Rapid Routes for Synthesis of CIGS Absorbers

MISSING IMAGE

Material Information

Title:
Rapid Routes for Synthesis of CIGS Absorbers
Physical Description:
1 online resource (402 p.)
Language:
english
Creator:
Krishnan, Rangarajan
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Chemical Engineering
Committee Chair:
Anderson, Timothy J
Committee Members:
Ren, Fan
Ziegler, Kirk
Craciun, Valentin
Park, Chinho

Subjects

Subjects / Keywords:
cigs -- kinetics -- pathways -- xrd
Chemical Engineering -- Dissertations, Academic -- UF
Genre:
Chemical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
The chalcopyrite solid solution Cu(InxGa1-x)Se2(CIGS) is a commercially emerging thin film absorber material based on thepromise of low manufacturing costs and high conversion efficiency (championcell now exceeds 20%).  The primarychallenge in achieving low processing cost is increasing the synthesis rate ofCIGS at lower temperature.  Recognizingthis challenge the national solar technology roadmap calls for decreasing theabsorber synthesis time to 2 min by 2015 while retaining the efficiency.  However, the synthesis of CIGS is not limitedby rate of deposition of elements, but is rather limited by rate of reaction ofstarting precursors that form CIGS. Reaction pathways and kinetics were followedfor understanding CIS/CIGS formation both qualitatively and quantitativelyusing time-resolved X-ray diffraction for two common processes such asco-evaporation and metal-selenization. First, in-situ X-ray diffraction was used to understand metal-seleniumbinary system and the results were consistent with the sequence predicted bythermodynamic phase diagram. Then, the reaction pathways and kinetics ofternary CIS formation was followed using bilayer precursor structure ofglass/Mo/g-In2Se3/CuSeyand glass/Mo/g-In2Se3/b-Cu2Se.  Variation in kinetics with addition ofgallium to the group III sub-lattice was also developed.  In-situselenization pathways and kinetics were studied using X-ray diffraction forvarious metal precursor structures (bilayer metal precursors, stacked elementallayers, multilayer structures). The gallium distribution after annealing wasfound to be dependent on the precursor structure. Also the effect of nucleationin kinetics resulting with change in the order of deposition of metals was alsostudied.  The effect of sodium inpathways and kinetics of absorber formation was studied by using sodium-dopedmolybdenum substrates. Furthermore, reaction pathways and kineticsof MoSe2 formation was studied by performing in-situ selenization of molybdenum substrates. It was found thatoxygen presence leads to formation of MoO2 along with the formationof MoSe2.  The devicedegradation mechanism due to cadmium diffusion in to CIGS was studied for acompleted device without top contacts for industrial sample using in-situ X-ray diffraction.  Cadmium diffusion initiated for temperatures>400 ?C forming CuCd2 (In, Ga)Se4 compoundresulting in complete failure of the device.  A new process for high rate synthesis ofCIGS has been developed using nanoparticle based approach involving peritecticdecomposition of compounds to yield a liquid assisted growth of CIGS.  The nanoparticles involved in this study weresynthesized by simple reaction in alcoholic medium. Another process for highrate synthesis of CIGS has been developed using binary nanoparticles basedapproach emulating the co-evaporation process. The binary nanoparticles forthis study were synthesized by colloidal route using coordinatingsolvents.  An efficiency of 1.67% hasbeen obtained for the same process. Thermodynamic assessment of the CIS-CGS andCu2Se-Ga2Se3 pseudobinary was performed topredict the phase diagrams. Sub-lattice models were used to express Gibbsenergy as a function of temperature.  Themodel developed agrees well with the experimental data reported in theliterature.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Rangarajan Krishnan.
Thesis:
Thesis (Ph.D.)--University of Florida, 2012.
Local:
Adviser: Anderson, Timothy J.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-08-31

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2012
System ID:
UFE0044459:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 RAPID ROUTES FOR SYNTHESIS OF CIGS ABSORBERS By RANGARAJAN KRISHNAN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHI LOSOPHY UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 Rangarajan Krishnan

PAGE 3

3 To my family

PAGE 4

4 ACKNOWLEDGMENTS First of all, I would like to thank Dr. Timothy Anderson for providing tons of support both in professional and personal life. He has always take time out of his busy schedule and helped me in every way he can. I would also like to thank Dr. Andrew Payzant from Oak Ridge National Laboratory for his support in my professional life. I have learnt and applied lot of his advises in my PhD research. Par t of the research was conducted Program which is sponsored by the U.S. department of Energy, Office of Energy Efficiency and Renewable Energy, Vehicle Technologies Program. I would also to thank my committee members Dr. Chinho Park, Dr. Fan Ren, Dr. Kirk Ziegler and Dr. Valentin Craciun for all the help provided. Dr. Park has been of immense help providing guidance in my PhD work. We really had good time when he was here i n UFL on sabbatical. I would also to than k Melanie Kirkham (ORNL), Roberta Peascoe ( ORNL), Tom Watkins ( ORNL) and Edgar Lara Curzio (ORNL) for their help with my experiments at Oak Ridge National Laboratory. My special thanks to Dr. Paul Holloway and Dr. M ark Davidson for their help in training me with the RTA system. I would also like to thank Dr. Kyoung Kim, Dr. Rommel Noufi, Dr. Jeff Britt, Dr. Urs Schoop and Dr. Ryan Kaczynski for providing me samples whenever I needed. I would also like to thank Dr. Ch ristoph Adelham for p roviding molybdenum sodium substrates for my studies. I would also like to thank Tze Bin Song for his help wi th TEM characterization. I would like to all the group members for their help and insights in my research. I would also like to thank Shirley Kelly, Debbie Sandoval Jim Hinnant, Chuck Rowland, Dennis Vince, Christine Goudy and Carolyn Miller for all the support provided during my graduate work.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 10 LIST OF ABBREVIATIONS ................................ ................................ ........................... 21 ABSTRACT ................................ ................................ ................................ ................... 22 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 25 Photovoltaics ................................ ................................ ................................ .......... 25 Working of Solar Cell ................................ ................................ .............................. 25 Classification of Solar Cells ................................ ................................ .............. 27 CIGS Solar Cells ................................ ................................ .............................. 28 Crystal structure of CIGS ................................ ................................ ........... 29 Absorber composition and its effect on performance ................................ 30 Defect chemistry ................................ ................................ ........................ 31 Role of impurities in CIGS ................................ ................................ .......... 32 Absorber Growth Techniques ................................ ................................ ................. 34 Co evaporation from Elemental Sources ................................ .......................... 35 Selenization of Metallic Precursors ................................ ................................ .. 37 Nanoparticle Appro ach ................................ ................................ ..................... 38 Electrodeposition ................................ ................................ .............................. 39 Problem Statement ................................ ................................ ................................ 40 2 REACTION PATH WAYS AND KINETICS IN METAL SELENIUM BINARY SYSTEM ................................ ................................ ................................ ................. 56 Overview ................................ ................................ ................................ ................. 56 Experimental ................................ ................................ ................................ ........... 58 Results and Discussion ................................ ................................ ........................... 60 Glass/Cu/Se Bilayer Precursor ................................ ................................ ......... 60 Glass/Cu+Se Comixed Precursor ................................ ................................ ..... 61 Glass/In/Se Bilayer Precursor ................................ ................................ .......... 62 Glass/In+Se Comixed Precursor ................................ ................................ ...... 64 Glass/Ga/Se Bilay er Precursor ................................ ................................ ......... 65 Glass/Ga+Se Comixed Precursor ................................ ................................ .... 66 Isothermal Annealing and Reaction Kinetics ................................ ........................... 66 Summary ................................ ................................ ................................ ................ 70

PAGE 6

6 3 REACTION PATHWAYS AND KINETICS OF CIS FORMATION FROM BILAYER COMPOUND PRECURSORS ................................ .............................. 101 Overview ................................ ................................ ................................ ............... 101 Experimental ................................ ................................ ................................ ......... 103 Temperature Ramp Annealing ................................ ................................ .............. 105 Gl ass/Mo/In 2 Se 3 / CuSe+ Cu 2 Se/Se Precursor ................................ ........... 105 Glass/Mo/In 2 Se 3 / Cu 2 Se/Se Precursor ................................ ......................... 108 Glass/Mo/ (In,Ga) 2 Se 3 /CuSe Precursor ................................ ........................ 110 Isothermal Annealing and Reaction Kinetics ................................ ......................... 111 Glass/Mo/In 2 Se 3 / CuSe+ Cu 2 Se/Se Precursor ................................ ........... 111 Glass/Mo/In 2 Se 3 / Cu 2 Se/Se Precursor ................................ ......................... 115 Glass/Mo/ (In,Ga) 2 Se 3 /CuSe Precursor ................................ ........................ 116 Summary ................................ ................................ ................................ .............. 117 4 GALLIUM DISTRIBUTION STUDIES EMPLOYING BILAYER METALLIC PRECURSORS ................................ ................................ ................................ .... 142 Overview ................................ ................................ ................................ ............... 142 Experimental ................................ ................................ ................................ ......... 143 Temperature Ramp Annealing ................................ ................................ .............. 144 Glass/Mo/CuIn/CuGa Precursor ................................ ................................ ..... 144 Glass/Mo/CuGa/CuIn Precursor ................................ ................................ ..... 146 Glass/Mo/CuIn/CuGa/Se Precursor ................................ ............................... 147 Glass/Mo/CuGa/CuIn/Se Precursor ................................ ............................... 150 Glass/Mo/CuIn/CuGa Precursor + Se Vapor ................................ ................. 151 Glass/Mo/CuGa/CuIn Precursor + Se Vapor ................................ .................. 153 Isothermal Annealing and Reaction Kinetics ................................ ......................... 154 Summary ................................ ................................ ................................ .............. 155 5 REACTION PATHWAYS AND KINETICS OF STACKED ELEMENTAL LAYER FORMING CuIn x Ga 1 x Se 2 ................................ ................................ ..................... 186 Overview ................................ ................................ ................................ ............... 186 Experimental ................................ ................................ ................................ ......... 187 Temperature Ramp Annealing ................................ ................................ .............. 188 Glass/Mo/Cu/Ga/In/Se Precursor ................................ ................................ ... 188 Glass/Mo/Ga/In/Cu/Se Precursor ................................ ................................ ... 194 Isothermal Annealing and Reaction Kinetics ................................ ......................... 195 Summary ................................ ................................ ................................ .............. 197 6 REACTION PATHWAYS AND KINETICS OF MoS e 2 FORMATION .................... 214 Overview ................................ ................................ ................................ ............... 214 Experimental ................................ ................................ ................................ ......... 215 T emperature Ramp Annealing ................................ ................................ .............. 216 Isothermal Annealing and Reaction Kinetics ................................ ......................... 217

PAGE 7

7 Summary ................................ ................................ ................................ .............. 221 7 SELENIZATION PATHWAYS AND KINETICS OF REPEATED CU/GA/IN ELEMENTAL LAYERS ................................ ................................ ......................... 235 Overview ................................ ................................ ................................ ............... 235 Experimenta l ................................ ................................ ................................ ......... 236 Temperature Ramp Annealing of Metallic Precursor ................................ ............ 237 Temperature Ramp Selenization ................................ ................................ .......... 239 Isothermal Annealing and Reaction Kinetics ................................ ......................... 241 Summary ................................ ................................ ................................ .............. 242 8 EFFECT OF SODIUM ON SELENIZATION OF CU GA/IN METALLIC PRECURSOR ................................ ................................ ................................ ....... 261 Overview ................................ ................................ ................................ ............... 261 Experimental ................................ ................................ ................................ ......... 262 Tempe rature Ramp Selenization of Metallic Precursor ................................ ......... 263 Isothermal Annealing and Reaction Kinetics ................................ ......................... 266 Summary ................................ ................................ ................................ .............. 267 9 NANOPARTICLE ROUTE FOR SYNTHESIS OF CIS ABSORBERS .................. 284 Overview ................................ ................................ ................................ ............... 284 Experimental ................................ ................................ ................................ ......... 285 Temperature Ramp Annealing ................................ ................................ .............. 286 Copper Rich Nanoparticles ................................ ................................ ............. 286 Copper Poor Nanoparticles ................................ ................................ ............ 287 Nearly Stoichiometric Nanoparticle ................................ ................................ 288 Ink Formulation and Film Formation ................................ ................................ ..... 289 CIGS Device Fabrication using Binary Nanoparticles ................................ ........... 291 (In,Ga) 2 Se 3 Nanoparticle Synthesis ................................ ................................ 292 Cu 2 Se Nanoparticle Synthesis ................................ ................................ ....... 293 Rapid Thermal Annealing ................................ ................................ ............... 293 Summary ................................ ................................ ................................ .............. 2 94 10 DEVICE DEGRADATION STUDIES OF CIGS SOLAR CELLS ............................ 317 Overview ................................ ................................ ................................ ............... 317 Experimental ................................ ................................ ................................ ......... 318 Temperature Ramp Annealing ................................ ................................ .............. 318 Isothermal Annealing and Reaction Kinetics ................................ ......................... 320 Summary ................................ ................................ ................................ .............. 321 11 ELLINGHAM DIAGRAM OF CU IN GA SE O SYSTEM AND SELENIUM TRANSPORT STUDIES FOR GROWTH OF CIGS ................................ .............. 330 Overview ................................ ................................ ................................ ............... 330

PAGE 8

8 Thermodynamics ................................ ................................ ................................ .. 331 Ellingham Diagram for Cu In Ga Se O System ................................ .................... 334 Selenization Reactor: Fluent Simulations ................................ ............................. 337 Summary ................................ ................................ ................................ .............. 343 12 PHASE DIAGRAM ASSESMENT OF PSEUDOBINARY C u 2 S e Ga 2 Se 3 AND CuInSe 2 CuGaSe 2 SYSTEM ................................ ................................ ................. 368 Overview ................................ ................................ ................................ ............... 368 Thermodynamic Optimization ................................ ................................ ............... 370 Cu 2 Se Ga 2 Se 3 Pseu dobinary system ................................ ............................. 371 CIS CGS Pseudobinary system ................................ ................................ ..... 373 Summary ................................ ................................ ................................ .............. 374 13 CONCLUSIONS AND FUTURE WORK ................................ ............................... 384 LIST OF REFERENCES ................................ ................................ ............................. 389 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 402

PAGE 9

9 LIST OF TABLES Table page 1 1 Solar cell efficiency list based on different technologies ................................ ..... 47 2 1 Kinetics parameters calculated from Avrami and parabolic model ..................... 71 3 1 Mole fraction of as deposited and annealed precursor as determined by ICP OES ................................ ................................ ................................ .................. 119 3 2 Phases identified by XRD after RTA for 2 min ................................ .................. 120 5 1 ICP mole fraction of precursors before and after annealing ............................. 198 5 2 Rate constants of precursors obtained from Avrami model .............................. 199 7 1 SEM EDS composition measurement at matrix and island for as deposited and temperature ramp annealed samples ................................ ........................ 244 8 1 Phase fraction obtained from Rietveld refinement using high resolution XRD data ................................ ................................ ................................ .................. 269 12 1 Data used for optimization of Cu 2 Se Ga 2 S e 3 phase diagram ........................... 375 12 2 Optimized parameters for Gibbs energy for Cu 2 Se Ga 2 Se 3 pseudobinary system. ................................ ................................ ................................ ............. 376 12 3 Optimized paramet ers for Gibbs energy for CIS CGS pseudobinary system. .. 378 13 1 Summary of reaction kinetics from different precursor structures ..................... 388

PAGE 10

10 LIST OF FIGURES Figure p age 1 1 Spectrum of solar radiation for AM0 and AM1.5 conditions based on ASTM ..... 48 1 2 I V curve of a solar cell device ................................ ................................ ............ 49 1 3 Device structure of CIGS solar cell ................................ ................................ ..... 50 1 4 Crystal structure of CIGS solar cell ................................ ................................ ..... 51 1 5 NREL three stage co evaporation process for solar cell fabrication ................... 52 1 6 Cost of manufacturing of CIGS modules as a function of efficiency improvement ................................ ................................ ................................ ....... 53 1 7 Cost of manufacturing of CIGS modules with increase in module efficiency ...... 54 1 8 Cost of manufacturing of CIGS modules with increase in process scale ............ 55 2 1 Calculated Cu Se phase diagram using op timized Thermocalc database .......... 72 2 2 Calculated In Se phase diagra m using op timized Thermocalc database ........... 73 2 3 Calculated Ga Se phase diagram using o ptimized Thermocalc database ......... 74 2 4 Temperatu re ramp annealing of glass/Cu/Se bilayer precursor ......................... 75 2 5 TEM image of quenched bilayer precur sor of glass/Cu/Se ................................ 76 2 6 Temperat ure ramp annealing of MEE grown glass/Cu/Se bilayer precursor ...... 77 2 7 Temperature ramp annealing of co evaporated glass/Cu+Se comixed precursor ................................ ................................ ................................ ............ 78 2 8 TEM images of quenched bilayer precursor of glass/Cu+Se. ............................. 79 2 9 Temperature ramp annealing of MEE grown gl ass/Cu+Se comixed precursor .. 80 2 10 Temperature ramp annealing of co evaporated glass/In/Se bilayer precursor ... 81 2 11 Temperature ramp annealing of co evaporated glass/In/Se bilayer precursor ... 82 2 12 TEM images of quenched bilayer precursor of glass/In/Se. ............................... 83 2 13 Temperature ramp annealing of MEE grown gl ass/In/Se bil ayer precursor ........ 84

PAGE 11

11 2 14 Temperature ramp annealing of co evaporated glass/In+Se comixed precursor ................................ ................................ ................................ ............ 85 2 15 TEM images of quenched bilayer precursor of glass/In+Se. .............................. 86 2 16 Temperature ramp annealing of co evaporated glass/Ga/Se bilayer precursor 87 2 17 TEM images of quenched bilayer precursor of glass/Ga/Se. .............................. 88 2 18 Temperature ramp annealing of MEE grown gla ss/Ga/Se bilayer precursor ...... 89 2 19 Temperature ramp annealing of co evaporated glass/Ga+Se comixed precursor ................................ ................................ ................................ ............ 90 2 20 Temperature ramp annealing of MEE grown gla ss/Ga+Se comixed precursor ................................ ................................ ................................ ........................... 91 2 21 TEM images of quenched bilayer precursor of glass/Ga+Se. ............................. 92 2 22 Isothermal annealing of glass/In/Se bilayer precursor grown by evaporation at 290 o C ................................ ................................ ................................ ............. 93 2 23 Isothermal annealing of glass/In+Se comixed precursor grown by evaporation at 360 o C ................................ ................................ ......................... 94 2 24 Isothermal annealing of g lass/Cu/Se bilayer precursor grown by evaporation at 240 o C ................................ ................................ ................................ ............. 95 2 25 Isothermal annealing of glass/Cu+Se comixed precursor grown by evaporation at 220 o C ................................ ................................ ......................... 96 2 26 Isothermal annealing of glass/Ga+Se comixed precursor grown by evaporation at 236 o C ................................ ................................ ......................... 97 2 27 Avrami plot for samples from isothermal selenization studies. ........................... 98 2 28 Parabolic plot from isothermal sele nization studies ................................ ............ 99 2 29 Arrhenius plot using rate constants obtained from isothermal ann ealing experiments. ................................ ................................ ................................ ..... 100 3 1 Room temperature diffraction data of as deposited glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se ................................ ................................ ............................ 121 3 2 Sequence of diffraction patterns during temperature ramp annealing of glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se ................................ ............................ 122 3 3 Se 2 partial pressure (atm) temperature diagram for the Cu Se system ............ 123

PAGE 12

12 3 4 High resolution room temperature diffraction patterns of temp erature ramp annealed precursor ................................ ................................ ........................... 124 3 5 Cross sectional and surface images of temperature ramp annealed precursor. ................................ ................................ ................................ ......... 125 3 6 High resolution room temperature XRD of as deposited glass/Mo/ In 2 Se 3 / Cu 2 Se/Se ................................ ................................ ................................ .......... 126 3 7 Sequence of diffraction patterns collected during temperature ramp annealing of glass/Mo/ In 2 Se 3 /Cu 2 Se/Se ................................ ....................... 127 3 8 Sequence of diffraction patterns collected during temperature ramp annealing of glass/Mo/ In 2 Se 3 /Cu 2 Se/Se with selenium overpressure ........... 128 3 9 High resol ution r oom temperature diffraction data of as deposited glass/Mo/ (In,Ga) 2 Se 3 /CuSe ................................ ................................ ............................. 129 3 10 Sequence of diffraction patterns collected during temperature ramp annealing of glass/Mo/ (In,Ga) 2 Se 3 /CuSe ................................ ....................... 130 3 11 Isothermal annealing d iffraction data of glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se precursor ................................ ................................ .......................... 131 3 12 Solid state gr owth model plot of glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se precursor ................................ ................................ ................................ ......... 132 3 13 Arrhenius plot using rate constants obtained from Avrami and parabolic model for glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se ................................ ........... 133 3 14 Cross sectional and surface image of rapid thermal annealed glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se precursor. ................................ ............................ 134 3 15 Is othermal annealing d iffraction data of glass/Mo/ In 2 Se 3 / Cu 2 Se/Se precursor ................................ ................................ ................................ .......... 135 3 16 Solid state growth model plot of glass/Mo/ In 2 Se 3 / Cu 2 Se/Se precursor ....... 136 3 17 Arrhenius plot using rate constants obtained from Avrami and parabolic model for glass/Mo/ In 2 Se 3 / Cu 2 Se/Se ................................ ......................... 137 3 18 Cross sectional and surface imag e of rapid thermal annealed glass/Mo/ In 2 Se 3 / Cu 2 Se/Se precursor. ................................ ................................ ........... 138 3 19 Isothermal annealing d iffraction data of glass/Mo/ (In,Ga) 2 Se 3 /CuSe precursor ................................ ................................ ................................ .......... 139

PAGE 13

13 3 20 Solid state growth model plot of glass/Mo/ (In,Ga) 2 Se 3 /CuSe precursor ....... 140 3 21 Arrhenius plot using rate constants obtained from Avrami and parabolic mod el for glass/Mo/ (In,Ga) 2 Se 3 /CuSe precursor ................................ ........... 141 4 1 Room temperature X ray diffraction plot of as deposited glass/Mo/CuIn/CuGa precursor ................................ ................................ ................................ .......... 157 4 2 Temperature ramp sequence of glass/Mo/CuIn/CuGa precursor in forming gas ................................ ................................ ................................ .................... 158 4 3 Room temperature X ray diffraction plot of temperature ramp annealed precursor of glass/Mo/Cu In/CuGa ................................ ................................ ... 159 4 4 SEM images of temperature ramp annealed precursors. ................................ 160 4 5 EDS profile and TEM image of temperature ramp annea led precursors. ......... 161 4 6 Room temperature X ray diffraction plot of as deposited glass/Mo/CuGa/CuIn precursor ................................ ................................ ................................ .......... 162 4 7 Temperature ramp sequence of glass/Mo/CuGa/CuIn precursor in forming gas ................................ ................................ ................................ .................... 1 63 4 8 Room temperature X ray diffraction plot of temperature ramp annealed precursor of glass/Mo/CuGa/CuIn ................................ ................................ .... 164 4 9 Room temperature X ray diffraction plot of as deposited glass/Mo/CuIn/CuGa/Se precursor ................................ ................................ ... 165 4 10 Temperature ramp selenization sequence of gl ass/Mo/CuIn/CuGa/Se precursor ................................ ................................ ................................ .......... 166 4 11 Room temperature X ray diffraction plot of temperature ramp annealed precursor of glass/Mo/CuIn/CuGa/Se precursor ................................ ............... 167 4 12 EDS profile and TEM image of temperature ramp annealed glass/Mo/CuIn/CuGa/Se precursor ................................ ................................ ... 168 4 13 S EM images of temperature ramp annealed precursors. ................................ 169 4 14 Room temperature X ray diffraction plot of as deposited glass/Mo/CuGa/CuIn/Se precursor ................................ ................................ ... 170 4 15 Temperature ramp selenization se quence of glass/Mo/CuGa/CuIn/Se precursor ................................ ................................ ................................ .......... 171 4 16 Room temperature X ray diffraction plot of temperature ramp annealed precursor of glass/Mo/CuGa/CuIn/Se precursor ................................ ............... 172

PAGE 14

14 4 17 EDS profile and TEM image of temperature ramp annealed glass/Mo/CuGa/CuIn/Se precursor ................................ ................................ ... 173 4 18 Temperature ramp selenization sequence of glass/Mo /CuIn/CuGa precursor 174 4 19 Room temperature X ray diffraction plot of temperature ramp selenized precursor of glass/Mo/CuIn/CuGa ................................ ................................ .... 175 4 20 EDS profile and TEM image of temperature ramp selenized glass/Mo/CuIn/CuGa precursor ................................ ................................ ........ 176 4 21 SEM images of temperature ramp annealed precursors. ................................ 177 4 22 Temperature ramp selenization sequence of glass/Mo/CuGa/CuIn precursor 178 4 23 EDS profile and TEM image of temperature ramp selenized glass/Mo/CuGa/Cu In precursor ................................ ................................ ........ 179 4 24 Room temperature X ray diffraction plot of temperature ramp selenized precursor of glass/Mo/CuGa/CuIn ................................ ................................ .... 180 4 25 Isothermal annealing of glass/Mo/CuIn/CuGa/Se precursor with selenium overpressure at different temperatures. ................................ ............................ 181 4 26 Isothermal annealing of glass/Mo/CuGa/CuIn/Se precursor with seleniu m overpressure at different temperatures. ................................ ............................ 182 4 27 Isothermal selenization of glass/Mo/CuIn/CuGa precursor with selenium vapor at different temperatures. ................................ ................................ ........ 183 4 28 Isothermal selenization of glass/Mo/CuGa/CuIn precursor with selenium vapor at different temperatures ................................ ................................ ......... 184 4 29 Arrhenius plot plotting rate constants obtaine d from isothermal studies ........... 185 5 1 Room temperature X ray diffraction pattern of as deposited precursor. .......... 200 5 2 SEM and TEM ima ge of as deposited precursor. ................................ ............. 201 5 3 Temperature ramp sequence during annealing of SS/Ru/Mo/Cu/Ga/In/Se with overpressure of selenium ................................ ................................ .......... 202 5 4 Room temperature X ray diffraction pattern of temperature ramp annealed precursor. ................................ ................................ ................................ ......... 203 5 5 S elected area diffraction pattern of temperature ramp annealed SS/Ru/Mo/Cu/In/Ga/Se pr ecursor ................................ ................................ .... 204

PAGE 15

15 5 6 SEM images of temperature ramp annealed precursors with selenium overpressure. ................................ ................................ ................................ .... 205 5 7 Auger depth profile of te mperature ramp annealed precursor of SS/Ru/Mo/Cu/Ga/In/Se. ................................ ................................ ................... 206 5 8 Temperature ramp sequence during annealing of SS/Ru/Mo/Cu/Ga/In/Se precursor without selenium overpressure. ................................ ........................ 207 5 9 Temperature ramp sequence during annealing of SS/Ru/Mo/Ga/In/Cu/Se with overpressure of selenium ................................ ................................ .......... 208 5 10 Electron diffraction pattern of quenched sample during temperature ramp annealing of SS/Ru/Mo/Ga/In/Cu/Se. ................................ ............................... 209 5 11 Auger depth profile of temperature ramp annealed precursor of SS/Ru/Mo/Ga/In/Cu/Se. ................................ ................................ ................... 210 5 12 Sequence of isothermal annealing of SS/Ru/Mo/Ga/In/Cu/Se at 400 C. ......... 211 5 13 Avrami plot using fractional conversion from isothermal experime nts at different isothermal temperature. ................................ ................................ ...... 212 5 14 Arrhenius plot using rate constants obtained from Avrami model. .................... 213 6 1 Room t emperature X ray diffraction of molybdenum substrate ........................ 222 6 2 Temperature ramp selenization sequence of molybdenum in nitrogen atmosphere ................................ ................................ ................................ ....... 223 6 3 Temperature ramp selenization sequence of molybdenum in forming gas atmosphere ................................ ................................ ................................ ....... 224 6 4 Room temperature X ray diffraction scan of temperature ramp selenized molybdenum in form ing gas atmosphere ................................ .......................... 225 6 5 Isothemal selenization sequence of molybdenum in forming gas atmosphere 226 6 6 Fractional conversion of molybdenum with time at different isothermal temperature ................................ ................................ ................................ ...... 227 6 7 Avrami plot using fractional conversion from isothermal studies ...................... 228 6 8 P arabolic plot using fractional conversion from isothermal studies ................... 229 6 9 Arrhenius plot using rate constants obtained from solid state growth models. 230 6 10 TEM image of temperature ramp selenized and pure molybdenum. ................ 231

PAGE 16

1 6 6 11 X ray photoelectron spectroscopy of pure molybdenum. ................................ .. 232 6 12 Ellingham diagram for Mo O, Mo Se and CIS systems. ................................ ... 233 6 13 Electron diffraction of tempe rature ramp annealed precursor ........................... 234 7 1 Room temperature X ray diffraction scan of as deposited repeated layers of Cu/Ga/In. ................................ ................................ ................................ ......... 245 7 2 SEM images of as deposited repeated layers of Cu /Ga/In. .............................. 246 7 3 Temperature ramp annealing of 8 repeated layers of Cu/Ga/In in forming gas 247 7 4 Temperature ramp annealing of 4 repeated layers of Cu/Ga/In in forming gas 248 7 5 Room temperature X ray diffraction scan of temperature ramp annealed precursor. ................................ ................................ ................................ ......... 249 7 6 SEM images of temperature ramp annealed precursors of Cu/Ga/In. .............. 250 7 7 Temperature ramp selenization sequence of 8 repeated layers of Cu/Ga/In .... 251 7 8 Temperature ramp selenization sequence of 4 repeated layers of Cu/Ga/In .... 252 7 9 SEM images of temperature ramp selenized precursors of Cu/Ga/In. .............. 253 7 10 Room temperature XRD of temperature ramp selenized precursors of Cu/Ga/In ................................ ................................ ................................ ........... 254 7 11 AES depth profile of temperature ramp selen ized precursors of Cu/Ga/In. ..... 255 7 12 Isothermal selenization sequence of 4 repeated layers of Cu/Ga/In at different temperatures. ................................ ................................ ..................... 256 7 13 Isothermal selenization sequence of 8 repeated layers of Cu/Ga/In at different temperatures ................................ ................................ ...................... 257 7 14 Rate constant estimation for 4 repeated layers of Cu/Ga/In using solid state gr owth models. ................................ ................................ ................................ 258 7 15 Rate constant estimation for 8 repeated layers of Cu/Ga/In using solid state growth models. ................................ ................................ ................................ 259 7 16 Arrhe nius plot using calculated rate constants from solid state growth model. 260 8 1 Room temperature X ray diffraction scans of as deposited CuGa/In precursor. ................................ ................................ ................................ ......... 270 8 2 SEM image of as deposited CuGa/In precursor. ................................ .............. 271

PAGE 17

17 8 3 Temperature ramp selenization sequence of CuGa/In precursor deposited on sodium free substrate ................................ ................................ ....................... 272 8 4 Temperature ramp selenization sequence of CuGa/In precursor deposited on MoNa 3 substrate ................................ ................................ ............................... 273 8 5 Temperature ramp selenization sequence of CuGa/In precursor deposited on MoNa 5 substrate ................................ ................................ ............................... 274 8 6 Room temperature X ray diffraction of temperature ramp selenized CuGa/In precursor. ................................ ................................ ................................ ......... 275 8 7 SIMS depth profile of temperature ramp selenized CuGa/In precursor. .......... 276 8 8 Ga/III profile for temperature ramp selenized precursor deposited on sodium doped m olybdenum substrates ................................ ................................ ......... 277 8 9 SEM images of temperature ramp selenized precursor. ................................ ... 278 8 10 Isothermal selenization sequence of Cu Ga/In precursor deposited on sodium free substrates at different temperature ................................ ............................ 279 8 11 Isothermal selenization sequence of CuGa/In precursor deposited on MoNa 3 at different temperature ................................ ................................ .................... 280 8 12 I sothermal selenization sequence of CuGa/In precursor deposited on MoNa 5 at different temperature ................................ ................................ .................... 281 8 13 Avrami plot using fr actional conversion data from isothermal selenization experiments. ................................ ................................ ................................ ..... 282 8 14 Arrhenius plot using rate constants obtained from Avrami model ..................... 283 9 1 Schematic of CIS nanoparticle synthesis ................................ ......................... 296 9 2 Room temperature X ray diffraction pattern of as synthesized nanoparticles .. 297 9 3 TEM images of as synthesized nanoparticles. ................................ .................. 298 9 4 Temperature ramp sequence of copper rich nanoparticle with selenium overpressure ................................ ................................ ................................ ..... 299 9 5 Phase transformation of CIS (cubic) to CIS (tetragonal) ................................ ... 300 9 6 SEM image showing grain growth in temperature ramp copper rich nanoparticle ................................ ................................ ................................ ...... 30 1 9 7 Room temperature XRD of as synthesized copper poor nanoparticle .............. 302

PAGE 18

18 9 8 Temperature ramp sequence of copper poor nanoparticle with seleni um overpressure ................................ ................................ ................................ ..... 303 9 9 Room temperature X ray diffraction of as synthesized nearly stoichiometric nanoparticle ................................ ................................ ................................ ...... 304 9 10 Tempera ture ramp sequence of nearly stoichiometric nanoparticle with selenium overpressure ................................ ................................ ..................... 305 9 11 Temperature calibration of selenization tube furnace ................................ ....... 306 9 12 SEM image of CIS nanoparticle deposited on glass. ................................ ........ 307 9 13 X ray diffraction pattern of annealed film obtained from CIS nanoparticle ........ 308 9 14 SEM image showing grain growth of annealed film obtained from CIS nanoparticle ................................ ................................ ................................ ...... 309 9 15 X ray diffraction pattern of CIS before and after KCN etching .......................... 310 9 16 X ray diffraction pattern of as synthesized (In,Ga) 2 Se 3 nanoparticles .............. 311 9 17 TEM image of as synthesized nanoparticles of (In,Ga) 2 Se 3 ............................. 312 9 18 X ray diffraction pattern of as synthesized Cu 2 Se nanoparticles ...................... 313 9 19 TEM image of as synthesized na noparticles of Cu 2 Se ................................ ..... 314 9 20 X ray diffraction pattern of CIGS film obtained from rapid thermal annealing of nanoparticles ................................ ................................ ................................ 315 9 21 J V curve of device fabricated from nanoparticle approach .............................. 316 10 1 X ray diffraction pattern of as received and temperature ramp annealed sample of SS/Mo/CIGS/CdS. ................................ ................................ ............ 322 10 2 Temperature ramp sequence during annealing of SS/Mo/CIGS/CdS sample in nitrogen environment ................................ ................................ .................... 323 10 3 Temperature ramp sequence during annealin g of SS/Mo/CIGS/CdS/ITO sample in nitrogen gas environment ................................ ................................ 324 10 4 Temperature ramp sequence during annealing of SS/Mo/CIGS/CdS/ITO sample in forming gas environment ................................ ................................ .. 325 10 5 SIMS depth profile of as received and quenched sample at 460 C. ................ 326 10 6 Isothermal annealing of SS/Mo/CIGS/CdS sample at different temperature s .. 327

PAGE 19

19 10 7 Avrami plot using fraction conversion from isothermal annealing experiments 328 10 8 Arrhenius plot using rate consta nts obtained from Avrami model ..................... 329 1 1 1 Ellingham diagram for In Se system ................................ ................................ 344 11 2 Ellingham diagram for Ga Se system ................................ ............................... 345 11 3 Ellingham diagram for CIGS and MoSe 2 system ................................ .............. 346 11 4 Vapor pressure of selenium species over liquid selenium as a function of te mperature ................................ ................................ ................................ ...... 347 11 5 Molecular flux of selenium species from a evaporating liquid source ............... 348 11 6 Ellingham diagram for oxides of Cu Ga In O system ................................ ....... 349 11 7 Selenization reactor for roll to roll process ................................ ........................ 350 11 8 Hexahedral mesh design for selenization reacto r ................................ ............. 351 11 9 Calculated temperature contours of substrate for different inlet velocity. ........ 352 11 10 Calculated velocity contours of nitrogen over substrate for different inlet velocity. ................................ ................................ ................................ ............ 353 11 11 Temperature contours for counter current movement of substrate at a different gas velocity. ................................ ................................ ........................ 354 11 12 Temperature contours for co current movement of substrate at a different gas velocity. ................................ ................................ ................................ ...... 355 11 13 Velocity contours over substrate for counter current mov ement of substrate at a different gas velocity. ................................ ................................ ................. 356 11 14 Velocity contours over substrate for co current movement of substr ate at a different gas velocity ................................ ................................ ......................... 357 11 15 Temperature contours with radiation incorporated for counter current movement of substrate at a different gas velocity. ................................ ............ 358 11 16 Temperature contours wit h radiation incorporated for co current movement of substrate at a different gas velocity. ................................ ................................ 359 11 17 Velocity contours with radiation incorporated for counter current movement of substrate at a different gas velocity. ................................ ................................ 360 1 1 18 Velocity contours with radiation incorporated for co current movement of substrate at a different gas velocity. ................................ ................................ 361

PAGE 20

20 11 19 Temperature contours for counter current movement of substrate at a total gas velocity ................................ ................................ ................................ ....... 362 11 20 Temperature contours for co current movement of substrate at a total gas velocity ................................ ................................ ................................ ............. 363 11 21 Velocity contours for counter current movement of substrate at a total gas velocity ................................ ................................ ................................ ............. 364 11 22 Velo city contours for co current movement of substrate at a total gas velocity 365 11 23 Concentration contours for counter current movement of substrate at a total gas velocity ................................ ................................ ................................ ....... 366 11 24 Concentration contours for co current movement of substrate at a total gas velocity ................................ ................................ ................................ ............. 367 12 1 Experimental points used for optimization of Cu 2 Se Ga 2 Se 3 psedudobinary section. ................................ ................................ ................................ ............. 379 12 2 Experimental data points used for optimization of Cu 2 Se Ga 2 Se 3 pseudobinary section. ................................ ................................ ....................... 380 12 3 Experimental data points used for optimization of CIS CGS pseudobinary section. ................................ ................................ ................................ ............. 381 12 4 Comparison of experimental data and as predicted by the model for Cu 2 Se Ga 2 Se 3 system ................................ ................................ ................................ 382 12 5 Comparison of experimental data and as predicted by the model for CIS CGS pseudobinary system ................................ ................................ ............... 383

PAGE 21

21 LIST OF ABBREVIATION S AES Auger electron spectroscop y CALPHAD Calculation of phase diagram ED Electron diffraction EDS Energy dispersive spectroscopy FF Fill factor ICP OES Inductively coupled plasma optical emission spectroscopy ISET International solar electric technology LCOE Levelized cost of electrici ty MEE Migration enhanced epitaxy NREL National renewable energy laboratory NSTR National solar technology roadmap PV Photovoltaics RTA Rapid thermal annealing SAED Selective area electron diffraction SEM Scanning electron spectroscopy SIMS Secondary ion mass spectroscopy TEM Transmission electron spectroscopy TMI Trimethyl indium XPS X ray photoelectron spectroscopy XRD X ray diffraction

PAGE 22

22 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fu lfillment of the Requirements for the Degree of Doctor of Philosophy RAPID ROUTES FOR SYNTHESIS OF CIGS ABSORBERS By Rangarajan Krishnan August 2012 Chair: Timothy Anderson Major: Chemical Engineering The chalcopyrite solid solution Cu(In x Ga 1 x )Se 2 (CI GS) is a commercially emerging thin film absorber material based on the promise of low manufacturing costs and high conversion efficiency (champion cell now exceeds 20%). The primary challenge in achieving low processing cost is increasing the synthesis r ate of CIGS at lower temperature. Recognizing this challenge the national solar technology roadmap calls for decreasing the absorber synthesis time to 2 min by 2015 while retaining the efficiency. However, the synthesis of CIGS is not limited by rate of deposition of elements, but is rather limited by rate of reaction of starting precursors that form CIGS. Reaction pathways and kinetics were followed for understanding CIS/CIGS formation both qualitatively and quantitatively using time resolved X ray diffr action for two common processes such as co evaporation and metal selenization. First, in situ X ray diffraction was used to understand metal selenium binary system and the results were consistent with the sequence predicted by thermodynamic phase diagram. Then, the reaction pathways and kinetics of ternary CIS formation was followed using bilayer precursor structure of glass/Mo/ In 2 Se 3 /CuSe y and glass/Mo/ In 2 Se 3 / Cu 2 Se. Variation in kinetics with addition of gallium to the group III sub lattice was also

PAGE 23

23 developed. In situ selenization pathways and kinetics were studied using X ray diffraction for various metal precursor stru ctures (bilayer metal precursors, stacked elemental layers, multilayer structures). The gallium distribution after annealing was found to be dependent on the precursor structure. Also the effect of nucleation in kinetics resulting with change in the order of deposition of metals was also studied. The effect of sodium in pathways and kinetics of absorber formation was studied by using sodium doped molybdenum substrates. Furthermore, reaction pathways and kinetics of MoSe 2 formation was studied by performing in situ selenization of molybdenum substrates. It was found that oxygen presence leads to formation of MoO 2 along with the formation of MoSe 2 The device degradation mechanism due to cadmium diffusion in to CIGS was studied for a completed device without top contacts for industrial sample using in situ X ray diffraction. Cadmium diffusi on initiated for temperatures > 400 C forming CuCd 2 (In, Ga)Se 4 compound resulting in complete failure of the device. A new process for high rate synthesis of CIGS has been developed using nanoparticle based approach involving peritectic decomposition of compounds to yield a liquid assist ed growth of CIGS. The nanoparticles involved in this study were synthesized by simple reaction in alcoholic medium. Another process for high rate synthesis of CIGS has been developed using binary nanoparticles based approach emulating the co evaporation process. The binary nanoparticles for this study were synthesized by colloidal route using coordinating solvents. An efficiency of 1.67% has been obtained for the same process.

PAGE 24

24 Thermodynamic assessment of the CIS CGS and Cu 2 Se Ga 2 Se 3 pseu dobinary was pe rformed to predict the phase diagrams. Sub lattice models were used to express Gibbs energy as a function of temperature. The model developed agrees well with the experimental data reported in the literature.

PAGE 25

25 CHAPTER 1 INTRODUCTION Photovoltaics Solar C ells are electronic devices based on p n or p i n structures used for generating direct current through the photovoltaic effect. The photovoltaic effect was first observed by Alexander Edmond Bequerel in 1839 and is the process of converting photons to el ectricity [1] The first silicon solar cell was produced by Bell Laborator ies in 1954 using a p n junction. Other semiconductor materials such as CdTe, GaAs, InP, and CIGS were developed subsequently. Properties of the material change depending on the group and researchers take advantage of these differences per the applicatio n. The performance of the solar cell is measured in terms of efficiency of output power to incident radiation. The Shockley Queisser limit for a single junction solar cell is 33% [2] An efficiency of 28.2 % has been obtained for GaAs based devices [3] Table 1 1 compares the efficiency obtaine d using different technologies. Depending on the application, low or high efficiency modules can be used for different purposes. The important factor to be considered during installation of solar panels is the levelized cost of electricity (LCOE) which v aries with different technologies. LCOE takes into account module costs and balance of the system costs. An important goal of the research community is to improve solar cell efficiencies while decreasing the balance of system costs. Working of Solar Cell The solar radiation reaching the earth is similar to black body radiation at ~5800 K. atmospheric effects such as absorption and scattering at selective wavelengths. Figure

PAGE 26

26 1 1 shows the standard solar spectra used in photovoltaic (PV) applications. AM0 (air mass index) represents the solar spectra outside the atmosphere and AM1.5 represents the spectra at sea level. AM1.5 spectrum is accepted as a reference spectrum in PV applications. For the photovoltaic device to work, the first step involves the absorption of light. The absorption coefficient of the semiconductor material determines how deep the photons can travel in the material before it loses its energy. Ma terials with low absorption coefficients absorb photons poorly and thin materials with high absorption coefficients will be transparent to the corresponding wavelength. The key factor that determines the absorption of photons is the band gap of the materi al. If the energy of the photon is less than the band gap of the semiconductor material, it interacts weakly with the semiconductor and just passes through the material. When the photon energy is greater or equal to the band gap of the semiconductor, an electron hole pair is generated and electrons are excited to the conduction band. For the photovoltaic effect, the electron hole pairs needs to be separated and collected at external contacts. This requires an internal electric field which is obtained by the formation of a heterojunction structure (p n junction). At thermodynamic equilibrium, minority carriers reach the edge of the depletion region by diffusion after which they are immediately swept by the internal electric field which yields a current f low. The solar cell is characterized by three different parameters; open circuit voltage (V oc ), short circuit current density (J sc ), and the fill factor (FF). The maximum voltage available from a solar cell is the open circuit voltage and this happens at zero current conditions. The short circuit current is the maximum current from the solar cell at zero voltage. The fill factor measures the quality of the solar cell. Figure 1 2 shows the IV curve of a solar cell along with the

PAGE 27

27 parameters. The efficien cy of the solar cell is calculated using these parameters using the following relation, (1 1) where P in is a known quantity for a given air mass index. The major loss mechanisms in the solar cell are reflections from the surfa ce, electron hole recombination due to the crystal defects, and series and shunt resistance losses These losses affect the efficiency of the solar cell which depends on the structural and electrical properties of the grown solar cell Classification of S olar Cells Commercial solar cells are made of either monocrystalline or polycrystalline silicon. Most of the knowledge of silicon photovoltaics has been derived from the microelectronics industry in terms of material properties and manufacturing technique s. Moreover, electronic grade silicon has been available at a much lower price [4] Due to the indirect band gap of sil icon, it is not an ideal choice for the absorber. Semiconductor materials with indirect band gaps do not absorb solar radiation efficiently compared to direct band gap materials and hence thick layers are required to absorb the entire solar spectrum. For instance, ~100 m of silicon absorbs the same amount of light as ~1 m of GaAs. Owing to large thicknesses, solar cells produced should be of high quality in terms of carrier lifetimes and diffusion lengths, to minimize the recombination of the photogenerated carriers. These stringent conditions increase the manufacturing cost of the solar cells. The high manufacturing cost of silicon cells are compensated by their high efficiencies. Other III V materials such as In x Ga 1 x As, In x Ga 1 x P and GaAs are also dominated by hi gh costs [5 8]

PAGE 28

28 In order to meet the cost requirements, other thin film absorber materials such as cadmium telluride (CdTe), copper indium diselenide(CIS) and its alloys with gallium and amorphous silicon (a: Si H) h ave been extensively studied [9 11] Thin film solar cells have many advantages over crystalline silicon because of their very high absorption coefficient and direct band gap. The thickness of the absorber materia l are a few microns and they can be deposited by a variety of vacuum and non vacuum techniques on both flexible (e.g. stainless steel and polymers) and rigid substrates (glass). A serious problem with amorphous silicon is its tendency to undergo light ind uced degradation which decreases the efficiency from its initial value, resulting from the dangling bonds which act as recombination centers. The stabilized efficiency of amorphous silicon is about 13% [12] The polycrystalline compound semiconductors such as CdTe or CIGS do not undergo light induced degradation. The performance of CIGS based cells has shown improvement after light illumination under normal operating conditions. Record efficiencies of CdTe have reached 17.2% [13] and for CIGS, 20.3% efficiency has been obtained [14] CIGS Solar Cells Figure 1 3 shows a schematic representation of a CIGS solar cell. CIGS solar cells are generally grown in a substrate configuration and cell fabrication starts with deposition of the back electrode on substrate. The back electrode is molybdenum deposited by sputtering. Molybdenum (Mo) is grown in a two step fashion to optimize the adhesion and the resistivity. This is followed by deposition of the p type absorber layer, n type CdS buffer layer undoped ZnO highly conductive aluminum doped zinc oxide( Al:ZnO) and metal grids for minority carrier collection. The final device is encapsulated to protect it against moisture from the surroundings. The band gap of

PAGE 29

29 CuInSe2 is 1.04 eV and the band gap can be increased with the addition of gallium (Ga) in the group III sub lattice, or with the addition of sulfur(S) in the group VI sub lattice or with addition of silver (Ag) in the group I sub lattice. This flexibility allows control of the band gap of CIS from 1.04 eV to 1.67 eV with the addition of gallium (CuGaSe2), and 1.53 2.5 eV with the addition of sulfur (CuInS 2 ) ( CuGaS 2 ) [10] Table 1.2 compares the lattice constants and band gaps of the (Cu, Ag) In x Ga 1 x Se y S 2 y alloy system The structure of a CIGS device is quite stable, as all detrimental interfacial reactions are kinetically limited at room temperatures [15] It has been reported that moderate interdiffusion of CdS and CIGS occurs, which improves th e performance of the cell [16] The reaction of CIGS and ZnO has also been reported where Cu poor CIGS reacts with ZnO forming ZnSe and oxides of indium and gallium. Other c admium free buffer layers are also developed which makes the device toxic free. The other interfacial reaction that is observed is the formation of MoSe 2 It is because of this layer that the contact between CIGS and Mo is ohmic [17] Without MoSe 2 the contact becomes schottky and this decreases the efficiency of the device [18] The work function of CIGS and Mo 2 the value is 4.96 eV which makes it ohmic. The band gap of MoSe 2 is 1.4 eV [19] wider than CIGS and this reduc es recombination at the back contact. The other advantage of a MoSe 2 interface layer is adhesion of CIGS to the back contact [17] Crystal s tructure of CIGS It h as been reported that stoichiometric CIGS exists in different crystal structures [20] The crystal structure of CIGS is derived from a zinc blende structure common to many II VI semiconductors such as ZnSe [ 21] In ZnSe, each selenium atom is bonded to four zinc atoms and in CIGS each divalent zinc atom is replaced with monovalent

PAGE 30

30 copper and trivalent indium or gallium such that each selenium is bonded to two copper and two group III atoms. The bond distan ces are different which leads to a tetragonal distortion (space group I 42d) and hence the chalcopyrite crystal strucuture. When the copper and group III atoms are distributed randomly (cation disorder), this leads to a cubic structure, also termed as a s phalerite structure. Figure 1 4 shows the chalcopyrite structure of CIGS. Also, metastable structures with CuPt ordering have been reported in the literature [20] A bsorber composition and its e ffect on p erformance Although the add ition of gallium, silver, or sulfur improves the properties of solar cells, the most important factor that determines the quality of solar cells are the Cu/III and Ga/III ratios. For high efficiency CuIn x Ga 1 x Se 2 solar cells, the overall composition of th e absorber should be copper deficient and perhaps more copper deficient at the surface. The composition of the surface layer corresponds to CIGS also known as ordered vacancy compound (OVC) [22] The OVC layer is n type [23, 24] and the bulk of th e CIGS is p type, favoring homojunction formation. It is reported that the inverted surface minimizes the recombination at CIGS/CdS interface [25] The thickness of the OVC layer varies from 5 to 60 nm and the thickness of this layer depends on the copper content in the film. The band gap of the surface layer is wider than the bulk with a reported value of 1.23 and 1.3 eV [25, 26] The wider band gap of the surface layer increases the barrier for recombination and is essential for high efficiency devices. However, too thick of an OVC layer decreases the cell performance because of an increase in series resistance owing to its low conductivity [27] It has also been reported that the OVC layer forms only for co evaporated absorbers [28]

PAGE 31

31 It is expected that increasing the band gap of the absorber would increase the conversion efficiency of the device. This has been found to be untrue, as efficiency of absorbers based on CuInS 2 and CuGaSe 2 are lower than those achieved by CuIn x Ga 1 x Se 2 2 CuGaSe 2 and CuInS 2 are 14.5% [29] 10.2%, [30, 31] and 11.1 %, [32] respectively. The record efficiency (20.3%) of absorber and the resulting band gap is 1.12 eV [14] CuInSe 2 can be grown as either p type or n type depending on the Cu/III ratio, whereas CuGaSe 2 is always grown as p type preventing the formation of an OVC layer, and consequently decreasing the efficiency as discussed above For Cu poor material, the open circuit voltage (V oc ) varies inversely with defect densities of the absorbers [33] With the addition of gallium to CuInSe 2 ,open circuit voltage increases lin early with the band gap and beyond 30% gallium content, the increase of V oc slows down accompanied by an increase in defect density [33 35] The gallium content in the absorber is usually graded in such a way that more gallium is found near the rear contact than those near the surface [36] This helps enhance the separation of photogenerated carriers and reduces recombination at the back contact [37] A higher band gap at the surface can be achieved intentionally by increasing the Ga/In ratio towards the front contact or by sul furization of the surface [38] Generally, open circuit voltage increases and short circuit current decreases with an increase in the band gap; thus band gap grading is critical in optimizing cell performance. Defect c hemistry The most important factors that contribute to ele ctrical and chemical stability of CIGS solar cells are wide single phase domains and doping levels which are non

PAGE 32

32 degenerate over a wide composition range. These are the products of self compensation of the chalcopyrite compounds, that is, defects that are formed by deviations from stoichiometry are compensated by new defects to maintain the charge neutrality condition. Most of the defects remain inactive with respect to carrier recombination [15] An energetically favored isolated point defect has been reported [39] and it is the copper vacancy (V cu ) acting as a shallow acceptor, that contributes to the p type doping of CIGS. Indium on a copper site (In Cu ) is an antisite defect that would be a deep donor (2V Cu +In cu ) and is a favorable defect complex that prevents degenerate doping in an indium rich material. The OVC layer existence can be expla ined as repeating units of (2V Cu +In cu ). With the addition of gallium, the defect complex pair (2V Cu +Ga cu ) is less stable than the corresponding complex in CIS [40] Thus, formation of the OVC layer is more difficult in CGS than it is in CI S or CIGS with low gallium content, which is a reason for the low efficiency of a CGS based system. Other defects such as V Se Cu In Cu Ga exists in the system, however there is no experimental results that allows for the determination of the electrical pr operties of these defects. Role of i mpurities in CIGS The effect of sodium in CIGS was discovered in 1993, when high efficiencies were obtained for the material deposited on soda lime glass compared with borosilicate glass [41, 42] X ray photoelectron spectroscopy (XPS) and secondary ion mass spectroscopy (SIMS) showed the presence of sodium (Na) in the bulk and on the surface of the absorber [41] The presence of sodium du ring growth has been reported to improve the surface morphology [43] and increase the grain size of the absorber [43, 44] An increase in grain size is not always true as it depends on the precursor

PAGE 33

33 structure and the sodium concentration [45] An increase in carrier concentration leading to higher p type conductivity and enhanced crystallinity has also been reported for films prepared in the presence of Na [46 49] The increase in the p type conductivit y of Na containing films is because of the repression of donor type defects such as In Cu that act as majority carrier traps [47, 48, 50, 51] Sodium occupies the Cu sites owing to similar ionic radii, hence elimina ting the In Cu defect which results in increase in hole concentration [47, 51] Thus, removal of In Cu defects leads to a more ordered structure and hence enhanced ( 112 ) orientation. Lundberg et al. have reported th e suppression of indium and gallium with a Na presence which helps in obtaining a gallium gradient [52] The beneficial MoSe 2 formation between Mo and CIGS is aided in the presence of Na [19] The ideal amount of Na content for CIS and CIGS films is between 0.05 to 0.5 at %. Braunger et al. proposed a model for diffusion of sodium to the CIGS surface along grain boundaries which reacts with selenium to form sodium polyselenides (Na 2 Se x x=1, 2, 3, 4, 6) [53 ] At low selenium pressure, Na 2 Se formation dominates and the release of selenium is highly unlikely from this stable compound. During the absorber synthesis, sodium polyselenide formation dominates and they act as a selenium source during growth. In most cases, sodium diffuses from the soda lime process as well as on the properties of the Mo back contact and the glass itself [47, 49, 54, 55] In order to have better control and reproducibility over the sodium content, sodium can be supplied externally using sodium sources such as NaF [43, 45] Na 2 S [50, 5 5] Na 2 Se [55] Na x O [56] or molybdenum doped sodium [57] An external supply of Na through NaF helps in decreasing the absorber synthesis temperature with no loss in

PAGE 34

34 efficiency [58] The effect of other alkali fluorides such a s LiF, KF, and CsF has also been studied, but the impact of these salts was minimal when compared with NaF [47] The reason is because the ionic radius of Na helps in substitutional incorporation in the copper latti ce. Thus, NaF has the highest influence on film properties. Sodium also enhances the role of oxygen in CIGS based system [59 61] It was concluded in a recent study that oxygen is needed for diffusion of sodium f rom soda lime glass substrates [62] Sodium diffusion was suppressed at 10 8 Torr vacuum, whereas diffusion occurred in 10 5 Torr of air, oxygen or water. It is believed that oxygen passivates positively charged selenium vacancies (V Se ) that are present a t the surfaces and grain boundaries. Passivation of selenium vacancies is necessary as it decreases the effective p type doping of the film. Additionally, V Se serves as recombination centers for minority carriers. This is one of the reasons why air anne aling improves the efficiencies of the solar cell. Absorber Growth Techniques A wide range of deposition techniques exist for CIS based solar cells. The deposition method has a large impact on the resulting film properties as well as the manufacturing cos ts. Regardless of the deposition technique, the stoichiometry of the finally grown absorber is always copper poor and no additional phases such as Cu 2 x Se are allowed in the films. Cu 2 x Se is highly conductive as it is a degenerate semiconductor and caus es very high bucking currents. Before fabricating devices, the absorber layer is wet etched with CN ions to ensure removal of the Cu 2 x Se phase. The formation of photovoltaic quality films require high temperatures (>500 C) during film growth or post deposition annealing. The formation of films with a higher gallium concentration requires higher temperatures (>550 600 C) or longer reaction times than

PAGE 35

35 CIS. The higher processing temperature results in selenium loss because of its high vapor pressure a nd that should be compensated by maintaining a selenium overpressure environment. Co evaporation from Elemental Sources High efficiency small area absorbers (20.3%) were obtained using a three stage co evaporation process developed by National Renewable En ergy Laboratory (NREL). The process involves evaporating pure elements in the presence of selenium vapor. The Se/metal ratio is very important and this affects the grain size and morphology of the grown absorber. NREL 3 stage process is a variant of the 2 stage process developed by Boeing [63] which involves deposition of Cu rich CIGS layer at lower substrate temperature (450 C), followed by deposition of In rich layer at a higher temperature ( 550 C). The layers intermix at higher temperature and the overall compos ition is copper deficient. NREL modified the process t o 3 stages as shown in Figure 1 5. In the first stage (In, Ga) 2 Se 3 C). Depending on the vapor pressure of selenium, other compounds of the In Ga Se system s uch as, (In, Ga) Se, (In, Ga) 4 Se 3 can be formed on the substrate. The latter compound has not been reported in the literature, but with no gallium present, In 4 Se 3 can be formed. In the second stage, the temperature is increased to 600 C to evaporate Cu a nd Se yielding Cu rich CIGS. At (that) the processing temperature, there exists a eutectic point at 523 C [64] in which the Cu 2 Se is in equilibrium with liquid. The rate of formation of CIGS is higher because of increased mobility of species in the liquid phase and the high grains of CIGS are achieved by a vapor liquid solid growth mechanism [65] In the third stage, more (In, Ga) 2 Se 3 is added to make the

PAGE 36

36 overall compos ition copper poor and the substrate is cooled down in a selenium atmosphere. The champion cell process takes 60 minutes to deposit the absorber layer. Although the basic process remains the same, researchers differentiate the co evaporation process by va rying the substrate temperature [66] deposition time [54] gallium grading [52, 67] and sodium supply [68] Decreasing the deposition time to 4 minutes resulted in an absorber efficiency of 12.3 % which was attributed to increased recombination due to a smaller grain size. Also, single stage evaporation was carried out by different groups [69] and a highest effic iency of 16% was obtained. In co evaporation, the orientation of CIGS depends strongly on the ori entation of the underlying (In, Ga) 2 Se 3 layer [70] A higher flux of Se/(In+Ga) in the first stage resulted in an increased 220/204 orientation of CIGS When the Mo w as oriented in ( 110 ) the CIGS fi lm was found to be oriented in ( 220/204 ) direction, in the absence of Na [45] With a high concentration of Na, an increasing ( 112 ) orientation was observed and with the intermediate Na ( 220/204 ) orientation do minating. Most of the cha mpion cell processes have more ( 220/204 ) orientation and with this orientation, Cd 2+ ions diffusion takes place easily becaus e of fewer copper atoms on the (220/204) surface as compared to the (112) surfaces. Though high quality m aterial can be grown by the co evaporation process, problems still exist in terms of scaling the process. This is because of stern control requirements of the evaporation fluxes to achieve the desired film properties which are quite difficult for large ar ea substrates. In addition to that, this technology is dominated by a very high capital cost and incomplete utilization of the expensive source materials, such as indium and gallium.

PAGE 37

37 Selenization of Metallic Precursors Another absorber formation process i s the selenization process originally developed by ARCO Solar in 1981 [71] The process is divided in two steps: the deposition of the metal precursor layer at room temperature, followed by selenization at high temperature. Prior to selenization, metal l ayers are heated at lower temperatures to facilitate interdiffusion of metal precursors and alloy formation [67, 72 75] The metal layers can be deposited by a variety of processes such as sputtering [76 78] evaporation [74, 79 82] and electrodeposition [75, 83 85] Selenization is carried out in selenium containing atmospheres and typical te mperatures are greater than 500 C. Selenium can be in the form of pure elemental selenium or hydrogen selenide (H 2 Se) diluted with argon (Ar). Siemens Solar changed the metal precursor structure by alloying gallium with copper and then deposited a CuGa/In structure followed by rapid thermal annealing with selenium deposited by vacuum evaporation [86] Later, the selenization process was modified by using toxic H 2 Se followed by sulfurization, using dilute H 2 S gas to create a wide band gap at th e surface. Selenization time depends on thickness, structure, reaction temperature, the selenium source, and the composition of the film. Another approach is the deposition of modulated structures instead of a bilayer and the final film ends up with smoo ther surfaces and higher crystallinity upon annealing [79, 81, 87, 88] It has been reported that the rate of formation of CIS is greater than CGS [89, 90] As a result, depending on the precursor structure, two different phases of CIS and CGS may be present, if annealing temperature or annealing time is too little. Also, higher reaction temperatures facilitate formation of too much MoSe 2 which is detrimental to device performance [19, 81] It has been reported that the H 2 Se method of selenization is efficient for temperatures below 500 C and the best compositional

PAGE 38

38 uniformity with large grain size is obtained [81] MoSe 2 formation was found only when selenization is carried at 600 C using H 2 Se. The best module efficiency of 16 % was reported by Showa Shell using combined selenization and the surface sulfurization process. The alternative multi step approach to the synthesis of the absorber offers many advantages such as composit ional uniformity over large area substrates and high throughput compared with co evaporation process. Moreover, the metal selenization process is cost effective because of its high throughput, efficiency and low temperature. Energy pay back for CIS modul es is roughly about 2 years using the selenization of metals process [91 ] Nanoparticle Approach In order to decrease the cost, International Solar Electric Technology (ISET) employed a non vacuum process for mass production of CIGS solar cells [92, 93] The process uses a water base d precursor ink consisting of oxides of copper, indium and gallium. The ink is coated onto the substrate coated with molybdneum using the doctor blade technique. The oxide layer which is approximately ~3 m are reduced in forming gas in the temperature range of 475 to 525 C to obtain ~0.6 m of Cu Ga In. Finally the alloy is selenized using toxic H 2 Se gas at temperatures less than 500 C to yield CIGS. Solar cells with an efficiency of ~13 % have been re ported by this process. Though the efficiencies are low compared with the champion cell, advantages such as ease of scale up, high material utilization and low cost makes the process competitive with other technologies.

PAGE 39

39 The other nanoparticle process deve loped by Nanosolar involves the synthesis of metal nanoparticles coated on aluminum foil with molybdenum as a back contact [94] The metal nanoparticles are synthesized by top down and bottom up approach. The shape and size of the particle is not yet reported. The metals deposited are then oxidized using selenium vapor by rapid thermal annealing. Since the particles are in nanometer dimensions, the dif fusion distances are reduced and the rate of formation of CIGS is only kinetically limited. Small area efficiency of ~17.3 % has been reported by this process [95] Electrodeposition Electrochemical deposition for the synthesis of CIGS includes sequential deposition of elements in the form of a stacked layer and sequential deposition of binary compounds followed by annealing in reactive or inert atmospheres. The stacked elemental layer approach is similar to selenization of metallic precursors as discussed in previous sections. The electrodeposited films are usually amorphous or poorly crystalline in the as deposited state. Electrodeposition of CIS films are usually carried out under indium excess conditions. Since the standard electrode potentials are not the same for all the elements, preferential deposition of a single element takes place and the resulting films are copp er rich [96] Solopower uses the electrodeposition approach and a module efficiency of greater than 12% has been obtained. Scale up is still an issue and a roll to roll process is being e mployed to decrease the manufacturing cost. Other Deposition Techniques Absorber formation has also been achieved by evaporating or sputtering binary [97] ternary [98] and even quaternary compounds [99] This approach is relatively simple and easier to control than co eva poration of pure elements. A selenium rich

PAGE 40

40 environment is required during deposition to accommodate selenium loss due to volatilization. Shi et al have reported sputtering CIGS from a quaternary target and efficiency of 7.95% has been reported [100] CIS thin films have been de posited by atmospheric pressure metal organic chemical vapor deposition using Cu (hfac) 2 triethylindium (TMI) and H 2 Se as copper, indium and selenium sources, respectively. The deposition was carried out at 400 C and film had a preferred orientation in ( 112 ) direction. Solar cells fabricated by this technique had a open circuit voltage of 260 mV due to low shunt and increased series resistance [101] Though the process has the advantages of high throughput, it is still not scaled up attributed to the low efficiency of the devices. Optimization studies are being carried out to improv e the efficiency of the device. Spray pyrolysis is another technique in which salts of copper, indium, and gallium are dissolved in alcoholic solvents and selenium in the form of dimethylselenourea are used to deposit CIGS films. Films were deposited at 4 00 C, were group III rich, and the solar cell efficiencies reached 4 5 %. Reducing the temperature of deposition led to copper rich films along with contamination of carbon and chlorine impurities [102] Problem Statement CIGS based cells are a promising candidate for thin film solar cell applicat ions because of their high efficiency, excellent radiation hardness, and the potential use in a CIGS based tandem arrangement. Additionally, the high absorption coefficient of CIGS allows for thinner layers, thus decreasing materials cost and increasing t hroughput. The degrees of freedom offered by CIGS based cells are not realizable in other thin film technologies.

PAGE 41

41 Most of the industries utilizing CIGS based technology are differentiated on the basis of deposition of the absorber material. The manufa cturing cost variations between different technologies are related to the deposition of the absorber structure. The costs are further differentiated into materials cost and processing cost. The material cost depends on the absorber thickness and material utilization. For instance, the co evaporation process is dominated by wall deposition, thus increasing the cost of production. The capital cost is dominated by the complexity of the process and its scale. Again, complex processes such as co evaporation is dominated by higher capital costs because of the rigorous design to attain uniformity in large scale manufacturing. A reduction in the price of CIGS modules is required for broad insertion in the market. Currently, silicon dominates 80% of the market share and the remaining market is dominated by CdTe and amorphous silicon based technologies. CIGS based industries are still optimizing their process to improve the efficiencies. Since the CIGS champion cell efficiency (20.3%) exceeds that of other thi n film technologies (17.2% for CdTe, 12% for a Si: H), it also has the potential to achieve the highest module efficiency. Generally, when the module efficiency reaches 80% of the champion cell efficiency, the cost advantage of CIGS is highly competitive with silicon and other thin film technologies. Process economics suggest that cost can be decreased by increasing the process synthesis scale and decreasing th e manufacturing cost. Figure 1 6 shows that the levelized cost of electricity from CIGS solar c ells from a co evaporation process has the potential to be cost competitive with non renewable resources. The cost analysis of the nanoparticle based approach shows that the manufacturing cost decreases by 14% for every 1% increase in absolute efficiency f rom the initial value for

PAGE 42

42 a 5 0 MW plant as shown in Figure 1 7. The same benefits are achieved when the throughput of the product is increased, due to a decrease in the capital expenditure for an efficie ncy of 10% as shown in Figure 1 8. It is well know n that the rate of formation of CIGS does not depend on the rate of deposition of materials; rather, it is limited by diffusion controlled kinetics. The route to synthesize CIGS absorber is critical in achieving high efficiency and improving process scale s. Although the time required for synthesizing the commercial absorber is not yet published, these times are the bottleneck for reducing the cost of production. The national solar technology roadmap (NSTR) for CIGS based technology specifically calls for increasing the deposition rate to 30 40 m/hr for an absorber thickness less than a micron. This translates into a reduction of absorber synthesis time from ~15 to 36 min to ~1.5 to 2 minutes. It is realized that the deposition rate can be increased, bu t the kinetics of the synthesis process limits the rate. Scaling production capacity can be either achieved by building more identical process lines or increasing the capacity of the turnkey line. The latter approach has advantages in cost reduction, but scaling laws are not obvious. In particular, little is known about the thermochemistry and reaction pathways in the system of CIGS. The problems with CIGS based technology are the lack of predictive models that describe the formation of CIGS under diffe rent processing conditions and the lack of accelerated testing models that define the lifetime of the modules. It is thus necessary to develop qualitative and quantitative models that describe the formation of CIGS under processing conditions. Identifyin g precursor structures that have the highest rate will be a key step in increasing the synthesis rate. As discussed in the growth methods for absorber materials, a variety of processing

PAGE 43

43 techniques are available for synthesis (e.g. using stacked elemental layer, co deposition of elements, particle selenization). Most of the knowledge on understanding reaction pathways during CIGS synthesis is derived from ex situ studies. The use of ex situ studies create uncertainties with continued reactions during the cooling step, as well as the effects of exposure to the atmosphere. In this thesis, reaction pathways of CIGS formation are followed as a function of temperature using in situ high temperature X ray diffraction (HT XRD). Observing phase transformations du ring the reaction helps in understanding reaction pathways both qualitatively and quantitatively. To get the quantitative data, reaction pathways are followed as a function of time and rate parameters, such as activation energy and rate constant, are obta ined using solid state growth models. In Chapter 2, the reaction pathways and kinetics of binary metal selenides are investigated using in situ high temperature X ray diffraction and compared with the prediction from the phase diagram. The variation in re action pathways associated with the precursor structure is discussed. Understanding the binary metal selenide pathways helps in designing a bilayer process for high rate manufacturing of CIGS The bilayer process comprising binary compound precursors are h ighly important as they have the potential to decrease the manufacturing cost and increase the throughput. In C hapter 3, the reaction pathways and kinetics of CIS formation is deduced for glass/Mo/ In 2 Se 3 /Cu 2 Se. This is similar to the precursor structure used for the NREL three stage process. Subsequently, the pathways and kinetics for the glass/Mo/ In 2 Se 3 /CuSe y precursor is pursued. CuSe y is a two phase mixture of Cu 2 Se and CuSe. During mass manu facturing using the co evaporation process, it is possible

PAGE 44

44 to form both these compounds because of the variation in elemental flux across the web. The effects of gallium addition to the group III sub lattice in kinetics are also investigated. The reactio n pathways and kinetics of CIGS employing metal selenization using bilayer metal precursors (glass/Mo/CuIn/CuGa/Se, glass/Mo/CuIn/CuGa+ Se vapor, glass/Mo/CuGa/CuIn/Se and glass/Mo/CuGa/CuIn + Se vapor) are studied in Chapter 4. The gallium distribution i n the final absorber is studied using Rietveld refinement and TEM EDS. The rate constants and activation energies are estimated from solid state growth models by performing isothermal studies at selected temperatures. The stacked elemental layer approach is another route to synthesis CIGS for high rate manufacturing as discussed in Chapter 5. The order of deposition of metals plays an important role in nucleation a nd growth of CIGS. In Chapter 5 the effect of the order of deposition of copper on the kin etics and morphology of the final grown CIGS is examined. MoSe 2 plays an important role in the adhesion and efficiency of a solar cell. In Chapter 6, selenization pathways of molybdenum are monitored as a function of temperature, time, and carrier gas. The role of hydrogen in the reduction of molybdneum is discussed. In Chapter 7, selenization kinetics of modulated structures is studied. Modulated structures give smooth films and helps in junction formation, thus improving the efficiency of devices. The e ffect of the number of repeating units of elements on kinetics is discussed. Also, the microstructure dependence on the modulated structure is discussed.

PAGE 45

45 In Chapter 8, effect of sodium in selenization kinetics is studied. It is known that sodium improves the microstructure and efficiency of the final device. However, limited information is available with respect to kinetics. The sodium in this study is introduced through sodium doped molybdenum layer. A nanoparticle based approach has the potential to re duce the cost of solar cells. One of the challenges faced by the nanoparticle based approach is the longer retention time to evaporate the solvents and binders. In Chapter 9 a new process is discussed that uses binder free nanoparticles and uses peritecti c decomposition of precursors to help in grain growth of CIS. Also, a simulation of the bilayer process using nanoparticle based approach is discussed. In Chapter 10 high temperature X ray diffraction is utilized for a complete solar cell device. It is u nderstood from the literature that CdS diffuses into CIGS and improves the performance of the cell. T he formation of the compound is limited by the kinetics and rate constants estimated from the isothermal experiments. The reliability of the device is th en extrapolated to the panel operating temperature from the kinetics data. In Chapter 11, E llingham diagram and reactor design for selenization is discussed. Ellingham diagram gives the equilibrium partial pressure over a compound at a particular temperatu re which would be really helpful during processing of CIGS. Also Fluent simulations were performed on continuous selenization reactor to get the flow, temperature and concentration profiles. In Chapter 1 2 the thermodynamic description of CIS CGS pseudobin ary system is evaluated using Thermocalc and Pandat. This description is necessary to understand

PAGE 46

46 the solid solution formation at different processing temperatures and to determine the homogeneity of the chalcopyrite phase with the addition of gallium. A pseudobinary description of Cu 2 Se Ga 2 Se 3 in Cu Ga Se is also presented.

PAGE 47

47 Table 1 1. Solar cell efficiency list based on different technologies Classification Efficiency Area(cm 2 ) V oc (V) J sc (mA/cm 2 ) FF(%) Crystalline Silicon [103] 25 4.00 0.706 42.7 82.8 Multicrystalline silicon [104] 20.4 1.00 0.664 38.0 80.9 Amorphous silicon [105] 10.1 1.04 0.886 16.8 67.0 CIGS [14] 20.3 0.996 0.713 34.8 79.2 CdTe [106] 16.7 1.03 0.845 26.1 75.5 GaAs thin film [3] 28.2 0.998 1.111 29.4 85.9 Dye sensitiz ed [107] 10.9 1.008 0.736 21.7 68.0 Organic polymer [108] 8.3 1.031 0.816 14.5 70.2 Amorphous silicon(triple junction) [105] 12.4 1.050 1.936 8.96 71.5

PAGE 48

48 Figure 1 1. Spectrum of solar radi ation for AM0 and AM1.5 conditions based on ASTM

PAGE 49

49 Figure 1 2 I V curve of a solar cell device

PAGE 50

50 Figure 1 3 Device structure of CIGS solar cell

PAGE 51

51 Figure 1 4 Crystal structure of CIGS solar cell

PAGE 52

52 Figure 1 5 NREL three stage co evaporation process for solar cell fabrication(Provided by Dr.Noufi from NREL)

PAGE 53

53 Figure 1 6 Cost of manufacturing of CIGS modules as a function of efficiency improvement( Provided by Dr.Noufi from NREL)

PAGE 54

54 Figure 1 7 Cost of manufacturing of CIGS modules with increase in module efficiency

PAGE 55

55 Figure 1 8 Cost of manufacturing of CIGS modules with increase in process scale

PAGE 56

56 CHAPTER 2 REACTION PATHWAYS AN D KINETICS IN METAL SELENIUM BINARY SYST EM Overview Solar cells based on CIGS have the potential to be leader in cost because of its excellent properties such as high efficiency [14] (20.3%), excellent outdoor stability and potential use in tandem cells. Ther e are different receipe to grow the absorber material [38 93] and some of the process uses binary compound precursors to synthesis CIGS for ease in scale up [109] For instance in co evaporation process, the growth takes place in the pseudobinary section of In 2 Se 3 Cu 2 Se. The properties of the material grown in the pseudobinary section are excellent in terms of film microstructure, texture and orientation. There exists a eutectic valley at 523 C where Cu 2 Se is in equilibrium with liquid and helps in the grain growth of CIGS. Hence, it is necessary to understand the reaction pathways of the Cu Se system for designing high rate bilayer process for rapid growth of CIGS [64] Since most synthesis processes are diffusion limited, an understanding of phase equilibria in these binary systems is relevant. Fortunately, the phase diagrams of the Cu Se system are well established as summarized below The thermochemical properties and phase equilibria in the binary Cu Se system were reviewed by Glazov et al [64] and a thermodynamic assessment using the CALPHAD (CALculation of PHAse Diagram) softwar e was performed by our research group [110] and Du et al [111] Four intermediate binary compounds Cu 2 x Se, Cu 3 Se 2 CuSe, and CuSe 2 were experimentally identified as shown in the phase diagram reproduced in Figure 2 1. The Cu 2 x Se compound is known to melt congruently and shows two polymorphs: the low temperature stab le Cu 2 x Se phase and the high temperature modification ( i.e ., Cu 2 x Se) having a transition temperature of around 396K. Three

PAGE 57

57 CuSe polymorphs, CuSe, CuSe and CuSe, were also reported. Stoichiometric Cu 2 Se or non stoichiometric Cu 2 x Se can exist in different crystal structure including orthorhombic, monoclinic, cubic and tetragonal. In a study by Verma et al. [112] glass/Mo/Cu precursor were reacted with H 2 Se for different times at 400 C and analyzed by XRD (X ray diffraction) to determine the phase evolution. The selenized copper structures yielded single phase Cu 2 x Se (x~0.15), which is the most stable high temperature phase in the binary Cu Se system, over the entire reaction time range (1 to 45 min). In another study Lakshmikumar and Rastogi studied the selenization of Cu thin film under flowing Se vapor in the temper ature range 340 to 400 C for 5 to 30 min [113] These st udies showed that hexagonal CuSe is the primary phase formed in the temperature range 340 to 370 C, while either cubic Cu 2 x Se or hexagonal CuSe resulted from reaction at 400 C depending on the selenium flux. The binary In Se system was assessed by Li et al based on the evaluation of available literature data [114 ] Multiple intermediate compounds (In 4 Se 3 InSe, In 6 Se 7 In 9 Se 11 In 5 Se 7 and polymorphic and In 2 Se 3 ) were identified as shown in the phase diagram (Figure 2 2 ). Selenization of pure indium was carried out by Verma et al using H 2 Se gas for d ifferent times at 400 C and analyzed the crystalline phases by XRD to determine the phases formed [112] The selenization of In for shortest reaction time (1.5 min) yielded a film constituted by 3 phases (In 2 Se, InSe, and unreacted pure indium). In the films selenized for 3 and 5 min, however, the pure indium and In 2 Se phases disap peared leaving InSe and In 2 Se 3 fully selenized In 2 Se 3 was evident in the XRD pattern, which is consistent with the binary In Se phase diagram shown in Figure 2 2

PAGE 58

58 Selenization of indium was also carried out by La kshmikumar et al. using pure selenium vapor in the temperature range 340 to 400 C for 5 to 30 min [113] The studies showed direct formation of In 2 Se 3 over the entire temperature range studied. These studies did not report quantitative rate data, but they do suggest that the selenization pathways are not simp le and depend on the reaction conditions. Phase diagram evaluation and thermodynamic assessment of the binary Ga Se system were reported by Dieleman et al. [115] and Zheng [116] According to these re ports, only two binary compounds ( i.e ., GaSe and Ga 2 Se 3 ) are stable in the Ga Se sy stem, as illustrated in Figure 2 3 Furthermore, Ga 2 Se 3 has two polymorphs: a low temperature stable Ga 2 Se 3 phase and its high temperature modification ( i.e ., Ga 2 Se 3 ) w ith a transition temperature around 967K. It is necessary to understand the kinetics of formation of Ga Se compounds to better understand the gallium distribution in quaternary CIGS In this chapter, reaction pathways and kinetics of metal selenium binari es are studied using in situ high temperature X ray diffraction deposited in sequential and co deposited fashion. Experimental The samples were deposited by both co evaporation and migration enhanced molecular beam epitaxy (MEE) in which effusion cells ar e employed to generate elemental vapors in ultra high vacuum (10 7 to 10 8 Torr) The precursors were deposited on sodium free glass substrates at room temperature to ensure no initial reaction. Though the deposition was intended at room temperature, temp erature measured by thermocouple was around 50 C which was attributed to radiation from the

PAGE 59

59 source The precursor was deposited in both bilayer and comixed configuration. In case of bilayer approach, the metal was deposited first onto glass followed by selenium. In comixed approach, both metal and selenium were deposited at room temperature to minimize the reaction between metal and selenium. The selenium used was in excess to account for selenium loss due to volatilization at high temperature. Phase HTXRD without overpressure of selenium. The Scintag HTXRD includes Scintag PAD X vertical / goniometer, a Buehler HDK 2.3 furnace, and an mBraun linear position sensitive detector (LPSD). The advantage of having LPSD over conventional X ray detectors is data collection over 10 2 window reducing the data collection time dramatically. In co nventional X ray, point scanning detectors are used to collect data by scanning step by step from lower to higher angles. This allows for in situ time resolved studies of phase transformations, crystallization, and grain growth. Temperature is measured by type S thermocouple welded onto the bottom of a Pt/Rh strip heater and gives feedback to the PID temperature controller. Samples are mounted on the platinum heater strip using carbon paint to improve the thermal contact between the sample and heater stri p. The sample temperature is calibrated using lattice expansion of silver powder dispersed on a identical substrate and comparing the results with the suggested literature. The PANalytical HTXRD system is composed of a PANalytical / X ray diffractometer equipped with an Anton Paar XRK 900 furnace heater in PANalytical HTXRD. The temperature difference between the furnace and the sample differs by 1 C. Both HTXRD furnaces were purged by flowing N 2 /He. The

PAGE 60

60 sample holder was covered with graphite dome to minimize the loss of selenium due to volatilization. For samples with poor signal to noise ratio, PANalytical HTXRD was used as it had better resolution than Scintag HTXRD. Results and Discussion Glass/Cu/Se B ilayer P recursor The series of diffraction patterns collected during the temperature ramp study for bilayer glass/Cu/Se sample is shown in Figure 2 4 The sample was heated to 50 C at a ramp rate of 10 C /min after a initial room temperature scan. The data was collected for every 10 C step increment till 330 C The room temperature scan reveals Se as well as CuSe. The selenium deposited was crystalline in nature as observed in the Figure 2 4 Selenium disappeared abruptly at 160 C, well below its melting temperature, and CuSe was transformed to CuSe 2 The Se chemical potential is sufficient to drive formation of CuSe 2 The CuSe 2 then decomposes to CuSe around 260 C, which is below the pertitectic decomposition temperature (~331.8 C). Apparently this is driven by Se loss. At around 290 C, with further loss of selenium, Cu 2 x Se was formed, and again this temperature is well below the CuSe peritectic reaction temperature (~379.5 C). The ICP analysis result of selenized film ( i.e ., x(Se) = 0.32) suggests that the composition of Cu 2 x Se phase should approach that of Cu 2 Se as ann ealing temperature increases. The reactions are summarized as follows (2 1) (2 2)

PAGE 61

61 (2 3) Furthermore TEM ( transmission electron microscopy) characterization was done for samples quenched at 240 C for qualitative support of the pathway. Selected area diffrac tion patterns showed the presence of CuSe 2 and CuSe existing together consistent with the result obtained during high temperature XRD as shown in Figure 2 5 Similar experiments carried out for glass/ Cu/Se prepared by MEE (migration enhanced epitaxy) gav e identical results with the samples prepared by co evaporation as shown in Figure 2 6. Glass/Cu+Se C omixed P recursor Temperature dependent phase evolution of the co deposited Cu+Se mixture on a glass substrate ( glass/Cu Se precursor) prepared by co evap oration wa s investigated using the PANlytical HTXRD system. The glass/Cu + Se precursor was scanned at 25 C, followed by subsequent heating and data collection at every 1 0 C step increment till 330 C. The room temperature scan (Figure 2 7 ) shows no cryst alline phases. As the temperature was ramped, abrupt formation of CuSe and Cu 7 Se 4 Cu 2 x Se phase) occurred at 90 C. The Cu 7 Se 4 phase transformed to CuSe phase at 170 C along with CuSe 2 formation. This series of phase transformation were similar t o glass/Cu/Se after formation of CuSe 2 Also no major difference in pathway was observed. TEM characterization was done for glass/Cu+Se sample quenched at 240 C and both CuSe 2 and CuSe coexisted as supported by selected area diffraction pattern. Figure 2 8 shows the image and selected area diffraction patterns for the quenched glass/Cu+Se sample The reaction sequence in the temperature ramp study is summarized as below

PAGE 62

62 (2 4) (2 5) (2 6) Similar studies done by Kyoung et al. on MEE grown precursors showed similar results as shown in Figure 2 9 [110] Glass/In/Se B ilayer P recursor Temprature ramp studies were done for glass/In/Se deposited by co evaporation in PANalytical HTXRD. X ray diffraction data were collected for every 10 C step i ncrement in forming gas atmosphere. Initially the ramp rate was set 10 C/min till the temperature reached 150 C after which the ramp rate was increased to 40 C/min. The total scan time for each temperature was approximately 2 minutes and forming gas w as used as car rier gas. As shown in Figure 2 10, the as prepared sample showed strong reflections for indium and weak ones for the indium selenide compound, In 4 Se 3 During ramping Indium melted slightly below its melting temperature due to a small temper ature overshoot produced by the high ramp rate. Since the In reflections were very strong relative to the compounds, the patterns taken at temperatures below In melting were removed from the sequence to better illustrate the disappearance of In 4 Se 3 In 2 Se 3 at hig her temperature. Thus Figure 2 11 shows only the collected patterns for temperature >150 C. No peaks for selenium was observed during temperature ramp as noticed on glass/In/Se. It is seen that reflections for In 4 Se 3 disappear at ~180 In 2 Se 3 start to appear at the same temperature. In 2 Se 3 phase continue to grow until the experimental temperature reaches 300 In 2 Se 3 coexists along wi th

PAGE 63

63 In 2 Se 3. In 2 Se 3 phase disappears completely at ~330 C and the In 2 Se 3 peak intensities continue to grow. The In 2 Se 3 exists during stage 1 in the NREL 3 stage process, where In and Se evaporative fluxes react on a heated substrate maintained at ~400 C. According to In In 2 Se 3 transforms to In 2 Se 3 at around 200 C. Brummer et.al has reported the phase transition temperature interval around 375 C along with formation of InSe [117] No InSe phase formation was observed for our samples. The comp ositional analysis (x (Se) =0.57 ) of the completely anneale d sample supports the single phase stoichiometry of In 2 Se 3 The reactions occurring during ramp are summarized as follows (2 7) (2 8) (2 9) Additionally TEM char acterization was performed on samples quenched during the ramp experiments. The In/Se bilayer was quenched at 320 C and allowed to cool down naturally. The TEM image showed large grains of In 2 Se 3 as evident from XRD. It was expected that In 2 Se 3 sho uld coexist with In 2 Se 3 phase as observed in high temperature XRD. Due to the large grain size of In 2 Se 3 only hexagonal patterns were observed in selected area diffraction. Figure 2 12 shows the 6 fold symmetry of In 2 Se 3 phase. Similar studies wer e done for MEE grown precursor where In 2 Se 3 phase was not detected in the scan. The transformation of In 4 Se 3 In 2 Se 3 was observed as shown in Figure 2 13

PAGE 64

64 Glass/In+Se Comixed P recursor Similar experiments were carried out for glass/In+Se prepared by co evaporation samples in forming gas atmosphere. X ray diffraction data were collected for every 10 C step increment. Initially the ramp rate was set 10 C/min till the temperature reached 150 C after which the ramp rate was increased to 40 C/min and eac h scan was approximately 2 minutes. The room temperature scan showed no crystalline phases, apparently a result of the low temperature deposition (Figure 2 14 ). At 210 C, abrupt reaction between In and Se occurred forming In 2 Se 3 and InSe. At 340 C In 2 Se 3 starts to grow with (202) preferred orientation The rapid reaction and product formation in comixed samples is due to shorter diffusion times compared with the bilayer samples as observed for samples deposited by MEE [110] The final product formed has nearly single crystal orientation. Again no selenium crystallizatio n was observed for comixed samples deposited by co evaporation. Similar quench experiments were performed for In+Se comixed samples and the samples were quenched at 325 C and cooled to room temperature naturally. The grain size of In 2 Se 3 was again large and almost selected area diffraction at all the spots yielded single crystal patterns. Figure 2 15 shows the SAED pattern for quenched sample. The reaction sequence for comixed In+Se precursor can be summarized as (2 10) (2 11) (2 12)

PAGE 65

65 Glass/Ga/Se Bilayer P recursor The PANalytical HTXRD system was used again for in situ investi gation of the phase evolution of the glass/Ga/Se precursor. The same experimental scheme ( i.e ., temperature ramp profile and X ray scan sequence) as used for glass/In/Se precursor was applied to this bilayer structure. Initially the ramp rate was set 10 C/min till the temperature reached 150 C after which the ramp rate was increased to 40 C/min As temperature increases, however, selenium begins to crystallize until temperature reaches the Se melting temperature (~221 C), at which temperature the sele nium peaks abruptly disappear as shown in Figure 2 16 Melted selenium subsequently reacts with Ga to form GaSe as evidenced by the appearance of the GaSe reflection peaks. From the binary Ga Se phase diagram of Figure 2 3 Ga 2 Se 3 phase is slightly more stable than GaSe phase under selenium rich condition Experiments were done without a Se overpressure and thus possible at lower Se partial pressure. This would favor the formation of GaSe as opposed to Ga 2 Se 3 which forms at high Se partial pressure. Ra mp results suggest that GaSe is more stable than Ga 2 Se 3 given the fact that selenium loss occurs due to high vapor pressure at high temperature. Additionally TEM characterization was done to support the results from high temperature XRD. The crystallite size was of the order of 300 nm and each grain was almost single crystal. Figure 2 17 shows the image and selected area diffraction pattern for the glass/Ga/Se quenched sample consistent with formation of GaSe observed in the high temperature XRD. Simila r results were obtained for MEE grown precursor even though the growth occurred in selenium rich conditions which is attributed to selenium loss owing to high vapor pressure as shown in Figure 2 18

PAGE 66

66 The reaction pathway sequence can be summarized as (2 13) (2 14) (2 15) Glass/Ga+Se Comixed P recursor Similar experiments were conducted for glass/Ga+Se prepared by co evaporation in forming gas atmosphere. All the expe rimental conditions (ramp rate and data collection time) were similar to glass/Ga/Se bilayer precursor. Ga 2 Se 3 formation took place for comixed precursor owing to higher local selenium pressure as shown in Figure 2 19. Comparing with MEE grown precursor (Figure 2 20) Ga 2 Se 3 formation took place at higher temperature (360 C) No other major difference was observed in the pathway compared with the sample prepared by MEE. TEM characterization was done for samples quenched at 250 C and selected area dif fraction patterns showed the presence of polycrystalline grains of Ga 2 Se 3 as shown in Figure 2 2 1 The ring patterns corresponded to the d spacing of the Ga 2 Se 3 as observed in the XRD (2 16) (2 17) (2 18) Isothermal Annealing and Reaction K inetics Isothermal selenization was performed at selected temperature from temperature ramp experiments for each binary to extract the kinetic information in terms of rate

PAGE 67

67 constant and activation energy. Scan was performed in the same furnace with the co evaporated samples with no selenium overpressure. In the end, temperature was increased to complete the reaction. The 2 scan range was selected such that the major reflection of each binary lies within this range. The fractional reactions of the product was estimated using normalized peak area assuming maximum peak area represents complete reaction. For In/Se bilayer s ample, the 2 range was selected from 22 o to 32 o since the major reflection of the desired product lies within this range. The glass/In/Se samples were ramped to the isothermal temperature by manually controlling the ramp rate. The maximum ramp rate of t he PANalytical system was 40 C/min when controlled using a program, whereas by manual control a ramp rate in the order of 120 C/min was obtained. The same method has been followed for all the isothermal runs. The growth rate of In 2 Se 3 was followed as a function of time at different isoth er mal temperatures and Figure 2 22 shows the isothermal set collected for 29 0 C It is well known fact that during the first stage of three stage co evaporation process, In 2 Se 3 is formed at around 380 C. It was d ifficult to follow the growth kinetics of In 2 Se 3 as the peaks were originating from the background and the scan time was also on the order of 1 minute giving poor signal to noise ratio For the comixed In+Se sample, the 2 range was selected from 24 o t o 35 o As discussed above, it was again difficult to follow the kinetics of In 2 Se 3 One of the interesting observations was made during isothermal studies of com ixed In+Se sample. In Figure 2 2 3 the last scan shows a small fraction of different orie ntation in addition to the major reflection. This orientation had brownish red color where as the major reflection had a silvery white color. During co evaporation of In+Se, there were regions

PAGE 68

68 of non uniformity. Isothermal data was collected only for tho se samples which yielded silvery white color. This texture in In 2 Se 3 can lead to texture dur ing the growth of CIGS. For Cu/Se, the growth rate of CuSe was determined. The 2 range was selected from 25 o to 35 o since the major reflection of CuSe 2 and CuSe lies within this range. Both the disappearance and appear ance kinetics of CuSe 2 and CuSe was followed. The disappearance kinetics of CuSe 2 was calculated from the Cu/Se bilayer isothermal scan data set. It is noted the disappearance of CuSe 2 is faster than the appearance of CuSe. Figure 2 2 4 shows the isothe rmal set of glass/Cu/Se followed at 240 C. For Cu+Se, the growth rate of CuSe was determined. The 2 range was same as the one selected for g lass/Cu/Se bilayer. Figure 2 2 5 shows the isothermal set for glass/Cu+Se sample followed at 270 C. In case of glass/Ga/Se, the 2 range was selected from 20 o to 28 o since the major reflection of GaSe li es within this range. Figure 2 2 6 shows the isothermal set for glass/Ga/Se sample at 262 C. Two different solid state growth models, avrami and parabolic rate m odel were used to get the kinetic information in terms of rate co nstant and activation energy In A vrami model the kinetic parameters under isothermal conditions are obtained by (2 19) 1.5. The value of the avrami exponent gives an idea about the nucleation mechanism. For instance, lower avrami exponent predicts nucleation is instantaneous and higher one predicts nucleation is finite. In this model, isotropic growth is assumed and the product

PAGE 69

69 regions formed are spherical in nature. It has been shown earlier that this model fits rele vant experimental data satisfactorily. For In/Se, comixed and bilayer Cu Se system, the linear fit for experimental data was good, but for other systems the data fit was satisfactory. Figure 2 2 7 shows the avrami plot for all the binary system studied du ring isothermal selenization. solid materials. For further reaction in forward direction, the reactants have to diffuse through the product layer The reaction kinetics of the para bolic growth rate model is described by (2 20) the diffusion coefficient of the mobile spe Figure 2 2 8 sho ws the plot for parabolic model for selected isothermal temperatures. Parabolic model provided good fit of data for bilayer samples of In Se and Cu Se system. Rate constants obtained from both the models were used to extract activation energy using Arrhen ius equation, (2 21) Based on the result of two models, the process follows one dimensional diffusion controlled reaction with nucleation and growth sequence. The rate constants, exponents and activation energy calculat ed both the models are summarizes in Table 2 1. Figure 2 29 shows the Arrhenius plot based on Avrami and parabolic growth models.

PAGE 70

70 Summary The phase evolution of mixed and stacked bilayer Cu Se, In Se, and Ga Se precursor structures grown by co evaporation was examined u sing in situ high temperature X RD. The results show the overall phase transformation of binary metal (Cu, In and Ga) Se precursors qualitatively follow the sequence predicted by the thermodynamic phase diagram. The intermediate reaction pro ducts, however, depend on the as deposited precursor structure and starting compounds. For instance, the co mixed glass/Cu + Se precursor prepared by both the techniques takes the sequence of Cu 7 Se 4 CuSe Cu 2 x Se while the glass/Cu/Se bi layer precursor b eginning with the CuSe, presumably as an intermediate layer, follows the reaction path of CuSe CuSe 2 CuSe Cu 2 x Se For the glass/In/Se precursors, reaction sequence follows In 4 Se 3 In 2 Se 3 In 2 Se 3 where as for glass/In+Se, reaction sequence follows InSe+ In 2 Se 3 In 2 Se 3 (preferred). For glass/Ga/Se bilayer precursor, GaSe formation takes place and for glass/Ga+Se precursor stable Ga 2 Se 3 phase was observed. Isothermal soaki ng experiments were performed for the precursor structures grown by co evaporation and kinetic parameters were calculated from avrami and parabolic solid state growth models. Activation energy calculated from A vrami model were ; Cu/Se: 138.3 kJ/mole, Cu+S e:100.3 kJ/mole, In/Se: 155.9 kJ/mole, In+Se: 168.3 kJ/mole and Ga/Se: 198.9 kJ/mole. Parabolic model provided good fit of data for In/Se and Cu/Se precursor structure yielding activation energy in the order of 161kJ/mole and 144.3 kJ/mole. Due to amorph ous nature of Ga Se precursor, it was difficult to follow the kinetics

PAGE 71

71 Table 2 1. Kinetics parameters calculated from Avrami and parabolic model Precursor Avrami model Parabolic model k*10 5 (1/s) E a (kJ/mole) k*10 5 (1/s) E a (kJ/mole) T ( C) Glass/In/Se 5.71 155.9 1.5 161 280 11.76 4.3 290 23.78 9.6 307 65.24 20 318 80.64 25 328 Glass/In+Se 31.3 168.3 325 42.3 335 39.5 345 186.5 360 290.1 370 Gl ass/Cu/Se 70.6 138.4 24.3 144.3 240 153.0 54.5 250 245.7 86.5 260 440.2 166.0 270 Glass/Cu+Se 301.2 100.3 210 314.1 220 658.1 230 1201.1 240 Glass/Ga/Se 3.6 198.9 228 15.8 236 37.9 250 173.9 262 216.5 275

PAGE 72

72 Figure 2 1 Calculated Cu Se phase diagram using optimized Thermocalc database [110]

PAGE 73

73 Figure 2 2 Calculated In Se phase diagram using optimized Thermocalc data base [110]

PAGE 74

74 Figure 2 3 Calculated Ga Se phase diagram using optimized Thermocalc database [110]

PAGE 75

75 Figure 2 4 Temperature ramp annealing of g lass/Cu/Se bilayer precursor

PAGE 76

76 Fi gure 2 5 TEM image of quenched bilayer precursor of glass/Cu/Se A) image of quenched precursor, B) selected area diffraction of quenched precursor A B

PAGE 77

77 Figure 2 6 Temperature ramp annealing of MEE grown glass/Cu/Se bilayer precursor [110]

PAGE 78

78 Figure 2 7 Temperature r amp annealing of co evaporated glass/Cu+Se comixed precursor

PAGE 79

79 Figure 2 8 TEM images of quenched bilayer precursor of glass/Cu+Se A) image of quenched sample, B) selected area diffraction pattern of quenched sample B A

PAGE 80

80 Figure 2 9 Temperature ramp annealing of MEE grown glass/Cu+Se comixed precursor [110]

PAGE 81

81 Figure 2 10 Temperature ramp annealing of co evaporated glass/In/Se bilayer precursor

PAGE 82

82 Figure 2 11 Temperature ramp annealing of co evaporated glass/In/Se bilayer precursor (Data shown from 150 o C)

PAGE 83

83 Figure 2 12 TEM images of quenched bilayer precursor of glass/In/Se A) image of quenched sample, B) selected area diffraction pattern of quenched sample B A

PAGE 84

84 Figure 2 13 Temperature ramp annealing of MEE grown glass/In/Se bilayer precursor [110]

PAGE 85

85 Figure 2 14 Temperature ramp annealing of co evaporated glass/In+Se comixed precursor

PAGE 86

86 Figure 2 15 TEM images of quenched bilayer precursor of glass/In+Se A) image of quenched sample, B) selected area diffraction pattern of quenched sample B A

PAGE 87

87 Figure 2 16 Temperature ramp annealing of co evaporated glass/Ga/Se bilayer precursor

PAGE 88

88 Figure 2 1 7 TEM images of quenched bilayer precursor of glass/Ga/Se A) image of quenched sample, B) selected area diffraction pattern of quenched sample B A

PAGE 89

89 Figure 2 18 Temperature ramp annealing of MEE grown glass/Ga/ Se bilayer precursor [110]

PAGE 90

90 Figure 2 19 Temperature ramp annealin g of co evaporated glass/Ga+Se comixed precursor

PAGE 91

91 Figure 2 20 Temperature ramp annealing of MEE grown glass/Ga+Se comixed precursor [110]

PAGE 92

92 Figure 2 21 TEM images of quenched bilayer precursor of glass/Ga+Se A) image of quenched sample, B) selected area diffraction pattern of quenched sample B A

PAGE 93

93 Figure 2 2 2 Isothermal annealing of glass/In/Se bilayer precursor grown by evaporation at 290 o C

PAGE 94

94 Figure 2 23. Isothermal annealing of glass/In+Se comixed precursor grown by evaporation at 360 o C

PAGE 95

95 Figure 2 24. Isothermal annealing of g lass/Cu/Se bilayer precursor grown by evaporation at 240 o C

PAGE 96

96 Figure 2 25. Isothermal annealing of glass/Cu+Se comixed precursor grown by evaporation at 220 o C

PAGE 97

97 Figure 2 26. Isothermal annealing of glass/Ga+Se comixed precursor grown by evaporation at 236 o C

PAGE 98

98 Figure 2 27. Avrami plot for samples from isothermal selenization studies. A) glass/Cu/Se precursor, B) glass/Cu+Se precursor, C) glass/In/Se precursor, D) glass/In+Se precursor, E) glass/Ga/Se precursor A B C D E

PAGE 99

99 Figure 2 28. Parabolic plot from isothermal selenization studies. A) glass/Cu/Se precursor, B) glass/In/Se precursor A B

PAGE 100

100 Figure 2 29. Arrhenius plot using rate constants obtained from isothermal an nealing experiments. A) glass/Cu/Se precursor, B) glass/In/Se precursor, C) glass/Ga/Se precursor A B C

PAGE 101

101 CHAPTER 3 REACTION PATHWAYS AN D KINETICS OF CIS FO RMATION FROM BILAYER COMPOUND PRECURSORS Overview Chalcopyrite Cu (In x Ga 1 x ) Se 2 (CIGS) is a well establis hed absorber material for high efficiency thin film solar cells [38] that is currently undergoing intense commercialization. Remarkably, high efficiency CIGS absorber layers can be synthesized using a variety of pre cursor structures and a wide range of deposition methods. Indeed, the industry is largely differentiated on the basis of how the absorber material is formed. Synthesis processes can be single step (e.g., co deposition and reaction of the elements), two s tep (e.g., co deposition of metals followed by selenization using H 2 Se or Se) or a greater number of steps (e.g. printing of oxide nanoparticles followed by their reduction in H 2 and then oxidation in a Se ambient). These structures can be initially depos ited as a single layer or as stacked layers. Differentiation is also found by the method of deposition, which includes co evaporation [118] sputtering followed by selenization [119, 120] printing nanoparticle suspensions [92] and electrodeposition [109, 121] An obvious strategy to reduce CIGS cell manufacturing cos t is to shorten the absorber synthesis time. This strategy is supported by the U.S. national solar technology roadmap for CIGS PV, which specifically calls for the reduction in absorber synthesis time to the range 1.5 to 2 min [122] It is noted that the absorber synthesis rate is not limited by the rate of material deposition but rather by the rate of reaction to form high quality CIGS. Studies of reaction pathways and rates have indicated that synthesis rates are typically diffusion limited during most of the synthesis time [123 126] This suggests that approaches to increasing synthesis rates should focus on

PAGE 102

102 increasing the diffusivity of the limiting species (e.g., increasing the synthesis temperature, modifying the diffusion mechanism) or decreasing the diffusion distance (e.g. co evaporation of all elements). Of course achieving high synthesis rate does not necessarily produce an efficient absorber layer as other growth outcomes such as point defect concentration profiles, Ga distribution, and microstructure are also important. Reaction pathways that are not rapid, however, should be avoided based on lo w throughput concerns. In terms of synthesis rate the co evaporation process has the advantage of forming metal Se bonds as deposition proceeds. The first stage in the NREL 3 stage co evaporation process [127] deposits the sesquiselenides of Ga and In, III 2 Se 3 followed by co evaporation of Cu+Se in the 2 nd stage at conditions that would deposit Cu 2 x Se and yield a Cu rich composition for the combined two stages A 3 rd stage is then used to remove excess Cu Se phases and render the overall composition In rich. The earlier 2 stage Boeing recipe [128] reverses the first two stages of the NREL sequenc e but deposits an overall In rich composition in the 2 stages, thus also transitioning from Cu rich to In rich growth conditions during synthesis CIGS forms at the interface during the 2 nd stage and presents a growing diffusion barrier. There have been several reports in the literature describing the mechanism of CIGS formation utilizing the co evaporation process [129 131] It is clear that overall Cu rich conditions are needed to grow large grain absorbers with the beneficial ( 112) preferred orientation. It has been suggested that the monotectic reaction L 3 + Cu 2 x Se L 2 (523 C) or its ternary extension liquid phase assists growth [65]

PAGE 103

103 This study compa res the reaction pathway of the 2 nd stage of the NREL process [127] u sing in situ high temperature X ray diffraction (HT XRD) for two different Cu dep osition conditions, both of which can yield a liquid phase at elevated temperature. The first structure, glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se /Se incorporates the two phase mixture CuSe + Cu 2 Se The Cu Se phase diagram [132] indicates that as temperature increases CuSe transforms to CuSe ( C) (136.9 C ), then peritectic ally decomposes CuSe Cu 2 x Se + L 3 ( 379.5 C) The second sample, glass/Mo / In 2 Se 3 / Cu 2 Se /Se simulates the traditional process and contains only Cu 2 Se which as previously stated, monotectically decomposes at 523 C. Since the equilibrium Se 2 partial pressure of the peritectic reaction (0.23 Torr) is considerably lower t han that for the monotectic reaction (13.2 Torr) it is more experimentally accessible. The third sample glass/Mo/ (In,Ga) 2 Se 3 /CuSe studies the addition of gallium to the group III sub lattice and its effect on the rate. The anticipated beneficial effe cts of liquid phase assisted growth include increased synthesis rate, reduced processing temperature, improved microstructure, and increased net carrier concentration. Experimental The precursor films examined in this study were deposited by co evaporatio n of the elements in a chamber with a background pressure in the range 10 7 to 10 8 Torr Prior to precursor deposition, molybdenum (0.6 m) was deposited on 0.4 mm thick, sodium free glass substrates. Indium and Se were first co evaporated at 350 C to form In 2 Se 3

PAGE 104

104 C to avoid significant reaction between In 2 Se 3 and the subsequently co evaporated volatil ization loss during the temperature ramp studies. The Se source was maintained at 300 C and the beam pressure was ~10 5 Torr. It is noted that attempts to deposit CuSe 2 by co evaporation were not successful because the substrate could not be sufficientl y cooled to match the required Se vapor pressure to that available in the growth system. In addition, t he required Se partial pressure was likely higher given the low Se sticking coefficient For (In,Ga) 2 Se 3 higher substrate temperature of 450 C was used for deposition followed by low temperature deposition of CuSe at 150 C. The elemental composition of each precursor film was determined by inductively coupled plasma optical emission spectro scopy (ICP OES, Perkin Elmer Plasma 3200) and the results for the as deposited samples are summarized in Table 3 1. This table also displays the elemental abundances of the annealed samples from the temperature ramp experiments. All high temperature X ra y diffraction studies were performed in a PANalytical / X ray diffractometer and 900 reaction furnace that surrounds the sample. As described in more detail in Chapter 2, a graphite dome was placed over the sample to better contain Se and maintain a reasonable Se partial pressure. The substrate temperature was calibrated by measurement of the shift in the lattice parameter of Ag as temperature was varied with the graphite dome in position. The calibration results indicate th e difference between the furnace and the sample temperature is 1 C. An additional experiment was

PAGE 105

105 performed using a sealed stainless steel chamber that enabled higher Se partial pressure [133] In the temperature ramp experiments, forming gas (4 mol % H 2 / N 2 ) was used to blanket the system after purging three times prior to starting the experiments to reduce possible oxi des. In the isothermal studies the chamber external to the graphite dome was blanketed with N 2 again pressure purged 3 times. Temperature Ramp Anneali n g Glass/Mo/In 2 Se 3 / CuSe+ Cu 2 Se/Se Precursor Phase evolution during heating the In 2 Se 3 / CuSe+ Cu 2 Se/Se samples was followed in the PANalytical HT XRD system from room temperature to 500 C. The temperature was stepped in 5 C increments and diffraction data were acqui red at each step in the 2 range 20 to 55 in ~2 min. The temperature was stepped using a ramp rate of 20 C/min in the temperature range room temperature to 200 C, while the ramp rate was increased to 40 C/min above 200 C. This schedule was used to a void temperature overshoot at lower temperature The high resolution XRD pattern of the as deposited sample at room temperature (Figure 3 1) revealed reflections of In 2 Se 3 (hexagonal), CuSe (orthorhombic), Cu 2 Se (orthorhombic), selenium (orthorhombic ) and molybdneum (bcc). It is noted that the molybdenum had (110) preferred orientation. It has been reported that when molybdenum is textured in the (11 0 ) CIGS growth is preferred in the (220) [134] The phase evolution of the glass/Mo/In 2 Se 3 / CuSe+ Cu 2 Se/Se stack upon step heating is shown in Figure 3 2. This figure plots sequentially the set of XRD patterns as the sample temperature is incremented. As expected, elemental Se is seen to crystallize (peak sharpening) and melt (peak disappea rance) at the reported melting temperature of 221 C (i.e. a peak was observed

PAGE 106

106 at 220 C (2 = 46.8) and no peak observed at 225 C). This result also verifies the high accuracy of temperature control and calibration. A summary of the primary reactions an d their initiation temperature observed during the temperature ramp annealing experiment is summarized below. (3 1) (3 2) (3 3) (3 4) (3 5) The CuSe transformed to CuSe at 200 C (the equilibrium transformation is 140 C) suggesting the kinetics of transformation are rate limiting [64] Solid state transformations are anticipated to be slow at l ow temperature. The Se rich phase, CuSe 2 appears in the diffraction pattern at 195 C, consuming Cu 2 Se via reaction (3). Apparently, the Se overpressure provided by the cap is sufficient to form this phase at this relatively low temperature likely ne ar the top surface in contact with the selenium cap. The 2 phase CuSe + CuSe 2 region remains intact until 280 C, and then decomposes back to CuSe and Se loss to another phase at higher temperature. According to the Cu Se phase diagram at 140 C, C u 2 Se and CuSe are in equilibrium for an overall mixture slightly Cu rich compared to CuSe, and at 280 C CuSe 2 and CuSe are at equilibrium for slightly Se rich compositions. The peritectic reaction of CuSe 2 to CuSe and Se rich liquid does not occur until 331.8 C. This suggests that the partial pressure of Se in the system is too low at 331.8 C to support

PAGE 107

107 the existence of CuSe 2 and that at 280 C CuSe 2 decomposes to CuSe releasing Se to the vapor phase. From the P T diagram of the Cu Se system shown in Figure 3 3, the vapor pressure of Se 2 at 331.8 C is 0.2 0 Torr and total Se partial pressure 0.49 Torr. If the pressure is below this value, the reduction reaction can occur at lower temperature by evaporation of Se rather than forming a Se rich liquid. CIS forms at 275 C by the reaction of CuSe and In 2 Se 3 accompanied by the formation of InSe (Reaction 5) This is consistent with the large negative Gibbs energy of formation of CIS via reaction (5) R = 77 kJ/mol CIS at 548 K [135] ). The formation of CIS occurs at a slightly higher temperature than the literature reported value of ~2 60 C [123] A high resolution X ray diffraction pattern was collected on the cooled sample after the temper at ure ramp experiment (Figure 3 4 A ). The annealed sample showed CIS reflections with a strong ( 112 ) orientation. A trace of MoS e 2 (100) was detected indicating completion of the CIS formation reaction. This orientation is the preferred one for formation of an ohmic contact and adherent film. The SEM (scanning electron microscopy) image of the annealed sample shown in Figure s 3 5A reveals columnar grain growth with the average lateral grain size dimension estimated as ~2.5 m. The surface SEM image (Figure 3 5 B ) shows well connected grains. It is interesting that CIS formed with the desired texture and with such a large grain siz e, given that growth initiated at ~275 C and the final temperature was below the typical 550 C value. Furthermore, the rate of increase in the dominant CIS reflection (2 = 26.67 ) in Figure 3 2 suggests that CIS formation by reaction (5) is rapid. Su bsequent annealing appears to increase grain size and perhaps yields additional reaction as evidenced by peak sharpening above 415 C.

PAGE 108

108 Glass/Mo/In 2 Se 3 / Cu 2 Se/Se Precursor Phase evolution during heating the In 2 Se 3 / Cu 2 Se/Se samples was followed in a P ANalytical HT XRD system from room temperature to 500 C with the same condition as discussed above. The high resolution room temperature XRD pattern of the as deposited second precursor that emulates the traditional 2 nd stage process is shown in Figure 3 6. As expected this figure shows reflections assigned to In 2 Se 3 (hexagonal), Cu 2 Se (orthorhombic), Se (orthorhombi c), and molybdenum textured in (110) (bcc). The temperature ramp diffraction data show that Se oxidized Cu 2 Se at 140 C to yield CuSe. Around 180 C, CuSe then reacted with Se to form the Se rich compound CuSe 2 which continues to grow until 260 C, at which temperature it decomposes to CuSe and presumably Se vapor. Again the decomposition temperature of CuSe 2 is much lower than the peritect ic decomposition temperature (331.8 C ) The formation of CIS proceeds by the reaction of In 2 Se 3 with CuSe at 280 C (reaction 5) The CIS formation temperature was higher compared that observed in ramping the glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 S e /Se sample The reaction was complete at around 415 C as determined by the disappearance of InSe. With a further increase in temperature and subsequent Se loss, an additional peak (2 =30.8) was identified at ~480 C that does not belong to the CIS fam ily. The phase was identified as (Cu 2 Se) x (In 2 Se 3 ) 1 x as shown in Figure 3 7 which is reported as Cu poor ordered vacancy compound and also as CIS. This phase has been reported during CIS formation using Cu 2 Se and In 2 Se 3 as starting phases [136] After temperature ramp annealing, the sample was naturally cooled to room temperature and a post anneal XRD scan at was taken at 25 C. The high resolution

PAGE 109

109 scan showed reflections for InSe and (Cu 2 Se) x (In 2 Se 3 ) 1 x in addition to CIS as illustrated in Figure 3 4 B The reactions observed during temperature ramp selenization are summarized below: (3 6) (3 7) (3 8) (3 9) It is speculated that (Cu 2 Se) x (In 2 Se 3 ) 1 x forms because of Se loss at high temperature. Thus, another temperature ramp anneal was performed on this same bilayer stack but with a higher background Se pressur e. Loss of Se was not completely prevented using the graphite dome resting on the sample holder. Thus a custom gas tight stainless steel chamber using aluminum foil as an X ray transparent window was constructed to fully contain Se [133] The glass/Mo/ In 2 Se 3 /Cu 2 Se/Se precursor sample was placed in the reaction chamber along with additional Se powder (~ 40 mg) and the temperature ramp experiment repeated. The set of XRD scans taken during the temperature ramp are sequenced in Figure 3 8. It is first noted that reflections from the aluminum window are present in the patterns. The same scan temperature/time sequence was used as for the measurements shown in Figure 3 7. T he reaction pathway shows the same qualitative pathway as observed at lower Se pr essure The transformation temperatures, however, are higher for several reactions. For example, the InSe phase completely disappeared at 385 C, a much lower than in the ramp experiment at lower Se pressure. InSe was not detected in the high resolution scan at the end of the experiment. The peritectic

PAGE 110

110 decomposition of CuSe 2 and CuSe occurred at 315 and 375 C which are close to equilibrium de composition temperatures (331.8 and 379.5 C respectively ) consistent with no Se loss from the chamber The reaction was complete at 415 C as evidenced by the formation of MoSe 2 With a further increase in temperature, the molybdneum peak disappeared and Mo 3 Se 4 also formed indicating reaction of the excess Se by Mo An SEM image of the sample cross section (F igure 3 5 C ) shows columnar grain growth with a total film thickness of ~1 m. The SEM surface image in Figure 3 5 D shows densified grains with lateral dimensions on the order of ~2 m. Glass/Mo/ (In,Ga) 2 Se 3 /CuSe Precursor The high resolution XRD spect ra showed reflections of (In,Ga) 2 Se 3 (hexagonal), CuSe (hexagonal) and molybdenum as shown in Figure 3 9. Temperature ramp in custom built stainless steel reactor Initially, room temperature scan was taken at 25 C, followed by subsequent data collection at every 10C step increment. A constant ramp rate of 10C/min was used for the entire temperature ramp studies. Figure 3 10 shows the temperature ramp annealing of glass/M o/ (In Ga ) 2 Se 3 /CuSe. The CuSe transformed to CuSe 2 at 250 C which occurs by selenium incorporation from gas phase (The vapor pressure of selenium at 250C is around 0.05 torr). As observed in the previous study of glass/Mo/ In 2 Se 3 / Cu 2 Se, the CuSe 2 decomposes to CuSe and selenium rich liquid by a peritectic reaction at 330 C. The second peritectic reaction occurs at 380C, where CuSe transforms to Cu 2 Se and selenium rich liquid. CIGS formation takes place at 340C with no indium or gallium segre gation. The substrate

PAGE 111

111 used is alkali free glass and hence segregation of indium and gallium is not evident as reported in the literature. The rate of formation of CIGS increases after the second peritectic temperature indicating a liquid assisted growth of CIGS. The reaction was complete at 470 C with the onset of MoSe 2 formation. The following reactions were observed during temperature ramp (3 10) (3 11) (3 12) (3 13) Isothermal Annealing and Reaction Kinetics Glass/Mo/In 2 Se 3 / CuSe+ Cu 2 Se/Se Precursor To quantitatively determine the rate parameters for the formation of CIS, t ime resolved, high temperature X ray diffraction data were collected for a set of isothermal temperatures for glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se precur sor films. The specific temperatures were selected to give reasonable experiment times using the results of the temperature ramp experiments described in the previous section. The system was first purged with N 2 to minimize O 2 inside the chamber. Since t he maximum programmable ramp rate for the C/min, the temperature was manually increased at 120 C/min to the desired experimental temperature The 2 scan range was set from 24 to 32 such that the ma jor reflections of the product CIS ( 112 ) and the reactant In 2 Se 3 (110) lie within this range. The total

PAGE 112

112 scan time for each data set was 79 sec. After 90 such data sets, the temperature was ramped to 500 C and diffraction data were collected until no appreciable change in the CIS peak was evident (i.e., complete reaction) The fractional conversi on was obtained by calculating the area under the peak using the JADE software package followed by normalization assuming complete conversion and no change in the texture of CIS during isothermal soaking. T he isothermal data were collect ed in the temperat ure range 260 to 310 C with 10 C increment to provide 6 isothermal data sets. Figure 3 11 shows the selected isothermal annealing scans for the precursors annealed at different temperatures. It is noted that the CuSe phase had a ( 00 l ) texture. The calculated fractional co nversion values based on peak area were used to determine the rate parameters using two different solid state growth models, the Avrami and parabolic models as discussed in Chapter 2 Figure s 3 12 A and 3 12 B shows the Avrami and parabolic plot obtained us ing fractional conversion data calculated from isothermal studies. The value of Avrami exponent n varied from 0.8 to 1.1 for this particular precursor film. This range is consistent with a one dimensional diffusion controlled transformation. Figure 3 13 A a nd 3 13 B represents the kinetics obtained from Avrami and parabolic model It is noted that two independent experiments were performed at each temperature, differentiated by symbol style, to improve the statistical significance of the data The activation energy was estimated by simultaneous regression of both data The value calculated from Avrami model yielded activation energy 162 (7) kJ/mol, while that from the parabolic model gave the higher value 225 (16) kJ/mol. This significant difference in th e two estimates as well as the poor extrapolation of the conversion data to t=0 suggest that nucleation is sufficiently slow that the Avrami model

PAGE 113

113 results should be accepted. The approaches to selenization of metal precursors in industry use either elemen tal Se or H 2 Se. The latter is extremely toxic, more expensive, and likely of lower purity, although it is believed to produce superior results [137] It is possib le that the difference in the two approaches may be related to the reduction of surface species (e.g. oxides). To test this hypothesis relative to a 1 step process, a set of isothermal experiments was performed with the glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se /Se precursor using a 4 mol % H 2 /He mixture rather than N 2 and the graphite dome The isothermal HT XRD studies were performed at six different temperatures and the fractional conversions as a function of time was tested with both t he Avrami and parabolic growth models. Figure 3 13 A and 3 13 B compares the rate constant, k obtained by data analysis using both growth models. The activation energy estimated from the Arrhenius plot yielded values of 108 (8) kJ/mol using the Avrami m odel and 158 (16 ) kJ/mol using the parabolic model. As with the N 2 case, the H 2 /He background gas showed a large discrepancy between the two models, supporting the Avrami model results. Interestingly, the activation energy for the He/H 2 gas blanket is l ower than that of the N 2 case. The absolute rate in N 2 is the same as that the H 2 /He ambient at temperature of ~275 C and higher activation energy for the N 2 ambient case gives higher synthesis rate when extrapolated to higher temperature. The thermal conductivity of He/H 2 is higher than it is for N 2 so perhaps the temperature was lower. A temperature calibration was performed using a He/H 2 blanket with no change observed. It is also possible that H 2 plays a role in reducing selenides or oxides at the growth interface. The reduction of selenides by H 2 would be expected to reduce the growth

PAGE 114

114 rate while the reduction of oxides might be expected to increase the rate. These data, however, do not suggest a mechanism for the measured rate differences. The temperatures at which the extrapolate d rate yields a 2 min processing time for synthesis of the abs orber is indicated in Figure 3 13 (i.e. when ln(k) = 2.5) In this calculation, first order kinetics were assumed and represented as : (3 13) k is the first order rate constant, and t is time. Using Eq. 13 the value of k to achieve 99.99% conversion ( =0.9999) in 2 min is 0.077 s 1 (i.e. when ln(k) = 2.5) Extrapolating the kinetic parameters graphically shown in Figures 3 13 gives an estimate of the temperature required to synthesize CIS in 2 min, assuming there is no change in the reaction mechanism at higher temperature Using these temperature values, the precursors were annealed using rapid thermal annealing ( RTA ) to check if 2 min was suffic ient time to yield CIS. The reaction conditions along with phases identified by XRD after a 2 min anneal are reported in Table 3 2 As shown in this table glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se /Se sample annealed for 2 min showed incomplete reaction at an annealing temperature of 350 C and 370 C. The XRD pattern showed residual CuSe and InSe secondary phases co existing with CIS. At the higher temperature of 390 C, on ly CIS and MoSe 2 were evident, which is a signature for complete conversion. These results support the 2 min synthesis time in a real processing environment at higher temperature In the case of glass/Mo/ In 2 Se 3 / Cu 2 Se/Se, the reaction did not go to co mpletion at 350 C, but it did when annealed at 370 C. The data agree with the extrapolated temperatures to provide a sufficient rate

PAGE 115

115 to synthesize CIS in 2 min. Figure 3 14 shows the SEM images of synthesized samples using RTA for 2 m in at the highest t emperatu re. The grain structure of the precursor annealed at 390 C is clearly evident but not as columnar as typically observed in high efficiency absorbers Glass/Mo/In 2 Se 3 / Cu 2 Se/Se Precursor A set of isothermal temperatures for glass/Mo/ In 2 Se 3 /Cu 2 S e/Se precursor films were performed to determine the rate parameters for the formation of CIS using time resolved high temperature X ray diffraction. The same isothermal sequence was used as described for glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se precursor film s Data were collected in the temperature range 250 to 300 C for glass/Mo/ In 2 Se 3 / Cu 2 Se/Se to produce 6 data sets. Figure 3 1 5 shows the isothermal annealing scans for selected isothermal temperatures. It is noted that for this precursor structure, Cu Se had (102) preferred orientation. The fractional conversions were calculated based on peak area to determine the rate parameters from Avrami and parabolic model. T he Avrami exponent varied from 0.6 to 1.2 in the temperature range 250 to 300 C, which i s consistent with a one dimensional diffusion controlled transformation. Figures 3 1 6 A and 3 1 6 B shows the Avrami and parabolic plot for using the fractional conversion data obtained from normalization procedure. The rate constants were then calculated f rom the Avrami and parabolic model and then plotted against isothermal temperature in Arrhenius relationship as discussed in Chapter 2. The estim ated activation energy was 194 ( 1 0) kJ/mol and 20 3 ( 1 2) kJ/mole f or the Avrami and parabolic model s, respect ively as shown in Figures 3 1 7 A and 3 1 7 B The much closer values are consistent with rapid nucleation as well as being growth limited by 1 D diffusion across the reaction product

PAGE 116

116 (CIS). Although the CIS synthesis reaction is the same overall one for bot h glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se and glass/Mo/ In 2 Se 3 / Cu 2 Se/Se precursors, the estimated activation energies are different. The reason for this is not clear and more detailed mechanistic study is needed. It is noted that the overall composition is slightly different (Table 3 1) and that would modify the point defect concentrations and thus diffusion rates. Similar RTA experiments were performed to test the kinetics obtained from the solid stat growth models as discussed above. In the case of glass/Mo/ In 2 Se 3 / Cu 2 Se/Se precursor, the reaction did not go to completion at 350 C, but it did when annealed at 370 C. The data agree with the extrapolated temperatures to provide a sufficient rate to synthesize CIS in 2 min. Figure 3 18 shows the SEM image of annealed glass/Mo/ In 2 Se 3 / Cu 2 Se/Se precursor show ing large columnar grain growth at the low annealing temperature of 370 C The grain structure is similar to samples annealed at higher temperature in the literature. To our knowledge this is the lowest reported temperature for rapid (<2 min) growth of CIS using a bilayer approach Glass/Mo/ (In,Ga) 2 Se 3 /CuSe Precursor Isothermal annealing experiments were carried out in with custom built stainless steel reactor to determine the rate of formation of CIGS. The precursor film was held at an isothermal temperature and the growth of major reflection ( 112 ) of CIGS was followed as a function of time. A constant ramp rate of 120C/min was used to attain the isothermal temperature. In t he end, temperature w as increased to 580 C and scans were recorded for 10 minutes to ensure completion of the reaction. Individual scan time was approximately 73 seconds and the 2 range was selected such

PAGE 117

117 that the major reflection of the p roduct remains in this range. The temperature range for isothermal experiments was from 300 360C and isothermal data was collected from 20 C step increment. Figure 3 1 9 shows the isothermal a nnealing of precursor at different isothermal temperatures. The fractional conversion was calculated in a similar fashion as discussed above in the previous sections. The fractional conversion data was then used to calculate the rate constants from Avrami an d parabolic model. Figure 3 20 A and 3 20 B shows the Avrami and parabolic plot for the precursor and both the models yielded satisfactory fit to the data indicating one dimensional diffusion controlled transformation. Figure 3 21 shows the Arrhenius plo t for glass/ Mo/ (In,Ga) 2 Se 3 /CuSe precursor. The rate from the Avrami kinetics seems to be faster than the parabolic model with lower activation energy Summary The reaction pathway was investigated for three different bilayer precursor films using in s itu high temperature X ray diffraction. The effect of Se pressure on phase transformation or peritectic decomposition has been elucidated. The reaction pathway for each precursor structure differed at low temperature, but the CIS synthesis reaction was th e same. The rate for the In 2 Se 3 / CuSe+ Cu 2 Se precursor was slightly higher than that for the In 2 Se 3 / Cu 2 Se one. A quantitative model was established for all the precursor films using the Avrami and parabolic solid state growth models. Based on the mo deling results, the glass/Mo/ In 2 Se 3 / Cu 2 Se /Se precursors follows a one dimensional diffusion controlled reaction but the result for the glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se /Se precursor are consistent with a nucleation process followed by diffusion control led growth For glass/Mo/ ( In Ga) 2 Se 3 / CuSe the precursors follows

PAGE 118

118 one dimensional diffusion controlled transformation consistent with the obtained Avrami exponents.

PAGE 119

119 Table 3 1. Mole fraction of as deposited and annealed precursor as determined by ICP OES Atom Fraction Sample Cu In Ga Se Cu/In Se/Metal Glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se 0.1868 0.1924 0.0000 0.6208 0.970 1.64 Glass/Mo/ In 2 Se 3 / Cu 2 Se/Se 0.2566 0.2175 0.0000 0.5259 1.180 1.11 Glass/Mo/ (In,Ga) 2 Se 3 /CuSe 0.2304 0 .1907 0.0514 0.0514 0.951 1.12 Glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se Annealed 0.2309 0.2501 0.0000 0.5190 0.920 1.08 Glass/Mo/ In 2 Se 3 / Cu 2 Se/Se Annealed 0.2707 0.2547 0.0000 0.4746 1.06 0.90 Glass/Mo/ (In,Ga) 2 Se 3 /CuSe Annealed 0.1817 0.1480 0.0 383 0.6320 0.975 1.72

PAGE 120

120 Table 3 2 Phases identified by XRD after RTA for 2 min Sample RTA T ( o C) Phases Identified (XRD) Glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se 350 CIS, CuSe, InSe 370 CIS, CuSe, InSe 390 CIS,MoSe 2 Glass/Mo/ In 2 Se 3 / Cu 2 Se/S e 350 CIS, CuSe, InSe 370 CIS, MoSe 2

PAGE 121

121 Figure 3 1. Room temperature diffraction data of as deposited glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se

PAGE 122

122 Figure 3 2 Sequence of diffraction patterns during temperature ramp annealing of gl ass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se

PAGE 123

123 Figure 3 3 Se 2 partial pressure (atm) temperature diagram for the Cu Se system (Dotted lines indicate the pressure at peritectic and monotectic temperatures)

PAGE 124

124 Figure 3 4 High resolution room tem perature diffraction patterns of temperature ramp annealed precursor. A ) glass/Mo/ In 2 Se 3 /Cu 2 Se/Se B ) glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se A B

PAGE 125

125 Figure 3 5 Cross sectional and surface images of temperature r amp annealed precursor. A) cross se ctional image of glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se, B) surface image of glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se C) cross sectional of glass/Mo/ In 2 Se 3 /Cu 2 Se/Se D) surface image of glass/Mo/ In 2 Se 3 /Cu 2 Se/Se D C A B

PAGE 126

126 Figure 3 6 High resolution room temp erature XRD of as deposited glass/Mo/ In 2 Se 3 / Cu 2 Se/Se

PAGE 127

127 Figure 3 7 Sequence of diffraction patterns collected during temperature ramp annealing of glass/Mo/ In 2 Se 3 /Cu 2 Se/Se

PAGE 128

128 Figure 3 8 Sequence of diffraction patterns collected during temperature ramp annealing of glass/Mo/ In 2 Se 3 /Cu 2 Se/Se with selenium overpressure

PAGE 129

12 9 Figure 3 9. High resolution r oom temperature diffraction data of as deposited glass/Mo/ ( In ,Ga) 2 Se 3 /Cu Se

PAGE 130

130 Figure 3 10 Sequence of diffraction patterns collected during temperature ramp annealing of glass/Mo/ ( In ,Ga) 2 Se 3 /Cu Se

PAGE 131

131 Figure 3 11 Isothermal annealing d iffraction data of glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se precursor A) 260 o C, B) 270 o C, C) 290 o C, D) 310 o C A B C D

PAGE 132

132 Figure 3 12 Solid state growth model plot of glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se precursor A) Avrami model, B) parabolic model A B

PAGE 133

133 Figure 3 13 Arrhenius plot using rate constants obtained from Avrami and parabolic model for glass/Mo/ In 2 Se 3 / Cu Se+ Cu 2 Se/Se A) Avrami model, B) parabolic model A B

PAGE 134

134 Figure 3 14 Cross sectional and surface image of rapid thermal annealed glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se/Se precursor. A) cross sectional image B) surface image B A

PAGE 135

135 Figure 3 15 Isothermal annealing d iffraction data of glass/Mo/ In 2 Se 3 / Cu 2 Se/Se precursor A) 260 o C, B) 270 o C, C) 280 o C, D) 290 o C A B C D

PAGE 136

136 Figure 3 16 Solid state growth model plot of glass/Mo/ In 2 Se 3 / Cu 2 Se/Se precursor A) Avrami model, B) par abolic model B A

PAGE 137

137 Figure 3 17 Arrhenius plot using rate constants obtained from Avrami and parabolic model for glass/Mo/ In 2 Se 3 / Cu 2 Se/Se A) Avrami model, B) parabolic model A B

PAGE 138

138 Figure 3 18 Cross sectional and surface image of rapid thermal annealed glass/Mo/ In 2 Se 3 / Cu 2 Se/Se precursor. A) cross sectional image B) surface image B A

PAGE 139

139 Figure 3 19 Isothermal annealing d iffraction data of glass/Mo/ ( In ,Ga) 2 Se 3 / CuSe precursor A) 300 o C, B) 320 o C, C) 340 o C, D) 360 o C A B C D

PAGE 140

140 Figure 3 20 Solid state growth model plot of glass/Mo/ ( In ,Ga) 2 Se 3 / CuSe precursor A) Avrami model, B) parabolic model B A

PAGE 141

141 Figure 3 21 Arrhenius plot using rate constants obtained from Avrami and parabolic model for glass/Mo/ ( In ,Ga) 2 Se 3 / CuSe precursor

PAGE 142

142 CHAPTER 4 GALLIUM DISTRIBUTION STUDIES EMPLOYING BI LAYER METALLIC PRECURSORS Overview CuIn 1 x Ga x Se 2 based polycrystalline solid solutions are promising candidates for photovoltaic applications given the fact that the champion cell efficiency has exceeded 20 % by modifying the space charge region by tailoring the gallium distribution [38] A higher band gap at the CIGS surface increases the open circuit voltage of the device and the grading at the back contact increases the collection effi ciency due to back field Additionally, having a group III lattice concentration relative to a copper sub lattice facilitates the formation of OVC and pushes the metall urgical junction away from the hetero junction, thus reducing interfacial recombination. Increasing the gallium concentration increases the band gap to an optimum value of the solar spectrum; however the improvement in device efficiency has been realized o nly for a gallium Ga composition and distribution has been well supported by device modeling work [138, 139] Selenization of m etallic precursors is a widely used approach to produce high efficiency cells [140] After selenization, gallium rich region has been observed segregated towards the back contact. The gallium profile variation can be ac hieved intentionally during the growth process by manipulating the Ga/In ratio during deposition or unintentionally which occurs due to copper depletion and the difference in the reaction rates of indium and gallium. Gallium diffusion in CIS has been stud ied by Schroeder et al. and they suggest the dominant mechanism is vacancy hopping [141]

PAGE 143

143 Lundberg et al. have studied the diffusion of gallium and indium in both Cu rich and Cu poor CIS/CGS diffusion couple with the influence of sodium during growth. It is expected that gallium and indium vacancies under copper rich conditions controls the transport of group III metals. Additionally, the presence of a liquid phase increases the transport rates [142] Kyoung et.al have done a three step selenization using H 2 Se/Ar/H 2 Se and it was concluded that selenium vacancy occurs during annealing in argon which in turn creates V Ga and V In to maintain charge neutrality; hence promoting gallium diffusion [67, 143] Thus, a homogeneous distribution of gallium was achieved using three step selenization procedures. Furthermore, to lower the cost it is desirable to reduce the absorber thickness in conjunction with the gallium distributi on control. There has been limited information on pathways of gallium redistribution relative to the precursor structure dur ing se lenization. In this Chapter, reaction pathways and kinetics of selenization of bilayer metallic precursor grown under differe nt configurations are studied. The effect of selenium cap on gallium distribution is also reported. Experimental soda lime glass substrates using a molecular beam epitaxial rea ctor. The Ga/(Ga+In) ratio was around 0.31 and the Cu/III ratio was 0.92 as measured by inductively coupled plasma optical emission spectroscopy (ICP). The precursor sets were also coated with 0.6 m of selenium as a cap. The selenium was deposited at room temperature to ensure no formation of the solid solution in Cu Ga In system. Time resolved X ray system using a custom built stainless steel reactor. The detai ls of the system are given

PAGE 144

144 in Chapter 2. The sample was placed in a custom built stainless steel reactor with some selenium shots ( ~120 mg) and heated at low ramp rates of 10 20 o C/min. The temperature ramp study was performed from room temperature till 570 o C for selenization reactions. X ray data was collected for every 10 o C increment for approximately 2 minutes. Aluminum was used as an X ray transparent window and it should be noted that reflections from aluminum appear in the X ray spectrum. Nicke l could be an ideal choice as it is used to filter k radiation. However, with selenium present, the formation of nickel selenides are favored, which overlaps with the peaks of Cu In Se system. Prior to the temperature ramp, room temperature X ray diffraction was collected using high resolution X ray diff raction to identify the phases. The temperature ramp data collected is at low resolution; hence phases that appear in high resolution scan might not be visible in low resolution scans. Peak identification was ing the ICDD 2009 database. The sample characterization was done us ing scanning electron microscopy (SEM, JEOL JSM 6335F) for the microstructure and EDS (energy dispersive spectroscopy) line scans were done by TEM (Tecnai F30) to get the composition variat ion along the thickness direction. Temperature Ramp Annealing Glass/Mo/CuIn/CuGa Precursor The room temperature XRD of the as grown sample showed reflections of indium (In), metastable CuIn, Cu 9 Ga 4 a Cu 2 In solid solution, and molybdenum (Mo) ( black spec tra in Figure 4 1). The metastable phase has been observed previously during the deposition of a Cu/In bilayer at room temperature [144] A temperature ramp was done

PAGE 145

145 without selenium and XRD patterns were taken for 2 minutes at 5 o C temperature intervals to follow the reaction pathways of metallic precursors. The sample was placed on a ceramic p edestal instead of the stainless steel reactor. Forming gas was used a carrier gas and the ramp rate was set at 10 o C/min. Indium and the CuIn compound melted at 150 o C and the compound Cu 11 In 9 compound was evident at 160 o C. Traditionally, the Cu Ga allo y is deposited after molybdenum, followed by indium. The Cu 11 In 9 phase formation has been reported below 200 o C using the SAS method [120] The lattice parameter of Cu 9 Ga 4 shifted towards the higher value because of both temperature as well as mixing of the group III sub lattice. The diffusion pathway was followed till 400 o C after which the temperature was cooled down to room temperature. A final room temperature scan revealed an incomplete reaction giving textu red indium, a solid solution of Cu 9 (In x Ga 1 x ) 4 and traces of Cu 9 Ga 4 Figure 4 2 shows the temperature ramp sequence of the glass/Mo/CuIn/CuGa precursor. A high reso lution scan (Figure 4 3 ) was done at room temperature and the value of x was calculated f rom the 2 position using end members of Cu 9 In 4 and Cu 9 Ga 4 (the only end members available in the ICDD database). The composition of gallium in the solid solution was approximately 0.101 and 0.145 and the phase fraction was equally distributed. The solid solution corresponding to these gallium compositions can be designated as Cu 9 (In 0.77 Ga 0.23 ) 4 and Cu 9 (In 0.67 Ga 0.33 ) 4 The inset in Figure 4 3 shows the enlarged version of the two different gallium compositions. The final composition of the annealed precu rsor is the tie line connecting Cu 9 (In x Ga 1 x ) 4 and indium in the ternary phase relationship of Cu In Ga system [145] SEM images (Figure 4 4 A ) for annealed samples showed micron size particles (island) lying on a densified film structure (matrix). SEM

PAGE 146

146 spot EDS was done on both the island region and the matrix region and the co mposition of the individual elements were determined. The matrix and island region had similar composition of individual elements indicating a solid solution formation. The XRD results showed the presence of free indium; however its location could not be identified by this technique. TEM images (Figure 4 5 A ) showed micron size particles of solid solution of Cu In Ga and the line scan revealed the composition was uniform along the thickness direction based on relative intensity. The thickness of the meta llic layer was approximately 0.5 m and it is postulated that free indium was in the form of particles and was removed during the milling process. The free indium present can form binary selenides during selenization and can affect the reaction rate along with the gallium distribution. Glass/Mo/CuGa/CuIn Precursor The room temperature XRD of the as grown precursor had reflections of indium (In), Cu 9 Ga 4 a stable Cu 2 In solid solution and molybdenum (Mo) ( Figure 4 6 ). The precursor structure is similar to the one used in the SAS process except the top layer is a Cu In compound instead of only indium. A similar temperature ramp sequence was used as discussed for the previous precursor. Forming gas was used as a carrier gas to eliminate the formation of metallic oxides. Indium melted a t 150 o C and Cu 2 In disappeared at 160 o C. The formation of Cu 11 In 9 took place at 160 o C. It has been reported that the formation of Cu 11 In 9 from copper in liquid indium is a diffusion controlled process with activation energy of 16.9 kJ/ mole [146] The diffusion of indium and gallium were evident from temperature ramp studies until 400 o C as shown in Figure 4 7. The temperature was lowered to ro om temperature and a final scan was

PAGE 147

147 taken. The final product had a mixture of textured indium, Cu 9 (In x Ga 1 x ) 4 and Cu 9 Ga 4 similar to the previous precursor. The composition of gallium in the solid solution was 0.102 and 0.142 and the phase fractions were unequally distributed. This can also be designated as Cu 9 (In 0.77 Ga 0.23 ) 4 and Cu 9 (In 0.67 Ga 0.33 ) 4 solid solution The enlarged version of the solid solution is shown in inset of Figure 4 8 with unequal distribution of solid solution. The amount of free i ndium remaining in the sample was more when co mpared with glass/Mo/CuIn/CuGa SEM images (Figure 4 4 B ) showed matrix and island regions and spot EDS measurements were taken at both locations. The island region showed more indium composition than copper and gallium. However in matrix region, composition measurements suggest a solid solution formation consistent with the XRD results showing more phase fraction of solid solution and textured indium. Similar results were obtained from TEM EDS line scan meas urements showing uniform distribution of elements across the thickness direction and the amount of gallium in the solid solution is slightly more than the indium ( Figure 4 5 B) Indium, after melting, remains secluded from the molybdenum interface and gall ium out diffuses to form the solid solution with an increase in temperature. Once the temperature is brought back to room temperature, liquid indium starts to crystallize on the matrix of solid solution forming islands of indium particles. Table 4 1 comp ares the spot EDS measurements taken from SEM of temperature ramped annealed samples. Glass/Mo/CuIn/CuGa/Se Precursor The glass/Mo /CuIn/CuGa precursor was coated with 0.6 m of selenium and no crystalline phases of selenium were found in the room tempera ture XRD scan as shown in Figure 4 9 The motivation behind coating the precursor with a selenium cap was to

PAGE 148

148 see if a metallic solid solution formation or binary metal selenides formation was favorable at lower temperatures during the temperature ramp. It is also expected that selenium presence can affect the gallium redistribution and hence the properties of the grown material. Temperature ramp selenization (Figure 4 10 ) was carried out in a stainless steel reactor with excess selenium overpressure provi ded in the form of powder. Data was collected for every 10 o C step increment at a constant ramp rate of 10 o C/min. As observed previously indium and CuIn melted at 150 o C and the selenium cap, which was amorphous, began to crystallize around 100 o C. S elenium melted at its melting point of 220 o C providing an internal calibration for the newly designed system. It is understood from the temperature ram p experiments discussed for glass/Mo/CuIn/CuGa precursor the composition of the solid solution lies in the tie line connecting Cu 9 Ga 4 and indium. It is expected that liquid indium will be on the surface and with high selenium pressure, In Se binary compounds formation would be observed. However, no In Se binary compounds were observed below 300 o C It h as been reported in the SAS process that In Se and Cu Se binaries are formed at the surface and they react forming CIS at the surface creating a gradient in gallium distribution. The formation of a solid solution of Cu 9 (In x Ga 1 x ) 4 was preferred and after 300 o C, CuSe was formed in relatively small amount as shown in Figure 8 suggesting the solid solution is more stable and only excess Cu is selenized at that temperature. Perhaps the InSe is formed and is in glassy state which is not detected by X rays T he first peak of CIGS was observed at 300 o C whereas no CGS formation was observed. The reaction was continued till 570 o C and then the temperature was brought back to room temperature. No MoSe 2 peak was observed in the final room temperature scan.

PAGE 149

149 A fi nal high resolution room temperature ( Figure 4 11 ) scan was taken in conjunction with Rietveld refinement for temperature dependent (112) reflections to determin e the gallium composition. The ( 112 ) r eflection is shown in inset of Figure 4 1 1 to show the g allium redistribution. The gallium composition estimated from this analysis yielded a value of 44 to 53 at%. The following reaction sequence is observed during a temperature ramp, (4 1) (4 2) (4 3) (4 4) (4 5) TEM EDS line scan measurements (Figure 4 1 2 ) showed the formation of MoSe 2, although it was not detected by X rays indicating the MoSe 2 formed wa s amorphous in nature. Gallium was distributed more towards the back contact and indium was distributed more towards the surface, suggesting a lower band gap formed near the sur face. Comparing the XRD (Figure 4 1 1 ) and EDS data, it can be confirmed that C uIn 0.5 Ga 0.5 Se 2 is formed near the back contact. The grains were continuous for ~1 m from the surface and towards the back contact the grain size distribution exists. This was also verified independently by cross sectional scanning electron microscope ima ge as shown in Figure s 4 13A and 4 13B It is been reported that the gallium inhomogenities in CIGS become more pronounced during cooling, undergoing de mixing of group III sub lattice [147]

PAGE 150

150 Glass/Mo/CuGa/CuIn/Se Precursor Temperature ramp annealing was performed for glass/Mo/CuGa/CuIn coated with 0.6 m of a selenium cap. The room temperature pattern of as deposited precursor is shown in Figure 4 1 4 X ray diffraction data was collected for every 10 o C at a constant ramp rate of 10 o C/min with selenium overpressure supplied in the form of powder (Fig ure 4 15 ) Indium melted at this melting point and the peak shifted left from 80 o C because of sample displacement (note a change in the height and displacement of molybdenum peak). Crystallization of selenium initiated at 100 o C and selenium melting took place at its melting point (220 o C). The formation temperature of Cu 11 In 9 compound was similar to the one as previously observed Diffusion of indium and gallium progressed till 240 o C. At 250 o C, CuSe 2 and CuSe are formed in relatively smaller amount s, indicatin g fre e copper at the surface. CIS formation was observed at 260 o C which is consistent with the reported temperatures [123] It is believed that binary In Se formed is in a glassy state and is not detected by X rays, pro moting CIS formation. The CIS formed acts as a diffusion barrier for selenium to react with Cu 9 Ga 4 forming CGS. At 340 o C, CGS formation took place forming CIS/CGS diffusion couple as shown in Figure 4 1 5 The formation temperature of CGS is higher than the values reported in the literature [125] The reaction was con tinued till 570 o C following the diffusion pathway of CIS/CGS, after which the sample was brought back to room temperature. A high resolution scan was taken ( Figure 4 16 ) and the Rietveld refinement was performed and lattice parameters were calculated for the ( 112 ) orientation of CIGS. The gallium composition varied form 39 53% from the obtained

PAGE 151

151 lattice constants as shown in the inset of Figure 4 16 The following reaction sequence was observed during temperature ramp, (4 6) (4 7) (4 8) (4 9) (4 10) (4 11) It was expected that gallium would move towards the surface and fo rm CuIn x Ga 1 x Se 2 However, EDS line spectra (Figure 4 1 7 ) showed more gallium bowing towards the back contact i ndicating little out diffusion of gallium during the selenization reaction. The reaction rate of CIS is higher than the analog CGS and from the EDS s pectra, it is believed that indium near the surface reacts to form CIS and the r eaction front moves towards the gallium rich region forming CuIn x Ga 1 x Se 2 This is obser ved in the form of small grains as shown in the SEM image in Figure s 4 1 3 C and 4 1 3D It has been shown that phase segregation in CIGS based system irrespective of deposition s equence of the precursor is not dependent on diffusion rates of indium and gallium, but dependent on reaction rates of CIS and CGS [67] Glas s/Mo/CuIn/CuGa Precursor + Se V apor Temperature ramp selenization was done for the precursor glass/Mo/CuIn/CuGa but now with selenium vapor provided in the form of powder. No selenium cap was present during selenization. The seleniu m vapor pressure is temperature dependent

PAGE 152

152 and increases exponentially with temperature. The aim of this experiment was to see if there was any preferential reaction for indium during selenization reactions. Ramp was done at a constant rate of 10 o C/min an d X ray data was recorded for every 10 o C step increment. As observed in Figure 4 1 8 In and CuIn melted at 150 o C followed by formation of Cu 11 In 9 CIGS formation took place at 300 o C along with the formation of CuSe. No CGS formed during the temperat ure ramp. The temperature ramp was continued until 570 o C to see if there was any phase separation. No phases other than single phase CIGS was observed. MoSe 2 formation was observed at 450 o C indicating the completion of reaction consistent with the rep orted values in the literature. No other major changes were observed in the pathway; however, the gallium distribution was different than the one with the selenium cap as observed in the high resolution scan shown in spectrum F igure 4 1 9 Rietveld refine ment of the room temperature (112) reflection gave the gallium composition in a narrower range (44 45%) suggesting good mixing on group III sub lattice as shown in inset of Figure 4 1 9 Also, the peak corresponding to ( 112 ) orientations can be deconvolute d to CuIn 0.7 Ga 0.3 Se 2 and predominately CuIn 0.55 Ga 0.45 Se 2 It has been reported that for gallium homogenization a selenium free annealing step is required in an argon atmosphere for longer retention time [119] A T EM image (Figure 4 20 ) clearly showed regions of 300 400 nm of MoSe 2 formed at the interface of Mo and CIGS consistent with the observation of MoSe 2 peak in the XRD spectra. EDS line scan measurements were done in the continuous grain region of CIGS keepi ng MoSe 2 as the origin. EDS line spectra showed that both gallium and indium were distributed uniformly and there was no accumulation of gallium towards the back contact consistent with the results obtained from Rietveld refinement.

PAGE 153

153 The interface relatio nship between Mo MoSe 2 and MoSe 2 CIGS could not be determined because of the amorphous nature of MoSe 2 The surface of the absorber was highly rough and the little grain growth was observed as shown in Figure 4 21 A and 4 21B The following reactions were observed during the temperature ramp, (4 12) (4 13) (4 14) (4 15) (4 16) Glass/Mo/CuGa/CuIn Precursor + Se V apor Similar temperature ramp experiments were done for the glass/Mo/CuGa/CuIn precursor with selenium vapor as a function of temperature with a ramp rate of 10 o C/min. Figure 4 22 shows the temperature ramp selenization of glass/Mo/CuGa/CuIn precu rsor. As observed earlier, In melted and Cu 2 In transformed to Cu 11 In 9 followed by interdiffusion of indium and gallium. CIGS formation took place at 280 o C alo ng with the formation of CuSe. The reaction was complete at 560 o C evidenced by MoSe 2 formatio n. No CGS formation took place because of initial mixing of group III sub lattice till 280 o C, after which selenization reactions become more dominant. Co mparing with the results of previous sections the gallium distribution was almost uniform with a lowe r Ga/ (Ga+In) ratio near the surface as obt ained from TEM EDS ( Figure 4 23 ) Indium was found to be more towards the surface and gallium was nearly uniform, thus providing a graded structure. Rietveld refinement was done on the ( 112 ) orientation

PAGE 154

154 taken from a high resolution X ray scan (Figure 4 24 ) and revealed 44 50 % gallium incorporation in the group III sublattice. The grain growth was non uniform as observed in Figure 4 21 C and 4 21D The following reaction was observed during the temperature ra mp, (4 17) (4 18) (4 19) (4 20) (4 21) Isothermal Annealing and Reaction Kinetics Time resolved high temperature X ray diffraction data was collected for a set of temperatures selected from the temperature ramp experiments. The data was collected for four different temperatures for the samples discussed in temperature ramp experiments. Isothermal temp erature was achieved by manually increasing the temperature instead of using the controller. This was because the maximum ramp rate that could be used in PAN a lytical system was 40 o C/min, and to avoid pre reaction of precursors, it is necessary to reach th e isothermal temperature very quickly. Ramp rates in the order of 100 120 o C/min could be achieved by manual increases of the temperature instead of controller. The 2 scan range was reduced from 21 o 32 o to have a high resolution scan of the ( 112 ) orie ntation. The data was collected in time mode and each scan time was 95 seconds. After 80 minutes, the temperature was

PAGE 155

155 increased to 570 o C to complete the reaction and the data was collected until MoSe 2 formed; as it has been customary to use MoSe 2 format ion as end point detection for completion of the reaction. The fractional conversion was obtained by calculating the area under the curve using normalization, assuming complete conversion at higher temperature and no change in the texture during isotherma l annealing. Once the fractional conversion was obtained, rate parameters in terms of rate constant and activation energy were calculated using the Avrami model as discussed in Chapter 2. Figure 4 25 to 4 28 compares the isothermal selenization of precurs ors discussed in temperature ramp annealing experiments The rate constant obtained from the model was plotted (Figure 4 29 ) using the Arrhenius equation, (4 22) As seen in the Figure 4 28 the kinetics of the sample wi th and without selenium cap are different. According to the model, the samples coated with a selenium cap had a higher rate than the one without a selenium cap. One of possible reasons for higher activation energy for samples without a selenium cap is th e mixing of group III sub lattice. This was observed in the TEM results, where gallium was distributed uniformly when selenization was done without a selenium cap. Summary Reaction pathways were followed for selenization of bilayer metallic precursors u sing in situ high temperature X ray diffraction. The effect of a selenium cap on the gallium distribution was elucidated. The pathways showed the formation of Cu 11 In 9 which then undergoes group III sub lattice diffusion transformation with Cu 9 Ga 4 to form

PAGE 156

156 a solid solution of Cu 9 (In x Ga 1 x ) 4 The value of x and phase fraction depends on the order of deposition of the precursor. CuSe formation was minimal indicating the solid solution is more stable than the selenides. No binary phases of In Se and Ga Se we re detected in the scan. TEM EDS results revealed that selenization done without a cap gives uniform distribution of gallium irrespective to the order of deposition, thus forming a graded structure. High temperature annealing was not required to get a un iform gallium distribution. Reaction kinetics for the formation of CIGS was obtained by conducting isothermal studies. For the glass/Mo/CuIn/CuGa precursor, the activation energy estimated from the isothermal studies give values of 76(14) and 107 (15) kJ/mole without and with selenium cap, respectively. For the glass/Mo/CuGa/CuIn precursor, the estimated activation energy yielded values of 93(4) and 101 (9) kJ/mole without and with selenium cap, respectively. A higher activation energy was required for selenization without a selenium cap because of the initial mixing of group III sub lattice

PAGE 157

157 Figure 4 1 Room temperature X ray diffraction plot of as deposited g lass/Mo/CuIn/CuGa precursor

PAGE 158

158 Figure 4 2 Temperature ramp sequence o f glass/Mo/CuIn/CuGa precursor in forming gas

PAGE 159

159 Figure 4 3 Room temperature X ray diffraction plot of temperature ramp annealed precursor of glass/Mo/CuIn/CuGa (inset shows the solid solution distribution)

PAGE 160

160 Figure 4 4 SEM images o f temperature ramp annealed precursors. A) surface image of temperature ramp annealed precursor of glass/Mo/CuIn/CuGa, B) surface image of temperature ramp annealed precursor of glass/Mo/CuGa/CuIn A B

PAGE 161

161 Figure 4 5 EDS profile and TEM image of temperature ramp annealed precursors. A) EDS line profile and TEM image of temperature ramp annealed glass/Mo/CuIn/CuGa, B) EDS line profile and TEM image of temperature ramp annealed glass/Mo/CuGa/CuIn A B

PAGE 162

162 Figure 4 6 Room temperature X ray diffractio n plot of as deposited glass/Mo/CuGa/CuIn precursor

PAGE 163

163 Figure 4 7 Temperature ramp sequence of glass/Mo/CuGa/CuIn precursor in forming gas

PAGE 164

164 Figure 4 8 Room temperature X ray diffraction plot of temperature ramp annealed precursor of gl ass/Mo/CuGa/CuIn (inset shows the solid solution distribution)

PAGE 165

165 Figure 4 9 Room temperature X ray diffraction plot of as deposited glass/Mo/CuIn/CuGa/Se precursor

PAGE 166

166 Figure 4 10 Temperature ramp selenization sequence of glass/Mo/CuIn /Cu Ga/Se precursor

PAGE 167

167 Figure 4 11 Room temperature X ray diffraction plot of temperature ramp an nealed precursor of glass/Mo/CuIn/CuGa/Se precursor (inset shows the gallium redistribution in (112) orientation )

PAGE 168

168 Figure 4 12 EDS profile and TEM image of temperature ramp annealed glass/Mo/CuIn/CuGa/Se precursor

PAGE 169

169 Figure 4 13 SEM images of temperature ramp annealed precursors. A) surface and cross sectional image of temperature ramp annealed precursor of glass/Mo/CuIn/CuGa/Se, B) s urface and cross sectional image of temperature ramp annealed precursor of glass/Mo/CuGa/CuIn/Se A C D B

PAGE 170

170 Figure 4 14 Room temperature X ray diffraction plot of as deposited glass/Mo/CuGa/CuIn/Se precursor

PAGE 171

171 Figure 4 15 Temperature ramp seleniz ation sequence of glass/Mo/CuGa/CuIn/Se precursor

PAGE 172

172 Figure 4 16 Room temperature X ray diffraction plot of temperature ramp annealed precursor of glass/Mo/CuGa/CuIn/Se precursor (inset shows the gallium redistribution in (112) orientation)

PAGE 173

173 Figure 4 17 EDS profile and TEM image of temperature ramp annealed glass /Mo/CuGa/CuIn/Se precursor

PAGE 174

174 Figure 4 18 Temperature ramp selenization sequence of glass/Mo/CuIn/CuGa precursor

PAGE 175

175 Figure 4 1 9 Room temperature X ra y diffraction plot of temperature ram p selenized precursor of glass/Mo/Cu In/CuGa (inset shows the gallium redistribution in (112) orientation)

PAGE 176

176 Figure 4 20 EDS profile and TEM image of temperature ramp selenized glass/Mo/CuIn/CuGa precursor

PAGE 177

177 Figure 4 21 SEM images of temperature ramp annealed precursors. A) surface and cross sectional image of temperature ramp selenized precursor of glass/Mo/CuIn/CuGa, B) surface and cross sectional image of temperature ram selenized precursor of glass/Mo/CuGa/CuIn A C B D

PAGE 178

178 Figure 4 22 Temperature ramp selenization sequence of glass/Mo/CuGa/CuIn precursor

PAGE 179

179 Figure 4 23 EDS profile and TEM image of temperature ramp selenized glass/Mo/CuGa/CuIn precursor

PAGE 180

180 Figure 4 24 Roo m temperature X ray diffraction plot of temperature ramp selenized precursor of glass/Mo/CuGa/CuIn (inset shows the gallium redistribution in (112) orientation)

PAGE 181

181 Figure 4 25 Isothermal annealing of glass/Mo/CuIn/CuGa/Se precursor with sel enium overpressure at different temperatures. A)270 o C, B) 285 o C, C) 300 o C, D) 320 o C A C B D

PAGE 182

182 Figure 4 26 Isothermal annealing of glass/Mo/CuGa/CuIn/Se precursor with selenium overpressure at different temperatures. A)275 o C, B) 290 o C, C) 300 o C, D) 315 o C C A D B

PAGE 183

183 Figure 4 27 Isothermal selenization of glass/Mo/CuIn/CuGa precursor with selenium vapor at different temperatures. A)260 o C, B) 280 o C, C) 300 o C, D) 320 o C A C B D

PAGE 184

184 Figure 4 28 Isothermal selenization of glass/Mo/Cu Ga/CuIn precursor with selenium vapor at different temperatures A)260 o C, B) 280 o C, C) 300 o C, D) 320 o C C A D B

PAGE 185

185 Figure 4 29. Arrhenius plot plotting rate constants obtained from isothermal studies

PAGE 186

186 CHAPTER 5 REACTION PATHWAYS AN D KINETIC S OF STACKED ELEMENTAL LA YER FORMING CUIN X GA 1 X SE 2 Overview Thin film solar cells based on Cu(In, Ga)Se 2 is a promising photovoltaic absorber material due to its excellent properties such as direct band gap, high absorption coefficient, and high power co nversion efficiency which is 20 % close to polycrystalline silicon cells [38] Despite having all the merits, the market penetration of CIGS thin film solar cells is still an issue. The challenge is to get high process yield with columnar grain growth to attain high efficien cy in a commercial scale. High efficiency cells require vacuum based approach causing a barrier for lowering $/W p Different approaches such as electro deposition, reactive sputtering, selenization of metallic alloys, coating of oxide nanoparticle/reduct ion followed by rapid thermal annealing have been tried for lowering the cost [92, 148, 149] Though high efficiency cells were obtained by co evaporation, the process is difficult to scale up due to non uniformity and material utilization. It is necessary to maintain exact stoichiometry throughout the film for better performance. The phase relationship in Cu In Se system at 500 C shows a wide homogeneity range in composition of CIS and there exist 8 different solid phases as well as selenium rich liquid in equilibrium with CIS [150 152] A small change in the chemical potential changes the equilibrium relationship of CIS and hence its properties. Irrespective of the process used for growth, a detailed understanding of CIGS formation, together with equilibrium phase diagram is necessary for ease in scale up and process optimization. The re exists no phase diagrams of quaternary systems, but it is well known fact that addition of gallium or sulfur changes the final properties of the absorber

PAGE 187

187 [153] It is also known that mos t of the solid state reactions are diffusion controlled. One could increase the reaction rates by raising the temperature. The chances of finding the compound that exist at low temperature would be eliminated in such case. Other way to speed up the proc ess is to reduce the diffusion distances by intimately mixing the precursors such as nanoparticles. This would require larger retention times to remove the solvents/binders. The best way would be identify new precursor structures where activation energy could be reduced by passing through fluid like phase leading to good films with uniform composition and better grain growth. Of several approaches discussed above, stacked elemental approach would have the potential for efficient low cost in continuous ro ll to roll processing for large area substrates. One of the advantages of stacked elemental layer in terms of large area processing are controlling the flux of elements separately instead of controlling the flux required for forming the binary or ternary phases. This approach has been tried by many research groups and has reported efficiency over 12% [154 156] This Chapter discusses the reaction pathways and kinetics of selenization of stacked elemental layers. I n one of the approach, copper was deposited first followed by gallium, indium and seleniu m. In other approach, gallium was deposited first, followed by indium, copper and selenium. The rationale behind this approach was to study the effect the nucleation on the growth kinetics of CIGS. Experimental Precursor films were prepared by evaporation in which effusion cells are employed to generate elemental vapors in high vacuum (10 7 to 10 8 Torr) Elemental stacked precursors were deposited on molybdenum co ated on stainless steel with ruthenium as

PAGE 188

188 the diffusion barrier material to prevent incorporation of chromium and iron in to the absorber. The precursors were deposited at room temperature to ensure no reaction between them. The thickness of copper, gall ium, indium and selenium deposited were 1300, 1000, 2160 and 10000 for both the samples. The phase transformation studies were done in both Scintag and PANlytical system. The details of the system have been described in Chapter 2. Inductively coupled pla sma optical emission spectroscopy (ICP OES, Perkin Elmer Plasma 3200) was used to measure the overall film composition. Scanning electron microscop y (SEM, JEOL JSM 6335F) was used to get the surface morphology and cross section of thin films. SEM EDS mea surements were done to get the composition information from the surface. EDS measurements were microscopy (TEM, JEOL 2010 CF) was used to get electron diffraction patterns fo r quenched and annealed samples. The crystalline phases were identified by Philips ray diffraction. The data obtained from high temperature X ray 2 009 database. Temperature Ramp Annealing Glass/Mo/Cu/Ga/In/Se Precursor Phase evolution of glass/Mo/Cu/Ga/In/Se precursor was studied during The atomic composition of the precursor was determined by ICP OES as shown in Table 5 1. The system was purged with nitrogen to remove residual oxygen from the system. Before temperature ramp, room temperature X ray diffraction was done to

PAGE 189

189 identify the initial phases present after dep osition. Reflections of indium (In), Cu 2 Se, Mo, CuIn compound and Cu 9 Ga 4 phases were identified in the precursor as shown in Figure 5 1 A Molybdenum was polycrystalline and had (002) preferred orientation significantly different from the films deposited on the glass as observed in previous Chapters. This is because glass is amorphous, whereas stainless steel has texture which causes molybdenum to have preferred orientation. Though the deposition was done at room temperature, the nominal temperature of s ubstrate was higher than room temperature due to radiation from the sources as evidenced by the crystalline phases in room temperature scan It was interesting to note the formation of Cu 2 Se phase in this particular precursor even when copper is covered wi th thin layers of gallium and indium. SEM of as pre annealed samples showed flower like crystals on the surface as shown in Figure 5 2 A TEM image ( F igure 5 2 B) of the same precursor showed that flower like crystals were present throughout the thickness of the film. SEM EDS measurements were done on this flower like crystals and the results indicated it as Cu 2 Se in accordance with XRD. The isothermal section of Cu In Ga system at 350 C as reported by purwins et al. shows that with Ga/(Ga+In) ratio of 0.37, the only phases present are Cu 9 Ga 4 and indium consistent with observation of the phases in room temperature X ray diffraction [145] There are also reports for CuIn metastable phase during Cu In and Cu In Ga preparation [157, 158] Temperature ramp was performed from room temperature till 600 C with overpressure of selenium and data was taken for 2 min (2 20 to 54 ) at every 10 C interval. Parallel beam optics was used to eliminate the graphite peak from the dome. It should be noted that using such optics the intensity drops down because of absorption in graphite which accounts for

PAGE 190

190 approximately 60 %. The background pressure of selenium in the chamber was estimated to be 1 torr approximately. The initial ramp rate was set at 20 C /min to avoid overshoot of temperature and after tempera ture reached 150 C, heating rate was increased to 40 C /min Figure 5 3 shows the temperature ramp annealing of the precursor with selenium overpressure. Around 100 C, selenium begins to crystallize and after 220 C selenium peak disappears consistent with the melting point of selenium (221 C). It has been reported that activation energy for crystallization of amorphous selenium is 77(6) kJ/mole by isot hermal DSC analysis. Also the A vrami exponent increases with increase in temperature indicating se lenium follows nearly three dimensional growth [159] The crystallization velocities are anisotropic and are l ower through the film thickness and higher lateral across the substrate. Indium peak disappears after 150 C consistent with the thermodynamic melting temperature of indium (156.5 C). The scans were taken at every 10 C step interval and at 160 C no peak is observed for indium indicating accuracy of temperature measurement in the system. At 150 C, selenium reacts with indium to form In 4 Se 3 Hergert et al. have reported the In 4 Se 3 is the first selenide of indium formed during RTP of stacked elemental precursors [160] The phase In 4 Se 3 is on indium rich side of In Se phase diagram. The vapor pressure of selenium in its solid state is lower and diffusion of selenium across liquid indium is also small. Hence selenium poor phase, In 4 Se 3 is formed first by solid state reaction of liquid indium and crystallizin g selenium. Once the selenium melts, two different reactions are observed. The In 4 Se 3 reacts with liquid

PAGE 191

191 selenium to form (In, Ga) Se and also Cu 2 Se reacts with liquid selenium forming copper perselenide (CuSe 2 ). The In 4 Se 3 going to (In, Ga) Se instead of n 2 Se 3 is because of competitive reactions of indium and copper with selenium. Liu et al. has observed the formation of In 2 Se 3 from In 4 Se 3 using resistivity change during temperature ramp annealing [161] In the presence of copper, the chemical potential of selenium is low and hence the formation of (In, Ga) Se. The intensity of Cu 2 Se drops ste adily and disappears completely at 250 C. CIS starts to form at 260 C followed by formation of CGS 310 C. The pathway for CGS formation is different from CIS formation. For CGS formation, first a sub lattice of Cu 2 Se is formed where gallium atoms have good solubility. This was proved i n the last study done by Kim et al. during reaction pathway observation of glass/GaSe/CuSe precursor strucuture [125] The reason CGS forming later is because the second peritectic reaction involving formation of Cu 2 x Se from CuSe takes place at a temperature higher than formation of CIS at 260 C. The peritectic decomp osition temperature of CuSe 2 is 331.8 C giving CuSe and a selenium rich liquid. CuSe 2 decomposed at 260 C earlier than the peritectic temperature forming CuSe and selenium rich liquid. With selenium loss, peritectic reactions can occur at lower tem perature than those predicted by the phase diagram. The decrease in the peritectic temperature indicates a loss of selenium from the system. The CuSe phase decomposed to Cu 2 x Se at 330 C and selenium rich liquid again lower than the second peritectic reaction (379.5 C). It is also noted that rate of formation of CIS is very rapid due to change in the reaction mechanism. It is well known fact that the there exists a liquid like phase in the second stage of NREL 3 stage process. Such rapid reactions

PAGE 192

192 are attributed to liquid assisted growth which increases the grain size of CIGS by vapor liquid solid growth mechanism. The CIS CGS diffusion couple is formed after formation of CGS in which indium and gallium inter diffuse to form CuIn 1 x Ga x Se 2 at 550 C. The following reactions take place during temperature ramp, (5 1) (5 2) (5 3) (5 4) (5 5) (5 6) (5 7) (5 8) (5 9) (5 10) (5 11) (5 12) The reactio n was not complete as Cu 9 Ga 4 phase existed till the end. Similar observation was made by Hanket et al. during two step process involving selenization and sulfurization [162] This phase was found towards the back contact next to molybdenum. These findings are very important in terms of large ar ea production, because different starting composition of the precursor can cause a shift in the three

PAGE 193

193 phase field of Cu In Ga system and hence cause inhomogenities of the absorber material. A high resolution XRD scan was taken after the temperature ramp a nd phase fractions were calculated from intensity ratios. Figure 5 4 A shows the room temperature XRD of temperat ure ramp annealed precursor The amount of Cu 9 Ga 4 phase remaine d after selenization was approximately 0.6 %. No secondary phase Cu 2 x S e was detected consistent with the fact the as deposited samples were copper poor. The composition of gallium in CuIn 1 x Ga x Se 2 law [163] The 2 value of pure CIS (112) and CGS (112) are 26.624 and 27.746 respectively. The value of x can be obtained from 2 value of CuIn 1 x Ga x Se 2 (112) as shown below. (5 13) Where 26.895 correspond to 2 value of CuIn .75 Ga .25 Se 2 and y corresponds to measured value of 2 by XRD. The value of x calculated from above equation yields x (Ga)=0.35 for temperature ramp annealed precursor B. Hi g h resolution TEM image was taken for samples to further understand the gr ain structure. T he grain size was small as observed by lots of grain boundaries indicating a de crease in the crystallite size. Single crystal patterns were obtained and most of the grains were (112) oriented as shown in Figure 5 5 S EM images showed the annealed precursor had rough surface and were similar to previously observed morphologies of stacked elemental layer ( Figure s 5 6 A and 5 6 B ) Samples were further analyzed by auger electron spectroscopy (AES) to get the distribution of gallium across the film thickness. Initial survey was done and elements were identified. Sputtering was done by 3 point method and data were

PAGE 194

194 collected after sputtering The surface was indium rich a s observed in the depth profile. The composition of selenium was almost constant throughout the film. Gallium distribut ion was more uniform along the growth direction indicating a higher bang gap absorber structure. Figure 5 7 shows the AES depth profile of annealed sample under selenium overpressure. Similar experiments we re carried out without selenium overpressure in Scintag system. A different reaction pathway was observed compared with overpressure of selenium. Indium melted lower than its melting point as observed previously. Selenium crystallization was similar to the one observed with selenium overpressure. Selenium reacts with indium forming In 4 Se 3 which decomposes to InSe as observed previously. The formation of CIS takes place at 260 C by reaction of CuSe and InSe. CuSe forms at 280 C by decomposition of CuSe 2 The only difference observed during temperature ramp with no selenium overpressure was the formation of In 2 Se 3 and Cu 4 In metallic compounds Figure 5 8 shows the tempera ture ramp sequence with no selenium overpressure. Glass/Mo/Ga/In/Cu/Se Precursor Phase evolution of glass/Mo/Ga/In/Cu/Se precursor was studied in PANlytical The initial composition of the film was determined using ICP OES as shown in Table 5 1. The room temperature X ray diffraction showed all the reflections similar to the previous precursor except the metastable CuIn was detected as Cu 2 In. Molybdenum had (002) preferred orientatio n as observed earlier. Figure 5 1 B shows the room temperature XRD of as deposi ted precursor. SEM (Figure 5 2 C ) images showed crystals present on the surface of the precursor which were detected as Cu 2 Se from SEM EDS

PAGE 195

195 measurements. The similar data collect ion receipe followed for previous precursor was used for this precursor. Figure 5 9 shows the temperature ramp annealing of glass/Mo/Ga/In/Cu/Se precursor with selenium overpressure. No major difference in pathways was observed except for recrystallizati on as observed by sharp peaks in the XRD spectrum compared with previous precursor. All the reactions discussed above is valid for this precursor too. The precursor sample was quenched during temperature ramp annealing at 280 C for 25 minutes with seleni um overpressure to identify the phases present with electron diffraction. Figure 5 1 0 shows the electron diffraction pattern and ring patterns were obtained. The rings 1,3,5,8 and 9 corresponds to polycrystalline CIS and and CIS, and the rings 2 and 4 co rresponds to CuSe. The d spacings of ring 6 and 7 matched the d spacing of InSe. Spots were observed for CuSe 2 consistent with the results obtained from high temperature XRD. The reaction was not complete as Cu 9 Ga 4 was observed in the high resolution sc an. The phase fraction calculated from intensity ratios accounted 1.4% for Cu 9 Ga 4 phase. gallium in group III sub lattice. The grains were smooth for this annealed precursor compar ed with the precursor having copper deposited first as shown in Figures 5 6 C and D Gallium distribution determined from AES showed more gallium segregated towards back contact thus causing gradient in gallium distr ibution as well as the band gap as sho wn in Figure 5 11. Isothermal Annealing and Reaction Kinetics In situ high energy powder diffraction was employed by Brummer et.al to determine the phase transformation of metal selenium binaries and elementally stacked

PAGE 196

196 layers of glass/Cu/In/Se [117] They reported only qualitative information with respect to phase transformation and its tempe rature. Wolf et.al has reported activation energy in the order of 160 kJ/mole for major selenization reaction by calorimetric studies [164] Using time resolved high temperature XRD measurements, growth rate and kinetic parameters were computed at isothermal tem peratures. Isothermal selenization was performed at selected temperature between 290 400 C for both the precursor discussed in temperature. Figure 5 12 shows the is othermal scan taken for SS/Ru/Mo/Ga/In/Cu/Se at 400 C. The scan time for each time interval was approximately 62 sec and data was collected for approximately an hour. The 2 sc an range was selected from 23 34 since the major reflection of CIGS (112) lies within this range. The fractional reactions of the product was estimated using normalized 112 peak area assuming maximum peak area represents complete reaction. Avrami soli d state growth model was used to get the kinetic information in terms of rate constant and activation energy [165] as discussed in Chapter 2 Figure 5 13 compares the A vrami plot of precursors discussed during temperature ramp selenization. A decrease in the local avrami exponent is due to two reasons; inhomogeneous distribution of nuclei and there exists an interdiffusion of indium and gallium atoms at elevated temperatures. The inhomogeneous nuclei distribution directly affects the grain size and its distribution. It is there very important to control the nucleation process to get a good gr ain structure. This type of behavior has been previously observed for other systems such as metallic glasses [166] (5 14)

PAGE 197

197 Figure 5 14 compares the Arrhenius plot of precursors discussed above The rate constants obtained were plotted using Arrhenius equation and activation energy of 78 (6) (copper first ) and 116 (5) ( Gallium first) kJ/mole was obtained Table 5 2 shows the rate constants obtained from Avrami model at different temperature. Summary Phase transformations were studied for stacked elemental layer precursor structure with change in order of deposition of copper using in s itu high temperature X ray diffraction with and without selenium overpressure. No major difference was observed in the pathways between the two different structures as the initial phases present in both the elemental stacks were the same. With selenium o verpressure, the reaction does not go to completion as Cu 9 Ga 4 was found to be present along with the product phase. With no selenium overpressure, a lot of mixed phases such In 2 Se 3 Cu 4 In, Cu 9 Ga 4 were present with product phase. Isothermal studies were p erformed without selenium overpressure and kinetics of formation of CuIn 1 x Ga x Se 2 was followed. The data was fitted with Avrami solid state growth model and activation energy in the order of 78 (6) [copper first] and 116 (5) [gallium first] kJ/mole was ob tained by changing the order of deposition of precursors L ower Avrami exponents can be explained because of inhomogeneous distribution of nuclei.

PAGE 198

198 Table 5 1 ICP mole fraction of precursors before and after annealing Precursor As Deposited Annealed Cu In Ga Se Cu In Ga Se Cu/Ga/In/Se 0.2075 0.2511 0.1205 0.4209 0.1606 0.1875 0.1164 0.5355 Ga/In/Cu/Se 0.2004 0.2417 0.1453 0.4126 0.1629 0.2271 0.0978 0.5122

PAGE 199

199 T able 5 2. Rate constants of precursors obtained from Avrami model Precursor Avrami Model Temperature ( C ) k10 5 (1/s) E a (kJ/mole) Cu/Ga/In/Se 290 315 340 370 400 33 61.6 150.5 156.7 645.4 78 (6) Ga/In/Cu/Se 290 315 340 370 400 8.78 59.8 264 258.5 670.7 116 (5)

PAGE 200

200 Figure 5 1 Room temperatur e X ray diffraction pattern of as deposited precursor. A) SS/Ru/Mo/Cu/Ga/In/Se, B) SS/Ru/Mo/Ga/In/Cu/Se A B

PAGE 201

201 Figure 5 2 SEM and TEM imag e of as deposited precursor. A) SEM image of the sur face of SS/Ru/Mo/Cu/ Ga/In/Se B)TEM image of SS/Ru/Mo/Cu/Ga/In/Se, C) SEM image of SS/Ru/Mo/Ga/In/Cu/Se, D) TEM image of SS/Ru/Mo/Ga/In/Cu/Se A Ga/In/Cu/Se Mo Cu 2 Se Cu 2 Se Mo Cu/In/Ga/Se B C D

PAGE 202

202 Figure 5 3 Temperature ramp sequence during annealing of SS/Ru/Mo/Cu/Ga/In/Se with overpressure of selenium

PAGE 203

203 Figure 5 4 Room temperature X ray diffraction pattern of temperature ramp annealed precursor. A) SS/Ru/Mo/Cu/Ga/In/Se, B) SS/Ru/Mo/Ga/In/Cu/Se A B

PAGE 204

204 Figure 5 5 Selected area diffraction pattern of temperature ramp annealed SS/Ru/Mo/Cu/In/Ga/Se p recursor

PAGE 205

205 Figure 5 6 SEM images of temperature ramp annealed precursors with selenium overpressure. A) surface image of SS/Ru/Mo/Cu/Ga/In/Se, B) cross sectional image of SS/Ru/Mo/Cu/Ga/In/Se, C) surface image of SS/Ru/Mo/Ga/In/Cu/Se, D) c ross s ectional image of SS/Ru/Mo/Ga/In/Cu/Se A B C D

PAGE 206

206 Figure 5 7 Auger depth profile of temperature ramp annealed precursor of SS/Ru/Mo/Cu/Ga/In/Se.

PAGE 207

207 Figure 5 8 Te mperature ramp sequence during annealing of SS/Ru/Mo/Cu/Ga/In/Se precursor with out selenium overpressure.

PAGE 208

208 Figure 5 9 Temperature ramp sequence during annealing of SS/Ru/Mo/Ga/In/Cu/Se with overpressure of selenium

PAGE 209

209 Figure 5 10 Electron diffraction pattern of quenched sample during temper ature ramp annealing of SS/Ru/Mo/Ga/In/Cu/Se A) Image of quenched sample, B) selected area diffraction patterns A B

PAGE 210

210 Figure 5 11 Auger depth profile of temperature ramp annealed precursor of SS/Ru/Mo/Ga/In/Cu/Se.

PAGE 211

211 Figure 5 12 Seque nce of isothermal annealing of SS/Ru/Mo/Ga/In/Cu/Se at 400 C.

PAGE 212

212 Figure 5 1 3 Avrami plot using fractional conversion from isothermal experiments at dif ferent isothermal temperature. A) SS/Ru/Mo/Cu/Ga/In/Se, B) SS/Ru/Mo/Ga/In/Cu/Se. A B

PAGE 213

213 Figure 5 1 4 Arrhenius plot using rate constants obt a ined from Avrami model. A) SS/Ru/Mo/Cu/Ga/In/Se, B) SS/Ru/Mo/Ga/In/Cu/Se. A B A =78(6) kJ/mole B=116(5) kJ/mole

PAGE 214

214 CHAPTER 6 REACTION PATHWAYS AN D KINETICS OF MOSE 2 FORMATION Overview CIGS has gained significant attention in thin film PV because of its exceptional properties such as direc t band gap, high absorption coefficient, and high conversion efficiency. A number of processes have been used to synthesize CIGS and often molybdenum is used as a back contact material. An early study was performed on Mo CIS interface using polycrystalli ne and single crystal CIS. The interface suggested formation of a Schottky barrier limiting the open circuit voltage [18] Recently it has been reported that the interface of Mo/CIGS junction is ohmic due to the presence of thin MoSe 2 layer at the interface between Mo and CIGS [19, 167] It has also been reported that having a MoSe 2 layer at Mo/CIGS interface improves the adhesion properties of the CIGS. MoSe 2 has a hexagonal crystal structure consisting of planes of molybdenum atoms bonded covalently to selenium atoms and the planes are weakly bonded to each other by Van der Waal forces. The orientation of the planes with respect to molybdenum plays an important role in electrical properties as well as adhesion of CIGS. The adhesion is go od when these planes are oriented perpendicular to the molybdenum substrate since there exists a strong covalent bond between selenium atom and the underlying molybdenum at the interface. It is found that the series resistance is increased when the orient ation is changed from perpendicular to parallel [168] It is hence necessary to determine the growth rate of MoSe 2 and its orientation with respect to the surface. It is also known the MoSe 2 is not a good conductor and it is important to keep the thickness of the interfacial layer as small as possible. There exists no quantitative model describing the growth of MoSe 2 and

PAGE 215

215 reaction mechanism. In this Chapter, reaction pathways and kinetics of selenization of polycrystalline molybdneum using time resolved high temperature X ray diffraction [123] The formation of MoO 2 as a result of residual oxygen during processing of molybdneum for CIGS solar cells is also explored. Reaction kinetics of MoSe 2 formation has been performed by Abou Ras et al. using ex situ technique indicat ing one dimensional diffusion controlled transformation [17] Experimental free glass substrates by rf m agnetron sputtering. Molybdenum samples of ~1 cm 2 were loaded into the HT XRD system consisting / X ray diffractometer equipped with an Anton Paar XRK state detector. Selenium in t he form of powder was placed under the sample in the recess area to provide uniform selenium pressure upon volatilization. The sample holder was covered with graphite dome to contain the selenium overpressure during temperature ramp selenization. Parallel beam optics was employed to eliminate the graphite peak from the XRD spectrum. The chamber was purged with either N 2 or forming gas to remove residual oxygen. The glass/Mo layer was first heated to 80 C at the rate of 10 C /min and then X ray diffracti on data were collected to a substrate temperature of 150 C. The ramp rate was increased to 40 C /min and X ray data were subsequently collected in 10 C increments until a final temperature of 600 C was reached. The total scan time for each temperature set was approximately 4 minutes. A final high resolution scan was recorded after cooling down the sample to room temperature. Isothermal data set was collected with temperatures selected from ramp

PAGE 216

216 studies, following phase evolution of MoSe 2 as a functio n of time. The kinetic parameters were determined using isothermal data using Avrami and parabolic solid state growth models. TEM EDS were done to determine the composition of the reaction products. Selected area diffraction patterns were collected to s upport the results obtained from HT XRD. Temperature Ramp Annealing The room temperature XRD of as deposited molybdenum as shown in Figure 6 1 was polycrystalline with texture along the <110> direction. A sample of size 1 cm 2 was loaded in high temperatur e furnace with some selenium powder. The selenium pressure maintained in the system was approximately 1 torr. High temperature X ray diffraction data was collected in a sequence as mentioned in the experimental section with selenium overpressure in nitro gen environment as shown in Figure 6 2. No reaction was observed in the pathway till the temperature reached 410 C. As temperature reached 420 C, polycrystalline MoSe 2 (100, 110) and MoO 2 (111,211) appeared simultaneously. Also a sudden drop in inten sity of molybdenum was observed. MoO 2 pattern selected has a cubic crystal structure where as MoSe 2 had a hexagonal structure. With further increase in temperature, both the MoSe 2 and MoO 2 reflections continued to increase and were complete at 520 C with no change in their intensities. The selectivity of the reactions is still not known at this point. The sample was cooled down to room temperature and peak shift in the final scan was due to lattice contraction during cool down process. The following inte rfacial reaction was observed during temperature ) ramp, (6 1)

PAGE 217

217 (6 2) It is anticipated that formation of MoO 2 is either from residual oxygen present in the sample or in the furnace. Also from the phase relationship of Mo O and Mo Se, it is found that both oxygen and selenium have limited solubility is molybdenum [169, 170] From the reactions ( 6 1) and ( 6 2), it is observed that oxygen will react with MoSe 2 to form the molybdenum oxide. It is also well known fact that for an ideal metal semiconductor contact, both layers are assumed to have a good contact a t the atomic level with no oxide layers and impurities such as carbon present at the interface. Thus a second experiment was performed using a mixture of 4% H 2 /N 2 to purge the system as well as used as a carrier gas. As shown in Figure 6 3, in the presence of hydrogen, formation of Mo O 2 was suppressed. No other major change was observed in the pathway except the disappearance of MoO 2 There has been several selenides of molybdenum reported in the literature which includes MoSe 2 Mo 3 Se 4 Mo 9 Se 11 and M o 15 Se 19 [171, 172] In our study no other MoSe y species was observed except MoSe 2 A high resolution XRD scan was taken after the ramp (Figure 6 4) and growth was mostly in ( 100 ) and (110) orientations The (004), (006) and (103) reflections emerged from the background level at low intensity indicating the presence of small grains in those directions. These planes are responsible for high series resistance encountered during CIGS device fabrication. Isothermal Anne aling and Reaction Kinetics In situ high temperature X ray diffraction has been used by our group to determine the kinetics using solid state growth models. We performed the isothermal selenization experiments at selected temperatures between 350 430 C t o determine the rate

PAGE 218

218 constants and activation energy for MoSe 2 formation. Scans were performed in the same furnace and each scan time was approximately 2 minutes covering the 2 range from 30 to 42 o The 2 range was selected such that the major reflecti on of MoSe 2 i.e. 100 lies within the specified range. The fractional conversion of product MoSe 2 was calculated using normalized peak area of molybdenum assuming maximum peak area represents complete conversion. Using the fractional conversion data, rate constants and activation energy was computed using Avrami and parabolic growth rate models as discussed in Chapter 2 [173, 174] The isothermal scan taken at 430 C is shown in Figure 6 5. Figure 6 6 shows the fractional conversion as a function of time f or different isothermal temperatures. The slope of fractional conversion should give the growth rate of MoSe 2 film. As temperature increases, fractional conversion also increases. The rate constants and exponent s were calculated Avrami model The Avram i exponent varied from 0.7 0.92 indicating one dimensional diffusion controlled reaction. Figure 6 7 shows the A vrami plot at selected isothermal temperatures. In case of parabolic model as discussed in Chapter 2 planar surface is assumed between two s olid materials. Product is formed at the interface and for further reaction in forward direction, selenium has to cross the product boundary (MoSe 2 ) and then react with molybdenum. Figure 6 8 shows the plot for parabolic model for different isothermal te mperatures. The rate constants obtained from the models was used to determine the activation energy using Arrhenius relation, (6 3)

PAGE 219

219 The data was represented well by both Avrami and parabolic model yielding activation energy 101.2 (8) and 101.7 (6) kJ/mole from Avrami and parabolic model respectively. Figure 6 9 A and 6 9 B compares the Arrhenius plot from two different models. The previous reported activation energy was in the order of 48 kJ/mole using quartz spring ba lance method [175] To further understand the orientation and composition of the grown MoSe 2 layer, TEM characterization was done. Figure 6 10 A shows the image for glass/Mo/MoSe 2 layered structure. The thickness of the MoSe 2 point scans were done from pure MoSe 2 layer to pure molybdenum layer. The selenium to molybdenum counts was nearly twice in the pure molybdenum selenide region indicating the compound formed was MoSe 2 Spectrum 7 in TEM image (Figure 6 10A) showed diffusion of selenium t hrough the molybdenum layer without reacting with the molybdenum layer for first 100 nm. The reason for this observation was not understood until TEM image was taken for as deposited molybdenum (Figure 6 10 B ). It was found that the within the molybdenum existed two different orientations and at the interface of the orientation, oxide layer was present. Hence oxygen at the interface was replaced by selenium by selenization reaction. It should be also noted that MoO 2 formed during ramp under nitrogen ambi ence would be because of this existing oxide layer. XPS (X ray photoelectron spectroscopy) depth profile was done on pure molybdenum sample to determine the chemical bonding states of oxygen with molybdenum. Data was collected till pure molybdenum was obt ained. Initial scan showed presence of MoO 3 at the surface with Mo 3d 5/2 (232.6 eV) and Mo 3d 3/2 (235.8 eV) emission lines were consistent with the

PAGE 220

220 reported literature values. Within the first 10 nm for molybdneum, a higher oxide of molybdneum is formed at the surface (Figure 6 11 A ). Further down the depth of molybdenum Mo 3d 5/2 emission lines at E B = 228.60 228.40 eV and oxygen 1s emission lines at E B = 530.6 530.66 eV was obtained consistent with the literature values of MoO 2 [E B (Mo 3d 5/2 ) = 228.8 232 eV and E B (O 2 1s) = 529.9 530.7 eV] [176, 177] as shown in Figure 6 11 B To further understand the system, E llingham diagram was con structed for Mo with selenium and oxygen as gas phase species. Unlike oxygen, selenium is a solid a room temperature. The diagram gives the partial pressure of gaseous species required to oxidize/ selenize the metal at a given temperature (Figure 6 12). It is found that at given temperature, oxides of molybdenum are more stable than selenides. At 700 K, the partial pressure of selenium (Se 2 gas species) required to oxidize the metal is 10 17 atm and that of oxygen is 10 33 atm indicating a small amount o f oxygen is sufficient to form the molybdneum oxide. One of the major observations noticed in the diagram is that the MoSe 2 forms after the reaction of CIS is complete where as MoO 2 are more stable than CIS. Selected area diffraction patterns were taken fo r molybdenum, MoSe 2 and Mo MoSe 2 interface. The ring patterns for molybdenum matched the spectra with XRD. For MoSe 2 some streaks were observed in the pattern indicating preferred orientation as observed in high resolution XRD scan after selenization of molybdenum. The d spacing of the MoSe 2 rings matched well with the XRD spectra. At the interface of Mo MoSe 2, the ring patterns of MoSe 2 matched with 100 and 110 orientation indicating the c axis of the grown MoSe 2 is parallel to molybdenum surface. Fig ure 6 13 shows the ED (electron diffraction) for all the patterns. This type of structure has been previously reported by Wurz et.al for MoSe 2

PAGE 221

221 formed during growth of CuGaSe 2 by CVD. However, this study was mainly focused on the MoSe 2 without the presenc e of copper, indium and selenium. It is understood that in the presence of other elements the orientation of MoSe 2 is affected. It is thus suggested to grow a thin MoSe 2 layer followed by growth of CIS/CIGS. Summary Reaction pathways during selenization of molybdenum were followed using in situ high temperature XRD in two different gas configurations. The formation of MoO 2 was observed when the system was run in nitrogen atmosphere. Formation of MoO 2 was suppressed when forming gas was used. XPS data s uggested formation of higher oxides (MoO 3 ) on the surface which then reduces to MoO 2 towards the depth of the molybdenum. Isothermal experiments were carried out at selected temperatures to extract the growth kinetics using time resolved diffraction. Rat e constant and activation energies were computed using avrami and parabolic solid state growth models. Both models yielded almost identical value for the activation energy. SAED (selected area diffraction pattern) patterns were taken at the interface of M o MoSe 2 showing 100 and 110 patterns of MoSe 2 existed at the interface.

PAGE 222

222 Figure 6 1 Room temperature X ray diffraction of molybdenum substrate

PAGE 223

223 Figure 6 2 Temperature ramp selenization sequence of molybdenum in nitrogen atmosphere

PAGE 224

224 Figure 6 3 Temperature ramp selenization sequence of molybdenum in forming gas atmosphere

PAGE 225

225 Figure 6 4 Room temperature X ray diffraction scan of temperature ramp selenized molybdenum in forming gas atmosphere

PAGE 226

226 Figure 6 5 Isothemal selenization sequence of molybdenum in forming gas atmosphere

PAGE 227

227 Figure 6 6 Fractional conversion of molybdenum with time at different isothermal temperature

PAGE 228

228 Figure 6 7 Avrami plot using fractional conversion from isother mal studies

PAGE 229

229 Figure 6 8 Parabolic plot using fractional conversion from isothermal studies

PAGE 230

230 Figure 6 9 Arrhenius plot using rate constants obtained f rom solid state growth models. A) Avrami model, B) parabolic model B A

PAGE 231

231 Figure 6 10 TEM image of temperature ramp selenized and pure molybdenum. A) image of MoSe 2 grown on molybdenum substrate, B) image of pure molybdenum A B

PAGE 232

232 Figure 6 11 X ray photoelectron spectroscopy of pure molybdenum. A) b onding states at the surface of molybdenum, B) bonding states after sputtering the surface A B

PAGE 233

233 Figure 6 12 Ellingham diagram for Mo O, Mo Se and CIS systems.

PAGE 234

234 Figure 6 13 Electron diffraction of temper ature ramp annealed precursor. A) electron d iffraction of pure MoSe 2 B) electron diffraction of pure Mo, C) electron diffraction at interface of Mo MoSe 2 A B C

PAGE 235

235 CHAPTER 7 SELEN IZATION PATHWAYS AND KINETICS OF REPEATED CU/GA/ IN ELEMENTAL LAYERS Overview CIGS based solar cells have gained attention in thin film PV because of its excellent properties such as high efficiency (20.3% AM1.5), high optical absorption coefficient and direct band gap [14] In spite of such good properties, the absorber commercializati on is still in progress for more than a decade. Two different scenarios targeted to reduce the overall cost of manufacturing are efficiency and throughput. Of different processes available for absorber growth [92, 1 09, 119, 178] selenization of metallic precursors have shown potential to increase both the targets discussed above [137] Key challenge in absorber synthesis is increasing the rate to allow for realized of NREL three stage co evaporation process [36] Higher gallium content at the surface and rear of the absorber has increased the overall efficiency due to reduction in interfacial recombination and better collection efficiency of minority carriers supported by device modeling work [1 38] With devices approaching wide band gap of 1.45eV to better match the solar spectrum with more gallium addition and reduced thickness, distribution of gallium becomes even more important. In this Chapter repeated elemental layers of Cu/In/Ga (4 ti mes and 8 times) were deposited using molecular beam epitaxy to study the absorber synthesis rate along with gallium redistribution. Absorber synthesis rate were studied using in situ high temperature X ray diffraction by reacting the metallic precursor w ith el emental selenium in gas phase. Depositing elemental layers in modulated structure helps in decreasing diffusion distances and better efficiencies have been reported for CuGa/In multiple

PAGE 236

236 stacks [179] It has also been reported that smooth and dense films are obtained employing modulated structure and more importantly single phase chalcopyrite structure was obtained [180 182] Experimental soda lime glass substrates using molecular beam epitaxial reactor. The precursor layers of Cu/Ga/In were repeated 4 times (70/20/50 nm) and 8 times (40/10/30 nm) a nd deposition was carried out at room temperature. The total thickness of the sample was 0.56 m and 0.64 m for repeated layers of 4 and 8 times. The Ga/ (Ga+In) was ~ 0.31 and Cu/III ratio was ~0.92 as measured by inductively coupled plasma optical emission spectroscopy (ICP). The high temperature X ray diffraction studies were done in PANalytic system have been described in Chapter 3 [110] The sample was heated in a custom built stainless steel reactor with 120 mg of selenium powder and heated at low ramp rate of 10 C/min. Data was collected from room temperature till 570 C for every 10 C step increment. Aluminum was used a sealant for stainless steel reactor which also served as a X ray transparent window. It should be noted that reflections of aluminum in the X ray data appears due to the same reason. Before temperature ramp experiments, room temperature X ray diffraction data was collected to identify the using ICDD 2010 d atabase. The sample was further characterized usi ng scanning electron microscopy ( SEM, JEOL JSM 6335F) for microstructure. The gallium

PAGE 237

237 distribution along the thickness was obtained from depth profile using auger electron spectroscopy (AES). Temperature R amp Annealing of Metallic Precursor The room temperature XRD of as deposited precursor s is shown in Figure 7 1. Reflections of indium, Cu 9 Ga 4 Cu 11 In 9, metastable C uIn and molybdenum were and 26% as determined by inductively coupled plasma optical emission spectroscopy. Scanning electron images were taken for sample before annealin g and as deposited samples had matrix and island structure (Figure 7 2). The composition of the island and matrix were determined from spot EDS measurements and the results are presented in Table 7 1. Temperature ramp annealing was performed for metallic p recursors to observe the phase transformation in Cu In Ga system. First, a room temperature scan was recorded and then the temperature was ramped to 50 C and data was collected for every 10 C s tep increment. The ramp rate was kept low at 10 C/min until the temperature reached 150 C after which the ramp rate was increased to 20 C/min. Forming gas was used a carrier gas to prevent oxidation of the sa mple. Figure 7 3 and 7 4 shows the temperature ramp annealing of 8 and 4 repeated elemental metallic precursors. For elemental layers repeated 8 times, indium and metastable CuIn melted at 150 C substantiating the temperature calibration of the system. Th e solid solution Cu 9 (In x Ga 1 x ) 4 forms immediately after melting of indium. The only compound that is detected at temperatures above 150 C is Cu 9 (In x Ga 1 x ) 4 Solid solution formation is complete at around 240 C with no evidence of diffusion of group II I sub lattice. The

PAGE 238

238 reaction was followed till 400 C and the temperature was brought back to room temperature to detect if any additional phases were present. The final sc an revealed indium textured in ( 002 ) plane Similar experiment was carried out for elemental layers repeated 4 times and the only difference observed was the completion of diffusion of group III sub lattice at 370 C much higher than for 8 times repeated elemental layers. High resolution X ray diffraction data was recorded for temperatu re ramp annealed samples to identify the crystalline phases that were not observed in the high te mperature low resolution scan. Figure 7 5 shows the room temperature scan after temperature ramp annealing for both 8 and 4 repeated layers of Cu/Ga/In Th e gallium composition in the solid solution Cu 9 (In x Ga 1 x ) 4 using Cu 9 In 4 and Cu 9 Ga 4 as the end members. Cu 9 In 4 and Cu 9 Ga 4 is the only end members included in the Cu In Ga metallic system as discussed in Chapter 4 The gal 9 (In 0.55 Ga 0.45 ) 4 and Cu 9 (In 0.64 Ga0 .36 ) 4 for 8 and 4 repeated layers of Cu/Ga/In. SEM image (Figure 7 6) showed matrix and island structure for temperature ramp annealed precursors and the composition determined from spot EDS measurements are reported in Table 7 1. After annealing the gallium content in the sample were similar in both matrix as well as island regions. The indium detected in the XRD spectrum could not be traced by EDS It has been reported the formation of Cu 9 (In x Ga 1 x ) 4 takes place near the back contact and leads to incomplete reaction during selenization [183] and thus creating voids which increases the series resistance and hence decreasing the efficiency. It is this necessary to avoid t his phase formation because of preferential reaction of indium over gallium leading to gallium segregation towards the back contact

PAGE 239

239 Temperature Ramp Selenization explained in t he experimental section. Figure 7 7 shows the temperature ramp selenization of 8 repeated layers of Cu/Ga/In. Indium and CuIn melted at 150 C, followed by formation of solid solution Cu 9 (In x Ga 1 x ) 4 till 300 C. During the temperature ramp of the same metallic precursor, composition of gallium in the solid solution was around 0.2 at 300 C. At 300 C, CIGS forms and continues to grow and th en diffusion of group III sub lattice takes place in the chalcopyrite phase. The pathway to CIGS formation involves formation of CuSe and InSe occurring at 300 C. The inset in the Figure 7 7 shows the formation of InSe and CuSe during temperature ramp. The reaction is complete at 450 C with evidence of MoSe 2 formation. Also observed is the sudden decrease in the intensity of molybdenum peak. No CGS formation takes place before of initial mixing of group III sub lattice prior to selenization. The follow ing reactions are observed during temperature ramp, (7 1) (7 2) (7 3) (7 4) (7 5) (7 6)

PAGE 240

240 Similar pathways were obtained for temperature ramp selenization of 4 repeated elemental layers of Cu/Ga/In including the phase transformation temperatures as shown in Figure 7 8. No major difference in the pathways was observed by increasin g the repeating layers. In NREL 3 stage process, most of the growth happens in the tie line connecting Cu 2 Se and In 2 Se 3 thus favoring ordered vacancy compound formation having wider band gap at the surface [184] However for the samples in this study, the growth happens in tie line connecting InSe CuSe. The formation of the OVC compound at the surface from this pathway is highly impossible and perhaps one of the reasons why selenized champion cells are behind co evaporat ed champion cells. This could also be the reason why sulfurization step is done to make the band gap wider at the surface. Scanning electron microscope images were taken for samples after temperature ramp annealing to see the difference in microstructur e (Figure 7 9). For samples with 8 repeated elemental layers, the surface morphology showed densified grain structure compared wit h 4 repeated elemental layers. More voids were observed near the back contact for sample with 4 repeated elemental layer s wh ich was minimized in sample with 8 repeated elemental layers. The grain growth in both the samples was almost the same showing similar grain size distribution. A high resolution scan was taken in the end and Rietveld refinement was performed to determine the gallium distribution (Figure 7 10). For 4 repeated layers of Cu/Ga/In, a graded structure of CIS and CIGS with 25 28% was obtained based on refined lattice parameters. In case of 8 repeated layers more grading was observed with gallium content of 28 30%. Auger electron spectroscopy was performed to determine the Ga/III profile as a function of depth. For 4 repeated layers, higher gallium concentration was observed towards back

PAGE 241

241 contact and for 8 repeated layer gallium concentration was u niform throug hout the thickness. The gradient Ga/III was similar for both 4 and 8 repeated layers of Cu/Ga/In Figure 7 11 compares the AES depth profile for 4 and 8 repeated layers of Cu/Ga/In. Isothermal Annealing and Reaction Kinetics Quantitative rate data was obta ined by following the CIGS phase formation as a function of time at different isothermal temperatures. The isothermal temperatures were selected from temperature ramp experiments. The maximum ramp rate that can be as 60 C/min. The ramp rate can be increased to 120 C/min by manual control. The 2 scan range was selected such that the major reflection (112) lies within the range. The ramp rate was controlled manually to rapidly achieve the desired isothermal temperatu re. Figure 7 12 and 7 13 shows the isothermal annealing of 4 and 8 repeated layers of precursors at different temperatures. The scan time for each isothermal data set was around 75 seconds and data was collected for approximately 70 minutes. In the end, temperature was increased to 580 C to complete the reaction and fractional conversion was estimated by normalizing area under the curve. It is assumed that no change in texture happ ens during the growth process. The calculated fractional conversions are used to determine rate constant and activation using solid state growth models discussed in Chapter 2 Figure s 7 14 and 7 15 shows the Avrami and parabolic plot for selenization of 4 and 8 repeated layers of Cu/Ga/In. The rate constant estimated from both models were plotted using the Arrhenius expression (7 7)

PAGE 242

242 Where, E a is the activation energy k the first order rate constant and A the pre exponential factor. Figure s 7 16 shows the Arrhenius plot using Avrami and parabolic kinetics. It is observed that pr ecursors with 8 repeated layers of Cu/Ga/In had more activation energy than 4 repeated layers. In case of 4 repeated layers for Cu/Ga/In, estimated activation energy from Avrami model gave 110(19) kJ/mol e and parabolic model yielded 122 ( 27 ) kJ/mole. Fo r 8 repeated layers of Cu/Ga/In, activation energy obtained from Avrami model was 122(27) kJ/mole and 129 (6) kJ/mole from parabolic model. Both the models agree well with each other within the error limits. Higher activation energy is due to mixing of the group III sub lattice and higher rate is obtained for 8 repeated layers of Cu/Ga/In due to decrease in the diffusion distances. Since most of the solid state reactions are diffusion limited, decreasing the diffusion distance helps in increasing the s ynthesis rate. Summary Reaction pathways were followed for selenization of 4 and 8 repeated layers of Cu/Ga/In using in situ high temperature X ray diffraction. It is evident from temperature ramp experiments that stable solid solution Cu 9 (In x Ga 1 x ) 4 i s in equilibrium with liquid indium. In case of 8 repeated layers of Cu/In/Ga, formation of the single phase of Cu 9 (In x Ga 1 x ) 4 was complete at 300 C due to reduced diffusion distances for mixing for group III sub lattice. The selenization pathways show ed the formation of InSe and CuSe during the temperature r amp experiments. The CIGS formation occurs in tie line InSe CuSe, different than that observed for NREL three stage processes. The selenization pathways can be summarized as Cu 11 In 9 + Cu 9 Ga 4 Cu 9 ( In x Ga1 x ) 4 + InSe(s) + CuSe(s) CuIn x Ga 1 x Se 2 Isothermal selenization was performed at selected

PAGE 243

243 temperatures and quantitative data in terms of activation energy and rate constant were estimated using Avrami and parabol ic solid state growth models. Est imated activation energy from Avrami model gave 110( 19) and 111( 1) kJ/mole for 4 and 8 repeated layers for Cu/In/Ga. Parabolic model yielded 122( 27) and 129 ( 6) kJ/mole for 4 and 8 repeated layers of Cu/In/Ga agreeing with the Avrami model indicat ing one dimensional diffusi on controlled transformation. Higher rate was obtained for precursor with 8 repeated layers of Cu/Ga/In because of decreased diffusion distances during the selenization process

PAGE 244

244 Table 7 1. SEM EDS composition me asurement at matrix and island for as deposited and temperature ramp annealed samples Sample Matrix (at %) Island (at%) Cu In Ga Cu In Ga Cu/Ga/In (4 times) 48.8 38.7 12.5 32.9 58.4 8.7 Cu/Ga/In ( 8 times) 46.1 40.5 13.4 44.2 44.7 11.1 Cu/Ga/In (4 tim es) Annealed 33.7 53.2 13.1 38.2 45.9 15.9 Cu/Ga/In ( 8 times) Annealed 34.1 50.9 15.0 43.2 42.4 14.4

PAGE 245

245 Figure 7 1 Room temperature X ray diffraction scan of as deposite d repeated layers of Cu/Ga/In. A) 8 repeated layers of Cu/Ga/In, B) 4 repeated layers of Cu/Ga/In A B

PAGE 246

246 Figure 7 2 SEM images of as deposite d repeated layers of Cu/Ga/In. A) 8 repeated layers of Cu/Ga/In, B) 4 repeated layers of Cu/Ga/In A B

PAGE 247

247 Figure 7 3 Temperature ramp annealing of 8 repeated layers of Cu/Ga/In in forming gas

PAGE 248

248 Figure 7 4 Temperature ramp annealing of 4 repeated layers of Cu/Ga/In in forming gas

PAGE 249

249 Figure 7 5 Room temperature X ray diffraction scan of temperature ramp annealed precursor. A) 4 repeated layers of Cu/G a/In, B) 8 repeated layers of Cu/Ga/In B A

PAGE 250

250 Figure 7 6 SEM images of temperature ramp an nealed precursors of Cu/Ga/In. A) 8 repeated layers of Cu/Ga/In, B) 4 repeated layers of Cu/Ga/In A B

PAGE 251

251 Figure 7 7 Temperature ramp selenization seq uence of 8 repeated layers of Cu/Ga/In

PAGE 252

252 Figure 7 8 Temperature ramp selenization sequence of 4 repeated layers of Cu/Ga/In

PAGE 253

253 Figure 7 9 SEM images of temperature ramp sel enized precursors of Cu/Ga/In. A) surface image of 8 repeated lay ers of Cu/Ga/I n, B) cross sectional image of 8 repeated layers of Cu/Ga/In, C) surface image of 4 repeated layers of Cu/Ga/In D) cross sectional image of 4 repeated layers of Cu/Ga/In B C A D

PAGE 254

254 Figure 7 10 Room temperature XRD of temperature ramp sel eniz ed precursors of Cu/Ga/In (Inset shows the 112 orientation of CIGS). A) 4 repeated layers of Cu/Ga/In, B) 8 repeated layers of Cu/Ga/In A B

PAGE 255

255 Figure 7 11 AES depth profile of temperature ramp seleniz ed precursors of Cu/Ga/In. A) 4 repeated layer s of Cu/Ga/In, B) 8 repeated layers of Cu/Ga/In B A

PAGE 256

256 Figure 7 12 Isothermal selenization sequence of 4 repeated layers of Cu/Ga/In at different temperatures. A) 260 o C, B) 280 o C, C)300 o C, D) 320 o C C A B D

PAGE 257

257 Figure 7 13 Isothermal selenization sequence of 8 repeated layers of Cu/Ga/In at different temperatures A) 260 o C, B) 280 o C, C)300 o C, D) 320 o C A D B C

PAGE 258

258 Figure 7 14 Rate constant estimation for 4 repeated layers of Cu/Ga/In us ing solid state growth models. A) Avrami mo deI, B) parabolic model A B

PAGE 259

259 Figure 7 15 Rate constant estimation for 8 repeated layers of Cu/Ga/In us ing solid state growth models. A) Avrami modeI, B) parabolic model A B

PAGE 260

260 Figure 7 16 Arrhenius plot using calculated rate constant s from solid stat e growth model. A) Avrami modeI, B) parabolic model A B

PAGE 261

261 CHAPTER 8 EFFECT OF SODIUM ON SELENIZATION OF CUGA /IN METALLIC PRECURS OR Overview Recently, CIGS solar cells have reached efficiencies exceeding 20% [14] It is understood that addition of sodium plays an important role in improving the electrical properties of solar cell. Perhaps the reason of growing absorber materials on soda lime glass is justified. Diffusion of sodium from soda lime glass occurs during synthesis of absorber layer at high temperature and is highly dependent on the microstructure of molybdenum back contact. Activation energy of 8.6 kcal/mole has been reported by Zellner at al. for sodium diffusion through the mol ybdenum layer using time dependent XPS studies [62] The effect of oxygen and water vapor on sodium diffusion has been described in the same study. For flexible substrates such as polyimide and stainless steel which are free of sodium source, sodium additi on is achieved by adding a sodium containing precursor such as sodium fluoride (NaF) before or after growth [45] CIGS films deposited on NaF show adhesion issues [43, 47, 185] and limit the sodium suppl y compared with sodium supply from soda lime glass. Highest efficiency of 18.8% has been obtained on polymer foils with post deposition of sodium using NaF [58] Irrespective of supply methods of sodium, i t has been reported that addition of sodium shows increased ( 112 ) texturing [186] change in the reaction rate [187] and increased hole concentrations by neutralizing selenium vacanci es by chemisorptions of oxygen [59] Most of the CIGS industries use flexible substrates and a uniform sodium supply is essential. Till date only few studies have explored the effect of sodium on kinetics of CIGS formation [57, 188] The re exists no quantitative model describing the kinetic effects upon sodium addition. In this study we have used the sodium doped

PAGE 262

262 molybdenum substrates from Plansee which helps in supplying constant sodium source during absorber formation reaction We have explored the reaction pathways and kinetics during selenization of metallic CuGa/I n precursor deposited on sodium free and sodium doped molybdenum substrate ( 3 and 5 at%). Experimental Metallic precursors of CuGa/In were deposited on three different so dium free glass substrates of 1 mm thickness i.e., SFG/Mo, SFG/MoNa/Mo (with 3 and 5 at% Na). From this point on MoNa 3 and MoNa 5 will be used as terminology for 3 and 5 at% sodium. The precursors were deposited by sequential sputtering of CuGa alloy (24 wt % Ga) and In targets. Total thickness of the precursor was approximately 0.6 m. The atomic compositions of the precursors as determined by ICP showed Cu/III=0.95 1.0 and Ga/III= 0.2. stem with selenium vapor. The description of the system can be found in Chapter 2 The sample was placed in a custom built stainless steel reactor with 120 mg of selenium powder and heated at a ramp rate of 10 C/min. Data was collected from room tempe rature till 580 C for every 10 C step increment. Aluminum was used as an X ray transparent window and reflections of aluminum can be observed in the data. Before annealing the precursors, high resolution room temperature XRD data was collected. The cr ystalline 010 database. The elemental distribution in the finally annealed absorber was determined by SIMS (secondary ion mass spectroscopy) analysis.

PAGE 263

263 Temperature Ramp Selenization o f Metallic Precursor The room temperature X ray diffraction of as deposited precursor (Figure 8 1) showed reflections of indium, CuIn, Cu 9 Ga 4, Cu 3 Ga, Cu 16 In 9 and molybdenum. Molybdenum showed preferred orientation in <110> direction. The as deposited pat tern had matrix and island structure with indium distributed over the island. Figure 8 2 shows the SEM image of as deposited precursors. Temperature ramp annealing experiments were performed on all the precursors to see if the effect of sodium in the reac tion pathway. At first, reaction pathway of SFG/Mo/CuGa/In was followed during selenization. The sample was loaded in the stainless steel reactor with 120 mg of selenium powder. The amount of selenium is supplied twice the amount required Figure 8 3 sho ws the temperature ramp selenization of SFG/Mo/CuGa/In precursor. Indium and metastable CuIn melted at its melting point. After melting, Cu 11 In 9 formation took place at 160 C. There was no diffusion of indium and gallium as observed earlier in our prev i ous work. The formation of CIS takes place at 290 C followed by formation of CGS at 320 C The CIS is perhaps, formed at the surface acting as a diffusion barrier for selenium to react with gallium to form CGS. The formation of CGS at higher temperat ure than CIS is because of preferential reaction of selenium with indium rather than gallium. Since, there was no mixing of the grou p III sub lattice, CIS and CGS formed separa tely rather than solid solution. At 440 C, diffusion of gallium in CIS starte d to happen with broad distribution occurring near the end of the CIS peak. The reaction was complete at 540 C with the formation of MoSe 2 Though MoSe 2 formation is not evident, but the intensity of molybdenum went down indicating MoSe 2 formation.

PAGE 264

264 Simil ar selenization studies were done with MoNa 3 and MoNa 5 and no change in the reaction pathways was observed as shown in Figures 8 4 and 8 5. Comparing sodium free selenized sample with MoNa 3 and MoNa 5 the rate of increase in CIGS formation is more for sam ples selenized in presence of sodium. With MoNa 3 MoSe 2 formation was not observed during temperature ramp selenization. However, a final high resolution scan revealed MoSe 2 Perhaps, thickness of the MoSe 2 was low to be detected in the low resolution s can during temperature ramp selenization. In case of MoNa 5 the reaction was complete at 450 C with MoSe 2 forming at the same temperature. Figure 8 6 compares the final high resolution scan of tempera ture ramped selenized samples. For sodium free selen ized CIGS, a broad distribution of ( 112 ) orientation (shown in the inset of Figure 8 6) indicates homogeneous distribution of group III lattice. With sodium presence, a graded structure is observed consistent with the observation in SIMS analysis. Figur e 8 7 shows the SIMS depth profile for films selenized with MoNa 3 and MoNa 5 as substrates. Molybdenum is detected earlier in the scan owing to the roughness of the selenized film. The sodium concentration decreases from the back contact to the film indica ting some sodium has diffused from the back contact. For MoNa 5 sodium and oxygen follows the similar profile with sodium concentration increasing towards the surface. The increase in the sodium concentration at the surface suggest that suppression of o rdered vacancy compound as sodium presence at the surface of the absorber inhibits the formation of OVC phase as reported in the literature [51] Figure 8 8 shows the Ga/III ratio as a funct ion of depth of the absorber. For both the samples depos ited on sodium doped molybdneum the front For

PAGE 265

265 films selenized with MoNa 5 more gallium is segregated towards back contact compared with MoNa 3 For films selenized with MoNa 3 as back contact, gallium distribution is uniform towards the back contact, followed by an increase and then dropping to a minimum value at the front side. However changing the order of deposition of precursor structure before selenization helps in obtai ning the desired gallium distribution as obtained for NREL champion cell process [36] Rietveld refinement was performed on the high resolution scan and the phase fractions of compounds are listed in Table 8 1. Hence, addition of sodium helps in reducing t he processing temperature of selenization reaction as well as helps in obtaining a graded structure. The following reactions are observed during temperature ramp selenization, (8 1) (8 2) (8 3) (8 4) (8 5) The SEM image were taken for temperature ramped selenized samples to see the effect of sodium on the microstructure of CIGS film. Figure 8 9 compares the microstructure of CIGS film after temperature ramp selenization of CuGa/In precursor. For the control sample with no sodium, grain growth was no evident at all. With MoNa 3 some grain growth was observed but cannot be differentiated distinguishly wi th the control sample. Some voids were observed near the back contact which can be related to sodium diffusion from the MoNa 3 layer during the temperature ramp selenization.

PAGE 266

266 However with MoNa 5 improved microstructure was observed because of increase in the flux of sodium diffusion. It has been reported that excessive sodium diffusion from soda lime results in poor adhesion of CIGS to the molybdneum layer with formation of soluble sodium compounds at the interface [187] However no adhesion problems were observed after selenization using MoNa 3 and MoNa 5 substrates. Grain growth was apparent compared with sodium free and MoNa 3 samples. Clearly more sodium incorporation helps in attaining good grain growth for CuGa/In prec ursor. Isothermal Annealing and Reaction Kinetics The rate of increase of (112) orientation of CIGS was followed as a function of time isothermally at different temperatures. The isothermal temperatures were selected from temperature ramp experiments. T he precursor samples were loaded in the stainless steel reactor with selenium shots (120 mg) and the ramp rate was manually increased to 120 C/min. The 2 scan range was selected such that the major reflection of CIGS (112) lies within the range. Each data set in the isothermal scan was collected for 75 seconds and the data was collected for approximately 70 minutes. After 70 minutes, the t emperature was increased to 580 C to complete the reaction as evidenced by formation of MoSe 2 at higher temperature. Figures 8 10 to 8 12 compares the isothermal plots for selenization of CuGa/In deposited on different substrates. The fractional convers ion was calculated by normalizing the area under the curve with maximum area under the curve representing complete conversion. The calculated fractional conversions are then applied to Avrami model as discussed in Chapter 2 to determine the re action rate parameters. Figure 8 13 compares the Avrami plot for selenization of precursors done with different sodium content.

PAGE 267

267 The rate constant obtained from the Avrami model are then plotted against i sothermal temperatures (Figure 8 14 ) using Arrhenius relationsh ip given by, (8 6 ) With increase in sodium content, the rate of formation of CIGS increases and higher rate is obtained for precursor selenized with NaMo 5 With MoNa 3 not much change is observed in kinetics compared with sodiu m free sample. The activation energy estimated from Avrami model gave 92 (12) kJ/mol, 103 (11) kJ/mol and 113 (20) kJ/mol for CIGS formation from selenization of SFG/Mo/CuGa/In, SFG/MoNa 3 /Mo/CuGa/In and SFG/MoNa 5 /Mo/CuGa/In respectively. Rudmann et al have reported that sodium influences the gallium and indium distribution during the second stage of the three stage co evaporation process [189] Simi lar effect is observed in the same study during selenization of metallic precursor forming a graded structure with increase in the kinetics of CIGS formation as opposed to decrease in kinetics reported by Kim et al. for bilayer stacked In 2 Se 3 /CuSe precurso r [57] The effect of sodium on kinetics is thus a function of precursor structure, sodium content and the type of process used for forming abs orber layer. Hergert et al. have reported that sodium presence forms sodium polyselenides on the surface assisting selenization of Cu Ga In precursor [160] Summary Temperature ramp selenization was fo llowed for CuGa/In precursor deposited on sodium free and sodium doped molybdneum substrates. There was no change in the reaction pathway in the presence of sodium within the detectable limits of X ray

PAGE 268

268 diffraction. The reaction rate dependence on sodium content was studied by performing isothermal selenization studies. With increase in the doping concentration of molybdenum, increase in the rate was measured quantitatively using Avrami solid state growth model. Improvement in the microstructure was obser ved for precursor having NaMo 5 as the underlying sodium supply layer. Another effect of increased sodium concentration in the MoNa was the formation of MoSe 2 at 450 C compared with 540 C for sodium free substrate. Hence, sodium helps in reducing the pr ocessing temperature for selenization of CuGa/In precursor when supplied through sodium doped molybdenum layer. The effect of sodium content on device performance and defects is in progress and will be reported in future studies.

PAGE 269

269 Table 8 1. P hase fraction obtained from Rietveld refinement using high resolution XRD data Precursor Structure Phases Identified Phase fraction (%) x(Ga) % SFG/Mo/CuGa/In CuIn x Ga 1 x Se 2 CIS Mo MoSe 2 75 12.5 8.1 4.4 0.19 0.30 SFG/NaMo 3 /Mo/CuGa/In CIGS CIS Mo MoSe 2 74 .3 2.2 18.7 4.8 0.23 0.27 SFG/NaMo 5 /Mo/CuGa/In CIGS CIS Mo MoSe 2 72.1 8.2 13.9 5.9 0.23 0.25

PAGE 270

270 Figure 8 1 Room temperature X ray diffraction scans of as deposited CuGa/In precursor. A) precursor deposi tion on sodium free substrate, B) pre cursor deposition on MoNa 3 substrate, C) precursor deposition on MoNa 5 substrate A B C

PAGE 271

271 Figure 8 2 SEM image of a s deposited CuGa/In precursor. A) precursor deposi tion on sodium free substrate, B) precursor deposition on MoNa 3 substrate, C) precu rsor deposition on MoNa 5 substrate A B C

PAGE 272

272 Figure 8 3 Temperature ramp selenization sequence of CuGa/In precursor deposited on sodium free substrate

PAGE 273

273 Figure 8 4 Temperature ramp selenization sequence of CuGa/In precursor deposited on MoNa 3 substrate

PAGE 274

274 Figure 8 5 Temperature ramp selenization sequence of CuGa/In precursor deposited on MoNa 5 substrate

PAGE 275

275 Figure 8 6 Room temperature X ray diffraction of temperature ram p selenized CuGa/In precursor. A) on MoNa 5 substrate B) on MoNa 3 s ubstrate, C) on sodium free substrate A B C

PAGE 276

276 Figure 8 7 SIMS depth profile of temperature ram p selenized CuGa/In precursor. A) on MoNa 3 substrate, B) on MoNa 5 substrate A B

PAGE 277

277 Figure 8 8 Ga/III profile for temperature ramp selenized precursor depos ited on sodium doped molybdenum substrates

PAGE 278

278 Figure 8 9 SEM images of temperature ramp selenized precursor. A) surface image on sodium free sub strates, B) cross sectional image on sodium free substrates, C) surface image on MoNa 3 D ) cross sectional image on MoNa 3 E) surface image on MoNa 5 F) cross sectional image on MoNa 5 A B C D E F

PAGE 279

279 Figure 8 10 Isothermal selenization sequence of CuGa/In precursor deposited on sodium free substrates at different temperature A) 260 o C, B) 280 o C, C) 300 o C, D) 320 o C D B A C

PAGE 280

280 Figure 8 11 Isothermal selenization sequence of CuGa/In precursor deposited on MoNa 3 at different temperature A) 260 o C, B) 280 o C, C) 300 o C, D) 320 o C C D B A

PAGE 281

281 Figure 8 12 Isothermal selenization sequence of CuGa/In precu rsor deposited on MoNa 5 at different temperature A) 260 o C, B) 280 o C, C) 300 o C, D) 320 o C C D B A

PAGE 282

282 Figure 8 13 Avrami plot using fractional conversion data from isothermal selenization experiments. (A) sodium free substrates, (B) MoNa 3 (C) MoN a 5 A B C

PAGE 283

283 Figure 8 14 Arrhenius plot using rate constants obtained from Avrami model

PAGE 284

284 CHAPTER 9 NANOPARTICLE ROUTE F OR SYNTHESIS OF CIS ABSORBERS Overview One of the approaches to reduce the cost of manufacturing is by increasing the throughput of the system. The formation of CIS is a diffusion controlled process and the diffusion distances can be reduced by employing a nanoparticle based approach. A nanoparticle based route has the potential to increase the throughput of the system and reduce th e cost. Several researchers have used nanoparticle based processes to synthesize CIGS [95, 190 193] Guo et al have synthesized quaternary nanoparticles of CIGS (sulfur based) and then fabricated the device by an nealing it in a selenium environment to substitute the sulfur sub lattice with selenium [194] An efficiency of 5.55 % has been reported using the same process. Yoon et al used the core shell of InSe/CuSe nanoparticles to synthesize CIS with an efficiency of 1.11% [178] Kapur et al coated oxide nanoparticles of Cu In Ga and then reduced the oxide to form the metal on the substrate [92] The metal is then selenized in toxic H 2 Se to form the absorber layer. Most of the processes discussed above use binders to avoid particle a ggregation during high solid loading of nanoparticles. Dispersion of the particles in solution is necessary for uniform coatings on the substrate. During the annealing step, the residues of binders remain in the film, thus affecting the efficiency of the final device. In addition to the annealing time in a selenium atmosphere, the film has to undergo heat treatment in argon for binder removal [178] Control of the absorber grain size, structural quality, texture, and composition profile in the growth direction is important to achieve reliable and high efficiency devices.

PAGE 285

285 In this chapter, CIS nanoparticles with secondary phases, such as CuSe and CuSe 2 are used for rapid formation of CIS by peritectic decomposition of CuSe and CuSe 2 Also, the simulation of the bilayer process using bilayer compound nanoparticles synthesized by the colloidal route is discussed. In situ p hase transformation of nanoparticles is followed as a function of temperature to understand the reaction pathways. Experimental The nanoparticle synthesis is carried out by simple reaction in alcoholic medium. Figure 1 shows the schematic of the synthesis procedure. Copper chloride (CuCl, 0.01 mole) and indium chloride (InCl 3 0.01 mole) are prepared separately in 20 ml ethanol and 25 ml propanol, respectively, at room temperature and stirred for 24 hours. The resulting solutions are then mixed for 2 hou rs and refluxed at 70 C under inert argon atmosphere. A selenium (Se, 0.02 mole) solution is prepared in 40 ml of ethylenediamine at room temperature and mixed with the copper and indium chloride solution. The mixture containing copper, indium and selenium is refluxed under a n inert argon atmosphere at 110 C for 5 hours to obtain CIS nanoparticles, along with some secondary phases. The resulting nanoparticles are then centrifuged with methanol at 10,000 RPM and dried at 115 C under vacuum. Depending on the precursor and its ratio, the nanoparticle can be synthesized either in copper rich or copper poor compositions as summarized in Table 9 1 Grain growth experiments were done with an overpressure of selenium in system. Selenium was loaded in the recess area of the sample holder and covered with graphite dome Initially, the sample was scanned at room

PAGE 286

286 temperature and then the temperature was ramped to 150 C (at 15 C/min) and data was collected for every 10 C increment. After 150 C the ramp rate was increased to 40 C/min. CIS nanoparticles were loaded on a quartz plate and the sample holder was covered with a graphite dome. The scan length at each tem perature was approximately ~3 minutes. The system was purged with forming gas to avoid oxygen interference during the reactions. Temperature Ramp Annealing Copper Rich Nanoparticles The nanoparticle composition was determined by inductively coupled plasma optical emission spectroscopy (ICP OES). The nanoparticle was copper rich as shown in Table 9 1. Room temperature X ray diffraction of the as synthesized nanoparticles showed CIS(cubic), CuSe 2 (orthorhombic), and excess selenium which is consistent with the ICP results as shown in Figure 9 2 Figure 9 3 A shows the particle size around ~30 50 nm from a high resolution transmission electron microscope (HR TEM) image. The dark spots in the image indicate the excess copper selenide phase. Temperature ramp a nnealing was carried out with the procedure as discussed above. Figure 9 4 shows the temperature ramp annealing of the copper rich nanoparticle. With an increase in temperature, selenium melts at its melting point (221 C). CuSe 2 continues to grow till 330 C, after which it undergoes peritectic decomposition as detected by the CuSe phase. The CuSe phase formed is polycrystalline with no preferred orientation. According to Cu Se phase equilibria, CuSe 2 decomposes to Cu Se and selenium rich liquid at 331.8 C which is consistent with the observation. It is expected that in the nanoparticles based system, transformation temperatures should

PAGE 287

287 be lower because of high surface energy. Tetragonal CIS formation occurs at 260 C and the cubic to tetragonal transformation occurs with no bond breaking as shown in Figure 9 5. Cubic CIS refers to the structure where the copper and the indium atoms are distributed randomly within the crystal structure. With a further increase in temp erature, CuSe decomposes to Cu 2 Se and a selenium rich liquid via a second peritectic reaction at 380 C consistent with the thermodynamic temperature of 379.5 C. After the second peritectic reaction, the rate of formation of CIS is rapid as observed by an increase in the crystall inity of the ( 112 ) orientation. It is evident from the results that the grain growth of CIS follows a vapor liquid solid growth mechanism. The SEM image showed significant grain growth after temperature ramp annealing of the nanoparticles as shown in Fig ure 9 6. The following reactions are observed during the temperature ramp, (9 1) (9 2) (9 3) Copper Poor Nanoparticles Temperature ramp annealing of these na noparticles was carried out under similar copper rich nanoparticles The room temperature XRD showed reflections of CIS (cubic), CuSe ( h exagonal), InSe ( h exagonal), and selenium as shown in Figu re 9 7. The nanoparticle composition was copper poor as indicated by the Cu/In ratio in Table 9 1. Since the selenium in the nanoparticle was in excess, no additional selenium pressure was supplied. The particle size was around ~100 nm as shown in Figure 9 3 B Figure 9 8 shows the temperature

PAGE 288

288 ramp annealing of the copper poor nanoparticles. During the temperature ramp annealing, selenium melted at its melting point followed by the formation of CIS at 300 C. The growth of CIS was complete at 380 C as observed by no change in the intensity or FWHM. After the completion of CIS formation, growth of In 2 Se 3 takes place in the ( 202 ) orientation It has been reported that an ordered vacancy compound is formed when grown under indium rich conditions. No ord ered vacancy compound was detected during the ramp. The following reaction products are detected during temperature ramp, (9 4) (9 5) (9 6) Nearly Stoichiomet ric Nanoparticle The nearly stoichiometric nanoparticle had a Cu/In ratio of 30% more than stoichiometry as shown in Table 9 1. The selenium to metal ratio was low, and hence a selenium overpressure was supplied during the temperature ramp. The room t emperature XRD (Figure 9 9 ) showed reflections of CIS (tetragonal), CuSe (hexagonal), InSe (hexagonal), and selenium. The TEM image (Figure 9 3 C ) showed a mesh like structure and the diameter o f the mesh was around ~20 nm. The temperature ramp was perf ormed in a similar way as discussed before and is shown in Figure 9 1 0 Selenium melted at its melting point, followed by grain growth of CIS at 240 C. It is believed that the formation of CIS is accompanied by the reaction of InSe and CuSe. The format ion of CIS is complete at 300 C as no change in peak intensity is observed after 300 C. After CIS formation, a reflection of Cu 2 Se is also observed which is

PAGE 289

289 consistent with the literature. It has been reported that the Cu 2 Se phase is formed when CI S is grown under copper rich conditions [65] The following reactions occur during temperature ramp (9 7) (9 8) (9 9) I nk Formulation and Film Formation The nanoparticles were synthesized in a polar sol vent such as ethanol and propanol. The nanoparticle synthesis was scaled up and the composition of the nanoparticles was made to be copper rich. Table 9 2 shows the ICP results of the scale up of the nanoparticle synthesis. Since no surfactant was used dur ing synthesis, the nanoparticles were polar in character because they were prepared from an alcoholic source. Different solvents were tried to prepare the ink for the coating of nanoparticles on the substrates. The solvent selection was done on the basis of boiling point, viscosity, and particle dispersion. It is necessary to boil off the solvent after coating the nanoparticles onto the substrate. To reduce the heat load in this step, a low boiling point solvent is preferred. The viscosity of the solvent also plays an important role in coating. It is difficult to coat the particles with solvents having a similar viscosity to water. It is also necessary that particles in the solvent do not aggregate. It would be ideal that the dispersion of particles occ urs in the liquid phase during the synthesis procedure. This type of ink formulation has been reported for sulfur based synthesis of nanoparticles. In order to prepare nanoparticle dispersions from dry materials, the

PAGE 290

290 affinity between solvent and the surfa ce of the particle (also termed as wetting of nanoparticles) is important. The wettability of the nanoparticles was measured by the magnitude of dispersion of the nanoparticles in different solvents. Table 9 3 shows the list of solvents used to measure d ispersion for ink formulation. The nanoparticles settled down in non polar solvents such as toluene and hexane suggesting the surface group of the nanoparticles to be polar. Of all the solvents used, isobutanol provided the best dispersion of nanoparticl es; hence it was used for ink formulation. The ink was drop casted onto molybdenum coated glass after 1 hour of ultrasonication. The as prepared sample was kept in an oven at 70 C for 6 hours to remove the solvent. The film was annealed in a custom built tube furnace with selenium overpressure (provided in gas phase and transported by nitrogen). The centerline temperature of the tube furnace was measured and a calibration curve was constructed as shown in Figure 9 1 1 First, the sample was heated to 110 C in forming gas for 30 minutes and then annealed at 500 C with selenium transported from the source at 400 C. After several trials, grain growth of the UF11 1 series was f ound as shown in Figure 9 1 2 The discontinued structure in the grain is because of poor coating of nan oparticles onto the substrate. It is seen that the CIS formed had a (112) preferred orientation as shown in the XRD spectrum (Figure 9 13). The sam e solution was now coated using a commercial ultrasonic spray coater and a uniform grain size of ~3 m was obtained (Figure 9 14). As discussed above, under copper rich conditions the CIS is in equilibrium with Cu 2 Se and the removal of Cu 2 Se is essential, either by cyanide etching or by adjusting the composition by adding indium, for device fabrica tion. KCN etching was performed on one such sample and XRD showed a decrease in the ( 112 )

PAGE 291

291 orientation with no change in the ( 220 ) orientation indicating lattice matching of cubic Cu 2 Se (0.576 nm) with tetragonal CIS (.578 nm) [ICDD reference pattern given in Figure 9 15]. It has been reported that Cu 2 Se has an epitaxial relationship with CIS in the island region of Stranski Krastanov growth mode for the samples grown by MBE under copper rich conditions. Device fabrication using these nanoparticles is in progress and will be discussed in future wo rk. CIGS Device Fabrication using Binary Nanoparticles The NREL 3 stage co evaporation process produced the champion cell and replicating the process in large scale is still a challenge because of low throughput and uniformity issues. The co evaporation p rocess is dominated by capital costs because of its unique design to attain uniform flux of elements over large area substrates. As discussed in Chapter 1, the process is divided in three stages and growth of CIGS occurs in the tie line connecting In 2 Se 3 and Cu 2 Se. In Chapter 3 it was shown that the kinetics of absorber formation is utilizing bilayer compound precursors. In an attempt to simulate the NREL 3 stage process, binary nanoparticles of (In, Ga) 2 Se 3 and Cu 2 Se were synthesized by the colloidal route. It is believed that in the first stage of the NREL 3 stage process, the solid solution (In, Ga) 2 Se 3 is formed and has a hexagonal crystal structure. The synthesis procedure is a hot injection method which is a widely used approach for synthesis of nanocrystals [195 197] The advantage of going to binary particles from ternary particles (as mentioned in the literature) is control of the composition over large area substrates. The other advantage is in the s ynthesis itself as synthesis of ternary particles involves maintaining supersaturation conditions and good mixing to ensure the product formed has only the

PAGE 292

292 ternary phase. While maintaining the synthesis to binaries, depending on the reaction conditions, s ingle phase reaction products can be obtained on high yield. The only disadvantage of the hot injection method is the residue of the coordinating solvent that remains in the nanoparticle which can affect efficiency of the devices compared to the PVD proce sses. (In,Ga) 2 Se 3 Nanoparticle Synthesis 0.15 M of InCl 3 was prepared in 2.5 ml of oleylamine and 0.05 M of GaCl 3 was prepared in 2.5 ml of oleylamine under nitrogen ambience. 0.25 M of selenium in 4 ml of oleylamine was prepared separately and added to th e reaction mixture of InCl 3 and GaCl 3 in a three neck flask. The mixture was freezed, pumped, and heated (4 cycles) to remove volatile impurities. The temperature of the reaction mixture was raised to 265 o C under argon atmosphere for 1 hour. The initial black solution turned brick red after 10 minutes of heating with an increase in solution viscosity over that period of time. The reaction mixture was quenched using toluene and ethanol. The reaction product was washed with ethanol and centrifuged for 10 m in at 8,000 rpm (4cycles) and dried overnight at 70 o C. The particles were then characterized by XRD (phase identification) and TEM (particle size and shape). The particles were then redispered into a nonpolar solvent such as hexane or toluene for ink for mation. The XRD pattern obtained for as synthesized (In, Ga) 2 Se 3 nanoparticles is shown in Figure 9 16 The phases correspond to In 2 Se 3 with peak shift owing to gallium addition in the group III sub lattice. The ICP results showed x ( Ga) =0.08, x (In) =0.31, and x (Se) =0.61 which is consistent with the results of XRD. The gallium content in the group III sub lattice corresponds to ~20 21%. The transmission electron microscope (TEM) image shows

PAGE 293

293 nanoparticles with distortions in the regular hexagonal structure with a size of 60 ( 10) nm (Figure 9 17 ). Cu 2 Se Nanoparticle Synthesis 0.1 M of CuCl solution was prepared in a 10ml mix ture of oleylamine and 1 octadecene or pure oleylamine. 0.5 M of selenium was prepared separately in oleylamine. The CuCl solution mixture was degassed under vacuum at 80 o C for 30 minutes and then the reaction temperature was raised to 280 320 o C under a n argon atmosphere. Once the temperature reaches the set point, 1ml of selenium solution was injected into the reaction mixture. A sudden change in viscosity was observed and the reaction was continued for another 15 20 minutes. The product obtained was washed with an ethanol and iso propanol mixture and centrifuged for 10 minutes at 10,000 rpm. The washing procedure was repeated 4 times to ensure removal of octadecene. ICP analysis was performed and the mole fraction of selenium [x(Se)] obtained was 0 .32, which is consistent with the Cu 2 x Se phase observed in Cu Se phase diagram. Figure 9 18 shows the room temperature X ray diffraction, revealing the cubic phase of Cu 2 x Se and a particle size around 20 (5) nm (Figure 9 19 ); indicating good cont rol over the size. Rapid Thermal Annealing The as synthesized nanoparticles were mixed together in a ratio such that the overall composition yielded copper poor conditions[x (Cu) =0.24, x (In) +x(Ga) = 0.26, x (Se) =0.5]. The particles were dispersed i n toluene and the ink was reddish black in color. The dispersion of the particles was a problem due to a difference in the particle size and molecular weight of the binaries. The ink was drop casted on the molybdenum

PAGE 294

294 coated glass substrate and dried over night at 70 o C. The precursor sample was annealed at 250 o C for 5 minutes and then the temperature was gradually raised to 500 o C and was held for 10 minutes. This procedure led to an incomplete reaction upon annealing because of the ink dispersion and t he results could not be replicated. Hence, a new procedure was followed replicating NREL 3 stage process. First, (In, Ga) 2 Se 3 was coated and allowed to dry followed by a coating of Cu 2 Se. The procedure was repeated till the thickness was approximately 1 1.5 m. The sample was then dried in an oven at 70 o C to remove the solvent. The sample was then rapid thermal annealed in a selenium atmosphere for 5 minutes at 500 o C to form CIGS. The XRD pattern showed pure CIGS (tetragonal phase) with MoSe 2 forma tion (Figure 9 20 ). Device fabrication was completed by depositing 50 nm of n type CdS by chemical bath deposition, followed by sputtering i ZnO (50 nm) and 500 nm of aluminum doped zinc oxide (AZO) on the device. The Ni/Al (50/300 nm) electrical contact was deposited by e beam deposition using a shadow mask and had an efficiency of 1.68%. Figure 9 21 shows the I V characteristics of the fabricated solar cell. Summary The reaction pathway was investigated for three different bilayer precursor films using in situ high temperature X ray diffraction. The effect of Se pressure on phase transformation or peritectic decomposition has been elucidated. The reaction pathway for each precursor structure differed at low temperature, but the CIS synthesis reaction was the same. The rate for the In 2 Se 3 / CuSe+ Cu 2 Se precursor was slightly higher than that for the In 2 Se 3 / Cu 2 Se one. A quantitative model was established for all the precursor films using the Avrami and parabolic solid state growth models. Based on the

PAGE 295

295 modeling results, the glass/Mo/ In 2 Se 3 / Cu 2 Se /Se precursors follows a one dimensional diffusion controlled reaction but the result for the glass/Mo/ In 2 Se 3 / CuSe+ Cu 2 Se /Se precursor are consistent with a nucleation process followed by diffusion controlled growth For glass/Mo/ ( In G a) 2 Se 3 / CuSe the precursors follows one dimensional diffusion controlled transformation consistent with the obtained Avrami exponents.

PAGE 296

296 Figure 9 1 Schematic of CIS nanoparticle synthesis

PAGE 297

297 Figure 9 2 Room temperature X ray di ffraction pattern of as synthesized nanoparticles

PAGE 298

298 Figure 9 3 TEM images of as synthesized nanoparticles. A) copper rich nanoparticle, B) copper poor nanoparticle, C) nearly stoichiometric nanoparticle A B C

PAGE 299

299 Figure 9 4 Temperature ramp sequen ce of copper rich nanoparticle with selenium overpressure

PAGE 300

3 00 Figure 9 5 Phase transformation of CIS (cubic) to CIS (tetragonal)

PAGE 301

301 Figure 9 6 SEM image showing grain growth in temperature ramp copper rich nanoparticle

PAGE 302

302 Figure 9 7 Room temperature XRD of as synthesized copper poor nanoparticle

PAGE 303

303 Figure 9 8 Temperature ramp sequence of copper poor nanoparticle with selenium overpressure

PAGE 304

304 Figure 9 9 Room temperature X ray diffraction of as synthesized nearl y stoichiometric nanoparticle

PAGE 305

305 Figure 9 10 Temperature ramp sequence of nearly stoichiometric nanoparticle with selenium overpressure

PAGE 306

306 Figure 9 11 Temperature calibration of selenization tube furnace

PAGE 307

307 Figure 9 12 SEM imag e of CIS na noparticle deposited on glass. A) before annealing, B) after annealing A B

PAGE 308

308 Figure 9 13 X ray diffraction pattern of annealed film obtained from CIS nanoparticle

PAGE 309

309 Figure 9 14 SEM image showing grain growth of annealed film ob tained from CIS nanoparticle

PAGE 310

310 Figure 9 15 X ray diffraction pattern of CIS before and after KCN etching

PAGE 311

311 Figure 9 16 X ray diffraction pattern of as synthesized (In,Ga) 2 Se 3 nanoparticles

PAGE 312

312 Figure 9 17 TEM image of as synthesiz ed nanoparticles of (In,Ga) 2 Se 3

PAGE 313

313 Figure 9 18 X ray diffraction pattern of as synthesized Cu 2 Se nanoparticles

PAGE 314

314 Figure 9 19 TEM image of as synthesized nanoparticles of Cu 2 Se

PAGE 315

315 Figure 9 20 X ray diffraction pattern of CIGS film o btained from rapid thermal annealing of nanoparticles

PAGE 316

316 Figure 9 2 1 J V curve of device fabricated from nanoparticle approach

PAGE 317

317 CHAPTER 10 DEVICE DEGRADATION S TUDIES OF CIGS SOLAR CELLS Overview Within the last two years, tremendous pro gress has been made in improving the performance of CuIn 1 x Ga x Se 2 based polycrystalline solar cells and modules. The champion cell efficiency has exceeded 20% [14] and a module efficiency of 15.7 % has been reported The device structure commonly incorporates an n type CdS buffer layer using chemical bath deposition (CBD). It has been reported that interdiffusion occurs across the CIGS/CdS interface, dominated by Cd +2 exchange with Cu +1 in the Cu poor near surface r egion of CIGS. It is further suggested that the near surface region of the CIGS absorber is converted to n type, thus burying the electrical junction below the metallurgical one to reduce recombination at interfacial defects [198] Nakada et al. have observed diffusion of Cd in the first 10 nm of the CIGS surf ace using small spot EDX measurements [16] The high diffusivity of Cd into Cu poor CIGS is consistent with the associated high V Cu concentration. In addition, Kijima and co workers have reported diffusion of Cd and Zn from the sulfides into CIGS using SIMS and EBIC measurements [199] The same group also reported dramatic reduction in cell performance after annealing the device at 400 C accompanied by diffusion of Cd throughout the full thickn ess of the CIGS film. Unfortunately, the exact failure mechanism is unknown as well as quantitative rate expressions describing the interdiffusion and reaction process to allow estimating the mean time to failure. T his Chapter reports in situ high tempera ture XRD measurements of a CIGS device stack, CuCd 2 (Ga x In 1 x )Se 4 at elevated temperature for the first time.

PAGE 318

318 Experimental CIGS thin film structures SS/Mo/CIGS/CdS and SS/Mo/CIGS/CdS/ITO were obtained from Global Solar Energy, I nc. The CIGS was deposited using a roll to roll production tool, followed by chemical bath deposited CdS. Sputtered ITO was used as the transparent conductor and no top contacts were deposited. Time resolved X ray diffraction studies were carried out in a 2 or forming gas (4% H 2 /N 2 mixture) ambient. The system includes an X ray tube and a solid state detector along with an Anton Paar hot stage. The temperature was controlled by a PID controller and the measurement thermocoupl e was calibrated by comparing the measured lattice expansion of silver to the temperature dependence reported in the literature [123] The system was purged with N 2 for removal of oxygen from the system prior to measurement. Tempera ture Ramp Annealing The room temperature high resolution XRD patterns of the as grown and annealed SS/Mo/CIGS/CdS sample are shown in Figure 10 1. As anticipated the room temperature scan contained reflections associated with CIGS, Mo and MoSe 2 (Figure 1A ) The reflections of the thin CdS buffer layer were not observed in the scan. As an initial experiment, a series of lower resolution diffraction patterns was collected in 10 C increments (~1 min scan time) over the temperature range 25 to 600 C. The s ample temperature was increased at a ramp rate of 10 C/min to avoid temperature overshoot. The room temperature high resolution diffraction pattern for the same sample after step annealing to 600 C is also shown in Figure 1 B In addition to the reflecti ons observed in the room temperature scan, additional peaks were evident. These reflections,

PAGE 319

319 labeled in Figure 1 B were assigned to the solid solution CuCd 2 (Ga x In 1 x )Se 4 which exhibits a cubic (sphalerite) crystal structure as reported in the ICDD database (PDF# 04 001 4706). It is noted that the lattice spacing of the three observed reflections planes are closely matched to those of CIGS. Interestin gly, the peak positions did not match well with sulfide of the stoichiometry. The FWHM of the reflection from the (112) consistent with increased crystallinity. Figure 10 2 sh ows the set of XRD patterns obtained upon temperature ramp annealing of the SS/Mo/CIGS/CdS/ITO test structure in a N 2 atmosphere. With increase in temperature, no discernible change was observed until 400 C, at which CuCd 2 (Ga x In 1 x )Se 4 CuCd 2 (Ga x In 1 x )Se 4 increased with increasing CuCd 2 (Ga x In 1 x )Se 4 A decrease in the intensity of the (112) reflection of CIGS was CuCd 2 (Ga x In 1 x )Se 4 while consuming CIGS. A temperature ramp study was also conducted for a sample with an ITO top layer (SS/Mo/CIGS/CdS/ITO) using the same conditions as applied for annealing the sample with data shown in Figure 10 2. The temperature ramp anneal of the structure capped with ITO is shown in Figure 10 3. No change in the reaction pathway was observed with ITO present suggesting ITO does not react with the buffer layer to 600 C and that Se loss was not a factor. The same ITO capped structure was also examined in a temperature ramp study under reducing conditions. Figure 10 4 shows the temperature ramp experiment under a forming gas flow (4% H 2 /N 2 mixture). No change w as evident

PAGE 320

320 in the reflections attributed to ITO. CuCd 2 (Ga x In 1 x )Se 4 merged into the (112) peak of CIGS. At high temperature, ~580 C, complete overlap was observed. CuCd 2 (In,Ga)Se 4 shows not only complete miscibility on the group III sublattice but also significant Cd and to a lesser extent Cu solubility [200, 201] Thus increasing the Ga or Cu content would contract the lattice to match that of CIGS. The role of the forming gas on the compound formation is not understood and further investigations are in progress. SIMS analysis was carried out for as deposited and sample quenched at 460 C. Figure 10 5 shows the SIMS profile for both the as deposited and quenched sample. Before annealing, a sharp signal corresponding to CdS is seen in the Figure 10 5A After quenching, cadmium diffusion is observed along with a dip in gallium profile at t he surface indicating some compound formation consistent with the X ray diffraction measurements (Figure 10 5B) Isothermal Annealing and Reaction Kinetics Isothermal experiments were carried out in the temperature range of 420 to 480 C to measure the rate of formation of CuCd 2 (Ga x In 1 x )Se 4 using the SS/Mo/CIGS/CdS structure. The time evolutions of the diffraction patterns at four temperatures are shown in Figure 10 6 and noting that the time of the experiment at 480 C was slightly less since the reaction rate was faste r. As expected, these figures show a progression in the time of the appearance of the peak associated with the new compound as well as the subsequent rate of increase of the peak height as isothermal temperature is increased. These data were then analyzed by applying the Avrami model [165] to the measurements. The fractional conversion of CdS CuCd 2 (Ga x In 1 x )Se 4

PAGE 321

321 CuCd 2 (Ga x In 1 x )Se 4 as a function of time to that measured after complete reaction. Figure 10 7 shows the Avrami plot at the four isothermal me asurement temperatures The rate constants estimated using the Avrami model are plotted in Figure 10 8 vs. CuCd 2 (Ga x In 1 x )Se 4 formation. Values of 233.5 (45) kJ/mol and 8.310 13 s 1 were est imated for the activation energy and pre exponential factor, respectively. These values were then CuCd 2 (Ga x In 1 x )Se 4 at normal operating temperature (50 o C) for a period of 30 years. The estimated frac tional conversion was essentially zero, and thus should not be a failure mechanism under normal operation. Summary A typical CIGS cell structure on stainless steel (SS/Mo/CIGS/CdS/ITO) was thermally stressed at elevated temperature and the appearance or disappearance of phases was observed using in situ high temperature XRD. The only new phase to appear during ramp annealing to 600 C was the solid solution CuCd 2 (Ga x In 1 x )Se 4 The onset of its formation was apparent in XRD beginning at 400 C. This result identifies the compound that was responsible for the previously reported device degradation. Isothermal annealing experiments were carried out and the r ate parameters for compound formation were determined using the Avrami model. Activation energy of 233.5 ( 45) kJ/mol and pre exponential factor 8.310 13 s 1 were obtained for the compound formation. Given the relatively high activation energy for

PAGE 322

322 format ion of CuCd 2 (Ga x In 1 x )Se 4 it is not surprising that the compound is not expected to form during the lifetime of a module under normal operating temperature Figure 10 1 X ray diffraction pattern of as received and temperature ramp annealed sample of SS/ M o/CIGS/CdS. A) temperature ramp annealed sample, B) as received sample A B

PAGE 323

323 Figure 10 2 Temperature ramp sequence during annealing of SS/Mo/CIGS/CdS sample in nitrogen environment

PAGE 324

324 Figure 10 3 Temperature ramp sequence during annealing of S S/ Mo/CIGS/CdS/ITO sample in nitrogen gas environment

PAGE 325

325 Figure 10 4 Temperature ramp sequence during annealing of SS/Mo/CIGS/CdS/ITO sample in forming gas environment

PAGE 326

326 Figure 10 5 SIMS depth profile of as received an d quenched sample at 460 C. (A) as received sample, (B) quenched sample A B

PAGE 327

327 Figure 10 6 Isothermal annealing of SS/Mo/CIGS/CdS sample at different temperatures A) 420 o C, B) 440 o C, C) 460 o C, D) 480 o C. A B C D

PAGE 328

328 Figure 10 7 Avrami plot using fraction conversion from isothermal annealing experiments

PAGE 329

329 Figure 10 8 Arrhenius plot using rate constants obtained from Avrami model

PAGE 330

330 CHAPTER 11 ELLINGHAM DIAGRAM OF CU IN GA SE O SYSTEM AND SELENIU M TRANSPORT STUDIES FO R GROWT H OF CIGS Overview CIGS solar cells fabricated using NREL 3 stage processes have achieved highest lab scale efficiency and highest module efficiency has been reported for metal selenization process. Higher success in manufacturing has been achieved for met al selenization process. The selenization is essentially a two step process, where the chalcogen (selenium or H 2 Se) is supplied to oxidize the metal deposited in the first step. Higher overpressure of selenium is required to synthesize CIGS owing to the h igher vapor pressure of selenium. Selenium supply itself is not a problem, but excess selenium deposition occurs at the cold spots in the chamber causing significant problem with the process control. Sputtering the metals with selenium similar to co evap oration process can solve the problem to some extent. However contamination of targets has been reported using this route [202] Absence of selenium during sputtering of metals has shown to improve the utilization of indium owing to higher vapor pressure of In 2 Se [203] The common supply method employed by majority of industries is H 2 Se which is highly toxic and can add a downtime in processing after selenization Variation in the process can include such as deposition of metals (order of deposition) and kinetics associated with reacting the metals with selenium. Nevertheless it is important to have an estimate of equilibrium vapor pressure of selenium over the different compounds formed during annealing step. In this section, we discuss the Ellingha m diagram for oxides and selenides of Cu In Ga system. Ellingham diagram has been reported for oxides of other systems in the literature [204] The diagram gives the information such as thermodynamic driving force for a given set of reactions as a

PAGE 331

331 function of temperature. It is also used to determine the relative stabilities of the element for oxidation in oxidizing atmospheres such as oxygen, selenium and sulfur. Thermodynamics According to second law of thermodynamics, at constant pressure and temperature, Gibbs energy of t he system should be minimum. The Gibbs energy can be defined as (11 1) Where, H is the enthalpy in kJ/mol and S is the entropy in kJ/mol K. The enthalpy can be written as (11 2) Wh ere, U is the internal energy kJ/mol, P the pressure and V being the volume. The change in the free energy, when the system is changed by infinitesimal amount can be obtained by differentiating equations 11 1 and 11 2, (11 3) (11 4) Applying first and second law, (11 5) Where, Q is the heat input to the system and W being the work done. and (11 6) Where S is the entropy of the system Substituting equations 11 6, 11 5 and 11 4 in 11 3 we get (11 7)

PAGE 332

332 At constant temperature, equation 11 7 reduces to (11 8) Assuming ideal gas law, we can express volume as a function of pressure for 1 mole of gas by (11 9) Substituting 11 9 in 11 8 and integrating equation 11 8, we get (11 10) Where G o is the standard free en ergy at the standard pressure P o For a reaction, A(s) + B(g) C(s) the change in the free energy of the system can be expressed as, (11 11) Where A is a metal, B is oxidizing gas, C the condensed phase product free energy change. If the free energy of the products is less than free energy of the reactants then there will be driving force for the reaction to happen. For a closed system, the concentration of reactants decrease and the concentration of the product increases. The change is the free energy of the system is then determined by relative quantities of reactant and product (equation 11 11). The reaction continues as long as the free change is negative. The system undergoes a free energy gradient and this gradient is a function of composition providing driving force for the reaction. The reaction proceeds until global minima in free energy are reached and the gradient

PAGE 333

333 (slope) becomes zero. The global minimum point is termed as equilibrium and at equilibrium; free energy change for the reaction is zero. Hence equation 11 11 becomes (11 12) For the reaction at equilibrium, the quotient in equation 12 is equilibrium constant K p for the reaction at constant pressure. The El o of the reaction as a function of temperature. The graph is a straight line as obtained from equation 11 13 and gradient of the lines is the standard entropy change for the reaction. The positive slope of the line indicates a l arge decrease in entropy involving elimination of gas. The reactions generally involve the reaction of gas phase with the condensed phase (metals or oxidized compounds). Using this diagram, the standard free change of any reaction can be found at any tem perature and is plotted against temperature. (11 14) Using equation 12, the equilibrium constant is given by (11 15) At equilibrium, the total free energy of compou nd in vapor phase must be equal to the free energy in the solid phase. From the law of mass action, the equilibrium constant is related to activity of the species. For the reactions discussed above, the metal and metal oxides are in their standard state a nd therefore the activity can be taken as 1. The standard state of the gas is defined as 1 atm and hence the equilibrium constant is inversely proportional to the pressure of the gas. Hence equation 11 14 can be re written as,

PAGE 334

334 (11 15) The equilibrium partial pressure of gas over the condensed phase product can be found out using (11 16) The equilibrium partial pressure of the gas is the pressure at which the driving force for the reaction is zero. From equation 1 1 16 if the partial pressure of the gas is greater than the driving force requirement, then the free energy change for the reaction is negative and the metal gets oxidized and the partial pressure of the gas drops down pressure of gas is below the equilibrium value, oxidation of the metal does not take place. The equilibrium partial pressure is obtained by select ing a temperature T and drawing a line through 0K. The slope of the line is related to partial pressure of the gas. E llingham Diagram for Cu In Ga Se O S ystem As discussed in the introduction of this chapter, there are two different processes for synthesi zing CIGS, co evaporation and selenization. In a co evaporation process, the typical base pressures are 10 6 Torr which provides sufficient mean free path between the evaporation source and substrate. Operating at low pressures also minimizes oxidation a nd contamination effects in the source and the film. The vapor pressure of the material to be deposited is in the range 10 2 Torr which gives a deposition rate of 1 10 /s. The sticking coefficients of metals are typically 1 and for selenium the value ha s been reported as 0.28 0.5 [205] Hence, the deposition rate of selenium is typically kept 2 2.5 times higher than t he metal rates. The selenium

PAGE 335

335 polymer (Se n) evaporated from the source get adsorbed on the substrate and then participates in the surface reaction. In a typical co evaporation process, the substrate temperatures are greater than the source temperature. At these conditions, the adsorbed Se n polymers leave the surface by dissociating to Se n j (where j=0 to n 1). During the adsorption of selenium polymer, the internal energy increases which results in the increase in the pressure favoring desorption. The mos t favorable species then diffuses in to the vacant lattice sites thus driving the growth process. This is of primary concern for metal selenization process where maintaining Cu/In ratio less than one after selenization is highly important. During seleniza tion, the selenium reacts with copper and indium to form various binary compounds before going to CIS. Selenium reacts with copper forming CuSe, CuSe 2 or Cu 2 x Se depending on the selenium pressure. Also indium reacts slowly with selenium to form In 2 Se at 250 C and if sufficient selenium is not provided to form InSe or In 2 Se 3 indium loss can occur during selenization. While for device manufacturing, Cu/In ratio is highly important, but selenium poor conditions can deteriorate the cell performance and can also make the transition from p type to n type. Estimation of equilibrium partial pressure over various binary compounds helps in providing alternative routes for providing selenium which can have a major impact with costs associated with operation and mainten ance. Figure 11 1 plots the Gibbs energy of formation of In Se system per mole of Se 2 as Se 2 is considered as stable adsorbed species during absorber formation. Of all the In Se, In 4 Se 3 is the most stable species followed by InSe. In 2 Se 3 appears to be le ast stable in the In Se system. The change in the slope is due to the melting point of selenium (494 K). In case of Ga Se, only two stable species GaSe and Ga 2 Se 3 has been reported. GaSe appears to be more stable

PAGE 336

336 than Ga 2 Se 3 as shown in Figure 11 2. In ca se of CIGS system, CGS seems to be more stable than CIS. MoSe 2 appears to be stable at low temperatures; however at high temperatures CIS becomes more stable. During the reaction pathways studies done in this work, MoSe 2 forms after the completion of CIS consistent with observation made from Figure 11 3. Figure 11 4 shows the selenium vapor pressure over liquid selenium as a function of temperature. It is seen that a processing temperature of CIGS the overall vapor pressure of selenium is just below ~ 1 atm. The selenium vaporization can happen from the substrate and hence overpressure of selenium is required. Using the Hertz Langmuir equation, flux of selenium species is estimated over a range of temperature (11 17) A being the vapor pressure of selenium at selected temperature and the molecular weight of polymeric selenium species. The calcu lation results are shown in Figure 11 5 and it is seen that at processing temperature of CIGS, Se 2 is the most dominant species. Some of the CIGS industries do lot of processing in ambient conditions and the effect of oxygen on formation of different phase s can influence the electronic properties of solar cell. For instance, ISET (CIGS manufacturer) starts with oxides of the Cu In Ga, followed by reduction of oxides, again followed by re oxidation in H 2 Se. The complete reduction of the oxides is really imp ortant before starting the selenization reaction as oxygen can affect device performance. Also oxygen plays an important role in back contact forming oxides of molybdenum. It has been reported that band bending

PAGE 337

337 takes place at CIS Mo interface in the prese nce of MoO 2 causing a schottky type behavior [206] Figure 11 6 shows the stability of oxides as a function of temperature. It is seen that MoO 2 are more stable than MoO 3 However under oxygen rich conditions MoO 3 formation occurs at the surface as seen from X PS studies in chapter 9. Kronik et al. has proposed a defect model stating that the oxygen presence to some extent helps in compensating by neutralizing the defects of selenium by forming oxides of indium [59] Comparing the stable oxides in In Ga O system, it is seen that Ga 2 O 3 is more stable than the In 2 O 3 The stability of oth er oxides such as InO, GaO, In 2 O and Ga 2 O increases with temperature but are less stable than In 2 O 3 and Ga 2 O 3 Seleniz ation R eactor: Fluent S imulations Currently, selenization reactors used by the industries is a batch reactor where the metals coated on t he substrates are oxidized with H 2 Se or Se vapor. The reaction time varies from 8 12 hours depending on t he processing conditions used. The reaction kinetics studied in this thesis shows that some precursors do have the highest rate and the reaction can go to completion in 2 minutes at low temperature. In order to increase the throughput, the reaction time should be brought down with still retaining the high efficiency. It is necessary to have roll to roll process for selenization as opposed to batch p roce ss to increase the throughput. Till date, no reactor design has been reported for selenization and part of the reason is lack of reaction kinetics. The reactor design helps in understanding the gas phase dynamics (which is function of temperature, g as velocity and concentration) occurring during selenization reaction. As seen in the Figure 11 7, the reactor is a rectangular duct with heaters present above and below the substrates. The reactor can used to process two substrates separated

PAGE 338

338 by small d istance and the gas flows in between the substrates. The two substrates can move either co current or countercurrent direction to the gas flow. Also selenium can be supplied by coating one of the substrate (either top or the bottom), while selenizing the other one. In that case, the temperature of each substrate can be controlled independently to achieve a good process control and maximizing the selenium utilization rate. The substrate material used for this study is a glass and the reactor is made of st ainless steel. The gases used in the simulation were N 2 and H 2 Se. The material properties used for the simulation are tabulated in Table 1. Simulations on flow and thermal patterns in the reactor were done using computational fluid dynamics package (FL UENT TM ). The equations for conservation of mass, momentum and energy were solved with boundary conditions specific to the reactor geometry as shown in Figure 11 7 The reactor geometry was made using GAMBIT TM with Cartesian coordinates in a three dimensio nal format. A hexahedral mesh grid was employed and the boundary conditions were specified during the mesh design step. For reactor inlet, inlet flow velocity was used and for reactor outlet, outflow with weighting factor of 1.0 was used as the system us ed had a single inlet and outlet. The surface wall temperature was used for the heater surface. Figure 11 8 shows the mesh design for the selenization reactor. A segregated based solver was used for solving continuity equation are solved sequentially ra ther than simultaneously. equations to algebraic equations and solved numerically. The solution domain is subdivided in to finite number of control volumes and the conservation equations discussed below a re applied to each control volumes. The solution domain is

PAGE 339

339 subdivided into a finite number of con tiguous control volumes (CV). This technique involves integrating the governing equations about each CV and yielding discrete equations to conserve each quant ity on a CV. The equation for conservation of mass or continuity can be written as (11 18) Equation 11 1 8 is the general form of the mass conservation equation and is valid for both compressible and incompressib le flows assuming no accumulation in the system. divergence is net rate of mass o f flux per unit volume. The conservation of momentum in a non acc elerating reference frame is (11 19) The equation describes the rate of change of momentum per unit volume. The term [ The rate of momentum addition by molecular transport is given by two terms p and The stress tensor in equation 11 1 9 is given by (11 20) The conser vation of energy is given by (11 21)

PAGE 340

340 Where, (11 22) h is the enthalpy defined for ideal gas. Equation 11 2 2 represents rate of increase of kinetic energy per unit volume. The term done by the pressure of surrounding fluid per unit volume. The term on the right side of equation 20 represents energy transfer due to conduction, species diffusion and vi scous dissipation. The term S h represents heat of chemical re action involved in the reaction. Simulation was performed on flow and thermal pattern in the reactor using FLUENT. The operating pressure was fixed at 1 atmosphere which is typically used for se lenization reactions. Simulations were performed for different velocity of inlet gas (different Reynolds number). The Reynolds number was calculated using equivalent diameter of the rectangular reactor. The temperature of the heater was set at 1200 K and the heat flux boundary condition was used. For walls sharing the geometry, coupled boundary condition was used. The substrate was held stationary and only nitrog en gas was supplied at inlet. Figure 11 9 shows calculated contours of temperature for di fferent inlet velocity of nitrogen. With increase in velocity of inlet gas, cooling of the substrate occurs causing a large gradie nt in the temperature. The velocity contour between the substrates is shown in Figure 11 1 0 At a lower velocity of 0.025 m/s, velocity of the gas decreases at the center of the reactor with sudden increase towards the outlet. In second set of simulations, substrate was moved in a counter current direction of gas flow at a velocity of 1cm/sec. Figure 11 1 1 shows the temper ature contour for

PAGE 341

341 counter current movement of the substrate. Again no radiation was incorporated in the model. A gradient in the temperature was observed across the length of the substrate with cooling taking place near the inlet. At very low Reynolds n umber (280), uniform substrate temperature (1130 1170 K) was observed half way through the substrate when the h eater is maintained at 1200 K. However at high Reynolds number (2800) the maximum temperature reached by the substrate was 700 K. Typically se miconductor processing is carrie d out in laminar flow regimes. When the movement of substrate was changed to co current direction to the gas flow, variations on substrate temperature was observed. Though the maximum temperatures reached on the substrate were essentially the same for the same Reynolds number, but the gradient was different from the counter current movement of the substrate as shown in Figure 11 12 The velocity contours between the parallel substrate for counter current movement of subst rate is shown in Figure 11 1 3 At very low Reynolds number, the velocity of the gas becomes zero at the center of the reactor and with increase in Reynolds number; velocity of the gas increases and becomes u niform over the entire reactor. However with c o current movement of substrate uniform velocity of gas was observed even for low Reynolds number as shown in Figure 1 1 14 With radiation incorporated in the model, temperature of the substrate was almost equal to the temperature of the heater irrespecti ve of the movement of the substrate. No change in the velocity profile was obtained, however absolute change in velocity magnitude was observed. Again uniform velocity of gas was observed for co curr ent movement of the substrate. Figure 11 1 5 to 11 18 s hows the temperature and velocity

PAGE 342

342 contours for substrate moving in counter current and co current fashion with incorporation of radiation in the model. The material properties of selenium such as viscosity, density, molecular weight were added to the Fluen t database to determine the concentration variation of selenium across the substrate. Simulation were done for both counter current and co current movement of substrates at a velocity of 1 cm/s. Different temperature distributions on the substrate were ob tained with addition of selenium vapor for the same Reynolds number (Figure 11 19 and 11 20) compared with no selenium vapor (Figure 11 1 1 and Figure 1 1 12 ). Similar observation was made for the velocity contours as shown in Figure 11 21 and Figure 11 22. The concentration of selenium decreased from inlet through outlet at very low Reynolds number irrespective of the movement of the substrate as show n in Figure 11 23 and Figure 11 24. Considering all the simulation results, it is suggested that operat ing the reactor in co current mode helps in maintaining selenium vapor pressure close to the equilibrium value as suggested by the phase diagram. In case of counter current mode operation, velocity of the gas becomes zero indicating stagnant zones causing depletion of selenium on incorporation of reaction rate in the model. The reaction rate for CIGS formation increases with increase in temperature and equilibrium selenium pressure should be maintained to prevent selenium loss. For co current mode, the te mperature of the substrate at the outlet increases to the set point value. The concentration of selenium at the outlet in terms of pressure (Se 2 species) units for Reynolds number of 700 equates to 63 Torr much above the equilibrium value (~49 Torr).

PAGE 343

343 Su mmary Ellingham diagram for the Cu In Ga Se O system was obtained using the Thermocalc database. The equilibrium partial pressure required for compounds to be stable at a particular temperature was calculated from the diagram. Selenization reactor design w as done using Fluent and simulations were done to get the flow, temperature and concentration profiles at different Reynolds number. Both counter current and co current flow configurations were considered and maximum utilization of the gas was obtained fr om counter current operation. Uniform temperature distribution over the substrates was obtained by incorporating radiation in the model. Simple heat transfer calculation showed the time required to reach the temperature was less than 2 seconds.

PAGE 344

344 Figure 11 1 Ellingham diagram for In Se system

PAGE 345

345 Figure 11 2 Ellingham diagram for Ga Se system

PAGE 346

346 Figure 11 3 Ellingham diagram for CIGS and MoSe 2 system

PAGE 347

347 Figure 11 4 Vapor pressure of selenium species over liquid selenium as a function of temperature

PAGE 348

348 Figure 11 5 Molecular flux of selenium species from a evaporating liquid source

PAGE 349

349 Figure 11 6 Ellingham diagram for oxides of Cu Ga In O system

PAGE 350

350 Figure 11 7 Selenization reactor for roll to roll process

PAGE 351

351 Figure 11 8 Hexahedral mesh design for selenization reactor

PAGE 352

352 Figure 11 9 Calculated temperature contours of substrate for different inlet velocity. A) 0.025 m/s, B) 0.05 m/s, C) 0.1 m/s A B C

PAGE 353

353 Figure 11 10 Cal culated velocity contours of nitrogen over substrate for different inlet velocity. A) 0.025 m/ s, B) 0.05 m/s, C) 0.1 m/s A B C

PAGE 354

354 Figure 11 11 Temperature contours for counter current movement of substrat e at a different gas velocity. A)0.01m/s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s C D B A

PAGE 355

355 Figure 11 12 Temperature contours for co current movement of substrat e at a different gas velocity. A)0.01m/s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s A C B A D

PAGE 356

356 Figure 11 13 Velocity contours ove r substrate for counter current movement of substrat e at a different gas velocity. A)0.01m/ s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s A C B D

PAGE 357

357 Figure 11 14 Velocity contours over substrate for co current movement of substrat e at a different gas velo city. A)0.01m/s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s A C B D

PAGE 358

358 Figure 11 15 Temperature contours with radiation incorporated for counter current movement of substrat e at a different gas velocity. A)0.01m/s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s A B C D

PAGE 359

359 Figure 11 16 Temperature contours with radiation incorporated for co current movement of substra te at a different gas velocity. A)0.01m/s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s A C D B

PAGE 360

360 Figure 11 17 Velocity contours with radiation in corporated for counter current movement of substrate at a different gas velocity. A)0.01m/s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s A D C B

PAGE 361

361 Figure 11 18 Velocity contours with radiation incorporated for co current movement of substrat e at a differ ent gas velocity. A)0.01m/s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s D B C A

PAGE 362

362 Figure 11 19 Temperature contours for counter current movement of substrate at a total gas velocity( Mass frac tion of selenium in gas =0.2). A)0.01m/s, B) 0.025 m/s, C) 0. 05 m/s, D) 0.1 m/s D B C A

PAGE 363

363 Figure 11 2 0 Temperature contours for co current movement of substrate at a total gas velocity( Mass frac tion of selenium in gas =0.2). A)0.01m/s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s D B A C

PAGE 364

364 Figure 11 2 1 Vel ocity contours for counter current movement of substrate at a total gas velocity( Mass frac tion of selenium in gas =0.2). A)0.01m/s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s A C B D

PAGE 365

365 Figure 11 2 2 Velocity contours for co current movement of substrate at a total gas velocity ( Mass fraction of selenium in gas =0.2). A)0.01m/s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s A C B D

PAGE 366

366 Figure 11 2 3 Concentration contours for counter current movement of substrate at a total gas velocity( Mass frac tion of se lenium in gas =0.2). A)0.01m/s, B) 0.0 25 m/s, C) 0.05 m/s, D) 0.1 m/s A C B D

PAGE 367

367 Figure 11 2 4 Concentration contours for co current movement of substrate at a total gas velocity( Mass frac tion of selenium in gas =0.2). A)0.01m/s, B) 0.025 m/s, C) 0.05 m/s, D) 0.1 m/s A C B D

PAGE 368

368 CHAPTER 1 2 PHASE DIA GRAM ASSESMENT OF PS EUDOBINARY CU 2 SE GA 2 SE 3 AND CUINSE 2 CUGASE 2 SYSTEM Overview Thermodynamic modeling of the CIGS system provides a base to understand the reaction pathways that lead to absorber formation process. By varying the band gap as a function of depth in CIGS film, performance of solar cell can be increased as it better matches the solar spectrum. In some processes like evaporation gallium grading is done intentionally by evaporating gallium at a higher rate initially followed by increasing the indium amount. In other processes, like metal selenization, gallium gets distributed higher towards the back, creating a back surface field which helps in reducing recombination and hence increase in effici ency. This is sometimes referred as unintentional grading. Also to improve the cell efficiency, tandem cell approach has been proposed with CGS as the top cell and CIGS as the bottom cell. It is necessary to have thermodynamics description for pseudo binar y CIS CGS for addition of gallium in CIS. The CuGaSe 2 was first synthesized by Hahn et al [207] A number of exp erimental studies on the phase relations of the Cu Ga Se system as well as the pseudo binary section of Cu 2 Se Ga 2 Se 3 system exist in the literature. Most of the experimental data are from DTA and XRD measurements [208, 209] Kyoung et al. have performed a phase diagram assessment in ternary Cu In Se system. The pseudo binary sections of Cu 2 Se In 2 Se 3 and Cu 2 Se Ga 2 Se 3 systems show characteristic similarities; however, Cu Ga Se system is relatively less studied and considerable uncertainty still exists especially for selenium rich sections of phase eq uilibria. The binary systems of Cu Ga, Ga Se and Cu Se have already been assessed ; however, the stability of ternary compounds and phase diagram along Cu 2 Se, Ga 2 Se 3 line have not been critically

PAGE 369

369 assessed. The unstable results obtained in preliminary EMF ex periments performed on Cu Ga Se system suggest that pseudobinary section of Cu 2 Se Ga 2 Se 3 may be analogous to Cu 2 Se In 2 Se 3 Similar observation were made by Palatnik and Belova [210] in the gallium rich region of Ga 2 Se 3 indicating stability of CuGa 5 Se 8 between 70 88 mole % of Ga 2 Se 3 as shown in Figure 12 1 The phase CuGa 3 Se 5 has never been reported in the literature. Later Mickelsen et al. inve stigated the ternary Cu Ga Se phase diagram by DTA and XRD and revised the solid solution on Ga 2 Se 3 rich side to extend from 70 100 mole % Ga 2 Se 3 as shown in Figure 12 2 [209] Hence only four phases, Cu 2 Se, CuGaSe 2 Ga 2 Se 3 rich solid solution and liquid phase were considered in their model. Mushin et al have performed EMF studies and the Gibbs energy of formation of CuGaSe 2 has been reported [135] Bodnar and Bologa [211] have the reported that CuInSe 2 and CuGaSe 2 crystallize in chalcopyrite structure and a complete solid solution is possible which was verified by performing experiments in double quartz ampoules. The phase transformation temperatures were reported by performing DTA studies. As shown in Figure 12 3 CuGaSe 2 undergoes phase transformation at 1318 and 1361 K. Similarly, CuInSe 2 undergoes phase transformation at 1083 K and 1259 K The phase transformation at 1045 C for CuGaSe 2 was assumed to be related to cation cati on disordering by analogous assessment of Palatnik and Rogacheva [212] th at was referred for the phase transformation of CuInSe 2 at 810 C. The phase transformation temperature of 1045 C is consistent with value of 1050 5 C by the earlier report of Palatnik and Belova [213] which is represented by the peritectic phase reaction. In this Chapter, optimization

PAGE 370

370 is performed by defining specific models to express Gibbs energy of each phase using CALPHAD based approach. Thermodynamic Optimization The CALPHAD method was applied to obtain a consistent thermodynamic description by using the PanOptimizer of the PANDAT package. The PANDAT uses CALPHAD method based on two principles; the total Gibbs energy of a syste m will be at global minimum at thermodynamic equilibrium and the chemical potential of every element in different equilibrium phases is the same. CALPHAD based approach has the following advantages, Predicsts phases diagrams and thermodynamic properties of system where no experimental data is available. Calculates phase diagrams of higher order systems based on the models of its lower order sub systems. Calculated metastable phase diagram. Determines the chemical potential of phase transformation. The st andard Gibbs Energy function G i for the element i the equation (12 1) Where, H i SER is the molar enthalpy of element i at 298.15 K and 1 bar in its standard phase. The parameters for the elemental functions were taken from the SGTE compilation.

PAGE 371

371 Cu 2 Se Ga 2 Se 3 Pseudobinary system The liquid phase in the Cu 2 Se Ga 2 Se 3 system is described by associatio n model developed by Sommer [214] The associated model can be expressed as ( Cu 2 Se, Ga 2 Se 3 ) I The Gibbs energy of this phase can also be expressed as, (12 2) Where ref G Liq is given as (12 3) Where y i represents the site fractions of the species i. specifically for this case. The term (12 4) G i Liq represents the Gibbs energy of pure liquid phase. The ideal mixing Gibbs energy id G Liq in equation 12 3 is given as, (12 5) The excess Gibbs energy E G Liq in equation 12 3 is given as, (12 6) Where the L with superscripts represents the interactions between the species and are expressed as function of temperature. The non stoichiometric chalcopyrite compound CuGaSe 2 is described by sub lattice model. The model incorporates anti site defect where Cu 2 Se lattice is occupied by Ga 2 Se 3 The Gibbs energy for pure CuGaSe 2 has been reported in the literat ure.

PAGE 372

372 The sublattice model for the chalcopyrite phase can be written as (Cu 2 Se, Ga 2 Se 3 ) I (Cu 2 Se, Ga 2 Se 3 ) I I (12 7) (12 8) (12 9) Where represents the non stoich iometric chalcopyrite phase. y i and y i represent the site fraction of the component i in the first and second sub lattice respectively. The term 0 G Cu 2 Se : Ga 2 Se 3 is the standard Gibbs energy for the stoichiometric compound CuGaSe 2 The Ga 2 Se 3 rich solid s olution is described by substitutional regular solution model. The Gibbs energy of such solutions can be expressed as, (12 10)

PAGE 373

373 W here represents the Ga 2 Se 3 rich solid solution. The data points shown in Figure 12 2 is used as experimental data points for optimization of thermodynamic model parameters of all phases considered in the system. The summary of the data is shown in Table 12 1. The optimization procedure is started by creating a POP file and then using the Pan Optimizer. To start with, the Gibbs energy of all the phases except liquid are fixed and optimization is carried out for liquid phase only. The liquid phase optimized coe fficients are updated in the database, and then the Cu 2 Se rich solid solution is optimized. In a similar way, all the other phases such as CuGaSe 2 and Ga 2 Se 3 rich solid solution are optimized. Table 12 2 shows the optimized parameters for the Cu 2 Se Ga 2 Se 3 pseudobinary system. Figure 12 4 compares the optimized curve wit h the experimental data points. CIS CGS Pseudobinary system The CALPHAD method described in previous section was used to obtain a consistent thermodynamic description of the CIS CGS pseudo bi nary using commercially available software PANDAT. The experimental data used for optimization in shown in Figure 1 2 3. For the solid phase, CIS and CGS, three sub lattice models were used. The first sub lattice for the copper, second for the gallium or indium and third sub lattice for selenium was used. The composition of copper and selenium were fixed and the addition of gallium was varied. The same sub lattice model was used for the phase. For the liquid phase ionic sub lattice model was used. Two s ub lattice were considered with the first sub lattice occupied by Ga +3 and In +3 ions. The second sub lattice had CuSe 2 3 and the sub lattice can be represented as, (Ga +3 In +3 ) p (CuSe 2 3 ) q

PAGE 374

374 The optimization was carried out in Pan Optimizer starting with liq uid phase followed by solid phase. Table 12 3 shows the optimized parameters for the CIS CGS system. Figure 12 5 shows the good fit of the model with the experimental data for the CIS CGS system. Summary Phase diagram assessment of pseudobinary system of Cu 2 Se Ga 2 Se 3 and CIS CGS was done using PANDAT using the experimental data in the literature and previous evaluation. Sub lattice models were used to represent the solid phase. For liquid phase in Cu 2 Se Ga 2 Se 3 association model was used. For CIS CGS syst em, liquid phase was represented by ionic sub lattice model. An expression of Gibbs energy was obtained and the model fits the experimental data more accurately. The models used in the system can be easily extended higher order systems.

PAGE 375

375 Table 12 1. Data used for optimization of Cu 2 Se Ga 2 Se 3 phase diagram Equilibrium Temperature (K) Reference Ga 2 Se 3 rich solid solution/Liquid; Congruent melting 1378 [208, 209] Eutectic; Liquid Cu 2 Se+ CuGaSe 2 1250 [209] Peritectic; Liquid +Ga 2 Se 3 CuGaSe 2 1360 [208, 209]

PAGE 376

376 T able 12 2 Optimized parameters for Gibbs energy for Cu 2 Se Ga 2 Se 3 pseudobinary system. Phase Parameters Liquid Cu 2 Se rich solid solution Ga 2 Se 3 rich solid solution

PAGE 377

377 Table 12 2 Continued Phase Parameters CuGaSe 2

PAGE 378

378 Table 12 3 Optimized parameters f or Gibbs energy for CIS CGS pseudobinary system. Phase Gibbs Energy (J/mol) Interaction Parameter Liquid G CIS G CGS G CIS + 114383 8 4.62 T G CGS + 104963 85.461T L 0 L 1 5279.14 7.657T 6006.84 5.478T CIS and CGS G CIS G CGS G CIS G CGS L 0 L 1 1 865.35 +3.6021T 1963.85+2.4433T CIS and CGS G CIS G CGS G CIS G CGS L 0 L 1 359.6 .3219T 220.66+0.0379T

PAGE 379

379 Figure 12 1 E xperimental points used for optimization of Cu 2 Se Ga 2 Se 3 psedudobinary section.

PAGE 380

380 Figure 12 2 Experimen tal data points used for optimization of Cu 2 Se Ga 2 Se 3 pseudobinary section.

PAGE 381

381 Figure 12 3 Experimental data points used for optimization of CIS CGS pseudobinary section.

PAGE 382

382 Figure 12 4 Comparison of experimental data and as predicted by the model for Cu 2 Se Ga 2 Se 3 system

PAGE 383

383 Figure 12 5 Comparison of experimental data and as predicted by the model for CIS CGS pseudobinary system

PAGE 384

384 CHAPTER 13 CONCLUSIONS AND FUTURE WORK In situ high temperature X ray diffraction was used to study the reaction pathways and kinetics in the CIGS based system. A systematic study was conducted starting with understanding the reaction pathways and kinetics in the metal selenium binary system. The overall phase transformation for the metal seleniu m binary system qualitatively follows the sequence predicted by the phase diagram. This work wa s substantiated with the TE M measurements. The kinetics of formation of binaries was estimated and as a future work, this data will be used to estimate the s peci es mobilities using DICTRA. Estimating species mobilities as a function of temperature, pressure and composition will predict the reaction pathways which will lead to high rate growth of CIGS. Reaction pathways and kinetics were followed for CIS formation using high temperature XRD employing no vel bilayer compound precursors A new reactor for HT XRD was built to see the effect of selenium pressure on the peritectic decomposition reactions involving CuSe and CuSe 2 Maintaining selenium pressure close to t he equilibrium value helps in attaining peritectic decomposition close to the temperature predicted by the phase diagram. The peritectic decomposition reactions help in attaining liquid assisted grain growth of CIS at very low temperatures. Having liquid phase assistance during crystal growth reduces the processing temperature and minimizes the point defects along the growth direction. The metal selenization precursors always have a difference in gallium redistribution compared to the co evaporation proc ess. It is necessary to have higher gallium content near the surface to improve the open circuit voltage. It is also necessary to simulate the gallium redistribution in metal selenization processes as

PAGE 385

38 5 obtained in champion cell process. Gallium redistribut ion studies were systematically investigated during selenization using bilayer metal and stacked elemental layers The gallium distribution in the final film was characterized by TEM AES and SIMS. It was found that gallium redistribution was highly depend ent on the order of deposition of precursor structure. The CuIn/CuGa/Se precursor was found to emulate the champion cell co evaporation process in terms of gallium redistribution. One of the other issues found in metal selenization process was creation o f voids near the back contact. The void creation is due to development of tensile stress happening during volume expansion upon selenization. The stress relief is obtained by creation of voids which increases the series resistance of the absorber. This is perhaps one of the reasons for lower efficiencies obtained in metal selenization process compared with the co evaporation process. The voids in the film can avoided by modulating the structure of the metal as discussed in Chapter 7. Modulating the structu re helps in relieving the tensile stress and higher efficiencies has been reported using the similar structure. Sodium addition is CIGS has been studied by various groups and till date no quantitative model exists describing its role on kinetics of absorbe r formation. Addition of sodium through molybdenum doped sodium was also studied to see the change in the gallium distribution and its kinetics as discussed in Chapter 8 It was found that sodium addition increases the rate of formation of CIGS along with grain growth. The results of the reaction kinetics using Avrami and parabolic models are summarized in Table 13 1. Formation of MoSe 2 after completion of CIGS formation is observed, though the temperature of formation varies with the precursor structure. The

PAGE 386

386 role of MoSe 2 has been identified in CIGS as discussed in Chapter 6. Formation temperature of MoSe 2 is reported as 420 o C and rate parameters have been identified using solid state growth model. In the future, more TEM characterization work is needed for the reactions stopped at various intermediate stages to further understand the reaction mechanisms. Also devices need to be fabricated to get baseline efficiency for the precursor structure studied. In addition, a new nanoparticle process has been deve loped using the peritectic decomposition of CuSe 2 and CuSe. To have a uniform composition in large area manufacturing, a binary nanoparticle based approach is presented. Efficiency of 1.67% has been obtained employing binary nanoparticle approach after annealing the nanoparticles for 5 minutes. A lower open circuit voltage is primarily due to high interface recombination. Also the transparent conducting oxides (TCO) improvements are needed to improve the efficiency. Preliminary calculations done on the 1.67% cell shows that a 5% cell can be obtained by just improving the properties of TCO. Additionally, electrical characterization such as DLTS, hall measurements should be done to get the defects and mobilities. The current processing time for CIGS depo sition is in the order of 25 30 minutes. Using the above mentioned nanoparticle approaches, the throughput can be increased which directly decreases the cost of manufacturing. A reliability study was carried out in Chapter 10 using high temperature XRD o n a complete device and a new compound CuCd 2 (Ga,In)Se 4 was detected. Reaction kinetics obtained from Avrami model suggests that device failure due to the compound formation at normal operating temperature of the module is not possible. In future, TEM studi es are required for identification of this compound in the device.

PAGE 387

387 Selenization process uses 4 8 hours of reaction time which decreases the throughput and increases the cost of production. A continuous selenization reactor was simulated using Fluent as dis cussed in Chapter 11 to get the composition, flow and temperature profiles in the reactor. As a part of future study, kinetics developed using high temperature XRD studies should be incorporated in the model and then selenium concentration profiles over t h e substrate should be simulated. A thermodynamic description of pseudobinary system of Cu 2 Se Ga 2 Se 3 and CIS CGS is reported in Chapter 12 Sub lattice and regular solution models were used to represent the Gibbs energy of different phases in the system. T he model agrees well with the experimental data reported in the literature. Future work would require developing quaternary description of CIGS system incorporating the models developed in this study.

PAGE 388

388 Table 13 1 Summary of reaction kinetics fro m different precursor structures Precursor Growth Method Activation Energy (kJ/mole) Avrami Parabolic SS/Mo/Cu/Ga/In/Se Evaporation 78( 6) N/A SS/Mo/Ga/In/Cu/Se Evaporation 116( 5) N/A Glass/Mo/In 2 Se 3 /Cu 2 Se Evaporation 194( 10 ) 203( 12) Glass/Mo/I n 2 Se 3 /CuSe y Evaporation 162( 7) 225( 16) Glass/Mo/(In,Ga) 2 Se 3 /CuSe MBE 116( 6) 148( 10) Glass/Mo/CuGa/CuIn/Se MBE 101( 9 ) N/A Glass/Mo/Cu In/Cu Ga / Se MBE 107( 15 ) N/A Glass/Mo/CuGa/CuIn+ Se MBE 93( 4 ) N/A Glass/Mo/CuIn/CuGa+ Se MBE 76( 14 ) N/A Glass/Mo /CuGa/In+ Se Sputtering 92( 12 ) N/A Glass/Na 3 Mo/Mo/CuGa/In+ Se Sputtering 103( 11 ) N/A Glass/Na 5 Mo/Mo/CuGa/In+ Se Sputtering 113( 20 ) N/A Glass/Mo/Cu/Ga/In ( 4)+ Se MBE 110( 19 ) 111( 1 ) Glass/Mo/Cu/Ga/In ( 8 )+ Se MBE 122( 27 ) 129( 6 ) Glass/Mo + Se Spu ttering 101.2( 8 ) 101.7( 6 ) SS/Mo/CIGS/CdS Co evaporation 233.5( 45 ) N/A

PAGE 389

389 LIST OF REFERENCES [1] A.E. Becquerel, Competus Rendus 9 (1839) 561. [2] W. Shockley, H.J. Queisser, Journal of Applied Physics 32/3 (1961) 510. [3] B.M. Ka yes, H. Nie, R. Twist, S. Spruytte, F. Reinhardt, I. Kizilyalli, G. Higashi, Proceedings of 37th IEEE Photovoltaic Specialist Conference, Seattle, 2011. [4] A. Goetzberger, C. Hebling, Solar Energy Materials and Solar Cells 62/1 2 (2000) 1. [5] R.K. Jain, Photovoltaic Energy Conversion, 2003. Proceedings of 3rd World Conference on, 2003, p. 75. [6] R.K. Jain, D.J. Flood, Transactions of the ASME 115 (1993) 106. [7] H.K. Kang, S. H. Park, D.H. Jun, C.Z. Kim, K.M. Song, W. Park, C.G. Ko, H. Kim, Semiconductor Science and Technology 26/7 (2011) 075009. [8] R.R. King, D.C. Law, K.M. Edmondson, C.M. Fetzer, G.S. Kinsey, H. Yoon, R.A. Sherif, N.H. Karam, Applied Physics Letters 90/18 (2007). [9] A. Goetzberger, J. Luther, G. Willeke, Solar Energy Materials and Sol ar Cells 74/1 4 (2002) 1. [10] B. Robert W, Solar Energy Materials and Solar Cells 65/1 4 (2001) 17. [11] R.W. Birkmire, R.B. Hall, J.E. Phillips, Conference Record of the 17th IEEE Photovoltaic Specialists Conference Kissimmee, Florida, 1984, p. 882. [12] J. Yang, A. Banerjee, S. Guha, Applied Physics Letters 70/22 (1997) 2975. [13] M. Kanellos, 2011. [14] P. Jackson, D. Hariskos, E. Lotter, S. Paetel, R. Wuerz, R. Menner, W. Wischmann, M. Powalla, Progress in Photovoltaics: Research and Applications (2011 ) 894. [15] J. F.o. Guillemoles, L. Kronik, D. Cahen, U. Rau, A. Jasenek, H. W. Schock, The Journal of Physical Chemistry B 104/20 (2000) 4849. [16] T. Nakada, A. Kunioka, Applied Physics Letters 74/17 (1999) 2444. [17] D. Abou Ras, G. Kostorz, D. Bremaud, M. Klin, F.V. Kurdesau, A.N. Tiwari, M. Dbeli, Thin Solid Films 480 481 (2005) 433. [18] P.E. Russell, O. Jamjoum, R.K. Ahrenkiel, L.L. Kazmerski, R.A. Mickelsen, W.S. Chen, Applied Physics Letters 40/11 (1982) 995.

PAGE 390

390 [19] T. Wada, N. Kohara, S. Nishiwaki T. Negami, Thin Solid Films 387/1 2 (2001) 118. [20] B.J. Stanbery, Critical Reviews in Solid State and Materials Sciences 27/2 (2002) 73. [21] F. Benkabou, H. Aourag, M. Certier, Materials Chemistry and Physics 66/1 (2000) 10. [22] H. W. Schock, R. Nouf i, Progress in Photovoltaics: Research and Applications 8/1 (2000) 151. [23] S. Susanne, Thin Solid Films 403 404/0 (2002) 1. [24] R.W. Birkmire, E. Eser, Annual Review of Materials Science 27/1 (1997) 625. [25] D. Schmid, M. Ruckh, F. Grunwald, H.W. Schoc k, Journal of Applied Physics 73/6 (1993) 2902. [26] T. Negami, N. Kohara, M. Nishitani, T. Wada, T. Hirao, Applied Physics Letters 67/6 (1995) 825. [27] S.H. Kwon, S.C. Park, B.T. Ahn, K.H. Yoon, J. Song, Solar Energy 64/1 3 (1998) 55. [28] I.M. Ktschau, H.W. Schock, Journal of Physics and Chemistry of Solids 64/9 10 (2003) 1559. [29] J.A.M. AbuShama, S. Johnston, T. Moriarty, G. Teeter, K. Ramanathan, R. Noufi, Progress in Photovoltaics: Research and Applications 12/1 (2004) 39. [30] A. Jehad, R. Noufi, S. Johnston, S. Ward, X. Wu, Conference Record of the 31st IEEE Photovoltaic Specialists Conference, 2005, 2005, p. 299. [31] T.J. Coutts, D.L. Young, X. Li, W.P. Mulligan, X. Wu, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films 18/6 ( 2000) 2646. [32] J. Klaer, J. Bruns, R. Henninger, K. Siemer, R. Klenk, K. Ellmer, D. Brunig, Semiconductor Science and Technology 13/12 (1998) 1456. [33] U. Rau, M. Schmidt, A. Jasenek, G. Hanna, H.W. Schock, Solar Energy Materials and Solar Cells 67/1 4 (2001) 137. [34] W.N. Shafarman, R. Klenk, B.E. McCandless, Journal of Applied Physics 79/9 (1996) 7324. [35] G. Hanna, A. Jasenek, U. Rau, H.W. Schock, Thin Solid Films 387/1 2 (2001) 71.

PAGE 391

391 [36] I. Repins, S. Glynn, J. Duenow, T.J. Coutts, W.K. Metzger, M. A. Contreras, Thin Film Solar Technology, SPIE, San Diego, CA, USA, 2009, p. 74090M. [37] T. Dullweber, O. Lundberg, J. Malmstrm, M. Bodegrd, L. Stolt, U. Rau, H.W. Schock, J.H. Werner, Thin Solid Films 387/1 2 (2001) 11. [38] S. Niki, M. Contreras, I. R epins, M. Powalla, K. Kushiya, S. Ishizuka, K. Matsubara, Progress in Photovoltaics: Research and Applications 18/6 (2010) 453. [39] S.B. Zhang, S. H. Wei, A. Zunger, H. Katayama Yoshida, Physical Review B 57/16 (1998) 9642. [40] S. H. Wei, S.B. Zhang, A. Zunger, Applied Physics Letters 72/24 (1998) 3199. [41] J. Hedstrom, H. Ohlsen, M. Bodegard, A. Kylner, L. Stolt, D. Hariskos, M. Ruckh, H.W. Schock, Photovoltaic Specialists Conference, 1993., Conference Record of the Twenty Third IEEE, 1993, p. 364. [42] L. Stolt, J. Hedstrom, J. Kessler, M. Ruckh, K. O. Velthaus, H. W. Schock, Applied Physics Letters 62/6 (1993) 597. [43] K. Granath, M. Bodegrd, L. Stolt, Solar Energy Materials and Solar Cells 60/3 (2000) 279. [44] M. Bodegard, K. Granath, L. Stolt, Thi n Solid Films 361 362/0 (2000) 9. [45] D. Rudmann, G. Bilger, M. Kaelin, F.J. Haug, H. Zogg, A.N. Tiwari, Thin Solid Films 431 432/0 (2003) 37. [46] M. Lammer, U. Klemm, M. Powalla, Thin Solid Films 387/1 2 (2001) 33. [47] M. Contreras, B. Egaas, P. Dippo, J. Webb, J.E. Granata, K. Ramanathan, S. Asher, A. Swartzlander, R. Noufi, Proceedings of 26th IEEE Photovoltaics Specialists Conference Newyork, 1997, p. P 359. [48] B. Keyes, F.S. Hasoon, P. Dippo, A. Balcioglu, F. Abulfotuh, Conference Record of 26th I EEE Photovoltaic Specialists Conference, Newyork, 1997, p. p479. [49] T. Nakada, D. Iga, A. Kunioka, Jpn. J. Appl. Phys., 36 (1997) 732. [50] R. Kimura, T. Nakada, P. Fons, A. Yamada, S. Niki, T. Matsuzawa, K. Takahashi, A. Kunioka, Solar Energy Materials and Solar Cells 67/1 4 (2001) 289. [51] S. H. Wei, S.B. Zhang, A. Zunger, Journal of Applied Physics 85/10 (1999) 7214. [52] O. Lundberg, M. Edoff, L. Stolt, Thin Solid Films 480 481 (2005) 520.

PAGE 392

392 [53] D. Braunger, D. Hariskos, G. Bilger, U. Rau, H.W. Schock Thin Solid Films 361 362/0 (2000) 161. [54] O. Lundberg, M. Bodegrd, L. Stolt, Thin Solid Films 431 432 (2003) 26. [55] A. Rockett, K. Granath, S. Asher, M.M. Al Jassim, F. Hasoon, R. Matson, B. Basol, V. Kapur, J.S. Britt, T. Gillespie, C. Marshall, So lar Energy Materials and Solar Cells 59/3 (1999) 255. [56] M. Ruckh, D. Schmid, M. Kaiser, T. Walter, H.W. Schock, Conference Record of the 26th IEEE Photovoltaic Specialists Conference, Newyork, 1994. [57] S.C. Kim, H. Park, E. W. Lee, J.S. Han, S.H. Lee, C. W. Jeon, D. Jung, J. Jeong, W.K. Kim, Thin Solid Films 519/21 (2011) 7250. [58] Fella, L. Kranz, J. Perrenoud, S. Seyrling, R. Verma, S. Nishiwaki, Y.E. Romanyuk, G. Bilger, A.N. Tiwari, Nat Mater 10/11 (2011) 857. [59] L. Kronik, D. Cahen, H.W. Sch ock, Advanced Materials 10/1 (1998) 31. [60] D. Cahen, R. Noufi, Applied Physics Letters 54/6 (1989) 558. [61] L. Kronik, U. Rau, J. F. Guillemoles, D. Braunger, H. W. Schock, D. Cahen, Thin Solid Films 361 362/0 (2000) 353. [62] M.B. Zellner, R.W. Birkmir e, E. Eser, W.N. Shafarman, J.G. Chen, Progress in Photovoltaics: Research and Applications 11/8 (2003) 543. [63] R.A. Mickelsen, W.S. Chen, Applied Physics Letters 36/5 (1980) 371. [64] V.M. Glazov, A.S. Pashinkin, V.A. Fedorov, Inorg. Mater. 36/Copyright (C) 2010 American Chemical Society (ACS). All Rights Reserved. (2000) 641. [65] R. Klenk, T. Walter, H. W. Schock, D. Cahen, Advanced Materials 5/2 (1993) 114. [66] J. Kessler, C. Chityuttakan, J. Schldstrm, L. Stolt, Thin Solid Films 431 432/0 (2003) 1 [67] M. Marudachalam, R.W. Birkmire, H. Hichri, J.M. Schultz, A. Swartzlander, M.M. Al Jassim, Journal of Applied Physics 82/6 (1997) 2896. [68] Perrenoud, L. Kranz, R. Verma, D. Guettler, S. Nishiwaki, Y.E. Romanyuk, G. Bilger, D. Brmaud, A.N. Tiwari, Progress in Photovoltaics: Research and Applications 19/5 (2011) 560. [69] M. Turcu, U. Rau, Journal of Physics and Chemistry of Solids 64/9 10 (2003) 1591.

PAGE 393

393 [70] M.A. Contreras, B. Egaas, D. King, A. Swartzlander, T. Dullweber, Thin Solid Films 361 362/0 (2000) 167. [71] V.K. Kapur, U.V. Choudary, A.K.P. Chu Pat ent 1986. [72] A.C. Rastogi, K.S. Balakrishnan, R.K. Sharma, K. Jain, Thin Solid Films 357/2 (1999) 179. [73] .F. Yksel, B.M. Basol, H. Safak, H. Karabiyik, Applied Physics A: Materials Science & Processing 73/3 (2001) 387. [74] M.S. Sadigov, M. zkan, E Materials Science 34/18 (1999) 4579. [75] H.P. Fritz, P. Chatziagorastou, Thin Solid Films 247/1 (1994) 129. [76] M. Marudachalam, H. Hichri, R. Klenk, R.W. Birkmire, W.N. Shafarman, J.M. Schultz, Appl. Phys. Lett. 67/26 (1995) 3978. [77] F.O. Adurodija, J. Song, S.D. Kim, S.H. Kwon, S.K. Kim, K.H. Yoon, B.T. Ahn, Thin Solid Films 338/1 2 (1999) 13. [78] C. Guilln, M.A. Martnez, J. Herrero, Vacuum 58/4 (2000) 594. [79] C. Guilln, J. Herrero, Solar Energy Ma terials and Solar Cells 73/2 (2002) 141. [80] B.M. Basol, V.K. Kapur, Applied Physics Letters 54/19 (1989) 1918. [81] J. Bekker, V. Alberts, M.J. Witcomb, Thin Solid Films 387/1 2 (2001) 40. [82] V. Alberts, M. Klenk, E. Bucher, Thin Solid Films 387/1 2 (2 001) 44. [83] V.K. Kapur, B.M. Basol, E.S. Tseng, Solar Cells 21/1 4 (1987) 65. [84] A.M. Fernandez, P.J. Sebastian, A.M. Hermann, R.N. Bhattacharya, R.N. Noufi, M.A. Contreras, Renewable Energy 8/1 4 (1996) 396. [85] A.M. Hermann, C. Gonzalez, P.A. Ramakr ishnan, D. Balzar, N. Popa, P. Rice, C.H. Marshall, J.N. Hilfiker, T. Tiwald, P.J. Sebastian, M.E. Calixto, R.N. Bhattacharya, Solar Energy Materials and Solar Cells 70/3 (2001) 345. [86] C.L. Jensen, D.E. Tarrant, J.H. Ermer, G.A. Pollock, Conference Reco rd of the 23rd Photovoltaic Specialists Conference, 1993., 1993, p. 577. [87] K.T. Ramakrishna Reddy, P.K. Datta, M.J. Carter, physica status solidi (a) 182/2 (2000) 679. [88] S. Bandyopadhyaya, S. Roy, S. Chaudhuri, A.K. Pal, Vacuum 62/1 (2001) 61.

PAGE 394

394 [89] R. Caballero, C. Guilln, Thin Solid Films 403 404 (2002) 107. [90] M. Klenk, O. Schenker, V. Alberts, E. Bucher, Semiconductor Science and Technology 17/5 (2002) 435. [91] K. Knapp, T. Jester, Solar Energy 71 (2001) 165. [92] V.K. Kapur, A. Bansal, P. Le O.I. Asensio, Thin Solid Films 431 432/Copyright (C) 2010 American Chemical Society (ACS). All Rights Reserved. (2003) 53. [93] V. Kapur, R. Kemmerle, A. Bansal, J. Haber, J. Schmitzberger, P. Le, D. Guevarra, V. Kapur, T. Stempien, 33rd IEEE Photovoltai c Specialists Conference, 2008., 2008, p. 1. [94] J. Woodruff, J.K.J. Van Duren, R. Robinson, B.M. Sager Patent 2010. [95] www.nanosolar.com [96] M. Pourbiax (Ed.), Atlas of Electrochemical Equilibria in Aqueous Solutions, Permagon Press, Newyork, 1966. [9 7] A. Ashida, Y. Hachiuma, N. Yamamoto, T. Ito, Y. Cho, Journal of Materials Science Letters 13/16 (1994) 1181. [98] C. Y. Su, W. H. Ho, H. C. Lin, C. Y. Nieh, S. C. Liang, Solar Energy Materials and Solar Cells 95/1 (2011) 261. [99] J.M. Merino, M. Len, F. Rueda, R. Diaz, Thin Solid Films 361 362/0 (2000) 22. [100] J.H. Shi, Z.Q. Li, D.W. Zhang, Q.Q. Liu, Z. Sun, S.M. Huang, Progress in Photovoltaics: Research and Applications 19/2 (2010) 160. [101] M.C. Artaud, F. Ouchen, L. Martin, S. Duchemin, Thin Sol id Films 324/1 2 (1998) 115. [102] S. Shirakata, Y. Kannaka, H. Hasegawa, T. Kariya, S. Isomura, Japan. J. Appl. Phys. 38 (1999) 4997. [103] J. Zhao, A. Wang, M.A. Green, F. Ferrazza, Applied Physics Letters 73/14 (1998) 1991. [104] O. Schultz, S.W. Glunz, G.P. Willeke, Progress in Photovoltaics: Research and Applications 12/7 (2004) 553. [105] S. Bengali, D. Borrello, E. Vallat Sauvian, J. Meier, U. Kroll, J. Hotzel, S. J., S. J., C. L., D. Y., 24th European Photovoltaic Solar Energy Conference, Hamburg, 2 009.

PAGE 395

395 [106] X. Wu, J.C. Keane, R.G. Dhere, C. DeHart, D.S. Albin, A. Duda, T.A. Gessert, S. Asher, D.H. Levi, P. Sheldon, Proceedings of the 17th European Photovoltaic Solar Energy Conference, Munich,Germany, 2001, p. p.995. [107] N. Koide, R. Yamanaka, H. Katayama, MRS Proceedings 2009, 2009, p. 1211. [108] www.konarka.com [109] S. Aksu, J. Wang, B.M. Basol, Electrochemical and Solid State Letters 12/5 (2009) D33. [110] W.K. Kim, Chemical Engineering, University of Florida, Gainesville, 2006. [111] Z. Du, C Guo, M. Tao, C. Li, International Journal of Materials Research 3 (2008) 294. [112] S. Verma, N. Orbey, R.W. Birkmire, T.W.F. Russell, Progress in Photovoltaics: Research and Applications 4/5 (1996) 341. [113] S.T. Lakshmikumar, A.C. Rastogi, Solar Energ y Materials and Solar Cells 32/1 (1994) 7. [114] J. B. Li, M.C. Record, J.C. Tedenac, Metallkd 94/4 (2003) 381. [115] J. Dieleman, F.H.M. Sanders, J.H.J. Van Dommelen, Philips.J.Res 37 (1982) 204. [116] F. Zheng, J.Y. Shen, Y.Q. Liu, W.K. Kim, M.Y. Chu, M. Ider, X.H. Bao, T.J. Anderson, Calphad 32/2 (2008) 432. [117] A. Brummer, V. Honkimki, P. Berwian, V. Probst, J. Palm, R. Hock, Thin Solid Films 437/1 2 (2003) 297. [118] I. Repins, M.A. Contreras, B. Egaas, C. DeHart, J. Scharf, C.L. Perkins, B. To, R. Noufi, Progress in Photovoltaics: Research and Applications 16/3 (2008) 235. [119] B.M. Basol, V.K. Kapur, A. Halani, C.R. Leidholm, J. Sharp, J.R. Sites, A. Swartzlander, R. Matson, H. Ullal, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films 14/4 (1996) 2251. [120] K. Kushiya, S. Kuriyagawa, T. Kase, M. Tachiyuki, I. Sugiyama, Y. Satoh, M. Satoh, H. Takeshita, Photovoltaic Specialists Conference, 1996., Conference Record of the Twenty Fifth IEEE, 1996, p. 989. [121] A. Ihlal, K. Bouabid D. Soubane, M. Nya, O. Ait Taleb Ali, Y. Amira, A. Outzourhit, G. Nouet, Thin Solid Films 515/Copyright (C) 2010 American Chemical Society (ACS). All Rights Reserved. (2007) 5852. [122] v.B. Roedern, National Solar Technology Roadmap: CIGS PV, NREL, 2007

PAGE 396

396 [123] S. Kim, W.K. Kim, R.M. Kaczynski, R.D. Acher, S. Yoon, T.J. Anderson, O.D. Crisalle, E.A. Payzant, S.S. Li, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films 23/2 (2005) 310. [124] W. Kim, S. Kim, E. Payzant, S. Speakman, S. Yoo n, R. Kaczynski, R. Acher, T. Anderson, O. Crisalle, S. Li, Journal of Physics and Chemistry of Solids 66/11 (2005) 1915. [125] W. Kim, E. Payzant, S. Kim, S. Speakman, O. Crisalle, T. Anderson, Journal of Crystal Growth 310/12 (2008) 2987. [126] W. Kim, E Payzant, S. Yoon, T. Anderson, Journal of Crystal Growth 294/2 (2006) 231. [127] R. Noufi, A. Gabor, J. Tuttle, A. Tennant, M. Contreras, D. Albin, J. Carapella, USA Patent 1994. [128] R.A. Mickelsen, W.S. Chen, Proceedings of the 15th IEEE Photovoltaic Specialists Conference, New York, 1981, p. 800. [129] J.R. Tuttle, M. Contreras, M.H. Bode, D. Niles, D.S. Albin, R. Matson, A.M. Gabor, A. Tennant, A. Duda, R. Noufi, Journal of Applied Physics 77 (1995) 153. [130] J. Kessler, C. Chityuttakan, J. Lu, J. S chldstrm, L. Stolt, Progress in Photovoltaics: Research and Applications 11/5 (2003) 319. [131] S.M. Schleussner, T. Trndahl, M. Linnarsson, U. Zimmermann, T. Wtjen, M. Edoff, Progress in Photovoltaics: Research and Applications 20/3 (2012) 284. [132] J.Y. Shen, W.K. Kim, C. H. Chang, W. Zhuang, T.J. Anderson, Submitted for publication. [133] S. Shian, K.H. Sandhage, Review of Scientific Instruments 80/11 (2009) 115108. [134] T. Schlenker, V. Laptev, H.W. Schock, J.H. Werner, Thin Solid Films 480 481/Co pyright (C) 2010 American Chemical Society (ACS). All Rights Reserved. (2005) 29. [135] M. IDER, Chemical Engineering, University of Florida, Gainesville, 2003. [136] D. Chen, D. Wang, physica status solidi (a) 208/10 (2011) 2415. [137] K. Kushiya, Oyo But suri 75/Copyright (C) 2010 American Chemical Society (ACS). All Rights Reserved. (2006) 870. [138] J. Song, T.J. Anderson, S.S. Li, Photovoltaic Specialists Conference, 2008. PVSC '08. 33rd IEEE, 2008, p. 1.

PAGE 397

397 [139] J. Song, S.S. Li, C.H. Huang, O.D. Crisall e, T.J. Anderson, Solid State Electronics 48/1 (2004) 73. [140] K. Kushiya, M. Tachiyuki, T. Kase, I. Sugiyama, Y. Nagoya, D. Okumura, M. Sato, O. Yamase, H. Takeshita, Solar Energy Materials and Solar Cells 49/1 4 (1997) 277. [141] D.J. Schroeder, G.D. Be rry, A.A. Rockett, Applied Physics Letters 69/26 (1996) 4068. [142] O. Lundberg, J. Lu, A. Rockett, M. Edoff, L. Stolt, Journal of Physics and Chemistry of Solids 64/9 10 (2003) 1499. [143] W.K. Kim, G.M. Hanket, W.N. Shafarman, Photovoltaic Specialists Co nference (PVSC), 2009 34th IEEE, 2009, p. 000844. [144] J.S. Chen, E. Kolawa, M.A. Nicolet, R.P. Ruiz, Solar Cells 30/1 4 (1991) 451. [145] M. Purwins, R. Enderle, M. Schmid, P. Berwian, G. Mller, F. Hergert, S. Jost, R. Hock, Thin Solid Films 515/15 (200 7) 5895. [146] D. G. Kim, C. Y. Lee, S. B. Jung, Journal of Materials Science: Materials in Electronics 15/2 (2004) 95. [147] C.D.R. Ludwig, T. Gruhn, C. Felser, T. Schilling, J. Windeln, P. Kratzer, Physical Review Letters 105/2 (2010) 025702. [148] J.A. Thronton, T.C. Lommasson, D.G. Cornog, J. Sol. Energy Eng. 108/4 (1986) 259. [149] T. Nakada, K. Migita, A. Kunioka, Jpn. J. Appl. Phys., Part 2 32/8B (1993) L1169. [150] T. Godecke, T. Haalboom, F. Ernst, Z. Metallkd. 91/8 (2000) 622. [151] T. Godecke, T. Haalboom, F. Ernst, Z. Metallkd. 91/8 (2000) 635. [152] T. Godecke, T. Haalboom, F. Ernst, Z. Metallkd. 91/8 (2000) 651. [153] S.S. Kulkarni, G.T. Koishiyev, H. Moutinho, N.G. Dhere, Thin Solid Films 517/7 (2009) 2121. [154] S.N. Kundu, D. Bhattacharyya, S. Chaudhuri, A.K. Pal, Materials Chemistry and Physics 57/3 (1999) 207. [155] J. Palm, V. Probst, F.H. Karg, Solar Energy 77/6 (2004) 757. [156] V. Probst, W. Stetter, W. Riedl, H. Vogt, M. Wendl, H. Calwer, S. Zweigart, K.D. Ufert, B. Freienstein, H. Cer va, F.H. Karg, Thin Solid Films 387/1 2 (2001) 262.

PAGE 398

398 [157] T. Pisarkiewicz, H. Jankowski, Vacuum 70/2 3 (2003) 435. [158] H.K. Song, S.G. Kim, H.J. Kim, S.K. Kim, K.W. Kang, J.C. Lee, K.H. Yoon, Solar Energy Materials and Solar Cells 75/1 2 (2003) 145. [159 ] A.A. Abu Sehly, S.N. Alamri, A.A. Joraid, Journal of Alloys and Compounds 476/1 2 (2009) 348. [160] F. Hergert, R. Hock, A. Weber, M. Purwins, J. Palm, V. Probst, Journal of Physics and Chemistry of Solids 66/11 (2005) 1903. [161] W. Liu, J. G. Tian, Q. He, F. Y. Li, C. J. Li, Y. Sun, Thin Solid Films 519/1 (2010) 244. [162] G.M. Hanket, W.N. Shafarman, B.E. McCandless, R.W. Birkmire, Journal of Applied Physics 102/7 (2007) 074922. [163] L. Wei, et al., Journal of Physics D: Applied Physics 42/12 (2009) 1 25303. [164] D. Wolf, G. Mller, Thin Solid Films 361 362 (2000) 155. [165] M. Avrami, Journal of Chemical Physics 7 (1939). [166] N.X. Sun, X.D. Liu, K. Lu, Scripta Materialia 34/8 (1996) 1201. [167] N. Kohara, S. Nishiwaki, Y. Hashimoto, T. Negami, T. W ada, Solar Energy Materials and Solar Cells 67/1 4 (2001) 209. [168] R. Wrz, D. Fuertes Marrn, A. Meeder, A. Rumberg, S.M. Babu, T. Schedel Niedrig, U. Bloeck, P. Schubert Bischoff, M.C. Lux Steiner, Thin Solid Films 431 432 (2003) 398. [169] B. Predal, Springer Materials The Landolt Bornstein Database, ( www.springermaterials.com) [170] B. Predal, Springer Materials The Landolt Bornstein Database, ( www.spr ingermaterials.com [171] F. Le Berre, D. Tshimanga, A.L. Guilloux, J. Leclercq, M. Sergent, O. Pea, R. Horyn, A. Wojakowski, Physica B: Condensed Matter 228/3 4 (1996) 261. [172] B.D. Davis, W.R. Robinson, Journal of Solid State Chemist ry 85/2 (1990) 332. [173] S.F. Hulbert, J.Br.Ceram.Soc 6 (1969) 11. [174] M. Avrami, Journal of Chemical Physics 8 (1940). [175] N. Wakatsuki, Y. Sasaki, Bull.Chem.Soc.Jpn. 67/5 (1994).

PAGE 399

399 [176] B. Brox, I. Olefjord, Surface and Interface Analysis 13/1 (1988 ) 3. [177] Y. Okamoto, H. Tomioka, T. Imanaka, S. Teranishi, Journal of Catalysis 66/1 (1980) 93. [178] S. Yoon, T. Yoon, K. S. Lee, S. Yoon, J.M. Ha, S. Choe, Solar Energy Materials and Solar Cells 93/6 7 (2009) 783. [179] I. Ktschau, T. Hahn, A. Kampman n, J. Hinze, S. Gorse, O. Pursche, H. Stillrich, CIGSe Process and Equipment Design using real time X ray Monitoring Methods, PVcomB, Berlin, 2011. [180] V. Alberts, R. Swanepoel, Journal of Materials Science: Materials in Electronics 7/2 (1996) 91. [181] R. Caballero, C. Guilln, Applied Surface Science 238/1 4 (2004) 180. [182] V. Alberts, J.H. Schon, M.J. Witcomb, E. Bucher, M. Ruhle, H.W. Schock, Journal of Physics D: Applied Physics 31/20 (1998) 2869. [183] R. Kamada, W.N. Shafarman, R.W. Birkmire, Sol ar Energy Materials and Solar Cells 94/3 451. [184] M. Br, I. Repins, M.A. Contreras, L. Weinhardt, R. Noufi, C. Heske, Applied Physics Letters 95/5 (2009) 052106. [185] R. Caballero, C.A. Kaufmann, T. Eisenbarth, M. Cancela, R. Hesse, T. Unold, A. Eicke, R. Klenk, H.W. Schock, Thin Solid Films 517/7 (2009) 2187. [186] V. Probst, J. Rimmasch, W. Riedl, W. Stetter, J. Holz, H. Harms, F. Karg, H.W. Schock, Photovoltaic Energy Conversion, 1994., Conference Record of the Twenty Fourth. IEEE Photovoltaic Specia lists Conference 1994, 1994 IEEE First World Conference on, 1994, p. 144. [187] F. Kessler, D. Rudmann, Solar Energy 77/6 (2004) 685. [188] F. Hergert, R. Hock, A. Weber, M. Purwins, J. Palm, V. Probst, Journal of Physics and Chemistry of Solids 66/11 (2 005) 1903. [189] D. Rudmann, M. Kaelin, F.J. Haug, F. Kurdesau, H. Zogg, A.N. Tiwari, 3rd World Conference on Photovoltaic Energy Conversion, Osaka, Japan, 2003, p. 376. [190] Y.G. Chun, K.H. Kim, K.H. Yoon, Thin Solid Films 480 481/0 (2005) 46. [191] S. A hn, K. Kim, K. Yoon, Current Applied Physics 8/6 (2008) 766. [192] Q. Guo, S.J. Kim, M. Kar, W.N. Shafarman, R.W. Birkmire, E.A. Stach, R. Agrawal, H.W. Hillhouse, Nano Letters 8/9 (2008) 2982.

PAGE 400

400 [193] J.H. Lee, J. Chang, J. H. Cha, Y. Lee, J.E. Han, D. Y. J ung, E.C. Choi, B. Hong, European Journal of Inorganic Chemistry 2011/5 (2011) 647. [194] Q. Guo, G.M. Ford, H.W. Hillhouse, R. Agrawal, Nano Letters 9/8 (2009) 3060. [195] M.G. Panthani, V. Akhavan, B. Goodfellow, J.P. Schmidtke, L. Dunn, A. Dodabalapur, P.F. Barbara, B.A. Korgel, Journal of the American Chemical Society 130/49 (2008) 16770. [196] Hens, Chemistry of Materials 21/8 (2009) 1743. [197] L. Li, N. Coates, D. Moses, Jo urnal of the American Chemical Society 132/1 (2009) 22. [198] M.A. Contreras, M.J. Romero, B. To, F. Hasoon, R. Noufi, S. Ward, K. Ramanathan, Thin Solid Films 403 404/0 (2002) 204. [199] S. Kijima, T. Nakada, Applied Physics Express 1 (2008) 075002. [200] L.P. Marushko, Y.E. Romanyuk, L.V. Piskach, O.V. Parasyuk, I.D. Olekseyuk, S.V. Volkov, V.I. Pekhnyo, Journal of Alloys and Compounds 505/1 101. [201] O.F. Zmiy, I.A. Mishchenko, I.D. Olekseyuk, ChemInform 35/24 (2004) 49. [202] J.H. Ermer, R.B. Love, A.K Khanna, S.C. Lewis, F. Cohen, Proceedings of 18th Photovoltaic Specialists Conference, Las Vegas, NV, 1985, p. 1655. [203] C.D. Lokhande, G. Hodes, Solar Cells 21/1 4 (1987) 215. [204] H.J.T. Ellingham, Journal of Society of Chemical Industry 63/5 (1944) 125. [205] C. Chatillon, J. [206] D. Schmid, M. Ruckh, H.W. Schock, Solar Energy Materials and Solar Cells 41 42/0 (1996) 281. [207] H. Hahn, G. Frank, W. Klinger, Z. Anorg. Allg. Chem., 271 (1953) 153. [208] L.S. Palatnik, E.K. Belova, IZV. Akad. Nauk. SSSR. Neorgan. Mater 3 (1967) 2194. [209] J. Mikkelsen, Journal of Electronic Materials 10/3 (1981) 541. [210] L.S. Palatnik, E.K. Belova, Inorg. Mater. 1 (1966) 1703. [211] I.V. Bodnar, A.P. Bologa, Cryst al Research and Technology 17/3 (1982) 339.

PAGE 401

401 [212] L.S. Palatnik, E.I. Rogacheva, Izvestiya Akademii Nauk SSSR, Neorganischke Materialy 2 (1966) 659. [213] L.S. Palatnik, E.K. Belova, Inorg. Mater. 2/657 662 (1966). [214] F. Sommer, Z.Metallkd 73 (1982) 72.

PAGE 402

402 BIOGRAPHICAL SKETCH Rangarajan Krishnan was born in 1982 in India. He earned his undergraduate in chemical engineering in May 2004 from UDCT, Mumbai After his undergraduate gineering from University of Toledo and graduated in August 2007. He then joined PhD program in chemical engineering at University of Florida and worked for Professor Tim Anderson and he chose photovoltaics as his research topic. Upon graduation, he will w ork for Intel Corporation at Hillsboro, Oregon.