Citation
Thermodynamic Assessments and Atomic Layer Deposition for CIGS-Based Thin Film Solar Cell

Material Information

Title:
Thermodynamic Assessments and Atomic Layer Deposition for CIGS-Based Thin Film Solar Cell
Creator:
Li, Zhi
Publisher:
University of Florida
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Chemical Engineering
Committee Chair:
ANDERSON,TIMOTHY J
Committee Co-Chair:
REN,FAN
Committee Members:
JIANG,PENG
MYERS,MICHELE V

Subjects

Subjects / Keywords:
acigs
ald
annealing
buffer
calphad
cigs
diagram
phase
solar
tco
thermodynamic
zno

Notes

General Note:
The thermodynamic database for the Ag-Cu-Ga-Se quaternary system was optimized against all available experimental data using the CALPHAD (CALculation of PHAse Diagram) method in order to aid the development of (Ag,Cu)(In,Ga)Se2 selenization processes. The Bragg-Williams random mixing model, sublattice model and ionic liquid model were used to simulate the Gibbs free energy of the system. A set of consistent parameters was obtained by the CALPHAD method, and reasonable agreement was reached between the thermodynamic properties for various phases and phase relations of this system. Zinc Oxysulfide (ZnO1-xSx) thin film was deposited using Atomic Layer Deposition (ALD) on CIGS thin film substrate, serving as an alternative buffer layer material to CdS for wide-bandgap CIGS thin film solar cells. XPS studies of the deposited films indicated that the concentration of sulfur in the buffer layer film deviated negatively from the precursor pulse ratio. The deposited film had a polycrystalline structure and the conformality of the deposited film was confirmed. Cells were fabricated and tested with variable buffer layer composition and thickness. The highest efficiency, 11.58%, was achieved with 0.9 H2S pulse fraction. The efficiency decreases with decreasing and increasing S concentration, reaching 0 % at S/(S+O) ratio of 0.5. A lower fill factor and short circuit current density is primarily responsible for the reduced performance, which may be caused by parasitic resistances induced by recombination at the absorber/buffer layer interface or the conduction bandgap offset. In and Ga were doped into ZnO thin film. The comparable minimum resistivity 1.65 x 10-3 ohm cm was reached using 0.1 and 0.05 TMIn and TEGa pulse fraction for ZnO:In and ZnO:Ga, respectively. For the samples with minimum resistivity, the ZnO:In shows high carrier concentration but low mobility while ZnO:Ga shows high mobility but relatively lower carrier concentration. The efficiency of complete cells using ZnO:In and ZnO:Ga as TCO materials was measured as 6.03 and 11.51 %. The cell with ZnO:Ga shows better VOC and JSC because the Ga-doped ZnO shows better transmission and its larger bandgap and moderate carrier concentration may lead to a more favorable band offset at the interface.

Record Information

Source Institution:
UFRGP
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
12/31/2017

Downloads

This item has the following downloads:


Full Text

PAGE 1

1 THERMODYNAMIC ASSESSMENTS AND ATOMIC LAYER DEPOSITION FOR CIGS BASED THIN FILM SOLAR CELL By ZHI LI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 201 6

PAGE 2

2 201 6 Zhi Li

PAGE 3

3 To my f amily

PAGE 4

4 ACKNOWLEDGMENTS I would like to express my appreciation to my advisory committee chair, Dr. Timothy J. Anderson for his support during my Ph.D. program at University of Florida. I learned not only scientific knowledge but also the way to do research as an engineer and Ph.D under his excellent guidance Furthermore, I was definitely inspired by Dr. Anderson on how to utilize creative thinking and solve problems independently when I met dea d ends in my res earch. These two things I learned will give me a lot of strength whenever I meet more difficulties in future. I also would like to thank all other my committee members, Dr. Fan Ren, Dr. Peng Jiang and Dr. Michele Manuel for the ir support an d guidance during the past years. When I just joint the group, I knew nothing about the semiconductor research, I was lucky enough to have m any awesome group members who ga ve me help and guidance without preservation. I would like to thank Dr. Rangarajan K rishnan for being my mentor. Without his help, I would not master the knowledge about PV and thin film in such a short time. Also I would like to thank Dr. Chris topher for teaching me the knowledge about the vacuum deposition system and Madam She n for her generous help as well as excellent teaching and guidance on thermodynamics assessment. Furthermore, I must thank Dr. Paul Carpinone Dr. Eric Lambers, Dr. Paul Holloway and Dr. Mark Davidson for giving me the access to and training on all these c haracterization and deposition tools. I would also like to thank Dr. Urs Schoop and Dr. Ryan Kaczynski for providi ng me with the samples whenever I needed and all the group members for their help and insights in my research. Finally, I would like to expres s my special thanks to all my family for their support, without which I would not be able to make this achievement.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURE S ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 14 ABSTRACT ................................ ................................ ................................ ................... 16 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 18 Background ................................ ................................ ................................ ............. 18 Problem Statement ................................ ................................ ................................ 19 Present Contribution ................................ ................................ ............................... 22 2 THERMODYNAMIC OPTIMIZATION OF THE AG CU SYSTEM ........................... 27 Ag Cu System in Literature ................................ ................................ ..................... 27 Phase Diagram ................................ ................................ ................................ 27 Thermodynamic Properties ................................ ................................ .............. 29 Thermodynamic Modeling ................................ ................................ ................ 29 Thermo dynamic Models ................................ ................................ .......................... 30 Pure Elements ................................ ................................ ................................ .. 30 Liquid Phase and Solid Phase ................................ ................................ .......... 30 Result and Discussion ................................ ................................ ............................ 31 Summary ................................ ................................ ................................ ................ 34 3 THERMODYNAMIC OPTIMIZATION OF THE AG SE SYSTEM ........................... 43 Literature Review ................................ ................................ ................................ .... 44 Phase Diagram ................................ ................................ ................................ 44 Thermodynamic Properties ................................ ................................ .............. 46 Thermodynamic Models ................................ ................................ .......................... 46 Unary Phases ................................ ................................ ................................ ... 46 Liquid Phases ................................ ................................ ................................ ... 46 Non stoichiometric Compound Ag 2 Se ................................ .............................. 47 Calculations of Gibbs Energy ................................ ................................ .................. 49 Thermodynamic Modeling and Results ................................ ................................ ... 51 Summary ................................ ................................ ................................ ................ 54 4 THERMODYNAMIC OPTIMIZATION OF THE AG CU SE SYSTEM ..................... 67

PAGE 6

6 Ag Cu Se System in Literature ................................ ................................ ............... 67 Ag Cu System ................................ ................................ ................................ .. 67 Ag Se System ................................ ................................ ................................ .. 68 Cu Se System ................................ ................................ ................................ .. 70 Ag Cu Se System ................................ ................................ ............................. 71 Phase diagram ................................ ................................ ........................... 71 Ternary compound AgCuSe ................................ ................................ ....... 72 Thermodynamic Models ................................ ................................ .......................... 74 Unary Terminal Phases ................................ ................................ .................... 74 Liquid Phases ................................ ................................ ................................ ... 74 Compound Phases ................................ ................................ ........................... 75 Results and Discussion ................................ ................................ ........................... 76 Optimization Results ................................ ................................ ........................ 76 Isopleth Diagram Prediction ................................ ................................ ............. 78 Summary ................................ ................................ ................................ ................ 79 5 THERMODYNAMIC OPTIMIZATION OF THE AG GA SE, AG IN SE AND HIGHER ORDER SYSTEM ................................ ................................ .................... 92 Literature Review ................................ ................................ ................................ .... 92 Ag Ga Se System ................................ ................................ ............................ 92 Ag In Se System ................................ ................................ .............................. 93 Thermodynami c Models ................................ ................................ .......................... 94 Unary Terminal Phases ................................ ................................ .................... 94 Liquid Phases ................................ ................................ ................................ ... 94 Compound Phases ................................ ................................ ........................... 95 Assessment Results ................................ ................................ ............................... 96 Ag Ga Se System ................................ ................................ ............................ 96 Ag In Se System ................................ ................................ .............................. 97 Higher Order System and Practical Application ................................ ...................... 98 6 THE EFFECT OF RAPID THERMAL ANNEALING ON THE PERFORMANCE OF CIGS CELLS WITH A N ITO LAYER ................................ ............................... 114 Literature Revie w ................................ ................................ ................................ .. 114 Experimental Method ................................ ................................ ............................ 115 Results and Discussion ................................ ................................ ......................... 116 Summ ary ................................ ................................ ................................ .............. 122 7 ZNO X S 1 X THIN FILM AS ALTERNA TIVE BUFFER LAYER MA TERIAL USING ATOMINC LAYER DEOPSI TION ................................ ................................ ......... 133 Literature Review ................................ ................................ ................................ .. 133 Experiment al Procedures ................................ ................................ ...................... 134 Experiment Results and Discussion ................................ ................................ ..... 135 ALD Process Windows ................................ ................................ ................... 135 Film Composition ................................ ................................ ............................ 137

PAGE 7

7 Morphologi cal and Structural Properties ................................ ........................ 139 Optical Properties ................................ ................................ ........................... 142 ZnO x S 1 x Thin Film on CIGS with Stainless Steel Substrate ........................... 144 Complete Cell Performance ................................ ................................ ........... 145 Su mmary ................................ ................................ ................................ .............. 148 8 ATOMIC LAYER DEPOSITION OF IN DOPED ZNO THIN FILM ......................... 173 Literature Review ................................ ................................ ................................ .. 173 Experiment al Procedures ................................ ................................ ...................... 174 Experiment Results and Discussion ................................ ................................ ..... 175 Film Composition ................................ ................................ ............................ 175 Structural Properties ................................ ................................ ....................... 177 Morphological Properties ................................ ................................ ................ 179 El ectrical Properties ................................ ................................ ....................... 180 Complete Cell Performance ................................ ................................ ........... 1 81 Summary ................................ ................................ ................................ .............. 182 9 COMPARIASON STUDY OF ZNO:IN AND ZNO:GA T HIN FILM FOR PV USING ATOMINC LAYER DEOPSITION ................................ ............................. 191 Literature Review ................................ ................................ ................................ .. 191 Experimental Procedures ................................ ................................ ...................... 192 Results and Discussion ................................ ................................ ......................... 193 Electrical Properties ................................ ................................ ....................... 193 Crysta lline Structure ................................ ................................ ....................... 195 Morphological Properties ................................ ................................ ................ 198 Optical Properties ................................ ................................ ........................... 199 Cell Performance ................................ ................................ ............................ 202 Summary ................................ ................................ ................................ .............. 204 10 CONCLUSION ................................ ................................ ................................ ...... 220 Thermodynamic Assessment of Ag Cu In Ga Se System ................................ .... 220 Buffer and TCO Layer Deposition for CIGS Thin Film Solar Cell Using ALD ........ 221 Suggestions for Future Work ................................ ................................ ................ 222 Thermodynamics Properties and Phase Relation Measurement .................... 222 Deposition Condition, Post Annealing and Deposition Method ....................... 223 LIST OF REFERENCES ................................ ................................ ............................. 224 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 245

PAGE 8

8 LIST OF TABLES Table page 2 1 Special Points in the Ag Cu System ................................ ................................ ... 35 2 2 Thermodynamic model parameters of the Ag Cu system ................................ ... 35 3 1 Summary of experimentally measured thermodynamic properties of Ag 2 Se ...... 55 3 2 Summary of experimentally measured enthalpy of transition of Ag 2 Se .............. 55 3 3 Details of DFT calculations for each compound including the space group. ....... 56 3 4 Relative energy with respect to fcc Ag and Trigonal Se from DFT calculations. ................................ ................................ ................................ ........ 56 3 5 Optimized parameters for the Ag Se system ................................ ..................... 57 3 6 Temperatures and compositions of the invariant reactions in the Ag Se system ................................ ................................ ................................ ................ 59 4 1 Phase transformation temperature of AgCuSe ................................ ................... 80 4 2 Optimized parameters for the Ag Cu Se system ................................ ............... 80 4 3 Calculated and measured Invariant reaction for Ag 2 Se and Cu 2 Se pseudo binary system. ................................ ................................ ................................ .... 80 5 1 Thermodynamic properties of the ternary compounds in Ag Ga Se system. ... 101 5 2 Optimized parameters of Ag Ga Se and Ag In Se system. .............................. 102 6 1 Summary of Change of Device Properties after Dark Annealing ...................... 123 7 1 Parameter table from J V curves of devices using buffer layer prepared with various H 2 S pulse fraction. ................................ ................................ ............... 149 7 2 Parameter table from J V curves of devices using buffer layer prepared with H 2 O/H 2 S pulse ratio 3/1. ................................ ................................ ................... 149 8 1 Electrical Properties of ZnO:In (ALD) and ZnO:Al (Sputtering) ......................... 183 8 2 Properties of Complete Solar Cells. ................................ ................................ .. 183 9 1 Parameters of the complete cell performance ................................ .................. 205 9 2 XPS binding energy of ZnO:Ga, ZnO:In, CdS and TCOs/CdS interface .......... 205

PAGE 9

9 LIST OF FIGURES Figure page 1 1 Change of the source of energy consumption worldwide. ................................ .. 26 1 2 Research cell efficiency records. ................................ ................................ ........ 26 2 1 Assessed Ag Cu phase diagram. ................................ ................................ ....... 36 2 2 Assessed Ag Cu phase diagram and enlarged phase diagram between 0 and 0.11 in mole fraction of Cu with selected experimental data. .............................. 37 2 3 Activity of Cu in solid phase on silver rich side ................................ .................. 38 2 4 Enthalpy of mixing of Ag Cu system in solid phase. ................................ ........... 39 2 5 Excess Gibbs free energy of Ag Cu system in liquid phase at 1400 K. .............. 40 2 6 Activity of Ag and Cu in liquid phase at 1400 K ................................ ................. 41 2 7 Enthalpy of mixing of Ag Cu system in liquid phase at 13752 K. ...................... 42 3 1 Ag Se phase diagram a ssessed by Karakaya et al ................................ ........... 60 3 2 Calculated phase diagram of the Ag Se system ................................ ................ 61 3 3 Low relative energies of phase with compositions of Ag 3 Se 2 Ag 4 Se 2 and Ag 5 Se 2 ................................ ................................ ................................ ................ 62 3 4 Calculated phase diagram in the Ag Se system near the Ag 2 Se polymorphic transformation region in comparison with experimental data .............................. 63 3 5 Calculated phase diagram near the invariant reaction Ag 2 Ag 2 Se. ................................ ................................ ................................ ................ 64 3 6 Calculated activities of Ag and Se at 1273 K in comparison with experimental data ................................ ................................ ................................ .................... 65 3 7 Calculated heat capacity of Ag 2 Se versus temperature in comparison with experimental data ................................ ................................ ............................... 66 4 1 Pseudobinary phase diagram of Ag 2 Se and Cu 2 Se proposed by Agaev et al. ... 81 4 2 The miscibility gap in liquid phase of Ag Cu Se system at 1473 K proposed by Asano et al. ................................ ................................ ................................ .... 82 4 3 Pseudobinary phase diagram Ag 2 Se Cu 2 Se proposed by Schafer et al. ............ 83

PAGE 10

10 4 4 Calculated Ag 2 Se Cu 2 Se pseudo binary phase diagram. ................................ ... 84 4 5 Calculated Heal Capacity of AgCuSe along with experimental data. .............. 85 4 6 Isothermal diagram of Ag Cu Se system at 1247.15 K (1200 C) along with experimental data. ................................ ................................ .............................. 86 4 7 Calculated Isopleth Diagram with 15 wt.% Se superimposed with experimental data. ................................ ................................ .............................. 87 4 8 Calculated Isopleth Diagram of Ag/(Ag+Cu)=0.79 ................................ .............. 88 4 9 Calculated Isopleth Diagram of Ag/(Ag+Cu )=0.5 ................................ ................ 89 4 10 Calculated Isopleth Diagram of Ag/(Ag+Cu)=0.7 ................................ ................ 90 4 11 Calculated Isothermal Diagram of Ag Cu Se system at 873 K. .......................... 91 5 1 Gibbs free energy of AgGaSe 2 Ag 9 GaSe 6 and the two terminal compounds Ag 2 Se and Ga 2 Se 3. ................................ ................................ ........................... 104 5 2 Heat Capacity of AgGaSe 2 with experimental data. ................................ .......... 105 5 3 AgGaSe2 Se cross section superimposed with experimental data .................. 106 5 4 Calculated pseudobinary system of Ag 2 Se Ga 2 Se 3 with experime ntal data ..... 107 5 5 Heat Capacity of AgInSe 2 superimposed with experimental data ..................... 108 5 6 Gibbs energy of AgInSe 2 as the function of temperature. ................................ 109 5 7 Isothermal diagram of Ag In Se system at 450 C superimposed from the experimental data ................................ ................................ ............................. 110 5 8 Phase fraction change with the increasing Se composition. Ag/I=1, I/III=1, Ga/III=0.25. T=450 C. ................................ ................................ ..................... 111 5 9 Phase fraction change with the increasing Se composition. Ag/I=0.25, I/III=1, Ga/III=0.25. T=450 C. ................................ ................................ ..................... 112 5 10 Phase fraction change with the increasing Se composition. Ag/I=0.75, I/III=1, Ga/III=0.25. T=450 C. ................................ ................................ ..................... 113 6 1 Schematic of the RTA system ................................ ................................ .......... 124 6 2 Change in V O C with ann ealing time at 4 temperatures. ................................ ..... 124 6 3 Change in J SC with annealing time at 4 temperatures. ................................ ..... 125

PAGE 11

11 6 4 Change of Fill Factor with annealing time at 4 temperatures. ........................... 126 6 5 Change of efficiency with annealing time at 4 temperatures. ........................... 127 6 6 Quantum Efficiency of sample annealed at 50 C for 300 s. ............................ 128 6 7 Quantum Efficiency of sample annealed under 100 C for 600 seconds. ......... 129 6 8 Quantum Efficiency of sample annealed under 100 C for 900 seconds. ......... 130 6 9 Quantum Efficiency of sample annealed under 150 C for 900 seconds. ......... 131 6 10 Electrical properties of samples annealed at 85 C with RH 85%. ..................... 132 7 1 Schematic setup of ALD system design. ................................ .......................... 150 7 2 3d view schematic of the newly designed quartz substrate carrier. .................. 151 7 3 ALD window for ZnO and ZnS deposition. ................................ ........................ 152 7 4 XPS spectrum of ZnO and ZnS thin fil m without doping process. .................... 153 7 5 XPS spectra for film with different cycle ratio of ZnS to ZnO. ........................... 154 7 6 Sulfur concentration on the Group VI sublattice vs. pulse ratio. ....................... 155 7 7 Multiplex Scan on ZnO x S 1 x thin flim for O 1s, Zn 2p3 and S 2p peaks. ........... 156 7 8 AFM im age of ZnO thin film surface deposited with various DEZn and H 2 O pulse time. ................................ ................................ ................................ ........ 157 7 9 Surface roughness of ZnO thin film vs. DEZn and H 2 O pulse time. .................. 158 7 10 SEM surface of H 2 S pulse rich ZnO x S 1 x samples. ................................ ........... 159 7 11 SEM surface of H 2 O pulse rich ZnO x S 1 x samples. ................................ ........... 160 7 12 XRD spectrum of ZnO x S 1 x thin film with carious H 2 O/H 2 S pulse ratio. ............. 1 61 7 13 Crystallite size extracted from XRD FHMW of ZnO x S 1 x thin film deposited with various H 2 O pulse fraction. ................................ ................................ ........ 162 7 14 Transmission of ZnO x S 1 x with various H 2 O pulse fraction. ............................... 163 7 15 Square of absorption over wavelength of ZnO x S 1 x thin film. ............................ 164 7 16 Bandgap of ZnO x S 1 x thin film with various sulfur composition. ......................... 165 7 17 H 2 O pulse rich ZnO x S 1 x thin film on CIGS substrate without pretreatment. ..... 166

PAGE 12

12 7 18 XPS spectrum of ZnO x S 1 x thin film on CIGS substrate without any treatment. 166 7 19 Modified heater assembly showing the insertion of a magnetic plate between the heater surface and CIGS/Mo/SS GSE sample for the ALD reactor chamber. ................................ ................................ ................................ .......... 167 7 20 Cross section of CIGS sample with ZnO x S 1 x buffer layer deposited using ALD. ................................ ................................ ................................ ................. 167 7 21 XPS spectrum of CIGS substrate with ZnO x S 1 x buffer layer deposited using ALD. ................................ ................................ ................................ ................. 168 7 22 Curve of complete cells using buffer layer of various sulfur concentration. ...... 168 7 23 Change of complete CIGS solar cell performance with buffer layer deposited using various H 2 S pulse ratio. ................................ ................................ ........... 169 7 24 QE of complete cell deposited with various H 2 S pulse fraction. ........................ 170 7 25 Curve of complete cells using ZnO x S 1 x buffer layer with various thickness. .... 171 7 26 Transmission and thickness of ZnO x S 1 x thin film using H 2 O/H 2 S pulse ratio 3:1 deposited with different number of cycles. ................................ .................. 172 8 1 The layer structure of CIGS substrate prepared by Global Solar Energy. ........ 184 8 2 XPS spectrum of ZnO:In thin film on Si sputtered with Ar + for various time length. ................................ ................................ ................................ ............... 184 8 3 In composition measured with different technique. ................................ ........... 185 8 4 XPS peak posi tions of the elements in ZnO:In thin film. ................................ ... 185 8 5 XRD spectra of pure ZnO and ZnO:In with different In levels. .......................... 186 8 6 The change of d spacing of ZnO 100 and 101 direction with increasing In doping level. ................................ ................................ ................................ ..... 186 8 7 Estimated stress and crystallite size of In doped ZnO film v s. TMIn pulse fraction. ................................ ................................ ................................ ............. 187 8 8 Surface SEM image of ZnO:In with different In doping level. ........................... 187 8 9 different In doping Level. ................................ ................................ .................. 188 8 10 The roughness surface evolution with the In doping level of ZnO:In films as ................................ ........................... 189

PAGE 13

13 8 11 Hall measurements results for ZnO:In thin film. ................................ ................ 189 8 12 J V curve for two completed cells. ................................ ................................ .... 190 9 1 Schematic diagram of CIGS solar cell. ................................ ............................. 205 9 2 Schematic diagram of atomic layer deposition cycle sequence. ....................... 206 9 3 Resistivity change with the In or Ga doping level in ZnO thin flim. ................... 206 9 4 Electr ical properties of In(Ga) doped ZnO thin film with various doping level. .. 207 9 5 XRD spectrum of doped ZnO with different doping level. ................................ 208 9 6 GIXRD on doped ZnO thin film with 0.1 TMIn(TEGa) pulse fraction. ................ 209 9 7 Lattice Parameters change vs. the precursor pulse fraction. ............................ 210 9 8 Stress of In(Ga) doped ZnO vs. precursor pulse fraction. ................................ 211 9 9 Crystallite size and Microstrain of In(Ga) doped ZnO thin film v s. the precursor pulse fraction ................................ ................................ .................... 212 9 10 SEM images of doped ZnO with different precursor pulse fraction. .................. 213 9 11 AFM images of doped ZnO with different precursor pulse fraction. .................. 214 9 12 Surface roughness of In and Ga doped ZnO thin film as a function of precursor pulse fraction. ................................ ................................ ................... 214 9 13 Optical transmission spectra of doped ZnO thin film wit h various precursor pulse fraction. ................................ ................................ ................................ ... 215 9 14 doping level. ................................ ................................ ................................ ..... 216 9 15 Optical bandgap of In and Ga doped ZnO thin film. ................................ ......... 217 9 16 J V curve of complete cells using In and Ga doped ZnO thin film as TCO. ...... 218 9 17 XPS analysis on bulk CdS, ZnO:Ga, ZnO:In and ZnO:Ga (4 nm)/CdS, ZnO:In (4 nm)/CdS bilayer samples. ................................ ................................ ............ 219 9 18 Band diagram of ZnO:Ga/CdS and ZnO:In/CdS ................................ ............... 219

PAGE 14

14 LIST OF ABBREVIATIONS ACIGS Ag Cu In Ga Se AFM Atomic force microscopy ALD Atomic layer depos i tion CALPHAD Calculation of phase diagram CIGS Cu(In 1 x Ga x )Se 2 CVD Chemical vapor deposition DEZn Diethylznic DSC Differential scanning calorimetry DTA Differential thermal analysis E g Band gap of a semiconductor FF Fill factor GIXRD Gra zing incidence X ray diffraction ICP AES Inductively coupled plasma atomic emission spectroscopy J SC Short circuit current density J V Current density voltage NREL National renewable energy laboratory PV Photovoltaics PVD Physical vapor deposition QE Exter nal quantum efficiency RTA Rapid thermal annealing SEM Scanning electron spectroscopy SGTE Scientific Group Thermodata Europe TEGa Triethylgallium TEM Transmission electron spectroscopy

PAGE 15

15 TMA Trimethyaluminum TMIn Trimethylindium UV Vis Ultraviolet visible spectroscopy XPS X ray photoelectron spectroscopy XRD X ray diffraction

PAGE 16

16 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THERMODYNAMIC ASSESSMENTS AND ATOMIC LAYER DEPOSITION FOR CIGS BASED THIN FILM SOLAR CELL By Zhi Li December 2016 Chair: Timothy J. Anderson Major: Chemical Engineering The thermodynamic database for the Ag Cu Ga Se quaternary system was op timized against all available ex perimental data using the CALPHAD (CALcu lation of PHAse Diagram) method in order to aid the development of (Ag,Cu)(In,Ga)Se 2 selenization processes The Bra gg Williams random mixing model, sublattice model and ionic liquid m odel were used to simulate the Gibbs free e nergy of the system A set of consistent parameters was obtained by the CALPHAD method, and reasonable agreement was reached between the thermodynamic properties for various phases and phase relations of this syst em. Zinc Oxysulfide (ZnO 1 x S x ) thin film was deposited using Atomic Layer Deposition (ALD) on CIGS thin film substrate, serving as an alternative buffer layer material to CdS for wide bandgap CIGS thin film solar cells. XPS studies of the deposited films indicated that the concentrati on of sulfur in the buffer layer film deviated negatively from the precursor pulse ratio. T he deposited film had a polycrystalline structure and the conformality of the deposited film was confirmed Cells were fabricated and tested with variable buffer lay er composition and thickness. The highest efficiency, 11.58%, was achieved with 0.9 H 2 S pulse fraction The efficiency decreases with decreasing and

PAGE 17

17 increas ing S concentration, reaching 0 % at S/(S+O) ratio of 0.5. A lower fill factor and short circuit curr ent density is primarily responsible for the reduced performance, which may be caused by parasitic resistances induced by recombination at the absorber/buffer layer interface or the conduction bandgap offset. In and Ga were doped in to ZnO thin film The co mparable minimum resistivity 1.65 x 10 3 cm was reached using 0.1 and 0.05 TMIn and TEGa pulse fraction for ZnO:In and ZnO:Ga, respectively. For the samples with minimum resistivity, the ZnO:In shows high carrier concentration but low mobility while ZnO: Ga shows high mobility but relatively lower carrier concentration. The efficiency of complete cells using ZnO:In and ZnO:Ga as TCO materials was measured as 6.03 % and 11.51 %. The cell with ZnO:Ga shows better V OC and J SC because the Ga doped ZnO shows bett er transmission and its larger bandgap and moderate carrier concentration may lead to a more favorable band offset at the interface.

PAGE 18

18 CHAPTER 1 INTRODUCTION Background S ince the past few centuries, the demand of the energy has grown exponentially and is expected to grow further in future due to the increasing population and the technology development of the whole world. Although in 201 5 the energy demand increased by only 0. 9%, which is the slowest rate since 1998, it still put a lot of pressure on the conventional energy sources such as oil, gas and coal because this type of energy counts up to more than 80% energy consumption in the world. On the contrary, renewable energy counts less than 10%. Among all different kinds of renewable energy, hydroelectric energy reached the shares of global primary energy consumption 6.8% and the others are around 2.5% ( Figure 1 1 ). The increasing utilization of traditional fossil fuel will not meet the energy demand because it is nonrenewable and has very limited quantity. Furthermore, the consumption traditional fossil fuel releases large amount CO 2 whi ch is a huge concern in terms of the environment protection. The increasing concentration of CO 2 in atmosphere causes the rise of the temperature and the acidification of ocean. CO 2 absorbs the infrared part of radiation from the earth and re radiates it b ack to the earth, areas due to melting of ice at poles could happen if the greenhouse effect cannot be controlled in future. In order to meet the increasing demand of energy while minimize the damage to the environment, it is necessary to utilized renewable or green energy source. And solar energy is a promising energy source. Compared to other type of renewable energy

PAGE 19

19 source such as hydroelectricity, wind energy, solar energy is abundant and widely spread all over the world. Thus, the utilization of solar energy would be low cost and unrestricted geographically. Furthermore, during the use of solar energy, no waste and emission is generated and the operating and mainten ance cost is relatively low. There are several different photovoltaics materials, which can convert the solar energy directly to electricity such as Crystalline Si, thin film CIGS and CdTe, single junction GaAs etc as shown in Figure 1 2 Among all these materials, CIGS thin film solar cell has the advantage of low manufacture cost, suitability for flexible substrate and better resistance to heat. Currently, the CIGS thin film technology has demonstrated laboratory solar cell efficiency 22.3%. This performance approaches that of the Si solar cell technology, but lacks significantly in large scale production. The average commercial module performance is about 12%, and thus a large gap exists between the laboratory state of the art solar cells, and power modules from production. So the question is what are the technical barriers and the cost reduction opportunities that need to be tackled to both increase module efficien cy and reduce manufacturing cost. Problem Statement The chalcopyrite alloy Cu(In 1 x Ga x )Se 2 has received tremendous success as an absorber for thin film PV devices, demonstrating the highest efficiency of any polycrystalline thin film material [1] Th e absorber composition of this champion cell, however, is not at the bandgap energy. This has motivated research on alloying Ag with Cu to increase the bandgap energy while also i ncreasing cell efficiency. The substitution of Ag for Cu in CIGS absorber layers shifts the bandgap energy higher and towards the optimal value (1.4 eV) while decreasing materials cost. In addition to increasing the

PAGE 20

20 bandgap, researchers have begun investig ating (Ag x Cu 1 x )(In x Ga 1 x ) Se 2 alloys as a means to reduce processing temperature, improve crystalline quality, improve wetting, and attain greater power conversion efficiencies [2, 3] The synthesis process of CI GS has been well studied [4 8] and several researches have been done on the assessment of Cu In Ga Se system [9 17] to aid the opt imization of the synthesis process of CIGS. The phase relation change during the synthesis process of CIGS is complicated and the implementation of Ag would cause change of the material phase and reaction temperature. However, the effect of the Ag addition on the phase relation change is not clear, which causes the difficulties to optimize the synthesis reaction pathway for the ACIGSe. Therefore, the thermodynamic description of ACGISe system is of interest to better understand the absorber synthesis proces s. Furthermore, w ith this substitution and higher bandgap, an alternative buffer layer material to CdS is needed to retain a favorable band alignment with the absorber. For CdS buffer layer, it has been shown that an electrically resistive i ZnO layer is beneficial for reducing the shunts effect [18] However, the extra layer not only reduce the photon transmission but also increasing the manufacturing cost. Therefore, eliminating i ZnO layer would simplify the fabrication process and beneficial for the cost control. Extensive studies have bee n done on the alternative Cd free buffer layer materials [19 24] Zinc Oxysulfide (ZnO 1 x S x ) is a candidate buffer layer material since its electrical and optical properties are tunable by varying the oxygen to sulfur ratio in the film. Furthermore, the wider band gap of ZnO 1 x S x compared to CdS and the thinner film possible with conformal ALD compared with conventional chemical bath deposited

PAGE 21

21 CdS promises higher transmission especially in high wavelength range. It has been applied on various solar materials as a b uffer layer using different deposition technique [22, 25, 26] Among all these deposition technique, atomic layer deposition shows significant advantage due to its strong capability of thickness, doping concentration and film conformality control. Furthermore ALD usually requires low deposition temperature and generate less waste. This would minimize the degradation of the cell performance induced by the thermal diffusion [27] Beside the buffer layer, t he properties of transparent conducting oxides (TCO) can affect the performance of the solar cell because it not only conducts the photocurrent generated from the underlying cell but also, in the meantime, transmits the photon t o the absorber layer. Therefore, good transmission and low resistivity are two favorable properties of TCO for photovoltaic application. Stadler et al. [28] did a comprehensive review on various TCO materials including ITO, ZnO:Al etc. These n type TCOs are important for thin film solar cell production and have been extensively studied [29 33] Among all these TCO materials, ZnO based material has drawn a lot of attention for PV application due to its wide bandgap of 3.3 eV [34] Compared with Al depant Ga and In can be a better dopant candidates due to the r elatively smaller difference between their ion radius and Zn 3+ ion radius. Both In (ZnO:In) and Ga doped ZnO (ZnO:Ga) has been studied using various deposition method such as sputtering [35 42] chemical vapor Deposition [43 46] chemica l spray [47 49] pulsed laser deposition [50, 51] and sol gel method [52 54] Compa red to the deposition technique mentioned above, atomic payer deposition (ALD) shows many advantage such as the ability to

PAGE 22

22 finely control the doping level and film thickness, typically lower deposition temperature, good film uniformity and conformality. Th ere are several studies about ZnO:Ga using ALD with various Ga precursor, such as trimethylgallium (TMGa) [55] gallium(III) isopropoxide (GTIP) [56] and tri ethylgallium (TEGa) [57 59] However, studies on ZnO:In were much less. Hsia et al. and Kim et al. used trimethylindium (TMIn) [60 ] with N 2 O and dimethylamino dimethylindium (DMLDMIn) [61] with H 2 O for atomic layer deposition respectively. Although these studies developed the deposition process, the deposition using CIGS/CdS bilayer substrate has never been conducte d before. Furthermore, the high deposition temperature (above 300 C) leads to the degradation of the solar cell [27] Present Contribution Present contribution contains three parts: the assessment and thermodynamic modeling of Ag Cu In Ga Se system, the use of atomic layer deposition (ALD) to form ZnO 1 x S x as buffer layer and ZnO:Ga/In as TCO layer, the effect of rapid thermal annealing on th e complete cell. The Ag Cu system was reassessed in Chapter 2. A new self consistent model was established based on newly published Cu activity data in the FCC solid phase. In Chapter 3, o n the basis of thermochemical and phase boundary data, the Ag Se bi nary system was thermodynamically assessed and modeled The 2 sublattice and 3 sublattice models were used to model the low and high temperature phase of the intermetallic compound Ag 2 Se, respectively, based on their special order disorder structure transi tion. DFT calculation was used to calculate the Gibbs free energy of formation of the end members in the models. A set of modeling parameters was obtained by the CALPHAD method and reasonable agreement was reached between

PAGE 23

23 the thermodynamic properties for v arious phases and phase relations of this system. In Chapter 4 and Chapter 5, an assessment of the thermochemistry and phase equilibria of the Ag Cu Se, Ag Ga Se, Ag In Se and higher order system was performed based on available experimental data and predi ction s of temperatur e dependence of the Gibbs energies of formation for the intermetallic compounds. With the complete thermodynamic database, the implications for selenization of metal precursors are explored. In Chapter 6, r apid thermal annealing studies were conducted on SS/Mo/CIGS/CdS/ITO/Ag devices as a function of anneal time in dry N 2 in the temperature range 50 to 200 C, as well as in humidified N 2 (85% RH) at 85 C. Interestingly, dry N 2 annealing produced an enhancement in cell performance for l ow thermal budget, predominantly due to increased J SC and to a lesser extent higher V OC As examples, CIGS cell efficiency increased by 1.80.8% after annealing at 50 C for 300s and 2.02.3% when annealed at 100 C for 600s. At higher anneal temperatur e or longer time the cell performance deteriorated with significant decrease in FF and V OC The results for the 85% RH at 85 C anneals were similar to the dry anneal studies but with less dramatic changes. Cell performance results after dark annealing ind icate light soaking effects induced by the lamp are not important at these relatively short anneal times. Measurement of quantum efficiency for the annealed samples supports the assumption that the buffer layer/absorber interface degrades at a relatively l ow temperature and that ITO degradation takes place at higher thermal budgets In Chapter 7, Zinc Oxysulfide (ZnO 1 x S x ) thin film was deposited using Atomic Layer Deposition (ALD) on CIGS thin film substrate, serving as an alternative buffer

PAGE 24

24 layer material to CdS for wide bandgap CIGS thin film solar cells. XPS studies of the deposited films indicated that the concentration of sulfur in the buffer layer film deviated negatively from the precursor pulse ratio. The surface morphology and film structure were e xamined using SEM and XRD, respectively. Results showed that the deposited film had a polycrystalline structure and the conformality of the deposited film was excellent. Cells were fabricated and tested with variable buffer layer composition and thickness. The highest efficiency, 11.58%, was achieved with 90.9% S/(S+O) ratio. The efficiency decreases with decreasing and increas ing S concentration, reaching 0 % at S/(S+O) ratio of 0.5. A lower fill factor and short circuit current density is primarily respons ible for the reduced performance, which may be caused by parasitic resistances induced by recombination at the absorber/buffer layer interface or the conduction bandgap offset. Doping of ZnO with In during ALD by periodic exposure to TMIn was studied with an emphasis on the structural properties of the films (Chapter 8) XRD results show compressive stress in the film induced by the larger In atom, which leads to smaller grain size. Based on these results stacked layers of ZnOS/ZnO:In and sputtered sputter ed ZnOS/ZnO:Al were deposited on CIGS thin films to compare the performance of the ALD TCO to that of our standard TCO. The complete cell with ZnO:In as TCO shows relatively low J SC and Rse compared with the cell using sputtered ZnO:Al as TCO. In Chapter 9 In and Ga doped ZnO thin films with various doping level were deposited using atomic layer deposition on Si (100), borosilicate glass and CdS/CIGS/Mo/Stainless Steel substrate. The properties of ZnO:In and ZnO:Ga were characterized and the performance of the complete cell was tested and analyzed.

PAGE 25

25 ZnO:Ga was found to be a better TCO material than ZnO:In. Both films show minimum resistivity around 1.65 x 10 3 concentration and higher carrier mobility compared t o ZnO:In. XRD spectra reveal that both films exhibit hexagonal ZnO structure. In and Ga dopants induce compressive and tensil stress to the film, respectively. Compared to ZnO:Ga, ZnO:In shows less transmission and smaller bandgap due to the higher free ca rrier concentration, higher doping level and less bonding strength. The conduction band offset of ZnO:In/CdS and ZnO:Ga/CdS was measured as 0.8 eV and 0.23 eV, respectively. The complete solar cell with ZnO:In as TCO layer shows lower efficiency due to the relatively lower J SC and V OC. This can be attribute to the relatively lower transmission and higher conduction band offse t at the ZnO:In/CdS interface.

PAGE 26

26 Figure 1 1 Change of the source of energy consumpt ion worldwide [62] Figure 1 2 Research cell efficiency r ecords

PAGE 27

27 CHAPTER 2 THERMODYNAMIC OPTIMIZATION OF THE AG CU SYSTEM CIGS based solar cells have shown more than 20% photovoltaic (PV) conversion efficiency [63, 64] In order to further increase their efficiencies and minimize their production costs, Ag can be added to substitute part of Cu to form a PV material with larger band gap [65, 66] However, the reaction pathway during the synthesis process of CIGS is complicated and adding Ag into CIGS makes it even more difficult to be optimized. Fully understanding the Ag Cu In Ga Se thermodynamic system and phase equilibrium will be helpful in discovering the most efficient reaction pathway. In this work, the Ag Cu system was thermodynamically optimized as a part of the complete database and a good agreement between the calculated thermodynamic properties and experimental data was archived. Ag Cu System in Literature Phase Diagram Ac cording to extensive experimental results, the phase diagram of Ag Cu system is a simple eutectic diagram without any miscibility gap in the liquid phase. In the current work, the maximum solubility of Cu in Ag and Ag in Cu are assessed as 0.100 and 0.046 in mole fraction of Cu and Ag, respectively. The calculated phase diagram and the comparison with selected experimental data [67 69] are shown in Figure 2 2 A Special points from selected previous assessment work and from the current work are shown in Figure 2 2 B The liquid phase of Ag Cu system has been investigated by many groups since 1875 [67, 70 78] Various experimental methods such as DTA, EMF measurements, pyrometry, etc. were used in these measurements and no miscibility gap was reported.

PAGE 28

28 Among these values, the experimental data from Heycock et al. [70] were widely accepted since it has been confirmed by other experimental results obtained from using high purity metal and low oxygen contamination from Rosina et al. [67, 76] The detailed experimental values were collected in Subramanian et al. [67] The most recent eutectic point was reported at 1053K in Bienzle et al. [69] Before that, several groups [70 72, 79 81] measured the eutectic temperature ranging from 1050.15 K to 1053 K with various methods [67] In this work, the eutectic temperature 1052.10.08 K from Moser et al. [81] was used as suggested by Subramanian et al. [67] beca use the eutectic temperature purposed by in Bienzle et al. [69] was extrapolated whereas Moser et al. [81] carefully measured the temperature with high purity metal. The composition of eutectic point reported as 0.3990.002 in mo le fraction of Cu by Hansen [82] was used for optimization in this work because according to Subramanian et al. [67] previous experimental results were all in a small range between 0.398 and 0.410 of Cu in mole fracti on. Bahari et al. [83] reinvestigated the Ag Cu system by using Tamman plot. A similar results as Subramanian et al. [67] were obtained. The solidus and solvus of Ag Cu system have been reported by many groups [71 73, 75, 76, 78, 80, 84 88] No intermetallic compounds were reported in previous work. Selected experimental data are summarized in Elliott et al. [86] The experimental data on Cu rich side generally show good agreement while on the Ag rich side there was significant scattering in the data. The most recent solidus and solvus was reported by Bienzle et al. [69] In order to adopt the thermodynamic data measured in Bienzle et al. [69] these constitutional data were used for optimization in the current work but with

PAGE 29

29 relative lower weight because they were es timated by extrapolating EMF value to zero. Witusiewicz et al. [68] reported another set of solidus and solvus data measured by Sommer [89] However, these data are unpublished. Due to a lack of detailed experimental method and condition, they were not used for optimization in current work. Thermodynamic P roperties Several researchers [75, 90, 91] have conducted measurement of the thermodynamic properties of Ag Cu system in solid phase. The experimental data measured before 1990 were summarized by Subramanian et al. [67] Besides these summarized data, the activities of Cu and integral thermodynamics dat a of the solid phase on Ag rich side were measured by Bienzle et al. [69] with EMF measurement by using high purity metal. In this work, the enthalpy of mixing in solid phase at 1052 K from Subramanian et al. [67] were used since these data were in good agreement with previous experimental results [75] Furthermore, ot her newly measured thermodynamic data from Bienzle et al. [69] including the enthalpy of mixing in solid phase at 1053K and the activity of Cu in solid phase at various temperatures were used. Subramanian et al. [67] completed a summary about the thermodynamic properties in the liquid phase and some of the measurements were repeated. The published results are identical to previous work [92 98] After that, the enthalpies of mixing in liquid phase of Ag Cu system were re determined using a new in situ mixing technique at 13752 K by Fitzner et al. [99] Thermodynamic M odeling Before the current work, several groups have assessed the thermodynamic database of Ag Cu system [68, 100 104] Researches have shown that Murray et al. [100] Lim et al. [102] and Kusoffsky [103] generate deviation from the experimental

PAGE 30

30 data [67, 68] Currently, the data or the thermodynamic data reported by Hayes et al. [101] was widely used in other thermodynamic assessment research. However, the calculated data using Hayes et al. [101] shows noticeable deviation compared with experimental values reported in by Bienzle et al. [69] and Fitzner et al. [99] Although the results from Witusiewicz et al. [68] and He et al. [104] can fit the thermodynamic values from Fitzner et al. [99] their calculated results do not show satisfactory agreement with the experimental data reported by Bienzle et al. [69] Thermodynamic Models Pure Elements The description of the Gibbs energy of pure Ag and Cu was taken from the most recent complication Dinsdale [105] They were mod eled by using the following equ ation with different sets of parameters for correspondi ng temperature intervals: ( 2 1 ) where H SER means enthalpy of standard element reference referring to the enthalpy of the stable state of the pure element at 298.15 K, T is the absolute temperature and A~J are the coefficients. Liquid P hase and Solid Phase Extensive experiments have shown that ther e is no miscibility gap in the liquid phase and no stoichiometric compound was found in the solid phase. In this work, the A1) of Ag Cu system are described by Bragg Williams random mixing model. The Gibbs energy of these phases can be expressed as:

PAGE 31

31 ( 2 2 ) where refers to the phase to be described, x means element mole fraction and can be described with a simple Redlich Kister polynomial expansion: ( 2 3 ) where refers to the interaction coefficient for phase in an A B binary system. In the current work, these parameters are expressed as the first two terms of E quation 2 3 : ( 2 4 ) where a and b are the parameters for each phase to be optimized with PANDAT 8.1 software. Result and Discussion The thermodynamics parameters of the binary Ag Cu binary system were opt imized with Optimization Function in the PANDAT 8.1 software. For the rough search process, all experimental values including phase equilibrium data, enthalpy of mixing and partial molar activity in liquid phase, components activity at various temperature and enthalpy of mixing on silver rich side in solid phase, are taken into consideration. Finally, in the normal optimization process, the constitutional data on Ag solvus were excluded because these data are calculated by extrapolating the EMF value to zer o and based on the assumption that the concentration of Cu is nearly zero, which could generate large error especially for these points near 0.1 mole fraction of Cu. The phase diagram calculated using optimized database in this study was shown in Figure 2 1 Figure 2 2 A illustrates the comparison between the calculated ph ase diagram and the experimental phase equilibrium data. An enlarged phase diagram in the range

PAGE 32

32 of 0 to 0.11 mole fraction of Cu is plotted in Figure 2 2 B The optimiz ed thermodynamic parameters are summarized in Table 2 2 The calculated phase diagram generally shows a good agreement with the experimental phase equilibrium data. Ho wever, as shown in Figure 2 2 B on the Ag rich side in fcc A1 phase, the solidus and solvus deviate from some of the experimental data because the composition temperat ure data proposed by Bienzle et al. [69] were extrapolated. The extrapolation for solvus was based on the assumption that the concentration of Cu in Ag was very small and thus can be neglectable. Therefore, these data may have relatively lower accuracy and need to be confirmed by other methods, e.g. dilatometric measurements, DTA etc. Table 2 1 shows the calculated and experimental special points in the Ag Cu system including melting points, eutectic react ion and maximum solubility. It is worth noting that most of these calculated points show good agreement with the experimental data except the maximum solubility of Cu on silver rich side. This is because first, on the silver rich side, the experimental dat a were significantly scattered and secondly, in order to match with the measured activity data by Bienzle et al. [69] the extrapolated maximum solubility of Cu in Ag in the same work, which was not reconfirmed by other experimental means, was used for optimization. The activity of Cu on the silver rich side in FCC phase and selected integral thermodynamic value are plotted in Figure 2 3 and Figure 2 4 Compared with the activity calculated with Hayes et al. [101 ] the cooper activity curves generated with current work show better agreement with the experimental data measured by Bienzle et al. [69] except one data point at 0.04 mole fraction of copper at 1053 K, whi ch is considered as statistical outlier and thus was used with less weight.

PAGE 33

33 The comparison between calculated and experimental excess Gibbs free energy and partial molar activity in liquid phase at 1400 K are illustrated in Figure 2 5 and Figure 2 6 respectively. These experimental values, which have a good agreement with previous experiment results, are taken from Subramanian et al. [67] Figure 2 5 shows that the excess Gibbs free energy curve generated w ith the optimized database in current work has a good agreement with the experimental data. However, the calculated curve from Hayes et al. [101] shows negative deviation from t he experimental value and its maximum value shifts to the silver rich side, which is different from the fact that the experimental data show symmetry about 0.5 mole fraction of copper. Compared with the curve calculated with Hayes et al. [101] the optimized database generates better agreement with the experimental data in both value and symmetry aspects. The calculated and experimental data [99] of enthalpy of mixing in the liquid phase at 13752 K are plotted in Figure 2 7 As shown in the figure, the calculated enthalpy curve is identical to the experimental data, while there is a large discrepancy between the curve calculated with Hayes et al. [1 01] and the experimental data [99] The pha se diagram of Ag Cu system is an exception of the Hume Rothery theory because limited solubility is shown on both Ag rich and Cu rich side. The main reason can be that the atomic radii difference between silver and copper is 12.5%, which is very close to t he empirical value p roposed by Hume Rothery 14% 15% [106] When Ag atoms substitute the solvent Cu atoms or vice versa, extra energy will be required since relatively large lattice di stortion takes place due to the difference atomic sizes of silver and copper. Therefore, the completely solution is not thermodynamically favorable for Ag Cu system.

PAGE 34

34 Summary The thermodynamic database for the binary Ag Cu binary system was optimized based on available experimental data using the CALPHAD (CALculation of PHAse Diagram) method. The Bragg Williams random mixing model was used to describe both liquid phase and fcc A1 phase. The latest measured activity of Cu and enthalpy of mixing were used in o ptimization process. A simple eutectic Ag Cu phase diagram was reached using the optimized database. The calculated thermodynamic properties and phase diagram in the current work shows better agreement with the latest experimental data than those generated from previous databases.

PAGE 35

35 Table 2 1 Special Points in the Ag Cu System Reaction Composition Temperature Reaction type Reference mole fraction of Cu K 0 1234.98 Melting [67] 0.100 0.399 0.954 1052.25 Eutectic This work 0.129 0.416 0.955 1054.4 [101] 0.121 0.398 0.953 1053.4 [68] 0.141 0.397 0.951 1055.79 [104] 0.136 0.398 0.952 1054 [100] 0.141 0.399 0.951 1052.25 [67] 0.134 0.399 0.078 1053.05 [83] 1 1357.77 Melting [67] Table 2 2 Thermodynamic model parameters of the Ag Cu system Phase Thermodynamic Parameters Liquid fcc A1

PAGE 36

36 Figure 2 1 Assessed Ag Cu phase diagram

PAGE 37

37 Figure 2 2 Assessed Ag Cu phase diagram and enlarged phase diagram between 0 and 0.1 1 in mole fraction of Cu with selected experimental data The black line is the assed phase boundary.

PAGE 38

38 Figure 2 3 Activity of Cu in solid phase on silver rich side (reference state: fcc A1 phase of Cu) [69] [69] [69] [69] [69] 1150 K. Blue lines are the activity calculated using the database from [101] Red lines refer to the activity derived from the optimized database from

PAGE 39

39 Figure 2 4 Enthalpy of mixing of Ag Cu system in solid phase indicates experimental data measured by [69] and red line the data calculated from [101] Black line refers to the Enthalpy derived from this study.

PAGE 40

40 Figure 2 5 Excess Gibbs free energy of Ag Cu system in liquid phase at 1400 K. The [67] and the red dash line refers to Hayes et al. [101] The black curve is extract from the database assessed in this study.

PAGE 41

41 Figure 2 6 Activity of Ag and Cu in liquid phase at 1400 K (reference state: liquid phase of Ag and Cu, respectively) is the activity of Ag and Cu, respectively, taken from [67] The blue dash line refers to [101] and the red line is the activity calculated using the optimized database in this study.

PAGE 42

42 Figure 2 7 Enthalpy of mixing o f Ag Cu system in liquid phase at 13752 K. The solid [99] and the red dash line refers to [101] The black line is the curved extracted from the optimized database in this study.

PAGE 43

43 CHAPTER 3 THERMODYNAMIC OPTIMIZATION OF THE AG SE SYSTEM CuIn x Ga 1 x Se 2 ( CIGS ) based solar cells have been confirmed showing a 22.3% photovol taic (PV) conversion efficiency one of the highest efficiencies of PV thin film materials [107] The addition of Ag to substitute part of Cu can adjust the bandgap closer to the theoretical optimum band gap 1.4 eV for Air Mass ( AM ) 1.5 solar spectrum for further increas ing the conversion efficiencies [65, 66, 108] The prediction of phase equilibrium using the Ag Cu In Ga Se (ACIGS) thermodynamic database will be helpful in discov ering the most efficient reaction pathway and process conditions, making the present Ag Se system vital importan t. Furthermore, the low temperature phase, Ag 2 Ag 2 Se, is a superionic conductor [109 111] Both of them are widely used in photographic films thermochromic materials and photochargable secondary batteries [112] T he phase diagram and thermochemical properties of Ag Se system w ere extensively studied [113 116] and previously reviewed [117 119] but the thermo dynamic model has not been established yet, which makes it difficult to predict the effect of Ag on the phase relationships of the CIGS material. In this study, the thermodynamic model was set up using CALPHAD ( calculation of phase diagram ) approach [ref] combined with DFT (density functional theory) calculations based on evaluation of experimental data in the Ag Se system. A five sublattice model is applied for the high temperature phase of Ag 2 Se compound based on the study of its disordered structure with large vacancy occupancy [120, 121] The Gibbs energy of formation of endmembers in the sublattice model was o btained using DFT calculations. A two sublattice model was used for low temperature phase of Ag 2 Se and the Gibbs free

PAGE 44

44 energy of formation was also calculated with DFT approach for the endmembers. A good agreement between the calculated phase diagram and th ermodynamic properties with experimental data were archived using the self consistent thermodynamic model established in this work. Literature R eview Phase Diagram The latest Ag Se phase diagram was constructed by Karakaya et al. [117] as shown in Figure 3 1 According to this evaluation, the Ag Se system contains one intermetallic compounds Ag 2 Se with two polymorphs of Ag 2 Se with orthorhombic structure at low temperatures (space group < 405.42 K) and Ag 2 Se with body centered cubic (bcc) structure at high temperatures (space group > 405.42 K), two terminal solid phase face centered cubic (fcc) fcc Ag and trigonal Se, and two miscibility gaps in the liquid phase. Several investigations determined the liquid phase equilibrium. The existence of two miscibility gaps in the liquid phase was reported by Friedrich [122] and Pellini [123] and was confirmed by Houphouet Boigny et al. [124] as shown in Figure 3 2 Ohno et al. [125] Tsuchiya et al. [126] and Wiggett et al. [127] measured the phase boundaries of the miscibility gaps using conductivity measurement, ray attenuation and sample quenching method, respectively. These three measurements show large discrepancy, except the composition and temperature of the monotectic reaction between the liquid, fcc Ag 2 Se phase. Ohno et al. [125] and Tsuchiya et al. [126] proposed a converged miscibility gap in Ag rich liquid phase, while Wiggett et al. [127] proposed that t he boundary of miscibility gap does not converge even at high temperatures.

PAGE 45

45 Friedrich et al. [128] showed a very small solubility ( less than 0.27 at.% Se ) of Se in fcc Ag without details. Therefore, it is not considered in the present modeling work. N o solubility of fcc Ag in Trigonal Se was reported. Extensive studies have been conducted for the phase equilibrium and thermodynamic properties of Ag 2 Ag 2 Se phases Ag 2 Se was measured using thermal analysis [123] differential thermal analysis ( DTA ) [129] and conductivity measurement [125] and these results show a good consistency as 1170 K, 1167 K and 1168 K, respectively. M easured transition temperature s Ag 2 Ag 2 Se are mostly consistent expect for the data obtained by Voronin et al. [130] who used e lectromotive force (EMF) measuremen t in solid electrolyte (AgCl) or a glycerol solution with constant stirring to get the transformation temperature of 397.5 K. However, Echmaeva et al. [131] Nasar et al. [132] and Feng at el. [133] using EMF measurement, reported a higher transformation temperature as 405.5 K, 406 K and 407.7 K, respectively. The transformation temperature of 406 1 K was recently reported by Asadov et al. [134] using microstructural and X ray analysis Oehsen et al. [115] reported a small deviation from stoichiometry in Ag 2 Se and A g 2 Se phases, and the homogeneity range was determined using the solid state galvanic cell Ag/Ag 3 RbI 5 /Ag 2+ Se/C between 348 K and 463 K. Grnvold et al. [135] investigated thermodynamic properties of the Ag 2 Se and Ag 2 Se transformation region using adiabatic calorimetry and proposed a different phase diagram around th is phase transf ormation region.

PAGE 46

46 Thermodynamic P roperties The standard Gibbs energy of formation ( ) and the standard enthalpy of formation ( ) of Ag 2 Se was extensively studied using EMF measurement [115, 130 133] and adiabatic calorimetry [135 ] These results show good consistency, except that the standard Gibbs energy of formation of Ag 2 Se reported by Olin et al. [136] shows a slightly lower value (2 kJmol 1 ). The enthalpy of transformation from Ag 2 Se to Ag 2 Se has been measured using d ifferential scanning calorimetry (DSC) [114, 130, 133, 137] and adiab atic calorimetry [135, 138] The results from DSC scatter are between 6.06 kJmol 1 and 7.68 kJmol 1 while the value obtained with adiabatic calorimetry shows good consistency at ~7 kJmol 1 These experimental data were summarized in Table 3 1 and Table 3 2 Blachnik et al. [139] deter mined the activity of Se in liquid phase at 1273 K using the modified dew point method. Thermodynamic M odel s Unary Phases The description of Gibbs energ ies for pure Ag was taken from the compilation by Dinsdale [105] and that of Se was taken from Chang [140] using the following equation: ( 3 1 ) where H SER represents the standard element reference (SER) referring to the enthalpy of the stable state of the pure element at 298.15 K and 1 bar T the absolute temperature and A ~ J the modeling parameters Liquid Phases For liquid phase, the ionic liquid two sub lattice model was used by the following expression:

PAGE 47

47 ( 3 2 ) where is the Gibbs energy of formation for liquid Ag 2 Se. and are the Gibbs energies of formation of liquids Ag and Se, respectively. The values of P and Q are determined using: ( 3 3 ) ( 3 4 ) where the value 2 in Equation 3 3 and 1 in Equation 3 4 are the charges of Se 2 and Ag + ion, respectively. The term refers to the interaction between Se 2 and vacancy (Va) on the Ag rich side of the system and indicates the interaction between Se 2 and neutral Se atoms on the Se rich side of the system. These two terms are expressed by the following expression : ( 3 5 ) where and are modeling parameters to be determined in the present work using available experimental data Non stoichiometric C ompound Ag 2 Se Ag 2 Se shows an orthorhombic structure with space group of In this structure, two Ag atoms are crystallographically independent and are distributed between the Se layers [141] However, at high temperature s

PAGE 48

48 Ag 2 Se shows bcc structure with space group of In this structure, Se anions are arranged in bcc packing of 2a Wyckoff site and Ag cation s locates partially in three kinds of interstitial positions: the octahedral 6b site, the tetrahedral 12d site, and the triangular 24h site [120, 121] Small homogeneity range was reported in both Ag 2 Se and Ag 2 Se [114, 115, 135] Based on these structure information, Ag 2 Se phase was described by a two sublattice model (Ag, Va) 2/3 (Se,Va) 1/3 and Ag 2 Se phase can be described by the sublattice model based on the experimental ly measured structure [120] (Ag,Va) 6 (Ag,Va) 12 (Ag,Va) 24 (Se) 2 ( 3 6 ) where the four subscripts correspond to Wyckoff sites 6b, 12d, 24h, and 2a of space group respectively. Since there are 44 atom sites in the lattice, it is easier to model the structure by assuming the whole lattice as one atom site. Thus the model can be modified to (Ag,Va) 3/22 (Ag,Va) 3/11 (Ag,Va) 6/11 (Se) 1/22 ( 3 7 ) Based on the results of DFT calculatio n, for the convenience of modeling, the model was further modified as following (Ag,Va) 3/22 (Ag,Va) 2/11 (Ag,Va) 1/11 (Ag,Va) 6/11 (Se) 1/22 ( 3 8 ) The details can be found in section Calculations of Gibbs Energy The Gibbs energy of Ag 2 Se phases are described using the following expression: ( 3 9 )

PAGE 49

49 Each term in Equation 3 9 is expressed using the expressions: ( 3 10 ) ( 3 11 ) ( 3 12 ) The Gibbs energy of Ag2Se phases are described using the similar expression but with five sublattices. The terms in Equation 3 12 are the interaction parameters between the constitutions on the corresponding sublattices and were optimized in this work. Calculation s of Gib bs Energy Experimental data in the iterature combined with DFT based first principles calculations were employed to obtain Gibbs energy of the non stoichiometric Ag 2 Se Ag 2 Se phases. The DFT calculation were conducted by our collaborator, Shun Li Shang and Zi Kui Liu in the department of materials science and e ngineering at t he Pennsylvania State University First principles calculations in the present work were conducted by Vienna Ab initio Simulation Package (VASP 5.4.1) [142, 143] with the electron ion interaction described by the projector augmented wave method [144] The generalized gradient approximation developed by Perdew Burke E rnzerhof (PBE) [145] together with t he van der Waals correction of the D3 method (i.e., PBE+D3) [146 ] were selected to describe the exchange correlation (X C) energy functional. In addition, the improved GGA PBE for densely packed solids and their surfaces, i.e., the PBEsol [147] was also employed, since both PBE+D3 and PBEsol could describe well structural as well as phonon properties of chalcogenides [148]

PAGE 50

50 During VASP calculations, eleven electrons (5 s 1 4 p 10 ) were treated as valence electrons for Ag and six (4 s 2 4 p 4 ) for Se. The employed plane wave cutoff energies, k points meshes, and the smearing methods for Ag 2 Se as well as Ag and Se are shown in Table 3 3 phase), its equilibrium properties ( for example energy, volume, and bulk modulus) were determined by fitting the DFT energy ver sus volume ( E V ) data points according to a four parameter Birch Murnaghan equation of state (EOS) [149] ( 3 13 ) where a 1 a 2 a 3 and a 4 are fitting parameters. Seven or eight data points within the volume range of 10% < ( V V 0 )/ V 0 < 10% were typically used for the present EOS fittings. Other details of DFT calculations are the same as our previous calculations [148] Figure 3 3 shows the relative energies for some low energy structures of Ag 3 Se 2 Ag 4 Se 2 Ag 5 Se 2 in terms of (Ag,Va) 6 (Ag,Va) 12 (Ag,Va) 24 (Se) 2 (see E quation 3 6 ), where the reference states are the lowest energ y structures of Ag 3 Se 2 and Ag 5 Se 2 and the employed X C functional is PBEsol. It is seen that Ag 4 Se 2 A g 2 Se) is more stable with respect to Ag 3 Se 2 and Ag 5 Se 2 agreeing with Oliveria et al. [150] It is found that the lowest energy structure of Ag 4 Se 2 is the configuration with 4 Ag atoms in site 12d (1/3 site occupancy by Ag, not shown here) and this structure is set as the stable structure of Ag 4 Se 2 Ag 2 Se). Therefore, f or the convenience of thermodynamic modeling the second lattice (Ag,Va) 3/11 in E quation 3 7 can be treated as (Ag,Va) 1/11 (Ag,Va) 2/11 and the endmember ( Va) 3/22 ( Va) 2/11 (Ag ) 1/11 ( Va) 6/11 (Se) 1/22 refers to the stoichiometric Ag 2 Se Thus, t he model for Ag 2 Se phase can be modified as shown in E quation 3 8

PAGE 51

51 The second lowest structure of Ag 4 Se 2 is the configuration with one Ag in site 6b, two Ag in site Ag 2 Se occupy all of the sites 6b, 12d, and 24h, but the major one is the site 12d. The present finding agrees well with experimental observation [120, 150] shown in Table 3 4 wh ere the reference states are fcc Ag and Trigonal Se Since the Ag2Se cannot be fully occupied by Ag, otherwise, the distance between some Ag Ag atomic pairs are quite small (<1 ), resulting in unreasonably large energy. DFT calculations fail for some of these structures, for example, the three interstitial sites (6b, 12d, and 24d) are fully occupied by Ag. Ag 6 Se 2 Ag 12 Se 2 A g 24 Se 2 (site 24h fully occupied by Ag), and Ag 18 Se 2 (both sites 6b and 12d fully occupied by Ag), the predicted relative energies are 13.9, 91.73, 20.12, 150.532 kJ/mol atom (see data in Table 3 4 ), respectively. Thermodynamic M odeling and Results Thermodynamic optimization for Ag Se system was conducted using the Pandat Software. Phase diagram and thermodynamics properties were calculated using the Poly 3 module in Thermo Calc Software. The optimization procedure is shown as following. Firstly, the Gibbs energy of stoichiometric Ag 2 Se was derived based on the available e xperimental data. The terminal phases, fcc Ag and Trigonal Se, were treated as unary phases as no mutual solubility was reported. The interaction parameters of liquid phase were then optimized using the phase equilibrium data including liquid phase and the rmodynamic properties of liquid phase. The data published by Ohno et al. [125] and Tsuchiya et al. [126] were not used in this optimization since their phase

PAGE 52

52 boundary data derived from liquid physical properties show lar ge uncertainty and inconsistent with the measured Se activity in liquid phase [151] Secondly, the Gibbs Ag 2 Ag 2 Se phases obtained from DFT calculations was used for the endmembers in the subla Ag 2 Se and Ag 2 Se phases ( Table 3 5 ) All the available experimental data were used together to finalize the parameters in the mo del. The thermodynamic parameters of the Ag Se system modelled in the present work are listed in Table 3 5 Calculated Ag Se phase diagram is presented in Figure 3 2 along with the experimental data [122, 123, 125 127, 152] A good agreement between the experimental data and the calculated results was generally achieved Ag 2 Se in equilibrium with liquid phase around 900 1000 K due to the scattered experimental data in the literature. It lies close to the data of Pellini et al. [123] at the region between 0.35 0.39 at.% Se. With Se content increasing, the calculated phase boundary moves close to the data reported by Pelabon et al. [152] Calculated miscibility gap in Ag rich liquid phase shows a good agreement with the data measured using the sample quenching method [127] The enlarged phase diagram superimposed with experimental data [114, 115, 123, 135, 153, 154] between 33.28 and 33.35 in at.% Se is shown in Figure 3 4 The calculated homogeneity range of Ag 2 Se shows a good agreement with the experimental data measured by coulometric titrations with the solid state galvanic cell Ag/Ag 4 RbI 5 /Ag Se/C [115] exc Ag 2 Se in equilibrium with Se solid phases. It is worth noting that there is a two Ag 2 Ag 2 Se phases during Ag 2 Se phase transition reported by Oehsen et al. [115] As shown in

PAGE 53

53 Figure 3 5 the calculated two phase equilibrium region agrees with experimental data by Oehsen et al [115] The calculated phase diagram differs from the one proposed by Grnvold et al. [135] where the proposed phase boundaries were extrapolated instead of direct measurement. Table 3 6 shows the temperatures and compositions of the experimental and calculated values for all invariant reactions A good agreement was generally achieved. For reaction L3 (Se rich liquid phase as shown in Figure 3 2 ) Ag 2 Se+ Trigonal Se Karakaya et al. [117] reported that this reaction could be a congruent melting of Se. However, based on the phase equilibrium dat a reported by Pellini et al. [123] the temperature that Ag 2 Se +liquid transforms to Ag 2 Se +Se is 490 K, which is lower than the Se melting temperature of 494.2 K [136] This indicates the existence of a eutectic reaction. In this work, the eutectic composition and temperature of liquid were calculated as x(Liquid, Se)=0.9714 and T=489.75 K, respectively, agreeing with the experimental data by Pellini et al. [155] Calculated activities of Se and Ag at 1273 K were compared with experimental data of Blachnik et al. [139] as shown in Figure 3 6 The calculated values fit the experimental data well in the region of Se atomic fraction larger than 0.312, but in the region of Se atomic fraction smaller than 0.312, the calculated values are lower than the experimental data. The discrepancy may due to the relatively low accuracy of the vapor pressure measurements while the vapor pressure is low. The measured and calculated standard enthalpies of formation, standard entropies and standard G ibbs energy of formation for the intermetallic compound Ag 2 Se

PAGE 54

54 agrees well as shown in Table 3 1 The calculated value agrees with all of the experiment al data as the deviation of the experimental data is small. The temperature and Ag 2 Ag 2 Se transformation are calculated as 405.42 K and 7.71 kJ/mol as shown in Table 3 6 and Table 3 2 respectively. The experimentally measured enthalpy of this transformation scatters between 6.06 kJmol 1 and 7.71 kJmol 1 and the values obtained using DSC [114, 130, 133, 137] scatter and deviate significantly from the data measu red by a diabatic methods, which could be attributed to the h ysteresis effect during heating or cooling. The calculated heat capacity is in excellent agreement with the measured ones as shown in Figure 3 7 Summary A self consistent thermodynamic description for the Ag Se system was firstly established in this work. The ionic liquid model and sublattice model were used for the liquid phase and the inte rmetallic compounds Ag 2 Se, respectively. The Gibbs energy of the endmember with stoichiometric composition in Ag 2 Ag 2 Se phases was derived from the experimentally measured thermodynamic properties. The Gibbs free energy of the endmembers off stoi chiometric composition was calculated using first principle calculations and was directly used in the sublattice model. The calculated phase diagram and thermodynamic properties demonstrate a reasonable agreement with the measured data available in the lit erature.

PAGE 55

55 Table 3 1 Summary of experimentally measured thermodynamic properties of Ag 2 Se Compound f G, kJ mol 1 S, J K 1 mol 1 f H, k J mol 1 References Ag 2 Se 49.47 149.99 42.73 [131] 48.9 148.2 42.7 [132] 49.19 150.08 42.41 [130] 49.59 149.2 43.09 [130] 49.02 155.54 40.62 [115] 46.9 149.9 40.1 [136] 49.41 151.03 42.36 [135] 49.24 154.6 41.12 [133] 49.47 149.98 42.73 This Work Ag 2 Se 47.43 169.01 35.02 [131] 47.58 169.44 35.04 [130] 47.48 169.09 35.05 [135] 47.64 169.57 35.06 [133] 47.52 169.38 35.00 [115] 47.87 148.2 42.7 [132] 47.43 169 35.01 This Work Table 3 2 Summary of experimentally measured e nthalpy of transition of Ag 2 Se (kJ mol 1 ) Method References 6.06 DSC [133] 6.82 0.2 Adiabatic [135] 7.37 DSC [130] 7.68 0.1 DSC [114] 6.8 0.14 DSC [137] 7.0 0.6 Adiabatic [138] 7.71 0.55 EMF [131] 7.12 compilation [156] 7.06 compilation [157] 7.71 Evaluated This Work

PAGE 56

56 Table 3 3 Details of DFT calculations for each compound including the space group, total atoms in the supercell, k points mesh, cutoff energy ( E cut in eV), and the smearing method. Compound Space group Atoms k mesh E cut a Smear b Ag 2 Se P 2 1 2 1 2 1 12 1398 250 360 1 5 Ag 2 Se 2 ~ 44 c 121212 250 360 1 5 Ag 1 292929 360 1 5 Se P 3 1 21 3 232316 210 360 1 5 a The energy of 250 (or 210) eV used for structural relaxations and the energy of 360 eV for final calculations. b Methfessel corrections for final calculations to get accurate energy. c Atomic numbers depend on vacancy included in (Ag,Va) 6 (Ag,Va) 12 (Ag,Va) 24 (Se) 2 Table 3 4 Relative energy ( E, kJ/mole atom) with respect to fcc Ag and Trigonal Se from DFT calculations. Material Method E 4a 6b 4a 12d 4a 24h Note Ag 2 Se PBE+D3 9.936 Ag Ag Se Stable phase at low T PBEsol 5.869 Ag Ag Se Ag PBE+D3 0.518 Ag Ag Va PBEsol 0.561 Ag Ag Va Se PBE+D3 22.869 Va Va Se E V data points are scattering PBEsol 21.468 Va Va Se Ag 2 Se PBE+D3 4.127 Va Partial Ag Va The lowest energy structure PBEsol 3.196 Va Partial Ag Va Se PBE+D3 53.402 Va Va Va Bcc Se PBEsol 43.164 Va Va Va Bcc Se Ag 6 Se 2 PBE+D3 13.937 Ag Va Va PBEsol 22.092 Ag Va Va Ag 12 Se 2 PBE+D3 91.738 Va Ag Va PBEsol 97.680 Va Ag Va Ag 24 Se 2 PBE+D3 20.119 Va Va Ag PBEsol 20.653 Va Va Ag Ag 18 Se 2 PBE+D3 150.532 Ag Ag Va PBEsol 149.370 Ag Ag Va

PAGE 57

57 Table 3 5 Op timized parameters for the Ag Se system (J/mol formula as shown in the model ) Phase T (K) Function Reference Liquid 298.15
PAGE 58

58 Table 3 5 Continued Phase T (K) Function Reference Ag 2 Se 298.15
PAGE 59

59 Table 3 6 Temperatures and compositions of the invariant reactions in the Ag Se system Reaction Composition (at.% Se) T( K ) Reaction Type Reference 12/0/31.3 1163 Monotectic [71, 123] 8.2/0/31.2 1163 [127] 11/0/31 1163 [158] 7.6 /0/ 30.7 1163.01 This Work Ag 2 Se 32/0/33.3 1113 Eutectic [122, 123] 30.87 6/0/33.3 1112.53 This Work Ag 2 Se 33.3 1170 Congruent [155] 33.3 1168 [125] 33.3 1167 [129] 33.3 1170.12 This Work Ag 2 Se+L3 44.5/33.3/95.5 889 Monotectic [152, 158] 43.90 /33.3/ 94.02 887.31 This Work L3 Ag 2 Se+Se 100/33.3/ -490 Eutectic [117] 97.14 /33.3/1 00 489.75 This Work Ag 2 Se Ag2Se 33.3 40 5.4 Polymorphic Transformation [131] 33.3 406 [136] 33.3 397.5 [130] 33.3 407.7 [133] 33.3 406 [132] 33.3 405.42 This Work

PAGE 60

60 Figure 3 1 Ag Se phase diagram assessed by Karakaya et al. [117] and summarized by Predel et al. [118]

PAGE 61

61 Figure 3 2 Calculated phase diagram of the Ag Se system in the present work in comparison with experimental data [122, 127, 152, 155]

PAGE 62

62 Figure 3 3 Low relative energies of phase (see Equation 3 13 ) with compositions of Ag 3 Se 2 Ag 4 Se 2 and Ag 5 Se 2 where the reference states are the lowest energies of Ag 3 Se 2 and Ag 5 Se 2 respectively.

PAGE 63

63 Figure 3 4 Calculated phase diagram in the Ag Se system near the Ag 2 Se polymorphic transformation region in comparison with experimental data [114, 115, 135, 153, 154]

PAGE 64

64 Figure 3 5 Calculated phase diagram near the invariant reaction Ag 2 Se Ag+ Ag 2 Se.

PAGE 65

65 Figure 3 6 Calculated activities of Ag and Se at 1273 K in comparison with experimental data [151]

PAGE 66

66 Figure 3 7 Calculated heat capacity of Ag 2 Se versus temper ature in comparison with experimental data [135]

PAGE 67

67 CHAPTER 4 THERMODYNAMIC OPTIMIZATION OF THE AG CU SE SYSTEM Energy is the underlying factor to human economic activity, and more energy is projected to be needed in the near future and photovoltaics provide a means to supply that energy. CIGS based solar cells have been confirmed showing 21% photovol taic (PV) conversion efficiency [159] In order to further increase their efficiencies and minimize their production costs, Ag can be added to substitute part of Cu to form a PV material with larger band gap [65, 66] However, the reaction pathway during the synthesis process of CIGS is complicated and adding Ag into CIGS makes it even more difficult to be optimized. Fully understanding the Ag Cu In Ga Se thermodynami c system and phase equilibrium will be helpful in discovering the most efficient reaction pathway. In this work, the Ag Cu Se system was thermodynamically optimized as a part of the complete database and a good agreement between the calculated thermodynami c properties and experimental data was archived. Ag Cu Se System in Literature Ag Cu System The Ag Cu system was optimized [67, 101, 160] against phase equilibrium data [67] the enthalpy of mixing in liquid [92] In the most recent assessment, as shown in Chapter 2 the measurements of Cu activity in fcc A1 (Ag) phase using electro motive force (EMF) measurement were used. The calculate d solidus of fcc A1 Ag phase are shown close to the data reported by Sommer et al. [89] This set of thermodynamic s database was used directly without any modification for this ternary phase evaluation because firstly, no more new data was reported and also, the EMF measurements

PAGE 68

68 provides more accuracy than other phase equilibrium measurements such as DTA or microanaly sis. Ag Se System The Ag Se system was firstly assessed in this work and the details of this model was presented in Chapter 3 Before this assessment, only Karakaya et al. [117] proposed a phase diagram based on the available phase equilibrium data. T he Ag Se system contains one intermetallic compounds Ag 2 Se with two polymorphs of Ag 2 Se with orthorhombic structure (low temperature) and Ag 2 Se with body centered cubic structure (high temperature), two terminal solid phase FCC Ag and Trigonal Se, and two miscibility gaps in the liquid phase. Several invest igations have been conducted to determine the liquid phase equilibrium. The existence of two miscibility gaps in the liquid phase on was firstly reported by Friedrich [122] and Pellini [123] and was confirmed by Houphouet Boigny et al. [124] The composition of the two miscibility gap was reported between 11 31 at.% Se at 1163 K and 45 95 at.% Se at 8 89 K, respectively. Ohno et al. [125] Tsuchiya et al. [126] and Wiggett et al. [127] measured the phase boundaries of the miscibility gap using conductivity measurement, gamma ray attenuation and sample quenching method, respectively. The results from these three measurements show large discrepancy, except the composition and temperature of the monotectic reaction between the liquid, FCC Ag 2 Se phase. Ohno et al. [125] and Tsuchiya et al. [126] proposed a converged miscibility gap in Ag rich liquid phase, while Wiggett et al. [127] propose d that the boundary of miscibility gap does not converge rapidly.

PAGE 69

69 Friedrich et al. [128] showed small solubility ( less than 0.27 at.% Se ) of Se in FCC A without more detail. Therefore, it is not considered in this assessment. N o solubility of FCC Ag in Trigonal Se was reported. Extensive studies have been conducted for the phase equilibrium and thermodyn amic properties of Ag 2 Ag 2 Se phase Ag 2 Se was measured using thermal analysis [123] DTA [129] and conductivity measurement [125] and the results show good consistency as 1170 K, 1167 K and 1168 K, respectively. However, the measured Ag 2 Ag 2 Se shows large discrepancy Voronin et al. [130] used EMF measurement in solid electrolyte (AgCl) or a glycerol solution with con stant stirring to get the transformation temperature of 397.5 K. However, Echmaeva et al. [131] Nasar et al. [132] and Feng at el. [133] using EMF measurement, reported a higher tr ansformation temperature as 405.5 K, 406 K and 407.7 K, respectively. The transformation temperature of 406 1 K was recently reported by Asadov et al. [134] using microstructu ral and X ray analysis Oehsen et al. reported small deviation from stoichiometry in Ag 2 Se and Ag 2 Se phase, and the homogeneity range was determined using the solid state galvanic cell Ag/Ag 3 RbI 5 /Ag 2+ Se/C between 348 K and 463 K. Grnvold et al. [135] investigated the thermodynamic properties of the transformation region usi ng adiabatic calorimetry and proposed a different phase diagram around the Ag 2 Se and Ag 2 Se phase transformation region. Most of studies regarding the thermodynamic properties of Ag Se system focused on the intermetallic compound Ag 2 Se. The standard Gib bs energy of formation ( ) and the standard enthalpy of formation ( ) was extensively

PAGE 70

70 studied using EMF measurement [115, 130 133] and adiabatic calorim etry [135] Their results shows good consistency, except that the results for Ag 2 Se from Olin et al. [136] shows a slightly lower value (2 kJmol 1 ) for the standard Gibbs energy of formation. The enthalpy of transformation from Ag 2 Se to Ag 2 Se has been measured using DSC [114, 130, 133, 137] and adiabatic calorimetry [135, 138] The results from DSC scatter between 6.06 kJmol 1 and 7.68 kJmol 1 while the value obtained with adiabatic calorimetry show good consistency at ~7 kJmol 1 The experimental data are summarized in Table 3 1 and Table 3 2 Blachnik et al. [139] determined the activity of Se in liquid phase at 1273K by measuring the vapor pressure using modified dew point method. Besides the experimental data ment ioned above, no new experimental data is available. In this model, the intermetallic compound Ag 2 Se was modeled based on it lattice structure, which gives the physical meaning to the extrapolation to the ternary phases. The model was directly used in this ternary assessment without further modification. Cu Se System There are several investigations in terms of the Cu Se system [15, 71, 161] The latest assessment was done by Du et al. [15] However, they used the associate model for their thermodynamic datab a se, which is different from the ionic liquid model used for Ag Cu and Ag Se system Thus, it is not used in this ternary system assessment. Shen et al. [162] reassessed the system in 201 Gu 2 Se 3 was This binary system consists of two terminal phases, FCC A1 Cu and TRIGONAL_A8 Se. Three intermetallic compounds Cu 3 Se 2 CuSe, CuSe 2 were confirmed. The liquid phase of this system exhibits two miscibility gaps. In the In the

PAGE 71

71 most recent assessment [163] the compound CuSe 2 CuSe and Cu 3 Se 2 were treated as line compound. The homogeneity range of both Cu 2 x Se (low temperature phase) and Cu 2 x Se (high temperature phase) were modeled using sublattice model based on their la ttice structure. The calculated thermodynamic properties of these intermetallic compounds agrees with the available experimental data, including t he standard enthalpies of formation and entropies heat capacity etc. To sum up, a set of self consistent ther modynamic parameters describing the Gibbs energy of each individual phase as a function of composition and tempe rature were obtained. The calculated phase diagram and thermo chemical properties including enthalpy, entropy and chemical potentials agree we ll with the experimental values, thus it is used in this ternary assessment without further modification. Ag Cu Se System Phase d iagram Agaev et al. [164] firstly constructed the phase diagram for Ag 2 Se Cu 2 Se system over the temperature range 600 1200 C. The detailed phase diagram was shown in Figure 4 1 In this assessment, the ternary compound AgCuSe was proposed. This AgCuSe at 200 C. At 760 C, AgCuSe shows incongruent melting. There is one eutectic point shown at 730 C and two eutectoid reactions was reported at Ag 2 Se rich and Cu 2 Se rich portion respectively. One thing worth noticing is that, in this proposed phase diagram, both Ag 2 Se and Cu 2 Se shows considerably large solubil ity with each other, but no detailed experimental data was shown. In this assessment, only the liquidus was measured using thermal analysis, which leaves the proposed phase relation in the solid phase questionable.

PAGE 72

72 Another complete different pseudo binary phase diagram was proposed by Schfer et al. [165] as shown in Figure 4 3 The phase boundary was determined using Differential Thermal Analysis. In this assessment, no solubility was found for both terminal phases Ag 2 Se and Cu 2 Se, which was further con firmed by Chen et al. using XRD [166] The intermetallic compound AgCuSe melts at 750 C congruently. Two erected reaction was measured at 728 C and 745 C on Ag 2 Se rich and Cu 2 Se rich side. Compared to Agaev et al. and Asano et al, the phase equilibrium data shown in Schfer et al. are more reliable due to the lack of specific expe rimental detail and missing intermetallic compound in Agaev et al. and Asano et al, respectively. One important feature of Ag Cu Se system is the liquid miscibility gap at high temperature. The monotectic solubility gaps extend across the entire system. As ano et al. [167] determined the tie line in the miscibility gap at 1473 K as shown in Figure 4 2 Besides the phase equilibrium data in the liquid phase, Asano et al. also reported several isopleth diagram. However, the existence of the ternary compound AgCuSe was not shown in any of the phase diagrams. Thus, their isopleth diagram was not considered in this assessment. Ternary c ompound AgCuSe AgCuSe exists as an uncommon mineral eucairite, in nature [168] Although it has been studied by many researche rs [161, 169 174] it was shown in the phase di a gram proposed by three investigators [164, 165, 175] AgCuSe has two differ ent crystal structures: disordered high temperature structure ( phase) and ordered low temperature structure ( phase). The high temperature phase was characterized as cubic structure. In this structure, the Ag and Cu cations randomly distributed over

PAGE 73

73 t etrahedral and trigonal site and the Se anions forming an fcc l attice [176] In terms of the structure of the low temperature phase, Earley et al [168] reported a tetragonal structure but later, Frueh et al. [169] using XRD, reported a pseudo tetragonal orthorhombic structure based on fives subcells with tetragonal unit stacking along one a t ax is. No homogeneity range was reported for this compound. The available experimental data was summarized in The phase transformation temperature has been s tudied by several investigators [161, 169, 176 178] Trots et al. [177] reported the tr ansformation temperature as 464 K and 480 K using DSC cooling and heating, respectively. Frueh et al. [169] did not state specific temperature but the transformation temperature range w as determined between 463.15 K and 468.15 K according to the XRD analysis. The result measured with DSM 2M DSC by Glazov [161] shows that AgCuSe undergoes the phase transition at 484 K, which is close to the results repored by Trots et al. derived from the DSC heating curve. The same method was also used by Chrisafiss et al. [176] and the temperature was determined as 474.9 K. Balapanov et al. [178] reported a relatively higher transition temperature as 603 K using EMF. The reason could be the extra 0.1% Ag added to the sample, which results in the low purity of the sample. Thus, this value was not considered in this assessment. The thermodynamic properties of AgCuSe was incomplete. The only available data was the standard heat of formation 49.95 9.5 kJ/mol and entropy 156 3 J/mol for the AgCuSe. The heat capacity of AgCuSe was also reported in th e range between 298 K and 480 K [161]

PAGE 74

74 Thermodynamic Models Unary Terminal Phases The description of the Gibbs energy of pure Ag Cu and Se was taken from the most recent complication Dinsdale [105] and Change [140] They were modeled by using the follow ing equ ation with different sets of parameters for corresponding temperature intervals: ( 4 1 ) where H SER means enthalpy of standard element reference referring to the entha lpy of the stable state of the pure element at 298.15 K, T is the absolute temperature and A~J are the coefficients. Liquid Phases For the liquid phase, the ionic liquid two sub lattice model was used by the following expression: ( 4 2 ) where and is the Gibbs energy of formation of liquid Ag 2 Se and Cu 2 Se, re s pectively. and are the Gibbs energies of formation per mole of atoms of liquids Ag, Cu and Se, respectively. The value of P and Q are determined using: ( 4 3 ) ( 4 4 )

PAGE 75

7 5 w here the number 2 in Equation 4 3 and 1 in Equation 4 4 is the charge of Se and Ag or Cu ion, respectively. The term refers to the interaction between Ag + and Se 2 on the Ag rich side on the system and indicates the inter action between Ag + Se 2 and Se atoms on the Se rich part of the system. and indicates the interaction between Cu + and Se 2 and indicates the ternary inte raction between Ag + Cu + and Se 2 These terms are expressed by the following expression: ( 4 5 ) where and are the terms which was assessed in this work using available experiment data Compound Phases In the Ag Cu Se system, all compounds exist as selenide. There compounds CuSe, CuSe 2 and Cu 3 Se 2 were treated as line compound [162] The Ag 2 Se and Cu 2 Se shows certain homogeneous range and was modeled using sublattice model described as the model mentioned in Chapter 3 and (Cu,Va) 1 (Se,Va) 1 (Cu) 1 respectively. The ternary compound AgCuSe was modeled using sublattice model (Ag) 1 (Cu) 1 (Se) 1 and no vacancy site was added into the model since no homogeneity rage was reported. The Gibbs energy of AgCuSe was expressed using the following equation: ( 4 6 )

PAGE 76

76 Results and Discussion Optimization Results Thermodynamic optimization for Ag Cu Se system was conducted and the phase diagram and thermodynamics properties are calculated using Parrot and Poly 3 Module in Ther mo Calc Software., respectively [179] The optimization procedure is shown as following. Firstly, the optimized Ag Cu, Ag Se, Cu Se binary thermodynamic database was combined as one database. The ternary compound AgCuSe and the parameters for the liquid phase was optimized against available pseudo binary Ag 2 Se Cu 2 Se experiment data. The standard enthalpy of formation and the standard entropy AgCuSe as a function of temperature (noted as GAGCUSEB). Then, another function a+b*T was AgCuSe. The value of a and b were determined together with the liquid parameter using the transfor mation and melting temperature of AgCuSe and other phase equilibrium data in the liquid phase. Finally, the liquid parameter were optimized against the available ternary liquidus data and tie line of the liquid miscibility gap at 1473.15 K. The thermodynamic parameters of the Ag Cu Se system evaluated in the present work are listed in T able 4 2 The calculated Ag 2 Se Cu 2 Se pseudobinary phase diagram is presented in Figure 4 4 a long with the experimental data [164, 165] A good agreement between the experimental data and the assessment was achieved generally. The liquid phase equilibrium at the Ag 2 Se rich portion, due to the scattered experimental data. The measured and calculated invariant reaction using the opti mized database was listed in Table 4 3

PAGE 77

77 Glazov et al. [161] calculated the standard enthalpy of formation and the AgCuSe as 49.95 1.5 kJ/mol and 156.3 3.0 J/molK, respectively, by using the thermodynamic qua ntities for silver and copper selenides AgCuSe was calculated as 52.8 kJ/mol and the standard entropy was calculated as 151.7 J/mol K The deviation between calculated enthalpy of fo rmation in this work and Glazov et al. can be considered reasonable because other than directly measurement, the value in Glazov et al. was calculated using the thermodynamic quantities of two terminal compounds from other literature, which may increase th e inaccuracy of the results. Good agreement was reached between the standard entropy calculated in this work and AgCuSe agrees with the experimental data as shown in Figure 4 5 One of the most important features of Ag Cu Se system is the extent of the miscibility gap from Ag rich to Cu rich portion of the diagram. The calculated isothermal diagram at 1473.1 K shows good agreement with the measured experimental data by Asano et al. as shown in Figure 4 6 The relatively large deviation shown at the Cu, Se rich corner of the diagram is attributed to the disagreement of the measured miscibility gap boundary of Cu Se binary system and the Ag Cu Se system. An isopleth diagram with 15 wt.% of Se was calculated and shown in Figure 4 7 Generally, a good agreement between calculated and experimental data was reached. However, some disagreement was found at 1100 K between 5 and 20 wt.% Cu. Furthermore, the calculated liquidus at Cu rich part shows higher t emperature than the measured data. The deviation can be explained by the missing of the ternary compound AgCuSe in the

PAGE 78

78 diagram proposed by Asano et al., which will definitely result in the wrong phase relation and equilibrium for this ternary system. Isopl eth Diagram Prediction Several isopleth diagram with different Ag/(Ag+Cu) ratio were calculated using the optimized database in this work. The isopleth diagram is helpful in understanding the possible composition and reaction path way of Ag+Cu salinization process. Figure 4 8 shows the isopleth diagram with Ag/(Ag+Cu)=0.79 and the Se mole fraction ranges from 0 to 0.33. At 800 K, starting from the composition with pure Ag and Cu, the Cu Cu 2 Se (region 11 ). As more Se incorporating in to the sample, the mole fraction of Cu in FCC_A1 phase will keep decreasing and eventually below the maximum solubility of Cu in Ag, leading to the disappearance of the miscibility gap in FCC_A1 phase (region 7). Then, Cu 2 AgCuSe (region 12) until there AgCuSe existing in the sample (region 13). Then the rest Ag will be Ag 2 Se. Other Cu related compounds, such as CuSe, Cu 2 Se 3 and C uSe 2 will not form when the Se composition in the sample is less than 0.33 mole fraction. The order of the reactions are shown as following: ( 4 7 ) The other two isopleth diagram of Ag/(Ag+Cu)=0.5 and Ag/(Ag+Cu)=0.2 shown in Figure 4 9 and Figure 4 10 reveals the same salinization reaction order and one thing worth noticing is the shit of FCC_A1+CU2SE_B two phase region to the higher Se mole fraction position due to the decreasing Ag composition.

PAGE 79

79 The calculated isothermal diag ram of Ag Cu Se system at 873 K is shown in Figure 4 11 According to the diagram, there is a miscibility gap extending from Cu rich to Ag rich region with Se mole fraction between 0.95 and 0.52. At the metal rich region, Cu tends to react with Se before Ag gets selenized and the ternary compound will not Cu 2 Se. Summary A self consistent thermodynamic descript ion for the Ag Cu Se system is presented using the two sublattice ionic liquid model for the liquid phase and sublattice model for intermetallic compound with disordered high temperature structure ( phase) and ordered low temperature structure ( phase). A consistent set of thermodynamic parameters was derived. The calculated phase diagram and thermodynamic properties demonstrate a reasonable agreement with the measured data available in the literature.

PAGE 80

80 T able 4 1 Phase transformation temperature of AgCuSe Transition Temperature (K) Method Reference 464 Cooling Trots [177] 480 Heating Trots [177] 463.15 468.15 XRD Frueh [169] 484 DTA Glazov [161 ] 603 EMF Balapanov [178] 474.920.2 DSC Chrisafiss [176] 467.9 Optimized This Work T able 4 2 Optimized parameters for the Ag Cu Se system (J/mol of formula) Phase T (K) Function Liquid 298.15
PAGE 81

81 Figure 4 1 Pseudobinary phase diagram of Ag 2 Se and Cu 2 Se proposed by Agaev et al.

PAGE 82

82 Figure 4 2 The miscibility gap in liquid phase of Ag Cu Se system at 1473 K proposed by Asano et al.

PAGE 83

83 Figure 4 3 Pseudobinary phase diagram Ag 2 Se Cu 2 Se proposed by Schafer et al.

PAGE 84

84 Figure 4 4 Calculated Ag 2 Se Cu 2 Se pseudo binary phase diagram.

PAGE 85

85 Figure 4 5 Calculated Heal Capacity of AgCuSe along with experimental data. Glazov et al. [161]

PAGE 86

86 Figure 4 6 Isothermal diagram of Ag Cu Se system at 1247.15 K (1200 C) along with experimental data.

PAGE 87

87 Figure 4 7 Calculated Isople th Diagram with 15 wt.% Se superimposed with experimental data.

PAGE 88

88 Figure 4 8 Calculated Isopleth Diagram of Ag/(Ag+Cu)=0.79

PAGE 89

89 Figure 4 9 Calculated Isopleth Diagram of Ag/(Ag+Cu)=0.5

PAGE 90

90 Figure 4 10 Calculated Isopleth Diagram of Ag/(Ag+Cu)=0.7

PAGE 91

91 Figure 4 11 Calculated Isothermal Diagram of Ag Cu Se system at 873 K.

PAGE 92

92 CHAPTER 5 THERMOD YNAMIC OPTIMIZATION OF THE AG GA SE, AG IN SE AND HIGHER ORDER SYSTEM Literature Review Ag Ga Se System Palarnick et al. firstly published a pseudobinary Ag 2 Se Ga 2 Se phase diagram [180] 850 K at Ga 2 Se 3 rich region. Based on this assessment, Feigelson [181] did further study mainly focused on the AgGaSe 2 and the homogeneity range was firstly reported. Mikkelsen [182] did extensive study on full range of this pseudo binary system using differential thermal analysis, x ray diffraction, and crystal growth studies. A new compound Ag 9 GaSe 6 was reported and its homogeneity range was measured. A full summary of the work above can be found in Prince et al. [183] Zhu et al. successfully synthesized high qu ality AgGaSe 2 single crystal using Bridgman Technique and confirmed the solubility of Ag 2 Se in Ga 2 Se 3 [184] The maximum solubility shows the composition AgGa 7 Se 11 Furthermore, the detailed homogeneity range of AgGaSe 2 was measured by Zhu et al. T he results are identical as reported by Badikov et al. [185] The isothermal section at 300 K, vertical sections of AgGaSe 2 Se and Ag 9 GaSe 6 Se were also measured using DTA, XRD, Microhardness meas urements [186] Mller and Neumann et al. published two identical papers [187, 188] reporting the heat capacity of AgGaSe 2 in the range of 300 500 K. Sharma et al. also reported a set of heat capacity data between 0 900 K estimated using ab initio calculations. The data re ported by Sharma et al. are almost double value reported by Mller and Neumann et al. The difference may be caused by the error of the experiment or the

PAGE 93

93 immature calculation method used in their research. Therefore, the accuracy of these two sets of data n eeded to be evaluated. There are only two reports in terms of the thermodynamic properties of AgGaSe 2 and Ag 9 GaSe 6 as listed in Table 5 1 [189] The melting point and the e nthalpy of melting of AgGaSe 2 was measured as 819 844 C and 16.6 kcal/mol using DTA [190] Ag In Se System The only available summary of Ag In Se system was conducted by Schmid et al. [191] The Gibbs free energy of formation for AgIn 5 Se 8 and AgInSe 2 was reported as 850 kJ/mol and 188 kJ/mol [192, 193] However, Aristi et al. [194] reported a comp lete different value as 24225.12 kJ/mol. Later, Hiroaki et al. [190] reported the melting and transformation temperature of AgInSe 2 as 771 C and 654 667 C and the heat of melting and transformation was determined as 11 and 3.87 kcal/mol. The heat capacity of AgInSe 2 within the range of 0 300 K was reported by Kopytov et al. with the density functional theory using the pseudopotential method [195] However, Sharma et al. reported complete different value within the range of 0 800 K [196] Therefore, similar as AgGaSe 2 the accuracy of these two sets of data thus were evaluated in this assessment. The pseudobinary system Ag 2 Se In 2 Se 3 was reported by several researchers. Rigan et al. [197] report ed the homogeneity range for AgInSe 2 the phase boundary for the liquid phase for the first time. Olekseyuk et al. map out the phase diagram for the full pseudo binary system [198] The phase relation of the system at 450 C was reported by Chen et al. [163]

PAGE 94

94 Thermodynamic Models Unary Terminal Phases The description of the Gibbs energy of pure Ag Ga, In and Se was taken from the most rec ent complication Dinsdale [105] and Change [140] They were mode led by using the following equ ation with different sets of parameters for correspondin g temperature intervals: ( 5 1 ) where H SER means enthalpy of standard element reference referring to the enthalpy of the stable state of the pure element at 298.15 K, T is the abs olute temperature and A~J are the coefficients. Liquid Phases For the liquid phase, the ionic liquid two sub lattice model was used by the following expression: ( 5 2 ) where and is the Gibbs energy of formation of liquid Ag 2 Se and Ga 2 Se 3 or In 2 Se 3 respectively. and are the Gibbs energies of form ation per mole of atoms of liquids Ag, Ga, In and Se, respectively. The value of P and Q are determined using: ( 5 3 )

PAGE 95

95 ( 5 4 ) w here the number 2 in Equation 5 3 and 1 in Equation 5 4 is the charge of Se and Ag or Ga(In) ion, respectively. The term refers to the interaction between Ag + and Se 2 on the Ag rich side on the system and indicates the inter action between Ag + Se 2 and Se atoms on the Se rich part of the system. and indicates the interaction between Ga(In) 3 + and Se 2 and indicates the ternary interaction between Ag + Cu + and Se 2 These terms are expressed by the following expression: ( 5 5 ) where and are the terms which was ass essed in this work using available experiment data Compound Phases In the Ag Ga Se and Ag In Se system, all compounds exist as selenide. The Gibbs energy of the stoichiometric compound was expressed using the following equation: ( 5 6 ) The vacancy site was added using sublattice model accordingly. For AgGaSe 2 the model (Ag,Ga) 1 (Ga) 1 (Se) 2 (Se,Va) 1 was used in order to add solubility of Ga 2 Se 3 in to AgGaSe 2 phase. The last sublattice (Se,Va) indicates the Se atoms sitting on the interstitial site. The Ag 9 GaSe 6 was directly treated as line compound since no solubility was reported. AgInSe 2 and AgIn 5 Se 8 was also treated as line compounds due to lack of

PAGE 96

96 thermodynamic data. And their Gibbs energy was estimated based on the available phase diagram and phase relation. Assessment Results Ag Ga Se System The calculated standard Gibbs energy of formation, standard enthalpy of formation and standard entropy of AgGaSe 2 and Ag 9 GaSe 6 are listed in Table 5 1 and a good agreement between the experimental data and calculated value generally. It is worth noting that th ere is relatively large difference between the calculated and experimental value of the standard Gibbs energy of formation because according to the phase relation, the Gibbs energy of Ag 9 GaSe 6 has to meet the following relation in order to be stable: ( 5 7 ) Therefore, the value reported by Babanly et al. does not make Ag 9 GaSe 6 stable at 298.15 K. Thus this value was discarded in this assessment. Also Aresti et al. stated that there could be over estimation in terms of the heat of formation of AgGaSe 2 Therefore, this value was assessed using less weight in this study. The Gibbs free energy of AgGaSe 2 and Ag 9 GaSe 6 as the function of temperature was plot in Fig ure 5 1 Both of the assess Gibbs energy follows the relation with the terminal compounds. And the heat capacity of AgGaSe 2 superimposed with experimental data are s hown in Figure 5 2 A good agreement between calculated value and experimental data is achieved. Figure 5 3 shows the calculated AgGaSe 2 Se isopleth diagram. Babanly et al. [186] does not measure the composition in the liquid miscibility gap region. However, according the caution, two liquid miscibility gap was found in the isopl eth diagram.

PAGE 97

97 Figure 5 4 Shows the calculated pseudobinary Ag 2 Se Ga 2 Se phase diagram. The calculated results aggress with the experimental data well. One thing worth noticing is that Mikkelsen et al. [182] does not report any phase change in terms of the binary Ga 2 Se 3 which was proven by Ider et al. and Zheng et al. [16, 17] Therefore, the two and need to be confirmed by experiment in future. The optimized parameters are listed in Table 5 2 Ag In Se System Due to lack of experimental data, especially the thermodynamics data of two ternary compounds AgInSe 2 and AgIn 5 Se 8 It is impossible to model the whole system under the current experimental data availability. However, based on the available data and phase relation. It is possible for us to make an estimation of the Gibbs energy and the value can be confined in a small range. As shown in Figure 5 5 the calculated AgInSe 2 shows good agreement with Sharma et al. and the data from Kopytov et al. is too low [195, 196] The calculated standard Gibbs energy of formation, standard enthalpy of formation and standard entropy of AgInSe 2 are shown in Table 5 1 The standard enthalpy of formation of AgInSe 2 agrees with the data from Aresti et al but there are 29.97 kJ/mol difference between the experimental and calculated data in term of the sta ndard Gibbs energy of formation [193] According to the phase relation, the Gibbs free energy of the compound in Ag In Se system has the following relation: ( 5 8 ) ( 5 9 ) ( 5 10 )

PAGE 98

98 The value reported from Peraldo et a l. does not fit the relation above and thus was not considered in the assesment. The estimated Gibbs free energy of AgInSe 2 and AgIn 5 Se 8 is shown in Figure 5 6 and the calculated Gibbs energy match the relation mentioned above. The calculated isothermal diagram at 450 C ( Figure 5 7 ) shows good agreement with experimental data from Chen et al. [163] Higher Order System and Practical Application The experimental reported for higher order system is rare. The most important property is the complete solubility of AgInSe 2 an d AgGaSe 2 [199] And the solubility of Ag in CuInSe 2 or Cu in AgInSe 2 has not be en reported. Therefore, the higher order system is a combination of the system previously assessed, including Ag Cu Ga In [10] Cu Ga In Se [13, 14, 16, 17, 162] and the subsystems Ag Cu, Ag Se Ag Cu Se, Ag Ga Se and Ag In Se described in this work. And with the assessed thermodynamic database, the phase relation can be redacted. However, due to the large freedom, it is impossible to directly plot the phase relation of the quinary system unless some of the conditions are restrained. The phase change during the selenization process can be calculated by set the fixed value of temperature, pressure, the ratio between some of the elements in the system. Several different phases were involved during the selenization and these phases named with Greek letters are intruded as following. phase refers to the phase of Ag, Cu, In and Ga metal in HCP structure. phase and _HT phase refer to the metal alloys with the composition of (Ag,Cu) 9 (In,Ga) 4 and (Ag ,Cu) 2 (Ga,In) repectively. phase is the compound (Ag,Cu)(Ga,In)Se 2 that were desired for PV material. Phase is described as comprised of 3 compounds, Cu 2 ln 4 Se 7 Cu 8 ln 18 Se 32 and Cu 7 ln 19 Se 32 but is considered to be a defect chalcopyrite structure of

PAGE 99

99 sol id solutions within a composition ra nge bounded by the compounds Culn Se 2 and Culn 5 Se 8 [200] An example for the calculation was performed for T=450 C, metal group I/III=1, Ga/III=0.25 and the value for Ag/I was set at 1, 0.75 and 0.25. Figure 5 8 shows the salinization process when Ag/I=1. phase does not appear until x(se) is over 0.05 and Se first tend s to react with the liquid alloy forming phase and GaSe. InSe and FCC metal alloy phase appears during the salinization process but eventually they will react with the Se, forming phase A surge of phase fraction can be observed with x(se)>0.35, due to the reaction: ( 5 11 ) Once x(se) reach 0.5, the only phase that exists is AgIn 0.75 Ga 0.25 Se 2 and further Se incorporation will not have at even higher Se composition. Another calculation was performed on the same condition but Ag/I ratio was changed to Cu rich value 0.25 and the result is shown in Figure 5 9 Similar as the results in Figure 5 8 at the initial state, the liquid metal phase is unavoidable. However, phase phase coexists with the liquid phase in the sample. These two phases can directly react with Se, f orming phase _HT phase appears at x(Se)=0.1 but eventually will be selenized completely at x(Se)=0.32. One thing worth noticing is although at x(Se)=0.5, it is possible to get single phase, it is possible for further selenization to take place, formi ng phase and Ag 9 GaSe 6 Therefore, when selenizing Cu rich ACIG alloy, it is very important to control the Se incorporation quantitatively in order to avoid the formation of undesired phases.

PAGE 100

100 The third calculation was performed for the same condition exce pt the Ag/I was set as 0.25 which indicates the Cu poor condition ( Figure 5 10 ). Different from the case with no Cu, at the initial state the alloy contains which a llows the formation of phase at the beginning of the selenization. GaSe is not observed in the selenization process. reached at x(Se)=0.5 with the composition of Ag 0.75 Cu 0.25 I n 0.75 Ga 0.25 Se 2 No further selenization happens with increasing Se composition. Based on the calculation results, the existence of Cu aids the formation of formation of GaSe compounds at the beginning of selenization. However, for the alloy with Cu rich composition, the selenization process needs to be carefully controlled in 9 GaSe 6 defect phase, which may reduce the film quality.

PAGE 101

101 Table 5 1 Thermodynamic properties of the ternary compounds in Ag Ga Se system Compound Reference kJ/mole JK 1 mole 1 AgGaSe 2 2373.4 2395.6 159.611.2 [189] 238.733 239.836 164.12 This Work 446.31 75.36 [194] Ag 9 GaSe 6 4334.1 413.110.9 742.932.5 [189] 594.965 413.104 742.88 This Work AgInSe 2 242 25.12 [194] 188 [193] 219.97 241.8 94.56 This Work

PAGE 102

102 Table 5 2 Optimized parameters of Ag Ga Se and Ag In Se system Phase T (K) Parameters Reference AgGaSe 2 298.15 1500 This Work Ag 9 GaSe 6 298.15 1500 AgInSe 2 298.15 940 940 6000 AgIn 5 Se 8 298.15 6000 AGGASE2 298.15 1500 In 2 Se 3 298.15 6000 [201] GHSERSE 298.15 760 [140] 760 1500 GHSERAG 298.15 1234.93 [105] 1234.93 3000 GHSERGA 298.15 302.9

PAGE 103

103 Table 5 2 Continued Phase T (K) Parameters Reference GHSERGA 302.9 4000 [105] Liquid 298.15 6000 This Work

PAGE 104

104 Fig ure 5 1 Gibbs free energy of AgGaSe 2 Ag 9 GaSe 6 and the two terminal compounds Ag 2 Se and Ga 2 Se 3

PAGE 105

105 Figure 5 2 Heat Capacity of AgGaSe 2 with experimental data. Sharma et al. [202] and Neumann et al. [187, 188]

PAGE 106

106 Figure 5 3 AgGaSe2 Se cross section superimposed with experimental data from Babanly et al. [186]

PAGE 107

107 Figure 5 4 Calculated pseudobinary system of Ag 2 Se Ga 2 Se 3 with experimental d ata from Mikkelsen et al. [182]

PAGE 108

108 Figure 5 5 Heat Capacity of AgInSe 2 superimposed with experimental data Sharma et al. [196] and Kopytov et al. [195]

PAGE 109

109 Figure 5 6 Gibbs energy of AgInSe 2 as the function of temperature.

PAGE 110

110 Figure 5 7 Isothermal diagram of Ag In Se system at 450 C superimposed from the experimental data from Chen et al. [163]

PAGE 111

111 Figure 5 8 Phase fraction change with the increasing Se composition. Ag/I=1, I/III=1, Ga/III=0.25. T=450 C.

PAGE 112

112 Figure 5 9 Phase fraction change with the increasing Se composition. Ag/I=0. 2 5, I/III=1, Ga/III=0.25. T=450 C.

PAGE 113

113 Figure 5 10 Phase fraction change with the increasing Se composition. Ag/I=0. 7 5, I/III=1, Ga/III=0. 2 5. T=450 C

PAGE 114

114 CHAPTER 6 THE EFFECT OF RAPID THERMAL ANNEALING ON THE PERFORMANCE OF CIGS CELLS WITH AN ITO LA YER Literature Review CIGS thin film cells are an attractive choice for use in photovoltaic applications because of their higher power efficiency and potential lower production cost [203] CIGS cells use transparent conducting oxides such as Al:ZnO (AZO) and indium tin oxide (ITO) as the window laye r. AZO is often preferred due to its low er cost but reaction with H 2 O has raised reliability concerns [204] Although less expensive AZO is generally selected as the transparent w indow layer for commercial CIGS production ITO is still considered a better window layer material for its outstanding electrical and optical properties (e.g., low resistivity ( ~ 210 4 and high transparency (>80%) [205] wide band gap energy (3.5 4.3 eV) and long term stability at high temperature in humid environment s [206] ) Several investigations have been conducted on diffusion across the CdS/CIGS interface [207, 208] Krishnan et al. showed that the solid solution CuCd 2 (Ga x In 1 x )Se 4 reaction product appears when the sample temperature is elevated to ~400 C [5] Study of the degradation process on complete d device and the interaction between ITO and CdS after annealing are, however, less studied. Martel et al. showed that during the deposition of CdS on ITO using e beam evaporati on, S and In diffuses into ITO and CdS layer, respectively [209] Furthermore, the diffusion rate for each element wa s determined as 10 17 c m 2 /s [210] at T= 150 C This study reports on degradation of a complete d CIGS device as a function of anneal temperature and time with ITO as the window layer. The change in electrical and optical properties after annealing in dry N 2 and at 85% RH are also reported.

PAGE 115

115 Experimental Method The cells used in this study were prepared by co evaporation of CIGS on a Mo coated stainless steel substrate, followed by chemical bath deposition of the CdS buffer layer and then the sputtered ITO (SS/Mo/CIGS/CdS/ITO). These structures were fabricated by Global Solar Energy, Inc. and produced in a single run. A series of cells for testing were completed at the University of Florida by depositing an Ag contact layer (300 nm) on the ITO using electron beam ev aporation with a shadow mask. Four c ells were sc ribed, isolated, and characterized on a single larger area of SS/Mo/CIGS/CdS/ITO to establish pre measurement values of device parameters. The area of each cell was ~0.409 cm 2 The rapid thermal annealing (RTA) system consisted of a quartz tube equipped w ith heating lamps above and below the sample holder ( Figure 6 1 ). Cells were annealed for 5 10, 15 and 20 min at each temperature (50 100, 150 and 200 C ). It is not ed that the heating time from 25 to 200 C was ~30s. Flowing d ry nitrogen was used a s the gas ambient during annealing in one series of studies In an other set of experiments, samples were exposed to a flowing N 2 /H 2 O gas mixture ( 85% RH ) at 85 C. The rel ative humidity was controlled by mixing a water saturated N 2 flow with a dry N 2 flow, which were maintained at 70 and 12.5 sccm, respectively. The device parameter s were again measured after annealing the device. In a final set of experiments, aluminum foil was wrapped around the outside of the sample boat to block direct exposure to the heater lamp to block direct adsorption. These samples were annealed under dry N 2 flow and the device parameters measured to check for ligh t soaking effects

PAGE 116

116 Each cell was characterized before and after annealing with J V and QE measurements. The measurement temperature was controlled at 25 C during the J V measurements and the illumination area was maintained at the 1 sun AM 1.5 condition. After the J V measurement, these samples were annealed under the conditions previously described above and J V traces were immediately collected. QE measurements were also taken at Oak Ridge National Laboratory one day after the annealing. Results a nd Dis cussion Four cells were simultaneously annealed at each annealed condition. J V and QE measurements were taken for each cell before and after annealing. T he mean value and standard deviation of difference between the after and before anneal values were co mputed and the differences were plotted The mean and standard deviation values for all the non annealed cells are V OC SC = 35.269 mA/cm 2 there is a relatively large variation in performance from cell to cell and thus statistics on the differences were analyzed. The measured open circuit voltage (V OC ) difference at each anneal temperature is shown in Figure 6 2 as a function of anneal time Cells annealed at lower thermal budget showed an increase in V OC. A nnealing the device at higher thermal budget, however, significantly reduced the value of V OC T he largest inc rease in mean value of V OC ( 9.266.07 mV ) occurred after annealing at 50 C for 300, although similar increases were observed in individual cells at higher anneal temperature and time Loss in the mean V OC as a function of annealing temperature at constant anneal time ( not shown) reveals a decreasing trend (note OC scale differences in Figure 6 2 ). As

PAGE 117

117 expected, higher temperature promotes a faster degrad ation process. The results at the two highest temperatures show a decreasing V OC for the first 3 anneal times followed by a slight improvement when annealed for 900s, suggesting a plateau was reached E xperiments for longer annealing time are needed to better defined the trend Figure 6 3 shows the value change of short circuit current density as a function of annealing time at the 4 anneal temperatures. At all meas urement temperature s the mean value of J SC increases when the annealing time was less than 900s except when annealed at 200 C for 900s P revious studies on the influence of temperature on ITO properties indicate that ITO shows better crystallinity hi gher carrier concentration and higher roughness due to the rearrangement of atoms after annealing [211] The larger grain size can decrease the boundary scattering, and thus increase the carrier mobility [212] Ther efore, t he improvement in J SC and FF can be explained by improved crystallinity of the ITO decreasing grain boundary resistance, and smoothing of the interface; hence enhanced collection efficiency. At higher temperature the transmittance begins to decrea se due higher carrier concentration and associated free carrier adsorption. The difference between the annealed and control mean values of the fill factor (FF) and efficiency are plotted in Figure 6 4 and Figure 6 5 respectively. The results point to a decrease in FF with annealing, although a slight increase in the mean value of FF was observed at 3 anneal temperatures for short time (~1% change) The efficiency of each sample was calculated based on the measured V OC J SC and FF As shown in Figure 6 5 a positive efficiency change upon annealing is observed for both the 300 and 600s anneals at 50 and 10 0 C as well at 150 C 600s. The maximum

PAGE 118

118 improvement occurred at the lowest thermal budget studied ( 50 C for annealing time of 300s ) A lso, the magnitude of increment of efficiency show ed a dependence on the annealing temperature. A high thermal budget clearly lowered the mean cell efficiency. The r esults of dark annealing study ( Table 6 1 ) show that the dark annealing are values similar to those observed after the normal annealing condition, and thus the effect of light soaking is insignificant. Since the annealing time is rel atively short in this work, it is unlikely the absorber layer transitioned to the metastable state associated with light soaking to produce an increased efficiency [213] When the an nealing time exceeds 600s the electrical properties of the cell begin to deteriorate Kumar et al. [214] also found si gnificant electrical property decrease when the complete CIS cell was annealed at 164 345 C It has been suggested that diffusion of Cd +2 into CIGS and possibly Cu +1 into the CdS buffer layer [215] The Cd +2 would act as a donor in the upper layer of the CIGS, potentially leading to type inversion and moving the p n junction below the metallurgical one. If defects at the CIGS/CdS interface provide a major recombination pathway, the interdiffusion would have a positive impact. U sing energy dispersive x ray spectroscopy Nakada et al. [207, 208] confirmed that Cd +2 substitute s on Cu site s at temperature as low as 80 C Krishnan et al. recently used in situ X ray diffraction (XRD) to study the degradation mechanism of the CIGS device struc tures from the same source as this study. I n both N 2 and forming gas ambient, the collected diffraction patterns revealed formation of the solid solution CuCd 2 (Ga x In 1 x )Se 4 at ~400 C [5] This temperature is much higher than this study, but a significant amount of CuCd 2 (Ga x In 1 x )Se 4 is needed to be observed by XRD. It is possible that there a trade off between improvement at low thermal budget by CIGS

PAGE 119

119 type inversion and degradation by CuCd 2 (Ga x In 1 x )Se 4 formation may contribute to the observed results. To better understand the effect of annealing on cell performance, quantum efficiency measurements we re performed on selected cells. Figure 6 6 shows the quantum efficiency of a cell annealed at 50 C for 300 s The improvement of QE induced by annealing agrees with the efficiency measurement. The positive effect can be observed in a large range of wavelength (350 1020 nm). The increase in the blue wavelength response can be attributed to the reductio n of front surface recombination due to CdS/CIGS mixing at the interface. The most significant benefit appeared in the green responds region (500 950 nm), possibly due to increased transmittance of the ITO layer [212] and a higher electron mobility produced by the better crystallinity of both the ITO and CIGS layers. The QE curve of a cell annealed at 100 C for 600 s is shown in Figure 6 7 Compared with Figure 6 6 the benefits in the blue responds region have largely disappeared, which may indicate that formation of CuCd 2 (Ga x In 1 x )Se 4 or CIGS/CdS interface defects that act as a recombination center. The decrease in efficiency in the green region may also be attributed to increasing defect concentration, which decreases the diffusion length of the electrons. As shown in Figure 6 7 the increase of QE in the green region is about half the amount shown in Figure 6 6 This may also be caused by carrier concentration or structural change in the ITO induced by annealing. A previous report showed that higher carrier concentration in ITO layer can lead to lower transmittance [212]

PAGE 120

120 At an annealing temperature of 100 C and 9 00s, there is minimal improvement of QE in the green region while the QE is in blue region reduced slightly ( Figure 6 8 ). When the annealing temperature was increased to 150 C however, the QE in the green region decreased more than that in the blue region ( Figure 6 9 ). From Figure 6 8 and Figure 6 9 it is seen that RTA on CIGS cells will first decrease the response of the cell to high energy photons and then lower the cell performance by reducing its response in the green region. Since absorption coe fficient is larger for high energy photons than low energy photons, it is possible that at low thermal budget, the ITO and absorber layer crystallinity is improved and the CIGS/CdS interface contains fewer defects. Both of these effects will benefit the pe rformance of the cell. When the anneal conditions are more aggressive, it is possible that structural changes take place in ITO layer as well as at the ITO/CdS interface, which will decrease transmittance and electrical performance. Thus, the QE of the sa mple in green response region decreases significantly. The electrical properties of samples annealed at 85 C with relative humidity 85% are shown in Figure 6 10 Compared to the results obtained from dry annealing at 50 and 100 C the change of V OC J SC and efficiency reach ed a maximum at 300s and decreased thereafter. The fill factor however, show ed the largest decrease (1.80.3%) at 300s. The devices start ed to fail between 300s and 600s, which is a shorter time range than annealing with dry N 2 flow This suggests the acceleration of degradation caused by water vapor. ITO is typically deposited by RF magnetron sputtering. It has been shown that when ITO is sputtered at temperatures less than 200 C CIGS performance increases

PAGE 121

121 because of decrease in ITO resistance. If the temperature exceeds 200 C however, the interfacial reaction at CIGS/CdS incr eases and the dev ice degradation become apparent [216, 217] The increased conductivity of ITO is due to increased carrier concentration rather than high er mobility. The observed low mobility of ITO compared to bulk In 2 O 3 can be explained in terms of ionized impurit ies or grain boundaries scattering. Enhanced crystallization of films has been observed for films deposited at higher substrate temperature which was shown to increase the mobility. Using TEM and electron diffraction Sree nivas et al. observed large stacki ng faults of ITO films deposited on glass substrate s [33] The high transmission of the ITO is related to the wide band gap of the semiconductor. The absorption edge of ITO films with higher free carrier concentration shifts to shorter wavelengths due to the Moss Burstein shift. R.X. Wang et al. [218] show ed that air annealing at temperatur e less than 200 C increase s the transmittance of ITO films because oxygen incorporation and decomposition reactions substantially influence the carrier concentration and chemical states. Furthermore, it was found that the roughness of films does not have a close relationship with t heir transmittance. Previous reports indicate the ITO/CdS structure shows much lower sheet resistivity and a wider light transmission range than a single CdS layer. Furthermore, p ost annealing of ITO can also change its properties [32, 219] Annealing i n vacuum or reducing atmosphere improves the overall properties of thick ITO. Barlow et al. [32] showed that an nealing at low temperature even at 50 C for 5 minutes, increases the 1 cm 1 This study also showed that the

PAGE 122

122 optical transmission of the ITO films increase d after annealing under oxygen poor condition. These improvements of ITO film properties may contribute to the overall performance of the solar cell. Further characterization ( e.g. TEM) of the sample is necessary to understand the structural change mechanis m in ITO layer as well as the ITO/CdS interface Summary CIGS devices provided by Global Solar Energy, Inc. were annealed using rapid thermal annealing. The efficiency of the devices was found to increase by 1.8 0.8% after annealed at 50 C for 300s. W hen annealed at even higher thermal budget ( 100 C for 600 s ) the efficiency of the devices increased by 2.0 2.3%. This could be attributed to improvement of ITO properties. As the temperature and annealing time was further increased the performance of th e devices declined sharply This degradation process also could be accelerated by annealing in a humid environment. The results of dark annealing shows that annealing using RTA for no more than 600s will not induce a significant light soaking effect. Quant um e fficiency measurement indicates that the during the annealing process, the front area o f the absorber layer degrades at a relatively lower temperature and ITO related degradation takes place under more aggressive annealing condition s Further characte rization s such as TEM and SIMS depth profile measurement are necessary to better understand the underlying reasons for the changes in device performance induced by annealing.

PAGE 123

123 Table 6 1 Summary of Change of Device Properties after Dark Annealing Temperature Time Mean Difference in Value After and Before Annealing C s V OC (V) I SC (A) J SC (mA/cm 2 ) Fill Factor Efficiency 52.8 300 0.0030 0.0014 0.0011 0.0001 2.4709 0.2589 0.1003 0.0811 0.0324 0.0174 103.3 600 0.0054 0.0128 0.0014 0.0003 3.3508 0.6331 0.0737 0.0749 0.0278 0.0170

PAGE 124

124 Figure 6 1 Schematic of the RTA system Figure 6 2 Change in V OC OC scale varies with anneal temperature

PAGE 125

125 Figure 6 3 Change in J SC with annealing time at 4 temperatures

PAGE 126

126 Figure 6 4 Change of Fill Factor with annealing time at 4 temperatures. Note the

PAGE 127

127 Figure 6 5 Change of efficiency with annealing time at 4 temperatures

PAGE 128

128 Figure 6 6 Quantum Efficiency of sample annealed at 50 C for 300 s

PAGE 129

129 Figure 6 7 Quantum Efficiency of sample annealed under 100 C for 600 seconds

PAGE 130

130 Figure 6 8 Quantum Efficiency of sample annealed under 100 C for 9 00 seconds

PAGE 131

131 Figure 6 9 Quantum Efficiency of sample annealed under 150 C for 900 secon ds

PAGE 132

132 Figure 6 10 Electrical properties of samples annealed at 85 C with RH 85%

PAGE 133

133 CHAPTER 7 ZNO X S 1 X THIN FILM AS ALTERNA TIVE BUFFER LAYER MA TERIAL USING ATOMINC LAYER DEOPSI TION Literature Review CIGS thin film cells are an attractive choice for use in photovoltaic applications because of their higher power efficiency and potential lower production cost [203] The substitution of Ag for Cu in CIGS absorber layers shifts the bandgap energy higher and towards the optimal value (1.4 eV) while decreasing materials cost. With this substitution and higher bandgap energy, an alternative buffer layer material to CdS is needed to retain a favorable band alignment with the absorber For CdS buffer layer, it has been shown that an electrically resistive i ZnO layer is beneficial for reducing the shunts effect [18] However, the extra layer not only reduce the photon transmission but also increasing the manufacturing cost. Therefore, eliminating the i ZnO layer would simplifying the fabrication process and beneficial for the cost control. Extensive studie s have been done on the alternative Cd free buffer layer materials [19 24] Zinc Oxysulfide (ZnO 1 x S x ) is a candidate buffer layer material since its electrical and optical properties are tunable by varying the oxygen to sulfu r ratio in the film. Furthermore, the wider bandgap of ZnO 1 x S x compared to CdS and the thinner film possible with conformal ALD compared with conventional chemical bath deposited CdS promises higher transmission especially in high wavelength range. It ha s been applied on various solar materials as a buffer layer using different deposition technique [22, 25, 26] Among all these deposition technique, atomic layer deposition shows significant advantage due to its strong capability of thickness, doping concentr ation and film conformality control. Furthermore, ALD usually requires low deposition temperature

PAGE 134

134 and generate less waste. This would minimize the degradation of the cell performance induced by the thermal diff usion [27] In this study, the ZnO 1 x S x thin film with various H 2 O:H 2 S pulse ratio was prepared on Si (111), borosilicate glass and CIGS/Mo/Stainless Steel substrate. The morphological, structural and optical propertie s of the film were characterized. The complete cell performance was measured and the effect of the thickness of the buffer layer was discussed. Experimental Procedures The CIGS substrate used in this study was manufactured by Global Solar Energy (GSE) usin g co evaporation on Mo/stainless steel sheet. The ZnO x S 1 x buffer layer was deposited on Si and Borosillicate Glass for ALD temperature window determination and film characterization, using an atomic layer deposition reactor with reactants diethylzinc (DEZn) H 2 S, and H 2 O at 150 C. The reactant sequence for depositing ZnO 1 x S x film was set as ( ( DEZn purge H 2 O purge) x ( DEZn purge H 2 S purge) y ) n where x and y are number of times each Zn O and Zn S cycle are respectively executed. The number n indicates the total number of supercycle. To fabricate test cells, the buffer layer was deposited on CIGS substrate and then a layer of aluminum doped zinc oxide (AZO) was sputtered on the buffer layer, followed by deposition of an Ni/ Al contact grid using e beam ev aporation. The buffer layer thickness is precisely controlled by the number of deposition cycles. The deposited films were characterized using XPS, SEM, reflectometer and the complete cell were characterized using solar simulator under AM 1.5 one sun condi tion. As shown in the Figure 7 1 The ALD reactor on the left permits up to five gas or liquid reaction precursor or purge flows to be delivered to the deposition cha mber. T he

PAGE 135

135 top line is devoted to purge gas delivery and the adjacent line is equipped to deliver H 2 S delivery line. The next three lines, shown from top to bottom, are deployed to deliver diethylzinc (DEZn), biscyclopentadienyl magnesium (MgCp 2 ) and H 2 O re spectively. The flow rate of each line is regulated by a mass flow controller, which downstream of a gas purifier. The ALD deposition cycles are controlled by pneumatic valves located just prior to the entrance of the chamber. The switching of the pneumat ic valve is automatically controlled by Labview. The right chamber of the system is designed for ZnO based TCO material ALD deposition. Thus, connected with a transfer chamber, the whole system can be used for continuous ALD deposition of buffer layer and TCO without exposure to the atmosphere. The heater was made of Inconel 600 alloy and t he top heater surface was coated with a protective A l N film. A 3d view plot of the substrate carrier is shown in Figure 7 2 Quartz was selected as the material for the substrate carrier due to its high heat capacity and low cover s the entire heating surface and forms a good thermal contact. The substrate will be placed on the top surface. A customized fork can clear the lower diameter section to allow transfer of the carrier from one chamber to the other. The surface temperature will be calibrated before using this substrate carrier. Experiment Results and Discussion ALD Process Windows The ALD process window was identified by conducting a series of measurements of film growth rate vs. temperature, pulse time for the Zn, O, or S p recursors, and purge time to determine conditions that yielded self limiting reaction [220] During deposition, the carrier gas flow rate was maintained at 60 sccm and the pressure and bubbler

PAGE 136

136 temperature were controlled at 1 atm and 10 C, respectively. The base con dition for ALD fixed the DEZn and H 2 O pulse time at 0.2 and 0.1s, respectively with all purge times at 10 s. The base deposition temperature was 130 C. These settings were then individually varied about the base condition to determine the window for that process variable and base condition. As examples, Figure 7 3 shows the growth rate variation ( /cycle ) for 4 of the process variables ( H 2 O purge time, DEZn and H 2 S pu lse time, and temperature). As shown in Figure 7 3 A, t he ZnO growth rate decreased linearly from 2.4 to 2.05 /cycle for H 2 O purge variation in the time range 2 to 20 s and then was independent of purge time between 20 and 30 s ( 2. 0 5 /cycle ), which is consistent with ALD growth mode Figure 7 3 B shows a small ALD window for DE Zn pulse time (0.17 to 0.24s) for which the growth rate remained constant. The DEZn pulse time is an important parameter for ALD. With longer DEZn pulse, the growth rate increases and then tends to reach a constant, limited by background adsorbed H 2 O adsor ption during a given cycle. Figure 7 3 D shows the film growth per cycle as the temperature is changed for both growth of ZnO and ZnS. It is observed for ZnO (red sym bols) that the growth rate increases linearly from 0.5 to 2.1 /cycle with increasing temperature in the range 60 to 12 0 C This behavior is consistent with reaction limited growth conditions, presumably due to partial decomposition of DEZn on the surface T he growth rate reaches a plateau in the temperature range 120 to 200 C which is the temperature range for ALD ZnO film deposition. The data in Figure 7 3 C indicates that the H 2 S saturation window between 0.09 and 0.12s H 2 S pulse time is reached rapidly. When the pulse time is longer than 0.12s, the grow th rate increases linearly, which indic ates growth is in the continuous chemical

PAGE 137

137 vapor deposition mode The temperature variation of ZnS growth is provided in Figure 7 3 D (blue symbols). The growth rate de creases exponentially in the 60 to 100 C range. This is likely attributable to the condensation of DEZn on the surface at this relatively low temperature. B ased on the data for ZnS deposition in Figure 7 3 D however, a true plateau is not observed and the growth rate decreases slowly with temperature increase above 12 0 C. This is consistent with the observation by Bakke et al. [221] using in situ generated H 2 S. Stuyven et al. [222] attributed this decrease to decreasing number of reactive sites cause by the desorption of ethyl ligands from the film surface at high temperature. Film Composition X ray Photoelectron Spectroscopy (XPS) was used to study the composition of the deposited film. Before acquiring an XPS spectrum, the film was sputter with Ar+ to remove the surface contamination. The Ar sputtering gun was attached to the XPS ultra high vacuum chamber with a base pressure of 7x10 9 Torr. During sputtering, the Ar gun pressure was set at 20 mTorr and the chamber pressure was increased to 3x10 7 Torr. The sputtering time (5 min) was de termined by previous experience to remove the surface carbon. The XPS scan results of the ALD deposited ZnO film are shown in Figure 7 4 A The film was deposited at 1 25 C with 250 cycles. The spectrum was aligned using Oxygen 1s (532.8 eV) peak as a reference. Before initiating the XPS survey, 10 min Ar+ sputtering was performed to remove the adventitious surface carbon and oxygen. According to the XPS spectrum, no i mpurities are found in the deposited film within the detection limit of XPS (0.1 atm.%). In other words, no carbon atoms were induced into the ZnO ALD deposited film from DEZn precursor. Currently, the existence of H atoms in the film possibly induced by H 2 O remains unknown due to the limit of

PAGE 138

138 XPS. It is necessary to conduct further characterization (e.g. UPS) to confirm the film is free of hydrogen. Figure 7 4 B shows the XPS spectrum of ZnS thin film. It is worth noticing that there are around 5% Oxygen residue in the ZnS. It can be attribute to the oxygen and moisture residue in the deposition chamber as the base pressure of the chamber was kept at 0.17 Torr. A lso, due the low deposition thickness per cycle, it is easier for the oxygen incorporation to occur. The XPS spectra collected for samples with different cycle ratio of ZnO to ZnS are shown in Figure 7 5 It is noted that the Zn, O and S peaks are located at 1022 eV, 533.2 eV and 162.1 eV, respectively. No carbon peak was found within the XPS detection limit, which indicates that during the deposition the surface dieth ylznic was oxidized cleanly by H 2 O or H 2 S. As shown in Figure 7 5 the sulfur peak intensity decreases with the decreasing of the ratio of ZnS to ZnO cycle and consist ently, the oxygen peak becomes stronger with increasing the number of ZnO cycles. The intensity of Zn peaks does not change significantly, which means Zn concentration in the film does not change with the variation of the ratio of ZnO to ZnS cycles. The ac tual S/(S+O) value calculated from the XPS spectra are plotted as a function of the pulse ratio in Figure 7 6 The 45 line in this figure represents the condition wh ere the film composition matches the overall exposure composition. Difference from the pulse ratio, the concentration of sulfur in the deposited film deviated negatively from the precursor pulse ratio. When 90% H 2 S pulse was used, the sulfur concentration in the film dropped below 50%. And when H 2 S pulse ratio was below 0.3, the concentration of S in the film is even less than 5%. This negative deviation of sulfur concentration indicates the substitute reaction

PAGE 139

139 ( 7 1 ) could take place during the deposition. The Gibbs free energy of the reaction to the right direction was calculated as 78 kJ/mole at room temperature, using Gibbs free energy of formation of ZnO(s), H 2 S (g), ZnS(s) and H 2 O(g) as 318.32, 33.56, 201.29 and 228.59 kJ/mol. This means the reaction to the right direction is thermodynamically favorable [223] However, the ALD process is far from equilibrium because for every pulse only one type of precursor or oxidizer was introduced in to the reaction, which may enhance the reverse reaction. Higher temperature will incre ase the reaction rate as excessive H 2 S was delivered in to the deposition chamber every pulse [26] Platzer Bjrkman et al. reported similar behavior and thus the actual S concentration is highly depend on the deposition process even the parameters were chosen wi thin the ALD window. Multiplex scan were performed for O, S and Zn peaks only as shown in Figure 7 7 It is noted that the Zn, O and S peaks are located at 1022 eV, 5 30.4 eV and 162.1 eV, respectively. The sulfur 2p peak can be spilt into two peaks for 2p3 and 2p1 electron, located at 161.5 eV and 163.1 eV respectively. All these measured binding energy value matches the binding energy value for ZnO and ZnS in XPS bind ing energy database, which indicates all the O and S atoms are bounded to Zn. Morphological and Structural Properties In order to test the effect of the precursor pulse time on the surface feature of the deposited thin film. The AFM measurement was conduct ed on the samples of ZnO thin film using various DEZn and H 2 O pulse time. The AFM images were shown in Figure

PAGE 140

140 7 8 and the change of the surface roughness with precursor pulse time was plot in Figure 7 9 According to the AFM images, the film deposited with ALD windows shows quantitative analysis, the fi lm deposited within ALD window shows minimum surface roughness at 3.5 nm. The surface roughness increases monotonically with the increasing precursor pulse time and reach 17.8 nm for the sample deposited using 2 s and 3.5 s for DEZn and H 2 O pulse time. One induced to the substrate surface. This deposition feature will result in relatively less surface roughness. The topographical SEM i mages of ZnO x S 1 x thin film with different S concentration were shown in Figure 7 10 (H 2 S pulse rich) and Figure 7 11 (H 2 O pulse rich). The two films show distinctive surface feature. The grains of the H 2 S pulse rich ZnO x S 1 x thin film shows round shape feature and the grain size slightly increases with the increasing H 2 S pulse fr action from ~45 nm to ~80 nm. Even when the H 2 S pulse fraction reaches 0.5, the feature remains the same. However, different from the H2S pulse rich ZnO x S 1 x thin film, the H 2 O pulse rich samples show elongated grains on the surface. This results is similar as the feature reported by Sanders et al. [224] And small incorporation of S does not change the grain feature. All three samples shown in the Figure 7 11 shows sim ilar grain size between 80 110 nm.

PAGE 141

141 In order to analysis the structural information, XRD was conducted on samples with various S concentration ZnO x S 1 x thin film and the spectrum was shown in Figure 7 12 As shown in Figure 7 12 A H 2 O pulse rich ZnO x S 1 x thin film shows hexagonal Wurtzite ZnO structure. F our major peaks are observed at 31.86, 34.4873, 36.3324 and 56.7762, corresponding to (100), (002), (101) and (110) plane. The intensity of the peaks for (100), (002) and (101) are much higher. These four peaks indicate the thin film is multicrystallin e All the peaks show the shift to the lower angle with the increasing H 2 S pulse fraction. According to ( 7 2 ) is the scattering angle n is a positive integer and The d spacing value for all these four plane increases as the incorporation of S atoms into the ZnO thin film. On the country, the XRD spectrum of H2S pulse rich ZnO x S 1 x t hin film ( Figure 7 12 B ) shows only one strong peak at 28.65 This indicates that H2S pulse rich ZnO x S 1 x thin film has a cubic structure with preferred orientation of (111) direction. As the O composition in the film increasing, the peak shifts to higher angle, which suggests the decreasing of the d spacing. Although there is only one obvious peak shown in the XRD spectrum, the film is multicrystalline based on the sur face SEM image. The radii of O 2 ion (1.42 ) is less than that of S 2 ion (1.84 ). Therefore, the incorporation of O into ZnS or S into ZnO thin film will cause the decreasing or increasing of the lattice constant, respectively. Figure 7 12 C shows the XRD spectrum of ZnO x S 1 x thin film prepared using H 2 O:H 2 S ratio 1:1 and no peak is observed. This indicates that film is amorphous.

PAGE 142

142 The crystallite size was extracte d from XRD spectra using Scherrer equation The crystallite size of H2S pulse rich ZnO x S 1 x thin film was measured between 40 and 50 nm with relatively small deviation while that of H 2 O pulse rich ZnO x S 1 x thin film firstly increases from 28 nm to 80 nm fr om with the decreasing H 2 O pulse fraction and then drops back to 50 nm. All the data on the H 2 O pulse rich side shows large deviation due to the different growth direction of the grains in the film. Optical Properties Transmission of ZnO x S 1 x thin film dep osited with various H 2 S pulse fraction was plotted in Figure 7 14 It is worth noting that the thickness of the film was controlled at 50 nm but with slightly variation because the total thickness was controlled by the number of supercycles. And the growth per supercycle was different for films with different H 2 S pulse ratio. This could be one of the reason for the variation of the transmission for these thin fi lm. The red shift effect can be observed for the incorporation of both S and O. To derive the optical bandgap, further analysis was performed as following. The adsorption can be calculated using ( 7 3 ) where A is absorption with arbitrary unit and T is the transmission. The band gap of the samples were exacted from absorption spectrum using Absorption Spectrum Fitting (ASF) method proposed by Ghobadi et al. [225] The absorption coefficient can be described as a function of wavelength using the equation below, ( 7 4 )

PAGE 143

143 where g constant, and velocity of light, respectively. B is constant and m is the index which can have different values of 1/2, 3/2, 2, and 3 [226] It has been approved that the best fitting can be achieved for m=1/2 [225] and using the Beer Equ ation 7 4 becomes ( 7 5 ) The bandgap can be calculated using the date obtained by extrapolating the line to =0 and the equation as following ( 7 6 ) The derived optical bandgap for both materials was plot in Figure 7 16 The bandgap of ZnO x S 1 x thin film shows li near relation with the sulfur composition. The bandgap of the films using only H 2 O or H 2 S as oxidizer was measured as 3.20 and 3.62 eV, respectively. This value is similar to that reported in the literature (3.18 and 3.62 eV for ZnO and ZnS, respectively) [227, 228] The minimum bandgap can be observed on the film with 1:1 H 2 O/H 2 S pulse ratio, corresponding to 0.03 sulfur composition in the film and further increasing the sulfur composition leads to the rise of the bandgap of the film. This trend was also repor ted in several previous studies [25, 26, 229] The most important feature of ZnO x S 1 x thin film is its tunable properties and by adjusting the S composition in the film the film bandgap can be adjusted between 2.95 and 3.62 eV. The bandgap can be affected by the lattice structure of the film, the s composition in film as well as the (OH) group residue [25] However, based on the film composition analysi s, no O H bound was observed. Therefore, the crystallinity and S

PAGE 144

144 composition are the two main reasons varying the bandgap. With small H 2 O or H 2 S fraction, the crystallite size remains between 30 50 nm but the film turns in to amorphous when 1/1 H 2 O/H 2 S puls e ratio was used and the bandgap reaches minimum. Therefore, it is highly possible that for multicrystalline film, S composition is the main reason leads to the variation of the bandgap and the phase change from multicrystalline to amorphous may further de crease the bandgap. ZnO x S 1 x Thin Film on CIGS with Stainless Steel Substrate Although the mixed ZnO x S 1 x film was successfully deposited on Si substrate, the deposition of ZnO x S 1 x film on CIGS does not show complete surface coverage and good confomality. As example, Figure 7 17 shows multiple particles with 30 to 50 nm diameter attached on the CIGS surface rather than complete film coverage as was the case when depositing on Si According to the XPS spectrum shown in Figure 7 18 Zn, O and S are not the only elements detected on the surface. Other elements from underneath layer, such as Cu and In were detected as well. This indicates the incomplete coverage of the ZnO x S 1 x film Of course the surface structure and termination of the substrate influences the A LD window. Also, the thickness uniformity of the film decreases with the increasing of S conce ntration of the film. The non uniform thickness pattern suggests that contact resistance between the heater and substrate was not uniform. The samples were provid ed by GSE and the CIGS was thus grown on a SS substrate. This suggests that a surface temperature difference variation was induced by the poor therm al contact between the heater and the stainless steel substrate as a result of a slight convex curvature It is further noted that films grown on rigid Si substrates were uniform in thickness across the substrate.

PAGE 145

145 According to the ALD window of ZnO and ZnS film shown in Figure 7 3 when the temperature is lower than the ALD region, the ZnO growth rate decreases and the ZnS deposition rate increases rapidly. The CIGS sample is based on ~0.5 mm thin stainless steel sheet which is flexible. The slight curvature in the sample thus leads to the variable thermal contact between the CIGS sample and the heater under vacuum. T hus a temperature gradient forms from the center to the two sides of the sample, causing poor uniformity of ZnO x S 1 x film. T o better initiate the first layer o f reaction at the sample surface, t he sample was first washed with warm acetone to remove any organic contamination on the surface and then was dipped into hot water maintained at 85 C for 5 minutes to remove the sodium salts and other soluble contaminati ons on the CIGS surface. To establish a more uniform temperature profile the substrate heater assembly was modify by inserting a magnet ic plate ( 1/16 inch thickness Neodymium ) between the heater surface and the SS substrate of the GSE bare CIGS sample as schematically shown in Figure 7 19 As shown in Figure 7 20 after the pretre atment, the deposited ZnO x S 1 x film shows good coverage and uniformity on CIGS substrate and the absence of the Cu, In peaks in the XPS spectrum ( Figure 7 21 ) confirms the good coverage. Complete Cell Performance To make complete cell, a layer of ZnO:Al was sputtered on top of the buffer layer using ZnO target containing 1% Al doping. The oxygen flow rate during the sputtering was controlled at 3 sccm in order to control the oxygen vacancy in the TCO to achieve goo d conductivity. The total sputtered time was controlled at 25 minus and the thickness of TCO layer was measured as 239.5 nm. Finally, a layer of Al/Ni metal contact grip was

PAGE 146

146 deposited on the top of TCO layer using E beam evaporation. And for all samples, t he buffer layer was controlled around 50 nm. The J V curve of complete cells shown in Figure 7 22 was measured under AM 1.5 one sun c ondition using solar simulator. Table 7 1 shows the detailed parame ters of the measured curve. The buffer layer of the reference cell was CdS deposited using Chemical Bath Deposition (CBD). The TCO and metal contact grid of the reference cell were deposited using the same methods as the cell with ZnO x S 1 x buffer layer. As shown in Figure 7 23 at H 2 S pulse rich side, the cell using pure ZnS thin film as buffer layer does not show any efficiency and the highest efficiency, 11.58%, was a chieved with 0.91 H 2 S pulse fraction. The efficiency then decreases with increasing S concentration, reaching 0% at 0.5 H 2 S pulse fraction. At the H 2 O pulse rich side, the highest efficiency was measured as 8.5% with 0.25 H 2 S pulse fraction. The efficiency of the cells with H 2 O pulse rich buffer layer does not show significant degradation with the decreasing H 2 S pulse fraction. On the H 2 O pulse rich side, the increasing efficacy can be attribute to the increasing of V OC since the J SC of the cells does not s how large variation. The relatively higher V OC may also lead to better Fill Factor since the higher V OC can decrease the ration of rounded area in the J V curve. Furthermore, from the QE results shown in Figure 7 24 A the cells with 3:1 H 2 O:H 2 S pulse fraction buffer shows higher response by 10% than other cells between 700 1000 nm. For the samples with H2S pulse rich buffer layer, significant decreasing in J SC down to less than 1 mA/cm 2 with the decreasing H 2 S pulse fraction is observed, which contributes to the large part of the efficiency loss. According to the QE measurement shown in Figure 7 24 B the excessive sulfur composition lead to the decrease over the entire spectrum, especially

PAGE 147

147 between 400 1000 nm. The reduction in QE between 350 400 nm is relatively smaller. Compared to the cell using CdS as buffer layer the cell using ZnO x S 1 x thin film is 1. 6 2% lower in efficiency. A lower fill factor is primarily responsib le for the reduced performance. Beside the aging of the absorber during the sample transportation and treatment, the parasitic resistances induced b y recombination at the absorber/buffer layer interface o r the conduction bandgap offset may also lowering the efficiency. A serious of complete cells using ZnO x S 1 x buffer layer deposited with different number of supercycles was fabricated in order to find out the optimum buffer layer thickness. The H 2 O/ H 2 S pulse ratio was controlled at 3/1. The highest efficiency 8.44% was measured on the cells deposited using 39 supercycles and the efficiency decreasing with the increasing number of supercycles. The J V c urve and the detailed parameters are shown in Figure 7 25 and Table 7 2 As shown in Table 7 2 the increase of buffer layer thickness does not affect J SC and fill factor significantly but shows negative effect on the V OC 17% reduction in V OC from 0.47 V to 0.3 7 V was observed when the number of supercycles increased from 3 9 to 91. The transmission of the buffer layer deposited using 4 difference number of supercycles is plot in Figure 7 26 A and the thickness is shown in Figure 7 26 B According to the results, the thickness of ZnO x S 1 x generally shows linear relation with the total number of supercycles, which is a good indication of ALD growth mode The cut off edge of the transmission curve does not change with the number of supercycle which means the optical bandgap of the thin film does not change with its thickness and the sulfur composition in the film is consisten t as long as the H 2 O/ H 2 S pulse ratio is fixed. Compared to the film using 39 supercycles (253 cycles), the

PAGE 148

148 transmission of the film deposited with 91 super cycles (396 cycles) shows more than 10% reduction between 500 800 nm. The reduced transmission sugge sts less photon reach the absorber, which leads to the decrease of electron hole pair generation and open circuit voltage. Summary In this study, ZnO x S 1 x thin film was prepared using spercycles atomic layer deposition. The S composition shows strong negat ive deviation from the H 2 O/H 2 S pulse ratio, indicati ng strong exchange reaction of O for S The H 2 O/H 2 S pulse ratio shows strong effect on the morphology and the structure of the thin film. Elongated and round shape grains were observed for H 2 O pulse rich and H 2 S pulse rich film. Both film structure and S composition affect the bandgap properties of the thin film and a bowl shaped relation was mapped out between the bandgap and sulfur composition. Maximum complete cell efficiency 11.58% was measured with th e cell using ZnO x S 1 x thin film with 1/10 H 2 O/H 2 S pulse ratio. Compared to the cell using CdS, the relatively lower fill factor is primarily responsible for the reduced performance, which may be caused by parasitic resistances induced by recombination at t he absorber/buffer layer interface or the conduction bandgap offset. Reducing the buffer layer thickness to 30 nm can further increase the cell efficiency due to the increasing transmission between 500 800 nm.

PAGE 149

149 Table 7 1 Parameter table from J V curves of devices using buffer layer prepared with various H 2 S pulse fraction H 2 S Pulse Fraction J SC V OC Fill Factor Efficiency ( mA/cm 2 ) ( V ) % % 0 33.097 0.38 49.55 6.24 0.16667 30.313 0.44 52.91 7.19 0.25 33.69 0.46 54.67 8.502 0.5 0.955 0.19 17.11 0.031 0.75 17.107 0.4698 17.33 1.392 0.90909 29.36 0.618 63.77 11.58 0.96774 0 0 0 0 1 0 0 0 0 Table 7 2 Parameter table from J V curves of devices using buffer layer prepared with H 2 O/ H 2 S pulse ratio 3/1. Number of Supercycles V OC J SC Fill Factor Efficiency (V) ( mA/cm 2 ) % % 39 0.47149 30.12793 0.594183 8.440376 69 0.3885667 29.4593 0.617987 7.074036 78 0.38737 29.17231 0.623583 7.046789 91 0.37267 28.88534 0.603384 6.495252

PAGE 150

150 Figure 7 1 Schematic setup of ALD system design.

PAGE 151

151 Figure 7 2 3 d view schematic of the newly designed quartz substrate carrier

PAGE 152

152 Figure 7 3 ALD window for ZnO and ZnS deposition

PAGE 153

153 Figure 7 4 XPS spectrum of ZnO and ZnS thi n film without doping process.

PAGE 154

154 Figure 7 5 XPS spectra for film with different cycle ratio of Zn S to Zn O The value of ZnS:ZnO was set at 9:1(black), 7:1(red), 5:1(blue), 3:1(pink) and 1:1(green). Note that film deposition began with ZnO.

PAGE 155

155 Figure 7 6 Sulfur concentration on the Group VI sublattice vs. pulse r atio

PAGE 156

156 Figure 7 7 Multiplex Scan on ZnO x S 1 x thin flim for O 1s, Zn 2p3 and S 2p peaks.

PAGE 157

157 Figure 7 8 AFM image of ZnO thin film surface deposited with va rious DEZn and H 2 O pulse time. A) DEZn 0.1 s H 2 O 0.08 s B) DEZn 0.5 s H 2 O 0.1 s C) DEZn 0.8 s H 2 O 0.5 s D) DEZn 1.3 s H 2 O 1 s E) DEZn 1.5 s H 2 O 1.8 s F) DEZn 2 s H 2 O 3 s

PAGE 158

158 Figure 7 9 Surface roughness of ZnO thin film vs. DEZn and H 2 O pulse time.

PAGE 159

159 Figure 7 10 SEM surface of H 2 S pulse rich ZnO x S 1 x samples. H 2 O/H 2 S Pulse Ratio : A) 0:1 B) 1:5 C) 1:1

PAGE 160

160 Figure 7 11 SEM surface of H 2 O pulse rich ZnO x S 1 x samples. H 2 O/H 2 S Pulse Ratio: A) 1:0 B) 30:1 C) 15:1

PAGE 161

161 Figure 7 12 XRD spectrum of ZnO x S 1 x thin film with carious H 2 O/H 2 S pulse ratio.

PAGE 162

162 Figure 7 13 Crystallite size extracted from XRD FHMW of ZnO x S 1 x thin film deposited with various H 2 O pulse fraction.

PAGE 163

163 Figure 7 14 Transmission of ZnO x S 1 x with various H 2 O pulse fraction.

PAGE 164

164 Figure 7 15 Square of absorption over wavelength of ZnO x S 1 x thin film.

PAGE 165

165 Figure 7 16 Bandgap of ZnO x S 1 x thin f ilm with various sulfur composition.

PAGE 166

166 Figure 7 17 H 2 O pulse rich ZnO x S 1 x thin film on CIGS substrate without pretreatment. Figure 7 18 XPS spectrum of ZnO x S 1 x thin film on CIGS substrate without any treatment.

PAGE 167

167 Figure 7 19 Modified heater assembly showing the insertion of a magnetic plate between the heater surface and CIGS/Mo/SS GSE sample for the ALD reactor c hamber. Figure 7 20 Cross section of CIGS sample with ZnO x S 1 x buffer layer deposited using ALD.

PAGE 168

168 Figure 7 21 XPS spectrum of CIGS substrate with ZnO x S 1 x buffer layer deposited using ALD Figure 7 22 C urve of complete cells using buffer layer of various sulfur concentration.

PAGE 169

169 Figure 7 23 Change of c omplete CIGS solar cell performance with buffer layer deposited using various H 2 S pulse ratio. A) S hort circuit current density B) O pen circuit voltage C) F ill factor D) E fficiency Red line refers to the value of CIGS reference cell using CdS as buffer layer.

PAGE 170

170 Figure 7 24 QE of complete cell deposited with various H 2 S pulse fraction.

PAGE 171

171 Figure 7 25 C urve of complete cells using ZnO x S 1 x buffer layer with various thickness.

PAGE 172

172 Figure 7 26 Trans mission and thickness of ZnO x S 1 x thin film using H 2 O/H 2 S pulse ratio 3:1 deposited with different number of cycles. A) Transmission. B) Thickness.

PAGE 173

173 CHAPTER 8 ATOMIC LAYER DEPOSITION OF IN DOPED ZNO THIN FILM Literature Review Al doped ZnO has been the primary TCO used in CIGS thin film devices [230] although it has been suggested that Ga doped ZnO has potential advantages [41] Identifying the optimal tradeoff between transparency and conductivity as well as preventing pin hole formation are generally accomplished by detailed experiment. ALD has the potent ial to achieve controlled substitutional dopant incorporation, improve crystallinity, and yield excellent coverage. There has also been interest in developing a Cd free device and thus a buffer layer based on ZnO has the possibility of replacing the n CdS buffer layer as well as providing the TCO using a single deposition process. Given the presence of In in the CIGS absorber, the influence of In on the buffer layer is also of interest. Unfortunately, growth of ZnO:In has received little attention and onl y one report by ALD, which was at a temperatu re too high for CIGS processing [61] T here has also been considerable interest in developing a variable bandgap buffer layer to better align the band edges to CIGS (e.g. addition of more Ga, Ag, and S). ZnO 1 x S x (ZnOS) is a candidate buffer layer material sinc e its electrical and optical properties are tunable It is critical in depositing both ZnOS and ZnO:In to precisely control the O/S and Zn/In atomic ratio in the film. Atomic layer deposition typically provides precise and effective doping levels as well a s vertical placement of the ionized dopants since residence time during deposition can be varied during the deposition of each layer. In this study, ZnOS and ZnO:In thin films were deposited using ALD with diethylzinc (DEZn), trimethylindium (TMIn), H 2 S and H 2 O as precursors. The ALD windows for deposition of ZnO and ZnS thin films were determined and the

PAGE 174

174 microstructural properties of ZnO:In thin film were examined. In addition 2 cells were fabricated by depositing ZnO:Al or ZnO:In using sputtering and c ontinuous super cycle ALD, respectively, on a stacked layer of ZnOS/CIGS/Mo/S.S. to analyze the complete cell performance. Experimental Procedures In doped ZnO thin film were deposited on Si (100), Borosilicate Glass and CIGS sample terminated with CdS o r ZnOS buffer layer. The CIGS sample terminated with CdS buffer layer were prepared using Co evaporation and Chemical Bath Deposition method by Global Solar Energy. The CIGS substrate structure were shown in Figure 8 1 Before deposition, all substrate s were first cleaned using Electronic Grade t richloroethylene acetone and methanol. The Si (100) substrate were finally etched with BOE. CIGS substrate and b orosilicate glass were rinsed in warm de ionized water for 4 minutes. All substrates were dried with N 2 before placed in to the reaction chamber. During the deposition, DEZn, TMIn and water were used as precursors. The film deposition temperature s were set at 150 C and 200 C to test the properties of the film deposited under different temperature. The reactant sequence for depositing ZnO:In film was set as ( DEZn purge H 2 O purge) x ( TMIn purge H 2 O purge) y where x and y are number of times each Zn O and In O cycle are respectively executed. The value of x and y depends on the desired In doping level in the ZnO:In film. The deposition parameter s were optimized in order to keep the deposition as ALD mode. The background pressure was 0.171 Torr and was kept at 0.4 Torr during film growth. The film thickness was measured suing Filmetrics F20 thin film analyzer. The Zn/In ratio in the bulk films were determined with I CP/AES Perkin Elmer Optima 3200 RL The bounding status of each element were measures with Perki n Elmer 5100 XPS

PAGE 175

175 System using Mg X ray source. XRD were performed on samples with 200 nm thickness using Panalytical XPert Powder system. Shimadzu UV 2401PC spectrophotometer was used for absorption spectroscopy in the wavelength range of 350 nm to 800 nm. The sheet resistance, electron concentration (n) and mobility () were determined by Hall measurements under a magnetic field of 10 kG with Lakeshore 7507 The surface metrology was measured using SPM/AFM Dimension 3100 and FEI Nova NanoSEM 430 Experimen t Results and Discussion Film Composition XPS scan was performed on ZnO:In thin film deposited using ( DEZn purge H 2 O purge) 40 + ( TMIn purge H 2 O purge) 1 supercycle sequence at 200 C. The chamber pressure was kept at ~10 9 Torr and the detection range was set between 1100 eV to 0 eV. The results of XPS measurement were shown in Figure 8 2 The carbon 1s peak on the spectrum of the sam ple surface was due to the contamination such as organics and carbon dioxide from the ambient air. After five minutes sputtering using Ar + no carbon peak was found within the detection limit of XPS (1%), which indicates that all precursors reacted with th e oxidizer completely during the deposition. Furthermore, the decomposition of the precursor at the substrate were not observed. The In atomic percentage were analyzed using the following equation: ( 8 1 ) where I i is the intensity of the photoelectron peak for a certain element, N i is the average i is Scofield factor

PAGE 176

176 and K is the all other factors related to quantitative detection of a signal. The calculation was performed using AugerScan Version 3.22 and the results were plotted in Figure 8 3 A. According to the result, the measured In atomic percent in the film increases with the Ar + sputtering time gradually from 0.6% to 1.6%. One reason is that the DEZn and TMIn were introduced into the reactor separately. Therefore, the deposit ed film will have a layered structure, which was referred as type doping [231] In this experiment, the deposition sequence ends with ( TMIn pu rge H 2 O purge ) cycle and after 5 mins sputtering, the In at top surface were thus removed. Generally, the XPS detection depth limit is between 3 10 nm and 40 cycles of DEZn purge H 2 O purge deposition created ZnO layer with 8.8 nm thickness. Therefore, very limited electrons from In atoms can escape the surface of the film. However, with the long sputtering time, the thickness of ZnO top layer decreases, which leads to the stronger signal from the In atoms beneath the ZnO layer. The increasing detected In co mposition with the sputtering time confirms that the deposited film has a layered structure and therefore, it is necessary to perform other bulk measurement method such as ICP AES. In order measure the In/Zn ratio of the film, the samples were fully digest ed in 5% Nitric Acid for ICP AES measurement and the measured In atomic composition in the film vs. TMIn pulse fraction was plotted in Figure 8 3 B According to the re sult, the measured In composition shows 1:1 linear correlation to the TMIn pulse fraction within 0 0.4 TMIn pulse fraction range, which indicates that the deposition was controlled lf

PAGE 177

177 behavior and during the deposition, each cycle will deposit one monolayer film on to the substrate. Therefore, the In/Zn ratio in the film is cl ose to the TMIn pulse fraction. Multiplex XPS scan was performed within 421 428 eV, 527 534 eV and 1021 to 2016 eV range, in order to exam the atom binding status of In, Zn and O atoms. The samples were prepared using DEZn:TMIn=5:1 recipe, in order to increase the signal noise ratio of In peak. The step size was set as 0.125 eV. The spectrum were aligne d using the standard carbon peak at 284.6 eV and were shown in Figure 8 4 It is noted that the Zn, O and In peaks are located at 1022 .16 eV, 529.97 eV and 444.71 eV, respectively The O 1s peak can be spilt into two peaks, l ocated at 529.97 eV and 531.47 eV, stand ing for Zn O and In O bound respectively. All these measured binding energy value matches the binding energy value for Zn O and In O in XPS binding energy database, which indicates all the Zn and In atoms are bounded to O in the film According to the resu lts of quantitative analysis for O peak. The ratio of Zn O bound to In O bound is around 4, which is 20% deviation from the cycle ratio 5:1. However, as stated above, the quantitative results from XPS can be inaccurate due to the layered structure of the f ilm and low detection depth of XPS. Structural Properties The XRD patterns of pure ZnO thin film and In doped ( ZnO:In ) on borosilicate glass are shown in Figure 8 5 A ll sampl es exhibit three obvious peaks at 31.87, 36.35 and 53.82 which correspond to reflections from the ZnO (100), (101) and (110) planes A relatively small peak can be seen at 34.56 corresponding to the ZnO (002) plane As the In doping level inc reas es the relative intensity of the 36.35 centered peak increases while that of the one at 31.87 decreases slowly. This indicates the preferred plane o f ZnO: In film changes with the incorporation of In into the ZnO thin film The XRD

PAGE 178

178 spectr a indicates that the In and ZnO mix homogeneously on the Wurtzite lattice and no evidence of In 2 O 3 formation. The d spacing value for two major peaks of ZnO (100) and (101) was extracted from the peak position using Bragg's law and plotted in Figure 8 6 The d spacing for plane (100) increases from 2.805 to 2.818 and for (101) the d spacing value also increases from 2.465 to 2.480 The results sugges ts the lattice parameter of ZnO:In thin film expands for both a and c axis. This can be explained by the large r In 3+ ion ic radius (0.94 ) compared to Zn 2+ (0.74 ). For hexagonal crystals, the linear stress components can be calculated using [232] ( 8 2 ) where C ij are the elastic stiffness constants and e ij is the linear strain in the i th direction. Based on (2), Maniv et al. [233] derived the biaxial strain model as ( 8 3 ) w here c is the measu red lattice constant and c 0 is the lattice constant for pure ZnO, which is 2.80739 [ICOD 01 079 0205 ]. By plug in the values of the stiffness constant for ZnO, C 11 =C 33 =2.1E11 N/m 2 C 12 =1.2E11 N/m 2 C 13 =1.05E11 N/m 2 [233] the equ ation becomes, ( 8 4 )

PAGE 179

179 The stress calculated using the biaxial strain model was plotted vs. the TMIn pulse fraction in Figure 8 7 A T he undoped ZnO film shows max imum stress, which is mainly due to the lattice mismatch between the substrate and thin film as well as that produced upon cooling With increasing TMIn pulse fraction, the stress value decreases significantly and reaches a plateau when the TMIn pulse fraction exceeds 0.1. This exponential correlation can be explained by the incorporation of the In 3+ ions, which occupy the substitutional sites in the hexagonal W urtzite ZnO lattice. The negative value of the stress indicates the compressive stress type, which can be explained by the lattice expansion due to the relatively larger radius of In 3+ ion. Singh et al. [40] reported similar results for 1% I n doped ZnO:In sputter ed at 700 C The crystallite size was As shown in Figure 8 7 B the crystallite size increases significantl y when the (100) peak shi f ts from 31.7222 to 31.7452, corresponding to TMIn pulse fraction of 0.16 and 0.05, respectively. However, the crystallite size remain s at the same level for peak position 31.8977 and 31.7425, corresponding to TMIn pulse fracti on of 0 and 0.05, respectively. The decreasing of the crystallite size was caused by the relaxation of the film stress when the TMIn pulse fraction exceeds 0.01. Morphological Properties The morphology of ZnO:In thin film with various In doping level was e xamined using SEM as shown in Figure 8 8 A and Figure 8 8 B corresponding to DEZn/TMIn ratio 90/1 and 5/1, respectively. The ZnO:In thin film with low In doping level shows elongated pyramidal grain around 52 nm on the surface, which corresponds to t he hexagonal structure of ZnO, while the sample with high doping level shows round grain shape with size more than 200 nm and on the grain surface there are small nucleation

PAGE 180

180 site with 10 nm size. The results confirm the structural and crystallite size info rmation obtained from XRD spectra. According to the crystallite size calculated with XRD spectrum, the grains in low In doping level ZnO tend to be single crystalline and as the increasing of In doping level, the grains convert into multi crystalline. of ZnO:In with various In doping level was shown in Figure 8 9 and the change of surface roughness was plotted with TMIn pulse fraction in Figure 8 10 According to the measurement, the roughness of pure ZnO sample was 3.5 nm and increases to 38.3 nm when TMIn pulse fraction reaches 0.33. The roughness shows linear correlation with TMIn pulse fraction with slop of 9.98 nm/0.1 TMIn pulse fraction. According to the SEM and AFM image, the film growth after the deposition of the initiation layer follows Volmer Weber mode due to the existence of the small nucleation isla nd observed on the surface. Although Baji et al. [234] reported that undoped ZnO thin film sho ws Volmer Weber growth mode with Si (100) substrate throughout the film deposition process, the growth mode for the initial layer of ZnO:In remains unclear and ne ed to be further characterized. Electrical Properties The electrical properties of ZnO:In film s at various In doping level s is displayed in Figure 8 11 These data are consistent with n type ZnO:In. The minimum value is measured a s 1.65x10 3 cm when DEZn /TMIn pulse ratio was controlled at 1 0/1 Further changing the ratio t o 5/1 significantly increase s the resistivity When the ratio decreases to less than 60/1 the sample resistivity is constant at ~ 3 x10 3 cm The bulk carrier concentration (n) of the sample shows monotonic correlation with the In doping level and increa ses significantly when the DEZn /TMIn pulse ratio is less than 30/1, while the carrier mobility ( ) decreases with increasing In doping level

PAGE 181

181 Although the incorporation of In leads to the surge of the carrier concentration as shown in Figure 8 11 the substitutional In atoms also provide extra stress to the fil m, resulting in smaller grains, which can lower carrier mobility due to the grain boundary scattering. In the case of 1 0/1 DEZn /TMIn pulse ratio the low er carrier mobility is compensated by high er carrier concentration, leading to the optimum she et resistance among all samples. Besides the grain boundary scattering, the electron electron scatt ering and ionized impurity scattering can also be the reasons for the low carrier mobility, especially for high doping level condition. The significance of these scattering mechanism will be tested with temperature dependent Hall measurement in the future work. The detailed electrical parameters of ZnO:In using ALD and one of the typical TCO material ZnO:Al using sputtering are listed in Table 8 1 Compared to the ZnO: In, ZnO:Al shows smaller resistivity by one magnitude. The main drawback of ZnO:In thin film is the low carrier mobility. Besides the reason mentioned above, the different deposition technique may place a significant effect on the film properties. Comple te Cell Performance T he J V curve for two complete d cell s with sputtered ZnO:Al and ZnO:In using ALD were measured. The results are shown in Figure 8 12 and the detailed parameters are listed in Table 8 2 The cell using ZnO:Al as TCO outperforms th at with ZnO:In thin film by 4.1 %. These two cells show similar open circuit vol tage at 0.55 mV but the sample with In doped ZnO shows lower short circuit current and larger series re sistance, leading to relatively lower Fill Factor. This is largely attributed to the low er mobility (1.68 cm 2 /Vs ) of ALD In ZnO compared with Al doped Z nO film ( 9.82 cm 2 /Vs ). Furthermore, the high carrier

PAGE 182

182 concentration in the ZnO:In thin film may leads to lower transmission due to the plasma oscillation effect of the bulk free electrons The reduced number of photon reaching the absorber leads to the redu ction of electron hole pair generation and thus shows negative effect on the J SC Summary ZnOS and ZnO:In films were prepared using ALD and the ALD deposition windows for ZnOS and ZnO:In were determined. Importantly, no secondary In 2 O 3 phase was observed. The minimum resistivity was measured as 1.65x10 3 cm with the main drawback is the low carrier mobility attributed to grain boundary scattering. The completed cells using ZnO:In as TCO show relatively low open circuit current densit y and high series resistance compared with the cell using sputtered ZnO:Al thin film as TCO.

PAGE 183

183 Table 8 1 Electrical Properties of ZnO:In (ALD) and Z n O:Al (Sputtering) (ohmcm) n (cm 3 ) (cm 2 /Vs) ZnO:Al (sputter) 5 .0 E 4 1.15 E+21 9.82 ZnO:In (ALD) 1.8E 3 1.95 E+21 1.68 Table 8 2 Properties of Complete Solar Cells TCO Material

PAGE 184

184 Figure 8 1 The layer structure of CIGS substrate prepared by Global Solar Energy. ZnOS buffer layer were prepared using ALD at UF Figure 8 2 XPS spectrum of ZnO:In thin film on Si sputtered w ith Ar + for various time length.

PAGE 185

185 Figure 8 3 In composition measured with different technique. A) XPS. B) ICP AES Figure 8 4 XPS peak positions of the elements in ZnO:In (pulse ratio Zn:In 5:1) thin film. A) In. B) O C) Zn

PAGE 186

186 Figure 8 5 XRD spectr a of pure ZnO and ZnO:In with different In level s The black bar s indicate the expected peak position for In 2 O 3 Figure 8 6 The change of d spacin g of ZnO 100 and 101 direction with increasing In doping level.

PAGE 187

187 Figure 8 7 Estimated stress and crystallit e size of In doped ZnO film vs. TMIn pulse fraction. Figure 8 8 Surface SEM image of ZnO:In wit h different In doping level.

PAGE 188

188 Figure 8 9 different In doping Level. DEZn/TMIn ratio: A) Pure ZnO B) 4 0 C) 2.

PAGE 189

189 Figure 8 10 The roughness surface evolution with the In doping level of ZnO:In films as Figure 8 11 Hall measurement s results for ZnO:In thin film

PAGE 190

190 Figure 8 12 J V curve for two completed cells.

PAGE 191

191 CHAPTER 9 COMPARIASON STUDY OF ZNO:IN AND ZNO:GA THIN FILM FOR PV USING ATOMINC LAYER DEOPSITION Literature Review CIGS thin film solar cell has demonstr ated high efficiency over 22.3 % [107] The properties of transparent conducting oxides (TCO) can affect the performance of the solar cell because it not only conducts the photocurrent generated from the underlying cell but also, in the meantime, transmits the photon to the absorber layer. Therefore, good transmission and low resistivity are two f avorable properties of TCO for photovoltaic application. Stadler et al. [28] did a comprehensiv e review on various TCO materials including ITO, ZnO:Al etc. These n type TCOs are important for thin film solar cell production and have been extensively studied [29 33] Among all these TCO materials, ZnO based material has drawn a lot of attention for PV applic ation due to its wide bandgap of 3.3 eV [34] Compared with Al doped ZnO, Ga and In can be a better dopant candidates due to the relatively smaller difference between their ion radius and Zn 3+ ion radius. Both In (ZnO:In) and Ga doped ZnO (ZnO:Ga) has been studied using various deposition method such as sputtering [35 4 2] chemical vapor Deposition [43 46] chemical spray [47 49] pulsed laser deposition [50, 51] and sol gel method [52 54] Compa red to the deposition technique mentioned above, atomic payer deposition (ALD) shows many advantage such as the ability to finely control the doping level and film thickness, typically lower deposition temperature, good film uniformity and conformality. Th ere are several studies about ZnO:Ga using ALD with various Ga precursor, such as trimethylgallium (TMGa) [55] gallium(III) isopropoxide (GTIP) [56] and tri ethylgallium (TEGa) [57 59] However, studies on ZnO:In were much less. Hsia et al. and Kim et al. used trimethylindium (TMIn) [60 ] with N 2 O and dimethylamino dimethylindium (DMLDMIn) [61] with H 2 O for atomic layer

PAGE 192

192 deposition respectively. Although these studies developed the deposition process, the deposition using CIGS/CdS bilayer substrate has never been conducte d before. Furthermore, the high deposition temperature (above 300 C) leads to the degradation of the solar cell [27] In this study, we deposited ZnO:In and ZnO:Ga by thermal ALD (Th ALD) using diethylznic (DEZn), TMIn, TEGa and H 2 O as precursors and oxidizer. The supercycle deposition recipe was adopted in this study [235] The properties of the deposited thin film were characterized and compared. Finally, these two TCOs were deposited directly on CIGS/CdS bilayer substrate. The conduction band offset (CBO) at the ZnO:X/CdS interface and the complete cell performance was measured. Experimental Procedures ZnO:In and ZnO:Ga thin film were deposited on Si (100), Borosilicate Glass and CIGS sample terminated with CdS using supercycle ALD. The CIGS sample terminated with CdS buff er layer were prepared using Co evaporation and Chemical Bath Deposition method by Global Solar Energy. The CIGS substrate structure were shown in Figure 9 1 Before deposition, all substrates were pre cleaned with Electronic Grade trichloroethylene, acetone and methanol. The Si (100) substrate were then etched with buffer oxide etch CIGS substrate and borosilicate glass were rinsed in warm de ionized w ater for 4 minutes in order to further remove the surface contaminates. All substrates were dried with N 2 before placed into the reaction chamber. During the deposition, DEZn, TMIn, TEGa were used as precursors and water was used as oxidizer. The film deposition temperature was set at 200 C. The reactant se quence for depositing ZnO:In and ZnO:Ga thin film was set as (DEZn purge H 2 O purge) x (TMIn(TEGa) purge H 2 O purge) y where x and y are number of times each Zn

PAGE 193

193 O and In(Ga) O cycle are respectively exec uted. The schematic deposition sequence was shown in Figure 9 2 The value of x and y depends on the desired doping level in the TCO film. The depositi on parameters were optimized in order to keep the deposition as ALD mode. The background pressure was 0.171 Torr and was kept at 0.4 Torr during film growth. The resistivity electron concentration (n) and mobility () were determined by Hall measurements under a magnetic field of 10 kG with Lakeshore 7507. The film thickness was measured suing Filmetrics F20 thin film analyzer. XRD were performed on samples with 200 nm thickness using Panalytical XPert Powder system. Shimadzu UV 2401PC spectrophotometer wa s used for absorption spectroscopy in the wavelength range of 350 nm to 800 nm. The surface metrology was measured using SPM/AFM Dimension 3100 and FEI Nova NanoSEM 430. The bounding status of each element was measures with Perkin Elmer 5100 XPS System usi ng Mg X ray source. Results and Discussion Electrical Properties The resistivity of ZnO:In and ZnO:Ga is shown in Figure 9 3 Compared with intrinsic ZnO, doping In and Ga can lower the resistivity of ZnO thin film to the order of 1.65 x 10 3 frac tion while for ZnO:Ga, the TEGa precursor pulse fraction 0.05 was used to archive the lowest resistivity. However, the resistivity tends to rise when the pulse ratio was beyond 0.15 for both films. With 0.17 TMIn and TEGa pulse fraction, the resistivity of the thin film increases to 6.6 x 10 3 2 indicates the induce of Ga into Zn site leads to the stronger increase of resistivity than

PAGE 194

194 that of In at high doping level (>0.1 pulse fraction). The resistivity of the th in film is a function of carrier concertation and Hall mobility [236] ( 9 1 ) mum sheet resistance can be reached by adjusting the carrier concertation and mobility. These two parameters were measured using Hall Measurement and the results are shown in Figure 9 4 The bulk carrier concentration of both materials monotonically increases with the doping level. At low doping level, the bulk carrier concentration of materials shows significant increase by one order of magnitude, up t o 2.3 x 10 20 cm 3 When the pulse fraction exceeds 0.05, the carrier concertation of ZnO:Ga stabilizes around 6.6 x 10 20 cm 3 while, for ZnO:In thin film, the carrier concertation ascends up to 2.7 x 10 21 cm 3 .The change of carrier concentration indicates th at both In and Ga dopants act as extrinsic donors and at high doping level, ZnO:In shows higher bulk carrier convention than ZnO:Ga. Although higher doping level results in higher carrier concentration, the mobility shows monotonically decreasing behavior with the increasing TMIn(TEGa) pulse fraction, from 15 m 2 /Vs (intrinsic ZnO) to 1.1 m 2 /Vs when the TMIn(TEGa) pulse ratio exceeds 0.15. Compared to ZnO:Ga, the bulk carrier concertation of ZnO:In reaches 2.3 x 10 20 cm 3 level using relatively small dopant precursor pulse fraction, but In dopant shows stronger negative effect on the carrier mobility than Ga dopant. As a result, it is

PAGE 195

195 necessary to incorporate more In dopant into intrinsic ZnO thin film in order to archive the minimum resistivity. Crystalline Structure The XRD patterns of intrinsic ZnO thin film and doped ZnO thin film on borosilicate glass are shown in Figure 9 5 The thickness of all fil ms were kept at 200nm in order to get optimum signal to noise ratio. As shown in Figure 9 5 all samples exhibit two obvious peak at ~31.87 and ~36.35 which are corresponding to the (100) and (101) plane of ZnO. A relatively small peak is shown at 34.56 corresponding to (002) plane. The intrinsic ZnO shows a strong preferential direction as (100), but as the doping level increasing, both samples dis plays a change of the preferential plane since the intensity of (101) peak gradually rises. The XRD spectrum of ZnO:Ga with DEZn/TMGa=5 (i.e. 0.17 TEGa pulse fraction) shows strong peak broadening effect and low peak intensity, indicating the deceasing of the crystallinity. Compared with ZnO:Ga, the peak broadening effect is less significant for ZnO:In, which suggests that Ga shows a stronger effect than In in term of the decreasing of the film crystallinity. In order to confirm the results, GIXRD was perfo rmed on samples with 0.1 precursor pulse fraction as shown in Figure 9 6 The results suggest both ZnO:In and ZnO:Ga shows hexagonal wurtzite ZnO structure and no peaks related to In2O3 or Ga2O3 was observed. The lattice parameters were derive d using the following equation: ( 9 2 ) where a and c are the lattice parameters and h,k and l are the miller index of the plane. The d spacing can be

PAGE 196

196 ( 9 3 ) Figure 9 7 The lattice constant of a axis for ZnO:In increases to 3.252 nm and remains at the constant value with further increasing doping level. However, for the c axis the lattice constant slightly rises and then monotonically decreasing with the increasing doping level. Different from the ZnO:In thin film, Ga doping does not show strong effect on the a axis lattice parameter but causes significant reduction in the latti ce constant of c axis. The stress induced by the dopant was analyzed. For hexagonal crystals, the linear stress components can be calculated using [232] : ( 9 4 ) where Cij are the elastic stiffness constants and e ij is the linear strain in the i th direction. Based on Equa tion 9 4 Maniv et al. [233] derived the biaxial strain model as ( 9 5 ) where c is the measured lattice constant and c0 is the lattice constant for pure ZnO, which are 2.80739 and 2.46903 for palne (100) and (101) respectively [ICOD 01 079 0205]. By plug in the values of the stiffness constant for ZnO, C 11 =C 33 =2.1E11 N/m 2 C 12 =1.2 E11 N/m 2 C 13 =1.05E11 N/m 2 [233] the equation becomes, ( 9 6 )

PAGE 197

197 The calculated value was plotted with the precursor pulse fraction in Figure 9 8 For the pure ZnO, the film shows maximum stress due to the lattice mismatch between the substrate and thin film as well as the thermal effect. For In doped samples, significant negative stre ss value was found for even low TMIn pulse fraction. However, stabilized at 1.01 GPa. Different from the In doped samples, the ZnO:Ga thin film exhibits positive stress aroun d 0.8 GPa. Based on the calculated stress value, the In dopant results in the compressive stress in the ZnO thin film while the incorporation of Ga leads to the tensile stress in the thin film. Compare to the radius of Zn 2+ ion (0.74 ), In3+ ion has relat ively larger radius (0.94 ) but the radius of Ga 3+ is smaller (0.62 ). The difference in the radius between these ions will causes the distortion of the lattice, leading to compression or tensile stress when the dopant ion is larger or smaller than the Z n ion, respectively. Singh et al. [40] reported that using 1% ZnO:In thin film sputtering at 700 C shows zero stress and then the value of stress turns to negative with increasing In doping level. However, for ZnO:Ga, both compressive and t ensile was reported using various deposition method (e.g. magnetron sputtering, CVD, etc.) [37, 41, 54] The negative stress is mainly caused by structural defect [37] and Negi et al. reported that annealing ZnO:Ga film can decrease the defect concentration and cause the film stress shift to positive value [54] The naturally tensile stress of ZnO:Ga may indica tes less lattice defect and the Ga 3+ ions directly occupy Zn 2+ position in the deposited film.

PAGE 198

198 The crystallite size and lattice strain were analyzed using the (100) peak, which shows maximum intensity in the XRD spectrum. The crystallite size was calculate d using ( 9 7 ) The results are shown in Figure 9 9 The crystallite size of ZnO:In remains at 44 nm until the precursor pulse fraction reaches 0.1 but Ga doped sample s shows small fluctuation around 40 nm. However, when the precursor pulse fraction is over 0.1, the crystallite size of both samples drops significantly, which may due to the increasing of the thin film microstrain caused by the oversaturated dopant. Morph ological Properties The surface morphology of the deposited thin film was studied using AFM and SEM. The SEM and AFM images of the deposited thin film were shown in Figure 9 10 and Figure 9 11 respectively. With 0.02 precursor pulse fraction, both ZnO:In and ZnO:Ga show grains with pyramidal shape which is corresponding to the hexagonal wurtzite ZnO structure revealed by the XRD spectra. However, the grains of ZnO:In show elongated shape around 43 nm while the grains of ZnO:Ga show relatively more round shaped with larger size around 236 nm. When th e precursor pulse fraction increases to 0.2, the surface morphology changed significantly. On the surface of ZnO:In, the pyramidal shapes of the grains deteriorate to many round shaped islands around 10 20 nm and these islands agglomerate, leading to large particles measured

PAGE 199

199 around 200 350 nm. For ZnO:Ga, similar effect was observed but the size of the particles is around 794 nm. The coalesce effect leads to higher surface roughness which was confirmed by AFM measurement ( Figure 9 12 ). The roughness of both thin films shows monotonically increasing relation with the increasing doping level. With the same precursor pulse fraction, ZnO:Ga shows higher roug hness than ZnO:In and the difference between the roughness of these two thin films increases with the precursor pulse fraction. Compared to undoped ZnO thin film with roughness around 2.2 nm, the roughness of ZnO:In and ZnO:Ga, with 0.16 precursor pulse fr action, rises to 17 and 4 5 nm, respectively. According to the analysis of XRD spectrum, both macrostress and microstrain increases with the precursor pulse fraction. Therefore, with high doping level, it is more likely for lattice relaxation to take place, which results in the smaller crystallite site and the sharp hill and valley structure throughout the film surface. For ZnO:In with 0.02 precursor pulse fraction, the observed particles sizes are identical to the results analyzed using XRD, which indicates that for ZnO:In thin film with low doping level, the grains are more likely to be single crystalline. However, as the In concentration increases, the grains tends to become multi crystalline as the size of observed grains are serval times of the crystalli te size derived from XRD measurement. For ZnO:Ga, even at low doping level, the crystallite tends to agglomerate forming large particles. Optical Properties Transmission of ZnO:In and ZnO:Ga thin film with various doping level was plotted in Figure 9 13 And the adsorption was calculated using

PAGE 200

200 ( 9 8 ) where A is absorption with arbitrary unit and T is the transmission. The band gap of the samples was exacted from absorption spectra using Absorption Spectrum Fitting (ASF) method proposed by Ghobadi et al. [225] The absorption coefficient can be described as a function of wavel ength using the equation below, ( 9 9 ) constant, and velocity of light, respectively. B is constant and m is the index which can have different values of 1/2, 3/2, 2, and 3 [226] It has been approved t hat the best fitting can be achieved for m=1/2 [225] and using the Beer Lamber ation 9 9 becomes ( 9 10 ) Talking m=1/2, the linear part of Figure 9 14 The bandgap was calculated using the data obtained by extrapolating the line to and the equation as following: ( 9 11 ) The derived optical bandgap for both materials was plot in Figure 9 15 As shown in Figure 9 13 generally speaking, both films show good transmission in the visible region (400 800 nm) and the ZnO:Ga shows relatively higher transmission than the ZnO:In. Both free carrier concentration and the surface roughness can affect the optical

PAGE 201

201 properties [54, 237, 238] Higher electron density may lower the transmission of the thin film due to the absorption of light via plasma oscillatio n of the bulk free electrons and higher surface roughness exhibits the same effect due to the optical scattering. Although ZnO:Ga shows higher surface roughness, the optical transmission is still relatively higher than ZnO:In. Considering the free electron density of ZnO:In is one magnitude higher than that of ZnO:Ga thin film, it is reasonable to assume the free carrier concentration other than the roughness of the thin film, plays a significantly role in changing the optical properties. At room temperature, the bandgap of undoped ZnO is 3.3 eV [239] and it was re Ga 2 O 3 and In 2 O 3 was 4.9 eV [240] and 2.93 eV [241] respectively. With the increasing doping level, the blue and red shift effect can be observed on ZnO:Ga and ZnO:In thin film, respectively. The bandgap shifts to 3.55 eV and 3.14 eV for ZnO:Ga a nd ZnO:In with 0.17 precursor pulse fraction. The change of the bandgap corresponds to the bandgap difference between the oxides of In/Ga and Zn. For ZnO:Ga, the blue shift of the absorption edge can be attribute to the Burstein Moss effect which means the lowest states of conduction band are occupied by the free careers induced by Ga dopant and thus more energy is required f or valance electrons excitation [29, 41, 242] Different from the effect on bandgap induced by Ga dopant, In dopant cause red shift of the absorption edge. This is because the when the In atoms gets close enough in the film, the wavefunctions of the electrons bound to the impurity atoms start to overl ap, forming new recombination centers with lower emission energy [52, 243, 244] Furthermore, the change of bonding strength due to size difference between the In 3+ /Ga 3+ ion and Zn 2+ could also affect the bandgap.

PAGE 202

202 Cell Performance The top contact Ni/Al bilayer grid was deposited using E beam evaporator. The measured J V curve was plot in Figure 9 16 and the detailed parameters are l isted in Table 9 1 The cell with ZnO:Ga as TCO layer outperforms that with ZnO:In. The ideal TCO material for solar cell generally shows wide bandgap and low roughness to allow maximum photon transmission, moderate carrier concentration to control the plasma oscillation of the bulk free electrons and high carrier mobility to maintain high conductivity. According to the results of Hall measurement, ZnO :Ga shows much higher carrier mobility than ZnO:In when the resistivity reaches minimum. The low carrier mobility can be attribute to grain boundary scattering due to small crystallite size, ionized impurity scattering due to higher dopant concentration an d electron electron scattering due to high electron concentration. Both films with minimum resistivity show similar crystallite size around ~40 nm. Thus the grain boundary density could be similar for both films. However, ZnO:Ga thin film was deposited usi ng less precursor pulse fraction, leading to relatively lower dopant level. Thus, it is concluded that the high carrier concentration of ZnO:In thin film is the major reason for the relatively lower carrier mobility. And the carrier mobility of both films can be further improved by forming larger crystallite size using higher deposition temperature or post annealing method [27, 54, 238, 245, 246] Although ZnO:In thin film shows relatively lower surface roughness, due to its relatively high carrier concentration, its optical transmission is still around 8% lower than ZnO:Ga thin film in the visible 400 800 nm region [211] The reduction in transmission may directly decrease the light generated current, leading to the lower open circuit

PAGE 203

203 voltage and short circuit current. Furthermore, the ZnO:In shows high carrier concentration but low mobility while ZnO:Ga shows opposite properties. The high carrier concentration of ZnO:In, may also raise the fermi level and leads to larger CBO at the heterojunction interface [36, 247] In order to figur e out the band alignment at ZnO:X/CdS interface, the valance band offset (VBO) was firstly measured using XPS. The detailed XPS spectra are shown in Figure 9 17 XPS analysis on bulk CdS, ZnO:Ga, ZnO:In and ZnO:Ga (4 nm)/CdS, ZnO:In (4 nm)/CdS bilayer samples. and the detailed values are shown in Table 9 2 XPS binding energy of ZnO:Ga, ZnO:In, CdS and TCOs /CdS interface The VBO of the ZnO:X/CdS was determined from the energy separation between the core levels in the ZnO:X/CdS sample and the Valance Band Maximum (VBM) to core level separations in the thick layer ZnO:X a nd CdS samples, as described by ( 9 12 ) where refers to the binding energy of n in sample i. With the bandgap of ZnO:X was extracted from Figure 9 15 the CBO can be calculated using ( 9 13 ) where is the bandgap of material i. The detailed band diagram was plot in Figure 9 18 The CBO of ZnO:Ga/CdS shows 0.57 eV less than that of ZnO:In due to its relative small VBO and larger bandgap. Considering the CBO of CdS/CIGS is 0.2 0.3 eV, the CBO between ZnO:In and absorber CIGS layer could be larger than 0.4 eV, which may result in the double diode characteristic for the cell, decreasing the J SC and F ill Factor [248] The relatively lower FF factor of the cell with ZnO:In thin film can be

PAGE 204

204 attribute to the relatively low open circuit voltage since the resistivity o f the ZnO:In and ZnO:Ga thin film are identical, which eliminate the difference caused by parasitic resistance. Summary In and Ga were doped into ZnO thin film using atomic layer deposition. The comparable minimum resistivity 1.65 x 10 cm was reached u sing 0.1 and 0.05 precursor pulse fraction for ZnO:In and ZnO:Ga, respectively. For the samples with minimum resistivity, the ZnO:In shows high carrier concentration but low mobility while ZnO:Ga shows high mobility but relatively lower carrier concentrati on. Based on the characterization using XRD, AFM and SEM, the relatively lower carrier mobility of ZnO:In can be mainly attributed to ionized impurity scattering and electron electron scattering. The efficiency of complete cell using ZnO: In and ZnO:Ga as TCO material was measured as 6.03 and 11.51 %. The cell with ZnO:Ga shows better open circuit voltage and short circuit current because the ZnO:Ga shows better transmission property and a more favorable conduction band offset at the ZnO:X/C dS interface.

PAGE 205

205 Table 9 1 Parameters of the complete cell performance TCOs V OC (V) J SC (mA/cm 2 ) FF Efficiency (%) ZnO:In 0.57 18.11 0.58 6.03 ZnO:Ga 0.63 28.30 0.64 11.51 Table 9 2 XPS binding energy of ZnO:Ga, ZnO:In, CdS and TCOs /CdS interface ZnO:Ga ZnO:In CdS ZnO:In/CdS ZnO:Ga/CdS (eV) Zn 2p 3/2 1022.95 1022.75 1022.45 1022.52 VBM 2.71 2.83 0.75 Cd 3d 5/2 405.15 405.45 405.4 Figure 9 1 Schematic diagram of CIGS solar cell.

PAGE 206

206 Figure 9 2 Schematic diagram of atomic layer deposition cycle sequence. Figure 9 3 Resistivity change with the In or Ga doping level in ZnO thin flim.

PAGE 207

207 Figure 9 4 Electrical properties of In(Ga) doped ZnO thin film with various doping level.

PAGE 208

208 Figure 9 5 XRD spectrum of doped ZnO with different doping level. A) In doped ZnO. B) Ga doped ZnO.

PAGE 209

209 Figure 9 6 GIXRD on do ped ZnO thin film with 0.1 TMIn(TEGa) pulse fraction. A) In doped ZnO. B) Ga doped ZnO.

PAGE 210

210 Figure 9 7 Lattice Parameters change vs. the precursor pulse fraction A) A axis. B) C axis.

PAGE 211

211 Figure 9 8 Stress of In(Ga) doped ZnO vs. precursor pulse fraction.

PAGE 212

212 Figure 9 9 Crystallite size and Microstrain of In(Ga) doped ZnO thin film vs. the precursor pulse fraction. The black square indicates the crystallite size and microstrain of Ga doped ZnO and the blue round indicates In doped ZnO.

PAGE 213

213 Figure 9 10 SEM images of doped ZnO with different precursor pulse fraction A) ZnO:In with 0.02 precursor pulse fraction. B ) ZnO:In with 0.2 precursor pulse fraction. C) ZnO:Ga with 0.02 precursor pulse fraction. D ) ZnO:Ga with 0.2 precursor pulse fraction.

PAGE 214

214 Figure 9 11 AFM images of doped ZnO with different precursor pulse fraction A) ZnO:In with 0.02 precursor pulse fraction. B ) ZnO:In with 0.2 precursor pulse fraction. C) ZnO:Ga with 0.02 precursor pulse fraction. D ) ZnO:Ga with 0.2 precursor pulse fraction. Figure 9 12 Surface roughness of In and Ga doped ZnO thin film as a function of precursor pulse fraction.

PAGE 215

215 Figure 9 13 Optical transmission spectra of doped ZnO thin film with various precursor pulse fraction. A) In doped ZnO. B) Ga doped ZnO.

PAGE 216

216 Figure 9 14 doped ZnO thin film showing the bandgap change with the doping level. A) In doped ZnO. B) Ga doped ZnO.

PAGE 217

217 Figure 9 15 Optical bandgap of In and Ga doped ZnO thin film.

PAGE 218

218 Figure 9 16 J V curve of complete cells using In and Ga doped ZnO thin film as TCO.

PAGE 219

219 Figure 9 17 XPS analysis on bulk CdS, ZnO:Ga, ZnO:In and ZnO:Ga (4 nm)/CdS, ZnO:In (4 nm)/CdS bilayer samples. Figure 9 18 Band diagram of ZnO:Ga/CdS and ZnO:In/CdS using 0.05 and 0.1 TEGa and TMIn pulse fraction, respectively.

PAGE 220

220 CHAPTER 10 CONCLUSION Thermodynamic Assessment of Ag Cu In Ga Se S ystem CALPHAD method was u sed to conduct critical assessment for Ag Cu In Ga Se system. For Ag Cu system, new experimental data were collected and used in the assessment. The solubility of Cu in Ag in FCC phase was evaluated based on not only phase equilibrium data but also the activity of Ag and Cu in the solid phase. The incorporation of the th ermodynamics data enables the database more physically founded. The Ag Se thermodynamic database was established in this work for the first time. The homogeneity range of the intermetallic compound and the liquids was well assessed based on available exper imental data. This assessed databased will be of high importance for many other Ag Se related higher order system as may AgSeX (X=Cu, As, S, Al, Ge, etc.) compounds or alloys show practical use in semiconductor or superionic fields. The higher order ternar y systems Ag Cu Se, Ag Ga Se and Ag In Se was assessed based on the binary subsystems and the ternary experimental data. A good agreement has been reached between the calculated and experimental data. The assessed database for Ag Cu In Ga Se system was use d for the prediction of phase relations during the selenization process. It was found that, the existence of Cu in the unsele (Ag,Cu) 9 (In,Ga) 4 phase, which will 2 phase even a t low Se composition. In the meantime, with the present of Cu, GaSe can be avoided at the beginning stage of selenization. However, one thing worth noticing is that the excessive incorporation of Se can happen for sample with Ag to group I meatal ratio at 0.25, forming defect phase.

PAGE 221

221 Therefore, it is of vital importance to control the incorporation of Se during the selenization process. Buffer and TCO Layer Deposition for CIGS Thin Film Solar Cell Using ALD ZnO x S 1 x thin fi lm was prepared using spercycle at omic layer deposition. The S composition shows strong negative deviation from the H 2 O/H 2 S pulse ratio, indicating strong exchange reaction of O for S. The H 2 O/H 2 S pulse ratio shows strong effect on the morphology and the structure of the thin film. Elongat ed and round shape grains were observed for H 2 O pulse rich and H 2 S pulse rich film. Both film structure and S composition affect the bandgap properties of the thin film and a bowl shaped relation was mapped out between the bandgap and sulfur composition. Maximum complete cell efficiency 11.58% was measured with the cell using ZnO x S 1 x thin film with 1/10 H 2 O/H 2 S pulse ratio. Compared to the cell using CdS, the relatively lower fill factor is primarily responsible for the reduced performance, which may be c aused by parasitic resistances induced by recombination at the absorber/buffer layer interface or the conduction bandgap offset. Reducing the buffer layer thickness to 30 nm can further increase the cell efficiency due to the increasing transmission betwee n 500 800 nm. In and Ga were doped in to ZnO thin film using atomic layer deposition The comparable minimum resistivity 1.65 x 10 3 cm was reached using 0.1 and 0.05 precursor pulse fraction for In doped ZnO and Ga doped ZnO, respectively. For the sampl es with minimum resistivity, the In doped ZnO shows high carrier concentration but low mobility while Ga doped ZnO shows high mobility but relatively lower carrier concentration. Based on the characterization using XRD, AFM and SEM, the relatively

PAGE 222

222 low carr ier mobility of In doped ZnO can be attributed to ionized impurity scattering and e lectron electron scattering. The efficiency of complete cell using In doped ZnO and Ga doped ZnO as TCO material was measured as 6.03 % and 11.51 %. The cell with Ga doped ZnO shows better open circuit voltage and short circuit current because the Ga doped ZnO shows better transmission property due to the relatively lower carrier concentration. Also the large bandgap and moderate carrier concentration may lead to a more favorab le band offset at the interface. Suggestions for Future Work Thermodynamics Properties and Phase Relation Measurement More experiment s on the thermodynamic properties for the higher order system can be done in order to further improve the accuracy of the database. For example, a s stated in Chapter 3 and Chapter 5 some thermodynamic properties of AgIn 5 Se 8 AgCuSe remains unclear and th e Gibbs energy of formation of the compounds AgGaSe 2 Ag 9 GaSe 6 show s large error. Also, there is no isothermal phase relation data available for Ag Cu Se system at low temperature (e.g 500 K). Therefore, more work can be done in terms of thermodynamic properties measurement. D ifferential scanning calorimetry (DSC) and electromotive force measurement (EMF) can be used to measure the h eat capacity and Gibbs energy of the compounds as a function of temperature. Differential thermal analysis (DTA) and microscopic analysis can be used to determine the phase relation for the ternary system at certain temperature. Furthermore, phonon calcula tion is a useful and only tool to calculate the Gibbs energy of the defect termina l in the sublattice model used in this work. With these

PAGE 223

223 experimental data, the accuracy of the thermodynamic database can be fu rther improved. Deposition Condition Post Anne aling and Deposition Method For the buffer layer ZnO x S 1 x thin film the effect of S composition in the buffer layer has been studied. However, the effect of the deposition temperature remains unclear The major problem for the buffer layer is its band ali gnment with the absorber. It is possible to find the most favorable band alignment by varying the deposition temperature since it could affect the film electronic and morphological properties. According to the ALD windows mentioned in Figure 7 3 wide temperature range between 125 C to 150 C can be try for future work. Similarly, In doped ZnO and Ga doped ZnO were studied at 200 C and the effect of the deposition tem perature on the TCO thin film will also be an interesting topic Furthermore, in this study it has been cleared that t he main problem of In doped ZnO lies in the low free carrier mobility and high free carrier concentration. It is possible to improve the e lectronic properties as TCO material by annealing the sample with lower In doping level. This way, the sample may form larger grains and maintain the free carrier concentration at a relatively lower level (~10 20 cm 3 ). In this study, for both buffer layer and TCO layer materials, the supercycle ALD type layered doping. T his feature may lead to longer average free carrier travel path and reduce the mobility. Thus, a nother deposition technique that introducing H 2 O H 2 S or TMIn/TEGa DEZn simultaneously is worth trying because it can lead to more homogeneous doping. This may result in different interface between layers and improve the electric properties for the deposited film which is of vital importance for TCO mater ials.

PAGE 224

224 LIST OF REFERENCES [1] P. Jackson, D. Hariskos, R. Wuerz, O. Kiowski, A. Bauer, T. M. Friedlmeier et al. "Properties of Cu(In,Ga)Se 2 solar cells with new record efficiencies up to 21.7%," physica status solidi (RRL) Rapid Resea rch Letters, vol. in press, 2014. [2] J. Boyle, G. Hanket, and W. Shafarman, "Optical and quantum efficiency analysis of (Ag,Cu)(In,Ga)Se 2 absorber layers," in Photovoltaic Specialists Conference (PVSC), 2009 34th IEEE 2009, pp. 001349 001354. [3] Y. Tauc hi, K. Kim, H. Park, and W. Shafarman, "Characterization of (AgCu)(InGa)Se 2 absorber layer fabricated by a selenization process from metal precursor," IEEE Journal of Photovoltaics, vol. 3, pp. 467 471, 2013. [4] R. Krishnan, D. Wood, V. U. Chaudhari, E. A Payzant, R. Noufi, S. Rozeveld et al. "Reaction routes for the synthesis of CuInSe2 using bilayer compound precursors," Progress in Photovoltaics: Research and Applications, vol. 20, pp. 543 556, 2012. [5] R. Krishnan, G. Tong, W. K. Kim, R. Kaczynski, U. Schoop, E. A. Payzant et al. "Device degradation studies of CIGS solar cells using in situ high temperature X ray diffraction," in Photovoltaic Specialists Conference (PVSC), 2012 38th IEEE 2012, pp. 001970 001973. [6] R. Krishnan, H. Tong, Z. Li, C Muzzillo, W. K. Kim, E. A. Payzant et al. "Effect of Na doped Mo on selenization pathways for CuGa/In metallic precursors," in 2013 IEEE 39th Photovoltaic Specialists Conference (PVSC) 2013, pp. 0392 0397. [7] Z. Li, R. Krishnan, G. Tong, R. Kaczynski U. Schoop, and T. J. Anderson, "The effect of rapid thermal annealing on the performance of CIGS cells with an ITO layer," in 2013 IEEE 39th Photovoltaic Specialists Conference (PVSC) 2013, pp. 1136 1141. [8] J. Koo, S. Jeon, M. Oh, H. i. Cho, C. Son, a nd W. K. Kim, "Optimization of Se layer thickness in Mo/CuGa/In/Se precursor for the formation of Cu(InGa)Se2 by rapid thermal annealing," Thin Solid Films, vol. 535, pp. 148 153, 5/15/ 2013. [9] C. P. Muzzillo, C. E. Campbell, and T. J. Anderson, "Cu Ga I n thermodynamics: experimental study, modeling, and implications for photovoltaics," Journal of Materials Science, vol. 51, pp. 3362 3379, 2015. [10] C. P. Muzzillo, "Chalcopyrites for Solar Cells: Chemical Vapor Deposition, Selenization, and Alloying," Ph D, Chemical Engineering, University of Florida, 2015.

PAGE 225

225 [11] J. C. Mikkelsen, "Ternary phase relations of the chalcopyrite compound CuGaSe2," Journal of Electronic Materials, vol. 10, pp. 541 558, 1981. [12] H. S. Liu, X. J. Liu, Y. Cui, C. P. Wang, I. Ohnum a, R. Kainuma et al. "Thermodynamic assessment of the Cu In binary system," Journal of Phase Equilibria, vol. 23, pp. 409 415, Oct 2002. [13] M. Ider, R. Pankajavalli, W. Zhuang, J. Y. Shen, and T. J. Anderson, "Thermochemistry of the Cu2Se In2Se3 system ," Journal of Alloys and Compounds, vol. 604, pp. 363 372, 8/15/ 2014. [14] J. Shen, W. Kim, S. Shang, M. Chu, S. Cao, and T. Anderson, "Thermodynamic description of the ternary compounds in the Cu In Se system," Rare Metals, vol. 25, pp. 481 487, 2006. [1 5] Z. Du, C. Guo, M. Tao, and C. Li, "Thermodynamic modeling of the Cu Se system," International Journal of Materials Research, vol. 99, pp. 294 300, 2008. [16] M. Ider, R. Pankajavalli, W. Zhuang, J. Y. Shen, and T. J. Anderson, "Thermochemistry of the Ga Se System," ECS Journal of Solid State Science and Technology, vol. 4, pp. Q51 Q60, January 1, 2015 2015. [17] F. Zheng, J. Y. Shen, Y. Q. Liu, W. K. Kim, M. Y. Chu, M. Ider et al. "Thermodynamic optimization of the Ga Se system," Calphad, vol. 32, pp. 432 438, 2008. [18] U. Rau and M. Schmidt, "Electronic properties of ZnO/CdS/Cu(In,Ga)Se2 solar cells aspects of heterojunction formation," Thin Solid Films, vol. 387, pp. 141 146, 5/29/ 2001. [19] B. T. Ahn, L. Larina, K. H. Kim, and S. J. Ahn, "Development of new buffer layers for Cu(In,Ga)Se2 solar cells," Pure and Applied Chemistry, vol. 80, 2008. [20] J. Pettersson, "Modeling Band Gap Gradients and Cd free Buffer Layers in CIGS Solar Cells," Ph. D., Department of Engineering Sciences, Solid State Electronics, Acta Universitatis Upsaliensis, Uppsala, 2012. [21] K. Kim, L. Larina, J. H. Yun, K. H. Yoon, H. Kwon, and B. T. Ahn, "Cd free CIGS solar cells with buffer layer based on the In2S3 derivative s," Phys Chem Chem Phys, vol. 15, pp. 9239 44, Jun 21 2013. [22] J. Lindahl, J. T. Wtjen, A. Hultqvist, T. Ericson, M. Edoff, and T. Trndahl, "The free Cu(In,Ga)Se2solar cells," Progress in Photovoltaics: Research and Applications, vol. 21, pp. 1588 1597, 2013. [23] thin film solar cells," Materials Research Express, vol. 1, p. 016202, 2014.

PAGE 226

226 [24] W. Witte, S. Spieri ng, and D. Hariskos, "Substitution of the CdS buffer layer in CIGS thin film solar cells," Vakuum in Forschung und Praxis, vol. 26, pp. 23 27, 2014. [25] T. Kobayashi, T. Kumazawa, Z. Jehl Li Kao, and T. Nakada, "Cu(In,Ga)Se2 thin film solar cells with a c ombined ALD Zn(O,S) buffer and MOCVD ZnO:B window layers," Solar Energy Materials and Solar Cells, vol. 119, pp. 129 133, 2013. [ 26 ] C. Platzer Bj rkman, T. T rndahl, D. Abou Ras, J. Malmstr m, J. Kessler, and L. Stolt, "Zn(O,S) buffer layers by atomic layer deposition in Cu(In,Ga)Se 2 based thin film solar cells: Band alignment and sulfur gradient," Journal of Applied Physics, vol. 100, p. 044506, 2006. [27] Z. Li, R. Krishnan, G. Tong, R. Kaczynski, U. Schoop, and T. J. Anderson, "The effect of rapid th ermal annealing on the performance of CIGS cells with an ITO layer," in Photovoltaic Specialists Conference (PVSC), 2013 IEEE 39th 2013, pp. 1136 1141. [28] A. Stadler, "Transparent Conducting Oxides An Up To Date Overview," Materials, vol. 5, pp. 661 683 2012. [29] Q. You, H. Cai, Z. Hu, P. Liang, S. Prucnal, S. Zhou et al. "Blue shift in absorption edge and widening of band gap of ZnO by Al doping and Al N co doping," Journal of Alloys and Compounds, vol. 644, pp. 528 533, 9/25/ 2015. [30] Z. Baji, Z. Labadi, M. Fried, K. Vad, J. Toth, and I. Barsony, "Al Doped ALD ZnO for CIGS Buffer Layer," in 26th European Photovoltaic Solar Energy Conference and Exhibition 2011 pp. 2992 2997. [31] W. S. Leung, Y. C. Chan, and S. M. Lui, "A study of degradation o f indium tin oxide thin films on glass for display applications," Microelectronic Engineering, vol. 101, pp. 1 7, 2013. [32] F. Barlow, M. Abdel Naby, A. Joshi, and A. Elshabini Riad, "Comprehensive evaluation of ITO thick films produced under optimum anne aling conditions," Solar Energy Materials and Solar Cells, vol. 33, pp. 63 71, 1994. [33] K. Sreenivas, T. S. Rao, A. Mansingh, and S. Chandra, "Preparation and characterization of rf sputtered indium tin oxide films," Journal of Applied Physics, vol. 57, pp. 384 392, 1985. [34] et al. "A comprehensive review of ZnO materials and devices," Journal of Applied Physics, vol. 98, p. 041301, 2005. [35] G. J. Exarhos, A. Rose, and C. F. Windisc h, "Spectroscopic characterization of processing induced property changes in doped ZnO films," Thin Solid Films, vol. 308 309, pp. 56 62, 1997.

PAGE 227

227 [36] C. E. Kim, P. Moon, S. Kim, J. M. Myoung, H. W. Jang, J. Bang et al. "Effect of carrier concentration on optical bandgap shift in ZnO:Ga thin films," Thin Solid Films, vol. 518, pp. 6304 6307, 9/1/ 2010. [37] Y. Li, Q. Huang, and X. Bi, "Stress dependent properties of Ga doped ZnO thin films prepared by magnetron sputtering," Journal of Materials Science: Mat erials in Electronics, vol. 24, pp. 79 84, 2013. [38] K. Djessas, I. Bouchama, J. L. Gauffier, and Z. B. Ayadi, "Effects of indium concentration on the properties of In doped ZnO films: Applications to silicon wafer solar cells," Thin Solid Films, vol. 555 pp. 28 32, 2014. [39] A. Singh, S. Chaudhary, and D. K. Pandya, "On the temperature dependence of mobility in hydrogenated indium doped ZnO thin films," Acta Materialia, vol. 77, pp. 125 132, 2014. [40] B. Singh and S. Ghosh, "Highly Conducting Transpare nt Indium Doped Zinc Oxide Thin Films," Journal of Electronic Materials, vol. 43, pp. 3217 3221, 2014/09/01 2014. [41] M. C. Jun, S. U. Park, and J. H. Koh, "Comparative studies of Al doped ZnO and Ga doped ZnO transparent conducting oxide thin films," Nan oscale Research Letters, vol. 7, pp. 1 6, 2012. [42] Y. K. Moon, S. H. Kim, D. Y. Moon, W. S. Kim, and J. W. Park, "Enhancement of the Electrical Properties of Ga doped ZnO Thin Films on Polycarbonate Substrates by Using a TiO 2 Buffer Layer," Journal of th e Korean Physical Society, vol. 51, pp. 1732 1735, 2007. [43] J. Nishino, T. Kawarada, S. Ohshio, H. Saitoh, K. Maruyama, and K. Kamata, "Conductive indium doped zinc oxide films prepared by atmospheric pressure chemical vapour deposition," Journal of Mate rials Science Letters, vol. 16, pp. 629 631, Apr 15 1997. [44] A. Behrends, A. Wagner, M. A. M. Al Suleiman, H. J. Lugauer, M. Strassburg, R. Walter et al. "Transparent conductive Ga doped ZnO films fabricated by MOCVD," physica status solidi (a), vol. 2 09, pp. 708 713, 2012. [45] J. Zhong, S. Muthukumar, Y. Chen, Y. Lu, H. M. Ng, W. Jiang et al. "Ga doped ZnO single crystal nanotips grown on fused silica by metalorganic chemical vapor deposition," Applied Physics Letters, vol. 83, pp. 3401 3403, 2003. [46] J. Hu and R. G. Gordon, "Atmospheric pressure chemical vapor deposition of gallium doped zinc oxide thin films from diethyl zinc, water, and triethyl gallium," Journal of Applied Physics, vol. 72, pp. 5381 5392, 1992. [47] S. Major and K. L. Chopra, Indium doped zinc oxide films as transparent electrodes for solar cells," Solar Energy Materials, vol. 17, pp. 319 327, 1988.

PAGE 228

228 [48] A. Hafdallah, F. Yanineb, M. S. Aida, and N. Attaf, "In doped ZnO thin films," Journal of Alloys and Compounds, vol. 509, pp. 7267 7270, 6/30/ 2011. [49] T. P. Rao and M. C. S. Kumar, "Resistivity Stability of Ga Doped ZnO Thin Films with Heat Treatment in Air and Oxygen Atmospheres," Journal of Crystallization Process and Technology, vol. 2, pp. 72 79, 2012. [50] N. Naghavi, L. Dupont, C. Marcel, C. Maugy, B. La k, A. Rougier et al. "Systematic study and performance optimization of transparent conducting indium zinc oxides thin films," Electrochimica Acta, vol. 46, pp. 2007 2013, 2001. [51] J. M. Phillips, R. J. Cava, G. A. T homas, S. A. Carter, J. Kwo, T. Siegrist et al. "Zinc indium oxide: A high conductivity transparent conducting oxide," Applied Physics Letters, vol. 67, pp. 2246 2248, 1995. [52] M. Caglar, S. Ilican, and Y. Caglar, "Influence of dopant concentration on the optical properties of ZnO: In films by sol gel method," Thin Solid Films, vol. 517, pp. 5023 5028, 7/1/ 2009. [53] K. J. Chen, F. Y. Hung, S. J. Chang, and Z. S. Hu, "Microstruc tures, optical and electrical properties of In doped ZnO thin films prepared by sol gel method," Applied Surface Science, vol. 255, pp. 6308 6312, 2009. [54] S. Negi, M. P. S. Rana, S. K. Gautam, R. G. singh, F. Singh, and R. C. Ramola, "Structural and opt ical modification of Ga doped zinc oxide thin films induced by thermal annealing," Indian Journal of Pure & Applied Physics, vol. 54, pp. 236 240, 2016. [55] T. Nam, C. W. Lee, H. J. Kim, and H. Kim, "Growth characteristics and properties of Ga doped ZnO ( GZO) thin films grown by thermal and plasma enhanced atomic layer deposition," Applied Surface Science, vol. 295, pp. 260 265, 3/15/ 2014. [56] W. J. Maeng and J. S. Park, "Growth characteristics and film properties of gallium doped zinc oxide prepared by atomic layer deposition," Journal of Electroceramics, vol. 31, pp. 338 344, 2013. [57] Y. L. Lee, T. H. Huang, C. L. Ho, and M. C. Wu, "The Sandwich Structure of Ga Doped ZnO Thin Films Grown via H2O O2 and O3 Based Atomic Layer Deposition," ECS Journa l of Solid State Science and Technology, vol. 2, pp. Q182 Q186, January 1, 2013 2013. [58] P. R. Chalker, P. A. Marshall, S. Romani, J. W. Roberts, S. J. C. Irvine, D. A. Lamb et al. "Atomic layer deposition of Ga doped ZnO transparent conducting oxide s ubstrates for CdTe based photovoltaics," Journal of Vacuum Science & Technology A, vol. 31, p. 01A120, 2013.

PAGE 229

229 [59] K. Saito, Y. Hiratsuka, A. Omata, H. Makino, S. Kishimoto, T. Yamamoto et al. "Atomic layer deposition and characterization of Ga doped ZnO thin films," Superlattices and Microstructures, vol. 42, pp. 172 175, 7// 2007. [60] H. Chi Ying, Y. Jing Hsung, G. Jyh Rong, L. Dong Yuan, L. Tai Yuan, L. Cheng Tao et al. "Transparent conductive IZO films prepared by atomic layer deposition using DEZn/ TMIn and N 2 O," in Nanoelectronics Conference (INEC), 2010 3rd International 2010, pp. 1337 1338. [61] D. Kim, T. Nam, J. Park, J. Gatineau, and H. Kim, "Growth characteristics and properties of indium oxide and indium doped zinc oxide by atomic layer depo sition," Thin Solid Films, vol. 587, pp. 83 87, 2015. [62] "BP Statistical Review of World Energy," June 2016. [63] M. A. Contreras, B. Egaas, K. Ramanathan, J. Hiltner, A. Swartzlander, F. Hasoon et al. "Progress toward 20% efficiency in Cu(In,Ca)Se(2) polycrystalline thin film solar cells," Progress in Photovoltaics, vol. 7, pp. 311 316, Jul Aug 1999. [64] I. Repins, M. A. Contreras, B. Egaas, C. DeHart, J. Scharf, C. L. Perkins et al. "199% efficient ZnO/CdS/CuInGaSe2 solar cell with 812% fill fact or," Progress in Photovoltaics: Research and Applications, vol. 16, pp. 235 239, 2008. [65] H. Matsuo, K. Yoshino, and T. Ikari, "Characterization of AgGaSe2 thin films grown by post annealing method," Thin Solid Films, vol. 515, pp. 505 508, 2006. [66] J. L. Shay and J.H. Wernick, Ternary chalcopyrite semiconductors: Growth, electronic properties and applications Pergamon: Oxford, 1975. [67] P. Subramanian and J. Perepezko, "The Ag Cu (silver copper) system," Journal of Phase Equilibria, vol. 14, pp. 62 7 5, 1993. [68] V. T. Witusiewicz, U. Hecht, S. G. Fries, and S. Rex, "The Ag Al Cu System. Part 1. Reassessment of the Constituent Binaries on the Basis of New Experimental Data," ChemInform, vol. 36, pp. no no, 2005. [69] M. Bienzle, T. Oishi, F. Sommer, a nd K. Ono, "Thermodynamic Study of the Silver Rich Ag Cu Solid Solution," Materials Transactions Jim, vol. 33, pp. 51 56, Jan 1992. [70] C. T. Heycock and F. H. Neville, "Complete Freezing Point Curves of Binary Alloys Containing Silver or Copper Together with Another Metal," Philosophical Transactions of the Royal Society of London. Series A, Containing Papers of a Mathematical or Physical Character, vol. 189, pp. 25 69, 1897. [71] K. Friedich and A. Leroux, "Copper, Silver, and Lead," Metallurgie, vol. 4, pp. 293 315, 1907.

PAGE 230

230 [72] R. Hirose, "On the Silver Copper Alloys," Rept. Imp. Mint, vol. 1, pp. 1 74, 1927. [73] W. Broniewski and S. Kostacz, "On the alloys of Ag Cu," Compt. Rend., pp. 937 975, 1932. [74] C. S. Smith and W. E. Lindlief, "The Equih'brium Diagram of the Copper Rich Copper Silver Alloys," Trans. AIME, pp. 101 118, 1932. [75] B. Dobovisek and A. Paulin, "Report on Caloric Measurements with the DTA at High Temperatures," MOt Met. Quart, pp. 27 3 2, 1962. [76] A. Rosina, N. Smajic, and B. Dobovisek, "Application of Differential Thermal Analysis in Calorimetry," Mikroochim. Acta, pp. 626 638, 1967. [77] U. V. Choudary and A. Ghosh, "Thermodynamics of Liquid Copper Silver Alloys by a Solid Electrolyt e Cell," J. Electrochem. Soc., pp. 1024 1028, 1970. [78] J. Markali and P. Thoresen, "The Ag Rich Side of the Binary Phase Diagrams Ag/Au and Ag]Cu as Studied by Zone Refining," Acta. Chem. Scand., pp. 31 35, 1961. [79] W. F. Roessner, "Thermo Electric Tem perature Scales," U.S. Bur. Stds. J. Res., pp. 343 358, 1929. [80] D. Stockdale, "The Solid Solutions of the Copper Silver System," J. Inst. Metals, pp. 127 155, 1931. [81] H. Moser, J. Otto, and W. Thomas, "Gasthermometrische Messungen bei hohen Temperatu ren. III," Zeitschrift fr Physik A Hadrons and Nuclei, vol. 175, pp. 327 336, 1963. [82] M. Hansen, "Die Hrte silberreicher Kupfer Silberlegierungen. Bestimmung der Lslichkeit von Kupfer in Silber mit Hilfe von Hrtemessungen," Zeitschrift fr anorganis che und allgemeine Chemie, vol. 186, pp. 41 48, 1929. [83] Z. Bahari, M. Elgadi, J. Rivet, and J. Dugu, "Experimental study of the ternary Ag Cu In phase diagram," Journal of Alloys and Compounds, vol. 477, pp. 152 165, 2009. [84] N. Ageew, M. Hansen, and G. Sachs, "Entmischung und Eigenschaftsnderungen bersttigter Silber Kupferlegierungen," Zeitschrift fr Physik A Hadrons and Nuclei, vol. 66, pp. 350 376, 1930. [85] E. A. Owen and J. Rogers, "X Ray study of Copper SIlver Alloys," J. Inst. Metals, pp. 257 266, 1935. [86] Copper) system," Journal of Phase Equilibria, vol. 1, pp. 41 45, 1980.

PAGE 231

231 [87] E. Schmid and G. Siebel, "Uber die MischkristaUbildung bei einand VielkristaUinem Material," Z. Phys., vol. 85, pp. 36 55, 1932. [88] E. Schmid and G. Siebel, Z. Phys., vol. 85, pp. 41 55, 1933. [89] F. Sommer, "Unpublished data as quoted in 2004 Witusiewicz et al," ed: Max Plank Instiut fr Metallforschung, 2000. [90] R. A. Oriani and W. K. Murphy, "Differential Calorimeter for Heats of Formation. of Solid Alloys. Heats of Formation of Alloys of the Noble Metals," The Journal of Physical Chemistry, vol. 62, pp. 327 331, 1958/03/01 1958. [91] J. Kucera, L. Dvorak, and Z. Kudelasek, "Tensimetric Measurement of Thermodynamic Functions of Silver in Ag Cu Alloys by Method of Isotope Exchange," Ceskoslov. Casopis Fys., pp. 277 284, 1961. [92] R. K. Edwards and J. H. Downing, "The Thermodynamics of the Liquid Solutions in the T riad Cu Ag Au. I. The Cu Ag System," J. Phys. Chem., vol. 60, pp. 108 111, 1956. [93] J. Golonka, J. Botor, and M. Duiat, "Study of Copper Silver Liquid Solutions by Combined Effusion Vaporization and Mass Spectrometry Sensing," Metals Tech., vol. 6, pp. 2 67 272, 1979. [94] O. P. Mohapatra and M. G. Frohberg, "Calculation of Activities of Binary Metallic Systems from Distribution Equilibria Measurements with Fused Salt Mixtures," Z. Metallkd, vol. 65, pp. 58 62, 1974. [95] U. V. Choudary and A. Ghosh, "Ther modynamics of Liquid Copper Silver Alloys by a Solid Electrolyte Cell," Journal of the Electrochemical Society, vol. 117, pp. 1024 &, 1970. [96] S. Wagner, G. Sodeck, and A. Neckel, "Thermodynamic Excess Quantities of Liquid Binary Silver copper by Mass Sp ectrometry," High Temp. Sci., vol. 3, pp. 481 490, 1971. [97] S. M. Howard, "Direct Activity Measurements in Liquid Ag Cu Alloys," Metall. Trans. B, vol. 20, pp. 845 852, 1989. [98] R. Hultgren, P. D. Desai, D. T. Hawkins, M. Gleiser, and K. K. Kelley, Sel ected Values of the Thermodynamic Properties of Binary Alloys Metals Park, Ohio: American Society for Metals, 1973. [99] K. Fitzner, Q. Guo, J. Wang, and O. J. Kleppa, "Enthalpies of liquid liquid mixing in the systems Cu Ag, Cu Au and Ag Au by using an i n situ mixing device in a high temperature single unit differential calorimeter," Journal of Alloys and Compounds, vol. 291, pp. 190 200, 1999.

PAGE 232

232 [100] J. Murray, "Calculations of Stable and Metastable Equilibrium Diagrams of the Ag Cu and Cd Zn Systems," Me tallurgical and Materials Transactions A, vol. 15, pp. 261 268, 1984. [101] F. H. Hayes, H. L. Lukas, G. Effenberg, and G. Petzow, "A Thermodynamic Optimisation of the Cu Ag Pb System," Zeitschrift fr Metallkunde, vol. 77, pp. 749 754, 1986 NOV 1986. [102] M. S. S. Lim, J. E. Tibballs, and P. L. Rossiter, "An assessment of thermodynamic equilibria in the Ag Al Cu Mg quaternary system in relation to precipitation reactions," Zeitschrift Fur Metallkunde, vol. 88, pp. 236 245, Mar 1997. [103] A. Kusoffsky "Thermodynamic evaluation of the ternary Ag Au Cu system including a short range order description," Acta Materialia, vol. 50, pp. 5139 5145, 2002. [104] X. C. He, H. Wang, H. S. Liu, and Z. P. Jin, "Thermodynamic description of the Cu Ag Zr system," Cal phad, vol. 30, pp. 367 374, 2006. [105] A. T. Dinsdale, "SGTE DATA FOR PURE ELEMENTS," CALPHAD, vol. 15, pp. 317 425, 1991. [106] W. Hume Rothery, "XVIII. The lattice constants of the elements," Philosophical Magazine Series 7, vol. 10, pp. 217 244, 1930/0 8/01 1930. [107] K. Maraun and A. Yamiya, "Solar Frontier Achieves World Record Thin Film Solar Cell Efficiency: 22.3%," in Record cell independently verified by the Fraunhofer Institute ed: Showa Shell Sekiyu K.K., 2015. [108] W. Shockley and H. J. Queis ser, "Detailed Balance Limit of Efficiency of p n Junction Solar Cells," Journal of Applied Physics, vol. 32, pp. 510 519, 1961. [109] J. P. Ge, S. Xu, L. P. Liu, and Y. D. Li, "A Positive Microemulsion Method for Preparing Nearly Uniform Ag2Se Nanoparticl es at Low Temperature," Chemistry A European Journal, vol. 12, pp. 3672 3677, 2006. [110] Y. J. Glanville, D. G. Narehood, P. E. Sokol, A. Amma, and T. Mallouk, "Preparation and synthesis of Ag2Se nanowires produced by template directed synthesis," Journ al of Materials Chemistry, vol. 12, pp. 2433 2434, 2002. [111] S. Y. Zhang, C. X. Fang, Wei, B. K. Jin, Y. P. Tian, Y. H. Shen et al. "Synthesis and Electrochemical Behavior of Crystalline Ag2Se Nanotubes," The Journal of Physical Chemistry C, vol. 111, pp. 4168 4174, 2007/03/01 2007. [112] B. Gates, Y. Wu, Y. Yin, P. Yang, and Y. Xia, "Single Crystalline Nanowires of Ag2Se Can Be Synthesized by Templating against Nanowires of Trigonal Se,"

PAGE 233

233 Journal of the American Chemical Society, vol. 123, pp. 11500 115 01, 2001/11/01 2001. [113] R. Hultgren, P. D. Desai, D. T. Hawkins, M. Gleiser, and K. K. Kelley, "Ag Se," in Selected Values of the Thermodynamic Properties of Binary Alloys ed Metals Park, Ohio, 1973, pp. 95 98. [114] A. K. Shukla, P. Sen, and D. D. Sar ma, "Transformations in Ag S and Ag Se: A Calorimetric Study," Berichte der Bunsengesellschaft fr physikalische Chemie, vol. 86, pp. 198 202, 1982. [115] U. V. Oehsen and H. Schmalzried, "Thermodynamic investigations of Ag2Se," Berichte der Bunsengese llschaft fr physikalische Chemie, vol. 85, pp. 7 14, 1981. [116] B. Predel, "Ag Se (Silver Selenium)," in Landolt Brnstein Group IV Physical Chemistry vol. 5a, O. Madelung, Ed., ed: Springer Verlag Berlin Heidelberg, 1991, pp. 1 4. [117] I. Karakaya a nd W. T. Thompson, "The Ag Se (Silver Selenium) system," Bulletin of Alloy Phase Diagrams, vol. 11, pp. 266 271, 1990/06/01 1990. [118] B. Predel, "Ag Se (Silver Selenium)," in Ac Ag ... Au Zr vol. 12A, B. Predel, Ed., ed: Springer Berlin Heidelberg, 20 06, pp. 1 2. [119] V. Tomashik, "Ag Se Binary Phase Diagram Evaluation," in Binary Evaluations G. Effenberg, Ed., ed. Stuttgart: MSI, Materials Science International Services GmbH, 2002. [120] P. E. Tomaszewski, "Structural phase transitions in crystals. I. Database," Phase Transitions, vol. 38, pp. 127 220, 1992/05/01 1992. [121] H. L. Tuller, M. Balkanski, and T. Takahashi, Solid State Ionics : Elsevier Science, 2012. [122] K. Friedrich and A. Leroux, "The Fusion Diagram of Binary System Cu Cu2Se, Ag Ag2S e and Pb Pb2Se," Metallurgie, vol. 5, pp. 355 358, 1908 [123] G. Pellini, "Le combinazioni del selenio con l'argento," Gazzetta chimica Italiana, vol. 45, pp. 533 539, 1915. [124] D. Houphouet Boigny, "Study of the Ternary Diagram Ag -As -Se. I. -Descript ion of the Triangle Se -Ag 2 Se -As 2 Se 3 ," Journal of Alloys and Compounds, vol. 98, p. 11, 1984. [125] S. Ohno, A. C. Barnes, and J. E. Enderby, "The electronic properties of liquid Ag 1 x Se x," Journal of Physics: Condensed Matter, vol. 6, p. 5335, 1994.

PAGE 234

234 [ 126] Y. Tsuchiya, "Velocity of sound and high ray attenuation in liquid Ag Se alloys," Journal of Physics: Condensed Matter, vol. 8, p. 1897, 1996. [127] S. M. Wiggett, D. R. Swinbourne, R. H. Eric, and G. G. Barbante, "Thermodynamics of the mis cibility gap in the Ag Se system," Metallurgical and Materials Transactions B, vol. 30, pp. 589 595, 1999/08/01 1999. [128] K. Friedrich and A. Leroux, "The Fusion Diagram of Binary System Cu Cu2Se, Ag Ag2Se and Pb Pb2Se," Metallurgie, pp. 355 358, 1908. [ 129] B. M. Kulwicki, "The Phase Equilibria of Some Compound Semiconductors by Differential Thermal Analysis (DTA) Calorimetry," PhD, Department of Chemical and Metallurgical Engineering, The University of Michigan, 1963. [130] M. Voronin and E. Osadchii, "Determination of thermodynamic properties of silver selenide by the galvanic cell method with solid and liquid electrolytes," Russian Journal of Electrochemistry, vol. 47, pp. 420 426, 2011. [131] E. G. Osadchii and E. A. Echmaeva, "The system Ag Au Se: Phase relations below 405 K and determination of standard thermodynamic properties of selenides by solid state galvanic cell technique," American Mineralogist, vol. 92, pp. 640 647, Apr 2007. [132] A Nasar and M. Shamsuddin, "Studies on the thermodynamic stability of silver selenide," Metallurgical and Materials Transactions B, vol. 28, pp. 519 522, 1997/06/01 1997. [133] D. Feng, P. Taskinen, and F. Tesfaye, "Thermodynamic stability of Ag2Se from 35 0 to 500K by a solid state galvanic cell," Solid State Ionics, vol. 231, pp. 1 4, 2013. [134] Y. G. Asadov, Y. I. Aliyev, and A. G. Babaev, "Polymorphic transformations in Cu2Se, Ag2Se, AgCuSe and the role of partial cation cation and anion anion replaceme nt in stabilizing their modifications," Physics of Particles and Nuclei, vol. 46, pp. 452 474, 2015. [135] F. Grnvold, S. Stlen, and Y. Semenov, "Heat capacity and thermodynamic properties of silver(I) selenide, oP Ag2Se from 300 to 406 K and of cI Ag2Se from 406 to 900 K: transitional behavior and formation properties," Thermochimica Acta, vol. 399, pp. 213 224, 2003. [136] Olin, B. Nolng, L. O. hman, E. Osadchii, and E. Rosn, "Chemical thermodynamics of selenium," 2005. [137] R. Blachnik and G. Gu nia Paul, "Enthalpien von Kupfer und Silberchalkogeniden / Enthalpies of Copper and Silver Chalcogenides," in Zeitschrift fr Naturforschung A vol. 33, ed, 1978, p. 190.

PAGE 235

235 [138] Y. Baer, G. Busch, C. Frhlich, and E. Steigmeier, "Wrmeleitfhigkeit, elektri sche Leitfhigkeit, Hall Effekt, Thermospannung und spezifische Wrme von Ag2Se," in Zeitschrift fr Naturforschung A vol. 17, ed, 1962, p. 886. [139] R. Blachnik and G. Bolte, "Aktivitten von Se in geschmolzenen Cu2Se und Ag2Se Mischungen," Journal of th e Less Common Metals, vol. 57, pp. 21 28, 1978. [140] C. h. Chang, "Processing and characterization of copper indium selenide for photovoltaic applications," 9956558 Ph.D., University of Florida, Ann Arbor, 1999. [141] J. Yu and H. Yun, "Reinvestigation of the low temperature form of Ag(2)Se (naumannite) based on single crystal data," Acta Crystallographica Section E: Structure Reports Online, vol. 67, pp. i45 i45, 07/15 2011. [142] G. Kresse and J. Furthmuller, "Efficient iterative schemes for ab initio to tal energy calculations using a plane wave basis set," Physical Review B, vol. 54, pp. 11169 11186, Oct 15 1996. [143] G. Kresse and J. Furthmuller, "Efficiency of ab initio total energy calculations for metals and semiconductors using a plane wave basis s et," Computational Materials Science, vol. 6, pp. 15 50, Jul 1996. [144] G. Kresse and D. Joubert, "From ultrasoft pseudopotentials to the projector augmented wave method," Physical Review B, vol. 59, pp. 1758 1775, Jan 15 1999. [145] J. P. Perdew, K. Burk e, and M. Ernzerhof, "Generalized gradient approximation made simple," Physical Review Letters, vol. 77, pp. 3865 3868, Oct 28 1996. [146] S. Grimme, J. Antony, S. Ehrlich, and H. Krieg, "A consistent and accurate ab initio parametrization of density funct ional dispersion correction (DFT D) for the 94 elements H Pu," Journal of Chemical Physics, vol. 132, p. 154104, Apr 21 2010. [147] J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuseria, L. A. Constantin et al. "Restoring the density gr adient expansion for exchange in solids and surfaces," Physical Review Letters, vol. 100, p. 136406, 2008. [148] S. L. Shang, Y. Wang, P. W. Guan, W. Y. Wang, H. Z. Fang, T. Anderson et al. "Insight into structural, elastic, phonon, and thermodynamic pro perties of alpha sulfur and energy related sulfides: a comprehensive first principles study," Journal of Materials Chemistry A, vol. 3, pp. 8002 8014, 2015 2015. [149] S. L. Shang, Y. Wang, D. Kim, and Z. K. Liu, "First principles thermodynamics from phono n and Debye model: Application to Ni and Ni 3 Al," Computational Materials Science, vol. 47, pp. 1040 1048, Feb 2010.

PAGE 236

236 [150] M. Oliveria, R. K. McMullan, and B. J. Wuensch, "Single crystal neutron diffraction analysis of the cation distribution in the high te Ag2Se," Solid State Ionics, vol. 28, pp. 1332 1337, 1988/09/01 1988. [151] R. Blachnik and G. Bolte, "Aktivitten von Se in geschmolzenen Cu Se und Ag Se Mischungen," Journal of the Less Common Metals, vol. 57, p p. 21 28, 1978. [152] M. H. Pelabon and M. H. Moissan, "Sur le sulfure, Le seleniure et le tellurure d'argent," C. R. Acad. Sc. Paris, vol. 143, pp. 294 296, 1906. [153] N. Valverde, "Ionen und Elektronen Bewegung in festem Silbersulfid, Silberselenid und Silbertellurid," Zeitschrift fr Physikalische Chemie, vol. 70, pp. 128 138, 1970. [154] I. Rom and W. Sitte, "Composition dependent ionic and electronic conductivities and chemical diffusion coefficient of silver selenide at 160 o C," Solid State Ionics, v ol. 101 103, pp. 381 386, 1997. [155] G. Pellini, "Compound of Selenium with Silver," Gazz. Chim. Ital, vol. 45, pp. 533 539, 1915. [156] F. D. Rossini, D. D. Wagman, W. H. Evans, S. Levine, and I. Jaffe, Selected values of chemical thermodynamic propertie s Washington, DC: U.S. Government Printing Office, 1952. [157] K. C. Mills, Thermodynamic Data for Inorganic Sulphides, Selenides and Tellurides : University Microfilms, 1991. [158] D. Houphouet Boigny, R. Eholi, R. Ollitrault Fichet, and J. Flahaut, "tude du diagramme ternaire Ag As Se I: Description du triangle Se Ag2Se As2Se3," Journal of the Less Common Metals, vol. 98, pp. 11 31, 1984. [159] M. A. Green, K. Emery, Y. Hishikawa, W. Warta, and E. D. Dunlop, "Solar cell efficiency tables (version 47)," Progress in Photovoltaics: Research and Applications, vol. 24, pp. 3 11, 2016. [160] R. Hultgren, P. D. Desai, D. T. Hawkins, M. Gleiser, and K. K. Kelley, "Ag Cu," in Selected Value s of the Thermodynamic Properties of Binary Alloys ed Metals Park, Ohio: American Society for Metals, 1973, pp. 44 49. [161] V. M. Glazov, A. S. Pashinkin, and V. A. Fedorov, "Phase equilibria in the Cu Se system," Inorganic Materials, vol. 36, pp. 641 65 2, 2000. [162] J. Y. Shen, W. K. Kim, C. H. Chang, S. J. Wilson, and T. J. Anderson, "Thermodynamic optimization of the Cu Se binary system," ed, 2016.

PAGE 237

237 [163] S. w. Chen, J. s. Chang, S. m. Tseng, L. c. Chang, and J. y. Lin, "Phase diagrams of the Ag In Se photovoltaic material system," Journal of Alloys and Compounds, vol. 656, pp. 58 66, 1/25/ 2016. [164] M. I. Agaev, S. M. Alekperova, and M. I. Zargarova, "Physicochemical Study of Ag2X Cu2X (X = S, Se) Systems," Dokl. Akad. Nauk Az. SSR, vol. 27, pp. 20 2 3, 1971. [165] Schafer, "Experimental investigation in the Cu Ag S Se system," Neues Jahrbuch fur Mineralogie Abhandlungen, vol. 169, pp. 301 304, 1995. [166] L. Chen and W. Shafarman, "Ag content induced secondary phase formation in (Ag,Cu)(In,Ga)Se 2 thin film grown by three sttage co evaporation," ed, 2016. [167] N. Asano, K. Wase, and T. Nomura, "Equilibrium Diagram of the system Copper Silver Selenium and the Microscopic Structure of Copper Containing Silver and Selenium," Journal of the Mining and Meta llurgical Institute of Japan, vol. 87, pp. 485 490, 1971/01/04 1971. [168] J. W. Earley, "Description and Synthesis of the Selenide Minerals Eucairite, AgCuSe," American Mineralogist, vol. 35, pp. 337 364, 1950 [169] A. J. Frueh, G. K. Czamanske, and C. K night, "The Crystallography of Eucairite,CuAgSe," Zeitschrift fr Kristallographie, vol. 108, pp. 389 396, 1957. [170] S. Y. Miyatani, Y. Miura, and H. Ando, "Mixed Conduction in AgCuSe," Journal of the Physical Society of Japan, vol. 46, pp. 1825 1832, 19 79. [171] S. N. Mostafa, S. R. Selim, and F. M. Ismail, "X Ray Diffraction and Electrical Conductivity Studies on Ag Cu Se System," Journal of Materials Science & Technology, vol. 11, pp. 53 55, 1995. [172] P. Villars, K. Cenzual, R. Gladyshevskii, O. Shch erban, V. Dubenskyy, V. Kuprysyuk et al. "CuAgSe," in Structure Types. Part 11: Space groups (135) P42/mbc (123) P4/mmm vol. 43A11, P. Villars and K. Cenzual, Eds., ed: Springer Berlin Heidelberg, 2012, pp. 252 252. [173] A. J. Hong, L. Li, H. X. Zhu, X. H. Zhou, Q. Y. He, W. S. Liu et al. "Anomalous transport and thermoelectric performances of CuAgSe compounds," Solid State Ionics, vol. 261, pp. 21 25, 8// 2014. [174] X. Wang, P. Qiu, T. Zhang, D. Ren, L. Wu, X. Shi et al. "Compound defects and th ermoelectric properties in ternary CuAgSe based materials," Journal of Materials Chemistry A, vol. 3, pp. 13662 13670, 2015. [175] S. y. Miyatani, "Electronic and Ionic Conduction in (AgxCu1 x)2Se," Journal of the Physical Society of Japan, vol. 34, pp. 42 3 432, 1973.

PAGE 238

238 [176] K. Chrissafis, N. Vouroutzis, K. M. Paraskevopoulos, N. Frangis, and C. Manolikas, "Phase transformation in CuAgSe: a DSC and electron diffraction examination," Journal of Alloys and Compounds, vol. 385, pp. 169 172, 12/28/ 2004. [177] D M. Trots, A. N. Skomorokhov, M. Knapp, and H. Fuess, "High temperature behaviour of average structure and vibrational density of states in the ternary superionic compound AgCuSe," The European Physical Journal B, vol. 51, pp. 507 512, 2006. [178] M. K. B alapanov, R. K. Ishembetov, Y. K. Yulaeva, and R. A. Yakshibaev, "Thermodynamic properties of solid solutions of superionic copper, silver, and lithium chalcogenides," Russian Journal of Electrochemistry, vol. 47, pp. 1337 1342, 2011/12/01 2011. [179] J. O Andersson, T. Helander, L. Hglund, P. Shi, and B. Sundman, "Thermo Calc & DICTRA, computational tools for materials science," Calphad, vol. 26, pp. 273 312, 6// 2002. [180] L. S. Palatnick and E. K. Belova, Izv. Akad, Nauk, SSSR Neorg. Mater, vol. 1, p. 843, 1965. [181] R. S. Feigelson, "The Growth of Ternary Semiconductor Crystals Suitable for Device Applications," Le Journal de Physique Colloques, vol. 36, pp. C3 57 C3 66, 1975. [182] J. C. Mikkelsen, "Ag2se Ga2se3 Pseudobinary Phase Diagram," Material s Research Bulletin, vol. 12, pp. 497 502, 1977. [183] P. A., "Silver Galium Selenium," in Ternary Alloys -a Comprehensive Compendium of Evaluated Constitutional Data and Phase Diagrams vol. 2, G. Petzow and G. Effenberg, Eds., ed: VCH Publishers, 1988, p p. 150 152. [184] S. F. Zhu, J. Liu, B. J. Zhao, H. A. Jiang, and Z. H. Li, "Crystal Growth and Differential Thermal Analysis of AgGaSe2," Crystal Research and Technology, vol. 30, pp. 1165 1168, 1995. [185] V. B. Valerii, B. L. Vladimir, V. L. Panyutin, A R. Evgenii, G. S. Shevyrdyaeva, and O. B. Shcherbina, "Growth and optical properties of nonlinear silver selenogallate single crystals," Soviet Journal of Quantum Electronics, vol. 22, p. 722, 1992. [186] M. B. Babanly, G. I. Ibragimova, N. B. Babanly, and Y. M. Shikhiev, "The Ag2Se Ga2Se3 Se system," Russian Journal of Inorganic Chemistry, vol. 46, pp. 1589 1591, Oct 2001. [187] H. Neumann, G. Khn, and W. Mller, "High temperature specific heat of AgInS2 and AgGaSe2," Crystal Research and Technology, vol. 20, pp. 1225 1229, 1985.

PAGE 239

239 [188] W. Mller, G. Khn, and H. Neumann, "High temperature specific heats of chalcopyrite type semi conductors," Thermochimica Acta, vol. 93, pp. 669 672, 9/15/ 1985. [189] M Babanly, N. Babanly, and Y. Yusibov, The EMF method with solid state electrolyte in the thermodynamic investigation of ternary copper and silver chalcogenides : INTECH Open Access Publisher, 2011. [190] M. Hiroaki, E. Saburo, and I. Taizo, "Thermodynamica l Properties of I III VI 2 Group Chalcopyrite Semiconductors," Japanese Journal of Applied Physics, vol. 30, p. 1181, 1991. [191] R. Schmid Fetzer, "Ag In Se Ternary Phase Diagram Evaluation," in Ternary Evaluations G. Effenberg, Ed., ed. Stuttgart: MSI, Materials Science International Services GmbH, 1988. [192] L. Peraldo Bicelli, "Thermodynamic evaluation of the n AgIn5Se8 and n CuIn5S8 stability in photoelectrochemical cells," Solar Energy Materials, vol. 15, pp. 77 98, 1987/02/01 1987. [193] L. Perald o Bicelli, "Thermodynamic stability of silver indium selenide (n AgInSe2) in photoelectrochemical cells," The Journal of Physical Chemistry, vol. 92, pp. 6991 6997, 1988/12/01 1988. [194] A. Aresti, L. Garbato, and A. Rucci, "Some cohesive energy features of tetrahedral semiconductors," Journal of Physics and Chemistry of Solids, vol. 45, pp. 361 365, 1984/01/01 1984. [195] A. V. Kopytov and A. V. Kosobutsky, "Thermodynamic and elastic properties of AgInSe2 and AgInTe2," Physics of the Solid State, vol. 52, pp. 1359 1361, 2010. [196] S. Sharma, A. S. Verma, and V. K. Jindal, "First principles studies of structural, electronic, optical, elastic and thermal properties of Ag chalcopyrites (AgInX2: X=S, Se)," Physica B Condensed Matter, vol. 438, pp. 97 108, Apr 1 2014. [197] M. Y. Rigan, T. V.I., S. N.P., and L. G. Novikova, "PREPARATION, FORMATION REGION, AND SOME PROPERTIES OF AgInSe2," Vysokochistye Veshchestva, pp. 172 176, 1991. [198] I. D. Olekseyuk and O. V. Krykhovets, "The Ag2Se In2Se3 SnSe2 system," Jo urnal of Alloys and Compounds, vol. 316, pp. 193 202, 2001. [199] I. V. Bodnar, "Properties of AgGa x In 1 x Se 2 Solid Solutions," Inorganic Materials, vol. 40, pp. 914 918, 2004. [200] C. H. Chang, A. Davydov, B. J. Stanbery, and T. J. Anderson, "Thermodynami c assessment of the Cu In Se system and application to thin film photovoltaics," in

PAGE 240

240 Photovoltaic Specialists Conference, 1996., Conference Record of the Twenty Fifth IEEE Washington, DC, 1996, pp. 849 852. [201] "SGTE substances database (SUB94)," SGTE, E d., ed, 1993. [202] S. Sharma, A. S. Verma, and V. K. Jindal, "Ab initio studies of structural, electronic, optical, elastic and thermal properties of silver gallium dichalcogenides (AgGaX2: X = S, Se, Te)," Materials Research Bulletin, vol. 53, pp. 218 23 3, 5// 2014. [203] P. Jackson, D. Hariskos, E. Lotter, S. Paetel, R. Wuerz, R. Menner et al. "New world record efficiency for Cu(In,Ga)Se 2 thin film solar cells beyond 20%," Progress in Photovoltaics: Research and Applications, vol. 19, pp. 894 897, 2011 [204] J. Pern, Stability issues of transparent conducting oxides (TCOs) for thin film photovoltaics [electronic resource] / John Pern Golden, Colo. :: National Renewable Energy Laboratory, 2008. [205] K. L. Chopra, P. D. Paulson, and V. Dutta, "Thin fil m solar cells: an overview," Progress in Photovoltaics: Research and Applications, vol. 12, pp. 69 92, 2004. [206] R. Sundaramoorthy, F. J. Pern, C. DeHart, T. Gennett, F. Y. Meng, M. Contreras et al. "Stability of TCO window layers for thin film CIGS so lar cells upon damp heat exposures: part II," 2009, p. 74120J. [207] T. Nakada, "Nano structural investigations on Cd doping into Cu(In,Ga)Se 2 thin films by chemical bath deposition process," Thin Solid Films, vol. 361, pp. 346 352, Feb 21 2000. [208] T. Nakada and A. Kunioka, "Direct evidence of Cd diffusion into Cu(In, Ga)Se[sub 2] thin films during chemical bath deposition process of CdS films," Applied Physics Letters, vol. 74, pp. 2444 2446, 1999. [209] A. Martel, F. Caballero Briones, A. I. Oliva, R. Castro Rodriguez, A. Iribarren, P. Bartolo Perez et al. "Depth profiling study of the CdTe/CdS/ITO/glass heterostructure with AES and GIXRD," Physica Status Solidi B Basic Research, vol. 220, pp. 261 267, Jul 2000. [210] A. Ordaz Flores, P. Bartolo Pere z, R. Castro Rodriguez, and A. I. Oliva, "Annealing effects on the mass diffusion of the CdS/ITO interface deposited by chemical bath deposition," Revista mexicana de fsica, vol. 52, pp. 15 19, 02/01 2006. [211] A. Pokaipisit, M. Horprathum, and P. Limsuw an, "Influence of Annealing Temperature on the Properties of ITO Films Prepared by Electron Beam Evaporation and Ion Assisted Deposition," Nat. Sci., pp. 362 366, 2008.

PAGE 241

241 [212] H. C. Lee, "The behaviors of the carrier concentrations and mobilities in indium tin oxide thin films by DC and RF superimposed DC reactive magnetron sputtering at the various process temperatures," Applied Surface Science, vol. 252, pp. 2647 2656, 2006. [ 213] S. Lany and A. Zunger, "Light and bias induced metastabilities in Cu(In,Ga)Se 2 based solar cells caused by the (V Se V Cu ) vacancy complex," Journal of Applied Physics, vol. 100, p. 113725, 2006. [214] A. P. Kumar and K. V. Reddy, "Cadmium diffusion in CIS thin films," Semiconductor Science and Technology, vol. 12, p. 966, 1997. [215] M. A. Contreras, M. J. Romero, B. To, F. Hasoon, R. Noufi, S. Ward et al. "Optimization of CBD CdS process in high effi ciency Cu(In,Ga)Se2 based solar cells," Thin Solid Films, vol. 403 404, pp. 204 211, 2002. [216] D. H. Cho, Y. D. Chung, K. S. Lee, N. M. Park, K. H. Kim, H. W. Choi et al. "Influence of growth temperature of transparent conducting oxide layer on Cu(In,G a) Se 2 thin film solar cells," Thin Solid Films, vol. 520, pp. 2115 2118, Jan 1 2012. [217] X. Wang, S. S. Li, W. K. Kim, S. Yoon, V. Craciun, J. M. Howard et al. "Investigation of rapid thermal annealing on Cu(In,Ga)Se2 films and solar cells," Solar En ergy Materials and Solar Cells, vol. 90, pp. 2855 2866, 2006. [218] R. X. Wang, C. D. Beling, S. Fung, A. B. Djurisic, C. C. Ling, and S. Li, "Influence of gaseous annealing environment on the properties of indium tin oxide thin films," Journal of Applied Physics, vol. 97, p. 033504, 2005. [219] H. N. Cui and S. Q. Xi, "The fabrication of dipped CdS and sputtered ITO thin films for photovoltaic solar cells," Thin Solid Films, vol. 288, pp. 325 329, Nov 15 1996. [220] S. M. George, "Atomic Layer Deposition a n Over View," Chemical reviews, vol. 110, pp. 111 131, 2009. [221] J. R. Bakke, J. S. King, H. J. Jung, R. Sinclair, and S. F. Bent, "Atomic layer deposition of ZnS via in situ production of H 2 S," Thin Solid Films, vol. 518, pp. 5400 5408, 2010. [222] G. S tuyven, P. De Visschere, A. Hikavyy, and K. Neyts, "Atomic layer deposition of ZnS thin films based on diethyl zinc and hydrogen sulfide," Journal of Crystal Growth, vol. 234, pp. 690 698, Feb 2002. [223] A. Blackman and L. Gahan, Aylward and Findlay's SI Chemical Data : John Wiley & Sons Australia Ltd, 2013.

PAGE 242

242 [224] B. W. Sanders and A. Kitai, "Zinc Oxysulfide Thin Films Grown by Atomic Layer Deposition," Chemistry of Materials, vol. 4, pp. 1005 1011, Sep Oct 1992. [225] N. Ghobadi, "Band gap determination us ing absorption spectrum fitting procedure," International Nano Letters, vol. 3, p. 2, 2013. [226] N. Chopra, A. Mansingh, and G. K. Chadha, "Electrical, Optical and Structural Properties of Amorphous V2o5 Teo2 Blown Films," Journal of Non Crystalline Solid s, vol. 126, pp. 194 201, Dec 1990. [227] M. M. Ba Abbad, A. A. H. Kadhum, A. B. Mohamad, M. S. Takriff, and K. Sopian, "Visible light photocatalytic activity of Fe3+ doped ZnO nanoparticle prepared via sol gel technique," Chemosphere, vol. 91, pp. 1604 16 11, 6// 2013. [228] H. C. Warad, S. C. Ghosh, B. Hemtanon, C. Thanachayanont, and J. Dutta, "Luminescent nanoparticles of Mn doped ZnS passivated with sodium hexametaphosphate," Science and Technology of Advanced Materials, vol. 6, pp. 296 301, 4// 2005. [ 229] X. F. Fan, Z. X. Shen, Y. M. Lu, and J. L. Kuo, "A theoretical study of thermal New Journal of Physics, vol. 11, p. 093008, 2009. [230] M. A. Contreras, B. Egaas, K. Ramanathan, J. Hiltner, A. Swartzlander, F. Hasoon et al. "Progress toward 20% efficiency in Cu(In,Ga)Se2 polycrystalline thin film solar cells," Progress in Photovoltaics: Research and Applications, vol. 7, pp. 311 316, 1999. [231] A. A. Van Gorkum, K. Nakagawa, a nd Y. Shiraki, "Atomic Layer Doping (ALD) technology in Si and its application to a new structure FET," Journal of Crystal Growth, vol. 95, pp. 480 483, 1989/02/02 1989. [232] R. W. Hoffman, "Mechanical Properties of Non Metallic Thin Films," in Physics of Nonmetallic Thin Films vol. 14, A. C. C. H. S. Dupuy, Ed., ed: Springer US, 1976, p. 273. [233] S. Maniv, W. D. Westwood, and E. Colombini, "Pressure and angle of incidence effects in reactive planar magnetron sputtered ZnO layers," Journal of Vacuum Sci ence & Technology, vol. 20, pp. 162 170, 1982. [234] Z. Baji, Z. Labadi, Z. E. Horvath, G. Molnar, J. Volk, I. Barsony et al. "Nucleation and Growth Modes of ALD ZnO," Crystal Growth & Design, vol. 12, pp. 5615 5620, Nov 2012. [235] R. W. Johnson, A. Hul tqvist, and S. F. Bent, "A brief review of atomic layer deposition: from fundamentals to applications," Materials Today, vol. 17, pp. 236 246, 2014.

PAGE 243

243 [236] Y. Igasaki and H. Saito, "The effects of zinc diffusion on the electrical and optical properties of Z nO:Al films prepared by r.f. reactive sputtering," Thin Solid Films, vol. 199, pp. 223 230, 1991/04/15 1991. [237] Z. M. Gibbs, "Band Engineering in Thermoelectric Materials Using Optical, Electronic, and Ab Initio Computed Properties," California Institut e of Technology, 2015. [238] A. Wang, T. Chen, S. Lu, Z. Wu, Y. Li, H. Chen et al. "Effects of doping and annealing on properties of ZnO films grown by atomic layer deposition," Nanoscale Research Letters, vol. 10, p. 75, 2015. [239] M. K. Debanath and S Karmakar, "Study of blueshift of optical band gap in zinc oxide (ZnO) nanoparticles prepared by low temperature wet chemical method," Materials Letters, vol. 111, pp. 116 119, 11/15/ 2013. [240] L. Li, W. Wei, and M. Behrens, "Synthesis and characterizat Ga2o3 prepared from aqueous solutions by controlled precipitation," Solid State Sciences, vol. 14, pp. 971 981, 7// 2012. [241] P. D. C. King, T. D. Veal, F. Fuchs, C. Y. Wang, D. J. Payne, A. Bourlange et al. "Band gap, electronic s tructure, and surface electron accumulation of cubic and rhombohedral ${ \ text{In}}_{2}{ \ text{O}}_{3}$," Physical Review B, vol. 79, p. 205211, 05/29/ 2009. [242] E. Burstein, "Anomalous Optical Absorption Limit in InSb," Physical Review, vol. 93, pp. 632 633, 02/01/ 1954. [243] S. Y. Bae, H. C. Choi, C. W. Na, and J. Park, "Influence of In incorporation on the electronic structure of ZnO nanowires," Applied Physics Letters, vol. 86, p. 033102, 2005. [244] J. Jie, G. Wang, X. Han, Q. Yu, Y. Liao, G. Li et al. "Indium doped zinc oxide nanobelts," Chemical Physics Letters, vol. 387, pp. 466 470, 4/1/ 2004. [245] M. L. Addonizio and C. Diletto, "Doping influence on intrinsic stress and carrier mobility of LP MOCVD deposited ZnO:B thin films," Solar Energy Materials and Solar Cells, vol. 92, pp. 1488 1494, 11// 2008. [246] H. Hejin Park, R. Heasley, and R. G. Gordon, "Atomic layer deposition of Zn(O,S) thin films with tunable electrical properties by oxygen annealing," Applied Physics Letters, vol. 102, p. 132110, 2013. [247] A. Franciosi and C. G. Van de Walle, "Heterojunction band offset engineering," Surface Science Reports, vol. 25, pp. 1 +, // 1996. [248] M. Masashi, C. Jakapan, A. Naoki, H. Daisuke, and M. Takashi, "Influence of conduction band minimum difference between transparent conductive oxide and

PAGE 244

244 absorber on photovoltaic performance of thin film solar cell," Japanese Journal of Applied Physics, vol. 54, p. 032301, 2015.

PAGE 245

245 BIOGRAPHICAL SKETCH Zhi Li was born in 1986 in China. He was admitted to a dual degree program for his undergraduat e education and earned B.S. in chemical e ngineering from East China University of Science and Technology China and Diploma (FH) in environmental e ngineering from Fachhochschule L beck Germany, in August 2009. After his undergraduate degree, he w ent on to earn his M.S in chemical e ngineering from Columbia University and graduated in February 2011. He then joined Proctor & Gamble for internship. After that, he joined the Ph D pr ogram in chemical e ngineering at University of Florida and worked for Professor Timothy J. Anderson and he chose photovoltaics as his research topic. Upon graduation, he plans to work for Intel Corporation in Portland, Oreg on as a senior process engineer. H is main role is to work on the n ext g eneration 10 nm t ri g ate p rocess