortunately, the daughter products
are also contaminants of concern. However, laboratory and field observations of partial vinyl chloride dechlorination to ethylene and ethane (Freedman and Gossett, 1989;
Major et al., 1991; de Bruin et al., 1992) and the virtually complete transformation
Sec. 9.4
303
of high levels of PCE to ethylene in laboratory enrichments (DiStefano et al., 1991)

provide optimism that useful in situ anaerobic bioremediation schemes will be forthcoming.
9.5 CONCLUSIONS
Many halogenated compounds are susceptible to abiotic transformations in the groundwater environment. The hydrolysis of bromomethane and several haloethanes including bromoethane, chloroethane and I, I, I-trichloroethane and I, I ,2,2-tetrachloroethane,
should limit the occurrence of such compounds in groundwaters. Some reactions produce
chloroethylenes, so hydrolysis alone may not completely detoxify contaminated groundwaters. Nucleophilic substitution of sulfur for both bromide and chloride also produces
mixed benefits, with mercaptans and vinyl chloride among the reported products.
Reductive reactions are perhaps the most significant reactions in groundwaters, with
both abiotic and biologically-mediated pathways. It may not be possible, and perhaps not
useful, to distinguish the abiotic versus biotic nature of these reactions, since biological
reactions are almost always required to produce the strongly reducing conditions under
which reduction reactions are favored. Only in anaerobic plumes should we expect significant abiotic transformation of halocarbons. Persistence under aerobic conditions is
anticipated. Many chlorinated solvents enter the subsurface without easily metabolized
substrates and these may produce persistent halocarbon plumes in aerobic groundwater.
The coupling of reductive dechlorination with FeD oxidation appears to be an exciting in
situ remedial possibility in such cases.
Microbial transformation of haloaliphatic compounds is not rare, and although in
many instances our knowledge of such processes is not derived from groundwater environments, we now have unequivocal evidence that biotransformation reactions critically
influence the subsurface behavior of halogenated compounds. Although our understanding of the various biotransformations that have been observed is sometimes incomplete,
available evidence suggests that it is possible to manipulate the environment of the microbial population(s) in ways that will encourage (or discourage) these microbially-mediated
reactions. Such manipulation can be used to help control the behavior of contaminant
plumes. In situ, a major problem will be controlling the hydrogeologically complex
subsurface environment so that the desired conditions are established and effectively
maintained throughout the contaminated area (see Chapter 2).
ACKNOWLEDGEMENTS
304
Chemical! Microbiological Transformations and Degradation
Chap. 9
9.6 REFERENCES
Acton, D. W. (1990) "Enhanced in situ biodegradation of aromatic and chlorinated aliphatic hydrocarbons in anaerobic, leachate-impacted groundwaters. M.Sc. Thesis, University of Waterloo,
Waterloo, Ontario.
Alexander, M. (1985) "Biodegradation of organic chemicals", Environ. Sci. Technol. 18, 106-111.
Alvarez-Cohen, L. and P. L. McCarty (1991) "Product toxicity and cometabolic competitive inhibition modeling of chloroform and trichloroethylene transformation by methanotrophic resting
cells", Appl. Environ. Microbiol. 57, 1031-1037.
Baek, N. H. and P. R. Jaffe (1989) "The degradation of trichloroethylene in mixed methanogenic
cultures", J. Environ. Qual. 18,515-518.
Bagley, D. M. and J. M. Gossett (1990) "Tetrachloroethene transformation to trichloroethene and
cis-dichloroethene by sulfate-reducing enrichment cultures", Appl. Environ. Microbiol. 56,25112516.
Balkwill, D. L., J. F. Fredrickson, and 1. M. Thomas (1989) "Vertical and horizontal variations
in the physiological diversity of the aerobic chemoheterotrophic bacterial microflora in deep
southeast coastal plain subsurface sediments", Appl. Environ. Microbiol. 55, 1058-1065.
Barbash, J. E. and M. Reinhard (1989) "Abiotic dehalogenation of 1,2-dichloroethane and 1,2dibromoethane in aqueous solution containing hydrogen sulfide", Environ. Sci. Technol. 23,
1349-1357.
Barrio-Lage, G. A., F. Z. Parsons, R. S. Nassar, and P. A. Lorenzo (1986) "Sequential dehalogenation of chlorinated ethenes", Environ. Sci. Technol. 20, 96-99.
Barrio-Lage, G. A., F. Z. Parsons, R. S. Nassar, and P. A. Lorenzo (1987) "Biotransformation of
trichloroethene in a variety of subsurface materials", Environ. Toxicol. Chem. 6, 571-578.
Baxter, R. M. (1989) "Reductive dehalogenation of environmental contaminants: a critical review",
Water Poll. Research J. Canada 24, 299-322.
Belay, N. and L. Daniels (1987) "Production of ethane, ethylene, and acetylene from halogenated
hydrocarbons by methanogenic bacteria", Appl. Environ. Microbiol. 53, 1604-1610.
Boethling, R. S. and M. Alexander (1979) "Effect of concentration of organic chemicals on their
biodegradation by natural communities", Appl. Environ. Microbiol. 37, 1211-1216.
Bouwer, E. J., B. E. Rittmann, and P. L. McCarty (1981) "Anaerobic degradation of halogenated
1- and 2-carbon organic compounds", Environ. Sci. Technol. 15, 596-599.
Bouwer, E. J. and P. L. McCarty (1983a) "Transformation of I-and 2-carbon halogenated aliphatic
organic compounds under methanogenic conditions", Appl. Environ. Microbiol. 45, 1286-1294.
Bouwer, E. J. and P. L. McCarty (l983b) "Transformations of halogenated organic compounds
under denitrification conditions", Appl. Environ. Microbiol. 45, 1295-1299.
Bouwer, E. J. and J. P. Wright (1988) "Transformation of trace halogenated aliphatics in anoxic
biofilm columns", J. Contam. Hydrol. 2, 155-169.
Brunner, W., D. Staub, and T. Leisinger (1980) "Bacterial degradation of dichlorornethane", Appl.
Environ. Microbiol. 40, 950-958.
Castro, C. E. and N. O. Belser (1990) "Biodehalogenation: oxidative and reductive metabolism of
I, I ,2-trichloroethane by Pseudomonas putida-biogeneration of vinyl chloride", Environ. Toxicol. Chem. 9,707-714.
Sec. 9.6
References
305
Cline, P. V. and D. R. Viste (1985) "Migration and degradation patterns of volatile organic compounds", Waste Management & Research 3, 351-360.
Cooper, W. J., M. Mehran, D. J. Riusech, and J. A. Joens (1987) "Abiotic transformation of
halogenated organics. I. Elimination reaction of 1,1,2,2-tetrachloroethane and formation of
I, I ,2-trichloroethane", Environ. Sci. Technol. 21, 1112-1114.
Criddle, C. S. and P. L. McCarty (1991) "Electrolytic model system for reductive dehalogenation
in aqueous environments", Environ. Sci. Technol. 25, 973-978.
Criddle, C. S., P. L. McCarty, M. C. Elliott, and J. F. Barker (1986) "Reduction of hexachloroethane
to tetrachloroethylene in groundwater", J. Contam. Hydrol. 1, 133-142.
Criddle, C. S., J. T. DeWitt, D. Grbic-Galic, and P. L. McCarty (1990a) "Transformation of carbon
tetrachloride by Pseudomonas sp. strain KC under denitrification conditions", Appl. Environ.
Microbiol. 56, 3240-3246.
Criddle, C. S., J. T. DeWitt, and P. L. McCarty (1990b) "Reductive dehalogenation of carbon
tetrachloride by Escherichia coli K-12", Appl. Environ. Microbiol. 56, 3247-3254.
Dalton, H. and D. 1. Stirling (1982) "Co-metabolism", Phil. Trans. R. Soc. Lond. B 297, 481-496.
Davis, J. W. and C. L. Carpenter (1990) "Aerobic biodegradation of vinyl chloride in groundwater
samples", Appl. Environ. Microbiol. 56, 3878-3880.
De Bruin, W. P., M. J. J. Kotterman, M. A. Posthumus, G. Schraa, and A. J. B. Zehnder (1992)
"Complete biological reductive transformation of tetrachloroethene to ethane", Appl. Environ.
Microbiol. 58, 1996-2000.
Dilling, W. L., N. B. Tefertiller, and G. J. Kallos (1975) "Evaporation rates and reactivities of
methylene chloride, chloroform, 1,1, I-trichloroethane, trichloroethylene, tetrachloroethylene,
and other chlorinated compounds in dilute aqueous solutions", Environ. Sci. Techno!. 9, 833-838.
DiStefano, T. D., J. M. Gossett, and S. H. Zinder (1991) "Reductive dechlorination of high concentrations of tetrachloroethene to ethene by an anaerobic enrichment culture in the absence of
methanogenesis", Appl. Environ. Microbiol. 57, 2287-2292.
Dolfing, J. and J. M. Tiedje (1987) "Growth yield increase linked to reductive dechlorination in a
defined 3-chlorobenzoate degrading methanogenic coculture", Arch. Microbiol. 149, 102-105.
Egli, c., R. Scholtz, A. M. Cook, and T. Leisinger (1987) "Anaerobic dechlorination of tetrachloromethane and 1,2-dichloroethane to degradable products by pure cultures of Desulfobacterium sp. and Methanobacterium sp.", FEMS Microbiol. Lett. 43, 257-261.
Egli, C; T. Tschan, R. Scholtz, A. M. Cook, and T. Leisinger (1988) "Transformation of tetrachloromethane to dichloromethane and carbon dioxide by Acetobacterium woodii", Appl. Environ. Microbiol. 54, 2819-2824.
Ensign, S. A., M. R. Hyman, and DJ. Arp (1992) "Cometabolic degradation of chlorinated alkenes
by alkene monooxygenase in a propylene-grown Xanthobacter strain", Appl. Environ. Microbiol.
58, 3038-3046.
Fathepure, B. Z. and S.A. Boyd (l988a) "Reductive dechlorination of perchloroethylene and the
role of methanogens", FEMS Microbiol. Lett. 49, 149-156.
Fathepure, B. Z. and S.A. Boyd (l988b) "Dependence of tetrachloroethylene dechlorination on
methanogenic substrate consumption by Methanosarcina sp. strain DCM", Appl. Environ. Microbiol. 54, 2976-2980.
Fathepure, B. Z. and T. M. Vogel (1991) "Complete degradation of polychlorinated hydrocarbons
by a two-stage biofilm reactor", Appl. Environ. Microbiol. 57, 3418-3422.
306
Chap. 9
Fathepure, B. Z., J. P. Nengu, and S. A. Boyd (1987) "Anaerobic bacteria that dechlorinate perchloroethene", Appl. Environ. Microbiol. 53, 2671-2674.
Fliermans, C. B., T. J. Phelps, D. Ringelberg, A. T. Mikell, and D. C. White (1988) "Mineralization
of trichloroethylene by heterotrophic enrichment cultures", Appl. Environ. Microbiol. 54, 17091714.
Fogel, M. M., A. R. Taddeo, and S. Fogel (1986) "Biodegradation of chlorinated ethenes by a
methane-utilizing mixed culture", Appl. Environ. Microbiol. 51, 720-724.
Folsom, B. R., P. 1. Chapman, and P. H. Pritchard (1990) "Phenol and trichloroethylene degradation
by Pseudomonas cepacia G4: kinetics and interactions between substrates", Appl. Environ.
Microbiol. 56, 1279-1285.
Folsom, B. R. and P. J. Chapman (1991) "Performance characterization of a model bioreactor for
the biodegradation of trichloroethylene by Pseudomonas cepacia G4", Appl. Environ. Microbiol.
57, 1602-1608.
Freedman, D. L. and J. M. Gossett (1989) "Biological reductive dechlorination of tetrachloroethylene and trichloroethylene to ethylene under methanogenic conditions", Appl. Environ. Microbiol.
55,2144-2151.
Freeze, R. A. and J. A. Cherry (1979) Groundwater, Prentice-Hall, Englewood Cliffs, NJ.
Fulthorpe, R. A. and R. C. Wyndham (1989) "Survival and activity of a 3-chlorobenzoate-catabolic
genotype in a natural system", Appl. Environ. Microbiol. 55, 1584-1590.
Galli, R. and P. L. McCarty (l989a) "Biotransformation of I, I, I-trichloroethane, trichloromethane
and tetrachloromethane by a Clostridium sp.", Appl. Environ. Microbiol. 55, 837-844.
Galli, R. and P. L. McCarty (1989b) "Kinetics of biotransformation of I, I, I-trichloroethane by
Clostridium sp. strain TCAIIB", Appl. Environ. Microbiol. 55, 845-851.
Gantzer, C. J. and L. P. Wackett (1991) "Reductive dechlorination catalyzed by bacterial transitionmetal coenzymes", Environ. Sci. Technol. 25, 715-722.
Genthner, B. R. S., W. A. Price, and P. H. Pritchard (1989), "Anaerobic degradation of chloroaromatic compounds under a variety of enrichment conditions", Appl. Environ. Microbiol. 55,
1466-1471.
Ghiorse, W. C. and D. L. Balkwill (1985) "Microbiological characterization of subsurface environments", In: Ground Water Quality, C. H. Ward, W. Giger, and P. L. McCarty (Eds.) Wiley,
New York, pp. 387-398.
Ghiorse, W. C. and J. T. Wilson (1988) "Microbial ecology of the terrestrial subsurface", Adv.
Appl. Microbiol. 33, 107-172.
Gillham, R. W. and S. F. O'Hannesin (1992) "Metal-catalyzed abiotic degradation of halogenated
organic compounds", presented at the 1992 International Assoc. Hydrogeologists Conf. "Modem
Trends in Hydrogeology", Hamilton, Ontario, Canada, May 10-13. Available as paper Q-23 from
Waterloo Centre for Groundwater Research, Univ. of Waterloo, Waterloo, Ontario N2L 3G 1.
Gillham, R. W., R. C. Starr, and D. 1. Miller (1990) "A device for in situ determination of
geochemical transport parameters. 2. Biochemical reactions", Ground Water 28, 858-862.
Gossett, J. M. and D. L. Freedman (1989) "Biotransformation of dichloromethane in methanogenic
systems", Third Int. Meeting, NATO/CCMS Pilot Study on Demonstration of Remedial Action
Technologies for Contaminated Land and Groundwater, Nov 6-9, 1989, Montreal.
Grady, C. P. (1985) "Biodegradation: Its measurement and microbiological basis", Biotechnol.
Bioeng. 27, 660-674.
Sec. 9.6
References
307
Haag, W. R. and T. Mill (1988) "Some reactions of naturally occurring nucleophiles with haloalkanes in water", Environ. Toxicol. Chem. 7, 917-924.
Haber, C. L., L. N. Allen, S. Zhao, and R. S. Hanson (1983) "Methylotrophic bacteria: biochemical
diversity and genetics", Science 211, 1147-1153.
Harker, A. R. and Y. Kim (1990) "Trichloroethylene degradation by two independent aromatic
degrading pathways in Alcaligenes eutrophus JMP134", Appl. Environ. Microbiol. 56, 1179118\.
Henry, S. M. and D. Grbic-Galic (1991) "Influence of endogenous and exogenous electron donors
and trichloroethylene oxidation toxicity on trichloroethylene oxidation by methanotrophic cultures from a groundwater aquifer", Appl. Environ. Microbiol. 57,236-244.
Henson, J. M., M. V. Yates, and J. W. Cochran (1989) "Metabolism of chlorinated methanes,
ethanes, and ethylenes by a mixed bacterial culture growing on methane", J. Ind. Microbiol. 4,
29-35.
Henson, J. M., M. V. Yates, J. W. Cochran, and D. L. Shackleford (1988) "Microbial removal of
halogenated methanes, ethanes, and ethylenes in an aerobic soil exposed to methane", FEMS
Microbiol. Ecol. 53, 193-201.
Hirsch, P. and E. Rades-Rohkohl (1983) "Microbial diversity in a groundwater aquifer in Northern
Germany", Dev. Ind. Microbiol. 24, 183-200.
Holliger, c., G. Schraa, E. Stupperich, A. J. M. Starns, and A. J. B. Zehnder (l992a) "Evidence for the involvement of corrinoids and factor F430 in the reductive dechlorination of
1,2-dichloroethane by Methanosarcina barkeri", J. Bacteriol. 174,4427-4434.
Holliger, c, S. W. M. Kengen, G. Schraa, A. J. M. Starns, and A. 1. B. Zehnder (1992b) "Methylcoenzyme M reductase of Methanobacterium thermoautotrophicum H catalyzes the reductive
dechlorination of 1,2-dichloroethane to ethylene and chloroethane", J. Bacteriol. 174, 44354443.
Hou, C.T., R. N. Patel, A. I. Laskin, and N. Barnabe (1979) "Microbial oxidation of gaseous
hydrocarbons: Epoxidation of C2 to C4 n-alkenes by methylotrophic bacteria", Appl. Environ.
Microbial. 38, 127-134.
Hungate, R. E. (1969) "A roll tube method for cultivation of strict anaerobes", Methods Microbiol.
3B, 117-132.
Ingraham, J. L., O. Maaloe, and F. C. Neidhardt (1983) Growth of the Bacterial Cell, Sinauer
Associates Inc., Sunderland, MA.
Jafvert, C. T. and N. L. Wolfe (1987) "Degradation of selected halogenated ethanes in anoxic
sediment-water systems", Environ. Toxicol. Chem. 6, 827-837.
Janssen, D. B., A. Scheper, L. Dijkhuizen, and B. Witholt (1985) "Degradation of halogenated
aliphatic compounds by Xanthobacter autotrophicus GJlO", Appl. Environ. Microbiol. 49, 673677.
Janssen, D. B., D. Jager, and B. Witholt (1987) "Degradation of n-haloa1kanes and a, w-dihaloalkanes by wild-type and mutants of Acinetobacter sp. strain GJ70", Appl. Environ. Microbiol.
53, 561-566.
Jeffers, P. M., L. M. Ward, L. M. Woytowitch, and N. L. Wolfe (1989) "Homogeneous hydrolysis
rate constants for selected chlorinated methanes, ethanes, ethenes, and propanes", Environ. Sci.
Technol. 23, 965-969.
Jones, R. E., R. E. Beeman, and J. M. Suflita (1989) "Anaerobic metabolic processes in the deep
terrestrial subsurface", Geomicrobiol. J. 7, 117-13\.
308
Chap. 9
Kastner, M. (1991) "Reductive dechlorination oftri- and tetrachloroethylenes depends on transition

from aerobic to anaerobic conditions", Appl. Environ. Microbiol. 57, 2039-2046.
Kleopfer, R. D., D. M. Eastley, B. B. Haas, T. G. Deihl, D. E. Jackson, and C. J. Wurrey (1985)
"Anaerobic degradation of trichloroethylene in soil", Environ. Sci. Technol. 19, 277-280.
Kobayashi, H. and B. E. Rittmann (1982) "Microbial removal of hazardous organic compounds",
Environ. Sci. Technol. 16, 170A-181 A.
Kriegman-King, M. R. and M. Reinhard (1992) "Transformation of carbon tetrachloride in the
presence of sulfide, biotite, and vermiculite", Environ. Sci. Technol. 26,2198-2206.
Krone, U. E., K. Laufer, R. K. Thauer, and H. P. C. Hogenkamp (1989a) "Coenzyme F430 as a possible catalyst for the reductive dehalogenation of chlorinated C, hydrocarbons in methanogenic
bacteria", Biochemistry 28, 10061-10065.
Krone, U. E., R. K. Thauer, and H. P. C. Hogenkamp (1989b) "Reductive dehalogenation of
chlorinated Cj-hydrocarbons mediated by corrinoids", Biochemistry 28, 4908-4914.
Kuznetsov, S. I., G. A. Dubinina, and N. A. Lapteva (1979) "Biology of oligotrophic bacteria",
Annu. Rev. Microbiol. 33, 377-387.
Lanzarone, N. A. and P. L. McCarty (1990) "Column studies on methanotrophic degradation of
trichloroethene and 1,2-dichloroethane", Ground Water 28,910-919.
LaPat-Polasko L. T., P. L. McCarty, and A. 1. B. Zehnder (1984) "Secondary substrate utilization
of methylene chloride by an isolated strain of Pseudomonas sp.", Appl. Environ. Microbiol. 47,
825-830.
Large, P. J. (1983) Methylotrophy and Methanogenesis, American Society for Microbiology, Washington, D.C.
Leisinger, T. (1983) "Microorganisms and xenobiotic compounds", Experientia 39, 1183-1191.
Lesage, S., R. E. Jackson, M. W. Priddle, and P. G. Riemann (1990) "Occurrence and fate of
organic solvent residues in anoxic groundwater at the Gloucester Landfill, Canada", Environ.
Sci. Technol. 24, 559-566.
Little, C. D., A. V. Palumbo, S. E. Herbes, M. E. Lidstrom, R. L. Tyndall, and P. J. Gilmer
(1988) "Trichloroethylene biodegradation by a methane-oxidizing bacterium", Appl. Environ.
Microbiol. 54, 951-956.
Lyman, W. J., W. F. Ruhl, and D. H. Rosenblatt (1982) Handbook ofChemical Property Estimation
Methods, McGraw-Hili, New York, New York.
Mabey, W. and T. Mill (1978) "Critical review of hydrolysis of organic compounds in water under
environmental conditions", Phys. Chern. Ref Data 7, 383-415.
Major, D. W., E. W. Hodgins, and B. J. Butler (1991) "Field and laboratory evidence of in
situ biotransformation of tetrachloroethene to ethene at a chemical transfer facility in North
Toronto". In: On-site Bioreclamation Processes for Xenobiotic and Hydrocarbon Treatment, R.
E. Hinchee and R. F. Olfenbuttel (Eds.), Butterworth-Heinemann, Boston, pp. 147-171.
Mayfield, C. I. and R. L. Aldworth (1974) "Acetylene reduction in artificial soil aggregates amended
with cellulose, wheat straw, and xylan", Can. 1. Microbiol. 20, 1503-1507.
McCarty, P. L., M. Reinhard, and B. E. Rittmann (1981) "Trace organics in groundwater", Environ.
Sci. Technol. 15, 40-51.
McCarty, P. L., B. E. Rittmann, and E. J. Bouwer (1984) "Microbiological processes affecting
chemical transformations in groundwater". In: Ground Water Pollution Microbiology, G. Bitton
and C. P. Gerba (Eds.), Wiley and Sons, Toronto, pp. 89-115.
Sec. 9.6
References
309
McNabb, J. F. and W. J. Dunlap (\975) "Subsurface biological activity in relation to ground-water

pollution", Ground Water 13, 33-44.
Mikesell, M. D. and S. A. Boyd (1990) "Dechlorination of chloroform by Methanosarcina strains",
Appl. Environ. Microbiol. 56, 1198-120 I.
Miller, R. E. and F. P. Guengerich (\ 982) "Oxidation of trichloroethylene by liver cytochrome P450: evidence for chlorine migration in a transition state not involving trichloroethylene oxide",
Biochemistry 21, 1090-1097.
Morgan, P. and R. J. Watkinson (\ 989) "Microbiological methods for the cleanup of soil and
ground water contaminated with halogenated organic compounds", FEMS Microbiol. Rev. 63,
277-300.
Nelson, M. J. K. and J. V. Kinsella (\ 990) "In situ degradation of trichloroethylene", Abstract of a
paper presented at the First International Symposium on Microbiology of the Deep Subsurface,
Orlando, Fla., Jan. 15-19, 1990.
Nelson, M. J. K., S. O. Montgomery, E. J. O'Neill, and P. H. Pritchard (1986) "Aerobic metabolism
of trichloroethylene by a bacterial isolate", Appl. Environ. Microbiol. 52, 383-384.
Nelson, M. J. K., S. O. Montgomery, W. R. Mahaffey, and P. H. Pritchard (\987) "Biodegradation
of trichloroethylene and involvement of an aromatic biodegradative pathway", Appl. Environ.
Microbiol. 53, 949-954.
Nelson, M. J. K., S. O. Montgomery, and P. H. Pritchard (\988) "Trichloroethylene metabolism
by microorganisms that degrade aromatic compounds", Appl. Environ. Microbiol. 54,604-606.
O'Hannesin, S. F. and R. W. Gillham (\992) "A permeable reaction wall for in situ degradation
of halogenated organic compounds", presented at the 1992, 45th Can. Geotech. Soc. Conf.,
Toronto, Ontario, Oct. 25-28, 1992. Available as paper Q-25 from Waterloo Centre for Groundwater Research, Univ. of Waterloo, Waterloo, Ontario N2L 3G I.
Oldenhuis, R., J. Y. Oedzes, J. J. van der Waarde, and D. B. Janssen (\991) "Kinetics of chlorinated
hydrocarbon degradation by Methylosinus trichosporium OB3b and toxicity of trichloroethylene", Appl. Environ. Microbioi. 57,7-14.
Parsons, F. and G. B. Lage (\985) "Chlorinated organics in simulated groundwater environments",
J. Am. Water Works Assoc. 77,52-59.
Parsons, F., P. R. Wood, and J. DeMarco (1984) "Transformations of tetrachloroethene and
trichloroethene in microcosms and groundwater", J. Am. Water Works Assoc. 76, 56-59.
Phelps, T. J., D. Ringelberg, D. Hedrick, J. Davis, C. B. Fliermans, and D. C. White (\988)
"Microbial biomass and activities associated with subsurface environments contaminated with
chlorinated hydrocarbons", Geomicrobiol. J. 6, 157-170.
Phelps, T. J., J. J. Niedzielski, R. M. Schram, S. E. Herbes, and D. C. White (\990) "Biodegradation
of trichloroethylene in continuous-recycle expanded-bed bioreactors", Appl. Environ. Microbial.
56, 1702-1709.
Pignatello, J. J. (1986) "Ethylene dibromide mineralization in soils under aerobic conditions", Appl.
Poindexter, J. S. (\981) "Oligotrophy. Fast and famine existence", Adv. Microb. Ecol. 5,63-89.
Postgate, J. R. (\984) The Sulphate-Reducing Bacteria, 2nd edition, Cambridge University Press,
Cambridge, U.K.
Rasche, M. E., M. R. Hyman, and D. J. Arp (l990a) "Biodegradation of halogenated hydrocarbon
fumigants by denitrifying bacteria", Appl. Environ. Microbiol. 56, 2568-2571.
310
Chap. 9
Rasche, M. E., R. E. Hicks, M. R. Hyman, and D. J. Arp. (l990b) "Oxidation of monohalogenated

ethanes and n-chlorinated alkanes by whole cells of Nitrosomonas europaea", J. Bacteriol. 172,
5368-5373.
Rasche, M. E., M. R. Hyman, and D. J. Arp (1991) "Factors limiting aliphatic chlorocarbon
degradation by Nitrosomonas europaea: Cometabolic inactivation of ammonia monooxygenase
and substrate specificity", Appl. Environ. Microbiol. 57, 2986-2994.
Reinhard, M., G. P. Curtis, and M. R. Kriegman (1990) "Abiotic reductive dechlorination of
carbon tetrachloride and hexachloroethane by environmental reductants", Project Summary,
EPA/600/S2-90/040. Sept. 1990.
Roberts, A. L. and P. M. Gschwend (1991) "Mechanism of pentachloroethane dehydrochlorination
to tetrachloroethylene", Environ. Sci. Technol. 25, 76-86.
Roberts, A. L., P. N. Sanborn, and P. M. Gschwend (1992) "Nucleophilic substitution reactions of
dihalomethanes with hydrogen sulfide species", Environ. Sci. Technol. 26, 2263-2274.
Roberts, P. V., M. N. Goltz, and D. M. Mackay (1986) "A natural gradient experiment on solute
transport in a sand aquifer. III. Retardation estimates and mass balances for organic solutes",
Water Resources Research 22, 2047-2058.
Roberts, P. V., L. Sernprini, G. D. Hopkins, D. Grbic-Galic, P. L. McCarty, and M. Reinhard (1989)
"In-situ aquifer restoration of chlorinated aliphatics by methanotrophic bacteria", Technical
Report No. 310, June 1989, Department of Civil Engineering, Stanford University, Stanford,
California.
Roberts, P. V., G. D. Hopkins, D. M. Mackay, and L. Semprini (1990) "A field evaluation of insitu biodegradation of chlorinated ethenes: Part 1. Methodology and field site characterization",
Ground Water 28,591-604.
Schmidt, S. K. and M. Alexander (1985) "Effect of dissolved organic carbon and second substrates
on the biodegradation of organic compounds at low concentrations", Appl. Environ. Microbiol.
49, 822-827.
Scholz-Muramatsu, H., R. Szewzyk, U. Szewzyk, and S. Gaiser (1990) "Tetrachloroethylene as
electron acceptor for the anaerobic degradation of benzoate", FEMS Microbiol. Lett. 66, 81-86.
Schwarzenbach, R. P., W. Giger, C. Schaffner, and O. Wanner (1985) "Groundwater contamination
by volatile halogenated alkanes: abiotic formation of volatile sulfur compounds under anaerobic
conditions", Environ. Sci. Technol. 19, 322-327.
Sernprini, L., P. V. Roberts, G. D. Hopkins, and D. M. Mackay (1987) "A field evaluation of
in-situ biodegradation methodologies for the restoration of aquifers contaminated with chlorinated aliphatic compounds: results of a preliminary investigation", Dept. of Civil Engineering,
Stanford Univ. Technical Report No. 302.
Semprini, L., P. V. Roberts, G. D. Hopkins, and P. L. McCarty (1990) "A field evaluation of
in-situ biodegradation of chlorinated ethenes: Part 2. The results of biostimulation and biotransformation experiments", Ground Water 28, 715-727.
Shields, M. S. and M. J. Reagin (1992), "Selection of a Pseudomonas cepacia strain constitutive
for the degradation of trichloroethylene", Appl. Environ. Microbiol. 58, 3977-3983.
Shields, M. S., S. O. Montgomery, P. J. Chapman, S. M. Cuskey, and P. H. Pritchard (1989)
"Novel pathway of toluene catabolism in the trichloroethylene-degrading bacterium G4", Appl.
Simkins, S. and M. Alexander (1984) "Models for mineralization kinetics with the variables of
substrate concentration and population density", Appl. Environ. Microbiol. 47, 1299-1306.
Sec. 9.6
References
311
Slater, J. H. and A. T. Bull (1982) "Environmental microbiology: biodegradation", Phil. Trans.

Roy. Soc. Lond. B, 297, 575-597.
Strandberg, G. W., T. L. Donaldson, and L. L. Farr (1989) "Degradation of trichloroethylene
and trans-I,2-dichloroethylene by a methanotrophic consortium in a fixed-film, packed-bed
bioreactor", Environ. Sci. Technol. 23, 1422-1425.
Stucki, G., R. Galli, H.-R. Ebersold, and T. Leisinger (1981) "Dehalogenation of dichloromethane
by cell extracts of Hyphomicrobium DMS", Arch. Microbiol. 130, 366-371.
Stucki, G., U. Krebser, and T. Leisinger (1983) "Bacterial growth on 1,2-dichloroethane", Experientia 39, 1271-1273.
Suflita, J. M., S. A. Gibson, and R. E. Beeman (1988) "Anaerobic biotransformations of pollutant
chemicals in aquifers", J. Ind. Microbiol. 3, 179-194.
Tabak, H. H., S. A. Quave, C. 1. Mashni, and E. F. Barth (1981) "Biodegradability studies with
organic priority pollutant compounds", J. Water Poll. Control Fed. 53, 1503-1518.
Tardif, G., C. W. Greer, D. Labbe, and P. C. K. Lau (1991) "Involvement of a large plasmid in the
degradation of 1,2-dichloroethane by Xanthobacter autotrophicus", Appl. Environ. Microbiol.
57,1853-1857.
Tsien, H.-C., G. A. Brusseau, R. S. Hanson, and L. P. Wackett (1989) "Biodegradation of trichloroethylene by Methylosinus trichosporium OB3b", Appl. Environ. Microbiol. 55, 3155-3161.
Vannelli, T., M. Logan, D. M. Arciero, and A. B. Hooper (1990) "Degradation of halogenated
aliphatic compounds by the ammonia-oxidizing bacterium Nitrosomonas europaea", Appl. Environ. Microbiol. 56, 1169-1171.
Ventullo, R. M. and R. J. Larson (1985) "Metabolic diversity and activity of heterotrophic bacteria
in ground water", Environ. Toxicol. Chem. 4, 759-771.
Vogel, T. M. and P. L. McCarty (1985) "Biotransformation of tetrachloroethylene to trichloroethylene, dichloroethylene, vinyl chloride, and carbon dioxide under methanogenic conditions",
Appl. Environ. Microbiol. 49, 1080-1083.
Vogel, T. M. and P. L. McCarty (1987) "Abiotic and biotic transformations of 1,1, I-trichloroethane
under methanogenic conditions", Environ. Sci. Technol. 21, 1208-1213.
Vogel, T. M., C. S. Criddle, and P. L. McCarty (1987) "Transformation of halogenated aliphatic
compounds", Environ. Sci. Technol. 21, 722-736.
Wackett, L. P. and D. T. Gibson (1988) "Degradation of trichloroethylene by toluene dioxygenase
in whole-cell studies with Pseudomonas putida Fl ", Appl. Environ. Microbiol. 54, 1703-1707.
Wackett, L. P. and S. R. Householder (1989) "Toxicity of trichloroethylene to Pseudomonas putida
Fl is mediated by toluene dioxygenase", Appl. Environ. Microbiol. 55, 2723-2725.
Waksman, S. A. (1916) "Bacterial numbers in soil at different depths, and in different seasons of
the year", Soil Sci. 1, 363-380.
Wang, T. C. and C. K. Tan (1990) "Reduction of halogenated hydrocarbons with magnesium
hydrolysis process", Bull. Environ. Contam. Toxicol. 45,149-156.
Wang, Y.-S., R. V. Subba-Rao, and M. Alexander (1984) "Effect of substrate concentration and organic and inorganic compounds on the occurrence and rate of mineralization and cometabolism",
Appl. Environ. Microbiol. 47, 1195-1200.
Weintraub, R. A., G. W. Jex, and H. A. Moye (1986) "Chemical and microbial degradation of 1,2dibromoethane (EDB) in Florida ground water, soil, and sludge". In: Evaluation of Pesticides
312
Chap. 9
in Ground Water W. Y. Gamer, R. C. Honeycutt, and H. A. Nigg (Eds.), American Chemical

Society, Washington, D.C., pp. 294-310.
Wiggins, B. A., S. H. Jones, and M. Alexander (1987) "Explanations for the acclimation period
preceding the mineralization of organic chemicals in aquatic environments", Appl. Environ.
Microbiol. 53,791-796.
Wiggins, B. A. and M. Alexander (1988) "Role of chemical concentration and second carbon
sources in acclimation of microbial communities for biodegradation", Appl. Environ. Microbiol.
54, 2803-2807.
Wilson, B. H., G. B. Smith, and J. F. Rees (1986) "Biotransformations of selected alkylbenzenes
and halogenated aliphatic hydrocarbons in methanogenic aquifer material: a microcosm study",
Environ. Sci. Technol. 20, 997-1002.
Wilson, J. T. and B. H. Wilson (1985) "Biotransformation of trichloroethylene in soil", Appl.
Wilson, J. T., J. F. McNabb, D. L. Balkwill, and W. C. Ghiorse (1983) "Enumeration of bacteria
indigenous to a shallow water-table aquifer", Ground Water 21, 134-142.
Wilson, J. T., J. F. McNabb, J. W. Cochran, T. H. Wang, M. B. Tomson, and P. B. Bedient (1985)
"Influence of microbial adaptation on the fate of organic pollutants in ground water", Environ.
Toxicol. Chern. 4, 721-726.
Winter, R. B., K.-M. Yen, and B. D. Ensley (1989) "Efficient degradation of trichloroethylene by
a recombinant Escherichia coli", Biotechnology 7, 282-285.
Wyndham, R. C. (1986) "Evolved aniline catabolism in Acinetobacter calcoaceticus during continuous culture of river water", Appl. Environ. Microbiol. 51,781-789.
Zeikus, J. G., R. Kerby, and J. A. Krzycki (1985) "Single-carbon chemistry of acetogenic and
methanogenic bacteria", Science 227, 1167-1173.
Zylstra, G. J., L. P. Wackett, and D. T. Gibson (1989) "Trichloroethylene degradation by Escherichia coli containing the cloned Pseudomonas putida FI toluene dioxygenase genes", Appl.
10
The Effects of Chlorinated
Solvents on the Permeability
of Clays
Tammy A. Middleton 1 and John A. Cherry?
1 Golder
Associates Ltd.
180 Columbia St. W.
Waterloo, Ontario
Canada, N2L 3L3

Waterloo, Ontario
Canada N2L 3G1
Page
10.1
10.2
10.3
10.4
10.5
10.6
10.7
10.8
Introduction
10.2.1 General
Water-Wet V5. Dry Clays, and the Issue of Wettability
Relevant Properties of Organ ic Chemicals
10.5.2.1 General
determ ination of permeab iIity
10.5.2.4 Permeability tests on "dry" clays/organic mixtures
10.5.2.5 Permeability tests of water-wet clays and NAPLs
10.5.2.6 Permeability tests using aqueous solutions of NAPL chemicals
10.5.2.7 Permeability tests of clays and water-miscible organic fluids
10.5.2.8 The influence of clay mineral type on permeability to
organic fluids
Field Studies
Conclusions
References
314
315
315
315
316
317
318
318
318
319
319
320
322
323
323
328
328
330
330
331
333
313
314
The Effects of Chlorinated Solvents on the Permeability of Clays
Chap. 10
ABSTRACT
There is disagreement in the scientific community as to whether liquid chlorinated solvents and other organic non-aqueous phase liquids (NAPLs) can cause clay soils to shrink
and crack, thereby greatly increasing their permeability. However, a careful consideration of factors such as water miscibility of the organic liquids of interest as wel1 as the
details of the laboratory experiments that have been carried out allows a resolution of
many of the contradictions that exist in the literature. It can be shown that in hydrated
(i.e., wet) clay soils, chlorinated solvents and other organic NAPLs apparently do not
cause permeability increases. This is likely due to the exclusion of the organic liquid
from the clay double layers. When NAPL solvents are forced through clay samples, the
permeability increases that have been observed can be attributed to the mechanical formation of fractures due to large fluid gradients and low lateral confining stresses on the
samples. Exposure to aqueous solutions of these organic compounds does not influence
permeability because solutions at or below the solubility limit are too dilute to affect clay
double layers. It can be concluded that chlorinated solvents in NAPL form are unlikely to
cause shrinking, cracking, or permeability increases in natural or engineered clay barriers
under field conditions. However, further study is required on the chemical and physical
effects of such liquids on smectitic clay slurries.
10.1 INTRODUCTION
In preceding chapters of this book, the movement and fate of chlorinated solvents in
aquifers and aquitards have been described. Various conceptual model situations have
been discussed, including cases where accumulations of free-phase solvent rest on lowpermeability strata, or penetrate into fractures in low-permeability matrix materials. In
many field situations, low-permeability strata or matrix materials contain clay minerals.
In the early 1980s, laboratory research suggested that some types of industrial organic
chemicals, including chlorinated solvents, could chemical1y interact with clay-rich materials in such a way that shrinkage and cracking would take place in these materials
(Green et aI., 1981; Brown et al., 1984). It was concluded that such cracks would lead to
large permeability increases in these clays. Sponsored in part by the U.S.EPA, these results had much influence, even though they were preliminary and controversial in nature.
Thus, many groundwater scientists and engineers came to believe that accumulations
of chlorinated solvent chemicals on clay strata and in fractured, clayey deposits could
cause increases of permeability in these materials, resulting in deeper movement of the
chemicals. These results have prompted further laboratory studies of this topic.
The published results of the initial and subsequent studies of the effects of organic
chemicals on clays reveal several important inconsistencies, and as a result, some of the
various conclusions are contradictory. This chapter provides a review of the published literature in this area. Areas of contradiction and controversy are identified and, where possible, explanations for the inconsistencies are provided. The study of clay/organic chemical
interactions spans many disciplines, including col1oid and surface science, geotechnical
engineering, clay mineralogy, organic chemistry, and hydrogeology; the interdisciplinary
Sec. 10.2
315
nature of this topic has contributed to the lack of agreement in the literature. Although
few of the studies used chlorinated solvents, many of them used other organic chemicals
that provide results directly relevant to solvent-clay interactions. With rare exceptions,
the literature is based on laboratory studies. The extension of the results from the laboratory experiments to field situations will need to be carried out carefully because field
conditions such as soil fabric, confining stress, fluid gradient, and time scales of exposure
cannot easily be replicated in the laboratory.
This chapter first presents some necessary background information on the chemical mechanisms controlling the swelling and shrinkage of clay minerals, the differences
between "dry" and "water-wet" clays, and the properties of organic chemicals. This is
followed by a review of laboratory and field investigations. The term "permeability" is
used here to denote the intrinsic permeability. This is distinct from the "hydraulic conductivity" which is the proportionality constant in Darcy's Law between groundwater
velocity and hydraulic head (Freeze and Cherry, 1979). Since the hydraulic conductivity
value is partially dependent on the effects of the density and viscosity of the permeating
fluid, a change in this value does not necessarily imply a change in the actual transmissive properties of the porous medium. For this reason it preferable to use the intrinsic
permeability value, which depends only on the porous medium.
10.2 MECHANISMS CONTROLLING THE SWELLING OF CLAYS

10.2.1 General
It is well known that brines and other liquids with high electrolyte concentrations can
increase the permeability of clay soils by chemically causing shrinking and cracking
to occur (Acar and Seals, 1984; Mitchell and Madsen, 1987; and others). Chemical
interactions can also occur between organic chemicals and clay minerals. There are
two types of processes to be considered here: 1) interlayer processes (i.e., intraparticle
processes) that occur in "swelling clays"; and 2) interparticle processes (i.e., between
particle processes) that can occur in all clay minerals. Both types of processes can cause
shrinkage of the bulk clay soil and, as a result, enlargement of pore spaces or the opening
of tension cracks.
The "swelling clays" (e.g., the smectites and some vermiculites) undergo intracrystalline
expansion or contraction in the crystal c-axis direction which can result in significant
volume changes of the clay mass. The basal spacing, which is the degree of intracrystalline expansion of clays, can be affected by organic liquids. The basal spacings of
swelling clays depend in part on the number and type of molecules adsorbed between
the structural layers of the clays. The intrusion of organic molecules into the interlayer
spaces of swelling clays can potentially result in either an increase or a decrease in the
basal spacing. Van Olphen (1977) and Theng (1974) provide reviews of this process.
316
Chap. 10

Interparticle swelling, also referred to as double layer swelling, affects all types of clay.
This process is described by the Gouy-Chapman theory of the diffuse double layer ("double layer theory"). Van Olphen (1977) and Mitchell (1976) provide reviews of this theory.
In short, the surfaces of clay minerals carry a net negative charge. Cations accumulate at
the clay surfaces due to electrostatic attraction, but there is also a tendency for the cations
to diffuse away from the surfaces into areas of lower cation concentration. Anions are
electrostatically repelled from the negatively charged surfaces. The negatively-charged
clay surface together with the adjoining, equilibrium, aqueous distribution of cations and
anions is the diffuse double layer. When two negatively-charged clay particles approach
one other, there is an electrostatic repulsion. The repulsion is roughly proportional to the
inverse of the separation distance between the particles (Van Olphen, 1977). These repulsive forces influence the arrangement of individual clay particles, and therefore the
permeability of bulk clays.
Some aspects of the behavior of clay particles in soils have been inferred from
the study of clay particles in colloidal suspensions. The study of clay colloids shows
that as the thickness of the double layers increases, the repulsive forces between clay
particles also increases, and the clays tend to form a dispersed fabric with evenly spaced
particles and small pore sizes. As the thickness of the double layer decreases, the repulsive
forces between clay particles decrease, and the particles tend to form a flocculated fabric
containing aggregates of particles with large pores in and between the aggregates (Van
Olphen, 1977). A soil with a flocculated fabric has a greater permeability than a soil
with a dispersed fabric.
One of the factors that determines the thickness of the double layer is the dielectric
constant of the pore liquid. The dielectric constant of a liquid is an inverse measure of
the permeability of the liquid to electrical forces (Brownlow, 1979). A more polar liquid
will have a higher dielectric constant and will result in a thicker double layer. Water,
the usual pore liquid in the subsurface, has a high dielectric constant (see Table 10.1).
Some common light non-aqueous phase liquids (LNAPLs) such as benzene, xylene
and toluene, and dense non-aqueous phase liquids (DNAPLs) such as chlorinated solvents like trichloroethylene (TCE), perchloroethylene (PCE), and carbon tetrachloride
have very low dielectric constants.
If an organic liquid of low dielectric constant enters the pore spaces of a clay
soil and the organic molecules replace the water molecules in the double layers, then
theory predicts that there will be a shrinkage of the double layer surrounding each
clay particle, and a tendency for the particles to move more closely together. When
there are no significant effective stresses on the soil and clay aggregates in the soil,
then cracks or macropores may open around the aggregates with the result that there
will be an increase in permeability. In contrast, a significant effective stress acting
on the soil can prevent the opening of cracks and macropores by causing soil consolidation when double layer shrinkage weakens soil strength. This can result in bulk
soil shrinkage and a reduced permeability. Effective stress is considered in more detail
below.
Sec. 10.3
vs. Dry Clays, and the Issue of Wettability
Water-wet
TABLE 10.1
20-25 -c.
317
Properties of Some Common Nonionic Organic Compounds at
Compound
Water
Chemical
formula
Water solubility
(rng/L)
Density
(g/cm ')
Dielectric
constant
1.00
80.4
completely
miscible
0.79
21.5
C2HS OH
completely
miscible
0.79
25.0
Methanol
C H30 H
completely
miscible
0.79
33.6
Dioxane
C4 H S02
completely
miscible
1.03
2.2
2.3
Acetone
H2O
C3H60
Ethanol
Benzene
C6H6
1791
0.88
m-Xylene
CsHIO
146
0.88
2.3
Toluene
421
0.87
2.4
Cyclohexane
C7HIO
C6H12
55
0.78
2.0
Heptane
C7H16
0.68
1.9
Nitrobenzene
C6HsN2
1900
1.20
35.7
CCI4
805
1.59
2.2
Tetrachloroethylene
C2Cl4
150
1.62
2.2
Trichloroethylene
C2HCl3
llOO
1.46
3.3
Values taken from Howard (1990), Mitchell and Madsen (1987) and Abdul et al. (1990)
10.3 WATER-WET VS. DRY CLAYS, AND THE ISSUE OF WETTABILITY

It is important to differentiate between studies of organic liquid interactions involving
dehydrated clays, and those involving water-wet clays. The term "water-wet" is used
here to denote clay soils that have water as a pore liquid (as in nearly all field situations);
the soil pore space need not be 100% occupied by water. A hydrated clay mineral has a
different crystal structure and different behavior than does the corresponding "dry" clay.
This is particularly true for the swelling clays. Clays in their natural state are always
hydrated to some extent because polar water molecules are strongly attracted to the cations
adsorbed on the surfaces of clays; the complete dehydration of a clay mineral requires
temperatures in excess of 100C (Mitchell, 1976). Thus, although studies of dry clays in
contact with organic liquids provide fundamental information on relevant chemical and
physical interactions, this information is often not directly applicable to field situations.
In clay soils, polar water molecules are attracted to negatively-charged clay surfaces
more strongly than are non-polar organic molecules. Thus, in a water-wet clay, an organic
liquid phase will generally be the non-wetting liquid, and water will be the wetting fluid.
(Water wetting is also generally observed for other types of geologic media besides clays.)
However, in a dry clay, an organic liquid phase will be the wetting fluid and air will be
the non-wetting fluid.
318
Chap. 10
The ability of a low-water-solubility organic liquid to "wet," i.e., directly contact

a clay mineral is a fundamental control on the interactions between the clay and the
organic liquid. When an organic liquid phase penetrates a water-saturated (i.e., a waterwet) clay, interfacial tension (i.e., surface tension) will restrict the flow of the organic
phase to the largest pores in the clay. Although water will be displaced from the centers
of the pores, a film of water will remain to surround the clay, separating the clay from
the intruded organic liquid. This phenomenon was observed by Mossop (1980) in the
Athabasca tar sands where oil is the non-wetting liquid. In clays, under these types of
circumstances, there is a film of water between the organic liquid and the clay surfaces,
and the organic liquid is prevented from contacting or entering the clay double layers,
other than in dissolved form (Middleton, 1990).
10.4 RELEVANT PROPERTIES OF ORGANIC CHEMICALS
There are three broad categories of organic liquids: organic acids, organic bases, and
nonionic (i.e., uncharged) organic compounds (Anderson et al., 1985a). In this chapter,
we are concerned only with neutral compounds, the class which includes the chlorinated solvents. Neutral organic compounds have sometimes been further subdivided into
"water-miscible" and "water-immiscible" sub-classes. However, it is very important to
note that no organic liquid is totally immiscible with water: every organic liquid possesses
a finite solubility in water, and so if a small enough amount of any liquid is brought into
contact with water, all of that amount will dissolve and there would be no organic liquid phase. We can conclude that the adjective "water-immiscible" is oversimplified and
should be avoided; "low-water-solubility" is preferable, and will be used here. There is
no problem with the adjective "water-miscible," as we realize that it means a compound
that is completely soluble in water.
Low-water-solubility organic liquids, including the chlorinated solvents, are nonpolar and have low dielectric constants. Water-miscible organic liquids (such as low
molecular weight alcohols and acetone) are generally polar molecules and have moderate
to high dielectric constants. Exceptions to these classifications do occur. For example,
although dioxane is generally considered to be non-polar, as a relatively low molecular
weight cyclic ether, it is miscible with water over all proportions.
10.5 LABORATORY INVESTIGATIONS OF CLAY/ORGANIC
INTERACTIONS
A variety of approaches have been employed to study the effects of organic liquids
on clay permeability. X-ray diffraction studies and laboratory permeability tests have
been the most successful. Intracrystalline swelling of the smectitic clay minerals can
be measured directly by X-ray diffraction (XRD). The degree of crystal expansion is
commonly described by the basal "d-spacing" of the clay crystal. Various researchers
have used XRD techniques to measure the effects of organic liquids on the expansion
Sec. 10.5
319
of swelling clays. Griffin et al. (1984) conducted the only published study of the effect
of low-water-solubility organic chemicals on the intracrystalline swelling of hydrated
swelling clays. In that study, undiluted organic chemicals were added to aged, water-clay
slurries. Exposure of the hydrated smectite to carbon tetrachloride resulted in no change
in the d-spacing. When clay gels were exposed to acetone, which is totally miscible
with water, increased intracrystalline swelling was observed, while exposure to alcohols
resulted in a decrease in the d-spacing.
Barshad (1952) immersed dehydrated smectite and vermiculite clays in pure organic
liquids, then analyzed the clay-organic mixtures by XRD. Tests were performed with pure
organic liquids including low-water-solubility compounds (hydrocarbons and ethers) as
well as high solubility compounds (low molecular weight alcohols and ketones). Mixtures
of organic liquids were also tested. No tests were conducted with chlorinated solvents.
The results showed that the degree of crystal expansion is strongly influenced by both
the morphology of the organic molecule and the dielectric constant of the organic liquid.
Increasing the dielectric constant of a mixture of fluids by varying the composition of
the mixture resulted in an increase in the basal d-spacing. However, there was no simple
relationship between basal d-spacing and dielectric constant which held for all organic
fluids. Barshad's (1952) work with low-water-solubility organic compounds has limited
applicability to water-wet clays, where it is unlikely that a low-polarity organic molecule
will displace highly polar water molecules from clay interlayer spaces.
Barshad (1952), Brindley et al. (1969), Brown eta!. (1984), and Brown and Thomas
(1987) provide details of XRD studies of clays exposed to water-miscible organic fluids. XRD studies performed using various concentrations (0 to 100%) of water-miscible
organic liquids are more relevant in determining the behavior of water-wet clays. Tests
have shown that variations in the organic content of an aqueous solution produce variations in clay d-spacing, but no simple relationship between the dielectric constant of
the organic liquid and the degree of intracrystalline swelling has been observed. At low
concentrations of alcohols or acetone, the basal d-spacing of clays have been observed
to increase over that obtained with pure water. At some "critical" compound- and claydependent concentration, an abrupt decrease in basal spacing occurs, to a value that is
less than that obtained with pure water. Reported critical concentrations for alcohols and
acetone range from about 20% to 50% by volume (7% to about 50% by mole fraction).
There is no agreement between researchers on the exact values of the critical concentrations, or on a corresponding "critical dielectric constant" of the organic liquid/water
mixture. The concentration and the morphology of water-miscible organic chemicals are
apparently strong influences on intracrystalline swelling, and concentrated solutions of
some water-miscible organics can cause shrinkage of hydrated swelling clays.
10.5.2.1 General
Most of the data on the effect of organic liquids on the permeability of clays comes from
laboratory permeability tests, and much of the controversy about the effects of organic
liquids on clays derives from these studies. Most of the reported permeability tests have
320
Chap. 10
been conducted on compacted clay samples; few tests on natural undisturbed clay (i.e.,
non-compacted or non-remolded clay) have been performed. Laboratory permeability
tests conducted to measure the effects of exposure to organic chemicals on the bulk permeability of clay soils can indicate changes to bulk permeability such as those caused by
changes in clay fabric, but they cannot identify the mechanisms of the underlying chemical interactions. Interpreting the results of these tests is often complicated by influences
from factors such as the permeameter type, the imposed hydraulic gradient, the confining
stress placed on the specimen, and the nature of the test fluid (fully or only partly water
miscible). We attribute many of the apparent inconsistencies in the reported effects of
clay-organic chemical interactions to the influence of one or more of these factors.

determination of permeability
One of the major influences on the outcomes of laboratory permeability tests is the specific
design of the permeameter equipment. Daniel et al. (1985), Boynton and Daniel (1985),
Bowders et al. (1986), and Mitchell and Madsen (1987) have reviewed the various types
of permeameters appropriate for testing fine grained soils, and their effect on measured
permeability.
Laboratory permeameters are of two general types: rigid-wall cells (see Figure 10.1), and flexible-wall cells (see Figure 10.2). Rigid-wall permeameters have various
designs: materials can be tested inside the original compaction mold or sampling tube,
trimmed into a rigid-wall chamber, or tested inside a consolidation cell. Rigid-wall permeameters are less expensive and are simple to use. The principal weakness of these
devices is the potential for "side-wall leakage". Because the hydraulic gradient used
is commonly as high as 50 to 1000, test results are often suspected of having been influenced by side-wall leakage. Another limitation of the rigid-wall permeameter is the
inability to monitor or control stress on the sample. When clay/permeant reactions result
in shrinkage of the specimen, a rigid-wall device allows the formation of fractures or
macropores. The specimen may also pull away from the walls of the permeameter, causing a large increase in permeability. Under such conditions, the results of the tests cannot
be applied directly to field situations of interest because it is unlikely that the conditions
that permitted the fracture formation in the laboratory specimen would be representative
of the stress conditions existing at the field scale.
Some rigid-wall permeameters allow application of vertical confining stresses and
the measurement of sample volume changes in the vertical direction. Vertical stresses in
a permeameter simulate the overburden load at depth in a field situation, and to some
extent prevent fracture formation and side-wall leakage. Stiff or stony soil samples are
nevertheless susceptible to side-wall leakage even under vertical confining stress, though
these permeameters perform well on softer soils (Daniel et al., 1985).
Flexible-wall permeability tests can be conducted in a modified triaxial cell. The
specimen is contained within a flexible (usually latex) membrane. Pressurized fluid surrounding the membrane applies a confining stress on the soil, and holds the membrane
against the side of the sample, thereby preventing side-wall leakage while allowing sam-
Sec. 10.5
PRESSURE
321
VENT
Figure 10.1 Schematic diagram of a

typical rigid-wall permeameter.
pie deformation. Stress conditions on the sample are controlled and monitored. Disadvantages of this approach include the cost and complexity of the equipment, the potential
for incompatibility between the permeant fluids (e.g., a chlorinated solvent of interest)
and the flexible membrane, and the need for a lateral confining pressure larger than the
influent fluid pressure. The last feature may result in excessive effective stress at the
effluent end of the specimen.
In a flexible-wall permeameter, the applied confining stress should cause consolidation of the specimen when chemical interactions between soil and permeant induce
a reduction in interparticle repulsion forces. Fractures and macropores are therefore less
likely to form than in a rigid-wall cell. Instead, sample consolidation may occur, and permeability values will not increase to the same degree as in an unconfined rigid-wall test.
In fact, the measured permeability may decrease. We note then that a close monitoring
of specimen volume changes during organic fluid permeation may help identify changes
in the specimen. At the same time, sample consolidation and concomitant decreases in
permeability observed in flexible-wall permeameter tests may not be representative of
field situations, particularly when the effective confining stresses in the laboratory are
larger than the in situ field stresses.
322
ACRYLIC
CELL WALL ~
Chap. 10
O-RING
TOP CAP
LATEX
MEMBRANE
BASE CAP
POROUSSTONE
INfLUENT
LINE
Figure 10.2
EffLUENT LINE
Schematic diagram of a typical flexible-wall permeameter.

Natural groundwater gradients in clayey strata are generally less than I. Gradients across
compacted clay liners beneath landfills are unlikely to exceed 20 (Dunn and Mitchell,
1984). However, in almost all laboratory permeability tests, limited testing times and
small specimen sizes necessitate the use of large hydraulic gradients, much higher than
those found in the field, e.g., gradients of 50 to 1000.
In both rigid-wall and flexible-wall tests of compacted clay, with water as the
perrneant, increased gradients commonly result in slightly lower measured permeability
values (Foreman and Daniel, 1986; Dunn and Mitchell, 1984). High gradients can cause
particle migration within a specimen due to seepage forces (Dunn and Mitchell, 1984) or,
in the case of flexible-wall tests, specimen consolidation by increased maximum effective
stress. In permeability tests using water-wet clay and NAPLs, the relationship between
hydraulic gradient and measured permeability is more complex, particularly for tests
using rigid-wall permeameters. There is evidence that an increased NAPL gradient can
cause increases in measured permeability under some conditions (Quigley and Fernandez,
1989; Middleton, 1990). Consideration of the physical flow of a NAPL provides a possible
explanation. Forces of interfacial tension will prevent entry of a NAPL into the influent
end of a clay specimen until the entry pressure for the NAPL phase is exceeded. The
magnitude of the entry pressure will depend, in part, on the size of the voids (pores,
fissures, etc.) in the clay specimen. With the NAPL as the non-wetting fluid in the system,
the NAPL will enter the largest available voids in the clay, typically the largest pores and
fractures. If a sufficient entry pressure is achieved, "breakthrough" of the NAPL occurs
Sec. 10.5
323
soon thereafter, and the NAPL begins to exit from the effluent end of the clay specimen.
At this point, the NAPL is present in one or several small connected flow channels within
the specimen, but the relative saturation of the organic fluid expressed on a bulk sample
basis is typically very small. In almost all reported tests, a large value of NAPL hydraulic
gradient! was necessary to achieve breakthrough of the immiscible-phase fluid through
clay specimens. The evidence suggests that large values of NAPL hydraulic gradient
create physical forces which open flow channels in clay under low confining stress (as
with rigid-wall permeameters). Perhaps the physical mechanism relates to the difference
in fluid pressure between the clay pore water and the intruding NAPL fluid, causing
a change in the effective stress on the NAPL flow channels within the specimens (see
below).
10.5.2.4 Permeability tests on "dry" clays/organic mixtures

Permeability tests have been conducted with dry clay soils compacted or mixed with
various organic fluids. For these dry soil tests, the issue of water miscibility is not relevant. Therefore, both polar and non-polar organic chemicals are discussed together here.
Permeability tests on dry clay/organic fluid mixtures are expected to show the extreme
effects of organic fluids on clay fabric. Michaels and Lin (1954) and Mesri and Olson
(1971) performed rigid-wall permeability tests and consolidation tests, respectively, using dry clays mixed with pure organic fluids including alcohols, benzene, and carbon
tetrachloride. They found that the hydraulic conductivity decreased as the polarity (and
dielectric constant) of the test fluid increased. Using rigid-wall permeability tests, Fernandez (1984, 1989), Fernandez and Quigley (1985), and Schramm et al. (1986) obtained
similar results. These results are consistent with double layer theory: fluids of low dielectric constant cause double layer contraction, causing a flocculated soil fabric with
large inter-aggregate pore spaces, and greater permeability.
10.5.2.5 Permeability tests of water-wet clays and NAPLs

In nearly all field situations, soils are at least partly saturated with water before they
become contaminated with organic chemicals. Thus, permeability tests with water-wet
materials are more relevant to actual field sites than are tests with soils that are totally
water-dry. In many investigations of the permeability of water-wet clays to organic fluids,
low-water-solubility organic fluids were not considered to be distinct from organic fluids
that are totally water miscible. Since wettability has a significant effect on clay-organic
fluid interactions, laboratory studies using organic liquids that are totally water miscible
should not be expected to provide a reliable indication of the permeability behavior of
organic liquids that are not.
Most of the published permeability results using low-water-solubility organic liquids have been obtained using LNAPLs (Table 10.1) and compacted clay specimens.
tThe NAPL hydraulic gradient on the clay specimen is directly related to the NAPL "head drop" across
the clay specimen, which is equal to the difference between the fluid pressures of NAPL at the influent and
effluent ends of the specimen.
324
Chap. 10
Tests using LNAPLs are relevant to our study of chlorinated solvents because many of
the LNAPLs and the chlorinated solvents have similar dielectric constants and solubilities, and so are expected to interact with the double layers of clay minerals in similar
manners. Early research sponsored in part by the U.S. EPA showed large increases of
the hydraulic conductivities of water-wet compacted clays when permeated with lowwater-solubility organic liquids in rigid-wall permeameters (Brown et aI., 1984; Brown
et al., 1986; Anderson et al., 1985a). Hydraulic conductivity increases of up to three
orders of magnitude over the values obtained with water were measured during passage
of only ~ 0.25 pore volumes of flow of the organic liquids through the columns. These
results were obtained for tests performed using both swelling and non-swelling clays,
with xylene and heptane, and using gradients of 31 to 361 (Figure 10.3). Brown et al.
(1986) also obtained increases in hydraulic conductivity in similar tests using various
low-water-solubility commercial petroleum products. As confirmation of the observed
permeability changes caused by xylene and heptane, Anderson et al. (l985a) reported
that the macroscopic structure of clay specimens changed from a massive structure to a
blocky structure.
In rigid-wall tests of non-swelling compacted clays using heptane and TeE, Daniel
and Broderick (1985) obtained results that were similar to those discussed above. In
clays compacted with Standard Proctor effort, the hydraulic conductivity values increased
before even 0.2 pore volumes of the test fluids had passed through the permeameter.
MICA
XYLENE
NONSATURATED
GRADIENT 18 1
REP 1
o REP 2
6::::>
o
z
ou
-8
10
LAB VALUE VITH WATER
1.5 x 10- 8
-9
10
-f----r-----r-----,
PORE VOLUME
Figure 10.3 The hydraulic conductivity of

water-wet, compacted illitic clay permeated
with concentrated xylene, as measured in a
rigid-wall permeameter. (Reprinted with
permission from Brown et al., 1986.)
Sec. 10.5
325
The imposed gradient ranged from 100 to 300. However, an increase in the degree of
compaction of the clay prior to permeation resulted in no breakthrough of the organic
fluids: the entry pressure of the organic fluids could not then be achieved with their
testing apparatus.
Brown et al. (1986) tested the effect of varying the hydraulic gradient in rigid-wall
tests. Tests were conducted using water, a fully water-miscible liquid (acetone), and a
low-water-solubility liquid (xylene) as permeants through water-wet clay. They concluded
that no relationship could be found between the gradient and hydraulic conductivity.
Measured hydraulic conductivity values for each organic fluid varied over two orders of
magnitude with no apparent pattern. Each measurement was performed on a different
sample of compacted clay. However, the possibility that variability between samples
obscured a gradient vs. hydraulic conductivity relationship was not examined.
Green et al. (1981) also used rigid-wall permeameters and smectite and kaolinite
clays in tests with benzene, m-xylene, TCE, and carbon tetrachloride. In the majority
of their tests, the measured hydraulic conductivity values obtained with the organic
fluids were less than that obtained with water. However, they reported that although
decreases in hydraulic conductivity were common, some tests with carbon tetrachloride,
benzene and xylene exhibited a "breakthrough phenomenon" which consisted of a rapidlyoccurring, large increase in permeability. The hydraulic gradients used in these tests
were not reported. Green et al. concluded that the low dielectric constant, low-watersolubility liquids which they studied have the potential to cause bulk sample shrinkage
and consequently increases in hydraulic conductivity in clays.
Anderson et al. (1985b) used a double-ring (rigid-wall) permeameter to test a soilbentonite-water slurry with pure xylene under a hydraulic gradient of 36. Large increases
in measured hydraulic conductivity were recorded immediately after the introduction of
the xylene. Increases occurred in both the side wall area and in the central portion of the
specimens, indicating that the increases were not caused solely by side-wall leakage.
Abdul et al. (1990) used rigid-wall permeameters to permeate vacuum-compacted
clay slurries with nine low-water-solubility organic liquids, including TCE, PCE, benzene, and xylene. A smectitic clay was tested with all nine organic liquids, and a kaolin
clay was tested with all liquids except the two chlorinated solvents. Tests were performed with relatively low permeant fluid heads of about I meter, on samples about 5
to 6 centimeters in length. In the smectitic clay, both chlorinated solvents caused almost
immediate large hydraulic conductivity increases and visible cracking. All of the other
concentrated organic liquids caused visible cracking and large hydraulic conductivity
increases in both the smectitic clay and the kaolin. Abdul et al. did not correlate the
observed results with any specific properties of the liquids, nor discuss mechanisms of
clay/organic liquid interactions. However, all of the organic liquids used in the study
have dielectric constants much lower than that of water. These are the only reported tests
that have been conducted using relatively low gradients in which low-water-solubility
liquids penetrated and caused permeability increases in clays. In contrast to other studies
which used physically compacted clays, the clay specimens in this study were prepared
as a slurry and vacuum compacted. This mode of compaction may have had an influence
on the test results.
326
Chap. 10
Evans et al. (1985) obtained similar results in testing a vacuum-compacted

bentonite-sand-water slurry. The test medium was permeated with carbon tetrachloride
in a flexible-wall permeameter, using a hydraulic gradient of approximately 100 and an
average effective stress of 207 kPa. The hydraulic conductivity of the test medium increased by two orders of magnitude approximately three days after introduction of the
solvent. Data on sample consolidation during the test were not presented.
Comparing the results from different types of permeameters, Bowders (1985)
and Foreman and Daniel (1986) found that water-wet clays permeated with low-watersolubility fluids (heptane and TCE) showed significant hydraulic conductivity increases in
rigid-wall tests. In contrast, no entry of the organics was achieved in flexible-wall cells.
They attributed these results to bulk soil shrinkage and the opening of flow channels
which could occur in the rigid-wall cell but not in the flexible-wall cell. Foreman and
Daniel (1986) performed additional tests and determined that some, but not all, of the hydraulic conductivity increase in rigid-wall tests could be attributed to side wall flow. They
also determined that a finite "gradient" (i.e., entry pressure) existed below which the lowwater-solubility organic fluids would not penetrate the clay specimens in rigid-wall cells.
Other studies using flexible-wall cells to test compacted, water-wet clays with low-watersolubility permeants have also either obtained large hydraulic conductivity decreases, or
have been unable to achieve a sufficient entry pressure to allow penetration of the organic fluid through the clay (Acar et al., 1985; Uppot and Stephenson, 1989). Bowders
(1985) also reported one case where large hydraulic conductivity values were measured
in a flexible-wall cell when the permeant was able to "short circuit" the specimen by
leaking along the clay/Teflon interface associated with the Teflon film placed between the
soil sample and the latex membrane to prevent degradation of the membrane by organic
fluids; the Teflon film may have formed a semi-rigid cylinder and allowed side-wall flow.
Fernandez (1984; 1989) and Quigley and Fernandez (1989) tested compacted clays
in a rigid-wall permeameter and measured hydraulic conductivity values for several lowwater-solubility organic liquids (cyclohexane, benzene, and xylene). The results were
similar to or less than the value obtained with water. Channel flow and early breakthrough of the pure organic liquid were observed in some tests. They concluded that
unflawed, water-wet clays do not exhibit increased hydraulic conductivity when permeated by such liquids: interfacial tension effects create high breakthrough pressures and
flow can only occur by displacement of water from pre-existing macropores. However,
the introduction of a totally-water-miscible liquid such as acetone, or the addition of
a surfactant to the water/clay/organic fluid system, each resulted in decreased breakthrough pressures and increases in measured hydraulic conductivity. The same authors
reported flow-rate dependent hydraulic conductivity measurements during permeation of
compacted clay with cyclohexane (a low-water-solubility liquid) in a rigid-wall, constant
flow permeameter. The measured permeability increased with increasing gradient (Fernandez, 1989; Quigley and Fernandez, 1989). They concluded that the cyclohexane was
forced along microfissures or channels which expanded or contracted in proportion to
the gradient of the immiscible fluid.
In tests similar to those discussed above, Middleton (1990) found that permeation of
compacted illitic clay with PCE in rigid-wall permeameters did not result in chemically-
Sec. 10.5
327
induced permeability increases. The flow of the solvent was restricted to microfractures
and channels in the clay. Tests of unconfined compacted clay showed that permeability
values varied with the flow rate of the solvent through the sample. Tests were conducted
by establishing a constant flow of NAPL through the clay specimens. The measured
permeability increased as the flow rate increased, and decreased again as the flow rate
decreased. This behavior was attributed to physical deformation of the clay by large
solvent fluid pressures, whereby forces created by two-phase flow within the deformable
clay acted to increase the volume of flow channels as the pressure of the non-wetting
fluid increased. This is an explanation for why gradient-dependent permeability values
are observed only when a low-water-solubility liquid penetrates a water-wet clay, and
not in the case of the permeation of water itself.
Middleton et al. (1992) reported tests on cores of naturally fractured, illitic clay
that were permeated with PCE. The tests were performed in a flexible-wall permeameter
under low effective stresses and low gradients. Under constant effective stresses, the
permeability to the chlorinated solvent was less than or equal to the permeability to water,
and no significant volume changes of the samples were measured during permeation with
the solvent (Figure lOA). These are the only reported tests on an undisturbed clay with
natural fractures.
The results of laboratory permeability tests of clays with NAPLs can be summarized
as follows. In tests where a sufficient entry pressure was achieved to allow penetration
of a NAPL into the clay, all flexible-wall tests and some unconfined rigid-wall tests
have shown either no increase, or a decrease in measured permeability relative to that
1E-09 . , - - - - - - - - - , - - - - - - - - - - - - - - - - - - ,
1E-10
~
E
1E-11
15
co
Ql
E
1E-12
-31~ :'41-:-52-
- - 5 2 - - - :-31-:'41':
-52,
,
,
.
Qj
o,
* Head drop, in em of water
1E-13
..... .. .. ... Water

50
100
150
PERC ..........................................................................................
200
250
300
350
400
450
500
Flow Volume (mL)

Figure 10.4 Permeability vs. cumulative flow for sequential permeation of water and
tetrachloroethylene (PER) in illitic clay under a low gradient (~ 10) and low effective
stress (4.1 kPa). (Reprinted with permission from Middleton et al., 1992.)
328
Chap. 10
of water. Early breakthrough of the pure organic fluids occurred in all cases. The pure
organic fluid, because it was restricted by interfacial tension to macro-pore channels in
the specimen, occupied only a small portion of the total pore volume of the specimen.
In some unconfined rigid-wall tests, penetration of NAPL into the specimens resulted
in rapid breakthrough of the pure organics and large (up to three orders of magnitude
or more) increases in measured permeability. Some but not all of the NAPL flow was
due to side wall flow. These large permeability increases may be due to the formation
of fractures caused by physical (not chemical) phenomena that are attributable to the
high NAPL fluid pressures used in the tests. The exceptions to the above are the results
of Abdul et al. (1985) when chlorinated solvents caused large permeability increases in
swelling clays in both rigid-wall and flexible-wall tests. The mode of clay compaction
may have had an important influence on these results.
10.5.2.6 Permeability tests using aqueous solutions of NAPL
chemicals
Research on the permeation of water-wet clays with aqueous solutions of low-watersolubility organic chemicals (usually at their solubility limit) have been in good agreement. No changes in hydraulic conductivity were observed in any flexible-wall or rigidwall tests (Daniel et al., 1988; Bowders, 1985; Acar et al., 1985; Evans et al., 1985;
Abdul et al., 1990). This is consistent with double layer theory inasmuch as dilute solutions of organic chemicals do not have dielectric constants significantly lower than that
of water.
10.5.2.7 Permeability tests of clays and water-miscible organic
fluids
There is general agreement among researchers that concentrated, water-miscible, neutral
organic chemicals such as low molecular weight alcohols and acetone can increase the
permeability of water-wet clays. This applies to both swelling and non-swelling clays.
Increases in permeability have been explained in terms of the ability of these liquids to
fully mix with water, permeate the clay, and decrease the dielectric constant of the pore
fluid, thereby causing changes to the clay fabric and pore sizes as predicted by double
layer theory. Fernandez and Quigley (1985), Acar et at. (1985), and Foreman and Daniel
(1986) have provided some of the more recent work with water-miscible organic liquids.
The application of sufficient effective stress to clay specimens permeated with
water-miscible organic fluids generally results in chemically-induced consolidation of the
clay, and either no increase, or even a decrease in the measured permeability. Figure 10.5,
taken from Acar et al. (1985), illustrates that the final measured hydraulic conductivity
in tests with a water-miscible organic fluid decreases as the average effective confining
stress on the specimen is increased. This indicates that chemically-induced permeability
increases can be prevented by the application of sufficient effective stress on the clay.
Therefore, if chemically-induced desiccation of a clayey stratum or liner occurs at depth
where the overburden load is sufficient, whether the overburden is waste material in a
landfill or natural deposits, the permeability would not increase.
Sec. 10.5
329
,.......,
o(J.)
sao
VI
<,
RIGID WALL
PERMEAMETER
E
o
OJ
I
x
'-'
100
so
FLEXIBLE WALL
PERMEAMETER
>I5
6
~
--0---
10
u
u
@cy----o
, ,Cf
,D
~
o
>I
_..- --
o1
PORE VOLUME
Figure 10.5 The effect of effective stress and permeameter type on the measured hydraulic conductivity of water-wet compacted kaolinite permeated with concentrated acetone. (Adapted with permission from Acar et al., 1985.)
Water-miscible organic fluids can contact clay soils at concentrations in water that
range from 0% to 100%. Because water has a dielectric constant that is large relative
to organic compounds, the dielectric constant of an aqueous solution will decrease as
the fractional organic content increases (Brown and Thomas, 1987). As illustrated in
Figure 10.6, recent work indicates that the concentrations of the water-miscible organics must be above about 50-80% in order for laboratory-measured permeability values
to increase over the permeability to water (Fernandez and Quigley, 1988; Brown et
al., 1984; Bowders, 1985). In terms of double-layer theory, contraction of clay double
layers may not occur until a "critical" dielectric constant of the pore fluid is reached.
The critical dielectric constant may vary depending on the clay minerals and soil fabric.
For swelling clays, water-miscible organics could also cause changes in intracrystalline
swelling, which may also affect the critical concentration of the organic necessary to
cause clay permeability increases.
As with low-water-solubility compounds, tests using very dilute (less than 5%)
solutions of water-miscible organic compounds show no increases in hydraulic conductivity or permeability (Acar et al., 1985; Daniel et al., 1988; Bowders, 1985). This is as
expected since the dielectric constant of water is not changed much by the addition of
small amounts of water-miscible organic chemicals.
330
Chap. 10
10- 6
.........
OJ
KAOLINITE
CIl
<,
E
u
........
>f-
i=
U
10- 7
::::>
0
u
u
::J
FLEXIBLE-WALL CELL
::::>
~
0
>I
10- 8
0
20
40
60
PERCENT METHANOL
80
100
Figure 10.6 The measured hydraulic

conductivity of water-wet, compacted
kaolinite permeated with aqueous solutions
of methanol. (Reprinted with permission
from Bowders, 1985.)
10.5.2.8 The influence of clay mineral type on permeability to

organic fluids
Little information is available on the influence of clay mineral type on the permeability of
soils to organic fluids. Experimental results suggest that smectite clays, as swelling clays,
may undergo larger increases in permeability than illite or kaolinite clays on exposure to
water-miscible organic liquids (Anderson et al., 1982; Foreman and Daniel, 1986; Acar
and D'Hollosy, 1987). This may be caused by the chemical structure of the clay, or by
physical properties such as plasticity.
10.6 FIELD STUDIES

Despite a broad interest in chemical interactions between low-water-solubility organic
chemicals and clays, there have been few reported field investigations of this topic,
probably because of the difficulty in directly observing the effects of organic chemicals on
clays. At most sites, there is only limited potential for direct examination of clays exposed
to organic chemicals. This is especially true for sites contaminated with chlorinated
solvents due to the required depth of excavation and the risk of causing further movement
of the solvents while obtaining soil samples.
Field experiments of DNAPLIclay interactions are being conducted by the University of Waterloo at a site near Sarnia, in southwestern Ontario. The site is located
Sec. 10.7
Conclusions
331
in an extensive illitic clay deposit of glacial origin. Free-phase TCE was placed at the
bottom of boreholes below the water table, in both shallow weathered clay and in deeper
unweathered clay. Soil core samples which had been in direct contact with the solvent
in shallow boreholes for two years currently show no visible cracking or fabric changes
in the clay. TCE in contact with the unweathered clay in one borehole for II months
and in another nearby borehole for 6.5 years moved downward in the clay by molecular
diffusion, but showed no evidence of transport through fractures or by advection in the
porous medium (Poulsen et al., 1993). There was no evidence to suggest that exposure
to the solvent had caused increased permeability of the clay in any of the boreholes. In
the laboratory, samples of the same illitic clay immersed in TCE or PCE and stored for
one year show no visible fractures or changes to the clay fabric, even under examination using a scanning electron microscope. Also, after one year, the solvent showed no
wetting of the clay surfaces (Middleton, 1990).
At a site in Smithville, Ontario, dense oils containing chlorinated solvents and
chiorobenzenes were found to have moved through a natural clay deposit and entered
an underlying fractured limestone aquifer. Initially, it was suspected that chemical interactions between the oils and the clay had facilitated their movement through the clay.
However, it was later concluded that the oils had moved downward through pre-existing
fractures in the clay deposit. At a site at Wilsonville, Illinois, organic chemicals including chlorinated solvents, pesticides, and aromatic compounds have migrated through a
clayey glacial till at rates 100 to 1000 times greater than had been predicted (Griffin
et al., 1984; Herzog et al., 1989). However, the primary reason for the rapid migration
was discovered to be the presence of pre-existing fractures in the till. Nevertheless, the
investigators concluded that organic-clay interactions may have caused permeabilities
larger than those attributable to the natural fractures.
10.7 CONCLUSIONS
In laboratory experiments on water-wet clays, water-miscible organic chemicals have a

different effect on clay permeability than low-water-solubility organic chemicals. Watermiscible organics above a critical concentration can increase the permeability of clay
soils, apparently due to the ability of these fluids to fully penetrate into the clay and shrink
clay double layers by reduction of the solution dielectric constant. Critical concentrations
for low molecular weight alcohols and acetone are on the order of 50% to 80% organic
content by volume in water. In the case of swelling clays, intracrystalline swelling or
shrinkage may also affect permeability.
Permeation of clays by aqueous solutions of chlorinated solvents and other lowwater-solubility liquids at concentrations at or below their solubility limits (usually less
than 2% by volume) does not influence the permeability of clays. This agrees with double
layer theory because low concentrations of these organic chemicals do not decrease the
dielectric constant of the aqueous solution to any significant extent.
Low-water-solubility organic liquids, including the chlorinated solvents, apparently do not chemically increase the permeability of water-wet clay soils through double layer effects or changes to intracrystalline swelling. These liquids require a certain
332
Chap. 10
breakthrough pressure to overcome interfacial tension effects and penetrate into water-wet
clays. When penetration occurs, flow of the immiscible liquid is restricted to (pre-existing)
macro-pores or fractures in the clay. In laboratory permeability tests under very low confining stresses, mechanical fracturing of clay soils under large organic fluid gradients can
result in increased permeability. Although the chemical or physical mechanisms causing the permeability increases may not be well understood, we do know that the mode
of compaction and therefore the strength of the clay specimens may be an important
influence on such test results.
Based on the results of the laboratory studies, conclusions can be drawn regarding
field situations. Exposure to chlorinated solvents and other low-water-solubility liquids is
unlikely to cause shrinking, cracking, or permeability increases in natural clay strata or
compacted clay barriers. This is because interfacial tension prevents these liquids from
entering the clay porous medium, and because the solubilities of these chemicals are
too low to affect the dielectric constant of the fluid. In situations where the solubility
of chlorinated solvent in groundwater is increased by the presence of co-solvents or
surfactants, chemically-induced shrinkage of clayey materials may occur. However, if
this takes place at depth below ground surface, confining stresses due to the overburden
load would likely prevent an increase in the permeability. Care should therefore be
taken when using surfactants or alcohols for in-situ remediation to ensure that enhanced
solubility effects do not cause permeability increases.
Particular concern for permeability increases caused by organic chemicals pertains
to borehole seals and engineered barriers such as landfill liners and cutoff walls. Borehole
seals and engineered barriers commonly contain chemically-sensitive clays such as srnectites (e.g., bentonite) that may shrink considerably when exposed to high concentrations
of water-miscible organic chemicals under low confining stress conditions. However, we
discovered no documented cases where such permeability increases have occurred at field
sites.
The conclusions presented here differ substantially from those of Barbee (1994) in
his recent review of the environmental fate of chlorinated hydrocarbons. He concludes
that chlorinated hydrocarbon liquids are able to rapidly penetrate low permeability clays
and significantly increase clay hydraulic conductivity. However, Barbee (1994) based his
conclusions on an incomplete review of laboratory studies. Further, he did not distinguish
between studies using low-water-solubility and water-miscible organic liquids: one of his
primary references was the laboratory study using acetone and ethanol that was conducted
by Brown and Thomas (1987).
This discussion has shown that by grouping the results of laboratory research
according to solubility of the organic fluid, "dry" or "water-wet" clays, type of perrneameter, and other factors, many apparent contradictions in the literature can be resolved.
Nevertheless, further research is needed. The influence of specific clay mineral types on
clay/organic interactions should be determined with a focus on the chemical and physical
interactions between swelling clays and low water solubility organic liquids. The influence of the mode of compaction of clay specimens and the effective stress conditions
on permeability test results should be determined in order to more confidently apply
laboratory test results to field problems.
Sec. 10.8
References
333
ACKNOWLEDGEMENTS
10.8 REFERENCES
Abdul, A. S., T. L. Gibson, and D. N. Rai (1990) "Laboratory studies of the flow of some organic
solvents and their aqueous solutions through bentonite and kaolin clays", Ground Water, 28,
524-533.
Acar, Y. B. and E. D'Hollosy (1987) "Assessment of pore fluid effects using flexible wall and
consolidation permeameters". In: Geotechnical Practice for Waste Disposal, 1987, R. D. Woods
(Ed.), Geotechnical Special Publication No. 13. American Society of Civil Engineers, 231-245.
Acar, Y. B., A. Hamidon, S. P. Field, and L. Scott (1985) 'The effect of organic fluids on
the hydraulic conductivity of compacted kaolinite". In: Hydraulic Barriers in Soil and Rock,
American Society for Testing and Materials, Special Technical Publication 874, A. 1. Johnson, R.
K. Frobel, N.J. Cavalli, and CB. Pettersson (Eds.), American Society for Testing and Materials,
Philadelphia, 171-187.
Acar, Y. B. and R. K. Seals (1984) "Clay barrier technology for shallow land waste disposal
facilities", Hazardous Waste, 1, 167-181.
Anderson, D. C, K. W. Brown, and J. W. Green (1982) "Effect of organic fluids on the permeability
of clay soil liners". In: Land Disposal of Hazardous Waste, Proceedings of the Eighth Annual
Research Symposium, Ft. Mitchell, Kentucky, EPA-600/9-82-002, 179-190.
Anderson, D. C., K. W. Brown, and J. C Thomas (l985a) "Conductivity of compacted clay soils
to water and organic liquids", Waste Management and Research, 3, 339-349.
Anderson, D. C., W. Crawley, and J. D. Zabcik (1985b) "Effects of various liquids on clay
soil: bentonite slurry mixtures". In: Hydraulic Barriers in Soil and Rock, American Society
for Testing and Materials, Special Technical Publication 874, A. Johnson, R. K. Frobel, N. J.
Cavalli, and C B. Pettersson (Eds.), American Society for Testing and Materials, Philadelphia,
93-103.
Barbee, G. C (1994) "Fate of chlorinated aliphatic hydrocarbons in the vadose zone and ground
water", Ground Water Monitoring and Remediation, 9, No.1 (Winter), 129-140.
Barshad, 1. (1952) "Factors affecting the interlayer expansion of vermiculite and montmorillonite
with organic substances", Soil Sci. Soc. Amer. Proc., 16, 176-182.
Bowders, J. J. (1985) The Influence of Various Concentrations of Organic Liquids on the Hydraulic
Conductivity of Compacted Clay, Geotechnical Engineering Dissertation GT85-2, The University of Texas at Austin, Austin, Texas.
Bowders, J. J., D. E. Daniel, G. P. Broderick, and H. M. Liljestrand (1986) "Methods for testing the compatibility of clay liners with landfill leachate". In: Proceedings: Hazardous and
Industrial Solid Waste Testing: Fourth Symposium, American Society for Testing and Materials,
Philadelphia, Special Technical Publication 886, 233-250.
334
Chap. 10
Boynton, S. S. and D. E. Daniel (1985) "Hydraulic conductivity testing of compacted clay", J.

Geotech. Eng., 3, 465-478.
Brindley, G. W., K. Wiewiora, and A. Wiewiora (1969) "Intracrystalline swelling of montmorillonite in some water-organic mixtures. Clay organic studies XVIII", Amer. Mineralogist, 54,
1635-1644.
Brown, K. W. and J. C. Thomas (1987) "A mechanism by which organic liquids increase the
hydraulic conductivity of compacted clay materials", Soil Sci. Soc. Amer. J., 51, 1451-1459.
Brown, K. W., J. C. Thomas, and 1. W. Green (1984) "Permeability of compacted soils to solvents,
mixtures, and petroleum products". In: Land Disposal of Hazardous Waste, Proceedings of the
Tenth Annual Research Symposium, Ft. Mitchell, KY. EPA-600/9-007, 124-137.
Brown, K. W., J. C. Thomas, 1. W. Green (1986) "Field cell verification of the effects of concentrated organic solvents on the conductivity of compacted soils", Hazardous Waste and Hazardous
Materials, 3, 1-19.
Brownlow, A. H. (\979) Geochemistry, Prentice-Hall, Englewood Cliffs, N.1.
Daniel, D. E. and G. P. Broderick (1985) "Stabilization of compacted clay liners against attack by
concentrated organic chemicals", Unpublished report to Chemical Manufacturers Association,
Washington, D.C.
Daniel, D. E., D. C. Anderson, and S. S. Boynton (1985) "Fixed-wall versus flexible-wall permeameters", In: Hydraulic Barriers in Soil and Rock, A. Johnson, R. K. Frobel, N. J. Cavalli and
C. B. Pettersson (Eds.), Special Technical Publication 874, American Society for Testing and
Materials, Philadelphia, pp. 107-126.
Daniel, D. E., H. M. Liljestrand, G. P. Broderick, and J. J. Bowders (1988) "Interaction of earthen
liner materials with industrial waste leachate", Hazardous Waste and Hazardous Materials, 5,
93-108.
Dunn, R. 1. and J. K. Mitchell (\ 984) "Fluid conductivity testing of fine-grained soils", 1. Geotech.
Eng., American Society of Civil Engineers, 110, 1648-1665.
Evans, J. C; H.-Y. Fang, and J. J. Kugelman (\985) "Organic fluid effects on the permeability
of soil-bentonite slurry walls". In: Proceedings: National Conference on Hazardous Wastes and
Environmental Emergencies, May 14-16, Cincinnati, OH, 267-271.
Fernandez, F. (1984) The Effect of Organic Hydrocarbon Liquids on the Hydraulic Conductivity of
Natural Soils, M.E.Sc. Thesis, University of Western Ontario, London, Ontario.
Fernandez, F. (1989) The Effects of Waste Leachates on the Hydraulic Conductivity of Natural
Clays, Ph.D. Thesis, University of Western Ontario, London, Ontario.
Fernandez, F. and R. M. Quigley (1985) "Hydraulic conductivity of natural clays permeated with
simple liquid hydrocarbons", Canadian Geotech. J., 22, 205-214.
Fernandez, F. and R. M. Quigley (1988) "Viscosity and dielectric constant controls on the hydraulic
conductivity of clayey soils permeated with water-soluble organics", Canadian Geotech. J., 25,
582-589.
Foreman, D. E. and D. E. Daniel (1986) "Permeation of compacted clay with organic chemicals",
J. Geotech. Eng., American Society of Civil Engineers, 112, 669-681.
Freeze, R. A. and J. A. Cherry (1979) Groundwater, Prentice-Hall, Englewood Cliffs, NJ.
Green, W. 1., G. F. Lee, and R. A. Jones (1981), "Clay-soils permeability and hazardous waste
storage", 1. Water Pollut. Control Fed., 53, 1347-1354.
Sec. 10.8
References
335
Griffin, R. A., R. E. Hughes, L. R. Follmer, C. J. Stohr, W. J. Morse, T. M. Johnson, J. K.

Bartz, J. D. Steele, K. Cartwright, M. M. Killey, and P. B. Dumontelle (1984) "Migration
of industrial chemicals and soil-waste interactions at Wilsonville, Illinois". In: Land Disposal
of Hazardous Waste. Proceedings of the Tenth Annual Research Symposium, Ft. Mitchell, KY.
EPA-600/9-84-007, 61-77.
Herzog, B. L., R. A. Griffin, C. J. Stohr, L. R. Follmer, W. J. Morse, and W. J. Su (1989) "Investigation of failure mechanisms and migration of organic chemicals at Wilsonville, Illinois",
Ground Water Monitoring Review, 9, No.2 (Spring), 82-89.
Howard, P. H. (1990) Handbook of Environmental Fate and Exposure Datafor Organic Chemicals,
Volume II, Solvents, Lewis Publishers, Boca Raton, FL.
Mesri, G. and R. E. Olson (1971) "Mechanisms controlling the permeability of clays", Clays and
Clay Minerals, 19,151-158.
Michaels, A. S. and C. S. Lin (1954) "Permeability of kaolinite", Ind. Eng. Chem., 46, 1239-1246.
Middleton, T. A. (1990) The Effect of Chlorinated Solvents on the Permeability of Fractured Clay.
M.Sc. Thesis, University of Waterloo, Waterloo, Ontario.
Middleton, T. A., J. A. Cherry, and R. N. Quigley (1992) "The effect of tetrachloroethylene on the
permeability of a fractured clay under constant stress conditions". In: Proceedings [published
1992] of the Conference on Subsurface Contamination By Immiscible Fluids, K. U. Weyer (Ed.),
International Association of Hydrogeologists, Calgary, Alberta, April 18-20, 1990. Order from
A. A. Balkema, Old Post Road, Brookfield, Vermont, 05036.
Mitchell, J. K. (1976) Fundamentals of Soil Behavior, John Wiley and Sons, New York, NY.
Mitchell, J. K. and F. T. Madsen (1987) "Chemical effects on clay hydraulic conductivity". In:
Proceedings of the Specialty Conference on Geotechnical Practice for Waste Disposal, 1987, R.
D. Woods (Ed.) American Society of Civil Engineers, Geotechnical Special Publication, No.
13,87-116.
Mossop, G. D. (1980) "Geology of the Athabasca oil sands", Science, 207, 145-152.
Poulsen, M. M., R. L. Johnson, B. L. Parker, and 1. A. Cherry (1993) "Diffusion of trichloroethylene
in a natural clay: A field experiment". In: Contaminated Soil '93, F. Arendt, G. J. Annokkee, R.
Bosman, and W. J. van den Brink (Eds.), Kluwer Academic Publishers, Netherlands, 547-548.
Quigley, R. M., and F. Fernandez (1989) "Clay/organic interactions and their effect on the hydraulic
conductivity of barrier clays". In: Contaminant Transport in Groundwater, H. E. Kobus and W.
Kinzelbach (Eds.), Balkema, Rotterdam, 117-124.
Schramm, M., A. W. Warrick, and W. H. Fuller (1986) "Permeability of soils to four organic
liquids and water", Hazardous Waste and Hazardous Materials, 3, 21-27.
Theng, B. K. G. (1974) The Chemistry of Clay-Organic Reactions, Adam Hilger, Bristol, England.
Uppot, J. O. and R. W. Stephenson (1989) "Permeability of clays under organic permeants", J.
Geotech. Eng., 115, 115-131.
Van Olphen, H. (1977) An Introduction to Clay Colloid Chemistry, John Wiley and Sons, New
York.
11
Physics Governing the
Migration of Dense
(DNAPLs) in Fractured Media
Bernard H. Kueper' and David B. McWhorter2
1 Department of Civil Engineering
Queen's University, Kingston, Ontario, Canada K7L 3N6
2
Department of Agricultural and Chemical Engineering

Colorado State University, Fort Collins, CO, USA
Page
11.1
11.2
11.3
11.4
11.5
11.6
11.7
11.8
Introduction
Rates of DNAPL Migration Through Fractures
DNAPL Migration Through Fracture Networks
Conclusions
References
337
338
343
347
349
351
352
352
ABSTRACT
DNAPL will enter a water-saturated fracture if the capillary pressure at the leading edge
of the DNAPL body exceeds the fracture entry pressure, as determined by the properties
of the fluid and of the medium. Once in a fracture network, DNAPL migration can be
rapid, and will be governed by geological structure. Pooled DNAPL can be mobilized
by groundwater pumping, which can be desirable in a waterflooding operation, but undesirable if mobilization occurs towards previously uncontaminated areas. Unlike LNAPL,
the height of DNAPL accumulated in a borehole gives little indication of conditions in
the formation.
11.1 INTRODUCTION
At many sites throughout North America and other parts of the world where a dense nonaqueous phase liquid (DNAPL) is known to be present in the subsurface, a fractured
aquitard or bedrock underlies a relatively shallow overburden. Because the bulk retention
capacity of most types of porous media is relatively low, even small volumes of DNAPL
337
338
Physics of DNAPL Migration in Fractured Media
Chap. 11
POOI_~~.........;;=-~~~
;
~
~
DNAPL pool in
fractures
DNAPLresidual in
fractures
Figure 11.1 DNAPL accumulation above fractured bedrock.
will in general have the potential to migrate down to the base of these overburden
deposits. DNAPL will then be exposed to fractures and the issue arises as to whether or
not the DNAPL has the potential to enter the fractures that are present. Such a scenario
is illustrated in Figure 11.1 where DNAPL has come to rest as pools in local bedrock
depressions. Also of interest is the case where DNAPL has been placed in a lagoon
excavated directly into fractured rock, or where DNAPL has been placed in a lined
waste impoundment where the liner of interest is constructed of a material such as clay
which may contain fractures.
This chapter will focus on the issue of DNAPL migration through fractured media
below the water table. It has been our experience that at the majority of sites where
DNAPL migration through fractures is an issue, the problem of concern has been DNAPL
entry into water filled fractures. As in the case of porous media, fractures in clay or rock
exhibit capillary properties. The concepts of interfacial tension, wettability, capillary
pressure, and relative permeability discussed in Chapter 3 are therefore all relevant to
the case of DNAPL migration through fractured media. The reader is encouraged to
review those concepts since they will not be repeated here. This chapter will in general
be restricted to a discussion of DNAPL as an immiscible phase, and will not address the
issue of DNAPL dissolution or matrix diffusion. These topics are treated separately in
Chapters 7 and 12, respectively.
11.2 DNAPl ENTRY INTO FRACTURES

In many cases, the volume of DNAPL released at ground surface will be sufficient to
bring about migration through overburden deposits such that the DNAPL comes to rest
above a fractured aquitard (or fractured bedrock) in the form of one or more pools. In
Sec. 11.2
339
order for DNAPL to enter a fracture underlying a pool, the capillary pressure at the base
of the pool must overcome the entry pressure of the fracture. Assuming that the DNAPL
is non-wetting with respect to water and that the system is at hydrostatic equilibrium,
the capillary pressure at the base of the pool is given by
(11.1)
where P; is the capillary pressure, Pnw is the non-wetting phase (DNAPL) density, Pw
is the wetting phase (water) density, g is the gravitational constant, and h is the height
of the pool. Eq.(l1.1) assumes that the top of the pool exists at a capillary pressure
of zero. The entry pressure of a rough-walled fracture is approximated by (Kueper and
McWhorter, 1991)
2CJcos e
Pc = - - e
(11.2)
where P; is the entry pressure of interest, CJ is the interfacial tension between DNAPL
and water, e is the contact angle on the fracture walls, and e is the largest fracture
aperture. If it is assumed that water in the fracture is in hydraulic equilibrium with water
in the porous medium overlying the fracture, Eqs.( 11.1) and (11.2) can be equated to
yield the minimum pool height H giving rise to entry into a fracture:
2CJcos e
H=-----
(Pnw -pw)ge
(11.3)
As discussed in Kueper and McWhorter (1991), this expression can be adjusted for cases
where the top of the pool does not necessarily exist at zero capillary pressure.
Eq.(11.3) shows that the minimum pool height required for entry into a fracture is
directly proportional to the DNAPL-water interfacial tension, and inversely proportional
to the difference in fluid densities and the fracture aperture. It is also important to note
that the critical pool height given by Eq.(11.3) is independent of depth below the water
table. This is consistent with the fact that capillary pressures dictate the behavior of
two-phase systems, with absolute pressures having very little influence in near-surface
environments.
Figure 11.2 presents a plot of minimum pool height vs. fracture aperture invaded
for a variety of interfacial tensions and DNAPL densities. Cases A and B represent
tetrachloroethylene (i.e., perchloroethylene, or PCE), a common chlorinated solvent having a high density (1620 kg/m"), Case A is assigned an interfacial tension of 0.045 N/m,
which is typical of industrial grade PCE which has not been exposed to impurities. Case B,
on the other hand, represents PCE having an interfacial tension of only 0.009 N/m
as a result of exposure to a surfactant or detergent. The Case A curve shows that
pool heights on the order of 0.2-1.0 m have the potential to invade fracture apertures on the order of 10-100 usi: (microns), while the Case B curve shows that much
smaller apertures are invaded for the same pool height as a result of the lower interfacial tension. Cases C and D in Figure 11.2 are assigned a DNAPL density of
1100 kg/m", and a DNAPL-water interfacial tension of 0.045 N/m and 0.009 N/m, respectively. It is clear that much greater pool heights are required for a low density DNAPL
340
Chap. 11
10.00 - r . . . . . . . . , - - - - r - - - - - - - - - - - - ,
-----E=
.!2fJ
1.00
==
-==
=--
~Z
0.10
10
100
1000
Aperture Invaded (microns)
10000
Figure 11.2 Pool height vs. fracture

aperture invaded.
to initiate entry into fractures, particularly when the interfacial tension has not been
lowered.
Once a DNAPL has entered a fracture or fracture network, progressively smaller
aperture fractures will be invaded if the DNAPL is allowed to extend itself vertically while remaining a continuous phase. In other words, the capillary pressure at the
base of the DNAPL accumulation is a function of not only the pool height in the overburden overlying the fractures in question, but also the height of DNAPL accumulated
in fractures beneath this pool. Consider, for example, the scenario illustrated in Figure 11.3. Shown is a 0.25 m high accumulation of trichloroethylene (TCE) in a local
bedrock depression. The TCE-water interfacial tension is assumed to be 0.04 N/m. The
capillary pressure at point A is 1,127.7 Pa, implying that fractures as small as 70 /Lm
will be invaded (fracture entry pressure given by 2a/e). In Figure 11.3, the aperture
at point A is greater than this and DNAPL has entered the fracture network. If the
DNAPL were to stop moving and reach a static distribution as shown, the capillary
pressure at points Band C would be 5,638.5 Pa and 10,149.2 Pa, respectively. This
would require that the fracture apertures at points Band C be no larger than 14.2 /Lm
and 7.9 /Lm, respectively. Since such a rapid reduction in fracture aperture with depth
is not likely in a rock environment, it is probable that the pool will not come to rest as
shown, but rather will continue to drain into the fracture network until it is completely
depleted.
In some cases, DNAPL may be pooled above a fractured aquitard as illustrated in
Figure 11.4 such that the capillary pressure at the base of the DNAPL pool is insufficient
to bring about entry into fractures. Figure 11.4 shows a pumping well situated in a lower
aquifer beneath the fractured aquitard. If aggressive pumping is carried out in this well,
water pressures in the aquitard will be lowered from below, resulting in an increase
in capillary pressure at the base of the DNAPL pool. This will promote the entry of
DNAPL into fractures, thereby worsening the extent of contamination at the site. If
the height of the DNAPL pooled above the fractures is very near to the critical pool
Sec. 11.2
341
Release
__ _
....u'if~Vk----:t-
-llL-
sz
BJ'
1m
.L~C
Figure 11.3 Pool of TCE in bedrock depression. TCE can enter progressively smaller
aperture fractures with depth as a result of an increase in capillary pressure brought about
by vertical accumulation.
height, even a slight lowering of water pressure from below may bring about entry into
the previously uninvaded fractures.
At many sites where overburden covers fractured bedrock or a fractured aquitard,
soil samples can be obtained and analyzed to determine if an actual DNAPL phase is
present. If a soil sample obtained from immediately above a fractured medium is found
to contain residual or pooled DNAPL, this can be used as an indication of whether or
not the DNAPL had the potential to invade a given size fracture. DNAPL found in a
~ \It~
r?
'\7
Pooled DNAPL ~
UpperAquifer
i~
Fractured Clay
Aquitard
~
Lower Aquifer
Figure 11.4 DNAPL pooled above a fractured aquitard. Pumping from below aquitard
will promote the downward re-mobilization of DNAPL.
342
Chap. 11
silt as either residual or pools, for example, must have existed at either a high capillary
pressure or a low interfacial tension to be able to invade such a fine grained material.
These same conditions promote DNAPL entry into small fractures.
Figure 11.5 presents a plot of aperture invaded vs. DNAPL saturation in porous
media of varying permeability immediately above a fracture. To construct this plot it was
necessary to establish the relationship between capillary pressure and saturation for the
porous medium and the fluid pair of interest. A Brooks and Corey (1964) capillary pressure relationship and a DNAPL-water interfacial tension of 0.045 N/m has been assumed
here. The Brooks-Corey parameters are based on PCE-water capillary pressure curves
measured on samples of sand obtained from the Borden aquifer near Alliston, Ontario
(Kueper and McWhorter, 1991). The relationship between capillary pressure and permeability is assumed to follow a modified form of the Leverett function (Leverett, 1941) as
established by Kueper and Frind (1991). The figure clearly shows that DNAPL present
in progressively lower permeability porous media has the potential to enter progressively
smaller aperture fractures.
In the above examples it has been assumed that the water beneath a DNAPL pool
is in hydrostatic equilibrium with water above the pool. Figure 11.6 presents a scenario
where this is not the case. Shown is a lined lagoon containing both DNAPL and water.
Because the liner is comprised of a low permeability material such as clay, the water table
in the formation outside of the liner is not at the same elevation as the water inside the
Sec. 11.3
343
Rates of DNAPL Migration through Fractures
lagoon. In this case, the height of the water table outside the liner has a direct influence
on the critical pool height of DNAPL that can be supported inside the lagoon prior to
penetration of fractures. DNAPL will invade the fractures when the following condition
is satisfied (Kueper et al., 1992):
Pnw
2acos
A+B--C::,:--Pw
epwg
(11.4)
where A is the thickness of the water layer inside the lagoon, B is the thickness of the
DNAPL layer, and C is the height of the water table above the base of the lagoon. It is
clear that maintaining a high water table outside of the liner is a step that can be taken
to minimize the risk of DNAPL entry into fractures.
Porous Media
Aquifer
Figure 11.6 DNAPL accumulation in a lined waste disposal pond. (Reprinted with
permission from Kueper et al. (1992).)
11.3 RATES OF DNAPL MIGRATION THROUGH FRACTURES

Once having entered a fracture, DNAPL will preferentially migrate through the larger
aperture pathways of a fracture plane since these offer the least capillary resistance
to movement. The overall saturation of DNAPL achieved in the fracture during this
migration will be a function of the maximum capillary pressure generated. Recent laboratory measurements by Reitsma and Kueper (1994) have demonstrated that rough-walled
fractures exhibit capillary behavior analogous to porous media. Figure 11.7 presents a
capillary pressure-saturation curve measured in a single fracture in limestone. The fracture was subjected to a normal stress of 12.5 kPa. The hydraulic aperture of the fracture
at this loading was measured to be 112 fLm, which is substantially less than the 189 fLm
aperture corresponding to initial invasion of the non-wetting fluid (fracture entry pressure). The drainage curve shown in the figure was found to be well represented by a
Brooks-Corey type of porous media capillary pressure curve. This is not surprising given
the analogy between the pores and pore throats in a porous medium and the variable
aperture regions in a rough-walled fracture. The figure clearly shows that higher nonwetting saturations are attained in the fracture plane at higher capillary pressures. It is
344
Chap. 11
12
III
-'Q)
CO
o imbibition curve
10
@ point on capillary
pressure curve
corresponding to
respective aperture
E
e
primary drainage curve
11
hydraulic aperture (112 urn)
C.
Brooks-Corey
entry pressure
aperture (189 urn)
0-+---,-----,-----.---,-------,------,----,-----,---,------,
10
Volume of Water Removed From Sample (ml)

Figure 11.7 Capillary pressure-saturation curve for a limestone fracture. (Reprinted with permission from Reitsma and Kueper (1994)).
important to note, however, that the fracture exhibits a virtually non-existent residual
non-wetting phase saturation upon wetting.
To investigate the rates at which DNAPL can traverse a fractured aquitard, numerical simulations were carried out in the solution domain illustrated in Figure 11.8 using
the DNAPL fracture migration model developed by Kueper and McWhorter (1991). The
figure shows DNAPL pooled at the base of overburden deposits which overlie a 5 m
thick fractured aquitard. The DNAPL of interest is assigned a viscosity of 0.00057 Pa-s
and an interfacial tension with water of 0.045 N/m. Figure 11.9 presents a plot of the
time taken for DNAPL to first reach the lower aquifer as a function of fracture aperture.
The DNAPL pool height is set at 0.5 m. Two cases are shown, one for a DNAPL density
of 1200 kg/m ' and one for a density of 1460 kg/m'. It is clear that the first arrival times
are very sensitive to fracture aperture, consistent with the fact that fracture transmissivity
is proportional to the cube of the aperture (de Marsily, 1986). For the higher density
case (properties of TeE), very short first arrival times are exhibited for fracture apertures
larger than 50 /Lm. The lower curve approaches an asymptote at an aperture of 39.9 /Lm,
which is the smallest aperture that can be invaded by the 0.5 m high pool. For the lower
density case the first arrival times are longer and the asymptotic aperture is higher since
the 0.5 m pool height generates a smaller capillary pressure, and since the gravity driving
345
Rates of DNAPL Migration through Fractures
Sec. 11.3
"Release Location
'Sr
'Upper Aquifer
Clay Aquitard
Lower Aquifer
Figure 11.8 DNAPL pooled above fractured aquitard. Solution domain for results presented in Figure 11.9.
250,..-------------,
,--...
Vl
~ Density= 1200 kg/m~
co 200
GB-8-eI
Density= 1460 kg!m
E 150
'--""
Q)
L
~
100
Q)
50
0...
-c
10
20
Time (hours)
30
Figure 11.9 Time taken for first arri val of

DNAPL at lower aquifer versus fracture
aperture.
force is reduced. It follows that larger pool heights and lower viscosities will in general
bring about quicker first arrival times.
To examine the rate at which DNAPL can migrate along horizontal to sub-horizontal
fractures, numerical simulations were carried out in the solution domain illustrated in
Figure 11.10. Shown is a pool of DNAPL which has accumulated in a bedrock depression.
A set of parallel, slightly dipping fractures is shown to extend a large horizontal distance
from the pool. The DNAPL pool height is set at 0.25 m, and the DNAPL is assigned the
346
sz
Chap. 11
--
-=-
/Bedrock
Figure 11.10 DNAPL pool feeding horizontal fractures in bedrock. Solution domain
for results presented in Figure 11.\1.
<=
... ...<= ...<= ... ... ...
Q
Q
Q
Q
Q
Q
<= Q<= QQ
<=
Time (days)
Figure 11.11 Distance vs. time for

leading edge of DNAPL migration in a set
of fractures.
properties of TCE with an interfacial tension of 0.030 N/m. Figure 11.11 presents a plot of
distance vs. time for the leading edge of the DNAPL front for three cases. Case A, which
represents a set of parallel 100 {tm fractures dipping at 2 degrees, shows very high rates of
migration. Case B, which represents the same set of fractures with a horizontal orientation,
shows comparatively slower rates because of the reduced gravity driving force. Case C
represents a set of 75 {tm fractures with a horizontal orientation; the rates of migration
are substantially slower than the other cases because of the reduced intrinsic permeability
of the fractures, and because of a reduced relative permeability to the TCE as a result
of a higher fracture entry pressure. The rates of migration for all three cases would be
higher for a greater pool height at the inflow because of an increase in capillary pressure.
Sec. 11.4
347
11.4 RESIDUAL AND POOLED DNAPL

Once the source of the DNAPL to a fracture has been depleted, the DNAPL will redistribute itself in the form of residual and pools. Although quantitative studies are
scarce, residual DNAPL is expected to form in fractures due to by-passing and snap-off
mechanisms analogous to residual formation in porous media. The underlying premise
here is that the rough-walled nature of fractures will provide enough variability in topology to allow these processes to take place. However, since the topology of a rough-walled
fracture is not as irregular as in a porous medium, it may be reasoned that residual in the
form of small, disconnected blobs and ganglia of organic liquid will not be abundant.
This is consistent with the small residual contents in simulated fractures constructed
from glass plates by Schwille (1988), and the extremely low residual non-wetting phase
saturation exhibited by the capillary pressure curve measured by Reitsma and Kueper
(1994).
Pools are distinct from residual in that they are formed when a migrating DNAPL
body can no longer overcome capillary resistance at the leading edge. Figure 11.12
illustrates a pool of DNAPL which has formed in an inclined fracture as a result of
a narrowing of the fracture at the leading edge of the pool. By carrying out a onedimensional force balance and assuming that groundwater flow in the fracture is at
steady-state, the pool will remain stationary when
!'1p L sina + h(O) _ h(L) :s PeeL) - Pc(O)
(11.5)
Pw
Pwg
where L is the pool length, !'1p is the difference in density between DNAPL and water,
a is the dip of the fracture, h (0) is the hydraulic head at the trailing edge of the pool,
h(L) is the hydraulic head at the leading edge of the pool, PeeL) is the entry pressure
of the fracture at the leading edge of the pool, and Pc(O) is the capillary pressure at the
trailing edge of the pool.
Figure 11.12
DNAPL pool in a fracture.
348
Chap. 11
The first term in Eq.( 11.5) represents the influence of gravity acting down-dip.
Applied to a horizontal fracture, this term is zero. The second term represents the influence
of the groundwater gradient across the pool, and reduces to zero for a hydrostatic wetting
phase distribution. The term on the right-hand side represents the capillary resistance of
the fracture. Eq.(11.5) can be re-arranged to solve for L, the maximum stable pool length
for a given set of flow conditions. It is clear that upward hydraulic gradients result in a
longer stable pool length, and that longer pools can be sustained in more shallow dipping
fractures. It can also be seen that an in-situ lowering of interfacial tension will reduce
the capillary resistance of a fracture, resulting in possible pool mobilization.
In the case where groundwater pumping is initiated in the vicinity of pooled
DNAPL, the increase in hydraulic gradient across the pool will lead to the possibility
that the pool will be displaced further along the fracture. Such a scenario can arise if
a pump-and-treat system is operated in the vicinity of pooled DNAPL. Figure 11.13 illustrates a hypothetical site where DNAPL pools exist in fractures. A series of pumping
wells need to be placed to bring about hydraulic interception of the dissolved phase
plume which is evolving, but it is not known how far downstream of the DNAPL
pools the wells must be placed if mobilization of the DNAPL in the fractures is to
be avoided. Eq.(11.5) demonstrates that any increase in hydraulic gradient will mobilize DNAPL if pools exist at their maximum stable length. For pool lengths below the
maximum stable length, a certain incremental increase in hydraulic gradient can be tolerated.
In practice it is generally not possible to determine whether or not pools exist at
or near their maximum stable length. A conservative approach would therefore be to
sz
/ " Residual DNAPL
~-.:=-
=-----
-+~
~
~
...
DNAPL Pooled
in Fractures
Figure 11.13
~\
safe distance?
~l
Pool mobilization due to operation of pump-and-treat recovery wells.
Sec. 11.5
DNAPL Migration through Fracture Networks
349
locate plume interception wells far downgradient from suspected pool locations so that
the impact of pumping on the pools themselves will be minimal. An alternative approach
would be to place plume interception wells directly within the region of pooled DNAPL.
In this case any pool mobilization would be back towards the already-contaminated
region. The latter approach has the added advantage of minimizing the extent of aquifer
that will be impacted by dissolved contamination during the operation of the pump-andtreat system. Care must be taken, however, to ensure that capture zones extend far enough
downstream to encompass all pooled DNAPL.
11.5 DNAPL MIGRATION THROUGH FRACTURE NETWORKS

To date, there has been very little research carried out examining the migration of DNAPL
through fracture networks. This is in part due to the fact that DNAPL migration through
fractured media is in itself a relatively new area of interest, and in part because both
laboratory and field experimentation are very difficult in fracture networks. In addition,
the development of numerical models to simulate DNAPL migration through fracture
networks is only recently underway. The work which has been performed to date examining DNAPL in single fractures, however, can be used to direct a general discussion.
Once initial entry into a fracture network has taken place, a DNAPL will preferentially migrate through the larger aperture fractures, entering intersecting fractures only
if the local capillary pressure exceeds the entry pressure of the newly encountered fractures. The overall pattern of DNAPL migration will be governed by both fluid and media
properties. Figure 11.14 shows a sequence of horizontally bedded sedimentary rocks with
a waste lagoon containing DNAPL at ground surface. The rock units differ from each
other in overall bulk permeability. The upper unit is highly fractured while the lower
unit is less fractured and contains fractures with smaller apertures. Figure 11.14.a shows
the possible DNAPL migration pathways when the DNAPL has a low interfacial tension.
As can be seen, a large number of individual fractures are invaded because the fracture
entry pressures are low. This results in little lateral spreading through the fracture system
and large ultimate depths of migration.
Figure 11.14.b illustrates the migration pathways which can arise from a DNAPL
having a high interfacial tension with water (e.g., 0.045 N/m). In this case not as many
individual fractures are invaded because fracture entry pressures are higher. The pool
height in the lagoon and the density driving force can only generate capillary pressures
sufficient to enter the largest aperture fractures. In addition, substantial lateral flow takes
place above the less permeable rock unit since high capillary pressures are required for
penetration. Even if the lower permeability rock unit becomes penetrated, lateral flow
will continue above it since the relative permeability to DNAPL will be higher in the
overlying, more fractured unit. As shown in the figure, large amounts of lateral spreading can occur through major bedding partings. Because DNAPL-water interfacial tension
largely dictates what the ultimate lateral and vertical extent of DNAPL migration will
be in a fracture network, care should be taken to have this value measured whenever a
sample of the DNAPL of interest can be obtained.
350
Chap. 11
JI' Waste Lagoon
II.l".
I
I-I
I
I
\ \
low-K
rock unit
I
I
(a)
JI' Waste Lagoon
11.".. . .
I
high-K
rock unit
I
I
\ \
Lateral
Migration
~
I
I
(b)
Figure 11.14 a. DNAPL migration pathways for a low interfacial tension system;
b. DNAPL migration pathways for a high interfacial tension system.
In addition to the properties of the DNAPL itself, it should be clear that fracture connectivity, spacing, and orientation also have pronounced effects on DNAPL
migration pathways through fracture networks. As in porous media, DNAPL migration will be largely structure controlled. DNAPL will migrate through those pathways
which offer the least capillary resistance. These pathways will not necessarily be the
same as those involved in the flow of the groundwater. Moreover, since hydraulic gradients in most natural groundwater systems are typically very low, they will have little
influence on the directions of DNAPL migration through the fractures. A comprehensive monitoring program in fractured media should therefore focus on having boreholes
placed according to structure, not necessarily according to the direction of groundwater
flow.
Sec. 11.6
351
11.6 HEIGHT OF DNAPL IN A WELL
In the event that DNAPL is observed in a borehole in a fractured medium, it can obviously be concluded that the site is a DNAPL site. Beyond this, however, it is difficult to
infer exactly what the conditions are in the surrounding formation. Figure 11.15 presents
a series of schematics where boreholes have intersected a DNAPL pool in fractured
media. In Case A, the presence of DNAPL in the wellbore corresponds to the presence
of DNAPL in the formation at the same general elevation. If the top of the pool in
the formation exists at a capillary pressure of zero, the elevation of the water-DNAPL
interface in the wellbore wiIl equal the elevation of the top of the pool in the formation. What is not known, however, is the distance to which the pool extends below the
borehole.
In Case B, the identical height of DNAPL exists in the borehole as in Case A,
but no DNAPL is present in the formation at this elevation. In this case, DNAPL has
entered at an upper elevation and accumulated at the base of the borehole. The DNAPL
does not exit the borehole because the height of accumulation is insufficient to generate
a capiIlary pressure in excess of the local fracture entry pressures. Case C illustrates a
case of borehole short-circuiting where the capillary pressure has exceeded the fracture
entry pressures, resulting in penetration of fractures at lower elevation.
(a)
open
borehole
residual
DNAPL
(b)
(c)
Figure 11.15 Height of DNAPL in a well: a. DNAPL in well corresponds to DNAPL in formation;
b. DNAPL in well has entered from an upper elevation and does not correspond to DNAPL in
formation at lower elevation; c. borehole short-circuiting.
352
Chap. 11
11.7 CONCLUSIONS
Although research pertaining to multiphase flow in fractured media is still in its infancy,
much has been learned by drawing analogies with the flow of immiscible fluids in porous
media. In particular, the concepts of wettability, capillarity, and relative permeability
apply as well to fractured media as they do to porous media. The underlying reason
for this is the close analogy which exists between the distribution of apertures in a
fracture plane or network, and the distribution of pores and pore throats in a porous
medium.
Laboratory and numerical experimentation to date has focused primarily on the
flow of two immiscible fluids through a single fracture. While advances in the numerical modeling of multiphase flow through fracture networks will certainly be made in
the near future, controlled physical experimentation in actual fracture networks will
not likely be commonplace because of the inherent difficulty in artificially creating
such networks in the laboratory, and because intentional releases of DNAPL at actual field sites is usually prohibited. The study of DNAPL behavior at accidental spill
sites generally does not serve as a fully adequate alternative because of uncertainty in
the volumes, rates, times, location, condition, and composition of the liquid(s) released.
Our overall understanding of DNAPL migration through fracture networks will therefore be based largely on theoretical analyses and inferences obtained at accidental spill
sites.
ACKNOWLEDGEMENTS
11.8 REFERENCES
Brooks, R. H. and A. T. Corey (1964) "Hydraulic properties of porous media", Hydrology Paper
No.3, Civ. Eng. Dept., Colorado State University, Fort Collins, CO, U.S.A.
de Marsily, G. (1986) Quantitative Hydrogeology, Academic Press, New York, NY.
Kueper, B. H. and E. O. Frind (1991) "Two-phase flow in heterogeneous porous media. 2. Model
Application", Water Resources Research, 27, 1058-1070.
Kueper, B. H. and D. B. McWhorter (1991) "The behavior of dense, non-aqueous phase liquids
in fractured clay and rock", Ground Water, 29, 716-728
Kueper, B. H., C. S. Haase, and H. King (1992) "Leakage of DNAPL from waste impoundments constructed in fractured rock and clay: Theory and case history", Canadian Geotechnical
Journal, 29, 234-244.
Sec. 11.8
References
353
Leverett, M. C. (1941) "Capillary behavior in porous solids", Trans. A/ME, Petrol. Div., 142,
152-169.
Reitsma, S. and B. H. Kueper (1994) "Laboratory measurement of capillary pressure-saturation
curves in natural rock fractures", Water Resources Research, In press.
Schwille, F. (1988) Dense Chlorinated Solvents in Porous and Fractured Media: Model Experiments. Translated by James F. Pankow, Lewis Publishers, Boca Raton, Florida.
12
Effects of Molecular
The
Diffusion on DNAPL Behavior
in Fractured Porous Media
Beth L. Parker, John A. Cherry, and Robert W. Gillham
Waterloo Centre for Groundwater Research

Waterloo, Ontario, Canada N2L 3G1
Page
12.1
12.2
12.3
12.4
12.5
12.6
Introduction
12.2.1 Approach
12.2.5 Limitations
12.3.5 Remediation
12.4.1 Example 1: Solute Transport Through a Vertically Fractured
Clay Aquitard
Conclusions
References
356
361
361
364
365
371
375
378
378
379
380
381
381
382
382
386
387
390
ABSTRACT
In fractured media, diffusion of contaminant mass from highly conductive fractures into
the relatively immobile pore water of the matrix between fractures affects the persistence of dense non-aqueous phase liquids (DNAPLs) in fractures. This diffusion causes
a redistribution of the contaminant mass within source areas in fractured porous media such as unlithified clay-rich deposits and sedimentary rocks. A theoretical analysis
using parameter values typical for chlorinated solvents and fractured geologic media indicate that the chemical mass storage capacity of the matrix for dissolved and sorbed
355
356
The Effects of Molecular Diffusion on DNAPL Behavior in Fractured Porous Media
Chap. 12
chemical mass often greatly exceeds the DNAPL-phase storage capacity of the adjacent fractures. This condition allows complete disappearance of the DNAPL phase from
fractures by dissolution and subsequent diffusion into the porous matrix. For idealized
fracture systems, these diffusion rates are shown to be very rapid for the common chlorinated solvents (dichloromethane, trichloroethylene, and tetrachloroethylene), ranging
from a few minutes to a few years for fracture and matrix properties typical of clay-rich
deposits, and days to a few decades for sedimentary rock systems. Hence, the contaminant mass can undergo a complete change in state and position as DNAPL located
in high permeability fractures dissolves and moves into the matrix. These conclusions
have many implications regarding site monitoring and interpretations for DNAPLs in
fractured source zones. For example, mass removal by in-situ methods is likely to be
controlled by reverse diffusion and desorption from the low permeability matrix between
fractures rather than by dissolution and mobilization of DNAPL from difficult-to-find
DNAPL-Iaden fractures.
12.1 INTRODUCTION
The flow of dense non-aqueous phase liquids (DNAPLs) in fractured media is discussed in Chapter 11. A simplifying assumption made in that chapter so as to allow
one to focus on the physics of the flow was that the material making up the fractured
medium has insufficient matrix porosity to allow significant contaminant mass, either
as the DNAPL phase itself, or dissolved in groundwater, to enter the matrix. For certain
field situations this assumption is valid. However, for others, the porosity of the matrix
material between the fractures can have important effects on the fate and behavior of
DNAPLs in fracture networks. The focus of this chapter will therefore be on the entry
of dissolved contaminants into the matrix porosity. Fractured geologic media in which
the matrix porosity is most likely to have significant influence on DNAPL behavior include fractured clayey or silty deposits as well as sedimentary rocks in which the matrix
porosity is appreciable. The latter include shales, siltstones, sandstones, limestones, and
dolomites. Crystalline rocks such as granites and marble generally have insignificant
matrix porosity.
Although fractures will be conceptualized in this chapter as smooth parallel plates,
in reality they have rough surfaces with irregular apertures which are closed over parts
of their areas (Figure 12.1.a). This conceptualization, though a severe simplification of
reality, allows useful analyses of contaminant behavior in fractured porous media to be
made. Also, while the DNAPL phase can enter into and move within a fracture network,
we will assume that it cannot enter as a phase into matrix blocks between the fractures
(Figure 12.l.b). Thus, for matrix blocks, we will be assuming that the entry pressure for
the DNAPL is not exceeded under field conditions. This assumption is valid for many
though not all situations in fractured porous media.
In addition to the matrix porosity, important properties of fractured porous media
include the fracture aperture, the fracture porosity, the sorption capacities of the
matrix material for the contaminants of interest, and the effective diffusion coefficients
of those contaminants in the matrix. The matrix porosity (m) (also referred to as the
Sec. 12.1
357
Introduction
IDEAL
REAL
FRACTURE APERTURE
(b)
(0 )
-I 2 bl--
2b
--I I--
Figure 12.1 a. Idealized smooth parallel-plate fracture with uniform aperture VS. actual rough-walled, variable aperture fractures. b. Schematic diagram of fracture network
within a geologic medium, and definition of terms.
"primary porosity") is the void space between the individual solid grains in the matrix
blocks. The fracture porosity (I) is the void space offered only by the fracture network.
The parameter I is a function of the mean fracture aperture and the mean fracture
spacing. Few reliable measurements have been made for fracture porosity or fracture
aperture because of the exceptional difficulties involved in making such determinations.
The literature suggests that for fractured clayey or silty deposits in the groundwater zone,
fracture porosity, expressed as the percent of void space provided by fractures per unit
volume of bulk medium, generally ranges from 0.0001 to 0.1 %; for fractured sedimentary
rocks, with the exception of solution-channelled or karstic limestones or dolomites, the
upper end of the range is 1%.
Estimated values for fracture porosities and apertures are normally derived from
field measurements of bulk hydraulic conductivity carried out in boreholes or wells, combined with information on the frequency of fractures (distances between hydraulicallyconductive fractures) derived from cores, borehole geophysical logs or outcrops. Figure 12.2.a, adapted from Hoek and Bray (1981), illustrates the dependence of the bulk
hydraulic conductivity on the fracture spacing and the fracture aperture. The relations are
for a set of parallel fractures with uniform spacing and smooth parallel-plate apertures.
The bulk hydraulic conductivity increases directly with the number of fractures per unit
length, and directly with the cube of the fracture aperture. Figure 12.2.b illustrates the
dependence of the fracture porosity on fracture aperture and the fracture spacing.
As an example, a fracture porosity on the order of 0.1 % would be relatively large,
and if the fractures were aligned with the direction of groundwater flow, they would
normally cause a very large bulk hydraulic conductivity (greater than 6 x 10- 4 mls).
(Note that a 1.0 mm wide fracture spaced every meter of rock would represent a 0.1 %
fracture porosity case). In contrast, matrix porosities are commonly much larger than
358
Chap. 12
1%: non-indurated clayey deposits typically have matrix porosities in the range of 20 to
70%; many sedimentary rocks have matrix porosity values of 5 to 15%.
Based on the above information, it can be concluded that the void space in the
matrix of fractured porous media is commonly orders of magnitude larger than the void
space provided by the fracture network. This observation has important implications for
I-D Parallel Fracture System
a.
0.1
0.01
!o
0.001
'"8
0.0001
IE-05
.;:
::l
.~ IE-
:r:
"''3"
trl
IE-
10
100
1000
10000
Fracture Aperture (microns)
b.
I-D Parallel Fracture System

0.01,.-------------"'7-----,....-----:71
0.001
o 0.0001
&
~
IE-OS
IE-
10
100
1000
10000
Fracture Aperture (microns)

Figure 12.2
a. Bulk hydraulic conductivity (after Hoek and Bray, 1981), and b. fracture
porosity as a function of fracture spacing and fracture aperture for a set of parallel
fractures in one direction.
Sec. 12.1
359
Introduction
how DNAPLs behave in fractured porous media. In particular, the chemical mass storage
capacity of the fractures themselves per unit volume of geologic medium will often be
very small. Therefore, a small volume of DNAPL released into a fractured porous medium
will often spread extensively throughout the fracture network and eventually occupy a
large volume of bulk medium. In addition to the very small total void space due to
fractures, the non-wetting phase liquid saturation within the fractures is likely to be less
than 1.0. (As described in Chapter 11, this saturation value is a function of the fluid
pressures and of the variations in fracture aperture along the fracture surface.) In nearly
all cases, it is expected that the DNAPL of interest will be the non-wetting fluid, and
therefore that a thin film of water (the wetting fluid) will exist between the DNAPL and
the fracture walls (Figure 12.3.a). Because fracture apertures are typically quite small, a
DNAPL in fractures will exhibit a large surface area in contact with the water.
Molecular diffusion is the process by which chemical molecules move due to
random, "Brownian" motion. If a concentration gradient is present, this motion causes
molecules to move from regions of high concentration to regions of low concentration.
This process is described mathematically by Fick's First Law as
JD
acw
ax-
-De -
(12.1)
where J D is the diffusive flux, is the system porosity (in this case m), De is the
effective diffusion coefficient in the porous medium, C w is the dissolved aqueous-phase
concentration, and x is distance. Eq.(12.1) is one of several gradient-flux laws used in the
physical sciences to describe processes that occur at the molecular scale. Other examples
include Darcy's Law for fluid flow through a porous medium due to pressure gradients,
Fourier's Law for conduction of heat due to temperature gradients, and Newton's Law
of viscosity where the magnitude of fluid shear stress created by velocity gradients is
directly proportional to the rate of deformation of the fluid particles.
It was Foster (1975) who first recognized the important effect of diffusive loss of
dissolved chemical mass from fractures into rock matrices. Foster (1975) used this process
to explain the distribution of tritium in fractured limestone (chalk). Following Foster
(1975), there have been numerous studies of the diffusive loss of dissolved chemical
mass from fractures to the matrix in fractured rock and in fractured clayey deposits. For
the latter, these include field studies (Day, 1977; Goodall and Quigley, 1977; McKay
et al., I993a, 1993b, and 1993c), laboratory studies (Grisak et al., 1980; Sudicky et al.,
1985; and Starr et al., 1985), and mathematical modeling studies (Grisak and Pickens,
1980 and 1981; Tang et al., 1981; Sudicky and Frind, 1982; Germain and Frind; 1989;
Sudicky and McLaren, 1992). These studies have established the dominant influences
of fracture aperture, matrix porosity, and De for the contaminant on the movement of
dissolved species in fracture networks.
Most chlorinated solvents have limited, but nevertheless significant, aqueous solubilities that allow the DNAPL phase to slowly dissolve across the contact area between a
DNAPL and adjacent water. Parker et al. (1994) and Parker and McWhorter (1994) have
discussed how chemical mass will dissolve into the adjacent water film in a fracture, that
a chemical concentration gradient will then be established in the aqueous phase, and that
360

FRACTURE APERTURE
(0 )
( b)
-12bJ-
Chap. 12
FRACTURE APERTURE
-12bJ-
cf>m
I-FRACTURE .,
SPACING
(c)
FRACTURE APERTURE
-12bJ-
SPACING
Figure 12.3 Conceptual model for the distribution of DNAPL in a saturated, fractured porous
medium with DNAPL as the non-wetting fluid in the fractures. a. Early time: interconnected DNAPL
phase. b. Intermediate time: disconnected DNAPL blobs. c. Late time: diffusion haloes in the matrix
around fractures that once contained DNAPL.
contaminant mass will then move into the porous matrix by molecular diffusion. Using
I-D and 3-D models for idealized systems, they showed that in some fractured porous
media, some DNAPLs (particularly the chlorinated solvents), can disappear from fractures in very short time periods, resulting in nearly all of the contaminant mass ultimately
moving into the matrix as dissolved and sorbed contaminant.
The above process is described schematically in Figure 12.3, which illustrates the
distribution of the contaminant mass in a fractured porous medium at early, intermediate,
and late times. At early times, immobile DNAPL and water co-exist in the fractures;
the DNAPL is the non-wetting fluid and is present in the largest aperture regions in the
Sec. 12.2
361
fractures, in accordance with capillary pressure considerations. The DNAPL is surrounded

by a film of water (as the wetting fluid), and dissolution of the DNAPL phase into
the aqueous phase occurs. The concentration of the chemical dissolved in the water
film immediately adjacent to the DNAPL quickly becomes saturated with that chemical,
and dissolved mass concurrently diffuses into the immobile pore water in the matrix.
In Figure 12.3.b, a significant amount of the DNAPL has dissolved, thereby causing
the remaining DNAPL to disconnect and redistribute within the larger fracture aperture
regions so as to achieve minimal interfacial energy between the two fluids. Eventually
(see Figure 12.3.c), the DNAPL dissolves completely, resulting in its disappearance as
a separate phase. The chemical mass then exists as diffusion haloes around the fractures
that once contained DNAPL. The time periods required for complete DNAPL phase
disappearance are specific to the chemical and porous medium conditions. Subsequent
flushing of the fractures with uncontaminated groundwater will cause diffusion out of
the matrix and into the moving groundwater as the concentration gradients are reversed.
Based on the work of Parker et al. (1994) and Parker and McWhorter (1994), we
now describe the processes and consequences of DNAPL-phase disappearance from fractures. This is followed by considerations of the effects of matrix diffusion on dissolvedphase plumes migrating through fractured porous media under the influence of advection
and dispersion in the fracture network, and diffusion and sorption in the matrix. The
implications for migration rates towards downgradient receptors and for current practices
used in subsurface investigations and remediation are discussed.
12.2 DNAPL PERSISTENCE IN FRACTURES
12.2.1 Approach
Three sets of calculations are presented in this section to illustrate the relative importance
of the diffusive disappearance of an organic liquid phase containing a single component
from fractures within a porous medium. The first set of calculations simply compare the
mass storage capacity for a DNAPL material within fractures to the mass storage capacity
within the matrix blocks that are located between the fractures of a well-characterized
fracture network in a particular natural deposit of clay till. The second and third sets
of calculations illustrate how quickly mass transfer by diffusion into the porous matrix
blocks occurs for typical porous geologic media. Methods to calculate the times for
DNAPL-phase disappearance in the fractures due to dissolved-phase diffusion into the
matrix are presented. Two idealized fracture cases are considered: a single, parallel-plate
fracture adjacent to an infinite porous medium, and sets of parallel fractures in one and
three dimensions with corresponding matrix blocks having finite dimensions (i.e., finite
distances between fractures).
For this study, four DNAPL chemicals and three fractured media types were selected to represent the range of values typical for real field sites and common DNAPL
chemicals. These calculations are also used to illustrate the sensitivity effects of various
parameters typical of these environments on the time for DNAPL phase disappearance.
The four chemicals selected for these calculations include three chlorinated aliphatics
362
Chap. 12
(tetrachloroethylene (PCE), trichloroethylene (TCE), and dichloromethane (DCM, and

one chlorinated aromatic (1,2,4-trichlorobenzene (TCB. These compounds are commonly used in industry as single-component liquids. Their aqueous solubilities span
a wide range that encompasses nearly all of the DNAPL-type contaminants found in
groundwater. Table 12.1 gives the relevant physical-chemical properties for the calculations carried out here. The relevant geologic media parameters include the matrix porosity,
dry bulk density, fractional organic carbon content, and tortuosity. Parker et at. (1994)
summarize measured values obtained from the literature for a variety of fractured geologic media. Based on these values, ranges in parameters were selected to be typical of
a clay, a sandstone/shale, and a crystalline rock; Table 12.2 presents the values selected
for the calculations.
TABLE 12.1 Chemical Properties of Selected Halogenated and Non-Halogenated
Organic Compounds Used for Diffusive Mass Loss Calculations (20-25C).
Compound
Molecular
Formula
SIw
mglL
OCM
TCE
PCE
1,2A-TCB
CHzCh
Cz HCI3
C2Cl4
C6H 3 CI3
20,000
1.4204
240 4
305
x.;
mUg
11M
92M
3806a
3,2006b
MW
85.0
131.5
166.0
181.5
1.33
1.46
1.63
1.46 5
12.4
10.1
9.4
5.3
x \0-6
x \0-6
x \0-6
x \0-6
Notes:
I Schwille
(1988)
2Bonoli & Witherspoon (1968) for 20C

3Wilke & Chang (1955) for 20
DC
4Broholm et al. (1992) for 23-24C

5Mercer & Cohen (1990)
"Schwarzcnbach et al. (1993)
a
logK om
log K om
=
=
-0.701ogSw +0.35
-0.93 log Sw - 0.17
where K oc = 1.724K nm (Fetter, 1993)
The scientific literature provides laboratory-measured values for the free-solution

diffusion coefficients Do of many organic chemicals and mixtures of chemicals. However, no such values could be found for the four compounds considered here. Do values
for the four compounds were therefore calculated using a correlation equation developed
by Wilke and Chang (1955) for dilute solutions of single organic solutes in water. It is
the apparent tortuosity factor r which relates the free-solution diffusion coefficient for
a chemical species to its effective diffusion coefficient De in a porous medium. The
value of r can range between zero and unity. Correlation equations for r as a function
of porosity (in this case 4Jm) are found in the literature in the general form
De
- = r =::
Do
4JP
(12.2)
Sec. 12.2
363
TABLE 12.2 Geologic Parameters for Three Types of Fractured Media Used for Comparison of
TCE Mass Loss Rates Due to Diffusion and Sorption to Matrix Solids.
Type of Geological Medium

Parameter
Clay!
Shale/Sandstone?
Granite 3
porosity:
0.35
0.10
0.006
bulk density: Ph (g/cm')
1.6
2.4
2.63
fraction organic carbon: foe
0.01
0.001
0.33
0.10
0.06
apparent tortuosity:
Combined Parameters: TCE with Geologic Medium

effective diffusion coefficient: De (cm 2/s)
retardation factor: R
3.3 x 10- 6
1.0 x 10- 6
5.2
3.2
6.0 x 10- 7
NOTES:
!Johnson et al. (1989)
2Feenstra et al. (1984); Barone et al. (1990)
3 Skagius
and Neretnieks (1986)
where the exponent p varies between 1.3 to 5.4, depending on the type of porous geologic
medium (Dullien, 1992; Lerman, 1979; and Millington and Quirk, 1961). Low porosity
values result in small r values and low De values. Eq.(l2.2) is useful because, in many
cases, measured De values are rarely available.
The chemical mass in the matrix consists of the dissolved mass in the pore water
plus the mass sorbed to the matrix solids. The ratio between the two is given by the
retardation factor R
R = 1 + Pb K d
(12.3)
<Pm
where K d (mUg) is the distribution coefficient for the compound of interest between
the solid and aqueous phases. The value of R will be a constant for a particular chemical/porous medium combination. For non-ionic organic compounds, the equilibrium distribution coefficient K d (mUg) for the compound of interest between the solid and
aqueous phases is commonly predicted using the fraction of organic carbon foe in the
porous matrix and the partition coefficient of the compound of interest to organic carbon
K oe (mUg) together with the equation
(12.4)
K oe values can be predicted using a correlation equation such as
log K oe = -a log Sw + b
(12.5)
where the appropriate correlation coefficients a and b and the aqueous solubility values
Sw of interest can be obtained from the literature (Schwarzenbach et al., 1993; Fetter,
1993; Broholm et al., 1992; Schwille, 1988). Chapter 8 provides additional information
regarding the assumptions and limitations involved in using correlation equations like
364
Chap. 12
Eqs.(12.4) and (12.5). R values for the calculations discussed here were obtained using
Eqs.(12.3)-(12.5).

With the above conceptual model in mind for the disappearance of a DNAPL phase in a
fractured porous medium, chemical mass in the medium can exist as: 1) DNAPL phase
in the fractures; and 2) dissolved and sorbed chemical mass in the matrix blocks between
the fractures. The relative storage capacities of these two regions in a given fractured
porous medium will determine the potential for complete DNAPL phase disappearance
in that medium. The fracture and matrix storage capacities for a given DNAPL will
depend on the DNAPL density, aqueous phase solubility, and the partition coefficient to
the matrix solids. Since the maximum possible aqueous concentration is given by the
solubility, then Mm , the maximum chemical mass storage capacity in the matrix per unit
volume of fractured porous medium (dissolved plus sorbed), is given by
(12.6)
The chemical mass storage capacity in the fractures per unit volume of fractured medium
M f (as DNAPL), is given by
(12.7)
where cPf is the porosity due to the fractures, and p is the density of the DNAPL. When
M m :::: M], then the potential exists for all of the DNAPL mass in the fractures to dissolve
and diffuse into the surrounding matrix material.
To illustrate the above principles for a static case, an example of relative contaminant mass storage capacities is now presented using the fracture characteristics of
a surficial clay till deposit in southwestern Ontario, Canada. Field-measured values for
fracture apertures and fracture spacings at the site were determined by McKay (1991)
and McKay et at. (1993a). We will assume a representative fracture aperture of 15 /Lm
(based on detailed hydraulic data, the fracture apertures were found to be log-normally
distributed between 1 and 43 /Lm). We will also assume a mean fracture spacing of
0.06 m over the depth of visible fractures (6 m). These values yield a calculated fracture porosity of 10- 3.6 . This is in the upper range of values calculated by McKay et al.
(1993a). Mass storage capacity calculations were performed for three common chlorinated solvents assuming matrix parameters measured for this clay till (Johnson et al.,
1989). In a bulk volume of the fractured clay with dimensions of 10 m x 10 m and 6 m
deep, the total fracture void volume equals 0.145 m', and the matrix void volume equals
222 m3 (cPm = 0.37). Results for this example, given in Table 12.3, show that M m > Mf
for all three solvents.
Using the same fracture network, calculations can also be performed for a typical
sedimentary rock. We now let cPm = 0.10, and foe = 0.001. Although the matrix storage
capacity is reduced for this case due to the lower values of cPm and foe, we still have
that M m > M, for DCM and TCE (Table 12.3), again indicating the potential for the
complete disappearance of these two DNAPL liquids from such fractures.
Sec. 12.2
365
TABLE 12.3 Contaminant Mass Storage Capacities in Fractures and in the Matrix for Natural
Fractured Systems. (The fracture network characteristics are based on concepts presented by
McKay (1991 ).)
Geologic Medium
Type-Clay
Type-Sedimentary Rock
Chemical
DNAPL Mass
in Fracture
kg
Dissolved
Mass in Matrix
kg
Sorbed Mass
in Matrix
kg
Total Mass in
Matrix
kg
DCM
TCE
PCE
DCM
TCE
PCE
193
212
235
193
212
235
4,440
311
53
1,200
84
14
1,064
1,067
528
228
229
113
5,504
1,378
582
1,428
313
128
Notes: Clay properties after Johnson et al., 1989:

</J = 0.37 Ph = 1.6 g/cm'
foe = 0.01.
Shale/Sandstone Properties after Feenstra et al. (1984) and Barone (1989):
</J = 0.10 Ph = 2.4 g/cm'
foe = 0.001
Chemical properties from Schwille (1988);
Aqueous solubility for TCE and PCE from Broholm et al. (1992).
An analytical solution for l-D diffusion of chemical mass from a fracture surface into the
pore water of the matrix was used to calculate DNAPL disappearance times for singlecomponent liquids. Figure 12.4, which is the conceptual model corresponding to the
analytical solution, involves a single, parallel-plate fracture containing stagnant DNAPL
as a non-wetting liquid. The DNAPL is assumed to fill all of the fracture aperture except
for a thin film of water (as the wetting phase) that is assumed to be present along the
fracture walls.
The chemical mass flux (MIL2_T) from the fracture surface into the porous matrix
is described by the l-D form of Fick's First Law, Eq.(l2.l). The flux is proportional to
<Pm, to De. as well as to the negative of the concentration gradient of the contaminant in
the aqueous phase (-acw/ax). It is the concentration gradient that is the driving force
for mass transfer. Combining Fick's First Law with the equation for conservation of mass
yields Fick 's Second Law,
De a2cw
Ii ax
acw
at
(12.8)
where the retardation factor R accounts for partitioning to the solid matrix materials. The
analytical solution for diffusive flux at the fracture surface (x = 0) at any time t is
iDeO, t) = <pSw
JRD
xt
(12.9)
where it is assumed that the porous medium is of infinite extent in the direction of
diffusion, with constant cPm, De, R, and that the concentration at the fracture surface
boundary (i.e., at x = 0) is constant and equal to the aqueous solubility of the chemical
366
Chap. 12
(b)
)(.
(c )
x+
Figure 12.4 a. Physical model used for I-D diffusion calculations for DNAPL-phase
mass loss from a single, parallel-plate fracture with a constant aperture equal to 2b. The
initial conditions are illustrated with the concentration equal to the aqueous solubility
limit at the fracture surface and zero concentration in the matrix. b. Schematic of future
conditions and concentration profiles at various time periods II < 12 < I D. c. Conversion
of dissolved and sorbed mass in matrix into an equivalent thickness of DNAPL film in
a smooth parallel-plate fracture.
(C w (0, t) = Sw). The initial aqueous phase concentration in the matrix pore water is
assumed to be zero (Cw(x,O) = 0).
Figure 12.4.a illustrates the initial conditions with the concentration at the fracture
surface established at the aqueous solubility, and zero concentration in the porous matrix.
At t > 0, diffusion into the porous matrix results in concentration vs. distance profiles
Sec. 12.2
367
of the type shown in Figure 12.4.b. As chemical mass penetrates further and further
into the matrix, the concentration gradient (and therefore the driving force for diffusion)
decreases with time. According to Eq.(12.9), the mass flux at the fracture surface iDeO, t)
is inversely proportional to the square root of time. Analytical solutions for a variety of
other initial and boundary conditions can be obtained from the literature (see Crank,
1975, and Carslaw and Jaeger, 1959).
Eq .(12.9) can be integrated over the time period of interest to provide the total
mass diffused into the matrix per unit area of fracture face (Mt ) . For the parallel-plate
fracture case illustrated in Figure 12.4, the solution is
Mt
= 4> Sw
4
ft/RDet
(12.10)
Eq.(12.10) accounts for diffusion occurring in both the positive and negative directions
(diffusion from both fracture surfaces). The time that elapses between the first arrival
of the DNAPL in the fracture and the time required for complete disappearance of the
DNAPL is the disappearance time t o- An expression for to can be obtained directly
from Eq.(12.1O). In particular, when t = to, then M, will correspond to all of the DNAPL
originally in the fracture, i.e., it will correspond to the product of the void volume of the
fracture with the density of the DNAPL, all divided by the surface area in the fracture.
At that point in time, M, can then be expressed in terms of the parallel plate fracture
aperture 2b yielding
to =
rrp2
16S~4>?;,DeR
(2b)2
(12.11)
Figure 12.4.c thus illustrates the conversion of the mass diffused into the matrix at t D
into an equivalent DNAPL thickness (2b') for a single parallel-plate fracture.
As indicated by Eq.(12.1O), M, is directly proportional to 4>m and Sw, and is proportional to the square root of R, De' and t. However, an examination of the sensitivity
of M, to the various characteristics of a porous medium/chemical combination is made
difficult by the fact that a particular property can affect another parameter. For example, Sw affects R (see Eqs.(12.4) and (12.5. Similarly, 4>m affects both T and R (see
Eqs.(12.2) and (12.3. To de-couple the secondary effects of Sw and 4>m on R, we will
first consider the mass loss to a porous medium when R = 1 (no sorption). This case is
conservative in the sense that when R > I, the rate of mass loss is faster and therefore
t o values will be smaller. Other more complicated cases will be considered later. The
range of fracture apertures used in the calculations vary from 0.01 to 1000 /Lm, which is
consistent with fracture apertures easily invaded with reasonable heights of continuous
DNAPL phase (see Kueper and McWhorter (1991), and also Chapter II).
Figures 12.5.a and 12.5.b show, for selected times, the equivalent fracture aperture
2b' from which various chlorinated solvents would disappear as a function of Sw in clay
and sandstone/shale when R = 1. For example, in a clay (Figure 12.5.a), TCE, with a
solubility of 1,400 mg/L, would disappear from a fracture with an aperture of 13 fLm in
10 days. The maximum time considered in the figure is 50 years since this would correspond to the period during which solvents have been widely used (i.e., since the 1940's).
368
(a)
1pOO
(b)
a.
e
::l
.... 1
a.
::l
1:
8-
-c
u,
u,
C
II>
C
II>
10
iii
::l
.~
::l
c:r
WI
~I
I
I
I
I
I
I
I
l!1
100
1:
all
ul
Chap. 12
I
I
c:r
1
10
100
1,000
Solubility (mg/L)
lOPOO
1~
1
10
100
iooo
10pD0
100p00
Solubility (mg/l.)
Figure 12.5 Amounts of DNAPL-phase loss expressed in terms of equivalent fracture aperture as
a function of DNAPL solubility in water for selected times. No sorption (R = I). a. Type clay.
b. Type sandstone/shale. (Fx = fracture.)
DNAPL disappearance for the higher solubility compounds is rapid. For example,
in the clay considered here, within 100 days, DCM in a 706 /Lm fracture and TCE
in a 41 /Lm fracture would completely disappear by diffusion into the matrix. Smaller
amounts of PCE and TCB would disappear in the same length of time due to their lower
solubilities. Within 2.7 years in clay, all of the PCE in a 19 /Lm fracture and all the TCB
in a 2 /Lm fracture would diffuse into the matrix. Within 50 years, DNAPL in fractures
with apertures ranging from 9 /Lm (TCB) to 10 mm (DCM) would disappear completely
into the clay matrix.
As noted above, the fracture apertures in the weathered, clay-rich deposits within 1
to 4 meters of ground surface at the site in southwestern Ontario mentioned above ranged
between 1 and 43 /Lm (McKay et a!', 1993a). The same authors provide a summary
of hydraulic apertures for other near-surface fractured, unlithified clay deposits, with
values ranging from 1 to 100 /Lm. Thus, over the range of solubilities typical of singlecomponent DNAPL chemicals, disappearance times presented in Figure l2.5.a are short
(a few days to a few years) for fracture apertures typical of these deposits.
Figure l2.5.b presents results for a sedimentary rock (e.g., a sandstone or a shale)
with a matrix porosity of 10%. As seen in Eq.(l2.1l), the to value for a given fracture
aperture is inversely proportional to </J;;'. With a typical sandstone/shale porosity being
approximately one-third that of a typical clay, and also because De is smaller for the
sedimentary rock, to is a factor of 40 larger for the sandstone/shale than it is for the
clay. Therefore, for a 13 /Lm fracture, TCE would disappear into the clay matrix in
10 days, but would require 400 days to disappear into the sandstone/shale matrix. Fieldmeasured values for hydraulic fracture apertures in moderate-permeability sedimentary
rocks range from 80-200 /Lm for a sandstone aquifer on Prince Edward Island (Francis
and Gale, 1988) and 120-200 /Lm for a shale aquitard in southern Ontario (Novakowski
and Cherry, 1988) to 120-588 /Lm for a dolomite aquifer in southern Ontario (Reichart,
1992; Feenstra, 1993). Considering these aperture values and Figure 12.5.b, complete
Sec. 12.2
369
disappearance of the common chlorinated solvents would require on the order of a few
years or longer in these materials.
The disappearance times for TCE can also be calculated for a relatively nonporous rock matrix like granite (Figure 12.6). The geologic parameters used for these
calculations are summarized in Table 12.2. Crystalline rocks such as granite are relatively
non-porous with m values that are usually one to three orders of magnitude lower
than for sedimentary deposits (Skagius and Neretnieks, 1986); foe values are near zero.
Figure 12.6 illustrates that long time periods are required for significant amounts of
chlorinated solvents to be lost from fractures in crystalline rocks. After 27 years, only a
3 fLm fracture-equivalent of TCE will be able to diffuse into the granite as compared to
116 fLm and 940 fLm for the sedimentary rock and clay, respectively. Considering that
fracture apertures in rock aquifers can commonly exceed 100-300 fLm, complete DNAPL
disappearance in crystalline rocks will generally require several centuries.
The results of these initial R = I simulations clearly show that diffusion into the
matrix of a fractured, porous geologic material can be a very significant process causing
much or all of the initial DNAPL mass to disappear from the fractures. The results further
show that the rate of disappearance is strongly dependent on Sw and m. Considering
typical materials, chlorinated solvent phase loss from fractures in clay can be very rapid.
In sedimentary rock, it is much slower, but nevertheless significant. In crystalline rock,
such loss is likely to be minor.
0.1 +----r-.,--r-r--rr>TT""<==-r-,r-T'"TTT"T'1'T"-..,--..".-rrTTn----.--.-r-nn'TTi
100
10p00
1
10
1,000
TIme (days)
Figure 12.6 Comparison of TCE mass loss rates for three geologic media expressed in
terms of equivalent DNAPL-phase fracture aperture and time for DNAPL disappearance:
Type clay with R = 5.2, type sandstone/shale with R = 3.2, and type granite R = I.
Parameter values provided in Table 12.2. (Fx = fracture.)
370
Chap. 12
As noted above, the solubility of a DNAPL has a direct effect on its mass loss from
a fracture, and an indirect effect on its mass loss through its influence on the partitioning to
the matrix solids. As shown in Eq.(12.5) and in Table 12.1, low solubility chemicals have
larger K oe values and thus higher preference for the organic carbon present in the matrix.
The effects of sorption on chemical mass loss from a fracture is a result of its effect on
the retardation factor R (see Eqs.(12.3) and (12.4)). For a particular chemical, the value
of foe in the porous matrix determines the value of R, and hence the degree of sorption.
Values of foe for clay-rich deposits typically range between 0.001-0.01 (e.g., see Johnson
et al., 1989; Myrand et al., 1992; Anderson and Pankow, 1986; Rowe and Barone, 1991;
and Russell, 1993); lower foe values are likely for 1ithified sedimentary deposits such as
shales and sandstones, with typical values expected to range from 0.0001 to 0.001.
The sorption effect on mass loss from fractures is demonstrated for four chlorinated solvents in Figures 12.7.a-12.7.d by varying foe for a clay. Because of the low
absolute partitioning for DCM, varied partitioning has a small effect on tD values for
this compound (Figure l2.7.a). Increasing effects are observed for TCE, PCE, and TCB
since within this group, K oe increases (Sw decreases). We also note that sorption has a
greater effect when there is more organic carbon in the porous matrix. For example, for
TCB, after 2.7 years, a clay matrix with foe values of 0.0, 0.001, and 0.01 will result in
the disappearance of 2 p,m, 8 p,m and 24 p,m worth of this DNAPL.
In order that the effects of various chemical and porous media properties on disappearance times could be demonstrated, the calculations presented thus far have been
for a single fracture in an infinite porous medium. However, in a real porous medium,
distances between fractures are finite, causing diffusion to occur in all directions. We
also note that the matrix material between fractures has geometries that are variable from
site to site, and possibly from location to location at a particular site. Nevertheless, in
some cases, the matrix volume can be represented by blocks or columns.
The matrix block size becomes relevant when chemical mass invades the matrix
to the extent that the concentration gradients become reduced by the chemical mass that
is diffusing into the block from the opposite side. Figure 12.8 shows orthogonal sets of
parallel fractures in all three dimensions (XI, Xz, and X3) with generic diffusion profiles
in the matrix for three time periods. At time t], the diffusion profiles essentially do
not overlap, and the rate of mass loss from both fractures can be calculated using the
semi-infinite column approach for the single fracture case presented above. However,
once the profiles begin to overlap to any significant extent (tz and t3), the rate of mass
loss decreases due to decreased concentration gradients. The point in time when these
competing diffusion profiles overlap to a significant extent in the matrix blocks depends
on the distance between the fractures as well as on the sorption properties of the chemical
and the porous medium. The analytical solutions for calculating disappearance times for
cases with finite distances between fractures are presented below in Section 12.2.4.
As discussed in Section 12.2.2, the possibility of complete DNAPL-phase disappearance can be addressed by comparing the maximum (i.e., solubiity-limited) chemical
mass storage capacity in the matrix (Mm ) to the DNAPL storage capacity in the fractures
(Mf), without consideration of the diffusion time. The values of the critical (i.e., minimum) distance between parallel fractures for DNAPL-phase disappearance are presented
Sec. 12.2
371

TCE
OCM (Koc=ll)
(0 )
lQOOO
( b)
~
:J
1::
1::
"
.a.
e:J
"
-c
0.
<t
)(
u..
0.
1,000
100
~"
"
iii
>
'5
C"
w
'5
C"
W
100
10
100
1,000
lime (days)
PCE
(e)
(Koc=92)
lPOO
iooo
lQOOO
100
lime (days)
1,2,4-TCB (Koc=3,200)
(Koc=380)
100
(d)
E
.a.
100
.a.
~
1::
:J
:J
1::
"0.
<t
<t
"
"
~
8-
10
Oi
.~
:J
C"
W
'5
C"
ui
1
10
100
1,000
lime (days)
lQOOO
100
1,000
lime (days)
lQOOO
Figure 12.7 Effects of sorption (R > I) on DNAPL phase disappearance times in a clay deposit
for four chlorinated organic chemicals with different K oc values and a range of fractional organic
carbon contents (foe = 0.00 to 0.01): a. methylene chloride (DCM). b. trichloroethylene (TCE).
c. perchloroethylene (PCE). d. I,2,4-trichlorobenzene (TCB). (Fx = fracture.)
in Figures 12.9.a through 12.9.d for foe = 0.001 as a function of <Pm and fracture aperture. The results for the four chemicals in a 100 p.m fracture in a 10% porosity matrix
(representative of a porous sedimentary rock) show critical distances of < 0.0 1, 0.1, 0.7
and 5 m for DCM, TCE, PCE and TCB, respectively. These results can be useful when
interpreting conditions at actual field sites, e.g., when evaluating the possible presence
of DNAPL-phase chemicals in fractures whose apertures have been estimated based on
bulk hydraulic conductivities and fracture spacing.
An analytical solution for the 3-D diffusion of chemical mass from a fracture into pore
water in matrix blocks represented as rectangular, parallelepipeds can be used to calculate
372
2b
-j
Chap. 12
---1( f--(------1'( ---(-
X1 = L1
DISTANCE
Figure 12.8 Schematic diagram of orthogonal sets of parallel fractures in three dimensions with finite dimension matrix cubes. Generic concentration vs. distance profiles for
three time periods illustrating the effect of competing diffusion profiles at time steps 12
and 13.
disappearance times for single-component DNAPLs. Figure 12.8 shows the conceptual
model for the most general case wherein a matrix block that is bounded by fractures on
all sides has finite lengths in each principal direction (XI = L I, X2 = L 2 , X3 = L 3 ) . In
this 3-D case, Fick's Second Law for a sorbing solute is
(12.12)
Constant values for De. R, and <Pm will be assumed. Parker and McWhorter (1994)
present the details regarding the development of the analytical solution to Eq.(12.l2) for
the following initial and boundary conditions: 1) the concentration at the fracture surfaces
is constant at the aqueous phase solubility Sw (i.e., e w = Swat Xi = 0, and Xi = L, or
Sec. 12.2
373
TCE Koc=92 foc=O.1%
DCM Koc=11 foc=O.1%

(0 )
10
10
(b)
Cl
Cl
.~
.~
c:
c.
C.
(/)
(/)
!'!
!'!
u..
0;
"
Sao
u..
"'rn
0.1
0;
0.1
:;:;
i.'5
''''0
0.01
0.01
0.1
TCB Koc=3200 foc=O.1%
PCE Koc=380 foc=O.1 %

10
(c)
(d)
Cl
Cl
'0
"c.
c.
(/)
(fJ
!'!
"
~
0;
10
c:
.~
u..
0.1
Matrix Porosity
Matrix Porosity
!'!
u..
0.1
0;
0.1
~
o
~
0.1
0.1
Matrix Porosity
Matrix Porosity
Figure 12.9 The critical fracture spacing required for complete DNAPL phase disappearance in an
equivalent fracture aperture calculated as a function of the matrix storage capacity with fo," = 0.001
for four compounds: a. methylene chloride (DCM). b. trichloroethylene (TCE). c. perchloroethylene
(PCE). d. 1,2,4-trichlorobenzene (TCB). The critical fracture spacing is the minimum spacing that
allows complete disappearance and results in an aqueous concentration in the fracture and in the
matrix equal to the aqueous solubility limit.
= 1, 2, and 3 throughout the time period of interest); and 2) initially, the concentration
in the matrix pore water is zero (i.e., Cw(x\, X2, X3) = 0).
The solution for calculating the total mass diffused at any selected time (M t ) is
(12.13)
where
(12.14)
374
Chap. 12
for i = 1, 2, 3, representing the three principle directions. P is described as the "part

remaining" by Glover (1974) for the groundwater drainage problem, and represents the
fractional part of the matrix storage capacity remaining at time t for the diffusion problem
presented here. P is a function of the dimensionless time variable Det/ R L 2 . The quantity
Mft is the mass that remains as DNAPL in the fracture at time t. Mft equals the initial
mass in the fracture minus the amount of chemical mass that has diffused into the fracture
so that
(12.15)
or
(12.16)
Eq.(12.l6) expresses the mass remaining in the fracture at a particular time as a fraction
of the original amount in the fractures at t = 0, and can be used for any idealized
rectangular parallelepiped matrix block size.
The results of calculations based on the above equations are presented for two
specific cases in Figure 12.10: 1) equally-spaced parallel fractures in l-D (Figure l2.1O.a)
so that L 2 = L3 = 00, and therefore P2 = P3 = 1; and 2) 3-D matrix block cubes where
Figure 12.10 Block diagram illustrating idealized fractures networks. a. Parallel fractures in one
principal direction (l-D finite distance case). b. Orthogonal sets of parallel fractures in all three
principal directions (3-D matrix cube case).
Sec. 12.2
375
the fracture sets are orthogonal and the distances between parallel fractures in each
dimension are equal (Figure 12.IO.b) so that L] = L 2 = L 3 and therefore PI = P2 = P3.
The ratio of chemical mass remaining in the fracture at any time relative to the initial
mass in the fractures equals Mft/fpLIL2L3, and is a function of the mass storage
capacity ratio and the dimensionless time variable (Det / R L 2), both of which depend on
the properties of the porous medium and the chemical of interest. As noted above in
Section 12.2.2, for complete DNAPL disappearance in the fractures to occur, we require
that M m :::: M J . Figures 12.11.a and 12.11.b show that the fraction of DNAPL remaining
in the fractures decreases logarithmically toward zero as the dimensionless time increases
for the I-D and 3-D cube cases, respectively.
It is evident from Figures 12.ll.a and 12.11.b that the dimensionless disappearance
time (Detv/ R L 2) decreases as the mass storage capacity ratio increases. When M m/ M J =
1, the curve becomes asymptotic to the x-axis, and the dimensionless disappearance time
approaches infinity. Comparing the results for the I-D finite distance case to the 3-D
cube case, for equal values of Mm/MJ , the rate of mass removal from the fractures for
the 3-D cube case is greater due to its larger surface area for diffusion than for the 1-D
finite distance case. This results in shorter DNAPL phase disappearance times for the 3-D
case for specific chemical and porous media conditions. For example, for M m / M J = 4,
the dimensionless disappearance time for the I-D case is "-'0.01 compared to "-'0.001 for
the 3-D case.
Disappearance times can be calculated using the dimensionless variables by substituting actual parameter values for the chemical and geologic properties. For TCE and PCE
in a fractured clay and in a sedimentary rock, the chemical and porous media parameters
may be taken to be the same as in Tables 12.1 and 12.2. For a fracture porosity of 10-4,
disappearance times for the two solvents in a 1-D parallel fracture system and in a 3-D
orthogonal fracture network are presented for the two porous media types in Table 12.4.
Since M m / M J > 1 in all four combinations, the time periods for DNAPL disappearance
from the fractures are finite in every case. For any specified fracture porosity, numerous
combinations of fracture aperture and fracture spacing are possible. Three examples are
used for each of the four cases to show the effect of fracture aperture on rates of DNAPL
phase disappearance. Smaller, more closely-spaced fractures result in larger surface to
volume ratios for the DNAPL, enhancing the dissolution rate. This effect is also observable in disappearance times for the I-D and 3-D cube cases where for constant fracture
porosity and fracture spacings, the fracture apertures differ between the two systems. For
a I-D parallel fracture network, the disappearance times are rapid for TCE and PCE in
a clay matrix with a fracture porosity of 10-4. For TCE and PCE, these disappearance
times range from a few minutes to several weeks, increasing to five years for widely
spaced, larger aperture fractures. Disappearance times in fractured sedimentary rocks are
longer, and range up to several years to decades.
12.2.5 Limitations
The results presented thus far are based on a number of simplifying assumptions, including single-component DNAPLs, and single idealized fractures or simple fracture
376

I-D FINITE DISTANCE CASE
-----
(a) 1
0.9
~e0.8
u..
.5 0.7
---
bll
:~ 0.6
5 0.5
.....-
10
...J
Chap. 12
0.4
Q
~
0.3
s 0.2
'.:;:l
~ 0.1
0.0001
0.001
0.01
0.1
Dimensionless Time
-----
3-D CUBE CASE

(b) 1
0.9
1;l 0.8
~
.50.7
eo
-e>-
:~ 0.6
---
5 0.5
10
...l
0.4
5 0.3
~
''8
0 .2
0.1
0.0001
0.001
0.01
0.1
Dimensionless Time
Figure 12.11 Fraction of DNAPL mass remaining in the fractures vs. dimensionless
time as a function of various chemical mass storage capacity ratios (matrix/fracture).
a. 1-D finite distance case. b. 3-D matrix cube case.
networks. The effects of changing various chemical and porous media properties on the
disappearance times are easily interpreted for such cases. The values of the chemical
properties used in the calculations pertain to 20-25C. This is warmer than the subsurface temperatures at most North American sites. Although the magnitude and direction
TABLE 12.4 DNAPL Phase Disappearance Times for 1-D and 3-D Finite Dimension Matrix Block Cases. (Porous media and chemical
parameter values are provided in Tables 12.1 and 12.2. Both porous media have fracture porosities of 1O- 4 .j
Type-Clay
Mass Storage
Ratio
Chemical
TCE
MatrixlFracture
17.7
Matrix
Block
Dimension
L (em)
to
3-D Cubes
Mass Storage
Ratio
(days)
MatrixlFracture
0.01
0.002
10
1.1
0.15
9.5
to
I-D Finite
Distance
(days)
100
PCE
Type-Sedimentary Rock
10
100
113
0.15
15.0
1500
1
5.3
15
180
2.8
to
I-D Finite
Distance
(days)
0.44
10
44
100
4,400
0.02
1.8
Matrix
Block
Dimension
L (em)
5.8
to
3-D Cubes
(days)
0.06
5.6
560
0.8
10
580
85
100
58,000
8,500
Note: For a constant fracture porosity, varying the matrix block dimensions results in changes to the fracture aperture for both cases. For fracture
spacings of 1, 10 and 100 em, the fracture apertures are I, 10 and 100 J1.m respectively for the I-D finite distance block case, and 0.33, 3.3, and
33 J1.m respectively for the 3-D cube case.
IN
'I
'I
378
Chap. 12
of the necessary temperature corrections depend on the chemical and the system parameters, they are small in comparison to the order of magnitude approximations utilized
here for the pertinent geologic parameters (e.g., fracture spacing, fracture aperture, matrix tortuosity, and matrix sorption). Also, assumptions have been made by representing
the fractures as smooth parallel-plates. In addition, we have ignored situations in which
advection of dissolved contaminant occurs in the water in the fractures, and situations in
which the DNAPL in the fractures is replenished from above by downward flow. When
relaxed, some of these assumptions have opposite effects on calculated disappearance
times.
Thus far, only single-component DNAPLs have been considered. At actual field
sites, however, DNAPLs composed of two or more chemicals are common, e.g., mixtures
of TCE, PCE, and other solvents. Evaluating matrix diffusion from such multi-component
liquids requires consideration of changes in the chemical mole fractions in the DNAPL
phase with time as individual components leave by diffusion at different rates due to
different aqueous solubilities (thus different concentration gradients), and different diffusion coefficients. The effective solubilities for each component at the fracture surface
thus change continuously as the composition of the DNAPL changes with time. Also,
physical properties of the DNAPL (e.g., DNAPL density) change due to the changes in
composition.
12.3 IMPLICATIONS FOR DNAPL DISAPPEARANCE FROM FRACTURES
Issues concerning real fractured porous media contaminated with DNAPLs include matters ranging from the interpretation of site conditions to the selection and design of
remediation systems. The actual presence of a DNAPL phase in monitoring wells or in
boreholes in fractured porous media occurs only seldomly, and when it does occur, it
often does so only once (Feenstra and Cherry, 1988; Mercer and Cohen, 1990; U.S.EPA,
1992). This is often taken as evidence of the difficulty in characterizing the fractured media sufficiently well to identify the discrete DNAPL transport pathways. At many sites,
an alternative explanation is that much of the DNAPL that may have once filled the
fractures has been lost by diffusion into the matrix. Indeed, the calculations presented in
this chapter indicate that in many fractured media, a DNAPL solvent phase may persist
in the fractures only for a short time relative to the typical age of a DNAPL site.
For the common chlorinated solvents, the concentrations in groundwater obtained
from monitoring or extraction wells located within or close to DNAPL source zones are
seldom at or approaching the aqueous solubility limit(s). This is commonly attributed
to dilution caused when wells are purged and sampled; it is often assumed that some
fractures contain groundwater that are at or near the solubility limit, and other fractures
that are less connected to the DNAPL source have much lower concentrations. Thus
upon sampling, a mixed or diluted value is obtained. However, based on the conceptual
model of DNAPL disappearance discussed above, we can envision situations where
DNAPL at one time occupied many or all of the fractures in the zone of a monitoring
Sec. 12.3
379
well, but has disappeared due to diffusive mass loss to the matrix, with the aqueous phase
concentrations then declining below the solubility limit. In such a situation, concentrations
in a monitoring well will be below the solubility limit regardless of where the well intake
is situated. Moreover, if the monitoring well is sampled after the point in time when
the DNAPL originally entered the monitoring zone, the concentrations in the well could
decline progressively with time. This could give the erroneous impression that the overall
system is in the process of being "cleaned up" due to flushing by the groundwater flow.

The flow of NAPLs in rough-walled fractures has been evaluated using 2-D numerical models (Pruess and Tsang, 1990; Kueper and McWhorter, 1991; and Murphy and
Thomson, 1993), and using laboratory models (Schwille, 1988). The model simulations
indicate that the DNAPL travels in single, rough-walled fractures at rates ranging from
minutes to several hours per one meter length of fracture. Kueper and McWhorter (1991)
(see also Chapter 11) have examined the influence of various parameters on DNAPL entry into and flow rates through water-saturated fractures. The parameters include fracture
aperture and aperture distribution, groundwater gradient, fracture dip from horizontal,
DNAPL pool height above the fracture, and DNAPL fluid properties. Given the numerous site specific influences on DNAPL migration through real fractures and the current
infeasibility of validating these above referenced models in relation to real fractures, a
comparison of DNAPL disappearance times with simulated DNAPL travel times is not
yet possible.
Although diffusive mass loss to matrix materials was not considered in the studies
discussed in the preceding paragraph, diffusive mass loss into the porous matrix causes the
region available for mass distribution to be significantly larger than the fracture porosity.
For matrix diffusion to have an effect on DNAPL flow rates and distribution does not
require complete DNAPL phase disappearance from fractures, but rather only sufficient
mass loss to have a measureable affect on the DNAPL driving forces and permeability
in the fracture by changing fluid pressures and resultant fluid saturations.
At the fracture-network scale, the rate of advance of a DNAPL front will be slowed
by DNAPL mass loss to the matrix. Assuming a continuous DNAPL phase, the maximum
capillary pressure of a DNAPL at its advancing front in the fracture network is a function
of the total depth of penetration. Thus, if DNAPL mass in the fractures is continually
lost into the matrix, the ability of the DNAPL to invade smaller and smaller fractures
with depth will be reduced. The driving forces for penetration of DNAPL in the fracture
network will be reduced further if DNAPL mass loss from the fractures results in a disconnection of the DNAPL phase in the fractures (Figure 12.3.b). For a multi-component
DNAPL, the driving forces are also affected by changes in the DNAPL composition over
time (Parker et al., 1994). Depending on the initial chemical composition, the DNAPL
may become heavier or lighter as dissolution proceeds.
For many old industrial waste disposal sites on fractured media, there is concern
that free-phase DNAPL continues to flow, causing expansion of the DNAPL zone to
locations increasingly distant from the site. There is also a concern of re-mobilizing
380
Chap. 12
free DNAPL during drilling or groundwater pumping within source zones. It is often
not possible to determine conclusively from site monitoring or mathematical modeling
whether DNAPL flow has ceased. However, the probability that DNAPL flow has ceased
is highest at sites where the potential for diffusion into the matrix is greatest; likelihood
that DNAPL can be remobilized within source zones is also reduced by such conditions.
Ultimately, matrix diffusion affects the distribution of DNAPL in the subsurface
by reducing the portion of the fracture network invaded before immobilization of the
DNAPL. Thus, the "source zone" is smaller than it would otherwise be, and a portion
or all of the contaminant mass then exists in the dissolved and sorbed states within the
matrix blocks between the fractures that once contained the DNAPL.

Chapter 2 indicated that sites where DNAPL contamination of groundwater is present
are generally characterized by two regions of contamination, the source zone and the
plume. In the simplest context, the source zone is the domain where DNAPL exists as
residual or as pools, and the plume is the domain where contamination exists only in the
dissolved and sorbed phases. The plume is caused by dissolution and transport of dissolved contaminant mass away from the immiscible liquid in the source zone. This simple
conceptualization of the source zone and its plume is applicable to sites where DNAPL
persists with minimal mass loss in porous media as is typical of DNAPLs in sand or
gravel, or in fractured non-porous rock such as granite. At sites where DNAPL has entered
fractured clay or fractured sandstone, the above discussion has illustrated that the zone
initially occupied by the DNAPL can be diminished in size due to diffusion to the matrix.
We now know that in some cases, the DNAPL originally present in fractures can
disappear completely due to dissolution and transfer to the matrix. In such a case, the
zone that was initially the DNAPL source zone will still contain the contaminant mass,
but the contaminant is now all in the dissolved and sorbed phases. While these are the
same two phase types that contain the portion of the contaminant that is in the plume,
they still act as a source zone inasmuch as chemical mass will be released continuously
from the matrix due to the reversed concentration gradients that develop once lower
concentrations occur in the fractures of the source zone. We also note that the dissolved
concentrations that will ultimately occur in the fractures in this zone after disappearance
of the immiscible phase may be small, making it very difficult to determine that this
zone was formerly a zone of DNAPL occurrence.
An increasingly-common approach for the remediation of DNAPL sites is the
placement of vertical cut-off wall enclosures around the source zones so that the plumes
are prevented from causing continued contamination. Once a cut-off wall is in place,
the plume outside the enclosure can be controlled and ultimately removed by pumpand-treat. However, at fractured-porous-media sites where appreciable disappearance of
the DNAPL has occurred, and also where it is difficult to determine where DNAPL
still exists, the potential success of attempts at source-zone isolation can be difficult to
determine.
Sec. 12.3
381
The selection of an appropriate conceptual model for interpreting site data is essential
for the effective design of remedial actions. The effective evaluation of the likelihood
of DNAPL persistence in fractured porous media requires a discrete-fracture approach
and certain types of field and laboratory data that are rarely obtained. Nevertheless, a
basis for evaluating the relative importance of matrix diffusion can be obtained from calculations for DNAPL disappearance times of the type presented in Sections 12.2.3 and
12.2.4 using representative data for the geologic formation, the contaminant(s) present,
and an estimate of the elapsed time since the initial contamination by DNAPL. Unfortunately, reliable information on fracture porosity and aperture is exceptionally difficult to
obtain, making it difficult to accurately predict disappearance times. However, methods
for estimating fracture apertures and fracture porosities exist using various combinations
of site data. For example, fracture patterns and frequency with depth can be estimated
from core logs, borehole geophysics, and outcrop mapping. Hydraulic data obtained
from aquifer tests and well tests can be combined with information regarding fracture
frequency to calculate equivalent hydraulic fracture apertures or fracture porosities. At
some sites, additional site-specific data should be obtained, including the collection of
site-specific samples to determine matrix pore-water concentrations, sorption characteristics, and effective diffusion coefficients. Accounting for the mass that has diffused
into the matrix will, in many fractured porous media cases, change the interpretation of
the contaminant mass distribution, the processes controlling mass removal, and cleanup
times.
12.3.5 Remediation
The restoration of fractured porous media contaminated by DNAPL compounds is a

task which is exceptionally difficult, and in many cases, even futile. The difficulty
in remediating such sites derives from three main factors: I) complex fracture networks cause the initial distribution of DNAPL mass to be difficult to predict or locate;
2) dead-end fractures or fractures not well-connected to active groundwater flushing impede cleanup by pump-and-treat systems; and 3) the existence of much or nearly all
the contaminant mass in the relatively immobile pore water of the matrix as a result
of matrix diffusion greatly increases the time scales required for cleanup. New technologies have been put forth for flushing DNAPLs out of fracture networks such as
flushing with surfactants or alcohols with the goal of greatly enhancing the DNAPL
solubility and hence increasing the mass removal rates. To be effective, these technologies depend on the presence of and access to DNAPL in the fracture network.
They are therefore unsuitable for sites at which the DNAPL has dissolved and diffused into the matrix. Technologies having the best prospects for restoration of fractured
porous media are those that rapidly extract dissolved and sorbed contaminant mass from
the matrix, or that destroy the contaminant mass by reactions that occur within the
matrix.
382
Chap. 12
12.4 DISSOLVED-PHASE TRANSPORT IN FRACTURED POROUS MEDIA
As with the flow of DNAPLs in fractures, groundwater flow in fractured porous media occurs predominantly in the fractures. If the fracture apertures are large, then the
groundwater velocity and down-gradient transport of contaminants can be rapid. Dissolved contaminants are subject to diffusion into and out of the matrix.
Plumes of dissolved contaminants can result from groundwater infiltrating through
zones of dissolved or DNAPL-phase contamination above the water table, from the dissolution of DNAPLs in source zones below the water table, or from dissolved chemical
mass residing in the matrix blocks between the fractures. As with the dissolution and
diffusion of DNAPL material in fractures, the influence of matrix diffusion on the overall
solute transport in fractures depends upon the combined chemical and geologic characteristics of the system of interest.
For illustrative purposes, two hypothetical scenarios of solute transport in fractured media have been created using a 2-D numerical model developed by Sudicky and
McLaren (1992). This model simulates steady-state groundwater flow and transient contaminant transport in discrete parallel-plate fractures within a porous medium. In the
first example (Harrison et al. 1992), the numerical model was used in a cross-section
analysis to assess the importance of vertical planar fractures on solute transport downward through a 15-m thick aquitard overlying a sand aquifer. The second scenario is a
plan-view analysis to evaluate two cases of plume development in porous and relatively
non-porous fractured rock.
12.4.1 Example 1: Solute Transport Through a Vertically
Fractured Clay Aquitard
Figure 12.12 shows the conceptual setting adopted by Harrison et al. (1992). The fractured clayey aquitard with a shallow water table overlies a sand aquifer where flow is
horizontal. Flow is downward through the aquitard. Contaminants emanate from a nearsurface waste zone in the aquitard. The waste zone is generic in the sense that it could
represent buried waste, a lagoon, industrial leakage, or some other type of source for
which dissolved contaminants enter the groundwater flow system from a localized source
zone near the water table. The aquitard provides the aquifer with a degree of protection
from contamination. This is a common scenario in many regions. The simulation of the
contaminant migration though the aquitard into the underlying aquifer provides insight
into how a small number of narrow vertical fractures in an aquitard can allow significant
leakage of dissolved contaminants through the aquitard.
The simulations for this case were performed for the hydrogeologic cross section
shown in Figure l2.13.a. The water table is at the top of the aquitard. The waste zone is
5 m deep and 50 m long. The top of the waste zone is 1.5 m below ground surface. The
aquitard has a network of fractures, where nearly all fractures are vertical, with decreasing
frequency at depth. Few fractures penetrate the entire thickness of the aquitard. Three
vertical fractures extend from the bottom of the waste zone all the way through the
aquitard to the aquifer. The fracture frequency and the properties of the clay matrix
Sec. 12.4
383
Waste
Facility
-e
"
o
..
'"
..
Figure 12.12 Schematic diagram of vertically fractured clay aquitard with contaminant source. Solute transport through the fractures creates a contaminant plume in the
underlying aquifer. (Adapted from Harrison et at. (1992).)
are assumed to be similar to what has been found by Ruland et al. (1991), McKay
(1991), McKay et al. (I 993a and 1993c). The waste zone and the overlying cap material
are assumed to be devoid of fractures. The simulations were performed using chemical
properties typical of the chlorinated solvents, particularly TeE. Table 12.5 provides the
hydraulic conditions and porous media properties used as input parameters.
TABLE 12.5 Numerical Model Input Parameters for Fractured Clay Aquitard Example: 2-D
Vertical Flow and Solute Transport (after Harrison et el., 1992).
Parameter
Aquitard
Aquifer
Waste Zone
Cap
Matrix porosity (m)

Hydraulic conductivity (K), cmfs
Longitudinal dispersivity (al), m
Transverse dispersivity (at), m
Tortuosity (T)
Effective diffusion coefficient (De), cm 2fs
Distribution coefficient! (Kd), mUg
Retardation factor/ (R)
Fraction of organic carbon (foe)
0.50
1.0 X 10- 8
0.0
0.0
0.2
1.49 X 10- 6
1.26
4.2
0.01
0.35
1.0 X 10- 2
1.0
0.02
0.7
5.22 X 10- 6
0.025
1.12
0.0002
0.40
1.0 X 10- 6
1.0
0.02
0.7
5.22 X 10-6
1.26
5.5
0.01
0.50
1.0 X 10- 8
0.0
0.0
0.2
1.49 X 10- 6
1.26
4.2
0.01
locKo,,: K oc
2R = I
+ PbKd/t/>m
Kd
126 mUg
The boundary conditions for flow in the calculations consist of specified hydraulic
head values along the surface which decline linearly from 26.5 m on the upper left corner
to 25.0 m in the upper right corner. This results in a horizontal gradient of 1.67 x 10- 3
in the aquifer. These heads were also fixed to provide an average downward hydraulic
gradient of 0.95 through the aquitard. Initially, the contaminant mass is assumed to exist
entirely in the waste zone at a relative aqueous phase concentration of Cw/C o = 1. In
384
25.01~~ilIl
25
20.0
20
17.5
17.5
'" 15.0
..
.~
12.5
c;
;;:;'"
15
12.5
.~
>
(b)
22.5
22.5
Chap. 12
10.0
10
>
7.5
5.0
2.5
2.5
0(0
30
60
7.5
5
A qu i fe r
90
120 150 180 210 240 270
Horizontal Distance x (m)
300
30
60
90
120
150
180
210
240
270
:lOb
25,,---------------------,
(c)
22.5
20
17.5
15
c: 12.5
0
10
'"
W
7.5
>
5
2.5
0
,
0
30
60
9'0
120
150
180
210
240
270
300
Figure 12.13a,b,c a. Computer generated 2-D fracture network in vertical cross-section. Numerical
simulations for five scenarios: b. 25 f.Lm fractures/25 years solute transport. c. 25 f.Lm fractures after
50 years.
the following results, the outer isoconcentration line represents a relative concentration
of CUI/CO = 0.01 and the inner concentration contour represents CUI/CO = 0.10. For
the case where the contaminant is TCE and the source concentration equals the aqueous
solubility (1,400 mg/L), the outer contour represents 14 mg/L of TCE, which is four
orders of magnitude above the maximum TCE contaminant level allowed by the U.S. EPA
in public drinking water supplies.
Figures 12.13.b-f show patterns of contaminant migration for five simulations.
Groundwater flow causes downward transport of dissolved contaminants from the waste
zone into the aquifer. Diffusion to the matrix causes the downward rate of movement
of the contaminant front to be greatly retarded relative to groundwater velocity in the
fractures. Greater retardation occurs in the smaller aperture fractures. When the fractures
are assigned aperture values ranging from very small (5 /Lm) to slightly larger (25 /Lm)
Sec. 12.4
25
22.5
25
17.5
..
15
(d)
.: ~', ',,:.
rsJ
'.'
20
385
.........
(e)
22.5
20
'
17.5
..
15
e 12.5
.~
0
10
>
.
;;;
c: 12.5
.~
.
0
.;>
7.5
>
10
0
iii 7.5
IP
o
2.5
30
60
90
do
Horizontal
HIO
2.5
"
1110
210
0
240
270 300
80 90
30
Distance x (rn)
do
liio
180
2io
240
270 300
25,,.-----------------,
(f )
22.5
20
17.5
..
15
12.5
.~
c;
..
>
10
iii 7.5
1)
2.5
Horizontol Distonce x (rn)
Figure 12.13d,e,f d. 5 J-Lm fractures/50 years. e. IO J-Lm fracture/50 years. f. 50 J-Lm fractures/50 years.
(Adapted from Harrison et al. (1992).)
in size, there is a large difference in the rate of contaminant migration. Over a 50 year
period, 5 or 10 !tm fractures cause no significant impact on the aquifer (Figures l2.13.d
and 12.13.e), whereas fractures of 25 !tm or larger cause a large impact on the aquifer
in less than 25 years (Figure l2.13.b). In cases where the fractures have apertures less
than 10 uiu, the contamination front moves downward very slowly not only because
the groundwater average velocity in the fracture is reduced by the square of the fracture
aperture, but also because the mass of contamination in each fracture is small relative to
the surface area of the fracture across which mass transfer by molecular diffusion occurs.
Although the 25 !tm fractures that fully penetrate the aquitard cause severe contamination
of the aquifer, these fractures are nevertheless very small in the context of the hydraulic
conductivity that they impart to the aquitard. For example, the bulk vertical hydraulic
conductivity of the aquitard when the fractures have 25 !tm apertures is on the order of
10- 9 mls.
386
Chap. 12

This example is taken from Sudicky et al. (1993) and Moncur (1992). It uses the same
conceptual model as the previous case, and involves 2-D numerical simulations of a plume
of dissolved contaminants for two contrasting types of fractured rock, a moderatelyporous fractured sandstone and a relatively non-porous fractured granite. Groundwater
flow occurs in a square horizontal flow domain, 200 m by 200 m by 10 m thick. The
flow domain has fractures generated randomly with the constraint of a minimum fracture
spacing of 3 m, and minimum and maximum fracture lengths of 20 to 50 m, respectively
(Figure 12.14). Matrix porosity values of 0.20 and 0.02 and hydraulic conductivities of
10-6 and 10- 9 m/sec were used for the sandstone and granite simulations, respectively.
The hydraulic characteristics were selected so as to be generally representative of these
two different media types. All other hydraulic parameters and system dimensions were
the same for all simulations for both media types, with a constant fracture aperture of
100 /lm, a fracture dispersivity of 0.5 m, a longitudinal dispersivity of 0.10 m, a transverse
dispersivity of 0.01 m, an effective diffusion coefficient of 10-6 cm 2/sec, and no decay.
The resulting fracture porosity calculated for this system is 10- 4 .8 . These model input
values are summarized in Table 12.6.
200
I---
180
160
f-
It
140
100
0
lJl
Cl
t--
I
I
120
Q)
r--
....--.
I
I
80
60
~
I--
40
f-
Source
- t--
LJ
20
I U
f--
'--
20
40
60
I---
I
I
80 100 120 140 160 180 200
Distance (m)
Figure 12.14 Computer-generated horizontal fracture network in two dimensions with
contaminant source area. (After Sudicky et al. (1993) and Moncur (1992).)
Sec. 12.5
387
Conclusions
TABLE 12.6 Numerical Model Input

Parameters for Fractured Rock Example: 2-D
Horizontal Flow and Solute Transport (after
Sudicky et al. 1993, and Moncur, 1992).
Fracture System (both cases)

Hydraulic Gradient
0.01
Fracture Aperture 2b
100j.Lm
Fracture Porosity
10-4 . 8
tPf
10 m
Thickness
No Retardation (R
I)
Sandstone Case
Matrix Porosity
tPm
Matrix Hydraulic Conductivity
0.20
10- 6 cmls
Diffusion Coefficient De
Granite Case:
Matrix Porosity
tPm
Matrix Hydraulic Conductivity
0.02
10- 9 cmls
Diffusion Coefficient De
Groundwater flow from left to right in Figure 12.14 is driven by a hydraulic gradient
of 0.01. The contaminant source is a 20 m by 50 m zone near the left (upgradient) side
of the system, and has an isotropic hydraulic conductivity of 10- 2 em/sec. The source is
considered to be a "slug source" with initial solute concentration Cw/C o = 1, allowing
the source to be depleted with time. The simulations presented in Figure 12.15 show
the effects of matrix diffusion for 20 and 100 year periods. The sandstone has an order
of magnitude greater matrix porosity than does the granite. This increases the effective
diffusion coefficient, and hence the rate of solute mass transfer from the fracture to the
matrix. Thus, the downgradient travel distances of the plume in the fractured granite
cases far exceed the distances of migration for the sandstone cases.
12.5 CONCLUSIONS
Matrix diffusion from DNAPLs in fractures can cause a significant portion of chemical
mass to be transferred to matrix materials with low hydraulic permeabilities. The most
important parameters governing the rate of disappearance of a given DNAPL compound
are its water solubility, the matrix porosity, and the matrix sorption capacity. This becomes evident when comparing disappearance times from single fractures for equivalent
DNAPL film thicknesses for various chemicals, and for clay materials vs. sedimentary
388
(0
( bl 200
200
180
180
160
160
140
140
~ 120
~ 120
OJ
c;
Chap. 12
u
'"
c
100
100
80
80
Cl
Cl
60
60
40
40
20
20
0
2'0
4'0
6'0
8'0 100120140160180200
,
0
Distance (rn)
(d) 200
(c) 200
E
'"
u
co
Distance (rn)
180
180
160
160
140
140
120
~ 120
OJ
100
c
a
80
80
Cl
100
Cl
60
60
40
40
20
20
,
0
0
2'0
4'0
6'0
8'0 100 120 140 160 180 2 0 0
Dis tance (m)
,
0
2'0
4'0
6'0
8'0 100
do 140
160 180200
Dis tance (m)
Figure 12.15 Computer simulations of contaminant plumes after 20 and 100 years of transport
in 100 f1m fractures for: a. granite aquifer after 20 years. b. sandstone aquifer after 20 years.
c. granite aquifer after 100 years. d. sandstone aquifer after 100 years. (After Sudicky et al. (1993)
and Moncur (1992).) The outside contour represents CUI/CO = 0.01 for the 20-year simulations and
C",/C" = 0.1 for the 100-year simulations.
rocks. The pertinent parameters for the chlorinated solvents and fractured media of interest vary over several orders of magnitude. Therefore, the values selected for specific
calculations have a strong influence on the calculated DNAPL disappearance times from
fractures. For typical fractures in unlithified, clay-rich deposits, matrix diffusion can cause
the common chlorinated solvents to completely dissolve and disappear as DNAPLs in
time periods that vary from a few months to years. This is also the case for small to
medium size fractures in sedimentary rocks. Hence, matrix diffusion is an important
Sec. 12.5
Conclusions
389
process in the conceptual model used to describe DNAPL flow and distribution in fractured porous media. Indeed, calculations show that there is considerable potential for
chlorinated solvent DNAPLs to completely disappear from fractures, that is to undergo a
complete phase change from DNAPL in the fractures to the dissolved and sorbed phases
in the matrix. This suggests that at many industrial sites where chlorinated solvents and
other modestly soluble DNAPLs have entered fractured porous media in past decades, it
is likely that little or no DNAPL remains in the fractures.
The concepts discussed above apply to single as well as multi-component DNAPLs,
and also to NAPLs that are lighter than water (i.e., to "LNAPLs"). For larger fracture
porosities where the matrix/fracture mass storage capacity ratio is less than one, complete
disappearance will not occur by diffusion into the matrix alone. However, advection of
solutes away from the DNAPL zone(s) in the fracture network will enhance the potential
for such disappearance. Examples of geological conditions where rapid disappearance
is not expected are fractured rocks with very large fractures (e.g., dissolution channels
in karst formations), and rocks with matrix porosities of less than 1% (e.g., granite and
most other crystalline rocks). In rocks with low matrix porosities, DNAPL disappearance
from fractures of typical aperture could take centuries to millennia. Also, complete disappearance is not expected to occur in situations where flow of the DNAPL (or LNAPL)
replenishes the liquid lost by diffusion to such a degree that the storage capacity in the
matrix is exceeded.
In low-permeability deposits that are not fractured, or have fractures of very small
apertures, matrix diffusion plays a dominant role in determining the migration distances
of dissolved organic solutes. The result is long time frames for contaminant migration,
even over short distances. In cases where contaminants are released to aquitards overlying
aquifers, a high degree of protection is provided to the underlying aquifers.
The presence of small-aperture fractures in otherwise low-permeability deposits
has been found to increase the bulk hydraulic conductivities by one to three orders of
magnitude. Also, given that the average linear groundwater velocity in such fractures can
be large, the travel times for water to underlying aquifers are very short. However, due
to diffusive mass loss of the contaminant from the fracture into the matrix, chlorinated
solvents moving in the fractures will advance only slowly relative to the average linear
groundwater velocity in the fracture network. Chemicals that sorb to the fracture surfaces
and matrix solids advance more slowly than non-sorbing solutes due to increased rate of
mass loss to the matrix. We can conclude that if realistic transport rates of contaminants
to groundwater receptors are to be estimated, then it will be important to determine the
nature and frequency of any fractures present in the media of interest as well as the
properties of the matrix (porosity, sorption capacity, and diffusion characteristics).
DNAPL loss from fractures and contaminant distribution in the matrix has important implications with respect to the interpretation of data from monitoring wells, assessment of ultimate risk to receptors, and selection and design of remedial measures. We
note though that two inherent difficulties exist with interpreting the transport and fate
of NAPLs and dissolved contaminants in fractured porous media: 1) the contaminant
mass in the fractures can be difficult to quantify due to the difficulty in characterizing the fractures; and 2) the contaminant mass in the matrix is rarely measured in site
390
Chap. 12
investigations. Even though detailed determinations of this type are not generally feasible, it should be possible to make significant advances in transport and fate assessments
by acquiring information of this nature in site investigations.
ACKNOWLEDGEMENTS
12.6 REFERENCES
Anderson, M. R. and J. F. Pankow (1986) "A case study of a chemical spill: polychlorinated
biphenyls (PCBs). 3. PCB sorption and retardation in soil underlying the site", Water Resources
Research, 22, 1051-1057.
Barone, F. S., R. K. Rowe, and R. M. Quigley (1990) "Laboratory determination of chloride
diffusion coefficient in an intact shale", Canadian Geotech. J., 27, 177-184.
Bonoli, L. and P. A. Witherspoon (1968) "Diffusion of aromatic and cycloparaffin hydrocarbons
in water from 2 to 60", J. Phys. Chem., 72, 2532-2534.
Broholm, K., J. A. Cherry, and S. Feenstra (1992) "Dissolution of heterogeneously distributed
solvents residuals". In: Proceedings: Subsurface Restoration Conference, 3rd International Conference on Ground Water Quality Research, June 21-24, 1992, Dallas, Texas.
Carslaw, H. S. and J. C. Jaeger (1959) Conduction of Heat in Solids, Oxford University Press, New
York.
Crank, J. (1975) The Mathematics of Diffusion, Second Edition, Clarendon Press, Oxford.
Day, M. J. (1977) Analysis of Movement and Hydrochemistry of Groundwater in the Fractured
Clay and Till Deposits of the Winnipeg Area, Manitoba, M.Sc. Thesis, Dept. of Earth Sciences,
University of Waterloo, Waterloo, Ontario Canada.
Dullien, F. A. L. (1992) Porous Media: Fluid Transport and Pore Structure, 2nd Ed. Academic
Press, Inc., San Diego, California.
Feenstra, S., J. A. Cherry, E. A. Sudicky, and Z. Haq (1984) "Matrix diffusion effects on contaminant migration from an injection well in fractured sandstone", Ground Water, 22, 307-316.
(DNAPL) chemicals". Presented at the International Groundwater Symposium, International
Association of Hydrogeologists, Halifax, Nova Scotia, May 1-4, 1988.
Feenstra, S. (1993) Waterloo Centre for Groundwater Research Short Course Lecture Notes: Dissolved Organic Contaminants in Groundwater, Feb. 15-18 in Chicago, Illinois.
Fetter, C. W. (1993) Contaminant Hydrogeology, Macmillan Publishing Company, New York.
Foster, S. S. D. (1975) "The chalk groundwater tritium anomaly-a possible explanation", J. Hydrology,25, 159-165.
Sec. 12.6
References
391
Francis, R. and J. Gale (1988) "Permeability distribution in a fractured sedimentary aquifer system".
Presented at the International Groundwater Symposium of the International Association of
Hydrogeologists, Canadian National Chapter, Atlantic Region, Halifax, Nova Scotia, May 1-5,
1988.
Germain, D. and E. O. Frind (1989) "Modelling of contaminant migration in fracture networks: Effects of matrix diffusion". In: Proceedings: International Symposium on Contaminant Transport
in Groundwater, Stuttgart, Germany, April.
Glover, R. E. (1974). Transient Ground Water Hydraulics. Unpublished report, Dept. of Civil
Engineering, Colorado State University, Fort Collins, Colo.
Goodall, D. C. and R. M. Quigley (1977) "Pollutant migration from two sanitary landfill sites near
Sarnia, Ontario", Canadian Geotech. J., 14,223-236.
Grisak, G. E. and J. F. Pickens (1980) "Solute transport through fractured media. 1. The effects of
matrix diffusion", Water Resources Research, 16, 719-730.
Grisak, G. E., J. F. Pickens, and 1. A. Cherry (1980) "Solute transport through fractured media:
2. Column study of fractured till", Water Resources Research, 16,731-739.
Grisak, G. E. and 1. F. Pickens (1981) "An analytical solution for solute transport through fractured
media with matrix diffusion", J. Hydrology, 52, 47-57.
Harrison, B., E. A. Sudicky, and J. A. Cherry (1992) "Numerical analysis of solute migration
through fractured clayey deposits into underlying aquifers", Water Resources Research, 28,
515-526.
Hoek, E. and J. W. Bray (1977) Rock Slope Engineering, 2nd Ed., Institution of Mining and
Metallurgy, London.
Johnson, R. L., 1. A. Cherry and J. F. Pankow (1989) "Diffusive contaminant transport in natural clay: A field example and implications for clay-lined waste disposal sites", Environ. Sci.
Technol., 23, 340-349.
Kueper, B. H. and D. B. McWhorter (1991) "The behavior of dense, nonaqueous phase liquids in
Lerman, A. (1979) Geochemical Processes: Water and Sediment Environments, John Wiley & Sons,
Inc. New York.
McKay, L. D. (1991) Groundwater Flow and Contaminant Transport in Fractured Clay Till, Ph.D.
Thesis, Dept. of Earth Sciences, University of Waterloo, Waterloo, Ontario.
McKay, L. D., J. A. Cherry, and R. W. Gillham (1993a) "Field experiments in a fractured clay till:
I. Hydraulic conductivity and fracture aperture", Water Resources Research, 29, 1149-1162.
McKay, L. D., J. A. Cherry, R. C. Bales, M. T. Yahya, and C. P. Gerba (1993b) "A field example
of bacteriophage as tracers of fracture flow", Environ. Sci. Technol., 27, 1075-1079.
McKay, L. D., R. W. Gillham, and J. A. Cherry (1993c) "Field experiments in a fractured clay
till: 2. Solute and colloid transport", Water Resources Research, 29, 3879-3890.
Millington, R. J. and J. P. Quirk (1961) "Permeability of porous solids", Trans. Faraday Soc., 57,
1200-1207.
Moncur, M. C. (1992) Simulation of Contaminant Movement and Pump and Treat Remediation in
Fractured Porous Media, M.Sc. Project, Dept. Earth Sciences, University of Waterloo, Waterloo,
Ontario.
392
Chap. 12
Murphy, J. R. and N. R. Thomson (1993) "Two-phase flow in a variable aperture fracture", Water
Myrand, D., R. W. Gillham, E. A. Sudicky, S. F. O'Hannesin, and R. L. Johnson (1992) "Diffusion of volatile organic compounds in natural clay deposits: Laboratory tests. J. Contaminant
Hydrology, 10,159-177.
Novakowski, K. S. and J. A. Cherry (1988) "Evaluating groundwater velocity in a low-permeability
fractured shale". In: Proceedings: Technology Transfer Conference, Ontario Environment,
Canada, Nov. 28-29, 1988, pp. 273-282.
Parker, B. L. and D. B. McWhorter (1994) "Diffusive disappearance of immiscible phase organic
liquids in fractured media: Finite matrix blocks and implications for remediation". Presented at
the International symposium on "Transport and Reactive Processes in Aquifers," ETH Zurich,
Switzerland, April 11-15.
Parker, B. L., R. W. Gillham, and J. A. Cherry (1994) "Diffusive dissappearance of dense, immiscible phase organic liquids in fractured geologic media", Ground Water, 32, 805-820.
Pruess, K. and Y. W. Tsang (1990) "On two-phase relative permeability and capillary pressure of
rough-walled rock fractures", Water Resources Research, 26,1915-1926.
Reichart, Todd M. (1992) Influence of Vertical Fractures in Horizontally-Stratified Rocks. M.Sc.
Thesis, Dept. of Earth Sciences, University of Waterloo, Waterloo, Ontario.
Rowe, R. K. and F. S. Barone (1991) Diffusion tests for chloride and dichloromethane in Halton Till,
Halton Waste Management Site, Unpublished report to Gartner Lee, Ltd., Markham, Ontario.
December, 1991.
Ruland, W. W., J. A. Cherry, and S. Feenstra (1991) "The depth of fractures and active groundwater flow in a clayey till plain in southwestern Ontario", Ground Water, 29, 405-417.
Russell, B. M. (1993) Nitrate Persistence in Slightly Permeable Sediments in Southern Ontario,
M.Sc. Thesis, Dept. of Earth Sciences, University of Waterloo, Waterloo, Ontario.
Schwarzenbach, R. P., P. M. Gschwend, and D. M. Imboden (1993) Environmental Organic Chemistry, John Wiley and Sons, New York.
Schwille, F. (1988) Dense Chlorinated Solvents, translated by James F. Pankow, Lewis Publishers,
Boca Raton, Florida.
Skagius, K. and 1. Neretnieks (1986) "Porosities and diffusivities of some nonsorbing species in
crystalline rocks", Water Resources Research, 22, 389-398.
Starr, R. C; R. W. Gillham, and E. A. Sudicky (1985) "Experimental investigation of solute
transport in stratified porous media: 2. The reactive case", Water Resources Research, 21, 10431050.
Sudicky, E. A. and E.O. Frind (1982) "Contaminant transport in fractured porous media: Analytical
solutions for a system of parallel fractures", Water Resources Research, 18, 1634-1642.
Sudicky, E. A., R. W. Gillham, and E. O. Frind (1985) "Experimental investigation of solute
transport in stratified porous media: I. The nonreactive case", Water Resources Research, 21,
1035-1041.
Sudicky, E. A. and R. G. McLaren (1992) "The Laplace Transform Galerkin technique for largescale simulation of mass transport in discretely fractured porous formations", Water Resources
Research, 28, 499-514.
Sudicky, E. A., R. G. McLaren, and J. VanderKwaak (1993) Characterization of contaminant
migration processes in fractured geologic media and numerical analysis of pump-and-treat re-
Sec. 12.6
References
393
mediation, Progress Report on Project No. 596G, Ontario Ministry of the Environment, March
8, 1993.
Tang, D. H., E. O. Frind, and E. A. Sudicky (1981) "Contaminant transport in fractured porous
media: Analytical solution for a single fracture", Water Resources Research, 17, 555-564.
U.S.EPA (1992). Dense Nonaqueous Phase Liquids-A Workshop Summary, EPA/600/R-92/030,
Dallas, Texas, April 16-18, 1991.
Wilke, C. R. and P. Chang (1955) "Correlation of diffusion coefficients in dilute solutions", Am.
Inst. Chem. Eng. J., 1, 264-270.
13
Diagnosis and Assessment of
DNAPL Sites
Stan Feenstra 1 and John A. Cherry1 Applied
Groundwater Research Ltd.

2550 Argentia Road
Canada, L5N 5R1

Waterloo, Ontario
Canada, N2L 3G1
Page
13.1
13.2
13.3
13.4
Introduction
13.2.4.5 DNAPL penetration in fractured media
13.3.3.2 Investigations using the "outside-in" approach
13.3.4.1 Test pits
13.3.4.3 Geophysics
Soil Gas Surveys
13.4.1 Description
397
398
398
403
404
405
405
406
409
409
410
412
413
413
413
413
418
421
421
422
423
425
425
426
427
429
429
430
395
396
13.5
13.6
13.7
13.8
Diagnosis and Assessment of DNAPL Sites

13.5.1 Soil Samples
13.5.2.1 General
13.5.2.4 Example calculation: contamination with PCBs
in hydrocarbons
13.5.3 Direct Determination of DNAPL
13.6.1 General
13.6.4 Persistence and Temporal Variability of Dissolved Concentrations
13.6.5 Response of Dissolved Concentrations During Aquifer Pump-and-Treat
13.6.6 Case History-Chlorinated Solvent Plume, Dayton, New Jersey
13.6.6.1 Background
Conclusions
References
Chap. 13
432
432
432
435
437
437
438
440
442
443
443
444
444
446
449
449
449
454
459
460
462
462
462
464
465
ABSTRACT
At many sites that are contaminated with chlorinated solvent compounds, there is no
direct visual evidence of solvent in the form of dense non-aqueous phase liquid (DNAPL)
either in soil cores or in monitoring wells. A lack of direct evidence does not, however,
necessarily mean that DNAPL material is not in the subsurface. Indeed, spatial variability
in the subsurface distribution of DNAPLs as well as the nature of the field investigation
methods typically used to investigate DNAPL sites can prevent subsurface DNAPL from
being found, even when there are large amounts present. The likelihood that DNAPL will
be identified if encountered in borings can be improved through the use of: I) careful
visual examinations during drilling; 2) enhanced visual inspections using dyes and/or UV
light; 3) vapor analyses; 4) chemical analyses of soils and rock samples; and 5) borehole
geophysics. The identification of DNAPL in wells will also be enhanced by use of
appropriate drilling, well construction, and well sampling procedures. The presence and
distribution of DNAPL can be assessed indirectly by evaluating the character of dissolved
plumes, and in some circumstances, through the use of soil gas surveys. Overall, a number
of lines of evidence can be pursued at any given site to evaluate the possible presence
Sec. 13.1
Introduction
397
of subsurface DNAPL. In any given circumstance, however, usually just a few of those
lines of evidence yield valuable information. The challenge at a given site is to render
an opinion, based on the available evidence, regarding the likely presence or absence of
subsurface DNAPL.
13.1 INTRODUCTION
In order to develop rational and cost-effective plans for remediation at sites having
groundwater contaminated by chlorinated solvents, it is essential to determine if any
dense non-aqueous phase liquid (DNAPL) is present in the subsurface, and if so, to
also determine the nature of that DNAPL contamination. The purpose of this chapter is
to describe the use of various types of site information for the diagnosis and assessment
of the presence and distribution of DNAPLs in the subsurface.
The term "DNAPL site" will be used here to designate any site at which appreciable
DNAPL (either residual or free (i.e., mobile) phase) exists below the water table. At many
sites there is no direct visual evidence from soil cores or from samples from monitoring
wells that DNAPL exists in significant quantities in the groundwater zone. However,
because of the potential variability in the spatial distribution of DNAPLs and of the
investigative methods commonly used, this need not mean that DNAPL is not present.
Many problems can arise when a DNAPL site is mis-diagnosed as a non-DNAPL
site. For example, a failure to diagnose the presence of DNAPL can result in the failure
of groundwater pump-and-treat measures intended to provide aquifer restoration. Failure
to identify the presence of DNAPL during drilling through source zones or during the
design of remedial measures in source zones can result in remobilization of the DNAPL
and further spreading of the DNAPL, and worsening of the groundwater contamination
problem.
The diagnosis of DNAPL sites can occur in different ways. Until recently, many diagnosis exercises involved consideration of various types of existing data obtained during
normal hydrogeological site investigations. In some of these cases, DNAPL diagnosis has
been part of a retrospective exercise after remedial actions such as groundwater pumpand-treat have already begun. Frequently, the diagnosis exercise is conducted during a
post-mortem to determine the "cause of death" of the remedial action. In other cases,
supplemental field investigations are conducted which focus specifically on the issue of
the presence of DNAPL. Ideally, the consideration of the potential presence of DNAPL
and the collection of the appropriate field information to allow DNAPL diagnosis should
be a fundamental part of the initial site investigation.
The primary goal in the diagnosis and assessment of DNAPL sites is to establish the
likelihood that significant quantities of DNAPL are present in the subsurface. Of the techniques that can be considered at the present time, only the visual observation of DNAPL
or measurement of dissolved contaminants at saturation levels provides unequivocal evidence of DNAPL presence. Until recently, this was the only type of information which
would have lead site investigators and regulators to recognize the presence of DNAPL.
Many lines of evidence can contribute to the diagnosis of DNAPL sites, but normally, only a few contribute strongly to the diagnosis of any particular site. At most sites,
398
Chap. 13
sufficient data are not available or conditions are not appropriate to allow all possible
techniques to be used. The site information that may be most convincing at one site may
be less convincing at another site. The challenge at any given site is to assess the various
lines of evidence to in order to render a judgment based on the preponderance of all of
the available evidence. Because the diagnosis of DNAPL sites considers much evidence
that requires careful interpretation, the conclusions depend frequently on the experience
of the site diagnostician. It is our observation that less experienced site assessors are less
likely to conclude that DNAPL is present, whereas site assessors with greater experience
at DNAPL sites are more likely to conclude that DNAPL is present.
This chapter is intended to provide a comprehensive description of the investigative
and interpretive techniques for the diagnosis of the presence of DNAPL. These techniques
include: assessment of the site with regard to the chemicals used, nature of handling and
disposal, and subsurface conditions; observations made during drilling and in wells, and
results from analyses of soil, rock, and groundwater samples. While it might be preferred
by regulators that a more formal procedure for the determination of DNAPL presence
be developed, many of the possible lines of evidence are subjective in nature and it
is unlikely that a quantitative assessment of the probability of DNAPL occurrence is
possible.
The simple recognition that DNAPL is present somewhere in the groundwater zone
will not be sufficient for a detailed analysis of past or future DNAPL migration, or the
detailed design of remedial measures which involve containment or in situ treatment of
the source zone. These types of activities require definition of the spatial distribution of
the DNAPL zone vertically and laterally, including the location and size of lenses, pools,
and residual zones. We note in this context that many of the field investigation techniques
used to assess DNAPL presence can also be utilized to assess the spatial distribution of
DNAPLs. For a detailed assessment of DNAPL migration, information on the chemical
composition, density, viscosity and interfacial tension for the DNAPL will be required,
and efforts will often be undertaken specificallyto collect DNAPL samples for laboratory
testing. Such samples might also be required to allow bench-scale testing of remediation
technologies such as surfactant or alcohol flooding.
13.2 GENERAL SITE ASSESSMENT
A general assessment of an industrial or waste disposal site can be made to determine

the potential presence of a DNAPL in the groundwater zone. This general assessment
can usually made early in a site investigation program using existing information about
the site. The site assessor must consider if the chemicals found at the site, or expected
at the site, could comprise DNAPLs, and if the activities at the site could have resulted
in the release of significant quantities of DNAPL into the subsurface.
The first step in the assessment of a potential DNAPL site is a consideration of the
types of chemicals which are found, or might be present at the site. For example, in-
Sec. 13.2
399
formation that large quantities of chlorinated solvents were used or disposed of at a

site is an obvious warning of the possible presence of a DNAPL phase in the subsurface.
A wide variety of chemical products and wastes may comprise a DNAPL. For a
chemical to be regarded here as a potential DNAPL in a given circumstance, it must
have a fluid density greater than l.01 g/cm", a solubility in water of less than 2% (or
about 20,000 mg/L), and a vapor pressure of less than 300 torr. For a potential DNAPL
to be an actual DNAPL at a given site, then the amount of the chemical that was released
to the subsurface will need to have been in excess of the total amount that could have
dissolved into the water and sorb to the soil materials in the source zone. Compounds
which are either more soluble than 2% or more volatile than 300 torr would not generally
be expected to persist as DNAPL phases in the subsurface.
Chlorinated solvents are regarded as the most important class of DNAPL chemicals.
Other classes of DNAPL chemicals include some halogenated aromatics, phthalates, coal
tar, creosote, PCB Aroclors, and some pesticides (see Table 1.1). Some of these other
compound classes have chemical and physical properties that are comparable to those
of the chlorinated solvents, but they have been produced in smaller quantities and have
been used in fewer industries.
As is discussed in detail in Chapter 1, the chlorinated solvents have been produced,
transported and used in very large quantities by a wide variety of industries for almost fifty
years. The U.S. production of the four principal chlorinated solvents of perchloroethylene
(PCE), trichloroethylene (TCE), 1,1,I-trichloroethane (l, 1,1-TCA), and dichloromethane
(DCM) have ranged from 250 million kg to 350 million kg per year. As a result of
occupational health considerations, TCE use has declined in favor of 1,1,1-TCA for
many degreasing and cleaning operations. 1,1, 1-TCA use will, in turn, begin to decline
as a result of reductions in the production of ozone-depleting chemicals.
Most chlorinated solvents have densities that range from 1.1 g/cm'' (as for chiorobenzene) to 1.63 g/cm' (as for PCE). Viscosities are less than or similar to that of water.
These high densities and low viscosities make this class of DNAPLs the most mobile as
a separate non-aqueous phase, and therefore the one possessing the greatest potential for
migration in the subsurface. Also, while the chlorinated solvents have solubilities that
are as much as a hundred thousand times higher than their respective drinking water
standards, these solubilities are low in absolute terms, thus favoring the persistence of a
separate DNAPL phase following a spill.
Coal tar is a waste product of coal or oil gasification, and iron and steel production.
Coal and oil gasification was used to produce gas for heating and lighting in most North
American cities from the mid-1800's until about 1950 when natural gas became widely
available. From the beginning of gasification operations until about 1900, coal tar was
generally disposed of on-site at the gasification plants. Coal tar is currently produced
almost exclusively as a by-product from blast furnace coking in iron and steel production.
Creosote is typically distilled from coal tar. Creosote, in its pure form or mixed with
petroleum hydrocarbons (such as fuel oil), has been used as a wood preserving chemical
in North America for more than one hundred years. Coal tar and creosote have physical
and chemical properties which are very different from those of the chlorinated solvents.
400
Chap. 13
Their densities are only slightly greater than water (1.01 g/cnr' to 1.05 g/cm'), and
their viscosities are greater than water (10 to 100 cP). These properties make coal tar
and creosote the least mobile of the common DNAPLs. However, coal tar and creosote
sites are generally characterized by the presence of very large volumes of DNAPL.
The components of coal tar and creosote such as the polycyclic aromatic hydrocarbons
(PAHs) have very low solubilities, and sorb strongly to geologic media. As a result,
coal tar/creosote DNAPL sites seldom create large dissolved contaminant plumes in
groundwater.
Polychlorinated biphenyls (PCBs) have been used as dielectric fluids in electrical transformers and capacitors, and have had a variety of other industrial uses from
about 1930 until the mid-1970s when the manufacture of these chemicals was discontinued in the U.S. In the late 1960s, PCB production reached almost 35 million kg
per year with about 50% of the production used in transformers and capacitors. A significant portion of the PCB inventory produced is still in service in transformers and
capacitor installations. Aroclors 1221, 1232, 1242, and 1248 are fluid PCB oils with
densities ranging from 1.18 g/cm' to 1.42 g/crrr'. Some commercial formulations for
dielectric fluids were comprised of mixtures of one or more Aroclors with trichlorobenzenes. For example, the product Askarel Inerteen 70-30 was formulated as 70% Aroclor 1254 and 30% trichlorobenzenes, and is a liquid with a density of 1.49 g/cm'.
The viscosities of the Aroclors range from about 5 cP for Aroclor 1221 to 65 cP for
Aroclor 1248. The Aroclors and PCB/trichlorobenzene mixtures have densities comparable to the chlorinated solvents, but their viscosities are higher. Consequently, the
Aroclors would be expected to be somewhat less mobile as a DNAPL phase than the
chlorinated solvents, but more mobile than coal tar or creosote. The PCB compounds
present in the Aroclor mixtures have very low solubilities and are strongly sorbed on
geologic media. As a result, PCB DNAPL sites are usually not characterized by large
dissolved contaminant plumes in groundwater. A general comparison of the mobility of
chlorinated solvents, coal tar/creosote, and the Aroclors as DNAPL phases is given in
Figure 13.1.
A significant number of pesticide products are dense liquids. These products exhibit very wide ranges of physical and chemical properties as compared to other compound classes discussed above, and it is not possible to generalize regarding their mobilities as DNAPL phases. Their densities can range from close to 1.0 g/crrr' to more
than 2.0 g/cnr'. The mobilities of pesticides dissolved in groundwater can also range
considerably because of varying solubility, potential for sorption on geologic media, and
degradability.
The DNAPLs released to the subsurface at many manufacturing sites may be
relatively pure. The chemical composition of the DNAPL found in a leaking disposal
lagoon at an aerospace manufacturing facility is given in Table 13.1.a. In this case, the
DNAPL was predominantly TCE. The small proportions of 1,1,l-TCA and DCM may
reflect impurities in the original TCE product, or the use of smaller amounts of these
solvents at the site. The trans-I,2-dichloroethylene may have been an impurity in the
original TCE, or may have formed by chemical transformation of the TCE during use,
or in the lagoon.
Sec. 13.2
1.7
401
u, JllUl
1.6
IPCB1254
1.5
g
E 1.4 t-:t'
~
c:
~
1.3 t-t11
II TCE
PC8-1248
I-
IPC8-1242
.TC~
oS
...J
~ 1.2
I----- I----
PCB1232
f-~
II
.s
1.1
Chlorobenzene
1.0
,Jr
0.9
0.1
PC8-1221
Typical Coal Tarl

Craosote
10
100
1000
NAPL Viscosity (cP)
Figure 13.1 Relative mobility of selected DNAPL chemicals presented by the NAPL
density and viscosity.
TABLE 13.1.a Chemical Composition of
a DNAPL Recovered From a Disposal
Lagoon at an Aerospace Manufacturing
Facility. DNAPL Density = 1.45 g/crn '.
Compound
Weight %
Trichloroethylene
1,1,l-Trichloroethane
Trans-I,2-dichloroethylene
Dichloromethane
94.1
3.3
1.45
1.15
At waste disposal sites, the DNAPLs encountered in the subsurface are frequently
complex mixtures of many organic chemicals, reflecting the disposal of various used
chemical products. Some of the constituent chemicals of these mixed DNAPLs can be
DNAPLs themselves in their pure forms. Other constituents of the DNAPLs may be
light non-aqueous phase liquids (LNAPLs) in their pure forms (e.g., benzene). Yet other
constituents of the DNAPLs might be miscible in water (e.g., acetone). The chemical
and physical properties of such mixed DNAPLs will depend on their exact chemical
composition. Except for coal tar and creosote, mixed DNAPLs are more dense than water
generally because of the presence of substantial concentrations of halogenated organics
such as the chlorinated solvents. Mixed DNAPLs at waste disposal sites commonly
402
Chap. 13
have densities ranging between 1.05 g/cm" to 1.3 g/cm", which is lower than for pure
chlorinated solvents. The lower densities are usually due to the presence of petroleum
hydrocarbons and other non-halogenated organics. The viscosities of mixed DNAPLs are
generally in the 10 to 100 cP range, which is significantly greater than that of either water
or pure chlorinated solvents. Tables l3.1.b gives the complex chemical composition of
a DNAPL recovered from a landfill used for the disposal of wastes from a chemical
manufacturing facility. The majority of the identified components can themselves be
DNAPLs, and as a result the density of the mixture is relatively high. This DNAPL
mixture was also found to contain a very large proportion of components that could not
be characterized using conventional analytical methods such as gas chromatography/mass
spectrometry (GC/MS). This is typical for the DNAPLs found at landfills and waste
disposal sites. Table l3.1.c gives the chemical composition of the DNAPL recovered
from beneath an illegal chemical and septic waste lagoon. In this case, the mixture
is a DNAPL because of the high concentration of 1,2,3-trichloropropane. All the other
identified components are less dense than water. The density of the mixture is 1.12 g/crrr'
compared to the pure-phase density of 1.39 g/cm'' for 1,2,3-trichloropropane. As for
the DNAPL considered in Table l3.1.b, a large proportion of this DNAPL could not
characterized by GC/MS.
Table 13.1.d gives the chemical composition of the DNAPL recovered from beneath
a PCB storage facility. The mixture is a DNAPL because of the large proportion of
Aroclors and trichlorobenzenes. The physical movement and behavior of this DNAPL
is controlled by properties imparted by the major components. However, despite the
fact that TCE comprises less than 2% of this DNAPL, it is the dominant groundwater
contaminant because it is much more soluble than the PCB compounds, and because it
sorbs so weakly to aquifer materials.
TABLE 13.1.b Example of Chemical
Composition of DNAPL Recovered from a
Landfill at a Chemical Manufacturing Facility.
DNAPL Density = 1.5 g/crn '.
Compound
Weight %
Chi oro benzene

Tetrachloroethylene
Toluene
Trichlorofluorotoluenes
1,2-Dimethyl benzene
1,3-Dimethyl benzene
I, I,2,2- Tetrachloroethane
Trichloroethylene
Dichlorotoluene
1,1'-Thiobisdodecane
Pentachloroethane
Methyl ester (p-chlorophenyl)
phenyl acetic acid
Unresolved by GCIMS Analysis
4.6
4.5
3.9
3.0
2.5
2.4
0.92
0.75
5.1
1.7
0.87
0.62
61.1
Sec. 13.2
403

TABLE B.1.e Example of Chemical
Composition of DNAPL Recovered from an
Illegal Chemical and Septic Waste Dump.
DNAPL Density = 1. 12 g/crn '.
Compound
Weight %
1,2,3- Trichloropropane
Xylenes
Ethyl benzene
Toluene
Unresolved by GC/MS Analysis
23
17
3.8
4.2
52
TABLE B.1.d. Example of Chemical

Composition of DNAPL Recovered from a PCB
Storage Facility. DNAPL Density = 1.2 g/cm:'
Compound
PCB Aroclors 1242, 1248, 1254
Trichloroethylene
Mineral Oils
Weight %
47
9.2
2.9
1.8
40
The presence of substantial concentrations of chlorinated solvents or other halogenated organics in groundwater or soil is commonly the first indication that DNAPL
may be present at a site. The evaluation of what chemical concentrations in groundwater
and soil might reflect the presence of DNAPL in the subsurface will be discussed in
following sections.

Site assessors must understand not only the types of chemicals which may comprise
DNAPLs, but also how these chemicals have been, or are currently being used. In many
cases, site assessors must consider industrial or waste disposal sites where releases of
DNAPL may have occurred many years ago. In such cases, there may be no specific
records of the chemicals used or wastes received, and there be no past employees with
useful recollections of past activities.
When site assessors (often hydrogeologists) consider the past or present use of
DNAPL chemicals at industrial and commercial facilities, it must be recognized that
many chemical products are known by a wide variety of chemical synonyms and product names. As has been described in Table 1.2, even the four most common chlorinated
solvents are known by a large number of names, many of which would not suggest their
common chemical name. For example, in the examination of historical records of chemical purchases at an industrial site to look for evidence of the use of DNAPL chemicals,
references to products like Nema (PCE), Westrosol (TCE), and Chlorothene (I,l,l-TCA)
404
Chap. 13
should not go unnoticed. Information on chemical synonyms and product names for
chemicals of relevance to groundwater contamination can be found in handbook sources
such as Howard (1989) and Montgomery and Welkom (1990). General information on
synonyms and product names for chemicals can be found in the Merck Index as compiled
by Merck & Co. (1989).
At old waste disposal facilities, many of the industries that contributed the wastes
may now be defunct. In these cases there may only be information on the general type of
business conducted by these industries, and the nature of their wastes must be estimated.
The principal uses for the wide range of organic chemicals which can be DNAPLs are
found in Tables 104-1.10. There are separate tables for: the most common chlorinated
solvents, other common chlorinated organics, halogenated organics, substituted aromatics
and phthalates, PCBs, and pesticides. Of particular note are the uses of: I) chlorinated solvents for metal cleaning and degreasing; 2) PCE for dry cleaning of fabrics and clothing;
3) 1,2-dichloroethane in the manufacture of PVC; 4) chlorobenzene as a chemical intermediate for the manufacture of other chemical products; 5) PCB Aroclors in hydraulic
fluids.
With an appreciation for the varied uses of DNAPL chemicals, a site assessor
can identify general industrial and commercial activities that should be associated with
DNAPL use. A summary of some of the key industrial and commercial activities which
use DNAPL chemicals is found in Table 1.11. These activities are separated into four
categories: 1) production and distribution; 2) manufacturing uses; 3) service uses; and
4) waste disposal. Obvious potential sources of DNAPL contamination are activities associated with the production and distribution of DNAPL chemicals themselves or the
use of DNAPL chemicals in the production of other products. The key manufacturing
activities which used DNAPL chemicals involved the use of chlorinated solvents for
metal cleaning and degreasing. Wood preserving activities used very large quantities of
creosote. The manufacture of electrical transformers and capacitors is a well known use
for PCB Aroclors. A lesser known use for PCB Aroclors is in fire-resistant hydraulic
fluids used in foundries, metal casting, iron and steel production, and in natural gas
pipeline compressors. The key service uses of DNAPL chemicals is probably commercial and retail dry cleaning using PCE and l,l,l-TCA presently, and TCE and carbon
tetrachloride in earlier times. Most pesticide products which are DNAPLs are not applied
and used as DNAPLs, but releases of DNAPL pesticides can occur during distribution,
mixing, and preparation for application. Activities such as the past operation of coal and
oil gasification facilities and past and present blast furnace coking operations produce
very large quantities of DNAPLs as waste products.

For an industrial site, the assessor needs to determine the type(s) of manufacturing operations as well as the types and quantities of chemicals that have been used at the site.
Specific information should be sought on the locations and manner of chemical storage,
distribution, use areas, wash-down processes, waste stream handling, and the means of
ultimate disposal. Possible information sources include: 1) interviews with present or
Sec.
13.2
405
former employees; 2) records of chemical purchases, off-site waste disposal, or waste

received; 3) historical site engineering drawings; and 4) archival aerial photographs.
Records of the volumes of DNAPLs that were produced, used, or disposed of at
a site are an important line of evidence. However, such records are often not available.
Estimates of usage can sometimes be made from anecdotal evidence from past employees.
The level of chemical usage at different sites may vary widely. For example, a small
metal fabrication plant with a single TCE degreaser may have used only a few drums
(about 200 L) of solvent per month. In contrast, a large electronics plant involved in the
manufacture of semiconductors may have used several hundred thousand liters per month.
Estimates of the quantities of DNAPL chemicals which might have actually been released
to the subsurface are much more difficult to obtain. Claims by site owners or operators that
a site has had no known or suspected major spills of DNAPL are generally not reliable.
A common cause of DNAPL releases to the subsurface are small repeated spills
or disposals that occur routinely as "de minimus" losses. At industrial sites, common
locations for such releases are chemical off-loading areas, floor drains, sumps and catch
basins, septic tanks and leach fields, process lines and sewers, and storage tanks. In many
hydrogeologic situations, a DNAPL can easily penetrate below the water table if released
as drips or very small periodic discharges at the same location. Prior to the mid-1980s, the
standard operating practices of numerous solvent-using industries was such that routine
or de minimus DNAPL losses were common. Poulsen and Kueper (1992) illustrated the
significance of such small leaks or spills by means of field experiments in a stratified
sandy aquifer. They found that a release of 6 L of PCE at the ground surface could
penetrate the vadose zones to depths of 2.0 to 3.2 m. For soil conditions comparable to
the experimental site and a release of one barrel (~200 L) of PCE, Poulsen and Kueper
(1992) estimated a penetration depth of 70 to 110m.
For disposal facilities such as landfills and lagoons, the locations and nature of the
liquid waste must be determined. In a landfill or disposal pit, liquids may have been
introduced as bulk liquids, or in drums or other containers. DNAPLs dumped as bulk
liquids generally have a greater potential for migration out of the disposal site and into the
subsurface. Bulk liquids wastes were received at many older landfills, and were frequently
discharged into specific pits or bermed areas within the landfill. Similarly, in the past,
many industrial facilities had on-site pits and lagoons for waste disposal operations,
but the locations were not recorded. Examination of historical aerial photographs can be
particularly useful in locating such areas. At old waste disposal facilities there are seldom
records of the types and quantities of waste received. The possible types of wastes and
the potential for the disposal of DNAPLs may need to be estimated based on information
about the industries which used the facility.
An estimation of the potential depth of DNAPL penetration through the vadose zone and
into the groundwater zone is an important component of site assessment. Estimates of the
potential depth of DNAPL penetration are commonly attempted during initial site studies
406
Chap. 13
based on general information about the site operations and the subsurface conditions. The
depth of DNAPL penetration will depend on the properties of the DNAPL, the nature
(e.g., the rate, areal distribution, etc.) of the DNAPL release, and the properties and
geological structure within the vadose and groundwater zones. Unfortunately, it will never
be possible to predict precisely the extent or rate of DNAPL migration. Indeed, even if
complete and reliable information about the DNAPL properties and release are available,
the spatial scales of the geological properties which will control DNAPL migration
are much smaller than can be characterized by site investigations. Nevertheless, it is
important to recognize, in general terms, the factors which will influence the depth of
DNAPL penetration through the vadose and groundwater zones. These factors and the
circumstances in which they might be important are summarized in Tables 13.2 and 13.3.
It has been common practice in the assessment of petroleum hydrocarbons spills (de
Pastrovich et al., 1979) to estimate the depth D of penetration of a DNAPL or LNAPL

spill through the vadose zone on the basis of the volume V released, the area A of the
release, and the retention capacity R, of the soil according to
V
D=AR s
(13.1)
TABLE 13.2 Factors Which Facilitate DNAPL Penetration Through the Vadose and
Groundwater Zones.
Factors facilitating DNAPL penetration
Typical circumstances
High DNAPL density
Chlorinated solvents.
PCB Aroclors.
Low interfacial tension
Surfactants or miscible co-solvents such as

methanol, methyl ethyl ketone or acetone
in the DNAPL.
Surfactants and co-solvents in aqueous wastes
or groundwater.
Complex DNAPL mixtures.
Low viscosity
Chlorinated solvents
Disposal of bulk liquid wastes in landfills,
lagoons.
Catastrophic spills and on-going leakages.
Disposal of bulk liquid wastes in landfills,
lagoons.
On-going leakages.
Large DNAPL volume release
Long duration DNAPL release
High permeability
Sand, sand & gravel, fractured rock.
Vertical and sub-vertical geological structure
Angled beddings in sandy aquifers.

Fractures, fissures, erosional windows in
fine-grained aquitards.
Fractured rock.
Sec. 13.2
407
TABLE 13.3 Factors Which Inhibit DNAPL Penetration Through the Vadose and
Groundwater Zones.
Factors inhibiting DNAPL penetration

Low DNAPL density
High interfacial tension
Typical circumstances
Coal tar / creosote.
Chlorinated organics at low concentrations in
petroleum hydrocarbons.
Relatively pure chemical products.
High viscosity
Coal tar / creosote.

PCB Aroclors.
DNAPL mixtures with high concentrations of
high molecular weight hydrocarbons.
Small DNAPL volume release
Small spills and leaks.
Short duration DNAPL release
Small one-time spills and leaks.
Low permeability
Unfractured silt and clay aquitards.

Unfractured rock.
Horizontal geological structure
Horizontal bedding in sandy aquifers.

Horizontal silt and clay aquitards.
Horizontal bedding plane partings in
sedimentary rock.
Eq.(l3.1) assumes that the NAPL penetrates directly downward from the release area
with negligible lateral spreading, and forms a regular prismatic or cylindrical residual
NAPL zone. The residual saturations of NAPLs measured in the laboratory for sandy
aquifer materials range from about 5 to 25% of the pore space (Mercer and Cohen, 1990).
For a total porosity of 30%, this range of residual saturations yields retention capacities
of 15 to 75 L'rrr', or 1.5 to 7.5% of the bulk volume of the soil. This range of values
is typically used for estimating NAPL penetration through the vadose zone. Thus, for
a release of 6 L of DNAPL over an area of 0.1 m", the depth of penetration would
be estimated to range from 0.8 to 4.0 m. The higher the retention capacity, the smaller
the depth of DNAPL penetration. Obviously, such calculations assume that the residual
saturations measured in the laboratory are applicable to undisturbed soils, and that the
residual zone in the soil is homogeneous.
The reliability of the above type of calculation can be examined by comparison
with the results of two controlled experimental releases of PCE into a stratified sandy
aquifer (Poulsen and Kueper, 1992). In one experiment, referred to as the instantaneous
(or "ponded") release, 6 L of PCE was spilled at the ground surface over an area of
0.1 m2 over a period of 90 s. In the other experiment, referred to as the "drip" release,
6 L of PCE was released over an area of about 1 cm 2 over a period of 100 min. For
both spills, the PCE was dyed red and the distribution of PCE residual was mapped after
excavation both visually and by analysis of soil samples. In both spills, the migration of
PCE was strongly influenced by the angled bedding structure of the sands. PCE migrated
preferentially along those layers with the highest permeability, and exhibited a highly
variable distribution on a scale of millimeters. In both cases, the migration of PCE along
small-scale angled bedding (laminations) caused a substantial lateral displacement of the
408
Chap. 13
resulting residual zone away from the release location. The instantaneous release penetrated to a depth of 2.0 m and the drip release penetrated all the way to the water table
at a depth of 3.2 m. Had the bedding been more horizontal, greater lateral spreading and
less vertical penetration would have been expected.
In the Poulsen and Kueper (1992) spills, the PCE residual saturations were measured in small, ~ I cm' soil samples at numerous locations within the residual zones.
The average residual saturation at various depths are shown for the two experiments
in Figure 13.2. The highest residual saturations are comparable to previously-reported
laboratory values (i.e., 1.5 to 7.5%), but many values are considerably lower. When the
volume of the release is divided by the volume of the DNAPL residual zone, the average
retention capacity of the soil is calculated to be 1.3% and 0.5% for the instantaneous
and the drip releases, respectively. These experiments clearly illustrate the difficulties in
estimating the potential depths of DNAPL penetration through the vadose zone. Indeed,
the bulk retention capacities of soils may be significantly lower than laboratory values
Instantaneous Release
PCE Residual (bulk volume %)
234
0.0
0--....
0.5
1.0
Xc:::::
._ ..
I1.5
-=g.
2.0
5
X
---- --
;x~
2.5
3.0
----
-------
3.5
Drip Release
234
0.0
.~~-
_'\
0.5
..-::::::::-
1.0
I1.5
s:
~ 2.0
o
2.5
3.0
3.5
_rl-~-------
.....
-~II-
.~
----
~
----
--~-
- - -
---
--_.._-
Figure 13.2
Vertical profiles of DNAPL
residual content resulting from experimental
releases of peE in the vadose zone. Data

from Poulsen and Kueper (1992).
Sec. 13.2
409
for residual saturation so that the use of laboratory values in penetration calculations can
lead to underestimates of the depth of DNAPL penetration. At the same time, however,
this effect can be partly compensated for by geological structures which can cause a
substantial degree of lateral spreading which can expand the volume of the DNAPL zone
and restrict the depth of penetration.
The depth of DNAPL penetration through the vadose zone can be influenced by the
vaporization of the DNAPL components. If the vadose zone is thick and the vapor
pressure of the DNAPL high, vaporization may deplete the DNAPL mass sufficiently to
prevent the DNAPL from reaching the groundwater zone. Vaporization will create a halo
of lower level contamination in the soil gas, soil water, and soil around the DNAPL zone.
Numerical modeling by Mendoza and Frind (1990) of NAPL sources in the vadose zone
can be used to gain some insight into the persistence of DNAPLs having relatively high
vapor pressures. The numerical modeling considered a cylindrical NAPL source zone
situated within a sandy vadose zone. The source zone was assumed to have a radius of
1.05 m, a height of 0.9 m, and to contain 125 kg of NAPL. A generic organic chemical
with a vapor pressure of 182 torr was considered. For comparison, the vapor pressures
of DCM, TCE, l,l,l-TCA, and PER are 353, 58, 96, and 8 torr at 20C, respectively. In
the simulation, 83% of the source was depleted after 8 days, and 94% of the source was
depleted after 16 days. Higher vapor pressures and smaller NAPL source zones would
require shorter periods of time to vaporize, and vice versa.
The factors which influence penetration of DNAPL into the groundwater zone are generally comparable to those which influence penetration in the vadose zone. The effects of
layered geological structures on DNAPL migration were clearly illustrated in a laboratory
experiment by Kueper (1989) and Kueper et al. (1989), and are shown in Figure 13.3. In
this experiment, even relatively small differences in hydraulic conductivity were found
to inhibit downward migration and cause lateral spreading of PCE.
Stratification and variability in geological structure in the groundwater zone will
cause highly variable DNAPL distributions. This was illustrated clearly in a field experimental release of 231 L of PCE below the water table in a stratified sandy aquifer by
Kueper et al. (1993). The experiment was conducted within a steel-sheet-pile test cell
measuring 3 m by 3 m, and anchored into an underlying clay aquitard to a depth of
3.5 m. The PCE was dyed red. Following the release, the upper 1 m of the cell was
excavated for visual examination and analysis of soil samples. Three continuous cores
were collected from the lower 2.5 m of the cell for the analysis of soil samples. The
cores were divided into 5 ern segments and a small sample (""'-'1 em") was removed from
each for determination of the degree of PCE saturation. The results from one of these
cores is shown in Figure 13.4. The highest PCE saturations occur in the upper portion,
the middle portion and the lower portion of the core. Each of these zones is interpreted
to represent layers (or pools) of PCE which accumulated on horizontal zones of lower
410
Chap. 13
Source
~ ~ ~
Figure 13.3 An illustration of the effect

of geologic variability on the migration of
DNAPL in the saturated zone. Results
from a laboratory experiment performed
by Kueper et al. (1989). Hydraulic
conductivity of each sand layer is indicated
in cmls times 10- 2 . Adapted from Kueper
(1989).
o
198.5
198.0
I
c
o
[~
197.5
:~-
197.0
j--
iii 196.5
196.0
195.5
-- --
10
198.5
.~
..-
I":~-I
tsr
50,000
198.0
~-
197.5
i---.
197.0
UJ
196.5
100,000
I~
I~~
~
.
r
196.0 ,
195.5
- -
PCE (mg/kg)
Figure 13.4 Vertical profiles of DNAPL residual content resulting from an experimental
release of peE in the groundwater zone. Data from Kueper et al. (1993).
permeability. The deepest pool was found on the clay aquitard. Between the layers, the
PCE saturations were lower and highly variable. Measurable PCE could not be found in
some zones. A smaller experiment yielded similar results (see Figure 13.5). The variability exhibited in these experiments illustrates that estimating the depth of penetration
of DNAPL into the groundwater zone will be as difficult as in the vadose zone.
13.2.4.5 DNAPl penetration in fractured media

Layers of fine-grained media such as silt or clay can be effective barriers to vertical
DNAPL migration, and substantial amounts of potentially mobile DNAPL can accumulate
on such barriers. The same applies to intact low-permeability rock. However, these strata
Sec. 13.2
(a)
(b)
Figure 13.5 Example of scale and spatial variability in DNAPL distribution resulting
from an experimental release of PCE in the groundwater zone. PCE is dyed red using
Sudan TV dye. (Photographs by Kim Broholm.) a) Horizontal view of top of layer;
b) vertical section through PCE layer.
411
412
Chap. 13
will not act as barriers if they contain fractures or other discontinuities. Silt and clay
strata may contain fractures as a result of desiccation and weathering, stress relief, and
tectonic stresses; almost all sedimentary and crystalline rocks contain fractures.
Calculations and modeling conducted by Kueper and McWhorter (1991) indicate
that DNAPLs can penetrate fractures having apertures as small as 10 to 20 /lm under the
types of DNAPL heads that are likely to occur at chemical spill and waste disposal sites.
Detailed studies of clay strata by Rudolph et at. (1991) and McKay et al. (1993) have
found that vertical fractures of this aperture are commonly present in strata normally considered to be competent aquitards. Although it might be anticipated that vertical fractures
should be closed at some small distance below the water table as a result of confining
stress imparted by the overlying geologic material, field studies have shown that open
fractures can exist in many types of clayey deposits to considerable depths, many tens of
meters in some cases. Such vertical fractures can provide important pathways for DNAPL
migration downward through silt and clay strata. Similarly, vertical fracture orientations,
fracture zones, and fault zones in rock strata can provide pathways for the downward
migration of DNAPLs. The variability of fracture characteristics, the difficulty of field
measurements of fracture characteristics, and the complexity of fracture networks prevent
all but the simplest of assessments of the depth of DNAPL penetration in fractured media. Such assessments will in most cases be limited to simple determinations of whether
downward migration is "likely" or "unlikely".
As an example, consider a situation where a shale layer separates two carbonate
rock aquifers. Liquid PCE has been found pooled at the base of the upper aquifer and,
dissolved PCE has been found in the lower aquifer. The question to be addressed is
whether liquid PCE has penetrated into the lower aquifer, or whether the PCE in the
lower aquifer is due solely to the leakage of dissolved PCE from the upper aquifer.
Hydraulically, the shale unit behaves like an aquitard with a consistent and significant
difference in head between the aquifers. A pumping test of the lower aquifer and monitoring of the upper aquifer suggest a bulk hydraulic conductivity of 2 x 10- 7 cm/s for
the shale. Exposure of the shale in a nearby outcrop exhibits continuous vertical fractures spaced at about 2 to 5 m apart. A simplified relationship between bulk hydraulic
conductivity and fracture frequency for smooth planar fractures (see Figure 12.2.a) can
be used to estimate a range of fracture apertures, and the result is 20 to 30 /lm. Using
an available relationship between fracture aperture and DNAPL head (see Figure 11.2),
a head of only 0.1 to 0.7 m would be necessary, depending on the interfacial tension, to
allow liquid PCE to penetrate into these fractures. Thus, given that pooled liquid PCE is
present on top of the shale unit, these calculations suggest that it is likely that liquid PCE
has penetrated into the lower aquifer, and could be the source of the dissolved PCE there.

The variable distributions of DNAPLs expected within the vadose zone and the groundwater zone make simple estimates of DNAPL penetration depths unreliable. However,
provided that these uncertainties are recognized, such calculations can be useful in assessing the range of possibilities at a DNAPL site. More rigorous calculations of DNAPL
Sec. 13.3
413
migration are possible using two-phase and three-phase flow numerical modeling techniques for homogeneous geological media and for simple heterogeneous systems. We
note though that the scale of the geological variability that can influence DNAPL migration is far smaller than that which can currently either be characterized in the field, or
accounted for in numerical models.
13.3 DETERMINING THE PRESENCE OF DNAPL BASED ON FIELD

OBSERVATIONS
Valuable direct and indirect evidence of the presence and distribution of DNAPLs can
be obtained by careful observation and field testing during drilling and recovery of soil
and rock samples. There are some circumstances where the presence of a DNAPL is
obvious. Abundant dark-colored DNAPL at high residual saturation, or in layers and
pools is usually readily visible in soil cores. This is frequently the case for coal tar and
creosote which are very dark and have sometimes been released to the subsurface in very
large quantities (see Sections 13.2.1 and 13.2.2). It is less common to find clear visual
evidence of liquid chlorinated solvents since the source quantities are usually smaller
and the source zones are more variable; in most cases, careful centimeter-by-centimeter
examination of soil cores is essential for any visual identification of DNAPL presence.
In the shallow vadose zone, there may commonly be dark staining of soil, but no
visual evidence of DNAPL. In the case where the original DNAPL was a mixture of
volatile and less volatile compounds, such staining may be due to the presence of the
less volatile residues following vaporization of the volatiles. Such observations would
suggest that DNAPL had been present in the past and had migrated at least to the depth
of the observed staining. Visual evidence of DNAPL or staining in soil samples can
provide clear information on the vertical distribution of DNAPL because the depth of
the observations can be defined.

Enhanced Visual Identification. Visual identification of DNAPL in soil samples can
be enhanced using methods such as ultraviolet (UV) fluorescence and various types of
soil-water shake tests. These methods have not yet been used widely, but have been
evaluated in controlled laboratory studies by Cohen et at. (1992). The UV fluorescence
method involves the examination of soil samples under a portable UV light. Since many
aromatic and polycyclic aromatic hydrocarbons fluoresce, UV fluorescence has been
used for decades in the petroleum industry to identify crude oil in cores, drill cuttings,
and drilling mud. Many unsaturated aliphatic hydrocarbons such as TCE and PCE also
fluoresce. Examination by UV fluorescence can be accomplished in transparent plastic
bags so that the samples can be manipulated without excessive loss of volatiles. Samples
of soil known to be uncontaminated should also be examined as controls because some
minerals and shell fragments also fluoresce.
414
Chap. 13
The soil-water shake test involves the transfer of a soil sample to a clear centrifuge
tube with an equal volume of water. The tube is stoppered and the mixture shaken by
hand. The DNAPL is identified by examination of the tube walls and bottom. This test
can be enhanced by centrifugation of the mixture to facilitate the accumulation of the
DNAPL at the bottom of the tube. It can also be enhanced by the addition of a small
amount of hydrophobic dye such as red Sudan IV or Oil Red a to the mixture. (These
dyes are insoluble in water, but are soluble in many organic liquids.) Any DNAPLs
present in the sample thereby become red in color. These methods can be performed
readily in the field during drilling and soil sampling activities.
Cohen et al. (1992) compared the performance of unaided-visual identification of
NAPLs in soil samples with UV fluorescence and four types of soil/water shake tests.
The methods were compared using prepared soil samples. Using three different soil
types, 78 samples were mixed with varying quantities of kerosene, chlorobenzene, and
PCE. The residual NAPL saturations ranged from 1 to 23% of the soil pore space; 22
of the 78 soil samples were blanks. Each sample was assessed using each test method
by examiners unaware of their contents. The results of these tests are summarized in
Figure 13.6. Unaided-visual examination correctly identified the presence of NAPL only
30% of the time, whereas the enhancements permitted correct identification 55 to 90% of
the time. UV fluorescence and the addition of dye to the soil/water shake test provided
correct identification of NAPL 80% of the time. These laboratory tests suggest that simple
and effective means of enhancing direct visual identification of some DNAPLs in soil
samples are available.
Vapor Analyses. The determination of organic vapor concentrations is commonly
performed during drilling and sampling to determine compliance with health and safety
conditions in the working environment, and for screening of soil samples for contami-
100
90
c
0
~
o
EQl
:E
tl
80
70
60
50
40
EQl
30
Ql
n,
20
10
0
OJ
::;]
Ul
I_ NAPL Presence D NAPL Absence I
Figure 13.6 Comparison of the

effectiveness of various field testing
methods for the identification of NAPLs in
soil samples. Results are from laboratory
testing of spiked samples performed by
Cohen et al. (1992).
Sec. 13.3
415
nation by volatile organics. Such analyses are often performed using a portable organic
vapor analyzer (OVA), and may be useful in the assessment of the presence of DNAPLs.
Many DNAPL chemicals have high vapor pressures and correspondingly high saturated vapor concentrations (see Table 13.4). The saturated vapor concentration of each
individual component of a multi-component DNAPL will be lower than its pure phase
saturated vapor concentration. The individual saturated vapor concentrations from a mixture can be estimated using Raoult's Law
(13.2)
where Sg,i is the saturated vapor concentration of component i in equilibrium with the
multi-component DNAPL, Xi is the mole fraction of i in the mixture, and S~,i is the
pure-phase saturated vapor concentration of component i. As an example, at 20C, TCE
has a saturated vapor concentration of 76,000 ppmV. Thus, for a mixture containing a
5% molar concentration of TCE in a mineral oil (XTCE = 0.05), the saturated vapor
concentration of TCE would be ~3,800 ppmV.
Data on the vapor pressure of pure compounds are commonly reported for temperatures of 20 or 25C. However, the vapor pressure of any given compound will decrease
strongly as the temperature decreases. At 0 DC, the vapor pressures for the common
chlorinated solvents are lower by a factor of 1/3 to 1/4 of what they are at 20 or 25 DC.
This effect must be taken into consideration when interpreting vapor concentration data.
Because of their nature and use, and because of the complex chemical composition of vapors expected at most sites, OVA instruments usually provide only a semiquantitative measure of organic vapor concentrations at most field sites. OVA instruments can utilize flame ionization detectors (FIDs) or photo-ionization detectors (PIDs)
for the determination of organic vapor concentrations. FID and PID instruments have
different sensitivities for a given organic compound. PID instruments have different
sensitivities depending on the ionization energy of the lamp utilized in the detector.
OVA instruments are commonly capable of indicating vapor concentrations from about
I ppmV to a full-scale reading of 1,000 or 2,000 ppmV. In addition, OVA instruments
are normally calibrated to a gas standard containing a single compound which may
TABLE 13.4
Saturated Vapor Concentrations at 20C for Selected DNAPL Compounds.
Compound
Vapor pressure
(torr)
Vapor cone. in air

(mg/L)
Vapor cone. in air

(ppmV)
Tetrachloroethylene
14
127
19,300
Trichloroethylene
58
417
76,700
100
349
90
151
730
1,620
132,000
459,000
758
987
118,000
199,000
Chlorobenzene
1,2-Dichlorobenzene
12
74
19
15,700
3,100
1,2-Dichloroethane
61
331
80,400
I, I, I-Trichloroethane
Dichloromethane
Carbon Tetrachloride
Chloroform
2.3
416
Chap. 13
or may not be a compound of interest. Consequently, for a complex mixture of organic

vapors, the total concentration indicated by an OVA measurement may be substantially
different than the actual total concentration. In most circumstances, an OVA measurement
must be considered to be an "order-of-magnitude" estimate of the actual total concentration. However, despite these limitations, OVA measurements can still provide useful
information on the presence of a DNAPL because OVA measurements can be performed
readily in the field, and because OVA instruments can measure relatively low vapor
concentrations compared to the vapor concentrations emitted from many DNAPLs.
The field screening of vapor concentrations emitted from soil samples can be performed in several ways. One way is to remove a soil core from the sampler and pass the
inlet tip of the OVA slowly along the length of the core while observing the indicated vapor concentrations. We note that the observed concentrations may be lowered by dilution
in the air, and by slow emission of vapors from the samples. Dilution effects can be reduced by transferring the sample into a bag or jar equipped with access for the OVA inlet
tip. The emission of vapors is slow from samples at low temperatures and/or with high water contents. The transfer and temporary storage of samples in a bag or jar may minimize
these effects by permitting the sample to warm and vapors to accumulate in the container.
Cohen et al. (1992) have shown in laboratory studies that the vapor concentrations
emitted from a soil sample containing a pure, volatile DNAPL may be substantially
lower than the corresponding saturated vapor concentration. A total of 56 soil samples
were prepared consisting of three different soil types. The porosity saturations of PCE
and chlorobenzene ranged from 1% to 23%. The saturated vapor concentrations for PCE
and chlorobenzene at 20C are 19,300 ppmV and 15,700 ppmV respectively. Of the
56 samples, 11 were blanks and 7 contained only chemicals dissolved in the pore water.
Each of these samples was prepared in a sealed polyethylene bag and tested by inserting
the inlet tip of the OVA with an FID into the bag. The results of the tests are shown
in Figure 13.7. The OVA readings for the blanks ranged from 1 ppmV to 20 ppmV.
Readings for the samples containing only dissolved chemicals ranged from 1.5 ppmV to
30 ppmV. For the samples containing PCE and chlorobenzene in the form of DNAPL,
OVA measurements were much higher, but variable, ranging from 80 ppmV to the fullscale reading of 1,000 ppmV. These tests illustrate the difficulty in relating OVA readings
to the presence of DNAPL in a quantitative manner. However, as a "rule of thumb", fullscale OVA readings of 1,000 to 2,000 ppmV are probably a reasonable indication of the
presence of DNAPL when the DNAPL is comprised of one or more chlorinated solvent.
Sampling Drilling Water. The sampling, examination, and analysis of drilling
water can be useful in the identification of a DNAPL phase during drilling. As will
be described below, this type of inspection may be needed in order to avoid drilling
through DNAPL zones and causing contamination of deeper zones by "short circuiting".
The presence of a DNAPL phase can be inferred from observations of oily films and
sheens on the drilling water returned to the ground surface. The determination of appropriate indicator chemicals in the drilling water, or using methods such as field gas
chromatography can provide real time information during drilling.
Unfortunately, not all drilling methods are compatible with sampling and analysis
during drilling. Methods that employ mud or that add water to the formation during
Sec. 13.3
417
1000
I--X
X
I.e.
e>
lIl:
100
XCB
c:
'"
~CC
<
r-t
:::
OPCE
>f
10
Blank
~
r-t
1000
_.._,
~X
100
25
20
-X
'[
15
10
NAPL Saturation (% of pore space)
n
~
Lr-i.
1-
!)L
g>
CC
<
>
XCS
OPCE
I
10
X
Blank
ill
1
20,000
40,000
60,000
80,000
Soil Concentration (mglkg)
100,000
Figure 13.7 Effectiveness of organic

vapor analyzer (OVA) measurements for
identification of NAPLs in soil samples.
Results are from laboratory testing of
spiked samples performed by Cohen et al.
(1992).
drilling are unlikely to yield water samples that reflect the composition of the groundwater. Similarly, methods that employ air flush are not compatible with the determination
of VOCs in the return water because of losses to the air stream. Sampling and analysis
of the return air using an OVA or GC is possible, but the degree of dilution resulting
from the high air flow makes the detection of VOCs difficult.
An alternative to sampling of drilling water during drilling is to stop drilling at
frequent intervals, purge the water from the drill string, and collect a sample for analysis
or inspection. This is possible when using drilling methods which advance a hollow
drill string or casing. Taylor and Serafini (1988) describe the use of a specially-designed
hollow-stem auger for the delineation of dissolved contaminant plumes. The first auger
flight was constructed with a slotted screen to allow collection of groundwater for analysis
in the field.
418
Chap. 13
At some sites, DNAPL can be found in one or more monitoring or pump-and-treat wells.
Although this finding confirms that the site is a DNAPL site, it is generally impossible to
relate the thickness of DNAPL in such wells to the volume of DNAPL in the subsurface.
In addition, it is generally not possible relate the finding of DNAPL in wells to the
vertical distribution of DNAPL in the subsurface. The only circumstance in which it
might be possible to relate DNAPL thickness in wells to the thickness of DNAPL pools
in the subsurface is when DNAPL occurs in large discrete pools in granular media.
This has been found at some creosote and coal tar sites where very large volumes of
product have been released to the subsurface, but it is rare for chlorinated solvent DNAPL
sites. Even when large pools of solvent are found, wells must be installed precisely to
intersect the entire vertical thickness of the pool. If the wells do not extend to the full
thickness of the pool, the observed DNAPL thickness will be erroneously small. If the
wells extend deeper than the base of the pool, the observed DNAPL thickness will be
erroneously large. And, even when wells are installed appropriately within the pool, the
DNAPL thickness in the well can only be related to the pool thickness through the use of
information on the capillary properties of the formation and of the well material, together
with the properties of the DNAPL. This type of information is seldom available.
Of course, if DNAPL free product has not been detected in wells, it cannot be
concluded that DNAPL is not present in the aquifer. At many sites, the zones of freeproduct DNAPL may be very small in areal extent relative to the spacings of borings.
Consequently, the DNAPL zones may be intersected only infrequently by borings. Even
when DNAPL zones are intersected, there are numerous circumstances in which the
DNAPL may not be found in the well. For example, a DNAPL zone may be thin in
vertical extent, and may not contact the open interval of a monitoring well which passes
through it. Also, most of the DNAPL in the subsurface may be present only at residual
saturation, and such DNAPL will not be able to enter the boring or monitoring well.
Figure 13.8 illustrates several circumstances that will inhibit identification of the
presence and distribution of a DNAPL using monitoring well observations. In most
circumstances, the sand pack or filter pack material used in well construction is more
permeable than the geologic media and should allow unrestricted entry of DNAPL into
the well bore. However, DNAPL will always migrate along the most permeable pathways,
and will commonly accumulate in relatively permeable zones that are bounded by lower
permeability zones. If DNAPL occurs in thin layers or pools, there may not be sufficient
fluid potential in the DNAPL to overcome the capillary resistance of the sand pack and
allow entry into the boring (see Figure l3.8.a). To minimize this possibility, sand pack
material can be sized to promote DNAPL entry.
DNAPL layers and pools are generally thin compared to the open interval of a
boring, and may not be identified during drilling operations. If DNAPL enters a boring,
it will sink to the bottom (see Figure l3.8.b). If the DNAPL accumulation is sufficiently
large, it will enter the well screen and may be found by well sounding or sampling.
However, it will not be possible to relate the elevation of the DNAPL in the monitoring
well to the elevation of DNAPL zones in the formation because DNAPL could have
entered anywhere along the open interval of the boring.
Sec. 13.3
419
a.
b.
Well screen
DNAPL layer
DNAPL in well
DNAPL must be able to overcome capillary

resistance of the sand pack to enter the boring.
c.
DNAPL from upper layers will sink to the bottom

of the well.
d.
Well screen
No DNAPL in well
DNAPL may not accumulate sufficiently to enter

the well, or may leak out of the boring.
Figure 13.8
? DNAPL in well
Small volumes of DNAPL may be mixed with

fine sediment in the botton of wells.
Various conditions required for the accumulation and identification of
DNAPL in a monitoring well.
DNAPL that enters the boring and sinks to the bottom will not enter the well screen
unless the DNAPL accumulates as high as the bottom of the well screen. Monitoring
wells are commonly constructed so that the well screen is some distance above the bottom
of the sand pack. If the volume of DNAPL that enters a boring is small, it may not be
sufficient to rise up to the well screen. If the formation surrounding the bottom of the
boring is relatively permeable, it is also possible that the DNAPL may exit the boring
before it can accumulate sufficiently to enter the well screen (see Figure 13.8.c). In this
case, not only will the DNAPL go unidentified in the monitoring well, but the boring
will become a conduit for DNAPL migration deeper into the subsurface. Niemeyer et al.
420
Chap. 13
(1992) proposed that these situations could be avoided by the use of a cement basket
and sump attached to the bottom of the well screen (see Figure 13.9). The cement basket
is a funnel-like device used to set grout seals for well construction. Any DNAPL that
migrates down the sand pack is deflected by the cement basket through the well screen
and accumulates in the sump. Although this type of construction may avoid making
the borings a pathway for downward DNAPL migration, it will not prevent downward
migration that might occur during drilling and well installation.
The accumulation of DNAPL in the bottom of a monitoring well can be determined
most effectively by use of interface probes, or by sampling at the bottom of the well.
Interface probes are down-hole sensors on a graduated cable or tape. They commonly
use an optical sensor for detecting the air-water interface, and a conductivity sensor for
detecting a waterJDNAPL interface. (In wells that also contain an LNAPL, the optical
sensor would detect the airlLNAPL interface, and the conductivity sensor would detect
the LNAPLIwater interface.) Under ideal conditions, a DNAPL thickness of about 0.5 cm
or greater can be detected. However, a DNAPL can be difficult to detect by these means
when it is mixed with fine sediment at the bottom of the well, when it is conductive, or
when it preferentially wets the probe materials.
Sampling of the bottom of a well can provide direct visual evidence of the presence of a DNAPL. It can also provide samples for measurement of the chemical composition and fluid properties of the DNAPL. Samples can be collected using pumps,
bottom-loading bailers, and discrete-depth canister samplers. In general, evacuated discrete-depth samplers provide the most reliable method with the least degree of sample
disturbance. Common DNAPL chemicals can degrade many pump materials. For relatively high density DNAPLs, the depth from which suction pumps, inertial-lift pumps,
or positive-displacement pumps can be operated may be limited. When using a bottom-
1==::j~+-'-'~- Well screen
:t-c---
Sand pack
Cement basket
attached to well
screen
Grout
-eump
'---',~-:-DNAPL
DNAPL is deflected by cement basket into the well screen
and accumulatesin the sump.
Figure 13.9 Schematic of a funnel

collector device to enhance the
accumulation and identification of DNAPL
in a monitoring well. Concept from
Niemeyer et at. (\992).
Sec. 13.3
421
loading bailer to sample DNAPLs with densities greater than about 1.3 g/cm", PVC and
Teflon will not provide adequate valve seating and the DNAPL may leak out as the bailer
is retrieved. (PVC has a density of about 1.4 g/cm", and Teflon has a density of about
2.2 g/cm-.) Thus, for sampling of a high density DNAPL, a steel ball valve should be
used to ensure adequate valve seating.
Although significant quantities of DNAPL may accumulate in some wells, frequently the DNAPL is present only as thin layers or globules at the bottoms of the wells.
In cases where a large quantity of fine-grained sediment is also recovered from the bottom
of a well, some type of phase separation or chemical analysis may be required to identify
the presence of the DNAPL. Separation may be as simple as mixing the sediment with
clean water and "panning" for DNAPL. The soil-water shake methods and the addition
of hydrophobic dyes described in Section 13.3.1.2 for soils (Cohen et al., 1992) can also
be used.
At sites where DNAPL has been found in one or more monitoring wells, it is
common that many of the monitoring wells were installed before a single well exhibited
any DNAPL. Then, in some cases, many more wells needed to be installed before a
second encounter occurred. At most sites, once many monitoring wells are installed
without a single DNAPL encounter, it is not reasonable to continue installing wells so as
to determine definitively that the site is a DNAPL site. The collection and use of other
lines of evidence may be a more effective approach.

When a boring or monitoring well installation intersects and also extends deeper than a
DNAPL layer or pool, there is a potential for further downward migration of the DNAPL
and expansion of the zone of contamination. This situation is sometimes referred to as
"short circuiting", and is obviously to be avoided. The potential for short circuiting is
also of concern in the interpretation of sampling results from most wells.
DNAPL may move downward within a boring along the open borehole during
drilIing and sampling before welI installation, along the sand pack of a completed well
instalIation, and along an unsealed boring. The potential for short circuiting is greatest
when borings intersect DNAPL accumulated in thick lenses or pools, especially when
the DNAPL has a low viscosity and/or a high density. An example of short circuiting
along the sand pack of a completed well is illustrated in Figure 13.8.c. The experimental
release and subsequent excavation of a PCE spill in a stratified sandy aquifer (Broholm
et al., 1994) has provided clear visual evidence of DNAPL short circuiting. Figure 13.10
provides an illustration of downward leakage of peE along a unsealed boring, and along
the outside of a core tube that intersected a thin (I em to 2 em) DNAPL pool. We
can conclude that short circuiting of DNAPLs can occur along the contact between the
formation and: 1) the drill casing; and/or 2) the augers or core tubes during drilling and
sampling operations.
Given the nature of DNAPL behavior, it is likely that many of the borings and wells
installed in the past at DNAPL sites have caused short circuiting of DNAPL, thereby
422
Chap. 13
Figure 13.10 Example of short circuiting

of DNAPL downward along an unsealed
boring (circular area) and along the outside
of a core tube. Both the boring and core
tube intersected the thin PCE layer shown
in Figure 13.5 and allowed PCE to migrate
to deeper levels. PCE is dyed red using
Sudan IV dye. (Photograph by Kim
Broholm.)
leading to misleading conclusions regarding the original morpology of the DNAPL in

the subsurface. In some cases, short circuiting down borings may have contributed significantly to expanded zones of contamination.
13.3.3.2 Investigation using the "outside-in" approach

Obviously, the potential for remobilization of DNAPL along borings can be reduced most
easily by not drilling in areas that are known or suspected to be DNAPL source zones.
However, site investigations often focus first on the source zone(s), and only then expand
to include the dissolved plume(s). In the context of a site investigation strategy, emphasis
should often rather be placed on defining the extent of the dissolved plume(s) first, then
working back toward the source zone(s). This is referred to as the "outside-in" approach.
This approach is also attractive because the migration of dissolved plumes almost always
Sec. 13.3
423
represents the most important pathway leading to exposure to chemicals from the subsurface, and should thus be focus for site investigations and risk assessments in any event.
In many circumstances, even the evaluation and design of DNAPL remedial measures do not require work that involves drilling directly through the source zone(s). As
will be described in subsequent sections, information on the general spatial distribution
of DNAPL zone(s) can often be inferred from soil gas surveys and from monitoring of
the dissolved groundwater plume(s) outside the source zone(s). This level of source zone
characterization is then generally sufficient for the design of remedial measures such as
the hydraulic containment of the source zone(s) by groundwater pump-and-treat.
When undertaking remedies involving the in situ treatment of a DNAPL source
zone, it may be necessary to drill directly into the source zone in order to obtain information on the physical and chemical characteristics of the DNAPL, as well as detailed
information on the nature and shape of the source. Because of the potential for short
circuiting, such drilling should be conducted using suitable precautions, and only when
absolutely necessary.

Various precautions can be taken to minimize the potential for short circuiting when
drilling through DNAPL zones. In most geological settings, although it is possible to
minimize it, one can probably never eliminate any chance of short circuiting. It is especially difficult to prevent short circuiting in fractured rock formations having complex
fracture networks. However, in all cases where drilling through known or suspected
DNAPL zones takes place, efforts should be made to reduce the length of the boring
interval that remains open as well as the time that borings remain open during drilling,
sampling, and well installation. In some cases, the precautions taken to minimize DNAPL
short circuiting may limit the usefulness of the borings for the collection of other types
of hydrogeological information.
The entry of DNAPL into a boring can be inhibited by high drilling water levels
within the boring, or by the use of high-density drilling muds. In relatively permeable
strata, the former will result in substantial loss of drilling water into the formation. Large
volumes of drilling water in the formation around the well may then interfere with the
interpretation of dissolved phase data for groundwater samples collected soon after well
installation. The use of high density drilling muds may form a filter cake on the boring
walls that is difficult to remove during well development. This remaining filter cake may
interfere with the interpretation of inorganic and metals data for the groundwater, and
with the measurement of the hydraulic conductivity using single-well response tests. The
filter cake might also inhibit DNAPL migration into the completed monitoring wells.
One strategy for minimizing DNAPL short circuiting during drilling is to terminate boring at the first evidence of a DNAPL. However, this method requires the ability
to obtain accurate evidence of a DNAPL by visual inspection, vapor monitoring, or
field-testing of borings during drilling. However, such techniques may not be sufficiently
accurate to eliminate the potential for drilling through DNAPL zones. Another drilling
strategy that may be used to minimize DNAPL short circuiting is to terminate borings
424
Chap. 13
at the top of geological barriers such as at the top of a major aquitard stratum. In this
way, any short circuiting of DNAPL that did occur would be halted at the barrier stratum. However, the application of this strategy is limited to those sites where previous
drilling has determined the presence and location of the major local geological barriers;
this information does not exist at many sites.
A common drilling strategy employed to minimize DNAPL short circuiting is the
use of telescoped casing to isolate a specific depth interval from the zone above. This
strategy involves the setting of casing and the advancing of progressively smaller diameter casing to the required depth of the monitoring interval (see Figure 13.11). This
strategy will be effective in a multi-layer system of aquifers and aquitards where the
end of each casing segment can be sealed in an aquitard stratum. In this manner, any
mobile DNAPL that exists in upper aquifers cannot move down through the open boring
or along the sand pack into deeper aquifers. (In situations when aquitard strata cannot
be defined, the depths of the casing intervals must be selected arbitrarily.) This drilling
strategy is slow and more costly than conventional drilling. There is also a practical limit
to the number of casing segments that can be telescoped within a single boring, and that
limit will depend on the drilling rig employed and the required diameter of the mon-
Upper
Aquifer
Lower
Aquifer
FirstStage Grout
m,
FinalStage Grout
Second Stage Grout
.. ..
~
FilterPack
, -, ,
~
Figure 13.11 Illustration of the use of multiple casings for drilling to prevent short
circuiting of DNAPL in areas of suspected DNAPL source zones. This method is most
suited to sites were aquitard units can be accurately defined. a. Extend large diameter
boring into top of aquitard unit. b. Set and grout large diameter casing into top of aquitard.
c. Extend smaller diameter boring through grouted casing and through the formation into
the next aquitard unit. d. Set and grout casing into top of aquitard. e. Extend srnaller
diameter boring through grouted casing and through the formation into the aquifer unit.
f. Complete monitoring well.
Sec. 13.3
425
itoring interval. Such installatiOif seldom employ more than three or four casing seg"
ments.
The drilling strategy described above can be modified in fractured rock with the
use of pressure grouting of the boring at sequential intervals until the required depth of
the monitoring interval is reached (see Figure 13.12). In this manner, mobile DNAPL
that exists in fractures in the upper portions of the rock cannot enter the boring and
migrate downward along the open boring.
13.3.4.1 Test pits
Test pits in overburden materials can provide valuable information that cannot be readily
obtained by means of borings. Test pits can excavated using a backhoe or a power shovel,
and can be dug to depths of about 5 to 8 m. The depth of test pits will be limited by
the equipment used, sidewall stability, and worker safety regulations. Test pits afford
a
.........
,.,.
........
"-r/"
.........
r/"r/"
,.,.
.......
.......
.......
,.fI'..
.........
"r/"
.................
............
............
""r/""
..-:j.-:j:'j."
............
.
r/"""'"
,.,.,.,.
,.,."..,.
,.,.,.,.
. ..
Fractured Bedrock
~ FirstStage Grout
~
~
Pressure Grout
~ Second Stage Grout
Figure 13.12 Illustration of the use of sequential grouting for drilling to prevent short
circuiting of DNAPL in areas of suspected DNAPL source zones in fractured rock. a. Set
and grout large diameter casing to top of rock. b. Where upper part of bedrock is highly
weathered and fractured, set and grout second casing through to intact rock. c. Extend
smaller diameter boring through grouted casing and through the formation to the target
depth. d. Pressure grout the open interval in the rock to seal the fractures in the formation
around the boring. e. Ream out the grouted interval and extend the boring to the next
target depth. f. Pressure grout the lower interval to seal the fractures is this interval. Steps
e and f are repeated to the final depth.
426
Chap. 13
the opportunity for direct visual examination of stratigraphy and geological structure. In
particular, vertical fractures in clayey strata, and the lateral continuity of fine-grained
layers can be identified in test pits. Such features are difficult to identify in borings. Test
pits also allow the visual examination of DNAPL distribution and the relationship between
the geological structure and the DNAPL distribution. Because DNAPL distribution and
geological structure may be visible in the test pits in at least two dimensions, soil samples
selected for chemical analysis have the potential to be much more representative than
those collected from borings.

Monitoring wells and sampling points can be constructed using drive-point devices.
Drive-point wells consist of an intake fitted with a conical point at the end. A tubing
or pipe extends from the intake to the ground surface. The drive-point assembly is
pushed into the ground on drill rods to the desired depth using a hydraulic ram or
vibratory hammer. Drive-point installations are suitable for use in most sand, silt and
clay strata to depths of about 30 to 40 m under ideal conditions. The depth of penetration
will be limited by the soil density and the presence of gravel, cobbles, or boulders.
Most drive-point devices have relatively short intakes 0.5 m) and the likelihood of
intersecting and recovering DNAPL directly is probably low unless DNAPL pools are
thick. Drive-point installations are used most commonly as permanent installations in
the vadose zone for soil-gas sampling, and in the groundwater zone for groundwater
sampling. Although drive-point installations avoid short-circuiting of DNAPL along an
open borehole, DNAPL short circuiting may still occur along the sides of the drill rods
or casing.
Several types of removable drive-point devices for obtaining soil-gas or groundwater samples are commercially available (Edge and Cordry, 1989). These devices are
pushed into the subsurface, and soil gas or groundwater samples are collected at a specific depth. Devices such as the Hydrol-unch'v are removed after collection of a sample
from a single depth. Other devices such as the BAT sampler allow the collection of
sequential samples at increasing depth within the same hole; following completion of
the sampling, the device is removed and the hole may be grouted. Comparable sampling
devices can be used in combination with a cone penetrometer (Litherland et al., 1985;
Smolley and Kappmeyer, 1991; Chiang et al., 1992). Cone penetrometers have been used
for many years in geotechnical investigations to interpret continuous vertical profiles of
soil type, stratigraphy, and geotechnical properties. A cone penetrometer consists of a
conical point of standardized size and shape that is pushed into the ground at a steady
rate using a hydraulic ram. The cone has electronic sensors to measure the resistance
to penetration at the tip and along the sides of the cone. The magnitude of the tip resistance and side friction resistance and the ratio of the resistances are used to assess
the soil type. For example, sands will exhibit high tip resistance and low side friction,
whereas clays will exhibit low tip resistance and high side friction. A combination cone
penetrometer/sampler can provide a range of information. As for permanent drive-point
installations, the likelihood of direct intersection and recovery of DNAPL is low. The in-
Sec. 13.3
427
formation from a cone penetrometer is very valuable to interpreting the geologic structure
that may control DNAPL migration.
Drive-point devices have an advantage over conventional monitoring wells in that
they are less costly and do not generate contaminated drill cuttings or drill water that
require special handling and disposal. In addition, drive-point devices can be installed
manually, or with light-weight equipment to allow access in restricted spaces (e.g., inside
buildings). However, the use of drive-point devices does not eliminate the potential for
short circuiting of DNAPL when working directly within DNAPL zones. When permanent drive-point devices are used, there remains the potential for downward migration
along the contact between the casing and the formation as shown in Figure 13.10. In
the case of removable drive-point devices, there is the additional potential for DNAPL
short circuiting down the open hole after the device is withdrawn, even if the hole is
subsequently grouted.
The use of drive-point devices is probably more useful in the detailed delineation
of dissolved contaminant plumes emitted from DNAPL source zones than it is in the
investigation of DNAPL source zones directly. Drive-point devices that allow the collection of sequential samples with increasing depth can be used to compile detailed vertical
profiles of dissolved contaminant concentrations. Information on the vertical distribution of dissolved concentrations can provide insight regarding the location and nature of
DNAPL zones below the water table. Examples of the use of vertical profiles of dissolved
concentrations in the interpretation of DNAPL zones are described in Section 13.6. The
results of such one-time vertical concentration orofiles can then be used to determine
optimum sampling intervals if permanent monitor. '~ wells are required.
13.3.4.3 Geophysics
Electromagnetic (EM) resistivity logs can be run in conjunction with natural gamma logs
to aid interpretation of the stratigraphy and target zones in which DNAPL accumulation
were mostly likely to occur. However, although surface and borehole geophysical methods such as seismic refraction and reflection, EM resistivity, and ground penetrating radar
(GPR) can provide valuable information on subsurface geological structure, the ability
of geophysics to detect DNAPLs in the subsurface is still unclear.
Field experiments involving the release of peE into a 3 m thick sand aquifer showed
that borehole electrical methods, borehole dielectric permittivity, and surface GPR methods were able to detect the migration and final distribution of the DNAPL (Schneider
and Greenhouse, 1992; Brewster et al., 1992). These electrical methods were successful
because most DNAPLs have an electrical conductivity much lower than geologic media.
However, these experiments were conducted under ideal conditions at shallow depth in
a relatively homogenous sand, and with background measurements having been taken
at the exact same locations before the release. Schneider and Greenhouse (1992) and
Brewster et al. (1992) concluded that while these methods were effective in assessing
changes in the DNAPL distribution, they may not have been able to detect the presence
and corresponding distribution of the DNAPL without the benefit of the exact background
measurements. This is especially the case because the DNAPL distribution was greatly
428
Chap. 13
affected by horizontal layers which are only a few centimeters thick, with lateral continuities of as little as a meter as less. Research on the use of these methods is continuing.
For attempts at monitoring DNAPL occurrence or potential DNAPL migration,
borehole EM resistivity logging tools (e.g., the Geonics EM39) are commercially available and may be applicable. These tools focus EM energy outside the radius of a typical
borehole (> 10 to 150 em) to measure the formation conductivity outside the borehole.
Measurements can be taken within an open boring immediately following drilling, within
a PVC monitoring well, or within a sealed PVC casing. Measurements can be taken
within installations as small as 5 ern in diameter. A possible application of this method
might involve the identification of DNAPL layers with high pore saturations that have
accumulated above fine-grained beds in aquifers. Surveys could be performed in existing
monitoring wells, yielding data from the entire vertical depth, not just the screened intervals of the wells. Zones of DNAPL accumulation would exhibit electrical conductivities
that are higher than in the underlying aquifer materials. The minimum vertical resolution
of such logs is probably about 0.5 m. As caveats, we note that: 1) DNAPL accumulations
thinner than 0.5 m will probably not be detectable; 2) when the change in the electrical
conductivity response of a zone of DNAPL accumulation is small relative to the natural
variabilities in electrical conductivity, DNAPL accumulations will be difficult to identify;
3) zones of residual DNAPL will not likely exhibit electrical conductivities sufficiently
different from that of the host medium to be detected.
Experimental work is underway to examine the combination of cone penetrometer
testing with in situ resistivity measurements for the detection of DNAPLs (Westinghouse Savannah River Company, 1992). The objective of this method is to obtain a
profile of the geologic conditions together with a resistivity profile. As with borehole
EM logging, DNAPL accumulations should exhibit low conductivities. Field testing of
this technique at a site expected to have large quantities of DNAPL in the subsurface has not demonstrated that DNAPL accumulations can be identified in this way.
However, it is not certain whether the technique is not effective, or just that no significant DNAPL accumulations were present at the tested locations. Further experimental work is underway at this site to assess the use of a cone penetrometer probe with
a fiber-optic Raman spectroscopy probe for direct in situ analysis of the presence of
DNAPL.
In the field of soil science, neutron logging is employed to measure in situ moisture
contents in soils, and in the petroleum exploration field to determine the water-filled
porosity in reservoirs. A neutron logging tool is comprised of a fast neutron source
(typically americium-beryllium) with a neutron and/or gamma detector. Fast neutrons
emitted from the source enter the formation and are slowed (moderated) by the hydrogen
in the water. The slowed or thermal neutrons are then susceptible to capture by other
atomic nuclei in the formation with the consequent emission of gamma rays. The detector
measures either the neutrons or the gamma rays. The measured thermal neutron counts
increase with increasing water content of the formation. Since chlorine has a high neutron
capture cross section, it reduces the neutron counts; it also causes gamma emissions.
Therefore, in petroleum reservoir logging, corrections must be made to neutron logs
to account for the chloride in reservoir brines. Because many DNAPLs are chlorinated
Sec. 13.4
Soil Gas Surveys
429
compounds, many DNAPLs are effective in capturing thermal neutrons. Thus, a zone
in which a DNAPL occupies a substantial proportion of the pore space will exhibit a
neutron count that is lower than the surrounding formation where the pore space is filled
with water or air. Although a zone of low neutron count can also be due lower porosity
or lower moisture content, the presence of a DNAPL can lower neutron counts beyond
the range of likely variations in porosity or moisture content. The peE release field
experiments examined by Schneider and Greenhouse (1992) and Brewster et at. (1992)
have illustrated that borehole neutron logging can be an effective method for the direct
detection of DNAPLs comprised of chlorinated compounds.
13.4 SOIL GAS SURVEYS
13.4.1 Description
Soil gas surveys to delineate source zone areas and dissolved-phase plumes in the groundwater zone are now a common component of site investigations. The general principles
of soil gas surveys and several case studies are described by Marrin and Thompson
(1987), and by Marrin and Kerfoot (1988). Discussions of some of the limitations of this
method are given by Marrin (1988). Soil gas surveys involve the collection of soil gas
samples from the shallow subsurface at locations distributed across a site. Grab sampling
of soil gas may be accomplished from permanent or removable probes installed below
ground surface. Surveys typically employ probes installed at a consistent depth of from
0.3 to 3 m below ground surface, but in some cases also employ clusters of probes
installed at various depths through the vadose zone. Soil gas samples are captured in
suitable containers, or passed through sorbent cartridges. Samples can be analyzed using
on-site equipment, or in off-site analytical facilities. Soil gas surveys that employ both
laterally-distributed and vertically-distributed sampling locations are those most likely to
provide useful information for the identification of DNAPL in the vadose zone. However, recent investigations of the soil gas from controlled field experiments conducted by
Rivett (1995) indicate that soil gas surveys are unlikely to be capable of identifying the
location of dissolved plumes that exist more than a meter or two below the water table.
Soil gas surveys are expected to reflect only the effects of the migration of vapor
away from source zones and the effect upward vapor migration from very shallow dissolved-phase plumes at the vadose zone/groundwater zone interface. If residual zones
or accumulations of volatile DNAPL are present in the vadose zone, they will emit
vapors at saturated vapor concentrations. The finding of vapor concentrations that approach expected saturated vapor concentrations (as based on the compounds present, the
composition of the DNAPL, and the temperature) would yield convincing evidence that
DNAPL is present in the vadose zone. Evidence of DNAPL deep in the vadose zone
would be valuable in the further assessment of the potential for DNAPL to have penetrated into the groundwater zone. However, there are many processes that will act to
reduce vapor concentrations in the vadose zone so that, in most circumstances, vapor
concentrations close to saturation should only be expected directly within or very close
to DNAPL source zones. Given the spacing of sample locations in most soil gas surveys
430
Chap. 13
and the potential spatial variabilities in DNAPL distributions, a lack of saturated-level

vapor concentrations in soil gas cannot be interpreted to mean that DNAPL is not present.
The pattern of vapor concentrations emanating from a DNAPL source zone will depend
on many factors related to the nature of the DNAPL source zone and the vadose zone.
In most circumstances, diffusion will be the dominant migration process and will cause
vapors to migrate outward in all directions and create a halo of lower vapor concentrations
around the DNAPL zone. Many chlorinated solvents have vapor pressures sufficiently
high to cause the contaminated air to have a density much greater than the surrounding
uncontaminated air. This high relative vapor density may induce downward advection of
vapors that is much more rapid than downward migration by diffusion. Advective sinking
of vapors can be important in media having hydraulic conductivities greater than about
10- 2 cm/s. In media with lower hydraulic conductivities, diffusion will be the dominant
migration process.
The present understanding of the size of the vapor halo that may form around a
DNAPL source zone and the persistence of a DNAPL source in the vadose zone under
different circumstances is based on what is currently a limited number of experimental
and computer modeling studies. Under favorable conditions, vapor migration through the
vadose zone may be rapid, and the size of the vapor halo around a DNAPL source zone
may be relatively large. Controlled field experiments were conducted with source zones
of residual TCE in the vadose zone of a sand aquifer (Hughes, 1991; Hughes et al.,
1992; Mendoza et al., 1992). Lateral profiles of TCE vapor concentrations vs. distance
from the source zone are shown in Figures 13.13 and 13.14. Vapor concentrations that
approach the saturated vapor concentration (88,000 ppmV at the ambient temperature)
are found only close to the source zone. The concentrations decline exponentially with
distance, suggesting a transient diffusion profile. Detectable (> I0 ppmV) concentrations
were found at a distance of about 7 m from the source zone after only 18 days, though
it should be recognized that the rate of lateral diffusion outward from a DNAPL source
zone will decline with time because the concentration gradient around the source zone
will decline with time. We also note that the above example reflects the most favorable
conditions for rapid vapor migration: the total porosity was relatively high (0.4), the
moisture content was relatively low (25% saturation), and the sorption coefficient for
TCE on the aquifer material was low. Higher moisture contents will substantially lower
the effective diffusion coefficient for organic vapors (see Figure 13.15), and will reduce
the rate and extent of vapor migration.
The effects of variability of the spatial distribution of the DNAPL source zone,
spatial variability in the properties of the vadose zone, and scenarios for vapor migration through low permeability/high moisture content media have not yet been assessed,
either experimentally or by computer modelling. Although there have been attempts to
relate measured soil gas concentrations to measured soil concentrations at actual field
sites (i.e., Siebenmann, 1993; Sohn et al., 1993; Einberger et al., 1993), there is considerable scatter in most of the data sets and quantitative correlations are poor. However,
Sec. 13.4
431
Soil Gas Surveys
TeE concentrations in soil gas (ppmv)

Cover
Source
Zone
+
2
to
Depth
(m)
10
Radial Distance (m)

Figure 13.13 Example of lateral transport of TCE vapor in a sandy aquifer resulting
from a experimental placement of a DNAPL source in the vadose zone. This vapor
distribution developed 18 days after placement of a source zone that contained 42 L of
TCE. Adapted with permission from Hughes et al. (1991); described further in Hughes
et al. (1992).
90,000
80,000
>' 70,000
E
~ 60,000
(/)
co 50,000
c.!'
~ 40,000
c
UJ
30,000
20,000
[~
1\
\
\
\
\
h
"-r\.
<,
10,000
..... t---n..
a
0.0
1.0
2.0
3.0
L..I
"'-C
4.0
5.0
1IL..I"
6.0
Radial Distance from Centre of Source (m)

Example of the decline in TeE vapor concentrations with distance from
the DNAPL source in the vadose zone. This profile corresponds to the section line shown
in Figure 13.13. Data from Hughes et al. (1992).
Figure 13.14
7.0
432

~
1E-01
..
C
.~ 1E-02
:E
Q)
----
r--.< ,
<,
r-. -,
o
c
'gj
1E-03
:1=
gj
-,
1\
Q)
s:
~
1E-04
Q)
>
t5
L.U
1E-05
0.00
Chap. 13
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
Water-filled Porosity
Figure 13.15 Relationship between effective air-phase diffusion coefficient (De) and the
percent water saturation in soil. The effective air-phase diffusion coefficient is calculated
according to De = Do r , where the tortuosity r = eJ33 Ie? is as defined by Karimi et al.
(1987); eu is the air-filled porosity, and ef is the total porosity.
with regard to the delineation of DNAPL sources in the vadose zone, we can conclude
that: I) the finding of vapor concentrations that approach saturated vapor concentration
for the compound of interest is evidence that DNAPL is likely to be present near the
sampling point; and 2) because DNAPL source zones may be small in relation to the
lateral and vertical spacing of sampling points and because of the rapid decline in vapor
concentrations that can result from vapor transport by diffusion, the absence of saturated
vapor concentrations is not necessarily evidence that DNAPL is absent. Comparable
conclusions with regard to groundwater concentrations are discussed in later sections.
13.5 CHEMICAL ANALYSIS OF SOIL AND ROCK
13.5.1 Soil Samples
Chemical analyses of soil samples can be useful in determining the presence and distribution of DNAPL in both the vadose and groundwater zones. However, such analyses
may be ineffective unless the soil sampling and analysis methods used are suitable for
the specific task at hand. The critical issues related to soil sample collection are: I) the
lateral and vertical spacings of the samples; 2) cross-contamination during sample collection; and 3) the selection of subsample portions for analysis. The critical factor related
to the chemical analysis of soil samples is the loss of volatiles during sample collection,
handling, and storage.
Sec. 13.5
433
As has been discussed above, the migration pathways and distribution of DNAPLs
in the subsurface are strongly influenced by variations in geologic structure at both
small and large scales. Consequently, in most circumstances, the spatial distribution of
a DNAPL will be highly irregular. In stratified unconsolidated deposits, much of the
DNAPL mass may be present as thin horizontal accumulations on less permeable strata.
Depending on the geologic variability, these accumulations may be as thin as a few centimeters (see Figures 13.4 and 13.5). Reliable identification of such accumulations can
be made only through the collection of vertically-continuous soil samples. Therefore,
the common practice of recovering of a single 2 ft. long split-spoon sample for every
5 ft. of drilling will not likely be effective in understanding the distribution of DNAPLs
in the subsurface.
Vertically-continuous soil samples can be collected in a number of ways. Conventional 2 ft. long split-spoon samples or Shelby-tube samples can be collected on a
continuous basis, though such sampling can be laborious and costly. Alternatively, there
are a variety of soil sampling devices available that will allow recovery of continuous
cores that are 5 ft. in length or longer. Some of these devices allow sample collection
while advancing a boring with hollow-stem augers (Central Mine Equipment Co.; George
E. Failing Co.). Other devices are thin-tube piston-type samplers that are advanced separately, or ahead of drill casing or augers (Zapico et al., 1987; Starr and Ingleton, 1992).
The recovery of samples of cohesionless sand from below the water table is generally
unreliable with devices that rely on mechanical traps or catchers to hold the soil in the
sampling barrel. This can be overcome with the use of piston-type samplers.
Special problems may arise in sampling soils containing mobile DNAPL. During
sample recovery, it may be possible for mobile DNAPL from a specific layer or accumulation to cross-contaminate other portions of the sample. An example of this form
of cross-contamination is shown in Figure 13.16. In this case, a thin-walled aluminum
core tube was used to recover soil samples from an experimental spill of PCE. Because
the PCE was dyed red, it is clear that DNAPL migrated along the interface between the
sample and the core barrel. The potential for this type of cross-contamination is particularly high for cases where the DNAPL has a high density and low viscosity, as is
generally the case for chlorinated solvents. The effect of such cross-contamination on the
results of soil analyses can be reduced by subsampling the interior of a soil core. The
elimination of cross-contamination and the effective recovery of DNAPL pore fluids can
also be achieved using cryogenic core samplers that freeze the sample in place in the
core tube. Dumford et al. (1991) demonstrated the effectiveness of such a sampler for
the recovery of an LNAPL product at the water table, and it would likely be similarly
effective for DNAPLs residing in potentially mobile accumulations or pools.
The volume of soil needed for a chemical analysis is generally much smaller than
the volume of soil recovered during sampling. Typical soil analysis methods require only
5 to 20 g of soil, though some methods can accommodate up to 250 g of soil. A 24 in.
long split-spoon sampler contains 2,100 to 2,500 g of soil, and typical sampling programs
would only yield one soil sample per split-spoon. Because the spatial distribution of
a DNAPL can vary at the centimeter scale, even within a single split-spoon sample,
DNAPL may be present in some portions but not in others. Consequently, the results of
434
Chap. 13
Figure 13.16 Example of cross-contamination of soil core resulting from migration of

DNAPL PCE along the inside of a core tube. PCE is dyed red using Sudan IV dye.
(Photograph by Stan Feenstra.)
the chemical analysis will depend strongly on where the subsample(s) are collected, and
it becomes difficult to decide where the subsample(s) should be taken to best reflect the
chemical concentrations in the overall section of core.
If one of the goals is to determine the presence/absence of DNAPL, subsampling
should be directed at the most contaminated zones, and subsamples of the soil cores
need not be large. If the goal is to attempt to estimate the chemical mass in the subsurface, the subsampling approach needs to be more complex. If DNAPL is present in
localized zones in the cores, subsampling of only the DNAPL zones will result in serious overestimation of the chemical mass. At the same time, if the DNAPL zones are
missed by the subsampling, the chemical mass in the core will be seriously underestimated. Because of the potential centimeter scale of variability in DNAPL distributions,
subsampling from multiple locations within a core followed by compositing may be then
be advantageous. However, composite subsampling requires more sample handling and
therefore risks volatilization losses. Another approach might be to collect and analyze
relatively large subsamples of the core in order to "average" the effects of small-scale
variabilities. These issues are discussed further in the next section.
Sec. 13.5
435

The results of chemical analyses of soil samples for VOCs such as chlorinated solvents
can be greatly affected by the methods used for sample handling, preservation, and
analysis. Conventional methods suitable for the determination of inorganic chemicals
and semi-volatile organic compounds are not suitable for VOCs because of the potential
for volatilization losses from the sample before the analysis can be completed.
Methods for the sampling and determination of VOCs in soil have involved the
following steps: 1) collection of soil cores using split-spoon or comparable core tubes;
2) examination of the samples in the field; 3) transfer of a small subsample to a 40 mL vial
or larger jar; 4) storage at 4 C during shipment and storage before analysis; 5) opening
of the jar in the laboratory to remove a small portion for analysis; and lastly 6) extraction
of the VOCs with methanol for high-concentration samples, or with water followed by
purge-and-trap for low-concentration samples. Direct analysis of the headspace (air) in
contact with the sample may be possible in either the original sample vial, or following
transfer of the sample to a suitable container in the laboratory.
Voice and Kolb (1993) compared purge-and-trap, direct headspace, and methanol
extraction in the laboratory under controlled conditions. Samples of three different natural
soils were spiked with known amounts of various VOCs and analyzed by the three
methods. The results for the chlorinated VOCs are given in Table 13.5. All the methods
were found to underestimate the actual soil concentrations, i.e. the recoveries were less
than 100%. The direct headspace analyses were more effective than the purge-and-trap
analyses. The methanol extraction method was not consistent, being as effective as direct
headspace for some compounds, but less effective for other compounds.
Although there are differences among the results obtained using different soil analysis methods, these differences are relatively small compared to the differences that can
result from sample handling. Siegrist and Jenssen (1990) studied the effect of different
sample handling methods under controlled laboratory conditions. A spiked soil column
was sampled simultaneously using various handling methods. Of particular interest is
their comparison of the effect of using a scoop or a tube to obtain a subsample from
the soil and transfer it to the sample container, and whether the sample was then immersed in methanol. In all cases, the samples were contained in 40 mL vials. For scoop
TABLE 13.5 Laboratory Comparison of Percent Recoveries by Purge-and-Trap,

Direct Headspace, and Methanol Extraction Methods for the Determination of
VOCs in Soil. (Data are from Voice and Kolb (1993). Percent recoveries are
averages of 8 to 10 replicate spikes each on 1 to 3 different soil types.)
Purge and trap
Direct headspace
Methanol
extraction
Dichloromethane
35%
71%
44%
I,1,1-Trichloroethane
25%
72%
57%
Trichloroethylene
35%
73%
61%
Tetrachloroethylene
33%
76%
72%
Compound
436
Chap. 13
sampling, the soil was collected from the soil column with a stainless steel scoop and
transferred to the vial; for the tube sampling, a small stainless steel tube was used to
remove soil from the column and the subsample was extruded directly into the vial. The
vials were later re-opened for removal of a portion for analysis by isopropanol-pentane
extraction. For the methanol immersion, tube sampling was used to extrude the subsample
directly into methanol in the vial. The results for the chlorinated VOCs using these methods are given in Table 13.6. The highest, and presumably most accurate, concentrations
were obtained when the methanol immersion method was used. Use of scoop sampling
generally yielded the lowest concentrations. The concentrations obtained using methanol
immersion range from 1.2 to 6.7 times higher than those obtained using scoop sampling.
TABLE 13.6 Laboratory Comparison of Soil Concentrations in mg/kg after
Sample Handling by Scoop Sampling, Tube Sampling, and Methanol Immersion.
(Data are from Seigrist and Jenssen (1990). Results are averages of duplicate
samples.)
Scoop sampling
Tube sampling
Methanol
immersion
Dichloromethane
6.1
4.9
7.2
0.28
0.36
1.9
Trichloroethylene
0.42
0.55
2.3
Chlorobenzene
0.58
0.69
0.76
Compound
Urban et al. (1989) provided a field comparison of different soil sample handling
methods. Portions of the same split-spoon sample were analyzed by field methanol immersion/extraction, and by scoop sampling followed by analysis using either laboratory
methanol extraction or purge-and-trap depending on the concentration level. The results
of these analyses are given in Table 13.7. The TCE concentrations obtained using the
field methanol immersion/extraction method were found to be consistently the highest,
ranging from 2.5 to 45 times higher than those obtained using the other methods. Unlike the controlled laboratory studies described above, true duplicate samples could not
be collected in the field. Consequently, some of the differences between the sampling
methods may have been due to true differences among the samples.
At the present time, there is no recognized standard method for the handling and
analysis of soil samples for VOCs (Lewis et al., 1991). The most reliable methods
are those that minimize the loss of volatiles during handling. Field methanol immersion/extraction has been shown to be effective in this regard. The field methanol method
will be particularly useful in the collection and analysis of large samples (> 100 g), or
composite samples comprised of multiple subsamples from the same core. Immersion in
methanol provides a mechanism for the homogenization of such samples that is more
effective than mechanical mixing, and also reduces volatilization losses.
Field methanol immersion/extraction has the disadvantage of having a relatively
high detection limit ("'0.1 mg/kg) compared to: I) direct headspace or purge-and-trap
Sec. 13.5
437

TABLE 13.7 Field Comparison of Methanol Field Immersion/Extraction
Methods with Methanol Laboratory Extraction and Purge-and-Trap for the
Determination of TCE in Soil. (All values are in mg/kg. Data are from Urban et
a/. (1989). "J" suffix designates samples in which the TCE that was detected
below the limit of quantification.)
Sample code
A. 2.0
feet to 4.0 feet
6.0 feet to 7.8 feet

B. 1.95 feet to 4.0 feet
D. 1.0 feet
E. 3.0 feet
to 3.0 feet
to 5.0 feet
I. 1.5 feet to 3.5 feet
M. 2.0 feet to 4.0 feet
H. 2.0 feet to 3.5 feet
Field methanol
immersion!
extraction
124
79.9
212
5.31
0.81
0.66
Laboratory
methanol
extraction
Purge and
trap
2.7
5.5
47.2
0.27
0.32
0.015 J
0.28
0.035
0.096 J
0.065 J
0.34
0.11 J
0.015 J
< 0.025
(~0.01 mg/kg); and 2) the soil clean-up targets defined for numerous sites. This limitation is not significant when diagnosing the presence of DNAPLs, since much higher
concentrations are then of interest. However, large soil sampling programs or the collection of large volume samples can require substantial quantities of methanol. Use of the
field methanol method will consequently require special precautions by field personnel
to protect against toxicity and fire hazard both before and after extraction; it can also
create a disposal problem for the contaminated methanol.
An alternative to the field methanol method would be any technique that does
not require any handling or exposure of the sample to the atmosphere after it has been
collected. For example, direct headspace analyses can be performed on soil samples
collected in 40 mL vials with septum caps. (In this case, the only avenue for loss of
volatiles is diffusion through the Teflon/silicone septa during storage.) Similarly, modified
caps can be used on 40 mL vials so that purge-and-trap analysis can be performed without
having to open the vial in the laboratory. The possible loss of VOCs from soil samples
must be considered when interpreting historic soil analysis data.

13.5.2.1 General
Sampling of soil from the vadose and groundwater zones for quantitative chemical determinations has been a common procedure during site investigations for many years. The
analysis of soil samples provides a measure of the total chemical mass in the soil. The
results of soil analyses can be used in a quantitative manner to assess the possible presence of residual DNAPL contamination when there is no visual evidence of the presence
of a DNAPL. Feenstra et al. (1991) describe a method for assessing the possible presence
438
Chap. 13
of DNAPL in soil samples based on the results of chemical and physical analyses, and
the fundamental principles of chemical partitioning in soil.

This method is a simple semi-quantitative tool to allow the investigators of chemical spill
and waste disposal sites to assess whether soil chemical analyses indicate the presence of
residual DNAPL in the subsurface. Information that is needed for this method includes:
1) total chemical concentrations; 2) moisture content; 3) porosity; 4) sorption parameters
for the chemicals of interest; and 5) physical and chemical properties of the chemicals
of interest.
The chemical analysis of a soil sample for organic compounds can proceed by
extraction of the sample using water leaching, solvent leaching, or heating. The analysis
generally provides a measure of the total amount of the chemical of interest in the sample,
expressed typically as the mass of chemical per unit dry weight of soil. The result includes
the chemical that is: 1) dissolved in the pore water of the sample; 2) sorbed on the soil
solids; 3) in the soil gas; and 4) in any DNAPL phase. The analysis cannot indicate
directly how the chemical is distributed among these phases, nor whether any DNAPL
phase is present. The method can be used to calculate whether a DNAPL phase is likely
present in the sample.
The hypothetical pore-water concentration of the constituent of interest is calculated from the measured total soil concentration by assuming equilibrium chemical
partitioning between the solid phase, the pore water, and the soil gas, and assuming that
no DNAPL phase is present. This pore-water concentration C w (mg/L, or equivalently
f.J.,g/cm 3 ) can be expressed in terms of the total soil concentration C, (f.J.,g/g dry weight) as
c; =
CtPb
KdPb + ew + Hcea
(13.3)
where: Pb = dry bulk density of the soil sample (g/cm"); ew

water-filled porosity
(volume fraction); ea = air-filled porosity (volume fraction); K d = partition coefficient
between pore water and soil solids for the compound and temperature of interest (cnr'Zg);
and He = dimensionless Henry's Gas Law constant for the compound and temperature
of interest.
If no DNAPL is present, there is a maximum amount of chemical which can be
contained in a sample of soil that contains soil solids, pore water, and soil gas. This
maximum total soil concentration is determined by: 1) the solubility concentration Sw of
the chemical in water; 2) the concentration in the soil gas that would be in equilibrium
with Sw; and 3) the concentration sorbed to the solids that would be in equilibrium with
Sw. The only way that the chemical in the form of a DNAPL phase can be present in a
sample would be for the value of C w as calculated by Eq.(13.3) to be greater than Sw.
In that case, we know that at equilibrium, some DNAPL phase is present, and for the
true aqueous concentration we have that C w = Sw.
The parameters Ph, ew , and ea can be determined from the standard measurements
of soil density and water content that are often performed. Even when these parameters
must be estimated rather than measured, the possible value ranges are relatively narrow,
Sec. 13.5
439
and the uncertainty imparted to a value of C w calculated by Eq.(l3.3) will be correspondingly small. If possible, measured values of K d should be obtained from batch or column
experiments. However, in many situations, it will be necessary to estimate K d values using K oc and foe values (see Chapter 8). Although factors such as differences in the nature
of the soil organic material, sorption on mineral surfaces, intergranular diffusion, very
high dissolved chemical concentrations, and the presence of dissolved organic matter or
cos01vents in the ground water may cause K d values so obtained to be in error, the K oe
approach still provides the only general means of estimating Kd values without direct
measurement. For most compounds of interest here, He values can be obtained from published data (e.g., Montgomery and Welkom, 1990; Howard, 1989; Howard, 1990; and Appendix I in this book). In general, the largest uncertainty in the C w values calculated using
Eq.(l3.3) will be due to uncertainty in K d . Figure 13.17 presents some example calculations for the proportion of TCE (as a single component) in the gas, dissolved, sorbed, and
DNAPL phases for a typical sandy soil. At soil concentrations greater than several thousand mg/kg, these results show that the majority of the TCE is found in the DNAPL phase.
1.0 -,--------,-----,-- -.----,---~
+-------t----+-----\Jf---+-------
10
100
1,000
--DNAPL
---Sorbed
- - - - - - . Dissolved
- - - Vapor
10,000
TCE Concentration in Soil (mg/kg)
Figure 13.17 Relative proportion of TCE in soil gas, water, and soil solid phases
for a typical sandy soil. Calculated based on the partitioning relationships described in
Feenstra et al. (1991) assuming a total porosity e, of 0.4, a water-filled porosity ew of
0.1, a dimensionless Henry's Gas Law constant of 0.27, and a Kd of 0.12.
For a multicomponent DNAPL, the value of C w calculated according to Eq.(l3.3)

for one of the components of interest must be compared not to the pure phase solubility
Sw of that component, but rather to the estimated effective solubility of that component
from the mixture. If the calculated C w value exceeds that estimated solubility, then it can
be concluded that the multicomponent DNAPL phase could be present. If a sample of
the multicomponent DNAPL is available, then the water solubility of the component of
interest can be measured directly. Alternatively, in a manner that is totally analogous to
Eq.( 13.2), the effective solubility of one component of a DNAPL requires information on
440
Chap. 13
the chemical composition of the DNAPL and the use of Raoult's Law (see Chapter 7).
The chemical composition of the DNAPL can be determined by analysis of samples of
DNAPL recovered from the subsurface; in some cases, there may be suitable information
on the chemical composition of the product released at the site. Unfortunately, we note
that at many waste disposal and chemical spill sites: 1) the DNAPL contains a substantial
proportion of organic compounds which cannot be readily identified and quantified by
standard analytical methods; or 2) samples of the DNAPL are simply not available for
analysis. In these situations, assumptions must be made regarding the composition of the
DNAPL. Or, it may be possible to use chemical analysis results for soil samples which
exhibit visual evidence of DNAPL, or are clearly very highly contaminated (thousands
of mg/kg or more): at very high soil concentrations, almost all of the chemical mass will
be in the DNAPL phase.
In some circumstances, it will be useful to have an estimate of the DNAPL-filled
porosity eDNAPL to assess the potential mobility of the DNAPL. Such an estimate can
be made using the equation
e D N APL
CtPb - Se(KdPb + ew + Heea )

=-----------
(13.4)
CDNAPL
where S; is the effective solubility of the compound of interest from the DNAPL phase
3
CDNAPL (ftg/cm ) is the concentration of the component of interest in the DNAPL phase.
(= Sw in the case of a single-component DNAPL), and

Soil samples were collected from borings around an industrial facility in order to assess
the source of TCE contamination in the groundwater beneath the site. Several of these
results are given in Table 13.8. The TCE concentrations in these samples ranged from
210 to 3,100 mg/kg. There was no visual evidence of DNAPL in any of the samples.
The samples were all collected from near or below the water table. The value of Pb was
estimated to be ~1.86 g/cm', the total porosity was estimated to be 0.3, and ea was taken
to be zero. The value of K oe for TeE at ~20 C is 126, and foe was taken to be 0.001.
Using these parameter values and the measured values of Ct> the calculated C w values
were found to range from 730 to 11,000 mglL.
TABLE 13.8 Results of Trichloroethylene Analyses of Soil and Calculation
of Hypothetical Pore-Water Concentrations. (Parameters used for
calculations: ew = 0.3; ea = 0; Pb = 1.86 g/cm ': Koe = 126; foe = 0.001.)
Sample depth
(m)
Measured TCE
levels in soil
(mglkg)
Calculated pore
water cone.
({<gIL)
NAPL
saturation
(% bulk vol.)
0.3
2.13
3,100
11,000
6.10
420
1,500
9.14
210
730
Sec. 13.5
441
TCE was used as a degreasing solvent at this site; no other chemicals were detected
in the soil samples at significant concentrations. It is not known whether the TCE released
to the subsurface was pure solvent, or used solvent which may have contained substantial
proportions of oil and grease compounds. The effective solubility of TCE in a used solvent
mixture at 20C would be somewhat less than the 1,400 mg/L solubility of pure-phase
TCE at 20C.
The calculated pore-water concentration for the sample from 2.13 m depth is
much greater than the pure-phase solubility of TCE of 1,400 mg/L. We can therefore conclude that liquid TCE was present in this sample. The calculated pore-water
concentrations in the samples from 6.10 m and 9.14 m depths are slightly greater
and slightly lower respectively, than the pure-phase solubility of TCE. Given that the
effective solubility of TCE in used solvent would be lower than that of pure-phase
TCE, these results suggest that small quantities of liquid TCE may have also been
present in both of these samples. To summarize, for these samples, any soil concentration greater than ~ 300 mg/kg would be consistent with the presence of residual
liquid TCE.
A generic graph relating C, values in soil to the calculated C w values is given in
Figure 13.18. This graph is based on a typical water-saturated soil with ew = 0.3, and
Ph = 1.86 g/crrr'. Four lines are shown representing four chemicals with relatively low
K d values of 0.0, 0.1, 1.0 and 10, the range expected for the chlorinated solvents in most
soils. For the preceding TCE example, K d was ~O.l. Thus, the results in Table 13.8 can
be obtained from this graph by locating the intersection of a measured C, value with the
appropriate Kd curve to obtain the calculated C w value.
10,000
Kcl 10
Oi
.:.t.
---
0>
.s
c..i
Kcl=1.0
----- Kcl 0.1

- KclO
1,000
c
0
".
UJ
::::>
'0
U
,,'
100
"
(/l
<II
(])
::2:
10
,,'
10
"
100
1000
10000
Calculated Pore Water Cone. (mg/L)

Figure 13.18
Generic graph for the relationship between the pore water and soil con-
centrations for weakly sorbed contaminants in a typical sandy aquifer. Calculated based
on the partitioning relationships described in Feenstra et al. (1991) assuming a total

porosity e, of 0.3 and a water-filled porosity ew of 0.3.
442
Chap. 13
13.5.2.4 Example calculation: contamination with PCBs in

hydrocarbons
Soil samples were collected from borings around disposal pits in which PCB-contaminated
hydrocarbon liquids had been released. Results from some of the soil analyses are
given in Table 13.9. The total PCB concentrations in these samples ranged from 71
to 590 mg/kg, and the PCB compound pattern resembled that of the mixture Aroelor 1248. The samples were all collected from near or below the water table. The
value of Ph was assumed to be ~ 1.86 g/cm", the total porosity was taken to be 0.3,
and eo was taken to be zero. The average value of K oe for Aroclor 1248 reported
by Chou and Griffin (1986) is 54,626, and foe was taken to be 0.003. Using these
parameter values, the calculated C w values were found to range from 430 to
3,600 f,Lg/L.
TABLE 13.9 Results of PCB Analyses of Soil and Calculation of

Hypothetical Pore-Water Concentrations. (Parameters used for calculations:
w = 0.3; a = 0; Pb = 1.86 g/cm': Koe = 54,626; foe = 0.003.)
Measured PCB
levels in soil
(mg/kg)
Calculated pore
water cone.
0.63
0.91
590
3,600
1,900
2.0
310
1.04
170
1,000
0.6
1.77
71
430
0.2
Sample depth
(m)
(J1.glL)
NAPL
saturation
(% bulk vol.)
1.0
PCBs were contaminants in the hydrocarbon liquids released to the subsurface at

this site, and were believed to be present at concentrations up to 5% by weight. The overall chemical composition of the hydrocarbon mixture is not known, so we have assumed
that the PCB mole fraction in the hydrocarbon mixture was also approximately 5%. The
estimated effective solubility of PCB in hydrocarbon mixture would be 5% of the total
pure-phase solubility of 54 f,Lg/L for Aroclor 1248 (Chou and Griffin, 1986), or 2.7 f,Lg/L.
All of the calculated C w values greatly exceed both the pure-phase and estimated effective solubility of Aroclor 1248, indicating the presence of liquid PCB-containing waste
in all of the samples analyzed.
A generic graph relating measured total concentrations in soil to the calculated C w
values is given in Figure 13.19 for chemicals which sorb much more strongly to soil
than those considered in Figure 13.18. This graph is based on a typical water-saturated
soil having a ew of 0.3 and a Ph of 1.86 g/crrr'. Three lines are shown, representing compounds having Kd values of 100, 1,000, and 10,000, the range which would be expected
for typical soils for compounds such as those in the common PCB mixtures, and for
many polycyclic aromatic hydrocarbons (PAHs).
443
Sec. 13.5
10,000
Oi
--5
-'"
OJ
1,000
--
Kd=10,000
---
Kd=1,OOO
- - - --
Kd=100
c
0
0
'0
en
"0
1/
~
1I
1I
100
f---~
f-
1/
~F
1=
1=
:J
Ul
III
Q)
10
V
0.001
0.01
..
.
.
.,.
.
.
.
~
f-
1=
f---
f-
f-
f-
0.1
f---
10
Calculated Pore Water Cone. (mglL)
100
Figure 13.19 Generic graph for the

relationship between pore water and
soil concentrations for strongly sorbed
contaminants in a typical sandy aquifer.
Calculated based on the partitioning
relationships described in Feenstra et al.
(1991) assuming a total porosity f), of 0.3
and a water-filled porosity f)w of 0.3.
The method represented by Eq.(13.3) is intended to be a simple semi-quantitative tool
to allow the investigators of chemical spill and waste disposal sites to assess whether
soil chemical analyses indicate the presence of residual DNAPL. At the present time,
such calculations are the only way to assess the presence of residual DNAPL in soil
samples using simple, conventional soil analysis methods. Moreover, the direct laboratory
confirmation of the presence of low levels of DNAPL by any method is currently difficult.
Conclusions regarding the presence of DNAPL in soil samples are most reliable for
cases when: I) the values of C, are either very high or very low; 2) the soil/pore water Kd
values can be measured for the compounds and soil(s) of interest; 3) the composition and
solubility of the DNAPL phase is well understood. Conclusions regarding the presence of
a DNAPL phase are less reliable when parameters such as K d and the effective solubility
must be estimated.
13.5.3 Direct Determination of DNAPls

As has been described in Section 13.3.1.2 on testing in the field, for qualitative visual
identification, methods such as UV fluorescence, soil/water shake tests, and shake tests
with the addition of a hydrophobic dye can be used (Cohen et al., 1992). These analyses
provide only a determination of the presence or absence of a NAPL. Although a presence/absence determination for a NAPL may be sufficient in many situations, there will
also be situations in which a quantitative measure of the degree of NAPL saturation is
desired. Cary et al. (1991) proposed a method for determining the NAPL content in a soil
by shaking a suspension of soil and water with a strip of porous polyethylene treated
with a light hydrocarbon oil. NAPL that is released from the soil into the suspension
is absorbed onto the polyethylene. The polyethylene strip is weighed before and after
the test to determine the NAPL content. Although this method has not been used with
444
Chap. 13
samples from actual field sites, laboratory tests with sands, loamy sand, and silt loam
soils have indicated recoveries of 82 to 99% for two LNAPLs, a mineral oil and a light
petroleum distillate. This method may therefore also be useful in the quantification of
DNAPL content in soil samples.

Determining the presence of DNAPL in fractured rock or clay is considerably more
complex than in soils. In DNAPL cases involving fractured media, the DNAPL will only
penetrate into the porous matrix of the rock between the fractures if the rock matrix is
itself highly permeable, or the pressures driving the DNAPL migration are very high; the
DNAPL will accumulate in the fractures in zones where capillary forces resist further
migration.
As is discussed in detail in Chapter 12, DNAPL in fractures will slowly dissolve
into the surrounding groundwater, including that in the matrix. Modeling studies discussed in Chapter 12 show that much or all of the original DNAPL mass can dissolve
and diffuse into the matrix in a relatively short period of time when the: 1) DNAPL
component(s) is (are) relatively soluble; 2) DNAPL mass in the fractures is sufficiently
small; and/or 3) the porosity of the rock matrix is sufficiently large. These studies have
shown that in many fractured sedimentary rock and fractured clay systems, DNAPLs
comprised of the common chlorinated solvents can disappear from most or all of the
fractures with a few years; the DNAPL phase can persist longer if comprised of low
solubility components such as PCBs and PAHs.

In a situation when DNAPL persists in the fractures, the challenge is to obtain samples
of rock or soil core without the loss of the DNAPL. In fractured clays, dark-colored,
viscous DNAPL may be visible within fractures or on fracture surfaces in conventional
soil cores (see Figure 13.20). In rock cores on the other hand, although dark-colored
and viscous DNAPL, staining, or oily sheens may be visible on the fracture surfaces, in
most circumstances, the water, air or mud drilling fluids will wash out any evidence of
DNAPL that was present within the fractures. In circumstances where major components
of the DNAPL are relatively non-volatile (as with the PCBs and PAHs), a "swab test"
of a fracture surface may be useful in assessing whether DNAPL had been present in the
fracture.
Swab tests on the surfaces of building interiors and equipment are commonly used
in the occupational health and industrial hygiene fields. For a fracture swab test, both
sides of the fracture are wiped with a surgical cotton gauze pad that has been soaked in a
solvent such as decane. Any small, but invisible, quantity of DNAPL that remains on the
fracture surface is thereby dissolved by the solvent in the swab. The swab is then placed
in a jar and shipped to the laboratory for analysis where it is extracted with solvent and
analyzed. The analysis is reported in terms of the mass of the compound(s) of interest per
swab. The concentration per unit area of fracture can then be determined. Although the
Sec. 13.5
445
Figure 13.20 Example of DNAPL

creosote observed along a fracture in a core
sample from below the water table at a
depth of about 10 m in a lacustrine clay
formation. (Photograph by John A. Cherry.)
results of swab tests are qualitative, they can provide information on DNAPL presence
that may not be obtainable by other means.
The results of swab analyses from an investigation of PCB contamination in fractured shale are shown in Table 13.10. PCB concentrations were found to range from
less than 0.05 to 1,300 jLg/l00 crrr'. The difficulty in the interpretation of swab tests is
determining what concentration might reflect the presence of a DNAPL phase prior to
drilling. As a means of comparison, even under natural conditions, the DNAPL residual
content within a fracture will likely be rather low. For example, laboratory studies of
chlorinated solvents in simulated fractures conducted by Schwille (1988) suggested a
residual content of 0.05 Llm 2 (= 0.5 mL/lOO em"). In the above example, the DNAPL
spilled was believed to be comprised of about 10% PCBs. If all the residual DNAPL
was retained on the fracture surface during drilling and was then extracted by the swab,
the PCB concentration would have been 45,000 jLg/lOO crrr'. However, most of any
446
Chap. 13
TABLE 13.10 Results of PCB Analyses

of Swab Test in a Fractured Shale
Formation.
Sample depth
(m)
PCB swab test

(/Lg/IOO cm 2 )
2.5
1,300
4.2
160
5.0
170
6.0
9.5
90
< 0.05
residual DNAPL would likely have been washed out of the fracture during drilling.
Because of this, concentrations in the range of several hundred to several thousand
tLg/lOO ern" would likely reflect the former presence of this DNAPL. At this site, then,
this DNAPL phase likely existed in the fractures to a depth of 5.0 m, but not beyond 9.5 m.
Sponge coring is a technology from the petroleum reservoir engineering industry
that may be applicable to the investigation of DNAPLs in fractured rock. Sponge coring
provides for the collection of undisturbed rock core and core fluids. The drill bit and
core barrel are specially designed to prevent the intrusion of drilling fluids into the core
barrel. The core barrel contains an inner sleeve of "oleophyllic" absorbent polyurethane
sponge which surrounds the rock core as it enters the core barrel. Any organic NAPL
present in the rock fractures or rock matrix is attracted to the sponge liner and trapped.
On recovery, the sponge liners are removed, split length-wise, and examined under UV
light. Any areas of the sponge that have absorbed a fluorescing NAPL will be evident.
This method is used in the study of petroleum reservoirs to estimate oil saturation since
many PAHs fluoresce strongly. In the study of a contaminated LNAPL site, Hardisty and
Johnson (1993) used sponge coring to investigate the distribution of LNAPL below the
water table in a fractured sandstone. Evidence of LNAPL was found in some fractures
but not in others. The zones of LNAPL presence as based on the sponge coring correlated
well with observations of LNAPL in monitoring wells. Research is required on sponge
coring to assess its potential for use in DNAPL investigations.

In situations where the DNAPL has dissolved and has diffused significantly into a rock or
clay matrix, it may be possible to determine the former presence of DNAPL by analysis
of the matrix material. This may be also be important in cases where other evidence of
DNAPL presence in the fractures is not available. The sampling and chemical analysis
for VOCs such as chlorinated solvents in rock or clay cores are prone to the same pitfalls
as for soil samples: handling and analysis methods must reduce the potential for losses
prior to analysis. We also note that sample handling and transfer to sample containers
is more difficult for consolidated rock or cohesive clay samples than with soil samples.
The field methanol immersion/extraction method is probably the method most suitable
Sec. 13.5
447
for the analysis of segments of rock core. Breaking of rock into small pieces to fit into a
40 mL vial for analysis by headspace or purge-and-trap would likely lead to significant
losses of YQCs.
If the purpose of the analysis of rock core is to determine whether a DNAPL is
present, core samples should be collected immediately adjacent to the fractures since that
is where the highest concentrations in the matrix will occur. Depending on the nature of
the matrix material and the length of time that the DNAPL might have resided in the
fracture, diffusion may have caused contamination of the matrix out to distances up to a
meter or more over a period of years. However, concentrations may decline very rapidly
with distance from the fracture.
The analysis of rock or clay core will yield the total concentration in the matrix. This
total will be the sum of the chemical mass dissolved in the pore water and sorbed on the
geologic solids. (In most circumstances, there will be no significant air or DNAPL phase
in the matrix.) The same partitioning calculations described in the preceding sections
for soil can be used to estimate the pore-water concentration in the matrix based on the
total matrix concentration. In this case, if the calculated pore-water concentration is a
substantial fraction of the effective solubility for the component of interest, then it is
likely that DNAPL was present in the adjacent fracture prior to diffusion into the matrix.
Lawrence et at. (1990) used the analysis of matrix samples to assess the presence
and distribution of a DNAPL in the fractured Chalk aquifer in eastern England. The
Chalk is a soft microporous limestone formation that is a major water supply aquifer.
The matrix of the Chalk has a porosity of 25% to 45% and a low hydraulic conductivity.
Primary and secondary fracture networks provide the avenues for groundwater flow in
the aquifer. The primary fractures comprise a regular set of orthogonal fractures spaced
at 0.05 m to 0.2 m apart. The secondary fractures are spaced at intervals of 5 m to 10 m
and represent primary fractures that have been enlarged by dissolution and weathering
processes.
Lawrence et at. (1990) collected cores from a 55 m deep boring in the Chalk
aquifer in an area where the groundwater had been found to be seriously contaminated
by PCE. A total of 21 cores, each 0.5 m in length, were collected from the boring. (The
Chalk is sufficiently soft to allow coring using a thin-walled tube sampler.) In a field
laboratory, each core was extruded and 6 subsamples were collected along its length.
Each subsample consisted of about 2 g of material, and was transferred to a pre-weighed
8 mL vial containing 6 mL of water and 0.5 mL of pentane. The vials were shaken in a
vortex mixer to disaggregate the material and then centrifuged to separate the sediment
from the pentane. The pentane was then analyzed using a gas chromatograph with a
photoionization detector (GC/PID). The PCE concentration in the chalk pore water was
calculated based on the measured PCE mass, and the weight and moisture content of the
sample.
The profile of the PCE concentrations in the Chalk matrix pore water are shown in
Figure 13.21. For each sample interval, the range in pore-water concentrations in the six
subsamples is shown by the solid bar. In the interval between 10 and 20 m in depth, PCE
concentrations in the matrix were as high as 40 mg/L ("'20% of saturation). Pore-water
concentrations as high as 5 mg/L (r- 2% of saturation) were found to a depth of 50 m.
448
Chap. 13
As will be described in the following section, dissolved concentrations of this magnitude

generally indicate proximity to a zone of DNAPL. On the basis of these concentrations,
Lawrence et al. (1990) concluded that the DNAPL had penetrated to a depth of ~50 m.
The higher concentrations found between 10 and 20 m coincide with a transition from a
weathered chalk stratum to a less weathered, less permeable stratum. In this zone, DNAPL
may have accumulated, occupying more of the fractures and permitting greater diffusion
into the matrix. At greater depths, DNAPL may have penetrated fewer fractures, or the
important fractures were more widely spaced. Consequently, pore-water concentrations
in the matrix at these depths were found to be lower.
In addition to the cores, samples of groundwater were obtained as the Chalk boring
was advanced. The PCE concentrations in these samples are given in Figure 13.21. The
boring water concentrations ranged from 10 to 25 mg/L (5 to 10% of saturation), i.e.,
sufficiently high to lead to the conclusion that DNAPL was present in the proximity of the
boring. However, although the highest concentrations in the boring water coincided with
the highest matrix pore-water concentrations at a depth of 10 to 20 m, the boring water
concentrations at greater depth were much higher than the pore-water concentrations.
This is likely the result of the migration of boring water from the upper portions of the
boring into the lower portions of the boring during drilling. If unrecognized, this crosscontamination would lead to the interpretation that abundant DNAPL had penetrated to
the full depth of 50 m, whereas the matrix pore-water concentration profile suggests that
the greatest accumulation of DNAPL may be in the 10 to 20 m depth interval. These
findings illustrate the potential effectiveness in using the analysis of matrix pore water
in assessing DNAPL distributions.
PCE Concentration (mg/L)
10
15
20
25
30
40
35
10
..
e - -- ---
-- ---
--
--:"e_
.e
40
Pore water
Ilromeore
e-
e_
Waterlrom'
.-e boring
e
50
--
Figure 13.21. Vertical profile of pore water PCE concentrations in core samples from
the Chalk aquifer in eastern England. The bars show the ranges in concentration from
six sub-samples collected from each 0.5 m section of core. The dashed curves shows
the PCE concentrations in groundwater collected from the boring during core sampling.
Data from Lawrence et al. (1990).
Sec. 13.6
449
13.6 CHARACTER OF DISSOLVED PLUMES

13.6.1 General
The spatial distribution of dissolved contaminant plumes in the groundwater zone

can yield valuable information when assessing the presence and spatial distribution of
DNAPLs in the subsurface. The spatial distribution of dissolved plumes within an aquifer,
the magnitude of the dissolved concentrations, the temporal variability in the dissolved
concentrations, and the response of dissolved concentrations to groundwater pump-andtreat operations can each be influenced by the nature of the DNAPL source zone(s) in the
subsurface. This section considers the theoretical and experimental evidence regarding
the character of dissolved plumes from DNAPL sources and then describes a case study
that exemplifies many of these findings.
The spatial distribution of the dissolved plume that develops at a DNAPL site in the
subsurface may be unique in several ways. Many DNAPL chemicals, including the chlorinated solvents, have high vapor pressures. The release of a significant quantity of
DNAPL can create a large plume or halo of contamination in the vadose zone as a result
of vapor diffusion. Groundwater contamination will then develop at the water table as a
result of infiltration of water through the vapor halo and the DNAPL zones in the vadose
zone. In most circumstances, the vapor halo will be much larger than the DNAPL zone,
and will result in a dissolved plume much larger than the DNAPL zone itself. Field experiments and computer modeling have shown that the groundwater contamination that
results from this type of release remains as a thin plume close to the water table.
In contrast to the above case, when DNAPL residual and/or pools are present in
the groundwater zone, the dissolved plume(s) will reflect the size(s) and depth(s) of the
source zone(s). The dissolved plumes may occur far below the water table, but will not be
significantly wider than the DNAPL source zones from which they are derived because
of weak transverse dispersion processes. As discussed in Section 13.3.3, attempts to
detennine the lateral extent and depth to which DNAPL has penetrated below the water
table by drilling within the DNAPL zone is unreliable and generally undesirable because
of the potential for short circuiting along borings. In most circumstances, lateral and
vertical delineation of dissolved plumes downgradient of DNAPL source zones is the
most effective and safest way of assessing the vertical distribution of DNAPL sources.
The contrast between dissolved plumes derived from DNAPL sources in the vadose zone and those derived from DNAPL sources groundwater zone is illustrated by
controlled field experiments that have been conducted at the Borden test site. The vadose
zone experiments are described by Conant et al. (1995), and the groundwater zone experiment is described by Rivett et at. (1995). Detailed monitoring utilizing hundreds of
sampling points spaced vertically at intervals of 0.2 to 0.3 m have permitted a detailed
understanding of the groundwater plumes created at the site. In the vadose zone experiments, cylindrical source zones were created in the unsaturated zone by mixing 42 L of
450
Chap. 13
TCE with sand. A source zone measured 1.2 m in diameter and 0.8 m in height. TCE
concentrations in soil gas measured 18 days after placement of the first source are given
in Figure 13.13; TCE concentrations above 30 ppmV (about 170 fLglL) extended as far
as 7 m from the source zone. This source remained in place for 1 month, and was then
removed. The area of the vapor halo was then treated by soil vapor extraction. A similar
experiment was conducted at the same location 20 months later. In this case, the TCE
source zone remained in place for 3 months before it was removed and the area treated
by soil vapor extraction. Despite the fact that the DNAPL source zones remained in
the vadose zone for only short periods of time, a very substantial dissolved TCE plume
was created at the interface between the vadose and groundwater zones. This interface
zone plume is shown in plan view in Figure 13.22, and in section view in Figure 13.23.
The interface plume in these figures is as much as 40 m wide, but only 1 to 2 m thick.
The width of the dissolved plume reflects the size of the vapor halo, not the size of the
DNAPL zone. Dissolved concentrations in the core of the plume were as high as 6 mglL,
reflecting the peak of Experiment #2, but extended over only very short vertical intervals
0.5 m).
The dissolved plume created by the groundwater zone source at Borden was substantially different from that generated by the vadose zone sources. In this experiment, a
rectangular source zone measuring 1.5 m long, 1 m high, and 0.5 m thick was created 1 m
below the seasonal low water table. The source zone contained 14.8 L of a solvent mixture comprised of 6% chloroform (trichloromethane, or TCM), 39% TCE, and 55% PCE.
After 11 months, the TCM and TCE plumes had migrated about 40 m downgradient. Due
to retardation processes, the PCE plume had migrated only about 30 m downgradient.
The TCE plume from the groundwater zone experiment is shown in plan view in Figure 13.22 and in section view in Figure 13.23. The maximum width of the groundwater
zone plume at 11 months was about 5 m, reflecting a small degree of lateral dispersion.
This occurred despite the fact the groundwater flow direction varied seasonally over an
arc of as much as 30 ", Without the seasonal fluctuation in flow direction, the width of
the plume would have been even smaller. The maximum vertical thickness of the plume
was about 3 m. Downward spreading likely resulted from density-induced sinking of the
plume in response to dissolved concentrations of up to 1 gIL in the core of the plume
close to the source zone, and from variations in hydraulic conductivity. The maximum
TCE concentrations in the plume close to the source ranged as high as 600 mg/L, which
was the effective solubility of TCE in the source-zone mixture; dissolved concentrations
downgradient of the source zone were considerably lower as a result of dispersion.
Although the Borden experiments represent idealized cases, they do reflect the
general features of dissolved plumes that would be anticipated at a site where volatile
DNAPL was released to the subsurface and entered the groundwater zone. Compared to
the original DNAPL zone in the vadose zone, the interface plume created by a vapor halo
in the vadose zone would be expected to be broad in width but vertically thin. Within this
type of thin interface plume, dissolved concentrations will generally be much lower than
saturation concentrations. In contrast, a plume created by DNAPL in the groundwater
zone would be expected to reflect the size and vertical location of the DNAPL zone. If
the DNAPL zone is narrow, the resultant dissolved plume will be narrow. If the DNAPL
Vadose lone Experiments;
#1 - Source in place 1 month

#2 - Source in place 3 months
om-o.
Peak Irom
Exp.#1
Peaklrom
Exp.#2
---
--.L"-------------+-
I-
Source size'
f'0>
37m
0.8
high
1.2 m
m dla.
i:;-
f .;:".
'i7
~ Water table range
sr
't---
.~
Plan View of
TCE Plume
20 months after
Exp. #1 and
11 months after
Exp.#2
"
Groundwater lone Experiment:

Continuous source
t-
Ground
2.5
YWater table range
\l
~
_
1.0 m
0.01 mg/L I TCE Concentration

Contours
Plan View of
TCE Plume
after 11 months
SoYrce S i z e . '
1.5 m wide
1.0 m high
0.5 mthick
...
01>0
\Il
Figure 13.22 Plan views of dissolved plumes derived from vadose-zone and groundwater-zone
experimental sources in the Borden aquifer. The vadose zone experiment was performed by Conant
et al. (1995), and the groundwater-zone experiment was performed by Rivett et al. (1995).
2,0
metres
4p
Vadose lone Experiments:
..,.
#1 - Source in place 1 month

#2 - Source in place 3 months
\i1
N
1Q2j
Iii
Source
~ 100
Qi
EBB
:;:;
~
~
TCECOnCeIltrati
0.01 mg/L
0.1 mg/L 1 mg/L
~
~~
~
,/
96
Section View of
TCE Plume
Contours
I
Peak trorn Exp.jrt
~ Limit of
~ Mon~oring
94
20
40
60
20 months after
Exp. #1 and
11 months after
Exp.#2
100
80
metres
Groundwater lone Experiment:

Continuous source
102
Iii
~
Qi
I
100
10 mg/L
Ground
TCE Concentration
Contours
1 mg/L
98
Section View of
TCE Plume
96
after 11 months
c:
~
>
Q)
W 94
92
20
metres
40
Figure 13.23 Section views of dissolved plumes derived from vadose-zone and groundwater-zone
experimental sources in the Borden aquifer. The vadose-zone experiment was performed by Conant
et al. (1994), and the groundwater-zone experiment was performed by Rivett et al. (1994).
Sec. 13.6
453
zone extends to a substantial depth in the aquifer, the dissolved plume will extend to a
similar depth.
Because of the thin vertical extent of interface plumes, groundwater samples obtained from such a plume using conventional wells will generally exhibit low dissolved
concentrations due to dilution with uncontaminated water from below the plume: well
intakes are usually long relative to the thickness of an interface plume. Monitoring using
conventional wells will fail to identify the plume if the well intake does not straddle the
water table. For plumes derived from DNAPL sources in the groundwater zone, because
of the limited vertical and lateral extents of such plumes, groundwater monitoring in this
case can: 1) also yield low dissolved concentrations as a result of dilution with uncontaminated water from above or below the plume when well intakes are long; or 2) fail to
identify the plume if the well intake does not intersect the plume. The manner in which
an interface zone plume and deeper plumes derived from DNAPL in the groundwater
zone can affect the results of groundwater monitoring at an actual field site is discussed
in Section 13.6.5 based on groundwater monitoring at three different depths within the
aquifer.
Good depth resolution and therefore information regarding the vertical distribution
of DNAPL below the water table can be obtained using drive-point devices to obtain a
vertical profile of the dissolved plume downgradient of a DNAPL source zone. Clausen
et al. (1993) used a drive-point device to collect groundwater samples from discrete
depths to assess the presence of liquid TCE within a confined sand and gravel aquifer.
The aquifer was about 9 m thick, and the top of the aquifer occurred at a depth of
about 13 to 20 m below ground surface. Four groundwater samples were collected at
approximately 2 m intervals through the aquifer. TCE concentrations were 4.1 mg/L at
the top, ~8.2 mg/L in the middle and 16 mg/L at the bottom of the aquifer. The increase
in TCE with depth and the relatively high concentrations led Clausen et al. (1993) to
conclude that liquid TCE had likely penetrated to the base of the aquifer, to a total depth
of about 30 m.
Pitkin et al. (1994) used detailed vertical profiling to delineate the dissolved plume
in a sand aquifer created by a release of PCE from a retail dry cleaner. A drive-point
device was used to collect groundwater samples at vertical intervals of 0.1 to 0.25 m,
to depths as great as 15 m. A total of 23 vertical profiles were collected from 3 transects across the dissolved plume. An example of two of the vertical profiles is shown
in Figure 13.24. Profiles Bland B2 were located about 20 and 50 m down gradient,
respectively, of the suspected source zone. The profiles indicate that the dissolved plume
occurs below the water table in the zone between 7 and 10 m deep. Dissolved concentrations in the center of the plume are as high as 40 mg/L, The highest PCE concentrations
were found in a zone of silty fine sand within the sand aquifer, which may represent
the stratum in which the DNAPL source resides. It is not clear why PCE concentrations
were found to be very low above the silt and peat layer. It is possible that the lateral
direction of groundwater flow in this upper zone differs from the zone below, and the
dissolved plume in this zone has not yet been found. Below the core of the dissolved
plume in profile B2, concentrations were found to decline rapidly to 0.01 mg/L or less
within a vertical distance of only 1 to 2 m. In profile B 1, concentrations beneath the
454
Chap. 13
PCE Concentration (mgIL)
0.001
0.01
0.1
100
10
o
Water Table
2
4
E.
s 10
Silt and Peat Layer
--
AJ
12
14
Silty Fine Sand
~-
-#0
I--
Profile 61
c~
16
PCE Concentration (mgIL)
0.001
0.01
0.1
10
100
-~
-~
4 f----
E.
Silt and Peat Layer
- " - - f---
~ 10
12
14
16
Water Table
~f---
-:--&-...
~-1-
\Profile 62
---- - -
Silty Fine Sand

Sand
_J __
Figure 13.24 Vertical profiles of a peE plume obtained using a drive-point device.
Data from Pitkin et al. (1994).
core of the plume were found to decline less rapidly. PCE concentrations in the range
of 0.1 down to 0.01 mg/L were found for a distance of about 4 m below the plume.
Because of its location nearest the suspected source zone, it is not known whether this
dissolved profile reflects the presence of minor amounts of DNAPL at greater depth, or
the effect of cross-contamination of the deeper samples by groundwater from the core
of the plume. If conventional monitoring wells with 3 to 10 m long intake intervals had
been used in this case, much lower dissolved concentrations would have been found and
the localized nature of the source zone would not have been evident. Such a result may
have led to a mistaken interpretation that DNAPL was not present at this site.

One of the simplest tasks that can be performed to assess the likelihood that a DNAPL
phase is present in the subsurface at a given site involves the comparison of dissolved
concentrations in groundwater with pure-phase solubilities or estimated effective
solubilities. Chapter 7 describes the present understanding of the mechanisms of DNAPL
dissolution as based on theoretical, experimental, and computer-modeling studies. This
understanding indicates that when groundwater moves at less than ~ 1 m/day through at
least a few centimeters of a DNAPL source zone that is at a typical residual saturation,
then that groundwater will exit that zone saturated (at the effective solubility) with the
Sec. 13.6
455
compound(s) in the DNAPL phase. Consequently, if groundwater samples are collected

in close proximity to the DNAPL zone and the intake length of the monitoring wells is
comparable to the size of DNAPL residual zone, near-saturated dissolved concentrations
can be expected. On the other hand, if the DNAPL residual zone is small relative to
the spacing of the monitoring wells or relative to the intake length of the wells, then
less-than-saturated concentrations are to be expected: dilution by less contaminated and
uncontaminated groundwater will occur.
Studies of the dissolution of DNAPL pools indicate that saturated concentrations in
groundwater will be found only: 1) immediately above a pool: and 2) in a thin layer at the
elevation of the pool in the near-downgradient area. This is a result of the facts that: 1) a
horizontal pool presents little cross-sectional area to on-coming, horizontal groundwater
flow; and 2) the high DNAPL-saturation levels within a pool inhibit groundwater flow
directly through the pool itself. These effects are illustrated in Figure 13.25 for which the
dissolved concentrations above a hypothetical DNAPL pool in a typical sandy aquifer
have been calculated using an analytical solution from Johnson and Pankow (1992).
The 10 m by 10m pool was located at the base of a 15 m thick aquifer. The linear
groundwater velocity was set at 0.5 mid, and the vertical dispersivity was set at 0.001 m.
At the downgradient edge of the pool, contamination is limited to elevations of less than
1.0 m above the pool. Each of the five hypothetical monitoring wells now considered is
located at the downgradient edge ofthe pool. For a well that is screened over the lower 3 m
of the aquifer, the dissolved concentration is only 4.6% of the saturation concentration,
regardless of the compound. For a longer well intake length, the concentration exhibited
in the well would be even lower; for an intake length equal to the full thickness of the
aquifer, the dissolved concentration would be less than 1% of saturation. If the intake
is terminated 0.3 m above the pool, the dissolved concentration in the well would be
only 0.1% of saturation; if the intake is terminated 0.5 m above the pool, the dissolved
concentration in the well would be 0.001 % of saturation, or barely above normal detection
limits for most chlorinated solvents.
Based on computer modeling studies of DNAPL migration in heterogeneous media,
and based on controlled field experiments in horizontally-stratified deposits, it is likely
that much of the DNAPL mass present in the subsurface at a site will be distributed in
horizontal layers or pools that have accumulated on zones of lower permeability. The
zones in which the DNAPL is present in residual form would be the vertical columns
or "fingers" that connect the horizontal layers. Anderson et al. (1992) used a 3-D analytical model to simulate the dissolved plumes that would develop in a 15 m thick
sandy aquifer from a complex DNAPL source zone comprised of multiple horizontal
pools and vertical residual zones. Figure 13.26 is a section view of the simulated source
zone, looking upgradient. Figure 13.27 gives the dissolved concentrations that would
be exhibited in monitoring wells downgradient of the source zone. The hypothetical
wells each had 2 m long intake intervals located in the upper, middle and lower portions of the aquifer. The highest dissolved concentration exhibited in a well was 12%
of saturation, and that occurred at the bottom of the aquifer close to the source zone.
Despite the fact that the wells were located relatively close to, and directly downgradient
of a relatively large source zone, most of the concentrations were only a few percent
456
Chap. 13
Average Concentration
% Saturation
4.6%
Gro~ndwater
Flow
LO
1.5%
0.9%
0.1%
0.001%
..
I
I
I
I
I
I
10 m
Hypothetical wells located at

down gradient edge of pool
tIm
Concentration Profile
above Pool
j~
1.5
2.0
gj
1.0
~ c,
0.5
0.0
11l
.~
Well Completion Details

a 3 m screen to top of pool.
b - 9 m screen to top of pool.
c - 15 m screen to top of pool.
d - 3 m screen to 0.3 m above pool.
25
50
75
% Saturation
100
e - 3 m screen to 0.5 m above pool.
Figure 13.25 Dissolved concentrations observed in five hypothetical monitoring wells

at the downgradient edge of a DNAPL pool. Each well is located at the same place, but
is screened over a different interval. Dissolved concentrations were calculated using the
analytical solution from Johnson and Pankow (1992).
of saturation. The dissolved plumes are described in longitudinal and transverse crosssections in Figure 13.28. The contamination that extends furthest downgradient is that
derived from the pools. However, the contamination from a given pool is relatively thin;
even monitoring wells with relatively short well intakes situated directly across such
plumes would exhibit dissolved concentrations of only a few percent of saturation, at
most.
We can conclude that conventional monitoring wells are not likely to allow detection of peak dissolved concentrations at DNAPL sites because: 1) the intakes are
generally too long; and/or 2) an insufficient number of wells is used in each well cluster
or nest. Drive-point devices used to collect detailed vertical profiles of dissolved concentration provide the highest probability for detecting peak concentrations. These devices
Sec. 13.6
457
o
3
-o
.J::.
cQ). 9
12
15
-10
-5
10
Width (rn)
Figure 13.26 Configuration of a complex DNAPL source zone for dissolved plume
modeling. Adapted with permission from Anderson (1988); described further in Anderson
et al. (1992).
DNAPL Source Zone
Concentrations in % saturation
0.4
1.9 - - - - -20 m
3.8
- - - - - - - - - - - - - - - - - - - - -50 m
1!l!I4---u.5
... _~-1.4
2.8
------------------100 m
Figure 13.27 Dissolved concentrations in hypothetical monitoring wells down gradient

of the complex DNAPL source zone. Well screens are 2 m in length, situated in the
upper, middle and lower portion of the aquifer. Concentrations are shown as percent of
saturation. Adapted from Anderson et al. (1992).
make it possible to sample groundwater at vertical intervals of 0.5 m or less (Pitkin et

al., 1994), and the resulting data is of great assistance in the assessment of the location
and morphology of the DNAPL source zone(s).
In fractured rock, the distribution of the DNAPL and its dissolution into groundwater is very poorly understood. However, on a conceptual basis, we can expect that
458
a.
Chap. 13
Longitudinal Section
>25% saturation
>2.5% saturation
I..c
Q.
Q)
Cl
15
100
0
Distance (m)
b.
Transverse Section
>2.5% saturation >0.5% saturation
Or---~~---,.-~~...,
15 L....._.....:::::.........~==-.:::::......J
-20
20
o
Width (m)
Figure 13.28 Longitudinal and transverse sections downgradient of the complex DNAPL
source zone. The longitudinal section (a) extends from the center of the source zone. The
transverse section (b) is situated 100 m downgradient of the source zone. Adapted from
Anderson et al. (1992).
DNAPL will be present in some fractures and absent from others. The groundwater captured in a monitoring well in fractured rock may therefore comprise some water that
has contacted DNAPL, and other water that has not. It is not possible to predict the
resultant dissolved concentrations, but they will in general be much lower than saturation. As an example, Table 13.11 gives the dissolved concentrations in a monitoring
well and a pumping well in a fractured sandstone aquifer (EPA, 1992). DNAPL was
recovered from each of these wells. The principal component of the DNAPL was 1,2,3TABLE 13.11 Dissolved 1,2,3-Trichloropropane Concentrations Found in Wells From
Which DNAPL Was Recovered in a Fractured Sandstone Formation.
Monitoring well
W-81
Pumping well
EW-4
35 m to 41 m
9 m to 55 m
Effective Solubility (mglL)
1,400
1,400
Initial Concentration (mglL)
150
240
Intake Interval
Concentration While Pumping (mg/L)
50
Sec. 13.6
459
trichloropropane having an effective solubility of approximately 1,400 mg/L, Both of

these wells are situated directly within a large bedrock zone known to contain DNAPL
to considerable depth. The dissolved concentration in well W-8I was about 11% of
saturation. The dissolved concentration in well EW -4 was 17% of saturation at the start
of pumping, but declined during pumping to about 3.5% of saturation. Although this
well was installed directly within a large DNAPL zone, this reduction was likely due
to dilution by less contaminated groundwater. Thus, wells installed downgradient of a
DNAPL source zone in a fractured aquifer can exhibit concentrations that are much lower
than saturation.
These findings have critical implications for the interpretation of groundwater monitoring data. Before it was recognized how DNAPLs can behave in the subsurface, the
presence of a DNAPL phase in the subsurface was not considered unless DNAPL was actually recovered directly, or groundwater concentrations in monitoring wells approached
saturation levels. It is now clear that less-than-saturated concentrations will occur in
monitoring wells in many, or even most cases of DNAPL contamination. During the
mid-1980s, the finding of dissolved concentrations of 10% of saturation was suggested
as a "rule-of-thumb" for inferring the presence of a DNAPL phase (Feenstra and Cherry,
1988). Recently, it has been suggested that this rule-of-thumb concentration be reduced to
1 % of saturation (Newell and Ross, 1991). We believe that this is reasonable. This value,
however, should not be viewed as strict criterion since, based on the above discussion,
we know that the magnitude of the dissolved concentrations observed in a monitoring
well will depend on factors other than the presence or absence of DNAPL phase source
zones (e.g., well location and well intake length, and size and nature of the source
zone). The use of a 1% rule-of-thumb in any assessment of the spatial distribution of
DNAPL zones must be performed cautiously, particularly in the downgradient direction.
For example, the dissolved plume emitted from a very large DNAPL zone may exhibit
dissolved concentrations above 1% of saturation for a substantial distance downgradient
of the source zone. A conclusion that the size of the DNAPL zone corresponds to the size
of the dissolved plume exceeding 1% of saturation, in such a case, could be an extreme
overestimate.
13.6.4 Persistence and Temporal Variability of Dissolved

Concentrations
The DNAPL phase in the vadose and groundwater zones at many sites probably represents
sufficient chemical mass to cause dissolved plumes to persist for centuries or longer (see
also Chapter 7). Many dissolved plumes at DNAPL sites have now been monitored for
more than a decade and show no indication of diminishing concentration or plume size.
During the early stage (perhaps the first few decades) of dissolution of a DNAPL zone,
the dissolved concentrations found in the groundwater may remain relatively constant,
with the highest concentrations found close to the source zone. In contrast, plumes
derived from the leaching of sorbed-phase contamination on soil will likely exhibit everdeclining dissolved concentrations. Similarly, dissolved plumes derived from past releases
of aqueous-phase contaminants will likely exhibit areal maximum concentrations, or
460
Chap. 13
peaks, someplace downgradient of the release area. Thus, the persistence of elevated
groundwater concentrations over monitoring periods of 5 to 10 years, particularly close
to suspected source areas, is generally indicative of a DNAPL source zone.
In the case of fractured clay or sedimentary rock, significant dissolved concentrations may persist in the groundwater even long after the DNAPL has disappeared. As
described in Chapter 12, the dissolution of DNAPL in fractures in a fractured porous
medium will be enhanced by diffusion of dissolved contaminants into the matrix. When
the DNAPL disappears and less-contaminated groundwater flushes through the fractures,
dissolved contaminants will diffuse back out of the matrix and into the water flowing
through the fractures. As shown in Figure 13.21 for the Chalk aquifer in eastern England,
the pore water in the matrix between fractures can contain high dissolved concentrations
that represent a significant continuing source of contaminants to the groundwater.
Extreme temporal variations in dissolved concentrations observed in monitoring
wells may be another indication of a DNAPL source zone. Because the dissolved plume
derived from a DNAPL source in the groundwater zone will reflect the size of the
DNAPL zone, such plumes may be of limited lateral extent with very steep concentration
gradients between the margins and the core of the plume (see Figures 13.22 and 13.23). If
a monitoring well is situated close to the margin of a plume, fluctuations in groundwater
flow direction may cause the plume to move into or out of the zone of capture of the
well, thereby resulting in large variations in the dissolved concentrations observed in the
well; dissolved plumes derived from larger and less localized source zones would be less
likely to cause this type of effect.
13.6.5 Response of Dissolved Concentrations During Aquifer

Pump-and-Treat
The time-dependent behavior of dissolved concentrations in groundwater pumping
wells can provide evidence regarding the presence/absence of DNAPL in the groundwater
zone. During the 1980s, aquifer pump-and-treat systems were installed at many sites of
groundwater contamination. The apparent failure of many of these systems to achieve
complete aquifer restoration is the result of DNAPL zones remaining in the subsurface
(EPA, 1992). A typical pattern in dissolved concentrations in a pumping well that is
influenced by the presence of a DNAPL source zone is a rapid decline in concentration
as a result of dilution effects followed by a plateau at a significant concentration level.
Effects such as slow desorption from aquifer solids and diffusion out of finer-grained
layers can inhibit restoration by pump-and-treat, but these effects are more likely to cause
long tails of slowly declining concentration rather than a long plateau at an elevated
concentration.
An example from Rivett (1993) is shown in Figure 13.29 for a pump-and-treat
system applied to the experimental plume in Figures 13.22 and 13.23. In this case, three
pumping wells were installed along the axis of the plume; PW-2 is nearest the source
zone, PW-4 is located at the leading edge of the plume, and PW-3 is located between
the other two. The wells were each pumped at a rate of 5 L/min. Based on data from the
monitoring network, it was found that the TCE plume between the wells was substantially
TCE (mg/L) in Pumping Wells

SOURCE ZONE
1.
10
20
PW-2
4
Il
3
metres
2
1
I "'1\
I
IV
50
100
Plume not monitored in detail in this area
200
250
200
250
200
250
PW-3
1 mg/L
150
Days
50
100
150
Days
PW-4
-'
Plume Configurations
50
57 days
before
pumping
..."'"
After 27 days
of pumping
After 106 days

of pumping
After 399 days

of pumping
Figure 13.29 Response of dissolved concentrations in pumping wells for pump-and-treat remediation of the plume from the groundwater zone experimental source in the Borden aquifer. Data
from Rivett (1993).
100
150
Days
462
Chap. 13
removed, with concentrations reduced from initial levels of 1 to 10 mg/L down to 0.01
to 0.1 mglL. The highest concentrations remained in the zones of hydraulic stagnation
between the wells, and in the finer-grained beds of the aquifer. At the leading edge
well (PW-4), the concentration rose initially as the plume was drawn toward the well,
then the concentration declined to less than 0.01 mg/L. In the middle well (PW-3), initial
concentrations were about 3 mglL, then declined to below 0.01 mglL after about 100 days.
The lateral zone of capture of each of the pumping wells was several times larger than
the width of the plume. Therefore, much of the initial rapid decline in concentrations in
the pumping wells was due to dilution from water outside the plume. The concentration
response in PW-2 was considerably different. Concentrations were initially about 5 mg/L,
and then declined rapidly for about 100 days, after which concentrations leveled off at
between 0.5 mg/L and I mglL; these elevated, plateaued concentrations reflect the ongoing contribution to PW-2 from dissolution of the source zone.
13.6.6 Case History-Chlorinated Solvent Plume, Dayton, New

Jersey
13.6.6.1 Background
Several of the important factors related to the diagnosis of DNAPL in the subsurface
can be illustrated by examining the groundwater contamination and remedial measures
taken at an industrial plant site near Dayton, New Jersey (EPA, 1989). The plant site was
used for the manufacture of inked printer ribbons and punch cards for computers. The
aquifer consists of about IS m of coarse sand and gravel overlying a discontinuous layer
of interbedded sand and clay about 1 to 3 m thick. The water table occurs at a depth of
about 6 to 7 m below ground surface, and the average hydraulic conductivity is about
8 x 10- 2 cm/s. In December 1977, PCE and I, 1,1-TCA were discovered at concentrations
of about 0.2 and 0.6 jLglL, respectively, in a municipal water supply well located about
500 m east of the plant. The municipal well was shut down and investigations of the plant
site began in early 1978. The actual contribution of this site to the contamination of the
municipal well cannot be determined precisely because other groundwater contamination
sites are present in the neighborhood.

A plume of PCE and 1,l,I-TCA moving eastward towards the municipal well was found
on the plant site. The groundwater concentrations ranged as high as 6.1 mglL for PCE and
9.5 mglL for I,I,I-TCA (see Figure 13.30). These maximum concentrations represent
2.5% of saturation for pure-phase PCE and 0.8% of saturation for pure-phase I,l,l-TCA.
However, when these studies were conducted in 1978, no understanding of DNAPL behavior in the subsurface existed in the groundwater profession. Therefore, the magnitude
of these concentrations did not lead to the conclusion that DNAPL was present in the
aquifer.
Chlorinated solvents had been used at the plant and it was believed that the groundwater contamination originated from releases of solvent product from the chemical offloading and storage areas. Records from the plant indicated that only PCE had been pur-
Sec. 13.6
463
GW-16B
1978 Before
Pump-and-Treat
"I"'~-l-_--.-
//
GW-32 and 42s.i.d
GW-16B
1984 - At End of
Pump-and-Treat
.\.
o
100
I ---....
m
GW-32 and 42s.i.d
"'o~'<;\
GW-16B
1987 - After Regrowth \

of Plume
100
"----'
\
GW-32 and 42s,i,d
Pumping Well
_.
Monitoring Well
Figure 13.30 PCE plume configuration for the Dayton site in 1978 before the implementation of the pump-and-treat remedy, in 1984 at the end of the initial pump-and-treat
effort, and in 1987 after regrowth of the plume.
chased. For a time it was thought the I, 1,1-TCA may have originated from another site.
Subsequently, an analysis of the solvent used at the plant revealed that the PCE contained
about 15% I,I,I-TCA. This illustrates the difficulties that can be encountered in assessing
the chemical composition of DNAPL releases. The composition suggests that the effective
solubilities of PCE and l,l,l-TCA for the site are about 200 and 190 mg/L, respectively.
As a result, the maximum observed groundwater concentrations noted above represent
~3% of saturation for PCE and ~5% saturation for l,l,l-TCA. In the light of our current understanding, finding dissolved concentrations of this magnitude would lead us to
464
Chap. 13
consider the presence of DNAPL below the water table. Indeed, the maximum groundwater concentrations were observed in wells screened across the full thickness of the aquifer.
With the dilution expected from these long-screened wells, dissolved concentrations of
this magnitude are very strong evidence of the presence of DNAPL in the aquifer.
No soil sampling/analysis for volatiles was conducted until 1985-1986. At that
time, the highest level of soil contamination was found at a depth of about 7 m, close
to the water table. The total VOC concentration of 13 mg/kg is considerably lower than
levels that would suggest the presence of DNAPL based on partitioning calculations (i.e.,
Eq.(13.3)). However, in the assessment of these historical data, we must remember that
this older soil concentration data probably suffered from volatilization losses due to the
sample handling and analysis methods used at that time.
The plume was found to be wide compared to the size of the suspected release
area. There are several possible explanations for this observation: 1) time-dependent
variations in off-site pumping in the municipal well to the east and other private wells
to the north may have acted to draw the plume back and forth; 2) a broad plume of
relatively low concentration (10 to 100 J..Lg/L) in the groundwater could have resulted
from the initial transport of vapors through the vadose radially outward from the release
area; and/or 3) the broad dissolved plume was a composite plume caused by several
different sources on the site. As regards the second possibility, we note that vapors in
the vadose zone would have resulted in a thin, relatively high concentration groundwater
plume immediately at the water table, any thin plume at the water table would have been
diluted significantly during sampling from the fully-screened wells.
Remedial measures at the site commenced at the Dayton site in mid-1978 with the
removal of the chemical storage tanks, excavation of several tens of cubic meters of soil
from around the tanks, and the installation of a network of groundwater pumping wells
on the plant site. Since it was believed at the time that the dissolved plumes were due to
leaching of contaminants from the shallow soil zone, removal of the tanks and shallow
soil together with removal of the dissolved plume by groundwater pump-and-treat were
expected to provide full and permanent restoration of the aquifer within a few years.
However, it was not recognized that DNAPL phase solvent had penetrated below the
water table where it would comprise an on-going source of contamination for many years.
The remedial measures for the overall plume involved on-site pumping wells, offsite pumping wells, and a resumption of pumping of the municipal water supply well.
We now consider only the performance of the on-site pumping because this area includes
the source zone, and because on-site monitoring was much more detailed than off-site
monitoring. For simplicity, only the PCE plumes will be considered. The on-site pumping
system consisted of up to 13 wells. Seven of the wells operated almost continuously
from mid-1978 until late-1984 when on-site pumping ceased. The other wells operated
for shorter intervals during this period. Three of the pumping wells yielded from 250 to
390 Umin, and the remaining 10 wells yielded from 20 to 40 Umin. A typical pumping
rate for the overall system was about 1,100 Umin. The discharge from the pumping
Sec. 13.6
465
system was treated by air stripping and recharged to the aquifer by spray irrigation along
the western side of the plant site.
By late 1984, dissolved concentrations in the aquifer had been reduced significantly
from levels as high as several thousand jlglL down to 10 jlglL or less (see Figure 13.30).
During this period, a total of 3.4 billion liters of groundwater and an equivalent of 7,500 L
of solvents had been removed by the on-site pumping system. Reductions down to low
jlglL concentrations occurred in monitoring wells and pumping wells throughout the
plume zone except in the area close to the suspected release. In that source zone, the
solvent concentrations in two pumping wells were reduced significantly, but declined only
to concentrations of 300 to 500 jlglL. The differences in response between the plume
zone and the source zone during pumping from 1978 to 1984 are shown in Figure 13.31
for pumping wells GW-32 and GW-16B respectively. Although not recognized at the
time, the elevated, plateaued concentrations in the pumping wells near the suspected
source zone indicated an on-going source of contamination, probably DNAPL below the
water table. The reduction in concentrations in the pumping wells in that zone that were
observed were likely due to dilution by water drawn from outside the source zone.
Although chlorinated solvents are only weakly sorbed to geologic media and can
therefore produce large contaminant plumes, their recovery by groundwater pump-andtreat is facilitated by their weak sorption. At the Dayton site, groundwater pump-and-treat
resulted in a very substantial reduction in the size of the dissolved plume by removing
~7 plume volumes of groundwater in 6.5 years. The groundwater pump-and-treat also
provided containment of the source zone by preventing further release of dissolved solvents to the plume.
Although it was not recognized at the time, groundwater pump-and-treat did not provide
restoration of the source zone. In late 1984, the state environmental regulatory agency
determined that the remedial measures had reached the appropriate clean-up target for
the majority of the plume area, and had reached the point of diminishing return for
the release area. With this conclusion, groundwater pump-and-treat was stopped. Within
about 6 to 9 months, monitoring indicated that the concentrations had begun to rise in
the wells close to the source zone (e.g., GW-32). Eventually, concentrations in several of
the monitoring wells rose to levels higher than were observed in 1978 (see Figure 13.32).
By mid-1987, about 2.5 years after pumping had ceased, the dissolved solvent plume had
grown as far as the eastern plant boundary (Figure 13.30). The regrowth of the plume
was, in itself, irrefutable evidence of a DNAPL source zone below the water table.
Further evidence of the nature of the source zone is the aquifer was derived from
multi-level monitoring wells. The majority of the monitoring at the site was conducted in
wells screened over the entire saturated thickness of the aquifer. In 1985, three multilevel
wells (42s, 42i, and 42d) were installed in the upper, middle and lower portions of the
aquifer at about 20 m from the suspected source zone. These samplers were monitored
regularly during the period after the groundwater pump-and-treat operations ceased; the
results are shown in Figure 13.33. At the beginning of 1985, dissolved contamination
466

Start of On-S~e
Pumping
Chap. 13
Source Zone GW-32
3,500
3,000
2,500
.=;
w
o0-
2,000
1,500
1,000
500
0
a>
~
--,
Date of 6-Month Means
Plume Zone GW-168

3,500
3,000
2,500
.;;
~
c.
2,000
Start of On-Site
Pumping
1,500
1,000
Figure 13.31 Response of dissolved concentrations in pumping wells during pumpand-treat remediation at the Dayton site. GW-32 is situated within 20 to 40 m of the
suspected source zone. GW-16B is situated about 200 m downgradient of the source
zone.
was found only at the bottom of the aquifer. However, with regrowth of the plume,
moderate concentrations developed during 1985 and 1986 in the top part of the aquifer,
very low concentrations were found in the middle part of the aquifer, and the highest
concentrations developed at the bottom of the aquifer. These findings suggest that a
significant accumulation or pool of DNAPL solvent exists at the bottom of the aquifer.
The source zone in the aquifer likely consists of multiple accumulations or layers of
DNAPL at various depths within the aquifer as well as a pool at the bottom of the aquifer.
These layers and pools were connected, at some point in time, by vertical pathways
("fingers") which likely drained to residual saturation. However, after as many as ten to
Sec. 13.6
467
Source Zone GW-32

14,000
End of On-Site
Pumping
12,000
10,000
i
w
8,000
6,000
Start of On-Site
Pumping
4,000
2,000
1=t=1=
co
"-
"-
co
ex;
C\l
='l
='l
='l
='l
='l
Q)
co
co
'"
='l
.,.co
~
co
<ll
to
<Xl
"<Xl
<Xl
='l
='l
='l
='l
cc
Date of 6-Montn Means
Plume Zone GW-16B

3,500
3,000
Start of On-Site
Pumping
End of On-Srte
Pumping
2,500
2,000
.:!,
w
o 1,500
n,
1,000
Figure 13.32 Regrowth of the dissolved plume following cessation of pump-and-treat

at the Dayton site as reflected by the increase in concentrations in well GW-32 close to
the source, and GW-16B downgradient of the source.
twenty years of dissolution by groundwater, it is cannot be known whether significant

zones of residual DNAPL in fingers still exist in the aquifer.
In 1990, the groundwater pump-and-treat system was reactivated with a network
of four pumping wells: one near the source, one close to the middle of the plume (near
GW-16B), and two at the eastern boundary of the plant site. The goals of this system
are containment of the plume to prevent further migration off-site, and restoration of the
aquifer in the current plume zone. It is hoped that in several years the dissolved plume
will be again eliminated, and containment of the source zone can be provided by a single
pumping well.
468
Chap. 13
o
JJ85 JD-85 JJ-86 JD86 JJ87 JD87
6
(J)
+-'
Q)
..........
.,.,-",-"-,.,,,.,.,-,
428
E
JJ85 JD-85 JJ-86 JD-86 JJ87 JD-87
12
15
18
42d
clay
JJ85 JD85 JJ-86 JD86 JJ87 JD-87
Date of 6Month Means
Figure 13.33 Vertical distribution of dissolved concentrations in the aquifer at the Dayton site
following the cessation of pump-and-treat.
13.7 CONCLUSIONS
In order to develop rational and cost-effective plans for the remediation of soil and
groundwater contamination at many industrial and waste-disposal sites, a first goal in
diagnosis and assessment is the determination of whether DNAPL is present in the
subsurface, particularly if it exists below the water table. Preliminary lines of evidence
that should be examined at any possible DNAPL site include the types and volumes of
the chemicals that may have been released to the subsurface. At an industrial facility,
this would include examination of how and where the chemicals were used, stored, and
disposed of at the site.
There are many types of site information that can contribute to the diagnosis and
assessment of the presence and distribution of DNAPLs in the subsurface. At many
sites, there is no direct visual evidence from soil cores or from samples from monitoring wells that DNAPL exists in significant quantities in the groundwater zone. However, because of the potential variability in the spatial distribution of DNAPLs and the
Sec. 13.7
Conclusions
469
investigative methods commonly used, this does not necessarily mean that DNAPL is
not present.
The investigative methods that can be used to improve the likelihood that DNAPL
will be identified if encountered in borings and monitoring wells include careful observations during drilling by means of visual inspection, enhanced visual inspection using
dyes or UV light, vapor analyses, the chemical analysis of soil and rock samples, the
use of borehole geophysics. The identification of DNAPL in wells can be made more
reliable by using appropriate well construction and well sampling procedures. Special
care during drilling and well construction in DNAPL zones is also necessary to prevent
"short-circuiting" of DNAPL deeper levels into the subsurface. In many circumstances,
drilling directly into suspected DNAPL zones should be avoided in favor of indirect
means which do not risk remobilization of the DNAPL. At many sites, uncertainties in
the location of any DNAPL source zone(s) are sufficiently great that inadvertent drilling
through DNAPL zones is very possible. There is therefore a need for improved methods
for drilling, sampling, and in situ analyses that will enable the detection of DNAPL while
drilling is in progress.
There are many investigative methods that can be used to indirectly assess the
presence and distribution of DNAPLs in the subsurface. These methods include soil gas
surveys and the evaluation of the character of the dissolved plumes emitted from the
source area. In support of these methods, we now have a sufficiently good understanding
of DNAPL migration patterns, dissolution processes, dissolved contaminant migration
processes, as well as adequate monitoring methods to allow a meaningful evaluation of
the possible presence of DNAPL at many field sites.
A conclusion that DNAPL is present in the groundwater zone will influence the
selection of remedial measures. For example, the mere existence of a DNAPL source
zone will usually eliminate any likelihood of complete aquifer restoration in the foreseeable future, and force reliance on remedial measures which provide containment of the
source zone. A recognition that DNAPL is present somewhere in the groundwater zone
will not be sufficient to permit a detailed analysis of past or future DNAPL migration.
Such analyses require information on the amounts and types of DNAPL released to the
subsurface, and the spatial distribution of DNAPL in the subsurface in the case of predictions of future migration. The selection and design of remedial measures which involve
containment and/or in situ treatment of DNAPL source zones will also require this level
of information. Unfortunately, at most sites, such information is difficult or impossible
to acquire.
We conclude that many lines of evidence can potentially contribute to the diagnosis
of DNAPL sites, but normally, only a few contribute strongly at a particular site. The
challenge is to assess the various lines of evidence and then render a judgment based on
the preponderance of all of the available evidence.
ACKNOWLEDGEMENTS
Solvents-in-Groundwater Research Program. Sponsors of the Program between 1988 and
470
Chap. 13
1994 have included: The Boeing Company, Ciba-Geigy Corporation, Dow Chemical
13.8 REFERENCES
Anderson, M. R., R. L. Johnson, and J. F. Pankow (1992) "Dissolution of dense chlorinated solvents
into groundwater. 3. Modeling of contaminant plumes from fingers and pools of solvent",
Brewster, M. L., A. P. Annan, J. P. Greenhouse, G. W. Schneider, and J. D. Redman (1992) "Geophysical detection of DNAPLs: Field experiments". In: Proceedings: International Association
of Hydrogeologists Conference, Hamilton, Ontario, May 1992, pp. 176-194.
Broholm, K., S. Feenstra, and J. A. Cherry (1994) "Solvent release into a sandy aquifer. 1. Overview
of source behavior and distribution", in preparation.
Cary, J. W., 1. F. McBride, and C. S. Simmons (1991) "Assay of organic liquid contents in
predominantly water-wet unconsolidated porous media", J. Contaminant Hydrology, 8, 135142.
Chiang, C. Y., K. R. Loos, and R. A. Klopp (1992) "Field determination of geological/chemical
properties of an aquifer by cone penetrometry and headspace analysis", Ground Water, 30,
428-436.
Chou, S. F. J. and R. A. Griffin (1986) "Solubility and soil mobility of polychlorinated biphenyls",
In: PCBs and the Environment, J. S. Waid (Ed.), CRC Press, Boca Raton, Florida.
Clausen, 1. L., J. Zutman, and N. Farrow (1993) Characterization of the Northwest Plume Utilizing
a Driven Discrete-Depth Sampling System. Report by Martin Marietta Energy Systems Inc. to
the U.S. Dept. of Energy, Document No. KYIER-22.
Cohen, R. M., A. P. Bryda, S. T. Shaw, and C. P. Spalding (1992) "Evaluation of visual methods
to detect NAPL in soil and water", Ground Water Monitoring Review, 12, No.4, 132-141.
Conant, B. H., M. M. Broholm, J. A. Cherry, and R. W. Gillham (1995) "Shallow groundwater
contamination in "the interface zone" associated with chlorinated solvent vapor transport", in
preparation.
de Pastrovich, T. L., Y. Baradat, R. Barthel, A. Chiarelli, and D. R. Fussell (1979) Protection
of Groundwater from Oil Pollution, CONCAWE, Water Pollution Special Task Force No. 11,
NTIS PB82-174608.
Dumford, D., J. Brookman, 1. Billica, and J. Milligan (1991) "LNAPL distribution in a cohesionless
soil: A field investigation and cryogenic sampler", Ground Water Monitoring Review, 11, No.2,
115-122.
Edge, R. W. and K. Cordry (1989) "The Hydrol'unch!": An in situ sampling tool for collecting
ground water from unconsolidated sediments", Ground Water Monitoring Review, 9, 177-183.
Einberger, C. M., G. C. Patrick, A. S. Burgess, and K. M. Angelos (1993) "Correlating soil and
soil vapor concentrations at suspect DNAPL sites: A case study", Ground Water, 31, No.5,
837 (abstract).
EPA (1989) Evaluation of Ground-Water Extraction Remedies: Volume 2. Case Studies /-19, EPA
540/2-89/054b, Washington, D.C.
EPA (1992) Evaluation of Ground-Water Extraction Remedies: Phase II, PB92-963346.
Sec. 13.8
References
471
(DNAPL) chemicals". In: Proceedings: International Groundwater Symposium, International
Association of Hydrogeologists, May 1-4, 1988, Halifax, Nova Scotia, pp. 62-69.
Feenstra, S., D. M. Mackay, and J. A. Cherry (1991) "Presence of residual NAPL based on organic
chemical concentrations in soil samples", Ground Water Monitoring Review, 11, No.2, 128-136.
Halogenated Solvents Industry Alliance (1987) personal communication.
Hardisty, P. E. and P. M. Johnson (1993) "Characterization of occurrence and distribution of
LNAPL in fractured rocks". In: Proceedings: Conference on Petroleum Hydrocarbons and
Organic Chemicals in Ground Water: Prevention, Detection and Restoration, Houston, Texas,
November 10-12,1993, pp. 113-129.
Howard, P. H. (1989) Handbook of Environmental Fate and Exposure Datafor Organic Chemicals
Volume I. Large Production and Priority Pollutants, Lewis Publishers, Chelsea, Michigan.
Howard, P. H. (1990) Handbook of Environmental Fate and Exposure Data for Organic Chemicals
Volume II. Solvents, Lewis Publishers, Chelsea, Michigan.
Hughes, B. M. (1991) Vapor Transport of Trichloroethylene in the Unsaturated Zone: A Field
Experiment, M. Sc. Thesis, Dept. of Earth Sciences, University of Waterloo, Waterloo, Ontario.
Hughes, B. M., R. W. Gillham and C.A. Mendoza (1992) "Transport of trichloroethylene vapours
in the unsaturated zone: A field experiment". In: Proceedings: International Association of Hydrogeologists Conference on Subsurface Contamination by Immiscible Liquids, Calgary, Alberta,
A. A. Balkema, Rotterdam, pp. 81-88.
Johnson. R. L. and 1. F. Pankow (1992) "Dissolution of dense chlorinated solvents into groundwater.
Karimi, A. A., W. J. Farmer, and M. M. Cliath (1987) "Vapor-phase diffusion of benzene in soil",
J. Environ. Qual., 16, 38-43.
Kueper, B. H. (1989) The Behavior ofDense Non-Aqueous Phase Liquid Contaminants in Heterogeneous Porous Media, Ph.D. Thesis, Dept. of Earth Sciences, University of Waterloo, Waterloo,
Ontario.
Kueper, B. H., W. Abbott, and G. Farquhar (1989) "Experimental observations of multiphase flow
in heterogeneous porous media", J. Contaminant Hydrology, 5, 83-96.
Kueper, B. H. and D. W. McWhorter (1991) "The behavior of dense, non-aqueous phase liquids
in fractured clay and rock", Ground Water, 29, 716-728.
Kueper, B. H., Redman, D., Starr, R. c., Reitsma, S., and Mah, M. (1993) "A field experiment to
study the behavior of tetrachloroethylene below the water table: Spatial distribution of residual
and pooled DNAPL", Ground Water, 31, 756-766.
Lawrence, A. R., P. J. Chilton, R. J. Barron, and W. M. Thomas (1990) "A method for determining
volatile organic solvents in chalk pore waters (southern and eastern England) and its relevance
to the evaluation of groundwater contamination", J. Comtaminant Hydrology, 6, 377-386.
Lewis, T. E., A. B. Crockett, R. L. Seigrist, and K. Zarrabi (1991) Soil Sampling and Analysis
for Volatile Organic Compounds, Ground-Water Issue Paper, U.S. Environmental Protection
Agency, EPN540/4-911001, Washington, D.C.
Litherland, S. T., T. W. Hoskins, and R. L. Boggess (1985) "A new ground water survey tool:
The combined cone penetrometer/vadose zone vapor probe". In: Proceedings: Conference on
Petroleum Hydrocarbons and Organic Chemicals in Ground Water Houston, Texas, National
Water Well Association, Nov. 13-15, 1985, pp. 322-330.
472
Chap. 13
Marrin, D. L. and G. M. Thompson (1987) "Gaseous behaviour of TCE overlying a contaminated

aquifer", Ground Water, 25, 21-27.
Marrin, D. L. (1988) "Soil-gas sampling and misinterpretation", Ground Water Monitoring Review,
8, 51-54.
Marrin, D. L. and H. B. Kerfoot (1988) "Soil-gas surveying techniques", Environ. Sci. Technol.,
22,740-745.
Mendoza, C. A., B. M. Hughes, E. O. Frind (1992) "Trichloroethylene vapours in the unsaturated
zone: Numerical analysis of a field experiment". In: Proceedings: International Association
of Hydrogeologists Conference on Subsurface Contamination by Immiscible Liquids, Calgary,
Alberta, A. A. Balkema, Rotterdam, 221-227.
Merck & Co., Inc. (1989) The Merck Index: An Encyclopedia of Chemicals, Drugs and Biologicals,
Eleventh Edition, Merck & Co. Inc., Rahway, New Jersey.
Montgomery, J. H. and L. M. Welkom (1990) Groundwater Chemicals Desk Reference, Lewis
Publishers, Boca Raton, Florida.
Montgomery, J. H. (1991) Groundwater Chemicals Desk Reference: Volume 2, Boca Raton, Florida.
Newell, C. and R. R. Ross (1991) Estimating Potential for Occurrence of DNAPL at Superfund
Sites, Quick Reference Guide Sheet, U.S. Environmental Protection Agency, publication number
9355.4-07FS, Washington, D.C.
Niemeyer, R. A., M. A. Palmer and D. R. Hargis (1992) "An innovative well design for DNAPL
recovery from the saturated zone", Ground Water, 30, 793 (abstract).
Pitkin, S., R. A. Ingleton, and J. A. Cherry (1994) "Use of a drive-point sampling device for
detailed characterization of a PCE plume in a sand aquifer at a dry cleaning facility". In:
Proceedings: Outdoor Action Conference, National Ground Water Association, Minneapolis,
Minn., May 23-25.
Poulsen, M. and Kueper, B. H. (1992) "A field experiment to study the behavior of tetrachloroethylene in unsaturated porous media", Environ. Sci. Technol., 26, 889-895.
Rivett, M. O. (1993) "A field evaluation of pump-and-treat remediation". In: Proceedings: National
Conference on Environmental Engineering, joint conference of the Canadian Society of Civil
Engineering and the American Society of Civil Engineering, Montreal, Quebec, July 12-14,
1993, pp. 1171-1178.
Rivett, M. O. (1995) "Soil-gas signatures from volatile chlorinated solvents: Borden field experiments". Ground Water, 33, 84-98.
Rivett, M. 0., S. Feenstra and J. A. Cherry (1995) "A natural gradient experiment on solute
transport from a residual solvent source. I. Overview", in preparation.
Rudolph, D. L., J. A. Cherry and R. N. Farvolden (1991) "Groundwater flow and solute transport
in fractured lacustrine clay near Mexico City", Water Resources Research, 27, 2187-2201.
Schneider, G. W. and J. P. Greenhouse (1992) "Geophysical detection of perchloroethylene in a
sandy aquifer using resistivity and nuclear logging techniques". In: Proceedings: Symposium
on the Application of Geophysics to Engineering and Environmental Problems, Society of Engineering and Mineral Exploration Geophysics, April 26-29, 1992, pp. 619-628.
Schwille, F. (1988) Dense Chlorinated Solvents in Porous and Fractured Media - Model Experiments, translated by J. F. Pankow, Lewis Publishers, Chelsea, Michigan.
Sec. 13.8
References
473
Siebenmann, K. (1993) "Problems with the use of near-surface soil gas results for source identification". Ground Water, 31, 850 (abstract).
Seigrist, R. L. and P. D. Jenssen (1990) "Evaluation of sampling method effects on volatile organic
compound measurements in contaminated soil", Environ. Sci. Technol., 24, 1387-1392.
Smolley, M. and J. C. Kappmeyer (1991) "Cone penetrometer tests and HydroPunch sampling:
A screening technique for plume definition", Ground Water Monitoring Review, 11, No.2,
101-106.
Sohn, M. D., S. R. Custance, J. Ryer-Powder, and M. J. Sullivan (1993) "Comparison of measured
and predicted trichloroethylene concentrations in soil gas resulting from ground water", Ground
Water, 31, 850 (abstract).
Starr, R. C. and R. A. Ingleton (1992) "A new method for collecting core samples without a drilling
rig", Ground Water Monitoring Review, 12, No. I, 91-95.
Taylor, T. W. and M. C. Serafini (1988) "Screened auger sampling: The technique and two case
studies", Ground Water Monitoring Review, 8, No.3, 145-152.
U.S. International Trade Commission (1991) Synthetic Organic Chemicals: United States Production and Sales, USITC Publication 2470, Washington, D.C.
Urban, M. J., J. S. Smith, and R. K. Dickinson (1989) "Volatile organic analysis for a sample
for a soil, sediment or waste sample", In: Proceedings: Fifth Annual Waste Testing and Quality
Assurance Symposium, U.S. EPA, Washington, DC, pp. II87-IIlOI.
Voice, T. C. and B. Kolb (1993) "Static and dynamic headspace analysis of volatile organic
compounds in soils", Environ. Sci. Technol., 27, 709-713.
Westinghouse Savannah River Company (1992) Assessing DNAPL Contamination, AIM Area, Savannah River Site: Phase I Results, Report to the U.S. Department of Energy, Report No.
WSRC- RP-92-1302.
Zapico, M., S. Vales, and J. A. Cherry (1987) "A wireline piston core barrel for sampling cohesionless sand and gravel below the water table", Ground Water Monitoring Review, 7, No.3,
74-82.
14
Concepts for the Remediation
of Sites Contaminated with
Dense Non-Aqueous Phase
Liquids (DNAPLs)
John A. Cherry 1 , Stan Feenstra 1 ,2, and Douglas M. Mackay 1
1 Waterloo
Centre for Groundwater Research

Waterloo, Ontario
Canada N2L 3Gl
2Applied Groundwater Research, Ltd.

2550 Argentia Road
Mississagua, Ontario
Canada L5N 5Rl
Page
14.1
14.2
14.3
14.4
14.5
14.6
14.7
14.8
14.9
Introduction
Remediation Goals
Pump-and-Treat
In-Situ Treatment for Migration Control
14.7.2.1 General
14.7.2.3 In-situ mass removal
14.7.2.4 Mass-destruction technologies
Conclusions
References
476
478
482
482
482
483
484
489
491
494
494
496
496
496
497
501
502
503
505
475
476
Concepts for the Remediation of Sites Contaminated with DNAPLs
Chap. 14
ABSTRACT
At each site where a dense non-aqueous phase liquid (DNAPL) has contaminated the
local groundwater, there are two principal components to the problem: a subsurface
source zone and a groundwater plume. Usually, most of the contaminant mass is in
the source zone, but at solvent sites, the plume usually occupies a much larger portion
of the aquifer. Therefore, for most sites, unless essentially all of the contaminant mass
(>99.9%) is removed from the source zone, permanent aquifer restoration to drinking
water standards will not be achievable in the near term. Pump-and-treat methodologies
remove contaminant mass too slowly, and excavation is generally not practical because
of the depths to which the DNAPL has moved.
New technologies being developed for source-zone restoration are of two types,
those that bring contaminant mass to the surface for treatment or disposal (e.g., the
flush technologies that use steam, air, surfactants, or co-solvents), and those that destroy DNAPLs in situ (e.g., chemical oxidation, chemical reductive dehalogenation, and
enzyme-catalyzed reductive dehalogenation.) The efficacies of the in-situ technologies
have not yet been assessed in the field. Heterogeneities in the geology and in the DNAPL
distribution will likely often severely limit the performance of remediation technologies,
especially the flush technologies.
Pump-and-treat and cutoff-wall enclosures used alone or in combination with other
remediation steps are proven technologies for source-zone containment. However, while
pump-and-treat can be used for containing or removing plumes, its long-term costs are
often large due to the stringent water treatment standards that are frequently applied.
Moreover, in cases where diffusion from many low-permeability zones severely restricts
the rate of mass removal, plume removal by aggressive pump-and-treat can take many
decades or longer.
In-situ treatment in conjunction with cutoff walls and funnel-and-gate systems comprise new systems to control dissolved contamination. The contaminant flux in most
plumes is sufficiently small for these in-situ treatment systems to offer good potential
for the containment of plumes in cases involving contamination in shallow porous-media
aquifers. In contrast, for most fractured rock sites and for porous-media sites with deep
DNAPL, no permanent restoration technologies exist and pump-and-treat remains as the
only proven containment option.
14.1 INTRODUCTION
The realization that dense non-aqueous phase liquids (DNAPLs) are common at industrial and military sites and that they pose exceptional problems at these sites came
gradually, and evolved from a recognition by a few groundwater scientists in the early
1980s to broad recognition by the groundwater profession in the early 1990s (see the
related discussion in Chapter 1). Subsurface contamination at DNAPL sites poses much
greater technical problems and financial burdens than exists at sites contaminated with
other materials. Most sites in North America that have received or will soon receive
major remediation work are afflicted with DNAPLs, the most common being chlorinated
Sec. 14.1
Introduction
477
solvents and mixed organic liquid industrial wastes that are quite similar to chlorinated
solvents. Meeting the clean-up criteria specified by regulatory agencies has been and will
continue to be elusive at most DNAPL sites. Progress towards the cost-effective management of this immense groundwater problem has been slow. A major impediment to
progress has been the difficulty of moving beyond conventional concepts and strategies
for subsurface remediation to new approaches best suited for the particular characteristics of DNAPL problems. Approaches that have long served for remediation of sites
contaminated with light non-aqueous phase liquids (LNAPLs) and for the remediation
of non-NAPL sites are generally ineffective for DNAPL sites. Advances in technology
for the remediation of DNAPL sites are urgently needed.
The immense need for effective subsurface remediation strategies for either containment or restoration at DNAPL sites has resulted in the accelerated application of new
and emerging technologies without the attention to independent and rigorous performance
assessment that technology developments normally require. We do not attempt in this
chapter to describe technologies in detail or to evaluate technologies. Rather, our goal is
to describe conceptually the nature of subsurface remediation problems at DNAPL sites
as well as the remediation approaches and technologies that are now available or that are
being developed. For more detailed descriptions of many of the technologies, the reader
is referred to the report of Grubb and Sitar (1994).
For our discussions of DNAPL site remediation, it is important to establish the
following definitions for technology development:
Proven Technology: A technology for which: 1) a considerable base of site experience and success currently exists; 2) commercial organizations offer the technology
in the marketplace; and 3) the performance of the technology is reasonably predictable.
Emerging Technology: A technology for which considerable laboratory research
and development have been conducted, and the technology has been used with
encouraging results at one or more field sites. The technology is available in the
commercial marketplace, but actual site performance and costs for specified performance have considerable uncertainty. More research and development and detailed
assessments of field trials are needed.
Experimental Technology: A technology which is new and offers promise, but
which is still being investigated primarily in the laboratory and has had few, if any,
successful prototype or pilot-scale field trials.
Figure 14.1 describes the stages according to which remediation technologies can
evolve from the initial experimental stage, through the emerging stage, to the proven
technology stage. For a technology to become proven, it must evolve from the bottom
of the Figure 14.1 triangle, where numerous technologies can exist as concepts or in the
initial experimental stage, to the top of the triangle, where few arrive because of technical failure or excessive cost. To enhance commercial prospects, technology vendors
commonly claim that their technologies rank higher in the progression of categories on
the triangle than the actual performance data warrant. Moreover, there is a paucity of
478
PROVEN
TECHNOLOGY
COMMERCIAL
ASSESSMENT
SCALE - UP
ENGINEERING
EMERGING
TECHNOLOGY
PILOT- SCALE
SITE TRIALS
EXPERIMENTAL
TECHNOLOGY
Chap. 14
SMALL- SCALE FIELD

EXPERIMENTS I PROTOTYPES
LABORATORY PROOF OF CONCEPT
RIGOROUS
MASS BALANCE
STUDIES
CONCEPT
SCREENING
CONCEPT INDENTIFICATION
Figure 14.1 Stages in the evolution of a new technology for site remediation; many
technologies exist in the concept stage, but few survive to the proven stage.
published, peer-reviewed literature on new and emerging technologies for which successful application to DNAPL sites has been proven.
14.2 CHARACTERISTICS OF DNAPL SITES
The major differences between groundwater contamination at non-NAPL, LNAPL, and
DNAPL sites are shown in Figure 14.2. DNAPL sites usually differ from non-NAPL and
LNAPL sites by the vertical position within the hydrogeologic system of the principal
source of the contamination. In particular, at most DNAPL sites, much or all of the
contaminant source mass causing the plume is DNAPL situated below the water table.
Commonly, much of the mass is far below the water table where it causes deep groundwater contamination. In contrast, at non-NAPL sites, waste material or contaminated soil
that cause groundwater contamination are usually situated entirely above the water table.
At most LNAPL sites, the source material is hydrocarbon liquid which can be found in
the soil zone, in the capillary fringe, and in the upper part of the groundwater zone.
Many industrial sites are characterized by a combination of two or three of the
conceptual models shown in Figure 14.2. For example, at some non-NAPL sites, a
portion of the waste may be below the water table. Horizontal groundwater flow through
the waste then contributes contaminant mass to the plume in much the same way as does
flow through DNAPL below the water table. At some sites, both LNAPL and DNAPL
are present, though we note that the DNAPL is usually the cause of the most severe or
extensive groundwater contamination.
Permanent subsurface restoration of a typical non-NAPL site is relatively simple,
although not necessarily inexpensive. Remediation possibilities include: 1) remove the
contaminated soil or waste by excavation; or 2) remove the soil/waste contaminant mass
by in-situ methods so that the "source" is removed, then extract the plume by pump-
Sec. 14.2
479
(a)
RAIN
CONTAMINATED SOIL I WASTE
~~~---------~---. ....
D
-..
PL~ME..~
(b)
LNAPLPOOL
,~~~:=-:::..
SZ
BTEX PLUME
(e)
=-=,","",':="'= LEAKY TANK
=_~_=_=
TCE PLUME
Figure 14.2 Schematic representations of simple cases for point-source groundwater

contamination: a) non-NAPL case; b) LNAPL case; and c) DNAPL case.
and-treat, or let it dissipate by natural groundwater flushing. Alternatively, if there

are no volatile contaminants, and if the contaminants are located only in the vadose
zone, the ability of the source to generate a plume can be eliminated by placing an
impervious cover over the source area. The benefits to the groundwater environment
provided by impervious covers on non-NAPL sites of the type shown in Figure 14.2.a
are clear. However, the benefits to groundwater from impervious covers at DNAPL sites
are generally much less evident. At many DNAPL sites, such covers will provide no
measurable reduction in groundwater contamination.
At LNAPL sites, significant contaminant source mass commonly extends all the
way down to the water table, thereby causing excavation to be difficult or excessively
expensive, and impervious soil covers to be less effective for source control. Fortunately,
480
Chap. 14
most LNAPLs are comprised of petroleum hydrocarbons for which natural microbiological processes cause gradual but effective destruction of the plume and ultimately of the
source zonel. As a result of biodegradation, an unremediated plume from an LNAPL
source rarely extends more than a few hundred meters from the source zone. Provision
of oxygen to an LNAPL source zone and/or to its plume will often enhance the rate of
subsurface restoration.
At most DNAPL sites, the source zone has multiple DNAPL entry points or
entry areas where the DNAPL was spilled onto soil or leaked into the soil (Figure 14.3).
Source zones may have buried drums, sludge, or other waste material located primarily
DNAPL ENTRY
LOCATIONS
(a)
DISSOLVED PHASE PLUME
-J;;:--b
t L'L-+-_--:::-r I
GROUNDWATER - ,
FLOW
PROPERTY
BOUNDARY
~,-",
)
_......... /-..
EXTERIOR OF
DNAPL SOURCE ZONE
PLUME OUTLINE
(USUALLY BASED ON
DRINKING WATER MCL
OR DETECTION LIMIT)
FRONT ADVANCES
TOWARDS HYDROLOGIC
BOUNDARIES SUCH AS
WELLS, OR STREAMS
(b)
INFILTRATION
SLUDGE
FILL
DRUMS
CONTAMINATED
SOIL
---.
GROUNDWATER
ZONE
. ---.
DISSOLVED
CONTAMINATION - - - .
.---.
-'"'''-.......II!iI!IilIlIl_ _...;..........,;---.
IMPERVIOUS LAYER
Figure 14.3 Anatomy of a DNAPL site on unconsolidated deposits: a) plan view; and
b) cross-sectional view.
tIn this chapter, we use the term source zone to refer collectively to the entire subsurface zone where
DNAPL contamination exists, including both the vadose and groundwater zones.
Sec. 14.2
481
or exclusively in the vadose zone, as well as DNAPL below the water table. If DNAPL
is below the water table, that portion of the source has typically contributed most of
the mass to the plume and presents the greatest obstacles to remediation. Contaminant
contributions to groundwater from DNAPL in the vadose zone cause shallow plumes
at some sites. These shallow plumes are likely to terminate at local natural hydrologic
barriers such as streams, while contamination from DNAPL below the water table is more
likely to feed plumes that follow regional groundwater ftowpaths extending to major
rivers, lakes, or wells beyond the local hydrologic barriers. Therefore, the remediation
of deep DNAPL source zones can be particularly important but difficult.
At DNAPL sites on porous media such as sand or gravel, the source zone includes
the entire volume wherein DNAPL is found. However, the regulatory and legal communities have more generally perceived the "significant source" on an industrial property to
be buried drums or contaminated soil or sludge which are close to the ground surface, and
typically all above the water table. Since regulatory and legal viewpoints have governed
most remedial actions, at many DNAPL sites, large remedial efforts have commonly
been directed at the drum or sludge portion of the source, with little effort directed at
the DNAPL below the water table. At many DNAPL sites, this confusion in the concept
of source has resulted in remediation efforts that, while very expensive, have provided
minimal risk reduction or groundwater protection.
Defining the source zone at solvent DNAPL sites on fractured porous media (e.g.,
fractured clayey deposits or fractured porous sandstone or shale) is complicated because
the DNAPL mass initially in the fractures may have been depleted fairly quickly by
diffusive transfer to the dissolved and sorbed phases in the matrix (see Chapters 2 and
12, as well as Parker et al., 1994). Therefore, in a fractured porous medium, the size of
the subsurface zone in which DNAPL exists can diminish with time. In some cases, the
DNAPL can disappear altogether. However, even with total DNAPL disappearance, the
initial zone will contain essentially all of the contaminant mass, and will thus represent a
long-term source zone as contaminants are slowly released by diffusion from the matrix
to groundwater flowing through the fracture network. As noted above, at DNAPL sites
on porous media such as sand or gravel, the "source zone" includes the entire volume
wherein DNAPL is found. In contrast, at DNAPL sites on fractured porous media, we use
the term "source zone" to refer to the zone containing the relatively large and immobile
or nearly immobile contaminant mass, regardless of whether DNAPL still exists in the
fractures.
Groundwater plumes have two components of contaminant mass, the dissolved and
sorbed components. Estimates of the dissolved contaminant mass in large plumes in sand
and gravel aquifers at solvent DNAPL sites are in the range of 0.5 to 70 drums (55 US
gallons) of equivalent DNAPL (see Chapter 2). Annual plume fluxes are usually onetenth or less of the total dissolved mass. The amount of the sorbed mass in these plumes
is generally unknown; for moderately hydrophobic DNAPLs such as trichloroethylene
(TCE), tetrachloroethylene (PCE), and I, I, I-trichloroethane (1, 1,1-TCA), it is expected to
be of about the same order of magnitude as the dissolved mass. To permanently clean up
a plume, sufficient dissolved and sorbed mass must be removed so that aquifer-restoration
criteria can be met even if some remaining mass desorbs to the groundwater.
482
Chap. 14
14.3 REMEDIATION GOALS

Risks from groundwater contamination can be reduced or eliminated by controlling the
migration of the aqueous- and vapor-phase plumes to achieve plume containment. By
definition, when complete or full containment of a plume is achieved, the affected volume
of the aquifer ceases to grow. In some cases, the condition of complete containment
is imposed naturally by hydrologic boundaries. An example would be a plume that
discharges into a nearby stream (Figure 14.4.a). Although such a plume may cause the
contamination of surface water, the zone of aquifer contamination does not expand.
At many DNAPL sites, complete containment cannot be achieved unless engineered
controls (e.g., pump-and-treat) are applied for as long as the DNAPL source persists
(Figure 14.4.b). At many sites, this is expected to be many decades or longer.
(a) NATURAL HYDROLOGIC BOUNDARY

PLUME ENTRY
ACROSS RIVER
BOnOM
FLOW
PLUME
- - - - - - - PLUME
CONTAMINANTS
IN RIVER
PUMPING WELL
FOR PLUME FRONT
CONTAINMENT
(b) PUMP AND TREAT
L!iI.
~ ~_L
__
FLOW
PLUME
:..-
.3.J
_
Figure 14.4 Plume containment: a) a river as a natural hydrologic boundary; and b)

pump-and-treat captures the front of a plume.

Aquifer restoration is much different from plume containment. Complete aquifer restoration requires subsurface source removal as well as plume removal to a degree sufficient
to allow the original beneficial use of the aquifer, which is usually as a source of drinking
water. A completely restored aquifer is one that requires no on-going remedial activity.
To accomplish complete aquifer restoration at a DNAPL site, the plume must be removed, either by groundwater extraction, by in-situ treatment, or by natural flushing.
Sec. 14.3
483
Remediation Goals
Sufficient contaminant mass must also be removed from the source-zone so that groundwater flowing through the source zone no longer creates a plume downgradient of the
source zone at concentration levels above regulatory limits. Of the thousands of sites
that are contaminated with DNAPLs in North America, while successful measures for
source zone or plume containment are on-going, none has been fully restored to drinking
water standards. Generally, claims of complete aquifer restoration have either not been
confirmed by adequate testing, or the cleanup requirements have been relaxed to values
above drinking water standards (National Research Council, 1994).
Although complete aquifer restoration is an elusive goal, better prospects exist for restoring parts of aquifers by removing all or part of the plume while at the same time isolating
the source zone in place. For a portion of the aquifer to remain restored, contamination
from the source zone must not be allowed to enter that portion: therefore, source-zone
containment using proven technologies is needed. The source zone can be isolated
from groundwater flow by a low-permeability barrier, or by hydraulic capture using
groundwater extraction wells (Figure l4.5.a,b). Once source-zone containment is effected,
----.. ;,
(a)
(b)
----..
----.. ----..
----..
~
----..
'-..
"
..........
(c )
.~
FLOW
----..
---i~
. .r,
. l
(d)
----..
----..
----..
FLOW
PERMEABLE
REACTION CURTAIN
FUNNEL & GATES
Figure 14.5 DNAPL source zone isolation provided by: a) a low permeability enclosure; b) hydraulic containment by pump-and-treat; c) a permeable reaction curtain; and
d) a funnel-and-gate.
484
Chap. 14
plume removal can be accomplished naturally, as the plume will eventually disappear
as it is flushed by groundwater through the aquifer to a natural hydrologic barrier or into
water supply wells. After the source zone is isolated, pump-and-treat can play one of two
roles: 1) plume-front capture, until natural flushing removes the contamination extending
from the source zone to the plume front; or 2) aggressive removal of the plume using
more pumping wells placed throughout the plume (Figure 14.6). In the second case,
removal of the plume is accomplished sooner, but by a greater effort. In addition, the
aggressive pumping causes much uncontaminated water to be removed from the aquifer
and mixed with plume water. This can represent a substantial waste of the water resource
if the treated water is discharged to sewers or surface water rather than being used or
reinjected into the aquifer.
(a)
(c)
(b)
z
o
~
ex:
CONTAINMENT
WELL
I-
z
o
ui
so
------MCL
10
100
;'i:
EXTRACTION
WELLS
@8~O~
~~P",,~MCl
10
100 years
Figure 14.6 Remediation of a source-zone

and its plume by pump-and-treat: a)
containment well at the source zone
and multiple wells in the plume for
aggressive plume removal; b) response of
containment well; and c) response envelope
for plume-extraction wells.
14.4 LOW PERMEABILITY ENCLOSURES
A low permeability enclosure placed around a DNAPL source zone will divert much of
the groundwater flow around the enclosure. The lower the permeability of the enclosure
walls, the smaller will be the flow through the DNAPL zone inside the enclosure (Figure 14.7). If desired, enclosure walls of very low permeability can be constructed. Many
types of engineered barrier technologies are available for creating low-permeability enclosures around DNAPL source zones (Table 14.1). Mutch and Ash (1993) describe
several of these technologies. Most types of barriers are proven tecbnologies in that,
when constructed with stringent quality-control procedures, they can be depended upon
to provide very low permeability enclosures for a long time. Two main design options
Sec. 14.4
485
(a)
-[j)-
DNAPL
(b)
(c)
Figure 14.7 Decreasing groundwater

flow through a source-zone enclosure with
decreasing hydraulic conductivity of the
enclosure walls.
exist for enclosures around source zones: the keyed enclosure, and the hanging enclosure (Figure 14.8). The keyed enclosure has the wall set into a low-permeability layer
(aquitard) at the bottom of the aquifer, whereas the bottom of the hanging enclosure
does not terminate in a low permeability zone. The hanging enclosure has the wall set
at some level above the bottom of the aquifer. To provide a high degree of source-zone
containment, the enclosure need only extend somewhat deeper than the bottom of the
DNAPL source zone, particularly in cases where pumping inside the enclosure is used to
cause upward flow at the bottom of the enclosure. However, higher rates of groundwater
pumping will be required inside a hanging enclosure to provide effective containment at
the bottom of the enclosure. At some sites, the DNAPL zone does not reach the bottom
of the aquifer and therefore the cost of the enclosure can be minimized because the
enclosure need not penetrate to the bottom of the aquifer.
Enclosures are not a practical option at many sites. In cases where the bottom of the
DNAPL zone is very deep, enclosure construction to depths below the DNAPL zone is
not feasible. At other sites, difficulties in determining conclusively the maximum depth of
DNAPL causes unacceptable uncertainty for specification of the enclosure depth. Of the
technologies listed in Table 14.1, only borehole grouting is capable of creating enclosures
in hard bedrock. Where DNAPL occurs in fractured bedrock, the high cost of creating
enclosures by borehole grouting usually renders enclosures impractical, in which case
the only option available for source zone containment is groundwater pump-and-treat.
In situations where there is no pumping in or near a low permeability enclosure,
inward groundwater flow occurs across the upgradient enclosure wall face; outward flow
486
Chap. 14
TABLE 14.1 Major types of low-permeability cutoff walls for source-zone enclosures and
funnel-and-gate systems in unconsolidated deposits and soft bedrock.
General type
Materials options
Construction
Comments
I. Slurry Trench
soil/bentonite
soil/cement
soil/attapulgite
backhoe
backhoe and
clamshell
thick wall (0.5 to 1m),

large volume of earth
removed for disposal
2. Vibrated Beam Wall
bentonite
bentonite/cement
asphaltic mixtures
driven beam with

injection down
annulus of beam
thin wall (0.1 m),

no soil removed for
surface disposal
3. Plastic Sheets in
Slurry Trenches
high density
polyethylene
(HDPE) or PVC
sheets
sheets inserted into

slurry trenches;
sealed joints
between sheets
thin plastic sheets In

trench (0.3 to I m), large
volume earth removed
for disposal, exceptionally low permeability
4. Jet Grout Walls
bentonite/cement
mixture
high-pressure injection of slurry down

drill rods; drill holes
spaced 0.3 to 0.1
m apart
minimal soil removed

in return to surface of
excess slurry from
boreholes
5. Auger Mix Wall
soil/bentonite
bentonite/cement
injection of sealant
into auger mix zone
thick wall, minimal soil

removed
6. Steel Sheet Piling:

Conventional
steel sheets; thickness

options from 4 to 15
mm
driven directly into

ground using
electric hammer
vibration or diesel
hammers
no earth material
removed from ground;
cannot be driven
through large boulders,
hard soil or rock
7. Steel Sheet Piling:

Sealable Joints
same as in type 6,
except that joints have
larger annulus that is
grouted with bentonite,
cement, epoxy or other
sealants
same as in type 6
same as in type 6; the

annulus in joints can
be inspected before
grouting; exceptionally
low permeability
carries contamination across the downgradient wall face (Figure 14,7). Side faces can
have areas of inwards and outwards flow, The outflow parts of the enclosure cause
formation of a plume downgradient of the enclosure (Figure 14.9).
The normal design goal for enclosures is the achievement of extremely low permeability so that contaminant advection across the barrier will be insignificant. However,
molecular diffusion is another mechanism for contaminant migration across enclosure
walls. Chlorinated solvents can diffuse through slurry walls and even through high density polyethylene walls (HDPE). Devlin and Parker (1995) have shown that molecular
diffusion can cause outwards diffusion across low-permeability walls even when the
Sec. 14.4
487
LOW PERMEABILITY
WALL
(a)
__ 7 __
FLOW
AQUIFER
..:::.:.::..:......
AQUITARD
(b)
__ 7 __
'.~'
FLOW
:~i~EL i%rr.'
'~
Figure 14.8 Cutoff wall enclosure: a)

keyed enclosure-groundwater is diverted
laterally around the enclosure; and b)
hanging enclosure-groundwater is diverted
laterally and vertically (downward) around
the enclosure.
Figure 14.9 A plume of dissolved

contamination downgradient of a
source-zone enclosure. Dissolved
contaminants diffuse through the walls to
cause the plume. (Adapted from Devlin and
Parker, 1994).
hydraulic gradient is directed inwards due to pumping inside the enclosure. Indeed,
very low hydraulic conductivity increases the importance of outward diffusion by reducing inward advection, For any set of conditions, there is an optimum hydraulic
conductivity for the enclosure wall for which both higher and lower values result in
488
Chap. 14
greater contaminant flux through the wall. Steel sheet pile walls with grouted joints can
prevent both advective and diffusive flux through the enclosure walls.
Enclosures around source zones need not provide zero outward contaminant flux in
order to effectively isolate the source. The enclosure must only limit the outward flux sufficiently to result in concentrations that meet regulatory standards at a specified location
outside of the enclosure, e.g., at the downgradient site boundary. Figure 14.9 illustrates the
formation of a plume from an enclosure in a uniform flow field in a moderately permeable
sandy aquifer. Contamination diffuses outwards through the enclosure walls where it is
transported downgradient, forming a narrow, almost linear plume following the groundwater flowpath. Whether or not this plume exceeds water-quality criteria depends on many
site specific factors such as source concentration(s), aquifer dispersivity, the regulatory
standard(s) for the contaminant(s), and the spatial scale of the groundwater sampling.
Under conditions of natural groundwater flushing, many DNAPL source zones will
probably require centuries or longer for natural restoration (see Chapters 2 and 7). When
surrounded by an enclosure, the DNAPL will persist much longer. Therefore, questions
arise regarding the longevity of enclosures. Predictions are uncertain, but expectations
for most wall materials are a century or longer if the walls are designed initially to
be chemically compatible with their subsurface environment. It is reasonable to expect
that slow deterioration over hundreds of years will eventually make replacement of the
enclosure necessary, unless new technologies to destroy or remove the DNAPL become
available at or before that future time.
At many sites, the DNAPL source zone has no readily-definable boundary, particularly when the DNAPL is a chlorinated solvent because mobile or even residual DNAPL
is seldom encountered directly in boreholes or wells. The presence and approximate
spatial distribution of DNAPL must usually be inferred based on indirect evidence
(see Chapter 13). Obviously, for a low-permeability enclosure to be most effective, the
enclosure must surround all of the DNAPL. To accomplish this, the DNAPL residuals
or accumulations need not be located precisely. It is sufficient to drill at or beyond the
outermost boundary of the suspected DNAPL zone to establish a circumference which
DNAPL has not crossed. However, the enclosure designed on this basis would be larger
than would be needed if the exact morphology of the DNAPL zone were known. There
will be a point at which the cumulative cost of determining the precise extent of the
DNAPL zone exceeds the incremental cost of a larger enclosure.
Pump-and-treat and low-permeability enclosures are both proven technologies for
source-zone containment. Pump-and-treat can be used in any hydrogeologic environment
whereas enclosures are most practical for sites where the DNAPL zone is not deep and
is not in bedrock. At sites where enclosures around DNAPL source zones are technically
feasible, cost is usually the reason for selecting an enclosure rather than pump-and-treat as
the primary technology for source zone containment. Enclosures have larger initial (capital) costs, but lower long-term costs for operations and maintenance (O&M) than pumpand-treat. Depending on the capital costs, O&M costs, and interest rates, the predicted
cumulative cost (present worth) of the enclosure may be lower in some cases than the
predicted cumulative cost of pump-and-treat for operation times exceeding some critical
duration (T-critical in Figure 14.10). A low-permeability enclosure may also be selected
Sec. 14.5
489
Pump-and-Treat
en
c:
~
'E
~~
6
/
::cUi
~o
a:
--\'"' .........
~---\
Of!?
3:~ 4
~.2
z~
w
en
w
a:
a..
CUTOFF WALL ENCLOSURE

"T" CRITICAL
8
12
OPERATION TIME (years)
16
Figure 14.10 Comparison of hypothetical future cost relations for pump-and-treat system vs. a low-permeability enclosure for DNAPL source zone isolation.
because it may facilitate the use of DNAPL removal or destruction technologies inside
the enclosure, or because it would be perceived by the public as more secure containment.
14.5 PUMP-AND-TREAT
There have been numerous attempts across North America to restore DNAPL source
zones by pump-and-treat. Figure 14.6 shows the typical concentration response of a
pumping well near a source zone compared to the response of wells in the plume after
the source zone has been contained by pump-and-treat. At early time, wells at the DNAPL
source zone show a decline in concentration due primarily to dilution caused by capture
of clean water with source-zone water. There is little or no subsequent reduction of
concentration except over very long time. Concentrations in the pumping wells in a
source zone can be higher than those observed in nearby monitoring wells because
pumping draws flowlines from the source zone.
Pumping wells in or near solvent DNAPL source zones commonly exhibit concentrations in the range of 1 ppm (mg/L) to 100 ppm of total dissolved solvents. Wells with
this concentration range pumping at a relatively high rates of say 1000 Umin (250 gpm),
would extract between 2 and 200 drums of equivalent DNAPL mass per year. Clearly,
whether or not this annual mass extraction rate permits significant progress towards
source-zone restoration depends on the total mass of solvent in the source zone. Unfortunately, that source-zone mass is rarely known, even within an order of magnitude.
The fact that pump-and-treat systems have been operating for more than a decade at
many DNAPL sites without appreciable concentration decline is evidence that a large
ratio of source-zone mass to annual mass removed is typical at DNAPL sites. The rate
of mass removal at some sites can be increased by adding more pumping wells in or
near the source zone. However, the increase in mass removal will not be proportional
to the increase in the groundwater pumping rate because of limitations on the DNAPL
490
Chap. 14
dissolution kinetics, and because of further dilution with clean water drawn from outside
the source zone (Figure 14.11.c).
As noted above, the main capability that pump-and-treat offers at DNAPL sites is
plume containment or source zone containment. Therefore, the tactic should be to pump
at the lowest rate needed to achieve the desired capture of the contaminant flux from
the source zone. At most sites this pumping needs to yield a dissolved contaminant flux
of less than a few drums of equivalent DNAPL per year to achieve complete plume
capture or complete source zone containment. The rate of contaminant mass removal is
then equal to or only slightly greater than the mass flux emanating from the source zone
under natural conditions (Figure 14.11.b).
The segment of aquifer occupied by the plume at some sites is sufficiently large
and the groundwater resource sufficiently valuable to warrant removal of the plume
by "aggressive pump-and-treat" (Figure 14.6). For this purpose, wells are distributed
throughout the plume to remove multiple pore-volumes (i.e., plume volumes) of contaminated water much more rapidly than occurs by natural flushing. Clean water from outside
the plume is drawn into the plume zone as the plume is removed. If prior isolation of the
source zone has occurred, the pump-and-treat system need operate only until sufficient
- : -;.,: :-. .
*6
(b):
PLUME
CAPTURE
~.y WELL
. ...
.......
DNA~L .~~)
.......
_._._._~
(c)
.......
==-=-.0:- - - . . . . . . .
--+- __~y
. .:.~:It)
DNAPL
~;<;>r~
.... ~/"l
--+-----;//
-~-------
CLEAN
WATER
.......
Figure 14.11 Source-zone capture with

wells illustrating: a) an uncontrolled plume;
b) a well providing capture at the minimum
pumping rate; and c) excessive pumping
causing capture of clean water along with
clean groundwater.
Sec. 14.6
491
water has flushed the plume zone to cause contaminant concentrations to decline and
then remain below the regulatory value. The wells can then be shut off. The aquifer
beyond the source zone then remains restored for as long as the source zone containment
is maintained.
Plume restoration by aggressive pump-and-treat after source-zone isolation is simple in concept, but in practice there are several factors that commonly extend the time
period needed to attain regulatory limits throughout the plume, particularly if the limits
specified are the ppb (fLgIL) levels of drinking water standards discussed in Chapter 2.
These factors include non-optimal location of extraction wells, slow desorption of contaminants from soil particles, and slow release of contaminants by diffusion from low
permeability strata or lenses, or from the low-permeability matrix in a fractured porous
medium. In a situation where much of the contaminant mass is stored in the lower
permeability portions of an aquifer, slow diffusion out of these zones causes the decline in plume concentrations to approach the required restoration level only very slowly
(Figure l4.6.c). In some heterogeneous granular aquifers and in many fractured porous
aquifers, the prospects for attaining restoration levels by aggressive pump-and-treat within
several decades are uncertain, even if the source zone has been successfully isolated.
14.6 IN-SITU TREATMENT FOR MIGRATION CONTROL
Figure 14.5 illustrates four approaches for the containment of DNAPL source zones. Two
of the four are the proven approaches discussed above (Figures l4.5.a,b). The other two
(Figures l4.5.c,d) are emerging technologies, known as the in-situ treatment curtain
and the funnel-and-gate system. In the latter two systems, the plume moves due to the
natural hydraulic gradient through an emplaced permeable zone (curtain or gate). The
zone can be designed to: 1) destroy or immobilize contaminants using reactive media;
2) volatize contaminants (Pankow et al. 1993); or 3) add amendments to the water to
support biodegradation downgradient (see also below).
The permeability of the curtain or gate should be equal to or greater than the natural
permeability of the surrounding aquifer material. In concept, the contaminated water that
passes through the curtain or gate forms a different plume in the downgradient part of the
aquifer, one in which the chemical quality of the water at some point meets regulatory
standards even though the water chemistry may differ from the ambient water in the
aquifer. In practice, use of treatment curtains will be limited to shallow plumes due to
cost and installation difficulties.
At some sites where plumes are wide and/or deeper, funnel-and-gate systems offer
cost advantages. The funnel is a low-permeability wall placed across the plume with one
or more gaps (i.e., gates) through which plume water flows while receiving treatment.
The barriers available to form the low-permeability portion of a funnel are those listed
in Table 14.1. Treatment curtains and funnel-and-gate systems are emerging technologies
in the sense that the treatment materials and gate designs are still the subject of intensive
research and, as of 1995, only a few have been installed at field sites.
Research on treatment curtains and funnel-and-gate systems for organic contaminants began in the late 1980s with the first publications appearing a few years later (Burris
492
Chap. 14
and Cherry, 1992; Pankow et al., 1993; Starr and Cherry, 1994). Treatment curtains or
gates may in principle penetrate the entire aquifer, or they can be partially penetrating
or "hanging" to a depth sufficient to treat the entire plume (Figure 14.12). A funneland-gate system causes a rise in the upgradient water table which then induces some
downward flow at the gate. The bottom of the funnel-and-gate must therefore penetrate
to an impermeable layer, or to an appropriate depth below the plume so that no part of
the plume passes beneath the gate.
(a)
FUNNEL
-- --------PLUME
WITHOUT
CRITICAL
CONTAMINANTS
<,
-----------
( b)
K EY ED GAT E
'j)
'i)
-~----------AQUIFER
1>
FILL
------------
ot.--
01
o I
01
TREATED PLUME
\: :~-;~~~;;E~~---
/7#/////7//')7;.....,~--r~~?...0"7~"ry/;,..,./:""7":%:rI~//77/777//
/////////h
AQUITARD
I?'////.
(c)
AQUIFER
Figure 14.12 Schematic illustration of a funnel-and-gate system: a) plan view; b) crosssectional view, keyed case; and c) cross-sectional view, hanging case.
Sec. 14.6
493
Treatment curtains and funnel-and-gate systems have the same goal: to treat the
plume in situ in a manner that requires little or no water extraction, energy use, or
maintenance. The most passive in-situ treatment system involves no external energy or
chemical input, no extraction or forced fluid circulation, and no maintenance or replacement of reactive materials. Many different types of reactive materials can be placed in a
curtain/gate, depending on the chemical composition of the plume. Research and development is most advanced for chlorinated solvent compounds. Gillham and O'Hannesin
(1993, 1994) describe laboratory and field prototype trials of a solid-phase treatment
material (zero-valent, elemental iron filings) for the in-situ destruction of chlorinated
solvents. Iron offers good prospects to achieve passive treatment of solvent plumes over
long periods of time.
Approaches for in-situ plume control that are less passive than elemental iron are
emerging, particularly for use in gates. For example, an alternative to the use of solid
particles in a gate is the slow release and mixing of dissolved chemicals in a gate.
These chemicals react with the plume constituents causing destruction of contaminants
during passage through the gate, or in a reaction zone that develops in the aquifer
downgradient of the gate. Thus, the gate or curtain can function as both the source of
the treatment chemicals and the treatment zone, or simply as the chemical source that
induces treatment in an aquifer zone downgradient of the gate.
Microbiological systems are being investigated for use with curtains (Devlin and
Barker, 1994) and gates (Wilson and Mackay, 1995). The research is investigating
two paths: 1) the release of nutrients to the aquifer to enhance natural biodegradation of chlorinated solvents by aquifer bacteria; or 2) the emplacement of cultured or
genetically-engineered bacteria in the gate. All in-situ microbial approaches require
the provision of nutrients (and in some cases other chemicals) over long periods (either
continuously or pulsed) to sustain the necessary biodegradation processes. The technical
challenge for in-situ microbial approaches is to develop release systems for nutrients and
other microbial life support chemicals that can function in a relatively passive mode for
long-times with minimal maintenance and monitoring.
In a funnel-and-gate system in which iron particles are used for dehalogenation
of chlorinated solvents, the particles can be mixed with sand and/or other media and
deposited in a gate. If the need arises to replace the reactive medium, it can be excavated
by auger equipment. Other designs for gates have been developed wherein the treatment
medium is contained in a slotted or screened metal enclosure (treatment cassette) that
is inserted into the gate. The cassette can be hoisted from the gate when replacement is
necessary. Removable cassettes offer convenience for cultured or genetically-engineered
microbial treatment systems for which periodic microbe regeneration will likely be necessary.
The concepts presented here pertain specifically to DNAPL sites, however, they
also are relevant to many non-NAPL and LNAPL sites. They provide a spectrum of approaches for plume containment ranging from an active option like pump-and-treat, to the
most passive, which include the reaction curtain and the funnel-and-gate (Figure 14.13).
In-situ treatment curtains and funnel-and-gate systems are advancing rapidly from pilot scale trials to full-scale systems. The contaminant mass flux that these technologies
494
Chap. 14
WELL
REMEDIATED PLUME
ROW OF RELEASE
WELLS OR GATES
REACTION CURTAIN
REMEDIATED PLUME
- -1----...)
---
FUNNEL &
REMEDIATED PLUME
---1--)
---- ----- ----
Figure 14.13 A spectrum of options for containment of plumes, from no action in the
top case to a funnel with multiple gates in the lower case.
must remove from plumes is not large, and therefore the remaining technical challenges
are likely surmountable. However, even when these new technologies are fully developed, the hydrogeologic conditions at many sites, particularly those with deep plumes
or in fractured rock, will render them unsuitable, leaving only pump-and-treat as the
technically- and economically-viable option for source zone or plume control.
14.7 SOURCE ZONE RESTORATION
At many DNAPL sites, remediation efforts for the purpose of reducing or preventing
groundwater contamination have been directed at the vadose zone. As has been discussed in Section 14.2, vadose-zone remediation in the source zone commonly receives
Sec. 14.7
495
priority over deeper remediation in the groundwater zone, even though vadose-zone contamination usually causes much less degradation of groundwater quality than does the
deeper DNAPLt.
The common selection of the vadose zone as the remediation priority at DNAPL
sites derives, at least initially, from the common belief that a restored vadose zone will
appreciably diminish groundwater contamination, and, of course, from the belief that
successful restoration of the vadose zone is feasible. Although vadose zone restoration
will ultimately be required to achieve complete aquifer restoration, by itself the benefit
will generally be minimal or undetectable for as long as significant DNAPL mass persists
below the water table. Furthermore, where the primary contaminants are volatile organic
chemicals such as chlorinated solvents, transfer of these contaminants from the groundwater zone upwards to the restored vadose zone offers the potential at many sites to cause
parts of the vadose zone to become re-contaminated (Feenstra and Conant, 1994). Thus,
if the groundwater zone remains contaminated, much of the benefit from vadose-zone
restoration may be temporary.
In 1986, the U.S.EPA issued regulations that imposed severe restrictions on the
excavation of contaminated soil if the soil was to be deposited off-site. These restrictions,
and the high cost of available on-site treatment of excavated soil, prompted rapid development of new in-situ technologies for the vadose zone. At solvent DNAPL sites, the
emerging technology of choice during the late 1980s and early 1990s was soil vacuum
extraction (SVE). Unfortunately, there have been few examples ofrestoration to normal
cleanup standards. SVE involves the application of a vacuum to slotted pipes in the
vadose zone to draw air through the contaminated earth material. The air flow volatilizes
contaminants from the DNAPL, soil, and/or aqueous phases. Air injection wells can be
used in combination with air extraction wells to enhance the air flow. At some sites, the
depth at which SVE can be applied can be increased by lowering the water table with
groundwater pumping wells.
The ultimate goal of most SVE operations at solvent DNAPL sites is to remove
contaminant mass to the point where water infiltrating through the vadose zone will no
longer generate contaminant concentrations above drinking water standards. SVE has best
potential to achieve these stringent standards in relatively homogeneous sand or gravel
where rapid air flow can circulate throughout the contaminant zone. However, even in
such an ideal vadose zone, SVE is not effective in the capillary fringe. Fine-grained
layers or lenses in heterogeneous deposits also limit the effectiveness of SVE because
both high water contents and low intrinsic soil permeabilities will limit air flow. Slow
diffusion of contaminants out of these zones will limit the restoration rate and can cause
the time required to achieve full restoration of the vadose zone to stretch to decades or
longer.
lNear-surfuce contamination in the vadose zone can cause unacceptable risk to land users due to the
escape of vapor at the surface, diffusion into buildings, soil erosion, or by direct contact. At such sites, the
surficial soil (uppermost 1-3 m) is often removed and replaced by clean soil, or an impervious cover is placed
on the surface to reduce risk. Such efforts are unrelated to groundwater remediation, and are not considered in
this chapter.
496
Chap. 14
SVE commonly removes considerable contaminant mass rapidly in the early stage
of operations. This is followed by rapidly diminishing mass-removal rates that approach
asymptotic or nearly asymptotic mass-removal rates. When the asymptotic condition is
reached, much contaminant mass usually remains in the vadose zone, and the continued
use of SVE provides minimal benefit. At DNAPL sites, the history of SVE for vadosezone restoration is similar to that of groundwater pump-and-treat for the restoration of
source zones. Indeed, optimism was high for both of these approaches in the 1980s,
but by the 1990s, applications usually revealed that performance was much below initial
expectations.
Heat and steam are now being investigated as means to achieve enhancements of
SVE. Other adaptations are also being tried for low permeability deposits. These include
borehole injection technologies that create artificial fractures for the enhancement of
air flow. Although such enhancements offer potential for improvement in mass-removal
efficiency, the restrictions caused by heterogeneities of geology and DNAPL distribution
will continue to impose severe limitations at most sites.

14.7.2.1 General
As noted above, at the time of this writing, aggressive pump-and-treat has been underway
at many DNAPL sites for ten years and at a few sites for nearly twenty years. Unfortunately, it has proven to be largely ineffective source-zone restoration. An examination
of concentration vs. time records for wells at long-duration pump-and-treat DNAPL sites
indicates that progress towards aquifer restoration is slow and that the prospects for
achieving drinking water quality within decades (or even much longer) are minimal. The
only other approach that has been used at many DNAPL sites is excavation, but it rarely
extends deep enough to remove all of the known or suspected DNAPL.

Large pools of DNAPL have been located at a few chlorinated solvent DNAPL sites.
Some pools are large enough to allow pumping of free-product DNAPL from wells
screened at the bottom of the pool. Many thousands of gallons of DNAPL have been
pumped from some exceptionally large pools situated at the bottom of granular aquifers.
Free-product pumping can be enhanced by means of dual pumping wells, one in the
DNAPL pool, and one in the groundwater above the pool. The flow of water across
the pool enhances DNAPL flow towards the DNAPL pumping wells. Under the most
favorable circumstances, free-product pumping removes a maximum of one-half or twothirds of the DNAPL, leaving abundant DNAPL to serve as a long-term source.
Removal of free-product from a pool causes part of the pool to desaturate with
respect to DNAPL, leaving residual DNAPL where the former pool existed. This causes
an increase in the water permeability of this zone, and consequently larger groundwater
flux through the zone. This carries the potential to worsen water quality in the plume
(Figure 14.14). Ultimately, this enhanced contaminant flux from the former pool zone
will decline and the DNAPL in this zone will disappear sooner than would be the case
Sec. 14.7
497
-:---;------ ...
--.
FLOW
PLUME
.
---
__ .... ~
/
/
/
<,
MORE EFFECTIVE DISSOLUTION

DUE TO FREE-PRODUCT
DNAPl REMOVAL
""
""
""
NO FREE-PRODUCT
REMOVAL
""-,
<,
<,
<,
<,
- - - - - - - - - - - - - - - - - - - MCl
TIME ( decades)
Mel = maxilTlJm concentration levelfor drinking waler use.
Figure 14.14 Schematic illustration of the effect of free-phase DNAPL removal on the
downgradient plume concentration. After free-phase DNAPL is removed from pools, dissolution of the remaining DNAPL (residual) produces a temporarily greater contaminant
flux.
if it had not been pumped, Thus, in this case, to achieve better groundwater quality in
the future, it is necessary to have more severe contamination for an initial period; the
length of that period is difficult to predict.
The decision to pump DNAPL from a large pool discovered during the normal
course of a site investigation is easily made since such pumping will minimize the
potential for re-occurrence of DNAPL flow. Indeed, DNAPL movement might otherwise
be initiated by changing groundwater conditions around the pool as could be caused
by pump-and-treat or other influences on the groundwater flow system. The decision
to search aggressively for smaller pools so that they can be pumped is much more
difficult because: 1) search costs can be large; 2) the prospects are minimal for significant
mass removal from small, poorly delineated pools; and 3) there is considerable risk of
remobilizing DNAPL by drilling.
14.7.2.3 In-situ mass removal
The failure of pump-and-treat and the impracticality of excavation for the restoration of
DNAPL source zones provides motivation for developing new in-situ technologies for
498
Chap. 14
this task. We will consider these experimental technologies in two categories: 1) mass
removal; and 2) mass destruction (Figure 14.15). Mass-removal technologies include
circulating steam or water containing chemical additives through the DNAPL zone. Flow
is from injection wells to withdrawal wells where fluid is pumped into water treatment
facilities at the surface (Figure 14.16). These technologies are a type of enhanced pumpand-treat. Surfactants and co-solvents such as alcohols are currently receiving most
of the research attention. They are used to greatly increase the effective solubility of
the DNAPL components so that circulation of the water/chemical mixture through the
DNAPL zone will remove contaminant mass much more rapidly than conventional pumpand-treat. To avoid prohibitive cost, the chemical additive (surfactant or co-solvent) must
usually be recycled. The contaminants must therefore be removed from the effluent at the
surface. The efficient removal of contaminants from the effluent can be difficult. And,
once removed, the contaminants must be disposed of, or destroyed.
(a )
MASS REMOVAL
STEAM FLUSH
ON-SURFACE EFFLUENT TREATMENT AND RECYCLING
(contaminant destructionor disposal)
(b)
Example
<,
<,
<,
Example
Example
VITAMIN B-12
& REDOX DEPRESSION
KMn04
2-
S04 and
BENZONATE STIMULATION
I
<,
,/
,/
I
<,
,/
,/
<,
<,
OPTIONAL
RECYCLING
OF CHEMICAL
ADDITIVES
,/
,/
Figure 14.15 Categories of experimental technologies for the restoration of a DNAPL

source zones with examples: a) mass-removal technologies; and b) in-situ mass destruction technologies.
Sec. 14.7
499
0-
0_
0-
00-
INJECTION
WITHDRAWAL
DISSOLVED PHASE
Figure 14.16 Schematic illustration of a

surfactant or alcohol flush system for the
removal of DNAPL: a) plan view; and (b)
cross-sectional view.
Steam injection combined with vacuum extraction is another way to remove

DNAPL from the vadose and groundwater zones. When saturated steam is injected
into an aquifer, chlorinated solvents with boiling points lower than that of water are
vaporized efficiently. Vacuum extraction wells are used to remove the vapor. As steam
sweeps through the aquifer, condensate at the steam front mobilizes and pushes freephase DNAPL towards the extraction wells (Stewart and Udell, 1992). However, the
mobilized zone of DNAPL at the steam front may cause DNAPL to escape downwards
into deeper zones where it will not be captured by the extraction wells. Other approaches
described by Grubb and Sitar (994) exist that use heat, low-quality steam, or hot water
for the removal of DNAPL. Whether these other approaches offer distinct advantages
over steam flushing with vacuum extraction remains to be determined.
The use of surfactants, co-solvents, and steam for the restoration of DNAPL source
zones is associated with three types of technical difficulties. As with other methods,
the first is caused by heterogeneities in the permeability and DNAPL distributions. For
the DNAPL zone to be restored to the point where groundwater flushed through the zone
meets or approaches drinking water standards, nearly all of the DNAPL mass must be
removed. Delineation of all or even most of the DNAPL in the source zone is generally
not feasible at solvent sites. Therefore, the surfactant solution, the alcohol solution, or
the steam must be flushed through the entire zone where DNAPL is expected. After
some initial flushing, detailed monitoring can provide more specific delineation of the
DNAPL distribution so that flushing can be focused on these locations. For the flushing
of surfactants, co-solvents, or steam to remove DNAPL efficiently, the flushing fluid
must come into direct contact with the DNAPL to cause mobilization and then to carry
the contaminant mass to the extraction wells. A main impediment to the full restoration
of a DNAPL zone is therefore inadequate contact of the flushing fluid with DNAPL in
low permeability zones.
500
Chap. 14
DNAPL accumulations lying in depressions at the bottom of an aquifer or in

depressions on low-permeability lenses within an aquifer are particularly difficult to
remove. Fractured media and fractured porous media present even more severe limitations
for flush technologies because while the flushing fluid will flow primarily through the
larger connected fractures, considerable DNAPL mass can exist in small and dead-end
fractures. In fractured porous media (e.g., fractured clay or sandstone), much of the
DNAPL mass present initially in the fractures is, by the time remediation is started,
likely to exist in dissolved or sorbed form in the low-permeability matrix where flushing
is ineffective. In such cases, a chemical flush removes the contaminant mass from the
zones of active flushing, causing chemical concentration gradients from the stagnant
zones. Diffusion from these zones to the flush zones results in continual but slow mass
removal.
The second type of technical difficulty with the use of surfactants or co-solvents
is the potential for adverse impacts on groundwater quality from the enhanced mobility
of the DNAPL phase. For example, surfactants and co-solvents generally cause a decrease
in the interfacial tensions of solvent DNAPLs, and this may mobilize DNAPL in residual
or accumulation zones. Mobilization may cause the DNAPL to move deeper into formerly
uncontaminated zones, thereby exacerbating the problem. Whether DNAPL mobilization
could occur to an unacceptable degree will be site specific. Research on surfactant and cosolvent flushing for DNAPL site restoration is still at an early stage; reliable predictions
of the effects of DNAPL mobilization are not yet possible.
The third difficulty with use of surfactants or co-solvents derives from the necessity to capture the groundwater that has been contaminated with the injected chemicals.
The injection wells will cause distortion of the groundwater flow pattern, and it is necessary that the extraction wells provide complete capture of the resulting, new zone
of contamination. To minimize the potential risk of DNAPL mobilization, the developing technologies of surfactant or co-solvent flushing should be used first at sites where
deeper penetration of DNAPL would be acceptable, such as at sites where the deeper
groundwater is too saline for drinking, or where deeper aquifers simply do not exist.
Air sparging is another method for restoring solvent DNAPL sources. In this
method, air is injected into the aquifer below or within the DNAPL zone. As the air
travels through the DNAPL zone, contaminant mass moves from the DNAPL phase into
the surrounding water, then volatilizes into the air. The air then travels upwards through
the groundwater zone to the vadose zone where it is generally collected in vacuum
extraction wells. In the mid- to late-1980s, when air sparging was first applied in aquifers,
it was thought that the air travelled as bubbles dispersed through the contaminated zone
and that the bubbles cause efficient mass transfer from the DNAPL phase to the air.
However, recent laboratory and mathematical-modeling investigations show that the air
travels in air stream tubes or filaments (Ahlfeld et al., 1993; Goodman et al., 1993; Unger
et al., 1995; see also Figure l.b of Pankow et al., 1993). No well-documented case of
a successful restoration of a DNAPL source zone by this technology has been reported
in the literature. Studies are in progress to determine the effectiveness of air sparging
in heterogeneous porous media. Current indications are that air sparging shows more
promise for the remediation of petroleum product LNAPL source zones. In this case,
Sec. 14.7
501
the provision of air to the subsurface brings oxygen that permits aerobic biodegradation
of petroleum hydrocarbons. Unfortunately, oxygenation of solvent DNAPL zones is not
likely to stimulate significant aerobic biodegradation unless other organic compounds
(e.g., BTEX, phenol) are provided to support co-metabolic degradation.
14.7.2.4 Mass-destruction technologies

For in-situ mass destruction, chemicals are mixed with water that is injected into the
DNAPL zone. The hydraulic aspects of these technologies are similar in general terms to
the chemical flush technologies described in the prior section. The chemical/water mixture
is flushed through the DNAPL zone towards extraction wells that capture the treatment
solution. In most circumstances, the destruction is usually not complete on the first pass
through the DNAPL zone. We also note that the effluent from the extraction wells will
contain altered chemical additive concentrations. A surface facility is therefore generally
needed to adjust the chemistry of the flush solution prior to reinjection/recirculation.
Several mass-destruction technologies are currently in the experimental stage. We
have selected two promising methods from a broad range of new possibilities for discussion here. The first involves chemical oxidation; the second involves chemical reduction.
The use of permanganate (MnOi) to scavenge and oxidize organic contaminants has
a long history in water and wastewater treatment. Adapting this history to DNAPL remediation, Gonullu et al. (1994) and Truax et al. (1995) have shown KMn04 solutions
to be effective in destroying two solvent DNAPLs, trichloroethylene (TCE) and tetrachloroethylene (PCE). This oxidant cleaves the double bond in these compounds, forming
CO 2, Ch, cr, and Mn02(s), none of which pose a problem in groundwater at the levels
typically involved. These investigators conducted laboratory batch and column tests, as
well as two small field experiments. In the latter, water containing dissolved KMn04
near the solubility limit of 30 gIL was flushed through TCE and PCE DNAPL zones 1.0
and 8.0 L in volume, located in a moderately heterogeneous sand aquifer (the Borden
aquifer). After flushing six pore volumes through the DNAPL zones, the effluents from
the extraction wells were close to or below the drinking water limit for TCE and PCE
(5 /LglL). The KMn04 remaining in the aquifer was removed by passing clean water
though the flush zone after the destruction of DNAPL was complete. Pilot-scale tests
at actual PCErrCE DNAPL sites have been designed and are expected to proceed in
1995. KMn04 has good potential for use in the in-situ destruction of other chlorinated
alkenes, but not for chlorinated alkanes like 1,1,l-trichloroethane (1,1,1-TCA). Prospects
for other chemical oxidants to destroy alkanes and other DNAPLs are being assessed.
Lesage and Brown (1993) used laboratory batch and column experiments to demonstrate the effectiveness of enzymes in destroying chlorinated alkane and alkene solvents
by reductive dechlorination. The dechlorination process is catalyzed by vitamin B12, but
the process needs strongly reducing conditions to be effective. As in the other chemical technologies described above, it is envisioned that DNAPL destruction might be
accomplished by flushing a source zone with water that contains dissolved vitamin B12
plus chemicals for the control of the redox chemistry and the pH. This reductive dehalogenation technology offers more versatility than KMn04 oxidation in the restoration
502
Chap. 14
of solvent DNAPL sites because it has potential for the destruction of a larger variety
of DNAPL chemicals. However, its field implementation is complicated by the need to
control the subsurface redox and pH conditions. The cost of chemicals per unit mass
DNAPL destroyed will also probably be higher than with KMn04 oxidation. Research
on the vitamin B12 process has progressed to the field prototype trial stage.
Of the various approaches being explored for the restoration of aquifers contaminated by chlorinated solvents, in-situ microbiological processes have been investigated
the most. One category of microbiological approach involves flushing chemical additives through a DNAPL zone to stimulate reductive (Beeman et al., 1993) or oxidative
degradative processes carried out by bacteria inherent in the aquifer (intrinsic bacteria). In
a second category, cultured or genetically-engineered bacteria are injected into a DNAPL
zone along with bacterial life-support chemicals. However, the high aqueous contaminant concentrations in and near a DNAPL zone are toxic to bacteria, and this severely
reduces the likelihood that in-situ microbiological methods will become effective for the
restoration of DNAPL source zones. Microbiological methods offer better possibilities
for assisting in source-zone restoration when applied at a later stage, after other methods
have lowered the concentrations below microbial toxicity limits.
As has been discussed above, the task of removing or destroying sufficient contaminant
mass to achieve the restoration of a DNAPL source zone is often a formidable one. The
petroleum industry has spent billions of dollars on research and field trials to enhance the
recovery of petroleum LNAPLs from oil fields. The petroleum industry considers that it
has achieved exceptional success in favorable geologic situations when the efficiency of
oil recovery increases from the 5 to 15% common to conventional oil extraction to the 30
to 40% range that can sometimes be achieved using enhanced oil recovery technologies.
(The petroleum industry has tried many technologies in these pioneering efforts to extract
LNAPLs from porous and fractured media, including the use of polymers, surfactants,
and steam.) In contrast, when remediating a DNAPL source zone, we must normally aim
for >99.9% contaminant removal if we are to approach the restoration levels needed
to allow the use of an aquifer as a drinking-water supply.
The limitations on DNAPL mass removal and in-situ destruction imposed by the
combined effects of heterogeneities in the geology, heterogeneities in the DNAPL distribution, and occasionally complex DNAPL mixtures mean that only very limited
prospects exist for the complete restoration of DNAPL source zones. This judgement
remains reasonable even if we assume that research and development efforts in this field
will intensify in this decade. Conceptually, technologies that destroy DNAPL in-situ
should be preferred to those that remove DNAPL for above-ground treatment and disposal. In-situ technologies in which DNAPL destroying-chemical additives are capable
of seeking out DNAPL without having to be advected to the DNAPL have inherent
advantages. Oxidation of TCE and PCE by KMn04 is an example. KMn04 in solution
oxidizes TCE and PCE in solution. Thus, when the KMn04 solution arrives in the vicinity
of DNAPL TCE or PCE, this increases the concentration gradient of the KMn04 towards
Sec. 14.8
Concl usions
503
the DNAPL and increases the concentration gradient of dissolved solvents towards the
KMn04 solution. Therefore, chemicals causing in-situ destruction of DNAPL may need
only periodic or occasional flushing to replenish the supply of the reactive chemicals.
These technologies offer good possibilities for being relatively passive if the time period
allowed to achieve restoration is many years. In contrast, mass-removal methods using
surfactant or alcohol flushing depend on advective flushing for mass removal, and require continual or near-continual flushing and surface treatment of effluent for long-term
restoration.
14.8 CONCLUSIONS
A DNAPL site has two principle components, a subsurface source zone and a groundwater
plume. Most of the contaminant mass at a DNAPL site is expected to be in the source
zone, but the plume usually occupies a much larger portion of the aquifer. To achieve
complete restoration of an aquifer to drinking water standards, essentially all of the
contaminant mass must be removed from both the source zone and the plume. If the
plume is restored but the source is not, the plume will reestablish itself unless contaminant
flux from the source zone is halted. Pump-and-treat and low-permeability enclosures are
proven technologies for the containment of source zones. Pump-and-treat can also be used
to capture or remove plumes. Emerging technologies, known as in-situ treatment curtains
and funnel-and-gates, are becoming available for source-zone containment and plume
control. Since annual contaminant mass flux in plumes is generally small, the technical
challenge for these in-situ containment technologies is probably not insurmountable,
except at sites where DNAPL is very deep or where the DNAPL is in fractured rock.
No proven technologies exist for the restoration of DNAPL source zones. The
difficulty of this technical challenge is unprecedented in the field of groundwater engineering. We illustrate this challenge by means of conceptual diagrams. Figure 14.17
describes progress paths towards DNAPL source-zone restoration expressed as cumulative contaminant mass removed vs. time. For the experimental mass-removal technologies described earlier in this chapter, this mass-removal relation typically exhibits an
exponential decrease in removal when the level of removal effort per unit time remains
constant. Eventually, in the ideal case (Figure l4.l7.a), the mass-removal line intersects
the complete restoration line at which time source-zone restoration is achieved. Actual
situations differ greatly from the ideal. Firstly, technology performance is less than ideal
because of heterogeneity effects. Secondly, the position of the complete restoration line
is uncertain. Therefore, as the restoration effort proceeds, the rate of approach to the
desired end point is unknown. The benefit in risk reduction derived from continuing the
remediation effort for a specified time interval cannot, consequently, be predicted or later
determined quantitatively.
As noted above, the technologies available for the removal of DNAPL from the
groundwater zone at appreciable rates are still experimental, and no DNAPL source zone
of significant size has been fully restored using any of them. Thus, controversies exist
regarding the reasons or circumstances that would justify use of these experimental technologies. A technical argument for using an experimental technology derives from the
504
Chap. 14
(a)
/
REMOVAL NEEDED TO MEET

CLEANUP STANDARDS
'///////////////////////~lj".J ' / / / / / /
,,-
Ww
-0
~~
:::la:
::i:CI)
./
,,"
I CLEANUP
ACHIEVED
//..<....... DESIRED PERFORMANCE
:::lCl)
()c(
::i:
/
/
TIME
( b)
o
Ww
-0
~::i:
::Sw
::>a:
-~
~.E~~E
. .UNCERTAIINTY IN SOURCE ZONE MASS.
::i:CI)
:::lCl)
()c(
::i:
/./
---------- --
/~
ACTUAL PERFORMANCE
TIME
Figure 14.17 Graphical concepts for

progress paths towards DNAPL source zone
restoration: a) the ideal case--eontaminant
mass in the source zone is known and
mass removal or destruction is fully
effective; and b) the typical field case-the
contaminant mass is not known and mass
removal technology performance is not
100% effective.
necessity for site trials as a means of advancing the knowledge and effectiveness of new
technologies. However, the scale and monitoring of such site trials must be commensurate
with the goal of performance assessment and process elucidation. The degree of complexity at most DNAPL sites usually necessitates intense monitoring to enable rigorous
performance assessment. Insufficient monitoring and data analyses render trials wasteful,
and often confusing. When the trial is funded solely by the organization responsible for
the particular site cleanup, such high monitoring costs for performance assessment are
often deemed unwarranted because adequate performance is not guaranteed or, in some
cases, not expected. In many cases, the main benefits from intense monitoring accrue to
industry and government at large rather than the organization responsible for cleanup of
a particular site.
A philosophical argument for the application of experimental mass-removal or
mass-destruction technologies at actual sites derives from the premise that any amount
of mass removed is beneficial because it moves the site closer to restoration, and therefore
provides some inherent future benefit. A public-perception argument for the application
of experimental technologies for mass removal is founded on the premise that public
support for groundwater remediation or protection may wane unless observable and wellintentioned remedial actions involving mass reduction are pursued, regardless of the poor
Sec. 14.9
References
505
prospects for measurable or predictable benefit to groundwater quality in the next few
years or decades. For any particular DNAPL site, there is normally opposition to such
open-ended mass-removal actions from the financially responsible parties who base their
arguments on generic economic principles and/or site-specific cost-benefit analyses.
The dilemma for decision makers regarding mass removal from DNAPL source
zones centers on the relevance and value attributed to possible future benefits, such as
increased availability of potable water or environmental risk reduction. In many cases,
the benefits are not expected to be significant or even measurable until many decades
have passed, and estimates of these future benefits are unreliable. The degree to which
today's society should allocate financial resources to mass-removal efforts to achieve
undefinable benefits for future society is the essence of a debate that is tied to both
financial and ethical issues. Decisions regarding containment of source zones or plumes,
whether they involve hydraulic capture, cutoff-wall enclosures, in-situ reaction curtains
or funnel-and-gates are much less problematic because fewer technical difficulties exist
and because risk reduction or economic benefits can be achieved rapidly.
ACKNOWLEDGEMENT
14.9 REFERENCES
Ahlfeld, D., A. Dahami, E. Hill, J. Lin, and J. Wei (1993) "Laboratory study of air sparging: Air
flow visualization", Ground Water Monitoring and Remediation, 4, 115-126.
Beeman, R. E., J. E. Howell, S. H. Shoemaker, E. A. Solazar, and J. R. Buttram (1993) "A field
evaluation of in situ microbial reductive dehalogenation by biotransformation of chlorinated
solvents". In: Proceedings of the In Situ and On Site Bioremediation Conference, Battelle, San
Diego, April 1993, 19 pp.
Burris, D. and J. A. Cherry (1992) "Emerging plume management technologies: In situ treatment
zones". In: Proceedings of the Air and Waste Management Association, AnnualMeeting, Kansas
City, Missouri, June 21-26, 1992.
Devlin,J. F. and J. F. Barker (1994) "A semi-passive nutrientinjectionschemefor enhancedin-situ
bioremediation", Ground Water, 32, 374-380.
Devlin, J. F. and B. L. Parker (1995) "Optimum hydraulic conductivity to limit contaminant flux
through cutoff walls" (unpublished work, University of Waterloo).
Feenstra, S. and B. H Conant (1994) "Vaportransport: A practical source of soil contamination following completion of remedial measures". Conference on Toxic Substances and the Hydrologic
Sources, Austin, Texas, April 10-13, 1994.
506
Chap. 14
Gillham, R. W. and S. F. O'Hannesin (1993) "Metal enhanced abiotic degradation of halogenated

aliphatics: Laboratory tests and field trials". In: Proceedings of the HazMat Central Conference,
Chicago, Illinois, March 9-11.
Gillham, R. W. and S. F. O'Hannesin (1994) "Enhanced degradation of halogenated aliphatics by
zero-valent iron", Ground Water, 32, 958-967.
Gonullu, F., G. J. Farquhar, C. Truax, M. J. Schnarr, and B. Stickney (1995) "Studies on the
use of permanganate to oxidize chlorinated solvents in soil" (unpublished work, University of
Waterloo).
Grubb, D. G. and N. Sitar (1994) "Evaluation of technologies for in-situ cleanup of DNAPL
contaminated sites", U.S. Environmental Protection Agency, EPA/6001R-94/120, 173 pp.
Goodman, 1., R. E. Hinchee, R. L. Johnson, P. C. Johnson, and D. B. McWhorter (1993) "An
overview of in situ air sparging", Ground Water Monitoring and Remediation, 4, 127-135.
Lesage, S. and S. Brown (1993) "In-situ biochemical degradation of perchloroethylene present as
residual DNAPL". In: Proceedings of the I&EC Special Symposium on Emerging Technologies
for Hazardous Waste Management, ACS, Atlanta, Georgia, September 27-29.
Mutch, R. D. and R. E. Ash (1993) "Subsurface cutoff walls still valuable in site remediation role",
Harlvlat World, February, 37-44.
National Research Council (NRC) (1994) Alternatives for Ground Water Cleanup, Committee on
Ground Water Cleanup Alternatives, Commission on Geosciences, Environment and Resources,
National Academy Press, Washington, D.C., 315 pp.
Pankow, 1. E, R. L. Johnson, and 1. A Cherry (1993) "Air sparging in gate wells in cutoff walls
and trenches for control of plumes of volatile organic compounds", Ground Water, 31, 654-663.
Parker, B. L., R. W. Gillham, and 1. A. Cherry (1994) "Diffusive dissappearance of dense, immiscible phase organic liquids in fractured geologic media", Ground Water, 32, 805-820.
Starr, R. C. and J. A. Cherry (1994) "In situ remediation of contaminated ground water: The
funnel-and-gate-system", Ground Water, 32, 465-476.
Stewart, L. D. and K. S. Udell (1992) "Combined steam injection and vacuum extraction for aquifer
cleanup". In: Proceedings of the IAH Conference on Subsurface Contamination by Immiscible
Fluids, K. U. Weyer (Ed.), Calgary, Alberta, April 18-20, 1990, A A Balkema, Rotterdam,
The Netherlands.
Truax, C., C. J. Farquhar, M. J. Schnarr, and B. Stickney (1995) "Field experiments using permanganate to oxidize trichloroethylene and perchloroethylene spilled into soil", (unpublished
work, University of Waterloo).
Unger, A J. A, E. A Sudicky, and P. A Forsyth (1995) "Mechanisms controlling vacuum extraction coupled with air sparging for remediation of heterogeneous formations contaminated
by dense non-aqueous phase liquids", Water Resources Research, 31 (in press).
Wilson, R. D. and D. M. Mackay (1995) "A method for passive release of solutes from an unpumped
well", Ground Water, 33 (in press).
Appendix
Physical and Chemical

Properties of Dense
Non-Aqueous Phase Liquid
(DNAPL) Compounds
James F. Pankow and Richard L. Johnson
Department of Environmental Science and Engineering
P.O. Box 91000
Portland, Oregon 97291-1000
507
TABLE Al
Summary of Physical and Chemical Properties of DNAPL Compounds at 25Ct.
Compound
MW t
Vapor
Pressure
(p", torr)
dichlorometbane
chloroform
bromoform
trichlorofluoromethane
carbon tetrachloride
Ll-dichloroethane
1,2-dichloroethane
1,1,l-trichloroethane
1,1,2-trichloroethane
1,I,2,2-tetrachloroethane
1,I-dichloroethylene
1,2-dibromoethane (EDB)
cis-I,2-dichloroethylene
trans-I,2-dichloroethylene
trichloethylene (TCE)
tetrachloroethylene (PCE)
1,2-dichloropropane
trans-I,3-dichloropropylene
bis(chloro)methyletber
bis(2-chloroethyl)ether
bis(2-chloroisopropyl)ether
2-chloroethylvinyletber
chlorobenzene
a-dichlorobenzene
m-dichlorobenzene
84.9
119.4
163.8
208.3
282.8
137.4
153.8
99
99
133.4
133.4
167.9
97
187.9
97
97
131.5
165.8
113
110
115
143
171
106.6
112.6
147
147
415
194
64.2
17
6.21
796
109
221
82.1
124.6
24.4
6.36
603
13.8
205
315
75
18.9
52.3
34
30
1.11
0.73
34.3
11.7
1.39
2.25
Relative
Vapor
Density
Boiling
Point
(S, mg/L)
Henry's
constant
(H, atm-mvrnol)
20000
8000
4500
4000
3000
1100
825
5100
8500
1300
4400
2900
3350
4200
3500
6300
1100
200
2800
2800
22000
10200
1700
15000
500
140
119
0.00212
0.00358
0.00206
0.00115
0.000530
0.0888
0.0298
0.00543
0.0015
0.0167
0.00108
0.000459
0.0255
0.000680
0.00374
0.00916
0.00937
0.0174
0.00262
0.0013
0.00021
0.00013
0.00011
0.00025
0.00390
0.00198
0.00325
2.05
1.80
1.39
1.14
1.07
4.91
1.62
1.70
1.26
1.59
1.12
1.04
2.86
1.10
1.63
1.97
1.35
1.12
1.20
1.10
1.09
1.004
1.003
1.010
1.04
1.01
1.01
41
62
90
119
149
23.8
76.7
57.3
83.5
Calculated
Solubility
Literature
Solubility
(S, mg/L)
21867
8513.5
6716.9
4051.6
4360.0
1620.6
740.21
5301.7
7129.7
1309.6
3965.6
3061.1
3018.1
5017.5
6995.8
4389.1
1384.9
236.96
2968.0
2994.9
16572.0
1139.1
1049.3
15148.0
444.52
135.84
133.91
COe)
113.7
146.4
31.9
131.6
60
48
86.7
121.4
96.8
112
104
178
189
108
132
179
172
Absolute Kinematic
Density Viscosity Viscosity
(mUg) (g/cm ')
(cP)
(cS)
Koc
8.8
44
61
84
116
159
439
30
14
152
56
118
65
92
86
59
126
364
51
48
1.2
14
61
6.6
330
1700
1700
1.33
1.49
1.97
2.38
2.89
1.49
1.59
1.17
1.25
1.35
1.44
1.6
1.22
2.18
1.28
1.26
1.46
1.63
1.16
1.22
1.32
1.22
1.11
1.05
1.11
1.31
1.29
0.44
0.56
1.71
0.32
0.38
0.87
2.07
0.72
0.97
0.5
0.84
0.84
1.19
1.76
0.36
1.72
0.48
0.4
0.57
0.9
0.87
0.61
0.43
0.67
0.62
0.83
1.12
0.3
0.79
0.38
0.32
0.39
0.54
0.75
2.41
1.98
0.8
1.41
1.08
0.72
1.28
0.84
"Footnores:
Solubility (8, mg/L = ppm). There are many papers in the literature that discuss measurements of the solubilities of halogenated solvent compounds.
Unfortunately, not all of them agree. An effort was therefore made to select the best values for inclusion in column five of this table. The primary source used
in this effort was Horvath (1982). Mabey et al. (1982) was used to obtain 5 values for the chlorinated ether compounds.
For a given compound, of the three parameters v" (vapor pressure, torr = mm Hg), Henry's Gas Law constant H (arm-me/mol), and 5, knowledge of any
two allows calculation of the third. Therefore, in order to provide a set of p"; H, and 5 values that are consistent with one another, in addition to giving the
literature values for solubilities, 5 values were calculated from p" and H according to the equation
5(mgjL)
= pO MW j(760H)
where M W (g/mol) is the molecular weight. These calculated solubilities are given to five significant figures for those who wish to employ the p", H, and 5
values in computer modelling efforts.
Vapor Pressure (po, torr). The primary source for the vapor pressure data was Weast and Astle (1979). The p" values given therein were interpolated and
extrapolated as necessary to the desired temperature 25C. The interpolations and extrapolations were carried out in a log p" vs. liT format, where T is
temperature in degrees Kelvin. Mabey et at. (1982) was used to obtain po values for the chlorinated ether compounds.
Henry's Gas Law Constants (H). A number of different references were used to obtain H values as a function of temperature. Gossett (1987) was used for
10 compounds, namely: dichloromethane, chloroform, carbon tetrachloride, I, I ,-dichloroethane, I, I, I-trichloroethane, I, I-dichloroethylene, cis-dichloroethylene,
trans-dichloroethylene, trichlorethylene, and tetrachloroethylene. Howe et al. (1987) was used for 8 compounds, namely: 1,2-dichloroethane, 1,1,2-trichloroethane,
I, I ,2,2-tetrachlorethane, 1,2-dihromoethane (EDB), 1,2-dichloropropane, chlorobenzene, a-dichlorobenzene, and m-dichlorobenzene. Nicholson et al. (1984) was
used for bromodichloromethane. The H values for dibromochloromethane were obtained by averaging values from Howe et at. (1987) and Nicholson et at.
(1984). Munz and Roberts (1987) was used for bromoform. Hunter-Smith et at. (1983) was used for trichlorofluoromethane. Mabey et at. (1982) was used for
5 compounds, namely: trans-I,2-dichloropropylene, bis(chloromethyl)ether, bis(2-chloroethyl)ether, bis(2-chloroisopropyl)ether, and 2-chloroethyl vinyl ether.
Water/Organic Carbon Partition Coefficients (K oc) The primary source for K oc values was Mabey et at. (1982).
Relative Vapor Density (RVD) Values. As discussed by Pankow (1988), RVD values were calculated here as the ratio of the density of dry air that is saturated
(at 25C and I atm total pressure) with the compound of interest, to the density of dry air. The equation used was
RV D
LMW
+ (760-pC)29.0
760
760
29
where po is the saturated vapor pressure, MW (g/mol) is the molecular weight of the compound of interest, and 29.0 is the mean molecular weight of dry air.
This method of calculating the RVD is more meaningful than the more common approach of taking the RVD to be the relative vapor density at the boiling
point. That RVD equals the ratio of the MW of the compound to 29.0, the mean molecular weight of dry air.
r.n
\l:)
Vapor Pressure (pO, torr), Henry's Gas Law Constants (H, atm-rrr/rnol), and Solubilities (5, mglL) as a
Function of Temperaturel. 5 Values are Given as Calculated Based on po and H Values.
TABLE A2
Compound
dichloromethane
MW
5C
IODC
15C
20 DC
25C
30 DC
84.9 pO
H
173
0.000844
22898.
76.3
0.00117
10245.
23.8
0.000586
8753.5
5.72
0.000284
5520.2
1.93
0.000171
4199.8
371
0.0467
1436.3
41.5
0.0103
815.37
89.6
0.00201
5806.8
30.4
0.000797
4968.6
47.8
0.00616
1362.0
8.1
0.000657
2164.0
1.83
0.000225
1796.8
262
0.0104
3215.3
5.45
0.000266
5065.6
79.8
0.00136
7489.0
129
0.00334
4929.5
27.6
0.00296
1613.3
5.86
0.0053
241.21
218
0.00108
22549.
98.0
0.00157
9806.6
30.9
0.000817
8151.5
7.62
0.00041
5093.9
2.63
0.00023
4254.9
454
0.0553
1484.2
53.6
0.0136
797.57
115
0.00261
5739.6
40.0
0.000941
5537.2
61.7
0.00801
1352.1
11.0
0.000749
2577.8
2.54
0.000271
2070.6
326
0.0131
3176.2
6.85
0.000340
4981.1
102
0.00178
7313.7
163
0.00436
4771.5
36.2
0.00400
1565.9
7.97
0.00724
240.15
272
0.00136
22342.
124
0.00209
9321.1
39.8
0.00112
7658.9
10.0
0.000585
4685.1
3.53
0.000307
4278.6
551
0.0651
1530.2
68.6
0.0178
779.91
143
0.00336
5543.9
51.3
0.00111
6020.3
78.9
0.0103
1344.6
14.5
0.000849
2997.8
3.49
0.000325
2372.4
403
0.0165
3117.3
8.56
0.000432
4898.9
130
0.00230
7213.9
205
0.00563
4647.3
46.7
(J.00537
1504.7
10.7
0.00979
238.44
337
0.0017
22145.
156
0.00274
8944.7
50.8
0.00153
7156.0
13.1
0.000824
4357.3
4.71
0.000405
4327.4
664
0.0762
1575.4
87.0
0.0232
758.88
179
0.00429
5435.2
65.1
0.00129
6573.7
100
0.0132
1329.7
18.9
0.000959
3459.3
4.74
0.000387
2705.9
495
0.0206
3066.9
10.8
0.000544
4908.4
164
0.00294
7119.6
255
0.00721
4514.0
59.4
0.00713
1441.5
14.3
0.0131
238.14
415
0.00212
21868.
194
0.00358
8513.5
64.2
0.00206
6716.9
17.0
0.00115
4051.6
6.21
0.000530
4359.9
796
0.0888
1620.6
109
0.0298
740.21
221
0.00543
5301.7
82.1
0.0015
7129.7
124.6
0.0167
1309.6
24.4
0.00108
3965.6
6.36
0.000459
3061.1
603
0.0255
3018.1
13.8
0.000680
5017.5
205
0.00374
6995.8
315
0.00916
4389.1
75.0
0.00937
1384.9
18.9
0.0174
236.96
507
0.00262
21617.
240
0.00462
8161.3
80.7
0.00275
6324.7
21.8
0.00159
3757.8
8.12
0.000687
4398.1
948
0.103
1663.9
136
0.0381
722.36
272
0.00683
5187.6
103
0.00173
7755.6
154
0.021
1287.2
31.2
0.00121
4526.0
8.46
0.000540
3461.1
730
0.0314
2967.2
17.7
0.000843
5191.1
255
0.00472
6895.4
387
0.0116
4258.1
94.0
0.0122
1333.2
24.7
0.0228
236.34
S
chloroform
119.4 p"
H
163.8 p"
H
208.3 po
H
bromoform
S
282.8 po
H
S
trichlorofluoromethane
137.4
"
H
S
carbon tetrachloride
153.8
v"
H
I ,I-dichloroethane
S
99 po
1,2-dichloroethane
S
99 po
I, I, I-trichloroethane
S
133.4 p"
I, I,2-trichloroethane
S
133.4 po
I, I,2,2-tetrachloroethane
S
167.9 po
1, I-dichloroethylene
S
97 pO
H
1,2-dibromoethane
(EDB)
S
187.9 po
H
cis-I,2-dichloroethylene
S
97 po
trans-I,2-dichloroethylene
S
97 po
trichloroethylene
(TeE)
tetrachloroethylene
(PCE,PER)
S
131.5 po
H
S
165.8 po
H
TABLE A2 Cont.
Compound
1,2-dichloropropane
15C
MW
113
v"
H
5
trans-I ,3-dichloropropylene
bis(chloro )methylether
18.9
0.000841
3341.4
24.7
0.00113
3250.0
32.0
0.00151
3150.9
3.49
0.00209
247.40
0.348
0.00109
61.752
0.619
0.00183
65.429
4.8
0.00242
293.87
0.501
0.00127
76.302
0.87
0.00213
79.003
6.54
0.00285
339.98
0.712
0.00148
93.051
1.21
0.00246
95.138
110 p"
H
5
115 po
H
5
bis(2-chloroethyl)ether
143
v"
H
bis(2-chloroisopropyl)ether
2-chloroethylvinylether
S
171 p"
H
S
106.6 po
H
5
chlorobenzene
112.6 po
H
S
a-dichlorobenzene
147 po
H
5
m-dichlorobenzene
147 po
H
S
41.2
0.00200
3062.9
26.9
0.0013
2994.9
23
0.00021
16572.
0.787
0.00013
1139.1
0.513
0.00011
1049.3
27
0.00025
15148.
8.8
0.00334
390.36
1.00
0.00172
112.40
1.66
0.00284
113.06
52.3
0.00262
2968.0
66.0
0.00340
2886.2
11.7
0.00390
444.47
1.39
0.00198
135.78
2.25
0.00325
133.91
15.2
0.00452
498.23
1.91
0.00227
162.75
3.03
0.00372
157.54
lFootnotes:
Vapor Pressure (p", torr). The primary source for the vapor pressure data was Weast and Astle (1979). The p" values
given therein were interpolated and extrapolated as necessary to the desired temperatures of 5, 10, 15,20,25, and 30C.
The interpolations and extrapolations were carried out in a log po vs. liT format, where T is temperature in degrees
Kelvin. Mabey et al. (1982) was used to obtain p" values for the chlorinated ether compounds.
Henry's Gas Law Constants (H, atm-nr'rmol). A number of different references were used to obtain H values as a function of temperature. Gossett (1987) was used for 10 compounds, namely: dichloromethane, chloroform, carbon tetrachloride, I, I,-dichloroethane, I, I, I-trichloroethane, I, I-dichloroethylene, cis-dichloroethylene, transdichloroethylene, trichlorethylene, and tetrachloroethylene. Howe et al. (1987) was used for 8 compounds, namely: 1,2-dichloroethane, I, I ,2-trichloroethane, I, I,2,2-tetrachlorethane, 1,2-dibromoethane (EDB), 1,2-dichloropropane, chlorobenzene, a-dichlorobenzene, and m-dichlorobenzene. Nicholson et al. (1984) was used for bromodichloromethane. The
H values for dibromochloromethane were obtained by averaging values from Howe et al. (1987) and Nicholson et
al. (1984). Munz and Roberts (1987) was used for bromoform. Hunter-Smith et al. (1983) was used for trichlorofluoromethane. Mabey et al. (1982) was used for 5 compounds, namely: trans-I,2-dichloropropylene, bis(chloromethyl)ether,
bis(2-chloroethyl)ether, bis(2-chloroisopropyl)ether, and 2-chloroethyl vinyl ether.
Solubility (8, mgIL = ppm). For a given compound, of the three parameters p? (vapor pressure, torr = mm Hg),
Henry's Gas Law constant H (atm-rrr'zmol), and 5, knowledge of any two allows calculation of the third. S values were
calculated from po and H according to the equation S (mg/L) = po MW /(760H) where MW (g/mol) is the molecular
weight. These calculated solubilities are given to five significant figures for those who wish to employ the p"; H, and
S values in computer modelling efforts.
511
512
Appendix
REFERENCES
Gossett, 1. M. (1987) "The measurement of Henry's Law constants for C; and Cz chlorinated
compounds", Environ. Sci. Techno/., 21, 202-208.
Hovarth, A. L. (1982) Halogenated Hydrocarbons. Solubility-Miscibility with Water, Marcel Dekker,
New York, 889 p.
Howe, G. B., M. E. Mullins, and T. N. Rogers (1987) "Evaluation and prediction of Henry's Law
constants and aqueous solubilities for solvents and hydrocarbon fuel components. Vol. I: Technical discussion". Report prepared by: Research Triangle Institute, P.O. Box 12194, Research
Triangle Park, NC, 27709. Report prepared for: Engineering & Services Laboratory (ESL),
Air Force Engineering & Services Center, Tyndall Air Force Base, Florida, 32403. ESL report
number: ESL-TR-86-66Nol. I.
Hunter-Smith, R. J., P. W. Balls, and P. S. Liss (1983) "Henry's Law constants and the air-sea
exchange of various low molecular weight halocarbon gases", Tellus, 35B, 170-176.
Mabey, W. R., J. H. Smith, R. T. Podoll, H. L. Johnson, T. Mill, T.-W. Chou, J. Gates, and D.
Vandenberg (1982) Aquatic Fate Processes Data for Organic Priority Pollutants, EPA 440/481-014, Washington, D.C., December 1982.
Munz, A. and P. V. Roberts (1987) "Air-water phase equilibria of volatile organic solutes", J. Am.
Water Works Assoc., May, 62-70.
Nicholson, B. C., B. P. Maguire, and D. B. Bursill (1984) "Henry's Law constants for the trihalomethanes: Effects of water composition and temperature", Environ. Sci. Techno/., 18, 518523.
Pankow, J. F. (1988) "Translator's Appendix: Physical and Chemical Properties of Dense Solvent
Compounds". In: Dense Chlorinated Solvents in Porous and Fractured Media: Model Experiments, Lewis Publishers, Boca Raton, FL, 146 p.
Weast, R. C. and M. J. Astle (1979) CRC Handbook of Chemistry and Physics, section entitled
"Vapor Pressure; Organic Compounds (Pressures Less than One Atmosphere)", pp. D-203 to
D-217, CRC Press, Boca Raton, Florida.
Index
abiotic degradation,
at Borden site, 277
dehydrohalogenation, 269-272
dihaloelimination, 273
elemental iron, 268, 278
general, 78, 132, 268-278
hydrogenolysis, 273-274, 287
hydrolysis, 270-273, 276-277, 287
oxidation, 269, 476, 501-503
reaction with sulfide, 268-269,
272-274
reduction, 269
substitution, 269-273
ABS (alkyl benzene sulfonate)
detergents, 18
adhesives, DNAPL use in, 9
adsorption vs. absorption, definitions,
242-243
advection in gas phase, in physical
models, 170-173
advection-dispersion equation
in saturated zone, 250
in vadose zone, 187, 192
aerial photographs, DNAPL site
assessment, 405
aerobic biodegradation, 268, 283-286,
291-297
aerosol propellants, DNAPL use in, 9
air injection, remediation by, 476
air sparging, remediation by, 500-501
air stripping, 25, 41, 302
aircraft production, DNAPL use in, 13
alkyl benzene sulfonate (ABS)
detergents, 18
American Water Works Association,
26
anaerobic biodegradation, 78, 268,
274, 276, 280, 286-297, 302
analytical method development,
for chlorinated solvent
compounds, 3, 16-20, 30-40
annual production, chlorinated solvent
compounds, 2, 7-9, 29, 399
aquitards,
as DNAPL barriers, 72, 86, 407
fractures in, 72-75, 86, 337, 340
movement of DNAPL through, 72,
86
penetration by diffusion, 77
penetration by DNAPL, 99-101
puncturing of by boreholes, 72
stratigraphic windows in, 72, 75,
86,406
Aroclor PCBs,
as hydraulic and dielectric fluids,
404
density, 400, 402, 406
inference of presence of, 442
mobility as DNAPLs, 400
solubility of, 400
viscosity of, 400, 407
Aroclors and other PCBs, 5, 10, 12,
56, 60, 61, 238, 400, 402-404,
406-407,442
Askarel Inerteen PCBs, 400
assessment of DNAPL sites, 381,
395-469
atmospheric pumping, of vadose zone,
191
auger mix cutoff walls, 486
automobile production, DNAPL use
in, 13
bacterial counts, as early measure of
water contamination, 17
barometric pumping, of vadose zone,
191
bedrock,
fractured, 337
pools in depressions in, 338
biodegradation,
aerobic, 268, 283-286, 291-297,
301
anaerobic, 78, 268, 274, 276, 280,
286-297
at Borden site, 274
at Moffett Field Naval Air Station,
300
basic pathways, 283
bioremediation, 300-303
co-metabolically by aerobic
aromatic degraders, 30 I
co-metabolically by aerobic
methane oxidizers, 268,
284-285, 302
co-metabolically by copiotrophs,
283
facilitation of by retardation, 283

fermentatively, 286-288, 298
field studies, 299-300
general, 9-10, 15, 268-269, 278-303
genetically-engineered organisms,
301
inhibition by excess carbon, 279
inhibition by high concentrations,
268, 279, 283, 502
inhibition by toxic metabolites, 279
laboratory studies, 290-299
loss term in models, 132
plasmids in, 284
plume front rates, 283
primary co-metabolism, 281-282
pump and treat, 30 I
sorption effects, 234
suicide substrates, 284
temperature effects, 291
threshhold levels for, 282
treatment curtains for, 493
biological oxygen demand (BOD),
as early measure of water
contamination, 17
bioremediation, general, 300-303
BOD (biological oxygen demand),
as early measure of water
contamination, 17
boiling points, of chlorinated solvent
compounds, T508-T509
Borden site,
abiotic degradation at, 277
biodegradation at, 274
characteristics of, 166,241-242,
246, 249, 253, 277
field experiments with chlorinated
solvents at, 166-171, 174,
195-197,235-239,246,250,
405,407-410,427,430-431,
449-452
predicting depth of penetration in
vadose zone at, 167-168
borehole dielectric permittivity, in
DNAPL site assessment, 427
boreholes, puncturing of aquitards by,
72
breakthrough curves,
at wells, 236, 250
513
514
breakthrough curves, cont.
ideal and non-ideal, 239, 250-253,
259
Brooks-Corey,
Equation, 98, 104, 107, 155, 157,
190, 342
fractured media, 343-344
porous media, 117
Buckley- Leverett,
Equation, 90
solution for unsteady flow, 114-116
calibrating, numerical models, 137
California, 24-25, 27, 38, 41
capillary forces,
entry into fractured clay and rock,
338
formation of residual DNAPL in
vadose zone, 180
vs. viscous forces at certain fluid
velocities, 163
capillary fringe, collapse of due to
arrival of DNAPL, 151, 153
capillary number, definition, 112
capillary phenomena, and LNAPLs,
157
capillary pressure,
across air/water interface, 149
across oil/water interface, 149
and DNAPL entry into porous and
fractured media, 58, 156,
338-343
and DNAPL flow during field
spills, 167
and multi-phase flow, 90, 110-111
and remobilization of DNAPL
fingers (ganglia), 112
and unsteady multi-phase flow,
114,133-135
definition, 93
in a fracture due to DNAPL height
in fracture, 340, 347, 349
capillary pressure curve,
determination by solution of
differential equations, 99
general, 93-99, 342
in fractured limestone, 343-344
measurement of, 94-95, 155
capillary zone,
definition, 147
formation of pooled DNAPL on,
153, 180
movement across to the
Index
groundwater zone, 63,
172-173, 180-181, 193-195,
198
carbon dioxide, leading to advective
effects in the vadose zone, 191
carcinogens, 10
CERCLA (Superfund), 7, 12, 24, 28,
38,41,84,207,208
characterization of DNAPL sites, 90,
166
chemical manufacturing, DNAPL use
in, 9-12
chemical properties, of chlorinated
solvent compounds, 9-10, Tl5,
T508-T511
chemical safety data sheets, 21, 23,
25, 30
chemical transformations,
abiotic, 78, 132, 268-278, 287, 476,
501-503
biotic, 78, 132, 234, 268-303, 493,
502
dehydrohalogenation, 269-272
dihaloelimination, 273
general, 78, 132, 268-278
hydrogenolysis, 273-274, 287
hydrolysis, 270-273, 276-277, 287
oxidation, 269
reaction with sulfide, 268-269,
272-274
reduction, 269
substitution reactions, 269-273
using elemental iron, 268, 278, 493
chlorinated solvent compounds,
annual production, 2, 7-9, 29, 399
common and commercial names,S,
403-404
history in groundwater, 2-52
mobility as DNAPLs, 400
chlorination of drinking water, 19-20,
33, 35, 37, 39, 41
chlorofluorocarbons, 23
clay liners,
for impoundments, 338, 342-343
integrity in presence of DNAPLs,
314,322
clays, effects of organic compounds
on, 314-332
Clean Water Act,
Amendments, 37
development of analytical methods
for, 38
passage of, 19, 33

coal tar,
as a DNAPL, 10, 56, 399
density of, 400-40 I, 407
estimation of spill mass, 80
formation as a waste by-product,
13, 399
in wells, 418
mobility as DNAPL, 400
pool thickness, 102
viscosity, 400, 407
coffee, decaffeination of using
chlorinated solvents, 13
coking, 13
cold cleaning, DNAPL use in, 9
colloids, effects on sorption, 254-256
common and commercial names,
chlorinated solvents, 5,
403-404
Community Water Supply Survey
(CWSS), 26, 41
concentrations, dissolved, interp. of
reI. to prob. of DNAPL
presence, 68, 173, 204, 207,
214-215, 222-223, 229,
378-379, 398,437-443,447,
454-459, 462-464
concentrations, vadose zone, 185,
429-430, 432
cone penetrometer, use of in site
assessmemnt, 426-428
Consent Decree, 37-39
contact angle, definition, 92
contact angle, pressure difference
across interface as a function
of,93
correlation equations,
among K Ge , K Gw , and solubility,
244-245
for dissolution mass transfer,
212-214
Council on Environmental Quality
(CEQ), 25, 36, 38, 40-41
co-metabolism,
by aerobic aromatic degraders, 30 I
by aerobic methane oxidizers, 268,
284-285, 302
by copiotrophs, 283
in air sparging zones, 50 I
primary, 281-282
co-solvents,
bench-scale testing of for use in
Index
remediation, 398
dense immiscible liquids, definition,
effects on interfacial tension, 406
6, 90
effects on sorption, 234-235,
density,
255-256
chlorinated solvents, 399,
in remediation, 476, 499-500
T508-T509
with DNAPLs in fractures, 381
coal tar, 407
creosote,
creosote, 407
as a DNAPL, 10, 13, 56, 399
dependence on temperature, 185,
density of, 400-401, 407
191
estimation of DNAPL mass, 80
dissolved plumes, 76
in wells, 418
DNAPLs at a site under assessment,
mobility as a DNAPL, 400
398, 406-407
observations of in fractures, 445
DNAPL mixtures, 401-402
pools of, 102
effects on DNAPL movement, 15,
production and use, 399
54, 56, 59, 108, 149
viscosity of, 400, 407
entry into fractures, 67
wood preserving, 404
gas phase plumes, 63, 170, 172,
180, 185, 189-191, 197,430
critical micelle concentration, and
LNAPLs containing chlorinated
sorption to solids, 256
critical wavelength, for finger
organics, 407
(ganglia) formation, 149, 154
depth of penetration,
cross-contamination, in sampling
effects of heterogeneities, 62
DNAPL-contaminated soils,
force balance for flow, 75
in fractured media, 69-70, 86, 379
433-434
cutoff walls,
in granular media, 72
in site assessments, 405-413
diffusion across, 486-488
predictions of for vadose zone,
grout, 486
plastic sheet, 486
167-168,406-409
source containment by, 380,476,
retention capacity, implications, 62,
406
483-489
spill size effects, 63, 85-86, 405
CWSS (Community Water Supply
uncertainties in prediction, 62, 159,
Survey), 26, 41
175
Darcy's Law, 105-107, 132, 187-190,
desiccation of clays, opening of
315, 359
fractures for DNAPL flow, 75,
Dayton, NJ site, 462-468
314-332
dead end,
detergents, as early water pollutants,
fractures, 69
18, 31
pores and sorption, 252
diagnosis of DNAPL sites, 90, 381,
decaffeinated coffee, chlorinated
395-469
solvent use in manufacture of,
diffusion,
13
across cutoff walls, 486-488
degradability, implications for
effects of variable water content in
chlorinated solvent
vadose zone, 430
compounds, 2, 15
effects on sorption kinetics, 253
degradation,
from fractures into porous matrix
abiotic, 78, 132, 268-278, 287, 476,
materials, 55-57, 71-72,
501-503
355-390,446-448,481
biotic, 78, 132, 234, 268-303, 493,
in the gas phase
502
across the capillary zone,
products, hydrophobicity of, 282
degreasing, DNAPL use in, 9, 13, 21,
180-181
404-405 '
formation of vapor plumes, 60,
515
63,170-173,180,187-188
in the vadose zone, 180-198,
430-432
in water and dissolution rates, 130,
204-229
into/out of low permeability zones,
77-79,476,481,500
out of fractured porous media, 70,
356, 361
study of using physical models of
the vadose zone, 170
diffusion coefficient,
in aqueous phase, 188, 362
in free air, 188
in porous matrix materials, 356
in vadose zone and dependence on
water content, 189, 197
temperature effects, 188
dimensionless numbers, for
dissolution mass transfer,
212-214
disappearance, of DNAPL from
fractured porous media, 70,
355-390
dispersion,
and dilution of saturated
concentrations, 204
and dissolution of pooled DNAPL,
218-220
in capillary zone, 193
in fractured media, 76-77, 86
in porous media, 75, 86, 187
displacement pressure,
and entry into porous media, 95,
156
and hydraulic conductivity, 102
and multi-phase flow, 110
definition, 95
disposal practices, historical, 19, 21,
32, 33, 38, 40, 43, 139, 398,
400-402, 404-405
disposal wells, puncturing of aquitards
by, 72
dissolution,
and equilibrium partitioning, 59,
204,214
and plume mass flux, 82
by groundwater, 4, 29, 54-55,
59-60, 66-67, 70, 75, 82,
86-87, 125, 130, 173-174,
204-229
516
dissolution, cont.
diffusion limitation of, 130,
204-229
effects on plume size, 60
from DNAPL in fractures to
adjacent porous matrix,
355-390
heterogeneity effects, 222
in fractured porous media, 70
into pores in aquitard, 75
mathematical models of, 215-220
of pools of DNAPL, 173-174,
208-209, 215, 217-220
of residual DNAPL, 173, 208-209
solubilities and persistence, 60
study of at Borden site, 174
time to dissolve, 29, 66, 86,
174-175,204,220-222,229,
459-460, 488
dissolved organic matter (DaM),
as used by subsurface organisms,
279
effects on sorption, 234, 254-256
distributions of DNAPLs,
as affected by heterogeneities, 54,
61,64,75,87, 108, 125, 130,
146, 147, 149, 154-158,
166-170, 455
as affected by water content, 151,
153
DNAPLs,
source zones,
coming to understand their natures,
26-29
delineation of, 84
isolation of, 85
lifetimes, 29, 66, 86, 174-175, 204,
220-222, 229, 459-460, 488
chemical types, 3-5, 10-13,398-403
dielectric constants of, 316
double layers, and clay permeability,
314,316,323
drainage of DNAPL, 95
drains, use of to recover DNAPLs in
pools, 103
drilling,
analysis of water during, 416-417,
448
observations of DNAPL during,
396, 398, 413-417
remobilization of DNAPL by, 397,
416
Index
drinking water,
Evansville, IN, 33
haloforms in, 20
New Orleans, 33
drinking water limits, 2, 10-11, 14,
T14, 45, 76, 78, 82-83, 173,
253, 491
drive point devices, monitoring and
sampling, 426-427, 453-454,
456-457
dry cleaning, DNAPL use in, 7, 9-11,
13,404
dyes, use of to assist visualization of
free product DNAPL, 396,
413-414,421,443
electrical resistance logging, use of to
characterize DNAPL spills,
170
electromagnetic (EM) resistivity, in
DNAPL site assessment,
427-429
electron acceptors, and biological
activity, 278-279
electron capture detection, on GCs,
22,34,38,40
electron donors, and biological
activity, 278-279
electronics industry, DNAPL use in,
9,11,13,405
entry pressure,
definition, 93-95
fractures, 59, 337, 339, 347, 349
granular media, 59, 153
interfacial tension effects on, 322
multi-phase flow and, 90
porous matrix material, 356
variations of due to heterogeneities,
64
enzymes, in remediation, 476,
501-502
equilibrium distribution of a chemical,
59, 186, 192,204,214,234,
249-250
Evansville, IN, drinking water, 33
excavation, for remediation, 464, 478,
479, 495, 497
extraction, analysis of soils and rock
using methanol, 435-437
facilitated transport, 234, 254-256
faults, breaches in aquitards due to, 72
Federal Water Pollution Control
Administration, 18-19
fermentative biodegradation, 286-288,

298
fiber optics, in DNAPL site
assessment, 428-429
Pick's Laws of diffusion 133, 188,
210, 359, 365, 372
field experiments, use of to investigate
chlorinated solvent behavior,
146, 166-174, 195-197,
235-239, 246, 248, 250, 252,
259, 405, 407-410, 427,
430-431,449-452
film penetration model, for
dissolution, 210-212
fingers (ganglia) of DNAPL,
connecting DNAPL in horizontal
layers, 455-466
critical wavelengths and, 149, 154
formation,
at the capillary fringe, 152
due to flow instabilities, 123-125
upon entry into a lower
permeability zone, 159
general, 15,96, 149-150, 161,207
in a Hele-Shaw cell, 124
incipient motion of, 111-112
plumes from, 455-457
within pore-scale zone of higher
permeability, 154
flame ionization detection (FID),
of vapors, 415-416
on GCs, 38, 40
fluid contents, and multi-phase flow,
90
foe (fraction organic carbon), 186,
234-235,240-241,243-244,
246-248, 258, 363-364, 383,
439, 440, 442
Ford, Gerald, 20, 35
foundry operations and castings,
DNAPL use in, 13
fractured media,
diffusion to matrix, 55-57, 71-72,
355-390,446-448,460,481
dissolution of DNAPL in fractures,
457-460
dissolved phase transport, 382-387
DNAPLs in, 4, 54-57, 61, 67-72,
74, 331, 337-352, 355-390,
406
methods for identifying DNAPLs
in, 444-448
517
Index
problems in remediating, 500

fractures,
compromising of aquitards by,
72-75,86,337,340
DNAPL entry into, 74, 338-343,
412
DNAPL flow in, 74, 343-350
free phase DNAPL,
direct visual observation of,
104-105, 154,337,351,378,
396-398,413,418-421
entry pressure achievement by, 64
locating, 3
methods for identifying presence
of, 396-469
observations of during drilling,
413-417
observations of in wells, 104-105,
337,351,378,398,418-421
pore-scale flow, 59
rarity of direct observation, 84
recovery from pools, 496-497
recovery using drains, 103
summary of source characteristics,
54
Freundlich isotherm, 241-242, 252
fuel oil, 3
funnels and gates, in plume treatment,
476, 483, 491-493
gamma logging, in DNAPL site
assessment, 170, 427
gas analyses, to determine presence of
DNAPL,396
gasoline, as early indicator of
groundwater contamination, 3
Gaussian elimination, 135
GeIMS, 22, 32-35, 38, 40
gene transfer, and biodegradation, 281
genetic engineering, and
biodegradation, 30 I, 493
geophysics, in DNAPL site
assessment, 427-429
granite (fractured), DNAPLs in, 68
granular aquifers, sorption in, 235,
246-247
ground penetrating radar (GPR), in
DNAPL site assessment, 170,
427
Ground Water Supply Survey, 45
groundwater plumes,
character of, 449-468
dissolved mass in, 81
from
contamination in the vadose
zone, 63, 65, 85, 130
fingers of DNAPL, 455-457
pools, 449, 455-457
subsurface DNAPL, 450
vapor plumes, 449-450, 481
locating, 64-65, 193-195
mapping by vadose zone sampling,
193-195
sizes of, 64-65, 76-77, 81-82, 84,
86
spatial distribution, 449-454
velocity of, 84
grout cutoff walls, 486
grouting, and short circuiting of
DNAPL around wells in
fractured rock, 425
haloforms, in drinking water, 20-23,
25,33, 35-37, 39
headspace methods, analysis of soils
and rock by, 435, 437, 447
headspace methods, analysis of water
by, 32, 38, 40
health effects, 7, 10,41, 78, 82
heat transfer fluids, DNAPL use in, 10
Hele-Shaw cells, 124
Henry's Law, 134, 185-186, 192, 197,
438, T508-T511
heterogeneities,
and DNAPL dissolution, 222
and DNAPL distributions, 54, 61,
64,75,87, 108, 125, 130, 146,
147, 149, 154-158, 166-170,
314,406-410,433
effects on breakthrough curves,
251, 253-254
site assessment considerations,
406-407
high tech companies, DNAPL use by,
7
history, chlorinated solvent
compounds in groundwater,
2-52
Hyde Park Landfill, 27
hydrologic boundaries, for plumes,
82-83, 86
hydrophobicity,
and sorption to soil, 236, 238,
243-244
degradation products, 282
hysteresis, in capillary pressure
relationships, 96-97
Ideal Gas Law, 187
immobile water, and sorption,
252-253
incompressibility assumption, in the
vadose zone, 190
industrial solvent users, Tl3
industrial wastes, 19
infiltrating water, as downward
transport mechansim in vadose
zone, 180-181, 193, 198
injection, underground, 33
insurance industry, 16, 30, 32
interface probes, detection of
DNAPLs in wells, 420
interfacial tension,
effects on capillary phenomena, 58
effects on multi-phase flow, 90
waterlDNAPL,
pressure difference across
interface, 92-93, 99
general, 2, 15, 54, 59
entry into fractures, 67, 322,
338-340, 349
effects of surfactants and
co-solvents, 59, 406, 500
typical values, 92
site assessment, 398, 406-407
pore scale effects, 91, 92, 318
intrinsic permeability, and fluid flow,
105-106
ionic strength, effects on sorption,
254-255, 259
iron, remediation with, 268, 278, 493
karstic rock,
dissolved plumes in, 84
DNAPLs in, 67-68
Klinkenberg no-slip effect, 190
Knudsen diffusion, in the gas phase,
188
K oc (organic carbon/water partition
coefficient), 186, 244-248,
362-363, 383, 439-440, 442,
T508-T509
K ow (octanol/water partition
coefficient), 234, 236,
244-245, 256, T508-T509
lakes, dissolved contaminants in, 82,
83
lateral spreading of DNAPLs,
at the water table, 153, 180
due to heterogeneities, 64, 86,
Index
518
lateral spreading of DNAPLs, cont.
116-119,136,149,154-158,
166-170,314,407-410,433,
455
liability, 16
lifetimes of DNAPL,
in fractures, 367-375, 377, 460
source zones, 29, 66, 86, 174-175,
204, 220-222, 229, 459-460,
488
light immiscible liquids, definition, 6
limestone, 447-448, 460
linear sorption, to aquifer solids,
240-242
litigation, 16
LNAPLs,
at waste disposal sites, 40 I
biodegradation of, 480
dielectric constants of, 316
excavation, 479
permeability tests with, 323-328
remediation of sites contaminated
with, 477
local equilibrium assumption, 134
Long Island, 22, 26, 34, 36, 40
Louisiana, 32
Louisiana Stream Control
Commission, 18-19,30
Love Canal, 28
machine shops, DNAPL use in, 7
Maine, 22, 36
Manufacturing Chemists Association,
19,30
mass flux, in dissolved plumes, 82
mass storage capacity,
of porous matrix material adjacent
to fractures, 356, 361,
364-365, 370, 375
of fractures, 356, 361, 364-365,
370, 375
mass transfer rates,
between the dissolved and sorbed
phases, 239, 248, 249, 252,
253
of a DNAPL compound to air
phase, 60, 184
of a DNAPL compound to water
phase, 60, 204-229
Massachusetts, 22, 38
material safety data sheets, 21, 23, 25
maximum concentration limits
(MCLs), drinking water,
10-11, 14, Tl4, 45, 76, 78,

82-83, 173, 253
metal fabrication, DNAPL use in, 9,
13, 21, 404-405
methane, production of and advective
effects in the vadose zone, 191
methanol extraction, analysis of soils
and rocks by, 435-437
Miami, FL, 35
microbial ecology, 268, 278-281
microsolvent extraction, analysis of
water by, 34, 38
military bases, contamination on, 23
minerals, sorption to inorganic
surfaces of, 234, 247-248,
257-259
miscibility, 6, 90, 318
Mississippi River, organic
contaminants in, 18-20, 30-31,
33
mixtures of DNAPLs,
composition of, 401-403
density of, 401-403
due to impurities in solvents, 463
in field experiments, 450
interfacial tension of, 406
solubility of, 67, 174, 204-210,
222-229, 379, 439, 463
viscosity of, 402
volatility of, 134, 182-185,415
mobile water, and sorption, 252-253
mobility tensor, 133
mobilization of DNAPL, by water
pumping, 340-341, 348-349,
379-380
models, for general fate predictions,
182
Moffet Field Naval Air Station,
biodegradation at, 300
molecular weights, of chlorinated
solvent compounds,
T508-T509
monitoring, by drive point devices,
426-427, 453-454, 456-457
monitoring wells,
DNAPLs in, 104-105
use of to locate groundwater
plumes, 64
Monte-Carlo methods, 130, 138
multi-phase flow,
in fractured media, 337-352
in porous media, 90-126, 129-143,
146, 150-170
Nassau County, New York, 22, 36
National Organics Monitoring Study
(NOMS), 21, 35, 37, 43
National Organics Reconnnaissance
Study (NORS), 20-22, 33, 35
National Priority List (NPL) Sites, 29
National Water Commission, 30
neutron logging, in DNAPL site
assessment, 428
neutron probe, in DNAPL site
assessment, 170
New Hampshire, 41
New Jersey, 3, 22, 25-26, 39, 43,
462-468
New Orleans, drinking water organic
contaminants in, 18-20, 30, 32,
33,35
New York State, 22, 26-27, 34, 36,
38, 40
Newton-Raphson method, 135
Niagara Falls area, 27, 40
nitrate reduction, 286
Nixon, Richard M., 19,31
NOMS (National Organics Monitoring
Study), 21, 35, 37, 43
non-linear sorption,
to aquifer solids, 234, 241-242,
248, 252
to dry soils, 186
non-wetting fluid,
definition, 92
flow of a DNAPL as, 99, 156, 359
non-polar organic liquids, 317
on geologic solids, 58
pressure in, 93
retention in fractures, 159
NORS (National Organics
Reconnaissance Study),
20-22, 33, 35
NPL (National Priority List) Sites, 29
nucleophilic attack, degradation by,
269
numerical model,
calibration, 137
definition, 130
sensitivity, analysis, 138
testing, 138
validation, 137, 146
verification, 137
nutrients, and biological activity,
278-279
519
Index
one-dimensional (I-D) columns,

use of to study DNAPL
dissolution, 217
one-percent rule, for interpreting
dissolved cone. and DNAPL
presence, 222-223, 229, 459,
462-464
organic carbon, sorption to, 186,
234-235, 240-241, 243-244,
246-248, 258, 362-364, 383,
439, 440, 442
organic vapor analyzer (OVA), to
detect DNAPLs, 415-417
Otis Air Force Base, studies at, 238,
248
outside-in approach, evaluation of
DNAPL sites, 422-423
OVA (organic vapor analyzer), to
detect DNAPLs, 415-417
oxidation,
general, 269
PAHs (polycyclic aromatic
hydrocarbons ),
in coal tar and creosote, 400
solubility of, 400
sorption of, 400
paint removal, DNAPL use in, 9, II
PCBs,
accumulation in food chain, 15
annual production, 400
as DNAPLs, 5, 10
as DNAPLs in groundwater, 56
chlorinated benzenes in, 61
density, 400, 402
DNAPL mixtures, 402-403
inference of presence in soil, 442
rate of movement in groundwater,
60
solubility, 400, 402
sorption to soil, 60, 238, 400
uses, 12,400,403
viscosity, 400
Pennsylvania, 41, 207-208
perched DNAPL,
in saturated zone, 64, 75
in vadose zone, 62
persistence of, 67
permanganate, in remediation,
501-503
permeability,
and Darcy's Law, 105-107
and DNAPL flow, 59, 62, 64,

86, 116-119, 136, 149-151,
154-158, 166-170
effects of dissolved organics on
clays, 314, 329
effects on remediation efficiency,
499-500
field studies, 330-331
in pooled DNAPL, 66, 90
in residual DNAPL, 66, 90
influence on finger formation, 149
laboratory studies, 319-330
of clays to organic liquids, 75,
314-332
of dry clays to organic liquids, 323
site assessrnent considerations,
406-407
variability of and finger formation,
154
variation of for DNAPL flow due to
water content, 151, 153, 338
persistence, of DNAPL source zones,
29, 60, 66, 82, 86
pesticides,
as DNAPLs, 5, 10-11,400
as early indicators of water
contamination, 18, 30, 37
movement in fractures, 331
prohibition of solvent use in
pesticide formulations, 45
uses, 13
pH, effects on sorption, 254-255, 259
phase transfer processes, 59-61
photographic processing, DNAPL use
in, 7
photoionization detection (PID), of
vapors, 415
physical model experiments,
advantages of, 146
general, 145-175
physical properties, chlorinated
solvent compounds, 9-10, TIS,
T508-T511
physical properties, LNAPLs, 14
PID (photoionization detection), of
vapors, 415
plasmids, and biodegradation, 284
plastic sheet, cutoff walls, 486
plume mass fluxes, 82
plumes,
in groundwater zone
character of, 449-468
dissolved mass in, 81, 481

from fingers of DNAPL, 455-457
from pools, 449, 455-457
from subsurface, DNAPL, 450
from vapor plumes, 449-450, 481
mapping by soil gas sampling,
193-195,429-432
mapping using drive point
devices, 427
sizes of, 64, 65, 76-77, 81-82,
84,86
spatial distribution, 449-454
velocity of, 84
plumes,
in vadose zone,
detection by soil gas sampling,
194-195,426-427,429-432
formation by diffusion from
DNAPL mass in unsaturated
zone, 60, 63, 180-181,449-450
field experiments at Borden site,
195-197,
polycyclic aromatic hydrocarbons
(PAHs),
in coal tar and creosote, 400
solubilities of, 400
sorption of, 400
pools of DNAPL,
at Dayton, NJ site, 466
dependence of formation on
DNAPL density, 102
dissolution of, 173-174, 215,
217-220
distribution of DNAPL fluid in,
99-105,209
estimating volumes of, 103-104
groundwater plumes from, 449,
455-457
in horizontal fractures, 67
leading to DNAPL in wells, 418
locating in the saturated zone, 65
on bedrock, 91, 338, 341, 345
on low permeability layers in the
saturated zone, 15, 65, 75, 86,
90,99-105, 154-158, 169,
208-209, 340, 409, 455
on low permeability layers in the
unsaturated zone, 15, 149, 180
on top of fracture systems, 72-73,
338, 344, 347-350, 412
on top of the capillary zone, 180
520
and Recovery Act), 7, 24, 28,
pools of DNAPL, cont.
37-39,45, 84
persistence of in the saturated zone,
Reagan, Ronald, 41
67
problems in remediation caused by, reduction,
general, 269
500
recovery of DNAPL from, 496
Regional Water Quality Control Board
vaporization of and consequent
(RWQCB),27
groundwater contamination,
197
relative permeabilities,
variation of fluid pressure in, 10I
and multi-phase flow, 106-107,
pore doublet model, for residual
133-135
measurement of, 106-107
formation, 97
relative vapor densities, of chlorinated
porous media, DNAPLs in, 4, 54-55,
64-67,71-72, 139-142
solvent compounds,
porous polyethylene, use of to detect
T508-T509
free product DNAPL, 443-444 remediation,
presence of DNAPL, inference of, 68,
abiotically,277-278
80
biotically, 277
comparisons of different strategies,
printing industry, DNAPL use in, 7,
13,462
146
priority pollutants, 23, 37, 39, 302
cost of, 72
products of degradation,
deciding need for, 82
sorption of, 79, 80
design of, 138, 146
stability of, 78
in fractured porous media, 70, 381
pump and treat,
of contaminated soil, 85
air stripping, 41
of DNAPL sites, 90, 476-505
and diffusion from low permeability
using
zones, 78
air sparging, 500-50 I
co-solvents, 398, 476, 499-500
and DNAPL mobilization, 348-349
at Dayton, NJ site, 462-468
elemental iron, 268, 278, 493
enzymes, 476, 501-502
common failure of in restoring
DNAPL source zones, 85, 226,
excavation, 464, 478-479, 495,
397, 460, 496, 497
497
funnels and gates, 476, 483,
enhanced by surfactants or
co-solvents, 498
491-493
plume capture, 64, 467
oxidation, 476, 501-503
treatment of plume zones, 85, 204,
pump and treat, 85
234, 253, 460-463, 478-479
reduction, 476, 501-502
hydraulic containment, 423, 467,
soil vapor extraction, 450,
476,482-484
495-496, 499
in fractured systems, 381
surfactants, 476, 499-500
source zone containment, 488-491
remobilization of DNAPLs,
with biodegradation, 301
as fingers (ganglia), 112
purge and trap,
by changing water pressure, 86
analysis of soils and rock by,
by co-solvents, 500
435-437, 447
by drilling, 84, 397, 416, 419,
analysis of water by, 18, 20, 22, 32,
421-425,449
34-35, 38, 40
by injected steam, 499
random surface renewal, model for
by surfactants, 500
minimizatiion of by use of drive
dissolution, 210, 212
Raoult's Law, 182-184,205,415,440
points, 426
RCRA (Resource Conservation
residual DNAPL,
Index
and creation of dissolved plumes,

98, 207-208
as long term source, 170
detection by EM methods, 428
dissolution of, 173-174,215-222
distribution of, 62, 86
effects on permeability, 66
formation of, 59, 90
in fracture systems, 347
locating, 84
values in sand, 98
values in saturated zone, 66, 169
vaporization of and groundwater
contamination, 180, 197
vaporization of with consequent
advective flow, 191
residual saturation,
definition, 96, 129, 159
for NAPLs in vadose zone, 407-408
for water in presence of a NAPL,
162
Resource Conservation and
Recovery Act (RCRA), 7, 24,
28, 37-39,45,84
retardation,
and facilitation of biodegradation,
283
due to diffusion into porous matrix
material, 56
due to sorption to soil, 60, 171,
180, 235-240, 242, 246, 253,
258
due to sorption to water in the
vadose zone, 171, 180, 257
retardation factors,
and diffusion in matrix material
adjacent to fractures, 365-370
in the saturated zone, 236, 238,
240, 242, 246, 363, 365
in the vadose zone, 186-187,
257-258
retention capacity,
and heterogeneities, 61, 75
and predicting depth of penetration,
62,408-409
definition, 158
for DNAPLs in field spills at the
Borden site, 167
for LNAPLs in unsaturated sands,
61
in vadose zone in the field, 406-408
521
Index
limitations of lab values for,

408-409
of aquitards, 75
of dry homogeneous sands in
vadose zone, 61, 159
of fractured rock, 69, 86
of geological media for a DNAPL,
60,85
of moist stratified sands in vadose
zone, 61
of sands below the water table, 159
variability of, 87
rivers, dissolved contaminants in,
82-83
rock analyses, to determine DNAPL,
396, 398, 432-448
Ruckelshaus, William, 31
Safe Drinking Water Act, 17,20,24,
33, 35, 37, 39
sampling, by drive point devices,
426-427
San Francisco, 27, 40, 42, 44
San Gabriel Valley, Calif., 24, 38
sand and gravel aquifers, sorption in,
235, 246-247
sandstone, fractured, 54, 70
Santa Clara Valley, Calif., 24, 27, 42,
71
saturated zone, definition, 147
SchwiIle, Friedrich, 4-6, 16, 25,
27-28, 45, 61, 66, 69, 98,
146-147,149,150,154-155,
157-162,173,214-215,220,
347, 363, 379, 445
seismic methods, in DNAPL site
assessment, 427
sensitivity analysis, for numerical
models, 138, 197
septic tank cleaners, 22, 26, 34, 36, 40
sewage and sewage sludge, containing
solvent chemicals, I, 30
shale, fractured, 54, 70
sheet piling cutoff walls, 486
Shelby tube, sampling in DNAPL site
assessment with, 84, 433
shock front, in unsteady flow, 115
short circuiting of DNAPL,
caused by drilling, 416, 419,
421-425,449
minimized by use of drive points,
426
minimized by use of telescoping
casing, 424-425
minimized by use of grout in
fractured rock, 425
shrinkage, of clays due to organic
compounds, 75, 314-332
Silicon Valley, 24, 27, 29, 40, 42,71
site assessment, data needs, 381,
395-469
slurry trench cutoff waIls, 486
soil analyses, to determine DNAPL,
396, 398, 432-448, 464
soil gas analyses,
for identifying DNAPL, 414-417
for plume mapping and location of
DNAPL zones, 193-195,396,
429-432
soil vapor extraction (SVE), 450, 495,
496
soil/water shake tests, determination
of DNAPL presence with, 414,
421,443
Solid Waste Disposal Act, 37
solubility,
and definition of a DNAPL, 399
and human health, 98
and miscibility, 318
as a reference to establish probable
presence of DNAPL, 68, 173,
204,207,214-215,222-223,
229, 378-379, 398, 437-443,
454-459, 462-464
effects of temperature, 205,
T5 IO-T5I I
general, 2, Tl4, 15, 185
implications for sorption, 243-244
of a sub-cooled liquid, 205-206
of chlorinated solvent compounds,
T14, T508-T511
of mixtures, 67, 174,204-210,
222-229, 439
solubility/MeL ratio, Tl4
sorption,
effects of
colloids, 254-256
co-solvents, 234-235, 255-256
dead-end pores, 252
immobile water, 252-253
ionic strength, 254-255, 259
pH, 254-255, 259
surfactants, 256
isotherms, 239-240
kinetic effects, 253
mobile water, 252-253

non-linear, 186, 234, 241-242, 248,
252
of degradation products, 79-80
to mineral surfaces, 234, 247-248,
257-259
to solid materials, 2, 15, 59-61, 86,
180, 186-187,233-259,438
source types, 7, 9, TI4
source zone, isolation, 85
spill size, estimation of, 80
split spoon sampling, 84, 433
sponge coring, for detecting DNAPLs
in fractures, 446
stagnant film model, for dissolution,
210-211
Stanford-Waterloo study, 235-239,
246, 250
static distribution of DNAPL,
governing equations, 90
steam, in remediation, 476, 499
streams, dissolved contaminants in,
82-83
substitution reactions, 269-273
Suffolk County, New York, 22, 36
suicide substrates, and biodegradation,
284
sulfate reduction, 286-288
sulfide, reactions with, 268-269,
272-273
Superfund, 7, 12,24.28.38,41.45,
84,207,208
surfactants,
bench-scale testing of, 398
effects on biodegradation, 234
effects on sorption, 234-235,
256-257
effects on water/DNAPL interfacial
tension 59, 406
in pump and treat, 498
with DNAPLs in fractures, 381
SVE (soil vapor extraction), 450,
495-496, 499
swab tests, for detecting DNAPLs in
fractures, 444-446
swelling clays, and effects of
DNAPLs on permeability, 315,
318-319,324
swelling clays, and sorption to solids,

247
522
tailing,
as caused by diffusion limitations,
130
in soil vacuum extraction, 131
Teflon, sorption losses to in laboratory
work,248
telescoping casing, use of to minimize
short-circuiting of DNAPL,
424-425
temperature effects on
degradation rates, 271, 291
diffusion coefficients, 188
gas phase density, 185, 191
Henry's Law constants, 185,
T51O-5ll
solubility, 205, T51O-T5ll
sorption, 254, 259
vapor pressures, 184-185,415,
T5I0-T5I1
test pits, for locating DNAPLs,
425-426
testing numerical models, 138
textile manufacturing, DNAPL use in,
9, II
THMs (trihalomethanes), in drinking
water, 20-23, 25, 33, 35-37, 39
three-dimensional (3D),
numerical models, 131
physical models use of to study
DNAPL dissolution, 173
threshhold level, for biodegradation,
282
threshhold saturation, for formation
of connected pores and pore
throats, 106
time domain reflectometry, use of to
characterize DNAPL spills,
169-170
tool and die making, DNAPL use in,
13
tortuosity, 133, 188-189, 362
TaSCA (Toxic Substances Control
Act), 35
transformer production, DNAPL use
in, 13
trihalomethanes (THMs), 20-23, 25,
33, 35-37, 39
Tucson, AR, 35
two-dimensional (2D),
numerical models, axisymmetric,
131, 188
numerical models, plane, 131
underground injection, of wastes, 33
Index
underground storage tanks, 38, 45

unpredictability, of DNAPL
distributions, 64
unstable flow, and finger formation,
149
unsteady multi-phase flow, 112-123
urethane foam production, DNAPL
use in, 9
U.S. Government, 17-19
vadose zone,
definition, 147
DNAPL behavior in, 61, 147-149
remediation, 494-496
analyses to determine possible
presence of DNAPL, 396, 398
analyses of soil gas for plume
mapping and location of
DNAPL, 193-195,396
validating, numerical models, 137,
146
van Genuchten media, 117
vapor analyses, to determine presence
of DNAPL, 396, 414-417
vapor plumes, 60, 63, 67, 130,
170-173,180-181,187-188,
194-197,409,449-450
vapor pressure,
and definition of a DNAPL, 399
and plume size in the vadose zone,
63,180-181
effects of temperature, 184-185
of chlorinated solvents, 60, 185,
415, T508-T5lI
vapor transport, 170-173, 180-198
vaporization, effects on DNAPL
penetration, 409
vaporization, production of
groundwater contamination,
409
verifying numerical, models, 137
vertical dispersion, effects on transport
across capillary zone, 173
vibrated beam cutoff walls, 486
Ville Mercier, Quebec, 139-142
viscosity,
and entry into fractures, 67
and site assessment, 398, 406-407
DNAPL mixtures, 402
general, 2, 15,54, 59, 149
of chlorinated solvents, 399,
T508-T509
of PCBs, 407
viscous forces, and capillary forces at

certain fluid velocities, 163
volatility,
DNAPL mixtures, 134, 182-185,
415
effects of temperature, 184-185
of chlorinated solvents, 2, 15, 59,
180
volatilization,
from water to the atmosphere, 83
to soil gas, 59
to the atmosphere, 180
losses from soil and rock samples,
435-437,464
waste disposal sites, 40 I
water, partitioning from the gas phase
to 171, 180, 186-187, 196,
257-259,430
water, pumping and DNAPL
mobilization, 340-341,
348-349, 379-380
water table, fluctuations of,
and advection in the vadose zone,
191-192
and entrapment of vadose zone
contam., 63, 193, 198
waterflooding, remediation by, 90,
119-123,337
Welge tangent, 115
wells,
DNAPLs in, 104-105,337,351,
378,398,418-421
use of to locate groundwater
plumes, 64
use of to recover DNAPLs in pools,
103
wettability,
and entry into fractured media, 338
and entry into geological media, 58,
323
and multi-phase flow, 90, 92
measurement of, 92
wetting fluid,
behavior of an oil as, 149
definition, 92
in dry clays, 317
on geologic solids, 58
pressure in, 93
Wisconsin, 277
wood preserving, DNAPL use in, 13,
403
World War II, 7

Pankow Cherry Dnapl Book 1996

Uploaded by

Copyright:

Available Formats

Pankow Cherry Dnapl Book 1996

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pankow Cherry Dnapl Book 1996

Uploaded by

Copyright:

Available Formats

Who is the book dedicated to and why?

Who is the book dedicated to and why?

What is the book about?

What is the book about?

DENSE

Pankow, James F., Cherry, John A.

Editorial/production supervision: James F. Pankow

1996 by Waterloo Press.

Printed in the United States of America

Library of Congress Catalog Card Number: 95-61690

DENSE CHLORINATED SOLVENTS IN GROUNDWATER: BACKGROUND

MECHANICS AND MATHEMATICS OF THE MOVEMENT OF DENSE

CONCEPTUAL MODELS FOR THE BEHAVIOR OF DENSE NON-AQUEOUS

3.2.2 Interfacial Tension, Wettability and Capillarity

Capillary Pressure and Fluid Contents

3.3.3 Vertical Distribution of DNAPL in a Pool

3.3.4 DNAPL in a Monitoring Well

Distribution of DNAPL in Unsteady, Horizontal Flows

NUMERICAL SIMULATION OF THE MIGRATION OF DENSE NON-AQUEOUS

EXPERIMENTAL STUDIES OF THE MOVEMENT OF CHLORINATED SOLVENT

Solvent Migration in Fractured Media

VAPOR MIGRATION IN THE VADOSE ZONE

DISSOLUTION OF DENSE NON-AQUEOUS PHASE LIQUIDS (DNAPLS) IN

7.3.3 Modeling Studies

SORPTION OF DISSOLVED CHLORINATED SOLVENTS TO AQUIFER

CHEMICAL AND MICROBIOLOGICAL TRANSFORMATIONS AND

9.3.3 General Biotransformation Processes

Biotransformation of Haloaliphatic Compounds

9.4.5 Field Studies

THE EFFECTS OF CHLORINATED SOLVENTS ON THE PERMEABILITY OF

Mechanisms Controlling the Swelling of Clays

10.2.2 Interlayer Mechanisms

10.5.2.2 The influence of permeameter type on laboratory

10.5.2.3 Hydraulic gradients used in permeability tests

Permeability tests on "dry" clays/organic mixtures

PHYSICS GOVERNING THE MIGRATION OF DENSE NON-AQUEOUS PHASE

THE EFFECTS OF MOLECULAR DIFFUSION ON DNAPL BEHAVIOR IN

DIAGNOSIS AND ASSESSMENT OF DNAPL SITES

13.5.4 Sampling and Analysis of Fractured Media

13.6.4 Persistence and Temporal Variability of Dissolved Concentrations

14 CONCEPTS FOR THE REMEDIATION OF SITES CONTAMINATED WITH

of Environmental Science and Engineering

2Applied Groundwater Research, Ltd.

Dense Chlorinated Solvents: Background and History of Problem

1.7.2 Timelines in the History

Dense Chlorinated Solvents: Background and History of Problem

Organic Compounds That Are DNAPLs.

PeE, tetrachloroethylene, etbylene tetrachloride, Nema, Tetracap, Tetropil,

TCE, ethinyl trichloride, Tri-Clene, Trielene, Trilene, Trichloran, Trichloren,

TCA, metbyl chloroform, Chlorotbene, Solvent 111, TRI-ETHANE

DCM, metbylene chloride, methylene dichloride, metbylene bichloride

subsurface was being developed independently, largely without knowledge of Schwille's

Dense Chlorinated Solvents: Background and History of Problem

1.1.2 DNAPls and lNAPls

Source Types for Solvents in Groundwater

1.2 CHLORINATED SOLVENT USAGE

1.3 SOURCE TYPES FOR SOLVENTS IN GROUNDWATER

Dense Chlorinated Solvents: Background and History of Problem

Physical, Chemical, and Biological Properties

1986 U.S. Production of Chlorinated Organic Solvents.