d "73 - =_/ '-/'-/'-/'

NASA CONTRACTOR NASA CR-2163

REPORT

,¢¢ CASE FILE

,¢¢
Z COPY

HYDROGEN ENVIRONMENT
EMBRITTLEMENT OF METALS

by R. P..lewet6 R.j. Walter, IV. T. Chandler,

and R. P. Frohmberg

Prepared by

ROCKETDYNE

DIVISION OF NORTH AMERICAN ROCKWELL

Canoga Park, Calif. 91304

/or

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION ° WASHINGTON, D. C. ° MARCH 1973

 1. Report No. 2. Government Accession No. 3. Recipient's Catalog No.
NASA CR-2163
4. Title and Subtitle 5. Report Date
March 1973
HYDROGEN ENVIRONMENT _{BRI_T OF METALS
6. Performing Organization Code

7. Author(s)
8. Performing Organization Report No.
R. P. Jewett, R. J. Walter, W. T. Chandler and R. P. Froh_rg

10. Work Unit No.

9. Performing Organization Name and Address

P_cketdyne '11, Contract or Grant No.

Division of North American Rockwell
6633 Canoga Avenue NASS-19(C)
Cauo_a Park t CA 91304 13. Type of Report and Period Covered
12. Soonsorina Agency Name and Address
Contractor Report
National Aeronautics and Space Administration 14. Sponsoring Agency Code
Washington, DC 205h6

:15. Supplementary Notes

16. Abstract

Hydrogen environment embrittlement refers to metals stressed while explosed to a hydrogen

atmosphere. Tested in air, even after exposure to hydrogen under pressure, this effect
is not observed on similar specimens. Much high-purity hydrogen is prepared by evaporation
of liquid hydrogen, sad thus has low levels for potential impurities which could otherwise
inhibit or poison the absorbent reactions that are involved. Hlgh strength steels and
nickel-base allows are rated as showing extreme embrittlement; aluminum alloys and the
austenitic stainless steels, as well as copper, have negligible susceptibility to this
phen_enon.

The cracking that occurs appears to be a surface phenomenon, unlike that of internal
hydrogen embrittlement.

17. Key Words (Suggested by Author(s)) 18. D;stribution Statement

Hydrogen embrittlement, Hydrogen environment
embrlttlement, Testing in high pressure Unclassified - Unlimited
hydrogen, Effects of hydrogen environment on
mechanical properties of metals, Crack
behaviour in hydrogen

19. Security Classif. (of this report) 20. Security Classif. (of this page) 21. No. of Pages 22. Price"
Unclassified Unclassified 243 $3 .oo

For sale by the National Technical Information Service, Springfield, Virginia 22151
 FOREWORD

This report presents, describes, and discusses the nature of

hydrogen-environment embrittlement. The embrittling effects of hydro-

gen have long been known, but it was relatively recent failures of

large, high-pressure hydrogen storage vessels at NASA and NASA con-

tractor facilities that served to focus attention specifically on

hydrogen-environment embrittlement. The ensuing extensive investiga-

tions sponsored by NASA and other government agencies have consider-

ably expanded the knowledge and understanding of this type of

embrittlement.

Thus, in accordance with_the NASA technology utilization pro-

gram goal of "the rapid dissemination of information ... on techno-

logical developments ... which appear to be useful for general indus-

trial application," this report was designed to acquaint a wide range

of interested parties with hydrogen-environment embrittlement and to

assist them in developing a course of action to deal with it.

For this report, a literature search was made and various inves-

tigators were personally contacted. The data were critically evalu-

ated, collated, and compiled for their most effective presentation.

Also included are discussions of the characteristics and proposed

mechanisms for hydrogen-environment embrittlement, the extent and

limitations of our understanding, associated problems and hazards,

and possible means of preventing, minimizing, or compensating for

hydrogen-environment embrittlement.

iii
 AC KNOWLEDGblENTS

The study of hydrogen-environment embrittlement has grown initially from

the work of a few investigators to a substantial group of investigations spon-

sored, for the most part, by NASA and other government agencies. The authors

wish to acknowledge the help of the following individuals in obtaining the

basic information:

_. Bryan _cPherson NASA-Marshall Space Flight Center

Samuel Snyder NASA-Space Nuclear Propulsion Office, Washington

Dell P. Williams NASA-Ames Research Center

James J. Lombardo NASA-Space Nuclear Propulsion Office, Cleveland

Royce G. Forman NASA-Manned Spacecraft Center

James A. Donovan AEC-Savannah River Laboratory

Harris L. Marcus North American Rockwell-Science Center

James C. Williams North American Rockwell-Science Center

Peter P. Dessau Aerojet-General Corporation

R. L. Kesterson Westinghouse Astronuclear Laboratory

Robert P. Wei Lehigh University

M. C. VanWanderham Pratt & Whitney Aircraft

Joseph N. Masters The Boeing Company

A. W. Loginow United States Steel Corporation

Herbert H. Johnson Cornell University

v
 CONTENTS

Foreword .... ........ iii

Chapter i. Introduction

Operating Experiences .

Hydrogen and Metals

Scope . . .

Chapter 2. Significance • • • • • . • • . . • . 7

Availability of Low-Cost Liquid Hydrogen .... 7

Usage Outlook ............. 8

The Overall Problem .......... 9

Chapter 3. Overview ........... Ii

Internal Hydrogen Embrittlement .......... 13

Hydrogen-Environment Embrittlement .......... 17

Comparison ............ 19

Chapter 4. Early Work ............ 23

Early Service Failures Attributed to High-Pressure Hydrogen . 23

Mechanical Tests Performed in High-Pressure Hydrogen ...... 24

Recent Pressure Vessel Failures ............ 29

Chapter 5. Test Procedures ............ 31

Hydrogen Purification ............. 31

Pressurization and Line Flushing ............ 33

Mechanical Testing in Low-Pressure Hydrogen at

Cryogenic and Ambient Temperatures ............. 35

Mechanical Testing in High-Pressure Hydrogen at

Cryogenic and Ambient Temperatures ............ 37

vii
 Mechanical Testing in Hydrogenat Elevated Temperatures 48

Chapter 6. Tensile Properties ........ 55

Room-Temperature Results ......... 57

Effect of Hydrogen Pressure ....... 76

Influence of Notch Severity ...... 81

Effect of Hold Time .......... 87

Effect of Temperature .......... 92

Embrittlement of Welds in Pressure Vessel Steels . . . 99

Effect of Hydrogen on Stress-Strain Curves ..... 103

Chapter 7. Other Medlanical Properties 107

Creep and Stress Rupture 107

Fatigue Properties 107

Fracture Mechanics 128

Crack Growth . 134

Chapter 8. Metallography 149

Slightly Embrittled Specimens 150

Severely Embrittled Specimens 152

Extremely Embrittled Specimens 159

Fractography of Specimens Fatigue Tested in Hydrogen 164

Chapter 9. Preventive Measures 169

Coatings 169

Impurity Additions to Hydrogen . 174

viii
Chapter i0. Discussion . 177

General 177

Tensile Tests 179

Surface Cracking in Hydrogen 182

Crack Propagation in Hydrogen 185

Effect of HydrogenEnvironmentson Other Properties 189

Metals Susceptible to llydrogen-EnvironmentEmbrittlement 189

Effect of Various Parameterson Degreeof Hydrogen-

EnvironmentEmbrittlement 191

Mechanism 195

Chapter ii. Summary. 203

Chapter 12. Recommendations 207

References 211

Appendix: Workin Progress 219

ix
 CHAPTER i. INTRODUCTION

Operating Experiences

In 1965, a space program contractor experienced a failure of a large high-

pressure hydrogen tank. Shortly afterward, a vessel failed at a NASA facility

while being filled to rated pressure with gaseous hydrogen, and a total of six

failures in five tanks occurred in less than 1 year. These were all pressure

vessels which, according to general knowledge at that time, should not have failed.

These events were important because the failures were experienced in large,

laminated (1500-cu ft volume or greater), thick-walled tanks designed to store

gaseous hydrogen at 5000 to 10 000 psi, and the investment in a single tank could

be as high as a quarter-million dollars. Fortunately, the failures were by leak-

age, or "whistlerS," but the potential damage and loss of life caused by the

explosion and fire of such an installation are staggering. A typical facilities

installations is shown in Fig. I.

In 1968, during the final stages of preparation for the Mariner '69 mission

to Mars, it was realized that a nickel-base-alloy hydrogen tank was subject to

the same environmental embrittling effects of gaseous hydrogen. Subsequent lab-

oratory work showed that the threshold fracture toughness for the Inconel 718

alloy was reduced to 15 percent of its basic fracture toughness for the base metal

and to 22 percent for weldments when tested in hydrogen. This dramatic reduction

of fracture toughness in hydrogen, led to the replacement of Inconel 718 in these

pressure vessels with an aluminum alloy, whose threshold fmacture toughness was

much less affected, thus reducing the probability of a tank failure and possible

loss of an expensive mission to Mars.

 .r-_

.,-_
U_
U_

.,-4

J
-r,I

0
,....1

.r4

°r'l

ffl

[..-,

.,.q
m
IA
°_',1
tel
m
.H

<

.r',l
 The alarming feature of these experiences was that the prior knowledge of

hydrogen embrittlement was not adequate to explain these effects. The large

storage vessels were made of low-strength structural grades of steel, normally

considered unaffected by hydrogen. So also with Inconel 718 and other nickel-

base alloys. The extensive work on delayed failure and internal-hydrogen em-

brittlement had shown that nickel-base alloys were unaffected by hydrogen except

under the most severe hydrogen charging conditions. At that time, an awareness

was just emerging regarding the differentiation of effects between ionic hydro-

gen in interstitial solution in the metal lattice and that of gaseous hydrogen

external to and surrounding a nucleating and advancing crack front.

The many unique properties of hydrogen have led to widespread and increas-

ing usage in nuclear programs, space programs and, in some cases, advanced air-

craft programs. In addition to pure hydrogen, there are many cases where com-

bustion or reaction products contain or produce hydrogen, and hardware must

operate in these environments. Frequently, hardware development programs re-

veal the necessity for design changes during the component development stage,

where adjustments for reduced load-carrying ability can be made. However, the

penalty of a downrated or overweight piece of hardware in a high-performance

system is too high to accept. Add to this the fact that a hardware develop-

ment cycle which permits life and environment simulation--even in the labora-

tory--is not always available. Many times a system is unique, only one or two

will be built, and a long life is required. The foregoing events and facts

lead to only one conclusion. Since we must work with hydrogen, we must under-

stand all the relationships between hydrogen and metals and the ultimate effects

of these relationships on operating components and systems.

 Hydrogen and Metals

The relationship between hydrogen and metals may be broadly placed into

three categories:

i. Hydrogen chemical reactions

2. Internal-hydrogen embrittlement

3. Hydrogen_environment embrittlement

The first two of these have been fairly well explored, while the third is

the subject of this report. All three will be described briefly here and in

more detail in Chapter 3, as sometimes there is confusion between these very

different effects.

One case of hydrogen chemical reactions is typified by the long-recognized

"steam reaction" in copper, in which hydrogen reacts with oxygen in solid copper

to form H20. A similar situation occurs with steel, when carbon reacts with

hydrogen in the lattice to form methane. In such cases, conditions of tempera-

ture, pressure, and chemical environment must be conducive to such reactions,

and the resultant gas pressures lead to development of fissures, strength reduc-

tion, and cracking.

Another type of hydrogen reaction is found in the hydriding of the base

metal or alloying constituents. Thus, under appropriate exposure conditions,

metals such as titanium, columbium (niobium), and tantalum form hydrides. Since

the propert_s of the hydride compound are so very different, the mechanical

properties of the base metal or of an alloy containing these elements may degrade

or, in the extreme cases, the metallic part may be converted to the hydride.

4
 Internal-hydrogen embrittlement is the case where high-strength steels

containing hydrogen in interstitial solid solution experience delayed failure

in service. Many examples have been found in the manufacture of hardened

springs, washers, etc. Perhaps the most dramatic cases were seen in the early

1950's when high-strength steel aircraft landing gear struts were under develop-

ment. There were cases reported in which a test aircraft would successfully

sustain one or more hard landings, imposing severe loads on the landing gear,

but later would collapse while standing parked on the apron. This type of

embrittlement has been studied in some detail, and the phenomenon of delayed

_ailure due to hydrogen in high-strength steel has been well characterized.

The principal characteristic of this effect is that hydrogen in solution is the

cause of delayed failure, and the effect is essentially independent of its exter-

nal environment during test or in service.

Hydrogen-environment embrittlement, as the name implies, takes place while

the metal is stressed in hydrogen. A specimen exposed to high-pressure hydrogen,

then tested in air does not exhibit this environmental effect. Thus, it was not

until specimens were placed in pure hydrogen gas at pressures on the order of

10 000 psi and the specimen loaded to failure in this environment that the effect

was isolated and reproduced. The experimental difficulties involved can be

significant.

Scope

It is the purpose of this work to isolate and describe hydrogen-environment

embrittlement as a separate and distinct effect of hydrogen on metals. Engineers

5
are generally acquainted with internal-hydrogen embrittlement. Too often they

try to equate all hydrogen effects in metals or otherwise extend our understand-

ing of internal=hydrogen embrittlement to explain hydrogen=environment

embr_ttlement.

If tae engineering community is to live with hydrogen (and it must), it

is mandatory that it be clearly understood. Although other kinds of hydrogen

effects exist, and may coexist with hydrogen-environment embrittlement, the

presence of the latter as a separate phenomenon has been well established in the

space program. There have already been inquiries about the characteristics of

this phenomenon from other segments of industry, government, and the academic

community. It is clear that a greater segment of the technical community will

be working with hydrogen ...... and its effects ...... in the years to come.
 CHAPTER 2. SIGNIFICANCE

The hydrogen-environment embrittlement of metals and alloys is observed in

gaseous hydrogen, particularly in high-purity, high-pressure hydrogen at room

temperature. Thus, any application in which metals are exposedunder these con-

ditions may be vulnerable to this effect. These conditions have become far more

commonplace in industry and government hardware and facilities with the growing

use of liquid hydrogen in the space program.

The potential performance advantages of liquid hydrogen fuel/liquid oxygen

oxidizer rocket engines have been well known since the beginning of modern

rocketry. Aside from unfamiliarity and handling hazards, the loss due to boil-

off during storage and handling were prohibitive and prevented early usage. It

was not until the 1950's that several developments took place that formed a base

from which large-scale production and use of liquid hydrogen could grow (Ref. I).

Availability of Low-Cost Liquid Hydrogen

Since hydrogen was first liquefied in 1896, it has been used in small quan-

tities in universities and research institutes in cryogenic studies and basic

studies of atomic structure. _le quantities of liquid hydrogen required for

this kind of work were limited, and there was no stimulus to large-scale produc-

tion or price reduction.

With the emergence and growth of the space program in the 1950's, propel-

lant chemists examined a large variety of fuels and oxidizers. The fuel then in

common use was RP-1, an adaptation of jet engine fuel, which is essentially
kerosene. The vacuum specific impulse* for the RP-1/liquid oxygen propellant

combination is 340 seconds while, for liquid hydrogen/liquid oxygen, the specific

impulse is 450 seconds. Thus, for high performance, hydrogen offers distinct

advantages for chemical and nuclear rocket engines.

There were three major developments which provided liquid hydrogen in quan-

tity: (1) the availability of hydrogen from a modified version of the Fischer-

Tropsch process, the original goal of which was the synthesis of liquid gasoline;

(2) the recognition of the ortho and para forms of hydrogen and, more important,

the ability to convert ortho to the more stable para form. This permitted

longer storage periods consistent with large-scale operations. Normally, liquid

hydrogen boiloff losses were 1 percent/hour, which was excessive for widespread

shipping, storage, and use_ and (3) the invention of "super-insulation" which

permitted piping, valving, and storage with a minimum of heat transfer.

Usage Outlook

It is clear that the space program will continue to use hydrogen, both in

the chemical and nuclear rocket engines. Although there are very serious prob-

lems in developing a hydrogen-fueled, air-breathing engine, the fact remains

that the potential performance payoff is handsome enough that most propulsion

experts agree that it will ultimately become a reality. In the chemical syn-

thesis of fuels and polymers, pure hydrogen has been in use for some time, and

it can only be expected to grow.

It is not the liquid hydrogen per 8e which is of concern in hydrogen envi-

ronment embrittlement. It is the fact that hydrogen is liquefied, thus freezing

*Specific impulse is the ratio of thrust to the mass flowrate, and an overall
measure of the efficiency of a jet or rocket engine.
 the common contaminant gases such as oxygen, water, or nitrogen, and rendering

the liquid hydrogen very pure. On boiling, the resultant gaseous hydrogen is of

the highest purity. Thus, an adsorption-related effect, as hydrogen-environment

embrittlement appears to be, is in no way poisoned or deterred by the presence

of impurity elements. Thus, any user who works with liquid hydrogen boiloff,

particularly at high pressures, is faced with the question of environment em-

brittlement for metal parts operating at high stress levels.

The Overall Problem

The recognition and concern for hydrogen-environment embrittlement is part

of a growing recognition of the sometimes dramatic and even insidious effects

of chemical environments on mechanical properties and hardware performance.

The gross effects of corrosion and the less conspicuous effects of stress

corrosion are generally well known, and designers are alert to them.

The presence of hydrogen, water vapor, and other gases can (and does) have

a pronounced effect on crack initiation and crack growth rates. This in turn

affects conventionai mechanical properties, and can drastically reduce fracture

toughness. Recognition of this is seen in the fracture mechanics approach to

the design and development of high-performance hardware. The determination of

plane strain fracture toughness in the identical proof and service media is

essential to the realistic application of these concepts.

Thus, there is a growing awareness in science and engineering of the

effects of environment on material properties and performance in real hardware

situations. There is a rapidly increasing number of cases where behavior in

 air or other convenient environment cannot be used to accurately predict per-

formance. _lis _s especially significant in those cases where high performance

is necessary or where the cost of a development or service failure is prohibi-

tive. With the trend toward precision design, there are penalties for over-

performance as well as underperformance. This design philosophy is being

adopted in a greater number of cases outside tile aerospace industry, and is the

direction in which our technology is moving.

The recognition and need for an understanding of hydrogen-environment em-

brittlement is a part of this growing awareness, and reaches out well beyond

the space program in its ultimate impact.

I0
 CI_PTER3. OVERVIEW

There are three types of hydrogen embrittlement that have been identified

for purposes of this report. _ey are classified as hydrogen-reaction embrit-

tlement, internal-hydrogen embrittlement, and hydrogen-environment embrittle-

ment. Many reviews have been published on hydrogen embrittlement, but none

make this vital distinction between these types of embrittlement which will be

described in this chapter.

There is no doubt that hydrogen-reaction embrittlement occurs by a dif-

ferent mechanism than the other two. There are similarities and differences

between internal-hydrogen embrittlement and hydrogen-environment embrittlement.

Many of the differences arise from the source of the hydrogen and how it is

transported to the source of embrittlement. It is not resolved at this time

whether the failure mechanisms are identical. In spite of over 3000 published

papers on internal-hydrogen embrittlement, there is as yet no concensus on the

exact failure mechanism for this type of embrittlement.

The oldest known type of hydrogen effect on a metal is that in which some

form of chemical reaction occurs between the hydrogen gas and the metal. This

type of reaction is prevalent in the petroleum industry in refinery operations

and sour gas wells. It also has been encountered in other chemical industries

where the use of gaseous hydrogen is required. Some of these instances are the

manufacture of ammonia, methanol, edible oils, etc. The problems stem from

the fact that all of these processes require the use of gaseous hydrogen, albeit

not high-purity, at elevated temperatures and pressures.

The result of exposure of steels to high-pressure, high-temperature hydro-

gen is decarburization (with the formation of methane) and intergranular

ii
 cracking (Ref. 2). A severely attacked steel may undergo a reduction of tensile

strength from 60 ksi down to 25 ksi with a concomitant decrease in the percent

elongation from 30 percent to zero. The attack can be prevented by the appro-

priate qhoice of a steel containing carbide stabilizers. This type of hydrogen

embrittlement has been studied by many investigators.

A similar type of hydrogen embrittlement is the well-documented "steam

reaction" in tough pitch copper where hydrogen reacts with oxygen in solid

copper to form water. This reaction, and the methane reaction in steel, can

lead to extensive blistering.

In those materials that can absorb large quantities of hydrogen (e.g.,

titanium, columbium, tantalum, zirconium), embrittlement is associated with the

formation of a hydride phase. Hydride phase formation can occur by a spontan-

eous transformation or can be strain-induced. Normally, embrittlement of

these alloys is related to hydride formation, and is most severe at fast strain

rates, except for the strain-induced transformation which can occur at slow

strain rates. Delayed failure also can occur under sustained loading by the

strain-induced transformation. The amount of hydrogen absorbed during ex-

posure to gaseous hydrogen is a complex function of temperature. The hydrogen

solubility decreases, but the rate of hydrogen absorption increases with in-

creasing temperature. The embrittling hydride phases are most stable at

low temperatures, at which comparatively low hydrogen contents may cause

severe embrittlement. Hydrogen absorption and embrittlement at room tempera-

ture is dependent upon surface cleanliness, prior exposure to hydrogen,

and hydrogen purity. Although embrittlement of the hydride formers is

thought to be strictly from the hydride, experiments conducted by Bentle

and Chandler (Ref. 3) indicate that columbium (niobium) and tantalum are

12
 extremely embrittled when exposed to high-pressure hydrogen at temperatures in

the range of 1000 ° to 1900°F, although the hydride is stable only to about 400OF.

Internal Hydrogen Embrittlement

Observations of internal hydrogen embrittlement have been made in a number

of metals and alloys; however, the vast majority of these observations have

taken place in steels. A description of the effects of internal hydrogen in

steels will illustrate the effects observed in other metals and alloys.

Internal hydrogen may be present in steels from a variety of sources,

almost all of them related to processing. During the steelmaking process, part

of whatever hydrogen may be in the melt will be trapped during solidification in

the ingot and significant amounts of hydrogen may be retained in heavy sections.

Pickling baths, during processing can and do, by virtue of the surface electro-

lytic reaction, deliver hydrogen ions to the metal surface. This permits ready

transfer across that interface and buildup of hydrogen in the metal. Hydrogen

can be picked up in a weld joint, and can be particularly severe in heavy

sections, if combined water is present in welding flux.

Perhaps the most widely publicized origin of internal-hydrogen embrittle-

ment is electroplating. Cadmium plating is frequently used to protect low-

alloy, high-strength steel parts, and was identified as being responsible for

delayed failure in springs, washers, and aircraft landing gears.

A more recently recognized source of ionic hydrogen is found in corrosion

reactions. In the presence of water, a part of the corrosion reaction is the

liberation and deposition of hydrogen on the metal surface, making it available

to the corroded metal in solution, much as in the case of pickling.

13
 In these instances hydrogen is present in dissociated fom and it goes into

the metal. The effects can be observed long after the sources are removed.

The presence of "fisheyes" or flakes in heavy steel forgings, castings,

and weldments has been known for some time. This is a spontaneously occurring

internal fissure, a fom of delayed failure, and has been associated (Ref. 4)

with stress in the presence of hydrogen. The decomposition of austenite into

its several transformation products is a common origin of stress in heavy sec-

tions, although flaking may be traced to stress other than transformation

stresses if hydrogen is present in the steel. Antiflaking thermal cycles have

been devised to cool a heavy section under controlled conditions which eliminate

transformation or other stresses.

The phenomenon known as delayed failure has been studied extensively, and

early work dates back to the 1920's. This was first systematically studied and

identified as a form of hydrogen embrittlement in 1955 (Ref. 5) and subsequently

was clarified by Troiano and co-workers in 1960 (Ref. 6).

The now traditional representation of delayed failure due to internal

hydrogen is shown in Fig. 2. The essential features as stated by Troiano (Ref.

6) are:

"1. The notch tensile strength may be less than normal and directly

reflects the loss of ductility due to hydrogen.

2. Delayed failure may occur over a wide range of applied stress.

3. There is only a slight dependence of the time to failure upon the

applied stress.

14
 1 Perhaps the most significant characteristic of this stress rupture

relationship-lies in the fact that there is a minimum critical value

below which failure does not occur."

Figure 2. - Failure characteristics of a hydrogenated high-strength steel.

Another characteristic of this type of failure is that the crack initiates

below the metal surface_ in the region of highest stress concentration ahead of

the root of the notch (Fig. 3). The crack continues to grow in a discontinuous

fashion by repeating the nucleation step ahead of the notch. This takes place

when the average hydrogen concentration in the metal is relatively low, and

hydrogen diffuses to the region of triaxial stress concentration_ at the root of

a notch, a surface scratch, or a stress concentration region as encountered in

real hardware.

IS
Figure 3. - Cracks observed in notched specimens sectioned after static loading.
Longitudinal section at X lO0.
Top specimen-.u.OOltin' notch radius
Bgttom.specimen - O,OIq-in. notch radius

Delayed failure occurs almost exclusively in martensitic steels heat

treated to high-strength levels and subjected to conditions permitting hydrogen

to enter the metal. Normally, hydrogen concentrations in the order of I to 5

ppm with appropriate stress conditions are adequate to cause this. In some

other alloys, internal hydrogen embrittlement can be produced in the laboratory

by severe artificial charging; however, the problem does not arise from normal

metal processing.

16
 Delayed failure may be avoided by eliminating hydrogen or reducing its

level to a harmless value. Improved steelmaking andwelding techniques, par-

ticularly for susceptible grades of steel, have virtually eliminated this origin

of hydrogen. Pickling baths and procedures are controlled, and much work has

been done in the area of electroplating to minimize the hydrogen deposited and

to redistribute hydrogen in the part by postplating baking treatments.

Hydrogen-Environment Embrittlement

When early failures of bourdon tubes and similar parts subjected to service

in high-pressure hydrogen were reported, it was evident from the circumstances

of the failure that hydrogen was responsible. Such failures, although not

realized at the time, did not fit the pattern which had already been established

by flaking and ddayed failure in statically loaded parts.

In 1965, when a very large investment in facilities tankage in the space

program appeared threatened by failure in high-pressure gaseous hydrogen, a

series of studies was undertaken to at least characterize the embrittling

effects of high-pressure hydrogen. The first comprehensive report on this sub-

ject was by Walter and Chandler in February 1969 (Ref. 7).

This work was the first to recognize the wide variety of metals and alloys

affected by gaseous hydrogen, and the authors established four categories of

embrittlement upon exposure to hydrogen at 10 000 psi and room temperature.

These categories are:

I. Extreme Embrittlement. High-strength steels and nickel-base alloys are

in this category. Embrittlement is characterized by a large decrease

(from that in helium) of notch strength and notched and unnotched

17
 ductility, and some decrease of unnotched strength. In unnotched

specimens, failure is initiated by one surface crack which propagates

into the specimen, leaving a thin shear lip around the periphery except

at the location of the initiating crack.

2. Severe Embrittlement. This category contains the largest number of the

metals tested, including ductile, lower-strength steels, pure nickel,

and titanium alloys. Embrittlement is characterized by a considerable

reduction of notch strength and notched and unnotched ductility, but no

reduction of strength of unnotched specimens. Failure initiates from a

series of surface cracks which, in some cases, are close enough to be

continuous, and no shear lip is formed.

3. Slight Embrittlement. The metastable (tend to transform to martensite

during deformation) Type 300 stainless steels, beryllium-copper, and

commercially pure titanium are in this category. Embrittlement is char-

acterized by a small decrease (from that in helium) in notch strength

and notched ductility. Failure of unnotched specimens appears to initi-

ate within the specimen, and shear lips are well formed. Very small,

shallow, blunt surface cracks are frequently observed.

4. Negligible Embrittlement. Aluminum alloys, stable austenitic stainless

steels, and copper are in this category. These materials were essenti-

ally unembrittled and no surface cracks were observed.

Contrary to prior experience with internal hydrogen, delayed failure was

not observed. The embrittlement was immediate once a stress level greater

than tho yield strength was reached. There is ample evidence (to be developed

18
in subsequent chapters) that cracking originates at the surface, even in the

case of notched specimens, where a high triaxial stress state exists at a point

ahead of the notch.

The crack growth rate was found to be a function of temperature and pres-

sure. The observed relationship has been attributed to a rate-controlling

adsorption process (Ref. 8).

Perhaps the most disturbing feature of hydrogen-environment embrittlement

is the large number of metals and alloys affected. It is easier to list those

alloys unaffected (aluminum alloys, stable austenitic steels, pure copper) than

those which exhibit the effect. Indeed, among those closest to the work, the

feeling is that, in the absence of prior knowledge, it is safer to consider an

alloy subject to hydrogen-environment embrittlement until experimental verifica-

tion of immunity has been established.

Comparison

Table 1 shows a comparison between some of the characteristics of internal-

hydrogen embrittlement and hydrogen-environment embr_ttlement. Excluded from

this comparison is the embrittlement caused by a reaction between gaseous

hydrogen and a metal.

Usually no effect of the hydrogen will be observed if tensile specimens,

internally charged with hydrogen, are tested at ambient temperature immediately

after charging. If, however, these specimens are held under a relatively low

stress for a period of time, embrittlement will occur. This is not true in

hydrogen-environment embrittlement. If a tensile specimen is tested in high-

purity gaseous hydrogen at ambient temperature and pressure of 1 atmosphere or

19
 TABLE I

COMPARISON OF CHARACTERISTICS OF HYDROGEN-ENVIRONMENT EMBRITTLEMENT

AND INTERNAL HYDROGEN EMBRITTLEMENT (Ref. 9)

Parameter Internal Hydrogen Hydrogen Environment

Tensile Tests Not Embrittled* Embrittled

Delayed Failure Test Embrittled Not Embrittled

Crack Origin Inside Metal At Surface

Crack Rate Dependency Ilydrogen Diffusion Probably Hydrogen Adsorption

Most Embrittled Materials High-Strength Steels High-Strength Steels and High-

Strength Nickel-Base Alloys

Materials Not Embrittled Low-Strength Steels, Austenitic Stainless Steels,

Austenitic Stainless Aluminum Alloys, Copper
Steels, and Nickel-
Base Alloys

Surface Cracking Not Observed** Observed in Unnotched Specimens

*May be observed in high-strength steels with high hydrogen concentrations

**Observed in some cases in martensitic-austenitic and martensitic steels

(Ref. lO)and in nickel-base alloys (Ref. II) when hydrogen is concentrated
at the surface.

2O
 greater, immediate embrittlement occurs. The amount of embrittlement observed

will depend on the type of specimen or alloy and the hydrogen pressure.

The origin of failure is different between the two types of embrittlement;

with internal-hydrogen embrittlement, the crack origin is subsurface, as with

most tensile tests. With hydrogen-environment embrittlement, the crack origin

has been observed to be at the surface.

Probably the most striking difference between the alloys subjected to

brittle failure in the two types of embrittlement is the behavior of nickel and

nickel-base alloys. Inconel 718 and Rene' 41 (nickel-base alloys) were not

embrittled by severe electrolytic charging (Refo 12), while these alloys were

among the most embrittled in 10 000-psi hydrogen. The high-strength steels are

subject to both types of embrittlement. Stainless steels are not subject to

either type of hydrogen embrittlement except those that transform to martensite

after some deformation.

Again, it should be noted that many of the differences in characteristics

between the two types of embrittlement arise from the difference in source of

the hydrogen during embrittlement and the rate-controlling step for transport

of the hydrogen.

21
 CHAPTER 4. EARLY WORK.

The embrittling effects of high-pressure hydrogen have been known for some time.

It was not until recently, however, that hydrogen-environment embrittlement was

identified as being distinct from internal-hydrogen embrittlement. The early

tests were all designed with the latter type of hydrogen embrittlement in mind.

Specimens were exposed to high-pressure hydrogen for a period of time, removed

from the environment, and tested. It was not until the late 1950's that tests

were conducted in a hydrogen environment. The early tests, while showing that

an effect may exist, are of historic value and are summarized in this chapter.

Early Service Failures Attributed to High-Pressure Hydrogen

There have been many service failures over tile years attributed to high-

pressure hydrogen. Bourdon gages were found to fragment within an hour or so

after filling with hydrogen at pressures of 800 to i000 atmospheres (ii 800 to

14 700 psi) (Ref. 13). Mills and Edeskuty (Ref. 14) reported bourdon tube fail-

ures after room-temperature exposure to hydrogen for times as short as 1 minute

and at pressures as low as 2/3 of the full-scale rating, even though these gages

had been proof tested with oil and helium at full-scale pressure.

Dodge (Ref. 15) reported that an intensifier, which had been used repeatedly

to pump oil at 4000 atmospheres (58 800 psi) and several times to compress nitro-

gen to the same pressure, failed within minutes when used to compress hydrogen

at pressures not exceeding 3000 atmospheres (44 000 psi). Failure was defined

as the development of very fine cracks that were barely visible to the eye.

23
 Failures also have occurred in several pressure vessels when they were

pressurized with high-pressure hydrogen. Bridgeman (Ref. 16) found that a Cr-V

steel vessel used to contain hydrogen at 9000 atmospheres(IS2 000 psi) developed

submicroscopic fissures which later developed into cracks visible to the eye,

although the vessel had previously withstood liquids at 25 000 atmospheres

(368 000 psi) pressure without damage.

Poulter (Ref. 17)reported that a 200-cu ft hydrogen cylinder, presumably

with 2000-psi maximum pressure, failed after being in service 25 years. Exami-

nation of other hydrogen cylinders that had not failed showed small cracks

extending from the inside surface to a depth of 25 percent of the wall thickness.

Nitrogen pressure vessels which had been in service about the same length of

time did not contain cracks.

Mechanical Tests Performed in High-Pressure Hydrogen

Prior to the early 1960's, the preponderance of tests performed in high-

pressure hydrogen were carried out at elevated temperatures to study hydrogen

attack on, or decarburization of steel. These tests were designed to measure

decarburization effects and not hydrogen-environment embrittlement. Thus, the

collection of elevated-temperature data plotted by Nelson (Ref. 18),i.e., the

Nelson curves, showing safe operating conditions, are not applicable for pre-

dicting low-temperature compatibility of hydrogen with steels.

Room-temperature, high-pressure hydrogen burst tests were reported by the

du Pont Chemical Co. as early as 1952 (Ref. 19). These tests were performed

24
 using oil and hydrogen as the pressurizing media. It was stated that "the

steels involved covered a variety of compositions, heat treatments, and wall

thicknesses." The results are quoted as follows: "In the most striking

instances these results showed that pressure vessels suitable for operation

under oil at pressures in excess of 7000 atmospheres (103 000 psi) failed in a

brittle fashion under hydrogen at pressures as low as 2000 atmospheres (29 400

psi) and in periods of time as short as a few minutes. Stainless-steel-lined

vessels provided with continuous, shallow, spiral grooves to prevent hydrogen

pressure buildup between the liner and vessel also were tested. It was found

that with these liners the hydrogen pressure required for failure was the same

as the oil pressure required for failure, indicating that the liner grooved-ves-

sel approach was effective in preventing hydrogen embrittlement. Furthermore,

the use of stainless-steel liners for periods up to several months was success-

ful in eliminating brittle rupture."

Dodge and coworkers (Ref. 20 and 21) measured the susceptibility of various

alloys to hydrogen embrittlement by measuring ductility after exposure to high-

pressure hydrogen. Ductility was determined in air by measuring the number of

180 ° bends required for failure after exposure. The results of Van Ness and

Dodge are given in Table 2 (Ref. 20). Perlmutter and Dodge extended this study

to include approximately 50 metals at exposure conditions ranging from 7500 to

60 000 psi at room temperature. Some of their results for the more severely

embrittled alloys are summarized in Table 2. These authors found no embrittle-

ment for copper, zinc, titanium, and a number of their alloys, and for commercial

grade 2-S aluminum, and only slight embrittlement for AISI types 321 and 450 stain-

less steels.

25
 .to

, ¢) •,O oo c,]

0 0 _

e-,._ ,--.I

,-t .,-I
OZ
Z U
r,.Ll
tD
0

,.-, .ic
>_ 0 0 0.1
-m

ell

0,-_ 0

U3

m" m
t_

sic I._ O0
Oe
0
.-r.

o
).....i

_ tel I I t_

2 °0
.Ic
,<
[.-,
_ 0

"1-1
Z
i,..-4

•,._ _ _
[.-, v 0 ,1-_

<
[...,

Z
I./.1
•_ 0 IJ

0
•,-4 .,_ .,-4
m •,-I U_ U_ U_
•el •,"I ',_
,--1 0
.rl 0 0 0
0 0 0 0
-_1 c_ _1 _ _ 0 0 0 0 0 0 0 0
0 0 _ 0 0

0
U O0 _' g "_ 0"'_0''_O0 r-- I'-- ,--4u_ _
0 Z,-_

m m
U_ U_ U_
c' c
0
•Ix tc ._
_ i,._f'xl "_1._ _OL_ 0 0 ,Ic ,_
0
Z tc

26
 The results of Perlmutter and Dodge (Ref. 21) for AISI types 420 and 430

stainless steels and for coated alloys Alfer and Nifer* are of interest.

Figures 4a and 4b show the effect of exposure time in high-pressure hydrogen on

the bend ductility in ambient-pressure air of AISI types 420 and 430 stainless

steels, respectively. Embrittlement occurred immediately in the AISI type 420

stainless steel; whereas, a definite incubation period exists even at 60 000 psig

for the type 430 stainless steel. The authors attributed this behavior to the

greater amount of chromium present in the type 430 stainless steel. In the type

420 stainless steel, there was an "insufficient amount of chromium to form a

stable oxide and, as a consequence, embrittlement begins to take place immedi-

ately upon exposure." Also of interest is the fact that the type 420 stainless

steel "never loses its ductile character completely."

Figures 4c and 4d show the ductility behavior of Alfer and Nifer in air

after exposure to high-pressure hydrogen. Perlmutter and Dodge again attribute

this behavior to surface oxides.

Alfer has a stable coating of oxide (AI203 type); whereas, the Nifer does

not. Further, aluminum is not embrittled by gaseous hydrogen but nickel is.

Thus, Nifer is embrittled upon exposure to hydrogen and Alfer is not.

*Alfer and Nifer, Hetals Control Corp., are low-carbon steels clad with
aluminum (containing 1.24 Si) and grade 330 nickel, respectively (cladding
thickness, 10_ on each side).

27
 A

.o
20'

m
z
"L I
t5
..=.
KEY: •
ioF-- -t- -- -l----J KEY: 10 SS 430-LOT I
>- /"- 7,500 PSi
I 1 ] IL_ 7,500 PSi O 46,OO0 PSi
..J I-1 | 5,OO0 PSi
i 51__
I 4__.
[ ______1
I i C_ls_s!
ss,2o-Lo,,
A_52,ooo
Ps!
'1 I [ IO_;ooo PSi 060+000 PSi 5 _,_,ooo Psi

+ t t I
045,000 PSI
52,000 PSI
60,0OO PSI
o _ 4 9 16 25 36 49 54
9 16 25
TIME IN DAYS, SQUARE ROOT SCALE
TIME IN DAYSp SQUARE ROOT SCALE

a. AISI type 420 stainless steel b. AISI type 430 stainless steel

t KEY: ALFER t

],'% 7,500 P91

/0 15,oooPsiJ
z_ IO ,5 oooPSl j
/ • sz,ooo PS, / t ]
. I I ]/• ,o,ooo
es,j I I
_, I xx ,I \ '

}1 I\] ,_ I_,1
_,oL t x,_ ...._,_L_ _ _ 2c

3c_

0 I 4 16 25 36 k9 0 I , 9 16 25 36 k9 6_

TIME IN DAYS, SQUARE ROOT SCALE TIME IN DAYS_ SQUARE ROOT SCALE

c. Alfer d. Nifer

Figure Jr. Ductility of various alloys exposed to hydrogen

gas at room temperature(ref. 21)
 The results of Perlmutter and Dodge for Haynes alloy 25 (_50% cobalt) show

a difference in behavior between the annealed, 35% cold reduced, and 50% cold

reduced conditions. The annealed material showed considerable resistance to

embrittlement, and the 35% cold-reduced samples lost all their ductile character

after 20 days exposure to hydrogen. The 50% cold-reduced samples were quite

brittle in the as-received condition, but the ductility was further reduced in

hydrogen.

Perlmutter and Dodge also performed additional tests at elevated tempera-

tures, but the results reported are not sufficiently complete for analysis.

They did report that tensile tests indicated that Armco Iron (_0.03% C) is em-

brittled at room temperature by 1000-atmospheres hydrogen and progressively more

at higher temperatures. This result would indicate that pure iron is suscep-

tible to hydrogen-environment embrittlement at elevated temperatures where there

is insufficient carbon present to cause degradation of properties from chemical

changes, viz., the mefhane reaction.

Recent Pressure Vessel Failures

More recently (1965), failures were reported by an aerospace contractor in

1300-cu ft, 5000-psig hydrogen storage vessels (Ref. 22). There were seven

vessels in service at 5000 psig, four containing hydrogen and three containing

nitrogen. In the hydrogen vessels, four failures occurred in three vessels in

a 1-inch nozzle at the top of each vessel. These failures could not be pre-

dicted from the early work in hydrogen=environment embrittlement. The nozzles

were fabricated from a low-alloy, 0.23-percent carbon steel with a 70-ksi

29
elastic limit, a 100-ksi ultimate tensile strength, and 23.S-percent elongation.

These nozzles were welded into the approximately 6-inch-thick pressure vessel

with a 1-inch weld surrounding the nozzle. The weld was not stress relieved

and, because of the multilayer design of the pressure vessels, there were

numerous stress raisers at the weld. Failure analysis of the last of the four

failures showed that the failure initiated at the nozzle/weld interface, and

propagated in a partially ductile, partially brittle manner through the nozzle.

The weld and nozzle materials were found to be essentially free of defects and

to have good mechanical properties. The failure stress analysis indicated that

the internal stresses in the nozzle were a large fraction of the yield strength

of the material. The nitrogen vessels have been in continuous operation under

identical conditions at pressures up to 5000 psi. Another failure occurred in

the weld area of a T-1 steel vessel being pressurized for the first time. The

vessel was proof tested with water at 7500 psi and pressurized three times with

nitrogen to 4200 psi prior to filling with hydrogen. The vessel was pressurized

with hydrogen in a series of steps followed by depressurization. The last steps

were 3500, 3800, and 3900 psi. On reaching 3900 psi, an audible failure oc-

curred. On inspection of the vessel, a 50-inch-long crack was found which ex-

tended through the vessel along a longitudinal seam weld.

30
 Ct_PTER 5. TEST PROCEDURES

Because the nature of hydrogen-environment embrittlement was little under-

stood, it was some time before the necessity to conduct tests in the hydrogen

environment was realized. This requirement has led to some unique testing

facility problems.

Hydrogen Purification

High hydrogen purity is important because impurities, particularly oxygen,

decrease hydrogen-environment embrittlement. The inhibiting influence of impur-

ities on hydrogen-environment embrittlement will be discussed in Chapter 9,

Preventive Measures. Very high-purity hydrogen in the form of liquid hydrogen

is used in rocket engine service; therefore, testing in the highest possible

practical hydrogen purity is consistent with simulating actual service conditions.

The purity* of bottled hydrogen ranges from 99.5 percent for that obtained

from the electrolysis of water to 99.9995 percent obtained from the evaporation

of liquid hydrogen. The less pure grades, less than -97°F dew point, (l.96-ppm

H20), contain 4.0-ppm N2, 0.5-ppm 02, l-ppm CO or CO2, and 2-ppm total hydro-

carbons. Contaminants in the highest-purity grade are mostly inert gases.

The most common method used for removing oxygen from gaseous hydrogen is

by installation of a Deoxo Unit in the high-pressure hydrogen lines. This is a

pressure vessel containing a catalyst which catalyzes the 2 H 2 + 02 ÷ 2 H20

reaction. The manufacturer (Ref. 23) reports that the remaining oxygen impurity

in gaseous hydrogen passed through the unit will be less than 1 ppm. Water

*Information supplied by AIRCO, Division of Union Carbide.

31
vapor is removed downstream of the Deoxo Unit by means of a desiccant such as

BaO, which has a high affinity for water and is not reduced by hydrogen.

A second method for purifying hydrogen is by cryogenic adsorbers. Acti-

vated charcoal, activated alumina, activated silica, and molecular sieves at

-320°F, will all scavenge 02, N2, CO, and CO 2 and other gases from hydrogen and

helium. In flowing systems, these impurity gases are rapidly adsorbed in mole-

cular sieves at -320°F (Ref. 24) to partial pressures of less than 0.02 torr

which, in high-pressure flowing hydrogen, results in 02, N2, CO, and CO 2 levels

considerably below l-ppm (Ref. 25). With static systems, even higher-purity

levels are obtainable. The cryogenic adsorbers have the advantages of being

easily installed in high-pressure systems.

The highest-purity hydrogen can be obtained by hydrogen evolution from

metals which absorb large quantities of hydrogen. Hydrogen evolves from the

hydride-forming metals (uranium, palladium, columbium, tantalum, and zirconium)

at about 1000°F, which is considerably below the temperatures that other gases

(N 2 and 02) are evolved. The metals are first saturated with hydrogen by heat-

ing in a gaseous hydrogen environment. ]]le lines are then evacuated and high-

purity hydrogen is evolved by heating the hydrided metals.

For continuous hydrogen flow, a closed-end palladium alloy tube is used.

Bottled hydrogen is maintained at tile inside wall, and purified hydrogen is

evolved at about 550°C (662°F) from the outside of the tube. Two companies mar-

ket hydrogen purification units utilizing this principle. Re maximum outlet

pressure of these units is about 150 psig, and thus multistage compressors are

needed for high-pressure service.

32
 Pressurization and Line Flushing

Experience has shown that contamination during pressurization and passage

of hydrogen through pressure lines can be far in excess of the ordinary impuri-

ties present in bottled hydrogen unless considerable care is taken to avoid

it. Oil-lubricated piston compressors are the most common means of pres-

surizing. These compressors contaminate the pressurized gases with oil, and an

oil trap (molecular sieve) downstream of the compressor is needed to remove the

oil.

Nonlubricated piston compressors also are available. Air, however, will

pass through the sliding seals into the pressurized gas. The mechanism involves

air adsorption on the shaft during the extension cycle, and the air is transferred

into the high-pressure cavity during the compression cycle. To prevent this form of

contamination, the compressor can be enclosed inside a cavity filled with hydro-

gen or an inert gas.

In diaphragm compressors, the gas being compressed is completely isolated

from air or hydraulic oil because there are no sliding seals in this type of

compressor. The gases are compressed by metal diaphragms which are flexed by a

pulsating hydraulic pump. A series of static O-rings prevents the oil from leak-

ing around the diaphragm periphery into the gas side. Thus, of the three types

of compressors, the diaphragm type is preferred for maintaining purity during

pressurization.

Air entrapment in high-pressure lines and test vessels is the main source

of contamination in high-pressure systems. Unless air removal is systematic and

thorough, the value of hydrogen purification is lost.

33
 Methods used for removing air are (1) flowing hydrogen through the lines,

(2) pressurization/depressurization, and (3) evacuation. Purging lines with

flowing hydrogen is quite inefficient because of the large momentum difference

between hydrogen (the lightest of molecules) and the air. The pressure cycling

method is capable theoretically o£ removing all contaminating gases entrapped

in the lines. However, adsorbed gases are not effectively removed by this

technique. Evacuation is the only method that can ensure adequate removal o£

air. Unfortunately, high-pressure valves and fittings are not designed for

vacuum, and experience indicates that the best vacuum that is reasonably obtain-

able is about 20 microns. Even at this level, considerable effort is needed to

maintain a vacuum/high-pressure, leak-tight condition.

Flowing hydrogen and pressurization/depressurization are, of course, also

capable o£ removing entrapped and adsorbed air molecules. Vacuum, however, does

this most efficiently. In hardware application such as hydrogen-fueled propul-

sion systems, continuous flushing with flowing hydrogen should be every bit as

effective for removing entrapped and adsorbed gases as the evacuation treatment.

Thus, the final purity state from evacuation of test systems appears to most prop-

erly simulate that in flowing hydrogen systems.

The flushing procedure proved to be the most effective for high-pressure

work is a combination of pressure and vacuum cycling. Greater embrittling

effects were found with this procedure which has been adopted at Rocketdyne

(Ref. 7). The system is initially evacuated, then purged by a series of pressure-

vacuum cycles between 100 psig and 20 microns hydrogen. This is followed by sev-

eral pressurization/depressurization cycles between the test pressure (3000 to

10 000 psi) and approximately 1-atmosphere pressure. For tests conducted below

3000 psi, several additional pressurizations/evacuations are substituted for the

34
higher-pressure pressurization/depressurization purgings. The resulting hydrogen

purities of samples taken from the test cells following a mechanical test showed

about l-ppm N 2 and less than 0.5 ppm total of other gases (02, A, CO, C02, and

CH4) (Ref. 26). A comparable pressure-vacuum cycle is employed by others

(Ref. 27 and 28).

Considerable maintenance is required to maintain a vacuum leak-tight and

pressure leak-tight system. Experience indicates that if the system is capable

of being evacuated to 20 microns, it probably will be leak tight in the many

pressure fittings, as indicated with a helium leak check with 10 -7 std cc/sec

sensitivity.

To be sure that the hydrogen purity is adequate, the hydrogen present in

the test vesel must be analyzed. Also, precaution must be taken to ensure that

the sample represents the test environment. A sampling container (pressure ves-

sel) is installed in the pressure system, the sampling container and pressure

lines are evacuated several times as described above, and then are filled with

the test environment. The standard analysis technique is by gas chromatography

(Ref. 29 ). The hydrogen impurities are collected in a cryogenic trap of molecu-

lar sieves or activated charcoal, then released into a gas chromatography column.

Mechanical Testing in Low=Pressure Hydrogen

at Cryogenic and Ambient Temperatures

The apparatus used for performing tests in low-pressure (less than 1 atmos-

phere) hydrogen are an adaptation of vacuum testing systems. An example of a

versatile low-pressure test apparatus is that developed by Yarcus and Stoc1_er

(Ref. 30). The apparatus is illustrated in Fig. 5, and is basically a high-

vacuum system with gas inlet valves. The apparatus was constructed to follow

35
 LOADING
RAM

I GASINLET ION

BOURDON TUBE
BELLOWS _

PUMP

I ON PUMP

TO HEAT
EXCHANGER

0___
SORP" ION
PUMP

Figure 5. High-vacuum system used for performing tests in

controlled low-pressure environments (ref. 30)

36
crack growth in low-pressure environments in the tapered double cantilever beam-

type specimens shown in Fig. 5. Williams and Nelson (Ref. 31) used a similar

apparatus.

Important features of the system used by Marcus and Stocker (Ref. 30) are

(i) very high vacuum (10 -10 torr), which is sufficient to eliminate effects due

to other contaminating environments; (2)a heat exchanger that contacts the

specimen and in which a precisely temperature-controlled fluid circulates; and

(3) a bellows through which a tensile load is transmitted into the vessel with-

out contaminating the environment.

The test procedure involves initial bakeout of the vessel at 250°C (482°F)

and 10 -8 tort. The system then is cooled to the test temperature, and a speci-

al]y prepared high-purity sample of hydrogen is introduced into the vessel.

The influence of other gases, such as oxygen, on the hydrogen-environment effects

can be tested by subsequent introduction of low partial pressures of these gases.

The rate of crack extension is monitored optically through a glass port.

Other investigators (Ref. 32 and 33) have used a small environmental chamber

which clamps to the test specimen as described in a subsequent section of this

chapter.
Mechanical Testing in High-Pressure Hydrogen at

Cryogenic and Ambient Temperatures

Testing in high-pressure environments has its own unique problems. Apparatus

used can obviously be used at pressures down to 1 atmosphere. At high hydrogen

pressures transmittal of the load from outside the vessel to the specimen is a

37
 major consideration. Bellows that can withstand high internal pressures are

too stiff for attachment to a loading ram. Thus, sliding seals are genera]ly

preferred for high-pressure systems.

The high pressure inside the vessel tends to force the loading ram back out

of the vessel, and this load can be a significant percent of the total load

applied to a specimen. This pressure force actually can be utilized as the sole

tensile force for tensile testing specimens in high-pressure hydrogen. Figure 6

is a schematic representation of such an apparatus (Ref. 34). The test is con-

ducted by slowly increasing the pressure until the specimen fails and is accel-

erated out the tube. A major deficiency of this test method is that the pres-

sure and load cannot be varied independently without varying specimen size and,

since embrittlement is a function of hydrogen pressure, it is difficult to

interpret the test results.

The most common method of performing tensile tests in high-pressure hydrogen

environments is by extending the test specimens through sliding seals at both

vessel ends, as is illustrated in Fig. 7. This test concept was developed initi-

ally at Rensselaer Polytechnic Institute (Ref. 35) and has been used at Rocket-

dyne (Ref. 7) with modification to the sliding seal design to eliminate metal-

to-metal contact. Elimination of metal-to-metal contact at the sliding seals

reduced the sliding seal friction, and electrical isolation between the speci-

men and vessel made it possible to follow crack growth during the test by elec-

trical resistivity measurements.

The test vessel can be placed in any mechanical testing machine. The load-

ing device illustrated in Fig. 8 has been used for the tests conducted at

38
 /

2.0

t J
1
z. zSo = o. 8 I_-_'-
6.38

4340 STEEL
HEAT TREAT TO
ST
160,000-180,000 PSI SPECIMEN

9/16-18 THREAD
O-RING 7/16 DEEP

O-RING

3.25

I.25
O.O6

FOR 1/4-I_I. FLAT-W I

FOR 8-32
0. 386 SCREW HEAD SCREW

Figure , m Apparatus for tensile-testing specimens in high-pressure hydrogen.

The system pressure is the only. tensile force applied.

 _SEAL PLUG

STEEL _.___TEFLON O-RINGS

BACKUP RINGS
I

- - - I

I
_ GH2 VENT

GH2 IN

316 STAINLESS
! STEEL VESSEL

TEST SPEC

Figure 7. - Test vessel with cylindrical specimen (ref. 7).

 I
AND SEAL

21MEN

GH 2 VENT
GH 2 _

I
23 IN.
FRAME

SPRING

FORCE

THRUST BEARING

Figure 8. Test vessel and static-loading device

(ref. 35)-

41
 Rensselaer Polytechnic Institute and Rocketdyne. It is very useful for conduct=

ing long-duration hold tests because several loadingdevices can be set up with

specimens under load in the high-pressure environment at the same time. The

tensile load is applied by compressing the spring, and the spring deflection is

held by the nut riding on the thrust bearing.

Tensile tests are performed at Rocketdyne (Ref. 7) by compressing the spring

with a hydraulic ram. Strain and load pacing are obtained by controlling the

rate of travel of the hydraulic ram. Spring deflection is monitored for load

pacing of notched specimens, and the cross-head motion is monitored for strain

pacing of unnotched specimens. Load measurements are made by means of a load

cell located between the hydraulic ram and the spring, and the load is recorded

on a strip-chart recorder.

Calculation of the tensile load requires that the sliding seal friction and

tensile load from the high-pressure gas be considered. The following equation

is used to calculate the load at failure of unnotched specimens:

Load = applied load - friction + pressure (specimen area at sliding

seal - specimen area prior to necking)

The maximum combined tensile load is assumed to occur prior to necking.

For notched specimens, the final area at the base of the notch is used in place

of the "area prior to necking" in the above equation.

A second effect of extending the specimen ends outside the pressure vessel

is an increase of shear stress from the hydrostatic forces which act in two of

the three principal directions. It can be shown that the two-dimensional pres _

sure component increases the shear stress on the 45 ° maximum shear stress plane

42
 to the same extent as the uniaxial applied tension load. That is, the specimen

should have a I0 000-psi lower yield strength in I0 000-psi pressure than in l-

atmosphere pressure. This has been verified experimentally (Ref. 7) in tests

conducted on 35 various engineering alloys in air (I atmosphere), in I0 000-psi

helium, and in i0 000-psi hydrogen.

The same technique of gripping the specimen outside the vessel has been

used (Ref. 36) for testing sheet specimens. The vessel is clamped to the spec-

imen as shown in Fig. 9. The center of the specimen is exposed to high-pressure

hydrogen while the specimen ends are exposed to air. Tile same pressure and

sliding seal effects discussed above on the cylindrical specimens are also appli

cable. The O-ring area is considerably larger for the sheet specimen than for

the cylindrical specimens and, therefore, sliding friction is considerably

higher during the sheet specimen tests.

For other types of specimens and/or tests, it is generally necessary to

design the pressure vessel so that the entire specimen is inside the vessel and

the loading ram extends into the vessel through sliding seals. Thus, the vessel

contains the specimen and loading frame, and is comparatively large and expensive.

Figure i0, a schematic of a vessel used to perform bend test fracture toughness

measurements (Ref. 7 ), illustrates this point. The specimens were 1 x 2-1/2 x

14 inches and the vessel ID is 6 inches, while the inside length is 16 inches.

The load was applied by means of a 65 000-pound loading ram that extended into

the side (top of picture) of the vessel through sliding seals. A load cell

located inside the pressure vessel is attached to the loading ram. Thus, the

exact load applied to the specimens is measured directly without the need to

compensate for sliding seal friction or for pressure effects.

43
 SPECIMEN-TO-CONTAINER ENVIRONMENT CONTAINER
CLEARANCE = 0.001 TO 0.002 INCH BACKING PLATE

SURFACE FLAW TEST

SPECIMEN-TO-CONTAINER SPECIMEN
CLEARANCE SPACERS

VITON "A" O-RING SEAL

CONTAINER
CLAMPING \
BOLTS (4) \
\
\
\
\
\
l
ENVIRONMENT CONTAINER SPECIMEN CONTAINER
CLAMPING PLATE

5200-PSIG
HYDROGEN

Figure 9- Sheet specimen cls_ped in frame with surface flaw

exposed to 5LKX)-psigbydrogen.
 s_

-rl
¢1)
P_
o I
o
\ o
o
,-I

GJ

_J

_o
.,-I

o
,-I

45
 A special pressure vessel was fabricated to abrade cylindrical tensile

specimens in high-pressure hydrogen (Fig. II). An abrading tool located at the

longitudinal centerline around the vessel extends from the sides into the

middle of the vessel to contact the tensile specimen. The specimen is abraded

by rotating the specimen with the tool in contact with it. The electrical feed-

through openings are to facilitate electrical resistivity measurements while the

specimen is abraded, or during tensile or fatigue testing in high-pressure

hydrogen.

To perform tests at cryogenic temperatures in high-pressure hydrogen, the

pressure vessels are immersed in a cooling medium and the temperature inside the

vessel is monitored. Metal-sheathed thermocouples can be inserted into the

pressure vessels through high-pressure glands.

The sliding seal friction is considerably higher at cryogenic temperatures

and this increased friction decreases the accuracy of the load measurements if

a load cell is not located inside the test vessel. As an example of the magnitude

of the sliding seal friction, the static friction on a 0.506-inch-(shaft) diam-

eter specimen illustrated in Fig. 7 was measured to be 56, 85, 566, and 500

pounds at 140 °, 74 ° , -109° and -520°F. The measurements were made while the

vessel was pressurized with 2000-psi helium and with the sliding seals tightened

sufficiently to hold this pressure without measurable leakage. With increasing

sliding seal area, the sliding seal friction would increase proportionately.

46
 hi)
0

r/l

-i-I

°_
o

r-t
ffl

,¢-t

t'--

4'[
 With large pressure vessels, the compatibility of the pressure vessel mater-

ial with the high-pressure hydrogen is an important consideration. The _terials

that are not embrittled (Ref. 7) by hydrogen are copper alloys, alumintun alloys,

and the stable (will not transform to martensite during cold working ) austenitic

stainless steels. _e smaller vessels, such as shown in Fig. 7 and 11, are con-

structed of AISI type 316 stainless steel. _e large vessel shown in Fig. 10

contains a vented AISI 516 stainless-steel liner in contact with high-pressure

hydrogen. _e structural part of this vessel is constructed of low-alloy steel.

Hechanical Testing in Hydrogen at Elevated Temperatures

It is considerably more difficult to perform tests in high-pressure hydrogen

at elevated temperatures than at lower temperatures. 3_ne problems are difficulty

of transmitting the load into the vessel, and strength and compatibility of the

vessel materials in contact with hi_l-temperature, high-pressure hydrogen.

3"nese problems, however, are not particularly severe for hydrogen pressures

less than a few atmospheres. _e high-vacuum system described in Fig. 5 can be

used to 250°C (482°F) in low-pressure hydrogen. In this system, the specimen is

heated by contact with a heat exchanger. In higher-pressure hydrogen, the hydro-

gen gas would quickly transfer the heat to the vessel wall because of the high

theIlnal conductivity of gaseous hydrogen at high pressure.

A test apparatus designed for tensile testing at elevated ten_eratur_s in

vacuum or inert-gas environments may not be suitable for testing in low-pressure

hydrogen because the heat reflectors and heating elements are likely to be very

reactive with hydrogen. Of the refractory metals Ccolimbium, tantal_un, molyb-

denum, and tungsten), colu_ium and tantalum are the easiest to fabricate into

48
 heaters and heat reflectors, but these two metals will form hydride in the

presence of hydrogen. Molybdenum and tungsten, which do not hydride, are

therefore better suited for this type of service.

The most feasible method of performing tests in ambient-pressure hydrogen

at elevated temperatures is by heating a tubular test chamber from the outside.

Tensile tests at elevated temperatures (72 ° to 1500°F) in 1-atmosphere hydrogen

(Ref. 37) and in precisely controlled ambient-pressure hydrogen/water vapor mix-

tures (Ref. 38) have been performed in thiS type of apparatus. The test chamber

consisted of a nickel alloy tube with a removable water-cooled flange and a slid-

ing seal at each end of the tube. For the hydrogen/water vapor tests, hot sili-

cone oil was used in place of water as the coolant to prevent the water vapor

from condensing.

The same basic method can be used for elevated-temperature testing in high-

pressure hydrogen. Figure 12 is a schematic of a vessel to be used at Rocketdyne

for performing fracture toughness tests on compact-tension and modified WOL speci-

mens in 5000-psi hydrogen at cryogenic and elevated temperatures (-320 ° to 1200°F).

The test specimens are anchored to the vessel cover and loaded by the ram, which

extends through sliding seals located in the water-cooled necked region. It is

advisable that the sliding seals be maintained at room temperature during

cryogenic- and elevated-temperature testing because of increased sliding friction

at cryogenic temperatures and the temperature limitations of sliding seal materi-

als at elevated temperatures.

49
 1-_1 COVER , A-286

S I LVER-PLATED
_ _ _;:::_-STUDS AND NUT, A-286

_SEAL RING, 316 SS

I I BODY, A-286
I

]0 II

611 .__._1_

G
EXTENSION TUBE D A-2_

j___M I HIGH ALLOY

_ _ _ _._T,ON_EO_NO
_ _S INLET/VENT-1/4-1-'_-. HIGH PRESS.

_,_ I _ ELECTRODE (6 LEADS)

_ THERMOCOUPLE (3 COUPLES)
_'_ COOLING OUTLET - I/8-1N. NPT
V
Q COOLING INLET - I/8-1N. NPT

Figure 12. Pressure vessel used to test modified WOL

specimens in hlgh-pressure hydrogen at
from -320 ° to 1200 ° F.
 The pressure vessel is constructed of A-286, a precipitation-hardened aus-

tenitic stainless steel. A-286 is not particularly high strength (125-ksi yield

and 170-ksi ultimate), but it is the strongest alloy tested thus far which is

not severely embrittled by high-pressure hydrogen. High-temperature pressure

vessels would ordinarily be made of high-strength nickel-base alloys, but these

are extremely embrittled (Ref. 7 ) by high-pressure hydrogen.

Materials limitations, therefore, place definite restrictions on the tem-

peratures at which pressure vessel materials can be heated. To extend this

range, it is necessary that the vessels contain an internal heater and tfiat the

vessel walls be water cooled. A system of radiation shields is not effective

because of high heat transfer due to high thermal conductivity and rapid convec-

tion of gaseous hydrogen at high pressure. Outgassing from any ceramic insula-

tion, however, would make it virtually impossible to maintain any degree of

hydrogen purity unless the insulation is isolated from the test specimen.

The apparatus used at Rocketdyne for performing tensile tests on cylindrical

specimens at temperatures up to 2000°F is shown schematically in Fig. 13 (Ref. 26).

The test vessel has a double wall for water cooling, and includes a tube which

separates the specimen cavity from the furnace cavity. A water-cooled lower

extension of the vessel contains a load cell. The specimen cavity, furnace

cavity, and load cell cavity are separated from each other by Viton-A O-ring

seals. The same gas at the same pressures is fed into the three cavities during

the tests in order to minimize gas passage across the O-rings. Self-aligning

ball joints are located inside the vessel to ensure that a uniaxial load is

applied to the specimen.

51
 I"t-- GASEOUS HYDROGEN INLET

WATER-COOLED TEST
VESSEL (3If STAINLESS
VESSEL LIFT STEEL)
EXTENSION ROD
COUNTER WE IGHT--_N k
HEATING ELEMENTS

1 _______---_-SPECIMEN
SPECIMEN
P EXTENSION

INSULATION
GRIPS

I2 FEET

BALL JOINT

fESSEL DETACH POINT

PIPE SUPPORT

FRAME _

=,.-----VESSEL SUPPORT FRAME

_--VESSEL CONTAINING
LOAD CELL
I
I
I I

TENSILE DRIVE
MECHANISM

Figure 13. Apparatus for tensile-testing cylindrical

specimens in hlgh-pressure hydrogen at
elevated temperatures.

52
 Hollow specimens can be a. convenient method for performing tests in high-

pressure, high-temperature hydrogen; however, notching the inside of a cylindri-

cal specimen is difficult. Considerable care must be exercised in prepara-

tion of the hoIe and its surface. A disadvantage of this type of specimen is

data interpretation, since virtualIy all engineering data are determined on

solid cyIindrical specimens.

53
 CHAPTER
6. TENSILEPROPERTIES

This chapter reports the results of investigations on the tensile properties

of various metals and alloys in gaseous hydrogen environments. All of these data

cannot be collected into homogeneous figures and tables due to the wide variety

of methods used by the different investigators. As described in Chapter 5, the

test procedure can have a pronounced effect on the test results. For example,

if the test cells are not evacuated prior to admitting hydrogen, the full extent

of the embrittlement may not be observed. Some testing of tensile specimens has

been performed with various hold times in the hydrogen environment prior to

testing. This type of test has been performed with and without an applied

stress present during the hold time. The data on the effect of temperature on

the susceptibility to embrittlement are quite limited as is the effect of impur-

ities in the hydrogen atmospheres. Factors such as these have not always been

considered.

Among the materials that are embrittled by exposure to gaseous hydrogen are

the high- and medium-strength steels, high-strength nickel-base alloys, pure

nickel, some titanium and some cobalt alloys, and the metastable (with regard to

the austenite-martensite transformation) 300-series stainless steels. These are

the materials found to be susceptible to hydrogen-environment embrittlement, and

it is by no means a complete list of materials which may be affected by gaseous

hydrogen environments.

A large number of alloys have been investigated by Walter and Chandler, and

on the basis of tensile tests on unnotched and notched specimens in 10 O00-psi

hydrogen at room temperature, they have classified the susceptibility to hydrogen-

55
environment embrittlement into four categories, viz., extreme, severe, slight,

and negligible (Ref. 7). These categories are defined as follows and serve to

orient further discussion.

1. Extreme embrittlement: High-strength steels and high-strength nickel-

base alloys are in this category. Embrittlement is characterized by a

large decrease of notch strength and ductility, and some decrease of

unnotched strength. Large reductions of ductility are found for all

three common measures of ductility (elongation and reduction of area of

unnotched specimens, and reduction of area of notched specimens). Sur-

face cracks are not usually found in failed specimens.

2. Severe embrittlementi The majority of the metals tested are in this

category, including ductile, lower-strength steels, Armco iron, pure •

nickel, and the titanium-base alloys. Embrittlement is characterized

by a considerable reduction of notch strength and ductility, but no

reduction of unnotched strength. The measure of ductility most affected

by the high-pressure hydrogen environment is the reduction of area of

notched specimens. A large number of deep surface cracks occur in

failedspecimens.

3. Slight embrittlement: The nonstable AISI 300-series stainless steels

(AISI 304L and 305), beryllium-copper, and commercially pure titanium

are in this category. Embrittlement is characterized by a small

decrease of unnotched ductility. Infrequent shallow surface cracks

occur in failed specimens.

56
 4. Negligible embrittlement: The aluminum alloys, stable austenitic

stainless steels, A-286 (a precipitation-hardened austenitic stainless

steel), and OFHC copper are in this category. Surface cracking is not

observed in failed specimens.

The average ratio of tensile properties in hydrogen to helium are shown in

Table 3 and this classification will be used wherever possible.

TABLE 3

CHARACTERISTICS OF HYDROGEN-ENVIRONMENT EMBRITTLEMENT CATEGORIES

Average Ratio of Tensile Properties in

I0 000-psig tlydrogen to I0 000-psig Helium

=.
Notched Specimens Unnotched Specimens
Reduction Ultimate Reduction Ultimate"
Category of Area Strength of Area Elongation Strength

Extreme 0.16 0.29 0.11 0.07 0.77

Severe 0.51 0.75 0.62 0.79

Slight 0.80 0.91 0.96 0.98

Negligible 0.91 0.99 ].0 0.97 1.0

Tensile tests performed on columbium and tantalum in gaseous hydrogen en-

vironments resulted in severe embrittlement. However, embrittlement was associ-

ated with absorption of large quantities of hydrogen, and thus may be an internal-

hydrogen rather than a hydrogen-environment phenomenon. Internal-hydrogen embrit-

tlement of refractory metals has been reviewed by Chandler and Walter (Ref. 39).

Room-Temperature Results

Hofmann and Rauls performed the earliest experiments in which tensile tests

were conducted in gaseous hydrogen. Their results on the effect of gaseous

57
hydrogen on a 0.22% carbon steel in 1961 are shown in Table 4 (Ref. 40). No

strength values were reported. Purity of the hydrogen was 99.95%.

TABLE 4

TENSILE PROPERTIES OF LOW-CARBON STEEL IN HYDROGEN

Room Temp. UTS, ksi % Elong. Red. of Area, %

Air (ambient press.) 70.8 32 64

34.S 52
147 psig H2

33 47
294 psig H2

30 S0
735 psig H2

30 36.5
1470 psig H2

26 28
2205 psig H2

1470 psig Argon 36 62

At 147 psig, many cracks formed, with increasing frequency

of cracks to 2205 psig.
CK22 (0.22% C) steel normalized at 900°C, extension
rate of 1.75 mm/sec, gage section 6mm x 30 mm.

In a second series of tests utilizing co}d-drawn 0.22% carbon steel, llofmann

and Rauls (Ref. 41) investigated the effect on the degree of embrittlement occasioned

by various hydrogen pressures and various additions of oxygen and nitrogen to the

high-pressure hydrogen, This latter topic will be discussed later. A lowering

of the ultimate tensile strength and rupture elongation with increasing hydrogen

pressure was reported (Fig. 14). These authors also investigated (Ref. 42) the

effect of strain rate on ductility at a pressure of 2205 psig (Fig. 15). The

58
 128

_ I]4
xl
. I00
1-

Z
85

u_ t

"' 57 !

I
.J
1
I
z 43
p-

i I

ATMOS. 2 5 I0 20 50 O0 200
PSIG 14.7 147 1470
HYDROGEN PRESSURE

Figure 14. Effect of hydrogen pressure vs tensile

strength.

20 .....
I

"ii
o
,._1
""
I0
o
............
, .. I I
I I

_ 60,,

50
,
i I
] I''
I | I I
I
M -i
I
!!I
I I

< 40 iI
/
h

30 --- I
0
loll I

__ 20 I • 0 : : : " I ti
I
" 10

0 i: I111111
IIII
I00 200
3 5 10 20 5O

DEFORMATION RATE, % MIN.

Figure 15. Effect of deformation rate on elongation

(in 5D) and reduction in area (normalized
CK22 steel, exposed to hydrogen pressure
of 2205 psig).
59
largest effect of hydrogen is noted at the slowest strain rate and the effect

decreases with increasing strain rate with, however, a significant effect still

evident at the fastest strain rate.

These authors (Ref. 42) also reported on additional experiments with Armco

iron and 0.45% carbon steel. The ductility of these steels in hydrogen environ-

ments from 14.7 to 2205 psig is shown in Fig. 16. The effect of a 2205-psig

hydrogen environment on the ductility of various plain carbon steels is shown

in Fig. 17.

Hofmann and Rauls further investigated the effect of gaseous hydrogen with

notched tensile specimens of the 0.22% and 0.45% carbon steel and a high-strength

steel specimen. For the carbon steel, the specimens were 8 mm in diameter with

a circumferential notch of 60 ° included angle. The notch was l-mm deep and had

a root radius of 0.i ram (Kt = 5.1). For the high-strength steel the specimen

was 4.3 mm in diameter with a 60 ° notch, a 1.15-mm notch depth, and a 0.08-mm

root radius (Kt = 3.5). Results are shown in Table 5.

TABLE 5. NOTCHED AND UNNOTCHED TENSILE STRENGTH IN AIR AND HYDROGEN (Ref. 42)

UTS, ksi
Notch
Strength
Air 2205-psig H2
Ratio
Material UN N UN N I[2/Air

Armco Iron 51.4 71 48.6 51.4 0.72

i0.22 C 71 102.8 71 76.6 0.75

Normalized

0.45 C 96.2 119.6 96.2 93.4 0.79

iNormalized

High-Strength 2-33.3 312 233.3 156 0.50

Steel*

*In 1470-psig H 2. Chemical composition: 0.71C; 0.65 Si;

1.24 Mn; 0.035 P; 0.028 S; 0.005 N; 0.006 AI

6O
 awP ,51
=- 3O
0

< 25_
Z
0
J 20

15

• I"_

50
Z
0 go (
P
U
3O
I I , I
20 I I I a 0 0 ,_|
ATMOS. 2 5 I0 2O 50 I00 2OO 5OO
PSIG 147 1470
HYDROGEN PRESSURE

Figure 16. Ef_ct of high-_ess_e bydro_n on

elongation (in 5D) and reduction in area
(_eo iron, _ 22 N, _ C 45 N steels).

40
$ -i
.S 30

o
20

O 10
o
0
60
?0
60
.S
50 'T_
40
o
v
30 i
o
o 20 Tests in
e AiGot latin
I0
oHydroge nl_crt.
0 l_U orm
o QI O.2 0,3
Carbon Content in °/o

Figure IT. Effect of carbon content on elongation

and reduction in area in air and in
high-pressure hydrogen.

61
 A series of tests was conducted at Rensselaer starting in about 1957

(Ref. 3S). The test specimens used were cylindrical rods with threaded ends.

The diameter was 0.306 inch, with a 60 ° V-notch at the midplane of the specimen.

The diameter of the specimen at the root of the notch was 0.1S00 inch, and the

root radius was 0.0046 inch. This gave a stress concentration (after Peterson,

Ref. 43) of Kt = 4.0. Most of the tests were conducted with a hold time in the

hydrogen atmosphere and will be reported later. The purity of the hydrogen was

not reported.

Cavett and Van Ness reported in 1963 that K-monel and quenched and tempered

4140 steel are severely embrittled as compared to normalized 4140 steel (Table 6,

Ref. 44). Vennett and Ansell (Ref. 45) also investigated the effect of strain

rate on the tensile strength and percent elongation of AISI 304L stainless steel.

Their results are shown in Fig. 18. The degree of hydrogen-environment embrittle-

ment is again observed to decrease with increasing strain rate.

In the most comprehensive series of tests, Walter and Chandler (Ref. 7)

tested a large number of materials in l0 000-psig hydrogen, and the degree of

hydrogen-environment embrittlement was determined by comparison with the mechan-

ical properties determined in l0 000-ps_helium. The test specimens were 9 inches

long and 0.306 inch in diameter. The surface finish was 16 rms. The unnotched

specimens had a 1.2S-inch-long, 0.250-inch-diameter gage section. The notched

specimens had a 60 ° notch at the midplane. Specimen diameter at the root of the

notch was 0.150 inch, with a root radius of 0.00095. The stress concentration

factor (Kt) was approximately 8.5 (Ref. 43).

62
 Z

0
Z

LO 0 oo
C

[-,
A
Z •,--I _.t
tll
Z

'-r
._z
,l-J

z
[.I.]

'I_ If] 0 0xl i I I

DO 0
e,.,
o _
.-1 ,-.]
°r-I
•_ oO LTJ
[-" Z
0 Z

,.--, >..,
C_

_' ""_ '_ _ 0 _ _"

,I.-I

Z
U
[-., 0
0
Z 0

e-.

0 0
.el
• ,'_ 0

E _ "_

z0
_ _ "_._
• U

0 0 I_ _ 0 0

_-I I ! ._ .1¢

63
 '°°
I

70
I I, I I I .
_e
80-

o 70_ _ AIR -'_

60q
.J

50 • , i I , I ,
0.002 0.02. 0.2 2.0

CROSSHEAD SPEED, IN./MIN.

Figure 18. Tensile strength and elongation

versus crosshead speed for notched
(k% 4.0) specimens of 30_L stainless
steel tested in air and i0000 psi
hydrogen at various strain rates.
(Gage length is assumed to be 0.0985 in.)

6h
 The unnotched specimens of the AISI 504L stainless steel, AISI 305 stain-

less steel, and the OFHC copper had a gage section diameter of 0.150. The

majority of the tests were conducted using a strain rate of 0.002 min -1 to the

-1
yield point and 0.04 min from yield to fracture. The notched specimens were

-1
load paced at a rate corresponding to an unnotched strain rate of 0.0007 min

The notched specimens of AISI 504L and 430F stainless steel and 6061-T6 aluminum

alloy were tested with a strain rate corresponding to an unnotched strain rate
-I
of 0.005 min and the notched A-286 stainless steel was tested with a loading

rate corresponding to an unnotched strain rate of 0.0014 min -I The heat treat-

ments of the alloys tested are given in Table 7. Purity of the hydrogen, after

pressurization, was <0.5 ppm D2,<5 ppm total hydrocarbons, <0.5 ppm CO + C02_

<.0.5 ppm H20, and 2.6 ppm N 2. The results of these tests are shown in Table 8.

The high-strength steels, AISI 1042 (quenched and tempered), AISi 4140,

Fe-9Ni-4Co-0.20C, and 18 nickel (250) maraging steel, were all extremely em-

brittled by gaseous hydrogen as indicated by large reductions of notch strength

and unnotched and notched ducti!ity. Unnotched ductility of these specimens was

reduced over 75 percent, and the notch ductility was essentially zero for all of

the specimens tested in I0 000-psig hydrogen. Reduction of notch strength ranged

from 60 percent for AISI 4140 to 88 percent for 18 nickel (250) maraging steel.

The 18 nickel (250) maraging steel specimens were the most embrittled of ali the

materials tested. This steel had a 52-percent reduction in unntoched strength,

as well as an 88-percent reduction in notched strength.

The 400-series stainless steels (AISI 410, 440C, and 450F) were among the

most hydrogen-environment-embrittled steels tested. The AISI 410 and 440C are

in the extremely embrittled category, while the 450F is in the severely

65
 r_
,-.1

LL1

U.1

[--,

o
[-,

_ 333 3 3 _ _ _
'__,

_ _, "_ _ '_ '_ '_ _ '_ ,_ .-_ . .._° o ._

3_

66
 TABLE 8

RESULTS OF TESTS ON VARIOUS ALLOYS AT ROCKETDYNE (REF. 7 )

8nvt ronment Unnotched Speci_,ens NOtched Specimens

Stren8th Red. of Strength

Red. of Ratio Ratio
Press., YS, UTS, Elong., Area UTS, Degree
Area, H2/H e _ ksi H2/H e of Embritt.
_terial Type psig ksilksi %

lg% Hi (250) He 10 O00 240 250 8,2 55.0 -- 2.5 423 -- Extreme

HeragLng Steel H2 -°- ]71 O.2 2.5 O,b8 0.0 50 0,12

AISI Type 410 He 192,211 15.O 60.0 -- 2.2 3_5 --

Stainless Steel H2 °--106 1.5 12.0 0.79 0.6 82 0.22

.... 0.6 236 --

AISI 1042 Quenched He i 165 187 ' [[ .... 0.6 53 0.22
& Tempered H2 ---
I

17-7PH He 150 164 17.0 45.0 -° 0.6 302 --

Stainless Steel H2 --- 151 1.7 2.5 0.92 0.4 70 0.23

H-I[ He 244 299 8.8 30 -. 0.0 252 --

H2 --- 171 0.0 0.0 0.57 0o0 63 0.25

Fe-Ni-4CO-0.2C He 107 199 15.0 67.0 -- 6.3 367 --

H2 --- 175 O.S 150 086 02 ......

Rene 41 He 163 ]96 21.0 29.0 -- 3.4 280 --

H2 --- 165 4.3 11.0 O.84 0.2 77 0.27

Elec_roformed Ni He 7 000 .......... .... 31.0 420 --

.... 1.0 37 O.31
H2 ..........

AISI 4140 He 10 000 179 186 14.O 48.0 -- 2.8 313 --

H2 --o 178 2.b g.8 0.96 0.9 125 0.40 i

Inconei 718 He 102 2O7 17.0 26.0 -- 4.7 274 --

0.8 0.93 0.2 126 0.46
H2 --- 193 1.5

AISI Type 440C He 236 293 -- 3.2 -- O.2 149 --

0,0 0.40 O.0 74 0.50 I
Stainless Steel H2 o-- 119 --

Ti-6HJ-4V He 157 164 13.0 47.0 -- 2.2 228 =- Severe

(STA) H2 152 171 15.0 SO.O -- 0.9 133 0.58 I

A[5I Type 43OF He 72 8O 22.0 64.0 -- 1.9 152 -- i

Stainless Steel H2 --- 78 14.O 37.0 °- 0.6 104 0.68 !

Nickel 27O He i
°-- 88 56.O 89.0 -- 23.6 77 --
67.0 -- 6.9 54 0.70
H2 I --- 50 52.O
I

AS'O4 A-SIS Gr. 7O He 8$ 65 42.0 67.O -- 0.1 106 --

H2 43 64 29.O 35.O -- 3.0 77 0.73

HY-IOO He 97 113 20.0 76.0 -- 7.3 224 --

63.0 -- 3.8 164 0.73
H2 --- 115 18.0
i
ASTN A-372-1V He I 02 118 2O.O 53.0 -- 2.0 200 --
18.0 -- 2.0 140 0.74 i
H2 --- 116 9.6

Ti-6AI-4V He 132 156 15.0 48.0 -- 2.2 243 --

40.0 -- 1.0 183 0.79
(&_ealeci) H2 --- 159 15.0

HIS! 1042 He 50 90 29.O 59.0 -- 8.5 1S3 -- i

Normlized 142 --- 89 22,O 27.0 -- 2.8 145 0.75

ASTM A-SI7-F (T-I) He 409 148 10.0 65.0 -- 7.4 226 -°

H2 111 121 18.0 63.0 -- 2.1 173 0.77

ASh4 6533-0 He --- 119 19.O 66.0 -- 15.0 227 --

33.0 -- 2.8 176 0.78
H2 j --- 116 27.O

HY-gO He I 02 98 25.0 70.0 -- 8.6 190 --

60.0 -- 3.9 153 0.00
H2 8$ 99 2O.O

AISI 4020 He 41 63 40.O 68,0 -- 14.0 lOS --

45.0 -- 0.2 83 0.79
H2 4O 62 32.O

Ti-SHI-Z.SSn ELI He 106 i It3 20.0 45.0 -- 3.t 201 -- /

H2 ---114 18,0 39,0 -- 1.8 162 0.8[ [

Hrsco Iron He 54 56 10.0 83.0 -- 6,4 124 --

50.O -- 1.7 105 0.86

142 --- 57 15,0

A]S4 Type 304L He 24 77 06.O 78.0 -- 21,0 102 -- Slight

71.0 -- 11.0 89 0.87
Stainless Steel H2 --- 76 79.0

6IS4 Type 3O5 He 51 90 63,0 78.0 -- 19,0 165 --

75.0 0.97 17.0 147 0.89
Stainless Steel H2 --- 07 6_.O

Be-Co Alloy 25 He 79 04 22.0 72.0 -- 12.0 195 --

71.0 -- 13.O 481 0.93
H2 --- 93 22.O

Titaniu= He 53 63 32.O 61.0 -- i0.0 126 --

61.0 -- 7,3 420 0.95
(Couercially Pure) H 2 i ....... 31.O

AISI Type 310 He --- 77 56.O 64,0 - 20.0 116 - Negligible

62.0 -- 18,0 108 0.93
St=inless Steel H2 --- 70 560

A-286 He 123 158 26.0 44.0 -- 5.6 233 --

43.0 -- 6.2 227 0.97
StaLnless Steel R2 --- 162 29.O

7075-T73 He $4 66 15.0 37.0 -- 3,8 146 --

35.0 -- 2.3 114 0.90
Aluminum H2 --- 65 12.0

6061-T6 He 33 39 19.0 61.0 -- 9.5 72 --

66.0 -- 11.O 78 --
A4umiu H 2 ; --- 4O 19.0
11OO-0 He --- 16 42,O 93.0 -- 20,0 ]8 --

93.0 -- 21.0 25 --
Aluminu= H2 --- 16 39.0
94.0 -- 20.0 87 --
OFHC Copper He 39 42 2O.O
94.0 -- 24.0 86
H2 --- 41 20.0

AISI Type 316 He 64 94 59.0 72.0 -- 18.0 161 --

75.0 -- 19.0 161 --
Stitnless Steel H2 --- 99 56.O

67
 embrittled category. Among these steels, it appears that, the stronger the

steel, the greater the hydrogen-environment embrittlement, although the results

for AISI 440C are somewhat ambiguous because of its extreme notch sensitivity.

The ultimate tensile strength of unnotched 440C was 295 ksi, while the notch

strength in 10 000-psig helium was 149 ksi. In 10 O00-psig hydrogen, the notch

strength was 74 ksi, indicating a 50-percent reduction of notch strength.

There was also a reduction of strength of the unnotched specimens of the

400-series stainless steels in high-pressure hydrogen. Strength reductions were

2.5 percent for AISI 430F. Ultimate tensile strength reduction of unnotched

specimens of AISI 440C was considerably greater (60 percent) than observed for

any other material.

The low-alloy steels (Armco iron, AISI 1042 normalized, HY-100, and AS_4

A-372) were in the category of severe embrittlement. Reduction of ductility of .

the unnotched specimen was greatest for ASTM A-372 class IV, and least for HY-100.

Reduction of notch strength was about 27 percent for ItY-100 and ASTM AS33-B and

14 percent for Armco iron. Except for the AS_4 A-372, the notch ductility of

these steels was drastically reduced by the 10 000-psig hydrogen environment.

The AISI 504L and 305 stainless steels are in the slightly embrittled cate-

gory. The reduction of area of unnotched AISI 304L specimens was reduced approx-

imately 10 percent by the 10 000-psig environment. The notched specimens had a

strength reduction between 10 and 17 percent and a ductility reduction of 50 per-

cent due to the hydrogen environment. All of the specimens were reported to be

magnetic after testing. Benson, et al. (Ref. 46) reported a similar effect with

a 7.3-percent reduction of notched strength in AISI 304L stainless steel in 10 000-

psig hydrogen. Specimens of AISI 305 stainless steel were reported to be very

68
slightly magnetic after testing; no surface cracking was observed, and

there was only a slight indication of reduced ductility in the unnotched and

notched specimens (Ref. 7). Only a small reduction (about 4 percent) of

unnotched strength and ll-percent reduction of notch strength was observed.

Thus, AISI 305 stainless steel was somewhat less embrittled by 10 000-psig hydro-

gen than was AISI 304L stainless steel, which showed a reduction of ductility as

well as strength in high-pressure hydrogen. There was no indication of a reduc-

tion in strength or ductility in the unnotched and notched AISI 316 specimens

because of exposure to the high-pressure hydrogen environment.

The precipitation-hardened austenitic stainless steel (A-286 stainless

steel) showed negligible embrittlement in the i0 O00-psig hydrogen environment.

The tensile strength of the notched specimen was approximately 2-1/2 percent lower

in hydrogen than it was in 10 000-psig helium; there was no decrease in ductility

of the unnotched and notched A-286 alloy specimens.

Both nickel and nickel-base alloys were considerably embrittled by i0 O00-

psig hydrogen. Nickel 270 (99.4 percent Ni + Co) is severely embrittled in

10 000-psig hydrogen. The reduction of notch strength is approximately 30 per-

cent when compared to results obtained in 10 000-psig helium. The reduction in

unnotched tensile strength for Rene' 41 was 16 percent and, for notched specimens,

the reduction of tensile strength was 73 percent. These reductions are based on

tests conducted in 10 000-psig helium. As noted earlier, K-monel is also subject

to hydrogen-environment embrittlement (Ref. 44).

69
 In more intensive investigations, three heats of Inconel 718 were tested

for hydrogen-environment embrittlement (Ref. 47). Five different heat treatments

(Table 9) were used with these three heats (Table 10). The results of the ten-

sile tests on unnotched and notched specimens of Inconel 718 are given in Table 11.

The data for the room-temperature tests on notched specimens indicate that the

reduction of notch strength by hydrogen is a function of the heat-heat treatment

combination of the Inconel 718 tested. For example, specimens of heat C with heat

treatment 5 showed a 49-percent reduction of notch strength in 500-psig hydrogen,

while specimens of heat B with heat treatment 3 had only a IS-percent reduction

of notch strength in 2000-psig hydrogen. From these results, it is obvious that

the heat and/or heat treatment of Inconel 718 has a considerable effect on the

degree of hydrogen-environment embrittlement. Walter and Chandler (Ref. 47) pro-

posed that the lower embrittlement of heat B with heat treatment 3 was due to the

high annealing and aging temperatures used. VanWanderham (Ref. 48) also con-

cluded that this type of heat treatment yields Inconel 718 less susceptible to

embrittlement by high-pressure hydrogen. However, the results of recent tests

on heat C showed a high degree of embrittlement of specimens with heat treatment

5, which resembles heat treatment 3, and these specimens were more embrittled

than specimens with heat treatment 4, with its lower annealing and aging temper-

atures. These data indicate that, for some as yet unknown reason, heat B had a

relatively low susceptibility to hydrogen environment. It should be noted that

the three heats .tested were within composition specification limits for Inconel

718. There does not appear to be any correlation between the properties in

helium for the different heats and heat treatments and the degree of hydrogen-

environment embrittlement, although heat B did have the highest ductility and

notch strength/unnotched strength ratio and lowest hydrogen-environment embrittle-

ment. Additional work is in progress, see the Appendix.

7O
 TABLE 9

HEAT-TREAT CONDITION OF INCONEL 718 TEST _IATERIALS (Ref. 47)

Heat Treatment

Solution Annealing First Aging Second Aging

(In Argon) Temp. Temp.
Heat
Treatment °F minutes °F hours °F hours

1750, air cool 60 1325 8 1150 10

1850, air cool 60 1360 10 1175 10

1900, air cool 20 1400 11 1200 9

1800, air cool 60 1325 8 1150 12

1950, air cool 60 1350 9 1200 11

71
 I 0 0 0
0 0 0 0

0 0 0

0 0 0 0

c; o d o

m o ,--4

I • •

d d o d
0_ c_l 0

• !

C_
*,-4

o _ _
0
0
U ! e

0 0

0
cx) .,-4 _ _ 0 0
•.-4 -IJ
k ,<
.-I o o _ do
.-I U

E-. Z _ _ t._L_
0 0 .,-I
U "_
Z o d ,J o,-i
>
0 0 O0
o 0 _ _ ,.-.I cO _0 L_'_
0
z
0
• °r'l

_ 0
t_
o
0 0
U

U _ O_ _ O0

,-.1 o_
_kt_

,-,I 0 _ 0 0
0"_

U_
t.-I

tl "_0

72
 0 • 0

t_ C-,
Ol_ II_ lllO_ _ li_ i_

_ 0 II 0 II 0 llI-.'-'_O00 II 0 I'_
l-J .r,.I

D.,

o
o
•_ 0 0

0 • _ Q\o ° • • •
_o _ ___
0 Z _

e
_4 o_

Z
<
._"t
Z 0

m _
r'-,

0
,--1
0 ,M
_0_ _ 0000 O0
[.-,

_ 0r-I
e_ r./3
I

0 00_ 0000000 000 O0

0 0 O0 000000 O0 O0
Z 0
0
I-I 0 _ _ _

t_ ,r-I

U-I
Z
o

_ E_ 0
0

°_
% [.-. [.-,

73
 The notched, commercially pure titanium specimens were slightly embrittled.

Ti-6A1-4V and Ti-SA1-2.SSn ELI notched specimens were severely embrittled in

10 000-psig hydrogen. The unnotched, commercially pure titanium and Ti-6A1-4V

specimens were not embrittled by 10 000-psig hydrogen. Numerous surface cracks

were present on the unnotched, commercially pure titanium specimens tested in

hydrogen, but the cracks were not sufficiently deep to affect embrittlement, as

will be discussed further in Chapter 8, Metallography. The large (45 percent)

decrease of notch strength of Ti-6A1-4V (STA) in 10 000-psig hydrogen without a

reduction of unnotched properties is unique for the materials tested.

The results of the tests on notched Ti-6A1-4V (STA) specimens should be

noted because the mechanical properties show a very large dependency on the

method of purging the test cells. Tests in 10 000-psig hydrogen preceded only

by a pressurization/depressurization treatment did not cause embrittlement.

The large reduction of notch strength occurred only when the test cells were

previously evacuated. Thus, embrittlement of Ti-6AI-4V (STA) is evidently quite

sensitive to hydrogen purity.

Unalloyed aluminum (1100-0 aluminum) was tested in S000-psig hydrogen and

there was no indication of embrittlement; 6061-T6 aluminum alloy was not

embrittled in 10 000-psig hydrogen. _le ultimate tensile strengths of both

unnotched and notched 7075-T73 aluminum alloy were approximately the same in

hydrogen and helium environments. However, the average ductility of the un-

notched and notched specimens was slightly lower in 10 000-psig hydrogen than in

10 000-psig helium. In these tests, the test vessels were vacuum purged prior

to pressurization with 10 000-psig hydrogen for the 1100-0 aluminum tests, but not

for the other two aluminum alloys.

74
 Results of tests on Be-Cu alloy 25 indicated no embrittlement of the un-

notched specimens by high-pressure hydrogen. There was, however, a 7-percent

reduction of notch strength in 10 O00-psig helium containing 44-ppm hydrogen.

Notch ductility was not affected by high-pressure hydrogen.

75
 Effect of Hydrogen Pressure (Room Temperature)

The effect of pressure on the degree of hydrogen embrittlement has been

investigated for only a few alloys. Results of tests performed at various

hydrogen pressures are shown in Table 12 (Ref. _ 44, 46, and 49). The heat-

treated condition of these alloys is shown in Table 13.

_lere is no evidence of any reduction of strength or ductility of the un-

notched AISI 310 stainless steel due to hydrogen environment. However,

there appears to be a small decrease of strength and ductility of the notched

specimens, and the degree of embrittlement decreases with decreasing pressure.

Data at hydrogen pressures less than 10 000 psi are possibly within experimental

scatter. Benson, et al. (Ref. 46) report similar results for notched AISI 310

stainless steel, the degree of embrittlement at 10 000-psig hydrogen being less

than 1 percent. The notch ductility at lower hydrogen pressures is somewhat

lower on an average than the ductility in 10 000-psig helium, but this difference

may have been the result of the increase of ductility in the 10 000-psighelium

due to pressure effects.

A slight, reduction of unnotched strength for ASTH A533-B specimens occurs

when they are tested in 10 000-psig hydrogen. No strength reduction was observed

for specimens tested in hydrogen at lower pressures. A substantial loss of duc-

tility was noted in the unnotched and notched ASTM A533-B specimens tested in

76
 TABLE 12

EFFECT OF HYDROGEN PRESSURE ON VARIOUS ALLOYS

Notched Specimens

Environment Unnotched Specimens Strengtf

Material

AISI Type
Type

He
Press.,

psig

0
YS,

33.5
ksJ UTS,

89
ks: Elong.,

55.0
_
Red.

Area,

66.5
of

%
Red.

Area,

......
of

% UTS, ks!
Ratio

H2/He

8
71
310 Stain- 100
less Steel 1 000 35.0 90 54.0 67.5
I0 000 -- 77 56.0 64.0 19.8 116 --

0 35.0 87 53.5 65.5

1t 2 18.7 124 0.98
100 36.3 87 52.3 66.0 18.0 123 0.98
1 000 34.0 89 55.0 67.0 16.0 123 0.98
10 000 -- 78 55.0 63.3 18.0 108 0.93

He 10 000 -- 110 --
41
10 000
H2 -- 109.5 1.0

ASTH A533-E He 0 118.0 118 28.0 70.0 12.0 216 -- 8 7

100

1 000 119.5 116 23.5 71.5 14.0 215 --

I
10 000 -- 120 19.5 67.5 11.0 227 --
I
0 115.0 118 21.0 68.5 7.1 235 0.99
H2
100 118.0 120 22.3 67.3 7.4 212 0.98
1 000 113.3 117 20.0 57.7 4.2 218 0.92
10 000 36.0 116 16.7 32.7 2.9 176 0.78 t
Ti-6AI-4V Air 0 169.0 179 11.5 48.0 1.7 237 -- 8 43'
(STA) I
He 1 000 166.0 178 11.0 45.5 3.4 249 --
2 000 166.5 177 13.0 46.0 3.2 251 --

1 000 167.6 178 11.5 50.0 0.6 154.5 0.64

H2
2 000 164.3 175 11.3 • I
49.0 0.6 150.0 0.60
i
Ti-SA1- lie 1 400 3.8 186 -- 8
2.SSn ELI 10 000 106.0 113 20.0 45.0 3.1 201 --

1 400
H2 3.4 168 0.90
l0 000 -- 114 18.0 39.0 It I
1.8 162 0.81

Inconel Air 0 187.5 217 16.0 31.0 1.S 284 -- 8 421

718"
He 1 000 194.0 216 16.5 52.0 1.4 282 --

]t 2
1 000 186.3 215 17.3 i 28.0 1.3 169 0.58 t, I
i
Inconel Air 0 190.5 206 25.0 I 39.0 3.1 333 -- 8 421
718"*
lte 100 199.5 208 24.5 I I I
31.5 3.1 333 -- | i
1 000 -- 333 --
1 500
I I
2 000 187.5 208 24.0 I 37.5 2.8 333 --

100 199.5 208 24.5 I 31.5

H2 3.1 332 1.00
1 000 3.0 302 0.9J i i
1S00 2.6 299 0.90
2 000 -- 206 37.5 1.7 282 0.85 I
i
4140 Low _ir I 0 -- 135 -- ...... 4 39i
Strength%

1
10 000 -- 241 --
_2

6 000
_2 -- 207 --
10 000 -- 204 0.83

4140 High 0 -- 228 -- ...... 4 39i

Strength%
N 2
10 000 208.0 206 37.5 -- 362 --
I I
2 000
H 2 -- 135 --

10
6 000
000
--

j
121

89 0.22
--

i 11
"Heat Treatment 2 IT-q_ _ XIII)
**Heat Treatment 3' .Ta . XIII)
%Held in H2 environment 24 hours before testing

77
 Cb
y
,v

C O',
D.- O

U ',J
c_

;'40

O 4._ ,-.._
.C a_ 0 0
0 •-'10 U_
OO_o _"
a)
N _ 0

o 0
_-_
0 _,._ 0 00_ 0

0 _-_ f_ 0 _0 0
O D O _ _ _,
o_ 0 0 O0 0
,.O cO O
,.-I •-_ _ ..,

¢.J 4-_ -_ _ ,-_ _ _ 0 _,-_

0
"O
C_' 0 ._O _ c_ _ "-0
OO r'- O
c_ o_ ,,nO 0

< C
0 •,_ O _ ",_
c_ .<

u_
u_

O_ o0
U_ 0 o4

0 0
O_ c_

<
U_
c_

CqO

_U

>,C
.o O
tD _0 0
.C
>

.,_
_-_ .,_ _.._
0
U ,.._ ,--_
I 0
[-,,,I u_ 0
I 0 0
0

C 1"_ .r-I
< <

('4

C C
O _D _D
,--4 ,--_ 4_
r OO'_ CO cO _.0

I'.- _D r-- O_

o_
<
C
0 _'-_
0 _D .c

<tn

78
 hydrogen at i000- and 10 000-psig pressures (Fig. 19) (Ref. 7). A decrease in

the notched strength was also observed. The decrease in ductility and strength

is a linear function of the square root of hydrogen pressure and extrapolates to

zero effects at zero hydrogen pressure. It should be noted that surface cracking

was observed on unnotched ASTM A533-Bspecimens tested at all hydrogen pressures.

Surface cracking effects will be discussed in Chapter 8, Metallography.

The relationship between tensile properties and the square root of the

hydrogen pressure is consistent with a hydrogen adsorption rate-controlled embrit-

tlement process, as has been proposed by Williams and Nelson (Ref. 8). This point

will be discussed later in Chapter I0, Technical Discussion.

The influence of pressure was also determined to a limited extent on Inconel

718 (Ref. 47 and 49). The results of a series of tests for Inconel 718 in 100-

and 2000-psig hydrogen appear in Table 12.

These results indicate a small reduction of ductility for heat treatment 2,

as measured by the reduction of area in the unnotched specimens during exposure

to 1000-psig hydrogen, but no decrease is observed for heat treatment 3 at any

pressure. It should be noted that this is a considerable difference from the

almost zero ductility found for heat treatment 2 in tests conducted in 10 000-psig

hydrogen (Table 8). Walter and Chandler (Ref. 7 ) observed that the reduction

of ductility for the individual specimens in 1000-psig hydrogen was directly pro-

portional to the size of the surface crack causing fracture.

The notch strength of specimens from heat treatment 2 was reduced ap-

proximately 40 percent in 1000-psig hydrogen, compared to approximately 55

percent in 10 000-psig hydrogen when compared to simiiar helium pressures.

There was a small reduction in ductility Of notched specimens from expo-

sure to 1000-psig hydrogen, while the notched specimens tested in i0 O00-psig

79
 s::l=:,

N
t
i,-,=1 ¢v-_O

n o

O..

o _

1.1.1

O- o

/ O

>-
-I-
,--4

4-)

_o
a

WflI'I3H WO_:I NOI13rl(13_! IN.-lO_13d

hydrogen had almost zero ductility. For heat treatment 3, the notch strength was

reduced only 15 percent at 2000-psig hydrogen. The reduction of notch strength

for heat treatment 3 was a function of the square root of hydrogen pressure (as

shown in Fig. 20) up to 2000-psig hydrogen pressure.

There was no reduction of unnotched strength or ductility at room tempera-

ture for Ti-6A1-4V (STA) in i000- and 2000-psig hydrogen (Table 12). The reduc-

tion of notch strength was about 32 percent in 1000-psig hydrogen and about 45

percent at 2000-psig hydrogen. Accompanying the reduction of notch strength was a

reduction of notch ductility. For Ti-SAI-2.5Sn ELI the reductions of notch

strength in 1400-psig and 10 000-psig hydrogen were l0 and 19 percent, respectively.

Influence of Notch Severity

The effect of stress concentration factor on hydrogen-environment embrittle-

ment has been investigated at Rensselaer (Ref. 50 and 51] and at Rocketdyne

(Ref. 7). _le work at Rensselaer is summarized in Table 14. Two steels were

used, AISI 4140 and AISI 304L stainless steel. The 4140 was heat treated to two

strength levels. Two specimen designs were used, one with a stress concentration

factor of 2.0 and one with 4.0. The low-strength 4140 had approximately an 11-

percent loss of strength for the K t = 2.0 specimen when tested in i0 000 psig

hydrogen, with or without a hold time, when compared to a control specimen tested

in air. The K t = 4.0 specimen, soaked in i0 000-psig hydrogen for 20 hours be-

fore testing, showed an 18-percent loss of strength when compared to the control

specimen. The K t = 4.0 specimen without a hold time showed an anomalous 16-percent

gain in strength.

81
 eHI_H

) o

O-_

a..
,-I b0
ELi
O

tH
o
o

o,-I

m N

l l J J J O0

IN33_13d 'WIII13H NI IVHI WO_l-I

N3OO_I0,1_INI HION3_I.I.SH310N -IO NOI.L3N(]3_I

82
 TABLE 14

EFFECT OF NOTCH SEVERITY ON HYDROGEN-ENVIRONMENT EMBRITTLEMENT

Test Results

Strength "Ductility

Test Condition
Stress
Ratio Ave. % Ratio
Exposure Concentration Press., Strain I/TS, H2/H e Red. of I{2/He
Time psig Rate/Min gel.
Material Factor, Kt Environ. ksi (Air) Area (Air)

4140 Low 0 2,0 Air, RT 0 0.02 287 ...... 80

strength 20 hr 254 0.89 ....
3 rain H2, RT 10 000 crosshead
speed 252 0.88 ....

0 4.0 Air, RT 0 0.02 261 ......

20 hr 215 0.82 ....
3 min H2, RT 10 000 speed 304 1.16 ....
0
RT-321°F
Air, 000 0 361 ......
l H2, 10 crosshead
2 hr H2, -321°F I0 000 377 1.04 ....

4140 High 0 2.0 Air, RT 0 0,02 377 ......

Strength 20 hr crosshead 177 0.45 ....

S min H2, RT I0 000 speed 240 0.64 ....
I H2, RT I0 000

0 4.0 Air, RT 0 0.02 375 ......

20 hr crosshead 122 0.32 ....
0 A_r, -321°F : 0 265 ......
speed
H2, RT I0 000
20 hr H2, -321°F 10 000 247 0.93 ....

AISI Type 304 L 24 hr 2.0 Air 0 107.0 -- 60.0 --

Stainless Steel 24 hr N2 I0 000 108.4 ......
24 hr 98.0 0.92 ....
H2 2 500
24 hr 103.0 0.96 52.8 0.88
H2 S 000
24 hr H2 7 500 96.7 0.90 -- --
24 hr H2 10 000 98.6 0.92 54.4 0.92
3 min H2 10 000 102.2 0.96 ....

24 hr 4.0 Air 0 117.0 -- 60.0 --

24 hr 118.9 ......
I N2 10 000
24 hr 110.2 0.94 ....
H2 2 500
24 hr 99.5 0.8b 44.4 0.74
i H
H 22 S
24 hr 7 000
000 95.7 0.82 ....
24 hr 94.4 0.81 41.4 0.69
3 min i H 22 i0 000 100.8 0.86 ....
H 10 000

24 hr 1.0 Air 0 102.0 -- 60.0 --

89.0 0.87 45.7 0.76
!H
IH 2
2 i0S 000
000 85.0 0.83 44.3 0.74

AISI Type 310 0 8.3 He i0 O00 0.0007 116.0 -- 20.0 --

Stainless Steel 8.5 H2 i0 000 I 108.0 0.93 18.0 0.90
5.9 Air 0 125.0 -- 27.0 --
5.9 He i0 000 lll. S -- 29.0 --
s.9 ia2 lO ooo 112.0 1.00 29.0 l. O0
4.0 !Air
i
0 i 112.0 -- 37.0 --
4.0 IHe 10 000 iii.0 -- 32.0 --
]I 4.0 I H2 I0 000 111.0 1.00 31.7 1.00
I

ASTH AS33-B 0 7.8 !He 10 000 0.0007 227.0 -- 11,0 --

7.8 H2 I0 000 176.0 0.97 2.8 0,26

8.9 Air 0 221.0 -- 15.0 --
5.9 ;He 10 000 211.0 -- 16.0 --

5.9 !H_ 10 000 168.0 0.80 4.9 0.31

T

4.0 iA1r 0 221.0 -- 19.0

Ir Ir 4,0 He lO 000 207.0 -- 19.0 --

4.0 ll2 10 000 177.0 0.86 4.9 0.26

ASTM A-517 (T-l) 0 8.3 Air 0 0.0007 232,0 -- 7.0 --

8.7 He 10 000 221.5 5.7
8.3 H2 10 O00 173.o ;_78 31 ;_s_
5.8 Air 0 243.0 -- II.0 --
5.8 He 10 000 229.5 -- 11.5 --
172.0 0.45 2.0 0.]4
5.8 H2 10 000
3.75 Air 0 243.0 -- 13.0 --
3.75 He 10 000 226.5 -- 12.0 --

3.75 H2 10 000 181.0 0.20 2.8 0.23

1.0 Air 0 128.0 -- 65.0 --
Ir 1.0 H2 10 000 1 121.0 0.96 63.0 --

83
 Embrittlement for the high-strength 4140 was more spectacular. The Kt=2.0

specimen showed a 52-percent loss of strength, and the Kt=4.0 specimen showed a

68-percent loss of strength (as compared to air tests) when tested in i0 O00-psig

hydrogen after an approximately 20-hour soak.

The embrittlement of the AISI 304L stainless steel also showed a dependency

on the degree of stress concentration. For the Kt = 2.0 notch, the decrease in

ductility was 12.0 percent at 5000 psig, and 8.0 percent at I0 000 psig. For the

Kt = 4.0 notch, the decrease in ductility was 26.0 percent at 5000 psig, and

31.0 percent at I0 O00-psig hydrogen. These results show that for these two

materials, there was a small dependence of ductility on hydrogen pressure, but

the stress concentration factor had a pronounced effect.

Walter and Chandler (Ref. 7 ) have investigated the influence of notch geom-

etry for ASTI4 ASZ3-B,ASTM A-517, and AISI 310 stainless-steel specimens in 10 000-

psig hydrogen. The notch geometries investigated had stress concentration factors

(Kt) of approximately 3.9, 5.9, and 8.4. The results are tabulated in Table 14.

For AISI 310 stainless steel, no embrittlement was observed except for specimens

with a K t= 8.4 for which the loss of notch strength was less than i0 percent.

The results for ASTM A533-B are plotted in Fig. 21, and for ASTM A-517 (T-l)

in Fig. 22. The degree of embrittlement in 10 000-psig hydrogen increased for

low values of Kt, reached a maximum at Kt= 8.0 for ASTM A533-B and at K t= 6.0 for

ASTM A-517 (T-I).

Also plotted in Fig. 21 and 22 is tile reduction of strength for fatigue

precracked specimens. The Kt value for a notch containing a fatigue-induced

crack was arbitrarily assumed by Walter and Chandler to be between 19 and 20.

84
 22

_D
20
Z

18 0

16
O
Z
14
O

Z
O
12

10

8i
b--
Z
I,Ll
6
LAJ
a-
4

0 I I 1 i I I I _ _,_
1 2 3 4 5 6 7 8 PRE-CRACKED
STRESS CONCENTRATIONFACTOR
Pt gu_e 21. Reduction of Strength as a Fmlction of Notch Severity
_or AS_ A533-B (re_. 7)

85
 25

24

22

"1- 0
I-- 2O
Z
m,m
I-- 18
(J)

-r
15
I---
0
Z
14
mi
0

Z 12
0
m

I--
10
I.l.I
8
Z
m,m
5
O-
4

I 1 I I
O(
1 2 3 4 5 6 l 8 PRE-CRACKED
STRESS CONCENTRATIONFACTOR

P£8_re 22. Reduction in Strength of AS_ A-517 as a function of

notch severity.

86
 The precracked specimens of both materials are not as embrittled by I0 000 psig

hydrogen as the Kt = 6.0 and 8.0 geometries, indicating that embrittlement does

not increase indefinitely with decreasing notch radius.

Effect of llold Time (Room Temperature)

Early experimental work on hydrogen-environment embrittlement employed an

exposure time (unstressed) in the gaseous hydrogen prior to conducting tensile

tests in the environment (Ref. 50). Later work was done in which a stress was

applied during the hold time in hydrogen (Ref. 7 ).

All of the hold-time data (Table 15) with no applied stress showed that the

amount of hydrogen-environment embrittlement was not a function of the hold time

in the environment. Some of the data indicate that an inverse relation between

hold time and the degree of embrittlement exists, i.e., the longer the hold time,

the less the degree of embrittlement (Ref. 7). The inverse relationship was at-

tributed to a buildup of oxygen contamination on the surface of the sample during

the hold period (Ref. 7). The effect is evident from the data in Table 15 for the

AISI 310 stainless steel, where the degree of embrittlement with no hold time is

7 percent and is approximately 0 as the hold time increases. These tests were

run in test cells which had been pressurized/depressurized with hydrogen but had

not been evacuated. For titanium, data are available (Table 15) for pressurized/

depressurized test cells and for test cells which were evacuated prior to testing.

• In both cases, the degree of embrittlement is much greater with no hold time than

with prolonged exposure.

87
 •_ 0 0 0 0

0 i._
I I'_ I
i.l,e I • I *
'++_+ 0_00

_0_

00000

_DO_D

P-4

elm
r..4 _--_ r-4

IIIII IIIII _I +II

IIIII IIIII _I _II
IIIII IIIII _I MII

0 •
[...,
I
>
"m" "l=I
I 0 I
_'I P'-I

I
.,-I

88
 From the data in Table 16, no effect on the degree of hydrogen-environment

embrittlement of various steels and K-monel could be attributed to holding speci-

mens under stress in 10 000-psig hydrogen for periods up to 100 days before ten-

sile testing. The yield strength of unnotched AISI 1020 specimens increased with

hold time when held at a stress below the yield strength; h0wever, the specimens

had been prestrained and the yield strength increase was attributed to strain

aging. Yield strengths o£ unnotched HY-80 and ASTM A-51S specimens, which had not

been prestrained, were not affected by holding at a stress below the yield

strengths. When unnotched AISI 1020, ASTM A-51S, and HY-80 specimens were tensile

tested after holding at a stress above the yield strength, a new yield point was

found that increased with hold time. This increase in yield strength was attrib-

uted to strain or stress aging.

The degree of hydrogen-environment embrittlement, as measured by reduction

o£ notch strength in 10 000-psig hydrogen compared to that in 10 000-psig helium,

decreased with increasing exposure duration for the prestrained, notched AISI

1020 specimens. It was suggested (Ref. 1 ) that this embrittlement decrease

related to an increase of yield strength/ultimate strength ratio, resulting

from strain aging during the hold period. Evacuation of the test vessel.s prior

to pressurization/depressurization purging led to increased embrittlement from

the 10 0D0-psig hydrogen environment in tests involving no hold period. The

increased embrittlement was attributed to a reduction in the concentration of

oxygen adsorbed on the specimen surface. After holding for 1 day or longer,

the embrittlement in 10 000-psig hydrogen was the same whether or not tile test

vessels had been evacuated.

89
 TABLE16
EFFECT
OF EXPOSURE
TIME AND HOLD STRESS IN HYDROGEN

ON THE TENSILE PROPERTIES OF VARIOUS METALS

Test E_vironment Unnotcbed Notched

Haterial

AIS[ 1020
(Prestrained

ASTN A515

H2

HY-80

H-ll Tool
Steel

4140 Low

Strength

4140 High
Strength

AISI Type
304L
Stainless
Steel

C-102S

K-Hone1

'recip.
ardened

g-_onel
Annealed

4140 *

_AE 1018"

*Tests performed in air after exposure to 6000-psig hydrogen

90
 The strength of notched Ti-6A1-4V specimenswas considerably lower in
10 000-psig hydrogen as comparedto 10 000-psig helium. In almost all cases,
both annealed and STATi-6A1-4V notched specimensfailed in hydrogen shortly
aftera hold stress was established, rather than while the specimenswere being
loaded. Oncecracking was initiated in hydrogen, failure was rapid. There was
no indication of hydrogen-environmentembrittlement of unnotchedTi-6A1-4V STA

specimens, but an increase of yield strength occurred during the hold period
and was attributed to strain or stress aging.

Thus, no effect of hold time or applied stress on the degree of hydrogen-

environment embrittlement is evident from these results. Similar results were

found (Ref. 52) from an investigation of the effect of hold times of 1, 10, and
100 days on tensile properties in hydrogenof parent- and weld-metal specimens
of ASTM A-517, ASTM A515, and ASTM AS33-B.

In all cases reported, the degree of embrittlement of notched specimens was

approximately the same for the various hold times at constant applied stress

in I0 000-psig hydrogen. The degree of embrittlement has always been found to

be less for the long hold times, probably because of increased impurity effects.

In recent tests, Owens (Ref. 53) tested in air specimens that had been

exposed to 6000-psig hydrogen. Results obtained for notched specimens (t/r_3)

of SAE 4140 and 1018 steels showed essentially no effect of hold time under an

applied stress at 6000-psig hydrogen pressure. The hold stresses were 33 000 and

16 500 psi, respectively. The specimens were exposed to 6000-psig hydrogen for

25 and 50 days, taken out of the vessel, and tested immediately. In addition,

some specimens were allowed to remain out of storage for 1 and i0 days prior to

testing. The results are shown in Table 16. The results show that SAE 4140 was

91
embrittled by this procedure while the SAEI018 was not. It is of interest to

note that for specimensexposedfor 25 days to 6000-psig hydrogen, the ultimate

tensile strength was restored to the original value after i0 days storage at
ambient temperature.

No comprehensive study has been made on the effect of hold stress and

hold time. The limited data to date (Table 16) indicate that there is probably

no effect from these variables. Whatever effect might occur, however, may be

masked by the effect of contamination buildup during the course of these types

of tests.

Effect of Temperature

The data available on the tensile properties in hydrogen as a function of

temperature are limited. The data available are from the work of Hofmann and

Rauls (Ref. 42) and from Walter et al. on Inconel 718 and Ti-6A1-4V (Ref. 26,

47, and 49).

Hofmann and Rauls investigated the effect of temperature on hydrogen-

environment embrittlement between -130°F (-90°C) and 338°F (170°C) at 2205 psig

pressure for cold-drawn and normalized CK 22. The tensile deformation rate was

5%/min. No effect on the ultimate tensile strength was observed at any of the

test temperatures. The effect on ductility (percent elongation and reduction

of area), however, was pronounced, as shown in Fig. 23 and 24. In the cold-

drawn CK 22, the maximum degradation value of reduction of area occurred at O°C

in hydrogen, while in the normalized CK 22, the maximum degradation occurred

at -40°C.

92
 32

•_ 28

t_
16

c _ m.;o
50

"_ "..-.,o.._. _" rests

,_ m In Air, art farm
3O
! e_ In Hydroge4

30 et ;SO e_m

-80 - 40 _0 • 40 .80 * 120 *

remperotu_ "C

Figure 23. Effect of temperature on

elongation and reduction of
area of cold-drawn CK 22 steel
in air and in hydrogen.

32

o 20
r.
16

6O

•_ 50

'_ 40
f • ei IhAirrtl
rests re|m
• em._ k) HydrogelL
af ;$0 afro
• m

-40 tO .40 *80 *1_0 ,_0

r_rr_tr_tur, "C

Figure 2_. Effect of temperature on elongation

and reduction of area of normalized
CK 22 steel in air and in hydrogen.

93
 As can be seen from the data in Table 17 and Fig. 25, the reduction of notch

strength of Inconel 718 for heat B and heat treatment 3 specimens tested in 2000-

psig hydrogen was essentially zero at -109 ° and -320°F, and was essentially the

same (approximately 1S percent) at room temperature and 140°F. The reduction of

notch strength of heat C with heat treatment 4 specimens by S00-psig hydrogen

decreased from 34 percent at room temperature to 14 percent at 400°F, and down

to zero at 1200°F. Results from 1200°F tests may be invalid because of oxygen

contamination of the hydrogen, as was evidenced by surface discoloration of the

specimens. As discussed earlier, the susceptibility of heat B and heat treat-

ment 3 specimens to a hydrogen-environment embrittlement was much less than other

heat-heat treatment combinations of Inconel 718. Thus, greater temperature depend-

ence may be realized with other heat-heat treatment combinations (Ref. 47).

A limited number of hydrogen-environment embrittlement tests at tempera-

tures above room temperature also have been performed on Waspaloy, another

nickel-base alloy. _e reduction of notch strength for Waspaloy in 500-psig

hydrogen was approximately 36 percent at room temperature, 26 percent at 400°F,

and 6 percent at 1200°F. Again, the results from 1200°F tests might be affected

because of contamination. In any case, the data indicate that the hydrogen-

environment embrittlement of Inconel 718 (and probably other nickel-base alloys)

is a maximum in the vicinity of room temperature.

The results of the tensile tests of Ti-6AI-4V(STA) in hydrogen and helium

at various pressures and temperatures are given in Table 18 and Fig. 26. The

loss of notch strength in hydrogen at room temperature is severe. The sharp

reduction in the hydrogen-environment embrittlement of Ti-6AI-4V(STA) as the

94
 [-,

Z
0 q'O I_ _ _ _ i_o r_ i._

I-I
___ _ _ I _ I I _ ....
U
Z
o

._o

0
lllllllll
i.--I IIIIIIII
,..-1 _J

,.-1
_'___II__ii Iiiilll
........ I
, , I I

I , , I I

4..I-.I
Utl
,--1

Z .IJO
[--,

g5
 _H/ZH 0 _J.Vll HJ.'JN3111S
0
O0 O0 0'_ (3": 0

0 d c; c; d 0
0

0 r_

I CH
0 0
0
C_

0 I.L.

:Z)
I-.

I./,4
_8
a.
_r"

I-.
U .el

m
C !
0

e_
I

ii
o4

.LN33i:I3d 'Nflll3H NI J.VH.L HO_.-I

N3OO_(]AH t_l HJ.ON3_J.S H3J.ON 30 NOIJ.3fl(]3_

96
 _0 _ I I _ I C_I I LI_ I _ I I
0 I::: .,._ "_
• _ .l_ ' ' 0 " I 0 ' 0 " , O* I I
_0 _ c_ t_

I/I

0 0
Z

L_
0

_ _ -,,
0

_ 0 _

L¢5 0 cD

o_
• _ o\°

u_

0
u'l
_ Z
m _ 0 o,,_o
I_ _
_ ,_
r_l ,_
;-t _* _ __ ._ C_ _ _ _
o
,.-.1 0 o
r._ LO

.._

.r4 0 C_o

0 0

0 _

U'_ [--,
aa ._
0
_ m
0 0 '._ C_ _ t,_ ,--_ t._ _1 t_ _ t_ C-,,I _'_

E-,

0 0 C_ 0'_ 0 0

.t-t
m [--, I ! I I I I
Z
uJ

U _
0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0
oe,I 0 0 0 0 0 0 0 0 0 0
> ,.-'¢ .--I ,,--I 0..,I _ N _ ("xi Oq °r,t

m _

E-,

97
 aH/ZH OI±V_ HIgN31±S

0 0 0 0 0
0

d d d d
0
0

0
o
v

m -_

h o

I..LI

o -_
I:1
I.U
q-i .p
o_

0
0
Lm_
|
0
et

0
9----0 "
o

N3OOgOAH NI H±9N3_LS HOLON 30 NOli3n03_

98
test temperature is reduced below room temperature is evident from Fig. 26.

Some loss of notch strength occurred at -109°F, but virtually none at -3200F.

Embrittlement of Welds in Pressure Vessel Steels

The effect of high-pressure hydrogen on weldments and parent metal of three

pressure-vessel steels has been investigated by Walter and Chandler (Ref. 52).

The ASTM designations of the steels tested were: A-302-56 Gr. B modified with

nickel (A-302), A-517-64 Gr. F (T-I), and A-212-61T Gr. B-FB (A-212). The A-302

steel has since been redesignated AS33-B and the A-212 Steel as A515. The newer

designations are used in this report. Both longitudinal and girth welds were

tested, except for A515 in which the girth and longitudinal welds were similar.

The welded plates simulated actual pressure-vessel plates in thickness, weld

geometry, weld procedure, and heat treatment. _le thickness of the plates tested

were: A533-B, 5-1/2 inches; A-517, 1-3/8 inches; and A515, 4 inches. Both un-

notched and notched specimens were tested. The notched specimens had a Kt=4.2

(Ref. 43). The specimens from the weld panels were taken normal to the longi-

tudinal and the girth welds, and the notches were located either in the weld zone

or in the heat-affected zone. Tests were conducted with various hold times (1,

I0, and I00 days_ and at various hold stresses in the hydrogen atmosphere. No

effect of hold time or hold stress on embrittlement was observed. The results

of short-time tests are tabulated in Table 19.

No difference was observed between the strength of the unnotched specimens

tested in i0 000-psig hydrogen and in i0 000-psig helium for any of the three

alloys, or for the parent metal and the weld specimens. However, exposure to

high-pressure hydrogen resulted in a decrease in the reduction of area of approx-

imately 33 percent for the A533-B longitudinal welds and 50 percent for the A533-B

99
 U_
Z
ILl

I--4
U
I..I-I _o _ oo _o _ oo oo
o • i
...... _ _
_o _'_
I'M
U..1
, , _

II _ II i i i
,U..,)

0 I I ''_o,_ I I I o_I i

_'_ ', ll ', ', _ ll i i i i o iI ii i1

,--t Q in
U-1 Z i o
U-I .... _ o, o_ i
I
i
i
o 00
I_-
i
i
i
i o_ o_, i i , c I
i o_ o_ o,,
U
[.-, Z
0 LLI
L.9
0
-_#_
y._

L9
_ oo°° _
oo oo oo oo oo oo oo oo oo oo oo oo oo oo

U') oo oo oo oo oo oo oo _o oo oo oo oo oo oo
[--, r_
I
0
0
0 0

U..I
_._
,-.1 Z

Z
U-I =o o_
•_ o_

._ t..i

.,4 ._ ._ ¢_

"_: -_; .,_ ,

_o _
_ o o_, _ ._ _o= o= _ _ ._

Lq

100
girth welds, the A-517 longitudinal welds, and the A515 parent metal. The

longitudinal weld of the A515 showed a 25-percent decrease in the reduction of

area. A small effect of the hydrogen environment was observed for all mater-

ials except the A533-B girth weld and A515 parent metal. The reduction of area

data are shown in Fig. 27. The notched specimens that were tested in the hydro-

gen environment showed a reduction of both strength and ductility compared to

specimens tested in helium. The smallest strength reduction occurred in the

A533-B parent-metal specimens. The transverse parent-metal specimens had almost

no strength reduction, while the strength of the longitudinal weld was more em-

brittled than the parent metal. The girth-weld, heat-affected zone was the most

embrittled by the hydrogen environment. The A515 parent-metal specimens showed

a greater strength reduction than did specimens from the other two parent-metal

materials. The welds of A515 were not as embrittled as the parent metal. There

was a small effect of I0 000-psig hydrogen on the A-517 parent metal in the

A-517 longitudinal weld and longitudinal weld heat-affected zones. The A-517

girth welds were not as embrittled as the longitudinal wdlds.

Tests also were performed at 3000-, 5000-, and 7500-psig hydrogen pressures

on notched longitudinal parent-metal and notched longitudinal weld-metal speci-

mens of these three materials. Only for the notched A515 parent metal and the

A515 _pecimens containing a notch in the longitudinal-weld metal was there a

significant decrease of embrittlement as the hydrogen pressure was reduced from

i0 000 to 3000 psig. _le specimens tested at 3000 and 5000 psig had significantly

higher strengths than those tested at higher pressures. The ultimate strength of

the A515 notched parent-metal specimens was reduced by 10 percent in 3000-psig

101
 A533-B A-517 A515
A A

LW GW P LW P LW

10

DAYSSTRESS
AT HELD IHe H2 l,.
,2.
i""
,2 I..!,21
l,e,2
I,o.2
I,o'.2
UNDER PRESSURE-"-" I I 1 I I 1 !

HOLDING .....,,,.78 78 78 87 87 34 53
STRESS I KSI _

Figure 27. Reduction of area of unnotched parent metal and weld

specimens in lO000-psi hydrogen and helium. (P =
parent metal, LW = longitudinal weld; GW = girth weld.)

102
 hydrogenand by 30 percent in i0 000-psig hydrogen comparedto helium at cor-

responding pressures. There wasa correspondingly higher ductility of the A-515

notched specimenstested at 3000- and 5000-psig pressures than at higher hydrogen
pressures (Ref. 52).

Pittinato (Ref. 54) has determined the properties of Ti-6AI-4V ELI weld-

ments in forged plate, and found no environmental effects whentesting in 14.7-

and S0.0-psia pressure hydrogen. The ultimate strength was lowered at the most
by 2 percent by the hydrogen gas. The results are shownin Table 20. Pittinato
also investigated the effect of prestrain and abrasion with iron on the tensile

properties and found only an additional 1-percent loss of tensile strength (total
loss 3 percent) from prestraining and abrasion.

Effect of llydrogen on Stress-Strain Curves

An interesting effect of hydrogen on the stress-strain curves for OFHC cop-

per and high-purity silicon iron has been observed by Vennett and Ansell (Ref.55)

and Bernstein (Ref. 56). These results are shown in Fig. 28 and 29. For OFIIC

copper, an approximate loss of 16 percent in the ultimate tensile strength occurs

in I0 000-psig hydrogen. The hydrogen "damage" apparently begins before the onset

of necking (before the ultimate is reached) and most likely is related to the

onset of plastic deformation. The fracture surface also changes remarkably in

i0 000-psi hydrogen, as shown in Fig. 30. _le "double cup" geometry generally

found in ductile materials disappears in i0 000-psig hydrogen. The fracture sur-

face in both instances was ductile. For silicon iron, tile stress-strain curves

were almost identical, but the elongation and ductility changed drastically.

103
 0.,..4

Q;

.,,-I
U

_J
o_
E0,-_ • e. •

LQ I._ ct_ _J_

0 0
c_ ,io
0
I

n_ >-,
_O 0
I
.rl
F-,

Z n_ m 0
o 0 0 0
c_ 0 o

o9
0 o

0 0 0 0 0 _

,-1
r_

r'_
I._ o

Z n_
0 E vI vI vI vI
F_ r,, 00_
nO _-_

0
vt Vl Vl Vl

I"',. r-,. 0 _ 0

0 0
m
0

...-4
>

L3

104
 I OF CC : ,

_10" /I _----_'_ Cross-Head Speed: 0.02 in./mi..

....

0 I/ I I )0 ,rain. Expesucf
I e I

0 4 8 12
Elon|ation (10 "2) in,

Figure 28. Engineering stress vs elongation observed OFHC

copper tested in alr and hydrogen.

8O I ! I I I

7O

6O

AIR

"o 14.7o_sia
HYDROGEN
-. 4c ARGON
10 psia
10 psia

io;

I I I I
0 5 I0 15 2O 25
PLASTIC STRAIN, %

Figure 29. Effect of environment on ductility of Iron-3%

silicon polycrystals.

A B
AiR IO.OOO _ HYDROGEN

Figure 30. O_C copper tensile specimens-- smooth-bar fractures

in alr and in hydrogen at i0000 pslg.

lo5
Also, the fracture mode changed appreciably. For the sample tested in air, a

significant amount of ductility occurred and the major part of the brittle

region of the fracture surface was £ntergranular. In the sample tested in the

lO-psig hydrogen environment, there was no localized deformation, and the frac-

ture mode was predominately cleavage. This was particularly true near the frac-

ture initiation site.

106
 C}_PTER 7. OTHER MECitANICAL PROPERTIES

With the exception of tensile properties, the majority of mechanical prop-

erty data in hydrogen is on crack growth in hydrogen environments at various

pressures. Extremely limited data exist on creep, stress-rupture, fatigue life,

and fracture mechanics.

Creep and Stress Rupture

Klima, Nachtigall, and Hoffman (Ref. 57) investigated the effect of hydro-

gen on the stress-rupture properties of an iron-base alloy (A-286), a nickel-

base alloy (Inconel 700), and a cobalt-base alloy (S-816). The test specimens

were hollow tubes into which hydrogen gas (99 percent purity, -70°C dewpoint)

was supplied at 500 psig from pressurized cylinders. Comparative tests were run

with air at 500 psig. The results are given in Table 21. The 1500°F stress-

rupture strength of S-816 was significantly less in hydrogen than in air.

Hofmann, Rauls, and Vogt (Ref. 58) have shown that failure in smooth speci-

mens of a 0.7% carbon reinforcing steel occurred within a few minutes to several

hours in hydrogen at an applied stress of 200 ksi at 40°'C. Failure did not occur

within a 1000-hour test period when the same material was tested in air. The

0.02% yield stress of this alloy was 208 ksi. The results are shown in Fig. 51.

Tests with notched specimens of the same steel showed a sharp increase in time

to failure at a critical stress at -4°F (-20°C) (Fig. 32). At 68°F (+20°C), no

effect was observed.

Fatigue Properties

The effect of hydrogen at 1000 and I0 000 psig on the low-cycle fatigue

(tension-tension) strength of unnotched and precracked cylindrical specimens was

determined for AISI 310 stainless steel, ASTM A553-B, and ASTM A-517 steels (Ref. 7)

107
 TABLE 21

STRESS-RUPTURE LIVES OF THREE HIGH-TEMPERATURE

ALLOYS IN 500-PSIG GAS (Ref. 57)

Average Standard Confidence

Internal Stress, Temperature, Life, Deviation, Leve i,
Alloy Gas psi °F hr hr percenti le

A=286 Air $7 000 1200 264 +32

83

57 000 1200 215 +90

H2

Intone Air 40,000 1500 209 +15

7OO
2O

40 000 IS00 206 +40

H2

S-816 Air 19 000 IS00 215 _+32

98

19 000 ISO0 iSl +31

H2

108
 120

lO0

Z
0
- 60

Z
G
40

Figare 31. Creep of reinforcing

2O steel at 104 ° F (40 ° C)
in air and in 1470-psia
hydrogen.

O.01 O. I I.0 10 100 1OOO

TIME, HOURS

'!. 250

200

|50 [_ TESTS IN _

I- °HYDROGEN' _ --'--"I_ --

1oo, _147o PSi TEMPERATURE -20 C

o.ol o.i I.O 1o I_ IOOO o.ol o.I I.o 1o 1oo 1ooo

TIME, MINUTES

Figure 32. Creep of reinforcing steel with round notch in l_70-psia

hydrogen at 68°F (20 ° C), and -4° F (-20 ° C). Tensile
properties in air are shown for comparison.

lO9
It should be noted that the results from precracked specimens do not represent

conventional fatigue life, but represent comparative rates of cyclic crack

growth in the test environment. Low-cycle fatigue strengths of unnotched speci-

mens of all three materials and of precracked, AISI 310 stainless-steel specimens

were about the same as their tensile strengths in the same environment. The

low-cycle fatigue properties of the preeracked ASTMAS33-B (Fig. 33) and AS_4

A-517 (Fig. 34) specimens were quite similar to each other in all environments.

For both materials, the lO00-cycle fatigue strength in 10 O00-psig hydrogen was

about 1/3 that in 10 000-psig helium; and for the 10 000-psig hydrogen tests

there appeared to be a low-cycle fatigue limit that first occurred between 1000 and

2000 cycles. The fatigue strengths of precracked ASTMAS33-B and ASTM A-S17

specimens in lO00-psig hydrogen were about midway between the strengths in

10 O00-psig hydrogen and that in 10 000-psig helium.

The unnotched specimens were load cycled between zero load and a load in the

vicinity of their yield strengths. The results showed no environmental influence

on the fatigue life of these specimens.

Results of low-cycle fatigue tests on precracked specimens for Inconel 718

at room temperature are plotted in Fig. 35 (Ref. 49). Essentially no difference

exists between the fatigue strength in 100-psig hydrogen at room temperature and

that in air at 14.7-psia pressure. The fatigue strength for 100 cycles to fail-

ure was 22 percent lower in 2000-psig hydrogen than in 2000-psig helium at room

temperature. Results of fatigue tests at -260°F on precracked specimens of

Inconel 718, with a different heat treatment, are plotted in Fig. 36 (Ref. 59).

At -260°F, the fatigue strength is essentially the same in 800-psig hydrogen as

in 800-psig helium.

110
 4a

,.-I o
o
o
o

,-.-4 ro

rd'j

,.v.

._1

L,
co
o
I.--

....4
>.

co e_.j

I
0 _ _
I
_ _+_
I o
I
I _0_
I
I II
I II
I I
I I

I$_ ']0nllljI¢/ sS]WC$

1_1.1
 0
0
0

I -- 0
a

-,'-1.,'-4

0
0
- 0

t_-

W
_®---

0
. o 0

.
0 0
U

t'-O

U ¢J

g,_l 0 0

I 0
0

112
I S)l '3ONV_I SS3_!1S

].]-3
 220

210

200

190

18o
u

170

160

150
A
0 8O0
140

i 30

120

II0

I00
50 I00 1000
CYCLES TO FAILURE

_leare 36. Effect of 800-psl hydrogen and helium on the low-cycle fatigue
stren&_ch of precracked specimens of Inconel-718 (heat B, heat
treatment i) at -260 ° F. (See Tables IX and X for heat and
heat treatment .)

114
 Pittinato (Ref. 60) has also investigated the influence of gaseous hydrogen

on the fatigue life of Ti-6A1-4V using precracked specimens. Tests were con-

ducted in high-purity hydrogen, liquid hydrogen, liquid nitrogen, and helium for

comparative purposes. The results are shown in Fig. 37. The number of cycles

to failure was less in hydrogen than in nitrogen or helium at temperatures be-

tween -100°F and room temperature.

Van Wanderham and Harris (Ref. 61) have studies conventional low-cycle

fatigue in hydrogen. Results were obtained at 5000 psig using both hollow speci-

mens that were internally pressurized with hydrogen or helium and solid speci-

mens in a hydrogen-filled pressure vessel. Axial strain was measured and con-

trolled by means of a proximity probe extensometer.* The chemical composition

of the materials tested at Pratt and Whitney are given in Table 22. The type

of specimens used by these authors is shown in Fig. 38.

Results for S000-psig hydrogen and helium are given in Fig. 39 for two heats

of Inconel 718. Considerable degradation of fatigue properties is evidenced

from these data for heat code LDUL. For example, at a 1-percent total strain

range with hollow specimens from heat code LDUL, a life of approximately 3000

cycles would be predicted for helium while in hydrogen the lift would be only

150 cycles (Fig. 39). Data are not yet available for heat code BVTO in helium.

Figures 40 and 41 show the results of low-cycle fatigue tests for Inconel

625 conducted at 78°F and 1250°F, respectively. At room temperature and a

1-percent total strain r_,ge, a fatigue life of 2000 cycles in helium and

500 cycles in hydrogen i$ indicated. At 1250°F and a 1-percent total strain

range, a fatigue life Qf 1000 cycles in helium and of approximately 250 cycles

in hydrogen would be predicted. At a 1-percent plastic strain range, a life of

*Patent applied for by M. C. VanWanderham, J. A. Harris, Jr., and R. Bogart.

I15
 17 000

! 6 000 H2 ' I

15 000

l h 000

! ) 000
L IX
I NZ-\ '
12 000 I--
..--t ! ____-- _ %_ II2 -- N2-.--._-e
,,=I2 I ] 000 I--
Lm- I "2_
0
1.--
lo 000
, |
-- _ --E-
9 000
,L..)
>- 8 000
_) + ,
7 0o0 'LIQUI[ --" i , ,,4

__.H2_ 01 I
6 000 --E

5 ooo

4 000
I
3 ooo i

-400 -300 -200 -100 0 +lO0

TEMPERATURE, F

Figure 37. Fatigue curves of precracked Ti-6AI-4V

tested under tension-tension loading
using a lower stress of 12.5 ksi and an
upper stress of 62.5 ksi.

116
 I I I _ I

,'-4

_0
i l i 0 |
,..-1 I ! I I
d

_ _ _ _ ,
m
t _ 0 , !
N
I I _ I l

0
£-_ O _d

C_
_4
O O ,
I ,-4 I I
Z
_4 J ' o ' '
___
0

O I O _ _
cO O I O O O

U.)
C) o _ _ o
k
0 tl
_4 O O I,_

j <
O
_4 •_...I
I
O O
•e-I .el
m _
O o d ' d
0
Z

d ' ' o
0 t_

CO
Z ._ Uo
[-.,
0 N
U.1 o M

Z
0
a

{_J
U
!
_ _ _ _ '
O
_3

0 "'4 0 t'_ I 0

r.) O ,"4 O "'4 O

ir
°_.-I

117
 Test specimen used
for low-cycle fatigue
tests conducted in
5000-psig hydrogen.

118
 0
0
0

__ 0

¢;

0
0

1"4

0 0
0
_ 0

w,,,-
Z
,-'1

_ g
t,f

b-- 0
O.
1:1

m
_"T-

I --

Z 0
4-)

o.
)-
0 0

_ Z
_ - 0

& - 0

_ -

0
d

119
 i0.0
m
m

I°0
h

a_P
b
uJ

p,-
¢/I

Ool
SOLID SYMBOLS ,,, TOTAL STRAIN
OPEN SYMBOLS ,, PLASTIC STRAIN

u
OOSOLID SPECIMEN IN AIR
Ill--IHOLLOW, 5000-PSI HELIUM
00HOLLOW, 5000-PSI HYDROGEN

O.Ol
.1 I I l, llll . I I I I IIII I I, I I IIII
l0 lO0 1000 l0,000
CYCLES TO FAILURE, Nf

Figure 40. Effect of cyclic strain on fatigue life at 78 ° F of

Amuealed Inconel 625 (I_A 1040 Heat Co_e HUD_.

120
 0
0
0
0

q_
0

P,

(_

/
!
/
4.)
_v

Illll I 1 I IIII1 I I I 0

0 0 m

0 0

35N_1 N I _IIS

].2_l
less than 50 cycles in hydrogen would be predicted at room temperature and of

less than 10 cycles at 1250 F. The two data points at a given Nf value for

plastic strain range in Fig. 40 through 46 do not indicate data scatter, but

represent the observed change in the plastic strain range for that specimen.

Figures 42 and 43 show the low-cycle fatigue behavior of PWA 1027 Waspaloy

at room temperature and 540°F, respectively. At room temperature and at a l-

percent total strain range, a life of 7000 cycles would be predicted for helium

and a life of I000 cycles for hydrogen. At 540°F and at the same total strain, a

life of approximately 2250 cycles in helium and of approximately 200 cycles in

hydrogen would be predicted.

Degradation of fatigue properties by 5000-psig hydrogen also was encountered

in the cobalt base alloy Haynes 188 (Fig. 44). The degradation observed was not

as severe as in the nickel-base alloys. For the Haynes 188 alloy at 1-percent

total strain range, a life of approximately 6000 cycles would be predicted for

specimens in helium while in hydrogen a life of approximately 2500 cycles would

be predicted. At a 1-percent plastic strain, a fatigue life of between 500 and

600 cycles would be predicted in hydrogen.

Degradation of fatigue properties was not as severe in the Crucible Steel

Company Alloy CG-27 at room temperature as in the nickel-base alloys. At a 1-

percent total strain range, a fatigue life in helium and hydrogen of 4000 cycles

and 1100 cycles, respectively, would be predicted (Fig. 45). At S40°F and the

same total strain, a fatigue life of 5000 cycles would be predicted in helium

while, in hydrogen, a fatigue life of 1000 cycles would be expected. This would

indicate that degradation is more severe at S40°F than at room temperature (Fig. 46).

122
 I0.0
!
w 1850°F (4 HRS) OIL QUENCH OR WATER QUENCH
)550°F (4 HRS) AIR COOLED
b 14OO°F (16 HRS) AIR :OOLED

!o0

I--
fo')

0.1

b "-_ ._),' >\<

SOLID SYMBOLS = TOTAL STRAIN
b OPEN SYMBOLS = PLASTIC STRAIN

OOSOLID SPECIMEN IN AIR

II--IHOLLOW, 5000-PSl HELIUM
IOHOLLOW, 5OOO-PSI HYDROGEN

I
0.0! , .I I I IIIIII I I I I IIII I I I l llll
O I00 I000 10,O00

CYCLES TO FAILURE, Nf

Figure 42. Effect of cyclic strain on fatigue life at 78 ° F of We.spe_.o_"

(I_4A 1027 Heat Code LCCS).

123
 •
/ 0 I

/
/
/
/ /
/ /
.)
/
T
/. /
, / m

,i

/ 0

/
0 ,-_
-r ,-_ L.i
¢.) ud .J
Z-JO
f 0

l
eUO0

0'_

0"1
L_m_
i

0 0
0
0
0
30N_I NIV_IS
 0
0
0

-- 0
0

nd

0
0
0
==

_ 0_

..3

-J 0

)-

4J

•_ 0

0
0
0,--t

m 0 _

IIII I1 1, I1 .111 I I 0
0 0

0 ¢;
35NV_I N IV'diS

Z25
 ]0,0
M

_900 °F (i HR) OIL QUENCH

14OO°F (16 HRS) AIR COOLED
1200°F (16 HRS) AIR COOLED
m

).0 m
m

a.#

tl.i m

\
z
m \
\
\
I,-

\
0,1 .\

SOLID SYMBOLS = TOTAL STRAIN

b
OPEN SYMBOLS = PLASTIC STRAIN 0
"0
0
OOSOLIO SPECIMEN IN AIR
1DMOLLOW, 5000-PSI HELIUM

iIL=IQ_OHOLLOW _ 5000-PSI HYDROGEN

_....

) I ) )}Jl_ I ) I1_111
0.01 _ I I 11111
100 1000 10 _000

CYCLES TO FAILURE, Nf

Figure bS. Effect of cyclic strain on fatigue life at 78 ° F

of CG-27 alloy.

z_6
 0
0
0

0
0

¢u
!
0
0
8
0

0 o

/ o
/
/
I
I
/
Z

o/o I
/ LIJ r-i

::::)
--I
0 --
oo<LI.
C)
I--

o
LIJ
J
U
>-
(,_)

_q

0
0
0

,d

0
0 0
0
0

3_)ND_d N IV'dJ.S 0

127
 Figure 47 summarizes data for tests at a 1% total strain range. It can be

seen that all the alloys except A-286 had more than a 60% reduction in fatigue

life in 5000-psig hydrogen with Inconel 718 having the most severe reduction of

fatigue life. The ratio of the cycles to failure in helium to the cycles to

failure in hydrogen at 5000 psig also is shown in Fig. 47. The data shown for

A-286 Were obtained from tests run at 1% total strain range only.

Fracture Mechanics

Threshold fracture toughness (Kth_ values have been determined in high-

pressure hydrogen for Inconel 718 and 2219-T6E46 aluminum by Lorenz (Ref. 62).

This program was undertaken to determine which of these materials would be the

best for use in hydrogen pressure vessels on the Mariner '69 space vehicle.

Lorenz (Ref. 62) found for the Inconel 718 alloy that Klc (in air) was

150 ksi i¢_-_,based on the results for one specimen that fractured below the yield

strength of the material. For Inconel 718 weldments, the Klc value was

95.0 ksi _ .

Sustained flaw growth data derived from surface-flawed specimens tested in

5200-psig hydrogen indicated a Kth of 22 ksi i¢_. for Inconel 718 (Fig. 48) and

21.0 ksi i_. for Intone1 718 weldments (Fig. 49).

The Kic value (in air) for the 2219-T6E46 aluminum base metal was

35.7 ksi i¢_., and 32.0 ksi i_. for the weldments. Sustained flaw growth data

are shown in Fig. 50 and S1 and the data indicate Kth values of 28.0 ksi

and 26.0 ksi i¢_., for the base metal and the weldments, respectively. Thus, it"

128
 RATIO Nf H2

Nf He

O.Ok O.15 O.10 0,25 O.25 0.34 0.35

0.20 1.0
80OO
I I

5OOO-PSIG He

5OOO-PSIG HZ
7o00

ALL TESTS AT 78°F

EXCEPT WHERE NOTED*

6OOO

50oo

o 3oo0--

2000 --

1000 --,

MATERIAL co
r-._l

Figure _7- Relative low-cycle fatigue life at 1-percent total strain.

129
 ell

"I"
v

b-
0

0
0 0
..I o

u.I
%
Z

I,l.I

I-,-

4-_

8 8 8

(N_._ IS)l) U)I ' A.I.ISN:::I.I.NI SS:I_iJ,S 031"kld¥

Z30
_ _ o
'AIISN:ilNI SS::l_t.l.S (l::tl'lddV

131
 O

4a

_o

I.i.I

ILl

4O
O

IS)i) 11)1 '111SN:IINI SS3_IlS (]31"lddV

132
 O

,,-4

_D

!
O_

8,
c_

,el

_0
I

I:I

+_

_J

i
0
i,--

.',_J.ISN::IJ.NI SS:I_I.I.S C]:lrlddY

133
 can be seen that the aluminum retains its load-carrying capability in 5200-psig

hydrogen at stress-intensity levels of more than 80% of the KIc fracture tough-

ness values, while the Inconel 718 is drastically affected by hydrogen.

Crack Growth

Hancock and Johnson (Ref. 63); Spitzig et al. (Ref. 64); Walter and Chandler

(Ref. 47); Marcus and Stocker (Ref. 30); Williams and Nelson (Ref. 8); and Nelson

et al. (Ref. 65) have investigated the effect of gaseous hydrogen on crack growth

rate in steels, nickel and nickel-base alloys, and a copper-base alloy. Accel-

erated crack growth was found in hydrogen in all cases except in the copper-base

alloy.

Hancock and Johnson (Ref. 63) investigated the effect of hydrogen at atmos-

pheric pressure on crack growth in H-11 tool steel. They found that subcritical

crack growth initiated in hydrogen at lower stress field intensities and propa-

gated at faster rates than in wet argon, as shown in Table 23 and Fig. 52.

Hancock and Johnson further showed that additions of small amounts of oxygen

(0.6%) to the gaseous hydrogen environment completely stopped crack growth.

This effect is shown in Fig. 53. At the lower stress intensities, there appears

to be an incubation period before crack growth begins. This emphasizes the

importance of having clean initial cracks at the beginning of a test in hydrogen.

Spitzig et al. (Ref. 64) investigated the effect of dry and humid hydrogen

environments on crack growth in 18 Ni(250) maraging steel. The results are

shown in Fig. 54 in terms of (i/c O - I/c)in. -I versus the number of cycles N,

134
CRACK INITIATION (Ki) AND UNSTABLE FRACTURE (KIc)
STRESS FIELD INTENSITIES

Hydrogen Wet Argon Dry Argon

40

4O
(ksi in.) .(i) 40

(1)
Because of the long subcritical crack, could
not be measured. Klc

.og

.08

.06

.05
,07'
.04

_ .o3
.02

%
Time (mil%ul@$|

Figure 52. Subcritlcal crack growth of H-II steel in

molecular hydrogen and humidified argon.

I I IJ

C
.24
/
C
o
"_ 1.6
C
o
K

Rus
/Hydrogen
_" .08 0.6 °I,,Oxyg_
U
e
U

0 = | ! I
0 4 8 12 16
Time, (Minutel ]

Figure 53. Subcritical crack growth of H-II steel in

hydrogen-oxygen mixtures.

135
 Z _ 0
0
c_q
X
__ n, _
. .,s-. ,_.,,,

-I--!"

,O

1;I

Z Z
--I.14 0
ZO W
OF,," .,.I
__ ,,.._
l-->- u v

,_>.-

.,'..,I

O- <1
.

[] <_ •
!
[]-<_ --O--
.r-I

D<_ •

_-O

D_O-

S3H3NI ' 31 L-e311

i-

136
where c is the equivalent half-crack length (Ref. 66 and 67) and co is the ini-

tial equivalent half-crack length. The variation in the crack growth rate for

the "dry" hydrogen is believed to be caused by variations in the amount of water

contained in the hydrogen used. Duplicate tests in the humid hydrogen environ-

ment showed negligible differences.

Walter and Chandler (Ref. 47) have shown that cyclic crack growth rates in

Inconel 718 are faster in 2000-psig hydrogen than in 2000-Dsig helium and ambient

air. Their results are shown in Fig. 55. The chemistry and heat treatment of

the Inconel 718 used are given in Tables 9 and 10, Heat B, heat treatment 3.

Crack growth rates were determined by an electrical resistivity method, tlydrogen

purity was <0.5 ppm 02, <0.5 ppm H20, and 2.6 ppm N2.

Walter and Chandler (Ref. 7) used electrical resistivity measurements to

determine the onset of crack growth in tensile tests on notched cylindrical

specimens of ASTM A-517, Inconel 718, and Ti-6A1-4V. These results are presented

in Table 24.

TABLE 24

CRACK GROWTH DURATION DURING TENSILE TESTS OF NOTCHED SPECIMENS

TESTED IN HIGH-PRESSURE HYDROGEN

Stress at Time Period

Notch Notch Which Crack Over Which
Strength Strength Initiation Crack Growth
Pressure, in He, in H 2 , Occurred, Occurred,
Specime n psi ksi ksi ksi sec

A-517 10 000 251 183 173" 39*

Inconel 718 2 000 333 286 249 25

Ti-SAI-4V 2 000 25O 132 114 40

*The crack growth resolution during this test was quite marginal and thus
considerable crack growth may have occurred past this time period and
before this stress is reached.

137
 I
0

bO
I

ILl
.--I

C.)
"-I-

0
_._

_1 0

0_o

I i _-4

IS)I '39NV_I X.I.ISN31NI SS3_IIS

138
 The results indicate that failure is preceded by a considerable period of

crack growth for all three materials. The hydrogen-affected region as determined

from fractographic examination (see Chapter 8, Metallography) was, therefore,

formed over a significant period of the total time during which the applied load

was continually increasing.

The electrical resistivity was also monitored during comparative air and

helium tests. The results showed that failure occurred almost immediately fol-

lowing the first indication of crack formation.

Marcus and Stocker (Ref. 30) have investigated the effect of hydrogen on

crack growth rates in the pressure range of 10 -8 to 150 torr for Ni-200 and a

copper alloy, NARloy-Z% using tapered double-cantilever beam specimens (TDCB).

The apparatus used in the foregoing experiments has been described previously in

Chapter 5, Test Procedure. Tests were conducted at room temperature at a fre-

quency o£ 2 cycles per second with a maximum stress intensity of 27 ksi i_. and

a cyclic stress intensity of 11 ksi _ . The crack propagation rate of Ni-200

was 0.02 _/cycle at 10 -9 torr pressure, and increased to 0.i6 _/cycle in hydrogen

at a pressure of 150 tort. A crack growth rate of 0.2 _/cycle was established

-10
for the NARloy-Z in the 10 tort vacuum environment; this rate remained essen-

tially the same in the hydrogen environments tested. Further tests on crack

growth rates in Ni-200 at 150 torr hydrogen pressure between 77°F (25°C) and

-58°F (-S0°C) showed that the rate of crack growth was a maximum in the vicinity

of -13°F (-25 C). The crack growth rates were relatively frequency independent

in the range of 0.2 to Scps.

*Trademark of North American Rockwell Corporation

139
 Similar increases in crack growth rates were observed in hydrogen at room

temperature for titanium alloys (Ti-6AI-4V, Ti-SAI-IMo-IV, and titanium 75A)

(Ref. 30). Measurable increases in the crack growth rates were observed at

hydrogen pressures as low as i torr. For annealed Ti-6AI-4V in 150 torr hydrogen,

with a Cyclic stress intensity K of 13 ksi i_n., crack growth rates increased in a

linear fashion from 0.6 to 3.3 _/cycle as the stress intensity maxima were in-

creased from 20 to $0 ksi i_.. The crack growth rate in vacuum (10 -10 torr)

and 14.7-psia air was 0.4 _/cycle at a K max of 37 ksi _ .

Attempts to determine crack growth rates with TCDB specimens of columbium

_niobium) resulted in formation of cracks in the grip section and fracture of

the specimen normal to the desired direction. Any region of the sample which

was stressed beyond the yield point resulted in the formation of microcracks.

Williams and Nelson (Ref. 8) measured the stress intensity factors required

for the initiation of measurable slow crack growth (KscG) in fully hardened,

stress-relieved 4130 steel as a function of displacement rate. The results for

tests in air and at hydrogen pressures ranging from 127 to 709 tort are shown

in Fig. 56. At low-displacement rates, the KSC G values in hydrogen are about

one-half those in air. The KSC G values for air and for hydrogen at the faster

displacement rates are equal to the critical stress intensity factor K C. The

maximum embrittlement in low-pressure gaseous hydrogen was observed to occur at

slow strain rates.

Figure 57 shows the results of Williams and Nelson (Ref. 8) for crack prop-

agation rates as a function of the reciprocal of the absolute temperature. Each

datum point is the result of two measurements at the indicated temperature. All

measurements were taken using the same specimen. Crack growth began immediately

140
 55x 103

50 _- 760 tort(AIR) _'- /__/'Q

45- e/o /// //

-_" 40- /_7 // ///

k
30 //._.__09__ torr (H 2 )

25-

2 0__ 4 I I I , m I I 010-3
I I I u i 10-2
_ n i il

DISPLACEMENT RATE, inJsec

Figure 56 Effect of hydrogen pressure and displacement rate on the

embrlttlement of _130 steel.

141
 I
0
I

×
_t
®
U)

0 0

® 0
!

,,0-- I

® v
0

®
ILl I 4_
n
0

W 0-

O_
cU

O_

0 Q
Z
0
I
O-
(D W

I tl%

N
!
0 'o
i m

OOS/WO '1:1 '31V_ HIMO_9 NOV_O

after loading with no indication of an incubation period. From Fig. 57 it is

observed that the maximum crack propagation rate occursnear room temperature.

A similar temperature dependence has also been observed for internal-hydrogen

embrittlement (Ref. 68). Figure 58 shows the results for crack propagation rates

as a function of pressure in the temperature range of -76°F (-60°C) to 140°F (60°C).

Nelson et al. (Ref. 65) investigated the effect of partially dissociated

hydrogen on embrittlement. Figure 59 shows the effect of temperature on KSC G in

various environments. Again, KSC G in air is nearly independent of temperature

in the range studied. In a molecular hydrogen environment the results are simi-

lar to 'those in Fig. 56. In an atomic-molecular hydrogen environment the temp-

erature dependence and the magnitude of KSC G vary significantly from those in

molecular hydrogen. Figure 59 shows that KSC G is pressure dependent in the

vicinity of room temperature. The nature of this pressure dependence near room

temperature is shown in Fig. 60, where KSC G is shown as a function of molecular

hydrogen pressure. The results of crack-growth studies in an atomic-molecular

hydrogen environment at a constant hydrogen pressure of 8 x I0 -3 torr are given

in Fig. 61. The apparent activation energy of 6800 cal/g-atom was found for

crack growth in a partially dissociated hydrogen environment. The significance

of these results will be discussed in Chapter i0, Discussion.

143
 10-2 --
o 60 ° C (REGION a_
A 23°C(REGION2)

o 0-60°0 (REG ION 3)

,_. K = 36 kSl-In. I/2
• 13

_ n.= 1.0

_ 10-3_

10-4L -"i I I I I I I I I
I00 I000
PRESSURE, P, t0rr

Flgure 58. Effect of hydrogen pressure on the slow crack-growth rate

of 4130 steel.
 .,-I

0
t_

II

_g
o
0
0

0 o
_.,_

0,_

_g

0
!

0
X
0

_/iu!__u!lql,
6'os
>_
 60xlO 3

,o
t_l.
C:

.4:}
m T=23°C
40--
t_ = 3 x I0-3 in./min
{/)
v

30L , i iitllsl t i lll,,,I I i i IIIIII

10-3 10-2 i0-1 I0 C

HYDROGEN
PRESSURE, PH2' torr

Fi&nAre 60. Dependence of embrittlement o£ 4130 steel on molecular hydrogen

pressure in an atomic-molecular hydrogen environment.
 TEMPERATURE, °C

10-3 __.. 40 , 20 , ? , -20 _ -40

5,,\

•E
,- PHz = 8 x 10.3 torr
uJ - _ K = 41 ksi- in!/?.
'_tr I0 -4- _,._ Q = 6800 cal/mole

Ilg
_) Io-SL I I I I I I I
3.0 :3.2 3.4 :3.6 :3.8 4.0 4.2 4.4x10 -3

I/T, °K -I

Figure 61. Dependence of slow crack-growth rate in 4130 steel on temperature

in an atomic-molecular hydrogen environment.

i_7
 CHAPTER
8. METALLOGRAPIIY

The results of metallographic and fractographic examination of specimens

tested in hydrogen environments are described in this chapter, llydrogen environ-

ment embrittlement results from the formation of a surface crack or cracks and

the propagation of these cracks in a brittle manner. In the majority of em-

brittled metals, extensive surface cracking is observed near the fracture of

unnotched specimens. Hofmann and Rauls {Ref. 40) were the first to observe these

surface cracks; they associated the cracks with gaseous hydrogen pockets, "fish-

eyes," that form internally in steels containing internal hydrogen.

Walter and Chandler (Ref. 7) examined the surface of unnotched specimens

of 32 alloys tensile tested at ambient ten_eratures in air (ambient pressure),

i0 000-psig helium and I0 000-psig hydrogen environments. Surface cracks, ori-

ented approximately parallel to the fracture surface, were apparent on all the

specimens embrittled by i0 000-psig hydrogen. No cracks were observed on speci-

mens tested in air or in I0 O00-psig helium. It was shown (Ref. 7) that the

metallographic characteristics associated with hydrogen-environment embrittlement

can be categorized according to the four embrittlement categories presented ear-

lier (Table 8 ). The results for unnotched and notched specimens are presented

together and are described as a function of the degree of embrittlement.

The metallography shown in this chapter is taken, except where noted, from

Walter and Chandler (Ref. 7). The specimens tested in I0 000-psig hydrogen which

were negligibly embrittled did not contain surface cracks and the fracture mode

of these specimens appeared to be ductile.

149
 Slightly Embrittled Specimens

The slightly embrittled group of metals either had no surface cracks, as

observed for Be-CuAlloy 25, or contained a large numberof very small surface
cracks, as observed on titanium and AISI 304L stainless steel. The results for
commercially pure titanium are illustrated in Fig. 62. Figure 62(A) and (B) show
that the surface cracks on unnotchedspecimenswere numerous,shallow, and blunt.

The fracture of the unnotchedspecimensdid not, however, appear to originate

from these surface cracks. There wasno measurablereduction of ductility, and
the shear lips were well-formed indicating that the fracture initiated within
the specimens.

An electron fractograph of the machinedsurface/fractured surface interface

of a notched titanium specimentested in 10 000-psig hydrogenis shownin
Fig. 62(C). The fracture is brittle and transgranular with a considerable
amountof secondary cracking perpendicular to the fracture surface. The depth
of the brittle region was approximately 0.02 inch.

The surface cracks on AISI 304L stainless steel were similar in shape to
those observed on commercially pure titanium. The cracks were sufficiently deep,

however, to cause a slight (8 to 9 percent) reduction of ductility and the shear

lips on these specimenswere somewhatdiscontinuous. Bensonet al. (Ref. 46)
attributed the nucleation of surface cracks on AISI 304L stainless-steel speci-

menstested in 10 000-psig hydrogento strain-induced martensite platelets.

Blunting of the cracks was observed in the more ductile austenite matrix.

150
 (A) lOX

(B) 110X

'

(C) lO00X

Figure 62. Metallography of Commercially Pure Titanium Specimens

Tested in I0 000 ps_Hydrogen ((A) Photomacrograph (10X)
Showing Multitude of Small Surface Cracks; (B) Photo-
micrograph (IIOX) of Surface Cracks; (c) Electron Fracto-
graph (!000X) of Notched Specimen at the Machined Surface/
Fractured Surface Interface)

iSl
 Severely Embrittled Specimens

The majority of metals tested by Walter and Chandler (Ref. 7) were in the

severely embrittled category. The surface cracks on specimens of metals in this

category were less numerous than for the slightly embrittled metals, but were

larger and easily visible to the naked eye. The density and size of the cracks

increased with the degree of plastic strain, i.e., there were few cracks located

outside the necked-down region and most were located close to the fracture.

It appears that a critical plastic deformation is required to initiate sur-

face cracking. This critical deformation could be associated with rupturing

surface protective layers to expose a metal surface to the hydrogen. Abrading

ASTM A-517 specimens in hydrogen prior to testing increased the number of sur-

face cracks formed, but plastic deformation was still required before surface

cracks were initiated.

Figure 63 illustrates surface cracking of ASTMASSS-B steel specimens tested

in I0 000-psig, 1000-psig, and 14.7-paPa hydrogen environments. The cracks were

sufficiently deep to initiate fracture at the higher hydrogen pressures. There

were indications of surface cracks on the specimen tested in 14.7-psia hydrogen,

although the cracks were not sufficiently deep to affect ductility; however,

shear lips were complete. It appears from Fig. 63 that the amount of plastic

deformation needed to form a surface crack increases with decreasing hydrogen

pressure.

The metallography of several severely embrittled specimens is illustrated

in Fig. 6_ Fracture of unnotched specimens generally involved formation of

several surface cracks as is illustrated in Fig. 64_A). These surface cracks

152
 (A) lOX

(B) lOX

(C) 10X
Figure 63. Photomicrographs (10X) of ASTM AS33-B Specimens Tested
in Hydrogen at Various Pressures ((A) 10 000-psig Hydrogen;
(B) 1000-psig Hydrogen; (C) 14.7-psia Hydrogen)

153
 2$0X

(^) lox

(s) lOX

.(G) llOOX
lOX

Figure 64. Metallography of Severely Embrittled Specimens Tested in i0 O00-

psig Hydrogen (Photomacrographs Showing Fracture Origins (A) AISI
1020 Unnotched Specimen; (B) ASTM A533-B Notched Specimen; (C)
Ti-6A1-4V (STA) Notched Specimen; Photomicrographs of Surface
Cracks in (D) AISI 1020 (E) AISI 430F; Electron Fractographs of
(F) AISI 1042 Notched Specimen at the Machined Surface/Fractured
Surface Interface and (G) Ti-6A1-4V (STA) Notched Specimen at
Region Adjacent to Machined Surface)

1S4
 appearedas small, half-moon-shapedareas normal to the surface and gave the
appearanceof radial growth of a crack which initiated at the surface. In many
cases, the surface cracks were not distinct becauseof crack overlap. Shear lips
were not formed on specimensin this embrittlement category.

The surface cracks of metals in the severely embrittled category were con-
siderably sharper than those found in slightly embrittled metals. Figure 64
showscracking in AISI 1020 (D) and AISI 430F stainless steel (E). The surface
cracks were somewhatsharper in the more embrittled AISI 430F stainless
steel.

Crack branching is extensive in the severely embrittled category. As sho_en

in Fig. 64(E), branching can cause the crack to changedirection toward the
tensile axis, which decreases the stress intensity at the crack tip, and causes
the crack to ceasepropagating.

The brittle region extends uniformly around the entire fracture periphery
in the notched low- and moderate-strength low-alloy steels tested in 10 000-
psig hydrogen as illustrated in Fig. 64(B). The brittle outer region corresponds

to the region of surface cracking of the unnotched specimens, and evidently,

several surface cracks formed the brittle periphery. If the fracture had pro-

ceeded from a single surface crack, the brittle region would have been restricted

to one side of the specimen. Figure 64(F) is an electron fractograph showing

the machined surface (far left side)/fractured surface interface of a notched

AISI 1042 specimen tested in 10 000-psig hydrogen. The brittle region is trans-

granular with secondary cracking (cracks into the fracture surface) that corres-

ponds to branching of the surrace cracking in unnotched specimens.

155
 The fracture appearance of commercially pure titanium and the titanium-

base alloy specimens was somewhat different from that of the other severely em-

brittled specimens. The brittle, hydrogen-affected region appeared as a dark

band (Fig. 64(C)), often of crescent shape, on one side of the fracture periphery.

There was no shear lip on the regions containing the dark band; however, there was

a thin, sometimes broken shear lip on the remainder of the specimen. The size of

the band appeared to be directly proportional to the degree of embrittlement. That

is, the specimens with the most strength reduction also had the largest dark band.

Figure 64(G) is an electron fractograph from the dark region (Fig. 64(C) of

the fracture of a Ti-6A1-4V (STA) notched specimen tested in 10 000-psig hydrogen.

The fracture is brittle and transgranular. The electron fractography did not show

any evidence of a second phase (hydride); however, it is difficult to detect the

hydride phase, and hydride formation is usually characterized by intergranular

failures (Ref. 69).

Williams and Nelson (Ref. 311 have examined Ti-6A1-4V specimens that were

bend tested in hydrogen at 14.7-psia pressure. In the solution-treated and aged

condition, the specimens were little affected by the hydrogen environment, and

the overall appearance of the fracture indicates primarily a shearing mode of

failure (Fig. 65(A). Figure 65(A) also shows small amounts of intergranular and

transgranular fracture and large amounts of transgranular secondary cracking.

Ti-6AI-4V, solution treated and stabilized to form a microstructure con-

taining equiaxed and some acicular e in a 8 matrix, exhibited a large degree of

embrittlement in hydrogen, Figure 65(B), a fractograph of the fracture, shows

numerous regions of intergranular fracture along both the acicular and equiaxed

- B boundaries.

156
 (A) STANDARD SOLUTION-TREATED AND AGED CONDITION

(B) SOLUTION-TREATED AND STABILIZED TO OBTAIN A HICROSTRUCTURE

CONSISTING OF EQUIAXED AND ACICULAR e IN A 6 MATRIX

Figure 65. Scanning Electron Fractographs of Ti-6AI-4V Specimens Bend Tested

in a 14.7 psi Hydrogen Environment, Ref. 31 (2200XI

157
 (C) SOLUTION-TREATED IN THE B FIELD AND STABILIZED TO OBTAIN
COMPLETELY ACICULAR _ IN A B MATRIX

Figure 65 (Concluded)

158
 Figure 65(C) shows the fracture of a specimen which had been solution treated

and stabilized to form completely acicular _ in a _ matrix. The prime mode of

failure was along prior _ grain boundaries and along acicular a - _ boundaries

with significant amounts of transgranular, secondary cracking. An interesting

feature common to fracture of this structure in hydrogen is the fan-shaped,

terraced structure which, accourding to Williams and Nelson _Ref. 31), is indi-

cative of a stepwise crack growth process.

Extremely Embrittled Specimens

Metals extremely embrittled by hydrogen, are likely to form only one surface

crack which propagates to failure. Figures 66(A) and (B) are end and side views

of an unnotched Inconel 718 specimen showing the surface crack that initiated

fracture. In each of the extremely embrittled metals, thin shear lips were

observed which were complete except for the location of the single surface crack.

Specimens of the alloys tested in air or i0 000-psig helium had complete shear lips.

Formation of surface cracks other than the one leading to failure was infre-

quent in the extremely embrittled alloys. Two surface cracks in an AISI 4140

specimen are shown in Fig. 66(C). The cracks appear to be fairly broad initially,

but then narrow and remain sharp during propagation. Typically, surface cracks

in the extremely embrittled metals are initially sharp and remain sharp or become

sharp during propagation. Branching does not significantly impede crack propaga-

tion in these materials.

Electron fractographic examination of notched specimens in this embrittle-

ment category showed the fractures to be very brittle near the region of fracture

origin. Steinman, Van Ness and Ansell (Ref. 50) have shown that the brittle

regions in notched AISI 4140 specimens tested in I0 000-psig hydrogen are charac-

terized by transgranular cleavage. Figure 67 shows scanning electron fractograp_s

159
 CA) lOX
(B) lox

(c) 3oox

Figure 66. Metallography of Extremely Enflarittled Specimens Tested in I0 000-

psi Hydrogen ((A) and (B) are Orthographic Projections of Un-
notched Inconel 718 Specimen Showing Surface Crack that Initiated
Fracture (Surface Crack is Dark Region Without the Shear Lip in
(A) and Flat Region in (B)); (C) is Photomicrograph of Surface
Cracks in an AISI 4140 Specimen)

160
 (A) Ni-4.5_o AI TESTED IN 14.7 PSl PRESSURE HYDROGEN, REF. 3l (160X)

(B) INCONEL 718 TESTED IN 2000-PSIG HYDROGEN, REF. 7O (750X)

Figure 67. Scanning Electron Fractographs of Nickel-Base Alloys Tested

in Hydrogen [The machined surface is at the right in (A),
and at the far left in (B).]

161
of Ni-4.S% A1 (A) tested by Williams and Nelson (Ref. 31) in 14.7-psia hydrogen

and of an Inconel 718 specimen (B) tested by Walter and Chandler in 2000-psig

hydrogen. The fractographs are from the machined surfacefractured surface

interface and show that the hydrogen-affected region is intergranular with deep

secondary cracking. The layer of fine grains shown on the Ni-4.S% A1 fractograph

was caused by deformation during machining and subsequent recrystallization dur-

ing heat treating. The depth of the brittle region and the sharp transition be-

tween the brittle and ductile regions are illustrated in the Inconel 718

fractograph.

Walter and Chandler (Ref. 70) measured the brittle crack depth for Inconel

718 as a function of hydrogen pressure. It was found that increasing the hydrogen

pressure increased the depth of the brittle fracture but did not change its appear-

ance. Figure 68 shows that the brittle crack depth was a linear function of

the square root of hydrogen pressure and extrapolated to zero at zero hydrogen

pressure. Embrittlement, as measured by reduction of notch strength, of these

specimens was also found to be a linear function of the square root of hydro-

gen pressure.

162
 B

==..

-,-4

I
_--oo
I
I

£0[ X "NI 'l-U.d](i _'111.RIB

163
 Fractography of Specimens Fatigue Tested inHydrogen

Hydrogen environments reduce the fatigue strength and enhance the rate of

crack growth in those metals that are susceptible to hydrogen-environment embrit-

tlement. Marcus et al (Ref. 71) have shown that hydrogen, even at pressures con-

siderably less than 14.7 psia, increases the rate of crack propagation for

tapered DCB specimens during high-cycle fatigue in Ti-6A1-4V annealed. The frac-

ture of a tapered DCB specimen tested in 150 torr hydrogen is illustrated in Fig.

69. Secondary cracks are quite numerous, but the orientation of the cracks is ran-

dom to the direction of crack propagation. Fatigue striations were not observed.

Electron fractography examination was made by Walter and Chandler (Ref.7

and 70) on several precracked notched specimens tested in low-cycle fatigue.

The results are presented in Table 25. The fatigue strengths of the ASTM A-517

and ASTM AS33-B specimen in 10 O00-psig hydrogen were about 1/3 that in 10 000-psig

helium. The Inconel 718 specimens were tested in 2000-psig helium and hydrogen

environments. The fatigue strength was 22 percent less in hydrogen than in helium.

Examination of the fatigue-induced fractures indicated close similarity

between the specimen tested in air and helium environments; however, the regions

of fatigue crack growth were considerably more brittle in the specimens tested

in hydrogen than in those tested in air and helium environments. In fact, the

hydrogen-affected regions appeared essentially the same in the tensile- and

fatigue-type failures. Secondly, fatigue striations were easily seen in those

specimens fractured in air and helium, but were not observed in the low-alloy

steel specimens fractured in 10 O00-psig hydrogen, and were only faintly seen

164
 G

(A) Electron Fractograph

(B) Scanning Electron Fractograph

Figure 69. Fracture Surface From Annealed Ti-6A1-4V Fatigued in 150 Torr
H2 Gas (Ref. 71)

165
 TABLE 25

RESULTS OF ELECTRON FRACTOGRAPHY EXAMINATON OF SPECIHENS

TESTED IN LOW CYCLE FATIGUE (Ref. 7 and 70)

Initial Cycles
Stress, to
Material Test Environment ksi Failure Fracture Description

ASTM A-517 10 O00-psig ttelium 206 438 Fatigue striations observable.

ASTM-A-517 117 191 Transgranular, brittle or quasi-

10 000-psig H2
cleavage. No fatigue striations
observable.

75 3 281 Transgranular, brittle or quasi-

AS_.I A533-B 10 000-psig H2
cleavage. No fatigue striations
observable.

Inconel 718 Air (la.7 psia) 118 3 576 Fatigue striations clearly
observable

Inconel 718 2000-psig Helium 132 4 575 Fatigue striations clearly

observable

77 8 780 Intergranular and transgranular

Inconei 718 2000-psig 112
brittle fracture; faint indica-
tions of fatigue striations.

Inconel 718 2000-psig H2 87 14 919 lntergranular and transgranular

brittle fracture; faint indica-
tions of fatigue striations.

. .

166
in the Inconel 718 specimens fractured in 2000-psig hydrogen. Figure 70(A) and

(B) illustrate striations formed in Inconel 718 in air and hydrogen environments.

Those in Fig. 70(B) are the most evident striations seen on the hydrogen

embrittled specimens.

These observations agree with those of Spitzig, Talda, and Wei (Ref. 64)

who studied crack growth during fatigue testing of 18 Ni (250) maraging steel

I/4-inch-thick center-notched plate specimens while exposed to hydrogen at 14.7

psia. They observed that hydrogen considerably increased the rate of crack growth

in this material. Fractographic examination revealed that the hydrogen-affected

regions were quite brittle without indications of fatigue striations.

The relative absence of fatigue striations on the specimens tested in

hydrogen environments suggests that the mechanism proposed by Laird and Smith

(Ref. 72) is the most applicable of those which have been postulated to explain

fatigue striations. According to these authors, the striations form from plastic

deformation at the crack tip. The hydrogen environment reduces fracture ductility

and thereby reduces striation formation.

Meyn (Ref. 73) suggested that the brittle striation regions produced during

a low-amplitude fatigue are predominantly the result of environmental reaction at

the crack tip, and presumably fatigue striations in low-amplitude fatigue would

not occur in vacuum. Pelloux (Ref. 74) has shown that fatigue striations are not

found in certain aluminum and titanium alloys during fatigue crack growth in vac-

uum environments. Pelloux theorized that fatigue striations occur because oxi-

dation prevents reversible slip, and reversible slip is not prevented in vacuum

environments. According to Meyn and Pelloux, an embrittling environment should,

167
therefore, delineate striations and accelerate crack growth which, of course,

is not in accordance with what is observed from fatigue failure in hydrogen. It

should be noted, however, that the fatigue striations may have been present but

obscured by the brittle nature of the fractures.

(_) 3ooox (_) 3ooox

Figure 70. Electron Fractograph of Fatigue Tested Inconel 718

Specimens _A) 3576 Cycles to Failure in Air;
(B) 8780 Cycles to Failure in 2000-psig Hydrogen)

168
 Walter and Chandler (Ref. 37), and Walter and Ytterhus (Ref. 78) tested

various coatings to determine their effectiveness as hydrogen barriers. The

coatings were applied to tantalum and Ta-10W and columbium and B-66 (columbium

alloy), sheet (1/2 x 2-1/8 x 0.03 inches) specimens. The specimens were heat

treated in purified hydrogen at atmospheric pressure. The exposure consisted

of thermally cycling the specimens three times between 800 ° and 1500°F, holding

at 800°F for 1 hour, and quenching in oil at room temperature. The total

elapsed time for each heat treatment was approximately 6-1/2 hours.

The results of these tests are listed in Table 26. The coatings that were

effective hydrogen barriers were vapor-deposited tungsten, Sylcor R-505, tin-

aluminum, and Sylcor R-508 silver-silicon. The vapor-deposited tungsten and

Sylcor R-505 coatings were only 0.001 inch thick. Carburizing B-66 by vacuum

heat treating at 2000°F (1093°C) to 2500°F (1371°C) in contact with carbon also

formed a protective hydrogen barrier. However, a surface network of intergrau-

lar carbides formed during carburization and subsequent heat treatment in hydro-

gen, and this carbon penetration caused a fairly low bend ductility (30 ° bend

angle) following hydrogen exposure.

Many of the coatings were not adequate hydrogen barriers because of cracking

in the intermetallic coatings and/or incomplete adherence of the coating to the

substrate. It was observed that complete adherence was necessary for a coating

to be an effective hydrogen barrier.

Fidelle et el. (Ref. 79) evaluated a series of coatings for protecting

high-strength steels from hydrogen embrittlement. Uncoated disks of the steels

were burst with hydrogen at room temperature. The effectiveness of the protec-

tive coatings was evaluated by subjecting the disks on the coated side to a

170
 CHAPTER
9. PREVENTIVE
MEASURES

Nearly all of the high-strength metallic materials investigated have been

found to be embrittled whenexposedto high-pressure hydrogen. Thus, there is a
severe limitation on the combination of properties for metals that can be used in
highly stressed structures in contact with high-pressure hydrogen. If existing

high-strength alloys are to be used to their full potential in hydrogen, it is

necessary that methods of prevention of hydrogen-environment embrittlement be

established.

Two basic methods by which this embrittlement can be prevented are the use

of protective coatings and the addition of inhibitors to the hydrogen. Despite

the importance, the literature is quite limited regarding methods for preventing

embrittlement.

Coatings

Most of the published literature on hydrogen barrier coatings deals with

the permeability of hydrogen through various metals and nonmetaIlic compounds.

Generally, those metals with low hydrogen solubilities and covalent and ionic-

bonded materials, such as oxides and carbides, have low hydrogen permeabilities.

The aluminum and silicon intermetallic compounds also have been found to be

hydrogen barriers. The available data (Ref. 75 through 7D indicate that the

following materials have low hydrogen permeabilities in order of increasing per-

meability: aluminum oxide (lowest permeability), tungsten, gold, various

glasses, tungsten and molybdenum silicides, molybdenum, nickel aluminide,

aluminum Solaramic enamel, copper, platinum, and nickel.

169
 TABLE26

CHARACTERISTICS
OFCOATINGS
AS HYDROGEN BARRIERS (Ref. 37 and 78)

FROM 800 ° TO 1SO0°F *

Substrate
Coating B-66" Ta-IOW Tantalum Eolumbium

Oxidized Surface Poor Good

Anodized Surface Poor IPoor

Carburized Surface Excellent Poor

Copper Strike-Nickel Strike- Good

Gold Electrolytic Deposition

Ni (0. OOl") W(0. 0005-0. 002") Poor Poor

Electrolytic Deposition

Tungsten (Vacuum Deposited) Promising Poor

Tungsten (Vapor Deposited) Excellent Excellent

Platinum Poor Poor

Aluminized (Los Angeles Div.) Promising Poor

Aluminized-Siliconized Surface Poor Poor

Aluminized (Atomics Int'l.) Poor Promising

Tin-Aluminum R-505 Sylcor Coating Excellent Excellent Excellent

Silver-Silicon R-508 Sylcor Coating Exce 1 lent

Siliconized Poor Promising

W (Vacuum Deposited) and Siliconized Poor

Promising.]

*Barriers evaluated by measuring hydrogen absorbed in specimen substrate.

Coating resistance ranked as follows:

Excellent: essentially no hydrogen absorption at 800°F heat treatments

Good: measurable hydrogen absorption but the amount absorbed would not be
expected to appreciably affect the mechanical properties

Promising: the amount of hydrogen absorbed considerably less than would be

absorbed without the coating (without improvement, the coating
would probably not be an adequate barrier)

Poor: no significant reduction in hydrogen absorption due to the coating

171
 hydrogen pressure which would lead to immediate failure in the absence of the

coating. The effectiveness of the coatings was evaluated by observing the time

to failure. The influence of surface conditions on embrittlement was also eval-

uated. It was found that a rough surface accentuated embrittlement, and that

surface compression by shot peening decreased emhrittlement.

Fidelle and co-workers found that protection from high-pressure hydrogen

was obtained with gold, aluminum, cadmium, and, to a lesser extent, tin, but

not with nickel, zinc, chromium, or copper. Rupture of the protected disks

occurred after several months' exposure and after absorption of approximately

l-ppm'hydrogen in the steel.

Fidelle et al. (Ref. 7_ assumed that the failure of the coated disks was

caused by an internal-hydrogen embrittlement delayed failure mechanism. It is

possible, however, that the time delay was that needed to obtain sufficient

quantities of hydrogen at the coating/substrate interface to cause hydrogen-

environment embrittlement.

The disks that were uncoated or coated with a nonprotecting coating failed

during application of pressure or soon after the pressure was established.

Coatings were not protective if they contained microscopic porosity and lacked

good adherence. The nickel coating was ineffective because it was embrittled

by high-pressure hydrogen. Chromized coatings were evaluated because of the

protective chromium oxides. The chromized surfaces were brittle, however, and

failure occurred at hydrogen pressures about 1/2 the hydrogen burst pressure

of uncoated disks.

172
 The Sylcor R-505 tin-aluminum and vapor-deposited tungsten coatings (which,

as indicated above, are barriers to hydrogen) were further evaluated by Walter

et al. (Ref. 37 and 78) to determine their effectiveness for preventing embrit-

tlement of B-66 and Ta-10W during tensile testing in hydrogen. Coated sheet

(0.003-inch thick) tensile specimens were exposed to flowing hydrogen at 14.7-psia

pressure. When the test environment and temperature had been established, a ten-

sile stress of S0 percent of the room temperature yield strength was applied.

This stress was maintained for 30 minutes, then the specimens were tensile tested

to failure. The results indicated that the Sylcor R-SOS tin-aluminum coating was

not particularly effective for preventing embrittlement at room temperature and

400°F. The change of hydrogen barrier characteristics at 400°F was associated

with the 442°F tin-aluminum eutectic temperature. Below that temperature, the

coating was not self-healing to cracks present in the aluminide intermetalli_

diffusion zone. These cracks may have formed during application and solidifica-

tion, because cracking and absorption of hydrogen occurred at ambient temperature

even in undeformed specimens.

Similar tensile tests were performed at room temperature on vapor-deposited

tungsten-coated B-66 and Ta-10W specimens. The tungsten coating was not ductile

at this temperature, as evidenced by flaking of the coating during tensile defor-

mation and, as a result, this coating also was not protective.

Experiments are being conducted at Rocketdyne for developing a technique to

protect electroformed nickel during ambient-temperature exposure to 1200- and

7000-psig hydrogen. The results indicate that copper and gold coatings prevent

embrittlement of unnotched specimens tensile tested in these hydrogen environ-

ments. The coatings, however, gave limited protection to notched specimens.

173
 The investigations that have been performed to prevent embrittlement by

gaseous hydrogen indicate that protective coatings must have the following

characteristics:

1. Low hydrogen permeability

2. Nonporous, i.e., free of defects

3. Completely adherent to the substrate

4. Ductile or self-healing

Coatings that have shown the most potential for preventing hydrogen-

environment embrittlement are copper, gold, cadmium, and perhaps the Sylcor

R-508 silver-silicon coating and, for elevated-temperature service, the Sylcor

R-505 tin-aluminum coating.

Impurity Additions to Hydrogen

Hofmann and Rauls (Ref. 42) performed a series of tests to evaluate the

influence of air, nitrogen, oxygen, and argon on embrittlement by gaseous hydro-

gen. The tests were performed at ambient temperature on CK 22, a 22-percent

carbon, plain carbon steel in 1470 psia hydrogen.

Their results are shown in Fig. 71 and 72. Additions of argon and purified

nitrogen to hydrogen did not reduce embrittlement. However, the addition of

5-percent nitrogen (not shown in Fig. 71), which was not prepurified, partially

eliminated embrittlement. It was suggested that oxygen impurities in the nitro-

gen were responsible for the embrittlement reduction. The minimum oxygen content

needed to initiate reduction of embrittlement by the hydrogen environment was

about 10 -4 volume percent (I ppm) and embrittlement was completely eliminated

by 1-percent oxygen.

174
 o,,,o

o
d_

'
N
40

o
]'

]-- "
i

_ ......
5
I........"- [-']NITROGEN_

_."' I00_ 40 20 I0 5 2 I 0.5 0.2 100%

AIR FOREIGN GAS PORTION IN HYDROGEN, PERCENT H2

Figure 71. Effect of trace impurities in hydrogen under

lO0-atm partial pressure on the elongation
and reduction in area of CK 22 N steel.

_: 20 ,_-
Z
0
____i_Z_- 'I" I 'Ill ' I !_
u.I

A_I('H LIJJf
I !I]
IIIIIII
l__',i I0' 6,'I I! IIHI
"' 40 i IIII
"_ III] II IIII I III I I III-h'-'713 IIIIIII_
o 20 I I IIII I IIII I IIII I Ill
_. IJll , I I IIII I IIII I IIII I III I I IIII J ___.
" o IllJ
--
I-.-
Jill I II IIII I IIII I IIII I III I1 IJlJ I ___,
o lO0_ I0 I 10 -I I0 -2 10 -3 I0" 10 .5 100%
" AIR
I,l.,I H2
=C OXYGEN CONTENT IN HYDROGEN,PERCENT

Figure 72. Effect of oxygen additions to hydrogen under

lO0-atm partial pressure on the elongation
and reduction in area of CK 2"2 N steel.

175
 At the other extreme, Walter and Chandler (Ref. 7 ) showed that measurable

reductions in strength and ductility of notched high-strength steels and nickel-

base alloys occurred during tensile tests conducted in I0 O00-psig helium con-

taining 44-ppm (0.44-psi) hydrogen. Thus, hydrogen-environment embrittlement

is a function of the partial pressure of the hydrogen and is not affected by

dilution with inert gases.

In a qualitative manner, investigators have observed that hydrogen-environ-

ment embrittlement is reduced when impurities are present in the hydrogen envir-

onment. Williams and Nelson (Ref. 31) have performed tests in low-pressure

(_14.7'psia) hydrogen on iron, nickel, and titanium-base alloys and have observed

that impurities affect the degree of embrittlement of titanium alloys more than

the other two. Similarly, Walter and Chandler (Ref. 7) observed that embrittle-

ment of Ti-6AI-4V during tensile testing in 10 O00-psig hydrogen will occur only

if the test cells are evacuated and purged several times prior to testing. On

the other hand, purging by pressurization/depressurization without evacuation is

sufficient to obtain near-maximum embrittlement of iron and nickel-base alloys

by 10 O00-psig hydrogen (see Chapter 5, Test Procedures).

Hydrogen-environment embrittlement o£ high-strength steels appears to be

least influenced by hydrogen impurities. Sawicki and Johnson (Ref. 80) found that

embrittlement of high-strength steels by hydrogen at 14.7-psia pressure is not

influenced by oxygen contents of less than 200 ppm. This relatively high oxygen

contamination level suggests that the inhibiting influence of oxygen is a func-

tion of its absolute partial pressure regardless of the total pressure of the

system. That is, the higher the hydrogen pressure, the less the oxygen content

required to decrease or eliminate the embrittling influence of hydrogen.

176
 CHAPTER
i0. DISCUSSION

General

As described in previous chapters, a wide variety of metals and alloys are

embrittled by hydrogen environments. The embrittlement increases with increasing

hydrogen pressure and occurs over a range of temperatures, but is most severe in

the vicinity of room temperature. The general characteristics of hydrogen-

environment embrittlement are: (i) the effect is one of embrittlement while in

the hydrogen environments; (2) it is a surface effect; and (3) it is an immediate

effect.

The elastic properties, yield strength and, in many cases, the ultimate

tensile strength of unnotched specimens are the same in hydrogen as in air or in-

ert environments. The most significant effects of the hydrogen environments are

on tensile ductility, notch strength, and crack behavior. More precisely, the

effect of the hydrogen environment is to embrittle the surface. The embrittled

"surface" may be the immediate surface or a surface layer of some finite but

limited thickness. The metal surface in hydrogen cannot undergo plastic deforma-

tion to the same degree as it can in air or inert environments or, for that matter,

as can the interior of the metal whose surface is in contact with hydrogen. When

a susceptible metal is stressed in tension in hydrogen to some critical amount of

plastic deformation, the surface fractures, i.e., a surface crack forms, while

the interior of the metal is still capable of undergoing considerable additional

deformation. This surface cracking provides some of the most dramatic visual

evidence of hydrogen-environment embrittlement. For a susceptible metal, an

177
existing crack or one formed in the hydrogen environment will propagate at a

lower stress intensity and at a more rapid rate in hydrogen than in air or inert

environments.

The effect of the hydrogen environment is, for all practical purposes, immed-

iate, i.e., as soon as a susceptible metal is placed in the hydrogen it is

embrittled. No hold time in the hydrogen environment is required. This does not

mean that failure is immediate. Time to failure obviously will depend on the

type and rate of stress application and other factors. But, as soon after exposure

to hydrogen as the required critical surface deformation can be applied, the sur-

face will crack. The actual formation of the crack will take a finite but very

short time. Also, propagation of any existing crack will be more rapid as soon

as the hydrogen contacts the true metal surface. Crack propagation will be com-

plicated by the presence of surface layers, such as self oxides, which will have

to be broken to establish the hydrogen/metal contact. The extremely short time

required to establish the effect of the hydrogen environment is not inconsistent

with the fact that the degree of hydrogen-environment embrittlement is greater at

slower strain rates and may not occur at rapid strain rates. This results simply

from the fact that the hydrogen embrittling effects of surface crack formation

and crack propagation take a certain time and, at very rapid strain rates, they

do not have time to occur before failure due to tensile overload.

The amount of hydrogen-environment embrittlement of susceptible metals is

highly influenced by the nature of the surface cracking and crack propagation in

hydrogen. Extreme embrittlement is associated with the formation of a sharp sur-

face crack(s) and with the crack(s) remaining sharp during propagation. Lesser

178
degrees of embrittlement are associated with the formation of surface cracks which

are blunt and/or the blunting or branching of cracks during propagation.

Hydrogen-environment embrittlement is an environmental effect and occurs

only while the metal is in contact with hydrogen. A metal that is exposed to

hydrogen but then removed from the hydrogen is nct subject to hydrogen-environment

embrittlement. It may not seem necessary to dwell on the environmental nature of

hydrogen-environment embrittlement, except that the thinking on this embrittlement

tends to be colored by the long-standing familiarity with, and the massive litera-

ture on, embrittlement resulting from hydrogen charging. Thus, questions continu-

ally arise as to the required time of exposure and many tests have been performed

on specimens soaked in hydrogen, even at ambient temperatures, but removed before

testing. Obviously, exposure to hydrogen can result in other forms of hydrogen

embrittlement subsequent to removal from the hydrogen. Depending on the pressure,

temperature, and time of exposure, sufficient hydrogen could be absorbed to result

in internal-hydrogen embrittlement or the hydrogen could react with the metal to

form a hydride or with carbon or oxygen to form high-pressure gas pockets and

thus embrittle the metal even after it is removed from the hydrogen. Embrittle-

ment due to previously absorbed hydrogen may add to, or synergistically interact

with, hydrogen-environment embrittlement.

Tensile Tests

Much of the information on the characteristics of hydrogen-environment embrit-

tlement has come from tensile tests performed on unnotched and notched specimens in

hydrogen at ambient temperatures. The results of such tests will be considered in

this section to clarify or expand on some of the characteristics noted above and

to assist in understanding the meaning of the data obtained from these tests.

179
 As noted, no effects of hydrogen environments at ambient temperatures on

elastic properties or yield strength have been found. The deformation capability

of the metal surface is lower in hydrogen than in air or inert environments. The

interior of the metal is, of course, unaffected by the environment. Thus, as the

tensile test proceeds, a critical surface deformation is reached at which the sur-

face fractures in a brittle manner and a surface crack is formed. The most direct

evidence for the environmental nature of the embrittlement in hydrogen is the

brittle appearance of the fracture at the surface of unnotched and notched speci-

mens (as discussed in Chapter 8, Metallography) and the cracks formed on the sur-

face in deformed regions of unnotched specimens of susceptible metals. In

notched specimens, the region of highest deformation is so localized at tile base

of the notch that surface cracks, other than those resulting in failure, are nor-

mally not found.

From the time at which surface cracking begins, the test is no longer a

normal tensile test but is a rather complex test of a cracked specimen and this

must be considered in assessing the meaning and usefulness of the strength and

ductility data obtained from these tests. In alloys that are placed in the

extremely embrittled category (Table 8) the first surface crack to form propagates

to failure, usually without the formation of additional surface cracks. This

crack propagates at some finite rate that is more rapid in hydrogen than it would

be in air or an inert environment. Because the test is being conducted as a ten-

sile test, and the load is continuing to be increased, and because of the deepen-

ing of the crack, the stress intensity in front of the propagating crack is in-

creasing rapidly during the test. The crack propagates until the remaining area

fails due to tensile overload. The final catastrophic failure is a ductile fail-

ure apparently unaffected by the hydrogen environment. It can be seen that the

180
"ultimate tensile strengths" and "ductilities" determined from such tests on

unnotched and notched specimens of metals that are extremely embrittled in hydro-

gen are determined by the load or strain rate, the time (plastic deformation) at

which the surface crack is formed, and the rate of propagation of the crack in

hydrogen. Thus, these properties are far more sensitive to load or strain rates

than are tensile properties determined in air or inert environments.

For alloys in the severely or slightly embrittled categories (Table 8) the

first surface crack to form does not propagate to failure. The surface cracks in

these alloys are blunt as formed or tend to blunt or branch during propagation in

hydrogen. The first surface crack forms, propagates in a brittle manner to some

extent, blunts or branches, and stops. A second crack forms and the process con-

tinues until, finally, the remaining area has been reduced by the series of sur-

face cracks (e.g., AISI 1020, Fig. 64) so that catastrophic, ductile, tensile

overload failure occurs. The properties determined from these tests will be a

function of the same factors as listed in the above paragraph for the extremely

embrittled metals, and also will be a function of the extent of crack propaga-

tion before the crack is stopped by blunting or branching.

From the above discussion, it can be inferred that the different tensile

properties will show differing degrees of sensitivity to the hydrogen environments.

The degrees of sensitivity in decreasing order are: fracture mode (effects on

the appearance of the fracture have been noted when only negligible change of

properties occurred), reduction of area of notched specimens, ultimate strength

of notched specimens, reduction of area of unnotched specimens, elongation of

unnotched specimens, and ultimate strength of unnotched specimens. Only for

181
metals in the extremely embrittled category is there a reduction of ultimate

tensile strength of unnotched specimens in hydrogen. The fact that this property

is not affected by hydrogen for metals in the other embrittlement categories

simply reflects the relationship noted above between rate of loading (straining)

and the time for crack initiation and extension, remembering that crack blunting

and branching slow crack growth considerably in these metals. In fact, it appears

that in most cases, surface cracks do not form until the ultimate tensile strength

is reached, i.e., necking begins.

Although the ultimate strengths and ductilities determined from tensile tests

in hydrogen on unnotched and notched specimens of susceptible materials are of

limited direct usefulness in design, these tensile tests are valuable as rapid

and economical screening tests which provide a good indication of the relative

degrees of embrittlement of different metals by hydrogen environments, e.g., as

shown in Table 8. For design purposes, the tensile properties determined in

hydrogen can be used as a guide for establishing additional safety factors to

compensate for hydrogen-environment embrittlement.

Surface Cracking in Hydrogen

Metals susceptible to hydrogen-environment embrittlement undergo surface

cracking when stressed in hydrogen while nonsusceptible metals do not. The fact

that failure of susceptible metals in ambient temperature hydrogen initiates with

brittle surface cracking (as discussed in Chapter 8, Metallograph_is convincing

evidence of the environmental nature of the embrittlement (Ref. 7). Failure ini-

tiates with brittle surface cracking in other forms of environmental embrittlement

i.e., stress corrosion and liquid metal embrittlement. Without environmental

effects, tensile failures initiate inside the metal in the region of highest

182
triaxial stress and the fracture'propagates back to the surface in a ductile man-

ner forming the typical ductile shear lip. This behavior holds even for metals
embrittled by charging with hydrogenbut tested in air. Troiano (Ref. 6), by
optical metallography, and Phillips, Kerlins, and Whiteson (Ref. 81), by electron
fractography, showedthat in hydrogen-chargedspecimenstested in air, brittle
cracks formed inside the metal and propagated inward in a brittle manner. The
fracture finally propagated back to the surface in a ductile mannerand shear lips
were formed.

All results to date showthat surface cracking in hydrogen does not begin
until the yield strength has been exceededand plastic deformation has occurred.
No investigations have been reported in the literature of the amountof plastic
deformation which takes place before surface cracking begins. It has been

postulated (Ref. 52) that a critical plastic deformation is required to initiate

surface cracking in hydrogen and the important deformation is, of course, that at
the surface.

For hydrogen to cause surface cracking, it seemsaxiomatic that the hydrogen

must be brought to the metal surface and adsorbed, dissociated, and possibly
absorbed into the metal surface. If the metal has its normal oxide layer, the
hydrogen can reach the actual metal surface either by diffusing through the
oxide or by making direct contact with the surface after the oxide is ruptured.
Most metal oxides are good barriers to hydrogen and the diffusion of hydrogen
through the oxide is rather slow, particularly at ambient temperatures. The
amountof hydrogenwhich could reach the metal surface by diffusion through the
oxide is believed too small to cause significant embrittlement for most of the
tensile tests performed in hydrogenbecause of the short exposure times. Thus,
it appears reasonable that the critical surface deformation for initiation of

183
surface cracking in hydrogen is, at minimum, that surface deformation required

to rupture the oxide layer so that the hydrogen can contact the metal surface.

As discussed in Chapter 8, Metallography, specimens tensile tested in hydrogen

after abrasion in hydrogen to remove the oxide layer had a much larger number

of surface cracks than specimens abraded in air before testing in hydrogen

(Ref. 7). The amount of plastic deformation at crack initiation was not deter-

mined. However, a certain amount of plastic deformation of the metal surface

still was required before surface cracking initiated in specimens abraded in

hydrogen to remove the oxide before beginning the tensile test.

The abrading tests were performed on ASTM A-517, a steel placed in the

severely embrittled category. The more numerous surface cracks resulting from

abrading in hydrogen did not result in a greater degree of embrittlement as

measured by strength and ductility. This is attributed to the crack blunting

that occurs in such metals and results in slower crack propagation which then

determines the degree of embrittlement more than does crack initiation. It

should be cautioned that abrading in hydrogen of the metals in the extremely

embrittled category may increase significantly the amount of embrittlement be-

cause crack blunting does not occur in these metals.

The surface deformation required for surface crack initiation in hydrogen is

undoubtedly a function of many factors. The following are suggested: the metal,

metal condition (e.g., heat treatment and fabrication method), the nature of

the oxide (how adherent and protective it is, and how able it is to deform with

the metal), temperature (effects on metal, oxide, and rate of adsorption, dis-

sociation, and absorption of hydrogen into metal surface) hydrogen pressure

(higher pressures may be able to increase rate of movement of hydrogen through

protective surface layers and rate of hydrogen adsorption, etc.), prior surface

184
work hardening, surface preparation (surface finish, chemical polishing, oxide

removal, vacuum treatment, etc.), stress concentration factor (required surface

deformation may be different at base of notch than on unnotched specimen),

strain or load rate, crystal plane exposed, direction of loading relative to

direction of prior cold working, and hydrogen purity. There may be others.

Crack Propagation in Hydrogen

All metals susceptible to hydrogen-environment embrittlement undergo sur-

face cracking in hydrogen, but the degree of embrittlement in tensile tests is

determined by the rate of crack propagation and the tendency for the crack to

blunt or branch to reduce the applied stress intensity. Metals in the slightly

embrittled category form very rounded cracks which tend to propagate very little,

and effects on properties due to hydrogen are small. In fact, in pure titanium,

the effect of the very shallow, very rounded surface cracks is so small that a

ductile shear lip is formed. Metals in the severely embrittled category form

sharp surface cracks, but they tend to branch. The metals in the extremely

embrittled category form sharp surface cracks and these tend to remain sharp

and not branch as they propagate.

As discussed in Chapter 7, Other Mechanical Properties, crack behavior in

hydrogen has been investigated with precracked specimens for a number of metals.

In a few instances, the tests were designed to develop valid fracture mechanics

data and were performed in high-pressure hydrogen. However, most tests on pre-

cracked specimens have been conducted to develop more fundamental information on

the effect of various parameters on crack growth rates in hydrogen. Almost all

185
of these latter tests were performed in hydrogen at pressures near 14.7 psia and

below and at temperatures near room temperature. As to be expected, crack growth

rates were considerably faster in hydrogen than in air or inert environments for

materials classified from tensile tests as extremely or severely embrittled

(Table 8). Crack behavior of an aluminum alloy and a copper alloy was found to

be essentially unaffected by hydrogen environments, which is consistent with the

negligible embrittlement found for aluminum- and copper-base materials from ten-

sile tests in hydrogen.

The fracture toughness parameter, Kic, is generally considered to be a

property of a given material and not affected by the environment, including hydro-

gen environments. Direct evidence that KIc is not affected by hydrogen environ-

ments, particularly high-pressure hydrogen environments, is very limited. From

crack growth studies in hydrogen at pressures near 14.7 psia, Sawicki and Johnson

for fi-11 tool steel (see Appendix) and Williams and Nelson for AISI 4130 steel

(Ref. 8) have rather definitive evidence that KIc is not affected by hydrogen

environments at these pressures. Bixler (Ref. 28) conducted tests on surface-

flawed Inconel 718 and Ti-SA1-2.SSn ELI specimens in 1400-psig hydrogen at ambient

temperatures, and concluded that Kic was not affected by the environment. However,

the evidence is somewhat ambiguous since valid Kic values were not obtained. Thus,

it is believed that additional experiments are required to determine if KIc is

affected by high-pressure hydrogen environments. A program including such tests

is in progress (see Appendix). The definition of KIc is important in this respect.

A Kic determined using the secant intercept method may well be affected by a high-

pressure hydrogen environment and the time to perform the Kic test may be more

critical in hydrogen than in air or inert environments. In the determination of

186
Kic , it is important that the precrack be extended by cycling at low stress levels

in hydrogen before beginning the Kic test. Otherwise, the initiation of crack

movement in hydrogen would be a function of the rupturing of the surface oxide

layer. Extension of the crack by fatiguing in hydrogen brings the added compli-

cation that the amount of this extension will be difficult to measure visually or

metallographically because there is little difference in the appearance of frac-

ture surfaces produced by tensile loading or fatiguing in hydrogen (see Chaoter 8,

Metallography).

As expected, considerable effects of high-pressure hydrogen on Kth, the thres-

hold stress intensity for sustained-load flaw growth, have been found. Ratios of Kth

in high-pressure hydrogen at 70 F to Klc in air were shown to be less than 20 per-

cent for both Inconel 718 (Ref. 28 and 62) and Ti-5AI-2.5Sn ELI (Ref. 28). Con-

versely, this ratio was found to be about 80 percent for the 2219-T6E46 aluminum

alloy. Sawicki and Johnson (see Appendix) have found that Kth for H-II tool steel

in hydrogen decreases with increasing pressure for pressures near 14.7 psia. As

with KIc tests, it is important in Kth tests that the precrack be fatigue extended

in hydrogen before the start of the Kth test. As an example, if the bolt-loaded

modified WOL specimen is used, the stress intensity required to initiate crack

movement in air (or even in hydrogen but with the oxide present) may be large

enough to lead to specimen failure in the hydrogen. Even if failure does not

Occur, the crack will extend much further, which leads to a greater probability

that the crack will branch from the plane of crack growth which complicates Kth

determinations. This is especially true because, in some steels, the tendency

for crack branching is greater in hydrogen than in inert environments (Ref. 7).

187
 As noted in Chapter 7, Other Mechanical Properties, crack growth rates have

been found to be considerably greater in hydrogen environments, even at very low

pressures, than in air or inert environments for various high-strength steels,

pure nickel, nickel-base alloys, and titanium-base alloys, but not for a copper-

base alloy. These investigations were conducted in almost all cases with hydrogen

pressures of 14.7 psia and below, in most cases with sustained rather than cyclic

loading, and were designed to develop information on the nature of hydrogen-

environment embrittlement and parameters affecting it.

Williams and Nelson (Ref. 8) investigated sustained load, slow crack

growth of A'ISI 4130 steel in hydrogen at pressures below 14.7 psia, and at

temperatures from approximately -122 F (-80°C) to 176°F (80°C). They found the

crack growth rate to be a function of l/T, where T is in °K, to be a maximum near

room temperature, and to be a function of p1.5

H2 .at higher temperatures and p0.5H2 at

lower temperatures.

Sawicki and Johnson (see Appendix) found a similar temperature relationship

for H-11 tool steel. To explain their results, Williams and Nelson developed a

hypothesis that a thermally activated chemisorption process is rate controlling

for crack growth in hydrogen. Their theory predicts that crack growth rates would

be a function of p0.5
H2 at high pressures.

The investigations of Nelson et al. (Ref. 65) on crack behavior of AISI 4130

steel in partially dissociated hydrogen at low pressure showed that the presence

of the atomic hydrogen in the environment considerably increased crack growth

rates and altered the kinetics. They postulated that the presence of atomic

188
 hydrogen served to bypass the chemisorption rate-controlling process discussed

above and the rate-controlling process then became the absorption of the atomic

hydrogen into the metal.

Effect of Hydrogen Environments on Other Properties

Hydrogen environments most severely degrade the properties of precracked

specimens and thus have the large effects on Kth and crack growth rates described

in Chapter 7, Other Mechanical Properties. As discussed above, with smooth speci-

mens, the most severe property degradation will occur in tests involving plastic

strain. Thus, the considerable reductions of stress-rupture strengths and

cycles-to-failure in strain cycling tests in hydrogen are not surprising.

High-cycle fatigue tests in hydrogen have not been reported although some are

planned (see Appendix). However, it is expected that the effect of hydrogen'

environments on high-cycle fatigue will not be as significant. In high-cycle

fatigue, the strain per cycle is small, and thus the effect of hydrogen on the

number of cycles to crack initiation may be small and the main effect of the hydro-

gen would be on crack propagation. But generally, in high-cycle fatigue, approxi-

mately 90 percent of the time-to-failure is occupied with crack initiation.

Metals Susceptible to Hydrogen-Environment Embrittlement

A wide variety of metals, which unfortunately includes almost all of the

high-strength structural alloys, are susceptible in some degree to hydrogen-

environment embrittlement. Iron, nickel, and titanium and alloys based on these

metals a ad on cobalt have been found to be embrittled by hydrogen environments.

Thus, susceptible metals include those with body-centered-cubic, face-centered-

cubic, and hexagonal structures. Metals not embrittled are copper and aluminum

189
and their alloys, and the most stable stainless steels, i.e., those that remain

austenitic and do not transform to martensite during deformation.

The only apparent common denominator separating susceptible and nonsus-

ceptible metals is that all metals found to be susceptible to date are transi-

tion elements and their alloys. Metals not susceptible are not transition

metals, the exception being the austenitic stainless steels. In this regard,

it is simply noted that magnetic behavior is associated with transition metals,

and the ferromagnetic behavior of iron alloys is not present in the austenitic

stainless steels, which thus have lost one of the characteristics of transition

metal behavior. The less stable stainless steels tend to partially transform

to martensite during deformation and, where martensite forms on the surface,

surface cracking occurs in hydrogen (Ref. 46).

Within the group of metals susceptible to hydrogen-environment embrittlement,

the degree of embrittlement is not particularly predictable. In general, higher

strength alloys are sore embrittled than lower strength alloys. However, one of

the most embrittled metals is electroplated nickel, as plated. Also, in an inves-

tigation of three pressure vessel steels, the lowest strength one, a plain carbon

steel, was the most embrittled in tensile tests (Ref. 52). The difference between

yield and ultimate strengths was small in the two higher-strength steels, but rel-

atively large in the lower-strength steel. In accord with the earlier discussions

of surface cracking, the larger reduction of properties in hydrogen for the lower-

strength steel was attributed to the initiation of surface cracks at a stress which

was low relative to the ultimate tensile strength.

Different heats and different heat treatments for a given alloy can show

significant dlffe_ences in the degree of embrlttlement by hydrogen environments.

I_ the most striking instance, the reduction of notch sCrength of Inconel 718 in

190
1000-psig hydrogen at ambient temperatures was 40 percent for one heat-heat

treatment combination and only 10 percent for a different heat and heat treat-

ment. There was no obvious compositional or structural difference to which the

difference in embrittlement could be attributed, but the less embrittled mater-

ial did have the higher ductility and notch strength/unnotched strength ratio.

In general, for a given alloy, heat treating to a higher strength level results

in a greater degree of hydrogen-environment embrittlement.

As noted in the earlier discussion of surface crack formation and propaga-

tion, both the nature of the surface crack formed and the way in which it propa-

gates play a crucial role in determining the degree of hydrogen-environment

embrittlement of a material. The most extremely embrittled materials form sharp

cracks which remain sharp during propagation in hydrogen. Lesser degrees of em-

brittlement are associated with the initial formation of blunt cracks and with

crack blunting or branching during propagation.

Effect of Various Parameters on Degree of

Hydrogen-Environment Embrittlement

Hydrogen Pressure.-- The degree of hydrogen-environment embrittlement in-

creases with increasing hydrogen pressure. However, sizable effects, e.g.,

increases in crack growth rates, occur in hydrogen at pressures below 14.7 psia.

From their investigations of slow crack growth rates of AISI 4130 steel in hydro-

gen environments at pressures of 14.7 psia and less, and at temperatures near

room temperature, Williams and Nelson (Ref. 8) found that the crack growth rates

were a function of p0.S

H2 at temperatures below room temperature, and of
pi.5
H2 at temperatures above room temperature. From their hypothesis of a

191
rate-controlling, thermally activated chemisorption process, Williams and Nelson

further concluded that crack growth rates are a function of p0.5

H2 at high pressures
pl.5
and/or low temperatures, and a function of H2 at low pressures and/or high

temperatures.

As discussed in Chapter 6, Tensile Properties, an apparent relationship

was found between the tensile properties of ASK! AS33-B steel and Inconel 718 in

hydrogen and p0.5 (Ref. 47 and 70). This may be explained as follows if one
H2
accepts the hypothesis of Williams and Nelson. Tensile tests of susceptible

materials in hydrogen are ultimately complex tests of cracked specimens and the

proper£ies measured are a function of the rate of testing, the plastic deforma-

tion at which surface cracks initiate, and the rate of crack growth. The same

rate of testing was used for the tests at the different pressures. If the

amount of plastic deformation at which surface cracks initiate is relatively

insensitive to hydrogen pressure, which does not appear to be an unreasonable

assumption for the oxide-covered specimens, then the measured tensile properties

will essentially be a function of crack growth rates and, from Williams and

Nelson, p0.5 Such a relationship may not hold for specimens with a clean metal
H2 "
surface unless the time, i.e., critical plastic deformation, for surface crack

initiation is also a function of p0.5

H2 or insensitive to pressure, which seems

unlikely.

In tensile tests on electroformed nickel at pressures from 14.7 to 7000

psig, the degree of embrittlement in hydrogen increased with increasing pressure,

0S
but was a function not of PH2" , but more nearly the in PH2. With either rela-

tionship, the increase in the degree of embrittlement with a given increase of

pressure is much greater at low pressures than at high pressures.

192
It has been suggested (Ref. 31) that the effect of hydrogen pressure is more

significant for normal surfaces, i.e., surfaces covered with oxide and adsorbed

gases than for clean metal surfaces. It may be, especially for clean surfaces,

that the effect of pressure will reach a maximum and embrittlement will not

increase significantly above a certain high pressure.

Temperature.- There have been only very limited investigations of the rela-

tionship between hydrogen-environment embrittlement and temperature. However,

all the data obtained to date indicate that the degradation of mechanical prop-

erties, e.g., tensile, fatigue, or crack growth rate, by hydrogen environments

is a maximum near room temperature. It also appears that the degree of embrittle-

ment decreases more rapidly as the temperature is lowered below room temperature

than when it is raised an equivalent amount above room temperature. It should be

noted that the hydrogen environments considerably reduce the cycles-to-failure

in strain cycling tests at temperatures as high as 1250°F (677°C)(Ref. 61). As

discussed earlier, Williams and Nelson (Ref. 8) determined the effect of temper-

ature on the sustained load, slow crack growth rate of AISI 4130 steel in hydro-

gen at pressures below 14.7 psia and temperatures from approximately -II2°F (-80°C)

to 176°F (80°C_. They found the crack growth rate to be maximum near room tem-

perature and to be a function of I/T (°K) at lower and higher temperatures. The

temperature at which the maximum crack growth rate occurred increased with pres-

sure. A similar relationship between crack growth rate and temperature was

found by Sawicki and Johnson (Ref. 80) for H-II tool steel in a similar tempera-

ture/pressure regime. It might be expected that surface crack initiation is a

function of temperature, but no investigations have been conducted.

193
 Hydrogen Purity.-- The degree to which a metal will be embrittled by a hydro-

gen environment is a function of the purity of the hydrogen, particularly with

regard to oxygen content. Hofmann and Rauls (Ref. 42) found that the minimum

oxygen content at which a reduction in the embrittlement of a plain carbon steel

by 1470-psig hydrogen occurred was approximately 10 -4 volume percent (1 ppm) and

embrittlement was completely eliminated by 1-percent oxygen. Conversely, Sawicki

and Johnson (Ref. 80) found that crack growth rates of H-11 tool steel in 14.7-

psia hydrogen were not influenced by oxygen contents of less than approximately

200 ppm. Although comparisons between these two sets of results are somewhat

questionable because of the differences in tests performed, the results do sug-

gest that the inhibiting effect of oxygen is more a function of absolute quantity

present, i.e., partial pressure, than of the percentage content. Thus, the

higher the hydrogen pressure, the lower the percentage oxygen content that will

inhibit the hydrogen embrittlement. Such gases as nitrogen and argon (Ref. 42)

or helium (Ref. 7) do not inhibit hydrogen-environment embrittlement, so their

presence as an impurity in the hydrogen would not affect results. It would be

expected that other oxidizing gases, such as CO 2 and H20, would inhibit hydrogen-

environment embrittlement, but preliminary indications are that H20 is not as

effective an inhibitor as expected.

Surface Condition.-- Because hydrogen-environment embrittlement is a surface

effect, the condition of the surface would be expected to affect the degree of

embrittlement. Important factors would be surface finish; method of surface

finishing, i.e., mechanical or chemical machining, electroplating, swaging, shot

194
peening, etc.; thickness and adherence of oxide; vacuum treatment; and degree

of work hardening. Investigations of the effect of these factors are very lim-

ited. Walter and Chandler (Ref. 7) did find that abrading a steel in hydrogen

to remove the oxide prior to tensile testing in hydrogen increased the number of

surface cracks formed in the necked-down region.

Mechanism

The mechanism for hydrogen-environment embrittlement may be divided into

two areas for discussion: (1) the mechanism(s) for the transfer of hydrogen

from the gas phase to the metal at the location and in the form in which it can

embrittle the metal, and (2) the mechanism by which the hydrogen actually

causes the metal to be embrittled. The first of these, which will be desig-

nated as the transfer mechanism, is at present better understood than the sec-

ond, which will be designated as the embrittlement mechanism. It is generally

agreed that hydrogen-environment embrittlement is a surface or near-surface

effect. If the embrittlement results from penetration of hydrogen into and

embrittlement of a surface layer, the depth of that layer is very small.

In the discussion, hydrogen-environment embrittlement and internal-hydro-

gen embrittlement will be covered. These two terms refer only to the source of

the hydrogen for embrittlement. In the first case, the hydrogen comes from the

environment and, in the second, the hydrogen is dispersed through the metal but

may have to be concentrated in certain locations for embrittlement to occur.

The transfer mechanisms are obviously not identical for the two types of embrit-

tlement. However, the actual embrittlement mechanism may be the same, and many

investigators have found it attractive to so postulate such a unifying mechanism.

195
 The transfer of hydrogen to or into a metal will involve molecular hydro-

gen adsorption, which may be separated into physisorption and chemisorption,

dissociation of molecular hydrogen to atomic hydrogen, and possibly absorption

of the atomic hydrogen into the surface of the metal and diffusion into the

metal to the extent necessary. As discussed previously, Williams and Nelson

(Ref. 8) have concluded that chemisorption of hydrogen is the rate-controlling

transfer mechanism for embrittlement in a hydrogen atmosphere. By studying

crack growth rates using dissociated hydrogen, thereby eliminating the control-

ling adsorption reaction H 2 + 2H + + 2e-, they concluded that embrittlement was

controlled by absorption of hydrogen into the metal surface. For internal-

hydrogen embrittlement, the movement of hydrogen through the metal has been

shown to be the rate-controlling process. Thus, for hydrogen-environment em-

brittlement, adsorption of hydrogen has been postulated to be the rate-

controlling transfer mechanism while, for internal-hydrogen embrittlement, the

movement of hydrogen may be rate controlling; this does not, however, in any

way determine the final embrittlement mechanism.

Bulk diffusion cannot be the transfer mechanism for the rapid embrittlement

encountered in hydrogen environments unless the embrittlement mechanism can be

operative in the surface layer. Williams and Nelson (Ref. 8) have estimated

that the observed crack growth rates of 4130 using TDCB specimens in gaseous

hydrogen are two to five orders of magnitude faster than the bulk diffusion

rates to be expected, and they concluded that bulk transport cannot be a neces-

sary step in gaseous hydrogen environment. It is evident, therefore, that em-

brittlement is not due to the buildup of hydrogen in solution at a significant

distance ahead of the propagating crack. Walter, et al. (Ref. 9) have estimated

196
 the rate of movement of hydrogen in the metal as being faster than the rate of

propagation of cracks in tensile tests of unnotched specimens in i0 000-psig

hydrogen. If, however, embrittlement is confined to a surface layer, diffusion

in the surface layer (which can be considerably more rapid than bulk diffusion)

could allow the buildup of sufficient hydrogen in front of the crack to cause

embrittlement.

The exact mechanism by which hydrogen embrittles a metal has been the sub-

ject of many investigations. To date, no embrittlement mechanism has been

widely accepted for the actual embrittlement process. Some of the failure

mechanisms that have been suggested for internal-hydrogen embrittlement are:

i. Planar-pressure theory

2. Addition of hydrogen electrons to the 3-d band of the metal

3. Lowering of surface energy by hydrogen adsorption

The pressure theory, as originally proposed by Zappfe (Ref. 82) and modi-

fied by many investigators, e.g., Tetelman and Robertson (Ref. 83), cannot be

applicable to hydrogen-environment embrittlement since hydrogen pressures of

less than 1 atmosphere have been found to increase the rate of crack propaga-

tion. For specimens totally immersed in a hydrogen environment, the hydrogen

present in internal voids cannot have pressures greater than the applied hydro-

static pressure. Therefore, the external pressure must be at least equal to the

pressure in the gas pockets, and the pockets would have no tendency to enlarge

due to a pressure differential.

197
 Troiano (Ref. 6) has suggested that for iron-base alloys the increase of

energy due to the presence of hydrogen results from the addition of electrons

from the hydrogen atoms to the 3-d electron bands. The repulsive forces deter-

mining the interatomic distances in transition metals are due to overlapping of

these 3-d electron bands; increasing the number of electrons in the 3-d band

would increase the repulsive forces and thus lower the fracture resistance of

the metal. Such a mechanism may be applicable to both types of hydrogen embrit-

tlement since transition metals are observed to be the most susceptible to each

type. Nelson, et al. (Ref. 65) have related embrittlement to the solubility

of hydrogen in the metal surface, and thus a failure mechanism based on Troiano's

suggestion may be applicable.

Petch and Stables (Ref. 84) and Petch (Ref. 85) suggested that the failure

mechanism occurs by lowering the surface energy of internal cracks by adsorption

of hydrogen onto the crack surface and lowering the surface energy of the crack.

This theory is an extension of the Griffith crack theory for fracture of brittle

materials. Perch (Ref. 85) calculated the energy to initiate fracture in

hydrogen-charged and hydrogen-free steels, and found that the energy difference

was comparable with the energy of hydrogen adsorption. Walter, et al. (Ref. 9)

and Williams and Nelson (Ref. 8) have postulated that this embrittlement mecha-

nism can also be applicable to hydrogen-environment embrittlement wherein the

energy of adsorption of hydrogen at the crack tip compensates for the energy

required to form a new surface.

Thus, the embrittlement mechanism by which final failure occurs may be the

same for hydrogen-environment embrittlement as for internal-hydrogen embrittle-

ment. There is no evidence to indicate that the failure mechanism must be

198
different, and none to indicate they are the same. One difference that must be

explained by any mechanism is why some metals are more embrittled by hydrogen

environments than by internal-hydrogen charging. Nickel is a good example of

this. Nickel-base alloys are among the most embrittled by hydrogen environ-

ments, while they are generally not subject to internal-hydrogen embrittlement

when charged to higher hydrogen contents than can be absorbed from the embrit-

tling hydrogen environment. Table 27 shows a comparison between the suscepti-

bilities to each type of hydrogen embrittlement for a number of alloys (Ref. 9).

TABLE 27

SUSCEPTIBILITY OF VARIOUS MATERIALS TO INTERNAL-HYDROGEN

EMBRITTLEMENT AND HYDROGEN-ENVIRONMENT EMBRITTLEMENT

Internal Hydrogen
Degree of Electrolytically Charged Gaseous Hydrogen Environment
Embrittlement (Ref. 12) (Ref. 9)

Extreme H-11 Tool Steel 18 Ni(250) Maraging Steel

17-4 Plt Stainless Steel AISI Type 410 Stainless Steel
H-f1 Tool Steel
17-7 PH Stainless Steel
Ren_ 41
Inconel 718

Severe AM 355 Stainless Steel HY I00

18-Ni(250) Maraging Steel Nickel 270
AISI E 8740

Slight 17-7 PIt Stainless Steel AISI Type 304L Stainless Steel

Negligible Inconel 718 A-286 Super-Strength Alloy

r

Rene 41 AISI Type 316 Stainless Steel

199
 Internal-hydrogen embrittlement can be caused by a small amount of hydrogen

in high-strength steels. Figure 73, from Farre11 and Quarrel (Ref. 86), shows

the effect of hydrogen concentration on the reduction of area (R.A.) for a number

of steels with various initial ultimate tensile strengths. Walter and Chandler

(Ref. 52) and Benson, et al. (Ref. 46) have shown that hydrogen-environment em-

brittlement occurs with final hydrogen concentrations as low as 0.1 to 1.0 ppm.

Data for the R.A. of unnotched specimens tested in 10 O00-psig helium and hydrogen

at room temperature are also plotted in Fig. 75. It can be observed that the

high-strength steels are quite affected by internal-hydrogen contents of less than

1.5 ppm, while lower-strength steels are not as affected at this hydrogen concen-

tration. The data for specimens tested in i0 000-psig helium and hydrogen at

room temperature show that the R.A. is significantly affected for both high- and

low-strength steels by external hydrogen. Walter and Chandler (Ref. 7), Hoffman

and Rauls (Ref. 41), and Ansell, et al. (Ref. 50 and 51) have shown that a hold

period in hydrogen (either at zero hold stress or at a tensile hold stress) at

room temperature prior to testing in hydrogen has no effect on the amount of

embrittlement, indicating that no significant pickup of hydrogen occurs.

As shown in Fig. 75, embrittlement in the environment is much greater than

can be explained on the basis of internal-hydrogen embrittlement because of the

very small quantities of hydrogen that can be absorbed from these environments.

It has generally been found that the hold time in the hydrogen environment has

no effect on the degree of embrittlement. This is not to say there cannot be

synergistic effects between hydrogen-environment embrittlement and internal-

hydrogen embrittlement resulting from previously absorbed hydrogen or absorption

of hydrogen prior to exposure to the hydrogen environment.

200
 9O
-k

46
[
8O OPEN SYMBOLS ARE % RA
IN 10 O00-PSIG He

CLOSED SYMBOLS ARE % RA

IN lO O00-PSlG H2
7o

6O
Z

m_
LAJ
a_

_- 5o
I.iJ
aC
<(
UTS = 130 KSI
la.
O

z 40
156

uJ

30

2O

270

0240 I 196
0
0 0.5 1.0 1.5 2.0 2.5

HYDROGEN CONTENT, PPM

Figure 73. Effect of hydrogen content on the tensile

ductility of high-strength steel (hydrogen
content after exposure to lO000-psig
hydrogen is approximately 0.i ppm).
201
 In summary, the rate controlling transfer mechanism in hydrogen environments

is probably the adsorption of hydrogen onto the metal surface. In internal-

hydrogen embrittlement, it has been concluded to be the movement of hydrogen

through the lattice to concentrate the hydrogen in front of a crack. In both

types of embrittlement, the failure mechanism has been hypothesized only, and no

definite conclusions can be reached as to whether the failure mechanism is the

same or different for hydrogen-environment embrittlement and internal-hydrogen

embrittlement. Also, it is not possible to establish whether hydrogen-environment

embrittlement is simply a surface effect or if it involves absorption of hydro-

gen into the metal to some limited depth.

202
 CHAPTER 11 . SU_RY

A wide variety of pure metals and alloys has been found to be susceptible

to hydrogen-environment embrittlement. The embrittlement is an environmental

effect so that, as soon as a susceptible metal comes in contact with the hydro-

gen environment, the surface of the metal behaves in a brittle manner once a

certain amount of plastic deformation has occurred. Elastic properties, yield

strength, and, in many cases, the ultimate tensile strength are not affected by

the hydrogen environment. The most significant effects of the hydrogen environ-

ment are on tensile ductility, notch strength, and crack behavior.

Four categories of embrittlement, based on the results of tensile tests con-

ducted in I0 O00-psig hydrogen, have been established for classifying this suscep-

tibility of metals to hydrogen-environment embrittlement. These are:

I. Extreme embrittlement: Embrittlement is characterized by a large

decrease of notch strength and ductility and some decrease of

unnotched strength in I0 000-psig hydrogen, lligh-strength steels

and high-strength nickel-base alloys are in this category.

2. Severe embrittlement: Embrittlement is characterized by a considerable

reduction of notch strength and ductility, but no reduction of unnotched

strength. The majority of the metals tested are in this category,

including ductile, lower-strength steels, Armco iron, pure nickel, and

the titanium-base alloys.

3. Slight embrittlement: Embrittlement is characterized by a small decrease

of notch strength and a small or negligible decrease of unnotched ductility.

The nonstable AISI type 300 series stainless steels (AISI types 304L and

305), beryllium-copper, and commercially pure titanium are in this category.

203
 4. Negligible embrittlement: The aluminum alloys, stable austenitic

stainless steels, A-286 (a precipitation-hardened austenitic stainless

steel), and OFHC copper are in this category.

The effects of hydrogen environments on the tensile properties of metals

have been investigated as a function of the following test variables:

1. Deformation Rate--The degree of embrittlement decreases with increasing

rates of deformation. There are indications that no embrittlement is

observed at sufficiently fast testing speeds.

2. Hydrogen Pressure--The degree of embrittlement increases with increasing

hydrogen pressure, but can be considerable in some alloys even at

14.7-psia pressure.

5. Notch Severity--The hydrogen-environment embrittlement of low-alloy

steels is a function of stress concentration factor (Kt), increasing

with Kt at low Kt, passing through a maximum for Kt_6 to 9, and then

decreasing with Kt for higher Kt values.

4. Exposure Time in Hydrogen--Most investigations indicate that the degree

of hydrogen-environment embrittlement is essentialy independent of hold

time or stress for both notched and unnotched specimens.

5. Te_0erature--ttydrogen-environment embrittlement is greatest in the

vicinity of room temperature. The embrittlement decreases rapidly with

decreasing temperature, and less rapidlywith increasing temperatures.

Embrittlement has been observed at temperatures as high as 1250°F.

Crack growth measurements indicate that the rate of crack growth is

proportional to l/T, where T is the absolute temperature.

204
 6. Weldments--The effects of hydrogen on the mechanical properties of

metals may vary appreciably between the parent metal, heat-affected

zone, and weld metal. The weld zone and the heat-affected zone may

be more affected or less affected than the parent metal.

Low-cycle fatigue and fracture toughness measurements also have been per-

formed in hydrogen environments. Cyclic crack propagation rates are considerably

increased by hydrogen environments. As an example, the 1000-cycle, ambient tem-

perature fatigue strength of precracked, notched ASTbl AS33-B and ASTM A-S17 spec-

imens was reduced by about 6S percent in 10 O00-psig hydrogen. Tests conducted

on Inconel 718 and Ti-6A1-4V (STA) indicate that the effect of hydrogen on the

fatigue strength is also a maximum at room temperature and decreases with decreas-

ing temperature. Very low hydrogen pressures also can increase the rate of cyclic

crack growth. For example, measurable increases of cyclic crack growth rates

were observed in titanium at hydrogen pressures as low as 1 torr.

Hydrogen can increase the sustained as well as cyclic crack growth rate.

The sustained crack growth rate in 4130 steel was considerably increased by hydro-

gen at 14.7-psia pressure. The threshold stress intensity, Kth, which is the min-

imum stress intensity required for sustained flaw growth, also can be considerably

reduced by hydrogen environments. For example, Kth of Inconei 718 in S200-psig

hydrogen was determined to be 22 ksi i_n. compared to a Kic of 150 ksi ivY. in air.

Two methods for the prevention of hydrogen-environment embrittlement are

the use of protective coatings and the addition of inhibitors to hydrogen.

Coatings that have shown the most potential for preventing hydrogen-environment

embrittlement are copper, gold, cadmium, and perhaps Sylcor R-S08 (a silver-

20S
silicon coating) and Sylcor R-505 (a tin-aluminum coating) for elevated-

temperature service. The degree of hydrogen-environment embrittlement decreases

with the amount of oxygen in the hydrogen. For example, the addition of l-

percent oxygen has been shown to prevent embrittlement of a 0.22-Percent carbon

steel by high-pressure hydrogen.

Metallographic studies have shown that fracture initiation in high-pressure

hydrogen occurs at the metal surface. The fracture can be transgranular or inter-

granular. The surface-cracking characteristics are similar for metals in the

same embrittlement categories regardless of base metal, crystal structure, or

hydrogen adsorption capacities. The appearance of the brittle surface region

was found to be independent of hydrogen pressure, but the depth of the brittle

region was a linear function of the square root of hydrogen pressure.

The mechanism by which gaseous hydrogen embrittles metals has not been

established, but there are certain aspects regarding it that are known. Basic-

ally, hydrogen-environment embrittlement is believed to be a surface or near-

surface phenomenon. It has been shown that fracture initiation in hydrogen

occurs at the surface and after a certain amount of plastic deformation has

occurred.

206
 CHAPTER 12. RECO_ENDATIONS

This chapter contains recommendations for those concerned with the applica-

tion of structural materials in contact with hydrogen and, in particular, with

high-pressure hydrogen.

In designing structures for service in hydrogen, the use of materials embrit-

tled by hydrogen should be avoided. However, this will be possible only in a

limited number of structures and, in general, it will be necessary to use mater-

ials affected to some degree by hydrogen. The method used for factoring the en-

vironmental effect into the design will depend on the loading characteristics of

the structure. In some cases, the degradation of a given mechanical property,

e.g., low-cycle fatigue, can be used for establishing the design. However, the

problem is more difficult for designs based on tensile properties.

As discussed in Chapter 10, Technical Discussion, tensile tests in hydrogen

are actually complex crack propagation tests, and it is difficult to use "tensile

properties" from such tests in normal design procedures, ttowever, in some cases,

it may be appropriate to use degradation of'properties, i.e., reduction of notch

strength, to establish a safety factor to compensate for the effects of the hy-

drogen environment. Hydrogen-environment embrittlement (excluding high-cycle

fatigue) should not occur if one can truly design so that there is no material

yielding and the design is based on the yield strength rather than on the ulti-

mate strength of the material.

The most significant effect of environmental hydrogen on metals involves

crack initiation and brittle crack propagation. Therefore, fracture mechanics,

which is the study of crack behavior, is an attractive approach in designing for

207
service in hydrogen environments. The useful life of the structure can be pre-

dicated from proof tests in air, which measure the maximum existing flaw size,

and from the threshold stress intensity (Kth) for crack growth and/or cyclic

flaw growth data obtained from tests conducted in hydrogen environments.

It has been shown that most structural alloys are embrittled by hydrogen en-

vironments; therefore, those alloys (particularly those based on transition metals)

that have not been tested in hydrogen should be assumed to be susceptible to hy-

drogen environment embrittlement.

Good design practices are especially important for the design of structures

for ser¢ice in hydrogen environments. Notches or stress concentrations should be

reduced and, if possible, eliminated; surface finishes should be tightly controlled;

and all surface material defects removed. Inspection techniques should be thorough

and periodic to determine if sustained or cyclic flaw growth is occurring after

the structure has been placed in service.

It is evident that only properties measured in hydrogen environments should

be used in designing structures for use in hydrogen. The apparatus and procedures

for hydrogen environmental testing have been described in Chapter 5, Test Proced-

ures. The hydrogen used during the tests should have the same purity as the

highest hydrogen purity the structures would see in service. Prior to conducting

the tests, the test syste m should be thoroughly purged by evacuation and pres-

surization/depressurization cycling. Removal of adsorbed gases from the test

specimen simulates the condition of structures that will be exposed to high-

purity hydrogen or to flowing hydrogen. The cracked specimen s used for fracture

mechanics testing should be cycled at low stress intensity amplitudes to extend

208
the cracks in hydrogen prior to performing the tests. This is to ensure that a

clean crack will be exposed to hydrogen so that the true crack behavior in hydroL

gen will be measured.

In some instances, it may be possible to prevent embrittlement of a suscep-

tible material by coatings or by inhibition (impurity additions to the hydrogen

environment). Investigations or inhibition have been limited. If coatings are

used, extreme care must be taken to ensure that the coatings remain intact because

hydrogen will initiate and propagate cracks when contact is made with the metal

surface. Vented, stainless-steel liners have proved to be effective for the pro-

tection of structural steels in high-pressure hydrogen storage vessels.

209
 REFERENCES

i. Lessing, L.: Fortune, May 1961, 152.

2. Nelson, G. A.: Proc. Am. Pet. Inst., 29M(III), 1949, 163.

3. Bentle, G., and W. T. Chandler: Effects of Hzdrogen Environments on

Columbium and Tantalum Alloys, Rocketdyne, a division of North American

Rockwell Corporation, Canoga Park, California, AFML-TR-66-322, Part If,

Wright-Patterson AFB, Ohio, September 1968.

4. Dann, A. W., F. J. Shortsleeve, and A. R. Troiano: Trans. AIME, 203,

1955, 895.

5. Frohmberg, R. P., W. J. Barnett, and A. R. Troiano: Trans. ASM, 47, 1955,

892.

6. Troiano, A. R.: Trans. ASM, 52, 1960, 54.

7. Walter, R. J., and W. T. Chandler: Effects of Hi_h-Pressure Hydrogen on

Metals at Ambient Temperature, Final Report, Contract NAS8-19, NASA, MSFC,

Huntsville, Alabama, Rocketdyne, a division of North American Rockwell

Corporation, Canoga Park, California, Report R-7780-I, 2, 3, 1969.

8. Williams, D. P., and H. G. Nelson: Met. Trans., _, 1970, 63.

9. Walter, R. J., R. P. Jewett, and W. T. Chandler: Mater. Sci. Eng., _,

1969/70, 99.

10. Lounamaa, K., and G. Braggstrom: J. Iron Steel Inst., (London), 203,

1965, 702.

211
 11. Schaller, F. W., and A. R. Troiano: Hydrogen Embrittlement in Face-Centered

Cubic Alloys of Nickel, ARL-133, Wright-Patterson AFB, Ohio, December 1961.

12. Groenveld, T. P., E. E. Fletcher, and A. R, Elsea: A Stud_ of Hydrogen

Embrittlement of Various Alloys, Summary Report, Contract NAS8-20029,

NASA, MSFC, Huntsville, Alabama, 1966.

13. Tardif, H. P., and H. Marquis: High Pressure ltydrogen Effects on Steel,

CARDE Tech. Memo 609/61, May 1961.

14. Mills, R. L., and F. J. Edeskuty: Chem. Eng. progress, 52 (11), 1956, 477.

15. Dodge, B. P.: Trans. ASME, 75, 1955, 331.

16. Bridgeman, P. W.: "The Compressibility of Five Gases to High Pressure,"

Proceedings of the America n Society of Arts and Sciences, 59, 1924, 173.

17. Poulter, T. C.: Trans. ASME, J. Appl. Mech., 64, 1942, A31.

18. Nelson, G. A., and R. T. Effinger: Welding J., 1955, 12s

19. Cooper, C. M.: Chem. and Eng. News, 30, 1952, 2942.

20. Van Ness, H. C., and B. F. Dodge: Chem. Eng. Prog., 51 (6), 1955, 266.

21. Perlmutter, D. D., and B. F. Dodge: Ind. and En_. Chem., 48 (S), 1956, 885.

22. Laws, J.S., V. Frick and J. McConnell, Hydrogen Gas Pressure Vessel Problems

in the M-1 Facilities, NASA CR-1305, March 1969. (Aerojet-General Report

8800-67, April 1966.)

23. Engelhard Industries, Inc. Sales Brochures, 113 Astor St., Newark, New Jersey.

24. "Molecular Sieves for Selective Adsorption," Data Sheet No. 51Linde Co.

New York, New York (1957).

212
25. Walter, R. J.: Instrumentation Technology, 14 (9), 1967, 84.

26. Waiter, R. J., H. G. Hayes, and W. T. Chandler: Mechanical Properties

of Inconel 718, Waspaloy, A-286, and Ti-SA1-2.5Sn ELI in Pure Gaseous H2,

Prepared for Aerojet General Corp., Sacramento, California, Rocketdyne, a

Division of North American Rockwell Corporation, Canoga Park, California

Report No. R-8187, 10 April 1970.

27. Dessau, P. P.: Aerojet Nuclear Systems Company, Sacramento, California

Personal communication.

28. Bixler, W. D.: Flaw Growth of Inconel 718 and 5A1-2.5Sn (ELI) Titanium in

a High Purity Gaseous Hydrogen Environment, Boeing Document D180-10142-1,

Prepared for Aerojet General Corporation, Sacramento, California, Contract

L-80076, The Boeing Company, Seattle, Washington.

29. Lyszj, I.: "An Apparatus for the Determination of the Ortho-Para Isomer

Ratio and the Detection of Trace Impurities in Cryogenic Samples of

Hydrogen," ISA ProceedinGs , 1963, 41.

30. Marcus, H. L., and P. J. Stocker: North American Rockwell Science Center,

Thousand Oaks, California, Personal communication.

31. Williams, D. P., and H. G. Nelson: NASA Ames Laboratory, Moffett Field,

California, Personal communication.

32. Wei, R. P.: Lehigh University, Bethlehem, Pennsylvania, Personal

communication.

33. Forman, R. G.: NASA, MSC, Houston, Texas, Personal communication.

213
34. Hydrogen Embrittlement Tests of Pressurized Cylindrical and Tensile Specimens

Including Conventional Tensile Tests on ASTM A-302-B Steel, Nickel Modified

NASA, Kennedy Space Center TR-451, October 1966.

35. Honnigford, J.B.: M.S. Thesis, Rensselaer Polytechnic Institute, Troy,

New York, 1958.

36. Lorenz, P. M.: Effect of Pressurized Hydrogen Upon Inconel 718 and

2219 Aluminum, ASM Report No. W9-13.2, ASM Report System, _tetals Park, Ohio.

37. Walter, R. J., and W. T. Chandler: AIAA J., 4 (2), February 1966, 302,

38. Walter, R. J., J. A. Ytterhus, R. D. Lloyd, and W. T. Chandler: Effect

of Water Vapor/Hydrogen Environments on Columbium Alloys, AFML-TR-66-322,

Part I, Air Force Materials Laboratory, Wright-Patterson Air Force Base,

Ohio, December 1966.

39. Chandler, W. T. and R. J. Walter: "Hydrogen Effects in Refractory Metals

and Alloys," Refractory Metals and Alloys, Ed. by I. Machlin, R. T. Begley,

and E. D. Weisert, Plenum Press, New York, 1968, 197.

40. Hofmann, W., and W. Rauls: Archiv fSr das EisenhSttenwesen, _arz 1961, 1.

41. Hofmann, W., and W. Rauls: Dechema Monograph 45, 1962, 33.

42. Hofmann, W., and W. Rauls: Welding J,, May 1965, 225-S.

43. Peterson, R. E.: Stress Concentration Desi._n Factors, John Wiley and Sons,

New York, 1953.

44. Cavett, R. H., and H. C. Van Ness: Welding J., July 1963, 316-S.

45. Vennett, R. M., and G. S. Anse11: Trans. quart. A_i , 62, 1969, 1007.

214
46. Benson, R.B., R.K. Dann, and L.W. Roberts, Jr.: Trans AIME, 242, 1968, 2199.

47. Walter, R.J. and W.T. Chandler: Effect of Hydrogen Environments on Inconel

718 and Ti-6AI-4V (STA), Paper presented at AIME Meeting, Las Vegas, Nevada,

May II-14, 1970.

48. VanWanderham, M.C., Pratt and Whitney Aircraft, W. Palm Beach, Fla.,

Personal communication.

49. Walter, R. J., and W. T. Chandler: Effect of Hydrogen Environment on

Inconel 718 and Ti-6AI-4V Under Simulated J-2 Engine Operating Environments,

R-7920, Rocketdyne, a division of North American Rockwell, Canoga Park,

California, Contract NAS8-19C, NASA, MSFC, Huntsville, Alabama, 30 June 1969.

50. Steinman, J. B., li. C. Van Ness, and G. S. Ansell: Welding J, Hay 1965,

221-S.

51. Vennett, R. M., and G. S. Ansell: Trans. ASM, 60, 1967, 242.

52. Walter, R. J., and W. T. Chandler: Effect of lligh-Pressure Hydrogen on

Storage Vessel Materials, Paper presented at WESTEC Conference, Los Angeles,

Calif., ASM Report No. W 8-2.4, ASM Report System, Metals Park, Ohio, 44073,

March 1968.

53. Owens, W.H.: Effect of High Pressure Hydrogen on SAE 4140 and SAE 1018

Steels at Ambient Temperatures, AEDC-TR-66-269, Arnold Engineering Develop-

ment Center, Arnold Air Force Station, Tennessee, December 1966.

54. Pittinato, G.F.: The Effects of a Hydrogen Environment on the Mechanical

Properties of Forged Ti-6AI-4V ELI Weldments, McDonnell Douglas Corp.,

Huntington Beach, Calif., Contract NAS8-21470, NASA, MSFC, Huntsville,

Alabama, November 1969.

215
 55. Vennett, R. M., and G. S. Ansell: Trans. Quart. ASH, 62, 1969, 10u/.

56. Bernstein, I. M.: Mat. Science and Eng., 6, 1970, 1.

57. Klima, S. J., A. J. Nachtiga11, and C. A. Hoffman: Preliminary Investiga-

tion of Effect of Hydrogen on Stress-Rupture and Fatigue Properties of an

Iron-a a Nickel-, and a Cobalt-Base Alloy, NASA TN D-1458, December 1962.

58. Hofmann, W., W. Rauls, and J. Vogt: Acta. Met., 10, (7) 1962, 688.

59. Walter, R. J., and W. T. Chandler: Effect of High-Pressure Hydrogen on

Incone1718 at -260°F, Research Report RR 70-3, Rocketdyne, a division of

North American Rockwell Corporation, Canoga Park, California, 1 June 1970.

60. Pittinato, G. F.: Trans. Quart. ASM, 62, 1969, 410.

61. VanWanderham, M.C., and J.A. Ilarris, Jr.: Pratt and Whitney Aircraft, West

Palm Beach, Florida, Private communication.

62. Lorenz, P. M.: Effect of Pressurized Hydrogen Upon Incone1718 and 2219

Aluminum_ Boeing Document D2-I14417-I, The Boeing Company, Seattle, Wash-

ington, February 1969.

63. Hancock, G. G., and H. H. Johnson: Trans. AIME, 236, 1966, 513.

64. Spitzig, W. A., P. M. Talda, and R. W. Wei: Eng. Frac. Mechn., i, 1968, 155.

65. Nelson, II. G., D. P. Williams, and A. S. Tetelman: "Embrittlement of a

Ferrous Alloy in a Partially Dissociated Hydrogen Environment," Paper

accepted for publication in Met. Trans.

66. Isida, M., and Y. Itagaki: Stress Concentration at the Tip of a Transverse

Crack in a Stiffened Plate Subjected to Tension, 4th U.S. Congress Applied

Mechanics, 1962, 955.

216
67. Paris, P. C., and G. C. Sih: ASTM STP No. 381, Philadelphia, Pa., 1965, 30.

68. Toh, T., and W. M. Baldwin: Stress Corrosion Crackin_ and Embrittlement,

John Wiley and Sons, New York, 1956, 176.

69. Williams, D. N.: J. Inst. Metals, 91, 1962-63, 147-152.

70. Walter, R. J., and W. T. Chandler: Metallograph_ of Alloys Fractured in

Gaseous H_dro_en Environments, Paper presented at AIME Meeting, Las Vegas,

Nevada, 11-14 May 1970.

71. Marcus, H. L., B. S. Hickman, J. C. Williams, G. Garmong, and P. Stocker:

Slow Crack Growth in Titanium Alloys Exposed to Low Pressure of H 2 Gas,

Paper presented at AIME Meeting, Las Vegas, Nevada, I1-14 May 1970.

72. Laird, C., and G. C. Smith: Phil. Mas., 7, 1962, 847.

73. Meyn, D. A.: Trans. ASM, 61, 1968, 42.

74. Pelloux, R. M.: Trans. ASM, 62, 1969, 281.

75. Martini, W. R.: Flame Heated Thermionic Converter Research, Final Report

,July 1961-June 1963), Contract No. DA-36-039 SC-88982, Report AI-8681,

Atomics International, a division of North American Rockwell, Canoga Park,

California.

76. Steigerwald, E. A.: The Permeation of Hydrogen Through blaterials for the

Sunflower System, Final Report on Project No. 512-278131-08, 15 November

1963, TRW Electromechanical Division Report ER-5623.

217
 77. Dushman, S.: Scientific Foundations of Vacuum Technique, 2nd Edition,

John Wiley and Sons, New York, 1962.

78. Walter, R. J., and J. A. Ytterhus: Behavior of Columbium and Tantalum

in Hzdrosen Environments, Research Report No. 67-7, Rocketdyne, a division

of North American Rockwell, Canoga Park, California.

79. Fidelle, J. P., J. M. Deloron, C. Roux, and M. Rapin: "Influence of

Surface Treatments and Coatings on the Embrittlement of High-Strength

Steels by Hydrogen Under Pressure," Case of 35 NiCrMo 16 Steel. Reprint

from Proceedings of Interfinish '68 (7th International Congress on Metal

Surface Treatment, Hanover, 5-9 May 1968).

80. Sawicki, V., and W. Johnson, Cornell University, Private communication.

81. Phillips, A., V. Kerlins, and B. V. Whiteson: AFML-TDR-64-416, Wright-

Patterson Air Force Base, Ohio, 1965.

82. Zappfe, C. A.: Mater. Methods, 32, 1950, 58.

83. Tetelman, A. S. and W. D. Robertson: ActaMet, Ii, 1963, 415.

84. Petch, N. J. and P. Stables; Nature, 169, 1952, 842.

85. Petch, N. J.: Phil.Mag. _, 1956, 351.

86. Farrell, K. and A. G. Quarrell: J. _ron Steel Inst. (London) 202, 1964,

1002.

218
 APPENDIX:WORK
IN PROGRESS

Rocketdyne,
a division of North American Rockwell Corporation

Canoga Park, California

Program Title: Influence of Gaseous Hzdrogen on Metals

Principal Investigators: R. J. Walter and W. T. Chandler

Funding Agency: NASA under contract NAS8-25579

The program is designed to provide data on the mechanical properties of

candidate structural alloys in hydrogen under simulated Space Shuttle Engine

operating conditions. Testing will be conducted in 5000-psig hydrogen and helium

environments. The program is divided into the following three phases. Phase I

consists of determining the variation of hydrogen-environment embrittlement with

the material condition of Inconel 718. Three material heats in three different

heat-treatment conditions will be evaluated by tensile testing in the two environ-

ments. Inconel 718 weldments will also be tested.

Phase II consists of performing tensile tests at room temperature and

-200°F on Inconel 718, Inconel 625, AISI 321 stainless steel, A-286 stainless

steel, Ti-SAI-2.SSn ELI, 2219-T87 aluminum alloy, and OFHC copper specimens.

Phase III consists of measuring the threshold stress intensity, Kth at

room temperature and -200°F on Inconel 718, Inconel 625, AISI 521 stainless

steel, A-286 stainless steel, Ti-SA1-2.SSn ELI, 2219-T87 aluminum alloy, and

OFHC copper specimens.

219
Program Title: Space Shuttle Engine and Breadboard Thrust Chambers

Funding Agency: NASA

Severe hydrogen-environment embrittlement has been observed in electro-

formed nickel used to close out the coolant passages in hydrogen-fueled rocket

engine thrust chambers. A program is being conducted to determine the mechanical

properties of electroformed nickel in 1200- and 7000-psi hydrogen at -320 ° to

400°F. In 1200-psi hydrogen at room temperature, electroformed nickel has prac-

tically no ductility, and the notched strength is reduced about 70 nercent com-

pared to that in air. These results are particularly interesting because electro-

formed nickel is very ductile in air and in comparative helium environments.

Annealing decreases the embrittling effect of hydrogen and, as is mentioned in

Chapter 9, Preventive Measures, copper and gold coatings are partially effective

for preventing embrittlement.

Work is continuing to better define the properties of electroformed nickel

in these hydrogen environments by determining its fatigue strength, cyclic crack

growth rate, and fracture toughness. Experiments also are being conducted to

improv 9 techniques for preventing embrittlement.

220
 Pratt 6 Whitney Aircraft, Florida Research

and Development Center,

Division of United Aircraft Corp.

West Palm Beach, Florida

Program Title: Influence of Elevated Temperature on Metals In Gaseous Hydrogen

Responsible Engineer: J. A. Harris, Jr.

Funding Agency: NASA under contract NAS8-26191

The program consists of performing fatigue, fracture toughness, creep

rupture, and tensile tests in 5000-psi hydrogen and helium environments,

according to the matrix in Table 28.

221
 TABLE 28

TEST PROGRAM (PRATT 6 WHITNEY AIRCRAFT)

Low High
Cycle Cycle Fracture Creep Notched Smooth
Temp. Fatigue Fatigue Toughness Rupture Tensile Tensile
Material OF (H2/H e) (H2/He) (H2/He) (H2/He) (H2/He) (H2/He)

Inconel 718" 75 X X X X X
1250 X X X X X

Inconel 718"* -320 X

75 X X X X X
1250 X X X X X

Inconel 718 Welds** 75 X X X

Inconel 625 75 X X X X -
1250 X X X

Inconel 625 Welds 75 X X X

A-286 75 X X X X
1250 X X X X

AISI 347 -320 X X

75 X X X X X
1250 X X

X X
Hastelloy X 75 X
1250 X X

Ti-6A1-4V 75 X X X X
200 X X X X
(or maximum feasible temperature)

Ti-5A1-2.5Sn 75 X X X X X
200 X X X X
(or maximum feasible temperature)

*Anneal at 1750°F for 1 hour; air cooled, plus aged at 1325°F for 8 hours; furnace
cooled to l150°F; held at 1150°F for a total aging time of 18 hours; air cooled
**Solution at 1950°F for 1 hour; air cooled, plus aged at 1400°F for I0 hours;
furnace cooled to 1200°F; held at 1200°F for a total aging time of 20 hours;
air cooled

OR

Solution at 1900°F for 1 hour; air cooled 50°F/minute or faster, plus aged at 1325°F
for 8 hours; furnace cooled to l150°F; held at 1150°F for total aging time of 18 hours;
air cooled. Recent data indicate solution at 1900OF provides less hydrogen degrada-
tion in low cycle fatigue than anneal at 1750°F or solution at 1950°F.

222
 NASA, Ames Research Center

Moffett Field, California

Program Director: D. P. Williams

The mechanical properties of dispersion-strengthened nickei-base alloys are

being measured in hydrogen environments. The purpose of the program is to study

the intrinsic nature of embrittlement of nickel. Tests performed on Ni-4.5%AI,

which contained a fine dispersion of an aluminum-rich phase in an essentially

pure-nickel matrix, has shown considerable embrittlement. Fractography of one of

these specimens is presented in Chapter 8.

The energy of adsorption of hydrogen on metal surfaces is being measured

by two methods: (I) measuring the energy needed to remove adsorbed hydrogen by

ion bombardment; (2) evaporating and condensing a metal inside a calorimeter in

vacuum and hydrogen environments. The differences between the surface energy

in these environments is a measure of the heat of hydrogen adsorption.

The influence of straining in hydrogen on the surface dislocation morphology

of AISI 1020 steel is being determined. Cracks are propagated slowly in low-

pressure hydrogen, and the fracture surfaces are examined by transmission

electron microscopy.

223
 NASA, Manned Spacecraft Center

Houston, Texas

Principal Investigator: R. G. Forman

The sustained flaw growth in surface-flawed sheet specimens of parent metal

and weld metal Inconel 718 was measured in 700- to 1000-psi hydrogen and helium

environments. The fracture toughness Kic of these specimens in air as a function

of specimen strength is shown in Fig. 74. Even though a magnification factor

was used, the calculated fracture toughness of the parent metal specimens de-

in air was about I00 ksi ivan. for the

creased with increasing crack depth. KIc

parent metal and 55 ksi _-n. for the weld material.

The time to failure for specimens held under load at -IO0*F in hydrogen is

shown in Fig. 75. The threshold stress intensity, designated as Kiscc, is about

18 ksi _ for the weld metal and 25 ksi i_-. for the parent metal. The

crack growth rate at -100°F in these hydrogen environments is shown as a func-

tion of stress intensity in Fig. 76.

224
 110,

!00 i Kic FOR PARENT HATERIAL

V- FLAW NORMAL TO

WALL SURFACE
A

8o--
a
1
2c--
K c " i.1 Gcvr-_M K

70_ WHERE Q = f (a/_c)

HK = f (a/t)

6O

!K I C FOR WELD HATERIAL

q

-U 5o
_L

40

30 SYMBOLS TENSILE STRENGTHp ksi

-320°F =423°F
-320@F(LN2) -423°F(LH2) i

WELD • • ayp aULT Gyp GULT

20 _ PARENT 0 WELD 169,5 _ 202* 212 Z55

i I
PARENT 204,5 259 212 283
Gc - FRACTURE STRESS
J
Gyp - YIELD STRESS *WELDS WERE MACHINED FLUSH
10 .
GULT - ULTIMATE STRESS

OI
0.3 0.4
I
0.5 0.6 0.8 0.9 1.0

Figure T4. Fracture toughness and tensile strength of Inconel 718.

225
 O

SYNBOLS
WELD PARENT
• O CRACK GROWTH TO FAILURE
35 m o CRACK GROWTH >O.OO2-1MCH
A A POSSIBLE TRACE OF GROEr114
• o NO CRACK GEOVI'N

NUNBERS IN PARENTHESIS REFER TO

CRACK GROWTH IN INCHES
I I
3o
0(0.oo6)
I-1(o.oo3)
17 (0.004) 0(0.003)
KISCC FOR PARENT I

,oo.ooor
Q
0
A
w

v
©
2O

WECOS__ A
A
(KI¢- 45°O0O TO 55°OO0)

15

KISCC IS THE HAXIHUH KIi LEVEL AT WHICH CRACK

GRO_/TH WILL NOT OCCUR RESULTING FROM SUSTAINED

LOADING ]
IO

5 io 50 loo 5o0 I000

TIME AT LOAD, HOURS

Figure 75. Crack-growth behavior of surface-flawed Inconel 718

specimen tested under sustained load in gaseous hydrogen
at -i00 ° F.

_6
 45

4o

35
/
30
A

°I

25

20

151o-6' 1o-S io-4 io-3 10-2

CRACK GROWTH RATE (d(a/Q)/dt), IN./HR

Figure 76. Crack-growth rate of Ineonel 71_ parent metal

under sustained load in gaseous hydrogen at
-lO0° F.

227
 The Boeing Company

Space Division

Seattle, Washington

Program Title: Flaw Growth of Inconel 718 and 5AI-2.SSn (ELI) Titanium in a

High-Purity Gaseous H[drogen Environment

Principal Investigator: W. D. Bixler

Report No. : Boeing Document D180-I0142-I, prepared for Aerojet-General Corp.

Contract : L-80076, September 1970

This work has been completed but the final report was received too late to

be included in the text. Tests were conducted on surface-flawed Inconel 718 and

Ti-SA1-2.5Sn ELI sheet specimens in 1400-psi hydrogen at 70°{, -100°F, and -160°F.

The results indicated sustained-load threshold intensities in Inconel 718 of

19.8, _30, and 99.4 at 70, -I00, and -160°F, respectively. Similarly; the

sustained-load threshold intensities of Ti-SA1-2.SSn ELI were 21.5, _45, and

53.3 at 70, -i00, and -160°F, respectively. The threshold intensities for cyclic

flaw growth was not measured but was determined to be less than the sustained

load threshold values.

A continuation of this work will be performed under a direct NASA contract.

Sustained and cyclic flaw growth measurements will be performed on Ti-SA1-2.SSn

ELI specimens in 300- to 400-psi hydrogen at -100 to -423°F.

228
 Aerojet Nuclear Systems Company

Sacramento, California

Principal Investigator: P. P. Dessau

Funding Agency: NASA

A series of fracture mechanics-type measurements are being planned in

1400-psi hydrogen at room temperature. The tests will be conducted on

Fe-22Cr-15Ni-9Mn, titanium, Ineonel 718 and A-286 stainless steel. Measure-

ments will include plane strain and cyclic flaw growth tests on compact tension

and through cracked-sheet specimens and threshold measurements using the Novak-

Rolfe modified WOL-type specimen technique.

229
 U. S. Steel Corporation
Applied ResearchLaboratory
Monroeville, Pennsylvania

Principal Investigator: A. W. Loginow

The threshold stress intensity on various steels in high-pressure hydrogen

is being measured by placing preloaded WOL-type specimens in high-pressure hydro-

gen for relatively long durations.

230
 Cornell University

Ithaca, New York

Principal Investigator: V. Sawicki

Program Manager: H. H. Johnson, Department Materials Science and Engineering

Funding Agency: AEC

Professor Johnson is continuing his investigations into the influence of

environments on crack propagation. Tests are being conducted in l-atmosphere

to 200-mm Hg on H-If steel specimens. The work includes measuring Klc, Kth ,

and the influence of temperature, stress intensity, and oxygen inpurity content

on the rate of crack extension. Investigations are being made to determine

whether hydrogen penetration into the metal ahead of the crack is involved in

crack growth.

Some of the significant though preliminary results are (I) Kth decreases

as the hydrogen pressure increases, (2) the effect of temperature on crack

growth rate is essentially the same as observed by Williams and Nelson* on

AISI 4130 steel, (3) the crack growth rate increases with increasing stress

intensity but appears to reach saturation at high stress intensities and,

(4) oxygen impurity levels of less than 500 to SO0 ppm in 1-atmosphere hydro-

gen did not affect the measured Kth. Kth increased with increasing oxygen con-

tent for oxygen contents greater than S00 ppm.

Klc was measured by noting the stress intensity at which the sustained

or slow flaw growth rate reached instability. By this method Klc was found

to be independent of hydrogen pressure.

*Williams and Nelson results were discussed earlier in this report.

231
 Lehigh University

Bethlehem, Pennsylvania

Program Manager: R. P. Wei, Department of Mechanical Engineering and

Mechanics Funding Agency: ARPA

Professor Wei is continuing his investigations into the influence of

environments on crack growth. An interesting observation is that cyclic crack

growth rate in_O.O60-inch-thick Inconel 718 plate appears to be the same in

a 1-atmosphere hydrogen as in a 1-atmosphere argon environment.

232
 Westinghouse Electric Corporation

Astronuclear Laboratory

Pittsburgh, Pennsylvania

Principal Investigator: R. L. Kesterson

Funding Agency: NASA

A program has commenced in which Kth is being measured in hydrogen at atmos-

pheric pressure and temperatures ranging from -320°F to +300°F. The materials are

beryllium, TZM molybdenum alloy, tungsten, stainless steels, and some copper

and aluminum alloys.

A program is being scheduled to determine the influence of an ionized

hydrogen environment on the mechanical properties of beryllium, beryllium

copper, tungsten alloys, A-286 stainless steel, Cu-Bl0mixtures, BATH (aluminum

mixture), aluminum alloys, graphites, and AISI 301, 302, and 304 stainless

steels.

233
 Aerojet Nuclear Systems Company

Sacramento, California

Principal Investigator: C. E. Dixon

Funding Agency: NASA

Notched specimens of four materials will be irradiated in water (1S0*P)

prior to room-temperature tensile testing in 500- and 1S00-psig hydrogen. Con-

trol tests also will be conducted in helium after irradiation. The four mater-

ials to be tested are: Inconel 718, Ti-SA1-2.SSn, A-286 stainless steel (bolt

material), AISI type 301 stainless steel (full hard).

234
 AEC,SavannahRiver Laboratory
Aiken, South Carolina

Principal Investigator: J. A. Donovan

Funding Agency: NASA

The purpose of this program is to develop fundamental knowledgeabout the

mechanismof hydrogendamagein metals at near-ambient temperatures. Initial
work has concentrated on the hydrogen solubility, diffusivity, and permeation

in high-purity iron and nickel.

235