A Dissertation Entitled Development of Back Contacts for Cdte Thin Films Solar Cells by Fadhil Khalaf Dahash Alfadhili Submit

Home , Cadmium Zinc Telluride, Zinc telluride

A Dissertation

entitled

Development of Back Contacts for CdTe Thin Films Solar Cells

Fadhil Khalaf Dahash Alfadhili

Submitted to the Graduate Faculty as partial fulfillment of the requirements for the

Doctor of Philosophy Degree in Physics

Dr. Michael J. Heben, Committee Chair

Dr. Randy J. Ellingson, Committee Member

______Dr. Robert Collins, Committee Member

Dr. Nikolas Podraza, Committee Member

Dr. Daniel Georgiev, Committee Member

______Dr. Amanda C. Bryant-Friedrich, Dean College of Graduate Studies

The University of Toledo

May 2020

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

Development of Back Contacts for CdTe Thin Films Solar Cells

Fadhil Khalaf Dahash Alfadhili

Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Doctor of Philosophy Degree in Physics

The University of Toledo May 2020

Thin film solar cells based on polycrystalline p-type cadmium telluride (CdTe) represent one of these the most promising photovoltaic (PV) device due to high efficiency and low-cost production. Currently, CdTe solar cells provide the lowest cost electricity generation in utility-scale applications, which is a cost-competitive with the traditional power source, fossil fuel. CdTe thin film PV has attained 22.1 % of power conversion efficiency for small area scale and 18.6 % for modules scale. However, the high efficiency of CdTe devices has been achieved by increasing the photo-generated current by changing the traditional window layer (CdS) of CdTe to a wider bandgap material with better band alignment. The open-circuit voltage (VOC) remains below the theoretical limit due to a barrier at the back of the device due to the deep valence band edge of CdTe (-5.9 eV). Voc can be increased by adding a buffer layer between CdTe and the back electrode to decrease band banding and reducing carrier recombination at the back interface. In this thesis, several materials were investigated as a back-buffer layer, such as single-wall carbon nanotube (SWCNT), zinc telluride (ZnTe), tellurium (Te), and cadmium zinc telluride

(CZT) to minimize the bend bending at CdTe/back-buffer layer interface. An alternative method to reduce the carrier recombination at the rear surface, the use of aluminum oxide

iii (Al2O3) layer as a passivation layer was also demonstrated. Finally, an effective method of

CdCl2 treatment for CZT thin film was investigated. This method shows that zinc (Zn) can be maintained during the heat treatment.

Acknowledgements

First, I would like to express my deepest and sincere gratitude and appreciation to my advisor, Dr. Michael Heben, for his excellent guidance and patience through my academic study. I would also like to thank my committee members, Dr. Randy Ellingson,

Dr. Robert Collins, Dr. Nikolas Podraza, and Dr. Daniel Georgiev for their valuable advice and feedback. My special thanks go to Dr. Adam Phillips for his unlimited support and guidance throughout my academic research journey. Also, I would like to thank my colleagues, who helped me throughout this journey. Furthermore, I would like to thank all the faculty members and staff in the Department of Physics and Astronomy and the Wright

Center for Photovoltaics Innovation and Commercialization (PVIC). My appreciation and thanks also go to the Higher Committee for Education Development in Iraq (HCED) for sponsoring my studies.

Finally, special thanks and deepest gratitude to my wife, Jannah, for understanding the challenges, for her patience and helpfulness, as well as my kids: Zahraa, Narjes,

Hussien, and Zainab. Besides, I am thankful to my mother, brothers, and sisters for their support and encouragement.

Table of Contents

Abstract ...... iii

Acknowledgements ...... vi

Table of Contents ...... vii

List of Tables ...... xii

List of Figures ...... xiii

List of Abbreviations ...... xx

List of Symbols ...... xxii

1 Introduction and Motivation ...... 1

1.1 Motivation ...... 1

1.2 Solar Cell Basics ...... 3

1.3 Solar Cell Parameters ...... 4

1.4 CdTe Solar Cells ...... 7

1.5 CdTe Device Components ...... 7

1.5.1 Substrate and Front Contact ...... 8

1.5.2 Absorber layer CdTe ...... 8

1.5.3 Window layer (emitter) ...... 9

1.5.4 CdCl2 Activation ...... 9

1.5.5 Back Contact ...... 10

1.6 Dissertation Overview ...... 11

vi 2 Use of Single Wall Carbon Nanotube films doped with Triethyloxonium

Hexachlorantimonate as a Transparent Back Contact for CdTe Solar Cells ...... 13

2.1 Introduction ...... 14

2.2 Experimental Details ...... 16

2.3 Results and Discussion ...... 17

2.3.1 SWCNT on Glass Substrate ...... 17

2.3.2 SWCNT Film on CdTe ...... 18

2.3.3 Devices Performance ...... 19

2.3.4 The Differences in SWCNT Doping on Glass and CdTe Substrates21

2.3.5 The Activation Energy of OA Doping ...... 22

2.3.6 OA Interacting with CdTe ...... 24

2.4 Conclusion ...... 26

3 Controlling Band Alignment at the Back Interface of Cadmium Telluride Solar

Cells Using ZnTe and Te Buffer Layers ...... 27

3.1 Introduction ...... 28

3.2 Experimental Details ...... 29

3.3 Results and Discussion ...... 31

3.3.1 Raman Spectroscopy Analysis ...... 31

3.3.2 Device Performances ...... 32

3.3.3 Auger Depth Profile ...... 32

3.3.4 Back Barrier Height Analysis ...... 34

3.4 Conclusion ...... 36

4 Potential of Cd1-XZXnTe Thin Film Back Buffer Layer for CdTe Solar Cells ....38

vii 4.1 Introduction ...... 39

4.2 Experimental Details ...... 40

4.3 Results and Discussion ...... 40

4.3.1 Optical Properties of CZT Film ...... 40

4.3.2 Electric Properties of CZT Film doped with Cu ...... 41

4.3.3 Device Performance ...... 42

4.3.4 Barrier Height Measurements ...... 44

4.4 Conclusion ...... 46

5 Back Contact Passivation of CdTe Solar Cells by Solution-Processed Oxidized

Aluminum…… ...... 47

5.1 Introduction ...... 48

5.2 Experimental Details ...... 50

5.3 Results and Discussion ...... 53

5.3.1 The surface characterization ...... 53

5.3.2 Device Performance ...... 56

5.3.3 Device with Cu Doping ...... 59

5.3.6 External Quantum Efficiency Measurements with Bias Voltage .....62

5.4 Conclusion ...... 64

6 Development of CdCl2 Activation to Minimize Zn Loss from Sputtered Cd1-

xZnxTe Thin Films for Use in Tandem Solar Cells ...... 66

6.1 Introduction ...... 67

6.2 Experimental Details ...... 68

6.3 Results and Discussion ...... 68

viii 6.3.1 Optical Properties ...... 68

6.3.2 XRD Spectroscopy Analysis ...... 70

6.3.3 Auger Spectroscopy Analysis ...... 73

6.4 Conclusion ...... 75

7 Summary and Future Research ...... 77

7.1 Thesis Summary...... 77

7.2 Future Research ...... 80

7.2.1 The CZT Back Buffer Layer ...... 80

7.2.2 Using Al2O3 and SWCNTs as a Point Contact ...... 81

References ...... 84

List of Publications ...... 97

ix List of Tables

2.1 J-V characteristics of CdTe devices prepared with different back contacts ...... 20

4.1 Effect of Cu doping on the electrical properties of CZT thin-film ...... 42

4.2 J-V characteristics of CdTe devices prepared with different back contacts ...... 43

5.1 Auger depth signals of the CdTe without and with different coating cycles of

Al2O3...... 55

5.2 J-V performance data for devices fabricated with and without Cu and Al2O3

deposition/heating cycles. Data is presented for the best device and the population

of devices in each data set (n > 20)...... 66

List of Figures

1-1 Energy band diagram of a p-n junction at equilibrium ...... 4

1-2 Current density voltage curve of CdTe solar cell in dark and under light ...... 6

1-3 Series and shunt resistances in a solar cell circuit...... 6

1-4 CdS/CdTe solar cell in superstrate configuration (a) schematic and (b) a cross

section SEM image ...... 8

2-1 (a) The absorbance spectra of a 100 nm thick CoMoCat SWCNT film on glass

before and after doping with OA for 30 sec. (b) variation of sheet resistance with

doping time for SWCNTs films on glass ...... 17

2-2 The transmittance spectra of CdTe, CdTe/SWCNT and CdTe/SWCNT/OA ...... 19

2-3 J-V characteristic of the CdTe devices with different back contacts ...... 20

2-4 OA doping of SWCNT films at 70 °C as a function of time on (a) glass and (b) on

CdTe ...... 22

2-5 Plot of ln (D) vs. 1/KT to evaluate Ea of OA doping for SWCNT on glass and CdTe

...... 23

2-6 J-V characteristic of the CdTe devices with different back contacts ...... 24

2-7 SEM images of CdTe (a) before OA treated (b) after OA treated for 10 min ...... 25

3-1 Eight different back contact structures for CdTe thin film solar cells ...... 30

3-2 Raman spectra of as deposited ZnTe thin film and MAI treated with different

constructions ...... 31

xi 3-3 Performance parameters of CdTe devices with different back contacts (a)

efficiency, (b) Voc, (c) FF, and (d) Jsc ...... 33

3-4 Auger depth profile spectra of the CdTe/Te/ZnTe/Te/Cu/Au device ...... 36

3-5 a) Temperature dependence dark current-voltage (J-V) characteristics for

2 CdTe/Cu/Au device. (b) Plots of ln (J˳/T ) versus 1/KBT for the eight different back

contacts. (c) the values of the back-barrier height ...... 37

4-1 The plot of (α hυ)2 Vs hυ of as deposited CZT film ...... 41

4-2 Current density curves for CdTe devices with CZT/Cu/Au, CZT/Cu/Te/Au, and

standard Cu/Au back contacts ...... 43

4-3 Temperature dependent dark current voltage curves for CdTe devices with (a)

CZT/Cu/Au and (b) CZT/Cu/Te/Au back contacts. (c) Arrhenius plots of ln (J/T2)

vs 1/KT for CdTe/CZT/Cu/Au and CdTe/CZT/Cu/Te/Au devices ...... 45

5-1 SEM images of Al2O3 on CdTe for (a) 0 cycle, (b) 1 cycle, and (c) 5 cycles. (d)

XPS data for these samples ...... 55

5-2 (a) The AFM images of CdTe, (b) of CdTe/Al2O3 (1 cycle), and (c) of CdTe/Al2O3

(5 cycles) samples ...... 56

5-3 (a) J-V characteristics for CdTe/Au and CdTe/Al2O3(1,3,5,7, and 9cycles)/Au

devices. (b) average PCE vs average FF. (c) average VOC vs Tau2 ...... 58

5-4 (a) The average PL spectrum of the CdS/CdTe thin film solar cell passivated by

using the Al2O3. (b) TRPL decay of the CdS/CdTe thin film solar cell passivated

by using the Al2O3 ...... 59

5-5 (a) J-V carves for CdTe/Cu/Au and CdTe/Al2O3 (1 cycle)/Cu/Au devices with

different copper diffusion time ...... 60

xii 5-6 Bias dependent EQE for CdTe/Cu/Au and CdTe/Al2O3/Cu/Au devices. Slope of

EQE in infrared at MPP shows back surface passivation for CdTe/Al2O3/Cu/Au

device...... 64

6-1 Schematic of the system used for CdCl2 treatment ...... 69

2 6-2 The plot of (αhν) vs hν of as-deposited, pre-annealing at 450 °C in H2/He ambient,

CdCl2 activated without flowing gas, with flowing N2, and with flowing dry air

films ...... 70

6-3 (a) XRD spectra of as-deposited, pre-annealing at 450°C in H2/He ambient, CdCl2

activated without flowing gas, with flowing N2, and with flowing dry air films.

Panel (b) is zoom in of normalized of (111) peak ...... 73

6-4 (a) Auger depth profile spectra of as-deposited, (b) CdCl2 activated in a static inert

atmosphere, and (c) CdCl2 activated with flowing dry air films ...... 75

7-1 (a) The SEM image of 5 cycles on CdTe. (b) J-V curves of CdTe device with no

Al2O3 and 5 cycles...... 81

7-2 Schematic of CdTe with a SWCNT point contact/passivation layer (Al2O3) ...... 82

7-3 (a) Fill factor and (b) open circuit voltage of CdTe devices with back buffer layers

consisting of varying ratios of Al2O3 and SWCNTs. (c) J-V crves for the best CdTe

device with different SWCNT concentration with 6.15 mM Al(acac)3. (d) J-V

curves for the best CdTe devices with different SWCNT concentration and with

123 mM Al(acac)3 ...... 82

xiii

List of Abbreviations

Al(acac)3……………Aluminum acetylacetonate ALD…………………Atomic layer deposition AFM…………………Atomic force microscope AES………………….Auger electron spectroscopy

CoMoCat……………Cobalt Molybdenum Catalyst

DCE…………………Dichloroethane DI……...... De-ionized water eV……...... Electron Volt EQE ...... External Quantum Efficiency

IPA ...... Isopropanol IR……………………Infra-red ITO ...... Indium Tin Oxide

MAI ...... Methylammonium Iodide

NIR ...... Near Infrared

OA………………….Triethyloxnium hexachlorantimonate

PCE ...... Power Conversion Efficiency PL ...... Photoluminescence PV ...... Photovoltaic

SDBS….. ……………Dodecyl Benzene Sulfonate SEM ...... Scanning Electron Microscope

TCO...... Transparent Conducting Oxide TRPL ...... Time-Resolved Photoluminescence .

xiv

List of Symbols

η ...... Device efficiency θ ...... Angle 1 ...... Surface chare carrier life time 2 ...... Bulk chare carrier life time ϕb ...... Back barrier height q...... Charge of an electron FF ...... Fill Factor Jo ...... Dark saturation current density JL ...... Current generated JSC ...... Short circuit current density J-V ...... Current density–voltage J-V-T ...... Temperature dependent current Density–voltage K ...... Boltzmann constant MPP...... Maximum power point n...... Diode quality factor R ...... Universal gas constant RS ...... Series resistance RSH ...... Shunt resistance T ...... Temperature V ...... Applied voltage VOC ...... Open circuit voltag

Chapter 1

Introduction and Background

1.1 Motivation

The continuous growth of society and the economy has led to a rapid increase in world energy consumption, which now stands at 18.4 TW(average power used over one year)[1]. Most of this consumed energy is generated from burning fossil fuels such as oil, coal, and natural gas. The fossil fuels are a nonrenewable source of energy which have limited quantities on Earth and may not serve the future world energy market. Additionally, producing electricity from these resources is the main reason why our planet is polluted with carbon dioxide (CO2) and greenhouse gases, which are a significant cause of global warming. Therefore, scientists around the world are focused on developing more cost- effective and environmentally friendly renewable sources of energy that can meet the future energy demand.

One of the most promising forms of renewable energy is solar energy, which is the cleanest and most abundant source. Solar energy can be directly converted into electricity using photovoltaic solar cell technology. A solar cell is an electric device that is operating due to the concept of the photovoltaic effect. The first functional solar cell was

1 demonstrated in 1954[2]. Since then, there has been significant interest in solar cell devices, and scientists have done extensive studies to develop high efficiency and potentially inexpensive solar cells. Today, about 2.4 % (66.7 gigawatt-hour)[1, 3] of the world's electricity is produced by photovoltaic technology, and it is expected to grow to terawatt levels in the next ten years[4]. In general, existing solar cells can be classified in terms of different generations. Crystalline silicon (c-Si) based solar cells represents the first generation, which is identified as the well-developed solar cell and with wide commercially acceptance. The second generation of solar cell is based on thin film materials such as cadmium telluride (CdTe), copper indium gallium diselenide (CIGS), and amorphous silicon (a-Si), emerging as lower-cost solutions. The third generation includes perovskites, quantum dot, dye-sensitized, and organic solar cells. Although some third generation solar cells shows remarkably high conversion efficiency, they still have a problem with stability.

Among the available solar cells, CdTe device shows great potential in achieving high efficiency and low-cost production. Based on modules, it currently provides the most inexpensive kilowatt per hour cost with a power conversion efficiency of 19.1 %, which makes it highly valuable as an energy source. One of the key challenges in CdTe photovoltaic technology is to form a good back contact due to high work function of CdTe.

As an effort to develop the back contact, this dissertation is mainly focused on the engineering of high-performance back contact. Single wall carbon nanotubes (SWCNTs), zinc telluride (ZnTe), tellurium (Te), and cadmium zinc telluride (Cd1-XZnXTe) are utilized as a back buffer layer to attain the high efficiency. Additionally, passivation of the rear surface of CdTe devices using solution processing of aluminum oxide (Al2O3) layer was

2 demonstrated. Finally, minimization of zinc loss from CdZnTe thin film during the CdCl2 treatment was achieved.

1.2 Solar Cell Basics

The main interpretation that is used to describe the operation of a solar cell is based on the concept of a p-n junction. To form the junction, a p-type semiconductor layer and n-type are required. When the layers are brought together, the free electron in the n-region diffuses into the p-region (leaving a positive ion), and free holes diffuse in the opposite direction (leaving a negative ion). The diffusion process continues until the equilibrium condition is satisfied. As a result, a free mobile carriers’ region close to the junction is developed, which is known as space- charge region (SCR) or the depletion region, and a built-in potential is created across SCR (figure 1-1). When light (photons) with energy greater than the band gap of the absorber layer is incident on solar cells, electron hole pairs are generated. The produced electrons and holes are separated due to this built-in potential.

Electrons move towards the front electrode, while holes move towards the back electrode, producing the photocurrent in a circuit.

Figure 1-1. Energy band diagram of a p-n junction at equilibrium.

1.3 Solar Cell Parameters

The main parameters that are used to describe the performance of a solar cell are the open-circuit voltage, the short-circuit current density, fill factor, and photoconversion efficiency. These parameters can be evaluated from the current density-voltage (J-V) characteristic which shows all the possible combinations of current and voltage output of a solar cell. Figure 1-2 shows an example of J-V curve for a solar cell measured in the dark and under illumination (standard 1 kW/m2 spectrum). Under the dark condition, a solar cell acts as a diode, and the net current density (J) flowing through it is[5]:

푞푉 퐽 = 퐽0(푒푘푇 − 1) (1.1)

Where J0 is dark saturation current density, V is the applied voltage, q is the charge of an electron, k is Boltzmann's constant and T is an absolute temperature.

In the presence of light, the net current becomes[5]:

4 푞푉 퐽 = 퐽0 (푒푛푘푇 − 1) − 퐽퐿 (1.2)

Where n is a diode quality factor, and JL is a current generated in the presence of light. The negative sign indicates that the dark current and the photocurrent are in the opposite direction [6], [3]. When J = 0, then V in equation 1.2 represents open-circuit voltage (VOC), which is the voltage dropped across the solar cell when the external load is infinite. The

VOC is given by:

푛퐾푇 퐽퐼 푉푂퐶 = ln ( ) (1.3) 푞 퐽°

At zero voltage bias, V = 0 in equation 1.2, J= Jsc which can be defined as the current density flowing through the solar cell when the circuit is a short circuit.

From figure 1-2, fill factor (FF) represents the ratio of the maximum power (Pmax) from the solar cell to the product of Voc and Jsc. Graphically, it is given by the area of the largest rectangle which will fit in the J-V curve. Mathematically, it is provided by:

퐽 푉 푃 퐹퐹 = 푚푎푥 푚푎푥 = 푚푎푥 (1.4) 퐽푠푐푉표푐 퐽푠푐푉표푐

Using JSC, VOC, and FF, the photoconversion energy (PCE) can be determined, which is defined as the ratio of the maximum output power of the solar cell to the total input power incident (Pinc) on it. i.e.

푃 푉 퐽 퐹퐹 푃퐶퐸 = 푚푎푥 = 푂퐶 푆퐶 (1.5) 푃푖푛푐 푃푖푛푐

Figure 1-2. Current density-voltage curve of CdTe solar cell in dark and under light.

An ideal working solar cell consists of a parallel combination of a photocurrent source, a diode, and shunt resistance (RSH) with a series resistance (RS), as shown in figure

(1-3). Ideally, the shunt is infinitely large, and series resistance is close to zero for a high- performance solar cell.

Figure 1-3. Equivalent circuit of a solar cell showing the series and shunt resistances[5].

6 1.4 CdTe Solar Cells

CdTe semiconducting material is of great interest in thin film photovoltaic technology due to high efficiency and cost-effective manufacturing capabilities. Currently, it is the second most used solar cell in the world and provides the lowest cost production (

$0.29 /WP )[7], which is expected to reach $0.2/Wp[4] or below in the next 10 years. The current world record photoconversion efficiency (PCE) of CdTe thin film solar cell is 22.1

%[8]. However, the current efficiency of the CdTe device stays well below the theoretical limit (~ 30 %)[9] due to the losses of open-circuit voltage, which are arising from high recombination rates. Consequently, further improvements in the performance of CdTe solar cells are needed and represented the most promising contribution of the CdTe photovoltaic (PV) technology.

1.5 CdTe Device Components

Typically, a CdTe thin film solar cell consists of different semiconductors and metal layers that can be fabricated in either superstrate or substrate architecture. The common architecture used in the fabrication of CdTe devices is superstrate. This configuration permits light to enter through the substrate into the CdTe device. Hence, in this structure, the substrate should be highly transparent for light in the solar spectrum. While, in the inverted structure (substrate architecture), light does not enter through the substrate, and an opaque substrate such as metal foils or any conducting substrate can be used.

Figure 1-4 shows the essential layers that are used to fabricate the CdTe device such as glass, front contact, emitter, absorber, and back contact. Each of these layers has an essential role in the operation and the photoconversion efficiency of CdTe device. In the following sections, the properties of these different layers are discussed.

Figure 1-4. (a) Schematic diagram and (b) cross-sectional scanning electron microscopy

(SEM) image of CdS/CdTe solar cell in superstrate configuration.

1.5.1 Substrate and Front Contact

The most common substrate used in CdTe solar cells is 3.2 mm soda-lime glass

(SLG) with low-iron content to reduce the optical absorption. Due to low cost and stability at high temperatures (˃ 450 °C), SLG is identified as a primary substrate for CdTe devices.

The SLG substrate is coated with a transparent conducting oxide (TCO) layer that serves as a front electrical contact (FC). The basic requirements for the TCO layer are high electrical conductivity, high electron mobility, and high transmittance, greater than 80% in the visible spectrum. Different types of TCO can meet these requirements, such as fluorine- doped tin oxide (SnO2:F; FTO), tin-doped indium oxide (InO3: Sn, ITO), and cadmium stannate Cd2SnO4 (CTO) which are extensively used in thin film solar cells.

1.5.2 Absorber layer CdTe

CdTe material is a combination of group II and VI elements that can be n or p-type. p-type CdTe semiconductor has a direct bandgap of 1.45 eV at room temperature, which

8 is close to that of the ideal bandgap (1.4 eV) for solar cells. Moreover, CdTe has a high absorption coefficient, which is ~ 105 cm-1 in the wavelength 400-950 nm range. With this high absorption coefficient, about 2-micron thickness of CdTe thin film is effective enough to absorb most of the visible solar spectrum. On the other hand, because of the low absorption coefficient of c-Si, more than one hundred microns thick material of c-Si is required for complete light absorption. Thus, CdTe is highly attractive for thin film photovoltaic applications.

1.5.3 Window layer (emitter)

The emitter in CdTe photovoltaic devices needs to be an n-type semiconductor to form a p-n junction with the p-CdTe layer. Traditionally, CdS has been used as a window layer due to its thermal and chemical stability to CdTe solar cells deposition. However,

CdS does not facilitate the high performance of CdTe solar cells due to the absorption of light of wavelength below 500 nm and an unfavorable conduction band offset (CBO) at

CdS/CdTe interface which causes high interface recombination. Alternatively, using a wide band gap emitter (3.5 eV) such as zinc magnesium oxide (ZMO)[10] has shown significant improvement in the device performance due to reduction in the recombination at ZMO/CdTe interface and allowing the light of the low wavelengths (below 520 nm) to contribute in the photocurrent generation.

1.5.4 CdCl2 Activation

A cadmium chloride (CdCl2) activation step is essential to obtain high-efficiency polycrystalline CdTe thin film solar cells. This treatment assists recrystallization and grain growth of CdTe films[11], as well as the incorporation of Cl at the grain boundaries of

CdTe[12], all of which leads to higher performance of CdTe thin film solar cells.

1.5.5 Back Contact

Highly efficient back contact is a crucial ingredient for achieving high performance and long term stability of CdTe devices. However, CdTe semiconductor material has a deep valance band energy (~ 5.95 eV)[13] because of its high electron affinity (χ = 4.5 eV) and the bandgap (1.45 eV), which would make it harder to form a suitable ohmic contact.

Even with the noble metals available, it would still make a huge barrier for the holes at the back. The band bending in the conduction band would favor the electron flowing to the back, effectively increasing the recombination at the back interface and limiting the device performance. Historically, doping the back surface of CdTe with copper (Cu) has been used to manage the barrier. Therefore, this would result in reducing the barrier width so that the holes would make it to the back contact easily. However, Cu diffuses to the heterojunction (CdS/CdTe) and reduces the device performance over time[14].

Alternatively, adding a back buffer layer (BBL) between the CdTe and back electrode can effectively reduce the bend bending at the back. So, few electrons would be attracted to the

CdTe/BBL interface resulting in a much lower barrier for holes. Several types of back buffer layers (BBLs) have been demonstrated, such as ZnTe[15], tellurium (Te)[16], single-wall carbon nanotubes (SWCNT)[17], and iron pyrite (FeS2)[18]. With BBL, the device performance has increased even more due to reducing downward band banding at the back, which leads to reduction of the carrier recombination. Among the available BBLs,

ZnTe has widely used as a BBL due to high p-type conductivity and high electron affinity.

However, electrical measurement indicated that a barrier height of ~ 400 meV[19] still exists between CdTe and ZnTe interface, resulting in undesirable band bending

10 Based on previous modeling work [20], a BBL should have an initial Fermi level

(IFL) below that of CdTe to produce a barrier for electrons and no barrier for the holes, resulting in low back surface interface recombination.

1.6 Dissertation Overview

This dissertation focuses on the development of back interface engineering for

CdTe thin film solar cells using two different strategies. The first approach is the addition of semiconducting buffer layers such as single-wall carbon nanotubes (SWCNTs), ZnTe, and CZT to form low back surface recombination. The second approach uses an oxide passivation layer (Al2O3) to repel the electrons from the rear surface. The characterization of materials and devices are extensively studied. Additionally, controlling the zinc loss from CdZnTe thin film during CdCl2 activation was also examined.

Chapter 2 discusses the application of SWCNTs as a transparent back contact for

CdTe solar cells. The electrical and optical properties of SWCNTs thin film were improved through the doping with triethyloxonium hexachlorantimonate (OA). The performance of

CdTe devices with a different back electrode such as Cu/Au, SWCNT/Au, and SWCNT:

OA/Au were compared. Also, the interaction between OA and CdTe was determined.

Chapter 3 compares the use of ZnTe and Te as a BBL for the CdTe device. The performance of the device with ZnTe and Te back buffer layers are provided. Further, temperature dependent current density-voltage (JVT) measurements were applied to determine back barrier height between CdTe and ZnTe or Te.

Chapter 4 presents the use of CdZnTe thin film as BBL for the CdTe device. In this study, the effects of Cu doping on electric properties of CdZnTe film were studied. Also,

11 the performance of a device with Cu/Au, CdZnTe: Cu/Au, and CdZnTe: Cu/Te/Au are compared. Finally, back barrier height between CdTe and CdZnTe was determined using

JVT measurements of several BBL configurations.

Chapter 5 focuses on the study of the rear passivation surface of CdTe solar cells utilizing a solution processed Al2O3 layer. The formation of Al2O3 on the CdTe thin film is detected using atomic force microscope (AFM) and Auger electron spectroscopy (AES).

Also, the impact of the Al2O3 on device performance and carrier lifetimes were investigated.

Chapter 6 presents the development of CdCl2 treatment for wider bandgap CZT thin films to reduce zinc losses during typical processing. CdCl2 activation with different environments was studied. X-ray diffraction, optical measurements, and Auger spectroscopy were used before and after the CdCl2 treatment.

Chapter 7 concludes this dissertation with a summary and future considerations.

12 Chapter 2

Use of Single Wall Carbon Nanotube films doped with Triethyloxonium Hexachlorantimonate as a Transparent Back Contact for CdTe Solar Cells

This chapter represents the application of using the doped single wall carbon nanotube (SWCNT) films as a transparent back contact for CdTe-based devices. Highly conductive transparent films were formed by doping SWCNTs with triethyloxonium hexachlorantimonate (OA). This doping resulted in complete quenching of the SWCNT absorption features in the visible and infra-red (IR) range leaving an IR transparent film.

These doped SWCNT films were investigated for possible use as a transparent back contact for CdTe-based devices. One particular goal is the development of tunnel junction for use in the application of tandem cells. The results in this chapter have been published in

Alfadhili et al., 2017[21]. © [22] IEEE. Printed with permission, from Fadhil K. Alfadhili,

Jacob M. Gibbs, Geethika K. Liyanage, Patrick W. Krantz, Suneth C. Watthage, Zhaoning

Song, Adam B. Phillips, and Michael J. Heben, Use of Single Wall Carbon Nanotube films doped with Triethyloxonium Hexachlorantimonate as a Transparent Back Contact for

CdTe Solar Cells, 2017 IEEE 44th Photovoltaic Specialist Conference (PVSC), and Nov.

2018.

2.1 Introduction

Thin film tandem solar cells have attracted renewed attention recently because of the potential for an excellent combination of high-power conversion efficiency (PCE) and easy manufacturing[23, 24]. In their simplest form, tandem devices are a combination of two cells, a top cell consisting of a higher band gap material (e.g., ~1.7 eV), and a bottom cell with a lower energy band gap (e.g., ~1.1 eV). The top cell absorbs the high-energy photons and transmits infrared and near infrared light to the bottom cell[23, 24]. A recent analysis of commercially available technologies indicates that CdTe-based devices

(currently with a record efficiency of 22.1%[25] ) could be an important wide band gap partner for high efficiency thin film tandem cells[26]. Although CdTe could be paired with

CuInSe2 or GaSb, there is also consideration of widening the band gap of CdTe through the incorporation of Mg or Zn[26]. While challenges remain for increasing the band gap of

CdTe-based materials, progress has been made with recent studies achieving 9.3% efficiency for a CdMgTe solar cell with 1.6 eV band gap[27]. Even with such developments, however, a significant challenge for developing a high efficiency CdTe tandem solar cell is finding an optically transparent tunnel junction that could enable monolithic integration[26]. Recent work with substrate configuration CdTe devices suggests that SWCNT films could act as a tunnel junction[28]. SWCNTs have also been used as a transparent layer for CdTe[29].

As a one dimensional material, SWCNTs possess sharp optical absorptions associated with the van Hove singularities in the density of states[30].The wavelengths at which these absorptions occur are largely determined by the diameter of the SWCNTs.

14 Larger diameter SWCNTs, such as those produced by laser vaporization or arc discharge, have so-called S11 and S22 absorptions occurring at roughly 1800 nm and 1100 nm, respectively, while smaller diameter SWCNTs, such as those produced by the CoMoCat or

HipCO processes, have S11 and S22 absorptions at roughly 1000 nm and 550 nm, respectively[31]. The absorption peaks in SWCNT films could be highly disadvantageous for tunnel junctions contacts since there will be a loss of photo-generated current in the bottom cell in the S11 and S22 wavelength regions.

It is possible to quench the SWCNTs absorption peaks through doping. For example, HNO3 is a well-known p-type dopant that will completely quench the first absorption peak and partially quench the second absorption peak of large diameter tubes[32]. However, HNO3 will also oxidize, etch, and, thus, destroy CdTe devices. In an effort to explore a potentially more gentle dopant, the present work explores the use of

+ - triethyloxonium hexachlorantimonate ((C2H5)3O SbCl6 ; OA), which has been used to strongly dope SWCNTs[33] and completely quench the absorption features at all wavelengths shorter than 900 nm[34].Such doping would effectively remove all absorption features that would appear below the CdTe band gap, and would result in a near IR transparent SWCNT film suitable for tandem devices. The quenching of the optical absorption features would also result in lowering the Fermi level of SWCNTs films. Based on our understanding of the role of BBL, reducing in the IFLO between the absorber layer

(CdTe) and BBL would improve the device performance due to minimizing the carrier recombination at the CdTe/BBL interface. However, when this project was done, we did not understand the band bending requirements.

15 2.2 Experimental details

CoMoCat SWCNT (SG65i, Southwest Nanotechnologies) films were prepared by spraying[35] sodium dodecylbenzenesulfonate (SDBS)-dispersed SWCNTs at 300 µl/min under nitrogen flowing at 7 std l min−1 with an ultrasonic spray head (06-5108, Sonotek,

3W) onto heated (140 °C) soda lime glass and CdTe substrates. Twenty-five nanometers thick SWCNT films were prepared[17]. The SDBS was removed by soaking the films in water for 1 hour.

Doping of SWCNT films followed the method reported by Chandra et al[33]. The prepared SWCNT films were soaked in an OA/dichloroethane (DCE) solution (10 mg-ml-

1) for 30 to 300 s at 70 °C. Residual OA on the surface was removed by rinsing with acetone.

CdS and CdTe films were deposited on commercially available TEC 15M

(Pilkington NA) substrates using a commercial vapor transport deposition process (Willard

& Kelsey Solar Group; WK). These samples were activated by applying a CdCl2/methanol solution and heating at 387 °C for 30 min in a dry air flow. Samples were subsequently rinsed with methanol to remove the excess CdCl2.

CdTe/SWCNT devices were completed with a 40 nm thick Au film, which was deposited by thermal evaporation. For comparison, standard CdTe devices were completed by evaporating 3 nm Cu/40 nm Au and annealing at 150 °C in air for 45 minutes. A solar cell area of 0.079 cm2 was defined by laser scribing[36]. These devices were measured under simulated AM1.5 illumination to obtain the current density-voltage (J-V) characteristics. The sheet resistance of SWCNTs films were measured with a standard four- point probe system (Signatone-2400), and the optical absorbance of the sprayed SWCNT

16 thin films were obtained using a Perkin-Elmer Lambda 1050 spectrophotometer. The CdTe films were characterized using scanning electron microscopy (SEM; Hitachi S-4800) and energy dispersive X-ray spectroscopy (EDS).

2.3 Results and Discussion

2.3.1 SWCNT on Glass Substrate

Significant changes in the optical properties were observed after the SWCNT films were doped with OA (figure 2-1a). There was a significant decrease in the intensities of the absorption peaks at ~550 nm and ~1000 nm, corresponding to S22 and S11 transition energies for the SWCNTs, respectively. The quenching of the S11 and S22 features indicates that hole doping has occurred and that the Fermi level in the SWCNTs has been pushed deep into the valence band[33, 34]. The complete quenching of the S22 transitions demonstrated here indicates a higher degree of SWCNT doping than has been reported to date[33, 34].

Figure 2-1. (a) The absorbance spectra of a 100 nm thick CoMoCat SWCNT film on

glass before and after doping with OA for 30 sec. (b) variation of sheet

resistance with doping time for SWCNTs films on glass.

The doping also reduces the electrical resistance of the SWCNT film. As figure 2-

1b shows, strong doping of the SWCNT film occurs in less than 30 s. The increase in the conductivity of the SWCNT film is consistent with an increase in the hole carrier density.

The strong enhancement in optical transmission and conductivity of SWCNT films by OA doping, coupled with previous results[17, 29], qualitatively suggests that SWCNTs films could act as a highly transparent back contact for the CdTe device.

2.3.2 SWCNT Film on CdTe

To investigate the possibility of use the SWCNT as a transparent back contact,

SWCNT films were deposited on CdTe device stacks by ultrasonic spraying (see Materials and Methods). Figure 2-2 shows the transmittance spectra of a bare CdTe device stack

(TEC15M/ CdS (~150 nm)/ CdTe (~4 휇m)) and the same device stack with undoped and

OA-doped SWCNT films on it. The average transmittance of the bare device stack is 55% for the wavelengths range that is most likely to be important for tandem solar cells (850 to

1100 nm). However, when the 25 nm thick SWCNT film was added, the transmittance decreased across the same wavelength range to 40% with a minimum value of 36% corresponding to the S11 absorption peak at ~1000 nm. As expected, the OA doping induced quenching of the S11 peak such that the average transmittance below the CdTe band gap

(850 to 1100 nm) increased to 53% .These results suggest that OA-doped SWCNT films could provide a transparent back contact with better performance than the current state-of- the-art transparent back contact[29].

Figure 2-2. The transmission spectra of CdTe, CdTe/SWCNT and CdTe/SWCNT/OA.

2.3.3 Device Performance

To determine the effect of OA doping on the device performance, CdTe devices with Cu/Au, SWCNT/Au, and OA-doped SWCNT/Au back contacts were fabricated.

Figure 2-3 shows the J-V characteristics for the best device, and table 2-1 shows the PV parameters for devices prepared in the three device architectures. Note that more than 20 devices where measured and included in data set reported in table 2-1. As previously reported[17], the average efficiency of SWCNT back contact device matched the performance obtained from the standard Cu/Au devices, with an improvement in VOC being offset by a decrease in FF. However, after exposing the SWCNT layer to the OA doping process, the efficiency was significantly reduced. Previous device modeling[17] suggested that reducing the Fermi level in the SWCNTs through doping

Figure 2-3. J-V characteristic of the CdTe devices with different back contacts.

would result in improved device performance by increasing the VOC, and this was indeed observed. However, the FF was significantly impacted and the J-V curves displayed indications of high series resistance and, in some cases (not shown) a kink in the characteristic at VOC, inducting a barrier at the back of the device. While the increase in the

VOC was expected based on our understanding of the behavior of the SWCNT back contact and doping, the increase in RS (and accompanying reduction in FF) was not.

Table 2.1: J-V characteristics of CdTe devices prepared with different back contacts.

Back contact Voc Jsc FF (%) Eff (%) 2 (mV) (mA/cm ) Cu/Au Avg. 781± 3 20.3±0.3 74.7±1 11.8±0.4

Best 804 20.4 76.5 12.5 SWCNT /Au Avg. 790± 2 20±0.31 73.8±0.7 11.6±0.2

Best 789 20.5 74.2 12.0 SWCNT with OA for 5min/Au Avg. 805± 2 19.4±0.31 67.7±0.9 10.6±0.2

Best 805 19.8 68.9 10.9

2.3.4 The Differences in SWCNT Doping on Glass and CdTe Substrates

The data in figure 2-3 and table 2.1 compares the performance of the devices with

SWCNT films that were optimally doped. It is interesting to note that the kinetics of the doping of SWCNTs was vary depending on whether the SWCNT film was on CdTe or on glass. Figure 2-4 compares the time required to quench the S11 absorption transitions in

SWCNT films prepared on glass (figure 2-4a) and CdTe (figure 2-4b). Complete doping of films on glass is achieved in a matter of seconds, while minutes are required for doping of films on CdTe.

2.3.5 The Activation Energy of OA Doping

To quantify the differences in SWCNT doping behavior on glass and CdTe substrates, we measured the activation energy for the process. Because the doping of

SWCNTs on glass is rapid, it seems clear that doping occurs when the OA is in contact with the SWCNT. In this case, the rate-limiting step would be the diffusion of OA to the surface of the individual SWCNTs. As a result, a diffusion-limited model in the form of equation 2.1 was used.

퐿푛 [퐴푆11]푡 = −퐷푡 + 퐿푛 [퐴푆11]0 (2.1)

-1 Where D is the diffusion time (s ), [AS11]t is the area under S11 curve at time t, [AS11]0 is the area under S11 curve at time 0, and t is time (s). To determine the diffusion time (D) for each temperature (27, 50, and 70 ℃) the area under the S11 curve was calculated for several doping times, and the results were fit to equation (2.1). Figure 2-4 shows an example of the measured transmittance for doping completed at 70 °C.

Figure 2-4. Transmission spectra of OA doped SWCNT films at 70 °C as a function of

time on (a) glass and (b) on CdTe.

The thermally activated diffusion process is expected to follow an Arrhenius behavior, following equation 2.2.

퐸푎 − 퐷 = 퐴푒 퐾푇 (2.2) where Ea is the activation energy, T is the temperature, K is the Boltzmann constant, and

A is a constant. Fitting the diffusion data to equation 2.2 (figure 2-5) shows that the activation energy of OA doping of SWCNTs on CdTe is 2.45 times higher than that of doping SWCNT on glass. The significant difference in the activation energy indicates that the presence of CdTe is severely inhibiting SWCNT doping. This could be due to an interaction between the OA and CdTe, either repulsive or attractive, or due to an interaction between the SWCNTs and CdTe that prevents the OA from interacting with the SWCNTs.

Figure 2-5. Plot of ln (D) vs. 1/KT to evaluate Ea of OA doping for SWCNT on glass and

CdTe.

2.3.6 OA Interacting with CdTe

To determine if the CdTe is interacting with the OA, bare CdTe device stacks underwent the OA doping process prior to device completion with Cu/Au and

SWCNTs/Au. As shown in figure 2-6, the device performance severely degraded by the

OA processing. In both cases, the JSC and VOC decreased relative to the untreated sample.

In the case of the Cu/Au back contact devices, a kink in the J-V curve also appeared, indicating an increase in the barrier at the back contact[37]. These results indicate that there is an interaction between CdTe and OA, but do not suggest what type of interaction occurs.

To determine the type of interaction between CdTe and OA, the CdTe films were characterized using SEM and EDS. Figure 2-7a shows the SEM image of the pristine CdTe containing tightly packed grains with a size of ~1 μm. This morphology is commonly observed in the CdCl2-treated CdTe films. After 10 minutes of OA treatment, though, the

SEM image (figure 2-7b) looked very different. New features appear on the surface of the

23 CdTe. EDS of the sample shows the presence of ~1.5 wt.% of Cl that is not present in the untreated sample.

Figure 2-6. J-V characteristic of the CdTe devices with different back contacts

To determine if the CdTe is interacting with the OA, bare CdTe device stacks underwent the OA doping process prior to device completion with Cu/Au and OA doping

- of the SWCNTs is thought to be through the interaction of the SbCl6 ion[33]. Assuming the same mechanism for CdTe, the Cl detected in EDS is likely from this ion. The Sb was probably not detected because the sensitivity of EDS is not sufficiently high for such low loading levels. These results show that

Figure 2-7. SEM images of CdTe (a) before OA treated (b) after OA treated for 10 min.

a Cl-containing chemical residue forms on the CdTe surface after exposure to the OA doping process. The high activation energy for OA doping of SWCNTs on CdTe substrates, relative to on glass, is due to a preferential interaction of OA towards CdTe. This preferential interaction results in an OA film on the CdTe surface, which, in turn, reduces the device performance of SWCNT back contact devices exposed to an OA doping process.

Evidently, the OA interaction with CdTe is stronger than the interaction with SWCNTs, suggesting that the CdTe surface getters OA species which, in turn, impacts the rate at which SWCNT doping is achieved. Consequently, CdTe devices where the SWCNT back contact is OA doped after deposition onto the CdTe will not result in high performance transparent back contacts. Current data indicates that a partially insulating film may be formed at the CdTe/SWCNT interface. Such a layer might increase the series resistance and impact the FF, as observed in figure 2-3.

25 2.4 Conclusion

OA doped SWCNT films were investigated as a transparent back contact for CdTe solar cells. OA doping induces a complete quenching of S11 and S22 optical transitions resulting in increasing the transmittance of the SWCNT films and a decrease in the electrical resistance due to doping. The transmittance of the doped SWCNTs on CdTe was nearly identical to that of the bare CdTe sample after the SWCNTs were thoroughly doped.

The OA doping process, however, adversely affected the device performance. It was found that the OA preferentially interacted with CdTe during the doping process and formed a reaction film on the CdTe surface. This film is believed to be the cause of the reduced device performance. Consequently, OA doping of the SWCNTs after application to the

CdTe may not be a viable pathway towards preparing a transparent back contact or tunnel junction for CdTe solar cells.

26 Chapter 3

Controlling Band Alignment at the Back Interface of Cadmium Telluride Solar Cells Using ZnTe and Te Buffer Layers

This chapter focuses on the incorporation of ZnTe and Te as buffer layers to improve the device performance. In this study, we compare device performance using these two materials as buffer layers at the back of CdTe devices. We show that using Te in contact to CdTe results in higher performance than using ZnTe in contact to the CdTe. Low temperature current density-voltage measurements show that Te results is a lower barrier with CdTe than ZnTe, indicating that Te has better band alignment, resulting in less downward bending in the CdTe at the back interface than when ZnTe is used. The results in this chapter have been published in Alfadhili et al., 2019[19] and printed from MRS

Advances, Vol 4, Fadhil K. Alfadhili, Adam B. Phillips, Geethika K. Liyanage, Jacob M.

Gibbs, Manoj K. Jamarkattel, and Michael J. Heben, Controlling Band Alignment at the

Back Interface of Cadmium Telluride Solar Cells using ZnTe and Te Buffer Layers, Pages

913-919, Copyright (2019), with permission from Cambridge University Press with

License Number 4823660553336, Jan. 2019.

27 3.1 Introduction

Over the past few years, significant improvement in CdTe device efficiency has occurred with record values currently over 20%[25, 38]. Most of this performance improvement is due to increased current collection by changing the traditional window layer (CdS) of CdTe to a wider bandgap material with better band alignment [39].

However, the open circuit voltage (VOC) remains well below the theoretical limit due to the large barrier formed at the back of the device caused by the deep valence band edge of

CdTe (~-5.8 eV) [9].

Researchers have implemented different types of back-buffer layers (BBLs) to reduce the barrier, band bending, and carrier recombination at the back surface[16]. Among the candidates for buffer layers, zinc telluride (ZnTe) has been used due to the expected low valence band offset (VBO) between CdTe and ZnTe (~-0.14 eV)[40] which should result in a low barrier to holes at the back of the device[40, 41]. However, previous experimental work shows that the barrier height between the CdTe and ZnTe:Cu is ~-0.4 eV,[42] far higher than the expected value. Tellurium (Te) has also been used as a buffer layer at the back of CdTe, and been shown to improved device performance through, presumably, reduced band bending.[16] Te can be formed either by wet chemical etching- techniques [43], [44], [45], chemical bath deposition (CBD)[46], or physical vapor deposition- methods such as sublimation [47] or evaporation[48]. Recently, we developed a facile, self-limiting wet chemical method to form a Te layer at the back surface of CdTe by reacting the CdTe with methylammonium iodide (CH3NH3I, MAI).[49] The MAI treatment effectively reduce the interface defects by selectively removing Cd from the

CdTe surface. This leads to an increase in device efficiency up to 10% compared to devices

28 without the MAI treatment and a reduction of the barrier height at the back of the device up to 0.140 eV [49]. Furthermore, we found that the interaction between MAI and Zn is similar to that of MAI and Cd – the Zn2+ or Cd2+ ions react with MA+ and I- ions, resulting in Zn- or Cd-based perovskites.[50] Since the Zn2+ reacts with MAI similar to the Cd2+, selective removal of Zn from ZnTe films is expected. In this study, we demonstrated the use of MAI treatment to form a Te layer on top of CdTe and ZnTe films to fabricate back contacts for CdTe devices. Eight different back contact configurations for CdTe device were implemented to determine whether Te or ZnTe can result in the highest device performance. In addition to device performance, the barrier heights were measured to provide a better understanding of the energetics of the back contact of the devices. The results indicated that the back-contact barrier is reduced with the use of Te or ZnTe at the back of CdTe device which increases VOC by decreasing the band bending.

3.2 Experimental Details

WK samples were used and activated as described in chapter 2. For these experiments eight different back contact structures (BCs) were deposited to study how the band alignment changed with the buffer layer and how this affects the device performance.

Figure 3-1 shows the details of the eight different back contacts for CdTe solar cells. The

CdTe devices with only Au (BC1) and Cu/Au (BC2) were used as standard devices for comparison. Te layers were formed on top of the CdTe and ZnTe films using MAI treatment. Details of the interaction between CdTe and MAI were reported in our previous work [49]. 300 nm of ZnTe was deposited by RF-sputtering at room temperature onto

CdCl2 treated CdTe films (0.55 w/cm2, 10 mTorr) followed by post-deposition annealing at 400 °C for 2 minutes in N2 ambient. Finally, 5 nm Cu and 40 nm Au back contact were

29 evaporated for devices with Cu (BC4, BC6, and BC8) and annealed in N2 ambient at 400

°C for 1 minute to dope the CdTe and ZnTe. Devices without Cu (BC3, BC5, and BC7) did not undergo a second annealing.

Film surfaces were characterized using Raman spectroscopy obtained at room temperature from 50 to 600 cm-1 using a LabRam confocal scanning microspectrometer

(Horiba-Jobin Yvon) equipped with a He-Ne laser with line excitation at 632 nm. Auger electron spectroscopy (Perkin-Elmer PHI600) was used to characterize the composition of the CdTe/Te/ZnTe/Te/Cu/Au device. For low temperature measurements, the devices were placed in a helium cryostat, and the temperature was varied from 190 to 310 K. A

LabVIEW program controls the temperature controller, a Keithley 2400 source-meter, and a Keithley 7001 switch.

Figure 3-1. Eight different back contact structures for CdTe thin film solar cells.

30 3.3 Results and Discussion

3.3.1 Raman Spectroscopy Analysis

To verify that the MAI treatment is effective at selectively reacting with Zn in the

ZnTe, MAI with varying concentrations (125, 250, 357, and 500 mM) were applied to

ZnTe films and processed. Figure 3-2 shows Raman spectra of as-deposited ZnTe thin films on TECTM-15 glass substrates after these MAI treatments. The as-deposited ZnTe thin film exhibits the first order longitudinal optic (LO) mode at 204 cm-1 [51]. After MAI treatment, all spectra exhibit additional peaks at 122 cm-1 and 142 cm-1, likely due to A(Te) and E(Te) modes, respectively [52, 53]. Additionally, the intensity of the LO(ZnTe) mode at 204 cm-1 decreases with increasing MAI concentration while the intensity of A(Te) and

E(Te) modes increase with increasing MAI concentration. These results indicate that a Te layer forms on the surface of the ZnTe films and the volume of Te layer increases with increasing of MAI concentration.

Figure 3-2. Raman spectra of ZnTe thin films as-deposited and MAI treated with different

concentrations.

31 3.3.2 Device Performances

To understand how the buffer layer at the back of CdTe devices affect the device performance, CdTe devices with the eight back contact configurations shown in Figure 3-

1 were fabricated. Figure 3-3 shows the performance parameters for the best devices and the average for more than 25 devices. The CdTe device with ZnTe/Au (BC3) contact shows higher VOC than the device with a Au (BC1) contact. This result indicates that the Fermi level of ZnTe was lower than the Fermi level of Au which reduce the band banding and carrier recombination at the back surface. However, FF for CdTe device with ZnTe/Au was lower than CdTe device with Au due to the high electrical resistivity through the ZnTe film. Doping the ZnTe with Cu improved the efficiency due to increase FF for these devices, as seen clearly in the devices (BC4). Interestingly, for the CdTe devices with a Te layer between CdTe and ZnTe (BC5, BC6, BC7, and BC8), the Voc, FF, and, thus, efficiency improve. This may suggest that the valance band offset (VBO) between the

CdTe and Te was lower than the valance band offset between the CdTe and ZnTe, resulting in better band alignment and lower recombination at the interface. More importantly, the

VOC of Cu-free devices with Te/ZnTe/Te/Au contact (BC7) is notably higher than the CdTe device fabricated with standard Cu/Au and Te/ZnTe/Au contacts (BC2 and BC5, respectively). These results indicate that Te improves the band alignment between ZnTe and Au as well as between CdTe and ZnTe and results in reduced interface recombination despite not using Cu.

Figure 3-3. PV performance parameters of CdTe devices with different back contacts

including (a) efficiency, (b) Voc, (c) FF, and (d) Jsc.

3.3.3 Auger Depth Profile

To understand how the Cu distributes into the CdTe/Te/ZnTe/Te/Cu/Au device, an

Auger depth profile was carried out. Figure 3-4 shows that the Cu was uniformly distributed throughout the ZnTe which confirms the expectation that Cu doped ZnTe [40].

Additionally, the incorporation of Cu into the ZnTe film resulted in a decrease in resistivity from 108 Ω/sq to 103 Ω/sq, consistent with increased doping and fill factor for devices.

Figure 3-4. Auger depth profile spectra of the CdTe/Te/ZnTe/Te/Cu/Au device.

3.3.4 Back Barrier Height Analysis

To further understand the difference between the eight back contact configurations, the barrier height was measured by assuming that the front diode (CdS/CdTe) and the back diode (CdTe/BBL/Metal) can be treated as independent circuit elements with no interaction between them[54]. In this case, a forward bias voltage is divided between the two diodes.

The voltage across the main diode saturates at Voc, hence, any applied voltage above the

VOC will be dissipated across the back diode. As a result, barrier to hole flow at the front junction reduces with the forward bias while the back-contact barrier remains. Figure 3-5a shows the current density voltage as a function of temperature for the CdTe/Cu/Au device.

The thermionic emission model (TE) was used to extract the back barrier height as described by equation 3.1.

2 푞∅푏 퐽0 ∝ 푇 [exp ( )] (3.1) 푘퐵푇

34 where 퐽0 is the saturation current density, T temperature, q the electron charge, ∅푏 the back- barrier height, and kB the Boltzmann constant. The back-barrier height for the eight back

2 contact configurations was extracted using an Arrhenius plot of ln(J0/T ) versus 1/kBT, as shown Figure 3-5b. The CdTe device with only Au at the back (BC1) has barrier height

0.395 ± 0.011 eV, which is higher than the device with Cu/Au back contact (0.331 ± 0.004 eV) indicating that doping the back of CdTe with Cu reduces the back-barrier height. These results are consistent with reported values of the barrier height for Au and Cu/Au back contacts[54, 55]. When a ZnTe layer is added between the CdTe and Au (BC3), the barrier height was further reduced 0.293 ± 0.009 eV, suggesting that the the band alignment between CdTe and ZnTe is better than that of CdTe and Au. Interestingly, adding Cu between the ZnTe and Au for the CdTe/ZnTe/Cu/Au device only slightly changes the value of the back-barrier height (0.277 ± 0.010 eV). This indicates that the barrier at the

CdTe/ZnTe interface did not change even with doping of the ZnTe. These results show that the VBO at the CdTe/ZnTe interface was not as low as expected[40] and is close to what others have reported [42]. The back barrier was further reduced when a Te layer added between the CdTe and ZnTe. For all devices with CdTe/Te (BC5, BC6, BC7, and BC8), the barrier height values are the same within error at ~0.2 eV. These results show that Te has better band alignment with CdTe than ZnTe does. These results help support the idea that band alignment between the CdTe absorber layer and a BBL is necessary for high performance devices. Furthermore, it indicates that due to the relatively poor alignment between CdTe and ZnTe, ZnTe back contacts are not an optimal buffer material for high efficiency CdTe devices.

35 (a) -2 (b) 10 310 K CdTe:Cu/Au -16 -3 Temperature (K)

10 )

190 2 -

k -18

) 200

2 -4 2

10 210 -

m m

c 220

/ c

-5 -20

A 230 A

( 10 Back Contact

(

) 240

] Au

J 2

( -6 250

T -22 Cu/Au /

g 10

260 0 ZnTe/Au

J [

L 270 -7 190 K ZnTe/Cu/Au

n -24

10 280 Te/ZnTe/Au L 290 Te/ZnTe/Cu/Au -8 10 300 -26 Te/ZnTe/Te/Au 310 Te/ZnTe/Te/Cu/Au

0.2 0.4 0.6 0.8 1.0 1.2 1.4 40 45 50 55 60 -1 Bias voltage (V) 1/KBT (eV)

0.40

)

(

t 0.35

h g

i Back contact

e h

0.30 1- Au r

e 2- Cu/Au i

r 3- ZnTe/Au r

a 0.25 4- ZnTe/Cu/Au B 5- Te/ZnTe/Au 6- Te/ZnTe/Cu/Au 0.20 7- Te/ZnTe/Te/Au 8- Te/ZnTe/Te/Cu/Au

0 1 2 3 4 5 6 7 8 Back contact

Figure 3-5 (a) Temperature-dependent dark J-V characteristics for CdTe/Cu/Au device. (b)

2 plots of ln (J˳/T ) versus 1/KBT for the eight different back contacts. (c) the

values of the back-barrier height for each back contact configuration.

3.4 Conclusion

To summarize, we have shown that MAI treatment selectively removes Zn from

ZnTe films and can be used to form a Te layer at the back interface. The use of Te and

ZnTe as buffer layer at the back of CdTe devices improve device performance by increasing VOC due to reduce carrier recombination at the back of the devices. Back barrier height analysis shows that the barrier at the CdTe/Te interface (~ 200 meV) is lower than the barrier at the CdTe/ZnTe interface (~ 300 meV), indicating that Te reduces the downward band bending. This suggests that the Te layer can act as a low barrier buffer layer for the flow of holes. Additionally, using layers of Te/ZnTe/Te/Au at the back of

36 CdTe device, a Cu free back contact for CdTe device has been designed which has the same CdTe device efficiency as the standard device using Cu/Au back contact.

37 Chapter 4

Potential of Cdmium Zinc Telluride Thin Film Back Buffer Layer for CdTe Solar Cells

Based on the work of chapter 3, ZnTe and Te are not the best option to use as BBL for CdTe devices. CdTe/ZnTe and CdTe/Te interfaces show a VBO of 300 and 200 meV, respectively. These values suggest that the bend bending still exists at these interfaces, resulting in high recombination which would limit the device performance. However, using

BBL with IFL of less than that of CdTe would create band banding up, resulting in low surface recombination. In this chapter, I will discuss the potential of using CZT as BBL for

CdTe devices. Besides, controlling the p-type doping density of CZT films using Cu as a dopant to push the Fermi level down will be described in this study. The results in this chapter have been published in Alfadhili et al., 2019[56]. © [2019] IEEE. Printed with permission, from Fadhil K. Alfadhili, Adam B. Phillips, Manonoj K. Jamarkattel, Astin J.

Snder, Jacob M. Gibbs, Geethika K. Liyanage, and Michael J. Heben, Potential of CdZnTe

Thin Film Back Buffer Layer for CdTe Solar Cells, 2019 IEEE 46th Photovoltaic Specialist

Conference (PVSC), and Feb. 2020.

38 4.1 Introduction

Over the past few years, the power conversion efficiency (PCE) of CdTe thin film solar cells has been pushed to 22.1%[25]. Given the recently reported time-resolved photoluminescence (TRPL) results that indicate high carrier lifetime[57],open circuit voltage (VOC) is likely limited by recombination at the interfaces[58]. Recent published experiments have demonstrated reduced recombination at the front interface by replacing the traditional cadmium sulfide (CdS) emitter layer of CdTe devices with magnesium zinc oxide (MZO)[10]. This change improved the band alignment between the emitter and absorber, which resulted in a significant increase in the VOC and fill factor (FF). Analogous improvements to the back of the devices have not been conclusively demonstrated

Different types of back buffer layers (BBLs) have been investigated[18, 59, 60].

Zinc telluride (ZnTe) widely used[61] as a BBL due to low electron affinity, the capability of high hole density with Cu, and expected low valence band near that of CdTe [40].

However, barrier height measurements indicate that the valence band offset between CdTe and ZnTe is much higher than expected at ~400 meV[42, 62]. ZnTe and CdTe, though, form an alloy (Cd1-XZnXTe; CZT), which could produce a composition with a lower barrier at the back interface. Additionally, the p-type doping of ZnTe film by substitutional Cu is expected to be maintained for CZT film, which could lead to upward band bending at the interface [63].

Here, we explore using CZT as a BBL for CdTe solar cells. We show that high Cu- doping of a CZT film with band gap of 1.85 eV. We also apply this CZT film to CdS/CdTe device stack and show that the efficiency increases relative to Cu/Au back contact. The

39 limiting barrier height of the device is also measured and indicates that the valence band edge (VBE) of the CZT is close to that of CdTe.

4.2 Experimental Details

WK partial device stacks were performed in this study and activated as described in chapter 2. These samples were completed with a 250 nm thick Cd1-XZnXTe film deposited by RF sputtering at room temperature (1.99 W/cm2, 10 mtorr) from 2” diameter homemade target with 30 atomic % CdTe (99.99%, Beantown Chem.) and 70 atomic%

ZnTe (99.99%, Beantown Chem.)[64]. The stacks were than annealed at 400 °C for 2 minutes in N2 ambient. Subsequently, 5 nm Cu was thermally evaporated followed by an additional annealing in N2 at 400 °C for 1 minute to Cu dope the CdTe and CZT. For the device with Te layer, the Te layer was formed using methylammonium iodide (CH3NH3I,

MAI) treatment [49, 62]. Finally, the device was completed with 40 nm of Au. Electrical characterization of the CZT films deposited onto 3 mm soda-lime glass was performed using Hall measurements (MMR Technologies, Inc.) while the sheet resistance of as- deposited CZT film was measured with a standard four-point probe system (Signatone-

2400).

4.3 Results and Discussion

4.3.1 Optical Properties of CZT Film

To determine the composition X in the Cd1-XZnXTe film, the optical band gap of as deposited CZT film was calculated by fitting a Tauc plot (figure 4-1), which indicates a bandgap of 1.85 eV. The X value was determined from the optical band gap using the reported equation[65] for the CZT alloy. The X value of the as deposited film was 0.6

40 which is 9 % blow the value of the prepared sputtering target which is likely due to the different sticking coefficients of CdTe and ZnTe and sputtering rate[66].

Figure 4-1. The Tauc plot ((α hυ)2 vs hυ) of as deposited CZT film.

4.3.2 Electric Properties of CZT Film doped with Cu

To explore the potential doping properties, CZT films were deposited on soda-lime glass, coated with different Cu thickness (1,3,5 nm), then annealed at 400 °C for 1 minute in N2 ambient to drive the Cu into the CZT film. The sheet resistance measurement of as- deposited CZT film could not be measured, indicating a value on the order of 109 Ω/square.

After deposition varying thicknesses of Cu and the subsequent heat treatment, significant changes were observed in the electrical properties as shown in Table 4.1. The resistivity and sheet resistance of the films were reduced dramatically with increasing Cu thickness due to the increase in the carrier density. These results indicate that the p-type Cu doping has occurred, moving the Fermi level toward the valence band[38]. The resistivity of the

CZT film was as low as 20.6 Ω-cm and highest carrier concentration of 2.2 x 10^17 cm-3

41 occurred with 5 nm of Cu, indicating the high level of Cu doping was obtained. 5 nm Cu was used for further CdTe device fabrication.

Table 4.1: Effect of Cu doping on the electrical properties of CZT thin film.

Cu Resistivity Mobility Density Sheet Type of Thickness (Ω*cm) (cm2/Vs) (cm-3) Resistance carrier (nm) (Ω/cm2) 0 ND ND ND 5.3E+09 ND 1 152 2.2 1.90E+16 6.0E+06 h 3 53.7 3.1 3.70E+16 2.1E+06 h 5 20.6 1.4 2.20E+17 8.2E+05 h 7 23 1.2 2.20E+17 9.2E+05 h/e

4.3.3 Device Performance.

To examine the impact of CZT as a BBL on device performance, CZT films were sputtered on activated CdS/CdTe device stacks. Figure 4-2 shows the J-V curves for the highest efficiency CdTe devices with CZT/Cu/Au, CZT/Cu/Te/Au, and standard Cu/Au back contacts. The associated performance parameters shown in table (4.2) represent average values for more than 20 devices. As shown in Figure 4-2, the device performance increased when the CZT was incorporated between the CdTe and Au due to improvement in VOC. The FF, though, was slightly lower than the FF in the standard device. For the device with a Te layer between the CZT and Au, the VOC was further increased and the FF matched the FF of Cu/Au back contact, leading to 13.2% efficiency.

Figure 4-2. J-V curves for CdTe devices with CZT/Cu/Au, CZT/Cu/Te/Au, and

standard Cu/Au back contacts.

Table 4.2: J-V characteristics of CdTe Devices Prepared with Different Back Contacts.

Back contact V (mV) 2 FF PCE (%) OC J (mA/cm ) SC (%)

Cu/Au Avg. 759±10 21±0.2 73.9±1.7 11.8±0.2

Best 778 21.2 74.0 12.2 CZT/Cu/Au Avg. 814±7 21.2±0.5 70.5±0.8 12.2±0.3

Best 809 21.8 72.2 12.8 CZT/Cu/Te/Au Avg. 834±2 20.4±0.2 73.8±1.5 12.6±0.3

Best 830 21.0 75.6 13.2

43 4.3.4 Barrier Height Measurements

To help understand the J-V results with the addition of CZT and CZT/Te BBL, the limiting barrier heights of the two devices were measured using the method described by

Niemegeers and Burgelman[54]. Temperature dependent (J-V) was performed in dark with voltage sweeping from -0.5 to 1.5 V. The thermionic emission model was used to determine

2 the barrier height by fitting a line to the Arrhenius plot of ln(J0/T ) versus 1/kBT. The slope of this line is the limiting barrier height (Фb). Figure 4-3 (a) and (b) show the measured

JVT of CdTe devices with CZT/Cu/Au and CZT/Cu/Te/Au back contacts, respectively, for a temperature ranging from 180 to 310 K with step size of 10 K. Figure 4-3 (c) shows the

Arrhenius plot and Фb. The barrier height measurements show that there is a barrier of

0.392 ± 0.015 eV for the CdTe/CZT/Cu/Au device. This value is the largest barrier to carrier flow in forward bias and could be located at any interface. For this device, the barrier is likely at the CdTe/CZT interface or the CZT/Au interface. However, after applying MAI treatment to form a Te layer on top of CZT film, the barrier height was reduced to 0.234 ±

0.004 eV. Since only the CZT/Au interface was altered, the initial 0.392 eV barrier must be located at this interface. Interestingly, the barrier height value at the CZT/Au interface is consistent with reported value of barrier height at CdTe/Au[62] and the barrier height value at CZT:Cu/Te interface is in good agreement with the reported value of barrier height at the CdTe:Cu/Te[49] interface. This suggests that the valence band of CZT is close to that of CdTe. The barrier height measurements help explain the J-V results. Clearly the increase in VOC compared to the Cu/Au standard is due to reduced recombination at the back interface. The measured barrier heights indicate that the CZT valence band is close to that of CdTe, and the doping measurements indicate CZT is more highly doped than the

44 CdTe. This interface should, therefore, have upward band bending in CdTe[38]. However, because the film is thin, the large barrier between the CZT and Au may cause downward band bending the CZT/Au interface that extends into the CdTe. This would cause more recombination and a lower VOC and FF than expected for proper band alignment. The addition of Te splits the large barrier at the back of the device over 2 interfaces, improving the alignment at the back of the device and reducing the downward band bending in the

CZT, leading to improved VOC and FF.

Figure 4-3. Temperature-dependent dark J-V curves for CdTe devices with (a) CZT/Cu/Au

and (b) CZT/Cu/Te/Au back contacts. (c) Arrhenius plots of ln (J/T2) vs 1/KT

for CdTe/CZT/Cu/Au and CdTe/CZT/Cu/Te/Au devices.

45 4.4 Conclusion

This work represents the first reported attempt of using CZT film to make a back- buffer layer for CdTe devices. Electrical characterization shows that CZT film can be more highly doped with Cu than CdTe. This indicate that the Fermi level is closer to the valence band edge in CZT than CdTe. Employing CZT:Cu/Te as the BBL between CdTe and back electrode, lead to improve device performance through reduction of recombination at the back of the device. Temperature dependent current-voltage analysis suggests that the valence band edge of Cd0.4Zn0.6Te is close to that of CdTe. When taken together, these results indicate that it should be possible to develop a high performance back contact with the desired Fermi level alignment and upward band bending at the back of the device[12].

46 Chapter 5

Back Surface Passivation of CdTe Solar Cells by Solution‐Processed Oxidized Aluminum

This chapter presents an alternative strategy to develop the back contact for CdTe device by passivating the rear surface. This method can reduce minority recombination while still providing for efficient majority carrier extraction. Here, we present a solution- based process that reduces minority carrier recombination at the back surface of the device and increases the open circuit voltage (VOC). The process deposits very small amounts of oxidized aluminum in a nonconformal manner, and the Fill Factor (FF) and photoconversion efficiency (PCE) are improved when the total amount added corresponds to ~1 monolayer of alumina. The chemical bonding is presently unclear due to the very small quantities of Al3+ and the topography of the polycrystalline CdTe surface. Addition of further aluminum causes the FF and efficiency to drop as the interface becomes blocking to current flow. The optimized layer increases the average baseline PCE for Cu-free device stacks made with a commercial process from 10.4% to 11.7%, while the efficiency with

Cu doping was improved from 12.2% to 13.6%. The conclusion that interface recombination is reduced at the back surface is supported by time-resolved photoluminescence spectroscopy and quantum efficiency measurements performed at the maximum power point.

47 5.1 Introduction

The record photoconverson efficiency (PCE) of CdTe-based thin-film solar cells has reached 22.2 %[8, 25]. While the short circuit current density (Jsc) has approached bandgap limited values,[39] and fill factors above 79% have been achieved,[10] there is still substantial opportunity to increase the open circuit voltage (Voc).[67] As further improvements in the bulk lifetime are obtained, through approaches such as Cl passivation and Se incorporation,[39] and the interface recombination at the front surface is curtailed through emitter engineering,[10] the back contact will limit the device efficiency.[20]

Consequently, research is being focused on the development of back contact strategies that can reduce minority recombination while still providing for efficient majority carrier extraction. Back surface passivation can be accomplished either by reducing the concentration of electrically active defects at the interface, or by creating a back surface field that repels minority carriers through electrostatics or doping profiles.[58] Such strategies have been important for advancing Si and GaAs PV,[6, 68] but have not yet been successfully applied to thin-film polycrystalline materials such as CdTe.

Alumina is considered to be a promising material for back surface passivation due to a high density of negative charge [69] and a low degree of lattice mismatch (3.7%) between the unit cell of the (0001) surface of Al2O3 and the (111) surface of CdTe. These characteristics offer the potential for repelling minority carriers and creating a low defect density interface, respectively.[69, 70] Indeed, minority carrier lifetimes 27 and 430 ns have been demonstrated for CdTe and CdxSe1-xTe layers in alumina double heterostructures[71].

48 Recent efforts to use alumina to passivate the rear surface of CdTe solar cells have employed atomic layer deposition (ALD) [3] and sputtering [71, 72]. In their pioneering work, Liang et al. deposited alumina layers in thicknesses up to 5 nm by ALD and tracked the device performance characteristics. They stated that the baseline PCE of 10.7% was improved to 12.1% when the Al2O3 thickness was 1 nm, and further increases in Al2O3 thickness led to poorer device performance. The authors suggested that a 1 nm layer was sufficiently thin to allow holes to tunnel, but thick enough to present fixed charge that repelled minority carrier electrons. An improvement in the long wavelength response in the short-circuit external quantum efficiency (EQE) was presented as evidence for reduced back surface recombination, but no improvement in JSC was observed as would be expected for this proposed mechanism. Also inconsistent was the lack of evidence for current- blocking for the thicker Al2O3 layers. In contrast, Kephart et al. applied Al2O3 to CdTe back surfaces by sputtering and saw pronounced kinks in the J-V characteristic at layer thicknesses of 3 and 5 nm.[38] In this case, however, the device efficiency was not improved with the addition of Al2O3. Thus, while Al2O3 has indeed been shown to provide passivation for photoluminescence lifetime measurements[71], there has been no conclusive evidence to date for alumina providing an efficiency enhancement via back surface passivation in a CdTe PV device.

In this study we present a solution-based process for passivating the back surface of CdTe solar cells. Following a procedure developed for passivating silicon solar cells,[73] we employed aluminum acetylacetonate (Al(acac)3) dissolved in methoxyethanol to deposit alumina on the back surface of CdTe solar cells by spin-coating and heating. The amount of material deposited was increased by repeating the spin-

49 coating/heating cycle. Photoluminescence measurements showed that the minority carrier lifetime increased with the number of cycles, as did the Voc of the finished devices.

Scanning electron microscopy and atomic force microscopy revealed that the deposited aluminum was not formed in a uniform layer and was present in very small amounts, suggesting that the passivation effects maybe be site- and/or facet-specific. X-ray photoelectron and Auger electron spectroscopy studies (XPS and AES, respectively) revealed that the CdTe surfaces that produced the best performing devices had less than a monolayer of Al in a chemical state similar to that found for alumina. External quantum efficiency measurements made at the maximum power point of the J-V curves clearly supported the conclusion that the device improvement was due to back surface passivation.

The optimized solution process increased the average baseline efficiency for Cu-free devices made with a commercial process from 10.4% to 11.7%, while the efficiency for devices made with Cu doping improved from 12.2% to 13.6%.

5.2 Experimental details

CdTe device stacks composed of ~120 nm of CdS and ~3 μm of CdTe were deposited onto TECTM-15M coated soda-lime glass substrates in an unoptimized commercial vapor transport deposition process by Willard and Kelsey Solar Group. The

CdTe material was activated by applying a saturated solution of CdCl2 in methanol to the sample and heating to 390 °C in dry air for 30 minutes. Excess CdCl2 was removed by rinsing with methanol. The aluminum acetylacetonate Al(acac)3 (Sigma Aldrich Co. LLC,

99.999%) precursor solution was prepared by dissolving 400 mg of as-received powder in

20 mL of 2-methoxyethanol. 250 μl of the solution was pipetted onto a stationary sample which was then spun at 2000 rpm for 25 s[73]. Samples were then heated in laboratory air

50 to 300 °C for 10 minutes. The spinning/heating cycles were performed 1, 3, 5, 7, and 9 times to produce increasingly thick passivation layers. Devices were formed by depositing

40 nm of Au by thermal evaporation to form a back-metal electrode. Some samples also had a thin layer (3 nm) of Cu deposited prior to Au deposition to enhance doping and lower the back surface barrier. In these cases, a subsequent heating step at 150 °C was performed in air to promote Cu diffusion for times ranging between 40 and 80 minutes. Individual solar cells where precisely defined by laser scribing (0.06 cm2). Performance statistics were evaluated for relatively large data sets (n > 20). Additional experimental details regarding Auger electron and X-ray photoelectron spectroscopies, J-V characterization, and photoluminescence and other measurements can be found in the Supporting

Information.

Al2O3 was deposited onto CdCl2 activated CdTe samples (WK samples) using solution processing. The aluminum acetylacetonate Al(acac)3 (Sigma Aldrich Co. LLC,

99.999%) precursor solution was prepared by dissolving 400 mg of as-received powder in

20 mL of 2-methoxyethanol. 250 μl of the solution was pipetted onto a stationary sample which was then spun at 2000 rpm for 25 s[73]. Samples were then heated in laboratory air to 300 °C for 10 minutes. The spinning/heating cycles were performed 1, 3, 5, 7, and 9 times to produce increasingly thick passivation layers. Devices were formed by depositing

51 The surface morphology was examined by AFM (Veeco metrology group) and scanning electron microscopies (SEM). XPS and AES examinations were made with a

Perkin-Elmer PHI600. Both steady-state and time-resolved PL measurements were performed with 532 nm excitation of the film side of the samples at room temperature.

Current density voltage (J-V) curves were measured under simulated AM1.5G solar irradiation (Newport model 91195A-1000), and external quantum efficiency (EQE) measurements were acquired from wavelength range of 300−900 nm using a PV

Measurements Inc., model IVQE8-CQE system.

Samples were excited through the film side and PL signals were detected by a

Horiba Symphony-II CCD detector.

Time-resolved photoluminescence (TRPL) measurements were also performed at

532 nm, with a ~150 µm spot at an intensity of ~132 mW/cm2 with the repetition rate of 20

MHz when the samples are excited through the film side at the peak emission wavelength determined from the PL measurement. The TRPL measurements of CdTe samples were performed with time correlated single photon counting (TCSPC) module with integration time 300s bi-exponential PL decays were observed. Current density voltage (J-V) curves were measured under simulated AM1.5G solar irradiation (Newport model 91195A-1000) using a Keithley 2400 source meter. The external quantum efficiency (EQE) measurements were acquired from wavelength range of 300−900 nm using a PV Measurements Inc., model IVQE8-CQE system. with a cw laser with beam diameter ~100 µm at 3.1 W/cm2.

52 5.3 Results and dissection

5.3.1 The Surface Characterization

Figures 5-1a-c show scanning electron microscopy (SEM) images of the CdTe surfaces directly after CdCl2 processing and after the addition of 1 and 5 spin- casting/heating cycles, respectively. Prior to deposition (figure 5-1a), the image is characteristic of the polycrystalline films with grain size of ~300 nm to 1 micron. The images are only subtly changed after 1 spin-casting/heating cycle (figure 5-1b) and the grains appear to be smoothed with some edges appearing to be eroded or perhaps decorated with a thin deposit. There is no evidence of a conformal or complete coating. After five cycles, the surface morphology is significantly changed (figure 5-1c) and small protrusions, or nodules, are evident. These are ~100 to 200 nm in extent, and are present on the surfaces of the grains. In some locations, it appears that the grains have developed new terraces, suggesting that the surface energy may have been reduced due to reconstruction. AES mapping yielded very poor signal-to-noise Al maps with no clear correlation between the

Al content and the structures observed in the SEM images (table 5.1). However, aluminum was clearly detected in AES data that was acquired while integrating the Al signal while scanning 2 µm x 2 µm areas of both the 1-cycle and 5-cycle samples with higher Al signal strength for the latter sample. Interestingly, the nodules observed in the SEM images of the 5 cycle sample were not well correlated with the aluminum AES signals (table 5.1).

Figure 5-1d shows integrated Al 2p XPS data collected by scanning 2 micron by 2 micron areas of the samples. The aluminum signal is evident after only 1 spin/heating cycle, and becomes more intense after five spin/heating cycles. Based on alignment of the spectra by setting the Te2- peak to 572.4 eV,[74] the Al signal appears at a binding energy

53 that is consistent with Al3+ bonding in alumina.[ref] In an attempt to quantify the amount of Al on the surface, an alumina film prepared by ALD on a CdTe single crystal was examined as a calibration standard. The calibration standard was prepared on the CdTe native oxide using eight water/trimethylaluminum reaction cycles, resulting in an estimated thickness of 1.0 ± 0.5 nm.[74] The Al/(Cd+Te) signal ratio determined by XPS for the sample with 1 spin-casting/heating cycle was approximately 1/10th of that measured for the calibration standard. This result, coupled with the nonuniformities observed by SEM and

AFM (figure 5-2a-b), indicates that the passivation effect is realized through surface chemistry reactions at specific locations on the grains of the polycrystalline film.

Maruyama and Arai produced stoichiometric Al2O3 films on silicon wafers by heating Al(acac)3 powder to 150 ºC and introducing the entrained vapor to samples heated to temperatures between 250 and 600 ºC[75]. The films were adherent suggesting that the reaction occurred at the surface rather than in the gas phase. Our work suggests that the aluminum deposition reaction from Al(acac)3 is also surface specific in the presence of polycrystalline CdTe. However, at present, due to the rough topography of the surface and the very small amounts of material involved, we have no direct evidence for the formation of Al2O3 or information about the bonding or residues after the spin/heat cycles. X-ray diffraction and infrared spectroscopy analyses were inconclusive in identifying surface

Al2O3, even for samples made with 9 spin/heating cycles.

Figure 5-1. SEM images of Al2O3 on CdTe for (a) 0 cycle, (b) 1 cycle, and (c) 5 cycles.

(d) XPS data for these samples.

Table 5.1: Auger depth signals of the CdTe without and with different coating cycles of

Al2O3.

Sample Al % Cd % Te % CdTe 0.7 51.6 47.7

CdTe/Al2O3 5.5 48.9 45.6 (1 Cycle) CdTe/Al2O3 19.8 41.7 38.5 (3 Cycle) CdTe/Al2O3 25.8 37.9 36.3 (5 Cycle) CdTe/Al2O3 33.0 34.4 32.6 (7 Cycle) CdTe/Al2O3 50.7 25.2 24.1 (9 Cycle)

Figure 5-2. (a) The AFM images of CdTe, (b) of CdTe/Al2O3 (1 cycle), and (c) of

CdTe/Al2O3 (5 cycles) samples.

5.3.2 Device Performance

Figure 5-3 shows the J-V characteristics and key PV parameters as a function of the number of spin-coating/heating cycles. All samples were prepared from the center part of a 60 cm x 120 cm plate that was produced by a large-area commercial deposition process. Because the process produced uniform films, we can consider the initial starting

TECTM- 15M/CdS/CdTe device stacks for the individual experiments were essentially constant. Note that the standard deviation in the measured PV parameters for sets of 20 devices in each experiment showed typical variation of only 1-2% in Jsc, VOC, and FF, and

~3% in PCE (table 5-2). Figure 5-3a shows the J-V curves from the best devices for zero,

1, 3, 5, 7 and 9 spin-coating/heating cycles. The short circuit current density (Jsc) values are essentially constant within error, and small difference can be attributed to small deviations in the CdS emitter layer thickness. On the other hand, figure 5-3c shows that

st the VOC increases abruptly with the 1 cycle and stays nearly constant with additional cycles. The 1st cycle J-V curve also shows an improved FF, but increasing the number of cycles produces first a kink, and then a strong rollover/blocking effect. Figure 5-3b shows

56 that the overall efficiency trend is dominated by the FF, with a peak in PCE after the initial increase in the VOC. Data for the full population of devices can be found in table 5-2.

The trend of VOC with increasing spin/heat cycles can be clearly seen in figure 5-

3c. A similar trend is shown for the measured carrier lifetimes, which were extracted from biexponential fits to the PL transients produced by pulsed 532 nm excitation (figure 5-4c).

Since no changes are being made to the front of the cell, the fact that the two curves track each other can be attributed to changes in the back contact. Note that no Cu was used in these devices, so possible redistribution of dopants with the heating step can be discounted.

The PL (figure 5-4a) was excited and detected through the front glass of the device and the

1/e penetration depth for 532 nm is ~125 nm. Nevertheless, it seems clear that the minority carrier diffusion length is sufficient to allow a portion of the recombination kinetics to be influenced by the energetics at the back surface. Note that diffusion lengths several times longer than the 3 micron CdTe thickness used here have been reported for single crystals[76]and CdSeTe alloys in double heterostructures[71]. The increase in PL lifetime can be attributed to either a reduction in defect state surface density at the interface, or a reduction in rear-surface band bending due to affixed negative charge, either of which could produce passivation effects.[20].

Figure 5-3. (a) J-V characteristics for CdTe/Au and CdTe/Al2O3(1,3,5,7, and

9cycles)/Audevices. (b) average PCE vs average FF. (c) average VOC vs Tau2.

Figure 5-4. (a) The average PL spectrum of the CdS/CdTe thin film solar cell passivated

by using the Al2O3. (b) TRPL decay of the CdS/CdTe thin film solar cell

passivated by using the Al2O3.

5.3.3 Device with Cu Doping

In an effort to increase the PCE further, we used Cu to improve the level of p-type doping and reduce the Schottky barrier at the back surface[77-79] Figure 5-5 shows the J-

V curves for a CdTe device with an optimized standard Cu/Au back contact (3 nm thick evaporated Cu followed by 150 °C for 40 min in air) as compared devices fabricated with surface Al2O3 (1 cycle) and 3 nm of Cu and 40 nm Au with Cu diffusion times (150 ºC in air) of 40, 60 and 80 minutes. After the Cu diffusion process alone (no Al2O3) the PCE improved from 10.4 to 12.0 % (figures 5-1a and 5-3a, and table 5.2). Note that the performance of the devices with the optimized standard Cu/Au back contact exceeded the

PCE of the devices with one Al2O3 cycle and no Cu (12.0 versus 11.7%, respectively).

Adding 3 nm of Cu on top of the 1-cycle Al2O3 film and extending the heating time to 60 minutes increased the VOC and the FF without changing JSC, leading to an increase of the

PCE of the devices to 12.9%, and a best cell efficiency of 13.6% (fig. 5-5, table 5-1).

59 Recent device simulations which explored the use of back buffer layers for high efficiency CdTe solar cells make it clear that a combined increase in FF and VOC is one hallmark of a reduction in back surface recombination[20]. concluded that their 1 nm ALD alumina coating did produce back surface passivation,[72] but the change in slope of their

J-V curves at zero bias is not consistent with simulations and can be more appropriately explained by shunt passivation.

Figure 5-5. J-V characteristic of CdTe/Cu/Au and CdTe/Al2O3 (1 cycle)/Cu/Au with

different copper diffusion time.

60 Table 5.2: J-V performance data for devices fabricated with and without Cu and Al2O3

deposition/heating cycles. Data is presented for the best device and the

population of devices in each data set (n > 20).

Device VOC (mV) JSC FF PCE (mA/cm2) (%) (%) Without Cu Doping CdTe/Au Average 0.700±0.013 21.0±0.2 70.9±0.8 10.4±0.4

Best 0.726 21.5 72.0 11.3 CdTe/Al2O3 Average 0.756±0.008 21.2±0.3 73.1±0.9 11.7±0.3 (1 cycle)/Au Best 0.770 21.6 74.9 12.5

CdTe/Al2O3 Average 0.767±0.005 20.5±0.2 68.9±1.1 10.8±0.3 (3 cycles)/Au Best 0.771 20.7 70.4 11.3

CdTe/Al2O3 Average 0.755±0.010 20.5±0.2 58.8±1.4 9.1±0.3 (5 cycles)/Au Best 0.785 20.4 60.4 9.7

CdTe/Al2O3 Average 0.753±0.014 20.4±0.1 47.7±1.2 7.3±0.3 (7 cycles)/Au Best 0.772 20.6 50.1 7.9 CdTe/Al2O3 Average 0.781±0.037 20.3±0.4 37.9±3.7 6.0±0.7 (9 cycles)/Au Best 0.830 20.7 40.1 6.9 With Cu Doping CdTe/Cu/Au Average 0.783±0.004 21.1±0.3 72.5±0.9 12.0±0.2

150 °C 40 min Best 0.790 21.8 71.5 12.3 CdTe/Al2O3 Average 0.800±0.008 20.8±0.5 76.4±0.8 12.7±0.2 (1 cycle)/Cu/Au Best 0.806 21.8 75.6 13.3 150 °C 40 min CdTe/Al2O3 Average 0.818±0.007 20.7±0.4 76.2±0.7 12.9±0.3 (1 cycle)/Cu/Au Best 0.830 21.5 76.2 13.6 150 °C 60 min CdTe/Al2O3 Average 0.821±0.006 20.5±0.3 75.8±1 12.7±0.3 (1 cycle)/Cu/Au Best 0.829 20.7 76.4 13.1 150 °C 80 min

61 5.3.4 External Quantum Efficiency Measurements with Bias Voltage

To investigate whether back surface passivation is responsible for the improvements observed here we performed external quantum efficiency (EQE) measurements with the devices biased at their respective maximum power points (MPPs) under an AM1.5 light bias. It is most instructive to exam the wavelength-dependent carrier collection efficiency under this condition, as opposed to typical EQE measurements which are done at zero bias (short-circuit), without any additional light, because the energy bands in the emitter and absorber semiconductor layers will be fairly flat and poised in their normal operating condition. At MPP it is possible to examine back surface passivation effects by measuring the collection efficiency for carriers that are generated deeper in the device.

Figure 5-6c compares the data obtained at zero bias and at MPP for a device with an optimized standard Cu/Au back contact to that obtained from a device with an optimized

1-cycle Al2O3/Cu/Au back contact. We first note that the zero-bias data for the two devices overlaps completely, consistent with the JSC values being the same. This result is expected since the electric field in the absorber is high at zero bias and the impact of back surface recombination should not be reflected in the data.

Turning to the EQE data obtained at the MPP, the Cu/Au contact (MPP = 650 mV) shows higher response in the short wavelength region of the spectrum (< 500 nm) presumably due to Cu compensation of donor sites in CdS window layer[80]. The absence of an increase in the short wavelength response for the 1-cycle Al2O3/Cu/Au (MPP = 675 mV) is consistent with the thin Al2O3 layer reducing Cu diffusion into the CdS. In the long wavelength portion of the spectrum (> 500 nm), the Cu/Au contact shows a reduction in

62 the carrier collection efficiency as the band edge is approached. This is due to the downward band bending at the back surface, which causes minority carrier electrons that are generated nearby to be attracted to the back surface where they will more efficiently recombine. In contrast, the 1-cycle Al2O3/Cu/Au sample shows a long wavelength response that has the same spectral dependence (i.e. slope) as does the zero-bias data. The reduction in the EQE that is constant with wavelength in comparison to the zero-bias data is a reflection of the fact the Schockley-Read-Hall contribution to the overall recombination is larger throughout the device when the bands are flatter, as is the case at

MPP. The fact that the spectral dependence of the EQE is the same as in the zero-bias data, where the internal field is larger, indicates that the carriers generated throughout the device, regardless of depth, have a similar probability of being collected. Consequently, we can conclude that band bending near the back of the device no longer plays the same role in increasing back-surface recombination. This could be due to either the addition of fixed negative charge, which would set-up a so-called “back surface field”, or the passivation/elimination of surface states.

Figure 5-6. Bias dependent EQE for CdTe/Cu/Au and CdTe/Al2O3/Cu/Au devices. Slope

of EQE in infrared at MPP shows back surface passivation for

CdTe/Al2O3/Cu/Au device.

5.4 Conclusion

To summarize, the Al2O3 layer was spin coated by precursor Al(acac)3 on the back surface of CdTe to use as a passivation layer. With the optimized spin coating layer at 1 cycle, the Al2O3 partially covered the surface of CdTe and improved device performance by increasing VOC and FF due to reduce carrier recombination at the back of the devices.

Time-resolved photoluminescence spectroscopy measurements show that the carrier lifetime of the CdCl2 treated CdTe sample was greatly increased with the use of Al2O3,

64 indicating that Al2O3 reduces the interface recombination the back surface. This suggests that the solution processed Al2O3 layer can act as a passivation layer for the rear surface of

CdTe.

65 Chapter 6

Development of CdCl2 to Minimize Zn Loss from Sputtered CZT Thin Films for Use in Tandem Solar Cells

In addition to studying the back contact for CdTe solar cells, an investigation to reduce the Zn loss from CZT film was also completed. CZT thin films show great potential to use as a BBL for CdTe solar cells (as described in chapter four). Based on that project, the capability of making CZT targets in our lab was developed because, at that time, we do not have a commercial target and a homemade one was used. By utilizing a homemade target, it is possible to have targets with different band gaps by varying Zn% in the alloy, saving both time and money. With this advantage, we started to investigate the CZT as an absorber layer which is a promising candidate in tandem structures due to its tunable band gap (1.48-2.23 eV). However, one of the great challenges in CZT thin film solar cells is the activation of CdCl2 due to the loss of Zn. As an effort to address this issue, this chapter will discuss a route to do the CdCl2 treatment for CZT films while maintaining the Zn amount. The results in this chapter have been published in Alfadhili et al., 2018[64] and printed from MRS Advances, Vol 3, Fadhil K. Alfadhili, Geethika K. Liyanage, Adam B.

Phillips , and Michael J. Heben, Development of CdCl2 Activation to Minimize Zn Loss from Sputtered Cd1-xZnxTe Thin Films for Use in Tandem Solar Cells, Pages 9129-3134,

66 Copyright (2018), with permission from Cambridge University Press with License Number

4823670557440, Jal. 2018.

6.1 Introduction

Over the past five years, progress in CdTe solar cells have resulted in a record efficiency, 22.1% [81], approaching that of the Schockley-Queissar limit [9]. As the device performance has improved, the cost of manufacturing CdTe modules has decreased, leading to an average production cost of $0.50 /WP [82], a number that makes power production from CdTe modules competitive with power generated using traditional sources. The next step for improving the efficiency of thin film solar cells beyond 25% is the development of tandem devices [83]. Modelling work done by Coutts et al. [84] shows that a 15% top cell with 1.72 eV bandgap is required to achieve a 25% two terminal polycrystalline tandem. If the low-cost production methods used for CdTe solar cells can be adopted for a polycrystalline top cell, low cost tandems may be possible.

Cd1-xZnxTe (CZT) is a strong candidate for the top cell in polycrystalline tandems

[83]. It is made by alloying CdTe and ZnTe; therefore the device fabrication is expected to follow that of CdTe devices. The bandgap of CZT ranges from 1.48 eV to 2.23 eV and can be controlled by varying the Zn concentration [85]. While reported CZT device efficiency is below 15% [86, 87], improving the band alignment at the front of the device is expected to result in increased values [88]. In addition to low efficiency, the other challenge with

CZT is that Zn reacts during CdCl2 activation, reducing the band gap of the device [89].

To be used as the top cell in a tandem, this Zn loss must be eliminated or controlled.

Here, we investigated several CdCl2 activation methods. We show that activation in a flowing oxygen-containing atmosphere results in the formation of ZnO in the film.

67 Even without the oxygen, treatment with flowing gas results in significant Zn loss from the film. By using a closed system, we are able to minimize the Zn loss while still activating the film.

6.2 Experimental details

Partial device stacks were fabricated on fluorine-doped tin oxide (FTO) coated glass substrates. The substrates (TEC15M; Pilkington N.A.) were cleaned in an ultrasonic bath using diluted Micro-90 detergent and deionized water. Samples were then rinsed with deionized water and thoroughly dried using nitrogen gas. A 100 nm cadmium sulphide

(CdS; 99.999% Materion target) film was deposited on the cleaned substrates maintained at 250 °C using RF magnetron sputtering (0.41 W/cm2, 15 mTorr and 23 sccm of Ar). A 2

μm thick CZT film was deposited at 250 °C on top of the CdS film using RF magnetron

2 sputtering (1.23 W/cm , 10 mTorr). Prior to CdCl2 activation, all samples were annealed at 450 °C in H2/He ambient with 2.5% H2 for 30 min to passive the interface states [87,

90].

6.3 Results and Discussion

To determine the effects the CdCl2 activation process has on CZT films, three processes were completed. For all three activations, the CdCl2 was applied to the CZT film using three drops of a saturated CdCl2-methanol solution and allowed to dry. The heating process and ambient environments were varied. The first is a “standard” process for CdTe devices [17]. The sample was placed in a graphite box and purged with N2 for a total of four volume exchanges prior to heating. Dry air was slowly flowed through the box as it was heated to 387 °C and during the 30 minute heating process. After the heating period, the system was allowed to cool with flowing N2. The second activation process was

68 identical to the first, but N2 gas was used instead of dry air. For the third activation process, the samples were placed in a graphite box in a N2 filled glove box and sealed using a graphite gasket. In addition to the samples, Zn metal powder (99.9%; Sigma Aldrich) was included in the box, see figure 6-1 for a schematic diagram. The box was then placed on a hot plate and allowed to reach 387 °C. After heating at this temperature for 30 minutes, the graphite box was removed from the hot plate and placed on the table.

Figure 6-1. Schematic diagram of the system used for CdCl2 treatment.

6.3.1 Optical Properties

Figure 6-2 shows the Tauc plots for the CZT films before and after treatments. The bandgaps were calculated by fitting equation 6.1 for direct band gap semiconductor.

2 (αhν) =A (Eg-hν) (6.1)

All the films have sharp band edge which suggests the good crystalline quality of the films.

As-deposited films have a bandgap of 1.82 eV, which is about 9% lower than the expected value for the Cd0.3Zn0.7Te composition of the prepared sputter target. This discrepancy between the target and film composition is likely due to the different sputtering rates for

69 ZnTe and CdTe and different sticking coefficients when sputtering at elevated substrate temperatures [91]. After annealing the films at 450 °C for 30 minutes in a H2/He environment, the bandgap reduces slightly to 1.80 eV.

2 Figure 6-2. The plot of (αhν) vs. hν of as-deposited, pre-annealed at 450 °C in H2/He

ambient, CdCl2 activated without flowing gas, with flowing N2, and with

flowing dry air films.

Both CdCl2 activation processes completed in flowing gas show significant reduction in bandgap from 1.80 eV to ~1.61 eV. For the CZT films with CdCl2 treatment done in the enclosed box, only a small reduction in bandgap was observed to 1.76 eV.

These results suggest that CdCl2 activation in a closed box with sacrificial Zn can maintain the bandgap needed for the top cell of tandem devices.

6.3.2 XRD Spectroscopy Analysis

To determine how the different CdCl2 activation processes affect the crystal structure, the XRD patterns were obtained for all the CZT films and are shown in figure 6-

3 a. All the CZT films exhibit peaks for the cubic structure of the CZT alloy with the strong

70 preferred orientation (111) peak [86]. After the annealing, the intensity of peaks increases, indicating improvement of crystal quality of all the films. After all CdCl2 activation processes, new peaks in the diffraction spectra relating to (400), (331), (422), and (333) appear, indicating more randomization of the crystal structure.

Upon CdCl2 activation in flowing gas, the (111) peak position shifted to lower values of 2θ (figure 6-3b), indicating a loss of Zn from CZT alloy. As the XRD data shows

(figure 6-3a), both ZnO and ZnCl2 are present in these films. In addition, pure Te is detected, forming as the Zn reacts with the oxygen and chlorine. What is surprising is that

ZnO appears even when the activation process was completed in pure N2. It is likely that this system is leaky and that a small concentration of oxygen is present. It is possible that in absence of this leak, the bandgap of the CZT film would not change. This is unlikely because others have shown that CdCl2 activation in flowing inert gas still results in a reduction of the bandgap [87, 89].

When the activation occurred with a Zn over-pressure in a confined box, loss of

Zn did not occur. Clearly performing the treatment in an inert atmosphere will prevent ZnO formation, but the inclusion of Zn appears to prevent the formation of ZnCl2 in the CZT film. This could be due to the chemical stability of Cd and Zn. The oxidation potential of

Cd is lower than of Zn, so the interaction between CdCl2 and Zn in the CZT alloy is energetically favorable yielding equation 6.2.

ZnTe + CdCl2 → CdTe + ZnCl2 (6.2)

71 This equation indicates that the ZnTe can be easily converted to CdTe at high temperature in the presence of CdCl2 [92]. The results presented here suggest that the activation energy of interaction between ZnTe and CdCl2 without flowing gas, when the excess Zn can create a ZnCl2 overpressure, is higher than with flowing gas. Additionally, the vapor pressure of

ZnCl2 is higher than that of CdCl2 and CdTe at the treatment temperature [87], which could account for additional loss of Zn during activation in an inert flowing gas.

Interestingly, the CZT peaks in the XRD spectra split for films that underwent

CdCl2 activation in dry air. This suggests that there is distinct second CZT phase, but this phase is not pure CdTe. One of the phases matches well with the CZT phase of films activated in flowing N2. This may indicate that the CdCl2 interaction front propagates through the film at one rate while the oxygen interaction front propagates more slowly. If this is the case, the entire CZT film could have interacted with the CdCl2 but only a small portion would have interacted with oxygen. This, in turn, would leave the “top” CZT film that is Zn poor and ZnO rich, which would have a lower bandgap, and could account for the distinct curve shapes in the Tauc plot in figure 6-2.

Figure 6-3. (a) XRD pattern of as-deposited, pre-annealed at 450°C in H2/He ambient,

CdCl2 activated without flowing gas, with flowing N2, and with flowing dry

air films. (b) Expanded view of the (111) peak in (a).

6.3.3 Auger Spectroscopy Analysis

To further investigate how the composition of the CZT films evolved during the different CdCl2 activations, Auger depth profile spectra of the films were measured (figure

6-4 a, b, and c). All the CZT films are Zn-rich, which is consistent with XRD data and the band gaps of the films. The Zn fraction (x) in the Cd1-xZnxTe films for the as-deposited and film activated without flowing gas are 0.64 and 0.61 (figure 6-4a and b), which corresponds to 1.95 and 1.93 eV, respectively [16]. While these values are higher than the values obtained from optical measurements, these results further demonstrate that the reduction of Zn and, therefore, bandgap is small after CdCl2 treatment without flowing gas.

Interestingly, some of the Zn loss for the sample CdCl2 treated without flowing gas is due to Zn interaction with the CdS window layer, forming a ZnS phase. This could reduce

73 device performance and provides further support for replacing the CdS window with

MgZnO with the appropriate band gap [38].

The Auger depth profile spectrum of the sample activated in dry air is significantly different from the other two (figure 6-4c). The concentration of Zn varies throughout the depth of the film, indicating that the Zn diffuses during the CdCl2 process. As proposed above, there appears to be a ZnO-rich layer at the top of the film. While it is difficult to determine how much of the Zn remains in the CZT film, the Zn:O ratio appears to vary, suggesting that the composition of the CZT changes as well. This would lead to a graded band gap or even a multiphase layer, as suggested by the XRD data. Focusing on the front window area of the graph, it appears that the Zn signal increases faster than the O signal.

This would indicate the formation of ZnS at the front, similar to the film where the CdCl2 treatment was performed without flowing gas. In general, the Auger depth profile shows that activation in flowing dry air results in a series of interactions that are difficult to control, while activation in a static inert atmosphere is well behaved.

Figure 6-4. Auger depth profile spectra of (a) as-deposited, (b) CdCl2 activated in a static

inert atmosphere, and (c) CdCl2 activated with flowing dry air films.

6.4 Conclusion

Cd0.3Zn0.7Te films were sputtered onto TEC15M/CdS substrates to undergo various

CdCl2 activation processes. The as-deposited film had a preferred (111) orientation and a bandgap of 1.86 eV. After annealing in an H2/He environment the bandgap reduced slightly to 1.80 eV. All CdCl2 treatments improved the polycrystalline structure. However, the bandgap was shifted from 1.80 eV to 1.62 eV when activation occurred in flowing gas.

Furthermore, the Auger depth profile showed that the Zn reacts with both oxygen and sulfur, resulting in varying Zn concentrations throughout the depth of the film. The bandgap of CZT film treated at 390 °C for 30 min without flowing gas, on the other hand,

75 reduced slightly from 1.80 eV to 1.76 eV, while Auger depth profiling show that these films are well behaved even though and intermixing with the CdS layer occurred. The results suggest that CdCl2 activation of CZT films in a static, inert atmosphere is well behaved and can be used for a controlled device processing.

76 Chapter 7

Summary and Future Work

7.1 Thesis Summary

This thesis had one main topic and one secondary topic. The main one was investigating back contacts to CdTe solar cells and the second was investigating Zn loss in

CdZnTe thin films. The first was to use semiconducting materials such as SWCNT, ZnTe,

CdZnTe as a back buffer layer to reduce the back-barrier height and enhance hole collection and reduce electron attraction at the back electrode. The second approach was to explore back surface passivation using solution processing of an Al2O3 layer. The other focus was to investigate methods for CdCl2 activation of wide bandgap CdZnTe thin films in an attempt to reduce Zn loss from the films.

The first topics discussed in this thesis was using doped SWCNTs as a transparent back contact for CdTe devices. SWCNT films were doped using OA. This doping produced in a highly transparent and electrically conductive SWCNT thin film. The intensities of the absorption peaks at ~550 nm and ~1000 nm, corresponding to S22 and S11 transition energies completely quenched. The doping also reduced the electrical resistance of the

SWCNT film. These results inducted that hole carrier density increased and that the Fermi level in the SWCNTs pushed deep into the valence band. This improvement was expected

77 to reduce the band bending at the CdTe/SWCNT interface which could increase the Voc due to reduce the carrier recombination at the back interface. However, it was noticed that during the doping process, the OA interacted more strongly with the CdTe layer than the

SWCNTs, resulting in reduced device performance. The average power conversion efficiency decreased from 11.8% for a standard Cu/Au back contacted device to 10.6% for the OA-doped SWCNT back contacted device.

To further investigate how the alignment at the back can affect and improve the device performance, different semiconducting materials were applied. First, I incorporated

ZnTe and Te as the BBL to determine which material leads to the best device performance.

Using these two materials, eight different back contacts were deposited onto CdCl2 activated CdTe samples. ZnTe was deposited by RF sputtering while Te was formed by solution processing using the MAI treatment. This study showed that using Te in contact with CdTe resulted in higher performance than using ZnTe in contact with the CdTe. Low temperature current density-voltage measurements showed that Te resulted in a lower barrier with CdTe than ZnTe, indicating that Te has more suitable band alignment, resulting in less downward bending in the CdTe at the back interface than ZnTe does.

In addition, I investigated CdZnTe as a back contact for CdTe solar cells. We showed that the CZT film was easily doped with copper (Cu). Devices with CZT:Cu/Au and CZT:Cu/Te/Au back contacts demonstrated significant gains in VOC compared to

Cu/Au back contacts. Barrier height measurements indicated that the valence band edge of the CZT is close to that of CdTe, which, when coupled with the high dopant density, should result in upward band bending at the back of the device. In the case of the CZT:Cu/Au back contact, the large band offset between the CZT and Au results in a depletion region that

78 extends into the CdTe, limiting device performance. When the alignment was improved with the incorporation of Te, the efficiency improved.

The reduction of the interface recombination can also be accomplished by decreasing the concentration of electrically active defects at the interface. Passivation of the rear surface of CdTe solar cells plays a crucial role in improving the lifetime, which increases the photoconversion efficiency of CdTe devices. Incorporating an oxide layer to minimize interface recombination can significantly increase the carrier collection at the back contact of CdTe devices. Recently, researchers have developed a route to passivate the rear surface by incorporating aluminum oxide layer using atomic layer or sputtering deposition methods to improve CdTe device performance and carrier lifetimes. In this study, we showed that aluminum oxide can be formed on the CdTe surface using solution processing. Both atomic force microscope and Auger electron spectroscopy confirmed the formation of Al2O3 with as few as one cycle. The Al2O3 layer at the back of CdTe devices increases photoluminescence (PL) intensity and time-resolved photoluminescence (TRPL) decay lifetimes. The PCE for devices with a standard Cu/Au back contact was improved from 12.2% to 13.6 % with the addition of the solution-processed Al2O3 due to the improvement of the open-circuit voltage (VOC) and fill factor (FF), indicating that the solution-processed Al2O3 has the potential to reduce interface recombination.

Finally, the effect of CdCl2 treatment on wide band gap CdZnTe films was studied by varying the treatment atmosphere. CdZnTe can be achieved by alloying CdTe with

ZnTe. The bandgap of the CdZnTe film ranges from 1.48 eV to 2.23 eV and can be controlled by varying the Zn concentration. Like CdTe, the alloyed films are expected to allow for low-cost production, suggesting that CdZnTe could be an ideal top cell for mass-

79 produced tandem devices. However, the CdCl2 activation of the alloyed films results in a significant loss of Zn, thereby reducing the bandgap. In this study, we demonstrated a novel

CdCl2 activation method that did not result in significant Zn loss. By performing the activation step in a closed, inert environment we were able to avoid oxidation of the Zn in the CdZnTe film; furthermore, by including sacrificial Zn in the container, an overpressure of ZnCl2 forms limiting the amount of ZnCl2 formed in the film. X-ray diffraction, optical measurements, and Auger spectroscopy showed that the CdCl2 treatment with no flowing gas minimizes the loss of Zn from the CZT alloy.

7.2 Future Work

There are several interesting future works based on the work of this thesis. The first project could be linked to our successful back buffer layer of CZT: Cu/Te configuration.

The second future project is based on the SWCNTs film and Al2O3 layer. This project has already begun and expected to produce highly efficient back contact. In the following sections, these future works are summarized.

7.2.1 The CZT Back Buffer Layer

As described in chapter 4, inserting the buffer layer of CZT film between the absorber layer (CdTe) and back electrode reduces the band beinding and minimizes such recombination losses. However, in this study, a fixed composition of Cd and Zn in the film

(Cd0.4Zn0.6Te) was used. For the future work, we would like to investigate how the composition of Cd and Zn, as well as the dopant density in the CZT film, will affect the band alignment at the back of CdTe device.

80 7.2.2 Using Al2O3 and SWCNTs as a Point Contact

We showed in Chapter 5 that using the Al2O3 based on the solution processing has the potential to passivate the rear surface of CdTe. Carrier lifetime has improved from 0.37 to 1.99 ns with the addition of the five cycles of Al2O3. However, in these cases the Al2O3 fully covers the surface of CdTe ( figure 6-1a) and with this thickness the device performance suffers (figure 7-1b).

Figure 7-1. (a) The SEM image of 5 cycles on CdTe. (b) J-V curves of CdTe device with

no Al2O3 and 5 cycles.

To passivate the rear surfce of CdTe while potentially enabling the hole carrieres transport through the passivating layer, we need to form a contact bewteen CdTe and back electrode. To address this issue, we have invistigted the use of SWCNTs as a point contact approach. Figures 7-2 showes the schematic diagram for our point contact. We reduced this to practice by employing using a solution process. A solution of dispersed SWCNTs was mixed with a precursor solution for Al2O3 layer. Using this method, the SWCNTs were embedded in the passivation material (Al2O3) during deposition of the layer. Figure 7-3 shows how the device performance changed as we varied the concentration of the SWCNTs

81 and the Al2O3 precursor in solution. Further investigation is required to achieve the potential of this novel back contact.

Figure 7-2. Schematic of CdTe with a SWCNT point contact/passivation layer (Al2O3).

Figure 7-3. (a) Fill factor and (b) open circuit voltage of CdTe devices with back buffer

layers consisting of varying ratios of Al2O3 and SWCNTs. The black box on

the left shows a Cu/Au reference cell. None of the other cells include the

82 appropriate Cu doping, so the FF is suppressed. (c) J-V crves for the best CdTe device with different SWCNT concentration with 6.15 mM Al(acac)3. (d) J-V crves for the best CdTe device with different SWCNT concentration with 123 mM Al(acac)3.

83 References

[1] BP, "Statistical Review of World Energy," 2019, vol. 68th edition.

[2] D. M. Chapin, C. S. Fuller, and G. L. Pearson, "A New Silicon p‐n Junction

Photocell for Converting Solar Radiation into Electrical Power," Journal of

Applied Physics, vol. 25, no. 5, pp. 676-677, 1954.

[3] REN21, "Renewables 2019 Global Status Report (Paris: REN21 Secretariat,"

2019, Available: https://www.ren21.net/wp-

content/uploads/2019/05/gsr_2019_full_report_en.pdf.

[4] Y. Chen et al., "From Laboratory to Production: Learning Models of Efficiency

and Manufacturing Cost of Industrial Crystalline Silicon and Thin-Film

Photovoltaic Technologies," IEEE Journal of Photovoltaics, vol. 8, no. 6, pp.

1531-1538, 2018.

[5] s. Kosap, Principles of electronic Materials and Devices. McGraw-Hill NY,

2018.

[6] J. Schmidt, A. Merkle, R. Brendel, B. Hoex, M. C. M. v. de Sanden, and W. M.

M. Kessels, "Surface passivation of high-efficiency silicon solar cells by atomic-

layer-deposited Al2O3," Progress in Photovoltaics: Research and Applications,

vol. 16, no. 6, pp. 461-466, 2008.

[7] F. Solar, "Manufacturing Update," 2016.

84 [8] (2018, July 15). Solar cell efficiency chart. Available:

https://www.nrel.gov/pv/assets/images/efficiency-chart.png

[9] W. Shockley and H. J. Queisser, "Detailed Balance Limit of Efficiency of p‐n

Junction Solar Cells," Journal of Applied Physics, vol. 32, no. 3, pp. 510-519,

1961.

[10] A. H. Munshi et al., "Polycrystalline CdTe photovoltaics with efficiency over

18% through improved absorber passivation and current collection," Solar Energy

Materials and Solar Cells, vol. 176, pp. 9-18, 3// 2018.

[11] A. Abbas et al., "Cadmium chloride assisted re-crystallization of CdTe: The effect

of annealing over-treatment," in 2014 IEEE 40th Photovoltaic Specialist

Conference (PVSC), 2014, pp. 0701-0706.

[12] C. Li et al., "Grain-Boundary-Enhanced Carrier Collection in CdTe Solar Cells,"

Physical Review Letters, vol. 112, no. 15, p. 156103, 04/16/ 2014.

[13] J. Freeouf and J. Woodall, "Schottky barriers: An effective work function model,"

Applied Physics Letters, vol. 39, pp. 727-729, 12/01 1981.

[14] C. R. Corwine, A. O. Pudov, M. Gloeckler, S. H. Demtsu, and J. R. Sites,

"Copper inclusion and migration from the back contact in CdTe solar cells," Solar

Energy Materials and Solar Cells, vol. 82, no. 4, pp. 481-489, 5/30/ 2004.

[15] C. A. Wolden et al., "The roles of ZnTe buffer layers on CdTe solar cell

performance," Solar Energy Materials and Solar Cells, vol. 147, pp. 203-210, 4//

2016.

[16] T. Song, A. Moore, and J. R. Sites, "Te Layer to Reduce the CdTe Back-Contact

Barrier," IEEE Journal of Photovoltaics, vol. 8, no. 1, pp. 293-298, 2018.

85 [17] A. B. Phillips et al., "Wiring-up Carbon Single Wall Nanotubes to Polycrystalline

Inorganic Semiconductor Thin Films: Low-Barrier, Copper-Free Back Contact to

CdTe Solar Cells," Nano Letters, vol. 13, no. 11, pp. 5224-5232, 2013/11/13

2013.

[18] K. P. Bhandari et al., "Iron pyrite nanocrystal film serves as a copper-free back

contact for polycrystalline CdTe thin film solar cells," Solar Energy Materials

and Solar Cells, vol. 140, pp. 108-114, 9// 2015.

[19] F. K. Alfadhili, A. B. Phillips, G. K. Liyanage, J. M. Gibbs, M. K. Jamarkattel,

and M. J. Heben, "Controlling Band Alignment at the Back Interface of Cadmium

Telluride Solar Cells using ZnTe and Te Buffer Layers," MRS Advances, vol. 4,

no. 16, pp. 913-919, 2019.

[20] G. K. Liyanage, A. B. Phillips, F. K. Alfadhili, R. J. Ellingson, and M. J. Heben,

"The Role of Back Buffer Layers and Absorber Properties for >25% Efficient

CdTe Solar Cells," ACS Applied Energy Materials, 2019/07/08 2019.

[21] F. K. Alfadhili et al., "Use of Single Wall Carbon Nanotube films doped with

Triethyloxonium Hexachlorantimonate as a Transparent Back Contact for CdTe

Solar Cells," in 2017 IEEE 44th Photovoltaic Specialist Conference (PVSC),

2017, pp. 730-734.

[22] https://cleantechnica.com/2017/12/06/first-solar-unveils-series-6-thin-film-pv-

module-expected-2018-guidance/, ed, 2017.

[23] L. Kranz et al., "High-Efficiency Polycrystalline Thin Film Tandem Solar Cells,"

The Journal of Physical Chemistry Letters, vol. 6, no. 14, pp. 2676-2681,

2015/07/16 2015.

86 [24] R. W. Crisp et al., "Tandem Solar Cells from Solution-Processed CdTe and PbS

Quantum Dots Using a ZnTe–ZnO Tunnel Junction," Nano Letters, vol. 17, no. 2,

pp. 1020-1027, 2017/02/08 2017.

[25] http://www.nrel.gov/ncpv/images/efficiency_chart.jpg.

[26] J. P. Mailoa, M. Lee, M. Lee, T. Buonassisi, A. Panchula, and D. N. Weiss,

"Energy-yield prediction for II-VI-based thin-film tandem solar cells," Energy &

Environmental Science, 10.1039/C6EE01778A vol. 9, no. 8, pp. 2644-2653, 2016.

[27] O. S. Martinez, D. Reyes-Coronado, and X. Mathew, "Cd1-xMgxTe thin films

and top-cells for possible applications in tandem solar cells," 2009, vol. 7409, pp.

74090V-74090V-7.

[28] R. R. Khanal et al., "Substrate configuration, bifacial CdTe solar cells grown

directly on transparent single wall carbon nanotube back contacts," Solar Energy

Materials and Solar Cells, vol. 157, pp. 35-41, 2016/12/01/ 2016.

[29] T. M. Barnes et al., "Single-wall carbon nanotube networks as a transparent back

contact in CdTe solar cells," Applied Physics Letters, vol. 90, no. 24, p. 243503,

2007.

[30] R. Saito, M. Fujita, G. Dresselhaus, and M. S. Dresselhaus, "Electronic structure

of chiral graphene tubules," Applied Physics Letters, vol. 60, no. 18, pp. 2204-

2206, 1992.

[31] H. Kataura et al., "Optical properties of single-wall carbon nanotubes," Synthetic

Metals, vol. 103, no. 1, pp. 2555-2558, 1999/06/01/ 1999.

[32] H.-Z. Geng, K. K. Kim, K. P. So, Y. S. Lee, Y. Chang, and Y. H. Lee, "Effect of

Acid Treatment on Carbon Nanotube-Based Flexible Transparent Conducting

87 Films," Journal of the American Chemical Society, vol. 129, no. 25, pp. 7758-

7759, 2007/06/01 2007.

[33] B. Chandra, A. Afzali, N. Khare, M. M. El-Ashry, and G. S. Tulevski, "Stable

Charge-Transfer Doping of Transparent Single-Walled Carbon Nanotube Films,"

Chemistry of Materials, vol. 22, no. 18, pp. 5179-5183, 2010/09/28 2010.

[34] A. D. Avery et al., "Tailored semiconducting carbon nanotube networks with

enhanced thermoelectric properties," Nature Energy, vol. 1, p. 16033, 2016.

[35] R. C. Tenent et al., "Ultrasmooth, Large-Area, High-Uniformity, Conductive

Transparent Single-Walled-Carbon-Nanotube Films for Photovoltaics Produced

by Ultrasonic Spraying," Advanced Materials, vol. 21, no. 31, pp. 3210-3216,

2009.

[36] A. B. Phillips et al., "High speed, intermediate resolution, large area laser beam

induced current imaging and laser scribing system for photovoltaic devices and

modules," Review of Scientific Instruments, vol. 87, no. 9, p. 093708, 2016.

[37] Alex Niemegeers and Marc. Burgelman, "Effects of the Au/CdTe back contact on

IV and CV characteristics of Au/CdTe/CdS/TCO solar cells," Journal of Applied

Physics, vol. 81, 1997.

[38] http://files.shareholder.com/downloads/FSLR/1835294976x0x884412/1548B782-

59A0-4544-A452-989E1FA42BFE/FS_AnalystDay_ManufacturingUpdate.pdf.

[39] A. H. Munshi et al., "Polycrystalline CdSeTe/CdTe Absorber Cells With 28

mA/cm2 Short-Circuit Current," IEEE Journal of Photovoltaics, vol. 8, no. 1, pp.

310-314, 2018.

88 [40] T. A. Gessert, A. R. Mason, P. Sheldon, A. B. Swartzlander, D. Niles, and T. J.

Coutts, "Development of Cu‐doped ZnTe as a back‐contact interface layer for

thin‐film CdS/CdTe solar cells," Journal of Vacuum Science & Technology A,

vol. 14, no. 3, pp. 806-812, 1996.

[41] J. Li et al., "Controlled activation of ZnTe:Cu contacted CdTe solar cells using

rapid thermal processing," Solar Energy Materials and Solar Cells, vol. 133, pp.

208-215, 2// 2015.

[42] J. V. Li et al., "Electrical Characterization of Cu Composition Effects in

CdS/CdTe Thin-Film Solar Cells With a ZnTe:Cu Back Contact," IEEE Journal

of Photovoltaics, vol. 3, no. 3, pp. 1095-1099, 2013.

[43] Y. Kwon, J. Seo, Y. Kang, D. Kim, and J. Kim, "Bifacial CdS/CdTe thin-film

solar cells using a transparent silver nanowire/indium tin oxide back contact,"

Optics Express, vol. 26, no. 2, pp. A30-A38, 2018/01/22 2018.

[44] X. Li, D. W. Niles, F. S. Hasoon, R. J. Matson, and P. Sheldon, "Effect of nitric-

phosphoric acid etches on material properties and back-contact formation of

CdTe-based solar cells," Journal of Vacuum Science & Technology A, vol. 17, no.

3, pp. 805-809, 1999.

[45] D. L. Bätzner, A. Romeo, H. Zogg, R. Wendt, and A. N. Tiwari, "Development of

efficient and stable back contacts on CdTe/CdS solar cells," Thin Solid Films, vol.

387, no. 1, pp. 151-154, 2001/05/29/ 2001.

[46] R. Ochoa-Landín, O. Vigil-Galan, Y. V. Vorobiev, and R. Ramírez-Bon,

"Chemically-deposited Te layers improving the parameters of back contacts for

CdTe solar cells," Solar Energy, vol. 83, no. 1, pp. 134-138, 2009/01/01/ 2009.

89 [47] N. Abbas Shah, A. Ali, Z. Ali, A. Maqsood, and A. K. S. Aqili, "Properties of Te-

rich cadmium telluride thin films fabricated by closed space sublimation

technique," Journal of Crystal Growth, vol. 284, no. 3, pp. 477-485, 2005/11/01/

2005.

[48] A. Moore, T. Song, and J. Sites, "Improved CdTe Solar-Cell Performance with

An Evaporated Te Layer before The Back Contact," MRS Advances, vol. 2, no.

53, pp. 3195-3201, 2017.

[49] S. C. Watthage et al., "Selective Cd Removal From CdTe for High-Efficiency Te

Back-Contact Formation," IEEE Journal of Photovoltaics, vol. 8, no. 4, pp. 1125-

1131, 2018.

[50] S. C. Watthage et al., "Impact of Divalent Metal Additives on the Structural and

Optoelectronic Properties of CH3NH3PbI3 Perovskite Prepared by the Two-Step

Solution Process," MRS Advances, vol. 2, no. 21-22, pp. 1183-1188, 2017.

[51] R. G. Bohn, C. N. Tabory, C. Deak, M. Shao, A. D. Compaan, and N. Reiter, "RF

sputtered films of Cu-doped and N-doped ZnTe," in Proceedings of 1994 IEEE

1st World Conference on Photovoltaic Energy Conversion - WCPEC (A Joint

Conference of PVSC, PVSEC and PSEC), 1994, vol. 1, pp. 354-356 vol.1.

[52] A. Jarkov et al., "Photo-assisted electrodeposition of polypyrrole back contact to

CdS/CdTe solar cell structures," Thin Solid Films, vol. 535, pp. 198-201,

2013/05/15/ 2013.

[53] J.-W. Liu, F. Chen, M. Zhang, H. Qi, C.-L. Zhang, and S.-H. Yu, "Rapid

Microwave-Assisted Synthesis of Uniform Ultralong Te Nanowires, Optical

90 Property, and Chemical Stability," Langmuir, vol. 26, no. 13, pp. 11372-11377,

2010/07/06 2010.

[54] A. Niemegeers and M. Burgelman, Effects of the Au/CdTe back contact on IV and

CV characteristics of Au/CdTe/CdS/TCO solar cells. 1997, pp. 2881-2886.

[55] T. Shalvey, L. Phillips, K. Durose, and J. Major, A comparison of organic back

contact materials for CdTe solar cells. 2018.

[56] F. K. Alfadhili et al., "Potential of CdZnTe Thin Film Back Buffer Layer for

CdTe Solar Cells," IEEE 46th Photovoltaic Specialist

Conference (PVSC), 2019.

[57] M. Gloeckler, I. Sankin, and Z. Zhao, "CdTe Solar Cells at the Threshold to 20%

Efficiency," IEEE Journal of Photovoltaics, vol. 3, no. 4, pp. 1389-1393, 2013.

[58] T. Ablekim, E. Colegrove, and W. K. Metzger, "Interface Engineering for 25%

CdTe Solar Cells," ACS Applied Energy Materials, vol. 1, no. 10, pp. 5135-5139,

2018/10/22 2018.

[59] D. Rioux, D. W. Niles, and H. Höchst, "ZnTe: A potential interlayer to form low

resistance back contacts in CdS/CdTe solar cells," Journal of Applied Physics,

vol. 73, no. 12, pp. 8381-8385, 1993.

[60] K. P. Bhandari, S. C. Watthage, Z. Song, A. Phillips, M. J. Heben, and R. J.

Ellingson, "Applications of hybrid organic-inorganic metal halide perovskite thin

film as a hole transport layer in CdTe thin film solar cells," in 2017 IEEE 44th

Photovoltaic Specialist Conference (PVSC), 2017, pp. 748-751.

[61] Strevel Nicholas, Trippel Lou, Kotarba Chad, and K. Imran, "Improvements in

CdTe module reliability and long-term degradation through advances in

91 construction and device innovation," Photovoltaics International vol. 22, pp. 66-

74, 2014.

[62] F. K. Alfadhili, A. B. Phillips, G. K. Liyanage, J. M. Gibbs, M. K. Jamarkattel,

and M. J. Heben, "Controlling Band Alignment at the Back Interface of Cadmium

Telluride Solar Cells using ZnTe and Te Buffer Layers," MRS Advances, pp. 1-7.

[63] G. K. Liyanage, A. B. Phillips, F. K. Alfadhili, R. J. Ellingson, and M. J. Heben,

"Back Buffer Layer Requirements for >25% Efficient CdTe Solar Cells,"

Advanced Energy Materials, 2019.

[64] F. K. Alfadhili, G. K. Liyanage, A. B. Phillips, and M. J. Heben, "Development of

CdCl2 Activation to Minimize Zn Loss from Sputtered Cd1-xZnxTe Thin Films

for Use in Tandem Solar Cells," MRS Advances, vol. 3, no. 52, pp. 3129-3134,

2018.

[65] T. L. Chu, S. S. Chu, C. Ferekides, and J. Britt, "Films and junctions of cadmium

zinc telluride," Journal of Applied Physics, vol. 71, no. 11, pp. 5635-5640, 1992.

[66] S. H. Lee, A. Gupta, and A. D. Compaan, "Polycrystalline sputtered Cd(Zn,

Mn)Te films for top cells in PV tandem structures," physica status solidi (c), vol.

1, no. 4, pp. 1042-1045, 2004.

[67] W. K. Metzger et al., "Exceeding 20% efficiency with in situ group V doping in

polycrystalline CdTe solar cells," Nature Energy, vol. 4, no. 10, pp. 837-845,

2019/10/01 2019.

[68] Y. Jiang et al., "Passivation properties of alumina for multicrystalline silicon

nanostructure prepared by spin-coating method," Applied Physics A, vol. 124, no.

2, p. 95, 2018/01/09 2018.

92 [69] F. Werner et al., "Very low surface recombination velocities on p- and n-type c-Si

by ultrafast spatial atomic layer deposition of aluminum oxide," Applied Physics

Letters, vol. 97, no. 16, p. 162103, 2010.

[70] S. M. Jovanovic et al., "Optical characterization of epitaxial single crystal CdTe

thin films on Al2O3 (0001) substrates," Thin Solid Films, vol. 570, pp. 155-158,

2014/11/03/ 2014.

[71] J. M. Kephart et al., "Sputter-Deposited Oxides for Interface Passivation of CdTe

Photovoltaics," IEEE Journal of Photovoltaics, vol. 8, no. 2, pp. 587-593, 2018.

[72] J. Liang et al., "Rectification and tunneling effects enabled by Al2O3 atomic layer

deposited on back contact of CdTe solar cells," Applied Physics Letters, vol. 107,

no. 1, p. 013907, 2015.

[73] R. Watanabe, M. Kawashima, and Y. Saito, "Film properties of alumina

passivation layer for silicon solar cells prepared by spin-coating method," Thin

Solid Films, vol. 590, pp. 98-102, 2015/09/01/ 2015.

[74] Craig L Perkins et al., "Interfaces between CdTe and ALD Al2O3," in 7th World

Conference on Photovoltaic Energy Conversion, Waikoloa, HI, 2018.

[75] T. Maruyama and S. Arai, "Aluminum oxide thin films prepared by chemical

vapor deposition from aluminum acetylacetonate," Applied Physics Letters, vol.

60, no. 3, pp. 322-323, 1992.

[76] P. Ščajev et al., "Excitation-dependent carrier lifetime and diffusion length in

bulk CdTe determined by time-resolved optical pump-probe techniques," Journal

of Applied Physics, vol. 123, no. 2, p. 025704, 2018.

93 [77] D. L. Bätzner, R. Wendt, A. Romeo, H. Zogg, and A. N. Tiwari, "A study of the

back contacts on CdTe/CdS solar cells," Thin Solid Films, vol. 361–362, pp. 463-

467, 2/21/ 2000.

[78] J. Pan, M. Gloeckler, and J. R. Sites, "Hole current impedance and electron

current enhancement by back-contact barriers in CdTe thin film solar cells,"

Journal of Applied Physics, vol. 100, no. 12, p. 124505, 2006.

[79] G. S. Khripunov, "Effect of a rear contact on the electrical properties of the

CdS/CdTe-based thin-film solar cells," Semiconductors, journal article vol. 40,

no. 1, pp. 113-117, January 01 2006.

[80] T. A. Gessert et al., "Evolution of CdS/CdTe device performance during Cu

diffusion," in Conference Record of the Thirty-first IEEE Photovoltaic Specialists

Conference, 2005., 2005, pp. 291-294.

[81] M. A. Green et al., "Solar cell efficiency tables (version 49)," Progress in

Photovoltaics: Research and Applications, vol. 25, no. 1, pp. 3-13, 2017.

[82] F. Solar,

"http://files.shareholder.com/downloads/FSLR/1835294976x0x884412/1548B782

-59A0-4544-A452-989E1FA42BFE/FS_AnalystDay_ManufacturingUpdate.pdf,"

2014.

[83] J. P. Mailoa, M. Lee, I. M. Peters, T. Buonassisi, A. Panchula, and D. N. Weiss,

"Energy-yield prediction for II-VI-based thin-film tandem solar cells," Energy &

Environmental Science, 10.1039/C6EE01778A vol. 9, no. 8, pp. 2644-2653, 2016.

94 [84] T. J. Coutts, K. A. Emery, and J. Scott Ward, "Modeled performance of

polycrystalline thin-film tandem solar cells," Progress in Photovoltaics: Research

and Applications, vol. 10, no. 3, pp. 195-203, 2002.

[85] A. Rohatgi, S. A. Ringel, R. Sudharsanan, P. V. Meyers, C. H. Liu, and V.

Ramanathan, "Investigation of polycrystalline CdZnTe, CdMnTe, and CdTe films

for photovoltaic applications," Solar Cells, vol. 27, no. 1, pp. 219-230,

1989/10/01/ 1989.

[86] C. S. Ferekides, R. Mamazza, U. Balasubramanian, and D. L. Morel,

"Cd1−XZnXTe thin films and junctions," Thin Solid Films, vol. 480-481, pp. 471-

476, 2005/06/01/ 2005.

[87] S. Hyun Lee, A. Gupta, S. Wang, A. D. Compaan, and B. E. McCandless,

"Sputtered Cd1−xZnxTe films for top junctions in tandem solar cells," Solar

Energy Materials and Solar Cells, vol. 86, no. 4, pp. 551-563, 4/1/ 2005.

[88] T. Song, A. Kanevce, and J. R. Sites, "Emitter/absorber interface of CdTe solar

cells," Journal of Applied Physics, vol. 119, no. 23, p. 233104, 2016. Journal of

Applied Physics

[89] T. M. Shimpi et al., "Effect of the cadmium chloride treatment on RF sputtered

Cd0.6Zn0.4Te films for application in multijunction solar cells," Journal of

Vacuum Science & Technology A: Vacuum, Surfaces, and Films, vol. 34, no. 5, p.

051202, 2016.

[90] H. Kim and D. Kim, "Influence of CdS heat treatment on the microstructure of

CdS and the performance of CdS/CdTe solar cells," Solar Energy Materials and

Solar Cells, vol. 67, no. 1, pp. 297-304, 2001/03/01/ 2001.

95 [91] R. Dhere, T. Gessert, J. Zhou, S. Asher, J. Pankow, and H. Moutinho,

"Investigation of CdZnTe for Thin-Film Tandem Solar Cell Applications," MRS

Proceedings, vol. 763, p. B8.25, 2003, Art. no. B8.25.

[92] D. Mohanty, P.-Y. Su, G.-C. Wang, T.-M. Lu, and I. B. Bhat, "Effect of CdCl2

heat treatment on ZnTe back electron reflector layer in thin film CdTe solar

cells," Solar Energy, vol. 135, pp. 209-214, 2016/10/01/ 2016.

96 Appendix A

List of Publications

1. F. K. Alfadhili, A. B. Phillips, G. K. Liyanage, J. M. Gibbs, M. K. Jamarkattel, and

M. J. Heben, "Controlling Band Alignment at the Back Interface of Cadmium Telluride

Solar Cells using ZnTe and Te Buffer Layers," MRS Advances, vol. 4, no. 16, pp. 913-

919, 2019.

2. F. K. Alfadhili, A. B. Phillips, M. K. Jamarkattel, J. M. Gibbs, G. K. Liyanage, and

M. J. Heben, “Potential of CdZnTe Thin Film Back Buffer Layer for CdTe Solar Cells”

in 2019 IEEE 46th Photovoltaic Specialist Conference (PVSC) Chicago LI, 2019.

3. F. K. Alfadhili, G. K. Liyanage, A. B. Phillips, and M. J. Heben, "Development of

CdCl2 Activation to Minimize Zn Loss from Sputtered Cd1-xZnxTe Thin Films for

Use in Tandem Solar Cells," MRS Advances, vol. 3, no. 52, pp. 3129-3134, 2018.

4. F. K. Alfadhili ,J. M. Gibbs, G. K. Liyanage, P. W. Krantz, S. C. Watthage, Z. Song,

A. B. Phillips, and M. J. Heben "Use of Single Wall Carbon Nanotube films doped with

Triethyloxonium Hexachlorantimonate as a Transparent Back Contact for CdTe Solar

Cells," in 2017 IEEE 44th Photovoltaic Specialist Conference (PVSC), pp. 730-734,

2017.

97 5. F. K. Alfadhili, A. B. Phillips, M. K. Jamarkattel K. K. Subedi, G. K. Liyanage,

R. J. Ellingson⁠, and M. J. Heben, “Solution‐Processed Aluminum Oxide (Al2O3)

Layer to Passivate the Rear Surface of CdTe Solar Cells (In-progress).

6. E. Bastola, F. K. Alfadhili, A. B. Phillips, M. J. Heben, and R. J. Ellingson, "Wet

chemical etching of cadmium telluride photovoltaics for enhanced open-circuit

voltage, fill factor, and power conversion efficiency," Journal of Materials

Research, vol. 34, no. 24, pp. 3988-3997, 2019.

7. A. B. Phillips, G. K. Liyanage, F. K. Alfadhili, and M. J. Heben, “Understanding

the CdTe Device Performance as a Function of Band Alignment at the Front and

Back Interfaces” Proceedings of 46 th PVSC Chicago LI, 2019.

8. G. K. Liyanage, A. B. Phillips, F. K. Alfadhili, R. J. Ellingson, and M. J. Heben,

"The Role of Back Buffer Layers and Absorber Properties for >25% Efficient CdTe

Solar Cells," ACS Applied Energy Materials, 2019.

9. G. K Liyanage, A. B Phillips, F. K Alfadhili, Z.haoning Song, K. P. Bhandari, R.

J Ellingson, M. J Heben, “Modeling the Performance of CdTe Solar Cells with a

th CH3NH3Pb(I1-xBrx)3-like Back Buffer Layer”, Proceedings of 7 WCPEC, 2018.

10. G. K. Liyanage, A. B. Phillips, F. K. Alfadhili, and M. J. Heben, "Numerical

Modelling of Front Contact Alignment for High Efficiency Cd1-xZnxTe and Cd1-

xMgxTe Solar Cells for Tandem Devices," MRS Advances, vol. 3, no. 52, pp. 3121-

3128, 2018.

11. 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

98 hybrid sputtering/co-evaporation as a hole transport layer for sputtered CdS/CdTe

solar cells," Solar Energy Materials and Solar Cells 163, 277-284 , 2017.

12. S. C. Watthage, G. K. Liyanage, Z. Song, F. K. Alfadhili, R. B. Alkhayat, K. P.

Bhandari, R. J. Ellingson, A. B. Phillips, and M. J. Heben., "Novel, Facile Back

Surface Treatment for CdTe Solar Cells," in 2017 IEEE 44th Photovoltaic

Specialist Conference (PVSC), pp. 815-819, 2017.

13. Z. S. Almutawah, S. C. Watthage, R. H. Ahangharnejhad, F.K. Alfadhili, G. K.

Liyanage, N. Shrestha, A. B. Phillips, R. Ellingson and M. J. Heben, “Enhanced

Grain Size and Crystallinity in CH3NH3PbI3 Perovskite Films by Metal Additives

to the Single-Step Solution Fabrication Process.” MRS Advances,1-6, 2018.

14. S. C. Watthage, A. B. Phillips, G. K. Liyanage, Z. Song, J. M. Gibbs, F. K.

Alfadhili, R. B. Alkhayat, R. H. Ahangharnejhad, Z. S. Almutawah, K. P.

Bhandari, R. J. Ellingson, M. J. Heben, “Selective Cd Removal From CdTe for

High-Efficiency Te Back-Contact Formation.” IEEE J. Photovoltaics, 8 (4), 1125-

1131, 2018.

15. Awni, R. A., Li, D. , Grice, C. R., Song, Z. , Razooqi, M. A., Phillips, A. B., Bista,

S. S., Roland, P. J., Alfadhili, F. K., Ellingson, R. J., Heben, M. J., Li, J. V. and

Yan, Y. , The Effects of Hydrogen Iodide Back Surface Treatment on CdTe Solar

Cells. Sol. RRL, 2019