Investigation of the Alkaline Electrochemical Interface and Development of Composite

Metal/Metal-Oxides for Hydrogen and Oxygen Electrodes

by Michael Bates

B.S. Chemistry, Salem State University

A dissertation submitted to

The Faculty of

the College of Science of

Northeastern University

in partial fulfillment of the requirements

for the degree of Doctor of Philosophy

April 6, 2015

Dissertation directed by

Sanjeev Mukerjee

Professor of Chemistry and Chemical Biology

ii

Acknowledgements

I would like to begin by thanking the individual without whom I would never have made

it to this point in my scientific career: my wife Kathleen. Were it not for her unfaltering

encouragement and patience, I would certainly have chosen an easier yet significantly less

rewarding path for my life. In addition, the support provided by our families: Jane Sullivan,

Mary and Arthur Bates and Marcia and Jack Bender, has been invaluable. I am so very blessed to

have been raised by such wonderful parents who encouraged me to follow my dreams and

always believed that I would succeed in achieving my goals.

Next, I wish to express my gratitude to my advisor, Professor Sanjeev Mukerjee, who

provided me with every opportunity to succeed. I am very grateful to have been able to work for

him at the Northeastern University Center for Renewable Energy Technology (NUCRET). The

access to world-class analytical equipment has been truly rewarding. Also, Dr. Mukerjee has

instilled within me a strong sense of self-reliance and determination which I believe will serve

me very well in the challenging world of scientific research.

I would also like to thank all of my Thesis Committee members for their help and

support: Dr. David Budil, Dr. Max Diem and Dr. Geoffrey Davies. Also, thanks to all the

outstanding administrative staff of the Northeastern University Department of Chemistry and

Chemical Biology, in particular Cara Shockley and Andrew Bean for all their help and support

over the years. And thanks to the Northeastern University College of Science for giving me the

opportunity to conduct my PhD research.

While the path to earning a PhD is quite challenging and stressful, I was very lucky to

enjoy great support and camaraderie from my colleagues at NUCRET. I thank Dr. Nagappan

Ramaswamy for teaching me so many practical and fundamental lessons of electrochemical

iii

analysis. Special thanks to Kara Strickland, Dan Abbott and Urszula Tylus for their help, support

and friendship over the years. I would especially like to thank Bob Allen for sharing with me his

vast scientific knowledge, from inorganic synthesis to the development of practical

electrochemical devices, without Bob’s advice and mentorship I would have much less data in

this document and knowledge in my head. In addition, I wish to express my sincere gratitude to

Dr. Qingying Jia for his tireless work conducting meticulous analysis of the in-situ XAS data.

Without such high-quality XAS analysis, this research would certainly not contain the depth of

insight into the fundamental descriptors of the electrocatalysis provided by Jia’s efforts.

I would also like to acknowledge and thank the numerous industry collaborators, without

whose support (financial and managerial) none of these projects would have been possible. From

Proton On-Site, thanks to Dr. Katherine Ayers for sharing her knowledge of electrochemisty.

Also, thanks to Chris Capuano, Luke Dalton and Mike Niedzwiecki of Proton On-Site for their

help and support with the alkaline electrolysis research. Thanks to collaborators at Penn State,

Dr. Mike Hickner and Dr. Yongjun Leng for their discussions of the alkaline polymer electrolyte

research. On the hydrogen-bromine redox-flow-battery research, thanks to Dr. Gaungyu Lin and

Dr. Trung Van Nguyen for their advice and support. Finally, without financial support from

ARPA-E none of this research would have been possible.

iv

Abstract of Dissertation

Understanding the fundamentals of electrochemical interfaces will undoubtedly reveal a

path forward towards a society based on clean and renewable energy. In particular, it has been

proposed that hydrogen can play a major role as an energy carrier of the future. To fully utilize

the clean energy potential of a hydrogen economy, it is vital to produce hydrogen via water

electrolysis, thus avoiding co-production of CO2 inherent to reformate hydrogen. While

significant research efforts elsewhere are focused on photo-chemical hydrogen production from

water, the inherent low efficiency of this method would require a massive land-use footprint to

achieve sufficient hydrogen production rates to integrate hydrogen into energy markets. Thus,

this research has primarily focused on the water splitting reactions on base-metal catalysts in the

alkaline environment. Development of high-performance base-metal catalysts will help move

alkaline water electrolysis to the forefront of hydrogen production methods, and when paired

with solar and wind energy production, represents a clean and renewable energy economy. In

addition to the water electrolysis reactions, research was conducted to understand the de-

activation of reversible hydrogen electrodes in the corrosive environment of the hydrogen-

bromine redox flow battery. Redox flow batteries represent a promising energy storage option to

overcome the intermittency challenge of wind and solar energy production methods.

Optimization of modular and scalable energy storage technology will allow higher penetration of

renewable wind and solar energy into the grid.

In Chapter 1, an overview of renewable energy production methods and energy storage

options is presented. In addition, the fundamentals of electrochemical analysis and physical

characterization of the catalysts are discussed. Chapter 2 reports the development of a Ni-Cr/C

electrocatalyst with unprecedented mass-activity for the hydrogen evolution reaction (HER) in

v

alkaline electrolyte. The HER kinetics of numerous binary & ternary Ni-alloys and composite

Ni/metal-oxide/C samples were evaluated in aqueous 0.1 M KOH electrolyte. The highest HER

mass-activity was observed for Ni-Cr materials which exhibit metallic Ni as well as NiOx and

Cr2O3 phases as determined by ex-situ XRD and in-situ XAS analysis. The on-set of the HER is

significantly improved compared to numerous binary and ternary Ni-alloys – including state-of-

the-art Ni-Mo materials. It is likely that at adjacent Ni/NiOx sites, the oxide site facilitates

formation of adsorbed hydroxide (OHads) from the reactant (H2O) thus minimizing the high

activation energy of cleaving the H-OH bond to form the Hads HER intermediate on the metallic

Ni site. This is confirmed by in-situ XAS studies which show that the synergistic HER

enhancement is due to NiOx content and that the Cr2O3 appears to stabilize the composite NiOx

component under HER conditions (where NiOx would typically be reduced to metallic Ni0).

Furthermore in contrast to Pt, the Ni(Ox)/Cr2O3 catalyst appears resistant to poisoning by the

anion exchange ionomer (AEI), a serious consideration when applied to an anionic polymer

electrolyte interface. Furthermore a model of the double layer interface is proposed, which helps

explain the observed ensemble effect in the presence of AEI.

In Chapter 3, Ni-Fe and Ni-Fe-Co mixed-metal-oxide (MMO) films were investigated for

oxygen evolution reaction (OER) activity in 0.1M KOH on high surface area Raney-Nickel

supports. During investigations of MMO activity, aniline was identified as a useful “capping

agent” for synthesis of high-surface area MMO-polyaniline (PANI) composite materials. A Ni-

Fe-Co/PANI-Raney-Ni catalyst was developed which exhibits enhanced mass-activity compared

to state-of-the-art Ni-Fe OER electrocatalysts reported to date. Furthermore, in-situ XAS

analysis revealed charge-transfer effects of MMOs in which the average oxidation state of the

OER-active NiOx(OH)y sites is affected by the binary or ternary components (Fe &/or Co).

vi

Cyclic voltammetry results show changes in the potential of the Ni2+/3+

transitions in the presence

of binary or ternary metals. In-situ XAS analysis confirms that the redox peaks can be attributed

to the Ni sites and the shifts in the XANES peak as a function of applied potential indicates that

Fe acts to stabilize Ni in the 2+ oxidation state, while Co facilitates oxidation to the 3+ state.

The enhanced OER activity of the ternary Ni-Fe-Co/PANI-Raney catalyst is likely due to

“activation” of the conductive Ni(III)OOH phase at lower overpotential due to the charge-

transfer effects of the cobalt component. The morphology of the MMO catalyst film on

PANI/Raney-Ni support provides excellent dispersion of active-sites and should maintain high

active-site utilization for catalyst loading on gas-diffusion electrodes.

In Chapter 4, the de-activation of reversible-hydrogen electrode catalysts was

investigated and the development of a Pt-Ir-Nx/C catalyst is reported, which exhibits

significantly increased stability in the HBr/Br2 electrolyte. Initial screening of Rh- and Ru-

chalcogenides (oxides, sulfides and selenides) indicates that these non-Pt catalysts do not exhibit

sufficient hydrogen reaction kinetics for use in the hydrogen electrode of a H2-Br2 redox flow

battery (RFB). However, a standard Pt/C catalyst suffered from rapid and irreversible de-

activation upon high-voltage cycling or exposure to Br2. In contrast a Pt-Ir/C catalyst exhibited

increased tolerance to high-voltage cycling and in particular showed recovery of electrocatalytic

activity after reversible de-activation (presumably from bromide adsorption and subsequent

oxidative bromide stripping). Under the harshest testing conditions of high-voltage cycling or

exposure to Br2 the Pt-based catalyst showed a trend in stability: Pt < Pt-Ir < Pt-Ir-Nx.

Finally, Chapter 5 presents a summary of work and proposals for future developments of

composite metal/metal-oxide catalysts for the HER & OER in alkaline electrolyte, as well as

nitrogen-functionalized PGM catalysts for reversible hydrogen electrodes in H2-Br2 RFBs.

vii

TABLE OF CONTENTS

Acknowledgements ii

Abstract iv

Table of Contents vii

List of Figures ix

List of Tables xii

List of Abbreviations and Symbols xiii

Chapter 1 Introduction 1

1.1 Modern Energy Economy – Beyond Fossil Fuels 1

1.2 Electrochemical Energy Storage & Conversion Technologies 5

1.3 Hydrogen Production Methods 7

1.4 Fundamentals of Electro-analytical Techniques 9

1.5 Double-Layer Structure and Ionomer Effects 12

1.6 Electrocatalysis of Hydrogen Reactions Acid vs. Alkaline 14

1.7 Electrocatalysis of Oxygen Evolution Reaction 17

1.8 Physical Characterization: XRD, SAXS & SEM/EDS 19

1.9 Scope of Dissertation 23

1.10 References 25

Chapter 2 Composite Metal/Metal-Oxides

for Hydrogen Reactions at High pH 28

2.1 Introduction 28

2.2 Experimental 32

2.3 Results and Discussions 35

2.4 Conclusions 76

2.5 Acknowledgements 78

viii

2.6 References 79

Chapter 3 Alkaline Oxygen Evolution Reaction on Mixed-Metal-Oxides 86

3.1 Introduction 86

3.2 Experimental 92

3.3 Results and Discussions 94

3.4 Conclusions 117

3.5 Acknowledgements 118

3.6 References 119

Chapter 4 Development of Electrocatalysts for

H2-Br2 Redox Flow Batteries 124

4.1 Introduction 124

4.2 Experimental 129

4.3 Results and Discussions 131

4.4 Conclusions 152

4.5 Acknowledgements 153

4.6 References 154

Chapter 5 Thesis Summary and Future Directions 156

5.1 Introduction 156

5.2 Chapter 2 – Alkaline HER/HOR Electrocatalysis Conclusions 157

5.3 Chapter 3 – Alkaine OER Conclusions 158

5.4 Chapter 4 – HBr Reversible Hydrogen Electrode Conclusions 159

5.5 References 162

ix

List of Figures

Chapter 1:

Figure 1: Global renewable energy production capacity. 3

Figure 2 – Schematics of reversible fuel cell and redox flow battery systems. 6

Figure 3 – Model of Electrochemical Double-Layer. 12

Chapter 2:

Figure 1 – Electrolysis cell polarization curves. 30

Figure 2 –HER/HOR kinetics of standard Pt/C in acid vs. base. 36

Figure 3 –Hupd charge from standard Pt/C in acid vs. base. 38

Figure 4 – Hupd charge & and HER/HOR kinetics of various Pt-alloy samples. 40

Figure 5 –CV response & HER-HOR kinetics of Pt-Co/C before & after acid-washing. 41

Figure 6 – Pt particle size effects on HER-HOR activity. 42

Figure 7 – Hupd & “Hopd” observed for various Pt electrodes in 0.1 M KOH. 44

Figure 8 –Hupd & Hopd of Pt/C in presence of Nafion vs AS-4 ionomer binders. 46

Figure 9 – XRD analysis of Pt black and standard Pt/C catalysts. 48

Figure 10 - HER activity of Pt/C & Pt black in the presence of Nafion vs. AS-4 binder. 49

Figure 11 – Model of electrostatic effects of AEI on double-layer for alkaline MOR & HER. 50

Figure 12 – Initial screening of non-PGM catalysts for alkaline HER. 53

Figure 13 – Screening of select binary & ternary Ni-alloys for alkaline HER. 54

Figure 14 - Screening of composite metal/metal-oxides for alkaline HER. 55

Figure 15 – Steady-state alkaline HER performance of composite metal/metal-oxides. 56

Figure 16 – SEM images of Ni-Cr/C Ni/Mo-Ox/C catalysts. 58

Figure 17 – Steady-state alkaline HER performance of Ni-Cr/C vs Ni/Mo-Ox/C. 58

Figure 18 – Alkaline HER kinetics of catalysts with Nafion vs. AS-4 binder. 61

Figure 19 – Interface model for alkaline HER on pure metal and M/MOx surfaces. 64

Figure 20 – SEM image of Ni-Mo nanoparticles. 65

Figure 21 – Alkaline HER on Ni-Cr/C vs Ni-Mo nanoparticles at 1 mg/cm2 loading. 66

x

Figure 22 – Interfacial study of Ni-Mo nanoparticles. 67

Figure 23 – Comparison of alkaline HER activity on various Ni-based samples. 68

Figure 24 – XRD Ni-Cr/C sample. 68

Figure 25 – HR-TEM images of Ni-Cr/C sample. 70

Figure 26 – Comparison of alkaline HER on original vs. alternate Ni-Cr/C samples. 71

Figure 27 – XAS data from the Cr K-edge collected on Ni-Cr/C electrodes. 72

Figure 28 – XAS data from the Ni K-edge collected on Ni-Cr/C electrodes. 72

Figure 29 – EXAFS fitting results from “Original” Ni-Cr/C sample. 74

Chapter 3:

Figure 1 – SEM images of (8:1:1 at. ratio) Ni-Fe-Co/PANI-Raney catalyst. 89

Figure 2 –OER activity & Redox peaks from mono-metallic films on Raney-Ni. 95

Figure 3 –HT-effects of Ni-Fe on Raney-Ni support: RDE & XRD. 96

Figure 4 – HT-effects of Ni-Fe-Co on Raney-Ni support: RDE & XRD. 100

Figure 5 –HT effects of C-supported Ni-Fe & Ni-Fe-Co samples: RDE & XRD. 103

Figure 6 – Steady-State OER activity of Ni-Fe-Co/PANI-Raney. 104

Figure 7 –Ni2+/3+

redox peaks & OER activity from C-supported samples. 107

Figure 8 – Fe K-edge XAS results. 109

Figure 9 – Co K-edge XAS results. 109

Figure 10 – Ni K-edge Data collected from various MMO/C samples. 111

Figure 11 – Summary of XAS analysis: correlation with Ni2+/3+

redox peak shifts. 116

Chapter 4:

Figure 1: Volcano Plot of the HER activity and M-H bond strength. 125

Figure 2 –HER-HOR activity in 0.1M HClO4 for Pt, Ir & Rh catalysts. 132

Figure 3 - HER-HOR activity in 0.5M HBr for Pt, Ir & Rh catalysts. 135

Figure 4 – HOR-HER activity in 0.5 M HBr on selected samples. 136

Figure 5 – Optimization of RhS/C heat-treatments: XRD & H2-pump testing. 137

Figure 6 – H2-pump testing of Pt-Ir-Nx/C vs RhS/C, RuS/C and standard Pt/C. 138

xi

Figure 7 –Hupd & HER-HOR kinetics on standard Pt/C in HClO4 & HBr. 141

Figure 8 – Evaluation of reversible and irreversible de-activation of Pt-Ir/C catalyst. 143

Figure 9 – Soak test results for standard Pt/C in 0.5 M HBr/1 mM Br2. 144

Figure 10 – HER-HOR before & after high-voltage cycling on PGM Standards. 145

Figure 11 – Mild stability testing for Pt/C: soak in 1M HBr (with no Br2). 146

Figure 12 – Harsh stability testing for Pt/C: soak in 1M HBr & 1mM Br2. 147

Figure 13 – High-Voltage cycling of Pt/C. 147

Figure 14 – Mild stability testing for Pt-Ir/C: soak in 1M HBr (with no Br2). 149

Figure 15 - Harsh stability testing for Pt-Ir/C: soak in 1M HBr & 1mM Br2. 149

Figure 16 – High-Voltage cycling of the Pt-Ir-Nx/C sample. 151

Figure 17 - Harsh stability testing for Pt-Ir-Nx/C: soak in 1M HBr & 1mM Br2. 151

xii

List of Tables

Chapter 1:

Table 1: Comparison of EES Technology. 7

Chapter 2:

Table 1: ECSA Values for Pt Samples – Particle Size Effects. 42

Table 2: Summary of EXAFS results of Ni-Cr/C sample. 75

Chapter 3:

Table 1: Comparison of Ni-Fe OER Activities in the Literature. 99

Table 2: Profile-fitting Results from XRD Analysis. 102

Table 3: EXAFS Fitting Results from MMO/C Samples. 115

Chapter 4:

Table 1: Electro-Chemical Surface Area Comparisons. 134

Table 2: Catalyst HER-HOR Metrics in HBr. 136

xiii

List of Abbreviations and Symbols

Ads - Adsorbate

AEI – Anion Exchange Ionomer

AEM – Anion Exchange Membrane

AEMFC – Anion Exchange Membrane Fuel Cell

APE – Alkaline Polymer Electrolyte

CA – Chrono-Amperometry

CE – Counter Electrode

CL – Catalyst Layer

CV – Cyclic Voltammetry

ECSA – Electrochemical Surface Area

FT-EXAFS – Fourier-Transformed Extended X-ray Absorption Fine Structure

GC – Glassy Carbon

GDE – Gas-Diffusion Electrode

GDL – Gas-Diffusion Layer

Hupd – Under-potentially deposited hydrogen

Hopd – Over-potentially deposited hydrogen

HER – Hydrogen Evolution Reaction

HOR – Hydrogen Oxidation Reaction

HT – Heat-Treated

IHP – Inner Helmholtz Plane

MEA – Membrane Electrode Assembly

M/MOx – Metal / Metal-Oxide

MOR – Methanol Oxidation Reaction

OCP – Open-Circuit Potential

OCV – Open-Circuit Voltage

OER – Oxygen Evolution Reaction

OHP – Outer Helmholtz Plane

ORR – Oxygen Reduction Reaction

PEM – Proton Exchange Membrane

PEMFC – Proton-Exchange Membrane Fuel Cell

PGM – Platinum Group Metal

PZC – Potential of Zero Charge

PZFC – Potential of Zero Formal Charge

QA – Quaternary Ammonium

RDE – Rotating Disk Electrode

RE – Reference Electrode

RHE – Reversible Hydrogen Electrode

rds – rate-determining step

SEM – Scanning Electron Microscopy

TOF – Turn-over Frequency

TM – Transition Metal

WE – Working Electrode

XAS – X-ray Absorption Spectroscopy

XANES – X-ray Absorption Near-Edge Spectra

XRD – X-ray Diffraction

1

Chapter 1 Introduction

1.1 Modern Energy Economy – Beyond Fossil Fuels:

Due to the well-established concerns over global climate change and energy security,

most developed nations are investing substantial resources into the development of renewable

energy technology. The major sources of renewable energy generation are: Hydropower,

geothermal, biofuels, wind & solar. However, hydropower and geothermal are geographically

limited and even “renewable” biofuel solutions still produce CO2. Thus the most viable

solutions to solve the energy crisis are wind and solar. The term energy crisis is not intended to

incite alarmism, but rather to motivate the development of alternative energy technologies to

prepare for the inevitable depletion of fossil-fuels which currently power our global economy.

The most significant challenge of utilizing wind and solar energy is managing the intermittency

problem. Modern energy storage technology is geographically limited and thus the need arises

for modular and scalable electrochemical energy storage solutions. While massive battery stacks

offer a short-term solution to the energy storage needs, the low energy density and high cost of

battery technology will eventually limit the wide-spread use of battery technology for grid-scale

energy storage. In contrast, fuel cells and redox flow batteries (RFBs) offer flexible solutions to

manage grid-scale energy storage requirements. Fuel cells offer extremely high energy-density

(giving them a small “footprint”) and power-density (which provides fast reaction to grid-

buffering needs), but at a rather high capital cost. In contrast, RFBs offer low-cost energy

storage, but their relatively low energy-density translates into a large land-use footprint and their

low-power density translates into slow reaction times to meet spikes in energy consumption.

However, both technologies have large potential markets for modular and scalable grid-scale

energy storage. This research has examined the electrochemical interfaces of fuel cells and

2

RFBs and let to the development of high-performance and high-durability electrocatalysts to

advance these technologies towards wide-spread commercialization.

To compare the various renewable energy options we first examine the global storage

capacity of various technologies. As of 2010, global hydropower accounted for ~1000 GW of

production capacity[1] and geothermal plants accounted for ~10 GW of production capacity.[2]

This corresponds to ~2% and 0.2% of global production capacity for hydro and geothermal.

Interestingly, although Iceland is typically thought of as a global leader in geothermal power

because >60% of their national electricity needs are derived from geothermal plants,[3] the total

geothermal capacity in Iceland in 2010 was 575 MW,[4] while the top three nations in terms of

total geothermal capacity in 2010 were: USA, Philippines & Indonesia[4] with national

capacities of about 3, 2 & 1 GW,[4] respectively.

In contrast, biofuels are not as strictly limited by geography, but they suffer from three

major draw-backs: 1) combustion of renewable biofuels such as biodiesel and ethanol still

produce CO2 greenhouse gas emissions which contribute to global climate change. Although the

CO2 emissions from renewable biofuels consist of a so-called “closed-loop” CO2 cycle in which

plants consume CO2 from the atmosphere to produce the raw materials for biodiesel and ethanol,

thus the cycle adds no net CO2 to the atmosphere (as opposed to combustion of fossil-fuels

which does add net CO2 to the atmosphere). 2) Biofuel production requires large land-mass – and

competes with food production to upset regional and global markets. Thus, wide-spread adoption

of biofuel production could potentially create food-shortages in under-developed regions. 3) The

construction of biofuel farms is often concomitant with deforestation and thus the loss of forests

to produce biofuels detracts from the “closed-loop” CO2 argument in favor of biofuels. In

addition, the deforestation to produce biofuels threatens ecosystems and biodiversity. Thus,

3

while biofuels represent a possible bridging technology to transition away from the combustion

of fossil-fuels, they do not represent a “silver bullet” to solve the impending global energy crisis.

Just as hydropower, geothermal and biofuels are to varying extents geographically

limited, so are wind and solar (although to a lesser extent). However, in any location on earth

the wind will occasionally blow and the sun will occasionally shine. As of 2010 global wind

power station capacity was 318 GW and global solar-array capacity was 139 GW.[1] Wind and

solar represent the most versatile options for the adoption of renewable energy sources. In

addition to the versatility of wind and solar, these renewable energy production methods

represent the largest power capacity (as shown in figure 1) and are the fastest-growing renewable

energy production methods.

Figure 1: Global renewable energy production capacity. Data adapted from Chamorro et al.[2]

The primary challenge of harnessing wind and solar energy is the intermittency issue.

Photovoltaic arrays and wind farms use the solar and wind energy, respectively to produce

electricity. Electricity is a transient form of energy which must be used as soon as it is

generated. Unfortunately, the power-consumption trends of society do not match the power

4

production windows for wind and solar. While it may be possible to offer incentives for

industrial electricity use to match renewable production, the household needs of consumers

would be very difficult to match with renewable production, even with the development of

smart-grid infrastructure. Thus the development of wind and solar energy production facilities

must be linked to energy storage facilities so that the renewable energy can be stored and used

later, when it is needed.

Today, the most cost-effective energy storage technologies are “pumped hydro” and

compressed air.[5] Pumped hydro refers to the simple process of pumping large amounts of

water to a higher elevation to store energy and later releasing the water to flow downhill and turn

a waterwheel-driven turbine to use the energy on demand. This is by far the lowest-cost

technology to store energy, but like geothermal energy production, it is very geographically

limited. Natural differences in elevation (hills and mountains) are required to make pumped

hydro economical. For instance, pumped hydro wouldn’t make much sense in geographically

flat regions such as Kansas or Texas. In a particular interesting anecdotal example, the Swiss

government has benefited greatly from their natural resources (the Alps) because the relatively

flat country of Germany, which produces 30% of their national electricity via renewable wind

and solar, has decided to pay the Swiss a premium to use Germany’s wind and solar energy to

pump water up the Swiss Alps and then sell the energy back to Germany when they need it.

Like pumped-hydro, compressed-air storage is also geographically limited. Compressed-

air storage requires large subterranean cavities to store compressed gas, making this storage

technology prohibitively expensive except in locations with ideal natural geological conditions.

In contrast to the geographic limitations of pumped-hydro and compressed air, electrochemical

5

energy storage technologies offer an array of versatile, modular and scalable solutions to the

intermittency problems of wind & solar power.

1.2 Electrochemical Energy Storage & Conversion Technologies:

The primary technology candidates for electrochemical energy storage (EES) are: NaS

batteries, Li-ion batteries, redox flow batteries and fuel cells. While batteries offer a nice

modular “plug & play” advantage, they also suffer from relatively low energy density and thus

require very large facilities to meet the needs of grid-scale energy storage. Also, NaS batteries in

particular require constant heating to keep the systems above 300°C so that the molten salts don’t

crystallize. Thus, while NaS battery systems are currently being used for grid scale energy

storage, the low over-all system efficiency (when battery heating systems are accounted for) and

high cost will inevitably limit the wide-scale use of these systems. Li-ion batteries offer the

advantages of extremely high charge/discharge efficiency and operation at ambient temperature;

they suffer from extremely high capital costs compared to other EES options. In addition, Li-ion

batteries use non-aqueous, reactive organic electrolytes which present significant safety hazards.

The two remaining options for EES technologies are redox flow batteries (RFBs) and fuel

cells. To clarify, because the fuel cell itself is an energy conversion (not energy storage),

technology the term fuel cell in the context of this discussion refers to a reversible fuel cell &

electrolysis system, henceforth called reversible fuel cell (RFC). In addition, the focus of this

discussion relates to the “low temperature” polymer electrolyte fuel cells and not the high-

temperature solid oxide or molten carbonate fuel cells. The high temperatures required for the

operation of solid oxide or molten carbonate fuel cells make these systems viable candidates for

combined heat and power systems and possibly niche applications in grid-scale energy storage,

but the high-temperature systems are not considered for grid-scale EES in this discussion.

6

Figure 2 – Top: Schematic of RFC system from Park et al.[6], reproduced with

permission from the Royal Society of Chemistry. Bottom: Schematic of generic RFB system

from review by Weber et al.[7] Reproduced with permission from publisher.

The primary advantage of the RFC system is the extremely high theoretical energy

density. Because the H2 fuel is the smallest molecule and the oxidant (O2) is all around us in the

7

air, the energy density of oxidizing hydrogen gas is the highest of any EES technology.

However, RFC systems do suffer from numerous drawbacks. The three most significant being:

safety and cost of hydrogen storage, low charge/discharge efficiency and the requirement of

precious metal catalysts. On the other hand, RFBs offer the advantages of low-cost electrodes

and fuel storage tanks and high charge/discharge efficiency. However, the energy-density and

power-density of RFB systems is at least 10X less than fuel cell systems – thus requiring very

large RFB systems to match the energy and power capacity requirements of grid-scale energy

storage. Table 1 lists the pros & cons of the best EES options.

Table 1 – Comparison of EES Technology. [8-10]

System Type (Efficiency, %)

Advantages Drawbacks Practical

Specific Energy (Wh/kg)

NaS Battery (75-90 %)

High Efficiency Requires heating to 300°C 150 - 250

Li-ion Battery (>95%)

Very High Efficiency Safety concerns due to electrolyte,

High cost 75 - 200

Vanadium RFB (70-80%)

Low cost, Flexible Scalability

Low Power Density: 0.5 - 1 kA/m2 10 - 50

RFC (~50%)

High Power Density: 5 - 20 kA/m2

High cost of H2 fuel storage, Precious metal catalysts,

Slow ORR kinetics (low efficiency) 450+

1.3 Hydrogen Production Methods:

Greater than 90% of global production of hydrogen gas uses the method of steam

reforming of methane or other fossil fuels. Other hydrogen production methods include: coal

and biomass gasification, high-temperature thermo-chemical cycles (such as Cu-Cl and S-I), and

water electrolysis.[11] The two main drawbacks of steam reforming of fossil fuels to produce

8

“reformate hydrogen” are: the massive amounts of CO2 produced from this thermally intensive

process as well as the CO and CO2 impurities in the reformate hydrogen which poison fuel cell

catalysts. Most of the global production of hydrogen gas is used in industrial synthesis of

ammonia via the Haber process, thus the purity of the hydrogen isn’t a major concern. However,

literature reports have shown that even levels as low as 20 ppm CO in the reformate hydrogen

can poison fuel cell catalysts and lead to significantly decreased cell efficiency.[12] In contrast

to reformate hydrogen, water electrolysis produces hydrogen with no impurities. However, the

high energy costs and requirements of precious metal catalysts have prevented the wide-spread

use of water electrolysis for hydrogen production. However, the development of alkaline

polymer electrolyte membranes enables the use of non-precious metal catalysts for the water

electrolysis reaction, thus significantly decreasing the cost of hydrogen production. The

development of base-metal catalysts for alkaline electrolysis has been a major focus of this

research.

On a side-note, a hybrid of steam reforming and electrolysis is available in the form of

the direct methanol hydrogen pump. In this configuration, liquid methanol is oxidized at the

anode of a PEM electrolyzer and the protons are transported through the proton exchange

membrane to a reformate-tolerant (Pt-alloy) cathode where the hydrogen evolution reaction

occurs to produce hydrogen gas. Although this system does produce CO2 emissions, it is

theoretically much more efficient than the high-temperature steam reforming of methane and

produces hydrogen fuel with an intermediate level of CO & CO2 impurities. Further purification

could be conducted with a secondary electrochemical hydrogen-pump. In addition, the methanol

fuel could be produced from the fermentation of renewable biofuels, resulting in a closed-loop

9

CO2 system. However, the main focus of this work is water electrolysis and the hydrogen and

oxygen reactions at high pH.

1.4 Fundamentals of Electrochemical Analysis:

Electrochemical reactions are controlled by thermodynamic and kinetic effects. Each

reaction is fundamentally associated with a thermodynamic equilibrium potential. The

thermodynamics of an electrochemical reaction dictate whether a certain reaction is considered

spontaneous or non-spontaneous. The redox potential of any given half-reaction (i.e. oxidation

or reduction) is related to the Gibb’s free energy of the half-reaction by the equation:

(1.1) ΔG° = -nFE°

Where ΔG° is the standard Gibbs free energy of the reaction, n is the number of moles of

electrons transferred in the reaction, F is Faraday’s constant (96,485 C / mol e-) and E° is the

standard redox potential of the reaction. Traditionally, cathodic electro-reduction reactions of

analytes were catalogued by early electrochemists and thus tables of standard reduction

potentials are typically consulted or referenced in electrochemical research. However, the

standard oxidation potentials of any given reaction can be easily calculated because the reverse

reaction would simply change the sign (+/-) of ΔG° and thus E° such that E°oxid = -E°red.

Furthermore, the full-cell potential resulting from the pairing of two half-reactions can be

calculated from the sum of the reduction and oxidation potentials: Ecell = Ered + Eoxid. The sign

(+/-) of the full-cell potential dictates whether a set of given oxidation and reduction reactions is

spontaneous (-ΔG = +Ecell, “galvanic cell”) or non-spontaneous (+ΔG = -Ecell, “electrolytic cell”).

10

In addition to the standard values of the redox potential of any given reaction, the Nernst

equation allows for the determination of redox potentials under non-standard conditions (i.e. not

at 273 K, 1 atm, and 1 M [O/R]). The Nernst equation is shown here:

(1.2) Reduction reactions: E = E° - (RT/nF)*LnQ

(1.3) Oxidation reactions: E = E° + (RT/nF)*LnQ

In the Nernst equation, R is the gas constant (8.315 J*mol-1

*K-1

) and Q is the reaction

quotient of the redox reaction. It should be noted that the value of 8.315 J*mol-1

*K-1

is used for

R because a joule is a coulomb * volt, and as such the quotient of RT/nF results in a value in

units of voltage. Thus, the Gibb’s free energy of the reactions determines the thermodynamic

properties, and the Nernst equation can be used to correct the thermodynamic redox potential

under non-standard conditions. However, thermodynamic considerations simply indicate the

potential at which a certain reaction will occur and the spontaneity of a reaction. Practical

electrochemistry of real devices also requires the quantification of the electrochemical reaction

kinetics to determine the current that is produced at a given cell voltage (or vice versa).

To evaluate electrode kinetics, we turn to the Butler-Volmer equation:

(1.4) i = A* i° * [exp(-αfη) - exp((1-α)fη)]

Where A is the area of the electrode, i° is the exchange current density at the equilibrium

potential (i° is proportional to the kinetic rate constant, k° among other factors), α is the charge

transfer coefficient (unit-less, typically with a value of ~0.5), f represents the combination of the

three constants F/RT, and η is the over-potential applied to the system. The term over-potential

refers to the difference Eapplied – E°. The full Butler-Volmer equation, represented in Eq. 1.4,

accounts for the forward and reverse contributions to a given electrochemical reaction at

11

potentials very close to E°. Under conditions where a large magnitude of anodic or cathodic

over-potential is applied, the full Butler-Volmer equation simplifies to:

(1.5) Anodic η: i = A * i° * exp((1-α)*f*ηanodic)

or:

(1.6) Cathodic η : i = A * i° * exp(-α*f*ηcathodic)

In the case of Eq.s 1.5 & 1.6, under the force of an applied over-potential (η) beyond the

thermodynamic equilibrium potential (E°) the rate constant of the forward potion of the half-

reaction is much greater than the reverse reaction (i.e. kforward >> kreverse) and thus the contribution

of the reverse reaction to the measured current is negligible. Thus Eq.s 1.5 & 1.6 are typically

used to model the electro-kinetics of real-life electrodes under operating conditions.

Furthermore, the behavior of a “full-cell” redox device must account for anodic and cathodic

over-potential as well as Ohmic over-potential and mass-transport over-potential at high current

density. Therefore the performance of a full-cell device is characterized by a polarization curve

which measures current and voltage, from a maximum voltage (the open circuit potential,

determined by EOPC = E0

Cathodic – E0

Anodic) to the maximum current (the short circuit current, at

which ECell = 0 V). The factors contributing to the measured cell voltage are:

(1.7) ECell = E0 – ηanodic – ηcathodic – ηohmic – ηmass-transport

In Eq. 1.7, the anodic and cathodic over-potential is described above in Eq.s 1.5 & 1.6.

The Ohmic-overpotential is determined by Ohm’s law (V = iR), such that the voltage loss due to

resistance is proportional to the product of the cell current and the cell resistance. Finally, the

mass-transport over-potential occurs in situations where slow diffusion of reactants or products

limit the current density. The mass-transport behavior of an electrochemical system is

determined by many factors, including diffusion, migration and convection. However a

12

thorough discussion of the quantitative analysis of mass-transport phenomenon is beyond the

scope of this discussion. Suffice to say, the focus of this research is the kinetic over-potential of

anodic and cathodic reactions in alkaline electrolytes. In addition to this kinetic analysis, this

research included an analysis of the stability of catalysts for the reversible hydrogen reactions in

the highly corrosive environment of the HBr/Br2 electrolyte used in the H2-Br2 RFB.

1.5 Double-Layer Structure and Ionomer Effects:

Figure 3 – Model of Electrochemical Double-Layer. Reproduced from Jingjie et al.[13]

reproduced with permission from the Journal of the Electrochemical Society.

Electrochemical reactions occur at the complex interface of an electrically conducive

solid electrode and liquid electrolyte. Because a balance of charge must be maintained in at this

interface, ion transport (diffusion, migration and convection) as well as atomic-scale electrostatic

effects play a significant role in determining the electro-kinetics of a reaction. While the mass-

transport of charged ionic species has been thoroughly studied and documented in various

13

electrochemical engineering texts, the less obvious yet very significant electrostatics of the

electrochemical interface have been of particular interest in this research. On the metal side of

the interface, the large excess of free electrons in the conducting electrode act to rapidly

equilibrate and charge imbalance and thus the electrode is modeled as a planar surface

(neglecting roughness factors and nano-scale heterogeneity of composite surfaces such as metal /

metal-oxides) with a uniform charge, which is relative to the applied potential and the electrode

surface properties. In contrast, the transport properties of charge carriers (ions) on the liquid side

of the interface are much lower than for charge carriers (electrons and holes) in the metal. Thus

the focus on the interface is on the liquid side where build-up of charge occurs and effects

electro-kinetics. A typical electrochemical interface is referred to as a “double-layer”. This term

signifies the two layers identified as the Inner Helmholtz Plane (IHP) and the Outer Helmholtz

Plane (OHP). The IHP represents the closest region to the electrolyte and this plane contains a

single layer of solvent molecules (with their dipoles oriented wrt the electrode charge) and/or

specifically adsorbed ions. The OHP is the next layer away from the electrode surface and it

contains a layer of solvated ions which balance the applied charge on the electrode. Beyond the

OHP lies an intermediate region called the “diffuse layer” with properties similar to that of the

bulk electrolyte except in extreme conditions of very high concentration, voltage, etc. The

formal models of the double-layer have evolved from the Helmholtz model developed in the

mid-1800s through Guoy-Chapman, Stern and Grahame models of the early and mid-1900s to

the modern models of Bockris, Trasatti and Conway. Each model built on the last and more

accurately interpreted the atomic-scale phenomena at these complex interfaces.

Today we have a great understanding of the behavior of solvent and ionic species at the

interface of an electrolyte with a homogeneous electrode. However, this field of study is far

14

from complete as modern electrodes consist of heterogeneous surfaces with non-uniform

properties. Thus modeling of these composite interfaces, which have been shown to exhibit

greatly enhanced electro-kinetics, is the next stage in the development of electrochemical

interfaces.

1.6 Electrocatalysis of Hydrogen Reactions – Acid vs. Base:

In the water electrolysis process, H2O is split into hydrogen and oxygen gases.

Historically, industrial electrolysis was conducted at high pH in hot, concentrated caustic

electrolyte (20-30 wt% KOH, T >80°C). The water-splitting reactions could be achieved on

porous Raney-Ni electrodes. While the concentrated caustic electrolyte provided high

conductivity and thus relatively low Ohmic contribution to over-potential, operation of such

systems was very hazardous. In contrast, modern water electrolysis is conducted using proton-

exchange membrane (PEM) solid polymer electrolytes. These PEM systems offer the advantage

of high ionic conductivity without the hazards associated with concentrated liquid electrolytes.

However, the PEM systems are very acidic due to the high proton capacity and ionic

conductivity of the membranes. Thus, the Raney-Ni electrodes historically used in liquid caustic

electrolytes would rapidly dissolve in the acidic conditions of the PEM systems. Therefore, the

only catalytic metals stable in the highly acidic PEM systems are noble metals such as Pt, Ir and

Ru (and Rh, Re & Os to a lesser extent as Pt-alloys). Because of the high cost and low

availability of the platinum-group metal (PGM) catalysts, researchers have recently focused

substantial efforts into the development of anion-exchange membranes (AEM) as analogues to

PEMs, but which operate at the high pH (alkaline) conditions under which base metal catalysts

are stable. There have very recently been significant improvements in the ionic capacity,

15

conductivity and stability of such AEMs[14] and thus the research reported here has evaluated

base metal catalysts which exhibit high activity for the hydrogen and oxygen evolution reactions

as well as for the hydrogen oxidation reaction under alkaline conditions.

Operation in high vs. low pH (i.e. alkaline vs. acid electrolytes) offers both advantages

and challenges. The primary advantage of alkaline electrocatalysis is the enhanced facility of the

complex oxygen reactions While some researchers have claimed increased kinetics of the oxygen

reduction reaction (ORR) in alkaline media,[15] the observed facility of ORR on various

electrodes in alkaline media has more recently been explained in terms of stabilization of the

ORR intermediates for outer-sphere electron transfer (often favoring the undesired 2 e- ORR

pathway).[16] For the oxygen evolution reaction (OER) in alkaline electrolyte, the on-set

potential and reaction kinetics appear to be favorable on many metal oxide electrodes.[17, 18]

The inexpensive & abundant metal oxides exhibit OER performance which rivals PGM catalyst

activity. However, the primary challenge of alkaline electrocatalysis is the well-documented

decrease in hydrogen reaction kinetics. Gasteiger et al.[19] have quantified the two-order-of-

magnitude (i.e. 100-fold) decrease in HER & HOR kinetics on Pt in alkaline vs. acid electrolyte.

However, the exact cause of the decreased hydrogen reaction kinetics in alkaline is still not well-

defined. Most researchers generally agree that the cause of the decreased HER kinetics in

alkaline is due to the different proton-source in alkaline vs. acid. The HER proceeds via three

fundamental steps (shown for alkaline conditions, for acidic conditions H2O is replaced by H3O+):

Volmer: H2O + e- → Hads + OH

- (Electrochemical)

Heyrovsky: Hads + H2O + e- → H2 + OH

- (Electrochemical)

Tafel: 2Hads → H2 (Chemical)

16

The large Ea of formation of the Hads intermediate (Volmer reaction) is likely the cause of

the lower hydrogen kinetics in alkaline vs. acid media.

(1.8) In acid: H3O+ + e

- → Hads + H2O

(1.9) In alkaline: H2O + e- → Hads + OH

-

The activation energy of reaction (1.8) is presumably much lower than for reaction (1.9).

Although DFT studies have evaluated the energy of HER reactions in acidic media, the literature

appears devoid of any computational studies comparing the Ea of reactions (1.8) & (1.9) – likely

due to the complications in accurately modeling the solvent & pH effects when attempting to

quantify reaction energies in heterogeneous catalysis.[20] As noted above, this proton-source

hypothesis is generally accepted in the research community as the cause for the slower HER

kinetics in alkaline.

In contrast to the relatively apparent cause of the slow HER kinetics, the well-

documented decrease in HOR kinetics is a point of argument in the research community. It is

believed by some that the Pt-H bond strength dictates the HOR/HER activity. Recent reports

have argued that the positive shift in Hupd peak potential on Pt catalysts in base (compared to the

Hupd features on Pt in acid) is indicative of stronger Pt-H bond strength, thus inhibiting HER-

HOR reactions. In contrast to the theory that Pt-H bond strength dictates HOR activity, other

researchers believe that the decreased HOR kinetics in alkaline electrolyte are related to the

necessity of using OH- as a proton sink for oxidizing the Hads intermediate in the electrochemical

Volmer reaction mechanism. From a fundamental electrostatic perspective, under HOR

conditions, the electrode is negative of the potential of zero charge (pzc) and thus the negative

charge on the electrode surface would present a significant electrostatic barrier to transport of the

OH- anion from the bulk electrolyte through the double-layer to the electrode surface. While the

17

role of OH- is contested in the literature, preliminary studies reported in this research have

indicated that mixing the HOR-active Pt with an additional oxophilic surface increases the HOR

activity by providing OHads in the IHP on the oxophilic surface site. For the binary catalysts, the

pzc is not uniform across the entire surface and thus the oxophilic surface sites can form an

OHads passivation layer directly from the water solvent molecules present in the double-layer –

thus avoiding the electrostatic barriers inhibiting OH- transport through the double-layer.

Preliminary testing results have indicated that binary Pt-X catalysts where X is an oxophilic

element (Ru, Ir or Nb) exhibit alkaline HOR/HER kinetics similar to the “non-polarizable” acid

case. These results are in agreement with the theory of Markovic et al.[21-24] which suggest that

composite M/MOx surfaces show enhanced kinetics for numerous reactions in alkaline

electrolytes. However, the M/MOx theory has recently been challenged by Gasteiger,[25]

Zhuang[26] and Yan,[27] each of whom have argued that the positive shift in Hupd peak potential

with increasing pH indicates an increased Pt-H bond strength and thus decreased HOR kinetics.

1.7 Electrocatalysis of Oxygen Evolution Reaction (OER):

The Oxygen Evolution Reaction (OER) is critical for clean energy conversion and

storage. Electrochemical water splitting requires catalysts for hydrogen evolution and oxygen

evolution. Like the oxygen reduction reaction in fuel cells, the oxygen evolution reaction in

water electrolysis is the primary obstacle which requires advanced catalysts to increase system

efficiency. Recent breakthroughs in alkaline polymer electrolytes (APEs) have sparked a flurry

of research into non-platinum group metal (non-PGM) electrocatalysts for fuel-cells and

electrolyzers.[28-30] For the OER, research has identified three main classes of non-PGM

electrocatalyst materials: spinels (with AB2O4 crystal structure), perovskites (with ABO3 crystal

18

structure) and mixed metal-oxides (“MMO”, typically with MxM1-x(OH)y crystal structure). A

significant trend noted across all three types of OER materials is that samples with mixed metals

(Ni-Fe, Ni-Co & Fe-Co) in the active-site position (A-site for spinels, B-site for perovskites and

M-site for MMOs) appear to exhibit enhanced OER activity compared to samples with mono-

metallic active sites. This research has shown that these mixed-metal catalysts likely offer

enhanced reversibility of the redox potential of the active-site metal via charge-transfer effects to

increase the catalyst OER activity.

Although the OER is a complex, multi-step reaction, Trasatti has recently noted that the

Yeager mechanism (which is most relevant for alkaline OER on metal oxides) emphasizes that

surfaces with highly reversible redox activity are likely to exhibit enhanced OER activity.[31] In

a similar fashion, extensive work by Lyons et al.[32] has described how the formation of

hydrated non-stoichiometric oxy-hydroxide surfaces increases OER activity on bulk metallic

electrodes. The active sites in such non-stoichiometric oxy-hydroxide networks are referred to as

“surfaquo” groups and the Lyons-Brandon mechanism has emphasized that redox reversibility is

likely a vital step in the OER mechanism on the most active Ni surfaces.[33] Thus, the

reversibility of the redox active sites appears to play an integral role in the most active OER

catalysts.

As noted above, the Yeager mechanism emphasizes the redox reversibility of the active-

site as a key descriptor for OER activity. The Yeager mechanism is presented here:

1) 2Sz + 2OH

- 2S

z-OH + 2e- (Adsorption of OH

-)

2) 2Sz-OH 2S

z+1-OH + 2e- (Oxidation of metal)

3) 2 Sz+1

-OH + 2 OH- 2S

z + O2 + H2O (Acid-Base reaction)

This mechanism was specifically developed for explaining OER kinetics on thermally-

prepared metal-oxides in alkaline electrolyte due to Yeager’s observations of a Tafel slope of 40

19

mV/dec along with a reaction order of 1 with respect to hydroxide concentration. In addition to

the Yeager mechanism, Lyons and Brandon[33] have proposed more detailed OER mechanisms

on transition metal oxide surfaces – emphasizing the concepts of the surfaquo MOx(OH)y active-

site in hydrated oxy-hydroxide network and their proposed mechanisms support the concept of

redox reversibility as a key component of highly-active OER catalysts. In addition to redox

reversibility, they discuss the acid-base properties of the oxy-hydroxide network which facilitate

the proton-transfer processes that accompany the electrochemical steps.

Careful analysis of the Ni2+/3+

redox peaks studied using cyclic voltammetry and the

metal edge energy studied using X-ray Absorption Spectroscopy (XAS), have indicated that a

ternary Ni-Fe-Co MMO catalyst exhibits unprecedented mass-activity for the OER. It is

believed that the Fe & Co components of the ternary Ni-Fe-Co MMO catalyst act to increase the

redox reversibility of the Ni active sites, thus increasing the turn-over frequency of the OER and

the resulting current density of the MMO electrode.

1.8 Physical Characterization: XRD, XAS & SEM/EDS:

X-ray diffraction (XRD) is a powerful analytical tool for materials scientists. XRD

analysis can elucidate the crystal structure of inorganic materials. In contrast to standard

diffraction methods, in which a polychromatic light source is split into its component

wavelengths by a diffraction grating with uniformly spaced gratings – XRD uses a

monochromatic light source to evaluate crystalline samples. The crystal lattice planes act as

diffraction gratings and as the light source and signal detector are moved at different angles with

respect to the sample, different crystal planes act to diffract the monochromatic light at

20

characteristic angles. The lattice spacing of a diffraction signal can be determined from the angle

at which the diffraction signal is observed using Bragg’s Law:

(2.0) Bragg’s Law: n * λ = 2d*sinθ

In which n is an integer, λ is the wavelength of the monochromatic light, d is the lattice

spacing of the diffracting plane and θ is the angle between the source/detector and the sample

plane. It is said that Bragg’s law is satisfied (i.e. the incident X-rays are diffracted) when the

value of 2d*sinθ is equal to an integer number of wavelengths, such that reflection of the

incident X-rays off of the diffracting plane results in constructive interference. If the angle of the

source/detector is such that the value of 2d*sinθ does not satisfy Bragg’s law then the signal will

undergo constructive interference and a diffraction signal will not be observed.

The observation of the diffraction pattern for an inorganic sample allows for the

determination of the crystal phase of the sample. Identification of the crystal phase reveals

information about the physical and catalytic properties of the material. In addition, XRD

analysis is vital in the screening of new catalyst materials to identify structure-activity-

relationships.

Beyond identification of the crystal-phase of a sample, XRD can be used to determine the

average crystallite size of the particles in the sample. Nano-scale effects cause non-uniform

strain. The diffracting lattice planes near the outer shell of a nano-scale particle will exhibit

strain effects due to the non-uniform lattice spacing. This non-uniform strain creates wide, low-

intensity diffraction peaks, as opposed to the narrow, high intensity peaks produced by samples

with long-range order and uniform lattice spacing. Once instrumental peak broadening effects

are accounted for, the average size of the diffracting domains (crystallites) can be calculated

using the Scherrer equation:

21

(2.1) S = Kλ / (βcosθ)

Where, S is the average size of the diffracting domain, K is a shape factor with a value

close to unity (typically 0.9 is used for spherical crystallites), β is the full width at half max

(FWHM) of the diffraction peak, and θ is the Bragg angle of the diffraction peak. Just as

identification of the crystal phase of a catalyst can provide information of its physical properties,

identification of the crystallite size can provide information about the relative surface area of the

catalytic material. Electrocatalytic reaction rates (and the resulting current density) are

proportional to the surface area of the catalyst, thus small particles provide high surface area per

mass. This is especially important for precious metal catalysts, where the mass-activity must be

optimized to make the economics of the electrochemical device feasible.

Further in-depth analysis of structure-activity-trends can be facilitated by XAS analysis.

XAS studies are conducted at synchrotron facilities such as the National Synchrotron Light

Source (NSLS) at Brookhaven National Labs (BNL). In XAS analysis, the catalyst sample is

bombarded by a high intensity of X-rays in an energy spectrum from just below to well above

the binding energy of the core electrons of a particular element. When the core-level electrons

absorb the high-energy X-rays, the electrons are promoted to the unoccupied orbitals in the

Fermi level, leaving “holes” in the core-orbitals, which are subsequently occupied by valence

electrons. The element specificity allows for an atomic-scale analysis of the local environment

of the element being analyzed. The “edge energy” i.e. the minimum energy at which X-ray

absorption occurs is related to the HOMO & LUMO states of the metal and thus provides

information about the oxidation state of the metal. This region of the XAS spectrum is known as

the X-ray Absorption Near Edge Spectra (XANES). Beyond the XANES “edge energy”,

absorption of higher energy X-rays creates photo-electrons which are promoted beyond the

22

Fermi level orbitals into the continuum. These photo-electrons create a scattering pattern as they

interact with the atoms closest to the metal center. The scattering pattern is referred to as the

Extended X-ray Fine Structure (EXAFS) and the data can be used to determine the structural and

geometric environment of the metal center (i.e. bond lengths and coordination number).

The high intensity of X-rays available at synchrotron facilities affords the ability to

analyze catalyst samples in-situ, while under potential control in the presence of an electrolyte

which closely simulates the actual operating environment of the catalyst. Normally, the

electrolyte would attenuate most of the incident X-rays and the resulting XAS signal would be

very difficult to detect, however the high intensity of the synchrotron X-rays overcomes the

electrolyte attenuation effects and allows analysis of the catalyst under true operating conditions,

thus allowing a fundamental understanding of the active-sites.

Traditional XAS analysis measures the change in the absorption coefficient (μ) as a

function of the transmitted photon intensity. From the Beer-Lambert Law:

(2.2) It = I0 exp(-μt),

Where It is the transmitted photon intensity, I0 is the incident photon intensity, μ is the

absorption co-efficient and t is the thickness of the sample. Thus in typical transmission-mode

the absorption coefficient is plotted vs. ln(I0/It). However, due to the lower core-level binding

energy of the non-precious metal catalyst samples and the relatively high absorption co-efficient

of the 0.1 M KOH electrolyte used for in-situ XAS analysis, the fluorescence signal was used to

monitor μ (indirectly) as a function of incident photon energy. In fluorescence analysis, a

Germanium detector is used to collect fluorescent photons emitted from the sample which are

generated when the outer-shell electrons “fall” into the “holes” created by excitation of core

electrons. In this case μ is proportional to If/I0. Thus the use of a 32-element Ge-detector to

23

collect the fluorescence signal of the non-PGM catalyst samples investigated in this research

project allowed for in-situ analysis of the catalysts under operating conditions. This in-situ

analysis is vital for identification of the active site and structure-activity trends.

Analysis using scanning electron microscopy and energy dispersive spectroscopy

(SEM/EDS) provide insight into the morphology and composition of the samples. SEM analysis

is very important for the development of state-of-the-art electrocatalysts because the morphology

of the catalyst determines the mass-transport characteristics and the surface-density of available

active-sites. Furthermore, the EDS capability of the SEM in the NUCRET labs allows for

identification of composition-activity-trends. EDS analysis can confirm a successful synthesis

by validating the elemental composition of a sample as well as identifying trends in composition

and catalytic activity.

1.9 Scope of Dissertation:

The topics covered in this research focus primarily on the development of electrocatalyst

for operation in the high-pH (alkaline) environment and the development of highly active and

stable reversible hydrogen electrodes for operation in alkaline and HBr electrolytes. For the

alkaline system, non-PGM catalysts were investigated for the water electrolysis reactions

(hydrogen and oxygen evolution reactions). For the HBr system a durable reversible hydrogen

electrode was developed and the stability was studied in detail in the highly poisoning and

corrosive electrolyte. In both alkaline and HBr systems, novel catalyst (PGM & non-PGM) were

compared to commercial Pt/C standard. Catalysts were characterized using XRD to determine

crystal structure, SEM to determine morphology, EDS to validate elemental composition and

XAS to elucidate the nature of the active sites under operating conditions. For the hydrogen

24

evolution reactions (HER & HOR) in alkaline, the research supports the theory by Markovic et

al.[22, 23] that composite metal / metal-oxide surfaces exhibit greatly improved electro-kinetics.

For the oxygen evolution reaction (OER) in alkaline, the research results emphasizes the

importance of redox reversibility on the OER activity. Finally, for the hydrogen reactions in

HBr, the research has carefully evaluated the effects of catalyst de-activation, showing both

reversible deactivation from halide adsorption as well as irreversible deactivation (presumably

from catalyst dissolution) from high-voltage cycling or exposure of the electrode to Br2.

25

1.10 References:

1. Rybach, L., Geothermal power growth 1995–2013—A comparison with other

renewables. Energies, 2014. 7(8): p. 4802-4812.

2. Chamorro, C.R., et al., World geothermal power production status: Energy,

environmental and economic study of high enthalpy technologies. Energy, 2012. 42(1): p.

10-18.

3. Georgsson, L.S. and I.B. Fridleifsson, Geothermal energy in the world with some

emphasis on the status in Latin America and capacity building. 001226914, 2012.

4. Bertani, R., Geothermal power generation in the world 2005–2010 update report.

Geothermics, 2012. 41(0): p. 1-29.

5. Lindley, D., Smart grids: The energy storage problem. Nature News, 2010. 463(7277): p.

18-20.

6. Park, S., et al., Oxygen electrocatalysts for water electrolyzers and reversible fuel cells:

status and perspective. Energy & Environmental Science, 2012. 5(11): p. 9331-9344.

7. Weber, A.Z., et al., Redox flow batteries: a review. Journal of Applied Electrochemistry,

2011. 41(10): p. 1137-1164.

8. Soloveichik, G.L., Battery technologies for large-scale stationary energy storage. Annual

review of chemical and biomolecular engineering, 2011. 2: p. 503-527.

9. Yang, Z., et al., Electrochemical energy storage for green grid. Chemical reviews, 2011.

111(5): p. 3577-3613.

10. Eberle, U., B. Muller, and R. von Helmolt, Fuel cell electric vehicles and hydrogen

infrastructure: status 2012. Energy & Environmental Science, 2012. 5(10): p. 8780-8798.

11. Acar, C. and I. Dincer, Comparative assessment of hydrogen production methods from

renewable and non-renewable sources. International Journal of Hydrogen Energy, 2014.

39(1): p. 1-12.

26

12. Lee, S.J., et al., Electrocatalysis of CO tolerance in hydrogen oxidation reaction in PEM

fuel cells. Electrochimica Acta, 1999. 44(19): p. 3283-3293.

13. Wu, J., et al., Electrochemical reduction of carbon dioxide I. Effects of the electrolyte on

the selectivity and activity with Sn electrode. Journal of The Electrochemical Society,

2012. 159(7): p. F353-F359.

14. Varcoe, J.R., et al., Anion-exchange membranes in electrochemical energy systems.

Energy & Environmental Science, 2014. 7(10): p. 3135-3191.

15. Spendelow, J.S. and A. Wieckowski, Electrocatalysis of oxygen reduction and small

alcohol oxidation in alkaline media. Physical Chemistry Chemical Physics, 2007. 9(21):

p. 2654-2675.

16. Ramaswamy, N. and S. Mukerjee, Fundamental Mechanistic Understanding of

Electrocatalysis of Oxygen Reduction on Pt and Non-Pt Surfaces: Acid versus Alkaline

Media. Advances in Physical Chemistry, 2012. 2012: p. 17.

17. Gong, M., et al., An Advanced Ni–Fe Layered Double Hydroxide Electrocatalyst for

Water Oxidation. Journal of the American Chemical Society, 2013. 135(23): p. 8452-

8455.

18. Li, X., F.C. Walsh, and D. Pletcher, Nickel based electrocatalysts for oxygen evolution in

high current density, alkaline water electrolysers. Physical Chemistry Chemical Physics,

2011. 13(3): p. 1162-1167.

19. Sheng, W., 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): p. B1529-B1536.

20. Santos, E., P. Quaino, and W. Schmickler, Theory of electrocatalysis: hydrogen evolution

and more. Physical Chemistry Chemical Physics, 2012. 14(32): p. 11224-11233.

21. Subbaraman, R., et al., Enhancing Hydrogen Evolution Activity in Water Splitting by

Tailoring Li+-Ni(OH)2-Pt Interfaces. Science, 2011. 334(6060): p. 1256-1260.

22. Danilovic, N., et al., Enhancing the Alkaline Hydrogen Evolution Reaction Activity

through the Bifunctionality of Ni(OH)2/Metal Catalysts. Angewandte Chemie

International Edition, 2012. 51(50): p. 12495-12498.

27

23. Strmcnik, D., et al., Improving the hydrogen oxidation reaction rate by promotion of

hydroxyl adsorption. Nat Chem, 2013. 5(4): p. 300-306.

24. Subbaraman, R., et al., Trends in activity for the water electrolyser reactions on 3d

M(Ni,Co,Fe,Mn) hydr(oxy)oxide catalysts. Nat Mater, 2012. 11(6): p. 550-557.

25. Rheinländer, P.J., et al., Kinetics of the Hydrogen Oxidation/Evolution Reaction on

Polycrystalline Platinum in Alkaline Electrolyte Reaction Order with Respect to

Hydrogen Pressure. Journal of The Electrochemical Society, 2014. 161(14): p. F1448-

F1457.

26. Wang, Y., et al., Pt–Ru catalyzed hydrogen oxidation in alkaline media: oxophilic effect

or electronic effect? Energy & Environmental Science, 2015. 8(1): p. 177-181.

27. Sheng, W., et al., Correlating hydrogen oxidation and evolution activity on platinum at

different pH with measured hydrogen binding energy. Nature communications, 2015. 6.

28. Merle, G., M. Wessling, and K. Nijmeijer, Anion exchange membranes for alkaline fuel

cells: A review. Journal of Membrane Science, 2011. 377(1–2): p. 1-35.

29. Varcoe, J.R. and R.C. Slade, Prospects for Alkaline Anion‐Exchange Membranes in Low

Temperature Fuel Cells. Fuel cells, 2005. 5(2): p. 187-200.

30. Varcoe, J.R., et al., Anion-exchange membranes in electrochemical energy systems.

Energy & Environmental Science, 2014. 7(10): p. 3135-3191.

31. Guerrini, E. and S. Trasatti, Electrocatalysis in water electrolysis. Catalysis for

Sustainable Energy Production, 2009: p. 235-269.

32. Lyons, M.E.G. and M.P. Brandon, Redox switching and oxygen evolution electrocatalysis

in polymeric iron oxyhydroxide films. Physical Chemistry Chemical Physics, 2009.

11(13): p. 2203-2217.

33. Rebouillat, S., et al., Paving the way to the integration of smart nanostructures: part II:

nanostructured microdispersed hydrated metal oxides for electrochemical energy

conversion and storage applications. International Journal of Electrochemical Science,

2011. 6(11).

28

Chapter 2 Composite Metal/Metal-Oxides for Hydrogen Reactions at High pH

2.1 Introduction:

Many researchers have envisioned “hydrogen highways” to supply hydrogen to a

growing market of fuel-cell powered vehicles. A key challenge to the implementation of the

hydrogen highways is the immense cost of building an infrastructure for the delivery and storage

of hydrogen gas. However, the use of modular “plug & play” water electrolyzers significantly

reduces the cost of this infrastructure because it allows market-driven deployment of individual

hydrogen fuelling stations, wherever there is access to water and electricity. Although the cost of

state-of-the-art water electrolyzers is significant, the high price-tag is largely due to the use of the

acidic proton exchange membrane (PEM)-based system which requires rare and prohibitively

expensive platinum group metal (PGM) electrocatalysts. The requirement of PGM catalysts is a

significant factor preventing the commercialization of this technology. This is especially relevant

considering the order of magnitude higher loading of noble metals typically used in electrolyzers

in comparison to concomitant application in fuel cells.[1, 2] Alternative methods of hydrogen

production such as high-temperature reforming of natural gas and methane produce very “dirty”

reformate gasses, which require complicated clean-up steps and reformate tolerant PGM

catalysts in fuel cells to avoid catalyst deactivation from CO poisoning.[3-5] The development of

alkaline electrolyzers using anion-exchange membranes (AEMs) circumvents the stability

criterion restriction of the PEM systems and opens a pathway to utilize inexpensive and abundant

transition metal (TM) catalysts to produce pure H2(g) via the direct electrochemical reduction of

water. Although the inherent catalytic activity of TM electrocatalysts is much lower than PGM

catalysts – the development of composite metal/metal-oxide interfaces may hold the key to

unlocking the potential of PGM-free electrocatalysis.

29

Ni-Mo alloys have long been considered the most active non-PGM electrocatalysts for

alkaline hydrogen evolution[6-8] – however, the inherent oxophilicity of the Mo component

makes this material quite pyrophoric[7] and significantly complicates handling of the metallic

alloy for use in any commercial device. Furthermore, a careful analysis of the Mo Pourbaix

diagram indicates that metallic Mo exists only at potentials well negative of the HER on-set. In

fact, Mo has a small pH window where a passivation layer will form (near 0 V vs. RHE between

pH 4-8).[9] For pH values beyond this passivation window, Mo tends to leach out of the

metallic state, forming the MoO42-

oxyanion. Thus, while Ni-Mo electrodes have exhibited

promising performance in PGM-free AEM electrolyzers,[10] the long-term stability of Ni-Mo

electrodes in this environment has not yet been demonstrated. In contrast, Ni-Mo films on Ni-

mesh electrodes have shown long-term stability in an industrial application[11] – however the

question of start-up & shut-down stability remains a significant issue for modular electrolysis

systems where transient potential spikes could cause oxidation and subsequent dissolution of the

Mo component of Ni-Mo alloys.

Although operation in an alkaline environment circumvents the stability criterion of PEM

systems, one significant challenge of alkaline systems is the decreased hydrogen reaction

kinetics.[12, 13] In PEM systems, the direct involvement of highly mobile hydrated protons

makes both the hydrogen oxidation and evolution reactions extremely facile and the hydrogen

electrode in fuel cell and electrolysis devices requires negligible over-potential to achieve

operational current densities. In the absence of strongly adsorbing anions (anion “poisoning”) the

hydrogen reactions on Pt in acid are essentially non-polarizable (hence the use of the ubiquitous

Pt/H+/H2 reference electrode for many electrochemical applications). However, in alkaline

environment, the source of protons for hydrogen evolution is no longer H3O+ - but simply H2O.

30

Thus the initial absorption of a proton to produce the Hads HER intermediate requires much

larger activation energy to strip a strongly bound proton from water than the Ea to remove H+

from H3O+.[13, 14] This phenomenon manifests in increased over-potential for the hydrogen

evolution reaction (HER) required to achieve operational current densities with PGM

electrocatalysts in alkaline electrolyzers compared to the PEM MEAs. Figure 1 shows identical

PGM electrodes tested in PEM vs. AEM MEAs and the resulting increase in cell overpotential

for the AEM-system compared to the PEM-system, most of the contribution being a result of

higher overpotential for HER.

Figure 1 – Electrolysis cell polarization curves: Pt (HER cathode) & Ir (OER anode)

electrodes provided by Proton On-Site, PEM MEA used Nafion membrane; AEM MEA used

Tokuyama A-201 membrane. Fuel was DI water at a flow rate of 100mL/min. Steady-state

measurements were collected via a galvanostatic polarization test. Data points were recorded

after the measured potential remained stable for at least 1 min.

A further complication of alkaline electrocatalysis results from the interfacial issues

unique to the use of an anion-exchange ionomer (AEI) in contrast to the proton-exchange

ionomer (most commonly, solubilized Nafion) used in PEM systems. The solubilized ionomer is

31

intimately mixed with the catalyst layer (CL) during fabrication of the electrode. The presence of

the ionomer helps to gradually extend the electrified interface deeper into the CL, thus increasing

ionic conductivity and minimizing mass-transport limitations related to diffusion of the ionic

species. In the case of Nafion – the SO3- moiety has been shown to interact with the catalyst,

adsorbing onto the Pt surface,[15] but the effects of this adsorption on electrocatalyst

performance are negligible – particularly in the HER region where the negative charge on the

electrode surface electrostatically repels sulfonate anions. The analogous ion-exchange group

for OH- transport in AEMs is typically the quaternary ammonium (QA) functional group.

Although AEMs with QA functional groups show an optimized balance of stability (chemical &

mechanical) and hydroxide conductivity compared to other anion exchange moieties,[16-18] the

ammonium functionalities have recently been shown to poison Pt surfaces via electronic and/or

covalent interaction and inhibit electrocatalysis.[19] However, careful analysis of interfacial

phenomena suggest that the anion-exchange functional groups in the ionomer do not result in

loss of Hupd-derived electro-catalytic surface area (ECSA) as is typical of adsorbate “poisons”.

Instead, the AEI appears to affect the electrochemical properties of the interface, thus inhibiting

formation of HER intermediates and decreasing HER kinetics on Pt surfaces. In contrast, the

data shows that a Ni-Cr/C catalyst appears more resistant to inhibition by QA.

In this study we have screened a wide array of binary and ternary Ni-alloys and

composite Ni/MOx/C (MOx = transition metal oxide) samples and have identified a composite Ni-

NiOx-Cr2O3 material as a front-line catalyst with performance rivaling that of state-of-the-art Ni-

Mo. Furthermore, careful evaluation of the electrochemical response of benchmark Pt catalysts

in the presence of acidic vs. alkaline ionomer interfaces has provided a molecular level

perspective of the interface. This work supports previous studies indicating synergistic

32

enhancement of HER & HOR electrocatalysis on composite M/MOx surfaces and provides

guiding principles for the development of non-PGM electrified interfaces for energy conversion

and storage technology.

2.2 Experimental:

For HOR studies, various Pt/C and Pt-alloy/C electrodes were tested in acid and alkaline

electrolytes. The commercial catalysts used to evaluate alkaline HER-HOR kinetics were: 20%

Pt/C (Johnson-Matthey HiSPEC-1000, Alfa Aesar, Ward Hill, MA), 46% Pt/C (TEC 10E50E,

TKK, Tokyo, Japan), 20% Pt-Ru/C (E-tek), 20% Pt-Ir/C (Premetek, Wilmington, DE), Pt black

(provided by Proton On-Site, Wallingford, CT), HSA Pt/C and HSA Pt-Co/C were provided by

Automotive Fuel Cell Corporation (AFCC, Burnaby, BC, Canada). The Pt-Nb/C sample was

synthesized via co-precipitation of Pt and Nb oxides on Vulcan carbon (Cabot Corporation,

Haverhill, MA), followed by annealing in a tube furnace under argon. The 0.1M HClO4

electrolyte was diluted from 70% HClO4 (Veritas, double-distilled, GFS chemicals) using ultra-

pure H2O (18.2 MΩ, Millipore). Alkaline (0.1 M KOH) electrolyte was prepared using

potassium hydroxide pellets (semiconductor grade 99.99%, Sigma-Aldrich) and ultra-pure H2O.

For HER studies, various binary and ternary Ni-alloy electrocatalysts were synthesized

by a standard impregnation method using sodium borohydride reducing agent (Sigma-Aldrich).

Briefly, metal chloride or nitrate salts (Reagent Grade, Sigma-Aldrich) were dissolved in H2O

(18.2 MΩ Millipore), followed by addition of the carbon black support (Ketjen Black-EC600JD,

Akzo Nobel Polymer Chemicals) and the reaction solution was stirred for at least 1 hour. NaBH4

was then added drop-wise to chemically reduce the metal salts onto the carbon support. Upon

addition of NaBH4, bubbling was observed and the solution changed color from a greyish

suspension of carbon to dark black, indicating formation of nanoparticles. The solution was

33

stirred for at least 1 hour after addition of NaHB4 to ensure complete reaction of all metals. The

products were then vacuum-filtered using a Büchner funnel and filter paper and washed with DI

H2O. The solid product was dried in a vacuum oven overnight then heated in a tube furnace at

500°C for 6 hours under 5%H2/95%Ar forming gas to anneal the metal alloys. For comparison,

the Ni/MOx/C samples were prepared by precipitation of MOx onto the carbon support, followed

by NaBH4 chemical reduction of Ni onto the MOx/C composite support and then subjected to the

same sample work-up (filtering, heating, etc.). The MOx precipitation was achieved by adding

1M Na2CO3 (Sigma-Aldrich) drop-wise until the solution changed to a grey or milky color and

the solution pH was at least 10. For HER studies the standard Pt catalyst was Pt black supplied

by Proton On-site (Wallingford, CT). Other Ni-, Co- & Fe-oxide nanoparticle catalysts

referenced in the figures were purchased from QuantumSphere (Santa Anna, CA).

RDE analysis was conducted using inks composed of 5 mL H2O, 4.95 mL 2-

propanol and 50 μL of 5 wt% ionomer dispersion, mixed with an appropriate amount of catalyst.

The ionomer dispersions used were either “Nafion” (perfluorosulfonic acid-PTFE co-polymer,

Alfa Aesar, Ward Hill, MA) or Tokuyama AS-4 (Tokuyama Co, Japan). The inks were sonicated

for at least 30 min before a 10 μL aliquot was deposited on the tip of a polished glassy carbon

disk (5.61 mm diameter) to produce a loading of 50 μg(metal)/cm2. The catalyst layers were

spin-coated on the RDE tip using an inverted Pine Instruments rotator to ensure uniform

distribution of the catalyst. For examination of the ionomer effects on unsupported Pt black, the

CL contained 250 μg(Pt)/cm2 with a 5:1:1 mass ratio of catalyst:acetylene black

(Chevron):ionomer (i.e. 250 μL 5 wt% ionomer dispersion in ink). Electrochemical tests were

conducted with an Autolab (Ecochemie Inc., model-PGSTAT30) potentiostat/galvanostat. Tests

were conducted in a 100 mL jacketed 3 electrode cell using a water circulator (Neslab Exacal

34

EX-300) to maintain 50°C. Alkaline (0.1M KOH) electrolyte was prepared using potassium

hydroxide pellets (semiconductor grade 99.99%, Sigma-Aldrich). For each test, a freshly made

RHE was used as the reference electrode and a gold flag counter electrode was used to avoid Pt

contamination. The glassy carbon working electrode was rotated using an RDE setup from Pine

Instruments. Rotation rates of 2500 rpm were sufficient to remove the H2(g) product from the

surface of the working electrode and examine the kinetics well into the HER region. All results

were obtained after conditioning electrodes by performing cyclic voltammetry (CV) scans at 50

mV/sec from 0-1 V for at least 20-30 scans, or until stable features were observed.

XRD characterization was conducted using a Rigaku Ultima IV XRD with a Cu Kα

source (λ=1.541 Å) operated at 40 kV and 44 mA. 2θ/θ scans were conducted using a 0.05° step

size and 5 sec hold per step.

SEM characterization was conducted using a Hitachi S-4800 FE-SEM. EDS data was

collected using EDAX Genesis on the SEM to validate sample elemental composition. HR-TEM

images were collected on a JEOL 2010F TEM at 200kV.

XAS measurements were conducted at the X3B beamline of Brookhaven National Labs

and analysis was performed using the IFEFFIT suite. The in-situ XAS studies at the Ni & Cr K-

edges (8331eV & 5989eV, respectively) were performed at X3B beamline of National

Synchrotron Light Source (NSLS, Brookhaven National Laboratory, NY). Information on the in-

situ spectro-electrochemical cell can be found in the literature.[20] Working electrodes (1 cm x 3

cm) were prepared on carbon cloth GDL by brush-painting a catalyst ink containing 4:1 2-

propanol:DI water solvent and 100μL of 5wt% AEI binder. The metal loading was chosen to

give 0.1 transmission spectra edge heights at the Ni and Cr K edges. Prior to XAS analysis, the

GDEs were wetted with the 0.1M KOH alkaline electrolyte via vacuum impregnation. Nitrogen-

35

saturated 0.1M KOH was used as the electrolyte. A second carbon cloth GDL of larger surface

area was used as the counter electrode. A reversible hydrogen electrode (RHE) was used as a

reference and placed in the electrolyte reservoir and connected to the cell by a salt bridge

consisting of silicone tubing with a Vycor tip. Prior to the start of each experiment, the working

electrode was activated by potential cycling between 50 mV and 1 V for 3 cycles at 20 mV/sec,

the CV scan was followed by chronoamperometry to hold the working electrode at the desired

potential during the collection of XAS data. Before each measurement, the potential was held for

5 minutes to reach a steady state. Spectra were collected in fluorescence mode using a 32

element Ge detector. All of the experimental data was collected in conjunction with the

appropriate reference foils to aid in energy alignment and normalization.

XAFS Analysis: The data were processed and fitted using the Ifeffit-based Athena[21]

and Artemis[22] programs. Scans were calibrated, aligned and normalized with background

removed using the IFEFFIT suite.[23] The χ(R) were modeled using single scattering paths

calculated by FEFF6.[24]

2.3 Results and Discussions:

The key challenges to efficient operation of alkaline electrolyzers and fuel cells are

predominantly attributable to the hydrogen electrode reactions. While some researchers have

claimed increased kinetics of the oxygen reduction reaction (ORR) in alkaline media,[25] the

observed facility of ORR on various electrodes in alkaline media has more recently been

explained in terms of stabilization of the ORR intermediates for outer-sphere electron transfer

(often favoring the undesired 2 e- ORR pathway).[26] For the oxygen evolution reaction (OER)

in alkaline electrolyte, the on-set potential and reaction kinetics appear to be favorable on many

metal oxide electrodes.[27, 28] The inexpensive & abundant metal oxides exhibit OER

36

performance which rivals PGM catalyst activity. In contrast, for the hydrogen electrode, non-

PGM catalysts require large activation over-potential before the on-set of the HER kinetic

region. Ni-based catalysts have been reported as fuel cell anodes for the hydrogen oxidation

reaction (HOR) in anion exchange membrane fuel cells (AEMFCs) and RDE studies but even

these rather dubious reports claim very low HOR activity on Ni-based catalysts.[29, 30]

Furthermore, HOR has never been observed on any of the Ni-based catalysts tested during the

course of this project.

2.3.1 – Decreased HER-HOR Kinetics in Alkaline vs. Acid Electrolyte

Figure 2 – Left: HOR kinetics on Pt/C (20 μgPt/cm2) in acid vs alkaline electrolytes. Right: HER

kinetics on Pt/C (50 μgPt/cm2). Electrolytes were purged with H2 for at least 30 min prior to

testing. Testing was conducted at room temp (~23°C) on polished glassy carbon RDE substrate.

Rotation rates of 2500 were used to achieve high HOR mass-transport limiting currents and to

remove H2 gas (produced during HER) from the working electrode surface. EIS measurements

indicated that HFR values (used to calc. iR-comp.) were ~25 Ω for both electrolytes.

High surface-area platinum is the benchmark catalyst for HOR/HER reactions, although

Ir & Ru have also exhibited reversible hydrogen activity in alkaline media.[31-33] As noted

above, Pt suffers from two major challenges in alkaline media. The first and most fundamental

challenge is the decreased HOR/HER kinetics in alkaline electrolyte. The second challenge

37

relates to the electrostatic effects inherent to the interaction of the anion exchange ionomer with

the catalyst surface. Figure 2 shows the decreased HOR & HER kinetics on a commercial Pt/C

standard. For the HOR, the mass-transport limiting current is reached by +50 mV in acid, but not

until +400 mV in base. The slow HOR kinetics translates into significant anode polarization

losses in fuel cell operation. For the HER, a somewhat higher catalyst loading is typically used to

evaluate conditions closer to the high loading of real-world electrolyzers. Figure 2 shows a much

higher Tafel slope for HER in base than in acid. For HER, the decrease in kinetics is related to

the involvement of H2O as the proton source as opposed to H3O+ in acidic media.[13] While for

HOR, the cause of the decreased kinetics in alkaline is still a controversial issue in current

literature. It is believed by some that the Pt-H bond strength dictates the HOR/HER activity.

Recent reports have argued that the positive shift in Hupd peak potential in base indicates stronger

Pt-H bond strength, thus inhibiting HER-HOR reactions. The typical CV response of a Pt/C

sample is displayed in Figure 3. Notice the Hupd charge occurs almost entirely < 300 mV in acid,

while the Hupd peaks are observed at more positive potentials in base. It has been argued that the

Hupd peak potential is a descriptor for Pt-H bond strength and the resulting HOR/HER

activity.[34, 35]

38

Figure 3 – CV data (20 mV/sec) showing Hupd charge from 46% Pt/C (50 μgPt/cm2) in argon-

purged electrolytes at room temp. (~23°C).

In contrast to the theory that Pt-H bond strength dictates HOR activity, we believe that

the decreased HOR kinetics in alkaline electrolyte are related to the necessity of using OH- as a

proton sink for oxidizing the Hads intermediate in the electrochemical Volmer reaction

mechanism. From a fundamental electrostatic perspective, under HOR conditions, the electrode

is negative of the potential of zero charge (pzc) and thus the negative charge on the electrode

surface would present a significant electrostatic barrier to transport of the OH- anion from the

bulk electrolyte through the double-layer to the electrode surface. While the role of OH- is

contested in the literature, preliminary studies in our lab have suggested that mixing the HOR-

active Pt with an additional oxophilic surface increases the HOR by providing OHads in the IHP

on the oxophilic surface site. For the binary catalysts, the pzc is not uniform across the entire

surface and thus the oxophilic surface site can form an OHads passivation layer directly from the

water solvent molecules present in the double-layer – thus avoiding the electrostatic barriers

inhibiting OH- transport through the double-layer. Preliminary testing results have indicated that

39

binary Pt-X catalysts where X is an oxophilic element (Ru, Ir or Nb) exhibit alkaline HOR/HER

kinetics similar to the “non-polarizable” acid case. These results are in agreement with the theory

of Markovic et al.[14, 32, 36, 37] which suggest that composite M/MOx surfaces show enhanced

kinetics for numerous reactions in alkaline electrolytes. However, the M/MOx theory has recently

been challenged by Gasteiger,[38] Zhuang[34] and Yan,[35] each of whom have argued that the

positive shift in Hupd peak potential with increasing pH indicates an increased Pt-H bond strength

and thus decreased HOR kinetics.

2.3.2 – Validation of the M/MOx Theory for Alkaline HOR:

Figure 4 shows the Hupd features (in argon-purged electrolyte) and HOR/HER kinetics (in

H2-purged electrolyte) for a Pt/C standard and three different Pt-X/C samples. For all three Pt-

alloys, the binary metal is highly oxophilic and all are known to form an OHads passivation layer

when in contact with alkaline electrolyte. It can be clearly seen from the Hupd features of the

various samples that the Pt-alloys each have very different Hupd peak potentials. Thus if the Pt-H

bond strength theory is correct, we would expect to observe drastically different HOR activity

from each of the Pt-alloy samples. Also, the significantly larger double-layer charge of the Pt-

Ir/C sample is due to the surface Ir, which is known to exhibit a much larger capacitance than

any of the other materials (Pt, Ru, Nb or C). Examination of the HER & HOR kinetics under H2-

purged conditions indicates that all of the Pt-alloy catalysts exhibit a similar and significant

increase in activity compared to the low kinetics on the Pt/C standard. The specific activity is

reported to normalize the kinetics to the amount of Pt surface, thus removing any differences in

particle size and ECSA of the samples. While this topic certainly warrants a more thorough

investigation, the results shown in figure 4 tend to refute the Pt-H bond strength theory and

40

support the M/MOx theory (i.e. electrostatic barrier / OH- transport) for explaining the decreased

HOR kinetics at high pH.

Figure 4 – Left: CV scans scans (50 mV/sec) showing Hupd charge on various Pt samples in

argon-purged 0.1M KOH at 40°C. Right: HER-HOR polarization curves from 10 mV/sec CV

data in H2-purged 0.1M KOH. HER-HOR data is iR & mass-transport corrected. “Specific

activity” is reported by normalizing the current density to the ECSA value (calculated for each

catalyst from their respective Hupd charge). Pt/C: Johnson-Matthey, Pt-Ru/C: E-tek, Pt-Nb/C:

NEU, Pt-Ir/C: Premetek. For both plots, Pt/C loading is 10 μgPt/cm2.

In addition to the results shown in Figure 4, another preliminary study was conducted to

evaluate the validity of the M/MOx theory for alkaline HOR. Figure 5 shows the Hupd features (in

argon-purged electrolyte) and HOR/HER kinetics (in H2-purged electrolyte) for a commercial Pt-

Co/C sample before and after CV cycling in an acid electrolyte. The catalyst was first evaluated

at the “beginning of life” (BOL) in 0.1M KOH under argon- and H2-purged conditions. The

electrode was then removed from the alkaline electrolyte, washed gently with DI water to avoid

delamination of the catalyst layer, and placed in an argon-purged 0.1M HClO4 electrolyte. The

electrode was cycled 50 times at 100 mV/sec from 0 – 1.2 V in the acid electrolyte, at which

point the CV features remained quite stable from scan to scan. Following the “acid cycling”, the

electrode was removed from the acid electrolyte, again gently washed with DI water and placed

in a fresh, 0.1M KOH electrolyte and tested under argon- and H2-purged conditions. This Pt-Co

41

catalyst is well-known to have Co on the surface which is readily removed by voltage cycling in

acid.[39] From the argon-purged CV data we can clearly see that after acid-cycling, the redox

features characteristic of surface Co at ~0.5 & 0.7 V are no longer present. In particular, this

redox process is commonly attributed to a Co2+/3+

transition, with the exact stoichiometry of the

Co-oxide or –hydroxide being dependent on the calcination pretreatment of the Co surface prior

to electrochemical analysis. However, for the initially metallic Co in this Pt-Co sample, we can

assume a Co(OH)2 <--> Co(OH)3 transition.

Figure 5 – Left: CV scans scans (50 mV/sec) showing Hupd charge on a Pt-Co/C sample from

AFCC in argon-purged 0.1M KOH at 40°C. Right: HER-HOR polarization curves from 10

mV/sec CV data in H2-purged 0.1M KOH. HER-HOR data is iR & mass-transport corrected.

“Specific activity” is reported by normalizing the current density to the ECSA value (calculated

for each catalyst from their respective Hupd charge).

The presence of these redox peaks at BOL emphasizes the M/MOx nature of the Pt-Co/C

sample. After acid-cycling we observe loss of the Co redox peaks and a significant decrease in

the HOR/HER activity – without a shift in the Hupd peak potential. These results further support

the M/MOx theory for alkaline HOR and invalidate the Pt-H bond strength theory.

42

Figure 6 – Pt particle size effects on HER-HOR activity. Left: CV scans (50 mV/sec)

showing Hupd charge on various Pt samples in argon-purged 0.1M KOH at 40°C. Right: HER-

HOR polarization curves from 10 mV/sec CV data. HER-HOR data is iR & mass-transport

corrected. In addition the current density is normalized to the ECSA value calculated for each

catalyst from their respective Hupd charge. Pt/C – 1: Johnson-Matthey, Pt/C – 2: TKK, Pt/C – 3:

AFCC, Pt black: Proton On-Site. For both plots, Pt/C loading is 10 μgPt/cm2 and Pt black loading

is 20 μg/cm2.

Table 1 – ECSA Values for Pt Samples, Calculated from Hupd Charge in 0.1M KOH

Sample ECSA (m2/g)

Pt/C - 1 47

Pt/C - 2 22

Pt/C - 3 15

Pt black 4

Finally, it should be noted that an activity trend related to particle-size was observed for

HOR/HER activity on pure Pt samples. Figure 6 shows the Hupd features and HOR/HER kinetics

of four different Pt catalysts. The Hupd data (results presented in Table 1) indicates drastically

different ECSA values from the four Pt samples. The HOR/HER kinetic data in Figure 6 shows

an inverse relationship between specific activity and ECSA – i.e. the larger particles with lower

surface-area exhibit better HOR/HER kinetics than the small particles with high surface-area.

This result is counter-intuitive given that this is a surface-specific reaction which undeniably

requires the formation of a surface-adsorbed intermediate. The cause of this size-dependent

activity is unclear. Markovic et al.[40] have reported trends in the specific activity on Pt single-

43

crystal electrodes, indicating that Pt (110) > Pt (100) > Pt (111), and they attributed the trends in

activity to the relative oxophilicity of the single-crystal surfaces, further supporting the M/MOx

theory. Thus, the larger particles may exhibit a predominance of Pt (110) surface facets. Given

the opportunity (time & funding), it would be prudent to conduct a more detailed investigation of

the particle-size effects and more thoroughly evaluate the ECSA of the Pt-alloy samples using

Hupd and CO-stripping in acid and alkaline electrolytes to remove any possible particle-size

effects from the comparison of HOR/HER specific activity on M/MOx catalysts.

2.3.3 - AEI Poisoning of Pt:

As discussed previously, the decrease in hydrogen kinetics in alkaline electrolyte is not

the only challenge to the development of AEM fuel cells and electrolyzers. The second challenge

relates to the electrostatic interactions of the catalyst surface with the quaternary ammonium

(QA) exchange group of the anion exchange ionomer (AEI). While Kohl et al.[19] have reported

the apparent poisoning effects of the QA functionality on Pt surfaces, we have taken a closer

look at the interfacial phenomenon to develop a detailed model of the Pt/AEI interface.

We begin our discussion by examining the electrochemical response of Pt surfaces at

potentials relevant to HER & HOR. When scanning to potentials below the redox potential of the

H2O/H2 couple (i.e., 0 V vs. RHE), the electrochemical response is larger than the Hupd pseudo-

capacitive charge. This anodic feature is typically called over-potential deposited hydrogen

(Hopd). The increase in anodic peak charge is shown for various Pt surfaces in Figure 7.

44

Figure 7 – Hupd & “Hopd” observed for various Pt electrodes in 0.1 M KOH. All electrolytes are

argon-purged, testing was conducted at room temp (25°C) and CV scan rate was 20 mV/sec. All

catalyst layers contain the standard 14 wt% Nafion binder (except for the Pt disk).

The CV plots in Figure 7 show that the Hopd phenomenon is not exclusive to bulk or

nano-particle surfaces, but is observed on all Pt surfaces. However, although the high surface

area Pt/C samples initially exhibit larger magnitude Hupd charge (from scans with 0 mV cathodic

limit), the bulk Pt disk and bulk-like Pt black samples exhibit a larger ratio of Hopd to Hupd

charge. It is presumed that the larger Pt black nanoparticles and the bulk Pt surfaces are

dominated by high-index Pt (111) facets, in contrast with the high surface-area Pt/C

nanoparticles which likely consist of a much higher fraction of low-index edge & terrace surface

sites – Pt (100) & Pt (110). Thus the phenomenon of significantly higher Hopd/Hupd charge ratio

45

on surfaces dominated by Pt (111) facets lends support to the concept of “ensemble

requirements” for the HER. The phrase “ensemble requirements” refers to the requirement of

adjacent active sites, both with surface adsorbed intermediates, to facilitate a catalytic reaction.

In this case, two “close-packed” Pt (111)-Hads intermediates would likely exhibit faster HER

kinetics than the Pt (100)-Hads intermediates, which are separated by a larger atomic distance due

to the increased d-spacing of adjacent (100) facets compared to Pt (111) facets. These ensemble

requirements will be discussed further in the context of Figure 8 – vide infra.

While the exact nature of this Hopd response is rather ambiguous, it is generally accepted

that the charge corresponds to phenomenon occurring on the surface and NOT the formation of

absorbed, sub-surface Pt-hydrides. Thus, the “Hopd charge” may represent pseudo-capacitance in

excess of monolayer coverage from the adsorbed intermediate in the HER, or it may be

interpreted as a superposition of Hupd pseudo-capacitive charge and faradaic current produced

from the oxidation of hydrogen formed during the cathodic scan below 0 V. In either case the

charge observed in the anodic scan may not provide a suitable metric to accurately evaluate the

ECSA of Pt samples.

46

Figure 8 – Diagnostic CV scans (50 mV/sec) showing Hupd (+50 mV cathodic limit) & Hopd (-

100 mV cathodic limit) on 50 μgPt/cm2 Pt/C (a.) and 250 μgPt/cm

2 unsupported Pt black (b.) in

presence of Nafion vs AS-4 ionomer binders (incorporated during catalyst ink formulation).

Scans collected in RDE cell with argon-purged 0.1 M KOH at 23°C and 0 rpm.

The poisoning effects of the anion-exchange functional groups on Pt were noted in

previous work on the methanol oxidation reaction (MOR) in alkaline electrolyte by Kohl et

al.,[19] which showed decreased hydrogen stripping charge on polycrystalline Pt electrode and

decreased MOR current densities on bulk Pt and Pt/C in the presence of free quaternary

ammonium species in the electrolyte as well as poly tetra-methyl ammonium hydroxide ionomer

deposited on the electrode surface. Here, we have examined the interfacial phenomena on Pt/C

and unsupported Pt black in the presence of an AEI binder as compared to Nafion binder during

47

RDE testing. Interestingly, Figure 8 shows a negligible change in Hupd charge in the presence of

AEI. The Hupd is defined as the charge observed when the electrode is scanned in the potential

range (vs. RHE) from just above the thermodynamic standard reduction potential (+50 mV) to

roughly the beginning of the double-layer region (+500 mV). The inhibitory effects of the AEI

are only observed upon cycling to cathodic potential limits beyond the on-set of the HER. This

effect is observed upon cycling to cathodic potential limits of 0 V or -50 mV, but is most

pronounced upon cycling to -100 mV (as shown in Figure 8a & b). Thus, the decreased “Hopd

charge” does not represent a decrease in ECSA from specific adsorption of the QA moieties in

the AEI, but rather a significant decrease in the HER activity in the presence of the AEI. In

addition, Figures 8 & 10 clearly show particle-size effects on AEI adsorption. This particle-size

effect is evident from the difference of the HER & Hopd response of Pt black vs. the supported

Pt/C as shown in Figures 8 & 10. Difference in the ensemble size of the Pt/C and Pt black

samples is manifest in the XRD analysis displayed in Figure 9. The XRD response indicates

distinctly different crystallite size of the two Pt samples (based on diffracting domains)

indicating as expected a much smaller crystallite size for the supported Pt/C.

48

Figure 9 – XRD of Pt black and 46% Pt/C benchmark HER catalysts obtained from Proton On-

Site and Tanaka, respectively. Analysis was conducted using Rigaku PDXL software. Pt phase

ID matched ICDD PDF2010 DB#: 01-087-0636. For Pt black, crystallite size calculated using

Williamson-Hall method = 6 nm. Also, BET surface area was analyzed using a QuantaChrome

Monosorb single-point instrument. BET SA = 35 m2/g. For 46% Pt/C calculated crystallite size

was 2-3 nm. BET SA measurements were not performed on Pt/C because the response from the

Pt component cannot be isolated from the carbon-support.

The greater diminution of the HER & Hopd response in the presence of the AEI observed

on Pt black compared to Pt/C suggests differences in the inhibitory characteristics of the AEI

with reference to specific adsorption of QA moieties. In the presence of the AEI compared to the

case with the Nafion binder, for scans with cathodic limits below 0 V vs. RHE (As shown in

Figure 2), the subsequent anodic peaks at ~300 & 400 mV are not changed as dramatically as the

anodic response between 0-300 mV. This may point to a preferential blocking of the high index

Pt (111) facets, typically attributed to the peak near 0 V.[40, 41] This analysis supports the

particle-size effect wherein Pt black with more bulk-like (hence predominance of (111) sites) is

more severely effected as compared to Pt/C. While the details of the Pt/AEI interaction could be

further elucidated with in-situ spectro-electrochemical studies, the effect of HER/HOR inhibition

49

on Pt surfaces in contact with AEI is clearly evident here. The decreased HER activity in the

presence of the AEI binder is also seen in the HER CV scans on Pt catalysts in Figure 10 where

the effect of the AEI on the well-dispersed Pt/C sample is smaller relative to the unsupported Pt

black which exhibits substantial inhibition of the HER – as indicated by a ~20mV shift in the

half-wave potential on Pt black compared to ~5 mV shift for Pt/C.

Figure 10 - CV scans (50 mV/sec) showing HER activity on Pt/C (50 μgPt/cm2) and unsupported

Pt black (250 μgPt/cm2) in the presence of Nafion vs. AS-4 ionomer binder. Scans collected in

RDE cell with argon-purged 0.1 M KOH at 23 °C and 2500 rpm.

Aside from the apparent preferential blocking of the Pt (111) facets, the Hupd vs. Hopd

results in Figure 8 warrant a further clarification of the interfacial electrostatic effects of the AEI.

In particular, Kohl et al.[19] surmised that the AEI likely resided in the inner Helmholtz plane

(IHP) and exhibited a poisoning effect to decrease the ECSA and thus the MOR activity. This

prior report evaluated the ECSA by scanning to potentials well below the on-set of HER, thus

observing the combined Hupd & Hopd. Furthermore, they did not make the distinction between

Hupd & Hopd and thus concluded that the decrease in observed charge was due to poisoning of the

50

catalyst ECSA via direct chemisorption of the AEI in the IHP. In this study, careful control of

potential scan range indicates that the AEI likely straddles the double-layer interface such that

some of the QA moieties reside in the IHP on Pt (111) surface sites while the remainder of the

QA moieties of the AEI reside in the OHP – not specifically adsorbed in the IHP. This residence

of the AEI (in both IHP and OHP) also can account for the MOR results, where the electrode

potential (φm) is positive of the potential of zero formal charge (PZFC) and thus the positively

charged quaternary ammonium species on the AEI would be repelled by the positively charged

electrode. It is more likely that the inhibitory effect of the AEI on MOR is related to the

effective potential at the OHP (φ2).

Figure 11 – Qualitative representation of electrostatic effects of AEI on φ1 & φ2. The condition

studied previously under MOR conditions[19] is represented on the left, while the condition

studied in this paper is represented on the right. In both situations the chemical potential of the

AEI (μAEI

) increases the effective potentials at the IHP (φ1) & the OHP (φ2).

In the case of MOR (φm > PZFC) or HER/HOR (φm < PZFC ), φ2 is more positive than it

would be in the absence of AEI (φ2AEI

> φ2). As shown in Figure 11, for the relatively

51

concentrated electrolyte used in this study, the Guoy-Chapman-Stern model of the double-layer

interface predicts that the compact portion of the double-layer between the metal and the OHP

will behave like a double-plate capacitor and exhibit a linear decrease in potential between the

metal and the OHP.[42] Beyond the OHP, the model predicts an exponential decrease in

potential from the OHP to the diffuse layer. Based on our analysis of Hupd vs. Hopd above, it is

likely that for MOR and HER/HOR conditions, the majority of the metal surface is still solvated

by water molecules in the IHP. The effects of the AEI for the two situations are both related to

the electrostatics exerted by the chemical potential of the AEI (μAEI

) on the local environment.

For the MOR, the AEI in the OHP creates a higher magnitude φ2 which inhibits transport of OH-

from the OHP to the IHP. For MOR, the OHads is required to facilitate the removal of the COads

by-product of the MOR as per the well-known Langmuir-Hinshelwood mechanism for MOR. In

contrast, the relative change in the effective potential at the IHP (φ1) dictates the rate constant for

the HER, as the reactant is the water solvent layer in the IHP. Thus, for the HER, the AEI exerts

a similar dampening of the applied potential (φm) such that water molecules in the IHP

experience a slightly less negative potential than they would in the absence of AEI (φ1AEI

> φ1 or,

in terms of magnitude of negative charge: |φ1AEI

| < |φ1|). Thus the decreased HER activity in the

presence of the AEI can be accounted for by both specific adsorption of the AEI on the high-

index Pt (111) facets evidenced by the particle-size effects (Figs. 2&5) as well as the electrostatic

effects (Fig.6) which results in the H2O HER reactants in the IHP experiencing a lower

magnitude φ1 in the presence of AEI than in the presence of Nafion.

52

2.3.4 - HER Electrochemistry on Ni catalysts:

Upon establishing benchmark HER performance with the Pt black catalyst, RDE testing

was used to evaluate the kinetics of commercial samples of unsupported TM nanoparticles and

numerous binary and ternary carbon-supported Ni-alloys. Investigations also included Ni

catalysts deposited on composite metal-oxide/C supports including TiO2/C, CeO2/C, ZrO2/C,

WO3/C and Mo-Ox/C. Figures 7-10 show the activity of many of the most promising Ni-alloy

and Ni/Metal-Oxide/C samples. The best HER performance was observed from a binary Ni-

Cr/C sample with a 1:1 Ni:Cr atomic ratio.

Figure 12 shows early work in the HER screening studies. In particular, the HER activity

of the bare carbon support is negligible and an on-set of slight HER cathodic current is not

observed until nearly -600 mV. It was also noted that unsupported Ni-oxide nanoparticles (from

Quantumsphere) consistently exhibited greater HER activity than Fe-oxide or Co-oxide (not

shown) nanoparticles. This observation is in agreement with fundamental studies by Markovic et

al.[37] In addition, during early screening studies, Ni-Zr materials exhibited increased HER

activity compared to Ni-Mo catalysts.

53

Figure 12 – Early RDE screening results: 50 µgmetal/cm

2, 0.1 M NaOH, 50°C, 3600 rpm, 50

mV/sec. Note: negligible HER activity from bare Vulcan carbon support and much higher HER

activity from C-supported Ni-alloys than unsupported metal oxides from QuantumSphere (QS).

Although the literature has indicated that Ni-Mo catalysts exhibit optimized HER electro-

catalytic activity (in theory due to optimized H-bond strength), this was not observed during the

course of our testing. Figure 13 shows the HER activity of a wide array of binary and ternary Ni-

alloys. It can be seen that Ni-Mo samples (red) and Ni-Co samples (purple) exhibit rather low

HER activity with on-set potentials near -400 mV. In contrast, Ni-Fe and Ni-Cr samples exhibit

improved HER activity with on-set potentials of -300 mV. A ternary Ni-Zr-Mo catalyst exhibited

the best HER activity during early screening studies. The ternary Ni-Zr-Mo sample exhibited an

on-set potential of -200 mV – a significant increase in performance over Ni-Mo and Ni-Co

samples.

54

Figure 13 - RDE screening of binary & ternary Ni-alloys: 50 µgmetal/cm2, 0.1 M NaOH,

50°C, 3600 rpm, 50 mV/sec. Note: relatively low HER activity from most alloys. Best

alloys contain Cr or Zr (which were later determined to be oxides, not metallic alloys).

During the course of this study, the research facilities provided limited access to rapid

elemental analysis and XRD analysis could not confirm or quantify the presence of the Mo

component in Ni-Mo catalysts. Thus, the apparent low activity of the Ni-Mo samples may have

been due to deficiencies in the synthesis methods. Ideally, rapid X-ray fluorescence (XRF)

analysis could have provided validation of elemental composition of these catalyst samples, but

as XRF was not readily available, the nominal chemical composition was presumed to be the

accurate composition of the final synthesis products. It was not until much later in the project,

that EDS (essentially XRF from an SEM instrument) analysis became available to validate

sample elemental composition. The EDS results indicated that in general, the Mo content of Ni-

Mo samples was less than the expected nominal values. This low Mo yield is likely due to the

slow kinetics of Mo chemical reduction compared to Ni chemical reduction (i.e. differences in

55

kinetics do not allow for co-reduction of Ni & Mo metal salts). In addition, Mo is known to form

non-reducible MoO42-

oxy-anion under the elevated pH conditions caused by the use of NaBH4

which reacts to form an alkaline borate solution. Thus the apparent low HER activity of Ni-Mo

samples prepared early in the study may be due to low Mo-content and thus did not benefit from

the optimum M-H bond strength predicted in the literature.

The high HER activity of the ternary Ni-Zr-Mo sample let to a very careful analysis of

the XRD pattern of this sample. It was initially assumed that the Ni-alloy samples were reduced

to the bare metals during furnace heating under the reducing hydrogen atmosphere; however the

XRD results from Ni-Zr and Ni-Zr-Mo samples indicated the presence of a nearly amorphous

ZrO2 phase. This led to the development of an array of Ni/metal-oxide/C samples to evaluate the

synergistic enhancement of HER on these composite metal / metal-oxide (M/MOx) surfaces.

Figure 14 shows the HER activity of the M/MOx samples from CV testing. The HER

enhancement resulting from composite M/MOx surfaces is apparent from the -200 mV on-set

potential observed for all samples. Furthermore, the Ni/Cr-Ox/C sample exhibited the best HER

activity.

56

Figure 14 - RDE screening of Ni/Metal-Oxide/C composites: 50 µgmetal/cm2, 0.1 M

KOH, 50°C, 3600 rpm, 50 mV/sec. Note: all composite M/MOx samples exhibit higher

activity than best Ni-alloy sample.

Due to the high activity of the Ni/Cr-Ox/C sample as well as a previous Cr-decorated-Ni

sample evaluated early in the project, a binary Ni-Cr/C catalyst was synthesized and compared to

the M/MOx samples. Figure 15 shows the steady-state HER activity of the M/MOx samples

(which all out-perform the ternary Ni-Zr-Mo) compared to the new binary Ni-Cr/C sample,

which exhibits the best HER activity. Furthermore, the Ni-Cr/C catalyst achieved the target

performance of 5 mA/cm2 at -200 mV. This target was set in an effort to achieve 500 mA/cm

2

between 1.8 – 2.0 V in full-cell testing.

57

Figure 15 - RDE screening: steady-state chrono-amperometry data collected after 3 min at each

50 mV potential step (50 µgmetal/cm2, 0.1 M KOH, 50°C, 3600 rpm). Note: all materials show

nearly identical HER activity, however a co-deposited Ni-Cr/C catalyst exhibits the best HER

activity.

During the course of this project it was discovered that the target was based on the flawed

assumption that the mass-activity of the catalyst would scale from the RDE loading of 0.05

mg/cm2 to the target loading of 5 mg/cm

2 on gas-diffusion electrodes (GDEs) in the full-cell

testing. It should be noted that subsequent full-cell testing at Proton On-Site and at NUCRET

proved quite challenging. From a practical perspective, attempts to achieve 5 mg/cm2 catalyst

loading on GDEs in full-cell testing resulting in very thick catalyst layers which necessitated re-

optimization of cell gaskets, the result of which was that the optimum compression ratio of the

full membrane electrode assembly (MEA) was not maintained after changing multiple

component parameters (GDE & gasket thickness). The best full-cell testing only achieved 100

mA/cm2 below 2 V and a polarization of 2.5 V was required to achieve the operational target of

500 mA/cm2 using Ni-Cr/C cathode catalyst and a PGM anode catalyst (Ir black from Proton

58

On-Site). However, although the full-cell performance metrics were not achieved using this

particular catalyst, the inherent HER activity of the Ni-Cr/C material is still of fundamental

interest because it exhibits the highest mass-activity reported to date. Elucidating the mechanism

of HER enhancement should also allow for the optimization of Ni-Cr composition, which can

then be applied to various catalyst morphologies (i.e. Raney-Ni-Cr, Ni-Cr on carbon nanotubes,

epitaxially grown Ni-Cr nano-wire arrays, etc.) to increase the active-site utilization at mass-

loadings relevant to operation in full-cell GDEs. Figure 16 shows the morphology of the Ni-Cr/C

catalyst and the best Ni-Mo sample. The Ni-Cr catalyst exhibits nano-film morphology on the

carbon support, while the Ni-Mo catalyst has formed discrete nano particles.

Figure 16 – SEM images of 30% Ni-Cr/C (left) and 20% Ni / 50% Mo-Ox /C (right) catalysts.

59

Figure 17 – Steady-state HER Chronoamperometry: 50 mV step size with 60 sec hold-time/step

in 0.1 M KOH (Ar-purged) at 50°C and 2500 rpm. Catalyst loading is 50 µg(metal)/cm2 with 15

wt% AS-4 used as binder in catalyst layer.

Figure 17 shows the steady state HER performance of a Ni-Mo/C sample compared to the

Ni-Cr/C and the Pt black benchmark. The performance of the Ni-Mo/C sample is typical for Ni-

based catalysts in that it requires an overpotential (η) of 300-400 mV before the HER kinetic

region is observed. This large overpotential has been attributed to a potential-dependent

mechanism. The HER proceeds via three fundamental steps:

Volmer: H2O + e- → Hads + OH

- (Electrochemical)

Heyrovsky: Hads + H2O + e- → H2 + OH

- (Electrochemical)

Tafel: 2Hads → H2 (Chemical)

Previous literature has proposed that at low η the formation of the Hads intermediate is the

rds and the reaction proceeds through a Volmer (rds)-Tafel mechanism, while at sufficiently high

η, the surface is saturated by Hads and the reaction proceeds via a Volmer-Heyrovsky (rds)

mechanism.[6, 8, 43] However, a standard analysis of the kinetic mechanism from the Tafel

60

slope is not possible because we cannot accurately quantify the active surface area of non-PGM

catalysts. In addition, Tafel slope analysis is best conducted in the case of a planar electrode

surface. But it is likely that this potential-dependent mechanism holds true for our nanoparticle

systems – especially at the somewhat elevated temperature of the test conditions (50°C) where

the reverse of the Volmer reaction (desorption of Hads) likely has a higher rate constant than the

forward Heyrovsky reaction step – until sufficient η is applied such that saturation of Hads is

achieved.

While Ni does exhibit the most promising HER activity of any of the 3d TMs, it also

suffers from deactivation via Ni-hydride formation.[44] Literature has shown that alloying Ni

with other 3d TM elements helps to prevent the formation of Ni-hydride and increases the

durability of the electrode.[45] Furthermore, Zhuang et al.,[30] have theorized that decorating

the Ni surface with Cr-oxide alters the electronic density of states of the Ni d-band in such a way

to decrease the Ni-O bonding while retaining the Ni-H bond affinity. In addition, Jaksic has

written extensively on the volcano plots for hydrogen binding energy and HER activity of

numerous PGM & TM catalysts.16, 19

While this strategy of alloying Ni with other TMs has

shown HER activity and inferred HOR activity on Ni-alloys, the literature generally shows large

activation overpotential before the on-set of the kinetic region of hydrogen electrocatalysis.

Markovic et al.,[14, 32, 36, 37] have recently presented a new strategy to overcome the slow

hydrogen kinetics in alkaline media via tailored metal/metal-oxide (M/MOx) interfaces. They

have shown significant increases in HOR, ORR & CO-oxidation performance by decorating Pt &

Ni surfaces with TM-oxides. This strategy is similar to the spillover effects presented by Jaksic

et al.,[46, 47] for enhancing methanol, H2/CO and oxygen reaction kinetics on catalysts which

they describe as having “interactive hyper-hypo-d-electronic bonding”. For these composite

61

metal/metal-oxide materials, the formation of Pt-OHads species is substantially increased on

catalysts supported on hydrated metal-oxides (Pt/Ta2O5-TiO2/C or Pt/Nb2O5-TiO2/C). In this

configuration, OHads spills over onto the Pt sites to increase the turn over frequency (TOF) for

CO-stripping, methanol oxidation or ORR. We believe that, similar to the spillover phenomenon,

adjacent M/MOx sites afford catalytic synergy to increase the effective TOF of the HER. As

described by Markovic et al.,[14, 32, 36] the oxide site has an affinity to form OHads, thus

weakening the H-OH bond in the HER reactant to allow for the formation of the Hads HER

intermediate on the metallic site. The large Ea of formation of the Hads intermediate (Volmer

reaction) is likely the cause of the lower hydrogen kinetics in alkaline vs acid media.

(1) H3O+ + e

- → Hads + H2O (Acid)

(2) H2O + e- → Hads + OH

- (Alkaline)

The activation energy of reaction (1) is presumably much lower than for reaction (2).

Although DFT studies have evaluated the energy of HER reactions in acidic media, the literature

appears devoid of any computational studies comparing the Ea of reactions (1) & (2) – likely due

to the complications in accurately modeling the solvent & pH effects when attempting to

quantify reaction energies in heterogeneous catalysis.[48] The large Ea of (2) can be lowered by

weakening the H-OH bond. This can be achieved by carefully tailored surface sites with

adjacent M/MOx components. Catalytic surfaces with nano-scale heterogeneity could offer a high

density of adjacent M/MOx sites where the M site (metallic Ni0) has an affinity for H-bonding

and the MOx (NiOx) has a high affinity for OHads formation.

62

Figure 18 – CV showing HER kinetics of Pt (black), Ni-Cr/C (orange), 60% Ni/C (blue) and

60% Cr-Ox/C (red) catalysts in 0.1 M KOH (Ar-purged) at 50 °C and 2500 rpm with Nafion vs.

AS-4 binder tested in RDE cell. Ni-Cr/C (orange) is the only sample which exhibits increased

HER activity in presence of AEI. All samples were prepared with 50 μg/cm2 total metal loading.

As shown in the CV results in Figure 18, either the M or MOx surfaces alone requires a

large η to achieve kinetic-controlled HER behavior, while the composite M/MOx surface exhibits

catalytic synergy for the HER and the MOx increases the HER TOF on the metal site. Figure 18

shows the HER CV results for Ni/C, Cr-Ox/C and Ni-Cr/C compared to the Pt black benchmark

(in the presence of Nafion vs. AS-4 AEI). We observe that the Ni/C catalyst suffers from a large

η before the kinetic region is established. The Ni/C sample is representative of a bare M surface

(without MOx). It is well-known that metallic nickel electrodes are covered with a passivating

surface oxide layer – NiO or Ni(OH)2 – when in contact with alkaline electrolyte. But this

passivating surface oxide is electrochemically reduced at potentials below 0 V (vs. RHE).[49] In

fact, it is this passivating oxide layer that inhibits HOR on Ni electrodes. Thus, at potentials

positive of the E0 for HER, the Ni/C catalyst may actually be representative of the archetypal

63

MOx surface, but upon applying a sufficient cathodic η to achieve HER kinetic-controlled

behaviour the Ni/C sample is representative of the bare M surface. The Cr-Ox/C sample exhibits

negligible HER activity. This provides a base-line to show that the excellent HER activity on Ni-

Cr/C is not simply the sum of HER from Ni and Cr-Ox components. In contrast to Ni/C and Cr-

Ox/C the Ni-Cr/C sample achieves kinetic-controlled HER activity at very low η. Furthermore,

Ni-Cr/C is in fact representative of a composite M/MOx interface as shown by XRD & in-situ

XAS analysis (vide infra). Thus, the results of Figure 18 support Markovic’s theory of enhanced

HER on composite M/MOx surfaces.

2.3.5 – Interfacial studies on Ni catalysts:

In addition to the unprecedented mass activity, the Ni-Cr/C sample is the only sample

which exhibits increased HER performance in the presence of the AEI binder, when compared to

the HER activity in the presence of the Nafion binder. This apparent resistance to AEI poisoning

is evident from the CV results in Figure 18. Based on the mechanistic models proposed by Lasia

et al.[6, 8, 50, 51] this indicates that the AEI somehow inhibits formation of Hads (Volmer

reaction) on bare metals such as Ni or Pt, while on the composite Ni-Cr/C catalyst, the AEI

appears to promote the formation of Hads. Based on previous interfacial studies by Kohl et

al.,[19] and the above discussion related to Figure 11, we can infer that the AEI exerts an

electrostatic dampening of the negative charge on the electrode (φ1AEI

> φ1 or, in terms of

magnitude of negative charge: |φ1AEI

| < |φ1|). However, the increase in HER activity on the

composite M/MOx interface of the Ni-Cr/C catalyst indicates that the AEI effects are not simply

due to a dampening of the electrode charge. While the exact nature of the Ni-Cr/AEI interaction

cannot be confirmed without further investigation such as using in situ spectro-electrochemical

64

studies, we theorize that the AEI may affect the water dipole at the M/MOx interface. Figure 19

details how the orientation of H2O can facilitate formation of OHads on MOx sites (thus

facilitating the alkaline HER rds: cleavage of the H-OH bond) or inhibit formation of Hads on

bare metal sites. Quadrant 1 (Q1) in the lower left shows the standard orientation of H2O at the

bare metal interface. In the standard orientation, the water dipole is oriented with the positively

polarized hydrogen atoms pointed towards the negatively charged electrode. Q2 shows the

M/MOx interface without AEI, where the orientation of H2O makes it difficult to form OHads on

the MOx site. We propose that the chemical potential of the AEI (μAEI

) in the OHP decreases the

magnitude of the electric field exerted on H2O at the IHP (φ1) thus allowing a greater fraction of

the water solvent/reactant molecules in the IHP to orient their dipoles such that the hydrogen

points towards the AEI while the oxygen points towards the M/MOx interface – thus facilitating

formation of OHads on the MOx site (Q3). The presence of MOx in close concert with bare metal

surface benefits from these uniquely oriented water molecules, wherein the orientation of oxygen

close to MOx provides for easy replenishment of the oxy-hydroxides. Closely juxtaposed, bare

metals on the other hand have lower affinity to form hydrides in the context of the same

orientation (Q4). Future studies involving the development of M/MOx catalysts for alkaline

electrochemistry should be cognizant of the effects of ionomer on the interfacial electrostatics

and solvent dipole orientation.

65

Figure 19 – Proposed interface model for alkaline HER on pure metal and M/MOx surfaces. AEI

represented by archetypical quaternary ammonia polysulfone structure in OH-exchanged form.

Metal represented by fcc unit cell model and metal oxide represented by NiO unit cell model.

As noted above, Ni-Mo/C samples synthesized in the early stages of this project may

have contained Mo contents lower than the nominal values and therefore may not have exhibited

the properties characteristic of Ni-Mo samples reported in the literature. For a more thorough

comparison of Ni-Cr to Ni-Mo samples, Ni-Mo nanoparticles were synthesized as per recent

literature by Lewis et al.[7] The synthesis product was characterized using SEM as shown in

Figure 20 and the results appear very similar to the SEM images reported in the literature. In the

66

literature these unsupported Ni-Mo nanoparticles are tested for HER activity using much higher

loading than typical for examining fundamental electrocatalytic kinetics in RDE studies. Lewis

et al. evaluated the Ni-Mo catalyst at 1 mg/cm2 on Ti-foil substrate using a home-made

electrode. While this approach proved rather difficult to simulate, we began by testing the Ni-

Cr/C and Ni-Mo samples at 1 mg/cm2 loading on the RDE substrate.

Figure 20 – SEM image of Ni-Mo nanoparticles synthesized as per Lewis et al.[7]

67

Figure 21 - Data collected from RDE: 1 mg/cm

2, 0.1 M KOH, 50°C, 3600 rpm, 50 mV/sec.

Figure 21 shows the CV results from the high-loading RDE studies. In this case a 60wt%

Ni-Cr/C catalyst was used to avoid de-lamination of the catalyst layer. Catalysts with high

carbon content tend to flake off of the glassy carbon RDE substrate – especially at high mass

loading. The 1 mg/cm2 CV results indicate that the mass-activity of Ni-Cr/C catalyst is

significantly better then the state-of-the-art Ni-Mo nanoparticles reported in the literature. Both

catalysts appear to exhibit two Tafel slope regions. At low η the catalysts are presumably

operating under a Volmer(rds)-Tafel mechanism while at higher η, when the θH ~ 1 and the

mechanism likely switches to Volmer-Heyrovsky(rds) as discussed previously. The Ni-Cr/C

sample clearly reaches a state of surface saturation by Hads at lower η than the Ni-Mo sample as

the shift in Tafel slope of the Ni-Cr/C catalyst occurs ~50 mV before the shift for the Ni-Mo

catalyst.

68

Figure 22 – Interfacial study of Ni-Mo nanoparticles: Data collected from RDE:

50 µg/cm2, 0.1 M KOH, 50 °C, 3600 rpm, 50 mV/sec

In addition to the low mass-activity of unsupported Ni-Mo catalyst samples, the

Ni-Mo also exhibited decreased HER activity in the presence of AS-4 AEI (Figure 22). The

results of this interfacial study indicates that the presence of the AEI near the Ni-Mo surface

dramatically inhibits the HER activity of this catalyst. The electrostatic and interfacial models

presented in Figures 11 & 19, respectively could certainly explain the observed results on the Ni-

Mo catalyst, which is assumed to behave like a bare metal and not a composite M/MOx surface.

This indicates that while the d-band theory & volcano plots[43] have accurately predicted

optimized H-bonding energy on Ni-Mo metallic surfaces, these surfaces do not exhibit the AEI-

tolerance of the composite M/MOx materials.

Finally, Figure 23 shows a re-cap of the alkaline HER testing. The trend in activity is

quite clear. The Ni-Mo catalyst reported in the literature exhibits very low mass-activity and

requires a large η before the steep kinetic region is observed. In contrast, all carbon-supported samples

69

show significantly better HER than the relatively low surface area Ni-Mo catalyst. Thus, while the Ni-Mo

catalyst has recently garnered various high-impact publications, the loading required to achieve

operational current density in the best report of full-cell activity by Zhuang et al.[10] was 40 mg/cm2!!!

Figure 23 - Data collected from RDE: 50 µgmetal/cm2, 0.1 M KOH, 50 °C, 3600 rpm, 50 mV/sec

2.3.6 – Characterization of Ni-Cr/C structure & composition – XRD & TEM:

Figure 24 – XRD of 60% Ni-Cr/C sample. NiCr2O4 is mixture of NiO and Cr2O3. Peak

positions of standards retrieved from ICDD ref: 01-070-0989 (Ni), 01-085-0936 (NiCr2O4), 00-

004-0835 (NiO), 00-006-0504 (Cr2O3).

70

The high performance of the Ni-Cr/C catalyst was initially assumed to result from the

alloying of Ni with Cr, which would presumably result in a change of the surface electronic

structure. However, XRD & XAS studies show that even after heat-treatment at 500°C under a

reducing atmosphere, both the Ni & Cr components maintain some oxide character. Figure 24

shows the XRD profile from the Ni-Cr/C catalyst. The metallic Ni peaks are the primary phase in

the sample, but the wide peaks at 37 & 63° 2θ closely match the most prominent peaks in a

NiCr2O4 phase. This nickel chromite spinel has recently been reported to form during the

heating of NiO and Cr2O3 at 500 °C.[52, 53] However, other reports did not observe formation of

the NiCr2O4 phase until heating of NiO and Cr2O3 to 1200°C.[54, 55] The differences in the

literature are due to the size of the precursor particles. Heating microparticles requires a much

higher temperature to achieve phase transition to the mixed spinel phase, while meso-porous

mixed metal oxides and metal oxide nanoparticles with much higher surface area and less long-

range order allow for more rapid diffusion of the Ni into the Cr2O3 matrix – causing

rearrangement to the spinel phase. However, the XRD data alone cannot definitively confirm

whether the sample contains segregated NiO & Cr2O3 phases – or the spinel NiCr2O4 phase

because of the similarity of the peak profiles for each system. The literature would suggest that

the nano-scale of the material would allow for the low-temp formation of the NiCr2O4 spinel

phase, but the profile fitting of Ni & NiCr2O4 phases requires a large skew in the Ni peaks –

indicating that NiO phase is present in the sample. Nickel chromite has appeared in the literature

as a catalyst for NOx reduction[56] as well as benzene and CO hydrogenation.[57] So the spinel

clearly has a high affinity for formation of gas phase Hads in thermal reactors. Furthermore, the

literature on metal oxide stability shows that Cr2O3 is one of the most difficult oxides to reduce

71

as it exhibited no transition to metallic Cr in H2-TPR studies up to 800 °C.[58] Thus, the sample

likely contains a variety of metal oxide phases.

Figure 25 – HR-TEM images of Ni-Cr/C sample. Lower magnification (left) shows 10-20 nm

Ni-Cr particle size on Ketjen-600 carbon support. Higher magnification (right) shows

characteristic particle with possible core-shell structure. Although we cannot confirm core-shell

morphology due to lack of elemental mapping on HR-TEM, it is likely that the Cr2O3 phase may

form a porous protective shell over the Ni/NiOx composite surface

In addition to XRD analysis, the Ni-Cr/C sample was examined using HR-TEM. Selected

TEM results are shown in Figure 25. The images in Figure 25 are characteristic of the 60 wt%

Ni-Cr/C sample. In particular, the lower magnification image shows discrete particles as opposed

to the nano-film of Ni-Cr observed for the 30 wt% Ni-Cr/C sample examined using SEM (Fig.

16). However, both the 60 wt% and 30 wt% samples exhibited similar HER mass-activity and in

particular lower on-set of the steep slope kinetic region than other Ni-based samples. The really

interesting TEM data comes from the high magnification image in Figure 25. This image

indicates a core-shell morphology. The darker core of the particle indicates an electron-rich

metallic Ni core as compared to the much lighter shell, likely resulting from the lower electron-

density of the Cr2O3 phase. To further investigate the Ni-Cr catalyst, we obtained in situ XAS

72

data to analyze the chemical composition and atomic structure of the sample in contact with the

alkaline electrolyte.

2.3.7 – Characterization of Ni-Cr/C structure & composition - XAS:

In an effort to reveal the structural origin of the high HER activity of the Ni-Cr/C sample,

in situ XAS was performed on two Ni-Cr/C samples, comparing the sample with performance

reported thus far with an alternative Ni-Cr sample exhibiting significantly lower activity. Figure

26 shows the comparison of the HER activity of the two samples. The alternate Ni-Cr/C sample

exhibits HER activity characteristic of a pure Ni/C sample (i.e. bare metal as opposed to

composite M/MOx surface). This comparison is used to illustrate the relative importance of

moieties present on Ni and Cr surfaces and their role in the observed electrochemical response.

Figure 26 – Comparison of 2 Ni-Cr/C samples. Both samples produced similar XRD profiles –

but the original sample contained a much larger fraction of NiOx (as determined by XAS

analysis). Data collected from RDE: 50µgmetal/cm2, 0.1M KOH, 50°C, 3600 rpm. The high-

performance sample was synthesized as noted in the experimental section, while the alternate

sample with inferior HER performance was produced using a much higher concentration of pre-

cursor salts in the reaction solution.

For in-situ XAS studies, the samples were analyzed using 0.1 M KOH electrolyte with a

specially designed cell to determine the details of the changes to the coordination environments

73

around Ni and Cr under in situ conditions. The electrodes were analyzed at various potentials

from OCP = 0.8 V down to 0 V vs. RHE. No significant changes in XAS data were observed for

either Ni or Cr-edges upon changing the applied potential, this is expected based on the lack of

any redox changes predicted in Pourbaix diagrams or observed in CV results. Thus the

representative spectra collected at 0 mV is displayed in Figure 27 & 28 for discussion (attempts

to collect data below 0 V vs. RHE resulted in formation of H2-bubbles on the “window” of the in

situ cell, causing noisy spectra which could not be analyzed).

Figure 27 – XANES (left) and FT-EXAFS (right) data at the Cr K edge collected on Ni-Cr/C

electrodes. The data for Cr2O3 standard is also included for comparison purposes.

Figure 28 – XANES (left) and FT-EXAFS (right) data at the Ni K edge collected on Ni-Cr/C

electrodes. The XANES of Ni(OH)2 standard is also included for comparison purposes.

74

Figure 27 clearly shows that the XANES and FT-EXAFS for the Cr K-edge of the two

samples are essentially identical. In addition, the Cr-edge XANES spectra of the two samples

are very similar to the Cr2O3 standard. The lack of fine details in the Cr2O3 XANES spectra of

both samples compared to the reference indicates a lack of long-range order required to give

coherent scattering profiles. Furthermore, the lack of prominent peaks in the FT-EXAFS beyond

3 Å – compared to those of the Cr2O3 standard (Figure 27, right) indicates a lack of long-range

order structure – i.e., the Cr2O3 exists as very small, nearly amorphous particles. The similarity

of the Cr-composition for the high- & low-performance samples indicates that the Cr2O3

component is not providing the synergistic HER enhancement.

In contrast, the Ni-edge XANES and FT-EXAFS data for both samples are significantly

different (Figure 28). The radial coordination environment shown by the Fourier transform of the

Ni K edge shows the Ni-O interactions in the first peak (around 2 Å) followed with two specific

Ni-Ni interactions between 3 and 5 Å. Ni K edge XANES also is in close concert with core level

transition of s electrons and hence the electronic states near the Fermi level of Ni. Comparison

of the Ni K edge XAS, shows that while the alternate (poorly performing) sample exhibits only

slightly higher white line intensity at 8350 eV as compared to the reference Ni-foil, the original

(high-performance) sample exhibits a much higher white line intensity, approaching that of the

Ni(OH)2 standard (Figure 28, left). This strongly suggests the presence of higher degree of Ni-

oxides in the superior sample as compared to those present in the inferior HER sample. White

line intensity is directly attributable to charge transfer from the Ni to the oxygen neighbors,

hence higher degree of oxides is directly related to the magnitude of the Ni-O interactions and

hence peak magnitude. Consistently, the FT peaks of the inferior sample perfectly overlap those

of the reference Ni-foil out to 5 Å with only slightly lower intensity, confirming the sample is

75

dominated by metallic Ni (Ni0). On the other hand, the original (high-performance) sample

exhibits FT peaks at the same position as the Ni foil but with much lower intensity, as well as a

distinct peak at ~2.5 Å (without phase correction) that is not present in the inferior sample,

indicating the multiple-component nature of the high-performance sample.

Figure 29 – EXAFS fitting results from data collected from “Original” Ni-Cr/C sample. Ni edge

fitting confirms mixture of metallic Ni0 and Ni-oxide. While Cr-edge fitting excludes the

presence of NiCr2O4 since the Ni-Metal bond (2.98 Å) is in agreement with that in Ni-oxide

(2.95 Å in NiOx) but much smaller than the Ni-Metal bond distance (3.42 Å) in NiCr2O4.

Analysis of Figures 27 & 28 indicates that the Ni/NiOx surface is behaving like

Markovic’s proposed archetypal M/MOx to provide synergistic HER enhancement. Thus, the

Cr2O3 component likely stabilizes the NiOx component under HER conditions where it would

typically be reduced to Ni0. As the XRD results in Figure 5 could not distinguish between the

presence of NiCr2O4 and phase-segregated NiO & Cr2O3, EXAFS fitting was performed on the

XAS data from the original (high performance) sample. Figure 29 shows the results of the

EXAFS fitting which was conducted at the Ni and Cr K edges concurrently. The fitting details

and results are given in the Table 2. The fitting confirms that the peak at ~2.5 Å (present on the

original – but not the alternate sample) arises from the Ni-Ni scattering with a bond distance of

2.98 Å from Ni oxides (Figure 8, left). The Ni-Ni bond distance observed in the sample rules out

76

the existence of the spinel NiCr2O4 phase in which the Ni-Ni scattering from Ni oxides would

show a bond distance of 3.42Å. Thus, the EXAFS fitting can definitively confirm the presence of

phase-segregated NiO and Cr2O3 – and exclude the presence of any significant quantity of

NiCr2O4 phase.

Table 2: Summary of EXAFS results of Ni-Cr/C (original) sample, together with the data of

several standard references for comparison.

Sample tested Standard references

Ni-Cr/C (Original) Cr

2O

3 NiCr

2O

4

R (Å) N

σ2

×10-3

N(Å2

) R (Å) N R (Å) N

Cr-Cr 2.98±0.02 4±2 9±8 2.8 4 2.94 6

Cr-O 1.99 3.7±0.8 0.5±2 2.01 6 1.99 6

Ni foil

Ni-Ni

(metallic Ni) 2.49±0.01 4.7±1.4 6±2 2.49 12

NiO

Ni-O

scattering 2.05±0.03 3.1±1.6 6±7 2.08 6 1.96 4

Ni-Ni

scattering

(NiOx)

2.98±0.02 3.1±1.1 6±2 2.95 12 3.42 8

S02 fixed at 0.83 and 0.80 for Ni and Cr, respectively as obtained by fitting the reference foils.

Fits were done in R-space, k1,2,3

weighting. For Ni, 1.0< R < 3.4 Å and Δk = 3.16 - 12.48 Å-1

were used; for Cr, 1.0 < R < 3.0 Å and Δk = 2.73 – 10.73 Å-1

were used.

The most significant difference in the XAS analysis of the two samples is that the

original (high-performance) Ni-Cr/C sample exhibited a much larger NiOx content which can be

clearly seen in the much larger white line intensity and the FT peak at 2.5 Å in Figure 7. These

results clearly indicate that the NiOx and not the Cr2O3 must be responsible for the increased

HER activity of the original sample. Thus it may be that the amorphous Cr2O3 acts to stabilize

the NiOx, but doesn’t appear to enhance HER activity directly. This data is supported by HR-

77

TEM results (Fig. S15) that appear to show a core-shell structure of Ni-NiO/Cr2O3 particles in

the Ni-Cr/C sample. It is quite feasible that a porous Cr2O3 shell could stabilize the NiO

component in the oxide state under HER conditions where the NiO would typically be reduced to

metallic Ni0. The stability of the composite M/MOx surface under HER conditions was observed

in the stability of the HER performance during repeated CV cycling.

2.4 Conclusions:

After screening the HER activity of numerous binary and ternary Ni-alloys and

composite Ni/MOx/C materials, a Ni-Cr/C sample was identified which exhibits unprecedented

mass-activity for the HER in 0.1 M KOH electrolyte. In particular, the Ni-Cr/C sample required

the least amount of HER overpotential to achieve what literature indicates to be a direct Volmer-

Heyrovsky mechanism. In addition, the effects of the AEI binder on the HER have been

investigated for Pt, Ni-Cr and Ni-Mo catalysts. For Pt catalysts, careful control of CV potential

range limits indicates that the QA+ moiety of the AEI likely straddles the double-layer interface

and does not reside exclusively in the IHP. CV results from larger (~6 nm) unsupported Pt black

and smaller (~2-3 nm) supported Pt/C catalysts indicates that at high-index Pt (111) surface sites,

the QA+ moiety of the AEI is specifically adsorbed in the IHP, while for other Pt (100) & (110)

surface sites, the QA+ moiety likely resides in the OHP, but still exerts an electrostatic effect,

dampening the HER activity of the catalyst. Furthermore, the AEI-inhibition of the HER is

observed on Ni-Mo, but not on Ni-Cr/C. The cause of the AEI-tolerance by Ni-Cr is not

confirmed but a model is proposed surmising that the AEI relaxes the dipole orientation

constraints of H2O in the IHP thus allowing more H2O reactant molecules to assume an

orientation which facilitates H-OH cleavage on adjacent M/MOx surface sites.

78

Various analytical techniques (XRD, EDS, XAS) have identified that the Ni-Cr/C sample

actually contains metallic nickel as well as nickel- and chromium-oxides. XAS analysis of two

different Ni-Cr/C samples indicates that the HER enhancement is related to the adjacent Ni/NiOx

surface sites and that the Cr2O3 phase may act to stabilize the NiOx under the reducing HER

conditions. The HER enhancement on composite Ni/NiOx surfaces was observed by Lasia et

al.[51] by cycling a polycrystalline Ni electrode from HER to OER conditions, thus roughening

the surface with NiOx, but the effects were not stable because the NiOx was quickly reduced back

to the metallic state under HER conditions. More recently, enhanced HER has been reported on

Ni/NiOx by Markovic et al.[36] and Gong et al.[59] – both of whom suggested that at adjacent

metal/metal-oxide sites, the metal-oxide facilitates the formation of OHads, thus weakening the

H-OH bond in the HER reactant and reducing activation energy for the Volmer reaction.

However, the Ni Pourbaix diagram indicates that the NiOx component of these composite

surfaces is not stable below 0 V (vs. RHE) and will quickly reduce to metallic Ni, thus losing the

synergistic HER enhancement of adjacent Ni/NiOx surface sites. All HER data in this paper was

collected after confirming stable CV performance after at least 20 scans from -0.6 to +1.0 V (i.e.

well before the on-set of surface roughening under OER conditions). Thus, according to the

XAS data shown here it is likely that the Cr2O3 component stabilizes the NiOx component well

into the HER operating voltage, allowing the catalyst to retain the enhanced HER activity from

synergistic Ni/NiOx surface sites. Further studies of the effects of the work-up conditions can be

conducted to optimize the structure and activity of this M/MOx catalyst. This type of composite

metal/metal-oxide electrocatalyst may be the key to realizing the promise of alkaline

electrochemistry for low-cost, high-purity hydrogen production.

79

2.5 Acknowledgements:

This work funded by the Advanced Research Projects Agency – Energy (ARPA-E), U.S.

Department of Energy, under Award No. DE-AR0000121. We also thank Tokuyama Corp. For

providing AS-4 ionomer solution and Proton On-Site for providing Pt black catalyst. Use of the

National Synchrotron Light Source (beamline X3B), Brookhaven National Lab was made

possible by the Center for Synchrotron Biosciences grant, P30-EB-009998, from the National

Institute of Biomedical Imaging and Bioengineering (NBIB). Support from beamline personnel

Dr. Erik Farquhar (X3B) is gratefully acknowledged. Thanks to Dr. Qingying Jia for analysis of

the XAS data and the related discussions. Also, thanks to Dr. Nagappan Ramaswamy for his help

reviewing the electrochemical discussions and for his assistance developing the electrostatic

model of the double-layer interface. Special thanks to Dr. Wentao Liang for collecting HR-TEM

data and to Samuel Boylston for his assistance in developing the figure for the interfacial model.

80

2.6 References:

1. Carmo, M., et al., A comprehensive review on PEM water electrolysis. International

Journal of Hydrogen Energy, 2013. 38(12): p. 4901-4934.

2. Kriston, A., T. Xie, and B.N. Popov, Impact of Ultra-low Platinum loading on Mass

Activity and Mass Transport in H2-Oxygen and H2-Air PEM Fuel Cells. Electrochimica Acta,

2014. 121(0): p. 116-127.

3. Urian, R.C., A.F. Gullá, and S. Mukerjee, Electrocatalysis of reformate tolerance in

proton exchange membranes fuel cells: Part I. Journal of Electroanalytical Chemistry, 2003.

554–555(0): p. 307-324.

4. Mukerjee, S., et al., Electrocatalysis of CO Tolerance by Carbon-Supported PtMo

Electrocatalysts in PEMFCs. Journal of The Electrochemical Society, 2004. 151(7): p. A1094-

A1103.

5. Ehteshami, S.M.M., et al., The role of electronic properties of Pt and Pt alloys for

enhanced reformate electro-oxidation in polymer electrolyte membrane fuel cells.

Electrochimica Acta, 2013. 107(0): p. 155-163.

6. Birry, L. and A. Lasia, Studies of the Hydrogen Evolution Reaction on Raney Nickel—

Molybdenum Electrodes. Journal of Applied Electrochemistry, 2004. 34(7): p. 735-749.

7. McKone, J.R., et al., Ni–Mo Nanopowders for Efficient Electrochemical Hydrogen

Evolution. ACS Catalysis, 2012. 3(2): p. 166-169.

8. Miousse, D., A. Lasia, and V. Borck, Hydrogen evolution reaction on Ni-Al-Mo and Ni-

Al electrodes prepared by low pressure plasma spraying. Journal of Applied Electrochemistry,

1995. 25(6): p. 592-602.

9. Pourbaix, M., Atlas of Electrochemical Equilibria in Aqueous Solutions. 1966, Brussels:

Pergamon Press.

10. Xiao, L., et al., First implementation of alkaline polymer electrolyte water electrolysis

working only with pure water. Energy & Environmental Science, 2012. 5(7): p. 7869-7871.

81

11. Brown, D.E., et al., Preparation and characterization of low overvoltage transition metal

alloy electrocatalysts for hydrogen evolution in alkaline solutions. Electrochimica Acta, 1984.

29(11): p. 1551-1556.

12. Bagotzky, V.S. and N.V. Osetrova, Investigations of hydrogen ionization on platinum

with the help of micro-electrodes. Journal of Electroanalytical Chemistry and Interfacial

Electrochemistry, 1973. 43(2): p. 233-249.

13. Sheng, W., 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): p. B1529-B1536.

14. Subbaraman, R., et al., Enhancing Hydrogen Evolution Activity in Water Splitting by

Tailoring Li+-Ni(OH)2-Pt Interfaces. Science, 2011. 334(6060): p. 1256-1260.

15. Kendrick, I., et al., Elucidating the Ionomer-Electrified Metal Interface. Journal of the

American Chemical Society, 2010. 132(49): p. 17611-17616.

16. Zhang, H. and P.K. Shen, Recent Development of Polymer Electrolyte Membranes for

Fuel Cells. Chemical Reviews, 2012. 112(5): p. 2780-2832.

17. Pan, J., et al., Designing Advanced Alkaline Polymer Electrolytes for Fuel Cell

Applications. Accounts of Chemical Research, 2011. 45(3): p. 473-481.

18. Li, N., et al., Highly Stable, Anion Conductive, Comb-Shaped Copolymers for Alkaline

Fuel Cells. Journal of the American Chemical Society, 2013. 135(27): p. 10124-10133.

19. Ünlü, M., et al., Analysis of Double Layer and Adsorption Effects at the Alkaline Polymer

Electrolyte-Electrode Interface. Journal of The Electrochemical Society, 2011. 158(11): p.

B1423-B1431.

20. Arruda, T.M., et al., Fundamental Aspects of Spontaneous Cathodic Deposition of Ru

onto Pt/C Electrocatalysts and Membranes under Direct Methanol Fuel Cell Operating

Conditions: An in Situ X-ray Absorption Spectroscopy and Electron Spin Resonance Study. The

Journal of Physical Chemistry C, 2009. 114(2): p. 1028-1040.

21. Newville, M., IFEFFIT: interactive XAFS analysis and FEFF fitting. Journal of

synchrotron radiation, 2001. 8(2): p. 322-324.

82

22. Ravel, B. and K. Gallagher, Atomic structure and the magnetic properties of Zr-doped

Sm2Co17. Physica Scripta, 2005. 2005(T115): p. 606.

23. Newville, M., et al., Near-edge x-ray-absorption fine structure of Pb: A comparison of

theory and experiment. Physical Review B, 1993. 47(21): p. 14126.

24. Ankudinov, A., et al., Real-space multiple-scattering calculation and interpretation of x-

ray-absorption near-edge structure. Physical Review B, 1998. 58(12): p. 7565.

25. Spendelow, J.S. and A. Wieckowski, Electrocatalysis of oxygen reduction and small

alcohol oxidation in alkaline media. Physical Chemistry Chemical Physics, 2007. 9(21): p. 2654-

2675.

26. Ramaswamy, N. and S. Mukerjee, Fundamental Mechanistic Understanding of

Electrocatalysis of Oxygen Reduction on Pt and Non-Pt Surfaces: Acid versus Alkaline Media.

Advances in Physical Chemistry, 2012. 2012: p. 17.

27. Gong, M., et al., An Advanced Ni–Fe Layered Double Hydroxide Electrocatalyst for

Water Oxidation. Journal of the American Chemical Society, 2013. 135(23): p. 8452-8455.

28. Li, X., F.C. Walsh, and D. Pletcher, Nickel based electrocatalysts for oxygen evolution in

high current density, alkaline water electrolysers. Physical Chemistry Chemical Physics, 2011.

13(3): p. 1162-1167.

29. Hu, Q., et al., Alkaline polymer electrolyte fuel cell with Ni-based anode and Co-based

cathode. International Journal of Hydrogen Energy, 2013. 38(36): p. 16264-16268.

30. Lu, S., et al., Alkaline polymer electrolyte fuel cells completely free from noble metal

catalysts. Proceedings of the National Academy of Sciences, 2008. 105(52): p. 20611-20614.

31. Danilovic, N., et al., The Effect of Noncovalent Interactions on the HOR, ORR, and HER

on Ru, Ir, and Ru0.50Ir0.50 Metal Surfaces in Alkaline Environments. Electrocatalysis, 2012.

3(3-4): p. 221-229.

32. Strmcnik, D., et al., Improving the hydrogen oxidation reaction rate by promotion of

hydroxyl adsorption. Nat Chem, 2013. 5(4): p. 300-306.

83

33. Ohyama, J., et al., Size Specifically High Activity of Ru Nanoparticles for Hydrogen

Oxidation Reaction in Alkaline Electrolyte. Journal of the American Chemical Society, 2013.

135(21): p. 8016-8021.

34. Wang, Y., et al., Pt–Ru catalyzed hydrogen oxidation in alkaline media: oxophilic effect

or electronic effect? Energy & Environmental Science, 2015. 8(1): p. 177-181.

35. Sheng, W., et al., Correlating hydrogen oxidation and evolution activity on platinum at

different pH with measured hydrogen binding energy. Nature communications, 2015. 6.

36. Danilovic, N., et al., Enhancing the Alkaline Hydrogen Evolution Reaction Activity

through the Bifunctionality of Ni(OH)2/Metal Catalysts. Angewandte Chemie International

Edition, 2012. 51(50): p. 12495-12498.

37. Subbaraman, R., et al., Trends in activity for the water electrolyser reactions on 3d

M(Ni,Co,Fe,Mn) hydr(oxy)oxide catalysts. Nat Mater, 2012. 11(6): p. 550-557.

38. Rheinländer, P.J., et al., Kinetics of the Hydrogen Oxidation/Evolution Reaction on

Polycrystalline Platinum in Alkaline Electrolyte Reaction Order with Respect to Hydrogen

Pressure. Journal of The Electrochemical Society, 2014. 161(14): p. F1448-F1457.

39. Jia, Q., et al., Activity Descriptor Identification for Oxygen Reduction on Platinum-Based

Bimetallic Nanoparticles: In Situ Observation of the Linear Composition–Strain–Activity

Relationship. ACS Nano, 2015. 9(1): p. 387-400.

40. Markovića, N.M., et al., Hydrogen electrochemistry on platinum low-index single-crystal

surfaces in alkaline solution. J. Chem. Soc., Faraday Trans., 1996. 92(20): p. 3719-3725.

41. Markovic, N., H. Gasteiger, and P.N. Ross, Kinetics of Oxygen Reduction on Pt(hkl)

Electrodes: Implications for the Crystallite Size Effect with Supported Pt Electrocatalysts.

Journal of The Electrochemical Society, 1997. 144(5): p. 1591-1597.

42. Böckris, J., A. Reddy, and M. Gamboa-Aldeco, Modern Electrochemistry, 2A:

Fundamentals of Electrodics. Modern Electrochemistry, 2A: Fundamentals of Electrodics, 2000.

43. Jakšić, J.M., M.V. Vojnović, and N.V. Krstajić, Kinetic analysis of hydrogen evolution at

Ni–Mo alloy electrodes. Electrochimica Acta, 2000. 45(25–26): p. 4151-4158.

84

44. Soares, D.M., O. Teschke, and I. Torriani, Hydride Effect on the Kinetics of the Hydrogen

Evolution Reaction on Nickel Cathodes in Alkaline Media. Journal of The Electrochemical

Society, 1992. 139(1): p. 98-105.

45. Mauer, A.E., D.W. Kirk, and S.J. Thorpe, The role of iron in the prevention of nickel

electrode deactivation in alkaline electrolysis. Electrochimica Acta, 2007. 52(11): p. 3505-3509.

46. Jaksic, J.M., et al., Novel Spillover Interrelating Reversible Electrocatalysts for Oxygen

and Hydrogen Electrode Reactions. The Journal of Physical Chemistry C, 2010. 114(43): p.

18298-18312.

47. Jaksic, M.M., et al., Primary Oxide Latent Storage and Spillover Enabling

Electrocatalysts with Reversible Oxygen Electrode Properties and the Alterpolar Revertible

(PEMFC versus WE) Cell. The Journal of Physical Chemistry C, 2014. 118(17): p. 8723-8746.

48. Santos, E., P. Quaino, and W. Schmickler, Theory of electrocatalysis: hydrogen evolution

and more. Physical Chemistry Chemical Physics, 2012. 14(32): p. 11224-11233.

49. Hall, D.S., C. Bock, and B.R. MacDougall, The Electrochemistry of Metallic Nickel:

Oxides, Hydroxides, Hydrides and Alkaline Hydrogen Evolution. Journal of The Electrochemical

Society, 2013. 160(3): p. F235-F243.

50. Hitz, C. and A. Lasia, Experimental study and modeling of impedance of the her on

porous Ni electrodes. Journal of Electroanalytical Chemistry, 2001. 500(1–2): p. 213-222.

51. Lasia, A. and A. Rami, Kinetics of hydrogen evolution on nickel electrodes. Journal of

Electroanalytical Chemistry and Interfacial Electrochemistry, 1990. 294(1–2): p. 123-141.

52. Stoia, M., et al., Studies regarding the formation from metal nitrates and diol of NiM 2 III

O4 spinels, inside a silica matrix. Journal of Thermal Analysis and Calorimetry, 2012. 108(3): p.

1041-1049.

53. Ştefănescu, M., et al., The obtaining of NiCr2O4 nanoparticles by unconventional

synthesis methods. Journal of Thermal Analysis and Calorimetry, 2013. 111(2): p. 1121-1127.

54. Cheng, X.-d., et al., Preparation of high emissivity NiCr2O4 powders with a spinel

structure by spray drying. International Journal of Minerals, Metallurgy, and Materials, 2012.

19(2): p. 173-178.

85

55. Greskovich, C., Kinetics of NiCr2O4 Formation and Diffusion of Cr3+ Ions in NiO.

Journal of the American Ceramic Society, 1970. 53(9): p. 498-502.

56. Gassan-Zadeh, G.Z., M.Y. Woode, and T.G. Alkhazov, Effect of oxygen on NO reduction

over nickel chromite catalysts. Reaction Kinetics and Catalysis Letters, 1988. 36(1): p. 33-37.

57. Simentsova, I.I., et al., The effect of the precursor structure on the catalytic properties of

the nickel—chromium catalysts of hydrogenation reactions. Russian Chemical Bulletin, 2010.

59(11): p. 2055-2060.

58. Ren, Y., P.G. Bruce, and Z. Ma, Solid-solid conversion of ordered crystalline

mesoporous metal oxides under reducing atmosphere. Journal of Materials Chemistry, 2011.

21(25): p. 9312-9318.

59. Gong, M., et al., Nanoscale nickel oxide/nickel heterostructures for active hydrogen

evolution electrocatalysis. Nat Commun, 2014. 5.

86

Chapter 3 Alkaline Oxygen Evolution Reaction on Mixed-Metal-Oxides

3.1 Introduction:

The Oxygen Evolution Reaction (OER) is critical for clean energy conversion and

storage. Electrochemical water splitting requires catalysts for hydrogen evolution and oxygen

evolution. Like the oxygen reduction reaction in fuel cells, the oxygen evolution reaction in

water electrolysis is the primary obstacle which requires advanced catalysts to increase system

efficiency. Ideally, intermittent solar or wind energy could be stored in the form of hydrogen gas

via electrochemical water splitting. Hydrogen as an energy carrier could be used to power zero-

emission fuel-cell vehicles,[1] residential combined heat & power systems,[2-4] or even grid-

scale energy-buffering in unitized regenerative fuel-cells or H2-Br2 redox flow battery

systems.[5, 6] Although photochemical systems which convert photons directly to hydrogen and

oxygen are under development, the relative H2-production rates are orders of magnitude lower

than electrochemical systems[7-9], thus the photo-electrochemical systems would require

massive land-use “footprint” for large-scale hydrogen production, while water electrolysis

systems offer a near-term route to developing modular H2-production systems for niche markets.

Recent breakthroughs in alkaline polymer electrolytes (APEs) have sparked a flurry of

research into non-platinum group metal (non-PGM) electrocatalysts for fuel-cells and

electrolyzers.[10-12] Non-PGM materials consist mainly of nitrogen-doped carbons and/or group

3d transition metals and represent a two order-of-magnitude cost reduction compared to their

PGM counterparts (i.e. “pennies on the dollar”). Operation in the high-pH environment of APEs

circumvents the “stability criterion” of traditional proton-exchange membrane (PEM) systems,

and allows for the use of metals such as Ni, Fe, Co, etc. which would rapidly dissolve in the

highly acidic environment of PEM systems.

87

In traditional PEM systems, Pt is used for the hydrogen evolution reaction (HER) while Ir

is used for the OER.[13] In an effort to replace these prohibitively expensive materials which

preclude commercialization of water electrolysis, recent research efforts have examined non-

PGM electrocatalysts in APE systems. On the HER cathode side, various non-PGM candidates

such as Ni-Mo, Ni-Cr and other composite metal / metal-oxide surfaces are under development

and show very promising mass-activity and stability for alkaline HER.[14-17]

For the OER, research has identified three main classes of non-PGM electrocatalyst

materials: spinels (with AB2O4 crystal structure), perovskites (with ABO3 crystal structure) and

mixed metal-oxides (typically with MxM1-x(OH)y crystal structure). Singh et al.[18, 19] (among

many others) have conducted extensive screening of a wide range of spinel OER catalysts with

results indicating that catalysts containing both Ni & Fe exhibit the highest OER activity.

However, typical spinels materials reported in the literature are unsupported nanoparticles with

moderate surface area (30-50 m2/g) and thus exhibit only moderate mass-activity at relevant

operating potentials. As such, fundamental properties and activity descriptors have not been

examined in much detail on spinels. In contrast, perovskites have played a starring role in recent

literature because they have shown very promising fundamental properties, such as high

“specific activity” (activity normalized to their physical surface area). Unfortunately, the

calcination temperatures required to produce the perovskites-phase (700-1100°C) also result in

significant particle sintering resulting in physical surface area typically below 1 m2/g, and thus

the mass-activity for these samples is lower than mass-activity reported for MMO catalysts.[20-

23] However, Hardin et al.[24] have recently reported a novel method to produce La(Ni,Co)O3

perovskites with surface-area of ~10 m2/g which exhibit OER activity approaching state-of-the-

art MMO catalysts. In contrast to the thermal treatments required to form spinel and perovskites

88

phase catalysts, MMO materials do not require high temperature heat treatment and can be

fabricated by electrochemical deposition or simple (& “green”) hydrothermal methods to

produce the transition metal oxide films.[25-27] A significant trend noted across all three types

of OER materials is that samples with mixed metals (Ni-Fe, Ni-Co & Fe-Co) in the active-site

position (A-site for spinels, B-site for perovskites and M-site for MMOs) appear to exhibit

enhanced OER activity compared to samples with mono-metallic active sites. Thus, this study

investigated ternary combinations of Ni, Fe & Co.

Historically, alkaline electrolysis has been conducted in hot, liquid caustic electrolyte

with Raney-Ni electrodes.[28, 29] The porous Raney-Ni provides high surface area and

relatively high electronic conductivity (i.e. low Ohmic contribution to over-potential). These

systems have proven long term durability, but the hot caustic systems are not favored due to

safety concerns related to leaking of the highly corrosive liquid electrolyte, these safety concerns

have motivated the development of APEs for water electrolysis. Due to the proven activity and

stability of Raney-Ni electrodes, this material was chosen as a starting point for development of

MMO OER electrocatalysts. The addition of the PANI component provides an added degree of

active-site dispersion, as well as increased conductivity of the catalyst layer. While it has been

shown that PANI nano-fibers form spontaneously in the presence of metal salts as mild

oxidants,[30] SEM investigations of these MMO/PANI-Raney catalysts show the presence of a

high surface-area PANI film on the Raney-Ni substrate but no indication of discrete PANI nano-

fibers (ESI Figure S8). Thus, additional optimization of synthesis conditions to achieve PANI

nano-fiber morphology may further increase OER activity.

89

90

Figure 1 – SEM images of (8:1:1 at. ratio) Ni-Fe-Co/PANI-Raney catalyst. Note the presence of the high

surface area MMO-PANI film on the Raney-Ni substrate.

Although the presence of carbon based materials in water electrolysis anodes is typically

avoided because the high voltage operating conditions would invariably lead to carbon corrosion,

recent literature has shown that adding carbon nanotubes or quantum dots to OER catalysts

increases the conductivity and OER activity of these materials which have shown stability under

OER operating conditions in lab scale RDE testing.[26, 27]

Screening of mono-metallic metal-oxide films by Markovic et al.[31] previously

indicated that of the group 3d metals, Ni is the most active for OER. However, it has recently

been claimed that the high apparent activity of “pure Ni-oxide” samples in previous reports was

due to contamination from trace Fe in the electrolytes.[32] Numerous studies reveal that Ni-Fe

mixed oxides provide a greatly increased OER-activity compared to pure Ni electrodes – or other

binary mixed oxides.[33, 34] Various studies of Ni-Fe systems have shown that Ni-rich samples

(in the range of 60-90 wt% Ni) exhibit optimized activity.[25, 35-39] The wide range reported

for “optimum” Ni content is likely due to the various synthesis methods and heat treatments to

which the samples were subjected. In particular, it has been shown that layered double-

hydroxides (LDH) are more active than the calcined spinel-phase nickel ferrite catalysts.

Furthermore, it has been shown that non-conductive FeOOH phases nucleate above a certain Fe-

content threshold.[37, 40, 41] Until very recently, it was widely assumed that Ni was the OER

active-site and the role of the Fe in Ni-Fe catalysts was rather ambiguous. The addition of Fe was

certainly beneficial in facilitating formation of the LDH phase, thus increasing the effective

ECSA. In addition, Boettcher et al.[32] conducted in-situ conductivity measurements of Ni-Fe

films which showed that Fe increases the conductivity of the Ni-oxide films but the increased

91

conductivity could not completely account for the increased OER activity of Ni-Fe films and

thus the further OER enhancement of Ni by Fe was not well understood. However, very recent

in-situ XAS analysis of Ni-Fe-oxides and DFT calculations by Bell et al.[40] have indicated that

Fe – not Ni, is the OER active-site in Ni-Fe catalysts. Furthermore, Boettcher et al.[41]

conducted in-situ conductivity testing of Co1-xFexOOH films which indicated that the pure

FeOOH films are not conductive until >1.62 V (vs. RHE) and thus, a conductive matrix (of Co

or Ni oxide) is required to “activate” the Fe sites.

It is in line with this discussion that we turn our attention to the observed shift in the

Ni2+/3+

redox peak potentials. The anodic peak typically occurs ~1.45 V vs. RHE for mono-

metallic Ni samples, however we have observed a shift to more positive potentials (>1.45 V vs.

RHE) when mixing Ni with Fe and to less positive potentials (<1.4 V vs. RHE) when mixing Ni

with Co. These observations of the electrochemical response of MMOs are supported by in-situ

XAS analysis which shows that Fe acts to stabilize Ni in the 2+ oxidation state, while Co

facilitates oxidation to the 3+ state. It is well-known that Ni(OH)2 is insulating, while the

oxidized NiOOH phase is significantly more conductive.[42] In this study, a ternary Ni-Fe-Co

MMO sample exhibits the highest OER activity of any binary or ternary sample evaluated to

date. Thus, in reference with the above-mentioned results of Bell and Boettcher, we propose that

the charge-transfer effects of the Co component facilitate oxidation of the insulating Ni(OH)2 to

the conductive NiOOH, thus “activating” the Fe OER active-sites in the conductive Ni/Co-oxide

host.

Additionally, it has been observed that for thin-film MMOs on carbon supports, heat

treatment decreases the observed Ni2+/3+

redox charge, but does not significantly affect OER

activity, while the MMO films on PANI/Raney-Ni support exhibit an increase in OER activity

92

upon heating. However, Geng et al.[43] have reported that amorphous Ni-Fe-oxide microspheres

exhibit decreased activity upon calcination (likely due to nucleation of non-conductive Fe-oxide

crystallites). Thus, it is presumed that the increase in OER activity upon heating our

MMO/PANI-Raney samples may be the result of two possible phenomena: 1) Removal of

inactive synthesis by-products (i.e. “bare” PANI). TGA results show ~15% decrease in mass

upon heating (ESI figures S4 & S5), indicating the removal of “bare” PANI or aniline oligomer

components, leaving only OER-active MMO surfaces. 2) Diffusion of Fe (&/or Co) from the

MMO surface film into the Raney-Ni substrate during heat-treatment. The diffusion of Fe &/or

Co into the Raney-Ni substrate may increase the electrocatalytic surface area by facilitating the

formation of the more active LDH structure deeper into the pores of the Raney-Ni support.

3.2 Experimental:

Raney-Ni supported catalysts were synthesized via a modified impregnation method

(IM) technique,[44] in the presence of aniline to form a MMO – polyaniline (PANI) composite

film on the Raney-Ni support. To a slurry of Raney-Ni (Raney-2800, Sigma-Aldrich) in H2O

(18.2 MΩ Millipore) was added an appropriate amount of metal nitrate salts (Reagent Grade,

Sigma-Aldrich) to produce a film of 40 wt% MMO on Raney-Ni. The composite MMO-PANI

film was produced by adding 10 molar excess (w.r.t. metal salts) aniline (Sigma-Aldrich) to the

reaction solution and stirring gently for ~1 hour prior to drop-wise addition of a 3 molar excess

(w.r.t. metal salts) of sodium borohydride reducing agent (Sigma-Aldrich). The reaction solution

was then left to stir overnight and subsequently filtered, washed and dried before heat-treatment

in a tube furnace. In addition, various carbon-supported binary and ternary Ni-alloy

electrocatalysts were synthesized by a standard impregnation method (IM) using NaHB4. The

93

carbon support was Vulcan XC-72R (Cabot Corporation, Billerica, MA). The support was

dispersed in the reaction solution and stirred for at least 1 hour, followed by addition of metal

nitrate salts and an additional 1 hour stirring. Finally, NaHB4 was added drop-wise under

vigorous stirring. The solution was stirred for at least 1 hour to ensure complete reaction. Both

Raney-Ni and C-supported samples were vacuum-filtered through a Büchner funnel and washed

with 500 mL DI H2O. The Raney-Ni supported samples were also washed with 200mL of

acetone to remove any free aniline or aniline oligomers. The solid products were dried in a

vacuum-oven at 80°C overnight. Heat-treatment studies were performed in a tube furnace under

air for calcination or argon for annealing.

For RDE studies, inks composed of 3 mL H2O, 6.95 mL 2-propanol and 50 μL 5wt%

ionomer dispersion were mixed with an appropriate amount of catalyst. The ionomer dispersion

used was “Nafion” (perfluorosulfonic acid-PTFE co-polymer, Alfa Aesar, Ward Hill, MA). The

inks were sonicated for at least 30 min before a 10 μL aliquot was deposited on the tip of a

polished glassy carbon disk (5.61 mm diameter) to produce a loading of 50 μg(MMO)/cm2 for C-

supported catalysts and 250 μg/cm2 for Raney-Ni supported catalysts. The catalyst layers were

spin-coated on the RDE tip using an inverted Pine Instruments rotator to ensure uniform

distribution of the catalyst. For examination of the Raney-Ni supported catalysts, CLs also

contained 50 μg/cm2 Acetylene Black (Chevron) to maximize the dispersion and increase the

electronic conductivity of the CL. Electrochemical tests were conducted with an Autolab

(Ecochemie Inc., model-PGSTAT30) potentiostat/galvanostat. Tests were conducted in a 100

mL 3 electrode cell. Alkaline (0.1 M KOH) electrolyte was prepared using potassium hydroxide

pellets (semiconductor grade 99.99%, Sigma-Aldrich) and ultra pure H2O (18.2 MΩ Millipore).

For each test, a freshly made RHE was used as the reference electrode and a gold flag counter

94

electrode was used to avoid Pt contamination. The glassy carbon working electrode (WE) was

rotated using an RDE setup from Pine Instruments. Rotation rates of 2500 rpm were sufficient to

remove the O2(g) product from the surface and examine the kinetics well into the OER region.

All results were obtained after conditioning electrodes with 100 mV/sec CV scans (from 1-1.8

V) for 20-30 scans, or until stable features were observed.

XRD characterization was conducted using a Rigaku Ultima IV XRD with a Cu Kα

source (λ=1.541 Å) operated at 40 kV and 44 mA. 2θ/θ scans were conducted using a 0.05° step

size and 5 sec hold per step. SEM characterization was conducted using a Hitachi S-4800 FE-

SEM. EDS data was collected using EDAX Genesis on the SEM to validate sample elemental

composition. XAS measurements were conducted at the X3B beamline of Brookhaven National

Labs and analysis was performed using the IFEFFIT suite. Additional details of XAS

measurements are provided in supporting information.

3.3 Results & Discussion:

3.3.1 – Validation of Metal Trends for OER:

In an effort to validate previous trends of OER activity on mixed metal oxides, we first

synthesized mono-metallic films on the combined PANI/Raney-Ni support. Figure 2 shows the

results of cyclic voltammetry testing on the films of Ni, Mo, Co and Fe on PANI/Raney-Ni

support. The results indicate that the Mo surface on Raney-Ni did not enhance the OER activity

and the lack of redox peaks on the Mo/Raney sample indicate that the Mo surface may be

completely passivating the redox activity of the Raney-Ni substrate. Furthermore, the Fe/Raney-

Ni sample also showed muted Ni2+/3+

redox peaks indicating the presence of an iron-rich surface.

It has been previously reported that iron-rich surfaces do not exhibit the OER enhancement

95

typical of mixed Ni-Fe systems.[37] Finally, it should be noted that the redox peaks observed on

the Co/Raney sample occur at a much lower potential than the redox peaks characteristic of the

Ni2+/3+

transitions. The redox peak potentials observed on the Co/Raney surface are very similar

to those reported by Berlinguette et al.[38] for the Co2+/3+

transition. Figure 2 also clearly shows

enhancement in OER activity by addition of Fe or Co to the Raney-Ni substrate. These results

were conducted on unheated samples to avoid the formation of crystalline spinel-phase materials.

Figure 2 – CV data (20mV/sec) showing Ni2+/3+ redox peaks & OER activity from Raney-Ni-supported

samples. 250μg/cm2 catalyst loading on GC RDE substrate. Tested in O2-purged 0.1M KOH at room temp.

(~23C).

3.3.2 – Investigations of Heat-Treatment Part 1: Annealing vs Calcination of Ni-Fe:

Following identification of the OER enhancement from Fe & Co addition to the Raney-

Ni support, preliminary MMO composition optimization studies were conducted on carbon-

supported MMO films. Carbon-supported MMOs were studied because the standard IM

synthesis and evaluation on RDE provided a rapid screening platform. Nickel-rich samples with

9:1 Ni:Fe and 8:1:1 Ni:Fe:Co atomic ratios exhibited maximum OER activity of the carbon-

96

supported materials screened for this study. It should be noted that the MMO films could

certainly be further optimized for binary and ternary Ni-Fe-(Co) composition on the Raney-Ni

support. Subsequent studies were conducted to evaluate the effects of heat-treatment on the

Raney-Ni and C-supported catalyst samples.

Figure 3 – a) CV data (20 mV/sec) showing HT-effects of Ni-Fe on Raney-Ni support. 0.25 mg/cm2

catalyst loading, tested in O2-purged 0.1M KOH at room temp. (~23°C). b) XRD data from Ni-Fe/PANI-

Raney-Ni Sample after heat-treatments at 400°C under Ar or air forming gas.

Figure 3 shows the changes in electrochemical response and the XRD characterization for

the (9:1 at. ratio) Ni-Fe/PANI-Raney sample after annealing in argon or calcination in air at

400°C. Figure 3 (left) shows that the Ni-Fe sample annealed under argon retains a relatively high

Ni2+/3+

charge capacity and exhibits significantly increased OER activity compared to the

unheated sample. In particular, the on-set of the OER for the annealed Ni-Fe sample appears to

be shifted to lower over-potential by 30-40 mV compared to the unheated sample. In contrast,

calcination of the Ni-Fe sample under air results in a significantly decreased Ni2+/3+

redox peak

charge compared to the unheated sample. The calcined Ni-Fe sample also exhibits OER activity

similar to the bare Raney-Ni substrate. Figure 2b shows that the unheated sample contains

97

numerous Fe-oxide impurities. It has been previously noted that phase-segregated Fe-oxides do

not increase the OER activity of Ni-Fe samples[37] and thus this helps to explain the decreased

OER activity of the unheated sample compared to the bare Raney-Ni support. The primary

difference between the crystal structure of the annealed and calcined Ni-Fe samples in Figure 3

(right) is that the calcined sample (with lower OER activity) exhibits a higher ratio of NiO to

metallic Ni, based on the increased relative intensity of the peaks – particularly those at ~37°

(NiO) and ~52° (Ni). While an investigation of the crystal structure under in-situ operating

conditions would obviously provide a more definitive analysis of the active crystal phases, these

ex-situ results indicate that calcination of Ni-Fe increases the content of the crystalline oxide

phase which decreases the OER activity compared to samples annealed under non-oxidizing

argon atmosphere. Presumably, the annealing allows for the diffusion of Fe into the Ni structure

without the formation of the less-active oxide phase. Furthermore, it is possible that the unheated

samples contained a significant mass-fraction of inactive PANI or aniline oligomers. The TGA

results (ESI Figures S4 & S5) show a mass-loss of ~15% for both Ni-Fe/PANI-Raney and Ni-Fe-

Co/PANI-Raney catalyst samples. The loss of 15% of the as-synthesized mass presumably acts

to clean the catalyst by removing the inactive bare PANI or partially-reacted aniline oligomers

which were not removed during the acetone washing procedure. The remaining PANI is likely

protected by the MMO film and thus less exposed to thermal degradation. As noted above, in

addition to the cleaning effects, the heating may facilitate diffusion of Fe from the MMO surface

film into the Ni(oxide) lattice of the Raney-Ni substrate. The XRD analysis of the bare Raney-Ni

substrate exhibits both Ni and NiO crystal phases and can be seen in ESI Figure S6. It has been

noted by Song et al.[45] that incorporation of Fe3+

into the Ni(OH)2 lattice increases the average

lattice spacing – even allowing large anions to intercalate into the resulting “layered double

98

hydroxide” (LDH). Recently literature has reported Ni-Fe LDH catalysts with extremely high

OER mass-activity.[26, 27, 45] To our knowledge, the highest mass-activity to date, reported by

Song et al.[45] has resulted from exfoliation of these Ni-Fe LDH materials to form sub-

nanometer films of LDH which were described as 2D-networks of MO6 clusters. This exfoliated

LDH catalyst material has shown unprecedented mass-activity on the order of 300-400 mA/mg

at η = 320 mV, although this was in 1M KOH, as opposed to the 0.1M KOH standard electrolyte

concentration used in most recent literature reports. In addition, these exfoliated LDH nano-

flakes were tested at relatively low mass loading (70 μg/cm2). Under the conditions reported by

Song et al., the utilization of the active sites likely approaches 100%. However, in a real

electrode it is very unlikely that the mass-activity of the exfoliated LDH will scale linearly – i.e.

active-site utilization will be much lower once the exfoliated LDH is stacked in a thick film and

the apparent mass-activity will not be maintained above a certain mass-loading threshold where

active-site utilization remains near 100%. Thus, although the mass-activity reported by Song et

al. acts as an excellent guide to show the inherent OER activity of LDH catalysts, it should be

emphasized that optimization of catalyst morphology will be paramount to leverage the high

mass-activity of this type of catalyst. Thus, the MMO-film on composite PANI/Raney-Ni

support reported here provides an example of a novel morphology which may allow high active-

site utilization at higher catalyst loading on a real electrode. As a basis for comparison, Table 1

reports the mass-activity of the leading perovskites, PGM and MMO catalysts under nearly

identical testing conditions. Table 1 shows that while the perovskites reported by Shao-Horn et

al.[21] and Hardin et al.[24] exhibit mass-activity approaching the Ir PGM standard[46] at high

over-potential, the MMO samples (Ni-Fe LDHs) exhibit much higher mass-activity at low over-

potential resulting from the very low on-set potential for OER on these catalysts.

99

Table 1

Comparison of recently reported OER activities under identical conditions:

0.1M KOH @ 25°C

Catalyst Sample size (nm) CL LD

(mg/cm2)

MA

(mA/mg)

@ 1.55 ViR-comp.

Reference

BSCF

Perovskites

(Ba,Sr,Co,Fe,O)

micrometer 0.25 20 21

La(Co/Mn)O3

on N-doped-C 20 - 50 0.05 30 24

IrO2

nanoparticles 6 0.05 40 46

Ni-Fe LDH 20 x 500

plates 0.07 70 45

Ni-Fe LDH / CNT 5 x 50 plates 0.2 80 26

Ni-Fe LDH / CQD 5 x 50 plates 0.2 100 27

Ni-Fe-Co /

PANI-Raney 2-5 0.25 110 This work

These studies of intrinsic OER activity should be understood as guideposts for the

development of functional OER catalysts. For alkaline water electrolysis to be an effective

modular energy storage technology, commercial devices need to operate in the range of 100-

1000 mA/cm2 to effectively stand out from photo-electro-chemical water splitting systems which

operate at current densities of ~10 mA/cm2 under maximum illumination. Thus, as noted above,

the optimization of catalyst morphology is a primary challenge due to the high catalyst loading

(in the range of mg/cm2) typical for full-cell water electrolysis electrodes.

100

3.3.3 –Heat-Treatment Study Part 2: Annealing vs Calcination of Ni-Fe-Co:

The results of heat-treatment studies on the binary Ni-Fe MMO film shown in Figure 3

indicate that annealing under argon provides increased OER activity compared to the unheated or

calcined samples. Thus, a similar study was conducted on the (8:1:1 at. ratio) Ni:Fe:Co MMO

film on the composite PANI/Raney-Ni support. Figure 4 shows the electrochemical response and

XRD characterization results for the Ni:Fe:Co/PANI-Raney sample.

Figure 4 – a) CV data (20 mV/sec) showing HT-effects of Ni-Fe-Co on Raney-Ni support. 0.25 mg/cm2

catalyst loading, tested in O2-purged 0.1M KOH at room temp. (~23°C). b) XRD data from Ni-Fe-Co/PANI-

Raney Sample after heat-treatments at 400°C under Ar or air forming gas.

Figure 4 (left) indicates that the ternary Ni-Fe-Co MMO does not exhibit decreased OER

activity resulting from calcination as did the binary Ni-Fe sample. In fact, the calcined sample

exhibits the largest Ni2+/3+

charge and the OER activity is identical to the Ni-Fe-Co sample

annealed under argon. Figure 4 (right) shows that the unheated sample contains a distinctive

Ni(OH)2 component indicated by the peaks at 34° and 60.6° and the shoulder at ~72°. The

Ni(OH)2 component is not observed after annealing or calcination. However, unlike the iron

oxide impurities in the unheated Ni-Fe sample which decreased the OER activity relative to the

Raney-Ni substrate, the Ni(OH)2 phase in the Ni-Fe-Co sample exhibits OER activity higher than

101

the Raney-Ni substrate. Also, as with the Ni-Fe sample, calcination in air increases the NiO

content relative to the unheated or annealed sample. The intensity of the NiO peak at 37°

compared to the intensity of the metallic Ni peak at 52° indicates that the relative amount of

NiO:Ni is greater in the calcined sample than in the annealed sample. While the XRD data could

provide a more quantitative analysis of the mass ratios of the two phases with the use of an

internal standard, this analysis may not be descriptive of the true MMO surface due to diffraction

from the Raney-Ni substrate. Thus, the XRD data from the Ni-Fe-Co MMO samples provides a

more qualitative result, indicating that the increased metal oxide content of the calcined Ni-Fe-

Co MMO does not decrease the OER activity of the ternary sample as it did for the binary Ni-Fe

MMO. However, quantitative analysis of the crystallite sizes (xs) of the various phases present in

each sample is possible from the XRD data. Table 2 presents the results of profile fitting and the

average xs calculated from the FWHM of each peak using the Scherrer equation. Table 2

presents the results of profile fitting and the average xs calculated from the FWHM of each peak

using the Scherrer equation. Table 2 shows that for the binary Ni-Fe samples, heating under

argon does not increase the Ni xs but calcination in air does increase the Ni xs from ~5 nm to 6-

10 nm. For the ternary Ni-Fe-Co sample, the Ni xs increases from 5 nm (unheated) to 8 nm

(annealed) and 6-10 nm (calcined), thus indicating that calcination of the ternary Ni-Fe-Co

appears to increase the OER “specific-activity” (wrt physical surface area) to compensate for the

loss of physical surface area inferred by the particle growth. For all samples the NiO xs is in the

range of 2-3 nm.

102

Table 2 – Profile-fitting Results from XRD Analysis:

Sample crystallite size (nm)

Ni NiO Ni(OH)2

Raney-Ni 5 2 -

Ni-Fe/R - unHT 5 2-3 -

Ni-Fe/R - HT-Argon 5 2-3 -

Ni-Fe/R - HT-Air 6-10 3 -

Ni-Fe-Co/R - unHT 5 - 3

Ni-Fe-Co/R - HT-Argon 8 2-3 -

Ni-Fe-Co/R - HT-Air 6-10 3 -

3.3.4 –Heat-Treatment Study Part 3: Annealing C-supported MMOs:

It should be noted that determination of the local structure of the active MMO surface is

not particularly clear from XRD results where diffraction from the Raney-Ni substrate may be

the dominant contribution to the observed response. Thus, in an effort to more clearly identify

the phase changes in the OER-active MMO surface, we subjected MMO samples on carbon

supports to an annealing heat treatment. The calcination heat treatment was not studied using

carbon supported MMO catalysts due to the issue of carbon degradation in an oxidizing

environment at elevated temperature.

103

Figure 5 – Top: CV data (20 mV/sec) showing HT effects of C-supported Ni-Fe & Ni-Fe-Co samples.

Both “HT” samples were annealed at 400°C under argon for 30 min. Bottom, XRD data from Left: Ni-

Fe/C sample, Right: Ni-Fe-Co/C sample, after annealing heat-treatment at 400°C under argon forming

gas. Profile fitting indicates 2-3 nm crystallite size for the Ni(OH)2 phase in the unheated sample and 5

nm crystallite size for the NiO phase in the heated sample.

Figure 5 (top) shows that both the binary Ni-Fe and ternary Ni-Fe-Co MMO films on carbon

supports exhibited a significant decrease in the redox peak charge upon annealing, however at operating

voltages (1.7-1.8 V vs. RHE), the OER activity of the annealed samples was very similar to the unheated

samples. Figure 5 (bottom) shows the XRD results for the Ni-Fe sample (left) and the Ni-Fe-Co sample

104

(right). For both the Ni-Fe and Ni-Fe-Co samples on carbon supports, the XRD analysis shows a clear

transition from Ni(OH)2 phase with a crystallite size of 2-3 nm to NiO phase with crystallite size of ~5 nm

after annealing in argon. Finally, the inset of Figure 4a examines the Ni2+/3+

redox peaks of the heated

samples. The inset shows that the on-set of OER is nearly the same, but the redox peak potentials of the

Ni-Fe sample are observed at higher potentials than for the ternary Ni-Fe-Co sample. This observation is

in agreement with the observed shifts in peak potentials which will be discussed in more detail later.

3.3.5 – Comparison of Steady-State OER Activity:

Figure 6 – Steady-State OER chronoamperometry results: 20 mV steps with 1 min hold/step.

0.25 mg/cm2 catalyst loading, tested in O2-purged 0.1M KOH at room temp. (~23°C)

The OER activity results reported above in Table 1 are taken from the data shown in

Figure 6 (left). Although many recent reports evaluating the fundamental properties of OER

catalysts chose to report the mass-activity at over-potentials of 350 or 400 mV, we chose to

evaluate OER performance only up to 1.8 V (vs. RHE) applied potential, because this translates

to full-cell operation below 2 V assuming 100-200 mV cathodic overpotential from the HER

105

catalyst. Figure 6 (left) shows the clear increase in OER activity by adding Ni-Fe and Ni-Fe-Co

MMO films to the Raney-Ni substrate. This increased activity is evident both from the raw data

and iR-comp. data. It should be noted that all catalysts reported in Figure 5a exhibited ~36 Ω

high-frequency resistance (HFR) evaluated using potential-scan EIS at 50 mV intervals from

1.55 – 1.7 V vs. RHE (20 kHz – 100 mHz with 10 mV amplitude). Thus the same HFR value

was used to calculate the iR-compensated over-potential for all samples and the observed

increase in iR-comp. activity of the MMO/PANI-Raney samples is not due to differences in

conductivity of the catalyst layers, but is representative of a true increase in OER activity. In

addition to increasing the OER activity beyond that of the Raney-Ni substrate, the addition of the

MMO films produces catalyst layers which exhibit increased performance compared to an Ir

black standard provided by Proton On-Site. Thus, the MMO catalysts presented here show

significant improvement over a commercial PGM standard catalyst. Furthermore, Figure 6

(right) shows the stability of the ternary Ni-Fe-Co/PANI-Raney catalyst during 1 hour of

continuous operation at 1.7 V applied potential (1.54 V iR-comp.). Although the data is

somewhat noisy due to occasional bubble formation on the tip of the working electrode, the

durability of this sample is apparent. During the course of the entire hour the average current was

18.2 mA/cm2 with a standard deviation of 0.07 mA/cm

2 which translates to an average mass

activity of greater than 70 mA/mg over the course of the 1 hour test, with a loss of less than 1%

after an reaching a stable steady-state at ~2 min.

106

3.3.6 – Observations of Charge-Transfer Effects and Discussion of Mechanism

It has been noted by Gong et al.[26] and Bell et al.[39] that the inclusion of Fe in Ni-

oxide catalysts appears to create a more disordered local structure, characteristic of the α-

Ni(OH)2 phase in the α-Ni(OH)2 γ-NiOOH transition. In addition to observation of Raman

features indicating a more disordered surface structure of Ni-oxide layers upon inclusion of Fe,

Bell et al.[39] noted that the average oxidation state of Ni was lower in films with Fe than pure

Ni-oxide films – i.e. inclusion of Fe into the Ni-oxide film increases the potential at which the

Ni2+/3+

transition occurs. Furthermore, the effects of Co on the Ni redox peaks in co-precipitated

Ni-Co-oxide films has been noted by Corrigan et al.[47] In particular they noted a cathodic shift

of the Ni2+/3+

redox features in the presence of Co.

Figure 7 shows the Ni2+/3+

peaks for MMO films on carbon support for which there is no

contribution to the redox signal from the Raney-Ni support used in the previously discussed

catalyst samples. The results indicate that Co facilitates oxidation of Ni2+

to Ni3+

at lower

potentials, effectively presenting an electron withdrawing effect. In contrast, Fe appears to

stabilize the Ni2+

oxidation state, presumably via an electron donation to the Ni2+

sites. It is

interesting to note that addition of Co to the Ni-oxide film does not appear to enhance the OER

activity, however both samples containing Fe exhibit significantly enhanced OER activity.

Boettcher et al.[32] noted that while Fe does increase the conductivity of the Ni-oxide film, this

effect cannot entirely account for the enhancement of OER activity on Ni-Fe surfaces. Thus, in

light of the recent DFT results by Bell et al.[40] it appears that this data supports their assertion

that Fe is the active site. While FeOOH alone is not sufficiently conductive to act as a catalyst,

Fe3+

sites in a sufficiently conductive host matrix (Ni- or Co-oxide) have been shown to exhibit

ideal binding energy with –OH and –OOH intermediates. Thus, the further OER enhancement of

107

the Ni-Fe-Co catalyst is likely due to “activation” of the conductive NiOOH phase (compared to

the non-conductive Ni(OH)2 phase) at lower over-potential due to charge-transfer effects from

the Co component.

Figure 7 – CV data (20 mV/sec) showing Ni2+/3+ redox peaks & OER activity from C-supported samples

used for XAS analysis

3.3.7 – Observations of Charge-Transfer Effects from in-situ XAS Analysis:

In an effort to further examine the charge-transfer effects in MMOs, in-situ XAS analysis

was conducted on the C-supported MMO samples shown in Figure 7. Analysis of the high

surface-area C-supported samples ensured that the observed XAS response can be attributed

predominantly to the surface-active metal sites, with negligible contribution from sub-surface

metal. Furthermore, high-quality XAS data was collected at potentials above 1.4 V (i.e. in the

OER potential region). The XAS data shows distinct differences in the potential-dependence of

the Ni K-edge energy from the various MMO samples. In contrast, no shift in edge-energy is

108

observed from the Fe K-edge and the Co K-edge shows only a minor shift in the edge-energy

with changes in applied potential. Careful analysis of the potential at which the Ni K-edge

energy shift occurs on the various samples indicates that the Fe & Co components exhibit

charge-transfer effects on the Ni component, with Fe apparently providing an electron-donating

effect (thus stabilizing Ni in a lower average oxidation state), while Co appears to exhibit an

electron-withdrawing effect (thereby facilitating a higher average Ni oxidation state).

Figure 8 shows Fe K-edge data from the ternary Ni-Fe-Co sample at various applied

potentials. From both the XANES & FT-EXAFS analysis, it is clear that the Fe does not appear

to undergo any change in the oxidation state or local geometry at potentials up to 1.45 V. In

addition, the general shape of the XANES features as well as the metal-metal and metal-oxide

scattering peaks from FT-EXAFS plot, very closely match the features of a Fe2O3 standard,

suggesting that the Fe maintains an oxidation state of 3+ under the conditions tested. Previous

reports have also observed that Fe exists in the 3+ oxidation state under similar conditions, with

the onset of Fe oxidation to a mixed-valence Fe3+/4+

occurring above 1.55 V.[26, 40]

109

Figure 8 – Fe K-edge XAS results. Data collected from ternary Ni-Fe-Co(oxide)/C. Left:

XANES data most closely matches a Fe(III) oxide standard. Right: FT-EXAFS plot shows very

similar Fe-O scattering at ~1.5 Å and Fe-Fe scattering at ~2.8 Å (without phase correction)

indicating that the sample exhibits similar local geometry to the Fe(III) oxide standard.

Figure 9 – Co K-edge XAS results. Data collected from ternary Ni-Fe-Co(oxide)/C. Left:

XANES data most closely matches a Co(III) oxide standard. Right: FT-EXAFS plot shows very

similar Co-O scattering at ~1.5 Å and Co-Co scattering at ~2.4 Å (without phase correction)

indicating that the sample exhibits similar local geometry to the Co(III) oxide standard.

For in-situ XAS analysis of the Fe and Co K-edges, the ternary Ni-Fe-Co/C sample was

examined and the results are shown in figure 8 (Fe K-edge) and figure 9 (Co K-edge). From the

XANES data in figure 9 it appears that the Co is partially oxidized, beginning at 1.3 V and to a

greater extent at 1.5 V. However, the shift in the Co K-edge energy (1-2 eV) is rather small

110

compared to the larger shifts in Ni-edge energy (4-5 eV). Furthermore, the Co-edge signal is

very similar to the XAS response of a Co2O3 standard, even up to 1.3 V. The XANES data in

figure 9 shows a strong qualitative similarity with the Co2O3 standard, indicating a similar

oxidation state. In addition, the FT-EXAFS analysis of the sample was compared to the Co2O3

standard. The sample shows very similar Co-O scattering at ~1.5 Å and Co-metal scattering at

~2.4 Å. However, the Co K-edge data exhibits a more pronounced shoulder at ~1 Å and a

significant decrease in the intensity of the scattering peak at ~3 Å, compared to the Co2O3

standard. Both of these features indicate the presence of a mixed-phase cobalt-oxide in the

sample. For example, the thin-film of cobalt in the sample may exhibit more CoOOH character

than Co2O3, therefore accounting for the tightly bound oxide characteristic of the shoulder at ~1

Å and the Co-Co scattering characteristic of the peak at ~3 Å. The data at 1.5 V did not produce

coherent EXAFS fitting results, most likely due to the mixed oxidation state (mostly 3+ with

some 4+) of the Co at this potential.

111

112

Figure 10 – Ni K-edge Data collected from various samples. 1st row: Ni-Ox/C, 2

nd row: Ni-Fe-

Ox/C, 3rd

row: Ni-Co-Ox/C, 4th

row: Ni-Fe-Co-Ox/C. Left: XANES & Right: FT-EXAFS fitting

results.

The Ni K-edge was studied in the most detail for this research. Figure 10 shows the Ni K-

edge XANES data and FT-EXAFS analysis results for all four samples: Ni-(Ox)/C, Ni-Co-

(Ox)/C, Ni-Fe-(Ox)/C and Ni-Fe-Co-(Ox)/C. To focus on the most relevant data, figure 11

compares the Ni K-edge XANES data from each sample at 1.45 V applied potential, where the

difference in edge-energy of the various samples is most obvious. Figure 11 also provides a

depiction of the relative shift in the Ni K-edge energy of each sample as a function of applied

potential. The results support the CV data shown in Figure 7 which indicate that Fe-stabilizes Ni

in the 2+ oxidation state, while Co facilitates oxidation to Ni 3+ at lower applied potential.

The many graphs in figure 10 show the differences in potential-dependent shifts in Ni

oxidation state (XANES data, on the left) and the relative changes in the geometry of the metal

and oxide neighbors (EXAFS fitting results, on the right). Data was collected at 50 mV intervals

between 1 – 1.6 V for all samples, however only the most relevant data points are shown in the

graphs in figure 10 for clarity. In particular, for the XANES data, the best-quality low-voltage

data point is shown first, followed by points at potentials immediately below and above the

observed edge-shift. For the FT-EXAFS analysis, the low-voltage data is also omitted such that

the FT-EXAFS analysis emphasizes the redox transition and subsequent oxidation for each

sample. For the Ni-(Ox)/C sample (1st row), an apparent change in oxidation state begins at 1.45

V and appears stable by 1.5 – 1.55 V. This corresponds with the Ni2+

NI3+

transition which is

well-documented in the literature. The FT-EXAFS analysis for the Ni-(Ox)/C sample shows a

clear contraction of the Ni-O and Ni-Ni nearest-neighbors upon oxidation (peaks at ~1.7 1.4

Å and 2.7 2.4 Å, respectively).

113

For the Ni-Co-(Ox)/C sample (2nd

row), the change in the oxidation state begins by 1.35

V (100 mV lower than for pure Ni-Ox) and appears stable by .45 – 1.5 V. The FT-EXAFS

analysis of the Ni-Co-(Ox)/C sample shows a less intense decrease in the Ni-O and Ni-metal

scattering peaks (i.e. less bond contractions), but a significant increase in the intensity ratio of the

Ni-metal peak (~2.4 Å) to the Ni-O peak (~1.4 Å). This change in the relative intensity of the

peaks from 1.16:1 Ni-O:Ni-M at 1.2 V to 0.82:1 at 1.5 V indicates an increase in the

metal:metal-oxide ratio of the nickel. This is an interesting point because the Ni-Co-(Ox)/C

sample is the only sample which exhibits such a dramatic increase in the metallic character of the

nickel content. This may indicate that in addition to exhibiting an electron-withdrawing effect on

nickel, the cobalt component may in addition act as a more oxophilic site, effectively decreasing

the average Ni-O coordination number. This hypothesis is supported by the EXAFS fitting

values presented in table 3, which contain the fitting results of the Ni-(Ox)/C sample as well as

the results from the binary Ni-Co-(Ox)/C and Ni-Fe-(Ox)/C samples. For the binary Ni-Co-

(Ox)/C sample, the average coordination number of the Ni-O scattering path decreases more than

in either the Ni-(Ox)/C or Ni-Fe-(Ox)/C samples. However, the Ni-M coordination number also

decreases dramatically with increased potential for the Ni-Co-(Ox)/C sample, which cannot be

readily explained.

For the Ni-Fe-(Ox)/C (Figure 10, 3rd

row), the XANES data shows that the shift in the Ni

K-edge doesn’t occur until 1.5 V (50 mV higher than for monometallic Ni-Ox and 150 mV

higher than for Ni-Co-oxide). The edge energy appears stable from 1.5 – 1.55 V, however the

FT-EXAFS analysis indicates slightly more contraction of the Ni-metal scattering peak (2.7

2.3 Å) and increase in the relative ratio of Ni-O to Ni-M peaks from a ratio of ~1.08:1 Ni-O:Ni-

M at 1.4 V to 1.39:1 at 1.55 V. It is interesting to note that the cobalt appears to exhibit greater

114

electronegativity and oxophilicity than the iron when comparing the results of the binary samples

– as predicted by the standard periodic trends in electronegativity.

Finally, for the ternary Ni-Fe-Co-(Ox)/C sample (Figure 10, 4th

row), the XANES data

appears nearly identical to the XANES data for the monometallic Ni-(Ox)/C sample, with a shift

in the Ni-edge beginning at 1.45 V and stabilizing by 1.5 through 1.55 V. In contrast, the FT-

EXAFS analysis shows that the ternary Ni-Fe-Co-(Ox) maintains a nearly 1:1 ratio of Ni-O to

Ni-M scattering peaks after complete oxidation to Ni3+

(i.e. beyond 1.45 V). The mono-metallic

Ni-(Ox)/C sample maintains a higher ratio of Ni-O to Ni-Ni scattering once in the Ni3+

state.

115

Table 3 – EXAFS Fitting Results

Ni-oxide

sample Ni-Ni path Ni-O path

Potential

(mV) N R (Å) σ

2

(Å)×10-3

N R (Å) σ2

(Å)

1100 8.4±1.9 3.11±0.01 7±2 6.8±0.7 2.07±0.01 5±2

1350 8.0±1.6 3.11±0.01 6±1 6.9±0.7 2.07±0.01 5±2

1400 7.6±2.0 3.10±0.01 6±2 6.9±0.9 2.05±0.01 5±2

Ni-Co

sample Ni-M (M=Ni or Co) path Ni-O path

Potential

(mV) N R (Å) σ

2

(Å)×10-3

N R (Å) σ2

(Å)

1000 9.8±2.4 3.04±0.01 8±2 6.7±0.8 2.05±0.01 4±1

1200 9.3±1.9 3.03±0.01 9±2 7.0±0.8 2.04±0.01 4±1

1450 4.5±1.0 2.80±0.01 2±1 6.0±1.0 1.88±0.01 3±2

1500 4.3±1.1 2.80±0.01 1±1 6.0±1.0 1.88±0.01 3±2

Ni-Fe

sample Ni-M (M=Ni or Fe) path Ni-O path

Potential

(mV) N R (Å) σ

2

(Å)×10-3

N R (Å) σ2

(Å)

1100 9.4±2.4 3.11±0.01 7±2 6.7±1.0 2.07±0.01 5±2

1200 9.7±2.7 3.10±0.01 8±2 6.6±1.0 2.06±0.01 5±2

1300 8.9±2.8 3.11±0.01 7±2 6.4±0.9 2.07±0.02 4±2

1350 8.9±2.8 3.11±0.01 7±2 6.4±0.9 2.07±0.02 4±2

1400 9.8±2.8 3.10±0.01 8±2 6.4±0.9 2.06±0.01 4±2

1450 6.0±2.1 3.11±0.01 7±3 6.4±1.1 2.07±0.02 6±3

The XANES data and FT-EXAFS analysis of the ternary sample indicate that both the

charge-transfer effects and the relative oxophilicity of the Fe vs. Co components tend to cancel

116

out each other’s influence on the electronic and geometric state of Ni. In an effort to summarize

the XAS results from the Ni-edge data, figure 11 compares the XANES data for each sample at

1.45 V. This comparison emphasizes the charge-donating effects of the Fe in Ni-Fe sample

(lower Ni K-edge energy) and the charge-withdrawing effects of Co in the Ni-Co sample (higher

Ni K-edge energy). In addition, the ternary and monometallic samples appear identical indicating

that the effects of the Fe and Co components cancel each other in the ternary sample. The graph

on the right in figure 11 shows the shift in Ni K-edge energy relative to the literature value of

8113 eV. The dramatic differences in edge-shifts are most apparent at 1.45 and 1.5 V, where the

Co and Fe components of the binary samples tend to shift the Ni edge energy positive and

negative of the monometallic & ternary samples by +/- 1 eV.

Figure 11 – Summary of in-situ XAS analysis showing charge-transfer effects of Fe & Co on Ni

K-edge energy. Left: lower Ni K-edge energy of Ni-Fe sample indicates lower average oxidation

state while higher Ni K-edge energy of Ni-Co sample indicates higher average oxidation state.

Right: summary of relative shifts in Ni K-edge energy (relative to 8113 eV for metallic Ni)

comparing the various MTMO films.

117

3.4 Conclusions:

Binary and ternary mixed-metal-oxide (MMO) films on a commercial Raney-Ni support

were evaluated for alkaline OER activity in 0.1M KOH at room temperature (~23°C).

Electrochemical testing in RDE configuration showed that a ternary Ni-Fe-Co-oxide (with 8:1:1

atomic ratio of metals) catalyst exhibited the best OER activity. The mass-activity was 110

mA/mg at 1.55 V(iR-comp.). In addition, the ternary Ni-Fe-Co catalyst showed less than 1% loss

of OER activity at 1.7 V applied potential (1.54 V iR-comp.) during one hour of continuous

operation. Analysis of the heat-treatment conditions reveals that the ternary Ni-Fe-Co sample is

less sensitive to loss of activity from exposure to oxidizing conditions than was the binary Ni-Fe

catalyst which suffered significant de-activation due to an oxidizing heat-treatment. The

tolerance to oxidizing conditions may in part be rationalized by in-situ XAS studies which

indicate that MMO films which contained cobalt exhibited less oxidation of the nickel-

component under the oxidizing high-voltage conditions of the OER, presumably due to a higher

oxophilicity of cobalt than nickel. In addition, the high OER activity of the ternary Ni-Fe-Co

sample can be rationalized in light of recent in-situ XAS and conductivity measurements of Ni-

Fe and Co-Fe films. While previous trends have shown higher OER activity on Ni than on Fe, it

has very recently been proposed that iron impurities in the electrolyte often produce Ni-Fe

catalysts for which Fe is the true active site.[48] In fact, recent DFT calculations have proposed

that Fe3+

sites provide ideal binding energy for the -OH and -OOH OER intermediates, whereas

Ni3+

sites bind these adsorbates too weakly.[48] The results of this study indicate the most active

catalysts are those which contain Fe (Ni-Fe & Ni-Fe-Co). The further enhancement of OER

activity on the ternary Ni-Fe-Co sample compared to the binary Ni-Fe sample may be due to

multiple factors. Primarily, the charge-transfer from the Co component facilitates oxidation of Ni

118

to the more conductive NiOOH phase at lower over-potential, thus effectively “activating” the Fe

sites which are otherwise dormant in the non-conductive Ni(OH)2 phase. In addition, EXAFS

fitting of the in-situ XAS data shows a decrease in the Ni-metal and Ni-oxide bond distances in

samples containing cobalt. Bell et al.[40] showed that Fe substitutes into the Ni-oxide lattice in

Ni1-xFexOOH (for samples with x<0.25). They further indicated that the NiOOH local geometry

was not noticeable affected by the replacement of Fe in the Ni sites and that the resulting Fe-

metal and Fe-O bond distances were ~3% smaller than in a pure FeOOH lattice. Thus our results

showing that the presence of Co in the MMO films further decreases the metal-metal and metal-

oxide bond distances may indicate that the shrinking of the local geometry may further increase

the inherent OER activity on the Fe active-sites.

3.5 Acknowledgements:

This work funded by the Advanced Research Projects Agency – Energy (ARPA-E), U.S.

Department of Energy, under Award No. DE-AR0000121. Use of the National Synchrotron

Light Source (beamline X3B), Brookhaven National Lab was made possible by the Center for

Synchrotron Biosciences grant, P30-EB-009998, from the National Institute of Biomedical

Imaging and Bioengineering (NBIB). Support from beamline personnel Dr. Erik Farquhar

(X3B) is gratefully acknowledged.

119

REFERENCES:

1. Eberle, U., B. Muller, and R. von Helmolt, Fuel cell electric vehicles and hydrogen

infrastructure: status 2012. Energy & Environmental Science, 2012. 5(10): p. 8780-8798.

2. Brown, J.E., C.N. Hendry, and P. Harborne, An emerging market in fuel cells?

Residential combined heat and power in four countries. Energy policy, 2007. 35(4): p.

2173-2186.

3. Lokurlu, A., et al., Fuel cells for mobile and stationary applications—cost analysis for

combined heat and power stations on the basis of fuel cells. International journal of

hydrogen energy, 2003. 28(7): p. 703-711.

4. Dodds, P.E., et al., Hydrogen and fuel cell technologies for heating: A review.

International Journal of Hydrogen Energy, 2015.

5. Cho, K.T., et al., High performance hydrogen/bromine redox flow battery for grid-scale

energy storage. Journal of The Electrochemical Society, 2012. 159(11): p. A1806-

A1815.

6. Braff, W.A., M.Z. Bazant, and C.R. Buie, Membrane-less hydrogen bromine flow

battery. Nat Commun, 2013. 4.

7. McCrory, C.C.L., et al., Benchmarking HER and OER Electrocatalysts for Solar Water

Splitting Devices. Journal of the American Chemical Society, 2015.

8. Pinaud, B.A., et al., Technical and economic feasibility of centralized facilities for solar

hydrogen production via photocatalysis and photoelectrochemistry. Energy &

Environmental Science, 2013. 6(7): p. 1983-2002.

9. Rodriguez, C.A., et al., Design and cost considerations for practical solar-hydrogen

generators. Energy & Environmental Science, 2014. 7(12): p. 3828-3835.

10. Merle, G., M. Wessling, and K. Nijmeijer, Anion exchange membranes for alkaline fuel

cells: A review. Journal of Membrane Science, 2011. 377(1–2): p. 1-35.

11. Varcoe, J.R. and R.C. Slade, Prospects for Alkaline Anion‐Exchange Membranes in Low

Temperature Fuel Cells. Fuel cells, 2005. 5(2): p. 187-200.

120

12. Varcoe, J.R., et al., Anion-exchange membranes in electrochemical energy systems.

Energy & Environmental Science, 2014. 7(10): p. 3135-3191.

13. Carmo, M., et al., A comprehensive review on PEM water electrolysis. international

journal of hydrogen energy, 2013. 38(12): p. 4901-4934.

14. McKone, J.R., et al., Ni–Mo Nanopowders for Efficient Electrochemical Hydrogen

Evolution. ACS Catalysis, 2012. 3(2): p. 166-169.

15. Bates, M.K., et al., Composite Ni / NiO-Cr2O3 Catalyst for Alkaline Hydrogen Evolution

Reaction. The Journal of Physical Chemistry C, 2015.

16. Gong, M., et al., Nanoscale nickel oxide/nickel heterostructures for active hydrogen

evolution electrocatalysis. Nat Commun, 2014. 5.

17. Danilovic, N., et al., Enhancing the Alkaline Hydrogen Evolution Reaction Activity

through the Bifunctionality of Ni(OH)2/Metal Catalysts. Angewandte Chemie

International Edition, 2012. 51(50): p. 12495-12498.

18. Kumar, M., et al., New ternary Fe, Co, and Mo mixed oxide electrocatalysts for oxygen

evolution. international journal of hydrogen energy, 2011. 36(15): p. 8831-8838.

19. Singh, A. and R. Singh, Effect of V substitution at B-site on the physicochemical and

electrocatalytic properties of spinel-type NiFe 2 O 4 towards O 2 evolution in alkaline

solutions. International Journal of Hydrogen Energy, 2010. 35(8): p. 3243-3248.

20. Zhu, Y., et al., SrNb0.1Co0.7Fe0.2O3−δ Perovskite as a Next-Generation

Electrocatalyst for Oxygen Evolution in Alkaline Solution. Angewandte Chemie

International Edition, 2015: p. n/a-n/a.

21. Grimaud, A., et al., Double perovskites as a family of highly active catalysts for oxygen

evolution in alkaline solution. Nat Commun, 2013. 4.

22. Risch, M., et al., La0.8Sr0.2MnO3−δ Decorated with Ba0.5Sr0.5Co0.8Fe0.2O3−δ: A

Bifunctional Surface for Oxygen Electrocatalysis with Enhanced Stability and Activity.

Journal of the American Chemical Society, 2014. 136(14): p. 5229-5232.

121

23. Suntivich, J., et al., A Perovskite Oxide Optimized for Oxygen Evolution Catalysis from

Molecular Orbital Principles. Science, 2011. 334(6061): p. 1383-1385.

24. Hardin, W.G., et al., Tuning the Electrocatalytic Activity of Perovskites through Active

Site Variation and Support Interactions. Chemistry of Materials, 2014. 26(11): p. 3368-

3376.

25. Trotochaud, L., et al., Solution-Cast Metal Oxide Thin Film Electrocatalysts for Oxygen

Evolution. Journal of the American Chemical Society, 2012. 134(41): p. 17253-17261.

26. Gong, M., et al., An Advanced Ni–Fe Layered Double Hydroxide Electrocatalyst for

Water Oxidation. Journal of the American Chemical Society, 2013. 135(23): p. 8452-

8455.

27. Tang, D., et al., Carbon Quantum Dot/NiFe Layered Double-Hydroxide Composite as a

Highly Efficient Electrocatalyst for Water Oxidation. ACS Applied Materials &

Interfaces, 2014. 6(10): p. 7918-7925.

28. Kjartansdóttir, C.K., L.P. Nielsen, and P. Møller, Development of durable and efficient

electrodes for large-scale alkaline water electrolysis. International Journal of Hydrogen

Energy, 2013. 38(20): p. 8221-8231.

29. Schiller, G., et al., High performance electrodes for an advanced intermittently operated

10-kW alkaline water electrolyzer. International Journal of Hydrogen Energy, 1998.

23(9): p. 761-765.

30. Ding, H., M. Wan, and Y. Wei, Controlling the diameter of polyaniline nanofibers by

adjusting the oxidant redox potential. Advanced Materials, 2007. 19(3): p. 465-469.

31. Subbaraman, R., et al., Trends in activity for the water electrolyser reactions on 3d

M(Ni,Co,Fe,Mn) hydr(oxy)oxide catalysts. Nat Mater, 2012. 11(6): p. 550-557.

32. Trotochaud, L., et al., Nickel–Iron Oxyhydroxide Oxygen-Evolution Electrocatalysts: The

Role of Intentional and Incidental Iron Incorporation. Journal of the American Chemical

Society, 2014. 136(18): p. 6744-6753.

33. Li, X., F.C. Walsh, and D. Pletcher, Nickel based electrocatalysts for oxygen evolution in

high current density, alkaline water electrolysers. Physical Chemistry Chemical Physics,

2011. 13(3): p. 1162-1167.

122

34. McCrory, C.C.L., et al., Benchmarking Heterogeneous Electrocatalysts for the Oxygen

Evolution Reaction. Journal of the American Chemical Society, 2013. 135(45): p. 16977-

16987.

35. Chen, J.Y., et al., Inverse spinel NiFeAlO 4 as a highly active oxygen evolution

electrocatalyst: Promotion of activity by a redox-inert metal ion. Energy &

Environmental Science, 2014. 7(4): p. 1382-1386.

36. Gerken, J.B., et al., A survey of diverse earth abundant oxygen evolution electrocatalysts

showing enhanced activity from Ni–Fe oxides containing a third metal. Energy &

Environmental Science, 2014. 7(7): p. 2376-2382.

37. Landon, J., et al., Spectroscopic Characterization of Mixed Fe–Ni Oxide Electrocatalysts

for the Oxygen Evolution Reaction in Alkaline Electrolytes. ACS Catalysis, 2012. 2(8): p.

1793-1801.

38. Smith, R.D., et al., Water oxidation catalysis: electrocatalytic response to metal

stoichiometry in amorphous metal oxide films containing iron, cobalt, and nickel. Journal

of the American Chemical Society, 2013. 135(31): p. 11580-11586.

39. Louie, M.W. and A.T. Bell, An investigation of thin-film Ni–Fe oxide catalysts for the

electrochemical evolution of oxygen. Journal of the American Chemical Society, 2013.

135(33): p. 12329-12337.

40. Friebel, D., et al., Identification of Highly Active Fe Sites in (Ni,Fe)OOH for

Electrocatalytic Water Splitting. Journal of the American Chemical Society, 2015.

137(3): p. 1305-1313.

41. Burke, M.S., et al., Cobalt–Iron (Oxy)hydroxide Oxygen Evolution Electrocatalysts: The

Role of Structure and Composition on Activity, Stability, and Mechanism. Journal of the

American Chemical Society, 2015. 137(10): p. 3638-3648.

42. Natan, M.J., et al., pH-sensitive nickel (II) hydroxide-based microelectrochemical

transistors. Journal of Physical Chemistry, 1987. 91(7): p. 1834-1842.

43. Kuai, L., et al., A Reliable Aerosol‐Spray‐Assisted Approach to Produce and Optimize

Amorphous Metal Oxide Catalysts for Electrochemical Water Splitting. Angewandte

Chemie, 2014. 126(29): p. 7677-7681.

123

44. Chan, K.-Y., et al., Supported mixed metal nanoparticles as electrocatalysts in low

temperature fuel cells. Journal of Materials Chemistry, 2004. 14(4): p. 505-516.

45. Song, F. and X. Hu, Exfoliation of layered double hydroxides for enhanced oxygen

evolution catalysis. Nature communications, 2014. 5.

46. Lee, Y., et al., Synthesis and Activities of Rutile IrO2 and RuO2 Nanoparticles for

Oxygen Evolution in Acid and Alkaline Solutions. The Journal of Physical Chemistry

Letters, 2012. 3(3): p. 399-404.

47. Corrigan, D.A. and R.M. Bendert, Effect of coprecipitated metal ions on the

electrochemistry of nickel hydroxide thin films: cyclic voltammetry in 1M KOH. Journal

of The Electrochemical Society, 1989. 136(3): p. 723-728.

48. Soloveichik, G.L., Battery technologies for large-scale stationary energy storage. Annual

review of chemical and biomolecular engineering, 2011. 2: p. 503-527.

124

Chapter 4 Development of Electrocatalysts for H2-Br2 Redox Flow Batteries

4.1 Introduction:

H2-Br2 redox flow batteries (HB-RFBs) present a viable energy storage solution to solve

the intermittency issue of clean energy production from solar and wind farms. RFBs in general

offer the advantages of modular and independently scalable energy and power. The energy can

be scaled by the size (volume) of the fuel tanks. The power density can be scaled by the size of

the electrodes in the stack. Thus, RFBs are ideal for tailoring the system to suit the needs of

specific large-scale energy applications. The two primary advantages of HB-RFBs include the

low cost of the HBr electrolyte as well as the extremely fast kinetics of both half-cell reactions

leading to >90% charge/discharge efficiency for this system. In addition to the fast kinetics, the

Br2/Br- reaction involves and outer-sphere heterogeneous electron transfer (OSHET) process,

which means that the reaction does not require an adsorbed intermediate. The half-reactions on

the bromine/bromide electrode are:

Br2 + 2e- 2 Br

- (Discharge)

2Br- Br2 + 2e

- (Charge)

The benefit of OSHET reactions is that they proceed readily in the presence of any

electronically conducting substrate, such that expensive platinum-group metal (PGM) electrodes

are not required to catalyze the reaction. As such, the bromine/bromide reactions can achieve

high current density from carbon-based electrodes. The current density is proportional to the

physical surface-area of the electrode and thus research efforts have investigated the use of

carbon nanotube arrays on carbon-paper electrodes to provide extremely high current density and

electrode stability at a very low cost.

125

The primary research challenge for the HB-RFB systems is related to the development of

high activity and stability hydrogen electrodes. Unlike the Br2/Br- reaction, the H2/H

+ reaction is

an inner-sphere heterogeneous electron transfer (ISHET) process. This means that the reaction

proceeds via a mechanism which involves an adsorbed intermediate. ISHET reactions typically

require electrocatalysts with very specific binding energies relative to the adsorbed

intermediate(s). The hydrogen oxidation reaction (HOR) and hydrogen evolution reaction (HER)

mechanistic steps are presented here:

2 Pt + H2 2 Pt-Hads (Chemical dissociative adsorption, “Tafel step”)

2Pt-Hads + 2H2O 2Pt + 2H3O+ + 2e

- (Electrochemical oxidation, “Volmer step”)

For the HOR, the reaction proceeds via the Tafel-Volmer path while for HER, the reverse

Volmer-Tafel path produces H2 gas from hydronium ions. It is the Tafel reaction mechanism

which presents the challenge in development of highly active hydrogen electrocatalysts. For

adsorption of the intermediate to occur, the surface energy of the catalysts must favor the

formation of a bond. This phenomenon has been well-documented and has led to the

development of “volcano curves” which plot the HER-HOR activity of various metals as a

function of their hydrogen bond strength. An example of a Volcano curve is presented in figure

1. In these studies, the metal-hydrogen bond strength is measured experimentally via enthalpy of

hydride formation or calculated via DFT methods. The volcano curves show that some metals

bind too weakly to hydrogen to form the M-Hads intermediate and other metals bind to strongly to

allow for the desorption of the M-Hads intermediate. It has long been observed that the PGMs

exhibit the optimum metal-hydrogen bond strength to allow a high turn-over frequency (TOF) of

HER & HOR and as such, achieve high current density with minimal over-potential. Therefore,

126

one limitation of the HB-RFBs is the requirement of PGMs to catalyze the hydrogen reactions. In

addition to the high cost of PGM catalysts, the hydrogen electrode also presents stability

challenges to achieve long-term operation required for grid-scale applications.

Figure 1: Sabatier-type “Volcano Plot” of the HER activity of various metals as a function of M-

H bond strength from Koper et al.[1] (Reproduced with permission from the Royal Society of

Chemistry).

The hydrogen electrode of H2-Br2 flow cells suffers from well-known stability issues due

to the halide adsorption on the catalyst surface and Pt dissolution during the “short-circuit”

conditions created by Br2 cross-over or hydrogen starvation. The halide adsorption effects have

been well documented for ORR applications in the literature.[2-5] While the halide “poisoning”

effects may be semi-reversible (i.e. the halides can be removed by the application of a

momentary large over-potential to the H2 electrode resulting in oxidation of adsorbed halides)

the more pressing issue is the apparent corrosion of Pt electrodes in the concentrated HBr/Br2

electrolyte. This corrosion effect has been observed to occur under H2-starvation conditions. The

presumed mechanism involves a shift in the open-circuit potential (OCP) of the H2-electrode.

127

Upon removal of the protecting H2 atmosphere and exposure to air, the OCP increases to a value

positive of the potential of zero charge (pzc) of the Pt catalyst. In this potential region, the Pt

surface is positively charged and becomes poisoned by Br- ions in the electrolyte. This poisoning

effect is due to the electrostatic interactions between the positively charged electrode surface and

the negatively charged adsorbate. These electrostatic poisoning effects are also known to occur

in the presence of other large oxyanions (SO42-

, PO43-

, etc.) as well as molecules with highly

polarizable free electron clouds such as CO.[4, 6] As the catalyst surface approaches saturation

by halide adsorbates, chemical oxidation of the halides can occur. The resulting diatomic halide

molecules produce a change in the chemical potential of the system – i.e. the system equilibrates

to match the equilibrium potential of Br2 (1.07 V). At this elevated voltage, the Pt surface is

electrochemically oxidized to the now acid-soluble PtO. A fundamental property of Pt is it’s

resistance to corrosion in acid. However, in the presence of oxidizing and acidic conditions (such

as with the use of “aqua regia” HCl & HNO3 mixtures) Pt will dissolve.

Even during normal cell operation, with hydrogen flowing, some amount of Br2-

crossover is inevitable. The presence of Br2 molecules approaching the Pt surface may cause a

transient voltage spike on the atomic scale (i.e. not over the entire electrode, but only for the Pt

atoms in contact with the Br2 molecules) and result in dissolution of the catalyst. Given these

challenges, the focus of this project was to identify bi-functional (HER-HOR) catalysts resistant

to Br- poisoning.

During the course of this project, an array of Ru & Rh-based catalysts were synthesized

and tested to include metals, metal-oxides, -sulfides and –selenides. Ru & Rh chalcogenides

were investigated due to the previously reported resistance to adsorbate poisoning on these

chalcogenide-type materials under oxygen-reduction conditions.[7] These catalyst materials were

128

screened for hydrogen evolution reaction and hydrogen oxidation reaction (HER-HOR) activity

in a rotating disk electrode (RDE) cell configuration, first in non-adsorbing HClO4 and then in

HBr. In addition to RDE testing, the most promising materials were screened in a 5cm2 H2-pump

configuration to evaluate the HER & HOR kinetics under operating cell conditions. RDE and H2-

pump testing indicated that most Ru-based materials exhibited moderate HER but no HOR

activity. However, a Ru-Nx sample did exhibit reversible HER-HOR activity in perchloric acid

(and KOH), but no HOR activity was observed in the strongly adsorbing HBr electrolyte.

Because Ru-based materials did not exhibit HOR activity in the HBr electrolyte they were not

pursued as viable catalysts for reversible hydrogen electrodes. Only the best RhS/C sample

exhibited reversible HER-HOR activity but with much slower reaction kinetics than the Pt/C

standard, thus severely limiting the charge-discharge efficiency when operated in full-cell

configuration. A 30% RhS/C catalyst heated to 800°C exhibited reversible hydrogen activity,

albeit with only ~50% of the target exchange current density (RhS i0 = 0.3 A/mg) measured in

H2-pump testing. In addition to the Ru & Rh-based catalysts, Pt & Ir nitrides were also

investigated. In particular, a 40 wt% Pt-Ir-Nx/C sample exhibited i0 well above the 0.6 A/mg

target for this project. This Pt-Ir- Nx catalyst was also subjected to stability testing in 1M HBr

with 1mM Br2. The Pt-Ir-Nx/C sample exhibited increased stability compared to a commercial

Pt/C catalyst which was immediately de-activated upon being subjected to the stability protocol.

The stability of the Pt-Ir-Nx/C catalyst was investigated in more detail using an RDE soak-testing

protocol. RDE studies of pure Ir/C catalyst showed that the Ir-component is active for both HER

& HOR, but Ir alone (like Pt) is not stable under the HBr/Br2 stability testing conditions.

129

4.2 Experimental:

A 46% Pt/C sample from Tanaka corp. was chosen as a commercial standard by which to

compare the reversible HER-HOR activity of all other samples. The metallic Rh/C & Ru/C

samples were synthesized via direct borohydride reduction of the Metal(III)Cl3 *xH2O salts (Alfa

Aesar, Ward Hill, MA) in the presence of Vulcan XC-72R carbon support (Cabot Corp.

Haverhill, MA). The RhSe/C & RuSe/C samples were synthesized via a modified borohydride

reduction in the presence of SeO2 pre-cursor. For the RhS & RuS materials, the metal salt

precursors were dissolved, refluxed overnight and then a solution of ammonium thiosulfate was

added in aliquots to produce the metal sulfide. All samples were subsequently subjected to

pyrolysis in a tube furnace at either 500°C or 800°C under an inert argon atmosphere. Finally,

the Ru-Nx, Pt-Nx & Pt-Ir-Nx materials were synthesized mixing aqueous solutions of the

appropriate metal chloride salts in the presence of 1,3 propylendiamine and Vulcan carbon

support, followed by filtering, washing, drying in a vacuum-oven and annealing at 700°C under

argon in a tube furnace.

For RDE studies, inks composed of 5 mL H2O, 4.95 mL 2-propanol and 50 μL of 5wt%

“Nafion” (perfluorosulfonic acid-PTFE co-polymer, Alfa Aesar, Ward Hill, MA) ionomer

dispersion were mixed with an appropriate amount of catalyst. The inks were sonicated for at

least 30 min before a 10 μL aliquot was deposited on the tip of a polished glassy carbon disk (5

mm diameter) to produce a loading of 50 μg(metal)/cm2. Electrochemical tests were conducted

with an Autolab (Ecochemie Inc., model-PGSTAT30) potentiostat/galvanostat. Tests were

conducted in a 500 mL jacketed 3 electrode cell at room temperature. Electrolytes were prepared

using concentrated 70% HClO4 (Veritas, double-distilled, GFS chemicals) or 48% HBr solutions

(Alfa Aesar, Ward Hill, MA). A silver/silver-chloride reference with saturated KCl filling

130

solution was used as the reference electrode and a home-made iridium counter electrode was

used to avoid deactivation of a Pt counter electrode. The Ir counter electrode consisted of a Ti-

foil substrate, loaded with ~3 mg/cm2 Ir black catalyst (from Proton On-Site, Wallingford, CT)

and attached to a copper wire using silver paint. The copper wire was protected from corrosion in

HBr by placing it in a modified glass pipette and the junction of the Cu-wire, glass pipette and

Ti-mesh was sealed with epoxy so that only the Ir black surface remained exposed to electrolyte.

The glassy carbon WE was rotated using an RDE setup from Pine Instruments. Rotation rates of

2500 rpm were sufficient to remove the H2(g) product from the surface of the WE and examine

the kinetics well into the HER region. All results were obtained after conditioning electrodes by

conducting 50 mV/sec CV scans (from -100 mV to +300 or +800 mV vs. RHE) for at least 20-30

scans, or until stable features were observed.

For H2-pump testing, experimental conditions simulated the experiments detailed by

Nguyen et al.[8] gas gas-diffusion electrodes (GDEs) were prepared by depositing 0.5

mg(metal)/cm2 onto carbon-paper diffusion layers (GDLs). Hydrophobic Toray GDLs were used

to mitigate catalyst flooding from the humidified gas fuel. The GDLs were hot-pressed onto a

commercial Nafion 212 membrane using a Carver heated press. The experimental catalysts

(RuS, RhS, Pt-Ir-Nx) were each tested against a gas diffusion electrode (GDE) with commercial

Pt/C used in the catalyst layer. This Pt/C GDE served as a combined counter & reference

electrode (C/RE) to analyze the HER & HOR polarization curves of the experimental catalysts.

The cell was operated at room temperature (~23°C) with 500 mL/min H2 at 100% relative

humidity and 18 psi back pressure on both electrodes. The HER was evaluated with the negative

terminal on the GDE with the experimental catalyst. The +/- terminals were switched to evaluate

HOR kinetics.

131

XRD characterization was conducted using a Rigaku Ultima IV XRD with a Cu Kα

source (λ=1.541 Å) operated at 40 kV and 44 mA. 2θ/θ scans were conducted using a 0.05° step

size and 5 sec hold per step. SEM characterization was conducted using a Hitachi S-4800 FE-

SEM. EDS data was collected using EDAX Genesis on the SEM to validate sample elemental

composition. HR-TEM images were collected on a JEOL 2010F TEM at 200kV.

4.3 Results and Discussions:

The search for highly active and stable catalysts for use as reversible hydrogen electrodes

in the H2-Br2 system began by identifying catalysts which exhibited reversible HER & HOR

activity in ideal (non-poisoning) systems. Thus, various PGM metal, chalcogenide and “nitride”

samples were tested in 0.1 M HClO4 electrolyte to evaluate their ability to catalyze both the HER

& HOR. It should be noted, that the use of the term “nitride” or Nx throughout this chapter is not

meant to infer the existence of a formal crystalline PGM-nitrides. While crystalline transition

metal-nitrides can readily be formed readily by heating precursors under NH3 in a tube furnace at

ambient pressure and temperatures below 1000°C, crystalline PGM-nitrides require much more

extreme conditions of 5-50 GPa and 1000-3000 K,[9-11] although some very recent reports have

produced PGM-nitrides via sputtering.[12, 13] Instead of the crystalline PGM-nitrides reported

in recent literature, the wet chemical synthesis used to fabricate the catalysts in this report more

likely produced a “nitrogen-functionalized” catalyst. However, experiments have not been

conducted to verify the location of the nitrogen functionality in the catalyst. Therefore, the

nitrogen could possibly reside in the carbon support, as a bridging functionality between the

carbon and the metal, or even doped into the metal in the tetrahedral or octahedral sites. The

absence of the PGM-nitride crystal phase is emphasized by XRD results which show that the

132

samples are dominated by the metallic diffracting domains and show no presence of diffracting

domains characteristic of the crystalline nitrides.

4.3.1 Baseline HER-HOR Testing in RDE Configuration:

Figure 2 – Left: Area-normalized, Right: Mass-normalized activity of various bifunctional HER-

HOR catalyst in 0.1M HClO4 at room temp. (~23°C). CV scans (20 mV/sec). 50 ug(metal)/cm2

for Pt, Ir & Rh catalysts, 200 ug/cm2 for Ru catalyst.

Figure 2 shows the HER-HOR activity of select samples in the (presumably) non-

adsorbing 0.1 M HClO4 electrolyte. The general trend of HER & HOR activity is: Pt > RhS > Ir

> Ru-Nx in the HClO4 electrolyte. In addition, figure 1 shows that Pt is much more active than

any of the other samples. The dramatic differences in mass-transport limiting current for the non-

Pt samples indicate that the non-Pt samples suffer from a very low density of active sites. To

examine this phenomenon, we must consider the true surface area of each catalyst. As the test

conditions were identical for evaluation of each catalyst, the mass-transport limiting current is

determined by Eq. [1]:

[1]: ilim = A*(nFD / δ ) * Cbulk

133

Where A is the area of the electrode, n is the number of moles of electrons transferred in

the reaction, F is Faraday’s constant (96,485 C / mol e-), D is the diffusion constant for the

analyte (H2(g)), δ is the diffusion layer thickness and Cbulk is the concentration of the analyte in

the bulk electrolyte. For Eq. [1], all variables should remain constant given the identical

temperature, scan-rate and concentrations of electrolyte and analyte used to test each sample.

Therefore the only difference in the measured current is the “true area” of the electrode. This

value is typically referred to as the electrochemical surface area (ECSA). For Pt electrocatalysts,

the ECSA can be evaluated using the well-known method of determining the Hupd charge and

dividing by the reference value of 210 μC/cm2. This value corresponds to the charge passed from

anodically striping a monolayer of adsorbed hydrogen atoms from 1 cm2 of Pt surface. The Pt

sample tested in figure 2 exhibited an ECSA of 7.48 cm2 and thus an ECSA-normalized limiting-

current-density of 0.126 mA/cm2

ECSA for the HOR. Thus, although there are no well-defined

methods for experimentally determining ECSA on Ir, Ru and RhS catalyst surfaces, we can

determine the ECSA of these catalysts by comparing the ratio of HOR limiting-currents and

comparing this ratio to the well-defined Pt ECSA. These ECSA values are reported in Table 1. In

comparison to this method, some researchers report the physical surface area of non-Pt catalyst.

This value can be determined directly from BET or SAXS measurements or indirectly from

XRD, assuming spherical or hemi-spherical particle surfaces, with volume and surface area

relative to XRD crystallite size. Table 1 indicates only small differences in crystallite size for the

samples shown in figure 2 (except for RhS). For most metal or chalcogenide nanoparticle

samples we can assume that the crystallite size (xs) is nearly synonymous with particle-size.

Therefore, by assuming spherical nanoparticle morphology, we can calculate the particle volume

from xs, the number of particles using the catalyst density and mass-loading on the electrode and

134

the total exposed particle surface area using spherical or hemi-spherical surfaces for each

particle. Table 1 shows that for Pt, the ratio of ECSA to XRD-derived surface area is nearly unity

when we consider an average of spherical and hemi-spherical surface area. This indicates that

~3/4 of the Pt particle surface is exposed to the reacting electrolyte or active for catalysis (i.e.

some of the Pt surface may be buried in the carbon support or otherwise inactive for catalysis). In

comparison to the Pt ECSA, the Ru & Ir samples of nearly identical xs exhibit only 20 and 30 %

utilization of their respective XRD-derived surface-area. The RhS sample shows a somewhat

higher density of active-sites at 60% ECSA:XRD-SA ratio, however, it should be further noted

that Ru & Ir samples reach HOR ilim conditions by +100 mV, the RhS/C sample does not reach

the limiting current until at least 200 mV. Thus indicating slower kinetics of HOR in RhS.

Table 1 – Electro-Chemical Surface Area Comparisons

Sample HOR ilim

(mA) ECSA (cm2)

ECSAm (m2/g)

XRD xs

(nm)

XRD SA (sphere)

XRD SA (hemi-sphere)

XRD avg. SA

ECSA vs.

XRD SA

46% Pt/C 0.749 7.48 79 3 9.29 4.64 6.97 1.07

40% Ir/C 0.304 3.03 31 2 13.07 6.53 9.80 0.31

30% RhS/C 0.445 4.44 45 8 9.67 4.84 7.25 0.61

50% Ru-Nx/C 0.304 3.03 8 2 23.71 11.85 17.78 0.17

Figure 3 shows the HER-HOR activity of the same catalysts from Figure 2, but in a 0.5

M HBr electrolyte. It is clear that the Rh & Ru samples exhibit negligible HOR activity and

significantly inhibited HER activity in HBr electrolyte. Interestingly, the Ir/C sample appears to

increase in both HER & HOR activity in the HBr electrolyte. The differences in HER & HOR

activity on the Ir/C sample may be due to the increased proton concentration in HBr.

135

Additionally, the ClO4- anion, while well-known to be non-adsorbing on Pt surfaces, may have a

higher adsorption affinity for the Ir/C catalyst. In either case, the Ir/C sample exhibits sufficient

HER & HOR activity for consideration as a candidate for the reversible hydrogen electrode of

the HB-RFB cell.

Figure 3 - Left: Area-normalized, Right: Mass-normalized activity of various bifunctional HER-

HOR catalyst in 0.5 M HBr at room temp. (~23°C). CV scans (20 mV/sec). 50 ug(metal)/cm2 for

Pt, Ir & Rh catalysts, 200 ug/cm2 for Ru catalyst.

Figure 3 shows the HBr testing of an array of Rh & Ru chalcogenides. Previous reports

by MacFarlane et al.[14] indicated that Rh & Ru chalcogenides exhibited some tolerance to

bromide adsorption, however their results investigated primarily HER activity and largely

ignored the HOR activity of these chalcogenide samples. In comparison, the Pt, Pt-Nx and Pt-Ir-

Nx catalysts exhibit highly reversible hydrogen reaction activity (i.e. high HER & HOR activity).

In particular, the Pt-Ir-Nx sample appears to exhibit HOR kinetics superior to the Pt/C standard.

136

Figure 4 – Left: HOR-HER testing in H2-purged 0.5 M HBr on selected Rh & Ru samples

compared to Pt/C standard. Note: all Rh & Ru-based samples show moderate HER activity and

no HOR activity under H2-purged conditions. Right: mass-normalized reversible HER-HOR

activity observed on Pt-Nx and Pt-Ir- Nx catalysts.

Table 2: Catalyst HER-HOR in HBr

Sample HER:

j @ -150 mV

HOR:

j @ +150 mV

(A/g) (A/g)

46% Pt/C -1030 40

30% Rh/C -145 6

30% RhS/C -340 6

30% RhSe/C -86 5

30% Ru-Nx/C -113 0

30% RuS/C -115 0.3

30% RuSe/C -18 0.05

30% Pt-N/C -525 65

60% Pt-Ir-N/C -1190 65

Table 2 summarizes HER & HOR activity of the samples evaluated under RDE testing

conditions. The goal of this project was to identify stable, bi-functional (HER-HOR) catalysts for

the H2-Br2 system. Thus, although the Ru and Rh-based catalysts have previously exhibited

excellent halide-poisoning resistance in ORR applications such as ODC for Cl2 generation,[15]

137

these chalcogenides of Rh & Ru show almost complete de-activation of HOR activity in the HBr

electrolyte. Furthermore, the low HOR activity is likely not due to low surface area, as the xs

were calculated to be 5-10 nm for all samples using the Williamson-Hall method built-in to

Rigaku’s PDXL analysis software. These xs values translate to physical surface areas of 50-100

m2/g, very similar to state-of-the-art Pt/C commercial standard. Given the high physical surface

area and resulting low HER-HOR activity we can confirm that Ru & Rh samples suffer from a

low density of active sites, as noted by the low HOR limiting currents discussed above, as well

as slow HER & HOR kinetics. These kinetics were further investigated in H2-pump cell testing.

4.3.2. H2-pump Testing to Quantify HER-HOR Kinetics:

Figure 6 shows the results of H2-pump cell testing on the most promising non-Pt

catalysts. Because the initial RhS/C sample exhibited the best HER & HOR activity of any of the

Ru & Rh chalcogenide catalysts, an optimization study was conducted on the RhS. Previous

literature indicates that heating the RhS catalyst increases the metallic Rh character and results in

improved ORR activity.[7] This optimization of crystal structure and electrocatalytic activity via

thermal treatments was confirmed in our study of HER-HOR activity on RhS catalysts.

138

Figure 5 – Optimization of RhS/C heat-treatments. Left: XRD analysis of RhS/C samples.

Right: H2-pump data shows that optimized HER-HOR activity results from heating to 800C.

Figure 6 shows the investigations of the RhS/C catalyst samples. The XRD data shows

that for the samples heated to 500°C or 800°C, the primary crystal phase is R17S15 and that

heating to higher temperatures results in an increase in the metallic Rh phase as indicated by the

increase in intensity of the diffraction peak at ~41 degrees. The H2-pump testing shows that the

RhS sample heated to 800°C (blue / triangles) is the only sample which exhibits sufficient HOR

activity to operate in a full-cell configuration. However, subsequent full-cell testing by our

collaborators at TVN Systems indicates that while Pt/C catalysts can achieve 500 mA/cm2 at 0.9

V during discharge, even the best RhS/C catalyst only achieve 150 mA/cm2 at 0.9 V during

discharge. Furthermore, the best RhS/C catalysts require charge/discharge voltages of 1.3 V &

0.5 V respectively to achieve the operational target of 500 mA/cm2, which translates to a voltaic

efficiency of only 38%. Thus the slow kinetics of the RhS samples effectively preclude their use

as catalysts in the HB-RFB system, which requires high efficiency to compete as a viable EES

technology option.

Figure 6 – H2-pump analysis of down-selected catalysts. a.) Pt-Ir-N/C vs RhS & RuS.

139

b.) Pt-Ir-Nx/C vs commercial Pt/C. Calculated Pt-Ir-N/C i0 = 1.8 A/g

In contrast to the slow HER/HOR kinetics on RhS/C catalysts, Figure 7 shows the

kinetics of the Pt-Ir-Nx/C sample compared to the non-Pt catalysts and the Pt/C commercial

standard. Figure 7 shows that the Pt-Ir-Nx/C sample exhibits much faster HER & HOR kinetics

than the non-Pt samples and in fact the HOR activity appears to be slightly greater than that of

the Pt/C standard. To quantify the kinetics, we evaluated the exchange current based on the

micro-polarization region near the E0 for HER/HOR. This micro-polarization region is

characterized by a nearly linear i-V response at +/- 5 mV from the thermodynamic equilibrium

potential (0 V vs. RHE). This linear i-V behavior is in contrast to the exponential i-V response

typical of Butler-Volmer electrode kinetics. However, it simplifies quantification of the

exchange current density.

Equation (3) from Nguyen et al.[8] was used to quantify the exchange current density of

each catalyst. Equation (3) is reproduced here to show how the area-normalized exchange

current density (j0) was calculated:

(3) j0 = (RT/2F) * (m/2)-1

Where R is the gas constant, T is the temperature in Kelvin, F is Faraday’s constant and

m is the slope of the linear region of HER-HOR polarization curve. The value j0 was then

converted to the mass-normalized i0 by dividing A/cm2 by the mass loading in mg(metal)/cm

2

thus providing units of i0 = A/mg(metal). For commercial Pt/C the results shown in Figure 7

validate that our testing method matched the reported value of 1.2 A/mg reported by Nguyen et

al.[8] The Pt-Ir-N/C sample exhibited an even higher i0 of 1.8 A/mg, while RhS/C exhibited

only 0.3 A/mg and RuS showed no HOR activity and a HER i0 of 0.2 A/mg.

140

On a side-note it should be noted that that the value (m/2) in Eq. 3 should be replaced

with (m-mPt) for testing conditions which evaluate a non-Pt catalyst using at Pt C/RE, with mPt

being the slope of the Pt polarization curve corresponding to the contribution of the Pt C/RE to

the measured potential. From the Nguyen et al.[8]:

“dE/di is half (because there are two electrodes and reactions involved) of the slope of the

IR-corrected hydrogen pumping curve…” (pg F334)

The assumption of Eq. 3 is true for MEAs in which both electrodes contain the same

catalyst and we can assume that HER & HOR polarization curves are symmetric. However, the

use of the Pt C/RE in H2-pump testing of a non-Pt catalyst means that the measured dE/di is

actually the sum of the polarization from the experimental electrode and the Pt C/RE, thus it

would be more accurate to subtract the dE/di of the Pt/C baseline from the measured dE/di to

isolate the contribution from the experimental catalyst.

4.3.3. Evaluation of Halide Poisoning and Catalyst Stability:

RDE testing was used to further evaluate the effects of Br-poisoning on catalysts which

exhibited sufficient activity for the reversible hydrogen reactions. Figure 7 shows the Hupd charge

(under argon-purged conditions) and HER-HOR results (under H2-purged conditions) for a

standard commercial catalyst (46% Pt/C from Tanaka, TKK, Japan). The Hupd response is

typically used to quantify the ECSA of Pt/C catalyst. Thus, figure 7 shows that the Pt/C standard

does suffer some loss of EASA in the presence of the Br-. Because the electrode was scanned to

potentials positive of the pzc, the loss of ECSA is expected. Because of the well-known

electrostatic attractions between halides in electrolyte and positively charged electrode surfaces,

the electode was only cycled very briefly to potential positive of the pzc of Pt to obtain the data

141

in figure 7. Under H2-purged conditions, figure 7 shows that the HER activity is in fact increased

in the concentrated HBr electrolyte compared to the non-adsorbing HClO4, and the HOR activity

in H2-purged electrolytes are nearly identical. The similarity of the HOR activity is explained by

the observations of Gasteiger et al.[16] that in RDE testing in acid electrolyte, only the mass-

transport limiting currents will be observed. This effect is due to the fact that the HOR kinetics

of Pt in acid are much faster than the diffusion of H2(g) to the electrode, thus the observed

diffusion-limited response for the anodic reaction occurs before the kinetic behavior can be

observed or quantified.

Figure 7 – left: conditioning CV @100 mV/sec showing Hupd on Pt/C standard catalyst as well

as Br-/Br2 faradaic current ~1V vs. RHE in argon-purged 0.1M HClO4 and 0.5 M HBr. Right:

HER-HOR CVs under H2-purged conditions @20 mV/sec.

Subsequent stability testing studies on various Pt & Ir catalysts indicated a somewhat

higher degree of stability from a commercial Pt-Ir/C catalyst (vide infra). Thus, figure 8 shows

the reversible and irreversible deactivation of the commercial Pt-Ir/C standard. Very similar

results were also observed on the Pt/C standard, but the most detailed analysis was conducted on

the Pt-Ir/C sample and thus these results are shown here. Figure 8 show the results of CV cycling

142

at various potential ranges on a Pt-Ir/C catalyst (with enhanced stability) in 1 M HBr electrolyte.

Initially the sample was cycled from -50 mV to +300 mV (left, black lines) from which a rapid

decrease in the Hupd charge (i.e. ECSA) can be observed over the course of only 20 cycles. After

this initial low-voltage cycling, the electrode was subjected to mid-voltage cycling from -50 mV

to +800 mV. The first scan was started at +250 mV and a large anodic feature can be observed in

the first scan with a peak at ~0.55 V. This peak is attributed to the oxidative stripping of the

adsorbed bromide, after which the Hupd charge returned to its initial magnitude. During the

course of each subsequent mid-voltage cycling scan, a stable Br-stripping feature was observed

and the initial Hupd was maintained. After the low- and mid-voltage cycling, the electrode was

subjected to high-voltage cycling with an anodic limit of 1.2 V. The on-set bulk bromine

evolution (2Br- + 2e

- Br2) is observed at just below 1 V. The inset of the right graph shows the

rapid loss of Hupd charge (i.e. ECSA) during each of 20 scans to 1.2 V, followed by the mid-

voltage scans from -50 mV to +800 mV which did not result in regaining of the initial Hupd

charge. These results indicate that while the bromide adsorption is reversible, the high-voltage

condition appears to irreversibly deactivate the catalyst, presumably by dissolution of the metals

under the high-voltage, oxidizing conditions.

143

Figure 8 – Evaluation of reversible and irreversible de-activation of Pt-Ir/C catalyst. Left:

reversible deactivation with progressive loss of Hupd charge during initial low-voltage cycling

from -50 mV +300 mV, followed by reactivation and increase in Hupd charge upon cycling to

+800 mV potential limit. Note the wide anodic peak between ~0.5-0.7 V, which represents

stripping of adsorbed bromide. Right: irreversible deactivation from initial cycling to a 1.2 V

followed by cycling to +0.8 V. In the latter case the loss of Hupd was not reversible and the

catalyst presumably dissolved upon being cycled to 1.2V. The same potential-dependent

reversibility of deactivation was observed on the Pt/C standard but is not shown here for brevity.

Once it had been determined that the bromide adsorption alone did not irreversibly de-

activate the Pt/C or Pt-Ir/C catalysts in the course of typical RDE testing, an overnight soak test

was devised to evaluate the stability in a mixed HBr / Br2 electrolyte. Figure 9 shows the “soak-

test” results for Pt/C and Pt-Ir-Nx/C catalysts. Both electrodes were cycled to high-voltage limits

followed by addition of Br2 to the electrolyte and then subjected to a 24-hr soak. In the soak-test,

the HER-HOR activity was first evaluated in H2-purged 0.5 M HBr, followed by high-voltage

cycling in argon-purged electrolyte. After the initial evaluation, liquid Br2 was added to the

electrolyte to create a solution of 0.5 M HBr and 1 mM Br2. The electrode was then kept under

argon-purged conditions overnight and re-tested ~24 hours later. The results in figure 9 indicate

that the Pt/C standard suffers from complete loss of ECSA and HOR activity (although it retains

144

moderate HER activity). In contrast, the Pt-Ir-Nx/C catalyst retained a high degree of ECSA as

well as HER & HOR activity.

Figure 9 – Soak test data. Electrodes were initially tested in 0.5M HBr under argon-purged and

then H2-purged conditions. Following the initial testing, 5μL Br2 was added to 100 mL of

electrolyte for a 1 mM Br2 concentration. The H2-purging was stopped and the cell was left open

to air for 24-hrs. After 24-hr soak, the 0.5 M HBr + 1 mM Br2 electrolyte was purged with argon

for the diagnostic test followed by H2-purge for HER-HOR testing.

The initial 24-hr soak-testing in 1M HBr / 1mM Br2 showed that the Pt-Ir-Nx/C sample

exhibited ~50% loss in Hupd charge and HOR mass-transport limited current. Because the sample

contains 50:50 ratio of Pt:Ir, a commercial Ir/C sample was subjected to high-voltage cycling to

evaluate the stability of the Ir component. Evaluation of the Ir/C catalyst was conducted to

identify whether the apparent stability of the Pt-Ir-Nx/C was simply a result of increased bromine

tolerance of the Ir-component, the results in Figure 10 a. & b. show that both the pure Pt/C and

Ir/C samples exhibit complete loss of HOR activity after high-voltage cycling. In contrast, the Pt-

Ir-Nx/C sample retains HOR activity immediately after high-voltage cycling (Fig. 9c.) and only

suffers partial loss of HOR activity after overnight soaking in HBr (Fig 9d.).

145

Figure 10 – HER/HOR testing of PGM catalysts. All samples tested in H2-purged 1 M HBr at

room temp. (~23°C) with 50 μg(metal)/cm2 at a scan-rate of 20 mV/sec.

4.3.4 Mild, Mid & Harsh Stability Testing of Pt/C Standard:

Figures 10 - 12 show the stability of the Pt/C standard under various testing conditions.

Figure 11 shows that under the most mild testing conditions, which involve overnight-soaking in

argon-purged 1M HBr, with no Br2 present, the Pt/C standard exhibits very stable HER-HOR

activity, yet suffers a loss of ~40% of the ECSA (derived from Hupd-stripping charge). Under the

most mild testing conditions used in figure 11, it is assumed that the Pt/C sample suffers from

reversible bromide adsorption but no dissolution of the catalyst. Figure 12 shows the somewhat

more harsh testing condition of soaking in argon-purged 1 M HBr with 1 mM Br2 overnight.

146

Even with the argon-purging of the electrolyte, the measured open circuit voltage (OCV) goes to

~0.95 V (vs. RHE) during this overnight soak. Figure 12 shows complete loss of HER/HOR

activity and ECSA during the soak with Br2-present. For the results presented in figure 12, the

electrode potential was carefully controlled during CV cycling to ensure that the applied

potential did not exceed +800 mV (vs. RHE). Thus, even without the high-voltage cycling

conditions, the Pt/C sample appears to have been dissolved by the addition of Br2 to the

electrolyte. It is likely that under these conditions, the OCV is high enough to cause oxidation of

the catalyst to a metal oxide and subsequent dissolution of the metal oxide catalyst surface.

Further ICP-MS studies of the concentration of dissolved precious metals in the electrolyte

would validate this assumption.

Finally, in Figure 13 we see that the harshest testing conditions involve cycling the

electrode to 1.2 V. Figure 13 shows the immediate and complete loss of HER/HOR activity and

ECSA after cycling to 1.2 V. It is likely that the high voltage causes formation of acid-soluble

Pt-oxide.

147

Figure 11 – Mild stability testing conditions for Pt/C Standard: Overnight soak in 1M HBr (with

no Br2). Main figure shows HER-HOR activity of H2-purged electrolyte. Inset shows Hupd from

Argon-purged electrolyte and resulting ECSA calculated by integrating anodic H-stripping peak.

Figure 12 – Harsh stability testing conditions for Pt/C standard: Overnight soak in 1M HBr &

1mM Br2. Left: HER-HOR activity in H2-purged electrolyte. The HOR limiting current

decreased dramatically from ~2 mA/cm2 to <0.4 mA/cm

2 after soaking in HBr / Br2 solution.

Right: Hupd charge in Argon-purged electrolyte. The “Low-Voltage” data was cycled repeatedly

between -50 mV and +300 mV and exhibited progressive loss of ECSA, stabilizing ~14 m2/g.

The “Mid-Voltage” data was cycled between -50 mV and +800 mV, during which a Br-stripping

peak was observed between 0.50.7V. The Mid-Voltage cycling immediately increased the

observed ECSA which remained stable at ~42 m2/g.

Figure 13 – High-Voltage cycling of Pt/C standard: Main figure shows rapid loss of Hupd charge

during high-voltage cycling in argon-purged electrolyte and complete loss of ECSA after 24-hr

148

soak. Inset shows nearly complete loss of HER & HOR activity in H2-purged electrolyte

immediately following high-voltage cycling.

4.3.5 Mild, Mid & Harsh Stability Testing of Pt-Ir/C:

Figures 13 & 14 evaluate the stability of a commercial Pt-Ir/C sample under various

testing conditions. The mild testing conditions shown in Figure 14 show stability after overnight

soaking in 1M HBr, with no Br2 present. Figure 14 indicates that the Pt-Ir/C sample (like the

Pt/C standard) exhibits stable HER & HOR activity when protected from high-voltage conditions

by the argon-purging. Under these mild conditions, both Pt/C and Pt-Ir/C samples were observed

to have an OCV of ~0.5 V vs RHE after overnight soaking in the argon-purged 1 M HBr (with

no Br2 present). Similar to the Pt/C standard, the Pt-Ir/C sample shows negligible change in

HER/HOR activity, but a slightly higher 60% loss in ECSA after the mild overnight soak.

Figure 15 shows the stability results under the more harsh conditions of soaking in 1M

HBr with 1mM Br2. As with Pt/C, the measured OCV after overnight soak in the presence of Br2

was ~0.95 V (vs. RHE). In this case, the Pt-Ir/C sample suffers a loss of ~70% ECSA and the

HOR activity decreases by ~50%. After the overnight soak in HBr / Br2, the HER activity of the

Pt-Ir/C sample also decreased significantly, however the HER activity was almost completely

recovered upon cycling the soaked electrode to +0.8 V, presumably due to stripping of any

adsorbed bromide. The high-voltage cycling shown in the right-hand graph of figure 8 shows an

increased tolerance for high-voltage dissolution by Pt-Ir/C compared to the immediate and

complete deactivation of Pt/C upon high-voltage cycling in figure 13. In addition, the Pt-Ir/C

sample retained a high degree of HER/HOR activity under the harsh HBr / Br2 soak conditions

compared to a complete loss of activity by Pt/C after HBr / Br2 soak (Figure 15 vs. Figure 12).

Since a commercial Ir/C standard was observed to suffer complete loss of HER/HOR activity

149

after high-voltage cycling (shown in figure 10), the increased stability of the Pt-Ir/C catalyst is

likely due to an effect from alloying Pt & Ir.

Figure 14 – Mild stability testing conditions for Pt-Ir/C: Overnight soak in 1 M HBr (with no

Br2). Main figure shows HER-HOR activity of H2-purged electrolyte. Inset shows Hupd from

Argon-purged electrolyte and resulting ECSA calculated by integrating anodic Hupd charge.

Figure 15 - Harsh stability testing conditions for Pt-Ir/C: Overnight soak in 1M HBr &

1mM Br2. Right: HER-HOR activity in H2-purged electrolyte. The HOR limiting current

decreased by ~50% from ~2mA/cm2 to <1mA/cm

2 after soaking in HBr/Br2 solution. Left: Hupd

charge from argon-purged electrolyte. The “Low-Voltage” data was cycled repeatedly between -

50 mV and +300 mV and exhibited progressive loss of ECSA. The “Mid-Voltage” data was

cycled between -50 mV and +800 mV. The Mid-Voltage cycling immediately increased the

150

observed ECSA which remained stable at ~52 m2/g at BOL. However, the ECSA decreased by

~70% during HBr / Br2 soaking.

4.3.6 Mild, Mid & Harsh Stability Testing of Pt-Ir-Nx/C:

Figures 16 & 17 evaluate the stability of the Pt-Ir-Nx/C sample under various testing

conditions. Figure 16 shows the relative stability of the Pt-Ir-Nx/C sample to high-voltage

cycling. The sample exhibits an initial loss of Hupd charge, but the surface area appears to

stabilize after an initial mild loss of ECSA. Furthermore, the Pt-Ir-Nx/C sample retains almost

100% HER/HOR activity after high-voltage cycling. This is in stark contrast to the complete loss

of HER/HOR activity after high-voltage cycling by Pt/C in figure 13. Finally, figure 17 shows

the results of the harsh HBr / Br2 stability test on the Pt-Ir-Nx/C sample. Similar to the

commercial Pt-Ir/C sample, the Pt-Ir-Nx/C sample suffered ~50% loss of HER/HOR activity, but

only ~30% loss of ECSA, compared to ~70% loss of ECSA for Pt-Ir/C and 100% loss of ECSA

for the Pt/C standard. Although these final results appear to indicate that the Pt-Ir-Nx/C sample is

more stable than the commercial Pt-Ir/C sample, it may also be possible to explain the apparent

increased stability via particle-size effects. We assume that the catalyst dissolution is a first-

order reaction and that the rate of dissolution is proportional to the exposed surface area. Thus

smaller, more finely dispersed particles would exhibit higher dissolution-rate than large particles

with low ECSA. The Pt-Ir-Nx/C sample has lower inherent ECSA and larger XRD-derived

crystallite size of 4-5 nm, compared to the extremely finely dispersed Pt/C and Pt-Ir/C samples

which exhibit 1-2 nm crystallite size. Thus, while it is certainly possible that the nitrogen

functionality is enhancing the catalyst stability, we cannot at this point rule out particle-size

effects.

151

Figure 16 – High-Voltage cycling of the Pt-Ir-Nx/C sample: Left: only mild loss of Hupd charge

in argon-purged electrolyte upon repeated cycling to high-voltage. Inset shows negligible loss of

HER & HOR activity in H2-purged electrolyte immediately after high-voltage cycling. Right:

only moderate loss of ~50% ECSA after high-voltage cycling followed by 24-hr soaking of the

Pt-Ir-Nx/C catalyst.

Figure 17 - Harsh stability testing conditions for Pt-Ir-Nx/C: Overnight soak in 1M HBr & 1mM

Br2. Right: HER-HOR activity in H2-purged electrolyte. HOR limiting current decreased by

~50% from ~1.2 mA/cm2 to 0.6 mA/cm

2 after soaking in HBr / Br2 solution. Left: Hupd charge

from argon-purged electrolyte. The “Low-Voltage” data was cycled repeatedly between -50 mV

and +300 mV but exhibited negligible loss of ECSA. The “Mid-Voltage” data was cycled

between -50 mV and +800 mV. Although the mid-voltage cycling exhibited a Br-stripping peak,

the Hupd charge after mid-voltage cycling did not increase – instead, both low-voltage and mid-

voltage cycling exhibited very stable Hupd charge indicating a resistance to halide adsorption.

152

4.4 Conclusions:

During the course of this project we have developed a more thorough understanding of

the deactivation of the reversible hydrogen electrode in the HB-RFB. In particular, if the

electrode potential is not allowed to shift positive of ~0.8 V (or perhaps 1 V), even the Pt/C

standard should exhibit a high-degree of stability. However, although this indicates the

possibility of over-voltage protection as an engineering solution to the catalyst deactivation

problem, it is presumed that no real membrane will be able to completely prevent Br2-crossover

to the hydrogen electrode. It has been shown that a stability trend exists for catalysts exposed to

high-voltage conditions (either through potential cycling or exposure to Br2) such that Pt/C < Pt-

Ir/C < Pt-Ir-Nx/C. The stability tests results conducted in RDE configuration in this lab have

been validated in full-cell testing by collaborators at TVN Systems and the Lawrence Berkeley

National Lab. Further catalyst stability testing could be conducted in full-cell configuration

under potential hold during shut-down to evaluate the stability of the catalyst with real Br2-

crossover conditions.

It should also be noted that attempts to optimize the dispersion of the Pt-Ir-Nx/C sample

were complicated by synthetic challenges. In particular, it was discovered that the 1,3-

propanediamine nitrogen complexing agent is very caustic and in many synthesis attempts

addition of too much 1,3-propanediamine caused a drastic increase in pH which greatly

decreased Ir-yield in the final product. The Ir yield could be recovered by heating the reaction

solution, but at the cost of a loss of Pt yield from heating. Thus we believe that investigation of

addition of very dilute 1,3-propanediamine may allow simultaneous complexing of the PtCl42-

and IrCl62-

precursors by the nitrogen precursor – thus increasing yield of both metals. Additional

synthetic research on Pt-Ir-Nx/C catalyst may investigate the use of alternate nitrogen precursors.

153

4.5 Acknowledgements:

The authors deeply appreciate financial assistance from TVN Systems Inc. under an ARPA-E

grant. We also gratefully acknowledge full-cell testing conducted by collaborators at TVN

Systems Inc. and the Lawrence Berkeley National Lab.

154

4.6 References:

1. Calle-Vallejo, F., M.T. Koper, and A.S. Bandarenka, Tailoring the catalytic activity of

electrodes with monolayer amounts of foreign metals. Chemical Society Reviews, 2013. 42(12):

p. 5210-5230.

2. Arruda, T., et al., In Situ XAS Investigation of Electrocatalysts Surface Poisoning by

Halides. ECS Transactions, 2007. 11(1): p. 903-911.

3. Arruda, T.M., et al., Investigation into the competitive and site-specific nature of anion

adsorption on Pt using in situ X-ray absorption spectroscopy. The Journal of Physical Chemistry

C, 2008. 112(46): p. 18087-18097.

4. Strmcnik, D., et al., The effect of halide ion impurities and Nafion on electrooxidation of

CO on platinum. Solid State Ionics, 2005. 176(19): p. 1759-1763.

5. Ferro, S. and A. De Battisti, The bromine electrode. Part I: Adsorption phenomena at

polycrystalline platinum electrodes. Journal of applied electrochemistry, 2004. 34(10): p. 981-

987.

6. Huang, J., W. Ogrady, and E. Yeager, The effects of cations and anions on hydrogen

chemisorption at Pt. J. Electrochem. Soc.;(United States), 1977. 124.

7. Ziegelbauer, J.M., et al., Fundamental Investigation of Oxygen Reduction Reaction on

Rhodium Sulfide-Based Chalcogenides. The Journal of Physical Chemistry C, 2009. 113(17): p.

6955-6968.

8. Kreutzer, H., V. Yarlagadda, and T. Van Nguyen, Performance Evaluation of a

Regenerative Hydrogen-Bromine Fuel Cell. Journal of The Electrochemical Society, 2012.

159(7): p. F331-F337.

9. Crowhurst, J.C., et al., Synthesis and characterization of the nitrides of platinum and

iridium. Science, 2006. 311(5765): p. 1275-1278.

10. Ding, Z., et al., High pressure synthesis and characterization of noble metal nitride IrNx.

Materials Letters, 2013. 107(0): p. 382-385.

155

11. Kawamura, F., H. Yusa, and T. Taniguchi, Synthesis of rhenium nitride crystals with

MoS2 structure. Applied Physics Letters, 2012. 100(25): p. 251910.

12. Bouhtiyya, S., et al., Application of sputtered ruthenium nitride thin films as electrode

material for energy-storage devices. Scripta Materialia, 2013. 68(9): p. 659-662.

13. Veith, G.M., et al., Evidence for the Formation of Nitrogen-Rich Platinum and Palladium

Nitride Nanoparticles. Chemistry of Materials, 2013. 25(24): p. 4936-4945.

14. Ivanovskaya, A., et al., Transition Metal Sulfide Hydrogen Evolution Catalysts for

Hydrobromic Acid Electrolysis. Langmuir, 2012. 29(1): p. 480-492.

15. Jin, C., et al., Rh-RhSx nanoparticles grafted on functionalized carbon nanotubes as

catalyst for the oxygen reduction reaction. Journal of Materials Chemistry, 2010. 20(4): p. 736-

742.

16. Sheng, W., 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): p. B1529-B1536.

156

Chapter 5 Conclusion

5.1 Introduction:

The focus of this research has been the development of high-performance, high-durability

electrocatalysts for alkaline water electrolysis and the reversible hydrogen electrode in the H2-

Br2 redox flow battery. In the alkaline environment, Markovic et al.[1, 2] have demonstrated

enhanced electro-kinetics on composite metal/metal-oxide surfaces. The research described here

has supported the theories of Markovic et al. and demonstrated unprecedented mass-activity on a

composite Ni(Ox)/Cr2O3 catalyst for the hydrogen evolution reaction. In addition, investigations

of Pt-X alloys (where X is an oxo-philic metal) have demonstrated enhanced alkaline hydrogen

oxidation reaction (HOR) electro-kinetics on materials with vastly differing Hupd features, thus

supporting Markovic’s metal/metal-oxide theory and invalidating the hypothesis of Gasteiger et

al.[3] which suggests that Pt-H bond strength is the sole descriptor for alkaline HOR activity.

These observations suggest that, while the standard Sabatier principle is valid for homogeneous

electrode surfaces, a more complex analysis of composite surfaces may lead to the development

of next-generation electrocatalysts. Recent theoretical studies have begun expanding the

conceptual underpinnings of electrocatalysis beyond the simple thermodynamics of reactant

adsorption enthalpy to consider the geometric effects of various crystalline facets and the

interplay of ensemble requirements, such as for the oxidation of organic molecules.[4] It is our

belief that the next stage of electrocatalysis must consider composite catalysts which contain

adjacent active-sites which work in concert. This is particularly relevant for complex multi-

electron, multi-proton transfer reactions such as the oxygen reduction and evolution reactions,

which are of paramount importance to achieving a sustainable, renewable energy future.

157

5.2 Chapter 2 – Alkaline HER/HOR Electrocatalysis Conclusions:

After screening the HER activity of numerous binary and ternary Ni-alloys and

composite Ni/MOx/C materials, a Ni-Cr/C sample was identified which exhibits unprecedented

mass-activity for the HER in 0.1 M KOH electrolyte. In particular, the Ni-Cr/C sample required

the least amount of HER over-potential to achieve what literature indicates to be a direct

Volmer-Heyrovsky mechanism. In addition, the effects of the AEI binder on the HER have been

investigated for Pt, Ni-Cr and Ni-Mo catalysts. For Pt catalysts, careful control of CV potential

range limits indicates that the QA+ moiety of the AEI likely straddles the double-layer interface

and does not reside exclusively in the IHP. CV results from larger (~6 nm) unsupported Pt black

and smaller (~2-3 nm) supported Pt/C catalysts indicates that at high-index Pt (111) surface sites,

the QA+ moiety of the AEI is specifically adsorbed in the IHP, while for other Pt (100) & (110)

surface sites, the QA+ moiety likely resides in the OHP, but still exerts an electrostatic effect,

dampening the HER activity of the catalyst. Furthermore, the AEI-inhibition of the HER is

observed on Ni-Mo, but not on Ni-Cr/C. The cause of the AEI-tolerance by Ni-Cr is not

confirmed but a model is proposed surmising that the AEI relaxes the dipole orientation

constraints of H2O in the IHP thus allowing more H2O reactant molecules to assume an

orientation which facilitates H-OH cleavage on adjacent M/MOx surface sites.

Various analytical techniques (XRD, EDS, XAS) have identified that the Ni-Cr/C sample

actually contains metallic nickel as well as nickel- and chromium-oxides. XAS analysis of two

different Ni-Cr/C samples indicates that the HER enhancement is related to the adjacent Ni/NiOx

surface sites and that the Cr2O3 phase may act to stabilize the NiOx under the reducing HER

conditions. The HER enhancement on composite Ni/NiOx surfaces was observed by Lasia et

al.[5] by cycling a polycrystalline Ni electrode from HER to OER conditions, thus roughening

158

the surface with NiOx, but the effects were not stable because the NiOx was quickly reduced back

to the metallic state under HER conditions. More recently, enhanced HER has been reported on

Ni/NiOx by Markovic et al.[1] and Gong et al.[6] – both of whom suggested that at adjacent

metal/metal-oxide sites, the metal-oxide facilitates the formation of OHads, thus weakening the

H-OH bond in the HER reactant and reducing activation energy for the Volmer reaction.

However, the Ni Pourbaix diagram indicates that the NiOx component of these composite

surfaces is not stable below 0 V (vs. RHE) and will quickly reduce to metallic Ni, thus losing the

synergistic HER enhancement of adjacent Ni/NiOx surface sites. According to the XAS data it is

likely that the Cr2O3 component stabilizes the NiOx component well into the HER operating

voltage, allowing the catalyst to retain the enhanced HER activity from synergistic Ni/NiOx

surface sites. Further studies of the effects of the work-up conditions can be conducted to

optimize the structure and activity of this M/MOx catalyst. This type of composite metal/metal-

oxide electrocatalyst may be the key to realizing the promise of alkaline electrochemistry for

low-cost, high-purity hydrogen production.

5.3 Chapter 3 – Alkaline OER Conclusions:

Binary and ternary mixed-metal-oxide (MMO) films on a commercial Raney-Ni support

were evaluated for alkaline OER activity in 0.1M KOH at room temperature (~23°C).

Electrochemical testing in RDE configuration showed that a ternary Ni-Fe-Co-oxide (with 8:1:1

atomic ratio of metals) catalyst exhibited the best OER activity. The recorded mass-activity was

110 mA/mg at 1.55 V (iR-comp.). In addition, the ternary Ni-Fe-Co catalyst showed less than

1% loss of OER activity at 1.7 V applied potential (1.54 V iR-comp.) during one hour of

continuous testing. Analysis of the heat-treatment conditions showed that the ternary Ni-Fe-Co

159

sample was less sensitive to loss of activity from exposure to oxidizing conditions than was a

binary Ni-Fe catalyst which suffered significant de-activation due to an oxidizing heat-treatment.

The tolerance to oxidizing conditions may in part be rationalized by in-situ XAS studies which

showed that MMO films which contained cobalt exhibited less oxidation of the nickel-

component under the oxidizing high-voltage conditions of the OER, presumably due to a higher

oxophilicity of the cobalt component than the nickel component. In addition, the high OER

activity of the ternary Ni-Fe-Co sample can be rationalized due to the high redox reversibility of

the active sites. Both the Yeager and Lyons-Brandon OER mechanisms emphasize the

importance of redox reversibility for achieving high OER activity. While previous trends have

shown the higher OER activity on Ni than on Fe, it has very recently been proposed that iron

impurities in the electrolyte often produce Ni-Fe catalysts for which Fe is the true active site.[7]

In fact, recent DFT calculations have proposed that Fe3+

sites provide ideal binding energy for

the OH and OOH OER intermediates, whereas Ni3+

sites bind these adsorbates too weakly.[7]

However, whether the Ni3+

or Fe3+

sites bind the intermediates, the results of this study indicate

that the redox reversibility of Ni sites certainly plays a role in OER catalysis. It may in fact be

the case that the OER occurs most rapidly on composite active sites where Fe may facilitate

binding of the intermediates while Ni facilitates proton transfer from a neighboring oxy-

hydroxide in the hydrated networks described by Lyons et al.[8]

Further development of catalyst morphology (nano-wire arrays, optimized Raney-type

porosity, etc.) will enable high-efficiency water splitting in alkaline electrolysis cells. In addition,

although the ORR activity of these MMO is not shown, preliminary results indicated that the

ORR kinetic region is not achieved until ~0.7 V, so while the current composition is not ideal as

a catalyst for reversible oxygen electrodes (because they exhibit slightly greater than 1 V

160

separation between ORR & OER kinetic regions, unlike optimized Mn-Ox which can achieve

~900 mV reversible kinetic regions).[9] These MMO catalysts may aid in the rational

development of reversible oxygen electrodes for unitized regenerative fuel cells or rechargeable

metal-air batteries which represent orders-of-magnitude greater energy density than the state-of-

the-art Li-ion battery.

5.4 Chapter 4 – HBr Reversible Hydrogen Electrode Conclusions:

During the course of this project we have developed a more thorough understanding of

the deactivation of the reversible hydrogen electrode in the HB-RFB. In particular, if the

electrode potential is not allowed to shift positive of ~0.8 V (or perhaps 1 V), even the Pt/C

standard should exhibit a high-degree of stability. However, although this indicates the

possibility of over-voltage protection as an engineering solution to the catalyst deactivation

problem, it is presumed that no real membrane will be able to completely prevent Br2-crossover

to the hydrogen electrode. It has been shown that a stability trend exists for catalysts exposed to

high-voltage conditions (either through potential cycling or exposure to Br2) such that Pt/C < Pt-

Ir/C < Pt-Ir-Nx/C. The stability tests results conducted in RDE configuration in this lab have

been validated in full-cell testing by collaborators at TVN Systems and the Lawrence Berkeley

National Lab. Further catalyst stability testing could be conducted in full-cell configuration

under potential hold during shut-down to evaluate the stability of the catalyst with real Br2-

crossover conditions.

It should also be noted that attempts to optimize the dispersion of the Pt-Ir-Nx/C sample

were complicated by synthetic challenges. In particular, it was discovered that the 1,3-

propanediamine nitrogen complexing agent is very caustic and in many synthesis attempts

161

addition of too much 1,3-propanediamine caused a drastic increase in pH which greatly

decreased Ir-yield in the final product. The Ir yield could be recovered by heating the reaction

solution, but at the cost of a loss of Pt yield from heating. Thus we believe that investigation of

addition of very dilute 1,3-propanediamine may allow simultaneous complexing of the PtCl42-

and IrCl62-

precursors by the nitrogen precursor – thus increasing yield of both metals. Additional

synthetic research on Pt-Ir-Nx/C catalyst may investigate the use of alternate nitrogen precursors.

162

References:

1. Danilovic, N., et al., Enhancing the Alkaline Hydrogen Evolution Reaction Activity

through the Bifunctionality of Ni(OH)2/Metal Catalysts. Angewandte Chemie International

Edition, 2012. 51(50): p. 12495-12498.

2. Strmcnik, D., et al., Improving the hydrogen oxidation reaction rate by promotion of

hydroxyl adsorption. Nat Chem, 2013. 5(4): p. 300-306.

3. Rheinländer, P.J., et al., Kinetics of the Hydrogen Oxidation/Evolution Reaction on

Polycrystalline Platinum in Alkaline Electrolyte Reaction Order with Respect to Hydrogen

Pressure. Journal of The Electrochemical Society, 2014. 161(14): p. F1448-F1457.

4. Li, H., et al., Bond-Making and Breaking between Carbon, Nitrogen, and Oxygen in

Electrocatalysis. Journal of the American Chemical Society, 2014. 136(44): p. 15694-15701.

5. Lasia, A. and A. Rami, Kinetics of hydrogen evolution on nickel electrodes. Journal of

Electroanalytical Chemistry and Interfacial Electrochemistry, 1990. 294(1–2): p. 123-141.

6. Gong, M., et al., Nanoscale nickel oxide/nickel heterostructures for active hydrogen

evolution electrocatalysis. Nat Commun, 2014. 5.

7. Friebel, D., et al., Identification of highly active Fe sites in (Ni, Fe) OOH for

electrocatalytic water splitting. Journal of the American Chemical Society, 2015.

8. Rebouillat, S., et al., Paving the way to the integration of smart nanostructures: part II:

nanostructured microdispersed hydrated metal oxides for electrochemical energy conversion

and storage applications. International Journal of Electrochemical Science, 2011. 6(11).

9. Gorlin, Y. and T.F. Jaramillo, A bifunctional nonprecious metal catalyst for oxygen

reduction and water oxidation. Journal of the American Chemical Society, 2010. 132(39): p.

13612-13614.