Growth kinetics and processings of copper indium diselenide-based thin films

MISSING IMAGE

Material Information

Title:
Growth kinetics and processings of copper indium diselenide-based thin films
Physical Description:
Book
Creator:
Kim, Suku
Publication Date:

Record Information

Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 30309979
oclc - 52731337
System ID:
AA00022321:00001

Table of Contents
    Title Page
        Page i
        Page ii
    Dedication
        Page iii
    Acknowledgement
        Page iv
        Page v
    Table of Contents
        Page vi
        Page vii
        Page viii
    List of Tables
        Page ix
    List of Figures
        Page x
        Page xi
        Page xii
        Page xiii
        Page xiv
    Abstract
        Page xv
        Page xvi
    Chapter 1. Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
    Chapter 2. Background and literature review of solar cell and CuInSe2-based materials
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
    Chapter 3. Reaction kinetics and pathways of CuInSe2 growth from bilayer precursor films: Time-resolved high temperature X-ray diffraction study
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
    Chapter 4. Secondary grain growth of CuInSe2 in the presence of CuxSe driven by interfacial forces
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
    Chapter 5. Epitaxial growth of CuInSe2 and CuGaSe2 films on single crystalline GaAs substrate using a migration enhanced epitaxy reactor
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
    Chapter 6. Thermal modeling of a rotating platen and substrate in a migration enhanced epitaxy reactor
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
        Page 113
        Page 114
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
        Page 120
        Page 121
        Page 122
        Page 123
    Chapter 7. Future work
        Page 124
        Page 125
        Page 126
        Page 127
        Page 128
        Page 129
        Page 130
        Page 131
        Page 132
        Page 133
        Page 134
        Page 135
        Page 136
        Page 137
        Page 138
        Page 139
    List of references
        Page 140
        Page 141
        Page 142
        Page 143
        Page 144
        Page 145
        Page 146
    Biographical sketch
        Page 147
        Page 148
        Page 149
        Page 150
Full Text










GROWTH KINETICS AND PROCESSING
OF COPPER INDIUM DISELENIDE-BASED THIN FILMS












By

SUKU KIM


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2003



























Copyright 2003

by

Suku Kim



























To my Lord, my wife, and my family
for their unconditional love.















ACKNOWLEDGMENTS

I am very grateful that I had such a precious time during my Ph.D. study here in

Gainesville. Along with the productive research experiences, I have been blessed with so

many things that God has planned for me. When I look back upon my last years, I realize

that everything eventually became better than the best that I could have wished for.

First and foremost, I would like to thank my family. My wife, Aggie, has always

shown the most sincere love and support. She has been the source of my strength

especially during my times of frustration. I am the luckiest man to have her next to me in

my life.

I thank my parents Tae Hwa Kim and Young Soon Lee, for their devotion and tender

love. They have always believed in me. I thank my grandma for her unconditional love

towards me. My sweet sister and brother, I thank both for their love and support.

I extend my most sincere thanks to my adviser, Dr. Anderson, for his guidance and

care. He always emphasized on being an independent researcher, and he leaded me in

that way. I would like to express my appreciation to every member in my supervisory

committee. They helped me in every possible way to find a right direction of my

research. I send my special thank to Dr. Crisalle for his time and sincere help. He and I

worked together on many projects, and he always tried to be there for me. Dr. Li and Dr.

Holloway, their knowledge and sincere help were often the inspirations for my research.

I had an opportunity to work for Dr. Ren, and it was one of the most productive and

pleasant experiences.









I must acknowledge my colleagues for their kind help and valuable contributions.

Seokhyun Yoon, Woo Kyoung Kim, Ryan Kaczynski, Ryan Acher, Lei Li, and Jiyon

Song, they are wonderful people to work with. Especially, I thank Seokhyun and Woo

Kyoung for their contribution to many of my works. I also thank former students, Dr.

Serkan Kincal and Dr. Billy Stanberry for their tremendous help.

I acknowledge the National Renewable Energy Laboratory and the Oak Ridge

National Laboratory for providing excellent research opportunities. I extend my special

thanks to Dr. Andrew Payzant for his sincere help on high-temperature XRD study.

I would like to thank all the staffs of Microfabritech for their wonderful jobs on

creating the best research environment.

I have been blessed with many friends that I met during my study. The people from

Korean Baptist Church of Gainesville are not just friends but brothers and sisters to me. I

express my special thanks to pastor Sohn and Mrs. Sohn for their love. I know that they

never stopped praying for me and my wife.

And finally, all my thanks go to my Lord.















TABLE OF CONTENTS
page

ACKNOWLEDGMENTS ...................................................................................................... iv

LIST OF TABLES........................................................................................................... ix

LIST OF FIGURES............................................................................................................ x

A B ST R A C T ...................................................................................................................... xv

1 INTRODUCTION ....................................................................................................... 1

1.1 System Description............................................................................................... 4
1.2 Statement of Problems............................................................................................. 9

2 BACKGROUND AND LITERATURE REVIEW OF SOLAR CELL AND CuInSe2-
BASED MATERIALS.................................................................................................. 10

2.1 Solar Cell Fundamentals........................................................................................ 10
2.2 Cu(In,Ga)Se2-based Thin Film Solar Cells............................................................ 14
2.3 Fundamentals of CuInSe2-based Materials............................................................ 17
2.3.1 Crystalline Structure and Solid State Electronics of CIS-based Materials.. 17
2.3.2 Structure, Chemistry, and Processing of CIS-based Thin Films.................. 20
2.3.2.1 Structure and chemistry of CIS-based thin films................................ 20
2.3.2.2 Processing of CIS-based solar cells.................................................... 22
2.4 Characterization Techniques for CIS-based Films................................................ 25

3 REACTION KINETICS AND PATHWAYS OF CuInSe2 GROWTH FROM
BILAYER PRECURSOR FILMS: TIME-RESOLVED HIGH TEMPERATURE X-
RAY DIFFRACTION STUDY..................................................................................... 28

3.1 Introduction............................................................................................................ 28
3.2 Experiments ........................................................................................................... 29
3.2.1 Preparation of Precursor Films..................................................................... 29
3.2.2 Time-resolved High Temperature X-ray Diffraction................................... 30
3.2.3 Calibration for Absolute Temperature of Samples...................................... 32
3.3 Results and Discussion .......................................................................................... 33
3.3.1 Reaction Kinetics of CuInSe2 Growth from InSe/CuSe Bilayer................. 33
3.3.2 Engineering of Reaction Pathways.............................................................. 39
3.4 C onclusions............................................................................................................ 40









4 SECONDARY GRAIN GROWTH OF CuInSe2 IN THE PRESENCE OF CuxSe
DRIVEN BY INTERFACIAL FORCES...................................................................... 54

4.1 Introduction............................................................................................................ 54
4.2 Kinetic Aspect of Growth Mechanism of CuInSe2-based Thin Films................... 55
4.2.1 Vapor-Liquid-Solid Growth Model for CIS-based Thin Films................... 55
4.2.2 Experimental Evidence of the Liquid Phase during Growth Process.......... 59
4.3 Experim ents ........................................................................................................... 62
4.4 Results and Discussion .......................................................................................... 63
4.5 C onclusions............................................................................................................ 68

5 EPITAXIAL GROWTH OF CuInSe2 AND CuGaSe2 FILMS ON SINGLE
CRYSTALLINE GaAs SUBSTRATE USING A MIGRATION ENHANCED
EPITAXY REACTOR.................................................................................................. 84

5.1 Introduction............................................................................................................ 84
5.2 Experim ents ........................................................................................................... 85
5.3 Results and Discussion .......................................................................................... 87
5.4 C onclusions............................................................................................................ 91

6 THERMAL MODELING OF A ROTATING PLATEN AND SUBSTRATE IN A
MIGRATION ENHANCED EPITAXY REACTOR................................................. 106

6.1 Introduction.......................................................................................................... 106
6.2 Experim ents ......................................................................................................... 107
6.3 Modeling Equations and Strategy........................................................................ 109
6.4 Results and D discussion ........................................................................................ 112
6.5 C onclusions.......................................................................................................... 114

7 FU TU RE W O RK ......................................................................................................... 124

7.1 High Temperature X-ray Diffraction................................................................... 124
7.1.1 Study of Reaction Kinetics of Binary Phases Formation........................... 124
7.1.2 Reaction kinetics of Cu(In,Ga)Se2 Formation From Precursor Films....... 125
7.1.3 In-situ Investigation of Grain Growth and Epitaxy.................................... 126
7.2 Study of the Relationship between Electrical Properties and Surface Phase of
CuIn(S, Se)2 Thin Films: Engineering of Bandgap and Surface Phase................ 128
7.2.1 B background ................................................................................................ 128
7.2.2 Electronic and Structural Properties of CuInS2 Thin Films....................... 129
7.2.3 Bi-layer Structure for CuInS2-based Absorber Layer................................ 134
7.3 Continuous Deposition of Cd-free Buffer Layers without Breaking Vacuum.... 136
7.3.1 B background ................................................................................................ 136
7.3.2 Alternative Buffer Layers.......................................................................... 136
7.3.3 Growth Process and Properties of Indium Sulfide and Indium selenide ... 137

LIST OF REFERENCES................................................................................................. 140










BIOGRAPHICA L SKETCH ...........................................................................................147















LIST OF TABLES


Table page

3.1 Thermocouple reading and corrected temperature of the specimen measured by the
therm al expansion m ethod ..................................................................................... 42

3.2 Estimated rate constants and Avrami exponents for CuInSe2 formation from bilayer
precursor film s ....................................................................................................... 42

3.3 Estimated rate constants and the exponents based on the parabolic rate law for
CuInSe2 formation from bilayer precursor films ................................................... 42

4.1 Summary annealing conditions.................................................................................. 69

4.2 Composition (ICP) and X-ray diffraction analysis results......................................... 70

5.1 Process conditions for CIS and CGS film growth ..................................................... 92

5.2 Compositional variation of samples CIS 350 and CGS 353 ..................................... 93















LIST OF FIGURES


Figure page

1.1 Schematic top view of the plasma-assisted migration enhanced epitaxy (PMEE)
system ......................................................................................... ............................. 6

1.2 Schematic structure of the platen and the control thermocouple. (a) Top view of the
platen showing two of the 9 substrate-holders and the position of the heater, (b)
Cross-sectional view showing the mounting of the heater above the platen, the
position of a control thermocouple in the open gap, and the position of a glass
substrate fitted in a holder that lies in an incision made in the platen ..................... 7

2.1 Spectral distribution of sunlight. (Figure taken from Moller 1993) ......................... 11

2.2 Band diagrams of an unbiased p-n junction solar cell: (a) in the dark and (b) under
illum ination.......................................................... ............................................... 12

2.3 Schematic diagram of a p-n junction solar cell.......................................................... 12

2.4 Current-voltage characteristics of a p-n junction diode in the dark and when
illuminated, defining the basic parameters ............................................................ 14

2.5 Equivalent electrical circuit of a p-n junction for the two-diode model with diffusion
Is and recombination current ISR, series Rs and shunt Rsh resistance, and light-
generated current IL ......................................................... ...................................... 14
2.6 Absorption coefficient as a function of the photon energy for selected semiconductor
materials (Figure taken from M6ller 1993) .....................................................15

2.7 Schematic structure for a CuInSe2 thin film solar cell..........................................16

2.8 Tetragonal unit cell of an ABX2 chacopyrite lattice: A = Cu, B = In, and X = Se for
CuInSe2 (Figure taken from Bube 1998)................................ ...............................18

2.9 Carrier concentration of p- and n-type CuInSe2 single crystals as a function of the
composition ratio, [Cu]/[In]. Black symbols are n-type and white symbols are p-
type. (Figure taken from Moller 1993)................................... ............. ............. 20

3.1 Structure of the precursor film, InSe/CuSe: (a) schematic drawing of the precursor
film structures, (b) cross-sectional SEM image of the precursor film (x40,000)..43









3.2 Time-resolved in-situ X-ray Diffraction for isothermal heating at 218 C. The last 20
data were collected while heating at 340 C for 30 minutes ................................. 44

3.3 Fractional reaction of CuInSe2 formation during the isothermal runs as a function of
tim e and tem perature..............................................................................................45

3.4 Fractional reaction of CuSe transformation during the isothermal runs as a function
of tim e and tem perature .........................................................................................46

3.5 Sum of the mole fractions of the reactant (CuSe) and the product (CuInSe2) through
isotherm al heating.................................................................................................. 47

3.6 Avrami plots for the isothermal reactions at different temperatures ......................... 48

3.7 Arrhenius plot for the isothermal reactions based on the Avrami analysis: apparent
activation energy for CuInSe2 growth reaction, Ea = 66.0 kJ/mol......................... 49

3.8 Plots based on the parabolic rate law for the isothermal reactions at different
tem peratures........................................................................................................... 50

3.9 Arrhenius plot for the isothermal reactions based on the parabolic rate law: apparent
activation energy for CuInSe2 growth reaction, Ea = 65.2 kJ/mol......................... 51

3.10 Time-resolved in-situ X-ray Diffraction with increasing temperature:
transformation from a-InSe/CuSe to CuInSe2........................................................ 52

3.11 Time-resolved in-situ X-ray Diffraction with increasing temperature:
transformation from a-InSe/Cu:Se to CuInSe2 ...................................................... 53

4.2 Schematic structure of a growing Cu-rich CIS thin film based on the vapor-liquid-
solid m odel............................................................................................................. 71

4.3 SEM images of different regions on the surface showing an apparent difference in
grain size in (a) matrix region and (b) droplet region............................................ 72

4.4 Droplet structures on an as-grown Cu-rich CulnSe2 film with annealing: T = 450 C,
[C u]/[In] = 1.03....................................................................................................... 72

4.5 Schematics of sample structures: (a) without the CuSe layer, control sample, (b)
w ith the CuSe layer................................................................................................ 73

4.6 Cross-sectional SEM images of CuInSe2 films: (a) annealed without the copper
selenide layer on the top of the CuInSe2 layer....................................................... 74

4.6 Cross-sectional SEM images of the CuInSe2 films: (b) annealed with the copper
selenide layer. Both samples were treated with KCN solution to selectively
remove the copper selenide layer after annealing.................................................. 75









4.7 Cross-sectional SEM images of CuInSe2 films: (a) annealed without the copper
selenide layer on the top of the CuInSe2 layer....................................................... 76

4.7 Cross-sectional SEM images of the CuInSe2 films: (b) annealed with the copper
selenide layer. Both samples were treated with KCN solution to selectively
remove the copper selenide layer after annealing.................................................. 77

4.8 CIS (112) diffraction peaks: (a) annealed at approximately 450 C without selenium
overpressure. The sample 359-8K was not annealed. The sample 362-3K and
363-3K were annealed with the copper selenide layer, while sample 362-4K and
363-4K were annealed without the copper selenide layer. All the samples were
treated by KCN to selectively remove the copper selenide after annealing..........78

4.8 CIS (112) diffraction peaks: (b) same temperature condition, but with selenium
overpressure. The sample 359-8K was not annealed. The sample 362-3K and
363-3K were annealed with the copper selenide layer, while sample 362-4K and
363-4K were annealed without the copper selenide layer. All the samples were
treated by KCN to selectively remove the copper selenide after annealing..........79

4.9 CIS (204)/(220) diffraction peaks: (a) annealed at approximately 450 C without
selenium overpressure. The sample 359-8K was not annealed. The sample 362-
3K and 363-3K were annealed with the copper selenide layer, while sample 362-
4K and 363-4K were annealed without the copper selenide layer. All the samples
were treated by KCN to selectively remove the copper selenide
after annealing........................................................................................................ 80

4.9 CIS (204)/(220) diffraction peaks: (b) same temperature condition, but with
selenium overpressure. The sample 359-8K was not annealed. The sample 362-
3K and 363-3K were annealed with the copper selenide layer, while sample 362-
4K and 363-4K were annealed without the copper selenide layer. All the samples
were treated by KCN to selectively remove the copper selenide
after annealing........................................................................................................ 81

4.10 XRD peak areas: (a) annealed at approximately 450 C without selenium
overpressure. The sample 359-8K was not annealed. The sample 362-3K and 363-
3K were annealed with the copper selenide layer, while sample 362-4K and 363-
4K were annealed without the copper selenide layer. All the samples were treated
by KCN to selectively remove the copper selenide after annealing...................... 82

4.10 XRD peak areas: (b) same temperature condition, but with selenium overpressure.
The sample 359-8K was not annealed. The sample 362-3K and 363-3K were
annealed with the copper selenide layer, while sample 362-4K and 363-4K were
annealed without the copper selenide layer. All the samples were treated by KCN
to selectively remove the copper selenide after annealing..................................... 83









5.1 Multiple temperature growth schedule: nucleation at a lower Tub, enhanced surface
migration using a higher Tsub, subsequent growth at a lower Tsub to grow meta-
stable phase with low dislocation density.............................................................. 94

5.2 SEM images ofCuInSe2 (CIS 350): atomic ratio, [Cu]/[In] = 1.16, (a) grown on
(001) GaAs (x3000), (b) grown on Mo-coated soda lime glass (x3000). .............95

5.3 SEM images ofCuInSe2 (CIS 350) grown on (001) GaAs: atomic ratio, [Cu]/[In] =
1.16, (a) dark background area of Figure 5.2 (a) (x20000), (b) bright island area of
Figure 5.2 (a) (x20000).......................................................................................... 96

5.4 AFM images ofCuInSe2 (CIS 350) grown on (001) GaAs: atomic ratio, [Cu]/[In] =
1.16, (a) 10gxl0j (b) 5p.x5 ................................................................................ 97

5.5 SEM images ofCuGaSe2 (CGS 353): atomic ratio, [Cu]/[In] = 1.29, (a) grown on
(001) GaAs (x3000), (b) grown on Mo-coated soda lime glass (x3000). .............98

5.6 SEM images ofCuGaSe2 (CGS 353) grown on (001) GaAs: atomic ratio, [Cu]/[In]
= 1.29, (a) plain view (x20000), (b) side view (x20000) ...................................... 99

5.7 XRD 0-20 scan diffraction data of CuInSe2 film (CIS 354) grown on (001) GaAs:
atom ic ratio, [Cu]/[In] = 0.97............................................................................... 100

5.8 XRD 0-20 scan diffraction data of CuInSe2 film (CIS 350) grown on (001) GaAs:
atom ic ratio, [Cu]/[In] = 1.15............................................................................... 101

5.9 XRD 0-20 scan diffraction data of CuGaSe2 film (CGS 353) grown on (001) GaAs:
atom ic ratio, [Cu]/[In] = 1.29............................................................................... 102

5.10 XRD 0-20 scan diffraction data of CuGaSe2 film (CGS 355) grown on (001) GaAs:
atom ic ratio, [Cu]/[In] = 1.09............................................................................... 103

5.11 SEM images of CuInSe2 (CIS 350) grown on (001) GaAs after KCN treatment, (a)
background area (x20000), (b) island area (x20000) .......................................... 104

5.12 SEM images ofCuGaSe2 (CGS 353) grown on (001) GaAs after KCN treatment:
atomic ratio, [Cu]/[In] = 1.29, (a) plain view (x20000),
(b) side view (x20000)......................................................................................... 105

6.1 Schematic top view of the migration enhanced epitaxy reactor .............................. 115

6.2 Schematic structure of the platen and the control thermocouple. (a) Top view of the
platen showing two of the 9 substrate-holders and the position of the heater, (b)
Cross-sectional view showing the mounting of the heater above the platen, the
position of a thermocouple in the open gap, and the position of a glass substrate
fitted in a holder that lies in an incision made in the platen ................................ 116









6.3 Time-varying view factors between the main heater and the platen at 20 rpm....... 117

6.4 Time-varying view factors for the main heater at 20 rpm ....................................... 118

6.5 Temperature profile on the platen/substrates: thermocouple reading = 650 C,
rotation speed = 20 rpm with thermal-break region............................................. 119

6.6 Temperature profile along the angular position: thermocouple reading = 650 C. .120

6.7 Temperature profile along the radial position at 140 with and without a thermal-
break region: thermocouple reading = 650 C, rotation speed = 20 rpm............. 121

6.8 The effect of the thermal-break region on the pattern of temperature profile within a
substrate at 0 = 140: thermocouple reading = 650 C,
rotation speed = 20 rpm ....................................................................................... 122

6.9 Comparison of the modeling result to the experiment: thermocouple reading = 650
C, rotation speed = 20 rpm .................................................................................123

7.1 Schematic drawing of the precursor film structures: (a) Cu/Se, (b) In/Se,
(c) C u:Se .............................................................................................................. 125

7.2 Schematic drawing of the precursor film structures: (a) CuInSe2/CuSe, (b) mixture
of elemental copper, indium, and selenium ......................................................... 127

7.3 A typical process for CuInS2 absorber layer: Cu-rich CulnS2 thin film is grown and
the secondary phase on the surface is etched out by KCN solution .................... 132

7.4 Schematic structure of an as-grown single layer of the In-rich CuInS2................... 132

7.5 Bi-layer process for CuInS2 absorber layer deposition to prevent the formation of the
In-rich CuInS2 secondary phase, CuIn3S5 ............................................................ 135















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

GROWTH KINETICS AND PROCESSING
OF COPPER INDIUM DISELENIDE-BASED THIN FILMS

By

Suku Kim

May, 2003


Chair: Timothy J. Anderson
Major Department: Chemical Engineering



CuInSe2 (CIS)-based compound semiconductors are increasingly important absorber

layer materials for thin film solar cells. A better understanding of the growth kinetics of

CuInSe2 thin films as a function of the process parameters would benefit the development

of this technology.

The reaction kinetics for formation of CuInSe2 from the bilayer structure InSe/CuSe

was studied in-situ by high-temperature X-ray diffraction. The reaction pathway

produces a diffusion barrier layer that can be schematically represented as InSelCuSe -+

InSe|CuInSe2|CuSe. Two different analyses based on the Avrami and the parabolic rate

laws suggest that the reaction is one-dimensional diffusion controlled. The estimated

apparent activation energy from each model is 66.0 and 65.2 kJ/mol, respectively. The

result demonstrates that the time-resolved high temperature X-ray diffraction provides a

powerful method for studying the reaction kinetics of CuInSe2 growth.









The thermodynamic driving force for formation of copper selenide phase and the grain

size distribution in CuInSe2 films was investigated. Large grains (- a few im) were

observed in the CuInSe2 films annealed with a CuSe layer while films annealed without

this layer exhibited very small grain size (< 0.2 jm). This result suggests a secondary

grain growth mechanism driven by the surface-energy anisotropy is responsible for the

increased grain size.

Epitaxial growth of CuInSe2 and CuGaSe2 on (001) GaAs substrates was attempted.

The result shows that the crystalline structure and its quality strongly depends on the film

stoichiometry, especially the [Cu]/[III] atomic ratio, with Cu-rich compositions showing

higher crystalline quality.

A two-dimensional model of heat transfer in the growth reactor was developed for a

rotating platen/substrate in the molecular beam epitaxial reactor that was used for film

growth. Time-varying view factors were included in the model to solve the problem

dynamically and to account for the fact that the platen rotating at a given angular speed.

The modeling results predict the temperature uniformity on the substrate surface is good

(< 10 C) at a set point temperature in the range 200 to 500 C and rotation rate of 20

rpm.














CHAPTER 1
INTRODUCTION

This chapter presents an overview of this dissertation, description of the deposition

system for processing CuInSe2 (CIS) thin films, and finally a statement of problems

addressed in this dissertation.

The second chapter reviews the key principles associated with CIS thin film

photovoltaic technology. In particular, it presents the basic principles of solar cell device

operation as well as specific features and recent progress in CuInSe2-based solar cell

applications. The material properties, crystal structure, chemistry, and processing of

CuInSe2 thin films are also discussed. Finally, techniques to characterize CuInSe2 thin

films are summarized.

The third chapter presents the result of a study on the reaction kinetics and pathways

of CuInSe2 film growth. The reaction kinetics for formation of CuInSe2 from precursor

films consisting of stacked binary M-Se layers were studied using high-temperature X-

ray diffraction analysis. Different reaction pathways and phase transformations were

observed depending on the deposited precursor film structure. For example, the

isothermal phase evolution of the InSelCuSe couple film was observed at different

temperatures by in-situ time-resolved X-ray diffraction. The pathway produces a

diffusion barrier layer that can be schematically represented as InSelCuSe -+

InSe|CuInSe2|CuSe. This product layer also serves as a nucleation barrier. As a result,

amorphous and crystalline phases simultaneously grow during the isothermal heating.

The rate of reaction exhibits a deceleratory behavior consistent with a diffusion-









controlled reaction mechanism. Two different analyses based on the Avrami and the

parabolic rate laws were performed. The Avrami exponent for each isothermal reaction

is between 0.5 and 0.8, which indicates that the growth reaction is dominantly one-

dimensional, diffusion controlled. The estimated apparent activation energy for this

reaction is 66.0 kJ/mol. The result from the parabolic rate law is very consistent with the

Avrami analysis and an almost identical apparent activation energy (65.2 kJ/mol) was

obtained.

In the fourth chapter, a thermodynamic analysis of the effect of a copper selenide

phase on the grain size distribution of CuInSe2 films is presented. The CuInSe2 films

were deposited onto molybdenum-coated soda lime glass substrates using a molecular

beam epitaxy system. In this study, the composition of the as-grown CuInSe2 films was

controlled to be stoichiometric (possibly slightly Cu-rich). The a layer of CuSe was

deposited on top of the as-grown CuInSe2 films. These bi-layer samples, CuInSe2/CuSe,

were annealed under various conditions. The grain size distribution of the annealed films

was estimated using SEM images and the full width of half maximum (FWHM) of XRD

peaks. Large grains (a few p.m) were observed from samples annealed with the copper

selenide layer while the samples annealed without this layer showed very small grains (<

0.2 tm). After annealing the CuInSe2/CuSe films, the peak area of the (112) orientation

consistently increased, while that of the (220)/(204) orientations decreased, suggesting a

difference in surface energy is important. The results were interpreted by assuming a

thermodynamic driving force, surface-energy anisotropy, is responsible for the grain size

increase of the CuInSe2 film.









In the fifth chapter, preliminary results on the epitaxial growth of CuInSe2 and

CuGaSe2 films on (001) GaAs substrates are discussed. Previous work in our research

group on epitaxial growth of CuInSe2 films on GaAs (001) was reported (Stanbery 2002).

Under certain conditions, crystal structure of the CuInSe2 films was found to be Cu-Au

(CA) structure rather than more commonly observed chalcopyrite (CH) structure. The

results reported in this chapter show that the crystal structure and its quality depended on

the film stoichiometry, especially the [Cu]/[III] atomic ratio. SEM and AFM

examination revealed that the Cu-rich CuInSe2 and CuGaSe2 films consisted of a highly

oriented array of facetted islands on the surface, and that a textured background region

lies adjacent or beneath those islands. The specific surface morphology was an indication

of the epitaxial growth or textured structure of the film. It is believed that Cu-rich

compositions quality of the epitaxially grown films. There was a structural and

compositional distinction between the islands and the background regions of Cu-rich

films. More extensive study is needed to establish detailed process parameters and

improve the quality of the epitaxially grown films.

The final chapter presents a thermal mode of the migration enhanced epitaxial (MEE)

deposition system that was used to produce the CuInSe2-based films. The finite element

method was used to model the heat transfer in a MEE reactor. The thermal domain

included is a rotating platen that holds nine substrates. Time-varying view factors were

employed to solve the problem dynamically, and account for the rotational motion. A

two-dimensional temperature profile of the platen and the substrates was obtained. The

relationship between the actual substrate temperature and the control thermocouple was

also modeled. It is demonstrated that the existence of thermal contact resistance between









the platen and the substrates improves the temperature uniformity of the substrate. The

modeling results predict that the temperature distribution of the substrate surface showed

good uniformity of the given conditions, a set point temperature in the range 200 to 500

C and rotation rate of 20 rpm.

1.1 System Description

The photovoltaic research group at the University of Florida developed a plasma-

assisted migration enhanced epitaxy (PMEE) deposition system to produce CuInSe2-

based absorber layers for thin film solar cell applications. The system is basically a

variant of the molecular beam epitaxy (MBE) system. In addition to the capabilities of a

typical MBE, it is capable of processing nine samples situated on a large rotating platen.

Square substrates of 2 in.x2 in., 2 in diameter wafers (Si or GaAs) or 1 cmxl cm of

square substrates can be loaded onto the platen. The migration enhanced feature of the

system is established by the sequential deposition of each source through a revolution of

the platen rather than simultaneous co-deposition of all the sources.

As a modified MBE system, it creates an ultra high vacuum environment and

molecular beam fluxes of the elemental sources. The pumping unit consists of three

mechanical pumps, one large capacity diffusion pump, a turbo molecular pump (TMP),

and a liquid nitrogen cryogenic pump. The base pressure can be maintained as low as

8x 10'9 Torr with a cryogenic pumping. The pressure during deposition is typically in the

range 10-8 to 10-7 Torr depending on the magnitude of the source fluxes. The pressure

can elevate as high as 10-5 Torr range when an inert gas is introduced to the system to

generate plasma. In standby mode, the system pressure is maintained as low as 5x l07

Torr by the turbo molecular pump and a back-up mechanical pump.









The deposition chamber is divided into four zones, as shown in Figure 1.1. It consists

of a heater zone, a metal deposition zone, a load-lock zone, and a chalcogen deposition

zone (as reviewed in counter-clockwise direction). As described above, the system

adopts a rotating platen that delivers the substrates to all the zones sequentially and

periodically. A radiation heater is located in the heater zone; hence, the substrates and

the platen are heated while they pass through the heater zone. In other zones, the

substrates are slowly cooled since there is no direct heating source. A slight non-uniform

temperature distribution on the platen and substrates is expected due to the complex

design. Two effusion source cells are located in the metal deposition zone. Impingement

of Cu and In fluxes occurs while the heated substrates pass through the metal deposition

zone. Rotation of the platen continuously delivers the substrates to the cooling zone (or

load-lock zone) where neither deposition nor heating occurs. Finally, the substrates enter

the chalcogen deposition zone and deposition of Se or S occurs. The cycle is repeated

with the rotational motion of the platen. The above description about the rotational

motion and sequential deposition assumes counter-clockwise rotation of the platen. The

direction of the rotation, of course, can be either clockwise or counter-clockwise so that

the sequence of deposition may be reversed. The selenium deposition zone was isolated

from the other zones to minimize selenium deposition in the other parts of the system.

Selenium has the highest vapor pressure among all the elemental sources and

correspondingly creates the highest background pressure; hence, it can cause serious

contamination of other parts, such as the ion gauges. The substrates are loaded into the

PMEE system through the load-lock zone. A separate chamber, the load-lock, is








connected to the main chamber, which makes it possible to move the substrates in and out

without venting the main chamber.


>Thermal cracker
/\ Heater zone / ^ ,

Metals (C
deposition Chalcogen zone
zone 0
\ Substrate Plasm cracker
\v<_^/ \ rotation y

\ Ga) 1.



Load lock zone

Figure 1.1 Schematic top view of the plasma-assisted migration enhanced epitaxy
(PMEE) system
Five effusion source heaters (Cu, In, Ga, Se, and S) are located in the main chamber.

Three thermal evaporation sources produce create copper, indium, and gallium fluxes.

Due to the high melting temperatures and the high sticking coefficients of copper and

gallium, a dual filament system was adopted in these sources to ensure a completely

melted surface and prevent condensation on the wall near the exit of the effusion cell.

The source heater for selenium was more complex. It consisted of a cracker and a

crucible. It is known that selenium vapor evaporates as a mixture of several molecular

phases (e.g., Se, Se2, Se6, and Ses). The high molecular weight species do not react as

readily as the other species on the substrate even at high temperature.











Top View Side View

0 = 900
0.434 m
0.252 m

Heater

m Control Thermocouple
0=180 -i 0=00

eater\ .Substrate (Glass)

Platen (Mo)
Inner boundary ring

0 = 2700 Outer boundary ring


Figure 1.2 Schematic structure of the platen and the control thermocouple. (a) Top view
of the platen showing two of the 9 substrate-holders and the position of the heater, (b)
Cross-sectional view showing the mounting of the heater above the platen, the position of
a control thermocouple in the open gap, and the position of a glass substrate fitted in a
holder that lies in an incision made in the platen.
The thermal cracker was designed to reduce the concentration of the high molecular

weight species. The temperature condition for the cracker during deposition is usually in

the range 350 to 1000 C. The second selenium source is located in the chalcogen

deposition zone. This source was designed for plasma cracking of a selenium molecular

flux to yield lower molecular weight Sex. Electron cyclotron resonance (ECR) plasma is

generated and maintained in a sapphire tube that works as a wave-guide and for the

passage of selenium molecular flux. This source cell can be used for both selenium and

sulfur. Both sequential and simultaneous deposition can be conducted by controlling

these effusion cells.

The fluxes from the copper and the indium sources are measured in-situ by EIES

sensors. A closed loop feed-back control scheme implemented using the EIES sensors









for the Cu and the In sources. The EIES sensors can be calibrated by quartz crystal

monitors (QCM) that are located over the source cells. There is no sensor that directly

measures the chalcogen (selenium and sulfur) flux rate in-situ; hence, the flux rate is

calculated by depositing the chalcogens on a substrate at room temperature and

measuring the film thickness. For the chalcogen sources, a closed loop feed-back control

based on temperature is used.

Along with the unique features of the system described above, the current growth

system has both advantages and limitations compared to other deposition techniques.

First, it has the advantages of the MBE system, including a clean ultra high vacuum

environment and a relatively precise control over the flux. Combined with the shutter

operation and the rotational motion of the platen, an operation of atomic layer deposition

mode is also possible. In addition, the design has a relatively high sample throughput

since nine samples processed in one batch. Adopting the sequential deposition scheme

gives more versatile tool, i.e., the rotation speed and the rotation direction can be varied.

There exist some disadvantages of this system as well. Due to the rotational motion of

the platen/substrates, direct in-situ measurement of the substrate temperature is very

difficult. The measurement thermocouple is currently located in the gap between the

platen and the heater, and reads a temperature at a fixed position independent of rotation.

The localized heater location in a rotation system creates a time-dependent non-uniform

temperature distribution on the substrate surface. The growth rate is significantly limited

by the chalcogen flux delivery. Even with high chalcogen flux rate ([Se]/[Cu]+[In] > 5),

it is a challenge to deliver sufficient chalcogen incorporation for growth at high

temperature. This is because the chalcogen deposition zone is removed from the metal









deposition zone, and the transit time between the zones allows the high-vapor-pressure

material to be easily re-evaporated from the surface. As a result, the maximum flux rates

of copper and indium should be also limited, thus making it difficult to grow at high rate.

1.2 Statement of Problems

Cu(In,Se)Se2 (CIGS)-based compound semiconductors are increasingly important as

the absorber layer material of thin film solar cells. The energy conversion efficiency of

the CIS-based solar cells has reached that of the best solar cells made of crystalline

silicon and gallium arsenide. While there have been many studies of the growth

mechanism of the CuInSe2 film, it is not yet fully understood partly due to complex phase

evolution and defect chemistry in that material system. In a Cu-In-Se ternary system,

various binary and ternary phases can be present and transformed into many different

phases during synthesis of the absorber film. For these reasons, the fundamentals of the

growth mechanism for CIGS synthesis are still ambiguous. Such lack of scientific and

engineering basis leads to difficulties in scaling up promising laboratory processes.

Among many important issues, the following five problems are addressed in this

dissertation.

1. Quantitative investigation of the reaction kinetics of CuInSe2 thin film growth using
in-situ time-resolved, high-temperature X-ray diffraction.

2. Study of the grain growth mechanism of CuInSe2 thin films: influence of the CuSe
phase.

3. Epitxial growth of CuInSe2 and CuGaSe2 films on (001) GaAs substrates.

4. Thermal modeling of the PMEE growth system to better understand the relations
between the process parameters and the material properties of the grown films.


5. Development of a suitable growth process of CulnSe2-based absorber layer for
material property improvement and process scale-up.














CHAPTER 2
BACKGROUND AND LITERATURE REVIEW OF SOLAR CELL AND CUINSE2-
BASED MATERIALS

This chapter discusses the key principles and technologies in CIS thin film solar cells.

In particular, section 2.1 describes basic principles and fundamentals of the solar cell

device, and section 2.2 presents specific features and recent progress in CuInSe2-based

solar cell applications. Section 2.3 describes the material properties, crystal structure,

chemistry, and processing of CuInSe2 thin films. Section 2.4 summarizes some of

characterization techniques for CuInSe2 thin films.

2.1 Solar Cell Fundamentals

A photovoltaic device or solar cell converts sunlight (or photon energy) directly into

electricity using the photovoltaic properties of suitable materials. The photovoltaic effect

was discovered by Becquerel, in 1839. Solar cell technology has developed

tremendously during the last three decades (Moller 1993). Silicon was the first

commercially used solar cell material and is still the most usually produced by industry.

A wide range of semiconductors, such as GaAs, CdTe, InP, CdS, and CuInSe2-based

materials, is currently being explored for their potential use in photovoltaic applications.

The wavelength distribution of the sunlight follows approximately the radiation

distribution of a black body at 6000K; however, there are some deviations in the actual

spectral distribution due to absorption effects in the atmosphere. The radiation power is

usually measured in air mass (AM) units. For example, AMO corresponds to the spectral

distribution measured outside the atmosphere perpendicular to the direction of the sun,









and AMI corresponds to the one measured on the earth's surface. When the sun is about

42 above the horizon, the path length is 1.5 atmospheres. The corresponding spectral

distribution is referred to as AM1.5. In the U.S. photovoltaic program, the spectral

distribution for an AM 1.5 radiation source with a total power density of 0.855 kW/m2 has

been adopted as a standard spectral distribution (Bube 1998). Those spectral

distributions of the radiation are displayed in Figure 2.1.



N 2.5 -
2 Black body radiation 6000 K
S2.0 /\
gAMO radiation
1. AM1.5 radiation
S1.0

0.5 .


0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Wavelength jam]

Figure 2.1 Spectral distribution of sunlight. (Figure taken from M61oer 1993)

In photovoltaic materials, illumination (photon energy) generates electron-hole pairs,

and recombination of the pairs can occur at the same time. Both are non-equilibrium

processes and controlled by the device parameters. The photovoltaic device has a built-in

electrical field that separates the photo-generated electrons and holes. The internal

electric field is built by formation of p-n junction, Schottky barrier, or a MIS (metal-

insulator-semiconductor) structure in the device. For most solar cells, the p-n junction

formation is used to build the internal electric field. In thermodynamic equilibrium,

when no electrical current flows, the separated minority carriers are accelerated by the

electric field to the p-type side of the junction (Li 1993). Electrical current should flow









under illumination by this mechanism when those two sides of the p-n junction are

connected externally. A band diagram and schematic view of a p-n junction solar cell are

depicted in Figures 2.2 and 2.3.

Important solar cell parameters can be determined from the current-voltage curve

(Figure 2.4). The short-circuit current, Isc, is the current flow when the voltage drop

across the junction is zero. From the equivalent electrical circuit of a p-n junction (Figure

2.5), it is found that Isc is equal to the photo-generated current, IL, when the series

resistance (Rs) is zero, and Rs reduces Isc.



E EVb ,----- Vb-
.....................EF.. E,
*.................................................... *...................... EF Efp ................................V .-Ef




(a) Dark (b) Under illumination

Figure 2.2 Band diagrams of an unbiased p-n junction solar cell: (a) in the dark and (b)
under illumination.


Front surface contact


n-region


p-region



Back surface contact

Figure 2.3 Schematic diagram of a p-n junction solar cell.









The open-circuit voltage, Voc, corresponds to the voltage drop when the current flow

is zero. The electrical power output is P = IV, which is equal to the area of the rectangle

in Figure 2.4. The maximum area of the rectangle will be Pma. = VIpp for a given current-

voltage curve. The shape of the curve depends on the load resistances, R, and Rsh. The

fill factor, FF, is the parameter that determines the shape of the curve. It is defined by


FF = VPI (2.1)
VIoc l sr


A more square-like curve leads to a larger fill factor, FF. The value of FF varies

between 0.7 and 0.9 for the cells with acceptable efficiencies.

The energy conversion efficiency, q, is determined by

VpIp FF Vpx l,
r _7= = pill (2.2)
'irn irn

where P,, is the total power generated by illumination (photon energy). Therefore, the

value of each parameter in equation (2.2) significantly affects the performance of the

solar cell. These parameters are not independent, and are controlled by the structural and

electrical properties of the solar cell device.

To make solar celsl a competitive energy source, several challenges need to be

overcome. The challenges include achieving high performance (energy conversion

efficiency), low production cost, and high durability to the environment, and long term

reliability. Development action should consider these factors.










A Dark current




""VP


Illuminated


Voc U


Figure 2.4 Current-voltage characteristics of a p-n junction diode in the dark and when
illuminated, defining the basic parameters.



I rI R s--- -
is ISR Rs
IL A S RSh u




Figure 2.5 Equivalent electrical circuit of a p-n junction for the two-diode model with
diffusion Is and recombination current ISR, series Rs and shunt Rsh resistance, and light-
generated current IL.

2.2 Cu(In,Ga)Se2-based Thin Film Solar Cells

A wide range of semiconductors, including GaAs, CdTe, InP, CdS, and CIGS-based

materials, is being explored for their potential use in photovoltaic applications. Among

those candidates, copper indium diselenide, the CIGS-based thin film solar cell, is

considered the most promising technology due to its structural and electrical properties,

including an extremely high absorption coefficient and stable electro-optical properties










(M6ller 1993, Bube 1998, Li 1993). The absorption coefficient is one of the most

important parameters that determine the performance of the solar cell device and

production cost. With higher absorption coefficient, absorber films can be thinner, which

is directed related to the material cost. The values for CuInSe2, silicon and other

important semiconductors are shown in Figure 2.6.

Other advantages of CIGS-based materials include a band gap energy (1.0 ~ 1.7 eV)

that is suitable for either homojunction or heterojuntion device types. A lattice structure

and an electron affinity that matches well with common n-type window materials are also

positives of this material (M6ller 1993). A schematic structure of a traditional CuInSe2-

based solar cell is depicted in Figure 2.7.

Wavelength Ipmj
1.2 1.0 0.9 0.8 0.7 0.6 0.5


CulnSe2 [
103



104 a-Si

SGaAs Imono-Si
IS 103



102 I CdS



1.0 1.5 2.0 2.5
Photon energy |eVI


Figure 2.6 Absorption coefficient as a function of the photon energy for selected
semiconductor materials (Figure taken from M61ller 1993).










Al grid Al grid




ZnO
CdS
Poly-CuInSe2



Substrate


Figure 2.7 Schematic structure for a CuInSe2 thin film solar cell.

Wagner et al. produced an early high-efficiency CuInSe2/CdS solar cell (12.5%) for

the first time by evaporating CdS onto a single crystal of CuInSe2 in 1974 (Birkmire

2001). Numerous efforts have been made to develop processes that improve the device

performance, e.g., using the alloys of Cu(In,Ga)Se2 and CuIn(S,Se)2, which are wider

energy band gap materials. A graded energy band gap structure was developed to

increase the open-circuit voltage while maintaining the short-circuit current.

Recent results show that the cells can be prepared by a variety of methods, while

obtaining efficiencies approximately 13% for CuInS2 and approaching or exceeding the

16% level for Cu(In,Ga)Se2 (Nadenau 1995). The most recent efficiency reported by

NREL is 18.8% (Cu(In,Ga)Se2) for an area of 0.44 cm2, which is similar to that of the

best multicrystalline silicon devices of 19.8% (Birkmire 2001, Zhao 1998). These results

suggest that the CIGS-based solar cell is a very promising alternative to the crystalline-

silicon technologies. NREL has also reported a 43.9 cm2 area device with an efficiency

of 13.7%. Furthermore, the high efficiency devices have been produced by several

groups, and have shown no evidence of degradation of the CIS layer with time or light









exposure. These results clearly show the high potential of CIS-based thin film solar cells

in fabricating high efficiency solar cell modules with low production cost and high

stability, the most important factors in commercializing solar cell modules.

In the industrial sector, Shell Solar, Inc. is at the most advanced stage, offering

Cu(In,Ga)Se2-based 12V modules to customers. Global Solar Energy is developing

flexible module manufacturing systems using polyimide or stainless steel as substrates.

In addition to the companies mentioned above, there are a number of smaller companies

that are at early stages of commercialization of CIS-based modules. Unisun and

International Solar Electric Technology, Inc. are developing CIS-based photovoltaic

technology using non-vacuum deposition processes for the absorber formation. Their

technologies involve spraying/printing of precursor powders onto a Mo-coated glass

substrate, followed by reactive sintering (Birkmire 2001, Roedemrn 1999, Norsworthy

2000).

It is still a challenge to produce large-area modules reproducibly at low cost for

electric power generation. The lack of understanding of the quantitative correlation

between the device performance and the material properties has limited progress towards

manufacturing the high efficiency photovoltaic modules at large scale (Rockett 1994).

2.3 Fundamentals of CuInSe2-based Materials

2.3.1 Crystalline Structure and Solid State Electronics of CIS-based Materials

CuInSe2 crystallizes in a diamond-like lattice structure with a face-centered tetragonal

unit cell that is termed the chalcopyrite structure. The crystal structure is considered as

isoelectronic analogs to binary II-VI or III-V semiconductors with the zinc-blende

sphaleritee) structure (Fig. 2.8) (Givargizov 1991). Each selenium atom serves as the

center of a tetrahedron of two Cu and two In atoms, and each metallic atom is surrounded









by a tetrahedron of selenium atoms. Each anion (selenium) has two A (copper) and two

B (indium) cations as nearest neighbors, whereas in the zinc-blende structure, each anion

has four cations of the same kind as nearest neighbors (M61ler 1993). The lattice

parameters of CuInSe2 crystal structure are a = 0.5784 nm and c = 1.1614 nm.







c





B
a
A X
a

Figure 2.8 Tetragonal unit cell of an ABX2 chacopyrite lattice: A = Cu, B = In, and X =
Se for CuInSe2 (Figure taken from Bube 1998)

There are twelve intrinsic point defects that can form in the ABX2 chacopyrite lattice.

They are three interstitials (Ai, Bi, X1), three vacancies (VA, VB, Vx), and six anti-site

defects. Some of these intrinsic defects are shallow-level defects that can be used to dope

the crystal. In the CIS-based materials, the electrical properties (carrier type, carrier

concentration, and mobility) are determined by the intrinsic defects depending on the

crystal composition instead of a doping by shallow level impurities (extrinsic doping).

An essential feature of chalcopyrite semiconductors is the high density of point defects

that influence not only the optical properties, but also the conductivity and other electrical

properties. Numerous efforts have been made to reveal the physics and roles of the

intrinsic defects (Chang 1999, Burgelman 1997, Sch6n 2000, Klais 2000). It is believed









that Vse, Vcu, Cul,, and the defect pair (2Vcu-Incu) play a dominant role in determining

the electrical properties in CuInSe2 (M6ller 1993, Givargizov 1991). No conclusive

experimental or theoretical results exist, however, that allow the determination of the

structural and electronic properties of these defects in CuInSe2 or any other copper

ternaries and quatemrnaries. This issue remains ambiguous.

Both CuInSe2 and CuInS2 can be made n- or p-type by changing the stoichiometry of

the crystals, whereas CuInTe2 and CuGaSe2 can so far only be prepared as a p-type

material. It makes it difficult to fabricate a solar cell device with these materials. As an

effort to improve the controllability on the electrical properties of the chalcopyrite thin

films, extrinsic doping technology has been studied. It was aimed at improving the

performance of CuInS2-based solar cells; however, it has not been successfully employed

for device fabrication to date.

Typical values for the net hole carrier concentration in the CuInSe2 absorber layer

(single- and polycrystalline) are in the range 1016 to 10 'cm'3. The carrier concentration

sensitively changes depending on the stoichiometry of the materials. Variation of carrier

concentration as a function of the composition of the films is depicted in Figure 2.9

(M6ller 1993). The graph shows that the carrier concentration drops by several orders of

magnitude in a narrow range of composition, and the conductivity changes from p- to n-

type behavior. For this reason, controllability and reproducibility on the composition are

very important issues.










n/p
[cm-3] Poly CulnSe2
10 20
1019
10 18 -
1017
10io16 *

1015 *U
1014 U
I I I I I I I
0.7 0.9 1.1 13
[Cu) / (In]

Figure 2.9 Carrier concentration of p- and n-type CuInSe2 single crystals as a function
of the composition ratio, [Cu]/[In]. Black symbols are n-type and white symbols are p-
type. (Figure taken from Miller 1993)

2.3.2 Structure, Chemistry, and Processing of CIS-based Thin Films

2.3.2.1 Structure and chemistry of CIS-based thin films

CIS-based materials exhibit a distinct transition in morphology and phase segregation

behavior passing through the stoichiometry ([Cu]/[In] = 1). The distinction between the

Cu-rich and the indium-rich films includes the growth mechanism as well as morphology

and micro-structure of the films. In accordance with the pseudobinary (Cu2Se-In2Se3)

phase diagrams (Nadenau 1995), phase segregation in Cu-rich CIS thin film is attributed

to the formation of copper-chalcogenide secondary phases. The secondary phases are

degenerate p-type semiconductors, and dominate the optoelectronic and the surface

properties of Cu-rich films. Most growth techniques used in depositing thin films,

especially physical vapor deposition processes, are generally far from equilibrium, thus,

to know when and where the second phases form is critical to fabrication of CIS-based

solar cells and development of CIS growth process. In Cu-rich CIS films, the individual

grains are significantly larger than those in indium-rich films. These grains prefer to be









(112) oriented in Cu-rich films whereas indium-rich films additionally exhibit strong

(220)/(204) orientation.

For Cu-rich CIS system, the secondary phase observed is copper chalcogenide or

copper selenide, mostly Cu2.xSe, and CuSe (Rockett 1994, Ramanathan 1998, Niemi

1990). Surface composition analysis revealed that even a minor excess of copper in the

bulk leads to very Cu-rich surfaces (Nadenau 1995, Rockett 1994, Ramanathan 1998,

Tuttle 1995). It seems to be caused by secondary phase segregation at the film surface.

X-ray Diffraction (XRD) also indicated segregation of these phases on the surface. The

Cu-rich secondary phases can be selectively removed by etching with KCN solution. The

compositional variation of Cu-rich films due to the KCN treatment has been shown to be

consistent with removal of the copper selenide and to be extremely rapid, as expected for

removal of a near-surface layer (Rockett 1994). CIS-based thin film is often grown

through deposition of Cu-rich films and etching the secondary phases. Several models

for the growth mechanism of CIS thin films under a Cu-rich composition have been

proposed. Klenk et al. (Klenk 1993) reported a significant difference in surface and bulk

compositions in co-evaporated Cu(In,Ga)Se2 (CIGS) polycrystalline films grown under

Cu-rich conditions, and proposed a growth model that the grain growth is controlled by

the interaction of vapor-liquid-solid phase (VLS model). Wada et al. (Wada 1997)

suggested that growth of the CIS thin film takes place through the segregated Cu2Se solid

phase by a "topotactic reaction", after they observed a Cu2Se phase on the CIS grain

surfaces by transmission electron microscopy (TEM). Drastic changes of average

molecularities from m>l to m=0.92-0.93 (m = [Cu]/[In] atomic ratio) and hole

concentrations from p> 1019cm'3 to as low as p=7.5 x 1016cm'3, have been observed from a









epitaxially grown CuInSe2 by MBE before and after the KCN treatment. High hole

concentration and Cu-rich compositions of the as-grown films were attributed to the

copper selenide phase. Their result strongly suggests that the stoichiometry of CIS films

(the ratio [Cu]/[In]) is slightly below unity and independent of the supplied Cu- and In-

flux; the stoichiometry of the CIS films are self-limiting so as not to exceed unity under

Cu-rich growth conditions by operating in the 2-phase region of the phase diagram.

For indium-rich thin films, the phase segregation behavior and the growth mechanism

are evidently different from those of Cu-rich thin films. Indium-rich crystallites are

tetrahedrally shaped, which indicates an anisotropy in the growth velocity of the { 112 }

planes (Nadenau 1995). This anisotropy, ascribed to limitations in chalcogen supply, is

not observed in Cu-rich films. For a wide range of indium-rich compositions, the indium

chalcogen secondary phase is not observed in the films. The whole region consists of

single phase chacopyrite. At the surface, however, a secondary phase has been observed

in the near surface region of indium-rich films. It was originally termed an ordered

defect crystal (ODC) phase. The phase (CuIn3Ses) is a defect chalcopyrite structure and

n-type due to strong compensation of the defects. The lattice constant of this phase

nearly matches that of the CuInSe2 lattice. The ODC was found in most high efficiency

CuInSe2 solar cells and believed to do an important role in determining the electrical

properties of the device. Further study is needed to better understand the fundamentals of

CuInSe2-based thin film.

2.3.2.2 Processing of CIS-based solar cells

Co-evaporation is one of the most widely used processes for growth of CIS-based thin

films. The technique delivers a flux of each source to the substrate simultaneously. This

is often considered as an appropriate way to grow high quality films; however, it also has









limitations for achieving uniform film deposition on a large area substrate. A typical

substrate temperature during deposition is in the range 450 to 550C, which is needed to

form the absorber from the elements and to form large grain size. The simplest co-

evaporation process design is the 'single-layer process'. In this process, the flux rate of

each source is maintained constant. It delivers a film that is homogeneous across the

thickness since the elemental evaporation ratios directly determine the final film

composition. Better controllability and reproducibility are expected due to its simplicity

and directness. It was recognized early, however, that a single-layer does not result in

optimum cell performance.

The empirical approach of depositing a bilayer, i.e. indium-rich layer onto a Cu-rich

layer, was introduced to improve the film quality (Nadenau 1995). The co-evaporation

process allows changing the film composition across the film thickness by changing the

elemental flux rates with time. As pointed out above, the Cu-rich thin film is known for

its large grain size and higher crystallinity. The CIS films is grown under Cu-rich

conditions during most of the growth process to achieve the benefits of large grain size

and high crystallinity. The desired composition, slightly indium-rich, can be attained

through the second layer deposition. By this way, the bilayer process can grow a large

grain CIS thin film that has slightly indium-rich composition.

Tuttle et al. (Tuttle 1995) proposed a more detailed growth mechanism in their growth

model that is based on the VLS model. It consists of (a) initial atomistic accommodation,

reaction, and nucleation, (b) CuInSe2 and CuxSe island formation, (c) CuInSe2

coalescence with vertical phase separation, and (d) CuxSe conversion and local epitaxial

growth. The phase segregation in (b) is lateral, that is, CuInSe2 and CuxSe co-exist side









by side. According to their model, the lateral phase segregation is transformed to a

vertical one where two phases exist as separate layers, layer by layer. Many variants of

the bilayer process have been used to grow high quality CIS thin films.

The sequential deposition process has been also widely used and studied. Its

advantages are its potential for uniform deposition on a large area substrate and the

simplicity of the process. The precursors, elemental layers or binary phase layers (In-Se

and Cu-Se) are usually deposited at low temperature. It is followed by selenization and

annealing at high temperature. Several research groups have fabricated the CIS-based

photovoltaic devices with efficiency over 12% using the sequential process. The

precursor layers can be grown by various techniques, such as sputtering, evaporation,

electro-deposition, and e-beam evaporation.

Another critical process that significantly affects the device performance is the

deposition of the buffer layer. To date, the highest conversion efficiencies of CIS-based

solar cells have been achieved using a CdS buffer layer deposited by chemical bath

deposition (CBD), followed by deposition of a ZnO window layer. There have been

many studies to better understand the roles of the CdS buffer layer in the solar cell

operation. The Cd atoms are believed to form Cdcu interstitial donor defect, providing

the required near-interface charge and restoring the positive surface charge. Kronik et al.

postulates that the donor defect creates significant band-bending at the interface (Kronik,

2000). Several research groups identified the diffusion of Cd into the absorber layer.

Nakada et al. (Nakada 1999) provided direct evidence of Cd diffusion using high

resolution transmission electron microscope (TEM) equipped with energy dispersive x-

ray spectroscopy (EDS). CBD grown CdS buffer layer has its own advantages, the









simplicity of the process, good reproducibility, and potential for large area processing.

However, there is a great interest in replacing CdS by a cadmium-free buffer layer for

environmental reasons and possible gains in efficiency associated with wider band gap

energy. For same processes, a vacuum deposition technique would be competitive with

other processes. CBD grown In(OH)xSy shows similar result to the standard CBD grown

CdS (Birkmire 2001, Braunger 1996, Hariskos 1996). NREL recently reported 13.5%

efficiency Cu(In,Ga)Se2 solar cell without any buffer layer (Ramanathan 1998).

Deposited films of InSe and InS are also potential candidates since it can eliminate the

step of breaking vacuum after the absorber layer growth. It will significantly reduce

processing time and contamination problem.

2.4 Characterization Techniques for CIS-based Films

The surface morphology and the cross sectional structure of CIS thin films can be

studied using scanning electron microscopy (SEM) (Tuttle 1991, Walter 1996, Rastogi

1999). The instrument uses electrons accelerated by tens to hundreds kilovolts to

produce the image, and it can resolve micro-scale objects. It is a versatile instrument for

examining the microstructure of solid surfaces, because it combines high spatial

resolution with depth of field in the same image, and requires minimal sample

preparation (Runyan 1998). Atomic force microscope (AFM) can provide more detailed

information on the surface morphology, e.g., microstructures and roughness of the

surface.

SEM can be equipped with energy dispersive spectroscopy (EDS) or wavelength

dispersive X-ray spectroscopy (WDS) that are compositional analysis instruments. Both

techniques have been widely used for measuring the bulk composition of CIS thin films.

Reliability of the results significantly depends on the film structure since those









instruments collect the information of local region of the objects. Both techniques,

hence, are convenient and reliable for compositional analysis of homogeneous films.

Those techniques, however, may not be very reliable for multi-layer structures or

sequential precursor layers that have compositional variation with film depth.

Auger electron spectroscopy (AES) has been used for obtaining depth profiles of

elemental composition in CIS thin films. Auger electron spectroscopy is a surface

sensitive probe and uses an incident electron beam of 3 to 5keV that is somewhat lower

than that used for the electron microscope, but like a microprobe, the beam scatters

within a small sample volume. To depth profile, sputter etching is used while an Auger

spectrum is being collected. As the etch proceeds, Auger electrons from specimen

provide a measure of the elemental concentration of a freshly exposed layer. To achieve

the best result, the etch rate should be carefully calibrated considering composition,

structure, and chemistry of the sample. Different phases along the film depth can be

detected by analyzing the compositional variation. Phase segregation on the surface or at

the back contact has been observed and analyzed using AES (Rockett 1994, Nakada

1999, Scheer 1994).

X-ray diffraction (XRD) is one of the most commonly used techniques for

characterization of crystalline phases, e.g., thin films, single crystals, and powders. It can

provide information about crystalline structure, stress, and existence of specific phases in

the samples. One of its advantages is that there exists an enormous amount of reference

for XRD spectrum on many crystalline phases.

Recently, Raman spectroscopy has been used as another tool to detect various phases

in CIS films. Some of the secondary phases in CIS thin films have overlapping XRD









peaks; hence, it is not easy to resolve the phase. Raman spectroscopy can provide

additional information to assist in detection of these phases. Micro-Raman technique has

been employed to detect secondary phases in a micro-structure on the surface of CIS thin

film (Stanbery 1999).

Inductively coupled plasma (ICP) is known as one of the most reliable quantitative

analytical techniques for measuring the overall composition of various samples. Sample

preparation step is critical to achieving a reliable result. The samples, e.g., thin film,

single crystalline, or powder, need to be completely dissolved in an appropriate solvent.

The dissolved sample is fed into the instrument and fragmented to the elemental level by

ICP. Photons are released from the reaction in the plasma. With a unique wavelengths

for each element, the intensity of the peaks is proportional to the absolute amount of the

corresponding element in the sample. It is noted that only the overall composition of the

samples can be obtained by measuring the peak intensity. Our research group used

combination of EDS (or WDS) and ICP to achieve compositional analysis.














CHAPTER 3
REACTION KINETICS AND PATHWAYS OF CUINSE2 GROWTH FROM
BILAYER PRECURSOR FILMS: TIME-RESOLVED HIGH TEMPERATURE X-RAY
DIFFRACTION STUDY

3.1 Introduction

Compound semiconductors based on CuInSe2 are increasingly gaining acceptance as

absorber-layer materials for thin film solar cells. The energy conversion efficiency of

CuInSe2-based solar cells has already reached that of the best solar cells made of

crystalline silicon and of gallium arsenide. While there have been many studies of the

growth mechanism for CuInSe2 films, (Tuttle 1995, Rockett 1994, Klenk 1993, Nadenau

1995) to date key details remains not fully understood partly due to the complex phase

evolution and defect chemistry present in the material system. Various binary and

ternary phases can be simultaneously present in a Cu-In-Se ternary system, (Klenk 1993,

Nadenau 1995, Change 1999) and furthermore, phase transformations may occur during

the growth processes. Given that important transformations occur during the growth

process, ex-situ studies may fail to reveal key aspects of the reaction kinetics of CuInSe2

formation. A more promising approach is to conduct an in-situ study of the phenomena

using time-resolved high temperature X-ray diffraction analysis. This technique infers

kinetic information from changes observed in the XRD data as a function of time as the

film is heated. Careful analysis of in-situ XRD data with an advanced software JADE 6

(JADE) permits the quantitative investigation on the reaction kinetics.

The process followed to generate a CuInSe2 film consists of starting with a stack of

two binary films serving as a precursor. Examples of stacked binary films are InSe/CuSe









(i.e., CuSe film grown over an InSe film), InSe/Cu:Se (i.e., an elemental mixture of

copper and selenium deposited over an InSe film), and CuSe/In:Se. In all cases, the

bottom binary films were grown over a thin (0.4 mm) sodium free glass substrate. The

films were deposited in a modified molecular beam epitaxy reactor, and the total

thickness of the stack was approximately 800 nm. The precursor stack is then heated to

initiate a reaction leading to the formation of CuInSe2 film, and the evolution of the

reacting system is observed via in-situ high temperature XRD.

This article presents the results of an in-situ isothermal time-resolved X-ray diffraction

study of CuInSe2 film growth from the binary precursor stack CuSe/InSe. Quantitative

analyses were performed to estimate the reaction order and the apparent activation

energy, as well as to elucidate mechanistic details of the film growth pattern.

3.2 Experiments

3.2.1 Preparation of Precursor Films

The deposition method used to grow the precursor films is called migration enhanced

epitaxy (MEE), a variant of the classical molecular beam epitaxy (MBE) approach. As in

MBE, an ultra high vacuum environment and effusion cells are employed to generate

molecular beam fluxes of all the elemental sources. In contrast with a typical MBE

system, the MEE reactor used is capable of accommodating multiple substrates (up to

nine) positioned on a large rotating platen. Another unique feature of the MEE system is

the sequential deposition of each source material through a revolution of the platen, rather

than a simultaneous co-deposition from all the sources. A most relevant difference,

however, lies on the fact that the rotating platen of the MEE system passes each substrate

through a flux-free zone, namely a relaxation zone that enhances the potential for

adsorbed atoms to migrate seeking the most thermodynamically favored configurations.









The direction of rotation can be either clockwise or counter-clockwise, so that the

sequence of deposition may be reversed. The base pressure of the system can be

maintained as low as 8x 10-9 Torr, and the pressure during deposition is in the range of

10-7-108 Torr depending on the operation conditions. Further details of the deposition

technique and experimental apparatus are given elsewhere (Stanbery 2002).

For the InSe/CuSe structure, first an indium selenide layer was grown on top of a thin

(-0.4 mm) sodium-free thin glass substrate in the MEE system with a substrate

temperature of approximately 250 C. The films were made to have a slightly selenium-

rich composition ([In]/[Se] 0.95). Second, a copper selenide layer was deposited on the

as-grown InSe layer at a lower temperature condition (-150 C) to minimize any

potential reactions between the InSe and CuSe layers. The final stacked binary structure

is depicted in Figure 3.1. Room temperature XRD data shows that the InSe phase is

amorphous, and that the CuSe phase is polycrystalline. The cross-sectional SEM image

in Figure 3.1 (b) also shows that no grain structure is observed in the bottom InSe layer,

while the top CuSe layer has a large-grain structure. The composition of the films was

measured using the inductively coupled plasma technique.

3.2.2 Time-resolved High Temperature X-ray Diffraction

Time-resolved high-temperature X-ray diffraction data were collected while the

precursor InSe/CuSe bilayers were isothermally heated in stagnant air. The experimental

temperatures used ranged from 220 to 270 C. Initially, the samples were mounted on a

platinum strip heater, and XRD data was collected at room temperature. Then the

samples were heated at rate of 120 C /min until the temperature reached a point 20 or 30

C below the desired target temperature. At this point the temperature was allowed to









stabilize, and then the system was again heated again at 120 C /min until the target

temperature was reached. XRD data was collected for a period of time, typically an hour,

at the target temperature. Then all substrates were subjected to a final isothermal

temperature treatment at 340 C for 30 minutes in order to ensure the completion of all

reactions (i.e., accomplish a total conversion of the reactants into the final product phase).

XRD data was also collected during this last isothermal treatment.

The high-temperature X-ray diffractometer used consisted of a Scintag PAD X vertical

0/0 goniometer, a Buehler HDK 2.3 furnace, and an mBraun linear position sensitive

detector (LPSD). The LPSD was centered at 28 20 and covered a 10 range, i.e., from

23 to 33. In contrast to conventional X-ray point scanning detectors that perform the

scanning step-by-step from lower to higher angles, the LPSD collects the XRD data

simultaneously over the 10 window, dramatically shortening the data collection time.

This permits in-situ time-resolved studies of phase transformations, crystallization, and

grain growth. The collection time for the 10 20 window was set to be either 19.5 or 34.5

seconds, depending on the temperature range. Figure 3.2 displays the time-resolved

XRD data collected for a sample isothermally heated at 218 C. To obtain the fractional

reaction (a), the areas for the CuSe (006) peak and the CuInSe2 (112) peak were

estimated from the diffraction data using the JADE software. The values were

normalized assuming that the reactants are completely transformed to crystalline CuInSe2

after each run, and that the texture of the CuInSe2 does not appreciably change through

the entire heating process. Figure 3.3 displays the fractional reaction as a function of

time for all the temperatures considered.









Non-isothermal heating was used to investigate the phase evolution of the samples, as

follows. The InSe/CuSe samples were heated to several target temperatures. The

samples were held at isothermal conditions for 15 minutes, and sequential longer-range

(20-55 20) high-temperature X-ray diffraction scans were collected every 85 seconds.

After 15 minutes, the temperature was raised to the next target value, and the process was

repeated. The longer-range high-temperature X-ray diffraction scans were used to trace

all the intermediate phases and their transformations, and they were realized by

combining high-temperature X-ray diffraction data from four discrete LPSD positions.

3.2.3 Calibration for Absolute Temperature of Samples

Since the thermocouple used to measure the temperature was welded to the bottom of

the strip heater, a temperature calibration procedure was required to estimate the actual

film temperature. Accurate measurements of the thin film temperatures were achieved by

calibrating to the known thermal expansion of silver, chosen due to its high coefficient of

expansion (19.5x1O-6 /K) (Brand 1955). A fine silver powder was carefully dispersed on

the surface of a 0.4 mm glass substrate, and was then annealed at a temperature well

above the range of the experimental conditions to make a solid thermal contact and

minimize any undesired thermal effects during the calibration runs. For each run, a

control program was set to heat the samples at 60 C/min to the target temperatures, and

followed by a hold period until the thermocouple reading became constant. The position

of the silver (331) peak was measured from room temperature to 440 C thermocouplee

reading) and used to calculate the lattice parameter L(T) from the expression

L(T)=dhk Vh 2+12+k2 (3.1)









where h=3, k=3, and 1=1, and dhki is the spacing between atomic planes inferred from the

peak position.

The thermal expansion ratio was taken from the correlation reported by Touloukian et

al. (Touloukian 1977)
L(T)-Lo
L(T)-L0 = -0.515+1.647xlO-3T+3.739x10-7T2+6.283xl0-"T3 (3.2)
Lo

where Lo is the lattice parameters of specimen at To = 293 K, and T is measured in

degrees Kelvin. The correlation is valid in the temperature range 200 K < T < 1200 K.

Substituting (3.1) into (3.2) yields

d, h22 +k2 L = -0.515+1.647x10-3T+3.739xl 0-7T2+6.283x10-"T3 (3.3)
L0

The left-hand side of (3.3) contains known quantities. The resulting cubic equation

(3.3) can be directly solved for the unknown value T, which represents the surface

temperature of the film. Table 3.1 displays the deviation between the thermocouple

reading and the temperature of the specimen estimated by the thermal expansion method.

The difference between the thermocouple reading and corrected temperatures reported in

Table 3.1 ranges from 52-59 C, and this difference can be most likely attributed to the

low thermal conductivity of the glass substrate. The calibration experiments were

replicated, and the results were reproducible within 2 C.

3.3 Results and Discussion

3.3.1 Reaction Kinetics of CuInSe2 Growth from InSe/CuSe Bilayer

According to the room temperature XRD analysis, the top CuSe layer is a

homogeneous crystalline phase with an atomic ratio 1:1, and the bottom InSe layer is









either an amorphous or a nanocrystalline phase with an atomic ratio 1:1.05. The expected

interfacial reaction is

CuSe + InSe -> CuInSe2 (3.4)

Time-resolved high-temperature X-ray diffraction scans showed that the

transformations produce a crystalline CuInSe2 phase product, as expected from reaction

(3.4), and that there is no evidence of other intermediate phases.

It is well known that in multilayer thin film systems, the product layer grown at the

initial interface of the reactants acts as a nucleation barrier as well as a diffusion barrier.

The nucleation barrier often forces a meta-stable amorphous phase to form prior to any

crystalline phase (Ma 1991, Johnson 1986). Amorphous and crystalline phases may grow

simultaneously during subsequent reaction (Clevenger 1990). Additional heating may

cause a further transformation of the amorphous phase to crystalline phase and may also

induce additional nucleation reactions.

Analysis on the changes in time of the XRD peaks shows that both the product phase

(CuInSe2) formation and crystalline reactant-phase (CuSe) consumption follow the same

deceleratory reaction trend (Figure 3.3 and 3.4), which can be attributed to diffusion-

controlled reaction kinetics. Figure 3.5 demonstrates that the sum of the mole fractions

of the CuSe reactant and of the product quickly falls below unity during the early stage of

the isothermal heating process, because the rate of the CuSe consumption is faster than

that of the CuInSe2 formation. This implies that there is an intermediate phase present

before the final product is fully formed. As the long-range high-temperature X-ray

diffraction scans showed no evidence of intermediate crystalline phases, the hidden

intermediate phase is most likely amorphous CuInSe2. Under this perspective, during the









initial stage of the heating process the sum of the mole fractions falls significantly below

unity, an effect that can be explained by the formation of the amorphous CuInSe2 phase,

which is initially prevented from evolving into a crystalline phase due to the nucleation-

barrier effect. As the heating energy progressively overcomes the nucleation barrier, the

CuInSe2 material also progressively crystallizes, and the sum of mole fractions starts to

rise towards unity. The combined amorphous and crystalline CuInSe2 interfacial layer

also acts as a diffusion barrier, opposing the one-dimensional diffusion of the reactants

shown on the left-hand side of reaction (3.4). In summary, the InSe|CuInSe21CuSe

structure introduces both a nucleation and a diffusion barrier. As the isothermal heating

continues further, the sum of the mole fractions now rapidly increases towards to unity

(Figure 3.5). This indicates that the amorphous phase fully transforms into a crystalline

phase. The XRD data of Figure 3.5 shows that for all cases the amorphous phase is

completely consumed during the final heating step at 340. Another plausible

explanation for the mole sum to be lower than unity is that the CuSe phase becomes a

solid-solution before the crystalline CuInSe2 phase forms. Further study, perhaps

including detailed microscopic analysis, will be needed to better understand the initial

stage of the growth mechanism.

The analysis proposed above is consistent with the prevailing perspectives on the

growth mechanism documented in the literature. It has been claimed that during the

initial steps of solid-solid reaction, the formation of product is rapidly accomplished on

the attainment of reaction temperature (Bamford 1980). The initial nucleation reactions

at the original interface of the bilayer structure may occur even before the temperature

reaches the target values likely via a fast two-dimensional growth mode, rapidly evolving









to impingement. This gives rise to a nucleation barrier that would dramatically reduce

the nucleation rate, so that the nucleation reaction is expected to be rapidly deceleratory.

The growth of the CuInSe2 phase is governed by the planar nature of the precursor

film structure. During the early stage of the reaction, heterogeneous nucleation and fast

saturation is expected to occur at the original interface. The in-plane material transport is

likely to be sustained by rapid interface diffusion, so that at the initial instants the growth

at the original interface plane may not be limited by diffusion. The nucleated CuInSe2

phase grows and coalesces into a continuous layer. After the initial transient stage, the

product layer thickens by a diffusion-limited process that is likely to be one-dimensional

and perpendicular to the original interface (Ma 1991, Clevenger 1990). The subsequent

growth of the product phase, the main reaction stage, is now controlled by the diffusion

of one or more species through the product layer.

The kinetics of the growth reactions in terms of activation energy and reaction order

have been investigated using two solid-state reaction models, namely, an Avrami model

and a parabolic rate model. Avrami analysis is a widely used method for the preliminary

identification of the growth rate law. It has been shown that the method yields

satisfactory fits to relevant experimental data (Bamford 1980). The transformation

kinetics under isothermal heating is described by

a = l-exp(-(kt)" (3.5)

or equivalently, by

ln(-ln( 1-a)) = nlnt + nlnk (3.6)

where the fractional reaction a represents the volume fraction transformed, k is kinetic

rate constant, and n is the Avrami exponent. This analysis has been advocated by Hulbert









(Hulbert 1969), who showed that the Avrami exponent can vary between 0.5 and 1.5 in

the case of one-dimensional diffusion-controlled reactions. The value of n is close to 0.5

if the nucleation is instantaneous, and close to 1.5 if the nucleation rate is constant

throughout the reaction. It is well known that thickening of thin planar structures after

complete edge impingement is realized through one-dimensional growth mode (Christian

1975).

Figure 3.6 displays the Avrami plots for the isothermal heating of the precursor films

at different temperatures. Data were taken only for 0.1 < a < 0.95 to minimize the

experimental error. Clearly, the data are fit well by the Avrami model (3.6). Table 3.2

shows the kinetic rate constant and Avrami exponent estimated from the plot. The value

of the Avrami exponent is between 0.5 and 0.8, and consistently increases with

temperature. This indicates that the growth mechanism is through one-dimensional

thickening of the product layer, and that it is also diffusion-controlled. The increase of

the n value with temperature is believed to be caused by the enhancement of the

nucleation processes with temperature. Higher temperature conditions can overcome the

nucleation barrier, so that the nucleation rate during the subsequent growth of the product

phase increases with temperature. The temperature effect on the Avrami exponent has

been previously reported via simulation studies (Pascual 1996).

The Avrami parameters and the Arrhenius equation

k = AexpE El'^
T) (3.7)

were used to estimate the apparent activation energy E,, for the CuInSe2 formation

reaction from the bilayer precursor films. Figure 3.7 shows the high linearity of the









logarithmic Arrhenius plot for In k. The slope of the line yields the estimate Ea = 66.0

kJ/mol.

A second study was performed based on a diffusion-controlled rate law. The simple

parabolic kinetic model

a = (kt)X (3.8)

describes well a process where the interface area is constant and the diminution of

reaction rate is a consequence of increasing thickness of the diffusion barrier. The above

equation can be written in the equivalent logarithmic form

1 1,
Ina = -Int + -Ink
2 2 (3.9)

This expression (3.9) has been shown to be obeyed by one-dimensional diffusion

process, such as oxidation of metals, where the reactant is in the form of a thin sheet.

Figure 3.8 shows the plot of In a vs. In t for the same XRD data used in the Avrami

analysis. Table 3.3 gives the slope of each curve and the resulting kinetic rate constants

extracted from the slopes of the plot. The slope is approximately equal to 0.5 for all the

runs throughout most of the reaction period, which strongly indicates that the reaction is

one-dimensional diffusion controlled. This result is consistent with the conclusions

drawn from the Avrami analysis. The Arrhenius equation and the parabolic law

parameters were used to produce another estimate of the apparent activation energy

(Figure 3.9). The result yielded the apparent activation energy Ea,, = 65.2 kJ/mol, which is

consistent with the value Ea,, = 66.0 kJ/mol obtained from the Avrami analysis. Hence,

the results from two different analyses are consistent and lead to an identical conclusion,

namely that the reaction is one-dimensional diffusion controlled.









3.3.2 Engineering of Reaction Pathways

As described above, non-isothermal heating was used to investigate phase evolution

and reaction pathways of the samples as follows. Different precursor samples,

InSe/CuSe, and InSe/Cu:Se, were heated to several set points. The top layer of the

second sample, Cu:Se means mixture of elemental copper and indium that was prepared

depositing copper and indium simultaneously at room temperature. At each set point, the

samples were held for 15 minutes during which sequential longer-range (20-55 2theta)

HTXRD scans were collected every 85 seconds. After 15 minutes, the temperature was

raised to the next set point at a ramp rate of 60 C/min, and the process was repeated.

The longer-range XRD scan was selected to trace all the intermediate phases and their

transformations, and was accomplished by combining HTXRD data from four discrete

LPSD positions. It was observed that the precursor films go through different reaction

pathways depending on the precursor film structure. This indicates that the reaction

pathway can be controlled designing the precursor film structure.

Figure 3.10 displays the high temperature X-ray diffractions for InSe/CuSe precursor

films heated under vacuum conditions. It depicts well the phase evolution with time

through the reaction. The CuInSe2 phase forms directly from the reactants and no

intermediate phase is detected. During the entire reaction, CuInSe2 is the only crystalline

phase to form according to the HTXRD data. The apparent reaction pathway is CuSe +

InSe -- CuInSe2. With an identical heating process under air and helium atmosphere, the

reaction was not completed at the end of the run while the reactants were completely

transformed to CuInSe2 phase under vacuum well before the run ends. Higher pressure

conditions seems to slow down the reaction. It was previously reported that the low









pressure leaded to completion of the reaction at lower temperature (Lakshmikumar 1995,

Lakshmikumar 1996), which is consistent with our result. Further study is needed to

understand the mechanism.

Another precursor film, InSe/Cu:Se, went through different reaction pathway. The top

layer of this precursor film consists of elemental copper and selenium instead of a binary

phase. Figure 3.11 demonstrates that Cu7Se4 phase forms first from the top layer. It is

then transformed to CuSe phase. Main reaction of CuInSe2 formation begins after CuSe

phase forms probably because temperature did not reach high enough to trigger the

CuInSe2 formation until that point. It appears that formation Cu7Se4 and CuSe are

sensitively affected by temperature conditions. These results demonstrate that the

reaction pathway can be engineered using various precursor film structures. The

precursor film structure and the process parameters can be optimized through the

HYXRD study.

3.4 Conclusions

Time-resolved high temperature X-ray diffraction analysis has been successfully

employed to conduct quantitative analyses of the reaction kinetics of the CuInSe2 phase

formation from bilayer InSe/CuSe precursor films. Transformation from the precursor

films into the final phase was clearly observed by in-situ X-ray diffraction scanning

during isothermal heating. The analysis results based on the Avrami and the parabolic

rate law models support the conclusion that the reaction mode is one-dimensional

diffusion controlled. The estimated apparent activation energies from these analyses are

66.0 kJ/mol, and 65.2 kJ/mol, respectively.

Careful analysis of XRD data showed that there was a non-crystalline intermediate

phase, most likely amorphous CuInSe2, during the initial stage of the isothermal heating









process. The combined amorphous and crystalline CuInSe2 interfacial layer functions as

a diffusion barrier as well as a nucleation barrier. Further research should perhaps focus

on resolving the exact nature of the non-crystalline intermediate phases, so that a more

fundamental understanding of the initial growth kinetics can be achieved.

The reaction pathway for CuInSe2 formation could be altered using various precursor

film structures, which was confirmed using the high temperature XRD. These results

demonstrate that the time-resolved high temperature X-ray diffraction provides a

powerful method for studying the reaction kinetics of CuInSe2 growth from the precursor

films.











Table 3.1 Thermocouple reading and corrected temperature of the specimen measured
by the thermal expansion method.

Thermocouple reading Absolute temperature of the specimen (C)
(C) Air Vacuum He
270 218 218 211
290 236 234 227
310 253 251 244
F330 271 268 261


Table 3.2 Estimated rate constants and Avrami exponents for CuInSe2 formation from
bilayer precursor films.

Temperature (C) k (s') Avrami exponent (n)
218 0.000298 0.573
236 0.000562 0.649
253 0.000945 0.722
271 0.00148 0.760


Table 3.3 Estimated rate constants and the exponents based on the parabolic rate law for
CuInSe2 formation from bilayer precursor films.

Temperature (C) k (s-) Exponent (n)
218 0.000116 0.446
236 0.000190 0.450
253 0.000347 0.502
271 0.00539 0.514

















(a)





















(b)

Figure 3.1 Structure of the precursor film, InSe/CuSe: (a) schematic drawing of the
precursor film structures, (b) cross-sectional SEM image of the precursor film (x40,000).












CuInhISe2 (112)


CuSe (006)


_1 6 ^27
-r~ ') o(0


0 2.4


Figure 3.2 Time-resolved in-situ X-ray Diffraction for isothermal heating at 218 C. The
last 20 data were collected while heating at 340 C for 30 minutes.


32 33




























Time (sec)


Figure 3.3 Fractional reaction of CuInSe2
function of time and temperature.


formation during the isothermal runs as a












1.0-

0.8-
U

0.6-
*\


% --% 006I
\\, **^. ^*
0.2 -0 A.. S ,
-" A AAAAA-A-
AA^AAAA^
0.0- vV I --

-500 0 500 1000 1500 2000
Time (sec)


-- 218 C
--- 236 C
-A- 253 C
-v- 371 C


2I 300 3500
2500 3000 3500


Figure 3.4 Fractional reaction of CuSe transformation during the isothermal runs as a
function of time and temperature.












1.0
TV~o- _
,-- --/A,.. o/ m*i-,. ,
0.8- A y ""


o 0.6-
U_
A0.4
U0-
04- --- 218C

--- 236 C
0.2 -- 253 C

--- 271 C
0.0
-500 0 500 1000 1500 2000 2500 3000 3500
Time (sec)


Figure 3.5 Sum of the mole fractions of the reactant (CuSe) and the product (CuInSe2)
through isothermal heating.







48




2.0

1.5

1.0

0.5-

-0.0 o

-0.5-
C -1.0o 218C
236 C
-1.5
A 253 C
-2.0
v 271 C

-2.5 --
5 6 7 8
In t


Figure 3.6 Avrami plots for the isothermal reactions at different temperatures.















-6.5-



-7.0-



-7.5-



-8.0-



-8.5-


0.00180


I.185 0.00190 0.00195 I0.00200 .20 I 00I 210
0.00185 0.00190 0.00195 0.00200 0.00205 0.00210


1I/T(K1)


Figure 3.7 Arrhenius plot for the isothermal reactions based on the Avrami analysis:
apparent activation energy for CuInSe2 growth reaction, Ea = 66.0 kJ/mol.





















t -1.5 -

218 C
-2.0 236C
236 C

A 253 C
-2.5-
v 271 C

-3.0
5 6 7 8
In t



Figure 3.8 Plots based on the parabolic rate law for the isothermal reactions at different
temperatures.






















C
-8.5




-9.0


0.00180 0.00185 0.00190 0.00195 0.00200 0.00205

1 /IT (K"1)



Figure 3.9 Arrhenius plot for the isothermal reactions based on the parabolic rate law:
apparent activation energy for CuInSe2 growth reaction, Ea = 65.2 kJ/mol.
















CuInSe2 (112)


CuSe (108)


CuInSe2 (220)/(204)/

f/CuInSe2 (312)/(116)


so


Figure 3.10 Time-resolved in-situ X-ray Diffraction with increasing temperature:
transformation from a-InSe/CuSe to CuInSe2.


CuSe(006)
\ A. .iA ,i


25 -


35


40
20(0) 4s














CuInSe2 (112)


CuSe (006)


Cu7Se4 (321)
/ -Cu7Se4 (320)


25


Figure 3.11 Time-resolved in-situ X-ray Diffraction with increasing temperature:
transformation from a-InSe/Cu:Se to CuInSe2.


30"''" g














CHAPTER 4
SECONDARY GRAIN GROWTH OF CUNSE2 IN THE PRESENCE OF CUxSE
DRIVEN BY INTERFACIAL FORCES

4.1 Introduction

It is well known that the secondary phase, copper selenide, plays an important role in

grain growth of CIS-based materials (Rockett 1994, Klenk 1993). The existence of

copper selenide during CIS film growth is connected to increases in the grain size and

improvements in the electrical properties of the film. Some researchers hypothesized that

the secondary phase exists as a liquid-like phase during the process and enhance transport

of the reactants from vapor phase into the surface of growing CIS phase (Klenk 1993,

Tuttle 1995). This is often called vapor-liquid-solid (V-L-S) growth model, and it

emphasizes the kinetic factor during film growth. The kinetic growth models seem to

explain acceptably the dramatic grain size change related with composition range,

however, there has not been any direct evidence that the kinetic factor is the only driving

force for grain growth. Furthermore, the kinetic growth models are valid only when there

is supply of reactants from vapor phase.

For other thin film systems, it is well established that the driving force for grain

growth is often the reduction of the interfacial free energy that accompanies reduction in

total grain boundary area (Atkinson 1988, Hillert 1965, Thompson 1985, Srolovitz 1983).

It is hence quite plausible to assume that the grain growth mechanism in the CIS-based

material is also related to a reduction of interfacial free energy. In particular, the

dramatic increase of the grain size in the presence of the copper selenide phase during









growth shows some similarities to the so called, "secondary grain growth" or "abnormal

grain growth" phenomenon for other film growth processes (Wong 1986, Palmer 1987,

Thompson 1984, Thompson 1988). In this mechanism, the secondary grain growth is

induced by the selectivity in driving force provided by surface-energy anisotropy. That

is, to minimize the total energy of the system, the grains with the orientations that have

minimum surface energy grow and become prevalent by consuming the grains with other

orientations that have higher surface energy (Cahn 1970, Thompson 1985).

If the copper selenide phase enhances the surface-energy anisotropy of the CuInSe2

phase, in other words, if the copper selenide appreciably lowers the interface free energy

of a specific orientation, it will lead to large grains with a specific crystallographic

texture. The other feature of the secondary grain growth, such as reduction of some

orientation phase and development of bimodal grain size distribution, will be observed.

In this chapter, it is reported that existence of the copper selenide phase dramatically

increases the average grain size of the CuInSe2 film during an annealing process, and it is

suggested that a thermodynamic driving force, surface-energy anisotropy drives the

process.

4.2 Kinetic Aspect of Growth Mechanism of CuInSe2-based Thin Films

4.2.1 Vapor-Liquid-Solid Growth Model for CIS-based Thin Films

The quality of CIS-based thin films strongly depends on the process parameters,

especially the metal atomic ratio, [Cu]/[In], and the growth temperature. Many research

groups have incorporated the Cu-rich growth steps to produce larger grain size. It is well

known that the surface of even slightly Cu-rich CIS thin film is covered with CuxSe

phase (Rockett 1994, Niemi 1990, Tuttle 1991). Many researchers suggested that a

copper selenide phase, CuxSe, exists as liquid phase during a high temperature growth









process, and it enhances the grain growth due to its superior transport mechanism

(Birkmire 2001, Nadenau 1995). There has not been, however, any direct proof for the

existence of the liquid phase during the growth process.

Droplet structures have been reproducibly observed (Figure 4.1) on the surface of as-

grown Cu-rich CIS thin films when certain conditions are fulfilled, namely Cu-rich

composition during growth stage and sufficient Se flux during the growth and cooling

stages. Other features of the as-grown samples and the process strongly suggest that the

droplet structure is closely related to the CuxSe liquid phase. The droplet structure seems

to be significantly affected by the substrate type, temperature, final thickness of the film

as well as the composition. This study should help to better understand the fundamental

growth mechanism of CIS-based thin films.

It is well known that growing thin films with different stoichiometries ([Cu]/[In] >1

and <1) grow by entirely different mechanisms and result in distinguished crystalline

structure from each other. Klenk et al. (Klenk 1993) proposed a model for the

characteristic growth mechanism, 'vapor-liquid-solid model'. It explains the role of the

secondary copper-chalcogenide phases in CIS-based thin-film growth, and several

researchers refined the model (Tuttle 1991, Tuttle 1995).

It has been shown that some properties of Cu-III-VI2 materials undergo a sudden

change as soon as even a small amount of excess copper ([Cu]/[III] >1) is present in the

film. Early in the development of CIS photovoltaic device, the so-called "Boeing recipe"

was employed to grow CIS thin films with large grain size. The process was first

growing Cu-rich thin film and turning it into an overall Cu-poor thin film by reducing the

[Cu]/[In] during latter part of the process. The enhanced grain size was attributed to the









superior grain growth mechanism of the bottom layer, Cu-rich CIS. As-grown Cu-rich

thin films showed much larger grain size (2 to 10 .tm) than that of as-grown Cu-poor

samples (<1 tm) (Tuttle 1994, Klenk 1993). The shape of the grains also changed

depending on the composition ratio. The grain structure of Cu-rich samples was more

round-shaped while that of Cu-poor samples is very facetted. This anisotropy, attributed

to limitations in chalcogen supply, is not observed in Cu-rich films. It suggests that the

Cu-rich and the In-rich films go through entirely different growth mechanism. Many

researchers believe that the growth mechanism of the CIS thin films is strongly affected

by the presence of the secondary phase, CuxSe.

The primary assumptions of the V-L-S model are: (1) the presence of liquid phase on

the surface during growth, and (2) the grain growth is enhanced by the liquid phase to

give superior transport. According to the phase diagram of the Cu-In-Se ternary system,

the existence of a liquid phase of CuxSe is possible and very likely at high temperature

(>530 C) (Nadenau 1995), although it is realized that the growth process is not at

equilibrium. This reasoning coincides with several experimental results that display high

crystalline structure and large grain size above this high temperature (Tuttle 1995, Stolt

1993). The absorber layers of the best efficiency solar cells have been grown at these

conditions.

It is also well known that CuxSe phase exists primarily at grain boundaries and

surface; hence, it is plausible that vapor, liquid, and solid phases coexist, and the ternary

chalcopyrite grains grow by way of the vapor-liquid-solid growth mechanism. Figure 4.2

depicts schematic structure of a growing Cu-rich CIS thin film in view of the V-L-S

growth model. The solid bottom layer phase is a growing CuInSe2. The vapor species









impinge and condense on the surface of the binary liquid phase. They are then

transported to the liquid/solid interface and incorporated in the film. The excess Cu and

Se remain in the liquid phase, so the solid ternary phase will be nearly stoichiometric

CuInSe2. If the vapor is deficient in Cu, then the Cu-Se phase will be consumed to form

the solid ternary phase, and its amount is reduced. If the Cu deficiency in the vapor

phase is maintained, the liquid CuxSe phase will finally disappear, and an In-rich surface

is formed (Ramanathan 1993). It is believed that the key role of the liquid phase is to

enhance transport of the adsorbed species into the growing surfaces.

Tuttle et al. (Tuttle 1995) proposed that a growing CIS thin film goes through two

different phase segregations in sequence depending on the thickness of the film.

According to their model, phase separation between CuxSe(l) and CuInSe2(s) phases

esists lateral to the growth plane during initial stage of the growth process. The liquid

secondary phase enhances mobility of the adatom on surface and mass transport of the

adsorbed species, which result in enhanced grain growth. As the CuInSe2 solid coalesces,

the surface energy between the liquid and the solid phases increases and reaches a critical

value, at which time the phase separation converts to an orientation normal to the growth

plane. This mechanism reflects the result and reasoning of Adams (Adams 1992) applied

for phase separation of Al-Ge thin films during co-deposition. The result clearly shows

that Al-Ge phase separation was lateral and transformed to transverse as the film grows.

This phenomena can be explained in view of relative thermodynamic stability of the co-

existing phases in the thin film structure. Tuttle et al. also proposed that the presence of a

liquid phase is possible even at low temperature condition (< melting point of CuSe;

523C). It has been observed in other thin film material systems (Givargizov 1991). The









thermodynamic driving forces and rates on surface of thin films are quite different from

those of bulk material. As a rule of thumb, thin films can be condensed in the liquid state

above temperature of approximately 2/3 Tm (the melting point of the bulk material)

(Givargizov 1991). These results support the hypothesis that the presence of liquid phase

at the surface region is quite plausible during CIS thin film growth at even less than

523C. Another research group also proposed the possibility of the liquid phase at low

temperature condition (Nadenau 1995).

Although experimental results and theoretical evidence exists for the presence of a

secondary liquid phase during growth process, there has not been any direct proof that the

liquid phase actually exists and enhances grain growth mechanism. Recently, a unique

film surface morphology was observed in the CIS films deposited by MEE. It may

provide a more direct evidence for the presence of liquid phase during film growth. The

result, however, are likely to be attributed to the unique feature of MEE growth system.

4.2.2 Experimental Evidence of the Liquid Phase during Growth Process

In this work, a unique experimental result was obtained. It may be related to the

vapor-liquid-solid growth mechanism. To the best of our knowledge, no such result has

been reported to date. It may be a strong and direct evidence of the V-L-S growth model

according to what the analysis results indicate.

Cu-rich CIS thin films were grown at various temperature and elemental flux ratios. A

reproducible surface morphology is obtained when both conditions of Cu-rich metal flux

and sufficient selenium flux are supplied to the growth of CIS films. Droplet-shaped

islands formed on the surface of Cu-rich CIS thin films (Figure 4.2). According to the

AES and EDS composition measurements, the islands (droplet region) have different

composition than the rest of the region (matrix region). The composition of the droplet









region is much more Cu-rich than that of the plane region. When the overall composition

of the film is slightly Cu-rich, it seems that only the droplet region has Cu-rich

composition while the matrix region is slightly Cu-poor or almost stoichiometric (Cu/In

<1). This suggests that there is a phase segregation lateral to the growth plane during the

growth process.

Careful observation by SEM revealed that the grain size of the droplet region is larger

than that of the matrix region. It suggests that the superior grain growth mechanism

occurred in the droplet region. XRD analysis shows that the thin films with the droplet

structures have Cu-rich secondary phase, CuxSe. It is very likely that only the droplet

region has this secondary CuxSe phase since the composition of that region is Cu-rich and

the composition of the plane region is slightly Cu-poor. The droplet structure remains

even after etching the secondary phase with KCN solution, which means that the larger

grains are CuInSe2 as well as the Cu-rich secondary phase.

It is our hypothesis that the droplet structure is caused by the presence of a liquid

phase and its superior grain growth mechanism. The shape of the structure is very similar

to that of liquid droplets (Figure 4.2). The Cu-rich composition in the droplet region

indicates that the region was covered by a low melting temperature Cu-rich secondary

phase. The difference of the grain size between the droplet region and the matrix region

becomes larger when an annealing step is introduced at the end of the deposition

probably because the annealing step enhances surface migration (Figure 4.3).

The droplet structure did not appear when the selenium composition (the

[Se]/([Cu]+[In]+[Se]) ratio) was below 48%, even though the film is very Cu-rich. Cu-

Se phase such as Cu2Se has a very high melting temperature compared to the selenium-









rich eutectic at, 523C. It is thus possible that a deficiency in selenium prevented

formation of the low melting eutectic liquid. The shape of the droplet increasingly

resembles that of a liquid droplet as the growth temperature is increased or annealing

process is introduced (Figure 4.4).

For the hypothesis described above, however, there are still some questions that need

to be answered. Firstly, the growth temperature (40030C) seems to be too low to

sustain a liquid phase. The answer may come from the example of a different material

system. Givargizov introduced a result that is somewhat commonly observed in thin film

structure (Givargizov 1991). The thermodynamics and kinetics on the surface can be

quite different from that of bulk state. It is possible that some species is condensed as a

liquid or semi-liquid phase above approximately 2/3 of the melting temperature (K). If it

can be applied to CIS thin films, 2/3 of the CuSe melting point, 236C, is lower than the

growth temperature.

Another important question is why the specific result is only found when deposition

occurs by MEE system. To answer this question, the unique features of the MEE system

need to be understood.

The differences from typical co-deposition systems are discussed in the system

description section (Chapter 1). The MEE system adopts a rotating platen that

sequentially delivers the metal and the chalcogen elements, therefore, each source

impinges on the substrate in an intermittent and periodic fashion. There is also certain

time period (load lock zone) that the substrate is not exposed to a flux. The system was

designed to enhance the surface migration of species so that a lower growth temperature

or potentially better material quality could be achieved.









A liquid in contact with a solid will either completely wet the surface or form a

spherical droplet depending on the surface free energies, YSL, ysv, and YLV. It has been

observed that the appearance of the droplet surface morphology depends on the substrate

used. The structure clearly formed on Mo-coated substrates while it did not appear on

bare soda lime glass substrates. A hint about what is occurring can be found from the

experiment of Adams on phase segregation during Al-Ge deposition (Adams 1992). The

phase segregation is lateral during initial stages of growth, and then becomes transverse

as the film continues to grow. It is argued in this work that the transformation occurs to

make the system thermodynamically more stable.

4.3 Experiments

The migration enhanced epitaxy (MEE) growth system was used to grow CIS films,

and the equipment is discussed in Chapter 1.

Approximately 0.6 jLm of CuInSe2 was grown on molybdenum-coated soda lime

glasses in the MEE system. The substrate temperature was 450 C. The films were

grown very slightly Cu-rich ([Cu]/[In] = 1.02). Thin layer of CuSe (-04).1 aim) was then

deposited on the as-grown CIS film at a low temperature (-150 C) in the MEE system.

This temperature is sufficiently low to prevent reaction between CuSe layer and the

underlying CIS layer. The final sample structure before the annealing step is

glass/Mo/CIS/CuSe as depicted in Figure 4.5. The films with and without the CuSe layer

were annealed in the MEE system under the conditions given in Table 4.1. The

composition of the films was measured using ICP. The SEM analysis was performed

with a JEOL JSM-6335F field emission scanning electron microscope. XRD data were

analyzed using the software JADE, a program for powder diffraction data analysis.









4.4 Results and Discussion

Composition of the as-grown CuInSe2 films was intended to be very slightly copper

rich ([Cu]/[In] = 1.02) to prevent Cu incorporation into the CuInSe2 phase leading to

additional formation of CuInSe2 during the annealing step. The phases that exist in CIS

strongly depend on the film composition (Klenk 1993, Tuttle 1995, Rockett 1994). When

the composition is slightly indium-rich, the phase diagram indicates that the material is

single phase, CuInSe2. The two-phase CuxSe-CuInSe2 boundary lies very close to the

[Cu]/[In] =1 composition. Addition of Cu or other element will form CIS within the limit

of the single phase region. Two phases, CuInSe2 and CuxSe, co-exist when the film

composition is stoichiometric or Cu-rich. It is important to know that the excess copper

is not incorporated into the CuInSe2 phase but consumed to form the secondary phase,

copper selenide.

The CuxSe phase in the film can be selectively etched by a potassium cyanide (KCN)

solution. When the as-grown CuInSe2 films were treated with 10 % KCN aqueous

solution, the film composition changed from slightly Cu-rich ([Cu]/[In]-l1.02) to slightly

indium-rich ([Cu]/[In]~0.99: Table 4.2). X-ray diffraction spectra show small peaks of

the CuSe phase before the etching, and those peaks disappear after the KCN treatment.

The ICP and XRD results indicate that the as grown CIS samples have a small amount of

the CuxSe phase.

The sample groups with and without the additional copper selenide layer were

annealed at the same time in the MEE system. The exact annealing conditions are

described in Table 4.1. Selenium flux with source maintained at 250 C was maintained

to prevent selenium loss by evaporation from the samples, which keeps the composition









of both CIS and CuSe constant throughout the annealing process. After the annealing

process, both the samples with and without the CuSe layer were treated by the KCN

solution to remove the CuxSe phase and the composition of the sample was determined

by ICP. The results shows that the film composition after annealing and the KCN

treatment (sample group C-KCN) is very close to that of the KCN-treated CIS samples

that was not annealed (sample group A-KCN). This result indicates that there was not

incorporation of any element into the CuInSe2 phase during the annealing step. As

mentioned earlier, it is important to know that the excess copper is not incorporated into

the CuInSe2 (i.e., this phase is saturated with reagent of Cu) but remains in a secondary

phase (CuxSe). Overpressure of the selenium was maintained during the CuInSe2 film

growth and for selected samples during the annealing process, hence; it can be reasonably

assumed that there was not significant addition or loss of selenium during the process.

The compositional analyses and the phase diagram strongly suggest that there is

negligible or no CIS phase formation through incorporation of new elements during the

annealing process. This minimizes the possibility that the grain growth occurred by the

mass transport between the CIS and the CuxSe phases (i.e., a kinetic effect) during the

annealing step.

Figures 4.5 and 4.7 display SEM images of the samples that were annealed with (b)

and without (a) the CuSe layer, and with a Se overpressure during annealing. These

images clearly show that grain growth results from the annealing in the presence of CuSe

layer, compared to annealing process without the CuSe layer. Some of the grains in the

samples annealed with the CuSe layer (362-3K and 363-3K) are larger than the film

thickness.









Full width at half maximum (FWHM) of the (112) x-ray diffraction peak was

measured to better quantify crystallinity and grain size of the samples comparing samples

of group C (annealed in the presence of CuSe layer) and group D (annealed without CuSe

layer). Table 4.2 shows that the FWHM values decreased by ~ 30% by annealing in the

presence of CuSe layer. The samples annealed with the CuSe (sample group C) show

remarkably greater peak intensity and smaller FWHM values of(112) peak compared to

the sample groups that did not have the CuSe layer during annealing (Table 4.2 and

Figure 4.8). These results strongly suggest that existence of the CuSe layer during the

annealing process enhanced the (112) oriented grain growth. The XRD result is

consistent with the SEM images (Figures 4.6 and 4.7) that show an increase in the grain

size.

The samples that did not have the CuSe layer during annealing also shown an increase

of the (112) peak area, but much less compared to the samples with the CuSe layer

(Figure 4.8). It is likely because those samples without the CuSe layer had small amount

of CuSe phase due to the slightly Cu-rich composition of the as-grown CuInSe2 films.

The small amount CuSe phase is likely to cover the film surface in part and exist at the

grain boundaries. This hypothesis specifically explains the difference between samples

362 and samples 363. Figure 4.8 clearly shows that 363-4K acquired more increase of

the (112) peak area compared to 362-4K. Those two samples did not have the CuSe layer

during the annealing process. The only difference is that the samples 363's had

overpressure of selenium during the annealing process while the samples 362's did not.

It is likely that absence of the selenium supply during the annealing process caused

selenium evaporation from the CuSe phase of the samples 362's and resulted in phase









transition from CuSe to the selenium poor CuxSe phase. It could reduce the effect of the

CuSe phase in the sample 362-4K. The difference in the peak area between the sample

362-3K and the 363-3K is negligible. Both samples had a CuSe layer during annealing.

Figure 4.9 displays the (220)/(204) reflections of CIS samples that were annealed with

and without the CuSe layer. Interestingly, it is observed that peak area of the (220)/(204)

orientated grains decreased more when the CuSe layer was deposited prior to annealing

(Table 4.2 and Figure 4.9). This indicates that large portion of the (220)/(204) oriented

grains were consumed to contribute to (112) oriented grain growth. This tendency is

clearly demonstrated in Figure 4.10. From the shape of the curves, extent of the increase

of the (112) peak area is consistent with the decrease of the (204)/(220) peak area.

Specific peak area is proportional to the amount of the corresponding crystalline phase

assuming other effects (e.g., stress or strain effects) are the same. This is a strong

evidence for secondary grain growth. It is also observed that the FWHM decreases after

annealing so that (204) and (220) peaks are split in the samples 362-3K and the 363-3K.

Those two peaks could not be clearly distinguished before annealing since they closely

overlapped and the crystallinity was poor.

As pointed out earlier, secondary grain growth can be induced by a selectivity in

driving force provided through a surface-energy anisotropy. To minimize the total free

energy of the system, grains with orientations that minimize the surface energy grow to

consume the grains with other orientations (Cahn 1970, Thompson 1985). The surface-

energy anisotropy can play a dominant role in determining which grains become

secondary grains in thin films. As a result, a specific orientation may grow and









sometimes become dominant while other orientations diminish or disappear. Existence

of secondary phase and defect can affect the status of the surface energy anisotropy.

In the chalcopyrite structure material systems such as CuInSe2, the (112) plane is

likely to have the lowest surface energy for free surfaces (Zhang 2001, Liao 2002). As a

result, XRD spectra of CIS films often show that the CIS films are textured along the

(112) plane parallel to the substrate. The second strongest peak is usually the

(204)/(220). It is possible that presence of the CuxSe increases the surface-energy

anisotropy by selectively lowering the surface free energy of the (112) orientation phase.

From the XRD results, all the samples annealed with the CuSe layer showed an

increase of the (112) orientated phase and a decrease of the (220)/(204) phase compared

to the as-grown CuInSe2 films. The extent of the increase of the (112) peak area is

consistent with that of a decrease in the (220)/(204) peak area. The SEM images show

that annealing with the CuSe layer increased the grain size of the CIS films. These

results are consistent with a secondary grain growth mechanism assisted by the CuSe

phase.

It is proposed that the CuSe phase enhances the surface-energy anisotropy of the

CuInSe2 films and consequently induces the secondary grain growth of the (112)

orientated phase. This is likely to occur during many other growth processes of the

CuInSe2-based materials when the overall composition is Cu-rich. The thermodynamic

factor (the surface-energy anisotropy) as well as the kinetic factor (mobility

enhancement) is believed to be primary driving forces for the enhanced grain growth to

result in the large grain size of CuInSe2 phase.









4.5 Conclusions

The effect of a secondary CuSe phase on the grain growth of CuInSe2 films was

investigated. CuInSe2 films were deposited onto molybdenum-coated soda lime glass

using a molecular beam epitaxy system. The composition of the as-grown CuInSe2 films

were intended to be stoichiometric or slightly Cu-rich. On selected films, thin layer of

the CuSe was deposited on the as-grown CuInSe2 films. These bi-layer samples,

CuInSe2/CuSe, were then annealed at two different temperatures and with and without Se

overpressure. The grain size was observed to increase (up to a few g~m) in samples

annealed with the CuSe layer as compared to samples annealed without the CuSe layer.

After annealing the CuInSe2/CuSe films, the peak area of the (112) reflections increased

while that of the (220)/(204) reflections decreased, consistent with (112) oriented grain

growth at the expense of the (220)/(204) grains. It is proposed that there is a

thermodynamic driving force for the grain size increase, known as "secondary grain

growth", assisted by the surface-energy anisotropy.













Table 4.1 Summary annealing conditions


Annealing conditions
Sample Sample Structure before KCN R
,.------- i ------ _Remarks
group number annealing Se treatment
Temperature
overpressure
As grown
A 359-8 CuInSe2 No annealing No CIS, No
________ annealing
No
A-KCN 359-8KO CulnSe2 No annealing Yes annealing,
A-KCNKCN
treatment
Deposition
B 360-2 CulnSe2/CuSe No annealing No of CuSe
layer
KCN
B-KCN 360-2K CulnSe2 No annealing Yes treatment of
_________ group B
Annealing
C-KCN 362-3K CuInSe2/CuSe -450 C No Yes with CuSe
____ _________ _______ ____________ layer
Control
D-KCN 362-4K CulnSe2 -450 C No Yes Control
sample I
Annealing
C-KCN 363-3K CulnSe2/CuSe -450 C Yes Yes with CuSe
___________layer
Control
D-KCN 363-4K CuInSe2 -450 C Yes Yes Control
sample II
Annealing
C-KCN 364-3K CulnSe2/CuSe -370 C Yes Yes with CuSe
___________ ___________________________ layer

a: K indicates that the sample was treated with KCN to remove CuxSe phase.












Table 4.2 Composition (ICP) and X-ray diffraction analysis results


Atom (%)
Sample group Sample %)FWHM Peak area of Peak area of
number Cu In Se Cu/In of(112) (112) (204)/(220)

A 359-8 25.7 25.1 49.2 1.02 -

A-KCN 359-8K 25.2 25.5 49.3 0.99 0.134 3.40x105 1.36x105

B 360-2 28.3 22.5 49.2 1.25 -

B-KCN 360-2K 25.2 25.4 49.4 0.99 0.130 3.53xl05 1.40x 105

C-KCN 362-3K 25.3 25.3 49.4 1.00 0.085 1.01x106 4.65x104

D-KCN 362-4K 25.3 25.4 49.3 1.00 0.126 4.63x 105 1.24x 105

C-KCN 363-3K 25.3 25.5 49.2 0.99 0.084 9.10x105 5.77x104

D-KCN 363-4K 25.4 25.5 49.1 1.00 0.115 6.80x105 9.89x104

C-KCN 364-3K 25.2 25.6 49.2 0.98 0.089 1.06x106 4.91x 104


























Figure 4.1 Droplet structures on an as-grown Cu-rich CuInSe2 film: T = 350 C,
[Cu]/[In] = 1.03.



Cu, In, Se (Vapor)




CuSe (Liquid



CuInSe2 (Solid)

Mo

Soda lime glass


Figure 4.2 Schematic structure of a growing Cu-rich CIS thin film based on the vapor-
liquid-solid model.























Figure 4.3 SEM images of different regions on the surface showing an apparent
difference in grain size in (a) matrix region and (b) droplet region.


Figure 4.4 Droplet structures on an as-grown Cu-rich CuInSe2 film with annealing: T =
450 C, [Cu]/[In] =1.03.


pil









CuSe


CuSe layer


Figure 4.5 Schematics of sample structures: (a) without the CuSe layer; control sample,
(b) with the CuSe layer.










































(a) Sample 362-4K


Figure 4.6 Cross-sectional SEM images of CuInSe2 films: (a) annealed without the
copper selenide layer on the top of the CuInSe2 layer.


:..- .:: Ki












4


(b) Sample 362-3K


Figure 4.6 Cross-sectional SEM images of the CuInSe2 films: (b) annealed with the
copper selenide layer. Both samples were treated with KCN solution to selectively
remove the copper selenide layer after annealing.



































(a) Sample 363-4K


Figure 4.7 Cross-sectional SEM images of CuInSe2 films: (a) annealed without the
copper selenide layer on the top of the CuInSe2 layer.



































(b) Sample 363-3K


Figure 4.7 Cross-sectional SEM images of the CuInSe2 films: (b) annealed with the
copper selenide layer. Both samples were treated with KCN solution to selectively
remove the copper selenide layer after annealing.











60000 39
------ 359-8K

50000- 362-4K
\ '362-3K

40000


30000
C
.C 20000

10000 -,




26.0 26.5 27.0

20 (o)


(a)

Figure 4.8 CIS (112) diffraction peaks: (a) annealed at approximately 450 C without
selenium overpressure. The sample 359-8K was not annealed. The sample 362-3K and
363-3K were annealed with the copper selenide layer, while sample 362-4K and 363-4K
were annealed without the copper selenide layer. All the samples were treated by KCN
to selectively remove the copper selenide after annealing.











60000 39
------ 359-8K

50000- 363-4K
-363-3K

40000-


.> 30000


20000-

10000 '"

0-

26.0 26.5 27.0

20 (0)


(b)

Figure 4.8 CIS (112) diffraction peaks: (b) same temperature condition, but with
selenium overpressure. The sample 359-8K was not annealed. The sample 362-3K and
363-3K were annealed with the copper selenide layer, while sample 362-4K and 363-4K
were annealed without the copper selenide layer. All the samples were treated by KCN
to selectively remove the copper selenide after annealing.






80



3000- ----- 359-8K
A,'. 362-4K
i' i 623
2500- 362-3K

2000
>, /
1500-
--0
1500000,-'"





43.5 44.0 44.5 45.0
20 (o)

(a)
Figure 4.9 CIS (204)/(220) diffraction peaks: (a) annealed at approximately 450 C
without selenium overpressure. The sample 359-8K was not annealed. The sample 362-
3K and 363-3K were annealed with the copper selenide layer, while sample 362-4K and
363-4K were annealed without the copper selenide layer. All the samples were treated by
KCN to selectively remove the copper selenide after annealing.













------ 359-8K
3000 363-4K
II I
2500 363-3K
I.
I 8I1
2000 -


2 1500-
(D)
C
I



1000


500



43.5 44.0 44.5 45.0

20()


(b)

Figure 4.9 CIS (204)/(220) diffraction peaks: (b) same temperature condition, but with
selenium overpressure. The sample 359-8K was not annealed. The sample 362-3K and
363-3K were annealed with the copper selenide layer, while sample 362-4K and 363-4K
were annealed without the copper selenide layer. All the samples were treated by KCN
to selectively remove the copper selenide after annealing.







82




-u-(112)
1100000-
1000000 --A- (204)/(220) 140000
1000000- 1400(X

900000- 120000

800000-
100000
(I) 700000-

( 600000- 80000

500000- -
60000
400000 -^

300000 40000

359-8K 362-4K 362-3K
Sample Number


(a)

Figure 4.10 XRD peak areas: (a) annealed at approximately 450 C without selenium
overpressure. The sample 359-8K was not annealed. The sample 362-3K and 363-3K
were annealed with the copper selenide layer, while sample 362-4K and 363-4K were
annealed without the copper selenide layer. All the samples were treated by KCN to
selectively remove the copper selenide after annealing.














1000000-

900000-

800000-

700000-

600000-

500000-

400000-

300000-


-- (112)

-A- (204)/(220)


359-8K


363-4K


140000


120000


100000


80000


60000


363-3K


Sample Number


(b)

Figure 4.10 XRD peak areas: (b) same temperature condition, but with selenium
overpressure. The sample 359-8K was not annealed. The sample 362-3K and 363-3K
were annealed with the copper selenide layer, while sample 362-4K and 363-4K were
annealed without the copper selenide layer. All the samples were treated by KCN to
selectively remove the copper selenide after annealing.














CHAPTER 5
EPITAXIAL GROWTH OF CUINSE2 AND CUGASE2 FILMS ON SINGLE
CRYSTALLINE GAAS SUBSTRATE USING A MIGRATION ENHANCED
EPITAXY REACTOR

5.1 Introduction

This chapter presents preliminary results on the epitaxial growth of CuInSe2 (CIS) and

CuGaSe2 (CGS) films. While there have been several studies of the growth mechanism

of CuInSe2 films, it is still not fully understood in part due to complex phase evolution

and defect chemistry in that material system. The Cu(In,Ga)Se2 (CIGS)-based solar cells

are known to produce high efficiency when grown by a number of techniques (Devaney

1990, Bodegard 1995, Cahen 1989). The mechanism responsible for the improvements,

however, are not fully understood. It has been difficult to produce samples in a

completely reproducible way. To accurately identify the effects of process parameters

and defect chemistry on the device performance, one must first grow high quality films

reproducibly. For example, the inconsistent effect from grain boundaries needs to be

eliminated or at least minimized to investigate the relationship between the electrical

properties and the process parameters. The best approach of preparing such high quality

CIS and CGS fims is to grow the films epitaxially on a single crystalline substrate, such

as a GaAs wafer.

In addition to the above reasons, the need for single crystalline CIGS-based solar cells

are growing due to its potential for a tandem solar cell structure. The concept of the

tandem solar cell has been explored during the last decade. The structure consists of

double junctions that have different energy band gaps. The single crystalline CGS film is