Heteroatom-containing Carbon Nanostructures as Oxygen Reduction Electrocatalysts for PEM and Direct Methanol Fuel Cells

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

By

Dieter von Deak, B.S.ChE

Graduate Program in Chemical Engineering

* * * *

The Ohio State University

2011

Dissertation Committee:

Professor Umit S. Ozkan, Advisor

Professor David Wood

Professor James Rathman

Copyright by

Dieter von Deak

2011

2

ABSTRACT

The main goal of this work was to undertake a fundamental investigation of precious metal-free carbon catalysts nano-structure modification to enable their use as oxygen reduction reaction (ORR) catalysts in proton exchange membrane (PEM) fuel cells. The sluggish ORR is accelerated by fiscally prohibitive loadings of Pt catalyst. The expense and availability of platinum motivate the development of non-precious metal carbon-nitroge-based ORR catalysts (CNx). The project targets the nature of oxygen reduction reaction active sites and exploring ways to create these sites by molecular tailoring of carbon nano-structures.

CNx grown with phosphorous had a significant increase in the ORR active site density. CNx catalyst growth media was prepared by acetonitrile deposition over a Fe and

P impregnated MgO. Rotating Ring Disk Electrode (RRDE) Activity and selectivity showed a significant increase in oxygen reduction current with CNx grown with less than a 1:1 molar ratio of P:Fe. Selectivity for the full reduction of dioxygen to water trended with increasing ORR activity for phosphorous grown CNx catalysts. Phosphorus growth altered the morphology of carbon-nitride graphite formed during pyrolysis.

The role of the transition metal used to form non-noble metal electrochemical oxygen reduction CNx catalysts was investigated through sulfur and carbon monoxide

ii treatments of the CNx and Pt/carbon electrocatalysts. The intent of poisoning was to show the existence of a non-iron containing electrocatalytic active site in CNx. The sulfur treatment increased the overpotential on a platinum catalyst, but enhanced the current density of the CNx catalyst while leaving the CNx iron phase unchanged. CO in the present of oxygen was found to strongly adsorb to platinum and completely eliminate all oxygen reduction. Under identical conditions, CNx showed a displacement of oxygen due to CO and no oxygen reduction poisoning effect. This suggests that either iron-based active site is sulfur and CO tolerant or that this active site does not participate in the electrocatalytic reduction of oxygen in CNx catalysts.

Density functional theory (DFT) calculations of small polycyclic aromatic hydrocarbons (PAHs) that have a similar electronic structure to carbon-nitride catalyst materials were preformed. A strong correlation between B3LYP method N 1s energies and experimental N 1s energies was established for the PAHs studied. Additionally, experimental ionization potentials that would correspond to electron donation trended strongly with the DFT adiabatic and vertical ionization potentials.

The testing and setup of fuel cell test station was accomplished. Bench scale membrane electrode assemblies (MEAs) were fabricated cell and achieved comparable performance to a commercial MEA constructed from similar materials. A MEA was constructed with a CNx cathode and was found to have fuel cell performance of the same order of magnitude as other graphitic carbon-nitrogen catalysts heat-treated in the presence of a transition metal.

iii

Vulcan carbon and CNx catalysts were compared in accelerated carbon corrosion by examining the current of the electrochemically active surface species hydroquinone/quione with cyclic voltammetry after extended potential holds. CNx was found to be more corrosion resistant than Vulcan carbon that is the most commonly used support in fuel cell electrodes.

iv

DEDICATION

To my friends and family, without your support this would not be possible.

v

ACKNOWLEDGMENTS

I gratefully acknowledge the teaching, guidance, and camaraderie I have received in the laboratory from all the past and present heterogeneous catalysis research group members.

I am especially indebted to those who began this project: Paul Matter, PhD and his undergraduate assistants Eugenia Wang and Maria Arias. A special thanks is given to

Eugenia Wang and Maria Arias for leading me to graduate school and towards Dr.

Ozkan's research group.

Thank you to those who I have worked with on the PEM Team: undergraduates

Doug Knapke, Jesaiah King, Katie Luthman, Hilary Marsh, Dan Valco, and Judi Keys along with high school student Kate Baker, and of course, fellow graduate students

Elizabeth Biddinger, who pushed me to organize and think critically, and Deepika Singh, who will carry on great work after me.

This coherence of a great research group would not be possible without the guidance and dedication of my advisor Professor Umit S. Ozkan. Thank you for taking me into your care and helping me to grow.

Finally, a deep thanks to my family and friends for supporting me as a graduate student. I thank my grandmother von Deak for softening the financial burden of my undergradute degree which allowed me to focus on my studies. I am indebted to Troy

Vogel and Adam Burley who accompanied me through undergraduate and graduate school and Katie Richards for her love and companionship.

vi

VITA

March 31, 1983 ………………. Born, Richmond, Virginia

June 2001 ………………………… H.S. Diploma, Buckeye Sr. High School, Medina,

Ohio

Winter 2005-Spring 2006………… Chemical Engineering Internship, Entrotech,

Columbus, Ohio

June 2006 …………………………. B.S. Chemical Engineering, Ohio State University,

Columbus, Ohio

June 2011 …………………………. M.S. Chemical Engineering, Ohio State University,

Columbus, Ohio

Septemper 2011 ……………………Ph. D. Chemical Engineering, Ohio State University,

Columbus Ohio

vii

PUBLICATIONS

Invited Papers

 E.J. Biddinger, D. von Deak, U.S. Ozkan, “Nitrogen-Containing Carbon Nanostructures as Oxygen-Reduction Catalysts,” Topics in Catalysis, 52 (2009), 1566-1574.

Journal Articles

 E.J. Biddinger, D.S. Knapke, D. von Deak, U.S. Ozkan, “Effect of Sulfur as a Growth Promoter for CNx Nanostructures as PEM and DMFC ORR Catalysts,” Applied Catalysis B – Environmental, 96 (2010), 72-82.

 D. von Deak, D. Singh, E.J. Biddinger, J.C. King, B. Bayram, J.T. Miller, U.S. Ozkan, "Investigation of sulfur poisoning of CNx oxygen reduction catalysts for PEM fuel cells," Journal of Catalysis, (2011), submitted.

 D. von Deak, E.J. Biddinger, K.A. Luthman, U.S. Ozkan, "The effect of phosphorus in CNx catalysts for the oxygen reduction in PEM fuel cells," Carbon, 48(12) (2010), 3637-3639.

 D. von Deak, E.J. Biddinger, U.S. Ozkan, "Carbon corrosion characteristics of CNx nanostructures in acidic media and implication for ORR performance," Journal of Electroanalytical Chemistry, (2011) accepted.

viii

 E.J. Biddinger, D. von Deak, H. Marsh, U.S. Ozkan, "RRDE catalyst ink aging effects on selectivity to water formation in ORR," Electrochemisty Solid-State Letters, 13 (2010), B98-B100.

 E.J. Biddinger, D. von Deak, D. Singh, H. Marsh, B. Tan, D.S. Knapke, U.S. Ozkan, "Examination of catalyst loading effects on the selectivity of CNx and Pt/VC ORR catalysts using RRDE," Journal of Electrochemistry Society, (2011) accepted.

 X. Bao, D. von Deak, E.J. Biddinger, U.S. Ozkan, C.M Hadad, "A computational exploration of the oxygen reduction reaction over a carbon catalyst containing a phosphinate functional group," Chemistry Communications, 46(45) (2010), 8621- 8623.

FIELDS OF STUDY

Major Field: Chemical Engineering

Area of Interest: Heterogeneous Catalysis

ix

TABLE OF CONTENTS

ABSTRACT ...... ii

DEDICATION ...... v

ACKNOWLEDGMENTS ...... vi

VITA ...... vii

LIST OF FIGURES ...... xvii

LIST OF TABLES ...... xxv

CHAPTER 1. Executive Introduction of Carbon-based Oxygen Reduction

Electrocatalysts ...... 2

CHAPTER 2. Research Objectives...... 3

CHAPTER 3. Literature Review of Oxygen Reduction Catalysts in Fuel Cells ...... 7

3.2 Introduction to Polymer Electrolyte and Direct Methanol Fuel Cell Technologies . 7

3.2.1 Motivation for Fuel Cell Research ...... 7

3.2.2 Introduction of Polymer Electrolyte and Direct Methanol Fuel Cell ...... 9

3.3 Fuel Cell Testing and Performance ...... 13

3.3.1 Polarization Performance Curves ...... 13

x

3.3.2 Resistance Correction with Impedance ...... 15

3.4 Current State of Technology for PEM and Direct Methanol Fuel Cells ...... 23

3.4.1 The Nafion Membrane ...... 23

3.4.2 Hydrogen Polymer Electrolyte Fuel Cells ...... 24

3.4.3 Direct Methanol Fuel Cells ...... 25

3.5 Precious Meal Oxygen Reduction Catalysts ...... 26

3.6 Non-precious Meal Oxygen Reduction Catalysts ...... 27

3.6.1 Macrocycle Non-precious Metal Oxygen Reduction Catalysts ...... 28

3.6.2 Heat Treated Macrocycle Oxygen Reduction Catalysts ...... 31

3.6.3 Non-precious Metal Oxygen Reduction Catalysts Derived From Elemental

Precusors ...... 32

3.6.4 Non-precious Metal Oxygen Reduction Catalyst Active Site Dispute ...... 53

3.6.5 Computational Chemistry ...... 64

CHAPTER 4. Experimental Methods ...... 71

4.2 Catalyst Synthesis ...... 71

4.2.1 CNx Catalyst Synthesis...... 71

4.2.2 Phosphorus Grown CNx Catalyst Synthesis ...... 74

4.3 H2S Treatment of CNx Catalyst ...... 75

4.4 Rotating Ring Disk and Rotating Disk Electrode Techniques ...... 75

xi

4.4.1 Oxygen Reduction Activity and Selectivity Testing ...... 75

4.4.2 Collection Efficiency Testing...... 81

4.4.3 Carbon Corrosion Testing ...... 84

4.4.4 Carbon Monoxide Poisoning Testing ...... 86

4.4.5 Chronoamperometric CO Poisoning Experimentation ...... 88

4.5 Fuel Cell Testing and Membrane Electrode Assembly Fabrication ...... 90

4.5.1 Membrane Electrode Assembly Fabrication ...... 90

4.5.2 Electrochemical Fuel Cell Testing ...... 99

4.6 Transmission Electron Microscopy ...... 109

4.7 X-ray Photoelectron Spectroscopy ...... 109

4.8 Raman Spectroscopy ...... 110

4.9 X-ray Absorption Spectroscopy ...... 110

4.9.1 Pellet Preparation ...... 111

4.9.2 X-ray Absorption Spectroscopy Ex-situ Collection ...... 112

4.10 Physisorption Techniques ...... 112

4.10.1 Brunauer-Emmett-Teller Surface Area Method ...... 113

4.10.2 Barrett, Joyner, and Halenda Pore Size Distribution ...... 114

4.11 Computational Analysis ...... 114

4.12 Temperature Programmed Oxidation Experiments...... 116

xii

4.13 Temperature Programmed Desorption-TPO Experiments ...... 116

CHAPTER 5. Phosphorus Doped CNx Oxygen Reduction Catalyst ...... 118

5.2 Phosphorus Doping Introduction ...... 118

5.3 Phosphorus Doping Motivation ...... 119

5.4 CNxPy Grown Over Iron Acetate and Triphenylphosphine ...... 122

5.4.1 CNxPy Grown Over Iron Acetate and Triphenylphosphine Synthesis ...... 122

5.4.2 Electrochemical Oxygen Reduction Activity Testing ...... 122

5.4.3 Transmission Electron Microscopy Imaging of CNx and CNxPy Catalysts ...... 128

5.4.4 Raman Spectroscopy Characterization of CNx and CNxPy catalysts ...... 131

5.4.5 X-ray Photoelectron Spectroscopy of CNx and CNxPy catalysts ...... 133

31 5.4.6 P Nuclear Magnetic Resonance of CNx and CNxPy ...... 134

5.4.7 Temperature Programmed Oxidation of CNx and CNxPy ...... 135

5.4.8 Extended X-ray Absorption Fine Structure of CNx and CNxPy ...... 137

5.5 Phosphorus Grown Graphites Synthesized with Different Reactant Materials .... 141

5.5.1 Phosphorus-Doped CNx Grown From Iron Phosphate ...... 142

5.5.2 Phosphorus-Doped CNx Grown From Iron Phthalocyanine and

Triphenylphosphine ...... 144

5.5.3 Phosphorus-Doped CPy Grown From Triphenylphosphine on Vulcan Carbon . 147

5.6 Phosphorus Doping Conclusions ...... 148

xiii

CHAPTER 6. Probing the Oxygen Reduction Active Site Through Catalytic Poisons . 150

6.2 Investigation of Sulfur Poisoning of CNx ORR Catalyst for PEM Fuel Cells ...... 151

6.2.1 Introduction ...... 151

6.2.2 H2S Poisoning Introduction ...... 152

6.2.3 Results and Discussion ...... 154

6.3 The Effect of Carbon Monoxide on Oxygen Reduction PEM Fuel Cell Catalysts170

6.3.1 Introduction ...... 170

6.3.2 Experimental ...... 174

6.3.3 Results and Discussion ...... 177

6.3.4 CO poisoning Conclusions ...... 185

CHAPTER 7. Examination of Rotating Ring Disk Electrode Methodology and its Impact on Reported Selectivity ...... 186

7.2 Introduction ...... 186

7.3 The Rotating Ring Disk Method of Calculating Selectivity ...... 189

7.4 Rotating Disk Electrode Koutecky-Levich Selectivity Analysis ...... 195

7.5 Comparison of RDE and RRDE Selectivity ...... 198

CHAPTER 8. Computational Chemistry ...... 199

8.2 Research Motivation ...... 199

8.3 Introduction ...... 200

xiv

8.4 Determination of Ionization Potential ...... 207

8.5 Conclusions ...... 225

CHAPTER 9. Membrane Electrode Assembly Fabrication and Fuel Cell Testing ...... 226

9.2 Catalyst Application on PEM: Hot-Press vs. Direct Application ...... 226

9.2.1 Membrane Electrode Assembly Fabrication Using the Hot-Press Method ...... 227

9.2.2 Membrane Electrode Assembly Fabrication Using Direct Ink Application ...... 229

9.3 Anode and Cathode Electrode Composition ...... 230

9.4 Gas Diffusion Lay Electrode Contact ...... 231

9.5 Bench Scale Fabrication of a Commercial MEA ...... 233

9.6 Bench Scale Fabrication of a Non-Precious Metal Catalyst MEA ...... 235

CHAPTER 10. Carbon Corrosion Characteristics of CNx nanostructures in Acidic Media and Implications for ORR performance ...... 237

10.2 Introduction to Carbon Corrosion in Fuel Cells ...... 237

10.3 Carbon Corrosion Results and Discussion ...... 239

10.3.1 Carbon corrosion on CNx materials and Vulcan carbon (Corrosion tests) ...... 241

10.3.2 Effect of Corrosion on ORR (ORR-Corrosion-ORR tests) ...... 248

10.4 Carbon Corrosion Testing Conclusions ...... 255

CHAPTER 11. Conclusions...... 256

11.2 Phosphorus Doped CNx Catalysts ...... 256

xv

11.3 Probing Oxygen Reduction Active Site Though Catalytic Poison ...... 257

11.3.1 Investigation of Sulfur Poisoning on CNx ORR Catalysts ...... 257

11.3.2 The Effect of Carbon Monoxide on Oxygen Reduction PEM Fuel Cell Catalysts

...... 258

11.4 Rotating Ring Disk Electrode Methodology's Impact on Selectivity ...... 260

11.5 Computational Chemistry ...... 261

11.6 Fuel cell testing and MEA Fabrication ...... 262

11.7 Corrosion Testing ...... 264

CHAPTER 12. Recommendations for Future Work ...... 265

12.2 Summary of the Cathode PEM Fuel Cell Catalyst Research Progress ...... 265

12.3 Improving the Activity and Current Density of CNx Materials ...... 269

12.4 In-situ X-ray Absorption on the Fe K-edge During Oxygen Reduction ...... 270

CHAPTER 13. Glossary of Acronyms ...... 275

References ...... 279

xvi

LIST OF FIGURES

Figure 1. Diagram of Direct Methanol and Polymer Electrolyte Fuel Cell. Diagram of

Direct Methanol and Polymer Electrolyte Fuel Cell...... 9

Figure 2. Polarization Curve for a Typical PEM Fuel Cell ...... 13

Figure 3. Electronic Impedance Experimental Setup; Z =voltage source, Iload=electronic load, U=cell voltage, u(t)=cell voltage response due to current frequency, i(t)=small signal harmonic current...... 16

Figure 4. Static I/V characteristics with impedance perturbation and signal around an operating current of 0.5 A/cm2 on a basic polymer electrolyte membrane fuel cell polarization curve...... 17

Figure 5. PEM fuel cell Niquest plot taken using a full frequency impedance analysis37

...... 22

Figure 6. The structure Heme B found in hemoglobin ...... 29

Figure 7. Cobalt prophine (right), Iron Phthalocyanine (left) Provided by Paul H. Matter

...... 30

Figure 8. Oxygen Functional Groups Present of the Surface of Graphite Materials ...... 40

Figure 9. The Electrochemically Active Quinone (right) /Hydroquinone (left) Oxygen

Species ...... 44

xvii

Figure 10. Oxygen Functional Groups Detected in X-ray Photoelectron Spectroscopy

(shown in energies for oxygen 1s blue are the oxygen species that have a binding energy centered at that value)11 ...... 45

Figure 11. An Illustration of the Possible Types of Nitrogen Present in Graphitic Carbon and their Corresponding N 1s Binding Energies ...... 48

Figure 12. Transmission Electron Microscopy Image of a CNx Encased Iron

Nanoparticle ...... 50

Figure 13. Carbon Nanostructure Morphologies.211 ...... 51

Figure 14. Iron-N2/C type oxygen reduction active site (courtesy of Paul H. Matter) .... 57

Figure 15. Iron-coordinated oxygen reduction active site thought to reside in graphite . 60

Figure 16. Quaternary and pyridinic nitrogen ORR active sites proposed by

Subramanian et al...... 62

Figure 17. Two step electron donation type oxygen reduction mechanism...... 67

Figure 18. Graphical Representation of Vertical and Adiabatic Ionization Potentials ... 69

Figure 19. Pyrolysis CNx Growth Experiment Setup ...... 72

Figure 20. Pyrolytic Growth of CNx (Courtesy of Elizabeth J. Biddinger) ...... 73

Figure 21. Diagram of the Rotating Ring Disk Electrode Experimental Setup ...... 78

Figure 22. Illustration of Oxygen Reduction Activity and Selectivity on an RRDE setup

...... 79

Figure 23.. Illustration of Collection Efficiency RRDE setup ...... 82

Figure 24. Electrochemical Rotating Disk Electrode Carbon Corrosion Setup (inset; the hydroquinone/quinone reaction and standard reaction potential) ...... 85

xviii

Figure 25. Electrochemical Rotating Disk Electrode Corrosion Setup...... 87

Figure 26. Diagram of Hydrogen Polymer Electrolyte Fuel Cell...... 100

Figure 27. Static I/V characteristics with impedance perturbation and signal around an operating current of 0.5 A/cm2 on a basic polymer electrolyte membrane fuel cell polarization curve with a small sinusoidal perturbation ...... 107

Figure 28. Cathodic scans of CNxPy catalyst-coated RRDE in Oxygen Saturated

Electrolyte ...... 123

Figure 29. CNxPy Selectivity measured during activity testing. Legend: molar ratio of

P/Fe in the growth media to enhanced ...... 124

Figure 30. Cyclic Voltammograms of CNx and CNxPy RRDE in Argon Saturated

Electrolyte ...... 125

Figure 31. Full Fuel Cell Testing of CNx and CNxPy with Oxygen and Hydrogen

Reactants ...... 127

Figure 32. TEM Images of CNx and CNxPy Carbon Nanofibers ...... 129

Figure 33. TEM Images of CNx and CNxPy Carbon Nanostructures ...... 130

Figure 34. Raman Spectra of CNx and CNxPy catalysts, normalized by G-band intensity

...... 132

Figure 35. Temperature Programmed Oxidation of CNxPy catalysts in 10%O2/He ...... 136

Figure 36. EXAFS comparison of Fe K-edge in CNx Fe:0.33P Verses Standard Materials

...... 138

Figure 37. EXAFS comparison of Fe K-edge in CNx and CNx P-doped materials ...... 140

xix

Figure 38. Cathodic scans of CNxPy-Phosphate catalyst-coated RRDE in Oxygen

Saturated Electrolyte ...... 144

Figure 39. Cathodic scans of CNxPy-N/Fe-P-C and P/Fe=0.33 CNxPy-triphenylphosphine catalyst-coated RRDE in Oxygen Saturated Electrolyte ...... 146

Figure 40. Cathodic scans of CPy catalyst-coated RRDE in Oxygen Saturated Electrolyte

...... 148

Figure 41. Oxygen reduction activity of Pt/Vulcan carbon catalysts. Inset: Selectivity for water formation ...... 155

Figure 42. Oxygen reduction activity of CNx catalysts. Inset: Selectivity for water formation ...... 157

Figure 43. X-ray photoelectron spectra in the N 1s region for untreated and heat treated

CNx catalysts under different atmospheres ...... 159

Figure 44. X-ray photoelectron spectra of the S 2p region for untreated and H2S treated

CNx catalysts under different atmospheres ...... 161

Figure 45. Evolution of CO2 (m/z = 44 amu), NO2 (m/z = 46 amu) and SO2 (m/z =64 amu) as a function of temperature during temperature programmed oxidation of untreated, H2S, N2, and H2 treated CNx...... 163

Figure 46. XANES Fe K-edge spectra for untreated, H2S-, and H2-treated CNx ...... 165

Figure 47. XANES Fe K-edge spectra for, iron foil, iron pthalocyannine, iron carbide and untreated CNx. Inset: TEM of iron nanoparticle encapsulated in CNx catalyst ...... 166

Figure 48. EXAFS Fe K-edge spectra for H2S treated, and untreated CNx ...... 167

xx

Figure 49. EXAFS Fe K-edge spectra for untreated CNx, iron phthalocyanine, iron metal and iron carbide...... 168

Figure 50. CNx, 20 wt% Pt on Vulcan carbon, 2 wt% Fe Iron phatholocyanine on Vulcan carbon RDE at 0.3 V vs. NHE initially the electrolyte saturated with oxygen at time 1590 seconds, 1000 rpm, and 0.5 M H2SO4 (aq)...... 177

Figure 51. CNx (left) and Platinum on Vulcan Carbon (right) RDE at 0.3V vs. NHE initially the electrolyte saturated with oxygen at time <1.5 min, at 1.5 min gas flow changed to argon (a), carbon monoxide (b), argon and oxygen (c) or carbon monoxide and oxygen (d), at 26.5 min gas flow changed to oxygen, 1000 rpm, 0.5M H2SO4 (aq).180

Figure 52. a) CNx and b) 20wt%Platinum on Vulcan carbon RDE cathodic potential scan at 5 mV/s from 1.2 V to 0.0 V vs. NHE in oxygen saturated 0.5 M H2SO4 before and after

0.3 V vs. NHE potential hold in O2-CO-O2 at 1000 rpm...... 182

Figure 53. CNx RDE cathodic potential scan at 5mV/s from 1.2 V to 0.0 V vs. NHE in oxygen, carbon monoxide & oxygen, oxygen after treatment, then argon saturated 0.5 M

H2SO4 at 1000rpm...... 183

Figure 54. Carbon monoxide, 28 atomic units, pulse chemisorption on high performance

20 wt%Pt/VC (BASF) at 25 ºC spectra, 9 pulses shown (provided by Deepika Singh) . 184

Figure 55. The ring and disk currents in oxygen saturated 0.5 M H2SO4 of CNx (left) and

20wt%Pt on Vulcan carbon (right) (Insets: Selectivity of the oxygen reduction reaction)

...... 191

xxi

3- Figure 56. Disk and ring currents as a function of disk potential for the [Fe(CN)6]

4- 2 /[Fe(CN)6] redox couple. Curves shown are for 426 μg/cm catalyst loading. a) CNx, b)

20 wt%Pt on Vulcan carbon.330 ...... 193

Figure 57. Experimental collection efficiency as a function of loading. The dashed line indicates the theoretical collection efficiency as reported by the manufacturer. a) CNx catalyst b) Pt/VC catalyst.330 ...... 194

2 Figure 58. Disk currents in oxygen saturated 0.5 M H2SO4 of CNx with 426 μg/cm catalyst loading ...... 196

Figure 59. Koutecky-Levich Analysis of CNx in Oxygen Saturated 0.5 M H2SO4 with

426 μg/cm2 catalyst loading. The slopes for 2 and 4 electron transfer for ORR were taken from:316,331 ...... 197

Figure 60.. Purposed causes of ORR catalytic activity in graphitic carbon-nitrogen catalysts; a) Stelko,172,177,337 b) Lefévre, 92 c) Montoya,261 Pels,256, d) Subramanian 169

(figures b), c), and d) courtesy of Paul H. Matter) ...... 203

Figure 61.. Types of carbons that are commonly de-convoluted in XPS ...... 204

Figure 62. XPS N-1s spectra of CNx catalyst (Courtesy of Paul H. Matter) ...... 206

Figure 63. Illustration of the Difference in Vertical and Adiabatic Ionization Potentials

...... 208

Figure 64. Comparison of Experimental and Adiabatic AM1 / B3LYP/6-31G* Scaled

Ionization Potentials grouped by nitrogen atoms...... 220

Figure 65. Comparison of Experimental and Vertical AM1 / B3LYP/6-31G* Scaled

Ionization Potentials grouped by nitrogen atoms...... 222

xxii

Figure 66. Comparison of Experimental and Vertical B3LYP/6-31G* Scaled Ionization

Potentials Grouped by the Number of Heavy Atoms ...... 223

Figure 67. Nitrogen 1s Energies, B3LYP/6-31G* verses XPS...... 224

Figure 68.. Polarization Curve of a Commercial MEA 1 mgPt/cm2 anode and cathode,

Nafion 212 membrane; Single Cell test in an Arbin 5cm2 geometric area fuel cell fixture,

(Break-in procedure and electrochemical testing are the same as Figure 69.) ...... 232

Figure 69. MEA Commercial Compared to Lab Fabricated; Single Cell test in an Arbin

5cm2 geometric area fuel cell fixture, Break-in procedure taken from Electrochem MEA manufacturer ...... 234

2 Figure 70. Un-optimized CNx catalyst fuel cell test comparison; Single Cell 5cm geometric area fuel cell fixtures...... 235

Figure 71. Carbon corrosion and oxygen reduction reactions and their standard potentials19,25 ...... 239

Figure 72. The Mechanisms of Deactivation of Platinum Supported on Carbon Catalysts

...... 240

Figure 73. Quinone (left) / Hydroquinone (right) Couple on Carbon Graphite...... 243

Figure 74. Evolution of quinone/hydroquinone species on Vulcan Carbon-XC72 ...... 244

Figure 75. Evolution of the quinone/hydroquinone species on CNx grown on Fe/Vulcan carbon ...... 245

Figure 76. Carbon Corrosion of CNx Catalysts and Vulcan Carbon Support reported as change in the intensity of the quinone/hydroquinone electrochemically active couple (at

xxiii

0.627 V vs. NHE) in the anodic current of intermittent CVs with potential hold of 1.2 V vs. NHE in a rotating disk electrode experimental setup ...... 246

Figure 77. Oxygen Reduction Activity of four catalysts at (1000rpm, 10mV/sec, 426

μg/cm2 catalyst loading) ...... 247

Figure 78. RRDE activity and selectivity comparison before and after corrosion testing,

(above) the ring current (below) disk current density, data collected in oxygen-saturated

2 0.5 M H2SO4 (1000 rpm with a catalyst loading 426 μg/cm , 10 mV/s.) and the ring current at the Pt ring held at 1.2 V vs. NHE in oxygen saturated 0.5 M H2SO4. After Ar background subtraction ...... 249

Figure 79. Number of electrons transferred per O2 molecule calculated from the ring and disk currents from the oxygen reduction activity experiments shown in Figure 77...... 250

Figure 80. Quinone/hydroquinone detection CVs after 0 and 48 hr 1.2 V vs. NHE potential holds for; a) fresh and ORR tested CNx grown on Fe/Vulcan carbon. Tested in

2 0.5 M H2SO4, at @ 1000 rpm, 10mV/sec, 426 μg/cm catalyst loading...... 252

Figure 81. Quinone/hydroquinone detection CVs after 0 and 48 hr 1.2 V vs. NHE potential holds for fresh and ORR tested Vulcan carbon XC-72.Tested in 0.5 M H2SO4, at

@ 1000 rpm, 10 mV/sec, 426 μg/cm2 catalyst loading...... 253

Figure 82. RRDE activity and selectivity comparison before and after corrosion testing disk current. Data collected in oxygen-saturated 0.5 M H2SO4 (1000 rpm with a catalyst loading 426 μg/cm2, 10 mV/s.) and the ring current at the Pt ring held at 1.2 V vs. NHE in oxygen saturated 0.5 M H2SO4. Data are presented after argon background is subtracted...... 254

xxiv

LIST OF TABLES

Table 1. Energy Storage Fuels and Materials for Mechanical Transportation Energy taken from Jaouen et al.4 ...... 7

Table 2. XPS surface concentration of CNx and CNxPy catalysts ...... 134

Table 3. Compositional analysis of CNx catalyst X-ray photoelectron Spectroscopy ... 162

Table 4. Computationally determined energies of compounds at AM1 level of theory 209

Table 5. Computationally determined energies of compounds at B3LYP/6-31G* level of theory ...... 213

Table 6. Ionization Energies of Various Polycyclic Aromatic Hydrocarbons ...... 216

Table 7. The types of Rotating Disk Electrochemical Experiment Performed ...... 242

25

CHAPTER 1. Executive Introduction of Carbon-based Oxygen Reduction Electrocatalysts

While fuel cells offer a more efficient and effective energy conversion alternative than combustion engines, there are still important technological barriers that must be overcome before their wide-spread application, many of which need to be addressed through fundamental research. For PEM fuel cells most of these barriers are associated with ORR that takes place over the cathode electrocatalysts. The ORR kinetics are slow, even with Pt-based catalysts. The enormous cost and the limited availability of platinum motivate the development of non-precious metal alternatives for ORR catalysts.

Heteroatom containing carbon-based catalysts have been shown to have promising oxygen reduction activity although the source of this activity and the mechanistic steps of oxygen reduction on these materials is largely unknown.

This work investigated of the nature of active sites in heteroatom containing nano- structured carbon catalysts in ORR in PEM fuel cell cathodes using fundamental approaches based on catalysis and electrochemistry. The main objective was arriving at a molecular level understanding of the electrocatalytic phenomena involved in the oxygen reduction reaction over nano-structured carbon catalysts which contain one or more 1 heteroatoms, especially nitrogen and phosphorus. The nature of the active sites and the ways to create these active sites for oxygen reduction reaction was investigated in an effort to control electrocatalytic properties by molecular tailoring of the carbon nanostructure and its surface moieties. The practical relevance of the work lies in the need for developing precious-metal free ORR catalysts for PEM fuel cells and direct methanol fuel cells (DMFC) and the long-term objectives include building a knowledge base that would facilitate the production of active, stable and inexpensive cathode electrocatalysts that would realize the wide-spread application of fuel cells for energy utilization.

2

CHAPTER 2. Research Objectives

A targeted investigation of non-precious metal, carbon-based catalysts used to electrocatalytically reduce oxygen in direct methanol and polymer electrolyte fuel cells encompasses the scope of research performed. Carbon-based oxygen reduction catalyst materials have been demonstrated as a viable, although less catalytically active, alternative to prohibitively expensive platinum-based ORR catalysts in PEM and direct methanol fuel cells.1 The existing platinum loadings used to accelerate the sluggish ORR roughly encompass 60% of the total fuel cell capital cost limiting PEM fuel cells to only niche applications.2 Although carbon catalysts are much less expensive than Pt and Pt alloys, the origin of oxygen reduction is debated, thereby limiting the development of non-precious metal catalysts.3 Greater insight into the nature of electrocatalytic dioxygen reduction on carbon-based catalysts could elucidate cause of their intrinsic activity and lead to the further development of these materials.

Non-precious metal catalysts are primarily composed of nanostructured graphines that have heteroatoms, incorporated on the surface and/or the bulk which alter material properties and imparts electrocatalytic oxygen reduction activity. The synthesis

3 and composition of these materials vary greatly, but a nitrogen and a carbon source, a transition metal, and oxygen-free heat treatment of these precursors are necessary to form a highly active catalyst.1,3,4

The research efforts have been focused on addressing three major questions:

I. The nature of the active site and the role of transition metal left in the carbon structure from the growth process

The high-temperature treatment used to stabilize non-noble metal catalyst materials facilitates the formation of many atomic arrangements any one of which could be responsible for the oxygen reduction activity, due to this effect the exact bonding configuration that facilitates high ORR activity is still actively debated.1,3,5 The active site debate can be split into two main camps; one believes that the metal growth catalyst forms a Metal-N2 or Metal-N4 type active site during the oxygen free heat treatment, and one that believes that nitrogen itself doped into the graphitic matrix imparts oxygen reduction activity. Two common metal poisons were used in an exclusionary approach by attempting to deliberately poison the catalysts with sulfur or carbon monoxide to determine whether a transition metal is present in the active site for oxygen reduction in

CNx materials. Sulfur is a well-known catalyst poison that deactivates iron-based catalysts for Fischer-Tropsch, water-gas shift, ammonia synthesis, ammonia decomposition, and iron carburization.6-9 Similarly, a hydrogen sulfide treatment can be expected to have a similar poisoning effect on any iron containing active sites present on non-noble metal ORR electrocatalysts. Fe-N4 systems such as hemoglobin and

4

10 myoglobin have ~200 times higher binding affinity for carbon monoxide than O2. This suggests that the competitive adsorption of carbon monoxide on graphitic carbon nitride oxygen reduction electrocatalyst in the presence of oxygen could be used as a probe for

Fe-N2 and Fe-N4 type active sites in catalysts. If the oxygen reduction activity of these catalysts is reduced by a carbon monoxide or sulfur poisoning treatment than a Metal-N2 or Metal-N4 type oxygen reduction active site would plausible.

II. The role of heteroatoms incorporated into the carbon matrix

The inclusion of different heteroatoms during the synthesis of carbon nanostructures not only affects the properties of the formed material but also the carbon deposition of that material. For example, it has been demonstrated that sulfur doping of carbon-nitrogen graphite enhanced the catalyst yield while leaving oxygen reduction activity unchanged.11 Sulfur was thought to bond with transition metal iron in the supported growth media used catalytically to grow the carbon nanostructures, lowering the eutectic point and allowing for a more mobile and facile growth catalyst.12

Phosphorus was hypothesized to alter carbon growth through a similar but opposite mechanism that reduces carbon growth.13 Phosphorus and nitrogen are both n-type semiconducting electron donors when present in a graphitic matrix.13 Phosphorus incorporation into CNx effects the electrocatalytic reduction of dioxygen. Phosphorous doping promotes large corrugation in graphitic structures due to the positive curvature of incorporation14, which is evident in relative intensity in the Raman spectroscopy disorder band.13,15,16 Phosphorus incorporation into the graphite matrix could therefore result in

5 more chemically active carbon edge planes and textural surfaces where reactions are more likely to proceed.

III. Corrosion characteristics of CNx catalysts

Carbon supports are commonly used as a support for platinum and platinum- alloyed cathode catalysts in PEM fuel cells. Carbon support materials are chosen due to their low cost, high surface area, thermal and electrical conductivity, and high stability under fuel cell operating conditions.17,18 The decrease in fuel cell performance during operation is often attributed to the carbon corrosion in the cathode catalysts layer where oxygen reduction takes place.17 Fuel cell operating conditions appear to make the carbon support prone to carbon corrosion, in practice, the actual carbon corrosion is slow.17,19

Carbon corrosion has been studied for various carbon support materials with20,21 and without Pt catalysts,22,23 but the corrosion of nitrogen-doped carbon oxygen reduction catalysts was not previously studied. To elucidate the corrosion resistance of Pt-free, nitrogen-doped graphite oxygen reduction catalysts as they compare to Vulcan carbon, the most prevalent type of carbon support, an electrochemical acidic medium simulating the cathode oxygen reduction fuel cell environment in fuel cells was examined.

6

CHAPTER 3. Literature Review of Oxygen Reduction Catalysts in Fuel Cells

3.2 Introduction to Polymer Electrolyte and Direct Methanol Fuel Cell Technologies

3.2.1 Motivation for Fuel Cell Research

Fuel cells have the potential to radically alter the contemporary energy landscape, but several infrastructural, economic, and technological challenges must be overcome for fuel cells to compete with contemporary energy conversion devices. Since fuel cells directly convert electrical energy from stored chemical energy, they can achieve a high power density while still remaining highly portable, making fuel cells advantageous for distributed and mobile power application, Table 1. It can be seen that combustion engine fuels (gasoline, biodiesel, ethanol) have a much lower

Table 1. Energy Storage Fuels and Materials for Mechanical Transportation Energy taken from Jaouen et al.4 row Gasoline Biodiesel Ethanol Compressed Lithium-ion Nickel-metal H2 (690 bars) battery hydride battery Energy 13.4 11.1 8.3 39.36, 1.87 0.4 - 14, <0.25 0.244, <0.15 density1 (kWh/kg) Density 0.7 0.88 0.79 0.06156, - - (kWh/L) 0.0227 Energy 9.7 9.8 6.6 2.46, 0.97 -, <0.45 -, <0.255 Density1 (kWh/L) Efficiency (%) 20.02 25.02 20.02 50.0c 85.0 85.5 1Based on higher heating value. 2with internal combustion engine. 3with fuel cell. 4electrode materials only. 5with packaging. 6H2 only. 7Assuming a storage system with gravimetric and volumetric capacities of 4.5 wt.% H2 and 22 gH2/L, respectively. energy density and efficiency than hydrogen when used in a fuel cell. Hydrogen fuel cells also have a much higher energy density that lithium-ion and nickel-metal hydride batteries, although batteries have a higher efficiency. Fuel cells utilize a coupled

7 oxidation-reduction reaction where electrons are created by one reaction, consumed by the other, and flow through an external circuit to do work. The oxygen reduction reaction is almost universally used as cathodic reaction in fuel cells because oxygen is abundant, free, air-born. Since most fuels use the oxygen reduction reaction, they tend to be differentiated by the anode fuel for the oxidation reaction. The particular fuel used typically dictates the membrane, operating temperature, and supporting infrastructure of the fuel cell. For instance, a hydrogen fuel cell utilizes gas phase hydrogen fuel for the oxidation reaction coupled with the oxygen reduction reaction (ORR). The fuel distribution infrastructure that currently exists is tailored to liquid phase petroleum, which hinders the market uptake of any competitive performance gas phase reactant fuel cells.

Although there are several fuel cell types, the United States' Department of Energy has focused on the development of polymer electrolyte membrane (PEM) fuel cells due to their potential application in the portable energy sector. PEM and direct methanol fuel cells (DMFCs) offer adaptable sources of power utilization that could be applied to batteries and combustion engine markets. The largest developmental barrier for PEM fuel cells is associated with the oxygen reduction over the cathode electrocatalyst. The sluggish cathode ORR is accelerated by fiscally prohibitive loadings of platinum and platinum alloys which are necessary at the low temperatures (< 100 ºC) that are defined by the boiling of water within the proton conducting membrane. The expense and availability of platinum motivate the development of non-precious metal ORR catalysts.

8 3.2.2 Introduction of Polymer Electrolyte and Direct Methanol Fuel Cell

Direct methanol and polymer electrolyte membrane, also known as hydrogen, fuel cells have a very similar structure. Both fuel cell types have a "sandwich" type of configuration where two catalyst layers are adhered to either side of an electrically

Figure 1. Diagram of Direct Methanol and Polymer Electrolyte Fuel Cell. Diagram of Direct Methanol and Polymer Electrolyte Fuel Cell. insolating, proton conducting polymer electrolyte membrane, Figure 1. A humidified sulfonated tetrafluoroethylene based fluoropolymer-copolymer, commercially available as Dupont's Nafion, acts as the polymer electrolyte membrane. Stationary fluoride anions within electrically insulating Nafion polymer electrolyte membrane conduct protons

9 produced in either the methanol or hydrogen anode oxidation reaction to the cathode of the fuel cell.

Each catalyst layer is composed of a catalyst supported on an electrically conducting carbon graphite. Carbon support materials are used due to their low cost, high surface area, thermal and electrical conductivity, and high stability under normal PEM fuel cell operating conditions.17,18 On the outside of each catalyst layer is a gas diffusion layer composed of polytetrafluoroethylene treated carbon paper, Toray paper, which is used to diffuse the reactants in a controlled manner, regulate water management, and conduct electrons that participate in cathode and anode reactions.24 Polytetrafluoroethylene, more commonly known as Dupont's Teflon, adds hydrophobicity to gas diffusion layer allowing liquid water to be effectively diffused away from catalyst layers. Outside of the gas diffusion layers are graphite current collection plates with serpentine gas flow channels carved into the side that faces the gas diffusion layer, shown in grey in Figure 1.

Electrical interconnects are attached to these electrically conducting plates to complete the circuit between the coupled anode and cathode reactions. The flow channels are used to uniformly deliver the humidified reactant gasses or methanol/water mixture. The components of hydrogen fuel cells and their effect on performance will be further described in CHAPTER 9: Development of Highly active non-noble metal MEAs.

The overall efficiency of polymer electrolyte and direct methanol fuel cells is restricted by the kinetics of the oxygen reduction reaction on the cathode. Although polymer electrolyte and direct methanol fuel cells are similar they are differentiated by their anode oxidation reactions. PEM and DMFCs anode utilize the hydrogen and

10 methanol oxidation reactions, respectively. The standard potentials at which anode oxidation reactions take place are below.

+ - 25 4H2  4 H + 4e (0.000 V vs. NHE)

+ - 25 CH3OH + H2O  6H + 6e + CO2 (0.043 V vs. NHE)

Both reactions shown above proceed quickly with nearly no efficiency losses during normal operation, which is indicated by the nearly zero standard reaction potential verses the normal hydrogen electrode (NHE) for both reactions. Both anode reactions are catalyzed by a platinum-based, carbon-supported catalyst layer with a low (0.05 mgPt/cm2) platinum loadings.26 The DMFC anode catalyst is typically platinum-alloy with ruthenium to prevent carbon monoxide poisoning.26,27 Ruthenium is thought to alter the interaction between the support and metal catalyst, which can modify the electric structure of catalytic metals, and in turn retards coking during methanol oxidation.27 The hydrogen fuel for PEM fuel cell anodes is typically humidified so that proton transport can be conducted through hydration shell in the polytetrafluoroethylene (PTFE) membrane. Likewise, methanol fuel in DMFC anodes is mixed with water to hydrate the

PTFE membrane and limit methanol diffusion through the membrane.28

Protons generated by the anode are conducted through a hydrated, electrically insulating (PTFE) membrane. The hydrated sulfonic acid groups supported on a tetrafluorethylene backbone in PTFE allows protons to hop from one acid site to another to allow proton transport through the PTFE. The PTFE has the duel purposed of

11 electrically separating the anode and cathode reactions while facilitating the transport of protons from the anode that are reacted in the cathode.

Protons generated at the anode pass through the electrically insulating polymer electrolyte membrane while the electrons produced at the anode pass through an external circuit. The electrons and protons produced at the anode are recombined with oxygen and consumed in the oxygen reduction reaction in the cathode. The consumption of the electrons and protons drives the flow of electrons which do work through an external circuit. The ORR, that takes place in the cathode where oxygen reduction reaction proceeds to form water and hydrogen peroxide, consumes the protons and electrons produced in the anode.

+ - 25 O2 + 4 H + 4 e  2 H2O (1.229 V vs. NHE)

+ - 25 O2 + 2 H + 2 e  H2O2 (0.695V vs. NHE)

The lower standard reaction potential and the corrosive nature of hydrogen peroxide reduces the overall fuel performance making hydrogen peroxide an undesired product.

Since both reactions simultaneously take place on the cathode surface, the selectivity of

ORR is important in PEM fuel cells.

12 3.3 Fuel Cell Testing and Performance

3.3.1 Polarization Performance Curves

The amount of usable work available from a fuel cell is determined by the rate of electron consumption and the potential difference across the anode and cathode. The highest theoretical potential difference achievable is ~1.2 V vs. NHE when O2 is completely reduced to water in the cathode and H2 is completely oxidized in the anode.

However, the actual voltage is significantly lower due to several factors that lead to

"potential loss" or "overpotential".

Figure 2. Polarization Curve for a Typical PEM Fuel Cell

13 The highest achievable potential in a working membrane electrode assembly (MEA) is when an extremely small amount of current is passed through the cell and is known as the open cell potential, Figure 2. Even when operating at open cell potential the small electrical load creates a small loss in kinetic performance making the theoretical potential unachievable. Reaction rate limited power losses occur abruptly at high potentials where work is most thermodynamically useful and are attributed to catalytic limitations in the

PEM fuel cell. Power losses due to Ohmic resistances are due to the inherent electrical resistance of the fuel cell components. Mass transfer limited power losses are mainly due to proton transfer through the PEM and product water management restricting reactant transfer to the electrocatalysts in the cathode and anode. Polarization graphs are used as a common measure of fuel cell performance because the power density of a fuel cell can be determined at any potential of interest by multiplying that potential by the geometric current density at that potential. If higher powers are required for a particular application the amount of operational fuel cell area can be increased accordingly.

Fuel cell performance curves are obtained through ether chronoamperometric potential holds or potential sweep electrochemical techniques. To obtain steady state fuel cell polarization curves, chronoamperometric techniques are recommended by the United

States Fuel Cell Council.29 This council attempted to create a uniform PEM fuel single cell testing protocol by working with the national laboratories and industry leaders in an effort to standardize fuel cell performance polarization curves. This was done in an effort to accelerate PEM fuel cell development by establishing a common metric of performance. In brief, the protocol calls for the fuel cell to be humidified under stochiometric reactant flows until the open circuit potential stabilized. After the current 14 stabilized at open cell potential the fuel cell potential was stepped down from the open cell potential with 15 minute potential holds while the current was collected. The last 5 minutes of collected current data at each potential hold step was averaged generate a single point on the polarization curve. To correct for the internal fuel cell resistance at the end of each current hold, a full impedance measurement with a range of frequencies is typically made. It has also been demonstrated that a transient linear voltage scan can be used to generate the performance polarization curve.1,30,31 It should be noted that capacitance was included in the recorded resistance when using non-steady state values to determine the polarization curve. This will of course translate to a higher performing polarization curve than was actually achievable during fuel cell use. The impedance measured during the voltage scan will also be transient if measured during the dynamic voltage scan.24 Only a single frequency was used during this type of analysis since there would not be enough time at each potential for a full frequency impedance analysis.

Although with this technique is possible to perform an impedance analysis, it is not commonly used by most researchers in the hydrogen and direct methanol fuel cell field.

3.3.2 Resistance Correction with Impedance

3.3.2.a Introduction to Electrochemical Impedance Resistance

The total resistance of the fuel cell and each of its components can be determined through electrochemical impedance spectroscopy (EIS). The processes that occur in PEM

15 fuel cell systems involve complicated multi-step, inter-dependant reactions that can proceed along several parallel reaction pathways. In the case of advanced multi- component electrodes, no one-reaction step is rate limiting and dictates overall fuel cell performance. The contribution of each reaction step to the overall fuel cell performance is governed through the complex interconnectivity mass transport and reactions, the operating conditions, and physical design of the fuel cell. In order to compare the fuel cell polarization curves using differing fuel cell fixtures and operating conditions, the contributions of internal resistance must be determined for each fuel cell through EIS and corrected for in the resistance-free polarization performance curve.

Figure 3. Electronic Impedance Experimental Setup; Z =voltage source, Iload=electronic load, U=cell voltage, u(t)=cell voltage response due to current frequency, i(t)=small signal harmonic current. 16 A basic experimental arrangement of EIS is shown in Figure 3. The electrochemical fuel cell is represented by a voltage source with a non-linear polarization resistance, Z. The system current source is operated at a well-defined operating point along its I/V characteristics. This allows for a frequency current or voltage to be passed through the fuel cell fixture to receive a voltage or current response, respectively. Figure

4 shows the frequency voltage response for a frequency current input on a polarization curve at a fixed current density.

Figure 4. Static I/V characteristics with impedance perturbation and signal around an operating current of 0.5 A/cm2 on a basic polymer electrolyte membrane fuel cell polarization curve This technique works by passing an alternating current signal with a small sinusoidal perturbation through a fuel cell at steady state and measuring the phasor quality of the voltage output. A small-signal harmonic current; 17 i(t) = i0cos(t) is added to the direct current, i, by the summing amplifier. The steady state current at which the fuel cell is operating is represented by i0. Variable time is represented by t. The frequency of the input current is . In polymer electrolyte membrane fuel cell experiments, the current source is usually operated in a galvanostatic mode, electronically controlling the applied current that passes through the fuel cell.

The alternating current causes a response voltage signal, shown in the equation below, that is measured by an amplifier with filter circuits that eliminate the direct current components of the signal.

u(t)= u0() cos(t+()).

The constant value voltage, u0, and voltage phase shift, , are functions of the input frequency, . The phase shift response caused by impedance tested is represented by . The response voltage of chronovoltammetric i0 is u0.

Theoretically, the fuel cell being tested can also be operated potentiostatically by applying a given potential difference to the electrodes. In this case, the response variable would be amperometric. The sinusoidal current function causes a phase shift in the resulting wave response potential function at current load chosen for the impedance test,

Figure 4. Hydrogen and direct methanol fuel cell testing was typically impedance corrected by operating the fuel cell in chronovoltammetic mode while passing an constant current with a small frequency wave through the fuel cell.

18 The electrochemical interface was used to generate the perturbation signal and the

frequency response analyzer determines impedance response of this perturbation. The

impedance is given by the ratio of the Laplace transforms of u(t) and i(t) for the particular

sample frequency used, .

  st t  u0 cos[t  ]e dt  e dt 0 u0 j 0 j( ) Z    e  | Z() |e  Re{Z()}  j Im{Z()} st i0 t  i0 cos(t)e dt  e dt 0 0 Where; s =  + j,  < 0

 The imaginary number j is equal to -11/2.

The resulting impedance is a complex number representing both the amplitude

and phase signal output with regard to the input sinusoidal current which can be used

over a range of frequencies. The sample frequencies should be chosen such that they

cover the dynamic range of the relevant processes acting in the system under

investigation. However, the choice of frequencies is limited by the experimental

conditions and capability of the equipment used. The frequency range should be sampled

logarithmically, with about five to ten samples per frequency decade.24 This gives the

optimal resolution to the relevant impedance features in PEM fuel cells.

Alternatively to sine wave excitation, step current fluctuation functions can also

be used as the input perturbation. For this instance the input and response signals are then

recorded in the time domain. In principle, the investigation of dynamic systems in the

time-domain delivers the same amount of information as in the frequency domain due to

the related response of the impedance by Fourier transformation. However, these

19 methods are less accurate since sample frequencies are not directly controlled; yet a

Fourier analysis can calculate the impedance values of these frequencies. To cover a wide bandwidth of relaxation frequencies with an acceptable resolution a flexible sampling rate is required, making the analysis highly dependent on the sample being studied. This makes time-domain methods less suitable for the characterization of a complex electrochemical fuel cell.

Impedance measurements are typically represented by a Nyquist-plot, the imaginary part verses the real part of impedance on the complex impedance plane. For these graphs the frequency is the implicit variable. Each data point represents one distinct frequency. Because impedance in PEM fuel cells typically displays capacitive behavior rather than inductive behavior, it has become customary to plot the negative imaginary part rather than the positive imaginary part. The radial distance of the graphed Nyquist- plot, see Figure 5, is the total resistance of the fuel cell, R. Using Ohm's law;

V=I*R at a fixed current the resistance corrected potential can be determined, where the potential in volts is "V", the current in amps is "I", and the radial resistance in Ohm's determined by the Nyquist-plot is "R".

3.3.2.b How Impedance Measurement are Made in PEM Fuel Cells

20 The area-specific total resistance of a PEM fuel cell should be in the order of 0.1-

1 /cm.224 Therefore, commercially available frequency response analyzers have to be operated at the limit of their measurement capacity if the electrode area exceeds 10 cm2.

If the total resistance of the cell is significantly smaller than 100 m, the contribution of individual electrochemical processes can hardly be detected.

There are two common methods in PEM fuel cell testing used to correct for the

EIR during fuel cell testing to determine the resistance free potential, the Ecell-iR. Both methods determine the overall fuel cell resistance, but the frequencies and currents studied are different. The two electrochemical methods used to obtain the polarization curves require the use different impedance techniques to obtain a resistance free potential polarization curve.

To correct for the internal fuel cell resistance at the end of each current hold, a full impedance measurement with a range of frequencies is typically made. Most researchers use full impedance to determine PEM fuel cell resistance, ~10 kHz-5 mHz, during each current step.32-36 The resistance is determined by finding the resistance diameter of the generated Nyquist plot, Figure 5. In Figure 5 the resistance was found at

21 the open cell potential, but resistance

Figure 5. PEM fuel cell Niquest plot taken using a full frequency impedance analysis37 measurements can be taken at any operating potential. Other researchers have used a single high frequency impedance measurement at the end of each current hold to determine the fuel cell resistance and construct a resistance-free polarization curve.2 The advantage of a single frequency technique is that the galvanostat only needs one impedance frequency instead of the full frequency range capability needed to generate a

Nyquist plot. The resistance-free current is resolved through the following equation;

2 EiR-free= Ecell + Eohmic = Ecell + iR

The resistance-free potential is represented by EiR-free. The measured fuel cell potential is

Ecell. The Ohmic potential loss is Eohmic. The Ohmic potential loss was determined by taking the product of the current, i, at the measured potential with the impedance-

22 determined resistance, R. This method of determining the fuel cell resistance at each current that was used to generate the polarization curve is the preferred method because the fuel cell resistance is determined at each current.

Some researchers use full impedance ranges at open cell potential to determine the full fuel cell resistance at open cell potential.37 Other researchers use chronoamperometric full impedance ranges at a fixed potential to determine the fuel cell resistance.38,39 Both of these techniques rely on determining only a single resistance of the fuel cell, but this does not capture the behavior of the complex interconnectivity governing fuel cell performance at each operating condition. The assumption is that the fuel cell resistance is uniform over the entire potential range, but this has been shown to be inaccurate.32-36

3.4 Current State of Technology for PEM and Direct Methanol Fuel Cells

3.4.1 The Nafion Membrane

Humidification in the anode feed is required due to water removal from hydration shell proton transport from the anode to the cathode. The most ubiquitous supplier of this membrane material is Nafion produced by Dupont. The thickness of this membrane dictates the diffusive cross-transport from anode to cathode and the converse. A thinner membrane is preferred to decrease the mass transport resistance of the fuel cell, but the membrane cannot be too thin because PEM membrane degrades in the acidic, oxidizing, 23 high potential environment on the cathode side of these fuel cells. The proton conductivity of the Nafion membrane is a strong function of the relative humidity of the reactant gases. For instance, at 45 ºC with reactant gases with a relative humidity of 20%, the proton conductivity is 1 mS/cm and at 80 ºC with reactant gases with a relative humidity of 100%, the proton conductivity is 100 mS/cm.24 The lower end of the fuel cell operating temperature is limited by Nafion's humidification requirement for high proton conductivity, otherwise operation of the fuel cell will be severely mass transfer limited by the protons needed for oxygen reduction at the cathode. At temperatures greater than 100 ºC liquid water within the Nafion could boil causing mechanical destruction of the polymer electrolyte membrane layer, so to avoid the effect of localized hot spots, the maximum operating temperature for direct methanol and hydrogen fuel cell is 80 ºC.24

3.4.2 Hydrogen Polymer Electrolyte Fuel Cells

The current state of hydrogen PEM fuel cell technology uses a very thin sulfonated tetrafluoroethylene proton conducting membrane, ~2 Mils, with an anode with low precious metal catalyst loading, ~0.05 mgPt/cm, and a cathode with a very high platinum or platinum alloy loading, 0.4-0.1 mgPt/cm2.26 There is approximately 10 times more platinum in the cathode to accelerate the sluggish oxygen reduction reaction that is five- orders-of-magnitude slower than anode's hydrogen oxidation reaction.4 Platinum is considered the best fuel cell catalyst in low temperature fuel cells because it has the 24 lowest overpotential and the highest selectivity for dioxygen reduction to form water.2

Unfortunately, even on pure platinum electrocatalsyts ~0.3 V is lost from the thermodynamic potential of 1.2 V vs. NHE for oxygen reduction due to the water activation and sluggish kinetics of the oxygen reduction reaction.40 If platinum specific power density is 0.2 gPt/kW, U.S. Department of Energy target for 2015, at operating voltages greater of 0.65 V then the high fuel cell energy conversion efficiencies of greater than 55% can be maintained.2 If the operating potentials are much less than 0.65 V than the energy conversion will be closer to that of a combustion engine effectively eliminating fuel cell‟s primary advantage. The development of ORR catalysts to either mitigate the overpotential or reduce the noble metal loading is pinnacle for PEM fuel cell research.

3.4.3 Direct Methanol Fuel Cells

The current state of direct methanol fuel cells technology uses a thicker sulfonated tetrafluoroethylene proton conducting membrane to mitigate methanol crossover, ≥ 2

Mils, with an anode with low precious metal catalyst loading, ~0.05 mgPt/cm, and a cathode with a very high platinum, 0.4-0.1 mgPt/cm2, and ruthenium alloy , 0.2-0.05 mgRu/cm2 loadings. The anode technology of the direct methanol fuel cell is nearly the same as the anode for a hydrogen fuel cell. The cathode in DMFC favors a ruthenium- platinum alloy over a pure platinum catalyst. Although, ruthenium is less active for oxygen reduction than platinum, ruthenium is also much more tolerant to methanol

25 oxidation crossover effects than platinum.41 Due to the strongly absorbed oxygen species on the surface of Ru-based catalysts methanol cannot readily absorb or react.40 A thicker sulfonated tetrafluoroethylene proton conducting membrane is used in direct methanol fuel cells to reduce the amount of methanol that crosses over to the cathode from the anode due to a difference in concentration. When methanol crosses over to the cathode and the cathode catalyst is active for both oxygen reduction and methanol oxidation, like platinum, both the methanol oxidation and oxygen reduction reactions occur causing a mixed current is created due to ORR consuming electrons whereas the methanol oxidation reaction produces electrons greatly decreasing the performance of a direct methanol fuel cell. Other classes of non-platinum catalysts such as Ru-Se clusters, Ru-N chelate compounds, nanostructured Ru, Co-C-N, Fe-C-N, TaOxNy, and MnOx are currently being explored as cathode oxygen reduction catalysts with high methanol oxidation tolerance.40 The most promising class of cathode alloys are palladium,

~$700/oz,42 and ruthenium, ~$180/oz,42 due to the lower price and decreased scarcity as compared to platinum,~$1,500/oz.43

3.5 Precious Meal Oxygen Reduction Catalysts

Both PEM and DMFCs have a similar construction and that could potentially benefit from the development of non-precious metal ORR catalysts. The cathode catalyst for both PEM and DMFCs is a platinum or platinum alloy supported on a porous carbon support with fairly high platinum loadings in the range of 0.4-0.1mgPt/cm2 with 26 negligible performance loss.26 To put things in perspective, the United States Department of Energy PEM fuel cell catalyst target for 2015 is 0.2 gPt/kW. If PEM fuel cells were to be used to strictly replace the current automotive fleet a production of 100 million cars per year would be needed to replace the 73 million cars produced in 2007.44 If each vehicle were rated for 50 kW approximately 1000 tons of Pt per year would be need.44

Contrarily, the total platinum production in 2007 was 204 tons.44 Most of this platinum was used in production of catalytic converter monoliths to clean the exhaust of combustion engines and other mature catalysis markets. In addition to the scarcity of platinum the current cost is $1,500/oz45, which would translate to a raw Pt material cost of $530 per vehicle if the 2015 United State PEM fuel cell target is satisfied. It is unlikely that platinum prices would remain static if a large portion was consumed by a formerly nonexistent business sector. For instance, the cost of platinum ballooned from

$600/oz (2003) to $2100/oz (2008) when demand increased with stagnant production.

The scarcity and high cost of platinum used to catalyze the oxygen reduction reaction in

PEM and DMFCs necessitate the development of alternative non-noble metal catalysts to facilitate oxygen reduction.

3.6 Non-precious Meal Oxygen Reduction Catalysts

There are several types of non-precious metal oxygen reduction catalysts but as a class of materials they are defined as oxygen reduction catalysts that do not contain; platinum ($1500/oz), palladium ($740/oz), rhodium ($2300/oz), iridium ($1050/oz), and 27 to a lesser extent ruthenium ($180/oz).42 In addition of being much less expensive than precious metals these catalysts are composed of abundant materials and as a consequence have different U.S. Department of Energy activity target of 8 A/cm3 at 0.8 V (impedance resistance corrected) in 2005. In 2005 platinum-based catalysts the U.S. Department of

Energy target was 0.55 A/cm3 at 0.9 V (impedance resistance corrected). These targets have different current density goals a different potentials because non-precious metal catalysts are currently less active for oxygen reduction than noble metal based catalysts, but due to the drastic differences in cost non-precious metal catalysts are being developed to meet the U.S. Department of Energy price point of 5 $/kW with 5,000 hrs of durability for 2015.

3.6.1 Macrocycle Non-precious Metal Oxygen Reduction Catalysts

The oxygen reduction reaction not only occurs in fuel cells but is commonly found in life on Earth. Nature has devised cytochrome enzymes, where an Fe-iron is coordinated to four surrounding nitrogen atoms to form a Fe-N4 complex, to catalyze the reduction of oxygen at low temperatures.46 Non-precious metal oxygen reduction research was inspired from these naturally occurring organic macrocycles like hemoglobin found in blood where oxygen adsorption readily occurs. The heme site in hemoglobin readily adsorbs oxygen on an iron center that is coordinated

28 by four surrounding nitrogen functional groups at low temperatures, Figure 6.47 Heme-

Figure 6. The structure Heme B found in hemoglobin type atom serve many simultaneous biological functions including; adsorbing oxygen, acting as catalysts for oxygen reduction, and electron transfer. The aromatic structure of heme molecules allows them to readily distribute electronic charge allowing them to act as an electron donor or participate in redox chemistry. In 1964 Jansinski was the first to discover that cobalt phthalocyanine had considerable oxygen electroreduction activity in alkaline solution by impregnating this compound on porous nickel disks.48 Cobalt phthalocyanine cathode electrode porous nickel supports were replaced with carbon black due chemical instability of nickel with solvents that dissolve and disperse pthalocyanines.48 The first non-precious metal cathode catalyst fuel cell was a hydrogen oxidation / oxygen reduction fuel cell at 80 C where electrolyte was 35 % KOH in an

29 asbestos matrix with a anode of 9 mgPt/cm2 and the cathode electrode was composed of cobalt phthalocyanine, acetylene carbon black, and Polytetrafluoroethylene as a binder.49

To prevent carbonate formation from the oxidation of carbonaceous fuels, such as methanol, and carbon dioxide in air an acidic electrolyte was used.50 It was later discovered that heme-like macrocycle molecules, such as iron pthalocyanines (shown in

Figure 7) and porphyrin, were active for ORR under hydrogen fuel

Figure 7. Cobalt prophine (right), Iron Phthalocyanine (left) Provided by Paul H. Matter cell cathode but quickly deactivated.50-53 The activity loss of these materials was attributed to the destruction of the Fe-Nx active sites in the strongly acidic cathode environment.54 It was found that the coordinated metal species was readily dissolved in the acidic and oxidizing conditions found in the hydrogen fuel cell cathode. Comparative studies using chelates other than N4- with other types of coordination ( N2O2-, O4-, N2S2, and S4-) were found to catalyze the reduction of oxygen although they were found to be

30 much less active nitrogen ligands.50 Within a given coordination-chelate the type and nature of the bonded metal atom strongly influenced the catalytic activity more so than the type of chelate used in the macrocycle.50 Experimental results during this period indicated that coordinated metal atom was the site of oxygen reduction.

3.6.2 Heat Treated Macrocycle Oxygen Reduction Catalysts

Due to the inherent instability of macrocycles in acid media, in 1976 Jahnke et al. published the first reports of improved activity and stability through pyrolysis, a heat treatment in an inert atmosphere. The lifetime of these macrocycle materials was improved with a pyrolysis temperatures above 600 C that were used to stabilize macrocycle structure on a carbon support.51-53,55-64 The activity of these heat-treated macrocycles was largely dependent on the certain metal centers used, in order of oxygen reduction activity Fe > Co > Ni  Cu.50 Ruthenium and iridium macrocycles were also found to have comparably high activity but studies were limited due to the cost of the coordinated metal.60,64 It was even shown that carbonized hemoglobin could function as a oxygen reduction electrocatalyst.65 For each macrocycle studied there was an optimum temperature treatment where the catalyst oxygen reduction activity would be the highest between 400 C and 950 C.50,63 At temperatures higher than the optimum the oxygen reduction activity would quickly become much worse. The optimal metal loading of pyrolyzed macrocycle materials was typically between 2 wt% and 10 wt%.59,66,67 The

31 selectivity of these catalysts for the complete reduction of dioxygen to only produce water could be achieved for certain materials.68,69 The lifetimes of these heat-treated macrocycle catalysts were limited due to direct chemical attack on the central coordinated metal atom, which dissolved into the solution as ions. Lifetime was dependant on the oxidation state, the type of metal used, the temperature of treatment, and chemical precursors used. A comparison of the effluent during pyrolysis for different macrocycle materials showed that the pyrolized metal containing macrocycle species with more nitrogen species had higher oxygen reduction activity.63 To produce the most active pyrolyzed macrocycle catalysts it was found that a coordinated metal, organic nitrogen- containing macrocycle, and carbon support must be present during the pyrolytic heat treatment.59,70,71 Interestingly, it was shown that the metal had to be present during the pyrolysis, but did not necessarily be coordinated with the ligands in the macrocycle.71

The pyrolysis was later found to completely destroy the coordinated-metal-type active site in the precursor yet the resulting catalyst materials were found to be more ORR active and stable.72 This increase in activity and stability was thought to be due to the formation of a new unknown active site.

3.6.3 Non-precious Metal Oxygen Reduction Catalysts Derived From Elemental Precusors

Based on the finding that metal had to be present during pyrolysis and not coordinated with the macrocycle to,71 it was also discovered that it was unnecessary to use organic macrocycle precursors create highly active and stable oxygen reduction

32 catalyst.73 In 19 86 Johansson and Larsson were the first report the thermal activation of an non-macrocycle, non-precious metal oxygen reduction catalyst, which was synthesized from a polymer complex doped with cobalt.74 This discovery was remarkable because this was the first instance where the precursors before pyrolysis did not have any inherent oxygen reduction activity, yet after a heat treatment above 600 ºC the material became active for ORR. There was a great deal research preformed following this original finding based on nitrogen-containing, electrically conductive polymers, such as polyacrylonitrile and polypyrrole, that could be pyrolyized with metal salts and high surface area carbons to form highly active non-precious metal oxygen reduction catalysts.75-80 More recently, it was demonstrated that a highly porous nitrogen containing polymer aerogel in the presence of a transition metal could be carbonized to form a highly porous graphite non-precious metal oxygen reduction catalyst without a carbon support.81 These works inspired the large body of work by the Dodelet group starting in

1997 that demonstrated that non-precious metal oxygen reduction catalysts could be formed during pyrolysis with simple precursors, such as carbon graphites, ammonia or acetonitrile, and iron or cobalt salts.1,44,46,59,82-123 Non-precious metal catalysts formed from simple precursors were shown to have equivalent oxygen reduction activity and selectivity as pyrolized macrocycle catalysts.92,93,95,104 These studies suggested that all that was necessary to create an active non-precious metal oxygen reduction catalyst was a metal ion, nitrogen and carbon precursors, and a heat treatment in inert atmosphere between 400 ºC and 1000 ºC.

33

3.6.3.a Synthesis Techniques Used to Create Pyrolized C-N-Metal Catalysts

The synthesis of non-precious metal oxygen reduction catalysts varies greatly within the body of literature since the only requirement to create material an active catalyst is that a metal ion, nitrogen source, and carbon source be contact during pyrolysis. The steady development of increasing more active oxygen reduction materials shaped the direction of this research field. Major publications often explored a series of materials then came to a hypothesis consistent with established literature theories to elucidate the apparent differences in the materials studied forming a new basis of knowledge and a new research directions. Several representative synthesis techniques will be discussed to illustrate the types and breadth of treatments used. Many non- precious metal ORR catalyst reviews have been performed and can provide a more exhaustive list of synthesis techniques that are commonly used.3,124-128

Due to the native oxygen reduction activity of metal-macrocycles, they were historically used as the simultaneous metal, nitrogen and carbon sources during pyrolysis.

1976 Jahnke et al. published the first reports of improved oxygen reduction activity and stability through the pyrolysis of series of Fe, Co, Ni, Cu and Mn chelates.50 In the same publication a metal chelates are deposited electrically conductive carbon-supports then pyrolyzed to disperse the active material chelates. This innovation achieved higher current densities while utilizing less chelate reactant material. Although, several types of chelates were studied the nitrogen-coordinated chelates were found to be the most active 34 for oxygen reduction so further research efforts were directed on toward nitrogen containing macrocycles. To produce pyrolyzed macrocycle catalysts with the highest activity with it was found that a coordinated metal, organic nitrogen-containing macrocycle, and carbon support must be present during the pyrolytic heat treatment.59,70,71 Many researchers have continued in this vein to create non-precious metal catalysts by pyrolyzing Metal-N4 macrocycles on carbon supports. Koslowski et al. created an ORR active material by pyrolizing 310 mg of 5,10,15,20-tetrakis(4- methoxyphenyl)-21H,23H-porphine iron(III)chloride, an iron porphyrin, on Ketjen carbon black at 800 ºC under N2 flow for 30 min followed by an etching treatment in 1 M

HCl with a post-rinsing step.129 The etching step was used to remove amorphous carbon and increase the surface area of the material.130 Some researchers resorted to continuously post-treating their material to enhance the catalytic activity of the catalyst material. For instance, Kosloski et al. synthesized an ORR catalyst by pyrolyzing a precursor containing 1 gram of 5,10,15,20-tetrakis(4-methoxyphenyl)-21H,23H-porphine iron(III)chloride or 5,10,15,20-tetrakis(4-methoxyphenyl)-21H,23H-porphine cobalt(III)chloride, 5 grams of iron oxalate dihydrate and 0.3 grams of sulfur in N2 at 450

ºC for 15 min then 800 ºC for 45 min followed by an etching treatment in 1 M HCl for 12 hrs. The resulting material was then subjected to a secondary heat treatment at 800 C in

10% H2/N2 for 30 min and etched again in 1 M HCl. After rinsing and drying the resulting powder was subjected to a tertiary heat treatment at 800 ºC in CO2 for 30 min to synthesize the finished catalyst.129 Multiple washing and pyrolysis steps were used to etch away amorphous carbon to increase the surface area and functionalize the carbon surface, respectively. To study the biological heme, the inspiration for non-precious metal

35 carbon-based catalysts, Maruyama et al. created a oxygen reduction active material from pyrolized, dried and ground hemoglobin from bovine blood in argon below 600 ºC for 10 hours. The resulting material was finely ground again and pyrolyzed again in 10 %

CO2/Ar at 900 ºC for 2 hours. The sample was then ground for a third time then acid- washed in boiling H2SO4 for 1 hour and rinsed. This study displayed that macrocycles derived from biological systems could be modified to form ORR active electrocatalysts.

Although the heme site was not isolated and supported on a carbon support the other blood constituents during pyrolysis will carbonize to effectively create a carbonaceous support during pyrolysis.

Some non-precious metal catalyst synthesis techniques used a metal macrocycle that was templated on a non-carbon support. Recently, Olson et al. synthesized an ORR active catalyst by pyrolyzing the precipitant of 0.5 grams cobalt tetramethoxyphenylporphyrin dissolved in 100 mL of tetrahydrofuran with 0.5 g of amorphous fumed silica at 700 ºC in N2 for 4 hours. The silica was then dissolved with

KOH, rinsed and dried.131 Silica was used as a high surface area support that acted as a template for the formation of a high surface area non-precious metal catalyst. Since the silica is electrically insolating the pyrolyzed material was treated in KOH to dissolve the silica.

Kaisheva et al. was the first to show that the metal had to be present during heat treatment, but not necessarily coordinated with the ligands of the macrocycle was shown to form an oxygen reduction active material.71 This finding inspired new synthesis routes where simpler metal, carbon, and nitrogen precursors replaced the nitrogen coordinated

36 metal compounds. In 1997 Fournier et al. showed that a dispersion of Fe(OH)2,

FeSO4*7H2O, FeCl3 dried on a 6.0 grams graphite or Vulcan carbon black support to from a 7-10 wt% Fe loading on carbon could be treated in H2 at 600 ºC then in acetonitrile saturated argon at 1000 ºC for hours to form non-precious metal oxygen reduction catalyst.82 From this study it was now apparent that a large variety of precursors could be used to for create non-precious metal catalysts causing the research field to explode. The Ozkan group was one of the first to form a non-noble metal catalyst grown entirely from a carbon and nitrogen rich atmosphere.132 This was accomplished by a pyrolysis of 2 wt% Fe acetate, Co acetate , or Ni acetate impregnated aluminum oxide or Vulcan carbon support at 900 ºC in acetontrile saturated nitrogen for 2 hours.73,133 The resulting material was etched in 1 M HCl at 60 ºC for 1 hr, rinsed, and dried to form the active catalyst. This was one first studies show that highly porous non-carbon support oxide could be used to template a highly porous catalyst. In order to remove the non- electrically conducting oxide support a post-pyrolysis acid or base treatment could be used to dissolve the support and increase the mass activity of the resulting non-precious metal catalyst.

Another class of non-precious metal catalyst materials was synthesized by only pyrolyzing simple solid phase precursors in keeping with the methodology for pyrolyzing macrocycles. Olson et al. used a 1 wt% Fe and Cu ground glucose, adenine, FeII gluconate dihydrate, and CuII mixture with equal molar ratios of FeII to CuII and glucose to adenine as the growth media. The physical mixture was heated to 150 ºC in air for 24 hours to complete glucose dehydration. The resulting material was then re-ground and headed in argon at 1000 ºC for 2 hours then etched in boiling 0.5 M H2SO4 for 1 hour, 37 rinsed, and dried to form non-precious metal ORR catalyst.131 A bi-metallic mixture was used to express the formation of desired nanostructures in the non-precious catalyst formed.134 This method also used glucose dehydration to form a highly porous template for carbon formation and to disperse the Fe and Cu pyrolysis catalyst. Garsuch et al. placed a 0.1 M ironIII chloride solution on mesoporous silica template that was then dried at 80 ºC for 12 hours. One gram of impregnated silica was then loaded with 1 mL pyrrole and exposed to HCl vapor to polymerize the pyrrole. After polymerization was complete the sample was pyrolized in argon at 900 ºC for 2 hours. The resulting material was then treated in a solution of 5% HF for 1 hr to remove non-electrically conducting silica template and non-activated iron, rinsed, and dried to form non-precious metal ORR catalyst.79,135 In this synthesis technique the carbon, nitrogen, and metal were dispersed on a highly porous template support to form a highly porous material. In order to achieve a high nitrogen content in the resulting material pyrrole was polymerized to be in close proximity of the metal. When non-conductive templates are used they are often removed with a dissolving treatment to increase the mass current density of the resulting catalyst.

Another synthesis approach for non-precious metal ORR catalysts used a metal salt as the exclusive metal precursor and ammonia atmosphere during pyrolysis as the exclusive nitrogen precursor. The carbon source for this scenario comes for a carbon support impregnated with the metal precursor. Charreteur et al. prepared a pyrolysis media by impregnating 6.24 mg of iron acetate with an aqueous solution on 1 gram of pristine furnace carbon black to obtain a 0.2 wt% iron loading. 110,111 The solution was stirred for 2 hours, then dried at 80 ºC. The resulting media was then heat treated at 950

ºC under ammonia for 40 min after which time the media was cooled in argon to 38 synthesize iron-carbon-nitrogen catalyst. High temperature ammonia treatments simultaneously impart nitrogen functional groups and corrode carbon materials.5 By imparting nitrogen function groups separately from carbon source this allowed the researchers to optimize the nitrogen functional groups on the catalyst surface. Jaouen et al. synthesized non-precious metal catalyst by planetary ball milling 0.5 g of carbon black Black Pearls 2000, 0.5 g of perylene tetra-carboxylic dianhydride, and 6.24 mg iron

44 acetate in N2. The mixture was then pyrolyzed at 1050 ºC with NH3 for 5 min. Ball milling the precursor material was thought to thoroughly mix the material and allow perylene tetra-carboxylic dianhydride to fill the carbon pores to form highly disordered catalyst material after pyrolysis.

3.6.3.b Characteristics of C-N-Metal Catalysts

There are several characteristics that work in concert to effect the oxygen reduction activity of non-precious metal carbon-based catalysts. Carbon-graphite oxygen reduction catalyst materials are principally composed of carbon, nitrogen and transition metal. Although carbon constitutes most of the non-precious metal catalyst material. An intricate surface functionalized carbon network with almost limitless chemical combinations comprises most ORR catalysts which complicates the significance of attributes that express catalytic activity, so only reoccurring literature themes will be discussed.

39 1) Oxygen Functional Groups

Oxygen functional groups are readily imparted on carbon materials when exposed to air.136-139 The chemically active edge planes are where most oxygen functional groups reside. There are several types of oxygen functional groups that may be present on the carbon surface, Figure 8. Some oxygen functional groups such as

Figure 8. Oxygen Functional Groups Present of the Surface of Graphite Materials lactols, phenols, and carboxyls impart acidic character to the carbon material.140 The presence of acidic groups makes the carbon surface more hydrophobic,141 which will effect water transport in the fuel cell catalyst layers. Furthermore, the presence of other surface oxygen groups is thought to decrease the water contact angle.142 These oxygen functional groups are fairly stable on the carbon surface up to 400K at which point CO2 desorption starts to take place.143 The concentration of oxygen functionalities on carbons

40 is known to increase with strong aqueous acid treatments and oxidizing gas treatments at elevated temperatures.130,144-148

The effects of oxidation has been studied on several carbons types. They include carbon blacks,85,136,138,142,149-151 activated carbons,140,148,152-156 carbon fibers,115,139,150,157-163 and carbon nanofibers.130,141,145,150,163-166 Carbon type has been correlated to the type of oxygen functional groups species that can be imparted on the carbon surface.167 The pore size distribution, ratio of amorphous to graphite carbon, and surface area can affect the type and degree of oxygen functionalization on carbon surfaces with comparable treatments.

It has been established that increasing the amount oxygen functionality of carbon supports has aided in the nucleation of platinum on the surface of carbon to form anode and cathode electrodes in PEM fuel cells.152,168 Active metal phase dispersion is thought to increase with the amount of oxygen functional groups present on the carbon surface.141

The impact of oxygen functional groups in non-precious metal oxygen reduction catalysts has also been studied.85,94,149,169,170 These studies have shown that carbon materials containing more oxygen functional groups possess greater oxygen reduction activity than carbon materials with lesser amounts of oxygen functional groups.85,94,149,169,170 Oxygen functional groups are also thought to be an indication of carbon corrosion within the catalyst layer.17,30,171 Oxidation of carbon nanofibers is thought to increase the surface area and pore volume of the resulting carbon material.130 Many researchers believe that oxygen functionalized carbon itself does not have oxygen reduction activity but acts as a

41 chemical medium that allow for other active sites to readily form on the carbon surface.85,94,169,170

Several publications have demonstrated a correlation with oxygen reduction activity and concentration of oxygen functional groups. The Popov group investigated the oxygen reduction activity of pyrolized un-oxidized carbon blacks and oxidized carbon blacks with a metal-free chelate and observed a 100 mV greater take off potential for the oxidized carbon black.18,22,169,170 Stelko et al. showed that graphites with mixed oxygen and/or nitrogen incorporation into the graphite structure had high electron donation properties and subsequent oxygen reduction activity.172 Woods et al. found a correlation between oxygen reduction and oxidative dehydrogenation with the amount of oxygen functional groups present on the surface for a carbon-nitride graphite catalyst.173

Oxygen functional groups can be characterized using complementary methods, but no single method can be used to distinguish each oxygen group. Oxygen functional groups on carbons are routinely characterized with temperature programmed desorption, acid and base neutralizing titration methods, electrochemical, X-ray photoelectron spectroscopy, and infrared methods.

Temperature programmed desorption in an inert atmosphere can be used to characterize oxygen functional groups. The emission of gaseous species containing oxygen (water, carbon monoxide or carbon dioxide) at different temperatures and relative concentrations can indicate the distribution of oxygen functional groups present.153,163

For instance, carboxylic group desorbs CO2 from 373 - 673 K, phenol group desorbs CO from 873 - 973 K, Anydride group desorbs CO and CO2 from 623 - 900 K, quinone 42 groups desorbs CO from 973- 1173 K, lactone group desorbs CO2 from 463 - 923 K, carbonyl group desorbs CO from 973 to 1173 K, and ether group desorbs CO at

153 approximately 973 K. Deconvoluting overlapping CO and CO2 spectra over a range of temperatures during temperature programmed desorption spectra is challenging, but there have been several methods presented in literature on techniques to identify oxygen functional groups on carbons.143,153,163

Chemical titration experiments, particularly Boehm titrations,136,174 were used to detect acidic and basic oxygen functional groups on carbon surfaces.158 Boehm titrations separate the surface carbon species into two groups, acidic and basic, and use specific titrants that are able to react with each group. For example, acidic carbonyl groups are thought to susceptible to OEt- attack from sodium ethoxide in ethanol. Other acidic groups include; carboxyl, lactols, and other groups that have phenolic hydroxyl groups on the carbon surface. Basic groups are comprised of ethers and γ-pyrone structures. The relative strength of the acidic or basic functional groups and the concentration effect oxygen functional group titration method accuracy.148 If oxygen functional groups are located within micropores the aqueous titrant might not be able to interact with these functional group resulting in an underestimate of the oxygen surface species on highly porous carbons.148

The presence of quinone groups has been determined by quantifying the electrochemically active quinone/hydroquinone redox couple at 0.55 V vs. NHE during

43 cyclic voltammetry in an aqueous acid solution, Figure 9.22,142,175 An increase quinone

Figure 9. The Electrochemically Active Quinone (right) /Hydroquinone (left) Oxygen Species groups can be observed during potentiostatic17,22 and cyclic voltametric30 oxygen reduction conditions. Since this is a one electron reaction redox couple the changes in current can be used to quantify the amount of each species present on the electrochemically active surface.

Oxygen species have also been readily identified through X-ray photoelectron spectroscopy. Although, the deconvolution of the oxygen 1s spectra to determine the specific oxygen groups can be challenging, Figure 10. X-ray photoelectron spectroscopy binding functional groups include; physically adsorbing oxygen (530.1 eV),176 carbonyls and quinones (531.0 eV),115,153 ethers or lactones and C=O in anhydrides (532.3 eV),115,153 and oxygen in anhydrides of lactones (533.3 eV).115,153 Since many of the

44 oxygen functional groups have multiple oxygen atoms with different binding energies the

Figure 10. Oxygen Functional Groups Detected in X-ray Photoelectron Spectroscopy (shown in energies for oxygen 1s blue are the oxygen species that have a binding energy centered at that value)11

determination of the concentration of a specific oxygen group is difficult. This matter is further complicated since the resolution of X-ray photoelectron spectroscopy is on the order of 0.1 eV for oxygen 1s species so typically the envelope of the oxygen spectra is deconvoluted into four main groups. The oxygen functionalized carbon 1s spectra has also been used to identify the carbon-oxygen species which tends to be less difficult than identification of the oxygen species in the of XPS.159 Since it is difficult to determine the concentration of any one oxygen group typically only the surface oxygen concentration is reported.

For infrared characterization of carbon materials the material must be highly oxygen functionalized to overcome the high infrared absorption of the carbon material.137,153 Few studies have been done on carbon materials that are of practical importance for fuel cell electrodes because to achieve such high degrees of surface

45 oxygen the materialsn would need to be largely composed of amorphous carbon, but to be stable in a fuel cell environment graphitic carbon is desired.

2) Nitrogen Functional Groups

It is generally agreed that nitrogen incorporation into the graphite matrix is needed to form a highly active carbon-based non-precious metal oxygen reduction catalyst. Strelko et al. used computational calculations on graphitic carbons doped with nitrogen, phosophorus and nitrogen and determined that each dopant species imparted enhanced electron donation properties which was considered a rate limiting step for oxygen reduction.177 However, many researchers consider the active site for oxygen reduction active metal center that is stabilized by nitrogen ligands, a similar structure to macrocycles compounds, that is formed during pyrolysis.1,55,118,178-184 Another group of researchers believe that graphitic carbon nitrogen groups have inherent oxygen reduction activity without any metal present.64,133,185-187 Both hypothesized active site types would have enhanced ORR activity with the inclusion of nitrogen into the hexagonal network.188-190 Most integrated nitrogen graphite species are described as either, pyridinic-nitrogen, located as a substituted heteroatom on the graphitic edge plane or quarternary-nitrogen, a substituted nitrogen deeper in the graphitic plane bonded to at least three carbons.44,73,97,132,133,191-194 Other nitrogen groups, including pyrrolic-nitrogen, have also been identified in carbon graphites, but appear to be less abundant nitrogen doped graphites.44,97,191,194. The typical total nitrogen surface content for nitrogen doped 46 carbon graphite oxygen reduction catalysts is approximately from 1 to 9 atomic

%.44,73,97,132,133,191-194 Carbon-nitride materials with higher nitrogen concentrations undergo a distinct a phase change and have more semiconductor like properties.125

Many researchers have developed techniques to create nitrogen incorporated graphites.

Several different methods can be used to incorporate nitrogen in the graphite structure.

Nitrogen can be introduced during carbonization pyrolysis14,195-206 or in a post-treatment step with ammonia1,85,97,103,116,120,184,207,208. The Dodelet group also studied the effect of pretreatment nitrogen functionalization of carbon blacks on their Fe-N-C ORR catalysts.85 The oxidized carbons were heat treated in ammonia, then an iron salt was added to the carbon and heat treated. It was found that the presence of oxygen functional groups on the carbon black increased the nitrogen content of the support after ammonia treatment. Several carbon and nitrogen containing reactants have been used during pyrolysis to create nitrogen doped graphites including; acetonitrile,133,189,196,198-

200,203,204,209-211 melamine,14,195,202,212-214 polyacrylonitrile,76,77,215,216 polyaniline,80,186,217-220 and pyridine199,201,204-206,213,221-224. Only a few nitrogen-containing carbons synthesis methods are discussed in section Synthesis Techniques Used to Create pyrolized C-N-

Metal catalysts since a wide variety of treatment conditions can be used.

Surface nitrogen species on carbons are most often characterized by nitrogen 1s

X-ray photoelectron spectroscopy. When nitrogen species are incorporated into the carbon graphite some broadening of the graphitic carbon 1s peak has been reported.84,193

Since the nitrogen species in carbon graphites have similar X-ray photoelectron binding energies their spectra is deconvoluted to determine the relative amount of each surface

47 species. The nitrogen species of the N 1s X-ray photoelectron spectroscopy spectra are either pyridinic oxide (402-405 eV), quarternary (401.3 eV), pyridinium (401.2 eV), pyrrolic (400.5 eV), or pyridinc (398.6 eV) nitrogen, Figure 11.58,225-228 In highly graphitic materials quaternary and pyridinic nitrogens

Figure 11. An Illustration of the Possible Types of Nitrogen Present in Graphitic Carbon and their Corresponding N 1s Binding Energies nearly comprise the entirety of the nitrogen species on the surface making nitrogen 1s spectra deconvolution relatively simple.133 It should be pointed out that the resolution of the nitrogen 1s X-ray photoelectron spectra (~0.2 eV) is not high enough to distinguish the difference in quarternary and pyridinium nitrogen groups.

3) Carbon Nanostructure Morphology

48 Carbon nanostructure morphology can affect the mass activity of carbon-based non-precious metal oxygen reduction catalysts. Observing this phenomena is consistent with theories where the oxygen reduction active site is located on the carbon edge planes.

The metal species needed to form highly active non-precious metal ORR catalysts are also pyrolic carbon nanostructure growth catalysts although this not often discussed non- precious metal ORR literature.229-232 Some researchers hypothesize that carbon growth starts from supported metal particles that stay anchored to the support while carbon fibers grow away from the metal particle while a carbon source is constantly supplied from a high temperature atmosphere.134 Other researchers believe that the metal particle is transient and that forming the carbon nanostructure pushes the metal particle along to grow additional nanostructure.199 Most of the evidence for these theories comes from transmission electron spectroscopy of the formed materials well after carbon growth has occurred. In both instances the metal nanoparticles used to growth the carbon structurse

49 are encased in layers of graphitic carbon, Figure 12. Matter et al. was one of the first

Figure 12. Transmission Electron Microscopy Image of a CNx Encased Iron Nanoparticle to demonstrate that nanostructure of non-precious metal catalysts could influence ORR activity with the proportion of nanostructures present that have a large amount of exposed nitrogen functionalized edge planes.132 This was done by characterizing the distribution of the carbon nanostructures formed and correlating the oxygen reduction activity with the prevalence of high-edge plane nanostructures, stacked cups or bamboo carbon and herring bone structures . Several carbon morphologies can be formed during carbon

50 growth and deposition, Figure 13. Graphite

Figure 13. Carbon Nanostructure Morphologies.211 sheets are represented by solid shapes in Figure 13. Carbon nanostructures have high electrical conductivity within each carbon sheet, and poor electrical conductivity between sheets. Carbon nanostructures with a high aspect ratio are often referred to as carbon nanofibers. Multi-walled nanotubes have multiple cylindrical shaped cabon nanotubes placed inside one another. These structures have high electrical conductivity across the main longitudinal axis but have very little chemically active carbon edge plane exposure.

There are many synthesis routes to fabricate carbon nanofibers, most of which involve a high temperature treatment of gas phase hydrocarbon reactants and metal catalysts particles in an inert atmosphere.231,233-237 Herringbone nanofibers are comprised of cone- shaped graphitic sheets that are stacked on the longitudinal axis of the nanofiber. This orientation has high electrical conductivity through the main longitudinal axis while leaving a large amount of chemically active graphitic edge plane exposure. Yoon et al. showed that herringbone nanofibers could be synthesized by decomposing ethylene with

51 hydrogen over a bimetallic copper-nickel catalyst.238 Stacked platelet nanofibers are composed of sheets of carbon stacked on top of one another along the longitudinal axis.

They have poor conductivity across the main longitudinal axis and a high degree of edge plane exposure. Stacked platelets have been synthesized decomposing carbon monoxide in the presence of hydrogen gas streams over iron and iron-copper growth catalysts.238,239

Ribbon nanofibers are composed of sheets of carbon stacked on top of one another along the axial axis. They are high conductivity across the main longitudinal axis and a fairly high degree of edge plane exposure. Stacked cup or bamboo carbon nanostructures are composed of cup-shaped carbon sheets that are stacked on top of one another along the longitudinal axis of the nanofiber. They are fairly conductivity across the main longitudinal axis and have a high degree of edge plane exposure. When iron was used as the nanofiber growth catalyst and a carbon-nitrogen precursor was used for growth, stacked cup nanofibers were commonly observed.14,195,197-199,202 Nakajima et al. made stacked cup fibers with high edge plane exposure using a system of acetonitrile/nitrogen for carbon fiber growth over Ni catalysts.204 Carbon nano-onion structures are composed on concentric orb-like sheets of carbon. They are poorly conductivity across the main longitudinal axis and have very little edge plane exposure. Kvon et al. observed nanofibers with high degree of edge exposure while using a system of methane/hydrogen/pyridine for carbon deposition over Ni-Cu catalysts.205 Carbon nano- onions nanostructures are typically observed when a metal catalyst particle is not involved in carbon growth or reactants are supplied in excess.

52 3.6.4 Non-precious Metal Oxygen Reduction Catalyst Active Site Dispute

The breadth of precursors and treatments used create precious-metal free ORR catalysts combined with the limited characterization techniques available lead to many active site hypotheses. Without a definitive active site(s) the development of precious metal catalysts is hindered due to an inability to normalize data over the catalytic active area. In metal-based electrochemical research hydrogen absorption is used to determine the electroactive surface area and normalize catalyst materials accordingly,240 but electrochemical techniques used to detect the electroactive surface area for non-metal systems have been largely inconclusive. Efforts to normalize the non-noble metal catalysts oxygen reduction activity the search for a detectable active site have been ongoing in the literature.44 The determination of the site(s) active for oxygen reduction on non-noble metal catalysts would greatly advance the research progress of these materials.

It was thought that the active coordinated metal center (cobalt, iron, and ruthenium) in pyrolyzed macrocycles was retained after heat treatment. For these materials there exists an optimal pyrolysis temperature between 400 - 800 ºC where oxygen reduction activity is the highest. At higher temperatures activity quickly decreases which is attributed to the destruction of the coordinated metal center oxygen reduction active site.55,241 X-ray photoelectron spectroscopy demonstrated that the surface content of nitrogen and metal were the most prevalent for the most ORR active pyrolized macrocycle catalysts, suggesting that the more coordinated macrocycles were present in the more active catalysts.242,243 Mossbauer and X-ray absorption spectroscopy that

53 specifically target the metal atom have also shown that the coordination number of the metal does is not altered during heat treatment for the most active oxygen reduction catalysts until pyrolysis temperature in excess of 700 ºC.55,244,245 At higher temperatures the metal coordination number changes and losses in ORR activity are reported.

In opposition, there are many opposing views that suggest that the coordinated metal center belonging to the pyrolyzed macrocycle is completely destroyed during the pyrolysis. In one such study Gojkovic et al. pyrolized iron (III) tetramethoxyphenyl porphyrin on a high surface area carbon support and observed an increase in ORR activity/stability but an absence of Fe-N4 centers characterized by X-ray photoelectron spectroscopy.68,246 It should be noted that the lower limit using X-ray photoelectron spectroscopy for cobalt and iron is approximately 0.1 % surface concentration and it has been shown that concentrations as low as 0.02 wt% significantly increase the ORR activity.107 Since X-ray photoelectron spectroscopy is a surface sensitive technique and catalytic activity takes place on the surface this suggests that less than 0.1 % of the surface needs to have stabilized macrocycles to have sufficient oxygen reduction activity.

Conversely, dispersed nano-platinum needs to be than greater than 0.5 % surface platinum to have high oxygen reduction activity (0.4 mgPt/cm2, with 10 wt%Pt on

Vulcan carbon in cathode catalyst).1 This would suggest that stabilized macrocycles are more active than nano-platinum catalysts although it has been established that nano- platinum is the most active ORR catalyst.1,2,247 TEM imaging of the resulting pyrolized macrocycle catalysts were found to have agglomerated metal nanoparticles encased entirely within carbon sheets.68 Complimentary Fe57 Mössbauer and X-ray absorption spectroscopy have shown the destruction of metal-N4 center and the formation of metal 54 54,248 carbide. This suggests that some portion of the metal-N4 macrocycle was destroyed during pyrolysis to form these carbon encased metal particles, but inexplicably the resulting materials had greater activity and stability for the oxygen reduction reaction.

In summary, the literature belief was that non-precious metal oxygen reduction electrocatalyst derived from macrocycles could be obtained with a heat-treatment from

500 - 1000 ºC with elemental metal ions (Fe, Co, Cu, etc), a source of carbon (carbon support, macrocycle), and a source of nitrogen (metal-N4 macrocycle, H2-N4 macrocycle) that are all simultaneously present during pyrolysis. Currently, the development of macrocycle-based non-precious metal oxygen reduction electrocatalysts is still ongoing.

65,78,129,249

Some researchers purposed that a nitrogen-coordinated metal ORR active site(s) structurally different from a macrocycle is formed.86,103,105,248 Although all these research groups agree that metal is part of the active site, there is considerable disagreement in what the structure this site is, where it is located, and the reaction behavior of the purposed site. The Bron group declared that the active site was a metal phenylporphyrin chemically bonded to a graphite support macrocycle-like site.189 The Maruyama group supported this hypothesis by showing that the amount of non-noble metal precursor used effects the active site density in the final material.180 Initially, some researchers thought the macrocycle type clusters where shielded from the metal leaching acid electrolyte by layers of carbon that were not so thick that they would inhibit mass transfer to and from

76,250 the active site. This argument was supported by the absence of detectable Metal-N4 structures with surface-sensitive X-ray photoelectron spectroscopy and graphite

55 formation observed in transmission electron imaging.246,250 If this theory were plausible than increasing amounts of Fe-N4 precursor would create catalysts with ever greater active site density and increased oxygen reduction current density, which is not the case since the optimal iron loadings for optimal ORR activity can be as low as 0.2 wt%.92 The most acknowledged theory is that the new active site formed during pyrolysis is a metal-

N4 and/or metal-N2 centers which has the same bonding as metal chelates. The main evidence for the metal-Nx sites comes from the prevalence of Fe-N4/Co-N4 and Fe-

N2/Co-N2 ions of heat-treated materials correlate with ORR activity detected by time-of- flight secondary ion mass spectrometry.100 Additional time-or-flight studies showed at

92 correlation between Fe-N4C8 and Fe-N2C4 ions and oxygen reduction activity. It is important to note that time-of-flight secondary ion mass spectrometry uses a high energy pulsed ion beam to ionize species and only records stable ion fragments, such as metal-N4 and metal-N2, from the material‟s surface, but since highly active non-precious metal graphite materials are known to process high nitrogen content, and cobalt or iron it is not surprising that these ion fragments are detected.3,44,73 This suggests that the ions measured during time-of-flight secondary ion mass spectrometry may not be in the metal-

N4 and metal-N2 bonding configurations in the native highly active materials but are instead formed by the technique. It was first thought that three active sites were present in non-noble metal catalyst; an iron oxide site, an Fe-N4/C site, and an Fe-N2/C site that

56 was thought to be the most active, shown in Figure 15.94 In these types of

Figure 14. Iron-N2/C type oxygen reduction active site (courtesy of Paul H. Matter) structures it is believed that nitrogen atoms that are incorporated into the graphitic carbon matrix stabilize metals on the carbon edge planes. Additional evidence for metal-nitrogen moieties in active non-precious metal ORR catalysts comes from the presence of metal- nitrogen bonds that was determined by investigating the local iron environment with extended X-ray absorption Fourier–transform analysis and Mössbauer Fe57 spectroscopy.56 These analyses that specifically target iron atoms demonstrated that the majority of the iron in non-precious metal catalysts is located with a lighter weight scatter, which could be nitrogen, but standards were not used to eliminate carbon or oxygen as the lightweight scatters observed. Nitrogen 1s X-ray photoelectron spectroscopy performed by Faubert et al. detected pyridinic and pyrrolic nitrogen species

57 that reside on the graphite edge planes and could therefore coordinate transition metal ions.83 Oxygen reduction nitrogen-carbon graphites pyrolyzed in the presence of iron formed more pyridinic nitrogen groups as compared to the other nitrogen species present on the surface detected by nitrogen 1s XPS.83 In a later publication, time-of-flight secondary ion spectroscopy was used to investigate the surface species present of non- precious metal pyrolyzed Co-N/C ORR catalysts, but surprisingly the materials were found to be active for ORR yet Co-N2 and Co-N4 type ions were not detected on the surface.100 These ORR active materials were synthesized by impregnating cobalt (II) acetate and cobalt porphyrins on high surface area carbon supports that were then pyrolyzed in an ammonia atmosphere at temperatures greater than 500 C. Although these Co-N2 and Co-N4 sites were not detected on fresh catalyst materials, one theory is that they could be generated in the acidic solution during oxygen reduction testing where the metal is soluble, and can relocate to coordinating nitrogens.105,251,252 This theory is supported by a recent publication that investigated the N-ls spectra with X-ray photoelectron spectroscopy on fresh catalyst material and after electrochemical oxygen reduction.253 It was found that pyridinic nitrogen groups became quaternary nitrogen groups after fuel cell testing, which could suggest that the pyridinic nitrogen groups are unstable in acidic cathode, that pyridinic nitrogen groups are additional bonded by metals atoms leached into the solution or protonated by the acid solution to form pyridinium.

Since the X-ray photoelectron spectroscopy N-ls quaternary and pyridinium species have nearly same binding energy and standardized materials for high-resolution de- convolution do not currently exist each possible explanation is equally plausible.

58 Many researchers consider the new active site formed during pyrolysis to have a similar structure to macrocycles compounds, with a metal coordinated to nitrogen ligands.1,55,118,178-184 In an effort to improve the active site density of nitrogen-coordinated metal ORR catalyst highly microporous carbon supports were pyrolyzed with 0.2 wt% Fe in a pure ammonia atmosphere at 950 ºC.112 To form a material that was active for ORR nitrogen and iron needed to be present during the pyrolysis of these microporous carbons.

Moreover, a strong correlation between the volume of micropores and oxygen reduction activity was found provided there was greater than 1 at% surface nitrogen.112 Microporus carbon is defined as having pore size distribution with diameters less that 2 nm determined by a specialized non-local density fuctional BET theory assuming a slit pore geometry.112 Conventional Brunauer, Emmett, and Teller model isotherm analysis accounts for molecularly adsorbed multi-layers that can only measure pores with diameters greater than 2 nm due to different the heats of adsorption in macro- and micro- pores. Due to the correlation between ORR activity and microporus area, in 2008 the

Dodelet group hypothesized that the active site formed during pyrolysis was a metal coordinated with pyridinic nitrogen groups housed in the mircropores of a graphitic

59 catalyst material, Figure 15.110,112 The

Figure 15. Iron-coordinated oxygen reduction active site thought to reside in graphite micropore was thought to stabilize the coordinated metal ion so that it could not easily dissolve into the bulk solution. It was also thought that there was another Fe-N4 type active site on the surface of the Fe-N-C non-precious metal catalysts which was evident by ORR activity that correlated with Fe-N4/Co-N4 and Fe-N2/Co-N2 ions detected by time-of-flight secondary ion mass specronomy.100 It was shown by Herranz et al. that material with a higher concentration of the Fe-N4-C active site performed well in basic electrolyte and poor in acidic electrolyte whereas materials with a higher concentration of the Fe-N2+2-C active site located in the micropore was ORR active in both acid and

118 base. This suggested the bridged Fe-N2+2-C active site that can only be located in the micropore was less prone dissolution in an acid solution than the Fe-N4-C active site that would be present on the surface. It was thought that either N-C or iron ions were limiting

60 formation of materials with higher active site density in F-Nx-C materials. Non-precious metal catalysts were pyrolyzed in ammonia on different carbon supports while keeping a nominal 0.2 wt% Fe loading constant resulting in materials with ORR activities ranging in several orders of magnitude.94 This suggested that main limitation in the formation of

Fe-Nx-C type active sites was the availability of surface nitrogen content. Although, it was unknown why some carbon supports lead to higher nitrogen contents than others. As a partial explanation, it was later shown that the oxygen functional groups present on the surface of carbon support facilitates the incorporation of nitrogen during ammonia pyrolysis.5 For non-noble metal catalysts with extremely low transition metal concentrations if the metal was sole contributor to ORR the mass activity would be similar to that of the best performing platinum metal catalyst1, which would suggest that another non-metal active site is present.

Other researchers maintain that there is an ORR active site does not necessary contain a metal. Based on studies of heat-treated carbon-supported macrocycles, Weisner and co-workers concluded that the metal in the organometallic compounds acts as a catalyst for the formation of a non-metal containing electrocatalytic oxygen reduction active site during pyrolysis.70,254 Goureck et al. presented work on heat-treated charcoal supported Co-tetraazaannulene and concluded that the nitrogen species were the source of oxygen reduction through protonation and subsequent oxidation of the nitrogen ions.61

Gojkovic et al. went on to propose that an N-O species were responsible for the ORR activity present in heat-treated Fe porphyrin-type catalysts and that metal, metal oxide, or metal carbide particles could act as the catalyst for the formation of these active sites.60,68

This work was the first to suggest that the metal was involved in the formation of oxygen 61 reduction active sites, but was not a part of the active sites formed. The metal type present during pyrolysis was known to effect the ORR activity of the resulting material, but it was also found to effect the morphology of the carbon nanostructures formed.193

Metals that formed nanostructures that had higher amounts of edge plane exposure were found to be more catalytically active. This suggested that the growth of the nitrogen- containing carbon nanostructures during pyrolysis were responsible for the increase in activity observed. In 2009, Subramanian et al. hypothesized that quaternary and pyridinic nitrogen may act as active sites in graphitic carbons whereby oxygen reduction is facilitated by electron donation from the conjugated nitrogen π bond, Figure 16.188 The pyridinic nitrogen group was thought be more

Figure 16. Quaternary and pyridinic nitrogen ORR active sites proposed by Subramanian et al. catalytically active than the quaternary nitrogen site due to its facile donation of

133,255 electrons. It should be noted that this observation supports the competing metal-N4 62 and metal-N2 type active site theory where pyridinic nitrogen sites within close proximity are also thought to stabilize metal ions. Active materials formed by pyrolyzing either metallic or non-metallic macrocycles clearly demonstrated that the same oxygen reduction catalytic properties could be obtained except that a higher pyrolysis temperature were required.51,254 This appears to refute the hypotheses where the active site is a coordinated metal ion coordinated by pyridinic nitrogen groups, yet surface nitrogen groups on these nitrogen-doped carbons could explain the differences in activity.254 The ORR activity of untreated carbon supports was improved with the addition of nitrogen-containing H2-macrocycles during pyrolysis demonstrating that metal was not necessarily participating in the reduction of oxygen. Detractors to this study suggest that the improvement in activity was caused by metal contamination found in the carbon support. In response to this protest, another metal-free carbon support synthesized by polyvinylidene chloride decomposition displayed a comparable increase in ORR activity when pyrolized with metal-free macrocycles.64 Matter et al. also prepared metal-free carbon-nitrogen graphite ORR catalyst with considerable activity by pyrolyzing acetonitrile over sol-gel alumina support.133

Although, these metal-free materials had improved oxygen reduction activity compared to untreated carbons, the power density of these materials was considered too low for practical high power density applications.133,188 To obtain non-precious metal materials with high oxygen reduction activity and current density a metal species must be present during pyrolysis. Nitrogen functionalized heat-treated carbons have been studied extensively by several researchers201,256-261, but no consensus was reached on the structure responsible for oxygen reduction. 63 The metal atom may indeed catalyze the formation of active sites on carbon supports for non-noble metal ORR catalysts. Both hypothesized active site types are enhanced by the inclusion of nitrogen into the hexagonal network.188-190 However, the structure of these active sites and how the metal enhances the formation of these sites during pyrolysis remains unknown and warrants further investigation for catalyst development.

3.6.5 Computational Chemistry

During the past decade, significant computational efforts were devoted to study the oxygen reduction reaction mechanism on various catalyst surfaces. These efforts were focused on improving the performance of existing catalysts and to aid in the rational design of novel catalysts.

Anderson et al.1 first applied ab initio method to investigate the potential- dependent activation energy of elementary reaction steps for ORR processes.

Subsequently, a single platinum atom and a platinum dimer model was used for coordination of O2, HOO, H2O2, and HO intermediates in order to consider the effect of platinum for the conversion of these reactive oxygen species.2-3 These studies concluded that the first electron transfer to the adsorbed molecular oxygen was the rate- limiting step in the ORR mechanism. Balbuena and co-workers4 used larger realistic nanoparticle platinum cluster models to analyze the conversion from O2 to HOO with

64 consideration of an applied electric field like what is found in the oxygen reduction cathode and came to a similar conclusion. In addition, periodic boundary density functional theory (DFT) studies, using a slab model, were also used to study the ORR mechanism on Pt.5-11 Norskov et al.8 employed periodic DFT methods to investigate the binding energy of the reactants and intermediates involving in the ORR process. Trends in oxygen reduction activity as a function of the oxygen binding energy were demonstrated with a volcano plot, binding energies and temperature space houses chemical species points. Neurock and co-workers developed the double-reference method12,13 and then applied this method to determine the energetic dependence of the initial electroreduction steps of oxygen over a fully hydrated Pt(111) surface for various electrode potentials.14 Recently, computational studies were also performed to rationalize and to design Pt-based alloy electrocatalysts15-17 – these catalysts have demonstrated higher ORR activity than pure Pt surfaces.

Computational studies have also been applied to the non-metallic graphene catalysts. For instance, understanding the detailed mechanism of ORR in CNx materials has attracted increasing interest from a computational point of view. Sidik et al. reported that carbon radical sites formed adjacent to quaternary graphitic nitrogen are active for O2

18 electroreduction to H2O2 via an adsorbed HOO intermediate. Ikeda et al. used a Car-

Parinello molecular dynamics approach to study the possible O2 adsorption sites for eight

19 different model structures of appropriate CNx catalysts. The simulation results showed that O2 preferentially adsorbed in an associative manner to the graphene-like C sites, zigzag edges and more so if a graphite-like N was located nearby. Then two water molecules could be generated through the subsequent O–O bond cleavage.19 However, 65 these simulations did not support the ORR catalytic activities on C atoms adjacent to N atom in the basal plane.19 However, the conclusions drawn from the CPMD simulation were not consistent with extant quantum calculations with cluster models.18 Additional theoretical investigations with a cluster model of nitrided graphite edges by Vayner et al. did not support the activity of pyridine-like N toward the two- and four-electron reduction of oxygen.20 Although it has been suggested that a trace amount of transition metal (such as Fe, Co) might play an important role in the ORR process,20-23 elucidating how the CNx material have intrinsic catalytic activity on their own is still unclear.

Providing an accurate structure for the theoretical study is critical for computational theories to be valid. Thus, on the basis of these seminal contributions, computational approaches should be valuable in aiding experimental studies for obtaining a molecular-level insight into an effective understanding of the ORR mechanism by the catalyst‟s structure, but one does need to have good structural information about the catalyst‟s structure and three-dimensional morphology.

The high-temperature treatment necessary to achieve stable carbon-nitride non- precious metal materials facilitates the formation of a multitude bonding configurations, thereby obscuring the identification of an active site which further constrains the optimization these materials. Several correlations with increased oxygen reduction activity for non-precious metal catalyst were made despite little knowledge of how they operated. For instance, it was established that incorporated nitrogen is essential to achieve high oxygen electroreduction activity in acidic media.3,87,262 Furthermore, it was shown that significant ORR activity can be achieved with a metal-free (< 1 ppm) CNx

66 catalyst, suggesting that there are either multiple active sites, one of which does not have a coordinated metal.133 Oxygen reduction activity was also shown to increase with the graphitic character of CN catalyst.193 These observations led to the hypothesis of a graphitic carbon phase with nitrogen atoms replacing carbons in the graphite matrix being the active phase for oxygen electroreduction.

Less computationally expensive poly-aromatic hydrocarbons were used as probe molecules to investigate the active site hypotheses in which electron donation is rate

Figure 17. Two step electron donation type oxygen reduction mechanism

67 determining, such as described by Strelko et al. shown in Figure 17.172 The Strelko et al. mechanism through a graphite material with free electrons donating an electron to dioxygen to form a superoxide. This super oxide is then protonated by available water if in a basic solution or free protons in an acid solution to form a hydroperoxyl. This hydroperoxyl would then get another donated electron from the graphite material to

- - become H2O . H2O would then get protonated from the acidic and/or aqueous solution.

Another electron donation from graphite or decomposition could dissociate the hydrogen peroxide into water and dioxygen. It should be noted that the active site hypothesized by

Strelko et al. has electron donation for graphites doped with oxygen instead of nitrogen atoms. The mechanism in essence would be the same for nitrogen-doped graphites except that the positively charged graphite species would be more stable due to the additional electronegativity found nitrogen as opposed to oxygen. If the graphite materials were operating in fuel cell environment the positive change would be neutralized by electrons provide by the oxidation reaction taking place in the anode.

To determine the probability of electron donation for model polycyclic aromatic hydrocarbons (PAHs) the adiabatic and vertical ionization potentials (Ip) were calculated.

Vertical ionization potentials are calculated by taking the difference in total energy of the optimized the geometry of molecule in a neutral, ground electronic state and the same optimized geometry in the cationic state. Since only on optimized geometry is used to calculate the vertical ionization potential this technique is relatively computationally inexpensive. Adiabatic ionization potentials are calculated by taking the difference in total energy of the optimized the geometry of molecule in a neutral, ground electronic state and the optimized geometry of the molecule in the cationic, ground state. Adiabatic 68 ionization potentials are much more computationally expensive than vertical ionization potentials due the additional optimized geometry calculation of the cationic state of the molecule. The difference in vertical and adiabatic ionization potentials are shown in

Figure 18. The cationic state for the PAHs is typically at a higher energy than the

Figure 18. Graphical Representation of Vertical and Adiabatic Ionization Potentials neutral state. There are several vibrational states possible for each molecular geometry and state with the lower energy vibrational states being the most statistically probable at low temperatures and therefore the most chemically relevant to physical phenomena.

Since the vertical ionization potential does not take into account the relation of the molecule after electron donation the vertical ionization potential can be larger than the adiabatic ionization potential as shown in Figure 18. Through the calibration of known structures with experimental and computational IP values, one can then extend this computational approach to incorporate other CNx catalysts in which various forms of N

69 are considered for the PAH, including pyridinic, pyrrolic, N-oxide, and quaternary N centers.

With an understanding of the local nitrogen bonding environment and the role of doped nitrogen atoms on the ionization potential for eventual O2 reduction to superoxide

– radical anion (O2 ), the active site of ORR can be explored using computational and experimental methods in tandem.

70

CHAPTER 4. Experimental Methods

4.2 Catalyst Synthesis

4.2.1 CNx Catalyst Synthesis

The fabrication of CNx oxygen reduction catalyst can be separated into two parts; preparation of the growth media and pyrolytic growth of nitrogen-carbon nanostructures with after treatment. CNx growth media was prepared by incipient wetness impregnation of 2 wt % Fe onto a nanopowder magnesia support (Sigma-Aldrich) using an aqueous solution of iron (II) acetate (Sigma-Aldrich). The resulting growth media was dried overnight at 110 °C in air. The purpose of incipient wetness impregnation was to deposit the iron (II) acetate solution into the pore volume, determined by the Barrett, Joyner and

Halenda technique, of the high surface area nanopowder magnesia support. Two grams of dried growth media was deposited into a quartz calcination boat that was then placed inside of a quartz calcination tube in a high temperature furnace, shown in Figure 19. A

71 quartz calcination tube (2 inch outer diameter) was purged with

Figure 19. Pyrolysis CNx Growth Experiment Setup high purity nitrogen at 150 mL/min for 30 min and then heated at 10 °C /min until 900 °C was reached. Upon reaching 900 °C, acetonitrile (CH3CN, Fisher, Optima grade) saturated nitrogen gas at 150 mL/min was streamed over the growth media for 2 hrs to grow CNx nanostructures. After the 2 hrs of CNx growth treatment, the system was cooled under nitrogen to room temperature. A separate 1 inch outer diameter pyrolysis tube system was also used to fabricate carbon nanostructures. The previously described procedure was identical except that the high purity nitrogen flow was 65 mL/min for heating and CNx nanostructure growth. Figure 20 illustrates the growth conditions of CNx catalyst. The resulting carbon nanostructures were treated in 1 M HCl at 60 °C to remove

72

Figure 20. Pyrolytic Growth of CNx (Courtesy of Elizabeth J. Biddinger) the non-electrically conducting magnesia support and any exposed iron. After the acid treatment the nanostructures were rinsed with ~1 L deionized and distilled water while under aspirator vacuum filtration. The resulting mater was washed off the filter paper with dionized and distilled water into a glass beaker and then dried in a convection oven at 110 °C for 24 hrs. The resulting dry material was considered nitrogen-containing carbon ORR catalyst (CNx). All materials referred to as CNx are prepared in this manner unless otherwise noted.

73

4.2.2 Phosphorus Grown CNx Catalyst Synthesis

CNxPy catalysts were synthesized by pyrolyzing acetonitrile over nanopowder magnesia support (Aldrich) incipient-wetness-impregnated with 2 wt% Fe using an iron acetate (Aldrich) aqueous solution, following the same procedure reported for

73 phosphorus-free CNx previously . When doped with phosphorus, an additional incipient wetness impregnation step with triphenylphosphine (Sigma Aldrich) in diethyl ether was used. Triphenylphosphine was dissolved in a volume of diethyl ether that was equal to the Barrett, Joyner and Halenda technique pore volume of the nanopowder magnesia support. One gram of growth media was then weighed into a quartz calcination boat and placed inside of a quartz calcination tube in a high temperature furnace, shown in Figure

19. The quartz calcination tube (1 inch outer diameter) was purged with high purity nitrogen at 65 mL/min for 30 min and then heated at 10 °C /min until 900 °C was reached. Upon reaching 900 °C, acetonitrile (CH3CN, Fisher, Optima grade)-saturated nitrogen gas at 65 mL/min was streamed over the growth media for 2 hrs for catalyst growth. Following the high temperature acetonitrile treatment the furnace was cooled in high purity nitrogen flowing at 65 mL/min to room temperature. The resulting material was treated in 1M HCl (aq) for 1 hr at 60 C to remove exposed magnesia and metals, then rinsed with approximately 1 L of excess deionized and distilled water under vacuum filtration with an aspirator. The resulting carbon nanostructures were dried in a convection oven at 110 °C in air over night. Upon drying, the resulting graphitic carbon-

74 nitrogen ORR catalyst was denoted as “CNx” if it is phosphorus-free and “CNxPy” if the synthesis involved a triphenylphosphine impregnation step.

4.3 H2S Treatment of CNx Catalyst

To study the effects of sulfur poisoning, previously fabricated CNx was post- treated in a variety of atmospheres. Between two loosely plugged ends 150 mg of CNx was placed into a 0.25 inch inner diameter quartz reactor tube in a high temperature furnace. The quartz tube was purged with high purity nitrogen at 33 ml/min for 30 min and then, heated at 10 °C /min from room temperature until 350 °C was reached. At 350

°C, the catalyst was treated with 1050 ppm H2S/N2, 5.7 %H2/N2, or high purity nitrogen at

33 ml/min for 2 hrs. At the end of 2 hr treatment the system was cooled in nitrogen gas at

33 ml/min to room temperature. Control experiments were carried out over 20 wt%

Pt/Vulcan Carbon (BASF) with the same treatment procedures.

4.4 Rotating Ring Disk and Rotating Disk Electrode Techniques

4.4.1 Oxygen Reduction Activity and Selectivity Testing

Oxygen reduction activity and selectivity of the electrocatalysts were determined by electrochemical half-cell testing with a rotating ring disk electrode (RRDE).

Electrodes were suspended in oxygen saturated sulfuric acid to mimic the protons, 75 reactant dioxygen, and sulfonic acid groups found in Nafion that would be present in the cathode of a direct methanol or polymer electrolyte fuel cell. Catalyst inks were prepared using a composition of 5 mg catalyst, 50 mg 5% Nafion in aliphatic alcohols (Dupont), and 800 mg 100% ethanol. Inks were low-energy sonicated in a 1.5 mL capped vial for

30 min. Three 5 L-aliquots of catalyst ink were applied to a model 636 RRDE setup

(PAR) with a 0.1642 cm2 glassy carbon disk, resulting in a catalyst loading of 426

µg/cm2 assuming a homogeneous mixture. One 6 and one 5.5 L-aliquots of catalyst ink were applied to a model 616 rotating disk electrode setup (PAR) 0.1256 cm2 glassy carbon disk, resulting in a catalyst loading of 427 mg/cm2. A model 636 RRDE setup was connected to a Princeton Applied Research Bistat for the RRDE electrochemical testing.

A model PAR 616 rotating disk electrode setup was connected to a Princeton Applied

Research Bistat for RDE electrochemical testing. An Ag/AgCl (saturated KCl) reference electrode and a Pt wire counter electrode were used for the half-cell system. Prior to testing the Ag/AgCl (saturated KCl) reference electrode used was compared against a separate master Ag/AgCl (saturated KCl) reference electrode to ensure that the potential reported was within 5 mV. All reported potentials are referenced versus the normal hydrogen electrode (NHE). The standard electrode potential of an Ag/AgCl (saturated

KCl) reference electrode is 0.197 V vs. NHE at 25 ºC. The half-cell electrolyte was 0.5 M

H2SO4. It has been reported that there is no significant difference in ORR testing of

44,60 nanostructured carbon nitrogen catalysts between H2SO4 and HClO4 solutions. This

- - demonstrates that there is no difference in the adsorption of the HSO4 or ClO4 anions on the active site of carbon-nitrogen graphitic catalysts. It should be noted that H2SO4 is

76 thought to poison platinum based catalysts 263, but all electrochemical methods were identical for all platinum supported on Vulcan carbon catalysts studied, so the deactivation would be equivalent, making any activity/selectivity differences attributable to the chemical treatments used.

ORR catalyst testing used cyclic voltammetry (CV) on the catalyst coated glassy carbon disk while monitoring hydrogen peroxide products with the surrounding platinum ring held at a constant potential of 1.2 V vs. NHE. All disk CVs were scanned from 1.2 V to 0.0 V to 1.2 V vs. NHE for CNx catalysts. All disk CVs were scanned from 1.2 V to

0.2 V to 1.2 V vs. NHE for platinum-based catalysts. The rotating ring disk electrode experimental setup is composed of a rotating Teflon cylinder with a circular, test catalyst coated, gassy carbon electrode in the center surrounded by a thin platinum ring electrode, shown in Figure 21. The disk rotates in a controlled manner to facilitate a convective flow of dissolved reactant to the catalyst coated disks surface. An 100 mL H2SO4 electrolyte solution, allows for the transfer of current through the test solution. To remove any gaseous oxygen from the catalyst pores and fully wet the catalyst surface and

77 pores oxygen reduction activity testing the electrolyte was saturated with oxygen and

Figure 21. Diagram of the Rotating Ring Disk Electrode Experimental Setup disk CVs were run at 10 mV/s. Prior to the collection of the background, the half-cell was purged with argon for 30 min to remove all significant amounts of dissolved oxygen from the electrolyte. Disk CVs were run at a scan rate of 50 mV/s to remove adsorbed impurities from the catalyst coated glassy carbon electrode surface and twenty platinum ring CVs were run at 100 mV/s from -0.1 to 1.8 V to clean the platinum ring.

Immediately after the last platinum ring scan, a disk CV was run at 10 mV/s paired with the ring held at 1.2V vs. NHE while the assembly was rotating at 100 rpm to collect the argon saturated solution background.

78

Then, the electrolyte was saturated with oxygen for 30 min to determine activity and water formation selectivity of the catalysts. Once again, disk CVs were run at a scan rate of 50 mV/s to eliminate impurities from the catalyst coated glassy carbon electrode surface and ten ring CVs were run at 100 mV/s from -0.1 to 1.8 V to clean the platinum ring. Immediately following the last ring scan, oxygen was bubbled through the electrolyte for 1 - 2 min to re-saturate the solution, following which a 1000 rpm oxygen- saturated electrolyte disk CV at 10 mV/s was performed while holding the ring at 1.2 V

Figure 22. Illustration of Oxygen Reduction Activity and Selectivity on an RRDE setup

vs. NHE. Additional CVs were run at 100 rpm and 0 rpm in oxygen-saturated electrolyte using a scan rate of at 10 mV/s.

79

The rotating ring and disk electrodes on the tip of the RRDE setup determine the oxygen reduction selectivity and activity of the studied catalyst, Figure 22.

The rotation of the cylindrical electrode brings oxygen saturated electrolyte to the oxygen reduction catalyst coated disk where it undergo two possible reactions 25:

+ - O2 + 4 H + 4 e  2 H2O (1.229 V vs. NHE)

+ - O2 + 2 H + 2 e  H2O2 (0.695V vs. NHE)

Due to the lower standard reaction potential and corrosive nature of the undesired hydrogen peroxide product the overall fuel cell performance and lifetime are decreased, respectively. Since both reactions take place simultaneously on the cathode, the selectivity of ORR on the catalyst surface is important in PEM and direct methanol fuel cells. Selectivity of the catalyst is reported as the number of electrons transferred per oxygen molecule (n). The n value is calculated by comparing the ring current (IR), corrected by collection efficiency of the ring (N), to the disk current (ID) using the following equation:

4I n  D ID  (IR /N)

An n value of 4.0 is equivalent to 100% water formation, while an n value of 2.0 is  equivalent to 100% hydrogen peroxide formation. The collection efficiency was reported by the manufacturer as N = 0.22.

80

4.4.2 Collection Efficiency Testing

3- 4- A potassium ferricyanide (Sigma-Aldrich) [Fe(CN)6] / [Fe(CN)6] redox couple was studied in 0.1 M NaOH to determine the collection efficiency of the platinum ring surrounding the catalyst-coated, glassy carbon disk working electrode. The ferricyanide metal couple is commonly used to study the experimental collection efficiency of RRDE systems.240,264-266 An Hg/HgO (sat. KOH) reference electrode with a saturated KOH filled

Luggin capillary with a molecular sieve tip was used to determine potential. A Hg/HgO

(sat. KOH) reference electrode was used because it is more stable than an Ag/AgCl (KCl) reference electrode in basic solutions. The standard reference electrode potential of a

Hg/HgO (sat. KOH) electrode is 0.2412 V vs. NHE at 25 ºC. Catalyst inks were prepared using a composition of 5 mg catalyst, 50 mg 5% Nafion in aliphatic alcohols (Dupont) and 800 mg 100% ethanol. Inks were low-energy sonicated in a 1.5 mL capped vial for

30 min. The total volume of the catalyst ink was approximately 1 mL. Three 5 L- aliquots of catalyst ink were applied to a model 636 RRDE setup (PAR) with a 0.1642 cm2 glassy carbon disk, resulting in a catalyst loading of 426 µg/cm2 assuming a homogeneous mixture. A fresh 0.1 M NaOH solution was sparged with argon gas for a time exceeding 30 minutes prior to testing and was continuously sparged during all collection efficiency tests. The 0.1 M NaOH needed to be fresh for accurate and reproducible collection efficiency data, due to the glass etching that occurs when a strong base is kept in a glass receptacle like the RRDE electrolyte basin.267 Electrochemical testing began with consecutive CVs on the disk from 0.72V to -0.43V to 0.72 V vs. NHE 81 at 50 mV/s while holding the ring at 0.72 V vs. NHE until reproducible scans of both the ring and disk were achieved. Next, a baseline for the disk was obtained by sweeping from 0.72V to -0.43V to 0.72 V vs. NHE at 10 mV/s in the argon saturated solution at a

100 rpm rotation rate. The half cell solution was quickly changed to argon saturated 0.1M

NaOH and 50 mM K3Fe(CN)6. Once again, consecutive CVs were run on the disk from

0.72V to -0.43V to 0.72 V vs. NHE at 50 mV/s while holding the ring at 0.72 V vs. NHE until steady state ring and disk collection was achieved in the new ferricyanide containing solution. Collection efficiency data was taken by sweeping the disk from 0.72V to -

0.43V to 0.72 V vs. NHE at 10 mV/s in the argon sparged solution at a 100 and 1000 rpm rotation rates while simultaneously holding the ring at 0.72 V vs. NHE.

Figure 23.. Illustration of Collection Efficiency RRDE setup 82

3- Collection efficiency was determined by comparing the current of [Fe(CN)6]

4- reduced at the catalyst coated disk to the current [Fe(CN)6] oxidized on the surrounding platinum ring, Figure 23. [Fe(CN)6]3- can only undergo one reduction reaction so it can be assumed that 100 % of the current at the disk contributes to the reduction of

3- 4- 3- [Fe(CN)6] to form [Fe(CN)6] . The amount of [Fe(CN)6] reduced at the disk was determined by subtracting the background current collected in 0.1M NaOH from the collection current taken in 0.1M NaOH and 50 mM K3Fe(CN)6. The current of the ring

4- was established by subtracting the base capacitance before [Fe(CN)6] oxidation from the ring current. The selectivity of the catalyst covered glassy carbon disk (n) was known to be 1 because the only reaction that could occur was the one-electron transfer of

3- 4- [Fe(CN)6] to form [Fe(CN)6] . The collection efficiency of the ring (N) was calculated by comparing the ring current (IR), corrected by, to the disk current (ID) using the following equation:

1 I n  D I D  (I R / N)

The collection efficiency reported by the manufacturer was N = 0.22 for the model MT28

Series ThinGap RRDE (Pine Research) with a 0.1642 cm2 glassy carbon disk and platinum ring.

83

4.4.3 Carbon Corrosion Testing

Cyclic voltammograms were performed on a catalyst coating rotating disk electrode to measure the Faradaic current of the electrochemically active quinone/hydroquinone redox reaction to indicate the degree of carbon corrosion using a

PAR bipotentiostat. The electrochemically detectable quinone/hydroquinone redox couple is a likely intermediate in the oxidation of carbon graphites.23 By examining this reactive intermediate on the electrode surface the degree of carbon corrosion can be studied on the catalyst during normal electrochemical conditions.

To simulate the fuel cell environment, half-cell corrosion tests were performed in fresh 0.5 M H2SO4 with a rotating ring disk electrode (RRDE) (PAR Model 636

Electrode Rotator). An Ag/AgCl (Sat. KCl) reference electrode (Gamry Instruments) was used to determine reference potential. A platinum wire counter electrode was also used in the half-cell setup, Figure 24. Argon was continuously sparged at 5 ccm through the 0.5 M H2SO4 electrolyte

84

Figure 24. Electrochemical Rotating Disk Electrode Carbon Corrosion Setup (inset; the hydroquinone/quinone reaction and standard reaction potential) during testing and at least 30 min prior to testing so that changes in the oxygen reduction activity would not mask the detection of the quinone/hydroquinone species. A catalyst ink was prepared by mixing 1:10:160 catalyst: 5% Nafion in aliphatic alcohols

(Electrochem): 100% ethanol (Decon) by weight and was immediately sonicated for 30 min. After sonication, three 5 μL aliquots of catalyst ink were applied to the glassy carbon (GC) disk electrode of the RRDE (MT28 Series ThinGap RRDE Pine Research), allowing the ink to dry between applications. The catalyst loading deposited on the glassy carbon RRDE disk was 426 μg/cm2 assuming uniform ink dilution. Once the catalyst- coated electrode had dried, it was wetted with DI-H2O before being submerged in the 0.5

M H2SO4 electrolyte. Holding the catalyst-coated electrode at 1.2 V vs. NHE for 48 hours 85 accelerated carbon corrosion. The progress of carbon corrosion was monitored by intermittent CVs from 1.2 V to 0.2 V to 1.2 V vs. NHE at 10 mV/sec after 0, 2, 4, 8, 16,

24, and 48-hours of accelerated carbon corrosion potential hold. A series of 5 CVs was used at each carbon corrosion interval to ensure that the recorded signal had reached steady state. All corrosion comparisons were done using steady state CVs. The intermittent CVs were used to measure the evolution of the hydroquinone/quinone with time of potential hold. All corrosion tests were performed at 1000 rpm and at room temperature, 23  1 C. Similar investigations have been performed that demonstrate carbon corrosion on Pt on carbon23,268 and carbon nanotubes.20 It should be pointed out that in a membrane electrode assembly the carbon gas diffusion layer is especially susceptible to carbon corrosion17, so the following study has focused solely on the occurrence of carbon corrosion on cathode catalyst materials.

4.4.4 Carbon Monoxide Poisoning Testing

A model PAR 616 RDE setup was connected to a Princeton Applied Research

Bistat for RDE electrochemical corrosion and activity testing. All carbon monoxide testing were performed in a platinum free electrochemical system to avoid deactivation of platinum electrodes due to carbon monoxide adsorption, shown in Figure 25. An

Ag/AgCl (saturated KCl) reference electrode and a carbon graphite counter electrode were used in the half-cell system. Prior to testing the Ag/AgCl (saturated KCl) reference

86 electrode used was compared against a separate master Ag/AgCl (saturated KCl) reference electrode to ensure that the potential reported was within 5 mV. All reported potentials are referenced versus the normal hydrogen electrode (NHE). The standard electrode potential of an Ag/AgCl (saturated KCl) reference electrode is 0.197 V vs.

NHE at 25 ºC. The half-cell electrolyte was 0.5 M H2SO4.

Figure 25. Electrochemical Rotating Disk Electrode Corrosion Setup.

Oxygen reduction cathodic linear voltage scans were collected by cyclic voltammetry on the catalyst coated glassy carbon disk from 1.2 V to 0.0 V to 1.2 V vs.

NHE at 10 mV/s, 1000 rpm in 0.5M H2SO4 (aq). Before electrochemical testing, the 100 mL 0.5 M H2SO4 electrolyte was saturated with oxygen by diffusing gaseous oxygen through the electrolyte. Immediately, after the electrode was submerged in the electrolyte a CV was collected to ascertain the initial activity of the catalyst.

87

After the initial activity CV a 0.3 V vs. NHE potential hold induced on catalyst surface 1000 rpm RDE the electrolyte was saturated with different gasses. At the beginning of the potential hold the electrolyte was saturated with oxygen. After 1.5 minutes at 0.3 V vs. NHE the gas diffused through the electrolyte was changed to either

30 ccm oxygen, 30 ccm argon, 30 ccm carbon monoxide, 30 ccm oxygen and 30 ccm carbon monoxide, or 30 ccm argon and 30 ccm oxygen. After 26.5 minutes at 0.3 V vs.

NHE the diffusion gas was switched back to oxygen 30 ccm. The electrolyte was allowed to re-saturate with oxygen while the disk potential was held at 0.3 V vs. NHE. The 0.3 V vs. NHE potential hold test was terminated after 65 minutes. The current density varied slightly for each catalyst application even within catalyst types. For the potential hold experiments it was necessary to normalize the recorded currents by largest current in each data set so that trends relative to the electroactive area could be compared.

After the potential hold had concluded an oxygen reduction cathodic linear voltage scan was collected by cyclic voltammetry from 1.2 V to 0.0 V to 1.2 V vs. NHE at 10 mV/s while the electrode rotated at 1000 rpm to determine the post treatment activity of the catalyst.

4.4.5 Chronoamperometric CO Poisoning Experimentation

The 100 mL 0.5 M H2SO4 electrolyte was saturated with oxygen by diffusing gaseous oxygen through the electrolyte. Immediately after the electrode was submerged 88 in the electrolyte CV from 1.2 V to 0.0 V to 1.2 V vs. NHE at 10 mV/s on the catalyst coated glassy carbon disk at 1000 and 0 rpm was collected ascertain the initial activity of the catalyst and oxygen functional groups, respectively.

After the first CV concluded the gas that was diffused through the electrolyte was changed to either CO at 30 ccm or both 30 ccm CO and 30 ccm O2 for 30 minutes. Thirty minutes was sufficient to saturate the electrolyte as demonstrated by the chronoamperometric experiments as previously described. Then, gas treatment CV from

1.2 V to 0.0 V to 1.2 V vs. NHE at 10 mV/s on the catalyst coated glassy carbon disk at

1000 and 0 rpm was collected.

Then, gaseous oxygen was diffused through the electrolyte at 30 ccm through the electrolyte for 30 minutes. A post treatment CV from 1.2 V to 0.0 V to 1.2 V vs. NHE at

10 mV/s on the catalyst coated glassy carbon disk at 1000 and 0 rpm was collected in the oxygen saturated electrolyte.

Finally, gaseous argon was diffused through the electrolyte at 30 ccm through the electrolyte for 30 minutes. A background CV from 1.2 V to 0.0 V to 1.2 V vs. NHE at 10 mV/s on the catalyst coated glassy carbon disk at 1000 and 0 rpm was collected in the argon saturated electrolyte.

89

4.5 Fuel Cell Testing and Membrane Electrode Assembly Fabrication

4.5.1 Membrane Electrode Assembly Fabrication

Direct methanol and polymer electrolyte membrane electrode assemblies are composed of an anode and cathode catalyst layer opposite of one another separated by a sulfonated tetrafluoroethylene based fluoropolymer-copolymer membrane. Gas diffusion layers composed of carbon paper treated with a hydrophobic synthetic fluoropolymer of tetrafluoroethylene are placed on top of the anode and cathode catalyst layers to deal with water management. The unit comprising all of these components is known as the membrane electrode assembly (MEA).

There are many different ways to manufacture MEAs all of which affect the performance of the MEA differently under different operating conditions. MEA manufacturing techniques can be separated into two main groups; direct application of the catalyst ink to the sulfonated tetrafluoroethylene based fluoropolymer-copolymer membrane or catalyst ink application to substrate with a heated pressing step needed to adhere the catalyst layer to the sulfonated tetrafluoroethylene based fluoropolymer- copolymer membrane. A further discussion of the difference in performance caused by these two fabrication techniques is found in Fuel Testing Section. Both bench scale techniques have been developed in the Heterogeneous Catalysis Research Group laboratory at The Ohio State University. 90

4.5.1.a Hot Press Fabrication of Membrane Electrode Assemblies

Catalyst inks where painted onto a fiberglass decal, dried and hot pressed onto a cation exchanged Nafion membrane.

Nafion membranes were prepared before the catalyst coated decals and submerged in DI-H2O for storage. The Nafion membranes were cut into 2.5 inch squares, which were deemed of sufficient size to achieve a gas tight seal during fuel cell testing.

Nafion 115, 117, and 212 membranes (Electrochem) were cleaned by boiling 3 % hydrogen peroxide for one hour while the membrane layers were sandwiched between two holey Teflon disks. After the boiling treatment the 3 % hydrogen peroxide was cooled to room temperature then disposed as chemical waste. During the cleaning process the Nafion casting sheets, if present, delaminated from the membranes. These rigid, clear, shiny casting sheets were discarded once the 3 % hydrogen peroxide solution cooled to room temperature. The purpose of the chemically inert, holey Teflon disks is to keep the

Nafion membranes submerged in the solution without affecting the chemistry of the cation exchange. The Nafion membranes and holey Teflon disks were then rinsed with deionized and distilled water. Next, the Nafion membranes were treated in boiling deionized and distilled water for one hour while the membrane layers were sandwiched between two holey Teflon disks to remove any absorbed hydrogen peroxide. The solution 91 was then cooled to room temperature and discarded as chemical waste. Next, the Nafion membranes were treated in boiling 1 % NaOH or 0.5 M H2SO4 for one hour while the membrane layers were sandwiched between two holey Teflon disks to cation exchange the fluoride ions in the membrane to the Na+ or H+ state, respectively. The Na+ cation exchange Nafion is more thermally stable which allows for higher temperature pressing conditions without damaging the proton conductivity of the protonated or untreated

Nafion membranes.24 Membranes that have been treated in 1 % NaOH will be considered

+ Na cation exchanged. Membranes that have been treated in 0.5 M H2SO4 will be considered H+ cation exchanged. The solution was cooled down to room temperature and disposed of as chemical waste. The Nafion membranes and holey Teflon disks were then rinsed with deionized and distilled water. Next, the Nafion membranes were treated in boiling deionized and distilled water for one hour while the membrane layers were sandwiched between two holey Teflon disks to remove any unbound cations and anions.

The solution was cooled to room temperature and disposed of as chemical waste. The

Nafion membranes were then submerged in deionized and distilled water in a labeled sealed container until use.

The active areas of the fuel cell the Electrochem and Arbin fuel cell fixtures were

5 cm2 so fiberglass decals of the same dimension were cut out of a woven fiberglass sheet. The fiberglass decals were then coated with a synthetic fluoropolymer of tetrafluoroethylene, also known as Dupont's Teflon, and dried in air at 110 ºC overnight so that the catalyst layers would not adhere to the decals. The dry Teflon coated 5 cm2 fiber glass sheets were stored and sealed in labeled plastic bag until use.

92

Anode and Cathode inks of differing compositions were thoroughly mixed than painted onto the Teflon coated fiberglass sheets and dried until the desired catalyst loadings were reached. The anode ink mixture of ~5 : 11-25 : 6-13 , C1-20 20 wt% HP Pt on VC XC-72 (BASF): 5 wt% Nafion in aliphatic alcohols (electrochem) : glycerol

(Fisher Scientific), was mixed together in a small 1.5 mL glass vial. A small Teflon coated magnetic stirrer was placed into the catalyst ink vial then the vial was capped. The catalyst ink was magnetically stirred for 1 hr to make the ink a uniform mixture. At the end of the stirring treatment 50 mg of Tetrabutylammonium hydroxide in 1 M methanol

(Acros Organics) was added for every gram of 5 wt% Nafion in aliphatic alcohols used in the ink solution to Na+ cation exchange the Nafion in the catalyst ink. The catalyst ink was capped and magnetically stirred for another hour to make the ink a uniform mixture.

After stirring the catalyst ink was painted with a fine camel-haired paintbrush onto a dry

Teflon coated 5 cm2 fiber glass sheet decal. After each thin coat of catalyst ink the decal was dried in air at 140 ºC for one hour. If the decals are dried at higher temperatures the proton conductivity of Na+ cation exchanged Nafion was decreased leading to poorer fuel cell performance. Catalyst ink painting was done in alternating directions in an effort to evenly disperse the catalyst on the decal. After each drying treatment the decal was weighed until a loading of 1 mgPt/cm2 was accomplished. The cathode ink mixture of ~5

: 33.3 - 300 : 17-150 , non-precious metal CNx catalyst: 5 wt% Nafion in aliphatic alcohols (electrochem) : glycerol (Fisher Scientific), was mixed together in a 1.5 mL glass vial. A small Teflon coated magnetic stirrer was placed into the catalyst ink vial then the vial was capped. The catalyst ink was magnetically stirred for 1 hr to make the

93 ink a uniform mixture. At the end of the stirring treatment 50 mg of

Tetrabutylammonium hydroxide in 1 M methanol (Acros Organics) was added for every gram of 5 wt% Nafion in aliphatic alcohols used in the ink solution to Na+ cation exchange the Nafion in the catalyst ink. The catalyst ink was capped magnetically stirred for another hour to make the ink a uniform mixture. After stirring the catalyst ink was painted with a fine camel-haired paintbrush onto a dry Teflon coated 5 cm2 fiber glass sheet. After each thin coat of catalyst ink the decal was dried in air at 140 ºC for one hour. Catalyst ink painting was done in alternating directions in an effort to evenly disperse the catalyst on the decal. After each drying treatment the decal was weighed

2 until a loading of 1 to 10 mgCNx/cm was accomplished. After the decals had reached the desired catalyst loading they were kept at room temperature until hot pressing.

The dried anode ink was composed of 10 to 20 wt% Nafion with the balance being C1-20 20 wt% HP Pt on VC XC-72 (BASF) which is 20 wt% nanoparticle

Platinum on high surface area Vulcan carbon support. The dried cathode ink was comprised of 25 to 75 wt% Nafion with the balance being C1-20 20 wt% HP Pt on VC

XC-72 (BASF) or 25 to 75 wt% Nafion with the balance being non-precious metal carbon-nitrogen graphite nanostructures.

Once the anode and cathode decals were complete they were hot pressed to the

Na+ cation exchanged Nafion membrane. The top and bottom plates of the heated press were heated to 210 ºC while the plates were kept apart. The Na+ cation exchanged Nafion membrane was removed its DI-H2O storage container and blotted dry with a clean

Kimwipe. The following stacked assembly from bottom to top was placed into the heated 94 press; red furan sheet to disperse the pressure, aluminum plate, a Teflon coated fiberglass sheet larger than the Nafion membrane, catalyst coated cathode decal facing upward, Na+ cation exchanged Nafion membrane, catalyst coated anode decal facing down opposite the cathode decal, a Teflon coated fiberglass sheet larger than the Nafion membrane, aluminum plate, and a red furan sheet to disperse the pressure. The purpose of the Teflon coated fiberglass sheets were to keep the Nafion from binding to other components. The stacked assembly was quickly but carefully placed between the heated press plates. The stacked assembly was pressed at 550 lbs for 5 min. The stacked assembly was removed and allowed to cool to room temperature with 5 lbs pressure. Once at room temperature the stacked assembly was carefully taken apart. The Teflon coated decals were slowly peeled away from catalyst layers, which after hot pressing are adhered to the Na+ cation exchanged membrane. The removed decals were then weighed to determine the mass of electrode deposited on the membrane electrode assembly.

After the decals were removed the membrane electrode assembly was cation exchanged from the Na+ to the H+ cation. The Na+ membrane electrode assemblies were sandwiched between two holey Teflon disks and placed in a boiling 0.5 M H2SO4 for one hour to cation exchange the fluoride ions in the membrane from the Na+ to the H+ state.

Next, the membrane electrode assemblies were sandwiched between two holey Teflon disks and treated in boiling deionized and distilled water for one hour while to remove any unbound cations and anions. The solution was allowed to cool to room temperature and membrane electrode assembly was removed and blotted dry with a clean Kimwipe.

95

The dried membrane electrode assembly was placed in a sealed labeled plastic bag until further use.

4.5.1.b Direct Catalyst Ink Application on Membrane MEA Fabrication

Nafion membranes were prepared before the catalyst coated decals and submerged in DI-H2O for storage. The Nafion membranes were cut into 2.5 inch squares, which was deemed sufficient size to get a gas tight seal during fuel cell testing. Nafion

115, 117, and 212 membranes (Electrochem) were cleaned by boiling 3 % hydrogen peroxide for one hour while the membrane layers were sandwiched between two holey

Teflon disks. After the boiling treatment the 3 % hydrogen peroxide was cooled to room temperature then disposed as chemical waste. During the cleaning process the Nafion casting sheets delaminated from the membranes. These rigid, clear, shiny casting sheets were discarded once the 3 % hydrogen peroxide solution cooled to room temperature.

The Nafion membranes and holey Teflon disks were then rinsed with deionized and distilled water. Next, the Nafion membranes were treated in boiling deionized and distilled water for one hour while the membrane layers were sandwiched between two holey Teflon disks to remove any absorbed hydrogen peroxide. The solution was then cooled to room temperature and discarded as chemical waste. Next, the Nafion membranes were treated in boiling 0.5 M H2SO4 for one hour while the membrane layers were sandwiched between two holey Teflon disks to cation exchange the fluoride ions in

96

+ the membrane to the H state. Membranes that have been treated in 0.5 M H2SO4 will be considered H+ cation exchanged. The solution is cooled down to room temperature and disposed of as chemical waste. The Nafion membranes and holey Teflon disks were then rinsed with deionized and distilled water. Next, the Nafion membranes were treated in boiling deionized and distilled water for one hour while the membrane layers were sandwiched between two holey Teflon disks to remove any unbound cations and anions.

The solution was cooled to room temperature and disposed of as chemical waste. The

Nafion membranes were then submerged in deionized and distilled water in a labeled sealed container until use.

Anode and Cathode inks of differing compositions were thoroughly mixed then painted onto the Teflon coated fiberglass sheets and dried until the desired catalyst loadings were reached. The anode ink mixture of ~5 : 11-25 : 6-13 , C1-20 20 wt% HP Pt on VC XC-72 (BASF): 5 wt% Nafion in aliphatic alcohols (electrochem) : deionized and distilled water, was mixed together in a small flat bottom glass vial. The catalyst ink was low energy sonicated for 1 hr in an ice bath to make the ink mixture uniform. A H+ cation exchange Nafion membrane was centered onto a heated vacuum table (Nuvant) with a smaller piece of Teflon coated fiberglass underneath to disperse the vacuum pressure on the MEA active area. A red furan decal covered the surface of the vacuum table with a square 5.75 cm2 area cutout was placed on top of the H+ Nafion membrane such that the cutout was centered. The H+ Nafion membrane was then heated to 80 ºC under vacuum.

After sonication the catalyst ink was painted with a fine camel-haired paintbrush into the furan decal cutout where the H+ Nafion membrane was at 80 ºC. After each thin coat of

97 catalyst ink the decal was dried in air at 70 ºC on the vacuum table for five minutes.

Catalyst ink painting was done in alternating directions in an effort to evenly disperse the catalyst on the decal. The catalyst ink was kept in a ice during the painting process to prevent catalyst-Nafion agglomerations from forming. The entire contents of the catalyst ink were deposited onto the Nafion membrane to reach the desired catalyst loading of 1 mgPt/cm2. After the last anode ink coating the H+ Nafion membrane was heated to 80 ºC to cure the anode catalyst layer for 30 min. The catalyst loading for the Nafion membrane was calibrated for each ink composition by measuring the weight gain for a fiberglass sheet under the same conditions. After anode curing the membrane was cooled to room temperature under table vacuum. The table vacuum was tuned off and the membrane was now ready for H+ Nafion membrane cathode coating on the reverse side of the anode. A cathode ink mixture of ~5 : 33.3 - 300 : 17-150 , non-precious metal CNx catalyst: 5 wt%

Nafion in aliphatic alcohols (electrochem) : deionized and distilled water, was mixed together in a small flat bottom glass vial. The catalyst ink was low energy sonicated for 1 hr in an ice bath to make the ink mixture uniform. A cured anode coated H+ cation exchange Nafion membrane was centered onto a heated vacuum table (Nuvant) with a smaller piece of Teflon coated fiberglass underneath with the anode face down. A red furan decal covered the surface of the vacuum table with a square 5 cm2 area cutout centered on the H+ Nafion membrane such that the cutout was in the center of the anode electrode on the opposite side. The membrane was then heated to 70 ºC under vacuum.

After sonication the catalyst ink was painted with a fine camel-haired paintbrush into the furan decal cutout where the H+ Nafion membrane was at 70 ºC. After each thin coat of

98 catalyst ink the decal was dried in air at 70 ºC on the vacuum table for five minutes.

Catalyst ink painting was done in alternating directions in an effort to evenly disperse the catalyst on the decal. The entire contents of the catalyst ink were deposited onto the

2 Nafion membrane to reach the desired catalyst loading of 1 - 6 mgCNx/cm . The catalyst loading for the Nafion membrane was calibrated for each ink composition by measuring the weight gain of a fiberglass sheet under the same conditions. After the last anode ink coating the H+ Nafion membrane was heated to 80 ºC to cure the anode catalyst layer for

30 min. After curing the membrane electrode assembly was cooled to room temperature under table vacuum. The table vacuum was tuned off and the membrane was now ready fuel cell testing.

4.5.2 Electrochemical Fuel Cell Testing

The membrane electrode assembly was installed into a fuel cell fixture for full fuel cell testing. PEM fuel cell has a "sandwich" type of configuration where two catalyst layers are adhered to either side of an electrically insolating yet proton conducting polymer electrolyte membrane, shown in Figure 26. A humidified sulfonated

99

Figure 26. Diagram of Hydrogen Polymer Electrolyte Fuel Cell. tetrafluoroethylene based fluoropolymer- copolymer, commercially available as Dupont's

Nafion 115, 117, 212, was used as the polymer electrolyte membrane. The last two digits of the Nafion membrane number as received from the Dupont indicate the thickness of the membrane in Mils. The first number indicates the polymer manufacturing process for each membrane material. The stationary fluoride anions in Nafion conduct protons created in the anode of the membrane electrode assembly through the electrically

100 insolating polymer electrolyte membrane. The anode composition remains constant as 1 mgPt/cm2 in the MEAs tested with a Nafion content of 10 – 20 wt% in the electrode layer in order to study the effect of other phenomena. Both the anode and cathode catalyst layers are composed of catalyst supported on electrically conducting carbon graphite. On the outside of each catalyst layer a gas diffusion layer composed of polytetrafluoroethylene treated carbon paper, carbon Toray paper 060 (electrochem), was used to diffuse the reactants in a controlled manner, regulate water management, and conduct electrons that participate in respective reactions.24 The carbon Toray paper was cut into 5 cm2 area squares that were custom sized to fit into the 5 cm2 square holes in the gasket material where the electroactive are of the MEA was positioned during operation.

Gaskets composed of chemically stable silicone (Electrochem) or Teflon (Electrochem or

McMasterCarr) with differing thicknesses was used to form a gas tight seal around the 5 cm2 electroactive area. It should be noted that the active area of the anode was manufactured to be slightly larger than the active are of cathode layer so that the cathode layer would not overlap the anode layer which would cause a large potential difference across the MEA causing membrane degradation that leads to catastrophic failure due to holes formed in overlapping positions of the MEA.269 Membrane electrode assemblies were closed with 70-150 in-lbs of torque to reach an 18-25 % compression to assure optimal connectivity of the gas diffusion and catalyst layers.44 The compression of the

MEA fixture was determined using the following equation;

44 MEA compression (%) = 100% (tMEA – tmembrane – tgasket ) / (tMEA – t membrane).

101

The thickness of the free standing MEA not under compression as measured by a digit outside micrometer is tMEA. The thickness of the non-humidified Nafion membrane as measured by a digit outside micrometer is tmembrane. The combined thicknesses of the cathode and anode gaskets are tgasket. The compression of Teflon gaskets is assumed to be zero so the manufactures thickness is used for tgasket. The compression of silicon gaskets is a function of closure pressure so the MEA compression equation above was necessary to determine MEA percent compression. If the compression of the MEA was too high the gas diffusion layer was crushed causing poor transfer of reactant gasses and a decrease in the current density achievable due to increase mass transfer limitations. Due to variations in the MEA thickness, tMEA, gaskets of different thicknesses from 2 – 12 Mils were used to achieve a MEA compression of 18 – 25 %. Electrically conductive graphite flow channels in fuel cell fixture (Arbin or Electrochem) were located outside of the carbon

Toray paper gas diffusion layers had a serpentine flow channels that was used to uniformly deliver the humidified reactant gasses to both the anode and cathode, shown in grey in Figure 26. Electrical interconnects were attached to the electrically conducting current collection plates places outside of the conductive graphite flow channels of the 5 cm2 active area fuel cell test fixture to complete the circuit between the two coupled anode and cathode reactions (Arbin or Electrochem).

Since an operational fuel cell simultaneously uses both oxygen and hydrogen gases it was safety critical to perform leak checking before testing to ensure that reactive gases do not mix during testing which could potentially cause a fire or explosion. After the membrane electrode assembly was installed into the fuel cell fixture gas leak and

102 crossover testing were completed in high purity nitrogen gas. The gas inlets and exits of the fuel fixture were attached to the 10 Watt fuel cell test station (Arbin) through stainless steel Swagelok or plastic hand-tightened interconnects for the Arbin fuel cell fixture or

Electrochem fuel cell fixture, respectively. The anode and cathode sides of the fuel cell fixture were simultaneously pressurized to 25 psig in high purity nitrogen gas with a flow rate of 200 ccm. The pressure difference between the anode and cathode sides of the

MEA greater than 5 psig could potentially cause a mechanical perforation in the MEA.

Once the anode and cathode pressures reached 25 psig ball valves at the exit of the fuel cell fixture were closed and the gas flows were stopped. If the pressures of the anode and cathode did not decrease by more than 0.5 psig over 15 min than the fuel cell fixture was deemed to have passed the gas leak check. The fuel cell fixture was depressurized to ambient pressure and nitrogen gas flow was restored to 200 ccm for the cathode only.

The cathode side of the MEA was pressurized to 4 psig while keeping the anode at ambient pressure. Once the cathode pressure reached 4 psig a ball valve at the exit of the fuel cell fixture was closed and the gas flow was stopped. If the pressure cathode did not decrease by more than 0.1 psig over 10 min than the fuel cell fixture was deemed to have passed gas crossover testing. This leak check procedure was adapted from the Single Cell

Test Protocol authored by the United States National Laboratories.29

After leak checking, the membrane electrode assembly was conditioned at the desired fuel cell operating conditions at low potential to humidify the Nafion and create proton conduction channel within the MEA. After conditioning a polarization performance curve was generated by chronoamperometric current holds that were

103 resistance corrected with electrochemical impedance. The leak checked fuel cell fixture was slowly heated to 80 ºC in high purity nitrogen gas while the anode and cathode where simultaneously pressurized to 21.75 psig. The inlet gas lines were heated to 75 ºC and the anode and cathode deionized and distilled water bubbling humidifiers were heated to 70 ºC. Once all of these zones had reached their set point temperature high purity nitrogen gas was allowed to flow through the bubbling humidifiers. The 5 ºC temperature difference in the zones was used to prevent water condensation in successive components where humidified gas was present. Once all of the temperature zones restabilized after turning on the humidifiers nitrogen flow was changed to high purity hydrogen flow for the anode and nitrogen flow was changed to high purity oxygen flow for the cathode. It should be noted that hydrogen flow should be switched on before oxygen flow, otherwise a high potential could be induced in the cathode would cause carbon oxidation degradation in area between the carbon sublayer and the catalyst layer.17

Chronoamperometric, where the current is changed with time and the potential is recorded, electrochemical fuel cell testing the oxygen flow is 0.0066 slpm and the hydrogen flow is 0.0279 slpm. This corresponds to a 9.5 stoichiometric flow of oxygen and 2 stoichiometric flow of hydrogen at 0.2 Amps according to the following equation;

Flow Rate = St  F  N  I.

“Flow Rate” was the flow rate of the reactant gas to the fuel cell fixture. The desired stoichiometric ratio of the reactant gas was “St” in units of stoichiometric coefficients. The stoichiometric flow rate calibrated for the Arbin fuel cell test station

104 was “F”, which was equal to 0.0034824 for pure oxygen and 0.0069648 for pure hydrogen. The number of test cells in the fuel cell stack is “N” in integer values, for the single fuel cell testing performed this value was one. The chronoamperometric current that the fuel cell is held at was “I” in units of Amps. The gas flows at open cell potential were chosen at 9.5 = St for oxygen and 2 = St for hydrogen at 0.2 Amps because these are the lowest controllable flow rates for the mass flow controllers (Brooks) installed in the Arbin fuel cell test station. The open cell potential was allowed to stabilize for 15 to

40 minutes. The fuel cell was then chronopotentiometrically, held at a fixed potential while the current was measured, conditioned from 0.2 to 0.4 V at the 0.2 Amp St = 9.5 for oxygen and St = 2 for hydrogen flow rates until the current stabilized in 4 to 48 hrs.

After the fuel cell was conditioned, the fuel cell was allowed to stabilize at open cell potential for 15 to 40 minutes. After the OCV treatment the current was increased to 0.2

Amps for 15 minutes while the stoichiometric flow was maintained at 9.5 for oxygen and

2 for hydrogen. After the 0.2 Amp treatment the current was increased to 0.3 Amps for

15 minutes while the stoichiometric flow was maintained at 9.5 for oxygen and 2 for hydrogen. After the 0.3 Amp treatment the current was increased to 0.5 Amp for 15 minutes while the stoichiometric flow was maintained at 9.5 for oxygen and 2 for hydrogen. After the 0.5 Amp treatment the current was increased to 1.0 Amp for 15 minutes while the stoichiometric flow was maintained at 9.5 for oxygen and 2 for hydrogen. After the 1.0 Amp treatment the current was increased to 1.5 Amp for 15 minutes while the stoichiometric flow was maintained at 9.5 for oxygen and 2 for hydrogen. After the 1.5 Amps treatment the current was increased to 4 Amp for 15

105 minutes while the stoichiometric flow was maintained at 9.5 for oxygen and 2 for hydrogen. After the 4 Amp treatment the current was increased to 4.5 Amp for 15 minutes while the stoichiometric flow was maintained at 9.5 for oxygen and 2 for hydrogen. After the 4.5 Amp treatment the current was increased to 5 Amp for 15 minutes while the stoichiometric flow was maintained at 9.5 for oxygen and 2 for hydrogen. After the 5 Amp treatment the current was increased to 5.5 Amp for 15 minutes while the stoichiometric flow was maintained at 9.5 for oxygen and 2 for hydrogen. After the 5.5 Amp treatment the current was increased to 6.5 Amp for 15 minutes while the stoichiometric flow was maintained at 9.5 for oxygen and 2 for hydrogen. The potential of the last 5 minutes of each treatment was averaged to generate a single data point at that current. At the end each chronoamperometric hold an impedance measurement at 1000 Hz was performed to determine the resistance of the fuel cell fixture, shown in Figure 27. The impedance technique works by passing an alternating current signal through the fuel cell at steady state and measuring the phasor quality of the voltage output. A small-signal harmonic current i(t)=i0cos(t) was added to the direct current, i, by the summing amplifier. The steady state current at which the fuel cell is operating is represented by i0. Variable time was represented by t. In these polymer electrolyte membrane fuel cell experiments the current source was operated in galvanostatic mode, electronically controlling the applied current that passes through the fuel cell. The alternating current causes a response voltage signal u(t)= u0() cos(t+()) which was measured by the amplifier with filter circuits that eliminate the

106 direct current components of the signal. The phase shift response caused by impedance

Figure 27. Static I/V characteristics with impedance perturbation and signal around an operating current of 0.5 A/cm2 on a basic polymer electrolyte membrane fuel cell polarization curve with a small sinusoidal perturbation tested was represented by . The response voltage of chronovoltammetric i0 was u0. The u0 voltage and phase shift, , are functions of the response frequency, , which are used to determine the total resistance of the fuel cell fixture at determined potential. If the potential of the fuel cell fixture went below 0.4 V then the chronoamperometric fuel cell test was terminated. After the last chronoamperometric hold the fuel cell fixture was depressurized then purged with high purity nitrogen gas and allowed to cool to room temperature.

107

A negative potential scan from open cell potential can also be used to generate a transient polarization performance curve. The MEA was first installed in the fuel cell fixture and leak checked then conditioned at the fuel cell at operating condition at low potential to humidify the Nafion and create proton conduction channels within the MEA.

The leak checked fuel cell fixture was slowly heated to 80 ºC in high purity nitrogen gas while the anode and cathode where simultaneously pressurized to 21.75 psig. The inlet gas lines were heated to 75 ºC and the anode and cathode deionized and distilled water bubbling humidifiers were heated to 70 ºC. Once all zones reached their set point temperatures high purity nitrogen gas was allowed to flow through the bubbling humidifiers. Once the temperature zones restabilized after switching flow through on the humidifiers the anode nitrogen flow was changed to high purity hydrogen and cathode nitrogen flow was changed to high purity oxygen flow. The open cell potential was allowed to stabilize for 15 to 40 min. The fuel cell was then chronopotentiometrically, held at a fixed potential while the current is measured, conditioned at 0.2 to 0.4 V until the current stabilized in 4 to 48 hrs. After the fuel cell was conditioned it was allowed to stabilize at open cell potential for 15 to 40 minutes. The fuel cell potential was then scanned at -0.0005 V from the open cell potential until 0.0 V was reached while the current was recorded. After the voltage scan was completed the fuel cell fixture was depressurized and purged with high purity nitrogen gas then allowed to cool to room temperature.

108

4.6 Transmission Electron Microscopy

An FEI Tecnai F20 XT Transmission Electron Microscope (TEM) was operated at 200 kV to image CNx catalyst. The material was prepared for imaging by sonicating a suspension of CNx catalyst in ethanol, which was then deposited drop wise on a lacey- formvar carbon supported on a 200 mesh copper TEM grid. The solution was allowed to dry on the TEM grid. The grid was added to the Tecnai TEM sample load chamber and imaged. The TEM images species shown were representative of the total species found.

4.7 X-ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy (XPS) was used to examine the binding energies and local environments of surface oxygen, nitrogen, carbon, sulfur, phosphorus, magnesium, and iron when applicable of CNx catalysts. The CNx catalyst samples were pressed in a stainless steel cup with a fired glass tip that made them loosely pelletized.

These samples were then hand pressed with tool and die so that they would not be fluidized during evacuation. A Kratos Ultra Axis Spectrometer with a monochromated aluminum anode source was used to collect a survey spectrum, O 1s, N 1s, C 1s, S 2p, P

2p, Mg 2p and Fe 2p regions for each CNx catalyst sample when applicable. The survey spectrum was used to determine what species were of interest on the CNx surface.

Analysis of the collected data was performed using XPSPeak 4.1® with a Shirley 109 background baseline correction and a combination of Lorentzian-Gaussian peaks for curve fitting. For quantitative comparison of the surface composition atomic sensitivity factors that for the specific Kratos Ultra Axis Spectrometer with a monochromatic aluminum anode source used were used to calibrate the relative counts for each elemental species. The relative XPS peak areas were determined by taking into account the relative sensitivity factors for the XPS instrument for each species.

4.8 Raman Spectroscopy

Raman spectra were acquired with Horiba Jobin Yvon labRam 300 confocal

Raman, which employed a helium-ion laser (633.81 nm). Obtaining a clear image of the powder surface with a white light source was used to focus the beam. The spectra were calibrated with naphthalene. The spectra were obtained at 298 K in air by averaging 20 scans each with a 30 sec exposure time. Spectra were collected without a filter. The laser power was held constant at 1 mW during spectra collection.

4.9 X-ray Absorption Spectroscopy

X-ray absorption spectroscopy techniques were used to determine the local bonding environment of metal phase within CNx catalysts. The deviations in the local 110 bonding environment were studied by observing changes in the Fe 1s binding energies, which correspond to the K absorption edge of iron. X-ray absorption is a bulk technique that can be turned to a single element as long as the absorption edge is with the limits of the experimental apparatus used.

4.9.1 Pellet Preparation

The samples were palletized and placed in the path of the X-ray beam for X-ray absorption analysis. The pellet composition and thickness for X-ray absorption was calculated with a Henke analysis using XAFSmass software using the binding edge of interest (Fe K-edge 7112 eV and Co K-edge 7709 eV). The optimal pellet thickness was dependant on the X-ray absorption of the sample of interest. Typically the absorption input, d, was between 2 and 3 to accomplish a 17.5 % X-ray absorption level for the first detection chamber and a 50 % X-ray absorption level for the second chamber. The difference in X-ray detection for the chamber before the sample and X-ray detection chamber after the sample were used to determine the X-ray absorption of the sample of interest. Pellet dies with ground power materials were pressed at approximately 100

Newtons for 30 - 40 seconds to form pellets. Ex situ tested pelletized were mixed with situ boron nitride (Sigma Aldrich) and nanoporous silicon dioxide (Sigma Aldrich) to form rigid pellets.

111

4.9.2 X-ray Absorption Spectroscopy Ex-situ Collection

The X-ray absorption fine-structure spectroscopy (XAFS) measurements were made on the bending magnet beam line of the Dow-Northwestern-Dupont Collaborative

Access Team (DND-CAT, Sector 5) at the Advanced Photon Source in Argonne National

Laboratory. Extended X-ray near Edge spectra and X-ray absorption near-edge analyses of the iron carbide sample used for comparison were collected at the Materials Research

Collaborative Access Team (MRCAT, Sector 10) beam line at the Advanced Photon

Source, Argonne National Laboratory. Measurements were made in the transmission mode with ionization chambers optimized for maximum current with linear response.

Standard procedures were used to extract the XAFS data from the absorption spectra using WINXAS97 software.270 Samples were ground and cast into self-supporting pellets. The XAFS data were collected at room temperature in air.

4.10 Physisorption Techniques

Nitrogen physisorption was performed on the catalyst materials to determine the

Brunauer, Emmett and Teller surface area and Barrett, Joyner and Halenda pore size distribution and cumulative desorption pore volume. Catalytic reactions take place on the catalyst surface so surface area corresponds to the quantity of catalyst active sites. For

112

CNx electrocatalytic materials an increase in surface area could correspond to a decrease in the electrical connectivity of the material leading to an increase in electrical resistance.

A distribution of the pore volume also can be used in understanding of the transport of reactants and water to and from the catalyst during fuel cell operation. A Micromeritics

ASAP 2020 was used for both of these analyses.

First an empty sample tube was evacuated, filled with nitrogen, and then massed.

Next, a quartz analysis tube was filled with 70 – 150 mg of CNx catalyst. Catalyst that statically clung to the sidewalls of the analysis tube was removed with a clean cue tip.

The amount of material varied with surface area of the material in accordance with the sensitivity of the instrument. The tube with catalyst was filled with nitrogen and weighed to ensure that the degas treatment did not physically alter the catalyst material. The sample was evacuated and then heated to 130 °C for 24 hrs under high vacuum to degas the sample. The degas step was necessary to remove physically adsorbed water from the microsporous catalyst material which would affect the calculation of Brunauer, Emmett and Teller surface area and Barrett, Joyner, and Halenda Pore size distribution.

4.10.1 Brunauer-Emmett-Teller Surface Area Method

Brunauer, Emmett and Teller surface area was computed by nitrogen physisorption on degassed high surface area catalysts using a Micromeritics ASAP 2020.

After degassing, the sample was backfilled with nitrogen and massed. The sample weight 113 after degassing was used in the surface area and pore volume analysis. The Brunauer,

Emmett and Teller surface area analyses were performed in a liquid nitrogen bath.

4.10.2 Barrett, Joyner, and Halenda Pore Size Distribution

After degassing the adsorption and desorption of nitrogen gas was used to determine the Barrett, Joyner, and Halenda Pore size distribution of the catalysts studied.

The experimental data for Brunauer, Emmett and Teller surface area and Barrett, Joyner, and Halenda Pore size distribution analyses were simultaneously obtained during the same sample treatment. The cumulative pore volume was calculated from the nitrogen desorption.

4.11 Computational Analysis

Electron donation was hypothesized as a critical step in the reduction of dioxygen to water over carbon-nitrogen non-precious metal catalysts. Unfortunately the computational resources needed to investigate the plethora of proposed active sites that could be responsible for oxygen reduction over non-precious metal catalysts was not currently not feasible. To determine the electron donation of carbon-nitrogen graphite materials the vertical and adiabatic ionization potentials of smaller saturated polycyclic 114 aromatic hydrocarbons with and without nitrogen that were thought to be chemically similar to non-precious metal catalysts were computed.

To justify the use of the Becke (three-parameter) Lee-Yang-Parr exchange- correlation functional (B3LYP) computational method for large graphitic carbons, experimental values of small polycyclic molecules were evaluated to validate the use of the B3LYP method. The geometry of each molecule was roughly optimized at the Austin

Model 1 theory. All molecules explored were studied with highest level of symmetry available for each species, the morphology of all compounds were planar as a result. For closed-shell calibration, a single gas-phase Reichard‟s catalyst geometry optimization, frequency analysis, and full orbital computation at the B3LYP/6-31G* level of theory took approximately 50 hours on 4 parallel processors utilizing the Ohio Supercomputer

Center. For open-shell calibration, a single gas-phase quaterrylene geometry optimization, frequency analysis, and full orbital computation at the B3LYP/6-31G* level of theory took approximately 59 hours on 4 parallel processors, utilizing the Ohio

Supercomputer center. To obtain both the adiabatic and vertical ionization potentials at least two additional computations were completed. To determine the vertical ionization potential a one closed-shell and to obtain the adiabatic ionization potential one open-shell computation were performed for each molecule of interest at the neutral electronic state and at the ionized state. These computations were fist completed at Austin Model 1 lower level of theory to explore the energetic difference in vertical and adiabatic ionization energies. Some structures had imaginary frequency vibrations and required that the starting geometry be re-optimized to the lowest energy state to ensure that the

115 optimized geometries were global minima so that properties of these molecules translate to what are the most likely states present in actual structures. The experimental Ionization

Potentials (IPs) of small polycyclic aromatic hydrocarbons (PAHs) were used to compare computational adiabatic and vertical IPs obtained without a rigorous frequency analysis for comparison. Approximately, ten-percent of the molecules studied had an imaginary vibrational state and were re-optimized to a stable state.

4.12 Temperature Programmed Oxidation Experiments

Temperature programmed oxidation (TPOs) experiments were performed using a

Setaram 111 TGA/DSC (theromogravimeteric analyzer/differential scanning calorimeter) instrument coupled to a MKS Cirrus residual gas analyzer to monitor the oxidation temperature and composition of the CNx catalysts. For the TPO experiments, approximately 5 mg of CNx was loaded into the platinum cup of the Setaram 111

TGA/DSC. The temperature was ramped at 5°C/min to 750°C while flowing 10%

O2/He. Product gas mass signals from 1-100 were monitored with a MKS Cirrus mass spectrometer.

4.13 Temperature Programmed Desorption-TPO Experiments

116

Temperature programmed desorption (TPD) – temperature programmed oxidation

(TPO) coupled experiments were performed using a Micromeritics Autochem II system coupled to a MKS Cirrus residual gas analyzer to study the CNx catalyst surface and bulk species, as well as, the stability of the catalyst species. TPD-TPO experiments were performed with approximately 25 mg of catalyst. A helium gas stream was flowed over the catalyst bed while heating at 5 °C/min to 900 °C and cooling using forced convection back to room temperature for the TPD portion of the experiment. Following the TPD, a

TPO was immediately ran in 5 %O2/He at 5 °C/min to 750 °C to characterize the catalyst material remaining after the TPD. The product gas mass signals of the entire experiment were monitored from 1-100.

To study the effect of the presence of water on the catalyst, 10 mol% water vapor was introduced into the TPD temperature ramp from room temperature to 900 °C. The catalyst bed was cooled in He and then a TPO was performed as described above for the dry TPD-TPO experiment. The experiments involving water during the TPD are named wet TPD-TPO experiments.

117

CHAPTER 5. Phosphorus Doped CNx Oxygen Reduction Catalyst

(This chapter is partially adapted from 271)

5.2 Phosphorus Doping Introduction

Phosphorus was added to the growth media of nitrogen-containing carbon nanostructures and found to dramatically improve the electrocatalytic activity for oxygen reduction in acidic electrolyte. Phosphorus doping was achieved by growing carbon- nitrogen-phosphorus catalyst (CNxPy) over triphenylphospine- and iron acetate- impregnated high surface area magnesia support in nitrogen saturated with CH3CN atmosphere, at 900 ºC. Catalysts grown in phosphorus-containing media displayed an improved onset of oxygen reduction activity, increased current density and higher selectivity for water formation. The incorporation of both phosphorus and nitrogen into graphite materials allows for the tailoring of the physical and electrochemical properties.

118

5.3 Phosphorus Doping Motivation

The development of non-Pt metal catalysts for PEM fuel cells is motivated by the high platinum loading needed to accelerate the sluggish ORR. The inherent expense and scarcity of platinum limit the applicability of PEM fuel cell technologies. Conversely, carbon-based ORR catalysts are inexpensive and elementally abundant, but the nature of the electrocatalytic activity is not well understood, limiting their development.1

The inclusion of heteroatoms atoms (e.g., N, O, P) into the hexagonal networks of graphitic carbons modifies the electronic and chemical properties due to variations in the electronic structure and nanostructure growth.272 Nitrogen is the most commonly doped heteroatom in graphite structures.273 The electronic band gap of nitrogen-doped carbon nanofibers is known to be narrowed by metal inclusion in the nanostructure.273 The narrower band gap allows for the more facile donation of electrons facilitating electron donation reactions on the surface. Nitrogen doping is also known to decrease the

14 mechanical strength of graphine nanostructures such at CNx. Furthermore, nitrogen doping in single-walled carbon nanotubes has been shown to renormalize the electron and photon states by causing non-localized band gap effects.13 The C/N ratio in carbon nanostructures is a function of temperature,199 growth metal,134,199 and reactants.199 Both phosphorus and nitrogen are n-type electron donators that add electron density to the graphite structure.13 Phosphorus doped carbon materials have increased electrical and electrochemical behavior due to the facilitated interfacial electron behavior at moderate phosphorus loadings.15 It is a natural extension that the inclusion of nitrogen and 119 phosphorus heteroatoms into graphitic ORR electrocatalysts would alter the electrocatalytic properties graphitic carbon nanostructures formed.

Impurities of phosphorus and sulfur during carbon fiber growth are known to affect yield of carbonaceous material.12 These heteroatom impurities are thought to alter the temperature at which the metal growth catalyst becomes sufficiently ductile to facilitate carbon fiber growth, known as the eutectic point. The presence of sulfur during the metal-catalyzed pyrolyic growth of carbon is known to increase the amount of carbon fibers grown at a given temperature.11,274-276 The observed increase in structures is thought to be due to the lowered eutectic point of iron to approximately 988 C, thus allowing the iron to be more ductile and therefore facilitate easier carbon growth.12

Although, the oxygen reduction activity and nanostructure was found to be unchanged

11 when sulfur was used to grow CNx materials. Conversely, phosphorus doping was shown to alter the physical properties of the carbon by modifying the growth of carbon nanostructures.12 Low phosphorus levels are thought to lower the eutectic point of transition metals used to grow carbon nanostructures during pyrolysis, promoting carbon yield in a similar manner as sulfur.12,272 For instance, the eutectic temperature of iron with

16.9% phosphorus is 1048 C.12 Phosphorus doping levels for optimal carbon growth are thought to be P/Fe≤1 molar ratio of phosphorus to transition metal.12,272 Alloys and metal phosphides have been used to tailor the growth of the carbon nanostructures. Cruz-Silva et al. showed that the increasing the phosphorus content less than 3.5 % decreased the length and yield of nitrogen-doped, multiwalled carbon nanotubes.272 Furthermore,

Jourdain et al. used a Ni/Fe/P or Fe/Co/P carbon growth catalyst with less than 1.5 at %

120 phosphorus to form carbon nanotubes with periodically encapsulated metal nanoparticales along the length of the nanotubes.229 The combination of alloy and phosphorus dopant was optimized to alter the morphology of the carbon nanostructures.

During vapor deposition carbon nanofiber growth phosphorus reactants at higher than 1

P/Fe atomic ratios results in a become stiff and brittle carbon material.12

Although the incorporation of phosphorus into a carbon-nitrogen nanostructure lattice has previously been reported,272 the catalytic ORR activity of these structures was not studied. Phosphorus, like nitrogen, can lead to an increase in electron donation in graphitic carbon catalysts. Earlier modeling studies by Strelko et al. have predicted an increased ORR activity when graphites are doped with Phosphorus.177 The computational study by Strelko et al. was based on the assumption that electron donation was the rate determining step in dioxygen reduction, so as a result atoms with higher electron density, such as phosphorus, that replace carbon atoms in the graphite matrix increase the oxygen reduction activity. Increases in activity are to be more pronounced when multiple heteroatoms are doped into the graphite.

121

5.4 CNxPy Grown Over Iron Acetate and Triphenylphosphine

5.4.1 CNxPy Grown Over Iron Acetate and Triphenylphosphine Synthesis

In this study, CNxPy catalysts were synthesized by pyrolyzing acetonitrile over high surface area MgO (Aldrich) incipient-wetness-impregnated with 2wt.% Fe following

73 a procedure previously reported for phosphorus-free CNx. When doped with phosphorus, the support went through an additional impregnation step with triphenylphosphine in diethyl ether. The resulting growth medium was heated in N2 (g) to 900 C, and then exposed to acetonitrile for 2 hr. The resulting catalyst was treated in

1M HCl (aq) for 1hr at 60 C to remove exposed magnesia and metal, then rinsed with

DI-H2O. Upon drying, the resulting graphitic carbon-nitrogen ORR catalyst is denoted as

“CNx” if it is phosphorus-free and “CNxPy” if the synthesis involved a triphenylphosphine

IWI.

5.4.2 Electrochemical Oxygen Reduction Activity Testing

ORR performance was determined using cyclic voltammagrams (CVs) in an O2-

11 saturated half-cell of 0.5 M H2SO4 on the RRDE, as described before. RRDE testing showed a dramatic increase in activity for phosphorus loadings lower than P/Fe = 20 in

CNxPy ORR catalysts, Figure 28. All RRDE testing was performed in 0.5 M H2SO4 at

122

1000 rpm saturated with oxygen at room temperature with 10 mV/s potential sweeps. The

catalyst loading was 426 g/cm2.

1

:

3

0

Figure 28. Cathodic scans of CNxPy catalyst-coated RRDE in Oxygen Saturated Electrolyte

The P/Fe ratios reported correspond to those used in the growth media. The greatest

activity improvements were observed when P/Fe ratios were less than 3. To isolate the

effect of the second IWI step used to impregnate triethylphosphine into the growth media,

a CNx catalyst with a second impregnation of diethyl ether without triphenylphosphine

was prepared, P/Fe = 0 blank, Fig. 1 a). The P/Fe = 0 blank CNx catalyst had less activity

than CNx grown on P-free media which confirmed that triphenylphosphine was altering

123 the ORR activity, not diethyl ether. ORR selectivity was determined by measuring the hydrogen peroxide produced during ORR activity testing. H2O2 formation is undesired

- + because it can corrode fuel cell components and only consumes 2e and 2H per O2 molecule (as opposed to 4 in H2O formation), leading to a net power loss. CNxPy grown with a molar ratio of P/Fe ≤ 1 had similar, if not improved selectivity, Figure 29. An increase in selectivity would correspond

Figure 29. CNxPy Selectivity measured during activity testing. Legend: molar ratio of P/Fe in the growth media to enhanced

124 activity, however, the observed activity increases cannot be explained by the selectivity differences alone.

It can be seen that there is a considerable difference in the Faradaic capacitance catalysts which can be obtained in the argon saturated electrolyte when oxygen reduction is not present, Figure 30. The magnitude of the Faradaic capacitance is

Figure 30. Cyclic Voltammograms of CNx and CNxPy RRDE in Argon Saturated Electrolyte proportional to the electroactive area of the non-precious metal catalyst.277 Since all catalysts were applied to the electrode with the same loading by weight the differences in capacitance suggest that the mass density of the materials is greatly altered by

125 phosphorus doping. Those catalysts with the highest capacitance correspond to the catalysts with the highest oxygen reduction currents and takeoff potentials. The correlation of oxygen reduction activity with Faradaic capacitance has also been observed in other publications.44 This finding is also supported by the hypothesis that the non- precious metal catalyst active site resides within the microporous regime of the nitrogen- doped carbon catalyst.112

Full fuel cell testing was used to confirm the relative trend in activity observed in the simulated cathode half-cell testing. The membrane electrode assemblies comprised;

2 + 2 060 carbon Toray paper, a 1 mgPt/cm anode, H Nafion 212 membrane, a 1 mgCNx/cm cathode, and 060 carbon Toray paper. The assemblies were fabricated using the heated, vacuum table. Before fuel cell testing each assembly was leak and crossover checked in nitrogen gas at room temperature. The leak checked fuel cell fixture was slowly heated to

80 ºC in high purity nitrogen gas while the anode and cathode where simultaneously pressurized to 21.75 psig. The inlet gas lines were heated to 75 ºC and the anode and cathode deionized and distilled water bubbling humidifiers were heated to 70 ºC. Once all zones reached their set point temperatures high purity nitrogen gas was allowed to flow through the bubbling humidifiers. Once the temperature zones re-stabilized after switching gas flow through on the humidifiers the anode nitrogen flow was changed to high purity hydrogen and cathode nitrogen flow was changed to high purity oxygen flow.

The open cell potential was allowed to stabilize for 15 to 40 min. The fuel cell was then chronopotentiometrically held at 0.3 V while the current was measured it stabilized after

4 to 48 hrs. After the fuel cell was conditioned it was allowed to stabilize at open cell

126 potential for 15 to 40 minutes. The fuel cell potential was then scanned at -0.0005 V from the open cell potential until 0.0 V was reached while the current was recorded. After the voltage scan was completed the fuel cell fixture was depressurized and purged with high purity nitrogen gas then allowed to cool to room temperature. Full fuel cell testing had the same activity trend as what was observed in half-cell testing, Figure 31. The differences in activity were much closer in

Figure 31. Full Fuel Cell Testing of CNx and CNxPy with Oxygen and Hydrogen Reactants full fuel cell testing than in half-cell testing, which could be due to water management in the microporus regime. It is important to note that the membrane electrode assembly fabrication is not optimized for the CNxPy catalysts tested.

127

5.4.3 Transmission Electron Microscopy Imaging of CNx and CNxPy Catalysts

The transmission electron microscopy images for P-doped and P-free catalysts displayed an increasing degree of disorder in the carbon nanofibers with increasing phosphorus doping, Figure 32. Phosphorus doping is known to promote large corrugation in graphitic structures due to the positive curvature of incorporated phosphorus when tetragonally bonded.14 This is likely due to the larger relative size of phosphorus atoms as compared to graphitic carbon and nitrogen atoms, which causes more strain on the hexagonal carbon framework.13 Phosphorus prefers a more pyramidal bonding structure than nitrogen when bonded in graphitic carbons.278 Phosphorus doped into nitrogen

128 containing carbon nanotubes is also thought to widen the tube ends causing a more non-

P/Fe=0

P/Fe=20

P/Fe=0.33 P/Fe=30

Figure 32. TEM Images of CNx and CNxPy Carbon Nanofibers uniform nanotube diameter.14 The carbon fibers with less phosphorus doping have a much more defined and regular diameter. The morphology as phosphorus doping increases goes from a stacked cup or bamboo like structure to more of a broken-walled nanotube structure. The prevalence of nanofibers was considerably less at the higher phosphorus loadings. The morphology of the other carbon structures was also affected by

129 higher phosphorus loadings, Figure 33. The disorder in the structures likely arises from the increased instability in the carbon nanostructures with increased

P/Fe=0 P/Fe=20

P/Fe=0.3 P/Fe=30

3

Figure 33. TEM Images of CNx and CNxPy Carbon Nanostructures phosphorus loadings. These hallow cube-like structures in Figure 33 were formed by templated carbon growth over the cubic magnesia support that was subsequently 130 dissolved during the acid washing step after catalyst growth.73,192 The iron used to catalyze the growth of these nanostructures was exclusively found encased within several layers of graphitic carbon, which is consistent with other reports.193,279 The phosphorus doping appears to disrupt the stability of these cubic carbon structures and causes them to be much rounder and non-uniform.

5.4.4 Raman Spectroscopy Characterization of CNx and CNxPy catalysts

Raman D- and G-band intensities were used to determine the extent of defects in the nanostructures. Individually nitrogen133 and phosphorus13 incorporation into graphite are known to cause a substantial increase in the relative D-band intensity. The Raman

Spectra of CNx and CNxPy catalysts, normalized by G-band intensity, was acquired by a

Horiba Jobin Yvon LabRam 300 confocal Raman spectrometer, using an argon-ion laser

(632.81 nm). The spectra were obtained at 298 K by averaging 20 scans each with a 30 s exposure time. The power has held constant at 6 mW. A rise in disorder would likely correspond to an increase in edge plane

131

Figure 34. Raman Spectra of CNx and CNxPy catalysts, normalized by G-band intensity

exposure, or heteroatom incorporation. CNxPy catalysts did not show much change in the relative intensities of the D and G bands except at very high P/Fe ratios, where they exhibited a slight decrease in the D-band Figure 34, suggesting that the disorder of the of the carbon nanostructures decrease or heteroatom incorporation into the graphitic matrix increased. This could be due to less heteroatom doping with larger amounts of phosphorus present during catalyst growth or an increase in the relative amount of graphitic carbon in the catalyst that is formed.

132

5.4.5 X-ray Photoelectron Spectroscopy of CNx and CNxPy catalysts

The surface composition of CNx and CNxPy was determined by X-ray photoelectron spectroscopy. The binding energies and environments of oxygen, nitrogen, phosphorus, and carbon were characterized. A Kratos Ultra Axis Spectrometer with a monochromated aluminum anode source was used to collect a survey spectrum, O 1s, N

1s, C 1s, and P 2p regions for each CNx or CNxPy catalyst sample. The survey spectrum was used to determine if additional species such as iron or magnesia were present.

Analysis of the collected data was done on XPSPeak 4.1® using a Shirley background for baseline correction and a combination of Lorentzian-Gaussian peaks for curve fitting. For quantitative comparison of the surface composition CNx catalysts that underwent different treatments, the atomic sensitivity factors were used to correct the XPS peak areas for the relative sensitivity for the XPS instrument used. The incorporation of nitrogen atoms into the carbon hexagonal network is known to impart ORR activity through a debated mechism.188-190 It has also been hypothesized that an increase in oxygen functional groups within graphitic carbons could increase ORR activity through electron donation.172 To determine the extent that surface functional groups affect the oxygen reduction activity of the CNxPy catalysts X-ray photoelectron spectroscopy was used to characterize the surface functional groups, Table 2. The nitrogen and oxygen surface concentrations vary across the various phosphorus loadings there are no clear trends that correspond with ORR activity. It was noted that carbon materials enriched with nitrogen have a higher capacitance than nitrogen-free carbon materials,280 but the

133 nitrogen content does not trend with the observed capacitance observed in the argon saturated electrolyte cyclic voltammograms in Figure 30. Since the nitrogen content did not trend with the observed capacitance the differences in activity likely attributed to textural and morphological changes in the carbon nanostructures.

Table 2. XPS surface concentration of CNx and CNxPy catalysts

C 1s (%) N 1s (%) O 1s (%) CNxPy 200.1 eV 398.8 eV 401.2 eV 402.0 eV 532.9 eV P/Fe=0 89.0 1.5 3.8 2.7 3.0 P/Fe=0.33 96.0 0.4 0.8 0.6 2.2 P/Fe=1 86.6 1.2 6.6 1.7 3.9 P/Fe=3 89.3 0.6 4.0 2.8 3.4 P/Fe=30 90.4 1.2 2.5 3.8 2.0

31 5.4.6 P Nuclear Magnetic Resonance of CNx and CNxPy

Solid-state 31P nuclear magnetic resonance (NMR) was used in an attempt to quantify the amount of phosphorus in the formed CNxPy nanostructures. Nearly 100 % of all phosphorus is in the NMR active state. 31P NMR is known to work well on activated amorphous carbons.281 This could be due to the physical absorption of phosphorus on the activated carbon surface281 or facile bonding environment for phosphorus on amorphous

15,282 73 carbons. Unfortunately CNx catalyst materials are highly graphitic and as a result

134 are unlikely to have high levels of phosphorus incorporated. Since there was also a fair

31 amount of ferromagnetic iron, 1-5 wt.%, in the CNxPy catalyst samples no P NMR spectra were able to be resolved.

5.4.7 Temperature Programmed Oxidation of CNx and CNxPy

To determine the stability and relative bulk composition of CNx and CNxPy catalysts a temperature-programmed oxidation was performed while measuring the exhaust gas with a mass spectrometer. Temperature programmed oxidations of CNxPy catalysts were performed in a Setaram TGA/DSC 111 with 30 ccm of 10%O2/He flow. A mass of ~6 mg of CNxPy catalyst was placed in Pt crucible with an empty Pt crucible counterbalance on the reference. The temperature was ramped from 20 to 750 ºC at a rate of 5 ºC/min under 10 %O2/He then held at 750 ºC for 5 min before ramping back down to

20ºC. A Cirrus residual gas analyzer was used to follow the gas species evolved. When the CNxPy catalysts were subjected to temperature-programmed oxidation, the onset of carbon oxidation (indicated by mass number 44, attributed to CO2) was similar for all catalysts studied, Figure 35. However, the majority of CO2 evolution moved to higher temperatures at higher phosphorus doping levels suggesting an increase in stability. Cruz-

Silva et al. also reported a decrease in the ease of oxidation when phosphorus/nitrogen doped multiwalled carbon nanotubes contain less than 0.5 atom % phosphorus.272 Both

CNx and CNxPy catalysts were seen to contain significant levels of nitrogen incorporated

135 into the graphite indicated by m/z= 30 evolving with CO2 in Figure 35. The nitrogen signal contribution was from the bulk and surface with the

Figure 35. Temperature Programmed Oxidation of CNxPy catalysts in 10%O2/He m/z=30 signal at lower temperatures being more from the surface. The nitrogen signals during temperature-programmed oxidation starkly contrast the quantities of surface nitrogen detected during X-ray photoelectron spectroscopy. It is not clear why this is the case, but surface functionalities are thought to contribute more strongly to catalytic properties than bulk species. Unfortunately mass numbers associated with phosphorus were below the detection limit for all catalysts studied. One possible explanation for lack of phosphorus signal during temperature programming oxidation is that the phosphorus 136 present was in an alloy with the iron that remained in the platinum crucible after the treatment.

5.4.8 Extended X-ray Absorption Fine Structure of CNx and CNxPy

To determine the local bonding environment of the iron phase that resides within

CNx and CNxPy catalysts X-ray absorption fine-structure spectroscopy (XAFS) measurements were made on the bending magnet beam line of the Dow-Northwestern-

Dupont Collaborative Access Team (DND-CAT, Sector 5) at the Advanced Photon

Source, Argonne National Laboratory. The extended X-ray absorption fine structure of the iron carbide sample used for comparison was collected at The Materials Research

Collaborative Access Team (MRCAT, Sector 10) at the Advanced Photon Source,

Argonne National Laboratory. All measurements were made in the transmission mode with ionization chambers optimized for maximum current with linear response. Standard procedures were used to extract the XAFS data from the absorption spectra using

WINXAS97 software.270 Samples were ground and mixed with small amounts of boron nitride then pressed into self-supporting pellets. The XAFS data were collected at room temperature in air.

Since transition metal must be present during pyrolysis to form an active non- precious metal carbon-based ORR catalyst, it is difficult to determine the role that plays iron in ORR activity. Many researchers believe that the oxygen reduction active site on 137 nitrogen-doped carbons contains a coordinated metal.1,55,118,178-184 Another group researchers believe that nitrogen functionalized heat-treated carbons have inherent oxygen reduction activity that does not necessary contain a transition metal.201,256-261 X- ray absorption spectroscopy was used to probe the local bonding environment of the iron in CNx oxygen reduction catalysts. The Fourier transform obtained at the Fe K edge and k2 weighted EXAFS displayed a similar local bonding environment to that of an iron carbide, Figure 36. None of the CNx samples studied resembled

Figure 36. EXAFS comparison of Fe K-edge in CNx Fe:0.33P Verses Standard Materials the iron phthalocyanine standard, which was similar to the nitrogen-bonded transition

1,180,283 metal active site proposed by some researchers. The iron contained within CNx

138

Fe:0.33P has a small metallic phase at roughly 2 Å and an iron phase bonded to a lighter element shoulder at approximately 1.5 Å that could be due to nitrogen, oxygen, phosphorus and/or carbon. It is important to note that these values are not absolute due to a work function inherent with the instrument used for X-ray detection. It can be seen that the CNx Fe:0.33P Fe-Fe bond length is less than that of bulk iron, Figure 36. This is likely due iron particles being much smaller in CNx catalyst than of the bulk iron foil.

The iron-iron bonding shoulder at ~1.5 Å in CNx catalyst coincides with Fe-N bond distance in iron phthalocyanine, but also aligns with the Fe-C bond distance in iron carbide. Due to the similar scattering spectra of periodic table 2p elements, it cannot be determined if the CNx peak at ~1.5 Å is due to an iron nitrogen, carbon, or oxygen bond.

However, it does not appear that this shoulder is a due to a Fe-P phase. The EXAFS for iron carbide displayed the same peak structure and locations as the CNx catalysts studied.

This is in agreement with an investigation of the CNx iron phase through Mössbauer spectroscopy which revealed metallic iron (a paramagnetic -Fe phase, which would be

279 unstable in air) and some cementite phase (Fe3C) in CNx. Jacobs et al. developed an

EXAFS theoretical structural model was by using ab initio multiple scattering calculations of X-ray absorption fine structure for cementite (Fe3C), Hägg carbide

(Fe5C2), -carbide (Fe3C), and -carbide (Fe2C) to generate the Fourier transform magnitude versus bond distance for the iron K-edge in order to characterize the iron

284 carbides present in Fischer-Tropsch synthesis catalysts. CNx oxygen reduction electrocatalyst most closely resembles a combination of the -carbide (Fe3C) and Hägg carbide (Fe5C2) when compared to the simulated carbide EXAFS spectra. A comparison

139 of the EXAFS spectra over various phosphorus doping showed that phosphorus had a profound effect on the iron phase, Figure 37. The most obvious difference is that the Fe-

Figure 37. EXAFS comparison of Fe K-edge in CNx and CNx P-doped materials

Fe bond distance is shifted to lower values with increasing phosphorus doping. This suggests that the size of the iron nanostructures within the carbon layers become smaller with phosphorus doping, but a critical Fe-Fe bond distance is reached around 1.6 Å. The bonding distance of the lighter element Fe K-edge shoulder decreases with phosphorus doping, but in an irregular manner. This could be due to the inclusion of multiple heteroatoms within the close proximity of iron such as phosphorus and carbon or the

140 extremely small size of the iron nanoparticles, which would cause non-regular scattering patterns. It was also shown that Fe3P and Fe2P readily dissolve carbon during pyrolysis at

272 850 °C, so it likely that mixed phosphide, carbide iron phase is present in CNxPy. It was also shown that the majority of the phosphorus that is incorporated into CNxPy nanostructures is located in the iron phase.272 These findings coupled with complicated crystallography of iron could partially explain the non-monotonic trend in the lighter element shoulder in the iron K-edge EXAFS with phosphorus loading.

5.5 Phosphorus Grown Graphites Synthesized with Different Reactant Materials

To elucidate the effective promotion of oxygen reduction activity with the inclusion of phosphorus during the growth of CNx nanostructured catalyst alternative synthesis techniques were explored. A variety of synthesis techniques have been used to form carbons doped with phosphorus. For instance, Cruz-Silva et al. pyrolyzed in a two- step spray of triphenylphosphine, ferrocene, and benzylamine at 720 – 840 °C in argon for 2 minutes.272 Maciel et al. used a pulse laser vaporation at 950 °C of ethanol, ferrocene, benzilamine, or triphenylphosphine to form P-doped graphites.13 Benissad-

Aissani et al. synthesized vapor grown carbon fibers by the catalytic decomposition of

CH4 and H2 with H3PO4 in ethanol over Fe3(CO)12 doped on graphite supports at 950 °C to 1150 °C for 30 min.12 Lui et al. incorporated phosphorus in tetrahedral amorphous carbons by using a filtered cathodic vacuum arc over carbons with phosphine gas flows.15

141

These synthesis techniques vary the source of phosphorus and the type of metal catalyst used during carbon growth. Since these various synthesis techniques incorporate phosphorus into graphites with and without nitrogen different other synthesis methods were studied to determine what role phosphorus has in imparting increase oxygen reduction activity.

5.5.1 Phosphorus-Doped CNx Grown From Iron Phosphate

To determine whether phosphorus needs to be in contact with the supported iron an iron phosphate hexahydrate was used as the carbon growth catalyst. In this study,

CNxPy-phosphate catalysts were synthesized by pyrolyzing acetonitrile over MgO

(Aldrich) incipient-wetness-impregnated (IWI) with 2wt% Fe with a combination of iron phosphate hexahydrate and iron acetate to vary the phosphorus doping. The resulting growth medium was heated in N2 (g) to 900 C, and then exposed to N2 saturated acetonitrile for 2 hr. The resulting catalyst was treated in 1 M HCl (aq) for 1hr at 60 C to remove exposed magnesia and metal, then rinsed with DI-H2O. Upon drying, the resulting graphitic carbon-nitrogen ORR catalyst is denoted as “CNxPy-phosphate”. For the sake of comparison the CNxPy referred to previously section will be referred to as

CNxPy-triphenylphosphine. This synthesis technique is very similar to how the CNxPy- triphenylphosphine catalyst that was previously described was grown, except that the source of phosphorus is different.

142

The electrochemical oxygen reduction activity was determined by CVs in an O2-

11 saturated 0.5M H2SO4 half-cell on the RRDE, as described elsewhere. All RRDE testing

was performed in 0.5 M H2SO4 at 1000 rpm saturated with oxygen at room temperature 1 with 10 mV/s potential sweeps. The catalyst loading was 426 g/cm2. The testing : conditions were identical to those used for the CN P -triphenylphosphine catalyst. The 3 x y oxygen reduction RRDE testing of CN P -phosphate showed a similar activity trend to 0 x y

that of CNxPy-triphenylphosphine, Figure 38. There was an optimal phosphorus doping

that maximizes the ORR activity of the CNxPy phosphate catalyst materials. Interesting,

the optimal phosphorus doping for CNxPy-triphenylphosphine is P/Fe = 0.33 and for

CNxPy-phosphate P/Fe = 1/4 = 0.25. The similarity in these values suggests that the same

mechanism imparts increased oxygen reduction activity to the graphitic nitrogen

containing structure when phosphorus in the growth media is present. This is somewhat

surprising because the oxidation state of phosphorus is different in the

triphenylphosphine and iron phosphate hexahydrate precursors, but it is likely that the

acetonitrile rich hydrocarbon environment during pyrolysis reduces the reactants and

catalyst to the lowest oxidation state. A low concentration of phosphorus in the iron

carbon growth media increases the oxygen reduction activity for both CNxPy-

triphenylphosphine and CNxPy-phosphate catalysts.

143

Figure 38. Cathodic scans of CNxPy-Phosphate catalyst-coated RRDE in Oxygen Saturated Electrolyte

5.5.2 Phosphorus-Doped CNx Grown From Iron Phthalocyanine and

Triphenylphosphine

To determine whether phosphorus has to be in contact with the iron on the support

an iron phthalocyanine was used as the carbon growth catalyst. Iron phthalocyanine has

an iron center that is coordinated by 4-surrounding nitrogen atoms. This would limit the

144

formation of an iron phosphide phase when growing CNxPy catalyst. In this study,

CNxPy-N/Fe-P-C catalysts were synthesized by pyrolyzing acetonitrile over Vulcan

carbon (Cabot) incipient-wetness-impregnated with 2 wt% Fe with iron phthalocyanine.

When doped with phosphorus, the support went through an additional IWI step with

triphenylphosphine in diethyl ether. The resulting growth medium was heated in N2 (g) to

900 C, and then exposed to N2 saturated with acetonitrile for 2 hr. The resulting catalyst

was treated in 1M HCl (aq) for 1hr at 60 C to remove exposed magnesia and metal, then

rinsed with DI-H2O. Upon drying, the resulting graphitic carbon-nitrogen ORR catalyst is

denoted as “CNxPy-N/Fe-P-C”.

The electrochemical oxygen reduction activity was determined by CVs in an O2-

11 saturated half-cell of 0.5 M H2SO4 on the RRDE, as described elsewhere. All RRDE

testing was performed in 0.5 M H2SO4 at 1000 rpm saturated with oxygen at room 1 temperature with 10 mV/s potential sweeps. The catalyst loading was 426 g/cm2. The : electrochemical testing is that same 3

0

145

Figure 39. Cathodic scans of CNxPy-N/Fe-P-C and P/Fe=0.33 CNxPy-triphenylphosphine catalyst-coated RRDE in Oxygen Saturated Electrolyte as all other RRDE testing performed in this chapter. When the iron access is partially occluded by the phthalocyanine macrocycle the phosphorus doping ORR activity trend is different, Figure 39. Phosphorus doping was found to reduce the oxygen reduction activity of the CNxPy catalyst material. It appears that if phosphorus is not in close proximity of iron in the growth media an opposite mechanism decreases the ORR activity of the resulting CNxPy-N/Fe-P-C catalyst material. This observation is not surprising considering that all phosphorus-doped carbon in the literature have the opportunity for phosphorus to be in close proximity with the metal growth catalyst.13,272,278,281,282,285 The

CNxPy-N/Fe-P-C catalysts are categorically less active than the most active CNxPy- triphenylphosphine catalyst. 146

5.5.3 Phosphorus-Doped CPy Grown From Triphenylphosphine on Vulcan Carbon

To determine whether phosphorus had its own inherent oxygen reduction activity

when doped into a graphitic carbon, CPy catalysts where synthesized by a reactant

nitrogen-free pyrolysis. In this study, CPy catalysts were synthesized by pyrolyzing

toluene over a high surface area MgO (Aldrich) incipient-wetness-impregnated with

2wt.% Fe from iron acetate (Aldrich). When doped with phosphorus, the support went

through an additional IWI step with triphenylphosphine in diethyl ether. The resulting

growth medium was heated in N2 (g) to 900 C, and then exposed to nitrogen saturated

with toluene for 2 hr. The resulting catalyst was treated in 1M HCl (aq) for 1 hr at 60 C

to remove exposed magnesia and metal, and then rinsed with DI-H2O. Upon drying, the

resulting graphitic carbon-nitrogen ORR catalyst is denoted as “CPy”.

The electrochemical oxygen reduction activity was determined by CVs in an O2-

11 saturated half-cell of 0.5M H2SO4 on the RRDE, as described elsewhere. All RRDE

testing was performed in 0.5 M H2SO4 at 1000 rpm saturated with oxygen at room 1 temperature with 10 mV/s potential sweeps. The catalyst loading was 426 g/cm2. The : ORR activity was determined in the same manner as all other RRDE experiments in this 3 chapter. The phosphorus used to grow CP catalysts did not appear to impart additional 0 y activity to the resulting catalyst materials, Figure 40. It can be seen that compared to the

graphite material grown with only toluene that all levels phosphorus doping of in CPy

147 decreased the oxygen reduction activity. The take off potential for the CPy catalysts are all much lower than CNx and all CNxPy catalysts, suggesting that phosphorus, iron, and a nitrogen source have to be present during carbon growth for increased in oxygen reduction activity. This study demonstrates that phosphorus incorporated into graphites in this manner does not form oxygen reduction active sites on graphitic carbons.

Figure 40. Cathodic scans of CPy catalyst-coated RRDE in Oxygen Saturated Electrolyte

5.6 Phosphorus Doping Conclusions

148

Phosphorus inclusion in the growth media of CNx catalysts enhanced the ORR activity in acidic electrolyte. This increase in activity was directly correlated with the

Faradaic capacitance of the catalyst material on the rotating ring disk electrode.

Phosphorus also slightly increased the stability and order of the catalyst materials. The presence of phosphorus was found to alter the morphology of the CNx nanostructures formed during pyrolyic growth. Increasing levels of phosphorus appeared to increase the disorder in the nanostructures formed. No trend in nitrogen or oxygen surface composition correlated with the oxygen reduction activity of the phosphorus doped CNx electrocatalysts. Although, temperature programmed oxidation shown a similar nitrogen composition in the bulk of all catalysts studied. X-ray absorption showed that phosphorus introduced to the iron growth media decreases the size of the iron nanoparticles found encased in carbon layers within the resulting CNxPy catalysts. Unfortunately, none of the techniques used thus far were able to conclusively detect phosphorus. To synthesize catalyst material with increased oxygen reduction activity the iron, phosphorus, and nitrogen source must be present during the growth of the carbon nanostructured catalyst.

The inclusion of phosphorus and nitrogen dopants into graphite materials allows for the tailoring of electrocatalytic and physical properties of carbon nanostructures.

149

CHAPTER 6. Probing the Oxygen Reduction Active Site Through Catalytic Poisons

Since there is no consensus for the oxygen reduction active site on the non-

precious metal carbon-based catalysts, a deductive poisoning methodology was used to

eliminate potential metal containing active sites. Many researchers consider the active

site for oxygen reduction an active metal center that is stabilized by nitrogen ligands with

a structure similar to macrocycles compounds formed during pyrolysis.1,55,118,178-184

Another group of researchers believe that graphitic carbon nitrogen groups have inherent

oxygen reduction activity without any metal present.64,133,185-187 Both hypothesized active

site types would have enhanced ORR activity with the inclusion of nitrogen into the

carbon graphite hexagonal network making experimental active site observations

inconclusive.188-190 Conventional catalytic metal poisons were used to probe the validity

of active metal center type hypotheses.

150

6.2 Investigation of Sulfur Poisoning of CNx ORR Catalyst for PEM Fuel Cells

(adapted from 286)

6.2.1 Introduction

The role of the transition metal used to form non-noble metal electrochemical oxygen reduction CNx catalysts was investigated through sulfur treatment, a well-known poison for transition metal based catalysts, on the CNx electrocatalyst. The intent of sulfur poisoning was to show the existence of a non-iron containing electrocatalytic active site in CNx. The sulfur treatment increased the overpotential on a platinum catalyst but enhanced the current density of the CNx catalyst. The deposition of sulfur onto CNx catalyst was verified by temperature-programmed oxidation and X-ray photoelectron spectroscopy. Iron K-edge X-ray absorption near edge structural analysis suggested that the CNx iron phase, which was primarily composed of nanometer-sized metallic particles, was unchanged by sulfur poisoning. This suggests that either the iron-based active site is sulfur tolerant or that this active site does not participate in the electrocatalytic reduction of oxygen in CNx catalysts.

151

6.2.2 H2S Poisoning Introduction

Fuel cells have the potential to radically alter the contemporary energy landscape, but many economic and technological challenges must first be overcome. Economically prohibitive loadings of platinum are used in the cathodes of PEM fuel cells to overcome the slow kinetics of the ORR.

Non-noble metal catalysts have been shown to have significant oxygen reduction activity and could potentially reduce or replace platinum in PEM fuel cells. Research on the oxygen reduction reaction over non-noble metal electrocatalysts was inspired by naturally occurring organic macrocycles in hemoglobin where oxygen adsorption readily occurs at low temperatures. In hemoglobin, oxygen adsorbs onto an iron center (heme site) that is coordinated by four surrounding nitrogen functional groups at 37 C.47 It was discovered that heme-like molecules, such as iron pthalocyanines and macrocycles, were active for ORR in the fuel cell cathode, but quickly deactivated in acidic environments

50-53 where the Fe-Nx active sites are destroyed. To improve the stability of these materials, pyrolysis, heat treatment in an inert atmosphere, above 600 C was used to stabilize the organic macrocycles on a carbon support.51-53,55-64 Although the pyrolysis step was found to completely destroy the coordinated metal type active site, the catalyst materials prepared through pyrolysis were found to be more active and stable than macrocycles, leading researchers to believe that a new active site had been formed during

152 the heat treatment.72 The structure of this new active site has not been definitively determined, although researchers have hypothesized a number of different sites ranging from metal atoms stabilized by nitrogen groups, to non-metallic sites grown catalytically by the presence of a metal during pyrolysis.51,61,63,94,103,283

The high-temperature treatment used to stabilize non-noble metal catalyst materials facilitates the formation of nearly all possible bonding configurations, thereby allowing many atomic arrangements any of which could be responsible for ORR activity.

It has been established that a carbon source, a transition metal, and a nitrogen source need to be present during pyrolysis to achieve high ORR activity,75,82-86,90,92-95,97,287 but the exact bonding configuration that facilitates high ORR activity is still debated.

We have previously shown that significant ORR activity can be achieved in a metal-free (< 1 ppm Fe) CNx catalyst and in light of those findings, it was suggested that iron was not a part of the active site, but plays the role of a growth catalyst for carbon nanostructures.132,133 Although the presence of multiple active sites, i.e., one with a metallic center and one without, cannot be ruled out, the activity observed over these materials being solely due to the presence of Fe centers in very low concentrations (100 ppm or less) is less likely. If this were the case, the intrinsic activity of these metal centers would be exceptionally high. In this paper, we adopt an exclusionary approach by attempting to deliberately poison the catalysts with sulfur to determine whether a transition metal is present in the active site for oxygen reduction in CNx materials. CNx catalysts prepared using a number of different transition metals during synthesis have been shown to have significant activity with the CNx catalyst grown on Fe or Co- 153 containing media having the highest activity.73,283,288-291 Sulfur is a well-known catalyst poison, which has been shown to deactivate iron-based catalysts for Fischer-Tropsch, water-gas shift, ammonia synthesis, ammonia decomposition, and iron carburization.6-9 A hydrogen sulfide treatment can be expected to have a similar poisoning effect on any iron containing active sites present on non-noble metal ORR electrocatalysts. To the best of our knowledge, this is the first time there have been any reports on the sulfur poisoning attempts of CNx oxygen reduction catalysts.

6.2.3 Results and Discussion

6.2.3.a H2S Poisoning Treatment Effect on Oxygen Reduction Activity

A hydrogen sulfide treatment was used to poison any iron containing active sites present on CNx and platinum oxygen reduction catalysts. It has been shown that CNx activity is correlated with microporous (< 2 nm) pore distribution, leading to the hypothesis that the ORR active site(s) is housed within Fe/N/C-type centers located in the micropores.116 To effectively poison the proposed microporous Fe/N/C active site, the sulfur poison will need physical access to the sites while being chemically active.

Oxygen has a collision diameter (O2) of 3.467 Å while hydrogen sulfide has a collision

292 diameter (H2S) of 3.623 Å. This would suggest that hydrogen sulfide would have nearly the same access to micropores as oxygen in CNx catalyst. Furthermore, hydrogen sulfide is known to readily bind to the oxygen adsorption iron site in heme at 37 °C,47 and 154

9 to poison Fe-based water gas shift catalysts in 300-450 °C range so the H2S treatment at 350 °C used in this study should be sufficient to bind and/or activate H2S on the

Fe/N/C type active sites.

Oxygen reduction activity testing was carried out by cyclic voltammetry in a rotating ring disk electrode setup placed in an acidic solution to mimic the cathode environment in a direct methanol or hydrogen fuel cell. To eliminate the heat treatment and reduction effects of H2S poisoning separate samples of both CNx and 20 wt%Pt supported on Vulcan carbon were also heat treated under the same conditions in a 5.7

Figure 41. Oxygen reduction activity of Pt/Vulcan carbon catalysts. Inset: Selectivity for water formation

155

%H2/N2 atmosphere. A reducing environment of hydrogen was used to capture the reducing effects of H2S. Figure 41 shows a comparison of the oxygen reduction activity of an untreated Pt/Vulcan carbon with the same catalyst that underwent different treatments, i.e. H2-N2 or H2S, prior to the testing of the activities. The decrease in activity shown for the platinum catalyst verifies that the H2S treatment has a detrimental effect on the ORR activity under the treatment conditions used in this study. The decrease in activity can be observed in both the lower limiting current and the lower take off potential (i.e., higher overpotential) compared to the untreated platinum catalyst.

Although there was some current density decrease following the treatment in H2-N2, the performance loss was much more pronounced when the catalyst was treated in an H2S- containing atmosphere. This observation is not surprising since sulfur has been previously reported to poison platinum ORR catalysts 8,293. The current density loss under

H2-N2 could be due to sintering of the active phase during heat treatment or a disorder increase of the Pt phase 294. The nearly identical take off potentials of the H2-N2 and untreated platinum catalysts demonstrated that the activity was nearly the same for these catalysts. For all treatments, the selectivity of the platinum catalyst remained high, with number of electrons transferred per O2 molecule being ~3.9 or above. (Figure 41, inset).

The Pt catalyst experiments were important in showing that the H2S treatment was sufficient to impart a measurable poisoning effect; hence a similar procedure was used for CNx catalysts.

Figure 42 shows similar linear polarization curves for the untreated CNx and catalysts that were treated under different atmospheres, i.e. H2-N2 or H2S, prior activity 156

Figure 42. Oxygen reduction activity of CNx catalysts. Inset: Selectivity for water formation

testing. The sulfur poisoning treatment was found not to reduce the activity of CNx oxygen reduction catalyst. The CNx ORR current density and take off potential were found to improve with H2S treatment (Figure 42). The treatment in H2-N2 atmosphere also led to an improved performance, but the improvement was much more pronounced with H2S treatment. This result is significant in showing that either the iron containing active site is sulfur tolerant or does not participate in the electrocatalytic reduction of oxygen. This result is not surprising considering it has already been shown that sulfur compounds introduced during pyrolysis could enhance CNx catalyst growth while the 157 oxygen reduction activity remained largely unchanged.11 Although it is not clear why an enhancement in CNx ORR activity was observed with H2S treatment, Chung et al. reported a similar activity increase on a non-precious metal catalyst synthesized by pyrolyzing cyanamide and iron sulfate.295 If N-sites incorporated into the carbon nanostructure have ORR activity, it is conceivable that other heteroatoms such as sulfur can fulfill a similar role. In fact, results suggesting ORR activity of other heteroatom containing carbon structures, such as phosphorus or oxygen have been reported

11,173,271,296 earlier. The selectivity for the formation of H2O on CNx remained relatively low compared to platinum for all CNx catalysts regardless of treatment, but the formation of H2O was preferred over H2O2.

The selectivity of CNx was slightly lower (n = 3.6 - 3.8, Figure 42, inset) than the platinum supported on Vulcan carbon catalyst (n > 3.9, Figure 41, inset) studied.

Although the primary product for both catalysts is H2O, the undesired product hydrogen peroxide that reduces fuel cell performance with its lower standard reaction potential and corrosive nature was produced. The stronger oxidation properties of hydrogen peroxide compared to dioxygen is thought to accelerate the degradation of internal fuel cell components although the effect on fuel cell lifetime are also material and operational dependant.269

6.2.3.b X-ray Photoelectron Spectroscopy

CNx catalysts that underwent heat treatment under different atmospheres were

158

Figure 43. X-ray photoelectron spectra in the N 1s region for untreated and heat treated CNx catalysts under different atmospheres

159 further characterized with XPS and temperature-programmed oxidation experiments to investigate the effect of such treatments on the surface and bulk properties of these catalysts and to determine the extent and location of sulfur incorporation to the carbon phase of the CNx catalysts. The sulfur poisoning of the iron phase in CNx catalyst was further studied by X-ray absorption spectroscopy.

Iron and magnesium were not detectable in the survey scans of any of the CNx catalysts studied. This finding is consistent with our previous reports 279 and suggests that the acid washing step for CNx fabrication removed most of the surface iron and magnesia. The XPS analysis of CNx catalysts found similar nitrogen species surface compositions, regardless of the

160

Figure 44. X-ray photoelectron spectra of the S 2p region for untreated and H2S treated CNx catalysts under different atmospheres treatment (Figure 43). Figure 44 shows a comparison of the X-ray photoelectron spectra in the S 2p region for an H2S treated and an untreated CNx catalyst. Although the counts in this region is low due to the low intrinsic atomic sensitivity of the instrument for sulfur, photoelectron peaks arising from the presence of sulfur species is observed in the

S 2p envelope for the H2S treated catalyst. Curve fitting for providing insights to the nature and relative surface concentration of sulfur species has not been possible due to weakly resolved envelope. However, it is possible to discern two different photoelectron

161 peaks located at 163.5 eV and 167.5 eV, which can be associated with sulfur species in

2- 2- 297 sulfide (S ) and sulfite (SO3 ) environments, respectively. It should be noted that the splitting between 2p3/2 and 2p1/2 was not observed for the S 2p envelope due to small spin orbit splitting for sulfur (1.18 eV). Formation of sulfite species can be associated with interaction of adsorbed sulfur with oxygen from ambient air during sample transfer, although an interaction between sulfur and the oxygen-containing functional groups present on the CNx catalysts leading to formation of sulfite species cannot be ruled out.

The elemental surface compositions of the CNx samples determined from the quantitative analysis of XPS data are presented in Table 3. The results showed a

Table 3. Compositional analysis of CNx catalyst X-ray photoelectron Spectroscopy

redistribution of the elemental surface concentration with H2S treatment. The relative amount of surface oxygen was less for CNx catalysts exposed to a reduction treatment. It was determined that changes in the oxygen 172 and nitrogen 86,192 surface species can affect ORR electrocatalytic activity, which could contribute to the changes observed in

ORR activity with H2-N2 or H2S treatment.

6.2.3.c Temperature-programmed Oxidation 162

Temperature programmed oxidation experiments were carried out over the untreated CNx catalyst as well as catalysts that were heat-treated under different atmospheres (Figure 45). Regardless of the pretreatment conditions, the CO2 (m/z =

Figure 45. Evolution of CO2 (m/z = 44 amu), NO2 (m/z = 46 amu) and SO2 (m/z =64 amu) as a function of temperature during temperature programmed oxidation of untreated, H2S, N2, and H2 treated CNx.

44) and NO2 (m/z = 46) traces closely followed each other and broad evolution bands for these species, which are associated with the decomposition of the CNx structure, were observed in 510 °C - 720 °C region. Over the H2S treated catalyst, evolution of SO2 (m/z

= 64) that is concomitant to CO2 and NO2 was observed. This indicates that sulfur species 163 that were retained on the surface after the H2S pretreatment were strongly bonded to the surface rather than being physically adsorbed.

6.2.3.d X-ray Absorption Spectroscopy

Since a transition metal is needed during the pyrolic formation of the non-noble metal active site, it is difficult to ascertain the role of iron in ORR activity. X-ray absorption spectroscopy was used to probe the local bonding environment of the iron in

CNx oxygen reduction catalysts. The XANES spectra for all CNx catalysts, Figure 46, were nearly identical

164

Figure 46. XANES Fe K-edge spectra for untreated, H2S-, and H2-treated CNx regardless of the post-synthesis treatment conditions indicating that the local Fe structure is unchanged by H2S treatment. All CNx XANES spectra appeared to strongly resemble iron in the metallic or carbide phase, Figure 47. None of the CNx samples studied resembled the iron phthalocyanine standard, which is similar to the nitrogen-bonded

1,180,283 transition metal active site proposed by some researchers. TEM of CNx catalysts revealed iron entirely encased within several carbon sheets, Figure 47, inset. Since the iron phase is isolated from the chemical

165

Figure 47. XANES Fe K-edge spectra for, iron foil, iron pthalocyannine, iron carbide and untreated CNx. Inset: TEM of iron nanoparticle encapsulated in CNx catalyst

treatments by layers of carbon it is not surprising that the Fourier transform obtained at the Fe K edge and k2 weighted EXAFS displayed a nearly identical local bonding environment for untreated and H2S treated CNx, Figure 48. The iron contained within

CNx has a small metallic phase and an iron phase bonded to a lighter element (nitrogen, oxygen, and/or carbon). It can be seen that untreated CNx Fe-Fe bond length is less than that of bulk iron, Figure 49. This is likely due iron particles being much smaller in CNx catalyst than of a bulk iron foil. The iron-iron bonding shoulder at ~1.5 Å in CNx catalyst coincides with Fe-N bond distance in iron 166

Figure 48. EXAFS Fe K-edge spectra for H2S treated, and untreated CNx phthalocyanine, but also aligns with the Fe-C bond distance in iron carbide. Due to the similar scattering spectra of 2p elements, it cannot be determined if the CNx peak at ~1.5

Å is due to an iron nitrogen, carbon, or oxygen bond. The EXAFS for iron carbide displayed the same peak structure and locations as the CNx catalysts studied. This is in agreement with an investigation of the CNx iron phase through Mössbauer spectroscopy which revealed iron to be metallic (a paramagnetic -Fe phase, which would be unstable in air) and some cementite phase (Fe3C), which is likely to be formed during the carbon

167 fiber growth.279 Working iron catalysts are often a complex mixture of iron carbide compounds, often containing a fraction of iron oxides.284 An

Figure 49. EXAFS Fe K-edge spectra for untreated CNx, iron phthalocyanine, iron metal and iron carbide

EXAFS theoretical structural model was prepared by Jacobs et al. using ab initio multiple scattering calculations of X-ray absorption fine structure for cementite (Fe3C), Hägg carbide (Fe5C2), -carbide (Fe3C), and -carbide (Fe2C) to generate the Fourier transform magnitude versus bond distance for the iron K-edge in order to characterize the iron

284 carbides present in Fischer-Tropsch synthesis catalysts. CNx oxygen reduction 168 electrocatalyst resembles a combination of the -carbide (Fe3C) and Hägg carbide (Fe5C2) when compared to the simulated carbide EXAFS spectra. Although, CNx catalysts are grown in a reducing atmosphere, the presence of iron oxides cannot be ruled out.

6.2.3.e H2S poisoning Conclusions

A sulfur poisoning treatment was performed on both platinum and CNx oxygen reduction catalysts to investigate the role of the transition metal in the CNx ORR active site. The intent of sulfur poisoning treatment was to effectively eliminate iron from the proposed electrocatalytic oxygen reduction active sites of CNx. The sulfur treatment was found to reduce ORR activity on the platinum catalyst, which demonstrated the validity of the sulfur poisoning treatment. The H2S treatment was found to increase the activity on the CNx catalyst, but this is likely due to a reorganization of the catalyst surface species.

The incorporation of sulfur into the CNx catalyst was verified by TPO and XPS spectra.

The iron phase within the CNx catalysts was investigated with X-ray absorption and was found to be similar for a treatments studied, suggesting that iron was not abundantly present on the catalyst surface. XANES analysis of the CNx Fe K-edge showed the iron phase in CNx to be very metallic and no spectra resembled the iron macrocycle standard, which is similar to the hypothesized nitrogen bonded transition metal ORR active site.

169

6.3 The Effect of Carbon Monoxide on Oxygen Reduction PEM Fuel Cell Catalysts

6.3.1 Introduction

Hydrogen fuel cells have the potential to transform energy conversion in portable devices if several scientific obstacles could be overcome. Currently, the proton exchange membrane (PEM) fuel cell cathode utilizes high loadings of high-priced platinum supported on carbon to overcome the slow kinetics of the oxygen reduction reaction

(ORR).298 Less expensive, non-noble metal catalysts consisting of a graphitic carbon doped with nitrogen (CNx) have also shown significant oxygen reduction activity under fuel cell conditions and could potentially reduce or replace platinum in PEM fuel cells1

Unfortunately, CNx catalysts deactivate during fuel cell operation and attempts to stabilize these catalysts have largely relied on chemical intuition.4,123,299 To elucidate the catalytic behavior and active site of CNx, a carbon monoxide treatment was introduced to

CNx and platinum catalysts during electrochemical oxygen reduction to investigate the competitive adsorption of carbon monoxide in the presence of oxygen.

The active site ambiguity in CNx-type electrocatalysts originates from the historical development of these materials. Non-noble metal oxygen reduction catalysts were inspired from hemoglobin where oxygen adsorbs onto a four-way nitrogen coordinated iron center (heme site) at low temperatures.47 Molecules that had a coordinated transition metal, such as iron or cobalt pthalocyanines, are active for ORR in the fuel cell cathode, but quickly deactivate in acidic solution due to metal dissolution

50 causing the destruction of the metal-N4 oxygen reduction site. In an effort to increase 170 the stability, the metal-N4 active site materials were placed on a carbon support and pyrolysized, heat treatment in an inert atmosphere, above 600 C.51,52,64 The pyrolyzed metal-N4 materials were found to be more stable and surprisingly, more active, but the heat treatment was also found to destroy the coordinated metal type active site, making it apparent that new active site had been formed.72 Researchers have hypothesized many different active sites, ranging from metal atoms stabilized by nitrogen groups to non- metallic sites grown catalytically by the presence of a metal during pyrolysis, but the exact bonding configuration that facilitates ORR activity is still debated.51,61,94,103,283

The performance of oxygen reduction electrocatalysts diminishes during long- term operation. This phenomenon is not only present for non-precious metal catalysts, but also occurs for nano-dispersed platinum catalysts. For a carbon supported platinum oxygen reduction site to be electrochemically active it must have simultaneous access to protons, reactant oxygen, and electrons at the operating potential.24 If any one of these conditions is not satisfied the platinum particle will not participate in oxygen electroreduction. The high potential, strongly acidic, and oxidizing environment of the

PEM fuel cell cathode causes the electron conducting carbon support to corrode and platinum to separate which decreases the activity of the catalyst layer 17,300. Platinum has actually been shown to catalyze the corrosion of the surrounding carbon leading to the

19,300 local formation of CO and CO2. On carbon supported platinum, hydrogen peroxide evolution, which is known to degrade fuel cell components, becomes apparent at < 0.6 V vs. NHE, so fuel cell operation is at higher potentials to avoid hydrogen peroxide evolution, but unfortunately the formation of carbon monoxide is thermodynamically

171 possible (>0.518 V vs. NHE).18,265 The carbon monoxide that is generated can adsorb more strongly to the platinum catalyst than oxygen, inhibiting oxygen reduction. Fuel cell performance is also diminished by dissolution of platinum in the cathode which either decreases the electroactive area by Ostwald ripening 301 or deposits platinum away from the electrolyte membrane.302

Like platinum, non-precious metal carbon-based CNx catalysts lose activity during fuel cell operation. However, the active site for CNx is unknown and as a consequence the mechanism of deactivation is largely unknown. The graphitic nature of carbon materials correlates with corrosion resistance, so unsurprisingly the stability of CNx catalysts increases with graphitic character of the material.123,303 The catalytic oxygen

112 reduction activity of CNx also increases with microporous surface area, suggesting that chemically active graphitic carbon edge planes are responsible for activity and stability. It is generally agreed that nitrogen incorporation into the graphite matrix is needed to form a highly active carbon-based non-precious metal oxygen reduction catalyst. However, many researchers consider the active site for oxygen reduction an active metal center stabilized by nitrogen ligands, similar to macrocycles compounds, formed during pyrolysis.1,55,118,178-184 Another group of researchers believe that graphitic carbon nitrogen groups have inherent oxygen reduction activity without any metal present.64,133,185-187

Both hypothesized active site types would have enhanced ORR activity with the inclusion of nitrogen into the hexagonal network.188-190

Carbon monoxide is not thought to interact with the oxygen reduction active site.

Several researchers believe that the active site for oxygen reduction in CNx materials is a 172 transition metal coordinated by nitrogens. Although, Fe-porphyrin, a commonly used precursor.123,304,305 has approximately ~2 x 104 times more binding affinity for CO than

122 O2. In porphyrins, the hydrophobic pocket created by N-H groups around Fe-N4 sites that could improve the O2 affinity, but would consequently also show a similar affinity

10 for CO. No difference in voltammetric electrochemical capacitance of a CNx pyrolyzed cobalt porphyrin electrocatalyst was observed when the acid electrolyte was saturated

277 with CO or N2. Additionally, no CO adsorption or oxidation peaks were observed in electrochemical capacitance of this CNx when the electrolyte was saturated with carbon monoxide. However, on a different pyrolyzed metal porphyrin in acidic electrolyte a carbon monoxide adduct was detected by X-ray absorption by analyzing the pre-edge features, but this adduct disappeared at > 0.6 V vs. NHE.306 It may be expected that the binding of CO to the metal porphyrin would block the active site for dioxygen reduction and render the electrocatalyst inactive, but the electrochemical oxygen reduction activity of a non-precious metal catalysts was found to be reversible after carbon monoxide exposure 122. This suggests that carbon monoxide does not strongly interact with the electrode surface, although the competitive adsorption of carbon monoxide on CNx electrocatalysts in the presence of oxygen has not been previously reported. By determining the effect of the competitive adsorption of carbon monoxide in the presence of oxygen on CNx, the active site for ORR can be either be classified as metal-containing or metal-free.

173

6.3.2 Experimental

6.3.2.a Catalyst Synthesis

The synthesis of the non-precious metal nitrogen doped graphite oxygen reduction electrocatalyst used for this study has been described in greater detail elsewhere.307

High performance 20wt% platinum on Vulcan carbon (BASF) was used as received.

6.3.2.b Electrochemical Testing

1) Catalyst Application and Electrochemical Setup

The ORR activity and selectivity of catalysts was determined by electrochemical half-cell testing with a rotating disk electrode. Catalyst inks were prepared using a composition of 1:10:160 (by mass) catalyst mixture: 5% Nafion in aliphatic alcohols:

100% ethanol. Inks were low-energy sonicated for 30 min. Catalyst ink was applied to a

0.1256 cm2 glassy carbon disk, resulting in a catalyst loading of 427 mg/cm2. A model

PAR 616 RDE setup was connected to a Princeton Applied Research Bistat for electrochemical testing. An Ag/AgCl (saturated KCl) reference electrode and carbon

174 graphite counter electrode were used for the half-cell system. All reported potentials are referenced versus the normal hydrogen electrode (NHE). The acidic electrolyte used was

0.5 M H2SO4 (aq).

2) Chronoamperometric CO Poisoning Experimentation

The 100 mL 0.5 M H2SO4 electrolyte was saturated with oxygen by diffusing gaseous oxygen through the electrolyte. Immediately after the electrode was submerged in the electrolyte, CV from 1.2 V to 0.0 V to 1.2 V vs. NHE at 10 mV/s on the catalyst coated glassy carbon disk at 1000 and 0 rpm was collected ascertain the initial activity of the catalyst and oxygen functional groups, respectively.

After the first CV concluded, the gas diffused through the electrolyte was changed to either CO at 30 ccm or 30 ccm CO and 30 ccm O2 for 30 minutes. Thirty minutes was sufficient to saturate the electrolyte as demonstrated by the chronoamperometric experiments as previously described. Then, gas treatment CV from 1.2 V to 0.0 V to 1.2

V vs. NHE at 10 mV/s on the catalyst coated glassy carbon disk at 1000 and 0 rpm was collected.

Then, gaseous oxygen was diffused through the electrolyte at 30 ccm through the electrolyte for 30 minutes. A post treatment CV from 1.2 V to 0.0 V to 1.2 V vs. NHE at

10 mV/s on the catalyst coated glassy carbon disk at 1000 and 0 rpm was collected in the oxygen saturated electrolyte. 175

Finally, gaseous argon was diffused through the electrolyte at 30 ccm through the electrolyte for 30 minutes. A background CV from 1.2 V to 0.0 V to 1.2 V vs. NHE at 10 mV/s on the catalyst coated glassy carbon disk at 1000 and 0 rpm was collected in the argon saturated electrolyte.

6.3.2.c Carbon Monoxide Pulse Chemisorption

Carbon monoxide pulse chemisorption experiments where performed at room temperature on platinum supported on Vulcan carbon and CNx using a chemisorption analyzer Autochem II 2920. 50 mg of 20 weight% Pt-VC was placed in a quartz u-tube reactor and purged with He for 30 minutes to remove all oxygen from the system. In order to carry out a reduction-pretreatment, a temperature ramp of 10oC/min to 400oC was initiated, upon which 5% H2-He was introduced for 30 minutes. Thereafter, the reactor was cooled to 35 oC, and 9 pulses of pure CO each comprising of 574 l, were introduced at intervals of 2 minutes into the reactor. A similar experiment was performed with 50 mg of CNx, using a lower concentration of carbon monoxide (10% CO in

Helium) to negate the effect of not being able to see an appreciable difference in pulses by fully saturating the iron sites, which are less than 2% in concentration, if pure CO was used.

176

6.3.3 Results and Discussion

Carbon monoxide is a known poison for oxygen reduction over platinum electrocatalysts 308,309, so the electrochemical activity differences with CO will be ascribed to carbon monoxide poisoning and compared to CNx under identical conditions.

In an electrolyte continuously saturated with oxygen while the electrodes were held at

0.3V vs. NHE, all three catalysts deactivate at varied rates, Figure 50. The negative

Figure 50. CNx, 20 wt% Pt on Vulcan carbon, 2 wt% Fe Iron phatholocyanine on Vulcan carbon RDE at 0.3 V vs. NHE initially the electrolyte saturated with oxygen at time 1590 seconds, 1000 rpm, and 0.5 M H2SO4 (aq). normalized current that corresponds to oxygen reduction activity for the iron

177 phthalocyanine catalyst decreases the most during the potential hold experiment, indicating that this catalyst is the least stable of those studied. Platinum and CNx both decrease in activity at a similar rate, but Pt is slightly more stable. It should be noted that

263 H2SO4 is thought to poison platinum catalysts through sulfide poisoning, but since the electrochemical experiments are identical for all platinum on Vulcan carbon methods studied, the deactivation would be equivalent thus making any activity differences attributable to the chemical treatments used. Furthermore, it was reported that there is no

44,60 difference in ORR activity for CNx catalysts in H2SO4 and HClO4 aqueous solutions, so it is unlikely that sulfide and chloride atom significantly interact with the CNx active site.

Carbon monoxide poisoning experiments were studied on a platinum catalyst to highlight the difference between CO poisoning and oxygen displacement. Carbon monoxide strongly adsorbs to platinum and completely eliminates all oxygen reduction.

A similar catalyst poisoning effect was observed for a 2% CO/H2 saturated electrolyte when investigating the hydrogen oxidation reaction over high surface area platinum.309

This phenomenon was observed when the oxygen reduction activity diminished entirely upon introduction of carbon monoxide to an oxygen-saturated electrolyte, Figure 51 b) & d) Platinum. The decrease in activity was the same whether the bubbling gas was switched from O2 to CO (O2-CO-O2) or when both gases were bubbled at the same rate to study the competitive adsorption of CO and O2 (O2-CO&O2-O2). Oxygen reduction activity reversibly recovered for both CO treatments when the electrolyte was re- saturated with oxygen. To compare the effects of CO poisoning to oxygen displacement

178 chemically inert argon was sparged through the electrolyte while oxygen was switched off (O2-Ar-O2) or when oxygen flowed continuously (O2-Ar&O2-O2), Figure 51 a) & c)

Platinum. The oxygen reduction activity decreased at a much slower rate when sparging argon, consistent with displacement of oxygen from the electrolyte, as compared to the abrupt activity change observed with CO poisoning in corresponding experiments.

Carbon monoxide experiments over CNx electrocatalyst showed no oxygen reduction poisoning effect. When CO (O2-CO-O2) or Ar (O2-Ar-O2) were diffused through the electrolyte, the oxygen reduction behavior was similar for both experiments and the oxygen reduction activity of CNx returned to the same level suggesting that CO poisoning, if present, was reversible, Figure 51 a) & b) CNx. Birry et al. also reported a recoverable oxygen reduction activity loss with CO treatment during chonoampermetric

179

a CNx Platinum )

b CN x Platinum )

c CNx Platinum )

d CN x Platinum )

Figure 51. CNx (left) and Platinum on Vulcan Carbon (right) RDE at 0.3V vs. NHE initially the electrolyte saturated with oxygen at time <1.5 min, at 1.5 min gas flow changed to argon (a), carbon monoxide (b), argon and oxygen (c) or carbon monoxide and oxygen (d), at 26.5 min gas flow changed to oxygen, 1000 rpm, 0.5M H2SO4 (aq). 180

122 half-cell testing . The decreased ORR current for both the (O2-CO-O2) or (O2-Ar-O2) were very similar suggesting that loss in ORR current was due to the displacement of oxygen by an inert argon or carbon monoxide species. Argon displaced oxygen at a slower higher rate initially but allowed oxygen to resaturate the electrode at a faster rate than CO, which is to be expected because argon is less soluble in aqueous solutions than

310 CO. When CO&O2 or Ar&O2 were bubbled through the electrolyte the oxygen reduction current had a similar normalized decrease and behavior indicating that CO was not preferentially adsorbed on CNx in the presence of oxygen like what was observed on platinum in Figure 51 c) & d). The ORR current was less for CO&O2, which was likely due to the higher solubility of carbon monoxide in aqueous solutions as compared to argon. This suggests that the CNx oxygen reduction active site does not appreciably bond to carbon monoxide. Carbon bonding tolerance is also evident when CNx materials preferentially reduce oxygen in the presence of methanol whereas platinum catalysts are poisoned by an intermediate carbon monoxide formed from the methanol oxidation reaction.307

181

a)

b)

Figure 52. a) CNx and b) 20wt%Platinum on Vulcan carbon RDE cathodic potential scan at 5 mV/s from 1.2 V to 0.0 V vs. NHE in oxygen saturated 0.5 M H2SO4 before and after 0.3 V vs. NHE potential hold in O2-CO-O2 at 1000 rpm.

182

The cyclic voltammograms taken in oxygen saturated electrolyte before and after the gas treatments were found to nearly identical for platinum and CNx catalysts regardless of gas treatment, Figure 52. This demonstrated that the deactivation seen during the treatment gases was recoverable for both platinum and CNx catalysts. There is a slight decrease in activity this can be attributed to the deactivation of the catalyst due to

Figure 53. CNx RDE cathodic potential scan at 5mV/s from 1.2 V to 0.0 V vs. NHE in oxygen, carbon monoxide & oxygen, oxygen after treatment, then argon saturated 0.5 M H2SO4 at 1000rpm. the oxygen reduction that takes place on the electrode during the 0.3 V vs. NHE potential hold, as seen in Figure 50. This was confirmed by an experiment where CVs where

183 performed without a potential hold in electrolyte saturated with oxygen, oxygen and carbon monoxide, oxygen, then argon, Figure 54.

Room temperature (25 ºC) Carbon monoxide pulse chemisorption experiments confirmed that CO is adsorbed to platinum supported on Vulcan carbon and not adsorbed on CNx catalysts, Figure 54. A series of nine 574 l pulses at 35 ºC pure carbon monoxide pulses and the platinum supported catalyst partially absorbed the first two pulses in Figure 54. The CO pulse chemisorption experiment on CNx catalyst was done with

Figure 54. Carbon monoxide, 28 atomic units, pulse chemisorption on high performance 20 wt%Pt/VC (BASF) at 25 ºC spectra, 9 pulses shown (provided by Deepika Singh)

184

10%CO/He to increase the absorption sensitivity to capture the effects of very low concentrations of iron on the surface, but no CO adsorption was observed. Past work has shown that the iron used to synthesize CNx is mainly is encased in carbon sheets and in the carbide phase, hence it is not surprising that a chemical treatment targeted to poison iron would be ineffective, due to the inability of the poison to reach iron buried within the carbon sheets.279

6.3.4 CO poisoning Conclusions

During oxygen reduction carbon monoxide preferentially adsorbed to the platinum electrocatalysts causing all oxygen reduction to cease, but was reversibly recovered once carbon monoxide left the system. Carbon monoxide does not preferentially adsorb to the oxygen reduction active site in CNx during oxygen reduction. Deactivation of the electrocatalysts was found to only take place during oxygen reduction under potential load. Carbon monoxide pulse chemisorption confirmed that CO is adsorbed to platinum on Vulcan carbon and not adsorbed on CNx.

185

CHAPTER 7. Examination of Rotating Ring Disk Electrode Methodology and its Impact on Reported Selectivity

7.2 Introduction

Oxygen reduction catalyst research is typically conducted through a high material

throughput process where catalysts are first screened for activity and selectivity for water

formation with a RDE and RRDE half-cell experimental setups that mimics the cathode

fuel cell environment. The best performing catalysts evaluated by RRDE or RDE are then

manufactured into membrane electrode assemblies and tested in full fuel cells. For this

process to function well the RDE or RRDE methods must determine activity and

selectivity accurately and reproducibly. For oxygen reduction catalyst testing RDE and

180,311-314 RRDE half-cells that use an acidic electrolyte such as HClO4,

41,83,120,179,240,315 186 185 H2SO4, HCl, or HNO3 are saturated with oxygen to simulate the

highly acidic, oxidizing cathode environment of a PEM fuel cell. It is known that the

kinetics of the oxygen reduction reaction on the catalyst surface can be affected by the

- anions in the electrolyte, but it is generally accepted that ClO4 is the least absorbing

anion on platinum and platinum-based catalysts and therefore gives the highest Pt

activity.263,265 So to compare the activity of platinum catalyst to one the same electrolyte

should be used while HClO4 is preferred to express the highest performance. For non-

precious metal carbon-based oxygen reduction electrocatalysts there is no significant 186

difference in ORR activity/selectivity testing in H2SO4 and HClO4 electrolyte

44,60 - - solutions. This suggests that the HSO4 and ClO4 anions absorb equally on the

unknown activity site on CNx materials during oxygen reduction. For non-precious metal,

nitrogen-doped carbon oxygen reduction catalysts a direct comparison of activities can be

made with H2SO4 and HClO4 half-cell electrolytes.

The selectivities determined by the RDE and RRDE techniques require that the

hydrodynamic and convective-diffusion equations be solved for the steady state

condition. The spinning disk for both techniques drags the laminar electrolyte fluid along

the electrode disk surface and due to the centrifugal force expels the solution outward

from the center in a radial direction. The expelled fluid is then replenished by a normal

flow to the surface of the disk electrode. The mathematical solution to this flow pattern

was solved using the cylindrical coordinate symmetry of the system.264 Once the velocity

profile solution was determined the convective-diffusion equation was solved using

reasonable boundary conditions. The first assumption was that the system was at steady

state. The second assumption was that the concentration of reactants far away from the

electrode is approximately equal to the concentration in the bulk. Other assumptions

were; that the radial velocity did not vary with radial distance and across the face of the

circular electrode and the concentration change of reactants in the radial direction is equal

to zero normal to the electrode surface. Upon solving the velocity profile with the

convective diffusion equation, the Levich equation was obtained;

2/3 1/2 1/6 * 25 il,c  0.62nFADO   CO.

 187

The limiting current of the disk at steady state is “il,c”. The current in the absence of mass-transfer effects is “iK”. The number of electrons transferred per reactant molecule is

“n”. Faraday‟s constant is “F”. The geometric area of the electrode is “A”. The diffusion coefficient of the reactant molecule through the electrolyte is “DO”. The rotation of the disk is “”. The viscosity of the electrolyte is “”. The reactant concentration in the bulk

* is “CO ”. The Levich equation applies to the totally mass-transfer-limited condition at the

* 1/2 RDE or RRDE and predicts that il,c is proportional to CO and  . Although the RDE technique is well established, there have been reports in the literature that the technique may not always give comparable results for selectivity when catalyst is coated on the disk.46,105,316,317 The behavior of porous of material coated on the rotating disk electrode is expected to alter the flow pattern. A survey of the literature shows that the catalyst loadings vary widely for both carbon-nitrogen and Pt-based catalysts using the rotating disk techniques. Catalyst loadings on the disk can vary from 70 g/cm2 318 to 4370

g/cm2 73, with the majority of the studies fall in the 100 g/cm2 to 1100 g/cm2 range.57,103,294,319-323 Studies examining porous electrodes have shown deviations from

Levich behavior with increasing catalyst layer thickness, which may be a contributing factor in the variation in RDE selectivity results.324-326 It is unlikely that the assumption that radial velocity does not vary with radial distance across the face of the circular electrode used to solve the Levich equation is satisfied when larger volumes of catalyst are deposited on the rotating disk. Recently, there have been reports that observed that the measured selectivity of the catalyst is effected with catalyst loading levels on the disk of an RRDE setup for carbon-nitrogen, Pt-based and chalcogenide ORR

188 catalysts.109,312,327,328 Interestingly, it was recently reported that the selectivity of Fe/N/C catalyst loading is dependent upon the specific preparation of the catalyst.109,120 To determine how catalyst loading affects selectivity, experimentally determined and calculated values were compared.

7.3 The Rotating Ring Disk Method of Calculating Selectivity

The high throughput RRDE half-cell technique is generally regarded as the best method for simultaneously testing the ORR activity and selectivity of the electrocatalysts directly. In this technique the disk electrode is coated with a dried catalyst and Nafion suspension to the desired loading. The coated disk electrode composition is similar to what is used in membrane electrode assemblies with Nafion acting as a binder and as proton conductor to effectively increase the electrochemical surface area. The catalyst-coated electrode is then placed into an oxygen saturated electrolyte solution. While the catalyst-coated disk is rotating, a cyclic voltammogram is performed to monitor the oxygen reduction activity of the catalyst, while a ring (typically

Pt) surrounding the disk is held at a constant potential. The ring at a high enough potential can oxidize hydrogen peroxide to water and can be used to determine selectivity. The two reactions that occur at the disk are the desired complete reduction of dioxygen to water:25

- + O2 + 4e + 4H  2H2O 1.23V vs. NHE (1) 189

And the undesired partial reduction of dioxygen to hydrogen peroxide:25

- + O2 + 2e + 2H  H2O2 0.695V vs. NHE (2)

The production of hydrogen peroxide is undesired not only because it decreases the efficiency due to a decreased number of electrons transferred per O2 molecule at a lower standard reaction potential, but also because it can corrode the components of the fuel cell.329

ORR Selectivity using the RRDE technique was determined by comparing the current produced on the catalyst-coated disk to the current collected on the ring. The disk current is a product of the complete and partial reduction of dioxygen to form H2O and

H2O2 on the catalyst surface, respectively. The rotation of the disk causes controlled convection of oxygen-saturated electrolyte to the disk and flow of products from the disk over the ring. The hydrogen peroxide in the products flows over the ring and is electrochemically oxidized creating a current that provides a measure of the hydrogen peroxide produced on the catalyst-coated disk. Only a fraction of the hydrogen peroxide produced on the disk is detected at the ring due to diffusion through the electrolyte. The fraction of hydrogen peroxide produce that can be reacted at the ring is known as the collection efficiency, N, of the RRDE technique. The selectivity of the electrocatalyst tested with s RRDE system is often reported as the number of electrons transferred per oxygen molecule, n. RRDE oxygen reduction selectivity is calculated using the following equation:

190

4I n  D . 263,264  I R  I D     N 

While n = 4 represents a catalyst with 100% selectivity to water formation, an n value of

2 is equivalent 100% selectivity to hydrogen peroxide formation. The steady-state disk current during oxygen reduction is “ID”. The steady-state ring current during oxygen reduction on the disk while the ring is held at a potential high enough to oxidized hydrogen peroxide is “IR”.

A typical ring and disk current for oxygen reduction for CNx and 20 wt.% platinum on Vulcan carbon are shown in Figure 55. The catalyst mass loading on the glassy carbon

Figure 55. The ring and disk currents in oxygen saturated 0.5 M H2SO4 of CNx (left) and 20wt%Pt on Vulcan carbon (right) (Insets: Selectivity of the oxygen reduction reaction)

191 disk for both experiments is the same, 426 µg/cm2, so it can be seen that the platinum on

Vulcan carbon has a higher mass activity than CNx grown over Fe/MgO due to the greater ORR activity current and higher take-off potential where the oxygen reduction current begins. It can also been seen that platinum appears more selective for the formation of water during oxygen reduction than CNx, but these selectivity calculations are based on the manufactures collection efficiency value, N = 0.22. Since both porous electrode materials are deposited on the grassy carbon electrode it is possible that the

Levich equation is not valid making the manufactures reported collection efficiency incorrect.

Experimental collection efficiencies were determined to correct for the possible error in collection efficiency with catalyst loading. A solution of 0.1 M NaOH electrolyte with 50 mM K3Fe(CN)6 was used to determine the experimental collection efficiency. A basic solution was used so that hydrogen cyanide would not be formed. The selectivity of the catalyst covered glassy carbon disk "n" was known to be 1 because the only

3- 4- reaction that can occur is the one electron transfer of [Fe(CN)6] to form [Fe(CN)6] . The collection efficiency of the ring "N" was calculated by comparing the ring current "IR", corrected by, to the disk current "ID" using the following equation:

1 I n  D I D  (I R / N)

The collection efficiency reported by the manufacturer for the model MT28 Series

ThinGap RRDE (Pine Research) with a 0.1642 cm2 glassy carbon disk and platinum ring

192 was N = 0.22. Figure 56 displays the catalyst coated disk with a loading of 426 µg/cm2

a) b)

3- Figure 56. Disk and ring currents as a function of disk potential for the [Fe(CN)6] 4- 2 /[Fe(CN)6] redox couple. Curves shown are for 426 μg/cm catalyst loading. a) CNx, b) 20 wt%Pt on Vulcan carbon.330 and ring currents for the reduction and oxidation of iron ferricyanide species, respectively. The iron ferricyanide disk and ring current densities and take-off potentials are fairly comparable for both catalysts, suggesting that CNx and 20 wt%Pt on Vulcan

3- carbon have comparable that the activity for [Fe(CN)6] reduction. The current of the

3- ring was non-zero at potentials higher than where [Fe(CN)6] reduction takes place, so this base capacitance current was taken as the background current when calculating the collection efficiency of the RRDE system. The collection efficiency of the RRDE system

193 was found to change with loading of catalyst material applied to the disk, Figure 57.

a)

b)

Figure 57. Experimental collection efficiency as a function of loading. The dashed line indicates the theoretical collection efficiency as reported by the manufacturer. a) CNx catalyst b) Pt/VC catalyst.330

There was a strong collection efficiency trend with catalyst loading for CNx. This can be attributed to CNx being less dense than 20 wt%Pt on Vulcan carbon, so for equal mass loadings the CNx catalyst would be more voluminous and affect the hydrodynamic flow more strongly. The collection efficiency for CNx at low catalyst loadings was higher than

0.22 value reported by the manufacturer, but at higher catalyst loadings the collection efficiency is consistently lower. Nearly all platinum on Vulcan carbon loadings had consistently higher collection efficiency values.

194

7.4 Rotating Disk Electrode Koutecky-Levich Selectivity Analysis

The rotating disk electrode technique can also determine electrochemical

activity/selectivity of catalysts. The RDE technique is similar to the RRDE technique but

does not have a ring that surrounds the disk for the direct determination of selectivity.

Alternatively, the Koutecky-Levich technique calculates selectivity by varying the

rotation rate of the disk through the following equation;

1 1 1 25 .   2/ 3 1/ 2 1/ 6 * i iK 0.62nFADO   CO

The total current of the disk is “i”. The current in the absence of mass-transfer effects, iK,  would be the current under kinetic limitation if the mass transfer were high enough to

keep the concentration at the electrode equal to the bulk at any rate of electrode reaction.

The number of electrons transferred per reactant molecule is “n”. Faraday‟s constant is

“F”. The geometric area of the electrode is “A”. The diffusion coefficient of the reactant

molecule through the electrolyte is “DO”. The rotation of the disk is “”. The viscosity of

* the electrolyte is “”. The concentration of the reactant in the bulk is “CO ”. Under

reaction conditions where RDE is used to determine selectivity the equation is simplified

to the Levich equation. To determine the selectivity of the catalyst material the rotation

195 rate of the disk is varied while the limiting current is measured, Figure 58. It can be seen

2 Figure 58. Disk currents in oxygen saturated 0.5 M H2SO4 of CNx with 426 μg/cm catalyst loading that the current increases with the disk rotation rate. The oxygen reduction cathodic potential scans in Figure 58 were not argon background subtracted so that capacitance changes with rotation rates do not confound the limiting current values. The transient region (0.8 – 0.4 V vs. NHE) of the cathodic scan shows overlapping current densities, which is likely due to the fluid flow effects of the catalyst-coated disk. For this reason

Levich analysis was performed on the lower potential region where the system is more

2/3 * likely to be at steady state. The 0.62FADO CO from the Levich equation can be grouped together as a constant while the n selectivity value is calculated from the slope,

Figure 59. The selectivity value calculated from the Levich equation, n = 3.42, was

196 similar to the selectivity value determined experimentally using the rotating ring disk electrode, n = 3.66. The literature values used for the grouped constants of the slope at the same experimental conditions used to determine selectivity can change considerably.

The Bron et al. and Pattabi et al. values were found to calculated selectivity values at n =

3.42 and n = 5.53, respectively.

Figure 59. Koutecky-Levich Analysis of CNx in Oxygen Saturated 0.5 M H2SO4 with 426 μg/cm2 catalyst loading. The slopes for 2 and 4 electron transfer for ORR were taken from:316,331

A selectivity value greater than 4 would mean that on average more than 4 electrons were transferred for every molecule of oxygen during the oxygen reduction reaction, which

197 cannot be rationalized chemically. This discrepancy is indicative of the uncertainty when using the Koutecky-Levich technique to determine selectivity.

7.5 Comparison of RDE and RRDE Selectivity

The RRDE technique is better suited for determining the selectivity of the oxygen reduction reaction of catalyst-coated disks than the RDE Koutecky-Levich analysis. The error in the collection efficiency values due to catalyst loading in the RRDE system is much less than the accumulated error caused in the grouped constants of the Koutecky-

Levich equation that are used to determine selectivity in the RDE technique. For the RDE technique the error associated with catalyst loading would be masked by the inherent error in the Koutecky-Levich equation making it difficult to separate effects. Although there was some change in the RRDE collection efficiency with catalyst loading the effects the propagated in error in the selectivity for the oxygen reduction reaction was very small.330

198

CHAPTER 8. Computational Chemistry

8.2 Research Motivation

Graphitic carbon-nitrogen compounds that are active for oxygen reduction could potentially replace or supplement expensive platinum catalysts in PEM and DMFC. The development of carbon-nitrogen carbon-based catalysts has been slow due to the difficultly characterizing blackbody materials with spectroscopic analytical techniques.137,332,333 The highly absorbing X-ray character of these materials makes it difficult to detect reactive intermediates and consequently the mechanism for which oxygen is reduces. This limits the intelligent design of more advanced carbon-nitrogen graphite ORR catalysts.

In order to elucidate electronic phenomena of nitrogen-doped graphites initial computational analyses were performed. To study the oxygen reduction activity of graphitic carbon-nitrogen compounds and the experimental XPS N 1s spectra computational calculations determined the IP and nitrogen 1s orbital energies of small

199

PAHs, respectively. PAHs are more computationally accessible than much larger graphitic carbon-nitrogen compounds, but the polycyclic aromatic bonding structure in

PAHs is similar to graphitic compounds. Electronic structure theory simulations were compared to experimental values in order to validate computational methods for small polycyclic hydrocarbons where experimental results were available. The validated methods could then be applied to larger systems that would be more like nitrogen-doped graphite compound used in oxygen reduction. Insight into the electronic structure of CNx catalysts though computational analysis could potentially aid in the design of enhanced carbon-based PEM fuel cell catalysts.

8.3 Introduction

The limited supply of petroleum has caused the transportation industry to search for other high-power density energy conversion devices such as the PEM and direct methanol fuel cells. Currently, these fuel cell technologies are limited by high material costs. A large component of this cost is allocated to the platinum catalyst, which is used to overcome the slow kinetics of the oxygen reduction reaction at the cathode of PEM and direct methanol fuel cells. Novel electro-catalytic carbon-based materials could offer a less expensive alternative cathode catalyst, which would aid in the mobilization of PEM and DMFC technology for automotive use. The ORR involves the reaction of oxygen molecules with protons and electrons to form water.

200

+ - 25 O2 + 4H + 4e  2H2O (1.229 V vs. NHE)

The above reaction is the preferred oxygen reduction reaction due to the high thermodynamic potential and consumption of electrons. The reaction above competes with the undesirable reaction that forms hydrogen peroxide below, which decreases fuel utilization efficiency by possessing a lower reaction potential and electrons consumed.

Furthermore, the hydrogen peroxide product in the reaction below could corrode fuel cell components decreasing the lifetime of the membrane electrode assembly. An ideal oxygen reduction cathode catalyst would have high selectivity for the formation of water and also be very catalytically active at low temperatures (< 100 C) and pressures (< 150 psig).3

+ - 25 O2 + 2H + 2e  H2O2 (0.695 V vs. NHE)

Even with the traditionally highly active platinum catalysts for ORR, the reaction is relatively slow and must overcome a large overpotential before kinetic barriers are reduced to have large amounts of dioxygen reduced. This effect thereby reduces the efficiency and maximum power of the fuel cell.

The development of non-precious metal ORR catalysts for fuel cell applications has been the focus of many studies for some time. Initial studies by Jansinski during the

1960‟s were inspired by the low-temperature oxygen adsorption in hemoglobin, where

201 oxygen was adsorbed on an iron center stabilized by graphitic nitrogen groups of a

334,335 macrocycle. These studies focused on the ORR activity of N4-chelates with cobalt and iron ions as the active centers. This class of catalysts was found to be active for only a limited duration in the PEM fuel cell due to decomposition via hydrolysis in the electrolyte or material degradation caused by peroxide intermediates.128 In order to stabilize the active macrocycle catalyst Jahnke et al. heat-treated macrocycle catalysts in an inert environment.50 This heat treatment not only improved the stability of the catalyst but also enhanced the activity. Generally, an active and stable catalyst could be made with a pyrolysis temperature from 400 C to 1000 C, which as a consequence would allow for the formation of many product types. To create a highly active catalyst, it was found that both nitrogen-containing macrocycles and a conductive support (long chained graphitic carbon) must be present during the pyrolysis.59,70,336 The increased activity after pyrolysis was attributed to the formation of more electrically connected active sites through bonding active macrocycles to conductive carbon support.64 What is of interest is that the metal component didn‟t necessarily have to be stabilized by the macrocycle to create a highly active catalyst.336 In opposition to the notion that the macrocycle catalyst site remains unaffected, many researchers believe that the active macrocycle center is completely destroyed during pyrolysis and have used TEM imaging, 68 Mössbauer spectroscopy,54 and XAS 248 to confirm a change in the oxidation state and coordination of the metal species. To depart further form macrocycle precursors, Gouerec et al. studied heat treated cobalt porphyrins for ORR and observed the same activity effects of heat treatment when using macrocyclic precursors as when using their simple elemental constituents; nitrogen, carbon and metal together.62

202

a) b)

c) d)

Figure 60.. Purposed causes of ORR catalytic activity in graphitic carbon-nitrogen catalysts; a) Stelko,172,177,337 b) Lefévre, 92 c) Montoya,261 Pels,256, d) Subramanian 169 (figures b), c), and d) courtesy of Paul H. Matter)

Unfortunately the high-temperature treatment that was necessary to create ORR catalytically active carbon-nitrogen catalysts confounded the cause of catalytic activity.

Figure 60 depicts some of the theories regarding ORR activity in these materials. The temperatures used to form active material allow for the formation of any of these types of active site, which is why there is much disagreement in the active site of CNx catalysts. It

203 is widely known in that the nitrogen in carbonaceous materials affects the catalytic activity of ORR,3 but there is some disagreement as to how nitrogen contributes to oxygen reduction. There are two main arguments for how nitrogen enhances ORR in

CNx materials; one in which a metal is supported by the graphitic nitrogens such as in the metalloporphyrins or the second in which the CNx material has intrinsic catalytic activity by its own accord. Our heterogeneous research group has shown ORR activity in a transition-metal-free (< 5 ppm) CNx catalyst suggesting that there is either multiple active sites, one of which does not have a coordinated metal, or that the coordinated metal active site is exceptionally active.132

Figure 61.. Types of carbons that are commonly de-convoluted in XPS

204

Further inquiry into the mystery surrounding the ORR activity of CNx and the more recent discovery of a metal-free CNx catalyst could be enhanced by a computational chemistry study of potential reaction mechanisms. Since our hypothesized active site would not have a coordinated metal, less computationally expensive methods were used to simulate materials that would have similar properties to CNx. These materials were selected so that they would have verifiable experimental values. Most density functional theory (DFT) methods are optimized for a series of small molecules where detailed experimental data are available, which would have similar bonding structures to CNx catalysts.338 Furthermore, due to the proliferation of DFT methods there was a precedent

172,259,339-341 of investigating the activity of CNx catalysts. The active site described by

Strelko et al. shows ORR activity proceeding through electron donation from CNx, shown in Figure 60 a). 177 The phenomena of electron donation is readily accessible for PAHs with and without nitrogen incorporation by examining the experimental IPs and computational by comparing the total energy difference in the neutral and charges states.

These PAHs have a known structure and have a similar electronic structure as CNx catalysts will be used to for this preliminary investigation. Computationally determined ionization potentials were used to predict the reactivity of persistent carbon-centered radicals with dioxygen, so as a natural extension ionization potentials of PAHs will preliminary explored for oxygen reduction activity. Similarly, IPs as they relate ORR activities could also be explored for the Montoya et al. pyridinic N+O- and Surbramanian et al. pyridinic-quarteranary graphitic nitrogen type structures, shown in Figure 60 d).

205

Figure 62. XPS N-1s spectra of CNx catalyst (Courtesy of Paul H. Matter)

Experimental XPS nitrogen 1s spectra can be de-convoluted to determine the nitrogen species on the surface of nitrogen-containing carbon graphites and PAHs. Many researchers have used XPS data to qualitatively characterize their carbon-nitride materials.193,210,223,228,257,342 The problem that arises with using XPS to quantify differences in nitrogen types is that the experimental de-convolution of the nitrogen 1s peak is not definitive, see Figure 61. There is narrow range of a few electron volts that differentiates the nitrogen 1s peak positions and their corresponding nitrogen types,

Figures 61 and 62, which in turn makes the collection and interpretation of these peaks prone to error. Even the number of peaks that are de-convoluted for the nitrogen 1s XPS spectra can range from three to six.223,343,344

206

Fundamental studies have investigated the energies N 1s species of carbon materials. Titantah et al. used DFT calculations on a series of amorphous carbon nitride structures generated by Monte Carlo methods to develop the nitrogen 1s energy spectra that would be expected in experimental XPS from first principals.344 Furthermore,

Gammon et al. compared the experimental XPS N 1s species assignments to amorphous carbon nitride films with reference organic compounds and found that nitrogen species could be separated into four energetic categories; 398.5 eV, 400.6 eV, 402.5 eV, and

223 404.8 eV. The highest energy at 404.8 eV does not appear on CNx materials so this will not be considered in this study.133,211 Since the most non-precious metal ORR catalyst activity site theories involve different types of nitrogen incorporated within the graphitic carbon, it is of paramount importance know where the nitrogens are located for theory validation and corresponding catalyst enhancement.

8.4 Determination of Ionization Potential

A computation study of CNx catalysts investigated the unique electronic phenomena of polycyclic aromatic hydrocarbon materials to elucidate electronic effects of nitrogen doping. These PAHs were chosen to represent CNx materials because they are computationally accessible, posses a similar bonding configuration, and have relevant experimental data available. For dioxygen reduction over CNx catalysts electron donation was postulated as likely the rate-determining step and likely to be linked to catalytic activity.172 Experimental ionization values are readily available for small

207 polycyclic aromatic hydrocarbons, with and without hetroatom nitrogen incorporation, and were used to substantiate the compiled computational data.

There are two methods to computationally calculate the ionization potential of a compound, Figure 63. Both methods start by optimizing the PAH‟s

Figure 63. Illustration of the Difference in Vertical and Adiabatic Ionization Potentials geometry in the ground, closed, neutral sate. The total energy optimization of these molecules was accomplished with the structures starting in planar or Cs symmetry to reduce the amount of computational resource units required. Since the CNx catalyst materials are highly graphitic and stable at low temperatures it is most likely that the structures are relatively planar and will be well represented by energy optimized PAHs.

To calculate the vertical ionization potential of a PAH the optimized geometry in the neutral state was used to calculate the energy of the compound in the positively charged state with the ionization potential being the total energy difference in the positively

208 charged state and the neutral state. The vertical ionization potential calculation assumes that structure did not charge orientation once in a higher charged state. This assumption may or may not be valid, but can be compared to experimental IP values. To account possible rearrangement of a charged molecule an adiabatic ionization potential can be calculated. An adiabatic ionization potentials was calculated by taking the difference of the re-optimized geometry in the positively charged state and optimized neutral state. By re-optimizing the structure in the charged the state the adiabatic ionization potential accounts for any rearrangements that may take place, but because of this re-optimization of the molecules geometry in the open plus changed state the adiabatic ionization potential was much more computationally expensive than a vertical ionization potential calculation.

To obtain an adiabatic and vertical ionization potential the neutral state energy, cationic state energy with neutral state geometry, and cationic state energy with an optimized geometry were computed. These computations were first completed at the

Austin Model 1 level of theory, Table 4.

Table 4. Computationally determined energies of compounds at AM1 level of theory

Neutral charge Adiabatic +1 Vertical +1 AM1 1st AM1 1st AMI 1st

vibrational vibrational vibrational

corrected corrected corrected

Chemical name Molecule (eV) (eV) (eV)

Pyridine C5H5N 3.727052623 13.01753221 13.39165433 1,2-Diazine C4H4N2 4.423801047 14.10161384 14.39487066 1,3-Diazine C4H4N2 3.924804491 13.60695117 13.85593366 1,4-Diazine C4H4N2 3.938025703 13.51522151 13.83881234

209 Continued

Table 4 Continued s-Triazine C3H3N3 4.214146465 14.48195716 14.7984015

1,2,3-Triazine C3H3N3 5.24328492 15.71667445 15.46664609 1,2,4-Triazine C3H3N3 4.719085088 14.64394882 14.91892128 s-Tetrazine C2H2N4 5.609390074 15.95185015 16.24575528 Azulene C10H8 7.53810364 14.6057937 14.85265623 Naphthalene C10H8 5.686273051 13.72621735 13.82908006 Quinoline C9H7N 5.878308712 14.80792566 15.21331641 Isoquinoline C9H7N 5.78837118 14.12800867 15.21935588 1,2-Diazanaphthalene C8H6N2 6.524677329 15.78361104 16.14606789 1,3-Diazanaphthalene C8H6N2 6.037810359 14.76880139 14.96126802 1,4-Diazanaphthalene C8H6N2 6.145103248 14.90581416 15.65241261 1,5-Diazanaphthalene C8H6N2 6.064506996 15.02072007 15.15117279 1,6-Diazanaphthalene C8H6N2 5.991996606 15.24146159 15.64076589 1,7-Diazanaphthalene C8H6N2 5.991143651 15.23842359 14.90886205

1,8-Diazanaphthalene C8H6N2 6.286012491 15.29670429 15.64518527 2,3-Diazanaphthalene C8H6N2 6.455428765 15.28592912 16.14679834 2,6-Diazanaphthalene C8H6N2 5.931149039 14.54893532 14.73963663 2,7-Diazanaphthalene C8H6N2 5.916189275 14.62026824 14.84425387 2,2'-Bipyridine C10H8N2 7.339081538 15.9454314 16.22423772 2,2'-Bipyrimidine C8H6N4 7.880914179 17.31174509 17.58280832 Acenaphtylene C12H8 7.777681081 15.55855278 15.7813218 Biphenylene C12H8 9.452791434 17.20837381 17.34191831 Fluorene C13H10 7.372251115 15.25183388 15.42521413 Pyrido[2,1,6-de]quinolizine C12H9N 8.971738362 15.06208944 15.2718183 Pyrimidine, 4-phenyl- C10H8N2 7.321315846 16.67845953 16.92170381 4,4'-Bipyridine C10H8N2 7.14122462 16.21019554 16.48685126 4,4'-Bipyrimidine C8H6N4 7.85785301 17.8573629 17.69653527

4,5'-Bipyrimidine C8H6N4 7.604322758 17.20421322 17.53722146 5,5'-Bipyrimidine C8H6N4 7.487671825 16.52380412 16.82761893 Anthracene C14H10 7.920404048 15.34434097 15.46771841 Phenanthrene C14H10 7.689090641 15.4974269 15.66552243 Acridine C13H9N 8.224172388 16.69191196 17.15534909 Phenanthridine C13H9N 7.836092521 16.02372455 17.15532188 Benzo[h]quinoline C13H9N 7.84534447 15.79898691 15.96260273 Benzo[f]quinoline C13H9N 7.885100865 15.96648805 17.08430558 6,7-Benzoquinoline C13H9N 8.113237435 15.78234731 15.91127619 5,6-Benzoisoquinoline C13H9N 7.782809553 15.80399519 15.98295091 Phenazine C12H8N2 8.593665686 17.51280579 16.88231091 Benzo[c]cinnoline C12H8N2 8.492093642 17.11881173 17.9146535 o-Phenanthroline C12H8N2 8.130837677 16.83039549 16.48656916 4,7-Phenanthroline C12H8N2 8.092860305 17.04104277 18.25653432

Diazene,diphenyl-,(E)- C12H10N2 9.422762905 17.23942578 17.75713763 Benzenamine, N-(phenylmethylene)- C13H11N 8.591748133 16.14562841 16.78140209 Pyrene C16H10 8.487385062 15.89298404 16.03126928 Fluoranthene C16H10 9.354559369 17.07734592 - 4-Azapyrene C15H9N 8.658727336 16.32592714 16.47970264 2-Azapyrene C15H9N 8.589471599 16.17129684 16.32353264 1-Azapyrene C15H9N 8.698778513 16.41859187 16.57390861 Tetracene C18H12 10.22406781 17.19611446 17.31492612 Chrysene C18H12 9.773963405 17.29451958 17.46770449 210 Continued

Table 4 Continued

Benz[a]anthracene C18H12 9.85653012 17.25685625 -

Triphenylene C18H12 9.73972408 17.5504759 18.81340238 Benzo[c]phenanthrene C18H12 9.996695758 17.58654753 17.67374257 3,4-Benzacridine C17H11N 10.15390907 18.75727228 - 1,2-Benzacridine C17H11N 10.15384285 18.80941553 - 1-Azabenz(a)anthracene C17H11N 10.01094245 17.46203596 17.59891135 4-Azachrysene C17H11N 9.867400476 17.64447334 17.84285966 1-Azachrysene C17H11N 9.968927305 17.70023761 25.54967798 2-Azachrysene C17H11N 9.86740557 17.59212112 17.77700698 Corannulene C20H10 13.07705151 20.82281791 - Benzo[a]pyrene C20H12 10.63148421 17.70533348 - Benzo[e]pyrene C20H12 10.4784592 17.91459258 18.04622671 Perylene C20H12 10.70935859 17.72428342 - 10-Azabenzo[a]pyrene C19H11N 10.83976195 18.17630637 19.71303182

Anthanthrene C22H12 11.5152646 18.34635023 - Benzo[g,h,i]perylene C22H12 11.16843739 18.38910898 - Pentacene C22H14 12.56397227 - - Pentaphene C22H14 11.99553828 - - Picene C22H14 11.82982979 19.18159857 - Benzo[b]chrysene C22H14 11.969486 19.14608798 - Dibenz[a,j]anthracene C22H14 11.82696959 - - Benzo[b]triphenylene C22H14 11.87566017 19.31102796 19.49254798 Dibenzo[c,g]phenanthrene C22H14 12.22740615 19.66465195 - Dibenz[a,h]anthracene C22H14 11.81870096 19.12509305 19.30238978 Benzo[a]naphthacene C22H14 12.12692681 19.06588822 19.60700794 Coronene C24H12 11.75467117 19.05704089 - Dibenzo[b,def]chrysene C24H14 12.83599628 - 19.76544999

Dibenzo[cd,lm]perylene C26H14 13.51709049 - 20.29171301 Hexacene C26H16 12.56397227 - - Bisanthene C28H14 14.68834858 20.8975148 21.10101961 Benzo[a]coronene C28H14 13.83841477 21.00787286 - Dibenzo[bc,kl]coronene C30H14 15.37119722 21.4265305 - Dibenzo[bc,ef]coronene C30H14 14.96615149 21.5709031 - Terrylene C30H16 15.72369059 - 22.32516104 Ovalene C32H14 15.38524139 22.04275137 22.26180589 Tetrabenzo[bc,ef,kl,no]coronene C36H16 18.75337187 24.01988112 - Circumbiphenyl C38H16 17.96048494 - - Circumanthracene C40H16 20.72798678 26.98556884 27.24430463 Quaterrylene C40H20 20.73510781 - - Circumpyrene C42H16 - - - Hexabenzocoronene C42H18 20.1979062 - -

Dicoronylene C48H20 22.8816082 - - Pentarylene C50H18 25.74484574 - 31.4239061 Circumcoronene C54H18 32.5012491 - - Circumovalene C66H20 35.82382743 - 68.95993497

211

It can be seen that the total energy of the system increases with the number of heavy atoms present, nitrogen and carbon. The ionized species have higher total energies than the neutral molecules with the vertical ionization potentials consistently higher than the adiabatic ionization potentials. Approximately, ten-percent of the molecules had an imaginary vibrational states and were re-optimized to a stable ground state. The re- optimization of imaginary frequencies was necessary to ensure that the states of the molecules were representative of the actual structures used in experiments. The calculated energies are very fairly consistent with experimental values due to the development of AM1 theory. Dewar modified the modified neglect of differential overlap

(MNDO) method to calculate nuclear repulsion energies between opposing nuclei by using Gaussian functions centered at various distances for C, H, O, and N atoms to modify the potential mean force between two atoms to account for the poor prediction of hydrogen bonding to form AM1.338 The AM1 molecular orbital model is considered reasonable robust over a large range of chemical functionality and is computational inexpensive so it was used for the initial molecular analysis. Unfortunately, the AM1 model is known to have average error in the ionization potential for organic compounds of 0.6 eV.338 As seen by the empty vertical and adiabatic total energy values empty values in Table 3, where repeated convergence errors had occurred. In addition, the AM1 model parameter optimization was accomplished in a stepwise manner, potentially accumulating errors, with a limited parameter space and parameter selection.338

To get more accurate IP values the Becke (three-parameter) Lee-Yang-Parr exchange-correlation functional (B3LYP) computational method was also used. The

B3LYP method uses exchange-correlation fuctionals that not only account for the

212 difference between the classical and quantum mechanical electron-electron repulsion, but also include the kinetic energy differences between fictitious non-interacting system and the desired system of interest.338 This means that the B3LYP model optimizes with respect to the electron density whereas AM1 molecular theory optimizes the electron wave function and as a consequence the B3LYP method is generally better at dealing with open shell systems, such as positively charged PAHs.338 Table 5 displays the calculated neutral, vertically ionized,

Table 5. Computationally determined energies of compounds at B3LYP/6-31G* level of theory

Neutral charge Adiabatic +1 Vertical +1 B3LYP/6-31G* B3LYP/6-31G* B3LYP/6-

1st vib 1st vib 31G*1st vib

corrected corrected corrected

Chemical name Molecule (eV) (eV) (eV)

Pyridine C5H5N -6754.001386 -6745.180452 -6744.535746 1,2-Diazine C4H4N2 -7189.76458 -7181.44057 -7180.912271 1,3-Diazine C4H4N2 -7190.755933 -7181.810485 -7181.388175 1,4-Diazine C4H4N2 -7190.58482 -7181.701954 -7181.376932 s-Triazine C3H3N3 -7627.570705 -7617.899752 -7617.312325 1,2,3-Triazine C3H3N3 -7625.756065 -7616.776244 -7616.177659 1,2,4-Triazine C3H3N3 -7626.439393 -7617.717603 -7617.22356 s-Tetrazine C2H2N4 -8062.14339 -8053.211831 -8052.682458 Azulene C10H8 -10495.59381 -10488.66371 -10488.53679 Naphthalene C10H8 -10497.04544 -10489.45769 -10489.3688 Quinoline C9H7N -10933.78195 -10925.68977 -10925.60885

Isoquinoline C9H7N -10933.73343 -10925.72182 -10925.62415 1,2-Diazanaphthalene C8H6N2 -11369.53957 -11361.68156 -11360.48911 1,3-Diazanaphthalene C8H6N2 -11370.54888 -11362.02889 -11361.79889 1,4-Diazanaphthalene C8H6N2 -11370.41984 -11361.92542 -11361.64272 1,5-Diazanaphthalene C8H6N2 -11370.50893 -11362.20901 -11361.33217 1,6-Diazanaphthalene C8H6N2 -11370.44109 -11361.92018 -11361.82596 1,7-Diazanaphthalene C8H6N2 -11370.44397 -11361.92722 -11361.81439 1,8-Diazanaphthalene C8H6N2 -11370.44382 -11361.65442 -11361.83476 2,3-Diazanaphthalene C8H6N2 -11369.59509 -11361.6579 -11361.1406

213 Continued

Table 5 Continued

2,6-Diazanaphthalene C8H6N2 -11370.35434 -11361.96581 -11361.85265

2,7-Diazanaphthalene C8H6N2 -11370.3815 -11361.88757 -11361.76928 2,2'-Bipyridine C10H8N2 -13476.05155 -13468.15744 -13467.85432 Dipyrido[1,2-b:1',2'-e][1,2,4,5- C10H8N4 -16453.38698 -16447.57871 -16447.57871 2,2'-Bipyrimidine C8H6N4 -14349.53323 -14341.47343 -14341.2496 ]tetrazine Acenaphtylene C12H8 -12570.16208 -12562.73433 -12562.58082 Biphenylene C12H8 -12568.66286 -12561.64854 -12561.49742 Fluorene C13H10 -13639.79714 -13632.45008 -13632.3195 Pyrido[2,1,6-de]quinolizine C12H9N -14106.65965 -14099.20867 -14100.67858 Pyrimidine, 4-phenyl- C10H8N2 -13476.29031 -13468.02527 -13467.59079 4,4'-Bipyridine C10H8N2 -13476.10184 -13467.51441 -13467.38853 4,4'-Bipyrimidine C8H6N4 -14349.54024 -14341.09349 -14340.74772 4,5'-Bipyrimidine C8H6N4 -14349.68584 -14341.04081 -14340.82069 5,5'-Bipyrimidine C8H6N4 -14349.57404 -14340.78283 -14340.70315

Anthracene C14H10 -14676.61485 -14669.799 -14669.73228 Phenanthrene C14H10 -14676.82842 -14669.53177 -14669.44714 Acridine C13H9N -15113.40379 -15106.13616 -15106.06965 Phenanthridine C13H9N -15113.59315 -15105.8679 -15105.81501 Benzo[h]quinoline C13H9N -15113.65701 -15106.16442 -15106.07986 Benzo[f]quinoline C13H9N -15113.57536 -15105.97969 -15105.90242 6,7-Benzoquinoline C13H9N -15113.35128 -15106.22321 -15106.14772 5,6-Benzoisoquinoline C13H9N -15113.53933 -15105.95613 -15105.85287 Phenazine C12H8N2 -15550.09306 -15542.4274 -15542.36596 Benzo[c]cinnoline C12H8N2 -15549.3947 -15541.81789 -15541.37238 o-Phenanthroline C12H8N2 -15550.18931 -15542.3857 -15542.29176 4,7-Phenanthroline C12H8N2 -15550.30623 -15542.32896 -15542.28413 Diazene,diphenyl-,(E)- C12H10N2 -15580.99583 -15573.74022 -15572.97689

Benzenamine, N-(phenylmethylene)- C13H11N -15145.02004 -15137.64193 -15137.50757 Pyrene C16H10 -16750.98481 -16744.12856 -16744.05339 Fluoranthene C16H10 -16750.37567 -16743.12795 -16743.07888 4-Azapyrene C15H9N -17187.70421 -17180.5357 -17180.44138 2-Azapyrene C15H9N -17187.65746 -17180.5357 -17180.45703 1-Azapyrene C15H9N -17187.71997 -17180.44659 -17180.37047 Tetracene C18H12 -18856.12012 -18849.82741 -18849.77458 Chrysene C18H12 -18855.72735 -18849.60306 -18849.53001 Benz[a]anthracene C18H12 -18856.46685 -18849.70723 -18849.64251 Triphenylene C18H12 -18856.51619 -18849.28928 -18848.68155 Benzo[c]phenanthrene C18H12 -18856.2772 -18849.31074 -18849.24644 3,4-Benzacridine C17H11N -19293.35031 -19286.25621 -19286.18436 1,2-Benzacridine C17H11N -19293.26641 -19286.1084 -19286.04562 1-Azabenz(a)anthracene C17H11N -19293.29077 -19286.47028 -19286.132

4-Azachrysene C17H11N -19293.23559 -19285.99252 -19285.92272 1-Azachrysene C17H11N -19293.29079 -19286.0979 -19285.98295 2-Azachrysene C17H11N -19293.25372 -19286.04434 -19285.92166 2,2':6',2"-Terpyridine C15H11N3 -20198.41792 - - 2,2'-Biquinoline C18H12N2 -21835.92969 -21828.58139 -21828.58139 9,10-Anthracenedicarbonitrite C18H8N2 -19696.88764 - - Corannulene C20H10 -20896.80229 - -20889.49299 Benzo[a]pyrene C20H12 -20930.62894 -20924.1313 -20924.06206 Benzo[e]pyrene C20H12 -20938.87831 -20923.8965 -20923.82828

214 Continued

Table 5 Continued

Perylene C20H12 -20930.42084 -20924.05254 -20923717484

10-Azabenzo[a]pyrene C19H11N -21367.37082 -21360.51137 -21360.44498 Anthanthrene C22H12 -23004.70389 -22998.44675 -22998.39076 Benzo[g,h,i]perylene C22H12 -23004.95162 -22998.37645 -22998.31178 Pentacene C22H14 -23035.59594 -23029.67851 -23029.63553 Pentaphene C22H14 -23036.13823 -23029.53549 -23029.46008 Picene C22H14 -23036.28049 -23029.48518 -23029.40779 Benzo[b]chrysene C22H14 -23036.1615 -23029.63819 -23029.58224 Dibenz[a,j]anthracene C22H14 -23036.28696 -23029.55558 -23029.50621 Benzo[b]triphenylene C22H14 -23036.18071 -23029.42072 -23029.3649 Dibenzo[c,g]phenanthrene C22H14 -23035.7625 -23028.95935 -23028.90867 Dibenz[a,h]anthracene C22H14 -23036.29766 -23029.61045 -23029.54231 Benzo[a]naphthacene C22H14 -23036.00384 -23029.71854 -23029.66592 Coronene C24H12 -25079.34989 -25072.62471 -25072.58401

Dibenzo[b,def]chrysene C24H14 -25110.22281 -25104.04614 -25103.98672 2,4,6-Tris(2-pyridyl)-s-triazine C18H12N6 -27794.15688 -27786.294 -27786.294 Dibenzo[cd,lm]perylene C26H14 -27184.48566 -27178.38536 -27178.32323 Hexacene C26H16 -23035.59594 -23029.67851 -23029.63553 Bisanthene C28H14 -29258.04026 -29252.38051 -29252.33856 Benzo[a]coronene C28H14 -29259.00795 -29252.48812 -29252.44768 Dibenzo[bc,kl]coronene C30H14 -31332.56954 -31326.76688 -31326.72277 Dibenzo[bc,ef]coronene C30H14 -31332.78571 -31326.87045 -31326.82481 Terrylene C30H16 -31363.86968 -31358.08613 -31358.02715 Ovalene C32H14 -33407.37203 -33401.22877 -33401.18432 Tetrabenzo[bc,ef,kl,no]coronene C36H16 - -37580.4471 -37580.42726 Circumbiphenyl C36H16 -39661.43169 -39655.20541 -39655.20541 Quaterrylene C40H20 -41797.34297 -41791.91966 -41791.86505

adiabatically ionized total energy of the PAHs with the B3LYP/6-31G* level of theory.

It is apparent is that the reference energies of AM1 and B3LYP are different. AM1 energetically takes into account the nuclei, interelectronic and internuclear repulsions.

Whereas, B3LYP density functional theory uses the effective nuclear point charges and electron density functionals of the molecule to determine energy of the system.

Therefore, energies determined by B3LYP should be taken with reference to comparable state of interest, Table 6. It should be noted that because B3LYP optimizes with regards to electron density many more of the vertical and adiabatic cationic states were computed without continuity errors.

215

Table 6. Ionization Energies of Various Polycyclic Aromatic Hydrocarbons

HF/6-31G AM1 first B3LYP/6-31+G* AM1 vibartional frequency B3LYP/6-31+G* (lin.Scaled) (lin.Scaled) IP (eV) first vibartional Experimental frequency IP (eV) IP (eV) IP (eV) IP (eV) IP (eV)

Chemical name Formula Adiabatic Vertical Adiabatic Adiabatic Adiabatic Adiab. Vert. Adiab. Vert. 345 Pyridine C5H5N - - - - 9.260.01 9.29 9.66 8.82 9.47 345 1,2-Diazine C4H4N2 - - - - 8.740.11 9.68 9.97 8.32 8.85 346 1,3-Diazine C4H4N2 - - - - 9.29 9.68 9.93 8.95 9.37 347 1,4-Diazine C4H4N2 - - - - 9.36 9.58 9.90 8.88 9.21 348 s-Triazine C3H3N3 - - - - 9.8 10.27 10.58 9.67 10.26 349 1,2,3-Triazine C3H3N3 - - - - 9.3 10.47 10.22 8.89 9.58 350 1,2,4-Triazine C3H3N3 - - - - 9.2 9.92 10.20 8.72 9.22 346 s-Tetrazine C2H2N4 - - - - 9.72 10.34 10.64 8.93 9.46 351 351 345 Azulene C10H8 7.15 7.27 - - 7.420.02 7.07 7.31 6.93 7.06 351 351 352 352 3 Naphthalene C10H8 7.80 7.89 7.99 7.98 8.1440.001 8.04 8.14 7.59 7.68 345 Quinoline C9H7N - - - - 8.630.02 8.93 9.34 8.09 8.17 45 345 Isoquinoline C9H7N - - - - 8.53 0.02 8.34 9.43 8.01 8.11 347 1,2-Diazanaphthalene C8H6N2 - - - - 8.51 9.26 9.62 7.86 9.05 347 1,3-Diazanaphthalene C8H6N2 - - - - 9.02 8.73 8.92 8.52 8.75 347 1,4-Diazanaphthalene C8H6N2 - - - - 8.99 8.76 9.51 8.49 8.78 353 1,5-Diazanaphthalene C8H6N2 - - - - 8.8 8.96 9.09 8.30 9.18 353 1,6-Diazanaphthalene C8H6N2 - - - - 9 9.25 9.65 8.52 8.62 353 1,7-Diazanaphthalene C8H6N2 - - - - 8.99 9.25 8.92 8.52 8.63 353 1,8-Diazanaphthalene C8H6N2 - - - - 8.8 9.01 9.36 8.79 8.61 353 2,3-Diazanaphthalene C8H6N2 - - - - 8.4 8.83 9.69 7.94 8.45 353 2,6-Diazanaphthalene C8H6N2 - - - - 8.87 8.62 8.81 8.39 8.50 Continued

216

353 2,7-Diazanaphthalene C8H6N2 - - - - 8.98 8.70 8.93 8.49 8.61 354 2,2'-Bipyridine C10H8N2 - - - - 8.6 8.61 8.89 7.89 8.20 355 Dipyrido[1,2-b:1',2'-e][1,2,4,5- C10H8N4 - - - - 6.10 - - 5.81 5.81 ]tetrazine

356 2,2'-Bipyrimidine C8H6N4 - - 8.3 9.43 9.70 8.06 8.28 351 351 357 Acenaphthylene C12H8 7.66 7.86 - - 8.220.04 7.78 8.00 7.43 7.58 351 351 345 Biphenylene C12H8 7.22 7.37 - - 7.580.03 7.76 7.89 7.01 7.17 351 351 352 352 345 Fluorene C13H10 7.56 7.69 7.94 7.98 7.910.02 7.88 8.05 7.35 7.48 358 Pyrido[2,1,6-de]quinolizine C12H9N - - - - 5.87 6.09 6.30 7.45 5.98 359 Pyrimidine, 4-phenyl- C10H8N2 - - - - 8.65 9.36 9.60 8.27 8.70 360 4,4'-Bipyridine C10H8N2 - - - - 9.10.02 9.07 9.35 8.59 8.71 356 4,4'-Bipyrimidine C8H6N4 - - - - 9 10.00 9.84 8.45 8.79 356 4,5'-Bipyrimidine C8H6N4 - - - - 9 9.60 9.93 8.65 8.87 356 5,5'-Bipyrimidine C8H6N4 - - - - 9 9.04 9.34 8.79 8.87 351 351 352 352 3 Anthracene C14H10 7.02 7.09 7.33 7.32 7.4390.006 7.42 7.55 6.82 6.88 351 351 352 352 3 Phenanthrene C14H10 7.53 7.63 7.89 7.88 7.8910.001 7.81 7.98 7.30 7.38 45 345 Acridine C13H9N - - - - 7.8 8.47 8.93 7.27 7.33 45 361 Phenanthridine C13H9N - - - - 8.31 0.02 8.19 9.32 7.73 7.78 361 Benzo[h]quinoline C13H9N - - - - 8.040.02 7.95 8.12 7.49 7.58 361 Benzo[f]quinoline C13H9N - - - - 8.140.02 8.08 9.20 7.60 7.67 362 6,7-Benzoquinoline C13H9N - - - - 7.60.1 7.67 7.80 7.13 7.20 362 5,6-Benzoisoquinoline C13H9N - - - - 8.30.1 8.02 8.20 7.58 7.69 361 Phenazine C12H8N2 - - - - 8.33 8.92 8.29 7.67 7.73 361 Benzo[c]cinnoline C12H8N2 - - - - 7.9 8.63 9.42 7.58 8.02 361 o-Phenanthroline C12H8N2 - - - - 8.3 8.70 8.36 7.80 7.90 361 4,7-Phenanthroline C12H8N2 - - - - 8.350.02 8.95 10.16 7.98 8.02 363 Diazene,diphenyl-,(E)- C12H10N2 - - - - 8.50.05 7.82 8.33 7.26 8.02 364 Benzenamine, N- C13H11N - - - - 8 7.55 8.19 7.38 7.51 (phenylmethylene)-

351 351 352 352 3 Pyrene C16H10 7.07 7.14 7.37 7.33 7.4260.001 7.41 7.54 6.86 6.93 352 352 345 Fluoranthene C16H10 - - 7.88 7.89 7.90.1 7.72 7.25 7.30 45 4-Azapyrene C15H9N - - - - - 7.67 7.82 7.17 7.26 Continued

217

Table 6 Continued

2-Azapyrene C15H9N - - - - - 7.58 7.73 7.12 7.20 1-Azapyrene C15H9N - - - - - 7.72 7.88 7.27 7.35 352 352 Benzo[b]fluorene C17H12 - - 7.67 7.72 - - - - - 352 352 Benzo[a]fluorene C17H12 - - 7.59 11.26 - - - - - 352 352 Benzo[ghi]fluoranthene C18H10 - - 7.99 7.97 - - - - - 352 352 Cyclopenta[cd]pyrene C18H10 - - 7.99 7.97 - - - - - 351 351 352 352 345 Tetracene C18H12 6.49 6.55 6.89 6.90 6.970.05 6.97 7.09 6.29 6.35 351 351 352 352 345 Chrysene C18H12 7.17 7.25 7.63 7.60 7.60.01 7.52 7.69 6.12 6.20 351 351 352 352 345 Benz[a]anthracene C18H12 6.98 7.05 7.42 7.41 7.450.05 7.40 6.76 6.82 351 351 352 352 345 Triphenylene C18H12 7.52 7.86 7.94 7.92 7.870.02 7.81 9.07 7.23 7.83 352 352 365 Benzo[c]phenanthrene C18H12 - - 7.64 7.69 7.6 7.59 7.68 6.97 7.03 362 3,4-Benzacridine C17H11N - - - - 8.10.1 8.60 - 7.09 7.17 362 1,2-Benzacridine C17H11N - - - - 8.10.1 8.66 - 7.16 7.22 1-Azabenz(a)anthracene C17H11N - - - - - 7.45 7.59 6.82 7.16 4-Azachrysene C17H11N - - - - - 7.78 7.98 7.24 7.31 1-Azachrysene C17H11N - - - - - 7.73 15.58 7.19 7.31 2-Azachrysene C17H11N - - - - - 7.72 7.91 7.21 7.33 2,2'-Biquinoline C18H12N2 ------7.35 7.35

9,10-Anthracenedicarbonitrite C18H8N2 ------7.31 351 351 345 Corannulene C20H10 7.62 7.73 - - 7.830.02 - - 6.50 6.57 351 351 352 352 345 Benzo[a]pyrene C20H12 6.71 6.78 7.12 7.09 7.120.01 7.44 7.57 - - 351 351 352 352 345 Benzo[e]pyrene C20H12 7.05 7.12 7.45 7.43 7.430.4 7.01 - 6.37 - 351 351 352 352 345 Perylene C20H12 6.57 6.64 7.00 7.02 6.960.001 - - - - 352 352 Benzo[k]fluoranthene C20H12 - - 7.47 7.52 - 7.34 8.87 6.86 6.93 10-Azabenzo[a]pyrene C19H11N - - - - - 6.83 - 6.26 6.31 351 351 345 Anthanthrene C22H12 6.46 6.52 - - 6.920.04 7.22 - 6.58 6.64 351 351 352 352 345 Benzo[g,h,i]perylene C22H12 6.79 6.86 7.19 7.21 7.170.02 - - 5.92 5.96 351 351 352 352 345 Pentacene C22H14 6.12 6.16 6.58 6.61 6.630.05 - - 6.60 6.68 352 352 357 Pentaphene C22H14 - - 7.41 7.41 7.340.04 - - 6.80 6.87 352 352 345 Picene C22H14 - - 7.62 7.58 7.510.02 7.18 - 6.52 6.58 352 352 357 Benzo[b]chrysene C22H14 - - 7.26 7.24 7.140.04 - - 6.73 6.78 352 352 366 Dibenz[a,j]anthracene C22H14 - - 7.50 7.50 7.40.02 7.44 7.62 6.76 6.82 352 352 367 Benzo[b]triphenylene C22H14 - - 7.50 7.50 7.39 7.44 6.80 6.85 Continued

218

Table 6 Continued

352 352 366 Dibenzo[c,g]phenanthrene C22H14 - - 7.59 7.57 7.510.02 7.31 7.48 6.69 6.76 352 352 345 Dibenz[a,h]anthracene C22H14 - - 7.47 7.47 7.510.02 6.94 7.48 6.29 6.34 352 352 366 Benzo[a]naphthacene C22H14 - - 6.99 7.01 6.970.02 7.30 - 6.73 6.77 351 351 352 352 345 Coronene C24H12 7.02 7.08 7.37 7.39 7.290.03 - - 7.31 351 351 367 Dibenzo[b,def]chrysene C24H14 6.39 6.45 - - 6.82 - - 6.50 6.57 2,4,6-Triphenyl-1,3,5-Triazine C21H15N3 ------6.93 6.18 6.24 351 351 345 Dibenzo[cd,lm]perylene C26H14 6.31 6.38 - - 6.720.02 - - 7.86 7.86 351 351 345 Hexacene C26H16 5.83 5.87 - - 6.360.02 - 6.77 6.10 6.16 351 351 368 Bisantrene C28H14 5.85 5.90 - - 6.3 - - 5.92 5.96 351 351 368 Benzo[a]coronene C28H14 6.75 6.81 - - 7.08 6.21 6.41 5.66 5.70 351 351 369 Dibenzo[bc,kl]coronene C30H14 6.01 6.05 - - 6.420.02 7.17 - 6.52 6.56 351 351 368 Dibenzo[bc,ef]coronene C30H14 6.12 6.17 - - 6.5 6.06 - 5.80 5.85 351 351 345 Terrylene C30H16 5.98 6.05 - - 6.420.02 6.60 - 5.92 5.96 351 351 370 Ovalene C32H14 6.36 6.41 - - 6.860.01 6.60 5.78 5.84 351 351 Tetrabenzo[bc,ef,kl,no]corone C36H16 5.41 5.44 - - - 6.66 6.88 6.14 6.19 351 351 345 Circumbiphenyl C38H16 6.46 6.52 - - 6.810.02 5.27 - - - ne 351 351 Circumanthracene C40H16 5.90 5.94 - - - - - 6.23 6.23 351 351 345 Quaterrylene C40H20 5.62 5.68 - - 6.110.02 6.26 6.52 351 351 Circumpyrene C42H16 6.06 6.10 ------351 351 Hexabenzocoronene C42H18 6.55 6.79 ------351 351 Dicoronylene C48H20 6.07 6.13 ------351 351 Pentarylene C50H18 5.37 5.43 ------351 351 Circumcoronene C54H18 6.14 6.35 ------351 351 Circumovalene C66H20 5.71 5.74 ------

219

Comparing the difference in the neutral and positively charges states allows for a direct comparison of B3LYP and AM1 computational methods to experimental ionization values, Table 6. The computational B3LYP/6-31+G* vertical and adiabatic ionization potentials in this study are less than the values reported by Malloci et al. This is likely could be due to the molecules in the Malloci et al. study not being constrained to the first vibrational frequency. Although, The Malloci et al. values are closer to the experimental values no nitrogen containing PAHs were computed. Literature and computed AM1

Figure 64. Comparison of Experimental and Adiabatic AM1 / B3LYP/6-31G* Scaled Ionization Potentials grouped by nitrogen atoms.

220 ionization values were closer to the experimental ionization potentials with adiabatic ionization values being the closest. This suggests that some rearrangement of the molecule takes place when a positive charge is induced. PAH molecules with nitrogen incorporated had a higher ionization potential than PAHs without nitrogen, Figure 64.

It should be noted that although the B3LYP/6-31G* and AM1 levels of theory under predict the ionization potentials for the PAHs studied, there was a strong linear correlation that can be calibrated for this impreciseness. The AM1 level of theory was also more noisy that B3LYP/6-31G* with a root mean square values of 0.10 and 0.09, respectively. The ionization potential increases with the number of nitrogens incorporated suggesting that nitrogen incorporation increases the energy required for electron donation, although it has been reported that nitrogen incorporation into graphites increases the ORR activity suggesting that electron donation is not the rate-limiting step for oxygen reduction.177 This trend was also apparent in the vertical ionization potentials,

221

Figure 65.

Figure 65. Comparison of Experimental and Vertical AM1 / B3LYP/6-31G* Scaled Ionization Potentials grouped by nitrogen atoms. The vertical ionization potential calculation with the AM1 level of theory was much more noisy that B3LYP/6-31G* with a root mean square values of 0.30 and 0.05, respectively.

It is not clear why the vertical ionization potential for the B3LYP/6-31G* had the lowest root mean square fit, but it is encouraging that further studies could be conducted without re-optimizing the geometry of the cationic molecule. Although, it appeared that nitrogen incorporation increased the energy required for electron donation in Figures 64 and 65, ionization potential was also strongly correlated to the amount of heavy atoms in the

222 molecule, Figure 66.

Figure 66. Comparison of Experimental and Vertical B3LYP/6-31G* Scaled Ionization Potentials Grouped by the Number of Heavy Atoms

The molecules with more heavy atoms had a lower ionization potential. This is not surprising when one considers that in a aromatic system the charge can be de-localized over the entire molecule and that larger the molecules can have more de-localization.

This trend was also observed for adiabatic ionization potentials. This suggests that to study the electron donation of CNx catalysts large (>30 heavy atoms) nitrogen doped polycyclic aromatic hydrocarbon systems need to be used capture the effects of electron donation.

223

B3LYP in addition to calculating the IPs was used to determine the energy of the nitrogen 1s orbital and compared to experimental N 1s binding energy collected during

XPS. Computational methods could provide further insight into the deconvolution of the

223,225,256,344 N 1s region for CNx, which is in debate. To demonstrate the practicality of this approach, B3LYP was utilized to investigate the adiabatic IP and XPS N 1s peak position of a variety of PAHs indeed, this method shows a very good correlation with the available experimental data, Figure 67.223,226,227,344-350,353-364,366-376 The energy of the N 1s orbitals are determined by

Figure 67. Nitrogen 1s Energies, B3LYP/6-31G* verses XPS

224 the local bonding environment of that particular nitrogen, so the three distinct groupings in Figure 67 indicate three types of nitrogen. Unfortunately, the comparison of the computational and experimental values had a sizable amount of noise. Only limited experimental N 1s values for PAHs were available, which made it difficult to ascertain the validity of using a computational analysis to determine XPS N 1s spectra for unknown structure. To protect the X-ray source XPS is a high vacuum system, ,so there was only a small amount of experimental data that was available on PAHs because most species evolve under vacuum.

8.5 Conclusions

The findings of these initial computational analyses can be extended to more complicated larger graphites that would be more similar to CNx catalysts than the PAHs studied here. It was shown that the B3LYP/6-31G* method using the vertical ionization potential had a stronger linear correlation than the AM1 vertical and adiabatic ionization potentials and B3LYP/6-31G* adiabatic ionization potential. It was also shown that for electron donation to be accurate the structure would have to be very large, in excess of 30 large atoms. B3LYP/6-31G* also has the additional advantage of making open shell calculations more easily than AM1. B3LYP/6-31G* can also determine N 1s energy levels and can be compared to experimental XPS N 1s spectra.

225

CHAPTER 9. Membrane Electrode Assembly Fabrication and Fuel Cell Testing

Initial fuel testing was attempted using a laboratory-built fuel cell test fixture with

temperature regions controlled by heat tapes connected to variable transformers while

electrochemical conditions were controlled using a 263A potentiostat/galvanostat (PAR)

with a 10 Watt power boost. It was found that the poor water management of the

laboratory-built fuel cell test fixture caused constant water accumulation in the MEA

causing electrochemical performance. For this reason a commercial Arbin 50 W PEM

fuel cell test stand with automated humidification control was obtained and used in all

reported fuel cell testing. The Arbin fuel cell test stand was installed and calibrated for

hydrogen fuel cell testing.

9.2 Catalyst Application on PEM: Hot-Press vs. Direct Application

Bench scale fabrication of MEAs was performed by two main methods; hot-

pressing catalyst-coated decals onto the membrane or direct application of the catalyst to

226 the membrane. The details of these two techniques are discussed at length in the experimental section. MEAs fabricated by hot-pressing the coating catalyst decals allows the catalyst loading to be precisely known by measuring decal mass difference before and after hot pressing. Having an accurate measure of catalyst loading may seem like a simple matter, but when one considers that the polymer electrolyte membrane is extremely hydroscopic and can drastically change weight with humidity but also becomes extremely brittle when dried it becomes apparent that a simple mass difference of MEA without catalyst and MEA with catalyst cannot be used to determine catalyst loading.377

When catalyst is applied directly to the Nafion membrane, typically by painting or spray, the exact catalyst loading is estimated from the difference in mass of the instruments used for coating. These estimates are prone to error but are considered serviceable when comparing a series of materials. These two methods will be compared and contrasted further in the next section.

9.2.1 Membrane Electrode Assembly Fabrication Using the Hot-Press Method

To effectively obtain ionic contact with the Nafion membrane and the catalyst layer hot pressing must be performed at temperatures greater than 150 C, which is the glass transition temperature of Nafion.24 Unfortunately, at temperatures higher than 150

C an acid-catalyzed degradation of the ionomer resulting water retention loss limiting proton conductivity can occur.24,377 Therdthainwong et al. investigated the hot pressing

227 parameters for a PEM MEA with a H+ cation exchanged Nafion and found that temperatures greater than 100 C resulted in lower MEA current densities.378 In order to preserve the integrity of the Nafion polymer electrolyte membrane during hot-pressing the Nafion material was Na+ cation exchanged. Cation exchanging Nafion to the Na+ state is thought to stabilize the sulfonic acid groups within the ionomer, which are responsible for proton transfer.377 This allows the membrane to be hot-pressed at temperatures from 150 C to 190 C without degradation of the Nafion membrane.377

Although it should be noted that cation ion exchanging Nafion with Na+ or K+ can result in lower ion exchange capacity, which can reduce the overall performance of the membrane electrode assembly.377 The optimal pressing temperature Na+ cation exchanged Nafion membrane is dependent on the duration and pressure of the hot- pressing treatment with optimum treatments close to 190 C, for less than 2 minutes and less than 16 psi.38,379,380 After hot-pressing the catalyst layers to the PEM membrane, the

MEA is boiled in an acid solution to cation exchange the Na+ sulfonic acid groups within

Nafion back to the H+ state before fuel cell operation.

A dispersion of catalyst, Nafion polymer, sodium ion salt, and solvent comprise the catalyst ink that is applied to decals that are pressed into the Nafion membrane.

Electrode decals were Teflon coated so that the catalyst layers release after being adhered to the Nafion membrane. As a consequence the surface of the decal has an extremely low coefficient of friction so for the ink to coat the decal, glycerol, which is extremely viscous, was used as the solvent in the catalyst ink. The decals were dried at a temperature in excess of 150 C to remove glycerol, but the drying temperature is limited

228 by the stability of the sulfoic acid groups in Na+ cation exchanged Nafion that denature at temperatures greater than 190 C. 168,381 It is thought that some of the sodium ions remain in the Nafion after cation exchange reducing the current density of the MEA during testing. It is also suspected that some glycerol remains in the catalyst layer and detracts from membrane electrode assembly performance.

9.2.2 Membrane Electrode Assembly Fabrication Using Direct Ink Application

Another way to obtain good contact between the catalyst layer and the Nafion membrane is to directly apply the catalyst ink to the Nafion membrane. Unfortunately, water also severely swells the Nafion membrane. The membrane is placed on a headed vacuum for ink application to counteract the rapid Nafion expansion. The heat hastens catalyst ink drying while the vacuum keeps the Nafion from expanding and delaminating form the catalyst layers. Direct catalyst applications are typically preformed at temperatures less than 100 C in order to preserve the proton conductivity of the H+ state

Nafion.382 Since the membrane is not exposed to temperatures where it would degrade it is not necessary to cation exchange the Nafion material to the Na+ state. Catalyst inks are typically applied to the membrane at temperatures near but not exceeding 100 C so that the membrane is normally expanded. For example, at 25 C the water uptake of Nafion is

377 0.37 gH2O/g, which considerably expands the Nafion membrane. At 80 C the water uptake of Nafion is only 0.22 gH2O/g keeping the Nafion membrane volume relatively 229 unchanged.377 At the higher temperatures such as 105 C the water uptake of Nafion is

377 0.18 gH2O/causing the membrane contract. Since all fuel cell testing was performed at

80 C, the catalyst ink was applied to the membrane at 70 C to 80 C to avoid catalyst layer delamination caused by Nafion swelling or contraction.302 A dispersion of catalyst,

Nafion polymer and solvent comprise the catalyst ink that is directly applied to the membrane. Water is the most commonly used ink solvent because it does not introduce additional chemical complexity and is easily regulated during fuel cell operation.

Direct application of catalyst layers to the proton exchange membrane is either accomplished by catalyst ink spray17,37,171,383 or painting123,277,384. Spray techniques are more suited for scaling up the MEA fabrication process for commercial applications, but

Nafion and the hydrophobic catalysts often form agglomerations that can foul the spray head. Catalyst sprays also deliver a more uniform catalyst coating, but for lab scale fabrication where only a few MEAs are being prepared much of the catalyst ends up being wasted in spraying equipment. Catalyst painting techniques require less equipment and catalyst material, so the performance of catalyst materials was only explored using the painting technique.

9.3 Anode and Cathode Electrode Composition

Both anode and cathode catalyst layers are composed of carbon-supported catalyst mixed with sulfonated tetrafluoroethylene based fluoropolymer-copolymer (most 230 commonly Nafion) to act as a binder, and increase the proton and water transport. The optimized Nafion content of the anode is typically from 10% to 30% by mass.29 The catalyst loading in the anode of a hydrogen fuel cell is typically 0.05 mg/cm2 platinum supported on carbon.26 The Nafion content in the cathode of a hydrogen fuel can vary from 25% to 75% by mass.2,378,379,385,386 The optimized Nafion content in the cathode varies due to the catalyst type and how the MEA is fabricated. For platinum supported on carbon cathode catalysts the optimized Nafion content is typically on the lower end from

25% to 50% by mass.2 Whereas, lower density non-precious metal catalysts that have a higher Nafion content.1 Nafion content increases the proton conductivity and water management ,but also decreases electrical conductivity in the catalyst layer limiting the electrochemical utilization of the catalyst. For catalysts with a lower density, such as non- precious metal catalysts, it is intuitive that the optimal Nafion loading tends to be higher than for platinum catalysts supported on carbons.

9.4 Gas Diffusion Lay Electrode Contact

Poor bonding between the electrode and gas diffusion layer interface can result is water retention which dramatically decreases fuel cell performance also known as

“flooding”.24,377 This flooding phenomenon is caused by a lack of efficient fluid transport from the cathode layer where water accumulates due to production during the oxygen reduction and hydration spheres during reactant proton diffusion. The water that

231 accumulates is thought to limit access of oxygen reactant gas to the electrocatalytic active sites causing poor fuel cell performance. This effect was demonstrated by using the same

MEA while changing the thickness of the silicone MEA gasket material at a constant closure pressure, Figure 68. The current density of the fuel cell decreased with

Figure 68.. Polarization Curve of a Commercial MEA 1 mgPt/cm2 anode and cathode, Nafion 212 membrane; Single Cell test in an Arbin 5cm2 geometric area fuel cell fixture, (Break-in procedure and electrochemical testing are the same as Figure 69.)

higher gasket thicknesses, which suggests that there is partial flooding limiting oxygen reduction in the cathode. Thinner gaskets than those used in Figure 68 could potentially increase current density, but at some point the gas diffusion layer would compress to a

232 point where reactant and product transport would be impaired. To obtain a uniform compression of the gas diffusion layer membrane electrode assemblies were closed with

70-150 in-lbs of torque to reach an 18-25 % compression of MEA.44 The compression of the MEA fixture was determined using the following equation;

44 MEA compression (%) = 100% (tMEA – tmembrane – tgasket) / (tMEA – t membrane).

The thickness of the free standing MEA not under compression as measured by a micrometer is tMEA. The thickness of the non-humidified Nafion membrane as measured by a micrometer is tmembrane. The combined thicknesses of the cathode and anode gaskets are tgasket. The compression of Teflon gaskets is assumed to be zero so the manufactures thickness was used for tgasket. Due to variations in the MEA thicknesses relatively incompressible Teflon gaskets of different thicknesses from 2 – 12 Mils were used to achieve a MEA compression of 18 – 25 %.

9.5 Bench Scale Fabrication of a Commercial MEA

The first milestone of full-fledged fuel cell testing was to fabricate a MEA from known components to achieve comparable performance to a commercial MEA made from the same materials. This MEA was constructed using a modified decal technique developed by Los Alamos National Laboratories.387 The decals were hot pressed to the a cation exchange Na+ Nafion 212 (Dupont) membrane, cooled, then cation exchanged in

233 sulfuric acid to be the H+ Nafion state before fuel cell testing. Figure 69 demonstrates technical ability of our group to construct MEAs of similar performance to commercial

MEAs. Both MEAs are constructed with the same type of Nafion membrane, carbon

Toray paper gas diffusion layers, and catalyst electrode composition. The fabricated cathode and anode electrodes were composed of ~25 wt.% Nafion with ~1 mgPt/cm2.

The assumed composition of the electrode ink obtained from Los Alamos National

Laboratories was 7.0422 mg/cm2 ink for 1 mg Pt/cm2 loading when using C1-20 20%HP

Pt on Vulcan carbon (BASF) with a 25 wt% Nafion content. Both MEAs were chronoamperometrically conditioned at 0.5 V until the fuel cell current was maximized under the same fuel cell conditions.

Figure 69. MEA Commercial Compared to Lab Fabricated; Single Cell test in an Arbin 5cm2 geometric area fuel cell fixture, Break-in procedure taken from Electrochem MEA manufacturer 234

9.6 Bench Scale Fabrication of a Non-Precious Metal Catalyst MEA

The next milestone in full fuel cell testing was to construct a CNx cathode catalyst

MEA. CNx cathode MEAs were constructed using a slightly modified procedure

387 optimized for constructing electrodes from 20 wt%Pt/VC. An un-optimized CNx fuel cell polarization curve was obtained, Figure 70. All three of the MEAs displayed in

Figure 70 are of the same catalyst class; graphitic carbon-nitrogen catalysts heat treated

2 Figure 70. Un-optimized CNx catalyst fuel cell test comparison; Single Cell 5cm geometric area fuel cell fixtures.

235

219,386 in the presence of a transition metal(s). The material used in the CNx catalyst chosen was picked because of its high performance in RRDE that mimics cathode fuel cell conditions.73 Although the components of the compared fuel cells are different the main limitation in performance was attributed to the cathode catalyst. The performance of this non-optimized CNx cathode MEA may be lower than the other compared non- noble metal catalysts, but it is still promisingly that it was of the same order of magnitude.

Since material properties of CNx catalysts differ considerably from 20 wt%Pt/VC, many multi-affect material difficulties must be overcome to construct high performance

CNx MEAs. Additional fabrication techniques are currently being explored to enhance performance of CNx catalyst in the complicated fuel cell environment.

236

CHAPTER 10. Carbon Corrosion Characteristics of CNx nanostructures in Acidic Media and Implications for ORR performance

(adapted from 388)

10.2 Introduction to Carbon Corrosion in Fuel Cells

Accelerated electrochemical corrosion of nitrogen-containing carbon oxygen reduction catalysts was performed by a chronoamperometric hold at 1.2 V vs. NHE in acidic electrolyte using a rotating disk electrode system. Cyclic voltammograms were used to measure the electrochemically active quinone/hydroquinone redox couple indicating the degree of carbon corrosion. Half-cell testing of CNx oxygen reduction catalyst showed superior carbon corrosion resistance compared to Vulcan carbon, the most ubiquitous cathode catalyst support. When oxygen reduction activity was measured before and after carbon corrosion, carbon corrosion resilience trended with the oxygen reduction activity. CNx catalysts subjected to carbon corrosion testing did not show a change in the onset of ORR activity potential, but has a slight reduction in current density. The selectivity of CNx favored the complete reduction of dioxygen to water.

237

Proton exchange membrane fuel cells have the potential to greatly impact the portable energy utilization sector, but there are technological hurdles that must be addressed before PEM fuel cells become market-ready. Currently, costly platinum and platinum alloy catalysts with high loadings are utilized to accelerate the slow kinetics of the oxygen reduction reaction on the cathode side of the fuel cell. Alternative nitrogen- doped carbon catalysts have been shown to have significant ORR activity,1,3,73,127,132,133,178,188,192,193,211,219,260,291,298,385,386,389 but understanding the process by which these catalyst materials deactivate is necessary for further development.44,298

Carbon materials are commonly used as a support for platinum and platinum- alloyed cathode catalysts in PEM fuel cells. Carbon support materials are chosen due to their economics, high surface area, thermal and electrical conductivity, and high stability under normal PEM fuel cell operating conditions.17,18 Corrosive conditions in PEM fuel cells occur at low pH (<1),300 high temperatures (50 - 90 C),300 high operating potential

(0.6 - 1.2 V)300 and low hydrogen feed concentrations.18 These conditions prevail especially when the fuel cell is at open circuit voltage (OCV), where exchange currents are high and the cathode is subjected to potentials in excess of 1.2 V vs. NHE.17,18

Although, fuel cell operating conditions appear to make the carbon support prone to carbon corrosion, in practice, the actual carbon corrosion is slow.17,19 During normal fuel cell operation an excess of protons are available in the cathode making oxygen reduction the preferred reaction compared to carbon oxidation, due to Le Chatelier's principle. The oxygen reduction and carbon corrosion reactions involved are shown in Figure 71.

238

Figure 71. Carbon corrosion and oxygen reduction reactions and their standard potentials19,25 Carbon corrosion has been studied for various carbon support materials with20,21 and without Pt catalysts,22,23 but the corrosion of nitrogen-doped carbon ORR catalysts in acidic medium simulating the PEM fuel cell environment, to our knowledge, has not been studied directly. The focus of this publication is to elucidate the corrosion resistance of

Pt-free, nitrogen-doped graphite ORR catalysts as they compare to Vulcan carbon, the most prevalent type of carbon support used in fuel cells.

10.3 Carbon Corrosion Results and Discussion

There are two main electrochemical methods employed to accelerate carbon corrosion for ORR catalysts in half-cell and fuel cell experiments. One method employs

CVs that cycle from 1.2 - 1.4 V vs. NHE to approximately 0 V vs. NHE at fixed rate with or without rest periods between cycles for the duration of the test.17,19,30,171,300,301,390

Cycling voltage in this manner has shown a reduction in fuel cell performance for noble metal ORR catalysts and was used to demonstrate the carbon corrosion resistance of support media for the noble metal.391 This reported activity decrease could have due to 239 several phenomenon; platinum detachment without/with carbon corrosion leading to electrically inaccessible catalyst, and/or platinum agglomeration that reduces the catalytically active surface area, Figure 72.

Figure 72. The Mechanisms of Deactivation of Platinum Supported on Carbon Catalysts

However, it has been shown that this cycling can lead to the detachment of Pt nanoparticles without corroding the carbon support perceptibly.390 Furthermore, repeated CV techniques were shown to preferentially corrode the carbon sublayer that is between the active catalyst layer and the gas diffusion layer due to high potential and low hydrogen concentration, instead of the cathode catalyst layer.17 The other electrochemical 240 technique used to measure carbon corrosion is a chronoamperometric hold at high potential (1.2 - 1.4 V vs. NHE) with intermediate activity or CV testing to monitor the progress of the carbon corrosion.18,20,22,23,142,268 The chronoamperometric technique has been shown to increase the surface oxide concentration,268 which is an intermediate step in the corrosion of carbon materials before the formation of carbon dioxide gas.21

Chronoamperometric techniques have also shown a decrease in activity of Pt/C ORR cathode catalyst with duration of potential hold.22,23,268

Although there have been several reports on the corrosion resistance of supported

Pt catalysts, these studies have not investigated the corrosion characteristics of Pt-free

CNx catalysts. In this study, we have chosen to examine the corrosion behavior of

Vulcan carbon as well as CNx catalysts. Furthermore, since fuel cell operation will experience high cathode potentials during OCV and lean fuel operation during startup, chronoamperometry was chosen to examine the carbon corrosion of CNx catalyst materials.

10.3.1 Carbon corrosion on CNx materials and Vulcan carbon (Corrosion tests)

The degree of carbon corrosion was determined by observing the evolution of the quinone to hydroquinone peak in the anodic scan of intermittent CVs while performing chronoamperometric potential holds at 1.2 V vs. NHE over 48 hours in a RDE experimental setup in 0.5 M H2SO4 electrolyte. Vulcan carbon has been shown to oxidize 241 in sulfuric acid with high potential holds and undergo the intermediate hydroquinone/quinone species during carbon corrosion20,23,142. Graphitic carbon materials all have some degree of oxygen functionalization at ambient conditions.136 As carbon is corroded, formation of additional oxygen functional groups can indicate carbon corrosion. Since Vulcan carbon and CNx ORR catalysts are both primarily graphitic carbon, it is reasonable that CNx would corrode through a similar mechanism and therefore corrosion resistance could be gauged by high potential holds in sulfuric acid.

Three different types of electrochemical rotating disk electrode experiments were performed to study the effects of carbon corrosion on Vulcan carbon and CNx, Table 7.

Table 7. The types of Rotating Disk Electrochemical Experiment Performed

Test Type Consecutive experimental procedures

ORR Ink application Oxygen reduction Testing

Corrosion Ink application Corrosion testing

Oxygen Oxygen reduction ORR-Corrosion-ORR Ink application reduction Corrosion testing testing testing

Catalyst materials were first applied to the glassy carbon electrode with a Nafion binder and were accessed for oxygen reduction activity/selectivity testing without any further treatment. In another testing scheme the materials were applied to the glassy carbon electrode where they were held at 1.2 V vs. NHE with intermitted cyclic voltammograms to determine the degree of corrosion experienced. In the last experimental scheme the

242 materials were applied to the glassy carbon electrode and tested for oxygen reduction activity then held at 1.2 V vs. NHE with intermitted cyclic voltammograms to monitor corrosion after which oxygen reduction activity was reassessed.

The quinone/hydroquinone oxidation reduction reaction can be seen in Figure 73.

Figure 73. Quinone (left) / Hydroquinone (right) Couple on Carbon Graphite.

The reversible electrochemical reaction from quinone to hydroquinone is a one electron transfer that takes place in a solution that has accessible protons. Figure 74 shows the

243

Figure 74. Evolution of quinone/hydroquinone species on Vulcan Carbon-XC72 intermittent CVs taken over Vulcan Carbon while performing chronoamperic potential holds. The quinone/hydroquinone peaks are evident by the increase in current at ~0.6 V vs. NHE23,142 with time in the anodic (upper) set of linear scans. The intensity of the peaks significantly increases as the duration of high-voltage hold increases indicated a large degree of carbon corrosion with the electrochemical treatment. Figure 75 shows similar CVs taken over CNx grown on Fe-doped

244

Figure 75. Evolution of the quinone/hydroquinone species on CNx grown on Fe/Vulcan carbon

Vulcan carbon. The intensity increase of quinone/hydroquinone peaks in CNx materials is smaller than Vulcan carbon, suggesting that these materials are more corrosion resistant.

Figure 76 shows a comparison of the corrosion behaviors of three different CNx catalysts grown on different supports and untreated Vulcan carbon. The extent of

245

Figure 76. Carbon Corrosion of CNx Catalysts and Vulcan Carbon Support reported as change in the intensity of the quinone/hydroquinone electrochemically active couple (at 0.627 V vs. NHE) in the anodic current of intermittent CVs with potential hold of 1.2 V vs. NHE in a rotating disk electrode experimental setup corrosion is expressed as percent change in the intensity of the quinone/hydroquinone electrochemically active couple (at 0.627 V vs. NHE) in the anodic current of intermittent

CVs with increasing duration of the potential hold of 1.2 V vs. NHE. The high increase in the quinone/hydroquinone peak over Vulcan carbon indicates that Vulcan carbon is much less corrosion resistant than CNx ORR catalyst materials. The small growth of the quinone/hydroquinone peak for „CNx grown on Fe/MgO‟ and „CNx grown on Fe/Vulcan

Carbon‟ indicates that iron catalyzed growth of CNx leads to an increase in carbon corrosion resistance of the ORR catalysts. The increased corrosion resistance may be due

246 to an increase in graphitic character of these materials as well as the increased nitrogen functionalities on the edges. Previous studies have shown that CNx nanostructures formed during pyrolysis have a higher degree of edge plane exposure and a higher degree of pyridinic nitrogen when transition metals (Fe, Co) are used to catalyze the growth CNx catalyst.192 Although carbon corrosion is thought to attack the chemically active edges of

23,130 graphite, the observation that CNx catalysts with high edge plane exposure have a higher resistance to carbon corrosion would suggest that the inclusion of nitrogen into graphite could potentially reduce the oxidation of the exposed edge planes and retard carbon corrosion.

Figure 77 shows a comparison of the ORR activity of the same four materials

Figure 77. Oxygen Reduction Activity of four catalysts at (1000rpm, 10mV/sec, 426 μg/cm2 catalyst loading)

247 used for corrosion testing. As expected, CNx catalysts were found to have much greater oxygen reduction activity than Vulcan carbon. Nitrogen functionalities imparted to these

CNx materials through pyrolysis in nitrogen and carbon containing atmospheres is known

3,84 to increase their oxygen reduction activity. CNx ORR catalysts grown over Fe-doped supports were found to have a higher ORR activity than CNx formed without iron- catalyzed growth. This increase in activity may be due to the formation of carbon- nitrogen nanostructures that have more edge plane exposure per mass,192 which is where oxygen reduction likely takes place.94,133,188,193 Another correlation that has been reported previously is between the ORR activity and the pyridinic nitrogen content of the

1,132,194 CNx materials.

The two comparisons shown in Figures 76 and 77 reveal that catalysts with higher

ORR activity exhibit the increased resistance to carbon corrosion. In other words, the catalysts with the highest oxygen reduction activity are the most resistant to carbon corrosion. This trend was persistent in all catalysts studied. This could be due to the catalyst‟s ability to preferentially facilitate the reduction of oxygen over the surface instead of carbon oxidation reactions that would lead to carbon corrosion.

10.3.2 Effect of Corrosion on ORR (ORR-Corrosion-ORR tests)

In an effort to observe the effect of carbon corrosion on ORR performance of these materials, the RRDE oxygen reduction activity and selectivity tests were performed 248

Figure 78. RRDE activity and selectivity comparison before and after corrosion testing, (above) the ring current (below) disk current density, data collected in oxygen-saturated 2 0.5 M H2SO4 (1000 rpm with a catalyst loading 426 μg/cm , 10 mV/s.) and the ring current at the Pt ring held at 1.2 V vs. NHE in oxygen saturated 0.5 M H2SO4. After Ar background subtraction

249 before and after electrochemical corrosion testing. The same electrocatalyst electrode was submitted to all three electrochemical tests in succession so currents could be compared directly and unambiguously. A comparison of the disk and ring currents obtained before and after corrosion tests is presented in Figure 78 for the most active (CNx grown on

Fe/MgO) and the least active (Vulcan carbon support) ORR catalysts. The selectivity of the catalyst materials before and after corrosion testing was performed was calculated using disk and ring currents and is expressed as number of electrons transferred per O2 molecule (n), shown in Figure 79. For the CNx grown on Fe/MgO, the ORR activity

Figure 79. Number of electrons transferred per O2 molecule calculated from the ring and disk currents from the oxygen reduction activity experiments shown in Figure 77. current density was slightly lower following corrosion testing although the onset of 250 activity did not change. Vulcan carbon did not show much ORR activity loss, however, the activity was extremely low to start with. Another observation is that the selectivity of the most active catalyst improved following corrosion testing (n increasing from ~3.7 to

3.9) while that of Vulcan carbon somewhat decreased. This could be due in part to an increase in oxygen function groups, which were observed electrochemically.11,211

Because the current density of Vulcan carbon was very low at voltages above 0.3 V vs.

NHE, the selectivity values were not reported in this range (Figure 79).

The effect of ORR activity testing on the corrosion behavior of the catalysts was examined by performing corrosion tests on fresh catalyst-coated electrodes and electrodes after ORR tests. Corrosion CVs taken on freshly applied catalysts and catalysts post-ORR testing after 0 hr or 48 hrs of hold time at high voltage are shown for CNx grown on

Fe/Vulcan carbon and untreated Vulcan carbon in Figures 80 and 81, respectively.

Vulcan carbon and CNx grown on

251

Figure 80. Quinone/hydroquinone detection CVs after 0 and 48 hr 1.2 V vs. NHE potential holds for; a) fresh and ORR tested CNx grown on Fe/Vulcan carbon. Tested in 2 0.5 M H2SO4, at @ 1000 rpm, 10mV/sec, 426 μg/cm catalyst loading.

Fe/Vulcan carbon catalysts both show a substantial decrease of the hydroquinone/quinone peak after ORR testing. The relative intensity increase of the quinone/hydroquinone current during corrosion testing (from 0 hr to 48 hrs of hold time) did not change much for Vulcan carbon following ORR testing. CNx grown on Fe/Vulcan carbon showed significantly less growth in the quinone/hydroquinone peak post ORR compared to fresh catalyst, but had a substantial decrease in the cathodic current at low potentials (Figure

4(a)). The decrease in cathodic current at low potentials (0.5 to 0.2 V vs. NHE) during

252

Figure 81. Quinone/hydroquinone detection CVs after 0 and 48 hr 1.2 V vs. NHE potential holds for fresh and ORR tested Vulcan carbon XC-72.Tested in 0.5 M H2SO4, at @ 1000 rpm, 10 mV/sec, 426 μg/cm2 catalyst loading.

corrosion testing for materials subjected to ORR testing was unique to CNx catalyst materials. This decrease in cathodic current at low potential for CNx grown on Fe/Vulcan carbon post-ORR testing can be attributed to a removal of surface oxides apart from the electrochemically active quinone/hydroquinone species.21,392 ORR testing appears to have consumed much of electrochemically active quinone/hydroquinone and low cathodic potential oxygen species present on the catalyst surface (Figures 80 and 81), suggesting a surface stabilization during ORR testing, which makes it less prone to carbon corrosion.

The decrease in activity observed for CNx catalysts was also observed on a commercial 20 wt% platinum catalyst supported on a Vulcan carbon, Figure 82. It should

253

Figure 82. RRDE activity and selectivity comparison before and after corrosion testing disk current. Data collected in oxygen-saturated 0.5 M H2SO4 (1000 rpm with a catalyst loading 426 μg/cm2, 10 mV/s.) and the ring current at the Pt ring held at 1.2 V vs. NHE in oxygen saturated 0.5 M H2SO4. Data are presented after argon background is subtracted. be noted that the platinum catalyst has a higher take-off potential and higher limiting current than the inexpensive, non-optimized CNx studied. Curiously, corrosion testing has a much more detrimental effect on oxygen reduction active, as evidenced by the lower take-off potential, on the platinum catalyst than CNx. This would suggest that although platinum catalysts are more active than CNx their oxygen reduction activity decreases at a higher rate under carbon corrosion prone fuel cell conditions.

254

10.4 Carbon Corrosion Testing Conclusions

CNx ORR catalyst materials showed superior carbon corrosion resistance compared to Vulcan carbon, which is the most commonly used carbon material in fuel cells. CNx materials grown on an iron-doped support had higher ORR activity than materials subjected to an acetonitrile atmosphere at high temperature without an iron growth catalyst. CNx catalysts subjected to carbon corrosion testing did not show a change in the potentials of activity onset only a slight reduction in current density, but exhibited improved ORR selectivity to the complete reduction of dioxygen to water.

Carbon corrosion resistance trended with increasing oxygen reduction activity, with the most active CNx catalysts showing the most resistance to corrosion. These CNx catalysts also became more resistant to carbon corrosion following ORR activity measurements, possibly due to surface stabilization during the oxygen reduction reaction.

255

CHAPTER 11. Conclusions

11.2 Phosphorus Doped CNx Catalysts

CNxPy catalysts were synthesized by pyrolyzing acetonitrile over an iron acetate and triphenylphosphine-doped magnesia support at 900 C for 2 hrs. The resulting material was treated in 1 M HCl(aq) for 1 hr at 60 C to remove any exposed magnesia and iron, then rinsed with DI-H2O. Rotating ring disk electrode ORR activity testing showed a dramatic increase in current density and take off potential for phosphorus loadings lower than Fe:P = 1 to 20 in CNxPy ORR catalysts. The greatest activity improvements were observed when P/Fe growth media molar ratios are less than 1. To isolate the effect of phosphorus growth, a growth media impregnated with the same method without phosphorus, Fe:P =1 to 0 blank, was prepared. The Fe:P = 1 to 0 blank

CNx catalyst had less activity than CNx grown on 2 wt%Fe/MgO which confirmed that triphenylphosphine effect was altering ORR activity. Low levels of phosphorous are thought to lower the eutectic point of transition metals used to grow carbon nanostructure during pyrolysis, which promotes carbon growth.12,272 Phosphorus affected the growth and electronic properties of the carbon materials as seen by the increasing disorder in the nanofiber morphology with the increasing molar ratio of phosphorus present during

256 pyrolysis. No trend in nitrogen or oxygen surface composition correlated with the oxygen reduction activity of the phosphorus doped CNx electrocatalysts. Although, temperature programmed oxidation shown a similar nitrogen composition in the bulk of all catalysts studied. X-ray absorption showed that phosphorus introduced to the iron growth media decreases the size of the iron nanoparticles found encased in carbon layers within the resulting CNxPy catalysts. Unfortunately, none of the techniques used thus far were able to conclusively detect phosphorus. To synthesize catalyst materials with increased oxygen reduction activity the iron, phosphorus, and nitrogen source must be present during the growth of the carbon nanostructured catalyst. The inclusion of phosphorus and nitrogen dopants into graphite materials allows for the tailoring of electrocatalytic and physical properties of carbon nanostructures.

11.3 Probing Oxygen Reduction Active Site Though Catalytic Poison

11.3.1 Investigation of Sulfur Poisoning on CNx ORR Catalysts

A sulfur poisoning treatment was performed on both platinum and CNx oxygen reduction catalysts to investigate the role of the transition metal in the CNx ORR active site. The intent of sulfur poisoning treatment was to effectively eliminate iron from the proposed electrocatalytic oxygen reduction active site(s) in CNx. The sulfur treatment was found to reduce the ORR activity on the platinum catalyst, which demonstrated the validity of the sulfur poisoning treatment. The H2S treatment was found to increase the 257 activity on the CNx catalyst, but this is likely due to a reorganization of the surface species. The incorporation of sulfur into the CNx catalyst was verified by TPO and XPS spectra. The iron phase within the CNx catalysts was investigated with X-ray absorption and was found to be similar for the treatments studied, suggesting that iron was not abundantly present on the catalyst surface. XANES analysis of the CNx Fe K-edge showed the iron phase in CNx to be very metallic and no spectra resembled the iron macrocycle standard, which is similar to the hypothesized nitrogen bonded transition metal ORR active site.

11.3.2 The Effect of Carbon Monoxide on Oxygen Reduction PEM Fuel Cell Catalysts

Researchers have hypothesized many different active sites on non-precious metal carbon-based ORR catalysts, ranging from metal atoms stabilized by nitrogen groups to non-metallic sites grown catalytically by the presence of a metal during pyrolysis, but the exact bonding configuration that facilitates ORR activity is still debated.51,61,94,103,283

Determining the effect of carbon monoxide competitive adsorption in the presence of oxygen on CNx, the active site for ORR can be either classified as metal containing or metal-free. During oxygen reduction carbon monoxide did not preferentially adsorb to the oxygen reduction active site in CNx, but does preferentially adsorb to platinum causing all oxygen reduction to cease.

258

In an electrolyte continuously saturated with oxygen, while the electrodes were held at 0.3V vs. NHE, platinum on Vulcan carbon, iron phthalocyanine, and CNx deactivated at varied rates. The negative normalized current that corresponds to oxygen reduction activity for the iron phthalocyanine catalyst decreases the most during the potential hold experiment, indicating that this catalyst is the least stable of those studied.

The deactivation of platinum on Vulcan carbon was similar to CNx with the same potential hold treatment.

Carbon monoxide experiments over CNx electrocatalyst showed no oxygen reduction poisoning effect. When CO & O2 were diffused through the electrolyte, the oxygen reduction behavior completely ceased for the Pt catalyst, but only diminished in accordance to a lower oxygen concentration for the CNx catalyst.

During oxygen reduction, carbon monoxide preferentially adsorbed to the platinum electrocatalysts stopping all oxygen reduction, but was reversibly recovered once carbon monoxide left the system. Carbon monoxide does not preferentially adsorb to the oxygen reduction active site in CNx during oxygen reduction. Carbon monoxide pulse chemisorption confirmed that CO is adsorbed to platinum on Vulcan carbon and not adsorbed on CNx.

259

11.4 Rotating Ring Disk Electrode Methodology's Impact on Selectivity

Electrochemical half-cell testing in acidic electrolyte through the use of both rotating disk electrode and rotating ring disk electrode techniques were used to determine the activity and selectivity of the ORR catalysts. Both RDE and RRDE allow for efficient screening of oxygen reduction catalysts so that only the most active oxygen reducing catalysts can be labor intensively fabricated into MEAs for full fuel cell testing.

With RDE the catalyst activity was determined through cyclic voltammatry scans (CVs) in oxygen-saturated electrolyte. Studying these CVs at multiple rotation rates allowed for the calculation of the ORR selectivity to water formation (as opposed to hydrogen peroxide formation) using the Koutecky-Levich equation. With the RRDE technique, the catalyst activity was measured from CVs on the disk electrode as hydrogen peroxide was simultaneously detected from the surrounding ring held at a constant potential, which allowed for a direct determination of catalyst selectivity.

The RRDE technique was better suited for determining the selectivity of the oxygen reduction reaction of catalyst-coated disks than the RDE Koutecky-Levich analysis. The error in the collection efficiency values due to catalyst loading in the RRDE system were much less than the accumulated error caused in the grouped constants of the

Koutecky-Levich equation that were used to determine selectivity in the RDE technique.

The collection efficiency of the RRDE technique was verified at various catalyst loadings

260 of CNx and platinum supported on Vulcan carbon using a iron ferricyanide redox couple in basic solution. For the RDE technique the error associated with catalyst loading would be masked by the inherent error in the Koutecky-Levich equation making it difficult to separate effects. Although there was some change in the RRDE collection efficiency with catalyst loading the effects the propagated in error in the selectivity for the oxygen reduction reaction was very small.330

11.5 Computational Chemistry

The high-temperature treatment necessary to achieve stable CNx materials facilitates the formation of multitude bonding configurations, thereby obscuring the atomic connectivity of these materials. It was established that nitrogen must be incorporated into the final material to achieve high activity. In our previous work we have demonstrated significant ORR activity in a metal-free (< 1 ppm) CNx catalyst, suggesting that there are either multiple active sites, one of which does not have a coordinated metal.132 Taking these findings into consideration the system of study was chosen to be large graphitic aromatic hydrocarbons with are thought to be phenomenological similar to CNx catalysts.

A fundamental computation study of CNx catalysts was initiated to investigate the unique electronic phenomena of these materials to elucidate electron donation catalytic that was hypothesized to be a rate-limiting step in oxygen reduction. Small polycyclic 261 aromatic hydrocarbons, with nitrogen incorporated when available, that had readily available experimental data were used to substantiate data determined by computational methods. The ability of CNx catalysts to donate electrons is likely to be linked to the

ORR catalytic activity, so this phenomena in polycyclic hydrocarbons was compared to the experimental adiabatic ionization energy and found to correlate strongly. The local bonding environment of nitrogen, characterized though XPS N 1s spectra, is thought to play a role into the oxygen reduction activity of CNx materials. To better understand the local electronic state of incorporated nitrogens, B3LYP with the 6-31 g* functional was utilized to investigate the XPS N 1s peak position of a variety of PAHs and this method was found to have a very good correlation with the available experimental data. This combined approach justifies the use of B3LYP density functional theory techniques for studying the electronic structure of these CNx materials in future studies.

11.6 Fuel cell testing and MEA Fabrication

To understand the integrated affects of CNx ORR catalyst in a fuel cell environment an Arbin 50W PEM and direct methanol fuel cell test stand was installed and electrochemical-testing procedure was established.

The first milestone of full fuel cell testing was to fabricate a MEA from known components with performance comparable to a commercial MEA made from the same materials. The MEA was constructed using a modified decal technique developed by Los 262

Alamos National Labs (LANL).387 The MEA using bench scale hot-press technique demonstrated of similar performance to a commercial MEA with the same platinum on

Vulcan carbon catalyst loadings for the anode and cathode.

The next milestone for full fuel cell testing was to fabricate a CNx cathode catalyst MEA. CNx cathode MEAs were fabricated using a slightly modified procedure

387 optimized for constructing electrodes from 20 wt%Pt/VC. An un-optimized CNx fuel cell polarization curve was obtained. The CNx cathode MEA fabricated had performance within the same order of magnitude as cathode catalysts within the same class; graphitic carbon-nitrogen catalysts heat treated in the presence of a transition metal(s).219,386 The material used in the CNx catalyst chosen was selected because of its high performance in

RRDE that mimics cathode fuel cell conditions.73 Although the components of the compared fuel cells were different from those in the literature the main limitation in performance was attributed to the cathode catalyst. The performance of this non- optimized CNx cathode MEA may be lower than the other compared non-noble metal catalysts, but it is still promisingly that a non-optimized synthesis technique was able to achieve a material with activity of the same order of magnitude.

Since material properties of CNx catalysts differ considerably from 20 wt%Pt/VC, many multi-effect material difficulties must be overcome to construct high performance

CNx MEAs. Additional fabrication techniques will be explored to enhance performance of CNx catalyst in the complicated fuel cell environment.

263

11.7 Corrosion Testing

The oxidizing and acidic environment of PEM and DMFC cathodes provides an additional challenge in the development of catalyst materials. The long-term stability of the carbon black used to support platinum catalysts in the cathode was studied. A precedent that was established to examine the corrosion properties of cathode materials including supports of carbon blacks and carbon nanostructures using accelerated half-cell

21,23 testing was used to study CNx catalysts. The electrochemical hydroquinone/quinone redox couple indicated the oxidation of carbonaceous material and were used to monitor the corrosion progress of CNx and Vulcan carbon. The advantage of performing accelerated carbon corrosion testing is that the catalyst can be examined separately without needing to fabricate an MEA.

CNx catalysts and carbon were compared in accelerated aging conditions using hydroquinone/quinone cyclic voltammetry. After 48 hrs of testing all CNx catalysts had less change in the hydroquinone/quinone peak intensity. CNx catalysts were found to be more corrosion resistant than Vulcan carbon: the most common catalyst support in fuel cells. Interestingly there appeared to be some correlation with corrosion resistance and

ORR activity, but more tests are needed to make the trend better resolved.

264

CHAPTER 12. Recommendations for Future Work

12.2 Summary of the Cathode PEM Fuel Cell Catalyst Research Progress

The focus of the non-precious metal, carbon-based, cathode catalyst ORR PEM fuel cell project at the Ohio State University has covered considerable amount of ground in this field. To add perspective the proposed future work this discussion will begin with a brief review of all the research covered by Dr. Ozkan's PEM fuel cell group in chronological order.

This project began by studying the decomposition of acetonitrile vapor over nickel and iron impregnated alumina supports to form an active oxygen reduction electrocatalyst.132 In this publication nanostructure edge-plane exposure was linked with catalytic ORR activity. The growth of the nanostructured CNx catalyst was investigated with a battery of techniques that demonstrated the following; most of the non-electrically conductive alumina support could be removed by a post-carbon-growth acid wash, CNx catalysts grown without a metal could have considerable ORR activity, and the presence

133 of a transition metal enhances the ORR activity of the resulting CNx catalyst.

Comparable ORR activity was attainable for carbon-nitrogen graphites grown over metal- 265 doped alumina supports and carbon-nitrogen graphites grown over metal-doped silica and magnesia supports.73 Furthermore, the type of metal and support was found to effect the

193 morphology of the CNx nanostructures and the resulting ORR activity. Later it was discovered that the amount nitrogen on the surface of the carbon-graphite materials corresponded with the oxygen reduction activity of that material.192 It had been suggested by other research groups that trace amounts of transition metal (such as Fe, Co) could be coordinated to nitrogen groups and play an important role in the ORR process.20-23 To investigate this claim 57Fe Mossbauer spectroscopy was used to probe the iron metal used to facilitate the growth during acetonitrile decomposition of CNx catalysts. It was found that the iron in the growth media was in the iron carbide phase encased in sheets

279 carbon. Next, CNx materials grown from the decomposition of acetonitrile over iron- impregnated magnesia were found to preferentially reduce oxygen in the presence of methanol unlike platinum-based catalysts in the direct methanol cathode fuel cell environment.307

At this point Dieter von Deak began to contribute to Dr. Ozkan's PEM fuel cell project. It was shown that post-treating CNx materials with strong acids improved the oxygen reduction current density by increasing the proportion of oxygen functional groups on the surface.211 Similarly, the concentration of oxygen functional groups on the surface of CNx materials correlated the activity of the catalyst for the oxidative dehydrogenation and oxygen reduction reactions.173 This demonstrated that oxygen functional groups contribute to the reactivity of carbon-nitrogen graphite materials.

266

At this point, the project intensely investigated the catalyst preparation effects on the ORR electrochemical testing in simulated cathode half-cell environment. It was found that the duration that non-precious metal carbon-nitrogen catalysts are exposed to ethanol in the catalyst ink solutions impacted the oxygen reduction activity as reported by RRDE half-cell methods.393 A further inquiry confirmed that increased loadings of catalyst on disk of the RRDE of the half-cell cathode system improved the measured ORR selectivity of the catalyst.330 This investigation suggested that a possible 2-step oxygen reduction mechanism could be responsible for the observed improvement in selectivity with catalyst loading.

Carbon corrosion was thought to be a contributing factor to the deactivation of Pt- based and non-precious metal carbon-based oxygen reduction cathode fuel cell catalysts, so CNx catalyst and Vulcan carbon were held at corrosive potentials in the half-cell cathode environment while the formation of surface oxygen functional groups were

388 measured. It was found the graphitic nature of CNx catalyst made the material much more carbon corrosion resistant than Vulcan carbon, the most commonly used support for platinum-based catalysts.

The original CNx synthesis parameters where were altered in an effort to elucidate the source of catalytic activity. Stacked platelet carbon nanostructures where partially oxidized, then treated with an ammonia pyrolysis to create an nitrogen functionalized active oxygen reduction catalyst to demonstrate that nitrogen-carbon edge planes were not the controlling factor in forming active non-precious metal ORR materials.5 This suggested that nitrogen sites in the graphite edge planes were not solely responsible for 267 oxygen reduction activity. The addition of thiophene to acetonitrile during the pyrolytic growth of CNx over iron impregnated magnesia was found to increase amount of material produced without effecting the oxygen reduction activity.11 This suggested that sulfur could be used as a carbon growth promoter to form CNx materials without effecting the electro catalytic properties or the carbon morphology of the nitrogen-doped carbon catalyst. Contrarily, CNx catalyst grown by acetonitrile decomposition over triphenylphosphine impregnated on iron impregnated magnesia was found to increase oxygen reduction activity and effect the carbon morphology.271 Although, phosphorus could not be detected in the phosphorus-grown CNx catalysts, it was evident that phosphorus-doping the growth media altered the carbon material formed effecting the catalytic ORR activity.

Conflicting hypotheses of active site on non-precious metal carbon-based ORR electrocatalyst motivated a research into the active site of CNx catalysts. A computational analysis of phosphorus-doped carbon graphites demonstrated that graphite phosphate groups could favorably absorb dioxygen leading to the formation of H2O2 and a hydroperoxyl radical which are mechanistic steps in oxygen reduction.296 To determine whether an iron group was primarily responsible for the oxygen reduction activity on

CNx catalysts the catalysts were treated with sulfur and carbon monoxide catalytic poisons and tested for oxygen reduction. It was found that CNx treated at 400 ºC with H2S increased the ORR current density of the catalyst without altering the local iron bonding environment.286 It was also found that the presence of carbon monoxide during oxygen reduction did not competitively adsorb to the active site in CNx electrocatalyst. These two

268 iron-type site poisoning studies suggest that the iron residue that remains in CNx materials after synthesis was not participating in the electroreduction of oxygen.

12.3 Improving the Activity and Current Density of CNx Materials

Although there has been significant progress in elucidating the cause of oxygen reduction on CNx electrocatalysts, it is necessary to develop of materials with improved catalytic properties. In the past few years the activity of non-precious metal carbon-based oxygen reduction electrocatalysts has greatly increased. Despite intense research and development efforts, in 2005 the oxygen reduction activity of non-precious metal carbon- based catalyst was not high enough to make them viable in transportation

2,128 applications. In 2005 the highest reported ORR activity for CNx materials was less than 1.3 A cm-3 at a resistance corrected potential of 0.80 V.4 To put this in perspective the U.S DOE's 2010 volumetric activity target for non-precious metal carbon-based ORR catalysts was 130 A cm-3 at a resistance corrected potential of 0.80 V. Prior to 2008 the highest volumetric activity was only 1.4 A cm-3.108 In 2008 a major breakthrough in activity was accomplished using multiple elevated temperature ammonia treatments in an inert atmosphere to achieve a highly nitrided carbon with an activity of 2.9 A cm-3 at 0.80

V (resistance corrected).394 From this initial finding, in 2008 the Institute of Natural

Resource Sustainability group fabricated a pyrolyzed iron doped graphite materials with a post ammonia treatment that had activities of 30 and 98 A cm-3 at 0.8 V (resistance

269 corrected).1 The Los Alamos National Laboratory group during the same time period fabricated a pyrolized iron-doped cyanamide catalyst with a secondary nitrogen pyrolysis with 23 and 50 A cm-3 at 0.80 V (resistance corrected).295 It seemed that in order to create a non-precious metal oxygen reduction electro catalyst with high activity a nitrogen source and a high surface area carbon-based support doped with nominal amount of iron (less than 1 wt.%) needed to be pyrolyzed at high temperature than re-pyrolyzed in an inert or ammonia containing atmosphere. It was found that the high temperature ammonia treatment constituted a continuous weight loss due to a gasification reaction

103 causing micropore formation and increased active area. The CNx discussed in this thesis has very low (less than 0.1 A cm-3) or no mass activity above 0.8 V (resistance corrected). Similar synthesis techniques as to those mentioned above could be tailored to the experimentation that is available in our laboratories. The introduction of a secondary pyrolysis step seems necessary to achieve materials with high activity. The BET surface area and microporous surface area will need be characterized in order to correlated activities to the surface are of the electrode material. Furthermore, the U.S Department of

Energy's 2015 target of 300 A cm-3 for non-precious metal oxygen reduction catalysts is fast approaching so considerable advances in the CNx catalyst material must completed in the near term.

12.4 In-situ X-ray Absorption on the Fe K-edge During Oxygen Reduction

270

There is considerable ambiguity in the role that iron plays in enhancing the oxygen reduction activity of nitrogen-doped carbon materials, so the local bonding environment of iron will be characterized during oxygen reduction to gain insight into the active site of CNx. Some researchers have hypothesized that a metal coordinated to nitrogens forms a ORR active site(s) different from a macrocycle is formed during pyrolysis.86,103,105,248 Although it is agreed that metal is part of the active site, there is considerable disagreement in how this site is formed and where it is located.

The other main ORR site hypothesis is that the active site formed does not contain a metal. Based on studies of heat-treated carbon-supported macrocycles, Weisner and co- workers concluded that the metal in the organometallic compounds acts as a catalyst for the formation of a non-metal containing electrocatalytic oxygen reduction active site70,254.

Goureck et al. presented work on heat-treated charcoal supported Co-tetraazaannulene and concluded that the nitrogen species were the source of oxygen reduction through protonation and subsequent oxidation of the nitrogen ions.61 Gojkovic et al. went on to propose that an N-O species was responsible ORR activity on heat-treated Fe porphyrin type samples and that metal, metal oxide, or metal carbide particles could act as the catalyst for the formation of these active sites.60,68 Nitrogen functionalized heat-treated carbons have been studied extensively by several researchers,201,256-261 but no consensus was reached on the structure responsible for oxygen reduction.

The metal atom may indeed catalyze the formation of active sites on carbon supports for non-noble metal ORR catalysts. Both metal-containing and metal-free hypothesized active site types are enhanced by the inclusion of nitrogen into the 271 hexagonal network188-190. However, the structure of these active sites and how the metal enhances the formation of these sites during pyrolysis remains unknown and warrants further investigation for catalyst development.

A X-ray absorption in-situ half cell with an acid liquid electrolytic cell to study the transformations of the iron coordination state during oxygen reduction in a simulated cathode environment could determine iron's involvement in the ORR activity of CNx.

Previous ORR electrochemical research on CNx in solution indicated that there was some dissolution of the iron metal into the electrolyte solution, but the extent of dissolution

279 over extended time remains unknown. Hence, any changes in activity of CNx for ORR in an acidic media over a range of potentials could be attributed to either a change in the iron phase/oxidation state , or complete dissolution of the metal.

Half-cell experiments with a flexible electrolyte would enable us to collect in-situ

EXAFS data that could probe the state of iron during ORR. Some research was performed using electrochemcal systems while collecting X-ray absorption and

Mössbauer data. In 1983 Scherson et al. designed an in situ half cell system to characterize cobalt in cobalt tetramethoxyphenl porphyrin during electrochemical oxygen reduction using Mössbauer spectroscopy.54 Later, Scherson et al. used Mössbauer and pyrolysis-mass spectrometry to investigate the thermal stability of iron phthalocyanine, a known oxygen reduction catalyst.63,395 In 1991, Kim et al. used an in situ XANES analysis to characterize an iron phorphyrin irreversibly adsorbed on an electrode surface.396 In 2003 Stefan et al. performed an in situ Fe K-edge X-ray absorption fine structure on an iron phthalocyanine adsorbed on the surface of a high surface area carbon 272 electrode and observed changes in the iron oxidation state at different potentials in an acidic electrolyte.397 A similar electrode design and half cell design could be modified to serve as a stationary electrode system to simulate the cathode PEM fuel cell environment for platinum and carbon-based oxygen reduction catalysts while simultaneously collection electrochemical and X-ray absorption data. This type of experimental setup could be used to monitor the local bonding environment of platinum and the metal residue in CNx catalysts during oxygen reduction. Bae et al. examined the formation of a carbon monoxide adduct on a iron porphyrin in acidic electrolyte using X-ray adsorption.306 Correspondingly, this technique could be used to examine the effects of anion absorption of different electrolytes or catalyst poisons during oxygen reduction on the platinum or Fe-Nx type active sites. An in situ X-ray absorption electrochemical half cell experimental setup could also be used to monitor carbon corrosion and the corresponding metal dissolution during oxygen reduction.

An in-situ X-ray absorption spectroscopy experimental setup that allows for the analysis of the electrocatalyst during full fuel cell operation could elucidate the state of the catalyst and how metal state effect performance. X-ray absorption spectroscopy was used to examine the Pt L3-edge during fuel cell operation which has a much higher

398-401 adsorption energy than the iron or cobalt K-edge found in CNx catalysts. Since the

Pt L3-edge is at a high energy the mean penetration depth of the X-ray is much higher allowing for a thicker fuel cell graphite flow plates. To investigate the iron or cobalt K- edge a much thinner more fragile flow plates will need to be manufactured to obtain high signal to noise X-ray absorption spectra. For example, Principi et al. used a set of

273 graphite flow plates that had a combined graphite plate thickness of 0.5 mm to examine the Co K-edge of a 20 wt%PtCo cathode fuel cell catalyst with a loading of 0.2 mgCo/cm2 and 0.6 mgPt/cm2.398 There have not been any publications to our knowledge that have used XAFS to study the catalytic changes observed in non-precious metal ORR catalysts during fuel cell operation. By carrying out in-situ fuel cell tests with EXAFS on

CNx catalysts, we expect to observe the interaction of the Fe-N-C species in CNx with the ionomers present in the proton conducting Nafion membrane. The current-voltage characteristics obtained from in-situ analyses will also elucidate the changes occurring in

Fe phase, while catalyst degradation transpires.

These in-situ X-ray absorption spectroscopy experiments could provide invaluable information on the nature and role of iron as an electrocatalyst in non-precious metal carbon-based ORR catalysts which cannot be characterized using X-ray diffraction or high-vacuum X-ray photoelectron spectroscopy. In situ X-ray absorption spectroscopy could also be used to detect the electrochemical phenomena of Pt based catalysts as well.

The combined analysis of the these in situ experiments will further the understanding of the non-precious metal active site and, the activation-deactivation of CNx oxygen reduction catalysts.

274

CHAPTER 13. Glossary of Acronyms

B3LYP. Becke, three-parameter, Lee-Yang-Parr hybrid exchange-correlation functional in density fuctional theory. Used in computation chemistry to investigate a system using electron density.

BET. Brunauer, Emmett and Teller. A method for determination of material surface area using a multi-layer adsorption theory.

CNx. Nitrogen-Containing Carbon Nanostructure. This is the general carbon- nitrogen catalyst studied in this dissertation.

CV. Cyclic Voltammetry or Voltammogram. An electrochemical method where potential is controlled by scanning at a known rate and range in both cathodic and anodic directions while monitoring current.

DFT. Density functional theory. A quantum mechanical modelling methode to determine the electronic structure of atoms and molecules.

DMFC. Direct Methanol Fuel Cell. A type of low-temperature fuel where fuel fed is 1-3M aqueous methanol and oxidant is oxygen.

275

DSC. Differential Scanning Calrimetry. A techniques that sences the heat flow to and from a body of interest.

EXAFS. Extended X-ray Absorption Fine Structure. Characterizing the X-ray absorption of an element by examing the absorption edge after the initial edge jump.

IP. Ionization Potential. The energy required for an atom or molecule to expell an electron.

IWI. Incipient Wetness Impregnation. To impart a solution into a support with the solution having an equal pore volume to the support with the assumption that the solution with reside in the pores of the support.

MEA. Membrane Electrode Assembly. The interior components of a fuel cell that combines the gas diffusion layers, catalyst layers and membrane electrolyte.

ORR. Oxygen Reduction Reaction. The reaction occurring at the cathode in both direct methanol and PEM fuel cells. In an acidic aqueous system at 25 °C, the reaction is:

+ - O2 + 4H + 4e  2H2O 1.23 V vs. NHE

276

NHE. Normal Hydrogen Electrode. Standard reference in which electrochemical potentials are reported. It is the potential at which hydrogen oxidizes in an aqueous acidic system at 25 °C.

PEM. Proton Exchange Membrane Fuel Cell. Also known as Polymer

Electrolyte Membrane Fuel Cell. This is a low temperature hydrogen fuel cell.

RRDE. Rotating Ring Disk Electrode. This is an electrochemical half cell testing method in which one reaction is performed on the disk electrode and its products are monitored on the ring during laminar flow. In the study of ORR catalysts, the disk reduces oxygen and the ring monitors any hydrogen peroxide products.

TEM. Transmission Electron Microscope. An electron microscope where electrons are transmitted through sample for detection.

TGA. Thermogravimetric Analysis. An experiment where the temperature is recorded with respect to the weight change of the sample.

TPD. Temperature Programmed Desorption. An experiment where temperature is controlled while flowing an inert gas over a sample. The products are analyzed for characteristic temperatures and species.

277

TPO. Temperature Programmed Oxidation. An experiment where temperature is controlled while flowing an oxidizing gas over a sample. The products are analyzed for characteristic temperatures and species.

XANES. X-ray Absorption Near Edge Structure. Characterization of the adsorption band of an element using the finger print near the adsorption edge jump.

XPS. X-Ray Photoelectron Spectroscopy. A surface sensitive method in which an X-ray beam is directed at a sample which then photoemits electrons at characteristic binding energies. These characteristic binding energies are based upon the chemical species and chemical bonding of the species.

278

References

1. Lefevre, M.; Proietti, E.; Jaouen, F.; Dodelet, J.-P., "Iron-based catalysts with improved oxygen reduction activity in polymer electrolyte fuel cells." Science 2009, 324, 71-74.

2. Gasteiger, H. A.; Kocha, S. S.; Sompalli, B.; Wagner, F. T., "Review - Activity benchmarks and requirements for Pt, Pt-alloy, and non-Pt oxygen reduction catalysts for PEMFCs." Appl. Catal. B-Environ. 2005, 56, 9-35.

3. Shao, Y.; Sui, J.; Yin, G.; Gao, Y., "Nitrogen-doped carbon nanostructures and their composites as catalytic materials for proton exchange membrane fuel cell." Appl. Catal. B-Environ. 2008, 79, 89-99.

4. Jaouen, F.; Proietti, E.; Lefevre, M.; Chenitz, R.; Dodelet, J. P.; Chung, H. T.; Johnston, C. M.; Zelenay, P., "Recent advances in non-precious metal catalysis for oxygen-reduction reaction in polymer electrolyte fuel cells." Energy and Environmental Science 2011, 4, (1), 114-130.

5. Biddinger, E. J.; Ozkan, U. S., "Role of graphitic edge-plane exposure in carbon nanostructures for oxygen reduction reaction." J. Phys. Chem. C 2010, 114, (36), 15306- 15314.

6. Arabczyk, W.; Moszynski, D.; Narkiewicz, U.; Pelka, R.; Podsiadly, M., "Poisoning of iron catalyst by sulfur." Catal. Today 2007, (124), 43-48.

7. Oudar, J., "Sulfur adsorption and poisoning of metallic catalysts." Catalysis Reviews: Science and Engineering 1980, 22, (2), 171-195.

8. Garsany, Y.; Baturina, O. A.; Swider-Lyons, K. E., "Oxygen reduction reaction kinetics of SO2-contaminated Pt3Co and Pt/Vulcan carbon electrocatalysts." J. Electrochem. Soc. 2009, 156, (7), B848-B855. 279

9. Zhang, L.; Millet, J. M. M.; Ozkan, U. S., "Deactivation characteristics of Fe-Al- Cu water-gas shift catalysts in the presence of H2S." J. Mol. Catal. A - Chem. 2009, 309, (1-2), 63-70.

10. Ruzie, C.; Even, P.; Ricard, D.; Roisnel, T.; Boitrel, B., "O2 and CO binding to tetraaza-tripodal-capped iron(II) porphyrins." Inorg. Chem. 2006, 45, 1338-1348.

11. Biddinger, E. J.; Knapke, D. S.; von Deak, D.; Ozkan, U. S., "Effect of sulfur as a growth promoter for CNx nanostructures as PEM and DMFC ORR catalysts." Appl. Catal. B-Environ. 2010, 96, 72-82.

12. Benissad-Aissani, F.; Ait-Amar, H.; Schouler, M.-C.; Gadelle, P., "The role of phosphorus in the growth of vapour-grown carbon fibres obtained by catalytic decomposition of hydrocarbons." Carbon 2004, 42, (11), 2163-2168.

13. Maciel, I. O.; Campos-Delgado, J.; Cruz-Silva, E.; Pimenta, M. A.; Sumpter, B. G.; Meunier, V.; López-Urías, F.; Muñoz-Sandoval, E.; Terrones, H.; Terrones, M.; Jorio, A., "Synthesis, electronic structure, and Raman scattering of phosphorus-doped single-wall carbon nanotubes." Nano Lett. 2009, 9, (6), 2267-2272.

14. Sumpter, B. G.; Huang, J.; Meunier, V.; Romo-Herrera, J. M.; Cruz-Silva, E.; Terrones, H.; Terrones, M., "A theroretical and experimental study on manipulating the structure and properties of carbon nanotubes using substitional dopants." Int. J. Quantum Chem. 2009, 109, 97-118.

15. Liu, A.; Zhu, J.; Han, J.; Wu, H.; Gao, W., "Influence of phosphorous doping level and acid pretreatment on the voltammetric behavior of phosphorus invorporated tetrahedral amorphous carbon film electrodes." Electroanal. 2007, 19, (17), 1773-1778.

16. Panwar, O. S.; Khan, M. A.; Kumar, M.; Shivaprasad, S. M.; Satyanarayana, B. S.; Dixit, P. N.; Bhattacharyya, R., "Characterization of boron- and phosphorous- incorporated tetrahedral amorphous carbon films deposited by the filtered cathodic vacuum arc process." Jpn. J. Appl. Phys. 2009, 48, 065501.

17. Young, A. P.; Stumper, J.; Gyenge, E., "Characterizing the structural degradation in a PEMFC cathode catalyst layer: Carbon corrosion." J. Electrochem. Soc. 2009, 156, B913-B922.

18. Kim, J.; Lee, J.; Tak, Y., "Relationship between carbon corrosion and positive electrode potential in a proton exchange membrane fuel cell during start/stop operation." J. Power Sources 2009, 192, 674-678.

280

19. Chaparro, A. M.; Mueller, N.; Atienza, C.; Daza, L., "Study of electrochemical instabilities of PEMFC electrodes in aqueous solution by means of membrane inlet mass spectrometry." J. Electroanal. Chem. 2006, 591, 67-73.

20. Shao, Y.; Yin, G.; Zhang, J.; Gao, Y., "Comparative investigation of the resistance to electrochemical oxidation of carbon black and carbon nanotubes in aquesous sulfuric acid solution." Electrochim. Acta 2006, 51, 5853-5857.

21. Li, L.; Xing, Y., "Electrochemical durability of carbon nanotubes at 80°C." J. Power Sources 2008, 178, 75-79.

22. Wang, X.; Li, W.; Chen, Z.; Waje, M.; Yan, Y., "Durability investigation of carbon nanotube as catalyst support for proton exchange membrane fuel cell." J. Power Sources 2006, 158, 154-159.

23. Li, L.; Xing, Y., "Electrochemical durability of carbon nanotubes in noncatalyzed and catalyzed oxidations." J. Electrochem. Soc. 2006, 153, (10), A1823-A1828.

24. Vielstich, W.; Gasteiger, H. A.; Lamm, A., Handbook of Fuel Cells - Fundamentals, Technology and Applications. John Wiley & Sons, Ltd.: 2003; Vol. 2: Electrocatalysis.

25. Bard, A. J.; Faulkner, L. R., Electrochemical methods: fundamentals and applications. John Wiley and Sons Ltd: New York, 2001.

26. Gasteiger, H. A.; Panels, J. E.; Yan, S. G., "Dependence of PEM fuel cell performance on catalyst loading." J. Power Sources 2004, 127, 162-171.

27. Wu, G.; Swaidan, R.; Li, D.; Li, N., "Enhanced methanol electro-oxidation activity of PtRu catalysts supported on heteroatom-doped carbon." Electrochim. Acta 2008, 2008, 7622-7629.

28. Wasmus, S.; Kuver, A., "Methanol oxidation and direct methanol fuel cells: a selective review." J. Electroanal. Chem. 1999, 461, 14-31.

29. Single Cell Test Protocol; US Fuel Cell Council: 2006.

30. Weng, F.-B.; Hsu, C.-Y.; Li, C.-W., "Experimental investigation of PEM fuel cell aging under current cycling using segmented fuel cell." Int. J. Hydrogen Energ. 2010, 35, 3664-3675.

31. Stevens, D. A.; Rouleau, J. M.; Mar, R. E.; Bonakdarpour, A.; Atanasoski, R. T.; Schmoeckel, A. K.; Debe, M. K.; Dahn, J. R., "Characterization and PEMFC testing of

281

Pt1-xMx (M = Ru, Mo, Co, Ta, Au, Sn) anode electrocatalyst composition spreads." J. Electrochem. Soc. 2007, 154, B566-B576.

32. Yang, M.-C.; Hsueh, C.-H., "Impedance analysis of working PEMFCs in the presence of carbon monoxide." J. Electrochem. Soc. 2006, 153, (6), A1043-A048.

33. Tang, Y.; Zhang, J.; Song, C.; Liu, H.; Zhang, J.; Wang, H.; Mackinnon, S.; Peckham, T.; Li, J.; McDermid, S.; Kozak, P., "Temperature dependent performance and in situ AC impedance of high-temperature PEM fuel cells using the Nafion-112 membrance." Journal of The Electrochemistry Society 2006, 153, (11), A2036-A2043.

34. Diard, J.-P.; Glandut, N.; Le Gorrec, B.; Motella, C., "Impedance measurement of each cell of a 10 W PEMFC stack under load." J. Electrochem. Soc. 2004, 151, (12), A2193-A2197.

35. Wagner, N., "Characterization of membrane electrode assemblies in polymer electrolyte fuel cells using a.c. impedance spectroscopy." J. Appl. Electrochem. 2002, 32, 859-863.

36. Diard, J.-P.; Le Gorrec, B.; Montella, C.; Poinsignon, C.; Vitter, G., "Impedance measurements of polymer electrolyte membrane fuel cells running on constant load." J. Power Sources 1998, 74, 244-245.

37. Zhang, J.; Zhang, L.; Bezerra, C. W. B.; Li, H.; Xia, Z.; Zhang, J.; Marques, A. L. B.; Marques, E. P., "EIS-assisted performance analysis of non-noble metal electrocatalyst (Fe-N/C)-based PEM fuel cells in the temperature range of 23-80°C." Electrochim. Acta 2009, 54, 1737-1743.

38. Zhang, J.; Yin, G.-P.; Wang, Z.-B.; Lai, Q.-Z.; Cai, K.-D., "Effects of hot pressing conditions on the performances of MEAs for direct methanol fuel cells." J. Power Sources 2007, 165, 73-81.

39. Ciureanu, M., "Effects of Nafion dehydration in PEM fuel cells." J. Appl. Electrochem. 2004, 34, 705-714.

40. Lee, J.-W.; Popov, B. N., "Ruthenium-based electrocatalysts for oxygen reduction reaction - a review." J. Solid State Electr. 2007, 11, 1355-1364.

41. Tributsch, H.; Bron, M.; Hilgendorff, M.; Schulenburg, H.; Dorbandt, I.; Eyert, V.; Bogdanoff, P.; Fiechter, S., "Methanol-resistant cathodic oxygen reduction catalysts for methanol fuel cells." J. Appl. Electrochem. 2001, 31, (7), 739-748.

42. Johnson Matthey Precious Metals Marketing "Platinum Today," Johnson Matthey Precious Metals Marketing, 2010, http://www.platinum.matthey.com/index.html

282

43. Serov, A.; Kwak, C., "Review of non-platinum anode catalysts for DMFC and PEMFC application." Appl. Catal. B-Environ. 2009, 90, 313-320.

44. Jaouen, F.; Herranz, J.; Lefevre, M.; Dodelet, J.-P.; Kramm, U. I.; Herrmann, I.; Bogdanoff, P.; Maruyama, J.; Nagaoka, T.; Garsuch, A.; Dahn, J. R.; Olson, T. S.; Pylypenko, S.; Atanassov, P.; Ustinov, E. A., "Cross-laboratory experimental study of non-noble-metal electrocatalysts for the oxygen reduction reaction." ACS Applied Materials & Interfaces 2009, 1, 1623-1639.

45. "Ore Reserves and Mineral Resources," Anglo Platinum Inc., 2005, http://www.angloplatinum.com/

46. Boulatov, R., "Billion-year-old oxygen cathode that actually works: Respiratory oxygen reduction and its biomimetic analogs." In N4-Macrocyclic Metal Complexes, Zagal, J. H.; Bedioui, F.; Dodelet, J.-P., Eds. Springer Science+Business Media, Inc.: New York, 2006; pp 1-40.

47. Alberts, B.; Johnson, A.; Lewis, J.; Raff, M.; Roberts, K.; Walter, P., Molecular biology of the cell. 4th ed.; Garland science: New York, 2002; Vol. 1423, p 383.

48. Jasinski, R., "A new fuel cell cathode catalyst." Nature 1964, 201, 1212.

49. Jasinski, R., "Cobalt phthalocyanine as a fuel cell cathode." J. Electrochem. Soc. 1965, 112, (5), 526-528.

50. Jahnke, H.; Schonborn, M.; Zimmerman, G., "Organic dyestuffs as catalysts for fuel cells." Fortschr. Chem. Forsch. 1976, 61, 133.

51. Wiesener, K., "N4-chelates as electrocatalysts for cathodic oxygen reduction." Electrochim. Acta 1986, 31, (8), 1073-1078.

52. Yeager, E., "Electrocatalysts for O2 reduction." Electrochim. Acta 1984, 29, (11), 1527-1537.

53. Yeager, E., "Dioxygen electrocatalysis: Mechanisms in relation to catalyst structure." J. Mol. Catal. 1986, 38, 5-25.

54. Scherson, D. A.; Gupta, S. L.; Fierro, C.; Yeager, E. B.; Kordesch, M. E.; Eldridge, J.; Hoffman, R. W.; Blue, J., "Cobalt tetramethoxyphenyl porphyrin-emission Mossbauer spectroscopy and O2 reduction electrochemical studies." Electrochim. Acta 1983, 28, (9), 1205-1209.

283

55. Bouwkamp-Wijnoltz, A. L., Visscher, W., van Veen, J. A. R., Boellaard, E., van der Kraan, A. M., Tang, S. C., "On active-site heterogeneity in pyrolyzed carbon- supported iron porphyrin catalysts for the electrochemical reduction of oxygen: an in situ mössbauer study." J. Phys. Chem. B 2002, 106, 12993.

56. Bouwkamp-Wijnoltz, A. L.; Visscher, W.; van Veen, J. A. R.; Tang, S. C., "Electrochemical reduction of oxygen: an alternative method to prepare active CoN4 catalysts." Electrochim. Acta 1999, 45, 379.

57. Bron, M.; Fletcher, S.; Hilgendorff, M.; Bogdanoff, P., "Catalysts for oxygen reduction from heat-treated carbon-supported iron phenantroline complexes." J. Appl. Electrochem. 2002, 32, 211.

58. Bron, M.; Radnik, J.; Fieber-Erdmann, M.; Bogdanoff, P.; Fiechter, S., "EXAFS, XPS and electrochemical studies on oxygen reduction catalysts obtained by treatment of iron phenanthroline complexes supported on high surface area carbon black." J. Electroanal. Chem. 2002, 535, 113-119.

59. Faubert, G.; Cote, R.; Guay, D.; Dodelet, J. P.; Denes, G.; Bertrand, P., "Iron catalysts prepared by high-temperature pyrolysis of tetraphenylporphyrins adsorbed on carbon black for oxygen reduction in polymer electrolyte fuel cells." Electrochim. Acta 1998, 43, (3-4), 341.

60. Gojkovic, S.; Gupta, S.; Savinell, R., "Heat-treated iron(III) tetramethoxyphenyl porphryin chloride supported on high-area carbon as an electrocatalyst for oxygen reduction Part II. Kinetics for oxygen reduction." J. Electroanal. Chem. 1999, 462, 63- 72.

61. Gouerec, P., Biloul, A., Contamin, O., Scarbeck, G., Savy, M., Riga, J., Weng, L.T., Bertrand, P., "Oxygen reduction in acid media catalysed by heat treated cobalt tetraazaannulene supported on an active charcoal: correlations between the performances after longevity tests and the active site configuration as seen by XPS and ToF-SIMS." J. Electroanal. Chem. 1997, 422, 61.

62. Gouerec, P.; Biloul, A.; Contamin, O.; Scarbeck, G.; Savy, M.; Barbe, J. M.; Guilard, R., "Dioxygen reduction electrocatalysis in acidic media: effect of peripheral ligand substitution on cobalt tetraphenylporphyrin." J. Electroanal. Chem. 1995, 398, 67- 75.

63. Scherson, D.; Tanaka, A. A.; Gupta, S. L.; Tryk, D.; Fierro, C.; Holze, Z.; Yeager, E. B., "Transition metal macrocycles supported on high area carbon: pyrolysis-mass spectrometry studies." Electrochim. Acta 1986, 31, (10), 1247.

284

64. van Veen, J. A. R.; van Baar, J. F.; Kroese, K. J., "Effect of heat treatment on the performance of carbon-supported transition-metal chelates in the electrochemical reduction of oxygen." Chem. Soc., Faraday Trans. I 1981, 77, 2827.

65. Maruyama, J.; Abe, I., "Carbonized hemoglobin functioning as a cathode catalyst for polymer electrolyte fuel cells." Chem. Mater. 2006, 18, 1303-1311.

66. Ladouceur, M.; Lalande, G.; Guay, D.; Dodelet, J. P.; Dignard-Bailey, L.; Trudeau, M. L.; Schulz, R., "Pyrolyzed cobalt phthalocyanine as electrocatalyst for oxygen reduction." J. Electrochem. Soc. 1993, 140, (7), 1974-1981.

67. Bron, M.; Fiechter, S.; Hilgendorff, M.; Bogdanoff, P., "Catalysts for oxygen reduction from heat-treated carbon-supported iron phenanthroline complexes." J. Appl. Electrochem. 2002, 32, (2), 211-216.

68. Gojkovic, S. L.; Gupta, S.; Savinell, R. F., "Heat-Treated iron(III) tetramethoxyphenyl porphyrin supported on high-area carbon as an electrocatalyst for oxygen reduction I. Characterization of the electrocatalyst." J. Electrochem. Soc. 1998, 145, (10), 3493-3499.

69. Claude, E.; Addou, T.; Latour, J.-M.; Aldebert, P., "A new method for electrochemical screening based on the rotating ring disk electrode and its application to oxygen reduction catalysts." J. Appl. Electrochem. 1998, 28, 57-64.

70. Franke, R.; Ohms, D.; Wiesner, K., "Investigation of the influence of thermal treatment on the properties of carbon materials modified ny N4-chelates for the reduction of oxygen in acidic media." J. Electroanal. Chem. 1989, 260, 63-73.

71. Kaisheva, A.; Gamburtsev, S.; Iliev, I., "Characteristics of air (oxygen) gas- diffusion electrodes with catalysts of active carbon and pyrolyzed compounds of the TMPP type. Acidic and neutral electrolytes." Soviet Journal of Electrochemistry 1982, 18, 127-131.

72. Gouerec, P.; Savy, M., "Oxygen reduction electrocatalysis: ageing of pyrolyzed cobalt macrocycles dispersed on an active carbon." Electrochim. Acta 1999, 44, 2653- 2661.

73. Matter, P. H.; Wang, E.; Ozkan, U. S., "Preparation of nanostructured nitrogen- containing carbon catalysts for the oxygen reduction reaction from SiO2 and MgO supported metal particles." J. Catal. 2006, 243, 395-403.

74. Johansson, L. Y.; Larsson, R., "Electrochemical reduction of oxygen in sulphuric acid catalyzed by porphyrin-like complexes." J. Mol. Catal. 1986, 38, 61-70.

285

75. Gupta, S.; Tryk, D.; Bae, I.; Aldred, W.; Yeager, E., "Heat-treated polyacrylonitrile-based catalysts for oxygen electroreduction." J. Appl. Electrochem. 1989, 19, 19.

76. Martin Alves, M. C.; Dodelet, J. P.; Guay, D.; Ladouceur, M.; Tourillion, G., "Origin of the electrocatalytic properties for O2 reduction of some heat-treated polyacrylonitrile and pthalocyanine cobalt compounds adsorbed on carbon black as probed by electrochemistry and X-ray Absorption Spectroscopy." J. Phys. Chem. 1992, 96, 10898-10905.

77. Ohms, D.; Herzog, S.; Franke, R.; Neumann, V.; Wiesener, K.; Gamburcev, S.; Kaisheva, A.; Iliev, I., "Influence of metal ions on the electrocatalytic oxygen reduction of carbon materials prepared from pyrolyzed polyacrylonitrile." J. Power Sources 1992, 38, 327-334.

78. Li, H.; Li, H.; Cheng, X., "A comparative study of Co-PPy-MWCNTs as a non- noble catalyst for PEM fuel cells." ECS Transactions 2008, 16, (2), 377-383.

79. Garsuch, A.; d'Eon, R.; Dahn, T.; Klepel, O.; Garsuch, R. R.; Dahn, J. R., "Oxygen reduction behavior of highly porous non-noble metal catalysts prepared by a template-assisted synthesis route." J. Electrochem. Soc. 2008, 155, B236-B243.

80. Martinez Millan, W.; Toledano Thompson, T.; Arriaga, L. G.; Smit, M. A., "Characterization of composite materials of electroconductive polymer and cobalt as electrocatalysts for the oxygen reduction reaction." Int. J. Hydrogen Energ. 2009, 34, 694-702.

81. Ye, S.; Vijh, A. K., "Cobalt-carbonized aerogel nanocomposites electrocatalysts for the oxygen reduction reaction." Int. J. Hydrogen Energ. 2005, 30, 1011-1015.

82. Fournier, J.; Lalande, G.; Cote, R.; Guay, D.; Dodelet, J. P., "Activation of various Fe-based precursors on carbon black and graphite supports to obtain catalysts for the reduction of oxygen in fuel cells." J. Electrochem. Soc. 1997, 144, (1), 218.

83. Faubert, G., Cote, R., Guay, D., Dodelet, J. P., Denes, G., Poleunis, C., Bertrand, P., "Activation and characterization of Fe-based catalysts for the reduction of oxygen in polymer electrolyte fuel cells." Electrochim. Acta 1998, 43, (14-15), 1969.

84. Cote, R.; Lalande, G.; Guay, D.; Dodelet, J. P.; Denes, G., "Influence of nitrogen- containing precursors on the electrocatalytic activity of heat-treated Fe(OH)2 on carbon black for O2 reduction." J. Electrochem. Soc. 1998, 145, (7), 2411.

286

85. Wang, H.; Cote, R.; Faubert, G.; Guay, D.; Dodelet, J. P., "Effect of the pre- treatment of carbon black supports on the activity of Fe-based electrocatalysts for the reduction of oxygen." J. Phys. Chem. B 1999, 103, (2042), 2042.

86. Lalande, G.; Cote, R.; Guay, D.; Dodelet, J. P.; Weng, L. T.; Bertrand, P., "Is nitrogen important in the formation of Fe-based catalysts for oxygen reduction in solid polymer fuel cells?" Electrochim. Acta 1997, 42, (9), 1379.

87. Lalande, G.; Guay, D.; Dodelet, J. P., "Electroreduction of oxygen in polymer electrolyte fuel cells by activated carbon coated cobalt nanocrystallites produced by electric arc discharge." Chem. Mater. 1997, 9, 784-790.

88. Cote, R.; Lalande, G.; Faubert, G.; Guay, D.; Dodelet, J. P.; Denes, G., "Non- noble metal-based catalysts for the reduction of oxygen in polymer electrolyte fuel cells." J. New Mat. Electr. Sys. 1998, 1, 7-16.

89. TamizhMani, G.; Dodelet, J. P.; Dignard-Bailey, L., "A rapid half-cell technique for the pre-screening of polymer fuel cell catalysts." J. Electroanal. Chem. 1998, 444, 121-125.

90. Faubert, G., Côté, R., Dodelet, J.P., Lefèvre, M., Bertrand, P., "Oxygen reduction catalysts for polymer electrolyte fuel cells from the pyrolysis of FeII acetate adsorbed on 3,4,9,10-perylenetetracarboxylic dianhydride." Electrochim. Acta 1999, 44, (15), 2589.

91. He, P.; Lefebvre, M.; Faubert, G.; Dodelet, J. P., "Oxygen reduction catalysts for polymer electrolyte fuel cells from the pyrolysis of various transition metal acetates adsorbed on 3,4,9,10-perylenetetracarboxylic dianhydride." J. New Mat. Electr. Sys. 1999, 2, 243-251.

92. Lefevre, M.; Dodelet, J. P.; Bertrand, P., "O2 reduction in PEM fuel cells: Activity and active site structural information for catalysts obtained by the pyrolysis at high temperatures of Fe precursors." J. Phys. Chem. B 2000, 104, 11238.

93. Lefevre, M.; Dodelet, J. P.; Bertrand, P., "Molecular oxygen reduction in PEM fuel cells: Evidence for the simultaneous presence of two active sites in Fe-based catalysts." J. Phys. Chem. B 2002, 106, 8705-8713.

94. Jaouen, F.; Marcotte, S.; Dodelet, J.-P.; Lindbergh, G., "Oxygen reduction catalysts for polymer electrolyte fuel cells from the pyrolysis of iron acetate adsorbed on various carbon supports." J. Phys. Chem. B 2003, 107, 1376-1386.

95. Lefevre, M.; Dodelet, J.-P., "Fe-based catalysts for the reduction of oxygen in polymer electrolyte membrane fuel cell conditions: determination of the amount of

287 peroxide released during electroreduction and its influence on the stability of the catalysts." Electrochim. Acta 2003, 48, 2749-2760.

96. Marcotte, S.; Villers, D.; Guillet, N.; Roue, L.; Dodelet, J. P., "Electroreduction of oxygen on Co-based catalysts: determination of the parameters affecting the two-electron transfer reaction in an acid medium." Electrochim. Acta 2004, 50, 179-188.

97. Villers, D.; Jacques-Bedard, X.; Dodelet, J. P., "Fe-based catalysts for oxygen reduction in PEM fuel cells - pretreatment of the carbon support." J. Electrochem. Soc. 2004, 151, (9), A1507-A1515.

98. Grigoriu, M. A.; Jaouen, F.; Villers, D.; Dodelet, J. P. In Catalysts for oxygen reduction in PEM fuel cells based on Co/Fe porphyrins/acetates, 207th Electrochemical Society Meeting, Quebec City, Canada, Quebec City, Canada.

99. Jaouen, F.; Dodelet, J. P. In Carbon structure and activity of non-noble catalysts for oxygen reduction in PEMFC, 207th Electrochemical Society Meeting, Quebec City, Canada, Quebec City, Canada.

100. Lefevre, M.; Dodelet, J. P.; Bertrand, P., "Molecular oxygen reduction in PEM fuel cell conditions: ToF-SIMS analysis of Co-based electrocatalysts." J. Phys. Chem. B 2005, 109, 16718-16724.

101. Guillet, N.; Roue, L.; Marcotte, S.; Villers, D.; Dodelet, J. P.; Chhim, N.; Trevin, S., "Electrogeneration of hydrogen peroxide in acid medium using pyrolyzed cobalt- based catalysts: Infulence of the cobalt content on the electrode performance." J. Appl. Electrochem. 2006, 36, 863-870.

102. Jaouen, F.; Charreteur, F.; Dodelet, J. P., "Fe-based catalysts for oxygen reduction in PEMFCs: Importance of the disordered phase of the carbon support." J. Electrochem. Soc. 2006, 153, A689-A698.

103. Jaouen, F.; Lefevre, M.; Dodelet, J.-P.; Cai, M., "Heat-treated Fe/N/C catalysts for O2 electroreduction: Are active sites hosted in micropores?" J. Phys. Chem. B 2006, 110, 5553-5558.

104. Medard, C.; Lefevre, M.; Dodelet, J. P.; Jaouen, F.; Lindbergh, G., "Oxygen reduction by Fe-based catalysts in PEM fuel cell conditions: Activity and selectivity of the catalysts obtained with two Fe precursors and various carbon supports." Electrochim. Acta 2006, 51, 3202-3213.

105. Zagal, J. H.; Paez, M. A.; Silva, J. F., "Fundamental aspects on the catalytic activity of metallomacrocyclics for the electrochemical reduction of O2." In N4-

288

Macrocyclic Metal Complexes, Zagal, J. H.; Bedioui, F.; Dodelet, J.-P., Eds. Springer Science+Business Median, Inc.: New York, 2006; pp 41-82.

106. Herranz, J.; Lefevre, M.; Larouche, N.; Stansfield, B.; Dodelet, J.-P., "Step-by- step synthesis of non-noble metal electrocatalysts for O2 reduction under proton exchange membrane fuel cell conditions." J. Phys. Chem. C 2007, 111, 19033-19042.

107. Jaouen, F.; Dodelet, J.-P., "Average turn-over frequency of O2 electro-reduction for Fe/N/C and Co/N/C catalysts in PEFCs." Electrochim. Acta 2007, 52, 5975-5984.

108. Ruggeri, S.; Dodelet, J.-P., "Influence of structural properties of pristine carbon blacks on activity of Fe/N/C cathode catalysts for PEFCs." J. Electrochem. Soc. 2007, 154, B761-B769.

109. Bonakdarpour, A.; Lefevre, M.; Yang, R.; Jaouen, F.; Dahn, T.; Dodelet, J.-P.; Dahn, J. R., "Impact of loading in RRDE experiments on Fe-N-C catalysts: Two- or four- electron oxygen reduction?" Electrochem. Solid-State Lett. 2008, 11, (6), B105-B108.

110. Charreteur, F.; Jaouen, F.; Ruggeri, S.; Dodelet, J.-P., "Fe/N/C non-precious catalysts for PEM fuel cells: Influence of the structural parameters of pristine commercial carbon blacks on their activity for oxygen reduction." Electrochim. Acta 2008, 53, 2925-2938.

111. Charreteur, F.; Ruggeri, S.; Jaouen, F.; Dodelet, J. P., "Increasing the activity of Fe/N/C catalysts in PEM fuel cell cathodes using carbon blacks with a high-disordered carbon content." Electrochim. Acta 2008, 53, 6881-6889.

112. Lefevre, M.; Dodelet, J.-P., "Fe-based electrocatalysts made with microporous pristine carbon black supports for the reduction of oxygen in PEM fuel cells." Electrochim. Acta 2008, 53, 8269-8276.

113. Proietti, E.; Dodelet, J. P., "Ballmilling of carbon supports to enhance the performance of Fe-based electrocatalysts for oxygen reduction in PEM fuel cells." ECS Transactions 2008.

114. Proietti, E.; Ruggeri, S.; Dodelet, J.-P., "Fe-based electrocatalysts for oxygen reduction in PEMFCs using ballmilled graphite powder as a carbon support." J. Electrochem. Soc. 2008, 155, B340-B348.

115. Zhang, G.; Sun, S.; Yang, D.; Dodelet, J.-P.; Sacher, E., "The surface analytical characterization of carbon fibers functionalized by H2SO4/HNO3." Carbon 2008, 46, 196- 205.

289

116. Barazzouk, S.; Lefevre, M.; Dodelet, J.-P., "Oxygen reduction in PEM fuel cells: Fe-based electrocatalysts made with high surface area activated carbon supports." J. Electrochem. Soc. 2009, 156, B1466-B1474.

117. Charreteur, F.; Jaouen, F.; Dodelet, J.-P., "Iron porphyrin-based cathode catalysts for PEM fuel cells: Influence of pyrolysis gas on activity and stability." Electrochim. Acta 2009, 54, 2009.

118. Herranz, J.; Jaouen, F.; Dodelet, J.-P., "Electrochemical evidence of two types of active sites for oxygen reduction in Fe-based catalysts." ECS Transactions 2009, 25, 117- 128.

119. Herranz, J.; Lefevre, M.; Dodelet, J.-P., "Metal-precursor adsorption effects on Fe-based catalysts for oxygen reduction in PEM fuel cells." J. Electrochem. Soc. 2009, 156, B593-B601.

120. Jaouen, F.; Dodelet, J.-P., "O2 reduction mechanism on non-noble metal catalysts for PEM fuel cells. Part I: Experimental rates of O2 electroreduction, H2O2 electroreduction, and H2O2 disproportionation." J. Phys. Chem. C 2009, 113, 15422- 15432.

121. Meng, H.; Jaouen, F.; Proietti, E.; Lefevre, M.; Dodelet, J.-P., "pH-effect on oxygen reduction activity of Fe-based electrocatalysts." Electrochem. Commun. 2009, 11, 1986-1989.

122. Birry, L.; Zagal, J. H.; Dodelet, J. P., "Does CO poison Fe-based catalysts for ORR?" Electrochem. Commun. 2010, 12, 628-631.

123. Meng, H.; Larouche, N.; Lefevre, M.; Jaouen, F.; Stansfield, B.; Dodelet, J. P., "Iron porphyrin-based cathode catalysts for polymer electrolyte membrane fuel cells: effect of NH3 and Ar mixtures as pyrolysis gases on catalytic activity and stability." Electrochim. Acta 2010, 55, 6450-6461.

124. Gewirth, A. A.; Thorum, M. S., "Electroreduction of dioxygen for fuel-cell applications: Materials and challenges." Inorg. Chem. 2010, 49, 3557-3566.

125. Su, D. S.; Zhang, J.; Frank, B.; Thomas, A.; Wang, X.; Paraknowitsch, J.; Schlogl, R., "Metal-free heterogeneous catalysis for sustainable chemistry." ChemSusChem 2010, 3, 169-180.

126. Centi, G.; Perathoner, S., "Problems and perspectives in nanostructured carbon- based electrodes for clean and sustainable energy." Catal. Today 2010, 150, 151-162.

290

127. Bezerra, C. W. B.; Zhang, L.; Lee, K.; Liu, H.; Marques, A. L. B.; Marques, E. P.; Wang, H.; Zhang, J., "A review of Fe-N/C and Co-N/C catalysts for the oxygen reduction reaction." Electrochim. Acta 2008, 53, 4937-4951.

128. Zhang, L.; Zhang, J.; Wilkinson, D. P.; Wang, H., "Progress in preparation of non-noble electrocatalysts for PEM fuel cell reactions." J. Power Sources 2006, 156, 171-182.

129. Koslowski, U. I.; Abs-Wurmbach, I.; Fiechter, S.; Bogdanoff, P., "Nature of the catalytic centers of porphyrin-based electrocatalysts for the ORR: A correlation of kinetic current density with site density of Fe-N4 centers." J. Phys. Chem. C 2008, 112, 15356- 15366.

130. Toebes, M. L.; van Heeswijk, J. M. P.; Bitter, J. H.; van Dillen, A. J.; de Jong, K. P., "The influence of oxidation on the texture and the number of oxygen-containing surface groups of carbon nanofibers." Carbon 2004, 42, 307-315.

131. Olson, T. S.; Chapman, K.; Atanassov, P., "Non-platinum cathode catalyst layer composition for single membrane electrode assembly proton exchange membrane fuel cell." J. Power Sources 2008, 183, 557-563.

132. Matter, P. H.; Ozkan, U. S., "Non-metal catalysts for dioxygen reduction in an acidic electrolyte." Catal. Lett. 2006, 109, (3-4), 115-123.

133. Matter, P. H.; Wang, E.; Arias, M.; Biddinger, E. J.; Ozkan, U. S., "Oxygen reduction reaction catalysts prepared from acetonitrile pyrolysis over alumina supported metal particles." J. Phys. Chem. B 2006, 110, 18374-18384.

134. Khavrus, V. O.; Leonhardt, A.; Hampel, S.; Taschner, C.; Muller, C.; Gruner, W.; Oswald, S.; Strizhak, P. E.; Buchner, B., "Single-step synthesis of metal-coated well- aligned CNx nanotubes using an aerosol technique." Carbon 2007 45, 2889-2896.

135. Garsuch, A.; Yang, R.; Bonakdarpour, A.; Dahn, J. R., "The effect of boron doping into Co-C-N and Fe-C-N electrocatalysts on the oxygen reduction reaction." Electrochim. Acta 2008, 53, 2423-2429.

136. Boehm, H. P., "Some aspects of the surface chemistry of carbon blacks and other carbons." Carbon 1994, 32, 759-769.

137. Fanning, P. E.; Vannice, M. A., "A DRIFTS study of the formation of surface groups on carbon by oxidation." Carbon 1993, 31, 721-730.

291

138. Krishnankutty, N.; Vannice, M. A., "Effect of pretreatment on surface area, porosity, and adsorption properties of a carbon black." Chem. Mater. 1995, 7, 754-763.

139. Alexander, M. R.; Jones, F. R., "Effect of electrocatalytic oxidation on the surface chemistry of Type A carbon fibres -- Part I, X-ray photoelectron spectrscopy." Carbon 1994, 32, 785-794.

140. Fletcher, A. J.; Uygur, Y.; Thomas, K. M., "Role of surface functional groups in the adsorption kinetics of water vapor on microporous activated carbons." J. Phys. Chem. C 2007, 111, 8349-8359.

141. Ros, T. G.; van Dillen, A. J.; Geus, J. W.; Koningsberger, D. C., "Surface oxidation of carbon nanofibres." Chemistry - A European Journal 2002, 8, 1151-1162.

142. Kangasniemi, K. H.; Condit, D. A.; Jarvi, T. D., "Characterization of vulcan electrochemically oxidized under simulated PEM fuel cell conditions." J. Electrochem. Soc. 2004, 151, (4), E125-E132.

143. Haydar, S.; Moreno-Castilla, C.; Ferro-Garcia, M. A.; Carrasco-Marin, F.; Rivera- Utrilla, J.; Perrard, A.; Joly, J. P., "Regularities in the temperature-programmed desorption spectra of CO2 and CO from activated carbons." Carbon 2000, 38, 1297-1308.

144. Pittman, C. U., Jr.; Jiang, W.; He, G.-R.; Gardner, S. D., "Oxygen plasma and isobutylene plasma treatments of carbon fibers: Determination of surface functionality and effects on composite properties." Carbon 1998, 36, 25-37.

145. Rasheed, A.; How, J. Y.; Dadmun, M. D.; Britt, P. F., "The efficiency of the oxidation of carbon nanofibers with various oxidizing agents." Carbon 2007, 45, 1072- 1080.

146. Felten, A.; Ghijsen, J.; Pireaux, J.-J.; Johnson, R. L.; Whelan, C. M.; Liang, D.; Van Tendeloo, G.; Bittencourt, C., "Effect of oxygen rf-plasma on electronic properties of CNTs." J. Phys. D. Appl. Phys. 2007, 40, 7379-7382.

147. Hopkins, R. J.; Tivanski, A. V.; Marten, B. D.; Gilles, M. K., "Chemical bonding and structure of black carbon reference materials and individual carbonaceous atmospheric aerosols." J. Aerosol Sci. 2007, 38, 573-591.

148. Salame, I. I.; Bandosz, T. J., "Surface chemistry of activated carbons: Combining the results of temperature-programmed desporption, Boehm, and potentiometric titrations." J. Colloid Interf. Sci. 2001, 240, 252-258.

149. Nabae, Y.; Yamanaka, I.; Otsuka, K., "Electro-catalysis of Cu/Carbon cathode for the reduction of O2 during fuel-cell reactions." Appl. Catal. A-Gen. 2005, 280, 149-155.

292

150. Frysz, C. A.; Chung, D. D. L., "Improving the electrochemical behavior of carbon black and carbon filaments by oxidation." Carbon 1997, 35, 1111-1127.

151. Maksimova, N. I.; Roddatis, V. V.; Mestl, G.; Ledoux, M.; Schlogl, R., "Oxidative dehydrogenation of ethylbenzene to styrene over carbonaceous materials." Eurasian Chemico-Technological Journal 2000, 2, (3-4), 231-236.

152. Natarajan, S. K.; Cossement, D.; Hamelin, J., "Synthesis and characterization of carbon nanostructures as catalyst support for PEMFCs." J. Electrochem. Soc. 2007, 154, B310-B315.

153. Figueiredo, J. L.; Pereira, M. F. R.; Freitas, M. M. A.; Orfao, J. J. M., "Modification of the surface chemistry of activated carbons." Carbon 1999, 37, 1379- 1389.

154. Takoaka, M.; Yokokawa, H.; Takeda, N., "The effect of treatment of activated carbon by H2O2 or HNO3 on the decomposition of pentachlorobenzene." Appl. Catal. B- Environ. 2007, 74, 179-186.

155. de la Puente, G.; Pis, J. J.; Menendez, J. A.; Grange, P., "Thermal stability of oxygenated functions in activated carbons." J. Anal. Appl. Pyrol. 1997, 43, (2), 125-138.

156. Rios, R. R. V. A.; Alves, D. E.; Dalmazio, I.; Bento, S. F. V.; Donnici, C. L.; Lago, R. M., "Tailoring activated carbon by surface chemical modification with O, S, and N containing molecules." Materials Research 2003, 6, (2), 129-135.

157. Hamerton, I.; Hay, J. N.; Howlin, B. J.; Jones, J. R.; Lu, S.-Y.; Webb, G.; Bader, M. G.; Brown, A. M.; Watts, J. F., "ToF SIMS and XPS studies of carbon fiber surface during electrolytic oxidation of 17O/18O enriched aqueous electrolytes." Chem. Mater. 1997, 9, 1972-1977.

158. Wu, Z.; Pittman, C. U., Jr.; Gardner, S. D., "Nitric acid oxidation of carbon fibers and the effects of subsequent treatment in refluxing aqueous NaOH." Carbon 1995, 33, (5), 597-605.

159. Yue, Z. R.; Jiang, W.; Wang, L.; Gardner, S. D.; Pittman, C. U., Jr., "Surface characterization of electrochemically oxidized carbon fibers." Carbon 1999, 37, 1785- 1796.

160. Alexander, M. R.; Jones, F. R., "Effect of electrocatalytic oxidation upon the surface chemistry of carbon fibers -- Part II, Analysis of derivatised surface functionalities by XPS, and ToF SIMS." Carbon 1995, 33, 569-580.

293

161. Alexander, M. R.; Jones, F. R., "Effect of electrolytic oxidation upon the surface chemistry of type A carbon fibres: III. Chemical state, source and location of surface nitrogen." Carbon 1996, 34, 1093-1102.

162. Wang, Z. M.; Yamashita, N.; Wang, Z. X.; Hoshinoo, K.; Kanoh, H., "Air oxidation effects on microporosity, surface property, and CH4 adsorptivity of pitch-based activated carbon fibers." J. Colloid Interf. Sci. 2004, 276, (1), 143-150.

163. Zhou, J.-H.; Sui, Z.-J.; Zhu, J.; Li, P.; Chen, D.; Dai, Y.-C.; Yuan, W.-K., "Characterization of surface oxygen complexes on carbon nanofibers by TPD, XPS and FT-IR." Carbon 2007, 45, 785-796.

164. Xia, W.; Wang, Y.; Bergstraber, R.; Kundu, S.; Muhler, M., "Surface characterization of oxygen-functionalized multi-walled carbon nanotubes by high- resolution X-ray photoelectron spectroscopy and temperature-programmed desorption." Appl. Surf. Sci. 2007, 254, 247-250.

165. McCaldin, S.; Bououdina, M.; Grant, D. M.; Walker, G. S., "The effect of processing conditions on carbon nanostructures formed on an iron-based catalyst." Carbon 2006, 44, (11), 2273-2280.

166. Choi, W.-K.; Park, S.-G.; Takahashi, H.; Cho, T.-H., "Purification of carbon nanofibers with hydrogen peroxide." Synthetic Met. 2003, 139, 39-42.

167. Dicks, A. L., "The role of carbon in fuel cells." J. Power Sources 2006, 156, 128- 141.

168. Guha, A.; Lu, W.; Zawodzinski Jr., T. A.; Schiraldi, D. A., "Surface-modified carbons as platinum catalyst support for PEM fuel cells." Carbon 2007, 45, 1506-1517.

169. Subramanian, N. P.; Kumaraguru, S. P.; Colon-Mercado, H.; Kim, H.; Popov, B. N.; Black, T.; Chen, D. A., "Studies on Co-based catalysts supported on modified carbon substrates for PEMFC cathodes." J. Power Sources 2006, 157, 56-63.

170. Kumaraguru, S. P.; Subramanian, N. P.; Colon, H.; Popov, B. N., "Non noble metal catalysts for oxygen reduction reaction in acidic medium." ECS Transactions 2006, 1, (6), 27-39.

171. Oh, H.-S.; Kim, K.; Ko, Y.-J.; Kim, H., "Effect of chemical oxidation of CNFs on the electrochemical carbon corrosion in polymer electrolyte membrane fuel cells." Int. J. Hydrogen Energ. 2010, 35, 701-708.

294

172. Strelko, V. V.; Kartel, N. T.; Dukhno, I. N.; Kuts, V. S.; Clarkson, R. B.; Odintsov, B. M., "Mechanism of reductive oxygen adsorption on active carbons with various surface chemistry." Surf. Sci. 2004, 548, 281-290.

173. Woods, M. P.; Biddinger, E. J.; Matter, P. H.; Mirkelamoglu, B.; Ozkan, U. S., "Correlation between oxygen reduction reaction and oxidative dehydrogenation activities over nanostructured carbon catalysts." Catal. Lett. 2010, 136, 1-8.

174. Boehm, H. P., "Surface oxides on carbon and their analysis: a critical assessment." Carbon 2002, 40, 145-149.

175. Shao, Y.; Kou, R.; Wang, J.; Kwak, J. H.; Viswanathan, V.; Wang, Y.; Liu, J.; Lin, Y., "Fast test for the durability of PEM fuel cell catalysts." ECS Transactions 2008, 16, (2), 361-366.

176. Datsyuk, V.; Kalyva, M.; Papagelis, K.; Parthenios, J.; Tasis, D.; Siokou, A.; Kallitsis, I.; Galiotis, C., "Chemical oxidation of multiwalled carbon nanotubes." Carbon 2008, 46, 833-840.

177. Strelko, V. V.; Kuts, V. S.; Thrower, P. A., "On the mechanism of possible influence of heteroatoms of nitrogen, boron, and phosphorus in a carbon matrix on the catalytic activity of carbons in electron transfer reactions." Carbon 2000, 38, 1499-1525.

178. Schilling, T.; Bron, M., "Oxygen reduction at Fe-N-modified multi-walled carbon nanotubes in acidic electrolyte." Electrochim. Acta 2008, 53, 5379-5385.

179. Bron, M.; Bogdanoff, P.; Fiechter, S.; Tributsch, H., "Enhancement of oxygen electroreduction activity via surface modification of carbon supported ruthenium nanoparticles: A new class of electrocatalysts." J. Electroanal. Chem. 2005, 578, 339- 344.

180. Maruyama, J.; Abe, I., "Fuel cell cathode catalyst with Heme-like structure formed from nitrogen of glycine and iron." J. Electrochem. Soc. 2007, 154, B297-B304.

181. Tyurin, V. S.; Radyushkina, K. A.; Levina, O. A.; Tarasevich, M. R., "Electrocatalytic properties of a composite based on cobalt porphyrin pyropolymer and nafion." Russ J. Electrochem. 2001, 37, 843-847.

182. Anderson, A. B.; Sidik, R. A., "Oxygen electroreduction on FeII and FeIII coordinated to N4 chelates. Reversible potentials for the intermediate steps from quantum theory." J. Phys. Chem. B 2004, 108, 5031-5035.

183. Jian, M.; Chou, S.-h.; Siedle, A., "In search for structure of active site in iron- based oxygen reduction electrocatalysts." J. Phys. Chem. B 2006, 110, 4179-4185.

295

184. Liu, G. C.-K.; Dahn, J. R., "Fe-N-C oxygen reduction catalysts supported on vertically aligned carbon nanotubes." Appl. Catal. A-Gen. 2008, 347, 43-49.

185. Kumar, S. A.; Chen, S.-M., "Direct electrochemistry and electrocatalysis of myoglobin on redox-active self-assembling monolayers derived from nitroaniline modified electrode." Biosens. Bioelectron. 2007, 22, 3042-3050.

186. Barsukov, V. Z.; Khomenko, V. G.; Katashinskii, A. S.; Motronyuk, T. I., "Catalytic activity of polyaniline in the molecular oxygen reduction: its nature and mechanism." Russ J. Electrochem. 2004, 40, (11), 1170-1173.

187. Kumar, S. A.; Chen, S.-M., "Electrocatalytic reduction of oxygen and hydrogen peroxide at poly(p-aminobenzene sulfonic acid)-modified glassy carbon electrodes." J. Mol. Catal. A - Chem. 2007, 278, 244-250.

188. Subramanian, N. P.; Li, X.; Nallathambi, V.; Kumaraguru, S. P.; Colon-Mercado, H.; Wu, G.; Lee, J.-W.; Popov, B. N., "Nitrogen-modified carbon-based catalysts for oxygen reduction reaction in polymer electrolyte membrane fuel cells." J. Power Sources 2009, 188, 38-44.

189. Prehn, K.; Warburg, A.; Schilling, T.; Bron, M.; Schulte, K., "Towards nitrogen- containing CNTs for fuel cell electrodes." Compos. Sci. Technol. 2009, 69, 1570-1579.

190. Lyth, S. M.; Nabae, Y.; Moriya, S.; Kuroki, S.; Kakimoto, M.-a.; Ozaki, J.-i.; Miyata, S., "Carbon nitride as a nonprecious catalyst for electrochemical oxygen reduction." J. Phys. Chem. C 2009, 113, 20148-20151.

191. Iwazaki, T.; Obinata, R.; Sugimoto, W.; Takasu, Y., "High oxygen-reduction activity of silk-derived activated carbon." Electrochem. Commun. 2009, 11, 376-378.

192. Matter, P. H.; Wang, E.; Arias, M.; Biddinger, E. J.; Ozkan, U. S., "Oxygen reduction reaction activity and surface properties of nanostructured nitrogen-containing carbon." J. Mol. Catal. 2007, 264, (1-2), 73-81.

193. Matter, P. H.; Zhang, L.; Ozkan, U. S., "The role of nanostructure in nitrogen- containing carbon catalysts for the oxygen reduction reaction." J. Catal. 2006, 239, 83- 96.

194. Wang, P.; Ma, Z.; Zhao, Z.; Jia, L., "Oxygen reduction on the electrocatalysts based on pyrolyzed non-noble metal/poly-o-phenylenediamine/carbon black composites: New insight into the active sites." J. Electroanal. Chem. 2007, 611, 87-95.

296

195. Liu, H.; Zhang, Y.; Li, R.; Sun, X.; Desilets, S.; Abou-Rachid, H.; Jaidann, M.; Lussier, L.-S., "Structural and morphological control of aligned nitrogen-doped carbon nanotubes." Carbon 2010, 48, 1498-1507.

196. Khavrus, V. O.; Ibrahim, E. M. M.; Leonhardt, A.; Hampel, S.; Oswald, S.; Taschner, C.; Buchner, B., "Conditions of simultaneous growth and separation of single- and multiwalled carbon nanotubes." J. Phys. Chem. C 2010, 2010, 843-848.

197. Garcia-Garcia, F. R.; Alvarez-Rodriguez, J.; Rodriguez-Ramos, I.; Guerrero-Ruiz, A., "The use of carbon nanotubes with and without nitrogen doping as support for ruthenium catalysts in the ammonia decomposition reaction." Carbon 2010, 48, 267-276.

198. Tang, Y.; Allen, B. L.; Kauffman, D. R.; Star, A., "Electrocatalytic activity of nitrogen-doped carbon nanotube cups." J. Am. Chem. Soc. 2009, 131, 13200-13201.

199. van Dommele, S.; Romero-Izquirdo, A.; Brydson, R.; de Jong, K. P.; Bitter, J. H., "Tuning nitrogen functionalities in catalytically grown nitrogen-containing carbon nanotubes." Carbon 2008, 46, 138-148.

200. Bulusheva, L. G.; Okotrub, A. V.; Kudashov, A. G.; Shubin, Y. V.; Shlyakhova, E. V.; Yudanov, N. F.; Pazhetnov, E. M.; Boronin, A. I.; Vyalikh, D. V., "Effect of Fe/Ni catalyst composition on nitrogen doping and field emission properties of carbon nanotubes." Carbon 2008, 46, 864-869.

201. Maldonado, S.; Morin, S.; Stevenson, K. J., "Structure, composition, and chemical reactivity of carbon nanotubes by selective nitrogen doping." Carbon 2006, 44, 1429-1437.

202. Trasobares, S.; Stephan, O.; Colliex, C.; Hsu, W. K.; Kroto, H. W.; Walton, D. R. M., "Compartmentalized CNx nanotubes: Chemistry, morphology, and growth." J. Chem. Phys. 2002, 116, (20), 8966-8972.

203. Brichka, S. Y.; Prikhod'ko, G. P.; Brichka, A. V.; Ogenko, V. M.; Chuiko, A. A., "Matrix synthesis of N-containing carbon nanotubes." Theoretical and Experimental Chemistry 2002, 38, (2), 114-117.

204. Nakajima, T.; Koh, M., "Synthesis of high crystalline carbon-nitrogen layered compounds by CVD using nickel and cobalt catalysts." Carbon 1997, 35, (2), 203.

205. Kvon, R.; Il'inich, G.; Chuvilin, A.; Likholobov, V., "XPS and TEM study of new carbon material: N-containing catalytic filamentous carbon." J. Mol. Catal. A - Chem. 2000, 158, 413.

297

206. Sen, R.; Satishkumar, B. C.; Govindaraj, A.; Harikumar, K. R.; Renganathan, M. K.; Rao, C. N. R., "Nitrogen-containing carbon nanotubes." Materials Chemistry Communication 1997, 7, (12), 2335-2337.

207. Niwa, H.; Horiba, K.; Harada, Y.; Oshima, M.; Ikeda, T.; Terakura, K.; Ozaki, J.- i.; Miyata, S., "X-ray absorption analysis of nitrogen contribution to oxygen reduction reaction in carbon alloy cathode catalysts for polymer electrolyte fuel cells." J. Power Sources 2009, 187, 93-97.

208. Wang, X.; Lee, J. S.; Zhu, Q.; Liu, J.; Wang, Y.; Dai, S., "Ammonia-treated ordered mesoporous carbons as catalytic materials for oxygen reduction reaction." Chem. Mater. 2010, 22, 2178-2180.

209. Pakula, M.; Swiatkowski, A.; Biniak, S., "Electrochemical behaviour of modified activated carbons in aqueous and nonaqueous solutions." J. Appl. Electrochem. 1995, 25, 1038-1044.

210. Lim, S. H.; Elim, H. I.; Gao, X. Y.; Wee, A. T. S.; Ji, W.; Lee, J. Y.; Lin, J., "Electronic and optical properties of nitrogen-doped multiwalled carbon nanotubes." Phys. Rev. B 2006, 73, (4), 045402/1-045402/6.

211. Biddinger, E. J.; von Deak, D.; Ozkan, U. S., "Nitrogen-containing carbon nanostructures as oxygen-reduction catalysts." Top. Catal. 2009, 52, 1566-1574.

212. Ozaki, J.-i.; Kimura, N.; Anahara, T.; Oya, A., "Preparation and oxygen reduction activity of BN-doped carbons." Carbon 2007, 45, 1847-1853.

213. Terrones, M.; Ajayan, P. M.; Banhart, F.; Blase, X.; Carroll, D. L.; Charlier, J. C.; Czerw, R.; Foley, B.; Grobert, N.; Kamalakaran, R.; Kohler-Redlich, P.; Ruhle, M.; Seeger, T.; Terrones, H., "N-doping and coalescence of carbon nanotubes: synthesis and electronic properties." Applied Physics A: Materials Science & Processing 2002, 74, (3), 355-361.

214. Terrones, M.; Terrones, H.; Grobert, N.; Hsu, W. K.; Zhu, Y. Q.; Hare, J. P.; Kroto, H. W.; Walton, D. R. M.; Kohler-Redlich, P.; Ruhle, M.; Zhang, J. P.; Cheetham, A. K., "Efficient route to large arrays of CNx nanofibers by pyrolysis of ferrocene/melamine mixtures." Appl. Phys. Lett. 1999, 75, (25), 3932-3934.

215. Zhu, Q.; Money, S. L.; Russell, A. E.; Thomas, K. M., "Determination of the fate of nitrogen functionality in carbonaceous materials during pyrolysis and combustion using X-ray absorption near edge structure spectroscopy." Langmuir 1997, 13, 2149- 2157.

298

216. Liu, G.; Li, X.; Ganesan, P.; Popov, B. N., "Development of non-precious metal oxygen-reduction catalysts for PEM fuel cells based on N-doped ordered porous carbon." Appl. Catal. B-Environ. 2009, 93, 156-165.

217. Alonso-Vante, N.; Cattarin, S.; Musiani, M., "Electrocatalysis of O2 reduction at polyaniline + molybdenum-doped ruthenium selenide composite electrodes." J. Electroanal. Chem. 2000, 481, (2), 200-207.

218. Wu, A.; Vanancio, E.; MacDiarmid, A. G., "Polyaniline and polypyrrole oxygen reversible electrodes." Synthetic Met. 2007, 157, 303-310.

219. Wu, G.; Chen, Z.; Artyushkova, K.; Garzon, F. H.; Zelenay, P., "Polyaniline- derived non-precious catalyst for the polymer electrolyte fuel cell cathode." ECS Transactions 2008, 16, 159-170.

220. Lei, Z.; Zhao, M.; Dang, L.; An, L.; Lu, M.; Lo, A.-Y.; Yu, N.; Liu, S.-B., "Structural evolution and electrocatalytic application of nitrogen-doped carbon shells synthesized by pyrolysis of near-monodisperse polyaniline nanospheres." J. Mater. Chem. 2009, 19, 5985-5995.

221. Matzner, S.; Boehm, H. P., "Influence of nitrogen doping on the adsorption and reduction of nitric oxide by activated carbons." Carbon 1998, 36, (11), 1697-1700.

222. Kapteijn, F.; Moulijn, J. A.; Matzner, S.; Boehm, H.-P., "The development of nitrogen functionality in model chars during gasification in CO2 and O2." Carbon 1999, 37, 1143-1150.

223. Gammon, W. J.; Kraft, O.; Reilly, A. C.; Holloway, B. C., "Experimental comparison of N(1s) x-ray photoelectron spectroscopy binding energies of hard and elastic amorphous carbon nitride films with reference organic compounds." Carbon 2003, 41, 1917-1923.

224. Abrasonis, G.; Gago, R.; Jimenez, I.; Kreissig, U.; Kolitsch, A.; Moller, W., "Nitrogen incorporation in carbon nitride films produced by direct and dual ion-beam sputtering." J. Appl. Phys. 2005, 98, 074907.

225. Deng, Z. W.; Souda, R., "Synthesis and thermal decomposition of carbon nitride films prepared by nitrogen ion implantation into graphite." This Solid Films 2002, 406, 46-53

226. Takahagi, T.; Shimada, I.; Fukuhara, M.; Morita, K.; Ishitani, A., "XPS studies on the chemical structure of the stabilizer polyacrylonitrile fiber in the carbon fiber production process." Journal of Polymer Science Part A-General Papers 1986, 24, 3101.

299

227. Bartle, K. D.; Perry, D. L.; Wallace, S., "The functionality of nitrogen in coal and derived liquids: and XPS study." Fuel Process Technol. 1987, 15, 351-361.

228. Jansen, R. J. J.; van Bekkum, H., "XPS of nitrogen-containing functional groups on activated carbon." Carbon 1995, 33, (8), 1021-1027.

229. Jourdain, V.; Simpson, E. T.; Paillet, M.; Kasama, T.; Dunin-Borkowski, R. E.; Poncharal, P.; Zahab, A.; Loiseau, A.; Robertson, J.; Bernier, P., "Periodic inclusion of room-temperature -ferromagnetic metal phosphide nanoparticles in carbon nanotubes." Physical Chemistry B 2006, 110, 9759-9763.

230. Liu, J.; Li, X.; Schrand, A.; Ohashi, T.; Dai, L., "Controlled syntheses of aligned multi-walled carbon nanotubes: catalyst particle size and density control via layer-by- layer assembling." Chem. Mater. 2005, 17, 6599-6604.

231. Saito, T.; Ohshima, S.; Xu, W.-C.; Ago, H.; Yumura, M.; Iijima, S., "Size control of metal nanoparticle catalysts for the gas-phase synthesis of single-walled carbon nanotubes." J. Phys. Chem. B 2005, 109, 10647-10652.

232. Lueking, A. D.; Gutierrez, H. R.; Fonseca, D. A.; Dickey, E., "Structural characterization of exfoliated graphite nanofibers." Carbon 2007, 45, 751-759.

233. Ci, L.; Wei, J.; Wei, B.; Liang, J.; Xu, C.; Wu, D., "Carbon nanofibers and single- walled carbon nanotubes prepared by the floating catalyst method." Carbon 2001, 39, 329-335.

234. Glerup, M.; Castignolles, M.; Holzinger, M.; Hug, G.; Loiseau, A.; Bernier, P., "Synthesis of highly nitrogen-doped multi-walled carbon nanotubes." Chem. Commun. 2003, 2542-2543.

235. Serp, P.; Corrias, M.; Kalck, P., "Carbon nanotubes and nanofibers in catalysis." Applied Catalysis, A: General 2003, 253, (2), 337-358.

236. Mohlala, M. S.; Liu, X.-Y.; Witcomb, M. J.; Coville, N. J., "Carbon nanotube synthesis using ferrocene and ferrocenyl sulfide. The effect of sulfur." Applied Organometallic Chemisty 2007, 21, 275-280.

237. Khanna, V.; Bakshi, B. R.; Lee, L. J., "Carbon nanofiber production: Life cycle energy consumption and environmental impact." Journal of Industrial Ecology 2008, 12, 394-410.

300

238. Yoon, S.-H.; Park, C.-W.; Yang, H.; Korai, Y.; Mochida, I.; Baker, R. T. K.; Rodriguez, N. M., "Novel carbon nanofibers of high graphitization as anodic materials for lithium ion secondary batteries." Carbon 2004, 42, (1), 21-32.

239. Carneiro, O. C.; Kim, M. S.; Yim, J. B.; Rodriguez, N. M.; Baker, R. T. K., "Growth of graphite nanofibers from the iron-copper catalyzed decomposition of CO/H2 mixtures." J. Phys. Chem. B 2003, 107, (18), 4237-4244.

240. Schmidt, T. J.; Gasteiger, H. A., "Rotating thin film method for supported catalysts." In Handbook of Fuel Cells - Fundamentals, Technology and Applications, Wolf Vielstich H.A., a. G., A. L, Ed. John Wiley & Sons: 2003; Vol. 2, pp 317-333.

241. Malakhov, I. V.; Nikitenko, S. G.; Savinova, E. R.; Kochubey, D. I.; Alonso- Vante, N., "In situ EXAFS study to probe active centers of Ru chalcogenide electrocatalysts during oxygen reduction reaction." J. Phys. Chem. B 2002, 106, (7), 1670-1676.

242. Widelov, A., "Pyrolysis of iron and cobalt porphyrins sublimated onto the surface of carbon black as a method to prepare catalysts for O2 reduction." Electrochim. Acta 1993, 38, 2493-2502.

243. Okada, T.; Gokito, M.; Yuasa, M.; Sekine, I., "Oxygen reduction characteristics of heat-treated catalysts based on cobalt-porphyrin ion complexes." J. Electrochem. Soc. 1998, 145, (3), 815-822.

244. Joyner, R. W., "Extended X-ray Absorption Fine Structure (EXAFS) study of cobalt-porphyrin catalysts supported on active carbon." J. Chem. Soc. Farad. T. 1 1982, 78, 1021-1028.

245. van Wingerden, B.; van Veen, J. A. R.; Mensch, C. T. J., "An Extended X-Ray Absorption Fine Structure study of heat-treated cobalt porphyrin catalysts supported on active carbon." J. Chem. Soc. Farad. T. 1 1988, 84, (1), 65-74.

246. Sun, G. Q.; Wang, J. T.; Gupta, S.; Savinell, R., "Iron(III) tetramethoxyphenylporphyrin (FeTMPP-Cl) as electrocatalyst for oxygen reduction in direct methanol fuel cells." J. Appl. Electrochem. 2001, 31, 1025-1031.

247. Alonso-Vante, N., "Novel nanostructured material based on transition-metal compounds for electrocatalysis." In Catalysis and Electrocatalysis at Nanoparticle Surfaces, Wieckowski, A.; Savinova, E. R.; Vayenas, C. G., Eds. Marcel Dekker, Inc.: New York, 2003; pp 931-958.

248. Martins Alves, M. C.; Dodelet, J. P.; Guay, D.; Ladouceur, M.; Tourillon, G., "Origin of the electrocatalytic properties for O2 reduction of some heat-treated

301 polyacrylonitrile and pthalocyanine cobalt compounds adsorbed on carbon black as probed by electrochemistry and X-ray Adsorption Spectroscopy." J. Phys. Chem. 1992, 96, 10898-10905.

249. Ziegelbauer, J. M.; Olson, T. S.; Pylypenko, S.; Alamgir, F.; Jaye, C.; Atanassov, P.; Mukerjee, S., "Direct spectroscopic observation of the structural origin of peroxide generation from Co-based pyrolyzed porphyrins for ORR applications." J. Phys. Chem. C 2008, 112, 8839-8849.

250. Martins Alves, M. C.; Tourillion, G., "Influence of complexation processes on the catalytic properties of some polymer-based cobalt compounds for oxygen electroreduction." J. Phys. Chem. 1996, 100, 7566.

251. Zagal, J. H., "Macrocycles." In Handbook of Fuel Cels - Fundamentals, Technology and Applications, Vielstich, W.; Gasteiger, H.; Lamm, A., Eds. John Wiley & Sons, Ltd.: 2003; Vol. 2: Electrocatalysis.

252. Tanaka, A.; Gupta, S. L.; Tryk, D.; Fierro, C.; Yeager, E. B.; Scherson, D. A., "Electrochemical and spectroscopic aspects of heat-treated transition metal macrocycles as electrocatalysts for oxygen reduction." In Structural Effects in Electrocatalysis and Oxygen Electrochemistry, Scherson, D.; Tryk, D.; Daroux, M.; Xing, X., Eds. The Electrochemical Society Inc.: Pennington, NJ, 1992; Vol. 92-11, pp 555-572.

253. Liu, G.; Li, X.; Ganesan, P.; Popov, B. N., "Studies of oxygen reduction reaction active sites and stability of nitrogen-modified carbon composite catalysts for PEM fuel cells." Electrochim. Acta 2010, 55, 2853-2858.

254. Gruenig, G.; Wiesner, K.; Gamburzev, S.; Iliev, I.; Kaisheva, A., "Investigations of catalysts from the pyrolyzates of cobalt-containing and metal-free dibenzotetraazaannulenes on active carbon for oxygen electrodes in an acid medium." J. Electroanal. Chem. 1983, 159, 155-162.

255. Titov, A. V.; Zapol, P.; Kral, P.; Liu, D.-J.; Iddir, H.; Baishya, K.; Curtiss, L. A., "Catalytic Fe-xN sites in carbon nanotubes." J. Phys. Chem. C 2009, 113, 21629-21634.

256. Pels, J. R.; Kapteijn, F.; Moulijn, J. A.; Zhu, Q.; Thomas, K. M., "Evolution of nitrogen functionalities in carbonaceous materials during pyrolysis." Carbon 1995, 33, 1641-1653.

257. Stanczyk, K.; Dziembaj, R.; Piwowarska, Z.; Witkowski, S., "Transformation of nitrogen structures in carbonization of model compounds determined by XPS." Carbon 1995, 33, 1383-1392.

302

258. Buckley, A. N., "Nitrogen functionality in coals and coal-tar pitch determined by X-ray photoelectron spectroscopy." Fuel Process Technol. 1994, 38, 165-179.

259. Montoya, A.; Gil, J.; Mondragon, F.; Truong, T. N., "Kinetics of nitric oxide desorption from carbonaceous surfaces." Fuel Process Technol. 2002, 77-78, 453-458.

260. Maldonado, S.; Stevenson, K. J., "Influence of nitrogen doping on oxygen reduction electrocatalysis at carbon nanofiber electrodes." J. Phys. Chem. B 2005, 109, 4707-4716.

261. Montoya, A.; Gil, J.; Mondragon, F.; Truong, T. N., "Oxygen adsorption on nitrogen containing carbon surfaces." Fuel Chemistry Division Preprints 2002, 47, (2), 424.

262. Wei, G.; Wainright, J. S.; Savinell, R. F., "Catalytic activity for oxygen reduction reaction of catalysts consisting of carbon, nitrogen, and cobalt." J. New Mat. Electr. Sys. 2000, 3, 121-129.

263. Hsueh, K.-L.; Gonzalez, E. R.; Srinivasan, S., "Electrolyte effects on oxygen reduction kinetics at platinum: A rotating ring-disc electrode analysis." Electrochim. Acta 1983, 28, 691-697.

264. Albery, W. J.; Hitchman, M. L., Ring-disc electrodes. Oxford University Press: London, 1971.

265. Paulus, U. A.; Schmidt, T. J.; Gasteiger, H. A.; Behm, R. J., "Oxygen reduction on a high surface area Pt/Vulcan carbon catayst: A thin-film rotating ring-disk electrode study." J. Electroanal. Chem. 2001, 495, 134-145.

266. Antoine, O.; Durand, R., "RRDE study of oxygen reduction on Pt nanoparticles inside Nafion: H2O2 production in PEMFC cathode conditions." J. Appl. Electrochem. 2000, 30, 839-844.

267. Mayrhofer, K. J. J.; Crampton, A. S.; Wiberg, G. K. H.; Arenz, M., "Analysis of the impact of individual glass constituents on electrocatalysis on Pt electrodes in alkaline solution." J. Electrochem. Soc. 2008, 155, (6), P78-P81.

268. Wang, J.; Yin, G.; Shao, Y.; Zhang, S.; Wang, Z.; Gao, Y., "Effect of carbon black support corrosion on the durability of Pt/C catalyst." J. Power Sources 2007, 171, 331-339.

269. Sompalli, B.; Litteer, B. A.; Gu, W.; Gasteiger, H. A., "Membrane degradation at catalyst layer edges in PEMFC MEAs." The Electrochemical Society 2007, 154, (12), B1349-B1357.

303

270. Ressler, T., "WinXAS: a new software package not only for the analysis of energy-dispersive XAS data." Journal de Physique IV 1997, 7, C2-269.

271. von Deak, D.; Biddinger, E. J.; Luthman, K. A.; Ozkan, U. S., "The effect of phosphorus in CNx catalysts for the oxygen reduction in PEM fuel cells." Carbon 2010, 48, (12), 3637-3639.

272. Cruz-Silva, E.; Cullen, D. A.; Gu, L.; Romo-Herrera, J. M.; Munoz-Sandoval, E.; Lopez-Urias, F.; Sumpter, B. G.; Meunier, V.; Charlier, J.-C.; Smith, D. J.; Terrones, H.; Terrones, M., "Heterodoped nanotubes: synthesis, and characterization of phosphorus- nitrogen doped multiwalled carbon nanotubes." ACS Nano 2008, 2, (3), 441-448.

273. Wang, X.; Chen, X.; Thomas, A.; Fu, X.; Antonietti, M., "Metal-containing carbon nitride compounds: A new functional organic-metal hybrid material." Adv. Mater. 2009, 21, 1609-1612.

274. Huang, H.; Kajiura, H.; Marakami, Y.; Ata, M., "Metal sulfide catalyzed growth of carbon nanofibers and nanotubes." Carbon 2003, 41, 615-618.

275. Tan, C. D.; Baker, R. T. K., "The effect of various sulfides on carbon deposition on nickel-iron particles." Catal. Today 2000, 63, 3-20.

276. Tibbetts, G. G.; Bernardo, C. A.; Gorkiewicz, D. W.; Alig, R. L., "Role of sulfur in the production of carbon fibers in the vapor phase." Carbon 1994, 32, 569-576.

277. Olson, T. S.; Blizanac, B.; Piela, B.; Davey, J. R.; Zelenay, P.; Atanassov, P., "Electrochemical evaluation of porous non-platinum oxygen reduction catalysts for polymer electrolyte fuel cells." Fuel Cells 2009, 5, 547-553.

278. Claeyssens, E.; Fuge, G. M.; Allan, N. L.; May, P. W.; Ashfold, M. N. R., "Phosphorus carbides: theory and experiment." Dalton T. 2004, 3085-3092.

279. Matter, P. H.; Wang, E.; Millet, J.-M. M.; Ozkan, U. S., "Characterization of the iron phase in CNx-based oxygen reduction reaction catalysts." J. Phys. Chem. C 2007, 111, 1444-1450.

280. Kodama, M.; Yamashita, J.; Soneda, Y.; Hatori, H.; Kamegawa, K., "Preparation and electrochemical characteristics of N-enriched carbon foam." Carbon 2007, 45, 1105- 1136.

281. Harris, R. K.; Thompson, T. V.; Norman, P. R.; Pottage, C., "Phosphorus-31 NMR studies of adsorption onto activated carbon." Carbon 1999, 37, 1425-1430.

304

282. Jibril, B.; Al-Maamari, R. S.; Houache, O.; Al-Aamri, M.; Al-Qalhati, A., "Effects of H3PO4 and KOH an pyrolysis of bituminous coal in preparation of activated carbon." Journal of Applied Sciences Research 2007, 3, (11), 1343-1351.

283. Olson, T. S.; Pylypenko, S.; Fulghum, J.; Atanassov, P., "Bifunctional oxygen reduction reaction mechanism on non-platinum catalysts derived from pyrolyzed porphyrins." J. Electrochem. Soc. 2010, 157, B54-B63.

284. Jacobs, G.; Sarkar, A.; Davis, B. H.; Cronauer, D.; Kropf, A. J.; Marshall, C. L., "Fischer-Tropsch synthesis temperature-programmed EXAFS/XANES characterization of the impact of Cu and alkali promoters to iron-based catalysts on the carbide formation rate." Advances in Fischer-Tropsch synthesis, catalysts, and catalysis 2009, 119-146.

285. Maciel, I. O.; Campos-Delgado, J.; Pimenta, M. A.; Terrones, M.; Terrones, H.; Rao, A. M.; Jorio, A., "Boron, nitrogen and phosphorous substitionally doped single-wall carbon nanotubes studied by resonance Raman spectroscopy." Physics Status Solidi B 2009, 246, 11-12.

286. von Deak, D.; Singh, D.; Biddinger, E. J.; King, J. C.; Bayram, B.; Miller, J. T.; Ozkan, U. S., "Investigation of sulfur poisoning of CNx oxygen reduction catalysts for PEM fuel cells." Journal of Catalysis Submitted 2011.

287. Ye, S.; Vijh, A. K., "Non-noble metal-carbonized aerogel composites as electrocatalysts for the oxygen reduction reaction." Electrochem. Commun. 2003, 5, 272- 275.

288. Kim, J.-H.; Ishihara, A.; Mitsushima, S.; Kamiya, N.; Ota, K.-I., "Catalystic activity of titanium oxide for oxygen reduction reaction as a non platinum catalyst for PEFC." Electrochim. Acta 2007, 52, 2492-2497.

289. Limoges, B. R.; Stanis, R. J.; Turner, J. A.; Herring, A. M., "Electrocatalyst (3+n)- materials for fuel cells based on the polyoxometalates [PMo(12-n)VnO40] (n = 0-3)." Electrochim. Acta 2005, 50, (5), 1169-1179.

290. Ozaki, J.-i.; Tanifuji, S.-i.; Furuichi, A.; Yabutsuka, K., "Enhancement of oxygen reduction activity of nanoshell carbons by introducing nitrogen atoms from metal phthalocyanines." Electrochim. Acta 2010, 55, 1864-1871.

291. Matter, P. H.; Biddinger, E. J.; Ozkan, U. S., "Non-precious metal oxygen reduction catalysts for PEM fuel cells." In Catalysis, Spivey, J. J., Ed. The Royal Society of Chemistry: Cambridge, UK, 2007; Vol. 20, pp 338-361.

292. Cussler, E. L., Diffusion mass transfer in fluid systems. 2nd ed.; Cambridge University Press: New York, 1997; p 580.

305

293. Garsany, Y.; Gould, B. D.; Baturina, O. A.; Swider-Lyons, K. E., "Comparison of the sulfur poisoning of PBI and Nafion PEMFC cathodes." Electrochem. Solid-State Lett. 2009, 12, B138-B140.

294. Yang, H.; Alonso-Vante, N.; Lamy, C.; Akins, D. L., "High methanol tolerance of carbon-supported Pt-Cr alloy nanoparticle electrocatalysts for oxygen reduction." J. Electrochem. Soc. 2005, 152, (4), A704-A709.

295. Chung, H. T.; Johnston, C. M.; Artyushkova, K.; Ferrandon, M.; Myers, D. J.; Zelenay, P., "Cyanamide-derived non-precious metal catalyst for oxygen reduction." Electrochem. Commun. 2010, 12, 1792-1795.

296. Bao, X.; von Deak, D.; Biddinger, E.; Ozkan, U. S.; Hadad, C. M., "A computational exploration of the oxygen reduction reaction over a carbon catalyst containing a phosphinate functional group." Chem. Commun. 2010, 46, (45), 8621-8623.

297. Moulder, J. F.; Strickle, W. F.; Sobol, P. E.; Bomben, K. D., Handbook of Photoelectron Spectroscopy. Perkin-Elmer Corporation: Eden Prairie, MN, 1992.

298. Gasteiger, H. A.; Markovic, N. M., "Just a dream - or future reality?" Science 2009, 324, 48-49.

299. Li, X.; Liu, G.; Popov, B. N., "Activity and stability of non-precious metal catalysts for oxygen reduction in acid and alkaline electrolytes." J. Power Sources 2010, 195, (19), 6373-6378.

300. Roen, L. M.; Paik, C. H.; Jarvi, T. D., "Electrocatalytic corrosion of carbon support in PEMFC cathodes." Electrochem. Solid-State Lett. 2004, 7, (1), A19-A22.

301. Liu, Z. Y.; Zhang, J. L.; Yu, P. T.; Zhang, J. X.; Makharia, R.; More, K. L.; Stach, E. A., "Transmission electron microscopy observation of corrosion behaviors of platinized carbon blacks under thermal and electrochemical conditions." J. Electrochem. Soc. 2010, 157, (6), B906-B913.

302. Xie, J.; Wood III, D. L.; Wayne, D. M.; Zawodzinski, T. A.; Atanassov, P.; Borup, R. L., "Durability of PEFCs at high humidity conditions." J. Electrochem. Soc. 2005, 152, A104-A113.

303. Kinoshita, K., Carbon, Electrochemical and Physiochemical Properties. Wiley Interscience: New York, 1988.

306

304. Serov, A. A.; Min, M.; Chai, G.; Han, S.; Seo, S. J.; Park, Y.; Kim, H.; Kwak, C., "Electroreduction of oxygen over iron macrocyclic catalysts for DMFC applications." J. Appl. Electrochem. 2009, 39, 1509-1516.

305. Li, X.; Liu, C.; Xing, W.; Lu, T., "Development of durable carbon black/titanium dioxide supported macrocycle catalysts for oxygen reduction reaction." J. Power Sources 2009, 193, 470-476.

306. Bae, I. T.; Scherson, D., "In situ x-ray absorption of a carbon monoxide-iron porphyrin adduct adsorbed on high-area carbon in an aqueous electrolyte." J. Phys. Chem. B 1998, 102, (14), 2519- 2522.

307. Biddinger, E. J.; Ozkan, U. S., "Methanol tolerance of CNx oxygen reduction catalysts." Top. Catal. 2007, 46, 339-348.

308. Bae, I. T.; Scherson, D., "In situ core-electron spectroscopy of carbon monoxide adsorbed on high-area platinum in an acid electrolyte." J. Phys. Chem. 1996, 100, (50), 19215-19217.

309. Schmidt, T. J.; Gasteiger, H. A.; Behm, R. J., "Rotating disk electrode measurements on the CO tolerance of a high-surface area Pt/Vulcan carbon fuel cell catalyst." J. Electrochem. Soc. 1999, 146, (4), 1296-1304.

310. Baysinger, G.; Berger, L. I.; Goldberg, R. N.; Kehiaian, H. V.; Kuchitsu, K.; Rosenblatt, G.; Roth, D. L.; Zwillinger, D., CRC Handbook of Chemistry and Physics. 74 ed.; CRC Press: Boca Raton FL, 2005.

311. Zhou, Y.; Pasquarelli, R.; Holme, T.; Berry, J.; Ginley, D.; O'Hayre, R., "Improving PEM fuel cell catalyst activity and durability using nitrogen-doped carbon supports: observations from model Pt/HOPG systems." J. Mater. Chem. 2009, 19, 7830- 7838.

312. Bonakdarpour, A.; Dahn, T. R.; Atanasoski, R. T.; Debe, M. K.; Dahn, J. R., "H2O2 release during oxygen reduction reaction on Pt nanoparticles." Electrochem. Solid- State Lett. 2008, 11, B208-B211.

313. Higuchi, E.; Uchida, H.; Watanabe, M., "Effect of loading level in platinum- dispersed carbon black electrocatalysts on oxygen reduction activity evaluated by rotating disk electrode." J. Electroanal. Chem. 2005, 583, (1), 69-76.

314. Wakabayashi, N.; Takeichi, M.; Itagaki, M.; Uchida, H.; Watanbe, M., "Temperature-dependence of oxygen reduction activity at a platinum electrode in an

307 acidic electrolyte solution investigated with a channel flow double electrode." J. Electroanal. Chem. 2005, 574, 339-346.

315. Maekawa, Y.; Ishihara, A.; Kim, J.-H.; Mitsushima, S.; Ota, K.-i., "Catalytic activity of zirconium oxynitride prepared by reactive sputtering for ORR in sulfuric acid." Electrochem. Solid-State Lett. 2008, 11, B109-B112.

316. Bron, M.; Fiechter, S.; Bogdanoff, P.; Tributsch, H., "Thermogravimetry/mass spectrometry investigations on the formation of oxygen reduction catalysts for PEM fuel cells on the basis of heat-treated iron phenanthroline complexes." Fuel Cells 2003, 2, (3- 4), 137-142.

317. Mayrhofer, K. J. J.; Strmcnik, D.; Blizanac, B. B.; Stamenkovic, V.; Arenz, M.; Markovic, N. M., "Measurement of oxygen reduction activities via the rotating disc electrode method: From Pt model surfaces to carbon-supported high surface area catalysts." Electrochim. Acta 2008, 53, 3181-3188.

318. Santos, L. G. R. A.; Oliveira, C. H. F.; Moraes, I. R.; Ticianelli, E. A., "Oxygen reduction reaction in acid medium on Pt–Ni/C prepared by a microemulsion method." J. Electroanal. Chem. 2006, 596, 141-148.

319. Baranton, S.; Coutanceau, C.; Roux, C.; Hahn, F.; Leger, J. M., "Oxygen reduction reaction in acid medium at iron phthalocyanine dispersed on high surface area carbon substrate: tolerance to methanol, stability and kinetics." J. Electroanal. Chem. 2005, 577, 223-234.

320. Liu, L.; Lee, J.-W.; Popov, B. N., "Development of ruthenium-based bimetallic electrocatalysts for oxygen reduction reaction." J. Power Sources 2006, 162, 1099-1103.

321. Tarasevich, M. R.; Chalykh, A. E.; Bogdanovskaya, V. A.; Kuznetsova, L. N.; Kapustina, N. A.; Efremov, B. N.; Ehrenburg, M. R.; Reznikova, L. A., "Kinetics and mechanism of oxygen reduction reaction at CoPd system synthesized on XC72." Electrochim. Acta 2006, 51, 4455-4462.

322. Sirk, A. H. C.; Campbell, S. A.; Birssa, V. I., "Oxygen reduction by sol derived [Co, N, C, O]-based catalysts for use in proton exchange membrane fuel cells." Electrochem. Solid-State Lett. 2005, 8, (2), A104-A107.

323. Ozaki, J.-i.; Anahara, T.; Kimura, N.; Oya, A., "Simultaneous doping of boron and nitrogen into a carbon to enhance its oxygen reduction activity in proton exchange membrane fuel cells." Carbon 2006, 44, 3358-3361.

324. Bonnecaze, R. T.; Mano, N.; Nam, B.; Heller, A., "On the behavior of the porous rotating disk electrode." J. Electrochem. Soc. 2007, 154, (2), F44-F47.

308

325. Ignaszak, A.; Teo, C.; Ye, S.; Gyenge, E., "Pt-SnO2-Pd/C electrocatalyst with enhanced activity and durability for the oxygen reduction reaction at low Pt loading: the effect of carbon support type and activation." J. Phys. Chem. C 2010, 114, (39), 1.

326. Nam, B.; Bonnecaze, R. T., "Analytic models of the infinite porous rotating disk electrode." J. Electrochem. Soc. 2007, 154, (10), F191-F197.

327. Bonakdarpour, A.; Delacote, C.; Yang, R.; Wieckowski, A.; Dahn, J. R., "Loading of Se/Ru/C electrocatalyst on a rotating ring-disk electrode and the loading impact on a H2O2 release during oxygen reduction reaction." Electrochem. Commun. 2008, 10, 611-615.

328. Feng, Y.; He, T.; Alonso-Vante, N., "Oxygen reduction reaction on carbon- supported CoSe2 nanoparticles in an acidic medium." Electrochim. Acta 2009, 54, 5252- 5256.

329. Shah, A. A.; Ralph, T. R.; Walsh, F. C., "Modeling and simulation of the degradation of perfluorinated ion-exchange membranes in PEM fuel cells." J. Electrochem. Soc. 2009, 156, B465-B484.

330. Biddinger, E. J.; von Deak, D.; Singh, D.; Marsh, H.; Tan, B.; Knapke, D. S.; Ozkan, U. S., "Examination of catalyst loading effects on the selectivity of CNx and Pt/VC ORR catalysts using RRDE." J. Electrochem. Soc. 2011, 158, (4), B402-B409.

331. Pattabi, M.; Castellanos, R. H.; Castillo, R.; Ocampo, A. L.; Moreira, J.; Sebastian, P. J.; McClure, J. C.; Mathew, X., "Electrochemical characterization of tungsten carbonyl compound for oxygen reduction reaction." Int. J. Hydrogen Energ. 2001, 26, 171-174.

332. Rodil, S. E., "Infrared spectra of amorphous carbon based materials." Diam. Relat. Mater. 2005, 14, (8), 1262-1269.

333. Pereira, M. F. R.; Orfao, J. J. M.; Figueiredo, J. L., "Oxidative dehydrogenation of ethylbenzene on activated carbon catalysts. 3. Catalyst deactivation." Appl. Catal. A-Gen. 2001, 218, (1-2), 307-318.

334. Jansinski, R., "A new fuel cell cathode catalyst." Nature 1964, 201, 1212.

335. Jansinski, R., "Cobalt phthalocyanine as a fuel cell cathode." J. Electrochem. Soc. 1965, 112, (5), 526-528.

336. Kaisheva, A.; Gamburtsev, S.; Iliev, I., "Characteristics of air (oxygen) gas- diffusion electrodes with catalysts of active carbon and pyrolyzed compounds of the

309 tetramethoxyphenylporphyrin type. Acidic and neutral electrolytes." Soviet Journal of Electrochemistry 1982, 18, 127-131.

337. Strelko, V. V.; Kuts, V. S., "The effect of oxygen impurities on the formation of intercalated compounds of graphite with an ultrahigh content of alkali metals under mild conditions." Theoretical and Experimental Chemistry 2002, 38, (1), 15-20.

338. Cramer, C. J., Essentials of Computational Chemistry: Theories and Models 2nd ed.; Wiley: New Jersey, 2006; p 596.

339. Vayner, E.; Anderson, A. B., "Theoretical predictions concerning oxygen reduction on nitrided graphite edges and a cobalt center bonded to them." J. Phys. Chem. C 2007, 111, 9330-9336.

340. Sidik, R. A.; Anderson, A. B.; Subramanian, N. P.; Kumaraguru, S. P.; Popov, B. N., "O2 reduction on graphite and nitrogen-doped graphite: experiment and theory." J. Phys. Chem. B 2006, 110, 1787-1793.

341. Ikeda, T.; Boero, M.; Huang, S.; Terakura, K.; Oshima, M.; Ozaki, J., "Carbon alloy catalysts: active sites for oxygen reduction." Journal of Physical Chemistry C 2008, 112, (38), 14706-14709.

342. Nakajima, T.; Koh, M.; Katsube, T., "Structure, chemical bonding and electrochemical behavior of heteroatom-substituted carbons prepared by arc discharge and chemical vapor deposition." Solid State Sci. 2000, 2, 17-29.

343. Deng, Z. W.; Souda, R., "Synthesis and thermal decomposition of carbon nitride films prepared by nitrogen ion implantation into graphite." This Solid Films 2002, 406, 46-53

344. Titantah, J. T.; Lamoen, D., "Carbon and nitrogen 1s energy levels in amorphous carbon nitride systems: XPS interpretation using first-principles." Diam. Relat. Mater. 2007, 16, 581-588.

345. Lias, S. G. "Ionization Energy Evaluation," National Institute of Standards and Technology, 2008, http://webbook.nist.gov

346. Gleiter, R.; Heilbronner, E.; Hornung, V., "Photoelectron spectra of azabenzenes azanaphthalenes: I. Pyridine, diazines s-triazine and s-tetrazine." Helvetica Chimica Acta 1972, 55, 255.

347. Dewar, M. J. S.; Worley, S. D., "Photoelectron spectra of molecules. II. The ionization potentials of azabenzenes and azanaphthalenes." J. Chem. Phys. 1969, 51, 263.

310

348. Shahbaz, M.; Urano, S.; LeBreton, P. R.; Rossman, M. A.; Hosmane, R. S.; Lenard, N. J., "Tri-s-triazine: synthesis, chemical behavior and spectroscopic and theoretical probes of valence orbital structure." J. Am. Chem. Soc. 1984, 106, 2805.

349. Gleiter, R.; Spanget-Larsen, J., "Photoelectron spectra of 1,2,3-triazine and its methyl derivatives." Chem. Phys. Lett. 1983, 97, 94.

350. Gleiter, R.; Kobayashi, M.; Nuenhoeffer, H.; Spanget-Larsen, J., "Photoelectron spectra of 1,2,4-traizine and some methyl derivatives." Chem. Phys. Lett. 1977, 46, 231.

351. Malloci, G.; Joblin, C.; Mulas, G., "Theoretical evaluation of PAH dication properties." Astronomy & Astrophysics 2008, 1, 1-14.

352. Modelli, A.; Mussoni, L., "Rapid quatitative prediction of ionization energies and electron affinities of polycyclic aromatic hydrocarbons." Chemical Physics 2007, 332, 367-374.

353. Van Den Ham, D. M. W.; Van Der Meer, D., "The photoelectron spectra of the diazanaphthalenes." Chem. Phys. Lett. 1972, 12, 447.

354. Dobson, B.; Hiller, I. H.; Connor, J. A.; Moncrieff, D.; Scanlan, M. L.; Garner, C. D., "Electronic structure and low-energy photoelectron spectra of 4,4'-disubstitured 2,2'- bipyridines." Journal of the Chemical Society-Faraday Transactions II 1983, 79, 295.

355. Lecoultre, J.; Heilbronner, E., "153. The ionization energies of tetrazinodi(heteroarenes)." Helvetica Chimica Acta 1987, 70, 1661.

356. Barone, V.; Cauletti, C.; Lelu, F.; Piancastelli, M. N.; Russo, N., "Relative ordering and spacing of n and pi levels in isomeric bipyrimidines. A theoretical and gas- phase UV photoelectron spectroscopic study." J. Am. Chem. Soc. 1982, 104, 4571.

357. Boschi, R.; Clar, E.; Schmidt, W., "Photoelectron spectra of polynuclear aromatics. III. The effect of nonplanarity in sterically overcrowded aromatic hydrocarbons." J. Chem. Phys. 1974, 60, 4406.

358. Palmer, M. H.; Leaver, D.; Nisbet, J. D.; Millar, R. W.; Egdell, R., "The electronic structure of some J. Heterocycl. Chem. with bridgehead nitrogen: photoelectron spectra and ab initio molecular orbital calculations." Journal of Molecular Structure 1977, (42), 85.

359. Barone, V.; Cauletti, C.; Commisso, L.; Lelj, F.; Piancastelli, M. N.; Russo, N., "Quantum-mechanical and ultraviolet photoelectron spectroscopic studies of

311 azabiphenyls. The case of 4-phenylpyridine." Journal of Chemical Research Synopsis 1984, 338.

360. Maier, J. P.; Turner, D. W., "Steric inhibition of resonance studied by molecular photoelectron spectroscopy. Part I. Biphenyls." Faraday Discuss. 1972, 54, 149.

361. Hush, N. S.; Cheung, A. S.; Hilton, P. R., "Binding energies of pi- and "lone pair" -levels in mono- and diaza-phenanthrenes and anthracenes: and He(I) photoelectron spectroscopic study." Journal of Electron Spectroscopy and Related Phenomena 1975, 7, 385.

362. Farrell, P. G.; Newton, J., "Ionization potentials of primary aromatic amines and aza-hydrocarbons." Tetrahedron Letters 1966, 5517.

363. Millefiori, S.; Millefori, A.; Pignataro, S.; Distefano, G.; Colonna, F. P., "Gas phase UPS investigation of trans-azobenzenes." Zeitschrift fur Naturorschung A: Journal of Physical Sciences 1979, 34, 1496.

364. Akaba, R.; Tokumaru, K.; Kobayashi, T.; Utsunomiya, C., "Electronic structures conformations of N-benzylideneanilines. II. Photoelectron spectral study." Bulletin of the Chemical Society of Japan 1980, 53, 2002.

365. Briegleb, G., "Electon affinity of organic molecules." Angew. Chem. Int. Ed. 1964, 3, 617.

366. Schmidt, W., "Photoelectron spectra of polynuclear aromatics. V. Correlations with ultraviolet absorption spectra in the catacondensed series." J. Chem. Phys. 1977, 66, 828.

367. Clar, E.; Schmidt, W., "Correlations between photoelectron and UV absorption spectra of polycyclic hydrocarbons. The pyrene series." Tetrahedron Letters 1979, 35, 1027.

368. Clar, E.; Schmidt, W., "Correlations between photoelectron and ultraviolet absorption spectra of polycyclic hyrdocarbons. The perylene, coronene and bisanthene series." Tetrahedron Letters 1977, 33, 2093.

369. Clar, E.; Schmidt, W., "Correlations between photoelectron and ultraviolet absorption spectra of polycyclic hydrocarbons. The terrylene and peropyrene series." Tetrahedron Letters 1978, 34, 3219.

370. Boschi, R.; Murrell, J. N.; Schmidt, W., "Photoelectron spectra of polycyclic aromatic hydrocarbons." Faraday Discuss. 1972, 54, 116.

312

371. Camalli, M.; Caruso, F.; Mattogno, G.; Rivarola, E., "Adducts of tin(IV) and organotin(IV) derivatives with 2,2'-azopyridine (AZP). II. Crystal and molecular structure of SnMe2Br2AZP and further Moessbauer and photoelectronic spectroscopic studies." Inorganica Chimica Acta 1990, 170, 225-231.

372. Yamaguchi, H.; Higashi, M.; MacBride, J. A. H.; Gleiter, R., "Photoelectron spectra of diazabiphenylenes." Journal of the Chemical Society-Faraday Transactions II 1985, 81, 1831.

373. Ryczkowski, J., "Application of infrared photoacoustic spectroscopy in catalysis." Catal. Today 2007, 124, 11-20.

374. Deng, C. Z.; Dignam, M. J., "Sputtered cobalt-carbon-nitrogen thin films as oxygen reduction electrocatalysts." J. Electrochem. Soc. 1998, 145, 3507-3512.

375. Nefedov, V. I.; Salin, J. V.; Walther, D.; Uhlig, E.; Dinjus, E., "ESCA studies of 1,4-diheterodiene complexes of nickel(0)." Zeitschrift fur Chemie 1977, 17, 190-191.

376. Brant, P.; Feltmam, R. D., "X-ray photoelectron spectra of aryldiazo derivatives of transition metals." Journal of Organometallic Chemistry 1976, 120, C53-C57.

377. Kocha, S. S., "Principles of MEA preparation." In Handbook of Fuel Cells - Fundamentals, Technology and Applications, Vielstich, W.; Gasteiger, H. A.; Lamm, A., Eds. John Wiley & Sons, Ltd.: 2003; Vol. 3: Fuel Cell Technology and Applications, pp 538-565.

378. Therdthianwong, A.; Manomayidthikarn, P.; Therdthianwong, S., "Investigation of membrane electrode assembly (MEA) hot-pressing parameters for proton exchange membrane fuel cell." Energy 2007, 32, 2401-2411.

379. Liu, P.; Yin, G.-P.; Wang, E.-D.; Zhang, J.; Wang, Z.-B., "Infuence of hot- pressing temperature on physical and electrochemical performance of catalyst coated membranes for direct methanol fuel cells." J. Appl. Electrochem. 2009, 39, 859-866.

380. Li, J.; Matsuura, A.; Kakigi, T.; Miura, T.; Oshima, A.; Washio, M., "Performance of membrane electrode assemblies based on proton exchange memebranes prepared by pre-irradiation induced grafting." J. Power Sources 2006, 161, 99-105.

381. Su, F.; Tian, Z.; Poh, C. K.; Wang, Z.; Lim, S. H.; Liu, Z.; Lin, J., "Pt nanoparticles supported on nitrogen-doped porous carbon nanospheres as an electrocatalyst for fuel cells." Chem. Mater. 2010, 22, 832-839.

313

382. Wee, J.-H.; Lee, K.-Y.; Kim, S. H., "Fabrication methods for low-Pt-loading electrocatalysts in proton exchange membrane fuel cells." J. Power Sources 2007, 165, 667-677.

383. Zelenay, P.; Ren, X.; Thomas, S.; Davey, J.; Gottesfeld, S. Catalyst inks and methods of application for direct methanol fuel cells. 31 October 2001, 2002.

384. Liu, H.; Shi, Z.; Zhang, J.; Zhang, L.; Zhang, J., "Ultrasonic spray pyrolyzed iron- polypyrrole mesoporous spheres for fuel cell oxygen reduction reduction electrocatalysts." J. Mater. Chem. 2009, 19, 468-470.

385. Chung, H. T.; Johnston, C. M.; Garzon, F. H.; Zelenay, P., "A non-precious electrocatalyst for oxygen reduction based on simple heat-treated precursors." ECS Transactions 2008, 16, (2), 385-391.

386. Bashyam, R.; Zelenay, P., "A class of non-precious metal composite catalysts for fuel cells." Nature 2006, 443, (7), 63-66.

387. Davey, J.; Wilson, M.; Valerio, J.; Bender, G. In Overview of Fuel Cell MEAs at Los Alamos National Laboratory, 1st Symposium on Manufacturing of MEAs for Hydrogen Applications, Dayton, OH, August 9-11; Dayton, OH.

388. von Deak, D.; Biddinger, E. J.; Ozkan, U. S., "Carbon corrosion characteristics of CNx nanostructures in acidic media and implications for ORR performance." J. Electroanal. Chem. submitted.

389. Nallathambi, V.; Lee, J.-W.; Kumaraguru, S. P.; Wu, G.; Popov, B. N., "Development of high performance carbon composite catalyst for oxygen reduction reaction in PEM proton exchange membrane fuel cells." J. Power Sources 2008, 183, 34- 42.

390. Mayrhofer, K. J. J.; Meier, J. C.; Ashton, S. J.; Wiberg, G. K. H.; Kraus, F.; Hanzlik, M.; Arenz, M., "Fuel cell catalyst degradation on the nanoscale." Electrochem. Commun. 2008, 10, 1144-1147.

391. Li, X.; Park, S.; Popov, B. N., "Highly stable Pt and PtPd hybrid catalysts supported on a nitrogen-modified carbon composite for fuel cell application." J. Power Sources 2010, 195, 445-452.

392. Kinoshita, K.; Bett, J. A. S., "Determination of carbon surface oxides on platinum-catalyzed carbon." Carbon 1974, 12, 525-533.

314

393. Biddinger, E. J.; von Deak, D.; Marsh, H.; Ozkan, U. S., "RRDE catalyst ink aging effects on selectivity to water formation in ORR." Electrochem. Solid-State Lett. 2010, 13, B98-B100.

394. Wood, T. E.; Atanasoski, R.; Schmoeckel, A. K. High performance, high durability non-precious metal catalyst for fuel cell cathodes. 4-12-07, 2008.

395. Scherson, D. A.; Fierro, C. A.; Tryk, D.; Gupta, S. L.; Yeager, E. B., "In-situ Mossbauer Spectroscopy and electrochemcial studies of the thermal stability of iron phthalocyanine dispersed in high surface area carbon." J. Electroanal. Chem. 1985, 184, 419-426.

396. Kim, S.; Bae, I. T.; Sandifer, M.; Ross, P. N.; Carr, R.; Woicik, J.; Antonio, M. R.; Scherson, D., "In situ XANES of an iron porphyrin irreversibly adsorbed on an electrode surface." Journal of American Chemical Society 1991, 113, (24), 9063-9066.

397. Stefan, I. C.; Mo, Y.; Ha, S. Y.; Kim, S.; Scherson, D. A., "In situ Fe K edge X- ray absorption fine structure of a nitrosyl adduct of iron phthalocyanine irreversibly adsorbed on a high area carbon electrode in an acidic electrolyte." Inorg. Chem. 2003, 42, (14), 4316-4321.

398. Principi, E.; Cicco, A. D.; Witkowska, A.; Marassi, R., "Performance of a fuel cell optimized for in situ X-ray absorption experiments." J. Synchrotron Radiat. 2007, 14, 276-281.

399. Witowksa, A.; Dsoke, S.; Principi, E.; Marassi, R.; Cicco, A. D.; Albertini, V. R., "Pt-Co cathode electrocatalyst behaviour viewed by in situ XAFS fuel cell measurements." J. Power Sources 2008, 178, 603-609.

400. Croze, V.; Ettingshausen, F.; Melke, J.; Soehn, M.; Stuermer, D.; Roth, C., "The use of in situ X-ray absorption spectroscopy in applied fuel cell research." J. Appl. Electrochem. 2008, 40, 877-883.

401. Tada, M.; Murata, S.; Asakoka, T.; Hiroshima, K.; Okumura, K.; Tanida, H.; Uruga, T.; Nakanishi, H.; Matsumoto, S.-c.; Inada, Y.; Nomura, M.; Iwasawa, Y., "In situ time-resolved dynamic surface events on the Pt/C cathode in a fuel cell under operando conditions." Angew. Chem.-Int. Edit. 2007, 46, (23), 4310-4315.

315