USF Libraries
USF Digital Collections

Impurity and back contact effects on CdTe/CdS thin film solar cells

MISSING IMAGE

Material Information

Title:
Impurity and back contact effects on CdTe/CdS thin film solar cells
Physical Description:
Book
Language:
English
Creator:
Zhao, Hehong
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Antimony
Phosphorus
Doping
ZTO
Barrier
Dissertations, Academic -- Electrical Engineering -- Doctoral -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: CdTe/CdS thin film solar cells are the most promising cost-effective solar cells. The goal of this project is to improve the performance for CdS/CdTe devices by improving the open circuit voltage Vsubscriptoc and current density Jsubscriptsc. Efforts focused on increasing the Vsubscriptoc, which include increasing the doping concentration by introducing Phosphorus and Antimony, finding and testing new back contact materials, and varying the ambient of CSS CdTe. In addition, the effect of Zn₂SnO₄ on the cells' performance was also studied. Electrical characterization of the thin films and completed devices were carried out by Current-Voltage (J-V), Capacitance-Voltage (C-V), and Spectral Response (SR) measurements. Structural/chemical characterization was done by SEM, XRD and EDS analysis. The ambient of CSS CdTe affects the growth rate, the grain size and electronic properties of CdTe. The N₂/O₂ mixture with varied ratio (N₂/O₂=9/1, 7/3, 5/5 and 1/9) was used in this study. The cells' performance and the net carrier concentration were studied as a function of the N₂/O₂ ratio. The net carrier concentration increases with the increasing O₂ concentration. The extrinsic impurities (P and Sb) were incorporated into CdTe layer. Phosphorus was directly introduced into CSS CdTe source. The Sb was incorporated into CdTe by a diffusion process. The effects of the annealing parameters, the excess Sb on CdTe surface, the CdCl₂ treatment and the depth of Sb in CdTe were studied. Higher doping concentration up to 10¹⁶ cm⁻³ has been achieved, however, Vsubscriptoc is still in the range of 830 mV.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2008.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Hehong Zhao.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 116 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001992327
oclc - 315968241
usfldc doi - E14-SFE0002377
usfldc handle - e14.2377
System ID:
SFS0026695:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200397Ka 4500
controlfield tag 001 001992327
005 20090317091534.0
007 cr mnu|||uuuuu
008 090317s2008 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002377
035
(OCoLC)315968241
040
FHM
c FHM
049
FHMM
090
TK145 (Online)
1 100
Zhao, Hehong.
0 245
Impurity and back contact effects on CdTe/CdS thin film solar cells
h [electronic resource] /
by Hehong Zhao.
260
[Tampa, Fla] :
b University of South Florida,
2008.
500
Title from PDF of title page.
Document formatted into pages; contains 116 pages.
Includes vita.
502
Dissertation (Ph.D.)--University of South Florida, 2008.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
3 520
ABSTRACT: CdTe/CdS thin film solar cells are the most promising cost-effective solar cells. The goal of this project is to improve the performance for CdS/CdTe devices by improving the open circuit voltage V[subscript]oc and current density J[subscript]sc. Efforts focused on increasing the V[subscript]oc, which include increasing the doping concentration by introducing Phosphorus and Antimony, finding and testing new back contact materials, and varying the ambient of CSS CdTe. In addition, the effect of ZnSnO on the cells' performance was also studied. Electrical characterization of the thin films and completed devices were carried out by Current-Voltage (J-V), Capacitance-Voltage (C-V), and Spectral Response (SR) measurements. Structural/chemical characterization was done by SEM, XRD and EDS analysis. The ambient of CSS CdTe affects the growth rate, the grain size and electronic properties of CdTe. The N/O mixture with varied ratio (N/O=9/1, 7/3, 5/5 and 1/9) was used in this study. The cells' performance and the net carrier concentration were studied as a function of the N/O ratio. The net carrier concentration increases with the increasing O concentration. The extrinsic impurities (P and Sb) were incorporated into CdTe layer. Phosphorus was directly introduced into CSS CdTe source. The Sb was incorporated into CdTe by a diffusion process. The effects of the annealing parameters, the excess Sb on CdTe surface, the CdCl treatment and the depth of Sb in CdTe were studied. Higher doping concentration up to 10 cm has been achieved, however, V[subscript]oc is still in the range of 830 mV.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor: Chris Ferekides, Ph.D.
653
Antimony
Phosphorus
Doping
ZTO
Barrier
690
Dissertations, Academic
z USF
x Electrical Engineering
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2377



PAGE 1

Impurity and Back Contact Effects on CdTe/CdS Thin Film Solar Cells by Hehong Zhao A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Electrical Engineering College of Engineering University of South Florida Major Professor:Chris Ferekides, Ph.D. Don Morel, Ph.D. Y. L. Chiou, Ph.D. Richard Gilbert, Ph.D. Nicholas Djeu, Ph.D. Date of Approval: December 5, 2007 Keywords: Antimony, Phosphorus, Doping, ZTO, Barrier Copyright 2008, Hehong Zhao

PAGE 2

DEDICATION I dedicate this research and this Dissertation to my family for their love and support.

PAGE 3

ACKNOWLEDGMENTS I would like to express my gratitude to my major professor, Dr. Chris Ferekides, for his endless support and guida nce throughout this research. I also thank Dr. Don Morel for all of his help and guidance over the past years. I would like to thank Dr. Nicholas Djeu, Dr. Yun-Leei Chiou, and Dr. Ri chard to be my committee members. I would like to express my special thanks to Dadra Hodges, Dr. Zhiyong Zhao, and professor Takhir for their help for writing th is dissertation. I am al so grateful to the many fellow research assistants at semiconduc tor lab: Vijay, Bhaskar, Harrshi, Sridevi, and all others. Finally I would like to thank my wife for her unconditionally support and love. This work was supported by National Rene wable Energy Laboratory (NREL), US Department of Energy.

PAGE 4

i TABLE OF CONTENTS LIST OF TABLES v LIST OF FIGURES vi ABSTRACT x CHAPTER 1 INTRODUCTION 1 CHAPTER 2 SOLAR CELL DEVICE PHYSICS 7 2.1 Semiconductors 7 2.2 P-N Junction 8 2.3 Heterojunction 9 2.4 Metal-semiconductor Contacts 14 2.4.1 Schottky Contact 14 2.4.2 Ohmic Contact 17 2.5 Solar Cells 18 2.5.1 Basic Parameters 20 CHAPTER 3 IMPORTANT ISSUES FOR CDTE SOLAR CELLS 25 3.1 Back Contact Effect 25 3.2 Doping Concentration Effect 28 3.2.1 Doping Limit 29 3.2.2 Doping Limitation Mechanism 31

PAGE 5

ii 3.2.2.1 Self-Compensation 31 3.2.2.2 Auto Compensation 34 3.2.2.3 Solubility Limits of Dopants 35 3.2.2.4 Ioization Energy Level Limit 35 3.2.3 Defects in CdTe 38 3.2.4 Acceptors in CdTe 38 3.2.4.1 Group I Elements 38 3.2.4.1.1 Lithium 38 3.2.4.1.2 Silver 39 3.2.4.1.3 Gold 40 3.2.4.1.4 Copper 41 3.2.4.1.5 Disadvantages of Group I Elements 42 3.2.4.2 Group V Elements 43 3.2.4.2.1 Antimony 43 3.2.4.2.2 Phosphorus 45 3.2.4.2.3 Arsenic 47 3.2.4.2.4 Nitrogen 48 3.2.4.2.5 Bismuth 49 3.3 How to Improve the Performance of CdS/CdTe 49 CHAPTER 4 EXPERIMENTAL METHODS 53 4.1 Chemical Bath Deposition 53 4.2 Close Space Sublimation 55 4.3 Sputtering Deposition 56

PAGE 6

iii 4.4 Device Characterization 57 4.4.1 I-V Measurements 57 4.4.2 Spectral Response 57 4.4.3 Capacitance-Voltage Measurement 58 CHAPTER 5 RESULTS AND DISCUSSION 59 5.1 Optimize the HgTe:Cu Doped Graphite Back Contact 59 5.1.1 Annealing Temperature Effect 59 5.1.2 Annealing Time Effect 61 5.1.3 The Effect of the Concentration of HgTe:Cu 63 5.2 High Work Function Materials 65 5.3 High Resistive Buffer Layers 67 5.4 Studies of the Effect of O2 Ambient for CSS CdTe 70 5.4.1 Device Fabrication 70 5.4.2 Structure Properties 70 5.4.3 O2 Partial Pressure Effects 73 5.4.3.1 Thicker CdS 73 5.4.3.2 Thinner CdS 75 5.4.4 Junction Properties with Different O2 Partial Pressure 77 5.4.5 C-V Measurement 78 5.5 Phosphorus Doped CdTe Devices 79 5.6 Antimony Doped CdTe Devices 81 5.6.1 Effect of Sb 82 5.6.2 Effect of the Sb Annealing Temperature 84

PAGE 7

iv 5.6.3 Effect of Annealing Time 89 5.6.4 The Effect of HCl Etching 91 5.6.5 CdCl2 Effect on CdTe:Sb Devices 93 5.6.6 Two-layer CdTe Devices 96 5.6.6.1 The Effect of the CSS Ambient of Second Layer CdTe 96 5.6.6.2 The Depth Effect of Sb in CdTe 99 5.6.7 The Effect of Back Contact 100 5.7 Voc Versus Doping Concentration 102 CHAPTER 6 CONCLUSION 107 REFERENCES 110 ABOUT THE AUTHOR End Page

PAGE 8

v LIST OF TABLES Table 1 Ionization Energies (in mV) of Various Foreign Donors in Different Znand Cd-Chalcodenides 36 Table 2 Ionization Energies (in mV) of Various Foreign Acceptors in Different Znand Cd-Chalcodenides 36 Table 3 Calculated Formation Energy 37 Table 4 Summarizes The Electrical Prope rties of Phosphorous-Doped CdTe and The Acceptor Energy Level of P Reported by Several Authors 46 Table 5 Comparison of World Record and Ideal Solar Cell 50 Table 6 Different Concentrations Used in This Study 64 Table 7 Summary of Doping Concentr ations for Varied Ratio of N2/O2 78 Table 8 Summary of Results for th e Devices with and without Sb 84 Table 9 The Process Conditions and Th e Cells’ Performance with Different Annealing Temperature 85 Table 10 Doping Concentration and Depl etion Width for The Cells Shown in Figure52 88 Table 11 The Effect of Annealing Time 89 Table 12 The Effect of HCl Etching 92 Table 13 Cdcl2 Treatment on Sb Doped Devices 94 Table 14 Summary of the Device s with Different Ambient for 2nd layer CdTe 97 Table 15 Comparative Device Data for Different Back Contact Conditions 100

PAGE 9

vi LIST OF FIGURES Figure 1 The Development History of Thin Film Solar Cells 3 Figure 2 The Growth of PV Modu les Installed in The World 6 Figure 3 Energy Band Diagrams of Various Material 7 Figure 4 The PN Junction Band Diagram at Equilibrium 9 Figure 5 The Forward (right) and Reverse (left) Biased Band Diagram 10 Figure 6 Energy Band Diagram for Tw o Isolated Semiconductors 12 Figure 7 Energy Band Diagram for A Heterojunction at Thermal Equilibrium 12 Figure 8 The Ideal Band Diagram at The CdS/CdTe Interface 14 Figure 9 Energy Band Diagram for Metal-Se miconductor Contact (A) Before contact (B) After Contact 15 Figure 10 Energy Band Diagram of A Me tal-Semiconductor Junction Under (A) Forward Bias (B) Reverse Bias 16 Figure 11 Band Structure of Ohmic Contact (A) Before Contact (B) After Contact 17 Figure 12 The Band Structure of Solar Cells 18 Figure 13 Theoretical Efficiency for A Given Band Gap (AM 1.5 Illumination) 19 Figure 14 I-V Characterizations for Ideal Solar Cell in The Dark and Under Illumination 20 Figure 15 Typical Spectral Respon se of CdS/CdTe Solar Cells 22 Figure 16 The Equivalent Circuit of A Sola r Cell (A) Ideal Case (B) With Series and Shunt Resistance Case 23 Figure 17 Relative Band Edge Positi ons of Various II-VI Compounds 30

PAGE 10

vii Figure 18 Calculated Maximu m Doping Efficiency Limited by Self-Compensation with Native Vacancy 33 Figure 19 The Effect of Back Barrier on Voc 50 Figure 20 Effect Of D oping Concentration on Voc 51 Figure 21 Configuration of CdTe Solar Cells as Used in this Dissertation 53 Figure 22 Diagram of CBD Set Up 54 Figure 23 Schematic Diagram of CSS Deposition Chamber 55 Figure 24 The Summary of the Effect of Temperature 60 Figure 25 Spectral Response of the Effect of Temperature 61 Figure 26 Light and Dark J-Vs for the Devi ces Treated at Different Temperatures 61 Figure 27 The Summary of the Effect of Annealing Time 62 Figure 28 Spectral Response for Devices with Different Annealing Time 62 Figure 29 Light and Dark J-Vs for the De vices with Different Annealing Time 63 Figure 30 The Summary of the Eff ect of HgTe:Cu Concentration 64 Figure 31 Spectral Response for Devices with Different Concentra tion of HgTe: Cu 64 Figure 32 Light and Dark J-V for the Devices with Different Concentration of HgTe:Cu 65 Figure 33 XRD Spectrum of A Selenized Ti Film 66 Figure 34 Spectral Response for Devices with ZTO Buffer Layer Anne aled at Different Temperatures 68 Figure 35 Spectral Response for Devices wi th ZTO (Zn/Sn=2) Buffer Layer Deposited at 400C 69 Figure 36 SEM Images of CdTe Films Deposited under Different Conditions 71 Figure 37 I-V Characteristics fo r Device Fabricated at Various O2 Concentrations 73 Figure 38 Spectral Response for De vices Fabricated at Various O2 Concentrations 74

PAGE 11

viii Figure 39 Summary Results for De vices Fabricated at Various O2 Concentrations 74 Figure 40 Dark I-V Characteristics for Device Fabricated at Various O2 Concentrations (Thinner CdS) 75 Figure 41 Spectral Response for Device Fabricated at Various O2 Concentrations (Thinner CdS) 76 Figure 42 Spectral Response for Device Fabricated at Various O2 Concentrations (Thinner CdS) 76 Figure 43 Diode Factor versus O2 Partial Pressure 77 Figure 44 Saturation Current versus O2 Partial Pressure 77 Figure 45 C-V Characteristics of Devices Fabr icated with Different Partial Ppressure 78 Figure 46 Voc for Cells With Phosphorus 80 Figure 47 FF for Cells With Phosphorus 81 Figure 48 Characteristics for Cells With Phosphorus 81 Figure 49 Light J-V Characteristics of CdTe Cells with and without Sb 83 Figure 50 Doping profile of CdTe Cells with and without Sb 83 Figure 51 Spectral Response of CdTe Cells with and without Sb 84 Figure 52 Spectral Response for Device wi th Different Annealing Temperature 86 Figure 53 The Effect of The Th ickness of N Layer CdTe on QE 86 Figure 54 The Effect of Donor Defects in CdTe Bulk on QE 87 Figure 55 Doping Concentration Vs. Depletion width (left) and C-V Characteristics 88 Figure 56 Dark (left) and Light (right) J-V Characteristics with Different Annealing Time 90 Figure 57 The QE of the Devices w ith Different Sb Annealing Time 90 Figure 58 Doping Concentration vs. Depletion width(left) and C-V Chacteristics(right) 91 Figure 59 Spectrsl Response for HC l and No HCl Etched Devices 93

PAGE 12

ix Figure 60 Light(left) and Dark(right)JV for HCl and No HCl Etched Devices 93 Figure 61 Spectral Pesponse for th e Devices With and Without CdCl2 Treatment 94 Figure 62 Light(left) and Dark(right)JV for the Devices With and Without CdCl2 Treatment 95 Figure 63 Light J-V of The Devices Fabricated in Different Ambients(2nd Layer) 97 Figure 64 Dark J-V of The Devices Fabricated in Different Ambients(2nd Layer) 98 Figure 65 Spectral Sesponse of The Devi ces Fabricated in Different Ambients(2nd Layer) 98 Figure 66 Dark J-V Characteristics of Cd Te Cells with Different Thickness of First Layer 99 Figure 67 Light J-V Characteristics of CdTe Cells with Different Thickness of First Layer 100 Figure 68 Light J-V for the Devices w ith Different Back Contact Conditions 101 Figure 69 Dark J-V for the Devices w ith Different Back Contact Conditions 102 Figure 70 Voc Versus Doping Concentra tion at Different Process Conditions 103 Figure 71 Simulations with Different Back Barriers 104 Figure 72 I-V Simulations with Different Back Barriers for 21016 cm-3 105

PAGE 13

x Impurity and Back Contact Effects on CdTe/CdS Thin Film Solar Cells Hehong Zhao ABSTRACT CdTe/CdS thin film solar cells are the most promising cost-effective solar cells. The goal of this project is to improve the performance for CdS/CdTe devices by improving the open circuit voltage Voc and current density Jsc. Efforts focused on increasing the Voc, which include increasing the doping concentration by introducing Phosphorus and Antimony, finding and testing new back contact materials, and varying the ambient of CSS CdTe. In addition, the effect of Zn2SnO4 on the cells’ performance was also studied. Electrical characterization of the thin films and completed devices were carried out by Current-Voltage (J-V), Capacitance-Voltage (C-V), and Spectral Response (SR) measurements. Structural/chemical charac terization was done by SEM, XRD and EDS analysis. The ambient of CSS CdTe affects the grow th rate, the grain size and electronic properties of CdTe. The N2/O2 mixture with varied ratio (N2/O2=9/1, 7/3, 5/5 and 1/9) was used in this study. The cells’ performan ce and the net carrier concentration were studied as a function of the N2/O2 ratio. The net carrier concen tration increases with the increasing O2 concentration.

PAGE 14

xi The extrinsic impurities (P and Sb) were incorporated into CdTe layer. Phosphorus was directly introduced into CSS Cd Te source. The Sb was incorporated into CdTe by a diffusion process. The effects of the annealing parameters, the excess Sb on CdTe surface, the CdCl2 treatment and the depth of Sb in CdTe were studied. Higher doping concentration up to 1016 cm-3 has been achieved, however, Voc is still in the range of 830 mV.

PAGE 15

1 CHAPTER 1 INTRODUCTION The evolution of human beings is also the history of finding and using new energy. Utilizing fire to get energy was a bi g step for man’s development. Two thousand years ago, men already learned how to use water to drive machines. Following that was the finding of fossil fuels. The invention of electric power is an important milestone of the evolution of man’s use of energy. Since then, almost ever y kind of energy had to be converted into electric power. There are various energy s ources that have been found and utilized by man. These energy sources have been classified as nonr enewable and renewable energy sources. The nonrenewable energy sources include fossil fuel s (used to generate 61% of the world’s electric power, 95% of the worl d’s total energy demands), such as coal, oil, natural gas, and nuclear power. The renewable energy sources include solar energy, wind, geothermal, hydropower, etc. Fossil fuels need millions of years to form. The main advantages of fossil fuels are low cost, easi ly transportable and st ore. However, burning coal produces sulphur dioxide, which causes acid rain and is harmful to the earth’s environment. Burning any kind of fossil fuels will produce carbon dioxide, which is believed to be the cause of the green house effect. Nonrenewability of fossils is its biggest disadvantage. Once we have burned it all, there isn’t any more unless you can wait for millions of years.

PAGE 16

2 Nuclear power provides about 17% of the wo rld’s electricity. It is a very clean energy source. The main problem or disadva ntage of nuclear power is the risk of radioactivity. Nuclear power accidents releasi ng radioactivity would be a major disaster like that of Chernobyl. The wast e is toxic for centuries and there was no safe, permanent storage facility for it until now. Wind energy is green and renewable. It is limitless, but it is not a stable energy source. It is only suitable for certain speci al areas. Hydropower doesn’t produce harmful byproducts like fossil fuels, so it is a clean energy. The problems and risk caused by hydropower is the dam. The dam can negatively affect the ecosystem where it is built. It is already well developed wherever is suitable to build hydropower plant. Solar energy is the cleanest energy we can use. Solar energy output could last 10 billion years. Using the photovoltaic effect, so lar cell converts solar energy directly into electricity. Edmund Becquerel first discovere d the photovoltaic effect in 1839. Thirtyfour years later in 1877, the fi rst solar cell made from se lenium was invented by W.G. Adams. Before 1954, the development of solar cells was very slow. In 1914, the efficiency was only about 1%. The breakth rough was made by Chapinetc with a 6% efficiency silicon solar cell in 1954. By 1958, the silicon solar cell efficiency had already reached 14%. Now single crystal silicon sola r cell efficiency is over 30%. But the single crystal silicon solar cell has its fatal disa dvantage: it is expens ive compared with conventional electricity. It was only used in space applications and some military related things where reliability is the main concern. In order to decrease the cost, polycrystalline silicon and amorphous silicon solar cells were developed. Hydrogenated amorphous silicon improves light absorption and the bandgap was increased. Hydrogenated

PAGE 17

3 amorphous silicon behaves like a direct ba ndgap material. The main problem is its instability under light illumination. Besides silicon solar cells, there is another kind of solar cell called thin film solar cells. In 1954, Reynolds [1] reported the first thin film solar cell with a 6% efficiency, the CuxS/CdS hetero-junction solar cell. Two main thin film solar cells were widely researched: CIGS and CdTe based solar cells. CIGS solar cells were deve loped from CIS. CIS solar cells show high current, but its efficiency is limited by its low bandgap ( 1.0eV). Researchers added Ga into CIS to open the bandgap to 1.1~1.2eV. So the Voc was increased. The effect resulted in a decrease of current. The efficiency was impr oved. Usually, CIGS so lar cells were made by coevaporation of Cu, In, Ga following the se lenization process. CIGS solar cells with 18.8% efficiency have been achieved. Figure 1 The Development History of Thin Film Solar Cells

PAGE 18

4 CdTe/CdS solar cell is the most promising thin film solar cell. CdTe has nearly ideal band gap of 1.45eV to covert sunlight into electricity. It is a direct band gap material, and the absorption coefficient in visible range is more than 105 cm-1, so only a few micrometers thick CdTe can absorb more than 90% of photons with energy above1.45eV. The material cost fo r solar cells is relatively low compared with Si solar cells, which need a much thicker film. On the other hand, CdTe thin film can be deposited by several low cost methods, such as close-space sublimation, spray deposition, electro deposition, screen-printing, and PVD. All these deposition methods can lead to high efficient devices. The first CdTe thin film solar cells were prepared by Cusano [2] in 1963 with the structure of CdTe/Cu2-xTe. In this structure CdTe was used as n layer and Cu2-xTe as the p layer. The cells have efficiencies close to 6% but not stable. The first CdTe/CdS solar cells were proposed by Andirovich [3] in 1968, but he only achie ved 1 % efficiency devices at that time. Due to the energy crisis in early 1970’s, more funding was put into the PV research. The progress of CdTe/CdS solar cell is very fast. In 1981, Kodak achieved 10% efficient devi ces. AveTEK achieved 12% de vices in 1990. After 1990, USF, Photon Energy and BP alternatively le ad the research. In 1992, Ferekides and J. Britt got 15.8% devices. The record was kept nearly 10 years. Now the world record efficiency is 16.5%, achieved by X.Wu [4] in NREL. Figure 1 shows the development histories of CuInSe2, CdTe and -Si solar cells [5]. All the reported high efficient CdTe/CdS solar cells rely on a special treatmentCdCl2 treatment [6-12]. It is still not comp letely understood. Studi es show that CdCl2 treatment leads to an intermixing and recrys tallization at the interface, the lattice

PAGE 19

5 mismatch is reduced. The carrier lifetime wa s increased, and the density of interface states is greatly reduced. There are two main technology issues re lated to further improve the cells performance. One is to increase the p-type doping of CdTe, the other is to get ohmic contact to p type CdTe. We will talk about th em in detail later. Long-term stability and the toxicity of cadmium are th e other two main issues. Copper diffusion is considered to be the main reason for the degradation [1315]. A copper free contac t will be the reliever to this issue. As to the toxi city of cadmium, it is not as what people think at first glance. In fact, based on the studies of V. Fthenaki s [16-17], most of the cadmium was captured into molten glass under 1000 C fire. It only produces 0.06mg/Gwh emission of cadmium, which is negligible within the context of life cycle analysis. On the other hand, end-oflife or broken modules can be recycled, so a ny environment concerns will be completely resolved. After more than 30 years of research a nd development, the fabrication of PV module technology becomes more mature. There are several CdTe PV module factories in the world, such as Antec, Primestar So lar and First Solar. The average PV module efficiency is 7-9%. There is a big efficien cy gap (7~9%) between the lab small cell and the large area PV modules. There is a long ro ad to walk to reduce this gap for the CdTe solar cell to become cost-competitive. Over the past 20 years, the demand for energy has grown consistently by 10~25% per year. The cost has reduced seven fold due to the progress of technology and the increased market volume. Even though, the el ectricity from solar cells is still not competitive compared with conventional electr icity, if you only consider it from financial

PAGE 20

6 point. Today, many governments recognized th e benefit of using solar power. More money has and will continue to be invested on re search in this area. This will lead to an increase in the technology’s progress. The co st will be reduced further and become competitive with conventional el ectricity in the not far future. Solar energy will occupy more and more portions of the whole en ergy demand. The futu re must belong to renewable energy. Figure 2 shows the growth of PV modules installed in the world. In 2005, 1460 Megawatts of PV were installed. Th is increased to 1744 megawatts in 2006. World Solar Photovoltaic Installations 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2000200120022003200420052006Megawatts Figure 2 The Growth of PV Modules Installed in the World

PAGE 21

7 CHAPTER 2 SOLAR CELL DEVICE PHYSICS 2.1 Semiconductors Materials can be classified into thr ee groups based on thei r conductivity: (1) Conductors, whose resistivity is <10-4 m (2) Insulators, whose resistivity is>1012 m (3) Semiconductors, whose resistivity is between conductors and insulators. The semiconductors are the corner stone of the modern electronic technologies. Semiconductors can be classified into P type and N type based on their majority carriers. The most important characteristic of semiconductors is th at their conductivity can be modified by adding impurities, which are called dopants. For example, adding one group III element into silicon makes the major ity carriers holes and the silicon becomes P type. Adding a group V element into Si make s the majority carriers electrons, silicon becomes N type. Ec Ec Ec Ef Ef Ef Ev Ev Ev INTRINSIC N-TYPE P-TYPE Figure 3 Energy Band Diagram s of Various Materials

PAGE 22

8 There are two important concepts about semiconductors: Fermi energy Ef and bandgap Eg. The Fermi Energy is defined as the energy at which the probability of a state is occupied by an electron is 0.5. The bandga p Eg is defined as the energy difference between the energy of the highest valence band and the energy of the lowest conduction band. In this energy gap, no allo wed energy states can exist. 2.2 PN Junction For P type material, the majority carri ers are holes. For N type material, the majority carriers are electrons. When these two types of material s contact with each other, a PN junction is formed. Because of the concentration gradient, the electrons diffuse into the P side and recombine with holes near th e junction, while holes diffuse into the N side and recombine with electrons near the junction. The diffusion movement of these carriers forms the diffusion current. Adjacent to the PN interface, there are no free carriers on both sides. This region is ca lled the depletion region (also called space charge region (SCR)). The result of the diffu sion leads to the build up of positive charge on the N side and negative charge on the P side, and an internal electric field is established. This internal electric field tries to drift holes back to th e p side and electrons back to the n side which led to the formation of drift curren t. Without any bias, the drift current is equal to the diffusion current, so the net current is zero. A flat Fermi level is established. Figure 3 shows the energy diagram for P type and N type material. Figure 4 shows the PN junction band di agram at equilibrium. Vbi is the band bending, called builtin potential, and can be expre ssed as the following equation: n N N q kT Vi D A bi 2ln (1)

PAGE 23

9 Where NA and ND are the doping concentrations in the P and N type semiconductors, ni is the intrinsic carrier concentration, k is Bo ltzmann’s constant, and T is the absolute temperature. Figure 4 The PN Junction Band Diagram at Equilibrium When a bias is applied across a PN j unction, the equilibrium state is broken. When a positive terminal is connected to P side, the bias is called a forward bias; otherwise, it is a reverse bias. Figure 5 shows the forward and reverse biased band diagram. The current-voltage relationship of a biased PN junction can be expressed as: 1 exp0AkT qV J J (2) Where J0 is the reverse saturation curren t, and A is the diode factor. 2.3 Heterojunction From the band structure shown in Figure 4, you can see that th e materials on both the P and N side are identical This kind of PN junction is called homojunction. There is another kind of PN junction called heterojunc tion. In a heterojunction, the material on the P side is different from that on the N side. Heterojunction can be classified as isotype

PAGE 24

10 heterojunction, in which the two semiconducto rs have the same type conductivity; and anisotype heterojunction, in which the tw o semiconductors have the different type conductivity. CdTe/CdS thin film solar cell s are anisotype hete rojunctions. Figure 6 shows the energy band diagram of n type materi al and p type materi al without contacting each other. EGN and EgP are the bandgaps, P, N are the electron affinities, and FN and FP are the Fermi levels. n and p are the displacement of the Fermi level from the conduction band edge and valence band edge in the n type and p type semiconductors. Figure 5 The Forward(right) and Re verse Biased (left)Band Diagram Due to the different materials’ properti es, there are discontinuities called band offset in the conduction band and in the va lence band. The conduction band offset can be defined as N P cE The valence band offset can be defined as E E E Ec GN gp v Ec can also be calculated as E V Egp p n D c Since P, N, Egp, EGN are independent of doping for nondegenerate semiconductors, so Ec and Ev are invariant for nondegenerate materi al. For an abrupt one-sided junction, VD can be obtained from the capacitance measurement. Once the VD is obtained, Ec and

PAGE 25

11 Ev can be obtained using the aforementioned e quations. When n type material contacts with the p type material, because FN is higher than FP, some electrons must flow from the N side to the P side, resulting in up band be nding for the N side and down band bending for the P side near the juncti on. Figure 7 shows the heteroju nction band structure after the junction formation. Compared with a homojun ction, the band structure is much more difficult to calculate because of the discontin uity and the interface state. The first and most popular model is proposed by Anderson [18] In this model, Anderson considers the material properties such as dielectric constant s, electron affinities a nd energy gaps. In this model, the current flow is entirely by in jection over the conduction or valence band barriers. The current has the form: kT V q kT V q kT V q A In p bp/ exp / exp / exp (3) Where A=XaqNDDn/Ln, Vbp is the band bending on the p side of the junction, Vp and Vn are applied voltage dropped on the p and n side of the junction. Further studies show that current-voltage behavior with temperature of most heterojunctions doesn’t conform to equation 3. This is because Anderson’ s model doesn’t consider tunneling and recombination effects. Later several models were proposed for recombination current, or tunneling current, or the combination of these mechanisms. Dolega proposed the recombination model assuming a lifetime appr oaching zero at the interface. The current is proportional to exp(qV/Akt), where A is th e diode factor and depending on the ratio of the impurity concentration. Riben developed the tunneling current m odel. The tunneling current has the form: J=J0exp ( KpV), where J0=BXNt exp (Vb), where Nt is the density of available tunneling states, B is a constant, and X is the transmission coefficient for electrons to cross the interface.

PAGE 26

12 Figure 6 Energy Band Diagram for Two Isolated Semiconductors Figure 7 Energy Band Diagram for A Het erojunction at Thermal Equilibrium The most important advantage of hetero junction devices is that the surface– recombination losses and sheet-resistance loss es are greatly reduced due to the deep junction position. A special property of the hete rojunction is that it makes it possible to use some materials which can only be doped eith er p-type or n-type to make solar cells. Without this constraint, many promising PV materials can be investigated to produce optimal cells. The window material can be made highly conductiv e, light-generated

PAGE 27

13 electrons can easily flow laterally in the wi ndow layer to reach an electrical contact. Also, the window layer band gap can be wider than th e absorber layer, allowing more light to be absorbed in absorber layer. Most of the heterojunction solar cells ha ve the structure of n type wide gap material and p type narrow gap material. One reason of selecting this structure is that most heterojunction pair’s Ec is considerably less than Ev, so the barrier for electrons flowing from p side to n side is lower. The second reason is that most of the current is caused by electrons from p type narrow ga p material, and the diffusion length of electrons is longer than that of holes. Of course, heterojunction devices have their own problems like lattice mismatch and the discontinuity of electron affin ity, which can produce interface states. The interface states act like recombination cent ers and should be minimized. Chemical compatibility, thermal expansion coeffici ent and long term stability are the other concerns. The typical band structure of a CdS/Cd Te solar cell is shown in figure 8. Ec is a little bit less than or close to zero. The nega tive value usually calls cliff. There exists a back barrier due to the insufficient work f unction of the contact material. The conduction band discontinuity does not limit the electron photocurrent from CdTe to CdS. The hole photocurrent from CdS to CdTe is not limite d either by the valence band discontinuity. James sites simulated the effect of Ec [19]. He found that even a small spike Ec could limit the ability to increase Voc with band gap. Photons with energy E< 3.5 eV (355 nm) will pass through the SnO2 layer. Part of photons (~ 30~60%, depending on the CdS thickness) with energy 2.43 eV
PAGE 28

14 1.45 eV
PAGE 29

15 Figure 9 Energy Band Diagram for Metal -Semiconductor Contact (A) Before Contact (B) After Contact lower energy levels, and accumulate near th e metal surface, leaving behind an electrondepleted region near the junction. In the de pletion region, the right side is positively charged donors, a built-in electric field pointi ng to the metal surface is developed. Finally this built-in potential prevents further accumulation of electr ons at the metal surface. At this point, equilibrium is established. The Fermi level throughout the whole solid is uniform. In the depletion region, electrons ar e depleted, so the energy band must bend up toward the junction. The band bending creates a barrier for electrons moving from metal to semiconductor. This barrier height is given by M B (4) Where is the electron affinity of the semiconduc tor. In practice it is very difficult to change the barrier height by va rying metal work functions. It is observed that the barrier

PAGE 30

16 height for some semiconductors is almost independent of the work function of the metal. This phenomenon can be explained by surface st ates. If the density of surface states is high enough, the barrier height is pinned at some energy. The surface states may be dangling bonds at the surface or some other type s of defects. They play an important role for contact formation. Figure 10 Energy Band Diagram of A Met al-Semiconductor Junction under (A) Forward Bias (B) Reverse Bias When a Schottky junction is in an open ci rcuit condition, the net current is zero. When it is reverse biased, the width of the depletion region is in creased. Figure 10 shows the energy band diagram of a metal-semic onductor junction under forward bias and reverse bias. The barrier for electrons from semiconductor to metal is increased to Vbi+VR, so the current from semiconductor to metal will decrease exponentially with VR. The current is essentially lim ited by the current from meta l to semiconductor. When the junction is forward biased, th e barrier for electrons from semiconductor to metal is

PAGE 31

17 decreased to Vbi-VF. The current is primarily limited by this current, which is exponentially increased with VF. 2.4.2 Ohmic Contact Figure 11 Band Structure of Ohmic Contact (A) Before Contact (B) After Contact Figure 11 shows the band diagrams of me tal and n type semiconductors before and after the contact. In this figure, the metal work function m is lower than the work function of semiconductor. The electrons in metal will tunnel into the semiconductor to seek lower energy levels. Electrons will accumulate in the semiconductor near the junction, and the band has to bend down to in crease the electron concentration near the junction. So from the metal to the end of semiconductor side there are always conduction electrons, meaning it is conductive throughout the whole solid. Ther e is no barrier for electrons in either directi on because the metal and the accumulation region have more

PAGE 32

18 electrons compared with the bulk of the semiconductor. The current is mainly limited by the bulk resistance of the semiconductor. Here the ohmic contact doesn’t mean linear I-V characteristics. It could be linear or quasi-linear. For a Ptype semiconductor, the work function of metal has to be greater than the work function of the semiconductor to achieve ohmic contact. Figure 12 The Band Structure of Solar Cells 2.5 Solar Cells A solar cell is a device that converts sunlight into electricity using the photovoltaic effect. Figure 12 shows the prin ciple of the photovoltaic device. When sunlight strikes on the solar cell, the photons with energy greater than the absorber bandgap are absorbed to produce electron-hole pairs. The excess energy over Eg just wastes as heat. The photons with energy less th an the bandgap will pass through the device and make no contribution to the cell output. The gene rated electron-hole pairs are transported to the junction region and separa ted by the built-in electric field of the

PAGE 33

19 junction. Based on this mechanism, to achie ve a high performance solar cell, several requirements must be met: (1) A large absorpti on coefficient, so only a thin film is enough to absorb most light. For CdTe, a 2m thick film is enough to absorb more than 90% of the photons with energy above its bandga p. Since the absorber layer is very thin, the requirement of high quality materials is released. (2) An appropriate band gap. This means Figure 13 Theoretical Efficiency for A Given Band Gap (AM 1.5 Illumination) the bandgap shouldn’t be too larg e or too small. Theoretical ideal bandgap for solar cells is 1.28 eV. If it is large, most of the light w ill pass through the absorber layer leading to a high Voc and a low Jsc. If it is too small, most part of the energy will be wasted as heat. The CdTe’s bandgap is 1.45 eV. It is almost ideal. The theoretical maximum efficiency is 29.2% with Voc=1.07 V, FF=89% and Jsc=29.5 mA/cm2. Figure 13 shows the theoretical efficiency—bandgap relationship. (3) A low re sistance ohmic contact, so most of the separated carrier can be trans ported to outside load device.

PAGE 34

20 40 0. 5 0. 5 Figure 14 I-V Characterizations for Ideal Solar Cell in The Dark and Under Illumination 2.5.1 Basic Parameters Whenever we evaluate a solar cell’s performance, we always use the conversion efficiency. It is defined as the ratio of th e maximum output electrical energy and the input light energy and is given by: in maxP P (5) Where Pmax is the maximum output electrical power, Pin is the incident light energy. In the laboratory, the light simulator is calibrated to 0.1W/cm2, and the efficiency becomes: FF J V J V Psc oc max max max (6) Where Voc is called open circuit voltage in Volt. It is the output voltage when the load has infinite resistance. Jsc is called short circuit current in mA/cm2, when the load circuit is shorted. Vmax and Jmax are the voltage and current wher e the output power is maximum. FF is called fill factor, it is defined as the follow: Dark Light Voc Pmax Jsc IL

PAGE 35

21 sc oc max maxJ V J V FF (7) From equation 6, in order to get higher c onversion efficiency, th e device must have higher Voc, Jsc and FF. Figure 14 shows the typical J-V curve of ideal solar cell measured both in the dark and under illumination. IL is the current produced by light generated electron hole pairs separated by the electric field. One should note that the current density from the J-V curve is not as accurate as the one calculated from the spectrum res ponse due to the imprecise area measurement. Figure 15 shows the typical spectral response of a CdS/CdTe solar cell. It shows quantum efficiency varies with the wavelength. Basica lly, quantum efficiency is defined as the ratio of the collected electrons to the incident photons at each wavelength. ) ) qF( ) J ) (L R SR (8) Where JL ( ) is the photo generated current as a function of the wavelength, F ( ) is the number of incident photons, and R ( ) is the reflected photons by the surface. If we take the derivative of function 8 with respect to we get: d R F d qSR d J dL 1 (9) If the SR ( ) is known, the exact photocurrent density can be obtained by the integration of the right side of the equation from 0 to the maximum corresponding to the absorber bandgap. In ideal case, the SR should be 1, but in practice, it is usually less than 1. The loss due to reflection is usually 5%. In Figure 15 the main loss is due to the absorption of TCO and CdS.

PAGE 36

22 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. Figure 15 Typical Spectral Response of CdS/CdTe Solar Cells Figure 16(a) shows the equiva lent circuit of an ideal solar cell which has zero series resistance and infinite shunt resistance. For practical solar cells, Rs is not zero and Rsh is not infinite. Figure 16(b) shows the equi valent circuit of a so lar cell with nonzero Rs and finite Rsh. For an ideal solar cell, from Figure16, the current-voltage relationship can be given by: I AkT qV I IL 1 exp0 (10) Set V=0, we get Isc= IL Set I=0, we get 1 ln0I I q AkT VSC oc (11) Where I0 is reverse saturation current a nd k is the Boltzmann’s constant. When we take into account the Rs and the Rsh (Frequently they are assumed to be constant, but they may be light sensitiv e and time variant), equation 10 becomes R R I V I I AkT R I V q I Ish sh L L s 1 exp0 (12)

PAGE 37

23 Equation 11 doesn’t change. So we can say series resistance doesn’t affect Voc and reduces the short circuit current. It reduces the maximum power available from the device. The series resistance consists of th e bulk resistance of the semiconductor, the bulk resistance of the contacts, and the cont act resistance between the contacts and the semiconductor. Several methods have been used to determine Rs, such as dV/dj at forward Figure 16 The Equivalent Circuit of A Solar Cell (A) Ideal Case (B) With Series and Shunt Resistance Case bias (Used in this dissertation), multiple light intensities, constant light intensity, and dI dV versus I 1 plot, etc.. Here we discuss the dI dV versus I 1 plot method in detail. From equation 12, we can get: 1 ln0 0R I R I V I I I R I V AkT qsh s L s (13)

PAGE 38

24 Usually RSh is very large, so sh sR I0IR V can be neglected. Equation 12 becomes: 1 ln0I I I R I V AkT qL s (14) From equation 14, we get 1 ln0I I I q AkT R I VL s (15) We take the derivative on both sides. I q AkT R dI dVs1 (16) If we plot dI dV versus I 1 the intercept with the y-axis will be Rs, and from the slope we can get the diode factor A. Us ually A is about 1-2, and it is voltage dependent. Usually the second term is small compared with Rs and can be neglected, so we can roughly get the Rs from dV/dJ. This is what we used in our lab. More accurate Rs can be determined from multiple light intensities I-V measurement. This method is independent of I0, A and Rsh and no limiting approximation, so the result is generally considered to be close to the real value. If there are leakage paths ac ross the junction, part of th e current will be bypassed by these paths, resulting in the reduction of shunt resistance. The pinholes and grain boundaries can lower the shunt resistance. The FF and Voc will be decreased with decreasing shunt resistance. We can get shunt resistance by taking th e reciprocal of the slope of the J-V curve at th e high reverse bias region.

PAGE 39

25 CHAPTER 3 IMPORTANT ISSUES FOR CDTE SOLAR CELLS 3.1 Back Contact Effect One of the most important issues for CdTe solar cells is it is very hard to get a good ohmic contact on p-type CdTe. This is because p-type CdTe has a high electron affinity (4.5eV) and a large band gap (1.45eV), so its work function is nearly 5.9eV. From contact theory, in order to get a barrier free contact, the work function of the metal must be greater than the work function of the p-type CdTe (5.9eV). But unfortunately, there is no such metal that satisfies this requir ement. In practice, there are three strategies to get a good ohmic contact. First is etching CdTe to adjust and control surface stoichiometry. Etching can be classified as wet and dry etching based on the medium used. For wet chemical etching, two types of solutions are usually used. One is an acid solution, such as 7K2Cr2O7:3H2SO4, K2Cr2O7:HNO3:H2O [91], and HNO3:H3PO4:H2O [92], the other one is Br2 /CH3OH. Dry etching can be classified into plasma etching, reactive ion etching and physical sputtering. The advantages of dry etching include: capability of anisotropic etching, eliminating handling, consumption and disposal of large amounts of liquid solvents used in wet chemi cal etching, and easily integrated into the production line. But the surface could be c ontaminated by redeposition of non-volatile species, and damaged by high-energy partic les’ bombardment. The second way is heavily doping the CdTe surface to produce a p+ layer; the barrier between CdTe and the

PAGE 40

26 contact will be narrowed so that free carriers can tunnel freely through the barrier, such as Au-Cu, Au-Li and Au-P. The third is us ing another high elec tron affinity, high conductivity p-type semiconductor as a buffer la yer, which contacts can be made more easily, such as HgTe and ZnTe. Currently, different kinds of back cont acts have been achieved, they can be classified into two categories: copper containing back contac ts and copper-free contacts. Copper-free contacts include Ni-P, Sb2Te3, Mo, and HgTe. Copper-containing contacts include Au-Cu alloy [20, 21], ZnTe:Cu [22, 23], HgTe:Cu doped graphite paste [24, 25], and Cu2Te [26, 27]. All of these kinds of contacts contain an elemental form of Cu or Cu compound. Copper is a fast diffuser. In a single crystal CdTe, Cu is considered to exist in the forms: Cui + and Cucd -. Cui + is a shallow donor with ac tivation energy 55 emV. Cucd is a deep acceptor with activation energy 0.28~0.34eV. C opper can also form a complex (Cui + +Vcd -2, Cui + + Cucd). H.C. Chou ’s studies show that Cu plays dual roles in CdTe solar cells [93]. The good one is that it is giving a good ohmic contact by increasing the p doping near back contact. The best cell with th e world record efficiency (16.5%) utilized a copper containing back contact. The negative effect is the migrati on of Cu from back contact to the main junction. It is responsible for de gradation of long term cell performance. Due to the stability problem caused by Cu Copper-free back c ontacts have gained increased attention. Several materials have been studied, such as Sb2Te3, VeSe2, and TiSe2 [32]. Sb2Te3 and Ni2P have shown promise as back contacts to CdTe solar cells [28-31].

PAGE 41

27 B.Ghosh et al. used electroless deposition Ni-P for contact fabrication to CdTe [28]. They found the optimum annealing temperature is 250C. At this temperature, Ni2P becomes more prominent. It is believed that the work function of this phase is similar as gold resulting in better contact properties. Above 250C, the contact resistance is increased. The group at USF used Ni2P powder mixed with graphite paste to fabricate back contact to CdTe and demonstrated Ni2P as a very promising candi date as a back contact material to CdTe [29]. In their study, they studied the effect of various processing parameters, such as annealing temperature, annealing time, and Ni2P concentration. The best processing parameters they achieved: 25%wt Ni2P, and 90 minutes annealing at 250C. The best device has Voc of 834 mV, FF of 70%, and 12% efficiency. N. Remeo et al. studied Sb2Te3 as a back contact to CdTe [30, 31]. They sputtered Sb2Te3 at 300C with a final layer of sputtered Moly. The resistivity is 10-4 -cm. The best cell has Voc of 857 mV, FF of 73% and effici ency of 15.1%. A stability study showed there is no change on cell performance after a 6 month, 1 sun soaking. D. Kraft et al. studied VSe2/CdTe and TiSe2/CdTe interface prope rties [32]. VSe2 and TiSe2 have high work functions 5.7eV and 5.6eV, respectively. They could form ohmic contacts to CdTe due to their high work functions. In this study, CdTe was deposited on VSe2 and TiSe2 substrates to form VSe2/CdTe and TiSe2/CdTe interfaces. They found an excess of Cd exists at the interf ace due to higher stic king coefficient of Cd atoms. Due to the relatively small work f unction of Cd (4.22eV), an interface dipole occurs resulting in a vac uum level discontinuity of Evac=0.65eV for CdTe/ VSe2 and Evac=0.55eV for TiSe2/CdTe. This leads to a nonohmic contact to CdTe. This suggests

PAGE 42

28 that if we can avoid the excess Cd by ch anging the processing conditions, the ohmic contact could be formed. 3.2 Doping Concentration Effect From equation 11, 1 ln0J J q AkT VSC oc, we can see Voc increases with decreasing J0. J0 can be given by: L n D q L p qD Jn p n p n p 0 0 0 (17) Where LP, Ln is the diffusion length of holes and electrons, respectively. Dp, Dn is the diffusion coefficient for holes and electrons, respectively. pn0 is the equilibrium hole concentration on the n side, given by n i nn n p2 0 np0 is the electron concentration on the P side, given by p i pp n n2 0. If carrier concentrations on one side or both sides are increasing, pn0 or np0 or both will decrease and Voc will increase. For example, if the doping concentration can be increased one order of magnitude on both sides, Voc could be increased about 90~120mV depending on diode f actor A. On the other hand, with the increasing doping concentration, the surface recombination effect will become weaker and practically become eliminated when it is more than 1016 -1017 cm-3. The recombination losses in space charge region w ill also decrease due to the increasing of the electric field with the increasing doping concentrati on. The collection will be improved, resulting in an increased FF.

PAGE 43

293.2.1 Doping Limit As we talked earlier, semiconductors resistivity is between conductors and insulators. In order to lowe r the resistivity, doping concen tration must be high enough. For some semiconductors, it is not a very easy task. Some semiconductors can be doped as high as 1021 cm-3 in both p type and n type. Some semiconductors can be doped with very high concentration in only p type or onl y n type, and are not dopable in the other type. Some can be doped in both type but with low concentration. CdTe is the only one II-VI compound that can be doped relatively eas y p type and n type. But doping ability is not same for n type and p type doping. For p type doping, the hole concentration becomes limited at 1017 cm-3. For polycrystalline CdTe, the hole c oncentration is even lower due to boundary effects. To obtain high doping concentration, three re quirements must be satisfied. First, the solubility of the dopant must be high e nough. For example, if you want to obtain 1021cm-3 doping concentration, the solubility in that semiconductor must be >=1021 cm-3 and these dopant atoms have to be at right place, generally in the subst itution position. Second, the formation energy levels have to be low enough so that dopants can be easily ionized at normal operation temperatures to pr ovide free carriers. Third, the incorporation of dopants in crystal must not spontaneously induce the oppositely charged defects or defect complexes to compensa te the intentional dopant. Several mechanisms have been proposed for explaining the doping limit. Sometimes one mechanism can explain the experimentally observed doping problems for one specific material, but not suitable for other materials. Sometimes two or more mechanisms have to combine together to e xplain the specific problem. What property of

PAGE 44

30 material plays the main role to determine the doping limit? Can we predict it? The answer is yes, maybe it is not 100% right, but at leas t the trend can be predic ted. This is the socalled phenomenon logical doping limit model. Th is model was first developed for III-V compounds by W. Walukiewicz [32-35], and th en further developed for II-VI compounds by. W. Faschinger et al. [36-37] and S.B. Zhang et al. [38]. Th is model tries to predict the doping limit in semiconductors without knowin g any compensating mechanism. It uses the conduction band minimum EC and valence band maximum EV, the vacuum energy level to determine Fermi stabilization energy levels. ESI, p,n. At this level, the new creation of defects doesn’t Figure 17 Relative Band Edge Positi ons of Various II-VI Compounds change the free energy of the material, the Fermi level was pinned there, so the doping level cannot be beyond that level. For p type doping, the maximum doping is achieved when EF is equal to ESI, p. In the same way, the maximum n doping is obtained when EF is p=1017 ESI,n ESI,p ZnS CdS ZnSe MgSe CdSe ZnTeMgTeCdTe Energy (eV)

PAGE 45

31 equal to ESI, n. The model predict that material with high electron affinity are relatively easy to be doped as n type, while materials with low Ev will be difficult to be doped as p type. Figure 17 shows the relative band edge positions of various II-VI compounds and calculated ESI, p,n. From this Figure, we can see that CdTe is the only II-VI compound that can be doped as n and p type. ZnS and CdS can’t be doped as p type. The ratio of concentrations between compensation defects Ncomp and active doping concentration Nactive is given by: E E N Nn p SI F active comp ,2 exp / (18) Where ESI,p,n is constant with respect to the va cuum level for all compounds as long as the compensation mechanism is the same. But Y. Marfaing [39] showed that ESI,p,n depends on the material properties and compensation mechanism. 3.2.2 Doping Limitation Mechanism 3.2.2.1 Self-compensation Self-compensation is defined as the intentional dopants were deactive by the spontaneous formation of native defects or s pontaneous change of the position of dopant atom. Native point defects include native va cancies, native interstitials, and native antisites. Native point defects are the most widely accepted as the main cause for the doping limit. These native defects could be in troduced during the crystal growth or by the post heat treatment. For wide band gap material it is energy favorable to the formation of compensating defects, because the recombin ation of free carriers from dopants with oppositely charged compensation defects will lower the total energy of the crystal.

PAGE 46

32 G. Mandel et al. [40] first proposed the self-compensation by spontaneous formed native vacancies. His prediction for n-ZnTe, n-CdTe and n GaAs is well in agreement with the experimental results. Later studies show that the number of possible native defects is much less<=1016cm-3. For the doped II-VI compounds, a high concentration of vacancies was observed by some researchers, bu t detail analysis showed they are dopantvacancy pairs, not isolated vacancies. U. V. Desnica [41] calculated the dopi ng efficiency limited by native vacancies compensation showed in Figure 18. In his ca lculation, he used Phillips-Van Vechlen’s two bands theory to calculate the vacancy formation energies and enthalpies. The calculated doping efficiency exactly followed the trend observed in experiments. We can also see that n type doping becomes easier a nd p type doping becomes more difficult with the increasing of the ratio of the atom size between II-VI group elements. This can be explained by the following: The formation en ergy of vacancy depends on the size of the vacancy. The larger the ratio, the bigger the ratio of the concentrations of the donor-like native vacancies to the acceptor-like native v acancies, leading to easier n doping. But the calculation is not sufficient to explain the ex tent of the experime ntally observed doping problem. This indicated that the isolated vacancies are not the only reason for doping problem. Native interstitials and antisites are the other two kinds of defects that could limit the doping effect. Up to today, there are no pos itive experimental re sults that can prove they play important roles for the doping limit in II-VI compounds. The formation of dopant-native vacancy complexes is another kind of selfcompensation. There are a number of experiment al results in II-VI compounds that could

PAGE 47

33 be explained by the dopant-native vacancy pair s. This self-compensation mechanism well explained the experimental results for In doped CdS [42], N doped ZnSe [43], n type ZnSe [44] and In doped CdTe [45]. It looks like it plays an important role for the doping limit. Figure 18 Calculated Maximum Doping Effi ciency Limited by Self-Compensation with Native Vacancy Dopant atoms usually have a different s ite and a different charge state. These dopant atoms disturb the periodicity of the crys tal structure. If the disturbance is strong Covalent radius ration rc(M)/rc(X) p/[A] n/[D]

PAGE 48

34 enough, a band near the dopant atom can break and form a charged compensation defect center(AX or DX). The AX center forms when two host anion-cation bonds break and an anion-anion bond forms between second neighbor atoms. This rebinding process leads to the conversion of the standard substitution atom into the dopant displaced from its lattice site, resulting in the converted dopant atom. a-+2h AX+ (19) Equation 19 describes the reaction in which an acceptor captures two holes and forms an AX center. The calculated formation energy of AX center for N, P, As and Sb acceptor dopants for CdTe are -0.1, 0.25, 0, and -0.3, respectively. Negative values imply the formation of an AX center is energetically favorable. The AX formation energy for P in CdTe has the highest value, so P could be a better dopant compared with other dopants. Further calculation by S. B. Zha ng et al. [46] indicates th at the formation of the AX center is not a limiting factor for p-type doping in CdTe. The DX center has been observed for a number of differe nt n type dopants in CdTe. 3.2.2.2 Auto Compensation Besides intrinsic defect compen sation for impurity doping, the doping concentration is also limited by the impurity itself. This is called auto compensation. For example, Cu could be either an acceptor or donor depending on the position it occupies. For CuCd, it is a p-type dopant. For Cui, it is an n-type dopant. If the formation energy of CuCd is lower than that of Cui, it acts as a p-type dopant. Vice versa, it acts as an n-type dopant. The formation energy will change with the changing of Fermi energy, which means if the impurity acts as a p-type dop ant initially, with the increasing doping

PAGE 49

35 concentration, the donor formation energy will increase or decrease, if it becomes lower than the formation energy of an acceptor, the further increasing of hole doping concentration will be limited. If the donor form ation energy is higher than the acceptor formation energy, no limit exists for p-type doping under this condition. Calculation shows the formation energy of Cui is always higher than that of CuCd. From this point of view, Cu could be a good dopant under the Te rich condition. 3.2.2.3 Solubility Limits of Dopants The solubility limit is the ultimate doping limit. For the wide band gap materials, the solubility of most dopants is relatively low due to the high incorporation energy. The solubility also depends on Fermi level. The cl oser the Fermi level to the band, the lower the solubility. Usually, the be st conditions for su ppressing compensating defects are also the worst conditions for solubility. Solubility calculation of Na, Li and N in ZnSe and ZnTe performed by D.B Laks at al. [47,48] well followed the experimental results. 3.2.2.4 Ionization Energy Level Limit To achieve a high carrier concentration, the dopant must have low ionization energy so that it can release electrons or holes at a normal operating temperature. Tables 1 and 2 list the ionization energy levels of acceptors and donors in Znand Cdchalcogenides, respectively. Comparing figure 17 tables 1 and 2, they fit each other very well. When the phenomenon logical model predic ts that one material is easily dopable, the experimental results show low ionization energy levels.

PAGE 50

36Table 1 Ionization Energies (in mV) of Va rious Foreign Donors in Different Zn-and Cd-Chalcodenides [41] ZnTe ZnSe ZnS CdTe CdSe CdS B 18.5 18.4 18.3 25.6 26 Al 25.6 26.3 26 100 74 14.05 Ga 27.2 27.0 27.9 400 13.88 33.1 In 28.2 28.1 28.9 400-600 14.5 14.15 33.8 F 28.2 28.8 29.3 13.67 35.1 Cl 20.1 26.2 26.1 26.9 14.1 14.48 14 32.7 Br 26.8 32.5 I 23.95 30.4 600 32.1 Lii 21 17 202 13.9 28 Nai 20 Ni 29.11 26 Table 2 Ionization Energies (in mV) of Va rious Foreign Acceptors in Different Znand CdChalcodenides [41] ZnTe ZnSe ZnS CdTe CdSe CdS Li 60.5 61 114 113 111 150 57.8 58 1098 1656 Na 62.8 126 128 190 58.8 28.7 1966 Cu 148 149 146 140 650 1250 147 146 140 1100 860

PAGE 51

37Table 2 (Continued) ZnTe ZnSe ZnS CdTe CdSe CdS Ag 121 123 430 113 720 113 108 107.5 92 123 260 Au 277 550 263 N 10015 110 111 56 21-31 O 80 110 46 116 P 63.5 85 85.3 550 600-700 60 68.2 50 836 120 600 As 79 78.5 110 113 92 750 Sb 57 280 Bi 300 Table 3 Calculated Formation Energy [46] Defect Formation energy(eV) Defect Formation energy(eV) VCd 2.67 CdTe 3.92 VTe 3.24 TeCd 3.70 Tei a 3.52 Cdi a 2.26 Tei c 3.41 Cdi c 2.04 NaCd 0.45 AlCd 1.17 CuCd 1.31 GaCd 1.23 AgCd 1.32 InCd 1.23 AuCd 1.30 FTe -0.08 NTe 2.62 ClTe 0.48 PTe 1.83 BrTe 0.62 AsTe 1.68 ITe 0.99 SbTe 1.72 Cui a 2.14 BiTe 1.96 Cui c 2.24 Nai a 0.60 Nai c 0.45 a: sites surrounded by anions c: sites surrounded by cations

PAGE 52

383.2.3 Defects in CdTe The defects in CdTe can exist in the fo rm of point defects or complexes. The possible native point defects include ne utral or ionized Cd interstitial Cdi, Cd vacancy Vcd, Te interstitial Tei, Te vacancy VTe, Te anti-site TeCd, and Cd anti-site CdTe. Point defects can also form defect associations, i.e. (TeCd-VCd). It can also form complexes with impurities, i.e., (ClTe-VCd)-. Some defects act like donors, such as Cdi, VTe and TeCd. Some act like acceptors, such as Tei, VCd. Cdi is the main donor def ect under the Cd rich condition, while VCd is often thought to be the ma in acceptor. Table 3 shows the calculated defect formation energy. For pol ycrystalline CdTe, the grain boundary acts like a donor. 3.2.4 Acceptors in CdTe Acceptor centers are obtained when group IA and IB elements are substituted for Cd atoms or when group V elements are substituted for Te atoms. They can also be donors when group IA and IB are substituted for Te atoms or being interstitial or when group V elements are substituted for Cd atoms. Which positions they occupy depend on the defect formation energy. 3.2.4.1 Group IA Elements 3.2.4.1.1 Lithium M. Restle et al. studied the Li atoms pos ition in CdTe using ion implantation [49]. They found Li atoms are immobile and occupy ma inly tetrahedral inte rstitial lattice sites below 130k and about 10% substitution Li abov e 130 K. They found a strong interaction

PAGE 53

39 of interstitial Li with vacancy defects resu lting in a formation of substitution Li until 470K. Above 470K, Li is unstable a nd diffuses out to the surface. H. Pauli et al. studied Li doped CdTe by hot-wall-beam epitaxy [50]. Because Li itself is a highly reactive element, they chose lithium nitride(Li3N) as a dopant source. In this study, they studied the e ffect of the dopant source temper ature, and the effect of the illumination. They found the hole concentration increased from 4x1016 to 2x1018cm-3 when the Li3N temperature increased from 350 to 550 C, and the mobility decreased from 100cm2/Vs to 30cm2/Vs. They also found illumination significantly enhances the doping efficiency below 500 C. L. Svob and C. Grattepain [51] studied the diffusion of lithium at 300C in CdTe. They found there were two diffusion forms, one with high diffusion coefficient depends on the cadmium vacancy concentration, while the other with low diffusion coefficient doesn’t. The fast one is interstitial Lii, and the slow one is substitution lithium, LiCd. L. Svob and Y. Marfaing [52] proposed a model of diffusion of lithium based on simultaneous diffusion of substitution Li accepto r and interstitial Li donors. This well explained the observation of the aging effect of Li doped CdTe. 3.2.4.1.2 Silver Silver can be easily introduced in high c oncentration as a p-type dopant in CdTe. Interstitial Ag is a fast diffusing impurity and leads to a uniform occupation of Cd vacancy even at room temperature. The di ffusion of Ag atoms depends on the external

PAGE 54

40 partial pressures of Cd and Te. Under Te pressure, the concentration of VCd is higher, the Ag atoms occupy the VCd position and show slow diffusion. H. Maqsood et al. studied the Ag-doped Cd Te fabricated by C SS [53]. Different doping concentrations were achie ved by dipping CdTe in AgNO3-H2O solution for different times as 0, 30s, 5 min, 10min, 20min, and 40minutes. The resulting Ag compositions (at%) were 0, 2.63, 8.23, 9.59, 12.25 and 17.6. The highest hole concentration exceeded 1017 cm-3 and mobility was 78 cm2/Vs. The resistivity was 1.69cm. The systematic decrease in resistivity wa s observed at room te mperature due to Ag doping. At high Ag concentration, the substitution AgCd acceptors and Agi donors are formed which passivate each other. H. Wo lf proposed self-compensation by lattice relaxation and rebinding, transforming 50% of the acceptor impurities into donor-like defects, which are called AXor double-broken-bond(DBB) centers [54]. 3.2.4.1.3 Gold Au doped CdTe has been investigated el ectrically and optically by MR Lorenz and B. Segall [55]. They measured the Au le vel at 0.34 eV and 0.4 eV above the valence band, respectively. W. Akutagawa et al studied Au doped CdTe [56]. Au was incorporated by diffusion into CdTe crystals from the surface with an Au thickness of about 10m by evaporation. They found subst itution Au solubility increases with both increasing temperature and decreasing PCd. The net Au concentration is about 1017 cm-3. But the Hall measurement shows the carrier concentration is less than 1010cm-3, indicating that substantial charge self-compensation o ccurred through the cadmium

PAGE 55

41 interstitial in the heavily doped CdTe. They found 50~80% of the Au atoms were in substitution. The other fraction may consist e ither of precipitated or interstitial Au. 3.2.4.1.4 Copper In single crystal CdTe, Cu is consider ed to exist as an interstitial Cui + or substitute Cd atom to form a deep acceptor state (CuCd -) with activation energy of 0.28-0.34eV. The mobility of interstitial Cu is very high. C opper can also be able to form complexes between Cu+ and VCd. These complexes are relatively shallow acceptors. Copper plays two important roles for CdTe/CdS solar cells. As an acceptor dopant, it dopes CdTe surface layer as p+ layer to assist the form ation of an ohmic contact by tunneling contact. But intersti tial Cu diffuses fast and accum ulates on the interface of CdTe/CdS and is believed to be the main r eason of degradation. But the study at USF [57], in which the Cu was introduced direc tly on CdS surface, showed that a suitable amount of Cu in the in terface also improves the device performance. C. R. Corwine et al. found the most pr obable Cu-related defect was a deep donor and may be a doubly ionized Cu interstitial ion (Cu2+)[58]. The highest concentration was at the CdTe/CdS interface and the substitu tion Cu impurities (acceptors) were only found in the bulk of the CdTe. T.D. Dzhafarov et al. studied the therma l and photo stimulated diffusion of Cu in CdTe thin films [59]. They found the diffusi on coefficient of photo stimulated diffusion of Cu is larger than that of thermal di ffusion by 2-4 times. For thermal diffusion: Dt is 7.3x10-7exp (-0.33/kT). For photo stimulated diffusion, Dph is 4.7x10-8exp(-0.20/kT). The activation energy for photo diffusion is smaller (0.2eV) than that of thermal diffusion

PAGE 56

42 (0.33eV). Hall measurements show the hole concentration is 3x1010cm-3, =40cm2/Vs, and =0.5 cm. 3.2.4.1.5 Disadvantages of Group I Elements B. Reinhold [60] and S.B. Zhang [61] f ound the dominant intrinsic defect that compensates acceptors is Cdi. In order to avoid the intrinsic compensation, p-type doping should be carried out at the Te-rich condition because of the relatively large formation energy of Cdi at this condition. IA and IB elemen ts have the lowest formation energy under Te-rich condition, so it is relatively easier to get highe r doping concentrations compared with group V elements. On the other hand, the fast diffusion of Ag, Li, Na and Cu at room temperature has been widely repor ted. N. V. Agrinskaya [62] found that the electrical conductivity and carrier concentration of Li and Na doped crystals decreased significantly after aging at room temperature for several months. For Li, Na, and Ag, this aging effect is attributed to the instability of these elements occupying Cd sites. This instability described by Mcd Mi+VCd. Further study by Y. Ishika wa et al. [63] found that besides the instability of these elements on Cd site, the formation of various complexes among native defects and impurities is also responsible for this aging effect. The diffusion of Cu in CdTe is believed to be responsible for the instability of CdTe/CdS solar cell’s performance. S. E. As her et al. [64] found a significant amount of copper diffused through the CdTe/CdS layer to compensate the shallow donor levels with deep acceptors in CdS.

PAGE 57

433.2.4.2 Group V Dopants (N, P, As, Sb, Bi) Su-Huai Wei et al. studied the doping limit in CdTe [65]. They found the calculated (o/-) transition energy levels are at 0.01, 0.05, 0.1, 0.23 and 0.3 eV for N, P, As, Sb, and Bi, respectively. The sh allow transition energy levels for NTe and PTe indicate that they could be very good dopants for CdTe but their defect formation energies are large because of the large site mismatch. Th eir studies showed that the doping limit is due to the very low solubility at the Te rich condition. 3.2.4.2.1 Antimony Sb is a suitable dopant because of its low ionization ener gy, low diffusion constant, and can easily incorporated into CdTe. At 400 C for 30 minutes, the diffusion length is about 1 m. A. Picos-Vcga et al. studied physical properties of CdTe-Sb thin films by RF sputtering [66]. Th ey found that the increase in Sb content in the film is accompanied by a decrease in the Cd and Te content. Both decreasing at approximately the same rate, this is evidence of the amphoter ic character of Sb substituting for Cd and Te in the CdTe lattice with about the same pr obability. If the concentration of Sb in CdTe is high (around 1019cm-3), the CdTe is a semi-insulating material. At low concentration, the Sb substitutes Te as an acceptor. By in creasing the concentration of Sb, more Sb atoms substitute Cd and compensate the acceptors of SbTe, SbTeSbCd was formed. A vapor phase technique has been tried for Sb doped CdTe, but it was unsuccessful. Sb is not incorporated at lo w vapor pressure, wher e at high pressure metallic Sb precipitates.

PAGE 58

44 R. N. Bichnell et al. [67] reported the first Sb doped p-type CdTe film by photo assisted MBE process. The film is hi ghly conductive. The hole concentration is 5x1016cm-3 and mobility is 40~45cm/Vs. The key point in their experiment is that they used an argon laser to assist the chemical reac tion. They attributed this to the argon laser providing the energy needed to surmount surface potential energy barriers at the growing film surface, thus increasing the degr ee of dopant activation. J. D. Benson [68] found laser-illumination enhances Te-desorption and produces more sites for Sb incorporation. Complementary Auger studies indicate a 20% in crease in Sb adsorption due to Sb filling both photons desorbed Te-sites and some equilibrium Te-site vacancies. O.Zelaya-Angel et al. studied the influen ce of the growth parameters of Sb doped CdTe [69]. In their study, Sb doped CdTe th in films were carried out with targets containing different amounts of antimony (CT:0, 2.5, 10 and 20 at.%). They found with low CT values the structure is a mixture of zinc blended ZB and hexagonal wurzite phases. Sb atoms in the CdTe lattice favor the stable ZB Phase of CdTe, and W is considered the metastable crysta lline phase. The re sistivity is 9x105 cm with 10% of CT at 100C, but a SIMS measurement shows the Sb is less than 1%. The resistivity decreases with the increasing of CT FWHM decreases when CT increases from 0 to 10% indicating the increasi ng of grain size. It was also found that Voc and Jsc increase with the increasing of substrate temperature. The excimer laser doping technique has b een used to obtain a heavy doping in IIVI semiconductors. Using excimer laser, th e semiconductor surface fi rst melts and then cools down rapidly. The bulk of the semiconducto r is not affected. Y.Ha tanaka et al. [70] reported Sb doped CdTe by excimer laser. The best doping characteristics are 0.027 cm

PAGE 59

45 resistivity, 5X1018 cm-3 hole concentration, and 45 cm2/Vs mobility at 60 mJ/cm2 irradiation. P.K. Pandey et al. [71] reported they developed p-type CdTe by a non-aqueous bath in ethylene glyeol. Their study show s Sb can reduce oxygen impurity in CdTe. A thermo-emf study shows the film is p-type but no hole carrie r concentration was reported. 3.2.4.2.2 Phosphorous Wei and Zhang have calculated p-type dopa nt energies in CdTe. N and P are the most attractive dopant elements because N a nd P create very shallow acceptor levels in CdTe. So if one can enhance the inco rporation of N and P in CdTe though nonequilibrium processes, the doping concentratio n could be significantl y enhanced. Table 4 summarizes the results of phosphorous doping of CdTe reported by various investigation groups. The highest hole concentration is 5x1019cm-3, reported by H. L. Hwang and coworkers [72], who used P+ implantation followed by pulsed electron beam annealing. Phosphorus atoms are often associated w ith native defects and form complexes and precipitates, such as Cd3P2, P2Te3. It might exist as Pi, PTe, and PCd in CdTe. It is an acceptor when it occupies the in terstitial site and surrounded by Cd atoms, or it occupies Te site. When it is surrounded by tellurium atoms, or occupies Cd site it becomes a donor. Based on Kroger’s model, the compensation mechanism can be described as : the shallow acceptors PTe are compensated by Cdi at high cadmium pressure annealing, while at low cadmium pressure annealing, the shallow acceptors VCd are compensated by PCd.

PAGE 60

46 At high doping concentration, the neutral or charged co mplexes become dominant leading to a large degree of self-compensation. A. D. Compan et al. [73] studied phosphorus doped CdTe films by reactive sputtering from CdTe/Cd3P2. In their study, they compared three structur e solar cells’ performance: CdS/CdTe/Au. CdS/CdTe:P /Au, and CdS/CdTe/CdTe:P/Au. They observed the effect of CdTe doping in impr oving device performance. A stability study shows the uniformly doped CdTe:P device has the poorest stability du e to the degradation in both Jsc and FF. The other two structure devices show almost same stability but with a different degradation mechanism. For the CdS/ CdTe:P/Au structure devi ce, this is due to degradation in both Voc and FF, while for CdS/CdTe/Au stru cture device, it is only due to the degradation in Voc. Table 4 Summarizes The Electrical Prop erties of Phosphorous-Doped CdTe and The Acceptor Energy Level of P Reported by Several Authors Dopant concentration (cm-3) Annealing condition (atm) Hole concentration (cm-3) Energy level EA (eV) 6.0x1016 Pcd, 0.01 6.0x1016 0.034 5.0x1019 Pcd, 0.01 1.4x1017 0.03-0.038 1.0x1019 Pcd, low 8.0x1016 0.05 1.0x1019 6.7x1016 0.11 1.2x1018 a 1.2x1016 0.5 2.5x1018 Pcd, 5x10-13 2.2x1017 0.051 Pcd, sat 3.0x1017 0.052 6.5x1014 b (cm-2) 600 5.2x1012(cm-2) 0.03 0.03 atomic percent Pcd, 2x10-13 2.2x1017 0.066 5.0x1019 a Pcd, 0.01 5.0x1017 0.037 5.0x1015 a(cm-2) Laser 1.6x1017 5.0x1015 a(cm-2) PEB 5.0x1019 a) p+ implantation, b) p+ and Cd+ co-implantation

PAGE 61

473.2.4.2.3 Arsenic A USF group [74] tried to use arsine as a dopant source to dope CdTe by MOCVD. The doping efficiency is very low an d most incorporated arsenic atoms were electrically inactive due to compensation and complex formation. S. K. Ghandhi et al. [75] first re ported arsenic doped CdTe films grown by organometallic vapor epitaxy using arsine as th e dopant gas. They studied the effect of arsine flow rate over the range from 8x10-9 to 3.5x10-7 atm. The doping concentration shows a linear relationship up to 2x1017 cm-3 and saturates as the arsine flow is increased further. The carrier concentration also de pends on the partial pr essure of DMCd and DETe. The acceptor ionization energy is about 624 meV. J. M. Arias el al [76] also reported arsenic doped CdTe by photoassisted MBE. The dopant source is As4. Acceptor concentration is in the range of 1014 – 1016 cm-3 and mobility is about 65 cm2/Vs. The acceptor ionization energy is 11816 meV. The doping efficiency is about 13%. M. Ekawa et al. found higher doping of As could be achieved by Cd rich [77]. At this condition, the sticking rate of As onto Cd increases, which leads to an effective incorporation of As into Te site. Hole concentration increased linearly from 2x1015 to 3x1016 cm-3 with the increasing flow rate of triethylarsine. It was reported that the CdTe doped with arsenic from AsH3 source contains neutral arsenic-hydrogen pairs. This leads to the low doping effi ciency in as-grown films. L.Svob et al. [78] found that short annealing leaded to the increase of hole concentration by dissociation of neutral As-H pairs. Almost full activation of the incorporated arsenic

PAGE 62

48 was achieved at low concentration, and the maximum hole concentration was limited to (1.5-2)x1017 cm-3. 3.2.4.2.4 Nitrogen K.Saminadayar et al. investigated the nitrogen doping efficiency in MBE of TeBased II-VI compounds [79]. They found the ni trogen doping efficiency is decreased as the decreasing of Zn content and the increasin g of Mg content. One of the reasons they proposed is that the length of Zn-Te bonds is smaller than that of Mg-Te and Cd-Te bonds, thus the lattice distortion induced by NTe is smaller and resulting in high doping efficiency. The other explanation they gave is by comparison of the formation enthalpies of the nitride compounds. Zn3N2 has lowest formation enthalpy while Mg3Te2 has the highest one. So there is less likely to form Zn3N2 resulting a higher doping efficiency. The highest doping concentr ation for CdTe-is 8X1016 cm-3. M. Niraula et al. studied N-doped CdTe using low-pressure plasma-radicalassisted MOCVD[80]. In this study, DMCd an d DETe were used as the sources of Cd and Te. By changing the ratio of Cd/Te, the resistivity of CdTe film first decreased with the increasing ratio of Cd/Te, but then incr eased with further increasing the ratio of Cd/Te beyond 4. With the ratio of Cd/Te betw een 1 and 2, the Cd rich condition makes more Te vacancies where N-atoms can easily reside. When N2 gas was used to produce nitrogen plasma source, the achieved CdTe film resistivity was not low enough to perform a Hall measurement. Due to the fact that it is difficult to dissociate nitrogen molecules into atomic nitrogen, by using a mmonia, the high conducti ve p type layer is

PAGE 63

49 successfully achieved. The highest doping concentration is 5x1018 cm-3 with the mobility 8.24cm2/V.s. N. Romeo et al. [81] reporte d N doped p-type CdTe films with resistivity of about 100 cm. CdTe films were deposited by co-eva porating cadmium and tellurium from two separate sources. During deposition, neutral hi gh energy nitrogen atoms were shot to the substrate from an atomic gun. The calc ulated activation energy is 55 meV. K. A. Dhese et al. [82] tried to use RF nitrogen plasma to get p-type nitrogen doped CdTe by MBE. Although PL measurements show nitrogen is incorporated in CdTe films and confirmed by SIMS measurements, electrical measurements show CdTe films are n-type with donor concentrations < 1015 cm-3. 3.2.4.2.5 Bismuth Recently O. Vigil-Galan et al. [83] st udied the physical properties of Bi doped CdTe. The CdTe films were deposited by CS VT using powdered CdTe:Bi crystals with nominal Bi doping conc entration from 1017 to 8x1018 cm-3. They found the resistivities were 1x108, 1x1010, and 6x105 cm, respectively, and for undoped, 1x1017, and 8x1018 cm-3. PL measurements show Bi atoms substitute Cd positions at low doping concentration. The VCd is compensated by BiCd, leading to a semi-ins ulating film. At high concentrations, Bi atoms occupy Te positions and the resulting film is p type. The resistivity can be reduced to 102-103 cm. 3.3 How to Improve the Performance of CdS/CdTe Table 5 compared the world-record CdTe cell and ideal one. We can see that Jsc is already 88% of theoretical value, while Voc is only 79%. It will be very difficult to

PAGE 64

50 increase Jsc and the space for Voc is bigger. It is relative easier to increase Voc than to increase Jsc. Table 5 Comparison of World Record and Ideal Solar Cell Word record Ideal Record/Ideal Voc(mV) 847 1070 79% Jsc(mA/cm2) 25.88 29.5 88% FF(%) 75.5 89 85% (%) 16.5 29.2 57% 500 600 700 800 900 -0.2-0.10.00.10.20.30.40.50.60.7Back Barrier Height [eV]Voltage [mV] Figure 19 The Effect of Back Barrier on Voc There are three main factors to limit the Voc. First, the back barrier. Figure 17 shows the Voc as the function of back barrier( B). At low back barrier(<0.35eV), the Voc is not significantly affected by back barrier height. When the back barrier is more than 0.35 eV, Voc is decreased a lot. When back barrier is less than 0, which means there is no back barrier exists and the band bending is up near the back surface, often referred to as an electron reflector. Even a small electron reflector, the Voc should increase

PAGE 65

51 significantly. There are two way to achieve ohm ic or near ohmic contact. One way is by heavily doped near back surface to get tunneling contact. Th e other way is by using high work function materials. The possible high wo rk function materials are TiSe2 and Vse2 ZnTe should work as an elec tron reflector material based on theoretical calculation. The second factor is the low hole concen tration. Figure 18 shows the simulated Voc results with increased hole co ncentration. We can see the Voc increases with the increasing of doping concentration. T ypical hole concentration is about 1x1014 cm-3 for the present CdTe solar cells. In this disserta tion, Sb is used as a dopant to increase the doping concentration for CdTe film. 700 800 900 1000 1.0E+121.0E+131.0E+141.0E+151.0E+161.0E+17 Doping Concentration[cm-3]Voltage [mV] Figure 20 Effect of Doping Concentration on Voc The third factor is the uniformity of CdS, CdTe, and back contact. If the CdS film is not uniform, for an example, there exit s pinholes, a weak diode between CdTe and TCO will form and reduce the Voc. To eliminate pinholes on CdS film, the thickness of CdS film has to be thicker, and thicker CdS will reduce Jsc. There is a trade-off for high Voc and Jsc for CdS. High resistive buffer laye r between TCO and CdS can reduce the

PAGE 66

52 required thickness of CdS and maintain high Voc. Undoped SnO2, In2O3 and Zn-Sn-O have been used as buffer layers. Further study on new buffer layer material has to be done to improve the cells’ performance. The nonuniform back contact will result in reach-through diode effect. This nonuniformity could be due to lo calized surface state, the diff usion of impurities, or the contamination from the process. Great care has to be taken prior to each deposition step to avoid contamination.

PAGE 67

53 CHAPTER 4 EXPERIMENTAL METHODS The basic structure of the CdS/CdTe solar ce lls used in this th esis is glass (7059) TCO/CdS/CdTe/back contact as shown in fi gure 19. The front contact TCO is ITO/SnO2 or bilayer SnO2. CdS is kept constant in all devices. All the other layers were changed as the experiment needed. Back contact CdTe absorber layer (4~10 m) CdS window layer In TCO Glass substrate (7059) Light Figure 21 Configuration of CdTe Solar Cells as Used in This Dissertation 4.1 Chemical Bath Deposition CdS forms the n-type junction partner of CdTe solar cells. Its bandgap is 2.4 eV, this allows the low energy light (above 510nm) can pass through it to reach CdTe absorber layer to generate photocurrent. Th e light absorbed by Cd S can not contribute

PAGE 68

54 photocurrent, so the thickness of CdS should be as low as possible. On the other hand, thin CdS leads to low Voc due to the pinholes of CdS. Cd S can be deposited by chemical bath deposition (CBD), closed-spaced sublim ation (CSS) and RF sputtering. CBD gives Figure 22 Diagram of CBD Set Up the best result of CdTe sola r cell. In this study, the Cd S is deposited by CBD. Cadmium acetate is used as cadmium source. Thiourea is used as sulfur source, ammonium acetate (NH4AC) and ammonium hydroxide (NH4OH) are used as buffers to control the PH value. The SnO2: F or ITO coated glass substrates are immersed in DI water. The SnO2: F or ITO should face up to eliminate bubbles on the surface. The temperature is maintained at 80-90C during the deposition. Figure 22 shows the diagram of the set up. Solution A is prepared by mixing sp ecific amount of CdAc, NH4AC and NH4OH solution. For every several minutes, 0.00003M CdAc and 0.000 12M thiourea are added into reaction Substrate Double jacketed beaker Stirring bar Hot plate HeatedFluid

PAGE 69

55 container. The thickness of CdS can be changed by varying the deposition time. J. Herrero et al. [84] proposed th e possible reactions as follows: Cd(CH3COO)2 Cd2++2CH3COO(20) NH3+HOH NH4 ++OH(21) Cd(NH3)4 2+ +2OH[Cd(OH)2(NH3)2]+2NH3 (22) [Cd(OH)2(NH3)2]+SC(NH2)2 [Cd(OH)2(NH3)2SC(NH2)2 (23) [Cd(OH)2(NH3)2SC(NH2)2] CdS(S)+CN3H5+NH3+2HOH (24) 4.2 Close Space Sublimation Figure 23 Schematic Diagram of CSS Deposition Chamber The CdTe absorber film was deposited by close space sublimation (CSS). The deposition is based on the followi ng reversible dissociation of CdTe at high temperature: 2CdTe(s) 2Cd(g) +Te2(g) (25) The CdTe source and substrat e are typically separated by a small space of several millimeters. The source temperature is maintained at around 100C higher than substrate temperature. At high temperature, the CdTe solid will dissociate into its elements gas (Cd

PAGE 70

56 and Te), then they recombine on the substrate surface because of its lower temperature. In this research, the CdTe chunks were first sublimated onto a 7059 glass substrate to get a very thick CdTe layer. This is so called sour ce substrate in this research. Before CdTe deposition, the CdS film was annealed at 390C for 10 minutes in H2 ambient to remove unpreferred oxide compound like CdO. Halogen la mps are used to heat up the source and substrate. The temperature was controlled by OMEGA temperature controller. Figure 23 shows the diagram of CSS set up. 4.3 Sputtering Deposition The moly back contact and antimony we re deposited by sputtering deposition. Sputtering deposition is widely used in semiconductor industry. The advantages of sputtering deposition include: low temperatur e, good adhesion and can be used for any material, good stoichiometric. Disadvantage in cludes: surface damage and stress. It can be classified as reactive spu ttering, which is by adding a ga s which reacts with a sputter material in the presence of the plasma, and non-reactive sputtering, which is the gas doesn’t react with the sputtered material. It can also classify as RF and DC sputtering. RF sputtering is suitable for both insulate and conductive materials and DC sputtering is only suitable for conductive materials. The basic theory is at high enough electric fi led, part of gas molecules are ionized, and the ionized particles are accelerated by the electric filed, these high speed particles bombard the target, knock loose the target surf ace material, transfer its energy to target molecules. When the target molecules impi nge the substrate, they diffuse along the substrate surface and settle down to form thin film.

PAGE 71

574.4 Device Characterization The fabricated solar cells were characteri zed by light and dark current-voltage (IV) measurements and spectral response measurements. 4.4.1 I-V Measurement All I-V measurements are done by using f our-terminal Kelvin connections. This minimizes measurement errors by eliminating voltage-drop losses that result from resistances due to cables, connections, series resistance. The data was collected by a lab view program, whic h can calculate Voc, FF, and efficiency automatically. The power supply is a Keithely 2410 source meter. The sun light simulator is calibrated using standard silicon solar cell. From light and da rk I-V plots, all cells parameters can be determined, such as Voc, Jsc, FF, Rs, Rsh, J0 and A. Color I-V measurement was also performed for selected devices. In this m easurement, Oriel color filters with 20nm bandwidth were used. The light intensity was adjusted to one sun intensity based on the current density calculated from SR for each wave length. 4.4.2 Spectral Response Oriel monochromater (model 74100) is used for spectral response measurement. The light source is GE400w/120v Quartz Line lamp (model 43707). The light intensity is calibrated using a standard si licon solar cell calibrated by th e National Renewable Energy Laboratory (NREL). The quantum efficiency is calculated by the following equation: QE I I QEference ference Device SampleRe Re (26) The device current density is calculated by inte grating the product of QE and the standard cell current density.

PAGE 72

584.4.3 Capacitance-Voltage Measurement The C-V measurement was performe d on selected devices. From C-V measurement, the doping concentration, dopi ng profile and depletion width can be calculated. It relies on the fact that the depl etion width of a revers e-biased P-N junction device depends on the applied voltage. From the definition of capacitance: dV dQ Cs (27) where dw w N q Q dA S we get: dV dw w N q CA/ (28) The depletion width can be given: N A q V V bi s w22 1 (29) Differentiating equation 28 with respect to vol tage and substituting dw/dV into equation 27 gives: dV c d A qK w Ns A/ ) / 1 ( 22 2 0 (30) Ideally, for a one-sided abrupt junction, a plot of 1/C2 versus V should be a straight line. In this measurement, a HP4194 Gain Phase Analyzer is used.

PAGE 73

59 CHAPTER 5 RESULTS AND DISCUSSION 5.1 Optimize the HgTe: Cu Do ped Graphite Back Contact The HgTe: Cu doped graphite paste is the standard back contact in our lab. It gives a good back contact resulting in good pe rformance of CdTe/C dS solar cells. The previous used graphite pa ste was Acheson Electrodag 00 3, but the company stopped producing this product. The new one is El ectrodag 502, which has the same nominal properties. But actually it has different pr operties. We cannot get good contact using same conditions as Electrodag 003.New conditions have to be developed. The annealing temperature effect, the annealing time effect, and the HgTe: Cu concentration effect were studied and optimized. 5.1.1 Annealing Temperature Effect The previous recipe was 37.5% wt. Hg Te:Cu doped graphite paste, and the annealing conditions were 270C and 25 minutes The same recipe was used first and then different annealing temperatures were used. Figure 24 shows the summary of the effect of the annealing temperature. From th e figure we can see the annealing temperature plays an important role. It affects the amount of copper in CdTe, leading to different doping concentrations. The JV characteristics for different annealing temperatures are shown in Figure 26. With the increasing annea ling temperature, the turn-on voltage shifts

PAGE 74

60 towards right which means the barrier across the main junction is increasing. This barrier is an indicator of doping c oncentration. This indicate s the doping concentration is increased with increasing temperat ure. The cells annealed below 270 C show similar series resistance, but the cell annealed at 270 C shows higher series resistance although the dark J-V shows it has higher doping concen tration. There are tw o possible reasons for this behavior. One reason is the high anne aling temperature introduces excess Cu to the extend that it compensates the p-type dopant. The other is the high temperature annealing leads to the detriment of the interface between CdTe and back contact.Compared with the 720 740 760 780 800 820 840 180200230270 Annealing Temperature [C]Voltage [mV]0 10 20 30 40 50 60 70 80FF [%] VOC FF Figure 24 The Summary of the Effect of Temperature previous graphite paste, the later reason is most probably re sponsible for th e higher series resistance. We also tried the annealing conditions at 270C for 5 minutes. The resulting devices are not good as those ann ealed at lower temperatures. Th is also verified the poor interface is the reason for higher series resist ance. Figure25 shows the SR of the devices annealed at different temperatures. The device annealed at 270C shows significant collection losses resulting in low Jsc and Voc. Based on these results it was concluded that

PAGE 75

61 high temperature has detrimental effect on the interface of the back contact and affect the doping in CdTe by changing the amount of Cu. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. 230C 200C 270C 180C Figure 25 Spectral Response of the Effect of Temperature -0.01 0.01 0.02 0.03 0.04 0.05 0.06 -0.50.00.51.01.5 Voltage [V]Current Density [A/cm2] 230C 200C 270C 180C -0.050 -0.025 0.000 0.025 0.050 0.075 0.100 0.125 0.150 -0.50.00.51.01.5 Voltage [V]Current Density [A/cm2] 230C 200C 270C 180C Figure 26 Light and Dark J-Vs for the De vices Treated at Different Temperatures 5.1.2 Annealing Time Effect From the studies of the effect of the a nnealing temperature, we found the best annealing temperature was 200 C, so the temperature was fixed at 200 C and the

PAGE 76

62 composition was kept as the same as before. Different annealing times (10, 20, 35, 45 and 60 minutes) were performed. 720 740 760 780 800 820 840 1020354560 Annealing Time [min]Voc [mV]0 10 20 30 40 50 60 70 80FF [%] VOC FF Figure 27 The Summary of the Effect of Annealing Time 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength[nm]Q.E. 10min 20min 35min 45min 60min Figure 28 Spectral Response for Devices with Different Annealing Time Figure 27 shows the summary of the devi ces’ performance for different annealing times. The cell annealed for 60 minutes showed highest FF with fairly high Voc. From the spectral response shown in Figure28, the cell annealed for 60 minutes has thin CdS, so

PAGE 77

63 the relatively lower Voc is mainly due to the thin CdS. The band gap was slightly increased with increasing the annealing time. The possible reason could be the formation of CdTe1-xSx. Figure 29 shows the J-V characterist ics of the devices annealed for different times. No shunting was observed, which indicated that even for 60minues annealing the amount of Cu is not excess. -0.01 0.01 0.02 0.03 0.04 0.05 0.06 -0.50.00.51.01.5 Voltage [V]Current Density [A/cm2] 45min 30min 20min 60min 10min -0.050 -0.025 0.000 0.025 0.050 0.075 0.100 0.125 0.150 -0.50.00.51.01.5 Voltage [V]Current Density [A/cm2] 45min 35min 20min 60min 10min Figure 29 Light and Dark J-Vs for the Devices with Different Annealing Time 5.1.3 The Effect of the Concentration of HgTe: Cu The concentration of HgTe: Cu affects the final amount of Cu in CdTe. Four concentrations were used (Table 6) to test the concentration effect. The devices were annealed at 200 for 60 minutes. The summary of the performance was shown in figure30. The performance is not very sensit ive to the concentration of HgTe: Cu. The A gave the best overall performance. Figur e 31 and figure 32 showed the SR and J-V characteristics for the devices with diffe rent contacts. They all have similar Jsc and Voc.

PAGE 78

64 from the dark J-V characteristics, A resu lted best doping of CdTe. Based on all the results, the optimized back contact recipe is using A and anneal at 200 for 60 minutes. Table 6 Different Concentrations Used in This Study A B C D Concentration (% wt.) 16.7 28.6 37.5 44.4 720 740 760 780 800 820 840 ABCD Concentration of HgTe: CuVoc [mV]0 10 20 30 40 50 60 70 80FF [%] VOC FF Figure 30 The Summary of the Eff ect of HgTe:Cu Concentration 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. A B C D Figure 31 Spectral Response for Devices with Different Concentration of HgTe: Cu

PAGE 79

65 -0.01 0.01 0.02 0.03 0.04 0.05 0.06 -0.50.00.51.01.5 Voltage [V]Current Density [A/cm2] A B C D -0.050 -0.025 0.000 0.025 0.050 0.075 0.100 0.125 0.150 -0.50.00.51.01.5 Voltage [V]Current Density [A/cm2] A B C D Figure 32 Light and Dark J-Vs for the De vices with Different Concentration of HgTe: Cu 5.2 High Work Function Materials One of the main issues for CdTe solar ce lls is the back contact. Developing a new back contact material with high work func tion, high conductivity and chemical inertness is one of the key point to further improve ce ll performance. Chemical inertness is desired to limit chemical reactions at the interface w ith CdTe and improve the long term stability, and high work function is required to enable the formation of ohmi c contacts. Potential compound candidates are the selenides TiSe2, VSe2, NbSe2 and TaSe, as well as their corresponding sulfides. Based on the existing deposition facilities the decision was made to first investigate TiSe2 films. There are several methods can be used to deposit TiSe2 films, however, the temperature has to be limited below 450C due to the requirement for good performance of solar cells. Early effort s focused on preparing this compound by selenizing Ti. Metallic Ti films were sputte red on glass slides to a thickness of 200-400

PAGE 80

66 , and subsequently exposed to a Se flux under high vacuum conditions; the selenization process is carried out in a CIGS chamber wh ich contains excess Se. It was observed that the formation of TiSe2 by this method depends on the substrate temperature. For temperatures below 400C no TiSe2 was detected. The Ti films appeared (to the naked eye) to change color at temperatures ar ound 400C. At selenizat ion temperatures of 425C the TiSe2 phase was detected; this can be s een in Figure 33. The XRD spectrum of a Ti selenized film clearly s hows diffraction peaks that have been found to correspond to TiSe2. Figure 33 XRD Spectrum of a Selenized Ti Film However, this is the case on glass substrat es. Efforts to selenize Ti films deposited directly on CdTe have been made but failed. This is due to the fact that the selenization conditions (elevated substrate temperature/high vacuum) cau se the evaporation of Ti film. The possibility of depositing TiSe2 by co-evaporation at lowe r temperatures is being considered.

PAGE 81

675.3 High Resistive Buffer Layers Typical CdTe thin film solar cells have short circuit curren t density of 22-26 mA/cm2, while the maximum theoretical Jsc is 30.5 mA/cm2. The most part of current losses is due to the absorption of CdS at s hort wavelength (<510 nm). The electron-hole pairs in CdS produced by photons do not make any contribution to photocurrent due to essentially zero collection. In order to mi nimize this current loss, the thickness of CdS has to be as thin as possible. However, Voc suffered from the thinning thickness of CdS. This is because a very thin and pinhole free film is difficult to deposit for polycrystalline films. If there is a pinhole, a weak diode will form between CdTe and TCO layer, and seriously harm the cells’ performance. Th e adoption of a high resistive buffer layer between the high conductive TCO and CdS window layer successfully solved or at least mitigate this dilemma stated above. With th e high resistive buffer layer, CdTe will not directly contact with the high conductive oxide and the chance for the weak diode formation is greatly suppressed. The commonly used high resistive buffer layer materials for CdTe/CdS solar cells are intrinsic SnO2, In2O3, and Zn2SnO4. They have excellent thermal and chemical stability and superior electrooptical properties. The buffer la yer used in the world record efficiency devices is Zn2SnO4. We studied the effects of deposition parameters (substrate temperature, Zn-Sn ratio, a nd post annealing after the Zn2SnO4 deposition) in its electrooptical properties. The ZTO film s were co-sputtered from SnO2 and ZnO targets. It was found that the resistivity of films deposited at room temperature incr eases with the Zn/Sn ratio, and it decreases with increasing anneali ng temperature. The highe r resistivity at low annealing temperatures can be attributed to the fact that the as-deposited films are

PAGE 82

68 amorphous and remain as such for annealing temperatures up to 550 C; they begin to crystallize at approximately 575 C [95]. All ZTO films with different Zn/Sn ratios contain both zinc stannate phases: Zn2SnO4 and ZnSnO3 [85]. Films deposited at 400C are polycrystalline films and post annea ling up to 600C didn’t affect the crystal structure. Besides Zn2SnO4 and ZnSnO3, the films deposited at 400 C also contain also contain the binary ZnO2 phase. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. 575C 550C As Dep 600C Figure 34 Spectral Response for Devices with ZTO Buffer Layer Annealed at Different Temperatures Zn2SnO4 has a unique property that other buffer layers don’t have, which is that Zn2SnO4 can consume CdS due to interdiffusion during the fabrication process. The extent of the consumption of CdS depends on the film structure which determined by the post annealing temperature as shown in figur e 34. As can be seen, based on the blue spectral response, the final th ickness of the CdS window layer is smaller for the devices with amorphous ZTO (as-deposited and ann ealed at 550 C). Since the starting CdS

PAGE 83

69 thickness for all these devices wa s identical, it is suggested that the consumption of CdS depends on the structure which is determ ined by the post annealing temperature. Nevertheless, the overall cell performance decreases due to losses in Voc and FF. In this case, ZTO films didn’t serve as effective buffer layers. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. Figure 35 Spectral Response for Devices with ZTO (Zn/Sn=2) Buffer Layer Deposited at 400C For the cells with ZTO films deposited at 400C, the blue spectral response at 450 nm is more than 70% as shown in figure 35. Since the starting thickness of CdS is the same as our baseline cells (QE 50%@450nm), the high blue spectral response clearly reflected the significant consumption of CdS. Moreover, not like the case aforementioned, the Voc and FF are maintained at high values. All these results show that ZTO film onl y acts as an effective buffer layer at a specific composition and deposition conditions. It provides another way to reduce the current loss: Since it is difficult to deposit a very thin and pinhole free film, we can start from a thick film and then consume it through fabrication process.

PAGE 84

705.4 Studies of the Effect of O2 ambient for CSS CdTe 5.4.1 Device Fabrication The substrate is 7059 boro-silicate glass. Th e front contact cons isted of sputtered ITO and MOCVD-deposited undoped SnO2 layer. Cadmium sulfide was deposited using chemical-bath deposition (CBD), followed by the deposition of CdTe by close-spaced sublimation (CSS) in different O2 partial pressure. Before the contact application, the cells were heated treated in the presence of CdCl2. Subsequently the CdTe surface was etched using a Br2/Methanol solution. Two kinds of contacts were used: HgTe:Cu carbon paste and moly. The moly back contact is sput tered at room temperature. Then the cells were subjected to a post deposition annealing in an inert ambient. Solar cells were characterized using dark and lig ht J-V and C-V measurements. 5.4.2 Structure Properties The deposition ambient has a profound effect on film continuity and microstructure. It affects the growth rate, th e grain size and some electronic properties. For thin film solar cell devices, both pinholes and grain size/grain boundary are known to have a direct effect on the conversion effici ency. For the case of CdTe growth by CSS, previous studies showed that the presence of O2 in the processing ambient leads to more uniform coverage, smaller grain size a nd condensed film [86]. Figure 36 shows representative SEM images for films deposite d under the four main process variations. As can be seen, film deposited at the lowest O2 concentration (N2/O2=9/1) has the largest grain size. At higher O2 content, and the grain size is reduced though the decrease is small. Although all films are observed as condensed films, there are more possibilities for

PAGE 85

71 films with larger grains to have more voi ds that can degrade the quality of the main junction and the cell performance. Figure 36 SEM Images of CdTe Films De posited Under Different Conditions N2/O2=9/1 N2/O2=7/3

PAGE 86

72 Figure 36 SEM Images of CdTe Films Deposited under Different Conditions (Continued) N2/O2=5/5 N2/O2=1/9

PAGE 87

735.4.3 O2 Partial Pressure Effects 5.4.3.1 Thicker CdS Figure 37 and figure 38 show the dark J-V and SR charac teristics for devices with the CdTe deposited in different O2 partial pressure. Except the O2 partial pressure, all other processing parameters are all same. From the dark J-V curve, we can see that at low O2 (9/1/)partial pressure, the device shows high J0, and it decreases with increasing O2 partial pressure indicating that the j unction quality is improved. This high J0 is the main reason of the low Voc for the device fabricated at the lowest O2 partial pressure. The SR data shows a small higher QE in blue region for high O2 partial pressure device, suggesting that more Cd S was consumed at higher O2 partial pressure since the starting thickness is same. Ther e is more collection loss in the range of 600-700nm for the device produced at ratio 9/1, while the one fabricated at ratio 7/ 3 has a collection loss at longer wavelengths. Both of losses can attribut e to the poor quality of CdTe layer. The Figure 37: Dark I-V Charac teristics for Devices Fabricated at Various O2 Concentrations

PAGE 88

74 overall current for highest O2 partial pressure is 0.7mA/cm2 higher than the one with lowest O2 content. The higher current is attributed to better collection, which resulted from the higher quality junctions. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. N2/O2=9/1 N2/O2=7/3 N2/O2=5/5 N2/O2=1/9 Figure 38 Spectral Response for Devices Fabricated at Various O2 Concentrations 720 740 760 780 800 820 840 9/17/35/51/9 Ratio of N2/O2Voltage [mV]0 10 20 30 40 50 60 70 80FF [%] VOC FF Figure 39 Summary Results for Devi ces Fabricated at Various O2 Concentrations Figure 39 shows the Voc and FF for the devices. It shows that performance increases and more uniform with increasing O2 partial pressure. The improvement of Voc

PAGE 89

75 is mainly due to the decrease of th e saturation current with increasing O2 partial pressure. It can also be observed th e saturation effect of O2. From 5/5 to 1/9, the performance difference is very small. The small difference can be attributed to the nature of the experiment variation. 5.4.3.2 Thinner CdS Figure 40 and figure 41 show the dark J-V and SR characteristics for the devices with thinner CdS. All other processing parame ters are all same as the devices discussed above. The dark J-V curve shows the same tr end as the device with thicker CdS: the reverse saturation current current decrease with increasing O2 partial pressure indicating better junction quality. For spect ral response, it clearly shows that the QE increases with decreasing O2 partial pressure in the range of 500600nm, this indicates that the extend of interdiffusion increases with increasing O2 partial pressure. Out of this range, their QEs 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 0.000.250.500.751.001.251.50 Voltage [V]Current Density [A/cm2] N2/O2=9/1 N2/O2=7/3 N2/O2=5/5 N2/O2=1/9 Figure 40 Dark I-V Characteristics fo r Device Fabricated at Various O2 Concentrations (Thinner CdS)

PAGE 90

76 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. N2/O2=9/1 N2/O2=7/3 N2/O2=5/5 N2/O2=1/9 Figure 41 Spectral Response for Device Fabricated at Various O2 Concentrations (Thinner CdS) 720 740 760 780 800 820 840 9/17/35/51/9 Ratio of N2/O2Voltage [mV]0 10 20 30 40 50 60 70 80FF [%] VOC FF Figure 42 Summary Results for De vice Fabricated At Various O2 Concentrations (Thinner CdS) are fairly same. Figure 42 shows the Voc and FF with different O2 partial pressure. The performance improves with increasing O2 partial pressure, but at high O2 partial pressure, no further improvement was observed.

PAGE 91

775.4.4 Junction Properties with Different O2 Partial Pressure The diode factor A is an indicator of th e quality of the main junction. Figures 43 and 44 show the parameters of the main junction versus the O2 partial pressure. The diode factor A first decreases with increasing O2 partial pressure and th en increases at higher 1.2 1.3 1.4 1.5 1.6 1.7 1.8 9/17/35/51/9 Ratio of N2/O2A value Figure 43 Diode Factor versus O2 Partial Pressure 1.0E-12 1.0E-11 1.0E-10 1.0E-09 1.0E-08 1.0E-07 9/17/35/51/9 Ratio of N2/O2Saturation current J0 [A] Figure 44: Saturation Current versus O2 Partial Pressure O2 partial pressure, the same trend of juncti on saturation current is observed. Based on these two junction parameters, the quality of junction improves with increasing O2 partial

PAGE 92

78 pressure and reach highest quality at some point then degrades. The higher quality of the main junction is attributed to the condensed CdTe film, in which the shunting path was suppressed. 5.4.5 C-V Measurement Figure 45 shows the C-V curves of the devices with different O2 partial pressure. Table 7 shows the calculated depletion widt h and doping concentration based on these CV measurements. In literature, O2 can increase the hole concentration in CdTe. We did observe a small increase on hole concentratio n The doping concentration increases from 1.98x1014 to 3.46x1014 cm-3 with increasing O2 partial pressure from N2/O2=9/1 to N2/O2=5/5 and then become saturated. 0.00E+00 5.00E+16 1.00E+17 1.50E+17 2.00E+17 2.50E+17 3.00E+17 3.50E+17 -2.00-1.50-1.00-0.500.000.50 Voltage [V]Capacitance [cm2/F2] 9/1 7/3 5/5 1/9 Figure 45: C-V Characterist ics of Devices Fabricated with Different Partial Pressures Table 7 Summary of Doping Concen trations for Varied Ratio of N2/O2 N2/O2=9/1 N2/O2=3/7 N2/O2=5/5 N2/O2=1/9 NA 1.98E14 2.77E14 3.46E14 3.4E14 W(m) 3.03 1.82 1.98 2.03

PAGE 93

795.5 Phosphorus Doped CdTe Devices The CdTe cells used for this study were fabricated on 7059 boro-silicate glass substrates. The front contact co nsisted of MOCVD-deposited SnO2: F and undoped SnO2 layer. Cadmium sulfide was deposited using chemical-bath deposition (CBD), followed by the deposition of CdTe by close-spaced subl imation (CSS) in an inert ambient. We tried to incorporate P during CSS CdTe deposition by the mixture of CdTe and P powders. In order to achieve evenly mixed CdTe and Cd2P3 powder, the Cd2P3 powder and CdTe powder (approximately 5% at. of P) were mixed and sealed in an evacuated quartz ampoule. The quartz ampoule was gr adually heated up to 700C for about 48 hours. Then the mixture was removed, repulverized and remixed. The mixed powder was referred to as CdTe: P and was used to deposit CdTe: P film by CSS in He ambient. It has to be noted that the high efficiency solar cells in our lab were fabricated in O2 ambient. Some CdTe films were subjected to CdCl2 treatment and some were not. Two different CdCl2 annealing temperatures were used. Subsequently the CdTe surface was etched using a Br2/Methanol solution. The moly back contact is sputtered at room temperature in order to dec ouple the effects of the impuri ties from the graphite back contact. Then the cells were subjected to a post deposition annealing in inert ambient. First we tried to deposit CdTe film direc tly from the powder mixture, but the film is unusually non-uniform; the reason for th e observed non-uniformity is not clear, probably due to the non-evenly mixed powder. In order to get uniform film we made a very thick CdTe film from the powder mixtur e as the CdTe source for the deposition of CdTe thin film. The thick CdTe film was analyzed using EDS. EDS detects 3% at. of P

PAGE 94

80 in the thick CdTe film, which is in relative ly good agreement with the initial amount of P added in CdTe powder (5%at.). 200 300 400 500 600 700 No CdCl2CdCl2,390C CdCl2,410C Voc [mV] Figure 46 Voc for Cells with Phosphorus Figure 46 and 47 shows the cells performance with different CdCl2 annealing conditions. All cells’ performance is poor compar ed with our baseline cells. The Voc is in the range of 300 to 620 mV; FF is in the ra nge of 30 to 45%. Th e best cells form this group of cells were those CdCl2 treated at 390C. Figure 48 shows the I-V curve. These I-V curves indicate shunting is a main problem for th e poor performance. The poor performance indicates the quality of the CdTe:P film is poor. One of these groups of CdTe:P films and again the thick source CdTe:P was analyzed using EDS. No P was detected; this doesn’t mean there are no any P atoms in CdTe film and probably due to the limitation of EDS. These EDS results showed that P and CdTe ha ve different transpor t rate during the CSS deposition process and the CdTe: P source may be depleted of P after several depositions. All these results showed the questionable quality of CdTe:P films indicating that CSS deposition technique is not suitable for the incorporation of P for doping purpose.

PAGE 95

81 20 25 30 35 40 45 50 No CdCl2CdCl2,390C CdCl2,410C FF [%] Figure 47 FF for Cells with Phosphorus -0.025 -0.015 -0.005 0.005 0.015 0.025 -1.00-0.75-0.50-0.250.000.250.500.751.001.251.50Voltage [V]Current Density [A/cm2] No CdCl2 CdCl2,390C CdCl2,410C Figure 48 I-V Characteristics for Cells with Phosphorus 5.6 Antimony Doped CdTe Devices Antimony (Sb) is another candidate of dopa nts for CdTe due to its low ionization energy, low diffusion coefficient, and easy of incorporation in to CdTe. It has been used

PAGE 96

82 as a back contact material to achieve high pe rformance of CdTe sola r cells [87]. Based on the experiments for P doped CdTe, we decide d to incorporate Sb into CdTe through a post diffusion approach. Following the depositio n of CdTe, a thin layer of Sb (20nm) was deposited onto CdTe by RF sputtering. Th e whole structure was subsequently heattreated in an inert ambient at the near atmosphere pressure to suppress the evaporation. This heat treatment drives Sb to diffuse into CdTe. The annealing temperatures were varied from 300 to 525C, and the dura tions from 20 to 160 minutes. The early experiments suggested that Sb could evaporat e during the heat trea tment; so another Sb coated glass (200nm) was used to cover the CdTe to act as an additiona l Sb source during the annealing. In order to inve stigate the role of Sb, the back contact was deposited by RF sputtered molybdenum at room temperature. 5.6.1 Effect of Sb In order to investigat e the Sb effect, two sets of cells were fabricated on the same substrate to eliminate experimental variati ons. The only difference between the two sets is that one set was subjected to the Sb-di ffusion process and one se t not. Figure 49 shows light J-V characteristics for cells with and without Sb. The Jsc was similar, but the Voc for the set with Sb is 50 mV higher. The current at forward bias beyond Voc was lower for the cells with Sb (Higher back barrier or roll-over). It is not clear at this moment what mechanism causes this behavior. One possibl e explanation is that the increased doping concentration near the back contact region resu lts in larger band bending at the interface. The hole current was limited by this larger band bending. The doping profile shown in figure 50 partially supports this explanation. It must be mentioned that there is no

PAGE 97

83 intentional copper in these devices. These re sults suggest that Sb could be used to increase the doping level in CdTe and avoi d using copper to im prove the long-term stability. -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 -0.50-0.250.000.250.500.751.00 Voltage [V]Current Density [A/cm2] Sb No Sb Figure 49 Light J-V Characteristics of CdTe Cells with and without Sb 1.0E+11 1.0E+12 1.0E+13 1.0E+14 1.0E+15 1.0E+16 12345 Depletion Width [m]NA [cm-3] Sb No Sb Figure 50 Doping Profile of CdTe Cells with and without Sb

PAGE 98

84 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. Sb No Sb Figure 51 Spectral Response of CdTe Cells with and without Sb The spectral response (figure 51) for the de vice without Sb is slightly higher than that of the device with Sb between 600 nm to approximately 825 nm. The difference becomes larger at longer wavele ngth although the absolute difference va lue is small. This is possibly because the higher carrier concentr ation in CdTe. Usually with the increasing of carrier concentration the collection effi ciency decreases due to the shrinking of depletion with. From 400 nm to 500 nm, the QE is higher for the device with Sb. This is most probably due to the th inning of CdS during the Sb diffusion process. Table 8 summarizes these results. Table 8 Summary of Results for the Devices with and without Sb Jsc[mA] Voc [mV] FF[%] Efficiency[%] With Sb 21.99 800 62 11 Without Sb 22.14 750 68 11.3 5.6.2 Effect of the Sb Annealing Temperature The diffusion depth and the amount of Sb in CdTe film depend on the annealing temperature and the duration. The annealing temperature also affects the properties of

PAGE 99

85 CdS, CdTe, and the interface between CdS a nd CdTe. The annealing time was fixed for 25 minutes and the temperature va ried from 400 to 525C. NO CdCl2 treatments were performed in this case. Table 9 summari zed the process condi tions and the cells’ performance. Table 9 The Process Conditions and The Cells’ Performance with Different Annealing Temperature 400C 450C 500C 525C CdTe O2 O2 O2 O2 Sb 200 200 200 200 HT 400C, 25min with 2K coated glass 450C, 25min with 2K coated glass 500C, 25min with 2K coated glass 525C, 25min with 2K coated glass CdCl2 N N N N Contact Molybdenum Molybde num Molybdenum Molybdenum Voc 710-730 740-750 590 470 FF 37-41 41-44 29 38 The cell performance first increases when the temperature increased from 400 C to 450 C and then drastically dropped with the fu rther increasing temperature. Figure 52 shows the spectral response with differe nt annealing temperatures. At 400 C and 450 C, the QEs are fairly normal, but the QEs dropped drastically when the temperature is above 450 C, further more, the QEs are constant with the wavelength up to 775nm. It looks like the cell’s structure is CdS (n type)-CdTe (n type)-CdTe (p type ), not the typical np junction structure. Figure 53 show s the QE simulation results of the nnp structure. In this simulation, the concentration of the n layer CdTe is fixed at 1x1016 cm-3. The thickness of this n layer CdTe is varied from 0.02 to 1 m. As can be seen, the QE drops with increasing thickness of n layer of CdTe. The difference between the simulated and the real cell’s is that the QE of the simula ted one is zero at short wavelengths and is

PAGE 100

86 asymmetric. Due to this difference, the real ca se is not just the assumed nnp structure. A modified simulation was done. In the modified simulation, the structure is np, but donor defects are introduced. The acceptor concentration in CdTe is fixed at 1x1014cm-3, donor 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. 400 C 450 C 500 C 525 C Figure 52 Spectral Responses for Device wi th Different Annealing Temperature 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. 0.02m 0.05m 0.5m 1m Figure 53 The Effect of The Thickness of N Layer CdTe on QE

PAGE 101

87 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. 1.00E+14 1.00E+15 1.00E+16 Figure 54 The Effect of Donor Defects in CdTe Bulk on QE defects were varied from 1x1014 to 1x1016 cm-3. Figure 54 shows the simulated QE curves. When the donor defects are higher than the acceptors, the QE shape is most likely close to the real case: constant below a specific wavelength depending on the level of donor defects. Based on this simulation, one tentative conclusion can be drawn: High temperature annealing results in high leve l donor defects in CdTe. The donor defects could be due to the diffusion of S from CdS at high temperature or due to the diffusion of Sb from the CdTe surface. Consideri ng the high deposition temperature (>560 C) of CdTe, the later one is closer to the truth: the high level donor defects are due to too much Sb diffused into CdTe and Sb atoms substitu te the Cd atoms. It is already known that CdTe becomes semi-insulating material at ve ry high concentration of Sb in CdTe and more Sb will occupy the Cd sites to compensate SbTe and VCd [88]. To verify this assumption, SIMS measurement has to be performed.

PAGE 102

88 0.E+00 1.E+17 2.E+17 3.E+17 4.E+17 5.E+17 6.E+17 7.E+17 8.E+17 9.E+17 1.E+18 -2.0-1.5-1.0-0.50.00.5 Voltage [V]Capacitance [cm2/F2] 400C 450C 500C 525C 1.E+11 1.E+12 1.E+13 1.E+14 1.E+15 1.E+16 6.07.08.09.0 Depletion Width [m]NA [cm-3] 400C 450C 500C 525C Figure 55: Doping Concentration Vs Depletion Width (left) and C-V Characteristics (right) Figure 55 shows the doping concentra tion and C-V characte ristics. The doping concentration is deduced from C-V measur ement. The higher annealing temperature leads to higher doping concentration. Tabl e 10 shows the doping c oncentration at zero bias. The doping concentrations for 500 and 525 C are up to 4x1014cm-3. Considering the high defect density as we said before ba sed on the QE, these high doping concentration may be over estimated, the real doping concentration should be lower or much lower than the deduced value from C-V measurem ent depending on the defect density. Table 10 Doping Concentration and Depl etion Width for the Cells Shown in Figure52 400 C 450 C 500 C 525 C WD 7.55 m 6.9m 8.14 m 7.83m NA-ND 4.1E13 1.2E14 3.9E14 3.8E14

PAGE 103

895.6.3 Effect of Annealing Time Due to the fact that high temperature anne aling leads to high de fect concentration, the Sb annealing temperature was fixed at 400 C, while the annealing time was varied from 40 to 160 minutes. Table 11 shows th e processing parameters and devices’ performance. The cells treated for 40 minutes have the lowest performance, while the cells treated for 80-160minutes show basically same performance. The light J-Vs for representative cells are shown in figure 56. The cell with the lowest performance treated for 40 minutes also shows lowest back barrier (least roll-over). Again, it also shows lowest net carrier concentration (the same as we observed before: the cell with higher back barrier also shows higher net carrier concentration). To fully understand this phenomenon will require numerical modeling that can find a set of parameters to fit the experimental results such as light and dark J-V and SR. The essentially same performance for 80-160 minutes annealing may be due to the low diffusion coefficient. Table 11 The Effect of Annealing Time 40min 80min 120min 160min CdTe O2 O2 O2 O2 Sb 300 300 300 300 HT 400C,40min with 3K coated glass 400C,80min with 3K coated glass 400C,120min with 3K coated glass 400C,160min with 3K coated glass CdCl2 Y Y Y Y Contact Molybdenum Molybde num Molybdenum Molybdenum Voc (mV) 700~730 750~770 730~770 740-770 FF(%) 58-63 60-62 61-63 61-64

PAGE 104

90 -0.05 0.05 0.15 0.25 0.35 0.45 0.55 0.65 -0.50.00.51.01.5 Voltage [V]Current Density [mA/cm2] 40min 80min 120min 160min -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 -0.50.00.51.01.5 Voltage [V]Current Density [A/cm2] 40min 80min 120min 160min Figure 56 Dark (Left) and Light (Right) J-V Characteristics with Different Annealing Time Figure 57 shows the spectral response of the devices shown in figure 56. The QEs’ shape are all similar, 160 minutes anneal ing gives the best QE. The lowest QE is from the device annealed for 120 minutes. It is almost parallel to the highest one from 400nm to 900 nm, so this lowest QE is because of the thicker CdS, not caused by the Sb annealing. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. 40min 80min 120min 160min Figure 57 The QE of the Devices with Different Sb Annealing Time

PAGE 105

91 0.E+00 2.E+16 4.E+16 6.E+16 8.E+16 1.E+17 1.E+17 1.E+17 2.E+17 -2.0-1.5-1.0-0.50.00.5 Voltage [V]Capacitance [cm2/F2] 40min 80min 120min 160min 1.0E+11 1.0E+12 1.0E+13 1.0E+14 1.0E+15 1.0E+161.02.03.04.0Depletion Width [m]NA [cm-3] 40min 80min 120min 160min Figure 58 Doping Concentration Vs Depletion Width (Left) and C-V Characteristics (Right) Figure 58 shows the doping c oncentration and C-V charac teristics with different annealing times. Although there is no single tr end, but the highest dop ing concentration is from the longest annealing time, the shortest annealing times results in the lowest doping concentration. The partial conclusion we can draw from these experi ments is that longer Sb annealing leads to higher dopi ng concentration due to the diffusion of Sb into CdTe absorb layer. 5.6.4 The Effect of HCl Etching After the Sb diffusion process, there is excess Sb (or Sb compound) left on the CdTe surface. It has been reporte d that Sb-Te can transform to Sb2Te3 under heat treatment [89]. In order to investigate its eff ect, the substrate was cut into two halves after Sb annealing; one half was s ubjected to HCl etching before the back contact application.

PAGE 106

92 No CdCl2 treatments were performed in this case. Table 12 summrizes the detail process conditions and devices’ performance. Table 12 The Effect of HCl Etching 400C, HCl Etch 400C, No HCl Etch 450C, HCl Etch 450C, No HCl Etch CdTe O2 O2 O2 O2 Sb 200 200 200 200 HT 400C,25min with 2K coated glass 400C,25min with 2K coated glass 450C,25min with 2K coated glass 450C,25min with 2K coated glass HCl Y N Y N CdCl2 N N N N Contact Moly Moly Moly Moly Voc,FF 680-690,37-38 710-730,37-41 630-680,32-37 740-750,41-44 Figures 59 and 60 show the spectral respon se and the dark and light J-V plots. Although the performance is poor fo r both sets due to the no CdCl2 treatment, the cells without HCl etching show higher Voc and Jsc. The dark and light JV show larger back barrier for the cells with HCl etching. For the HCl etched sample s, the higher barrier could be due to the lower work function of molybdenum. For unetched samples, because Sb2Te3 has higher work function (4 .93eV), it acts as a buffe r layer between CdTe (5.1 eV) and molybdenum (4.49 eV) metal layer, so the barrier is redu ced. The higher barrier for the HCl etched samples will limit the curre nt flow, resulting in lower collection. The QE shown in figure 59 is cons istent with this higher back barrier. It suggested that the excess “Sb” or Sb2Te3 compound at the surface of CdTe l eads to a reduction in the back contact barrier. Antimony telluride has been used as an effective back contact to achieve high performance CdTe solar cells [30].

PAGE 107

93 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. 400C,No HCl etched 450C,No HCl etched 400C,HCl etched 450C,HCl etched Figure 59: Spectral Response for HCl and No HCl Etched Devices -0.05 0.05 0.15 0.25 0.35 0.45 -0.50.00.51.01.5 Voltage [V]Current Density [mA/cm2] 400C,No HCl etched 450C,No HCl etched 400C,HCl etched 450C,HCl etched -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 -0.50.00.51.01.5 Voltage [V]Current Density [A/cm2] 400C,No HCl etched 450C,No HCl etched 400C,HCl etched 450C,HCl etched Figure 60 Light (Left) and Dark (Right) JV for HCl and No HCl Etched Devices 5.6.5 CdCl2 Effect on CdTe:Sb Devices The cadmium chloride (CdCl2) treatment is crucial in the device fabrication of polycrystalline CdTe/CdS solar cells. It can increase the grai n size and passivate the grain boundaries (GBs) in the CdTe active layer. It also redu ces recombination in these

PAGE 108

94 devices. Recent investigations showed that CdCl2 treatment affects the interior of CdTe grains by the potential incorporation of electri cally active impurities. Here we studied the effect of CdCl2 treatment on the Sb doped CdTe devi ces. Table 13 summaries the process and the comparative results. The devices with CdCl2 treatment have much better Voc and FF. The Voc is increased from mid 700 mV to 830mV, FF from 40 to 62%. Table 13 CdCl2 Treatment on Sb Doped Devices No CdCl2 No CdCl2 CdCl2 Treated CdCl2 Treated CdTe O2 O2 O2 O2 Sb 200 200 200 200 HT 400C,25min with 2K coated glass 450C,25min with 2K coated glass 430C,25min with 2K coated glass 430C,25min with 2K coated glass CdCl2 N N Y Y Contact Molybdenum Molybde num Molybdenum Molybdenum Voc, FF 710-730,37-41 740-750,41-44 800-810,61-61 810-830,61-62 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. No CdCl2 No CdCl2 CdCl2 CdCl2 Figure 61 Spectral Response for The Devices with and without CdCl2 Treatment Figure 61 shows the spectral response fo r the devices with and without CdCl2 treatment. The devices without CdCl2 treatment have lower QE for the whole range from 400nm to 800 nm, especially in 400nm to 550nm. We assume the beginning CdS

PAGE 109

95 thickness is same because of the same depositio n parameters of CdS. It clearly shows the final thickness of CdS is thinner for CdCl2 treated devices than those untreated (Based on the QE at 450nm). In the range of 500 to 600nm, the QEs for CdCl2 treated devices are higher than untreated devices, indicated that CdCl2 treatment promotes the formation of CdTexS1-x at the CdTe/CdS interface. This is c onsistent with the previous study from different groups. -0.05 0.05 0.15 0.25 0.35 0.45 -0.50.00.51.01.5 Voltage [V]Current Density [mA/cm2] No CdCl2 No CdCl2 CdCl2 CdCl2 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 -0.50.00.51.01.5 Voltage [V]Current Density [A/cm2] No CdCl2 No CdCl2 CdCl2 CdCl2 Figure 62 Light (Left) and Dark (Right) JV for the Devices with and without CdCl2 Treatment J-V characteristics (Figure 62) show that the back barrier for the CdCl2 treated devices is much lower than those untreated devices. This is primarily due to the passivation of the surface state. The untreated cells also show ed lower short circuit current density Jsc and FF. This is mainly due to the large back barrier which limits the hole current and the collection efficiency.

PAGE 110

96 We also compared the effect of CdCl2 treatment before Sb deposition and after the Sb annealing. When the CdCl2 treatment is done before Sb annealing, the performance is only a little bit better than untreated devices. This could be because Sb annealing leaves Sb related defects th at can’t not be passivated without CdCl2 treatment. 5.6.6 Two-layer CdTe Devices Based on the results for the different Sb annealing times, we suspected that Sb may not diffuse into the deep bulk of CdTe due to its low diffusion coefficient. A series of experiments have been done to investigat e the effect of the “Sb depth” in CdTe. A layer of CdTe with different th ickness was first deposited in O2 ambient (Labeled with deposition time), and then the Sb was spu ttered and diffused into CdTe. After Sb diffusion process, another layer of CdTe was deposited onto the first layer of CdTe. The thickness of the first layer (t he antimony depth) and the CSS ambient of second layer CdTe were varied. 5.6.6.1 The Effect of the CSS Ambient of Second Layer CdTe The CdTe of our baseline devices is deposited in O2 ambient. As we already know that O2 ambient can increase the doping concentration, reduce the grain size and improve the main junction properties, so the first layer of CdTe was deposited for two minutes in O2 ambient in order to have a good main junction. In this study, the deposition ambient of of second layer CdTe was varied (Table 14). The label “O2-O2” means the both CdTe layers are deposited in O2 ambient, O2-He means the second layer CdTe was deposited in He ambient. Figure 63 and Figure 64 show the light and dark J-V characteristics. The O2-

PAGE 111

97 He cells have higher Jsc, FF, and lower Rs than O2-O2 cells. The cells for both cases have essentially the same Voc. The roll-over observed in both li ght and dark curves at higher Table 14 Summary of the Devices with Different Ambient for 2nd layer CdTe 1st CdTe O2 O2 Sb 200 200 HT 430C,60min with 2K coated glass 4 30C,60min with 2K coated glass HCl etch N N 2nd CdTe He O2 CdCl2 Y Y Contact Moly Moly Voc,FF 820, 62 780~810,32-38 -0.025 -0.015 -0.005 0.005 0.015 -0.50-0.250.000.250.500.751.001.251.50 Voltage [V]Current Density [A/cm2] O2-O2 O2-He Figure 63 Light J-V of the Devices Fa bricated in Different Ambients (2nd Layer) forward bias, suggests that the contacts are rectifying. Howeve r at lower bias (close to Voc) for O2-He samples, the presence of a smaller barrier makes carrie r transport easier thereby causing no adverse effects on the fill factor of the device. The reverse saturation

PAGE 112

98 currents are similar. From the spectra l response (Figure 65), we can see the recombination rate for O2-O2 cell increases towards main junction, while O2-He cell doesn’t. This may be due to the formation of so me Sb-O related defects or a thin layer of Sb2O3. -0.05 0.05 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 -0.50-0.250.000.250.500.751.001.251.50 Voltage [V]Current Density [mA/cm2] O2-O2 O2-He Figure 64 Dark J-V of th e Devices Fabricated in Different Ambients (2nd Layer) 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 400500600700800900 Wavelength [nm]Q.E. O2-O2 O2-He Figure 65 Spectral Response of the Devices Fabricated in Different Ambient (2nd Layer)

PAGE 113

995.6.6.2 The Depth Effect of Sb in CdTe By varying the thickness of the first laye r of CdTe, the depth of Sb was varied. Figure 66 shows the dark J-V char acteristics for the representative cells. As can be seen, the turn-on voltages are almost identical, but the current was suppressed more for thinner first CdTe layer beyond 0.8 volt. From the light J-V shown in figure 67, the back barriers are almost same. It seems the Sb diffused in to the interface region and the properties of CdS was affected. The collection was increas ed at around the maximum power point for a thinner first CdTe layer. We can tentativ ely conclude that the increased doping level results in higher built-in pot ential that improved the coll ection. The calculated hole concentration from C-V measurement is 4X1016 cm-3. -0.05 0.05 0.15 0.25 0.35 0.45 0.55 0.65 0.75 -0.50-0.250.000.250.500.751.001.251.50 Voltage [V]Current Density [mA/cm2] 30S 45S 2min Figure 66 Dark J-V Characteristics of CdTe Cells with Different Thickness of First Layer

PAGE 114

100 -0.025 -0.015 -0.005 0.005 0.015 -0.50-0.250.000.250.500.751.001.251.50 Voltage [V]Current Density [A/cm2] 30S 45S 2min Figure 67 Light J-V Characteristics of CdTe Cells with Different Thickness of First Layer 5.6.7 The Effect of Back Contact The effect of back contact deposition conditions is also studied. In this study, the sputtering pressure and sputte ring rate are varied. Four co mbinations were used: high pressurehigh sputtering power (L-H), low pressure-low sputtering power (L-L), high pressure-low sputtering power (H-L), and high pressure-high s puttering power (H-H). Table 14 shows the comparative device data fo r different back contact conditions. Figure 68 and 69 show the J-V characteristics. Table 15 Comparative Device Data for Different Back Contact Conditions H-L L-L L-H H-H CdTe O2 O2 O2 O2 Sb 300 300 300 300 HT 400C,80min with 3K coated glass 400C, 80min with 3K coated glass 400C,80min with 3K coated glass 400C,80min with 3K coated glass CdCl2 Y Y Y Y Contact Molybdenum (HL) Molybdenum (LL) Molybdenum (LH) Molybdenum (HH) Voc (mV) 740 790 780 710 FF (%) 53 50 57 51

PAGE 115

101 It was observed that the best cell performance is from low pressure – high sputtering power condition and then the low pr essure-low sputtering power. Both devices fabricated at high-pressure s how poor performance and severe roll-over. It is clear that the low pressure is a key processing para meter to achieve a better back contact. G.Gordillollo’s study shows that the resistivity is decreased by decreasing the sputtering pressure and increasing the glow discharge power. The reduction of the resistivity is attributed to the increasing carrier mobility due to an increment of th e grain size. [90] Our studies also show the same trend. Although the decrease on resistivity is not obviously seen in the J-V curves due to the severe back barrier effect, it still could be an factor for the improvement of cell performance. A nother reason for the improvement of cell performance is that at low pressure and hi gh sputtering power, th e sputtered particles have higher energy. These high energy particle s can affect the surface state in a way which can lower the back barr ier (see figures 68 and 69). The adhesion, though not yet confirmed, is suspected to be improve d at low pressure-high power condition. -0.03 -0.02 -0.01 0.00 0.01 -0.50-0.250.000.250.500.751.001.251.50 Voltage [V]Current Density [A/cm2] 1-3B-4-12(H-L) 1-3B-4-21(L-L) 1-3B-4-31(L-H) 1-3B-4-42(H-H) Figure 68 Light J-V for The Devices Wi th Different Back Contact Conditions

PAGE 116

102 -0.05 0.05 0.15 0.25 0.35 0.45 -0.50-0.250.000.250.500.751.001.251.50 Voltage [V]Current Density [mA/cm2] 1-3B-4-12(H-L) 1-3B-4-21(L-L) 1-3B-4-31(L-H) 1-3B-4-42(H-H) Figure 69 Dark J-V for the Devices with Different Back Contact Conditions 5.7 Voc Versus Doping Concentration So far we studied the CSS CdTe am bient effect and Sb effect on doping concentration. In the ambient effect st udy, a weak trend was observed: the doping concentration and Voc increased with the O2 partial pressure. In the Sb effect study, the effect of processing parameters on the carri er concentration was studied by varying the annealing temperature and annealing tim e. Carrier concen trations up to 1016 cm-3 were observed. Figure 70 shows the summary of Voc versus doping concentration. Clearly, this graph can be partitioned into four regions (showed in figure 70). Regions 1 and 2 with empty legends show Voc versus doping concentration for the CdCl2 treated devices. Regions 3 and 4 with solid square legend is for the not CdCl2 treated devices. More than 100 mV Voc difference exists between the CdCl2 treated and not CdCl2 treated devices at similar doping concentrations.

PAGE 117

103 In regions 1 and 3, although there is fluctu ation, the general trend is up. While in regions 2 and 4, the trend reversed. It is also observed that depending on the process conditions, high Voc (830mv) can be achieved at both low (1014) and high concentrations (1016). In order to understand what factors are mainly responsible for the reduction of Voc at different doping concentrations, simula tion has been done to fit the curvature showed in figure 70 by varying back barriers, defect density, and carrier life time. 500 550 600 650 700 750 800 850 900 1.E+131.E+141.E+151.E+161.E+17 NA(cm-3)Voc(mV) Figure 70 Voc Versus Doping Concentrat ion at Different Process Conditions It was founded that back barrier played an important role at regions 1 and 3. When NA is 2.8 1014 cm-3 with perfect ohmic contact, the Voc is 855 mV. In order to achieve 830 mV Voc, the back contact barrier must be less than 0.13 eV. Above this value, even the CdTe is free of defects, the Voc is still below 830 mV. Figure 71 shows the simulation result in which the parameters are: NA: 2.8 1014 cm-3, no any defects, and back barrier: 0 eV, 0.13 and 0.2 eV. We can see that the Voc dropped to 780 mV when back barrier is 0.2 eV. This is the ideal case. Normally in CdTe solar cells, NA is around 21014 cm-3, and there always exist defects more or less depending on process 1 2 3 4

PAGE 118

104 parameters. This indicates that the copper doped graphite contact is an near ohmic contact and the defect density is very low and has minor effect on Voc. Due to the fact of the low work function of molybdenum (4.6eV), the Voc is difficult to be above 830 mV at low doping concentration for the cells with Moly b ack contact unless ther e is high density and shallow acceptor-like defects. For example, if there is 0.3eV back barrier, the following parameters have to be used to achieve 830 mV Voc with 2 1014 cm-3 doping concentration: NT=2 1016 cm-3, ET=Ev + 0.1 eV, n= p=10-18. -0.03 -0.02 -0.01 0.00 0.01 0.000.250.500.751.00 Voltage [V]Current Density [A/cm2] 0 V 0.1 eV 0.2 eV Figure 71 I-V Simulations with Different Back Barriers In order to increase doping concentration, external impurity has to be introduced into CdTe by any means. On the other hand, the introduction of external dopant atoms, it is unavoidable to introduce some defects to compensate the acceptors and the defects density depends on the amount of external dopant atoms and process parameters. In figure 70, it looks like Voc reaches its maximum value at the doping concentration of around 1015 cm-3. Simulation shows that at such high doping concentration and 0 eV back barrier, without any defects, the Voc could be 905mV. The experimental Voc is 840 mV.

PAGE 119

105 In order to fit this value, donor defect density up to 5 1014 cm-3 electron capture cross section up to 10-12 and with 0.2eV back barrier have to be used. In region 2, when the doping concentration is 2 1016 cm-3, the Voc dropped to 820 mV. While the ideal Voc is 968 mV. To fit this experime ntal value, parameters such as 0.37eV back barrier, electron and hole capture cross section up 10-11 and donor defect density up to 4 1013 cm-3 have to be used. It has to be noted that the Voc is not sensitive to low back barrier. When the back ba rrier was reduced to 0.32 eV, the Voc only changed several mVs, but the forward current increased this kind of shape IV was just what we observed for some devices( Figure 72). -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.000.250.500.751.00 Voltage [V]Current Density [A/cm2] 0.32 eV 0.37 eV Figure 72 I-V Simulations with Different Back Barriers for 21016 cm-3 J-V curves show higher back barrier for devices with higher doping concentration. At higher doping concentrati on, the Fermi level will sh ift downward, so the -work function of CdTe increases, re sulting in an increase of the back barrier. The work

PAGE 120

106 function of CdTe with 2 1016 cm-3 is 0.1 eV higher than th at of CdTe with 2 1014 cm-3. So the back back barrier of CdTe with 2 1016 cm-3 is 0.1 eV higher than that of CdTe with 2 1014 cm-3. How ever, if the doping concentration is high enough, tunneling contact will form and the barrier will actually decrease. It is still questionable whether the contact is tunneling or not, how ever, our results show that 1016cm-3 is not enough to form a tunneling contact. QE data show that the recombina tion rate is highe r at higher doping concentrations, which indicate that more defects were induced by the introduction of higher Sb content. This is cons istent with the simulation resu lts. Grain boundary exists in CSS CdTe in nature and defects tend to segregate there. Alt hough the Sb diffusion process may not change the grain boundary de nsity, the defect dens ity at grain boundary would increase due to this process. In general, it is much more likely that GBs lead to a majority carrier depletion in p-type material, the Fermi-level is close to the valence band and deep states are most likely positively charged or neutral. What electrons (minority carriers) see is a potential well, so the recombination will increase. CdCl2 treatment can passivate some defects at some degree. It is obvious that the defect s concentration is over the capability of CdCl2 treatment in this case. If the defects can be removed or reduced to some acceptable level, we should see an increase of Voc.

PAGE 121

107 CHAPTER 6 CONCLUSION The primary objective of this project is to explore new materials and processes that will improve the performance of CdTe/C dS solar cells. The main efforts focused on improving the open circuit volta ge by the following areas: (1) Investigation the effect of th e ambient of CSS CdTe on the performance and net doping con centration in CdTe layer. (2) Investigation the effects of extrinsi c impurities in CdTe absorber layer. In this case, we studied phos phorus (P) and antimony (Sb). (3) The front contact with the case focused on Zn2SnO4 (ZTO) buffer layer. Zn2SnO4 as a high resistive buffer layer was also studied. The effect of the deposition and annealing temperature on materi al structure and device performance was studied. The film deposited at 400C yielded ve ry high blue spectral response and high Voc at the same time. A more thorough study of the buffer layer mechanism will further help optimize device performance. The effect of the ambient of CSS CdTe was investigated. The mixture of N2 and O2 was used in this study. The morphology of the CdTe film s is independent on the ratio of N2/O2 as long as there is O2. The performance of the cells increases with the partial pressure of O2. The net carrier concentration is in the normal range of 1014 cm-3, but we

PAGE 122

108 did observe a small increase w ith the concentration of O2 and getting saturated at high content of O2. The quality of the main junction is improved with the increasing partial pressure of O2. The effects of phosphorus and antimony in Cd Te absorber layer were studied. The initial effort focused on incorporating phosphor us into CdTe by mixing P with CSS CdTe source. The attempt is not successful due to the inhomogenieties in the P-doped CdTe powder and different transportation rate of P and CdTe. The Sb effect was studied by diffusion approach. The effect of antimony anneali ng temperature and time, CdCl2 treatment, the excess antimony on CdTe, a nd the depth of antimony in CdTe on cell performance was studied. The amount and the depth of Sb in CdTe were not known at this point due to the inability of the SIMS measurement in our laboratory. The junction of the devices annealed above 500C crashed. It is not clear whethe r it is due to the diffusion of antimony or not. SIMS result and defects identification will help understand the causes. The excess Sb/orSb2Te3 is found to be helpful to reduce the back contact barrier. A consistent increase in the hole c oncentration was also observed. Our work clearly shows that the both Voc and the doping levels in th e CdTe:Sb device are higher (the Voc by 50 mV), suggesting increasing the dopi ng in CdTe further could potentially yield even higher Voc as expected. Since the doping study by this diffusion approach is not very successful, the future work should focus on other doping techniques such as sputtering and coevaporation and also other p type dopants. Co-evaporation is recommended to be first investigated since it is fa vorable for manufacture. Introdu cing ionized N in CSS CdTe process is also worth to study.

PAGE 123

109 In addition, an attempt has been made to synthesize TiSe2, which has high work function and could be used to form ohmic b ack contact. Although it is failed to selenize the Ti film on CdTe due to the high temperat ure at this stage, it is worth to try other means like co-evaporation and sputtering to deposit TiSe2 on CdTe to test the possibility as a candidate of an ohmic back contact material.

PAGE 124

110 REFERENCES [1] Reynolds D C, Leies G, Antes L T and Marburger R E, 1954, Phys. Rev. 96, pp. 533 [2] D. A. Cusano, 1963, SolidState Electronics, vol. 6, pp. 217-232 [3] E. I. Andirovich, Y. M. Yua bov, and G. R. Yagudaev, 1969, Sov. Phys. Semicond., 3, pp. 61 [4] Xuanzhi Wu, 2004, Solar Energy, Volume 77, Issue 6, pp. 803-814 [5] Tracie J. Berniard, 2004, CdTe overview, www.nrel.gov [6] ESM. Caraman, P. Gasin and S. Vatavu, 2005, Thin Solid Films, Volumes 480481, pp. 254-258 [7] M. Emziane, K. Durose, N. Romeo, A. Bosio and D.P. Halliday, 2005, Thin Solid Films, Volumes 480-481, pp. 377-381 [8] J. Fritsche, T. Schulmeyer, A. Thi en, A. Klein and W. Jaegermann, 2003, Thin Solid Films, Volumes 431-432, pp. 267-271 [9] J. Hiie, 2003, Thin Solid Films, Volumes 431-432, pp. 90-93 [10] V. Komin, B. Tetali, V. Viswanatha n, S. Yu, D. L. Morel and C. S. Ferekides, 2003, Thin Solid Films, Volumes 431-432, 1 May 2003, pp. 143-147 [11] P. D. Paulson and V. Dutta, 2000, Thin Solid Films, Volume 370, Issues 1-2, 17 July 2000, pp. 299-306 [12] Xiangxin Liu, A.D. Compaan and Jeff Terry, 2005, Thin Solid Films, Volumes 480-481, pp. 95-98 [13] S. Erra, C. Shivakumar, H. Zhao, K. Barri, D.L. Morel and C.S. Ferekides, 2007, Thin Solid Films, Volume 515, Issue 15, pp. 5833-5836

PAGE 125

111 [14] Steven S. Hegedus and Brian E. McCandless, 2005, Sola r Energy Materials and Solar Cells, Volume 88, Issue 1, pp. 75-95 [15] T. D. Dzhafarov, S. S. Yesilkay a, N. Yilmaz Canli and M. Caliskan, 2005, Solar Energy Materials and Solar Cells, Volume 85, Issue 3, pp. 371-383 [16] Vasilis Fthenkis, KenZweibel, 2003, NCPV meeting [17] V. M. Fthenakis, M.Fuhrmann, J.Heiser, W.Wang, 2004, 19th EPSC [18] Anderson Richard L., Zeidenbergs Girtz Huang Thomas Boyce David Davis Mark, 1967, Heterojunction Devices [19] James Sites and Jun Pan, 2007, Thin Solid Films, Volume 515, Issue 15, pp. 6099-6102 [20] H.C. Chou, A. Rohatgi, J. Electtrochem. 1995, Soc, 142, pp. 254 [21] H.C. Chou, A. Rohatgi, N.M. Jokerst, E.W. Thomas, and S. Kamra, 1996, J. Elect. Mat., 25, pp.1093 [22] J. Tang, D. Mao, T.R. Ohno, V. Kaydanov, and J.U. Trefny, 1997, 26th IEEE PVSC, pp. 439 [23] T.A. Gessert, S. Asher, S. Johns ton, M. Young, P. Dippo and C. Corwine, 2007, Thin Solid Films, Volume 515, Issue 15, 31 May 2007, pp. 6103-6106 [24] T.L. Chu, S.S. Chu, 1995, Solid-State Elect., 38, pp.533 [25] Britt, J., Ferekides, C.S., 1993, Appl. Phy. Lett., 62, pp.2851 [26] Jae Ho Yun, Ki Hwan Kim, D oo Youl Lee and Byung Tae Ahn, 2003, Solar Energy Materials and Solar Cells, Volume 75, Issues 1-2, pp. 203-210 [27] Kevin D. Dobson, Iris Visoly-Fishe r, Gary Hodes and David Cahen, 2000, Solar Energy Materials and Solar Cells, Volume 62, Issue 3, pp. 295-325 [28] B.Ghosh, S. Purakayastha, P.K. Datta, R.W. Miles, M.J. Carter, R. Hill, Semicond. 1995, Sci. Tech., pp.71-76. [29] V. Viswanathan, B. Tetali, P. Selvar aj, S. Jagannathan, D.L. Morel, Sally Asher, and C.S. Ferekides, 2000, Pr oc. 28th IEEE PVSC, pp.587-590. [30] N. Romeo, A. Bosio, R. Tedesc hi, A. Romeo and V. Canevari, 1999, Solar Energy Materials and Solar Cells, Volume 58, Issue 2, pp. 209-218

PAGE 126

112 [31] N. Romeo, A. Bosio, R. Te deschi and V. Caneva ri, 2000, Thin Solid Films, Volumes 361-362, pp. 327-329 [32] W. Walukiewicz, 1988, Phys. Rev. B37, pp.4760 [33] W. Walukiewicz, 1988, J. Vac. Sci. Technol. B6, pp.1257 [34] W. Walukiewicz, 1989, Appl. Phys. Lett. 54, pp.2094 [35] W. Walukiewicz, 1994, Phys. Rev. B50, pp.5221 [36] W.Faschinger, 1996, J. Cryst. Growth 159, pp.221 [37] W.Faschinger, S. Ferreira, H. Sitter, 1998, Appl. Phys. Lett. 66, pp. 2516 [38] S. B. Zhang, S. H. We i, A. Zunger, Appl. Phys. 83, pp.3192 [39] Y. Marfaing, 1996, J. Cryst. Growth, 161, pp.205 [40] G. Mandel, F.F. Morehead and P.R. Wagner, 1964, Phys. Rev..]36A, pp. 826 [41] U. V. Desnica, 1998, Prog. Cryst. Growth and Charact., Vol. 36, No. 4, pp. 291357 [42] R. Magede, M. Deicher, U. Desnica, R. Keller, W. Pfeiffer, F. Pleiter, H Skudlik, and Th. Wichert, 1991, Appl. Surf. Sci. 50, pp. 169 [43] K. Ohkawa, T. Karasawa and T.J. Mitsuyo, 1991, /. Cryst. Growth 111, pp. 797 [44] S. Pykk, M.J. Puska and R.M. Nieminen, 1998, Phys. Rev. B 57, pp. 12164 [45] L. Shcherbak, P. Feichouk, P. Fochouk and O. Panchouk, 1996, Cryst. Growth 161, pp.219 [46] Su-Huai Wei and S. B. Zha ng, 2002, PHYSICAL REVIEW B 66, pp. 155211 [47] D.B. Laks, CG. Van de Walle, G.F. Neumark, PE. Bloech and ST. Pantelides, 1992, Phys. Rev. B 45, pp. 10965 [48] C.G. Van de Walle, DB. Laks, GF Neumark and S.T. Pantelides, 1993, Phys. Rev. B 47, pp. 9425 [49] M. Restle, K. Bharuth-Ram, H. Quinte l, C. Ronning, H. Hofsass, U. Wahl, S. G. Jahn, 1996, J. of Crystal Growth, 161, pp. 168-171

PAGE 127

113 [50] H. Pauli, K. Hingerl, E. Abramof, H. Sitter, H. Zajicek, K. Lischka, 1993, J. Appl. Phys. 73(8), pp. 4061-4063 [51] L. Svob and C. Grattepain, 1977, Journal of Solid State Chemistry, Volume 20, Issue 3, pp. 297-303 [52] L. Svob, Y. Marfaing, 1 982, Journal of Crystal Growth, Volume 59, Issues 1, pp. 276-280 [53] N. Abbas Shah, A. Ali, A. K. S. Aq ili, A. Maqsood, 2006, Cryst. Growth 290, pp. 452-458 [54] H. Wolf, T. Filz, J. Hamann, S. Lany, V. Ostheimer, Th. Wichert, 1999, Phys. B 273-275, pp. 843-847 [55] MR Lorenz, B. Segall, 1963, P hysics Letters, vol. 7, Issue 1, pp.18-20 [56] W. Akutagawa, D. Turnbull ,W. K. Chu and J. W. Mayer, 1975, J. Physics and Chemistry of Solids, Volume 36, Issue 6, pp. 521-528 [57] K. BArri, M. Jayabal, H. Zhao, D. L. Morel, S. Asher, J. W. Pankow, M. R. Yong, C. S. Ferekides, 2005, IEEE 29th PVSC [58] C. R. Corwine, A. O. Pudov, M. Glo eckler, S. H. Demtsu, J. R. Sites, 2004, Solar Energy Materials and Solar Cells, Volume 82, Issue 4, pp. 481-489 [59] T. D. Dzhafarov, S. S. Yesilkay a, N. Yilmaz Canli and M. Caliskan, 2005, Solar Energy Materials and Solar Cells, Volume 85, Issue 3, pp. 371-383 [60] B. Reinhold, M. Wienecke, 1999, Physica B 273-274, pp. 856-860 [61] S.B. Zhang, S.-H Wei, A. Zunger, Physica B 273-274, pp. 976-980 [62] N. V. Agrinskaya, 1985, Sov. Phys. Semicond. 19, pp. 199 [63] B. Yang, Y. Ishikawa T. Miki, Y. Doumae, M. Isshik i, 1997, J. of Cryst. Growth 179, pp. 410-414 64] S.E. Asher, F.S. Hasoon, T.A. Gesse rt, M.R. Young, P. Sheldon, J. HiltneP and J. Sites, 2000, IEEE Photovoltaic Sp ecialists Conference, pp. 479-482 [65] Su-Huai Wei, S. B. Zhang, 2002, Physics Review B66, pp. 155211 [66] A. Picos-Vcga, R. Ramrez-Bon, F.J. Espinoza-Beltrn, O. Zelaya-Angel, O. Alvarez-Fregoso, 1996, Thin Solid Films 290-291, pp. 395-400

PAGE 128

114 [67] R. N. Bicknell, N. C. Giles, J. F. Schetzina, 1986, Appl. Phys. Lett. 49(25), pp. 1735-1737 [68] J. D. Benson, D. Rajavel, B. K. Wa gner, R. Benz, C. J. Summers, 2002, J. Cryst. Growth, Vlume 95, Issues 1-4, pp. 543-546 [69] J. Santos-Cruz G. Torres-Delgado, R. Castanedo-Prez, S. Jimnez-Sandoval, J. Mrquez-Marn, O. Zelaya-Angel, 2006, Solar Energy 80, pp.142147 [70] Y. Hatanaka, M. Niraula, A. Na kamura, T. Aoki, 2001, Applied Surface Science 175-176, pp. 462-467 [71] Jaya P. Nair, R. Jayakrishnan, Nandu B Chaure, Arun Lobo, S. K. Kulkarni, R. K. Pandey, 1999, Thin Solid Films 347, pp. 39-45 [72] H. L. Hwang, Klaus Y. J. Hsu, H..Y. Ueng, 1996, J. of Cryst. Growth 161, pp. 7381 [73] Alvin D.Compan, Vi ctor Karpov, Randy G. Bohn, Dean Giolando, 2002, Annual Technical Report [74] T.L. Chu, S.S. Chu, C. Ferekides, J. britt, C.Q. Wu, G. Chen, N. Schultz, 1991, Solar Cells,30, pp. 123. [75] S. K. Ghandhi, N. R. Taskar, an d I. B. Bhat, 1987, Applied Physics Letters Volume 50, Issue 14, pp. 900-902 [76] J. M. Arias, S. H. Shin, D. E. Coope r, M. Zandian, J. G. Pasko, E. R. Gertner, R. E. DeWames, 1990, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films Volume 8, Issue 2, pp. 1025-1033 [77] Mitsuru Ekawa, Kazuhito Yasuda Masahisa Okada, Touati Ferid, Akikazu Tanaka and Manabu Saji, 1992, Journal of Crystal Growth, Volume 117, Issues 14, pp. 254-258 [78] L. Svob, Y. Marfaing, B. Clerjaud, D. Cte, A. Lebkiri and R. Druilhe, 1996, Journal of Crystal Growth, Volume 159, Issues 1-4, pp. 72-75 [79] T. Baron, K. Saminadayar an d S. Tatarenko, 1996, Journal of Crystal Growth, Volume 159, Issues 1-4, pp. 271-275 [80] M. Niraula, T. Aoki, Y. Nakanish i and Y. Hatanaka, 1999, Journal of Crystal Growth, Volume 200, Issues 1-2, pp. 90-95

PAGE 129

115 [81] N. Romeo, A. Bosio, V. Canevari, C. Spaggiari and L. Zini, 1989, Solar Cells, Volume 26, Issue 3, pp. 189-195 [82] K. A. Dhese, D. E. Ashenford, J. E. Nicholls, P. Devine, B. Lunn, C. G. Scott and J. Jaroszy ski, 1994, Journal of Crystal Growth, Volume 138, Issues 1-4, pp. 443447 [83] O. Vigil-Galn, J. Sastr-Hernndez, F. Cruz-Gandarilla, J. Aguilar-Hernndez, E.Marn, G. Contreras-Puente, E. Saucedo, C.M. Ruiz, V. Bermdez and M. Tufio-Velzquez, 2006, Solar Energy Mate rials and Solar Cells, Volume 90, Issue 15, pp. 2228-2234 [84] J. Herrero, M. T. Gutirrez, C. Gu illn, J. M. Doa, M. A. Martnez, A. M. Chaparro and R. Bayn, 2000, Thin Solid Films, Volumes 361-362, pp. 28-33 [85] S. Gayam, S. Bapanapalli, H. Zhao L. Nemani, D.L. Morel and C.S. Ferekides, 2007, Thin Solid Films, Volume 515, Issue 15, pp. 6060-6063 [86] Alessio Bosio, Nicola Romeo, Samantha Mazzamuto, Vittorio Canevari, 2006, Crystal Growth and Characterization of Materials 52 pp. 247-279 [87] N. Romeo, A. Bosio, R. Tedesc hi, A. Romeo and V. Canevari, 1999, Solar Energy Materials and Solar Cells, Volume 58, Issue 2, pp. 209-218 [88] E. S. Nikonyuk, Z. I. Zakharuk, V. L. Shlyakhovyi, P. M. Fochuk, A. I. Rarenko, 2001, Semiconductors, Vol. 35, No. 4, pp. 405-408 [89] Daniel Kraft, Bettina Spth, Andreas, Thissen, W. Jaegermann, 2003, 3rd World Conf. on Photovoltaic Energy Conversion, pp. 450-453 [90] G. Gordillo, F. Mesa C. Caldern, 2006, Brazilian Journal of Physics, vol. 36, no. 3B, pp. 982-985 [91] W.J. Danaher, L.E. Lyons, M. Marychurch, G. C. Morris, 1986, Appd. Surface Science, pp.338 [92] M. Ritala, J. Sarlund, M. Leskela, E. Siponmaa, R. Zilliacus, 1996, Solar Energy Materials & Solar Cells, 44, pp.177 [93] H.C. Chou, A. Rohatgi, 1995, J. Electtrochem. Soc, 142, pp. 254 [94] M. Ritala, J. Sarlund, M. Leskela, E. Siponmaa, R. Zilliacus, 1996, Solar Energy Materials & Solar Cells, 44, pp.177

PAGE 130

116 [95] R. Mamazza, U. Balasabramanian, S. Gayam, S. Bapanapa lli, L. Nemani, M. Jayabal, H. Zhao, D. L. Morel and C. S. Ferekides 2005, 29th IEEE PVSC, pp.283-286

PAGE 131

ABOUT THE AUTHOR Mr. Hehong Zhao grew up in China. He obtained his Bachelor’s in Automation Control from Yan-Shan University. After receiv ing his Bachelor degree, he worked as an electrical engineer in Beijing Materials Handlin g Research Institute for nine years. Later, he came to the United States, to attend the Univ ersity of South Florid a, for his graduate study in Electrical Engineering.