ENGINEERING MOLYBDENUM SULFIDE ELECTROCATALYSTS

AND SILICON PHOTOCATHODES FOR HYDROGEN PRODUCTION

VIA SOLAR WATER SPLITTING

A DISSERTATION

SUBMITTED TO THE DEPARTMENT OF CHEMICAL ENGINEERING

AND THE COMMITTEE ON GRADUATE STUDIES

OF STANFORD UNIVERSITY

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

JESSE D. BENCK

AUGUST 2015

© 2015 by Jesse Daniel Benck. All Rights Reserved. Re-distributed by Stanford University under license with the author.

This work is licensed under a Creative Commons Attribution- Noncommercial 3.0 United States License. http://creativecommons.org/licenses/by-nc/3.0/us/

This dissertation is online at: http://purl.stanford.edu/df442wf0767

ii I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Thomas Jaramillo, Primary Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Stacey Bent

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Bruce Clemens

Approved for the Stanford University Committee on Graduate Studies. Patricia J. Gumport, Vice Provost for Graduate Education

This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file in University Archives.

iii

iv

Abstract

Hydrogen is a critical chemical reagent and energy carrier, but it is currently produced from fossil fuels, which are limited in supply and create harmful CO2 emissions when consumed. The development of new, sustainable methods for hydrogen production represents an important research challenge. Photoelectrochemical (PEC) water splitting, a process in which H2O is split into H2 and O2 using the energy from sunlight, is a promising technology for renewable hydrogen production. To make this process viable for widespread implementation, efficient, inexpensive, stable, and scalable PEC water splitting materials and devices must be developed. Creating active catalysts, strategies for corrosion prevention, and techniques for successfully integrating all required device components are especially important barriers to overcome.

The first part of this dissertation focuses on molybdenum sulfide catalysts for the electrochemical hydrogen evolution reaction (HER). We begin by analyzing a selection of state- of-the-art molybdenum sulfide HER catalysts to identify best practices for measuring catalytic activity and design principles for creating even more effective catalyst materials. Then, we perform a detailed investigation of an amorphous molybdenum sulfide (MoSx) catalyst. Using a new room temperature wet chemical procedure, we synthesize a highly active form of this amorphous MoSx. Then, we attempt to understand the physical and chemical characteristics of this material that result in its high activity. Using electrochemical measurements and ex situ spectroscopic characterization, we reveal that this material initially possess a composition of

MoS3, but after catalysis, the surface is reduced to a composition and chemical state resembling

MoS2.

To understand more about the mechanism of the catalyst transformation and the nature of the active phase under operating conditions, we use advanced in situ characterization techniques including ambient pressure X-ray photoelectron spectroscopy and environmental transmission electron microscopy to track the transformation of amorphous MoSx nanoparticles under hydrogen evolution conditions. These experiments show that the surface of the amorphous

MoSx catalyst is dynamic: the initial catalyst reduction forms the active surface of amorphous

MoS2, but further transformation continues during electrochemical operation, and some portions of the catalyst are eventually converted to crystalline MoS2. This process could contribute to the deactivation of the amorphous MoSx catalyst after prolonged operation.

v

Our efforts next shift to the development of integrated water splitting photocathodes, which incorporate both HER catalysts and semiconductor light absorbers. Silicon represents a particularly promising small band gap semiconductor for application in PEC water splitting devices, but this material possesses low catalytic activity and poor durability in aqueous electrolytes. We demonstrate that molybdenum sulfide nanomaterials can provide both corrosion protection and catalytic activity in silicon photocathodes. We engineer thin, conformal

MoS2 surface coatings to protect silicon absorbers, resulting in photocathodes that can operate for 100 hours with no loss in performance. We study the atomic-scale surface structure of these devices and identify the characteristics of the MoS2 layer that provide both catalytic activity and excellent stability. To further improve the performance of these structures, we incorporate a molybdenum sulfide molecular cluster catalyst and obtain further gains in the device performance.

Finally, we develop new Si photocathode architectures that address the challenge of successfully integrating multiple water splitting device components while retaining a very high photovoltage from the illuminated semiconductor. Silicon surfaces and interfaces control many key aspects of device performance. We focus on engineering these interfaces to reduce surface-mediated recombination using strategies inspired by high performance silicon photovoltaics. These efforts result in Si photocathodes with improved photovoltage and provide a platform for the fabrication of integrated, monolithic dual absorber water splitting devices.

In summary, this dissertation covers fundamental studies of molybdenum sulfide HER catalysts as well as device engineering efforts to create high performance silicon photocathodes. These results represent important advancements towards large-scale renewable H2 production using PEC water splitting.

vi

Acknowledgements

Graduate school is a challenging experience in many ways. I would not have been able to complete this dissertation without help and support from many wonderful mentors, colleagues, friends, and family members.

I would like to begin by thanking Professor Thomas Jaramillo, who has been the best doctoral advisor I could have hoped for. He is an outstanding scientist and engineer, with a remarkable knowledge of the literature and a clear, systematic approach to problem solving. He is also an unusually gifted communicator and generous scientific community member. Perhaps most unusual and impressive are his deep commitment to his students’ development and his incredibly positive and encouraging attitude. Tom has given me many amazing opportunities to interact with respected scientists, present my research results, and teach a course, all of which helped me to develop my skills as a researcher and educator. I will be forever grateful for the many ways in which he has deeply impacted my life during my time at Stanford.

I am also grateful for all the advice and supported provided by my reading committee members, Professors Stacey Bent and Bruce Clemens. Stacey helped me make the transition to graduate school when she took me into her group as a rotation student during my first quarter at Stanford. I also served as a teaching assistant for her course, Chemical Engineering 25E, for two years, and learned a great deal from Stacey’s incredible organization and carefully considered approach to teaching. Bruce has also had a great influence on me due to his calm and thoughtful demeanor, insightful questions, and wonderful sense of humor. I am very thankful that Bruce allowed me to become an honorary Clemens Group member during my last two of years at Stanford. My interactions with Stacey and Bruce were always encouraging and inspiring.

It would have been impossible for me to complete any work at all without the continual support of the Chemical Engineering Department staff members. Over the years, Jeanne Cosby has been an especially important source of guidance. Pam Juanes and Rob Rome always ensured that the department ran smoothly. I thank Victoria Lee, Annie Jensen, and Olayinka Popoola for providing administrative support directly to the Jaramillo Group. Many other staff members including Pamela Dixon, Sandra Handy, Andrea Hubbard, Eric Ngyuen, and others also provided critical help over the years. Several facilities staff members including Todd Eberspacher, Ray Hewlett, and Justine Sousa also helped me overcome many challenges.

vii

Much of the research that I completed at Stanford was performed at shared user facilities, which are truly remarkable resources, not only due to the state-of-the-art equipment they contain, but also because of the dedicated staff members that provide support to users. I would like to thank Richard Chin, Bob Jones, Chuck Hitzmann, Jeff Tok, and Arturas Vailionis of the Stanford Nano Shared Facilities for training and support on many characterization tools that were critical to my work. I also thank Mary Tang, Maurice Stevens, James Conway, Uli Thumser, Mahnaz Mansourpor, Usha Raghuram, Michelle Rincon, and Nancy Latta of the Stanford Nanofabrication Facility for training and support on semiconductor processing tools.

Throughout my time at Stanford, in addition to improving my research skills, it has also been important to me to develop my abilities as an educator. I am very lucky to have worked with several amazing teaching mentors. In particular, I am grateful to Lisa Hwang for her continual willingness to share thoughtful and practical advice on teaching. Lisa’s philosophy about teaching as a vehicle for personal development helped me benefit as much as possible from my time as a teaching assistant. In addition, Lisa created an incredible community of support for teaching-related activities through the Chemical Engineering Mentor TA Program, and I am lucky to have participated in this program for several years. Jennifer Schwartz-Poehlmann also had a great impact on me through the Mentors in Teaching Program, which allowed me to engage with many committed teaching assistants and instructors from across the university. I am especially grateful to the Chemical Engineering Department administration and to Professor Tom Jaramillo for giving me the opportunity to serve as a co-instructor for Chemical Engineering 25E. I could not have completed this course successfully without the support of two organized and tireless teaching assistants, Kevin Hurlbutt and Jeremy Feaster.

I have been fortunate enough to work with a large number of collaborators from other research groups, all of whom greatly influenced my scientific development. My first collaboration was with Hernan Sanchez-Casalongue, Sarp Kaya, Hirohito Ogasawara, and Professor Anders Nilsson at the Stanford Synchrotron Radiation Lightsource. Our work together focused on in situ ambient pressure x-ray photoelectron spectroscopy of an amorphous molybdenum sulfide catalyst. Hernan is extremely kind, very knowledgeable, and a great deal of fun to spend time with during overnight beam times. I also worked with Sang Chul Lee and his advisor, Professor Robert Sinclair, on a variety of transmission electron microscopy measurements. Sang Chul was very easy to work with and always produced incredible data in spite of the challenges associated with many of our measurements. Charlie Tsai and his advisors, Frank Abild-Pederson and

viii

Professor Jens Nørskov, contributed a great deal of fundamental understanding to various molybdenum sulfide-related studies through their density functional theory calculations. Charlie was always willing to take on new challenges and is already an impressively prolific researcher. On several occasions when I asked Charlie about a new idea, I discovered that he had already completed the necessary calculations! I also greatly enjoyed working with Vijay Parameshwaran and his advisor, Professor Bruce Clemens, on integrating molybdenum sulfide materials as catalysts and protective coatings on III-V semiconductors such as GaAs. While this work is not included in this thesis, it was an excellent learning experience. Through this process Vijay was always eager to share his knowledge of materials characterization techniques and semiconductor device physics. Finally, I have had the pleasure of working with Jieyang Jia, Yijie Huo, and Professor James Harris (along with fellow Jaramillo Group member Linsey Seitz and others) on H2 production through photovoltaic-driven electrolysis. While these experiments are not included in this thesis, this work challenged many of my fundamental assumptions about solar water splitting, and as a result, my understanding of the process is much deeper. Jieyang’s relentless enthusiasm and drive were also great source of motivation to push this project forward.

The relationships I have had with my fellow Jaramillo Group members are one of the most important elements of my experience at Stanford. The Jaramillo Group has supported me through difficult challenges and celebrated successes with me. Each and every Jaramillo Group member has been an outstanding colleague and treasured friend. I must especially thank my mentors from the photoelectrochemistry sub-group, Zhebo Chen and Blaise Pinaud, for teaching me the basics of this complex field and always maintaining the highest standards for quality in every element of their own work, which motivated me to strive to meet these standards as well. The other members of the PEC sub-group, including Linsey Seitz, Pong Chakthranont, Tommy Hellstern, Ieva Narkeviciute, Reuben Britto, and Arnold Forman also taught me a great deal and were always willing to help solve problems or brainstorm new ideas. Many members of other sub-groups became close friends as well, especially my former desk neighbors Yelena Gorlin, Jakob Kibsgaard, Ariel Jackson, Desmond Ng, and Maureen Tang. I also enjoyed spending time with all the rest of the Jaramillo Group, including Jeremy Feaster, Toru Hatsukade, David Abram, Kendra Kuhl, Etosha Cave, Benjamin Reinecke, Peter Vesborg, Chris Hahn, Sam Fleischman, Annalie Jongerius, Maria Escribano Escudero, Alessandro Gallo, Brenna Gibbons, Jon Snider, Alaina Strickler, Stephanie Nitopi, Mike Boyd, Bill Chen, Sung-Hyeon Baek Shin- Jung Choi, Zhi Wei Shi, Laurie King, and Ari Saha. I am especially grateful to have known and

ix worked with a terrific undergraduate mentee, Kara Fong, who has continually inspired me with her work ethic, ingenuity, and ability to take feedback and use it to rapidly improve her skills. I am also proud of all the younger PhD students I mentored, including Tommy Hellstern, Jeremy Feaster, and Reuben Britto, among others. All of these excellent students have become gifted scientists on whom I now rely for advice. After six years in the group, I have far too many wonderful memories of conferences, group socials, Danish Christmas Lunches, and more good times with all of these Jaramillo Group members to list here. I will forever value the experiences we shared.

All my friends at Stanford and beyond have been a great source of emotional support and much needed fun! In particular, I appreciate all the great times I spent with my Chemical Engineering cohort, including Amy Tsui, Katie Pickrahn, Will Durand, Elyse Coletta, Vivek Narsimhan, Jack Chai, Craig Buckley, Bonggeun Shong, Ooi Chin Chun, Mallory Hammock, and Briana Dunn. In particular, Amy was a remarkably organized and fun travel companion, and always impressed me with her excellent presentation skills as well as her desire to make a positive impact on the world around her. Katie is one of the most popular people I have ever met due to her kindness and good-hearted troublemaking. I enjoyed many meals and great conversations with Katie and her husband, Kevin. Although Will left Stanford to complete his PhD in Texas, he and his wife Lauren will be life-long friends. I truly enjoyed working on the Chemical Engineering Action Committee with Elyse and Vivek and sharing many other happy times with the rest of my classmates. I have also had lots of wonderful friends from outside my class at Stanford, including Peter Mulligan, Debora Lin, Ingrid Lawhorn, Tom Lampo, David Bergsman, and Andrew Doyle among many others. Away from the Stanford community, Eamonn Collins has been a great source of advice, emotional support, and laughter. He always challenged me to think deeply about social conventions, encouraged me to push myself out of my comfort zone, and made me stand a bit too close to the edge of various cliffs. I have also enjoyed time spent with my old friends Jan Ressl, Sarah Kuech, and Sam Rossmeissl.

My family has helped me achieve everything I have accomplished in my life. My parents have always been extremely supportive, encouraging, and loving. All my wonderful childhood experiences provided me with the self-confidence, sense of adventure, and openness to new experiences that have helped me ove rcome all the challenges I have faced in my life. My sister and all my cousins have been wonderful companions and friends through the years. All my aunts, uncles, and grandparents served as excellent models for how to live and contributed a

x great deal to my development. I genuinely enjoy spending time with all of them. I would never have made it to Stanford let alone completed this dissertation without my family.

Finally, I am incredibly grateful to have met Briana Dunn at Stanford. Briana is unfalteringly kind and compassionate. Her amazing talents as an educator have inspired me to better myself. She has always understood the challenges of graduate school and her sense of humor helps me keep these challenges in perspective. She encourages me to prioritize the elements of my life that I value most. I always have fun spending time with her, whether we are embarking on a new travel adventure, visiting a used book store, or just watching TV after a long day in lab. I look forward to many more wonderful years together!

xi

xii

Publications

1. J.D. Benck, Z. Chen, L.Y. Kuritzky, A.J. Forman, and T.F. Jaramillo. "Amorphous Molybdenum Sulfide Catalysts for Electrochemical Hydrogen Production: Insights into the Origin of their Catalytic Activity." ACS Catalysis, 2012. 2 (9): 1916-1923. http://dx.doi.org/10.1021/cs300451q

2. Y. Gorlin, B. Lassalle-Kaiser, J.D. Benck, S. Gul, S.M. Webb, V.K. Yachandra, J. Yano, and T.F. Jaramillo. "In Situ X-ray Absorption Spectroscopy Investigation of a Bifunctional Manganese Oxide Catalyst with High Activity for Electrochemical Water Oxidation and Oxygen Reduction." Journal of the American Chemical Society, 2013. 135 (23): 8525-8534. http://dx.doi.org/10.1021/ja3104632

3. B.A. Pinaud, J.D. Benck, L.C. Seitz, A.J. Forman, Z.B. Chen, T.G. Deutsch, B.D. James, K.N. Baum, G.N. Baum, S. Ardo, H.L. Wang, E. Miller, and T.F. Jaramillo. "Technical and economic feasibility of centralized facilities for solar hydrogen production via photocatalysis and photoelectrochemistry." Energy & Environmental Science, 2013. 6 (7): 1983-2002. http://dx.doi.org/10.1039/C3EE40831K

4. L.C. Seitz, Z.B. Chen, A.J. Forman, B.A. Pinaud, J.D. Benck, and T.F. Jaramillo. "Modeling the Practical Performance Limits of Photoelectrochemical Water Splitting Based on the Current State of Materials Research." ChemSusChem, 2014, 7 (5): 1372- 1385. http://dx.doi.org/10.1002/cssc.201301030

5. Y. Gorlin, C.J. Chung, J.D. Benck, D. Nordlund, L.C. Seitz, T.C. Weng, D. Sokaras, B.M. Clemens, and T.F. Jaramillo. "Understanding interactions between manganese oxide and gold that lead to enhanced activity for electrocatalytic water oxidation." Journal of the American Chemical Society, 2014, 136 (13): 4920-4926. http://dx.doi.org/10.1021/ja407581w

6. J.D. Benck, S.C. Lee, K.D. Fong, J. Kibsgaard, R. Sinclair, and T.F. Jaramillo. "Designing Active and Stable Silicon Photocathodes for Solar Hydrogen Production Using Molybdenum Sulfide Nanomaterials." Advanced Energy Materials, 2014, 4: 1400739. http://dx.doi.org/10.1002/aenm.201400739

xiii

7. J.D. Benck,* B.A. Pinaud,* Y. Gorlin, and T.F. Jaramillo. "Substrate Selection for Fundamental Studies of Electrocatalysts and Photoelectrodes: Inert Potential Windows in Acidic, Neutral, and Basic Electrolyte." PLOS ONE. 2014, 9 (10): e107942. http://dx.doi.org/10.1371/journal.pone.0107942

8. J.D. Benck,* T.R. Hellstern,* J. Kibsgaard, P. Chakthranont, and T.F. Jaramillo. “Catalyzing the Hydrogen Evolution Reaction (HER) with Molybdenum Sulfide Nanomaterials." ACS Catalysis, 2014, 4: 3957-3971. http://dx.doi.org/10.1021/cs500923c

9. H.G. Sanchez Casalongue, J.D. Benck, C. Tsai, R.K.B. Karlsson, S. Kaya, M.L. Ng, L.G.M. Pettersson, F. Abild-Pedersen, J.K. Nørskov, H. Ogasawara, T.F. Jaramillo, and A. Nilsson. “Operando Characterization of an Amorphous Molybdenum Sulfide Nanoparticle Catalyst During the Hydrogen Evolution Reaction." Journal of Physical Chemistry C, 2014, 118 (50): 29252-29259. http://dx.doi.org/10.1021/jp505394e

10. J.W.D. Ng, T.R. Hellstern, J. Kibsgaard, A.C. Hinckley, J.D. Benck, and T.F. Jaramillo. “Polymer Electrolyte Membrane Electrolyzers which Utilize Non-Precious Mo-based Hydrogen Evolution Catalysts.” ChemSusChem, accepted 2015.

11. J. Kibsgaard, C. Tsai, K. Chan, J.D. Benck, J.K. Norskov, F. Abild-Pedersen, and T.F. Jaramillo. “Trends in Hydrogen Evolution Reaction Activity of Transition Metal Phosphides.” Energy and Environmental Science, accepted 2015.

12. T.R. Hellstern,* J.D. Benck,* J. Kibsgaard, C. Hahn, and T.F. Jaramillo. “Engineering cobalt phosphide (CoP) thin film catalysts for enhanced hydrogen evolution activity on silicon photocathodes.” Submitted, 2015.

13. S.C. Lee, J.D. Benck, C. Tsai, J. Park, F. Abild-Pedersen, T.F. Jaramillo, and R. Sinclair. “Chemical and Phase Evolution of Amorphous Molybdenum Sulfide Catalysts for Electrochemical Hydrogen Production Directly Observed Using Environmental Transmission Electron Microscopy.” In Preparation, 2015.

*Equally contributing authors

xiv

Table of Contents

Abstract ...... v

Acknowledgements ...... vii

Publications ...... xiii

Chapter 1: Introduction ...... 1

1.1 Motivation ...... 1

1.2 Background ...... 2

1.2.1 Water Splitting Device Components ...... 3

1.2.2 Efficiency ...... 6

1.2.3 Hydrogen Cost ...... 9

1.2.4 Research Challenges ...... 11

1.3 Scope ...... 12

1.4 Copyright ...... 13

1.5 References ...... 13

Chapter 2: Catalyzing the Hydrogen Evolution Reaction (HER) with Molybdenum Sulfide Nanomaterials ...... 21

2.1 Abstract ...... 21

2.2 Introduction ...... 21

2.3 Background ...... 22

2.3.1 The Hydrogen Evolution Reaction ...... 23

2.3.2 Molybdenum Disulfide...... 25

2.4 Parameters for Catalyst Activity Comparison ...... 27

2.4.1 Total Electrode Activity Measurements ...... 28

2.4.2 Intrinsic Activity Measurements ...... 31

2.4.3 Complicating Factors ...... 36

2.5 State of the Art Molybdenum Sulfide Electrocatalysts ...... 37

xv

2.5.1 Crystalline MoS2 ...... 37

2.5.2 Amorphous Molybdenum Sulfides ...... 42

2.5.3 Molybdenum Sulfide Molecular Clusters ...... 43

2.5.4 Stability of Molybdenum Sulfides...... 45

2.5.5 Emerging Directions ...... 46

2.6 Comparison of MoS2 Catalysts ...... 48

2.6.1 Intrinsic Activity ...... 48

2.6.2 Total Electrode Activity ...... 50

2.7 Applications ...... 51

2.7.1 Molybdenum sulfide in photoelectrodes ...... 51

2.7.2 Acid Electrolyzers ...... 52

2.8 Conclusions ...... 54

2.9 Copyright ...... 55

2.10 Author Contributions ...... 55

2.11 Acknowledgments ...... 55

2.12 References ...... 56

Chapter 3: Amorphous Molybdenum Sulfide Catalysts for Electrochemical Hydrogen Production: Insights into the Origin of their Catalytic Activity ...... 67

3.1 Abstract ...... 67

3.2 Introduction ...... 67

3.3 Methods ...... 69

3.3.1 Catalyst Synthesis ...... 69

3.3.2 Physical and Chemical Characterization ...... 70

3.3.3 Electrochemical Characterization ...... 70

3.4 Results and Discussion ...... 72

3.4.1 Physical and Chemical Characterization ...... 72

3.4.2 Catalyst Activation and Hydrogen Evolution Activity ...... 80

xvi

3.4.3 Changes in Catalyst Chemical State after Electrochemical Testing ...... 83

3.4.4 Turn Over Frequency ...... 86

3.4.5 Relationship between Catalyst Mass Loading and Activity ...... 90

3.4.6 Catalyst Stability ...... 96

3.5 Conclusions ...... 98

3.6 Copyright ...... 99

3.7 Author Contributions ...... 99

3.8 Acknowledgments ...... 100

3.9 References ...... 100

Chapter 4: In Situ Observation of the Chemical and Structural Evolution of Amorphous

Molybdenum Sulfide Catalysts for Electrochemical H2 Production ...... 105

4.1 Abstract ...... 105

4.2 Introduction ...... 105

4.3 Methods ...... 107

4.3.1 Catalyst and Electrochemical Cell Preparation for APXPS Measurements ...... 107

4.3.2 APXPS Measurements and Data Analysis ...... 109

4.3.3 Catalyst Synthesis for TEM Measurements ...... 110

4.3.4 Electrochemical Catalyst Activation and Activity Measurement for TEM Samples ...... 110

4.3.5 TEM Sample Preparation and Imaging ...... 111

4.4 Results and Discussion ...... 112

4.4.1 Ambient Pressure X-ray Photoelectron Spectroscopy ...... 112

4.4.2 Transmission Electron Microscopy of Electrochemically Activated MoSx ...... 115

4.3.3 Environmental TEM Measurements of MoSx Activated by In Situ Annealing ..... 119

4.5 Conclusions ...... 124

4.6 Copyright ...... 125

4.7 Author Contributions ...... 125

xvii

4.8 Acknowledgments ...... 126

4.9 References ...... 127

Chapter 5: Designing Active and Stable Silicon Photocathodes for Solar Hydrogen Production using Molybdenum Sulfide Nanomaterials ...... 131

5.1 Abstract ...... 131

5.2 Introduction ...... 131

5.3 Methods ...... 133

5.3.1 Device Synthesis ...... 133

5.3.2 Physical and Chemical Characterization ...... 134

5.3.3 Electrochemical Characterization ...... 135

5.3.4 Hydrogen Quantification ...... 138

5.4 Results and Discussion ...... 140

5.4.1 Physical Characterization Using Electron Microscopy ...... 140

5.4.2 Chemical Characterization Using X-ray Photoelectron Spectroscopy ...... 145

5.4.3 Photoelectrochemical Hydrogen Evolution Activity ...... 147

5.4.4 Optical Absorption, Reflection, and Transmission ...... 148

5.4.5 Photon-to-Current Conversion Efficiency ...... 151

5.4.6 Photoelectrochemical Hydrogen Evolution Stability ...... 152

5.4.7 Identifying Performance Limitations ...... 155

2- 5.4.8 Improving Hydrogen Evolution Catalysis with [Mo3S13] Clusters ...... 156

5.5 Conclusions ...... 169

5.6 Copyright ...... 169

5.7 Author Contributions ...... 170

5.8 Acknowledgments ...... 170

5.9 References ...... 170

Chapter 6: Photovoltaic-Inspired Silicon Photocathodes with Interfaces Engineered for

High Performance Solar H2 Production ...... 175

xviii

6.1 Abstract ...... 175

6.2 Introduction ...... 175

6.3 Methods ...... 177

6.3.1 Photocathode Fabrication ...... 177

6.3.2 Physical and Chemical Characterization ...... 187

6.3.3 Photoelectrochemical Characterization ...... 187

6.4 Results and Discussion ...... 191

+ 6.4.1 “Generation 1” Pt/SiO2-n p Si photocathodes ...... 191

+ 6.4.2 “Generation 2” Pt/Si3N4-n p Si photocathodes ...... 194

+ + 6.4.3 “Generation 3” Pt/SiO2-n pp Si photocathodes ...... 197

6.4.4 Future Si photocathodes ...... 203

6.5 Conclusions ...... 205

6.6 Author Contributions ...... 205

6.7 Acknowledgments ...... 205

6.8 References ...... 205

Chapter 7: Conclusions and Future Directions ...... 213

7.1 Conclusions ...... 213

7.2. Future Directions ...... 214

xix

xx

List of Tables

Table 1.1. Water splitting half reactions as they occur in acid...... 3 Table 2.1. Elementary steps in the hydrogen evolution reaction...... 23 Table 6.1. Sheet resistance values measured for p-type silicon wafers as received and after various doping procedures. The sheet resistance of the SiO2-masked back side of the doped wafer matches the value measured for the undoped wafer within the experimental error, confirming that the SiO2 mask is effective at preventing the back surface from becoming doped...... 187

xxi

xxii

List of Figures

Figure 1.1. Schematic energy band diagram of a PEC water splitting system illustrating key device components and processes. Adapted from Chen, et al.25 ...... 4 Figure 1.2. Predicted maximum attainable STH efficiency as a function of top absorber and bottom absorber band gap for a stacked dual absorber water splitting device based on numerical modeling calculations. The dashed line represents the 1.12 eV band gap of silicon, indicating that water splitting devices incorporating silicon as the bottom absorber have the potential to achieve high STH efficiencies. Reproduced with permission from John Wiley and Sons.36 ..... 8 Figure 1.3. Schematic diagrams of the four reactor types considered in the technoeconomic analysis conducted to estimate the cost of H2 produced via hypothetical large-scale PEC water splitting facilities. Drawings not to scale. Reproduced by permission of The Royal Society of Chemistry.23 ...... 10 Figure 2.1. Exchange current density as a function of hydrogen adsorption free energy for various HER catalyst materials. a. The experimental “volcano plot” for HER is shown and Pt, with slightly negative hydrogen absorption energy, has the highest HER activity. 25 b. The Parsons theoretical HER volcano predicts catalysts with hydrogen binding energy equals zero will have the highest activity.26 Further details concerning metrics for catalyst activity are given in section 3.1. Reproduced with permission.25, 26 ...... 24

Figure 2.2. a. Structure of the 2H, 3R, and 1T polytypes of MoS2. b. Top view of the Mo edge and S edge of a bulk MoS2 crystal...... 26 Figure 2.3. Equation relating HER current and turn over frequency of each catalytic site. In practice, it is easiest to measure the current and number of surface sites, then derive the average turnover frequency per site.52 ...... 28 Figure 2.4. HER activity of two theoretical catalysts. a. Representative linear sweep voltamograms (LSV) plotting, current density as a function of potential. b. Tafel plot. These

2 catalysts require the same overpotential for -10 mA/cm electrode, so they appear the same based on the total electrode activity metric used in this perspective. However, these catalysts likely have different HER mechanisms based on their different Tafel slopes, and both could be better or worse than the other depending on the application. The catalyst represented in blue would

2 perform better for low current devices (< 10 mA/cm electrode) while the catalyst represented in

2 red would be superior for high current density devices (> 10 mA/cm electrode)...... 30

xxiii

Figure 2.5. 2D representation of MoS2 catalyst electrochemically active surface area and projected geometric surface area. For crystalline 2H-MoS2, only edge sites, indicated by green dots, are active for HER. Basal plane sites, indicated by blue squares, are not active for HER. Rougher surfaces should have more available edge sites and thus higher catalytic activity per projected geometric surface area. It should also be noted that a catalyst geometry with basal planes oriented perpendicular to the surface would result in a more active catalyst but the less active geometry is displayed here for visualization purposes...... 33

Figure 2.6. a. TEM image of MoS2/MoO3-x shown structurally (right). This catalyst shows no degradation after 10,000 potential cycles despite the fact that MoO3 is unstable in acid and only

3 layers (2-3 nm) of MoS2 protects it. b. SEM image of MoS2/MoO3-x nanowires after 200°C thermal sulfidization. The inset shows a photograph of the sulfidized sample. Reproduced with permission.52 ...... 38

Figure 2.7. MoS2 double gyroid mesoporous structure. a. TEM image of the [311] and [211] projections. b. TEM image of the [110]. c,d,e. Models corresponding to simulated TEM images of the [211], [311], and [110] projections, respectively. f. 3D model of the double dyroid. Reproduced with permission.56 ...... 40

Figure 2.8. Synthesis of MoS2 nanoparticles with and without reduced graphene oxide (RGO) sheets. a. Schematic diagram of MoS2/RGO nanoparticle synthesis. b. SEM and inset TEM images of the MoS2/RGO catalyst. c. Schematic of large, free MoS2 nanoparticle synthesis. d.

61 SEM and inset TEM images of the MoS2 particles with no RGO. Reproduced with permission...... 41

Figure 2.9. Vertically aligned MoS2 sheets. TEM image (left) and schematic diagram (right) indicating edge sites and terrace sites. Reproduced with permission.57 ...... 42 4+ 2- Figure 2.10. a,b. Structure of [Mo3S4] cubane and [Mo3S13] nanoclusters, respectively. Blue: Mo atoms, yellow: S atoms, and red: O atoms (from water ligands). c. STM image of anodized

2- HOPG surface after drop-casting [Mo3S13] clusters. d. Atom-resolved STM image of a single

2- 53, 94 [Mo3S13] cluster are shown on the right. Reproduced with permission...... 44 Figure 2.11. Turn over frequencies of different molybdenum sulfide catalysts normalized to the number of surface Mo atoms...... 49 Figure 2.12. Linear sweep voltammograms demonstrating the total electrode activity different molybdenum sulfide catalysts. The potential required to reach a current density of -10

2 mA/cm electrode is reported to the right. Polycrystalline Pt is shown for comparison...... 51

xxiv

Figure 3.1. SEM images showing (a) the catalyst film surface morphology and (b) a thickness cross section of a segment of the catalyst film that delaminated from the substrate surface. .. 73 Figure 3.2. Transmission electron micrograph of an isolated particle deposited from a diluted molybdenum sulfide catalyst suspension...... 74 Figure 3.3. X-ray diffraction spectra of a blank glass substrate (red trace), the molybdenum sulfide catalyst film as deposited (blue trace), and the catalyst film after sulfidization for 30 minutes at 500°C in 10% H2/90% H2S (green trace). (a) Wide scan range, (b) high resolution scan. The peak at 13.9° in the sulfidized material is attributed to crystalline MoS2. The absence of a peak at this position for the as-deposited sample suggests that this material does not possess the layered structure of MoS2...... 74

Figure 3.4. Raman spectra of a MoS2 crystal standard (red trace), the as-deposited molybdenum sulfide catalyst film (blue trace), the molybdenum sulfide catalyst after 10 minutes of reductive potential cycling (green trace), the molybdenum sulfide catalyst after 4 hours of reductive potential cycling (purple trace), and a blank glassy carbon substrate (black trace). The catalyst

36-38 does not exhibit the characteristic peaks that are observed for crystalline MoS2...... 75 Figure 3.5. X-ray photoelectron spectra of the catalyst surface before testing, after one reductive potential cycle, and after four hours of continuous reductive cycling. The data are scaled to normalize the Mo 3d peak areas. Before electrochemical testing, the shape and binding energy of the S 2p peak are indicative of amorphous MoS3. The Mo 3d peaks show that the majority of

(A) the molybdenum occurs in chemical state Mo Sx. The peak positions and shapes begin to change after the initial reductive potential cycle, corresponding to the catalyst activation observed during electrochemical testing. These changes are accentuated after extended electrochemical testing. In the tested samples, the S 2p peak shape and binding energy more

(B) closely resemble MoS2, and the majority of the molybdenum occurs in chemical state Mo Sx, with binding energies near the Mo 4+ peaks observed in MoS2...... 76 Figure 3.6. X-ray photoelectron spectra of the Mo 3p/N 1s and Na 1s regions of the catalyst as deposited and after resting in H2O or 0.5 M H2SO4 with no applied potential. As deposited, the

+ catalyst displays small N 1s (a) and Na 1s (b) peaks, which are attributed to residual NH4 and Na+ ions from the catalyst synthesis. Elemental quantification suggests that the concentration of sodium ions in the catalyst material is less than 1% by mole. Quantification of the ammonium ion content is more difficult due to convolution with the Mo 3p peaks, but we conservatively

+ estimate that the NH4 concentration is no greater than 5% by mole. After resting in H2O or 0.5

M H2SO4 for one hour (c, d, e, and f), the N 1s and Na 1s lines disappear, which shows that

xxv these residual ions dissolve into the liquid electrolyte. We therefore assume that these ions do not contribute to the reductive currents observed during the catalyst activation or hydrogen evolution...... 77 Figure 3.7. X-ray photoelectron spectra of the S 2p and Mo 3d regions of an untested catalyst sample 3, 12, and 19 days after synthesis. The increase in size of the Mo 6+ features over time (b, d, and f) suggests that the surface of the molybdenum sulfide catalyst forms a native surface oxide upon exposure to air. A ~25% decrease in the relative intensity of the S 2p peak is also observed (a, c, and e). We attribute this change to the attenuation of the sulfur signal due to surface coverage by a thin oxide. Aside from these changes, the Mo 3d and S 2p line shapes and peak positions remain very close to their original values, which suggests that the chemical state of the molybdenum sulfide phase does not change upon exposure to air...... 79 Figure 3.8. X-ray photoelectron spectra of the Mo 3d and S 2p regions of the natural crystalline

MoS2 standard. Mo 3d region (a) and S 2p region (b)...... 80 Figure 3.9. Cyclic voltammograms indicate that the molybdenum sulfide catalyst is activated during the cathodic sweep of the first cycle (solid red line). Enhanced activity is observed on the anodic sweep of the first cycle (dotted red line) and in subsequent cycles (blue lines). .... 81 Figure 3.10. Electrochemical activity of the molybdenum sulfide catalyst. Polarization curves (a) show that the catalyst exhibits high activity for the HER. A Tafel plot (b) shows the electrochemical activity of the wet chemical synthesized amorphous molybdenum sulfide catalyst along with digitized data of HER measurements of several other state-of-the-art materials from the time of the study for comparison, including MoS2 nanotriangles in pH 0.24

24 56 H2SO4 (orange trace), MoO3/MoS2 core-shell nanowires in 0.5 M H2SO4 (blue trace),

29 electrodeposited amorphous MoS3 in pH 0 electrolyte (green trace), wet chemical synthesized

33 amorphous MoS3/multi-walled carbon nanotube composite in 1 M H2SO4 (purple trace), and

28 MoS2 on reduced graphene oxide in 0.5 M H2SO4 (red trace)...... 82 Figure 3.11. Electrochemical activity of four duplicate catalyst samples. The overpotentials required for 10 mA/cm2 current density range from 198 to 204 mV. The values on the Turn Over Frequency axis are estimates derived from the catalyst roughness factor of 90 (calculation details provided below)...... 82 Figure 3.12. X-ray photoelectron spectra of the S 2p and Mo 3d regions of the catalyst as deposited and after resting in H2O or 0.5 M H2SO4 with no applied potential. As deposited (a, b), the Mo 3d region includes Mo 6+ features, which we attribute to a native surface oxide. After resting in H2O for one hour (c, d), the Mo 6+ feature is substantially diminished, leaving only

xxvi

(A) the Mo Sx doublet corresponding to the untested molybdenum sulfide material, suggesting that the native surface oxide dissolves in water. The relative intensity of the S 2p peak also increases slightly because this signal is no longer attenuated by the thin surface oxide layer. A separate sample allowed to rest in 0.5 M H2SO4 for one hour (e, f) displays a similar S 2p feature and

(B) also lacks and evidence of Mo 6+ peaks. However, in this case, a more reduced Mo Sx chemical state is also observed, which suggests that exposure to H2SO4 could chemically reduce some of the molybdenum sulfide material...... 85 Figure 3.13. Electrochemical capacitance measurements for determination of the molybdenum sulfide catalyst surface area. Cyclic voltammograms (a) were taken in a potential range where no Faradaic processes were observed to measure the capacitive current from double layer charging. The capacitive current measured at 0.30 V vs. RHE was plotted as a function of scan rate (b) for the wet chemical synthesized amorphous molybdenum sulfide and the MoS2 flat standard. The ratio of the capacitive currents for the molybdenum sulfide catalyst and the flat standard was used to determine the relative roughness factor...... 87 Figure 3.14. Electrochemical activity of catalyst samples with 25%, 50%, 100%, 200%, and 400% typical mass loading. (a) Roughness factor vs. nominal mass loading. (b) Polarization curves demonstrating the catalytic activity of each sample. (c) Catalytic activity vs. nominal mass loading. (d) Catalytic activity vs. roughness factor...... 93 Figure 3.15. SEM micrographs of catalyst films with (a) 25%, (b) 50%, (c) 100%, (d) 200%, and (e) 400% typical mass loading...... 94 Figure 3.16. Electrochemical stability of the molybdenum sulfide catalyst. The overpotential required to reach a current density of 10 mA/cm2 increases by only 57 mV after 10,000 reductive potential cycles, indicating that the catalyst remains highly active...... 96 Figure 3.17. SEM micrographs of the catalyst film (a) before electrochemical testing and (b, c) after 1,000 cycles electrochemical stability testing...... 98 Figure 4.1. (a) Diagram and (b) photograph of the electrochemical cell setup used for the APXPS measurements...... 108 Figure 4.2. Synthesis scheme for the polymer electrolyte membrane assembly used to measure the APXPS spectra of the MoSx catalyst...... 108

Figure 4.3. Cyclic voltammogram of the PEM assembly with MoSx on the working electrode.

The scan rate was 40 mV/s and the H2O vapor pressure was 5 Torr. The voltage represents the total cell potential for this two electrode measurement...... 109

xxvii

Figure 4.4. APXPS spectra of the S 2p region of (a) the amorphous MoSx nanoparticles and (b) the crystalline MoS2 control sample under various experimental conditions...... 113

Figure 4.5. S 2p spectral deconvolution for the amorphous MoSx catalyst under selected experimental conditions. (a) Curve-fitted S 2p spectrum of amorphous MoSx nanoparticles under open circuit conditions, Spectrum I in Figure 4.4(a). (b) Curve-fitted S 2p spectrum of amorphous MoSx nanoparticles under hydrogen evolution conditions, Spectrum VII in Figure

4.4(a). The green components correspond to sulfur in the chemical state of MoS3. The red components correspond to sulfur in the chemical state of MoS2...... 115

Figure 4.6. Electrochemical activity of the MoSx catalyst and a bare glassy carbon substrate...... 116

Figure 4.7. Cross-section view of MoSx catalysts on a glassy carbon substrate. (a) Low magnification TEM image and (b) HRTEM image of the as-deposited MoSx sample showing it to be amorphous. (c) Annular dark field (ADF) STEM image near the surface of the as-deposited sample with the Pt deposition clear at the top. (d) Background subtracted EEL spectra of sulfur

L2, 3 edge from regions 1 – 6 indicated on (c) showing the homogeneous chemical state. (e) Low magnification TEM image of the sample after 30 min of electrochemical testing showing the formation of pores and (f) HRTEM image of a representative pore. The inset in (f) shows a higher magnification TEM image of the area indicated by the yellow box indicating the nucleation of crystalline domain around the pore. (g) ADF STEM image of the sample after electrochemical testing. (h) Background subtracted EEL spectra from regions 1 – 6 indicated on (g) showing the change in chemical states of sulfur at positions 2 – 4 and 6, as indicated by the variations in the fine structure of the sulfur L2, 3 edges. Note that the ADF STEM image in (g) is acquired from the exact same region shown in (f)...... 117

Figure 4.8. (a) HRTEM image of a 2H-MoS2 control sample showing the characteristic layered structure and (b) Background subtracted EEL spectra of the sulfur L2,3 edge from the sample displaying the characteristic peak of crystalline MoS2 at 174 eV...... 118

Figure 4.9. Aberration-corrected TEM images showing the structural changes in the MoSx sample after in situ activation in a H2 gas environment. (a) Low magnification image of the sample at room temperature under high vacuum. (b) HRTEM image of the region indicated by the yellow box in (a) showing it to be amorphous. HRTEM images of the samples after exposure to hydrogen for 30 min at room temperature (c) and at 150 °C (d). Note that the images in (c and d) were acquired from the regions near to the box in (a). Increasing fraction of MoS2 crystallites near the SiO2 interface is revealed...... 121

xxviii

Figure 4.10. Structural and chemical changes in the MoSx after H2 exposure for 30 min at 300

°C. (a) HRTEM image showing more extensive MoS2 formation and (b) ADF STEM image of the samples after H2 exposure at 300 °C after H2 exposure for 30 min. (c) Series of background subtracted EEL spectra of sulfur L2, 3 edge from points 1-3 indicated on (b)...... 122

Figure 4.11. Aberration corrected HRTEM images of amorphous MoSx under high vacuum condition (a) at room temperature and (b) after annealing at 300 ºC for 1 hour. There are no observable changes and the image appearance is characteristically. (c) EEL spectra of the sulfur

L2,3 edge region of the amorphous MoSx samples shown in (a) and (b) confirming no change in chemical state in the thin film...... 123 Figure 5.1. Photograph of the electrochemical cell used to test the photocathodes...... 136 Figure 5.2. Spectral irradiance of 1000 W xenon lamp compared to the AM1.5G standard solar spectrum.34 Irradiance values were calibrated and measured as described above. a. 1000 W Xe light source calibrated to 760 W m-2 at wavelengths shorter than 1030 nm and “one sun” AM1.5G standard. This illumination condition was used for all tests unless otherwise noted. b. 1000 W Xe light source calibrated to 1000 W m-2, 760 W m-2, and 760 W m-2 with a 700 nm long-pass filter resulting in an irradiance of 274 W m-2 in the red and infrared portions of the spectrum. These illumination conditions were used to collect the linear sweep voltammograms displayed in Figure 5.26...... 137

Figure 5.3. Gas-tight two-compartment electrochemical cell used for H2 quantification measurements...... 139 + + Figure 5.4. MoS2-n p Si device structure. a. Diagram of MoS2-n p Si device. b-d. Cross-

+ sectional transmission electron micrographs of MoS2-n p Si surface region collected with a FEI Tecnai TEM operated at 200 kV. These images illustrate that the surface of the device consists of MoS2, Mo metal, amorphous MoxSi, and Si layers with no evidence of an insulating SiO2 layer. The arrows in panel d indicate locations where HER-active MoS2 edge sites are likely exposed at the device surface...... 140 + Figure 5.5. Aberration-corrected cross-sectional TEM images of MoS2-n p Si device surface region. Aberration corrected imaging was performed using a FEI Titan 80-300 Environmental Transmission Electron Microscope equipped with a spherical aberration (Cs) corrector in the image-forming (objective) lens and operated at 80 kV to minimize any damage which might be induced by the incident electron beam. Under these imaging conditions, all Mo-containing layers appear darker than the Si. Otherwise these images reveal an identical structure to those

xxix shown in Figure 1, which also suggests that the beam did not alter the sample morphology during imaging at 200 kV...... 142 Figure 5.6. a. Cross-sectional scanning transmission electron microscopy (STEM) image of

+ MoS2-n p Si device surface region collected using a FEI Titan 80-300 Environmental TEM operated at 80 kV. The yellow box indicates the region shown in the elemental maps created using electron energy loss spectroscopy (EELS) in the other panels. For these measurements, the probe size was 0.9 nm, the elemental map pixel width was 0.3 nm, and the acquisition time of each EELS spectrum was 0.05 s. b. Overlaid elemental maps of Si (blue), Mo (green), and S (red). c. Elemental map of Si. d. Elemental map of Mo. e. Elemental map of S. These images provide further confirmation of the layered structure observed in the bright field TEM images in Figures 5.4 and 5.5...... 143 + Figure 5.7. Scanning electron microscopy (SEM) images of MoS2-n p Si device surface viewed from the top. Before testing, the sample appears flat and featureless, as expected. After the 100 hour stability test, some particles are observed on the sample surface. These likely arise from residual sulfuric acid or contaminant salts from the electrolyte. The larger features may correspond to oxidized portions of the electrode (i.e. MoO3)...... 144 + Figure 5.8. Schematic energy band diagram of the MoS2-n p Si photocathode in the dark (not to scale)...... 145 + Figure 5.9. X-ray photoelectron spectroscopy measurements of the MoS2-n p Si device before and after electrochemical stability measurement. Before testing, the structure contains Mo in the 4+ oxidation state corresponding to MoS2. The second doublet arises from the Mo metal and the amorphous MoxSi intermixing layer. The expected Mo binding energies for these species are too close to enable further deconvolution.39, 42, 43 The sulfur binding energy matches the expected value for MoS2. The silicon exists as elemental Si and MoxSi; these species cannot be resolved independently.39, 42, 43 After testing, the composition and chemical state remain very similar. Most of the molybdenum still exists as MoS2, Mo metal, or MoxSi, but some Mo 6+ corresponding to MoO3 is also detected. In addition to sulfur in the MoS2, some S corresponding

2- to SO4 groups from residual electrolyte is observed. There is no evidence of additional SiO2 formed after testing...... 146 + Figure 5.10. Activity of MoS2-n p Si device and controls and hydrogen quantification

+ measurements. a. Linear sweep voltammograms (LSVs) of the MoS2-n p Si device under

+ simulated solar illumination with two controls, the MoS2-n p Si device in the dark and an illuminated n+p Si structure with no molybdenum sulfide layer. b. Hydrogen quantification

xxx

+ measurements on MoS2-n p Si photocathode. These measurements show that this device evolves hydrogen with effectively 100% Faradaic efficiency within the experimental error. 148 + Figure 5.11. Optical absorption, reflection, and transmission spectra of the MoS2-n p Si photocathode...... 149 Figure 5.12. Optical absorption, reflection, and transmission spectra of control samples. a.

+ MoS2-SiO2 control b. Bare n p Si control...... 150 Figure 5.13. Incident photon-to-current conversion efficiency (IPCE) and absorbed photon-to-

+ current conversion efficiency (APCE) measurements of MoS2-n p Si. Low absorption in the photoactive Si likely limits the IPCE and the saturation photocurrent density...... 152 + Figure 5.14. Electrochemical stability measurements of MoS2-n p Si photocathode. a. Chronoamperometry measurement at E = 0 V vs. RHE. b. LSVs collected before (0 hr) and after (100 hr) the constant potential measurements. The device shows no loss in performance after 100 hours of testing...... 153 + -2 Figure 5.15. Stability of MoS2-n p Si photocathode irradiated with 760 W m white light. Each LSV was performed after changing the electrolyte and rinsing the electrode. After 25 hours, the photocurrent onset decreased by 0.03 V, but the onset increased to equal its original value after 100 hours of testing...... 154

Figure 5.16. Dark catalysis controls demonstrating the HER activity of the flat MoS2 coating

2- and the improvement in onset potential resulting from the addition of the [Mo3S13] HER catalyst...... 155 + Figure 5.17. Diagram of Mo3S13-MoS2-n p Si photocathode structure...... 157 + Figure 5.18. Scanning electron microscopy (SEM) images of Mo3S13-MoS2-n p Si device surface viewed from the top. Before testing, some regions of the surface are covered with a

2- rough film of the [Mo3S13] clusters, but the coating is not perfectly uniform. This suggests that

+ it may be possible to increase the activity of the Mo3S13-MoS2-n p Si even further by improving

2- the uniformity of the cluster coating. After electrochemistry, very few [Mo3S13] clusters remain...... 158 + + Figure 5.19. Linear sweep voltammograms (LSVs) of MoS2-n p Si, Mo3S13-MoS2-n p Si, and

+ 9 2- Pt-n p Si devices under simulated solar illumination. Incorporating the [Mo3S13] HER catalyst increases the onset potential by 80 mV in both the dark control and the photocathode...... 160 + Figure 5.20. a. Optical spectra of Mo3S13-MoS2-n p Si. Optical absorption, reflection, and

+ transmission spectra. a. Mo3S13-MoS2-n p Si photocathode b. Mo3S13-MoS2-SiO2 control. . 161

xxxi

Figure 5.21. Comparison of the fraction of incident light absorbed by all photocathodes and control samples tested in this study...... 162 Figure 5.22. a. Photon-to-current conversion efficiency measurements. b. Hydrogen product

+ quantification. Both a. and b. show data for Mo3S13-MoS2-n p Si photocathodes...... 163 + -2 Figure 5.23. Stability of Mo3S13-MoS2-n p Si photocathodes irradiated with 760 W m white light. a. Linear sweep voltammograms taken every 10 minutes for 4 hours. b. Current measured with the photocathode held at 0 V vs. RHE between LSV measurements. The photocurrent onset potential and saturation photocurrent density both decrease during the course of testing due to

2- the accumulation of bubbles on the electrode surface as well as the desorption of the [Mo3S13] clusters. The electrolyte was changed and the electrode was rinsed to remove bubbles before the final linear sweep voltammogram was collected after 4 hours of testing. The activity of this

+ electrode matches the performance observed in the MoS2-n p Si photocathodes, indicating that

2- most, if not all, of the [Mo3S13] clusters desorbed...... 164 + -2 Figure 5.24. Stability of Mo3S13-MoS2-n p Si photocathodes irradiated with 760 W m white light. a. Linear sweep voltammograms taken every 25 hours for 100 hours. b. Current measured with the photocathode held at 0 V vs. RHE between LSV measurements. As expected, the linear sweep voltammograms show that the activity decreases after 25 hours of illuminated testing. In subsequent measurements, the activity remains stable and matches the performance observed in

+ stability tests of the MoS2-n p Si photocathodes. The decrease in performance after 25 hours is

2- caused by the desorption of the [Mo3S13] clusters...... 165 + Figure 5.25. X-ray photoelectron spectra of Mo3S13-MoS2-n p Si before and after electrochemical testing...... 166 Figure 5.26. Linear sweep voltammograms under different illumination conditions...... 168

Figure 6.1. Schematic diagram of the fabrication procedure for the “Generation 1” Pt disk/SiO2- n+p Si photocathodes (not to scale)...... 179 Figure 6.2. Diagram of the photomask used in the fabrication of all the photocathodes in this chapter. Clear areas are shown in white, dark areas are shown in black, and dimensions are shown in blue. This pattern repeats across the mask to fill an 8 cm by 8 cm square...... 179 Figure 6.3. Schematic diagram of the fabrication procedure for the “Generation 2” Pt

+ disk/Si3N4-n p Si photocathodes (not to scale)...... 182

Figure 6.4. Schematic diagram of the fabrication procedure for the “Generation 3” Pt disk/SiO2- n+pp+ Si photocathodes (not to scale)...... 186

xxxii

Figure 6.5. a. Photograph of the electrochemical cell used to test the silicon electrodes. b. Teflon and glass electrode positioning apparatus...... 188 Figure 6.6. Spectral irradiance of the 150 W xenon lamp compared to the AM1.5G standard solar spectrum.58 ...... 190 Figure 6.7. Graphical depiction of photocathode performance metrics used in this work. ... 191 + Figure 6.8. SEM images of the “Generation 1” Pt disk/SiO2-n p Si photocathode front surface collected using a secondary electron detector. a, b. Low magnification images showing that the

Pt disk pattern is successfully formed across a large area. c. Single Pt disk surrounded by SiO2. d. Image taken at 45° sample tilt showing the edge of the Pt disk in contact with the SiO2. . 192

Figure 6.9. Auger electron spectroscopy elemental maps of the “Generation 1” Pt disk/SiO2- n+p Si photocathode front surface. a. Silicon. b. Oxygen. c. Carbon. d. Platinum...... 193 + Figure 6.10. Activity of the “Generation 1” Pt disk/SiO2-n p Si photocathode measured under “one sun” simulated solar illumination. a. Full scan range. b. Photocurrent onset region. The

+ + + activity of the MoS2-n p Si, Mo3S13-MoS2-n p Si, and Pt-n p Si devices are reproduced for comparison.3, 22 ...... 194 + Figure 6.11. SEM images of the “Generation 2” Pt disk/Si3N4-n p Si photocathode front surface collected using a secondary electron detector. a. Low magnification image showing that the Pt disk pattern is successfully formed across a large area. b. Single Pt disk surrounded by Si3N4. The dark ring around the catalyst disk may be polymer remaining from the lithography process. c. Image taken at 45° sample tilt of a single Pt disk surrounded by Si3N4. d. Image taken at 45° sample tilt showing the edge of the dark ring surrounding the Pt disk...... 195 + Figure 6.12. Activity of the “Generation 2” Pt disk/Si3N4-n p Si photocathode measured under “one sun” simulated solar illumination. a. Full scan range. b. Photocurrent onset region. The “Generation 1” photocathode’s activity is also plotted for comparison...... 196 + + Figure 6.13. SEM images of the “Generation 3” Pt disk/SiO2-n pp Si photocathode front surface collected using a secondary electron detector. a. Low magnification image showing that the Pt disk pattern is successfully formed across a large area. b. Single Pt disk. c. Image taken at 45° sample tilt of a single Pt disk. d. Image taken at 45° sample tilt showing the edge of the Pt disk and the dark ring surrounding the Pt disk...... 198 Figure 6.14. SEM images of a single Pt disk on the front surface of the “Generation 3” Pt

+ + disk/SiO2-n pp Si photocathode. a. Image collected using a secondary electron detector. b. Image of the same region collected using a concentric backscatter detector, which produces contrast primarily based on atomic mass...... 199

xxxiii

+ + Figure 6.15. Activity of the “Generation 3” Pt disk/SiO2-n pp Si photocathode measured under “one sun” simulated solar illumination. a. Full scan range. b. Photocurrent onset region. The activity of the “Generation 1” and “Generation 2” photocathodes is also plotted for comparison...... 200 + + Figure 6.16. Stability of the “Generation 3” Pt disk/SiO2-n pp Si photocathode measured under “one sun” simulated solar illumination. a. LSVs collected every 10 min. b. Chronoamperometry measurement at E = 0 V vs RHE between the LSVs...... 201 + + Figure 6.17. SEM images of the “Generation 3” Pt disk/SiO2-n pp Si photocathode front surface collected using a secondary electron detector after electrochemical stability testing. a. Low magnification image showing that most of the Pt disks remain intact. b. Single hole etched in the SiO2 where the Pt disk delaminated. c. Single Pt disk covered with dark spots likely indicating surface contamination of the catalyst. d. Edge of a single Pt disk...... 202 + + Figure 6.18. Proposed “Generation 4” Pt disk/Si3N4-n pp Si photocathode structure. a. Perspective view. b. Cross-section view. The diagrams are not to scale...... 203 Figure 6.19. Schematic diagram of the proposed fabrication procedure for the “Generation 4”

+ + Pt disk/Si3N4-n pp Si photocathodes (not to scale)...... 204

xxxiv

Chapter 1: Introduction

1.1 Motivation

Molecular hydrogen, H2, is a critical chemical reagent that is required for many industrial processes fundamental to sustaining human civilization. Globally, H2 is consumed at a rate of

1, 2 approximately 55 billion kg per year. In the Haber-Bosch process, H2 is reacted with N2 to produce ammonia, a primary component of fertilizer.3, 4 This process has been called the “detonator of the population explosion,” and without its existence, roughly half of the world’s population would starve.5 Hydrogen is also used to synthesize many other useful chemicals such as methanol and to refine petroleum into the most important transportation fuels, diesel and

4 gasoline. Currently, more than 95% of the H2 consumed worldwide is produced from fossil fuels, with the largest fraction coming from steam reforming of natural gas.2, 6-8 Unfortunately, these production routes are not sustainable due to the Earth’s limited supply of oil, gas, and

6, 9 coal. The consumption of fossil fuels to produce H2 also results in the emission of CO2, a greenhouse gas and a primary cause of global climate change, at a rate of up to 9 kg CO2 per kg

2, 6, 8, 10, 11 H2. If atmospheric CO2 levels continue to increase, the Earth will experience more frequent extreme weather events such as extended droughts and damaging storms, rises in global sea level, destruction of life-supporting ecosystems, and many more severe consequences for

12, 13 all the planet’s inhabitants. As a result, the development of alternative H2 production processes that are sustainable and environmentally friendly is a critical challenge facing humanity.6, 14

Hydrogen can also be used for energy storage to help transition our society to a clean and sustainable energy economy.4, 6, 9, 14 The majority of the energy used in the world comes from fossil fuels, which are limited in supply and emit harmful greenhouse gases when consumed, as explained above.15, 16 World energy demand is projected to rise by 35% – 40% over the next 25 years, so increasing the use of renewable energy sources such as solar, wind, tidal, hydroelectric, and geothermal energy is critical.15, 16 Solar energy is an especially promising energy source due to its very large potential. The current rate of global energy consumption is 18 TW, while the total power of the sunlight continuously striking the Earth is more than 120,000 TW.15-18 Even

1 if only a small fraction of this sunlight can realistically be harnessed, this resource represents more than enough energy for all of civilization. In order to be useful for human activities, solar energy must be captured and converted into other forms of energy.17 Electricity generation using photovoltaics or concentrating solar thermal power plants is a very promising strategy, but these technologies are hindered by sunlight’s intermittency and the lack of storage capacity in the electric grid.17, 19 Capturing the sun’s electromagnetic energy and storing it in the molecular bonds of a chemical fuel could address these issues. Hydrogen is an ideal candidate fuel for solar energy storage because it has a very high gravimetric energy density of 120 MJ/kg, nearly three times higher than gasoline.20 Hydrogen can also be consumed via combustion in an engine

4, 6, 9, 14 or oxidation in a fuel cell with no direct CO2 emissions and water as the only product.

The potential benefits of using H2 as an energy carrier are great, but an efficient, cost-effective, and scalable process for generating H2 using solar energy must be developed if this vision is to become a reality.

There are many possible methods for producing H2 using the energy from sunlight, including biological, thermal, chemical, and electrochemical processes.2, 7, 8, 14 Photoelectrochemical (PEC) water splitting is one particularly promising process in which the energy from sunlight is

21-24 used to split H2O into H2 and O2. Though this process has been studied for many years, there remain substantial technical and economic barriers that must be overcome in order to make PEC water splitting viable for widespread implementation. If these challenges can be addressed and performance targets can be met, PEC water splitting could provide a transformative solution for some of the most important problems facing humanity.23

1.2 Background

Several thorough reviews of solar water splitting have been published elsewhere.21, 22, 24-27 Here, we briefly describe the fundamentals of PEC water splitting relevant to this dissertation.

The overall chemical reaction that occurs in any PEC water splitting device is:

1 H O  H  O 2 2 2 2

2

This reaction requires a minimum energy input of ΔG = 237.1 kJ/mol at standard conditions, which corresponds to a thermodynamic voltage requirement of 1.23 V, though due to kinetic and transport limitations, the voltage required in practice is much larger.28-30

The overall reaction takes place in the form of two half reactions, the hydrogen evolution reaction (HER) at the cathode and the oxygen evolution reaction (OER) at the anode.24 These reactions are shown in Table 1.1 as they occur in acidic electrolyte. PEC water splitting systems combine the processes of solar light capture and conversion with water electrolysis to produce

H2 in a single device.

  Hydrogen Evolution Reaction (HER) 4H  4e  2H 2

  Oxygen Evolution Reaction (OER) 2H 2O  O2  4H  4e

Table 1.1. Water splitting half reactions as they occur in acid.

Landmark studies in PEC water splitting were conducted in the early 1970s by Fujishima and

31, 32 Honda, who described light-driven water splitting using a TiO2 photoelectrode. Following this work, many subsequent studies focused on utilizing a single material to perform all the steps in the PEC water splitting process. As the field has developed, many newer devices have incorporated multiple components to perform different functions.24 These increases in complexity have resulted in devices with greatly improved performance, as discussed further in Section 1.2.2.24, 33, 34

1.2.1 Water Splitting Device Components

The PEC water splitting process can be understood by considering the functions of the various components required in efficient water splitting devices. These components are illustrated in Figure 1.1.

3

Figure 1.1. Schematic energy band diagram of a PEC water splitting system illustrating key device components and processes. Adapted from Chen, et al.25

Semiconductor Light Absorbers

PEC water splitting devices must include one or more semiconductors to absorb solar photons, resulting in the creation of excited electron/hole pairs.21, 24 The electrons and holes are separated by an electric field inside the semiconductor which may arise either from a buried solid state junction or an electrochemical junction formed at the semiconductor/electrolyte interface. The electrons must eventually reach one surface, where they can perform the HER, while the holes must reach the other surface to perform the OER. PEC water splitting devices can incorporate semiconductor electrodes in several configurations. Devices can include a semiconductor photoanode and dark cathode, a dark anode and semiconductor photocathode, or a photoanode and a photocathode in a “dual absorber” or “tandem” device as shown in Figure 1.1.21, 24

4

The semiconductor light absorbers in a PEC device must supply electrons and holes with a sufficient potential difference to overcome the thermodynamic and kinetic requirements of the water splitting reactions. The photovoltage provided by an illuminated semiconductor is limited by the band gap of the material, which means that the band gap of the semiconductor limits the attainable efficiency of a water splitting device as discussed further in Section 1.2.2.35-37 Hence an appropriate band gap is a key requirement for this device component. The light absorbers must also possess sufficiently large charge carrier transport lengths to enable carrier collection without significant recombination. Other requirements that may be helpful in certain device configurations include appropriate band edge alignment with respect to the redox potential of interest and long term stability under operating conditions.21, 24 However, it may be possible to circumvent these requirements through the use of buried solid state junctions and corrosion protection layers.24, 38-47

Electrocatalysts

After the electrons and holes are transported to the semiconductor surfaces, they must perform the water splitting half reactions.21, 24 The HER and OER proceed through reactive intermediates adsorbed at the cathode and anode surfaces, respectively. The properties of the electrode surfaces influence the strength of the adsorbate/surface bonds, and therefore affect the reaction rates.48-55 Unfortunately, the rates of the HER and OER are very slow on many of the best semiconductor light absorber materials.56-58 This means that large overpotentials are required to drive the reactions on these surfaces, so the devices do not operate efficiently. As a result, both the anode and the cathode must include catalysts to speed up the reaction rates and reduce voltage losses due to large kinetic overpotentials.

Following the Sabatier Principle, surfaces which bind reactive intermediates neither too weakly nor too strongly are generally most active for the HER and the OER.48-50, 59 Therefore, a key requirement for electrocatalysts is an appropriate binding strength to reactive intermediates, which is determined by the electronic structure and atomic scale morphology of the catalyst surface.50 Catalysts must also be selective for the desired product and stable under operating conditions.51 Depending on the device configuration and amount of catalyst used, the catalyst material may also need to have high electrical conductivity, optical transparency, and low cost.24, 55, 60, 61

5

Electrolyte

The electrodes in a PEC water splitting device must be immersed in an aqueous electrolyte, which conducts ionic current between the anode and cathode.24 Typically this electrolyte is a strong acid or strong base solution.24 High or low pH solutions are preferred because empirical measurements have shown that HER and OER electrocatalysts are usually most active under these conditions. In addition, acids and bases have higher ionic conductivity than neutral salt solutions, and pH gradients may build up and impede further current flow when neutral electrolytes are employed.24, 62 In the studies presented here, strong acid electrolytes are used, typically 0.5 M sulfuric acid solutions unless otherwise specified.

Optional Device Components

In addition to the semiconductor light absorbers, electrocatalysts, and electrolyte, which are necessary components of all high efficiency PEC water splitting systems, some devices employ additional components for a variety of purposes.24 For example, some devices use metal or polymer layers as adhesives to promote strong mechanical and electrical contact between device components.63, 64 Other devices include materials that create electronic junctions with the light absorbers, such as metal layers to form a Schottky contact or wide band gap semiconductors as electron or hole blocking layers.65, 66 Still further devices utilize passivation layers to reduce surface-mediated charge carrier recombination at semiconductor interfaces.67

One optional device component that is of particular interest for this dissertation is the corrosion protection layer. As discussed above, many semiconductors are susceptible to corrosion or oxidation in aqueous electrolytes.38, 68-72 To prevent these processes from occurring, some devices include a layer of protecting material that shields the semiconductor from coming into physical contact with the electrolyte.24, 38, 40-42 These corrosion protection layers must be conformal, stable, conductive, and in certain device configurations, optically transparent.73

1.2.2 Efficiency

The overall thermodynamic efficiency of the solar water splitting process is called the solar to hydrogen (STH) efficiency.25 This quantity is defined as the amount of chemical energy stored in the form of H2 divided by the amount of sunlight energy striking the device as shown in the following equation:25

6

(mol H / s) (237 kJ / mol) 2 STH   2 2  P (kW / m ) Area(m )  total  AM1.5G

Many studies in the field of PEC water splitting focus on individual device components or unoptimized systems which are not capable of splitting water without an externally applied bias. In these cases, it is not possible to calculate a STH efficiency.25 However, given that a primary goal of this field of research is to develop a device that can split water efficiently, it is crucial to determine whether any material considered for PEC applications could eventually enable the construction of a device with high STH efficiency.

The STH efficiency in PEC water splitting devices is subject to a current-voltage tradeoff that is related to, though in some ways different from, the Shockley-Queisser limit on the efficiency of photovoltaic cells.24, 36, 74 The photovoltage produced by a semiconductor is always less than the material’s band gap, so choosing a semiconductor with a larger band gap offers the potential to obtain a larger photovoltage. However, a larger band gap also decreases the fraction of solar photons with sufficient energy to excite electron/hole pairs, which reduces the photocurrent. In PEC water splitting devices, the semiconductor must supply a large enough photovoltage to overcome the thermodynamic and kinetic barriers to drive the water splitting reactions, as explained previously. For a single absorber PEC water splitting device, if the semiconductor band gap is too small, the photovoltage limits the device efficiency.36 In contrast, if the semiconductor band gap is too large, low photocurrent may limit the device efficiency.36 Unlike in photovoltaics, any additional photovoltage beyond the requirements for water splitting is wasted as heat.24, 36 The same concepts apply to PEC water splitting devices that incorporate multiple semiconductors.

Due to these tradeoffs, the semiconductor band gap is the most important determinant of the maximum attainable efficiency for a PEC water splitting device.24, 36 Previous studies have used numerical calculations to model the maximum attainable efficiency of PEC water splitting devices using realistic assumptions based on the performance of current semiconductor and catalyst materials.35-37 One relevant study showed that the maximum attainable efficiency for a single absorber device is approximately 11.2%, corresponding to a semiconductor band gap of 2.26 eV.36 Stacked dual absorber devices can use the energy in the solar spectrum more efficiently, and as a result have a predicted maximum attainable efficiency of approximately 22.8% for semiconductors with band gaps of 1.23 eV and 1.84 eV as shown in Figure 1.2.36

7

These insights are directly relevant to the design of PEC water splitting devices. Given that tandem devices have the potential to achieve much higher STH efficiencies than single absorber devices, materials relevant to this device configuration will be the focus of this dissertation.

Figure 1.2. Predicted maximum attainable STH efficiency as a function of top absorber and bottom absorber band gap for a stacked dual absorber water splitting device based on numerical modeling calculations. The dashed line represents the 1.12 eV band gap of silicon, indicating that water splitting devices incorporating silicon as the bottom absorber have the potential to achieve high STH efficiencies. Reproduced with permission from John Wiley and Sons.36

To date, no PEC water splitting device has ever achieved the maximum efficiency predicted by these numerical modeling calculations. The most efficient devices have relied on III-V absorber materials and precious metal catalysts, as shown by Turner and coworkers, who demonstrated

GaInP2/GaAs tandem water splitting devices with either one or two buried p-n junctions and STH efficiencies of 12.4% and 16.5%, respectively.33, 34 Devices based on other high quality semiconductor materials have also been successful. Results include 10.5% STH efficiency with a CIGS absorber75 as well as 6.2% and 7.8% STH efficiencies with triple junction amorphous Si absorbers.76-78 Devices incorporating other classes of semiconductors such as metal oxides have also been demonstrated, though the STH efficiencies achieved have been substantially

8 lower than those possible with high-quality materials due to the inferior charge carrier mobilities and increased recombination often found in low-quality semiconductors.79, 80 These trends suggest that utilizing semiconductors with excellent electronic properties is critical for high performance PEC devices.

1.2.3 Hydrogen Cost

The cost of H2 produced via PEC water splitting is another critical consideration. While no commercial PEC hydrogen production facilities currently exist, it is important to determine whether this process could eventually be economically competitive. To address this question, a team of PEC researchers and chemical plant design experts worked together to perform a

23 technical and economic analysis of hypothetical centralized solar H2 generation facilities. The four reactor designs considered are shown in Figure 1.3.

This study revealed that the levelized cost of hydrogen produced through PEC water splitting could be between $1.60 - $10.40 per kg H2, competitive with other H2 generation technologies

23, 81 and fossil fuels. However, the predicted H2 costs were highly sensitive to the assumptions for the device efficiency, cost, and lifetime.23 These results show that PEC water splitting has the potential to be successful on a commercial scale, but only if high efficiency, stable devices composed of scalable and inexpensive materials can be created.

9

Figure 1.3. Schematic diagrams of the four reactor types considered in the technoeconomic analysis conducted to estimate the cost of H2 produced via hypothetical large-scale PEC water splitting facilities. Drawings not to scale. Reproduced by permission of The Royal Society of Chemistry.23

10

1.2.4 Research Challenges

There are many research challenges that remain to be addressed in the field of PEC water splitting. As described in the previous section, for PEC water splitting devices to be viable for widespread implementation, they must be efficient, durable, scalable, and economical. No water splitting system created so far has fulfilled all of these criteria. The following three research challenges represent some of the greatest remaining barriers to the development of PEC systems that meet all these requirements.

Catalysis

The development of active, stable, selective, and earth abundant electrocatalysts for the HER and OER remains a significant challenge.24, 51, 55 Highly active catalysts are required to achieve the highest overall device efficiencies and to provide flexibility in system design. In acidic electrolyte, the most active known catalyst for the HER is platinum, while the best OER catalyst is iridium oxide.24, 51, 55, 82 Both of these materials are rare and expensive, so many efforts have been devoted to replacing them with earth abundant materials. Molybdenum sulfide nanomaterials represent a promising class of alternative HER catalysts.51, 55 These materials will be discussed further in Chapters 2, 3, and 4.

Semiconductor Corrosion Protection

The design of improved semiconductor corrosion protection layers is another critical challenge.38, 41, 42, 72 As discussed previously, many absorber materials have been utilized in PEC water splitting devices, the most successful of which have relied on high-quality semiconductors including silicon and III-V compounds. However, these materials are unstable in the strong acid and base electrolytes typically used in PEC water splitting cells. Some researchers have employed corrosion protection layers to address these challenges, but substantial work remains to develop new corrosion protection strategies that can prevent the degradation of semiconductor light absorbers over long-term operation. The use of thin, conformal films of crystalline molybdenum disulfide as corrosion protection layers will be discussed in Chapter 5.

Device Integration and Interface Engineering

Choosing the best possible catalysts and semiconductors is a necessary but insufficient condition for achieving high performance PEC water splitting. Successfully integrating all the

11 required components into a single device is a substantial challenge.24, 26, 44 The interfaces between device components must be designed carefully so that each component performs optimally without harming the performance of others. We employ various strategies to engineer interfaces in silicon photocathodes to maximize device performance in Chapters 5 and 6.

1.3 Scope

This dissertation focuses on the development of materials and devices for H2 production via PEC water splitting. Following the insights gained by examining the efficiency and economics of this process, we study materials that are scalable and have the potential to yield high performance water splitting devices. The remaining chapters in this thesis will cover the following material:

Chapter 2: Molybdenum sulfide materials as catalysts for the HER are analyzed. First, HER reaction mechanisms and the material properties that determine catalytic activity are considered. Best practices for experimentally determining the activity of HER catalysts are described. The most active molybdenum sulfide HER catalysts reported in the literature are compared, and these results are used to identify design principles for creating highly active catalysts.

Chapter 3: We describe a new room temperature wet chemical technique for synthesizing a highly active amorphous molybdenum sulfide HER catalyst. Following the recommended experimental procedures described in Chapter 2, we employ a variety of ex situ physical, chemical, and electrochemical techniques to characterize the catalyst and identify the physical and chemical properties that give rise to its high activity.

Chapter 4: The characterization methods employed in Chapter 3 provide many useful insights about the amorphous molybdenum sulfide HER catalyst, but these ex situ techniques cannot measure changes in the catalyst during the reaction. In this chapter, we use in situ ambient pressure x-ray photoelectron spectroscopy and environmental transmission electron microscopy to study the amorphous MoSx catalyst under operating conditions. The results reveal significant changes in the catalyst’s chemical state and physical structure after catalysis.

Chapter 5: Silicon is a promising candidate small band gap semiconductor for application in a dual absorber solar water splitting device, but its poor catalytic activity for the HER and susceptibility to corrosion in aqueous electrolyte are significant challenges. This chapter focuses

12 on integrating molybdenum sulfide nanomaterials with silicon light absorbers to make full photocathodes. We use multiple molybdenum sulfide nanomaterials designed to provide both catalytic activity and corrosion resistance to silicon light absorbers. The resulting devices show excellent activity and stability. We identify the atomic-scale features of the semiconductor/catalyst/electrolyte interfaces that result in this high performance.

Chapter 6: The molybdenum sulfide/silicon photocathodes described in Chapter 5 possess excellent catalytic activity and corrosion resistance, but achieving the best possible photovoltage from semiconductor absorbers is a remaining challenge. Here, we focus on engineering semiconductor interfaces to fabricate silicon photocathodes with very high performance. This work provides a foundation for the development of complete monolithic dual absorber PEC water splitting devices that use silicon as the small band gap absorber.

Chapter 7: A summary of the most important findings from the previous chapters and future research directions are discussed.

1.4 Copyright

Portions of this chapter reprinted with permission from:

J.D. Benck, T.R. Hellstern, J. Kibsgaard, P. Chakthranont, and T.F. Jaramillo. "Catalyzing the Hydrogen Evolution Reaction (HER) with Molybdenum Sulfide Nanomaterials." ACS Catalysis, 2014: 3957-3971. http://dx.doi.org/10.1021/cs500923c

Copyright 2014 American Chemical Society.

1.5 References

1. S.P.S. Badwal, S. Giddey, and C. Munnings. "Hydrogen production via solid electrolytic routes." Wiley Interdisciplinary Reviews: Energy and Environment, 2013. 2 (5): 473-487. http://dx.doi.org/10.1002/wene.50 2. C.M. Kalamaras and A.M. Efstathiou. "Hydrogen Production Technologies: Current State and Future Developments." Conference Papers in Energy, 2013. 2013: 9. http://dx.doi.org/10.1155/2013/690627 3. M. Appl. "Ammonia, 2. Production Processes," in "Ullmann's Encyclopedia of Industrial Chemistry." 2000, Wiley-VCH Verlag GmbH & Co. KGaA. http://dx.doi.org/10.1002/14356007.o02_o11

13

4. P. Häussinger, R. Lohmüller, and A.M. Watson. "Hydrogen, 6. Uses," in "Ullmann's Encyclopedia of Industrial Chemistry." 2000, Wiley-VCH Verlag GmbH & Co. KGaA. http://dx.doi.org/10.1002/14356007.o13_o07 5. V. Smil. "Detonator of the population explosion." Nature, 1999. 400 (6743): 415-415. http://dx.doi.org/10.1038/22672 6. M. Balat. "Potential importance of hydrogen as a future solution to environmental and transportation problems." International Journal of Hydrogen Energy, 2008. 33 (15): 4013-4029. http://dx.doi.org/10.1016/j.ijhydene.2008.05.047 7. P. Häussinger, R. Lohmüller, and A.M. Watson. "Hydrogen, 2. Production," in "Ullmann's Encyclopedia of Industrial Chemistry." 2000, Wiley-VCH Verlag GmbH & Co. KGaA. http://dx.doi.org/10.1002/14356007.o13_o03 8. J.D. Holladay, J. Hu, D.L. King, and Y. Wang. "An overview of hydrogen production technologies." Catalysis Today, 2009. 139 (4): 244-260. http://dx.doi.org/10.1016/j.cattod.2008.08.039 9. A. Midilli and I. Dincer. "Key strategies of hydrogen energy systems for sustainability." International Journal of Hydrogen Energy, 2007. 32 (5): 511-524. http://dx.doi.org/10.1016/j.ijhydene.2006.06.050 10. B. Gaudernack and S. Lynum. "Hydrogen from natural gas without release of CO2 to the atmosphere." International Journal of Hydrogen Energy, 1998. 23 (12): 1087- 1093. http://dx.doi.org/10.1016/S0360-3199(98)00004-4 11. G. Collodi and F. Wheeler. "Hydrogen Production via Steam Reforming with CO2 Capture." 2010. www.aidic.it/CISAP4/webpapers/7Collodi.pdf 12. Intergovernmental Panel on Climate Change. "Climate change 2014 synthesis report: approved summary for policymakers." 2014. http://www.ipcc.ch/report/ar5/syr/ 13. NASA. "The current and future consequences of global change." 2015. http://climate.nasa.gov/effects/ 14. N. Armaroli and V. Balzani. "The Hydrogen Issue." ChemSusChem, 2011. 4 (1): 21- 36. http://dx.doi.org/10.1002/cssc.201000182 15. BP. "Energy Outlook 2035." 2015. http://www.bp.com/en/global/corporate/about- bp/energy-economics/energy-outlook.html 16. International Energy Agency. "Key World Energy Statistics 2014." 2014. http://www.iea.org/publications/freepublications/publication/key-world-energy- statistics-2014.html 17. N. Armaroli and V. Balzani. "Energy for a sustainable world : from the oil age to a sun-powered future." 2011, Weinheim, Germany: Wiley-VCH. 18. Oak Ridge National Laboratory Basic Energy Sciences Advisory Committee. "Basic Research Needs to Assure a Secure Energy Future." 2003. http://science.energy.gov/~/media/bes/besac/pdf/Basic_research_needs_to_assure_a_s ecure_energy_future_feb_2003.pdf 19. G. Hegar. "Electrical Grids: a Primer." 2015. http://thetexaseconomy.org/natural- resources/articles/article.php?name=electricalGrid 20. U.S. Department of Energy Office of Energy Efficiency and Renewable Energy. "Lower and Higher Heating Values of Fuels." 2015. http://hydrogen.pnl.gov/tools/lower-and-higher-heating-values-fuels 21. T. Bak, J. Nowotny, M. Rekas, and C.C. Sorrell. "Photo-electrochemical hydrogen generation from water using solar energy. Materials-related aspects." International Journal of Hydrogen Energy, 2002. 27 (10): 991-1022. http://dx.doi.org/10.1016/S0360-3199(02)00022-8 22. Z. Chen, T.F. Jaramillo, T.G. Deutsch, A. Kleiman-Shwarsctein, A.J. Forman, N. Gaillard, R. Garland, K. Takanabe, C. Heske, M. Sunkara, E.W. McFarland, K.

14

Domen, E.L. Miller, J.A. Turner, and H.N. Dinh. "Accelerating materials development for photoelectrochemical hydrogen production: Standards for methods, definitions, and reporting protocols." Journal of Materials Research, 2010. 25 (01): 3- 16. http://dx.doi.org/10.1557/JMR.2010.0020 23. B.A. Pinaud, J.D. Benck, L.C. Seitz, A.J. Forman, Z. Chen, T.G. Deutsch, B.D. James, K.N. Baum, G.N. Baum, S. Ardo, H. Wang, E. Miller, and T.F. Jaramillo. "Technical and economic feasibility of centralized facilities for solar hydrogen production via photocatalysis and photoelectrochemistry." Energy & Environmental Science, 2013. 6 (7): 1983-2002. http://dx.doi.org/10.1039/C3EE40831K 24. M.G. Walter, E.L. Warren, J.R. McKone, S.W. Boettcher, Q. Mi, E.A. Santori, and N.S. Lewis. "Solar Water Splitting Cells." Chemical Reviews, 2010. 110 (11): 6446- 6473. http://dx.doi.org/10.1021/cr1002326 25. Z. Chen, H.N. Dinh, and E. Miller. "Photoelectrochemical water splitting: standards, experimental methods, and protocols." 2013, New York, NY: Springer. 26. S. Fiechter, P. Bogdanoff, T. Bak, and J. Nowotny. "Basic concepts of photoelectrochemical solar energy conversion systems." Advances in Applied Ceramics, 2012. 111 (1-2): 39-43. http://dx.doi.org/10.1179/1743676111Y.0000000041 27. Y. Tachibana, L. Vayssieres, and J.R. Durrant. "Artificial photosynthesis for solar water-splitting." Nature Photonics, 2012. 6 (8): 511-518. http://dx.doi.org/10.1038/nphoton.2012.175 28. M. Pagliaro and A.G. Konstandopoulos. "Solar Hydrogen: Fuel of the Future." 2012: Royal Society of Chemistry. 29. M.G. Walter, E.L. Warren, J.R. McKone, S.W. Boettcher, Q.X. Mi, E.A. Santori, and N.S. Lewis. "Solar Water Splitting Cells." Chemical Reviews, 2010. 110 (11): 6446- 6473. http://dx.doi.org/10.1021/cr1002326 30. M. Doyle, G. Rajendran, W. Vielstich, H. Gasteiger, and A. Lamm. "Handbook of Fuel Cells: Fundamentals, Technology and Applications." Vol. 3. 2003, Hoboken, NJ: Wiley. 31. A. Fujishima and K. Honda. "Electrochemical Evidence for the Mechanism of the Primary Stage of Photosynthesis." Bulletin of the Chemical Society of Japan, 1971. 44 (4): 1148-1150. http://dx.doi.org/10.1246/bcsj.44.1148 32. A. Fujishima and K. Honda. "Electrochemical Photolysis of Water at a Semiconductor Electrode." Nature, 1972. 238 (5358): 37-+. http://dx.doi.org/10.1038/238037a0 33. O. Khaselev, A. Bansal, and J. Turner. "High-efficiency integrated multijunction photovoltaic/electrolysis systems for hydrogen production." International Journal of Hydrogen Energy, 2001. 26 (2): 127-132. http://dx.doi.org/10.1016/S0360- 3199(00)00039-2 34. O. Khaselev and J. Turner. "A Monolithic Photovoltaic-Photoelectrochemical Device for Hydrogen Production via Water Splitting." Science, 1998. 280 (5362): 425-427. http://dx.doi.org/10.1126/science.280.5362.425 35. S. Hu, C.X. Xiang, S. Haussener, A.D. Berger, and N.S. Lewis. "An analysis of the optimal band gaps of light absorbers in integrated tandem photoelectrochemical water-splitting systems." Energy & Environmental Science, 2013. 6 (10): 2984-2993. http://dx.doi.org/10.1039/c3ee40453f 36. L.C. Seitz, Z. Chen, A.J. Forman, B.A. Pinaud, J.D. Benck, and T.F. Jaramillo. "Modeling practical performance limits of photoelectrochemical water splitting based on the current state of materials research." ChemSusChem, 2014. 7 (5): 1372-1385. http://dx.doi.org/10.1002/cssc.201301030

15

37. M.F. Weber and M.J. Dignam. "Efficiency of Splitting Water with Semiconducting Photoelectrodes." Journal of the Electrochemical Society, 1984. 131 (6): 1258-1265. http://dx.doi.org/10.1149/1.2115797 38. S. Hu, M.R. Shaner, J.A. Beardslee, M. Lichterman, B.S. Brunschwig, and N.S. Lewis. "Amorphous TiO2 coatings stabilize Si, GaAs, and GaP photoanodes for efficient water oxidation." Science, 2014. 344 (6187): 1005-1009. http://dx.doi.org/10.1126/science.1251428 39. N. Strandwitz, D. Comstock, R.L. Grimm, A.C. Nichols-Nielander, J.W. Elam, and N.S. Lewis. "Photoelectrochemical Behavior of n-type Si(100) Electrodes Coated with Thin Films of Manganese Oxide Grown by Atomic Layer Deposition." Journal of Physical Chemistry C, 2013. http://dx.doi.org/10.1021/jp311207x 40. A.B. Laursen, T. Pedersen, P. Malacrida, B. Seger, O. Hansen, P.C. Vesborg, and I. Chorkendorff. "MoS2: an integrated protective and active layer on n+p-Si for solar H2 evolution." Physical Chemistry Chemical Physics, 2013. 15 (46): 20000-20004. http://dx.doi.org/10.1039/c3cp52890a 41. B. Seger, T. Pedersen, A.B. Laursen, P.C. Vesborg, O. Hansen, and I. Chorkendorff. "Using TiO2 as a conductive protective layer for photocathodic H2 evolution." Journal of the American Chemical Society, 2013. 135 (3): 1057-1064. http://dx.doi.org/10.1021/ja309523t 42. B. Seger, D.S. Tilley, T. Pedersen, P.C.K. Vesborg, O. Hansen, M. Grätzel, and I. Chorkendorff. "Silicon protected with atomic layer deposited TiO2: durability studies of photocathodic H2 evolution." RSC Advances, 2013. 3 (48): 25902. http://dx.doi.org/10.1039/c3ra45966g 43. Y.W. Chen, J.D. Prange, S. Duhnen, Y. Park, M. Gunji, C.E. Chidsey, and P.C. McIntyre. "Atomic layer-deposited tunnel oxide stabilizes silicon photoanodes for water oxidation." Nature Materials, 2011. 10 (7): 539-544. http://dx.doi.org/10.1038/nmat3047 44. C.R. Cox, M.T. Winkler, J.J.H. Pijpers, T. Buonassisi, and D.G. Nocera. "Interfaces between water splitting catalysts and buried silicon junctions." Energy & Environmental Science, 2013. 6 (2): 532-538. http://dx.doi.org/10.1039/C2EE23932A 45. E.L. Warren, J.R. McKone, H.A. Atwater, H.B. Gray, and N.S. Lewis. "Hydrogen- evolution characteristics of Ni-Mo-coated, radial junction, n(+)p-silicon microwire array photocathodes." Energy & Environmental Science, 2012. 5 (11): 9653-9661. http://dx.doi.org/10.1039/c2ee23192a 46. S.W. Boettcher, J.M. Spurgeon, M.C. Putnam, E.L. Warren, D.B. Turner-Evans, M.D. Kelzenberg, J.R. Maiolo, H.A. Atwater, and N.S. Lewis. "Energy-Conversion Properties of Vapor-Liquid-Solid-Grown Silicon Wire-Array Photocathodes." Science, 2010. 327 (5962): 185-187. http://dx.doi.org/10.1126/science.1180783 47. S.W. Boettcher, E.L. Warren, M.C. Putnam, E.A. Santori, D. Turner-Evans, M.D. Kelzenberg, M.G. Walter, J.R. McKone, B.S. Brunschwig, H.A. Atwater, and N.S. Lewis. "Photoelectrochemical hydrogen evolution using Si microwire arrays." Journal of the American Chemical Society, 2011. 133 (5): 1216-1219. http://dx.doi.org/10.1021/ja108801m 48. J.K. Nørskov, T. Bligaard, A. Logadottir, J. Kitchin, J. Chen, S. Pandelov, and U. Stimming. "Trends in the exchange current for hydrogen evolution." Journal of the Electrochemical Society, 2005. 152 (3): J23-J26. http://dx.doi.org/10.1149/1.1856988 49. J.K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J.R. Kitchin, T. Bligaard, and H. Jónsson. "Origin of the Overpotential for Oxygen Reduction at a Fuel-Cell Cathode." Journal of Physical Chemistry B, 2004. 108 (46): 17886-17892. http://dx.doi.org/10.1021/jp047349j

16

50. J.K. Nørskov, F. Studt, F. Abild-Pedersen, and T. Bligaard. "Fundamental concepts in heterogeneous catalysis." 2014, Hoboken, NJ: Wiley. 51. J.D. Benck, T.R. Hellstern, J. Kibsgaard, P. Chakthranont, and T.F. Jaramillo. "Catalyzing the Hydrogen Evolution Reaction (HER) with Molybdenum Sulfide Nanomaterials." ACS Catalysis, 2014. 4 (11): 3957-3971. http://dx.doi.org/10.1021/cs500923c 52. J.O.M. Bockris and E.C. Potter. "The Mechanism of the Cathodic Hydrogen Evolution Reaction." Journal of the Electrochemical Society, 1952. 99 (4): 169-186. http://dx.doi.org/10.1149/1.2779692 53. L.A. Kibler. "Hydrogen electrocatalysis." ChemPhysChem, 2006. 7 (5): 985-991. http://dx.doi.org/10.1002/cphc.200500646 54. N.M. Markovic. "Electrocatalysis: Interfacing electrochemistry." Nature Materials, 2013. 12 (2): 101-102. http://dx.doi.org/10.1038/nmat3554 55. P.C.K. Vesborg, B. Seger, and I. Chorkendorff. "Recent Development in Hydrogen Evolution Reaction Catalysts and Their Practical Implementation." Journal of Physical Chemistry Letters, 2015. 6 (6): 951-957. http://dx.doi.org/10.1021/acs.jpclett.5b00306 56. K.D. Brummer. "Hydrogen Evolution on Single Crystal GaAs Electrodes." Journal of The Electrochemical Society, 1967. 114 (12): 1274-1281. http://dx.doi.org/10.1149/1.2426478 57. H. Gerischer, N. Muller, and O. Haas. "On the Mechanism of Hydrogen Evolution at GaAs Electrodes." Journal of Electroanalytical Chemistry, 1981. 119 (1): 41-48. http://dx.doi.org/10.1016/0368-1874(81)87043-8 58. J.R. McKone, E.L. Warren, M.J. Bierman, S.W. Boettcher, B.S. Brunschwig, N.S. Lewis, and H.B. Gray. "Evaluation of Pt, Ni, and Ni–Mo electrocatalysts for hydrogen evolution on crystalline Si electrodes." Energy & Environmental Science, 2011. 4 (9): 3573. http://dx.doi.org/10.1039/c1ee01488a 59. A.B. Laursen, A.S. Varela, F. Dionigi, H. Fanchiu, C. Miller, O.L. Trinhammer, J. Rossmeisl, and S. Dahl. "Electrochemical Hydrogen Evolution: Sabatier’s Principle and the Volcano Plot." Journal of Chemical Education, 2012. 89 (12): 1595-1599. http://dx.doi.org/10.1021/ed200818t 60. L. Trotochaud, T.J. Mills, and S.W. Boettcher. "An Optocatalytic Model for Semiconductor–Catalyst Water-Splitting Photoelectrodes Based on In Situ Optical Measurements on Operational Catalysts." Journal of Physical Chemistry Letters, 2013. 4 (6): 931-935. http://dx.doi.org/10.1021/jz4002604 61. V. Viswanathan, K.L. Pickrahn, A.C. Luntz, S.F. Bent, and J.K. Nørskov. "Nanoscale Limitations in Metal Oxide Electrocatalysts for Oxygen Evolution." Nano Letters, 2014. 14 (10): 5853-5857. http://dx.doi.org/10.1021/nl502775u 62. J. Jin, K. Walczak, M.R. Singh, C. Karp, N.S. Lewis, and C. Xiang. "An experimental and modeling/simulation-based evaluation of the efficiency and operational performance characteristics of an integrated, membrane-free, neutral pH solar-driven water-splitting system." Energy & Environmental Science, 2014. 7 (10): 3371-3380. http://dx.doi.org/10.1039/C4EE01824A 63. A. Paracchino, V. Laporte, K. Sivula, M. Gratzel, and E. Thimsen. "Highly active oxide photocathode for photoelectrochemical water reduction." Nature Materials, 2011. 10 (6): 456-461. http://dx.doi.org/10.1038/nmat3017 64. D.V. Esposito, I. Levin, T.P. Moffat, and A.A. Talin. "H2 evolution at Si-based metal- insulator-semiconductor photoelectrodes enhanced by inversion channel charge collection and H spillover." Nature Materials, 2013. 12 (6): 562-568. http://dx.doi.org/10.1038/nmat3626

17

65. S. Choudhary, S. Upadhyay, P. Kumar, N. Singh, V.R. Satsangi, R. Shrivastav, and S. Dass. "Nanostructured bilayered thin films in photoelectrochemical water splitting – A review." International Journal of Hydrogen Energy, 2012. 37 (24): 18713-18730. http://dx.doi.org/10.1016/j.ijhydene.2012.10.028 66. C. Hu, K. Chu, Y. Zhao, and W.Y. Teoh. "Efficient Photoelectrochemical Water Splitting over Anodized p-Type NiO Porous Films." ACS Applied Materials & Interfaces, 2014. 6 (21): 18558-18568. http://dx.doi.org/10.1021/am507138b 67. F. Le Formal, N. Tetreault, M. Cornuz, T. Moehl, M. Gratzel, and K. Sivula. "Passivating surface states on water splitting hematite photoanodes with alumina overlayers." Chemical Science, 2011. 2 (4): 737-743. http://dx.doi.org/10.1039/C0SC00578A 68. P. Allongue and S. Blonkowski. "Corrosion of III-V Compounds - A Comparative Study of GaAs and InP. 2. Reaction Scheme and Influence of Surface Properties." Journal of Electroanalytical Chemistry, 1991. 317 (1-2): 77-99. http://dx.doi.org/10.1016/0022-0728(91)85004-9 69. D.E. Aspnes and A.A. Studna. "Stability of (100)GaAs surfaces in aqueous solutions." Applied Physics Letters, 1985. 46 (11): 1071-1073. http://dx.doi.org/10.1063/1.95764 70. O. Khaselev and J.A. Turner. "Electrochemical stability of p-GaInP2 in aqueous electrolytes toward photoelectrochemical water splitting." Journal of the Electrochemical Society, 1998. 145 (10): 3335-3339. http://dx.doi.org/10.1149/1.1838808 71. M. Pourbaix. "Atlas of electrochemical equilibria in aqueous solutions." 2nd ed. 1974, Houston, TX: NACE International. 72. B. Seger, A.B. Laursen, P.C.K. Vesborg, T. Pedersen, O. Hansen, S. Dahl, and I. Chorkendorff. "Hydrogen Production Using a Molybdenum Sulfide Catalyst on a Titanium-Protected n+p-Silicon Photocathode." Angewandte Chemie International Edition, 2012. 51 (36): 9128-9131. http://dx.doi.org/10.1002/anie.201203585 73. B. Seger, I.E. Castelli, P.C.K. Vesborg, K.W. Jacobsen, O. Hansen, and I. Chorkendorff. "2-Photon tandem device for water splitting: comparing photocathode first versus photoanode first designs." Energy & Environmental Science, 2014. 7 (8): 2397-2413. http://dx.doi.org/10.1039/C4EE01335B 74. W. Shockley and H.J. Queisser. "Detailed Balance Limit of Efficiency of p‐n Junction Solar Cells." Journal of Applied Physics, 1961. 32 (3): 510-519. http://dx.doi.org/10.1063/1.1736034 75. T.J. Jacobsson, V. Fjällström, M. Sahlberg, M. Edoff, and T. Edvinsson. "A monolithic device for solar water splitting based on series interconnected thin film absorbers reaching over 10% solar-to-hydrogen efficiency." Energy & Environmental Science, 2013. 6 (12): 3676. http://dx.doi.org/10.1039/c3ee42519c 76. N. Kelly and T. Gibson. "Design and characterization of a robust photoelectrochemical device to generate hydrogen using solar water splitting." International Journal of Hydrogen Energy, 2006. 31 (12): 1658-1673. http://dx.doi.org/10.1016/j.ijhydene.2005.12.014 77. R.E. Rocheleau, E.L. Miller, and A. Misra. "High-efficiency photoelectrochemical hydrogen production using multijunction amorphous silicon photoelectrodes." Energy & Fuels, 1998. 12 (1): 3-10. http://dx.doi.org/10.1021/ef9701347 78. S.Y. Reece, J.A. Hamel, K. Sung, T.D. Jarvi, A.J. Esswein, J.J. Pijpers, and D.G. Nocera. "Wireless solar water splitting using silicon-based semiconductors and earth- abundant catalysts." Science, 2011. 334 (6056): 645-648. http://dx.doi.org/10.1126/science.1209816

18

79. F.F. Abdi, L. Han, A.H.M. Smets, M. Zeman, B. Dam, and R. van de Krol. "Efficient solar water splitting by enhanced charge separation in a bismuth vanadate-silicon tandem photoelectrode." Nature Communications, 2013. 4. http://dx.doi.org/10.1038/ncomms3195 80. J.H. Park and A.J. Bard. "Photoelectrochemical Tandem Cell with Bipolar Dye- Sensitized Electrodes for Vectorial Electron Transfer for Water Splitting." Electrochemical and Solid-State Letters, 2006. 9 (2): E5-E8. http://dx.doi.org/10.1149/1.2140497 81. B.D. James, G.N. Baum, J. Perez, and K.N. Baum. "Technoeconomic Analysis of Photoelectochemical (PEC) Hydrogen Production." 2009. https://www1.eere.energy.gov/hydrogenandfuelcells/pdfs/pec_technoeconomic_analy sis.pdf 82. C.C.L. McCrory, S. Jung, J.C. Peters, and T.F. Jaramillo. "Benchmarking Heterogeneous Electrocatalysts for the Oxygen Evolution Reaction." Journal of the American Chemical Society, 2013. 135 (45): 16977-16987. http://dx.doi.org/10.1021/ja407115p

19

20

Chapter 2: Catalyzing the Hydrogen Evolution Reaction (HER) with Molybdenum Sulfide Nanomaterials

2.1 Abstract

In this chapter, we discuss recent developments in nanostructured molybdenum sulfide catalysts for the electrochemical hydrogen evolution reaction. To develop a framework for performing consistent and meaningful comparisons between catalysts, we review standard experimental methodologies for measuring catalyst performance and define two metrics used in this dissertation for comparing catalyst activity: the turn over frequency, an intrinsic activity metric, and the total electrode activity, a device-oriented activity metric. We discuss general strategies for synthesizing catalysts with improved activity, namely increasing the number of electrically accessible active sites or increasing the turn over frequency of each site. Then we consider a number of state-of-the-art molybdenum sulfide catalysts, including crystalline MoS2, amorphous MoSx, and molecular cluster materials, to highlight these strategies in practice. Comparing these catalysts reveals that most of the molybdenum sulfide catalysts have similar active site turn over frequencies, so the total electrode activity is primarily determined by the number of accessible active sites per geometric electrode area. Emerging strategies to overcome current catalyst limitations and potential applications for molybdenum sulfide catalysts including photoelectrochemical water splitting devices and electrolyzers are also considered.

2.2 Introduction

Supplying the world’s growing population with clean, affordable energy is a critical challenge.1 Global energy demand is projected to rise from 17 TW in 2010 to 27 TW by 2040.2 Fossil fuels continue to dominate the global energy landscape, but increasing concerns over the effect of anthropogenic carbon dioxide on the earth’s climate make sustainable energy technologies, such as wind and solar, attractive options.3-6 However, the challenges of energy storage and integration into the current power grid have hindered the implementation of these intermittent renewable energy sources.4, 7, 8

21

As discussed in the previous chapter, electrochemical and photoelectrochemical water splitting are promising ways to store sustainable, intermittent energy resources (e.g. wind and solar) in the form of hydrogen, an energy vector with high energy density and potentially without carbon emissions.9-11 However, substantial technological advancements are necessary to make widespread implementation of water splitting economically viable. One critical requirement is the development of highly active, stable electrocatalysts composed of earth abundant materials.9

In this chapter, we discuss molybdenum sulfide electrocatalysts for the hydrogen evolution reaction (HER). Other dichalcogenides such as tungsten sulfide (WS2) have shown promise as HER catalysts,12-15 but they will not be discussed here. Due to their high activity, excellent stability, and precious metal-free composition, molybdenum sulfides represent a promising class of materials for making electrochemical hydrogen production feasible. There have been many important developments in MoS2 HER catalysis in recent years, many of which have been reviewed elsewhere.15-18 Here, we aim to provide a framework for evaluating and comparing strategies for improving the activity of molybdenum sulfide HER catalysts. We begin by discussing best practices for measuring the activity of HER catalysts, then highlight a number of few specific studies to illustrate important approaches for developing MoS2 catalysts with improved activity. We consider crystalline, amorphous, and molecular cluster molybdenum sulfide materials. To evaluate the efficacy of the various strategies for enhancing activity, we compare a number of the most successful catalysts using two metrics: the turn over frequency,

2 an intrinsic activity metric, and the overpotential required to reach 10 mA/cm electrode, a device- oriented total electrode activity metric. Finally, we discuss emerging directions in MoS2 catalysis.

2.3 Background

Electrochemical water splitting can be performed in a variety of devices, which can be broadly classified into two main categories: water electrolysis and water photolysis. Water electrolyzers, which include polymer electrolyte membrane (PEM), alkaline, and solid oxide electrolyzer configurations, require energy input from an external source of electricity to drive the water splitting process.19-21 As discussed in Chapter 1, photoelectrochemical (PEC) and photocatalytic water splitting devices rely on semiconductor materials to absorb sunlight and generate exited charge carriers, and can therefore split water without an external electricity input.9, 22 To perform

22 the water splitting reactions, catalysts are required for both the HER and the OER. Developments in OER catalysis have been discussed elsewhere;9, 23 in this perspective we focus on the HER.

2.3.1 The Hydrogen Evolution Reaction

The basics of the HER were introduced in Chapter 1. This reaction is thought to involve three possible reaction steps, as shown in Table 2.1:24

  1 Volmer step H  e  H ad

 2 Heyrovsky step H  H ad  H 2

 3 Tafel step 2H ad  H 2

Table 2.1. Elementary steps in the hydrogen evolution reaction.

23

Figure 2.1. Exchange current density as a function of hydrogen adsorption free energy for various HER catalyst materials. a. The experimental “volcano plot” for HER is shown and Pt, with slightly negative hydrogen absorption energy, has the highest HER activity. 25 b. The Parsons theoretical HER volcano predicts catalysts with hydrogen binding energy equals zero will have the highest activity.26 Further details concerning metrics for catalyst activity are given in section 3.1. Reproduced with permission.25, 26

The HER may occur via the Volmer-Heyrovsky mechanism or the Volmer-Tafel mechanism. In both cases, the reaction proceeds through hydrogen atoms adsorbed at the electrode surface,

Had, and thus the rate of the overall reaction is influenced by the free energy of hydrogen

26 adsorption, ΔGH, as originally described by Parsons. If the hydrogen to surface bond is too weak, the adsorption step will limit the overall reaction rate. If the hydrogen to surface bond is too strong, the reaction-desorption step will limit the overall reaction rate. Optimal HER catalysts have hydrogen adsorption energies close to ΔGH = 0, binding hydrogen neither too weakly nor too strongly.16,17,27-29 This principle gives rise to the “volcano” relationship in Figure 2.1, which shows the HER exchange current (a measure of catalytic activity) as a function of

24

ΔGH. To maximize the rate of the HER, a catalyst with appropriate surface properties must be employed. Several classes of materials have been investigated as active HER catalysts, including precious metals such as platinum, nickel alloys, metal oxides, metal phosphides, and metal sulfides; many of these efforts have been reviewed previously.9, 30

2.3.2 Molybdenum Disulfide

31 MoS2 has many interesting properties that allow it to be exploited as a lubricant , 2D-transistor

32 33 , and hydrodesulfurization catalyst , but this review will focus on MoS2 as an HER catalyst.

34 Interestingly, early work on the electrochemistry of bulk MoS2 crystals by Tributsch and others suggested that this material is not an active HER catalyst, but interest has been revived as studies have shown that nanostructuring MoS2 materials can significantly improve HER activity.25, 35

Bulk MoS2 is a hexagonally packed layered structure, similar to graphite, with a 6.5 Å van der Waals gap separating each sheet as shown in Figure 2.2.36-38 As a result of this crystal structure,

MoS2 possesses a variety of distinct surface sites, and electron and hole mobilities approximately 2200 times faster along a basal plane than perpendicularly between sheets.39 The surface of bulk MoS2 primarily consists of the thermodynamically-favored basal plane sites,

34, 40, 41 which are catalytically inert. In contrast, the edges of MoS2 layers have high activity for the HER.35

25

Figure 2.2. a. Structure of the 2H, 3R, and 1T polytypes of MoS2. b. Top view of the Mo edge and S edge of a bulk MoS2 crystal.

In their seminal work on MoS2 for the HER, using density functional theory (DFT), Hinnemann and coworkers calculated that the Mo( 1 010 ) edge of MoS2 possesses a hydrogen binding energy of approximately 0.08 eV at 50% H coverage, very close to the optimum value of 0 eV.35 This binding energy is similar to that observed on active precious metal catalysts, including Pt.35 This work was inspired by enzymes such as hydrogenases and nitrogenases, both of which are effective hydrogen producing catalysts.42-44 Both of these classes of enzymes are highly active, have hydrogen binding energies close to zero, and possess motifs containing Mo, Ni, and Fe with undercoordinated sulfur at their active sites. Guided by the MoS2 edge site prediction, experimentalists synthesized MoS2 on a high surface area carbon black support to expose a large number of edge sites.35 Their membrane electrode assembly setup, which reduced protons to

2 make hydrogen at the cathode and oxidized hydrogen at the anode achieved 10 mA/cm electrode

26 at about 175 mV overpotential, the most active, acid stable, nonprecious metal catalyst for the HER at the time.

In 2007, it was experimentally confirmed that the MoS2 edges are active catalysts for the HER.

Jaramillo and coworkers deposited single monolayer MoS2 nanoparticles on Au(111) by

25 physical vapor deposition of Mo in an H2S environment. After quantifying the nanoparticle area and edge length using scanning tunneling microscopy (STM), the authors measured the

HER activity and confirmed that the reaction rate scaled with MoS2 perimeter length rather than area. The per-site activity (turnover frequency) of MoS2 edges was very high, yet still 50-100 times lower than that of Pt.45 The insights gained by means of theory and fundamental experimental studies consequently sparked significant efforts aimed at developing MoS2 catalysts with improved activity approaching that observed for platinum surfaces.

2.4 Parameters for Catalyst Activity Comparison

There are many performance characteristics that may be important in evaluating an electrocatalyst material’s efficacy for a particular application, including its activity, stability, selectivity, cost, and optical or mechanical properties. Catalytic activity is almost always critical, however there is no universal method for assessing the activity of HER catalysts.46, 47 A variety of methodologies are described in the literature, and the differences among these strategies can make it challenging to compare different catalysts in a consistent manner.12, 16, 17, 25, 35, 48-74 Furthermore, all activity comparison approaches have advantages and disadvantages, so it is important to choose an appropriate strategy based on the relevant performance metrics for a given system. In this section, we first briefly review standard methodologies employed for measuring HER electrocatalyst activity. Then, we focus on two particularly important activity metrics to provide a fair and useful framework for comparing molybdenum sulfide HER catalysts.

The two primary categories of activity measurements of interest in this work are: “total electrode” activity (i.e. geometric electrode area-normalized measurements) and “intrinsic” activity (i.e. per site turn over frequency, TOF). Total electrode activity measurements are useful for practical device performance comparisons, but they are not ideal for fundamental studies of novel catalyst materials because they do not reveal the physical or chemical origins of an electrode’s activity. Intrinsic activity measurements provide the activity of the catalyst on a per-

27 site basis, and therefore contribute to the molecular-level structure-property-function relationships necessary to guide catalyst development. As shown in Figure 2.3, these two classes of activity measurements are fundamentally related, as the total electrode activity of any HER catalyst is determined by the product of two factors: the number of active sites and the intrinsic activity (turn over frequency) of each site. Intrinsic activity measurements deconvolute these two factors and thus provide more information, but such measurements can be challenging to perform accurately.

Figure 2.3. Equation relating HER current and turn over frequency of each catalytic site. In practice, it is easiest to measure the current and number of surface sites, then derive the average turnover frequency per site.52

2.4.1 Total Electrode Activity Measurements

Total electrode activity measurements are useful for comparing complete electrodes and are typically the first step in characterizing HER catalyst materials. These measurements are usually performed by supporting the catalyst on an inert, conducting substrate and performing cyclic voltammetry or linear sweep voltammetry to measure the catalytic current as a function of potential. For water splitting reactions, it is most appropriate to calibrate the potential scale to the reversible hydrogen electrode (RHE). The current is often normalized to the superficial geometric electrode area to facilitate comparison between materials tested under different experimental conditions.

28

While the full cyclic voltamograms (CV) or linear sweep voltamograms (LSV) contain the most complete information about catalyst activity, it is often helpful to report measurements at certain specific potentials to facilitate comparison among catalyst materials. Some special potentials are noted in Figure 2.4. One potential of interest is the “onset potential,” or the potential at which catalytic current is first observed. Without a strict definition, the ambiguity of the onset potential makes it a poor criterion for comparing different catalysts; different observers may assign different onset potentials to the same data and non-Faradaic capacitive current may comprise a significant fraction of the total current in the onset region. Thus, the onset potential should always be defined based on a specific current density. Depending on the surface area of the catalyst and scan rate of the LSV, a suitable current value for the onset could range from 0.05 – 5 mA/cm2. A more relevant metric by which to compare catalysts is the potential required to reach an operating current density of interest. In solar water splitting, the potential for 10

2 mA/cm electrode is a common figure of merit because this is the current density expected in a 12.3% efficient solar to hydrogen device, which is on the order of the efficiency that would be required for cost competitive photoelectrochemical water splitting.9, 48, 52, 75-78 In this dissertation, we use this metric to compare the total electrode activities of the MoS2 HER catalysts. The potential to reach a different current density may be more relevant for catalysts intended for application in water electrolyzers, which typically operate at current densities on the order of 1- 2 A/cm2.19

29

Figure 2.4. HER activity of two theoretical catalysts. a. Representative linear sweep voltamograms (LSV) plotting, current density as a function of potential. b. Tafel plot. These catalysts require the same 2 overpotential for -10 mA/cm electrode, so they appear the same based on the total electrode activity metric used in this perspective. However, these catalysts likely have different HER mechanisms based on their different Tafel slopes, and both could be better or worse than the other depending on the application. The 2 catalyst represented in blue would perform better for low current devices (< 10 mA/cm electrode) while the 2 catalyst represented in red would be superior for high current density devices (> 10 mA/cm electrode).

The exchange current density and Tafel slope are also frequently reported total electrode activity metrics.57, 61, 79, 80 The specific value of the electrode area-normalized exchange current density in the absence of other critical information is not an ideal metric for ranking catalyst activity since it is neither a relevant practical performance parameter nor a fundamental material property, as it depends on both the per-site turn over frequency and the total number of sites. However, recognizing that the site-specific exchange current density can vary by 10 orders of magnitude between a good catalyst and bad a catalyst, and that the total number of active sites on an electrode typically only varies by 2-3 orders of magnitude between a flat catalyst film and

30 a nanostructured film, a high total electrode exchange current density is a necessary but insufficient condition for highly active electrodes. Additionally, in combination with the Tafel slope, the electrode area-normalized exchange current density can be an informative parameter. When applied in the Tafel equation, these two metrics together can be used to calculate the overpotential required to achieve any current density. However, changes in the Tafel slope as a function of potential due to transport limitations, changes in catalyst surface structure, or other effects may introduce errors in this calculation, and the same information can usually be obtained more directly from the measured current-voltage curves. From a practical perspective, the Tafel slope determines the additional voltage required to increase the catalytic current by an order of magnitude, with units commonly reported in mV/decade. Ultimately, all electrodes are judged by the overpotential required to reach an operating current density, which is determined by both its Tafel slope and onset potential. Therefore, low Tafel slopes are desirable, especially for high current applications such as water electrolyzers. As demonstrated in Figure 2.4, catalysts with different Tafel slopes could still end up requiring the same overpotential to reach

2 10 mA/cm electrode. Under certain conditions using model HER catalyst systems, the Tafel slope may also provide evidence of a particular reaction mechanism. However, in most cases the Tafel slope alone cannot be used to unambiguously identify the mechanism in the absence of additional experiments.48, 52

In summary, for total electrode activity measurements the most relevant metric by which to

2 compare catalysts is the potential to reach a current density of interest, e.g. 10 mA/cm electrode for HER catalysts intended for solar fuels applications. As there is no normalization to the amount of catalyst loaded per geometric area, it follows that electrodes with very high catalyst loadings typically have the highest total electrode activity.

2.4.2 Intrinsic Activity Measurements

The ideal means to understand catalyst performance is by measuring the activity of each catalytic site and relating it to the site’s physical and chemical properties. For nonprecious metal catalysts, total electrode activity is the most important metric for device-oriented catalyst development. The intrinsic activity of each catalytic site is more important to develop a fundamental understanding of the origins of a material’s catalytic activity, which is necessary to design highly active electrodes. Unfortunately, limitations of current experimental techniques make this task challenging. Most practical catalysts include many different types of surface

31 sites, each with their own inherent activity, and there are few experimental techniques to probe individual sites. Different methods are required for different materials, measurements are often complicated by non-Faradaic current, catalyst instability, and/or transport limitations. The most common strategy for determining intrinsic per-site activity is to first measure the total electrode activity and then, through a separate measurement, to determine the total number of active sites and use these results to infer the average TOF. While less powerful than the ideal measurement, this strategy still enables the development of relationships between material properties and catalytic activity to drive further catalyst design efforts.

Measuring the Number of Active Sites

All intrinsic activity measurements include some effort to determine the number of active sites. In most cases, such a measurement is approximate rather than precise. Sometimes related quantities such as mass loading, total surface area, or electrochemically active surface area (ECSA) may also be employed in place of the number of active sites. Some of the relevant measures of surface area are illustrated in Figure 2.5.

Normalizing the total electrode activity by the catalyst mass is important as a practical performance metric for precious metal-based catalysts where the cost of the catalyst material is a primary concern.81 To perform these measurements, the electrode-supported catalyst material is often dissolved off of the electrode surface and its mass is measured using an established method of quantitative chemical analysis, for instance by means of inductively coupled plasma optical emission spectroscopy (ICP-OES) or a related technique. Mass normalization is less common for nonprecious materials like molybdenum sulfides because the catalyst mass is not as critical to the cost of the device. Catalyst mass measurements, however, may still be used as a means to estimate the total number of active sites, which is the key parameter for understanding catalyst property-performance relationships. While helpful, this technique is typically not the optimal strategy for intrinsic activity measurements because the number of active sites might not scale linearly with mass loading due to variations in catalyst morphology.

32

Figure 2.5. 2D representation of MoS2 catalyst electrochemically active surface area and projected geometric surface area. For crystalline 2H-MoS2, only edge sites, indicated by green dots, are active for HER. Basal plane sites, indicated by blue squares, are not active for HER. Rougher surfaces should have more available edge sites and thus higher catalytic activity per projected geometric surface area. It should also be noted that a catalyst geometry with basal planes oriented perpendicular to the surface would result in a more active catalyst but the less active geometry is displayed here for visualization purposes.

Measuring the total catalyst surface area or its electrochemical active surface area (as opposed to the superficial geometric electrode area) is one of the most common approaches for estimating the number of catalytic sites in nonprecious materials.48, 56, 68 This is an appropriate strategy because the number of active sites often scales with the catalyst surface area. Methods for measuring active surface area include both gaseous adsorption and electrochemical techniques. Gaseous adsorption techniques such as BET, which rely on the adsorption of probe molecules

(e.g. N2) at the catalyst surface, are excellent for heterogeneous catalyst materials and may be useful for some electrocatalysts, but these techniques can potentially overestimate the active surface area by adsorbing to some material that is not in electrical contact to the substrate and therefore cannot catalyze the reaction.82, 83 Overestimating surface area in this case would lead to a conservative estimate of the TOF. Electrochemical measurements are often preferred because they directly probe the electrochemically active surface area (ECSA).82 One popular technique involves measuring the non-Faradic current associated with electrochemical double layer charging at the catalyst surface.48, 56, 68, 82 This value can be compared to the material’s specific capacitance determined using a flat reference material to determine the catalyst's ECSA. Electrochemical probe molecule adsorption is another common technique. The most prevalent example of this strategy is hydrogen underpotential deposition, a standard method for measuring the surface area of platinum electrocatalysts.84 Other common probe molecule techniques

33 include underpotential deposition of lead or copper and carbon monoxide stripping, which involves adsorption and electrooxidation.85-88 Unfortunately these methods are not universally applicable to all materials. While the conditions used for these electrochemical surface area measurement techniques generally approximate HER operating conditions better than those used for gaseous adsorption techniques, these electrochemical measurements may still be imperfect because factors such as the catalyst surface structure or active site accessibility could change between the surface area measurement conditions and the operating conditions.

In some cases, the electrocatalyst contains structurally or chemically heterogeneous sites with substantially different TOFs. Under these circumstances, techniques that enable a direct determination of the number of active sites are preferred to surface area measurements because the number of active sites may not scale directly with the overall surface area or ECSA. This is the case with MoS2 HER catalysts, which generally consist of edge sites that are catalytically active for the HER and basal plane sites which are inert. In some studies, the number of active

MoS2 edge sites has been estimated using scanning tunneling microscopy or transmission electron microscopy imaging.25, 52 Other studies have employed electrochemical oxidation of the MoS2, yielding distinct electrochemical features corresponding to the edge sites and the basal plane sites.50 The number of edge sites was determined by integrating the appropriate oxidative feature. Other works have used probe molecules such as copper ions that adsorb selectively to the active sites,14 though the reliability of these techniques has not been well established for molybdenum sulfide materials. Any of these techniques could provide a useful estimate of the number of active sites, and the development of new methodologies for this task is still an active area of research, for MoS2 as well as for other electrocatalyst materials for the HER and other reactions.

In summary, measuring the total active surface area is the most common strategy for determining the number of active sites on many electrocatalysts and there exist many effective methods for doing so. For MoS2 materials techniques that specifically determine the number of active edge sites are particularly helpful. Counting the number of active sites is typically easiest for model systems with low catalyst loading. Additional approximations are often necessary for electrodes with high total activity.

34

Calculating the TOF of Each Site

The activity of each surface site, quantified as the TOF, is generally the most significant factor affecting the overall activity of an electrode for the HER, as TOFs can vary by more than 10 orders of magnitude for different electrocatalysts, while the density of surface sites on an electrode typically varies by only up to ~3 orders of magnitude.89, 90 Thus determining the active site TOF is especially critical in evaluating novel electrocatalyst materials. The TOF is primarily determined by the physical and chemical properties of the catalyst surface that ultimately determine parameters critical to the reaction chemistry at active sites, including the binding energies of reactive intermediates and the activation energies involved in the elementary steps. The TOF, however, is difficult to measure directly, so it is typically inferred from the measurements of total electrode activity and the number of sites as described above. While each unique active site may have a different turn over frequency, there is usually no practical means to measure the TOF of one site in isolation, so reported TOF numbers are almost always averages over many sites, i.e. the TOFavg as shown in Figure 2.3.

While it is generally assumed that hydrogen binds to undercoordinated sulfur edges sites, in this

-1 chapter, we use the average TOF in units of H2 s per surface Mo atom to compare the intrinsic activities of molybdenum sulfide HER catalysts. This avoids difficulties in distinguishing the activity among different sulfur edge sites. A number of different research groups have designed

MoS2 nanostructures with high edge site exposure and electrical accessibility. The surface Mo atom-normalized TOF provides useful insights about the efficacy of these strategies, as it can be used to infer the fraction of surface sites that is active. Another important research aim is to improve the TOF of each active site by modifying the properties of the molybdenum sulfide. Comparing the average TOF of the active sites for such catalysts would be ideal, but the data necessary to make this comparison are not readily available due to variations in the methods used to measure and report active site concentrations as well as differences in catalyst structure and stoichiometry. Performing a consistent comparison of the surface Mo atom-normalized TOF is practical given available data, and though this strategy is not perfect, it still provides useful insights about the intrinsic activity of the modified catalysts.

35

2.4.3 Complicating Factors

Slow mass transport can complicate the determination of the total electrode activity and active site TOF, but appropriate experimental procedures can usually reduce this effect. For HER catalysts operating in acidic electrolyte, the concentration of protons is typically high enough that diffusion limitations are not encountered, but slow mass transport may become significant in porous electrodes operating at large current densities. Hydrogen bubbles may also accumulate at the electrode surface and block some active area; a rotating disk electrode can often be helpful in this regard.

Electrical series resistance can also impact activity measurements through iR potential drops. Experimenters can reduce these effects by measuring the circuit resistance using electrochemical impedance spectroscopy and applying an appropriate correction during or after the experiment. This correction does not account for a non-uniform potential across the active surface if the substrate has a high sheet resistance or if the catalyst possesses a three-dimensional nanostructure that results in a range of charge transport path lengths. Nevertheless, when recognized, these effects can usually be mitigated sufficiently to enable a reasonably accurate TOF determination.

Selectivity and stability are other important considerations in measuring the efficacy of electrocatalysts in general. Measuring the current passed through the electrochemical circuit does not reveal which electrochemical reactions occurred, so independent measurements of

91 product concentration are essential, especially for cases such as CO2 reduction. For most HER catalysts, selectivity measurements are important but generally not as crucial as with other reactions because in acidic electrolytes, H2 is the only plausible reaction product from proton reduction that can be produced a significant rate. Nevertheless, under some circumstances, species other than protons might be reduced. Even the electrode material itself may be cathodically corroded, contributing to the measured Faradic current. Independent measurements of the evolved H2 concentration using volume displacement, gas chromatography, or an electrochemical hydrogen detector may be necessary to confirm selectivity. Coulometry measurements also provide a straightforward means of initial HER selectivity determination, i.e. if the total charge passed in a particular measurement is orders of magnitude higher than would be required to corrode the catalyst itself, this is strong evidence that most of the current must correspond to the HER.

36

2.5 State of the Art Molybdenum Sulfide Electrocatalysts

Recent years have seen many efforts to develop improved molybdenum sulfide HER catalysts in several forms, including crystalline, amorphous, and molecular cluster materials. To accomplish this goal, these studies employ two basic strategies corresponding to the two fundamental factors that determine overall electrode activity. First, the activity can be improved by increasing the number of electrically accessible active sites. This is non-trivial, as one must consider the catalyst nanostructure, which controls the atomic structure of the exposed surface, as well as the conductivity between the active sites and the conductive support. Electron and hole mobility is about 2200 times faster along a basal plane than between sheets,39 so catalytic activity may be hindered if the active sites are separated from the conductive substrate by MoS2 basal planes oriented parallel to the surface of the support.74 Second, the activity can also be improved by increasing the TOF of each individual site by modifying the physical or chemical properties of the molybdenum sulfide through surface structure doping, strain, support interactions, or other effects. Herein, we briefly review some recent results in state of the art molybdenum sulfide electrocatalyst design in the context of these activity enhancement strategies. Broadly, we classify these catalysts as: crystalline MoS2 materials, amorphous molybdenum sulfide films, or molecular molybdenum sulfide clusters.

2.5.1 Crystalline MoS2

Most recent works on crystalline MoS2 catalysts have focused on creating device-ready structures by controlling the catalyst nanostructure to increase the number of accessible edge sites per geometric area. MoS2 morphologies with high active site densities such as nanowires, nanoparticles, and modified thin films have all been developed. The materials discussed in this section have a 2H polymorph hexagonal crystal structure and are synthesized primarily through gas phase sulfidization of a molybdenum-containing precursor (molybdenum metal or molybdenum oxide) or by hydrothermal methods.36, 52, 56, 57, 61

MoS2 Nanowires

MoS2-MoO3 core shell nanowires, as seen in Figure 2.6, were synthesized by sulfidizing MoO3

52 nanowires in a 10% H2S/90% H2 atmosphere. This nanowire morphology was developed for its high surface area, which gives it the potential to expose a large number of active edge sites

37 per geometric electrode area. Furthermore, the conductive MoO3 core facilitates charge transport along the length of the nanowires, improving the electrical accessibility of the active

2 sites. This catalyst possessed reasonably high HER activity, achieving 10 mA/cm electrode at 250 mV overpotential. However, the surface Mo-atom averaged TOF determined using anodic stripping of the MoS2 and TEM imaging revealed a low density of exposed active edge sites, as the that MoS2 basal planes curved around the circumference of the nanowires. This result emphasizes one particular strategy to improve overall activity, designing the nanoscale morphology of the MoS2 to favor higher exposure of active edge sites rather than extended basal planes, a valuable insight for further catalyst design efforts.

Figure 2.6. a. TEM image of MoS2/MoO3-x shown structurally (right). This catalyst shows no degradation after 10,000 potential cycles despite the fact that MoO3 is unstable in acid and only 3 layers (2-3 nm) of MoS2 protects it. b. SEM image of MoS2/MoO3-x nanowires after 200°C thermal sulfidization. The inset shows a photograph of the sulfidized sample. Reproduced with permission.52

The outstanding stability of MoS2 in acidic electrolyte was another key insight provided by this structure. The thickness of the MoS2 shell was shown to depend on sulfidization temperature. At the lowest temperature tested, 150°C, the sulfidation was incomplete, leaving a shell with pinholes exposing the acid-unstable MoO3 cores and causing rapid degradation during potential cycling. Sulfidization at 200°C produced a MoS2 shell ~ 3 nm thick, and was found to have the highest HER activity among the samples investigated. This sample lost no activity over 10,000 potential cycles in 0.5 M H2SO4, demonstrating the ability of MoS2 to withstand very harsh testing conditions while protecting the underlying MoO3 core, a key asset for practical device integration.

38

Mesoporous Double Gyroid MoS2

The double gyroid mesoporous nanostructure, shown in Figure 2.7, exhibits a high surface area

56 despite minimal total film thickness. The double gyroid is an excellent structure for a MoS2 HER catalyst because its nanoscale curvature mitigates the formation of extended basal planes, resulting in a high density of exposed active edge sites. This MoS2 structure was synthesized by electrodepositing MoOx in a silica double gyroid template, sulfidizing the MoOx in 10% H2S/

90% H2 at 200°C, and removing the silica template with a hydrofluoric acid etch. The resulting

2 catalyst showed improved activity, achieving 10 mA/cm electrode at 220 mV overpotential. The electrochemically active surface area of this structure was determined using capacitance measurements, and the resultant surface Mo-atom averaged TOF showed that the fraction of active sites exposed at the surface of the double gyroid was 2x – 4x higher than in the MoO3-

MoS2 nanowires. An additional advantage of the MoS2 double gyroid is its tunable thickness. If

MoS2 catalysts are to be integrated into cost-effective PEC water splitting devices, synthesizing highly active thin film structures that do not absorb a large portion of incoming solar radiation is desirable. The double gyroid film thickness is tunable by varying the MoOx electrodeposition time, allowing for a tradeoff between thicker films that have higher activity and thinner films that have lower parasitic light absorption. A downside of the MoS2 double gyroid is an increase in resistive losses due to long path lengths for electron transport from the active site to the conducting or semiconducting substrate.

39

Figure 2.7. MoS2 double gyroid mesoporous structure. a. TEM image of the [311] and [211] projections. b. TEM image of the [110]. c,d,e. Models corresponding to simulated TEM images of the [211], [311], and [110] projections, respectively. f. 3D model of the double dyroid. Reproduced with permission.56

MoS2 Nanoparticles

MoS2 nanoparticles have been synthesized a variety of ways, including chemical exfoliation of

92 50, 93 a MoS2 precursor, thermal sulfidization of heptamolybdate precursor, and chemical

70 reduction of MoS3 nanoparticles. Various optimal loadings and activities were reported, likely resulting in various degrees of success with nanoparticle dispersion. Highly dispersed nanoparticles will allow access to more active sites than agglomerated particles. As shown in

Figure 2.8, the most active electrodes to date based on MoS2 nanoparticles were synthesized on reduced graphene oxide (RGO) nanosheets to increase the amount of active edge sites per geometric surface area. Solvothermal synthesis was carried out using (NH4)2MoS4 and hydrazine in dimethylformamide at 200°C. Coupling MoS2 nanoparticles with the RGO support showed a substantial improvement in activity compared to free MoS2 nanoparticles. The RGO support led to better nanoparticle dispersion and a different morphology compared to the agglomerated RGO-free nanoparticles. These differences ultimately led to superior active site access and likely better charge transport from the active site to the glassy carbon disk or porous

40 carbon paper onto which the catalyst was loaded for testing. The Tafel slope was 41 mV/decade,

67, 94 in line with the best Tafel slopes ever reported for MoS2 catalysts, and the catalyst achieved

2 10 mA/cm electrode at ~150 mV overpotential. The high total electrode activity of these catalysts

2 61 was due in large portion to a very high catalyst loading (1 mg/cm electrode). More recent work has shown that MoS2 has also been combined with other carbon nanomaterials to form very active electrodes.62, 63

Figure 2.8. Synthesis of MoS2 nanoparticles with and without reduced graphene oxide (RGO) sheets. a. Schematic diagram of MoS2/RGO nanoparticle synthesis. b. SEM and inset TEM images of the MoS2/RGO catalyst. c. Schematic of large, free MoS2 nanoparticle synthesis. d. SEM and inset TEM 61 images of the MoS2 particles with no RGO. Reproduced with permission.

Vertically Aligned MoS2 Thin Films

In an effort to expose a high density of edge sites and reduce resistive losses from electron transport perpendicular to MoS2 basal planes, Kong, et. al. synthesized vertically aligned

57 crystalline MoS2. The vertically aligned MoS2, shown in Figure 2.9, exposes primarily edge sites to the electrolyte and allows fast electron transport through a single MoS2 layer to the conducting substrate. Synthesis is achieved by exposing a 5 nm electron beam evaporated Mo film to 100 mTorr elemental sulfur vapor for 10 minutes at 550°C, a sulfurization temperature

57, 95 later found to produce active MoS2 nanosheets. The vertically oriented catalyst morphology is promising because MoS2 with only exposed edge sites should increase the number of active

-6 2 sites available to perform HER. The exchange current density of 2.2 x 10 A/cm electrode is high

41 and corresponds to a TOF of about 0.013 s-1 but the Tafel slope is also rather high (120

2 mV/decade). By extrapolation, the 10 mA/cm electrode is reached at 440 mV overpotential, resulting in a comparatively low catalyst activity on a geometric basis. However, these films show an exciting new geometry and may make extremely active catalysts if nanostructured or loaded on a higher surface area support.57

Figure 2.9. Vertically aligned MoS2 sheets. TEM image (left) and schematic diagram (right) indicating edge sites and terrace sites. Reproduced with permission.57

2.5.2 Amorphous Molybdenum Sulfides

Amorphous molybdenum sulfides have been known for several decades,96-102 but their excellent catalytic activity for the HER was discovered only recently.48, 67, 68, 70, 71, 103 These materials can be synthesized using electrodeposition or wet chemical reactions with no thermal sulfidization treatment, which may make them particularly advantageous for some applications.48, 67, 68, 70 This material has been successfully applied in several photoelectrochemical and photocatalytic water splitting devices.104-106

Amorphous molybdenum sulfide catalysts often possess high overall electrode activities, largely

48, 67, 68, 70, 71 due to their high surface area morphologies. Amorphous MoSx catalysts that

2 achieved 10 mA/cm electrode at -0.20 V vs. RHE were measured to have roughness factors of

17 2 nearly 100, corresponding to a total surface site density on the order 10 per cm electrode and a

-1 48 resulting TOF around 0.3 H2 s per site averaged over the total surface. Physical and chemical

42 characterization have revealed that as synthesized, the amorphous material typically possesses a composition close to MoS3, but upon applying a cathodic potential, the surface composition

48, 67, 70 changes to MoS2. Thus the electrochemically active surface likely resembles crystalline

MoS2 in chemical state, though no studies to date have shown evidence of crystallinity in this material. While it is difficult to identify the specific active sites on amorphous molybdenum sulfide surfaces due to their significant atomic-scale heterogeneity, these results suggest that the active sites probably have similar TOFs to those observed on the edges of crystalline MoS2.

The properties and performance of a wet-chemical synthesized amorphous MoSx catalyst will be discussed further in Chapter 3.

2.5.3 Molybdenum Sulfide Molecular Clusters

Molybdenum sulfide nanocluster compounds that bridge molecular and solid-state electrocatalysis when supported on electrode surfaces are another interesting class of HER catalysts. These inorganic clusters consist of small molecular units of molybdenum sulfide with under coordinated sulfur abounding at its surface, resembling MoS2 edges. Two examples, the

4+ 2- incomplete [Mo3S4] cubane and the thiomolybdate [Mo3S13] nanocluster, are shown in Figure 2.10.

43

4+ 2- Figure 2.10. a,b. Structure of [Mo3S4] cubane and [Mo3S13] nanoclusters, respectively. Blue: Mo atoms, yellow: S atoms, and red: O atoms (from water ligands). c. STM image of anodized HOPG surface 2- 2- after drop-casting [Mo3S13] clusters. d. Atom-resolved STM image of a single [Mo3S13] cluster are shown on the right. Reproduced with permission.53, 94

4+ The incomplete [Mo3S4] cubanes have been demonstrated to be active for electrocatalytic and

53, 107 4+ photoelectrochemical hydrogen production. STM images of HOPG-supported [Mo3S4] cubanes showed that individual clusters were uniformly scattered on the HOPG surface and the surface coverage was determined to be 1.0 (±0.1) ×1013 clusters/cm2. Together with an exchange

-7 2 current density of 2.2×10 A/cm electrode this surface coverage yielded an exchange TOF per

4+ -1 [Mo3S4] cluster of 0.07 s at 0 mV overpotential. However, the hydrophilicity of the cubanes made the activity decrease over time due to desorption of the catalyst.53 This was largely mitigated, provided that the concentration of O2 was kept low (≤15 ppb) in the system, in a later photoelectrochemical study by changing to methylcyclopentadienyl ligands to obtain less hydrophilic clusters and avoid deactivation by dissolution of the hydrophilic cubanes.107

2- Recently thiomolybdate [Mo3S13] nanoclusters have also been examined for the HER. These clusters are prepared by straightforward wet chemical methods and can be supported on a variety

4+ of substrates by simple drop-casting from a methanol solution. Similar to the [Mo3S4] cubanes,

2- the [Mo3S13] nanoclusters were supported on anodized HOPG and imaged with STM. The

2- 12 2- surface coverage of [Mo3S13] clusters was determined to be ~8 (±2) × 10 [Mo3S13] clusters

2 2- per cm HOPG. These HOPG supported [Mo3S13] clusters showed the highest HER turnover frequency of any molybdenum sulfide catalyst ever synthesized by scalable wet-chemical

2- methods (see section 2.6.1). Furthermore, the [Mo3S13] clusters have very low solubility in

108 4+ water and show better stability at similar current densities than the hydrophilic [Mo3S4] cubanes.

44

2.5.4 Stability of Molybdenum Sulfides

Catalyst stability is an important, but often overlooked, aspect of device-oriented electrocatalysts. The fuel cell research field has adopted standard methods for accelerated durability tests that applies to fuel cell membrane electrode assemblies,109 but no such standard protocols exist for catalysts in electrolyzers or photoelectrochemical water splitting devices, where MoS2 electrocatalysts might be utilized. In the MoS2 electrocatalysis literature, potential cycling has been the predominant method used to assess catalyst stability. Simplistically, potential cycling approximates startup-shutdown conditions in an electrolyzer and illumination intensity fluctuations in photoelectrochemical devices and in principle can provide a harsher and faster way to examine stability than with either chronoamperometry (constant potential) or chronopotentiometry (constant current) testing. However, it may also be beneficial to perform constant current or potential measurements which may better simulate long term operation conditions. Photoelectrochemical device degradation testing is further complicated by illumination conditions. The true potential at the semiconductor surface is related to both the applied potential and the photovoltage, a fact that must be accounted for when testing illuminated photoelectrodes.

Crystalline MoS2 materials have typically demonstrated greater stability than amorphous and molecular MoS2 catalysts. A commonly used protocol to access stability is 1,000 potential cycles, though various scan rates and potential ranges have been employed in various works.57, 61, 92 Nevertheless, all these catalysts showed negligible or near-negligible loss in activity after cycling. Intense stability testing of 10,000 potential cycles between +0.2 and -0.3 V vs. RHE at

50 mV/sec was conducted on the MoO3-MoS2 nanowires, which showed no measurable activity

52 loss, highlighting the exceptional stability of crystalline MoS2. Amorphous MoSx, though not as stable as crystalline MoS2 still exhibits reasonable stability. The potential to achieve 10

2 mA/cm electrode increased by only 57 mV after 10,000 potential cycles from +0.1 V to -0.25 V vs. RHE at 50 mV/sec, 30 mV of which was recovered by simply refreshing the electrolyte.48 The molecular cluster catalysts were the least stable form of molybdenum sulfide. The

4+ [Mo3S4] cubanes on Vulcan XC-72 carbon black lost significant activity between the first and

2 tenth potential cycle as the overpotential to reach 10 mA/cm electrode dropped from about -240

53 2- mV to -270 mV. This activity loss was attributed to catalyst desorption. The [Mo3S13] cluster catalyst has excellent stability for a molecular catalyst, only increasing by 13 mV overpotential

2 required to achieve 10 mA/cm electrode from -0.187 mV vs. RHE to -0.200 mV vs. RHE after

45

94 1,000 potential cycles. However, the Mo3S13 was supported on carbon paper so the interaction between the catalyst and support may be different than that between cubanes and carbon black, which could account for the measured stability differences. In both cases, desorption is likely the primary cause of the observed activity loss.

2.5.5 Emerging Directions

Most of the emerging directions in molybdenum sulfide research involve increasing overall edge site turnover frequency, inspired in large part by past successes in MoS2 research reported in the hydrodesulfurization (HDS) literature, a field in which carbon and alumina supported MoS2 has long been an economically viable catalyst.110-113 This reciprocity is not surprising because an active HDS catalyst must also bond hydrogen when removing sulfur as H2S from refined petroleum products. Four novel approaches in improving the intrinsic activity of MoS2 are activating the S-edge by doping, modifying the H-binding energy through substrate interactions, tuning electronic properties through Li+ ion intercalation, and utilizing the conductive 1T polymorph of MoS2.

Doping

It has been well established that adding small amounts of transition metal dopants including Co and Ni to MoS2 can increase its catalytic activity for HDS by more than an order of magnitude.114-117 Studies using STM have revealed that these dopants are located predominantly at the S-edges of doped MoS2 clusters, and that the activity enhancements arise from the dopant’s role in modifying the hydrogen bonding energy at the S-edges.118 Similar to the HDS reaction, doping with Fe, Co, and Ni has also been demonstrated to increase the HER activity

50, 68, 119 of the MoS2 S-edge. DFT calculations showed that in unmodified MoS2, the hydrogen binding energy, ∆GH, is 0.08 eV at the Mo-edge and 0.18 eV at the S-edge. Incorporating Co dopants decreased the binding energy at the S-edge from 0.18 eV to 0.10 eV, while the ∆GH at the Mo-edge was unaffected. Hence, the role of Co is to increase the number of active sites in nanostructured MoS2 catalysts by activating the S-edges. Thus, while these studies have demonstrated important activity enhancements, greater gains may be possible if the ∆GH at the active sites even can be reduced even further to approach 0.00 eV using new doping strategies.

Transition metal dopants including Ni, Co, and Fe have also been used to improve the activity of amorphous molybdenum sulfide catalysts.68 Cobalt doping was most successful, yielding in

46 changes in the catalyst morphology that increased the active surface area more than 300%,

2 68 resulting in electrodes that achieved 10 mA/cm electrode at -0.17 V vs. RHE in pH 0 electrolyte. Under these conditions, the doping did not affect the intrinsic activity of the catalyst, but the doping did improve the per-site TOF in pH 7 electrolyte.

Doping MoS2 with other transition metals to make mixed metal sulfides may also provide a promising path to higher HER activity. Mo1-xVxS2 catalysts (x ~ 0.05) were shown to be more

120 active for hydrogen evolution than similarly prepared MoS2. The activity enhancement resulted from lower resistivity and higher carrier concentrations. Further efforts to understand the effects of doping this material may enable additional improvements in its performance.

Substrate Interactions

Theoretical results from density functional theory have also predicted that H-binding at both the edge sites and the basal plane of single layered MoS2 could be modified by support interactions, thereby tuning HER activity.121, 122 The change in hydrogen binding is shown to arise from support interactions involving van der Waals forces. Stronger adhesion of the MoS2 onto the support leads to weaker hydrogen binding, which may yield several orders of magnitude difference in HER turn over frequency. These results may explain the lower than expected exchange current densities of supported MoS2 in electrochemical H2 evolution studies compared to theoretical predictions.121 It may be possible to optimally tune the hydrogen binding with a

121 support that binds MoS2 with a physisorption strength of approximately -0.30 eV.

Lithium Intercalation

Intercalation of Li ions into the van der Waals gaps tunes the electronic properties of MoS2, which leads to an increase in HER activity.14, 65, 69, 73 The Li electrochemical intercalation was

73 explained to have at least three effects on the electronic structure of MoS2. First, Li ions change the electronic band structure by increasing the layer spacing and eventually exfoliating MoS2 into individual layers. Second, the intercalated Li could change the d-band filling and reduce the oxidation state of Mo. This could significantly change the H-binding energy and improve the HER activity. Third, high Li content leads to a phase transition for MoS2 from the 2H semiconducting phase to the 1T metallic phase, which could lead to an increase in conductivity of the catalyst and potentially activate the otherwise inactive basal plane, as discussed further

69, 73 below. Li intercalation was applied to vertically aligned MoS2 on high surface area carbon

47 substrates deposited by atomic layer deposition of MoO3 followed by rapid sulfurization. This

2 catalyst demonstrated an impressively low overpotential of 168 mV at 10 mA/cm electrode and negligible degradation after 1,000 cycles.73

1T Phase of MoS2

Exfoliation of flowerlike MoS2 nanostructures was demonstrated using a reaction of n-

65 butyllithium and water, yielding metallic 1T-phase MoS2 nanosheets. These polymorph

2 catalysts showed required only 187 mV to reach 10 mA/cm electrode, surpassing the activity of similarly prepared 2H-phase catalysts by 130 mV. Lukowski et. al. argue that this is achieved by increasing the number of active edge sites and drastically improving the charge transfer resistance from 232 Ω to only 4 Ω. In another study by Voiry et al., the origin of active sites on

1T-phase MoS2 nanosheet was investigated by partially oxidizing both bulk 2H-phase MoS2

69 and MoS2 nanosheets. It was shown that the activity of 2H-phase MoS2 was significantly reduced after oxidation, presumably due to the oxidation of the edge sites. On the other hand, the 1T-phase MoS2 remains unaffected after oxidation. Voiry and co-workers suggested that edges of the 1T-phase MoS2 nanosheets are not the main active sites and the basal plane could

69 + be catalytically active. A recent catalyst consisting of Li -intercalated 1T-phase MoS2 on a high surface area carbon fiber paper demonstrated excellent total electrode activity, with a

2 79 current density of 10 mA/cm electrode achieved at 110 mV overpotential.

2.6 Comparison of MoS2 Catalysts

As discussed in Section 3, in this review we focus on two metrics for comparing the HER activity of the different molybdenum sulfide catalysts: intrinsic activity as measured by TOF per surface exposed Mo atom and total electrode activity measured by the potential to reach a

2 current density of 10 mA/cm electrode.

2.6.1 Intrinsic Activity

34 For 2H-phase MoS2 it is well established that the (0001) basal planes are catalytically inert, while only undercoordinated sulfur atoms at the edges are active. However, given the myriad of different S possible terminations in various state-of-the-art molybdenum sulfide catalysts, we

48 have found previously that calculating the TOF per surface Mo atom (as opposed to S atom) can help facilitate the direct comparison of different catalyst materials.94

The TOF plot in Figure 2.11 shows that all forms of nanostructured molybdenum sulfides exhibit high catalytic activity for the HER. On a TOF basis, the most active molybdenum sulfide

HER catalyst ever reported consists of UHV-deposited MoS2 nanoparticles on Au(111)

25 -1 substrates. The edge sites of these MoS2 nanoparticles exhibit TOFs (per edge site) of 1 s and 10 s-1 at overpotentials of approximately 0.10 V and 0.16 V, respectively.25 This sets the benchmark for HER catalysis on MoS2. For comparison, Pt catalysts have exchange current densities in the range of 0.4 – 400 mA/cm2.45, 123 These values correspond to exchange TOF

45, 123 values on the order of 1 – 1000 H2/s per site at 0 V overpotential. The TOF values for Pt reported previously vary over this wide range due to differences in the platinum preparation and measurement techniques employed.123 The extremely high activity of platinum makes it difficult of measure the TOF without influence of transport limitations, and under the carefully controlled conditions necessary to measure the TOF accurately, the Pt electrodes are typically not biased to potentials negative of -0.10 V vs. RHE.123 Therefore it is difficult to directly compare the

TOF values of Pt and MoS2 at the same overpotential. Nevertheless, these results indicate that

2 5 the intrinsic activity of Pt is approximately 10 – 10 times greater than that of MoS2 edges.

Figure 2.11. Turn over frequencies of different molybdenum sulfide catalysts normalized to the number of surface Mo atoms.

49

The intrinsic activity of other molybdenum sulfides typically falls one to two orders of magnitude below that of the UHV-prepared nanoparticles. The highest activity of a

2- molybdenum sulfide prepared by a scalable route is found for a sub-monolayer of [Mo3S13] clusters on an HOPG crystal. The high TOF recorded for these two particular systems, MoS2

2- nanoparticles on Au(111) and [Mo3S13] clusters on HOPG, may be due in large part to their submonolayer coverage, where mass transport and electrical accessibility are at a maximum.

2.6.2 Total Electrode Activity

Whereas TOF is the best figure of merit when comparing the intrinsic catalytic activity of a material, total electrode activity is useful to consider from a practical, device oriented point of view. For the HER it is useful consider the potential necessary to reach a current density of 10

2 mA/cm electrode electrode as discussed in section 2.4.1.

Figure 2.12 shows the LSV of different molybdenum sulfide materials and sample architectures.48, 52, 56, 60, 61, 67, 69, 73, 79, 94, 124 Not surprisingly, the trend in Figure 12 shows that electrodes with very high catalyst loadings generally have the highest overall activity. On a geometric area basis, electrodes with the highest demonstrated activity include: MoS2

61 60 nanoparticles on reduced graphene oxide, MoSx on piranha etched graphite paper, and

2- [Mo3S13] clusters on electrochemically anodized graphite paper (this work), all of which

2 require an overpotential of approximately 150 mV to reach 10 mA/cm electrode. Two electrodes

2 79, that each require only 110 mV overpotential to reach 10 mA/cm electrode were recently reported.

124 The first was composed of amorphous MoSx on nitrogen-doped carbon nanotubes, while the

+ second consisted of Li -intercalated 1T-phase MoS2 on carbon fiber paper. These electrode possess the highest geometric area-normalized activity reported to date. This activity is very impressive, indeed, in particular since molybdenum sulfides are earth-abundant, nonprecious metal catalysts with demonstrated stability in acid.

50

Figure 2.12. Linear sweep voltammograms demonstrating the total electrode activity different 2 molybdenum sulfide catalysts. The potential required to reach a current density of -10 mA/cm electrode is reported to the right. Polycrystalline Pt is shown for comparison.

2.7 Applications

2.7.1 Molybdenum sulfide in photoelectrodes

Molybdenum sulfide HER catalysts have been successfully applied in a number of photoelectrochemical devices. Early research focused on bulk MoS2 crystals as full photoelectrodes where the MoS2 acts as both light absorber and catalyst, but to date these devices have shown poor performance for hydrogen production.34, 125-131 A recent study suggests that MoS2 photocathodes may be inherently limited because the catalytically active edge sites may also serve as recombination centers.75 It may be possible to overcome this tradeoff with further research on surface engineering strategies.

51

Nevertheless, molybdenum sulfides have been successfully employed as the HER catalyst in combination with other semiconductor light absorbers for photocatalytic and photoelectrochemical water splitting.74, 104, 105, 107, 132-146 In particular, recent studies have shown that combining MoS2 with silicon can yield highly active complete water splitting photocathodes.104, 147 In this system, the catalyst must possess high activity for the HER while exhibiting minimal light absorption. The silicon surface must also be protected to prevent corrosion or oxidation in an aqueous electrolyte. To this end, a conformal layer of MoS2 on the Si photoelectrode surface can function as a protecting layer while simultaneously catalyzing the

HER. The flat MoS2 layers possessed only moderate HER activity because of the low density of active sites exposed at the photocathode surfaces, but the activity of the photocathodes was

2- increased by adding amorphous molybdenum sulfide or [Mo3S13] clusters to increase the HER active site density.104, 147 One drawback of these electrode designs is that the saturation photocurrent density is limited by undesired parasitic light absorption in the molybdenum sulfide layers. With further effort, molybdenum sulfide/silicon photocathodes could potentially match the performance achieved using platinum.

The fabrication, characterization, and hydrogen production performance of a MoS2-coated silicon photocathode will be discussed further in Chapter 4.

2.7.2 Acid Electrolyzers

Alkaline electrolyzers and polymer electrolyte membrane (PEM) eletrolyzers are the two main types of commercially available low temperature (< 100°C) water electrolyzers.148 Alkaline electrolyzers use nonprecious nickel-based catalysts and an aqueous KOH electrolyte with a diaphragm separating the electrode compartments.148 Commercial PEM electrolyzers use precious metal Pt and Ir catalysts with a solid proton transporting electrolyte membrane,

149 typically Nafion, between the electrodes. Though unexplored in the literature, MoS2 may have a future niche as a nonprecious alternative to Pt catalysts for PEM electrolyzer cathodes.

Alkaline electrolyzers are the industry standard for water electrolysis due to their nonprecious metal catalyst utilization and durable design.150 Though they are currently more expensive, PEM electrolyzers are superior to alkaline electrolyzers in a variety of ways which may lead to their broader implementation.19 Using an aqueous KOH electrolyte in a conventional alkaline electrolyzer means that the cell cannot be pressurized and it can slowly acidify over time due to

52 atmospheric CO2. The diaphragm that separates electrode compartments can also allow unwanted hydrogen crossover into the oxygen compartment. The combination of liquid electrolyte and diaphragm also lead to high ohmic losses which limit the maximum operating current density.

PEM electrolyzers, which typically use Nafion polymer electrolyte membranes, can be pressurized, potentially negating the need for an external compressor, have low reactant crossover, and run at higher current densities due to faster electrode kinetics and higher ion conductivity.19, 151 One of the major downsides of conventional PEM acid electrolyzers is their use of precious metal catalysts, which are the most active known acid-stable HER catalysts. Typically platinum is used as a cathode catalyst and iridium is used as an anode catalyst. Currently, the catalyst costs of PEM electrolyzers are a small fraction of the total capital costs,149 but as the prices of the other electrolyzer stack components decrease, replacing Pt will yield greater fractional cost savings. Additionally, the scarcity of Pt could hinder the scale-up of PEM electrolyzers to the TW scale; approximately 400,000 kg Pt, roughly equivalent to the entirety of global Pt production over 2 years,152 would be required to implement 1 TW of total electrolyzer capacity assuming 0.5 mg/cm2 platinum loading and operating current densities of 1 A/cm2. Developing a means to minimize or avoid Pt in PEM electrolyzers could help enable scale-up of this technology in the future. Ir also suffers from the same scarcity problems as Pt and a nonprecious metal alternative oxygen evolution catalyst must also be found to enable PEM electrolyzer scale-up.152

MoS2 is a good candidate nonprecious metal HER catalyst that could be used in PEM electrolyzers, as it is both active and stable in acid, especially if further research leads to even more active molybdenum sulfide catalysts that can compete with platinum. To our knowledge, no one has reported a true PEM electrolyzer which uses MoS2 as the cathode catalyst to evolve hydrogen. However, MoS2 catalysts loaded on gas diffusion electrodes have shown good

60 performance. New metrics will likely be necessary to evaluate MoS2 catalysts intended for use in PEM electrolyzers. The 10 mA/cm2 total electrode activity metric discussed in this work is not suitable because electrolyzers operate at approximately 1 A/cm2 current densities.19 Other performance metrics may also become more important. For example, a low Tafel slope, good catalyst adhesion to the conductive support under vigorous bubbling, and effective mass transport through the electrode structure will likely be essential for high current operation. Given the present state of high performance molybdenum sulfide catalysts, the future appears bright

53 in terms of developing gas-diffusion electrodes and water electrolyzer systems that incorporate these catalyst materials. For such studies, it is imperative that the electrodes reach current densities on the order of 1 A/cm2 in order to make for more direct comparisons to Pt-based cathodes designed for implementation in PEM electrolyzers.

2.8 Conclusions

Molybdenum sulfide is an exceptional HER catalyst if it is appropriately nanostructured to expose a high density of active edge sites. In this chapter, we described two main metrics for comparing HER catalysts, total electrode activity and intrinsic active site activity. Both are important in evaluating different catalysts to gain a fundamental understanding of the origins of catalysis as well as to compare activity among candidate materials for device integration. We made the comparisons for intrinsic activity and overall electrode activity for several classes of molybdenum sulfide catalysts, including crystalline MoS2, amorphous molybdenum sulfide films, and molecular molybdenum sulfide clusters. We described a number of approaches to engineer molybdenum sulfides to improve their intrinsic activity, pathways that may allow the material to approach the near-ideal HER activity of Pt. Finally, we described possible applications for MoS2 in water splitting, by means of water electrolyzers and photoelectrochemical (PEC) cells. Such technologies can help enable more sustainable approaches to energy production, storage and use, particularly involving intermittent renewable energy resources such as wind and solar. Given the importance of lowering the cost of such technologies to be able to compete against fossil fuels, molybdenum sulfide could play an important role as a leading scalable, nonprecious metal catalyst for hydrogen production.

Using the fundamental concepts and experimental techniques outlined here, in Chapter 3, we turn to a detailed investigation of one promising molybdenum sulfide HER catalyst, an amorphous MoSx material.

54

2.9 Copyright

Reprinted with permission from:

J.D. Benck, T.R. Hellstern, J. Kibsgaard, P. Chakthranont, and T.F. Jaramillo. "Catalyzing the Hydrogen Evolution Reaction (HER) with Molybdenum Sulfide Nanomaterials." ACS Catalysis, 2014: 3957-3971. http://dx.doi.org/10.1021/cs500923c

Copyright 2014 American Chemical Society.

2.10 Author Contributions

Thomas Hellstern and Jesse Benck contributed equally to the journal article based on this work. Jakob Kibsgaard, Pongkarn Chakthranont, and Thomas Jaramillo were also coauthors of the manuscript. All authors contributed to writing the paper.

2.11 Acknowledgments

This work was supported as part of the Center on Nanostructuring for Efficient Energy Conversion (CNEEC) at Stanford University, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DE-SC0001060. JDB and TRH acknowledge support from the National Science Foundation Graduate Research Fellowship Program. JDB and PC acknowledge Stanford Graduate Fellowships. JK acknowledges the Carlsberg Foundation for a postdoctoral fellowship and support from the US Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-SC0008685. PC was supported by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, under subcontract NFT-9-88567- 01 under prime contract no. DE-AC36-08GO28308.

55

2.12 References

1. S. Chu and A. Majumdar. "Opportunities and challenges for a sustainable energy future." Nature, 2012. 488 (7411): 294-303. http://dx.doi.org/10.1038/nature11475 2. U.S. Energy Information Administration. "International Energy Outlook 2013." 2013. http://www.eia.gov/forecasts/ieo/pdf/0484(2013).pdf 3. N.S. Lewis. "Toward cost-effective solar energy use." Science, 2007. 315 (5813): 798- 801. http://dx.doi.org/10.1126/science.1137014 4. N.L. Panwar, S.C. Kaushik, and S. Kothari. "Role of renewable energy sources in environmental protection: A review." Renewable & Sustainable Energy Reviews, 2011. 15 (3): 1513-1524. http://dx.doi.org/10.1016/j.rser.2010.11.037 5. B.D. Solomon and K. Krishna. "The coming sustainable energy transition: History, strategies, and outlook." Energy Policy, 2011. 39 (11): 7422-7431. http://dx.doi.org/10.1016/j.enpol.2011.09.009 6. K. Solangi, M. Islam, R. Saidur, N. Rahim, and H. Fayaz. "A review on global solar energy policy." Renewable & Sustainable Energy Reviews, 2011. 15 (4): 2149-2163. http://dx.doi.org/10.1016/j.rser.2011.01.007 7. K.L. Lim, H. Kazemian, Z. Yaakob, and W.W. Daud. "Solid‐state Materials and Methods for Hydrogen Storage: A Critical Review." Chemical Engineering Technology, 2010. 33 (2): 213-226. http://dx.doi.org/10.1002/ceat.200900376 8. Z. Yang, J. Zhang, M.C. Kintner-Meyer, X. Lu, D. Choi, J.P. Lemmon, and J. Liu. "Electrochemical energy storage for green grid." Chemical Reviews, 2011. 111 (5): 3577-3613. http://dx.doi.org/10.1021/cr100290v 9. M.G. Walter, E.L. Warren, J.R. McKone, S.W. Boettcher, Q.X. Mi, E.A. Santori, and N.S. Lewis. "Solar Water Splitting Cells." Chemical Reviews, 2010. 110 (11): 6446- 6473. http://dx.doi.org/10.1021/cr1002326 10. T. Abbasi and S. Abbasi. "‘Renewable’ hydrogen: Prospects and challenges." Renewable & Sustainable Energy Reviews, 2011. 15 (6): 3034-3040. http://dx.doi.org/10.1016/j.rser.2011.02.026 11. A.J. Bard and M.A. Fox. "Artificial Photosynthesis - Solar Splitting of Water to Hydrogen and Oxygen." Accounts of Chemical Research, 1995. 28 (3): 141-145. http://dx.doi.org/10.1021/ar00051a007 12. D.S. Kong, J.J. Cha, H.T. Wang, H.R. Lee, and Y. Cui. "First-row transition metal dichalcogenide catalysts for hydrogen evolution reaction." Energy & Environmental Science, 2013. 6 (12): 3553-3558. http://dx.doi.org/10.1039/c3ee42413h 13. A. Sobczynski, A. Yildiz, A.J. Bard, A. Campion, M.A. Fox, T. Mallouk, S.E. Webber, and J.M. White. "Tungsten Disulfide - a Novel Hydrogen Evolution Catalyst for Water Decomposition." Journal of Physical Chemistry, 1988. 92 (8): 2311-2315. http://dx.doi.org/10.1021/j100319a042 14. D. Voiry, H. Yamaguchi, J.W. Li, R. Silva, D.C.B. Alves, T. Fujita, M.W. Chen, T. Asefa, V.B. Shenoy, G. Eda, and M. Chhowalla. "Enhanced catalytic activity in strained chemically exfoliated WS2 nanosheets for hydrogen evolution." Nature Materials, 2013. 12 (9): 850-855. http://dx.doi.org/10.1038/nmat3700 15. J. Yang and H.S. Shin. "Recent advances in layered transition metal dichalcogenides for hydrogen evolution reaction." Journal of Materials Chemistry A, 2014. 2 (17): 5979-5985. http://dx.doi.org/10.1039/C3TA14151A 16. A.B. Laursen, S. Kegnaes, S. Dahl, and I. Chorkendorff. "Molybdenum sulfides- efficient and viable materials for electro - and photoelectrocatalytic hydrogen

56

evolution." Energy & Environmental Science, 2012. 5 (2): 5577-5591. http://dx.doi.org/10.1039/c2ee02618j 17. D. Merki and X.L. Hu. "Recent developments of molybdenum and tungsten sulfides as hydrogen evolution catalysts." Energy & Environmental Science, 2011. 4 (10): 3878-3888. http://dx.doi.org/10.1039/c1ee01970h 18. Y. Yan, B. Xia, Z. Xu, and X. Wang. "Recent Development of Molybdenum Sulfides as Advanced Electrocatalysts for Hydrogen Evolution Reaction." ACS Catalysis, 2014. 4 (6): 1693-1705. http://dx.doi.org/10.1021/cs500070x 19. M. Carmo, D.L. Fritz, J. Merge, and D. Stolten. "A comprehensive review on PEM water electrolysis." International Journal of Hydrogen Energy, 2013. 38 (12): 4901- 4934. http://dx.doi.org/10.1016/j.ijhydene.2013.01.151 20. J.D. Holladay, J. Hu, D.L. King, and Y. Wang. "An overview of hydrogen production technologies." Catalysis Today, 2009. 139 (4): 244-260. http://dx.doi.org/10.1016/j.cattod.2008.08.039 21. K. Zeng and D.K. Zhang. "Recent progress in alkaline water electrolysis for hydrogen production and applications." Progress in Energy and Combustion Science, 2010. 36 (3): 307-326. http://dx.doi.org/10.1016/j.pecs.2009.11.002 22. T. Bak, J. Nowotny, M. Rekas, and C.C. Sorrell. "Photo-electrochemical hydrogen generation from water using solar energy. Materials-related aspects." International Journal of Hydrogen Energy, 2002. 27 (10): 991-1022. http://dx.doi.org/10.1016/S0360-3199(02)00022-8 23. C.C.L. McCrory, S. Jung, J.C. Peters, and T.F. Jaramillo. "Benchmarking Heterogeneous Electrocatalysts for the Oxygen Evolution Reaction." Journal of the American Chemical Society, 2013. 135 (45): 16977-16987. http://dx.doi.org/10.1021/ja407115p 24. L.A. Kibler. "Hydrogen electrocatalysis." ChemPhysChem, 2006. 7 (5): 985-991. 25. T.F. Jaramillo, K.P. Jørgensen, J. Bonde, J.H. Nielsen, S. Horch, and I. Chorkendorff. "Identification of active edge sites for electrochemical H2 evolution from MoS2 nanocatalysts." Science, 2007. 317 (5834): 100-102. http://dx.doi.org/10.1126/science.1141483 26. R. Parsons. "The Rate of Electrolytic Hydrogen Evolution and the Heat of Adsorption of Hydrogen." Transactions of the Faraday Society, 1958. 54 (7): 1053-1063. http://dx.doi.org/10.1039/tf9585401053 27. J.K. Nørskov, T. Bligaard, A. Logadottir, J. Kitchin, J. Chen, S. Pandelov, and U. Stimming. "Trends in the exchange current for hydrogen evolution." Journal of the Electrochemical Society, 2005. 152 (3): J23-J26. http://dx.doi.org/10.1149/1.1856988 28. J. Greeley, I.E.L. Stephens, A.S. Bondarenko, T.P. Johansson, H.A. Hansen, T.F. Jaramillo, J. Rossmeisl, I. Chorkendorff, and J.K. Norskov. "Alloys of platinum and early transition metals as oxygen reduction electrocatalysts." Nature Chemistry, 2009. 1 (7): 552-556. http://dx.doi.org/10.1038/Nchem.367 29. J. Greeley, T.F. Jaramillo, J. Bonde, I.B. Chorkendorff, and J.K. Norskov. "Computational high-throughput screening of electrocatalytic materials for hydrogen evolution." Nature Materials, 2006. 5 (11): 909-913. http://dx.doi.org/10.1038/nmat1752 30. S. Trasatti. "Electrocatalysis of Hydrogen Evolution: Progress in Cathode Activation," in "Advances in Electrochemical Science and Engineering." 2008, Wiley-VCH Verlag GmbH. 1-85. http://dx.doi.org/10.1002/9783527616763.ch1 31. L. Rapoport, N. Fleischer, and R. Tenne. "Applications of WS2 (MoS2) inorganic nanotubes and fullerene-like nanoparticles for solid lubrication and for structural

57

nanocomposites." Journal of Materials Chemistry, 2005. 15 (18): 1782-1788. http://dx.doi.org/10.1039/b417488g 32. B. Radisavljevic, A. Radenovic, J. Brivio, V. Giacometti, and A. Kis. "Single-layer MoS2 transistors." Nature Nanotechnology, 2011. 6 (3): 147-150. http://dx.doi.org/10.1038/nnano.2010.279 33. G.M.K. Abotsi and A.W. Scaroni. "A Review of Carbon-Supported Hydrodesulfurization Catalysts." Fuel Processing Technology, 1989. 22 (2): 107-133. http://dx.doi.org/10.1016/0378-3820(89)90028-3 34. H. Tributsch and J.C. Bennett. "Electrochemistry and photochemistry of MoS2 layer crystals." Journal of Electroanalytical Chemistry, 1977. 81 (1): 97-111. http://dx.doi.org/10.1016/0368-1874(77)80287-6 35. B. Hinnemann, P.G. Moses, J. Bonde, K.P. Jørgensen, J.H. Nielsen, S. Horch, I. Chorkendorff, and J.K. Nørskov. "Biomimetic Hydrogen Evolution: MoS2 Nanoparticles as Catalyst for Hydrogen Evolution." Journal of the American Chemical Society, 2005. 127 (15): 5308-5309. http://dx.doi.org/10.1021/ja0504690 36. M. Daage and R.R. Chianelli. "Structure-Function Relations in Molybdenum Sulfide Catalysts - the Rim-Edge Model." Journal of Catalysis, 1994. 149 (2): 414-427. http://dx.doi.org/10.1006/jcat.1994.1308 37. W.O. Winer. "Molybdenum disulfide as a lubricant: a review of the fundamental knowledge." Wear, 1967. 10 (6): 422-452. http://dx.doi.org/10.1016/0043- 1648(67)90187-1 38. E. Benavente, M.A. Santa Ana, F. Mendizabal, and G. Gonzalez. "Intercalation chemistry of molybdenum disulfide." Coordination Chemistry Reviews, 2002. 224 (1- 2): 87-109. http://dx.doi.org/10.1016/S0010-8545(01)00392-7 39. G. Hagenbach, P. Courty, and B. Delmon. "Catalytic activity of cobalt and molybdenum sulfides in the hydrogenolysis of thiophene, hydrogenation of cyclohexene, and isomerization of cyclohexane." Journal of Catalysis, 1971. 23 (2): 295-298. http://dx.doi.org/10.1016/0021-9517(71)90052-2 40. A.B. Anderson, Z.Y. Al-Saigh, and W.K. Hall. "Hydrogen on molybdenum disulfide: theory of its heterolytic and homolytic chemisorption." Journal of Physical Chemistry, 1988. 92 (3): 803-809. http://dx.doi.org/10.1021/j100314a042 41. L.S. Byskov, M. Bollinger, J.K. Nørskov, B.S. Clausen, and H. Topsøe. "Molecular aspects of the H2 activation on MoS2 based catalysts — the role of dynamic surface arrangements." Journal of Molecular Catalysis A: Chemical, 2000. 163 (1–2): 117- 122. http://dx.doi.org/10.1016/S1381-1169(00)00404-0 42. A. Hemschemeier, S. Fouchard, L. Cournac, G. Peltier, and T. Happe. "Hydrogen production by Chlamydomonas reinhardtii: an elaborate interplay of electron sources and sinks." Planta, 2008. 227 (2): 397-407. http://dx.doi.org/10.1007/s00425-007- 0626-8 43. D. Das and T.N. Veziroǧlu. "Hydrogen production by biological processes: a survey of literature." International Journal of Hydrogen Energy, 2001. 26 (1): 13-28. http://dx.doi.org/10.1016/S0360-3199(00)00058-6 44. P.C. Hallenbeck and J.R. Benemann. "Biological hydrogen production; fundamentals and limiting processes." International Journal of Hydrogen Energy, 2002. 27 (11-12): 1185-1193. http://dx.doi.org/10.1016/S0360-3199(02)00131-3 45. N. Markovic, B. Grgur, and P. Ross. "Temperature-dependent hydrogen electrochemistry on platinum low-index single-crystal surfaces in acid solutions." Journal of Physical Chemistry B, 1997. 101 (27): 5405-5413. http://dx.doi.org/10.1021/jp970930d

58

46. N.M. Markovic. "Electrocatalysis: Interfacing electrochemistry." Nature Materials, 2013. 12 (2): 101-102. http://dx.doi.org/10.1038/nmat3554 47. C.F. Zinola, M.E. Martins, E.P. Tejera, and N.P. Neves. "Electrocatalysis: Fundamentals and Applications." International Journal of Electrochemistry, 2012. 2012. http://dx.doi.org/10.1155/2012/874687 48. J.D. Benck, Z.B. Chen, L.Y. Kuritzky, A.J. Forman, and T.F. Jaramillo. "Amorphous Molybdenum Sulfide Catalysts for Electrochemical Hydrogen Production: Insights into the Origin of their Catalytic Activity." ACS Catalysis, 2012. 2 (9): 1916-1923. http://dx.doi.org/10.1021/cs300451q 49. X.J. Bian, J. Zhu, L. Liao, M.D. Scanlon, P.Y. Ge, C. Ji, H.H. Girault, and B.H. Liu. "Nanocomposite of MoS2 on ordered mesoporous carbon nanospheres: A highly active catalyst for electrochemical hydrogen evolution." Electrochemistry Communications, 2012. 22: 128-132. http://dx.doi.org/10.1016/j.elecom.2012.06.009 50. J. Bonde, P.G. Moses, T.F. Jaramillo, J.K. Nørskov, and I. Chorkendorff. "Hydrogen evolution on nano-particulate transition metal sulfides." Faraday Discussions, 2009. 140: 219-231. http://dx.doi.org/10.1039/b803857k 51. T.Y. Chen, Y.H. Chang, C.L. Hsu, K.H. Wei, C.Y. Chiang, and L.J. Li. "Comparative study on MoS2 and WS2 for electrocatalytic water splitting." International Journal of Hydrogen Energy, 2013. 38 (28): 12302-12309. http://dx.doi.org/10.1016/j.ijhydene.2013.07.021 52. Z. Chen, D. Cummins, B.N. Reinecke, E. Clark, M.K. Sunkara, and T.F. Jaramillo. "Core–shell MoO3–MoS2 Nanowires for Hydrogen Evolution: A Functional Design for Electrocatalytic Materials." Nano Letters, 2011. 11 (10): 4168-4175. http://dx.doi.org/10.1021/nl2020476 53. T.F. Jaramillo, J. Bonde, J. Zhang, B.-L. Ooi, K. Andersson, J. Ulstrup, and I. 4+ Chorkendorff. "Hydrogen Evolution on Supported Incomplete Cubane-type [Mo3S4] Electrocatalysts." Journal of Physical Chemistry C, 2008. 112 (45): 17492-17498. http://dx.doi.org/10.1021/jp802695e 54. S.S. Ji, Z. Yang, C. Zhang, Z.Y. Liu, W.W. Tjiu, I.Y. Phang, Z. Zhang, J.S. Pan, and T.X. Liu. "Exfoliated MoS2 nanosheets as efficient catalysts for electrochemical hydrogen evolution." Electrochimica Acta, 2013. 109: 269-275. http://dx.doi.org/10.1016/j.electacta.2013.07.094 55. H.I. Karunadasa, E. Montalvo, Y.J. Sun, M. Majda, J.R. Long, and C.J. Chang. "A Molecular MoS2 Edge Site Mimic for Catalytic Hydrogen Generation." Science, 2012. 335 (6069): 698-702. http://dx.doi.org/10.1126/science.1215868 56. J. Kibsgaard, Z. Chen, B.N. Reinecke, and T.F. Jaramillo. "Engineering the surface structure of MoS2 to preferentially expose active edge sites for electrocatalysis." Nature Materials, 2012. 11 (11): 963-969. http://dx.doi.org/10.1038/nmat3439 57. D.S. Kong, H.T. Wang, J.J. Cha, M. Pasta, K.J. Koski, J. Yao, and Y. Cui. "Synthesis of MoS2 and MoSe2 Films with Vertically Aligned Layers." Nano Letters, 2013. 13 (3): 1341-1347. http://dx.doi.org/10.1021/nl400258t 58. V.W.H. Lau, A.F. Masters, A.M. Bond, and T. Maschmeyer. "Promoting the Formation of Active Sites with Ionic Liquids: A Case Study of MoS2 as Hydrogen- Evolution-Reaction Electrocatalyst." ChemCatChem, 2011. 3 (11): 1739-1742. http://dx.doi.org/10.1002/cctc.201100212 59. V.W.H. Lau, A.F. Masters, A.M. Bond, and T. Maschmeyer. "Ionic-Liquid-Mediated Active-Site Control of MoS2 for the Electrocatalytic Hydrogen Evolution Reaction." Chemistry - A European Journal, 2012. 18 (26): 8230-8239. http://dx.doi.org/10.1002/chem.201200255

59

60. A.B. Laursen, P.C.K. Vesborg, and I. Chorkendorff. "A high-porosity carbon molybdenum sulphide composite with enhanced electrochemical hydrogen evolution and stability." Chemical Communications, 2013. 49 (43): 4965-4967. http://dx.doi.org/10.1039/c3cc41945b 61. Y. Li, H. Wang, L. Xie, Y. Liang, G. Hong, and H. Dai. "MoS2 Nanoparticles Grown on Graphene: An Advanced Catalyst for the Hydrogen Evolution Reaction." Journal of the American Chemical Society, 2011. 133 (19): 7296-7299. http://dx.doi.org/10.1021/ja201269b 62. L. Liao, J. Zhu, X.J. Bian, L.N. Zhu, M.D. Scanlon, H.H. Girault, and B.H. Liu. "MoS2 Formed on Mesoporous Graphene as a Highly Active Catalyst for Hydrogen Evolution." Advanced Funcational Materials, 2013. 23 (42): 5326-5333. http://dx.doi.org/10.1002/adfm.201300318 63. T.W. Lin, C.J. Liu, and J.Y. Lin. "Facile synthesis of MoS3/carbon nanotube nanocomposite with high catalytic activity toward hydrogen evolution reaction." Applied Catalysis B, 2013. 134: 75-82. http://dx.doi.org/10.1016/j.apcatb.2013.01.004 64. Z.Y. Lu, H.C. Zhang, W. Zhu, X.Y. Yu, Y. Kuang, Z. Chang, X.D. Lei, and X.M. Sun. "In situ fabrication of porous MoS2 thin-films as high-performance catalysts for electrochemical hydrogen evolution." Chemical Communications, 2013. 49 (68): 7516-7518. http://dx.doi.org/10.1039/c3cc44143a 65. M.A. Lukowski, A.S. Daniel, F. Meng, A. Forticaux, L. Li, and S. Jin. "Enhanced hydrogen evolution catalysis from chemically exfoliated metallic MoS2 nanosheets." Journal of the American Chemical Society, 2013. 135 (28): 10274-10277. http://dx.doi.org/10.1021/ja404523s 66. X.J. Lv, G.W. She, S.X. Zhou, and Y.M. Li. "Highly efficient electrocatalytic hydrogen production by nickel promoted molybdenum sulfide microspheres catalysts." RSC Advances, 2013. 3 (44): 21231-21236. http://dx.doi.org/10.1039/c3ra42340a 67. D. Merki, S. Fierro, H. Vrubel, and X.L. Hu. "Amorphous molybdenum sulfide films as catalysts for electrochemical hydrogen production in water." Chemical Science, 2011. 2 (7): 1262-1267. http://dx.doi.org/10.1039/c1sc00117e 68. D. Merki, H. Vrubel, L. Rovelli, S. Fierro, and X. Hu. "Fe, Co, and Ni ions promote the catalytic activity of amorphous molybdenum sulfide films for hydrogen evolution." Chemical Science, 2012. 3 (8): 2515-2525. http://dx.doi.org/10.1039/C2SC20539D 69. D. Voiry, M. Salehi, R. Silva, T. Fujita, M. Chen, T. Asefa, V.B. Shenoy, G. Eda, and M. Chhowalla. "Conducting MoS2 Nanosheets as Catalysts for Hydrogen Evolution Reaction." Nano Letters, 2013. 13 (12): 6222-6227. http://dx.doi.org/10.1021/nl403661s 70. H. Vrubel, D. Merki, and X.L. Hu. "Hydrogen evolution catalyzed by MoS3 and MoS2 particles." Energy & Environmental Science, 2012. 5 (3): 6136-6144. http://dx.doi.org/10.1039/c2ee02835b 71. H. Vrubel, T. Moehl, M. Gratzel, and X.L. Hu. "Revealing and accelerating slow electron transport in amorphous molybdenum sulphide particles for hydrogen evolution reaction." Chemical Communications, 2013. 49 (79): 8985-8987. http://dx.doi.org/10.1039/c3cc45416a 72. D.Z. Wang, Z.P. Wang, C.L. Wang, P. Zhou, Z.Z. Wu, and Z.H. Liu. "Distorted MoS2 nanostructures: An efficient catalyst for the electrochemical hydrogen evolution reaction." Electrochemistry Communications, 2013. 34: 219-222. http://dx.doi.org/10.1016/j.elecom.2013.06.018

60

73. H.T. Wang, Z.Y. Lu, S.C. Xu, D.S. Kong, J.J. Cha, G.Y. Zheng, P.C. Hsu, K. Yan, D. Bradshaw, F.B. Prinz, and Y. Cui. "Electrochemical tuning of vertically aligned MoS2 nanofilms and its application in improving hydrogen evolution reaction." Proceedings of the National Academy of Sciences, 2013. 110 (49): 19701-19706. http://dx.doi.org/10.1073/pnas.1316792110 74. Y. Yu, S.-Y. Huang, Y. Li, S.N. Steinmann, W. Yang, and L. Cao. "Layer-Dependent Electrocatalysis of MoS2 for Hydrogen Evolution." Nano Letters, 2014. 14 (2): 553- 558. http://dx.doi.org/10.1021/nl403620g 75. Z. Chen, H.N. Dinh, and E. Miller. "Photoelectrochemical water splitting: standards, experimental methods, and protocols." 2013, New York, NY: Springer. 76. Z.B. Chen, T.F. Jaramillo, T.G. Deutsch, A. Kleiman-Shwarsctein, A.J. Forman, N. Gaillard, R. Garland, K. Takanabe, C. Heske, M. Sunkara, E.W. McFarland, K. Domen, E.L. Miller, J.A. Turner, and H.N. Dinh. "Accelerating materials development for photoelectrochemical hydrogen production: Standards for methods, definitions, and reporting protocols." Journal of Materials Research, 2010. 25 (1): 3- 16. http://dx.doi.org/10.1557/Jmr.2010.0020 77. B.A. Pinaud, J.D. Benck, L.C. Seitz, A.J. Forman, Z.B. Chen, T.G. Deutsch, B.D. James, K.N. Baum, G.N. Baum, S. Ardo, H.L. Wang, E. Miller, and T.F. Jaramillo. "Technical and economic feasibility of centralized facilities for solar hydrogen production via photocatalysis and photoelectrochemistry." Energy & Environmental Science, 2013. 6 (7): 1983-2002. http://dx.doi.org/10.1039/c3ee40831k 78. L.C. Seitz, Z. Chen, A.J. Forman, B.A. Pinaud, J.D. Benck, and T.F. Jaramillo. "Modeling practical performance limits of photoelectrochemical water splitting based on the current state of materials research." ChemSusChem, 2014. 7 (5): 1372-1385. http://dx.doi.org/10.1002/cssc.201301030 79. H. Wang, Z. Lu, D. Kong, J. Sun, T.M. Hymel, and Y. Cui. "Electrochemical Tuning of MoS2 Nanoparticles on Three-Dimensional Substrate for Efficient Hydrogen Evolution." ACS Nano, 2014. 8 (5): 4940-4947. http://dx.doi.org/10.1021/nn500959v 80. O. Savadogo, K. Lee, K. Oishi, S. Mitsushima, N. Kamiya, and K.-I. Ota. "New palladium alloys catalyst for the oxygen reduction reaction in an acid medium." Electrochemistry communications, 2004. 6 (2): 105-109. http://dx.doi.org/10.1016/j.elecom.2003.10.020 81. H.A. Gasteiger, S.S. Kocha, B. Sompalli, and F.T. Wagner. "Activity benchmarks and requirements for Pt, Pt-alloy, and non-Pt oxygen reduction catalysts for PEMFCs." Applied Catalysis B, 2005. 56 (1–2): 9-35. http://dx.doi.org/10.1016/j.apcatb.2004.06.021 82. S. Trasatti and O.A. Petrii. "Real surface area measurements in electrochemistry." Journal of Electroanalytical Chemistry, 1992. 327 (1–2): 353-376. http://dx.doi.org/10.1016/0022-0728(92)80162-W 83. S. Brunauer, P.H. Emmett, and E. Teller. "Adsorption of Gases in Multimolecular Layers." Journal of the American Chemical Society, 1938. 60 (2): 309-319. http://dx.doi.org/10.1021/ja01269a023 84. M. Shao, J.H. Odell, S.-I. Choi, and Y. Xia. "Electrochemical surface area measurements of platinum- and palladium-based nanoparticles." Electrochemistry Communications, 2013. 31 (0): 46-48. http://dx.doi.org/10.1016/j.elecom.2013.03.011 85. T. Hachiya, H. Honbo, and K. Itaya. "Detailed underpotential deposition of copper on gold (III) in aqueous solutions." Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, 1991. 315 (1): 275-291. http://dx.doi.org/10.1016/0022- 0728(91)80076-3

61

86. E. Herrero, L.J. Buller, and H.D. Abruna. "Underpotential deposition at single crystal surfaces of Au, Pt, Ag and other materials." Chemical Reviews, 2001. 101 (7): 1897- 1930. http://dx.doi.org/10.1021/cr9600363 87. K. Engelsmann, W. Lorenz, and E. Schmidt. "Underpotential deposition of lead on polycrystalline and single-crystal gold surfaces: Part I. Thermodynamics." Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, 1980. 114 (1): 1-10. http://dx.doi.org/10.1016/S0022-0728(80)80431-1 88. H.A. Gasteiger, N.M. Markovic, and P.N. Ross Jr. "H2 and CO electrooxidation on well-characterized Pt, Ru, and Pt-Ru. 1. Rotating disk electrode studies of the pure gases including temperature effects." Journal of Physical Chemistry, 1995. 99 (20): 8290-8301. http://dx.doi.org/10.1021/j100020a063 89. J.O.M. Bockris and E.C. Potter. "The Mechanism of the Cathodic Hydrogen Evolution Reaction." Journal of the Electrochemical Society, 1952. 99 (4): 169-186. http://dx.doi.org/10.1149/1.2779692 90. B.E. Conway, J.O. Bockris, apos, and M. "Electrolytic Hydrogen Evolution Kinetics and Its Relation to the Electronic and Adsorptive Properties of the Metal." Journal of Chemical Physics, 1957. 26 (3): 532-541. http://dx.doi.org/doi:http://dx.doi.org/10.1063/1.1743339 91. K.P. Kuhl, E.R. Cave, D.N. Abram, and T.F. Jaramillo. "New insights into the electrochemical reduction of carbon dioxide on metallic copper surfaces." Energy & Environmental Science, 2012. 5 (5): 7050-7059. http://dx.doi.org/10.1039/c2ee21234j 92. T.Y. Wang, L. Liu, Z.W. Zhu, P. Papakonstantinou, J.B. Hu, H.Y. Liu, and M.X. Li. "Enhanced electrocatalytic activity for hydrogen evolution reaction from self- assembled monodispersed molybdenum sulfide nanoparticles on an Au electrode." Energy & Environmental Science, 2013. 6 (2): 625-633. http://dx.doi.org/10.1039/c2ee23513g 93. J.C. Tokash and B.E. Logan. "Electrochemical evaluation of molybdenum disulfide as a catalyst for hydrogen evolution in microbial electrolysis cells." International Journal of Hydrogen Energy, 2011. 36 (16): 9439-9445. http://dx.doi.org/10.1016/j.ijhydene.2011.05.080 94. J. Kibsgaard, T.F. Jaramillo, and F. Besenbacher. "Building an appropriate active-site 2- motif into a hydrogen-evolution catalyst with thiomolybdate [Mo3S13] clusters." Nature Chemistry, 2014. 6 (3): 248-253. http://dx.doi.org/10.1038/nchem.1853 95. Z.Z. Wu, B.Z. Fang, Z.P. Wang, C.L. Wang, Z.H. Liu, F.Y. Liu, W. Wang, A. Alfantazi, D.Z. Wang, and D.P. Wilkinson. "MoS2 Nanosheets: A Designed Structure with High Active Site Density for the Hydrogen Evolution Reaction." ACS Catalysis, 2013. 3 (9): 2101-2107. http://dx.doi.org/10.1021/cs400384h 96. A.J. Jacobson, R.R. Chianelli, S.M. Rich, and M.S. Whittingham. "Amorphous Molybdenum Trisulfide - New Lithium Battery Cathode." Materials Research Bulletin, 1979. 14 (11): 1437-1448. http://dx.doi.org/10.1016/0025-5408(79)90087-4 97. K.S. Liang, S.P. Cramer, D.C. Johnston, C.H. Chang, A.J. Jacobson, J.P. Deneufville, and R.R. Chianelli. "Amorphous MoS3 and WS3." Journal of Non-Crystalline Solids, 1980. 42 (1-3): 345-356. http://dx.doi.org/10.1016/0022-3093(80)90035-6 98. C.H. Chang and S.S. Chan. "Infrared and Raman Studies of Amorphous MoS3 and Poorly Crystalline MoS2." Journal of Catalysis, 1981. 72 (1): 139-148. http://dx.doi.org/10.1016/0021-9517(81)90085-3 99. R.N. Bhattacharya, C.Y. Lee, F.H. Pollak, and D.M. Schleich. "Optical Study of Amorphous MoS3 - Determination of the Fundamental Energy Gap." Journal of Non- Crystalline Solids, 1987. 91 (2): 235-242. http://dx.doi.org/10.1016/s0022- 3093(87)80306-x

62

100. D. Belanger, G. Laperriere, and B. Marsan. "The Electrodeposition of Amorphous Molybdenum Sulfide." Journal of Electroanalytical Chemistry, 1993. 347 (1-2): 165- 183. http://dx.doi.org/10.1016/0022-0728(93)80086-w 101. T. Weber, J.C. Muijsers, and J.W. Niemantsverdriet. "Structure of Amorphous MoS3." Journal of Physical Chemistry, 1995. 99 (22): 9194-9200. http://dx.doi.org/10.1021/j100022a037 102. S.J. Hibble, R.I. Walton, D.M. Pickup, and A.C. Hamon. "Amorphous MoS(3): clusters or chains? The structural evidence." Journal of Non-Crystalline Solids, 1998. 232: 434-439. http://dx.doi.org/10.1016/s0022-3093(98)00393-7 103. X. Ge, L. Chen, L. Zhang, Y. Wen, A. Hirata, and M. Chen. "Nanoporous metal enhanced catalytic activities of amorphous molybdenum sulfide for high-efficiency hydrogen production." Adv Mater, 2014. 26 (19): 3100-3104. http://dx.doi.org/10.1002/adma.201305678 104. A.B. Laursen, T. Pedersen, P. Malacrida, B. Seger, O. Hansen, P.C.K. Vesborg, and I. Chorkendorff. "MoS2-an integrated protective and active layer on n(+)p-Si for solar H-2 evolution." Physical Chemistry Chemical Physics, 2013. 15 (46): 20000-20004. http://dx.doi.org/10.1039/c3cp52890a 105. B. Seger, A.B. Laursen, P.C.K. Vesborg, T. Pedersen, O. Hansen, S. Dahl, and I. Chorkendorff. "Hydrogen Production Using a Molybdenum Sulfide Catalyst on a Titanium-Protected n+p-Silicon Photocathode." Angewandte Chemie International Edition, 2012. 51 (36): 9128-9131. http://dx.doi.org/10.1002/anie.201203585 106. M.L. Tang, D.C. Grauer, B. Lassalle‐Kaiser, V.K. Yachandra, L. Amirav, J.R. Long, J. Yano, and A.P. Alivisatos. "Structural and Electronic Study of an Amorphous MoS3 Hydrogen‐Generation Catalyst on a Quantum‐Controlled Photosensitizer." Angewandte Chemie-International Edition, 2011. 123 (43): 10385-10389. http://dx.doi.org/10.1002/anie.201104412 107. Y.D. Hou, B.L. Abrams, P.C.K. Vesborg, M.E. Bjorketun, K. Herbst, L. Bech, A.M. Setti, C.D. Damsgaard, T. Pedersen, O. Hansen, J. Rossmeisl, S. Dahl, J.K. Norskov, and I. Chorkendorff. "Bioinspired molecular co-catalysts bonded to a silicon photocathode for solar hydrogen evolution." Nature Materials, 2011. 10 (6): 434-438. http://dx.doi.org/10.1038/nmat3008 108. A. Müller, E. Krickemeyer, A. Hadjikyriacou, and D. Coucouvanis. "Molybdenum- Sulfur Clusters," in "Inorganic Syntheses." A.P. Ginsberg, Editor. 2007, John Wiley & Sons, Inc. 47-51. http://dx.doi.org/10.1002/9780470132586.ch9 109. T.G. Benjamin, K. Epping-Martin, N.L. Garland, D.L. Ho, J.P. Kopasz, D.C. Papageorgopoulos, and W.F. Podolski. "US Department of Energy Polymer Electrolyte Membrane Fuel Cell Catalyst Development Activities." ECS Transactions, 2013. 50 (2): 1315-1320. http://dx.doi.org/10.1149/05002.1315ecst 110. S. Tauster, T. Pecoraro, and R. Chianelli. "Structure and properties of molybdenum sulfide: Correlation of O2 chemisorption with hydrodesulfurization activity." Journal of Catalysis, 1980. 63 (2): 515-519. http://dx.doi.org/10.1016/0021-9517(80)90109-8 111. R.R. Chianelli. "Fundamental-Studies of Transition-Metal Sulfide Hydrodesulfurization Catalysts." Catalysis Reviews-Science and Engineering, 1984. 26 (3-4): 361-393. http://dx.doi.org/10.1080/01614948408064718 112. Y. Okamoto, A. Maezawa, and T. Imanaka. "Active-Sites of Molybdenum Sulfide Catalysts Supported on Al2O3 and TiO2 for Hydrodesulfurization and Hydrogenation." Journal of Catalysis, 1989. 120 (1): 29-45. http://dx.doi.org/10.1016/0021-9517(89)90248-0 113. S. Bouwens, J. Van Veen, D. Koningsberger, V. De Beer, and R. Prins. "Extended X- Ray absorption fine structure determination of the structure of cobalt in

63

carbonsupported Co and Co-Mo sulfide hydrodesulfurization catalysts." Journal of Physical Chemistry, 1991. 95: 123-134. http://dx.doi.org/10.1021/j100154a028 114. J. Lauritsen, J. Kibsgaard, G. Olesen, P. Moses, B. Hinnemann, S. Helveg, J. Norskov, B. Clausen, H. Topsoe, and E. Lagsgaard. "Location and coordination of promoter atoms in Co- and Ni-promoted MoS2-based hydrotreating catalysts." Journal of Catalysis, 2007. 249 (2): 220-233. http://dx.doi.org/10.1016/j.jcat.2007.04.013 115. L.S. Byskov, B. Hammer, J.K. Nørskov, B.S. Clausen, and H. Topsøe. "Sulfur bonding in MoS2 and Co-Mo-S structures." Catalysis Letters, 1997. 47 (3-4): 177- 182. http://dx.doi.org/10.1023/A:1019009105792 116. F. Besenbacher, M. Brorson, B.S. Clausen, S. Helveg, B. Hinnemann, J. Kibsgaard, J.V. Lauritsen, P.G. Moses, J.K. Nørskov, and H. Topsøe. "Recent STM, DFT and HAADF-STEM studies of sulfide-based hydrotreating catalysts: Insight into mechanistic, structural and particle size effects." Catalysis Today, 2008. 130 (1): 86- 96. http://dx.doi.org/10.1016/j.cattod.2007.08.009 117. J. Lauritsen. "Atomic-Scale Structure of Co–Mo–S Nanoclusters in Hydrotreating Catalysts." Journal of Catalysis, 2001. 197 (1): 1-5. http://dx.doi.org/10.1006/jcat.2000.3088 118. J. Kibsgaard, A. Tuxen, K.G. Knudsen, M. Brorson, H. Topsøe, E. Lægsgaard, J.V. Lauritsen, and F. Besenbacher. "Comparative atomic-scale analysis of promotional effects by late 3d-transition metals in MoS2 hydrotreating catalysts." Journal of Catalysis, 2010. 272 (2): 195-203. http://dx.doi.org/10.1016/j.jcat.2010.03.018 119. H.T. Wang, D.S. Kong, P. Johanes, J.J. Cha, G.Y. Zheng, K. Yan, N.A. Liu, and Y. Cui. "MoSe2 and WSe2 Nanofilms with Vertically Aligned Molecular Layers on Curved and Rough Surfaces." Nano Letters, 2013. 13 (7): 3426-3433. http://dx.doi.org/10.1021/Nl401944f 120. X. Sun, J. Dai, Y.Q. Guo, C.Z. Wu, F.T. Hu, J.Y. Zhao, X.C. Zeng, and Y. Xie. "Semimetallic molybdenum disulfide ultrathin nanosheets as an efficient electrocatalyst for hydrogen evolution." Nanoscale, 2014. 6 (14): 8359-8367. http://dx.doi.org/10.1039/c4nr01894j 121. C. Tsai, F. Abild-Pedersen, and J.K. Nørskov. "Tuning the MoS2 Edge-site Activity for Hydrogen Evolution via Support Interactions." Nano Letters, 2014. http://dx.doi.org/10.1021/nl404444k 122. W. Chen, E.J.G. Santos, W.G. Zhu, E. Kaxiras, and Z.Y. Zhang. "Tuning the Electronic and Chemical Properties of Monolayer MoS2 Adsorbed on Transition Metal Substrates." Nano Letters, 2013. 13 (2): 509-514. http://dx.doi.org/10.1021/nl303909f 123. W. Sheng, H.A. Gasteiger, and Y. Shao-Horn. "Hydrogen Oxidation and Evolution Reaction Kinetics on Platinum: Acid vs Alkaline Electrolytes." Journal of the Electrochemical Society, 2010. 157 (11): B1529-B1536. http://dx.doi.org/10.1149/1.3483106 124. D.J. Li, U.N. Maiti, J. Lim, D.S. Choi, W.J. Lee, Y. Oh, G.Y. Lee, and S.O. Kim. "Molybdenum Sulfide/N-Doped CNT Forest Hybrid Catalysts for High-Performance Hydrogen Evolution Reaction." Nano Letters, 2014. 14 (3): 1228-1233. http://dx.doi.org/10.1021/nl404108a 125. A. Fujishima, Y. Noguchi, K. Honda, and B.H. Loo. "Photoelectrochemical Studies of MoS2 Electrode by Rotating Ring-disk Electrode Technique." Bulletin of the Chemical Society of Japan, 1982. 55 (1): 17-22. http://dx.doi.org/10.1246/bcsj.55.17 126. J. Gobrecht, H. Tributsch, and H. Gerischer. "Performance of Synthetical n-MoSe2 in Electrochemical Solar Cells." Journal of the Electrochemical Society, 1978. 125 (12): 2085-2086. http://dx.doi.org/10.1149/1.2131370

64

127. W. Kautek and H. Gerischer. "Photoelectrochemical Reactions and Formation of Inversion Layers at n-Type MoS2-, MoSe2-, and WSe2-Electrodes in Aprotic solvents." Berichte Der Bunsen-Gesellschaft-Physical Chemistry Chemical Physics, 1980. 84 (7): 645-653. http://dx.doi.org/10.1002/bbpc.19800840708 128. T. Kiesewetter, Y. Tomm, M. Turrion, and H. Tributsch. "Composite materials for photovoltaics: A realistic aim?" Solar Energy Materials and Solar Cells, 1999. 59 (4): 309-323. http://dx.doi.org/10.1016/s0927-0248(99)00046-x 129. C.P. Kubiak, L.F. Schneemeyer, and M.S. Wrighton. "Visible Light Driven Generation of Chlorine and Bromine. Photooxidation of Chloride and Bromide in Aqueous Solution at Illuminated n-Type Semiconducting Molybdenum Diselenide and Molybdenum Disulfide Electrodes." Journal of the American Chemical Society, 1980. 102 (22): 6898-6900. http://dx.doi.org/10.1021/ja00542a063 130. L.F. Schneemeyer and M.S. Wrighton. "Flat-Band Potential of n-Type Semiconducting Molybdenum Disulfide by Cyclic Voltammetry of Two-Electron Reductants: Interface Energetics and the Sustained Photooxidation of Chloride." Journal of the American Chemical Society, 1979. 101 (22): 6496-6500. http://dx.doi.org/10.1021/ja00516a002 131. J.P. Wilcoxon and G.A. Samara. "Strong quantum-size effects in a layered semiconductor: MoS2 nanoclusters." Physical Review B, 1995. 51 (11): 7299-7302. http://dx.doi.org/10.1103/PhysRevB.51.7299 132. F.A. Frame and F.E. Osterloh. "CdSe-MoS2: A Quantum Size-Confined Photocatalyst for Hydrogen Evolution from Water under Visible Light." Journal of Physical Chemistry C, 2010. 114 (23): 10628-10633. http://dx.doi.org/10.1021/jp101308e 133. Z.P. Huang, C.F. Wang, L. Pan, F. Tian, X.X. Zhang, and C. Zhang. "Enhanced photoelectrochemical hydrogen production using silicon nanowires@MoS3." Nano Energy, 2013. 2 (6): 1337-1346. http://dx.doi.org/10.1016/j.nanoen.2013.06.016 134. H.W. King. "Crystal Structures and Lattice Parameters of Allotropes of the Elements," in "CRC handbook of chemistry and physics : a ready-reference book of chemical and physical data." W.M. Haynes, Editor. 2013, CRC Press: Boca Raton, FL. 135. Y. Liu, Y.X. Yu, and W.D. Zhang. "MoS2/CdS Heterojunction with High Photoelectrochemical Activity for H-2 Evolution under Visible Light: The Role of MoS2." Journal of Physical Chemistry C, 2013. 117 (25): 12949-12957. http://dx.doi.org/10.1021/jp4009652 136. U. Maitra, U. Gupta, M. De, R. Datta, A. Govindaraj, and C.N.R. Rao. "Highly Effective Visible-Light-Induced H-2 Generation by Single-Layer 1T-MoS2 and a Nanocomposite of Few-Layer 2H-MoS2 with Heavily Nitrogenated Graphene." Angewandte Chemie International Edition, 2013. 52 (49): 13057-13061. http://dx.doi.org/10.1002/anie.201306918 137. L. Wei, Y.J. Chen, Y.P. Lin, H.S. Wu, R.S. Yuan, and Z.H. Li. "MoS2 as non-noble- metal co-catalyst for photocatalytic hydrogen evolution over hexagonal ZnIn2S4 under visible light irradiations." Applied Catalysis B, 2014. 144: 521-527. http://dx.doi.org/10.1016/j.apcatb.2013.07.064 138. Q.J. Xiang, J.G. Yu, and M. Jaroniec. "Synergetic Effect of MoS2 and Graphene as Cocatalysts for Enhanced Photocatalytic H-2 Production Activity of TiO2 Nanoparticles." Journal of the American Chemical Society, 2012. 134 (15): 6575- 6578. http://dx.doi.org/10.1021/ja302846n 139. X. Zong, G.P. Wu, H.J. Yan, G.J. Ma, J.Y. Shi, F.Y. Wen, L. Wang, and C. Li. "Photocatalytic H-2 Evolution on MoS2/CdS Catalysts under Visible Light Irradiation." Journal of Physical Chemistry C, 2010. 114 (4): 1963-1968. http://dx.doi.org/10.1021/jp904350e

65

140. X. Zong, H. Yan, G. Wu, G. Ma, F. Wen, L. Wang, and C. Li. "Enhancement of Photocatalytic H2 Evolution on CdS by Loading MoS2 as Cocatalyst under Visible Light Irradiation." Journal of the American Chemical Society, 2008. 130 (23): 7176- 7177. http://dx.doi.org/10.1021/ja8007825 141. Q. Xiang, J. Yu, and M. Jaroniec. "Synergetic effect of MoS2 and graphene as cocatalysts for enhanced photocatalytic H2 production activity of TiO2 nanoparticles." Journal of the American Chemical Society, 2012. 134 (15): 6575- 6578. http://dx.doi.org/10.1021/ja302846n 142. P.D. Tran, S.S. Pramana, V.S. Kale, M. Nguyen, S.Y. Chiam, S.K. Batabyal, L.H. Wong, J. Barber, and J. Loo. "Novel assembly of an MOS2 electrocatalyst onto a silicon nanowire array electrode to construct a photocathode composed of elements abundant on the Earth for hydrogen generation." Chemistry - A European Journal, 2012. 18 (44): 13994-13999. http://dx.doi.org/10.1002/chem.201202214 143. L. Gao, Y. Cui, J. Wang, A. Cavalli, A. Standing, T.T. Vu, M.A. Verheijen, J.E. Haverkort, E.P. Bakkers, and P.H. Notten. "Photoelectrochemical hydrogen production on InP nanowire arrays with molybdenum sulfide electrocatalysts." Nano Letters, 2014. 14 (7): 3715-3719. http://dx.doi.org/10.1021/nl404540f 144. Q. Ding, F. Meng, C.R. English, M. Cabán-Acevedo, M.J. Shearer, D. Liang, A.S. Daniel, R.J. Hamers, and S. Jin. "Efficient Photoelectrochemical Hydrogen Generation Using Heterostructures of Si and Chemically Exfoliated Metallic MoS2." Journal of the American Chemical Society, 2014. 136 (24): 8504-8507. http://dx.doi.org/10.1021/ja5025673 145. Y.-F. Zhao, Z.-Y. Yang, Y.-X. Zhang, L. Jing, X. Guo, Z. Ke, P. Hu, G. Wang, Y.-M. Yan, and K.-N. Sun. "Cu2O Decorated with Cocatalyst MoS2 for Solar Hydrogen Production with Enhanced Efficiency under Visible Light." Journal of Physical Chemistry C, 2014. 26: 14238–14245. http://dx.doi.org/10.1021/jp504005x 146. K. Chang, Z. Mei, T. Wang, Q. Kang, S. Ouyang, and J. Ye. "MoS2/Graphene Cocatalyst for Efficient Photocatalytic H2 Evolution under Visible Light Irradiation." ACS Nano, 2014. 8: 7078–7087. http://dx.doi.org/10.1021/nn5019945 147. J.D. Benck, S.C. Lee, K.D. Fong, J. Kibsgaard, R. Sinclair, and T.F. Jaramillo. "Designing Active and Stable Silicon Photocathodes for Solar Hydrogen Production Using Molybdenum Sulfide Nanomaterials." Advanced Energy Materials, 2014. 4 (18). http://dx.doi.org/10.1002/aenm.201400739 148. M. Pagliaro and A.G. Konstandopoulos. "Solar Hydrogen: Fuel of the Future." 2012: Royal Society of Chemistry. 149. K.E. Ayers, E.B. Anderson, C. Capuano, B. Carter, L. Dalton, G. Hanlon, J. Manco, and M. Niedzwiecki. "Research advances towards low cost, high efficiency PEM electrolysis." ECS Transactions, 2010. 33 (1): 3-15. http://dx.doi.org/10.1149/1.3484496 150. J. Ivy. "Summary of electrolytic hydrogen production: milestone completion report." 2004. http://energy.gov/eere/fuelcells/downloads/summary-electrolytic-hydrogen- production-milestone-completion-report 151. A. Goni-Urtiaga, D. Presvytes, and K. Scott. "Solid acids as electrolyte materials for proton exchange membrane (PEM) electrolysis: Review." International Journal of Hydrogen Energy, 2012. 37 (4): 3358-3372. http://dx.doi.org/10.1016/j.ijhydene.2011.09.152 152. P.C.K. Vesborg and T.F. Jaramillo. "Addressing the terawatt challenge: scalability in the supply of chemical elements for renewable energy." RSC Advances, 2012. 2 (21): 7933-7947. http://dx.doi.org/10.1039/c2ra20839c

66

Chapter 3: Amorphous Molybdenum Sulfide Catalysts for Electrochemical Hydrogen Production: Insights into the Origin of their Catalytic Activity

3.1 Abstract

We present a scalable wet chemical synthesis for a catalytically active nanostructured amorphous molybdenum sulfide material. The catalyst film is one of the most active non- precious metal materials for electrochemical hydrogen evolution, drawing 10 mA/cm2 at ~200 mV overpotential. To identify the active phase of the material, we perform x-ray photoelectron spectroscopy after testing under a variety of conditions. As deposited, the catalyst resembles amorphous MoS3, but domains resembling MoS2 in composition and chemical state are created under reaction conditions and may contribute to this material’s high electrochemical activity. The activity scales with electrochemically active surface area, suggesting that the rough, nanostructured catalyst morphology also contributes substantially to the film’s high activity. Electrochemical stability tests indicate that the catalyst remains highly active throughout prolonged operation. The overpotential required to attain a current density of 10 mA/cm2 increases by only 57 mV after 10,000 reductive potential cycles. Our enhanced understanding of this highly active amorphous molybdenum sulfide hydrogen evolution catalyst may facilitate the development of economical electrochemical hydrogen production systems.

3.2 Introduction

Hydrogen, mainly produced today from natural gas or coal, is an extremely important chemical utilized at the global scale for major industrial processes including petroleum refining and ammonia synthesis.1, 2 Hydrogen has also been proposed as a future energy carrier that could be used to power electronic devices, vehicles, and homes.3, 4 Developing methods to economically produce hydrogen from renewable energy resources could lead to substantial reductions in fossil

67 fuel consumption and lower global CO2 emissions. The uses of H2 and the potential impacts of renewable H2 production methods were explained more extensively Chapter 1.

A great deal of attention has recently been devoted to producing hydrogen from water through sustainable electrochemical processes such as photoelectrochemical water splitting or electrolysis driven by an external renewable source of electricity.4-7 The hydrogen evolution reaction (HER), a fundamental part of the water splitting process, requires the use of a catalyst to minimize losses due to kinetic overpotential, as discussed at length in Chapter 2.5, 8

The best known catalysts for the HER are precious metals such as platinum, ruthenium, and iridium, but the scarcity and high cost of these materials prohibit their wide-scale deployment.5, 9-11 Nickel alloy catalysts show high activity for the HER in alkaline electrolytes, but often degrade in acidic solutions.12-19 The development of a scalable, environmentally friendly synthesis for an inexpensive, highly active, acid-stable HER catalyst remains a major challenge.5, 20-23

A variety of molybdenum sulfide materials have shown high activity and stability for the HER in acidic environments.24-31 Many of these developments were reviewed in Chapter 2. One limitation of many previous crystalline MoS2 catalysts is that the procedures for synthesizing these materials have typically involved ultra-high vacuum processing,24 high temperature

27, 28 24, 27 29 treatment, sulfidization using H2S gas, or electrodeposition, which could limit the range of potential applications. Recent work has revealed scalable procedures for synthesizing highly active amorphous molybdenum sulfide catalysts.11, 28, 32, 33 However, much remains to be learned about the properties of amorphous molybdenum sulfide and the origins of its catalytic activity.

In this chapter, we report a facile wet chemical synthesis for a nanostructured amorphous molybdenum sulfide catalyst and aim to understand the origin of its catalytic activity. The synthesis technique is based on readily available precursors, avoids the need for high vacuum processing, high temperature treatment, or a separate sulfidization step, and enables straightforward catalyst deposition onto a wide range of substrates. Physical and chemical characterization suggests that the resulting catalyst is similar to other recently reported amorphous MoS3 materials in structure and composition, but this material has higher activity than most similar materials. To understand the origin of this material’s activity, we investigated changes in the catalyst morphology, composition, and activity during catalysis. These results

68 suggest that the high activity of this material likely arises from both the inherently favorable surface properties of the active molybdenum sulfide phase and the rough, nanostructured catalyst film morphology. Having enhanced the understanding of this material’s properties, we propose strategies for further improving its performance.

3.3 Methods

3.3.1 Catalyst Synthesis

The molybdenum sulfide catalyst was prepared via a room temperature wet chemical synthesis. All reagents were purchased and used as-received. A molybdenum precursor solution was prepared by adding 0.60 g ammonium heptamolybdate (Mo7O24(NH4)6·4H2O, Sigma-Aldrich) to 12 mL 0.2 M sulfuric acid (H2SO4, Aldrich) in Millipore water (18 MΩ cm). In a second container, a sulfur precursor solution was prepared by adding 0.075 g sodium sulfide (Na2S, Alfa Aesar) to 12 mL Millipore water. These two solutions were mixed, yielding a 24 mL solution of 0.02 M ammonium heptamolybdate, 0.04 M sodium sulfide, and 0.1 M sulfuric acid in water. Upon mixing the solutions, a suspension of nanoparticles was immediately formed. The suspension was centrifuged at 12,000 rpm (relative centrifugal force 17,400) for 30 minutes to collect the nanoparticles. After discarding the supernatant, the solid centrifuge pellet was rinsed once with 15 mL isopropanol and the rinsing liquid was discarded. The pellet was then redispersed in an additional 15 mL isopropanol via sonication for 10 minutes to give the final catalyst solution used for drop casting.

The catalyst samples with typical mass loading used for the scanning electron microscopy, x- ray photoelectron spectroscopy, Raman spectroscopy, and electrochemical testing were prepared by drop casting 10 μL of the redispersed suspension onto 5 mm diameter glassy carbon disks (Sigradur G, HTW Hochtemperatur-Werkstoffe GmbH). The sample used for x-ray diffraction analysis was prepared by drop casting 2 mL of the redispersed suspension onto a 50  75 mm glass microscope slide. The sample used for transmission electron microscopy was prepared by diluting the sonicated nanoparticle suspension by 400x in isopropanol and drop casting 2 μL of this mixture onto a holey carbon support (Ted Pella #01824). A series of samples with mass loadings ranging from 400% to 25% of the typical preparation were synthesized by redispersing the solid pellet formed after centrifugation in 3.75 mL isopropanol and using a 1:1

69 dilution series in isopropanol to make five solutions with 400%, 200%, 100%, 50%, and 25% of the typical catalyst concentration. A 10 μL aliquot of each solution was then deposited onto a 5 mm glassy carbon disk. All samples were dried under low vacuum (~25 torr) for several hours prior to characterization.

3.3.2 Physical and Chemical Characterization

Scanning electron microscopy (SEM) was performed before and after electrochemical characterization using a FEI Magellan XHR microscope operated with a beam voltage of 5.0 kV and current of 25 pA. Transmission electron microscopy (TEM) of the catalyst was collected using a FEI Titan 80-300 microscope operated at 300 kV.

X-ray diffraction (XRD) spectra were collected using a PANalytical X’Pert Pro diffractometer with a CuKα x-ray source operated at 40 kV and 45 mA. X-ray photoelectron spectroscopy (XPS) was performed using a Phi VersaProbe spectrometer with an AlKα source and binding energies were calibrated to the adventitious C 1s peak at 284.6 eV. Raman spectra were collected using a Cobolt Blues 473 nm blue diode pumped solid state laser at approximately 1 mW and an Andor charge-coupled detector held at -50 °C with a thermoelectric cooling plate.

Inductively coupled plasma – optical emission spectroscopy (ICP-OES) measurements were attempted to determine the bulk composition of the molybdenum sulfide catalyst. However, these measurements were unsuccessful due to the difficulty of obtaining a physically meaningful and reproducible measurement of the sulfur signal.

3.3.3 Electrochemical Characterization

Electrochemical measurements were performed in a three electrode electrochemical cell in a rotating disk electrode (RDE) configuration with a Bio-Logic potentiostat (VMP3). All tests were performed in 80 to 100 mL of 0.5 M sulfuric acid (H2SO4) electrolyte prepared in Millipore water (18 MΩ cm) with a Hg/Hg2SO4 in saturated K2SO4 reference electrode (Hach) and a graphite rod counter electrode (Ted Pella). The molybdenum sulfide catalyst film on a glassy carbon disk was used as the working electrode. The reversible hydrogen electrode (RHE) was calibrated to between -0.700 V and -0.706 V vs. the Hg/Hg2SO4 reference electrode as measured using platinum mesh working and counter electrodes under a H2 gas purge. The potential scale

70 was calibrated to RHE after each experiment to prevent contamination of the cell with platinum ions dissolved from the electrodes during this calibration step.

The series resistance through the electrochemical cell was determined before each activity measurement using electrochemical impedance. The measured series resistance was approximately 6 Ω on both the catalyst samples and a bare glassy carbon substrate, which suggests that the resistance contributed by the molybdenum sulfide film was small. After the testing, the data were corrected to remove the potential drop due to the series resistance.

Prior to further characterization, the catalyst was electrochemically activated by performing approximately three potential cycles between 0.10 V and -0.25 V vs. RHE at 20 mV/s. After the initial stabilization of the activity, the performance of the catalyst for the hydrogen evolution reaction was measured using a linear sweep voltammogram beginning at the open circuit potential of approximately 0.30 V vs. RHE and ending at -0.30 V vs. RHE with a scan rate of 5 mV/s. Throughout these measurements, the electrolyte was purged with H2 gas and the working electrode was rotated at 1600 rpm in the RDE to remove hydrogen gas bubbles formed at the catalyst surface.

To measure electrochemical capacitance, the potential was swept from 0.15 V to 0.35 V and back to 0.15 V three times at each of six different scan rates (10 mV/s, 20 mV/s, 40 mV/s, 80 mV/s, 160 mV/s, and 320 mV/s). The working electrode was held at each potential vertex for 20 s before beginning the next sweep. These scans were performed on the molybdenum sulfide catalyst while purging with N2 gas and rotating at 1600 rpm in the RDE. The rate of rotation was observed to have no effect on the measured capacitive current. The same set of scans was also performed on a flat standard MoS2 sample in a compression cell with a N2 gas purge and no rotation.

The flat standard sample of MoS2 was prepared by sputter-depositing approximately 50 nm molybdenum with a 10 nm titanium adhesion layer on a highly doped, conductive p-type silicon wafer. This sample was sulfidized at 300 °C for 60 min in a 90% H2/10% H2S atmosphere. The chemical composition of the surface was confirmed to be MoS2 by XPS.

Extended potential cycling was performed in order to investigate changes in composition during catalysis and to evaluate the catalyst’s durability. These tests were performed by taking linear sweep voltammograms after repeatedly cycling the potential between 0.10 V and -0.25 V vs.

71

RHE at 50 mV/s. The first 20 potential cycles were performed at 0 rpm and the next 5 cycles were performed at 1600 rpm. This pattern was repeated three additional times. Performing the majority of the cycles at 0 rpm prevented the RDE apparatus from overheating due to prolonged continuous rotation, while periodic rotation still enabled removal of H2 bubbles. After every 100 potential cycles, the potential was swept from 0.10 V to -0.30 V vs. RHE at 5 mV/s and 1600 rpm. The slow scans were used to minimize contributions from capacitance current and obtain a more accurate measure of the electrochemical activity. For electrochemical stability testing, this program was repeated 100 times for 10,000 total potential cycles at 50 mV/s (approximately 48 hours). The samples prepared for x-ray photoelectron spectroscopy measurements were cycled using this program for 4 hours. A continual H2 gas purge was used for the duration of these tests. As appropriate, these samples were rinsed in Millipore water to remove any residual sulfuric acid prior to spectroscopic measurements.

3.4 Results and Discussion

3.4.1 Physical and Chemical Characterization

The morphology of the catalyst film was characterized using scanning electron microscopy as shown in Figure 3.1. The film has a rough, nanostructured surface with feature sizes ranging from 50 – 100 nm. This structure arises from the agglomeration of the catalyst nanoparticles upon drop casting and is favorable for catalysis because it provides a high surface area interface between the catalyst and the electrolyte, which is conducive to high overall electrochemical activity. An image of a portion of the catalyst film that delaminated from the substrate shows that the film is approximately 5 - 7 μm thick assuming an observation angle between 0° and 45° with respect to the film surface. The amount of catalyst on the substrate is greater than the typical loading for precious metal electrocatalysts, which are often deposited as isolated nanoparticles rather than as a porous film.5, 34, 35 However, due to the low material cost and straightforward synthetic procedure of the molybdenum sulfide catalyst, this film structure could potentially be practical for wide-scale deployment.

72

Figure 3.1. SEM images showing (a) the catalyst film surface morphology and (b) a thickness cross section of a segment of the catalyst film that delaminated from the substrate surface.

The crystallinity of the film was investigated before electrochemical characterization using transmission electron microscopy (TEM) and x-ray diffraction (XRD). The TEM image of an isolated nanoparticle shown in Figure 3.2 indicates that the diameter of the catalyst particle is approximately 60 nm, which corresponds to the size of the features observed using SEM after the particles agglomerate into the rough, nanostructured film morphology. The absence of lattice or Moiré fringes observed in the bright field TEM image suggests that the material is amorphous, which was further confirmed by the lack of any peaks that would suggest a crystalline catalyst film in the XRD spectrum in Figure 3.3. As shown in Figure 3.4, Raman

1 spectroscopy revealed the absence of the distinctive A1g and E 2g vibrational peaks expected for crystalline MoS2, which suggests that crystalline MoS2 is not present in significant quantities in the as-deposited catalyst.36-38

73

Figure 3.2. Transmission electron micrograph of an isolated particle deposited from a diluted molybdenum sulfide catalyst suspension.

Figure 3.3. X-ray diffraction spectra of a blank glass substrate (red trace), the molybdenum sulfide catalyst film as deposited (blue trace), and the catalyst film after sulfidization for 30 minutes at 500°C in 10% H2/90% H2S (green trace). (a) Wide scan range, (b) high resolution scan. The peak at 13.9° in the sulfidized material is attributed to crystalline MoS2. The absence of a peak at this position for the as- deposited sample suggests that this material does not possess the layered structure of MoS2.

74

Figure 3.4. Raman spectra of a MoS2 crystal standard (red trace), the as-deposited molybdenum sulfide catalyst film (blue trace), the molybdenum sulfide catalyst after 10 minutes of reductive potential cycling (green trace), the molybdenum sulfide catalyst after 4 hours of reductive potential cycling (purple trace), and a blank glassy carbon substrate (black trace). The catalyst does not exhibit the characteristic peaks 36-38 that are observed for crystalline MoS2.

The x-ray photoelectron spectra collected before electrochemical testing in Figure 3.5 show that the as-deposited ("no testing") material resembles molybdenum trisulfide (MoS3). The sulfur 2p region shows a peak at a binding energy of 163.0 eV with a broad full width at half maximum (FWHM) of 2.2 eV and no evidence of the expected spin-orbit splitting doublet corresponding to the 2p1/2 and 2p3/2 lines, which suggests that the sulfur atoms near the surface exist in multiple oxidation states. These observations are consistent with previous reports of MoS3 materials,

2- 2- 11, 29, 33, 39, 40 which contain sulfur in a combination of S2 and S groups. A small peak observed near 169 eV corresponds to the binding energy of sulfur in a sulfate group41, 42 and likely arises

2- from residual SO4 from the sulfuric acid used in the catalyst synthesis. The presence of small

+ + nitrogen 1s and sodium 1s peaks indicates that there may also be some residual Na and NH4 from the catalyst synthesis, but these peaks disappear after the catalyst has been immersed in water or sulfuric acid, as shown in Figure 3.6.

75

Figure 3.5. X-ray photoelectron spectra of the catalyst surface before testing, after one reductive potential cycle, and after four hours of continuous reductive cycling. The data are scaled to normalize the Mo 3d peak areas. Before electrochemical testing, the shape and binding energy of the S 2p peak are indicative of amorphous MoS3. The Mo 3d peaks show that the majority of the molybdenum occurs in chemical (A) state Mo Sx. The peak positions and shapes begin to change after the initial reductive potential cycle, corresponding to the catalyst activation observed during electrochemical testing. These changes are accentuated after extended electrochemical testing. In the tested samples, the S 2p peak shape and binding energy more closely resemble MoS2, and the majority of the molybdenum occurs in chemical state (B) Mo Sx, with binding energies near the Mo 4+ peaks observed in MoS2.

76

Figure 3.6. X-ray photoelectron spectra of the Mo 3p/N 1s and Na 1s regions of the catalyst as deposited and after resting in H2O or 0.5 M H2SO4 with no applied potential. As deposited, the catalyst displays + + small N 1s (a) and Na 1s (b) peaks, which are attributed to residual NH4 and Na ions from the catalyst synthesis. Elemental quantification suggests that the concentration of sodium ions in the catalyst material is less than 1% by mole. Quantification of the ammonium ion content is more difficult due to convolution + with the Mo 3p peaks, but we conservatively estimate that the NH4 concentration is no greater than 5% by mole. After resting in H2O or 0.5 M H2SO4 for one hour (c, d, e, and f), the N 1s and Na 1s lines disappear, which shows that these residual ions dissolve into the liquid electrolyte. We therefore assume that these ions do not contribute to the reductive currents observed during the catalyst activation or hydrogen evolution.

77

In Figure 3.5, the molybdenum 3d region reveals a mixture of Mo oxidation states. Peak fitting shows that approximately 27% of the Mo signal corresponds to a 6+ oxidation state. XPS spectra of an untested catalyst sample aged in air show that this feature arises primarily from oxidation

(6+) to form a thin layer of Mo O3, as shown in Figure 3.7. The remaining 73% of the Mo signal

(A) arises from the molybdenum sulfide. This Mo exists in a chemical state labeled here as Mo Sx, which likely correspond to a lower oxidation state of either 4+ or 5+. The amorphous structure of this material prohibits the definitive determination of the formal oxidation state, as the observed binding energies are likely influenced by a number of factors including the oxidation state, sulfur coordination, and local bonding structure. In order to quantify the Mo and S components of the molybdenum sulfide film, the peak areas pertaining to the Mo 3d and the sulfide/persulfide S 2p lines were measured and calibrated versus a reference scan of a natural

MoS2 crystal shown in Figure 3.8, which served as a standard. This analysis reveals that the composition near the catalyst surface (~3 nm probe depth) is 22 % Mo and 78 % S as- synthesized (disregarding the S pertaining to sulfate as well as all other elements). These values closely match the 25% Mo and 75 % S expected for molybdenum trisulfide. Thus, the atomic composition and binding energies measured by XPS and the amorphous structure observed by TEM and XRD suggest that the as-deposited material is predominantly composed of amorphous

39, 40, 43-52 MoS3.

78

Figure 3.7. X-ray photoelectron spectra of the S 2p and Mo 3d regions of an untested catalyst sample 3, 12, and 19 days after synthesis. The increase in size of the Mo 6+ features over time (b, d, and f) suggests that the surface of the molybdenum sulfide catalyst forms a native surface oxide upon exposure to air. A ~25% decrease in the relative intensity of the S 2p peak is also observed (a, c, and e). We attribute this change to the attenuation of the sulfur signal due to surface coverage by a thin oxide. Aside from these changes, the Mo 3d and S 2p line shapes and peak positions remain very close to their original values, which suggests that the chemical state of the molybdenum sulfide phase does not change upon exposure to air.

79

Figure 3.8. X-ray photoelectron spectra of the Mo 3d and S 2p regions of the natural crystalline MoS2 standard. Mo 3d region (a) and S 2p region (b).

3.4.2 Catalyst Activation and Hydrogen Evolution Activity

We investigated the electrochemical properties of the molybdenum sulfide catalyst using a three-electrode electrochemical cell in a rotating disk electrode configuration. Catalyst activation is observed upon the first reductive potential cycle, as shown in Figure 3.9. Initially the onset of reductive current is observed near -200 mV vs. RHE, but after completing the first potential cycle, the onset of reductive current shifts to approximately -150 mV vs. RHE and remains stable in subsequent cycles. The current observed on the first reductive cycle is likely a combination of hydrogen evolution and electrochemical reduction of the catalyst material to form the catalytically active phase, the nature of which is discussed below. A similar reductive activation was also observed in other recent studies of amorphous molybdenum sulfide HER catalyst materials.29, 33

80

Figure 3.9. Cyclic voltammograms indicate that the molybdenum sulfide catalyst is activated during the cathodic sweep of the first cycle (solid red line). Enhanced activity is observed on the anodic sweep of the first cycle (dotted red line) and in subsequent cycles (blue lines).

A polarization curve demonstrating the HER activity of the molybdenum sulfide catalyst on a glassy carbon electrode is compared to a bare glassy carbon control in Figure 3.10(a). The catalyst shows high activity for the HER with ~200 mV overpotential necessary to achieve 10 mA/cm2 hydrogen evolution current density, which is a useful metric for comparing catalysts for solar hydrogen production.5, 53 Measurements performed on four duplicate samples show consistent activity, with the overpotentials necessary to achieve 10 mA/cm2 current density ranging from 198 to 204 mV as shown in Figure 3.11.

The activity of the wet chemical-synthesized amorphous molybdenum sulfide catalyst is compared to several other MoS2 and MoS3 materials representing state-of-the-art catalysts at the time of this study in the Tafel plot shown in Figure 3.10(b).24, 27-29, 33 This material shows the highest activity of any reported molybdenum sulfide catalyst synthesized using a room temperature wet chemical procedure. The activity is within ~50 mV of the best reported molybdenum sulfide catalyst synthesized using any technique.28 The average Tafel slope for this catalyst is 60 mV/decade, with individual slopes ranging from 53 to 65 mV/decade for the catalyst films shown in Figures 3.10 (a) and 3.11. Though the Tafel slope alone is insufficient to determine the specific mechanism of the HER,27, 54, 55 it does match several earlier reports for molybdenum sulfide catalysts, which also exhibited slopes of ~60 mV/decade.24, 27

81

Figure 3.10. Electrochemical activity of the molybdenum sulfide catalyst. Polarization curves (a) show that the catalyst exhibits high activity for the HER. A Tafel plot (b) shows the electrochemical activity of the wet chemical synthesized amorphous molybdenum sulfide catalyst along with digitized data of HER measurements of several other state-of-the-art materials from the time of the study for comparison, 24 including MoS2 nanotriangles in pH 0.24 H2SO4 (orange trace), MoO3/MoS2 core-shell nanowires in 56 29 0.5 M H2SO4 (blue trace), electrodeposited amorphous MoS3 in pH 0 electrolyte (green trace), wet chemical synthesized amorphous MoS3/multi-walled carbon nanotube composite in 1 M H2SO4 (purple 33 28 trace), and MoS2 on reduced graphene oxide in 0.5 M H2SO4 (red trace).

Figure 3.11. Electrochemical activity of four duplicate catalyst samples. The overpotentials required for 10 mA/cm2 current density range from 198 to 204 mV. The values on the Turn Over Frequency axis are estimates derived from the catalyst roughness factor of 90 (calculation details provided below).

82

3.4.3 Changes in Catalyst Chemical State after Electrochemical Testing

To understand the changes in the material during electrochemical testing and identify the active form of the catalyst, we performed x-ray photoelectron spectroscopy (XPS) after catalyst activation via the first reductive potential cycle and after four hours of continuous reductive cycling as shown in Figure 3.5. XPS spectra collected after electrochemical testing reveal pronounced changes in the material, both in the elemental ratio as well as in the chemical states of the Mo and S. The S 2p peak corresponding to the catalyst material after the initial activation cycle shifts from 163.0 eV to 162.4 eV and begins to resemble a single spin-orbit splitting doublet. After four hours of reductive cycling, the S 2p region displays a distinct spin-orbit splitting doublet with the S 2p3/2 peak at a binding energy of 162.2 eV, very close to the value observed in the crystalline MoS2 spectra in Figure 3.8. The same can be said for the S 2s line, which shifts from 227.4 eV before testing to 226.2 eV after testing, the binding energy observed in crystalline MoS2. This trend suggests that the sulfur in the sample is partially reduced during catalysis, and the dominant form of sulfur present after testing exists in a chemical environment similar to MoS2.

Significant changes are also observed in the Mo 3d region after electrochemical testing. First,

(6+) the amount of Mo O3 has decreased substantially, to a negligible quantity that obviates the need for fitted features in the deconvoluted spectra of the “after testing” samples in Figure 3.5; thus for the sake of clarity those particular features are omitted. This significant decrease in

(6+) Mo O3 was also observed in catalyst samples allowed to rest in water or 0.5 M H2SO4 with no applied potential as shown in Figure 3.12, so we attribute this difference to the chemical dissolution of the native surface oxide. This observation is consistent with the expectation that

27, 57 any MoO3 would dissolve in the electrolyte used during catalyst testing. The second noticeable change is a significant shift in the binding energies of the Mo peaks corresponding

(A) to the molybdenum sulfide. Prior to testing, the majority of the molybdenum exists as Mo Sx. After the initial activation and after four hours of reductive cycling, the majority of the near-

(B) surface molybdenum is found in a slightly more reduced state labeled here as Mo Sx, with 3d3/2 and 3d5/2 peaks near 232.6 and 229.5 eV, respectively. Approximately 1/3 of the Mo remains as

(A) (6+) (A) Mo Sx in both cases. A small amount of Mo O3 may be hidden within the Mo Sx peaks, causing them to appear at a slightly higher binding energy than observed before catalytic testing.

(A) (B) The shift from Mo Sx to Mo Sx corresponds to a change in chemical environment, and the direction of this shift is consistent with a reduction of the molybdenum, as expected based on

83 the reductive potentials applied during catalyst activation and HER catalysis. The Mo 3d3/2 and

(B) 3d5/2 binding energies in the Mo Sx are close to the values observed in Figure 3.8 for crystalline

MoS2, which has a Mo oxidation state of 4+. The small remaining difference in the binding energies may arise from the different sulfur coordination or local bonding structure in this catalyst. All the changes in the molybdenum spectra measured after catalysis show that the active form of the catalyst resembles MoS2, although it is likely not a pure phase since the Mo still exists in multiple chemical states.

Finally, quantification of the Mo and sulfide/persulfide S signals after catalysis reveals changes in the stoichiometry of the near-surface region. After the initial reductive cycle, the sample's composition has changed from 22 % Mo and 78 % S (consistent with molybdenum trisulfide) to a composition of 34 % Mo and 66 % S, closely matching the 33% Mo and 67% S expected in MoS2. After four hours of reductive cycling, the composition appears to have continued to change, albeit slightly, to 39% Mo and 61% S. These values, which are also within experimental error of MoS2, could suggest a subtle depletion of sulfur during catalysis. All of these observations regarding the quantification and chemical state of Mo and S suggest that the MoS3 material resulting from the wet chemical synthesis can be considered a “precatalyst.” This material undergoes major changes during operation to create the active phase, which more closely resembles MoS2, consistent with the trends observed in previous reports of amorphous

29, 33 MoS3 materials. However, as shown in Figure 3.4, Raman spectra collected after catalysis confirm that the sample still does not contain crystalline MoS2 in a detectable quantity.

Therefore, the reduced phase created during catalysis is likely an amorphous MoS2.

84

Figure 3.12. X-ray photoelectron spectra of the S 2p and Mo 3d regions of the catalyst as deposited and after resting in H2O or 0.5 M H2SO4 with no applied potential. As deposited (a, b), the Mo 3d region includes Mo 6+ features, which we attribute to a native surface oxide. After resting in H2O for one hour (A) (c, d), the Mo 6+ feature is substantially diminished, leaving only the Mo Sx doublet corresponding to the untested molybdenum sulfide material, suggesting that the native surface oxide dissolves in water. The relative intensity of the S 2p peak also increases slightly because this signal is no longer attenuated by the thin surface oxide layer. A separate sample allowed to rest in 0.5 M H2SO4 for one hour (e, f) displays a similar S 2p feature and also lacks and evidence of Mo 6+ peaks. However, in this case, a more (B) reduced Mo Sx chemical state is also observed, which suggests that exposure to H2SO4 could chemically reduce some of the molybdenum sulfide material.

85

Due to the amorphous and nanostructured nature of this catalyst, there is likely significant surface site heterogeneity in this material. It is therefore difficult to accurately determine the structure and relative concentration of various types of surface sites, which makes the identification of the active sites of this material particularly challenging. Previous work on many different types of molybdenum sulfide materials shows that this general class of catalyst can exhibit high activity for hydrogen evolution despite wide variations in atomic structure and chemical composition. 11, 23-25, 28-31, 33, 56, 58 The catalyst material presented in this work likely exhibits many types of surface sites due to its amorphous, nanostructured nature, some of which could resemble the active sites previously described in the studies cited above. It is therefore possible that many different types of surface sites participate in the reaction. The results of this study provide initial indications that domains resembling MoS2 are created during catalysis and may contribute to this material’s high electrochemical activity, but further characterization is necessary to confirm this hypothesis. Synchrotron techniques such as x-ray absorption spectroscopy could yield greater insights about the catalyst composition and structure.11, 48

3.4.4 Turn Over Frequency

As discussed in Chapter 2, the intrinsic per-site activity of a catalyst is an important metric necessary to compare catalyst materials and to guide catalyst development. To estimate this key figure of merit, we used electrochemical capacitance surface area measurements to determine the active surface area of the catalyst film.59, 60 When combined with a simplified model of the surface structure, this technique enables independent estimation of the density of electrochemically active sites and the average activity of each site, reported as a per-site turn over frequency (TOFavg). Though the analysis shown here utilizes known properties for MoS2, which may not perfectly reflect MoS3 or the reduced phase created during electrochemical testing, it nevertheless provides useful insights for future improvements to catalyst design.

86

Figure 3.13. Electrochemical capacitance measurements for determination of the molybdenum sulfide catalyst surface area. Cyclic voltammograms (a) were taken in a potential range where no Faradaic processes were observed to measure the capacitive current from double layer charging. The capacitive current measured at 0.30 V vs. RHE was plotted as a function of scan rate (b) for the wet chemical synthesized amorphous molybdenum sulfide and the MoS2 flat standard. The ratio of the capacitive currents for the molybdenum sulfide catalyst and the flat standard was used to determine the relative roughness factor.

To find the electrochemically active surface area, as shown in Figure 3.13(a), we measured the non-Faradaic capacitive current associated with electrochemical double layer charging upon repeated potential cycling. This double layer charging current, ic, is proportional to both the scan

61 rate, ν, and the electrochemically active surface area of the electrode, Aechem:

The capacitive currents for the molybdenum sulfide catalyst and for a flat MoS2 standard prepared by sulfidizing a flat Mo metal film on a conductive silicon wafer were measured as a function of scan rate as shown in Figure 3.13(b). A potential range of 0.15 V to 0.35 V vs. RHE was selected for the capacitance measurements because no obvious electrochemical features corresponding to Faradaic current were observed in this region. Furthermore, the dependence of the current on the scan rate in this region is linear for both materials, which is consistent with capacitive charging behavior. Current arising from a Faradaic process would yield a square root dependence with respect to scan rate as a result of mass transfer (reactant diffusion)

62 limitations. The ratio of the currents for the catalyst and the flat MoS2 standard was taken as the relative roughness factor, RF:

87

From this analysis, we measured a RF value of 90. This calculation relies on the assumption that the intrinsic, surface area-normalized capacitance of the molybdenum sulfide catalyst and the flat MoS2 standard are the same. Although the chemical and physical characteristics of MoS2 differ from the amorphous molybdenum sulfide catalyst, we chose to use a MoS2 flat standard due to the difficulty of fabricating a perfectly flat, well-defined analogue of the amorphous molybdenum sulfide catalyst. The capacitance measured for the flat MoS2, approximately 60 μF/cm2, is consistent with expectations for a flat electrode.59, 62 As the surface area-normalized capacitance associated with double layer charging is expected to be similar (i.e. within an order of magnitude) for many metallic and semiconducting materials in the same aqueous electrolyte,62 the resulting surface area estimate for our work presented here is accurate to within an order of magnitude or better.

To calculate the active surface site density and TOFavg for the amorphous molybdenum sulfide catalyst, we use the RF of the catalyst, the geometry of a MoS2 surface, and the hydrogen evolution current density.

The number of surface sites on the flat standard was calculated based on the geometry of MoS2.

Although the MoS2 standard used in this study may have a small finite roughness, as an approximation, we model this material as an atomically flat MoS2 surface. Previous experimental measurements of MoS2 surfaces showed that the sulfur-sulfur bond distance is

63, 64 3.15 Å. Based on the hexagonal arrangement of sulfur atoms at the MoS2 surface, this corresponds to an area of 4.296 Å2/S atom, which is used to calculate the surface area occupied by each MoS2 unit:

Å2 2 S atom Å2 4.296   8.593 S atom 1 MoS2 MoS2

This can be used to calculate the number of MoS2 units (the number of surface sites for the flat standard) per cm2 geometric area:

1 MoS 1016 Å2 MoS 2  1.1641015 2 8.593 Å2 cm2 cm2

88

While the atomic-scale surface geometry of the amorphous molybdenum sulfide catalyst is different from the flat MoS2 standard, as an approximation, we assume that the surface site density is the same for both materials. This approximation is reasonable given that all materials have a surface site density close to 1015 sites/cm2.65 Therefore, we calculate the number of electrochemically accessible surface sites on the molybdenum sulfide catalyst using the following formula:

# SurfaceSites (Catalyst) # SurfaceSites (Flat Standard)   Roughness Factor cm2 geometricarea cm2 geometricarea

The relative roughness factor of the amorphous molybdenum sulfide catalyst compared to the flat MoS2 standard was determined to be 90 using electrochemical capacitance measurements. Using the formula above, this corresponds to 1.0 * 1017 surface sites/cm2.

To find the per-site TOF, we use the following formula:

# Total HydrogenTurn Overs / cm2 geometricarea TOF per site  # SurfaceSites (Catalyst) / cm2 geometricarea

The total number of hydrogen turn over events is calculated from the current density using the following conversion:56

 mA  1 A 1C / s  1 mol e 1 mol H  6.022141023 molecules H   j     2  2  cm2 1000 mA  1 A  96,485.3 C  2 mol e  1 mol H        2  H / s mA  3.121015 2 per cm2 cm2

Based on the current density of 10 mA/cm2, the per site TOF is determined for a roughness factor of 90 as follows:

 H 2 / s  2  2  mA  1cm  H / s 3.121015 cm 10    0.3 2 mA cm2 1.01017 surfacesites  surfacesite      cm2 

Using this surface site density, we calculated a turn over frequency (TOF) of 0.3 H2/s per surface site at 200 mV overpotential (~10 mA/cm2 current density, calculated per geometric electrode

89 area). Conservatively, due to the order of magnitude inaccuracy of these calculations, the TOFavg may fall between 0.03 H2/s and 3 H2/s per surface site. This range is comparable to the TOFavg values reported for other highly active molybdenum sulfide catalysts at similar current

27, 29 densities. However, previous reports of well-defined MoS2 nanoparticles indicate that the

24, 27 intrinsic TOF for a crystalline MoS2 edge site is ~3 orders of magnitude higher. As discussed previously, there is likely significant surface site heterogeneity in this amorphous molybdenum sulfide material, which raises the possibility that only a fraction of the surface sites may be active for hydrogen evolution. Therefore, the true TOF of the most active sites could be orders of magnitude greater than the average value of 0.3 H2/s.

3.4.5 Relationship between Catalyst Mass Loading and Activity

In spite of the approximations necessary to determine the TOF, it is clear that the high surface area nanostructured morphology of the catalyst film plays an important role in providing a significant number of active sites. To further investigate the effect of electrochemically active surface area on the observed activity, we synthesized samples with mass loadings varying from 25% to 400% of the typical catalyst preparation.

We expect that increased mass loading should result in a thicker film, and therefore a higher surface area. If the catalyst film has a porous structure that is uniform throughout its thickness for all mass loadings, surface sites throughout the film should be electrochemically accessible and the measured roughness factor, RF, should increase linearly with the mass loading.

Furthermore, if we assume that the fraction of catalytically active surface sites is constant with respect to RF, the HER activity should increase with the RF due to the larger number of surface sites that are available to perform the catalysis.66 To make a quantitative prediction of the relationship between the film roughness and the electrochemical activity, we consider the Tafel equation,61, 66 which relates the current and overpotential for an electrocatalytic reaction such as hydrogen evolution:

  a  blogi

In this equation, η is the overpotential, i is the catalytic current, a is an intercept, and b is the Tafel Slope. A more complete version of this equation can be derived by simplifying the Butler- Volmer equation:61, 66

90

RT RT    lni  lni nF 0 nF

Here, R is the ideal gas constant, T is the absolute temperature, α is the transfer coefficient, n is the number of electrons transferred, F is the faraday constant, and i0 is the exchange current. This equation can be simplified to yield the version written above by making the following substitutions:

RT b  and a  bl ogi  nF l og (e) 0

The exchange current, i0, is related to the electrochemically active surface area, Aechem, and the electrochemically active surface area-normalized exchange current density, j0, as follows:

i0  j0  Aechem

This exchange current density is related to the fundamental surface properties of the catalyst material, so while the exchange current scales with the electrochemically active area of the electrode, the exchange current density is constant. Thus, a final alternative version of the Tafel equation which is illustrative for these purposes can be written as follows:

  b  logi  b  log j0  Aechem 

From this equation, it is apparent that the overpotential required to yield a given current depends on the Tafel slope, b, and the electrochemically active area, Aechem. As discussed above, the roughness factor, RF, is defined as

Catalyst Active Surface Area A RF   echem SubstrateGeometricSurface Area Ageom

To observe the effect of increasing the active surface area, we substitute RF* Ageom into the Tafel equation:

  blogi blogj0  RF * Ageom 

The change in the overpotential necessary to achieve the same current density is then

91

     b  log(RF )

Therefore, we expect that for every 10-fold increase in the measured RF, the overpotential necessary to drive a given current will decrease by the Tafel Slope, b. The value of the Tafel slope measured during catalytic testing of the amorphous molybdenum sulfide is ~60 mV/decade.

To experimentally test these predictions, we prepared and tested catalyst samples with 25%, 50%, 100%, 200%, and 400% of the typical mass loading. The roughness factor is plotted vs. the mass loading in Figure 3.14(a). As expected, for the samples ranging from 25% to 200% mass loading, the roughness factor increases linearly with mass loading. This observation supports our assumption that catalyst morphology is porous and surface sites throughout the film thickness are electrochemically accessible. The sample with 400% mass loading deviates dramatically from this linear trend, with a RF far lower than expected.

To explain the maximum in roughness factor at 200% loading and the subsequent decrease at 400% loading, we collected SEM images of these catalyst samples as shown in Figure 3.15. These images show regions of the catalyst film near the edge of the glassy carbon substrate. After deposition, a portion of each film cracks and peels away from the substrate. For the films ranging from 25% to 200% typical loading, the film remains adhered to the substrate across the majority of the surface. In contrast, the film with 400% typical mass loading cracks and delaminates from all parts of the substrate, presumably due to increased mechanical stresses arising from the increased film thickness. The decrease in surface area associated with the delamination of a large portion of the film results in the lower observed RF.

92

Figure 3.14. Electrochemical activity of catalyst samples with 25%, 50%, 100%, 200%, and 400% typical mass loading. (a) Roughness factor vs. nominal mass loading. (b) Polarization curves demonstrating the catalytic activity of each sample. (c) Catalytic activity vs. nominal mass loading. (d) Catalytic activity vs. roughness factor.

93

Figure 3.15. SEM micrographs of catalyst films with (a) 25%, (b) 50%, (c) 100%, (d) 200%, and (e) 400% typical mass loading.

94

To measure the HER activity of these samples, we collected the polarization curves displayed in Figure 3.14(b). The films exhibit significant differences in catalytic activity. As a metric for comparing the HER activities, we use the potential required to produce 1 mA/cm2 of hydrogen evolution current density. We selected this metric to minimize the influence of iR-correction on the observed catalytic activity. During the electrochemical testing, the measured values of the series resistance, Rs, varied from approximately 5 to 12 Ω. While all these values are reasonable, this range is larger than we observed for identically prepared samples with 100% mass loading, and we found no correlation between the Rs and the mass loading, activity, or RF. Since the magnitude of the iR-correction to the voltage is larger at higher current densities, we selected a current density of 1 mA/cm2 for our metric rather than the 10 mA/cm2 typically used to compare HER catalysts.

The HER activity is plotted vs. the mass loading in Figure 3.14(c). These data display an increase in activity from 25% to 200% mass loading and a subsequent decrease at 400% loading, which mirrors the trend observed in the RF.

To determine the correlation between the activity and roughness factor, we plotted the overpotential required for 1 mA/cm2 vs. log(RF) as shown in Figure 3.14(d). We found that the activity increases linearly with log(RF), as predicted using the Tafel equation. This linear relationship supports our assumption that the fraction of catalytically active surface sites is indeed constant with respect to RF. The slope of this relationship, however, is approximately 38 mV/decade RF which is close to but not exactly the expected ~60 mV/decade Tafel slope measured for hydrogen evolution catalysis on these samples. This discrepancy likely arises from mass or charge transport limitations in the thicker films that can play a role during catalysis but not during electrochemical capacitance measurements, indicating that the full benefits of increased surface area have not been realized. There is room for improvement.

These results confirm that the activity of this catalyst is strongly dependent on the film structure and morphology. The activity could be further increased by choosing a support that prevents film delamination at high mass loadings.

Ultimately, these data provide additional evidence that the high surface area morphology of the catalyst film contributes significantly to its high activity and may explain, in part, why this catalyst has activity superior to similar materials reported previously.29, 33 This result suggests that high catalytic current densities could be achieved by choosing a support that prevents film

95 delamination at high mass loadings. Additionally, utilizing a three-dimensional substrate architecture designed to preserve a high roughness factor while mitigating transport limitations could afford high activity from a lower catalyst loading.27

3.4.6 Catalyst Stability

The long-term stability of a catalyst is another important metric to consider for commercial applications. The stability of this catalyst film was assessed by repeated potential cycling for more than 10,000 cycles to replicate diurnal cycling experienced by a HER catalyst for solar water splitting. The lower bound of the potential cycles, -0.25 V vs. RHE, was chosen to reach a current density in excess of 10 mA/cm2, which is a useful metric for comparing catalysts for solar hydrogen production. The results of the stability testing are displayed in Figure 3.16, which shows that the overpotential required to attain 10 mA/cm2 current density increases by only 57 mV after 10,000 potential cycles. Though not quite as stable as some other molybdenum sulfide materials,27, 28 the absolute overpotential required to drive 10 mA/cm2 remains low.

Figure 3.16. Electrochemical stability of the molybdenum sulfide catalyst. The overpotential required to reach a current density of 10 mA/cm2 increases by only 57 mV after 10,000 reductive potential cycles, indicating that the catalyst remains highly active.

96

Several factors may contribute to the slight loss in activity. One hypothesis is that surface adsorbates, potentially from impurities in the electrolyte, reference electrode, or within the film itself may poison the active sites of the catalyst over the course of the stability testing. We tested this hypothesis by pausing the stability testing after the 10,000th cycle to replace the electrolyte solution, after which the catalyst immediately recovered 30 mV of activity during subsequent measurements. Delamination of the catalyst film from the substrate is another likely cause of activity loss, an effect confirmed by scanning electron microscope imaging performed after electrochemical testing as shown in Figure 3.17. This loss of catalyst loading could potentially be remedied by using a conductive binder to secure the catalyst to the glassy carbon substrate or by depositing this material upon a different substrate altogether. Although further study is required to detail the specific mechanisms of the catalyst degradation, our preliminary efforts indicate that factors other than the inherent properties of the catalyst material are responsible for the observed decrease in activity. Overall, this highly active material remains an excellent HER catalyst throughout the course of the rigorous accelerated durability test.

97

Figure 3.17. SEM micrographs of the catalyst film (a) before electrochemical testing and (b, c) after 1,000 cycles electrochemical stability testing.

3.5 Conclusions

We have developed a scalable wet chemical synthesis to produce a highly active, stable amorphous molybdenum sulfide HER electrocatalyst. This synthetic technique requires no high temperature processing or secondary sulfidization step and allows for direct catalyst deposition onto many types of substrates. We performed extensive spectroscopic and electrochemical characterization to understand changes in this material during catalysis and investigate the origin of the catalytic activity of amorphous molybdenum sulfide catalysts, for which there have been several recent literature reports. Domains resembling MoS2 in both composition and chemical

98 state are created during catalysis and may contribute to the high HER activity of this material. The high density of active sites that results from the rough, nanostructured surface morphology also contributes to the high geometric current densities. The catalyst's stability was also ascertained; it was found to retain its activity throughout extended reductive potential cycling. The catalyst activity and stability could be further improved by designing a substrate structure to increase the total surface area and prevent catalyst delamination during operation. This highly active, stable HER catalyst is a promising candidate material that could help to enable the widespread deployment of cost-effective systems for electrochemical hydrogen production.

Although the ex situ techniques employed here provided a great deal of useful information about this catalyst’s properties, these tools are unable to reveal the nature of the active surface under operating conditions. In Chapter 4, we discuss in situ spectroscopic and microscopic characterization of this catalyst to discover further insights about the amorphous MoSx HER catalyst.

3.6 Copyright

Reprinted with permission from:

J.D. Benck, Z. Chen, L.Y. Kuritzky, A.J. Forman, and T.F. Jaramillo. "Amorphous Molybdenum Sulfide Catalysts for Electrochemical Hydrogen Production: Insights into the Origin of their Catalytic Activity." ACS Catalysis, 2012. 2 (9): 1916-1923. http://dx.doi.org/10.1021/cs300451q

Copyright 2012 American Chemical Society.

3.7 Author Contributions

Zhebo Chen, Leah Kuritzky, Arnold Forman, and Thomas Jaramillo were coauthors of the journal article based on this work. Jesse Benck performed all experiments with the exception of the TEM measurements, analyzed the data, and wrote the manuscript. Zhebo Chen and Arnold Forman contributed to data interpretation and all authors edited the paper.

99

3.8 Acknowledgments

This work was supported as part of the Center on Nanostructuring for Efficient Energy Conversion (CNEEC) at Stanford University, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DE-SC0001060. JDB acknowledges support from the National Science Foundation Graduate Research Fellowship Program and a Stanford Graduate Fellowship. AJF was supported by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, under subcontract NFT-9-88567-01 under prime contract no. DE-AC36-08GO28308. The authors thank Benjamin Reinecke for performing the TEM imaging, Andrey Malkovskiy for help with Raman spectroscopy, and Yelena Gorlin for helpful discussions about electrochemical stability measurements.

3.9 References

1. P. Häussinger, R. Lohmüller, and A.M. Watson. "Hydrogen, 6. Uses," in "Ullmann's Encyclopedia of Industrial Chemistry." 2000, Wiley-VCH Verlag GmbH & Co. KGaA. http://dx.doi.org/10.1002/14356007.o13_o07 2. U.S. Energy Information Administration. "The Impact of Increased Use of Hydrogen on Petroleum Consumption and Carbon Dioxide Emissions." 2008. http://www.eia.gov/oiaf/servicerpt/hydro/pdf/oiafcneaf%2808%2904.pdf 3. J.A. Turner. "Sustainable hydrogen production." Science, 2004. 305 (5686): 972-974. http://dx.doi.org/10.1126/science.1103197 4. J. Nowotny, C.C. Sorrell, L.R. Sheppard, and T. Bak. "Solar-hydrogen: Environmentally safe fuel for the future." International Journal of Hydrogen Energy, 2005. 30 (5): 521-544. http://dx.doi.org/10.1016/j.ijhydene.2004.06.012 5. M.G. Walter, E.L. Warren, J.R. McKone, S.W. Boettcher, Q.X. Mi, E.A. Santori, and N.S. Lewis. "Solar Water Splitting Cells." Chemical Reviews, 2010. 110 (11): 6446- 6473. http://dx.doi.org/10.1021/cr1002326 6. T.R. Cook, D.K. Dogutan, S.Y. Reece, Y. Surendranath, T.S. Teets, and D.G. Nocera. "Solar Energy Supply and Storage for the Legacy and Non legacy Worlds." Chemical Reviews, 2010. 110 (11): 6474-6502. http://dx.doi.org/10.1021/cr100246c 7. J. Turner, G. Sverdrup, M.K. Mann, P.C. Maness, B. Kroposki, M. Ghirardi, R.J. Evans, and D. Blake. "Renewable hydrogen production." International Journal of Energy Research, 2008. 32 (5): 379-407. http://dx.doi.org/10.1002/er.1372 8. T. Bak, J. Nowotny, M. Rekas, and C.C. Sorrell. "Photo-electrochemical hydrogen generation from water using solar energy. Materials-related aspects." International Journal of Hydrogen Energy, 2002. 27 (10): 991-1022. http://dx.doi.org/10.1016/S0360-3199(02)00022-8 9. L.A. Kibler. "Hydrogen electrocatalysis." ChemPhysChem, 2006. 7 (5): 985-991. http://dx.doi.org/10.1002/cphc.200500646

100

10. R. Subbaraman, D. Tripkovic, D. Strmcnik, K.-C. Chang, M. Uchimura, A.P. Paulikas, V. Stamenkovic, and N.M. Markovic. "Enhancing Hydrogen Evolution Activity in Water Splitting by Tailoring Li+-Ni(OH)2-Pt Interfaces." Science, 2011. 334 (6060): 1256-1260. http://dx.doi.org/10.1126/science.1211934 11. M.L. Tang, D.C. Grauer, B. Lassalle-Kaiser, V.K. Yachandra, L. Amirav, J.R. Long, J. Yano, and A.P. Alivisatos. "Structural and Electronic Study of an Amorphous MoS(3) Hydrogen-Generation Catalyst on a Quantum-Controlled Photosensitizer." Angewandte Chemie-International Edition, 2011. 50 (43): 10203-10207. http://dx.doi.org/10.1002/anie.201104412 12. N.V. Krstajic, V.D. Jovic, L. Gajic-Krstajic, B.M. Jovic, A.L. Antozzi, and G.N. Martelli. "Electrodeposition of Ni-Mo alloy coatings and their characterization as cathodes for hydrogen evolution in sodium hydroxide solution." International Journal of Hydrogen Energy, 2008. 33 (14): 3676-3687. http://dx.doi.org/10.1016/j.ijhydene.2008.04.039 13. A. Lasia and A. Rami. "Kinetics of Hydrogen Evolution on Nickel Electrodes." Journal of Electroanalytical Chemistry, 1990. 294 (1-2): 123-141. http://dx.doi.org/10.1016/0022-0728(90)87140-f 14. A. Rami and A. Lasia. "Kinetics of Hydrogen Evolution on Ni-Al Alloy Electrodes." Journal of Applied Electrochemistry, 1992. 22 (4): 376-382. http://dx.doi.org/10.1007/bf01092692 15. S.A.S. Machado and L.A. Avaca. "The hydrogen evolution reaction on nickel surfaces stabilized by H-absorption." Electrochimica Acta, 1994. 39 (10): 1385-1391. http://dx.doi.org/10.1016/0013-4686(94)E0003-I 16. Y. Petrov, J.P. Schosger, Z. Stoynov, and F. de Bruijn. "Hydrogen evolution on nickel electrode in synthetic tap water - alkaline solution." International Journal of Hydrogen Energy, 2011. 36 (20): 12715-12724. http://dx.doi.org/10.1016/j.ijhydene.2011.07.049 17. E. Endoh, H. Otouma, T. Morimoto, and Y. Oda. "New Raney-nickel composite- coated electrode for hydrogen evolution." International Journal of Hydrogen Energy, 1987. 12 (7): 473-479. http://dx.doi.org/10.1016/0360-3199(87)90044-9 18. A.K.M.S. Huq, A.J. Rosenberg, and A.C. Makrides. "Electrochemical Behavior of Nickel Compounds." Journal of the Electrochemical Society, 1964. 111 (3): 278-286. http://dx.doi.org/10.1149/1.2426108 19. L. Birry and A. Lasia. "Studies of the hydrogen evolution reaction on Raney nickel- molybdenum electrodes." Journal of Applied Electrochemistry, 2004. 34 (7): 735-749. http://dx.doi.org/10.1023/B:JACH.0000031161.26544.6a 20. R.N. Dominey, N.S. Lewis, J.A. Bruce, D.C. Bookbinder, and M.S. Wrighton. "Improvement of photoelectrochemical hydrogen generation by surface modification of p-type silicon semiconductor photocathodes." Journal of the American Chemical Society, 1982. 104 (2): 467-482. http://dx.doi.org/10.1021/ja00366a016 21. Y. Li and G.A. Somorjai. "Nanoscale Advances in Catalysis and Energy Applications." Nano Letters, 2010. 10 (7): 2289-2295. http://dx.doi.org/10.1021/nl101807g 22. S.W. Boettcher, J.M. Spurgeon, M.C. Putnam, E.L. Warren, D.B. Turner-Evans, M.D. Kelzenberg, J.R. Maiolo, H.A. Atwater, and N.S. Lewis. "Energy-Conversion Properties of Vapor-Liquid-Solid-Grown Silicon Wire-Array Photocathodes." Science, 2010. 327 (5962): 185-187. http://dx.doi.org/10.1126/science.1180783 23. Y.D. Hou, B.L. Abrams, P.C.K. Vesborg, M.E. Bjorketun, K. Herbst, L. Bech, A.M. Setti, C.D. Damsgaard, T. Pedersen, O. Hansen, J. Rossmeisl, S. Dahl, J.K. Norskov, and I. Chorkendorff. "Bioinspired molecular co-catalysts bonded to a silicon

101

photocathode for solar hydrogen evolution." Nature Materials, 2011. 10 (6): 434-438. http://dx.doi.org/10.1038/nmat3008 24. T.F. Jaramillo, K.P. Jørgensen, J. Bonde, J.H. Nielsen, S. Horch, and I. Chorkendorff. "Identification of active edge sites for electrochemical H2 evolution from MoS2 nanocatalysts." Science, 2007. 317 (5834): 100-102. http://dx.doi.org/10.1126/science.1141483 25. T.F. Jaramillo, J. Bonde, J.D. Zhang, B.L. Ooi, K. Andersson, J. Ulstrup, and I. Chorkendorff. "Hydrogen Evolution on Supported Incomplete Cubane-type Mo3S4 (4+) Electrocatalysts." Journal of Physical Chemistry C, 2008. 112 (45): 17492- 17498. http://dx.doi.org/10.1021/jp802695e 26. J. Bonde, P.G. Moses, T.F. Jaramillo, J.K. Norskov, and I. Chorkendorff. "Hydrogen evolution on nano-particulate transition metal sulfides." Faraday Discussions, 2009. 140: 219-231. http://dx.doi.org/10.1039/b803857k 27. Z.B. Chen, D. Cummins, B.N. Reinecke, E. Clark, M.K. Sunkara, and T.F. Jaramillo. "Core-shell MoO(3)-MoS(2) Nanowires for Hydrogen Evolution: A Functional Design for Electrocatalytic Materials." Nano Letters, 2011. 11 (10): 4168-4175. http://dx.doi.org/10.1021/nl2020476 28. Y.G. Li, H.L. Wang, L.M. Xie, Y.Y. Liang, G.S. Hong, and H.J. Dai. "MoS2 Nanoparticles Grown on Graphene: An Advanced Catalyst for the Hydrogen Evolution Reaction." Journal of the American Chemical Society, 2011. 133 (19): 7296-7299. http://dx.doi.org/10.1021/ja201269b 29. D. Merki, S. Fierro, H. Vrubel, and X.L. Hu. "Amorphous molybdenum sulfide films as catalysts for electrochemical hydrogen production in water." Chemical Science, 2011. 2 (7): 1262-1267. http://dx.doi.org/10.1039/c1sc00117e 30. D. Merki and X. Hu. "Recent developments of molybdenum and tungsten sulfides as hydrogen evolution catalysts." Energy & Environmental Science, 2011. 4 (10): 3878- 3888. http://dx.doi.org/10.1039/c1ee01970h 31. A.B. Laursen, S. Kegnaes, S. Dahl, and I. Chorkendorff. "Molybdenum sulfides- efficient and viable materials for electro - and photoelectrocatalytic hydrogen evolution." Energy & Environmental Science, 2012. 5 (2): 5577-5591. http://dx.doi.org/10.1039/c2ee02618j 32. D. Merki, H. Vrubel, L. Rovelli, S. Fierro, and X. Hu. "Fe, Co, and Ni ions promote the catalytic activity of amorphous molybdenum sulfide films for hydrogen evolution." Chemical Science, 2012. 3 (8): 2515-2525. http://dx.doi.org/10.1039/C2SC20539D 33. H. Vrubel, D. Merki, and X. Hu. "Hydrogen evolution catalyzed by MoS3 and MoS2 particles." Energy & Environmental Science, 2012. 5 (3): 6136-6144. http://dx.doi.org/10.1039/C2EE02835B 34. R. Gilliam, D. Kirk, and S. Thorpe. "Part 1—Structure-Sensitivity of Nanoparticle Catalysts: Relating Current Theories to Experimental Data." Electrocatalysis: 1-7. http://dx.doi.org/10.1007/s12678-011-0077-7 35. H.A. Gasteiger, S.S. Kocha, B. Sompalli, and F.T. Wagner. "Activity benchmarks and requirements for Pt, Pt-alloy, and non-Pt oxygen reduction catalysts for PEMFCs." Applied Catalysis B: Environmental, 2005. 56 (1–2): 9-35. http://www.sciencedirect.com/science/article/pii/S0926337304004941 36. G. Eda, H. Yamaguchi, D. Voiry, T. Fujita, M. Chen, and M. Chhowalla. "Photoluminescence from Chemically Exfoliated MoS2." Nano Letters, 2011. 11 (12): 5111-5116. http://dx.doi.org/10.1021/nl201874w

102

37. C. Lee, H. Yan, L.E. Brus, T.F. Heinz, J. Hone, and S. Ryu. "Anomalous Lattice Vibrations of Single- and Few-Layer MoS2." ACS Nano, 2010. 4 (5): 2695-2700. http://dx.doi.org/10.1021/nn1003937 38. T.J. Wieting and J.L. Verble. "Infrared and Raman Studies of Long-Wavelength Optical Phonons in Hexagonal MoS2." Physical Review B, 1971. 3 (12): 4286-4292. http://dx.doi.org/10.1103/PhysRevB.3.4286 39. T. Weber, J.C. Muijsers, and J.W. Niemantsverdriet. "Structure of Amorphous MoS3." Journal of Physical Chemistry, 1995. 99 (22): 9194-9200. http://dx.doi.org/10.1021/j100022a037 40. J. Iranmahboob, S.D. Gardner, H. Toghiani, and D.O. Hill. "XPS study of molybdenum sulfide catalyst exposed to CO and H-2." Journal of Colloid and Interface Science, 2004. 270 (1): 123-126. http://dx.doi.org/10.1016/j.jcis.2003.11.013 41. C.D. Wagner and J.A. Taylor. "Generation of XPS Auger Lines by Bremsstrahlung." Journal of Electron Spectroscopy and Related Phenomena, 1980. 20 (1-2): 83-93. http://dx.doi.org/10.1016/0368-2048(80)85008-0 42. R.V. Siriwardane and J.M. Cook. "Interactions of SO2 with sodium deposited on CaO." Journal of Colloid and Interface Science, 1986. 114 (2): 525-535. http://dx.doi.org/10.1016/0021-9797(86)90438-8 43. A.J. Jacobson, R.R. Chianelli, S.M. Rich, and M.S. Whittingham. "Amorphous Molybdenum Trisulfide - New Lithium Battery Cathode." Materials Research Bulletin, 1979. 14 (11): 1437-1448. http://dx.doi.org/10.1016/0025-5408(79)90087-4 44. K.S. Liang, S.P. Cramer, D.C. Johnston, C.H. Chang, A.J. Jacobson, J.P. Deneufville, and R.R. Chianelli. "Amorphous MoS3 and WS3." Journal of Non-Crystalline Solids, 1980. 42 (1-3): 345-356. http://dx.doi.org/10.1016/0022-3093(80)90035-6 45. C.H. Chang and S.S. Chan. "Infrared and Raman Studies of Amorphous MoS3 and Poorly Crystalline MoS2." Journal of Catalysis, 1981. 72 (1): 139-148. http://dx.doi.org/10.1016/0021-9517(81)90085-3 46. R.N. Bhattacharya, C.Y. Lee, F.H. Pollak, and D.M. Schleich. "Optical Study of Amorphous MoS3 - Determination of the Fundamental Energy Gap." Journal of Non- Crystalline Solids, 1987. 91 (2): 235-242. http://dx.doi.org/10.1016/s0022- 3093(87)80306-x 47. D. Belanger, G. Laperriere, and B. Marsan. "The Electrodeposition of Amorphous Molybdenum Sulfide." Journal of Electroanalytical Chemistry, 1993. 347 (1-2): 165- 183. http://dx.doi.org/10.1016/0022-0728(93)80086-w 48. S.J. Hibble, D.A. Rice, D.M. Pickup, and M.P. Beer. "Mo K-Edge EXAFS and S K- Edge Absorption Studies of the Amorphous Molybdenum Sulfides MoS4.7, MoS3, and MoS3·nH2O (n ~ 2)." Inorganic Chemistry, 1995. 34 (21): 5109-5113. http://dx.doi.org/10.1021/ic00125a006 49. S.J. Hibble, R.I. Walton, D.M. Pickup, and A.C. Hamon. "Amorphous MoS(3): clusters or chains? The structural evidence." Journal of Non-Crystalline Solids, 1998. 232: 434-439. http://dx.doi.org/10.1016/s0022-3093(98)00393-7 50. S.J. Hibble, M.R. Feaviour, and M.J. Almond. "Chemical excision from amorphous MoS3; a quantitative EXAFS study." Journal of the Chemical Society-Dalton Transactions, 2001 (6): 935-940. http://dx.doi.org/10.1039/b008296l 51. H.J. Jiao, Y.W. Li, B. Delmon, and J.F. Halet. "The structure and possible catalytic sites of MO(3)S(9) as a model of amorphous molybdenum trisulfide: A computational study." Journal of the American Chemical Society, 2001. 123 (30): 7334-7339. http://dx.doi.org/10.1021/ja0034085

103

52. S.J. Hibble and G.B. Wood. "Modeling the structure of amorphous MoS3: A neutron diffraction and reverse Monte Carlo study." Journal of the American Chemical Society, 2004. 126 (3): 959-965. http://dx.doi.org/10.1021/ja037666o 53. M.F. Weber and M.J. Dignam. "Efficiency of Splitting Water with Semiconducting Photoelectrodes." Journal of the Electrochemical Society, 1984. 131 (6): 1258-1265. http://dx.doi.org/10.1149/1.2115797 54. N.M. Markovic, B.N. Grgur, and P.N. Ross. "Temperature-Dependent Hydrogen Electrochemistry on Platinum Low-Index Single-Crystal Surfaces in Acid Solutions." Journal of Physical Chemistry B, 1997. 101 (27): 5405-5413. http://dx.doi.org/10.1021/jp970930d 55. W.C. Sheng, H.A. Gasteiger, and Y. Shao-Horn. "Hydrogen Oxidation and Evolution Reaction Kinetics on Platinum: Acid vs Alkaline Electrolytes." Journal of the Electrochemical Society, 2010. 157 (11): B1529-B1536. http://dx.doi.org/10.1149/1.3483106 56. Z. Chen, D. Cummins, B.N. Reinecke, E. Clark, M.K. Sunkara, and T.F. Jaramillo. "Core-shell MoO(3)-MoS(2) Nanowires for Hydrogen Evolution: A Functional Design for Electrocatalytic Materials." Nano Letters, 2011. 11 (10): 4168-4175. http://dx.doi.org/10.1021/nl2020476 57. M. Pourbaix. "Atlas of electrochemical equilibria in aqueous solutions." 2nd ed. 1974, Houston, TX: NACE International. 58. H.I. Karunadasa, E. Montalvo, Y.J. Sun, M. Majda, J.R. Long, and C.J. Chang. "A Molecular MoS2 Edge Site Mimic for Catalytic Hydrogen Generation." Science, 2012. 335 (6069): 698-702. http://dx.doi.org/10.1126/science.1215868 59. R. Kotz and M. Carlen. "Principles and applications of electrochemical capacitors." Electrochimica Acta, 2000. 45 (15-16): 2483-2498. http://dx.doi.org/10.1016/s0013- 4686(00)00354-6 60. S. Trasatti and O.A. Petrii. "Real Surface-Area Measurements in Electrochemistry." Pure and Applied Chemistry, 1991. 63 (5): 711-734. http://dx.doi.org/10.1351/pac199163050711 61. A. Bard and L. Faulkner. "Electrochemical methods: fundamentals and applications." 2001, New York: Wiley. 62. B.E. Conway, V. Birss, and J. Wojtowicz. "The role and utilization of pseudocapacitance for energy storage by supercapacitors." Journal of Power Sources, 1997. 66 (1-2): 1-14. http://dx.doi.org/10.1016/s0378-7753(96)02474-3 63. D. Yang, S.J. Sandoval, W.M.R. Divigalpitiya, J.C. Irwin, and R.F. Frindt. "Structure of single-molecular-layer MoS2." Physical Review B, 1991. 43 (14): 12053-12056. http://dx.doi.org/10.1103/PhysRevB.43.12053 64. S. Helveg, J.V. Lauritsen, E. Laegsgaard, I. Stensgaard, J.K. Norskov, B.S. Clausen, H. Topsoe, and F. Besenbacher. "Atomic-scale structure of single-layer MoS2 nanoclusters." Physical Review Letters, 2000. 84 (5): 951-954. http://dx.doi.org/10.1103/PhysRevLett.84.951 65. K.W. Kolasinski. "Surface science: Foundations of Catalysis and Nanoscience." 2008, Chichester: John Wiley & Sons Ltd. 66. R.P. O'Hayre. "Fuel cell fundamentals." 2006, Hoboken, NJ: John Wiley & Sons.

104

Chapter 4: In Situ Observation of the Chemical and Structural Evolution of Amorphous Molybdenum Sulfide

Catalysts for Electrochemical H2 Production

4.1 Abstract

Amorphous MoSx is a highly active, earth abundant catalyst for the electrochemical hydrogen evolution reaction. While previous ex situ studies have shown that the nanoparticles are reduced from MoS3 to MoS2 when subjected to reaction conditions, the mechanism of this transformation and the structure and composition of the active phase during catalysis remain poorly understood. In this work, we employ in situ characterization techniques to address these questions. We track the transformation of amorphous MoSx nanoparticles under hydrogen evolution conditions through ambient pressure X-ray photoelectron spectroscopy. We also employ transmission electron microscopy to image amorphous MoSx catalysts activated using ex situ and in situ procedures. Our results suggest that the surface of the amorphous MoSx catalyst is dynamic: while the initial catalyst activation forms the primary active surface of amorphous MoS2, continued transformation to the crystalline phase during electrochemical operation could contribute to catalyst deactivation. These results have important implications for the application of this highly active electrocatalyst for sustainable H2 generation.

4.2 Introduction

In Chapter 3, we studied an amorphous molybdenum sulfide HER catalyst which is promising for applications in PEC water splitting devices because it is earth abundant, composed of inexpensive elements, synthesized using a scalable room temperature process, and highly active for the HER.1-15 As a result of these advantages, we investigated the origins of the amorphous

MoSx catalyst’s high performance using ex situ characterization techniques including X-ray photoelectron spectroscopy (XPS). These measurements showed that immediately after synthesis, the material possesses a composition of MoS3. When the material is subjected to

105 potentials negative of 0 V vs. RHE in an electrochemical cell, it is reduced to its catalytically active form with a composition close to MoS2 (sometimes labeled “MoS2+x” due to the persistence of some additional sulfur). These conclusions are consistent with ex situ characterization results from other researchers.1, 3, 6, 7, 9

Due to the ultrahigh vacuum conditions used for these ex situ XPS measurements, these studies were unable to reveal how the transformation takes place or provide insight into the nature of the surface under electrochemical operating conditions. It remains unclear which catalyst surface sites actively participate in the HER, which is crucial information for understanding the elementary steps in the reaction and the source of the high catalytic activity. In situ characterization techniques can shed light on these unknowns by probing the catalytic process and the catalyst material under reaction conditions.

Ambient pressure X-ray photoelectron spectroscopy (APXPS), sometimes also called ambient pressure photoelectron spectroscopy (APPES),16, 17 is a technique ideally suited for this task, as the chemical sensitivity of XPS enables the differentiation of distinct chemical compositions while the near-ambient pressure capabilities afford access to relevant reaction conditions for electrochemical processes.18-21 In this chapter, we use an electrochemical cell inside an APXPS

18, 19 system to investigate the changes that take place at the catalytic surface of amorphous MoS3 nanoparticles under operating conditions. We conclude that the catalyst changes from MoS3 to

MoS2 in a gradual manner, and the MoS2 phase is the active surface under HER conditions.

While the APXPS studies provide useful information about the composition and chemical state of the catalyst, they provide limited evidence for structural changes in the molybdenum sulfide and lack the atomic-scale spatial resolution necessary to observe variations in the chemical state or morphology of the catalyst at different locations within the material. To address these questions, in this chapter we also employ transmission electron microscopy (TEM), which can probe atomic arrangements and provide sub-nanometer spatially localized spectroscopic data, resulting in a complete picture of the catalyst chemical state, structure, and morphology.22, 23 We use the TEM to image catalysts activated under two conditions: ex situ via an applied potential in an aqueous electrochemical cell, and in situ via thermal annealing under H2 gas in an “environmental TEM” (ETEM). We directly observe how hydrogen enables the chemical transformation of amorphous MoS3→MoS2 and show that crystalline MoS2 domains are formed after both activation procedures. Based on the limited extent of the observed crystallinity in the

106 electrochemically activated catalyst, it is likely that the primary active sites are on the surface of the amorphous MoS2 rather than the crystalline MoS2 domains. The formation of the crystalline domains could contribute to the deactivation of the amorphous MoSx catalyst after extended operation.1, 2

The combination of these powerful in situ experimental techniques yields a thorough picture of the amorphous molybdenum sulfide catalyst’s properties and performance. As this material is among the most active nonprecious catalysts for the HER, the new insights discussed in this chapter may be useful for developing improved systems or for implementing this material into renewable fuels technologies.

4.3 Methods

4.3.1 Catalyst and Electrochemical Cell Preparation for APXPS Measurements

Building on previous operando electrochemical APXPS studies,18, 19 the experimental setup employed here comprises a PEEK polymer framework with two chambers separated by a polymer electrolyte membrane (PEM) as shown in Figure 4.1. The membrane is made of Nafion 115 and coated on one side with 4 mg/cm2 Pt nanoparticles (diameter 10 − 20 nm) supported on a Nafion/carbon black mixture, which serves as the counter electrode. This structure was purchased from Fuel Cell Store, Inc. The working electrode was fabricated on the other side of the membrane. First, 0.2 g Vulcan XC-72 Carbon black and 0.44 mL 5% Nafion 117 solution (Aldrich) were dissolved in approximately 5 mL of isopropanol and mixed by means of 30 min of sonication. 50 μL of this mixture was spread over a 1 cm2 area of the blank side of the membrane, resulting in a uniform carbon/Nafion coating. With this conductive support in place, ~1 mg of amorphous molybdenum sulfide nanoparticles suspended in isopropanol was drop cast onto working electrode. The amorphous molybdenum sulfide particles were synthesized using the procedure described in Chapter 3.1 A schematic diagram of this fabrication procedure is shown in Figure 4.2. A control sample was prepared by coating a second membrane using the same protocol but replacing the amorphous MoSx nanoparticles with 0.1 mg of crystalline MoS2 particles, which were obtained from 30 minutes of high-frequency sonication of a natural bulk

MoS2 crystal in isopropanol.

107

Figure 4.1. (a) Diagram and (b) photograph of the electrochemical cell setup used for the APXPS measurements.

Figure 4.2. Synthesis scheme for the polymer electrolyte membrane assembly used to measure the APXPS spectra of the MoSx catalyst.

108

4.3.2 APXPS Measurements and Data Analysis

The assembled cells were introduced into the APXPS system on beamline 13-2 at the Stanford Synchrotron Radiation Lightsource (SSRL), where electrode and membrane humidification were ensured by the introduction of saturated water vapor to both sides of the membrane. Water vapor was introduced to the counter electrode through small tubes connected to the PEEK framework and to the working electrode via variable leak valves in the gas cell. Both electrodes were connected to an external Pinewave potentiostat, allowing for the characterization of the electrochemical properties of the system as a function of applied voltage. Figure 4.3 shows a two-electrode cyclic voltammogram of the amorphous molybdenum sulfide electrode under a 5 Torr water vapor atmosphere. Current corresponding to the HER is observed at potentials below −1.0 V with respect to the counter electrode. The XPS and electrochemical measurements were performed at room temperature.

Figure 4.3. Cyclic voltammogram of the PEM assembly with MoSx on the working electrode. The scan rate was 40 mV/s and the H2O vapor pressure was 5 Torr. The voltage represents the total cell potential for this two electrode measurement.

All XPS spectra were collected from the molybdenum sulfide cathode, and their binding energies (BEs) were referenced to the Fermi level by calibrating the C 1s signal of the graphitic carbon measured spectrum to 284.5 eV. Each spectrum was corrected based on the C 1s peak position measured under the corresponding experimental conditions. The average collection

109 time for each experimental condition was approximately 3 hours. The signal intensity of the S 2p spectra is normalized to the background intensity. Spectral deconvolutions and background subtractions (linear backgrounds) were performed using the Igor Pro software. Each S 2p spectrum recorded under HER conditions was deconvoluted using asymmetric Gaussian−Lorentzian functions (60% Gaussian, 40% Lorentzian, 0.5 asymmetry), with the BE for each component chosen from experimental or literature references.

4.3.3 Catalyst Synthesis for TEM Measurements

The amorphous molybdenum sulfide catalyst used for TEM measurements was also synthesized following the procedure reported in Chapter 3.1 The nanoparticles were dispersed in isopropanol and deposited onto substrates for electrochemistry and microscopy by spray casting to achieve a uniform coating with a mass loading of approximately 5 × 10-5 g/cm2. For the electrochemical activity measurements and ex situ TEM studies, the catalyst was deposited onto 5 mm diameter glassy carbon disk substrates (Sigradur G, HTW Hochtemperature-Werkstoffe GmbH). For the environmental TEM measurements, the molybdenum sulfide catalyst was deposited onto (100) silicon wafer pieces with native surface oxide (WRS materials).

4.3.4 Electrochemical Catalyst Activation and Activity Measurement for TEM Samples

The MoSx catalyst was activated in a three-electrode electrochemical cell in a rotating disk electrode configuration. The working electrode, consisting of the molybdenum sulfide catalyst on a glassy carbon electrode, was continuously rotated at 1600 rpm during all electrochemical testing procedures to remove hydrogen bubbles formed during the HER. The counter electrode was a graphite rod (Ted Pella), the reference electrode was Hg/Hg2SO4 in saturated K2SO4

(Hach), and the electrolyte was 0.5 M H2SO4 (Aldrich) prepared in Millipore water and continuously purged with H2 gas. Before the activation, the potential of the reference electrode was calibrated to -0.695 V vs. the reversible hydrogen electrode (RHE) using platinum working and counter electrodes in H2-purged electrolyte. These measurements were performed in a separate cell to prevent platinum contamination during the activity measurement. To activate and measure the performance of the catalyst, the series resistance of the cell was first determined using electrochemical impedance spectroscopy. Then, the potential of the working electrode was continuously swept between -0.60 V and -0.95 V vs. the reference electrode at 50 mV/s for

110

30 minutes. After the measurements, the current-voltage data were mathematically corrected to remove the effects of the series resistance. The working electrode was removed from the electrolyte and rinsed in Millipore water before being prepared for TEM measurements.

4.3.5 TEM Sample Preparation and Imaging

TEM samples were prepared using a focused ion beam (FIB, FEI Helios NanoLab 600i DualBeam FIB/SEM) lift-out technique. Prior to ion milling, the samples were protected with a platinum capping layer deposited by an electron beam and then a gallium ion beam. The samples were then transferred to a Cu TEM grid using an Omniprobe manipulator and thinned by FIB milling closely parallel to the thin piece with a 30 keV Ga ion beam to a thickness of 80 nm. The Ga ion beam energy was reduced to 3 keV for the final thinning.

Aberration-corrected TEM imaging was performed using a FEI Titan 80-300 Environmental Transmission Electron Microscope equipped with a spherical aberration corrector in the imaging (objective) lens, and operated at 300 kV. The Cs image corrector was adjusted to approximately -10 μm, and all images presented were acquired at slightly over focus conditions (negative Cs imaging). For scanning transmission electron microscopy – electron energy loss spectroscopy (STEM – EELS) observations, the FEI Titan 80-300 ETEM was used in STEM mode and carried out on the same area where the TEM observations were made. The STEM probe size was approximately 0.5 nm.

ETEM experiments were performed using the same FEI 80-300 kV environmental TEM. A Gatan 652 Inconel heating holder was used to heat the samples inside the microscope. Hydrogen gas of research grade 6.0 (99.9999% purity, Air Liquide) was used. The environmental cell surrounds the specimen holder within the imaging objective lens and can allow gas pressures up to ~20 mbar. As there are no membranes in the electron beam path, the instrument’s sub-0.1 nm resolution is maintained.

111

4.4 Results and Discussion

4.4.1 Ambient Pressure X-ray Photoelectron Spectroscopy

Figure 4.4(a) shows the S 2p XPS spectra of the amorphous molybdenum sulfide working electrode surface probed with an incident photon energy of 900 eV under vacuum (10-6 Torr chamber pressure) and under water vapor pressures of 1 and 5 Torr. These measurements were performed with the electrochemical cell at open circuit and at different applied biases. The water vapor pressure, applied voltage, and cell current density are shown for each spectrum. The labels 0 through VIII correspond to the order in which the spectra were collected. The XPS spectra reveal that as the system is humidified and taken to operating HER conditions, there is an irreversible change in the catalyst composition. Initially the BE of the largest peak is ~163.5 eV, but as the testing continues, the feature at 162 eV grows larger. The peak shapes and BE shifts are is consistent with the trends observed in Chapter 3, though the exact BEs measured vary by a few tenths of an eV due to differences in the spectrometers and calibration procedures used for these measurements. The changes observed in Figure 4.4(a) are independent of X-ray beam exposure, and are slow under 1 Torr humidification and current densities in the μA/cm2 regime. The rate of change becomes more rapid once the system is taken to more extreme experimental conditions. Due to the two-electrode setup, the reported voltage values correspond to the total cell voltage (i.e., the potential required for the reaction at both electrodes and the potential drops due to series resistance). Nonetheless, with enough applied voltage, the cell reaches current density regimes on the order of mA/cm2, which remained constant during the XPS acquisition. The electrochemical data also show that humidification plays an essential role for the cell currents, with the transition from 1 to 5 Torr of water resulting in more double the current, even with a lower applied voltage. This is attributed to the condensation of a water layer entirely covering the surface as the relative humidity reaches 30%,24 which underscores the need for techniques such as APXPS that enable composition and chemical state measurements under the relevant operating conditions.

112

Figure 4.4. APXPS spectra of the S 2p region of (a) the amorphous MoSx nanoparticles and (b) the crystalline MoS2 control sample under various experimental conditions.

Figure 4.4(b) shows the S 2p spectra for the crystalline MoS2 control sample probed with the same photon energy and similar reaction conditions as the amorphous MoSx electrode. The water vapor pressure, applied voltage, and cell current density are shown for each spectrum.

113

The labels IX – XII correspond to the order in which the spectra were collected. The low measured currents are attributed to low catalyst loading. In contrast to the amorphous MoSx sample, the crystalline MoS2 sample shows no changes in the XPS spectra after being subjected to similar humidification and potential biases. This confirms that the changes are unique to the amorphous MoSx catalyst and do not occur solely due to the experimental conditions employed for the APXPS measurements.

Combining the results from the MoS2 crystalline control with reference spectra from the ex situ experiments in Chapter 3 and other sources,1, 4, 6 it is possible to deconvolute the spectra of the

MoSx nanoparticles into two components, as shown in Figure 4.5: First, shown in red, the sulfur atoms that match the chemical state observed in the crystalline MoS2 control with S 2p3/2 and S

2p1/2 signatures at 162.0 and 163.2 eV, respectively; second, shown in blue, the higher BE features at 163.5 and 164.7 eV, corresponding to S 2p3/2 and S 2p1/2 in the MoS3 phase. The broadening of the peaks with respect to the crystalline sample suggests the existence of a continuum of compositions between MoS3 and MoS2, which implies a highly defective structure.

The two panels in Figure 4.5 show the spectra for selected experimental conditions from Figure

4.4(a) and the fitted components described above. There is a significant increase of the MoS2 component observed in Figure 4.5(b), which was collected under HER conditions. By calculating the areas of the components corresponding to MoS2 and MoS3, it is possible to assess the relative composition of the surface of this MoSx catalyst under each condition. We observe that the as-synthesized nanoparticles consist of 65% MoS3 sites, but when used for hydrogen evolution, the MoS3 sites are reduced to MoS2 sites during the electrolysis (proton diffusion to the bulk or sulfur diffusion to the reduced surface being two possible mechanisms for the transformation), with the MoS2 sites eventually becoming 79% of the composition of the catalyst surface probed by these measurements. Taking into consideration that at 900 eV excitation energy, the inelastic mean free path for S 2p electrons is about 2.2 nm, we can conclude that the transformation occurs at the surface sites of the nanoparticles.

114

Figure 4.5. S 2p spectral deconvolution for the amorphous MoSx catalyst under selected experimental conditions. (a) Curve-fitted S 2p spectrum of amorphous MoSx nanoparticles under open circuit conditions, Spectrum I in Figure 4.4(a). (b) Curve-fitted S 2p spectrum of amorphous MoSx nanoparticles under hydrogen evolution conditions, Spectrum VII in Figure 4.4(a). The green components correspond to sulfur in the chemical state of MoS3. The red components correspond to sulfur in the chemical state of MoS2.

The transformation observed in Figure 4.4(a) occurs gradually throughout the experiment as the HER takes place. The spectra for the final experiments show that the surface under operating conditions primarily consists of sites with a chemical state resembling MoS2, consistent with the results of the ex situ experiments discussed previously. Based on the electrochemistry results in Chapter 3, which showed that the activity of the catalyst increased after the initial reduction of the MoS3 to MoS2, it is likely that the MoS2 surface is the main catalytically active phase, though these data cannot be used to rule out the possibility that the MoS3 phase also contributes some HER activity.

4.4.2 Transmission Electron Microscopy of Electrochemically Activated MoSx

The APXPS measurements revealed changes in the amorphous MoSx catalyst chemical state, but did not show changes in morphology or spatially resolved information indicating where or how the transformation occurs. To answer these questions, we use an analytical TEM to determine the atomic arrangement and chemical state of electrochemically activated MoSx. The HER activity of the catalyst during the activation procedure is displayed in Figure 4.6. The catalytic current density of the amorphous MoSx is much larger than the baseline measurement

115 of the bare glassy carbon substrate, confirming that the molybdenum sulfide is catalytically active for the HER. As observed in Chapter 3, the catalyst is activated by the first reductive potential cycle. The current during the first cathodic sweep corresponds to a mixture of the HER and reduction of the catalyst itself. This changes the surface composition and chemical state of

1, 7, 9 the material to form the catalytically active amorphous MoS2. During the second cathodic

2 sweep, a current density of 10 mA/cm geo is achieved at a potential of -0.21 V vs. RHE, indicating that the performance of the activated catalyst is consistent with the measurements in

1 2 Chapter 3. The slightly larger overpotential to reach 10 mA/cm geo here likely results from a reduced catalyst loading. After 30 minutes of cycling, the catalyst retains a high activity, with

2 the potential for 10 mA/cm geo changing by only 10 mV to -0.22 V vs. RHE. The extended cycling provides time for the catalyst to complete the transformation into the active phase.

Figure 4.6. Electrochemical activity of the MoSx catalyst and a bare glassy carbon substrate.

Figure 4.7(a) shows a bright field low magnification TEM image of the as-deposited sample.

This image shows several distinct layers: the glassy carbon substrate, the amorphous MoSx, and a platinum capping layer, which was deposited using a focused ion beam (FIB) to prevent damage to the catalyst during the TEM sample preparation. The MoSx layer appears quite uniform with no observable structural defects such as pores. The absence of lattice fringes in the high resolution TEM image in Figure 4.7(b) suggests that the material is fully amorphous.

116

Figure 4.7. Cross-section view of MoSx catalysts on a glassy carbon substrate. (a) Low magnification TEM image and (b) HRTEM image of the as-deposited MoSx sample showing it to be amorphous. (c) Annular dark field (ADF) STEM image near the surface of the as-deposited sample with the Pt deposition clear at the top. (d) Background subtracted EEL spectra of sulfur L2, 3 edge from regions 1 – 6 indicated on (c) showing the homogeneous chemical state. (e) Low magnification TEM image of the sample after 30 min of electrochemical testing showing the formation of pores and (f) HRTEM image of a representative pore. The inset in (f) shows a higher magnification TEM image of the area indicated by the yellow box indicating the nucleation of crystalline domain around the pore. (g) ADF STEM image of the sample after electrochemical testing. (h) Background subtracted EEL spectra from regions 1 – 6 indicated on (g) showing the change in chemical states of sulfur at positions 2 – 4 and 6, as indicated by the variations in the fine structure of the sulfur L2, 3 edges. Note that the ADF STEM image in (g) is acquired from the exact same region shown in (f).

In addition to observing the catalyst morphology, we used scanning transmission electron microscopy – electron energy loss spectroscopy (STEM-EELS) to measure the chemical state at different locations within the sample. The STEM image displayed in Figure 4.7(c) indicates six locations within the as-deposited MoSx where EEL spectra were collected, and Figure 4.7(d) shows the sulfur L2,3 edge fine structure EEL spectrum obtained at each location. The features in the post-edge region of the EEL spectra, which comprise the energy loss near-edge fine structure (ELNES), depend on the local coordination, bonding states, and valence of the sulfur atoms in the catalyst.22, 23, 25, 26 The spectra from the as-deposited catalyst are all very similar, indicating that the material is spatially homogeneous. The shape of the sulfur L2,3 edge observed

117 here is characteristic of the amorphous MoS3 starting phase. None of these spectra display the characteristic ELNES peaks observed at 174 eV in crystalline 2H-MoS2, as shown in Figure 4.8(b).25

Figure 4.8. (a) HRTEM image of a 2H-MoS2 control sample showing the characteristic layered structure and (b) Background subtracted EEL spectra of the sulfur L2,3 edge from the sample displaying the characteristic peak of crystalline MoS2 at 174 eV.

To understand changes in the catalyst after electrochemical activation and HER operation, we imaged a second MoSx catalyst sample after the 30 minute electrochemical activation displayed in Figure 4.6. The low magnification TEM image shown in Figure 4.7(e) reveals pores in the activated MoSx layer. The formation of pores likely results from a decrease in volume of the catalyst material during the activation process, which transforms the material from MoS3 to

MoS2. A HRTEM image of a representative pore in Figure 4.7(f) shows lattice fringes around the periphery which suggest that some crystalline domains have formed near the edges of pores, where the catalyst was in contact with the electrolyte. The linear structure of these fringes suggests that they may correspond to crystalline MoS2 imaged from the edge. The extent of crystallization is limited to within ~1 – 2 nm from the pore surfaces, and no obvious structural changes are observed in other regions including the surface of the catalyst material itself. To confirm that these crystalline domains were formed by the electrochemical activation procedure

118 and not by exposure to the electron beam, we investigated the stability of the catalyst under extended electron beam irradiation and observed no changes in morphology.

Figure 4.7(g) shows a STEM image of the activated MoSx material along with six locations where EEL spectra were collected. This is the same region of the sample shown in the HRTEM image in Figure 4.7(f). Though it is not possible to quantify the composition of the MoSx catalyst

26 using EELS due to the overlap of the sulfur L2,3 and molybdenum M4,5 spectral features, the

EEL spectra of the sulfur L2,3 edge region in Figure 4.7(h) reveal substantial spatial variations in the chemical state of the material as well as significant differences compared to the as- deposited material. At positions 2, 3, 4, and 6, near the pores and catalyst surface, a small peak appears at approximately 174 eV. This peak resembles the characteristic ELNES signature observed in crystalline MoS2, indicating that the material at the pore edges and catalyst surface more closely resembles 2H-MoS2 in chemical state after activation. In contrast, at positions 1 and 5, which are in the bulk of the material away from the catalyst/electrolyte interface, the EEL spectra more closely resemble that of the as-deposited amorphous MoS3. This shows that the change in the catalyst is a surface process and does not fully penetrate through the bulk of the material. The chemical state change is observed in the regions where crystalline MoS2 domains are formed, but also in other surface regions where there is no evidence of the formation of a crystalline phase. These results indicate that the catalyst activation likely begins with changes in the chemical state and composition from amorphous MoS3 to amorphous MoS2. Then, upon reaching this composition and chemical state, crystalline MoS2 nucleates in some regions. These chemical and structural transformations may result from the removal of sulfur via reaction with hydrogen or protons during the electrochemical activation.

4.3.3 Environmental TEM Measurements of MoSx Activated by In Situ Annealing

To directly test whether hydrogen is necessary for the transformation process, we imaged the activation of the catalyst in situ using aberration corrected ETEM. Given the challenges in exactly replicating the electrochemical conditions in the microscope, we employed in situ thermal annealing of the sample under a hydrogen atmosphere to approximate the electrochemical activation process. In an electrochemical environment, the chemical potential of adsorbed H atoms at an electrode surface depends on the electrode potential as expressed in the following equation, where μ is the chemical potential, q is the fundamental charge, and U is the applied electrode potential:27

119

During the electrochemical activation procedure, the negative applied potentials increase the chemical potential of adsorbed H (in the form of proton/electron pairs), which most likely results in higher surface coverage.27, 28 During the in situ thermal annealing activation procedure, rather than changing the electrode potential, raising the pressure of hydrogen gas increases the chemical potential of adsorbed hydrogen in a similar way. Direct adsorption of H2 molecules is expected to be highly unfavorable compared to dissociative adsorption, so under both activation conditions atomic hydrogen will likely be adsorbed at the catalyst surface. Since there are many important differences between the electrochemical activation procedure and the annealing under hydrogen gas related to the chemical potential of electrons, solvation effects, and other phenomena, this activation procedure enables us to test the hypothesis that hydrogen (rather than any uniquely electrochemical factor) is necessary for the activation of the catalyst.

One important consideration during the observation of in situ gas-solid reactions is the possible ionization of the gaseous species by the imaging electron beam. In this case, when the MoSx catalyst is continuously imaged under a H2 gas atmosphere, transformations in the morphology and chemical state of the sample occur rapidly even at room temperature due to the presence of atomic, ionized, and radical hydrogen species. These conditions therefore do not allow us to test the effects of thermal annealing at different temperatures in a straightforward manner. As a result, we performed the thermal annealing activation process without an imaging beam, and the changes in MoSx were observed after purging the gas from the chamber, a protocol which has been successfully adopted for studies of carbon nanotube oxidation.29 The regions of interest were identified for tracking under high vacuum. Then, with the electron beam blanked, 5 mbar of H2 was introduced into the ETEM for 30 min while keeping the temperature constant at room temperature. At the end of this cycle, the gas was purged from the system. The same regions were located and imaged to identify any differences after hydrogen exposure. The temperature was then increased to 150 °C, the reaction process was repeated, and the same regions were tracked and imaged at 150 °C after hydrogen was purged from the system. These reaction procedures were repeated again at 300 °C. Although the imaging was not performed under a gas atmosphere, we were able to carry out out the catalyst activation procedure inside the ETEM and repeatedly image the same location within the catalyst.

120

A second important consideration in the design of these measurements is the selection of the substrate. Glassy carbon, which was used for the electrochemically activated samples imaged previously, could form hydrocarbons during the reaction with hydrogen. As a result, we used silicon with a native SiO2 layer as the substrate rather than glassy carbon.

Figure 4.9. Aberration-corrected TEM images showing the structural changes in the MoSx sample after in situ activation in a H2 gas environment. (a) Low magnification image of the sample at room temperature under high vacuum. (b) HRTEM image of the region indicated by the yellow box in (a) showing it to be amorphous. HRTEM images of the samples after exposure to hydrogen for 30 min at room temperature (c) and at 150 °C (d). Note that the images in (c and d) were acquired from the regions near to the box in (a). Increasing fraction of MoS2 crystallites near the SiO2 interface is revealed.

121

Figure 4.9(a) shows a bright field low magnification TEM image of the as-deposited sample. Figure 4.9(b) shows the HRTEM image from the indicated region in Figure 4.9(a). This region possesses a fully amorphous structure without lattice fringes, as expected. Figure 4.9(c) shows a HRTEM image of the sample at room temperature after exposure to 5 mbar of H2 for 30 min, where crystalline MoS2 starts to nucleate at the interface between SiO2 and MoSx. As shown in

Figure 4.9(d), MoS2 domains start to grow from the SiO2/MoSx interface into the bulk of the

30 sample at 150 °C after exposure to H2 for 30 min. The high mobility of H2 inside the SiO2 layer likely allows for the exposure of MoSx to H2 at the SiO2/MoSx interface, leading to the reaction with H2. As a result, the crystallization of the catalyst is limited to the SiO2/MoSx interface.

Figure 4.10. Structural and chemical changes in the MoSx after H2 exposure for 30 min at 300 °C. (a) HRTEM image showing more extensive MoS2 formation and (b) ADF STEM image of the samples after H2 exposure at 300 °C after H2 exposure for 30 min. (c) Series of background subtracted EEL spectra of sulfur L2, 3 edge from points 1-3 indicated on (b).

Figure 4.10(a) shows a HRTEM image of the same region of the sample at 300 °C after exposure to 5 mbar H2 for 30 min. We observe that the crystalline MoS2 domains nucleated at the SiO2 interface grow further into the bulk. Separately, crystalline MoS2 domains start to nucleate within the bulk of the sample, which leads to the formation of crystalline MoS2 throughout the whole sample. The extensive crystalline MoS2 domain formation observed after the 300 °C treatment likely results from faster diffusion of H2 at this elevated temperature as well as greater probability of overcoming kinetic barriers to the formation of the crystalline phase. Figure

122

4.10(b) displays a STEM image of the same catalyst region as indicated in Figure 4.10(a). The

EEL spectra of the catalyst exposed to H2 at 300 °C taken at the points indicated in Figure 4.10(b) are shown in Figure 4.10(c). The three points selected for EELS show that the chemical state has changed at each point to more closely resemble crystalline MoS2. As discussed above, the transformation to crystalline MoS2 is observed at many points throughout the film, but there still exist some regions that remain amorphous. Points 1 and 2 correspond to crystalline MoS2 domains near the SiO2 interface and within the bulk of the film, respectively. These spectra both possess the characteristic sulfur L2,3 ELNES signature features observed in the crystalline MoS2 control sample in Figure 4.8(b). Point 3 corresponds to a domain of MoSx that appears amorphous in the HRTEM image. The EEL spectrum for this point looks similar to the crystalline MoS2 spectrum, but it does not possess the same feature, a “dip” at 180 eV, that is characteristic of the crystalline material. These measurements show that although the material is substantially transformed to the crystalline MoS2 phase in the presence of hydrogen at 300

°C, there still remain some amorphous domains that possess chemical states resembling MoS2 but with no observable crystallinity.

Figure 4.11. Aberration corrected HRTEM images of amorphous MoSx under high vacuum condition (a) at room temperature and (b) after annealing at 300 ºC for 1 hour. There are no observable changes and the image appearance is characteristically. (c) EEL spectra of the sulfur L2,3 edge region of the amorphous MoSx samples shown in (a) and (b) confirming no change in chemical state in the thin film.

A control set of MoSx samples was heated in the TEM under high vacuum conditions to 300 ºC for 1 hour to establish any effects of heating without H2 gas. Figure 4.11(a) displays a HRTEM

123 image of amorphous MoSx at room temperature, showing a fully amorphous structure without lattice fringes. Figure 4.11(b) shows a HRTEM image of the sample after annealing at 300 °C for 1 hour under high vacuum. The absence of lattice fringes indicates that the annealed sample also possesses a fully amorphous structure. There is no evidence of any structural changes such as crystallization after the treatment. EELS was carried out on the same regions to examine the chemical states before and after annealing under high vacuum condition. Figure 4.11(c) shows

EEL spectra from the MoSx at room temperature and after annealing under high vacuum. The shape of the sulfur L2,3 edge from both spectra are characteristic of the amorphous MoS3, and no chemical changes were observed after annealing without H2 gas. These results verify that the presence of hydrogen is critical to the structural and chemical changes during the catalyst activation process.

4.5 Conclusions

In this chapter, we used APXPS and TEM to study the activation of the amorphous MoSx HER catalyst. The results of all of our studies suggest that the surface of this catalyst is dynamic, with structural and chemical transformations occurring during the catalyst activation and continuing during operation. The APXPS study showed that the MoSx surface undergoes a gradual, irreversible change in chemical state from MoS3 to MoS2, which is the active surface during the

HER. The TEM studies confirm that the active phase has a chemical state resembling MoS2 based on EELS measurements. The limited extent of crystallization observed in the samples activated by 30 minutes of electrochemical potential cycling in Figure 4.8 suggests that the amorphous surface contributes most of the HER activity to this material.

The ETEM studies also showed that the phase transition to form crystalline MoS2 is enabled by the presence of hydrogen. During extended electrochemical operation, the formation of crystalline domains may continue, as the protons or hydrogen necessary for this transformation will always be present while the MoSx is catalyzing the HER. Since basal planes are the most energetically stable facets of crystalline MoS2, increased crystallization likely leads to the exposure of mostly catalytically inert basal plane sites, decreasing the total number of active surface sites. The crystalline domains observed in Figure 4.7(f) appear to be oriented with the basal planes parallel to the pore surface and facing the electrolyte. A similar preferential formation of basal plane sites has been observed in the synthesis of many crystalline MoS2

124 structures, which can result in MoS2 catalysts with relatively low electrode area-normalized HER activity unless special efforts are taken to create nanostructures with a high number of active edge sites exposed.31, 32 These observations suggest that the crystalline phase transition could be a contributing factor to the gradual deactivation of the amorphous MoSx catalyst observed in Figure 4.6 after 30 minutes of potential cycling and in Chapter 3 after extended electrochemical durability testing.1 The formation of crystalline domains could thus be a deactivation mechanism that is intrinsic to amorphous MoSx materials. This hypothesis is worthy of further investigation if this amorphous MoSx catalyst is to be implemented in successful PEC water splitting devices.

4.6 Copyright

Portions of this chapter reprinted with permission from:

H.G.S. Casalongue, J.D. Benck, C. Tsai, R.K.B. Karlsson, S. Kaya, M.L. Ng, L.G.M. Pettersson, F. Abild-Pedersen, J.K. Nørskov, H. Ogasawara, T.F. Jaramillo, and A. Nilsson. "Operando Characterization of an Amorphous Molybdenum Sulfide Nanoparticle Catalyst during the Hydrogen Evolution Reaction." The Journal of Physical Chemistry C, 2014. 118 (50): 29252-29259. http://dx.doi.org/10.1021/jp505394e

Copyright 2014 American Chemical Society.

4.7 Author Contributions

Hernan Sanchez-Casalongue was the first author of the journal article based on the ambient pressure photoelectron spectroscopy work. Jesse Benck, Charlie Tsai, Rasmus Karlsson, Sarp Kaya, May Ling Ng, Lars Pettersson, Frank Abild-Pedersen, Jens Norskov, Hirohito Ogasawara, Thomas Jaramillo, and Anders Nilsson were also coauthors of the journal article.

Hernan Sanchez-Casalongue developed the electrochemical cell used to study the MoSx catalyst inside the ambient pressure photoelectron spectroscopy tool. Jesse Benck synthesized the MoSx catalyst and fabricated the samples used for spectroscopic characterization. Hernan Sanchez- Casalongue and Jesse Benck performed the APXPS measurements with assistance from Sarp Kaya, May Ling Ng, Hirohito Ogasawara, and Anders Nilsson. Hernan Sanchez-Casalongue wrote the manuscript based on this work, and all authors edited the paper.

125

Sang Chul Lee was the first author of the journal article based on the environmental transmission electron microscopy measurements (in preparation). Jesse Benck, Charlie Tsai, Joonsuk Park, Ai Leen Koh, Frank Abild-Pedersen, Thomas Jaramillo, and Robert Sinclair were also coauthors of the journal article. Sang Chul Lee performed all TEM measurements. Jesse Benck synthesized the MoSx catalyst. Charlie Tsai performed the molybdenum sulfide phase stability and nudged elastic band DFT calculations. Jesse Benck and Sang Chul Lee wrote the manuscript and all authors contributed to editing the paper.

4.8 Acknowledgments

The APXPS study was supported the Joint Center for Artificial Photosynthesis, a DOE Energy Innovation Hub under Award DE-SC0004993. H.O. gratefully acknowledges the support from Precursory Research for Embryonic Science and Technology (PRESTO), Japan Science and Technology Agency (JST). Portions of this research were carried out at the Stanford Synchrotron Radiation Lightsource (SSRL), a division of SLAC National Accelerator Laboratory and an Office of Science user facility operated by Stanford University for the U.S. Department of Energy. For work on molybdenum sulfide catalyst synthesis and development (J.D.B. and T.F.J.), we acknowledge support by the Center on Nanostructuring for Efficient Energy Conversion (CNEEC) at Stanford University, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award DE-SC0001060. C.T, M.L.N., F.A-P., J.K.N., and A.N. acknowledge financial support from the U.S. Department of Energy (DOE), Office of Basic Energy Sciences to the SUNCAT Center for Interface Science and Catalysis.

The transmission electron microscopy study was supported as part of the Center on Nanostructuring for Efficient Energy Conversion (CNEEC) at Stanford University, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under award no. DE-SC0001060. F.A.-P. acknowledges financial support from the U.S. Department of Energy, Office of Basic Energy Sciences to the SUNCAT Center for Interface Science and Catalysis. Transmission electron microscopy was performed at the Stanford Nano Shared Facilities (SNSF). J.D.B. and C.T. acknowledge support from the National Science Foundation Graduate Research Fellowship Program. Additionally, J.D.B. acknowledges support from a Stanford Graduate Fellowship.

126

4.9 References

1. J.D. Benck, Z.B. Chen, L.Y. Kuritzky, A.J. Forman, and T.F. Jaramillo. "Amorphous Molybdenum Sulfide Catalysts for Electrochemical Hydrogen Production: Insights into the Origin of their Catalytic Activity." ACS Catalysis, 2012. 2 (9): 1916-1923. http://dx.doi.org/10.1021/cs300451q 2. J.D. Benck, T.R. Hellstern, J. Kibsgaard, P. Chakthranont, and T.F. Jaramillo. "Catalyzing the Hydrogen Evolution Reaction (HER) with Molybdenum Sulfide Nanomaterials." ACS Catalysis, 2014. 4 (11): 3957-3971. http://dx.doi.org/10.1021/cs500923c 3. D. Merki, S. Fierro, H. Vrubel, and X.L. Hu. "Amorphous molybdenum sulfide films as catalysts for electrochemical hydrogen production in water." Chemical Science, 2011. 2 (7): 1262-1267. http://dx.doi.org/10.1039/c1sc00117e 4. D. Merki and X.L. Hu. "Recent developments of molybdenum and tungsten sulfides as hydrogen evolution catalysts." Energy & Environmental Science, 2011. 4 (10): 3878-3888. http://dx.doi.org/10.1039/c1ee01970h 5. D. Merki, H. Vrubel, L. Rovelli, S. Fierro, and X. Hu. "Fe, Co, and Ni ions promote the catalytic activity of amorphous molybdenum sulfide films for hydrogen evolution." Chemical Science, 2012. 3 (8): 2515-2525. http://dx.doi.org/10.1039/C2SC20539D 6. H. Vrubel, D. Merki, and X.L. Hu. "Hydrogen evolution catalyzed by MoS3 and MoS2 particles." Energy & Environmental Science, 2012. 5 (3): 6136-6144. http://dx.doi.org/10.1039/c2ee02835b 7. H. Vrubel and X. Hu. "Growth and Activation of an Amorphous Molybdenum Sulfide Hydrogen Evolving Catalyst." ACS Catalysis, 2013. 3 (9): 2002-2011. http://dx.doi.org/10.1021/cs400441u 8. H. Vrubel, T. Moehl, M. Gratzel, and X.L. Hu. "Revealing and accelerating slow electron transport in amorphous molybdenum sulphide particles for hydrogen evolution reaction." Chemical Communications, 2013. 49 (79): 8985-8987. http://dx.doi.org/10.1039/c3cc45416a 9. C.G. Morales-Guio and X. Hu. "Amorphous Molybdenum Sulfides as Hydrogen Evolution Catalysts." Accounts of Chemical Research, 2014. 47 (8): 2671-2681. http://dx.doi.org/10.1021/ar5002022 10. C.G. Morales-Guio, S.D. Tilley, H. Vrubel, M. Gratzel, and X. Hu. "Hydrogen evolution from a copper(I) oxide photocathode coated with an amorphous molybdenum sulphide catalyst." Nature Communications, 2014. 5. http://dx.doi.org/10.1038/ncomms4059 11. H.G.S. Casalongue, J.D. Benck, C. Tsai, R.K.B. Karlsson, S. Kaya, M.L. Ng, L.G.M. Pettersson, F. Abild-Pedersen, J.K. Nørskov, H. Ogasawara, T.F. Jaramillo, and A. Nilsson. "Operando Characterization of an Amorphous Molybdenum Sulfide Nanoparticle Catalyst during the Hydrogen Evolution Reaction." Journal of Physical Chemistry C, 2014. 118 (50): 29252-29259. http://dx.doi.org/10.1021/jp505394e 12. B. Lassalle-Kaiser, D. Merki, H. Vrubel, S. Gul, V.K. Yachandra, X. Hu, and J. Yano. "Evidence from in Situ X-ray Absorption Spectroscopy for the Involvement of Terminal Disulfide in the Reduction of Protons by an Amorphous Molybdenum Sulfide Electrocatalyst." Journal of the American Chemical Society, 2015. 137 (1): 314-321. http://dx.doi.org/10.1021/ja510328m 13. P.C.K. Vesborg, B. Seger, and I. Chorkendorff. "Recent Development in Hydrogen Evolution Reaction Catalysts and Their Practical Implementation." Journal of

127

Physical Chemistry Letters, 2015. 6 (6): 951-957. http://dx.doi.org/10.1021/acs.jpclett.5b00306 14. D.J. Li, U.N. Maiti, J. Lim, D.S. Choi, W.J. Lee, Y. Oh, G.Y. Lee, and S.O. Kim. "Molybdenum Sulfide/N-Doped CNT Forest Hybrid Catalysts for High-Performance Hydrogen Evolution Reaction." Nano Lett., 2014. 14 (3): 1228-1233. http://dx.doi.org/10.1021/nl404108a 15. A.B. Laursen, P.C.K. Vesborg, and I. Chorkendorff. "A high-porosity carbon molybdenum sulphide composite with enhanced electrochemical hydrogen evolution and stability." Chemical Communications, 2013. 49 (43): 4965-4967. http://dx.doi.org/10.1039/c3cc41945b 16. M. Salmeron and R. Schlögl. "Ambient pressure photoelectron spectroscopy: A new tool for surface science and nanotechnology." Surface Science Reports, 2008. 63 (4): 169-199. http://dx.doi.org/10.1016/j.surfrep.2008.01.001 17. S. Kaya, H. Ogasawara, L.-Å. Näslund, J.-O. Forsell, H.S. Casalongue, D.J. Miller, and A. Nilsson. "Ambient-pressure photoelectron spectroscopy for heterogeneous catalysis and electrochemistry." Catalysis Today, 2013. 205 (0): 101-105. http://dx.doi.org/10.1016/j.cattod.2012.08.005 18. H.G. Sanchez Casalongue, S. Kaya, V. Viswanathan, D.J. Miller, D. Friebel, H.A. Hansen, J.K. Nørskov, A. Nilsson, and H. Ogasawara. "Direct observation of the oxygenated species during oxygen reduction on a platinum fuel cell cathode." Nature Communications, 2013. 4. http://dx.doi.org/10.1038/ncomms3817 19. H.G. Sanchez Casalongue, M.L. Ng, S. Kaya, D. Friebel, H. Ogasawara, and A. Nilsson. "In Situ Observation of Surface Species on Iridium Oxide Nanoparticles during the Oxygen Evolution Reaction." Angewandte Chemie International Edition, 2014. 53 (28): 7169-7172. http://dx.doi.org/10.1002/anie.201402311 20. R. Arrigo, M. Hävecker, M.E. Schuster, C. Ranjan, E. Stotz, A. Knop-Gericke, and R. Schlögl. "In Situ Study of the Gas-Phase Electrolysis of Water on Platinum by NAP- XPS." Angewandte Chemie International Edition, 2013. 52 (44): 11660-11664. http://dx.doi.org/10.1002/anie.201304765 21. C. Zhang, M.E. Grass, A.H. McDaniel, S.C. DeCaluwe, F.E. Gabaly, Z. Liu, K.F. McCarty, R.L. Farrow, M.A. Linne, Z. Hussain, G.S. Jackson, H. Bluhm, and B.W. Eichhorn. "Measuring fundamental properties in operating solid oxide electrochemical cells by using in situ X-ray photoelectron spectroscopy." Nature Materials, 2010. 9 (11): 944-949. http://dx.doi.org/10.1038/nmat2851 22. W. Sigle. "Analytical Transmission Electron Microscopy." Annual Review of Materials Research, 2005. 35 (1): 239-314. http://dx.doi.org/10.1146/annurev.matsci.35.102303.091623 23. D. Williams and C.B. Carter. "The Transmission Electron Microscope," in "Transmission Electron Microscopy." 2009, Springer US. 3-22. http://dx.doi.org/10.1007/978-0-387-76501-3_1 24. S. Kaya, D. Schlesinger, S. Yamamoto, J.T. Newberg, H. Bluhm, H. Ogasawara, T. Kendelewicz, G.E. Brown, L.G.M. Pettersson, and A. Nilsson. "Highly Compressed Two-Dimensional Form of Water at Ambient Conditions." Scientific Reports, 2013. 3. http://dx.doi.org/10.1038/srep01074 25. G.A. Camacho-Bragado, J.L. Elechiguerra, and M.J. Yacaman. "Characterization of low dimensional molybdenum sulfide nanostructures." Materials Characterization, 2008. 59 (3): 204-212. http://dx.doi.org/10.1016/j.matchar.2006.12.009 26. R.F. Egerton. "An Introduction to EELS," in "Electron Energy-Loss Spectroscopy in the Electron Microscope." 2011, Springer US. 1-28. http://dx.doi.org/10.1007/978-1- 4419-9583-4_1

128

27. J.K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J.R. Kitchin, T. Bligaard, and H. Jónsson. "Origin of the Overpotential for Oxygen Reduction at a Fuel-Cell Cathode." Journal of Physical Chemistry B, 2004. 108 (46): 17886-17892. http://dx.doi.org/10.1021/jp047349j 28. A.A. Peterson, F. Abild-Pedersen, F. Studt, J. Rossmeisl, and J.K. Norskov. "How copper catalyzes the electroreduction of carbon dioxide into hydrocarbon fuels." Energy & Environmental Science, 2010. 3 (9): 1311-1315. http://dx.doi.org/10.1039/C0EE00071J 29. A.L. Koh, E. Gidcumb, O. Zhou, and R. Sinclair. "Observations of carbon nanotube oxidation in an aberration-corrected environmental transmission electron microscope." ACS Nano, 2013. 7 (3): 2566-2572. http://dx.doi.org/10.1021/nn305949h 30. D.L. Griscom. "Thermal bleaching of x-ray-induced defect centers in high purity fused silica by diffusion of radiolytic molecular hydrogen." Journal of Non- Crystalline Solids, 1984. 68 (2-3): 301-325. http://dx.doi.org/10.1016/0022- 3093(84)90013-9 31. J.D. Benck, S.C. Lee, K.D. Fong, J. Kibsgaard, R. Sinclair, and T.F. Jaramillo. "Designing Active and Stable Silicon Photocathodes for Solar Hydrogen Production Using Molybdenum Sulfide Nanomaterials." Advanced Energy Materials, 2014. 4 (18). http://dx.doi.org/10.1002/aenm.201400739 32. Z. Chen, D. Cummins, B.N. Reinecke, E. Clark, M.K. Sunkara, and T.F. Jaramillo. "Core–shell MoO3–MoS2 Nanowires for Hydrogen Evolution: A Functional Design for Electrocatalytic Materials." Nano Letters, 2011. 11 (10): 4168-4175. http://dx.doi.org/10.1021/nl2020476

129

130

Chapter 5: Designing Active and Stable Silicon Photocathodes for Solar Hydrogen Production using Molybdenum Sulfide Nanomaterials

5.1 Abstract

Silicon is a promising photocathode for tandem photoelectrochemical water splitting devices, but efficient catalysis and long term stability remain key challenges. We demonstrate that with appropriately engineered interfaces, molybdenum sulfide nanomaterials can provide both corrosion protection and catalytic activity in silicon photocathodes. Using a thin MoS2 surface

+ protecting layer, we create MoS2-n p Si electrodes that show no loss in performance after 100 hours of operation. Transmission electron microscopy measurements show the atomic structure of the device surface, and reveal the characteristics of the MoS2 layer that provide both catalytic activity and excellent stability. In spite of a low concentration of exposed catalytically active sites, these electrodes possess the best performance of any precious metal-free silicon photocathodes with demonstrated long term stability to date. To further improve efficiency, we

2- incorporate a second molybdenum sulfide nanomaterial, highly catalytically active [Mo3S13] clusters. These photocathodes offer a promising pathway towards sustainable hydrogen production.

5.2 Introduction

As discussed in Chapter 1, photoelectrochemical (PEC) water splitting is a promising technology for using solar energy to produce hydrogen, an important industrial reagent and potential future fuel.1-3 Effective PEC water splitting devices must absorb a large fraction of incident sunlight, generate sufficient photovoltage to drive the electrochemical water splitting reactions, and possess long term stability under operating conditions. The most successful PEC water splitting devices have utilized multiple materials to meet the various requirements, harnessing the best properties of each while mitigating their deficiencies.3-5 In Chapters 2 – 4,

131 electrocatalysts for the hydrogen evolution reaction were discussed. The semiconductor light absorber is another critical device component for full water splitting photoelectrodes. Many recent studies have focused on developing semiconductors intended for application in dual absorber devices because this configuration enables higher potential efficiencies than single absorber devices.6-8 These light absorbers have been combined with a variety of surface protecting and electrocatalyst materials to enhance stability and provide efficient hydrogen and oxygen evolution.1, 3

Silicon is a promising candidate small band gap semiconductor for application in a dual absorber water splitting system due to its nearly ideal band structure, excellent charge carrier transport properties, and relatively low cost.9, 10 Several previous reports have described silicon photocathodes with very good performance, but poor stability and inefficient surface catalysis remain key challenges.9, 11-18 Long term durability is an important concern, as silicon can etch or oxidize in aqueous electrolytes, leading to device failure.13, 17, 19 Most prior efforts to prevent silicon photoelectrode corrosion have focused on coating the device surface with a metal oxide

17, 20-23 protecting layer. Catalysis is another important challenge, as unmodified SiO2/Si surfaces have low activity for the hydrogen evolution reaction (HER).14 Many previous works have combined silicon with precious metal catalysts such as platinum, but the high price and scarcity of these materials could limit the scalability of these device structures.9, 11, 13-15, 17, 24 To create durable, efficient, and economical silicon photocathodes, additional strategies for overcoming these challenges must be developed.

Molybdenum sulfide nanomaterials have the potential to enhance both stability and catalysis in silicon photocathodes. Due to the low permability through the basal planes of its layered structure and its excellent stability in acidic electrolyte, a thin coating of crystalline MoS2 can provide effective surface protection,25, 26 as shown in a recent study by Chorkendorff and

27 28 coworkers. The edges of MoS2 layers are also active hydrogen evolution catalysts. To achieve high geometric area-normalized HER activity, it is necessary to carefully engineer molybdenum sulfide nanostructures to expose high densities of catalytic active sites.29, 30 Thus, while a thin, planar coating of MoS2 is likely to be an ideal structure for surface protection, this morphology of MoS2 may have sub-optimal catalytic activity.

In this chapter, we use two forms of molybdenum sulfide nanomaterials to confer both excellent durability and efficient catalysis to silicon photocathodes, creating highly active, precious

132 metal-free devices. First, we employ a thin MoS2 coating as a surface protecting layer. In spite of the planar structure of the MoS2, this device shows high activity and importantly, excellent stability as no loss in performance was observed after 100 hours of operation. Key to this achievement is appropriately engineering the interface between the silicon photocathode and the MoS2 protecting layer. Using advanced characterization techniques including cross sectional transmission electron microscopy (TEM), we image the device structure at the atomic scale, elucidating the high quality of the interfaces involved; fabricating the device with uniform layers, few defects, and with no observable oxidation is crucial to its stable operation. While the layered MoS2 structure provides excellent stability, our studies show that the photocurrent onset potential is limited by the low density of catalytic sites at the surface of the device. Using a

2- 30 second form of molybdenum sulfide, highly active [Mo3S13] nanocluster HER catalysts, we increase the density of catalytic sites at the surface and improve the photocurrent onset potential substantially. These molybdenum sulfide/silicon photocathodes present a promising, precious metal-free path towards matching the performance achieved in platinum/silicon devices.

5.3 Methods

5.3.1 Device Synthesis

+ MoS2-n p Si photocathodes were synthesized from single crystal boron doped p-type CZ Si

+ (100) wafers with 0.1 – 0.9 Ω cm resistivity (WRS Materials). MoS2-n Si dark catalysis controls were synthesized from conductive single crystal arsenic doped n-type CZ Si (100) wafers with 0.001 – 0.005 Ω cm resistivity (Nova Electronic Materials). All wafers were 100 mm in diameter and 500 – 550 μm thick.

A standard Radio Corporation of America (RCA) procedure was used to clean the wafers prior

+ + 31 to creating the surface n p junction in the MoS2-n p Si samples. The p-type wafers were immersed in 5:1:1 H2O:H2O2:NH4OH at 50°C for 10 minutes and rinsed, then immersed in 50:1

HF at room temperature for 30 seconds and rinsed, immersed in 5:1:1 H2O:H2O2:HCl at 70 °C for 10 minutes and rinsed, and dried in a spin rinse dryer.

The wafers were doped with phosphorus from a gaseous POCl3 source in a tube furnace at 900 °C for 10 minutes following a previous report.16 The surface-doped p-type wafers and the n- type wafers were then cleaned a second time using a similar RCA procedure. The wafers were

133 immersed in 5:1:1 H2O:H2O2:NH4OH at 50 °C for 10 minutes and rinsed, immersed in 5:1:1

H2O:H2O2:HCl at 70 °C for 10 minutes and rinsed, immersed in 6:1 buffered oxide etch at room temperature for 45 seconds and rinsed, and dried in a spin rinse dryer.

To deposit a thin film of molybdenum metal, the wafers were transferred to a DC magnetron sputter coater within 10 – 15 minutes after the RCA cleaning procedure to limit the formation of native SiO2. A thin layer of Mo metal with intended thickness of 3.6 nm was deposited onto the cleaned wafers at a rate of 7.2 nm per minute. After Mo deposition, the wafers were diced and the pieces were sulfidized in 90% H2/10% H2S gas in a tube furnace held at 250 °C for 1

+ + hour to create the surface MoS2 layer, completing the MoS2-n p Si and MoS2-n Si synthesis.

2- The molybdenum sulfide cluster HER catalyst, [Mo3S13] , was deposited onto the surface of

+ + some devices to make the Mo3S13-MoS2-n p Si photocathodes and Mo3S13-MoS2-n Si dark catalysis controls. The catalyst precursor, (NH4)2Mo3S13·H2O, was synthesized following a

30, 32 2- previously reported procedure and re-dispersed in methanol to create a 0.33 mM [Mo3S13]

+ + solution. The clusters were deposited from this solution onto the MoS2-n p Si and MoS2-n Si devices using a spray deposition technique to achieve a cluster loading of approximately 100 μg

-2 cm geo.

5.3.2 Physical and Chemical Characterization

X-ray photoelectron spectra were collected before and after electrochemical characterization using a Phi VersaProbe spectrometer with an Al Kα source. Binding energies were calibrated to the adventitious C 1s peak at 284.6 eV.

Transmission electron microscopy (TEM) specimens were prepared using a conventional cross- sectional method.33 The cross-section specimen was glued and mechanically ground to 60 μm. The specimen was further ground using a Gatan-656 dimple grinder to reach a thickness of less than 15 μm at the center. The specimen was ion milled to electron transparency using a Gatan- 695 Precision Ion Polishing System. A 5 kV argon ion beam was used to create a hole in the center of the specimen with an incident angle of 5°. Then, the argon ion energy was reduced to 0.5 kV to remove the surface amorphous layer. Imaging was performed using a FEI Tecnai Transmission Electron Microscope operated at 200 kV or with a FEI Titan 80-300 Environmental Transmission Electron Microscope operated at 80 kV.

134

Scanning electron microscopy (SEM) images were collected using the secondary electron detector on a FEI Magellan XHR microscope operated with a beam voltage of 5.0 kV and current of 50 pA.

Optical transmission measurements were performed using a Varian Cary 6000i UV-Vis-NIR spectrophotometer. To perform the optical absorption measurements, the sample was placed inside an integrating sphere and illuminated with white light from a 1000 W xenon lamp. All light that was not absorbed was collected and measured using an Ocean Optics Jaz EL 200-XR1 spectrometer. Reflection data were calculated assuming that the absorbed, transmitted, and reflected light fractions sum to one.

5.3.3 Electrochemical Characterization

Electrical contacts were made to the back of the silicon electrodes by scratching a Ga-In eutectic (Aldrich) through the surface oxide using a diamond tip scribe to create an ohmic junction. The Ga-In eutectic was connected to a stranded wire using copper tape with a conductive adhesive. The electrodes were mounted in inert epoxy (Loctite Hysol 9462), cured for at least 24 hours to protect the back contact from exposure to the electrolyte, and stored in air for up to one week prior to testing. The active area of each sample was measured using a digital photograph. Working electrode areas were 0.2 – 0.8 cm2.

Electrochemical measurements were performed in a three-electrode cell configuration using a Bio-Logic potentiostat (VSP) in a two compartment glass cell (Adams and Chittenden Scientific Glassware). The electrochemical testing setup is illustrated in Figure 5.1. Both sides of the cell contained 0.5 M sulfuric acid prepared with Millipore water (18.2 MΩ cm). The two sides were separated by a proton-conducting Nafion membrane. The working electrode (MoS2-Si) and reference electrode (Hg/Hg2SO4 in saturated K2SO4) were placed in one compartment, and the counter electrode (IrOx/Ir wire) was placed in the other compartment to minimize cross- contamination. For photoactivity measurements, the working electrode was illuminated through a fused silica window. The working electrode compartment was purged with H2 gas prior to each measurement. Potentials were calibrated to the reversible hydrogen electrode (RHE) scale using platinum working and counter electrodes in H2-purged electrolyte. RHE calibrations were performed in a separate cell to prevent platinum ion contamination.

135

Figure 5.1. Photograph of the electrochemical cell used to test the photocathodes.

The light source used for illuminated measurements was a 1000 W xenon arc lamp with a water filter to reduce output in the infrared region. The irradiance incident on the electrode surface was measured using an Ocean Optics Jaz EL 200-XR1 spectrometer and calibrated to match the AM1.5G solar spectrum34 for “one sun” simulated solar illumination. The output of the 1000 W xenon arc lamp was directed onto the photocathode samples using a silica fiber bundle optic. The spectral irradiance was measured using a 3.9 mm diameter cosine corrector connected via an optical fiber to an Ocean Optics Jaz EL 200-XR1 spectrometer. To match the AM1.5G standard solar spectrum, the position of the optical fiber was adjusted until the intensity at the

-2 photoelectrode location was 760 ± 5 W m , the integrated power in the AM1.5G standard at wavelengths shorter than 1030 nm, the long wavelength detection limit of the Jaz spectrometer. The integrated photon count in the 200 – 1030 nm wavelength range also matched the AM1.5G standard within ± 3%. Some error was introduced into this calibration due to spatial variations in the light intensity. The light output from the silica fiber bundle optic forms a circular spot with the highest intensity in the center and the intensity gradually decreasing towards the edges. The silicon electrodes were larger than the cosine corrector used for intensity calibration, so the

136 edges of these electrodes were illuminated with a slightly lower intensity than the center of the sample. Therefore the actual average irradiance was slightly lower than the desired values. The spectral irradiance of the 1000 W Xe arc lamp compared to the AM1.5G standard solar spectrum is displayed in Figure 5.2.

Figure 5.2. Spectral irradiance of 1000 W xenon lamp compared to the AM1.5G standard solar spectrum.34 Irradiance values were calibrated and measured as described above. a. 1000 W Xe light source calibrated to 760 W m-2 at wavelengths shorter than 1030 nm and “one sun” AM1.5G standard. This illumination condition was used for all tests unless otherwise noted. b. 1000 W Xe light source calibrated to 1000 W m-2, 760 W m-2, and 760 W m-2 with a 700 nm long-pass filter resulting in an irradiance of 274 W m-2 in the red and infrared portions of the spectrum. These illumination conditions were used to collect the linear sweep voltammograms displayed in Figure 5.26.

137

The performance the photoactive electrodes was measured using linear sweep voltammograms (LSVs.). One or two LSVs were used to activate the electrodes and test for proper electrical connection. Then the current-voltage performance was measured using a LSV from +0.5 to -0.4 V vs. RHE at 10 mV s-1. The catalytic activity of the dark catalysis control samples prepared on the n+ doped Si was measured using LSVs. with no illumination. The potential was swept from 0.1 to -0.4 V vs. RHE at 10 mV s-1. We define the photocurrent or dark current onset as the potential at which a current density of 0.5 mA cm-2 is achieved.

Photon-to-current conversion efficiency measurements were made using monochromatic radiation (FWHM 18 ± σ = 1.7 nm) obtained using a monochromator (74100 Oriel Cornerstone). Appropriate long-pass filters were applied to eliminate light from higher order diffractions. During these measurements, the working electrode potential was held at 0 V vs. RHE and the wavelength of light was varied from 350 nm to 900 nm in 50 nm increments. The light was chopped on and off at a rate of 0.2 Hz. The dark current was subtracted from the illuminated current to find the photocurrent. The average photocurrent over three chopping cycles was used to calculate the photon-to-current conversion efficiency.

To measure the stability of the photoelectrodes, after the irradiance was calibrated to “one sun” as described above, the activity of the electrodes was measured periodically using LSVs +0.5 to -0.4 V vs. RHE at 10 mV s-1. Between these measurements, the potential was held at 0 V vs. RHE and the current was measured and averaged in 60 s increments. The electrolyte was replaced and the sample was rinsed every ~25 hours during the stability measurements. The irradiance changed by up to 10% during the 100 hour measurements due to drift in the lamp output.

5.3.4 Hydrogen Quantification

The Faradaic efficiency of hydrogen production on the illuminated photocathodes was measured using a commercial electrochemical hydrogen detector (Unisense H2-500 probe and 4-channel amplifier) in a gas-tight two compartment electrochemical cell designed specifically for product quantification pictured in Figure 5.3 (Adams and Chittenden Scientific Glassware). Before each measurement, the cell was filled with 0.5 M sulfuric acid electrolyte and the hydrogen detector was calibrated by injecting known volumes of hydrogen into the nitrogen-purged and sealed cell. The quantification measurements were performed in a three electrode configuration as

138 described above. The photocathode working electrode was held at a constant potential of 0 V vs. RHE and illuminated with a white LED until the desired amount of charge was passed through the circuit. After allowing the hydrogen concentration in the liquid and gas phases to equilibrate, the hydrogen concentration was measured. The experimental value was compared to the expected concentration, which was calculated by assuming 100% of the charge passed was directed to hydrogen evolution. The Faradaic efficiency was computed by dividing the slope of a linear regression fit of the measured data by the slope of the predicted values. The standard error was estimated to be ±6.2% based on control measurements of hydrogen evolved from a platinum working electrode.

Figure 5.3. Gas-tight two-compartment electrochemical cell used for H2 quantification measurements.

139

5.4 Results and Discussion

5.4.1 Physical Characterization Using Electron Microscopy

+ A diagram of the MoS2-n p Si photocathode is shown in Figure 5.4a. The device includes a surface n+p junction to maximize the photovoltage produced under illumination.9, 16, 35 As detailed in the previous section, the MoS2 was synthesized by thermally sulfidizing a sputtered thin film of Mo metal at 250 °C. This procedure was chosen to create a thin, conformal surface coating of MoS2 while minimizing the risk of oxidizing the silicon.

+ + Figure 5.4. MoS2-n p Si device structure. a. Diagram of MoS2-n p Si device. b-d. Cross-sectional + transmission electron micrographs of MoS2-n p Si surface region collected with a FEI Tecnai TEM operated at 200 kV. These images illustrate that the surface of the device consists of MoS2, Mo metal, amorphous MoxSi, and Si layers with no evidence of an insulating SiO2 layer. The arrows in panel d indicate locations where HER-active MoS2 edge sites are likely exposed at the device surface.

140

To confirm that the desired structure was successfully created, we performed cross-sectional transmission electron microscopy (TEM). As shown in Figure 5.4b, the TEM images reveal four layers at the device surface. The top layer corresponds to MoS2, as evidenced by the lattice fringes observed in Figure 5.4c with a spacing of 0.62 nm, very close to the expected MoS2

36 interlayer distance of 0.614 nm. The orientation of the lattice fringes shows that the MoS2 basal planes are aligned parallel to the silicon (100) surface, which is the thermodynamically favored orientation and is consistent with observations of other MoS2 materials prepared via

26 thermal sulfidization. Although the surface is mostly covered with MoS2 basal planes, some regions denoted by the arrows in Figure 5.4d show features indicative of overlapping or ending

37 MoS2 layers. These features likely expose MoS2 edges, which are the HER active sites.

In Figure 5.4b, below the MoS2 we observe a dark layer corresponding to unsulfidized molybdenum metal. Beneath the metal, there is a lighter layer that likely arises from amorphous molybdenum silicide (MoxSi), which has been observed in previous studies when molybdenum

38, 39 is sputtered onto silicon. The presence of the Mo metal and MoxSi layers, both of which are metallic conductors, may benefit device performance by improving adhesion of the MoS2 and/or creating a tunneling ohmic contact with the n+ doped silicon surface. Conversely, the Mo and MoxSi layers may also have some negative impacts on device performance, particularly due to undesired light absorption despite their thin nature, as discussed below. Finally, the TEM images also reveal crystalline Si underneath the Mo, MoxSi, and MoS2 layers. The fringe spacing is 0.54 nm, equal to the Si lattice parameter.40 The TEM images do not show any evidence of interfacial SiO2. The layered device structure is confirmed using aberration-corrected TEM as shown in Figure 5.5 as well as scanning transmission electron microscopy (STEM) elemental maps as shown in Figure 5.6. The scanning electron micrographs in Figure 5.7 show that the surface of the device is flat and featureless, as expected.

141

+ Figure 5.5. Aberration-corrected cross-sectional TEM images of MoS2-n p Si device surface region. Aberration corrected imaging was performed using a FEI Titan 80-300 Environmental Transmission Electron Microscope equipped with a spherical aberration (Cs) corrector in the image-forming (objective) lens and operated at 80 kV to minimize any damage which might be induced by the incident electron beam. Under these imaging conditions, all Mo-containing layers appear darker than the Si. Otherwise these images reveal an identical structure to those shown in Figure 1, which also suggests that the beam did not alter the sample morphology during imaging at 200 kV.

142

+ Figure 5.6. a. Cross-sectional scanning transmission electron microscopy (STEM) image of MoS2-n p Si device surface region collected using a FEI Titan 80-300 Environmental TEM operated at 80 kV. The yellow box indicates the region shown in the elemental maps created using electron energy loss spectroscopy (EELS) in the other panels. For these measurements, the probe size was 0.9 nm, the elemental map pixel width was 0.3 nm, and the acquisition time of each EELS spectrum was 0.05 s. b. Overlaid elemental maps of Si (blue), Mo (green), and S (red). c. Elemental map of Si. d. Elemental map of Mo. e. Elemental map of S. These images provide further confirmation of the layered structure observed in the bright field TEM images in Figures 5.4 and 5.5.

143

+ Figure 5.7. Scanning electron microscopy (SEM) images of MoS2-n p Si device surface viewed from the top. Before testing, the sample appears flat and featureless, as expected. After the 100 hour stability test, some particles are observed on the sample surface. These likely arise from residual sulfuric acid or contaminant salts from the electrolyte. The larger features may correspond to oxidized portions of the electrode (i.e. MoO3).

Based on the synthesis procedure and physical structure of this device, it is possible to construct

+ a schematic energy band diagram for the MoS2-n p Si photocathode in the dark as shown in Figure 5.8. The surface n-type doping of the silicon forms a buried p-n junction that should maximize the photovoltage produced by the device once it is illuminated. The Mo metal and/or metallic MoxSi are expected to form a tunneling ohmic contact to the silicon due to the high doping density of the n+ surface (on the order of 1020 cm-3).

144

+ Figure 5.8. Schematic energy band diagram of the MoS2-n p Si photocathode in the dark (not to scale).

5.4.2 Chemical Characterization Using X-ray Photoelectron Spectroscopy

To further characterize the composition and chemical state of the MoS2 coating and Si surface, we performed x-ray photoelectron spectroscopy (XPS) on the as-synthesized photocathode. As shown in Figure 5.9, the peaks observed in the Mo 3d region can be deconvoluted into two spin- orbit splitting doublets, indicating that the molybdenum exists in a mixture of oxidation states. Quantification reveals that the majority of the molybdenum measured by XPS exists in the 4+ oxidation state corresponding to MoS2 (61%), with the balance (39%) corresponding to metallic

Mo and MoxSi, which reflects the surface sensitivity of this technique in measuring the distinct layers observed by TEM in Figure 5.4.41 The binding energies of the molybdenum in the Mo metal and MoxSi differ by only ~0.2 eV, so these features are too close to enable further deconvolution.39, 42, 43 There is no evidence of Mo in a 6+ oxidation state, e.g. in the form of

26, 44, 45 MoO3. The S 2p region shows a single spin orbit splitting doublet with a binding energy

26, 44, 45 corresponding to sulfur in MoS2. Finally, one clear peak is observed in the Si 2p region with binding energy corresponding to bulk Si and MoxSi, which again cannot be deconvoluted because the Si binding energies for these compounds differ by only ~0.4 eV.39, 42, 43, 46, 47 A very

145 small feature at higher binding energy may indicate the presence of some Si in the 4+ oxidation

46, 47 state arising from SiO2, though that feature is within noise of the measurement. The intensity of all Si 2p peaks is low because this signal is attenuated by the Mo-containing layers above. Nevertheless, these data show that the Si-Mo interface was not substantially oxidized during the device synthesis and indicate that the surface protecting layer is effective at inhibiting the formation of SiO2 upon exposure to air.

+ Figure 5.9. X-ray photoelectron spectroscopy measurements of the MoS2-n p Si device before and after electrochemical stability measurement. Before testing, the structure contains Mo in the 4+ oxidation state corresponding to MoS2. The second doublet arises from the Mo metal and the amorphous MoxSi intermixing layer. The expected Mo binding energies for these species are too close to enable further 39, 42, 43 deconvolution. The sulfur binding energy matches the expected value for MoS2. The silicon exists 39, 42, 43 as elemental Si and MoxSi; these species cannot be resolved independently. After testing, the composition and chemical state remain very similar. Most of the molybdenum still exists as MoS2, Mo metal, or MoxSi, but some Mo 6+ corresponding to MoO3 is also detected. In addition to sulfur in the 2- MoS2, some S corresponding to SO4 groups from residual electrolyte is observed. There is no evidence of additional SiO2 formed after testing.

146

5.4.3 Photoelectrochemical Hydrogen Evolution Activity

+ The photoelectrochemical performance of the MoS2-n p Si photocathode structure was evaluated in a three electrode electrochemical cell with 0.5 M sulfuric acid electrolyte. The

+ MoS2-n p Si working electrode was illuminated with simulated solar irradiance calibrated to “one sun” based on the AM1.5G standard as described above.34 As shown in Figure 5.10a, the

+ MoS2-n p Si structure is a highly active, precious metal-free photocathode. The photocurrent onset potential, defined here as the potential required to reach -0.5 mA cm-2, is 0.32 V vs. the reversible hydrogen electrode (RHE), indicating a significant underpotential for hydrogen

+ evolution. To verify the performance enhancement offered by the MoS2-n p Si structure, we

+ also tested two control samples. First, the same MoS2-n p Si structure was tested with no illumination. A maximum current of -12 μA cm-2 was observed at -0.40 V vs. RHE, which shows that the device has a high shunt resistance and indicates that the semiconductor was not substantially defected due to unintentional doping or contamination during the device synthesis.

+ The second control, an illuminated n p Si with no Mo metal or MoS2, shows a maximum current

-2 of -0.6 μA cm at -0.40 V vs. RHE. The SiO2 surface layer results in low catalytic activity and increased series resistance in the electrochemical circuit, so the HER does not occur at a significant rate even at the most negative potentials tested. This control emphasizes the important role of the MoS2 as an active precious metal-free catalyst. We also performed product quantification measurements to confirm that the photocathode selectively evolves hydrogen. As

+ shown in Figure 5.10b, the Faradaic efficiency for H2 production on the MoS2-n p Si photocathode was measured to be effectively 100% within the experimental error.

147

+ Figure 5.10. Activity of MoS2-n p Si device and controls and hydrogen quantification measurements. a. + Linear sweep voltammograms (LSVs) of the MoS2-n p Si device under simulated solar illumination with + + two controls, the MoS2-n p Si device in the dark and an illuminated n p Si structure with no molybdenum + sulfide layer. b. Hydrogen quantification measurements on MoS2-n p Si photocathode. These measurements show that this device evolves hydrogen with effectively 100% Faradaic efficiency within the experimental error.

5.4.4 Optical Absorption, Reflection, and Transmission

Figure 5.10a shows that the saturation photocurrent density of the photocathode is ~17 mA cm- 2. This value, which is determined by the fraction of incident light absorbed by the silicon and the efficiency with which the absorbed photons are converted into hydrogen, is less than the predicted maximum of ~44 mA cm-2 based on a band gap of 1.12 eV due to incomplete light absorption in the silicon.34, 48

148

To quantify the fractions of incident light absorbed, reflected, and transmitted by the

+ photocathodes, optical spectra of the MoS2-n p Si and control samples were collected. A MoS2-

SiO2 control was synthesized in a similar fashion to the photocathodes with a quartz wafer used in place of the Si wafer. A bare n+p Si control sample was fabricated by cleaning and doping the p-type Si wafer but not carrying out the Mo metal deposition or sulfidization.

+ Figure 5.11. Optical absorption, reflection, and transmission spectra of the MoS2-n p Si photocathode.

+ As shown in Figure 5.11, only 44 – 61% of incident light is absorbed by the MoS2-n p Si photocathode in this wavelength range (350 – 900 nm); the rest of the light is reflected. Of the light absorbed by the device, a substantial fraction is absorbed in the Mo-containing layers. This is demonstrated by the MoS2-SiO2 control shown in Figure 5.12a. The light absorbed in the

MoS2, MoxSi, and Mo layers reduces the amount of light transmitted into the Si, where it can be converted into HER current. However, this control is insufficient to enable a precise determination of the fraction of incident light absorbed in the silicon. Due to differences between the Si and SiO2 wafers used to fabricate the photocathodes and these controls, the amorphous MoxSi intermixing layer may not be present in the controls. Additionally, reflection

+ likely occurs at multiple interfaces in the MoS2-n p Si photocathodes, as evidenced by the much higher reflection observed in the photocathode (Figure 5.11) compared to the MoS2-SiO2 control (Figure 5.12a). Complex optical effects may also arise from the nanoscale thickness of the surface coating layers, which are much smaller than the wavelengths of light used in these

149 experiments. Therefore a more detailed optical model would be necessary for further analysis. The increased absorption in the bare n+p Si wafer shown in Figure 5.12b demonstrates that the

MoS2 surface coating increases reflection losses in the photocathode.

Figure 5.12. Optical absorption, reflection, and transmission spectra of control samples. a. MoS2-SiO2 control b. Bare n+p Si control.

150

5.4.5 Photon-to-Current Conversion Efficiency

As displayed in Figure 5.13, the incident photon-to-current conversion efficiency (IPCE) reaches a maximum of 44% at 700 nm, while the absorbed photon-to-current conversion efficiency (APCE), calculated using the data from Figure 5.11, reaches a much higher maximum of 72%.

The trends observed in the IPCE and APCE measurements displayed in Figure 5.13 are similar to the quantum efficiency trends observed in silicon photovoltaics. The IPCE and APCE values at short wavelengths are low because these high energy photons have shallow absorption depths, and front surface-mediated recombination dominates conversion into hydrogen.15, 49, 50 The values increase at longer wavelengths as the absorption depth for these photons increases. The IPCE values are limited by low light absorption in the silicon due to surface reflection and parasitic absorption in the Mo and MoS2 layers. APCE values are lower than the desired ~100% because of absorption in the MoS2, Mo metal, and MoxSi layers. The APCE data account for losses due to surface reflection, but not parasitic absorption in the Mo and MoS2 layers. The “true” APCE based on the absorption in the silicon alone would yield even higher maximum values, likely approaching 100% for some wavelengths, but performing this calculation accurately is not possible in the absence of a complete optical model, as discussed above.

Despite the APCE values lower than the desired ~100%, these results confirm that the primary factor limiting the saturation photocurrent density is low light absorption in the Si. Strategies to reduce reflection due to planar surfaces and interfaces9, 15, 51 as well as to minimize the thickness of the MoxSi, Mo, and MoS2 layers could allow for substantial improvement in light absorption within the silicon.

151

Figure 5.13. Incident photon-to-current conversion efficiency (IPCE) and absorbed photon-to-current + conversion efficiency (APCE) measurements of MoS2-n p Si. Low absorption in the photoactive Si likely limits the IPCE and the saturation photocurrent density.

5.4.6 Photoelectrochemical Hydrogen Evolution Stability

+ To evaluate the stability of the MoS2-n p Si photocathode, we performed extensive illuminated electrochemical measurements and found that the device exhibits excellent stability. The constant potential measurements displayed in Figure 5.14a reveal that the HER current decreases over the first several hours of operation. This decrease is caused in part by bubble accumulation on the electrode surface, which blocks some active area, and by elevated temperature, which decreases the photovoltage generated by the semiconductor.52 Deactivation of catalytic sites due to poisoning or changes in surface structure may also contribute to the observed decrease in current. Nevertheless, most of the observed losses in activity are temporary, as the current increases to equal or greater than its original value after the electrolyte is replaced following each ~25 hr testing period. The linear sweep voltammograms (LSVs) in Figure 5.14b reveal that the device shows no loss in performance after 100 hours of testing. The photocurrent onset decreased slightly after 25 hours of testing, but increased to match its original value at the end of the experiment, as shown in Figure 5.15. The saturation photocurrent density increased slightly after extended testing, primarily as a result of slight changes in the illumination intensity due to drift in the lamp output.

152

+ Figure 5.14. Electrochemical stability measurements of MoS2-n p Si photocathode. a. Chronoamperometry measurement at E = 0 V vs. RHE. b. LSVs collected before (0 hr) and after (100 hr) the constant potential measurements. The device shows no loss in performance after 100 hours of testing.

153

+ -2 Figure 5.15. Stability of MoS2-n p Si photocathode irradiated with 760 W m white light. Each LSV was performed after changing the electrolyte and rinsing the electrode. After 25 hours, the photocurrent onset decreased by 0.03 V, but the onset increased to equal its original value after 100 hours of testing.

As shown in Figure 5.9, XPS spectra collected after the 100 hour stability test are similar to those observed in the untested sample. As before, the molybdenum exists primarily as MoS2 with some molybdenum metal and molybdenum silicide remaining. Here, some Mo (18%) is also observed in the 6+ oxidation state, which corresponds to MoO3. The relatively large full width at half maximum (FWHM) value of 3.0 eV for the Mo 6+ peaks suggests that some Mo 5+ species may also be present and incorporated into this doublet.26 It is likely that the oxide was formed in air after extraction from the electrolyte, as molybdenum oxide is highly unstable and dissolves rapidly in sulfuric acid.26 This oxide formation probably occurs at gaps in the

MoS2 coating, and may indicate that this device could be susceptible to corrosion over very long term operation (e.g. weeks, months, or years). In the S 2p region, in addition to the spin orbit splitting double corresponding to MoS2, a second doublet is observed with a binding

2- 53, 54 corresponding to sulfur in SO4 groups. This peak likely arises from residual sulfuric acid from the electrochemical measurements. As observed before testing, most of the silicon exists as elemental Si or MoxSi and a very small amount, just above detection limits, exists as SiO2.

These results show that the MoS2 coating provides good activity for the HER and effective protection for the photocathode surface, mitigating device degradation due to silicon oxidation or corrosion. This structure represents one of the most active precious metal-free silicon photocathodes to date with proven stability of more than 24 hours (in our case, with 100 hours

154 of proven stability). Nevertheless, there is room for improvement in the current-voltage performance. The best reported photocurrent onset for any silicon photocathode, approximately

9 + 0.56 V vs. RHE, was achieved using a Pt catalyst. The MoS2-n p Si device reported herein has a photocurrent onset of 0.32 V vs. RHE. Maximizing the onset potential is critical for successfully pairing this structure with a photoanode in a tandem water splitting device.

5.4.7 Identifying Performance Limitations

+ To improve the MoS2-n p Si photocurrent onset potential, we begin by identifying the device characteristics that limit this performance metric. The photocurrent onset potential is determined by the photovoltage produced by the illuminated semiconductor and the overpotential required

+ to drive the HER. To deconvolute these factors, we used a MoS2-n Si control to measure the dark catalytic activity of the MoS2 surface. As shown in Figure 5.16, the HER activity of the

MoS2 surface is reasonable, but lower than previously reported MoS2 HER catalysts. The MoS2-

+ n Si control has an onset potential of -0.25 V vs. RHE, while nanostructured MoS2 HER catalysts have onset potentials of -0.20 to -0.10 V vs. RHE.26, 29, 55

Figure 5.16. Dark catalysis controls demonstrating the HER activity of the flat MoS2 coating and the 2- improvement in onset potential resulting from the addition of the [Mo3S13] HER catalyst.

155

+ By calculating the difference in the onset potentials of the MoS2-n Si dark control and the

+ MoS2-n p Si under illumination, we infer that the semiconductor produces a photovoltage of 0.57 V, which is comparable to the best silicon photocathodes that have been previously

9 reported. These results demonstrate that the catalytic activity of the planar MoS2 surface is the primary factor that limits the photocurrent onset potential, thus engineering this interface further could lead to improved activity and efficiency of the photocathode.

+ The lower activity of the MoS2-n Si surface is not surprising in light of the low density of HER- active MoS2 edge sites observed in the cross-sectional TEM images shown in Figure 5.4. The

TEM image in Figure 5.4d suggests that there are likely ~5 – 10 places where MoS2 edge sites are exposed within the observed region. This corresponds to a linear density of ~0.1 – 0.2 exposed MoS2 edges per nm. The two-dimensional edge site number density cannot be inferred directly from this TEM measurement because these images do not provide information about the length of exposed MoS2 edges. To generate a lower bound estimate of the MoS2 edge site density, we assume that individual MoS2 edge sites exist in a square lattice on the cathode surface with a linear density of 0.1 per nm. This corresponds to an active site density of 1012

2 MoS2 edge sites per cm . The true active site density is likely substantially larger, as each MoS2 edge probably exposes many active sites. As an upper bound estimate of the active site density, we note that all flat solid materials have a surface site density of approximately 1015 per cm2.56

Since the roughness of the MoS2 layer is low and less than 100% of surface sites are catalytically active, we conclude that the active site density must be less than 1015 per cm2. As a result, it is reasonable that this surface exhibits some activity for the HER, but substantially lower than

26, 29, 55 reported for the best three-dimensionally nanostructured MoS2 catalysts.

2- 5.4.8 Improving Hydrogen Evolution Catalysis with [Mo3S13] Clusters

To improve the photocurrent onset, we employ an additional molybdenum sulfide catalyst,

2- [Mo3S13] thiomolybdate nanoclusters, to increase the number of HER active sites. These clusters have a high catalytic activity due to a structural motif resembling that of MoS2 edge

30 2- sites. The [Mo3S13] clusters can also be synthesized from inexpensive precursors and deposited onto any desired electrode, making them an attractive candidate catalyst. We

-2 2- deposited ~100 μg cm of the [Mo3S13] clusters onto the cathodes to make Mo3S13-MoS2-Si devices, as shown schematically in Figure 5.17.

156

+ Figure 5.17. Diagram of Mo3S13-MoS2-n p Si photocathode structure.

Based on the cluster molar mass of 704.7 g mol-1, the mass loading of ~100 μg cm-2 corresponds

16 2- 2 to a cluster density of 8 × 10 [Mo3S13] per cm . While undercoordinated sulfur atoms are considered to be the HER active site on molybdenum sulfide catalysts, we estimate that each

2- [Mo3S13] cluster provides up to three HER active sites, based on the number of Mo atoms per

30 cluster. The MoS2 layer contributes to the HER activity as well, but the number of MoS2 edge sites is approximately two orders of magnitude lower, and therefore does not contribute substantially to the active site density. Therefore, we assume an upper bound active site density

17 2 + of 2.4 × 10 sites per cm on the Mo3S13-MoS2-n p Si photocathodes. However, the scanning

2- electron micrographs in Figure 5.18 show that the [Mo3S13] clusters form a rough film upon deposition onto the electrode surfaces. As a result, it is likely that many of these active sites are not exposed to the electrolyte. As an approximate lower bound for the active site density, we assume that only one in ten active sites exposed to the electrolyte, resulting in an active site density of that 2.4 × 1016 per cm2. This range represents a ~1.5 – 2.5 order of magnitude increase

+ in active site density compared to the MoS2-n p Si samples.

157

+ Figure 5.18. Scanning electron microscopy (SEM) images of Mo3S13-MoS2-n p Si device surface viewed 2- from the top. Before testing, some regions of the surface are covered with a rough film of the [Mo3S13] clusters, but the coating is not perfectly uniform. This suggests that it may be possible to increase the + activity of the Mo3S13-MoS2-n p Si even further by improving the uniformity of the cluster coating. After 2- electrochemistry, very few [Mo3S13] clusters remain.

158

+ 2- The Mo3S13-MoS2-n Si dark catalysis control, shown in Figure 5.16, reveals that the [Mo3S13] clusters increase the HER onset potential to -0.17 V, an improvement of 80 mV. Based on the

2- 30 previously measured [Mo3S13] Tafel slope of 40 mV per decade, this improvement corresponds to a ~2 order of magnitude increase in the active site density, as expected based on the Tafel equation, which relates current and overpotential for electrocatalytic reactions such as the HER in a kinetically limited potential regime:[5]

  a  b  l og (i)

In this equation, η is the overpotential, i is the catalytic current, a is the intercept, and b is the

Tafel Slope. The intercept a can be defined in terms of i0, the exchange current density, as follows:57, 58

a  b  l og (i0 )

Then the Tafel equation becomes:

  b  log(i)  b  log(i0 )

Assuming each active site has the same turn over frequency, the exchange current density, i0, scales in proportion to the number of active sites. Thus, the Tafel equation shows that for each order of magnitude increase in the number of active sites, the overpotential required to reach a particular current density should decrease by the Tafel slope, b.

+ In this case, the active site density at the Mo3S13-MoS2-n p Si surface is ~1.5 – 2.5 orders of

+ magnitude greater than the MoS2-n p Si, so the onset potential should shift by 1.5b to 2.5b, where b = 40 mV per decade.30 Thus the expected increase in onset potential is 60 – 100 mV. The observed onset potential increase of 80 mV in the dark catalysis controls in Figure 5.16 falls within this range, confirming that the expected active site density increase was in fact achieved.

+ As shown in Figure 5.19, the photocurrent onset potential of the illuminated Mo3S13-MoS2-n p Si device is also increased by 80 mV to 0.40 V vs. RHE. This shows that the addition of the

2- [Mo3S13] clusters was an effective strategy for improving the performance of this photocathode.

159

+ + + 9 Figure 5.19. Linear sweep voltammograms (LSVs) of MoS2-n p Si, Mo3S13-MoS2-n p Si, and Pt-n p Si 2- devices under simulated solar illumination. Incorporating the [Mo3S13] HER catalyst increases the onset potential by 80 mV in both the dark control and the photocathode.

In addition to improving the photocurrent onset potential through improved HER catalysis, the addition of the clusters should block some of the incoming light. However, we observe little

+ difference in the saturation photocurrent density between the MoS2-n p Si and Mo3S13-MoS2-

+ n p Si devices in Figure 5.19. The optical spectra in Figure 5.20a show that the Mo3S13-MoS2-

+ + n p Si photocathodes do absorb slightly more light than the MoS2-n p Si, particularly at wavelengths shorter than 600 nm, but this difference is small. This light is likely absorbed by

2- the [Mo3S13] clusters as opposed to the Si, and as such is not converted to hydrogen evolution current. The lower reflection observed for the Mo3S13-MoS2-SiO2 control in Figure 5.20b emphasizes the differences between the photocathode and control samples that make it difficult to quantify the absorption in each layer without a more advanced optical model. The fractions of incident light absorbed by all the photocathodes and control samples tested here are shown

2- in Figure 5.21 to emphasize the small increase in light absorption caused by the [Mo3S13] clusters.

160

+ Figure 5.20. a. Optical spectra of Mo3S13-MoS2-n p Si. Optical absorption, reflection, and transmission + spectra. a. Mo3S13-MoS2-n p Si photocathode b. Mo3S13-MoS2-SiO2 control.

161

Figure 5.21. Comparison of the fraction of incident light absorbed by all photocathodes and control samples tested in this study.

+ The photon-to-current conversion efficiency curves for the Mo3S13-MoS2-n p Si device, shown

+ in Figure 5.22a, are similar to those for the MoS2-n p Si device in Figure 5.13. These data

2- explain why undesired light absorption in the [Mo3S13] clusters has little impact on the saturation photocurrent density. The clusters primarily absorb blue light, which is utilized inefficiently by the silicon.

+ Hydrogen quantification measurements show that the Mo3S13-MoS2-n p Si photocathode also produces H2 selectively. The data in Figure 5.22b indicate that the Faradaic efficiency for H2

+ production on the Mo3S13-MoS2-n p Si photocathode is 95%, within the experimental error of 100%.

162

Figure 5.22. a. Photon-to-current conversion efficiency measurements. b. Hydrogen product + quantification. Both a. and b. show data for Mo3S13-MoS2-n p Si photocathodes.

+ 2- A key limitation of the Mo3S13-MoS2-n p device is poor adhesion of the [Mo3S13] catalyst. After four hours of testing, the performance of this device decreases to match the performance

+ 2- observed in the MoS2-n p Si device with no [Mo3S13] , as shown in Figure 5.23. After a second sample was tested for 100 hours, the device remained stable with performance matching the

+ MoS2-n p Si photocathode’s activity, as shown in Figure 5.24. The observed degradation in

+ 2- performance to match the activity of the MoS2-n p Si devices suggests that the [Mo3S13] clusters may delaminate from the underlying MoS2 surface.

163

+ -2 Figure 5.23. Stability of Mo3S13-MoS2-n p Si photocathodes irradiated with 760 W m white light. a. Linear sweep voltammograms taken every 10 minutes for 4 hours. b. Current measured with the photocathode held at 0 V vs. RHE between LSV measurements. The photocurrent onset potential and saturation photocurrent density both decrease during the course of testing due to the accumulation of 2- bubbles on the electrode surface as well as the desorption of the [Mo3S13] clusters. The electrolyte was changed and the electrode was rinsed to remove bubbles before the final linear sweep voltammogram was collected after 4 hours of testing. The activity of this electrode matches the performance observed in the + 2- MoS2-n p Si photocathodes, indicating that most, if not all, of the [Mo3S13] clusters desorbed.

164

+ -2 Figure 5.24. Stability of Mo3S13-MoS2-n p Si photocathodes irradiated with 760 W m white light. a. Linear sweep voltammograms taken every 25 hours for 100 hours. b. Current measured with the photocathode held at 0 V vs. RHE between LSV measurements. As expected, the linear sweep voltammograms show that the activity decreases after 25 hours of illuminated testing. In subsequent measurements, the activity remains stable and matches the performance observed in stability tests of the + MoS2-n p Si photocathodes. The decrease in performance after 25 hours is caused by the desorption of 2- the [Mo3S13] clusters.

165

X-ray photoelectron spectra collected before and after the four hour stability test, shown in Figure 5.25, confirm that the decrease in performance is primarily attributable to cluster desorption from the underlying MoS2 surface. Before testing the Mo 3d region reveals that most of the Mo probed by XPS exists in the 4+ oxidation state (61%) with the rest existing as Mo

39, 42, 43 metal or MoxSi, which are indistinguishable in these measurements. In this case, the 4+

2- oxidation state corresponds to Mo in MoS2 and in [Mo3S13] . The S 2p region reveals a broad peak indicative of sulfur in a mixture of multiple oxidation states. This peak is deconvoluted into two spin orbit splitting doublets, one of which corresponds to sulfur in MoS2 and the second

2- 2- corresponds to [Mo3S13] . In reality, the sulfur in [Mo3S13] is expected to exist in a mixture of oxidation states, but the resolution of these XPS spectra are insufficient to enable a meaningful deconvolution of the observed sulfur features into multiple doublets, so a single S 2p doublet is

2- + assigned to [Mo3S13] to emphasize that the differences between the Mo3S13-MoS2-n p device

+ and the MoS2-n p Si spectra in Figure 5.9 arise from the addition of the cluster catalyst. As expected, the Si 2p peak reveals primarily bulk Si or MoxSi with a very small contribution from

SiO2.

+ Figure 5.25. X-ray photoelectron spectra of Mo3S13-MoS2-n p Si before and after electrochemical testing.

166

After electrochemistry, the Mo 3d region looks similar, but displays an additional doublet, indicating that a small fraction of the Mo (9%) now exists in the 6+ oxidation state corresponding to MoO3. In the S 2p region, the doublet arising from MoS2 remains, while the

2- doublet corresponding to [Mo3S13] has been greatly reduced in intensity. This confirms that

2- most of the [Mo3S13] desorbed from the sample surface during the electrochemical

2- characterization. A third small doublet arising from residual SO4 groups from the electrolyte is also observed. Finally, the Si 2p peaks are observed to have particularly low intensity in this region, but these data provide no evidence of the formation of additional SiO2. Further work to anchor the clusters in a more stable manner could lead to improved long-term durability of the

+ Mo3S13-MoS2-n p Si device.

+ + Figure 5.26 shows the performance of the MoS2-n p Si and Mo3S13-MoS2-n p Si photocathodes under different illumination conditions. Each sample was illuminated using 1000 W m-2 white light, 760 W m-2 white light, and 274 W m-2 red/infrared light. These irradiance values refer to the integrated power at wavelengths shorter than 1030 nm. Spectra of the incident light under

-2 each illumination condition are shown in Figure 5.2. The 274 W m red/infrared light illumination condition is designed to simulate the spectral irradiance of these Si photocathodes when paired with a ~1.8 eV band gap photoanode in a stacked dual absorber configuration. This illumination condition is similar, though not identical, to that employed by Chorkendorff and coworkers in their recent molybdenum sulfide/silicon photocathode study.27 These measurements show that, as expected, the saturation photocurrent density varies substantially with the illumination intensity, but the photocurrent onset potential varies much less significantly due to the logarithmic dependence of photovoltage on illumination intensity expected for a p-n junction device.59

167

Figure 5.26. Linear sweep voltammograms under different illumination conditions.

For the sake of comparison to a precious-metal based Si photocathode, Figure 5.19 plots the

+ activity of the Mo3S13-MoS2-n p Si photocathode along with that of the most active reported silicon photocathode to date, a Pt-n+p Si structure from Lewis and coworkers consisting of platinum nanoparticles on n+p silicon.9 The difference in the saturation photocurrent density between these devices can mostly be attributed to undesired absorption in the Mo-containing

+ + layers in the photocathodes studied herein. The MoS2-n p Si and Mo3S13-MoS2-n p Si photocathode activities compare favorably, especially considering that they consist entirely of

+ earth-abundant elements, with proven 100-hour stability in acid for the MoS2-n p Si system. Nevertheless, further improvements in non-precious metal HER catalyst development and device integration will be needed close the gap to match the activity of Pt-based systems.

168

5.5 Conclusions

The molybdenum sulfide/silicon devices presented herein are highly active precious-metal free

+ photocathodes for solar hydrogen production. A photocathode consisting of MoS2-n p Si represents the most active precious metal-free silicon photocathode to date with proven stability of more than 24 hours, exhibiting negligible losses after 100 hours of operation. Further

2- integration of [Mo3S13] catalyst clusters improved the photocathode activity substantially

+ through enhancements in surface catalysis. While the activity of the Mo3S13-MoS2-n p Si photocathode compares favorably to that of precious metal based systems, further device development will be needed to match the activity of Pt-based devices. The saturation photocurrent density can be increased by improving light absorption, which could potentially be accomplished through surface texturing to reduce reflection, and by decreasing the thickness of the MoxSi and Mo metal layers to mitigate parasitic light absorption. The photocurrent onset potential of the photocathodes could also be improved by further increasing the number of catalytic sites and by improving catalyst adhesion, e.g. with three-dimensionally nanostructured

29 MoS2. These strategies provide a pathway to match the high activity observed in Pt-containing Si photocathodes but using solely earth-abundant materials that exhibit proven durability.

The work in this chapter demonstrates that active HER catalysts are critical to achieving high performance silicon photocathodes. However, engineering interfaces in the silicon to reduce recombination in increase the photovoltage could also yield substantial improvements in performance. This approach will be the focus of Chapter 6.

5.6 Copyright

Portions of this chapter from:

J.D. Benck, S.C. Lee, K.D. Fong, J. Kibsgaard, R. Sinclair, and T.F. Jaramillo. "Designing Active and Stable Silicon Photocathodes for Solar Hydrogen Production Using Molybdenum Sulfide Nanomaterials." Advanced Energy Materials, 2014. 4 (18). http://dx.doi.org/10.1002/aenm.201400739

Reproduced with permission from Wiley & Sons Publishing.

169

5.7 Author Contributions

Sang Chul Lee, Kara Fong, Jakob Kibsgaard, Robert Sinclair, and Thomas Jaramillo were coauthors of the journal article based on this work. Jesse Benck synthesized the photocathodes and performed the physical, chemical, and electrochemical characterization except the following: Sang Chul Lee performed the TEM measurements, Kara Fong performed the

2- hydrogen quantification measurements, and Jakob Kibsgaard synthesized the [Mo3S13] clusters. Jesse Benck wrote the manuscript and all authors participated in editing the paper.

5.8 Acknowledgments

This work was supported as part of the Center of Nanostructuring for Efficient Energy Conversion (CNEEC) at Stanford University, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under award no. DE-SC0001060. J.K. was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences for support under contract no. DE-SC0008685. Photocathode synthesis was performed in part at the Stanford Nanofabrication Facility, which is supported by the National Science Foundation through the National Nanotechnology Infrastructure Network under grant no. ECS-9731293. Spectroscopy and microscopy were performed at the Stanford Nano Center (SNC) and Stanford Nanocharacterization Laboratory (SNL). J.D.B. acknowledges support from the National Science Foundation Graduate Research Fellowship Program and a Stanford Graduate Fellowship. K.D.F. acknowledges support from the Stanford Vice Provost for Undergraduate Education through the Chemical Engineering Summer Research Program.

5.9 References

1. T. Bak, J. Nowotny, M. Rekas, and C.C. Sorrell. "Photo-electrochemical hydrogen generation from water using solar energy. Materials-related aspects." International Journal of Hydrogen Energy, 2002. 27 (10): 991-1022. http://dx.doi.org/10.1016/S0360-3199(02)00022-8 2. B.A. Pinaud, P.C.K. Vesborg, and T.F. Jaramillo. "Effect of Film Morphology and Thickness on Charge Transport in Ta3N5/Ta Photoanodes for Solar Water Splitting." Journal of Physical Chemistry C, 2012. 116 (30): 15918-15924. http://dx.doi.org/10.1021/jp3041742

170

3. M.G. Walter, E.L. Warren, J.R. McKone, S.W. Boettcher, Q.X. Mi, E.A. Santori, and N.S. Lewis. "Solar Water Splitting Cells." Chemical Reviews, 2010. 110 (11): 6446- 6473. http://dx.doi.org/10.1021/cr1002326 4. O. Khaselev and J.A. Turner. "A monolithic photovoltaic-photoelectrochemical device for hydrogen production via water splitting." Science, 1998. 280 (5362): 425- 427. http://dx.doi.org/10.1126/science.280.5362.425 5. S.Y. Reece, J.A. Hamel, K. Sung, T.D. Jarvi, A.J. Esswein, J.J.H. Pijpers, and D.G. Nocera. "Wireless Solar Water Splitting Using Silicon-Based Semiconductors and Earth-Abundant Catalysts." Science, 2011. 334 (6056): 645-648. http://dx.doi.org/10.1126/science.1209816 6. S. Hu, C.X. Xiang, S. Haussener, A.D. Berger, and N.S. Lewis. "An analysis of the optimal band gaps of light absorbers in integrated tandem photoelectrochemical water-splitting systems." Energy & Environmental Science, 2013. 6 (10): 2984-2993. http://dx.doi.org/10.1039/c3ee40453f 7. L.C. Seitz, Z. Chen, A.J. Forman, B.A. Pinaud, J.D. Benck, and T.F. Jaramillo. "Modeling practical performance limits of photoelectrochemical water splitting based on the current state of materials research." ChemSusChem, 2014. 7 (5): 1372-1385. http://dx.doi.org/10.1002/cssc.201301030 8. M.F. Weber and M.J. Dignam. "Efficiency of Splitting Water with Semiconducting Photoelectrodes." Journal of the Electrochemical Society, 1984. 131 (6): 1258-1265. http://dx.doi.org/10.1149/1.2115797 9. S.W. Boettcher, E.L. Warren, M.C. Putnam, E.A. Santori, D. Turner-Evans, M.D. Kelzenberg, M.G. Walter, J.R. McKone, B.S. Brunschwig, H.A. Atwater, and N.S. Lewis. "Photoelectrochemical hydrogen evolution using Si microwire arrays." Journal of the American Chemical Society, 2011. 133 (5): 1216-1219. http://dx.doi.org/10.1021/ja108801m 10. C. Jacoboni, C. Canali, G. Ottaviani, and A. Alberigi Quaranta. "A review of some charge transport properties of silicon." Solid-State Electronics, 1977. 20 (2): 77-89. http://dx.doi.org/10.1016/0038-1101(77)90054-5 11. D.V. Esposito, I. Levin, T.P. Moffat, and A.A. Talin. "H2 evolution at Si-based metal- insulator-semiconductor photoelectrodes enhanced by inversion channel charge collection and H spillover." Nature Materials, 2013. 12 (6): 562-568. http://dx.doi.org/10.1038/nmat3626 12. Y.D. Hou, B.L. Abrams, P.C.K. Vesborg, M.E. Bjorketun, K. Herbst, L. Bech, A.M. Setti, C.D. Damsgaard, T. Pedersen, O. Hansen, J. Rossmeisl, S. Dahl, J.K. Norskov, and I. Chorkendorff. "Bioinspired molecular co-catalysts bonded to a silicon photocathode for solar hydrogen evolution." Nature Materials, 2011. 10 (6): 434-438. http://dx.doi.org/10.1038/nmat3008 13. C.U. Maier, M. Specht, and G. Bilger. "Hydrogen evolution on platinum-coated p- silicon photocathodes." International Journal of Hydrogen Energy, 1996. 21 (10): 859-864. http://dx.doi.org/10.1016/0360-3199(96)00023-7 14. J.R. McKone, E.L. Warren, M.J. Bierman, S.W. Boettcher, B.S. Brunschwig, N.S. Lewis, and H.B. Gray. "Evaluation of Pt, Ni, and Ni–Mo electrocatalysts for hydrogen evolution on crystalline Si electrodes." Energy & Environmental Science, 2011. 4 (9): 3573. http://dx.doi.org/10.1039/c1ee01488a 15. J. Oh, T.G. Deutsch, H.-C. Yuan, and H.M. Branz. "Nanoporous black silicon photocathode for H2 production by photoelectrochemical water splitting." Energy & Environmental Science, 2011. 4 (5): 1690. http://dx.doi.org/10.1039/c1ee01124c 16. B. Seger, A.B. Laursen, P.C. Vesborg, T. Pedersen, O. Hansen, S. Dahl, and I. Chorkendorff. "Hydrogen production using a molybdenum sulfide catalyst on a

171

titanium-protected n(+)p-silicon photocathode." Angewandte Chemie International Edition, 2012. 51 (36): 9128-9131. http://dx.doi.org/10.1002/anie.201203585 17. B. Seger, T. Pedersen, A.B. Laursen, P.C. Vesborg, O. Hansen, and I. Chorkendorff. "Using TiO2 as a conductive protective layer for photocathodic H2 evolution." Journal of the American Chemical Society, 2013. 135 (3): 1057-1064. http://dx.doi.org/10.1021/ja309523t 18. E.L. Warren, J.R. McKone, H.A. Atwater, H.B. Gray, and N.S. Lewis. "Hydrogen- evolution characteristics of Ni–Mo-coated, radial junction, n+p-silicon microwire array photocathodes." Energy & Environmental Science, 2012. 5 (11): 9653. http://dx.doi.org/10.1039/c2ee23192a 19. M. Pourbaix. "Atlas of electrochemical equilibria in aqueous solutions." 1966, Oxford; New York: Pergamon Press. 20. Y.W. Chen, J.D. Prange, S. Duhnen, Y. Park, M. Gunji, C.E. Chidsey, and P.C. McIntyre. "Atomic layer-deposited tunnel oxide stabilizes silicon photoanodes for water oxidation." Nature Materials, 2011. 10 (7): 539-544. http://dx.doi.org/10.1038/nmat3047 21. K. Jun, Y.S. Lee, T. Buonassisi, and J.M. Jacobson. "High photocurrent in silicon photoanodes catalyzed by iron oxide thin films for water oxidation." Angewandte Chemie International Edition, 2012. 51 (2): 423-427. http://dx.doi.org/10.1002/anie.201104367 22. M.J. Kenney, M. Gong, Y. Li, J.Z. Wu, J. Feng, M. Lanza, and H. Dai. "High- performance silicon photoanodes passivated with ultrathin nickel films for water oxidation." Science, 2013. 342 (6160): 836-840. http://dx.doi.org/10.1126/science.1241327 23. B. Seger, D.S. Tilley, T. Pedersen, P.C.K. Vesborg, O. Hansen, M. Grätzel, and I. Chorkendorff. "Silicon protected with atomic layer deposited TiO2: durability studies of photocathodic H2 evolution." RSC Advances, 2013. 3 (48): 25902. http://dx.doi.org/10.1039/c3ra45966g 24. P.C.K. Vesborg and T.F. Jaramillo. "Addressing the terawatt challenge: scalability in the supply of chemical elements for renewable energy." RSC Advances, 2012. 2 (21): 7933-7947. http://dx.doi.org/10.1039/c2ra20839c 25. E.W. Keong Koh, C.H. Chiu, Y.K. Lim, Y.-W. Zhang, and H. Pan. "Hydrogen adsorption on and diffusion through MoS2 monolayer: First-principles study." International Journal of Hydrogen Energy, 2012. 37 (19): 14323-14328. http://dx.doi.org/10.1016/j.ijhydene.2012.07.069 26. Z. Chen, D. Cummins, B.N. Reinecke, E. Clark, M.K. Sunkara, and T.F. Jaramillo. "Core–shell MoO3–MoS2 Nanowires for Hydrogen Evolution: A Functional Design for Electrocatalytic Materials." Nano Letters, 2011. 11 (10): 4168-4175. http://dx.doi.org/10.1021/nl2020476 27. A.B. Laursen, T. Pedersen, P. Malacrida, B. Seger, O. Hansen, P.C. Vesborg, and I. Chorkendorff. "MoS2: an integrated protective and active layer on n+p-Si for solar H2 evolution." Physical Chemistry Chemical Physics, 2013. 15 (46): 20000-20004. http://dx.doi.org/10.1039/c3cp52890a 28. T.F. Jaramillo, K.P. Jørgensen, J. Bonde, J.H. Nielsen, S. Horch, and I. Chorkendorff. "Identification of active edge sites for electrochemical H2 evolution from MoS2 nanocatalysts." Science, 2007. 317 (5834): 100-102. http://dx.doi.org/10.1126/science.1141483 29. J. Kibsgaard, Z. Chen, B.N. Reinecke, and T.F. Jaramillo. "Engineering the surface structure of MoS2 to preferentially expose active edge sites for electrocatalysis." Nature Materials, 2012. 11 (11): 963-969. http://dx.doi.org/10.1038/nmat3439

172

30. J. Kibsgaard, T.F. Jaramillo, and F. Besenbacher. "Building an appropriate active-site motif into a hydrogen-evolution catalyst with thiomolybdate [Mo3S13]2− clusters." Nature Chemistry, 2014. 6 (3): 248-253. http://dx.doi.org/10.1038/nchem.1853 31. W. Kern and D.A. Puotinen. "Cleaning Solutions Based on Hydrogen Peroxide for use in Silicon Semiconductor Technology." RCA Review, 1970. 31 (2): 187-206. 32. A. Müller, E. Krickemeyer, A. Hadjikyriacou, and D. Coucouvanis. "Molybdenum- Sulfur Clusters," in "Inorganic Syntheses." A.P. Ginsberg, Editor. 2009, Wiley. http://books.google.com/books?id=6yB2BSYfJUcC 33. J.C. Bravman and R. Sinclair. "The Preparation of Cross-Section Specimens for Transmission Electron-Microscopy." Journal of Electron Microscopy Technique, 1984. 1 (1): 53-61. http://dx.doi.org/10.1002/jemt.1060010106 34. ASTM International. "ASTM G173 - 03, Standard Tables for Reference Solar Spectral Irradiances: Direct Normal and Hemispherical on 37° Tilted Surface." 2013. West Conshohocken, PA. http://dx.doi.org/10.1520/G0173-03R12 35. E.L. Warren, S.W. Boettcher, M.G. Walter, H.A. Atwater, and N.S. Lewis. "pH- Independent, 520 mV Open-Circuit Voltages of Si/Methyl Viologen(2+/+) Contacts Through Use of Radial n(+)p-Si Junction Microwire Array Photoelectrodes." Journal of Physical Chemistry C, 2011. 115 (2): 594-598. http://dx.doi.org/10.1021/jp109147p 36. H. Tributsch and J.C. Bennett. "Electrochemistry and photochemistry of MoS2 layer crystals." Journal of Electroanalytical Chemistry, 1977. 81 (1): 97-111. http://dx.doi.org/10.1016/0368-1874(77)80287-6 37. T.F. Jaramillo, K.P. Jorgensen, J. Bonde, J.H. Nielsen, S. Horch, and I. Chorkendorff. "Identification of active edge sites for electrochemical H2 evolution from MoS2 nanocatalysts." Science, 2007. 317 (5834): 100-102. http://dx.doi.org/10.1126/science.1141483 38. K. Holloway, K.B. Do, and R. Sinclair. "Interfacial Reactions on Annealing Molybdenum-Silicon Multilayers." Journal of Applied Physics, 1989. 65 (2): 474-480. http://dx.doi.org/10.1063/1.343425 39. Y. Ijdiyaou, M. Azizan, E.L. Ameziane, M. Brunel, and T.A.N. Tan. "On the formation of molybdenum silicides in Mo-Si multilayers - the effect of Mo thickness and annealing temperature." Applied Surface Science, 1992. 55 (2-3): 165-171. http://dx.doi.org/10.1016/0169-4332(92)90106-8 40. H.W. King. "Crystal Structures and Lattice Parameters of Allotropes of the Elements," in "CRC handbook of chemistry and physics : a ready-reference book of chemical and physical data." W.M. Haynes, Editor. 2013, CRC Press: Boca Raton, FL. 41. W.A. Fraser, J.V. Florio, W.N. Delgass, and W.D. Robertson. "Surface sensitivity and angular dependence of X-ray photoelectron spectra." Surface Science, 1973. 36 (2): 661-674. http://dx.doi.org/10.1016/0039-6028(73)90410-X 42. W.A. Brainard and D.R. Wheeler. "An XPS study of the adherence of refractory carbide silicide and boride rf‐sputtered wear‐resistant coatings." Journal of Vacuum Science & Technology, 1978. 15 (6): 1800-1805. http://dx.doi.org/10.1116/1.569845 43. C.D. Wagner, D.E. Passoja, H.F. Hillery, T.G. Kinisky, H.A. Six, W.T. Jansen, and J.A. Taylor. "Auger and photoelectron line energy relationships in aluminum–oxygen and silicon–oxygen compounds." Journal of Vacuum Science & Technology, 1982. 21 (4): 933-944. http://dx.doi.org/10.1116/1.571870 44. J.H. Nielsen, L. Bech, K. Nielsen, Y. Tison, K.P. Jørgensen, J.L. Bonde, S. Horch, T.F. Jaramillo, and I. Chorkendorff. "Combined spectroscopy and microscopy of supported MoS2 nanoparticles." Surface Science, 2009. 603 (9): 1182-1189. http://dx.doi.org/10.1016/j.susc.2009.02.039

173

45. Z. Shuxian, W.K. Hall, G. Ertl, and H. Knözinger. "X-ray photoemission study of oxygen and nitric oxide adsorption on MoS2." Journal of Catalysis, 1986. 100 (1): 167-175. http://dx.doi.org/10.1016/0021-9517(86)90082-5 46. F.J. Grunthaner and J. Maserjian. "Experimental Observations of the Chemistry of the SiO2/Si Interface." IEEE Transactions on Nuclear Science, 1977. 24 (6): 2108-2112. http://dx.doi.org/10.1109/TNS.1977.4329175 47. G. Hollinger, Y. Jugnet, P. Pertosa, and T. Minh Duc. "X-ray photoelectron spectroscopy of thermally grown silicon dioxide films on silicon." Chemical Physics Letters, 1975. 36 (4): 441-445. http://dx.doi.org/10.1016/0009-2614(75)80276-4 48. Z. Chen, T.F. Jaramillo, T.G. Deutsch, A. Kleiman-Shwarsctein, A.J. Forman, N. Gaillard, R. Garland, K. Takanabe, C. Heske, M. Sunkara, E.W. McFarland, K. Domen, E.L. Miller, J.A. Turner, and H.N. Dinh. "Accelerating materials development for photoelectrochemical hydrogen production: Standards for methods, definitions, and reporting protocols." Journal of Materials Research, 2010. 25 (01): 3- 16. http://dx.doi.org/10.1557/JMR.2010.0020 49. A.G. Aberle. "Surface passivation of crystalline silicon solar cells: A review." Progress in Photovoltaics, 2000. 8 (5): 473-487. http://dx.doi.org/10.1002/1099- 159x(200009/10)8:5<473::aid-pip337>3.0.co;2-d 50. J.H. Zhao, A.H. Wang, M.A. Green, and F. Ferrazza. "19.8% efficient "honeycomb'' textured multicrystalline and 24.4% monocrystalline silicon solar cells." Applied Physics Letters, 1998. 73 (14): 1991-1993. http://dx.doi.org/10.1063/1.122345 51. A.I. Hochbaum and P.D. Yang. "Semiconductor Nanowires for Energy Conversion." Chemical Reviews, 2010. 110 (1): 527-546. http://dx.doi.org/10.1021/cr900075v 52. E. Skoplaki and J.A. Palyvos. "On the temperature dependence of photovoltaic module electrical performance: A review of efficiency/power correlations." Solar Energy, 2009. 83 (5): 614-624. http://dx.doi.org/10.1016/j.solener.2008.10.008 53. C.D. Wagner and J.A. Taylor. "Generation of XPS Auger Lines by Bremsstrahlung." Journal of Electron Spectroscopy and Related Phenomena, 1980. 20 (1-2): 83-93. http://dx.doi.org/10.1016/0368-2048(80)85008-0 54. X.-R. Yu, F. Liu, Z.-Y. Wang, and Y. Chen. "Auger parameters for sulfur-containing compounds using a mixed aluminum-silver excitation source." Journal of Electron Spectroscopy and Related Phenomena, 1990. 50 (2): 159-166. http://dx.doi.org/10.1016/0368-2048(90)87059-W 55. Y. Li, H. Wang, L. Xie, Y. Liang, G. Hong, and H. Dai. "MoS2 Nanoparticles Grown on Graphene: An Advanced Catalyst for the Hydrogen Evolution Reaction." Journal of the American Chemical Society, 2011. 133 (19): 7296-7299. http://dx.doi.org/10.1021/ja201269b 56. K.W. Kolasinski. "Surface science: Foundations of Catalysis and Nanoscience." 2008, Chichester: John Wiley & Sons Ltd. 57. A.J. Bard and L.R. Faulkner. "Electrochemical methods: fundamentals and applications." 2001, New York: Wiley. 58. R.P. O'Hayre. "Fuel cell fundamentals." 2006, Hoboken, NJ: John Wiley & Sons. 59. J. Nelson. "The physics of solar cells." Vol. 1. 2003: World Scientific.

174

Chapter 6: Photovoltaic-Inspired Silicon Photocathodes with

Interfaces Engineered for High Performance Solar H2 Production

6.1 Abstract

Many previous studies have focused on designing crystalline silicon photocathodes with enhanced catalytic activity and corrosion resistance, but developing photoelectrodes with improved photovoltage remains a critical challenge. In this work, we explore new Si photocathodes structures inspired by Si photovoltaics. These devices employ interface engineering strategies intended to minimize surface-mediated recombination. The best devices tested so far produce photovoltages of 600 mV, an improvement of 30 mV compared to conventional planar Si photocathodes with no surface passivation. The design strategies and fabrication procedures developed here provide a foundation for producing Si photocathodes with even higher performance, which will be necessary to enable high efficiency Si-based tandem water splitting systems.

6.2 Introduction

As discussed in Chapter 5, crystalline silicon is a particularly promising small band gap semiconductor for application in a dual absorber solar water splitting device due to its appropriate band structure, excellent charge transport properties, and relatively low cost.1-3 Many recent studies have focused on improving the catalysis and corrosion resistance of silicon photoelectrodes.3-51 We demonstrated that molybdenum sulfide nanomaterials can provide both hydrogen evolution reaction (HER) activity and stability in Si photocathodes.22 Many other researchers have combined Si with alternative HER catalysts including Pt, NiMo, and CoP as well as other protective coatings such as TiO2 to address these challenges, yielding many devices with very good performance.3, 5, 7, 8, 12, 14, 20, 25, 27, 33, 36, 46, 47 In spite of these significant

175 advancements, there remains substantial room for improvement in silicon photocathode performance, particularly in the photocurrent onset potential.

A silicon photocathode with a solid state “buried” p-n junction functions similarly to a silicon photovoltaic (PV) in series with a HER catalyst.15, 52-54 The photocurrent onset potential in this type of device depends on the activity of the HER catalyst as well as the photovoltage produced by the illuminated silicon. Conventional planar buried junction crystalline Si photoelectrodes have produced photovoltages of 480 – 580 mV,3, 13, 14, 16, 19, 20, 25, 33, 48, 55, 56 while the best crystalline Si PV produced a photovoltage of 740 mV under “one sun” AM1.5G illumination.57, 58 The superior performance of the PV is a result of advanced device engineering to reduce electron/hole pair recombination inside the silicon.59 Employing these strategies to reduce charge carrier recombination in Si photocathodes could result in a greatly improved photovoltage, potentially by as much as 160 mV. These gains will be necessary for high efficiency silicon-based dual absorber water splitting systems.60

Controlling the properties of Si surfaces and interfaces is critical for achieving excellent performance, as surface-mediated recombination is often a leading contributor to photovoltage losses in Si PVs and photoelectrodes.29, 33, 59, 61 Recently, some researchers have created Si water splitting photoelectrodes with doping profiles or surface passivation layers intended to reduce surface recombination.15, 28, 35, 40, 49, 62, 63 The best of these devices have produced photovoltages as high as 600 – 610 mV.35, 49 However, much more work is needed to realize the full potential of these strategies. Adapting design principles established through decades of research on Si PVs offers great potential for maximizing the performance of Si photoelectrodes.

In this chapter, we study new n+p-Si and n+pp+-Si photocathode structures inspired by high- performance Si PVs. We adapt strategies for interface passivation from PVs with the aim of minimizing surface-mediated carrier recombination, thereby increasing the photovoltage. We use photolithography to pattern catalyst disks on the photocathode surfaces, reducing the area of the metal/silicon contacts, which possess a high surface recombination velocity.59 Between the catalyst disks, the n+ Si emitters are coated with passivating layers of silicon oxide or silicon nitride to decrease the surface recombination velocity.59, 64-66 To reduce rear surface recombination, we employ a p+-doped back surface field created via aluminum deposition and rapid thermal annealing.67, 68 These photocathodes illustrate the power of PV-inspired device designs for improving photocathode performance.

176

6.3 Methods

6.3.1 Photocathode Fabrication

+ The “Generation 1” Pt disk/SiO2-n p Si photocathodes were fabricated using the procedure illustrated schematically in Figure 6.1. Each numbered step is described here.

1. The photocathodes were fabricated from prime grade, boron-doped p-type Czochralski single crystal Si wafers purchased from WRS Materials (San Jose, CA). These wafers were single side polished (SSP), 100 mm in diameter, 500 – 550 μm thick, (100) orientation, and 0.1 – 0.9 Ω cm resistivity.

2. The wafers were cleaned using a standard Radio Corporation of America (RCA)

69 procedure. The wafers were immersed in 5:1:1 H2O:H2O2:NH4OH at 50 °C for 10 min and rinsed, immersed in 50:1 HF at room temperature for 30 s and rinsed, immersed in 5:1:1

H2O:H2O2:HCl at 50 °C for 10 min and rinsed, and dried in a spin rinse dryer.

3. Within 10 min after the previous step, the wafers were transferred to a Tylan tube furnace

and doped with phosphorus from a gaseous POCl3 source at 900 °C for 10 min.

4. The wafers were cleaned using a RCA procedure. The wafers were immersed in 5:1:1

H2O:H2O2:NH4OH at 50 °C for 10 min and rinsed, immersed in 6:1 buffered oxide etch at

room temperature for 60 s and rinsed, immersed in 5:1:1 H2O:H2O2:HCl at 50 °C for 10 min and rinsed, and dried in a spin rinse dryer.

5. Within 10 min after the previous step, the wafers were transferred to a Thermco TMX 10000

furnace and oxidized in a dry O2 atmosphere at 1100 °C for 100 min. The thickness of the resulting oxide was determined to be 170 - 175 nm on both sides of the wafer using a J.A. Woollam variable angle spectroscopic ellipsometer.

6. The wafers were singed and primed with hexamethyldisilazane in a Yield Engineering

Systems oven at 150 °C for 35 min to improve photoresist adhesion to the SiO2 surface.

7. Photoresist was deposited on the front side of the wafers using a Silicon Valley Group automated resist coating system. Shipley 3612 positive photoresist was spin coated onto the

177

wafers at 3000 rpm for 60 s, resulting in a 1 μm thick layer. The wafers were baked on a hot plate at 90 °C for 60 s.

8. The photoresist was exposed for 1.1 s with 15 mW/cm2 365 nm wavelength light on a KarlSuss MA-6 contact aligner through a photomask consisting of a soda lime glass substrate with chrome coating purchased from Martin Photomask, Inc. (Escondido, CA). The mask pattern consists of 5 μm dark circles on a 25 μm pitch as shown in Figure 6.2.

9. The photoresist was developed on a Silicon Valley Group automated resist developer system. First, the wafers were baked on a hot plate at 110 °C for 60 s, then the resist was developed with Megaposit MF-26A developer for 60 s and rinsed in water, then the wafers were baked a second time on a hot plate at 110 °C for 60 s.

10. The thermal SiO2 was etched by immersing the wafers in a 20:1 buffered oxide etch solution for 7.5 min. The wafers were rinsed and dried in a spin rinse dryer. This procedure etched

holes in the front SiO2 through the patterned photoresist and removed all the SiO2 from the back of the wafer.

11. Within 20 min after the previous step, the wafers were transferred to a DC magnetron sputter coater. First, 10 nm of Ti was deposited at a rate of 4 nm/min. Then, 40 nm Pt was deposited at a rate of 34 nm/min.

12. The wafers were immersed in acetone for 60 s and then rinsed in acetone and isopropanol

and dried under a flow of N2. This procedure dissolved the photoresist and lifted off the Ti and Pt metal deposited on top of the polymer, leaving behind only the metal deposited in the open areas of the patterned photoresist.

13. The wafers were diced and electrical contacts were formed to the unpolished back side of the pieces by scratching a Ga-In eutectic (Aldrich) through the native oxide using a diamond tip scribe to create an ohmic junction. A stranded wire was pressed into the Ga-In eutectic and secured with copper tape.

14. These pieces were mounted in inert Loctite Hysol 9462 epoxy and cured for at least 24 hours to protect the back side of the device from exposure to the electrolyte during electrochemical testing.

178

+ Figure 6.1. Schematic diagram of the fabrication procedure for the “Generation 1” Pt disk/SiO2-n p Si photocathodes (not to scale).

Figure 6.2. Diagram of the photomask used in the fabrication of all the photocathodes in this chapter. Clear areas are shown in white, dark areas are shown in black, and dimensions are shown in blue. This pattern repeats across the mask to fill an 8 cm by 8 cm square.

179

+ The “Generation 2” Pt disk/Si3N4-n p Si photocathodes were fabricated using the procedure illustrated schematically in Figure 6.3. Each numbered step is described here.

1. The photocathodes were fabricated from prime grade, boron-doped p-type Czochralski single crystal Si wafers purchased from WRS Materials (San Jose, CA). These wafers were single side polished (SSP), 100 mm in diameter, 500 – 550 μm thick, (100) orientation, and 0.1 – 0.9 Ω cm resistivity.

2. The wafers were cleaned using a RCA procedure. The wafers were immersed in 5:1:1

H2O:H2O2:NH4OH at 50 °C for 10 min and rinsed, immersed in 50:1 HF at room

temperature for 30 s and rinsed, immersed in 5:1:1 H2O:H2O2:HCl at 50 °C for 10 min and rinsed, and dried in a spin rinse dryer.

3. Within 10 min after the previous step, the wafers were transferred to a Tylan tube furnace

and doped with phosphorus from a gaseous POCl3 source at 900 °C for 10 min.

4. The wafers were cleaned using a RCA procedure. The wafers were immersed in 5:1:1

H2O:H2O2:NH4OH at 50 °C for 10 min and rinsed, immersed in 6:1 buffered oxide etch at

room temperature for 60 s and rinsed, immersed in 5:1:1 H2O:H2O2:HCl at 50 °C for 10 min and rinsed, and dried in a spin rinse dryer.

5. Within 10 minutes after the previous step, the wafers were transferred to a Thermco TMX

10000 low pressure chemical vapor deposition furnace. A 150 nm layer of Si3N4 was

deposited on both sides of the wafers in 50 min at 785 °C under a mixture of SiCl2H2 and

NH3 gases at 250 mTorr. The thickness of the resulting layer was confirmed to be 150 – 155 nm using a J.A. Woollam variable angle spectroscopic ellipsometer.

6. The wafers were singed and primed with hexamethyldisilazane in a Yield Engineering

Systems oven at 150 °C for 35 min to improve photoresist adhesion to the Si3N4 surface.

7. Microposit LOL-2000 lift off layer polymer was deposited onto the front side of the wafers using a Headway manual spin coater at 3000 rpm for 60 s. The wafers were then baked at 150 °C for 3 min. The LOL-2000 polymer is not photoactive but can be dissolved in Megaposit MF-26A developer.

180

8. Photoresist was deposited on top of the LOL-2000 polymer on the front side of the wafers using a Silicon Valley Group automated resist coating system. Shipley 3612 positive photoresist was spin coated onto the wafers at 3000 rpm for 60 s, resulting in a 1 μm thick layer. The wafers were baked on a hot plate at 90 °C for 60 s.

9. The photoresist was exposed for 1.1 s with 15 mW/cm2 365 nm wavelength light on a KarlSuss MA-6 contact aligner through the photomask shown in Figure 6.2.

10. The photoresist was developed on a Silicon Valley Group automated resist developer system. First, the wafers were baked on a hot plate at 110 °C for 60 s, then the resist was developed with Megaposit MF-26A developer for 60 s and rinsed in water, then the wafers were baked a second time on a hot plate at 110 °C for 60 s. The LOL-2000 polymer was dissolved by the developer, undercutting the features in the exposed photoresist.

11. The Si3N4 on the front of the wafer was etched through the patterned photoresist using a 60

s dry etch in a mixture of CHF3 and O2 gases at 5 mTorr in a PlasmaTherm Versaline ICP high density plasma etch system.

12. Within 20 min after the previous step, the wafers were transferred to a DC magnetron sputter coater. A 50 nm layer of Pt was deposited at a rate of 34 nm/min.

13. The wafers were sonicated for 30 s in acetone and then rinsed in acetone and isopropanol

and dried under a flow of N2. This procedure dissolved the photoresist and lifted off the Ti and Pt metal deposited on top of the polymer, leaving behind only the metal deposited in the open areas of the patterned photoresist.

14. The LOL-2000 polymer cannot be not completely dissolved in acetone or isopropanol. To remove the remaining LOL-2000, the wafers were washed in Megaposit MF-26A developer for 90 s, rinsed in water, and dried.

15. The wafers were diced and electrical contacts were formed to the unpolished back side of

the pieces by scratching a Ga-In eutectic (Aldrich) through the Si3N4 using a diamond tip scribe to create an ohmic junction. A stranded wire was pressed into the Ga-In eutectic and secured with copper tape.

181

16. These pieces were mounted in inert Loctite Hysol 9462 epoxy and cured for at least 24 hours to protect the back side of the device from exposure to the electrolyte during electrochemical testing.

+ Figure 6.3. Schematic diagram of the fabrication procedure for the “Generation 2” Pt disk/Si3N4-n p Si photocathodes (not to scale).

+ + The “Generation 3” Pt disk/SiO2-n pp Si photocathodes were fabricated using the procedure illustrated schematically in Figure 6.4. Each numbered step is described here.

1. The photocathodes were fabricated from prime grade, boron-doped p-type Czochralski single crystal Si wafers purchased from University Wafer (South Boston, MA). These wafers were double side polished (DSP), 100 mm in diameter, 500 μm thick, (100) orientation, and 1 – 10 Ω cm resistivity.

2. The wafers were cleaned using a RCA procedure. The wafers were immersed in 5:1:1

H2O:H2O2:NH4OH at 50 °C for 10 min and rinsed, immersed in 50:1 HF at room

temperature for 30 s and rinsed, immersed in 5:1:1 H2O:H2O2:HCl at 50 °C for 10 min and rinsed, and dried in a spin rinse dryer.

182

3. Within 10 min after the previous step, the wafers were transferred to a Thermco TMX 10000

furnace and oxidized in a dry O2 atmosphere at 1100 °C for 2 hr. The thickness of the resulting oxide was determined to be 170 – 175 nm on both sides of the wafer using a J.A. Woollam variable angle spectroscopic ellipsometer.

4. The SiO2 on the front of the wafer was removed using a 50 s dry etch with a mixture of

CHF3, CF4, and Ar gases at 250 mTorr in an Applied Materials Precision 5000 magnetically- enhanced reactive ion etcher.

5. The wafers were cleaned using a RCA process. The wafers were immersed in 5:1:1

H2O:H2O2:NH4OH at 50 °C for 10 min and rinsed, immersed in 50:1 HF at room

temperature for 30 s and rinsed, immersed in 5:1:1 H2O:H2O2:HCl at 50 °C for 10 min and rinsed, and dried in a spin rinse dryer.

6. Within 10 min after the previous step, the wafers were transferred to a Tylan tube furnace

and doped with phosphorus from a gaseous POCl3 source at 900 °C for 10 min. The thermal

SiO2 on the back of the wafer served as a doping mask to so that only the front surface of

70-72 the wafer was exposed to phosphorus during this step. The efficacy of the SiO2 mask was confirmed using sheet resistance measurements, as shown in Table 6.1.

7. The wafers were cleaned using a RCA procedure. The wafers were immersed in 5:1:1

H2O:H2O2:NH4OH at 50 °C for 10 min and rinsed, immersed in 6:1 buffered oxide etch at

room temperature for 60 s and rinsed, immersed in 5:1:1 H2O:H2O2:HCl at 50 °C for 10 min and rinsed, and dried in a spin rinse dryer.

8. Within 10 min after the previous step, the wafers were transferred to a Thermco TMX 10000

furnace and oxidized in a dry O2 atmosphere at 1100 °C for 100 min. The thickness of the resulting oxide was determined to be 175 - 180 nm on the front side of the wafer using a J.A. Woollam variable angle spectroscopic ellipsometer.

9. The SiO2 on the back of the wafer was removed using a 70 s dry etch in a mixture of CHF3

and O2 gases at 5 mTorr in a PlasmaTherm Versaline ICP high density plasma etch system.

10. Within 20 min after the previous step, the wafers were transferred to an Innotec ES26C electron beam evaporator and a 300 nm thick layer of Al was deposited onto the back of the

183

wafers at a rate of 0.2 – 0.3 nm/s. The deposition was monitored in situ using a quartz crystal microbalance.

11. The wafers were annealed at 900 °C for 10 s under a forming gas atmosphere (5% H2, 95%

N2) using an All-Win 610 Rapid Thermal Process System to diffuse the Al into the silicon and form the p+-doped back surface field.

12. The wafers were cleaned by rinsing in acetone and isopropanol and dried under a flow of

N2 to remove particles from the front surface.

13. The wafers were singed and primed with hexamethyldisilazane in a Yield Engineering

Systems oven at 150 °C for 35 min to improve polymer adhesion to the Si3N4 surface.

14. Microposit LOL-2000 lift off layer polymer was deposited onto the front side of the wafers using a Headway manual spin coater at 3000 rpm for 60 s. The wafers were then baked at 170 °C for 3 min. The LOL-2000 polymer is not photoactive but can be dissolved in Megaposit MF-26A developer.

15. Photoresist was deposited on top of the LOL-2000 polymer on the front side of the wafers using a Silicon Valley Group automated resist coating system. Shipley 3612 positive photoresist was spin coated onto the wafers at 3000 rpm for 60 s, resulting in a 1 μm thick layer. The wafers were baked on a hot plate at 90 °C for 60 s.

16. The photoresist was exposed for 1.1 s with 15 mW/cm2 365 nm wavelength light on a KarlSuss MA-6 contact aligner through the photomask shown in Figure 6.2.

17. The photoresist was developed on a Silicon Valley Group automated resist developer system. First, the wafers were baked on a hot plate at 110 °C for 60 s, then the resist was developed with Megaposit MF-26A developer for 60 s and rinsed in water, then the wafers were baked a second time on a hot plate at 110 °C for 60 s. The LOL-2000 polymer was dissolved by the developer, undercutting the features in the exposed photoresist.

18. The SiO2 on the front of the wafer was etched through the patterned photoresist using a 70

s dry etch in a mixture of CHF3 and O2 gases at 5 mTorr in a PlasmaTherm Versaline ICP high density plasma etch system.

184

19. Within 20 min after the previous step, the wafers were transferred to a DC magnetron sputter coater. First, 10 nm of Ti was deposited at a rate of 4 nm/min. Then, 40 nm Pt was deposited at a rate of 34 nm/min.

20. The wafers were sonicated for 30 s in acetone and then rinsed in acetone and isopropanol

and dried under a flow of N2. This procedure dissolved the photoresist and lifted off the Ti and Pt metal deposited on top of the polymer, leaving behind only the metal deposited in the open areas of the patterned photoresist.

21. The LOL-2000 polymer cannot be not completely dissolved in acetone or isopropanol. To remove the remaining LOL-2000, the wafers were washed in Megaposit MF-26A developer for 90 s, rinsed in water, and dried.

22. The wafers were diced and electrical contacts were made to the Al on the back of the silicon electrodes using colloidal graphite and stranded wire. These electrodes were mounted in inert Loctite Hysol 9462 epoxy and cured for at least 24 hours to protect the back surface from exposure to the electrolyte during electrochemical testing.

185

+ + Figure 6.4. Schematic diagram of the fabrication procedure for the “Generation 3” Pt disk/SiO2-n pp Si photocathodes (not to scale).

186

Condition Sheet resistance

DSP p Si wafer, as received, front side 112 Ω/square

DSP p Si wafer, n-doped with no mask, front side 31 Ω/square

DSP p Si wafer, n-doped with SiO2 mask on the back, 25 Ω/square unmasked front side

DSP p Si wafer, n-doped with SiO2 mask on the back, 126 Ω/square masked back side (mask removed with buffered HF solution)

Table 6.1. Sheet resistance values measured for p-type silicon wafers as received and after various doping procedures. The sheet resistance of the SiO2-masked back side of the doped wafer matches the value measured for the undoped wafer within the experimental error, confirming that the SiO2 mask is effective at preventing the back surface from becoming doped.

6.3.2 Physical and Chemical Characterization

Scanning electron microscopy (SEM) was performed with a FEI Magellan XHR microscope. Images were collected using the secondary electron detector or concentric backscatter detector with a beam voltage of 5.0 kV and current of 50 pA.

Auger electron spectroscopy was performed using a Phi 700 Scanning Auger Nanoprobe. The Auger elemental maps have a resolution of 64 by 64 pixels.

6.3.3 Photoelectrochemical Characterization

Photoelectrochemical measurements were performed in a three-electrode electrochemical cell with a Bio-Logic VSP potentiostat. This setup is shown in Figure 6.5a. The silicon sample was used as the working electrode. The counter electrode was an iridium wire, the reference electrode was Hg/Hg2SO4 in saturated K2SO4, and the electrolyte was 0.5 M H2SO4 prepared in 18.2 MΩ cm Millipore water. The glass cell used for these measurements was divided into two compartments with a proton-conducting Nafion membrane separator. The working and reference electrodes were in one compartment, while the counter electrode was in the other

187 compartment to minimize cross-contamination. The working electrode compartment was continuously purged with H2 gas and stirred to remove bubbles from the silicon electrode surface.

After each measurement, the potential of the reference electrode was calibrated to the reversible hydrogen electrode (RHE) using platinum working and counter electrodes in H2-saturated electrolyte. All potentials are plotted with respect to the RHE. The currents are normalized to the active area of each sample, which was measured using a digital photograph. Active areas ranged from 0.25 – 0.75 cm2.

Figure 6.5. a. Photograph of the electrochemical cell used to test the silicon electrodes. b. Teflon and glass electrode positioning apparatus.

188

The photoactivity of the silicon electrodes was measured using a 150 W Xe arc lamp purchased from ABET Technologies (Milfort, CT). The white light from this lamp was passed through an assembly of lenses designed to increase the spatial uniformity of the light and an iris diaphragm to restrict the illuminated spot size to just larger than the active area of the sample. This light was directed onto the photocathode samples through a fused silica window in the electrochemical cell. The spectral irradiance of the light was measured using a 3.9 mm diameter cosine corrector connected via an optical fiber to an Ocean Optics Jaz EL 200-XR1 spectrometer. The irradiance at the sample was adjusted to match “one sun” based on the AM1.5G standard solar spectrum.58 With the cosine corrector positioned inside the electrochemical cell, the entire cell was moved closer to or farther from the light source using a screw-driven linear translating stage until the measured intensity was 760 ± 5 W/m2, which is the integrated power in the AM1.5G standard at wavelengths shorter than 1030 nm, the long wavelength detection limit of the Jaz spectrometer. The photon count in the measured wavelength range of 200 – 1030 nm was higher than the AM1.5G spectrum by about 6%. The spectral irradiance of the 150 W Xe arc lamp is compared to the AM1.5G standard solar spectrum in Figure 6.6. After the irradiance calibration, the cosine corrector was removed from the electrochemical cell, the electrolyte was added, and the silicon photoelectrode was positioned at the same location previously occupied by the cosine corrector using the custom- made Teflon and glass mounting apparatus shown in Figure 6.5b. As a result of this procedure, the irradiance at the silicon sample surface was likely slightly lower than the measured value due to light absorption in the electrolyte (path length ~4 cm).73 Some additional error may have been introduced into the calibration by spatial variations in the light intensity or the mismatch in size between the cosine corrector and silicon electrodes.

189

Figure 6.6. Spectral irradiance of the 150 W xenon lamp compared to the AM1.5G standard solar spectrum.58

The performance the electrodes was measured using linear sweep voltammograms (LSVs.). One or two LSVs were used to activate the electrodes and test for proper electrical connection. Then the performance of the photoelectrodes was measured using a LSV from approximately +0.6 to -0.3 V vs. RHE at 10 mV/s. To measure the stability of the photoelectrodes, after the irradiance was calibrated to “one sun” as described above, the activity of the electrodes was measured every 10 min using LSVs from +0.6 to -0.3 V vs. RHE at 10 mV/s. Between these measurements, the potential was held at 0 V vs. RHE and the current was measured.

The photoelectrodes were compared using several performance metrics illustrated graphically in Figure 6.7. For the purposes of this chapter, these metrics are defined as follows:

2 Onset potential (Eonset): The potential at which a current density of -0.5 mA/cm is reached for a photocathode or dark catalyst.

Open circuit potential (Eoc): The potential at which the net current is zero for a photocathode.

Photovoltage (Vph): The voltage supplied by the illuminated semiconductor in a photocathode. Measured as the difference between the photocurrent onset potentials for a photocathode and an appropriate dark catalysis control. The open circuit potential can be taken to as an estimate of the photovoltage for photocathodes that incorporate platinum catalysts because of the very small overpotential losses for the HER on Pt surfaces.

190

Figure 6.7. Graphical depiction of photocathode performance metrics used in this work.

6.4 Results and Discussion

+ 6.4.1 “Generation 1” Pt/SiO2-n p Si photocathodes

+ SEM images of the “Generation 1” Pt/SiO2-n p Si photocathode are shown in Figure 6.7. These images confirm that the photolithography process used to fabricate these structures was successful. A pattern of metal catalyst disks in a square array makes contact to the n+ Si emitter. This pattern extends across the entire surface of the photocathode sample. Between the catalyst disks, the Si surface is covered with a layer of thermal SiO2. The SEM images in Figure 6.8 show that the diameter of the catalyst disks is 4.8 – 4.9 μm, in good agreement with the dimensions of the photomask used to create the pattern.

The purpose of the catalyst disk pattern is to reduce the metal/Si contact area. Metal/silicon contacts typically possess very high surface recombination velocities, which can drastically

59 reduce the photovoltage produced by illuminated Si photodiodes. In contrast, thermal SiO2 can passivate the surface of the Si, reducing the surface recombination velocity by a factor of 100 or more.59 The disks in this pattern contact approximately 6% of the Si surface, while the

191

SiO2 covers the remaining 94%. This should provide this structure with enhanced photovoltage compared to Si photocathodes that have no surface passivation. However, this pattern also increases the local current density at the Pt HER catalyst by more than 16 times, which may reduce the photoelectrode’s fill factor.74 Future studies to optimize the catalyst surface coverage to balance the effects of light blocking, surface recombination, and catalytic activity could be useful, though prior studies suggest that the current design is reasonable.74

+ Figure 6.8. SEM images of the “Generation 1” Pt disk/SiO2-n p Si photocathode front surface collected using a secondary electron detector. a, b. Low magnification images showing that the Pt disk pattern is successfully formed across a large area. c. Single Pt disk surrounded by SiO2. d. Image taken at 45° sample tilt showing the edge of the Pt disk in contact with the SiO2.

For further confirmation that the desired structure was created successfully, Auger electron spectroscopy elemental maps of the “Generation 1” photocathode structure were collected, as shown in Figure 6.9. These maps show that the disks are primarily composed of platinum while

192 the area in between the disks is composed of silicon and oxygen, as expected. There may also be some carbon at the edges of the disks, possibly from remaining photoresist or another source of carbonaceous material.

+ Figure 6.9. Auger electron spectroscopy elemental maps of the “Generation 1” Pt disk/SiO2-n p Si photocathode front surface. a. Silicon. b. Oxygen. c. Carbon. d. Platinum.

+ The electrochemical activity of the “Generation 1” Pt/SiO2-n p Si photocathode is shown in Figure 6.10. This photocathode has an open circuit potential of 0.60 V vs. RHE, corresponding to a photovoltage of 600 mV. This is 30 mV larger than the photovoltage produced by a conventional Pt-n+p Si photocathode with no surface passivation.3 The photocurrent onset is also substantially more positive than the molybdenum sulfide/silicon photocathodes discussed in Chapter 5 due to both the superior activity of the Pt catalyst and the improved photovoltage

+ 22 of the Pt/SiO2-n p Si device. The saturation photocurrent density of the “Generation 1” photocathode is approximately 15 mA/cm2. This value is lower than the saturation photocurrent density measured on all the other structures, potentially due to thin film interference effects in the SiO2, which has a thickness that is not perfectly optimized as an antireflective layer.

193

+ Figure 6.10. Activity of the “Generation 1” Pt disk/SiO2-n p Si photocathode measured under “one sun” simulated solar illumination. a. Full scan range. b. Photocurrent onset region. The activity of the MoS2- + + + 3, 22 n p Si, Mo3S13-MoS2-n p Si, and Pt-n p Si devices are reproduced for comparison.

+ 6.4.2 “Generation 2” Pt/Si3N4-n p Si photocathodes

While the “Generation 1” photocathode demonstrates good performance, further gains are

+ possible. The “Generation 2” Pt/Si3N4-n p Si photocathode uses a stoichiometric silicon nitride,

Si3N4, in place of the thermal oxide passivation layer. Silicon nitride, which is commonly used as a surface coating in commercial Si PVs, could provide excellent passivation for the n+ Si

194 emitter.59, 66 The SEM images of the “Generation 2” device shown in Figure 6.11 indicate that the pattern of catalysts disks was created successfully. The diameter of the catalyst disks on these devices is 3.3 – 3.4 μm, slightly smaller than in the “Generation 1” device. This discrepancy may result from differences between the wet and dry etching procedures used in the “Generation 1” and “Generation 2” devices, respectively. Based on the observed dimensions, the Pt disks cover approximately 3% of the Si surface, while the Si3N4 covers the remaining 97%. The SEM images in Figure 6.11 also show rings of a dark material surrounding the catalyst disks. This material could be metal sputtered onto the Si3N4 or a carbonaceous polymer material.

+ Figure 6.11. SEM images of the “Generation 2” Pt disk/Si3N4-n p Si photocathode front surface collected using a secondary electron detector. a. Low magnification image showing that the Pt disk pattern is successfully formed across a large area. b. Single Pt disk surrounded by Si3N4. The dark ring around the catalyst disk may be polymer remaining from the lithography process. c. Image taken at 45° sample tilt of a single Pt disk surrounded by Si3N4. d. Image taken at 45° sample tilt showing the edge of the dark ring surrounding the Pt disk.

195

+ The activity of the “Generation 2” Pt/Si3N4-n p Si photocathode is shown in Figure 6.12. This photocathode possesses a photocurrent onset of 0.60 V vs RHE, the same as the “Generation 1” device, indicating that the Si3N4 layer did not result in a change in the photovoltage. However, the saturation photocurrent density is increased by 60% – 70% compared to the “Generation 1” device. This difference probably results from the improved antireflection properties of the Si3N4 layer, though this layer is also not optimized for this purpose.

+ Figure 6.12. Activity of the “Generation 2” Pt disk/Si3N4-n p Si photocathode measured under “one sun” simulated solar illumination. a. Full scan range. b. Photocurrent onset region. The “Generation 1” photocathode’s activity is also plotted for comparison.

196

+ + 6.4.3 “Generation 3” Pt/SiO2-n pp Si photocathodes

+ + In an attempt to improve the photovoltage even further, the “Generation 3” Pt/SiO2-n pp Si devices incorporate a back surface field, a p+ region intended to reduce surface-mediated recombination at the rear surface of the device. The overall structure of this photocathode is very similar to the Passivated Emitter Solar Cell (PESC) structure described by Blakers and Green, a Si PV which produced a photovoltage of 661 mV.79 Although the exact device structure and processing procedures used in the PESC solar cell are different than those used here, this comparison suggests that the “Generation 3” Si photocathode has the potential to produce a substantially improved photovoltage.

The back surface field in the “Generation 3” devices was created following previous reports.67, 80 A layer of Al was evaporated on the back side of the DSP Si wafer. Then this structure was rapidly annealed under a forming gas atmosphere to diffuse the Al into the Si, creating a highly- doped p-type region at the rear contact. To determine the effects of the back surface field formation parameters, samples were annealed at 800 °C and 900 °C. In addition, one control included an Al back contact but no anneal, and a final control sample used a Ga-In eutectic back

+ + contact. SEM images of the “Generation 3” Pt/SiO2-n pp Si device are shown in Figure 6.13. As before, the lithography process was successful. The catalyst disks possess a diameter of approximately 3.6 – 3.7 μm and cover slightly more than 3% of the device surface.

197

+ + Figure 6.13. SEM images of the “Generation 3” Pt disk/SiO2-n pp Si photocathode front surface collected using a secondary electron detector. a. Low magnification image showing that the Pt disk pattern is successfully formed across a large area. b. Single Pt disk. c. Image taken at 45° sample tilt of a single Pt disk. d. Image taken at 45° sample tilt showing the edge of the Pt disk and the dark ring surrounding the Pt disk.

Like the “Generation 2” devices, the SEM images of these “Generation 3” devices in Figure 6.13 also show dark rings surrounding the catalyst disks. To determine whether this material is metal sputtered onto the Si3N4 or a carbon-based polymer, we collected SEM images of the same Pt disk using both a secondary electron detector (SED) and concentric backscatter detector (CBD) as shown in Figure 6.14. While the SED produces contrast based on multiple factors including the topology, electrical conductivity, and elemental composition, the CBD produces contrast primarily due to the atomic mass of the elements in the sample.75-78 The dark ring is clearly visible in SED image, Figure 6.14a, but is not apparent in the CBD image, Figure 6.14b. This suggests that the dark ring is not composed of Pt. It is most likely that his material is fluoropolymer deposited during the Si3N4 dry etch step, which utilizes fluorine chemistry. Further studies are necessary to understand the origin and impacts of these features.

198

Figure 6.14. SEM images of a single Pt disk on the front surface of the “Generation 3” + + Pt disk/SiO2-n pp Si photocathode. a. Image collected using a secondary electron detector. b. Image of the same region collected using a concentric backscatter detector, which produces contrast primarily based on atomic mass.

The activity of the “Generation 3” photocathodes is shown in Figure 6.15. All the devices possess much more negative photocurrent onset potentials than the “Generation 1” and “Generation 2” devices. This discrepancy may results from the different Si wafers used for the “Generation 3” devices. These double side polished wafers were purchased from a different supplier and possessed nearly two orders of magnitude lower doping density than the single side polished wafers used for the “Generation 1” and “Generation 2” devices. In spite of the overall lower activity, it is clear that the Al back surface field improves the photocurrent onset potential of the “Generation 3” Si photocathodes relative to the control structures with no back surface field.

199

+ + Figure 6.15. Activity of the “Generation 3” Pt disk/SiO2-n pp Si photocathode measured under “one sun” simulated solar illumination. a. Full scan range. b. Photocurrent onset region. The activity of the “Generation 1” and “Generation 2” photocathodes is also plotted for comparison.

200

Figure 6.16 shows an electrochemical stability test of the best “Generation 3” device, the

+ + Pt/SiO2-n pp Si with aluminum back surface field annealed at 900 °C. While the photocurrent onset potential increased slightly over the course of this three hour test, the fill factor degraded, possibly due to catalyst delamination or poisoning, the formation of an insulating SiO2 at the catalyst/silicon interface, or other phenomena.

+ + Figure 6.16. Stability of the “Generation 3” Pt disk/SiO2-n pp Si photocathode measured under “one sun” simulated solar illumination. a. LSVs collected every 10 min. b. Chronoamperometry measurement at E = 0 V vs RHE between the LSVs.

201

Figure 6.17 shows SEM images of the same “Generation 3” device after the stability test. The images show that the majority of the catalyst disks remain intact, but are partially coated with a dark material that may poison the catalyst surface. This may be the same material that originally formed a dark ring around the catalyst disks. These results suggest that removing or preventing the formation of this dark ring could potentially improve the catalyst stability. Additionally, preliminary tests indicate that the stability of these structures are superior in perchloric acid electrolyte, which suggests that sulfate ion poisoning may contribute to the observed activity loss. The formation of an insulating silicon oxide layer at the catalyst/silicon interface is also a likely cause of activity loss. It may be possible to prevent these degradation mechanisms by using established strategies such as protecting layers of MoS2 or TiO2 in this device architecture.20, 22

+ + Figure 6.17. SEM images of the “Generation 3” Pt disk/SiO2-n pp Si photocathode front surface collected using a secondary electron detector after electrochemical stability testing. a. Low magnification image showing that most of the Pt disks remain intact. b. Single hole etched in the SiO2 where the Pt disk delaminated. c. Single Pt disk covered with dark spots likely indicating surface contamination of the catalyst. d. Edge of a single Pt disk.

202

6.4.4 Future Si photocathodes

The results presented so far suggest that if the advantages of the “Generation 1,” “Generation 2,” and “Generation 3” devices could be combined, a device with very high performance could eventually be created. Further work will focus on creating the “Generation 4” Pt disk/Si3N4- n+pp+ Si photocathode structure shown in Figure 6.18. This device will be fabricated using double side polished wafers that are otherwise identical to those used for the “Generation 1” and “Generation 2” devices. In addition, this device will utilize a passivated rear surface with a locally diffused emitter. This structure is very similar to the Passivated Emitter and Rear Contact (PERC) and Passivated Emitter Rear Locally Diffused (PERL) Si PV devices demonstrated by Green, et al., both of which produced photovoltages of 0.69 – 0.70 V.81-83 The “Generation 4” devices will be fabricated using the scheme shown in Figure 6.19. This processing procedure and fabrication scheme provide a platform for the development of high efficiency tandem solar

H2 production devices.

+ + Figure 6.18. Proposed “Generation 4” Pt disk/Si3N4-n pp Si photocathode structure. a. Perspective view. b. Cross-section view. The diagrams are not to scale.

203

Figure 6.19. Schematic diagram of the proposed fabrication procedure for the “Generation 4” Pt + + disk/Si3N4-n pp Si photocathodes (not to scale).

204

6.5 Conclusions

In this chapter, we explored several new silicon photocathodes with designs inspired by high performance PV devices. By incorporating features intended to reduce recombination at silicon interfaces, the best devices created so far have provided photovoltages of 600 mV, approximately equal to the most successful published crystalline Si water splitting photoelectrodes. With further efforts, the processing procedures developed here have the potential to create Si photocathodes with even better performance, a critical requirement for the development of high efficiency tandem water splitting devices.

6.6 Author Contributions

Thomas Hellstern, Pongkarn Chakthranont, Ieva Narkeviciute, and Rueben Britto contributed silicon photocathode fabrication procedure development and data interpretation. Arnold Forman provided helpful insights about silicon photocathode device concepts.

6.7 Acknowledgments

This work was funded by the National Science Foundation Directorate for Engineering and Department of Energy Office of Energy Efficiency and Renewable Energy through Award No. 1433442.

6.8 References

1. M.G. Walter, E.L. Warren, J.R. McKone, S.W. Boettcher, Q. Mi, E.A. Santori, and N.S. Lewis. "Solar Water Splitting Cells." Chemical Reviews, 2010. 110 (11): 6446- 6473. http://dx.doi.org/10.1021/cr1002326 2. C. Jacoboni, C. Canali, G. Ottaviani, and A. Alberigi Quaranta. "A review of some charge transport properties of silicon." Solid-State Electronics, 1977. 20 (2): 77-89. http://dx.doi.org/10.1016/0038-1101(77)90054-5 3. S.W. Boettcher, E.L. Warren, M.C. Putnam, E.A. Santori, D. Turner-Evans, M.D. Kelzenberg, M.G. Walter, J.R. McKone, B.S. Brunschwig, H.A. Atwater, and N.S. Lewis. "Photoelectrochemical hydrogen evolution using Si microwire arrays." Journal of the American Chemical Society, 2011. 133 (5): 1216-1219. http://dx.doi.org/10.1021/ja108801m

205

4. R.M. Candea, M. Kastner, R. Goodman, and N. Hickok. "Photoelectrolysis of Water - Si in Salt-Water." Journal of Applied Physics, 1976. 47 (6): 2724-2726. http://dx.doi.org/10.1063/1.322995 5. R.N. Dominey, N.S. Lewis, J.A. Bruce, D.C. Bookbinder, and M.S. Wrighton. "Improvement of photoelectrochemical hydrogen generation by surface modification of p-type silicon semiconductor photocathodes." Journal of the American Chemical Society, 1982. 104 (2): 467-482. http://dx.doi.org/10.1021/ja00366a016 6. A.Q. Contractor and J.O.M. Bockris. "Investigation of a protective conducting silica film on n-silicon." Electrochimica Acta, 1984. 29 (10): 1427-1434. http://dx.doi.org/10.1016/0013-4686(84)87022-X 7. Y. Nakato, H. Yano, S. Nishiura, T. Ueda, and H. Tsubomura. "Hydrogen photoevolution at p-type silicon electrodes coated with discontinuous metal layers." Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, 1987. 228 (1–2): 97-108. http://dx.doi.org/10.1016/0022-0728(87)80100-6 8. Y. Nakato and H. Tsubomura. "Silicon photoelectrodes modified with ultrafine metal islands." Electrochimica Acta, 1992. 37 (5): 897-907. http://dx.doi.org/10.1016/0013- 4686(92)85041-I 9. A.G. Muñoz and H.J. Lewerenz. "Advances in Photoelectrocatalysis with Nanotopographical Photoelectrodes." ChemPhysChem, 2010. 11 (8): 1603-1615. http://dx.doi.org/10.1002/cphc.200900856 10. Y.W. Chen, J.D. Prange, S. Dühnen, Y. Park, M. Gunji, C.E.D. Chidsey, and P.C. McIntyre. "Atomic layer-deposited tunnel oxide stabilizes silicon photoanodes for water oxidation." Nature Materials, 2011. 10 (7): 539-544. http://dx.doi.org/10.1038/nmat3047 11. Y.D. Hou, B.L. Abrams, P.C.K. Vesborg, M.E. Bjorketun, K. Herbst, L. Bech, A.M. Setti, C.D. Damsgaard, T. Pedersen, O. Hansen, J. Rossmeisl, S. Dahl, J.K. Norskov, and I. Chorkendorff. "Bioinspired molecular co-catalysts bonded to a silicon photocathode for solar hydrogen evolution." Nature Materials, 2011. 10 (6): 434-438. http://dx.doi.org/10.1038/nmat3008 12. J.R. McKone, E.L. Warren, M.J. Bierman, S.W. Boettcher, B.S. Brunschwig, N.S. Lewis, and H.B. Gray. "Evaluation of Pt, Ni, and Ni–Mo electrocatalysts for hydrogen evolution on crystalline Si electrodes." Energy & Environmental Science, 2011. 4 (9): 3573. http://dx.doi.org/10.1039/c1ee01488a 13. B. Seger, A.B. Laursen, P.C.K. Vesborg, T. Pedersen, O. Hansen, S. Dahl, and I. Chorkendorff. "Hydrogen Production Using a Molybdenum Sulfide Catalyst on a Titanium-Protected n+p-Silicon Photocathode." Angewandte Chemie International Edition, 2012. 51 (36): 9128-9131. http://dx.doi.org/10.1002/anie.201203585 14. E.L. Warren, J.R. McKone, H.A. Atwater, H.B. Gray, and N.S. Lewis. "Hydrogen- evolution characteristics of Ni–Mo-coated, radial junction, n+p-silicon microwire array photocathodes." Energy & Environmental Science, 2012. 5 (11): 9653. http://dx.doi.org/10.1039/c2ee23192a 15. C.R. Cox, M.T. Winkler, J.J.H. Pijpers, T. Buonassisi, and D.G. Nocera. "Interfaces between water splitting catalysts and buried silicon junctions." Energy & Environmental Science, 2013. 6 (2): 532-538. http://dx.doi.org/10.1039/C2EE23932A 16. D.V. Esposito, I. Levin, T.P. Moffat, and A.A. Talin. "H2 evolution at Si-based metal- insulator-semiconductor photoelectrodes enhanced by inversion channel charge collection and H spillover." Nature Materials, 2013. 12 (6): 562-568. http://dx.doi.org/10.1038/nmat3626

206

17. Z.P. Huang, C.F. Wang, L. Pan, F. Tian, X.X. Zhang, and C. Zhang. "Enhanced photoelectrochemical hydrogen production using silicon nanowires@MoS3." Nano Energy, 2013. 2 (6): 1337-1346. http://dx.doi.org/10.1016/j.nanoen.2013.06.016 18. M.J. Kenney, M. Gong, Y. Li, J.Z. Wu, J. Feng, M. Lanza, and H. Dai. "High- performance silicon photoanodes passivated with ultrathin nickel films for water oxidation." Science, 2013. 342 (6160): 836-840. http://dx.doi.org/10.1126/science.1241327 19. A.B. Laursen, T. Pedersen, P. Malacrida, B. Seger, O. Hansen, P.C. Vesborg, and I. Chorkendorff. "MoS2: an integrated protective and active layer on n+p-Si for solar H2 evolution." Physical Chemistry Chemical Physics, 2013. 15 (46): 20000-20004. http://dx.doi.org/10.1039/c3cp52890a 20. B. Seger, T. Pedersen, A.B. Laursen, P.C. Vesborg, O. Hansen, and I. Chorkendorff. "Using TiO2 as a conductive protective layer for photocathodic H2 evolution." Journal of the American Chemical Society, 2013. 135 (3): 1057-1064. http://dx.doi.org/10.1021/ja309523t 21. N. Strandwitz, D. Comstock, R.L. Grimm, A.C. Nichols-Nielander, J.W. Elam, and N.S. Lewis. "Photoelectrochemical Behavior of n-type Si(100) Electrodes Coated with Thin Films of Manganese Oxide Grown by Atomic Layer Deposition." Journal of Physical Chemistry C, 2013. http://dx.doi.org/10.1021/jp311207x 22. J.D. Benck, S.C. Lee, K.D. Fong, J. Kibsgaard, R. Sinclair, and T.F. Jaramillo. "Designing Active and Stable Silicon Photocathodes for Solar Hydrogen Production Using Molybdenum Sulfide Nanomaterials." Advanced Energy Materials, 2014. 4 (18). http://dx.doi.org/10.1002/aenm.201400739 23. M.J. Choi, J.-Y. Jung, M.-J. Park, J.-W. Song, J.-H. Lee, and J.H. Bang. "Long-term durable silicon photocathode protected by a thin Al2O3/SiOx layer for photoelectrochemical hydrogen evolution." Journal of Materials Chemistry A, 2014. 2 (9): 2928-2933. http://dx.doi.org/10.1039/C3TA14443G 24. Q. Ding, F. Meng, C.R. English, M. Cabán-Acevedo, M.J. Shearer, D. Liang, A.S. Daniel, R.J. Hamers, and S. Jin. "Efficient Photoelectrochemical Hydrogen Generation Using Heterostructures of Si and Chemically Exfoliated Metallic MoS2." Journal of the American Chemical Society, 2014. 136 (24): 8504-8507. http://dx.doi.org/10.1021/ja5025673 25. S. Hu, M.R. Shaner, J.A. Beardslee, M. Lichterman, B.S. Brunschwig, and N.S. Lewis. "Amorphous TiO2 coatings stabilize Si, GaAs, and GaP photoanodes for efficient water oxidation." Science, 2014. 344 (6187): 1005-1009. http://dx.doi.org/10.1126/science.1251428 26. S. Huang, H. Zhang, Z. Wu, D. Kong, D. Lin, Y. Fan, X. Yang, Z. Zhong, S. Huang, Z. Jiang, and C. Cheng. "Large-Area Ordered P-type Si Nanowire Arrays as Photocathode for Highly Efficient Photoelectrochemical Hydrogen Generation." ACS Applied Materials & Interfaces, 2014. 6 (15): 12111-12118. http://dx.doi.org/10.1021/am501168c 27. Z. Huang, J.R. McKone, C. Xiang, R.L. Grimm, E.L. Warren, J.M. Spurgeon, H.-J. Lewerenz, B.S. Brunschwig, and N.S. Lewis. "Comparison between the measured and modeled hydrogen-evolution activity of Ni- or Pt-coated silicon photocathodes." International Journal of Hydrogen Energy, 2014. 39 (28): 16220-16226. http://dx.doi.org/10.1016/j.ijhydene.2013.12.162 28. M.G. Kast, L.J. Enman, N.J. Gurnon, A. Nadarajah, and S.W. Boettcher. "Solution- deposited F:SnO2/TiO2 as a base-stable protective layer and anti-reflective coating for micro-textured buried-junction H2-evolving Si photocathodes." ACS Applied

207

Materials & Interfaces, 2014. 6 (24): 22830-22837. http://dx.doi.org/10.1021/am506999p 29. R. Liu, Z. Zheng, J. Spurgeon, and X. Yang. "Enhanced photoelectrochemical water- splitting performance of semiconductors by surface passivation layers." Energy & Environmental Science, 2014. 7 (8): 2504-2517. http://dx.doi.org/10.1039/C4EE00450G 30. B. Mei, A.A. Permyakova, R. Frydendal, D. Bae, T. Pedersen, P. Malacrida, O. Hansen, I.E.L. Stephens, P.C.K. Vesborg, B. Seger, and I. Chorkendorff. "Iron- Treated NiO as a Highly Transparent p-Type Protection Layer for Efficient Si-Based Photoanodes." Journal of Physical Chemistry Letters, 2014. 5 (20): 3456-3461. http://dx.doi.org/10.1021/jz501872k 31. B. Mei, B. Seger, T. Pedersen, M. Malizia, O. Hansen, I. Chorkendorff, and P.C.K. Vesborg. "Protection of p+-n-Si Photoanodes by Sputter-Deposited Ir/IrOx Thin Films." Journal of Physical Chemistry Letters, 2014. 5 (11): 1948-1952. http://dx.doi.org/10.1021/jz500865g 32. B. Seger, K. Herbst, T. Pedersen, B. Abrams, P.C.K. Vesborg, O. Hansen, and I. Chorkendorff. "Mo3S4 Clusters as an Effective H2 Evolution Catalyst on Protected Si Photocathodes." Journal of The Electrochemical Society, 2014. 161 (12): H722-H724. http://dx.doi.org/10.1149/2.0161412jes 33. K. Sun, S. Shen, Y. Liang, P.E. Burrows, S.S. Mao, and D. Wang. "Enabling Silicon for Solar-Fuel Production." Chemical Reviews, 2014. 114 (17): 8662-8719. http://dx.doi.org/10.1021/cr300459q 34. E.L. Warren, H.A. Atwater, and N.S. Lewis. "Silicon Microwire Arrays for Solar Energy-Conversion Applications." Journal of Physical Chemistry C, 2014. 118 (2): 747-759. http://dx.doi.org/10.1021/jp406280x 35. J. Yang, K. Walczak, E. Anzenberg, F.M. Toma, G. Yuan, J. Beeman, A. Schwartzberg, Y. Lin, M. Hettick, A. Javey, J.W. Ager, J. Yano, H. Frei, and I.D. Sharp. "Efficient and Sustained Photoelectrochemical Water Oxidation by Cobalt Oxide/Silicon Photoanodes with Nanotextured Interfaces." Journal of the American Chemical Society, 2014. 136 (17): 6191-6194. http://dx.doi.org/10.1021/ja501513t 36. X.-Q. Bao, M. Fatima Cerqueira, P. Alpuim, and L. Liu. "Silicon nanowire arrays coupled with cobalt phosphide spheres as low-cost photocathodes for efficient solar hydrogen evolution." Chemical Communications, 2015. 51 (53): 10742-10745. http://dx.doi.org/10.1039/C5CC02331A 37. X.-Q. Bao, D.Y. Petrovykh, P. Alpuim, D.G. Stroppa, N. Guldris, H. Fonseca, M. Costa, J. Gaspar, C. Jin, and L. Liu. "Amorphous oxygen-rich molybdenum oxysulfide Decorated p-type silicon microwire Arrays for efficient photoelectrochemical water reduction." Nano Energy, 2015. 16: 130-142. http://dx.doi.org/10.1016/j.nanoen.2015.06.014 38. Y. Chen, P.D. Tran, P. Boix, Y. Ren, S.Y. Chiam, Z. Li, K. Fu, L.H. Wong, and J. Barber. "Silicon Decorated with Amorphous Cobalt Molybdenum Sulfide Catalyst as an Efficient Photocathode for Solar Hydrogen Generation." ACS Nano, 2015. 9 (4): 3829-3836. http://dx.doi.org/10.1021/nn506819m 39. R. Fan, W. Dong, L. Fang, F. Zheng, X. Su, S. Zou, J. Huang, X. Wang, and M. Shen. "Stable and efficient multi-crystalline n+p silicon photocathode for H2 production with pyramid-like surface nanostructure and thin Al2O3 protective layer." Applied Physics Letters, 2015. 106 (1): 013902. http://dx.doi.org/10.1063/1.4905511 40. R. Fan, J. Min, Y. Li, X. Su, S. Zou, X. Wang, and M. Shen. "n-type silicon photocathodes with Al-doped rear p+ emitter and Al2O3-coated front surface for

208

efficient and stable H2 production." Applied Physics Letters, 2015. 106 (21): 213901. http://dx.doi.org/10.1063/1.4921845 41. J. Feng, M. Gong, M. Kenney, J. Wu, B. Zhang, Y. Li, and H. Dai. "Nickel-coated silicon photocathode for water splitting in alkaline electrolytes." Nano Research, 2015. 8 (5): 1577-1583. http://dx.doi.org/10.1007/s12274-014-0643-4 42. Q. Huang, Z. Ye, and X. Xiao. "Recent Progress in Photocathodes for Hydrogen Evolution." Journal of Materials Chemistry A, 2015. http://dx.doi.org/10.1039/C5TA03594E 43. L. Ji, M.D. McDaniel, S. Wang, A.B. Posadas, X. Li, H. Huang, J.C. Lee, A.A. Demkov, A.J. Bard, J.G. Ekerdt, and E.T. Yu. "A silicon-based photocathode for water reduction with an epitaxial SrTiO3 protection layer and a nanostructured catalyst." Nature Nanotechnology, 2015. 10 (1): 84-90. http://dx.doi.org/10.1038/nnano.2014.277 44. Y.-H. Lai, H.S. Park, J.Z. Zhang, P.D. Matthews, D.S. Wright, and E. Reisner. "A Si Photocathode Protected and Activated with a Ti and Ni Composite Film for Solar Hydrogen Production." Chemistry – A European Journal, 2015. 21 (10): 3919-3923. http://dx.doi.org/10.1002/chem.201406566 45. R. O’Connor, J. Bogan, N. Fleck, A. McCoy, L.A. Walsh, C. Byrne, P. Casey, and G. Hughes. "Growth and characterization of thin manganese oxide corrosion barrier layers for silicon photoanode protection during water oxidation." Solar Energy Materials and Solar Cells, 2015. 136 (0): 64-69. http://dx.doi.org/10.1016/j.solmat.2014.12.028 46. C.W. Roske, E.J. Popczun, B. Seger, C.G. Read, T. Pedersen, O. Hansen, P.C.K. Vesborg, B.S. Brunschwig, R.E. Schaak, I. Chorkendorff, H.B. Gray, and N.S. Lewis. "Comparison of the Performance of CoP-Coated and Pt-Coated Radial Junction n+p- Silicon Microwire-Array Photocathodes for the Sunlight-Driven Reduction of Water to H2(g)." Journal of Physical Chemistry Letters, 2015. 6 (9): 1679-1683. http://dx.doi.org/10.1021/acs.jpclett.5b00495 47. M.R. Shaner, S. Hu, K. Sun, and N.S. Lewis. "Stabilization of Si microwire arrays for solar-driven H2O oxidation to O2(g) in 1.0 M KOH(aq) using conformal coatings of amorphous TiO2." Energy & Environmental Science, 2015. 8 (1): 203-207. http://dx.doi.org/10.1039/C4EE03012E 48. K. Sun, M.T. McDowell, A.C. Nielander, S. Hu, M.R. Shaner, F. Yang, B.S. Brunschwig, and N.S. Lewis. "Stable Solar-Driven Water Oxidation to O2(g) by Ni- Oxide-Coated Silicon Photoanodes." Journal of Physical Chemistry Letters, 2015. 6 (4): 592-598. http://dx.doi.org/10.1021/jz5026195 49. K. Sun, F.H. Saadi, M.F. Lichterman, W.G. Hale, H.-P. Wang, X. Zhou, N.T. Plymale, S.T. Omelchenko, J.-H. He, K.M. Papadantonakis, B.S. Brunschwig, and N.S. Lewis. "Stable solar-driven oxidation of water by semiconducting photoanodes protected by transparent catalytic nickel oxide films." Proceedings of the National Academy of Sciences, 2015. 112 (12): 3612-3617. http://dx.doi.org/10.1073/pnas.1423034112 50. L. Zhang, C. Liu, A. Wong, J. Resasco, and P. Yang. "MoS2-wrapped silicon nanowires for photoelectrochemical water reduction." Nano Research, 2015. 8 (1): 281-287. http://dx.doi.org/10.1007/s12274-014-0673-y 51. Y. Zhao, N.C. Anderson, K. Zhu, J.A. Aguiar, J.A. Seabold, J.v.d. Lagemaat, H.M. Branz, N.R. Neale, and J. Oh. "Oxidatively Stable Nanoporous Silicon Photocathodes with Enhanced Onset Voltage for Photoelectrochemical Proton Reduction." Nano Letters, 2015. 15 (4): 2517-2525. http://dx.doi.org/10.1021/acs.nanolett.5b00086

209

52. M.R. Shaner, K.T. Fountaine, and H.-J. Lewerenz. "Current-voltage characteristics of coupled photodiode-electrocatalyst devices." Applied Physics Letters, 2013. 103 (14): 143905. http://dx.doi.org/doi:http://dx.doi.org/10.1063/1.4822179 53. T.J. Mills, F. Lin, and S.W. Boettcher. "Theory and Simulations of Electrocatalyst- Coated Semiconductor Electrodes for Solar Water Splitting." Physical Review Letters, 2014. 112 (14): 148304. http://link.aps.org/doi/10.1103/PhysRevLett.112.148304 54. A.C. Nielander, M.R. Shaner, K.M. Papadantonakis, S.A. Francis, and N.S. Lewis. "A taxonomy for solar fuels generators." Energy & Environmental Science, 2015. 8 (1): 16-25. http://dx.doi.org/10.1039/c4ee02251c 55. K. Walczak, Y. Chen, C. Karp, J.W. Beeman, M. Shaner, J. Spurgeon, I.D. Sharp, X. Amashukeli, W. West, J. Jin, N.S. Lewis, and C. Xiang. "Modeling, Simulation, and Fabrication of a Fully Integrated, Acid-stable, Scalable Solar-Driven Water-Splitting System." ChemSusChem, 2015. 8 (3): 544-551. http://dx.doi.org/10.1002/cssc.201402896 56. B. Seger, D.S. Tilley, T. Pedersen, P.C.K. Vesborg, O. Hansen, M. Grätzel, and I. Chorkendorff. "Silicon protected with atomic layer deposited TiO2: durability studies of photocathodic H2 evolution." RSC Advances, 2013. 3 (48): 25902. http://dx.doi.org/10.1039/c3ra45966g 57. M.A. Green, K. Emery, Y. Hishikawa, W. Warta, and E.D. Dunlop. "Solar cell efficiency tables (version 46)." Progress in Photovoltaics: Research and Applications, 2015. 23 (7): 805-812. http://dx.doi.org/10.1002/pip.2637 58. ASTM International. "ASTM G173 - 03, Standard Tables for Reference Solar Spectral Irradiances: Direct Normal and Hemispherical on 37° Tilted Surface." 2013. West Conshohocken, PA. http://dx.doi.org/10.1520/G0173-03R12 59. A.G. Aberle. "Surface passivation of crystalline silicon solar cells: A review." Progress in Photovoltaics, 2000. 8 (5): 473-487. http://dx.doi.org/10.1002/1099- 159x(200009/10)8:5<473::aid-pip337>3.0.co;2-d 60. L.C. Seitz, Z. Chen, A.J. Forman, B.A. Pinaud, J.D. Benck, and T.F. Jaramillo. "Modeling practical performance limits of photoelectrochemical water splitting based on the current state of materials research." ChemSusChem, 2014. 7 (5): 1372-1385. http://dx.doi.org/10.1002/cssc.201301030 61. J. Oh, H.C. Yuan, and H.M. Branz. "An 18.2%-efficient black-silicon solar cell achieved through control of carrier recombination in nanostructures." Nature Nanotechnology, 2012. 7 (11): 743-748. http://dx.doi.org/10.1038/nnano.2012.166 62. D.H. Lee, S.P.R. Kobaku, Y.R. Hong, and J.Y. Kwon. "An Efficient Photoelectrochemical Hydrogen Evolution System using Silicon Nanomaterials with Ultra-High Aspect Ratios." Energy Technology, 2014. 2 (11): 889-896. http://dx.doi.org/10.1002/ente.201402074 63. D. Bae, T. Pedersen, B.J. Seger, M. Malizia, A. Kuznetsov, O. Hansen, I. Chorkendorff, and P.C.K. Vesborg. "Back-illuminated Si photocathode: a combined experimental and theoretical study for photocatalytic hydrogen evolution." Energy & Environmental Science, 2015. 8 (2): 650-660. http://dx.doi.org/10.1039/C4EE03723E 64. W.D. Eades and R.M. Swanson. "Calculation of surface generation and recombination velocities at the Si‐SiO2 interface." Journal of Applied Physics, 1985. 58 (11): 4267- 4276. http://dx.doi.org/10.1063/1.335562 65. E. Yablonovitch, R.M. Swanson, W.D. Eades, and B.R. Weinberger. "Electron‐hole recombination at the Si‐SiO2 interface." Applied Physics Letters, 1986. 48 (3): 245- 247. http://dx.doi.org/10.1063/1.96570 66. M.J. Kerr, J. Schmidt, A. Cuevas, and J.H. Bultman. "Surface recombination velocity of phosphorus-diffused silicon solar cell emitters passivated with plasma enhanced

210

chemical vapor deposited silicon nitride and thermal silicon oxide." Journal of Applied Physics, 2001. 89 (7): 3821-3826. http://dx.doi.org/10.1063/1.1350633 67. S. Narasinha and A. Rohatgi. "Optimized aluminum back surface field techniques for silicon solar cells." Photovoltaic Specialists Conference 1997, 1997: 63-66. http://dx.doi.org/10.1109/PVSC.1997.653925 68. T. Fellmeth, S. Mack, J. Bartsch, D. Erath, U. Jäger, R. Preu, F. Clement, and D. Biro. "20.1% Efficient Silicon Solar Cell With Aluminum Back Surface Field." IEEE Electron Device Letters, 2011. 32 (8): 1101-1103. http://dx.doi.org/10.1109/LED.2011.2157656 69. W. Kern and D.A. Puotinen. "Cleaning Solutions Based on Hydrogen Peroxide for use in Silicon Semiconductor Technology." RCA Review, 1970. 31 (2): 187-206. 70. C.J. Frosch and L. Derick. "Surface Protection and Selective Masking during Diffusion in Silicon." Journal of The Electrochemical Society, 1957. 104 (9): 547- 552. http://dx.doi.org/10.1149/1.2428650 71. C.T. Sah, H. Sello, and D.A. Tremere. "Diffusion of phosphorus in silicon oxide film." Journal of Physics and Chemistry of Solids, 1959. 11 (3–4): 288-298. http://dx.doi.org/10.1016/0022-3697(59)90229-X 72. R.N. Ghoshtagore. "Silicon dioxide masking of phosphorus diffusion in silicon." Solid-State Electronics, 1975. 18 (5): 399-406. http://dx.doi.org/10.1016/0038- 1101(75)90041-6 73. H. Doscher, J.F. Geisz, T.G. Deutsch, and J.A. Turner. "Sunlight absorption in water - efficiency and design implications for photoelectrochemical devices." Energy & Environmental Science, 2014. 7 (9): 2951-2956. http://dx.doi.org/10.1039/C4EE01753F 74. Y. Chen, K. Sun, H. Audesirk, C. Xiang, and N.S. Lewis. "A quantitative analysis of the efficiency of solar-driven water-splitting device designs based on tandem photoabsorbers patterned with islands of metallic electrocatalysts." Energy & Environmental Science, 2015. 8 (6): 1736-1747. http://dx.doi.org/10.1039/C5EE00311C 75. D.E. Newbury and H. Yakowitz. "Contrast Mechanisms of Special Interest in Materials Science," in "Practical Scanning Electron Microscopy." J. Goldstein and H. Yakowitz, Editors. 1975, Springer US. 149-210. http://dx.doi.org/10.1007/978-1- 4613-4422-3_5 76. D.E. Newbury. "Image Formation in the Scanning Electron Microscope," in "Practical Scanning Electron Microscopy." J. Goldstein and H. Yakowitz, Editors. 1975, Springer US. 95-148. http://dx.doi.org/10.1007/978-1-4613-4422-3_4 77. J. Goldstein, D. Newbury, P. Echlin, D. Joy, C. Lyman, E. Lifshin, L. Sawyer, and J. Michael. "Electron Beam–Specimen Interactions," in "Scanning Electron Microscopy and X-ray Microanalysis." 2003, Springer US. 61-98. http://dx.doi.org/10.1007/978-1- 4615-0215-9_3 78. J. Goldstein, D. Newbury, P. Echlin, D. Joy, C. Lyman, E. Lifshin, L. Sawyer, and J. Michael. "Image Formation and Interpretation," in "Scanning Electron Microscopy and X-ray Microanalysis." 2003, Springer US. 99-193. http://dx.doi.org/10.1007/978- 1-4615-0215-9_4 79. A.W. Blakers and M.A. Green. "20% efficiency silicon solar cells." Applied Physics Letters, 1986. 48 (3): 215-217. http://dx.doi.org/10.1063/1.96799 80. S. Narasimha, A. Rohatgi, and A.W. Weeber. "An optimized rapid aluminum back surface field technique for silicon solar cells." IEEE Transactions on Electron Devices, 1999. 46 (7): 1363-1370. http://dx.doi.org/10.1109/16.772477

211

81. A.W. Blakers, A. Wang, A.M. Milne, J. Zhao, and M.A. Green. "22.8% efficient silicon solar cell." Applied Physics Letters, 1989. 55 (13): 1363-1365. http://dx.doi.org/10.1063/1.101596 82. A. Wang, J. Zhao, and M.A. Green. "24% efficient silicon solar cells." Applied Physics Letters, 1990. 57 (6): 602-604. http://dx.doi.org/10.1063/1.103610 83. J. Zhao, A. Wang, P.P. Altermatt, S.R. Wenham, and M.A. Green. "24% efficient PERL silicon solar cell: Recent improvements in high efficiency silicon cell research." Solar Energy Materials and Solar Cells, 1996. 41–42 (0): 87-99. http://dx.doi.org/10.1016/0927-0248(95)00117-4

212

Chapter 7: Conclusions and Future Directions

7.1 Conclusions

Hydrogen’s importance as a chemical reagent and energy carrier provides great motivation for developing sustainable processes for its production. Photoelectrochemical water splitting offers a promising strategy for renewable H2 generation from water using the energy from sunlight, but this technology is currently not viable for widespread implementation. This dissertation has focused on several key research challenges that must be addressed to make PEC water splitting a practical solution.

Efficient, stable, and scalable electrocatalysts for the HER are necessary to produce high efficiency water splitting devices, and many molybdenum sulfide materials are among the most active earth-abundant HER catalysts. Amorphous MoSx catalysts are especially interesting because they can be fabricated using scalable, room temperature synthesis techniques. We showed that these catalysts possess high HER activity in large part due to their nanoporous film morphology, which results in a high density of exposed catalytically active sites. In addition, the properties of amorphous MoSx surfaces result in high per-site turn over frequencies. The primary catalytic surface is an amorphous MoS2. This insight is derived from ex situ and in situ spectroscopic measurements, which reveal that the surface of the catalyst is reduced from MoS3 to MoS2 under operating conditions, and transmission electron microscopy measurements, which show that the surface remains largely amorphous. However, domains of crystalline MoS2 may form after extended operation, which could contribute to catalyst deactivation. Although the amorphous nature of this material prohibits the identification of a specific active site, these results provide further evidence that a wide range of molybdenum/sulfur species may exhibit excellent activity as HER catalysts.

Developing integrated photoelectrodes that combine semiconductor light absorbers with electrocatalysts in stables devices is another substantial challenge. Silicon was chosen as the light absorber for these studies due to its appropriate band gap for application in a tandem PEC water splitting device and its excellent electronic properties. However, silicon has poor catalytic activity for the HER and is susceptible to corrosion in aqueous electrolyte. We showed that

213 molybdenum sulfide nanomaterials can provide both corrosion protection and catalytic activity when integrated into silicon photocathodes. The most stable molybdenum sulfide/silicon photocathodes studied here showed no loss in performance after 100 hours of durability testing. We also developed silicon photocathodes with passivated interfaces that reduce surface- mediated recombination and improve the photovoltage generated by the illuminated semiconductor. These structures provide a platform for the fabrication of fully integrated, monolithic dual absorber solar water splitting devices.

7.2. Future Directions

Despite the significant progress towards successful PEC water splitting devices reported in this dissertation and in many other recent works, further advancements are necessary. Huge gains in the activity of earth-abundant catalysts have been achieved over the last several years, and molybdenum sulfides remain a promising class of catalyst materials. Many studies of molybdenum sulfide HER catalysts have focused on improving electrode-area normalized current densities by increasing the number of accessible active sites. Improving the turn over frequency of each site through new doping methods, strain effects, or catalyst/support interactions could provide even further advancements. Other classes of nonprecious HER catalysts are also promising. In particular, transition metal phosphides have received great attention because they demonstrate even better total electrode HER activity than most molybdenum sulfide materials. Understanding the physical and chemical properties that give rise to this high activity may enable even further advancements. Such studies would also contribute new fundamental understanding about the material properties that control HER activity. Comparing HER catalysts across material classes suggests that some physical properties aside from aside from the free energy of hydrogen adsorption lead to differences in activity. Understanding these effects may provide further levers for designing materials with improved performance.

As new, highly effective strategies for semiconductor corrosion protection are developed, it is increasingly important to test the limits of these approaches. So far, no corrosion protection schemes have successfully stabilized semiconductor absorbers for 10 years or more. Eventually, devices must be stable for this length of time to make PEC water splitting economically viable.

Applying MoS2 as a protective coating for III-V materials, which corrode even more readily

214 than silicon, could provide useful insights about the strengths and weaknesses of this design. Additionally, developing accelerated durability testing protocols that rely on potential cycling, elevated temperatures, or increased light intensity would be very useful for rapidly predicting the efficacy of protective materials over decades of operation. Finally, performing failure analysis to understand how and why photoelectrodes degrade will be critical for designing even more effective approaches.

The silicon photocathodes presented in Chapter 6 could enable the fabrication of monolithic dual absorber PEC water splitting devices, but developing the processes to integrate large band gap absorbers with silicon and appropriate catalysts without destroying any components will require further work. Constructing and testing such devices will reveal a great deal about the device integration strategies that are necessary for high performance. Additionally, using new approaches to characterize the interfaces between device components to better understand their effects on device performance will be very fruitful.

Alternative PEC water splitting device and system architectures could provide game-changing advancements towards economical devices. Both particulate photocatalyst systems and photovoltaic-electrolyzer configurations have inherent advantages which could make them less expensive than panel-based PEC water splitting systems. Novel device and reactor concepts could provide a means to circumvent the limitations of current materials and achieve the extremely high performance and low cost necessary for PEC water splitting to succeed in the marketplace.

Future generations of passionate and dedicated researchers will surely continue the rapid progress in this field. This is reason to have great hope that solar water splitting will eventually become a viable process for sustainable H2 production on a global scale.

215