A Dissertation Entitled Solution Processed High Efficiency Thin Film

A Dissertation

entitled

Solution Processed High Efficiency Thin Film Solar Cells: from Copper Indium

Chalcogenides to Methylammonium Lead Halides

Zhaoning Song

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. Robert W. Collins, Committee Member

______Dr. Randy J. Ellingson, Committee Member

______Dr. Terry P. Bigioni, Committee Member

______Dr. Lawrence S. Anderson-Huang, Committee Member

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

The University of Toledo

August 2016

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

Solution Processed High Efficiency Thin Film Solar Cells: from Copper Indium Chalcogenides to Methylammonium Lead Halides by

Zhaoning Song

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

The University of Toledo

August 2016

Photovoltaics (PV) is increasingly recognized as an important component of renewable energy sources after the rapid progress in the last decade due to increasing energy demand and reducing manufacturing costs. Despite the enormous growth of the PV market, the present solar technologies that are dominated by crystalline silicon are still limited by the relatively more expensive cost of electricity compared with power generation in the conventional fossil fuel plants. Consequently, there is an urgent need to increase the performance and reduce the manufacturing costs of solar cells. While the commercial thin film solar cells (CdTe and CuInGaSe2) have already demonstrated high efficiencies, the current fabrication processes heavily rely on intensive capital investment on expensive vacuum-based techniques. To reduce solar module costs, solution-processing techniques have been proposed as a promising route towards low cost, high throughput, large scale manufacturing of high efficiency thin film solar cells. In this thesis, we investigate the solution-processing of copper indium chalcogenides and methylammonium lead halides materials and their applications as high efficiency photovoltaic cells.

iii

In the first approach, we develop an ultrasonic spray deposition system to prepare the

CuIn(S,Se)2 thin films. Spray deposition is a controllable, scalable, and high throughput process that is suitable for industrial manufacturing. Here we first explore the Cu-In-S films prepared by an aqueous precursor ink. By controlling the precursor composition, we fabricate PV devices consisting of the n-type In2S3 window and p-type CuInS2 absorber layers and demonstrate 2% efficiency in the preliminary devices. After replacing the aqueous ink by a hydrazine-based precursor solution and incorporating a selenization process, we are able to fabricate high quality CuIn(S,Se)2 thin film solar cells in both conventional substrate and the backwall superstrate configurations. The efficiency of 7.2% has been achieved in the sprayed CuIn(S,Se)2 devices in the substrate configuration.

In the second approach, we investigate solution-processing of the inorganic-organic hybrid metal halide perovskites. We study the impact of reaction temperature and precursor composition on the formation of perovskite materials and propose a pseudo binary phase diagram to guide the processing of the materials. We develop a laser beam induced current (LBIC) technique to spatially resolve the photocurrent collection in the solution-processed devices. Processing defects and impurities phases have been identified as the origins of lower current generation. On the basis of these results, we apply advanced processing techniques in device processing and obtain the champion perovskite device with a 16% efficiency. Additionally, we image the photocurrent generated in the sub-cells of the Si/perovskite tandem devices. The result can be used to improve the design the device structure. Finally, to study the stability of perovskite solar cells, we investigate the spatial and temporal evolution of photocurrent collection across the devices and observe the partially reversible phase transition of perovskite in humid air.

I dedicate my dissertation to my wife Yachun Zhang and my parents. Their love gives me the strength to advance and makes it all worthwhile.

Acknowledgements

First, I would like to express my sincere gratitude to my advisor Dr. Michael Heben for his wise advice and guidance during my research and study at The University of

Toledo. I especially want to thank him for giving me the freedom to pursue my research interests and supporting me throughout the years. I would also like to thank Dr. Robert

Collins, Dr. Randy Ellingson, Dr. Terry Bigioni, and Dr. Lawrence Anderson for serving on my committee and providing valuable feedback and suggestions.

I am grateful to Dr. Adam Phillips for his countless efforts and contributions to my research. I also appreciate the help from our research group members and alumni, the faculty members, staff, and fellows of the Department of Physics and Astronomy and the

Wright Center for Photovoltaics Innovation and Commercialization (PVIC).

Additionally, I would also like to thank our collaborators, Dr. Antonio Abate, Dr. Ullrich

Steiner, Jérémie Werner, Dr. Bjöern Niesen, Dr. Stefaan De Wolf, Dr. Yanfa Yan, Ilke

Celik, Dr. Defne Apul, Dr. Paul Roland, Corey Grice, and Niraj Shrestha for making contributions to my research work.

Finally, my special thanks also go to my friends Yao Xie, Xinxuan Tan, Yue Yu,

Chuanxiao Xiao, Dr. Tingting Shi, Dr. Nanke Jiang, Dr. Zhu Wang, Dr. Rui Yang, Dr.

Changyong Chen, and Dr. Zhi Ren whose friendships have sustained me.

Table of Contents

Abstract ...... iii

Acknowledgements ...... v

Table of Contents ...... vi

List of Tables ...... xi

List of Figures ...... xii

List of Abbreviations ...... xx

List of Symbols ...... xxii

1 Introduction …...... 1

1.1 Background and Motivation ...... 1

1.2 Focus of the Dissertation ...... 7

1.3 Dissertation Overview ...... 9

2 Device Fabrication and Characterization Methods ...... 13

2.1 Materials Preparation and Device Fabrication ...... 13

2.1.1 Ultrasonic Spray Deposition ...... 13

2.1.2 Fabrication of CuInSe2 Based Solar Cells ...... 15

2.1.3 Fabrication of CH3NH3PbI3 Based Solar Cells ...... 17

2.2 Materials and Device Characterization ...... 18

2.2.1 Material Characterization Methods ...... 18

2.2.2 Device Characterization Methods ...... 18

2.2.3 Laser Beam Induced Current (LBIC) ...... 22 vi

3 Spray Pyrolysis of Copper Indium Sulfide Thin Film Solar Cells ...... 26

3.1 Introduction and Motivation ...... 26

3.1.1 History of CuInS2 Based Solar Cells...... 26

3.1.2 Structural Properties of CuInS2 Based Materials ...... 28

3.2 Cu-In-S Thin Films Prepared by Spray Pyrolysis ...... 31

3.2.1 Mechanism of Spray Pyrolysis of Cu-In-S Thin Films ...... 31

3.2.2 Spray Pyrolysis of Cu2-xS Thin Films ...... 33

3.2.3 Spray Pyrolysis of CuInS2 and In2S3 thin films ...... 35

3.3 Spray Pyrolysis of CuInS2/In2S3 Thin Film Solar Cells ...... 40

3.4 Factors Limited the Performance of the CuInS2 Solar Cells ...... 44

3.5 Conclusion ...... 46

4 Hydrazine Based Copper Indium Selenide Thin Film Solar Cells ...... 48

4.1 Introduction and Motivation ...... 48

4.2 Hydrazine-Based CuIn(S, Se)2 Prepared by Spray Deposition ...... 50

4.2.1 Precursor Solution Preparation ...... 50

4.2.2 Crystallinity of the Sprayed Films ...... 53

4.2.3 Morphology and Composition of the Sprayed Films ...... 56

4.2.4 Device Performance of the Sprayed CuInSe2 Solar Cells ...... 60

4.2.5 Selenization of the Sprayed CuInSe2 Solar Cells ...... 62

4.3 Sprayed CuInSe2 Solar Cells in the Backwall Superstrate Configuration ...... 66

4.3.1 Fabrication of Semi-Transparent Backwall Superstrate Devices .....67

4.3.2 Band Alignment of the Backwall Superstrate Devices ...... 68

4.3.3 Microstructure and Composition of the Devices ...... 70

vii

4.3.4 Device Performance of the Sprayed Backwall Superstrate CuInSe2

Cells… ...... 71

4.4 Summary and Outlook of the Sprayed CuInSe2 Solar Cells ...... 77

5 Emergence of Organic-Inorganic Lead Halide Perovskites: Literature Review ...79

5.1 Background of Perovskite Solar Cells ...... 80

5.2 Evolution of Device Architectures...... 83

5.2.1 Conventional “n-i-p” Structure ...... 84

5.2.2 Inverted “p-i-n” Structure ...... 87

5.3 Preparation Methods Evolution of Device Architectures ...... 88

5.3.1 Single-Step Solution Deposition ...... 88

5.3.2 Two-Step Solution Deposition ...... 90

5.3.3 Vapor-Assisted Solution Deposition...... 91

5.3.4 Thermal Vapor Deposition ...... 92

5.4 Advanced Engineering Techniques ...... 92

5.4.1 Solvent Engineering ...... 93

5.4.2 Process Engineering ...... 94

5.4.3 Composition Engineering ...... 97

5.4.4 Contact Engineering...... 97

5.5 Efficiency Improvement in Recent Years ...... 100

5.6 Issues and Challenges ...... 102

5.6.1 Device Stability ...... 103

5.6.2 Current-Voltage Hysteresis ...... 104

5.6.3 Toxicity and Pollution...... 107

viii

5.7 Conclusion ...... 108

6 Formation of the Perovskites: a Processing Phase Diagram Study on

Methylammonium Lead Iodide ...... 110

6.1 Introduction and Motivation ...... 111

6.2 Perovskite Sample Preparation ...... 113

6.3 Structural and Optical Properties of Perovskite Films ...... 114

6.3.1 Pure Phases of Perovskite Related Films ...... 114

6.3.2 Films with MAI-Poor Compositions ...... 118

6.3.3 Films with MAI-Rich Compositions ...... 120

6.4 A Pseudo-Binary Phase Diagram of CH3NH3PbI3 ...... 127

6.5 Conclusion ...... 133

7 Spatially-Resolved Current Generation Measurement of Solution-Processed

Perovskite Solar Cells ...... 134

7.1 Introduction and Motivation ...... 134

7.2 Perovskite Devices Prepared by Different Methods ...... 135

7.2.1 Perovskite Devices Preparation ...... 135

7.2.2 Perovskite Device Performance ...... 137

7.2.3 Morphology and Photocurrent Uniformity ...... 139

7.2.4 Current Loss Mechanisms...... 142

7.3 Perovskite Devices with Improved Performance ...... 145

7.4 Measurements on Perovskite/Si Tandem Cells...... 148

7.4.1 Purpose of the Work ...... 148

7.4.2 LBIC Measurement Results of Perovskite/Si Tandem Cells ...... 150

7.5 Conclusion ...... 156

8 Stability of Perovskite Solar Cells in Humid Air...... 157

8.1 Introduction and Motivation ...... 158

8.2 Experimental Details ...... 159

8.2.1 Perovskite Devices Fabrication ...... 159

8.2.2 In-situ LBIC Measurement ...... 161

8.3 Degradation Behaviors of Perovskite Solar Cells ...... 162

8.3.1 Four-Stage Process of Degradation ...... 162

8.3.2 Reversible Hydration Process ...... 171

8.4 Equilibria in the Ternary PbI2-CH3NH3I-H2O System ...... 174

8.5 Impacts of Charge Transporting Materials and Perovskite Compositions ....176

8.6 Approaches to Improve Devices Stability ...... 183

8.7 Conclusion ...... 185

9 Summary and Future Research ...... 187

9.1 Thesis Summary...... 187

9.2 Future Research ...... 188

9.2.1 Environmental and Economic Impacts ...... 188

9.2.2 Improving the stability of perovskite solar cells ...... 189

9.2.3 Solution-processed solar cells on flexible substrates ...... 190

9.2.4 Perovskite/CuInSe2 Tandem Solar Cells ...... 191

9.2.5 Fabrication of Perovskite Mini-Modules ...... 192

References ...... 193

A Publication List ...... 222

List of Tables

3.1 Cu2-xS thin films prepared by spray pyrolysis ...... 34

3.2 Summary of the sprayed solar cell performance ...... 43

4.1 Performance of CuInSe2 solar cells prepared by spray pyrolysis ...... 61

7.1 Device performance of perovskite solar cells ...... 139

7.2 Statistics of the LBIC measurement ...... 142

List of Figures

1-1 Electricity generation by fuel in the US ...... 2

1-2 Best efficiency chart of solution-processed emerging solar cells ...... 4

1-3 Solar cell performance triangle of efficiency, cost, and stability ...... 5

1-4 Solution-based deposition methods ...... 7

2-1 Schematic and photo of the built-in-house ultrasonic spray system ...... 14

2-2 Structure of a standard CuInSe2 solar cell in the substrate configuration ...... 15

2-3 Photo of an ongoing CdS chemical bath deposition ...... 16

2-4 Structure of CH3NH3PbI3 solar cell in the “n-i-p” superstrate configuration ...... 17

2-5 J-V curves and the equivalent circuit of a general photovoltaic device ...... 19

2-6 In-situ LBIC measurement setup ...... 23

2-7 Average laser power of the 532 nm laser ...... 23

2-8 Optical image, EQE curve, LBIC image, and LBIC histogram of the reference Si

photodiode ...... 24

2-9 Environmental chambers for in-situ LBIC and optical absorption measurements

...... 25

3-1 Chalcopyrite structure of CuInS2 ...... 28

3-2 Polytypes of chalcopyrite structure of CuInS2 ...... 29

3-3 Ternary phase diagram of Cu-In-S ...... 30

3-4 Schematic of the CuInS2 spray pyrolysis mechanism ...... 33 xii

3-5 Hole concentration and mobility of sprayed Cu2-xS thin films ...... 35

3-6 Optical transmittance and absorbance spectra of CuInS2 and In2S3 thin films .....36

3-7 XRD spectra of sprayed In2S3, CuInS2, and In2S3/CuInS2 thin films ...... 38

3-8 SEM morphological images of sprayed In2S3 and CuInS2 ...... 39

3-9 Fabrication process of sprayed CuInS2/In2S3 solar cells ...... 40

3-10 Cross-sectional SEM image of a CuInS2/In2S3/FTO device ...... 41

3-11 J-V curves and resistivity of the sprayed CuxIn1-xS2 device ...... 42

3-12 Optical band gaps of the sprayed Cu-In-S thin films derived from the UV-VIS

absorption spectra ...... 43

3-13 Band diagrams for the sprayed CuInS2/In2S3 solar cell ...... 44

3-14 J-V curves of a CuInS2 solar cell with respect to temperature ...... 45

4-1 Photos of precursor solutions involving with aqueous, anhydrous hydrazine, and

hydrazine hydrate as the solvent ...... 51

4-2 FTIR-Raman spectra of precursor solutions involving with aqueous, anhydrous

hydrazine, and hydrazine hydrate as the solvent ...... 52

4-3 XRD spectra of sprayed CuInSe2 thin films as various temperatures using the

aqueous, anhydrous hydrazine, and hydrazine hydrate solution ...... 54

4-4 The (112) direction texture coefficients and the crystallite sizes calculated from

XRD spectra for sprayed CuInSe2 thin films ...... 55

4-5 Schematics of thin films prepared by aqueous and hydrazine-based solutions ....56

4-6 Top view and cross-sectional SEM images of CuInSe2 films obtained by spray

pyrolysis using aqueous, anhydrous hydrazine and hydrazine hydrate solutions ..57

xiii

4-7 EDS spectra of CuInSe2 films obtained by spray pyrolysis using aqueous,

anhydrous hydrazine and hydrazine hydrate solutions ...... 58

4-8 EDS scan of the non-stoichiometric Cu2-xS defect in the CuInSe2 films obtained

by spray pyrolysis of the aqueous precursor solution ...... 59

4-9 J-V curves of sprayed CuInSe2 solar cells based on aqueous, anhydrous hydrazine

and hydrazine hydrate solutions ...... 61

4-10 SEM cross-sectional images of the as-sprayed CuInSe2 thin films and that after

selenization at 300, 350, 400, 450, 500, and 550 °C for 30 min ...... 64

4-11 SEM images of the sprayed CuInSe2 film after the 550 °C selenization ...... 65

4-12 J-V and EQE curves of the best sprayed CuInSe2 solar cells ...... 65

4-13 Schematic diagrams of sprayed CuInSe2 solar cells in the standard substrate and

(b) backwall superstrate configurations ...... 68

4-14 Energy band diagrams of substrate and backwall superstrate types CuInSe2 solar

cells at thermal equilibrium ...... 69

4-15 Cross-sectional SEM image and EDS compositional mapping of a typical

backwall superstrate device with a 600 nm sprayed CIS absorber ...... 71

4-16 J-V characteristics of sprayed CuInSe2 cells in the substrate and the backwall

superstrate configurations ...... 72

4-17 J-V curves of the best sprayed CuInSe2 cells in the substrate and the backwall

superstrate configurations ...... 74

4-18 Internal Quantum efficiency of the superstrate devices and control devices for the

CuInSe2 thickness of 1.2 μm, 0.9 μm, and 0.6 μm ...... 76

5-1 Crystal structure and photo of CH3NH3PbI3 perovskite crystal ...... 81

xiv

5-2 Schematic diagrams of perovskite solar cells in the n-i-p mesoscopic, n-i-p planar,

p-i-n planar, and p-i-n mesoscopic structures ...... 84

5-3 Deposition methods for perovskite thin films, including (a) single-step solution

deposition, (b) two-step solution deposition, (c) two-step hybrid deposition, and

(d) thermal vapor deposition ...... 89

5-4 SEM images of perovskite films prepared using various deposition techniques and

advanced engineering processes ...... 96

5-5 Diagram showing the energy levels, from left to right, for representative cathode,

n-type (ETM), absorber, p-type (HTM), and anode materials ...... 98

5-6 Efficiency mapping of recently reported the state-of-the-art perovskite solar cells

labeled with reference number and colored based on efficiency...... 102

5-7 Origins of J-V hysteresis in perovskite solar cells ...... 106

6-1 (a) XRD spectra of pure phases of PbI2, MAPbI3 and MAI. (b) XRD scan of the

tetragonal MAPbI3 phase. (c) MAPbI3 α and β phases ...... 115

6-2 OA spectra of pure phases of spin-coated PbI2, MAPbI3 and MAI ...... 117

6-3 Optical absorbance spectra and (b) contour plot of the room-temperature XRD

mapping of the spin-coated MAPbI3 ...... 117

6-4 Grain size in the MAPbI3 films as a function of reaction temperature ...... 118

6-5 XRD contour plots of the spin-coated perovskite composites with MAI percentage

(XMAI) of 0.09, 0.15, 0.23, and 0.38 ...... 119

6-6 OA spectra of the spin-coated perovskite composites with MAI percentage (XMAI)

of 0.09, 0.23, and 0.38 ...... 120

6-7 XRD contour plots of the spin-coated non-stoichiometric MAPbI3 composites

with MAI percentage (XMAI) of 0.58, 0.67, 0.75, and 0.80 ...... 122

6-8 OA spectra of the spin-coated non-stoichiometric MAPbI3 composites with MAI

percentage (XMAI) of 0.58, 0.67, 0.75, and 0.80 ...... 122

o 6-9 (a) TGA of MAI and MAPbI3 in N2. (b) TGA of MAI at 150 and 190 C. (c) DSC

plot of the MAI phase transition at around 150 oC ...... 123

6-10 Crystal structures of PbI2, MAPbI3 perovskites, 2D MA2PbI4 sheets, 1D MA3PbI5

chains, and 0D MA4PbI6 blocks ...... 124

6-11 (a) Measured and simulated XRD spectra of the 0D perovskite phase. (b) Crystal

structure of the 0D perovskite ...... 126

6-12 Compositional analysis of thin films prepared from various precursor

compositions at RT, 100 oC, and 150 oC ...... 128

6-13 (a) Pseudo-binary temperature/phase processing diagram for methylammonium

lead iodide. (b) Phase percentages of PbI2, perovskite, and LDP as a function of

MAI precursor concentration in the films reacted at 100 oC ...... 129

6-14 XRD contour plots of high temperature processed perovskites (XMAI = 0.75)

reacted on a hot plate and in a sealed graphite box ...... 130

6-15 SEM images of perovskite films (a) reacted at 100 oC on a hot plate and reacted at

150 oC in a sealed graphite box ...... 131

6-16 Proposed phase diagram constructed with high temperature reactions carried out

under a saturated MAI vapor pressure ...... 131

7-1 (a) cross-sectional SEM image and (b) energy level diagram of a perovskite

device with a structure of FTO/c-TiO2/mp-TiO2/MAPbI3/P3HT/Au ...... 137

xvi

7-2 (a) J-V characteristic curves and (b) EQE spectra of perovskite devices prepared

by different solution-based processes ...... 138

7-3 LBIC maps of perovskite devices prepared by (a) single-step spin, (b) single-step

spray, (c) sequential spins, and (d) sequential spray/spin methods ...... 140

7-4 SEM maps of perovskite films prepared by (a) single-step spin, (b) single-step

spray, (c) sequential spins, and (d) sequential spray/spin methods ...... 141

7-5 Histograms of the LBIC signals for perovskite devices prepared by different

methods ...... 142

7-6 (a) LBIC images of a spray deposited perovskite device before and after laser

scribing. SEM images of (b) the laser scribed line, (c) the low current spot “+”,

and (d) the high current spot “*” ...... 143

7-7 LBIC and SEM maps of perovskite devices prepared using a 40 mg/ml and a 100

mg/ml MAI solution ...... 145

7-8 Optical images of perovskite thin films prepared by the single-step methods using

conventional and advanced approaches ...... 146

7-9 Cross-sectional SEM image of the best performing perovskite solar cell ...... 147

7-10 J-V and EQE of the champion perovskite solar cell prepared by the advanced

single-step process ...... 147

7-11 J-V curves and schematic of the device structure of the perovskite/Si tandem solar

cells fabricated at PV-center, EPFL ...... 149

7-12 EQE spectra of two sub-cells taken with selective light biasing of a perovskite/Si

tandem device ...... 151

xvii

7-13 LBIC maps of a 1.22 cm2 perovskite/Si cell probed by 532 nm laser with red or

blue light bias ...... 152

7-14 LBIC maps of a 1.22 cm2 perovskite/Si cell probed by 1064 nm laser with blue

light bias ...... 153

7-15 LBIC maps and the line profiles of single-side textured and double-side polished

silicon cells ...... 154

7-16 LBIC maps of a 0.17 cm2 perovskite/Si cell probed by 532 nm laser with red or

blue light bias ...... 155

7-17 Histograms of the LBIC of the top and bottom cells of a large size and a small

region of perovskite/Si devices measured using the 532 nm laser ...... 155

8-1 EQE maps (at 532 nm) of a typical perovskite solar cell after exposure to 50 ± 5

% RH ...... 163

8-2 Areal average LBIC EQE (at 532 nm) as a function of time after exposure to

humidity ...... 164

8-3 Transmittance and ATR FTIR spectra of the stacked films of spiro-OMeTAD

/CH3NH3PbI3/TiO2/FTO after exposure to 50 % RH ...... 167

8-4 LBIC maps of perovskite devices aging under moist air or N2 flow of 80 RH ...... 167

8-5 Contour plots of the degradation front propagation of perovskite devices aged under (a)

50% RH and (b) 80% RH during Stage 3 ...... 168

8-6 (a) Photo of a perovskite solar cell and microscopic optical image of the scribing line

lose to the anode. (b-e) Schematics of water ingress into a perovskite solar cell ...... 169

8-7 J-V curves of a perovskite device aging under 80 % RH ...... 170

xviii

8-8 LBIC EQE maps of a typical perovskite solar cell after exposure to a moist N2

flow of 80 ± 5 % RH ...... 172

8-9 LBIC EQE maps of hydrated perovskite device while purging with dry air ...... 172

8-10 Areal average EQE (at 532 nm) of the perovskite device during hydration-

dehydration cycles ...... 173

8-11 Schematic diagram of the phase equilibria in the PbI2-CH3NH3I-H2O system ..174

8-12 Optical absorbance spectra of (a) FTO/TiO2/CH3NH3PbI3 and (b) FTO/TiO2/

CH3NH3PbI3/Spiro-OMeTAD aged under RH = 80% ...... 177

8-13 XRD spectra of (a) FTO/TiO2/CH3NH3PbI3 and (b) FTO/TiO2/CH3NH3PbI3/

Spiro-OMeTAD aged under RH = 80% for 30 min ...... 179

8-14 Evolution of LBIC maps of I/PTAA and I/Spiro devices after exposure to 80%

RH ...... 180

8-15 Evolution of LBIC maps of Br/PTAA and Br/Spiro devices after exposure to 80%

RH ...... 181

8-16 SEM image of aged (a) Br/PTAA and (b) Br/Spiro devices ...... 182

8-17 Degradation of integrated LBIC signal of the perovskite devices as a function of

exposure time ...... 183

8-18 LBIC EQE maps of a typical perovskite solar cell after exposure to dry air ...... 184

8-19 (a) photo of SWCNT/perovskite film stack. (b) Degradation behaviors of

perovskite/P3HT and perovskite/SWCNT in humid air ...... 185

xix

List of Abbreviations

CIGS ...... Copper Indium Gallium Selenide/Sulfide CZTS ...... Copper Zinc Tin Selenide/Sulfide

DSSC...... Dye-Sensitized Solar Cells DMF ...... Dimethylformamide DMSO ...... Dimethylsulfoxide

EDS ...... Energy Dispersive Spectroscopy EPFL ...... École Polytechnique Fédérale de Lausanne EPBT ...... Energy Pay Back Time EQE ...... External Quantum Efficiency ETM ...... Electron Transport Materials

FF ...... Fill Factor FTIR ...... Fourier Transform Infrared Spectroscopy FWHM ...... Full Width at Half maximum

HTM ...... Hole Transport Materials

LBIC ...... Laser Beam Induced Current LCA...... Life Cycle Assessment LCOE ...... Levelized Cost of Electricity LDP ...... Low Dimensional Perovskite Li-TSFI ...... Lithiumbis(trifluoromethanesulfonyl)imide

MAI ...... Methyl-Ammonium Iodide MMP ...... Maximum Power Point

OA ...... Optical Absorbance OPV...... Organic Photovoltaics

P3HT ...... Poly(3-hexylthiophene-2,5-diyl) PCE ...... Power Conversion Efficiency PEDOT ...... Poly(3,4-ethylenedioxythiophene) PSC ...... Perovskite Solar Cells PSS ...... Poly(styrene-sulfonate) PTAA ...... Poly(triarylamine) PV ...... Photovoltaics

QDC ...... Quantum Dots Solar Cells

RH ...... Relative Humidity

SCFH...... Standard Cubic Feet per Hour SEM ...... Scanning Electron Microscopy SLG ...... Soda Lime Glass SWCNT...... Single Wall Carbon Nanotube

TCO...... Transparent Conductive oxide TGA ...... Thermo-Gravimetric Analysis

XRD ...... X-Ray Diffraction

xxi

List of Symbols

α ...... Texture coefficient β ...... Full width at Half Maximum ε ...... Lattice strain η ...... Device efficiency θ ...... Angle λ ...... Photon wavelength σ ...... Conductivity τ ...... Average crystalline size Eg ...... Band gap FF ...... Fill factor hlk ...... index for crystal plane I ...... Current or intensity J0 ...... Saturation current density Jd ...... Dark current density JL ...... Light current density JSC ...... Short circuit current density k...... Boltzmann’s constant n...... Diode ideality factor Δn ...... Excess concentration of electrons NA ...... Acceptor concentration ni ...... Intrinsic carrier density Pin ...... Input power of illumination Δp ...... Excess concentration of holes q...... Electron charge RS ...... Series resistance RSH ...... Shunt resistance S ...... Incident photon flux density T ...... Temperature V ...... Applied voltage VOC ...... Open circuit voltage

xxii

Chapter 1

Introduction

1.1 Background and Motivation

With rapid industrial development and population growth, worldwide energy demand has been continuously increasing ever since the Industrial Revolution in the late 17th and early 18th centuries. Over the past two hundred years, large quantities of fossil fuels, such as petroleum, coal, and natural gas have been used as the primary energy source to power economic growth and the development of civilization. The vast consumption of fossil fuels over a short amount of time in human history has been accompanied by a large amount of greenhouse gas (e.g., CO2, CH4, and NOx) emissions. The increasing CO2 concentration forces a larger fraction of the energy from the irradiance from the Earth’s surface to remain in the Earth’s atmosphere.[1] This leads to global warming, the deterioration of the environment, the depletion of energy resources, and the destruction of ecosystems. These adverse environmental effects have raised great concerns about limiting the consumption of fossil fuels and reducing atmospheric carbon.

To establish a sustainable human civilization, it is necessary to expedite the development of environmentally-friendly alternative energy sources to meet the ever- growing energy demand. As one of the clean, sustainable energy sources that generate no 1

emissions during operation, solar photovoltaics (PV) is a means to directly convert solar radiation into electrical power. In the United States, solar (1%), together with other renewable energy sources, including hydropower (6.3%), wind (5%), biomass (1.57%), and geothermal (0.41%), contribute ~14% of total electricity generation (Figure 1-1).[2]

Although the electricity generation from solar currently contributes only a small fraction of the total electricity generation, the constantly increasing annual growth rate (~34%) shows a great potential for the large scale application of solar energy.[3] This rapidly increasing adoption of solar panels is mainly driven by the reduction of the installed costs, which has dropped by 63% since 2010. As long as the technological progress of PV goes on, this trend of decreasing solar energy cost and increasing installation capacity will continue. With such rapid progress, the percentage of electricity generation from solar is expected to grow remarkably reaching 11% in the U.S. by 2040. Thus, there is a great opportunity for various PV technologies.

Figure 1-1: Electricity generation by fuel in the U.S. Data was adapted from Annual

Energy Outlook 2016.[2] 2

The modern PV industry has a relatively short but diverse history. In 1954, Daryl

Chapin, Calvin Fuller, and Gerald Pearson developed the first practical PV device based on crystalline silicon at Bell Laboratories.[4] After several decades of progress, the PV industry has grown into a booming business, with 183 GWp global cumulative installation and 48 GW annual production at the end of 2014.[5] At present, poly- and mono-crystalline silicon (the first generation) dominate global PV market with 55% and 36% market share, respectively. Crystalline silicon (c-Si) is dominating the market due to high power conversion efficiencies (PCE, up to 25.6%),[6] robust stability for 30+ years, and low module costs. However, further reduction of c-Si PV manufacturing costs is limited by some intrinsic issues, such as high materials cost, energy-intensive processes (e.g. ingot preparation and dopant diffusion), poor materials utilization (e.g. wafer sawing), and high capital investment requirements and accompanying depreciation.

The remaining 9% of the market is split between thin film solar cells (the second generation) which include cadmium telluride (CdTe), amorphous silicon (a-Si), and copper indium gallium selenium [Cu(In,Ga)(S,Se)2 or CIGS]. In addition to the first and second generation PV technologies, single crystal III-V semiconductors (e.g., GaAs and InP) are also used for commercial PV. However, the application of III-V semiconductors is currently limited to the aerospace application due to the extremely high materials and manufacturing costs.

New solar photovoltaic technology alternatives have also emerged due to extensive research. The research focus has been prompted by a desire to find new approaches to developing lower cost, higher efficiency photovoltaics from, ideally, earth abundant and non-toxic materials. These emerging solar cells include earth-abundant inorganic solar 3

cells [e.g., Cu2ZnSn(S,Se)4, or CZTS], dye-sensitized solar cells (DSSCs), organic photovoltaic cells (OPV), quantum dots cells (QDCs), and perovskite solar cells (PSCs).

These new technologies have attracted significant attention around the world in the past few decades and some have rapidly developed in the laboratories and demonstrated potential to enter the PV market (Figure 1-2).[7]

Figure 1-2: Best efficiency chart of solution-processed emerging solar cells.

However, to gain market share from c-Si solar cells and well-established 2nd generation technologies and compete with conventional energy sources, the emerging PV technologies have to provide a desirable combination of high power conversion efficiency, low manufacturing costs, and excellent stability. This is the same tradeoff that more conventional PV technologies wrestle with, but new materials may offer new pathways to optimize. Typically, low-cost devices exhibit low efficiencies and often suffer from short

lifetimes, as well. On the contrary, the most stable, high-efficiency devices are typically costly in terms of materials or manufacturing, or both. Consequently, the routes that lead to the commercialization of new technologies have to be evaluated based on the performance triangle of efficiency, lifetime, and cost (Figure 1-3).

Figure 1-3: Solar cell performance triangle of efficiency, cost, and stability.

Commercial solar modules are overwhelmingly fabricated using vacuum-based processes. However, these capital- and energy-intensive manufacturing processes often pose economic and technological hurdles to the mass production of low-cost solar modules.

Bounded by huge capital investment, a significant number of solar companies have to be dependent on government subsidies to survive. In contrast, solution-based processes are cost-effective to manufacture solar cells and offer the potential for fabrication on flexible substrates.[8] Combining high-performance devices and solution processability may be able to lower the manufacturing costs on a per watt basis. If manufacturing costs can be lowered while maintaining high efficiencies, then a lower energy pay-back times (EPBT) 5

and a lower leveled cost of electricity (LCOE) can be achieved. Such technologies would have an edge over the competition in the marketplace.

To be more specific, solution-based deposition techniques offers several key advantages over the continued use of vacuum systems,[9] including significantly lower capital expenditure, simplified manufacturing processes, and reduced materials costs.

Solution-based processes occur at atmospheric pressure so that the facility costs associated with vacuum chambers and pumps can be reduced. Without the limited space in vacuum chambers, most of the solution-based processes are scalable and applicable to a high throughput roll-to-roll process. The adaptable manufacturing process amenable to roll-to- roll processing enables continuous and high throughput production of PV modules and diversifies the number of applications due to module flexibility. Additionally, solution- based approaches typically involve simplified processing that requires low energy input and the conservation of expensive raw materials, and thus, can reduce solar module manufacturing costs.

A variety of solution-based techniques can be employed for fabricating thin film PV devices. Figure 1-4 shows the most commonly used solution-based processes,[10] include chemical bath, spin coating, dip coating, doctor blade printing, metering rod printing, slot casting, spray coating, screen printing, inkjet printing, and aerosol printing. Among them, spray deposition, chemical bath, and spin coating are selected for the research work in this thesis due to the simplicity of the processes and applicable potential.

To date, five primary families of solution-processed materials that are suitable for solution processing have demonstrated more than 10% PCEs,[8] including dye- sensitizers,[11] organics,[12] bulk chalcogenide inorganics,[13] colloidal quantum 6

dots,[14] and inorganic-organic hybrid perovskites.[15] These solution-processed thin film materials are generally composed of nanostructured crystalline domains and abundant materials interfaces. To continue the progress toward high-efficiency solar cells, process control and chemical/materials management of the physics and optoelectronic properties of the light absorbing materials and charge transport properties within the materials and their interfaces require proper attention and optimization.

Figure 1-4: Various solution-based deposition methods. Adapted from Ref. [10] with

1.2 Focus of the Dissertation

The aim of the study is to develop a better understanding of the physics, chemistry, and materials science of solution processed materials and their interfaces, and to control the materials properties via materials chemistry and process engineering. In particular, this thesis focuses on investigating the solution-processing of copper indium chalcogenides and organometallic halide perovskites, and their photovoltaic application. 7

At the beginning of this thesis work (2010 - 2013), the most viable PV materials for solution-processing solar cells were in the CIGS family. Compared with the ~20% record efficiency of the former champion device,[16] the best solution-processed CIGS solar cell demonstrated a high PCE of 15.2%.[17] The relatively small efficiency gap between the vacuum- and solution-based CIGS solar cells demonstrates the potential of solution- processed solar cells to compete in performance with the counterpart produced by vacuum- based techniques. Consequently, we, along with numerous industrial and academic efforts, were focused on developing new techniques for solution processing of CIGS materials in the beginning of the thesis work.

The success of solution-processing for CIGS solar cells is the results of several factors.

The main reason is that the desired composition can be easily reached in the deposited materials. The problem with vacuum-based deposition of CIGS is the difficulty to achieve compositional uniformity of the quaternary compound over a large area by using directional deposition sources. In contrast, solution processing approaches provide a viable route to control stoichiometry at the molecular level in precursor solutions as a result of the nanoscale mixing of the various constituents, and thus has a potential to optimize the compositional uniformity in large area devices. Additionally, the development of pure solution technique using hydrazine as the solvent improves the solubility of metal chalcogenides in the precursor solution via dimensional reduction,[18] which allows the preparation of precursor solution with already formed metal-chalcogenide bonding and sufficiently high chalcogenide concentrations. The complete decomposition and dissociation of hydrazine from the precursor film can avoid external impurities (e.g., C, O,

N, and Cl) in final films. Last but not least, CIGS materials are forgiving in that the point 8

defects and grain boundaries are relatively benign, and form ordered defect compounds at off-stoichiometric composition.[19]

During the course of this thesis, two research groups reported PCEs of ~10% for perovskite cells in the solid-state DSSC structure in the latter part of 2012.[20, 21] Since then, perovskite-based PV device performance has rapidly progressed and the best efficiency record of over 22.1% was achieved in early 2016.[22] Such rapid progress has been remarkable and unprecedented in PV history. Surprisingly, the solution-processed perovskite cells are typically better than the vacuum deposited ones, making the perovskites a group of special materials of the interest.

The key problems in solution-processing of inorganic materials could be identified in the CIGS system, but these overcome in the organic-inorganic perovskite system. The excellent solution processability of the precursor materials in a variety of solvent systems allows the combination of different solvents to achieve the control of the desired solubility, viscosity, wettability, and evaporation rate, leading to a compact and conformal film morphology and controllable density/porosity. Additionally, compared with inorganic semiconductors, perovskite materials have much lower formation energy, making low- temperature processing possible for fabricating the devices. In addition, the materials exhibit a large degree of tolerance in synthesis with a wide range of precursor ratios, indicating a wide phase field that is compatible with solution processing.

1.3 Dissertation Overview

The thesis research has focused on studies of solution-processing of low cost, high- efficiency chalcogenide CuIn(S,Se)2 and halide CH3NH3PbI3 thin-film solar cells. 9

Solution-based processes, including ultrasonic spray, chemical bath deposition, and spin- coating have been used to prepare the thin films. The effects of processing conditions on electrical, optical, structural, and compositional properties and the related device performance have been investigated. Spatially resolved current generation measurements have been performed using laser beam induced current (LBIC) to study the non-uniformity in photocurrent generation and collection. The stability and degradation of perovskite solar cells have also been investigated. To provide a clear perspective and contributions of this thesis, a detailed outline is presented as follows.

In this chapter, Chapter 1, an introduction to the global PV market and different PV technologies was presented. A brief discussion that focused on solution-processing techniques was presented, which leads to the motivation behind the effort to developing cost-effective, solution-based processes for fabricating high-efficiency chalcogenide chalcopyrite and halide perovskite thin-film solar cells.

Chapter 2 provides the experimental methods of this work, including a thorough description of the deposition equipment utilized for materials preparation and devices fabrication with a special focus on the ultrasonic spray deposition system. The common procedures that used for the solution-based fabrication of entire CuIn(S,Se)2 and

CH3NH3PbI3 solar cells are described. In addition to the device preparation techniques, the common material and device characterization techniques are also introduced. In particular, the details of the built-in-house, high speed, large area LBIC system are presented.

Chapter 3 starts with a short introduction to the CuInS2-based materials. This is followed by a study of developing a solution-based approach for CuInS2 thin films using spray pyrolysis of aqueous precursor solutions with different chemical compositions. The 10

PV applications of the sprayed CuInS2/In2S3 film stacks are explored. The challenges and issues of this device fabrication approach are also discussed at the end of the chapter.

Chapter 4 focuses on a further development of the spray deposition methods by incorporating selenium and hydrazine into the precursor solution. Higher device efficiencies of CuIn(S,Se)2 thin film solar cells were achieved after replacing the aqueous precursor solutions by the hydrazine-based solutions. Next, the different annealing approaches, including tube furnace annealing and closed-space annealing, are presented.

Additionally, devices prepared in the back-wall superstrate configuration are studied. At the end of the chapter, a short summary is given to discuss the viability and future of the solution-processed CuInSe2 solar cells.

Chapter 5 introduces the organic-inorganic metal halide perovskites and discusses the benefits of the materials in comparison to inorganic thin films. A thorough literature review on the rapid emergence and progress of perovskite solar cells is presented in this chapter.

This review summarizes the excellent optoelectronic properties of the materials, the evolution of device architecture, the development of material deposition processes, and the advanced device engineering techniques that lead to recent improvement of perovskite solar cells.

Chapter 6 focuses on a study of the formation of methylammonium lead iodide perovskite material. The impact of processing temperature and precursor composition on the chemical phase and optical properties of perovskite film films are investigated. New phases associated with low-dimensional perovskites are identified by X-ray diffraction

(XRD), optical absorption spectroscopy, and differential scanning calorimetry

measurements. As a result of this study, a quasi-binary phase diagram was proposed to guide the processing of the perovskite materials.

Chapter 7 begins with a study of the non-uniform photocurrent generation in the perovskite solar cells processed by different solution-based processes. Laser-beam induced current technique is introduced as a powerful tool to characterize spatially resolved quantum efficiency. On the basis of this study, advanced processing techniques are applied to the perovskite device fabrication, and a 16% power conversion efficiency has been achieved. The LBIC mapping technique is also used to character semitransparent perovskite and tandem c-Si/perovskite solar cells fabricated by our collaborators.

Chapter 8 focuses on a study of the stability of perovskite solar cells under humidity.

A 4-stage degradation behavior has been observed via in-situ LBIC imaging. A quasi- ternary phase diagram is constructed to describe the system of PbI2-MAI-H2O. The impacts of different hole-transporting materials and absorber compositions are studied. Insight on the improvement of the material and device stability is provided.

Chapter 9 concludes this thesis with a summary and future research. Possible outlook in the future research directions and potential applications based on the significant results described in this thesis is provided.

Chapter 2

Device Preparation and Characterization Methods

Preparation and characterization methods for photovoltaic materials and devices are fundamentally important for investigating solar cells. This chapter introduces different kinds of experimental techniques for material preparation, device fabrication, and material/device characterization. Low-cost, solution-based deposition processes are developed to fabricate CuIn(S,Se)2 and CH3NH3PbI3 thin film solar cells which are then measured by a variety of characterization techniques to elucidate the underlying physics and chemistry of the materials and devices. Sections on the development of ultrasonic spray deposition and laser beam induced current (LBIC) systems are the focus of this chapter, while other techniques are briefly introduced.

2.1 Materials Preparation and Device Fabrication

2.1.1 Ultrasonic Spray Deposition

An aerosol ultrasonic spray deposition system (Figure 2-1) was built to fabricate semiconductor thin films.[23-25] To allow the use of air- and moisture-sensitive precursors and isolate toxic materials from operators, the spray system was enclosed in a nitrogen- filled plastic glove box that was connected to a gas circulator and purifier. The deposition process was automated using LabView software and circuitry developed in-house. 13

Figure 2-1: (a) Schematic diagram and (b) photo of the built-in-house ultrasonic spray

system.

A typical spray deposition process can be described as follows. A precursor solution of the material of interest was delivered to the spray head at a constant rate (0.1 to 1 ml/min) using an HPLC pump (Waters 510, modified for computer control). The liquid was atomized at the ultrasonic nozzle (Sonotek Impact Head) at a frequency of 20 kHz and a power of 2 to 3 W. The aerosols generated by ultrasonic atomization were transported to a heated substrate using directional nitrogen flow at ~12 SCFH. The substrate was placed on a heated copper plate held at a temperature of 30 to 600 °C and translated below the nozzle at a constant rate (1 to 10 mm/s). This configuration allows for precise control of the semiconductor thin film thickness, which is mainly controlled by the liquid flow rate, the substrate temperature, and the number of times the substrate is translated below the nozzle.

For the experiments reported in this thesis, all the CuIn(S,Se)2 and CH3NH3PbI3 based thin films were deposited using a flow rate of 300 μL/min, an ultrasonic power of 2.5 W,

and a constant raster speed of 5 mm/s. The substrate temperature and number of passes were independent variables that are separately controlled in each experiment.

2.1.2 Fabrication of CuInSe2 Based Solar Cells

Figure 2-2 shows the structure of a typical CuIn(S,Se)2 based solar cell in the substrate configuration.[26] Each layer of the device was fabricated or processed as discussed below.

Figure 2-2: Structure of a standard CuInSe2 solar cell in the substrate configuration.

(1) Soda-lime glass (SLG) substrates were cleaned sequentially in 1:50 (v:v) Micro-90 aqueous solution and De-Ionized (DI) water for 30 min each. The substrates were then rinsed with DI water and dried with nitrogen flow.

(2) Molybdenum (Mo) back contact was deposited by DC-sputtering at a power of 100

W at a low Ar pressure. A bi-layer structure, deposited at 20 mTorr for 2 min and 4 mTorr for 1 h, respectively, was used to achieve good adhesion to SLG and a good electrical conductivity. 15

(3) CuIn(S,Se)2 absorber layer was prepared by ultrasonic spray deposition. Details will be elaborated in the related chapters.

(4) CdS window layer was deposited by chemical bath deposition (CBD). The 400 mL bath solution consisting of 864 mg ammonium acetate (C2H3O2NH4), 637 mg thiourea

(CH4N2S), 341 mg cadmium acetate (CdC4H6O4), and 12 mL ammonium hydroxide

(NH4OH) was heated to 65 °C . The samples were immersed in the solution for 10 to 18 min for a thickness of 50 to 100 nm. Figure 2-3 shows the experimental setup for a typical

CdS deposition.[27]

Figure 2-3: Photo of an ongoing CdS chemical bath deposition.

(5) TCO layer was deposited by RF sputtering at a low Ar pressure (~4 mTorr). The standard TCO structure consisted of a 50 nm i-ZnO and 250 nm ZnO:Al, deposited at a

power of 80 W for 60 min and 150 W for 90 min, respectively. Alternatively, a 500 nm

ITO (In2O3/SnO2 90/10 WT%) was deposited at a power of 100 W for 60 min.

(6) Antireflection layers and metal grids were not used in this study for the simplicity of device fabrication. The individual cell area was defined by a 5 mm by 5 mm mechanically scribed grids.

2.1.3 Fabrication of CH3NH3PbI3 Based Solar Cells

The so-called “n-i-p” superstrate configuration consisting of FTO/c-TiO2/mp-TiO2/

CH3NH3PbI3/Spiro-OMeTAD/Au (Figure 2-4) was selected as the standard perovskite solar cell structure in this thesis.[15] All of the materials, except for Au, were deposited by spin coating in a nitrogen glove box (Mbraun). Au was deposited by thermal evaporation.

Deposition parameters and processing conditions will be provided in the experimental details when needed.

Figure 2-4: Structure of CH3NH3PbI3 solar cell in the “n-i-p” superstrate configuration.

2.2 Materials and Device Characterization

2.2.1 Material Characterization Methods

The thickness of the solution-processed thin films was measured with a profilometer

(Veeco Dektak 150). The crystal structure and preferred orientation of the films were characterized using a powder X-ray diffractometer (Rigaku Ultima III) with Cu Kα radiation. A typical spectrum was scanned from 5 to 80 o with a step size of 0.04o and a scan speed of 4o per minute in the Bragg-Brentano geometry. The morphology and microscopic structure were determined using a scanning electron microscope (Hitachi HD-

2300A). The chemical composition of the films was analyzed using the energy dispersive spectroscopy (EDS) equipment (Oxford Instruments) that is integrated with the electron microscope. The sheet resistance of the films was measured by the four-point probe (Pro4,

Lucas Labs). The mobility and carrier concentration of the films were obtained by Hall

Effect measurement (H-50 Hall system, MMR Technologies). The optical absorption spectra were measured in the range of 300 to 2000 nm using a UV-VIS spectrophotometer

(Perkin Elmer Lambda 1050). The infrared optical absorption spectra were measured using an FTIR (Thermo Scientific Nicolet 6700) in the attenuated total reflectance (ATR) mode.

The thermal phase transition of the films was measured by thermogravimetric analysis (TA

Instruments Q600).

2.2.2 Device Characterization Methods

The performance of the solution-processed photovoltaic devices was characterized through current density-voltage (J-V) measurement measured using a Keithley 2440 source meter and a solar simulator (Newport model 91195A-1000) configured to simulate AM1.5 18

illumination. Figure 2-5 shows the illuminated and dark J-V curves of a photovoltaic device and the equivalent electric circuit diagram. In general, the J-V curve under illumination is given by [28]:

푞(푉+퐽푅푆) 푉+퐽푅푆 퐽 = 퐽퐿 − 퐽푑 = 퐽퐿 − 퐽0 exp [ ] − (2.1) 푛푘푇 푅푆퐻 where JL is the light generated current density, Jd is the dark current density, J0 is the dark saturation (or leakage) current density, q is the electron charge, V is the applied voltage, RS is the series resistance, n is the diode ideality factor, k is Boltzmann’s constant, T is the temperature of the device, and RSH is the series resistance.

Figure 2-5: (a) J-V curves and (b) the equivalent circuit of a general photovoltaic device.

Several device parameters are used to measure the quality of a solar cell. The short circuit current density JSC (JL) is the photocurrent flows through the solar cell when the applied voltage is zero. JSC characterizes the generation and collection of photo-excited charge carriers, which can be affected by the spectrum of the incident light, the optical properties of the device (window and absorber), and the collection probability (carrier lifetime and mobility).

In addition to the direct J-V measurement, external quantum efficiency (EQE) measurement is an important means to determine the JSC. In this work, the EQE obtained using a PV Measurements setup (model IVQE8-C) was defined as the ratio of the number of collected charge carriers to the number of incident photons onto the solar cells at each wavelength. The short circuit current can be calculated by integrating the product of EQE and the spectrum of the incident light:

퐽푆퐶 = ∫ 푆(휆) 퐸푄퐸(휆) 푑휆 (2.2) where S(λ) is the incident photon flux density per unit wavelength.

The open circuit voltage VOC is the maximum working voltage of a solar cell when the current density is zero. Generally, VOC depends on the ratio of the light current over the saturation current, which can be found by setting Equation 2.1 to zero and neglecting the parasitic resistances (RS and RSH):

푛푘푇 퐽퐿 푉푂퐶 = ln ( + 1) (2.3) 푞 퐽0

Because the light current can be controlled, VOC is mainly determined by the saturation current, which depends on the amount of recombination in the solar cells. As a result, the

VOC is a measure of the quality of crystallinity of the materials and the density of non- radiative recombination centers. The VOC can also be determined by the Fermi energy splitting of electrons and holes, which are limited by the band gap of the absorber material and the carrier concentration. For a solar cell with a p-type absorber, the VOC can be estimated by [29]:

푘푇 푉 = ln [훥푛(푁 + 훥푝)/푛2 + 1] (2.4) 푂퐶 푞 퐴 𝑖

where NA is the dopant (acceptor) density, Δn and Δp are the excess concentration of electrons and holes, and ni is the intrinsic carrier density.

The fill factor (FF) is a parameter to define the ratio of maximum power output from the solar cell to the product of VOC and JSC. It is mostly affected by the defect density in the bulk of the absorber layer and the junction, which can be expressed as a function of VOC

[30]:

퐹퐹0 = [푉푂퐶 − ln(푉푂퐶 + 0.72)]/(푉푂퐶 + 1) (2.5) where VOC in this equation is normalized to the thermal voltage nkT/q. When the parasitic resistances are taken into consideration, the expression can be converted to [30]:

퐹퐹 = 퐹퐹0 (1 − 푅푆) (2.6)

퐹퐹 = 퐹퐹0 [1 − (퐹퐹0/R푆퐻)(푉푂퐶 + 0.7)/푉푂퐶] (2.7) where RS and RSH are normalized to the device characteristic resistance VOC/JSC.

The series resistance (RS), originating from interface resistance, emitter and base resistance, and contact resistance, reduces the FF of the device. Because RS strongly affects the portion of the J-V curve close to the VOC, it can be estimated by taking the slope of the

J-V curve at the VOC point. In contrast, shunt resistance (RSH) can cause significant power losses by creating an alternative current path for the photo-excited charge carriers. RSH is mainly due to manufacturing defects that limit the Fermi energy splitting and significantly influence on both FF and VOC. An estimate for RSH of a solar cell can be determined from the slope of the J-V curve near the JSC point.

The most important parameter, the power conversion efficiency (PCE), is defined as the ratio of maximum electrical power output to the power of incident light, which can be expressed as: 21

휂 = 푉푂퐶 퐽푆퐶퐹퐹/푃𝑖푛 (2.8) where Pin is the input power of illumination (the sun or a solar simulator). The PCE as the most straightforward way to evaluate the performance of a solar cell is used to compare different PV technologies and PV devices manufactured by different approaches.

2.2.3 Laser Beam Induced Current (LBIC)

A built-in-house LBIC system was used to spatially resolve the current collection efficiency map in solution-processed solar cells (Figure 2-6).[31-34] In these measurements, photocurrent is excited by two Nd:YAG diode-pumped lasers (Spectra-

Physics) with wavelengths of 1064 and 532 nm. The 532 nm laser is most often used because the photon energy (2.33 eV) is above the band gap for most solar cell materials.

The 1064 nm (1.17 eV) one can be used for Si and is particularly useful for probing the performance of the bottom cells in perovskite/Si tandem cells. In this case, additional light sources with appropriate filters can be readily included to provide needed light biasing.

Figure 2-7 shows the average output power of the 532 laser as a function of the laser diode pump current and Q-switch repetition rate. The right axis shows the average power density for a 40 μm diameter spot size. We have found that the laser output power is relatively stable over extended periods of time at low diode current and high repetition rates. For a standard device measurement, the 532 nm Nd:YAG laser operating at a repetition rate of 600 kHz was used to generate a light beam with an average power of 0.01 mW and a 40 μm diameter (corresponding to a power density of ~800 mW/cm2 or 2.14 ×

1018 photons/s-cm2). For the measurement of the bottom cell of a tandem device, the 1064

nm laser operating at 1 MHz was used with a neutral density filter with an optical density of 3 to generate a power density of 1 W/cm2 (5.36 × 1018 photons/s-cm2).

Figure 2-6: In-situ LBIC measurement setup.

Figure 2-7: Average laser power of the 532 nm laser as a function of laser current and

frequency. 23

During the LBIC measurement, computer controlled overhead galvanometers are used to scan the laser beam across the solar cells at a speed of 1 mm/s with a 30 μm spacing between two lateral scans. Individual devices are separately connected to a Keithley 2601

Sourcemeter through an electronic switch (Keithley 7001). Light-induced current signals are collected at an acquisition rate of 5 kHz and converted into local external quantum efficiencies using a reference Si photodiode (Model: S2281-8D083) with a calibrated EQE

(Figure 2-8). The EQE of a testing device can be determined by Equation 2.9:

퐸푄퐸 = 퐼 (퐸푄퐸푆𝑖/퐼푆𝑖) (2.9) where I is the LBIC data of the device, EQESi and ISi are the external quantum efficiency at the laser wavelength (Figure 2-8b) and the LBIC (Figure 2-8 c&d) of the Si photodiode.

Figure 2-8: (a) Optical image, (b) EQE curve, (c) LBIC image, and (d) LBIC histogram

of the reference Si photodiode.

For in-situ LBIC measurement, an air-tight environmental box (Figure 2-9a, size: 16 ×

11 × 8 cm3) was built to study the degradation of perovskite solar cells in humidity. The box was purged with a carrier gas (air or N2) through a water bubbler at a flow rate of 3 SCFH

(0.083 m3/min). The gas bubbler was kept at 30 and 60 °C to maintain a 50 ± 5 and 80 ±

5% relative humidity (RH), respectively. The RH value was monitored by a commercial hygrometer stored in an air-tight container that connected to the outlet of the environmental box. For the complementary in-situ optical spectroscopic measurement, an apparatus

(Figure 2-9b, size: 9 × 9 × 9 cm3) was used to control humidity in the similar approach.

Figure 2-9: Environmental chambers for in-situ (a) LBIC and (b) optical absorption

measurements.

Chapter 3

Spray Pyrolysis of Copper Indium Sulfide Thin Films for Photovoltaic Applications

The chalcopyrite semiconductor CuInS2 is a promising photovoltaic absorber material for fabricating low-cost, high-efficiency thin-film solar cells due to a direct optical band gap of ~1.55 eV that lies in the optimum range for solar spectrum,[35] a high optical absorption coefficient,[36] and the tolerance for compositional variations.[37] However, the device performance of CuInS2 based solar cell is inferior to its analog counterpart,

CIGS solar cell due to the difficulty to optimize the solid state chemistry and correlated electronic properties.[38] This chapter discusses the history and material properties of

CuInS2, spray pyrolysis of Cu-In-S thin films, and the fabrication of photovoltaic cells using sprayed In2S3/CuInS2 film stacks.[23] The goal of this study aims at developing a detailed understanding of mechanisms and experimental technique for solution-processed

Cu-chalcogenide materials and related photovoltaic devices.

3.1 Introduction and Motivation

3.1.1 History of CuInS2 Based Solar Cells

Historically, the ternary compound CuInS2 has received considerable attention during the 1970s. In 1971, Tell et al. at Bell Laboratories reported electrical and optical properties 26

of CuInS2 single crystals.[39] Their work necessitated the subsequent studies that showed p- or n-type doping of CuInS2 can be controlled by varying the sulfur content in the compound.[40] Early work demonstrated that a Schottky type solar cell incorporating n- type CuInS2 and a liquid electrolyte yielded a 9.7% efficiency.[41] Followed by this pioneering work, solid-state photovoltaic cells based on p-type CuInS2 were developed and devices with PCEs of 10 - 12% were reported in the early 1990s.[42-44] In these devices, the light absorption and charge carrier generation mainly occur in the p-type CuInS2 absorber, whereas the n-type window layer (e.g., CdS) is transparent to most of the visible light and used to form the p-n junction. Although the device performance of CuInS2 solar cells is inferior to CIGS devices, CuInS2 is still of interest due to the lack of toxic species and the feasibility for solution-based fabrication. With further development of the non- vacuum deposition techniques, an inexpensive pathway towards the commercialization of

CuInS2 based solar cells would be viable.[38]

A variety of deposition methods have been developed for the preparation of CuInS2 thin films, including co-evaporation,[45, 46] sputtering,[47, 48] electrodeposition,[49] chemical bath deposition,[50] spray pyrolysis,[51, 52] and sequential deposition.[53-55]

Among these, the most successful technique for preparing large-area CuInS2 thin film deposition is a two-stage process involving with sequentially (or simultaneously) sputtering (or evaporating) a Cu-In bilayer followed by a reactive sulfurization with the presence of S or H2S gas. More recently, increasing efforts have been devoted to developing alternative low-cost deposition methods. Particularly, in this study ultrasonic spray deposition was used for the preparation of CuInS2 thin films and devices.[23]

3.1.2 Structural Properties of CuInS2 Materials

The semiconductor CuInS2 is a member of the family of I-III-VI2 chalcogenides. This family of materials typically crystallizes in the chalcopyrite structure (Space group: I-

42d).[26] The chalcopyrite structure can be derived from the cubic zincblende structure for

ZnS after replacing the divalent Zn cations at the corners and face centers of the cubic lattice by an alternatively arranged monovalent Cu and trivalent In cations. Figure 3-1 shows the chalcopyrite unit cell of CuInS2, where each metal cation (Cu or In) is tetrahedrally coordinated to four S anions.

Figure 3-1: Chalcopyrite structure of CuInS2.

Depending on the growth conditions, CuInS2 films may possess various chemical phases (polymorphism).[56] In the chalcopyrite structure, the arrangement of non- isovalent Cu and In atoms exhibits a first order phase transition between disordered zincblende structure and ordered chalcopyrite structure. At a temperature higher than 1250

K, due to the random distribution of metal cations in the face-centered cubic (fcc) sublattice sites, the tetragonal chalcopyrite structure become the cubic zincblende structure.[57] 28

Besides the zincblende structure, another cation disordered polymorph of CuInS2 in the wurtzite (P63mc) structure can be prepared by a solvothermal route.[58]

In addition to the cation disorder, variation in the cation orderings can also lead to the

CuInS2 polymorphism (Figure 3-2). Alternating Cu and In cations in the (100) planes and the [111] direction lead to the CuAu phase (P-4m2) and CuPt phase (R-3m), respectively.[56] For the CuAu structure, each S anion is coordinated to 2 Cu and 2 In cations, similar to the chalcopyrite structure. For the CuPt phase, however, each S anion is surrounded by (3Cu+1In) or (1Cu+3In) coordination. Compared with the chalcopyrite phase, these variant phases show a reduction in the direct band gaps.[56]

The theoretical prediction of the polytypes and defect structure of chalcopyrite chalcogenides have shown the complexity of the crystalline phases of CuInS2 thin films.

In addition, the complexity of the material synthesis also lies in the high degree of the freedom in the ternary stoichiometry. Figure 3-3 shows the Gibbs phase diagram of ternary

Cu-In-S system.[59] Preparation of the CuInS2 thin films with different Cu to In ratios and with S sufficiency (or deficiency) can influence the structural and electronic properties of the film, and thus determine the device performance.

For instance, the benefits of excess Cu during the growth of CuInS2 thin films were well-known in the early 70s. PV devices fabricated with Cu-rich CuInS2 films typically showed a good device performance in contrast to Cu-poor films.[44] The most efficient

CuInS2 devices to date have also been achieved for a Cu-rich growth regime, where a secondary Cu2-xS phase is present and appear to be advantageous for improving the crystalline quality of the films.[60] However, the presence of this binary phase can adversely affect device performance due to high conductivity and the role as non-radiative recombination centers. The secondary phase segregates have to be removed by selective etching with KCN.[61] To avoid the etching step, the precursor composition has been shifted to the Cu-poor regime. However, Cu-poor phases are semi-insulating due to a high concentration of sulfur vacancies,[62] which also deteriorates solar cell performance.

Figure 3-3: Ternary phase diagram of Cu-In-S. Reprinted from Ref. [59] with permission.

Although the structural and phase properties of CuInS2 are complicated and difficult to precisely control, there is still an advantage in the Cu-In-S system. Specifically, there is tolerance to stoichiometric variation (Cu:In from 1.0 to 1.8) provided that there is sufficient sulfur present to promote p-type conductivity.[44] For solution-based methods, both metal and sulfur source concentrations can be controlled in the precursor solution. Consequently, it is feasible to study the formation of the CuInS2 thin films by varying the composition of the precursor solution, to provide routes to fabricate high-efficiency CuInS2 photovoltaic devices via solution-based processes.

3.2 Cu-In-S Thin Films Prepared by Spray Pyrolysis

In this study, ultrasonic spray pyrolysis is introduced as an alternative to the established deposition techniques that have been used to prepare CuInS2 based thin films. Spray pyrolysis provides a different approach to study the influence of the preparation conditions on the film quality and device performance, which is poorly understood. To develop a better understanding, it is important to investigate the growth of the material under different conditions and explore the Cu-In-S composite in different phases.

3.2.1 Mechanism of Spray Pyrolysis of Cu-In-S Thin Films

Figure 3-4 shows the mechanism of CuInS2 thin film deposition using the spray pyrolysis method. During the spray pyrolysis process, the substrate was maintained at an elevated temperature in the range of 300 to 500 ℃ to provide sufficient energy to promote the relevant chemical reactions. During the ultrasonic spray process, the precursor solution is ultrasonically atomized into a mist of fine droplets (90% within ~100 μm). These aerosol 31

droplets are comprised of the precursor solution and can act as micro-reactors where Cu-

In-S materials with the desired composition can nucleate and grow, while the volatile byproducts are vaporized and vented away. Thin films are deposited when the guiding gas transports these atomized aerosols to the heated substrate.

Although the spray process has many advantages such as ease of use, parameter flexibility, scalability, and material preservation, it exhibits some drawbacks that have to be addressed. First, localized compositional fluctuations in the nucleation stage may lead to the formation of secondary phases, such as Cu2-xS, CuIn5S8, and other ordered vacancy compounds. Moreover, the high entropy of mixing for a few undesired constituents, for example, O, C, N, and Cl in the films at elevated temperature makes incorporation of external impurities energetic favorable and places high demands on source chemical, solvent, and ambient purity. Additionally, if the arrival and evaporation are not carefully balanced, solvent may re-dissolve the deposited films, leading to the formation of cracks and voids and rough film morphology, or alternatively, if the evaporation rate is too slow,

“coffee stains” may be left on the surface of the film due to the capillary flow effects.[63]

Most of these impurities and processing defects in the films act as either recombination centers that interact with the photogenerated minority carriers or light scattering centers, and, thus, adversely affect the performance of optoelectronic devices fabricated using this technique. Therefore, the spray deposition parameters need to be finely tuned to avoid the formation of impurities and processing defects.

Figure 3-4: Schematic of the CuInS2 spray pyrolysis mechanism.

3.2.2 Spray Pyrolysis of Cu2-xS Thin Films

To study the formation conditions of secondary binary phases, Cu2-xS (0 < x < 1) thin films were prepared using copper chloride and thiourea in aqueous solution. By modifying the Cu to S precursor ratios and deposition temperatures, various combinations of Cu2-xS phases were obtained. Note that this binary composition also serves as a simplified material relative to CuInS2. Table 2.1 lists data for some Cu2-xS thin films prepared under different conditions, where the phase composition, electrical conductivity, and optical band gap were determined by XRD, four-point probe, and optical absorption spectroscopy.

The binary compound Cu2-xS exhibits an intricate manifold of phases,[64] where the crystal structure changes as a function of composition and temperature. Depending on the x value, the Cu2-xS films can exist as the chalcocite (Cu2S), djurleite (Cu1.94S), digenite

(Cu1.8S), or anilite (Cu1.75S) phase. Among them, the stoichiometric chalcocite phase has three different structures at different temperatures. The monoclinic phase (α-Cu2S) forms at a temperature below 100 oC. When the temperature is in the range of 100 to 450 oC, the

β-Cu2S phase, which has a hexagonal chalcocite structure, is the most stable phase. When

o the temperature is higher than 450 C, it transforms to the cubic structure (γ-Cu2S).

Table 3.1: Cu2-xS thin films prepared by spray pyrolysis

-1 Cu:S Temperature (℃) Phase composition σ (S cm ) Eg (eV)

2 1:2 300 Cu1.75S + β-Cu2S 7.5 ×10 1.27

3 1:2 400 CuS + Cu1.75S 3.8 ×10 1.31

2 1:2.5 300 Cu1.8S + β-Cu2S 8.4 ×10 1.21

3 1:2.5 400 CuS + Cu1.75S + β-Cu2S 2.7 ×10 1.26

2 1:3 100 α-Cu2S 1.1 ×10 1.12

2 1:3 200 α-Cu2S + β-Cu2S 3.6 ×10 1.15

2 1:3 300 Cu1.94S + β-Cu2S 7.5 ×10 1.25

3 1:3 400 Cu1.8S + β-Cu2S 2.1 ×10 1.18

3 1:3 500 CuS + Cu1.75S + γ-Cu2S 4.2 ×10 1.25

2 1:3.5 300 β-Cu2S 1.6 ×10 1.18

3 1:3.5 400 Cu1.8S + β-Cu2S 2.3 ×10 1.22

In addition to the complexity of polymorphism, the application of the sprayed Cu2-xS thin films in photovoltaic devices is limited by a high carrier concentration and a low mobility. Figure 3-5 shows the relationship between hole concentration and mobility in the sprayed Cu2-xS films. The high carrier concentration limits the depth of depletion region in the absorber, while the low hole mobility constrains the carrier diffusion length. Both parameters are not favorable for solar cell operation. As a result, Cu2-xS phases should be avoided as the absorber in the Cu-chalcogenide solar cells.

Figure 3-5: Hole concentration and mobility of sprayed Cu2-xS thin films.

3.2.3 Spray pyrolysis of CuInS2 and In2S3 thin films

To solve the issues related to Cu2-xS, ternary compounds with the incorporation of indium were introduced. In contrast to Cu2-xS, CuInS2 has stronger tetrahedral covalent bonding and is more stable and more suitable for solution-processed Cu-chalcogenide solar cells. Additionally, the use of indium provides an alternative window layer material In2S3, which is less toxic than the conventionally used CdS and is better suited for band alignment with CuInS2. Because both the window layer and absorber layer can be fabricated within the Cu-In-S system, the solar cells can be completed in a single spray process by adjusting the precursor ink composition.

The CuInS2 and In2S3 thin films were prepared by spray pyrolysis of the aqueous precursor solution.[23] The precursor sources, copper (II) chloride (CuCl2), indium chloride (InCl3) and thiourea [(NH2)2CS] were first dissolved in deionized (DI) water, targeting a final composition of (Cu+In)/S=1/3, with a nominal Cu concentration in the

solution of approximately 0.01 M. The stoichiometry of the resulting film is controlled by the cations providing that sufficient sulfur concentration is available.

Figure 3-6 shows the transmittance and absorption spectra of CuInS2 and In2S3 films prepared by spray pyrolysis. The direct bandgap of the CuInS2 thin film can be calculated to be ~1.53 eV, which is an optimized bandgap value for solar irradiance. In contrast, the

In2S3 film, which has a band gap of ~2.7 eV and a high optical transparency of greater than

85% at wavelengths above 510 nm is an appropriate candidate for the window layer of

CuInS2 based solar cells.

Figure 3-6: Optical transmittance and absorbance spectra of CuInS2 and In2S3 thin films.

To determine the chemical phase and structure of the sprayed CuInS2 and In2S3 thin films, XRD spectra of the films were measured. Figure 3-7 shows that all peaks from the tetragonal SnO2 (FTO glass) are preserved after the spray pyrolysis process. The robust chemical stability of SnO2 during the spray process makes superstrate CuInS2 solar cell configuration feasible. The structures of the In2S3 and CuInS2 layers were identified using

XRD. The crystal structure of sprayed β-In2S3 was confirmed to be the ordered defect

spinel structure (I41/amd) with the dominant peaks from (109) and (00,12) planes. The sprayed CuInS2 film was crystallized in the chalcopyrite phase with the principal peak from the close-packed (112) plane. No impurity phases were identified in the XRD spectra.

Additionally, the XRD spectrum for the CuInS2/In2S3/TCO structure shows the characteristic peaks of each semiconductor material indicating the existence of the three distinct layers in the completed device structure.

To estimate the average grain size of the sprayed films, Debye-Scherrer analysis was applied to the CuInS2 and In2S3 films sprayed on witness glass slides to the dominant peak for each material, which can be described by

Kλ 휏 = (3.1) βcosθ where τ is the average grain size, K = 0.9 is the coefficient for the Gaussian peak fitting, λ

= 0.154 nm is the Cu characteristic X-ray wavelength, β is the FWHM of the dominant

XRD peak, and θ is the angle of the XRD peak.

The average grain size of In2S3 and CuInS2 was estimated to be 56 ± 3 and 28 ± 2 nm, respectively, which were consistent with the grain size values measured from SEM images

(Figure 3-8).

Figure 3-7: XRD spectra of sprayed (a) In2S3, (b) CuInS2, and (c) In2S3/CuInS2 thin films

on FTO-coated glass substrate.

Figure 3-8: SEM morphological images (×50K) of sprayed (a) In2S3 and (b) CuInS2.

The measured crystal size is typical for films synthesized using solution-based processes and is known to limit the device performances due to the short mean free path and high grain-boundary density. Small crystal sizes are inherent in low-temperature routes because the thermal energy provided, while sufficient to overcome the nucleation barrier, is not adequate for allowing gain merging and growth. As a result, thin films deposited by spray pyrolysis are comprised of nanocrystalline grains.

The formation of nanocrystalline grains is a result of a high nucleation density and rapid transition to a stable solid-state phase. Stable bonding within the nuclei hinders grain growth. Efforts to produce larger crystals via high-temperature deposition or post- deposition annealing were ineffective and instead produced oxidation (reaction with the aqueous solvent and/or air), sulfur deficiency (higher volatility of sulfur at high temperatures), and Cu diffusion at the In2S3/CuInS2 interface, altering the p-n junction.

Thus all films used in these experiments were used as-deposited without any post- deposition treatment.

3.3 Spray Pyrolysis of CuInS2/In2S3 Thin Film Solar Cells

Spray pyrolysis of In2S3 and CuInS2 nanocomposites provides a viable route toward the scalable production of thin-film chalcogenide solar cells. Unfortunately, solar cells produced by these methods were limited to PCES of ~2%. Post-deposition thermal annealing is not involved in the device fabrication, so the process is simple and straightforward.

In this study, we investigated solar cells prepared by sequentially depositing In2S3 and

CuInS2 thin films on FTO-coated glass substrate using ultrasonic spray pyrolysis. The heterojunction of the Photovoltaic cells of the structure Ag/CuInS2/In2S3/FTO was fabricated followed by the fabrication process illustrated in Figure 3-9.

Figure 3-9: Fabrication process of sprayed CuInS2/In2S3 solar cells.

To investigate the photovoltaic performance of sprayed CuxIn1-xS2 thin films,

FTO/In2S3/CuxIn1-xS2/Ag solar cells in the superstrate configuration were fabricated with varying composition of the Cu1-xInxS2 films. Figure 2-10 shows a typical cross-sectional

SEM image of the device consisting of a 300 nm n-type In2S3 window layer and a 900 nm p-type CuInS2 absorber layer. The lack of visible grain structure in the window and absorber layers is indicative of small grain size of the sprayed films and is one of the inherent challenges for high-quality devices fabricated via low-temperature, solution-based methods.

Figure 3-10: Cross-sectional SEM image shows distinguishable layers in the

CuInS2/In2S3/FTO device.

The illuminated current density-voltage (J-V) measurements were taken for the completed devices. As a proof of concept, the sprayed CuInS2/In2S3 thin film stacks demonstrate pronounced photovoltaic responses. The statistics of the solar cell performance metrics (the average of 60 cells) for each absorber layer composition is

presented in Table 3.2 and the best J-V curves are shown in Figure 3-11. The VOC of the sprayed CuxIn1-xS2 devices decreases with increasing Cu composition. The decrease of VOC from Cu-poor to Cu-rich cells is mainly due to the decrease of the absorber bandgap and the increase in defect density. The Cu-rich samples demonstrate slightly higher JSC values as a result of a lower optical band gap (Figure 3-12). Additionally, the Cu-rich samples exhibit a higher doping density in the absorber layer which was confirmed by the sheet resistivity measurements on the witness samples deposited on glass substrates (Figure 3-

11 inset). However, as Cu concentration increases above 1.2, the formation of the semimetal Cu2-xS phase becomes a dominant effect that creates shunts (low RSH) through the devices, which are detrimental to the device performance. As a result, we found the optimal composition to be Cu1.1In0.9S2, and devices with this absorber layer demonstrated the best efficiency of 1.93%, which is comparable to the 4% efficiency devices with a similar structure.[65]

Figure 3-11: Current-voltage characteristics of several CuxIn1-xS2 devices under 1.5 AM

illumination. The inset is resistivity of Cu1-xInxS2 films. 42

Table 3.2: Summary of the sprayed solar cell performance

VOC JSC FF Efficiency RS RSH Absorber (V) (mA/cm2) (%) (%) (Ω cm2) (Ω cm2) 0.485 ± 4.08 ± 45.6 ± 0.91 ± 42.3 ± 451.6 ± Cu0.8In1.2S2 0.053 1.05 5.1 0.15 8.6 86.7 0.458 ± 4.37 ± 36.2 ± 0.76 ± 49.7 ± 239.2 ± Cu0.9In1.1S2 0.047 0.96 10.5 0.25 10.8 43.5 0.445 ± 5.24 ± 51.4 ± 1.19 ± 25.3 ± 248.5 ± CuInS2 0.035 1.13 6.7 0.18 5.1 48.8 0.451 ± 5.95 ± 61.5 ± 1.71 ± 11.9 ± 293.1 ± Cu1.1In0.9S2 0.057 1.53 3.8 0.17 2.4 61.3 0.392 ± 5.62 ± 49.5 ± 1.09 ± 25.1 ± 147.5 ± Cu1.2In0.8S2 0.071 1.15 8.1 0.21 2.9 28.3

Figure 3-12: Optical band gaps of the sprayed Cu-In-S thin films derived from the UV-

VIS absorption spectra.

3.4 Factors Limited the Performance of the Sprayed CuInS2 Solar Cells

A simple model for the VOC analysis is through the difference in Fermi energy between the p-type CuInS2 and the n-type In2S3. Due to the low doping density of In2S3, the Fermi energy for electrons lies in the middle of its band gap. Ideally, the Fermi energy would be driven near the conduction band through doping thereby increasing the VOC; however, since this is not the case, the VOC and the driving force for the forward bias electron movement is reduced. Another property that affects VOC for the sprayed CuInS2/In2S3 solar cells is the built-in field at the heterojunction. The carrier density of n-type In2S3 window layer is orders of magnitude lower than that of the p-type CuInS2 absorber layer, so a significant portion of the depletion region is located in the window layer rather than the absorber layer, as depicted in Figure 3-13. The shallow depletion region in the CuInS2 layer makes the less efficient diffusion of the photogenerated electron-hole pair, rather than drift, the dominant transport mechanism. This, again, reduces the VOC and could be partially remediated by increasing doping in the In2S3 layer.

Figure 3-13: Band diagrams for the sprayed CuInS2/In2S3 solar cell. 44

The adverse effects of the low photocurrent transport caused by the defects are responsible for the low JSC, which is a measure of how well the photogenerated carriers are collected. For films with small grains and defects, the carrier transport properties are likely dominated by the grain boundaries and impurities, which increase the recombination of photogenerated electrons and results in low current density. The JSC is further reduced by the high series resistance of the In2S3 window layer by limiting the carrier transport to the front contact.

To investigate the recombination mechanism of the sprayed CuInS2 devices, we measured temperature dependent J-V curves from 120 to 300 K at 20 K intervals (Figure

3-14, courtesy of Dr. Paul Roland). The curves were fitted to the single diode model

(Equation 3.1) to obtain a linear plot of n(T)Ln(J0) versus 1/kT. The slope of the plot gives the activation energy for recombination of 0.963 eV, which is less than the band gap of

CuInS2 (1.53 eV). Consequently, recombination at the interface (rather than in the space charge region) is the dominant recombination pathway for this sample.[66]

Figure 3-14: J-V curves of a CuInS2 solar cell with respect to temperature. 45

As another important parameter of solar cell performance, FF indicates the quality of solar cells. Typically the FF of sprayed CuInS2 solar cells is approximately 30%. For the best cells we prepared, a FF of 58% was obtained. From the empirical expression for FF, increasing FF is the result of the improvement of diode ideality factor, indicating the reduction of space-charge recombination. The more ideal p-n junction created in this work may be due to the optimization of spray parameters including a low solution supply rate and precisely controlled annealing time to limit Cu diffusion close to the interface.

Alternatively, a high FF could be a consequence of the ratio of shunt and series resistance.

It is possible that the highly resistive In2S3 layer is responsible for the high FF.

For the devices fabricated here, the maximum efficiency measured was 1.93%. This value is in agreement with other solution-processed CuInS2 solar cells. In all cases, the VOC is in the same range. However, the solar cells fabricated for this work exhibit lower JSC and higher FF. Based upon the above discussion, both of these variants could be due to the highly resistive In2S3 layer. Therefore, future work should be focused on using a thinner or more highly doped In2S3 layer.

3.5 Conclusion

In summary, we have studied the structure and crystallinity of sprayed chalcopyrite Cu-

In-S thin films with different compositions. Nanocrystalline grains are found to be the result of nucleation and growth from the liquid phase precursors at low temperatures. Using the spray pyrolysis approach, we have fabricated solar cells with the superstrate structure of Ag/CuInS2/In2S3/SnO2 and achieved the best efficiency of 2%.

The factors limited the performance of sprayed superstrate CuInS2 solar cells include general drawbacks from low-temperature solution-based processes, such as small grain size, impurities, and native defects, as well as the deficiency of window and back contact materials. As a result of these limitations, it is challenging to achieve high-efficiency solar cells by using spray pyrolysis. Even if production costs are low, a low efficiency is not likely to be suitable for photovoltaic application due to the non-manufacturing balance of systems and installation costs. Future research aimed to optimized the fabrication process and a efforts to better understand, control, and circumvent the limitations may solve these issues and promote the commercialization of sprayed CuInS2 cells.

Chapter 4

Hydrazine Based Copper Indium Selenide Thin Film Solar Cells

This chapter focuses on the development of a novel approach to prepare CuIn(S,Se)2 thin films through ultrasonically spraying a hydrazine-based precursor solution onto a heated substrate. Using hydrazine hydrate as a solvent allows the incorporation of selenium into the precursor solution, leading to improved crystallinity and morphology of the sprayed films. These films were used to fabricate CuInSe2 thin-film solar cells that demonstrated a photovoltaic response of 7.2%.[24] The effects of the deposition temperature and annealing conditions on the device performance were investigated.

Additionally, devices prepared in the back-wall superstrate configuration have been studied.[25] With optimization, the synthesis of high-quality CuIn(S,Se)2 films by spray pyrolysis from a hydrazine hydrate solution could demonstrate the viability for a low-cost, high-throughput, solution-based manufacturing process.

4.1 Introduction and Motivation

Low-cost fabrication techniques for the absorber layer in thin film photovoltaic (PV) devices are essential to enable thin film solar cells to compete with crystalline silicon panels in the renewable energy market. Among the thin-film PV technologies, the family 48

of CuIn1-xGax(Se,S)2 (CIGS) has attracted significant attention because a power conversion efficiency of nearly 22.3% has been achieved.[22] High-efficiency CIGS solar cells are conventionally fabricated using vacuum-based evaporation or sputtering approaches.[16,

67-70] However, the use of vacuum-based deposition facilities requires high capital investment and manufacturing costs, which impede the commercialization of CIGS solar cells. Recent progress of high efficient CIGS solar cells has focused on the investigation of low-cost solution-based deposition techniques,[9] using metal salt molecular precursor solutions [17, 71-73], metal–organic molecular precursors,[74, 75] and nanoparticles.[76-

79] All the approaches are applicable for scalable fabrication using roll-to-roll printing (or spraying) process. However, the commercial failure of the 17.1% [80] solar cells from

Nanosolar has revealed the limitation of the nanoparticle route, including incorporation of impurity elements, the complexity of synthetic and purification processes, and the requirement for a high temperature selenization step using toxic gases (e.g., H2Se), which makes the formation of a compositional uniform and phase pure CIGS layer very difficult.

In contrast, a pure solution-based method using hydrazine-based solution and yielding a high device efficiency of 15.2% demonstrates a great potential for manufacturing CIGS solar modules.[17]

The hydrazine-based approach has been developed by IBM to produce high-quality

CIGS and Cu2ZnSn(Se,S)4 (CZTS) thin film devices.[17, 81-83] In this approach, a molecular-based precursor solution comprising metal chalcogenide anions and

+ hydrazinium (N2H5 ) cations with the targeted composition was spin-coated onto a rigid substrate.[18] Although this approach is applicable to most solution-based processes, concerns about the toxicity of hydrazine have limited it to small area spin coating of the 49

absorber layer of photovoltaic devices. Compared with spin coating, the spray deposition process has several advantages including coating of arbitrarily large areas, capability to deposit on flexible and irregular shape surfaces, consistent control of substrate temperature, and continuity of processing. Thus, the goal of this study is to develop the spray deposition approach for CuInSe2-based thin film solar cells.

4.2 Hydrazine-Based CuIn(S, Se)2 Solar Cells Prepared by Spray Deposition

4.2.1 Precursor Solution Preparation

The solutions used in this study consisted of (a) metal salts in aqueous solution, (b) metal chalcogenides in anhydrous hydrazine solution, and (c) metal chalcogenides in hydrazine hydrate solution. The aqueous precursor solution (a) was prepared by dissolving

40 mM CuCl2, 48 mM InCl3, 240 mM (NH2)2CSe, and 80 mM (NH2)2CS in deionized (DI) water. A clear colorless solution was formed after filtration and used immediately to avoid aggregate formation (Figure 4-1a). For hydrazine-based solution (b) and (c), three types of precursor solution: 0.5 M Cu2S and 1 M S in hydrazine, 0.25 M In2Se3 and 1 M Se in hydrazine, and 2 M Se hydrazine solution were prepared following procedures reported by others.[82] NOTE: hydrazine should always be handled using precautions and proper protective equipment due to the extremely high toxicity. The overall chemical reactions involve with the hydrazine-based precursor solutions can be described by the following equations.[84]

Figure 4-1: Photos of precursor solutions with (a) aqueous, (b) anhydrous hydrazine, and

+ 2− 2 푆 + 5 푁2퐻4 → 푁2 + 4 (푁2퐻5) + 2 푆 (4.1)

+ 2− 6 퐶푢2푆 + 2 푆 + 5 푁2퐻4 → 푁2 + 4 (푁2퐻5) + 2 (퐶푢6푆4) (4.2)

+ 2− 2 퐼푛2푆푒3 + 2 푆푒 + 5 푁2퐻4 → 푁2 + 4 (푁2퐻5) + 2 (퐼푛2푆푒4) (4.3)

Solution (b) was prepared by mixing 1 ml of the Cu2S solution, 2.2 ml of the In2Se3 solution, 0.8 ml of the Se solution, and then diluting with 22 ml of anhydrous hydrazine

(Figure 4-1b). The less toxic solution (c) was prepared in the same way as solution (b), except DI water was used in place of the anhydrous hydrazine, yielding a 12% volume fraction hydrazine (Figure 4-1c). Upon addition of the water, the color changed from pale yellow to dark brown indicating the presence of increasing amounts of poly chalcogenide

+ species as N2H5 cation concentration decreased. This change is confirmed by measuring the chalcogenide complexes concentrations in the three precursor solutions.

Figure 4-2 shows the FTIR-Raman spectra of the three precursor solutions. The spectrum belonging to the aqueous solution (Figure 4-2a) exhibits two equally intense

peaks at 200 and 350 cm-1, which can be attributed to the In-Se and Cu-S stretching modes, respectively. Because of the sensitivity of In-Se mode is much higher than that of Cu-S, the similar intensity values indicate that In exists in the ionic form rather than as a complex in the aqueous solution. In contrast, the anhydrous hydrazine solution (Figure 4-2b) shows a dominant feature of the In-Se and a minor feature of Cu-S, which is close to the spectrum of equal amounts of Cu to In chalcogenide complexes in the solution.[84] The addition of water caused the aggregation of poly-chalcogenide species, leading to the reduction of both

2- + In2Se4 and N2H5 (Figure 4-2c). As a consequence, the viscosity of the solution was increased, which helps the formation of dense and smooth films during the spray deposition.

Figure 4-2: FTIR-Raman spectra of precursor solutions involving with (a) aqueous, (b)

anhydrous hydrazine, and (c) hydrazine hydrate as the solvent, respectively. 52

4.2.2 Crystallinity of the Sprayed Films

To compare the crystallinity of the spray deposited CuInSe2 films produced at various temperatures, the preferred orientation and crystallite size of the sprayed films were investigated. Figure 4-3 shows the XRD spectra of the CuInSe2 films prepared on soda- lime glass substrates at various temperatures for each precursor solution. All the sprayed films show characteristic peaks corresponding to the tetragonal chalcopyrite phase of

CuInSe2. The dominant closed-packed (112) plane and other minor peaks, including

(220)/(204) and (116)/(312) planes were observed in all the films regardless of growth conditions. Comparing the degree of preferred orientation, the films fabricated using aqueous solution exhibit random orientation, while hydrazine-based films demonstrate a strong preferential (112) orientation. To quantitatively evaluate the degree of preferred orientations, the Harris texture coefficient as described by Equation 4.4 was calculated.

퐼 (ℎ푘푙) / 퐼0(ℎ푘푙) 훼ℎ푘푙 = 1 (4.4) ∑푁 퐼 (ℎ 푘 푙 ) / 퐼 (ℎ 푘 푙 ) 푁 푖=1 푖 푖 푖 0 푖 푖 푖

where I(hkl) is the measured XRD intensity of the (hlk) plane, I0(hkl) is the intensity of the same peak from the standard power diffraction pattern , and N is the number of considered peaks. For this calculation, Powder Diffraction File (PDF) #65-7027 was referenced as standard power diffraction pattern of CuInSe2.

Figure 4-3: XRD spectra of sprayed CuInSe2 thin films at various temperatures using the

(a) aqueous, (b) anhydrous hydrazine, and (c) hydrazine hydrate solution.

Figure 4-4 shows the (112) plane texture coefficient for three precursor solutions as a function of the deposition temperature. All the coefficients for the films using the aqueous solution are close to 1, indicating that the grains within the films are randomly oriented.

The highest texture coefficient for the (112) plane is obtained for the films fabricated using the hydrazine hydrate solution. This preferred orientation along the (112) also holds when the CuInSe2 films are deposited on Mo using hydrazine hydrate solution and has been reported by others using hydrazine-based methods. At deposition temperature of 450 °C,

the average orientation factors of the (112), (220/204) and (116/312) planes change from

1.32: 0.85: 0.83 for the films fabricated using the aqueous solution to 1.79: 0.63: 0.58 and

2.50: 0.24: 0.26 for films fabricated using anhydrous hydrazine and hydrazine hydrate solution, respectively.

Figure 4-4: (a) The (112) direction texture coefficients and the (b) crystallite sizes

calculated from XRD spectra for sprayed CuInSe2 thin films.

To estimate the mean size of the crystalline domains within the polycrystalline CuInSe2 thin films, Williamson-Hall (W-H) analysis was used.

0.9휆 훽 cos(휃) = 4 휀 sin(휃) + (4.5) 휏

where β is the FWHM of the peak after instrument broadening correction at the Bragg angle θ, ε is the lattice strain, λ = 0.154 nm is the Cu X-ray wavelength, and τ is the average crystalline size.

Figure 4-4b shows the average crystallite size derived from Equation 4-5 as a function of deposition temperature. The crystallite size increases with increasing deposition temperature regardless of the precursor solution. The crystal sizes of as-sprayed films using hydrazine hydrate solution are larger than those using anhydrous hydrazine and aqueous solution. This is consistent with the early study which showed that the introduction of water in the CIGS deposition helps crystal growth. Combining the texture and grain size results, the aqueous solution leads to loosely packed, randomly orientated small grains, while the hydrazine-based solutions result in closely packed, the [112] direction oriented large grains, as illustrated in Figure 4-5.

Figure 4-5: Schematics of thin films prepared by aqueous and hydrazine-based solutions.

4.2.3 Morphology and Composition of the Sprayed Films

To further investigate the morphology of the deposited films, top and cross-sectional SEM images of the sprayed films were obtained (Figure 4-6). Films fabricated using aqueous solutions are comprised of mostly small grains ranging in size from 40 to 80 nm, while films fabricated using hydrazine solutions exhibit much larger grains. Grains ranging in size from 90 to 300 nm are present in films processed from anhydrous hydrazine solution. Grains of the films processed from hydrazine hydrate show similar grain size with several large grains (400 to 600 nm) present. The 56

same basic trend is also observed in the cross-sectional images. The larger grain size in hydrazine hydrate films is likely due to the higher viscosity of the solution. Instead of forming ring-stains during solvent evaporation, the high viscosity of the solution prevented the suspended solutes from flowing to the edge of the droplet during solvent evaporation and resulted in a uniform deposition.

In addition, the polyselenide species in the hydrazine hydrate solution created a locally high chalcogen concentration, which is believed to be the key factor to promote the chalcogenide crystal growth.

Although the hydrazine-based CuInSe2 films exhibit significant grain size enhancement, porosity in the films is observable. The presence of the isolated voids at the bottom of the film is very problematic since they lead to poor adhesion between the CuInSe2 films and the Mo back contact. In a CuInSe2 solar cell, these defects impede intergrain carrier transport through recombination, and thus limit the voltage of the device.

Figure 4-6: Top view and cross-sectional SEM images of CuInSe2 films obtained by

spray pyrolysis at 450 °C, using (a) aqueous, (b) anhydrous hydrazine and (c)

hydrazine hydrate solutions.

The atomic compositions of elements in the films fabricated at 450 °C were measured using energy dispersive X-ray spectroscopy (EDX). In this study, the precursor solutions were adjusted to yield a stoichiometry close to Cu0.9In1.1(S,Se)2. Regardless of the initial atomic ratio of S to Se, the resultant films were always found to be in the selenide phase, especially at temperatures higher than 350 °C, presumably due to the higher vapor pressure of sulfur. Figure 4-7 shows the EDS spectra of CuInSe2 films prepared using three precursor solutions. Both hydrazine-based solutions resulted in compositionally uniform films with the desired stoichiometry of Cu (24%), In (27%),

Se (48%) and S (1%), while impurities like oxygen (5%) and non-stoichiometric Cu-rich phases such as Cu2-xS were found in the films processed by aqueous solution.

Figure 4-7: EDS spectra of CuInSe2 films obtained by spray pyrolysis at 450 °C, using

(a) aqueous, (b) anhydrous hydrazine and (c) hydrazine hydrate solutions.

The non-stoichiometric Cu2-xS is the most detrimental defects in the sprayed

CuIn(S,Se)2 films. When the ~100 µm sized precursor solution droplets are transported to the surface of the heated substrates, the large temperature gradient between the droplet

(~50 °C ) and the substrate (~400 °C ) can cause chaotic motion of the solutes, leading to compositional perturbation inside the droplet. During the wetting process, the droplet spreads and dries, and the portion with high Cu concentration is likely to form micron- sized defects, as shown in Figure 4-8. These defects can deteriorate the device performance or directly shunt the whole device, thus should be avoided in the CuInSe2 device fabrication. Although these defects are widely presented in the CuInS2 devices prepared by the aqueous solution, using the hydrazine-based solutions can greatly reduce the density of these Cu2-xS aggregates.

Figure 4-8: EDS scan of the non-stoichiometric Cu2-xS defect in the CuInSe2 films

obtained by spray pyrolysis of the aqueous precursor solution.

4.2.4 Device Performance of the Sprayed CuInSe2 Solar Cells

As a proof-of-concept, preliminary CuInSe2 solar cells were fabricated with an active layer deposited using spray pyrolysis. An in-house spray deposition system contained inside a nitrogen glove box was used to deposit the CuInSe2 films. The precursor solutions were ultrasonically sprayed onto soda-lime glass slides (25 mm × 76 mm × 1 mm) held on a stage heated to an elevated temperature ranging from 200 to 450 °C. To fabricate photovoltaic devices, a 1.2 µm thick CuInSe2 film was sprayed onto a 1 µm thick sputtered

Mo films and annealed with selenium powder in an enclosure at 450 °C for 1 hour. 150 nm thick CdS film was deposited on the CuInSe2 using chemical bath deposition, followed by sputtering a 100 nm intrinsic ZnO buffer layer and a 350 nm Al-doped ZnO transparent conductive contact to complete the devices. For comparative devices, a 500 nm ITO layer was used in place of the top contact.

Table 4.1 lists the parameters of the highest performing cells and the corresponding J-

V curves are plotted in Figure 4-9. The spray-processed CuInSe2 films using traditional aqueous based solution exhibit a shunting problem likely due to small grain size, pinhole formation, and inhomogeneities. On the contrary, spray pyrolysis using a hydrazine-based solvent results in CuInSe2 films with larger grain size, reduced pinhole density, and better homogeneity. The device performance was degraded by primary problems with the sprayed films, including a high pin hole density, the presence of impurity phases, and non- stoichiometry. Defects in the absorber layer act as recombination centers, limiting open circuit voltage (VOC) and device efficiency. The poor fill factor (FF) and short-circuit current density (JSC) are mainly due to the reduced carrier transport in the buffer (CdS) and highly resistive transparent (i-ZnO) layers which were purposefully increased to avoid 60

device shunting. The formation of MoSe2 on the back contact due to long time selenization might also contribute to the high series resistance and lower the efficiency of the devices.

Table 4.1: Performance of CuInSe2 solar cells prepared by spray pyrolysis.

Precursor TCO VOC JSC FF Eff. RS RSH solution Layer (V) (mA/cm2) (%) (%) (Ωcm2) (Ωcm2) aqueous ZnO 0.151 6.27 27.3 0.26 19.6 30.0 solution anhydrous ZnO 0.317 16.58 28.1 1.47 14.9 35.3 hydrazine hydrazine ZnO 0.322 12.96 34.1 1.42 14.6 57.4 hydrate anhydrous ITO 0.368 13.11 39.8 1.92 13.0 170.4 hydrazine hydrazine ITO 0.397 21.94 50.0 4.15 5.6 209.9 hydrate

Figure 4-9: J-V curves of sprayed CuInSe2 solar cells based on (a) aqueous, (b)

anhydrous hydrazine and (c) hydrazine hydrate solutions.

Optimization of precursor solution, spray parameters, and selenization process are ongoing and should result in pinhole reduction, smoothening of thin film surfaces and fine tuning the film stoichiometry. The optimization of the sprayed film should relax the requirements on the buffer and highly resistive transparent layers, thereby increasing the

FF and JSC. The improvement of device performance was observed in comparative devices when more conductive ITO was used in place of ZnO. Although the performance of the preliminary devices is still poor, the cells using hydrazine hydrate solution show higher

VOC and efficiency than aqueous-based ones, indicating that less process optimization will be required for these films.

In summary, the properties of CuInSe2 thin films processed by spray pyrolysis using different precursor solutions were studied. Results showed that hydrazine-based films exhibited stronger preferable orientation, larger grain size, smoother surface, and better compositional uniformity than films produced without hydrazine. Using hydrazine hydrate instead of anhydrous hydrazine solution not only made the process less toxic, but the film quality also improved with this change. The performance of prototype CuInSe2 solar cells demonstrated the potential to fabricate high-efficiency cells through processing optimization.

4.2.5 Selenization of the Sprayed CuInSe2 Solar Cells

One of the critical steps that lead to high-efficiency CuInSe2 solar cells is the high- temperature annealing process with selenium sources. The so-called selenization process sinters a porous thin film with nanocrystalline structures to form a compact film consisting of larger grains. This process can greatly reduce the processing defects and grain 62

boundaries and tune the intrinsic doping of the CIGS materials into a shallow p-type, improving the electrical properties of the solution-processed CuInSe2 thin films.

The selenization process is commonly performed with hydrogen chalcogenide gases

(H2S or H2Se).[54, 70] However, due to the high toxicity, it may be difficult to use these gases to scale up for commercial production. Since the hydrazine-based precursors are binary chalcogenides in the low dimensional structures, a strong reducing agent such as

H2Se is not necessary for driving the selenization reaction.[82, 84] Thus, Se power can be used to replace the highly toxic H2Se in the selenization process.

To study the effect of selenization temperature on the microstructure of the CuInSe2 films, we prepared a batch of the CuInSe2 absorber layers by fixing the films deposition temperature at 300 °C and varying the annealing temperature from 300 to 550 °C . Varying the selenization temperature can change the selenium activity levels, and thus, control the defect density and grain size in the CIGS films.[85] Figure 4-10 shows the SEM cross- sectional images of these films. The as-sprayed films exhibit nano-sized grains (Figure 4-

10a), while after annealing the grain size significantly increased (Figures 4-10b to 4-10g).

The degree of grain size enhancement is proportional to the annealing temperature. The porosity and non-uniform grain sizes are observed in the films annealed at a temperature below 500 °C . For the sample heated at 550 °C for 30 min, large and dense grains are observed within the CuInSe2 films (Figure 4-11).

Figure 4-10: SEM cross-sectional images of (a) the as-sprayed CuInSe2 thin films and

that after selenization at (b) 300 °C, (c) 350 °C , (d) 400 °C , (e) 450 °C, (f)

500 °C , and (g) 550 °C for 30 min.

Figure 4-11: SEM images of the sprayed CuInSe2 film after the 550 °C selenization.

The optimal selenization condition was used for the sprayed CuInSe2 device fabrication. For the device with the structure of ITO/i-ZnO/CdS/CuInSe2/Mo, the champion device exhibited an efficiency η = 7.2%, with VOC = 0.488 V, JSC = 28.6 mA/cm2,

FF = 51% (Figure 4-12). This result is still significantly below the 12% reported for the state-of-the-art hydrazine-processed CuIn(S,Se)2 solar cell,[82] but is comparable to or even better than other reports (2 to 8%) .[73, 74, 76, 86]

Figure 4-12: (a) J-V and (b) EQE curves of the champion sprayed CuInSe2 solar cell.

4.3 Sprayed CuInSe2 Solar Cells in the Back-wall Superstrate Configuration

The progress in spray pyrolysis using hydrazine-based precursor solution demonstrates the potential for using less toxic hydrazine hydrate solution in a low-cost, high-throughput manufacturing process for fabricating high efficient CuIn(S,Se)2 thin film solar cells. In addition to the solution processing, the device structure can also be modified to simplify the manufacturing processes and further lower the total manufacturing cost. Recently, an innovative back-wall superstrate configuration was reported that demonstrates the promise of fabricating ultrathin CIGS solar cells on ITO-coated glass.[87] This new structure possesses several manufacturing benefits. Decreasing absorber thickness leads to reduced materials use and shorter deposition times, both of which reduce manufacturing cost.

Furthermore, the superstrate structure enables light trapping by using textured transparent conductive oxide (TCO) contacts and the use of back reflectors. The cost could be further lowered by replacing ITO coated glass by the industrially available TEC glass.

Additionally, semi-transparent devices, finished with TCO instead of a back reflector, have the potential to be used as solar windows.

This section focuses on an investigation of the hydrazine-based spray pyrolysis deposition of CuIn(S,Se)2 thin films as the absorber layer in semi-transparent back-wall superstrate PV devices.[25] We fabricated and compared the performance of solar cells in the conventional substrate and the back-wall superstrate configurations. The superstrate cells employed ZnO:Al (AZO) and SnO2:F (FTO) as top and bottom transparent conducting contacts, respectively, which allows for investigation of the photovoltaic response from both sides of the device.

4.3.1 Fabrication of Semi-Transparent Backwall Superstrate Devices

The CuInSe2 absorber layer was fabricated using a hydrazine-based spray pyrolysis method we developed.[24] The hydrazine hydrate precursor solution was prepared by mixing 1 ml of 0.5 M Cu2S-S hydrazine solution, 2.5 ml of 0.25 M In2Se3–Se hydrazine solution, 0.6 ml of 2 M Se hydrazine solution, 2 ml ethylamine and 30 ml of deionized water, where the metal chalcogenide hydrazine solutions were prepared following the procedures developed at IBM.[17, 81, 82] The addition of water changed the precursor solution from pale yellow to dark brown and significantly increased the viscosity due to the aggregation of poly-chalcogenide species. The CuInSe2 films were fabricated by ultrasonically spraying the precursor solution onto Mo-coated glass or TEC 15 glass

(Pilkington) substrates heated to 300 °C inside a nitrogen-filled glove box. The sprayed films were then annealed with selenium powder in a closed graphite box at 500 °C for 15 min to remove the precursor residue and promote grain growth within the sprayed films.

Two solar cell device structures were investigated to compare the effectiveness of the spray pyrolysis method on different device configurations. The conventional substrate structure, as shown in Figure 4-13a, consists of a 700 nm DC-sputtered Mo layer on a soda- lime glass slide, an ultrasonically sprayed CuInSe2 absorber layer, a 120 nm chemical bath deposited CdS layer, a 50 nm sputtered intrinsic ZnO layer, and a 350 nm sputtered AZO layer. The sprayed CuInSe2 layer thickness was varied from 1.2 to 0.6 µm to study the effect of the absorber layer thickness on the solar cell performance. The thickness of the

CdS layer was intentionally doubled to prevent shunting associated with the porosity of the sprayed CuInSe2 thin films. Semi-transparent back-wall superstrate configuration devices, as shown in Figure 4-13b, were fabricated in a structure similar to that used by others.[87] 67

Here, a 30 nm MoO3 layer was thermally evaporated on a TEC 15 glass substrate to form a transparent Ohmic contact with p-type CuInSe2. The deposition of CdS and top TCOs were performed simultaneously with the substrate configuration devices to allow for direct comparison of the two structures.

Figure 4-13: Schematic diagrams of sprayed CuInSe2 solar cells in the (a) standard

substrate and (b) back-wall superstrate configurations.

4.3.2 Band Alignment of the Back-wall Superstrate Devices

To predict the differences in charge carrier transport, SCAPS modeling software [88] was used to calculate the band alignment of the substrate and back-wall superstrate CuInSe2 solar cell configurations at thermal equilibrium (Figure 4-14). Both structures have identical band structure at the CuInSe2/CdS heterojunction. The band bending near the p- n junction indicates the formation of a built-in electric field in the space charge region.

This region within the CuInSe2 absorber layer, typically around 0.3 μm for the

16 -3 heterojunction with the CuInSe2 acceptor density of 2 × 10 cm and the CdS donor density of 1 × 1017 cm-3, is critical for separating photogenerated electron-hole pairs.

Because of a relatively short minority carrier lifetime and a low carrier diffusion length in the family of CIGS based materials,[89] charge carriers generated far away from the depletion region are difficult to collect. Therefore, improving long wavelength photon collection by reducing defect density is required for high-efficiency devices.

Figure 4-14: Energy band diagrams of (a) substrate and (b) back-wall superstrate type

CuInSe2 solar cells at thermal equilibrium.

The difference between the band diagrams is at the CuInSe2/back contact interface. In the substrate configuration, CuInSe2 forms an Ohmic contact with Mo and thus charge carriers can freely inject into the back contact of the device. However, in the back-wall superstrate configuration, a p-n junction in the reverse direction would form if the p-type

CuInSe2 layer is in direct contact with the n-type TCO. As a result, the electron extraction

and back surface recombination due to this reverse junction would be detrimental to the performance of the device. To prevent the formation of this junction and block electron extraction, a MoO3-x electron blocking layer is deposited onto the TCO prior to the deposition of CuInSe2. As shown in the band diagram (Figure 4-14b), MoO3-x has a 1.9 eV conduction band offset and a -0.25 eV valance band offset with CuInSe2, which results in a high energy barrier on the conduction band that blocks electron flow at the MoO3- x/CuInSe2 interface. The Coulomb field formed by the accumulated electrons would compensate the band bending so that effective recombination is reduced, and, thus, high- performance devices become viable.

4.3.3 Microstructure and Composition of the Devices

To characterize the structural and compositional properties of the sprayed CuInSe2 cells, SEM and EDS measurements were performed on the completed back-wall superstrate devices. Figure 4-15a shows cross-sectional SEM image of a typical device.

Each layer of the device could be identified except for the MoO3-x layer which was too thin to image (~30 nm). The compact grains of the CuInSe2 absorber layer deposited by the spray pyrolysis approach columnar grains and the AZO and FTO on top and bottom contacts, respectively, can also be observed.

Figure 4-15b shows the compositional map of the device measured by EDS. Based on the results, the average stoichiometry of sprayed CuInSe2 thin films has a Cu to In ratio of

0.8 and a chalcogen to metal ratio of ~1. The presence of CdS, ZnO and SnO2 are observed, but the presence of MoO3-x layer cannot be verified because the Mo characteristic peaks

cannot be discriminated from those of the more prevalent S. Similar microstructure and composition are observed in all other devices.

Figure 4-15: (a) Cross-sectional SEM image and (b) EDS compositional mapping of a

typical back-wall superstrate device with a 600 nm sprayed CIS absorber.

4.3.4 Device Performance of the Sprayed Backwall Superstrate CuInSe2 Cells

To investigate the device performance of the sprayed CuInSe2 cells, semi-transparent back-wall superstrate and substrate solar cells using CuInSe2 absorber layer with thicknesses of 1.2, 0.9, and 0.6 μm were fabricated and compared. Devices with CuInSe2

layers of 0.3 μm showed no photovoltaic response due to the poor crystal quality and porosity. The J-V parameters of the successfully completed devices are plotted in Figure

4-16, where the green triangles and blue squares represent the back-wall superstrate cells illuminated through the bottom (TEC side) and the top (ZnO side), respectively, while the red circles correspond to the substrate devices.

Figure 4-16: J-V characteristics of sprayed CuInSe2 cells in the substrate configuration

and in the back-wall superstrate configuration when illuminated from the

top (AZO side) and the bottom (FTO side).

As shown in Figure 4-16 the back-wall superstrate devices exhibit a reduced performance as compared to the substrate configuration. Because photons enter the device through the TEC glass, a majority of photon absorption occurs at the back side of CuInSe2

absorber far away from the depletion region. Due to the absence of built-in electric field in this quasi-neutral region and small charge carrier diffusion lengths, photogenerated electron-hole pairs are not efficiently separated and collected before recombination, leading to poor performance devices. Therefore, it is necessary to use ultrathin CIS absorbers (< 400 nm) to ensure the charge carrier collection in back-wall superstrate devices. However, these absorber thicknesses are limited by the high porosity and low crystal quality of the ultrathin sprayed CIS films. Future optimization of spray pyrolysis and application of shunt passivation techniques should allow for better device performance for devices using these ultrathin absorbers.

On the other hand, when the back-wall superstrate cells are illuminated through the top

(ZnO side), they exhibit the comparable performance as the substrate devices. In this case, however, the device is fabricated on a transparent substrate. Here, the majority of electron- hole pairs are generated in the space charge region and have a greater probability of contributing to power conversion. As a result, the efficiencies were higher when the superstrate devices were illuminated through the top.

All the superstrate cells show a low Fill Factor (FF, ~30%), suggesting a non-ideal combination of high series resistance, shunt conductance, and perhaps the unfavorable reverse back junction. This is likely due to the adverse effects of the FTO/MoO3-x contact, although band alignment analysis does not reveal a limitation. The higher sheet resistance of FTO/MoO3-x (~500 Ω/sq.), compared with that of Mo (~0.1 Ω/sq.), hinders the carrier transport and lowers the device efficiency. Moreover, possible degradation of MoO3-x during the spray and/or selenization processes may also be detrimental to the devices.

Further investigation is needed to clarify and solve this issue. 73

The performance of sprayed CuInSe2 cells in both configurations decreases with reduced absorber thickness. This is likely due to the degraded electronic quality, incomplete crystalline packing and imperfect interface passivation which is commonly observed in thin CIGS devices.[90] The porosity and rough morphology of sprayed thin films likely exacerbate these problems.

Figure 4-17 shows the best devices we have fabricated consisting of 1.2 μm sprayed

CuInSe2 absorber layers. The best substrate device shows a 5.6% efficiency with 409 mV

2 open circuit voltage (VOC), 26.31 mA/cm short-circuit current density (JSC), and 52% FF.

The best back-wall superstrate device has a 1.2% efficiency with 282 mV VOC, 12.51

2 mA/cm JSC, and 34% FF. When illuminated through the top (ZnO side), the efficiency

2 improves to 2.7%, with mV VOC, 26.82 mA/cm JSC, and 33% FF.

Figure 4-17: J-V curves of the best sprayed CuInSe2 cells in the substrate and the back-

wall superstrate configurations.

Although the present power conversion efficiency does not show the overall advantage of the back-wall structure, further improvements of the electronic quality of the sprayed films would make the ultrathin (< 500 nm) back-wall superstrate devices possible using the spray pyrolysis method. The performance of back-wall superstrate devices with ultrathin CIGS absorbers (0.1 to 0.5 μm) has been demonstrated to be superior to the performance in the conventional substrate geometry.[87]

To investigate the charge carrier generation and collection, the external quantum efficiencies (EQE) of devices with various absorber thicknesses were measured. The EQE was converted into internal quantum efficiency after the reflectance (R) and transmittance

(T) correction. As shown in Figure 4-18, the control devices exhibit a strong response to high energy photons (~550 nm) which are absorbed within the depletion region. The inflection at ~500 nm is due to absorption in CdS (Eg = 2.4 eV), while the reduced response at longer wavelengths is due to incomplete absorption of low energy photons and carrier collection losses. The spectral response decrease of the sprayed cells is worse than that in the state-of-the-art devices, indicating a short minority carrier lifetime that is likely due to the poor crystallinity and high defect density in the spray deposited films. Those defects may originate from high-density grain boundaries, impurities, or non-stoichiometry phases formed during the pyrolysis or selenization processes.

The spectral responses of the back-wall superstrate cells are much lower than that of the control devices, mainly due to inefficient carrier collection near the CuInSe2/MoO3-x interface. The majority of photons are absorbed in this region without contributing to power conversion. The spectral responses are slightly better at long wavelengths (>900 nm) because those low energy photons with longer penetration depth can be absorbed near the 75

depletion region. To achieve high-efficiency back-wall superstrate devices, thinner and better quality CuInSe2 absorbers are needed. The efficiencies could be further boosted if a front reflector (e.g., Ag) was applied onto the top contact.

Figure 4-18: Internal Quantum efficiency of the superstrate devices (green), superstrate

devices illuminated through the top (blue), and control devices (red) for the

CuInSe2 thickness of (a) 1.2 μm, (b) 0.9 μm, and (c) 0.6 μm. 76

When the superstrate cells are illuminated through the top, photons are absorbed near the space charge region. For thicker devices, the spectra are similar to that of the control samples, which confirms that both devices have an identical band structure near the p-n heterojunction. The higher response at long wavelengths (900-1100 nm) due to higher transmittance indicates incomplete photon absorption. For thinner devices, the junction at the back of the device influences the length of the depletion regions, leading to reduced charge collection and QE at shorter wavelengths. Similar spectral response curves were reported by others for a bifacial thin film solar cell structure with ~2 μm CIGS absorbers.[90]

4.4 Summary and Outlook of the Sprayed CuInSe2 Solar Cells

To conclude, we have shown progress in developing the spray deposition technique for

CuInSe2-based thin film solar cells as evidenced by fabricating devices with decent efficiencies, growing knowledge of the materials and processes, and exploring the novel device structures. This is a good reason to be optimistic that the solution-processed CIGS solar cells with efficiencies greater than 20% will be achieved in a foreseeable future and the deposition techniques that are viable for large-scale, high-yield CIGS module manufacturing will be improved. Still, the failure of Nanosolar Inc. in the pilot production stage reminds us that there are fundamental issues of solution-processing techniques and the understanding of the CIGS materials to be addressed.

The research will continue to solve the critical problems associated with solution- processed thin films and to validate the processes. Clearly, the development of processing 77

techniques based on well-controlled and reproducible engineering models is critical to achieve reliable scale-up manufacturing. For the CIGS materials, considering the complexity of the basic ternary system (Cu, In and Se) as well as other involved elements

(Ga, S, Cd, Na, Mo, Zn, and O), it is challenging to deposit thin films with a uniform stoichiometry over a large area. Additionally, morphological control of solution-processed thin films is a key factor to improve device performance. A better control of precursor solution viscosity and wettability on designed interface materials is critical to achieve a smooth surface and thus a better performance.[91] Last, well-developed engineering is needed in the solution-based processes to achieve high efficient PV modules.

Addition to the solution-based processing for the thin film devices, new diagnostic and process-control tools designed for the solution-based processes are important to reliably predict the performance of the final devices. These tools are available for most vacuum- based technologies, while are yet to be developed for solution processing.

Finally, if the solution-processing of the CIGS solar cells cannot directly lead to higher throughput, yield, and performance due to the nature of the materials, new materials that possess the similar properties as the CIGS, including the high device efficiency, long term stability, good solution processability, and tolerance to materials and process variations should be taken into consideration.

Chapter 5

Emergence of Organic-Inorganic Lead Halide Perovskites: Literature Review

Organic-inorganic lead halide perovskite solar cells have been the focus of intense research over the past five years and power conversion efficiencies have rapidly been improved from 3.8% to 22%. This chapter reviews major advances in perovskite solar cells that have contributed to the efficiency enhancements, including the evolution of device architecture, the development of material deposition processes, and the advanced device engineering techniques aiming to improve control over morphology, crystallinity, composition, and the interface properties of the perovskite thin films. The challenges and future directions for perovskite solar cell research and development are also discussed.

This review focuses on the recent advances that have allowed perovskite PV to improve to efficiencies greater than 20% in the time period of 2013-2015. Reviews that discuss early developments and the physical properties of materials can be found elsewhere.[92-

97] Here we mainly focus on the key factors that govern the device performance, including device architecture, preparation methods, and advanced device engineering. The challenges and issues for perovskite PV devices are also discussed.

5.1 Background of Perovskite Solar Cells

To gain market share from crystal silicon solar cells, alternative PV technologies have to provide a desirable combination of high power conversion efficiency, low manufacturing costs, and excellent stability. Recent research suggests that inorganic- organic halide perovskite solar cells (PSCs), with CH3NH3PbI3 (methylammonium lead iodide, or MAPbI3) being the prototypical example, have the potential to meet these conditions and become competitive in the marketplace. As a result of intensive research efforts across the globe over the past five years,[92-101] power conversion efficiencies of

PSCs are now comparable to other thin-film PV technologies (e.g., CIGS and CdTe), and the simple device processing promises lower manufacturing costs, suggesting the potential to challenge the prevailing silicon technology in the foreseeable future.[102]

The term perovskite refers to the crystal structure of calcium titanate (CaTiO3) which was discovered by German mineralogist Gustav Rose in 1839 and named in honor of

Russian mineralogist Lev Perovski.[103] In the field of optoelectronics, perovskites are a

+ + group of materials having the formula AMX3 where A is an organic cation (CH3NH3 ,

+ + 2+ 2+ NH2CH3NH2 , or Cs ), M is a divalent metal cation (Pb or Sn ), and X is monovalent halide anion (I-, Br-, or Cl-). Figure 5-1 shows the crystal structure and a single crystal of

+ MAPbI3. In a unit cell of the perovskite structure, eight A cations are located at the vertices of a cubic cage, an M+ cation is located at the center of the cube, and the latter species is octahedrally coordinated to six X- species that sit at the cube’s faces. The perovskite family of materials was studied in the 1990s due to their excellent optoelectronic properties and potential for solution-processed fabrication,[104-107] but the main goal of this early work

was to develop new materials for field effect transistors and organic light emitting diodes.[108, 109]

The first known use of perovskite was in a solar cell that reported a 3% PCE in

2009.[110] However, this PSC contained a liquid electrolyte and received little attention due to the low efficiency and poor stability. So-called perovskite fever[111] did not fully bloom until a solid state cell was developed and devices with ~10% efficiency were reported in 2012.[20, 21] Since then, PSC device performance has rapidly progressed and the best efficiency record of over 22.1% was achieved in 2016.[22] The pace of progress has been remarkable and unprecedented in PV history and is likely attributed to several factors related to inexpensive fabrication costs, ease of processing, and the excellent optoelectronic properties of the materials.[15, 92-97]

Figure 5-1: (a) Crystal structure of CH3NH3PbI3 perovskite. (b) Photo of CH3NH3PbI3

perovskite single crystal synthesized in our lab using the inverse temperature

crystallization method.[112]

As will be described in Section 3, high-quality perovskite thin films can be fabricated using a variety of processes including solution-[20, 113] and vapor-based[114-116] deposition methods. Many of these methods are compatible with low-cost, large-scale, industrial production techniques, which strengthens the potential for the commercialization of perovskite solar cells. Due to the ease of processing, many research groups from around the world have been attracted to work in the area. This includes groups that have past histories and relevant expertise in dye-sensitized solar cells (DSSC), organic photovoltaics

(OPV), and solution processing. Consequently, the learning curve for developing perovskite solar cells has been relatively short and progress has been very rapid.

In addition to flexibility in processing, the perovskite materials possess several outstanding optoelectronic properties that make them ideal choices for PV applications.

The 1.55 eV band gap of MAPbI3 is nearly ideal for single junction solar cells exposed to the solar irradiance spectrum, and it can be continuously varied in the range from 1.5 to 2.3 eV by exchanging the organic and halide ions.[117, 118] The optical absorption coefficient of MAPbI3 is higher than other PV materials such as Si, CdTe, CuGaxIn1-xSySe1-y, and amorphous Si:H, so the absorber thickness can be reduced to ~300 nm, thereby lowering the materials costs.[119, 120] In contrast to organic PV materials, the low exciton binding energy (30 to 50 meV) allows spontaneous exciton dissociation into free charges after light absorption.[121-123] Moreover, the high electron and hole mobility in the range of 10 to

60 cm2V-1s-1 and the long carrier lifetime (~ 100 ns) results in long diffusion lengths (~1

µm) so that charge carriers can be freely transported across the 300 nm thick perovskite absorber before recombination.[124-127] Finally, because the electronic defects are

shallow and relatively benign, the non-radiative recombination rates are low, allowing open-circuit voltages greater than 1 V can be achieved.[128, 129]

Although PSCs show great potential, there are several challenges that need to be addressed before commercialization will be possible. Perhaps most significantly, perovskites have not yet demonstrated the long-term stability that is necessary to compete with the 30 year lifetime of commercially available Si and CdTe solar panels. Secondly, there are questions about the current-voltage (I-V) hysteresis during voltage scanning, which could be problematic for large-scale deployment. There are also concerns associated with potential environmental impacts due to the fact that perovskites contain Pb.

5.2 Evolution of Device Architectures

The first perovskites materials employed in PV were used as direct replacements for the dye sensitizers in the DSSCs.[110, 130] The typical DSSC structure employs a several micron thick porous TiO2 layer that is coated and penetrated with an absorber dye material.

The electrode assembly is contacted by a liquid electrolyte containing a redox couple.[131]

In these devices, the TiO2 is used to collect and transport the electrons, while the electrolyte acts as a hole conductor. The original perovskite solar cells evolved from this same structure with the OHMP materials acting simply as a dye replacement.[20, 21] Interest increased when the so-called mesoscopic device structure (Figure 5-2a) was formed by replacing the liquid electrolyte with a solid-state hole conductor.[20, 21] This advance created great interest in the PV community and drew in experts from the thin-film PV and

OPV communities. As a result, planar device structures in which the perovskite layer is sandwiched between electron and hole transporting materials (ETM & HTM) were 83

developed. Depending on which transport material is encountered by the light first, these planar structures can be categorized as either the conventional n-i-p (Figure 5-2b) or the inverted p-i-n (Figure 5-2c) structures. Recently, a mesoscopic p-i-n structure (Figure 5-

2d) has also been developed.[132, 133] Due to processing differences, the device architecture determines the choice of charge transport (ETM & HTM) and collection

(cathode & anode) materials, the corresponding material preparation methods, and, consequently, the performance of the devices. To date, no perovskite devices with significant efficiency have been constructed on opaque substrates because the conventional deposition technologies for transparent conducting oxide (TCO) may lead to decomposition on the surface of perovskite.

Figure 5-2: Schematic diagrams of perovskite solar cells in the (a) n-i-p mesoscopic, (b)

n-i-p planar, (c) p-i-n planar, and (d) p-i-n mesoscopic structures.

5.2.1 Conventional “n-i-p” Structure

The mesoscopic n-i-p structure is the original architecture of the perovskite PV devices and is still widely used to fabricate high-performance devices. The structure (Figure 5-2a)

consists of a transparent conducting oxide cathode, a 50 to 70 nm thick compact electron transporting layer (typically TiO2), a 150 to 300 nm thick mesoporous metal oxide (mp-

TiO2 or mp-Al2O3) that is filled with perovskites, followed by a 300 nm perovskite capping layer, a 150 to 200 nm thick layer of 2,2’,7,7’-tetrakis(N,N-di-p-methoxyphenylamine)-

9,9’-spirobifluorene (Spiro-MeOTAD) which is a hole conductor, and 50 to 100 nm of a metal anode (Au or Ag).

In this structure, the mesoscopic layer is thought to enhance charge collection by decreasing the carrier transport distance, prevent direct current leakage between the two selective contacts, and increase photon absorption due to light scattering. Accordingly, the original mesoscopic perovskite devices used a thick (> 500 nm) porous layer to efficiently absorb the incident light.[21, 134] But because the grain growth of the perovskites is confined by the pores in the structure, a significant amount of the material is present in disordered and amorphous phases.[135] This leads to relatively low open-circuit voltage

(VOC) and short-circuit current density (JSC).[136] Surprisingly thinning the mesoporous layer to around 150 to 200 nm results in improved device efficiency due to enhanced crystallinity in the perovskite absorber. Additionally, the pore filling fraction and morphology of the perovskites is critically dependent upon the thickness of the mp-

TiO2.[137, 138] When the porous layer thickness is reduced to less than 300 nm, the pore filling fraction is increased and a perovskite capping layer forms on the top of the porous structure. Complete pore filling accompanied by the formation of a capping layer assures high charge transport rates and high collection efficiencies at the TiO2 interface. Once the charges are separated, recombination pathways between electrons in the TiO2 and holes in the HTM are blocked due to the relative positions in the energy of the respective conduction 85

and valance bands (vide infra).[137] The present record efficiency value (20.2%) was measured from a cell formed in the mesoscopic structure that had discrete perovskite nanocrystals embedded in the porous ETL film with an overlaying continuous and dense perovskite capping layer. Consequently, the meso-n-i-p structure is the most popular structure reported in the literature.[139] Interestingly, the mesoscopic structure exhibits a reduced J-V hysteresis in comparison to other structures, as will be discussed in Section

5.6.

The planar n-i-p structure (Figure 5-2b) is the natural evolution of the mesoscopic structure as the mesoporous layer thickness goes to zero. The mesoporous TiO2 layer was initially considered to be critical for high-efficiency perovskite devices,[140] however, more recent results demonstrate that high efficiencies can be achieved without a mesoporous layer due to long diffusion lengths for the charge carriers.[114, 124, 141]

Omitting the mesoporous layer simplifies the device structure, allows a variety of perovskite deposition methods to be used, and widens the choices for the ETM and HTM materials. To date, the best planar n-i-p device showed a 19.3% efficiency after the

ITO/TiO2 interface of the electron selective was engineered by polyethyleneimine ethoxylated (PEIE) and Yttrium.[141] Although the planar n-i-p perovskite solar cell usually exhibits enhanced VOC and JSC relative to a comparative mesoscopic device processed with the same materials and approach, the planar device usually exhibits more severe J-V hysteresis (see Section 5.6).[142] Thus, the state-of-the-art n-i-p devices usually include a thin (~150 nm) mesoporous buffer layer filled and capped with the perovskite.[139]

5.2.2 Inverted “p-i-n” Structure

When the deposition order is changed and the HTM layer is deposited first, the device is fabricated in the p-i-n structure (Figure 5-2c). In this case, the p-i-n type perovskite device is built on a 50 to 80 nm p-type conducting polymer such as poly(3,4- ethylenedioxythiophene) poly(styrene-sulfonate) (PEDOT:PSS) which is deposited on indium tin oxide (ITO) coated substrates. After depositing a 300 nm intrinsic perovskite thin film, the device is completed with a 10 to 60 nm organic hole-blocking layer [6,6]- phenyl C61-butyric acid methyl ester (PCBM), and a metal cathode (Al or Au). Early device design utilized a perovskite and fullerene (C60) donor-acceptor pair, which is typical in

OPV.[143] In fact, the commonality in structure has allowed OPV researchers to easily move into the field of perovskites. As the field has advanced, the organic acceptor has been omitted in favor of an ETM layer, leaving the planar perovskite absorber sandwiched between two opposite organic charge transporting materials.[144] Recently, the efficiency of the planar p-i-n device has improved significantly due to the use of more advanced material preparation methods such as a multi-cycle solution coating process and the best efficiency of 18.9% was achieved.[145]

Further development of the p-i-n device structure has expanded the selective contact options from organic to inorganic materials. For example, NiO and ZnO/TiO2 layers have recently been used for the hole and electron selective contacts, respectively, which makes the perovskite device distinct from its organic counterpart.[146, 147] Inorganic charge

2 extraction layers (NiMgLiO and TiNbO2) have been used to fabricate large-area (1 cm ) high efficiency (15%) perovskite cells, representing a potentially important step in the path toward commercialization.[147] The use of oxide HTMs also allows for construction of 87

the mesoscopic p-i-n device structure, in which NiO/mp-Al2O3 or c-NiO/mp-NiO are used as the HTM.[132, 133] The best mesoscopic p-i-n device with a nano-structured NiO film demonstrated a 17.3% efficiency.[148]

5.3 Preparation Methods

The device performance of most thin-film solar cells is mainly determined by the film quality of the absorber. High-quality perovskite films with appropriate morphology, uniformity, phase purity, and crystallinity are essential for high-performance PV device.

To meet these quality criteria, well-controlled crystallization and engineering of the composition and interface properties of perovskite films are required. Critical issues include the deposition approach, precursor composition, processing condition, and additive control, all of which can greatly affect the crystallization and quality of the perovskite films. Focusing first on the deposition approach, the preparation processes can be categorized as follows: single-step solution deposition,[20] two-step solution deposition,[113] two-step vapor-assisted deposition,[115] and thermal vapor deposition.[114]

5.3.1 Single-Step Solution Deposition

Single-step solution deposition (Figure 5-3a) is commonly used for perovskite thin film preparation due to ease of processing and low fabrication cost. Generally, organic halides

(methylammonium iodide, MAI) and lead halides (PbX2, X= I, Br, or Cl) are dissolved in gamma-butyrolactone (GBL), dimethylformamide (DMF) or dimethyl sulfoxide (DMSO) to prepare the precursor solution. The perovskite films can be prepared by spin-coating of 88

the precursor solution followed by a post-deposition heating at 100 to 150 °C. Since the perovskite tolerates composition variation,[149] high-efficiency devices can be fabricated through a wide range of MAI to PbI2 precursor ratio from MAI-poor (1:2)[150] to MAI- rich (3:1).[141] However, it is critical to choose appropriate processing temperatures and times based on differing precursor compositions to achieve the desired crystallinity, phase, and morphology of the perovskite films.[138, 149, 151] Besides the choice of precursor composition and processing temperature, the environment (oxygen and humidity levels), substrate material, and deposition parameters must also be controlled. The first solid state device prepared using the single-step solution process produced a perovskite device that exhibited 9.7% efficiency.[152] After developing advanced engineering techniques

(discussed in Section 4), the best efficiency of 19.7% has been achieved with single-step solution deposition.[153]

Figure 5-3: Deposition methods for perovskite thin films, including (a) single-step

solution deposition, (b) two-step solution deposition, (c) two-step hybrid

deposition, and (d) thermal vapor deposition.

In addition to spin coating, other solution-based deposition methods, including spray,[154] doctor-blade,[155] inkjet printing,[156] and slot-die printing[157] have also been employed to fabricate perovskite PV devices. These techniques demonstrate the potential for large-scale roll-to-roll manufacture of perovskite solar cells. However, the efficiency of devices prepared by these methods are still lower than that of spin-coated devices due to the difficulties associated with controlling the film morphology and compositional uniformity at present.

5.3.2 Two-Step Solution Deposition

The two-step solution deposition approach to prepare perovskites was first introduced by Mitzi et al. in 1998.[158] Following this pioneering work, Gratzel et al. developed a sequential deposition method (Figure 5-3b) to prepare perovskite solar cells that have resulted in efficiencies greater than 15%.[113] In a typical two-step solution procedure, a

PbI2 seed layer is spin-coated then converted to MAPbI3 by dipping the substrate into an

MAI/isopropanol solution.[113] Spin coating has also been used to introduce MAI molecules into the PbI2 network.[152] Compared with the single-step solution process, the two-step sequential deposition process results in more uniform and dense perovskite films.

The process can be well controlled and, consequently, has been extensively used to fabricate high-efficiency devices.[113, 139, 159, 160]

The two-step solution method provides a reproducible way to fabricate high-quality perovskite thin films. Through varying the MAI solution concentration, the perovskite grain size can be controlled.[152] However, one of the drawbacks of the two-step solution deposition method is the trade-off between perovskite grain size and surface smoothness. 90

Films with large perovskite grains typically exhibit poor surface coverage which can limit the performance of devices. The other issue with this method is incomplete perovskite conversion. The conversion from PbI2 to MAPbI3 rapidly occurs as the film is dipped into the solution because the layered structure of heavy metal halide is prone to the interaction with small molecules.[161] Thus, a dense perovskite capping layer usually forms on the surface of PbI2 and hinders the MAI diffusion to the underlying layer, leading to the incomplete perovskite conversion. These issues have been overcome by some new techniques that have been developed recently (Section 5.4), and now the champion cell efficiency using the two-step solution method has been improved to 20.2%.[139]

5.3.3 Vapor-Assisted Solution Deposition

In one modification to the two-step solution deposition method, MAI is introduced through a vapor deposition technique rather than through solution processing (Figure 5-

3c).[115] This deposition method allows better control of morphology and grain size via gas-solid crystallization and effectively avoids film delamination that can occur during the liquid-solid interaction. The perovskite films prepared by this method exhibits uniform surface coverage, large grain size, and full conversion. However, the use of this method is limited because the gas-solid reaction typically required tens of hours for the full conversion and devices prepared by this method exhibited only 10-12% efficiency.[115,

162]

5.3.4 Thermal Vapor Deposition

Vapor phase deposition is widely used for fabricating high-quality semiconductor thin films with uniform thickness and composition. The thermal vapor deposition of perovskite thin films was first demonstrated by Mitzi et al. in 1999.[105] After modifying the technique for dual-source thermal evaporation (Figure 5-3d), Snaith et al. prepared the first planar heterojunction MAPbI3-xClx perovskite solar cell with an efficiency that exceeded

15%.[114] Similar vapor-based deposition techniques such as sequential layer-by-layer vacuum sublimation[116] and chemical vapor deposition[163] have also been developed.

The perovskite films prepared by thermal vapor deposition are extremely uniform and pinhole-free. Compared with the incomplete surface coverage that can be found for perovskite films prepared by solution processing, vapor deposited perovskite layers can conformally coat TiO2 and PEDOT:PSS layers.[114, 144, 164] However, both the precursor sources and the products have low thermal stability, so the vapor deposition requires precise control over temperatures during deposition. Thus, only a few research groups have reported high-efficiency devices prepared by this method.[114, 116, 144, 164,

165]

5.4 Advanced Engineering Techniques

In early 2013, the state-of-the-art of perovskite solar cells prepared by various deposition techniques had demonstrated device efficiencies in the range of 12-15%.[113-

115] Since then, the efficiency has improved to 18-20%, mainly due to the advancements of several device engineering strategies.[96, 166, 167] These engineering strategies, which focused on controlling the precursor solution, processing condition, perovskite 92

composition, and interface properties, lead to smooth and pinhole-free perovskite thin films consisting of large grain with good crystallinity. The combination of these advanced engineering methods have improved the optoelectronic properties of the perovskite films, and, consequently, the device performance as well.

5.4.1 Solvent Engineering

The single-step spin coating is the simplest method for preparing perovskite thin films; however, it is difficult to achieve a homogeneous composition and uniform thickness over large areas. The reason for this is that single-step solution deposition using DMF and GBL solvents often results in the formation of the needle and spherical shaped colloidal intermediates.[113, 168] To improve the surface morphology of spin-coated perovskite films, several precursor solution additives have been employed to suppress the formation of deleterious intermediates.

Dimethyl sulfoxide (DMSO) is one of the best and widely used additives.[160, 169]

The precursor solution with added DMSO forms a uniform and flat MAI-PbI2-DMSO intermediate film when spin-coated. After a thermal treatment, the intermediate film is converted into a uniform perovskite film through a solid-state reaction. Several other additives, such as MACl,[170] HI,[171] I2,[172] NH4Cl,[173] H2O/HBr,[174] 1,8- diiodooctane (DIO),[175] aminovaleric acid,[176] and phosphonic acid ammonium[177] have also been used to improve the crystallinity and morphological uniformity of perovskite films.

The formation of uniform perovskite film by incorporating additives is the result of decoupling the nucleation and grain growth processes. For precursor solutions without 93

additives, these two processes occur simultaneously. Since grain growth favors large size nuclei (free energy of volume expansion eclipses that of interface formation), the unbalanced growth rate leads to the formation of large perovskite grains with a significant number of voids between grains. The introduction of additives retards the crystallization kinetics of perovskite formation and results in a uniform intermediate phase film during deposition. A thermal treatment provides the energy for conversion to the perovskite phase and promotes crystal growth to form pinhole-free films.

Additive incorporation was introduced to the two-step methods after its success in the single-step deposition. The precursor solution for PbI2 can be mixed with DMSO,[178]

H2O,[179] and low concentrations of MAI[159] to improve the surface coverage of the final perovskite film. As with the single-step deposition, the introduction of additives results in an intermediate state that retards the rapid reaction between MAI and PbI2 and effectively avoids the formation of a dense perovskite capping layer on the surface of the

PbI2 layer that hinders further conversion.

5.4.2 Process Engineering

In addition to modifying the precursor solution, improved device performance has been achieved by adapting the deposition and post-deposition processes. While slowing the crystal growth kinetics have resulted in higher quality films, the same results have been obtained by speeding the nucleation kinetics. Hot casting, in which crystallization of the perovskite film occurs immediately after a hot precursor solution is loaded onto the substrate at an elevated temperature, has been used to obtain pinhole-free perovskite films with millimeter scale grains.[180] Using this approach, the island-shaped grains rapidly 94

integrate into a dense perovskite film with millimeter-size grains following Volmer-Weber growth.[181] Devices with the efficiency of ~ 18% were fabricated using this technique.[180]

Another demonstration of process engineering for fabricating extremely uniform and dense perovskite films is adding an anti-solvent that does not dissolve perovskite films

(e.g., toluene) during the last few seconds of the spin process.[160] The introduction of toluene rapidly extracts DMF from the precursor solution, which results in a rapid precipitation of perovskite before the significant growth of the perovskite grains. Thus, a dense, small grain perovskite film can form uniformly across the entire substrate surface.

In addition to toluene, other anti-solvents, such as diethyl ether,[153] chlorobenzene, benzene, and xylene are also effective in forming highly uniform perovskite films.[182]

Since the grain growth kinetics are suppressed during the deposition, this process needs an optimized thermal annealing to achieve both smooth morphology and large grain size.

The post deposition grain growth process can also be engineered to achieve a uniform and high-quality perovskite film. Although thermal annealing helps increase grain size and improve crystallinity, it may cause decomposition of the perovskite phase and reduce surface coverage.[138, 149] Solvent annealing with DMF leads to recrystallization and regrowth of perovskite grains, resulting in improved crystallinity and electronic properties and enhanced device efficiency (15.6%).[183] Annealing with pyridine or MAI vapor has demonstrated enhanced luminescence and carrier lifetimes, indicating the formation of high-quality absorber material and the potential for high-efficiency devices.[184, 185]

Figure 5-4: SEM images of perovskite films prepared using various deposition techniques

and advanced engineering processes, including (a) vapor deposition, (b)

vapor-assisted deposition, (c) hot casting, (d) H2OH2O additive, (e) DMSO +

toluene, (f) chlorobenzene, (g) sequential deposition, and (h) solvent

of Photo Optical Instrumentation Engineers (SPIE).

The advanced solvent and process engineering techniques both aim to decouple the nucleation and growth processes so that the perovskite film formation can be precisely controlled. By applying one or a combination of these techniques, high-quality perovskite films with smooth morphology and large grains were prepared (Figure 5-4) and the devices with efficiencies of 15 to 19% were fabricated.

5.4.3 Composition Engineering

The tunability of the perovskite bandgap over a wide range of the solar spectrum has led to considerable improvement in device performance. Compositional engineering of

MAPbI3 perovskite can be achieved by exchanging the organic (formamidinium, FA), metal (Sn), or halide (Br or Cl) ions. The bandgap of the perovskite can be controllably tuned to cover almost the entire visible spectrum from 1.5 to 2.3 eV by introducing mixed halides (I and Br).[117, 118] The introduction of Br also enhances the water resistance of the perovskites.[117] Partial replacement of MA by FA for the alloyed MAxFA1-xPbI3 is an effective way to extend the absorption to longer wavelengths and enhance the thermal stability.[186] With the (FAPbI3)1–x(MAPbBr3)x absorber, solar cells with over 19% average efficiency have been fabricated with a high degree of reproducibility.[139]

In addition to varying the halide anions and the organic cations, the divalent metal cation may also be changed. Due to the concerns for lead toxicity, lead-free perovskites such as MASnX3 have attracted increasing attention.[187-189] The relatively lower bandgap of MASnI3 (~1.3 eV) allows absorption over a broader range and a higher JSC values up to 21 mA/cm2.[190] The efficiency (~ 6%) and stability of the lead-free perovskite-based device, though, are not currently comparable to their Pb-based counterparts at this time.

5.4.4 Contact Engineering

In addition to controlling and modifying the optoelectronic and structural properties of the absorber, as discussed above, the properties of the electron and hole collecting electrodes and their interfaces are also critical for improving perovskite device 97

performance. The importance of the interface properties has been revealed by electron beam induced current (EBIC) investigations, which show that efficient charge separation and collection occurs at the interfaces between the perovskite and both charge-selective layers.[191] The choice of the ETM and HTM is important to achieve a high degree of charge selectivity while maintaining a low surface recombination to minimize energy loss at the heterojunction interfaces. Recently, a variety of ETMs and HTMs have been explored for achieving high-efficiency perovskite devices. Figure 5-5 plots the energy levels for several representative components of some of the most common perovskite solar cells.

Figure 5-5: Diagram showing the energy levels, from left to right, for the representative

cathode, n-type (ETM), absorber, p-type (HTM), and anode materials.

Metal oxides are the most common ETMs. While TiO2 is predominant in the literature, many other materials can operate as either mesoporous or planar ETMs. Wide band gap metal oxides such as ZnO,[192] Al2O3,[20, 193] SrTiO3,[194] SiO2 and ZrO2[176, 195] 98

have been used to fabricate devices in the mesoscopic structure. An electrically insulating mesoporous layer allows high open circuit voltages (VOC) to be achieved if there is a lack of sub-band gap and surface electronic states.[20] A variety of ETMs have also been used to form compact layers in the planar n-i-p structure including ZnO,[196] SnO2,[197, 198]

CdSe,[199] CdS,[200] and TiO2-graphene.[201] Among them, SnO2 has been used to fabricate an 18% efficiency device presumably due to good band alignment.[198]

Commonly used HTMs fall into three categories: small molecules, organic polymers, and inorganics. Small molecules, especially spiro-MeOTAD, are very commonly used as the HTM in high-efficiency perovskite PV devices. The conductive organic polymer, poly(triarylamine) (PTAA), has recently emerged as a strong competitor to spiro-

MeOTAD and was employed in the highest efficiency perovskite device.[139] Poly(3- hexylthiophene-2,5-diyl) (P3HT) and other organic molecules and polymers have also been used to fabricate 12 -15% efficiency perovskite devices. A detailed review of HTMs can be found elsewhere.[202] Organic HTMs are typically doped with lithiumbis(trifluoromethanesulfonyl)imide (Li-TFSI) and 4-tertbutylpyridine (tBP) to improve hole conductivity, doping uniformity and device performance. Although these organic HTMs provide good carrier transport properties which lead to high performance, high materials costs and unproven long-term stability are major impediments to industrial application. In contrast, inorganic HTMs, such as CuSCN,[203, 204] CuI,[205] NiO[206], and Cu:NiOx[207] are promising for more cost-effective and stable performance. The highest efficiency reported to date for an inorganic material was 17.7% with Cu:NiOx as the HTM.[207]

It should be noted that HTM-free and ETM-free designs have also been attracted attention. The HTM is not a prerequisite for the perovskite solar cells when a high-quality perovskite layer with benign interface properties is presented. A high work function metal

(Au or C) may help to extract holes from the perovskite absorber alone. Several groups have demonstrated the HTM-free perovskite solar cells with efficiencies ranging from 5 -

12%.[162, 176, 208-210] In addition to the HTM-free devices, ETM-free devices with efficiencies of ~14% have been reported.[211, 212] In these devices, the interface properties of the TCO cathode were modified, and the perovskite solar cells were grown directly on FTO substrates without an ETM layer. Although the performance of these devices is inferior to the state-of-the-art perovskite devices due to a poor charge extraction and undesired surface recombination at the interface, the designs help improve the understanding of the device physics.

5.5 Efficiency Improvement in Recent Years

By combining a variety of advanced techniques, several research groups have achieved high-performance perovskite PV devices with efficiencies greater than 17% during the last two years. Over the past three years, the device performance has improved from 10%.

Figure 5-6 shows a mapping of the critical device performance metrics. It is clear that the evolution of the state-of-the-art perovskite device efficiency was achieved by the enhancements in JSC, VOC, and FF. Improvements in the 2012 – 2013 timeframe can be attributed to the developments of absorber preparation techniques that led to better morphology and surface coverage and contact engineering techniques that promised efficient charge separation and collection, and, consequently, higher photogenerated 100

current density (Figure 5-6). The device efficiency over this time frame was improved from

10 to ~15% by optimizing the basic processing of the perovskite absorber.

All high-performance perovskite devices share some common characteristics. First, a high JSC is typically the result of a dense and uniform perovskite film with appropriate thickness, good crystallinity, and large grain size. The high VOC is enabled by reducing intergrain and intragrain defect densities and good interface properties between the perovskite and the selective charge collectors. Fill factors are typically very high, with many devices having FFs in the range of 0.75 to 0.80. Additionally, compositional engineering of the perovskite absorber contributes to the better device performance.

Incorporating FA extends the absorption range to wavelengths longer than 800 nm and

2 hence enhances the JSC by ~4 mA/cm . The introduction of Br, on the other hand, increases the bandgap of the perovskites and reduces defect density and thus improve the VOC to around 1.1 V.

The highest efficiency devices typically employ a combination of several advanced engineering techniques. For instance, the champion 20.2% device was prepared by the intermolecular exchange process involving the reaction between the PbI2-DMSO intermediate phases and the FAI-MABr contained solution.[139] An extremely uniform and dense perovskite film was formed after annealing, and the device exhibited excellent performance. Other devices with over 18% efficiency are fabricated by spin-coating of the mixed PbI2-FAI-PbBr2-MABr precursor in the DMF/DMSO solution followed by anti- solvent quenching.[153, 198, 213]

101

Figure 5-6: Efficiency mapping of recently reported the state-of-the-art perovskite solar

cells labeled with reference number (as in Ref. [15]) and colored based on

efficiency.

5.6 Issues and Challenges

Perovskite solar cells have demonstrated high efficiency and are being investigated as a viable commercial option. However, crucial issues and challenges that limited the commercialization of perovskite-based PV remain. Long-term device stability during operation under stressed conditions (high humidity, elevated temperature, and intense illumination) have yet to be demonstrated. The existence of the J-V hysteresis limits the standardized characterization of device performance. Environmental impacts during the manufacturing, operational, and disposal phases of perovskite solar cells are unclear, leaving concerns of the toxicity and contamination associated with the water-soluble lead

102

compounds. Although the complexity of the diverse material preparation methods and device architectures makes it more difficult to address these issues, recent progress has provided insights into these issues and corresponding material properties.

5.6.1 Device Stability

One of the most important criteria for a practical solar cell is that the cell has to maintain a stable power output under a standard working condition. At present, the efficiency of perovskite devices is determined by the average of the forward and reverse scans or the steady-state power output close to the maximum power point (MPP). Although J-V hysteresis may exist, the current output of most perovskite devices quickly stabilizes at the maximum power point. Such steady state output shows the potential for the sustainable power generation and is now accepted as one of the criteria to characterize perovskite PV devices.[214] As a result, MPP tracking is used as a reasonable way to report stabilized efficiencies when characterizing devices with hysteresis.

Although stability data up to few hundred hours (1 - 4 months) has been reported,[21,

136, 176, 214] long-term stability that is comparable to the 30-year standard of commercial

PV panels has yet to be demonstrated. Early perovskite devices without encapsulation have shown stable operation up to hundreds of hours when stored in dark and measured infrequently. However, these devices rapidly degraded after sustained exposure to sunlight.[96] In addition to light exposure, elevated temperature and humidity may accelerate the degradation due to the moisture induced decomposition of perovskite crystals.[215] These stability issues, though, are being addressed by, for example, proper protective coatings. The stability of perovskite PV devices under high humidity and 103

temperature conditions was improved by employing a moisture resistant layer (e.g. carbon nanotubes or graphite) to prevent water ingress.[32, 176, 216, 217] Encapsulation techniques using glass sealing or laminate plastic films have also used to improve device stability to over 3000 hours at 60 °C under simulated sunlight.[218] Additionally, when

+ - incorporating larger ions (FA and Br ) into perovskite to form (FA1-yMAy)Pb(I3-xBrx), the thermal and moisture resistivity can be dramatically improved.[117, 118, 213] These results indicate that perovskite PV modules with appropriate composition and encapsulation have the potential to be stable. It should be noted that good stability of perovskite PV devices has recently been demonstrated under a hot outdoor condition in

Jeddah, Saudi Arabia.[217] After 3 months of operation at 80 °C the perovskite devices demonstrated impressively stable performance without measurable degradation.

5.6.2 Current-Voltage Hysteresis

One of the major issues that limit advancement of perovskite solar cells is the presence of the anomalous J-V hysteresis which is observed by varying the direction and the rate of voltage sweep (Figure 5-7a).[219] Holding a perovskite device at a forward bias voltage before measurement may result in a higher efficiency than that found when the device is held at the maximum power point or when the device has been reverse biased or held at the short circuit.[220] Measuring the device at a rate faster than its response time may also result in varying efficiency measurements.[221] This phenomenon undermines the accuracy of efficiency measurement and may lead to inflated and misleading in the literature.

104

Parameters affecting J-V hysteresis of perovskite solar cells have been investigated,[142, 222] however, the origins of hysteresis remain controversial. Three possible reasons, including ferroelectricity,[223, 224] ion migration,[220, 225] and unbalanced charge collection rates[226, 227] have been proposed to explain the origin of the J-V hysteresis. All of these hypotheses are related to a transient electrical polarization as a respond to the change of the external electrical field. Recent investigations have shown that the latter two are more likely the causes of the J-V hysteresis.[214, 228]

Ferroelectricity is one possible, but unlikely, origin for the I-V hysteresis.

Ferroelectricity may occur in organo-metal perovskites due to the shift of ions in the crystal away from their corresponding lattice point or the alignment of organic dipole moments.

Evidence of this was observed in polarization loops of MAPbI3 thin films and piezoelectric force microscopy.[229, 230] However, recent reports indicate that perovskites are not ferroelectric at room temperature (Figure 5-7b) and that the observed ferroelectric behavior is likely due to piezoelectric or electrochemical behavior.[231, 232]

Ion migration is another possible explanation of the J-V hysteresis. Under an external electric field, the positive and negative ionic species will migrate to the opposite sides of the device, forming space charge regions closed to the interfaces. Accumulation of the mobile ions changes the density of free electronic charge carriers and thus shifts the local quasi-Fermi level in the direction that is favorable (or unfavorable) to charge extraction under positive (or negative) bias (Figure 5-7c). Such ion migration has also been demonstrated in polarization-switchable perovskite devices, in which photocurrent direction could be switched by changing the voltage sweep direction.[233] Recent modeling work revealed that the ion migration is accompanied by the charge traps serving 105

as recombination centers.[228] Therefore, reducing the density of mobile ions or charge traps inside the absorber and at the interfaces may alleviate the hysteresis.

Figure 5-7: (a) J-V hysteresis measured using different scan speeds and directions of the

scan. Reprinted with permission from Ref.[220]. (b) Ferroelectricity of

CH3NH3PbI3 Perovskite. Reprinted with permission from Ref.[232]. (c)

Schematic diagrams indicating the influence of ion migration in the

perovskite solar cells. Reprinted with permission from Ref.[225]. (d)

Observation of the appearance of substantial JV hysteresis when cooling the

p-i-n perovskite device to 175 K. Reprinted with permission from Ref.[234].

Charge transfer rates at the interfaces of perovskite absorber also strongly influence J-

V hysteresis. If the unbalanced charge collection exists, i.e. if the charge transfer rates between perovskite and the n-/p-type selective contacts are quite different, charges will

106

accumulate on the interface with a lower charge collection rate and build up a transient capacitance. Evidence of trapped charges was found at two interfaces in the conventional n-i-p structure,[226] where the electron and hole mobilities in the ETM and HTM are differing, respectively.[227] Interestingly, the n-i-p device employing a thin mesoporous

ETM and an HTM with desired hole mobility typically exhibits negligible hysteresis, which is likely due to the enhanced surface area for electron injection and improved hole transport, respectively. In contrast, the inverted p-i-n cells exhibit much less J-V hysteresis, presumably due to a balanced charge carrier transport and surface passivation on the perovskite/fullerene interface.[227] However, it was demonstrated that the so-called hysteresis-free p-i-n devices exhibit substantial J-V hysteresis when the temperature was reduced to 175 K (Figure 5-7d).[234] Thus, changing the device architecture may not address the underlying mechanism of hysteresis in the perovskite materials themselves.

Moreover, as the devices aged, the J-V hysteresis is aggravated due to the degraded electronic quality of perovskite, especially at interfaces.[220] This shows the importance of improving the stability of the perovskite and the engineering at the interfaces to prevent materials degradation. Furthermore, compositional engineering may also reduce the J-V hysteresis. Unlike MAPbI3, FAPbI3 possesses an asymmetric charge transfer rate, which balances the charge extraction at either side of the perovskite and alleviates J-V hysteresis.[139] Further development may show that the J-V hysteresis could be reduced or eliminated.

5.6.3 Toxicity and Pollution

107

Because most perovskite solar cells are lead based, there are environmental concerns with the possibility of large-scale development. Recent environmental research should reduce these concerns. Ex-ante life cycle analysis and environmental impact assessment of perovskite solar cells have revealed that the lead bears very little proportion on the overall environmental impact during the manufacturing process.[235, 236] Compared with other lead emission sources, such as mining, fossil fuels, and the manufacture of common products (batteries, plumbing, soldering, electronics, etc.), the potential lead pollution from a 1 GW perovskite PV plant is insignificant, even assuming the worst case leakage scenario during operation.[237] in fact, perovskite PV may actually be able to reduce the amount of

Pb contamination in the environment by providing an opportunity to reuse it from other applications. Recently, perovskite PV devices were fabricated using lead sources recycled from used car batteries.[238] Although no industrial data exist at present, these results are based on the best projections of an industrial process and are likely an overestimate of the potential hazard.

5.7 Conclusion

The last five years have witnessed a rapid development of perovskite solar technology.

A variety of device architectures and material preparation methods have been developed for fabricating high-performance PV devices. Recent advances in engineering the bulk and interface properties of perovskite thin films and contacts have been tremendously effective in enhancing device performance. These advanced engineering techniques are beneficial to increase perovskite grain size and crystallinity, to improve surface coverage and film morphology, and to passivate surface and bulk defects. Further improvement of the 108

perovskite PV devices depends on a precise control of the processing of the organic and inorganic precursors and corresponding understanding of the fundamental material properties of the perovskites. With progress in device stability, perovskite solar cells may well be a very promising technology for the future PV market.

109

Chapter 6

Formation of the Perovskites: a Processing Phase Diagram Study on Methylammonium Lead Iodide

A delicate control of the stoichiometry, crystallographic phase, and grain structure of the photoactive material is typically required to fabricate high-performance photovoltaic

(PV) devices. Organic-inorganic metal halide perovskite materials, however, exhibit a large degree of tolerance in synthesis, and can be fabricated into high-efficiency devices by a variety of different vacuum- and solution-based processes, with a wide range of precursor ratios. This suggests that the phase field for the desired material is wider than expected, or that high device efficiency may be achieved with a range of phases. In this chapter, we investigate the structural and optical properties of the perovskites formed when a range of compositions of methylammonium iodide (MAI) and lead iodide (PbI2) were reacted at temperatures from 40 to 190 °C . The reactions were performed according to a commonly employed synthetic approach for high-efficiency PV devices, and the data was analyzed to construct a pseudo-binary, temperature-dependent, phase-composition processing diagram.[149] Escape of MAI vapor at the highest temperatures (150 – 190 oC) enabled a PbI2 phase to persist to very high MAI concentrations, and the processing diagram was not representative of phase equilibrium in this range. Data from reactions performed with a fixed vapor pressure of MAI allowed the high-temperature portion of the 110

diagram to be corrected and a near-equilibrium phase diagram to be proposed. The perovskite phase field is wider than previously thought under both processing conditions and extended by the existence of stacked perovskite sheet phases. Several aspects of the diagrams clarify why the perovskite materials are compatible with solution processing.

6.1 Introduction and Motivation

Despite a rapidly expanding literature that updates perovskite solar cell performance advances, several important fundamental questions regarding the material structures and properties have not been addressed. Clearly, high-performance perovskite devices can be fabricated through the empirical optimization of various preparation pathways, but the complexity of the film formation dynamics suggest that further advances would be enabled by the development of rational synthetic guidelines. In particular, with a long-term goal of scaling-up the materials synthesis to enable fabrication of large area devices and modules, a systematic study of the impact of processing temperature and composition on the formation of metal halide perovskites would be useful. Currently, such a study does not exist.

To address this need we have performed a comprehensive study of the reaction of methylammonium iodide (MAI) and lead iodide (PbI2), which are the two components used in the synthesis of a prototypical perovskite material used for PV, methylammonium lead iodide (MAPbI3). To probe the phase space, experiments were designed to examine the reaction of specific, known molar ratios of MAI to PbI2. Fourteen different molar ratios of the components were prepared by dissolving the components in N,N-

Dimethylformamide (DMF), and films were formed by spinning aliquots of the solutions 111

onto a 50 nm compact TiO2 layer that was prepared on soda-lime glass substrates (see

Experimental Section). Each sample composition was specified in terms of the mole fraction of MAI in the mixture (XMAI), and XMAI was varied between 0 and 1:

푋푀퐴퐼 = 푛푀퐴퐼/(푛푀퐴퐼 + 푛푃푏퐼2) (6.1)

Films were rapidly heated to the reaction temperature (40 to 190 oC) by placement on a preheated hot plate in a nitrogen glovebox and allowed to react for 30 minutes. After cooling to room temperature, the properties of the films were examined by X-ray diffraction (XRD), optical absorbance (OA) spectroscopy, scanning electron microscopy

(SEM), and energy-dispersive X-ray spectroscopy (EDS). The spin-coating method has led to high-efficiency PV devices,[15] and was chosen for this study to provide a homogeneous mixture of the two precursors with a well-controlled stoichiometry in a thin film form. The thin-film configuration enables rapid heat transfer and allows the phase transition study to be done on a relatively short timescale (samples were ~300 nm thick, as determined by profilometry). Experiments confirmed that 30 minutes was sufficient to allow complete reaction in the thin-film samples. Longer annealing times at temperatures below 150 oC produced no significant changes in the data, indicating near equilibrium conditions. At higher temperatures, however, MAI was lost from the film. To study the perovskite phase behavior under conditions close to equilibrium in this temperature range, additional data were collected from samples that were reacted in a sealed graphite box with excess MAI.

Note, that our results are expected to be similar to reactions of MAI and PbI2 in thicker films prepared by other means, including evaporation, though longer reaction times may be required.

112

6.2 Perovskite Sample Preparation

The perovskite methylammonium lead iodide (CH3NH3PbI3, MAPbI3) was prepared according to literature procedures.[20, 21] Methylamine iodide (CH3NH3PbI, MAI) was first prepared by reacting hydroiodic acid (HI, 57 wt% in water, Sigma-Aldrich) with methylamine solution (CH3NH2, 33 wt% in ethanol, Sigma-Aldrich) under a nitrogen atmosphere in a stirred ice bath for 2 hours. The resulting solution was then evaporated at

80 oC and the precipitate was washed three times using diethyl ether. After drying under vacuum, MAI was obtained as a white powder. The synthesized MAI and PbI2 (99.999%,

Sigma-Aldrich) were dissolved in anhydrous DMF at various molar ratios to prepare precursor solutions. Note that the PbI2 concentration was fixed at ~ 1 M while the MAI concentration was varied. Precursor solutions with PbI2 to MAI molar ratios of 1:0.1, 1:0.3,

1:0.6, 1:1, 1:2, 1:3 and 1:4 were prepared, corresponding to XMAI values of 0.09, 0.23, 0.38,

0.50, 0.67, 0.75 and 0.8, respectively. Films were also prepared from additional precursor solutions with PbI2 to MAI molar ratios of 1:0.9, 1:1.1, 1:1.2, 1:1.3, 1:1.4, 1:5 and 1:9 and reacted at various temperatures to prove the processing/composition space.

The 1” by 1” soda-lime glass substrates were sequentially cleaned with Micro-90 detergent and deionized water in an ultrasonic bath for 30 min, and dried with nitrogen gas.

A compact layer of titania (c-TiO2) was deposited by spin-coating titanium isopropoxide in ethanol solution (Sigma Aldrich).[113] The substrates were annealed at 500 oC in air for

30 minutes before the deposition of perovskite thin film composites.

To form the perovskite layer for X-ray diffraction and optical absorbance spectroscopy measurements, the precursor solutions were spin-coated on the c-TiO2 coated substrates at

2000 rpm in a nitrogen glove box. After spin-coating, the films were reacted at various 113

temperatures ranging from 30 to 190 oC with 10 oC intervals for 30 minutes. Standard samples were reacted on a hot plate covered with a glass petri dish, which is a commonly employed synthetic approach for high-efficiency PV devices. Additional samples were reacted in a sealed graphite box with an excess of MAI powder at temperatures ranging from 150 to 190 oC. The amount of MAI added to the graphite box was calculated to maintain a fixed MAI partial pressure. Samples were transported for characterization in an

N2 sealed storage box to minimize the air exposure.

6.3 Structural and Optical Properties of Perovskite Films

6.3.1 Pure Phases of Perovskite-Related Films

To experimentally obtain XRD data for the phases expected to be dominant in the study, we prepared samples with XMAI = 0, 0.5, and 1, and reacted the samples for 30

o minutes at 80 C. Figure 6-1a shows the XRD data for the PbI2 (XMAI = 0), stoichiometric

MAPbI3 (XMAI = 0.5) and pure MAI (XMAI = 1). The primary feature in the XRD spectra of the spin-coated PbI2 (XMAI = 0) is from the hexagonal 2H poly-type,[113] with a strong

o diffraction from the (001) planes at 2θ = 12.6 (d ≈ 6.95 Å). In contrast, the pure MAI (XMAI

= 1) phase is in the tetragonal rock salt phase (space group P4/nmm). Major features in the spectrum include diffraction peaks at 2θ = 9.84o, 19.74o, and 29.79o, which correspond to diffractions from (001), (002), and (003) planes,[239] respectively. The XRD spectrum of the samples prepared with XMAI = 0.5 corresponds to MAPbI3 perovskite (Figure 6-1b). In this case, a dominant diffraction peak from (110) planes of the tetragonal I4/mcm phase (β phase) is observed at 2θ = 14.08o (d ≈ 6.26 Å), with minor peaks at 2θ = 28.36o, 31.76o and

43.08o due to (220), (310) and (330) diffractions, respectively. 114

Figure 6-1: (a) XRD spectra of pure phases of spin-coated PbI2, MAPbI3, and MAI. (b) A

fine XRD scan of the tetragonal MAPbI3 phase. (c) Comparison between the

MAPbI3 α and β phases.

+ In this β phase, the c axis is slightly elongated due to the polar organic CH3NH3 cations and multiple twinning in comparison to the cubic Pm-3m structure adopted by inorganic

o perovskites such as CsPbI3.[240] Above 60 C the degree of distortion in the unit cell is reduced and a second tetragonal P4mm phase (α phase) becomes dominant.[240-243] This phase as well as the very similar cubic Pm-3m have both been referred to as the α phase[241-243] because the c-axis distortion is very small (0.05 Å) and difficult to

115

resolve.[240] The quasi-cubic α and tetragonal β phases have very similar structures due to a small difference in energy for the configurations, and only the minor diffraction from the (211) planes at 2θ = 23.5o can be used to distinguish between them.[242] The α phase can be observed by heating samples for a long time to improve the crystallinity and rapidly reducing the temperature by quenching in an ice bath prior to XRD analysis (Figure 6-1c).

However, for ease of experimentation, this was not done routinely. For the construction of the processing diagram and proposed phase diagram (vide infra), we simply used the fact

o that the α phase is stable above 60 C after verifying the literature reports for XMAI = 0.5.

As for the reactants, MAI does form a pre-melting state called an ionic plastic phase at 148 oC prior to becoming amorphous at 170 oC,[244] but this is not of significance to our study.

Also, no structural phase transformations are known for pure PbI2 in the temperature range we examined.

Turning to the OA data, Figure 6-2 shows a relatively featureless optical absorption spectrum for pure MAI (XMAI = 1) as expected for a wide bandgap halide salt. On the other hand, the spectrum for PbI2 (XMAI = 0), shows strong absorption at wavelengths shorter than

520 nm consistent with the reported direct bandgap of ~2.4 eV at room temperature.[245]

For the material prepared at XMAI = 0.5, we see the well-known band-to-band optical absorption edge at ~800 nm (~1.55 eV) for MAPbI3 perovskite, with absorption that plateaus at wavelengths below 500 nm.[124, 246]

When the reaction temperature was increased to 110 oC the bandgap transition became sharper (Figure 6-3a). At the same time, the dominant XRD peak at 2θ = 14.08o became more intense (Figure 6-3b), indicating a more complete conversion of the precursors and grain growth. In fact, the perovskite crystal size (estimated using the Scherrer equation) 116

increased from ~40 nm for films reacted at room temperature to around 100 nm for films formed at the higher temperatures (Figure 6-4). However, when the reaction temperature

o exceeded 150 C, MAPbI3 decomposed into PbI2 and MAI. The disproportionation was

o complete at 190 C and only PbI2 was observed in both OA and XRD spectra (Figure 6-3a and 6-3b, respectively) due to the evaporation of MAI.

Figure 6-2: Optical absorption (OA) spectra of pure phases of spin-coated PbI2, MAPbI3,

and MAI.

Figure 6-3: (a) Optical absorbance spectra and (b) contour plot of the room-temperature

XRD mapping of the spin-coated MAPbI3 (XMAI = 0.5). 117

Figure 6-4: Grain size in the MAPbI3 (XMAI = 0.5) films as a function of reaction

temperature.

6.3.2 Films with MAI-Poor Compositions

Understanding the behavior of the dominant phases assists in interpreting the XRD and

OA spectra of the other compositions. Looking at the XRD data for XMAI = 0.09, 0.15, 0.23, and 0.38 in Figures 6-5a to 6-5d, it is clear that MAI-poor films (XMAI < 0.5) are comprised of either pure PbI2 or mixtures of PbI2 and MAPbI3. As XMAI was increased to ~0.15 (Figure

6-5b), the perovskite feature emerged as a minor phase at 2θ ≈ 14.10o.

Below this threshold composition, e.g. at XMAI = 0.09, the OA spectra are representative of PbI2 at all temperatures (Figure 6-6a) and MAPbI3 could not be resolved in the XRD data (Figure 6-5a). On the other hand, at XMAI = 0.23, the OA spectra indicate the dominant presence of MAPbI3 at all but the highest temperature (Figure 6-6b) even though the XRD 118

feature at 2θ ≈ 14.10o is still weak (Figure 6-5c). Clearly, the high absorption coefficient of the perovskite [119] causes it to dominate the long wavelength portion of the OA spectrum even when present in small amounts. As a result, determination of the phase purity of a sample solely through the use of OA measurements can be misleading. Both data sets, however, make it clear that the fraction of the MAPbI3 phase in the sample decreases with increasing reaction temperature. This is due to the nucleation and growth of

PbI2 crystals from the amorphous precursor mixture at low temperatures and MAI vaporization at higher temperatures.

Figure 6-5: XRD contour plots of the spin-coated non-stoichiometric MAPbI3 composites

with MAI percentage (XMAI) of (a) 0.09, (b) 0.15, (c) 0.23, and (d) 0.38.

119

Figure 6-6: Optical absorption spectra of the spin-coated non-stoichiometric MAPbI3

composites with MAI percentage (XMAI) of (a) 0.09, (b) 0.23, and (c) 0.38.

More complex behavior is seen at XMAI = 0.38. The XRD (Figure 6-5d) and OA (Figure

6-6c) features of the low-temperature (< 70 oC) phases have not been reported before, to our knowledge, and cannot be assigned to any reported material. However, they are likely related to the intercalation of guest MAI or DFM molecules between I-Pb-I layers, leading to structures intermediate between PbI2 polytypes and perovskites.[247] Such intermediate structures may be similar to the MAI-PbI2-DMSO complex formed in solutions.[160]

Overall, from the results for MAI- poor films, it is clear that PbI2 cannot be completely converted into MAPbI3 perovskite, as was expected.

6.3.3 Films with MAI-Rich Compositions

While the phase evolution of the MAI-poor films is relatively straightforward to understand, MAI-rich films (XMAI > 0.5) exhibit more complicated structural and optical properties. This region of the phase space is perhaps most important because many

120

preparations for high-efficiency perovskite-based PV devices begin with XMAI values in the range of 0.5 to 0.75.[20, 124, 141, 176, 180, 209, 248] We will first focus on the high- temperature reactions since most high-efficiency devices are prepared with heat treatment in the temperature range of 80 – 150 oC.

Figures 6-7 and 6-8 show the XRD and OA data for samples prepared with XMAI = 0.67,

0.75, and 0.80. As with XMAI = 0.5 (Figure 6-3), the higher temperature reactions result in a MAPbI3 phase. Here, we see that the temperature required to form the MAPbI3 phase increases with increasing MAI content. For temperatures typically needed for high- efficiency devices, 80 – 150 oC, a perovskite phase is evident and dominant until 160 – 170 o C, at which point the PbI2 becomes dominant and MAPbI3 is no longer observed. As mentioned previously, there is no XRD evidence for MAI due to the lack of order, but MAI should still be present in the quasi-liquid phase (Figure 6-9c) below the melting temperature (190 oC).[244]

To examine the possibility of MAI evaporation, thermal gravimetric analysis (TGA) and differential scanning calorimetry (DSC) of MAI powder was performed. Both MAI

o and MAPbI3 exhibit good thermal stability at temperatures below 190 C (Figure 6-9a). To measure the vaporization rate, MAI samples were held at a fixed elevated temperature in

o flowing N2 and the weight loss was recorded as a function of time. At 150 C the weight loss rate was 0.002% per minute, while at the MAI melting point (190 oC) the weight loss rate was 0.032% per minute (Figure 6-9b). Though the vapor pressure is evidently low at these temperatures, the loss of the MAI indicates that the high-temperature portion of our phase study is clearly not in equilibrium.

121

Figure 6-7: XRD contour plots of the spin-coated non-stoichiometric MAPbI3 composites

with MAI percentage (XMAI) of (a) 0.58, (b) 0.67, (c) 0.75, and (d) 0.80.

Figure 6-8: OA spectra of the spin-coated non-stoichiometric MAPbI3 composites with

MAI percentage (XMAI) of (a) 0.58, (b) 0.67, (c) 0.75, and (d) 0.80.

122

Figure 6-9: (a) TGA of MAI and MAPbI3 in N2. The heating ramp is 2℃/min. (b)

Thermogravimetric analysis of MAI at 150 and 190 oC. (c) DSC plot of the

MAI rock-salt to ionic plastic phase transition at around 140 to 150 oC.

The XRD (Figure 6-7) and OA (Figure 6-8) spectra indicate that the resultant films contain different proportions of PbI2 and MAPbI3, depending on the initial composition.

As will be discussed in more detail later, the data indicates that the boundary of the MAPbI3 phase field persists farther than previously expected into the higher MAI content regions of the temperature/phase processing diagram. Consistently with the XRD data, EDS measurements on these samples revealed final compositions that were MAI-rich, in agreement with the precursor ratios. Overall, these findings conflict with the recent theoretical prediction of a narrow phase field for MAPbI3.[128]

The existence of non-stoichiometric perovskite phases can be explained by considering the two-dimensional organic-halide perovskite materials with a variable stoichiometry that have been known since the pioneering work of Mitzi et al.[106] So-called layered perovskites consist of a variable number of metal halide layers formed by corner-sharing

4- (PbI6) units with associated cations that are sandwiched between intercalated organic

123

cation bilayers. The cation bilayers keep the perovskite layers separated, retard the formation of bulk 3D perovskites, and introduce a stacking fault between adjacent metal halide sandwiches. As XMAI increases, the number of metal halide layers decreases as the density of both intercalated species and stacking faults increases. In essence, the 3D perovskite is gradually transformed into 2D layered perovskites by dissolution (Figure 6-

10).

Figure 6-10: Crystal structures of PbI2, MAPbI3 perovskites, 2D MA2PbI4 sheets, 1D

MA3PbI5 chains, and 0D MA4PbI6 blocks.

There exists a transitional phase between 3D and 2D perovskites where the perovskite has a limited number of stacking faults and can still persist as a 3D crystal. We term this phase “stacked perovskite sheets” (SPSs).[149] Thus, while a highly ordered perovskite phase may exist only in a narrow range close to the stoichiometric value for MAPbI3, the

124

perovskite structures for MAI-rich films persist to values of XMAI > 0.8 in the form of SPSs.

The crystallographic structures of perfect and stacked perovskites are not clearly distinguishable due to the same long-range periodicity of the 3D perovskite framework.

However, the SPSs are indeed MAI-rich as determined by energy dispersive X-ray spectroscopy. One might assume that defect levels in the SPS phases might lead to poor electronic quality, but a study on similar grain boundaries yielded no defect states within the band gaps.[119] Because of the benign nature of these defects, high-efficiency devices can be fabricated in a wider composition range than would be expected.

At the lower reaction temperatures, the MAI-rich films exhibit a distinctive XRD feature that is comprised of overlapping diffraction peaks from several components with slightly different d-spacings (2θ between 11.3o and 11.7o). These features are dominant at the higher MAI values (XMAI > 0.75) but quenched with the appearance of the perovskite feature as the MAI concentration is decreased. These diffractions are likely due to the

4- presence of low-dimensional perovskites (LDPs) that are built up from the same (PbI6) octahedral building blocks. When excess MAI is present in a large concentration, the basic

4- (PbI6) units exist in isolated form as zero-dimensional (0D) quantum dots in an MAI matrix. As the amount of MAI is decreased, the 0D units connect and build-up one- dimensional (1D) perovskite chains and 2D perovskite sheets through self-assembly, eventually leading to the assembly of 3D perovskite (Figure 6-10). The unique diffraction feature for LDPs arises because the structures are separated by a monolayer of intercalated

MA+ cations. Thus, the spacing between the metal halide octahedra is expanded by ~1.4 Å in comparison to the spacing in the 3D perovskite (d = 6.3 Å). Consequently, the corresponding XRD peaks shift to a lower angle. For the XMAI = 0.8 sample, the most 125

intense peak at around 11 to 12o in the XRD spectrum (Figure 6-11a) contains 4 discrete peaks, which were vaguely assigned to the precursors by previous researchers. Here we showed that similar XRD spectrum can be simulated by a 0D perovskite structure,[249]

4- + which contains discrete (PbI6) octahedra separated by 4 surrounding MA cations, as shown in (Figure 6-11b). Note that there is polydispersity in structure, so the determination of the exact configurations and ratio of species of the composites with XMAI = 0.5 to 0.8 is not straightforward.

Figure 6-11: (a) Measured and simulated XRD spectra of the 0D perovskite phase. The

inset is the zoom-in of the peak fitting at the most intense peak (~11o). (b)

Crystal structure of the 0D perovskite.

To support the interpretation of the XRD data, note that the OA spectra of the MAI- rich composites exhibit a very strong exciton absorption centered at 370 nm, corresponding to the ground-state exciton from the 0D perovskite structure.[249, 250] As XMAI is decreased, the exciton transition is quenched while the absorption in the range of 500 to

126

800 nm is increased, indicating that the fraction of quantum dots is decreased as the 3D perovskites are assembled. However, evidence for perovskite quantum dots persists even at XMAI = 0.67, which indicates that the LDP phase region is structurally and energetically complex.

6.4 A Pseudo-Binary Phase Diagram of CH3NH3PbI3

A general picture of the solid-state phase behavior for the system emerges if we consider the XRD and OA datasets for the full range of compositions and temperatures that were examined. When the MAI fraction is low (XMAI < 0.15), the only identifiable phase is

PbI2. Adding MAI results in a two-phase mixture consisting of PbI2 and perovskite

MAPbI3. The mixture tends towards PbI2 at higher temperatures, and at no point below

XMAI = 0.5 is the resultant material a pure perovskite. The simple fact that the perovskite signatures persist to high values of XMAI (> 0.8) indicates the existence of a wide phase field.

Despite the fact that EDS shows that the conventional perovskite phase (α) has an I:Pb ratio of 2.9 to 3.1, while the stacked perovskite sheet phase (α’) has an I:Pb ratio of 3.2 to

3.5 (Figure 6-12), α and α’ are indistinguishable by XRD and OA in our studies. The increased MAI content creates stacking faults within the perovskite and generates the SPSs.

With further increase in MAI concentration, the SPSs are broken-up and solvated by MA+ cations to form thin 2D sheets, 1D chains, and 0D quantum dots (Figure 6-10).

Interestingly, the XRD data for the MAI phase appears only when the precursor composition is very close to pure MAI (> 99%). Evidently, even small amounts of PbI2 disrupt the crystal structure and lead to LDPs within an amorphous MAI matrix. 127

Figure 6-12: Compositional analysis of thin films prepared from various precursor

compositions at RT, 100 oC, and 150 oC, respectively. Final compositions are

correlated to the specific phases in the phase diagram.

The findings can be compactly presented in a pseudo-binary temperature/phase processing diagram (Figure 6-13a). To construct the diagram, the percentages of the phases in the films prepared at each temperature were determined by calculating the integrated

XRD peak intensity ratios. For example, the amount of MAPbI3 perovskite was determined by comparing the integrated peak intensity at 2θ = 14.10o to the sum of the integrated peak

o o intensities for PbI2 (2θ ~ 12.60 ) and the low-dimensional perovskite (2θ ~11.60 ) peaks.

The percentages of the lead iodide and LDP phases were determined in a similar way.

Figure 6-13b is one example showing the proportions of the perovskite, PbI2, and LDP phases in the films reacted at 100 oC as a function of the starting precursor composition.

The perovskite curve was fitted by a Gaussian function to determine the pure phase region.

The pure PbI2 and LDP phases are defined as the regions where the perovskite percentage

128

was less than 0.1%. The MAI-poor side is the pure PbI2 region and the MAI-rich side belongs to the LDP phase. We then plotted the pure phase boundaries (stoichiometric limit of each pure phase) as a function of the reaction temperature (Figure 6-13a). The X and Y axes were swapped to correspond to a conventional view of a binary phase diagram.

Finally, we assumed that the field boundaries were smooth to generate the final processing diagram.

Figure 6-13: (a) Pseudo-binary temperature/phase processing diagram for

methylammonium lead iodide. (b) Phase percentages of PbI2 (purple

triangles), perovskite (red squares) and LDP (cyan circles) as a function of

MAI precursor concentration in the films reacted at 100 oC.

129

Figure 6-13a can be used to understand experimental processing pathways towards the formation of perovskites, but it is not necessarily descriptive of the equilibrium phase behavior. Although the data for the low temperature (< 150 oC) portion of the diagram is evidently obtained under near equilibrium conditions, the phase distribution at the higher temperatures may be affected by the evaporation of MAI. To address this possibility several high-temperature reactions were performed in a sealed graphite box that contained an equilibrium partial pressure of MAI. Interestingly, when the MAI ambient was maintained in the confined environment, we found that the α’ perovskite phase field

o extended fully to 190 C with no evidence for PbI2 (Figure 6-14). Additionally, the high vapor pressure of MAI promoted the grain growth of the perovskite films, which was revealed by SEM (Figure 6-15). Under these conditions, the system is much closer to equilibrium, and a phase diagram may be tentatively proposed (Figure 6-16).

Figure 6-14: XRD contour plots of high temperature processed perovskites (XMAI = 0.75)

reacted (a) on a hot plate and (b) in a sealed graphite box.

130

Figure 6-15: SEM images of perovskite films (a) reacted at 100 oC on a hot plate and (b)

reacted at 150 oC in a sealed graphite box.

Figure 6-16: Proposed phase diagram constructed with high-temperature reactions carried

out under a saturated MAI vapor pressure.

It is interesting to correlate our processing and proposed phase diagrams with the findings of other groups. In both diagrams, note that the perovskite phase is indeed quite

131

wide and extended into MAI-rich regions (up to XMAI = 0.8) when the reaction temperature

o is high (>120 C). This is consistent with the MAI to PbI2/PbCl2 precursor ratios of 1:1 to

3:1 (XMAI = 0.5 to 0.75) that have been used in numerous solution-based syntheses.[20, 21,

124, 141, 176, 180, 209, 248] Evidently, a relatively high reaction temperature provides sufficient thermal energy to eliminate alkyl groups from the spaces between perovskite nanostructures[13] and promote the LDP to SPS phase transition. While we find good agreement at the higher temperatures, the lower temperature (T< 120 oC) MAI-rich portion of the diagram conflicts with a few materials preparations for high-efficiency perovskite devices.[20, 141] This discrepancy may be explained by considering that these preparations were done in uncontrolled laboratory air rather than in, e.g. an inert gas environment. In these cases, it is possible that the MAI content during the film formation may have been lower than expected due to unaccounted for reactions with laboratory moisture and/or oxygen.[251] For example, it is well known that halide perovskites reactive with water vapor at relatively low temperatures to form lead iodide as a decomposition product.[117,

215] Accordingly, we can consider that low-temperature annealing of LDP phase samples in a moisture-containing environment can lead to a reduction in the effective MAI content during film formation/reaction. Although both low and high-temperature pathways can lead to perovskites, low-temperature annealing is more widely adopted because high- temperature annealing typically results in a poor surface morphology.[138, 243] The fine details of the morphological changes during these reactions will continue to be of interest going forward.

132

6.5 Conclusion

The formation of the perovskite MAPbI3 phase from PbI2 and MAI precursors was examined over a range of reaction temperatures and initial compositions. Based on the

XRD, OA, TGA, and EDS results, a processing diagram was determined and phase diagram was proposed. Both diagrams show that a pure perovskite phase occurs only in a narrow compositional space. However, an SPS phase extends the perovskite region to much higher values of XMAI. We showed that the SPS phase can be formed from LDP structures by reacting at high temperatures, or by decreasing the MAI concentration. The reaction pathways between the LDPs and the SPS phases provide easy routes for solution- processing via low-dimensional structures and self-assembly that allow the excellent optoelectronic properties of perovskites to be retained. Clearly, MAPbI3 perovskite is a forgiving material for polycrystalline thin film optoelectronic devices.

133

Chapter 7

Spatially-Resolved Current Generation Measurement of Solution-Processed Perovskite Solar Cells

The photovoltaic performance of organic-inorganic metal halide perovskite solar cells is significantly affected by perovskite film composition, morphology, and uniformity regardless of the deposition method. In this chapter, we used laser beam induced current

(LBIC) to spatially resolve the current collection in perovskite devices to compare the performance of perovskite solar cells prepared by different solution-based deposition methods and treatments. Devices with P3HT as the hole transport material (HTM) were used to study current generation non-uniformity. The results show that the non-uniformity is attributed to microscopic defects and impurity phases in the perovskite films. Based on the useful information provided by the LBIC measurement, we optimized the perovskite film growth conditions by adopting solvent and processing engineering advances and achieved a champion cell with 16% efficiency.

7.1 Introduction and Motivation

One of the advantages of perovskite solar cells is that they can be prepared by a great variety of deposition processes.[15] Among all the approaches, solution-based methods have a great potential due to low manufacturing costs and ease of processing. Due to the 134

phase behavior of perovskites,[149] heating to relatively low temperatures can convert the organic and inorganic precursors into the desired perovskite form. Although the process seems simple, precisely controlled procedures are required to produce high-quality devices because the photovoltaic performance is greatly affected by the morphology, stoichiometry, and crystallinity of the thin films. Thus, the choice of the deposition technique, preparation conditions, and subsequent treatments are of crucial importance.

To study the relationship between the solution processing method and the resultant morphology, composition, uniformity, and device performance, we employed LBIC imaging technique to investigate the spatial uniformity of the photocurrent collection in different solution-processed perovskite devices.

7.2 Perovskite Devices Prepared by Different Methods

7.2.1 Perovskite Device Preparation

The methylammonium lead iodide (MAPbI3) thin films were cast from precursor solutions of methylamine iodide (CH3NH3I, MAI) and lead iodide (PbI2) through various solution-based approaches including: (Device A) single step spin coating using a blended solution [5], (Device B) single step spray deposition using a blended solution [6], (Device

C) sequential deposition using spin-coated PbI2 and MAI [7] and (Device D) sequential deposition using sprayed PbI2 and spin-coated MAI. As for the single step coating methods

(A & B), the blended precursor solution was prepared by dissolving MAI and PbI2 at a molar ratio of 3:1 in anhydrous N,N-dimethylformamide (DMF). The molar concentrations of Pb2+ were fixed at 1 M and 0.2 M for spin and spray depositions, respectively. The blended precursor solution was first spin-coated at 3000 rpm for 30 s (A samples) or 135

sprayed at a rate of 50 µL/min for 30 minutes (B samples). Both types of samples were then annealed at 100 ℃ for 1.5 hours and 150 ℃ for 5 min in a nitrogen glovebox. The low-temperature annealing is crucial to achieve a better morphology [8] while the high- temperature treatment is necessary to remove the low-dimensional perovskite impurity phase [4]. As for the sequential deposition methods, PbI2 in DMF solution was first deposited by spin coating a 1 M PbI2 solution at 3000 rpm for 30 s (C samples) or spraying a 0.2 M PbI2 solution at a rate of 50 µL/min for 30 minutes (D samples). The PbI2 films were then over-coated with MAI by spin-coating a 40 mg/mL MAI isopropanol solution at

3000 rpm for 30 s. The perovskite conversion was completed by annealing at 150 ℃ for 5 min.

The fluorine-doped tin oxide (FTO) glass substrates (TEC8, Pilkington) were sequentially cleaned with Micro-90 detergent and deionized water in an ultrasonic bath and dried with nitrogen gas. A 60 nm compact layer of titanium dioxide (c-TiO2) was deposited on FTO by spin-coating 0.3 M titanium diisopropoxide bis(acetylacetonate) (Sigma

Aldrich) in ethanol solution followed by annealing at 500 ℃ for 30 minutes [9]. A 400 nm mesoporous titanium dioxide (mp-TiO2) layer was spin-coated at 5000 rpm for 30 s using ethanol-diluted TiO2 paste (18NR-T, Dyesol) and annealed at 500 ℃ for 30 minutes in the air. Perovskite thin films were deposited on the mp-TiO2 layer via the previously described approaches. Following this step, a 80 nm poly(3-hexylthiophene-2,5-diyl) (P3HT) was deposited as the hole transport layer (HTL) by spin-coating 15 mg/mL P3HT in dichlorobenzene. 3.4 µL tert-butylpyridine (tBP) and 6.8 µL lithium bis-

(trifluoromethanesulfonyl)imide (Li-TFSI) in acetonitrile (80 mg/ml) were added to a 1 mL P3HT solution to achieve the desired hole conductivity. Finally, a 60 nm gold electrode 136

was thermally evaporated onto the HTL. Figure 7-1 shows a cross-sectional SEM image and the corresponding energy level diagram of the perovskite device investigated in this study.

Figure 7-1: (a) cross-sectional SEM image and (b) energy level diagram of a perovskite

device with a structure of FTO/c-TiO2/mp-TiO2/MAPbI3/P3HT/Au.

7.2.2 Perovskite Device Performance

To compare the device performance of perovskite solar cells prepared by different solution-based processes, we fabricated more than 60 photovoltaic cells for each method.

Figure 7-2a shows the J-V data for the champion devices prepared by each method, and the statistics of critical parameters for the devices are listed in Table 7.1.

The perovskite devices exhibited moderate energy conversion efficiencies ranging from 5% to 11%. Devices prepared by the sequential spin-coating of PbI2 followed by

MAI, i.e. preparation C, resulted in the best performance. Devices prepared by the other methods showed slightly lower open circuit voltages (VOC) and fill factors (FF), but much lower short current densities (JSC). To investigate the device differences in greater detail, 137

EQE measurements were performed (Figure 7-2b). Devices prepared using the single-step methods (A & B) exhibited lower integrated current densities with a dramatic efficiency drop at the longer wavelengths (450 ~ 750 nm). Apparently, carriers generated near the front of these devices were collected much more effectively than those generated more deeply. The reduced current collection could be due to impurity phases, poor crystallinity, or poor surface coverage of the perovskite thin films. Although the performance of these devices could be improved by optimization of the deposition processes and post-deposition treatments, it is interesting to consider the impact of surface morphology and uniformity in the as-prepared samples.

Figure 7-2: (a) J-V characteristic curves and (b) EQE spectra of perovskite devices prepared by different solution-based processes.

We note that in comparison to the high-efficiency (> 18%) perovskite devices that have been reported with the commonly employed spiro-OMeTAD, P3HT is expected to yield lower device efficiencies mainly due to a lower electron lifetime and unfavorable valance band alignment [13]. In comparison to the smaller spiro-MeOTAD molecules, the large

138

molecular weight P3HT polymer may not conformally coat the relatively rough perovskite surfaces and may leave some grains unpassivated [14]. Consequently, devices using P3HT are more sensitive to the morphology, porosity, and grain size of the perovskite films.

Moreover, compared with a typical spiro-OMeTAD layer with a thickness of ~400 nm, the much thinner P3HT layer (~80 nm) that we employed allowed the surface morphology of the perovskite films to be imaged by SEM after the devices were fabricated. Thus, the use of P3HT allows us to more effectively probe the impact of perovskite film morphology on the device performance.

Table 7.1: Device performance of perovskite solar cells

Preparation Device Eff. (%) J (mA/cm2) V (V) FF (%) Method SC OC A Spin 6.7 ± 0.7 14.2 ± 1.7 0.862 ± 0.068 55.5 ± 7.1 B Spray 5.6 ± 1.8 12.2 ± 3.2 0.877 ± 0.076 49.9 ± 8.5 C Spin + Spin 10.8 ± 0.5 17.5 ± 1.3 0.963 ± 0.055 60.4 ± 5.2 D Spray + Spin 8.4 ± 1.2 15.2 ± 2.5 0.904 ± 0.069 60.5 ± 6.8

7.2.3 Morphology and Photocurrent Uniformity

To determine the local current collection efficiency we performed LBIC measurements of each device. Figure 7-3 shows LBIC mapping images of the perovskite devices prepared by the various approaches. Defects and non-uniformities can be clearly identified from the

LBIC images. Devices prepared by single step coating (A & B) exhibited lower and less uniform LBIC signals than those prepared using the sequential deposition methods (C &

D). This could be due to the loss of MAI in the blended precursor solution during the annealing process, which would cause the formation of voids and local defects. The sequential deposition methods retarded crystallization of perovskites within the inorganic

139

PbI2 network, which results in a more uniform perovskite surface, as shown in the SEM images (Figure 7-4).

Figure 7-3: LBIC maps of perovskite devices prepared by (a) single-step spin, (b) single- step spray, (c) sequential spins, and (d) sequential spray/spin methods.

140

Figure 7-4: SEM maps of perovskite films prepared by (a) single-step spin, (b) single- step spray, (c) sequential spins, and (d) sequential spray/spin methods.

The statistics of LBIC data are shown in Figure 7-5 and Table 7.2. The average LBIC signal is proportional to the JSC and EQE at 532 nm. The sequentially deposited perovskite devices (C & D) show a sharper current collection distribution than single-step coated ones

(A & B), indicating better film morphology and uniformity. The light-induced current generated in the devices fabricated using spray related methods (B & D) are limited by non-uniform surfaces where microscopic local defects (20 to 80 µm) are widely dispersed across the whole device.

141

Figure 7-5: Histograms of the LBIC signals for perovskite devices prepared by different methods.

Table 7.2: Statistics of the LBIC measurement

Preparation J EQE (%) LBIC mean LBIC FWHM Device SC Method (mA/cm2) @532nm (µA) (µA) A Spin 14.2 66.5 49.1 5.47 B Spray 12.2 60.5 42.1 5.49 C Spin + Spin 17.5 78.9 57.5 1.88 D Spray + Spin 15.2 65.1 51.3 2.02

7.2.4 Current Loss Mechanisms

To investigate the reasons for poor current collection, we compared the LBIC data to

SEM images obtained from specific device locations. Two kinds of current collection failures were identified and found to be related to microscopic defects and impurity phases.

Figure 7-6 shows one example of a perovskite device prepared using the single-step spray method. After measuring the LBIC data on the device, we marked the sample using 142

laser scribing (Figure 7-6a&b). The laser scribed mark allows us to track the morphological variations and surface non-uniformity using SEM. As for the low current (~35 µA) regions

(Figure7-6c), a significant amount of voids and defects are identified, and the perovskite grain size distribution is not uniform. In contrast, in the high current regions (Figure 7-6d), the surface is dense and smooth, and the perovskite grains are more compact and uniform in size. Note that although SEM images were measured over a relatively smaller area (~20

µm 2) compared with the LBIC images (~1300 µm2), the measurements do represent the large-area morphology as confirmed by a thorough survey and multiple spots image acquisition.

Figure 7-6: (a) LBIC images of a spray deposited perovskite device before and after laser scribing. SEM images of (b) the laser scribed line, (c) the low current spot “+”, and (d) the high current spot “*”.

143

This method provides us a useful tool to reliably identify the locations with processing defects and bad morphology. We speculate that the pinholes and microscopic voids are likely due to etching caused by precursor solution droplets during the spray process and the evaporation of MAI during the annealing. These may form randomly when the perovskite thin films are solution processed. However, finely tuned fabrication methods, e. g., sequential deposition, can alleviate the formation of these defects.

In addition to the microscopic defects, the formation of impurity phases also leads to a lower current collection. The most detrimental impurities are associated with the low- dimensional perovskites preferably formed in an MAI-rich stoichiometry [4]. These impurities consist of discrete inorganic quantum dot structures in a surrounding matrix of organic cations and hinder the current transfer and the collection of light-induced current.

Figure 7-7 shows the comparison of two devices without and with the impurity phases prepared by the sequential spin method using different MAI concentrations. The device prepared using 40 mg/ml MAI solution (Figure 7-7a & c) exhibits better performance and current uniformity than one prepared using 100 mg/ml MAI solution (Figure 7-7b & d). A high temperature (> 150 oC) treatment can convert the impurity phase into perovskite and improve the device performance.

144

Figure 7-7: LBIC maps of perovskite devices prepared using (a) a 40 mg/ml and (b) a 100 mg/ml MAI solution. SEM images exhibit (c) a smooth surface and (d) a rough surface with segregates of the low-dimensional perovskites for the 40mg/ml and 100 mg/ml samples, respectively.

7.3 Perovskite Devices with Improved Performance

The results of spatial photocurrent generation measurements elucidate the importance of controlling the morphology, crystallinity, and composition of solution-processed perovskite thin films. To fabricate high-efficiency perovskite solar cells, we optimized the device fabrication methods using advanced engineering techniques. The perovskite layer in the advanced devices was prepared by spin coating of a precursor solution of 0.8 M PbI2,

0.1 M PbBr2, and 0.8 M CH3NH3I in a mixed solution of DMF and DMSO (7:3 volume ratio) at 4000 rpm for 30 s, followed by a toluene wash at the end (~5 s) of the spin process. 145

The deposited film was then annealed at 100 °C for 30 min in a nitrogen glove box. The film prepared by this advanced technique possesses a shining surface (Figure 7-8), indicating a smooth morphology.

Figure 7-8: Optical images of perovskite thin films prepared by the single-step methods using conventional (left) and advanced (right) approaches.

The advanced method was used to fabricate the perovskite absorber for high-efficiency perovskite solar cells. The best performing device consists of a film stack of FTO/50 nm c-TiO2/150 nm meso-TiO2/350 nm CH3NH3Pb(I,Br)3/100 nm spiro-OMeTAD/80 nm Au, as shown in Figure 7-9. The spiro-OMeTAD film was deposited by spin coating of a solution of 0.07 M spiro-OMeTAD in chlorobenzene and doped with 50 mol% Li-TFSI

(50 mol%) and 300 mol% tert-butylpyridine.

The champion device fabricated using this method exhibited a 16% conversion efficiency under 1.5AM illumination, with VOC = 1.12 V, JSC = 20.1 mA/cm2, and FF =70%

146

(Figure 7-10a). Consistent with the J-V data, the device showed excellent photon response with over 80% EQE in the range of 400 to 750 nm (Figure 7-10b).

Figure 7-9: Cross-sectional SEM image of the best performing perovskite solar cell.

Figure 7-10: (a) J-V and (b) EQE of the champion perovskite solar cell prepared by the advanced single-step process.

147

7.4 LBIC Measurement on Perovskite/Si Tandem Cells

Perovskite/silicon tandem solar cells promise high power conversion efficiency and have the potential to enter the photovoltaic market in the near future. However, device design, fabrication, and integration techniques are critical for the full realization of the promise of these high-efficiency tandem devices. Here we used LBIC mapping technique to probe current collection uniformity in the individual sub-cells and investigate charge carrier generation and collection at different the interfaces. The LBIC technique reveals efficiency limiting processes related to defects and optical interference in the multilayer structure. The results show the importance of managing the optical properties of the interfaces and layers to develop high-efficiency perovskite/silicon tandem solar cells.

7.4.1 Purpose of the Work

The highest reported c-Si solar cell efficiency of 25.6% is very close to the theoretical and practical limits of single junction solar cells.[252] This value has only marginally improved in the last 15 years, and it is difficult to significantly improved in future.

Therefore, to become more competitive with conventional power generation technologies, a low-cost method to boost the efficiency of crystalline silicon solar cells is needed.

Recently, Werner et al. from the PV-center at EPFL reported a high-efficiency monolithic perovskite/silicon tandem solar cell with the best efficiency of 21.2% (Figure 7-11a).[253]

This achievement shows a great potential to further improve the performance of the well- established crystalline silicon PV technology. However, the fabrication of high efficiency monolithic tandem cells is very challenging due to the complexity of optical, electrical, and

148

mechanical design of the devices. To further improve the performance of the tandem cells, a better understanding of current generation in each sub-cells is needed.

Figure 7-11: (a) J-V curves and (b) schematic of the device structure of the perovskite/Si tandem solar cells fabricated at PV-center, EPFL. Adapted from Ref. [253] with permission. Copyright 2016 American Chemical Society.

The tandem solar cell consists of a perovskite top cell with the 7-layer structure of

ITO/IO:H/MoOx/Spiro-MeOTAD/CH3NH3PbI3/PCBM/PEIE, a crystalline silicon bottom cell with the 7-layer structure of p-aSi/i-aSi/n-Si wafer/i-aSi/n-aSi/ITO/Ag, and a IZO tunneling layer between the two sub-cells (Figure 7-11b). Both sub-cells need to generate the same current flowing through the tunneling layer under a given operating condition.

Thus, current collection non-uniformities due to processing defects and areas of weak diodes that can limit the performance of a sub-cell will limit the performance of the entire device. Since tandem devices consist of a complex device structure with up to 13 layers of materials prepared by different deposition techniques, it is very challenging to locate processing defects and weak diodes in the structure. Current nonuniformities can also arise

149

at the back contact of the perovskite due to the use of a novel transparent conductive recombination layer that is necessary for construction of the tandem structure. The optoelectronic quality of this intermediate layer is critical to the device performance.

Locating and understanding the nonuniformities in the device stack is critical to design and control of the fabrication processes and can lead to improved overall device performance.

Conventional external quantum efficiency and current-voltage measurements can be used to yield overall PV performance information for the complete device, but information about the performance of the individual sub-cells is more difficult to obtain for a tandem, particularly if there are nonuniformities in the current generation and collection.

Using the LBIC technique, we are able to probe the individual sub-cells and carrier transport across interfaces, and efficiency limiting defects can be identified. We identify processing defects and efficiency limiting spots and provide in-depth analysis on the assessment of current generation/transport in the individual sub-cells.

7.4.2 LBIC Measurement Results of Perovskite/Si Tandem Cells

The perovskite/Si tandem devices (courtesy of Jérémie Werner and Prof. Bjöern

Niesen) used for this measurement have an average 16.2% efficiency with VOC = 1670 mV,

2 JSC = 13.00 mA/cm , FF = 74.6%. Figure 7-12 shows the EQE spectra of the top and bottom subcells measured individually by light biasing the other sub-cell using a wide-band light source with infrared light bias (λ > 830 nm) and blue light bias (λ < 430 nm), respectively.

In the visible range, the top (perovskite) cell has a relatively flat photoresponse while the bottom (Si) cell exhibits a much lower response due to optical absorption in the top cell. In the infrared and near-IR ranges, current can only be generated by the bottom cell. 150

Figure 7-12: EQE spectra of two sub-cells of a perovskite/Si tandem device measured under different light-biases.

To probe the current generation uniformity in both top and bottom cells, we mapped the photocurrent excited by 532 nm and 1064 nm laser light under red or blue light bias, respectively, in a large area (1.22 cm2) device (Figures 7-13 and 7-14). When the 532 nm laser was used to probe the photoexcited current, the average quantum efficiencies (at 532 nm) of both sub-cells (59.1% for the top and 3.6% for the bottom cells) are close to the corresponding EQE at 532 nm values (59.2% and 1.3%, respectively). The higher value determined by LBIC for the bottom cell could be due to the small fraction of the infrared component from blue light bias, which was filtered by the lock-in amplifier in the EQE measurement while was recorded by the LBIC measurement. The current generation of the bottom cell as determined by LBIC at 532 nm with the blue light bias is uniform across the device although some lower current spots are identified (Figure 7-13a). In contrast, after changing to a red light bias, current generation non-uniformities due to processing defects

151

are clearly revealed in the top (perovskite) cell (Figure 7-13b). Some of the defects in the two images overlap, indicating poor current transport through the perovskite material or the recombination layer. These defects are likely formed during the solution processing due to aggregation of the solutes or impurities in the precursor solution. Others low current defects in the top cell may be simply due to incomplete perovskite conversion. This could be a result of decomposition or impurity phase formation at local spots due to uneven heating or off-stoichiometric composition. The formation of processing defects can cause a lower local current generation/collection rate, and as a consequence, lead to a lower overall short current density.

Figure 7-13: LBIC maps of a 1.22 cm2 perovskite/Si cell probed by 532 nm laser (a) with red light bias (the top cell), and (b) with blue light bias (the bottom cell).

The LBIC mapping with 1064 nm laser light exhibits unusual behavior for the perovskite/Si tandem solar cells (Figure 7-14). No signal was collected from the top cell

152

because the photon energy is below the band gap of the perovskite, while in the bottom cell, optical interference was clearly observed. We expected that the fringes may be due to a combination of effects of thickness variation in the device and related light path length variations for angular sweeping of the galvanostatically controlled raster mirrors.

Interestingly, the patterns persisted after 90 or 180o rotations and any movement of sample holder location, indicating the fringes are caused by the optical cavity formed by two parallel surface. Analysis of optical path length between adjacent maxima shows that thickness variation should be in the range of ~0.6 to 2 µm. This variation could be due to the particular wafers with a total thickness variation of 5 to 10 µm. This is confirmed by measuring the double-side polished and single-side textured crystalline silicon devices.

(Figure 7-15). Only the double-side polished sample exhibits the optical fringes.

Figure 7-14: LBIC maps of the bottom cell probed by 1064 nm laser.

153

Figure 7-15: LBIC maps and the line profiles of (a, c) a single-side textured silicon cell and (b, d) a double-side polished silicon cell.

To quantify the non-uniformity of the current generation in the sub-cells, we compared the LBIC data from both large (1.22 cm2) and small (0.17 cm2) area (Figure 7-16) devices.

Figure 7-17 shows the histograms of LBIC distribution of these two devices. The quantum efficiency distribution in the large area devices exhibited a dual-peak curve shape, which is the result of a high density of processing defects. In contrast, the small area device exhibits a single peak distribution, corresponding to more uniform current generation. In particular, the bottom cell of the small area device exhibits uniform photocurrent map

(Figure 7-16b) and sharp LBIC distribution (standard deviation < 5%). If the defects (lower current generation spots) in the large area device can be rectified by thorough control of the perovskite cell fabrication process, the overall current generation would be improved by up to 3 mA/cm2.

154

Figure 7-16: LBIC maps of a 0.17 cm2 perovskite/Si cell probed by 532 nm laser (a) with red light bias (the top cell) and (b) blue light bias (the bottom cell).

Figure 7-17: Histograms of the LBIC of the top and bottom cells of a large size (a, b) and a small region (c, d) of perovskite/Si devices measured using the 532 nm laser.

155

7.5 Conclusion

Perovskite thin film solar cells were prepared by several different solution-based processes. The difference in measured efficiency was primarily due to differing photon-to- current conversion efficiency and short-circuit current. The current generation was spatially resolved using LBIC characterization and was found to be non-uniform across the device. By combining LBIC date with SEM morphology measurements, the low-current locations were correlated to the occurrence of microscopic defects and impurity phases in the perovskite film. With the help of the LBIC technique, the perovskite film morphology and uniformity were further improved and the best efficiency of 16% was achieved. In addition to the single junction perovskite solar cells, we also examined the photocurrent collection in the sub-cells of the perovskite/Si tandem cells.

156

Chapter 8

Stability of Perovskite Solar Cells in Humid Air

After rapid progress over the past five years, organic-inorganic perovskite solar cells

(PSCs) currently exhibit power conversion efficiencies comparable to the best commercially available photovoltaic technologies and have the potential for low-cost manufacturing. However, instabilities in the materials and devices, primarily due to reactions with water, have kept PSCs from entering the PV marketplace.

In this chapter, we use laser beam induced current (LBIC) imaging to investigate the spatial and temporal evolution of the quantum efficiency of state-of-the-art PSCs under controlled humidity conditions. Several interesting mechanistic aspects are revealed as the degradation proceeds along a four-stage process involving with phase transformation of the perovskite material. Three of the four stages can be reversed, while the fourth stage leads to irreversible decomposition of the photoactive perovskite material. A series of reactions in the PbI2-CH3NH3I-H2O system explains the interplay between the interactions with water and the overall stability. This allows a detailed understanding of the mechanisms and evolution of moisture-induced perovskite degradation to be developed on a microscopic scale. Understanding of the degradation mechanisms of PSCs on a microscopic level gives insight toward improving the long-term stability. 157

8.1 Introduction and Motivation

Organic-inorganic perovskite solar cells (PSCs), mostly based on methylammonium lead iodide (CH3NH3PbI3), have shown remarkable progress in the past five years. In addition to power conversion efficiencies exceeding 22%,[22] PSCs are easily processed and promise low manufacturing costs.[15, 93, 94, 96] Consequently, PSCs are now considered as a viable alternative to other more established photovoltaic (PV) technologies such as crystalline Si, CdTe, and CuIn1-xGaxSe2. While these latter technologies have been successfully commercialized, large-scale deployment of PSCs is hindered by performance degradation. Although ~4 months of stability has recently been demonstrated,[254] a 25 -

30 year lifetime is required to achieve a low levelized cost of energy (LCOE) which will allow PSCs to compete with other more established PV and conventional power generation technologies.

Tremendous effort has been devoted to understanding the decay mechanisms and developing appropriate solutions to improve PSC device stability.[92, 254, 255]

Degradation can be due to humidity,[256] oxygen,[257] light,[136] and heat.[258] In nearly all studies, water has been implicated as a key culprit in causing instability in materials and devices. In early studies, the degradation of PSCs was attributed to hydrolysis reactions in which water molecules reacted with organic species in the parent material to drive

decomposition and the release of gas phase HI and CH3NH2.[215] More recently, it has been revealed that hydrated perovskite phases, CH3NH3PbI3 ∙ H2O and

(CH3NH3)4PbI6 ∙ 2H2O, may be formed during the initial degradation process.[259, 260]

Interestingly, the formation of the monohydrate phase can be fully reversed by exposing the hydrated perovskite to a dry environment.[260] To date, PSC degradation has been 158

investigated by monitoring the current-voltage (J-V) behavior as a function of time under controlled temperature and humidity conditions.[254] Changes in the PV performance were correlated with general structural and compositional changes in the materials.[254]

While instructive, this approach provides little information about the mechanisms and processes that control the causes and evolution of degradation on a microscopic level.

The understanding of the effects of spatially inhomogeneous chemical composition and crystallinity is particularly critical for designing strategies to improve the stability of PSCs.

To elucidate the details of the failure mechanisms and develop pathways to produce stable devices, we performed in situ laser beam induced current (LBIC) mapping on state-of-the- art PSCs in the presence of moisture. Very few LBIC studies have been performed on PSCs to date,[31, 32, 261] and this is the first detailed LBIC investigation of perovskite degradation processes.

8.2 Experimental Details

8.2.1 Perovskite Device Fabrication

The perovskite devices used for this study were fabricated at École Polytechnique

Fédérale de Lausanne (EPFL). To fabricate the devices, fluorine-doped tin oxide (FTO) coated glass substrates were cleaned sequentially with Hellmanex soap in an ultrasonic bath for 30 min, then washed with acetone, isopropanol and finally cleaned in an oxygen plasma for 5 min. A ~30 nm thick TiO2 compact layer was deposited on FTO by spray pyrolysis at 450 °C from a precursor solution prepared with acetylacetone (0.4 mL,

Aldrich), titanium diisopropoxide bis(acetylacetonate) solution (0.6 mL, 75% in 2-

159

propanol, Aldrich) and ethanol (9 mL). A ~150 nm thick mesoporous TiO2 layer was deposited by spin coating for 10 s at 4000 rpm with a ramp of 2000 rpm s-1, using 30 nm particle paste (Dyesol) diluted in ethanol. After spin coating, the substrate was sintered at

500 °C for 30 min, under dry air flow. Mesoporous TiO2 was then doped with lithium by spin coating a solution of bis(trifluoromethanesulfonyl)imide (Li-TFSI, 0.1 M, Aldrich) in acetonitrile at 3000 rpm for 30 s.[262] After spin coating, the substrate was sintered again at 500 °C for 30 min, under dry air flow. After cooling down to 150 °C, the substrates were immediately transferred in a nitrogen atmosphere glove box for depositing the perovskite films.

The perovskite films were deposited from a precursor solution containing methylammonium iodide (1.1 M) and lead iodide (1.2 M) in anhydrous dimethylformamide:dimethyl sulfoxide (DMF:DMSO) 4:1 (v:v).[263] In addition to I- based perovskite, CH3NH3PbBr3 films were prepared using a similar solution consisting of methylammonium bromide (1.1 M) and lead bromide (1.2 M). The perovskite solution was spin coated in a two steps, at 1000 and 4000 rpm for 10 and 30 s respectively. During the second step, chlorobenzene (100 μL) was poured on the spinning substrate 15 s prior the end of the program. The substrates were then annealed at 100 °C for 1 hour in the nitrogen filled glove box.

After the perovskite annealing, the substrates were cooled down for few minutes and a spirofluorene linked methoxy triphenylamines (spiro-OMeTAD, from Merck) solution was spun on at 4000 rpm for 20 s. The spiro-OMeTAD solution (0.07 M) was prepared in chlorobenzene, and doped with Li-TFSI (50 mol%) from a stock solution of Li-TFSI (1.8

M) in acetonitrile, tert-butylpyridine (tBP, 330 mol%, Aldrich) and Tris(2-(1H-pyrazol-1- 160

yl)-4-tert-butylpyridine)- cobalt(III) Tris(bis(trifluoromethylsulfonyl)imide) (Co-complex,

3 mol%, Dyesol) from a stock solution Co-complex (0.25 M) in acetonitrile.[264, 265] For the devices with an alternative hole transporting materials (HTM), a ~20 nm poly triarylamine (PTAA) film was spin coated using the similar method. Finally, 80 nm of gold was deposited by thermal evaporation under high vacuum, using a shadow masking to pattern the electrode.

8.2.2 In-situ LBIC Measurement

An air-tight environmental chamber was built to study the degradation of perovskite solar cells in humidity. The box was purged with a carrier gas (air or N2) through a water bubbler that was kept at 30 and 60 °C to maintain a 50 ± 5 and 80 ± 5% relative humidity

(RH), respectively.[32] The RH value was monitored by a commercial hygrometer stored in an air-tight container that connected to the outlet of the environmental box. Time- resolved LBIC maps were collected in 6 min intervals to study the evolution of the device degradation. Each LBIC map was constructed by measuring the photogenerated current induced by a 40 µm wide 532 nm laser beam that was scanned across the device. The laser power density was adjusted to be ~0.01 mW/cm2 to avoid heating. The measured current values were converted to EQE using a calibrated Si reference cell. A total of eight devices are discussed in this report, four of which were examined under 50% RH and four that were examined under 80% RH. The behavior was repeatable under each environmental condition, and consistent with other measurements that are not part of this study.

161

8.3 Degradation Behaviors of Perovskite Solar Cells

8.3.1 Four-Stage Process of Degradation

Spatially resolved EQE maps were obtained at 25 °C using LBIC. The EQE maps

Figures 8-1 and 8-2 show four distinct behaviors for devices exposed to a RH of 50% (1.6%

H2O). During Stage 1 the EQE of the PSCs increased slightly and peaked after a few minutes of exposure to humidity. During Stage 2, a rather uniform drop in photocurrent collection efficiency was observed over a time period of 18 to ~105 min. Stage 3 is characterized by the propagation of a front that sweeps from the periphery of the device over ~2 hours and a dramatic reduction in the EQE. Stage 4 occurred more slowly and, surprisingly, displayed a temporary increase in the EQE at some locations in the device.

Note that although variations from sample to sample were observed, all devices showed the same general behaviors.

A thorough analysis of the findings provides an understanding of the processes underlying each stage. In Stage 1, the EQE is not constant across the entire area even for these near-record devices (Figure 8-1a). Despite the fact that great care was taken to form uniform layers during processing, and to eliminate moisture during device manufacture, defective areas are observed as well as significant spatial variation in the EQE from ~60 to

~80% in regions of the device that appear to be defect-free. Thus, there is significant potential for improving the performance of these devices simply by improving the uniformity of the current collection. For example, if the EQE across the entire area of the device was optimized to give the maximum value (83%), this particular 16.2% device would have an efficiency of 18.5%.

162

Figure 8-1: EQE maps (at 532 nm) of a typical perovskite solar cell after exposure to 50

± 5 % RH for (a) Stages 1-3 (0 - 480 min) and (b) Stage 4 (480 – 1320 min)

shown with a higher resolution color scale.

163

Figure 8-2: Areal average LBIC EQE (at 532 nm) as a function of time after exposure to

humidity.

Although not easily discerned in the maps of Figure 1a, the spatially averaged EQE shown in the inset of Figure 8-2 exhibits a clear peak approximately 5 - 10 min. after the onset of humidity exposure. This somewhat surprising result can be explained by considering that recombination centers at the interface between spiro-OMeTAD and perovskite can be passivated by water molecules. Recently, it has been reported that uncoordinated ionic species on perovskite grain boundaries can be deactivated by hydrogen bonding with water, leading to reduced surface trap states and better carrier extraction.[266] Water has also been found to improve the crystallinity and morphology of films during deposition, leading to better device performance.[267] In the present case, the devices were prepared under rigorously dry conditions and the performance rapidly

164

improved upon exposure to water vapor. The route for water ingress could be diffusion along the perovskite/spiro-OMeTAD interface from the edges of the device. Note that the gold back-contact is dense and continuous and, therefore, should be impervious to water.

The LBIC results of Stage 2 are characterized by a uniform decline in the photocurrent across the device. The EQE decreases quickly during the first hour and then stabilizes before the onset of Stage 3. While the background EQE decreases, relatively large defects, such as the one seen at the left of the Stage 2 images in Figure 8-1a, may partially heal, presumably due to passivation by water, as seen in Stage 1. Overall, the spatially averaged data in Figure 1b shows that the EQE drops from ~72% to ~56% during Stage 2. To explain this we first consider that the conductivity in spiro-OMeTAD is produced by doping, which is achieved by oxidizing the molecular backbone.[264] In the present case, a lithium salt

(e.g. Li-TFSI) is used to promote an oxidative reaction between pristine spiro-OMeTAD and oxygen to produce mobile holes in the organic matrix.[264, 268, 269] Water incorporation has been shown to affect the electronic structure and hole mobility of doped spiro-OMeTAD in field effect transistor measurements,[270, 271] so it is reasonable to expect water to impact the ability of spiro-OMeTAD to collect holes in the PV device configuration as well. The time constant for the phenomena responsible for Stage 2 is much slower than the time required for Stage 1 enhancement, so the water transport mechanism is likely to be different. The spiro-OMeTAD layer has an amorphous structure and possesses internal channels that can efficiently transport O2 and H20,[254, 269] but the rate of transport is evidently slower than the transport along the perovskite/spiro-OMeTAD interface. Because the degradation occurs in a rather uniform manner across the area of the device, the water concentration in the film at any time should also be uniform. 165

To develop more insight, we performed optical transmission spectroscopy (Figure 8-

3a) and observed changes in the optical transmittance that were consistent with water incorporation and a loss in the degree of oxidation of the spiro-OMeTAD.[268] Fourier transform infrared spectroscopy (Figure 8-3b) revealed reversible water-induced quenching of the N-H+ functionality that is indicative of hole conductivity. Consequently, it seems clear that the change in EQE can be ascribed to bulk changes in the doping state and hole collection efficiency of the spiro-OMeTAD layer.

After stabilizing at the end of Stage 2, Stage 3 commences at ~105 min with a very large drop in EQE from ~56% to a plateau at ~6% where the EQE stabilizes once again. In contrast to Stages 1 and 2, Stage 3 degradation begins primarily at the top of the image and progresses downward (Figure 8-1a). In some cases, the degradation front propagates from the edges towards the center of the device (Figure 8-4). The speed at which the degradation front spreads is determined by the humidity, regardless of whether the carrier gas is dry air or N2, indicating that the degradation reaction is indeed related to the concentration gradient of water. With the capability to resolve the speed of propagation of the degradation front, which most likely corresponds to the rate of diffusion of water molecules in the perovskite film, degradation rates of 0.36 and 1.48 μm/s under 50 and 80% RH, respectively, were determined (Figure 8-5).

166

Figure 8-3: Transmittance and ATR-FTIR spectra of the stacked films of spiro-OMeTAD

/CH3NH3PbI3/TiO2/FTO after exposure to 50 % RH.

Figure 8-4: LBIC maps of perovskite devices aging under moist (a) air or (b) N2 flow of 80 ± 5

% RH. The time intervals are 12 min.

167

Figure 8-5: Contour plots of the degradation front propagation of perovskite devices aged under

(a) 50% RH and (b) 80% RH during Stage 3.

Figure 8-6 shows a schematic diagram of the cell as well as optical micrographs that show several of the device layers near a scribe line that is used to expose the front contact of the device for electrical connection. Evidently, Stage 3 response for the sample of Figure

1a is due to enhanced water ingress at the exposed CH3NH3PbI3 surface. In situ J–V measurements (Figure 8-7) shows that formation of the monohydrated perovskite phase dramatically decreases the short circuit current (JSC) and fill factor (FF) of the PSCs.

168

Figure 8-6: (a) Photo of a perovskite solar cell and microscopic optical image of the scribing

line close to the anode. (b-e) Schematics of water ingress into a perovskite solar cell.

169

Figure 8-7: J-V curves of a perovskite device aging under a moist N2 flow of 80 % RH.

The time intervals are 12 min.

After the first three stages of degradation (~330 min), the average EQE across the device was decreased from 71% to 6%, resulting in a completely failed device. These first three stages are, however, fully reversible (vide infra). These three stages, however, are followed by an irreversible fourth stage that has not yet been investigated in detail. During

Stage 4, the average EQE begins to increase at ~500 min from the Stage 3 plateau value of

6% to a maximum of ~14% at 900 min. At longer times, the EQE then declines to 0%

(Figure 2).

Figure 8-2 shows the Stage 4 response with an increased EQE contrast. As the EQE begins to peak we see individual grains within the device becoming more effective at generating significantly higher photocurrents than the average for the cell. After ~700 min, 170

these spots increase in intensity (higher EQE), while the background currents decreased.

The high-intensity EQE at spots indicates that the material at specific locations is becoming more effective at photocurrent generation as the degradation proceeds, possibly due to disproportionation and phase separation to produce unhydrated perovskite grains and poorly photoactive materials such as perovskite hydrates or PbI2. With continuing hydration (beginning at ~930 min) these unhydrated perovskite grains are evidently consumed as well. Then, the inactive regions expand, leading to morphological changes and, subsequently, to complete device failure (at ~1300 min). At this point, the color of the material has changed from dark brown to yellow.

8.3.2 Reversible Hydration Process

Stages 1 and 2 are clearly determined by the changes in charge carrier extraction

(surface recombination) and transport (hole mobility) properties of the spiro-OMeTAD layer and the perovskite interfaces, suggesting possible improvements through the development of improved hole transporting materials. Stages 3 and 4 are, on the other hand, related to reactions between water and the perovskite material and the resultant reactions and phase transformations. The hydration of perovskite responsible for Stage 3 behavior is a reversible reaction.[260] When the humidity is reduced, the metastable monohydrated perovskite phase loses crystalline water and converts back to CH3NH3PbI3.[162]

To investigate this reversibility with LBIC we mapped the photocurrent generation from devices subjected to hydration-dehydration cycles. Figures 8-8 and 8-9 show images for a device that was hydrated under a RH of 80 ± 5 % (3.4% water) and recovered by purging with dry air. Note that the higher humidity leads to a faster degradation rate 171

(Figures 8-8). In this case, the feature of Stage 2 cannot be separately distinguished due the rapid transition to Stage 3. This finding highlights that the rates of the two processes have different dependencies on the water concentration. When the atmosphere above the device was subsequently purged with dry air for ~5 hours, the photocurrent completely recovered.

Unlike the hydration which propagated along water diffusion pathways, the recovery of the photocurrent initiated at nuclei of recrystallized de-hydrated perovskite grains and spread outward from there as the photocurrent across the entire device increased gradually.

Figure 8-8: LBIC EQE maps of a typical perovskite solar cell after exposure to a moist

N2 flow of 80 ± 5 % RH.

Figure 8-9: LBIC EQE maps of hydrated perovskite device while purging with dry air. 172

Figure 8-10: Areal average EQE (at 532 nm) of the perovskite device during hydration-

dehydration cycles.

The hydration-dehydration cycles were repeated several times without significant loss of current generation (Figure 8-10). With multiple, deep hydration/dehydration cycles some degree of irreversibility begins to appear as defects form in the recrystallized

CH3NH3PbI3 grains. Under these conditions, morphological changes arise during the recrystallization of perovskite crystals, along with permanent deterioration in the current generation.[272]

Note that the degradation caused by the hydration of CH3NH3PbI3 is reversible only for the monohydrate phase that forms after short-time exposure to a low vapor pressure of water. Longer exposure times or exposures to higher H2O vapor pressures can quickly lead to irreversible damage to the devices. Therefore, the Stage 4 degradation must consist of at least one irreversible reaction, most likely a water-catalyzed decomposition that leads to solid PbI2 and volatile products.[215, 259, 260, 272]

173

8.4 Equilibria in the Ternary PbI2-CH3NH3I-H2O System

Here, we propose a quasi-ternary CH3NH3I-PbI2-H2O phase system (Figure 8-11) to explain the relationships between the chemical composition changes of the materials and optoelectronic response of the devices.

Figure 8-11: Schematic diagram of the phase equilibria in the PbI2-CH3NH3I-H2O

system.

As described by Equations 8.1 to 8.4, four primary reactions can be considered; (a) the perovskite formation reaction, (b) the reversible formation of a monohydrated chains of perovskite, (c) the formation of dihydrated perovskites with reduced dimensionality, and

(d) water-catalyzed decomposition to return to the starting components.

174

푃푏퐼2 + 퐶퐻3푁퐻3퐼 ⇔ 퐶퐻3푁퐻3푃푏퐼3 (8.1)

퐶퐻3푁퐻3푃푏퐼3 + 퐻2푂 ⇔ 퐶퐻3푁퐻3푃푏퐼3 ∙ 퐻2푂 (8.2)

(4 − 푛) 퐶퐻3푁퐻3푃푏퐼3 + n 퐶퐻3푁퐻3푃푏퐼3 ∙ 퐻2푂 + (2 − n)퐻2푂

⇔ (퐶퐻3푁퐻3)4푃푏퐼6 ∙ 2퐻2푂 + 3 푃푏퐼2 (8.3)

(퐶퐻3푁퐻3)4푃푏퐼6 ∙ 2퐻2푂 ⇔ 푃푏퐼2 + 4 퐶퐻3푁퐻3퐼 + 2 퐻2푂 (8.4)

Equation 8.1 (Figure 8-11a) specifies the basic formation (forward) and thermal decomposition (reverse) reactions of stoichiometric CH3NH3PbI3 in the absence of water.

While there is a rather wide composition space for high-efficiency perovskite solar cells,[149] here we only consider the stoichiometric perovskite phase. When a low concentration of water is introduced into the system, perovskite crystals can react to form monohydrated chains through Equation 8.2 (illustrated in Figure 8-11b). The reverse reaction can be driven by a reduction in the partial pressure of water.

The reaction between the monohydrate and dihydrated compounds occurs when the water vapor pressure is increased (Equation 8.3 and Figure 8-11c). The dimensional reduction from the 1D monohydrated chains to the 0D dihydrated perovskite phases requires the involvement of a larger number of CH3NH3 cations and I anions, resulting in the formation of PbI2 as a byproduct. This process redistributes organic and inorganic species, which causes an inhomogeneous composition, alters the volume of the grains and the morphology of the film, and subsequently degrades the optoelectronic properties of the

+ - devices. Meanwhile, the excess ions (CH3NH3 and I ) in the 0D dihydrated phase are able to migrate to neighboring locations containing PbI2 nuclei, leading to the formation of

175

CH3NH3PbI3. Thus, certain locations in the film can participate in the actual formation reaction (Equation 8.1) once again if the local water concentration is low. Under these conditions, individual CH3NH3PbI3 grains can co-exist with hydrated phases and PbI2 in locally phase-separated regions. This phase separation likely accounts for the increase in photocurrent at local spots the first part of Stage 4. If some of the reaction products are able to escape, the reaction becomes irreversible. For example, with gas flow removing

CH3NH3I from the solid-state system, the reaction of Equation 8.4 moves to the right resulting in the irreversible formation of PbI2 (Figure 8-11d).

In support of this mechanism, we note that white CH3NH3I residue was found at the gas outlet of our environmental chamber after ~30 devices were measured. As an aside, this suggests that CH3NH3I does not further decompose into CH3NH2 and HI in the presence of water as was suggested in early work,[215] which is consistent with reports of

CH3NH3I stability in aqueous solutions.[273]

8.5 Impact of Charge Transporting Materials and Perovskite Compositions

Understanding of the reactions in the PbI2-CH3NH3I-H2O system allows the stability of the perovskite thin films to be predicted under different conditions. However, the perovskite layers are very thin (~ 300 nm) so the interfaces with the charge transporting materials (TiO2 and spiro-OMeTAD) could alter the kinetics and thermodynamics of the system’s equilibria. To investigate the impact of the hole transporting material on the reaction between water and CH3NH3PbI3, we compared the optical absorbance of two different film stacks, one with and one without a spiro-OMeTAD layer, at RH = 80%.

Surprisingly, the FTO/TiO2/CH3NH3PbI3 sample was stable and exhibited no significant 176

change in absorption at the perovskite band (~1.6 eV) edge after 8 h (Figure 8-12a). From this result, it seems that the TiO2 layer may render the perovskite film resistant to water uptake. This stability enhancement is likely due to the improved crystallinity of the perovskite film when deposited on the TiO2 layer instead of a glass substrate. In contrast, the FTO/TiO2/CH3NH3PbI3/spiro-OMeTAD sample started to degrade immediately after water was introduced into the chamber (Figure 8-12b). Note that the loss of the band edge absorption is evidence for the formation of both mono and dihydrated perovskite species, both of which have a bandgap of 3.1 eV.[260] In addition, the increased absorption of photons with energies < 1.5 eV indicates the presence of water in the film stacks. Clearly, the presence of spiro-OMeTAD plays a role in accelerating the water ingress process and the decomposition of the perovskite material.

Figure 8-12: Optical absorbance spectra of (a) FTO/TiO2/CH3NH3PbI3 and (b)

FTO/TiO2/CH3NH3PbI3/Spiro-OMeTAD aged under RH = 80%.

177

Although it has been reported that both spiro-OMeTAD itself and the hygroscopic Li-

TFSI additive contribute to the instability of CH3NH3PbI3,[259, 274] the microscopic nature or origin of the accelerated degradation has not been clear. Combining the in-situ

LBIC and optical absorbance measurements make it clear that the degradation is inhomogeneous (Figure 8-1b). We speculate that the internal channels within the hygroscopic spiro-OMeTAD layer may provide microenvironments of varying effective water concentration in contact with the perovskite films. This inhomogeneous water distribution would provide spatial variation that allowed partial or full hydration in some locations (Equations 8.2 and 8.3), while other nearby environments with locally low H2O concentrations could pull Equation 8.4 to the right which could in turn drive the

CH3NH3PbI3 formation reaction (Equation 8.1). The fact that small molecules like H2O and CH3NH3I can be freely transported in open channels within spiro-OMeTAD suggests that the microenvironments could be in communication with one another to support the behavior observed in Stage 4. This speculation is consistent with X-ray diffraction data which shows that the dihydrated phase is more likely to form when spiro-OMeTAD is present (Figure 8-13).

178

Figure 8-13: XRD spectra of (a) FTO/TiO2/CH3NH3PbI3 and (b) FTO/TiO2/

CH3NH3PbI3/Spiro-OMeTAD aged under RH = 80% for 30 min.

As an alternative to spiro-OMeTAD, PTAA is widely used in perovskite solar cells and demonstrates high device efficiency. To probe the impact of the hole transport materials

(HTM) on the stability of perovskite devices, we compared the perovskite devices with spiro-OMeTAD and PTAA as the HTM. Figure 8-14 shows the comparison of the LBIC map evolution of two I-based perovskite devices measured simultaneously. It is clear that the spiro-OMeTAD device degraded much more rapidly than the PTAA device.

179

Figure 8-14: Evolution of LBIC maps of I/PTAA and I/Spiro devices after exposure to

80% RH. The time span between each frame is 12 min.

As previously discussed, the rapid degradation of the spiro-OMeTAD devices is due to the water ingress that started on the edge of the device and rapidly propagated inward to the center of the device. The formation of monohydrate perovskite (CH3NH3PbI3 ∙ H2O) as a result of the interaction with water dramatically changed the optoelectronic properties of the perovskite materials, leading to the decrease in current collection efficiency. In contrast, the PTAA device degraded in a different approach that the photocurrent generation in the

PTAA device decreased gradually and uniformly across the whole device area. Unlike the hygroscopic spiro-OMeTAD, the hydrophobic PTAA polymer chains can impede water ingress. The slow degradation rate is likely due to water permeation through pinholes in the thin (~20 nm) PTAA film. This result indicates that water resistive HTMs can improve

180

the stability of perovskite solar cells under high humidity, showing great potential for long- term stability when the perovskite/PTAA device structure is optimized.

To assess the role of the perovskite composition on device stability we measured the temporal evolution of the LBIC maps of Br-based perovskite devices (Figure 8-15).

Compared with the I-based devices, the Br-based devices exhibited improved stability under the high humidity condition due to higher resistance to water.[117]

Figure 8-15: Evolution of LBIC maps of Br/PTAA and Br/Spiro devices after exposure to

80% RH. The time span of each frame is 60 min.

The Br/PTAA device degraded earlier and faster than the Br/Spiro device, likely due to pin-hole formation on the Au back contact of the Br/PTAA sample (Figure 8-16).

Although the dynamics of water ingress through the HTMs should be the same, the reaction

181

between water and Br-based perovskite occurs at a much slower rate. This indicates a different degradation mechanism for the Br-based devices.

Figure 8-16: SEM image of aged (a) Br/PTAA and (b) Br/Spiro devices.

For both Br-based devices, degradation first occurred in the neck region (anode) and propagated toward the back of the devices. This may be related to the accumulation of

- - 2- negative ionic defects (Bri , VMA and VPb ) close to the anode. These defects favor the formation of a volatile organic halide and may stimulate the decomposition of the perovskite. It is also possible that Cu diffused from the external contact into the device and deteriorated the perovskite materials.

Figure 8-17 summarizes the evolution of integrated LBIC of the perovskite devices.

Based on the results, we hypothesize 3 mechanisms underlying the perovskite device degradation. (1) Rapid water ingress through the Spiro-MeOTAD accelerates the degradation of I-based perovskite. (2) Deformation of a thin PTAA film leads to pin-hole formation and opens pathways for water to enter. (3) Ionic defect (both intrinsic and extrinsic) accumulation on the anode side causes the decomposition of the perovskite materials, especially the Br-based perovskites. Thus, to enhance water resistance and 182

improve the stability of perovskite devices, a hydrophobic HTM with good electrical and mechanical properties is needed.

Figure 8-17: Degradation of integrated LBIC signal of the perovskite devices as a

function of exposure time.

8.6 Approaches to Improve Devices Stability

The phase equilibria in the PbI2-CH3NH3I-H2O system provides a comprehensive understanding of the stability of CH3NH3PbI3 in the presence of water, and, coupled with an understanding of water interactions with spiro-OMeTAD, can be used to understand device stability. To fabricate and maintain high efficiency devices it is critical to (1) reduce the water partial pressure to avoid the dimensional reduction from 1D monohydrate to 0D dihydrates, (2) avoid the use of hygroscopic materials in the device (e.g. spiro-OMeTAD), and (3) seal the device so that the volatile species cannot escape. Clearly, water impermeable encapsulation would provide an effective method to prepare devices with 183

long-term stability. Note that device was stable in dry air and exhibit no significant change in the EQE map after ~1000 h (Figure 8-18). However, due the reversible nature of Stages

2 and 3, PSCs are compatible with processing in moderately humid air so long as the device is dried prior to encapsulation.

Figure 8-18: LBIC EQE maps of a typical perovskite solar cell after exposure to dry air.

Another feasible approach to improve the perovskite device stability against humidity is to apply a hydrophobic thin film to avoid direct expose of the perovskite or HTM to the humid air. The addition of a poly(methyl methacrylate)/single-wall carbon nanotube

(PMMA/SWCNT) encapsulation layer can prevent degradation of the device in the moist air.[32] We have shown that improve the stability of perovskite device against humidity

(Figure 8-19). This suggests a route toward perovskite solar cells with improved operational stability and moisture resistance.

184

Figure 8-19: (a) photo of SWCNT/perovskite film stack. (b) Degradation behaviors of

perovskite/P3HT and perovskite/SWCNT in humid air.

8.7 Conclusion

In summary, we have investigated the spatial evolution of the EQE during the exposure of perovskite solar cells to humid atmospheres using LBIC measurements. The results show that the device behavior proceeds through 2 stages that are controlled by water- induced carrier extraction and transport changes in the spiro-OMeTAD layer and 2 stages that are due to phase transformation in the perovskite material. The changes in solar cell performance are explained by both reversible and irreversible processes. We also observed different degradation behaviors of the devices with different HTMs and compositions.

On the basis of these observations, we propose phase equilibria within the PbI2-

CH3NH3I-H2O system to understand the water-induced degradation and suggest that a dehydration process before device encapsulation will potentially improve the stability of perovskite solar cells. In addition, we propose that a thicker and optimized PTAA and an

185

optimized perovskite composition may greatly enhance the stability of perovskite solar cells.

186

Chapter 9

Summary and Future Research

9.1 Thesis Summary

Solution processed copper indium chalcogenides and methylammonium lead halides are PV technologies that appear to be worthwhile for further development towards producing high efficiency, low-cost photovoltaic power generation systems. Both technologies now achieve efficiency and cost levels that suggest that they may be able to compete with established PV cells such as CdTe and c-Si with further advances. Further research is required to develop a better understanding of the underlying physics and chemistry of the materials and devices and a better control over the fabrication processes.

The final goal is to produce low cost, high efficiency thin-film solar cells with good stability.

This thesis has investigated the solution processing of two different types of thin film solar cells. The study started with the inorganic chalcogenide thin film solar cells based on

CuIn(S,Se)2, and then extended to the new organic-inorganic CH3NH3PbI3 solar cells. The possibility to improve the manufacturability by using ultrasonic spray deposition for scaling up and allowing better control of stoichiometry of the materials was explored. LBIC mapping techniques were developed to study the spatial uniformity of photocurrent generation and collection in solution-processed thin film devices. The phase formation and 187

phase stability of the CH3NH3PbI3 perovskite materials were investigated to understand nucleation and growth and to develop insight to enhance the stability of the materials and devices.

9.2 Future Research

There are several interesting future research topics related to the work presented in this thesis. Some are ongoing projects and others are expected to continue as part of new efforts.

Potential research includes assessing the (1) environmental and economic impacts of thin- film solar cells manufacturing, (2) improving the stability of perovskite solar cells, (3) exploring solution-processed solar cells on flexible substrates, (4) CuInSe2/perovskite tandem solar cells, and (5) developing low-cost processing methods for perovskite mini- modules. These future works are summarized as follows.

9.2.1 Environmental and Economic Impact Assessments

Although the stability of the materials and devices is a concern, low-cost, high- efficiency perovskite solar cells have a great potential to enter the global PV market. To determine the future viability of perovskite solar cells, it is important to determine and compare the environmental and economic impacts of perovskite photovoltaic devices that have may be fabricated with a wide diversity of materials, preparation methods, and device structures. Rather than examining the spin coating process that is widely used in laboratories, we will consider spray or printing that are more amenable to scaling up low- cost manufacturing. The results can be used to compare with the environmental and

188

economic impacts of conventional c-Si PV technology, and thus provide a guideline for improving device and process designs for industrial level production.

Our preliminary life cycle assessments show that the environmental impacts from manufacturing of perovskite solar modules were lower than that of c-Si.[275] The energy payback time (EPBT) of perovskite PV technologies was estimated as 1 to 1.5 years and the cost of manufacturing perovskite module could be as low as $0.20 per watt, making perovskite solar cells competitive with mainstream electricity generating technologies in the foreseeable future. We also found the toxicity impacts of the lead used in the formation of the absorber layer were negligible.

Although the estimated PV module manufacturing costs are lower than that of established PV technologies, when the environmental impacts in terms of unit electricity generation were considered, perovskites were not as good as the other commercial PV technologies mainly because of the shorter lifetime. This result reveals the importance of improving the lifetime and stability of solution-processed thin film solar cells.

9.2.2 Improving the stability of perovskite solar cells

Our study on the partially reversible phase transitions of perovskite materials in humid air revealed the microscopic mechanisms of the water and perovskite reaction that is detrimental to the perovskite device stability.[33] Based on the results, it is clear that perovskite solar cells should be protected against moisture. One interesting future research area is developing a hydrophobic and impermeable coating that encapsulates the whole device to protect vulnerable perovskite and HTM. Potential coating materials include metal oxides, organic polymers, and nanomaterials. Besides encapsulation, designing and 189

engineering the perovskite materials to tune the thermal and chemical stability of the materials is also important to solve the intrinsic instability issues associated with the prototypical CH3NH3PbI3 material. Future studies will also include a comprehensive study of in situ measurement of perovskite device degradation under different conditions (e.g., illumination and heat), allowing for a better understanding of the origins of materials and device degradation.

9.2.3 Solution-processed solar cells on flexible substrates

A new approach to the design device architecture for perovskite solar cells is based on the use of flexible and light-weight substrates (e.g., flexible plastic foils). Such PV devices are of commercial interest for low-cost, large-scale roll-to-roll processing and applications as portable power sources and building/vehicle integrated materials. In the last two years, a great deal of effort has been made on perovskite PV devices on flexible substrates such as poly(ethylene terephthalate) (PET),[196, 276-279] polyethylene naphthalate

(PEN),[280] Ti foils,[281] Flexible perovskite PV mini-modules have demonstrated the potential to transfer laboratory-based perovskite techniques to industrial roll-to-roll processing,[282] and have been used to power aviation models.[283]

Future research is expected to focus on developing solution-processing techniques to control stoichiometric, phase, and morphological uniformities over a large area during the perovskite film formation. In addition, it will also be interesting to investigate the lifetime and degradation behaviors in different device structures on different flexible substrates.

190

9.2.4 Perovskite/CuInSe2 Tandem Solar Cells

Because perovskites have a tunable bandgap (Eg = 1.5 to 2.3 eV) and high VOC, there is great interest in incorporating them in tandem devices with crystal silicon or CIGS cells.[284] It is predicted that the ultimate efficiency of the monolithic tandem perovskite devices can exceed 35% in future.[92] Several designs of perovskite-based tandem devices have been reported, including two-terminal monolithic devices and four-terminal devices

(with a light splitting component).[252, 284-287] However, the overall efficiencies, 19.5% for the best four-terminal perovskite/CIGS device[287] and 18% for the best two-terminal perovskite/Si device,[288] are still lower than that of the state-of-the-art single junction perovskite devices due to the electric loss at the tunneling junction and the transparent electrode.

Practical fabrication of monolithic perovskite tandem device is challenging because the device efficiency is strongly determined by the choice of the materials and the processing methods. In particular, the tunneling layer needs to be adequately adjusted so that the optical transparency and electrical conductivity can be appropriately matched to both the top and bottom cells. Additionally, the high energy associated with the sputtering of the transparent electrode (ITO or AZO) may deteriorate the perovskite layer. Therefore the deposition processes will need to be well controlled to prevent any degradation of perovskite or organic HTM layers.

Regarding the potential application of this thesis work, the monolithic tandem solar cells consisting of solution-processed CH3NH3PbI3 and CuInSe2 sub-cells have a great potential to achieve a high efficiency. Although there are still challenges on the solution- processing of each cell, the advances in the understanding and development of solution- 191

processing techniques would lead to the breakthrough of both PV technologies and tandem devices with improved performance.

9.2.5 Fabrication of Perovskite Mini-Modules

Ever since the high efficiency (>20%) were demonstrated in the small area perovskite solar cells, the commercialization of solution-processed perovskite solar cells have become a potentially viable research topic and future direction of the development of perovskite

PV technology. The manufacturing of perovskite solar modules requires a scaling up solution-based approach rather than the spin coating method. The follow-up work of this thesis research will focus on developing a feasible large-area deposition technique that can be used to fabricate perovskite mini-modules. That is the same as the goal of this thesis: to develop low-cost, solution-processing methods for the fabrication of high-efficiency thin film solar cells.

192

References

[1] D. R. Feldman, W. D. Collins, P. J. Gero, M. S. Torn, E. J. Mlawer, and T. R. Shippert, "Observational determination of surface radiative forcing by CO2 from 2000 to 2010," Nature, vol. 519, pp. 339-343, 2015.

[2] US Energy Information Administration (EIA) (31 July 2016), "Annual energy outlook 2016 ", Available: http://www.eia.gov/forecasts/aeo/pdf/0383(2016).pdf.

[3] Solar Energy Industries Association (SEIA) (1 May 2016), "Solar industry data ", Available: http://www.seia.org/research-resources/solar-industry-data.

[4] 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, pp. 676-677, 1954.

[5] Fraunhofer Institute for Solar Energy Systems (11 March 2016), "Photovoltaic report ", Available: https://www.ise.fraunhofer.de/en/downloads-englisch/pdf- files-englisch/photovoltaics-report-slides.pdf.

[6] K. Masuko, M. Shigematsu, T. Hashiguchi, D. Fujishima, M. Kai, N. Yoshimura, T. Yamaguchi, Y. Ichihashi, T. Mishima, N. Matsubara, T. Yamanishi, T. Takahama, M. Taguchi, E. Maruyama, and S. Okamoto, "Achievement of more than 25% conversion efficiency with crystalline silicon heterojunction solar cell," IEEE Journal of Photovoltaics, vol. 4, pp. 1433-1435, 2014.

[7] M. A. Green, K. Emery, Y. Hishikawa, W. Warta, and E. D. Dunlop, "Solar cell efficiency tables (version 47)," Progress in Photovoltaics: Research and Applications, vol. 24, pp. 3-11, 2016.

[8] M. Graetzel, R. A. J. Janssen, D. B. Mitzi, and E. H. Sargent, "Materials interface engineering for solution-processed photovoltaics," Nature, vol. 488, pp. 304-312, 2012.

[9] C. J. Hibberd, E. Chassaing, W. Liu, D. B. Mitzi, D. Lincot, and A. N. Tiwari, "Non-vacuum methods for formation of Cu(In, Ga)(Se, S)2 thin film photovoltaic absorbers," Progress in Photovoltaics: Research and Applications, vol. 18, pp. 434- 452, 2010.

193

[10] R. M. Pasquarelli, D. S. Ginley, and R. O'Hayre, "Solution processing of transparent conductors: From flask to film," Chemical Society Reviews, vol. 40, pp. 5406-5441, 2011.

[11] M. K. Nazeeruddin, E. Baranoff, and M. Grätzel, "Dye-sensitized solar cells: A brief overview," Solar Energy, vol. 85, pp. 1172-1178, 2011.

[12] J. Nelson, "Polymer:Fullerene bulk heterojunction solar cells," Materials Today, vol. 14, pp. 462-470, 2011.

[13] D. B. Mitzi, "Solution-processed inorganic semiconductors," Journal of Materials Chemistry, vol. 14, pp. 2355-2365, 2004.

[14] E. H. Sargent, "Colloidal quantum dot solar cells," Nat Photon, vol. 6, pp. 133-135, 2012.

[15] Z. Song, S. C. Watthage, A. B. Phillips, and M. J. Heben, "Pathways toward high- performance perovskite solar cells: Review of recent advances in organo-metal halide perovskites for photovoltaic applications," Journal of Photonics for Energy, vol. 6, p. 022001, 2016.

[16] I. Repins, M. A. Contreras, B. Egaas, C. DeHart, J. Scharf, C. L. Perkins, B. To, and R. Noufi, "19·9%-efficient ZnO/CdS/CuInGaSe2 solar cell with 81·2% fill factor," Progress in Photovoltaics: Research and Applications, vol. 16, pp. 235- 239, 2008.

[17] T. K. Todorov, O. Gunawan, T. Gokmen, and D. B. Mitzi, "Solution-processed Cu(In,Ga)(S,Se)2 absorber yielding a 15.2% efficient solar cell," Progress in Photovoltaics: Research and Applications, vol. 21, pp. 82-87, 2013.

[18] D. B. Mitzi, "Solution processing of chalcogenide semiconductors via dimensional reduction," Advanced Materials, vol. 21, pp. 3141-3158, 2009.

[19] S. B. Zhang, S. H. Wei, and A. Zunger, "Stabilization of ternary compounds via ordered arrays of defect pairs," Physical Review Letters, vol. 78, pp. 4059-4062, 1997.

[20] M. M. Lee, J. Teuscher, T. Miyasaka, T. N. Murakami, and H. J. Snaith, "Efficient hybrid solar cells based on meso-superstructured organometal halide perovskites," Science, vol. 338, pp. 643-647, 2012.

[21] H.-S. Kim, C.-R. Lee, J.-H. Im, K.-B. Lee, T. Moehl, A. Marchioro, S.-J. Moon, R. Humphry-Baker, J.-H. Yum, J. E. Moser, M. Graetzel, and N.-G. Park, "Lead iodide perovskite sensitized all-solid-state submicron thin film mesoscopic solar

194

cell with efficiency exceeding 9%," Scientific Reports, vol. 2, 2012, 10.1038/srep00591.

[22] NREL (20 April 2016), "Solar cell efficiency chart ", Available: http://www.nrel.gov/ncpv/images/efficiency_chart.jpg.

[23] Z. Song, A. B. Phillips, Y. Xie, R. R. Khanal, and M. J. Heben, "The performance of nanocrystalline CuInS2/In2S3/SnO2 heterojunction solar cells prepared by chemical spray pyrolysis," in 2013 IEEE 39th Photovoltaic Specialists Conference (PVSC), 2013, pp. 2540-2544.

[24] Z. Song, A. B. Phillips, P. Krantz, T. Prabhakar, R. R. Khanal, Y. Xie, J. L. DeWitt, J. M. Stone, and M. J. Heben, "Spray pyrolysis of CuIn(S,Se)2 thin films using hydrazine-based solutions," MRS Proceedings vol. 1630, 2014, doi:10.1557/opl.2014.81.

[25] Z. Song, A. B. Phillips, P. W. Krantz, R. R. Khanal, and M. J. Heben, "Spray pyrolysis of semi-transparent backwall superstrate CuIn(S,Se)2 solar cells," in 2014 IEEE 40th Photovoltaic Specialist Conference (PVSC), 2014, pp. 1712-1717.

[26] W. N. Shafarman, S. Siebentritt, and L. Stolt, "Cu(InGa)Se2 solar cells," in Handbook of photovoltaic science and engineering, ed: John Wiley & Sons, Ltd, 2011, pp. 546-599.

[27] R. R. Khanal, A. B. Phillips, Z. Song, Y. Xie, H. P. Mahabaduge, M. D. Dorogi, S. Zafar, G. T. Faykosh, and M. J. Heben, "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.

[28] F. A. Lindholm, J. G. Fossum, and E. L. Burgess, "Application of the superposition principle to solar-cell analysis," IEEE Transactions on Electron Devices, vol. 26, pp. 165-171, 1979.

[29] R. A. Sinton and A. Cuevas, "Contactless determination of current–voltage characteristics and minority‐carrier lifetimes in semiconductors from quasi‐steady‐ state photoconductance data," Applied Physics Letters, vol. 69, pp. 2510-2512, 1996.

[30] M. A. Green, "Solar cell fill factors: General graph and empirical expressions," Solid-State Electronics, vol. 24, pp. 788-789, 1981.

[31] Z. Song, S. C. Watthage, B. L. Tompkins, G. K. Liyanage, A. B. Phillips, R. J. Ellingson, and M. J. Heben, "Spatially resolved characterization of solution processed perovskite solar cells using the LBIC technique," in Photovoltaic Specialist Conference (PVSC), IEEE 42nd, 2015. 195

[32] Z. Song, S. C. Watthage, A. B. Phillips, G. K. Liyanage, R. R. Khanal, B. L. Tompkins, R. J. Ellingson, and M. J. Heben, "Investigation of degradation mechanisms of perovskite-based photovoltaic devices using laser beam induced current mapping," Proc. SPIE, vol. 9561, p. 956107, 2015.

[33] Z. Song, A. Abate, S. C. Watthage, G. K. Liyanage, A. B. Phillips, U. Steiner, M. Graetzel, and M. J. Heben, "Perovskite solar cell stability in humid air: Partially reversible phase transitions in the PbI2-CH3NH3I-H2O system," Advanced Energy Materials, vol. 10.1002/aenm.201600846, 2016.

[34] A. Phillips, Z. Song, J. DeWitt, J. Stone, P. Krantz, J. Royston, R. Zeller, M. Mapes, P. Roland, M. Dorogi, S. Zafar, G. Faykosh, R. Ellingston, and M. Heben, "High speed, large area laser beam induced current imaging and laser scribing system for photovoltaic devices and modules," Review of Scientific Instruments, 2016.

[35] W. Shockley and H. J. Queisser, "Detailed balance limit of efficiency of p‐n junction solar cells," Journal of Applied Physics, vol. 32, pp. 510-519, 1961.

[36] Y. Ogawa, S. Uenishi, K. Tohyama, and K. Ito, "Preparation and properties of CuInS2 thin films," Solar Energy Materials and Solar Cells, vol. 35, pp. 157-163, 1994.

[37] H. Chen, C.-Y. Wang, J.-T. Wang, X.-P. Hu, and S.-X. Zhou, "First-principles study of point defects in solar cell semiconductor CuInS2," Journal of Applied Physics, vol. 112, p. 084513, 2012.

[38] R. Klenk, J. Klaer, R. Scheer, M. C. Lux-Steiner, I. Luck, N. Meyer, and U. Rühle, "Solar cells based on CuInS2—an overview," Thin Solid Films, vol. 480–481, pp. 509-514, 2005.

[39] B. Tell, J. L. Shay, and H. M. Kasper, "Electrical properties, optical properties, and band structure of CuGaS2 and CuInS2," Physical Review B, vol. 4, pp. 2463-2471, 1971.

[40] L. L. Kazmerski, M. S. Ayyagari, and G. A. Sanborn, "CuInS2 thin films: Preparation and properties," Journal of Applied Physics, vol. 46, pp. 4865-4869, 1975.

[41] H. J. Lewerenz, H. Goslowsky, K. D. Husemann, and S. Fiechter, "Efficient solar energy conversion with CuInS2," Nature, vol. 321, pp. 687-688, 1986.

[42] R. Scheer, T. Walter, H. W. Schock, M. L. Fearheiley, and H. J. Lewerenz, "CuInS2 based thin film solar cell with 10.2% efficiency," Applied Physics Letters, vol. 63, pp. 3294-3296, 1993.

196

[43] D. Braunger, D. Hariskos, T. Walter, and H. W. Schock, "An 11.4% efficient polycrystalline thin film solar cell based on CuInS2 with a Cd-free buffer layer," Solar Energy Materials and Solar Cells, vol. 40, pp. 97-102, 1996.

[44] J. Klaer, J. Bruns, R. Henninger, K. Siemer, R. Klenk, K. Ellmer, and D. Bräunig, "Efficient CuInS2 thin-film solar cells prepared by a sequential process," Semiconductor Science and Technology, vol. 13, p. 1456, 1998.

[45] R. Scheer, K. Diesner, and H. J. Lewerenz, "Experiments on the microstructure of evaporated CuInS2 thin films," Thin Solid Films, vol. 268, pp. 130-136, 1995.

[46] Y. L. Wu, H. Y. Lin, C. Y. Sun, M. H. Yang, and H. L. Hwang, "On the growth of CuInS2 thin films by three-source evaporation," Thin Solid Films, vol. 168, pp. 113- 122, 1989.

[47] A. N. Y. Samaan, S. M. Wasim, A. E. Hill, D. G. Akmour, and R. D. Tomlinson, "Preparation and properties of CuInS2 thin films deposited by rf sputtering," physica status solidi (a), vol. 96, pp. 317-324, 1986.

[48] T. Unold, I. Sieber, and K. Ellmer, "Efficient CuInS2 solar cells by reactive magnetron sputtering," Applied Physics Letters, vol. 88, p. 213502, 2006.

[49] S. Nakamura and A. Yamamoto, "Preparation of CuInS2 films with sufficient sulfur content and excellent morphology by one-step electrodeposition," Solar Energy Materials and Solar Cells, vol. 49, pp. 415-421, 1997.

[50] G. K. Padam and S. U. M. Rao, "Preparation and characterization of chemically deposited CuInS2 thin films," Solar Energy Materials, vol. 13, pp. 297-305, 1986.

[51] A. N. Tiwari, D. K. Pandya, and K. L. Chopra, "Electrical and optical properties of single-phase CuInS2 films prepared using spray pyrolysis," Thin Solid Films, vol. 130, pp. 217-230, 1985.

[52] D.-Y. Lee and J. Kim, "Deposition of CuInS2 films by electrostatic field assisted ultrasonic spray pyrolysis," Solar Energy Materials and Solar Cells, vol. 95, pp. 245-249, 2011.

[53] K. Siemer, J. Klaer, I. Luck, J. Bruns, R. Klenk, and D. Bräunig, "Efficient CuInS2 solar cells from a rapid thermal process (rtp)," Solar Energy Materials and Solar Cells, vol. 67, pp. 159-166, 2001.

[54] R. Klenk, U. Blieske, V. Dieterle, K. Ellmer, S. Fiechter, I. Hengel, A. Jäger- Waldau, T. Kampschulte, C. Kaufmann, J. Klaer, M. C. Lux-Steiner, D. Braunger, D. Hariskos, M. Ruckh, and H. W. Schock, "Properties of CuInS2 thin films grown

197

by a two-step process without H2S," Solar Energy Materials and Solar Cells, vol. 49, pp. 349-356, 1997.

[55] T. Nakabayashi, T. Miyazawa, Y. Hashimoto, and K. Ito, "Over 10% efficient CuInS2 solar cell by sulfurization," Solar Energy Materials and Solar Cells, vol. 49, pp. 375-381, 1997.

[56] S.-H. Wei, S. B. Zhang, and A. Zunger, "Band structure and stability of zincblende-based semiconductor polytypes," Physical Review B, vol. 59, pp. R2478- R2481, 1999.

[57] J. J. M. Binsma, L. J. Giling, and J. Bloem, "Order—disorder behaviour and tetragonal distortion of chalcopyrite compounds," physica status solidi (a), vol. 63, pp. 595-603, 1981.

[58] Y. Qi, Q. Liu, K. Tang, Z. Liang, Z. Ren, and X. Liu, "Synthesis and characterization of nanostructured wurtzite CuInS2: A new cation disordered polymorph of CuInS2," The Journal of Physical Chemistry C, vol. 113, pp. 3939- 3944, 2009.

[59] H. Metzner, M. Brüssler, K. D. Husemann, and H. J. Lewerenz, "Characterization of phases and determination of phase relations in the Cu-In-S system by g-g perturbed angular correlations," Physical Review B, vol. 44, pp. 11614-11623, 1991.

[60] R. Klenk, T. Walter, H.-W. Schock, and D. Cahen, "A model for the successful growth of polycrystalline films of CuInSe2 by multisource physical vacuum evaporation," Advanced Materials, vol. 5, pp. 114-119, 1993.

[61] M. Weber, R. Scheer, H. J. Lewerenz, H. Jungblut, and U. Störkel "Microroughness and composition of cyanide-treated CuInS2," Journal of The Electrochemical Society, vol. 149, pp. G77-G84, 2002.

[62] W. Takayuki and M. Masahiro, "Improved efficiency of CuInS2 -based solar cells without potassium cyanide process," Japanese Journal of Applied Physics, vol. 38, p. L1379, 1999.

[63] R. D. Deegan, O. Bakajin, T. F. Dupont, G. Huber, S. R. Nagel, and T. A. Witten, "Capillary flow as the cause of ring stains from dried liquid drops," Nature, vol. 389, pp. 827-829, 1997.

[64] Q. Xu, B. Huang, Y. Zhao, Y. Yan, R. Noufi, and S.-H. Wei, "Crystal and electronic structures of CuXS solar cell absorbers," Applied Physics Letters, vol. 100, p. 061906, 2012.

198

[65] M. Nanu, J. Schoonman, and A. Goossens, "Solar-energy conversion in TiO2/ CuInS2 nanocomposites," Advanced Functional Materials, vol. 15, pp. 95-100, 2005.

[66] J. Verschraegen, M. Burgelman, and J. Penndorf, "Temperature dependence of the diode ideality factor in CuInS2-on-Cu-tape solar cells," Thin Solid Films, vol. 480– 481, pp. 307-311, 2005.

[67] P. Jackson, D. Hariskos, E. Lotter, S. Paetel, R. Wuerz, R. Menner, W. Wischmann, and M. Powalla, "New world record efficiency for Cu(In,Ga)Se2 thin-film solar cells beyond 20%," Progress in Photovoltaics: Research and Applications, vol. 19, pp. 894-897, 2011.

[68] K. Ramanathan, M. A. Contreras, C. L. Perkins, S. Asher, F. S. Hasoon, J. Keane, D. Young, M. Romero, W. Metzger, R. Noufi, J. Ward, and A. Duda, "Properties of 19.2% efficiency Zno/CdS/CuInGaSe2 thin-film solar cells," Progress in Photovoltaics: Research and Applications, vol. 11, pp. 225-230, 2003.

[69] M. Kaelin, D. Rudmann, and A. N. Tiwari, "Low cost processing of CIGS thin film solar cells," Solar Energy, vol. 77, pp. 749-756, 2004.

[70] P. Reinhard, S. Buecheler, and A. N. Tiwari, "Technological status of Cu(In,Ga)(Se,S)2-based photovoltaics," Solar Energy Materials and Solar Cells, vol. 119, pp. 287-290, 2013.

[71] E. Lee, S. J. Park, J. W. Cho, J. Gwak, M.-K. Oh, and B. K. Min, "Nearly carbon- free printable CIGS thin films for solar cell applications," Solar Energy Materials and Solar Cells, vol. 95, pp. 2928-2932, 2011.

[72] W. Wang, Y.-W. Su, and C.-h. Chang, "Inkjet printed chalcopyrite CuInxGa1−xSe2 thin film solar cells," Solar Energy Materials and Solar Cells, vol. 95, pp. 2616- 2620, 2011.

[73] S. Ahn, C. Kim, J. H. Yun, J. Gwak, S. Jeong, B.-H. Ryu, and K. Yoon, " CuInSe2 (CIS) thin film solar cells by direct coating and selenization of solution precursors," The Journal of Physical Chemistry C, vol. 114, pp. 8108-8113, 2010.

[74] G. Wang, S. Wang, Y. Cui, and D. Pan, "A novel and versatile strategy to prepare metal–organic molecular precursor solutions and its application in Cu(In,Ga)(S,Se)2 solar cells," Chemistry of Materials, vol. 24, pp. 3993-3997, 2012.

[75] S. L. Castro, S. G. Bailey, R. P. Raffaelle, K. K. Banger, and A. F. Hepp, "Nanocrystalline chalcopyrite materials (CuInS2 and CuInSe2) via low-temperature

199

pyrolysis of molecular single-source precursors," Chemistry of Materials, vol. 15, pp. 3142-3147, 2003.

[76] Q. Guo, S. J. Kim, M. Kar, W. N. Shafarman, R. W. Birkmire, E. A. Stach, R. Agrawal, and H. W. Hillhouse, "Development of CuInSe2 nanocrystal and nanoring inks for low-cost solar cells," Nano Letters, vol. 8, pp. 2982-2987, 2008.

[77] A. R. Uhl, C. Fella, A. Chirilă, M. R. Kaelin, L. Karvonen, A. Weidenkaff, C. N. Borca, D. Grolimund, Y. E. Romanyuk, and A. N. Tiwari, "Non-vacuum deposition of Cu(In,Ga)Se2 absorber layers from binder free, alcohol solutions," Progress in Photovoltaics: Research and Applications, vol. 20, pp. 526-533, 2012.

[78] M. Kar, R. Agrawal, and H. W. Hillhouse, "Formation pathway of CuInSe2 nanocrystals for solar cells," Journal of the American Chemical Society, vol. 133, pp. 17239-17247, 2011.

[79] M. G. Panthani, V. Akhavan, B. Goodfellow, J. P. Schmidtke, L. Dunn, A. Dodabalapur, P. F. Barbara, and B. A. Korgel, "Synthesis of CuInS2, CuInSe2, and Cu(InxGa1-x)Se2 (CIGS) nanocrystal “inks” for printable photovoltaics," Journal of the American Chemical Society, vol. 130, pp. 16770-16777, 2008.

[80] G. Brown, P. Stone, J. Woodruff, B. Cardozo, and D. Jackrel, "Device characteristics of a 17.1% efficient solar cell deposited by a non-vacuum printing method on flexible foil," in Photovoltaic Specialists Conference (PVSC), 2012 38th IEEE, 2012, pp. 003230-003233.

[81] D. B. Mitzi, M. Yuan, W. Liu, A. J. Kellock, S. J. Chey, V. Deline, and A. G. Schrott, "A high-efficiency solution-deposited thin-film photovoltaic device," Advanced Materials, vol. 20, pp. 3657-3662, 2008.

[82] W. Liu, D. B. Mitzi, M. Yuan, A. J. Kellock, S. J. Chey, and O. Gunawan, "12% efficiency CuIn(Se,S)2 photovoltaic device prepared using a hydrazine solution process," Chemistry of Materials, vol. 22, pp. 1010-1014, 2010.

[83] T. K. Todorov, K. B. Reuter, and D. B. Mitzi, "High-efficiency solar cell with earth- abundant liquid-processed absorber," Advanced Materials, vol. 22, pp. E156-E159, 2010.

[84] C.-H. Chung, S.-H. Li, B. Lei, W. Yang, W. W. Hou, B. Bob, and Y. Yang, "Identification of the molecular precursors for hydrazine solution processed CuIn(Se,S)2 films and their interactions," Chemistry of Materials, vol. 23, pp. 964- 969, 2011.

[85] Q. Cao, O. Gunawan, M. Copel, K. B. Reuter, S. J. Chey, V. R. Deline, and D. B. Mitzi, "Defects in Cu(In,Ga)Se2 chalcopyrite semiconductors: A comparative study 200

of material properties, defect states, and photovoltaic performance," Advanced Energy Materials, vol. 1, pp. 845-853, 2011.

[86] S. Ahn, T. H. Son, A. Cho, J. Gwak, J. H. Yun, K. Shin, S. K. Ahn, S. H. Park, and K. Yoon, "CuInSe2 thin-film solar cells with 7.72 % efficiency prepared via direct coating of a metal salts/alcohol-based precursor solution," ChemSusChem, vol. 5, pp. 1773-1777, 2012.

[87] J. K. Larsen, H. Simchi, P. Xin, K. Kim, and W. N. Shafarman, "Backwall superstrate configuration for ultrathin Cu(In,Ga)Se2 solar cells," Applied Physics Letters, vol. 104, p. 033901, 2014.

[88] M. Burgelman, P. Nollet, and S. Degrave, "Modelling polycrystalline semiconductor solar cells," Thin Solid Films, vol. 361–362, pp. 527-532, 2000.

[89] S. S. Hegedus and W. N. Shafarman, "Thin-film solar cells: Device measurements and analysis," Progress in Photovoltaics: Research and Applications, vol. 12, pp. 155-176, 2004.

[90] T. Nakada, Y. Hirabayashi, T. Tokado, D. Ohmori, and T. Mise, "Novel device structure for Cu(In,Ga)Se2 thin film solar cells using transparent conducting oxide back and front contacts," Solar Energy, vol. 77, pp. 739-747, 2004.

[91] Z. Song, A. B. Phillips, Y. Xie, R. R. Khanal, J. M. Stone, and M. J. Heben, "The effect of substrate wettability on hydrazine-processed czts thin films," MRS Proceedings vol. 1648, pp. mrsf13-1648-hh03-03, 2014.

[92] Q. Chen, N. De Marco, Y. Yang, T.-B. Song, C.-C. Chen, H. Zhao, Z. Hong, H. Zhou, and Y. Yang, "Under the spotlight: The organic–inorganic hybrid halide perovskite for optoelectronic applications," Nano Today, vol. 10, pp. 355-396, 2015.

[93] P. Gao, M. Gratzel, and M. K. Nazeeruddin, "Organohalide lead perovskites for photovoltaic applications," Energy & Environmental Science, vol. 7, pp. 2448- 2463, 2014.

[94] M. A. Green, A. Ho-Baillie, and H. J. Snaith, "The emergence of perovskite solar cells," Nature Photonics, vol. 8, pp. 506-514, 2014.

[95] T. C. Sum and N. Mathews, "Advancements in perovskite solar cells: Photophysics behind the photovoltaics," Energy & Environmental Science, vol. 7, pp. 2518-2534, 2014.

[96] S. D. Stranks and H. J. Snaith, "Metal-halide perovskites for photovoltaic and light- emitting devices," Nat Nano, vol. 10, pp. 391-402, 2015. 201

[97] A. K. Chilvery, A. K. Batra, B. Yang, K. Xiao, P. Guggilla, M. D. Aggarwal, R. Surabhi, R. B. Lal, J. R. Currie, and B. G. Penn, "Perovskites: Transforming photovoltaics, a mini-review," Journal of Photonics for Energy, vol. 5, pp. 057402- 057402, 2015.

[98] H. J. Snaith, "Perovskites: The emergence of a new era for low-cost, high-efficiency solar cells," Journal of Physical Chemistry Letters, vol. 4, pp. 3623-3630, 2013.

[99] N.-G. Park, "Organometal perovskite light absorbers toward a 20% efficiency low- cost solid-state mesoscopic solar cell," Journal of Physical Chemistry Letters, vol. 4, pp. 2423-2429, 2013.

[100] R. F. Service, "Perovskite solar cells keep on surging," Science, vol. 344, p. 458, 2014.

[101] M. Gratzel, "The light and shade of perovskite solar cells," Nat Mater, vol. 13, pp. 838-842, 2014.

[102] V. S. Sivaram, Samuel D.; Snaith, Henry J. , "Perovskite solar cells could beat the efficiency of silicon," in Scientific American vol. 313, ed, 2015.

[103] L. J. Schmidt, "Tracking down the truth of perovski," presented at the 38th Rochester Mineralogical Symposium Program Notes, 2011.

[104] D. B. Mitzi, "Synthesis, crystal structure, and optical and thermal properties of (C4H9NH3)2MI4 (M = Ge, Sn, Pb)," Chemistry of Materials, vol. 8, pp. 791-800, 1996.

[105] D. B. Mitzi, M. T. Prikas, and K. Chondroudis, "Thin film deposition of organic−inorganic hybrid materials using a single source thermal ablation technique," Chemistry of Materials, vol. 11, pp. 542-544, 1999.

[106] D. B. Mitzi, C. D. Dimitrakopoulos, and L. L. Kosbar, "Structurally tailored organic−inorganic perovskites: Optical properties and solution-processed channel materials for thin-film transistors," Chemistry of Materials, vol. 13, pp. 3728-3740, 2001.

[107] J. Calabrese, N. L. Jones, R. L. Harlow, N. Herron, D. L. Thorn, and Y. Wang, "Preparation and characterization of layered lead halide compounds," Journal of the American Chemical Society, vol. 113, pp. 2328-2330, 1991.

[108] K. Chondroudis and D. B. Mitzi, "Electroluminescence from an organic-inorganic perovskite incorporating a quaterthiophene dye within lead halide perovskite layers," Chemistry of Materials, vol. 11, pp. 3028-3030, 1999.

202

[109] D. B. Mitzi, K. Chondroudis, and C. R. Kagan, "Organic-inorganic electronics," IBM Journal of Research and Development, vol. 45, pp. 29-45, 2001.

[110] A. Kojima, K. Teshima, Y. Shirai, and T. Miyasaka, "Organometal halide perovskites as visible-light sensitizers for photovoltaic cells," Journal of the American Chemical Society, vol. 131, pp. 6050-6051, 2009.

[111] Editorial, "Perovskite fever," Nat Mater, vol. 13, pp. 837-837, 2014.

[112] M. I. Saidaminov, A. L. Abdelhady, B. Murali, E. Alarousu, V. M. Burlakov, W. Peng, I. Dursun, L. Wang, Y. He, G. Maculan, A. Goriely, T. Wu, O. F. Mohammed, and O. M. Bakr, "High-quality bulk hybrid perovskite single crystals within minutes by inverse temperature crystallization," Nat Commun, vol. 6, 2015, 10.1038/ncomms8586.

[113] J. Burschka, N. Pellet, S.-J. Moon, R. Humphry-Baker, P. Gao, M. K. Nazeeruddin, and M. Graetzel, "Sequential deposition as a route to high-performance perovskite- sensitized solar cells," Nature, vol. 499, pp. 316–319, 2013.

[114] M. Liu, M. B. Johnston, and H. J. Snaith, "Efficient planar heterojunction perovskite solar cells by vapour deposition," Nature, vol. 501, pp. 395–398, 2013.

[115] Q. Chen, H. Zhou, Z. Hong, S. Luo, H.-S. Duan, H.-H. Wang, Y. Liu, G. Li, and Y. Yang, "Planar heterojunction perovskite solar cells via vapor-assisted solution process," Journal of the American Chemical Society, vol. 136, pp. 622-625, 2014.

[116] C.-W. Chen, H.-W. Kang, S.-Y. Hsiao, P.-F. Yang, K.-M. Chiang, and H.-W. Lin, "Efficient and uniform planar-type perovskite solar cells by simple sequential vacuum deposition," Advanced Materials, vol. 26, pp. 6647-6652, 2014.

[117] J. H. Noh, S. H. Im, J. H. Heo, T. N. Mandal, and S. I. Seok, "Chemical management for colorful, efficient, and stable inorganic-organic hybrid nanostructured solar cells," Nano Letters, vol. 13, pp. 1764-1769, 2013.

[118] G. E. Eperon, S. D. Stranks, C. Menelaou, M. B. Johnston, L. M. Herz, and H. J. Snaith, "Formamidinium lead trihalide: A broadly tunable perovskite for efficient planar heterojunction solar cells," Energy & Environmental Science, vol. 7, pp. 982-988, 2014.

[119] W. J. Yin, T. T. Shi, and Y. F. Yan, "Unique properties of halide perovskites as possible origins of the superior solar cell performance," Advanced Materials, vol. 26, pp. 4653-4658, 2014.

[120] S. De Wolf, J. Holovsky, S.-J. Moon, P. Loeper, B. Niesen, M. Ledinsky, F.-J. Haug, J.-H. Yum, and C. Ballif, "Organometallic halide perovskites: Sharp optical 203

absorption edge and its relation to photovoltaic performance," Journal of Physical Chemistry Letters, vol. 5, pp. 1035-1039, 2014.

[121] C. S. Ponseca, Jr., T. J. Savenije, M. Abdellah, K. Zheng, A. Yartsev, T. Pascher, T. Harlang, P. Chabera, T. Pullerits, A. Stepanov, J.-P. Wolf, and V. Sundstrom, "Organometal halide perovskite solar cell materials rationalized: Ultrafast charge generation, high and microsecond-long balanced mobilities, and slow recombination," Journal of the American Chemical Society, vol. 136, pp. 5189- 5192, 2014.

[122] V. D'Innocenzo, G. Grancini, M. J. P. Alcocer, A. R. S. Kandada, S. D. Stranks, M. M. Lee, G. Lanzani, H. J. Snaith, and A. Petrozza, "Excitons versus free charges in organo-lead tri-halide perovskites," Nature Communications, vol. 5, 2014, 10.1038/ncomms4586.

[123] S. D. Stranks, V. M. Burlakov, T. Leijtens, J. M. Ball, A. Goriely, and H. J. Snaith, "Recombination kinetics in organic-inorganic perovskites: Excitons, free charge, and subgap states," Physical Review Applied, vol. 2, 2014, 10.1103/PhysRevApplied.2.034007.

[124] S. D. Stranks, G. E. Eperon, G. Grancini, C. Menelaou, M. J. P. Alcocer, T. Leijtens, L. M. Herz, A. Petrozza, and H. J. Snaith, "Electron-hole diffusion lengths exceeding 1 micrometer in an organometal trihalide perovskite absorber," Science, vol. 342, pp. 341-344, 2013.

[125] G. Xing, N. Mathews, S. Sun, S. S. Lim, Y. M. Lam, M. Graetzel, S. Mhaisalkar, and T. C. Sum, "Long-range balanced electron- and hole-transport lengths in organic-inorganic CH3NH3PbI3," Science, vol. 342, pp. 344-347, 2013.

[126] C. Wehrenfennig, G. E. Eperon, M. B. Johnston, H. J. Snaith, and L. M. Herz, "High charge carrier mobilities and lifetimes in organolead trihalide perovskites," Advanced Materials, vol. 26, pp. 1584-1589, 2014.

[127] T. Leijtens, S. D. Stranks, G. E. Eperon, R. Lindblad, E. M. J. Johansson, I. J. McPherson, H. Rensmo, J. M. Ball, M. M. Lee, and H. J. Snaith, "Electronic properties of meso-superstructured and planar organometal halide perovskite films: Charge trapping, photodoping, and carrier mobility," Acs Nano, vol. 8, pp. 7147- 7155, 2014.

[128] W.-J. Yin, T. Shi, and Y. Yan, "Unusual defect physics in CH3NH3PbI3 perovskite solar cell absorber," Applied Physics Letters, vol. 104, p. 063903, 2014.

[129] W. Tress, N. Marinova, O. Inganas, M. K. Nazeeruddin, S. M. Zakeeruddin, and M. Graetzel, "Predicting the open-circuit voltage of CH3NH3PbI3 perovskite solar cells using electroluminescence and photovoltaic quantum efficiency spectra: The 204

role of radiative and non-radiative recombination," Advanced Energy Materials, vol. 5, 2015, 10.1002/aenm.201400812.

[130] J.-H. Im, C.-R. Lee, J.-W. Lee, S.-W. Park, and N.-G. Park, "6.5% efficient perovskite quantum-dot-sensitized solar cell," Nanoscale, vol. 3, pp. 4088-4093, 2011.

[131] B. O'Regan and M. Gratzel, "A low-cost, high-efficiency solar cell based on dye- sensitized colloidal TiO2 films," Nature, vol. 353, pp. 737-740, 1991.

[132] W. Chen, Y. Z. Wu, J. Liu, C. J. Qin, X. D. Yang, A. Islam, Y. B. Cheng, and L. Y. Han, "Hybrid interfacial layer leads to solid performance improvement of inverted perovskite solar cells," Energy & Environmental Science, vol. 8, pp. 629- 640, 2015.

[133] K.-C. Wang, J.-Y. Jeng, P.-S. Shen, Y.-C. Chang, E. W.-G. Diau, C.-H. Tsai, T.- Y. Chao, H.-C. Hsu, P.-Y. Lin, P. Chen, T.-F. Guo, and T.-C. Wen, "P-type mesoscopic nickel oxide/organometallic perovskite heterojunction solar cells," Scientific Reports, vol. 4, p. 4756, 2014, 10.1038/srep04756.

[134] J. H. Heo, S. H. Im, J. H. Noh, T. N. Mandal, C.-S. Lim, J. A. Chang, Y. H. Lee, H.-j. Kim, A. Sarkar, M. K. Nazeeruddin, M. Graetzel, and S. I. Seok, "Efficient inorganic-organic hybrid heterojunction solar cells containing perovskite compound and polymeric hole conductors," Nature Photonics, vol. 7, pp. 487-492, 2013.

[135] J. J. Choi, X. Yang, Z. M. Norman, S. J. L. Billinge, and J. S. Owen, "Structure of methylammonium lead iodide within mesoporous titanium dioxide: Active material in high-performance perovskite solar cells," Nano Letters, vol. 14, pp. 127-133, 2014.

[136] T. Leijtens, G. E. Eperon, S. Pathak, A. Abate, M. M. Lee, and H. J. Snaith, "Overcoming ultraviolet light instability of sensitized TiO2 with meso- superstructured organometal tri-halide perovskite solar cells," Nature Communications, vol. 4, 2013, 10.1038/ncomms3885.

[137] T. Leijtens, B. Lauber, G. E. Eperon, S. D. Stranks, and H. J. Snaith, "The importance of perovskite pore filling in organometal mixed halide sensitized TiO2- based solar cells," Journal of Physical Chemistry Letters, vol. 5, pp. 1096-1102, 2014.

[138] G. E. Eperon, V. M. Burlakov, P. Docampo, A. Goriely, and H. J. Snaith, "Morphological control for high performance, solution-processed planar heterojunction perovskite solar cells," Advanced Functional Materials, vol. 24, pp. 151-157, 2014. 205

[139] W. S. Yang, J. H. Noh, N. J. Jeon, Y. C. Kim, S. Ryu, J. Seo, and S. I. Seok, "High- performance photovoltaic perovskite layers fabricated through intramolecular exchange," Science, vol. 348, pp. 1234-1237, 2015.

[140] E. Edri, S. Kirmayer, A. Henning, S. Mukhopadhyay, K. Gartsman, Y. Rosenwaks, G. Hodes, and D. Cahen, "Why lead methylammonium tri-iodide perovskite-based solar cells require a mesoporous electron transporting scaffold (but not necessarily a hole conductor)," Nano Letters, vol. 14, pp. 1000-1004, 2014.

[141] H. Zhou, Q. Chen, G. Li, S. Luo, T.-b. Song, H.-S. Duan, Z. Hong, J. You, Y. Liu, and Y. Yang, "Interface engineering of highly efficient perovskite solar cells," Science, vol. 345, pp. 542-546, 2014.

[142] H.-S. Kim and N.-G. Park, "Parameters affecting i–v hysteresis of CH3NH3PbI3 perovskite solar cells: Effects of perovskite crystal size and mesoporous TiO2 layer," The Journal of Physical Chemistry Letters, vol. 5, pp. 2927-2934, 2014.

[143] J.-Y. Jeng, Y.-F. Chiang, M.-H. Lee, S.-R. Peng, T.-F. Guo, P. Chen, and T.-C. Wen, " CH3NH3PbI3 perovskite/fullerene planar-heterojunction hybrid solar cells," Advanced Materials, vol. 25, pp. 3727-3732, 2013.

[144] O. Malinkiewicz, A. Yella, Y. H. Lee, G. Minguez Espallargas, M. Graetzel, M. K. Nazeeruddin, and H. J. Bolink, "Perovskite solar cells employing organic charge- transport layers," Nature Photonics, vol. 8, pp. 128-132, 2014.

[145] Q. Dong, Y. Yuan, Y. Shao, Y. Fang, Q. Wang, and J. Huang, "Abnormal crystal growth in CH3NH3PbI3-xClx using a multi-cycle solution coating process," Energy & Environmental Science, vol. 8, pp. 2464-2470, 2015.

[146] J. You, L. Meng, T.-B. Song, T.-F. Guo, Y. Yang, W.-H. Chang, Z. Hong, H. Chen, H. Zhou, Q. Chen, Y. Liu, N. De Marco, and Y. Yang, "Improved air stability of perovskite solar cells via solution-processed metal oxide transport layers," Nat Nano, vol. 10.1038/nnano.2015.230

[147] W. Chen, Y. Wu, Y. Yue, J. Liu, W. Zhang, X. Yang, H. Chen, E. Bi, I. Ashraful, M. Grätzel, and L. Han, "Efficient and stable large-area perovskite solar cells with inorganic charge extraction layers," Science, vol. 350, pp. 944-948, 2015.

[148] J. H. Park, J. Seo, S. Park, S. S. Shin, Y. C. Kim, N. J. Jeon, H.-W. Shin, T. K. Ahn, J. H. Noh, S. C. Yoon, C. S. Hwang, and S. I. Seok, "Efficient CH3NH3PbI3 perovskite solar cells employing nanostructured p-type nio electrode formed by a pulsed laser deposition," Advanced Materials, vol. 27, pp. 4013-4019, 2015.

[149] Z. Song, S. C. Watthage, A. B. Phillips, B. L. Tompkins, R. J. Ellingson, and M. J. Heben, "Impact of processing temperature and composition on the formation of 206

methylammonium lead iodide perovskites," Chemistry of Materials, vol. 27, pp. 4612-4619, 2015.

[150] C. Roldan-Carmona, P. Gratia, I. Zimmermann, G. Grancini, P. Gao, M. Graetzel, and M. K. Nazeeruddin, "High efficiency methylammonium lead triiodide perovskite solar cells: The relevance of non-stoichiometric precursors," Energy & Environmental Science, vol. 8, pp. 3550-3556, 2015.

[151] Q. Wang, Y. Shao, Q. Dong, Z. Xiao, Y. Yuan, and J. Huang, "Large fill-factor bilayer iodine perovskite solar cells fabricated by a low-temperature solution- process," Energy & Environmental Science, vol. 7, pp. 2359-2365, 2014.

[152] J.-H. Im, I.-H. Jang, N. Pellet, M. Grätzel, and N.-G. Park, "Growth of CH3NH3PbI3 cuboids with controlled size for high-efficiency perovskite solar cells," Nat Nano, vol. 9, pp. 927-932, 2014.

[153] N. Ahn, D.-Y. Son, I.-H. Jang, S. M. Kang, M. Choi, and N.-G. Park, "Highly reproducible perovskite solar cells with average efficiency of 18.3% and best efficiency of 19.7% fabricated via lewis base adduct of lead(ii) iodide," Journal of the American Chemical Society, vol. 137, pp. 8696-8699, 2015.

[154] A. T. Barrows, A. J. Pearson, C. K. Kwak, A. D. F. Dunbar, A. R. Buckley, and D. G. Lidzey, "Efficient planar heterojunction mixed-halide perovskite solar cells deposited via spray-deposition," Energy & Environmental Science, vol. 7, pp. 2944-2950, 2014.

[155] Y. H. Deng, E. Peng, Y. C. Shao, Z. G. Xiao, Q. F. Dong, and J. S. Huang, "Scalable fabrication of efficient organolead trihalide perovskite solar cells with doctor- bladed active layers," Energy & Environmental Science, vol. 8, pp. 1544-1550, 2015.

[156] S.-G. Li, K.-J. Jiang, M.-J. Su, X.-P. Cui, J.-H. Huang, Q.-Q. Zhang, X.-Q. Zhou, L.-M. Yang, and Y.-L. Song, "Inkjet printing of CH3NH3PbI3 on a mesoscopic TiO2 film for highly efficient perovskite solar cells," Journal of Materials Chemistry A, vol. 3, pp. 9092-9097, 2015.

[157] K. Hwang, Y.-S. Jung, Y.-J. Heo, F. H. Scholes, S. E. Watkins, J. Subbiah, D. J. Jones, D.-Y. Kim, and D. Vak, "Toward large scale roll-to-roll production of fully printed perovskite solar cells," Advanced Materials, vol. 27, pp. 1241-1247, 2015.

[158] K. N. Liang, D. B. Mitzi, and M. T. Prikas, "Synthesis and characterization of organic-inorganic perovskite thin films prepared using a versatile two-step dipping technique," Chemistry of Materials, vol. 10, pp. 403-411, 1998.

207

[159] T. Zhang, M. Yang, Y. Zhao, and K. Zhu, "Controllable sequential deposition of planar CH3NH3PbI3 perovskite films via adjustable volume expansion," Nano Letters, vol. 15, pp. 3959-3963, 2015.

[160] N. J. Jeon, J. H. Noh, Y. C. Kim, W. S. Yang, S. Ryu, and S. Il Seol, "Solvent engineering for high-performance inorganic-organic hybrid perovskite solar cells," Nature Materials, vol. 13, pp. 897-903, 2014.

[161] V. Nicolosi, M. Chhowalla, M. G. Kanatzidis, M. S. Strano, and J. N. Coleman, "Liquid exfoliation of layered materials," Science, vol. 340, 2013, 10.1126/science.1226419.

[162] F. Hao, C. C. Stoumpos, Z. Liu, R. P. H. Chang, and M. G. Kanatzidis, "Controllable perovskite crystallization at a gas–solid interface for hole conductor- free solar cells with steady power conversion efficiency over 10%," Journal of the American Chemical Society, vol. 136, pp. 16411-16419, 2014.

[163] M. M. Tavakoli, L. Gu, Y. Gao, C. Reckmeier, J. He, A. L. Rogach, Y. Yao, and Z. Fan, "Fabrication of efficient planar perovskite solar cells using a one-step chemical vapor deposition method," Scientific Reports, vol. 5, p. 14083, 2015, 10.1038/srep14083.

[164] Q. Lin, A. Armin, R. C. R. Nagiri, P. L. Burn, and P. Meredith, "Electro-optics of perovskite solar cells," Nat Photon, vol. 9, pp. 106-112, 2015.

[165] D. Zhao, W. Ke, C. R. Grice, A. J. Cimaroli, X. Tan, M. Yang, R. W. Collins, H. Zhang, K. Zhu, and Y. Yan, "Annealing-free efficient vacuum-deposited planar perovskite solar cells with evaporated fullerenes as electron-selective layers," Nano Energy, vol. 19, pp. 88-97, 2016.

[166] T. Salim, S. Sun, Y. Abe, A. Krishna, A. C. Grimsdale, and Y. M. Lam, "Perovskite-based solar cells: Impact of morphology and device architecture on device performance," Journal of Materials Chemistry A, vol. 3, pp. 8943-8969, 2015.

[167] Y. Zhao and K. Zhu, "Solution chemistry engineering toward high-efficiency perovskite solar cells," The Journal of Physical Chemistry Letters, vol. 5, pp. 4175- 4186, 2014.

[168] K. Yan, M. Long, T. Zhang, Z. Wei, H. Chen, S. Yang, and J. Xu, "Hybrid halide perovskite solar cell precursors: Colloidal chemistry and coordination engineering behind device processing for high efficiency," Journal of the American Chemical Society, vol. 137, pp. 4460-4468, 2015.

208

[169] W. Li, J. Fan, J. Li, Y. Mai, and L. Wang, "Controllable grain morphology of perovskite absorber film by molecular self-assembly toward efficient solar cell exceeding 17%," Journal of the American Chemical Society, vol. 137, pp. 10399- 10405, 2015.

[170] Y. Zhao and K. Zhu, "Efficient planar perovskite solar cells based on 1.8 ev band gap CH3NH3PbI2Br nanosheets via thermal decomposition," Journal of the American Chemical Society, vol. 136, pp. 12241-12244, 2014.

[171] J. H. Heo, D. H. Song, H. J. Han, S. Y. Kim, J. H. Kim, D. Kim, H. W. Shin, T. K. Ahn, C. Wolf, T.-W. Lee, and S. H. Im, "Planar CH3NH3PbI3 perovskite solar cells with constant 17.2% average power conversion efficiency irrespective of the scan rate," Advanced Materials, vol. 27, pp. 3424-3430, 2015.

[172] H. Tsai, W. Nie, P. Cheruku, N. H. Mack, P. Xu, G. Gupta, A. D. Mohite, and H.- L. Wang, "Optimizing composition and morphology for large-grain perovskite solar cells via chemical control," Chemistry of Materials, vol. 27, pp. 5570-5576, 2015.

[173] Y. Chen, Y. Zhao, and Z. Liang, "Non-thermal annealing fabrication of efficient planar perovskite solar cells with inclusion of nh4cl," Chemistry of Materials, vol. 27, pp. 1448-1451, 2015.

[174] J. H. Heo, D. H. Song, and S. H. Im, "Planar CH3NH3PbBr3 hybrid solar cells with 10.4% power conversion efficiency, fabricated by controlled crystallization in the spin-coating process," Advanced Materials, vol. 26, pp. 8179-8183, 2014.

[175] P. W. Liang, C. Y. Liao, C. C. Chueh, F. Zuo, S. T. Williams, X. K. Xin, J. J. Lin, and A. K. Y. Jen, "Additive enhanced crystallization of solution-processed perovskite for highly efficient planar-heterojunction solar cells," Advanced Materials, vol. 26, pp. 3748-3754, 2014.

[176] A. Mei, X. Li, L. Liu, Z. Ku, T. Liu, Y. Rong, M. Xu, M. Hu, J. Chen, Y. Yang, M. Graetzel, and H. Han, "A hole-conductor-free, fully printable mesoscopic perovskite solar cell with high stability," Science, vol. 345, pp. 295-298, 2014.

[177] X. Li, M. Ibrahim Dar, C. Yi, J. Luo, M. Tschumi, S. M. Zakeeruddin, M. K. Nazeeruddin, H. Han, and M. Grätzel, "Improved performance and stability of perovskite solar cells by crystal crosslinking with alkylphosphonic acid ω- ammonium chlorides," Nat Chem, vol. 7, pp. 703-711, 2015.

[178] Y. Wu, A. Islam, X. Yang, C. Qin, J. Liu, K. Zhang, W. Peng, and L. Han, "Retarding the crystallization of PbI2 for highly reproducible planar-structured perovskite solar cells via sequential deposition," Energy & Environmental Science, vol. 7, pp. 2934-2938, 2014. 209

[179] C.-G. Wu, C.-H. Chiang, Z.-L. Tseng, M. K. Nazeeruddin, A. Hagfeldt, and M. Gratzel, "High efficiency stable inverted perovskite solar cells without current hysteresis," Energy & Environmental Science, vol. 8, pp. 2725-2733, 2015.

[180] W. Nie, H. Tsai, R. Asadpour, J.-C. Blancon, A. J. Neukirch, G. Gupta, J. J. Crochet, M. Chhowalla, S. Tretiak, M. A. Alam, H.-L. Wang, and A. D. Mohite, "High-efficiency solution-processed perovskite solar cells with millimeter-scale grains," Science, vol. 347, pp. 522-525, 2015.

[181] Y. C. Zheng, S. Yang, X. Chen, Y. Chen, Y. Hou, and H. G. Yang, "Thermal- induced volmer–weber growth behavior for planar heterojunction perovskites solar cells," Chemistry of Materials, vol. 27, pp. 5116-5121, 2015.

[182] M. Xiao, F. Huang, W. Huang, Y. Dkhissi, Y. Zhu, J. Etheridge, A. Gray-Weale, U. Bach, Y.-B. Cheng, and L. Spiccia, "A fast deposition-crystallization procedure for highly efficient lead iodide perovskite thin-film solar cells," Angewandte Chemie-International Edition, vol. 53, pp. 9898-9903, 2014.

[183] Z. Xiao, Q. Dong, C. Bi, Y. Shao, Y. Yuan, and J. Huang, "Solvent annealing of perovskite-induced crystal growth for photovoltaic-device efficiency enhancement," Advanced Materials, vol. 26, pp. 6503-6509, 2014.

[184] D. W. de Quilettes, S. M. Vorpahl, S. D. Stranks, H. Nagaoka, G. E. Eperon, M. E. Ziffer, H. J. Snaith, and D. S. Ginger, "Impact of microstructure on local carrier lifetime in perovskite solar cells," Science, vol. 348, pp. 683-686, 2015.

[185] B. S. Tosun and H. W. Hillhouse, "Enhanced carrier lifetimes of pure iodide hybrid perovskite via vapor-equilibrated re-growth (VERG)," The Journal of Physical Chemistry Letters, vol. 6, pp. 2503-2508, 2015.

[186] N. Pellet, P. Gao, G. Gregori, T.-Y. Yang, M. K. Nazeeruddin, J. Maier, and M. Graetzel, "Mixed-organic-cation perovskite photovoltaics for enhanced solar-light harvesting," Angewandte Chemie-International Edition, vol. 53, pp. 3151-3157, 2014.

[187] N. K. Noel, S. D. Stranks, A. Abate, C. Wehrenfennig, S. Guarnera, A.-A. Haghighirad, A. Sadhanala, G. E. Eperon, S. K. Pathak, M. B. Johnston, A. Petrozza, L. M. Herz, and H. J. Snaith, "Lead-free organic-inorganic tin halide perovskites for photovoltaic applications," Energy & Environmental Science, vol. 7, pp. 3061-3068, 2014.

[188] F. Hao, C. C. Stoumpos, C. Duyen Hanh, R. P. H. Chang, and M. G. Kanatzidis, "Lead-free solid-state organic-inorganic halide perovskite solar cells," Nature Photonics, vol. 8, pp. 489-494, 2014.

210

[189] Y. Ogomi, A. Morita, S. Tsukamoto, T. Saitho, N. Fujikawa, Q. Shen, T. Toyoda, K. Yoshino, S. S. Pandey, T. Ma, and S. Hayase, " CH3NH3SnxPb1-xI3 perovskite solar cells covering up to 1060 nm," The Journal of Physical Chemistry Letters, vol. 5, pp. 1004-1011, 2014.

[190] F. Hao, C. C. Stoumpos, P. Guo, N. Zhou, T. J. Marks, R. P. H. Chang, and M. G. Kanatzidis, "Solvent-mediated crystallization of CH3NH3SnI3 films for heterojunction depleted perovskite solar cells," Journal of the American Chemical Society, vol. 137, pp. 11445-11452, 2015.

[191] H. S. Kim, I. Mora-Sero, V. Gonzalez-Pedro, F. Fabregat-Santiago, E. J. Juarez- Perez, N. G. Park, and J. Bisquert, "Mechanism of carrier accumulation in perovskite thin-absorber solar cells," Nature Communications, vol. 4, p. 7, 2013.

[192] D. Y. Son, J. H. Im, H. S. Kim, and N. G. Park, "11% efficient perovskite solar cell based on ZnO nanorods: An effective charge collection system," Journal of Physical Chemistry C, vol. 118, pp. 16567-16573, 2014.

[193] J. M. Ball, M. M. Lee, A. Hey, and H. J. Snaith, "Low-temperature processed meso- superstructured to thin-film perovskite solar cells," Energy & Environmental Science, vol. 6, pp. 1739-1743, 2013.

[194] A. Bera, K. W. Wu, A. Sheikh, E. Alarousu, O. F. Mohammed, and T. Wu, "Perovskite oxide SrTiO3 as an efficient electron transporter for hybrid perovskite solar cells," Journal of Physical Chemistry C, vol. 118, pp. 28494-28501, 2014.

[195] D. Q. Bi, S. J. Moon, L. Haggman, G. Boschloo, L. Yang, E. M. J. Johansson, M. K. Nazeeruddin, M. Gratzel, and A. Hagfeldt, "Using a two-step deposition technique to prepare perovskite (CH3NH3PbI3) for thin film solar cells based on ZrO2 and TiO2 mesostructures," Rsc Advances, vol. 3, pp. 18762-18766, 2013.

[196] D. Liu and T. L. Kelly, "Perovskite solar cells with a planar heterojunction structure prepared using room-temperature solution processing techniques," Nature Photonics, vol. 8, pp. 133-138, 2014.

[197] W. Ke, G. Fang, Q. Liu, L. Xiong, P. Qin, H. Tao, J. Wang, H. Lei, B. Li, J. Wan, G. Yang, and Y. Yan, "Low-temperature solution-processed tin oxide as an alternative electron transporting layer for efficient perovskite solar cells," Journal of the American Chemical Society, vol. 137, pp. 6730-6733, 2015.

[198] J. P. Correa Baena, L. Steier, W. Tress, M. Saliba, S. Neutzner, T. Matsui, F. Giordano, T. J. Jacobsson, A. R. Srimath Kandada, S. M. Zakeeruddin, A. Petrozza, A. Abate, M. K. Nazeeruddin, M. Gratzel, and A. Hagfeldt, "Highly efficient planar perovskite solar cells through band alignment engineering," Energy & Environmental Science, vol. 8, pp. 2928-2934, 2015. 211

[199] L. Wang, W. F. Fu, Z. W. Gu, C. C. Fan, X. Yang, H. Y. Li, and H. Z. Chen, "Low temperature solution processed planar heterojunction perovskite solar cells with a cdse nanocrystal as an electron transport/extraction layer," Journal of Materials Chemistry C, vol. 2, pp. 9087-9090, 2014.

[200] J. Liu, C. Gao, L. Luo, Q. Ye, X. He, L. Ouyang, X. Guo, D. Zhuang, C. Liao, J. Mei, and W. Lau, "Low-temperature, solution processed metal sulfide as an electron transport layer for efficient planar perovskite solar cells," Journal of Materials Chemistry A, vol. 3, pp. 11750-11755, 2015.

[201] J. T.-W. Wang, J. M. Ball, E. M. Barea, A. Abate, J. A. Alexander-Webber, J. Huang, M. Saliba, I. Mora-Sero, J. Bisquert, H. J. Snaith, and R. J. Nicholas, "Low- temperature processed electron collection layers of graphene/TiO2 nanocomposites in thin film perovskite solar cells," Nano Letters, vol. 14, pp. 724-730, 2014.

[202] Z. Yu and L. Sun, "Recent progress on hole-transporting materials for emerging organometal halide perovskite solar cells," Advanced Energy Materials, vol. 5, p. 1500213, 2015, 10.1002/aenm.201500213.

[203] P. Qin, S. Tanaka, S. Ito, N. Tetreault, K. Manabe, H. Nishino, M. K. Nazeeruddin, and M. Grätzel, "Inorganic hole conductor-based lead halide perovskite solar cells with 12.4% conversion efficiency," Nat Commun, vol. 5, 2014, 10.1038/ncomms4834.

[204] S. Ye, W. Sun, Y. Li, W. Yan, H. Peng, Z. Bian, Z. Liu, and C. Huang, "Cuscn- based inverted planar perovskite solar cell with an average pce of 15.6%," Nano Letters, vol. 15, pp. 3723-3728, 2015.

[205] J. A. Christians, R. C. M. Fung, and P. V. Kamat, "An inorganic hole conductor for organo-lead halide perovskite solar cells. Improved hole conductivity with copper iodide," Journal of the American Chemical Society, vol. 136, pp. 758-764, 2014.

[206] X. Xu, Z. Liu, Z. Zuo, M. Zhang, Z. Zhao, Y. Shen, H. Zhou, Q. Chen, Y. Yang, and M. Wang, "Hole selective nio contact for efficient perovskite solar cells with carbon electrode," Nano Letters, vol. 15, pp. 2402-2408, 2015.

[207] J. W. Jung, C.-C. Chueh, and A. K. Y. Jen, "A low-temperature, solution- processable, Cu-doped nickel oxide hole-transporting layer via the combustion method for high-performance thin-film perovskite solar cells," Advanced Materials, vol. 27, pp. 7874-7880, 2015.

[208] L. Etgar, P. Gao, Z. Xue, Q. Peng, A. K. Chandiran, B. Liu, M. K. Nazeeruddin, and M. Graetzel, "Mesoscopic CH3NH3PbI3/TiO2 heterojunction solar cells," Journal of the American Chemical Society, vol. 134, pp. 17396-17399, 2012.

212

[209] W. A. Laban and L. Etgar, "Depleted hole conductor-free lead halide iodide heterojunction solar cells," Energy & Environmental Science, vol. 6, pp. 3249- 3253, 2013.

[210] J. Shi, J. Dong, S. Lv, Y. Xu, L. Zhu, J. Xiao, X. Xu, H. Wu, D. Li, Y. Luo, and Q. Meng, "Hole-conductor-free perovskite organic lead iodide heterojunction thin- film solar cells: High efficiency and junction property," Applied Physics Letters, vol. 104, p. 063901, 2014.

[211] W. Ke, G. Fang, J. Wan, H. Tao, Q. Liu, L. Xiong, P. Qin, J. Wang, H. Lei, G. Yang, M. Qin, X. Zhao, and Y. Yan, "Efficient hole-blocking layer-free planar halide perovskite thin-film solar cells," Nat Commun, vol. 6, 2015, 10.1038/ncomms7700.

[212] D. Liu, J. Yang, and T. L. Kelly, "Compact layer free perovskite solar cells with 13.5% efficiency," Journal of the American Chemical Society, vol. 136, pp. 17116- 17122, 2014.

[213] N. J. Jeon, J. H. Noh, W. S. Yang, Y. C. Kim, S. Ryu, J. Seo, and S. I. Seok, "Compositional engineering of perovskite materials for high-performance solar cells," Nature, vol. 517, pp. 476-480, 2015.

[214] T. Leijtens, G. E. Eperon, N. K. Noel, S. N. Habisreutinger, A. Petrozza, and H. J. Snaith, "Stability of metal halide perovskite solar cells," Advanced Energy Materials, vol. 5, p. DOI: 10.1002/aenm.201500963, 2015.

[215] J. M. Frost, K. T. Butler, F. Brivio, C. H. Hendon, M. van Schilfgaarde, and A. Walsh, "Atomistic origins of high-performance in hybrid halide perovskite solar cells," Nano Letters, vol. 14, pp. 2584-2590, 2014.

[216] S. N. Habisreutinger, T. Leijtens, G. E. Eperon, S. D. Stranks, R. J. Nicholas, and H. J. Snaith, "Carbon nanotube/polymer composites as a highly stable hole collection layer in perovskite solar cells," Nano Letters, vol. 14, pp. 5561-5568, 2014.

[217] X. Li, M. Tschumi, H. Han, S. S. Babkair, R. A. Alzubaydi, A. A. Ansari, S. S. Habib, M. K. Nazeeruddin, S. M. Zakeeruddin, and M. Grätzel, "Outdoor performance and stability under elevated temperatures and long-term light soaking of triple-layer mesoporous perovskite photovoltaics," Energy Technology, vol. 3, pp. 551-555, 2015.

[218] H. Snaith, "From nanostructured to thin film perovskite solar cells," in 42nd IEEE Photovoltaic Specialist Conference, 2015.

213

[219] H. J. Snaith, A. Abate, J. M. Ball, G. E. Eperon, T. Leijtens, N. K. Noel, S. D. Stranks, J. T.-W. Wang, K. Wojciechowski, and W. Zhang, "Anomalous hysteresis in perovskite solar cells," Journal of Physical Chemistry Letters, vol. 5, pp. 1511- 1515, 2014.

[220] W. Tress, N. Marinova, T. Moehl, S. M. Zakeeruddin, M. K. Nazeeruddin, and M. Graetzel, "Understanding the rate-dependent j-v hysteresis, slow time component, and aging in CH3NH3PbI3 perovskite solar cells: The role of a compensated electric field," Energy & Environmental Science, vol. 8, pp. 995-1004, 2015.

[221] E. L. Unger, E. T. Hoke, C. D. Bailie, W. H. Nguyen, A. R. Bowring, T. Heumueller, M. G. Christoforo, and M. D. McGehee, "Hysteresis and transient behavior in current-voltage measurements of hybrid-perovskite absorber solar cells," Energy & Environmental Science, vol. 7, pp. 3690-3698, 2014.

[222] Y. Shao, Z. Xiao, C. Bi, Y. Yuan, and J. Huang, "Origin and elimination of photocurrent hysteresis by fullerene passivation in CH3NH3PbI3 planar heterojunction solar cells," Nature Communications, vol. 5, 2014, 10.1038/ncomms6784.

[223] J. Wei, Y. Zhao, H. Li, G. Li, J. Pan, D. Xu, Q. Zhao, and D. Yu, "Hysteresis analysis based on the ferroelectric effect in hybrid perovskite solar cells," The Journal of Physical Chemistry Letters, vol. 5, pp. 3937-3945, 2014.

[224] J. M. Frost, K. T. Butler, and A. Walsh, "Molecular ferroelectric contributions to anomalous hysteresis in hybrid perovskite solar cells," APL Materials, vol. 2, p. 081506, 2014.

[225] C. Eames, J. M. Frost, P. R. F. Barnes, B. C. O/'Regan, A. Walsh, and M. S. Islam, "Ionic transport in hybrid lead iodide perovskite solar cells," Nat Commun, vol. 6, 2015, 10.1038/ncomms8497.

[226] V. W. Bergmann, S. A. L. Weber, F. Javier Ramos, M. K. Nazeeruddin, M. Grätzel, D. Li, A. L. Domanski, I. Lieberwirth, S. Ahmad, and R. Berger, "Real-space observation of unbalanced charge distribution inside a perovskite-sensitized solar cell," Nat Commun, vol. 5, 2014, 10.1038/ncomms6001.

[227] J. H. Heo, H. J. Han, D. Kim, T. K. Ahn, and S. H. Im, "Hysteresis-less inverted CH3NH3PbI3 planar perovskite hybrid solar cells with 18.1% power conversion efficiency," Energy & Environmental Science, vol. 8, pp. 1602-1608, 2015.

[228] S. van Reenen, M. Kemerink, and H. J. Snaith, "Modeling anomalous hysteresis in perovskite solar cells," The Journal of Physical Chemistry Letters, vol. 6, pp. 3808- 3814, 2015.

214

[229] J. Beilsten-Edmands, G. E. Eperon, R. D. Johnson, H. J. Snaith, and P. G. Radaelli, "Non-ferroelectric nature of the conductance hysteresis in CH3NH3PbI3 perovskite- based photovoltaic devices," Applied Physics Letters, vol. 106, p. 173502, 2015.

[230] B. Chen, J. Shi, X. Zheng, Y. Zhou, K. Zhu, and S. Priya, "Ferroelectric solar cells based on inorganic-organic hybrid perovskites," Journal of Materials Chemistry A, vol. 3, pp. 7699-7705, 2015.

[231] M. Coll, A. Gomez, E. Mas-Marza, O. Almora, G. Garcia-Belmonte, M. Campoy- Quiles, and J. Bisquert, "Polarization switching and light-enhanced piezoelectricity in lead halide perovskites," The Journal of Physical Chemistry Letters, vol. 6, pp. 1408-1413, 2015.

[232] Z. Fan, J. Xiao, K. Sun, L. Chen, Y. Hu, J. Ouyang, K. P. Ong, K. Zeng, and J. Wang, "Ferroelectricity of CH3NH3PbI3 perovskite," The Journal of Physical Chemistry Letters, vol. 6, pp. 1155-1161, 2015.

[233] Z. Xiao, Y. Yuan, Y. Shao, Q. Wang, Q. Dong, C. Bi, P. Sharma, A. Gruverman, and J. Huang, "Giant switchable photovoltaic effect in organometal trihalide perovskite devices," Nature Materials, vol. 14, pp. 193-198, 2015.

[234] D. Bryant, S. Wheeler, B. C. O’Regan, T. Watson, P. R. F. Barnes, D. Worsley, and J. Durrant, "Observable hysteresis at low temperature in “hysteresis free” organic–inorganic lead halide perovskite solar cells," The Journal of Physical Chemistry Letters, vol. 6, pp. 3190-3194, 2015.

[235] J. Gong, S. B. Darling, and F. You, "Perovskite photovoltaics: Life-cycle assessment of energy and environmental impacts," Energy & Environmental Science, vol. 8, pp. 1953-1968, 2015.

[236] N. Espinosa, L. Serrano-Luján, A. Urbina, and F. C. Krebs, "Solution and vapour deposited lead perovskite solar cells: Ecotoxicity from a life cycle assessment perspective," Solar Energy Materials and Solar Cells, vol. 137, pp. 303-310, 2015.

[237] B. Hailegnaw, S. Kirmayer, E. Edri, G. Hodes, and D. Cahen, "Rain on methylammonium lead iodide based perovskites: Possible environmental effects of perovskite solar cells," The Journal of Physical Chemistry Letters, vol. 6, pp. 1543- 1547, 2015.

[238] P.-Y. Chen, J. Qi, M. T. Klug, X. Dang, P. T. Hammond, and A. M. Belcher, "Environmentally responsible fabrication of efficient perovskite solar cells from recycled car batteries," Energy & Environmental Science, vol. 7, pp. 3659-3665, 2014.

215

[239] O. Yamamuro, T. Matsuo, H. Suga, W. I. F. David, R. M. Ibberson, and A. J. Leadbetter, "Neutron diffraction and calorimetric studies of methylammonium iodide," Acta Crystallographica Section B, vol. 48, pp. 329-336, 1992.

[240] C. C. Stoumpos, C. D. Malliakas, and M. G. Kanatzidis, "Semiconducting tin and lead iodide perovskites with organic cations: Phase transitions, high mobilities, and near-infrared photoluminescent properties," Inorganic Chemistry, vol. 52, pp. 9019-9038, 2013.

[241] A. Poglitsch and D. Weber, "Dynamic disorder in methylammoniumtrihalogenoplumbates (ii) observed by millimeter-wave spectroscopy," The Journal of Chemical Physics, vol. 87, pp. 6373-6378, 1987.

[242] T. Baikie, Y. N. Fang, J. M. Kadro, M. Schreyer, F. X. Wei, S. G. Mhaisalkar, M. Graetzel, and T. J. White, "Synthesis and crystal chemistry of the hybrid perovskite (CH3NH3)PbI3 for solid-state sensitised solar cell applications," Journal of Materials Chemistry A, vol. 1, pp. 5628-5641, 2013.

[243] A. Dualeh, N. Tétreault, T. Moehl, P. Gao, M. K. Nazeeruddin, and M. Grätzel, "Effect of annealing temperature on film morphology of organic–inorganic hybrid pervoskite solid-state solar cells," Advanced Functional Materials, vol. 24, pp. 3250-3258, 2014.

[244] H. Ishida, R. Ikeda, and D. Nakamura, "Pre-melting state of methylammonium iodide as revealed by proton magnetic resonance," physica status solidi (a), vol. 70, pp. K151-K154, 1982.

[245] A. E. Dugan and H. K. Henisch, "Fundamental optical absorption and photo- conduction in PbI2 single crystals," Journal of Physics and Chemistry of Solids, vol. 28, pp. 1885-1890, 1967.

[246] Z. Xiao, C. Bi, Y. Shao, Q. Dong, Q. Wang, Y. Yuan, C. Wang, Y. Gao, and J. Huang, "Efficient, high yield perovskite photovoltaic devices grown by interdiffusion of solution-processed precursor stacking layers," Energy & Environmental Science, vol. 7, pp. 2619-2623, 2014.

[247] P. A. Beckmann, "A review of polytypism in lead iodide," Crystal Research and Technology, vol. 45, pp. 455-460, 2010.

[248] S. Colella, E. Mosconi, P. Fedeli, A. Listorti, F. Gazza, F. Orlandi, P. Ferro, T. Besagni, A. Rizzo, G. Calestani, G. Gigli, F. De Angelis, and R. Mosca, "MAPbI(3- x)Clx mixed halide perovskite for hybrid solar cells: The role of chloride as dopant on the transport and structural properties," Chemistry of Materials, vol. 25, pp. 4613-4618, 2013.

216

[249] B. R. Vincent, K. N. Robertson, T. S. Cameron, and O. Knop, "Alkylammonium lead halides. Part 1. Isolated pbi64− ions in (CH3NH3)4PbI6•2H2O," Canadian Journal of Chemistry, vol. 65, pp. 1042-1046, 1987.

[250] X. Hong, T. Ishihara, and A. V. Nurmikko, "Dielectric confinement effect on excitons in PbI4-based layered semiconductors," Physical Review B, vol. 45, pp. 6961-6964, 1992.

[251] A. Dualeh, P. Gao, S. I. Seok, M. K. Nazeeruddin, and M. Graetzel, "Thermal behavior of methylammonium lead-trihalide perovskite photovoltaic light harvesters," Chemistry of Materials, vol. 26, pp. 6160-6164, 2014.

[252] P. Loper, S.-J. Moon, S. Martin de Nicolas, B. Niesen, M. Ledinsky, S. Nicolay, J. Bailat, J.-H. Yum, S. De Wolf, and C. Ballif, "Organic-inorganic halide perovskite/crystalline silicon four-terminal tandem solar cells," Physical Chemistry Chemical Physics, vol. 17, pp. 1619-1629, 2015.

[253] J. Werner, C.-H. Weng, A. Walter, L. Fesquet, J. P. Seif, S. De Wolf, B. Niesen, and C. Ballif, "Efficient monolithic perovskite/silicon tandem solar cell with cell area >1 cm2," The Journal of Physical Chemistry Letters, vol. 7, pp. 161-166, 2016.

[254] T. A. Berhe, W.-N. Su, C.-H. Chen, C.-J. Pan, J.-H. Cheng, H.-M. Chen, M.-C. Tsai, L.-Y. Chen, A. A. Dubale, and B.-J. Hwang, "Organometal halide perovskite solar cells: Degradation and stability," Energy & Environmental Science, vol. 9, pp. 323-356, 2016.

[255] T. Leijtens, G. E. Eperon, N. K. Noel, S. N. Habisreutinger, A. Petrozza, and H. J. Snaith, "Stability of metal halide perovskite solar cells," Advanced Energy Materials, vol. 5, p. 1500963, 2015.

[256] Y. Han, S. Meyer, Y. Dkhissi, K. Weber, J. M. Pringle, U. Bach, L. Spiccia, and Y.-B. Cheng, "Degradation observations of encapsulated planar CH3NH3PbI3 perovskite solar cells at high temperatures and humidity," Journal of Materials Chemistry A, vol. 3, pp. 8139-8147, 2015.

[257] N. Aristidou, I. Sanchez-Molina, T. Chotchuangchutchaval, M. Brown, L. Martinez, T. Rath, and S. A. Haque, "The role of oxygen in the degradation of methylammonium lead trihalide perovskite photoactive layers," Angewandte Chemie International Edition, vol. 54, pp. 8208-8212, 2015.

[258] G. Divitini, S. Cacovich, F. Matteocci, L. Cinà, A. Di Carlo, and C. Ducati, "In situ observation of heat-induced degradation of perovskite solar cells," Nature Energy, vol. 1, p. 15012, 2016.

217

[259] J. Yang, B. D. Siempelkamp, D. Liu, and T. L. Kelly, "Investigation of CH3NH3PbI3 degradation rates and mechanisms in controlled humidity environments using in situ techniques," ACS Nano, vol. 9, pp. 1955-1963, 2015.

[260] A. M. A. Leguy, Y. Hu, M. Campoy-Quiles, M. I. Alonso, O. J. Weber, P. Azarhoosh, M. van Schilfgaarde, M. T. Weller, T. Bein, J. Nelson, P. Docampo, and P. R. F. Barnes, "Reversible hydration of CH3NH3PbI3 in films, single crystals, and solar cells," Chemistry of Materials, vol. 27, pp. 3397-3407, 2015.

[261] S. Mastroianni, F. D. Heinz, J. H. Im, W. Veurman, M. Padilla, M. C. Schubert, U. Wurfel, M. Gratzel, N. G. Park, and A. Hinsch, "Analysing the effect of crystal size and structure in highly efficient CH3NH3PbI3 perovskite solar cells by spatially resolved photo- and electroluminescence imaging," Nanoscale, vol. 7, pp. 19653- 19662, 2015.

[262] F. Giordano, A. Abate, J. P. Correa Baena, M. Saliba, T. Matsui, S. H. Im, S. M. Zakeeruddin, M. K. Nazeeruddin, A. Hagfeldt, and M. Graetzel, "Enhanced electronic properties in mesoporous TiO2 via lithium doping for high-efficiency perovskite solar cells," Nature Communications, vol. 7, p. 10379, 2016.

[263] D. Bi, W. Tress, M. I. Dar, P. Gao, J. Luo, C. Renevier, K. Schenk, A. Abate, F. Giordano, J.-P. Correa Baena, J.-D. Decoppet, S. M. Zakeeruddin, M. K. Nazeeruddin, M. Grätzel, and A. Hagfeldt, "Efficient luminescent solar cells based on tailored mixed-cation perovskites," Science Advances, vol. 2, 2016.

[264] A. Abate, T. Leijtens, S. Pathak, J. Teuscher, R. Avolio, M. E. Errico, J. Kirkpatrik, J. M. Ball, P. Docampo, I. McPherson, and H. J. Snaith, "Lithium salts as "redox active" p-type dopants for organic semiconductors and their impact in solid-state dye-sensitized solar cells," Physical Chemistry Chemical Physics, vol. 15, pp. 2572-2579, 2013.

[265] A. Abate, D. R. Staff, D. J. Hollman, H. J. Snaith, and A. B. Walker, "Influence of ionizing dopants on charge transport in organic semiconductors," Physical Chemistry Chemical Physics, vol. 16, pp. 1132-1138, 2014.

[266] W. Zhou, Y. Zhao, C. Shi, H. Huang, J. Wei, R. Fu, K. Liu, D. Yu, and Q. Zhao, "Reversible healing effect of water molecules on fully crystallized metal–halide perovskite film," The Journal of Physical Chemistry C, vol. 10.1021/acs.jpcc.5b11465, 2016.

[267] G. E. Eperon, S. N. Habisreutinger, T. Leijtens, B. J. Bruijnaers, J. J. van Franeker, D. W. deQuilettes, S. Pathak, R. J. Sutton, G. Grancini, D. S. Ginger, R. A. J. Janssen, A. Petrozza, and H. J. Snaith, "The importance of moisture in hybrid lead halide perovskite thin film fabrication," ACS Nano, vol. 9, pp. 9380-9393, 2015.

218

[268] U. B. Cappel, T. Daeneke, and U. Bach, "Oxygen-induced doping of spiro-meotad in solid-state dye-sensitized solar cells and its impact on device performance," Nano Letters, vol. 12, pp. 4925-4931, 2012.

[269] Z. Hawash, L. K. Ono, S. R. Raga, M. V. Lee, and Y. Qi, "Air-exposure induced dopant redistribution and energy level shifts in spin-coated spiro-meotad films," Chemistry of Materials, vol. 27, pp. 562-569, 2015.

[270] L. K. Ono, P. Schulz, J. J. Endres, G. O. Nikiforov, Y. Kato, A. Kahn, and Y. Qi, "Air-exposure-induced gas-molecule incorporation into spiro-meotad films," The Journal of Physical Chemistry Letters, vol. 5, pp. 1374-1379, 2014.

[271] S. Fantacci, F. De Angelis, M. K. Nazeeruddin, and M. Grätzel, "Electronic and optical properties of the spiro-meotad hole conductor in its neutral and oxidized forms: A DFT/TDDFT investigation," The Journal of Physical Chemistry C, vol. 115, pp. 23126-23133, 2011.

[272] J. A. Christians, P. A. Miranda Herrera, and P. V. Kamat, "Transformation of the excited state and photovoltaic efficiency of CH3NH3PbI3 perovskite upon controlled exposure to humidified air," Journal of the American Chemical Society, vol. 137, pp. 1530-1538, 2015.

[273] J. Zhao, B. Cai, Z. Luo, Y. Dong, Y. Zhang, H. Xu, B. Hong, Y. Yang, L. Li, W. Zhang, and C. Gao, "Investigation of the hydrolysis of perovskite organometallic halide CH3NH3PbI3 in humidity environment," Scientific Reports, vol. 6, p. 21976, 2016.

[274] L. Zheng, Y.-H. Chung, Y. Ma, L. Zhang, L. Xiao, Z. Chen, S. Wang, B. Qu, and Q. Gong, "A hydrophobic hole transporting oligothiophene for planar perovskite solar cells with improved stability," Chemical Communications, vol. 50, pp. 11196- 11199, 2014.

[275] I. Celik, Z. Song, A. J. Cimaroli, Y. Yan, M. J. Heben, and D. Apul, "Life cycle assessment (LCA) of perovskite pv cells projected from lab to fab," Solar Energy Materials and Solar Cells, 2016, DOI: http://dx.doi.org/10.1016/j.solmat.2016.04.037.

[276] J. You, Z. Hong, Y. Yang, Q. Chen, M. Cai, T.-B. Song, C.-C. Chen, S. Lu, Y. Liu, H. Zhou, and Y. Yang, "Low-temperature solution-processed perovskite solar cells with high efficiency and flexibility," ACS Nano, vol. 8, pp. 1674-1680, 2014.

[277] P. Docampo, J. M. Ball, M. Darwich, G. E. Eperon, and H. J. Snaith, "Efficient organometal trihalide perovskite planar-heterojunction solar cells on flexible polymer substrates," Nature Communications, vol. 4, 2013, 10.1038/ncomms3761.

219

[278] C. Roldan-Carmona, O. Malinkiewicz, A. Soriano, G. Minguez Espallargas, A. Garcia, P. Reinecke, T. Kroyer, M. I. Dar, M. K. Nazeeruddin, and H. J. Bolink, "Flexible high efficiency perovskite solar cells," Energy & Environmental Science, vol. 7, pp. 994-997, 2014.

[279] D. Bryant, P. Greenwood, J. Troughton, M. Wijdekop, M. Carnie, M. Davies, K. Wojciechowski, H. J. Snaith, T. Watson, and D. Worsley, "A transparent conductive adhesive laminate electrode for high-efficiency organic-inorganic lead halide perovskite solar cells," Advanced Materials, vol. 26, pp. 7499-7504, 2014.

[280] B. J. Kim, D. H. Kim, Y.-Y. Lee, H.-W. Shin, G. S. Han, J. S. Hong, K. Mahmood, T. K. Ahn, Y.-C. Joo, K. S. Hong, N.-G. Park, S. Lee, and H. S. Jung, "Highly efficient and bending durable perovskite solar cells: Toward a wearable power source," Energy & Environmental Science, vol. 8, pp. 916-921, 2015.

[281] X. Y. Wang, Z. Li, W. J. Xu, S. A. Kulkarni, S. K. Batabyal, S. Zhang, A. Y. Cao, and L. H. Wong, "TiO2 nanotube arrays based flexible perovskite solar cells with transparent carbon nanotube electrode," Nano Energy, vol. 11, pp. 728-735, 2015, 10.1016/j.nanoen.2014.11.042.

[282] F. Di Giacomo, V. Zardetto, A. D'Epifanio, S. Pescetelli, F. Matteocci, S. Razza, A. Di Carlo, S. Licoccia, W. M. M. Kessels, M. Creatore, and T. M. Brown, "Flexible perovskite photovoltaic modules and solar cells based on atomic layer deposited compact layers and uv-irradiated TiO2 scaffolds on plastic substrates," Advanced Energy Materials, vol. 5, p. 1401808, 2015, 10.1002/aenm.201401808.

[283] M. Kaltenbrunner, G. Adam, E. D. Glowacki, M. Drack, R. Schwodiauer, L. Leonat, D. H. Apaydin, H. Groiss, M. C. Scharber, M. S. White, N. S. Sariciftci, and S. Bauer, "Flexible high power-per-weight perovskite solar cells with chromium oxide-metal contacts for improved stability in air," Nat Mater, vol. 14, pp. 1032-1039, 2015.

[284] C. D. Bailie, M. G. Christoforo, J. P. Mailoa, A. R. Bowring, E. L. Unger, W. H. Nguyen, J. Burschka, N. Pellet, J. Z. Lee, M. Graetzel, R. Noufi, T. Buonassisi, A. Salleo, and M. D. McGehee, "Semi-transparent perovskite solar cells for tandems with silicon and cigs," Energy & Environmental Science, vol. 8, pp. 956-963, 2015.

[285] C.-C. Chen, S.-H. Bae, W.-H. Chang, Z. Hong, G. Li, Q. Chen, H. Zhou, and Y. Yang, "Perovskite/polymer monolithic hybrid tandem solar cells utilizing a low- temperature, full solution process," Materials Horizons, vol. 2, pp. 203-211, 2015.

[286] J. P. Mailoa, C. D. Bailie, E. C. Johlin, E. T. Hoke, A. J. Akey, W. H. Nguyen, M. D. McGehee, and T. Buonassisi, "A 2-terminal perovskite/silicon multijunction solar cell enabled by a silicon tunnel junction," Applied Physics Letters, vol. 106, 2015. 220

[287] L. Kranz, A. Abate, T. Feurer, F. Fu, E. Avancini, J. Löckinger, P. Reinhard, S. M. Zakeeruddin, M. Grätzel, S. Buecheler, and A. N. Tiwari, "High-efficiency polycrystalline thin film tandem solar cells," The Journal of Physical Chemistry Letters, vol. 6, pp. 2676-2681, 2015.

[288] S. Albrecht, M. Saliba, J. P. Correa Baena, F. Lang, L. Kegelmann, M. Mews, L. Steier, A. Abate, J. Rappich, L. Korte, R. Schlatmann, M. K. Nazeeruddin, A. Hagfeldt, M. Gratzel, and B. Rech, "Monolithic perovskite/silicon-heterojunction tandem solar cells processed at low temperature," Energy & Environmental Science, vol. 9, pp. 81-88, 2016.

221

Appendix A

Publication List

[1] Z. Song, A. Abate, S. C. Watthage, G. K. Liyanage, A. B. Phillips, U. Steiner, M.

Graetzel and M. J. Heben, "Perovskite Solar Cell Stability in Humid Air: Partially

Reversible Phase Transitions in the PbI2-CH3NH3I-H2O System", Advanced Energy

Materials, DOI: 10.1002/aenm.201600846, 2016.

[2] Z. Song, S. C. Watthage, A. B. Phillips, B. L. Tompkins, R. J. Ellingson, and M. J.

Heben, “Impact of Processing Temperature and Composition on the Formation of

Methylammonium Lead Iodide Perovskites”, Chemistry of Materials, 27 (13), 4612–

4619, 2015. (ACS Editors’ Choice & Journal’s cover)

[3] Z. Song, S. C. Watthage, A. B. Phillips, and M. J. Heben, “Pathways toward High

Performance Perovskite Solar Cells: A Review on Recent Advances of Organometal

Halide Perovskites for Photovoltaic Applications”, Journal of Photonics for Energy, 6

(2), 022001, 2016. (Invited review)

[4] Z. Song, A. Abate, S. C. Watthage, G. K. Liyanage, A. B. Phillips, U. Steiner, M.

Graetzel and M. J. Heben, “In-Situ Observation of Moisture-Induced Degradation of

Perovskite Solar Cells Using Laser-Beam Induced Current”, in IEEE 43nd Photovoltaic

Specialists Conference (PVSC), 2016. (Best Student Paper Finalists)

222

[5] Z. Song, S. C. Watthage, A. B. Phillips, G. K..Liyanage, R. R. Khanal, and M. J.

Heben, “Investigation of degradation mechanisms of perovskite-based photovoltaic devices using laser beam induced current mapping”, Proc. SPIE, 9561, 956107, 2015.

(Newport Research Excellence Awards)

[6] Z. Song, S. C. Watthage, B. L. Tompkins, G. K. Liyanage, A. B. Phillips, R. J.

Ellingson and M. J. Heben, “Spatially Resolved Characterization of Solution Processed

Perovskite Solar Cells Using the LBIC Technique”, in IEEE 42nd Photovoltaic

Specialists Conference (PVSC), 1-5, 2015. (Best Student Paper Awards)

[7] Z. Song, A. B. Phillips, P. W. Krantz, and M. J. Heben, “Spray Pyrolysis of Backwall

Superstrate CuIn(S, Se)2 Solar Cells”, in IEEE 40th Photovoltaic Specialists Conference

(PVSC), 1712-1717, 2014.

[8] Z. Song, A. B. Phillips, P. W. Krantz, T. Prabhakar, R. R. Khanal, Y. Xie, J. L.

DeWitt, J. M. Stone, and M. J. Heben, “Spray Pyrolysis of CuIn(S, Se)2 Thin Films

Using Hydrazine-based Solutions”, MRS Proceedings, 1630, o01-03, 2014.

[9] Z. Song, A. B. Phillips, Y. Xie, R. R. Khanal, J. M. Stone, and M. J. Heben, “The

Effect of Wettability on Hydrazine Processed CZTS Thin Films”, MRS Proceedings,

1648, hh03-03, 2014.

[10] Z. Song, A. B. Phillips, Y. Xie, R. R. Khanal, and M. J. Heben, “The Performance of

Nanocrystalline CuInS2/In2S3/SnO2 Heterojunction Solar Cells Prepared by Chemical

Spray Pyrolysis”, in IEEE 39th Photovoltaic Specialists Conference (PVSC), 2540-2544,

2013.

223

[11] I. Celik, Z. Song, A. Cimaroli, Y. Yan, M. J. Heben, and D. Apul, “Life Cycle

Assessment (LCA) of perovskite PV cells projected from lab to fab”, Solar Energy

Materials and Solar Cells, DOI:10.1016/j.solmat.2016.04.037, 2016.

[12] A. B. Phillips, Z. Song, J. L. DeWitt, J. M. Stone, P. W. Krantz, J. Royston, R.

Zeller, M. Mapes, P. J. Roland, M. Dorogi, S. Zafar, G. Faykosh, R. J. Ellingston, and M.

J. Heben, "High speed, large area laser beam induced current imaging and laser scribing system for photovoltaic devices and modules," Review of Scientific Instruments, 2016.

[13] A. B. Phillips, R. R. Khanal, Z. Song, S. C. Watthage, K. R. Kormanyos, and M. J.

Heben, “Simultaneous pinhole protection and back contact formation for CdTe solar cells with spray-deposited single wall carbon nanotubes layers”, Applied Physics Letters, 107,

253901, 2015.

[14] R. R. Khanal, A. B. Phillips, Z. Song, Y. Xie, H. P. Mahabaduge, M. D. Dorogi, S.

Zafar, G. T. Faykosh, and M. J. Heben, “Substrate Configuration, Bifacial CdTe Solar

Cells Grown Directly on Transparent Single Wall Carbon Nanotube Back contacts”,

Solar Energy Materials and Solar Cells, 157, 35-41, 2016.

[15] A. B. Phillips, R. R. Khanal, Z. Song, R. M. Zartman, J. L. DeWitt, J. M. Stone, P. J.

Roland, V. V. Plotnikov, C. W. Carter, J. M. Stayancho, R. J. Ellingson, A. D. Compaan, and M. J. Heben, “Wiring-up carbon single wall nanotubes to polycrystalline inorganic semiconductor thin films: Low-barrier, copper-free back contact to CdTe solar cells”,

Nano Letters, 13 (11), 5224–5232, 2013.

[16] S. C. Watthage, Z. Song, G. K. Liyanage, A. B. Phillips, M. J. Heben, “Investigation on the Nucleation and Growth Mechanisms of Perovskite Formation in the Two-Step

Solution Process”, in IEEE 43nd Photovoltaic Specialists Conference (PVSC), 2016. 224

[17] I. Celik, Z. Song, M. J. Heben, Y. Yan, D. Apul, “Life Cycle Toxicity Analysis of

Unstable PV Cells”, in IEEE 43nd Photovoltaic Specialists Conference (PVSC), 2016.

[18] T. Prabhakar, Z. Song, M. J. Heben, and Y. Yan, “Synthesis of Single-Phase

Cu2ZnSnS4 Thin Films by Ultrasonic Spray Pyrolysis”, in IEEE 39th Photovoltaic

Specialists Conference (PVSC), 0409-0413, 2013.

[19] R. R. Khanal, A. B. Phillips, Z. Song, Y. Xie, H. P. Mahabaduge, M. D. Dorogi, S.

Zafar, G. T. Faykosh, and M. J. Heben, “CdTe/CdS Thin Film Solar Cells in the

Substrate Configuration on a Single-Wall Carbon Nanotube Back Contact”, in IEEE 39th

Photovoltaic Specialists Conference (PVSC), 1126-1130, 2013.

[20] R. R. Khanal, A. B. Phillips, Z. Song, V. V. Plotnikov, C. W. Carter, J. M.

Stayancho, and M. J. Heben, “Semiconducting Carbon Single-walled Nanotubes as a Cu-

Free, Barrier-Free Back Contact for CdTe Solar Cell” in IEEE 40th Photovoltaic

Specialists Conference (PVSC), 2348-2353, 2014.

[21] A. B. Phillips, B. L. Tompkins, Z. Song, R. R. Khanal, G. K. Liyanage, N. D. Gapp,

D. M. Wilt, and M. J. Heben, “Carbon Nanotube Reinforced Cu Metal Matrix

Composites for Current Collection from Space Photovoltaics” in IEEE 42nd Photovoltaic

Specialists Conference (PVSC), 2015.

225