A Dissertation
entitled
Optical and Microstructural Properties of Sputtered Thin Films for Photovoltaic
Applications
by
Dipendra Adhikari
Submitted to the Graduate Faculty as partial fulfillment of the requirements for the
Doctor of Philosophy Degree in Physics
______Dr. Nikolas J. Podraza, Committee Chair
______Dr. Robert W. Collins, Committee Member
______Dr. Yanfa Yan, Committee Member
______Dr. Michael Cushing, Committee Member
______Dr. Sylvain X. Marsillac, Committee Member
______John Plenefisch, PhD, Dean College of Graduate Studies
The University of Toledo
December 2019
Copyright 2019 Dipendra Adhikari
This document is copyrighted material. Under copyright law, no parts of this document may be reproduced without the expressed permission of the author.
An Abstract of
Optical and Microstructural Properties of Sputtered Thin Films for Photovoltaic Applications
by
Dipendra Adhikari
Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Doctor of Philosophy Degree in Physics
The University of Toledo December 2019
Thin film solar cells are promising candidates for generation of low cost and pollution-free energy. The materials used in these devices, mainly the active absorber layer, can be deposited in a variety of industry-friendly ways, so that the cost associated with manufacturing is generally lower than for competing technologies such as crystalline silicon. This dissertation will focus on the fabrication and characterization of nanocrystalline hydrogenated silicon (nc-Si:H) and polycrystalline cadmium telluride
(CdTe) thin films by industrially scalable, non-toxic, and comparatively simple magnetron sputtering. The performance of the solar cells incorporating these films as an active absorber layers are discussed. In this work, spectroscopic ellipsometry is used as the primary tool for the characterization of optical and structural properties of thin films and bulk material.
As a first case study, the anisotropic optical properties of single crystal strontium lanthanum aluminum oxide (SrLaAlO4) in the form of birefringence and dichroism is obtained from Mueller matrix ellipsometry. SrLaAlO4 exhibit uniaxial anisotropic optical properties and the indirect optical band gap of 2.74 eV. A parametric model consisting of
iii parabolic band critical points (CPs) for electronic transitions and a gap function is used to describe the complex dielectric function spectra in both the ordinary and extra-ordinary directions. The modeling in this case study has applications to both nc-Si:H, an indirect band gap semiconductor, and CdTe which may exhibit microstructural anisotropy depending upon the deposition method.
Fabrication and characterization of hydrogenated silicon (Si:H) thin films produced by reactive magnetron sputtering is the second case in this study. RTSE and a virtual interface analysis (VIA) are used to track the growth evolution of sputtered Si:H. From these studies, growth evolution diagrams depicting the nucleation of nanocrystallites from the amorphous phase and nanocrystallite coalescence are developed. Silicon-hydrogen bonding configurations are determined from absorption features extracted from infrared spectroscopic ellipsometry. Ultimately, this work provides a summary of how nanostructure can be manipulated in reactive magnetron sputtered Si:H films, which is necessary for implementation of these materials in photovoltaic (PV) devices.
Here we have demonstrated working PV devices produced with RF magnetron sputtered nc-Si:H absorber layers and compare overall device performances to those produced with the more conventional plasma enhanced chemical vapor deposition
(PECVD) absorbers. Different absorber layer deposition techniques and atmospheric exposure effects are studied to explain variations in the performance of single junction n- i-p substrate configuration nc-Si:H based solar cells. The cells with nanocrystalline
PECVD absorbers and an untextured (planar) back reflector serve as a baseline for comparison with power conversion efficiency near 6%. This efficiency is typical of this device configuration lacking optical enhancement due to scattering of light incident by the
iv back reflector. By comparison, cells with sputtered absorbers achieved efficiencies of about
1%. Comparison of dark / light current-voltage measurements and external quantum efficiency (EQE) indicate that lower performance in devices with sputtered absorbers may be attributed to both low electronic quality within the nc-Si:H absorber and also process incompatibility at the interfaces between the intrinsic layer made by sputtering with the doped layers made by PECVD. The collection probability profile obtained from EQE simulations show substantial carrier concentration losses are identified at the n-/i-interface.
The sputtered material itself requires further optimization to reach performance levels comparable to those achievable with PECVD, but this works serves as a baseline for future material and device studies.
Finally, the optical, electrical, and microstructural properties of magnetron sputtered CdTe films prepared by glancing angle deposition (GLAD) are studied. From cross-sectional micrographs, increasingly tilted columnar structure occurs with increasing incident oblique angle for as-deposited CdTe films. Films deposited at lower oblique angles closer to normal incidence have mixed (cubic + hexagonal) crystal structure and those prepared at more oblique higher angles have hexagonal wurtzite crystal structure. Post- deposition CdCl2 treated films show enhancement in grain size for samples prepared under all conditions. The optical response in the form of the complex dielectric function (ε = ε1 + iε2) spectra from 0.74 to 5.89 eV for the GLAD thin films are all qualitatively similar to single crystal CdTe. Higher angle deposited samples show columnar structure induced anisotropy in spectra in ε in the transparent spectral range. Application of GLAD CdTe interlayers between CdS and CdTe of the standard CdS/CdTe heterojunction design solar cell shows better performance with 0.9% absolute efficiency increase.
v
To my Parents and Families
vi
Acknowledgements
I want to express my gratitude to my parents. I am lucky to have you in my life. No matter where I am and what I am working on, you always give me enough freedom, and all the support you can offer. This keeps me working with a warm heart.
I am very grateful to my research advisor, Dr. Nikolas Podraza, for his guidance, suggestions, and encouragement throughout my research and his continuous advice and expertise is always appreciated. Without his innovative suggestions and continuous support, none of the works mentioned in this Thesis research would have been possible.
I am grateful to my committee members: Dr. Robert Collins, Dr. Yanfa Yan, Dr.
Michael Cushing, and Dr. Sylvain Marsillac for being in my committee and providing valuable suggestions. I am thankful to my friends: Prakash Koirala, Maxwell Junda, Kiran
Ghimire, Indra Subedi, Prakash Uprety, Niraj Shrestha, Ebin Bastola, and Biwas Subedi for their support and help.
I thank the National Science Foundation (NSF) and Office of Naval Research
(ONR) for the funding support.
Finally, to my wife, Binita, how could I express my acknowledgements to you? I could not imagine how my life would be without your support. You deserve all the honors
I received and will get. Without your love, support, motivation, and sacrifice, I could not have done this. Thanks to my son Aayan and daughter Aanvi for their patience and love during this time.
vii
Table of Contents
Abstract ...... iii
Acknowledgements ...... vii
Table of Contents ...... viii
List of Tables ...... xi
List of Figures ...... xiii
Chapter 1: Introduction ...... 1
1.1 Motivation and Background ...... 1
1.2 Dissertation Organization ...... 7
Chapter 2: Spectroscopic Ellipsometry and Instrumentation ...... 10
2.1 Introduction ...... 10
2.2 Basic Theories of Light Propagation in Materials ...... 12
2.3 Reflection from a Single Interface ...... 15
2.4 Reflection from Multilayer Stack and Matrix Formalism ...... 17
2.5 Effective Medium Theories in Spectroscopic Ellipsometry ...... 19
2.6 Global 흈̅-minimization Method for RTSE Data Analysis ...... 21
2.7 Measurement and Instrumentation in Spectroscopic Ellipsometry ...... 22
2.7.1 M-2000 Rotating Compensator Spectroscopic Ellipsometer ...... 22
2.7.2 RC2 Dual Rotating Compensator Multichannel Ellipsometer ...... 23
2.7.3 IR-VASE Fourier Transform Infrared Spectroscopic Ellipsometer .24
viii 2.8 Deposition Techniques...... 25
2.8.1 Plasma Enhanced Chemical Vapor Deposition (PECVD) ...... 25
2.8.2 RF Magnetron Sputtering ...... 27
2.8.3 MVsystem Cluster Tool Deposition System...... 29
2.9 Temperature Calibration for Substrate Temperature ...... 30
2.10 Thin-film and Solar Cell Characterization Tools...... 32
2.10.1 Atomic Force Microscopy (AFM) ...... 32
2.10.2 Current-Voltage (J-V) Measurement ...... 33
2.10.3 External Quantum Efficiency (EQE) Measurement ...... 35
2.10.4 X-ray Diffraction ...... 36
Chapter 3: Optical Anisotropy of SrLaAlO4 from Mueller Matrix Ellipsometry .....38
3.1 Introduction and Motivation ...... 38
3.2 Measurement Details ...... 40
3.3 Directional Dependent Optical Properties and Parameterization of Indirect
Band Gap Energy ...... 45
3.4 Summary ...... 56
Chapter 4: Growth Evolution and Analysis of Amorphous to Nanocrystalline
Transition in Sputtered Si:H Films ...... 57
4.1 Introduction and Motivation ...... 57
4.2 Film Deposition and Measurement Details...... 61
4.3 Virtual Interface Analysis (VIA) and Growth Evolution Diagram ...... 63
4.4 Deposition Rate of Sputtered Si:H ...... 69
4.5 X-ray Diffraction Study ...... 70
ix 4.6 Infrared Optical Properties ...... 73
4.7 Summary ...... 80
Chapter 5: n-i-p Solar Cell with Sputtered nc-Si:H Absorber ...... 82
5.1 Introduction and Motivation ...... 82
5.2 Experimental Details ...... 84
5.3 Device Fabrication ...... 89
5.4 Device Characterization and EQE Simulations ...... 91
5.5 Summary ...... 100
Chapter 6: Glancing Angle Deposited CdTe: Impact on Solar Cells ...... 102
6.1 Introduction and Motivation ...... 103
6.2 Film Deposition and Measurement Details...... 105
6.3 Microstructural and Optical Properties ...... 108
6.4 Solar Cell with GLAD CdTe Interlayer ...... 118
6.5 Summary ...... 125
Chapter 7: Conclusion and Future Work ...... 126
7.1 Conclusion ...... 126
7.2 Suggestion for Future Work...... 129
References ...... 131
x
List of Tables
3.1 Parameters describing spectra of for SrLaAlO4 in ordinary direction ...... 54
3.2 Parameters describing spectra of for SrLaAlO4 in extra ordinary direction with
one CPPB oscillator ...... 54
3.3 Parameters describing spectra of for SrLaAlO4 in extra ordinary direction
obtained with two CPPB oscillators ...... 54
4.1 Relative void fractions in low and high deposition rate series of sputtered Si:H
films ...... 76
5.1 Deposition conditions for the individual layers in the nc-Si:H n-i-p solar cell
configuration deposited on 15.24 cm × 15.24 cm soda lime glass (SLG)
substrates in the load-lock cluster tool (i.e., with plasma enhanced chemical
vapor deposition (PECVD) i-layers) ...... 86
5.2 Comparison of device performance parameters of highest efficiency n-i-p cells
incorporating nc-Si:H absorbers prepared with each absorber layer processing
method...... 93
6.1 Deposition conditions for the individual solar cell layers. “RT” denotes room
temperature ...... 108
6.2 Structural parameters including surface roughness thickness (ds), void fraction
percentage in each effective medium approximation (EMA) layer comprising
the surface roughness layer (fv), and bulk film thickness (db) for as-deposited
xi CdTe at different oblique angles (Φ) between the substrate normal and target
normal ...... 113
6.3 Comparison of device performance parameters of highest efficiency CdS/CdTe
solar cells with and without a 100 nm thick GLAD CdTe interlayer. Average
performance parameters with one standard deviation (1-σ) variation for the best
200 (out of 256) small area devices for each configuration are also given...... 121
xii
List of Figures
1 – 1 A p-n junction diode showing solar cell function ...... 2
1 – 2 Spectral irradiance just outside the Earth’s atmosphere (AM0), which has an
integrated power density of 1353 Wm-2 (red)A p-n junction diode showing solar
cell function ...... 3
2 – 1 Schematic of ellipsometry measurement setup in reflection mode ...... 12
2 – 2 Schematic of the electric field vector E(z0, t) for an elliptically polarized light
wave traced out in the x-y plane over time ...... 14
2 – 3 Schematic showing the interaction of light with a single interface between two
media ...... 16
2 – 4 Schematic showing the interaction of light with multi-layered structure ...... 18
2 – 5 Photograph showing single rotating multichannel ellipsometer ...... 22
2 – 6 Photograph of RC2 dual rotating compensator ellipsometer ...... 23
2 – 7 Photograph of IR-VASE spectroscopic ellipsometer ...... 25
2 – 8 Schematic of PECVD process for Si:H thin film deposition ...... 26
2 – 9 Schematic of sputtering system ...... 27
2 – 10 Schematic of sputtering in Glancing Angle Deposition (GLAD) System ...... 28
2 – 11 Schematic top-view of the MVsystem cluster tool ...... 30
2 – 12 Temperature calibration performed for the correct substrate temperature obtained
from CPs shift with temperature in c-Si ...... 31
xiii 2 – 13 Example of J-V curves for a solar cell in the dark and under illumination ...... 33
2 – 14 An example of External quantum efficiency (EQE) spectrum of a 2µm CdTe solar
cell ...... 36
3 – 1 Crystal structure of SrLaAlO4 ...... 40
3 – 2 A photograph of (001) and (100) oriented samples of SrLaAlO4 ...... 41
3 – 3 Schematic diagram showing arrangement for MM measurement at different
rotation azimuth angles for (100)-oriented sample ...... 42
3 – 4 Experimental ellipsometric spectra and model fit for (001)-oriented SrLaAlO4
single crystal over the spectral range from 0.74 to 5.90 eV ...... 45
3 – 5 Spectra in for the ordinary direction of SrLaAlO4 from 0.74 to 5.90 eV
determined from measurements of (001)-oriented SrLaAlO4 and extracted by
numerical inversion (open circles) and divided spectral range parameterization
(solid lines) using Sellmeier oscillator for the low energy region and assuming
critical point parabolic band (CPPB) transitions in the high energy region ...... 46
3 – 6 Experimental Mueller matrix spectra (open symbols) and model fit (solid lines)
obtained from (100)-oriented SrLaAlO4 crystal over the spectral range from 0.74
to 5.90 eV ...... 47
3 – 7 Spectra in for the extra-ordinary direction of SrLaAlO4 from 0.74 to 5.90 eV
determined from measurements of (100)-oriented SrLaAlO4 and extracted by
numerical inversion (open circles) and divided spectral range parameterization
(solid line) using Sellmeier oscillator for the low energy region and assuming
CPPB transitions in the high energy region ...... 48
xiv 3 – 8 Absorption coefficient () corresponding to the ordinary direction of SrLaAlO4 as
a function of photon energy ...... 49
1/2 2 3 – 9 α and α as a function of photon energy for SrLaAlO4 to obtain the indirect and
direct gaps in the ordinary direction ...... 50
3 – 10 Ordinary (black solid line) and extra-ordinary (black dotted and red solid line)
spectra in of SrLaAlO4 obtained by parameterization of numerical inverted in
Figures 3-5 and 3-7 ...... 53
3 – 11 Birefringence (ne − no) and dichroism (ke − ko) of SrLaAlO4 ...... 55
4 – 1 Schematic showing formation of nc-Si:H from a-Si:H via mixed (a+nc)-Si:H.....59
4 – 2 Schematic of layered optical/structural model fit to RTSE measurements for Si:H
film deposited on native oxide coated c-Si wafer as a substrate ...... 62
4 – 3 Variation of surface roughness thickness (ds) with bulk layer thickness (db)
obtained from analysis of real time spectroscopic ellipsometry (RTSE) data
collected for two Si:H films prepared at RF power = 100W and total gas pressure
= 30 mTorr with relative hydrogen gas concentration pH2 = [H2] / {[H2] + [Ar]} ×
100% = 15 and 20% ...... 64
4 – 4 Growth evolution diagrams of Si:H obtained from RTSE depicting thicknesses at
which the amorphous-to-amorphous roughening [a→a] (solid circles), amorphous
to mixed-phase amorphous+nanocrystalline [a → (a+nc)] (open circles), and
mixed-phase to single-phase nanocrystalline [(a+nc) → nc] (solid squares)
transitions occur for films prepared at (a) low and (b) high deposition rates as a
function of pH2 ...... 66
xv 4 – 5 Complex dielectric function, ε = ε1+iε2, spectra for a-Si:H and nc-Si:H used as
reference optical properties in the virtual interface analysis (VIA) ...... 67
4 – 6 Surface roughness and nanocrystallite fraction depth profile in the mixed-phase
(a+nc) regime obtained from VIA of RTSE data for a high deposition rate film
prepared at pH2 = 80% ...... 68
4 – 7 Deposition rates as a function of pH2 for low (black squares) and high (red circles)
rate sputtered Si:H...... 69
4 – 8 (a) Grazing incidence x-ray diffraction (GIXRD) patterns of high deposition rate
Si:H prepared at 250 W RF power and 10 mTorr total gas pressure with varying
pH2 as indicated. (b) Nanocrystalline grain size determined from GIXRD patterns
as a function of pH2 ...... 71
4 – 9 (a) Grazing incidence x-ray diffraction (GIXRD) patterns of low deposition rate
Si:H prepared at 100 W RF power and 30 mTorr total gas pressure with varying
pH2 as indicated. (b) Nanocrystalline grain size determined from GIXRD patterns
as a function of pH2 ...... 72
4 – 10 Possible modes of vibrations in Si:H ...... 74
4 – 11 Infrared (IR) spectra in ε, showing absorption features in for a high deposition
rate Si:H film produced at pH2 = 50%. The inset shows deconvoluted features at
2000 and 2090 cm-1 ...... 75
4 – 12 Integrated area under the absorption peaks centered at 640, 2000, and 2090 cm-1
as a function of pH2 for the (a) low and (b) high deposition rate film series ...... 77
4 – 13 Hydrogen content CH (at. %) as a function of pH2 comparison for low (lower
horizontal axis) and high (upper horizontal axis) rate Si:H ...... 79
xvi 5 – 1 Schematic diagram of nc-Si:H absorber based solar cell in the n-i-p substrate
configuration ...... 85
5 – 2 Three-dimensional atomic force micrographs (AFM) taken over an area of 2µm ×
2µm showing surface morphology of (a) Ag deposited on SLG/Cr and (b) ZnO
deposited on SLG/Cr/Ag...... 90
5 – 3 Current–voltage (J–V) measured under 100 mW/cm2 simulated AM1.5G
irradiation and in the dark for the highest efficiency n-i-p devices incorporating a
PECVD absorber without vacuum break (black squares, black dotted line),
PECVD absorber with vacuum break before and after intrinsic layer deposition
(red circles, red dotted line), and RF magnetron sputtered absorber also with
vacuum break before and after intrinsic layer deposition (blue triangles, blue
dotted line) ...... 92
5 – 4 Solar cell performance parameters including open circuit voltage VOC (a), fill
factor FF (b), short circuit current density JSC (c), power conversion efficiency
PCE (d), series resistance Rs (e), and shunt resistance Rsh (f) for devices
incorporating nc-Si:H absorbers prepared with different methods ...... 94
5 – 5 (a) External quantum efficiency (EQE) spectra for two n-i-p devices incorporating
PECVD nc-Si:H absorbers, one of which is deposited without breaking vacuum
(no air exposure, black solid line) and the other with vacuum breaks before and
after intrinsic layer deposition (air exposure, red line) to simulate the processing
of the samples with sputtered nc-Si:H absorbers (blue solid line). (b) Collection
probability profile used in EQE simulation accounting for the recombination
xvii losses of photogenerated carriers within the 1 µm thick sputtered nc-Si:H i-layer
...... 98
6 – 1 Schematic of glancing angle deposition (GLAD) sputtering ...... 106
6 – 2 Schematic diagram of CdS/CdTe heterojunction solar cells with and without a
thin GLAD CdTe interlayer between the wurtzite CdS and zinc blende CdTe
layers ...... 107
6 – 3 Cross-sectional and surface scanning electron micrographs (SEMs) of as-
deposited GLAD CdTe films prepared as functions of oblique angles with respect
to substrate normal ...... 109
6 – 4 X-ray diffraction (XRD) patterns of (a) as-deposited CdTe sputtered on soda-lime
glass at different oblique angles (Φ) and (b) Log scale XRD of as-deposited CdTe
sputtered at Φ = 0° ...... 111
6 – 5 Complex dielectric function (ε = ε1 + iε2) spectra of as-deposited GLAD
polycrystalline CdTe films produced at different oblique angles during sputtering
...... 112
6 – 6 Anisotropic optical properties parallel (extra-ordinary, e) and perpendicular
(ordinary, o) to the columnar principal axes in (top) ε1 and (bottom) birefringence,
ne-no in the transparent spectral range of CdTe prepared by GLAD at Φ = 60° and
80° ...... 116
6 – 7 Crystallite size deduced via the Scherrer equation as a function of CdCl2 treatment
time for CdTe deposited at different oblique angles ...... 117
6 – 8 Surface SEM images of CdTe film deposited at Φ = 0° (a) before and (b) after
CdCl2 treatment for 30 minutes ...... 118
xviii 6 – 9 XRD patterns of CdS (black), CdCl2 treated CdTe (Φ = 0°) (blue), and CdCl2
treated CdTe (Φ = 80°) (red) ...... 119
6 – 10 Device performance for the highest efficiency solar cells with and without a 100
nm GLAD CdTe interlayer: (a) light and dark current-voltage (J-V) and (b)
external quantum efficiency (EQE) for CdS/CdTe heterojunction solar cells with
(red) and without (blue) introduction of the GLAD interlayer ...... 121
6 – 11 (a) Open circuit voltage (VOC), (b) short circuit current (JSC), (c) fill factor (FF),
and (d) power conversion efficiency (PCE) parameter ranges for all CdTe PV
devices with and without GLAD CdTe interlayers ...... 122
6 – 12 Analysis of dark current density versus voltage characteristics for the CdS/CdTe
heterojunction solar cells with and without GLAD CdTe interlayers: (a) dV/dJ
versus 1/J plot for calculation of n and RS and (b) semi-log scale plot for the
calculation of J0 and n ...... 124
xix Chapter 1
Introduction
1.1 Motivation and Background
Renewable energy sources such as hydropower, wind, geothermal, solar thermal, and photovoltaics are promising non-polluting alternatives to replace the conventional fossil fuels for the need of energy. Among several renewable sources of energy, solar energy is the most reliable and consistent source of renewable energy.
Solar cell converts sunlight energy directly into electricity by means of photovoltaic effect. Basically, a solar cell is a p-n junction, transparent on at least one side, that is engineered for efficient generation, separation, and collection of charge carriers. The schematic diagram of a solar cell is shown in the Figure 1-1. Under solar illumination, the photons with energy higher than the band gap energy of the semiconductor are absorbed and generate electron-hole pairs. These electrons and holes are separated by drift or diffusion and results in the generation of electric current.
The sunlight reaching the earth surface has a spectral distribution that effects the potential efficiency of the solar cell. The solar spectrum measured just before entering earth’s atmosphere is called air mass zero (AM0). Due to the interaction of solar radiation
1
Figure 1-1: A p-n junction diode showing solar cell function.
with the earth’s atmosphere via absorption, reflection, and scattering, the incident radiation on the surface has substantially reduced intensity and slightly different spectral distribution. The spectrum incident on the earth’s surface at sea level with zenith angle of
37 degree has a normalized intensity of 1000 W/m2 and is denoted as air mass 1.5global
(AM1.5G) as shown in the Figure 1-2. The AM1.5G spectral distribution is considered as standard for terrestrial characterization.
The solar cell technology with the longest history, also referred to as the first generation PV technology, is based on single crystalline or multi-crystalline silicon and exhibits 26.7% cell efficiency and 24.4% module efficiency [Green et al., 2019]. Although solar cells made from crystalline silicon have higher PV conversion efficiencies, the cost of production of solar-grade silicon can be high. In attempts to reduce the module cost, PV technology based on thin films has been introduced, called second generation PV. The thin film materials include hydrogenated silicon (Si:H), cadmium telluride (CdTe), and copper
2
Figure 1-2: Spectral irradiance just outside the Earth’s atmosphere (AM0), which has an integrated power density of 1353 Wm-2 (red). Also shown are the irradiances for the terrestrial global spectrum (AM1.5G), having a power density 1000 Wm-2 (blue). All data are based on American Society for Testing and materials (ASTM) G-173 reference spectra.
indium-gallium diselenide [Cu(In,Ga)Se2 or CIGS]. Although the second generation PV technology have some drawbacks such as long term instability, materials toxicity, and limited materials availability for large scale production, it is benefited by wide variety of cost effective deposition techniques (sputtering, closed space sublimation, thermal evaporation, and plasma enhanced chemical vapor deposition etc.) and significantly less material (~ 0.3 - 3 μm thick) required in comparison to that of crystalline silicon (~ 300
µm thick) technology. In this dissertation, the optical and microstructural properties of Si:H and CdTe thin films deposited by RF magnetron sputtering and their relevance for PV application will be discussed.
3 Si:H is an attractive absorber material used as thin film PV devices. Silicon is an abundant element in nature and is non-toxic; its hydrogenated form can be produced in thin films. Several techniques including sputtering, plasma enhanced chemical vapor deposition
(PECVD), hot-wire gas dissociation etc., are available for the fabrication of Si:H. Two categories of Si:H materials can be used for solar cells: 1) Hydrogenated amorphous silicon
(a-Si:H) and 2) Hydrogenated nanocrystalline silicon (nc-Si:H). a-Si:H was one of the first thin film materials incorporated into solar cells and due to its amorphous structure and high above band gap absorption coefficient, the thickness of a-Si:H required in solar cells is the smallest (~ 0.3 µm) compared to other thin film materials. The disadvantage of a-Si:H technology, however, is the low efficiency and the light induced degradation, called
Staebler Wronski Effect (Staebler & Wronski, 1977). Si:H films produced by PECVD are widely studied (Fujiwara & Kondo, 2007; Karki Gautam et al., 2016; Collins et al., 2003) for application in PV devices, while the available literature on sputtered Si:H is limited
(Moustakas, 1984; Dutta et al., 2008; Tiedje et al., 1981). Sputtering is a simple technique which uses inert sputter gases like Ar and the use of toxic sources gases like silane and disilane as in conventional PECVD, can be completely avoided during Si:H fabrication.
Also, doping can be achieved in the composition of the silicon sputter target, eliminating the need for toxic dopant gases such as phosphine, diborane, and others. Additionally, sputtering of Si gives reasonable deposition rate and the microstructure of the film can be effectively controlled during deposition. In this dissertation, sputtered Si:H films are studied in depth, and the n-i-p device are fabricated with sputtered nc-Si:H and compared with conventional PECVD devices.
4 One of the most promising thin film solar cell technologies employs CdTe as the absorber material. Thin film CdTe is the PV technology on which the latter part of this dissertation will be focused. Research solar cells incorporating CdTe as the absorber layer have reached a record efficiency of 22.1% (NREL efficiency chart) and module efficiency of 18.6% (Green et al., 2019). CdTe is very stable material under outdoor conditions and can be deposited using various methods like sputtering, CSS, etc. However, there are some disadvantages with CdTe PV technology which includes, toxicity of Cd and availability of rare Te.
This dissertation introduces the overview of the thin film based Si:H and CdTe solar cells. Different methods of deposition for Si:H and CdTe thin films are discussed with magnetron sputtering as an alternative deposition technique for the active absorber layer for PV application. The subsequent chapters focus on the application of Mueller matrix spectroscopy for the characterization of anisotropic (directional dependent) optical properties of single crystal SrLaAlO4, growth evolution and optical characterization of reactive magnetron sputtered Si:H using real time spectroscopic ellipsometry (RTSE), and sputtered CdTe films produced by glancing angle deposition (GLAD). Solar cells with sputtered nc-Si:H absorber layer in n-i-p configuration is studied in detail. The effect of introducing thin layer of nanostructured engineered GLAD CdTe on CdS/CdTe solar cell is evaluated.
Si:H is currently used in applications including flat panel displays and PV. We have studied the optical, microstructural, and electronic properties of Si:H thin film produced by magnetron sputtering. Being a simple deposition technique, use of non-toxic gases, effective control over microstructure of the resulting film, and industrially scalable for
5 large scale production, are some factors that motivate us to use magnetron sputtering for the deposition of Si:H. The effect of different deposition parameters on the deposition rate was studied, and the growth evolution diagram was produced using real time spectroscopic ellipsometry (RTSE). As the film grows, several structural and phase transitions: amorphous to amorphous (a→a), amorphous to mix [a→(a+nc)], and mix to nanocrystalline [(a+nc)→nc] phase transitions are observed as the hydrogen to total gas ratio (pH2) is increased similar to those observed in conventional PECVD Si:H. Virtual interface analysis (VIA) uses Bruggeman effective medium approximation (EMA) with variable fraction of a-Si:H and nc-Si:H reference optical properties and fitted to RTSE measurements using linear regression analysis to minimize mean square error to extract nanocrystalline fraction in the mixed phase region. Finally, we have incorporated sputtered nc-Si:H absorber layer in a single junction n-i-p structure to make a complete solar cell device. A comparative study of device with nc-Si:H sputtered absorber with PECVD absorber fabricated with identical deposition conditions for all layers except absorber, is made.
CdTe is an II-VI semiconducting material and is important for low cost and high efficiency PV devices. CdTe solar cells are the dominant and the most industrially successful thin film based technology in the current market. Several deposition techniques are available to make CdTe including sputtering, close space sublimation (CSS), vapor transport deposition, etc. CdTe is a direct band gap semiconductor with band gap energy,
Eg ~ 1.5 eV, ideal for optimal solar energy conversion. It has high absorption coefficient
(α) > 104 cm-1 in the visible spectral range and hence a thin layer ~ 2-4 µm thickness is enough to absorb most of the incident light on its surface. Polycrystalline CdTe is
6 associated with grains and grain boundaries, which are the centers for charge carrier trapping and recombination. So, the chemistry and characteristics of grain/grain boundaries play a key role in PV device performance. The film density, grain size, crystal structure, optical properties, electrical properties, microstructure, etc. can be effectively manipulated using GLAD. The CdTe films studied here are prepared by magnetron sputtering using
GLAD at various sputtered source flux angle (oblique angle) ranging from 0° to 80° on soda-lime glass. The GLAD CdTe films were characterized by spectroscopic ellipsometry
(SE), scanning electron microscopy (SEM), and x-ray diffraction (XRD). Finally, we have successfully implemented GLAD CdTe interlayer on standard CdS/CdTe heterojunction solar cell and demonstrated improved device performance.
1.2 Dissertation Organization
This dissertation presents the comprehensive analysis of SE and its applications to obtain the optical properties of variety of materials including bulk single crystal to various thin films based PV devices. Chapter 1 provides a basic overview of thin film PV, especially nc-Si:H and CdTe PV.
In Chapter 2, different experimental methods employed throughout the research project are discussed. The main characterization tool is spectroscopic ellipsometry and the theoretical, instrumentation, and measurement procedures are explained in more details.
Other used characterization tools including XRD, atomic force microscopy (AFM), current-voltage measurement, external quantum efficiency measurement etc., are also briefly explained. The first section gives a description of RF magnetron sputtering,
7 PECVD, and GLAD techniques. The second section focuses on the theory of ellipsometry and the instrumentation in spectroscopic ellipsometry.
Mueller Matrix ellipsometry can be used to obtain the optical properties of anisotropic material. In Chapter 3, anisotropic optical properties of SrLaAlO4 single crystal in the form of birefringence and dichroism obtained. Divided spectral range analysis to extract the structural parameters of the sample commonly used for the isotropic samples have been extended to anisotropic material and works very well here. The indirect and direct band gaps, above band gap critical points, and absorption coefficient are obtained from the Mueller matrix data analysis. The birefringence and dichroism are obtained in the measured spectra range.
In Chapter 4, the growth evolution diagram of magnetron sputtered Si:H have been developed and the effect of deposition conditions on the Si:H film properties are studied using different characterization tools including SE and XRD. RTSE data collected during the film growth are analyzed to study microstructural evolution including a surface roughening and smoothening onsets associated with the amorphous-to-nanocrystalline phase transitions. The RTSE data are further analyzed using virtual interface technique to investigate the growth of nanocrystalline fraction from amorphous phase. IR extended spectroscopic ellipsometry (IR-SE) studies are performed for the sputtered Si:H films and different silicon-hydrogen bonding modes are evaluated. The hydrogen content in sputtered Si:H film are calculated.
Solar cells incorporating sputtered nc-Si:H as an absorber layer in n-i-p device configuration are fabricated and compared with conventional PECVD devices in Chapter
5. The functioning solar cell with sputtered absorber are fabricated in n-i-p substrate
8 configurations. Potential problems associated with solar cells with sputtered absorber are studied using external quantum efficiency (EQE) simulations, and a probability profile associated with incomplete collection of charge carriers is identified.
RF magnetron sputtered CdTe films produced by GLAD are discussed in Chapter
6. Optical and microstructural properties of GLAD CdTe deposited at various sputtered oblique angles are determined from SE, XRD, and SEM. More specifically, the variation in complex dielectric function, film morphology, crystal structure, and crystallite size of
GLAD CdTe films as a function of sputtered oblique angle are studied. The CdS/CdTe solar cells with and without GLAD CdTe interlayer are fabricated and the device performance parameters are compared.
In Chapter 7, we conclude the dissertation by highlighting some of the important results along with the suggestions for the future research directions. Some of the results provided in this dissertation are from ongoing projects and is taken as the basis of future works.
9 Chapter 2
Spectroscopic Ellipsometry and Instrumentation
2.1 Introduction
Spectroscopic ellipsometry (SE) is a non-contact powerful measurement tool based on modeling to obtain structural, optical, and electrical properties, applicable to wide variety of materials (metals, semiconductors, and dielectrics). More specifically, SE can be used to characterize the complex dielectric function (ε = ε1 + iε2), bulk layer thickness, surface roughness thickness, composition, crystalline nature, doping concentration, and electrical conductivity etc., of the thin films and bulk single crystals. In-situ real time spectroscopic ellipsometry (RTSE) has an additional advantage for studying thin films by monitoring film thickness and surface roughness of growing film before oxidation in air.
SE measures a change in polarization as light reflects or transmits from a sample.
The polarization change is represented as an amplitude ratio and a phase shift difference.
Upon interaction with a sample surface, the incident, reflected, and transmitted beams of light define a plane of incidence. Electric fields associated with light are defined as those polarized parallel (p) and perpendicular (s) to the plane of incidence. SE measures two parameters acquired from the polarized light: amplitude ratio (Ψ) and phase shift (Δ),
10 between the two principal polarizations before and after reflection. These two parameters are defined by the ellipsometric quantity ρ, the complex amplitude reflection ratio, = rp/rs
= tanΨ exp(iΔ). Here rp and rs are the complex reflection coefficients of p- and s-polarized fields, respectively. The biggest assets of the method are that one measurement yields two quantities and only ratios of measured quantities are analyzed, which avoids reference measurements. More importantly, ellipsometry measures the ratio of two values, not affected by the fluctuation in light source intensity making it a highly accurate and very reproducible measurement technique.
Ellipsometry is an indirect measurement procedure and relies on data analysis by building a realistic structural/optical model to extract sample information. The model describes the thickness and complex optical properties of each layer, as shown in Figure 2-
1. Any unknown properties are estimated and then varied to find best agreement between the model-generated response (simulations) and experimental data. A least square regression analysis with an unweighted error function σ, are applied to fit the experimental ellipsometric data with the model. The unweighted error function (σ) for ellipsometric spectra (in N, C, and S) is defined as follows:
푛 1 2 2 2 휎 = √ ∑(푁mod − 푁exp) + (퐶mod − 퐶exp) + (푆mod − 푆exp) (2.1) (3푛 − 푚) 푖 푖 푖 푖 푖 푖 푖=1 where n is the number of measured data points and m is total number of variable model parameters. Here, ‘exp’ and ‘mod’ refer to experimentally measured and model-simulated data, respectively. The ellipsometric parameters N, C, and S are related to Ψ and Δ as:
푁 = cos(2훹) (2.2a)
퐶 = sin(2훹)cos (훥) (2.2b)
11
Figure 2-1: Schematic of ellipsometry measurement setup in reflection mode.
푆 = sin(2훹)sin(훥) (2.2c)
The goal of SE data analysis is to find a unique model that minimizes error function σ while also minimizing the number of model fit parameters. More detail of the SE instrumentation and data analysis strategies will be discussed in the later sections.
2.2 Basic Theories of Light Propagation in Materials
The propagation of light in any medium can be described through its constituent electric and magnetic fields by Maxwell’s equations given as:
휌 ∇. 퐄 = (2.3a) 휀0
휕퐁 ∇ × 퐄 = − (2.3b) 휕푡
∇. 퐁 = 0 (2.3c)
휕퐄 ∇ × 퐁 = 휇 퐉 + 휇 휀 ( ) (2.3d) 0 0 0 휕푡
12 Here, E is the electric field vector, B is the magnetic induction vector, J is the current density vector, ρ is the charge density, ε0 and µ0 are permittivity and permeability of free space, respectively (Jackson, 1999). Making simplifying assumptions that the material is isotropic, homogenous, and non-magnetic, and that there are no external charges or currents providing additional field sources, it is well established that a solution to these equations is that of an electromagnetic plane wave (Humlíček, 2005). Using a right-handed, orthogonal coordinate system in which the wave is propagating in the positive z direction, a plane wave solution for the electric field can be written as:
푁풛 퐄(푧, 푡) = 퐄 푒푥푝 (푖휔 ( − 푡)) (2.4) 0 푐
Here E0 is the complex electric field vector of the polarization state of the wave in x-y plane and N is the complex refractive index of the medium through which the light propagates and is defined by
N = n + ik (2.5) where n is the real index of refraction and k is the extinction coefficient. The optical properties of the material in the form of complex dielectric function ε is written as:
2 2 ε = ε1 + iε2 = (n + ik) = N (2.6a)
2 2 ε1 = n – k , ε2 = 2nk (2.6b)
Also, the absorption coefficient (α) at a given wavelength λ can be obtained from extinction coefficient k as,
4휋푘 훼 = (2.7) 휆
Alternatively, n and k can be expressed in terms of ε1 and ε2 as,
13 2 2 2 2 (√(휀1 + 휀2 ) + 휀1) (√(휀1 + 휀2 ) − 휀1) 푛 = √ , 푘 = √ (2.8) 2 2
Now, in Equation (2.4), E0 can be written in terms of x- and y-components as
퐄0 = 퐄0푥푒푥푝(푖휑푥)풙 + 퐄0푦푒푥푝(푖휑푦)풚 (2.9) where E0x and E0y are the amplitudes, φx and φy are the phases, and x and y are unit vectors, along the x- and y-directions, respectively. Fundamentally, this formulation of E (z, t) provides the connection to ellipsometry since the relative phase and amplitude information contained in E0 are determined via measurement allowing access to the desired material optical properties contained in N.
The electric field of a plane wave of light, as given by Equation (2.9), traces out an ellipse in the x-y plane as a function of time t, as shown in Figure 2-2. The two special
Figure 2-2: Schematic of the electric field vector E(z0, t) for an elliptically polarized light wave traced out in the x-y plane over time. Q is the tilt angle measured in counterclockwise-positive sense when facing the light beam source and χ is the ellipticity angle, where χ > 0 corresponds to right-handed polarization and χ < 0 corresponds to left-handed polarization.
14 cases of polarization are:
(i) linear polarization with φx = φy and
(ii) circular polarization with φx = φy ± π/2 and E0x =E0y.
For the most general case, the polarization can be described by an ellipse having the tilt angle Q (–90° < Q ≤ 90°) between its major axis and the x-axis, and the ellipticity angle χ which is the inverse tangent of the ratio of the minor axis (b) to the major axis (a) given by, χ = tan-1(b/a) (–45° ≤ χ ≤ 45°) (Azzam & Bashara, 1977).
2.3 Reflection from a Single Interface
As the simplest case, consider the complex amplitude reflection coefficients for p- and s-polarized light when the light reflects from a planar interface between two materials semi-infinite isotropic, homogeneous, and uniform medium of complex index of refraction,
N0 and N1. Figure 2-3 shows a schematic of interaction of light with a single interface between two media and the p- and s- electric field components for the incident, reflected, and transmitted light is also shown. Snell’s law relates the angle of incidence (θ0) and angle of refraction (θ1) with the optical property of each medium: N0sin(θ0) = N1sin(θ1). In this case, the complex amplitude reflection coefficients for p- and s-polarized light are given by the Fresnel coefficients:
Erp 푁1cos(휃0) − 푁0cos(휃1) rp = = |푟푝| exp(i휑푟푝) = (2.10푎) Eip 푁1cos(휃0) + 푁0cos(휃1)
Ers 푁0cos(휃0) − 푁1cos(θ1) rs = = |푟푠| exp(i휑푟푠) = (2.10푏) Eis 푁0cos(휃0) + 푁1cos(휃1)
15
Figure 2-3: Schematic showing the interaction of light with a single interface between two media. The two media are defined by complex indices N0 and N1 with angle of incidence θ0, and angle of transmission θ1. The p and s are electric field components in the incident (i), transmitted (t), and reflected (r) waves.
Here, E represents the electric field complex amplitude, φ is the phase change upon reflection, and “i”, and “r” denote incident and reflected lights respectively. In the Eqs.,
2.10a and 2.10b, N0 is the real index of refraction of ambient and N1 that of medium and is given by N1 = n1 + ik1. n1 is the real index of refraction of medium and ks is its extinction coefficient.
As discussed earlier in section 2.1, the complex amplitude reflection ratio ρ is given as:
푟p |푟p| ρ = = exp (푖(휑p − 휑s)) = tan훹exp(i훥) (2.11) 푟s |푟s|
Here, tan훹 = |푟p|⁄|푟s| and Δ = φp - φs.
16 The final result, which relates the measured ellipsometry parameters to the complex dielectric function of the substrate medium, is given as:
1 − 휌 2 휀 = 휀 푠푖푛2(휃 ) + [1 + 푡푎푛2(휃 ) ( ) ] (2.12) 1 0 0 0 1 + 휌
2.4 Reflection from Multilayer Stack and Matrix Formalism
The equations described previously in this chapter can be more conveniently manipulated by making use of vector and matrix notation to describe the electric fields in each medium. The reflection and transmission behavior of a multilayer stack sample can be described using a S matrix formalism (Azzam & Bassara, 1977). A generalized schematic of multilayer stack structure consisting m total layers is shown in Figure 2-4 with each layer characterized by its complex index Nj and thickness dj.
In order to describe the effect of individual interfaces and layers of the entire film structure, a 2 × 2 scattering matrix for both p- and s-polarization can be expressed as a product of the interface matrices, I and layer matrices, L.
m S11 S12 푺 = [ ] = 퐈0,1 ∏ 퐋j 퐈j,j+1 (2.13) S21 S22 j=1 where index 0 represents the semi-infinite ambient, and m+1 represents the semi-infinite substrate. The interface and layer matrices for both p- and s-polarizations are given by
1 1 푟푗,푗+1 퐼푗,푗+1 = [ ] (2.14a) 푡푗,푗+1 푡푗,푗+1 1
푍푗 0 퐿푗 = [ −1] (2.14b) 0 푍푗
The complex Fresnel Coefficients for p- and s-polarizations are
17
Figure 2-4: Schematic showing the interaction of light with multi-layered structure.
푁j+1cosθj − 푁jcosθj+1 (rj,j+1)푝 = ; (2.15푎) 푁j+1cosθj + 푁jcosθj+1
푁jcosθj − 푁j+1cosθj+1 (rj−1,j)푠 = ; (2.15푏) 푁jcosθj + 푁j+1cosθj+1
2푁jcosθj (tj−1,j)푝 = ; (2.15c) 푁j+1cosθj + 푁jcosθj+1
2푁jcosθj (tj−1,j)푠 = ; (2.15d) 푁jcosθj + 푁j+1cosθj+1
In addition, Zj in the layer matrix of Equation 2.14b is given by,
18 2πidj 푍 = exp ( 푁 cosθ ) (2.16) j λ j j where dj is the thickness of layer j. The incident angle θj, for the interface between layer j and layer j+1 is obtained from successive applications of Snell’s Law as
푁0sinθ0 = 푁1sinθ1 = 푁jsinθj = 푁j+1sinθj+1 = 푁msinθm = 푁m+1sinθm+1 (2.17)
Hence, the complex Fresnel reflection coefficients of the entire structure can be written as
S 푟 = 21 (2.18) S11
Thus, the S matrix components can be related to the ellipsometric parameter by
rp S21(p) S11(s) 휌 = tanψ exp(iΔ) = = . (2.19) rs S11(p) S21(s)
2.5 Effective Medium Theories in Spectroscopic Ellipsometry
Bulk materials with homogeneous structures have well-defined optical responses in reflection from and transmission through specularly reflecting and transmitting interfaces. In many cases, samples may have interfaces or layers that contain mixtures of multiple component materials like in surface roughness, phase segregated materials, interfacial layers, etc. Effective medium approximations are used to estimate the complex optical response for a mixture of materials each with a unique set of optical properties and volume fraction within the mixture. Generally, an effective medium approximation (EMA) will only accurately reflect the effective optical properties of a mixture in the case where the nature of the mixing is either perfectly homogeneous or involves domains of constituent materials that are significantly smaller in size than the wavelength of probing light. A
19 general expression for the EMA, appropriate for the composite materials in which an inclusion is embedded within a host material, can be written as (Aspnes, 1982):
휀eff − 휀h 휀i − 휀h = ∑ 푓i (2.20) 휀eff + κ휀h 휀i + κ휀h i where εeff, εh, εi, and fi are the effective dielectric function, the host dielectric function, and the dielectric function and volume fraction of the ith component respectively. Here the quantity is given by κ = (1/q) − 1 , where q (0 ≤ q ≤ 1) is the screening parameter which describes the shape of the inclusions. If the fraction of different materials are similar and the material types are comparable, complex dielectric function spectra of the host medium will be equal to the effective medium complex dielectric function spectra. This is called Bruggeman approximation and is the most self-consistent choice and it is adopted for the full scope of this work. With this approximation, Equation 2.20 reduced to:
휀i − 휀eff ∑ 푓i = 0 (2.21) 휀i + 2휀eff i
To describe anisotropic dielectric functions, where the microstructures with non- spherical inclusions, there are two limiting forms of EMA (Aspnes, 1982; Fujiwara et al.,
2000a).
(i) no screening (q = 0); and
(ii) maximum screening (q = 1).
For the electric fields parallel to the component boundaries, there is no screening and this case arises if the inclusions are needle-like structures and lying flat within the plane of the film. The effective principal axis dielectric function (fields parallel to the film plane) becomes εeff = f1ε1 + f2ε2, assuming two components within the structure. Likewise, maximum screening is observed when the electric fields are perpendicular to the
20 component boundaries and this occurs if the inclusions are needle-like structures and standing on end within the film. The effective principal axis dielectric function for this case
-1 -1 -1 becomes, εeff = f1ε1 + f2ε2 .
2.6 Global 흈̅-minimization Method for RTSE Data Analysis
Global 휎̅-minimization method is used to analyze the multiple sets of ellipsometric spectra functions collected versus time. This method is developed by the combination of mathematical inversion solving for (Oldham, 1969) and least squares regression analysis
(An et al., 1990).
The first step in this type of analysis is to determine the optical properties and structural parameters of the substrate and any layers underlying the film under investigation before thin film deposition by performing least-squares analysis on each layer of the underlying stack. The properties retrieved for substrate and underlying layer (if any) are kept fixed and a suitable model for the analysis of thin film growth is adopted. This process is repeated until the ambient above the deposition is reached. The variable microstructural parameters for the growing layer of interest include the bulk layer thickness (db) and the surface roughness layer thickness (ds) and time independent spectra in is extracted using this approach, whereas, the optical response of the surface roughness layer is represented by a Bruggeman effective medium approximation (Aspnes, 1982; Fujiwara et al., 2000a).
21 2.7 Measurement and Instrumentation in Spectroscopic Ellipsometry
2.7.1 M-2000 Rotating Compensator Spectroscopic Ellipsometer
This is the mostly used ellipsometer for the ex-situ measurements of sample after deposition in this dissertation work. It is a single rotating compensator multichannel spectroscopic ellipsometer (J. A. Woollam Co., Inc. models M-2000XI and M-2000FI) covering the spectral range from near infrared (NIR) to ultraviolet (UV) wavelengths (0.74
– 5.9 eV). This type of ellipsometer is useful for the measurement to obtain sample thickness, optical properties, and mapping purposes. Figure 2-5 shows a schematic with the major components of M-2000 ellipsometer. The detail overview of this type of ellipsometer is presented in Lee et al., 1998 and Johs et al., 1999. It consists of a 75 W Xe arc-lamp providing intense, broadband light source, a quartz-biplate based compensator, and calcite Glan-Taylor type polarizers. The detector unit contains two separate diode array
Figure 2-5: Photograph showing single rotating multichannel ellipsometer (Model M 2000, J. A. Woollam).
22 spectrometers. The NIR portion of the spectrum covering 0.74 – 1.24 eV (1675 - 1000 nm) is measured by an InGaAs photodiode array whereas the visible to UV portion covering
1.24 – 5.9 eV (1000 – 210 nm) is measured by a Si CCD array. The usage of detector arrays allows for simultaneous measurement of ellipsometric data at 695 separate photon energies distributed over the full spectral range with a minimum acquisition time of less than one second.
2.7.2 RC2 Dual Rotating Compensator Multichannel Ellipsometer
The RC2 type ellipsometer consists of dual rotating compensators, which can measure all 16 sample Mueller matrix elements. It has a 150 W Xe arc-lamp as a light source covering spectral range from NIR to UV of 210 – 1690 nm wavelengths (0.74 ev to
0.89 eV). The RC2 ellipsometer is fixed angle base with 75° angle of incidence and the schematic of RC2 ellipsometer is shown in the Figure 2.6. It is useful to determine both
Figure 2-6: Photograph of RC2 dual rotating compensator ellipsometer showing major components.
23 isotropic and anisotropic complex dielectric function spectra ε related to electronic transitions as well as structural properties like layer and surface roughness thicknesses. The
RC2 uses the same dual detector configuration as the M2000 ellipsometer. A brief overview of this type of instrument is provided in Chen et al., 2004a. In our work, we have used RC2 ellipsometer for in-situ real time measurement during film deposition and ex- situ Mueller matrix measurement for anisotropic materials. RC2 ellipsometers are mounted to angled (70°), low-strain windows on deposition chambers to perform in situ real-time measurements during sample fabrication (as is discussed in Chapters 4 and 5). Typical measurement times during in situ, real-time measurement are usually 1-5 s, while ex situ measurement times for a single angle are often >10 s in order to improve the signal-to- noise ratio by averaging over a large number of optical cycles. The in situ measurement times are chosen such that a single measurement should capture the properties of a sample undergoing changes (i.e. film deposition, etching, phase change, etc.) at a single moment in time. Typical deposition rates for the processes described in this dissertation are on the order of 1 Å/s so the 1-5 s acquisition times used for in situ measurements are sufficiently short so that the sample does not undergo significant changes during any single measurement.
2.7.3 IR-VASE Fourier Transform Infrared Spectroscopic Ellipsometer
The FTIR spectroscopic ellipsometer used in this work (J. A. Woollam, Co.) is a single rotating compensator ellipsometer covering covering the mid-infrared (MIR) to NIR spectral range of 0.03 – 0.74 eV (41.3 – 1.6 μm). (J. A. Woollam Co., Inc. model IR-
VASE). This ellipsometer is sensitive to free carrier absorption and phonon modes related to bonding and lattice vibrations. In this dissertation, Si-Hn IR vibrational modes for Si:H
24
Figure 2-7: Photograph of IR-VASE spectroscopic ellipsometer (J. A. Woollam) showing the location of all major optical elements.
films produced by reactive magnetron sputtering are studied as discussed in chapter 4. A brief overview of this instrument is provided in Tiwald et al., 1998; Johs et al., 1999. Figure
2-7 shows the photograph of IR-VASE spectroscopic ellipsometer (J. A. Woollam) showing the location of all major optical elements. The major optical elements employed by this IR ellipsometer are a SiC glow bar light source housed within a Michaelson interferometer, a dual- rhomb achromatic rotating compensator, wire grid polarizers, and a deuterated-triglycine sulfate (DTGS) type detector.
2.8 Deposition Techniques
2.8.1 Plasma Enhanced Chemical Vapor Deposition (PECVD)
PECVD is a chemical vapor deposition technique in which high quality film can be deposited at comparatively lower temperature. All p-type and n-type nc-Si:H layers used
25 for this dissertation have been obtained using radio frequency PECVD. During PECVD process, silane (SiH4) is used as source gas and different dopant gases, phosphine (PH3) for n-type doping and diborane (B2H6) as p-type doping are used. Hydrogen (H2) gas is also used to dilute source gases which helps to produce different phase materials: amorphous
(at lower dilution), mixed phase (intermediate dilution), and nanocrystalline (higher dilution). Chemical reactions are involved in the process, which occur after creation of the plasma of the precursor gases. For the results presented in this thesis the thin silicon layers have been obtained using a capacitively coupled PECVD system, in which a plasma is created between the two electrodes by applying an AC frequency (RF ~ 13.56 MHz).
During Si:H film deposition, the source gas SiH4 diluted with H2 is admitted to the reaction chamber, gas phase chemical reactions take place in presence of plasma at relatively lower temperature (200°C), and finally a film is deposited on a substrate. The film properties (e.g.
Figure 2-8: Schematic of PECVD process for Si:H thin film deposition.
26 composition, structural, and optoelectronic properties) can be controlled by varying the gas-phase composition, substrate temperature, deposition pressure, applied power density, and excitation frequency. In this research, RF excitation (frequency 13.56 MHz) was used for all intrinsic and doped Si:H layer (n- and p-layer) deposition. The PECVD chambers consist of 7" × 7" cathodes and the spacing between the cathode and the substrate (anode) is 1.5 cm. Figure 2-8 shows a schematic diagram of PECVD process.
2.8.2 RF Magnetron Sputtering
Sputtering process is widely used and industrially scalable deposition technique for semiconductors, metals, and insulators. Sputtering is simple and allows the deposition of films having the same composition as the target source. Based on the type of sputter material, various kinds of power source can be applied (DC, RF, VHF, Pulse DC etc.) for effective deposition. Most of the layers discussed in this dissertation are deposited by RF magnetron sputtering. The simple schematic of sputtering process is shown in Figure 2-9.
Figure 2-9: Schematic of sputtering system.
27
Figure 2-10: Schematic of sputtering in Glancing Angle Deposition (GLAD) System.
In magnetron sputtering, the density of the plasma used in the process is boosted by a magnetic field oriented parallel to the surface of the target (or cathode), trapping energetic electrons in this region (Rossnagel et al., 1990) producing better sputtering yield.
Inert argon (Ar) gas is introduced into the vacuum chamber. The high energetic Ar ions in the plasma removes target atoms/molecules by transferring kinetic energy to a target and is deposited on the substrate as a film. In reactive sputtering, other reactive gases like H2,
O2, N2, etc are also used along with Ar. As an example, Ar with a very small amount of O2 is used during indium tin oxide (ITO) deposition which helps to obtain nearly stoichiometric TCO layer. Similarly, the sputtered Si:H layer discussed in chapters 4 and
5 are deposited by reactive sputtering of H2 in Ar.
Commonly during sputtering the sputtered source flux is oriented normal to the substrate plane. The microstructure, density, and optical property of the deposited film can be manipulated effectively by GLAD technique, at which the sputtered source flux is
28 oblique with respect to the substrate normal. The schematic of sputter GLAD process is shown in the Figure 2-10. Films having columnar structure inclined towards the incoming material flux and different microstructures and optical properties can be produced because of atomic scale self-shadowing effects (Adhikari et al., 2019) and is discussed in more details in chapter 6.
2.8.3 MVsystem Cluster Tool Deposition System
The cluster tool deposition system used for the fabrication of component layers of
Si:H solar cells has seven deposition chambers with a load lock (LL) as shown in the schematic Figure 2-11. Out of seven chambers, five chambers labeled as PL2, PL3, PL4,
PL5, and PL7 are used in our study for making n-i-p solar cells with nc-Si:H absorbers which is discussed in Chapter 4. The chamber PL2 is used to sputter metals (Cr & Ag) and
PL7 for transparent conducting oxides (TCOs) (ZnO & ITO). The three PECVD chambers
PL3, PL4, & PL5 are used for standard Si:H n-, i-, and p- layer depositions. These chambers are connected through a central transfer zone, called the isolation and transfer zone (ITZ) so that the thin film nc-Si:H solar cell layers Ag/ZnO/n/i/p/ITO can be deposited in succession without a vacuum break. Each chamber is equipped with ellipsometry ports for
RTSE measurement. Both the sputtered chambers PL2 and PL7 contains 11.25" × 2.25" sputtered targets of respective materials. The PECVD chambers PL3, PL4, and PL5 each consists of cathode of size 7" × 7" and the spacing between the cathode and the substrate is 1.5 cm. In the cluster tool deposition system, all parameters other than the heater temperature are controlled by computer through a programmable logic controller (PLC).
Both heaters (substrate & well) are controlled by proportional-integral-derivative (PID) loops.
29
Figure 2-11: Schematic top-view of the MVsystem cluster tool. This multi-chambered system consists of three PECVD reaction chambers (PL3, PL4, and PL5 for p-, i-, and n-layer deposition), a VHF-PECVD reactor (PL6 for i-layer deposition), two rf-sputtering reactors (PL2 for metals and PL7 for TCOs), and PL8 co-evaporation chamber. In this figure, PL1 (LL) stands for load lock chamber.
2.9 Temperature Calibration for Substrate Temperature
Sputtered Si:H films discussed later in chapter 4 and 5 are prepared in the K. J.
Lasker AXXIS vacuum chamber. The Si:H films are deposited at the substrate temperature of 200°C. The properties of the films highly depend on the deposition parameters. So, calibration of the deposition system to obtain the correct temperature at the substrate for a given heater set point is important. The temperature calibration was performed based on
30
Figure 2-12: Temperature calibration performed for the correct substrate temperature obtained from CPs shift with temperature in c-Si. Top figure shows the substrate temperature profile for the given heater set points. Bottom figure shows the substrate temperature as a function of heater set point and the linear fit.
31 the change in position of single crystal silicon (c-Si) critical points (Lautenschlager et al.,
1987).
The temperature calibration procedure starts with heating bare c-Si substrate (2 nm native oxide coated c-Si in this case) in a similar deposition condition during film deposition. The substrate heater is given a set point value and run for at least one hour to equilibrate temperature. RTSE analysis tracks the actual temperature at the substrate with time based on the critical point shift of c-Si with temperature. The correct substrate temperature for a given heater set point is obtained by taking average value of temperature at sufficiently later time after stabilization is reached. Figure 2-12 shows the RTSE determined substrate temperature as a function of time for a given range of heater set point.
Similarly, the intrinsic and n-doped PECVD nc-Si:H are deposited at substrate temperature of 200°C while p-doped nc-Si:H at 100°C. The temperature calibration from previous work
(Dahal, 2013) is used as reference for the actual substrate temperatures in the PECVD chambers.
2.10 Thin-film and Solar Cell Characterization Tools
2.10.1 Atomic Force Microscopy (AFM)
Atomic force microscopy (AFM) is used to study the surface morphology of the sample. AFM reveals a three-dimensional profile of a surface by using a cantilever with a very sharp tip to scan over a sample surface. As the tip approaches the surface, the close- range, attractive force between the surface and the tip cause the cantilever to deflect towards the surface. However, as the cantilever is brought even closer to the surface, increasingly repulsive force takes over and causes the cantilever to deflect away from the
32 surface. Root mean square value of the roughness (drms) is the extracted to compare the surface morphology between different samples. In our work, the rms roughness obtained from AFM measurement is compared with roughness values from SE.
2.10.2 Current-Voltage (J-V) Measurement
Measurement of the J-V curve is the most basic way to characterize a solar cell.
The goal of the J-V measurement is to measure the output power from a solar cell at standard test conditions. The test conditions are Air Mass 1.5 (AM 1.5) solar spectrum (100 mW/cm2) at 25 ºC. With no illumination (dark condition), the current through the solar cell
Id is given by a non-ideal diode equation:
푞푉 퐼 = 퐼 [exp ( ) − 1] (2.22) 푑 0 푛푘푇
Figure 2-13: Example of J-V curves for a solar cell in the dark and under illumination. The two curves are offset by the current density JL due to the current generated by illumination of the solar cell.
33 where I0 is the dark saturation current (or leakage current), V is the applied voltage, q is electronic charge, n is the diode ideality factor, k is Boltzmann’s constant, and T is temperature in Kelvins. Under illumination, the net current flowing through the solar cell is modified as (Honsberg & Bowden, 2010)
푞푉 퐼 = 퐼 −퐼 = 퐼 [exp ( ) − 1] − 퐼 (2.23) 푑 퐿 0 푛푘푇 퐿
The light generated current IL flows in opposite direction of dark current Id. An example of
J-V curve of a solar cell under dark and illumination is shown in the Figure 2-13. The quality of a solar cell is measured in terms of four different performance parameters.
A) Short circuit current density (JSC)
It is the maximum current from a solar cell when short circuited. The short circuit current results from the generation of charge carriers by light and the separation of those charge carriers by the built-in electric field. JSC depend on various factors like the spectrum and the irradiance of the light, the optical properties of the materials in the solar cell and their thicknesses, and the lifetime of minority carriers.
B) Open circuit voltage (VOC)
It is the maximum voltage available from the solar cell under open circuit condition.
VOC depends on several factors like bandgap of the active layer, behavior of the diode in the dark, defect density at the junction, and recombination losses.
C) Fill factor (FF)
The fill factor of the solar cell measures the squareness of the J-V curve and is given by the ratio of the maximum power output to the product of JSC and VOC.
푃 퐽 푉 퐹퐹 = 푀푃 = 푀푃 푀푃 (2.24) 퐽푆퐶푉푂퐶 퐽푆퐶푉푂퐶
34 Here 퐽푀푃 and 푉푀푃 are the current density and voltage at the maximum power point. The
FF also depends on defect density in the bulk of the semiconductor and the junction, as well as on the parasitic resistances, known as shunt and series resistances, present in the solar cell. A real solar cell device has series resistance RS due to layer and contact resistances and shunt resistance Rsh usually due to small shunt paths through or around device layers.
D) Power conversion efficiency (PCE)
The power conversion efficiency (PCE) of solar cell is the ratio of maximum power output to the incident power.
푃 퐽 푉푂퐶퐹퐹 푃퐶퐸 = 표푢푡 = 푆퐶 (2.25) 푃푖푛 푃푖푛 where FF is given in Equation 2.24 and Pin is the input power from the sun or a solar simulator.
2.10.3 External Quantum Efficiency (EQE) Measurement
The external quantum efficiency (EQE) measures the response of a solar cell as a function of the wavelength of light and is given by (Hegedus & Shafarman, 2004):
Number of electron − hole pairs collected per second 퐸푄퐸 (휆) = (2.26) Number of photons incident per second
The photons that are reflected or absorbed in the non-active layers (doped layers, TCOs, etc), and the charge carriers that recombine before collection do not contribute to the EQE.
An example of an EQE spectrum is shown in Figure 2-14 with various optical losses observed in the solar cells. Hence, quantum efficiency measurement is very helpful for the characterization of different loss mechanisms in the solar cell that could otherwise contribute to the photocurrent. The Jsc of the solar cell can also be obtained from the EQE
35
Figure 2-14: An example of External quantum efficiency (EQE) spectrum of a 2µm CdTe solar cell. The possible losses are also depicted in the EQE spectrum.
measurement by integrating the product of the measured EQE with the illumination spectrum (typically AM1.5) over the entire wavelength range.
퐽푆퐶 = 푞 ∫ 퐹(휆)퐸푄퐸(휆)푑휆 (2.27) where F(λ) is the photon flux density per unit wavelength. In an ideal solar cell, each photon incident on the solar cell produces one electron-hole pair to contribute to the photocurrent and, thus, the EQE is 100% above the band gap, and zero below the band gap. However, in a real solar cell, optical and electrical losses reduce the EQE spectrum significantly from that of the ideal case.
2.10.4 X-ray Diffraction
XRD is a simple and commonly used characterization technique providing information on crystallographic structure, crystal orientation, grain size, and stresses. X-
36 rays have wavelengths of few angstroms, which are in the same order of the inter-atomic distances in crystals. When an x-ray beam of wavelength λ is incident on parallel planes of atoms at a glancing angle of θ, the reflected beams interact either constructively or destructively depending upon the path difference. According to Bragg’s law, constructive interference takes place when the path difference is integral multiples of wavelength. It is given as,
2푑푠푖푛휃 = n휆 (2.28) where λ is the wavelength of the X-ray (usually Cu Kα 1.54059 Å), d is the inter planar spacing between the adjacent parallel planes, θ is the incident angle of x-ray, and n is an integer. In this dissertation, grazing incidence x-ray diffraction (GIXRD) is frequently used in which the x-ray beams are incident on the film surface at very small grazing angle
(usually 0.5° to 2°). The GIXRD measurement provides higher sensitivity to thinner films and no information from substrate is included in the measurement data.
37 Chapter 3
Optical Anisotropy of Strontium Lanthanum Aluminum Oxide (SrLaAlO4) from Mueller Matrix Ellipsometry
Mueller matrix ellipsometry is a powerful technique to study the optical anisotropy
(directional dependence) of bulk crystal as well as thin film materials. Investigation of anisotropic optical properties is a complex procedure as compared to isotropic materials and requires acquisition of additional data, such as more elements of the Mueller matrix.
Single crystal SrLaAlO4 is taken as a case study here to demonstrate how Mueller matrix ellipsometry can be used to obtain optical anisotropy in the material. The result presented in this Chapter have been published in Adhikari et al., 2016 and reprinted with permission from John Wiley & Sons, Ltd., (© 2016 WILEY-VCH Verlag GmbH & Co. KGaA,
Weinheim). The data acquisition and analysis procedure developed here for optical anisotropy characterization are adopted for the CdTe in subsequent chapters.
3.1 Introduction and Motivation
SrLaAlO4 is a tetragonal crystal of structure type K2NiF4 (space-group I4/mmm) with lattice constants: a = b = 3.756 Å and c = 12.635 Å (Shannon et al., 1992). The elementary cell of SrLaAlO4 is shown in the Figure 3-1 with the crystal built along the c- axis. The oxygen octahedral centers are occupied by Al-ions with Sr and La-ions randomly
38 placed between the octahedrons on nine coordinated C4v symmetric sites (Shannon et al.,
1992; Brown et al., 1990). This structure makes SrLaAlO4 uniaxially anisotropic.
SrLaAlO4 is a good substrate material for Y-Ba-Cu-O (YBCO) (Brown et al., 1990;
Sobolewski et al., 1991) and Bi-Sr-Ca-Cu-O (BSCCO) (Sobolewski et al., 1991) type high- temperature superconducting films used in microwave and far-infrared applications due to its good lattice matching and low thermal expansion mismatch. SrLaAlO4 crystals do not generally develop twins nor form intermediate phases at high temperature (Zimina et al.,
2003; Pajaczkowska & Gloubokov, 1998), and possess low dielectric constant and low dielectric losses (Shannon et al., 1992; Sobolewski et al., 1991). Several authors have investigated thermal, crystallization, and optical properties of SrLaAlO4 single crystal
(Brown et al., 1990; Zimina et al., 2003; Pajaczkowska & Gloubokov, 1998; Chen et al.,
2004b; Humlíček et al., 2000, Kamba et al., 1998; Hora et al., 1996; Gloubokov et al.,
1995; Ryba-Romanowski et al., 1995; El-Gohary et al., 2003). Single crystal SrLaAlO4 exhibits uniaxial anisotropic optical (Kamba et al., 1998; Hora et al., 1996; Ryba-
Romanowski et al., 1995; El-Gohary et al., 2003) and thermal properties (Pajaczkowska
& Gloubokov, 1998). The infrared optical properties have been investigated by J. Humlíček et al., 2000. J. Hora et al., 1996, studied the anisotropic optical properties and measured the birefringence of (100) SrLaAlO4 single crystal in the spectral range from 1.4 eV to 3.4 eV and reported a birefringence of 0.020 ± 0.005 at 1.96 eV. There are different values of the band gap energy reported for this material ranging from 2.31 eV to 5.0 eV (J. Hora et al., 1996; El-Gohary et al., 2003; Demsar et al., 2007; Jezierski, 1998). In our study, the near infrared to ultraviolet anisotropic optical properties of SrLaAlO4 have been investigated over the spectral range from 0.74 eV to 5.90 eV. In this work we have analyzed
39
Figure 3-1: Crystal structure of SrLaAlO4. Reprinted with permission from Adhikari et al., 2016 (© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim).
ellipsometric and Mueller matrix spectra in order to extract both the ordinary and extra- ordinary optical properties of SrLaAlO4. From this information, we have deduced the birefringence, dichroism, critical point transition energies, and information pertaining to the band gap.
3.2 Measurement Details
Single side polished (001)- and (100)-oriented SrLaAlO4 single crystals (10 x 10 x
1 mm3) have been optically characterized by spectroscopic ellipsometry. Ellipsometric spectra (in N = Cos2, C = Sin2Cos, S = Sin2Sin) at room temperature have been
40 collected ex situ at 70º angle of incidence in reflection mode using a single rotating compensator multichannel ellipsometer (Lee et al., 1998; Johs et al., 1999) (Model M-
2000, J. A. Woollam Co., Inc.) for the (001)-oriented sample. Similarly, room temperature
Mueller matrix spectra have been collected ex situ at 75º angle of incidence using a dual rotating compensator multichannel ellipsometer (Chen et al., 2004a) (Model RC2, J. A.
Woollam Co., Inc.) for the (100)-oriented sample from 210 nm (5.90 eV) to 1690 nm (0.74 eV) covering the near infrared to the ultraviolet wavelength range. The RC2 ellipsometer used in this study has a silicon CCD detector with 1 nm wavelength spacing and an InGaAs detector array with 2.5 nm wavelength spacing for wavelengths shorter and longer than
1000 nm, respectively. The unpolished backside of each sample is sufficiently rough so that only reflections from the initial crystal-ambient sides are collected. (001) and (100) of
SrLaAlO4 single crystals is shown in the Figure 3-2.
The optical response of uniaxial anisotropic samples like SrLaAlO4 is different with respect to crystal cut and orientation. Along the (001)-oriented sample, the electric field components of incident light oscillating parallel and perpendicular to the plane of incidence are primarily sensitive to the ordinary optical response in the form of the complex dielectric
Figure 3-2: A photograph of (001) and (100) oriented samples of SrLaAlO4.
41 function, = 1 + i2, spectra. The optic axis in the (001)-oriented sample is perpendicular to the sample surface and to first order behaves isotopically upon rotation of the crystal surface about its normal. Analysis of the ellipsometric spectra for the (001)-oriented crystal will yield spectra in along the ordinary direction. For the (100)-oriented sample, the electric field components are sensitive to both the ordinary and extra-ordinary spectra in as the optic axis lies parallel to the sample surface. Rotation of the (100)-oriented sample about the surface normal will affect the measured ellipsometric spectra. To characterize the extra-ordinary spectra in , the full Mueller matrix spectra of this sample are measured.
The Mueller matrix elements were obtained for the (100)-oriented sample at different azimuthal angle rotations (0o, 45o, 90o) of the optic axis with respect to the plane of incidence as shown in Figure 3-3. Ellipsometric spectra collected from the (001)-oriented sample are analyzed to obtain ordinary spectra in ε. The ordinary spectra in are used as a reference during the analysis of Mueller matrix spectra of the (100)-oriented sample to extract extra-ordinary spectra in .
Figure 3-3: Schematic diagram showing arrangement for Mueller matrix measurement at different rotation azimuth angles for (100)-oriented sample. The optic axis is on the X-Y plane and rotated about Z.
42 A structural model consisting of a semi-infinite anisotropic LaSrAlO4 substrate/surface roughness/semi-infinite air ambient, parametric expressions describing ordinary and extra-ordinary spectra in , and a least squares regression analysis with an unweighted error function () (Johs & Herzinger, 2008; Alterovitz & Johs, 1998) is applied to fit the experimental ellipsometric and Mueller matrix spectra. The unweighted error function (σ) for ellipsometric spectra (in N, C, and S) is given in the Equation (2.1) and in normalized Mueller matrix representation, the error function is written as,
푛 4 4 1 2 휎 = √ ∑ ∑ ∑ (푀mod − 푀exp) (3.1) (15푛 − 푚) 푘푙,푗 푘푙,푗 푗=1 푘=1 푙=1 where n is the number of measured data points, m is total number of variable model
th parameters, and Mkl,j are the individual Mueller matrix elements of the j data point. Here,
‘exp’ and ‘mod’ refer to experimentally measured and model-simulated data, respectively.
In order to obtain the correct value of structural parameters (surface roughness thickness, void fraction within the surface roughness layer), we have applied divided range analysis for both (001)- and (100)-oriented samples. In previous work (Karki Gautam et al, 2014;
Haneef & Podraza, 2014; Ghimire et al., 2015) this approach has been used for isotropic modeling but is being expanded for an anisotropic uniaxial material here. The full measured spectral range (0.74 to 5.90 eV) is divided into two segments: a transparent region from 0.74 to 2.60 eV and a highly absorbing region from 5.25 to 5.90 eV. In the transparent range, spectra in is represented using a Sellmeier oscillator (Collins &
Ferlauto, 2005) given as,
43 2 퐴퐸 ( ) 0 휀 퐸 = ( 2 2) (3.2) 휋 퐸0 − 퐸
where A is the amplitude and Eo is the resonance energy which is outside the measured spectral range. In the highly absorbing range, the spectra in is parameterized using critical point parabolic band (CPPB) oscillators (Aspnes, 1980)] described by,
퐴푒푖휑(0.5Γ) 휀퐶푃푃퐵(퐸) = 휇 (3.3) (퐸푛 − 퐸 − 0.5푖Γ)
where A is the amplitude, En is the critical point resonance energy, is the critical point broadening, ϕ is the phase projection factor, and exponent μ describes the dimensionality of the critical points. Two CPPB oscillators are used to model the spectra in ε and a constant additive term to ε1, represented by is also included in each parameterization.
The surface roughness ε is described by a Bruggeman effective medium approximation (Fujiwara et al., 2000a) consisting of two components with dielectric functions εmat and εvoid, and is defined by,
휀mat−ε 휀void−ε ( ) 푓푚푎푡 + ( ) 푓푣표푖푑 = 0 (3.4) 휀mat+2ε 휀void+2ε
where fmat and fvoid are the volume fractions of material and void respectively. In the divided spectral range analysis, independent optical models are used to fit experimental data in the appropriate spectral ranges, while structural fit parameters are held common in the analysis of these spectral ranges. No initial assumption is made for the line shape of in the weakly absorbing region from 2.60 - 5.25 eV, which is excluded from the divided spectral range analysis parametric fitting.
44 3.3 Directional Dependent Optical Properties and Parameterization of Indirect Band Gap Energy
The ordinary spectra in are obtained by analyzing the ellipsometric spectra from the (001)-oriented sample. Figure 3-4 shows experimental ellipsometric spectra fit using divided spectral range analysis. After obtaining structural parameters (surface roughness thickness = 20.7 ± 0.5 nm, void fraction = 0.157 ± 0.004, with σ = 7.95 x 10-4) from the divided spectral range analysis, these parameters are fixed, and numerical inversion is used to obtain ordinary spectra in ε (Oldham, 1969) as shown in Figure 3-5. There is good
Figure 3-4: Experimental ellipsometric spectra (open circles) and model fit (solid lines) for (001)-oriented SrLaAlO4 single crystal over the spectral range from 0.74 to 5.90 eV. Reprinted with permission from Adhikari et al., 2016 (© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim).
45 agreement between the ordinary numerically inverted spectra in ε and the parametric expressions in the appropriate spectral range. The same type of divided spectral range analysis is applied to analyze the Mueller matrix spectra for the (100)-oriented sample. To obtain the extra-ordinary optical properties of SrLaAlO4, the full measured Mueller matrix spectra is collected for the (100)-oriented sample at three different azimuthal angles between the optic axis and the plane of incidence: 0º, 45º, and 90º. Spectra in along the
Figure 3-5: Spectra in for the ordinary direction of SrLaAlO4 from 0.74 to 5.90 eV determined from measurements of (001)-oriented SrLaAlO4 and extracted by numerical inversion (open circles) and divided spectral range parameterization (solid lines) using Sellmeier oscillator for the low energy region and assuming critical point parabolic band (CPPB) transitions in the high energy region. Reprinted with permission from Adhikari et al., 2016 (© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim).
46 ordinary direction are fixed from the analysis of the (001)- oriented sample. Divided spectral range analysis has been applied to the 45º azimuthal angle measurement to obtain structural parameters (surface roughness thickness = 18.4 ± 0.2 nm, void fraction = 0.243
± 0.004, and σ = 3.45 x 10-2).
For fits to both ellipsometric and Mueller matrix spectra, the error bars reported are mathematically generated 90% confidence limits based on the uniqueness of each parameter and the overall fit quality between the model and the experiment. In general,
Figure 3-6: Experimental Mueller matrix spectra (open symbols) and model fit (solid lines) obtained from (100)-oriented SrLaAlO4 crystal over the spectral range from 0.74 to 5.90 eV. Reprinted with permission from Adhikari et al., 2016 (© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim).
47 there are always potential sources of error outside of the model that are not fully accounted for including spatial variation across the sample.
Experimental Mueller matrix spectra and fit from divided spectral range analysis is shown in Figure 3-6. After obtaining structural parameters, numerical inversion is used to extract spectra in ε for the extra-ordinary direction, shown in Figure 3-7 and in good agreement with the parametric models over the appropriate spectral ranges. As a consistency check, the Mueller matrix spectra collected at all orientations (0º, 45º, 90º)
Figure 3-7: Spectra in for the extra-ordinary direction of SrLaAlO4 from 0.74 to 5.90 eV determined from measurements of (100)-oriented SrLaAlO4 and extracted by numerical inversion (open circles) and divided spectral range parameterization (solid line) using Sellmeier oscillator for the low energy region and assuming CPPB transitions in the high energy region. Reprinted with permission from Adhikari et al., 2016 (© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim).
48 have been modeled with all structural parameters and spectra in fixed and only the azimuthal Euler angle varied. This analysis resulted in azimuth angle parameters each having fit errors of less than 5º and good correspondence to their respective nominal values of sample azimuthal orientation.
The numerically inverted spectra in in the ordinary direction are used to determine the extinction coefficient (k, from 1/2 = n + ik) and absorption coefficient ( = 4k/) which is plotted in Figure 3-8. Below the absorption edge where is low, ~102 cm-1 in this case, reflection-based ellipsometry lacks sensitivity (Haneef & Podraza, 2014; Ghimire et al., 2015). After obtaining , the indirect and direct band gaps are extracted by plotting
1/2 and 2 respectively as a function of photon energy and extrapolating 1/2 = 0 and 2 =
0 as shown in Figure 3-9 (Pankov, 1975; Tauc et al., 1966). An indirect band gap is identified due to the presence of two slopes in 1/2, indicating phonon interactions. Using this approach, indirect band gap energy is identified at 2.74 ± 0.01 eV and a direct gap at
5.05 ± 0.01 eV.
Figure 3-8: Absorption coefficient () corresponding to the ordinary direction of SrLaAlO4 as a function of photon energy.
49 There is a significant difference in the location between the two slopes observed in Figure
3-9 implying that phonons may not be solely responsible for absorption seen at photon energies less than the higher energy intercept at 3.01 eV. Absorption due to defects below the band gap and the presence of an Urbach tail have not been identified in the model.
There could be multiple effects including defect states and multiple phonon emission- absorption processes contributing to the shape of α in the vicinity of the band gap, however it does retain two linear slopes with a higher and lower energy intercept. The higher energy intercept at 3.01 ± 0.01 eV may be considered an upper limit for the indirect gap in this material regardless of the presence of absorption contributions from other processes or defect states. Hora et al., 1996; Demsar et al., 2007; and Qing et al., 2009 reported band gap values ranging from 4.8 to 5.0 eV. Other work has reported much lower band gap values. Theoretical calculations by Jezierski, 1998, indicated the indirect band gap energy at 2.31 eV and El-Gohary et al., 2003, reported indirect gap at 2.43 eV and direct gap at
2.84 eV from transmission/reflection measurements. The wide variations in the reported
1/2 2 Figure 3-9: α and α as a function of photon energy for SrLaAlO4 to obtain the indirect and direct gaps in the ordinary direction. Reprinted with permission from Adhikari et al., 2016 (© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim).
50 band gap energy may be due to the defects present and the distortions in the unit cell of
SrLaAlO4 as well as the range in magnitude of α used for linear extrapolation. The indirect and direct gaps reported here are consistent with the lower and higher energy values reported in the literature. The low magnitude of absorption between the indirect gap and the first direct critical point transition may have also made it difficult to identify the gap in other measurements.
Ordinary and extra-ordinary spectra in ε obtained by numerical inversion over the full range from 0.74 to 5.90 eV are fit to parametric models. To describe the shape of absorption onset in 2, the imaginary part of CPPB oscillators (Aspnes, 1980) have been modified by a band edge function Gc(E) (Ferlauto et al., 2002) with the result given as,
0 , 퐸 < 퐸푔 2 휀2 = { (3.5) 퐺푐(퐸)Im ∑ 휀퐶푃푃퐵(퐸), 퐸 ≥ 퐸푔 푛
Equation 3.5 is a modification of the form found in Ferlauto et al., 2002 as here the contribution from an Urbach tail is ignored. Sensitivity to the Urbach tail in this material is not present with the reflection based measurement techniques described here.
In this analysis, one dimensional critical points (CP) are assumed with = 0.5. The dimensionality of CPPB oscillators are obtained by fitting second derivative of numerically
2 2 inverted ε2 (d ε2 /dE ) in the ordinary direction for all possible combination of dimensions of CPs. The lower value of σ and better quality of fit are obtained with one dimensional
CPs. The band edge function Gc(E) in Equation 3.5 is defined as (Ferlauto et al., 2002),
2 (퐸 − 퐸푔) 퐺푐(퐸) = 2 (3.6) 2 (퐸 − 퐸푔) + 퐸푝
51 In this expression, Ep defines a transition energy, Ep + Eg. Above this energy, higher energy critical point behavior, reflected in the form of the CPPB oscillator, in 2 is retained such that Gc(E) →1 when E >> Ep + Eg. In the vicinity of band gap, Gc(E) →0 as E − Eg → 0.
The real part of , represented by 1 is obtained from the sum of Sellmeier oscillators, a constant additive term to 1 represented by , and Kramers-Kronig integration (Ferlauto et al., 2002) of 2,
2 퐴 퐸 퐴 퐸 2 ∞ 휉휀 (휉) 휀 = 휀 + ( 1 1 + 2 2 ) + 푃 ∫ 2 푑휉 (3.7) 1 ∞ 휋 퐸2 − 퐸2 퐸2 − 퐸2 휋 휉2 − 퐸2 1 2 퐸푔 where P is the Cauchy principle part of the integral. The resonance energy position of one
Sellmeier is fixed at 0.001 eV, a value well outside the measured spectral range.
Figure 3-10 shows a comparison of the parameterized ordinary and extra-ordinary spectra in , and Tables 3.1, 3.2 and 3.3 contain the parameters describing each set of .
Critical point energies and phase projection factors obtained from fitting numerically inverted ordinary spectra in have been fixed in the parametric fit of the numerically inverted extra-ordinary spectra in . This parametric model gives some basic interpretation of features in , and direct transitions have been observed at 5.54 and 6.01 eV, although the precise location of the higher energy is questionable as it resides outside the spectral range measured here. Eg is fixed at 2.74 eV, the indirect gap obtained from extrapolation
1/2 of α . The value of Ep has been constrained to the difference between the indirect gap energy and the lowest energy critical point, 5.54 eV in this case. Overall, the spectra show different amplitudes and broadenings of the direct transition features in the ordinary and extra- ordinary directions. The transition strength of feature at 5.54 eV is significantly
52
Figure 3-10: Ordinary (black solid line) and extra-ordinary (black dotted and red solid line) spectra in of SrLaAlO4 obtained by parameterization of numerical inverted in Figures 3-5 and 3-7. Reprinted with permission from Adhikari et al., 2016 (© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim).
suppressed in extra-ordinary direction as compared to the ordinary direction. Parametric models consisting of one and two CPPB oscillators have been developed separately and compared for the extra-ordinary direction. This analysis shows the model with two CPPB oscillators resulted in a slightly better fit and lower value of σ and is used to calculate birefringence and dichroism.
Figure 3-11 shows the birefringence (ne−no) and dichroism (ke−ko) in the ordinary
(no, ko) and extra-ordinary (ne, ke) indices of refraction and extinction coefficients.
53 Table 3.1: Parameters describing spectra of for SrLaAlO4 in ordinary direction obtained by using Equations 3.5 and 3.7. Two zero-broadened Sellemeier oscillators with A1 = 79 ± 4 eV, E1 = 0.001 eV, A2 = 15.2 ± 0.2 eV, and E2 -3 = 7.41 ± 0.03 eV are included with = 1.93 ± 0.02 and σ = 8.70 x 10 .
Parameter E1 E2
En (eV) 5.54 ± 0.01 6.01 ± 0.06 A (eV-1/2) 3.7 ± 0.1 8.6 ± 0.2 (eV) 0.37 ± 0.02 1.93 ± 0.08 ϕ (o) −285 ± 5 −8.3 ± 0.5
Table 3.2: Parameters describing spectra of for SrLaAlO4 in extra ordinary direction obtained with one CPPB oscillator by using Equations 3.5 and 3.7. Two zero-broadened Sellemeier oscillators with A1 = 0 eV, E1 = 0.001 eV, A2 = 23.8 ± 0.8 eV, and E2 = 7.66 ± 0.05 eV are included with = 1.28 ± 0.05 and σ = 3.98 x 10-2. The transition energy and phase projection factor are fixed to those of the ordinary direction.
Parameter E1 E2 A (eV−1/2) - 12.3 ± 0.1 (eV) - 1.99 ± 0.02
Table 3.3: Parameters describing spectra of for SrLaAlO4 in extra ordinary direction obtained with two CPPB oscillators by using Equations 3.5 and 3.7. Two zero-broadened Sellemeier oscillators with A1 = 0 eV, E1 = 0.001 eV, A2 = 22.0 ± 0.6 eV, and E2 = 7.53 ± 0.04 eV are included with = 1.40 ± 0.04 and σ = 3.08 x 10-2. The transition energy and phase projection factor are fixed to those of the ordinary direction.
Parameter E1 E2 A (eV−1/2) 2.3 ± 0.3 11.3 ± 0.2 (eV) 0.33 ± 0.06 1.57 ± 0.09
SrLaAlO4 exhibits a positive birefringence over the 0.74 to 5.90 eV spectral range. We have observed that the birefringence remains relatively constant near ~ 0.025 for the low photon energy range from 0.74 to ~ 3.5 eV, and then varies significantly as the material
54
Figure 3-11: Birefringence (ne − no) and dichroism (ke − ko) of SrLaAlO4. The triangle data point is the reported value of 0.02 at 1.96 eV from Hora et al., 1996. Reprinted with permission from Adhikari et al., 2016 (© 2016 WILEY- VCH Verlag GmbH & Co. KGaA, Weinheim).
becomes highly absorbing as reflected in the non-zero values of the dichroism. Hora et. al.,
1996 reported birefringence of 0.020 (ne = 1.979 and no = 1.959) for this material at 1.96 eV as obtained by null ellipsometry. This reported value is in close agreement with 0.019
(with ne = 1.993 and no = 1.974) at that energy reported here. The differences in the values of ne and no with previously reported values may be due to variations in sample purity and the quality of the sample surface.
55 3.4 Summary
The ordinary and extra-ordinary optical properties for single crystal LaSrAlO4 have been determined over the spectral range from 0.74 to 5.90 eV from analysis of ellipsometric and Mueller matrix spectra. The birefringence obtained in this work of about ~ 0.025 across the transparent region is in good agreement with the previously reported values in Hora et al., 1996 and Ryba-Romanowski et al., 1995. The indirect band gap has been identified at
2.74 ± 0.01 eV. However, the significant difference between the high and low energy intercepts of 1/2 = 0 may imply that additional absorption processes due to defects may be present. Therefore, the higher energy intercept at 3.01 ± 0.01 eV may be considered an upper limit for the indirect band gap here. Parametric models for ordinary and extra- ordinary spectra in have been developed. Direct transitions in have been identified at
5.54 ± 0.01 and 6.01 ± 0.06 eV with the 5.54 eV feature in the extra-ordinary direction exhibiting reduced amplitude and transition strength. This work shows an example of basic procedure to obtain the anisotropic optical properties and associated parameters using
Mueller matrix ellipsometry as well as development of a parametric expression for spectra in of indirect band gap crystalline semiconductors.
56 Chapter 4
Growth Evolution and Analysis of Amorphous to Nanocrystalline Transition in Sputtered Si:H Films
Hydrogenated silicon (Si:H) thin films are of use in photovoltaics as current- generating layers in amorphous (a-Si:H) and nanocrystalline (nc-Si:H) based thin film single junction and multijunction solar cells as well as for a-Si:H passivation layers in crystalline Si heterojunction (HIT) solar cells. Although Si:H used as layers in each of these devices is made by similar methods, structural nuances including transitions from amorphous to nanocrystalline characteristics and hydrogen bonding configuration ultimately impact the final solar cell device performance. In this paper we report how reactive magnetron sputtering process parameters can be used to manipulate the optical properties, structure, and evolution of Si:H thin films and the suitability of these materials for photovoltaic application is assessed. The results presented in this Chapter have been published in Adhikari et al., 2017a & Adhikari et al., 2017b and reprinted with permission from AIP publishing, © 2017 and IEEE.
4.1 Introduction and Motivation
a-Si:H and nc-Si:H are used in solar cells largely as a result of the Earth’s abundance of Si and cost effective, scalable deposition techniques. Both a-Si:H and nc-
57 Si:H are useful in single or multi-junction thin film photovoltaics (PV), while a-Si:H is also used as a passivation layer in single- or multi-crystalline Si heterojunction (“HIT”) cells
(Fujiwara and Kondo, 2007; Tanaka et al., 1992)). Incorporation of hydrogen into the a-Si matrix and at crystallite grain boundaries passivates dangling bonds and defect states to enhance the optoelectronic properties of the material and final device (Deng & Schiff,
2003; Yang & Guha, 1999). Thin film Si:H can be deposited using a variety of techniques including radio frequency (RF) plasma enhanced chemical vapor deposition (PECVD), very high frequency (VHF) PECVD, hot-wire gas dissociation, and others. The literature describing some of these techniques, PECVD in particular, is extensive (Fujiwara &
Kondo, 2007; Karki Gautam et al., 2016; Collins et al., 2003; Pearce et al., 2007; Koh et al., 1998b; Ferlauto et al., 2004; Wronski & Collins, 2004; Xu et al., 2011; Koh et al.,
2000; Podraza et al., 2006). In this study, we deposit thin Si:H films using RF magnetron sputtering of an undoped Si target in a mixed argon and hydrogen (Ar + H2) gas ambient.
Although existing literature describing RF-sputtered thin film Si is comparatively less comprehensive than other deposition techniques, sputtering potentially offers significant advantages (Moustakas, 1984; Dutta et al., 2008; Tiedje et al., 1981). Sputtering provides the opportunity to avoid the use of hazardous precursor gases (i.e. SiH4, Si2H6) associated with PECVD and other gas-phase dissociation techniques while maintaining effective control over many microstructural properties of resultant films including crystallinity, hydrogen incorporation, and surface morphology, all of which are of interest in the various applications of thin Si:H films.
The goal of this study is to investigate the phase evolution of sputtered Si:H films and to produce a-Si:H and nc-Si:H material at reasonable deposition rates, > 1 Å/s, for
58 possible future evaluation in PV devices. We employ in-situ, real time spectroscopic ellipsometry (RTSE) measurements of the growing film to non-destructively characterize optical properties and evolution of nanostructure in the material during deposition (Collins et al., 1991). Moreover, RTSE is especially useful when material grows with structural inhomogeneity such as crystallite nucleation from an amorphous matrix. Here, we use
RTSE to track the influence of H2 partial pressure or gas concentration on the amorphous roughening [(a → a)], amorphous to mixed-phase amorphous + nanocrystalline [a → (a + nc)], and mixed-phase to single-phase nanocrystalline [(a + nc) → nc] structural transitions.
The schematic showing formation of nc-Si:H from a-Si:H via mixed (a+nc)-Si:H is shown in the Figure 4-1. Although these transitions occurring during PECVD of thin film Si:H are already well characterized (Karki Gautam et al., 2016; Collins et al., 2003; Pearce et al., 2007; Koh et al., 1998b; Ferlauto et al., 2004; Wronski & Collins, 2004), RF sputtered
Si:H has not received the same attention and comparatively less is known about microstructural growth processes.
Ex-situ Raman spectroscopy and XRD measurements are commonly used experimental techniques to investigate the crystalline volume fraction in the material.
Figure 4-1: Schematic showing formation of nc-Si:H from a-Si:H via mixed (a+nc)- Si:H.
59 These techniques differ from RTSE in that the XRD measurement averages information over the full depth of a thin film sample, whereas Raman spectroscopy averages information from finite depth of sample. In some cases, ex-situ Raman spectroscopy may yield a significantly different result compared to XRD measurement (Houben et al., 1998).
Presently, there is growing interest in in-situ measurement including spectroscopic ellipsometry (Karki Gautam et al., 2016; Collins et al., 2003; Pearce et al., 2007; Koh et al., 1998b; Ferlauto et al., 2004; Wronski & Collins, 2004) and Raman spectroscopy
(Muthmann et al., 2011) to track the variation in crystallinity as a function of depth throughout a full film. We have collected XRD measurements of Si:H films fabricated under varying conditions to correlate crystallographic properties with RTSE results.
Complementary microstructural properties of Si:H films, relative to those from in-situ
RTSE, have been studied using XRD (Xu et al., 2011; Dutta et al., 2008; Wang et al.,
2011; Moustakas et al., 1985), which reveals several peaks corresponding to different Si crystallite orientations.
In previous studies, infrared (IR) spectroscopy (Xu et al., 2011; Wang et al., 2011;
Moustakas et al., 1985; Brodsky et al., 1977; Langford et al., 1992; Freeman & Paul, 1978;
Lucovsky et al., 1979; Smets et al., 2003; Smets & van de Sanden, 2007; Smets et al.,
2008; Stuckelberger et al., 2013; Stuckelberger et al., 2017; Melskens et al., 2013;
Melskens et al., 2012; Melskens et al., 2017) has been used to investigate the Si-Hn bonding configurations in Si:H. Observation shows that Si:H may contain several IR absorption features including stretching modes at 2000, 2090, and 2120 cm-1; bending modes at 850 and 890 cm-1; and wagging or rocking modes at 640 and 590 cm-1, respectively, with the distribution of these modes often depending on the method of
60 fabrication (Brodsky et al., 1977; Paul & Anderson, 1981). IR spectroscopic ellipsometry is used in this study, providing a method of identifying these IR vibrational modes in each film and inferring relative information about interrelationships between hydrogen incorporation and nanostructure when compared to RTSE and XRD results.
4.2. Film Deposition and Measurement Details
Si:H films have been deposited onto native oxide coated single crystal Si wafer substrates at 200°C using RF (13.56 MHz) magnetron sputtering of a 3-inch diameter undoped Si target with a target-to-substrate separation of 13.5 cm. H2 to total gas flow ratio
(pH2 = [H2]/{[H2] + [Ar]} × 100%) is maintained constant during each deposition. Two series of Si:H films have been deposited, one under conditions yielding a low deposition rate (~ 0.15 Å/s) and another with conditions resulting in a higher deposition rate (~ 1.3
Å/s). In each series pH2 was varied. The low rate series has been deposited at a total gas pressure ptot = 30 mTorr and RF power = 100 W and the high rate films at ptot = 10 mTorr and RF power = 250 W.
In-situ RTSE measurements are performed using a dual rotating compensator multichannel ellipsometer (Model RC2-XI, J. A. Woollam Co., Inc.) (Chen et al., 2004a) over the near infrared to ultraviolet spectral range of 0.74 - 5.90 eV at 70° angle of incidence. Real time ellipsometric spectra are collected at 4.18 s intervals during deposition. IR-range ellipsometric spectra are collected ex-situ by a Fourier transform, rotating compensator IR ellipsometer (Model IR-VASE, J. A. Woollam Co., Inc.) (Johs et al., 1999) over a spectral range of 0.05 to 0.62 eV (400 - 5000 cm-1). As has been demonstrated previously for PECVD Si:H, the optical response in the form of the spectra in the complex dielectric function, = 1 + i2, and microstructural properties are extracted
61 using a global ∑σ(t)-minimization procedure with an unweighted error function (Collins et al., 2003). An optical/structural model as shown in Figure 4-2, consisting of ambient / surface roughness / bulk thin film Si:H / interfacial layer / native oxide / semi-infinite single crystal Si is used to fit both RTSE and FTIR-extended measurements. The optical response of surface roughness and interfacial layer are described by a Bruggeman effective medium approximation (EMA) (Fujiwara et al., 2000a; Adhikari et al., 2016) consisting of two material components with optical properties εmat and εvoid, and defined by
휀mat − ε 휀void − ε ( ) 푓푚푎푡 + ( ) 푓푣표푖푑 = 0 (4.1) 휀mat + 2ε 휀void + 2ε where fmat and fvoid are the volume fractions of material and void, respectively, each fixed at 0.5 here. Each individual measurement has 1067 complex (Ѱ, Δ) pairs at different photon energies and in order to extract a common set of spectra in for the early stages of film growth 5 sets of spectroscopic measurements, selected from before any a→(a+nc) transitions are observed, are analyzed simultaneously. The structural parameters consist of
Figure 4-2: Schematic of layered optical/structural model fit to RTSE measurements for Si:H film deposited on native oxide coated c-Si wafer as a substrate.
62 bulk and surface roughness layer thickness for each experimental set of spectra along with an additional nucleation layer thickness fit to a common value for all 5 sets of spectra in the analysis. Spectra in are deduced at each photon energy. This type of procedure is described in Refs. (Karki Gautam et al., 2016; Collins et al., 2003; Pearce et al., 2007; Koh et al., 1998a; Koh et al., 1998b; Ferlauto et al., 2004; Wronski & Collins, 2004; Koh et al.,
2000; Podraza et al., 2006; Fujiwara et al., 2000b; Ferlauto et al., 2002; Fujiwara et al.,
2002). The surface roughness thickness can vary significantly over the deposition, depending on the structural evolution. A void-rich interfacial or nucleation layer has been previously observed in the early stages of growth for a-Si:H and nc-Si:H (Podraza et al.,
2009). The interfacial layer thickness ranges from 0 to 54.1 ± 0.4 Å for all films studied here, and remains static once a bulk film layer develops. GIXRD measurements are performed with a Rigaku/Altima-III X-ray diffractometer using Cu-Kα radiation (λ =
1.54059 Å) and provide complementary information about the overall crystallographic structure.
4.3 Virtual Interface Analysis (VIA) and Growth Evolution Diagram
As has been observed for PECVD Si:H (Karki Gautam et al., 2016; Collins et al.,
2003; Koh et al., 2000), RF sputtered Si:H films can exhibit structural transitions at different thicknesses during deposition. In particular, we identify the thicknesses of a→(a+nc) and (a+nc)→nc structural transitions corresponding to the nucleation of crystallites from the amorphous phase [a→(a+nc)] followed by their coalescence
[(a+nc)→nc] as well as the a→a roughening transition corresponding to a transition from a stable surface with unchanging roughness to a surface that is continuously roughening
63 within the amorphous growth regime. These structural transitions can be identified directly from their effects on the surface roughness thickness (ds) and spectra in as a function of bulk film thickness (db) during Si:H growth as obtained by fitting RTSE measurements as previously described. Optically determined surface roughness has been previously shown to correlate to values obtained from atomic force microcopy (Fujiwara et al., 2000a).
Figure 4-3 shows examples of ds evolution as a function of db for two different films fabricated among the low rate series. The film prepared at pH2 = 20% undergoes characteristic a→(a+nc) and (a+nc)→nc transitions at db corresponding to the abrupt increase in ds and at the maximum in ds, respectively. By contrast, the film prepared at pH2
= 15% features a smooth stable surface which is followed by slight roughening dictated by
Figure 4-3: Variation of surface roughness thickness (ds) with bulk layer thickness (db) obtained from analysis of real time spectroscopic ellipsometry (RTSE) data collected for two Si:H films prepared at RF power = 100W and total gas pressure = 30 mTorr with relative hydrogen gas concentration pH2 = [H2] / {[H2] + [Ar]} × 100% = 15 and 20%. Reprinted from Adhikari et al., 2017a with the permission from AIP Publishing.
64 atomic scale self-shadowing involved in the a→a transition (Podraza et al., 2006; Kryukov et al., 2009). Both a→a and a→(a+nc) transition thicknesses are identified as the db value at 1 Å increase in ds above the minimum ds thickness, however the a→(a+nc) transition is differentiated from the a→a transition in that the quality of fit is significantly poorer in the mixed phase regime as a result of the continued use of optical properties only describing a-Si:H and not a (a+nc)-Si:H mixture.
Figure 4-4 shows the growth evolution diagrams of sputtered Si:H prepared at substrate temperature of 200°C for both low and high deposition rates. The a→a, a→(a+nc), and (a+nc)→nc structural transitions observed in Figure 4-4 are qualitatively similar to those observed in PECVD Si:H (Collins et al., 2003). The films remain amorphous at low values of pH2, whereas nanocrystallites nucleate from the amorphous phase at relatively moderate to high pH2. In the case of both film series, the thicknesses corresponding to the a→(a+nc) and (a+nc)→nc transitions shift to lower bulk film thickness with increasing pH2, demonstrating that higher hydrogen gas concentrations create more favorable conditions for nanocrystallite nucleation and growth. These trends are qualitatively similar to those observed in PECVD where crystallite nucleation begins to occur at certain hydrogen-to-silicon-carrying-gas flow ratios and the thickness at which these transitions appear to shift to lower thicknesses with increased hydrogen present.
A virtual interface analysis (VIA) (Pearce et al., 2007; Koh et al., 1998b) is applied to RTSE data to extract a depth profile of nanocrystallite volume fraction in the mixed- phase regime. This is accomplished by treating the outermost ~ 50 Å of material as a discrete layer described by a Bruggeman EMA consisting of variable fractions of a-Si:H and nc-Si:H. Spectra in ε describing a-Si:H are obtained from fitting to RTSE
65
Figure 4-4: Growth evolution diagrams of Si:H obtained from RTSE depicting thicknesses at which the amorphous-to-amorphous roughening [a→a] (solid circles), amorphous to mixed-phase amorphous+nanocrystalline [a→(a+nc)] (open circles), and mixed-phase to single-phase nanocrystalline [(a+nc)→nc] (solid squares) transitions occur for films prepared at (a) low and (b) high deposition rates as a function of pH2. The upward arrows indicate that the associated transition had not yet been reached at the maximum accumulated thickness of the film measured. The dashed lines are included as a rough guide to the eye representing an approximate boundary between two regimes. Reprinted from Adhikari et al., 2017a with the permission from AIP Publishing.
66 measurements at db ≈ 200 Å, i.e. before nanocrystallite nucleation. The optical properties that best represent the nc-Si:H in the spectral range from 2.5 to 5 eV are obtained by fitting to RTSE measurements at the end of the deposition where the film transformed to single phase nanocrystalline material using the multi-step procedure described in Ferlauto et al.,
2004. The procedure begins by setting a trial value for the roughness layer thickness, generating the optical properties of pseudo-substrate, and using it as semi-infinite substrate with a nc-Si:H bulk over layer and the surface roughness. This step is repeated for various reasonable values of surface roughness thicknesses to extract spectra in of the nc-Si:H bulk layer. Finally, the extracted optical properties of the bulk layer are used in a three component EMA (a-Si:H, nc-Si:H, and void) along with ∑σ-minimization approach within
VIA to obtain the lowest error in the fitting for the mixed-phase regime. This minimum
Figure 4-5: Complex dielectric function, ε = ε1+iε2, spectra for a-Si:H and nc-Si:H used as reference optical properties in the virtual interface analysis (VIA). Spectra in for a-Si:H and nc-Si:H are obtained at different thicknesses from the film produced at RF power = 250 W, ptot = 10 mTorr, and pH2 = 80%. Reprinted from Adhikari et al., 2017a with the permission from AIP Publishing.
67 error identifies the correct surface roughness thickness and from it the most realistic optical properties of nc-Si:H. Figure 4-5 depicts a comparison of a-Si:H and nc-Si:H reference spectra in ε of a high deposition rate film with pH2 = 80%, used within VIA of RTSE data.
The optical response of nc-Si:H is characterized by the presence of two critical point features reminiscent of single crystal Si while a-Si:H has only one broad feature (Ferlauto et al., 2004). The depth profile of nanocrystallite fraction (fnc) in the mixed-phase regime obtained from a representative VIA of a high deposition rate film with pH2 = 80% is shown in Figure 4-6. The correspondence between the evolution of both ds and fnc seen in Figure
4-6 provides justification for the previously described technique of using ds as a function of db to identify phase transition thicknesses as has also been applied previously for
PECVD Si:H materials.
Figure 4-6: Surface roughness and nanocrystallite fraction depth profile in the mixed- phase (a+nc) regime obtained from VIA of RTSE data for a high deposition rate film prepared at pH2 = 80%.
68 4.4 Deposition Rate of Sputtered Si:H
Deposition rates are determined from the analysis of RTSE data using ε obtained from fitting to measurements corresponding to db ≈ 200 Å. A linear relationship is used to fit db as a function of deposition time, with the slope yielding the deposition rate of each sample studied. Figure 4-7 depicts the rates of deposition of Si:H samples for the low and high deposition rate series as a function of pH2. In general, increasing pressure typically decreases sputter deposition rates by increasing the likelihood that sputtered material is scattered away from the substrate by ambient Ar gas. By contrast, increases in RF power will increase overall deposition rates mainly by increasing the flux of material sputtered from the target. When RF power is increased, and total pressure decreased simultaneously, these two effects result in the significantly different deposition rates observed for each film series. As is seen in Figure 4-7, films with RF power of 100 W and total pressure of 30
Figure 4-7: Deposition rates as a function of pH2 for low (black squares) and high (red circles) rate sputtered Si:H. Reprinted from Adhikari et al., 2017a with the permission from AIP Publishing.
69 mTorr result in deposition rates ~ 0.15 Å/s compared to order of magnitude larger deposition rates of ~1.3 Å/s for films prepared with RF power of 250 W and total pressure of 10 mTorr. For films remaining in the amorphous phase, the deposition rate is relatively stable for both low and high rate series. This rate remains stable for slightly higher pH2 films that nucleate crystallites and evolve, however further increases in pH2 eventually result in a reduction in the growth rate. This reduction is seen in both series when crystallite nucleation occurs at < 150 Å and coalescence at < 400 Å. This behavior suggests that under higher pH2 conditions, hydrogen may not be as effectively incorporated into the crystallites compared to a-Si:H and grain boundary material produced at relatively lower pH2. The lower deposition rate observed for high pH2 after nanocrystallite coalescence may be due to etching of weakly bound material on grain boundaries by additional unincorporated hydrogen. PECVD Si:H also shows a similar trend, however decreasing growth rates are observed with increasing hydrogen dilution throughout the amorphous, mixed-phase, and nanocrystalline growth regimes (Karki Gautam et al., 2016).
4.5 X-ray Diffraction Study
Figure 4-8a shows the GIXRD patterns of films produced under high deposition rate conditions. Similar GIXRD patterns have been observed for low rate films as shown in Figure 4-9a. The (111) and (220) peaks of Si are observed at 2θ ≈ 28.4° and 47.3°, respectively, for films with significant nanocrystallite fractions grown at higher values of pH2. As has already been demonstrated in the RTSE results, the films remain amorphous for lower pH2 and nucleate nanocrystallites when pH2 is ≥ 17.5% and ≥ 75% for low and
70 high rate films, respectively. The broad feature observed at about 2θ ≈ 28.4° for samples with pH2 ≤ 15% (low rate) and pH2 ≤ 70% (high rate) is similar to that reported for a-Si:H elsewhere (Amrani et al., 2012). The XRD peaks change from broad, low amplitude peaks
(or none at all) to sharper, narrow peaks with increasing pH2 supporting the RTSE determined phase transformations depicted in Figures 4-4a and 4-4b. The significantly larger amplitude of the peak at 28.4° when compared to that at 47.3° indicates preferential crystallographic orientation along the (111) direction. The broadening of both diffraction peaks decreases, and the amplitude increases with increasing pH2. This behavior indicates
Figure 4-8: (a) Grazing incidence x-ray diffraction (GIXRD) patterns of high deposition rate Si:H prepared at 250 W RF power and 10 mTorr total gas pressure with varying pH2 as indicated. (b) Nanocrystalline grain size determined from GIXRD patterns as a function of pH2. Reprinted from Adhikari et al., 2017a with the permission from AIP Publishing.
71 that the size and fraction of nanocrystallites increases with increasing hydrogen concentration during deposition. The crystallite size (d) has been obtained using Lorentzian fitting of (111) peaks and the Scherrer equation, given by (Scherrer, 1918)
푘휆 푑 = (4.2) 훽푐표푠휃 where the shape factor k = 0.9, λ is the x-ray wavelength (in this case 1.54059 Å for Cuα),
β is the diffraction peak full width half maximum, and θ is the angle of diffraction. Figure
Figure 4-9: (a) Grazing incidence x-ray diffraction (GIXRD) patterns of low deposition rate Si:H prepared at 100 W RF power and 30 mTorr total gas pressure with varying pH2 as indicated. (b) Nanocrystalline grain size determined from GIXRD patterns as a function of pH2.
72 4-8b and 4-9b show the nanocrystallite grain size as a function of pH2 for the higher and low deposition rate series of Si:H sputtered films. The crystallite size increases with increasing pH2, again indicating that increased hydrogen concentration in the sputter ambient is conducive to improved crystallinity. Previous studies on Si:H films prepared from RF sputtering (Dutta et al., 2008; Moustakas et al., 1985), VHF-PECVD (Wang et al., 2011), and PECVD (Xu et al., 2011; Amrani et al., 2012; Ganguly et al., 1996) observed similar XRD patterns with (111) preferential orientation and comparable nanocrystallite size as obtained here. Observations in the literature also show the change of preferred orientation from (111) to (220) for different fabrication techniques including
RF sputtered (Achiq et al., 1998) and PECVD (Ganguly et al., 1996) material depending on the deposition parameters.
4.6 Infrared Optical Properties
The electronic properties of both a-Si:H and nc-Si:H benefit from hydrogen passivation of defects associated with dangling bonds or grain boundaries (Deng & Schiff,
2003; Karki Gautam et al., 2016; Brodsky et al., 1977). The transition strength of absorption features arising from silicon-hydrogen bonds provide a measure of the relative density of each bond type in the material. Several studies of the silicon-hydrogen (Si-Hn; n
= 1, 2, 3) bonding configurations’ manifestation in IR vibrational modes are used as a guide in the fitting of the IR ellipsometric spectra and interpreting the results in this work (Karki
Gautam et al., 2016; Xu et al., 2011; Wang et al., 2011; Moustakas et al., 1985; Brodsky et al., 1977; Langford et al., 1992; Freeman & Paul, 1978; Lucovsky et al., 1979; Smets et al., 2003; Smets & van de Sanden, 2007; Smets et al., 2008; Stuckelberger et al., 2013;
73 Stuckelberger et al., 2017; Melskens et al., 2013; Melskens et al., 2012; Melskens et al.,
2017). Three groups of vibrational modes corresponding to Si-H bonds are expected in
Si:H, namely: stretching (ωS = 2000, 2090, and 2120 cm-1), bending (ωB = 850 and 890 cm-1), and wagging or rocking (ωw = 640 and ωR = 590 cm-1). The possible different modes of Si-Hn vibrations are shown in the Figure 4-10. IR ellipsometric data are analyzed with a similar structural model to that used in RTSE data analysis with the IR absorption features in ε2 parameterized by Gaussian oscillators (De Meneses et al., 2006) each expressed as:
2 2 2√ln(2)(퐸−퐸 ) 2√ln(2)(퐸+퐸 ) −( 푛 ) −( 푛 ) Γ Γ 휀2(퐸) = 퐴푒 − 퐴푒 (4.3)
where En, A, and Г represent the resonance energy, amplitude, and broadening of each
Gaussian oscillator, respectively. The real part, ε1, is obtained by a Kramers-Kronig integration of ε2 given by (Ferlauto et al., 2002)
2 ∞ 휉휀 (휉) 휀 = 휀 + 푃 ∫ 2 푑휉 (4.4) 1 ∞ 휋 휉2 − 퐸2 퐸푔
where P is the Cauchy principal part of the integral and is a constant additive term to 1.
Figure 4-10: Possible modes of vibrations in Si:H (Reproduced from Brodsky et al, 1977).
74 Among the possible expected IR-absorption modes, we have identified sensitivity to those centered at 590, 640, 2000, and 2090 cm-1 for this series of Si:H samples.
The Fourier transform IR measurements and the analysis are performed for both low and high deposition rate film series. Figure 4-11 shows an example of IR ε spectra
Figure 4-11: Infrared (IR) spectra in ε, showing absorption features in for a high deposition rate Si:H film produced at pH2 = 50%. The inset shows deconvoluted features at 2000 and 2090 cm-1. Reproduced from Adhikari et al., 2017a with the permission from AIP Publishing.
75 - for pH2 = 50% a-Si:H from the high rate series. The absorption feature centered at 590 cm
1 -1 is attributed to the Si-H2 rocking mode, that at 640 cm is attributed to the Si-Hn (n = 1,
2, 3) wagging mode, and the modes at 2000 cm-1 and 2090 cm-1 are attributed to Si-H and
Si-H2 stretching, respectively (Brodsky et al., 1977). For each of these features, the area under the peak can be used as a relative measure of the prevalence of the associated bond in the material (Povolny et al., 2000). The index of refraction, n, of all films (both low and high rate) at 5000 cm-1, where Si:H is non-absorbing, is used to determine the relative density of each film compared to the densest material exhibiting the highest index of refraction, n = 3.1979, identified for Si:H produced at pH2 = 90%. The density of each film
2 -1 relative to the densest film at pH2 = 90% is obtained by using = n at 5000 cm and
Equation 4.1 with fmat = 1-fvoid. The low rate films are observed to have higher void fractions, lower material fraction, than high rate films relative to the densest film as presented in Table 4.1. This value of material fraction is used for the normalization of integrated area under the absorption peaks for both series of samples. Normalization is
Table 4.1: Relative void fractions in low and high deposition rate series of sputtered Si:H films.
Low deposition rate series High deposition rate series
pH2 (%) Void fraction pH2 (%) Void fraction 10 0.009 50 0.061 15 0.090 65 0.053 17.5 0.003 70 0.016 20 0.119 75 0.008 30 0.004 80 0.003 40 0.083 85 0.016 - - 90 0
76
Figure 4-12: Integrated area under the absorption peaks centered at 640, 2000, and 2090 -1 cm as a function of pH2 for the (a) low and (b) high deposition rate film series.
performed to ensure that comparison of relative hydrogen incorporation from transition strengths of absorption features between films is not biased by variations in film density.
The strength of the features related to silicon-hydrogen bonding in the optical response is
not artificially dampened due to the presence of voids in low density material decreasing the overall amplitude of spectra in .
Figure 4-12 shows the integrated area under the absorption peak as a function of pH2 for each of the 640, 2000, and 2090 cm-1 peaks for the high and low rate film series. absorption peaks first decrease then remain stable after the films become predominantly
-1 nanocrystalline at pH2 > 75%. The variations of the features at 2000 and 2090 cm within
77 the a-Si:H regime for the high rate films are generally close to the error limits. The area
-1 under the 2000 cm peak initially increases with increasing pH2, reaches a maximum at the
maximum hydrogen concentration prior to any crystallite nucleation at pH2 = 70%, and then decreases quickly as nanocrystallites form in the film. Since the 2000 and 2090 cm-1 peaks are associated with monohydride (Si-H) and dihydride (Si-H2) bonding configurations, respectively, these results suggest that monohydride bonding is more pronounced in a-Si:H and less so when the films becomes nanocrystalline (Fujiwara et al., 2002).
Similarly, for the low rate films, the area under each of the three peaks at 640, 2000,
-1 and 2090 cm initially increases with increasing pH2, reaches a maximum at the maximum hydrogen concentration prior to or at which point crystallites starts to nucleate at pH2 =
17.5%, and then decreases quickly as the film becomes nanocrystalline. Also for the low rate films, a similar trend in which monohydride bonding, more pronounced in a-Si:H, is suppressed as the material becomes predominantly nc-Si:H. More importantly, the area under all three peaks (640, 2000, and 2090 cm-1) are significantly lower than the values of corresponding peaks for high rate samples.
-1 The absorption peak at 640 cm is used to calculate the hydrogen content, CH (at.
%) in the film using an Equation (4.5) (Brodsky et al., 1977; Langford et al., 1992).
훼(휔) 푑휔 퐶퐻 = [퐴휔 (∫ )] × 100% (4.5) 휔 푁푆푖
19 - where () is the absorption coefficient, Aw is the oscillation strength (A640 = 2.1×10 cm
2 -1 for 640 cm wagging mode), ω is the frequency, and NSi is the atomic density of c-Si
22 -3 (5×10 cm ). Figure 4-13 shows the calculated CH (at. %) as a function of pH2 for both
78
Figure 4-13: Hydrogen content CH (at. %) as a function of pH2 for low (lower horizontal axis) and high (upper horizontal axis) rate Si:H. series and seems reasonable in consideration of the values reported in Langford et al., 1992.
CH decreases at higher pH2 for both series of samples, and the higher rate samples have significantly higher values of CH compared to low rate samples. All low rate samples have
CH < 7%, suggesting that hydrogen is not effectively incorporated into the low rate samples and may lack sufficient hydrogen to passivate defects and dangling bonds.
Predominant Si-H2 in a-Si:H is not desirable for solar cell absorber layer materials; however, both sets of films may still be suitable for grain boundary passivation in nc-Si:H layers or for passivation of crystalline silicon in wafer based devices. Significant features near 2100 cm-1 arising from the grain boundary material in nc-Si:H layers are not necessarily linked to poor performance of a solar cell incorporating an intrinsic nc-Si:H absorber layer (Smets et al., 2008; Remes et al., 1998) and are quite commonly reported in nc-Si:H films (Sekimoto et al., 2014; Fujiwara et al., 2000b; Cardona, 1983; Kroll et al.,
79 1996). Similarly, large amplitude high stretching mode features in the infrared spectrum of thin a-Si:H used as a passivants for crystal silicon have also been observed (Fujiwara &
Kondo, 2007; Macco et al., 2017). From the 640 cm-1 peak in the films studied here, which is associated with all possible wagging modes Si-Hn (n = 1, 2, 3), the hydrogen content relative to the densest film decreases with increasing pH2. This behavior may be due to the fact that grain size increases for higher pH2 films as observed from XRD and there is relatively less grain boundary surface area with fewer sites available for hydrogen incorporation. Previous studies of PECVD Si:H show that hydrogen incorporation increases with increasing hydrogen dilution within the amorphous growth regime.
However, during mixed phase or nanocrystalline growth, hydrogen incorporation in the film decreases with increasing hydrogen dilution (Fujiwara et al., 2002; Achiq et al., 1998).
The hydrogen content in the Si:H film depends strongly on the deposition parameters including temperature, pressure, and source gases (Fujiwara et al., 2002).
4.6 Summary
The growth evolution of Si:H films produced via magnetron sputtering show phase transformations from amorphous to mixed-phase then to single-phase nanocrystalline as the films grow. In studies of the growth evolution diagrams, deposition rates are not observed to decrease at the lowest pH2 at which crystallites nucleate. As a result, similar growth rates are obtained between a-Si:H and nc-Si:H phases near what would be considered the protocrystalline region in PECVD Si:H. From IR spectroscopic ellipsometry measurements and analysis, Si-Hn bonding configurations have been identified with absorption peaks centered at 590, 640, 2000, and 2090 cm-1. Samples with
80 higher pH2 and lower deposition rates after crystallite coalescence exhibit lower relative amounts of hydrogen incorporation into the film, which may be due to the etching of weak
Si-Si bonds on the growing surface by unincorporated reactive hydrogen (Moustakas et al.,
1985). Overall, these results indicate that sputtering can be a useful deposition technique for producing nc-Si:H of reasonable quality as assessed by growth evolution diagrams at rates that are generally comparable to many CVD processes. Additionally, a-Si:H made by sputter deposition may serve as a potential candidate for passivation of crystalline Si without the need for silicon carrying source gases.
81 Chapter 5 n-i-p Solar Cell with Sputtered nc-Si:H Absorber
Amorphous and nanocrystalline hydrogenated silicon (a-Si:H and nc-Si:H) thin films are of use as current-generating layers in thin film single junction and multijunction solar cells. Here we have demonstrated working PV devices produced with RF magnetron sputtered nc-Si:H absorber layers and compare overall device performances to those produced with the more conventional PECVD absorbers. Different absorber layer deposition techniques and atmospheric exposure effects are studied to explain variations in the performance of single junction n-i-p substrate configuration nc-Si:H based solar cells.
The results presented in this Chapter have been published in Materials, MDPI (Adhikari et al., 2019).
5.1 Introduction and Motivation
Hydrogenated silicon (Si:H) is an important and widely studied material for thin film photovoltaic (PV) applications because it is non-toxic, inexpensive, and earth abundant. Within the solar irradiance spectrum of interest for most solar cells, hydrogenated amorphous silicon (a-Si:H) shows a higher absorption coefficient compared to indirect band gap nanocrystalline silicon (nc-Si:H) and single crystal silicon (c-Si)
82 absorbers. Comparatively thin (~0.3 µm) a-Si:H, moderately thick (~1.8 µm) nc-Si:H, and much thicker (~200 µm) c-Si absorber layers are required for sufficient absorption of incident solar radiation. Among these two thin film absorbers, a-Si:H is vulnerable to the
Staebler–Wronski effect (Staebler & Wronski, 1997) resulting in degradation under illumination, whereas nc-Si:H suffers from less or no light induced degradation. Several techniques are available for the fabrication of Si:H thin films including plasma enhanced chemical vapor deposition (PECVD: RF and VHF) (Koh et al., 2000; Huang et al., 2015;
Gautam et al., 2016; Podraza et al., 2006; Pearce et al., 2007, Collins et al., 2003; Dahal et al., 2014; Junda et al., 2015), RF magnetron sputtering (Adhikari et al., 2017a; Moustakas et al., 1985; Dutta et al., 2008; Tiedje et al., 1981), hot-wire chemical vapor deposition
(Mahan et al., 1991; Heintze et al., 1996; Matsumura, 1986), and other technologies
(Masuda et al., 2012; Shindo et al., 1984). Among these techniques, Si:H PV devices with
PECVD absorbers are the most widely studied (Koh et al., 2000; Huang et al., 2015;
Gautam et al., 2016; Podraza et al., 2006; Pearce et al., 2007, Collins et al., 2003; Junda et al., 2015), but require toxic silicon carrying precursor gases (SiH4, Si2H6). By contrast, there is comparatively sparse literature describing sputtered Si:H, and nc-Si:H in particular
(Moustakas et al., 1985; Zhao et al., 2004), but the sputtering process from a solid silicon target in reactive hydrogen is completely nontoxic and fairly simple. In addition, sputtering
(Adhikari et al., 2017a) has been shown to provide effective control over many microstructural properties of the resultant films including crystallinity, hydrogen incorporation, and surface morphology, all of which are of interest in the various applications of thin Si:H films in PV and other devices. Sputter deposition also enables p- and n-type dopant atoms to be introduced in silicon targets, eliminating toxic dopant gases
83 needed in PECVD (Rajan et al., 2017). Solid silicon targets have also been used in pulsed laser deposition of Si:H (Hanabusa et al., 1997). Sputtering, however, is a demonstrated industrially scalable process already used for large area depositions needed for applications like PV (Greene, 2017), making it a strong potential alternative to PECVD in manufacturing. Here we have demonstrated working PV devices produced with sputtered nc-Si:H absorber layers and compare overall device performance to those produced with the more conventional PECVD absorbers. Although the PV performance of these devices can be improved, there is potential for development of nc-Si:H without the need for toxic source gases to reduce overall device fabrication cost. We report here one of the only studies of functioning solar cells incorporating sputtered nc-Si:H absorber layers.
5.2 Experimental Details
The general layered structure for the n-i-p configuration devices studied is a soda lime glass (SLG) supporting substrate, Cr adhesion layer, Ag metal back reflector, ZnO transparent conducting oxide diffusion barrier, nc-Si:H n-layer, nc-Si:H i-layer, nc-Si:H p- layer, and indium tin oxide (ITO) transparent conducting front dot contact defining the area of the cells. This structure is depicted schematically in Figure 5-1. The sputtered i-layers of primary interest to this work are prepared in a standalone sputter chamber (AXXISTM,
K. J. Lesker Co.), while all other layers are prepared in a multi-chamber, load locked cluster tool fully under vacuum (MV Systems). Thus, for cells prepared with sputtered i-layers, vacuum break and exposure to atmosphere occurs both immediately before and after the i- layer deposition. Three types of these devices are constructed with differently fabricated nc-Si:H i-layers, including (1) sputtered nc-Si:H layers, (2) PECVD nc-Si:H i-layers where
84
Figure 5-1: Schematic diagram of nc-Si:H absorber based solar cell in the n-i-p substrate configuration.
the devices were removed from vacuum into laboratory atmosphere both immediately prior to and after the i-layer deposition, and (3) otherwise identical PECVD nc-Si:H i-layers incorporated into devices fully prepared in the cluster tool without leaving vacuum. These three sample configurations provide the opportunity to compare the devices with sputtered i-layers to those deposited with PECVD that have the same vacuum breaks (2) and a control sample fabricated without breaking vacuum (3).
SLG substrates (Pilkington North America) are cleaned with detergent (Micro-90,
International Products Corp.) in a heated ultrasonic bath, rinsed several times with deionized water, and then dried with pure N2 gas in air before thin film deposition. The Cr,
85 Ag, ZnO, and ITO layers are all deposited via RF magnetron sputtering in the cluster tool.
The PECVD nc-Si:H n-, i-, and p-layers are deposited at radio frequency (13.56 MHz) using a H2 diluted SiH4 precursor gas mixture described by the dilution ratio R = [H2] /
[SiH4]. The dopant gas ratios for n- (D = [PH3]/[SiH4]) and p-layers (D = [B2H6]/[SiH4]), which can have considerable influence on the structural and electronic properties, are fixed at D = 0.0125. The chamber base pressure is ~10-7 Torr before each deposition. The substrate temperature, deposition pressure, plasma RF power density, and gas flows are listed for each layer in Table 5.1. It is worth noting that nanocrystallinity arising from these conditions in n-, i-, and p-layers is confirmed via in-situ real-time spectroscopic ellipsometry (RTSE). In-situ RTSE measurements have been performed over the near infrared to ultraviolet spectral range of 0.74–5.90 eV at a 70° angle of incidence using a chamber-mounted dual rotating compensator multichannel ellipsometer (RC2, J.A.
Woollam Co., Inc.) (Chen et al., 2004a). Modeling of the RTSE data allows for the determination of the thickness, surface roughness, and nanocrystalline volume fraction of
Table 5.1: Deposition conditions for the individual layers in the nc-Si:H n-i-p solar cell configuration deposited on 15.24 cm × 15.24 cm soda lime glass (SLG) substrates in the load-lock cluster tool (i.e., with plasma enhanced chemical vapor deposition (PECVD) i-layers). The dopant source gases are each 5% dopant gas in H2 is by volume. “RT” denotes room temperature.
RF Gas Flow (SCCM) Substrate Pressure Layer Power 5% O2 5% PH3 Temp.(°C) (mTorr) 2 Ar SiH4 H2 R = H2/SiH4 (W/cm ) in Ar or B2H6 in H2 Cr RT 15 0.920 10 - - - - - Ag RT 15 0.920 10 - - - - - ZnO RT 5 0.920 10 - - - - - n 200 1500 0.031 - - 2 0.5 PH3 200 100 i 200 1500 0.043 - - 5 - 125 25 p 100 1500 0.086 - - 2 0.5 B2H6 500 250 ITO 150 4 0.582 10 3 - - - -
86 each film. A set of p-, i-, and n-layers are deposited as a function of R and monitored by
RTSE to identify the appropriate R for nanocrystalline phase growth beginning immediately at the start of deposition (Huang et al., 2015; Karki Gautam et al., 2016;
Adhikari et al., 2017a). It has been observed that the highest electronic quality nc-Si:H films are produced with the lowest value of R at which the nanocrystalline phase dominates growth (Cao et al., 2008; Mai et al., 2005). Thus, each film is deposited with the minimum possible R that will still produce nanocrystalline growth from the start of deposition. It is determined that these minimum values of R required for immediate nanocrystalline growth for the n-, i-, and p-layers are R = 100, 125, and 250, respectively. The deposition rate of the PECVD nc-Si:H i-layer deposited on the top of a nc-Si:H n-layer is about 0.6 Å/s.
For the fabrication of sputtered i-layers, an undoped 7.62 cm diameter Si target
(99.999% purity, K.J. Lesker Co.) spaced 13.5 cm from the substrate is sputtered using RF power of 250 W in a controllable H2 + Ar environment with a substrate temperature of
200°C. Sputtering is performed at 10 mTorr gas pressure with a constant hydrogen-to- argon gas flow ratio, pH2 = 100% × [H2]/{[H2] + [Ar]} in upstream mode for each deposition. Previous work (Adhikari et al., 2017a) has successfully optimized these deposition conditions to produce reasonable growth rate material with controllable microstructure as identified by RTSE. Microstructure during sputtering is controlled by pH2 and typical growth rates are approximately 1.3 Å/s.
The n-i-p devices with PECVD i-layers both exposed and not exposed to air are fabricated on 15.24 cm × 15.24 cm glass substrates with 256 dot cells each having an active area of 0.0707 cm2. Devices incorporating sputtered nc-Si:H i-layers deposited with different pH2 (80, 85, and 90%) are fabricated, and results for pH2 = 90% sputtered layers
87 are included here. Because of the complexity of using larger substrates in the stand alone sputtering chamber, n-i-p devices with sputtered nc-Si:H i-layers are fabricated on 5.08 cm
× 5.08 cm glass substrates resulting in 25 dot solar cells. Photocurrent density versus voltage (J-V) characteristics are measured under simulated AM1.5G (100 mW/cm2) illumination from a 450 W Xenon light source (Oriel, Model 9119, Newport) with a digital source meter (Keithley 2440) in air at room temperature. Dark current is also measured using the same system. External quantum efficiency (EQE) measurements are performed
(model IVQE8-C, PV Instruments) to characterize spectrally resolved device performance.
Topography and roughness measurements of the Cr/Ag/ZnO back reflector (BR) structure are also performed using an atomic force microscope (AFM) operated in tapping mode
(Nanoscope V, Veeco). Ex-situ spectroscopic ellipsometry measurements of the BR structure are performed using a single rotating compensator multichannel ellipsometer (J.
A. Woollam M-2000FI) (Lee et al., 1998). EQE spectra of solar cells with i-layers fabricated by PECVD and sputtering are simulated by using complex dielectric function (ε
= ε1 + iε2) spectra and thicknesses of layers that are obtained using the same spectroscopic ellipsometry data analysis procedure explained in Karki Gautam et al., 2016. Spectra in ε for PECVD and sputtered i-layers are represented as Bruggeman effective medium approximations (Fujiwara et al., 2000a) of crystal silicon for simplicity (Herzinger et al.,
1998) and a-Si:H with a band gap of 1.8 eV (Ferlauto et al., 2002). Initially during simulation, it is assumed that each photon absorbed in the i-layer generates an electron– hole pair able to be collected without recombination losses. To account for the incomplete collection of charge carriers generated in the sputtered i-layer, a collection probability profile is introduced into the simulation to identify regions of reduced carrier collection.
88 5.3 Device Fabrication
Before depositing nc-Si:H layers, the sputtered BR structure is prepared. To obtain surface roughness thickness of the BR with and without the ZnO diffusion barrier (Ag deposited on SLG/Cr and ZnO deposited on SLG/Cr/Ag substrate), room temperature ellipsometric spectra (in N = cos 2Ψ, C = sin 2Ψ cosΔ, S = sin 2Ψ sin Δ) have been collected ex-situ at 70° angle of incidence over a spectral range from 0.74 to 5.90 eV. Experimental ellipsometric spectra have been fit to an optical model exactly the same as described in
Karki Gautam et al., 2016. From the analysis of ellipsometric spectra collected from a representative sample, the surface roughness thickness of the optically opaque Ag layer deposited on a SLG/Cr substrate is 10 ± 2 nm. The ZnO layer deposited on SLG/Cr/Ag substrate is found to have bulk layer and surface roughness thicknesses of 332 ± 3 nm and
11.3 ± 0.5 nm, respectively. The surface roughness and topography of both the SLG/Cr/Ag and full SLG/Cr/Ag/ZnO BR structures are also obtained using AFM, with the surface morphology shown in Figure 5-2. Using this technique, the root-mean squared (RMS) roughness of Ag and ZnO surfaces are found to be 6.8 and 9.4 nm, respectively. Previous reports comparing surface roughness from ellipsometry and RMS roughness from AFM have shown that ellipsometry determined roughness is about 1.5 times that from AFM (Koh et al., 1996; Koh et al., 1998a). Here, ellipsometry determined roughness is on average about 1.35 times the RMS value from AFM. These surface roughness thickness values indicate that the BR structures prepared with the processes described here are relatively smooth compared to intentionally textured BRs. In general, increasing the path length of light through the use of a rough, scattering BR is a common route to increasing current
89
Figure 5-2: Three-dimensional atomic force micrographs (AFM) taken over an area of 2µm × 2µm showing surface morphology of (a) Ag deposited on SLG/Cr and (b) ZnO deposited on SLG/Cr/Ag.
generated in and PV conversion efficiency of a-Si:H and nc-Si:H solar cells (Sivec et al.,
2013; Vetterl et al., 2000; Sai et al., 2015; Matsui et al., 2015; Sai et al., 2012). However, for the purposes of assessing the applicability of sputtered nc-Si:H i-layers, a relatively planar BR surface is sufficient for comparison among the three different i-layer fabrication configurations previously described. This configuration also avoids any cracking in nc-
Si:H films sometimes arising from deposition onto textured BRs (Sai et al., 2015; Sai et al., 2012) and will enable the most direct assessment of the electronic device quality of differently prepared intrinsic layers. Analysis of ellipsometric spectra shows similar surface roughness values of 3.3 1.7 nm and 2.7 0.3 nm of the p-layer at the top of the devices fabricated with PECVD and sputtered nc-Si:H i-layers, respectively. Similar light scattering at the top surface of the devices and nc-Si:H growth morphology are expected for devices with both PECVD and sputtered absorbers.
Optimization of thin film PV relies on characterizing the optoelectronic and structural properties of each layer and correlating these properties with device performance. Nanocrystallite growth in Si:H from the amorphous matrix is promoted by
90 using increased hydrogen dilution ratio, R, in PECVD (Karki Gautam et al., 2016) and hydrogen to total gas ratio, pH2, in RF magnetron sputtering (Adhikari et al., 2017a).
Growth evolution diagrams have been developed and used to guide production of materials in nc-Si:H n-i-p devices. For the cells incorporating sputtered nc-Si:H i-layers, the growth evolution diagram in Adhikari et al., 2017a is used as a guideline, to ensure that the material sputtered onto the underlying nc-Si:H n-layer will nucleate nanocrystallites immediately.
Sputtered i-layers on n-type PECVD nc-Si:H with pH2 = 80, 85, and 90% are all within the nanocrystalline growth regime according to the growth evolution diagram and RTSE monitoring of film growth, and separate full devices are fabricated with each of these pH2.
The devices fabricated with nc-Si:H i-layers at pH2 = 90% are found to result in the best performance. The optical response for the nc-Si:H film prepared at pH2 = 90% had the highest amplitude features in spectra in ε indicating the highest optical density among the nc-Si:H films prepared at pH2 = 80, 85, and 90% (Adhikari et al., 2017a). The n-i-p devices incorporating PECVD nc-Si:H i-layers prepared at R = 125, which is the minimum value of R required for immediate nanocrystalline growth for the i-layers deposited on the top of n-type PECVD nc-Si:H, have better device performance.
5.4 Device Characterization and EQE Simulations
Figure 5-3 displays J–V characteristic curves for the highest efficiency solar cell of each of the three different i-layer preparation procedures measured under simulated
AM1.5G illumination. The short circuit current (JSC), open circuit voltage (VOC), fill factor
(FF), and power conversion efficiency (PCE) corresponding to devices plotted in Figure
5-3 are reported in Table 5.2. The device performance parameters presented in Table 5-2
91
Figure 5-3: Current–voltage (J–V) measured under 100 mW/cm2 simulated AM1.5G irradiation and in the dark for the highest efficiency n-i-p devices incorporating a PECVD absorber without vacuum break (black squares, black dotted line), PECVD absorber with vacuum break before and after intrinsic layer deposition (red circles, red dotted line), and RF magnetron sputtered absorber also with vacuum break before and after intrinsic layer deposition (blue triangles, blue dotted line).
are obtained for 1.8 µm thick PECVD absorbers and a 1 µm thick sputtered absorber. The box plot showing the comparison of device performance of all n-i-p nc-Si:H devices with different absorbers are shown in Figure 5-4. Figure 5-4 includes device performance parameters for all measured solar cells. For devices based on sputtered and PECVD i-layers fabricated with air exposure between nc-Si:H layers, substantially increased shunting is observed via the span of cell shunt resistance. This behavior is expected for thin film solar cell devices when vacuum is interrupted during fabrication and corresponds to lower device yield. Spatial non-uniformity of all deposited thin film layers, particularly the nc-Si:H absorber, are also sources of variation in device performance parameters.
92 Table 5.2. Comparison of device performance parameters of highest efficiency n-i-p cells incorporating nc-Si:H absorbers prepared with each absorber layer processing method.
J V FF PCE R R Absorber i-layer preparation SC OC s sh (mA/cm2) (V) (%) (%) (Ω cm2) (Ω cm2) PECVD (no air exposure) 17.7 0.52 64.1 5.91 6.6 573.4 PECVD (air exposure) 18.5 0.50 54.2 5.08 7.3 574.1 Sputtered (air exposure) 6.2 0.32 45.6 0.92 15.4 445.6
As would be expected, the device with PECVD nc-Si:H i-layer fabricated without vacuum break is found to have highest PCE of 5.91%. By comparison, the cell incorporating a PECVD i-layer fabricated with vacuum breaks had PCE = 5.08%, and the device with a sputtered i-layer had 0.92%. For the device prepared with a PECVD nc-Si:H i-layer without breaking vacuum, VOC = 0.520 V and FF = 64.1% are consistent with values
2 typically reported in literature (Sai et al., 2012; Green et al., 2018). JSC = 17.7 mA/cm is relatively low compared to the theoretical maximum based on the band gap of crystalline silicon and is attributed to the use of a planar non-textured back reflector and an absorber only 1.8 m thick. This JSC value is consistent with other similarly designed devices (Sai et al., 2012; Sai et al., 2009; Kim et al., 2014).
The decreased efficiency of the device with the PECVD i-layer exposed to vacuum breaks relative to the device fabricated fully under vacuum is almost certainly due to oxidation and/or contamination of the interfaces on either side of the i-layer resulting from air exposure. The largest decrease in device performance parameters are in the VOC and FF which are reduced by 3% and 16%, respectively. This level of device performance reduction is expected considering a vacuum break during processing. Average JSC of the devices with i-layer exposed to air is lower compared to PECVD devices with absorbers
93
(a) (b)
(c) (d)
(e) (f)
Figure 5-4: Solar cell performance parameters including open circuit voltage VOC (a), fill factor FF (b), short circuit current density JSC (c), power conversion efficiency PCE (d), series resistance Rs (e), and shunt resistance Rsh (f) for devices incorporating nc-Si:H absorbers prepared with different methods.
94 prepared without air exposure (Figure 5-4). JSC of the highest efficiency cell with air exposure is slightly higher compared to corresponding highest efficiency cell without air exposure. This difference is likely the result of absorber layer thickness distribution due to spatial non-uniformity in each deposition and run-to-run variation between separate depositions.
Compared to devices with PECVD i-layers, those with sputtered i-layers have substantially lower device performance parameters. The primary reason for worse performance when incorporating a sputtered absorber is because of lower electronic quality of the sputtered material itself or of interfaces with the n- and p-layers. We can see that vacuum breaks themselves only result in a decrease in efficiency of less than 1% so this does not account for the substantially lower efficiency of devices with sputtered i-layers.
Average VOC, JSC, FF, and PCE of devices with a PECVD nc-Si:H absorber with no air exposure are slightly higher compared to PECVD nc-Si:H absorber with air exposure before and after deposition, while substantially lower mean device parameters are observed for sputtered absorber devices. For n-i-p nc-Si:H devices with sputtered absorber i-layers, higher series resistance (Rs) and lower shunt resistance (Rsh) are observed relative to devices with PECVD absorbers. The higher value of Rs and lower value of Rsh for devices with sputtered absorber results in lower FF and hence overall poor device performance compared to those with PECVD absorbers. The intersection between light and dark J–V curves for the example sputtered i-layer based device is more pronounced than those with
PECVD absorbers. This behavior could arise due to lower electronic quality of sputtered bulk layer, interfaces (n/i and p/i), or both (Ge et al., 2014; Ahmed et al., 2012).
95 Figure 5-5a shows EQE spectra over the range 300–1100 nm for the same three highest efficiency devices incorporating PECVD and RF magnetron sputtered nc-Si:H absorbers. There is little difference between either of the PECVD absorber device, both having nominally 1.8 µm thick absorbers. By contrast, the EQE spectra for a device with a
1µm thick magnetron sputtered nc-Si:H absorber is substantially lower for all measured wavelengths, more than that expected by the simple reduction in intrinsic layer thickness.
JSC calculated from integration of EQE simulated using the optical response of PECVD nc-
2 Si:H predicts a JSC reduction of about 4 mA/cm when the absorber layer thickness is decreased from 1.8 to 1 µm in n-i-p configuration devices. This indicates that the
2 experimentally measured value of JSC = 6.2 mA/cm obtained for the device with the 1 µm thick sputtered absorber is much lower than the expected JSC for a similarly thick absorber made by PECVD. The optical response of sputtered (Adhikari et al., 2017a) and PECVD nc-Si:H (Karki Gautam et al., 2016) are comparable and do not account for this additional discrepancy.
All these results indicate that the predominant limitation in devices with magnetron sputter deposited nc-Si:H absorbers stems from incomplete carrier collection originating from either or both recombination in the intrinsic bulk layer or at the interfaces with n- and p-type doped layers. Problems at the interfaces with the doped layers may arise from incompatibility of the two deposition processes for doped and undoped Si:H, PECVD, and sputtering. More energetic ion bombardment during sputtering of the i-layer on the PECVD n-layer may lead to surface damage. Similarly, the high R = 250 value for p-layer PECVD may lead to hydrogen etching of the underlying sputtered i-layer. The grain boundaries will etch preferentially over the crystalline grains, resulting in void rich and poorly passivated
96 material at the interface. Additionally, depending on grain size and any porosity present between grains, exposure to laboratory air may have a greater impact on sputtered nc-Si:H relative to its PECVD counterpart. The decrease in EQE at all wavelengths indicative of incomplete carrier collection lowers JSC by 66% when comparing the highest efficiency devices with sputtered material to similarly processed PECVD material but with vacuum breaks. This reduction in current generated is greater than that of VOC and FF at 37% and
16%, respectively. As FF may be indicative of bulk i-layer performance (Smets et al.,
2008; Stoke et al., 2008) and VOC of the interfaces, results here suggest that recombination may be taking place predominantly at those interfaces inhibiting current collection.
Simulations of EQE with a reduced carrier collection profile are used to further investigate the most likely sources of loss within the solar cell device incorporating the sputtered i-layer. The simulated EQE without collection losses (blue dashed line in 5-5a) for devices with a 1µm thick sputtered nc-Si:H absorber shows substantially higher values compared to measured EQE (blue solid line in 5-5a). This observed difference between simulated and measured EQE in the solar cell incorporating a sputtered i-layer can be attributed to incomplete carrier collection from portions of the i-layer. Modeling of the i- layer requires accounting for thickness variations in the relative amorphous and crystalline volume fractions and treating the PECVD and sputtered material slightly differently for agreement between simulation and measurement. The optical response of the 0.9 µm of the
PECVD i-layers adjacent to the p-layers is described by those of crystal silicon (Herzinger et al., 1998), and the 0.9 µm adjacent to the n-layers is described by a Bruggeman effective medium approximation (Fujiwara et al., 2000a) of 0.9 volume fraction crystal silicon and
0.1 volume fraction 1.8 eV band gap a-Si:H (Ferlauto et al., 2002). This bilayer approach
97
Figure 5-5: (a) External quantum efficiency (EQE) spectra for two n-i-p devices incorporating PECVD nc-Si:H absorbers, one of which is deposited without breaking vacuum (no air exposure, black solid line) and the other with vacuum breaks before and after intrinsic layer deposition (air exposure, red line) to simulate the processing of the samples with sputtered nc-Si:H absorbers (blue solid line). EQE spectra are simulated for a device with a PECVD i-layer (black dashed line), a device with a sputtered i-layer assuming complete carrier collection (blue dashed line), and a device with a sputtered i-layer with reduced collection probability (green dashed line). (b) Collection probability profile used in EQE simulation accounting for the recombination losses of photogenerated carriers within the 1 µm thick sputtered nc-Si:H i-layer.
98 provides a simplified model for Si:H growth evolution with increasing crystallinity as a function of accumulated film thickness (Karki Gautam et al., 2016; Collins et al., 2003;
Junda et al., 2015; Adhikari et al., 2017a; Stoke et al., 2008). The thinner 1 µm thick sputtered i-layer is described as a bilayer of the same material properties although the crystal silicon component of the i-layer adjacent to the p-layer is 0.1 µm thick, while the effective medium approximation of crystalline silicon and a-Si:H near the n-layer remains the same thickness and volume fraction as for the PECVD i-layers. The sputtered i-layer is divided into ten 0.1 µm thick sublayers to introduce a collection probability profile to further improve agreement between simulated and measured EQE (Ibdah et al., 2018). The collection probability profile is determined by fitting a variable fraction of carriers collected for each sublayer in a least squares regression to match the measured EQE. The resulting carrier collection probability profile as a function of depth from the p/i interface and simulated EQE spectrum incorporating that collection profile are shown in Figure 5-
5a (green dashed line) and b. Some limitations of this approach are the simplification of the collection probability structure itself and the assumed absorber layer optical response which are expected to vary somewhat with deposition conditions (Karki Gautam et al.,
2016; Adhikari et al., 2017a; Yuguchi et al., 2012; Fujiwara et al., 2001; Tong et al., 2012) as well as thickness (Huang et al., 2015; Fujiwara et al., 2001; Tong et al., 2012). A more advanced optical modeling approach accounting for these simplifications would lead to improved agreement between the model and measured spectra, however the improved qualitative agreement already helps to discern the sources of collection losses. Namely, collection of photogenerated carriers decreases after 100 nm depth into the sputtered nc-
Si:H i-layer reaching zero collection beyond 600 nm. Although the overall electronic
99 quality of the sputtered nc-Si:H i-layer is likely lower than that prepared by PECVD, this reduction in collection probability indicates greater collection losses at the n/i interface possibly due to damage by increased ion bombardment of the n-layer during sputtering of the over-deposited i-layer. In contrast, complete carrier collection is observed near the p/i interface indicating that PECVD of the p-layer does not cause as substantial damage to the underlying sputtered i-layer.
It should be noted that, regarding the relatively low efficiency of devices incorporating sputtered nc-Si:H absorbers, optimization of the material has been essentially limited to merely ensuring that the correct nanocrystalline phase is obtained. Variation of sputter deposition parameters would provide the potential to optimize the material further to achieve the best PV device performance. Material optimization principles and excursions in deposition parameter space similar to those explored for PECVD nc-Si:H can be adapted for optimization of sputtered nc-Si:H (Shah et al., 2003). Fabrication of complete devices fully under vacuum would be expected to increase performance as well.
Should incompatibilities exist between over-deposition by sputtering and PECVD, evaluation of sputtered doped layers (Rajan et al., 2017) may also be considered. Thus, these results are best interpreted as a starting point upon which substantial improvements in device performance could reasonably be expected upon further material optimization.
5.5 Summary
Thin film nc-Si:H devices in the n-i-p substrate configuration have been fabricated with different intrinsic Si:H absorbers: standard PECVD nc-Si:H i-layer prepared entirely under vacuum, PECVD nc-Si:H i-layer where the sample is removed from vacuum
100 between the deposition of the underlying n-layer and the overlying p-layer, and sputtered nc-Si:H i-layer with the same vacuum breaks. To our knowledge, this is the first report of a functioning PV device incorporating a sputtered nc-Si:H absorber layer. The devices with sputtered absorbers have low PCE in comparison to their counterparts incorporating
PECVD absorbers. The vacuum breaks associated with the fabrication process of the sputtered-absorber device are identified to account for only ~1% absolute decrease in PCE.
Further decreases in PCE for the sputtered absorber based devices are attributed to recombination losses in both the bulk and at interfaces. The collection probability profile obtained by comparing measured and simulated EQE for the device incorporating the sputtered nc-Si:H i-layer shows reduced carrier collection from partway through the bulk i-layer to the n/i interface indicating damage to the underlying n-layer during over- deposition of the sputtered nc-Si:H i-layer. These losses may be assessed and reduced by ensuring process compatibility and interfacial stability between deposition of the doped and undoped layers and performing all depositions in a load-locked system without unnecessary breaks from vacuum. The sputtered material itself also requires further optimization to reach performance levels comparable to those achievable with PECVD.
101 Chapter 6
Glancing Angle Deposited CdTe: Impact on Solar Cells
Optical, electrical, and microstructural properties of magnetron sputtered CdTe films prepared by glancing angle deposition (GLAD) technique are studied. GLAD is used to engineer different microstructures in thin films, resulting in variation in opto-electronic and micro-structural properties. GLAD magnetron sputtered CdTe thin films prepared at various oblique angle relative to the substrate normal are deposited on soda-lime glass at room temperature. The dependence of crystal structure, crystallite size, morphology, and optical properties of CdTe films on oblique angle are identified using different characterization tools including spectroscopic ellipsometry, x-ray diffraction, and scanning electron microscopy. The effect of post-deposition CdCl2 heat treatment time on the microstructure of GLAD CdTe is also studied. Finally, the impact of introducing a thin nominally 100 nm thick GLAD CdTe interlayer between CdS and CdTe of the standard
CdS/CdTe heterojunction design solar cell are evaluated. The results presented in this
Chapter have been published in Surface and Coatings Technology (Adhikari et al., 2019) and reprinted with permission, © 2019 Elsevier B. V.
102 6.1 Introduction and Motivation
Cadmium telluride (CdTe) is a II-VI compound semiconductor with high optical absorption >104 cm-1 in the visible spectral range and a direct band gap ~1.5 eV, making it an ideal absorber layer material in thin film photovoltaics (PV). CdTe is a stable material under outdoor conditions and can be deposited using various methods including sputtering
(Gupta & Compaan, 2004; Gupta et al., 2006; Koirala et al., 2016; Tuteja et al., 2015;
Paudel et al., 2012), spray pyrolysis (Boone et al., 1982), electrodeposition (Rakhshani,
1997), and close-space sublimation (Wu, 2004; Britt & Ferekides, 1993). Recently, the champion PV device efficiencies reached 22.1% for a thin film CdTe research solar cell and 18.6% for a module, respectively (Green et al., 2019). The best efficiencies of solar cells incorporating sputtered CdTe absorbers are about 14% (Gupta & Compaan, 2004;
Chinese Academy of Science, 2014). Current research in the field is focused on making highly efficient devices by alternative deposition techniques for better quality CdTe absorbers, improving back contact performance and compatibility, and modification of p/n interfaces (Dobson et al., 2000) to enhance photogenerated carrier collection.
The microstructure of CdTe films depends on deposition conditions and plays an important role in PV device performance, which depends upon the characteristics of both crystalline grains and surrounding grain boundary material. Previous reports indicate that depleted grain boundaries are beneficial to CdTe solar cell performance by helping to separate carriers, suppress recombination, and improve carrier collection (Tuteja et al.,
2015; Li et al., 2014; Visoly-Fisher et al., 2004). A CdS window layer is a commonly used n-type heterojunction partner with a p-type CdTe absorber. Cubic zinc blende CdTe is thermodynamically stable at room temperature, but exhibits a high lattice mismatch with
103 room temperature stable hexagonal wurtzite CdS (Dhere & Dhere, 2005). However, hexagonal wurtzite and mixed zinc blende + wurtzite CdTe films have been deposited at low temperature by vacuum evaporation and sputtering (Dhere & Dhere, 2005;
McCandless & Sites, 2003; Ehsani et al., 2013).
Glancing angle deposition (GLAD) produces films with varying preferential grain orientations and grain boundary configurations (Adhikari et al., 2018). GLAD involves separating the source flux and substrate target normal by an oblique angle. The dependence of crystal structure, grain size, grain orientation, and optical properties on oblique sputtering angle are identified. Previous works (Ehsani et al., 2013; Adhikari et al., 2018) show that opto-electronic and microstructural properties of the deposited film can be effectively manipulated using GLAD. The microstructure of CdTe films can potentially be engineered, resulting in the ability to tune opto-electronic properties for PV device optimization, such as by tailoring material structure at the p/n interface. This work involves study of such CdTe films deposited via radio frequency (RF) magnetron sputtering with material flux at a range of angles relative to the substrate normal and evaluation of PV device performance by incorporating GLAD CdTe. In particular, the influence of deposition angle on the microstructural and optical properties of sputtered CdTe film are studied and improvement of PV device performance by incorporating a 100 nm GLAD interlayer at the p/n junction is demonstrated. The crystal structure and the crystallite size are obtained from x-ray diffraction (XRD), cross-section and surface microstructures are imaged from scanning electron microscopy (SEM), and complex optical properties are extracted from spectroscopic ellipsometry. The oblique angle of deposition manipulates crystal structure, grain size, grain orientation, and optical properties. The effect of CdCl2
104 post-deposition processing on crystallite size enhancement as a function of treatment time on GLAD CdTe is tracked. A nominally 100 nm thick structurally engineered GLAD CdTe interlayer is introduced between the n-type CdS window layer and p-type CdTe absorber in solar cells. Power conversion efficiency (PCE), open circuit voltage (VOC), and fill factor
(FF) improve with introduction of this GLAD CdTe interlayer between CdS and CdTe layers of a conventional CdS/CdTe heterojunction solar cell.
6.2 Film Deposition and Measurement Details
RF magnetron sputtered (frequency = 13.56 MHz) CdTe films have been deposited at room temperature (RT) onto soda-lime glass substrates using GLAD in a standalone vacuum chamber (AXXISTM, K. J. Lesker Co.). The glass substrates are cleaned ultrasonically, rinsed with deionized water, and blown dry with nitrogen before film deposition. A schematic diagram of the GLAD process is shown in the Figure 6-1. Films are deposited at angles (Φ) of 0°, 20°, 40°, 60°, and 80° as defined by the angle between the normal extending from the center of the sputtering target and the normal extending from the center of the substrate.
A 7.62 cm diameter 99.999% purity CdTe target is sputtered for 70 minutes in a 15 mTorr pressure Ar ambient at 100 W RF target power. The distance from the center of the target to the center of the substrate is 13.4 cm. The substrate remains stationary during deposition. The deposition rate of material in the center of the substrate is 28.1 nm/min at
Φ = 0° and slightly decreases with increasing angle Φ to 22.1 nm/min at Φ = 80°. Post- deposition CdCl2 heat treatment is performed by wetting the film surface with a saturated solution of CdCl2 in methanol followed by heating at 387°C in dry air ambient for 10, 20,
105
Figure 6-1: Schematic of glancing angle deposition (GLAD) sputtering.
30, or 40 minutes (Koirala et al., 2016; Tuteja et al., 2015; Paudel et al., 2012). The CdCl2 process is standard for CdTe PV as it increases CdTe grain / crystallite size and results in higher PCE devices. Cross-sectional and surface images are obtained from SEM (Hitachi
S-4800). XRD patterns (Rigaku Ultima III diffractometer) are analyzed to obtain the crystal structure and crystallite size. A silicon sample is used for calibration of the diffractometer.
Ex-situ spectroscopic ellipsometric and Mueller matrix spectra have been collected by using single rotating compensator (M-2000FI, J. A. Woollam Co. Inc.) (Lee et al., 1998;
Johs et al., 1999) and dual rotating compensator (Chen et al., 2004a) (RC2, J. A. Woollam
Co. Inc.) multichannel ellipsometers. Optical spectra are collected from near the center of each sample. Complex optical response and optical anisotropy resulting from columnar structure are obtained from analysis of ellipsometric and Mueller matrix spectra collected for GLAD CdTe films.
106 CdS/CdTe heterojunction solar cells are fabricated in the superstrate configuration, as shown in the schematic diagrams in Figure 6-2. Solar cells are fabricated with and without a GLAD CdTe interlayer prepared at Φ = 80°, which exhibits wurtzite crystal structure. The superstrates are 15.24 cm × 15.24 cm soda lime glass coated by 300-400 nm of SnO2:F with an additional 100 nm of high resistivity transparent SnO2 to form the transparent conducting oxide (TCO) front contact to the device (TEC-15M, Pilkington
North America). The solar cell fabrication sequence begins with magnetron sputter deposition of 100 nm n-type CdS followed by a 2 µm p-type CdTe absorber. These layers are then subjected to the aforementioned CdCl2 heat treatment for 30 mins. The solar cell is completed by RT thermal evaporation of 3 nm Cu and 40 nm Au through a dot shadow mask to form the back electrical contact. The completed cell structure is annealed at 150°C in air for 30 minutes to promote Cu diffusion into CdTe. Cu does not exist as a discrete layer in the final devices as it fully diffuses into CdTe, lightly doping it. 256 individual dot
Figure 6-2: Schematic diagram of CdS/CdTe heterojunction solar cells with and without a thin GLAD CdTe interlayer between the wurtzite CdS and zinc blende CdTe layers.
107 Table 6.1: Deposition conditions for the individual solar cell layers. “RT” denotes room temperature.
Deposition parameters Deposition Target Layer type Substrate Pressure Ar flow method power tempr (°C) (mTorr) (SCCM) (W) CdS RF sputter 250 200 15 23 CdTe interlayer Φ = 80° GLAD RT 100 15 23 CdTe RF sputter RT 100 15 23 Cu Evaporation RT - 10-3 - Au Evaporation RT - 10-3 -
solar cells each having an active area of 0.126 cm2 are formed for each process. The deposition parameters used during the fabrication of each component layer of the complete solar cells are listed in Table 6.1. Photocurrent density versus voltage (J-V) are measured under simulated one sun AM1.5G (100 mW/cm2) illumination from a 450 W Xenon light source (Oriel, Model 9119, Newport) with a digital source meter (Keithley 2440) in air at room temperature. External quantum efficiency (EQE) measurements are performed
(model IVQE8-C, PV Instruments) to characterize spectrally resolved device performance.
6.3 Microstructural and Optical Properties
Figure 6-3 shows cross-sectional and surface SEM of as-deposited CdTe prepared at different oblique angles. The cross-sections show that the films have slanted columnar structure inclined towards the incoming material flux during sputtering. The angle of column inclination relative to substrate normal is much smaller than the incident oblique angle but increases with increasing angles (Ehsani et al., 2013). This behavior is expected as atomic scale self-shadowing effects are more pronounced at more oblique deposition
108
Figure 6-3: Cross-sectional and surface scanning electron micrographs (SEMs) of as- deposited GLAD CdTe films prepared as functions of source flux angles with respect to substrate normal.
angles, resulting in continued growth of initial nuclei on the substrate as lengthening columns due to low diffusion lengths of precursors on the surface (Ehsani et al., 2013;
Podraza et al., 2005a; Podraza et al., 2005b). In general, columnar GLAD films have lower
109 columnar tilt angle than oblique deposition angle with even further reduction in tilt angle when scattering in the gas phase becomes more substantial (Nieuwenhuizen & Haanstra,
1966; Dirks & Leamy, 1977; Tait et al., 1993; Lichter & Chen, 1986; Álvarez et al., 2011;
Morrow et al., 2006; Hawkeye & Brett, 2007; Alfonso et al., 2012). The surface microstructure of the films vary as the oblique angle is increased. The surface morphology of the films deposited at Φ = 60° and 80° appear substantially different from other films deposited at lower Φ, which may be due to higher columnar tilt angle, crystal phase transitions from mixed (cubic + hexagonal) to hexagonal at higher Φ, or both. The relatively low substrate temperature is expected to result in low diffusion lengths of precursors on the growing surface, enabling the formation of the metastable wurtzite phase instead of the stable zinc blende phase. At low angles the films are mixed phase. At higher oblique angles the low diffusion length of precursors coupled with enhanced atomic scale self-shadowing results in the metastable wurtzite phase formation. Surface view SEM of as-deposited films in Figure 6-3 shows variation in grain size from 126 ± 7 to 87 ± 5 nm among the films with the Φ = 80° sample consisting of the smallest grains.
XRD patterns for the as-deposited CdTe films are shown in Figure 6-4. The samples deposited with small oblique angles Φ < 40° show the highest intensity XRD peak at 2θ
~23.7° which is ascribed to both cubic (C) zinc blende and hexagonal (H) wurtzite phases with C(111) and H(002) miller index planes, respectively. Although literature reports that cubic CdTe is produced when sputtered normal to substrate (lower angle Φ) and at lower temperature close to room temperature (RT) (Gu et al., 2018; Kulkarni et al., 2017;
Hosseinpanahi et al., 2015), low intensity but distinct C(222), H(100), H(101), and H(103) peaks are observed here in as-deposited GLAD CdTe films at Φ < 40° as shown for Φ = 0°
110 in the Figure 6-4(b). Films deposited at Φ < 40° are considered to be mixed phase cubic and hexagonal CdTe. As the oblique angle is increased Φ ≥ 40°, diffraction peaks associated with only the cubic phase are not observed. Crystallite size obtained from the
XRD analysis for the same samples is lower compared to grain size obtained from SEM indicating that these grains are composed of multiple crystallites, the mean free path of electrons is limited by defects and smaller than the grain, or both.
Figure 6-4: X-ray diffraction (XRD) patterns of (a) as-deposited CdTe sputtered on soda-lime glass at different oblique angles (Φ) and (b) Log scale XRD of as-deposited CdTe sputtered at Φ = 0°. The reference lines are for standard cubic zinc blende CdTe [PDF # 97-010-8238] and hexagonal wurtzite CdTe [PDF # 97-015-0941] obtained from MDI JADE software.
111 Isotropic optical response in the form of the complex dielectric function (ε = ε1 + iε2) spectra of all as-deposited GLAD CdTe films are obtained by fitting parametric layered optical and structural models to measured ellipsometric spectra from 0.74–5.89 eV using a least squares regression that minimizes the unweighted error function (Alterovitz & Johs,
1998) between measured and model-simulated ellipsometric spectra. Structurally, the models consist of a semi-infinite glass substrate, bulk CdTe layer thickness db, and surface roughness thickness ds as shown schematically in Figure 6-5. The optical properties of bulk
CdTe are initially parameterized by a sum of critical point oscillators assuming parabolic bands (CPPB) (Aspnes, 1980; Collins & Ferlauto, 2005) and a Tauc-Lorentz oscillator
(Jellison & Modine, 1996) to describe electronic band-to-band transitions. The optical response of the surface roughness is described by multiple Bruggeman effective medium approximation (EMA) (Fujiwara et al., 2000a) layers consisting of fv volume fraction void and 1−fv fraction material identical to the underlying bulk CdTe. Three effective medium layers are needed to describe the complicated surface roughness of combined thickness ds.
Figure 6-5: Complex dielectric function (ε = ε1 + iε2) spectra of as-deposited GLAD polycrystalline CdTe films produced at different oblique angles during sputtering. The optical response of single crystal (c-CdTe) is included for comparison (Li et al., 2011), and the structural model schematic is shown in the inset.
112 Table 6.2: Structural parameters including surface roughness thickness (ds), void fraction percentage in each effective medium approximation (EMA) layer comprising the surface roughness layer (fv), and bulk film thickness (db) for as-deposited CdTe at different oblique angles (Φ) between the substrate normal and target normal.
Surface roughness thicknesses (ds) and void fractions (fv) Bulk EMA1 EMA2 EMA3 Eff. thick. Φ thick. Void Void Void (nm) (nm) Thickness Thickness Thickness fraction fraction fraction (nm) (nm) (nm) (%) (%) (%) 1758 ± 0° 151 ± 10 1.7 ± 0.3 49 ± 7 7 ± 2 28 ± 3 39 ± 2 1968 ± 27 9 1716 ± 20° 148 ± 15 1.2 ± 0.2 32 ± 10 6 ± 4 25 ± 4 37 ± 3 1908 ± 37 8 1544 ± 40° 175 ± 30 0.6 ± 0.3 40 ± 24 4 ± 3 29 ± 5 39 ± 3 1775 ± 72 15 1318 ± 60° 160 ± 11 1.0 ± 0.2 49 ± 8 4 ± 1 32 ± 2 35 ± 1 1545 ± 26 7 1307 ± 80° 166 ± 12 0.8 ± 0.2 53 ± 9 3 ± 1 31 ± 1 32 ± 1 1544 ± 29 7
fv of each layer is allowed to vary independently resulting in monotonically increasing fv towards the outermost EMA layer and the ambient interface. This three EMA layer model is found to be necessary to accurately describe the relatively rough surface where there is a gradient in fv. This multiple layer approach to describe surface roughness with depth is complicated even for normal incidence sputtered films (Zapien et al., 2005; Li et al., 2009;
Li et al., 2008). All the structural parameters including ds, fv, db, and effective film thickness for as-deposited CdTe at different Φ are reported in Table 6.2. The thickness and the relative void fraction in each EMA layer representing surface roughness are similar for all films, with the difference in effective film thickness primarily from the bulk film thickness.
Effective film thickness is defined as the sum of db plus each EMA layer thickness weighted by 1-fv monotonically and varies from 1968 ± 27 nm for Φ = 0° to 1544 ± 29 nm for Φ = 80°. All the films were deposited for 70 minutes, and this variation in effective film
113 thickness with Φ is expected for GLAD films. After initially using a parametric ε to determine structural parameters, ε describing the bulk CdTe is obtained by numerical inversion (Oldham, 1969) in Figures 6-5. For all films, ε1 at low photon energies in the transparent spectral range are substantially lower compared to single crystal CdTe (c-
CdTe). A Bruggeman EMA consisting of fractions of void and c-CdTe is fit to ε1 at 0.74 eV obtained for all films as a proxy for relative density of the films. These films become less dense with increasing Φ with 1−fv decreasing from 0.86 for Φ = 0° to 0.80 for Φ = 80°.
The observed columnar tilt angles and the relative density of the films in this study seems small compared to the range in Φ. Low substrate temperature results in low surface precursor diffusion lengths, and the process pressure results in high gas scattering. The relatively small columnar tilt angles are expected based on these process conditions and consistent with similar films in literature (Álvarez et al., 2011; Morrow et al., 2006;
Hawkeye & Brett, 2007; Alfonso et al., 2012). The decreasing density resulting from higher oblique angles is due to increased columnar self-shadowing (Ehsani et al., 2013), transition in crystal structure, or both. By fitting CPPB oscillators directly to the first derivative of numerically inverted ε2 (dε2/dE) with respect to photon energy, four critical point features describing the E0, E1, E1+Δ1, and E2 electronic transitions are determined to have resonance energies ranging from 1.49 ± 0.20 to 1.54 ± 0.20 eV, 3.24 ± 0.30 to 3.27 ±
0.30, 3.37 ± 0.06 to 4.00 ± 0.20 eV, and 4.20 ± 0.11 to 4.77 ± 0.30 eV, respectively. These values are close to those known for c-CdTe at 1.491, 3.310, 3.894, and 5.160 eV (Li et al.,
2011; Johs et al., 1998; Kim et al., 1997). Sensitivity to the E1+Δ1 and E2 critical point energies may be lost in the films due to the complicated surface roughness structure, small grain / crystallite size, low critical point amplitude, and wide critical point broadening. Zinc
114 blende cubic CdTe exhibits lower band gap energy compared to wurtzite hexagonal CdTe
(Jiménez-Sandoval et al., 1992; Hosseini, 2008). Here, the band gap energy (Eg) of all as- deposited CdTe as a function of Φ are obtained from spectroscopic ellipsometry. Lower band gap values Eg ~1.50 eV are observed for Φ < 40° and higher values Eg ~1.53 eV for
Φ ≥ 40°. Although these band gap values are close and may be impacted by film stress as well, the behavior corresponds to structural trends from XRD. All these observations indicate that the wurtzite hexagonal phase is favored at higher values of Φ.
The columnar structures of these films are substantially inclined away from substrate normal, about 12° in the cross-sectional SEMs for films deposited at Φ = 60° and
80°. In-plane optical anisotropy would be expected for the films deposited at high Φ because of the projection of sampling of the polar Euler angle rotated columns, = 12° in this case. Mueller matrix spectra are measured at three azimuthal angles ( = 0, 45, and
90°) of rotation about the substrate normal and simultaneous modeling of all three sets of spectra is performed using a similar procedure to that explained in Ref. (Adhikari et al.,
2016). Mueller matrix elements normalized to M11 sensitive to sample rotations (m13, m14, m23, m24, m31, m32, m41, m42) in the transparent spectral range are fit. The 0.74–1.23 eV photon energy range is chosen to simplify the optical property model by avoiding substantial absorption above the band gap due to electronic transitions and the highest contribution to absorption below the band gap due to Urbach tails. To describe optical anisotropy in the films deposited at Φ = 60° and 80°, a uniaxial model is employed. The parametric optical response for both of these films is obtained by fitting a Sellmeier expression (Collins & Ferlauto, 2005) to the ε1 spectra shown in Figure 6-5 over the 0.74–
1.23 eV transparent wavelength range. These spectra are then input into EMAs alongside
115 a variable fraction void to describe both optical properties parallel (extra-ordinary) and perpendicular (ordinary) to the columnar principal axes. For each film, the void fraction in both ordinary and extraordinary directions are held common. Limiting forms of effective medium theory are employed (Aspens, 1982) in which the ordinary direction exhibits maximum electric field screening (screening parameter = 1) and the extraordinary direction exhibits minimum electric field screening (screening parameter = 0). The azimuthal Euler angle values for each spectra collected at nominal 휙 are fit. Thicknesses are obtained from
Figure 6-6: Anisotropic optical properties parallel (extra-ordinary, e) and perpendicular (ordinary, o) to the columnar principal axes in (top) ε1 and (bottom) birefringence, ne-no in the transparent spectral range of CdTe prepared by GLAD at Φ = 60° and 80°.
116 the isotropic model and fixed. The polar Euler angle θ is fixed to that obtained from cross- sectional SEM. Figure 6-6 shows the anisotropic optical properties over the modeled range for the two as-deposited CdTe films sputtered at Φ = 60° and 80°. The birefringence, difference in the indices of refraction in extra-ordinary and ordinary direction (ne-no), is greater for Φ = 80° due to increased shadowing effects at higher angles of incidence. This behavior has also been observed in other GLAD polycrystalline thin films (Podraza et al.,
2005a; Podraza et al., 2005b; Wang et al., 2006; Tokas et al., 2016).
The post-deposition CdCl2 treatment process is an important step during the fabrication of CdTe solar cells for improving device performance. CdCl2 treatment results in recrystallization, grain growth, grain boundary passivation, and randomizing the crystal orientation (Dhere & Dhere, 2005; McCandless & Sites, 2003). CdCl2 treatments are applied to GLAD CdTe films at different oblique angles. Using XRD measurement and
Figure 6-7: Crystallite size deduced via the Scherrer equation as a function of CdCl2 treatment time for CdTe deposited at different oblique angles.
117
Figure 6-8: Surface SEM images of CdTe film deposited at Φ = 0° (a) before and (b) after CdCl2 treatment for 30 minutes.
Scherrer’s equation (Scherrer, 1918), the crystallite size variations with CdCl2 treatment time are shown in Figure 6-7. In the calculation, the diffraction peak centered around 23.7° indexed to C(111) and H(002) is considered. Crystallite size increases after CdCl2 treatment for all samples and increases with increasing treatment time. On average, the crystallite size for all sample increased by ~20% after CdCl2 treatment for the most commonly used 30 minutes treatment time (Koirala et al., 2016; Tuteja et al., 2015; Paudel et al., 2012) compared to as-deposited samples. Also, surface view SEM have shown enhanced grain size after CdCl2 treatment. Figure 6-8 shows an example of surface view
SEM images of CdTe film deposited at Φ = 0° before and after CdCl2 treatment for 30 minutes. As a result of CdCl2 treatment, the grain size increases from 122 ± 5 nm (before treatment) to 317 ± 7 nm (after treatment).
6.4 Solar Cell with GLAD CdTe Interlayer
CdS/CdTe heterojunction solar cells with and without nominally 100 nm thick
GLAD wurtzite interfacial layers between the n-type CdS window layer and p-type zinc blende CdTe absorber have been fabricated using Cu/Au back contacts (Fig. 6-2). All CdTe layers are deposited at RT followed by post-deposition CdCl2 treatment. Figure 6-9 shows
118 the XRD patterns of Φ = 0° CdTe absorber layer and Φ = 80° GLAD CdTe interlayer after
CdCl2 treatment along with that of the hexagonal CdS window layer. After CdCl2 treatment, Φ = 80° GLAD CdTe interlayer exhibits diffraction peaks corresponding to both cubic and hexagonal crystal structures. As some hexagonal phase CdTe crystallites exist after the CdCl2 process, better lattice matching with the underlying CdS can occur. For the absorber layer deposited at Φ = 0° after CdCl2 treatment, only C(111), C(200), C(220),
C(311), C(400), and C(331) index diffraction peaks associated with cubic CdTe are observed. Although the peak position of C(111) and H(002), C(220) and H(110), and
C(311) and H(112) are close, the absence of additional hexagonal peaks indicates that the
CdTe absorber layer deposited at Φ = 0° has cubic zinc blende crystal structure after CdCl2
Figure 6-9: XRD patterns of CdS (black), CdCl2 treated CdTe (Φ = 0°) (blue), and CdCl2 treated CdTe (Φ = 80°) (red). Peaks associated with CdxTeOy are also indicated (*). The vertical lines represent the reference diffraction peak positions for hexagonal CdS (PDF#97-015-4186), cubic CdTe (PDF#97- 010-8238), and hexagonal CdTe (PDF#97-015-0941) crystal structures obtained from MDI JADE software.
119 treatment. Also, after CdCl2 treatment, the presence of CdxTeOy peaks shows the oxidation of the surface, as expected for these processing conditions.
The Scherrer equation has been used to determine the crystallite size associated with each diffraction peak in as-deposited and CdCl2 treated (30 minutes) samples prepared at different Φ. Crystallite size associated with H(102) crystallites increases slightly for Φ
= 40° and 60° and remains approximately the same for Φ = 80° after CdCl2 treatment. The crystallite size associated with the H(103) peak increases for Φ = 80°, is approximately the same for Φ = 60°, and slightly decreases for Φ = 40°. Crystallite size associated with the most dominate crystallite orientations increase with CdCl2 treatment. Cubic phase crystallite orientations become more random after CdCl2 treatment as observed by the appearance of C(200) planes in addition to the C(220), C(111), and C(311) planes prior to treatment. Individual hexagonal phase crystallite orientations become either more or less pronounced after CdCl2 treatment depending upon the initial crystallite orientations present in the samples. These observations imply that the introduction of the thin GLAD CdTe interlayer between the hexagonal CdS and cubic CdTe absorber layer is improving lattice matching on both sides of the junction via alignment of crystal planes. The presence of the interlayer is expected to result in reductions in strain energy, density of unpassivated bonds, or both which is expected to reduce the defect density at the interface.
Figure 6-10(a) shows J-V characteristics of the highest performing CdS/CdTe solar cells with and without the GLAD CdTe interlayer in the solar cell structure. The open circuit voltage (VOC), short circuit current density (JSC), power conversion efficiency
(PCE), and fill factor (FF) corresponding to devices plotted in Figure 6-10 are shown in
Table 6.3. The best device without the GLAD CdTe interlayer has a PCE = 10.05% with
120 Table 6.3: Comparison of device performance parameters of highest efficiency CdS/CdTe solar cells with and without a 100 nm thick GLAD CdTe interlayer. Average performance parameters with one standard deviation (1- σ) variation for the best 200 (out of 256) small area devices for each configuration are also given.
V J PCE FF Solar cell parameter OC SC (V) (mA/cm2) (%) (%) Best cell 0.728 23.2 10.1 59.4 Without Average of 0.710 ± interlayer 20.0 ± 1.8 7.72 ± 1.04 54.1 ± 3.1 200 cells 0.032 Best cell 0.774 23.0 11.0 61.9 With interlayer Average of 0.730 ± 20.4 ± 1.3 8.10 ± 1.25 54.2 ± 3.7 200 cells 0.030
2 VOC = 0.728 V, JSC = 23.2 mA/cm , and FF = 59.4%. After incorporating structurally engineered GLAD CdTe between the n-type CdS window layer and normally sputtered p- type CdTe absorber, the best device has PCE = 11.02% with VOC = 0.774 V, JSC = 23.0 mA/cm2, and FF = 61.9%. Using the GLAD CdTe interlayer, PCE increased by 9% relative and 0.9% absolute, and this enhancement in device performance is attributed to an increase
Figure 6-10: Device performance for the highest efficiency solar cells with and without a 100 nm GLAD CdTe interlayer: (a) light and dark current-voltage (J-V) and (b) external quantum efficiency (EQE) for CdS/CdTe heterojunction solar cells with (red) and without (blue) introduction of the GLAD interlayer.
121
Figure 6-11: (a) Open circuit voltage (VOC), (b) short circuit current (JSC), (c) fill factor (FF), and (d) power conversion efficiency (PCE) parameter ranges for all CdTe PV devices with and without GLAD CdTe interlayers.
in VOC by 6% and FF by 4%. Device performance parameters of all CdS/CdTe heterojunction solar cells with and without GLAD CdTe interlayers are shown in Figure 6-
11. Out of 256 dot cells, some of the cells show substantially lower performance and hence only the 200 highest performing dot cells are included in the statistical average calculation and one standard deviation variation for both sets of solar cells given in Table 6.3. The average VOC, JSC, FF, and PCE of devices with GLAD CdTe interlayer are higher compared to device without GLAD interlayer. On average, PCE is higher for devices with the GLAD interlayer with the largest gain attributed to improved VOC. Figure 6-10(b) shows similar
EQE spectra over the 300-1000 nm wavelength range for the same highest efficiency devices indicating that the presence of the interlayer does not reduce the spectroscopic optical performance of these solar cells.
122 Solar cell diode parameters are obtained by fitting experimental J-V to the diode
equation (Hegedus & Shafarman, 2004; Bhandari et al., 2017):
푞 퐽 = 퐽0푒푥푝 [ (푉 − 푅푆퐽)] + 퐺푆퐻푉 − 퐽퐿 (6.1) 푛푘퐵푇
Here q is the electronic charge, n is the diode ideality factor, kB is Boltzmann’s constant, T
is temperature, J0 is the reverse saturation current, RS is the series resistance, GSH is the
shunt conductance, and JL is the current density at AM1.5G light illumination. Ideality
factors are calculated by plotting the derivative dV/dJ from the exponential diode equation
as a function of 1/J when the devices are not illuminated, with results for the highest
efficiency devices with and without the GLAD interlayer shown in Figure 6-12(a). The
slope and intercept of the linear fit give RS and n, respectively. The calculation results in
similar series resistances of 2.4 and 2.1 Ω cm2 for the devices with and without the GLAD
CdTe interlayers and substantially different ideality factors of 2.3 and 3.0, respectively.
Similarly, the plot between ln[J] and V-RSJ using Rs, allows for extraction of n and J0 for
the highest efficiency devices with and without the GLAD interlayer as shown in Figure
-6 2 6-12(b). These values of n are 2.0 and 2.5, and the values of J0 are 2×10 mA/cm and
7×10-5mA/cm2, respectively, for the devices with and without GLAD CdTe interlayers.
The values of n and J0 for our CdTe devices are consistent with values reported in literature
for similar devices (Hegedus & Shafarman, 2004; Bhandari et al., 2017; Marsillac et al.,
2007). Overall, lower values of Jo and n with comparable Rs for the device with the GLAD
CdTe interlayer indicates improvement in the diode performance parameters. The lower
values of n and J0 for the highest efficiency cell with the GLAD interlayer compared to the
highest efficiency device without the interlayer indicates lower recombination of charge
carriers due to the formation of an improved electronic quality interface (Marsillac et al.,
123
Figure 6-12: Analysis of dark current density versus voltage characteristics for the CdS/CdTe heterojunction solar cells with and without GLAD CdTe interlayers: (a) dV/dJ versus 1/J plot for calculation of n and RS and (b) semi-log scale plot for the calculation of J0 and n.
2007). The resulting improvement in diode parameters leads to increases in VOC, FF, and
overall solar cell performance as observed for devices incorporating the GLAD CdTe
interlayer. The better performance of device with the GLAD interlayer can be attributed to
improved n/p interface due to less lattice mismatch between wurtzite CdS and hexagonal
124 wurtzite phase crystallites in the GLAD CdTe interlayer, as both share the same crystal structure and more similar lattice constants when compared to zinc blende CdTe. As a result, wurtzite CdTe incorporated in these devices makes higher electronic quality, less defective interfaces with the underlying wurtzite CdS and over-deposited zinc blende CdTe
(Chou et al., 1996; Wu et al., 2002). All these results indicate that the idea of incorporating a thin GLAD CdTe interlayer between CdS and zinc blende CdTe serves as a means of interfacial tailoring at the heterojunction for improving PV device performance. This concept will be extended for devices incorporating more optimized CdTe after deposition of the interlayer, as opposed to RT normal incidence deposited CdTe.
6.5 Summary
GLAD magnetron sputtered CdTe shows variation in optical properties, density, and microstructure with incident oblique angle. Crystal phase structure transitions from mixed-phase cubic zinc blende and hexagonal wurtzite crystal structures to the hexagonal phase with increasing oblique angle. As-deposited GLAD CdTe films prepared at higher oblique angles show sensitivity to in-plane optical anisotropy from the projection of the tilted columnar axes and as identified from analysis of Mueller matrix spectra. CdCl2 treated CdTe shows enhanced crystallite size compared to as-deposited samples with crystallite size increasing with treatment time. PV devices incorporating GLAD CdTe interlayer show 0.9% absolute higher efficiency compared to devices without the interlayer. The higher efficiency of the device with GLAD CdTe interlayer is because of an improved n/p interface due to improved structural compatibility between CdS and wurtzite phase crystallites in the GLAD CdTe interlayer.
125 Chapter 7
Conclusion and Future Work
7.1 Conclusion
Mueller matrix ellipsometry have been applied to study the anisotropic optical properties of uniaxial single crystal SrLaAlO4. Ex-situ spectroscopic ellipsometry and in- situ RTSE have been applied to study the magnetron sputtered Si:H films and n-i-p solar cell with nc-Si:H absorbers. Similarly, the optical properties and microstructural properties of GLAD CdTe and the effect on CdS/CdTe heterojunction solar cell performance by including a thin layer of GLAD CdTe between n-type CdS window layer and p-type CdTe absorber is studied. Major results of this research dissertation include; i) finding optical, electrical and structural properties of sputtered Si:H and GLAD CdTe films, ii) study the growth of sputtered Si:H using RTSE during film deposition, iii) fabrication of n-i-p solar cells with nc-Si:H absorbers deposited by sputtering and PECVD and their characterization including EQE simulations, and iv) fabrication and characterization of CdS/CdTe solar cells with and without GLAD interlayer. Brief details of the major finding are summarized in this section.
The ordinary and extra-ordinary optical properties for single crystal LaSrAlO4 have been determined over the spectral range from 0.74 to 5.90 eV from analysis of ellipsometric
126 and Mueller matrix spectra. The birefringence of about ~ 0.025 has been obtained across the transparent region. The indirect band gap has been identified at 2.74 ± 0.01 eV.
Parametric models for ordinary and extra-ordinary spectra in have been developed. Direct transitions in have been identified at 5.54 ± 0.01 and 6.01 ± 0.06 eV with the 5.54 eV feature in the extra-ordinary direction exhibiting reduced amplitude and transition strength.
This work shows an example of basic procedure to obtain the anisotropic optical properties and associated parameters using Mueller matrix ellipsometry as well as a parametric model for indirect gap crystalline semiconductors.
Phase transformations from amorphous to mixed-phase then to single-phase nanocrystalline during growth have been observed for Si:H films produced via magnetron sputtering. In studies producing growth evolution diagrams, deposition rates are not observed to decrease at the lowest pH2 at which crystallites nucleate. As a result, similar growth rates are obtained between a-Si:H and nc-Si:H phases near what would be considered the protocrystalline region in PECVD Si:H. Different Si-Hn bonding configurations with absorption peaks centered at 590, 640, 2000, and 2090 cm-1 have been observed from IR spectroscopic ellipsometry. Samples with higher pH2 and lower deposition rates after crystallite coalescence exhibit lower relative amounts of hydrogen incorporation into the film, which may be due to the etching of weak Si-Si bonds on the growing surface by unincorporated reactive hydrogen (Moustakas et al., 1985).
The n-i-p solar cells in substrate configurations have been fabricated incorporating nc-Si:H intrinsic absorber layers deposited by sputtering and PECVD. The cells with nanocrystalline PECVD absorbers and an untextured (planar) back reflector serve as a baseline for comparison with power conversion efficiency near 6%. This efficiency is
127 typical of this device configuration lacking optical enhancement due to scattering of light incident to the back reflector. By comparison, cells with sputtered absorbers achieved efficiencies of about 1%. Comparison of dark / light current-voltage measurements and
EQE between the devices fabricated with different absorbers indicate that lower performance in devices with sputtered absorbers may be attributed to both low electronic quality within the nc-Si:H absorber and also process incompatibility at the interfaces between the intrinsic layer made by sputtering with the doped layers made by PECVD.
Substantial carrier concentration losses are identified at the n-/i-interface for the solar cell incorporating the sputtered absorber. Although the PV performance of these devices can be improved, there is potential for development of nc-Si:H without the need of toxic source gases to reduce overall device cost. The sputtered material itself requires further optimization to reach performance levels comparable to those achievable with PECVD, but this works serves as a baseline for future material and device studies.
GLAD magnetron sputtered CdTe thin films with different microstructures and optical properties have been produced on soda-lime glass by glancing angle deposition
(GLAD) at sputter oblique angles varying from 0° to 80° with respect to substrate normal.
From cross-sectional micrographs, increasingly tilted columnar structure occurs with increasing incident source flux angle for as-deposited CdTe films. Films deposited at lower source oblique angles closer to normal incidence have mixed (cubic + hexagonal) crystal structure and those prepared at more oblique higher angles have hexagonal wurtzite crystal structure. Post-deposition CdCl2 treated films show enhancement in grain size for samples prepared under all conditions. The optical response in the form of the complex dielectric function (ε = ε1 + iε2) spectra from 0.74 to 5.89 eV for the GLAD thin films are all
128 qualitatively similar to single crystal CdTe. Higher angle deposited samples show columnar structure induced anisotropy in spectra in ε in the transparent spectral range.
Application of GLAD CdTe interlayers between CdS and CdTe of the standard CdS/CdTe heterojunction design solar cell shows better performance with 0.9% absolute efficiency increase. This work will serve as the base for the ongoing project on GLAD CdTe thin film and CdS/CdTe heterojunction solar cells with a GLAD CdTe layer between the CdS and normal-incidence deposited CdTe.
7.2 Suggestion for Future Work
The Mueller matrix ellipsometry analysis technique presented here can be applied as a base line for the study of anisotropic optical properties of other similar materials. As an example, the TCOs and semiconductor thin films produced by GLAD may have anisotropy in the optical properties and the use of Mueller matrix ellipsometry can provide in-depth optical analysis of those materials. The n-i-p solar cells with sputtered nc-Si:H as an absorber discussed in Chapter 5 were fabricated on a flat BR structure. We have demonstrated the solar cells with the highest efficiency of about 1% for 1µm thick sputtered nc-Si:H absorber. There are several possibilities that the performance of this device can be increased. Using textured BR structure and making better n/i and p/i interfaces are some areas for the future study. Fabrication of all-sputtered nc-Si:H devices by optimizing sputtered p-type and n-type nc-Si:H is another future pathway. This way we can decrease the price without compromising the device quality.
The other important thin film based material to work with is CdTe. We have demonstrated that the GLAD technique can produce films with different microstructures,
129 optical property, and electrical property. So, this procedure can be very useful for other materials used in solar cell devices. As an example, GLAD TCOs may have better opto- electronic property and their application in solar cells may result in better device performances by minimizing optical losses (Leem & Yu, 2011). The results presented in
Chapter 6 are from the currently running research project and this may be the base for the future research.
130 References
Achiq et al., 1998 A. Achiq, R. Rizk, F. Gourbilleau, R. Madelon, B. Garrido, A. Perez-Rodrıguez, and J. R. Morante, “Effects of prior hydrogenation on the structure and properties of thermally nanocrystallized silicon layers,” J. Appl. Phys. 83, 5797 (1998).
Adhikari et al., 2016 D. Adhikari, M. M. Junda, P. Uprety, K. Ghimire, I. Subedi, and N. J. Podraza, “Near infrared to ultraviolet anisotropic optical properties of single crystal SrLaAlO4 from spectroscopic ellipsometry,” Phys. Status Solidi B 253, 2066 (2016).
Adhikari et al., 2017a D. Adhikari, M. M. Junda, S. X. Marsillac, R. W. Collins, and N. J. Podraza, “Nanostructure evolution of magnetron sputtered hydrogenated silicon thin films,” J. Appl. Phys. 122, 075302 (2017).
Adhikari et al., 2017b D. Adhikari, M. M. Junda, S. X. Marsillac, R. W. Collins, and N. J. Podraza, “Magnetron sputtered hydrogenated silicon thin films: assessment for application in photovoltaics,” In Proceedings of the 44th IEEE Photovoltaic Specialists Conference, Washington, DC, USA, 25–30 (2017); pp. 1–6.
Adhikari et al., 2019 D. Adhikari, M. M. Junda, E. Bastola, P. Koirala, R. J. Ellingson, R. W. Collins, and N. J. Podraza, “Glancing angle deposited CdTe: Nanostructured films and impact on solar cell performance,” Surf. Coat. Tech., (2019) (in press).
Adhikari, et al., 2018 D. Adhikari, P. Koirala, M. M. Junda, R. W. Collins, and N. J. Podraza, “Glancing angle deposited CdTe: optical properties and structure,” In Proceeding of IEEE 7th World Conference on Photovoltaic Energy Conversion (WCPEC) (2018), DOI: 10.1109/PVSC.2018.8547890.
Ahmed et al., 2012 S. Ahmed, K. B. Reuter, O. Gunawan, L. Guo, L. T. Romankiw, and H. Deligianni “A high efficiency electrodeposited Cu2ZnSnS4 solar cell,” Adv. Energy Mater. 2, 253 (2012).
131 Alfonso et al., 2012 E. Alfonso, J. Olaya, and G. Cubillos, “Thin film growth through sputtering technique and its applications,” M. Andreeta (Ed.), Crystallization-Science and Technology, Intech, Croatia (2012), pp. 397-432.
Alterovitz & Johs, 1998 S. A. Alterovitz and B. Johs, “Multiple minima in the ellipsometric error function,” Thin Solid Films 313-314, 124 (1998).
Álvarez et al., 2011 R. Álvarez, L. González-García, P. Romero-Gómez, V. Rico, J. Cotrino, A. R. González-Elipe, and A. Palmero, “Theoretical and experimental characterization of TiO2 thin films deposited at oblique angles,” J. Phys. D: Appl. Phys. 44, 385302 (2011).
Amrani et al., 2012 R. Amrani, P. Abboud, L. Chahed, and Y. Cuminal, “Low temperature growth of hydrogenated silicon prepared by PECVD from argon diluted silane plasma,” Cryst. Struct. Theory Appl. 1, 62 (2012).
An et al., 1990 I. An, H. V. Nguyen, N. V. Nguyen, and R. W. Collins, "Microstructural evolution of ultrathin amorphous-silicon films by real-time spectroscopic ellipsometry." Phys. Rev. Lett. 65, 2274 (1990).
Aspnes, 1980 D. E. Aspnes, Handbook of Semiconductors; Vol. 2, edited by M. Balkanski; North-Holland: Amsterdam, The Netherlands, 1980.
Aspnes, 1982 D. E. Aspnes, “Optical properties of thin films,” Thin Solid Films 89, 249 (1982).
Azzam & Bashara, 1977 R. M. A. Azzam and N. M. Bashara, "Ellipsometry and Polarized Light." Amsterdam, North Holland, (1977).
Bhandari et al., 2017 K. P. Bhandari, X. Tan, P. Zereshki, F. K. Alfadhili, A. B. Phillips, P. Koirala, M. J. Heben, R. W. Collins, and R. J. Ellingson, “Thin film iron pyrite deposited by hybrid sputtering/co-evaporation as a hole transport layer for sputtered CdS/CdTe solar cells,” Sol. Energy Mater. Sol. Cells 163, 277 (2017).
Boone et al., 1982 J. L. Boone, T. P. Van Doren, and A. K. Berry, “Deposition of CdTe by spray pyrolysis,” Thin Solid Films 87, 259 (1982).
132 Britt & Ferekides, 1993 J. Britt, and C. Ferekides, “Thin-film CdS/CdTe solar cell with 15.8% efficiency,” Appl. Phys. Lett. 62, 2851 (1993).
Brodsky et al., 1977 M. H. Brodsky, M. Cardona, and J. J. Cuomo, “Infrared and Raman spectra of the silicon-hydrogen bonds in amorphous silicon prepared by glow discharge and sputtering,” Phys. Rev. B 16, 3556 (1977).
Brown et al., 1990 R. Brown, V. Pendrick, D. Kalokitis, and B. H. T. Chai, “Low‐loss substrate for microwave application of high‐temperature superconductor films,” Appl. Phys. Lett. 57, 1351 (1990).
Cao et al., 2008 X. Cao, J. A. Stoke, J. Li, N.J. Podraza, W. Du, X. Yang, D. Attygalle, X. Liao, R. W. Collins, and X. Deng, “Fabrication and optimization of single-junction nc-Si:H n–i–p solar cells using Si:H phase diagram concepts developed by real time spectroscopic ellipsometry,” J. Non-Cryst. Solids 354, 2397 (2008).
Cardona, 1983 M. Cardona, “Vibrational spectra of hydrogen in silicon and germanium,” Phys. Status Solidi B 118, 463 (1983).
Chen et al., 2004a C. Chen, I. An, G. M. Ferreira, N. J. Podraza, J. A. Zapien, and R. W. Collins, “Multichannel Mueller matrix ellipsometer based on the dual rotating compensator principle,” Thin Solid Films 14 ,455 (2004).
Chen et al., 2004b G. X. Chen, Y. J. Ge, C. Z. Bi, X. G. Qiu, and B. R. Zhao, “Far-infrared optical properties of SrLaAlO4 single crystal,” J. Appl. Phys. 95, 3417 (2004).
Chinese Academy of Science, 2014 Research Progress Report, Institute of Electrical Engineering, Chinese Academy of Science, 2014.
Chou et al., 1996 H. C. Chou, A. Rohatgi, N. M. Jokerst, S. Kamra, S. R. Stock, S. L. Lowrie, R. K. Ahrenkiel, and D. H. Levi, “Approach toward high efficiency CdTe/CdS heterojunction solar cells,” Mater. Chem. Phys., 43, 178 (1996).
Collins & Ferlauto, 2005 R. W. Collins and A. S. Ferlauto, in: Handbook of Ellipsometry, edited by H. G. Tomkins and E. A. Irene (William Andrew, Norwich, NY, 2005) pp. 92-235.
133 Collins et al., 1991 R. W. Collins, I. An, H. V. Nguyen, and T. Gu, “Real time spectroscopic ellipsometry for characterization of thin film optical properties and microstructural evolution,” Thin Solid Films 206, 374 (1991).
Collins et al., 2003 R. W. Collins, A. S. Ferlauto, G. M. Ferreira, C. Chen, J. Koh, R. J. Koval, Y. Lee, J. M. Pearce, and C. R. Wronski, “Evolution of microstructure and phase in amorphous, protocrystalline, and microcrystalline silicon studied by real time spectroscopic ellipsometry,” Sol. Energy Mater. Sol. Cells 78, 143 (2003).
Dahal et al., 2014 L. R. Dahal, J. Li, J. A. Stoke, Z. Huang, A. Shan, A. S. Ferlauto, C. R. Wronski, R. W. Collins, and N. J. Podraza, “Applications of real-time and mapping spectroscopic ellipsometry for process development and optimization in hydrogenated silicon thin-film photovoltaics technology,” Sol. Energy Mater. Sol. Cells 129, 32 (2014).
Dahal, 2013 L. R. Dahal, “Spectroscopic Ellipsometry Studies of Thin Film a-Si:H Solar Cell Fabrication by Multichamber Deposition in the n-i-p Substrate Configuration,” Ph.D. Dissertation, The University of Toledo, Toledo, OH (2013).
De Meneses et al., 2006 D. De Meneses, M. Malki, and P. Echegut, “Structure and lattice dynamics of binary lead silicate glasses investigated by infrared spectroscopy,” J. Non-Cryst. Solids 352, 769 (2006).
Demsar et al., 2007 J. Demsar, A. Gozar, V. K. Thorsmolle, A. J. Taylor, and I. Bozovic, “Long-lived near-infrared photoinduced absorption in LaSrAlO4 excited with visible light,” Phys. Rev. B 76, 054304 (2007).
Deng & Schiff, 2003 X. Deng and E. A. Schiff, Handbook of Photovoltaic Science and Engineering, edited by A. Luque and S. Hegedus (John Wiley & Sons, 2003), pp. 505–565.
Dhere & Dhere, 2005 N. G. Dhere, and R. G. Dhere, “Thin-film photovoltaics,” J. Vac. Sci. Technol. A 23, 1208 (2005).
Dirks & Leamy, 1977 A. G. Dirks and H. J. Leamy, “Columnar microstructure in vapor-deposited thin films,” Thin Solid Films, 47, 219 (1977).
134 Dobson et al., 2000 K. D. Dobson, I. Visoly-Fisher, G. Hodes, and D. Cahen, “Stability of CdTe/CdS thin-film solar cells,” Sol. Energy Mater. Sol. Cells 62, 295 (2000).
Dutta et al., 2008 P. Dutta, S. Paul, S. Tripathi, Y. Chen, S. Chatterjee, V. Bommisetty, D. Galipeau, and A. Liu, “Comparative study of nc-Si:H deposited by reactive sputtering using crystalline and polycrystalline silicon targets,” in 33rd IEEE Photovoltaic Specialists Conference (2008), pp. 1-4.
Ehsani et al., 2013 M. H. Ehsani, H. Rezagholipour Dizaji, S. Azizi, S. F. Ghavami Mirmahalle and F. Hosseini Siyanaki, “Optical and structural properties of cadmium telluride films grown by glancing angle deposition,” Phys. Scr. 88, 025602 (2013).
El-Gohary et al., 2003 Z. El-Gohary, M. El-Nahass, H. Soliman, and Y. L. El-Kady, “Optical dispersion parameters with different orientations for SrLaAlO4 single crystals,” J. Mater. Sci. Technol. 19, 77 (2003).
Ferlauto et al., 2002 A. S. Ferlauto, G. M. Ferreira, J. M. Pearce, C. R. Wronski, R. W. Collins, X. Deng, and G. Ganguly, “Analytical model for the optical functions of amorphous semiconductors from the near-infrared to ultraviolet: Applications in thin film photovoltaics” J. Appl. Phys. 92, 2424 (2002).
Ferlauto et al., 2004 A. S. Ferlauto, G. M. Ferreira, R. J. Koval, J. M. Pearce, C. R. Wronski, R. W. Collins, M. M. Al-Jassim, and K. M. Jones, “Evaluation of compositional depth profiles in mixed phase (amorphous + crystalline) silicon films from real time spectroscopic ellipsometry,” Thin Solid Films 455–456, 665 (2004).
Freeman & Paul, 1978 E. C. Freeman and W. Paul, “Infrared vibrational spectra of rf-sputtered hydrogenated amorphous silicon,” Phys. Rev. B 18, 4288 (1978).
Fujiwara & Kondo, 2005 H. Fujiwara and M. Kondo, “Real-time monitoring and process control in amorphous ∕ crystalline silicon heterojunction solar cells by spectroscopic ellipsometry and infrared spectroscopy,” Appl. Phys. Lett. 86, 032112 (2005).
Fujiwara & Kondo, 2007 H. Fujiwara and M. Kondo, “Effects of a‐Si:H layer thicknesses on the performance of a‐Si:H∕c‐Si heterojunction solar cells,” J. Appl. Phys. 101, 054516 (2007).
135 Fujiwara et al., 2000a H. Fujiwara, J. Koh, P. I. Rovira, and R. W. Collins, "Assessment of effective medium theories in the analysis of nucleation and microscopic surface roughness evolution for semiconductor thin films." Phys. Rev. B 61, 10832 (2000).
Fujiwara et al., 2000b H. Fujiwara, Y. Toyoshima, M. Kondo, and A. Matsuda, “Structural study of initial layer for µc-Si:H growth using real time in situ spectroscopic ellipsometry and infrared spectroscopy,” J. Non-Cryst. Solids 266-269, 38 (2000).
Fujiwara et al., 2001 H. Fujiwara, M. Kondo, A. Matsuda, “Real-time spectroscopic ellipsometry studies of the nucleation and grain growth processes in microcrystalline silicon thin films,” Phys. Rev. B: Condens. Matter 63, 115306 (2001).
Fujiwara et al., 2002 H. Fujiwara, M. Kondo, and A. Matsuda, “Depth profiling of silicon–hydrogen bonding modes in amorphous and microcrystalline Si:H thin films by real-time infrared spectroscopy and spectroscopic ellipsometry,” J. Appl. Phys. 91, 4181 (2002).
Ganguly et al., 1996 G. Ganguly, T. Ikeda, T. Nishimiya, K. Saitoh, M. Kondo, and A. Matsuda, “Hydrogenated microcrystalline silicon germanium: A bottom cell material for amorphous silicon based tandem solar cells,” Appl. Phys. Lett. 69, 4224 (1996).
Ge et al., 2014 J. Ge, J. Jiang, P. Yang, C. Peng, Z. Huang, S. Zuo, L. Yang, J. Chu, “A 5.5% efficient co-electrodeposited ZnO/CdS/Cu2ZnSnS4/Mo thin film solar cell,” Sol. Energy Mater. Sol. Cells 125, 20 (2014).
Ghimire et al., 2015 K. Ghimire, H. F. Haneef, R. W. Collins, and N. J. Podraza, “Optical properties of single-crystal Gd3Ga5O12 from the infrared to ultraviolet,” Phys. Status Solidi B 252, 2191 (2015).
Gloubokov et al., 1995 A. Gloubokov, R. Jablonski, W. Ryba-Romanowski, J. Sass, A. Pajaczkowska, R. Uecker, and P. Reiche, “On the preparation and crystal growth of SrLaAlO4,” J. Cryst. Growth 147, 123 (1995).
Green et al., 2019 M. A. Green, Y. Hishikawa, E. D. Dunlop, D. H. Levi, J. Hohl‐Ebinger, M. Yoshita, and A.W.Y. Ho‐Baillie, “Solar cell efficiency tables (version 53),” Prog. Photovoltaics Res. Appl. 27, 3 (2019).
136 Green et al., 2018 M. A. Green, Y. Hishikawa, E. D. Dunlop, D. H. Levi, J. Hohl‐Ebinger, and A.W.Y Ho‐Baillie, “Solar cell efficiency tables (version 52),” Prog. Photovoltaics Res. Appl. 26, 427 (2018).
Greene, 2017 J. E. Greene, “Review Article: Tracing the recorded history of thin-film sputter deposition: From the 1800s to 2017,” J. Vac. Sci. Technol. A 35, 1(2017).
Gu et al., 2018 P. Gu, X. Zhu, H. Wu, and D. Yang, “Regulation of substrate-target distance on the microstructural, optical, and electrical properties of CdTe films by magnetron sputtering,” Materials 11, 2496 (2018).
Gupta & Compaan, 2004 A. Gupta, A. D. Compaan, “All-sputtered 14% CdS∕CdTe thin-film solar cell with ZnO:Al transparent conducting oxide,” Appl. Phys. Lett. 85, 684 (2004).
Gupta et al., 2006 A. Gupta, V. Parikh, A. D. Compaan, “High efficiency ultra-thin sputtered CdTe solar cells,” Sol. Energy Mater. Sol. Cells, 90, 2263 (2006).
Hanabusa et al., 1997 M. Hanabusa, Z. Liu, N. Nakamura, and H. Hasegawa, “Pulsed laser deposition of silicon films for solar cell applications,” Nucl. Instrum. Methods Phys. Res., Sect. B 121, 367 (1997).
Haneef & Podraza, 2014 H. F. Haneef and N. J. Podraza, “Optical properties of single crystal Bi4Ge3O12 from the infrared to ultraviolet,” J. Appl. Phys. 116, 163507 (2014).
Hawkeye & Brett, 2007 M. M. Hawkeye and M. J. Brett, “Glancing angle deposition: Fabrication, properties, and applications of micro- and nanostructured thin films,” J. Vac. Sci. Technol. A 25, 1317 (2007).
Hedström et al., 1993 J. Hedström, H. Ohlsen, M. Bodegard, A. Kylner, L. Stolt, D. Hariskos, M. Ruckh, and H. Schock, “ZnO/CdS/Cu(In,Ga)Se2 thin film solar cells with improved performance,” Proc. 23rd IEEE Photovoltaic Specialist Conf., p. 364 (1993).
Hegedus & Shafarman, 2004 S. S. Hegedus and W. N. Shafarman, “Thin-film solar cells: Device measurements and analysis,” Prog. Photovolt: Res. Appl. 12, 155 (2004).
137 Heintze et al., 1996 M. Heintze, R. Zedlitz, H. N. Wanka, M. B. Schubert, “Amorphous and microcrystalline silicon by hot wire chemical vapor deposition,” J. Appl. Phys. 79, 2699 (1996).
Herzinger et al., 1998 C. M. Herzinger, B. Johs, W. A. McGahan, J. A. Woollam, W. Paulson, “Ellipsometric determination of optical constants for silicon and thermally grown silicon dioxide via a multi-sample, multi-wavelength, multi-angle investigation,” J. Appl. Phys. 83, 3323 (1998).
Honsberg & Bowden, 2010 C. Honsberg and S. Bowden, www.pveducation.org/pvcdrom (2010).
Hora et al., 1996 J. Hora, K. Navrátil, J. Humlíček, and M. Berkowski, “Optical anisotropy of SrLaAlO4 and SrLaAl0.75Ga0.25O4 single crystals,” Phys. Status Solidi B 195, 625 (1996).
Hosseini, 2008 S. M. Hosseini, “Optical properties of cadmium telluride in zinc-blende and wurzite structure,” Physica B 403, 1907 (2008).
Hosseinpanahi et al., 2015 F. Hosseinpanahi, D. Raoufi, K. Ranjbarghanei, B. Karimi, R. Babaei, and E. Hasani, “Fractal features of CdTe thin films grown by RF magnetron sputtering” Appl. Surf. Sci. 357, 1843 (2015).
Houben et al., 1998 L. Houben, M. Luysberg, P. Hapke, R. Carius, F. Finger, and H. Wagner, “Structural properties of microcrystalline silicon in the transition from highly crystalline to amorphous growth,” Philos. Mag. A 77, 1447 (1998).
Huang et al., 2015 Z. Huang, L. R. Dahal, M. M. Junda, P. Aryal, S. Marsillac, R. W.Collins, and N. J. Podraza, “Characterization of structure and growth evolution for nc-Si:H in the tandem photovoltaic device configuration,” IEEE J. Photovoltaics 5, 1516 (2015).
Humlíček et al., 2000 J. Humlíček, R. Henn, and M. Cardona, “Infrared vibrations in LaSrGaO4 and LaSrAlO4,” Phys. Rev. B 61, 14554 (2000).
Humlíček, 2005 J. Humlíček, “Polarized Light and Ellipsometry”, Handbook of Ellipsometry, Ed. H. G. Tompkins, Norwich, NY: William Andrew, Inc., 2005, pp. 3-91.
138 Ibdah et al., 2018 A. R. A. Ibdah, P. Koirala, P. Aryal, P. Pradhan, M. J. Heben, N. J. Podraza, S. Marsillac, R. W. Collins, “Optical simulation of external quantum efficiency spectra of CuIn1−xGaxSe2 solar cells from spectroscopic ellipsometry inputs,” J. Energy Chem. 27, 1151–1169 (2018).
Jackson, 1999 J. D. Jackson, Classical Electrodynamics, New York, NY: Wiley, 1999, p. 2.
Jellison & Modine, 1996 G. E. Jellison and F. A. Modine, “Parameterization of the optical functions of amorphous materials in the interband region,” Appl. Phys. Lett. 69, 371 (1996).
and
G. E. Jellison and F. A. Modine, “Erratum: ‘Parameterization of the optical functions of amorphous materials in the interband region’ [Appl. Phys. Lett. 69, 371 (1996)],” Appl. Phys. Lett., 69, 2137 (1996).
Jezierski, 1998 A. Jezierski, “Band structure of LaSrGaO4 and LaSrAlO4 compounds,” Phys. Status Solidi B 207, 183 (1998).
Jiménez-Sandoval et al., 1992 S. Jiménez-Sandoval, M. Meléndez-Lira, and I. Hernández-Calderón, “Crystal structure and energy gap of CdTe thin films grown by radio frequency sputtering,” J. Appl. Phys. 72, 4197 (1992).
Johs & Herzinger, 2008 B. Johs and C. M. Herzinger, “Quantifying the accuracy of ellipsometer systems,” Phys. Status Solidi C 5, 1031 (2008).
Johs et al., 1998 B. Johs, C. M. Herzinger, J. H. Dinan, A. Cornfeld, and J. D. Benson, Development of a parametric optical constant model for Hg1-xCdxTe for control of composition by spectroscopic ellipsometry during MBE growth, Thin Solid Films 137, 313 (1998).
Johs et al., 1999 B. Johs, J. A. Woollam, C. Herzinger, J. Hilfiker, R. Synowicki, and C. Bungay, “Overview of Variable Angle Spectroscopic Ellipsometry (VASE), Part II: Advanced Applications,” SPIE Proc. CR72, 29 (1999).
139 Junda et al., 2015 M. M. Junda, A. Shan, P. Koirala, R. W. Collins, N. J. Podraza, “Spectroscopic ellipsometry applied in the full p-i-n a-Si:H solar cell device configuration,” IEEE J. Photovoltaics 5, 307 (2015).
Junda, 2017 M. M. Junda, “Spectroscopic Ellipsometry as a Versatile, Non-Contact Probe of Optical, Electrical, and Structural Properties in Thin Films: Applications in Photovoltaics,” Ph.D. Dissertation, The University of Toledo, Toledo, OH (2017).
Kamba et al., 1998 S. Kamba, E. Buixaderas, and A. Pajaczkowska, “Polarized infrared reflectivity spectra of SrLaAlO4 and SrLaGaO4 single crystals,” Phys. Status Solidi A 168, 317 (1998).
Karki Gautam et al., 2014 L. Karki Gautam, H. Haneef, M. M. Junda, D. B. Saint John, and N. J. Podraza, “Approach for extracting complex dielectric function spectra in weakly-absorbing regions,” Thin Solid Films 571, 548 (2014).
Karki Gautam et al., 2016 L. Karki Gautam, M. M. Junda, H. F. Haneef, R. W. Collins, and N. J. Podraza, “Spectroscopic ellipsometry studies of n-i-p hydrogenated amorphous silicon based photovoltaic devices,” Materials 9, 128 (2016).
Karki Gautam, 2016 L. Karki Gautam, “Spectroscopic Ellipsometry Studies of Thin Film Si:H aterials in Photovoltaic Applications from Infrared to Ultraviolet,” Ph.D. Dissertation, The University of Toledo, Toledo, OH (2016).
Kim et al., 1997 C. C. Kim, M. Daraselia, J. W. Garland, and S. Sivanathan, “Temperature dependence of the optical properties of CdTe,” Phys. Rev. B 56, 4786 (1997).
Kim et al., 2014 D. Y. Kim, R. Santbergen, K. Jäger, M. Sever, J. Krč, M. Topič, S. Hänni, C. Zhang, A. Heidt, M. Meier, R. A. C. M. M. van Swaaij, M. Zeman, “Effect of substrate morphology slope distributions on light scattering, nc-Si:H film growth, and solar cell performance,” ACS Appl. Mater. Interfaces 6, 22061 (2014).
Koh et al., 1996 J. Koh, Y. Lu, C. R. Wronski, Y. Kuang, R. W. Collins, T. T. Tsong,Y. E. Strausser, “Correlation of real time spectroellipsometry and atomic force microscopy measurements of surface roughness on amorphous semiconductor thin films,” App. Phys. Lett. 69, 1297 (1996).
140 Koh et al., 1998a J. Koh, H. Fujiwara, Y. Lu, C. R. Wronski, and R. W. Collins, “Real time spectroscopy ellipsometry for characterization and optimization of amorphous silicon-based solar cell structures,” Thin Solid Films 313, 469 (1998).
Koh et al., 1998b J. Koh, Y. Lee, H. Fujiwara, C. R. Wronski, and R. W. Collins, “Optimization of hydrogenated amorphous silicon p-i-n solar cells with two-step i layers guided by real-time spectroscopic ellipsometry,” Appl. Phys. Lett. 73, 1526 (1998).
Koh et al., 2000 J. Koh, A. S. Ferlauto, P. I. Rovira, R. J. Koval, C. R. Wronski, and R. W. Collins, “Evolutionary phase diagrams for the deposition of silicon films from hydrogen- diluted silane,” J. Non-Cryst. Solids 266–269, 43 (2000).
Koirala et al., 2016 P. Koirala, J. Li, H. P. Yoon, P. Aryal, S. Marsillac, A. A. Rockett, N. J. Podraza, and R. W. Collins, “Through-the-glass spectroscopic ellipsometry for analysis of CdTe thin-film solar cells in the superstrate configuration,” Prog. Photovolt: Res. Appl. 24, 1055 (2016).
Kroll et al., 1996 U. Kroll, J. Meier, A. Shah, S. Mikhailov, and J. Weber, “Hydrogen in amorphous and microcrystalline silicon films prepared by hydrogen dilution,” J. Appl. Phys. 80, 4971 (1996).
Kryukov et al., 2009 Y. A. Kryukov, N. J. Podraza, R. W. Collins, and J. G. Amar, “Experimental and theoretical study of the evolution of surface roughness in amorphous silicon films grown by low-temperature plasma-enhanced chemical vapor deposition,” Phys. Rev. B 80, 085403 (2009).
Kulkarni et al., 2017 R. Kulkarni, S. Rondiya, A. Pawbake, R. Waykar, A. Jadhavar, V. Jadkar, A. Bhorde, A. Date, H. Pathan, and S. Jadkar, “Structural and optical properties of CdTe thin films deposited using RF magnetron sputtering,” Energy Procedia 110, 188 (2017).
Langford et al., 1992 A. A. Langford, M. L. Fleet, B. P. Nelson, W. A. Lanford, and N. Maley, “Infrared absorption strength and hydrogen content of hydrogenated amorphous silicon,” Phys. Rev. B 45, 13367 (1992).
141 Lautenschlager et al., 1987 P. Lautenschlager, M. Garriga, L. Vina, and M. Cardona, “Temperature dependence of the dielectric function and interband critical points in silicon.” Phys. Rev. B 36, 4821 (1987).
Lee et al., 1998 J. Lee, P. I. Rovira, I. An, and R. W. Collins, “Rotating-compensator multichannel ellipsometry: applications for real time stokes vector spectroscopy of thin film growth,” Rev. Sci. Instrum. 69, 1800 (1998).
Leem & Yu, 2011 J. W. Leem and J. S. Yu, “Glancing angle deposited ITO films for efficiency enhancement of a-Si:H/μc-Si:H tandem thin film solar cells,” Optics Express, 19, A258 (2011).
Li et al., 2011 J. Li, J. Chen, and R. W. Collins, “Optical transition energies as a probe of stress in polycrystalline CdTe thin films,” Appl. Phys. Lett., 99, 061905 (2011).
Li et al., 2009 J. Li, J. Chen, M. N. Sestak, C. Thornberry, and R. W. Collins, “Spectroscopic ellipsometry studies of thin film CdTe and CdS: From dielectric functions to solar cell structures,” Proc. IEEE 34th Photovoltaics Specialists Conf., (2009), pp. 001982-001987 (DOI: 10.1109/PVSC.2009.5411520).
Li et al., 2008 J. Li, N. J. Podraza, and R. W. Collins, “Real time spectroscopic ellipsometry of sputtered CdTe, CdS, and CdTe1-X SX thin films for photovoltaic applications,” phys. stat. sol. (a) 205, 901 (2008).
Li et al., 2014 C. Li, Y. Wu, J. Poplawsky, T. J. Pennycook, N. Paudel, W. Yin, S. J. Haigh, M. P. Oxley, A. R. Lupini, M. Al-Jassim, S. J. Pennycook, and Y. Yan, “Grain- boundary-enhanced carrier collection in CdTe solar cells,” Phys. Rev. Lett., 112, 156103 (2014).
Lichter & Chen, 1986 S. Lichter and J. Chen, “Model for columnar microstructure of thin solid films,” Phys. Rev. Lett. 56, 1396 (1986).
Liu et al., 2016 W. Liu, L. Zhang, R. Chen, F. Meng, W. Guo, J. Bao, and Z. Liu, “Underdense a- Si:H film capped by a dense film as the passivation layer of a silicon heterojunction solar cell,” J. Appl. Phys. 120, 175301 (2016).
142 Lucovsky et al., 1979 G. Lucovsky, R. J. Nemanich, and J. C. Knights, “Structural interpretation of the vibrational spectra of a-Si: H alloys,” Phys. Rev. B 19, 2064 (1979).
Macco et al., 2017 B. Macco, J. Melskens, N. J. Podraza, K. Arts, C. Pugh, O. Thomas, and W. M. M. Kessels, “Correlating the silicon surface passivation to the nanostructure of low- temperature a-Si:H after rapid thermal annealing,” J. Appl. Phys. 122, 035302 (2017).
Mahan et al., 1991 A. H. Mahan, J. Carapella, B. P. Nelson, R. S. Crandall, and I. Balberg, “Deposition of device quality, low H content amorphous silicon,” J. Appl. Phys. 69, 6728–6730 (1991).
Mai et al., 2005 Y. Mai, S. Klein, R. Carius, J. Wolff, A. Lambertz, F. Finger, and X. Geng, “Microcrystalline silicon solar cells deposited at high rates,” J. Appl. Phys. 97, 114913 (2005).
Marsillac et al., 2007 S. Marsillac, V. Y. Parikh, and A. D. Compaan, “Ultra-thin bifacial CdTe solar cell,” Sol. Energy Mater. Sol. Cells 91, 1398 (2007).
Masuda et al., 2012 T. Masuda, N. Sotani, H. Hamada, Y. Matsuki, and T. Shimoda, “Fabrication of solution-processed hydrogenated amorphous silicon single-junction solar cells,” Appl. Phys. Lett. 100, 253908 (2012).
Matsui, 2015 T. Matsui, A. Bidiville, K. Maejima, H. Sai, T. Koida, T. Suezaki, M. Matsumoto, K. Saito, I. Yoshida, and M. Kondo, “High‐efficiency amorphous silicon solar cells: Impact of deposition rate on metastability,” Appl. Phys. Lett. 106, 053901 (2015).
Matsumura et al., 1986 H. Matsumura, “Catalytic chemical vapor deposition (CTL-CVD) method producing high quality hydrogenated amorphous silicon,” Jpn. J. Appl. Phys. 25, L949 (1986).
McCandless & Sites, 2003 B. McCandless and J. Sites, in Handbook of Photovoltaic Science and Engineering, edited by A. Luque and S. Hegedus, (Wiley, Chichester, England, 2003).
143 Melskens et al., 2012 J. Melskens, A. H. M. Smets, S. W. H. Eijt, H. Schut, E. Brück, and M. Zeman, “The nanostructural analysis of hydrogenated silicon films based on positron annihilation studies,” J. Non-Cryst. Solids 358, 2015 (2012).
Melskens et al., 2013 J. Melskens, A. H. M. Smets, M. Schouten, S. W. H. Eijt, H. Schut, and M. Zeman, “New insights in the nanostructure and defect states of hydrogenated amorphous silicon obtained by annealing,” IEEE J. Photovoltaics 3, 65 (2013).
Melskens et al., 2017 J. Melskens, S. W. H. Eijt, M. Schouten, A. S. Vullers, A. Mannheim, H. Schut, B. Macco, M. Zeman, and A. H. M. Smets, “Migration of open volume deficiencies in hydrogenated amorphous silicon during annealing,” IEEE J. Photovoltaics 7, 421 (2017).
Morrow et al., 2006 P. Morrow, F. Tang, T. Karabacak, P. I. Wang, D. X. Ye, G. C. Wang, and T. M. Lu, “Texture of Ru columns grown by oblique angle sputter deposition,” J. Vac. Sci. Technol. A 24, 235 (2006).
Moustakas et al., 1985 T. D. Moustakas, H. P. Maruska, and R. Friedman, “Properties and photovoltaic applications of microcrystalline silicon films prepared by rf reactive sputtering,” J. Appl. Phys. 58, 983 (1985).
Moustakas, 1984 T. D. Moustakas, in Semiconductors and Semimetals, edited by J. I. Pankove (Academic Press, N.Y., 1984), Vol. 21A, Ch. 4.
Muthmann et al., 2011 S. Muthmann, F. Köhler, M. Meier, M. Hülsbeck, R. Carius, and A. Gordijn, “Monitoring of the growth of microcrystalline silicon by plasma-enhanced chemical vapor deposition using in-situ Raman spectroscopy,” Phys. Status Solidi RRL 5, 144 (2011).
Nieuwenhuizen & Haanstra, 1966 J. M. Nieuwenhuizen and H. B. Haanstra, “Microfractography of thin films,” Philips Tech. Rev. 27, 87 (1966).
Oldham, 1969 W. Oldham, "Numerical techniques for the analysis of lossy films". Surf. Sci. 16, 97 (1969).
144 Pajaczkowska & Gloubokov, 1998 A.Pajaczkowska and A. Gloubokov, “Synthesis, growth and characterization of tetragonal ABCO4 crystals,” Prog. Cryst. Growth Charact. 36, 123 (1998).
Pankov, 1975 J. I. Pankov, “Optical processes in semiconductors: perturbation of semiconductors by external parameters," Dover, New York, 1975, p. 37.
Paudel et al., 2012 N. R. Paudel, K.A. Wieland, A.D. Compaan, “Ultrathin CdS/CdTe solar cells by sputtering,” Sol. Energy Mater. Sol. Cells 105, 109 (2012).
Paul & Anderson, 1981 W. Paul and D. A. Anderson, “Properties of amorphous hydrogenated silicon, with special emphasis on preparation by sputtering,” Sol. Energy Mater. 5, 229 (1981).
Pearce et al., 2007 J. M. Pearce, N. Podraza, R. W. Collins, M. M. Al-Jassim, K. M. Jones, J. Deng, and C. R. Wronski, “Optimization of open circuit voltage in amorphous silicon solar cells with mixed-phase (amorphous + nanocrystalline) p-type contacts of low nanocrystalline content,” J. Appl. Phys. 101, 114301 (2007).
Podraza et al., 2005a N. J. Podraza, Chi Chen, D. Sainju, O. Ezekoye, M. W. Horn, C. R. Wronski, and R. W. Collins, “Transparent conducting oxide sculptured thin films for photovoltaic applications,” Proc. Mat. Res. Soc. Symp. 865, F7.1.1-6 (2005).
Podraza et al., 2005b N. J. Podraza, Chi Chen, J. M. Flores, D. Sainju, llsin An, G. M. Ferreira, C. R. Wronski, M. W. Horn, R. Messier, and R. W. Collins, “Optical properties of transparent conducting oxide sculptured thin films silicon photovoltaics,” Proc. IEEE 31st Photovoltaics Specialists Conf., 1428 (2005) (DOI: 10.1109/PVSC.2005.1488409).
Podraza et al., 2006 N. J. Podraza, C. R. Wronski, and R. W. Collins, “Model for the amorphous roughening transition in amorphous semiconductor deposition,” J. Non-Cryst. Solids 352, 950 (2006).
Podraza et al., 2009 N. J. Podraza, J. Li, C. R. Wronski, E. C. Dickey, and R. W. Collins, “Analysis of controlled mixed-phase (amorphous + microcrystalline) silicon thin films by real time spectroscopic ellipsometry,” J. Vac. Sci. Technol. A 27, 1255 (2009).
145 Povolny et al., 2000 H. Povolny, P. Agarwal, S. Han, and X. Deng, “Comparison study of a-SiGe solar cells and materials deposited using different hydrogen dilution,” Mater. Res. Soc. Symp. Proc. 609, A30.3.1-6 (2000).
Rajan et al., 2017 G. Rajan, T. Miryala, S. Karki, R.W. Collins, N. Podraza, S. Marsillac, “Influence of deposition parameters on silicon thin films deposited by magnetron sputtering,” In Proceedings of the 44th IEEE Photovoltaic Specialists Conference, Washington, DC, (2017), pp. 1–3.
Rakhshani, 1997 A. E. Rakhshani, “Electrodeposited CdTe-optical properties,” J. Appl. Phys. 81, 12 (1997).
Remes et al., 1998 Z. Remes, M. Vanecek, P. Torres, U. Kroll, A. H. Mahan, and R. S. Crandall, “Optical determination of the mass density of amorphous and microcrystalline silicon layers with different hydrogen contents,” J. Non-Cryst. Solids 227-230, 876 (1998).
Rossnagel, et al., 1990 M. S. Rossnagel, J. J. Cuomo, and W. D. Westwood, Handbook of Plasma Processing Technology, Noyes Publication, Park Ridge, New Jersey, 1990.
Ryba-Romanowski et al., 1995 W. Ryba-Romanowski, S. Golab, I. Sokolska, W. A. Pisarski, G. Dominiak-Dzik, A. Pajaczkowska, and M. Berkowski, “Anisotropy of optical properties of SrLaAlO4 and SrLaAlO4:Nd,” J. Alloy. Compd. 217, 263 (1995).
Sai & Kondo, 2009 H. Sai and M. Kondo, “Effect of self-orderly textured back reflectors on light trapping in thin-film microcrystalline silicon solar cells,” J. Appl. Phys. 105, 094511 (2009).
Sai et al., 2012 H. Sai, K. Saito, and M. Kondo, “Enhanced photocurrent and conversion efficiency in thin-film microcrystalline silicon solar cells using periodically textured back reflectors with hexagonal dimple arrays,” Appl. Phys. Lett. 101, 173901 (2012).
Sai et al., 2015 H. Sai, K. Maejima, T. Matsui, T. Koida, M. Kondo, S. Nakao, Y. Takeuchi, H. Katayama, and I. Yoshida, “High‐efficiency microcrystalline silicon solar cells on honeycomb textured substrates grown with high‐rate VHF plasma‐enhanced chemical vapor deposition,” Jpn. J. Appl. Phys. 54, 08KB05 (2015).
146 Scherrer, 1918 P. Scherrer, Nachrichten von der Gesellschaft der Wissenschaften zu Göttingen, Mathematisch-Physikalische Klasse 2, 98 (1918).
Sekimoto et al., 2014 T. Sekimoto, M. Matsumoto, M. Hishida, and A. Terakawa, “Impact of microcrystalline-silicon surface-morphology on film quality and solar cell performance,” Jpn. J. Appl. Phys. 53, 095501 (2014).
Shah et al., 2003 A. V. Shah, J. Meier, E. Vallat-Sauvain, N. Wyrsch, U. Kroll, C. Droz, and U. Graf, “Material and solar cell research in microcrystalline silicon,” Sol. Energy Mater. Sol. Cells 78, 469 (2003).
Shannon et al., 1992 R. D. Shannon, R. A. Oswald, J. B. Parise, B. H. T. Chai, P. B. Yszewski, A. Pajaczkowska, and R. Sobolewski, J. Solid State Chem. 98, 90 (1992).
Shindo et al., 1984 M. Shindo, S. Sato, I. Myokan, S. Mano, and T. Shibata, “High rate preparation of a-Si:H by reactive evaporation method,” Jpn. J. Appl. Phys. 23, 273 (1984).
Sivec et al., 2013 L. Sivec, B. Yan, G. Yue, J. Owens-Mawson, J. Yang, S. Guha, “Advances in light trapping for hydrogenated nanocrystalline silicon solar cells,” IEEE J. Photovoltaics 3, 27 (2013).
Smets & van de Sanden, 2007 A. H. M. Smets and M. C. M. van de Sanden, “Relation of the Si-H stretching frequency to the nanostructural Si-H bulk environment,” Phys. Rev. B 76, 073202 (2007).
Smets et al., 2003 A. H. M. Smets, W. M. M. Kessels, and M. C. M. van de Sanden, “Vacancies and voids in hydrogenated amorphous silicon,” Appl. Phys. Lett. 82, 1547 (2003).
Smets et al., 2008 A. H. M. Smets, T. Matsui, and M. Kondo, “High-rate deposition of microcrystalline silicon p-i-n solar cells in the high pressure depletion regime,” J. Appl. Phys. 104, 034508 (2008).
Smets et al., 2008 A. H. M. Smets, T. Matsui, and M. Kondo, “Infrared analysis of the bulk silicon- hydrogen bonds as an optimization tool for high-rate deposition of microcrystalline solar cells,” Appl. Phys. Lett. 92, 033506 (2008).
147 Sobolewski et al., 1991 R. Sobolewski, P. Gierlowski, W. Kula, S. Zarembinski, S. J. Lewandowski, M. Berkowski, A. Pajaczkowska, B. P. Gorshunov, D. B. Lyudmirsky, and O. I. Sirotinsky, “High-T/sub c/ thin films on low microwave loss alkaline-rare-earth- aluminate crystals,” IEEE Trans. Magn. 27, 876 (1991).
Staebler & Wronski, 1977 D. L. Staebler and C. R. Wronski, “Reversible conductivity changes in discharge- produced amorphous Si,” Appl. Phys. Lett. 31, 292 (1977).
Stoke et al., 2008 J. A. Stoke, N. J. Podraza, J. Li, X. Cao, X. Deng, and R. W. Collins, “Advanced deposition phase diagrams for guiding Si:H-based multijunction solar cells,” J. Non-Cryst. Solids 354, 2435 (2008).
Stuckelberger et al., 2013 M. Stuckelberger, M. Despeisse, G. Bugnon, J. W. Schüttauf, F. J. Haug, and C. Ballif, “Comparison of amorphous silicon absorber materials: Light-induced degradation and solar cell efficiency,” J. Appl. Phys. 114, 154509 (2013).
Stuckelberger et al., 2017 M. Stuckelberger, R. Biron, N. Wyrsch, F. J. Huag, and C. Ballif, “Progress in solar cells from hydrogenated amorphous silicon,” Renewable Sustainable Energy Rev. 76, 1497 (2017).
Tait et al., 1993 R. N. Tait, T. Smy, and M. J. Brett, “Modelling and characterization of columnar growth in evaporated films,” Thin Solid Films 226, 196 (1993).
Tanaka et al., 1992 M. Tanaka, M. Taguchi, T. Matsuyama, T. Sawada, S. Tsuda, S. Nakano, H. Hanafusa, and Y. Kuwano, “Development of new a-Si/c-Si heterojunction solar cells:ACJ-HIT (artificially constructed junction-heterojunction with intrinsic thin- layer),” Jpn. J. Appl. Phys. 31, 3518 (1992).
Tauc et al., 1966 J. Tauc, R. Grigorovici, and A. Vancu, Optical properties and electronic structure of amorphous germanium, Phys. Status Solidi B 15, 627 (1966).
Tiedie et al., 1981 T. Tiedje, T. D. Moustakas, and J. M. Cebulka, “Effect of hydrogen on the density of gap states in reactively sputtered amorphous silicon,” Phys. Rev. B 23, 5634 (1981).
148 Tiwald et al., 1998 T. E. Tiwald, D. W. Thompson, J. A. Woollam, W. Paulson, and R. Hance, “Application of IR variable angle spectroscopic ellipsometry to the determination of free carrier concentration depth profiles,” Thin Solid Films 313-314, 661 (1998).
Tokas et al., 2016 R. B. Tokas, S. Jena, S. M. Haque, K. D. Rao, S. Thakur, and N. K. Sahoo, “Spectroscopic ellipsometry investigations of optical anisotropy in obliquely deposited Hafnia thin films,” AIP Conf. Proc. 1731, 060007 (2016) (DOI: 10.1063/1.4947813).
Tong et al., 2012 G. B. Tong, Z. Aspanut, M. R. Muhamad, and S. A. Rahman, “Optical properties and crystallinity of hydrogenated nanocrystalline silicon (nc-Si:H) thin films deposited by rf-PECVD,” Vacuum 86, 1195 (2012).
Tuteja et al., 2015 M. Tuteja, P. Koirala, S. MacLaren, R. Collins, and A. Rockett, “Direct observation of electrical properties of grain boundaries in sputter-deposited CdTe using scan- probe microwave reflectivity based capacitance measurements,” Appl. Phys. Lett. 107, 142106 (2015).
Vetterl et al., 2000 O. Vetterl, F. Finger, R. Carius, P. Hapke, L. Houben, O. Kluth, A. Lambertz, A. Muck, B. Rech, and H. Wagner, “Intrinsic microcrystalline silicon: A new material for photovoltaics,” Sol. Energy Mater. Sol. Cells 62, 97 (2000).
Visoly-Fisher et al., 2004 I. Visoly-Fisher, S. R. Cohen, A. Ruzin, and D. Cahen, “How polycrystalline devices can outperform single-crystal ones: thin film CdTe/CdS solar cells,” Adv. Mater., 16, 879 (2004).
Wang et al., 2006 S. Wang, X. Fu, G. Xia, J. Wang, J. Shao, and Z. Fan, “Structure and optical properties of ZnS thin films grown by glancing angle deposition,” Applied Surface Science 252, 8734 (2006).
Wang et al., 2011 D. Wang, Z. Yang, F. Li, and D. He, “The microstructure and optical properties of crystallized hydrogenated silicon films prepared by very high frequency glow discharge,” Appl. Surf. Sci. 257, 8350 (2011).
Wronski & Collins, 2004 C. R. Wronski and R. W. Collins, “Phase engineering of a-Si:H solar cells for optimized performance,” Sol. Energy 77, 877 (2004).
149 Wu, 2004 X. Wu, “High-efficiency polycrystalline CdTe thin-film solar cells,” Solar Energy 77, 803 (2004).
Wu et al., 2002 X. Wu, R.G. Dhere, Y. Yan, M.J. Romero, Y. Zhang, J. Zhou, C. DeHart, A. Duda, C. Perkins, and B. To, “High-efficiency polycrystalline CdTe thin-film solar cells with an oxygenated amorphous CdS (a-CdS:O) window layer,” Proc. IEEE 29th Photovoltaics Specialists Conf., (2002), pp. 1-4 (DOI: 10.1109/PVSC.2002.1190619).
Xu et al., 2011 L. Xu, Z. P. Li, C. Wen, and W. Z. Shen, “Bonded hydrogen in nanocrystalline silicon photovoltaic materials: Impact on structure and defect density,” J. Appl. Phys. 110, 064315 (2011).
Yanh & Guha, 1999 J. Yang and S. Guha, “Amorphous silicon alloy materials and solar cells near the threshold of microcrystallinity,” Mater. Res. Soc. Symp. Proc. 557, 239 (1999).
Yuguchi et al., 2012 T. Yuguchi, Y. Kanie, N. Matsuki, and H. Fujiwara, “Complete parameterization of the dielectric function of microcrystalline silicon fabricated by plasma-enhanced chemical vapor deposition,” J. Appl. Phys. 111, 083509 (2012).
Zapien et al., 2005 J. A. Zapien, J. Chen, J. Li, J. Inks, N. S. Podraza, C. Chen, J. Drayton, A. Vasko, A. Gupta, S. L. Wang, R. W. Collins, and A. D. Compaan, “Real time spectroscopic ellipsometry of thin film CdTe deposition by magnetron sputtering for photovoltaic applications,” IEEE 31st Photovoltaics Specialists Conf., (2005), pp. 461-464 (DOI: 10.1109/PVSC.2005.1488169).
Zhao et al., 2004 Z. X. Zhao, R. Q. Cui, F. Y. Meng, B. C. Zhao, H. C. Yu, and Z. B. Zhou, “Nanocrystalline silicon thin films prepared by RF sputtering at low temperature and heterojunction solar cell,” Mater. Letter 58, 3963 (2004).
Zimina et al., 2003 G. V. Zimina, A. V. Novoselov, L. P. Ogorodova, A. V. Filaretov, L. N. Komissarova, and A. Pajaczkowska, “Thermodynamic properties of SrLaAlO4– SrLaGaO4 solid solutions,” Inorg. Mater. 39, 63 (2003).
150