Googlepreview

Photovoltaic Engineering Handbook
Photovoltaic Engineering Handbook
France Lasnier and Tony Gan Ang
Asian Institute of Technology, Bangkok, Thailand
[dam
Adam Hilger, Bristol and New York

© IOP Publishing Ltd 1990
All rights reserved. No part of this publication may be reproduced, stored in

a retrieval system or transmitted in any form or by any means, electronic,
mechanical, photocopying, recording or otherwise, without the prior per
mission of the publisher. Multiple copying is only permitted under the terms
of the agreement between the Committee of Vice-Chancellors and Principals
and the Copyright Licensing Agency.
British Library Cataloguing in Publication Data
Lasnier, France
Photovoltaic engineering handbook.
1. Solar energy. Photovoltaic conversion
I. Title II. Ang, Tony Gan
621.31244
ISBN 0-85274-311-4
US Library of Congress Cataloging-in-Publication Data
Lasnier, France.
Photovoltaic engineering handbook/France Lasnier and Tony Gan
Ang.
568 p. 24 cm.
Includes bibliographical references.
ISBN 0-85274-311-4
1. Photovoltaic power generation. 2. Solar cells. 3. Solar batteries.
I. Gan Ang, Tony. II. Title.
TK2960.L27 1990
621.3T244—dc20 90-4286
Consultant Editor: Dr R H Taylor, CEGB
Published under the Adam Hilger imprint by IOP Publishing Ltd

Techno House, Redcliffe Way, Bristol BS1 6NX, England
335 East 45th Street, New York, NY 10017-3483, USA
Typeset by MS Filmsetting Limited, Frome, Somerset

Printed in Great Britain by Galliard (Printers) Ltd, Great Yarmouth
To Rose, Anita, our sisters, brothers, parents and grandparents
Contents
Preface xix
Part I Technological Processes 1
1 Solar Cell Technologies 3

1.1 Introduction 3
1.2 Large-scale production of silicon base material 7
1.2.1 Conventional technology process 7
1.2.2 Low-cost silicon alternative production processes 10
1.3 Silicon solar cells feedstock material 18
1.3.1 Single-crystal silicon fabrication techniques 18
1.3.2 Polycrystalline silicon fabrication techniques 24
1.3.3 Silicon-cutting and silicon-slicing techniques 36
1.3.4 Surface polishing and cleaning 42
1.3.5 p-n junction formation 43
1.3.6 Electrical contact formation 50
1.3.7 Anti-reflection coating 51
1.3.8 Cell encapsulation 52
1.3.9 Amorphous silicon thin-film technology 53
1.3.10 Options for acquisition of solar photovoltaic
technology by Third-World countries 60
Bibliography 65
Part II The Photovoltaic Generator 67
2 The Photovoltaic Generator 69

2.1 Introduction 69
2.2 Photovoltaic Generator Electrical Characteristics 70
2.2.1 Short-circuit current 71
2.2.2 Open-circuit voltage 71
2.2.3 Maximum-power operation 73
2.3 Loading conditions 73
2.4 Parametric analysis of the photovoltaic generator
current-voltage characteristics 76
viii Contents
2.4.1 Influence of solar irradiance intensity 79

2.4.2 Influence of temperature 80
2.5 Definitions 81
2.5.1 Peak power 82
2.5.2 Conversion efficiency 82
2.5.3 Fill factor 82
Bibliography 82
3 Network of Solar Cells, Modules and Arrays 83

3.1 Introduction 83
3.2 Solar cells in series 83
3.2.1 Identical solar cells in series 84
3.2.2 Non-identical solar cells in series 85
3.3 Solar cells in parallel 88
3.3.1 Identical solar cells in parallel 88
3.3.2 Non-identical solar cells in parallel 89
3.4 The photovoltaic array and its protection 91
3.5 Influence of temperature 92
Bibliography 97
Part III Photovoltaic Systems Engineering 99
4 Storage Batteries for Photovoltaic PowerSystems 101

4.1 Batteries suitable for photovoltaic power systems 101
4.2 Lead-acid batteries 101
4.2.1 Battery types 101
4.2.2 Grid alloy specifications 103
4.3 Lead-acid battery construction and chemistry 104
4.3.1 Battery construction 104
4.3.2 Battery chemistry 109
4.4 Lead-acid battery characteristics 111
4.4.1 Storage capacity 111
4.4.2 Cycle life 114
4.4.3 Charge and discharge voltages 115
4.4.4 Specific gravity of the electrolyte 118
4.4.5 Freezing point of the electrolyte 122
4.4.6 Battery self-discharge 122
4.5 Alkaline batteries 123
4.5.1 Nickel-cadmium batteries 123
4.5.2 Nickel-iron batteries 126
4.5.3 Sodium-sulphur batteries 128
4.6 Battery problem areas 130
4.6.1 Overcharging 130
Contents ix
4.6.2 Undercharging 130

4.6.3 Corroded terminals 130
4.6.4 Short circuits 131
4.6.5 Sulphation 131
4.6.6 Electrolysis and water loss 132
4.7 Battery maintenance and safety precautions 133
4.7.1 Maintenance 133
4.7.2 Safety precautions 134
4.8 Lead-acid battery modelling 135
4.8.1 Battery discharge and charge voltages 135
4.8.2 Decrease in state of charge 136
4.8.3 Water loss 136
4.8.4 Battery life before corrosion sets in 136
4.8.5 Cycle life 137
4.8.6 Cycle life as a function of temperature and depth of
discharge 137
References 138
Bibliography 138
5 Electronic Regulation 139

5.1 Introduction 139
5.2 Stand-alone photovoltaic system regulation 140
5.3 Basic types of regulator 141
5.3.1 Principle of self-regulation 141
5.3.2 Parallel (shunt) regulation 142
5.3.3 Series regulation with a semiconductor 145
5.3.4 Series regulation by electromechanical cut-off 147
5.3.5 Variants on parallel and series regulation 149
5.3.6 Automatic circuit breaking 150
5.3.7 Regulator with dc - dc converter 151
5.4 Conclusion 151
Bibliography 152
6 Power Conditioning 153

6.2 Voltage-receiver load 153
6.3 Current-receiver load 154
6.4 DC-DC converters 154
6.4.1 Buck converters 155
6.4.2 Boost converters 157
6.4.3 Design and sizing of D C -D C converters 158
6.5 Maximum-power point tracker 159
Bibliography 162
X Contents
7 Adaptation of a Positive-displacement Pump Directly Connected to a

Photovoltaic Generator 163
7.2 System characterization 163
7.3 Pulse-width-modulated series-switch step-downregulator 166
7.4 Experimental results 168
7.5 Conclusion 168
Nomenclature 170
Bibliography 170
8 Centrifugal Photovoltaic Pumping: Theoretical Study of Direct-

Coupled Pump 171
8.2 Model of the photovoltaic generator 172
8.3 Maximum-power point 174
8.4 Model of the DC motor 175
8.5 Model of the centrifugal pump 177
8.6 Starting torque of the DC motor 178
8.7 Evaluation of the theoretical operating speed 179
8.7.1 Direct-coupled systems 180
8.7.2 System with maximum-power point tracker 180
8.8 Evaluation of the theoretical efficiency 181
8.9 Results and discussion 182
8.9.1 Theoretical performance of the photovoltaic generator 182
8.9.2 Influence of temperature 183
8.9.3 DC motor analysis 185
8.9.4 Comparison of the theoretical performances of the
direct-coupled and maximum-power point tracker 190
8.9.5 Comparison of daily flow and efficiency 194
Bibliography 197
Part IV Characterization and Testing Methods 199
9 International Test Procedures for Photovoltaic Modules 201

9.2 Performance measurement 202
9.2.1 General principles 202
9.2.2 Reference spectral irradiance distribution 202
9.2.3 General measurement requirements 203
9.2.4 Current-voltage characteristics in natural sunlight 204
9.2.5 Current-voltage characteristics measurement using a
steady-state simulator 205
Contents xi
9.2.6 Current-voltage characteristics measurement using a

pulsed simulator 205
9.2.7 Transposition of current-voltage characteristics 205
9.2.8 Determination of temperature coefficients 206
9.2.9 Determination of internal series resistance 207
9.2.10 Determination of curve correction factor 208
9.2.11 Measurement of relative spectral response 209
9.2.12 Simulator requirements 211
9.2.13 Computation of spectral mismatch error 212
9.2.14 Reference solar cell 214
9.2.15 Calibration methods 214
9.3 Design qualification 216
9.3.1 Visual inspection (VIS) 217
9.3.2 Insulation test (INS) 217
9.3.3 Electrical performance at standard test conditions (PRS) 218
9.3.4 Determination of nominal operating cell temperature
(NCT) 218
9.3.5 Electrical performance at nominal operating cell
temperature (PRN) 218
9.3.6 Hot-spot endurance test (HSP) 218
9.3.7 Energy rating at standard day operation (ERS) 219
9.3.8 Pre-conditioning test (OPC) 219
9.3.9 Ultraviolet radiation test (UVT) 219
9.3.10 High-temperature storage test (HTS) 219
9.3.11 High-temperature high-humidity storage test (HTH) 219
9.3.12 Mounting twist test (MOT) 219
9.3.13 Mechanical loading test (MEL) 220
9.3.14 Hail resistance test (HAL) 220
9.3.15 Temperature-cycling test (TCY) 220
9.4 Module categories 221
9.5 Production acceptance 221
9.6 Quality assurance 221
References 222
10 Asian Institute of Technology Photovoltaic Module Test Bed System 223

10.2 Objective 223
10.3 Parts of the system 224
10.3.1 The photovoltaic module test bed 225
10.3.2 Electronic variable load 227
10.3.3 Ramp generator 229
10.3.4 Data measurement 232
10.4 PVMOD 3.0 234
10.4.1 Data file 237
xii Contents
10.4.2 Data-processing subprograms 238

10.4.3 Photovoltaic array performance subprogram 240
10.4.4 Plot graphs subprogram 242
10.4.5 Analysis subprogram 244
10.5 Conclusion 248
Appendix A p v m o d test bed system specification 248
Nomenclature 249
References 250
Bibliography 250
11 Testing of Photovoltaic Modules under Natural Conditions using the

Asian Institute of Technology Photovoltaic Module Test Bed System 251
11.2 Photovoltaic module test 252
11.2.1 Irradiance time window 253
11.2.2 General performance of the module 254
11.2.3 Module temperature 258
11.2.4 Module response to ambient parameters 260
11.3 Maximum-power point matching 269
11.4 Theoretical model of the test module 271
11.4.1 Prediction of parameters in equation (11.4) 271
11.4.2 Simulation of the tested module with equation (11.4) 277
11.5 Conclusion 280
Nomenclature 284
Reference 284
12 Characterization Procedure of a Photovoltaic Refrigeration System 285

12.2 Characterization procedure for a photovoltaic refrigeration
system 285
12.3 Instrumentation set-up 286
12.4 Experimental observation 287
12.5 Thermal energy balance of the refrigerator enclosure 290
12.6 Conclusion 290
Appendix A Determination of the thermal heat transfer
coefficient of the refrigerator enclosure 292
Nomenclature 295
13 Field Trial Procedure for a Photovoltaic Centrifugal Pumping System 297

13.1 Objective 297
13.2 Approach 297
13.3 Field trial procedure 299
Contents xiii
13.3.1 Characterization of components 300

13.3.2 Evaluation of overall system 302
13.3.3 Complementary tests 302
13.4 Test facilities 305
13.4.1 Outdoor test facility 306
13.4.2 Indoor test facility 307
13.4.3 Sample data sheets 307
13.5 Presentation of results 307
13.5.1 Characterization of components 307
13.5.2 Performance of overall system: instantaneous 313
13.5.3 Performance of overall system: long term 317
Appendix A Pump basics 318
Appendix B Basic equations for a DC motor 325
Appendix C Sensors 326
References 328
Bibliography 328
14 Battery-testing Method for Low-water-loss and Starting, Lighting and

Ignition (Automotive) Batteries 329
14.2 Testing low-water-loss batteries 329
14.3 Testing starting, lighting and ignition (automotive) lead-acid
batteries 332
14.3.1 Testing with adjustable-load tester andhydrometer 332
14.3.2 Testing with fixed-load tester and open circuit 334
Bibliography 336
Part V Sizing Procedure 337
15 The Sizing of Stand-alone Photovoltaic Power Systems: A Dynamic

Simulation Approach 339
15.2 Modelling of system components 340
15.2.1 Solar radiation data base 341
15.2.2 Model for ambient temperature 341
15.2.3 Model for array current-voltage 342
15.2.4 Model for array temperature 342
15.2.5 Model for load 343
15.2.6 Charge regulator 344
15.2.7 Model for battery 344
15.2.8 Back-up power-diesel generator 345
15.2.9 Reliability index of the system 345
15.3 Simulation results and sizing procedure 346
xiv Contents
15.4 Economic analysis of a photovoltaic hybrid system 349

15.5 Discussion 353
15.5.1 Effect of load profiles on sizing 353
15.5.2 Loss of load frequency and battery operations 354
15.6 Summary 356
Nomenclature 357
References 358
Bibliography 359
16 Prediction of Photovoltaic SystemPerformance using Cumulative

Frequency Curves of Irradiance 360
16.2 Daily and seasonal variations in the battery'sstate of charge 360
16.3 The significance of cumulative frequency 363
16.4 Obtaining cumulative frequency curves for irradiance values 364
16.5 Fractional time F 364
16.6 Description of photovoltaic system 365
16.6.1 Modules 365
16.6.2 Battery 366
16.6.3 Load consumption 366
16.7 Performance of the system 366
16.7.1 Ambient temperature 366
16.7.2 Performance parameters and indicators 367
16.8 Comparison and discussion of performance 368
References 370
Bibliography 370
Part VI Economic Analysis 371
17 Financial Evaluation of Renewable Energy Projects 373

17.2 Techno-economic system optimization 374
17.3 Data requirements 376
17.3.1 Interest rate 376
17.3.2 Discount rate 377
17.3.3 Inflation rate 379
17.3.4 Energy escalation rate 380
17.3.5 Service life of the installation 380
17.3.6 Investment costs 381
17.3.7 Residual value of plant 382
17.3.8 Major operating costs 382
17.3.9 Subsidies 383
17.3.10 Depreciation 383
Contents xv
17.4 Procedure of financial analysis 385

17.4.1 Net present value 388
17.4.2 Internal rate of return 390
17.4.3 Annuity method 393
17.4.4 Cost annuity comparison method 394
17.4.5 Pay-back calculation 396
17.5 Sensitivity analysis 397
Appendix A Case study of Thai village pumping system for
drinking water 398
Bibliography 399
18 Life-cycle Cost Comparisons of Photovoltaic-powered Systems and

Most Likely Conventional Alternative System 400
18.2 Methodology 400
18.2.1 Common base-case assumptions 401
18.3 Water-pumping analysis 406
18.3.1 Description of the base case 407
18.3.2 System costs 409
18.3.3 20 year life-cycle costs 413
18.3.4 Sensitivity analyses 413
18.4 Communication analysis 420
18.5 Vaccine refrigeration analysis 430
18.5.1 Description of the basic case 432
18.6 Lighting and home power analysis 444
18.7 Multi-use analysis 457
Appendix A Sizing and financial models 466
Reference 474
xvi Contents
19 A Comparative Assessment of Photovoltaics and Handpumps for

Rural Water Supply: A Case Study for Thailand 475
19.2 Technology configurations 475
19.3 Methodology 476
19.4 Base-case assumption 479
19.5 Case study for a shallow well (3 m head) 485
19.6 Case study for a shallow well (5 m head) 486
19.7 Case study for a deep well (20 m head) 488
19.8 Conclusion 491
References 492
Bibliography 492
20 Life-cycle Cost Comparison of an Alternative Power Supply for a

Portable Pocket-sized Stereo Cassette Tape Recorder 493
20.2 Description of the three alternative power supplies 494
20.2.1 Disposable alkaline primary cell 494
20.2.2 Rechargeable nickel-cadmium secondary cell with
grid electricity 495
20.2.3 Rechargeable nickel-cadmium secondary cell with
photovoltaic battery charger 495
20.3 Life-cycle cost comparison 496
20.3.1 Sensitivity to capital cost 498
20.3.2 Sensitivity to discount and inflation rate 499
Bibliography 500
Part VII Instrumentation 501
21 A Simple Metal-Oxide-Semiconductor Field-effect Transistor

Electronic Variable Load 503
21.2 Variable load 504
21.3 Ramp generator 506
21.4 Data acquisition interface 508
21.5 Photovoltaic array sensor 512
Appendix A Photovoltaic array sensor specifications 514
Bibliography 515
22 Equipment Accommodation 516

22.2 Equipment accommodation for remote and rural areas 516
22.2.1 Heavy-weight buildings 517
Contents xvii
22.2.2 Thermally thin buildings 517

22.2.3 Cabinets 518
22.3 Thermal performance 519
Bibliography 521
23 Terminology 522
23.1 Photovoltaic systems terminology 522
23.2 Battery terminology 529
23.3 Economic terminology 534
23.4 Solar geometry 537
23.5 Radiation terms 538
23.6 Radiation quantities 539
23.7 Radiation measurement 540
23.8 Radiation properties and processes 540
Index 543
Preface
This book is intended to be a comprehensive introduction to photovoltaic

( p v ) technology and systems engineering. Six years of teaching and research
experience at the Asian Institute of Technology (AIT)—a regional multi
cultural post-graduate technological institute located near Bangkok,
Thailand—have contributed to this achievement.
This book has been elaborated within the regional socio-economic scenery
which frames the industrial development of South and East Asian
economies, is aimed at assisting the growth of p v industry, and the dissemin
ation of p v generators in rural areas in particular. It provides the necessary
technical knowledge which is a preliminary condition for acquisition of this
technology by less developed countries.
Through a theoretical, practical and multi-disciplinary approach, it pro
vides a broad understanding of p v industrial options, system-designing
methods, economic appraisal as well as components field trial procedures
and characterization methods.
Owing to its multi-disciplinary approach, the authors believe this book will
meet the need of a broad range of readers in post-graduate institutions and
technical colleges, and will serve as a reference book in electronics and
electrical courses as well as in energy economics and planning and energy
technology related fields of study.
The materials contained in this book encompass the overall spectrum of p v
technology starting from the solar-cell-manufacturing processes to the analy
sis of the entire system of energy production. It is basically divided into seven
parts, i.e. technological processes, the p v generator, systems engineering, p v
characterization and testing methods, p v sizing procedure, economic analysis
and instrumentation. The order adopted is aimed at a progressive and global
understanding of photovoltaics and each part constitutes specifically a
perfect and independent learning entity.
Acknowledgment should be made to the world community of researchers
whose different publications and sharing of findings in international and
regional forums contributed to make the bibliographical sections of this
book well supported and densely informative. References and bibliographies
XX Preface
are listed at the end of each chapter for further in-depth research work.
Although many individuals have participated in one way or another, by
question or comment, to the quality of the present edition, a special word of
thanks has to be addressed to AIT students who helped the authors to find
the right focus of this book, i.e. to answer the regional needs in photovoltaics.
France Lasnier
Tony Gan Ang
June 1989
Part I
Technological Processes
1 Solar Cell Technologies
1.1 INTRODUCTION
It has been over 150 years since Becquerel in 1839 first discovered that a
photovoltage was developed when light was directed onto one of the
electrodes in an electrolyte solution. Adams and Day were the first to observe
the photovoltaic ( p v ) effect in a solid (selenium) in 1877. Early solid-state
researchers including Lange, Grondahl and Schottky did pioneering work on
selenium and cuprous oxide p v cells. However, it was not until 1954 that
scientific literature published results on the use of the p v effect in energy
conversion processes.
The wide spectrum of semiconductor-related technologies encompasses
the field of solar cells (see figure 1.5) and all manufacturing processes
described below are derived from and enjoy the latest advances made in
microelectronics. From this point of view, p v converters belong to the so-
called high-technology field and any improvement in that field contributes to
the progress in semiconductor technology as a whole.
The silicon solar-cell-manufacturing process involves basically three
stages.
(i) Material preparation and shaping.

(ii) Cell processing.
(iii) Cell interconnection and encapsulation.
The conventional technological process, the flow chart of which is presen

ted in figure 1.1, relies conventionally on the high-purity semiconductor-grade
silicon (SG-Si) available from electronics industry wastes. In this process,
quartzite as a raw material is reduced into metallurgical-grade silicon
(MG-Si), the purity of which lies in the 98-99% range. MG-Si is purified to
qualify as the so-called SG-Si or electronic-grade silicon (EG-Si). SG-Si is then
melted in a crucible and pulled out to form a single-crystal silicon ingot
by using the traditional Czochralski (cz) technique (figure 1.2). Ingots are
cut into wafers which are etched to remove the damage induced during
the slicing operation. Wafers are then processed into a complete solar cell
4 Solar cell technologies
(figure 1.3), i.e. junction diffusion, contact elaboration and anti-reflection

coating. Finally, cells are interconnected to achieve a practically useful
voltage and encapsulated into a module by using a lamination technique.
In 1982, silicon material contributed nearly 50% of the total cost of a p v
module whereas cell processing and module assembly contributed respec
tively 30% and 20% of the total cost breakdown. Since that time, critical
paths in the silicon technology process have been investigated intensively,
leading to major technical breakthrough and substantial cost reduction as
well as the definition of various specific silicon routes for solar cell fabric
ation (figures 1.4 and 1.5).
Figure 1.2 Production and slicing of silicon ingots.

Introduction
Solar Cells
Silicon- Chips
for Electronic
Industry
i
Ingot
j r
T hin Sheets Ribbon
i
Heat
i
Other
Casting Film s Exchange Approaches
Nearly D irectional
Single S o lid ific a tio n
Crystal
Sheets Die cost Big Thin Other

Grains Film s approaches
Hydrogenated Si - H Thin Ceram ic

Film s substrates
on other
substrates
Figure 1.4 Silicon and its various routes of utilization, particularly the solar-cell
route.
Figure 1.5 Spectrum of silicon technology.
A review of recent progress includes the following.

(i) Material.
(a) Preparation and definition of ‘solar-grade’ silicon (soG-Si) material
as a low-cost and solar-specific alternative to the expensive
semiconductor-grade material which will soon be depleted.
(b) Development of the polycrystalline technology route based on soG-Si
as an industrial alternative to the single-crystal technology and
including the ingot-casting technique, optimized slicing techniques
and, in particular, the continuous process to replace batch processes.
(c) Research and development efforts in sheet technology (ribbon, foil,
thick films, etc).
(ii) Cell processing.
(a) Surface texturization to decrease optical losses in single-crystal cells.
(b) Cold-junction processing ( c j p ) in which doping impurities are intro
duced by ion or molecular implantation or evaporated or spray
deposited on the wafer surface. They are then driven into the material
to form a p-n junction by pulsed annealing (laser or electron beam);
this process is amenable to full automation.
(c) Screen printing and to a lesser extent electroplating for deposition of
contacts. Research is being carried out to replace noble materials
(silver) by low-cost alternative materials (aluminium).
Moreover, substantial cost reductions have been achieved through process
automation, and it is widely recognized that the industrial dimension is of
paramount importance for the economic use of p v converters.
A further cost reduction scenario implies more fundamental changes and,
in the year 2000, amorphous silicon (a-Si) (or other material) thin-film
technology is anticipated to emerge as a key technology for future low-cost
Large-scale production of silicon base material 7
mass production of solar cells. In the last few years, remarkable progress has
been seen in both physics and technology development of hydrogenated
amorphous silicon (a-Si:H) solar cells which in a very short period has moved
through all the phases from discovery to industrial production. Nowadays
a-Si solar cells have a considerable market size particularly in the consumer
electronics field such as pocket-sized calculators, wrist watches and battery
rechargers. Although a-Si technical readiness has still to be proven to enter
the power production sector, the basics of the process are reviewed in the
present chapter.
1.2 LARGE-SCALE PRODUCTION OF SILICON BASE

MATERIAL
1.2.1 Conventional technology process

1.2.1.1 Quartzite to metallurgical-grade silicon
After oxygen, silicon is the most abundant element in the Earth’s crust.
Geological processes have provided different ways of concentrating silicon
dioxide in vast deposits all over the world. Quartzite rock and special strata
of quartz sand have been the most widely used raw materials for the
production of elemental silicon. Their inexpensive mining requires little
energy and has negligible environmental impact. The quality ranges of either
genuine or refined silicon dioxide have been computed in table 1.1. The
preference or necessity to use a certain quality depends largely on the type of
reduction and refining process sequence and possibly the geographical
Table 1.1 Different types of S i02 sources.
Mining Upgrading Nature of Impurities

product particles (wt.%)
Quartzite rock Standard Chunks «1

(technical grade)
Special silicates Chemical conversion Powder <0.3
Quartz sand Standard Grains <0.2

(technical grade)
Selected quartz Chemomechanical Grains <0.05
sands, pegmatites treatment
Special quartz veins, Standard Nuggets <0.03
rock crystal
situation. Prices of quartz product can be 50 times more expensive when

obtained from rock crystal (US$0.2 kg-1 in 1979) as opposed to selected
quartz sand.
Present commercial extraction processes use the crystalline form of silicon
dioxide, quartzite, with 90% or more silica (Si02) rather than sand whose
major constituent is also silica. Sand with its high amount of impurities is too
costly to process and therefore does not make an economic choice as a raw
material. The quartzite is reduced in large arc furnaces (figure 1.6) by carbon
to produce elemental silicon according to the following chemical reaction:
S i02(s) + 2C(s)—►Si(l) + 2CO(g).
Arc Furnace
automatic vertical regulator
The vapour reacts with impurities such as aluminium, carbon and mag
nesium which then leave the silicon as part of the gas. Through the reduction
of quartzite with coal, MG-Si is produced at a cost of US$1 kg" 1. The purity
of MG-Si is low (98-99%) owing to the low purity of the starting materials.
1.2.1.2 Metallurgical-grade silicon to electronic-grade silicon

The Siemens C process is at present the standard method for purifying MG-Si
to EG-Si. The silicon produced by this method is 99.9999% pure (electronic-
grade) polycrystalline form.
Si + 3HC1—>SiHCl3 + H 2
SiHCl3 + H 2—>Si + 3HC1.
In this process, trichlorosilane (SiHCl3) is first obtained. A bed of fine MG-

Si particles is fluidized and chlorinated with hydrochloric acid in the presence
of a copper catalyst to induce the reaction. The low purity of the MG-Si results
in impure SiHCl3 which must then undergo subsequent fractional distillation
to reduce impurities by a factor of 104-105. A chemical vapour deposition
( cv d ) method is subsequently used to isolate the pure silicon from the
electronic-grade SiHCl3. Vaporized SiHCl3 is decomposed and reduced with
hydrogen at about 1000 °C, resulting in the deposition of silicon precipitates
on the hot surface of a bridge in the shape of an inverted U. The bridge is
made of slim silicon rods and has been heated in a reactor by passing an
electric current through it. By this process, six polycrystalline silicon rods of
length 1 m and diameter 12 cm can be produced simultaneously.
The price increase from US$1 kg-1 for MG-Si to more than US$70 kg-1
for EG-Si reflects the costliness and energy-intensiveness of the Siemens C
Figure 1.7 Purification of MG-Si to make silicon wafers.

process, the most energy intensive being the last step, the deposition of silicon
from the gas phase (150 kWh (kg Si)-1). In addition, the deposition process
is characterized by a low yield (26%), a low throughput and a discontinuous
operation (figure 1.7). Analysis of the present module costs show that the
manufacturing of the silicon substrate alone amounts to nearly 50% of the
total cost per peak watt with the remainder being costs of cell fabrication and
packaging.
1.2.2 Low-cost silicon alternative production processes

The present worldwide production of crude silicon (Eastern European
countries excluded) is roughly identical by volume with the annual world
production rates of copper and zinc. Only 0.5% of it is used in the
semiconductor industry. Roughly 5% goes into silicon production whereas
the remaining is consumed by steel and aluminium industries in the form of
alloys or ‘silicon metal’.
Until now the p v industry has relied upon an inexpensive raw material, i.e.
electronics industry wastes, which constitutes a deposit of about
400 t year - 1, used to manufacture 50 MWp year - 1. This raw material capac
ity will be sufficient up to 1992 or 1995. However, owing to the restricted
nature of this deposit as well as the forecasted growth of the p v industry
itself, alternative technologies will have to produce a large quantity of
necessary silicon.
Different ways have been developed or re-explored to manufacture an
alternative low-cost silicon material. The target of these worldwide efforts is
to provide a bulk material characterized by the following features: abun
dance of resources, low-cost manufacturing (US$10 kg-1), impurity opt
imization (solar grade) and low-energy processing.
Figures 1.8 and 1.9 present flow charts of the different technological
processes investigated. Figure 1.8 shows a low-purity approach, i.e. all those
processes which start with MG-Si or at a corresponding impurity level. Figure
1.9 characterizes a high-purity approach based on an upgraded silicon
dioxide (Si02) starting material. Moreover two types of process sequence can
be distinguished: whether they are relying upon silicon dioxide (Si02) as a
primary silicon source or volatile compounds as secondary silicon sources.
All volatile products have one property in common that makes them very
attractive, namely the ease of refining, e.g. by fractionated distillation or
desorption. Among the volatile halides, silicon tetrafluoride and chloride are
the most economically interesting compounds. Monosilane (SiH4) can be
considered as an at least tertiary silicon source and characterized largely as
an intermediate product. These various processes produce intermediate
qualities of silicon (figure 1.10) which aim at further refining to reach the
‘solar-grade’ quality, i.e. a solar cell feedstock material qualified for the
manufacturing of solar cells with a high conversion efficiency (12%).
Large-scale production of silicon base material
Figure 1.8 Low-purity approaches to large-scale silicon production:

lp , low purity; HP, high purity; sog , solar grade.
Figure 1.9 High-purity approaches to large-scale silicon production:

h p, high purity.
Figure 1.10 Different categories of silicon quality: —, concentrations of fast-

diffusing metals.
1.2.2.1 The 'solar-grade' silicon concept

The SOG-Si concept comes out of the need for the p v industry to develop a
low-cost silicon material to replace the EG-Si material.
The development of SOG-Si came about as the result of the apparent
insensitivity of medium-efficiency solar cells to the presence of some typical
impurities; for example, copper and aluminium in n-type and phosphorus
and copper in p-type material which can be tolerated in excess of 1 ppm
without major cell degradation.
SOG-Si material has not to be seen as a second-rate material but as a specific
material for efficient solar cell production. Furthermore it is more accurate to
speak of solar-grade ‘synthesis’ through bulk segregration of impurities
during a unidirectional solidification process (figure 1.11).
In principle, each kind of crystal growth process, e.g. cz, floating-zone
( f z ), directional solidification or sheet growth, results in material purifi
cation, the amount of which differs for the different growth processes.
Appropriate silicon qualities should be associated with respective appropri
ate growth processes to produce SOG-Si material (figure 1.12).
Various processes in different stages of development are being explored to
produce partially refined silicon material for SOG-Si synthesis from a variety
of inexpensive industrially manufactured starting materials such as MG-Si,
hexafluorosilicic acid which is an inexpensive by-product of the phosphate
fertilizer industry and quartz sand (Si02).
Figure 1.11 Solar-grade synthesis from partially refined silicon by direc

tional solidification. For meanings of abbreviations, see text.
Figure 1.12 Different large-scale process versions to form solar-cell-

based material (foils or slices).
1.2.2.2 S i0 2 as the primary silicon source

The advanced carbothermic reduction process (Siemens; Dow Corning)
An alternative process for producing soG-Si is the advanced carbothermic
reduction ( a c r ) process. It requires the production of high-purity silica and
carbon as starting raw materials. The high-purity silica is obtained from
inexpensive quartz sand through a glass-melting-fibre-leaching process
(figure 1.13) while the high-purity carbon is obtained by leaching carbon
Crucible
Gloss Melt
Heating Element
Fiber
Thickness Gauge
Pulling Mechanism
Figure 1.13 Glass-melting and fibre-drawing apparatus.
black pellets with a hot HC1 solution. The basic chemical reaction of the
process is
S i02(s) + 2C(s)—>Si(l) + 2CO(g).
Figure 1.14 shows a 550 kV A three-electrode arc furnace used for the
carbothermic reduction of the purified silica and carbon in a pilot-scale
3 Graphite Electrodes
Si02 + C
High-Purity Graphite Lining

Heat Insulation
Tapping Spout
Mold for Liquid Silicon
Figure 1.14 Schematic diagram of the 550 kV A arc furnace used in the carbothermic
production of soG-Si (Siemens).
operation. With an inside diameter of 1.2 m it has a capacity for processing

I.2 tons of silica and carbon. Silicon is produced at an approximate rate of
15 k g h " 1, collected in graphite mould and then transferred into a cz crystal
puller for further purification. A cost of about US$10 k g " 1 for the soG-Si is
estimated for an industrial production level of 1000 t Si y ear"1 or more. The
best results for solar cells produced from single-crystal ingots obtained from
one cz pull were 13.4% (air mass 1 ( a m I ) ) for 2 cm x 2 cm test cells and
II.1% for cells 100 mm in diameter. A similar process using the carbo
thermic reduction of silica to produce soG-Si is used by the Dow Corning
Corporation and is presented in figure 1.15.
Use of Pure Raw

M aterials
S i Produ ction via
S i P u rific a tio n to
Fin al P o ly cry sta llin e
S o G - S i P ro d u ct
Figure 1.15 Schematic drawing of the Dow Corning process.
The aluminothermic reduction o f quartz sand (Heliotronic GmbH; Union

Carbide )
The production of silicon by aluminothermic reduction of S i02 does not
utilize chunks of quartz rock as required in the conventional electric arc
process but is based on particulate matter such as powder or sand, the latter
being available in large quantities and high purity.
The reduction of quartz with aluminium is a pronounced exothermic
reaction taking place in a temperature range 1100-1200 °C:
3Si02(l) + 4Al(0->2Al2O3(l) + 3Si(s).
A suitable technical process on this basis (figure 1.16) requires liquid
phases of both oxidic and metallic nature acting as solvents for the reaction
products. The two liquid and immiscible systems are represented by an
aluminium-silicon alloy and an Al2S3-A l20 3- S i0 2 slag system. Silicon is
S t ir r e r
Figure 1.16 Processing sequence of the aluminothermic reduction.
obtained in the form of platelets with typical impurity levels lower than
10 at.ppm. Prior to further refining, the product is milled and leached.
The semi-continuous character of the process is one of the preconditions
for guaranteeing low-cost production. Thus recycling of aluminium by
recrystallizing the by-product (A120 3) from the slag system and reducing it
by electrolysis is a rather important part of the process.
1.2.2.3 Volatile compounds as secondary silicon sources

A different class of reduction processes stems from the conversion of silicon
halides (halosilanes) into high-purity silicon. The following section describes
a composite of unique processes which aims at producing a low-cost high-
purity silicon for solar cell manufacturing. Some of these processes are
already at the ‘pilot plant’ stage. To some extent, the future potential of cost-
effective silicon p v cells is directly related to the success of one of these
emerging alternative silicon technologies.
Zinc reduction o f SiCl4: Bat telle Colombus Laboratories
In reviving the so-called duPont process of the early 1950s, a fluidized-bed
rea cto r ( f b r ) w a s a d a p te d to h a rv est silicon :
2Zn + SiCl4- ^ S i + 2ZnCl2.

Furthermore a non-aqueous electrolysis step for recovery of zinc was
incorporated:
ZnCl2 electrolysis +
In the proposed 50 m tyear-1 plant, chlorine gas (Cl2) will be reacted with
soda (NaOH) to form (NaOCl) a saleable product:

Cl2 + 2NaOH—NaOCl + NaCl + H 20 .
This process is a continuous multi-step closed-loop process in which the
raw feed is HG-Si and (SiCl4) the secondary source.
Silaneprocess (Union Carbide Corporation)
The Union Carbide process under development is a multi-step continuous
closed-loop process in which the raw material is MG-Si.
The primary-process steps are as follows.
(i) The preparation of chlorosilanes, predominantly (HSiCl3) by hydro
genation of a mixture of SiCl4, which is a recycled by-product, and MG-
Si:
Si + 3SiCl4 + 2H2 «c: : y„ MHSiCl,.
(ii) The two-step disproportionation of HSiCl3 into SiH4 with formation
of SiCl4 as the by-product:
2HSiCl3 ™lnc c,1,lys* H 2SiCl2 + SiCl4
2H2SiCl2 r inc c*““y" SiH4 + SiCl4.
(iii) The conversion of silane (SiH4) into silicon. Two deposition reactions
have been investigated, i.e. the free-space reactor and the f b r , the
fluidized-bed approach being the most attractive for a continuous
process. The silane process developed by Union Carbide is already in
the pilot (100 m tyear-1) and production plant (1200 m tyear-1)
stages. The flow diagram of the silane process is shown in figure 1.17.
Figure 1.17 Union Carbide silane process flow diagram for the production of
soG-Si.
Fluosilicic acid (SiH 2F6) reduction (S R I International)

The SRI process under development is the reduction of SiF4 with sodium.
SiF4 is obtained at a low cost from H 2SiF6 which is a by-product of the
fertilizer industry.
1.3 SILICON SOLAR CELLS FEEDSTOCK MATERIAL
An important effort has been devoted to the definition and the preparation of
soG-Si for replacing the expensive EG-Si. However, solar-grade as well as
electronic-grade materials have to be shaped prior to being processed for p v
cells production (figure 1.12).
Two kinds of method to produce silicon solar cell feedstock material are
investigated although their development stages are quite different.
(i) Bulk growth, leading to ingots which have to be cut into wafers with
optimized slicing techniques.
(ii) Ribbon pulling for which all the limitations bound to slicing are
definitely overcome. The silicon consumption in ribbon-pulling tech
niques is between two and eight times smaller than in sliced wafers
depending upon the different processes.
There is of course an economic advantage to using ribbons as against
ingots (to minimize waste material) but the ribbon-pulling techniques are
much less mature and most of them still have to be proved viable before
being industrialized. Moreover, bulk growth techniques are already in
industrial production, starting with a semiconductor-grade as well as an
electronic-grade feedstock material. Bulk growth techniques can then be
divided into two approaches.
(i) Ingot-pulling technology.
(ii) Ingot-casting technology.
In casting technology, the material is shaped by using a unidirectional
crystallization process in a fixed crucible. For EG-Si casting the process
involves a seed of crystalline silicon which is not required in the soG-Si
casting process.
Finally, two technological routes have to be distinguished. In the first case
leading technological processes for making single-crystal ingots are con
sidered. The second case corresponds to the elaboration of polycrystalline
structures of any kind since attractive physical and economical properties of
such material have been corroborated worldwide. The a-Si thin-film tech
nology, although specifically applied today for the consumer electronics
market, has yet to acquire the technical readiness which will present it as a
major competitor in the p v power market and this will be reviewed at the end
of this chapter.
1.3.1 Single-crystal silicon fabrication techniques

Crystal-growing processes produce single-crystal silicon in the form of
ingots, sheets or ribbons. The cz and f z techniques grow ingots from molten
silicon starting from single-crystal seeds, which must then be sawn into
Silicon solar cells feedstock material 19
wafers. New fabrication processes such as dendritic web growth and edge-
defined film-fed growth ( efg ) attempt to reduce cost by growing ribbons that
can be directly formed into usable wafers:
—c z and advanced c z methods

— ingot — — fz technique
-heat exchanger m ethod (hem)
single-crystal- -sh eet dendritic web growth
silicon
— ribbon----- efg .
1.3.L I Ingot-pulling technology

Czochralski and advanced Czochralski methods
The cz method is still the major commercial method of manufacturing ingots
of single-crystal silicon. A seed crystal of silicon is slowly withdrawn under
rotation from a bed of molten silicon contained in a crucible (figure 1.18(a)).
If the rate of rotation and pulling (normally less than 10 c m h " 1) is carefully
controlled, the pulled rod will solidify upon cooling into one large crystal
ingot of typically 7.5 cm diameter and 1 m in length. Normally, a
p-type crystal is manufactured directly by adding boron to the melt. A
schematic diagram of a conventional cz furnace is shown in figure 1.19.
The cz method is effective in segregating impurities as the impurities in the
molten silicon tend to remain in the melt rather than be drawn into the
crystal. Boron does not become segregated as much as other impurities and
becomes incorporated into the silicon crystalline structure, resulting in a
p-type crystal.
Significant cost reductions have been attained from improvements in the
Single - Crystal Polycrystalline

— Silicon Ingot Silicon
Graphite
^ Susceptor
Direction of Movement
Molten Silicon of Heater Winding
S i0 2 Liner S in g le -C r y s ta l Ingot
S in g le -C r y s ta l Seed
Figure 1.18 Converting polycrystalline silicon to single-crystal silicon:

(a) cz growth apparatus; (b) fz processing.
-P u ll rod
'o' R ing seal
Cooling water
x P ip e fla n g e seal
-Q u a r tz casin g
^Molybdenum chuck
.'See d crystal
^Q uartz crucible
.'Melt
.'Carbon sleeve
^Q uartz liner
1 A d ju stab le
; Cooling water
- G a s out
Therm ocouple
x Q uartz T. C. well
Figure 1.19 Diagram of a conventional cz furnace.
cz process which include a rechargeable feature allowing continuous oper

ation. Crystals with diameters as large as 15 cm and of length 1.5 m have
been grown using the continuous process. The rechargeable process also
allows the crucible to be re-used several times, thus reducing the cost of
crucible replacement. Subsequent sawing of the round or rectangular ingots
with diamond blades or slurry result in standard finished wafers 0.2-0.35 mm
thick. Kerf and etching losses are still greater than 50%.
Floating-zone technique
An ingot of polycrystalline silicon and a single-crystal silicon seed are needed
to start the process. The polycrystalline ingot is placed on top of the seed
and the interface between the two is melted by movable heating coils
(figure 1.18(6)). As the heating coils are moved up the stationary ingot, the
material below the molten interface solidifies into single-crystal silicon. The
process is called the ‘floating-zone’ process since the molten silicon is
supported only by its own surface tension. To ensure perfect uniformity of
the crystal, the heating coils may be moved up and down in several cycles.
A disadvantage is that the f z technique is less able to segregate impurities
than is the cz process since the volume of molten silicon is much less. On the
other hand, the contamination of the silicon crystal material from contact
with crucible material is avoided as the f z method is crucible free. In the cz
method, contamination of the silicon by oxygen can occur during the crystal-
pulling process.
1.3.1.2 Ingot-casting technology

Heat exchanger method
The hem is a low-cost directional solidification technique introduced by
Crystal Systems. It has been used for commercial production of sapphire
32 cm in diameter and of mass 50 kg. This method has been adapted for the
growth of silicon crystals and is actually a unique technique which gives a
single-crystal silicon ingot of square cross section.
In the hem process, molten silicon is allowed to freeze from a seed crystal
to form a nearly perfect cubic single-crystal ingot under highly controlled
conditions. The seed crystal is placed at the bottom of the crucible (made of
silica with a graded structure) which is seated on a high-temperature heat
exchanger. The seed is prevented from melting by forcing gaseous helium
through the heat exchanger. The hem is therefore the only directional
solidification technique in which there is independent control of the tempera
ture gradients in the solid and liquid with no moving parts. The ingots are
cast in a silicon crucible which breaks apart upon cooling. The process which
involves a helium gas heat exchanger has resulted in 15.5% efficient rect
angular solar cells fabricated from the ingot. The Crystal Systems process
produces silicon blocks of 34 cm x 34 cm x 17 cm which are more than 95%
single crystal at a 1.9 kg h " 1 growth rate with a yield of higher than 95%.
1.3.1.3 Ribbon growth: edge-definedfilm-fed growth

The efg method developed by Mobil Solar employs graphite shaping dies to
shape the crystal growth to a desired form as well as to separate the growth
interface from the main melt surface (figure 1.20). The best shaping dies for
silicon are made from dense high-purity graphite. As molten silicon rises by
capillary action in the die, it is drawn through a narrow slot, producing a
single-crystal ribbon. Mobil has achieved 40cm 2m in-1 growth of ribbons
10 cm wide. Cells with more than 12% efficiency have been produced from
simultaneous multi-ribbon growth of three ribbons at 10 cm width or five
ribbons at 5 cm width and 0.15 mm ribbon thickness.
The problem of edge control is avoided in a new growth technique by the
efg process whereby a hollow polygonal section shaped tube called a
nonagon of 150 mm diameter with nine flat sides, each of 50 mm width and
0.3 mm thickness, is grown. A capillary shaping die achieves shape control
over all crystal dimensions. Large-area growth rates are possible as a
consequence of the absence of discontinuities in the meniscus. Very wide
sheets of silicon can also be grown without the need for accurate control of
temperature over long linear distances. After growth, the nonagon is cut into
flat segments for solar cell fabrication by the use of high-intensity C 0 2 lasers
or high-speed diamond saws.
Inherent advantages of the efg process include efficient utilization of
Figure 1.20 efg : a seed crystal of single-crystal silicon contacts the

surface of the melt as it gets drawn through the die by capillary action.
As the crystal is slowly raised (at about 15 cm m in'1), the silicon cools
and freezes into a single-crystal ribbon.
silicon material with no loss involved in slicing or machining operations and

efficient utilization of consumable materials such as crucibles, shaping dies
and insulation elements. Disadvantages include problems with impurity
segregation which requires that the crystal ribbon be drawn from a highly
pure silicon melt and contamination by carbon impurities of the shaping
dies.
1.3.1.4 Sheet growth: dendritic web growth

A silicon ribbon forms between a pair of dendrites which provide stability to
the edges for width control (figure 1.21). The dendrites of silicon propagate
into a supercooled silicon melt during growth to shape the ribbon. This
process requires that the isotherm at the point of solidification be precisely
shaped so that supercooling under the dendrites takes place without super
cooling the solid-liquid interface where ribbon growth occurs (figure 1.22).
The method avoids the problem of impurity contamination since no die or
crucible is used and has yielded best cell efficiencies of 16.9%. The dendrites
cannot be incorporated into the final cell structure and must be cut off.
Disadvantages include necessity for precise temperature control and slow
growth rates. The dendritic web growth by Westinghouse Corporation has
achieved growth rates of 13 cm2 m in- 1 for short ribbon lengths. The ribbon
Figure 1.21 Dendritic web growth: a web of single-crystal silicon forms

as dendritic seed crystals are withdrawn from the melt.
surfaces are nearly perfect crystallographic (111) facets which require no

mechanical treatment prior to device fabrication. The growth apparatus
consists of a furnace chamber, a molybdenum susceptor which holds the
fused quartz crucible, a molybdenum lid assembly with a slot through which
the ribbon is withdrawn, and a reel which serves as a pull mechanism as well
as material storage.
* Bounding Dendrites
Vv.
Liquid
Position
Figure 1.22 Schematic illustration of a growing web compared with the

temperature profile at melt surface.
1.3.2 Polycrystalline silicon fabrication techniques

Polycrystalline silicon is made up of many grains of single-crystal silicon
whose size and quality determine solar cell performance (figure 1.23). The
grains should be as large as possible and should preferably be oriented
vertically running from top to bottom of the cell. Grain boundaries are then
less likely to hinder the flow of light-generated charge carriers and are
therefore less likely to cause recombination (figure 1.24).
O Silicon Atom S o |jd sin gle-

0 = 0 Bond Between Crystal
Atoms Material.
= Bound Electrons '
Figure 1.23 (a) Polycrystalline silicon is made up of randomly packed

grains, each of which is (b) single-crystal silicon. The disorder of
polycrystalline silicon is responsible for its performance problems.
(/um) Light
9 Electron
® Hole
Figure 1.24 (a) The lifetime of free charge carriers (which indicates the
time that they are likely to be free before recombining) is strongly
dependent on grain size in polycrystalline silicon. Since it is important
to keep charge carrier combination to a minimum, large-grained poly
crystalline material is necessary for good cell performance. (b) Wide
vertical grains spanning a cell from top to bottom aid cell performance
because light-generated carriers tend to flow vertically without en
countering grain boundaries.
In a process called directional solidification used to form columnar grains

(about a centimetre in size), molten silicon is cooled slowly along one
direction so that, as the silicon cools, the solidifying crystal structures are
orientated relatively uniformly along one direction. This process is analogous
to the hem method but is a seedless process.
The grain size is the major determinant of polycrystalline silicon cell
characteristics and efficiencies. To achieve efficiencies above 10%, grain sizes
greater than 0.5 mm are needed and grain boundaries must be orientated
perpendicular to the film.
Almost all methods used to make single-crystal silicon described in the
previous section can also be used to manufacture polycrystalline silicon
under less controlled conditions. Two additional methods unique to poly
crystalline silicon exist: edge-supported pulling ( esp ) and the sheet growth
methods. Also, cell fabrication processes after crystal growth (junction
formation, contacts and antireflection coating) are similar to those for single
crystal cells. The various growth methods are given below:
— heat exchanger
-Silso
—ingot- -casting—
method -Semix
-Polyx
- efg
-dendritic web growth
polycrystalline- -ribbon- -ribbon-to-ribbon growth
silicon -ESP
-ribbon-against-drop (rad ) process
-silicon-on-ceramic method
-silicon-on-silicon (sos) method
-silicon on inexpensive substrates
-sheet-
-supported web method
-ramp-assisted foil-casting technique (raft)
-horizontal supported web (hsw ).
1.3.2.1 Ingot-casting technology

During the past 10 years it has been demonstrated that melt-grown poly
crystalline silicon could be used for the fabrication of crystalline silicon
wafers as a base material (figure 1.25). Thus low-cost technologies which
produce, via a controlled directional solidification, multi-crystalline silicon
ingots provide a viable alternative to the conventional cz technique. The
Figure 1.25 Polycrystalline silicon solar cell route.
relatively low crystallization rate inherent to all ingot technologies can be

utilized for an in situ purification step, easing off the quality criteria for the
starting material that is soG-Si. On the other hand, the casting process
requires additional cutting and wafering steps which imply extra costs and
loss of silicon. A few technologies are described below with emphasis on
crucible re-usability.
Heat exchanger method (Crystal Systems)
The method developed by Crystal Systems is capable of producing poly
crystalline silicon solar cells of up to 15% efficiency using a seedless process.
Semix process ( Semix)
The ingot-casting process by Semix produces a granular or polycrystalline
ingot with large grain size and high intragrain order solidified from a vat of
molten silicon with or without a single-crystal seed at the bottom of the vat.
This method has been found to be effective in separating impurities and has
produced solar cells of up to 15% efficiency similar to the hem method.
Silso process (Wacker)
Polycrystalline ingots are produced through directional solidification in pre
heated moulds where the silicon has been melted previously in a quartz
crucible. In contrast with the procedures deriving from the Bridgman-
Stockbarger technique, the Wacker ingot-casting process uses two different

containers for melting and crystallization. It is a recharging process, with
possible re-usability of the mould after casting of the molten silicon. The use
of two separate containers for melting and crystallization allows for the
decoupling of the two processing cycles. In the upper part, melting and
casting take place whereas in the lower part the pre-conditioning of the
mould and the crystallization are carried out. The starting material is melted
using radio-frequency (r f ) heating; the melt crucible can be charged re
peatedly by recharging equipment. After casting of the molten silicon,
crystallization is carried out in a controlled manner so that a crystalline shell
is generated which caulks the mould from the silicon melt. This chilled layer
of polycrystalline silicon is produced first in the mould to prevent reaction of
the melt with the graphite walls. Crystallization of the remaining silicon melt
is then finished in this crystalline container. The method produces typically
polycrystalline ingots of size 43 cm x 43 cm and an efficiency of about
50 W k g " 1.
The ingot-casting equipment contained in a single water-cooled recipient
carries out all the processing steps involved in the melting and crystallization
cycles. A schematic diagram of casting equipment for the production of Silso
ingots is given in figure 1.26 where the various operations in the Wacker
ingot-casting process take place in a reduced inert gas atmosphere. The linear
Figure 1.26 Schematic diagram of casting equipment for the semi-

continuous production of Silso® ingots: 1, view port; 2, recharging
chamber; 3, rf coil; 4, graphite crucible; 5, quartz crucible; 6, silicon
melt; 7, mould; 8, thermal insulation; 10, heating element; 11, recipient.
growth rate of the polycrystalline silicon ingot should not exceed a critical
threshold of 1 m m m in-1 since the crystal structure begins to collapse at
higher growth rates.
This efficient ingot-casting technology is completed with highly efficient
cutting and wafering techniques which produce low-cost silicon wafers at
minimum material loss. Cutting and slicing of the finished ingots are
accomplished by band-sawing and multi-blade slurry sawing. The band-saw
technique contours and portions the ingots and is able to reduce kerf losses
to about 2 mm from the 8 mm incurred in outer-diameter ( o d ) sawing. By
using these techniques a 100 kg ingot can be cut into 16 columns of
10 cm x 10 cm cross section with 82% columns output.
Polyx process (Photowatt)
The furnace allows us to obtain multi-crystalline silicon ingots with a basaltic
structure by means of an unidirectional crystallization method in a fixed
crucible (figure 1.27). The seedless crystallization is operated in a crucible
stationary with respect to the working coils (direct induction in the crucible
walls). Regulation allows us to stabilize the silicon bath temperature, and
silicon solidification is obtained when opening the thermal insulation at the
bottom of the crucible. In comparison with the straight Bridgman process in
which the crucible moves down the thermal gradient, this modified process
leads to a steadier solidification rate and the shape of the solid-liquid
interface is more regular—nearly flat over a substantial extent of its surface.
When the crystallization is finished, the crucible is removed and can be used
again. The ingot is then slowly cooled naturally following a thermal
programme. In order to increase the capacity of industrial furnaces, the
section of the ingot should be increased preferably to increase the height.
Therefore an appropriate adaptation of the rf heating process has been
Figure 1.27 Polyx crystallization method.

developed. The rf coil is divided in two separate r f coils: a vertical one for
the crucible walls and a horizontal one for the crucible top wall. Therefore,
better heating control can be achieved by adjusting separately the induced
heat of the two coils. The graph in figure 1.28 shows an example of a
crystallization cycle used to obtain 30-60 kg ingots. The first phase of this
cycle is the fusion of the polycrystalline silicon and takes approximately
2-3 h; the next phase is stabilization which is carried out for j h. Then the
thermal insulation of the bottom is slightly opened in order to obtain heat
dissipation and to start the crystallization (6-8 h). The ingot is unmoulded
from the crucible at the end of the crystallization process and is allowed to
cool through controlled cooling.
Such an industrial casting furnace can produce multi-crystalline ingots of
mass up to 60 kg with 10% conversion efficiency for up to 90% of the ingot.
STA B ILIZATIO N
R ETU R N IN G -
UNMOLD
21 H TIME
Figure 1.28 30-60 kg ingot elaboration cycle.
1.3.2.2 Ribbon growth fabrication process

Edge-definedfilm-fed growth (Mobil Solar)
Graphite shaping dies confine and shape the silicon molten material into the
desired form just prior to solidification. The process is carried out under less
controlled conditions than are necessary for single-crystal growth.
Dendritic web growth ( Westinghouse Corporation)
Edge-stabilizing dendrites of silicon are used to grow ribbon-shaped crystals
from the melt. Specially designed radiation shields are used to shape the
thermal fields around the crystal liquid interface to achieve the differential
growth conditions required for the dendrites and the ribbon. Continuous
growth of the polycrystalline silicon ribbon is essentially possible through
melt replenishment.
Edge-supported pulling
esp differs from the efg method in that, here, a foreign material becomes an
integral part of the growing ribbon to stabilize the edges (figure 1.29).
Graphite filaments are immersed in the silicon melt and withdrawn at a
controlled rate, causing a web or ribbon to form between the filaments. The
rigid graphite filaments control the ribbon width and edge geometry in a
manner analogous to the silicon dendrites in the dendritic web process.
Figure 1.29 esp: in seeded esp, a seed crystal between graphite filaments
originally contacts the molten silicon. As the seed is pulled up, a nearly
single-crystal web solidifies between the filaments.
Ribbon-to-ribbon growth (Solavolt)

Ribbon-shaped crystals are not produced directly from the melt as in the
previous methods but, rather, silicon in the form of thin ribbons is deposited
on graphite or molybdenum substrates through c v d . The difference between
coefficients of thermal expansion of the two materials allows the silicon to be
physically separated from the substrate. The silicon is then recrystallized
using scanned laser or electron beams.
Ribbon-against-drop process
Ribbon growth processes can be divided into unsupported and supported
growth techniques, the r a d process being of the latter kind. The r a d process
is a vertical growth process whereby thin polycrystalline films crystallize on
both sides of a carbon ribbon which enters the melt via a slot in the bottom of
the silica crucible. A schematic diagram of the process designed for growth of
ribbons 5 cm wide is shown in figure 1.30. The crucible is heated by r f
induction in a graphite susceptor. A pyrocarbon coating 2-5 pm thick grown
in an open furnace by pyrolysis at 2100 °C gives the ribbon shaper its wetting
properties and its chemical resistance to molten silicon.
Figure 1.30 Growth of polycrystalline silicon ribbon by the rad

process.
Self-supported sheets of polycrystalline silicon are obtained by burning the

carbon ribbon in oxygen at high temperatures. The series of operations
involved in the preparation of unsupported silicon sheets is indicated in
figure 1.31. The carbon is burned off in a conventional resistance furnace at
1020 °C by exposing the as-grown r a d ribbons to an oxygen-bearing gas for
Laser Cutting of the

As-Grown Ribbon
(100 x5cm)
Sheets Ready for

Burn - Off
(12.5x4 cm)
Carbon Ribbon
Burn - Off Step
Etching of Silica
Layer Silicon
Solar Cell Fabrication

Figure 1.31 Concept of the preparation of unsupported sheets from
as-grown r a d ribbons.
1 h. The thin silica film formed on the surfaces of the sheets is then etched
away using an H F -H N 0 3-C H 3C 0 0 H mixture. Subsequent low-
temperature annealing and hydrogenation by implantation have been found
to improve solar cell performance greatly and 4 cm x 4 cm solar cells
fabricated from the polycrystalline silicon sheets have shown efficiencies of
10%.
1.3.2.3 Sheet growth fabrication process

Supported-web method
In the supported-web technique, a net of carbon fibres is pulled through
molten silicon which crystallizes on the net (figure 1.32), thus forming a sheet
of polycrystalline silicon. The size of the pulled supported-web sheet can be
typically 6 cm in width and 0.3 mm in thickness produced at a low-speed pull
rate of about 3cm m in_1. Through the conversion of carbon fibres into
silicon carbide (molten silicon reacts with the carbon), potential cracks in the
silicon sheets from thermal stresses caused by different thermal expansion
rates of carbon and silicon are avoided. Thermal stresses due to non-linear
thermal profiles during cooling of the ribbon can also be reduced by raising
the pulling speed for short time intervals, thus effectively interrupting the
silicon coating process on the net. This results in rectangular sheets of
polycrystalline silicon.
Figure 1.32 The supported-web technique for growing silicon ribbons:

1, carbon-fibre net; 2, crucible; 3, silicon melt; 4, crystallized silicon
ribbon.
Ramp-assisted foil-casting technique

The r a f t involves the direct production of self-supporting silicon wafers.
The slicing step necessary in ingot-casting techniques is therefore avoided
and feedstock silicon consumption is also reduced. Figure 1.33 shows the
basic principle involved in the r a ft and a schematic diagram of a con
tinuously operating set-up is given in figure 1.34. The method makes use of a
10x10 cm*
Silicon wafers Inspection Regeneration

Feed stock
Heater
«=>
Heater Preconditioning
Figure 1.33 Schematic diagram o f the raft : basic principle.
Consumable Layer
*
*
Silicon Foil -
*
4 \
Solid /Liquid
Interface
Ramps
Figure 1.34 Survey of complete raft processing.

temporary shaping guide called a ramp on which a silicon melt film is

crystallized. The ramp can be re-used several times before an expendable
coating has to be regenerated. 10 cm x 10 cm wafers have been grown on
ramps of the same surface area. Efficient removal of the heat of crystalliz
ation occurs through heat conduction rather than through heat radiation as
is normally the case in ribbon technologies. The solid-liquid interface
is inclined to allow for high casting rates at moderate temperature grad
ients. The best cell efficiencies obtained so far of 5.8-10.3% (AMl.5 for
5 cm x 5 cm cell) compare with those obtained from the standard Silso
process using the same device procedure.
Horizontal supported-web method (Siemens)
The h s w method is a modification of the supported-web process except here
a carbon fibre net is pulled almost horizontally over a silicon melt surface. A
schematic diagram of the h s w pulling apparatus is shown in figure 1.35. A
solid silicon film crystallizes on the moving net provided that a suitable
Si Feeder
C - Fiber Net
Figure 1.35 Schematic diagram of the hsw pulling apparatus.
temperature distribution has been established in the silicon melt contained in

a long shallow quartz crucible. The temperature gradient is established by
heating the crucible bottom with a resistance heater while keeping the melt
surface temperature close to the freezing point by surface radiation. The
thickness growth of the solid silicon takes place perpendicular to the pulling
direction and continues as long as the moving sheet is in contact with the
melt. With a pulling speed of 1 m m in-1 , silicon sheets of up to 6 cm in
width and with grain sizes up to 4 mm have been grown. Solar cells
(2 cm x 2 cm) made from the h s w silicon material by the conventional
diffusion process of phosphorus at 850 °C for 30 min show maximum
efficiencies of 9.4% under AM 1.5.
Silicon-on-ceramic method
Dip coating. Dip coating is one of the silicon-on-ceramic methods in which a
polycrystalline silicon film solidifies on a coated ceramic ribbon as it is
withdrawn from the silicon melt (figure 1.36). Polycrystalline silicon ribbons
about 100 gm thick have been pulled at a rate of 5-6 cm m in-1 . Cells with
large columnar grains 4-5 mm in diameter have been formed with cell
efficiencies of 10.5% for a film 100 /mi thick.
Aluminum Silicate
Figure 1.36 Dip coating is a simple soc process in which a coated

ceramic form is slowly withdrawn from a silicon melt while poly
crystalline silicon solidifies on it.
Silicon coating by inverted meniscus. In the process of silicon coating by

inverted meniscus ( scim ), ceramic sheets are coated with a polycrystalline
layer of silicon as they are drawn out through the scim furnace. The ceramic
sheets are drawn over a trough into which a controlled amount of molten
silicon has been released by a plunger lowered into a crucible containing the
liquid silicon (figure 1.37). The silicon forms a raised meniscus in the trough
Figure 1.37 s c im : a plunger lowered into a crucible of molten silicon

drives a controlled amount of the liquid into a trough, where it forms a
raised meniscus. The ceramic form drawn over the trough touches the
silicon, which cools and solidifies as it is drawn away.
which touches and coats the ceramic sheets as they are pulled past at a rate of
4 cm m in-1 . The method results in less exposure to impurity contamination
and ribbon silicon with thicknesses of 100-300 pm have been formed. The
s c im method developed by Honeywell has achieved growth rates of
60 cm2 m in“ 1 and cell efficiencies of 10.5%. Figure 1.38 shows how polycry
stalline silicon solar cells are fabricated from the sciM-grown silicon ribbon.
b
n - Contact Grid p-Contact Grid
Ceramic
Figure 1.38 (a) Contacts can be applied on both sides of an soc cell by
pre-slotting the ceramic base (which does not adversely affect the soc
process) or (b) contacts can be attached at the front by cutting through
the cell’s top layer. The series resistance resulting from the awkwardness
of contacting remains a problem in soc cells.
Silicon-on-silicon method
In the sos method, a polycrystalline silicon film is deposited on impure silicon
substrates through c v d by the reduction of SiHCl3 in H 2 at elevated
temperatures. The substrates can be cz or cast silicon and 10% efficient cells
over 100 cm2 in area have been grown. Cheap MG-Si may also be a suitable
substrate for deposition of purer polycrystalline films.
Silicon-on-inexpensive substrates
Silicon vapour deposition techniques on cheap, easily handled materials such
as steel or glass are applied with a best efficiency achieved of 23.6%.
1.3.3 Silicon-cutting and silicon-slicing techniques

An integrated line for low-cost production of crystalline silicon ingots
whether single-crystal or polycrystalline material has to incorporate efficient
cutting and wafering techniques. One disadvantage of using ingots rather
than thin or thick films in solar cell fabrication processes is that 50% of the
silicon is wasted by the wafering operation.
The technical state of the art in silicon slicing is represented by five
different methods. Some of them are used in practice while the others are still
in the laboratory testing stage.
(i) Internal-diameter ( i d ) sawing.
(ii) Conventional multi-blade slurry sawing ( m b s ).
(iii) Multi-wire slurry sawing ( m w s ).
(iv) High efficiency multi-blade slurry sawing ( h e s s ).
(v) Fixed-abrasive (multi-wire) sawing technique ( f a s t ).
Table 1.2 presents a comparative and qualitative evaluation on operational
conditions of alternative silicon-slicing methods.
Table 1.2 Silicon-slicing methods.
Slicing Mode of Production Yield Surface Technical

method application rate (%) quality reliability
(wafers h -1)
id sawing Conventional Average High Average High

id sawing Advanced High Very high Good Very high
MBS Conventional Low Average Average Good
MBS Advanced Average High Good Fair
MWS Conventional Low Average Good Fair
FAST Testing stage Average Average Good Low
Two trends in material removal processes are emerging from these

methods.
(a) Conventional sawing as in i d or f a s t methods.
(b) Slurry sawing as in m b s , m w s or h e s s methods.
1.3.3.1 Sawing process

In the case of the i d sawing technique, silicon stock is removed by synthetic
diamond grains which are firmly bonded in a nickel base applied on the
internal tool rim by electroplating. Thus the cutting action is in essence a
plunge grinding operation, characterized by high cutting speeds vs of
15-20 m s -1 and a rather large contact zone between tool and workpiece, i d
sawing results in specific removal rates Z' per tool of 50-80 cm2 m m '1 which
is up to 1000 times higher than in wire sawing with diamond-clad tools
( fa st ). Also with regard to the so-called wear ratio G (i.e. work material
removed divided by abrasive layer lost by wear), id sawing with G «
500000 cm2cm - 2 is superior to the fa st process with G % 7500 cm2cm -2 .
In spite of high specific removal rates, id sawing as well as other slicing
processes require a moderate application of coolant when cutting silicon.
As mentioned above, a similar removal process also using electroplated
diamond abrasives bonded to the surface of a wire-type tool is effective in the
f a s t . Figures 1.39 and 1.40 show the general principle of this method, the
main advantage of which is the relative low mass of the moving tool frame
and the relatively high cutting speed vs of 100-150 m m in-1 . However, in
spite of the pre-tensioning of the wires, the effective tool load is rather limited
owing to the elastic deformation of the wires perpendicular to the work
surface. The resulting specific normal forces do hardly exceed 5 gem -1 of
contact length while in the case of id sawing the respective values are 50-100
times higher. Thus, the necessary grooving action with a brittle type of
fracture to separate small particles from the work surface cannot take place
in fixed-abrasive wire sawing with suitable efficiency.
1.3.3.2 Slurry sawing process

A very different removal process is effective in the so-called slurry sawing
process and has not very much in common with a real sawing process which
is currently based on cutting like a lathe tool. In fact the slurry process
represents a specially modified lapping process, characterized by loose
abrasive particles which are conveyed into the contact zone between tool and
workpiece by means of a lapping suspension. The lapping grains are forced
to roll through the contact zone, affecting the work surface with an extremely
high impact frequency, which can reach a few hundred thousand impacts
(Hertzian fractures of the very brittle silicon) per second and per square
millimetre of the work surface, hess is a rather attractive sawing technology.
Pertinent machines with increased cutting speeds and blade loads have been
built and provide a promising alternative to conventional id sawing.
Technology o f high-efficiency slurry sawing
Basically there are two ways to increase the production rate of multi-blade
slurry: to increase the number of cutting blades or to increase the cutting rate.
It appears practically that blade packs with more than 300 blades are
impractical as tensioning forces become enormous and problems with tool
handling and tool accuracy intensify. Therefore the most promising way to
boost productivity and to lower production cost at the same time is to
increase the infeed rate of conventionally sized blade packs by means of
higher cutting speeds, to reduce the stroke and blade length and accordingly
to increase blade loads.
1.3.3.3 Band-saw technique

For the ingot-casting technique, a tailored technique for contouring and
portioning the 100 kg ingots with about 43 cm x 43 cm cross section is
essential. A band-saw technique is applied for contouring and portioning the
ingot in pre-shaped columns of 10 cm x 10 cm cross section; the columns
are then wafered using the hess technique.
Contouring and portioning of large ingots is favourably carried out with a
band-saw technique which allows the machining of larger dimensions than
does o d sawing.
The cutting of a 100 kg ingot (43 cm x 43 cm cross section) with a circular
saw requires a saw blade with a diameter of 1240 mm at least. Owing to the
necessary stabilization the core of the saw blade should have a thickness of
about 6 mm; the kerf loss is about 8 mm. In the case of a band-saw the
cutting width is only about 2 mm. As a result of this lower cutting width the
absolute cutting force in band-sawing is smaller than in o d sawing. Lower
cutting forces induce less strain in the brittle material silicon, i.e. the quality
of the cut and the yield are improved. The tool used in band-sawing is a
toothed endless steel belt with the diamond cutting segments soldered onto
Table 1.3 O perational criteria of different silicon-slicing techniques.
Surface Cutting force Surface cut In-feed Cutting rate Wafers Wafers Surface
speed (N) (m2) rate (cm2m in_1) per per damage
(m m in-1) (cmmin-1) time length Onn)
(IT 1) (cm "1)
Per Total Per Com- Per Com-

tool tool plete tool plete
pack pack
1000 28 — 30 5 50 30 14 10
1200 56 — 40 8 80 48 18 10
30 0.8 240 0.01 3.0 0.008 0.08 24 15 16 10

50 1.5 450 0.01 3.0 0.015 0.15 45 27 21 8
80 1.5 450 0.01 3.0 0.120 1.20 360 216 16 10

120 2.0 600 0.01 3.0 0.180 1.50 450 270 21 8
75 1.0 150 — 1.2 0.007 10 6 22 5

85 1.0 150 1.2 0.020 30 18 25 3
100 0.4 100 0.03 7.5 0.012 0.12 30 18 22 5

140 0.4 100 0.05 12.5 0.016 0.16 40 24 25 3
the carrier body. The carrier influences dominantly with its form and
material quality the lifetime of the tool and with it the overall output. The
saw band is made from low-carbon steel with a strength of about
MOON m m -2 . It has the dimensions 10500 mm x 80 mm x 1 mm. The
concentration and grain size of the abrasive diamond and the hardness of the
bronze bonding define the cutting rate and quality of the cut surface. In
general the abrasive grains are synthetic diamonds of medium grain size and
concentration. Cutting rates depend on the feeding rate and contact zone
between tool and workpiece. On contouring and portioning 100 kg ingots
with 43 cm x 43 cm cross section and a height of about 25 cm, cutting rates
of more than 120 cm2 min - 1 have been obtained (for a saw blade it is around
70 cm2 m in-1). The cut surfaces are flat; geometric deviations are less than
0.1 mm. The lifetimes of saw bands are comparable with those of o d saw
blades for silicon processing.
1.3.3.4 Trends and perspectives

The total production cost for solar silicon wafers is composed of different
cost elements, related to material, tool, machine, labour and minor expenses.
The dominant cost factor is the material. Then, of the remaining add-on
costs, machine and labour have the strongest influence while the tool costs
are of secondary importance. Therefore, minimal production costs are likely
to be achieved with slicing techniques which provide low value for wafer
thickness and kerf losses, and which at the same time produce a maximum
number of wafers per unit of time. A comparative survey of operational
criteria of different silicon-slicing techniques is presented in tables 1.3
and 1.4.
Many attempts have been made to increase cutting rates and yields mainly
by modification of existing slicing processes. From the technological and
Table 1.4 Comparison of performances of slicing techniques.
Surface Silicon Productivity Damage Kerf loss

Speed feed rate (wafers h - x) Ozm) 0*m)
(m m in- x) (cm min-1)
id saw 3000 5 30 10 350

MBS
(300 blades) 30 0.08 12 10 250
MWS
(150 wires) 100 0.02 20 5 220
Present mws
(300 wires x 2) 300 0.03 150 5 230
economic comparison of alternative silicon-slicing methods, two trends can

be extracted which most probably will determine the future state of appli
cation of slicing techniques.
(i) Regarding future mass production of solar silicon wafers, id sawing
with diamond-plated saw blades has an effective advantage against all
other processes based on its well introduced and accepted machine and
tooling technology. With increasing blade diameters (up to 27 in and
eventually more) the output will be increased by cutting larger ingots or
by cutting more than one ingot at the same time. In process control
systems, as already available with the most advanced id systems,
increasing yield and cutting quality considerably reduce the labour cost
as well.
(ii) h e s s is the second slicing technique which has a definite chance to play a
major role in the mass production of solar cells. Of all slicing techniques
it provides the highest production rate and the lowest add-on costs and
is expected to become an attractive alternative to id sawing.
1.3.4 Surface polishing and cleaning

A surface layer about 30 /nn thick is damaged after ingot sawing in
conventional cell technology. Generally this layer is removed isotropically
and rapidly by acid etching, e.g. typically a rate of 1 /mi m in-1 at 20 °C is
reached with the following mixture: 1HF (48%) -I- 1CH3C 0 2H (99.8%) +
5 H N 0 3 (65%).
The surface can also be polished with alumina powder of increasingly
smaller grain size and then with mechanochemical polishing such as 10 min
contact on a rotating belt with a mixture of equal parts of 25.9 g of N H 4F +
100 cm3 of H 20 and 9 g of (N 0 3)2Cu, 3H20 + 100 cm3 of H 20 .
Basic etches, i.e. K O H -N aO H -H 20 mixtures, are also used in industry
because they cause less pollution than acid etches do when used in large
quantities. Another advantage of basic etches is that somewhat texturized
surfaces (with smaller reflectivity) are obtained.
All these polishing etches are quite different from those used to reveal
grain boundaries and dislocations (1 volume of (50 g of C r0 3 + 100 ml of
H 20 ) + 1 volume of HF) or to remove silicon oxide (^HF + \ distilled
water).
Finally, after polishing, it is necessary to clean the surface from all
impurities, e.g. by one of the following procedures: simple cleaning (hot
trichlorethylene-hot acetone + hot methanol + rinsing in distilled water +
30 s in ^HF -f \ distilled water + rinsing + drying); ‘RCA cleaning’ (10 min
at 8 °C in each of 1H20 2 (30%) + 4S04H 2 (96%), 1H20 2 (30%) +
lN H 4OH (30%) + 5H20 , 1H20 2 (30%) + 1HC1(37%) + 5H20 and then
|H F + j distilled water).
1.3.5 p-n junction formation

A p-n junction can be formed through either a high-temperature diffusion
( h t d ) process or an ion implantation process (the c j p ). There exist two
different approaches for the diffusion process, i.e. diffusion from the vapour
phase, or diffusion out of a solid phase. One disadvantage of diffusion from
the vapour phase is that diffusion occurs on both sides of the wafer such that
the back parasitic junction has to be removed. Diffusion out of a solid phase
consists of the deposition of a dopant layer at ambient temperature followed
by the diffusion process itself at a high temperature (800-900 °C). The
deposition can be performed using, for example, the following techniques.
(i) CVD.
(ii) Spin-on process.
(iii) Spray-on process.
(iv) Screen printing (phosporous-doped paste).
Parasitic junctions at the edges of the wafer are removed by chemical
etching, by plasma etching or by laser scribing and breaking.
The cjp is still in the development stage; the doping species are introduced
by ion or molecular implantation or are evaporated or spray deposited onto
the surface of the semiconductor and driven in by pulsed annealing (laser or
electron beam).
Using this process, only a very thin layer is molten just underneath the
surface and the bulk of the material is kept at a low temperature, preventing
D E V IC E F A B R IC A T I O N
JU N C T IO N CO NTACT
F O R M A T IO N OTHERS
F O R M A T IO N
Th e rm a l Ion Vacuum S cre e n P la tin g

D iffu sio n Im planta- M etalli- Printing
(5) tion zation
Lapp in g Polishing A n tire fle ctiv e
Figure 1.41 Review of technical options for the solar cell fabrication
process: S, standardized.
d e g r a d a tio n o f th e m in o rity -ca rrier lifetim e; su ch a life tim e d e g r a d a tio n ca n

b e very im p o r ta n t d u rin g h t d in th e ca se o f a lo w er -g r a d e p o ly c r y sta llin e
silico n .
The main other advantages of cjp over htd are as follows.
(i) Lower energy consumption.
(ii) Process amenable to full automation.
(iii) Lower overall cost.
A review of the technical options for the solar cell device fabrication
process is presented in the flow chart in figure 1.41.
1.3.5.1 High-temperature diffusion

Standard solar cell technologies introduce a base layer dopant during crystal
growth. The p- or n-type silicon must then be formed upon the base layer
depending on the impurity type of the initial wafer (table 1.5). To form p-type
silicon, boron is generally used as the impurity from sources such as B20 3,
BC13 and BBr3. Sources of phosphorus, the n-type dopant, include P2Os and
POCl3. The process of delivery of the impurity source to the wafer are
dependent on whether it is a gas, liquid or solid.
Table 1.5 Impurity types.
Silicon Impurity Phase at room

type source temperature
P b 2o 3 Solid
P b c i3 Gas
P BBr3 Liquid
n p 2o
5 Solid
n POCl3 Liquid
A solid diffusion source such as P20 5 is heated in a chamber upstream

from the wafers. A carrier gas passed over the vaporizing P2Os carries the
impurities to the wafer surface. Various apparatuses for phosphorus diffu
sion are shown in figures 1.42(a), 1.42(6) and 1.42(c) for solid, liquid and
gaseous sources respectively.
For a liquid source such as POCl3, a carrier gas (a mixture of N 2 and 0 2 in
the ratio of 3:1) is bubbled through the liquid source at about 950 °C. After a
diffusion time of 10 min, a diffused layer of about 0.25 pm will form. Typical
apparatuses for boron diffusion are given in figure 1.43.
The formation of a p-n junction follows Fick’s laws of diffusion. Accord
ing to the laws of diffusion, if a base layer of doped silicon is coated with
Silicon solar cells feedstock material
D iffusion fu rn ace
to furnace to furnace
Figure 1.42 Apparatuses for phosphorus diffusion.
D iffu sio n furn ace
Q uartz tube
ziToT" Si wafer
\Q ,. 0 0 00
f low
meter
o
(o )
tl t
to fu rnace
(b ) (c)
Figure 1.43 Apparatuses for boron diffusion.

phosphorus and then heated, the phosphorus atoms will diffuse into the solid
silicon from regions of high concentration to regions of low concentration.
Diffusion is arrested by lowering the temperature after an n-type collector
layer of the proper thickness is formed. The junction area is the narrow
transition region in which the dominant dopant switches from phosphorus to
boron.
Figure 1.44 shows the variation in phosphorus concentration with depth
for various diffusion times at a diffusion temperature of 950 °C. The p-n
junction is defined to be located where the phosphorus and boron respective
concentrations become equal.
Phosphorus Phosphorus Diffuses to

Applied Here 1.2 /wm Depth,
Creating n-Type Layer
J u n c tio n '
U 1
lO20 I0 19 I0'8 Original Boron-Doped

Concentration (Atoms/cm3) p-Type Base
Figure 1.44 Four diffusion profiles for 950 °C diffusion of phosphorus into silicon.
The timing (10, 20, 30 or 60 min) can be regulated to produce the right amount and
depth of dopant. The base layer is 2 x 1016 B atoms cm -3 . The model cell (on the
right) shows that the junction region is located where the top-layer doping and the
base-layer doping are equal.
Owing to the high concentration of phosphorus at the cell surface even

after the diffusion process, the phosphorus and silicon react to form an alloy
called the ‘dead layer’ which is highly resistive and is a cause of electron-hole
recombination. The ‘dead layer’ is generally removed by acid etching.
Since over 50% of the sunlight incident on silicon is absorbed within the
first 3 pm of the surface and a depth of 300 pm is needed to absorb the rest, a
solar cell should be fabricated so that it has a shallow top collector layer (less
than 3 pm thick) and a thick base layer with a high mobility of charge carriers
in case light-generated carriers are formed far away from the junction.
1.3.5.2 Cold-junction processing

The ion implantation technique as used in c j p is still in the development stage
and provides an alternative method of forming a p-n junction to the
conventional, slow and energy-intensive h t d process. During diffusion, the
atomic dopants penetrate towards the bulk through an increase in thermal
vibrations under the influence of a gradient driving force (energy involved is
in the electronvolt range). In this case the dopants are ionized (except for
recoil implantation) and accelerated typically between 5 and 100 keV; they
penetrate the target through their kinetic energy. When impinging on the
solid, the projectiles lose their energy by electronic and nuclear collisions.
Nuclear collisions are responsible for the damage which is created in the
single-crystal or polycrystalline structure of the base material, i.e. silicon. In
order to anneal this damage and to put the dopants in substitutional sites, an
annealing procedure is necessary. The silicon wafer is exposed to a beam of
high-energy ions of the desired dopant by means of a glow discharge device;
for example, if the doped base layer is p type, it is exposed to a phosphorus
ion beam. The depth of penetration is adjustable depending on the strength
of the beam. The silicon surface must then be annealed with a high-energy
laser or electron beam in order to remove the damage caused by the
implantation on the crystal lattice and to put the dopant atoms in substi
tutional sites. The resultant p-n junction does not differ much from a
conventional diffusion-formed junction with depths generally about 0.25 /un.
Concentration of dopants varies from 1016 atoms cm "3 at the junction to
1021 atoms c m "3 at the surface. Advantages over the thermal diffusion
process include less energy usage; it is basically a cold process presenting a
greater reproducibility on a large scale because of its greater simplicity and
greater ease of control of junction depth and dopant concentration as well as
dopant profile. In addition, a thin layer of highly doped back-surface field
can be formed on the cell at the same time as the front-surface implantation.
Basically there exist four different procedures to perform dopant
implantation.
(i) Conventional ion implantation: dopant ions are produced by glow
discharge in a source starting from a gas or vapour compound
containing the dopant atoms (figure 1.45). These positive ions are then
iso la tio n
Figure 1.45 Classical ion implantation.

extracted, handled in the correct optics and accelerated before being

analysed either by a magnet or a speed filter in order to obtain an
isotropically pure ion beam. The typical implant dose is around (1-5)
x 1015 cm -2 .
(ii) Ion incrustation: this involves the direct use of glow discharge beams
issued from a gas including dopant atoms, in which both atomic and
molecular ions are implanted simultaneously without any mass selec
tion (figure 1.46). Ion sources of nearly any size can be realized using
multi-hole extraction structures as for ion-milling systems.
Figure 1.46 Glow discharge implantation.
(iii) Recoil implantation: in this procedure, dopants are first deposited at

the silicon surface, at thicknesses typically 50-200 A, by any technique.
Then the film is bombarded by heavy ions (generally rare gases). These
heavy ions push the dopants into the target with a yield which is larger
than unity, i.e. the number of recoiling dopant atoms is larger than the
incident beam intensity. The penetration is very shallow (typically
1000 A) before annealing; however, severe annealing problems exist
and the small gain in yield is not sufficient to make this procedure
attractive for cell manufacturing.
(iv) Ionized cluster beam deposition: the material to be deposited is
vaporized in a crucible and is adiabatically expanded when going
through an exit nozzle into about N « 1000 loosely coupled atoms. By
bombarding this cluster with a beam of either electrons or ions in the
kiloelectronvolt range, it attains a single positive charge and can be
accelerated. When impinging on the silicon surface, the ionized clusters
are broken into individual atoms, each having a mean energy in the
electronvolt range (figure 1.47). The mean energy in the systems at
present available is low (1-1000 eV); therefore the dopants are just
deposited at the surface, a situation intermediate between ion plating
and implantation (figure 1.48).
IN C ID E N C E
IO N IZED C L U S T E R N EU TRA L C LU S T E R
(W ITH A C C E L E R A T E D S P E E D ) 1 (W ITH E JE C T IO N S P E E D )
D E P O S IT M ATERIAL
SP U T T E R IN G
REEVAPORATION
O
S U B S T R A T E M ATERIAL
SP U TT ER IN G
Q
Figure 1.47 Break-up of ionized clusters in atoms.
Figure 1.48 Cluster beam deposition.
However, ion implantation is a costly procedure which has to be completed

with rapid thermal annealing by laser or strip heater. Strong cost reductions
are expected if large production rates can be obtained.
1.3.5.3 Screen-printing junction process

A phosphorus paste is spread on the wafer surface (p-type substrate) using a
screen-printing machine. Then diffusion of phosphorus impurities takes
place into a conveyor furnace at around 925 °C in an Azote atmosphere (N2).
The phosphorus paste contains basically three components: the active

material (normally P20 5); solvent (butylcarbitolacetate); thickener (very
pure S i02 particles 10-20 pm thick, sometimes called Aerosil).
Screen printing is also a commonly used industrial process for contact
elaboration and is accepted today as the most cost-effective method for cell
metallization in an industrial production environment.
1.3.6 Electrical contact formation

Electrical contacts must be attached to a cell in order to place the cell in an
electrical circuit. When placing the contacts, a trade-off must be made
between shading effects caused by the contacts and resistance losses. Top
surface contacts are usually in a grid pattern to allow good conduction while
at the same time shading not more than 10% of the cell surface. The different
techniques (vacuum vaporization and screen printing described below) are
being applied today for making top-surface grids. The back-surface contact
which can be simply a metal sheet is much less problematic. In certain newer
cell designs, light enters from both surfaces of the cell, thus requiring grid
contacts and anti-reflection coating on both front and back surfaces.
1.3.6.1 Vacuum vaporization

The metal to be deposited on the cell surface is heated in a vacuum to a
temperature high enough to cause melting and evaporation. The metallic
vapours are then deposited on a cell through a mask, forming a grid pattern
(figure 1.49). Titanium, palladium and silver are typically used for the front
S ilic o n w a fe r
O
(p la c e d in in tim a te
c o n t a c t w ith m a s k )
tI C r u c ib le
Figure 1.49 The use of a metal mask to define the top-metal grid
pattern.
contact with a nickel mask for grid definition. On the back side, aluminium
and silver are used successively to obtain Ohmic contacts. Evaporation still
leads to the most reliable contacts, but it is far too expensive and has been
abandoned by all manufacturers.
1.3.6.2 Screen printing

Two alternative methods for contact formation are electroplating and screen
printing. Both techniques should lead to similar costs, provided that one can
eliminate noble metals. Screen printing is actually the technique most
acceptable worldwide at the production stage.
The metal is painted over the entire front surface of the cell and
subsequently etched away from unwanted parts using a photographic
technique called photolithography (figure 1.50).
Ultraviolet Polymerized
Photoresist
Photoresist
iiiiiii / i \ mini umiui
CSSS muni
SIO,
Silicon chip
( a) (b) (c)
Figure 1.50 Photo-etching technique: {a) masking and exposure to ultraviolet
radiation; ( b) the photoresist after (a)\ (c) after etching.
The metal contact is usually made up of three separate layers of metal: a

thin layer of titanium as a bottom layer for good adherence; a top layer of
silver for low resistance and solderability; and a middle layer of palladium to
prevent reaction between the titanium and silver layers in the presence of
moisture. The contacts must then be sintered at 500-600 °C for good
adherence and low contact resistance after the metals have been deposited.
An aluminium paste is generally used for the back-side metallization instead
of silver and will produce a back-surface field as aluminium is an acceptor. It
is also advantageous to deposit the front metallization after the anti
reflection coating.
1.3.7 Anti-reflection coating

Untreated silicon reflects more than 30% of incident light. A single layer of
an anti-reflection coating such as silicon oxide is able to reduce cell surface
reflection to 10% of incident light while a double layer reduces it to below
3%. Materials such as silicon dioxide (Si02), titanium dioxide (T i02) and
tantalum pentoxide (Ta2Os) are also used as anti-reflection coatings. The
vacuum process described earlier for contact formation is also used in

depositing the anti-reflection coating. Sputtering and chemical spraying can
also be used. Figure 1.51 shows how texturing of a cell surface in addition to
an anti-reflection coating can reduce reflectivity and improve cell efficiency.
Figure 1.51 (a) Texturing a cell exposes the tetrahedral surfaces of the
silicon crystal lattice, (6) Light that strikes the cell nearly perpendicularly
can be either absorbed or reflected; if it is reflected, it will strike another
surface and have another chance to be absorbed. Note also that light that
is absorbed is bent at the treated surface, so that it penetrates the cell
obliquely. Because light travels through the cell at an angle, the cell need
not be as thick to absorb as much light, (c) The lower reflectivity and the
longer distance that the light travels through the cell give about a 15%
improvement in cell efficiency compared with smooth uncoated cells.
1.3.8 Cell encapsulation

Cell encapsulation is required to protect metallic contacts and interconnec
tions from corrosive elements in the atmosphere as well as to provide
electrical isolation from voltages generated by the solar panel. Encapsulation
provides mechanical and chemical protection to the cell and its durability
determines the operating life of the module.
Before encapsulation a structural layer at either the back or the front of the
module must be attached to provide rigidity. In one module design, cells are
attached directly to the structural back and then encapsulated in a flexible
pottant (figure 1.52(a)). The most commonly used materials for the struc
tural back are anodized aluminium, porcelained steel, epoxy board or
window glass. Silicone is widely used for the adhesive and pottant layers.
In another module design, the cells are enclosed and are supported by
layers of laminate applied first onto a structural back (figure 1.52(6)). The
laminate may be either polyvinyl butyral ( pv b ) or ethylene vinyl acetate
( eva ). A final layer at the top of the cell, possibly of low-iron glass, im
parts self-cleaning properties and in some designs provide protection from
moisture. For moisture-resistant cells, a soft silicone encapsulant may
be used instead with a harder silicone surface layer.
j — - top loyer
Pottant laminated layers
Cell Cell Cell

adhesive
v /////,
'&*))
(a) -Structural back- y (b )

Figure 1.52 Schematic diagrams of the encapsulation of solar cells.
1.3.9 Amorphous silicon thin-film technology

Efficient a-Si solar cells consist of a thin film of the order of 1 pm thick
whereas a typical classical cell thickness is 100-200 pm. This is because the
absorption coefficient in this disordered material structure is one order of
magnitude larger than in crystalline silicon. Moreover the growth tempera
ture for an amorphous material thin film is much lower (200-400 °C) than for
the crystalline technological process.
Specific features make a-Si cells very attractive as far as economics is
concerned.
(i) The saving of silicon material and consequently a high material yield.
(ii) A low-temperature technological process and consequently a reduction
in the energy pay-back period.
(iii) Deposition on inexpensive substrates such as glass, metal or plastic.
(iv) The possibility of producing industrially cells about a square metre in
size—a key feature in reducing production costs.
However, a-Si still suffers from a certain instability after exposure to
sunlight for some time, i.e. the Staebler-Wronski effect, which together with
its low efficiency limit its actual use to the power range of the consumer
electronics field such as pocket-sized calculators, wrist watches and battery
rechargers.
1.3.9.1 Amorphous silicon fabrication techniques

The a-Si atomic structure compared with single-crystal and polycrystalline
silicon does not have a preferential crystallographic orientation. Fabrication
of a-Si needs a high degree of control as slight alterations in conditions cause
a large change in cell structure. In most of the processes, a gaseous form of a
silicon-hydrogen alloy is decomposed by means of a glow discharge system
and deposited as a thin film on a substrate. Phosphorus or boron gaseous
compounds are included in the a-SjiH for doping the a-Si film. The substrate
must be kept at a temperature between 200 and 350 °C since at higher

temperatures a microcrystalline rather than an amorphous structure is
formed, while at lower temperatures the hydrogen and silicon do not form a
structure free of electronic defects. Fabrication of a-Si as a thin-film cell
is insensitive to substrate structure and can therefore be fully automated
and deposited on inexpensive substrates such as glass or stainless steel
(figure 1.53):
i i
L_ ix* u lt . 1
L igh t is absorbed with Ligh t passes Light is absorbed
a thickness of up to through
about 100/um
C rystal type Amorphous type

( f o r instance, C Z )
Figure 1.53 Features of an a-Si solar cell.
—glow discharge method

a-Si— thin-film------ sputtering method
techniques , ,
— cvd method.
In most a-Si fabrication processes, silane (S1H 4) is decomposed in an r f

glow discharge reactor near a heated substrate (200-300 °C) (see figure 1.56).
The exposure time, discharge power and flow rate all control the film
thickness. To produce p-type or n-type silicon, dopants can be introduced to
the SiH4 vapour, e.g. diborane (B2H 6) for p-type and phosphine (PH3) for n-
type silicon. To produce a-Si:H, SiH4 vapour is used, while incorporating
fluorine into the vapour produces fluorinated a-Si:H (a-Si:H:F). The struc
tural disorder of a-Si gives it a much higher absorption coefficient than
crystalline silicon which means that the thickness of a-Si need only be about
1 pm to absorb most of the incident light. The effective band gap is also
wider: 1.55-1.80 eV as opposed to 1.1 eV for crystalline silicon.
Major disadvantages lie in the short carrier lifetimes, lower efficiencies and
Light
Figure 1.54 Principle of improving the conversion efficiency of an a-Si

solar cell.
Top Grid C o n tact
L ig h t
A n tire fle ctio n
L a ye r, and
P rotective C o a tin g
S n O or Indium
T in O xide
p+ Am orphous 80 A
S ilico n
E le c tric
Undoped Field
( Intrin sic) 0.5pm Sp an s
Am orphous In trin sic
Silico n Region
n+ Amorphous
S ilic o n +t / +
r v r> 7 r / -
+ 200 A
Substrate and
Back Contact
( S te e l, Aluminum ,of
Conductively Coated
X/ G lass
/ / / / / / /* iu
Figure 1.55 A p -i-n device: most light passes through the top layer and
generates electron-hole pairs in the intrinsic region. An induced field
spans the intrinsic region, sweeping the charge carriers to opposite ends
of the cell.
degradation of cell efficiency under prolonged exposure to sunlight. The

Staebler-Wronski effect which causes the optically induced degradation of
a-Si solar cells is based upon the recombination effect of carriers generated
by light in the material. The main decrease in efficiency seems to occur in the
first 100 h of exposure after which cell parameters remain fairly constant. A
short anneal above 150 °C restores the original state of the cell. Changes due
to degradation consist mainly of a decrease in the fill factor and the short-
circuit current whereas the open-circuit voltage remains almost unaffected.
Common principles to improve the conversion efficiency of a-Si solar cells
are presented in figures 1.54 and 1.55.
Glow discharge deposition method
The glow discharge process involves the decomposition of a gas, i.e. source of
silicon material, in the electric field of a discharge plasma reactor (figure
1.56): a r f plasma is formed between a metallic cathode and the substrate on
which the a-Si will be deposited. Gas can be silane (SiH4) diluted in hydrogen
and, for the purpose of impurity addition, other gases such as B2H 6, PH 3,
FH 4 or CH4 may also be introduced into the chamber. The substrate is kept
at 250-300 °C which can be raised to obtain a microcrystalline layer instead
of an amorphous one. Glow-discharge-deposited films have a much lower
density of defect states in the gap owing to bond saturation by hydrogen
atoms. The hydrogen content which is typically 10-15% determines many of
the film properties such as microstructure, band gap and density of states.
^ J ~ ^ S i H 4 gas flow
R F coil
Heater
Figure 1.56 In the glow discharge process, silane gas (SiH4) is passed
slowly through a quartz tube in which is held a substrate heated to
200-300 °C. An r f field is applied inductively through an external coil,
which causes silicon to be deposited gradually as an extremely hard and
stable layer. The thickness of the film can be controlled by varying the
exposure and the flow rate. Doping elements are added to the silane gas
before starting the discharge.
Relatively low deposition temperatures of 220-300 °C allow for the

deposition of a-Si films on different kinds of inexpensive substrate ranging
from stainless steel sheets to polyimide films to already formed units such as
roof shingles.
For a high a-Si cell efficiency, the intrinsic layer which is the active part of
the solar cell device must have a low defect density and low impurity
contamination in order to provide good carrier lifetime. Deposition of high-
quality intrinsic material by glow discharge of silane is a slow process
occurring at a rate of about 0.1 n m s " 1 which will prove to be a limitation in
continuously running automated production lines. Efforts made to increase
the deposition rate include the use of disilane (Si2H 6) rather than monosilane
in the glow discharge deposition of intrinsic layers, which approximately
doubles the deposition rate, although the quality of films obtained so far
using higher silanes is still inferior to that from SiH4.
Multi-chamber a-Si deposition systems are currently perceived as the best
approach to the fabrication of high-quality a-Si solar cells as process gas
intermixing and dopant contamination in the intrinsic layer may result from
repetitive depositions in single-chamber systems.
A continuous mode of fabrication of flexible a-Si:H solar cells on
polyimide substrate by r f glow discharge deposition has been developed by
the 3M Company. A schematic diagram of the deposition system is shown in
figure 1.57. The best cells produced have a fill factor of 0.601, an open-circuit
voltage of 0.919 V and an efficiency of 6.2%. The flexible substrate consists
of a polyimide film 2 x 10 ~ 3 in thick which has been metallized by sputtering
200 A of TiN onto 1000 A of aluminium. The devices consist of p +-i-n +
layers 200, 5000 and 100 A thick on the Al/TiN bottom contact with a
quarter-wavelength indium tin oxide layer as a top contact. The multi
chamber r f glow discharge deposition system consists of four reactors within
one large chamber: the first reaction chamber for hydrogen plasma etching
and bake-out of the substrate before deposition, followed by three chambers
for deposition of p +- i- n + layers. Substrate temperatures during deposition
are 200-250 °C.
Figure 1.57 Commercial p -i-n cells fabricated by r f glow discharge are deposited
in layers as the cells are moved through consecutive isolated chambers.
Chemical vapour deposition, plasma chemical vapour deposition and

photochemical vapour deposition
Alternative methods of fabricating a-Si solar cells such as plasma c v d and
c v d (figure 1.58) for higher silanes have been successful in raising deposition
Pressure Sensor
Figure 1.58 cvd : appropriate gases are introduced into a reactor tube.
The gases react and are deposited on numerous substrate wafers.
rates to values higher than the 1-5 A s -1 rate of the conventional glow
discharge method; however, they have not as yet resulted in films of the same
quality.
A continuous mass production process of fabricating a-Si tandem cells by
the plasma c v d method has been jointly developed by Energy Conversion
Devices (ECD) Solar and Sharp. The roll-to-roll plasma c v d machine
schematically shown in figure 1.59 continuously processes a roll of stainless
steel substrate which advances from the pay-off chamber through the gas
gate into the reaction chamber. In the fabrication of the tandem thin-film p v
cell, six amorphous silicon layers are successively deposited onto a roll of
stainless steel substrate 400 mm wide and 500 m long. A typical tandem cell is
shown schematically in figure 1.60. Higher energy conversion efficiencies are
obtained with tandem thin-film cells where the bottom layer has a narrower
Stainless Steel
Gas Gate (anode)
P a y -O ff Take - Up
Vacuum Plasma
Pump
Cathode
>////// RF Generator
13.56 MHz
Figure 1.59 A schematic diagram of a roll-to-roll plasma cvd machine.
Collection
Figure 1.60 Schematic cross section of 1.8 eV/1.8 eV Sharp-ECD Solar

tandem p v cell produced by the roll-to-roll plasma c v d method.
band gap because a wider range of incoming solar spectrum is utilized. The
roll-to-roll process has been able to achieve high rate depositions of up to
18 A s“ 1. When the intrinsic layer is deposited at a rate higher than 10 A s“ \
conversion efficiencies of 8.5% and 8.11% have been obtained for the same
band-gap a-Si:H tandem cell and the dual-band-gap tandem cell using
a-Si,Ge:H and a-Si:H respectively ( a m i ; IOOmWcm-2). The roll-to-roll
process involves control of process gas isolation between adjacent reaction
chambers. Sharp-ECD Solar claim that through the introduction of proper
materials and techniques the process has been able to minimize the
Staebler-Wronski effect to the extent that cell degradation will be kept below
15% over a period of 20 years.
A new method called photo-cvD avoids bombardment of the films by high-
energy particles through the use of ultraviolet light from a low-pressure
mercury lamp to decompose the silanes. Admitting mercury vapour into the
deposition chamber greatly enhances the deposition rate. The mercury-
sensitized photo-cvD method has allowed preparation of a-Si:H layers with
properties very similar to those of glow-discharge-deposited films.
Sputtering
Sputtering deposition is the process whereby the coating material or a
constituent of the coating material is vaporized from a solid target by high-
energy particle bombardment and is deposited as a thin film on a suitable
substrate (figure 1.61). In the magnetron case, ions produced in a plasma
discharge that is magnetically confined to a region adjacent to the target are
High- Energy
Figure 1.61 Sputtering: particles strike a silicon target, sending a beam

of silicon ions towards a substrate. Hydrogen and dopant gases react
with the silicon so that a hydrogenated material forms on the substrate.
used as the bombarding particles. The composition of the film condensed

from the sputtered flux tends to be the same as that of the target which is an
important advantage of the sputtering process.
In reactive sputtering, at least one of the coating constituents is introduced
in the gas phase. Reactive sputtering of a-Si:H using silicon targets sputtered
in an argon-hydrogen atmosphere has several advantages which include the
following.
(i) The ability to control the hydrogen content by adjustment of the
hydrogen gas partial pressure.
(ii) The ease of application to alloys such as a-Si,Ge:H.
(iii) The avoidance of using SiH4 as a working gas.
The explosive nature of silane will necessitate expensive special handling
procedures in large production facilities involving glow discharge decompo
sition of SiH4.
1.3.10 Options for acquisition of solar photovoltaic technology by

Third-World countries
Solar p v technology is undergoing continuous technological development
mostly in the area of solar cell fabrication owing to efforts directed at
reducing solar cell cost per peak watt and gaining greater economic competi
tiveness as a decentralized energy source. The present section is aimed at
opening a debate on the conditions of acquisition of p v technology by Third-
World countries stressing the importance of relevant criteria as capital and
human resources availability, industry infrastructures and educational
environment.
pv m a n u f a c t u r e i n v o lv e s b a s i c a ll y t h e f o l lo w in g six c o m p o n e n t s :
(i) Silicon material: electronic grade or solar grade.

(ii) Other raw materials.
(iii) Solar p v cells.
(iv ) pv m o d u le s .
(v )Balance-of-system ( b o s ) components and subsystems.
(vi) Design and integration of effective p v energy systems.
The various aspects involved in the acquisition and effective utilization of
pv technology by Third-World countries are presented below.
1.3.10.1 Components o f photovoltaic technology

Silicon material
The most energy- and cost-intensive stage of the p v manufacturing process is
the preparation of the EG-Si in wafer or sheet form. The manufacture of EG-Si
feedstock through reduction of quartzite involves highly expensive and
complex technology while single-crystal silicon ingot and wafer production,
although regarded as high technology, is a relatively simple and adaptable
technology.
Other raw materials
(i) Various chemicals such as acids, alkalis, organic solvents and
diffusants.
(ii) Conductive pastes made of silver and aluminium.
(iii) High-transmission tempered ‘white’ glass sheets.
(iv ) Various laminating materials such as e v a and Tedlar.
(v) Various mechanical and electrical components such as junction boxes
and interconnections.
Except for the laminating materials which are produced only by a few
companies in the world, many of the Third-World countries have the
necessary infrastructure and expertise to produce the above raw materials.
Solar cells
The manufacture of high-quality solar cells on a commercial scale in the
Third-World countries requires acquisition of proper production equipment
suitable for conditions prevailing in these countries. Since solar p v tech
nology is not yet fully mature, production equipment which meets all the
stringent requirements of large-scale manufacture are generally not available
off the shelf. Most solar cell manufacturers build their own production
equipment according to their particular design requirements and manu
facturing techniques which may differ from company to company. Solar-cell-
and module-manufacturing equipment that are supplied by some companies
are more or less still in a prototype stage and are not very suitable for use
under developing-country conditions.
Photovoltaic modules
p v manufacturing technology based on single-crystal or multi-grain silicon
solar cells being offered by p v system manufacturers mostly include the

stages of production starting from module manufacture to integration of p v
systems. The production equipment available for module manufacture is
better designed and more reliable for developing-country conditions.
Balance o f system components and subsystems
For most efficient utilization of pv-generated power, b o s components and
subsystems should be specially designed for p v power systems. Two features
of p v systems, in particular, demand specially designed b o s components to be
integrated with the systems.
(i) The high cost of p v power requires that efficient b o s components be
used. The availability and abundance of conventional fossil fuels have
not encouraged optimization of the operational efficiency of electrical
and electronic components.
(ii) p v cells generate DC power. Most electrical appliances available off the
shelf use a c power since conventional fossil fuels, nuclear or hydro
power plants generate a c electricity.
With regard to the development and manufacture of b o s components
suitable for p v systems, the approach taken by developing countries differs
from that of the industrialized nations. The interest of the industrialized
nations is in improving overall operational efficiencies rather than on the
ruggedness of the products. In contrast with this, the lack of infrastructure in
the remote areas of Third-World countries necessitate minimal maintenance
of b o s components. In addition, the lower purchasing power of the popu
lation might dictate a compromise between achieving maximum operational
efficiencies and cost-effectiveness of the power system, b o s components for
use under Third-World conditions should therefore be rugged, maintenance
free and of low cost. A high-efficiency product manufactured in a developed
country and designed to be used under prevailing conditions there may fail to
perform with rated efficiency under less reliable Third-World country con
ditions and may require frequent maintenance and service, b o s components
and subsystems are therefore best designed and manufactured in the country
of application with due regard to the particular local needs and conditions.
Integration o f photovoltaic systems
Optimal p v system design must be carried out by local expertise in the
country of application since local environmental conditions as well as specific
load requirements must be considered for efficient utilization of pv-generated
power. The location-specific and load-specific nature of p v power system
design therefore make it crucial that design and integration of p v energy
systems be carried out by those familiar with local energy needs in all spheres
of activity: social, economic and educational.
The overall efficiency and cost-effectiveness of stand-alone p v systems can

be increased with an integrated approach to system design and electrical
energy utilization. Such an approach can only be attempted by the user
country itself since it requires familiarity and understanding of specific local
needs, life-styles, economic, social and educational activities as well as local
environmental conditions.
1.3.10.2 Technological options for acquisition o f photovoltaic technology

Components of p v manufacturing technology starting from the production
of pure silicon material to the finished p v modules and balance of systems are
shown in figure 1.62. The level of technical sophistication of the various
stages of production is also indicated in figure 1.63. Third-World countries
can enter into the p v manufacturing industry at any of the stages of
production depending on the available industrial infrastructure and technical
expertise.
The technology involved in the p v manufacturing processes becomes
progressively more complex as it goes up the path from the final stage, the
design and integration of p v energy systems, to the first stage, the production
of pure silicon feedstock for solar cell fabrication from quartzite. A par
ticular country therefore has the option to enter into the p v industry along
any of the stages starting from the lowest technology, b o s design and
Reduction to Si
and Purification
(a )
C rysta lliza tio n
to Wafer /Sheet Form
P V A rray
Technology
Device Fabrication — ( 3
Photovoltaic ( P V ) Module Assem bly ——( 4

Technology
(d )
Array B O S
(A rra y S tru ctu re ,
Module Netw ork)
Balance of
— System (B O S )
Technology
Nonarray BO S
(H igh E fficie n cy
Components, Power
Control Unit )
Figure 1.62 Components of p v manufacturing technology.

- Trich lo ro silan e P ro cess ( Siem ens )

Reduction to Si
and P u rificatio n S ilan e P ro cess ( Union Carbide )
-S ilic o n Flu o rid e P ro c e s s (M otorola)
C rystallization to Single - Crystal or P o lycrystallin e Ingots -

S licin g into Wafers
W a fe r/ S h e e t Form © “
_ Single - C r y s t a l, Poly - Si or a - S i
Sheets / Ribbons
_Therm al
_Ju n ctio n _ D iffusion
Formation
Ion
Im plantation
Vacuum Vaporization
Device Fabrication _ Contact Screen Printing

(2> ~ Formation
Plating
- Lapping
Polishing
A n tire fle c tiv e Coating
Cell Strin gin g
Module Assem bly

- Silicon Pottant
Module
Encapsulation
- Vacuum Lamination
Figure 1.63 Components of p v manufacturing technology and the

levels of technological complexity.
development for specific local needs and conditions to the high technology
of silicon purification and solar cell fabrication.
The following options for developing countries exist.
(i) The complete p v system can be purchased from p v system manu
facturers in developed countries with only the design and integration of
the systems performed by local expertise. This strategy is suitable for
countries with limited economic and technological resources and a
small potential for p v systems applications.
(ii) The b o s components can be developed and manufactured indig-
eneously. p v modules purchased from other countries can then be
designed into a complete system together with the locally made b o s
components. This will generate local employment opportunities with
out a need for high investment in pv among manufacturers.
(iii) Finished solar cells can be imported and assembled locally into
modules. A Bos-component-manufacturing industry will then exist
alongside the module assembly industry. This option has the potential
of reducing pv system cost without the need for very high investment in
an in d ig e n e o u s p v in d u str y a n d m a y b e su ita b le fo r c o u n tr ie s w ith a

su fficien t lo c a l m a rk et fo r p h o to v o lta ic s .
(iv) Imported single-crystal or polycrystalline wafers or sheets can be
fabricated locally into solar cells and assembled into modules. A higher
level of investment, technical expertise and infrastructure is necessary
for this option. The danger of technological obsolescence of imported
production equipment exists since p v technology is still in the develop
mental stage.
(v) Purified silicon feedstock material can be purchased and single-crystal
or polycrystalline silicon ingots or sheets grown from it. The local p v -
system-manufacturing industry therefore includes all the stages from
ingot or sheet crystal growth to module assembly to p v system design
and installation. Although a very high capital investment is needed for
this, the potential exists for supplying EG-Si to the semiconductor
industry as well as to the local p v industry. This approach is suitable if a
large potential demand for p v energy technologies exists locally.
(vi) The last option requiring the highest level of capital investment offers
the greatest security in terms of uncertainties in supply of either purified
silicon feedstock material or silicon wafers or sheets from pv-system-
manufacturing countries. Greater economic viability of the p v industry
exists here because of the possibility of supplying EG-Si both to the
larger semiconductor industry and to the p v industry. Export potential
to other developing countries also exists for any of the products from
the various stages of p v manufacture.
BIBLIOGRAPHY
Aulich H A 1986 Proc. 2nd Int. Photovoltaic Science and Engineering Conf., Beijing,
August 1986 pp 214-7
Aulich H A etal 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf., Kobe,
November 1984 pp 777-80
Boer K W 1987 Solar Wind Technol. 4 21-35
Bolton J R 1983 Solar Energy 31 483-502
Delahoy A E 1987 Solar Cells 21 153-66
Ermer W 1985 Proc. 6th Commission of the European Communities Conf. on
Photovoltaic Solar Energy, 15-19 April 1985 (Dordrecht: Reidel)
Ermer W et al 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf, Kobe,
Fuhs W 1987 Solar Wind Technol. 4 7-15
Grabmaier J G 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf., Kobe,
November 1984 pp 537—42
Grabmaier J G et al 1986 Proc. 2nd Int. Photovoltaic Science and Engineering Conf,
Beijing, August 1986 pp 174-7
Helmreich D 1986 Proc. 2nd Int. Photovoltaic Science and Engineering Conf., Beijing,
August 1986 pp 199-202
Hirobe T et al 1986 Proc. 2nd Int. Photovoltaic Science and Engineering Conf., Beijing,
August 1986 pp 214-7
Hoyler G 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf., Kobe,
Ingle W M 19— Recent Advances in Solar Silicon Purification Technology (Phoenix,
AZ: Semiconductor Products Group, Motorola Inc.)
Kuwano Y 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf, Kobe,
Marshall S L 1967 Microelectronic Technology (Boston, MA: Boston Technical
Publishers)
Mufti D 19— Silicon Technology Revolution—A Challenge for the Developing Coun
tries to Accept or Reject? (Islamabad: National Institute of Silicon Technology)
Panyakeow S 1984 Proc. 3rd Asian School on Solar Energy Harnessing (Bangkok:
Asian Institute of Technology—KMITL)
Pernisz U C et al 1987 Solar Cells 21 195-203
Ravi K V 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf, Kobe,
Siffert P 1982 Proc. 4th Commission of the European Communities Conf. on Photovol
taic Solar Energy, Stresa, 10-14 May 1982 (Dordrecht: Reidel)
Tabata K et al 1984 Proc 1st Int. Photovoltaic Science and Engineering Conf, Kobe,
Thorton J A 1987 Solar Cells 21 41-54
Vanier P E and Su F C 1987 Solar Cells 21 167-75
Vemstrom G D 1987 Solar Cells 21 141-6
Werner P G 1982 Proc. 4th Commission of the European Communities Conf. on
Photovoltaic Solar Energy, Stresa, 10-14 May 1982 (Dordrecht: Reidel)
October 1981 Westinghouse R& D Center; Monitor 8, vol II, issue IV
1980 Sunworld 4
1988 Solar Cell Technology, Proc. Workshop on Planning of Network Projects in
Materials Science and Solar Energy, Arusha, 21-26 November 1988 (Meudon:
Laboratoire de Physique des Solides, CNRS)
December 1985 Summary of Research and Development of Amorphous Solar Cells—
Sunshine Project (Solar Technology Development, New Energy Development
Organization)
1984 Basic Photovoltaic Principles and Methods (Golden, CO: Solar Energy Research
Institute)
Solar Cell Technologies
Aulich H. A. 1986 Proc. 2nd Int. Photovoltaic Science and Engineering Conf., Beijing, August
1986 pp 214217
Aulich H. A. et al 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf., Kobe,
November 1984 pp 777780
Ber K. W. 1987 Solar Wind Technol. 4 2135
Bolton J. R. 1983 Solar Energy 31 483502
Delahoy A. E. 1987 Solar Cells 21 153166
Ermer W. 1985 Proc. 6th Commission of the European Communities Conf. on Photovoltaic
Solar Energy, 1519 April 1985 (Dordrecht: Reidel)
Ermer W. et al 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf., Kobe,
Fuhs W. 1987 Solar Wind Technol. 4 715
Grabmaier J. G. 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf., Kobe,
Grabmaier J. G. et al 1986 Proc. 2nd Int. Photovoltaic Science and Engineering Conf., Beijing,
August 1986 pp 174177
66 Helmreich D. 1986 Proc. 2nd Int. Photovoltaic Science and Engineering Conf., Beijing,
August 1986 pp 199202
Hirobe T. et al 1986 Proc. 2nd Int. Photovoltaic Science and Engineering Conf., Beijing, August
1986 pp 214217
Hoyler G. 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf., Kobe, November
1984 pp 791795
Ingle W. M. 19Recent Advances in Solar Silicon Purification Technology (Phoenix, AZ:
Semiconductor Products Group, Motorola Inc.)
Kuwano Y. 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf., Kobe, November
1984 pp 571576
Marshall S. L. 1967 Microelectronic Technology (Boston, MA: Boston Technical Publishers)
Mufti D. 19Silicon Technology RevolutionA Challenge for the Developing Countries to Accept or
Reject? (Islamabad: National Institute of Silicon Technology)
Panyakeow S. 1984 Proc. 3rd Asian School on Solar Energy Harnessing (Bangkok: Asian
Institute of TechnologyKMITL)
Pernisz U. C. et al 1987 Solar Cells 21 195203
Ravi K. V. 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf., Kobe, November
1984 pp 4349
Siffert P. 1982 Proc. 4th Commission of the European Communities Conf. on Photovoltaic Solar
Energy, Stresa, 1014 May 1982 (Dordrecht: Reidel)
Tabata K. et al 1984 Proc 1st Int. Photovoltaic Science and Engineering Conf., Kobe, November
1984 pp 809812
Thorton J. A. 1987 Solar Cells 21 4154
Vanier P. E. and Su F. C. 1987 Solar Cells 21 167175
Vemstrom G. D. 1987 Solar Cells 21 141146
Werner P. G 1982 Proc. 4th Commission of the European Communities Conf on Photovoltaic
Solar Energy, Stresa, 1014 May 1982 (Dordrecht: Reidel)
October 1981 Westinghouse R&D Center; Monitor 8, vol II, issue IV
1980 Sunworld 4
1988 Solar Cell Technology, Proc. Workshop on Planning of Network Projects in Materials
Science and Solar Energy, Arusha, 2126 November 1988 (Meudon: Laboratoire de Physique
des Solides, CNRS)
December 1985 Summary of Research and Development of Amorphous Solar CellsSunshine
Project (Solar Technology Development, New Energy Development Organization)
1984 Basic Photovoltaic Principles and Methods (Golden, CO: Solar Energy Research Institute)
The Photovoltaic Generator
Backus C. E. (ed) 1976 Solar Cells (New York: IEEE)
Lasnier F. and Follea D. Lecture Notes (Bangkok: Asian Institute of Technology)
Rauschenbach H. S. 1980 Solar Cell Array Design Handbook (New York: Van Nostrand-
Reinhold)
Roger J. Lecture Notes (Lyon: University of Lyon)
Rousset P. 1983 Lecture, SDRL-Chulalongkorn University, Thailand, 22 June 1983 (Bangkok:
Asian Institute of Technology)
Network of Solar Cells, Modules and Arrays

Favolle D. P. 1979 Thesis Universit Claude Bernard Lyon I, France
Storage Batteries for Photovoltaic Power Systems

BCI 1982 Battery Service Manual (Battery Council International)
Paul T. D. 1981 How to Design an Independent Power System: Best Energy Systems
USDOE 1979 Handbook for Battery Energy Storage for PV Power Systems (Washington, DC:
US Department of Energy)
USDOE July 1980 Solar PV Applications Seminar: Design, Installation and Operation of Small,
Stand-alone PV Power Systems (Washington, DC: US Department of Energy)
Vinal G. W. 1955 Storage Batteries 4th edn
Bode H. 1977 Lead-Acid Batteries
Buresch M. 1983 PV Energy Systems: Design and Installation
Calloway T. M. 1986 Design of an Intermediate-Sized, Autonomous PV-Diesel Power Plant
(Albuquerque, NM: Sandia National Laboratories)
CIRDC 1980 Compatible Battery for Solar Electric Power Systems Application (Chloride India
Research and Development Centre)
Macomber H. L. and Ruzek J. B. 1981 Photovoltaic Stand-Alone Systems: Preliminary
Engineering Design Handbook (Washington, DC: US Department of Energy)
Electronic regulation
Lasnier F. 1985 Lecture Notes ET24 (Bangkok: Asian Institute of Technology)
Power Conditioning
Lasnier F. and Follea D. Lecture Notes ET24 (Bangkok: Asian Institute of Technology)
Adaptation of a Positive-displacement Pump Directly Connected to a

Photovoltaic Generator
Matsuo H. and Murozono M. 1984 Proc. 1st Int. Photovoltaic Science and Engineering Conf.,
Kobe, November 1984 Technical Digest
Roger J. 1979 Solar Energy 23 193198
Wagdy R. A. 1985 Solar Cells 14 2742
Centrifugal Photovoltaic Pumping: Theoretical Study of Direct-coupled
Pump
Easwarakhanthan T. 1981 Thesis No ET-8110 Asian Institute of Technology, Bangkok
International Test Procedures for Photovoltaic Modules

Anon 1977 ERDA-NASA Document No 1022/77/16 (Washington, DC: Energy Research and
Development Administration-National Aeronautics and Space Administration)
Anon 1980 CEC Specification No. 101 Issue 2; EUR Document 7978 EN (Luxembourg:
Commission of the European Communities)
Anon 1984a EUR Specification No 501 (Luxembourg: Commission of the European
Communities)
Anon 1984b EUR Specification No 502 Issue 1 (Luxembourg: Commission of the European
Communities)
Bird R. E. , Hulstrom R. L. and Lewis L. J. 1983 Solar Energy 30
Treble F. C. 1976 Proc. 2nd ERDA-NASA Terrestrial Photovoltaic Measurement Workshop,
Baton Rouge, LA
Treble F. C. 1985 Proc. 4th Asian School on Solar Energy Harnessing, Bangkok, 1220
December 1985 (Bangkok: Asian Institute of Technology)
Asian Institute of Technology Photovoltaic Module Test Bed System

Braunstein A. , Bany J. and Appelbaum J. 1977 Energy Conversion 17 16
Treble F. C. 1985 Proc. 4th Asian School on Solar Harnessing, Bangkok, 1220 December 1985
(Bangkok: Asian Institute of Technology)
Ashok S. and Pande K. P. 1985 Solar Cells 14 6181
Goldstein L. and Case G. R. 1987 Solar Energy 21 3743
International Rectifier hexfet Databook
Kar S. , Ashok S. and Fonash S. J. 1980 J. Appl. Phys. 51 3417
Lasnier F. and Ang T. G. 1987 Photovoltaic Test and Instrumentation Seriespvmod 1.0
(Bangkok: Energy Technology Division, Asian Institute of Technology)
Reinhold)
Testing of Photovoltaic Modules Under Natural Conditions Using the

Asian Institute of Technology Photovoltaic Module Test Bed System
Braunstein A. , Bany J. and Appelbaum J. 1977 Energy Conversion 17 16
Field Trial Procedure for a Photovoltaic Centrifugal Pumping System

Herrmann B. , Karl H. , Kopf E. and Lehner G. 1987 Solar Energy 4 275279
Easwarakhanthan T. 1981 Thesis No ET-81-10 Asian Institute of Technology, Bangkok
Lasnier F. and Follea D. 1983 Photovoltaic Systems Engineering Lecture Notes (Bangkok:
Asian Institute of Technology)
Myles S. Lecture Notes (Danfoss Cie)
Nir Z. Pumps and Pumping Lecture Notes (Bangkok: Asian Institute of Technology)
Warring R. H. Pumps Selection Systems and Applications (Morden, Surrey: Trade and
Technical Press)
Battery-testing Method for Low-water-loss and Starting, Lighting and
Ignition (Automotive) Batteries
BCI 1982 Battery Service Manual (Battery Council International)
Bode H. 1977 Lead-Acid Batteries
Buresch M. 1983 PV Energy Systems: Design and Installation
Calloway T. M. 1986 Design of an Intermediate-Sized, Autonomous PV-Diesel Power Plant
CIRDC 1980 Compatible Battery for Solar Electric Power Systems Application (Chloride India
Research and Development Centre)
Macomber H. L. and Ruzek J. B. 1981 Photovoltaic Stand-Alone Systems: Preliminary
Engineering Design Handbook (Washington, DC: US Department of Energy)
Paul T. D. 1981 How to Design an Independent Power System: Best Energy Systems
USDOE 1979 Handbook for Battery Energy Storage for PV Power Systems (Washington, DC:
US Department of Energy)
USDOE July 1980 Solar PV Applications Seminar: Design, Installation and Operation of Small,
Stand-alone PV Power Systems (Washington, DC: Department of Energy)
The Sizing of Stand-alone Photovoltaic Power Systems: A Dynamic

Simulation Approach
Ang T. G 1984 AIT Thesis No ET-84-10 Asian Institute of Technology, Bangkok
Chungpaibulpatana S. 1981 AIT Thesis No ET-81-9 Asian Institute of Technology, Bangkok
EPRI 1981 Reliability Index for Power Systems EL-1773 Research Project 1353-1 (Palo Alto,
CA: Electrical Power Research Institute)
Exell R. H. B. 1981 Solar Energy 26 161168
Gerald C. F. 1978 Applied Numerical Analysis 2nd edn (Reading, MA: Addison-Wesley)
Hoover E. R. 1980 Proc. 14th Photovoltaic Specialists Conf. San Diego, CA, 710 January 1980
(New York: IEEE) pp 12581261
Lasnier F. and Ang T. G. 1986 Computer Package for Photovoltaic System Sizing, Design and
Financial Evaluation (Bangkok: Asian Institute of Technology)
Rao S. S. 1984 Optimization Theory and Applications (New York: Wiley Eastern)
Reinhold)
Servant J. M. 1986 Proc. 9th Bienn. Congr. of the Internal Solar Energy Society (Oxford:
Pergamon)
Wang Y. J. 1987 The Sizing of Stand-Alone Photovoltaic Systems: AIT Research Study
(Bangkok: Asian Institute of Technology)
Prediction of Photovoltaic System Performance Using Cumulative

Frequency Curves of Irradiance
Lasnier F. and Ang T. G. 1988 Solar Photovoltaic Handbook (Bangkok: Asian Institute of
Technology)
Wang Y. J. 1987 The Sizing of Stand-alone Photovoltaic Systems (Bangkok: Asian Institute of
Technology)
Canavos G. C. 1984 Applied Probability and Statistical Methods (Little Brown)
Harnett D. L. 1982 Statistical Methods (Reading, MA: Addison-Wesley)
Kreyszig E. 1970 Introductory Mathematical Statistics (New York: Wiley)
Reddy T. A. 1987 The Design and Sizing of Active Solar Thermal Systems (Oxford: Clarendon)
Financial Evaluation of Renewable Energy Projects
Barish N. N. and Kaplan S. 1978 Economic Analysis for Engineering and Managerial Decision
Making (New York: McGraw-Hill)
Horst F. and Gerhard O. 1985 A Guide to the Financial Evaluation of Investment Projects in
Energy Supply (GTZ)
Jones B. W. 1982 Inflation in Engineering Economic Analysis (New York: Wiley)
Park W. R. 1973 Cost Engineering AnalysisA Guide to the Economic Evaluation of Engineering
Projects (New York: Wiley)
Perino A. M. 1979 Report No SAND 780931 (Albuquerque, NM: Sandia National Laboratories)
Sassone P. G. and Schaffer M. A. 1978 CostBenefit AnalysisA Handbook (New York: Academic
Press)
Wright J. C. 1983 Technoeconomics: Concept and Cases (Tokyo: Asian Productivity
Organization)
Life-cycle Cost Comparisons of Photovoltaic-powered Systems and

Most Likely Conventional Alternative System
Eskenazi D. , Kerner D. and Slominski L. 1986 Technical Report No SAND85-7018/2
A Comparative Assessment of Photovoltaics and Handpumps for Rural

Water Supply: A Case Study for Thailand
Cabraal A. , Siess A. , Slominski L. , Buresch M. and Kenna J. 1987 Technical Report No
SAND87-7015 (Albuquerque, NM: Sandia National Laboratories)
Lasnier F. 1989 AIT Report on Motor-Pump Market Survey (Bangkok: Asian Institute of
Technology)
United Nations Development Programme-World Bank 1986 Report No INT/81/026 (Bangkok:
Ministry of Public Health)
Arlosoroff S. , Tschannerl G. , Grey D. , Journey W. , Karp A. , Langeneffe O. and Roche R.
1987 Community Water Supply: The Handpump Option (Washington, DC: UNDP-World Bank)
Life-cycle Cost Comparison of an Alternative Power Supply for a

Portable Pocket-sized Stereo Cassette Tape Recorder
Eskenazi D. , Kerner D. and Slominski L. 1986 Technical Report No SAND85-7018/2
A Simple MetalOxideSemiconductor Field-effect Transistor Electronic

Variable Load
Millman J. and Taub H. 1965 Pulse, Digital and Switching Waveforms (New York: McGraw-Hill)
Equipment Accommodation
Heber G. and Seufert C. 1979 Solar Cooling of Buildings, German Appropriate Technology
Exchange
Wishart A. J. 1985 Telecommun. J. Aust. 35

Googlepreview

Uploaded by

Copyright:

Available Formats

Googlepreview

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Googlepreview

Uploaded by

Copyright:

Available Formats

Photovoltaic Engineering Handbook

Photovoltaic Engineering Handbook

France Lasnier and Tony Gan Ang

Asian Institute of Technology, Bangkok, Thailand

Adam Hilger, Bristol and New York

All rights reserved. No part of this publication may be reproduced, stored in

British Library Cataloguing in Publication Data

US Library of Congress Cataloging-in-Publication Data

Consultant Editor: Dr R H Taylor, CEGB

Published under the Adam Hilger imprint by IOP Publishing Ltd

Typeset by MS Filmsetting Limited, Frome, Somerset

Part I Technological Processes 1

1 Solar Cell Technologies 3

Part II The Photovoltaic Generator 67

2 The Photovoltaic Generator 69

2.4.1 Influence of solar irradiance intensity 79

3 Network of Solar Cells, Modules and Arrays 83

Part III Photovoltaic Systems Engineering 99

4 Storage Batteries for Photovoltaic PowerSystems 101

4.6.2 Undercharging 130

5 Electronic Regulation 139

6 Power Conditioning 153

7 Adaptation of a Positive-displacement Pump Directly Connected to a

8 Centrifugal Photovoltaic Pumping: Theoretical Study of Direct-

Part IV Characterization and Testing Methods 199

9 International Test Procedures for Photovoltaic Modules 201

9.2.6 Current-voltage characteristics measurement using a

10 Asian Institute of Technology Photovoltaic Module Test Bed System 223

10.4.2 Data-processing subprograms 238

11 Testing of Photovoltaic Modules under Natural Conditions using the

12 Characterization Procedure of a Photovoltaic Refrigeration System 285

13 Field Trial Procedure for a Photovoltaic Centrifugal Pumping System 297

13.3.1 Characterization of components 300

14 Battery-testing Method for Low-water-loss and Starting, Lighting and

Part V Sizing Procedure 337

15 The Sizing of Stand-alone Photovoltaic Power Systems: A Dynamic

15.4 Economic analysis of a photovoltaic hybrid system 349

16 Prediction of Photovoltaic SystemPerformance using Cumulative

Part VI Economic Analysis 371

17 Financial Evaluation of Renewable Energy Projects 373

17.4 Procedure of financial analysis 385

18 Life-cycle Cost Comparisons of Photovoltaic-powered Systems and

19 A Comparative Assessment of Photovoltaics and Handpumps for

20 Life-cycle Cost Comparison of an Alternative Power Supply for a

Part VII Instrumentation 501

21 A Simple Metal-Oxide-Semiconductor Field-effect Transistor

22 Equipment Accommodation 516

22.2.2 Thermally thin buildings 517

This book is intended to be a comprehensive introduction to photovoltaic

(i) Material preparation and shaping.

The conventional technological process, the flow chart of which is presen­

(figure 1.3), i.e. junction diffusion, contact elaboration and anti-reflection

Figure 1.2 Production and slicing of silicon ingots.

Sheets Die cost Big Thin Other

Hydrogenated Si - H Thin Ceram ic

Figure 1.5 Spectrum of silicon technology.

A review of recent progress includes the following.

1.2 LARGE-SCALE PRODUCTION OF SILICON BASE

1.2.1 Conventional technology process

The conventional technological process, the flow chart of which is presen

Figure 1.11 Solar-grade synthesis from partially refined silicon by direc

cz process which include a rechargeable feature allowing continuous oper