.13. .4.
. 2833“...

L.

.' m3...
Arm;

5.5.1.09”

..,.. . .
“mm: 4.7,: S
. h. 1h. , . fn‘WwHWMMM‘; .

a: ti
Rm. )

19hrmnhh .
. $.-
, . c (u). ‘
:Lufa

...
"3

l-
; Ain’t!»
. .72

x

.Wupu .. . V 5]....
3.6... hi}:

. 2.1.3 {luv}.

.

I

n (.5,
3

 

 

 

23.5%...

:4er .l

....;.p.ﬁ
of? 3.5.:

if
v lawaﬁvi .{ ,‘
.‘blvilu. W...\.!-lIw

THEM:-
WM

This is to certify that the
dissertation entitled

DEGRADATION MECHANISMS OF CARBON-BASED
ELECTROCATALYST SUPPORT MATERIALS AND
DEVELOPMENT OF AN ADVANCED SUPPORT BASED ON
ELECTRICALLY CONDUCTING DIAMOND

presented by

ANNE ELIZABETH FISCHER

has been accepted towards fulﬁllment
of the requirements for the

PhD. degree in Chemistry

 

 

7
”/1 "\
(ﬂex [/1] r‘éﬂut

J Major Professor’s Signature

Date

 

MSU is an Afﬁnnative Action/Equal Opportunity Institution

 

 

UBRARYM
Michigan State
University

 

 

 

....-v.-I-o-c-v-o--v-

«.0... —~-

o.-.-.-.--u-u-----:-.--.—

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

IMAY 0 8 2008

 

 

I324 '08.

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/CIRC/DateDue.indd-p.1

DEGRADATION MECHANISMS OF CARBON-BASED ELECTROCATALYST
SUPPORT MATERIALS AND DEVELOPMENT OF AN ADVANCED SUPPORT
BASED ON ELECTRICALLY CONDUCTING DIAMOND
By

Anne Elizabeth Fischer

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Chemistry

2005

ABSTRACT
DEGRADATION MECHANISMS OF CARBON-BASED ELECTROCATALYST
SUPPORT MATERIALS AND DEVELOPMENT OF AN ADVANCED SUPPORT
BASED ON ELECTRICALLY CONDUCTING DIAMOND
By
Anne Elizabeth Fischer

In this dissertation, the degradation mechanisms of spZ-bonded carbon
electrocatalyst supports were studied under potential and temperature conditions relevant
to the polymer electrolyte membrane fuel cell (PEMFC). In addition, an alternative
support was fabricated in two forms: electrically conducting diamond powder and paper
to overcome current material stability issues in the PEMF C.

Two structurally well-characterized spz-bonded carbon powders, graphite
(structurally well-ordered) and glassy carbon (GC, structurally disordered) were studied
under potentiostatic polarization from 1.0 to 1.6 V vs. Ag/AgCl at 25, 50, and 80 °C.
Characterization of the surface oxidation and microstructural changes (i.e., increase in the
exposed edge plane density) provided evidence for the so-called order/disorder
mechanism where structurally disordered carbons corrode more severely because of
oxidation and gasiﬁcation of the exposed edge plane. Microstructural changes for
graphite were heterogeneously distributed across the electrode surface. This is indicative
of a nucleation and growth process, where disordered regions and defects serve as active
sites for electrochemical corrosion, while other, more structurally ordered regions do not
corrode. Preliminary results for a high-surface-area carbon black, Vulcan XC-72, are

presented that show changes in the surface oxide content and also discuss the effect of

polarization potential on Pt activity.

The physical and electrochemical properties of two commercial boron-doped
diamond thin-ﬁlm electrodes were compared with microcrystalline and nanocrystalline
boron-doped diamond thin ﬁlm deposited in our laboratory. The electrochemical
response for F e(CN)63 '/ 4’, Ru(NH3)63+/2+, IrC162'/3', 4-methylcatechol, and F e3” 2+ was quite
reproducible from electrode type-to-type and from ﬁlm-to-ﬁlm for a given type. AE ,
if", and 1},“ values for Fe(CN)63'/4', Ru(NH3)63+/2+ on all electrodes were relatively
unaffected by pH.

Electrically conducting diamond powder was prepared by coating insulating
diamond powder (8-12 pm diam) with a thin boron—doped diamond layer using
microwave-assisted chemical vapor deposition (CVD). Increases in the electrical
conductivity aﬁer growth conﬁrmed that a conductive diamond overlayer formed. The
charge passed during anodic polarization at 1.6 V vs. Ag/AgCl and 25 °C for 1 h was
largest for GC powder (0.88 C/cmz) and smallest for conductive diamond powder (0.18
C/cmz), illustrating the dimensional stability of diamond powder compared to spz-bonded
carbon powder.

Boron-doped nanocrystalline diamond (BND) was coated on Toray® carbon paper
(TCP) via microwave-assisted CVD. Pt nanoparticles were deposited on TCP and BND
using a pulsed galvanostatic method. The stability of the bare TCP and BND substrates
and the composite Pt/TCP and Pt/BND electrodes were studied using potentiostatic
polarization in 0.1 M HClO4. The BND electrode exhibited superior morphological and
microstructural stability over TCP at 1.6 V vs. Ag/AgCl. Evidence was found for

dissolution and redeposition of Pt on composite electrodes, particularly for Pt/TCP.

COPyright by
ANNE ELIZABETH FISCHER
2005

In memory of
Jennifer Ruth White,
friend and inspiration.
November 5, 1977 — February 6, 2001

ACKNOWLEDGEMENTS

So many people have touched my life throughout my time and Michigan State
University. I would ﬁrst and foremost like to thank my advisor, Dr. Greg Swain, for his
guidance, support, and encouragement over the last few years. Thanks to my committee
members, Dr. Gary Blanchard, Dr. Jim McCusker, and Dr. Michael Mackay, for your
helpful comments about my research. I would also like to thank Dr. Simon Garrett and
Dr. Kathryn Severin for their support and mentoring.

Where would I be without the ﬁm, support, ﬁ'iendships, and frustrations that come
with being a Swain group member? Thanks to all along the way who helped shape the
scientist and the person that I have become. As Luther would say, 'thank-ye'
(alphabetically, of‘course) to Audrey, Doug, Elizabeth, Gloria, Grace, Hua, the Jasons,
Jesse, J inwoo, Luther, Martin, Show, Veronika, Yang, and Yingrui.

I have been fortunate to have a wonderful group of ﬁiends at Michigan State,
without whom, I am not sure I would have survived. Words alone cannot express my
gratefulness to have friends like Amber, Audrey, Bennett, Carl, Elizabeth, Heather, Mark,
Melissa (a.k.a., Melita and Phyllis), and Prema. You guys are the best.

Not to be forgotten are the non-chemistry (a.k.a. sane) people in my life that saw
me through the process. Allison, Katie, and Nick, your love and support were invaluable.
You helped me keep a healthy perspective (sometimes, at least) about what life is really
all about. Thanks so much. I would also like to extend a very special thanks to Katie for

all of her support during these last 5 years. I truly could not have done this without you.

vi

Finally, and most importantly, I would like to say thank you to my parents. I
would not be where I am today without your encouragement, your love, and your

support. Thank you for always being there for me.

-Anne

vii

TABLE OF CONTENTS
List of Tables
List of Figures

CHAPTER 1 — INTRODUCTION

1.1 Background

1.2 Fuel Cell Materials and Operation

1.3 Carbon Corrosion in Fuel Cells

1.4 spz-Bonded Carbon Structure

1.5 Boron-Doped Diamond

1.6 Chemical Vapor Deposition (CVD) of Diamond
1.6.1 Microcrystalline Diamond Growth Mechanism
1.6.2 Nanocrystalline Diamond Growth Mechanism

1.7 Research Goals and Objectives

1.8 Dissertation Outline

1.9 References

CHAPTER 2 — EXPERIMENTAL METHODS
2.1 Diamond Thin Film Growth
2.2 Material Characterization
2.2.1 Scanning Electron Microscopy (SEM)
2.2.2 X-Ray Diffraction Spectroscopy (XRD)
2.2.3 Raman Spectroscopy
2.2.4 Electrical Resistance
2.3 Electrochemical Characterization
2.3.1 Electrode Preparation and Experimental Setup
2.3.1.1 Diamond Thin-Film Electrodes on Si
2.3.1.2 Carbon Powder Electrodes
2.3.1.3 Carbon Paper Electrodes
2.3.2 Electrochemical Kinetics
2.3.3 Polarization Studies
2.3.4 Pt Electrodeposition
2.4 Reagents
2.5 References

CHAPTER 3 - STUDY OF THE DEGRADA TION MECHANISMS OF SP2-
BONDED CARBON PO WDERS DURING ELECTROCHEMICAL
POLARIZATION
Abstract
3.1 Introduction
3.2 Results and Discussion
3.2.1 Anodic Polarization of Graphite and Glassy Carbon (GC) Powders
3.2.2 Evaluation of the Extent of Surface Oxide in Terms of
Electroactive Oxide Formation

viii

S4
S4
56
6O
60

66

3.2.3 Evaluation of Microstructural Changes
3.3 Anodic Polarization of Carbon Black, Vulcan XC-72
3.4 Conclusions
3.5 References

CHAPTER 4 - THE ELECTROCHEMICAL PERFORMANCE OF
DIAMOND THIN-FILM ELECTRODES FROM DIFFERENT
COMMERCIAL SOURCES
Abstract
4.1 Introduction
4.2 Results and Discussion
4.2.1 Material Characterization
4.2.2 Electrochemical Characterization
4.2.3 pH Effects on Voltammetric Response
4.3 Conclusions
4.4 References

CHAPTER 5 — PREPARA TION AND CHARA CTERIZA TION OF BORON-
DOPED DIAMOND POWDER: A POSSIBLE DIMENSIONALL Y STABLE
ELE C TROCA TAL YS T SUPPORT MA TERIAL
Abstract
5.1 Introduction
5.2 Results and Discussion
5.2.1 Material Characterization and Electrical Measurements
5.2.2 Electrochemical Characterization
5.2.3 Morphological and Microstructural Stability
5.3 Conclusions
5.4 References

CHAPTER 6 — PREPARA TION AND CHARA CTERIZA T ION OF BORON-
DOPED DIAMOND PAPER WITH PT NANOPARTICLES
Abstract
6.1 Introduction
6.2 Results and Discussion
6.2.1 Nanocrystalline Diamond Deposition on Carbon Paper
6.2.1.1 Morphology and Microstructure
6.2.1.2 Electrochemical Characterization
6.2.2 Electrodeposition of Pt on TCP and BND
6.2.3 Evaluation of Pt/TCP and Pt/BND Stability
6.3 Conclusions
6.4 References

CHAPTER 7 — FUTURE RESEARCH DIRECTIONS

 

ix

69
75
84
85

88
88
90
91
91
96
107
109
111

115
115
117
119
119
127
137
140
142

144
144
146
149
149
149
153
155
165
174
176

178

Table 1.1

Table 2. I

Table 2.2

Table 2.3

Table 2.4

Table 3.1

Table 3.2

Table 3.3

Table 3.4

Table 3.5

Table 3. 6

Table 3. 7

Table 3.8

LIST OF TABLES

Crystallite width (La) and interlayer spacing (dooz) for some
common spZ-bonded carbons.

Growth conditions for microcrystalline and nanocrystalline boron-
doped diamond thin ﬁlm deposited on Si.

Film thickness, resistivity, carrier concentration, and carrier

mobility for typical microcrystalline and nanocrystalline boron-
doped diamond thin ﬁlm.

Growth conditions for coating insulating diamond powder with a
conducting diamond overlayer.

Growth conditions for coating carbon paper with boron-doped
nanocrystalline diamond.

Charge passed during the 1 h potentiostatic polarization of graphite
and GC powders at 1.0 and 1.6 V vs. Ag/AgCl (25 0C).

Charge passed (mC/cmz) during the 1 h potentiOstatic polarization
of graphite powder from 1.0 to 1.6 V vs. Ag/AgCl at 25, 50, and 80
°C.

Charge passed (mC/cmz) during the 1 h potentiostatic polarization
of GC powder from 1.0 to 1.6 V vs. Ag/AgCl at 25, 50, and 80 °C.

Change in current density (21], uA/cmz) for surface oxide reduction
before and after polarization on graphite powder as a function of
the polarization potential and temperature in 0.1 M HClO4 for 1 h.

Change in current density (AI, uA/cmz) for surface oxide reduction
before and after polarization on GC powder as a function of the
polarization potential and temperature in 0.1 M HClO4 for 1 h.

Ratio (n=6) of the D/Ezg Raman modes for graphite powder
electrodes at different polarization potentials and temperatures.

Ratio (n=6) of the D/Ezg Raman modes for GC powder electrodes
at different polarization potentials and temperatures.

Percent Pt lost (% L 03,) as a result of potentiostatic polarization of

Pt-loaded (20 wt. °/o) Vulcan XC-72 in 0.1 M HClO4 at different
polarization potentials and temperatures.

16

39

39

40

41

64

65

66

69

69

73

74

83

Table 4.1

Table 4.2

Table 5.1

Table 5.2

Table 6. I

Table 6.2

Table 6.3

Table 6.4

Cyclic voltammetric and heterogeneous electron-transfer rate
constant data for ﬁve redox systems at MSUMicro, Condias,
Sumitomo, and MSUNano boron-doped diamond thin-ﬁlm
electrodes.

Cyclic voltammetric and heterogeneous electron-transfer rate
constant data for Fe(CN)53'/4' and R11(NH3)53+/2+ at MSUMicro,
Condias, Sumitomo, and MSUNano boron-doped diamond thin-
ﬁlm electrodes, as a function of the solution pH.

Bulk conductivity values of graphite and CVD diamond powder
before (as-grown) and after (cleaned) washing in warm aqua regia
(30 min) and warm 30% H202 (30 min).

Cyclic voltammetric AEp values (mV) for 1 mM F e(CN)63'/ 4' at
conductive diamond powder, graphite powder, GC powder, and a
planar microcrystalline boron—doped diamond thin-ﬁlm electrodes
at three different scan rates.

Two different (A and B) current and charge conditions for the
pulsed galvanostatic deposition of Pt metal on carbon paper (TCP)
and boron-doped nanocrystalline diamond deposited on carbon
paper (BND) including the solution-exposed and actual surface
area (SA).

Electrode and Pt surface areas (SA) calculated for the Pt/carbon
composite electrodes (Pt/TCP-A, Pt/BND-A, Pt/TCP-B, and
Pt/BND-B) with the metal phase formed by pulsed galvanostatic
conditions A (0.146 mA/cmz or 0.0047 C/cmz) and B (1.25
mA/cm2 or 0.040 C/cmz) with the corresponding surface
enhancement factor, SEF.

Estimated speciﬁc surface area of Pt on Pt/TCP-A, Pt/BND-A,
Pt/TCP-B, and Pt/BND-B formed using pulsed galvanostatic
deposition conditions A (0.146 mA/cm2 Or 0.0047 C/cmz) and B
(1.25 mA/cm2 or 0.040 C/cmz) assuming 100% deposition
efﬁciency.

Peak position (Ev/2) and approximate limiting current (in-m) for the
oxygen reduction reaction on Pt/TCP-A, Pt/BND-A, Pt/TCP-B, and
Pt/BND-B in oxygen-saturated 0.1 M HCIO4 at 1 mV/s normalized
to the surface enhancement factor (in-m, 55p, uA/SEF) and to the
mass of Pt (tum, mm, mA/mgpt).

xi

103

108

125

132

157

160

162

164

Table 6. 5

Table 6. 6

Table 6. 7

Charge density (mC/cmz), change in faradaic current density for
electroactive surface-oxygen functional groups (21.1), and Raman
D/Ez8 ratio for bare carbon paper (TCP) and nanocrystalline
diamond deposited on carbon paper (BND) after polarization at
different potentials in 0.1 M HClO4 at 25 °C for l h.

Electrochemically—active Pt surface area (Pt/TCP-B and Pt/BND-B)
before and after anodic polarization at 1.4 V vs Ag/AgCl and 25 °C
in 0.1 M HClO4 for l h.

Half-peak position (Ep/z) and current (tum) for the oxygen reduction
reaction on Pt/TCP-B and Pt/BND-B in oxygen-saturated 0.1 M
HClO4 after anodic polarization at 1.4 V vs. Ag/AgCl and 25 °C in
0.1 M HClO4 for 1 h. The scan rate was 1 mV/s and the current
was normalized to the surface enhancement factor (tum, 55p,
uA/SEF).

xii

168

172

174

Figure 1.1

Figure 1.2

Figure 1.3

Figure 1.4
Figure 1.5
Figure 1.6

Figure I. 7

Figure 1.8

Figure 1.9

Figure 2.1

Figure 2.2

Figure 2.3

Figure 2.4

Figure 3.1

LIST OF FIGURES

History and projections of world energy use in British thermal units
(BTU).

(A) Schematic and (B) picture of a PEMFC.

Schematic of the triple phase boundary at the electrocatalyst/
electrocatalyst support/ionomer interface in a PEMFC.

Microstructures of (A) graphite, (B) GC, and (C) carbon black.
Schematic of proposed surface oxides on spz-bonded carbons.
Schematic of the bonding arrangement in sp3-bonded diamond.

Input, deposition process, reactor design, and internal variables that
affect output variables for CVD.

Simpliﬁed CVD scheme showing plasma composition, major
reactive components, and the primary reaction pathway for
microcrystalline diamond deposition.

Proposed mechanism for nanocrystalline diamond growth by
insertion of a C2 dimer into the diamond lattice.

Schematic microwave plasma-assisted CVD reactor used to grow
microcrystalline and nanocrystalline boron-doped diamond thin-
ﬁlm electrodes.

Schematic of the apparatus used to measure i—V curves for the
powder samples. This same general scheme was used to measure
the powder resistance with an ohmmeter, simply by removing the
power supply and arnmeter.

Design of the single compartment, three-electrode
electrochemical cell.

glass,

Carbon paper electrode preparation schematic.

Schematic of the oxidation of graphitic carbons including
intercalation and exfoliation of the graphite lattice, oxidation of

new exposed edge plane, and gasiﬁcation (corrosion) to CO and
CO2.

xiii

14

l6

18

22

24

26

37

44

46

48

59

Figure 3.2

Figure 3.3

Figure 3.4

Figure 3.5

Figure 3.6

Figure 3. 7

Figure 3.8

Figure 3. 9

Figure 4.1

Figure 4.2

Figure 4.3

Figure 4.4

Amperometric i—t curves for graphite and GC powders during a 1 h
polarization at 1.0 and 1.6 V vs. Ag/AgCl in 0.1 M HClO4 at 25 °C.
The top curve (A) is a plot of current and the bottom curve (B) is a
plot of the logarithm of the current.

Background cyclic voltammetric i-E curves for (A) graphite and
(B) GC in 0.1 M HClO4 at 0.1 V/s before and after polarization at 1
V vs. Ag/AgCl and 25 °C for 1 h.

Raman spectra for graphite and GC before anodic polarization
showing the E28 mode (~1580 cm'l) and D band (~1360 cm'l).

Charge density vs. polarization potential for Vulcan XC-72 as a
function of the temperature.

Change in faradaic current density for surface oxide reduction of
Vulcan XC-72 before and after polarization as a function of the
polarization potential and temperature in 0.1 M HClO4 for 1 h.

Background cyclic voltammetric i-E curve for Pt-loaded (20 wt. %)
Vulcan XC-72 in 0.1 M HClO4 at 0.005 V/s. Current is normalized
to the estimated Pt surface area.

Charge density vs. polarization potential for Pt-loaded (20 wt. %)
Vulcan XC-72 as a function of the temperature.

Background cyclic voltammetric i-E curves for Pt-loaded (20 wt.
%) Vulcan XC-72 in 0.1 M HClO4 at 0.005 V/s before and after
polarization from 0.8 to 1.4 V vs. Ag/AgCl at 25 °C. Current is
normalized to the estimated Pt surface area before polarization.

Scanning electron micrographs of (A) MSUMicro, (B) Condias,
(C) Sumitomo, and (D) MSUNano diamond thin ﬁlms.

Raman spectra for MSUMicro, Condias, Sumitomo, and MSUNano
diamond thin ﬁlms.

Cyclic voltammetric i-E curves for all four diamond types in 0.1 M
HClO4 at 0.1 V/s (geometric area = 0.2 cmz).

Cyclic voltammetric i-E curves for all four diamond types in (A)
0.1 mM Fe(CN)63'/4', (B) 0.1 mM Ru(NH3).,3“’2+ and (C) 0.1 mM
IrCl62'l3', all in 1 M KCl, and (D) 4-methylcatechol and (E) Fey/2+
in 0.1 M HClO4. Curves for scan rates of 0.1 and 1 V/s are shown
(geometric area = 0.2 cmz).

xiv

62

68

71

77

78

8O

81

83

93

95

97

101

Figure 4.5

Figure 5.1

Figure 5.2

Figure 5.3

Figure 5.4

Figure 5. 5

Figure 5.6

Figure 5. 7

Figure 5.8

Figure 5.9

Experimental and simulated (DigiSim® 3.03) cyclic voltammetric i-
E curves for (A) 0.1 mM Fe(CN)63"“‘ in 1 M KCl at the Sumitomo
diamond electrode (0.1 V/s) and (B) 0.1 mM Itch?” in 1 M KCl
at the MSUMicro diamond electrode (50 V/s). The experimental
voltammograms contain both the background and faradaic currents
while the simulated voltammograms contain just the faradaic
current.

Deposition scheme for coating insulating diamond powder with a
conductive, boron-doped diamond overlayer showing both full and
partial grain coverage as well as fusion of neighboring grains. The
shaded areas represent the conducting diamond overlayer.

Scanning electron micrographs of (A) bare diamond powder and
conductive diamond powder after (B) 1, (C) 2, and (D) 4 h coating
times.

X-ray diffraction spectrum of a 4 h grth diamond powder
sample.

Raman spectra of bare diamond powder and conductive diamond
powder after coating times of 1, 2, 4, and 6 h.

Scanning electron micrographs of a 4 h grth conductive
diamond powder (A) as-grown and (B) aﬁer acid washing.

I-V curves for a 4 h grth conductive diamond powder, graphite
powder, and a boron-doped diamond thin ﬁlm over a $10 V range.

(A) Background cyclic voltammetric i-E curves for conductive
diamond powder electrodes (2, 4, and 6 h deposition times) in 0.1
M HClO4 at 100 mV/s. (B) Enlarged view of an 800 mV region.
Current is normalized to the diamond powder surface area exposed
to solution.

(A) Cyclic voltammetric i-E curves for a conductive diamond
powder electrode (4 h coating time) in 1 mM Fe(CN)63'/4' at scan
rates between 10 and 500 mV/s and (B) a plot of the oxidation peak
current versus the scan ratem.

Cyclic voltammetric i-E curves for 1 mM Ru(NH3)63+/2+ for a
conductive diamond powder electrode (4 h coating time) at scan
rates between 10 and 300 mV/s with (A) 10 wt. % PTFE binder and
(B) 5 wt. % PTFE binder.

XV

106

119

120

122

124

126

127

129

134

136

Figure 5. 10

Figure 5. I I

Figure 6.1

Figure 6.2

Figure 6.3

Figure 6.4

Figure 6.5

Figure 6.6

Figure 6. 7

Chronoamperometric i-t curves for the 4 h conductive diamond
powder, graphite powder, and glassy carbon powder electrodes
recorded during potentiostatic polarization at 1.6 V vs. Ag/AgCl
and 25 °C for 1 h.

Scanning electron micrographs of a conductive diamond-coated
diamond powder electrode (4 h coating time) (A) before and (B)
after and GC powder electrode (C) before and (D) aﬁer constant
potential polarization of 1.6 V vs. Ag/AgCl for 1 h in 0.1 M
HClO4.

Scanning electron micrographs of (A) bare carbon paper (TCP), (B)
an individual bare carbon ﬁber within the paper, (C) boron-doped
nanocrystalline diamond deposited on the carbon paper (BND), and

(D) a cross-section of a nanocrystalline diamond-coated carbon
ﬁber.

X-ray diffraction spectra of bare (TCP) and boron-doped
nanocrystalline diamond-coated carbon paper (BND).

Raman spectra for the bare (TCP) and boron-doped nanocrystalline
diamond-coated carbon paper (BND).

Background cyclic voltammetric i-E curves for bare (TCP) and
boron-doped nanocrystalline diamond-coated carbon paper (BND)
in 0.1 M HClO4 at 100 mV/s. The current is normalized to the
electrode area estimated to be in contact with the solution.

Scanning electron micrographs of Pt/carbon composite electrodes
(Pt/TCP-A, Pt/BND-A, Pt/TCP-B, and Pt/BND-B) under two
different pulsed galvanostatic deposition conditions A (0.146
tnA/onn2 or 0.0047 C/cmz) and B (1.25 mA/cm2 or 0.040 C/cmz).

Cyclic voltammetric i-E curves of Pt/carbon composite electrodes
Pt/TCP-A, Pt/BND-A, Pt/TCP-B and Pt/BND-B deposited by
pulsed galvanostatic deposition (A = 0.146 mA/cm2 or 0.0047
C/om2 and B = 1.25 tnA/om2 or 0.040 C/cmz) in 0.1 M HCIO. at 50
mV/s. Currents are normalized to the carbon electrode surface
area.

Linear sweep voltammetric i-E curves for the oxygen reduction
reaction at Pt/TCP-A, Pt/BND-A, Pt/TCP-B, and Pt/BND-B in
oxygen-saturated 0.1 M HClO4 at 1 mV/s. Current is normalized to
the surface enhancement factor (SEF).

xvi

139

139

150

151

152

154

158

161

165

Figure 6.8

Figure 6.9

Figure 6.10

Figure 6. 11

Figure 6.12

Scanning electron micrographs of (A) BND and (B) TCP after
potentiostatic polarization at 1.6 V vs. Ag/AgCl at 25 °C for 1 h.

Raman spectra of BND before and after potentiostatic polarization
at 1.6 V vs. Ag/AgCl in 0.1 M HClO4 at 25 °C for 1 h.

Scanning electron micrographs of Pt/TCP-B and Pt/BND—B before
and after potentiostatic polarization at 1.4 V vs. Ag/AgCl and 25 °C
in 0.1 M HClO4 for 1 h.

Cyclic voltammetric i-E curves for Pt/TCP-B and Pt/BND-B in 0.1
M HClO4, recorded at 50 mV/s normalized to the electrode surface

enhancement factor (SEF), before and after a 1 h polarization at 1.4
V vs. Ag/AgCI at 25 °C.

Linear sweep voltammetric i-E curves for the oxygen reduction
reaction at Pt/TCP-B and Pt/BND-B in oxygen-saturated 0.1 M
HClO4 before and after polarization at 1.4 V vs. Ag/AgCl at 25 °C
in 0.1 M HClO4 for 1 h. Curves were recorded at 1 mV/s and are
normalized to the electrode surface enhancement factor (SEF).

xvii

167

169

171

172

173

Chapter 1

Introduction

1.1 Background

Energy use in the United States and abroad is increasing at an alarming rate. As a
result, the production of energy from fossil ﬁiels, such as oil and coal, and renewable
sources, such as water and wind, is being outpaced by demand. Each year, the Energy
Information Administration of the US. Department of Energy releases the International
Energy Outlook report, which details various aspects of the world energy market. Figure
1.1, adapted ﬁ'om the 2005 report, illustrates world energy use (by fuel type) from 1970
to 2004, as well as energy demand projections until 2025.1 As evident, world energy
consumption is expected to rise almost 70% by the year 2020 due to population growth
and emerging industrialized economies.2 Considering that the main source of world
energy is fossil fuels, this rise in consumption will lead to a 50% increase in the amount
of CO2 released into the environment each year.2 The atmospheric CO2 concentration has
already increased from 280 to 370 parts per million (ppm) during the last 150 years.3 If
this trend continues, the atmospheric CO2 concentration is expected to rise to over 550
ppm during the next 100 years.3 This prediction has detrimental environmental
implications in terms of global warming.4 In addition, oil production worldwide is
expected to peak within the next few decades.3 Although supplies will remain for some
time afterward, prices will soar. To minimize these monetary and environmental costs, as
well as to ensure that world energy supply can meet growing energy demand, it is

imperative that alternative energy sources be developed and utilized in the near term.

 

Ur
Ur
I

History Projections. ’

M
O
I I

 

 

 

 

 

. g I I
D 45 - a , ,
[-1 401 .. ir
m 35 L I \ , ’ : - - Petroleum
* I " \ '- " ’ 5 -----Natural Gas
c 30 I " I i I """""
.52 i . In”’. . - w Cbal
= 25 ;.-. .,---d'v-..i—". 1 -—Nuclear
': 20 F- \ '''' \'/'.\.-»’."." é Hydropower
F3 15 f 1 - - -- Renewable/Other
:3 .
O’

 

 

 

 

2010 2020

1970 1980 1990 2000
Year

Figure 1.1 History and projections of world energy use in British thermal units
(BTU). (Adapted ﬁ'om ref 1)

Hydrogen has received much attention in recent years as a clean, plentiful source
of energy that may serve as an alternative to commonly used fossil fuels.5 The principal
hydrogen energy conversion device used today, the polymer electrolyte membrane fuel
cell (PEMFC), is analogous to a combustion engine that burns gasoline. However, the
PEMFC converts the chemical energy of hydrogen directly into electrical energy, instead
of into heat as in the combustion engine. This direct conversion signiﬁcantly reduces the
amount of energy lost during the conversion cycle. In fact, the theoretical efﬁciency of a
fuel cell is 100%, while the combustion engine has only 40% efﬁciency under ideal
operating conditions and is reduced to about 25% in practice. From an environmental
perspective, the use of hydrogen would eliminate harmful emissions like CO2, NOX, and
SOx. Hydrogen could also be beneﬁcial economically and militarily by reducing U.S.

dependence on foreign oil.

Despite the attractiveness of hydrogen as an environmentally-friendly,
domestically-available energy source, there are remaining problems to be overcome
before it will be a viable energy alternative. These include production, transport, and
storage of the hydrogen fuel, in addition to signiﬁcant technical issues that must be
solved with the PEMFC in terms of cost, efﬁciency, and operational lifetimef"5 A
considerable amount of research has been done to address some of these problems. For
example, a survey of the literature using the SciFinder Scholar® search engine reveals
that over 20,000 articles have been published on fuel cell science and technology since
1970. The interested reader is referred to two reviews on the major advances in fuel cell
technologies from 1960 to 2000 by Costarnagna and Srinivasan.6’7
A global research effort is now focused on almost every aspect of PEMFC
development to improve and advance the so-called "hydrogen economy." The Ofﬁce of
Science of the US. Department of Energy published a report in February 2004 entitled
"Basic Research Needs for the Hydrogen Economy," which outlined 11 high-priority
research directions for the hydrogen fuel and the PEMFC.5 One of the key research
needs identiﬁed is the development of low-cost, highly-active, and durable electrodes
(electrocatalyst, carbon electrocatalyst support, and ionomer) for the PEMFC. Studies of
automotive PEMFC performance degradation have shown that voltage losses of >50
HV/h occur during operation, while target losses of less than 10 uV/h have been proposed
for a 5,000 h operating lifetime.5 These voltage losses are caused by degradation of the
electrode components, but the speciﬁc mechanisms of the degradation are not well-

understood. Thus, studies of the conditions under which these voltage losses occur, as

well as of the mechanisms involved in the electrode performance degradation are
important for the development of more durable electrode materials.

The research presented in this dissertation was conducted to address part of the
multifaceted electrode stability issue and speciﬁcally focused on corrosion of the carbon
electrocatalyst support. This chapter presents background information and provides a
context for the current work, including a description of PEMF C Operating principles and
electrode materials, speciﬁc stability criteria for electrocatalyst supports in PEMFCs, and
a description of current corrosion issues. The details of a proposed dimensionally stable
electrocatalyst support, electrically conducting diamond, including material properties,
established stability, and growth mechanisms are then presented. Next, the speciﬁc
research questions addressed in each chapter are given, followed by an outline of the

dissertation.

1.2 Fuel Cell Materials and Operation

Although there are several types of fuel cells that are typically classiﬁed by the
speciﬁc electrolyte employed, the discussion in this chapter and the research presented in
this dissertation will focus on the hydrogen-powered PEMFC. The PEMFC is a low-
temperature (~80 to 100 °C) energy conversion device that is being investigated for

transportation and decentralized stationary power systems. As shown in the simpliﬁed
schematic and picture in Figure 1.2A and B, respectively, the PEMFC consists of two
electrodes that contain an electrocatalyst (e.g., Pt) dispersed on a high-surface—area
Support (e.g., carbon black). The electrodes are connected through an external circuit and

Separated by a hydrated solid polymer electrolyte membrane, which is typically a ~10-30

pm thick perﬂuorosulfonic acid polymer, such as Naﬁon®,

(CF2CF2)X[CF(OR1803H)CF2]y.

 

 

 

 

Cathode I Anode
Electrolyte

Assembled fuel cell

1an plate

*Iililii , ’

plate

Anode/cathode

 

Figure 1.2 (A) Schematic (Adapted from ref 8) and (B) picture of a PEMFC.

In the PEMFC, hydrogen is directed into the fuel cell through a gas diffusion
layer (e.g., carbon paper), enters the anode, and reacts on an electrocatalyst to produce

two protons and two electrons, as shown in Equation 1.1:
H2 9 2H” + 2e' E° = 0.000 v vs. SHE (1.1)

where SHE is the standard hydrogen electrode (E° = 0.000 V). The electrons travel from
the electrocatalyst to the cathode through the electrocatalyst support material, bipolar
plate (e.g., Ti or stainless steel), and the external circuit of the cell. The protons travel to
the cathode through the hydrated solid polymer electrolyte (an ionomer) that is present
near the surface of the catalyst. The electrons and protons then react with oxygen from
either air or a pure gaseous feed to form water as the only byproduct, according to

Equation 1.2:
02 + 4H“ + 4e‘ —> 2H2O E° = 1.229 v vs. SHE (1.2)

The overall cell reaction (ECell = Emmi. — EAnode), shown in Equation 1.3, has a standard

potential of 1.229 V vs. SHE.
H2+ % 02 '9 H20 0 = 1.229 V VS. SHE (1.3)

Much research has been directed toward the improvement of fuel cell electrode
materials in terms of efﬁciency, stability, and cost. Fuel cell electrodes generally consist
of nanometer—sized electrocatalyst particles, such as Pt, that are highly dispersed onto a

high-surface-area (~100 mZ/g), spZ-bonded carbon powder support. Finely dispersed
electrocatalyst particles on a stable and conductive support increase the number of
electrocatalytically-active atoms available for the fuel cell reactions, thereby reducing the
loading requirements of expensive metallic catalysts.9’10 Although high-surface-area

graphitic carbons offer greater chemical stability and mechanical strength as

electrocatalyst supports, graphitization decreases the number of active sites for
electrocatalyst attachment. Instead, high-surface-area carbon blacks, such as Vulcan XC-
72, are typically used as the electrocatalyst support. To form the electrode, a catalyst-
loaded carbon black ink is sprayed or painted onto a hydrophobic carbon paper gas
diffusion electrode. Two of these layers are then hot-pressed under pressure into each
side of a Naﬁon® membrane to form the so-called membrane electrode assembly (MEA).
The interested reader is referred in an excellent review by Antolini that discusses the
recent advances in electrode preparation and materials for PEMFCs.”

Regardless of the manner in which the MBA is formed, the most important factor
in determining the electrochemical performance and lifetime of a fuel cell is the stability
of the interface where the fuel, electrocatalyst, and ionomer are all in direct contact. This
is so-called triple phase boundary (TPB). Figure 1.3 shows a simpliﬁed schematic of this
interface at the anode, where the dark circles represent the carbon support and the small
light circles represent the electrocatalyst particles. Only those electrocatalyst particles
that are in direct contact with the support material and with the polymer electrolyte will
be active for the hydrogen oxidation or the oxygen reduction reaction at the anode and
cathode, respectively. Considering this, it is not surprising that the morphological and
microstructural stability of the materials at this interface affect device performance,
possibly more than any other fuel cell component. Issues, such as overpotential
limitations and resistance effects (electronic and ionic), within the TPB strongly inﬂuence
fuel cell efﬁciency. Materials research to reduce the resistance (mass transport, kinetic,
and ohmic) at the TPB has focused on optimization of the properties of the conductive

polymer membrane electrolyte, reduction of the electrocatalyst loading, and improved

8,9,12,13

electrocatalyst utilization. However, the long—term chemical stability of all

components in the MBA is a critical problem that requires a more focused research effort.

 

I T
Electrocatalyst support

Figure 1.3 Schematic of the triple phase boundary at the electrocatalyst/
electrocatalyst support/ionomer interface in a PEMFC.

1.3 Carbon Corrosion in Fuel Cells

Although the thermodynamic potential for the fuel cell reaction is 1.23 V vs.
SHE, in practice, fuel cells operate at ca. 0.7 V vs. SHE, primarily because of sluggish
kinetics for the oxygen reduction reaction at the cathode. During start-up and shut-down,
however, the materials in the MEA are subject to potentials as high as 1.2 to 1.5 V vs.
SHE.M In the presence of liquid water and oxygen, as well as low pH (<1) and high
temperature (~80 °C), carbon corrosion is especially favorable at the fuel cell cathode at
these high potentials. Some work has been done to study the conditions that lead to

corrosive environments in a fuel cell. For example, it has been shown that ﬂooding of the

MEA, caused by factors such as insufﬁcient operating temperature, result in non-uniforrn
current distribution through the MEA.15 This creates potential "hot spots" and may lead
to high, localized corrosive potentials. Recent work by Reiser and co-workers has shown
that high potentials may be due, in part, to a mechanism called reverse-current decay.16
In this work, a one-dimensional model was used to show that a high interfacial potential
(1.44 V vs. SHE) is present at the anode during PEMFC startup.l6 This potential
develops because at some locations on the anode, the oxygen reduction reaction takes
place as a result of the presence of air that leaked in when the cell was off, while at others
locations, the hydrogen oxidation reaction takes place after introduction of the hydrogen
fuel.16

Research has shown that stability issues with both the metal electrocatalyst and
the carbon support, particularly at the cathode in the presence of Oxygen, exist at these
high potentials. Speciﬁcally, degradation of the carbon electrocatalyst support has been
shown to lead to loss of electrocatalyst activity, increased ohmic resistance, reduced
operational efﬁciency of PEMFCS, and eventually complete electrode failure.9"7'2'
Mathais and co-workers published an excellent review of the current issues with fuel cell
commercialization including both cost and stability limitations.” In this work,
production goals and benchmark stability targets were proposed. For use in automotive
applications, fuel cells are predicted to undergo ca. 30,000 on/off cycles. With a targeted
lifetime performance loss of less than 30 mV for a PEMFC Operating at 1.5 A/cm2 over
10 y and 5,500 h of operation, materials used in the MEA must be stable to potentials as

high as 1.5 V vs. SHE for up to ~100 h.14

It is important to develop an understanding of the mechanisms by which current
spZ-bonded supports degrade and corrode in a working-PEMFC environment. Many of
the early reports on the microstructural degradation and corrosion of spz-bonded carbons
were carried out in both alkaline and acidic media, typically at high temperatures (>100
°C).9’2"26 Consideration of these studies is useful for gaining insight into carbon

corrosion mechanisms. In acidic media, potential-dependent corrosion of spz-bonded

carbon materials occurs according Equation 1.4.9
C+2H209CO2+4H++46 E°=O.118Vvs. SHE (1.4)

Kinoshita and Bett measured the corrosion current and CO2 evolved from a hi gh-surface-
area carbon black (1000 m2/ g) in 96% H3PO4 under potentiostatic conditions (0.7 to 1 V
vs. SHE) at 135 °C.24 They found that electrochemical corrosion proceeded through two
independent processes, surface oxide formation and carbon gasiﬁcation to CO2.24 Unlike
high temperature gas-phase oxidation of carbon in which the surface oxide layer reduces
the active surface area and slows the rate of gasiﬁcation, electrochemical formation of
surface carbon oxides does not inhibit the carbon gasiﬁcation.24 In fact, a positive
correlation has been found between the percentage of surface oxides and the corrosion of
carbon black materials, perhaps because the oxygen atoms serve as nucleation sites for

degradation.21

Antonucci and co-workers reported an increase in the corrosion current
recorded at 1 V vs. SHE from ca. 0.03 mA/rng to 0.3 mA/mg when the O/C ratio was
raised from 2 to 14.21

Other work has related the graphitic nature of the carbon support to the extent of

corrosion. For example, Stonehart postulated that disordered regions of the carbon, such

as edge plane and defect sites, were oxidized ﬁrst, followed by slow attack of the more

10

ordered graphene sheets.25 In addition, heat-treated carbons (i.e., graphitized) were found
to be more resistant to corrosion than were the as-received materials, further supporting
the belief that the degree of microstructural order is critical for corrosion resistance.”
Similarly, Antonucci and co-workers found that as the dam-spacing of the carbon
increased, (i.e., as the material became less ordered), the corrosion current increased due

21 More detailed kinetic and mechanistic

to the so-called order/disorder dependence.
studies of carbon black oxidation in aqueous ozone have shown that oxidative
degradation proceeds through a three-stage process: 1) the formation of chemical
functional groups on the surface resulting in degradation products (e.g., polyphenols), 2)
oxidation of disordered carbon to CO2, and 3) further oxidation of the more ordered (i.e.,
graphitic) carbon to CO2.27 Regardless of the reagents used to study oxidative
degradation of carbon blacks, results have shown that the microstructure strongly
inﬂuences the extent of corrosion. The less-ordered portions of the material degrade ﬁrst
and more severely, followed by degradation of graphite-like regions during prolonged
exposure to harsh conditions.23

Appleby studied the corrosion of graphitized carbon black, graphite, and various
other carbon blacks in a phosphoric acid fuel cell environment.9 Corrosion of these
carbon supports, as measured by the corrosion current, was found to depend on the
carbon microstructure and increased with increasing potential, temperature, and pressure.
Operating at 1 atm, 190 °C, and 0.7 V vs. SHE in 98 % H3PO4, the corrosion rate of all
carbons was initially high, falling off to a pseudo-steady state level after about 100 hours.

The total mass loss of graphitized carbon black, graphite, and carbon black was

approximately 40, 50, and 75%, respectively. These differences were attributed to

11

structure-dependent corrosion rates. The more disordered regions are characterized by a
higher free energy and, hence, a higher reactivity.9

Recent work has focused on lower temperature systems (50 to 90 °C) to study
spz-bonded carbon oxidation/corrosion in an environment more similar to that in a
PEMFCH'W‘28 Kangasniemi and co-workers studied the electrochemical surface
oxidation of a carbon black, Vulcan XC-72, after polarization from 0.6 to 1.2 V vs. SHE
at both room temperature and 65 °C.‘7 They found evidence for surface oxidation at 0.8
V vs. SHE at 65 °C and at 1.0 V vs. SHE at room temperature.17 Roen and co-workers
examined the effect of Pt on spZ-bonded carbon support corrosion and found that CO2
evolution increased as the Pt mass fraction increased, suggesting that Pt accelerated the
rate of carbon corrosion.19 Maja and co-workers evaluated the lifetimes of carbon blacks
by monitoring the voltage stability under a constant current load of 200 mA/cmz.29
Vulcan XC-72 outperformed Black Pearls 3700 and Shawinigan Black AB50 (acetylene
black) with a constant potential output over 1000 h, while the others failed aﬁer ca. 50 h
and 400 h, respectively.29 More recently, Stevens and Dahn showed that combustion of
the cathode carbon support was enhanced when impregnated with Pt in a PEMFC.30 At
125 °C, a carbon support containing 57% Pt lost some 40% of its mass due to corrosion,

while the mass of the carbon support with no Pt was relatively unchanged.30

1.4 spz-Bonded Carbon Structure

It is obvious from the above discussion that corrosion of spz-bonded carbon
electrocatalyst support materials is a problem in fuel cells. However, until quite recently,

many researchers believed that these carbons were stable under PEMFC conditions

12

because of the low temperature (<100 °C) and operating potential (~0.7 V vs. SHE). The
mechanism of degradation is still not well-understood, especially with regard to new and
ongoing MBA and electrocatalyst development. It is important to gain a better
fundamental understanding of the chemical and electrochemical processes that occur
under PEMFC operating conditions, especially to aid in the development of an advanced
electrocatalyst support material that will provide enhanced stability with respect to spz-
bonded carbons. This understanding must be based on consideration of important
structural factors (e.g., edge plane density and surface oxide content) that affect the
chemical, physical, and electrochemical properties of carbons.

Figure 1.4 shows the microstructure of model spz-bonded carbon materials:
graphite, glassy carbon (GC), and carbon black. All structures are composed of six-
membered carbon rings in which each carbon atom is trigonally bonded to three others.
This bonding arrangement creates a planar honeycomb array called the basal plane, as
seen on the face of graphite in Figure 1.4A. The carbon-carbon bond distance in the
basal plane for all spZ-bonded carbons is 1.42 A.“ Graphite has the most well-ordered
spZ-bonded carbon microstructure, characterized by a stacking of the graphene sheets in
an ABAB... sequence. The ﬁrst structural model of GC, shown in Figure 1.4B, was
published by Jenkins and Kawamura in 1971.32 They proposed that the GC structure
consisted of tangled graphitic ribbons, connected by highly-strained carbon-carbon
covalent bonds, which resemble the polymer precursor backbone.32 More recent work by
Harris has suggested that GC is actually composed of imperfect, multilayered, porous
fullerene-related nanoparticles (i.e., structures with non-six-membered rings)” Carbon

black microstructure strongly depends on the powder preparation conditions, including

13

temperature, feedstock composition (i.e., impurities), and speed of the synthesis
reaction.34 However, in general, carbon blacks are turbostratic in nature, composed of a
bidimensional organization of stacked, graphitic planes that are arbitrarily rotated with
respect to one another, as shown in Figure 1.4C.23 High-resolution transmission electron
microscopy measurements have shown that carbon blacks are composed of small
spherical particles that join together to form agglomerates.34 The agglomerates coalesce

as a result of Van der Waals forces to form larger particles, called aggregates.34

 

 

C/
/

Figure 1.4 Microstructures of (A) graphite, (B) GC, and (C) carbon black. Note:
drawings not to scale. (Adapted from ref 35)

14

spz-bonded carbons are typically characterized by the dimensions of the crystallite
lattice determined from X-ray diffraction (XRD) measurements. In this characterization
scheme, Lc is the crystallite stacking height perpendicular to the graphene sheets, I... is the
crystallite width, and dim is the interlayer spacing. Comparative dam and La values for
highly oriented pyrolytic graphite (HOPG), polycrystalline graphite, GC, and carbon
black are listed in Table 1.1. HOPG, analogous to a single crystal metal, is a benchmark
model carbon to which the structural properties (LC, La, and (1002) of other carbons are
typically compared. For example, the dam-spacing is commonly used to quantify the
"graphitic nature" of spZ-bonded carbons; carbon microstructures with (1002 near that of
HOPG (3.354 A) are more "graphitic" than those with larger layer plane spacing. Carbon
blacks can be treated at high temperatures (~2000 °C), a process called graphitization, to
reduce the d002 spacing closer to that of HOPG. GC, in contrast, is a so-called non-
graphitizing carbon because heat treatment does not change the d002 spacing. La is also
an important parameter because carbons with a smaller crystallite width have a higher
proportion of edge plane density. L. for HOPG is greater than 104 A meaning that the
basal plane density in this material is high, while the edge plane density is low. La for
GC and carbon black are signiﬁcantly smaller, resulting in more disordered
microstructures with a higher edge plane density and thus, a higher electrochemical and
chemical reactivity.

The electrical and chemical properties of spZ-bonded carbons are anisotropic
because of the difference in bonding perpendicular and parallel to the basal plane. For
example, electrical conduction parallel to the basal plane in spz-bonded carbons is high

because three of the four carbon valence electrons are occupied within the extended 1:-

15

electron ring system. In contrast, electrical conduction perpendicular to the basal plane,
across the so-called edge plane, is lower because bonding between the layer planes is a
result of weak Van der Waals interactions. The resistivities of the basal and edge plane
of HOPG are 0.17 and 4 x 10'5 Q-cm, respectively.” . The chemical reactivity of the
respective graphitic planes also varies. Atoms in the basal plane have saturated valencies
and are typically unreactive, while the unsaturated edge plane atoms are active sites for
chemisorption of heteroatoms, such as oxygen. As seen in Figure 1.5, the edge plane of
spz-bonded carbon typically contains several types of oxides such as hydroxyls, ketones,
and quinones. The coverage and type of functional group depends on the type of spz-
bonded carbon and its past history.

Table 1.1 Cr‘ystallite width (La) and interlayer spacing (d002) for some common spz-
bonded carbons.3 ’32'35

 

 

 

Carbon Material doo2 (A) L. (A)
HOPG 3.354 >104
Polycrystalline graphite 3.355 >100
GC 3.48 ~50
Carbon black 3.43-3.57 ~20
Carboxylic acid
Lactone

Phenol Carbonyl 0 O 0\\
OH O H CH

0 o-quinone

p-quinone

  

Figure 1.5 Schematic of proposed surface oxides on spz-bonded carbons.

16

1.5 Boron-Doped Diamond

An ideal PEMFC electrocatalyst support material should have (i) electrical
conductivity (>10 S/cmz), (ii) high-surface-area (~100 mZ/g), (iii) mechanical strength,
(iv) chemical resistance in aqueous environments, (v) thermal stability (0 to 150 °C), and
(vi) microstructural stability at high potentials (e.g., >1 V vs. SHE) and current
densities.36 Carbon blacks commonly used in fuel cells, such as Vulcan XC-72, possess
many of these characteristics. However, as discussed above, these carbons are
susceptible to microstructural and morphological degradation under typical fuel cell
operating conditions and this affects the operational efﬁciency and electrode lifetime. It
is our supposition that boron-doped diamond can function as a dimensionally stable
support material, outperforming currently used spz—bonded carbOn materials.

Diamond is a technologically important material because of its unique properties,
such as extreme hardness, high electrical resistance, chemical inertness, high electron and
hole mobility, high thermal conductivity, and optical transparency.”38 Unlike the three-
fold bonding arrangement in graphite, carbon atoms in diamond are bonded in a
tetrahedral arrangement forming six-membered rings in a chair conformation, as shown
in Figure 1.6. Natural diamond is an electrical insulator with a band gap of ~5.5 eV and a
resistivity greater than 1012 Q-cm.39'40 However, synthetic diamond can be grown via
chemical vapor deposition (CVD) and made electrically conductive by doping with
boron. Boron-doped diamond (BDD) single crystals and polycrystalline thin ﬁlms can be
prepared with resistivities as low as ca. 0.005 Q—cm.”40 Both high-quality

microcrystalline and nanocrystalline thin ﬁlms can be produced by CVD methods.

17

Diamond thin ﬁlms that are hydrogen-terminated, hydrophobic, and conductive have

40-43

been used for numerous electrochemical applications.

 

Figure 1.6 Schematic of the bonding arrangement in sp3-bonded diamond.

BDD has many advantageous properties that make it an attractive candidate as an
electrocatalyst support including mechanical strength, chemical and thermal stability, and

corrosion resistance.40“‘2’44

For example, Swain compared the morphological,
microstructural, and electrochemical stability of BDD electrodes with that of HOPG and
GC in acidic ﬂuoride media.44 In this study, each electrode was exposed to
potentiodynamic cycling in 1.0 M HNO3 + 0.1 M NaF at 50 °C for 2 h,44 conditions
known to cause severe corrosion of HOPG and Grafoil®.45“17 It was found that structural
degradation, including delamination and fracturing of the HOPG graphene sheets, as well

as severe pitting of GC resulted, while the diamond microstructure was unaffected. The

potential window for BDD after potential cycling was larger than before with an

18

unchanging voltammetric background current, suggesting the surface had simply been
cleaned of nondiarnond carbon impurity. In contrast, the background current for both
HOPG and GC increased signiﬁcantly with evidence for redox-active surface oxide
formation at the edge plane sites. Similar results were found during the anodic
polarization of BDD thin-ﬁlms, HOPG, GC, and Grafoi1® in acidic chloride media.48
Each electrode was exposed to 1.0 M HNO3 + 2.0 M NaCl at constant current densities of
0.05 or 0.5 A/cmz. Severe structural and electrochemical degradation were observed for
the spZ-bonded carbon electrodes, while excellent stability was seen for BDD.48

As a ﬁrst step towards the development of diamond as a new electrocatalyst
support material, Pt/diamond composite electrodes were prepared using planar
microcrystalline and nanocrystalline boron-doped diamond thin ﬁlms.”52 Work is
ongoing to understand the metal nucleation and grth on diamond so that the
electrocatalyst deposition can be optimized in terms of catalyst particle size and

49-52

distribution. The Pt/diamond composite electrodes are active for the oxygen

reduction reaction, a prerequisite for possible use in PEMFCS.49‘51

Moreover, the
Pt/diamond composite electrode exhibited superior stability in hot concentrated
phosphoric acid compared to an spz-bonded carbon cloth electrode impregnated with Pt.52
In this study, the electrodes were exposed to anodic polarization in 85% H3PO4 at 170 °C
and 0.1 A/cm2 for two 1-h periods. Severe morphological and microstructural
degradation were observed for the spz-bonded carbon electrode, while no damage or loss
of catalyst activity was observed for the Pt/diamond composite electrode.52

Despite the advantages of diamond in terms of its morphological and

microstructural stability, the surface area of the thin-ﬁlm electrodes does not approach

19

the value necessary for a viable support material (10-100 mz/g). In addition, the solid,
planar thin-ﬁlm geometry is not suitable for use in PEMFCs because it is not gas
permeable. Research and development of high-surface-area, gas permeable (i.e., powder)

forms of conducting diamond is presented in this dissertation.

1.6 Chemical Vapor Deposition (C VD) of Diamond

sp3-bonded diamond is the thermodynamically stable form of carbon at high
temperature and high pressure, while spz-bonded graphite is the stable form under
ambient conditions.” Early efforts to synthetically produce diamond involved the costly
production from graphite, or other spZ-bonded carbon precursors, in the presence of
molten metal catalysts under both high pressure (7 to 10 GPa) and high temperature
(1700 to 2000 K).53 Diamond is, in fact, metastable with respect to graphite at low
temperature and pressure as the free energy difference between the two materials at 298
K and 1 atm is only 2.900 kJ/mol (~0.03 eV per atom).54’55 Thus, in theory, diamond
grth can be achieved in the metastable region of the carbon phase diagram as long as
graphite formation is suppressed.53

Low pressure diamond synthesis via CVD dates back to the 19508 and 60s when
research was conducted in both the former Soviet Union and the United States.”’56 Since
then, developments in CVD technology have enabled the production of diamond for a
number of applications, such as optical windows, wear-resistant coatings for cutting
tools, therrnistors, and electrodes.53 Several CVD growth methods exist for producing
diamond, all of which utilize a carbon-containing reactant source gas (e.g., methane and

hydrogen), an energy source to promote reactant dissociation (e.g., microwave, hot-

20

ﬁlament, or DC discharge), and a substrate (e.g., Si, W, Mo). High-quality (i.e., low
defect density and low nondiamond carbon impurity content) CVD diamond is typically
produced at low pressure (10 to 500 Torr), high gas temperature (>1700 °C), high
substrate temperature (700 to 800 °C), and low carbon concentration in the source gas
(e. g., 0.5%) with an energy source for gas dissociation and radical formation.”55

The morphology, microstructure, and properties of the thin-ﬁlm CVD diamond
are controlled by the method (e.g., excitation source, reactor design, and substrate
variables) and deposition conditions (e.g., source gas composition, temperature, and

pressure)_53,55-57

In general, the factors that affect CVD reactor performance and the
resultant diamond ﬁlm properties can be divided into four main categories: input, reactor
design, deposition process, and internal variables. The pertinent control parameters of
each category and the affected reactor output variables are shown in Figure 1.7.
Although the discussion of each variable is beyond the scope of this work, a number of
reports have been published that examine various CVD conditions and the resultant ﬁlm
characteristics and growth mecharrisms.37’53’5%1

Microwave plasma-assisted chemical vapor deposition (PACVD) is a common
deposition method that was utilized for all diamond grth presented in this dissertation.
In PACVD, microwave energy is coupled into a deposition chamber, causing the source
gas components to dissociate into reactive species (radicals). Both microcrystalline
(grain size ~1-3 um) and nanocrystalline (grain size ~10-15 nm) diamond thin ﬁlms can
be grown via PACVD by selection of the appropriate source gas composition, pressure,

and microwave power. These parameters control the relative nucleation rate and

subsequent crystallite size of the polycrystalline thin ﬁlms. A brief introduction to the

21

growth mechanisms for microcrystalline and nanocrystalline thin ﬁlms is provided

below. Details of the morphology and microstructure of these two diamond types will be

discussed in Chapter 4.

Reactor Design Variables

 

 

 

 

 

 

Type of Power Source
Reactor Conﬁguration
Reactor Wall Material
Substrate Biasing
Substrate Cooling/Heating
Controllable Input 1 1 Output Variables
. A. Reactor Performance
Variables Internal Variables Linear Growth Rate
Input Power Discharge Volume Total Growth Rate
Gas Chemistry Gas Temperature Carbon Conversion
13332:; Ohwhgts‘irre Electron Temperature b
Substrate Temperature Resldencc Tim B. Frlm Characteristics
_> Unrforrnrty
T I Structural Quality
Morphology
Deposition Process Variables Texture
Deposition Time

Substrate Seeding Procedure
Substrate Material and Size

Figure 1.7 Input, deposition process, reactor design, and internal variables that affect
output variables for CVD. (Adapted from ref 53)

I. 6.1 Microcrystalline Diamond Growth Mechanism
Many reports have been published on the CVD growth mechanism of
microcrystalline diamond.37’55'58’60'62 This discussion will be limited to microcrystalline

diamond deposited in a hydrogen-rich plasma with a low concentration of gas-phase

hydrocarbon (< 1%), such as methane. Increases in the CH4/H2 ratio are known to

22

increase the defect density and nondiamond carbon impurity content in CVD diamond
ﬁlms.53 Under conditions of low CH4/H2, the primary reactive species in the plasma are
atomic hydrogen and carbon-containing radicals. These species undergo a variety of
complex processes during diamond growth that include surface adsorption, diffusion,
reaction resulting in diamond nucleation, suppression of nondiamond carbon growth, and
eventually, growth of a diamond ﬁlm.62 Figure 1.8 shows a simpliﬁed schematic of these
processes that includes the main steps for diamond deposition: 1) hydrogen abstraction
from gas-phase CH4 by atomic hydrogen to form methyl radicals, 2) reaction of these
radicals with the substrate, and 3) propagation of diamond growth by abstraction of a
surface hydrogen and reaction with another methyl radical.53'55’57’62 The presence of a
high concentration of atomic hydrogen at the surface is important because it stabilizes the
diamond lattice during growth, maintains the presence of carbon radical reactive species
by hydrogen abstraction, reduces the number of defects in the diamond lattice, and
suppresses the formation of graphite on the substrate surface.53’54’56’57’60 More detailed
information including the distribution of reactive species in the plasma, the effect of
temperature and pressure on diamond growth, and the effect of diamond crystallographic

. . 7,53,55,59-62
orientation on growth rate can be found elsewhere.3

23

Microwave Energy

 

H2 + CH4

(1) 'H + CH4 -> H2 + CH,
'H 1 CH3

'CH3 'H
°H 'CH3 ,H °CH3 °H
-------------------------------------------------------------- D.fﬁl '
.CH3 TI H H 111i 1 81011

(3)

(2)1 lCH_,+ 'H -) (13H,+ ’CH3 -> Cﬂz—CH3

 

 

Substrate

 

 

Figure 1.8 Simpliﬁed CVD scheme showing plasma composition, major reactive
components, and the primary reaction pathway for microcrystalline diamond deposition.
(Adapted from ref 62)

1.6.2 Nanocrystalline Diamond Growth Mechanism

Deposition of nanocrystalline diamond thin ﬁlms is a relatively new ﬁeld of CVD
technology, developed throughout the mid-19905. In contrast to the H2-rich atmosphere
in which microcrystalline diamond is grown, nanocrystalline diamond is produced in an
Ar/CH4 plasma with little or no added H2. The gas composition (i.e., amount of added
H2) used for nanocrystalline diamond deposition directly affects structural properties
including surface morphology, ﬁlm roughness, and grain size.“64 The discussion

presented here will consider data from both experimental and theoretical studies of the

24

simplest mechanistic case, nanocrystalline diamond deposition from an Ar/CH4
plasma.6M7 Unlike microcrystalline diamond deposition in which methane radicals are
the primary growth species, insertion of the carbon dimer radical (C2) into the growing
diamond lattice is thought to be the major mechanistic pathway for growth of
nanocrystalline ﬁlms?“7 C2 insertion into a diamond lattice is shown schematically in
Figure 1.9. In this scheme, hydrogen abstraction of surface C-H bonds is not required.
Rather, it is thought that growth proceeds through 1) C2 insertion into a CH surface site,
2) rotation and insertion of the other C atom into an adjacent C-H site, 3) insertion and 4)
rotation of a second C2 molecule into an nearby surface site, and 5) single bond formation
between adjacent C2 dimers within the carbon lattice.68 This process is, in fact,
energetically more favorable than the hydrogen abstraction reaction required for CH4/H2
plasmas. The activation energy was calculated to be ~5 kcal/mol for insertion of the C2
dimer into a diamond (110) lattice,‘55 while hydrogen abstraction requires ~25 kcal/mol.53
Details of nanocrystalline diamond growth, including the effect of source gas
composition on ﬁlm morphology and measurement of C2 plasma density, can be found

elsewhere.64'7'

25

H H
H \ / \ /H
C: .\ C_—'.._C

H H H H
\C—‘C/ '\c c/
\. /C=C\C\. 5 :C/CéC\

Figure 1.9 Proposed mechanism for nanocrystalline diamond growth by insertion of a
C2 dimer into the diamond lattice.68

I. 7 Research Goals and Objectives

Understanding the mechanisms by which the microstructure of current spz-bonded
carbon support materials degrade (i.e., changes in surface chemistry, mass, and
microstructure) is one of the many technological challenges in modern fuel cell research.
Morphological and microstructural degradation of carbon electrocatalyst supports leads
to reduced fuel cell efﬁciency, reduced electrode lifetime, and eventually, complete

electrode failure. There is still much to be learned in terms of how spz-bonded carbon

26

supports degrade and, in particular, the structural and chemical property changes that lead
to decreased fuel cell performance. The research presented in this dissertation was
focused on developing a better understanding of the condition-speciﬁc limitations, as
well as the structural factors, that lead to carbon degradation and corrosion in PEMFCS.
In addition, work on an advanced material was undertaken to create a more stable
support. The following research questions are addressed in the following chapters of this

dissertation.

Chagter 3: Study of the Degradation Mechanisms of spZ-Bonded Carbon
Powders During Electrochemical Polarization

1) Under what conditions (potential and temperature) do spz-bonded carbon
electrocatalyst support materials commonly used in PEMFCs undergo
microstructural degradation and corrosion?

- The morphological and microstructural degradation of structurally well-
characterized spz-bonded carbon powders (graphite and GC) were tested
under controlled potential and temperature using chronoamperometry.

- The carbon powders were studied by cyclic voltammetry, Raman
spectroscopy, scanning electron microscopy (SEM), and XRD before and
after potentiodynamic polarization to more fully understand the nature of

the degradation.

27

2) What inﬂuence do microstructural degradation and corrosion have on the
electrocatalyst activity and what inﬂuence does the electrocatalyst have on the
carbon support degradation?

- The morphological and microstructural degradation of a PEMFC-relevant
carbon black, Vulcan XC-72, with and without Pt, was investigated under
controlled potential and temperature conditions using chronoamperometry.

- The bare and Pt-loaded Vulcan XC-72 powders were also fully
characterized by cyclic voltammetry, Raman spectroscopy, and SEM
before and after potentiodynamic polarization to better understand the

inﬂuence of carbon structure and Pt on the degradation.

Chagter 4: The Electrochemical Performance of Diamond T kin-Film
Electrodes from Different Commercial Sources

1) Is electrically conducting diamond a viable alternative material for ﬁle] cell
electrocatalyst supports in terms of the commercial availability of the material?
- The physical, chemical, and electrochemical properties of two commercial

boron-doped diamond electrodes were compared with those of two

electrodes (microcrystalline and nanocrystalline) grown in our laboratory.

28

Chapter 5: Preparation and Characterization of Boron-Doped Diamond
Powder: A Possible Dim ensionally Stable Electrocatalyst Support Material

and

Chapter 6: Preparation and Characterization of Boron-Doped Diamond
Paper with Pt Nanoparticles

1) Can a high-surface—area advanced electrocatalyst support material be made based
on electrically conducting diamond?

- Microcrystalline conductive diamond powder and nanocrystalline

diamond-coated carbon paper were grown via microwave plasma-assisted

chemical vapor deposition.

2) What are the physical, chemical, and electrochemical properties and stability of
these new materials with respect to spz-bonded carbons?
- The new diamond materials were studied using cyclic voltammetry, SEM,
Raman spectroscopy, and XRD.
- The long-term physical and electrochemical properties of these new 3173-
bonded materials were compared to spZ-bonded carbon supports, under a

variety of controlled potential and temperature conditions.

3) How does the electrodeposition of Pt on the new sp3-bonded carbon paper
electrocatalyst support compare to electrodeposition on spz-bonded carbon paper
in terms of the electrochemical activity of the catalyst as well as particle density

and size?

29

- Pt was electrodeposited onto diamond-coated carbon paper and bare
carbon paper using pulsed galvanostatic deposition.
- The physical and electrochemical properties of the metallized electrodes

were evaluated using cyclic voltammetry, SEM, and Raman spectroscopy.

4) Are the composite sp2- and sp3-bonded paper electrodes stable during
potentiostatic polarization?
- The Pt/diamond and Pt/carbon paper electrodes were subject to
polarization at 1.6 and 1.8 V vs. SHE.
- The physical and electrochemical properties of the electrodes were then

studied using cyclic voltammetry, SEM, and Raman spectroscopy.

1.8 Dissertation Outline

This dissertation describes studies of the degradation and corrosion of spZ-bonded
carbon support materials under controlled potential and temperature conditions, as well as
the development of a new material based on electrically conducting diamond that may
serve as an advanced electrocatalyst support material in fuel cells. In Chapter 2, the
experimental parameters of the work in this dissertation including materials,
instrumentation, and methods are presented. Evaluation of the morphological and
microstructural stability of two model carbon materials, graphite and GC, with that of a
high-surface-area carbon black, with and without Pt electrocatalyst, is presented Chapter
3. Chapter 4 establishes that high-quality boron-doped diamond electrodes are

commercially available with reproducible properties. In this chapter, the physical,

3O

chemical, and electrochemical properties of two commercial microcrystalline boron-
doped diamond thin ﬁlms, and microcrystalline and nanocrystalline boron-doped
diamond thin ﬁlm from our laboratory are compared. The fabrication and physical and
electrochemical characterization of a boron—doped diamond powder is described in
Chapter 5. The stability of this new material is compared to that of graphite and GC
powders under controlled potential. Chapter 6 presents work on the fabrication of
diamond-coated carbon paper to serve as a possible electrocatalyst support material. In
this chapter, the electrodeposition of Pt on bare and diamond-coated carbon paper is
described. The morphological and microstructural stability of this new diamond material
is compared with that of uncoated carbon paper, with and without Pt. Finally, future

directions for this research are proposed in Chapter 7.

31

12$?

(1)

(2)
(3)
(4)

(5)

(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)

(14)

(15)

(16)

.lbefsznncnas

Energy Information Administration, US. Department of Energy: Washington,
DC, 2005.

Service, R. F. Science 2004, 305, 962-963.
Service, R. F. Science 2004, 305, 958-961.
Hoffert, M. 1.; Covey, C. Nature 1992, 360, 573-576.

Report of the Basic Energy Sciences Workshop on Hydrogen Production, Storage,
and Use, Argonne National Laboratory: Argonne, IL, 2003.

Costamagna, P.; Srinivasan, S. J. Power Sources 2001, 102, 253-269.
Costamagna, P.; Srinivasan, S. J. Power Sources 2001, 102, 242-252.

Carrette, L.; Friedrich, K. A.; Stimming, U. Fuel Cells 2001, 1, 5-39.

Appleby, A. J. Corrosion (Houston) 1987, 43, 398-408.

Auer, E.; Freund, A.; Pietsch, J .; Tacke, T. Appl. Catal. A., 1998, I 73, 259-271.
Antolini, E. J. Appl. Electrochem. 2004, 34, 563-576.

Steele, B. C. H. J. Mater. Sci. 2001, 36, l053-1068.

Steele, B. C. H.; Heinzel, A. Nature 2001, 414, 345-352.

Mathias, M. F.; Makharia, R.; Gasteiger, H. A.; Conley, J. J .; Fuller, T. J .;
Gittleman, C. J .; Kocha, S. S.; Miller, D. P.; Mittelsteadt, C. K.; Xie, T.; Yan, S.
G.; Yu, P. T. Interface 2005, 14, 24-35.

Hottinen, T.; Noponen, M.; Mennola, T.; Himanen, 0.; Mikkola, M.; Lund, P. J.
Appl. Electrochem. 2003, 33, 265-271.

Reiser, C. A.; Bregoli, L.; Patterson, T. W.; Yi, J. S.; Yang, J. D.; Perry, M. L.;
J arvi, T. D. Electrochem. Solid-State Lett. 2005, 8, A273-A276.

32

(17)

(18)

(19)

(20)

(21)

(22)

(23)

(24)
(25)
(26)
(27)
(28)

(29)

(30)

(31)

(32)

(33)

Kangasniemi, K. H.; Condit, D. A.; J arvi, T. D. J. Electrochem. Soc. 2004, 151,
E125-E132.

Paik, C. H.; Jarvi, T. D.; O'Grady, W. E. Electrochem. Solid-State Lett. 2004, 7,
A82-A84.

Roen, L. M.; Paik, C. H.; J arvi, T. D. Electrochem. Solid-State Lett. 2004, 7, A19-
A22.

Antolini, E.; Passos, R. R.; Ticianelli, E. A. J. Power Sources 2002, 109, 477-482.

Antonucci, P. L.; Passalacqua, E.; Pino, L.; Giordano, N., Proceedings of the 2nd
Int. Conf. on Engineering Materials, Bologna-Modena, Italy 1988; 527-535.

Antonucci, P. L.; Romeo, F.; Minutoli, M.; Alderucci, E.; Giordano, N. Carbon
1988, 26, 197-203.

Donnet, J .-B.; Voet, A. Carbon Black: Physics, Chemistry, and Elastomer
Reinforcement; Marcel Dekker: New York, 1976.

Kinoshita, K.; Bett, J. Carbon 1973, 11, 237-247.

Stonehart, P. Carbon 1984, 22, 423-431.

Voet, A.; Donnet, J .-B.; Marsh, P. A.; Schultz, J. Carbon 1966, 4, 155-157.
Park, S.-M. J. Electrochem. Soc. 1984, 131, 363C-373C.

Stevens, D. A.; Dahn, J. R. Carbon 2005, 43, 179-188.

Maja, M.; Orecchia, C.; Strano, M.; Tosco, P.; Vanni, M. Electrochim. Acta 2000,
46, 423-432.

Stevens, D. A.; Dahn, J. R., Orlando, FL 2003; Proceedings of The
Electrochemical Society, 204th Meeting, The Electrochemical Society, Inc.

Kinoshita, K. Carbon: Electrochemical and Physicochemical Properties; John
Wiley & Sons: New York, 1988.

Jenkins, G. M.; Kawamura, K. Nature 1971, 231, 175-176.

Harris, P. J. F. Philos. Mag. 2004, 84, 3159-3167.

33

(34)

(35)

(36)

(37)

(38)

(39)

(40)

(41)

(42)

(43)

(44)

(45)

(46)

(47)

Ungar, T.; Gubicza, J .; Ribarik, G.; Pantea, C.; Zerda, T. W. Carbon 2002, 40,
929-937.

McCreery, R. L. In Electroanalytical Chemistry; Bard, A. J ., Ed.; Dekker: New
York, 1991; Vol. 17, pp 221-374.

Reed, M. W.; Brodd, R. J. Carbon 1965, 3, 241-246.
Angus, J. C.; Hayman, C. C. Science 1988, 241, 913-921.

Moustakas, T. D. In Synthetic Diamond: Emerging Science and Technology;
Spear, K. E., Dismukes, J. P., Eds.; John Wiley and Sons: New York, 1994, pp
145-192.

Xu, J .; Granger, M. C.; Chen, Q.; Strojek, J. W.; Lister, T. E.; Swain, G. M. Anal.
Chem. 1997, 69, 591A-597A.

Granger, M. C.; Witek, M.; Xu, J .; Wang, J .; Hupert, M.; Hanks, A.; Koppang, M.
D.; Butler, J. E.; Lucazeau, G.; Mermoux, M.; Strojek, J. W.; Swain, G. M. Anal.
Chem. 2000, 72, 3793-3804.

Show, Y.; Witek, M. A.; Sonthalia, P.; Swain, G. M. Chem. Mater. 2003, 15, 879-
888.

Wang, J .; Swain, G. M.; Mermoux, M.; Lucazeau, G.; Zak, J .; Strojek, J. W. New
Diam. Front. C. Tech. 1999, 9, 317-343.

Witek, M.; Wang, J .; Stotter, J .; Hupert, M.; Haymond, S.; Sonthalia, P.; Swain,
G. M.; Zak, J .; Chen, Q.; Gruen, D. M.; Butler, J. E.; Kobashi, K.; Tachibana, T.
J. Wide Bandgap Mater. 2001, 8, 171-189.

Swain, G. M. J. Electrochem. Soc. 1994, 141, 3382-3393.

Goss, C. A.; Brumﬁeld, J. C.; Irene, E. A.; Murray, R. W. Anal. Chem. 1993, 65,
1378-1389.

Bowling, R.; Packard, R. T.; McCreery, R. L. Langmuir 1989, 5, 683-688.

Cruickshank, B. J .; Gewirth, A. A.; Mohr Reynolds, R.; Alkire, R. C. J.
Electrochem. Soc. 1992, 139, 2829-2832.

34

(48)

(49)

(50)

(51)
(52)

(53)

(54)

(55)

(56)

(57)

(58)

(59)

(60)

(61)

(62)

Chen, Q.; Granger, M. C.; Lister, T. E.; Swain, G. M. J. Electrochem. Soc. 1997,
144, 3806-3812.

Bennett, J. A.; Show, Y.; Wang, S.; Swain, G. M. J. Electrochem. Soc. 2005, 152,
E184-E192.

Wang, J .; Swain, G. M.; Tachibana, T.; Kobashi, K. J. New Mat. Electrochem.
Systems 2000, 3, 75-82.

Wang, J .; Swain, G. M. J. Electrochem. Soc. 2003, 150, E24-E32.
Wang, J .; Swain, G. M. Electrochem. Solid-State Lett. 2002, 5, E4-E7.

Asmussen, J .; Reinhard, D. K., Eds. Diamond Films Handbook; Marcel Dekker:
New York, 2002.

Frenklach, M.; Speak, K. E. J. Mater. Res. 1988, 3, 133-140.

Angus, J. C.; Argoitia, A.; Gat, R.; Zhidan, L.; Sunkara, M.; Wang, L.; Wang, Y.
Philos. Topics 1994, 342, 195-208.

Kobashi, K.; Nishimura, K.; Kawate, Y.; Horiuchi, T. Phys. Rev. B 1988, 38,
4067-4084.

Wang, Y.; Evans, E. A.; Kovach, C. S.; Landau, U.; Angus, J. C. Mat. Res. Soc.
Symp. Proc. 1995, 363, 127-138.

Kovach, C.; Zeatoun, L.; Roozbehani, B.; Greber, 1.; Angus, J. C. In Advances in
New Diamond Science and Technology; Saito, S., Fujimori, N., Fukunaga, 0.,
Kamo, M., Kobashi, K., Yoshikawa, M., Eds.; MYU: Tokyo, 1994, pp 93-96.

Yarbough, W. A.; Messier, R. Science 1990, 247, 688-696.

Angus, J. G; Evans, E. A.; Wang, Y. In Adv. in New Diam. Sci. T ech.; Saito, S.,
Fujimori, N., Fukunaga, 0., Kamo, M., Kobashi, K., Yoshikawa, M., Eds.; MYU:
Tokyo, 1994, pp 141-144.

Lee, S.-T.; Lin, 2.; Jiang, x. Mat. Sci. Eng. 1999,25, 123-154.

Butler, J. E.; Woodin, R. L. Phil. Trans. R. Soc. Land. A 1993, 342, 209-224.

35

(63)

(64)

(65)

(66)

(67)

(68)

(69)

(70)

(71)

Zhou, D.; Gruen, D. M.; Qin, L. C.; McCauley, T. G.; Krauss, A. R. J. Appl.
Phys. 1998, 84, 1981-1989.

Jiao, S.; Surnant, A.; Kirk, M. A.; Gruen, D. M.; Krauss, A. R.; Auciello, O. J.
Appl. Phys. 2001, 90, 118-122.

Redfem, P. C.; Homer, D. A.; Curtiss, L. A.; Gruen, D. M. J. Phys. Chem. 1996,
100, 11654-11663.

Hiramatsu, M.; Kato, K.; Lau, C. H.; F oord, J. S.; Hori, M. Diam. Rel. Mater.
2003, 12, 365-368.

Homer, D. A.; Curtiss, L. A.; Gruen, D. M. Chem. Phys. Lett. 1995, 233, 243-
248.

McCauley, T. G.; Gruen, D. M.; Krauss, A. R. Appl. Phys. Lett. 1998, 73, 1646-
1648.

Liu, Y. K.; Tzeng, Y.; Liu, C.; Tso, P.; Lin, 1. N. Diam. Rel. Mater. 2004, 13,
1859-1864.

Liu, Y. K.; Liu, C.; Chen, Y.; Tzeng, Y.; Tso, P.; Lin, 1. N. Diam. Rel. Mater.
2004, 13, 671-678.

Gruen, D. M.; Liu, S.; Krauss, A. R.; Luo, J.; Pan, X. Appl. Phys. Lett. 1994, 64,
1502-1504.

36

Chapter 2

Experimental Methods

2.1 Diamond Thin Film Growth

Conductive diamond thin ﬁlm was grown on all substrates by microwave plasma-
assisted chemical vapor deposition (CVD, 1.5 kW, ASTeX Inc., Lowell, MA). A block
diagram of a typical reactor system is in shown in Figure 2.1. Brieﬂy, the source gases
(e.g., Ar, H2, CH4, and B2H6) enter the chamber at controlled rates regulated by mass
ﬂow controllers. Microwave energy is focused into the quartz cavity and ignites a
plasma. The plasma, which is in close proximity to the substrate, contains the reactive
species important for growth. The chamber pressure is maintained by a throttle exhaust

valve and rotary pump that is operated continuously during ﬁlm growth.

 

Microwave
Generator

 

 

 

1
Mass Flow __J I I l...!
Controllers Antenna
‘ Quartz
W _ . . z - - Chamber
/

 

 

 

 

    

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

     

 

 

 

 

 

'7 Plasma. /
Substrate
Holder
(: a Gas Inlet "'9 3'
B '- "‘ Pressure
U 8 9’ ‘8 ‘ Gauge
'5 U 13 : Exhaust
275' g g 7; Valve
.. e 5‘ g

 

 

 

 

 

 

 

 

 

Figure 2.1 Schematic microwave plasma-assisted CVD reactor used to grow
microcrystalline and nanocrystalline boron-doped diamond thin-ﬁlm electrodes.1

37

Microcrystalline and nanocrystalline diamond thin ﬁlms were grown on p—type Si
(111) (0.05 cm thick x 1 cm2 in area, ~10'3 Q-cm, Virginia Semiconductor Inc.,
Fredricksburg, VA). The substrates were pretreated by mechanically scratching with 0-
0.1 um diameter diamond powder (GE Superabrasives, Worthington, OH) on a felt
polishing pad for 5 min. The scratched substrates were then washed sequentially with
ultrapure water, isopropyl alcohol (IPA), acetone, IPA, and ultrapure water to remove
polishing debris from the surface. Embedded diamond powder, as well as the defects
introduced by the scratching, served as the initial nucleation sites.

The speciﬁc deposition conditions used for the growth of each diamond type on
Si are listed in Table 2.1. All source gases were ultrahigh purity (99.999%). At the end
of the deposition period, the CH4 and B2H6 gas ﬂows were stopped and the ﬁlms
remained exposed to a H2 (microcrystalline) or an Ar/H2 (nanocrystalline) plasma for an
additional 10 min. The plasma power and pressure were slowly reduced over a 15 min
period to cool the samples in the presence of atomic hydrogen to a temperature below
400 °C. The plasma power was then turned off and the ﬁlms were cooled to room
temperature under a ﬂow of H2 (microcrystalline) or Ar/H2 (nanocrystalline). Typical
diamond ﬁlm thickness, resistivity, carrier concentration, and carrier mobility are listed in
Table 2.2.

Boron-doped microcrystalline diamond thin ﬁlm, deposited on conducting Si, was
also purchased from CONDIAS Gmbh (DIACHEM®, 0.01 cm thick x 1 cm2, Germany)
and Surnitomo Electric Industries, Ltd. (0.5 cm thick x 7.3 cm2 in area, #SHMF-16128-1,

Japan). The exact deposition conditions were not disclosed.

38

Table 2.1 Growth conditions for microcrystalline and nanocrystalline boron-doped
diamond thin ﬁlm deposited on Si.

 

 

 

Microcrystalline Nanocrystalline
Microwave Power 1000 W 800 W
Chamber Pressure 45 Torr 140 Torr
Gas Mixture CH4/H2/B2H6 Ar/H2/CH4/B2H6
Methane Concentration 0.5% CH4 1% CH4
Dopant 10 ppm B2H6 10 ppm B2H6
Total Gas Flow 200 sccm 100 sccm
Substrate Temperature ~800 oC ~800 °C
Growth Time 10 h 2 h
Table 2.2 Film thickness, resistivity, can‘ier concentration, and carrier mobility for

typical microcrystalline and nanocrystalline boron-doped diamond thin ﬁhn.“3

 

 

Film Resistivity Carrier Carrier
Type Thickness (ﬂ-c ) Concentration Mobility
(um) m (cm") (cmz/V-s)
Microcrystalline ~5-7 pm 0.01 — 0.1 ~102° 10-40
Nanocrystalline ~4 um 0.2 ~1020 90

 

The diamond powder (8-12 pm diam [~2 mz/g], GE Micron Products) was
prepared for deposition by sequential cleaning in warm aqua regia (30 min) and warm
30% hydrogen peroxide (30 min) to remove nondiamond spz-bonded carbon and metallic
impurities. The clean powder was then copiously rinsed With ultrapure water, IPA, and
acetone, ﬁltered, and dried. A portion of the clean powder (ca. 15 mg) was thinly spread
over the surface of an undoped Si (100) wafer (Virginia Semiconductor Inc.,

Fredricksburg, VA) and placed in the CVD reactor. The deposition conditions are listed

39

in Table 2.3. At the end of deposition period the CH4 and B2H5 gas ﬂows were stopped
and the powder remained exposed to a H2 plasma for an additional 10 min. The plasma
power and pressure were slowly reduced over a 15 min period to cool the samples (< 400
°C) in the presence of atomic hydrogen. The plasma power was then turned off, and the

powder was cooled to room temperature under a ﬂow of H2.

Table 2.3 Growth conditions for coating insulating diamond powder with a
conducting diamond overlayer.

 

Diamond Powder

 

Microwave Power 1000 W

Chamber Pressure 45 Torr
Gas Mixture CH4/H2/B2H6

Methane Concentration 1% CH4
Dopant 20 ppm B2H6

Total Gas Flow 200 sccm

Substrate Temperature ~800 °C
Growth Time 1,2,4, or 6 h

 

While this approach led to both full and partial coating of the powder particles, it
was not the most ideal because of the difﬁculty ensuring uniform plasma/ gas contact with
the entire particle surface. Additionally, the variable height of the powder particles in the
plasma led to temperature variations during ﬁlm growth, which caused differences in
overlayer thickness and microstructural quality. Particles that experienced higher
temperatures had more nondiamond carbon impurity content.

Carbon paper (Toray TGP-H, 0.19 mm thick, ElectroChem, Inc., Wobum, MA)

was cut into ca. 1x1 cm2 squares and seeded for growth by sonication in a slurry

40

containing 2 mg of 0-0.1 um diameter diamond powder (GE Superabrasives,
Worthington, OH) in 100 mL ethanol, for 10 min. The substrates were then dried and
placed in the CVD reactor. The deposition conditions used to coat the carbon paper are
presented in Table 2.4. At the end of deposition period the CH4 and B2H6 gas ﬂows were
stopped and the paper remained exposed to an Ar/H2 plasma for an additional 10 min.
The plasma power and pressure were slowly reduced over a 15 min period to cool the
samples in the presence of atomic hydrogen to a temperature below 400 °C. The plasma
power was then turned off, and the paper was cooled to room temperature under a ﬂow of
Ar/l-I2. After growth on one side, the substrate was removed from the CVD reactor,
reseeded by sonication in the diamond/ethanol slurry for an additional 10 min, and placed
back in the CVD reactor. Growth was then completed on the backside using the same
deposition conditions. The same cool-down procedure was used aﬁer the second growth.

Table 2.4 Growth conditions for coating carbon paper with boron-doped
nanocrystalline diamond.

 

 

Carbon Paper
Microwave Power 800 W
Chamber Pressure 140 Torr
Gas Mixture Ar/CH4/H2/B2H6
Methane Concentration 1% CH4
Dopant 10 ppm B2H6
Total Gas Flow 100 sccm
Substrate Temperature ~800 °C
Growth Time 2 h

 

41

2. 2 Material Characterization

A number of analytical methods were used to characterize the diamond properties,
including crystallinity (X-ray diffraction), microstructure (Raman), morphology
(scanning electron microscopy), and electrical conductivity (current—voltage curves).
These same methods were also employed to assess the chemical and physical changes of

the sp2- and sp3-bonded carbon samples before and after applied potential.

2. 2.1 Scanning Electron Microscopy (SEM)

Sample morphology was probed by SEM using a J EOL—6400F microscope (J EOL
Ltd., Tokyo, Japan) with a 20 kV accelerating voltage and a working distance of 15 mm.
The diamond thin ﬁlm on Si and the carbon paper samples were mounted on aluminum
subs with conductive carbon tape. Carbon powder samples (on glass or Si) were attached
to aluminum stubs with conductive carbon tape and coated with a thin layer of gold to

reduce surface charging effects and to improve the SEM image quality.

2. 2.2 X-Ray Diffraction Spectroscopy (XRD)
Crystallinity of diamond (powder and paper) was studied by XRD. Spectra were
obtained by scanning 20 from 20 to 100 on a Rikagu Rotaﬂex RTP300 RC instrument.

X-rays were produced at 1.540 A from a Cu anode.

2. 2.3 Ram an Spectroscopy

Microstructure of the carbons was evaluated using Raman spectroscopy. Spectra

were acquired at room temperature, using a RAMAN 2000 spectrograph (Chromex, Inc.,

42

Albuquerque, NM) consisting of a diode-pumped, frequency-doubled CW Nd:YAG laser
(500 mW at 532 nm, COHERENT), with a holographic grating (f/4, 600 grooves/mm),
and a therrnoelectrically cooled 1024x256 element charge-coupled device (CCD) detector
(ANDOR Tech., Ltd.). Spectra were collected with an incident power density of ca. 500
kW/cm2 (100 mW at the sample and 5 um diameter spot size) and a 10 5 integration time.
A white-light spectrum was collected under the same conditions and served as the
background spectrum to which the spectra were referenced. The Raman instrument was
calibrated using a high-pressure, high-temperature (HPHT), single crystal diamond

sample (ﬁrst-order phonon position = 1332 cm'l).

2. 2.4 Electrical Resistance

Four-point probe measurements were used to evaluate the resistivity of the planar
diamond thin ﬁlm. Electrical measurements were used to evaluate the resistance of the
conductive diamond powders and were made by placing a ﬁxed quantity of the powder
(ca. 20 mg) between a metal plate (bottom) and a metal rod (top), and measuring the
contact-to-contact resistance with an ohmmeter. The powder resistance was also
measured from i—V curves recorded between :t10 V. The current was measured in series
with the resistive powder as a function of the applied dc voltage, as shown in Figure 2.2.
The plate-to-plate distance with the powder sample was ca. 2 mm. A 200 g weight was
placed on the top metal rod contact to ensure that a constant force was applied during a
measurement. The short-circuit resistance between the two plates was measured ﬁrst and

was always near 0 Q.

43

 

 

 

 

 

 

Figure 2.2 Schematic of the apparatus used to measure i-V curves for the powder
samples. This same general scheme was used to measure the powder resistance with an
ohmmeter, simply by removing the power supply and ammeter.

2. 3 Electrochemical Characterization

Electrochemical measurements were at the core of the material characterization
performed on the electrodes. Cyclic voltammetry was used to 1) assess the
electrochemical reaction kinetics for several redox systems at diamond thin ﬁlms
deposited on Si, 2) probe the electrochemical properties of conductive diamond powder
and diamond-coated carbon paper, 3) evaluate the microstructural stability of diamond
and commercial carbon powders, and 4) study the electrochemical properties of Pt
nanoparticles electrodeposited on carbon paper and diamond-coated carbon paper. The
morphological and microstructural stability of several sp2- and sp3-bonded carbon
powders was studied at PEMFC-relevant potentials and temperatures using
chronoamperometry. Finally, pulsed galvanostatic electrodeposition of Pt was used to

deposit the electrocatalyst on bare and diamond-coated carbon paper. Basic theory and

44

operating principles of the electrochemical techniques used in this work can be found
elsewhere.4'7

All of the electrochemical measurements, except Pt deposition, were performed
using a CHI650a computerized potentiostat (CH Instruments Inc., Austin, TX) with a
graphite rod counter electrode and a Ag/AgCl reference electrode (3 M NaCl, E °' = -45
mV vs. SCE). All measurements were made at room temperature (25 :t 1 °C), unless
otherwise noted, and all solutions were deoxygenated with N2 for at least 20 min prior to
a measurement and blanketed with the gas during the measurement. Details of the
electrode preparation procedures and the electrochemical cells used for each electrode
morphology are provided below. This is followed with a description of the experimental
approaches for assessing electrode reaction kinetics, morphological and microstructural

stability, and electrocatalytic activity.

2. 3. I Electrode Preparation and Experimental Setup

2.3.1.1 Diamond T kin-Film Electrodes on Si

The diamond thin ﬁlm electrodes on Si (~0.2 cm2 area) were sealed against the
bottom of a single-compartment electrochemical cell with a Viton® o-ring, as shown in
Figure 2.3.3 The backside of the Si substrate was scratched, cleaned, and coated with
graphite in order to ensure good ohmic contact with a metal current collector plate. All

electrodes were used after a 20 min soak in distilled isopropyl alcohol.2’8

45

Auxiliary

R f
e erence Electrode

Electrode

 

  

Worklng \.\\ : ,-
Electrode ~ ~ //’

Base ——+ ““4

Figure 2.3 Design of the single compartment, glass, three-electrode electrochemical
cell. (Adapted from ref 3)

2.3.1.2 Carbon Powder Electrodes

Conductive diamond, graphite (Superior Graphite Co., Bedford Park, IL), glassy
carbon (GC) powder (SIGRADUR® G, HTW GmBH, Germany), and carbon black
(Vulcan XC-72, Electrochem, Inc, Woburn, MA), with and without Pt (20 wt. %), were
thoroughly mixed with a (10 wt. %) polytetraﬂuoroethylene (PTFE, 60 wt. % in H2O,
Aldrich) binder and a few mL of distilled isopropyl alcohol. Each electrode contained
~10 mg carbon powder. The diamond, graphite, and GC slurries were cast on a clean
glass slide using a pipet and dried overnight in air. The Vulcan XC-72 slurry was spread
over a microcrystalline diamond thin ﬁlm (0.5 % CH4 in H2, 10 ppm B2H6, 6 h
deposition). During the solvent evaporation, the slurry was smoothed and spread over the
diamond electrode with successive passes of a Teﬂon-coated razor blade. This was
necessary to limit agglomeration of the carbon black particles and produce a thin, crack-

free ﬁlm. In terms of preparation of powder electrode with a binder, the Vulcan powder

46

was the most challenging to work with, requiring extreme care during electrode
fabrication. After the initial solvent evaporation in air, all electrodes were heated in an
oven (air atmosphere) for 30 min at 280 °C and for 30 min at 350 °C. This ﬁnal heat
treatment was found to be necessary to mechanically strengthen the electrodes.

The powder electrodes, mounted on glass slides, were pressed against a Viton® o-
ring (~0.2 cm2 area) at the bottom of a single compartment glass electrochemical cell.
Electrical contact was made to the diamond, graphite, and GC powders by pressing
copper foil against the surface of the powder outside the o-ring. Electrical contact with
the Vulcan XC-72 powder was made through the microcrystalline diamond thin-ﬁlm
electrode. The electrode area in contact with the solution was estimated from the cyclic .

voltammetric background current according to Equation 2.1:

i
A = -——c" 2.1
Cal) ( )

where A is the electrode area in cmz, id, is the anodic current at 800 mV in uA, Cd: is the
capacitance of the electrode (~5 uF/cm2 for diamond, ~35 uF/cm2 for graphite, GC, and
Vulcan XC-72),9 and u is the scan rate in V/s. 800 mV was chosen because it was the
ﬂattest region of the background cyclic voltammetric i-E curve for all carbon powders
and the only region in which the measured current could be assumed to be all capacitive
in nature. Note: The background current for the microcrystalline boron-doped diamond
thin-ﬁlm electrode was not subtracted from that of the Vulcan XC-72 because it was ca. 3

orders lower in magnitude and did not signiﬁcantly contribute to the electrochemical

signal.

47

 

2.3.1.3 Carbon Paper Electrodes

Conductive epoxy (Chemtronics, Kennesaw, GA) was used to seal a Cu wire
(Aldrich) to the bare and diamond-coated carbon paper. The epoxy was cured by heating
at 80 °C for 20 min in an oven. The bare Cu wire was wrapped in Teﬂon tape (Aldrich)
and the entire wire and epoxy section was coated with nail polish, as shown in Figure 2.4.
For the solutions used in this work, the nail polish was stable and electrically insulated
this section from solution contact. The sealed electrodes were fully immersed in
electrolyte contained in a beaker for electrochemical measurements. Unless otherwise
noted, the electrode area in contact with solution was estimated from the anodic current at

0 mV using Equation 2.2, as described for the carbon powder electrodes above.

Cu wrreN

\

Sealed with

_ nail polish
Conductive epoxy

 

 

Fig“ re 2.4 Carbon paper electrode preparation schematic.

48

2. 3.2 Electrochemical Kinetics

The electrochemical kinetics for a number of redox systems on microcrystalline

and nanocrystalline boron-doped diamond thin-ﬁlm electrodes are described in Chapter

4. The apparent heterogeneous electron-transfer rate constant, kgpp, for Fe(CN)63‘/4’,

Ru(NH3)63+/2+, and IrC162'/3' was determined from cyclic voltammetric data (AEp-v
dependence) using the method developed by Nicholson.7 According to theory, AEP for an
electrochemical reaction is dependent on the scan rate (v, V/s), transfer coefﬁcient (a),
and heterogeneous electron-transfer rate constant (k0 , cm/s). The behavior of

quasireversible reactions can be described by the Equation 2.2:

(Dox )a/Zko

Dred
”I: (2:0 1%)” (2'2)

 

where Box and Dr 8 d are the diffusion coefﬁcients of the oxidized and reduced species
(cmz/s), respectively, f is the Faraday constant divided by the ideal gas constant and
temperature (F/RT), and 9” is a dimensionless parameter. Nicholson derived an
experimental relationship between '1’ and 21Ep such that k0 can be calculated from the
above equation at different scan rates. In this research, 21Ep varied over the scan rate

range tested and the apparent heterogeneous electron-transfer rate constant, kgpp, was

statistically determined from data at several scan rates. The rate constants are referred to

as apparent, because no correction for any electric double layer effects was made. kgpp

was additionally determined through digital simulation of the cyclic voltammetric i-E

curves, using DigiSim® 3.03 (Bioanalytical Systems, Inc., W. Lafayette, IN). In all cases,

49

the simulated 21Ep for a given kg”, value differed from the experimental value by 9 mV,

or less, at scan rates up to 50 V/s. The following diffusion coefﬁcients were used for the
redox couples, assuming Dox=Dredz Fe(CN)63'/4', (7.6 x 1045 cmz/s), Ru(NH3)63+/2+ (5.5 x
10*5 cmz/s), Irc162"3' (6.8 x 1045 cmZ/s), 4-MC (6.0 x 1045 cmZ/s), and Few+ (9.0 x 10*5

2. 3.3 Polarization Studies

Carbon powder and paper electrodes were subjected to applied potentials from 1.0
to 1.6 V vs. Ag/AgCl at 25, 50, and 80 °C in 0.1 M HClO4 for 1 h periods. The
measurements were performed in the Chronoamperometric mode with the current (hence,
charge) being recorded as a function of time. The single compartment glass
electrochemical cell was wrapped with heating tape powered by a Variac for temperature
control. The solution temperature was measured before and after the polarization and
remained constant (4: 5 °C). The reference electrode temperature was not directly
controlled and was likely between that of room temperature and solution temperature.
The recorded current-time proﬁles were integrated to determine the charge passed at a
given potential. For a faradaic process, the charge passed is related to the number of
moles electrolyzed according to Equation 2.3:

Q = nFN (2.3)
where Q is the integrated charge (Coulombs), n is the number of electrons involved per

reaction, F is the Faraday constant, and N is the number of moles electrolyzed.

50

2. 3.4 Pt Electrodeposition

Electrocatalytic Pt particles were deposited on diamond-coated and bare carbon
paper electrodes from a solution of 1 mM potassium hexachloroplatinate(IV) dissolved in
0.1 M HClO4. A pulsed galvanostatic method was employed for deposition, as discussed
in Chapter 6. A homemade galvanostat, constructed with an OMNT 90 potentiostat
(Cypress Systems, Inc., Lawrence, KS), was interfaced with a PCI-6034E controller and
data acquisition board (National Instruments, Austin, TX). LabVIEW software (version
6.0, National Instruments) was used to control the pulse generation, which was performed
at 1 Hz with a 25% duty cycle, resulting in an on-time of 200 ms and an off-time of 800
ms per pulse. Cyclic voltammograms were recorded before and alter Pt electrodeposition
to assess the presence of Pt on the electrode surface. The Pt surface area was determined
by integrating the cyclic voltammetric charge (cathodic cycle) between 0.15 to -0.2 V vs.
Ag/AgCl, which is associated with H adsorption. This calculation assumes a lelPt
surface interaction and a surface charge of 210 uC/cm2 for hydrogen adsorption on an

atomically ﬂat Pt(111) surface.“13

2. 4 Reagents

All chemicals were analytical-grade quality, or better, and were used without
additional puriﬁcation. Solutions of 0.1 mM potassium ferrocyanide (Aldrich),
potassium hexachloroiridate (IV) (Aldrich), hexaammineruthenium (III) chloride
(Aldrich) in 1 M potassium chloride (Spectrum), and 0.1 mM ferric sulfate (Matheson,
Coleman, and Bell) and 0.1 mM 4-methylcatechol (Aldrich) in 0.1 M perchloric acid

(redistilled 99.999%, Aldrich) were made. Solutions of l M potassium phosphate buffer

51

with 0.1 mM potassium ferrocyanide and 0.1 mM hexaarnmineruthenium (III) chloride at
pH values of 1.90 and 7.35 were prepared using the appropriate molar ratios of
phosphoric acid (Aldrich), sodium phosphate dibasic (Columbus Chemical Industries,
Inc.), and sodium phosphate (Columbus Chemical Industries, Inc.). The appropriate
amount of potassium hexachloroplatinate(IV) (98%, Aldrich) was dissolved in 0.1 M
HClO4 to make a 1 mM solution. All redox analyte solutions were prepared daily using
ultrapure water (>18 MQ-cm) from an E-pure water puriﬁcation system (Bamstead) in
glassware that was cleaned by a three-step process: ethanol/KOH bath, alconox/ultrapure

water solution, and an ultrapure water rinse.

52

jinS

(1)

(2)

(3)

(4)

(5)

(6)

(7)
(8)
(9)

(10)

(11)

(12)

(13)

lizyﬂErznntnas

Swain, G. M. In Electroanalytical Chemistry; Bard, A. J ., Rubinstein, I., Eds.;
Marcel Dekker, 2003; Vol. 22, pp 181-277.

Show, Y.; Witek, M. A.; Sonthalia, P.; Swain, G. M. Chem. Mater. 2003, 15, 879-
888.

Granger, M. C.; Witek, M.; Xu, J .; Wang, J .; Hupert, M.; Hanks, A.; Koppang, M.
D.; Butler, J. E.; Lucazeau, G.; Mermoux, M.; Strojek, J. W.; Swain, G. M. Anal.
Chem. 2000, 72, 3793-3804.

Bard, A. J .; Faulkner, L. R. Electrochemical Methods: Fundamentals and
Applications, Second ed.; John Wiley & Sons, Inc.: New York, 2001.

Wang, J. Analytical Electrochemistry, Second ed.; Wiley-VCH: New York, 2000.

Skoog, D. A.; Holler, J. F.; Nieman, T. A. Principles of Instrumental Analysis,
Fiﬁh ed.; Harcourt Brace College Publishers: Philadelphia, 1998.

Nicholson, R. S. Anal. Chem. 1965, 37, 1351-1355.
Ranganathan, S.; Kuo, T.-C.; McCreery, R. L. Anal. Chem. 1999, 71, 3574-3580.
Xu, J .; Chen, Q.; Swain, G. M. Anal. Chem. 1998, 70, 3146-3154.

Kovach, P. M.; Deakin, M. R.; Wightrnan, R. M. J. Phys. Chem. 1986, 90, 4612-
4617.

Gerhardt, G.; Adams, R. N. Anal. Chem. 1982, 54, 2618-2620.

Essalik, A.; Amouzegar, K.; Savadogo, O. J. Appl. Electrochem. 1995, 25, 404-
407.

Perez, J .; Tanaka, A. A.; Gonzalez, E. R.; Ticianelli, E. A. J. Electrochem. Soc.
1994, 141, 431-436.

53

Chapter 3

Study of the Degradation Mechanisms of spz-
Bonded Carbon Powders During
Electrochemical Polarization

Abstract

Two microstructurally well-characterized carbon powders, graphite and glassy
carbon (GC), were studied under potentiostatic polarization using conditions that mimic
those of a polymer electrolyte membrane fuel cell (0.8 to 1.6 V vs. Ag/AgCl and 25, 50,
and 80 °C). Electrochemical (cyclic voltammetry) and spectroscopic (Raman)
characterizations were completed before and aﬁer polarization to determine the extent of
carbon oxidation and microstructural degradation. Several important results may provide
insight on the degradation mechanisms of spz-bonded carbon electrocatalyst supports
(e. g., carbon blacks). First, there was evidence of oxidation of graphite and GC at low
potential (1.0 V) and temperature (25 °C), caused by oxidation of exposed edge plane.
The extent of surface oxidation was larger for GC, a material with a higher edge plane
density. Second, as the potential and temperature were raised, the edge plane density for
graphite heterogeneously increased across the electrode surface, likely because of
intercalation and fracturing of the graphite lattice. The heterogeneity of this corrosion is
indicative of a nucleation and grth process, where disordered regions and defects serve
as the active sites for electrochemical corrosion, while other, more structurally ordered
regions do not corrode. Third, increases in the graphite edge plane density correlated

well with changes in the surface carbon-oxygen content, suggesting that fracturing of the

54

graphite lattice, perhaps as a result of intercalation, exposed additional carbon edge plane.
Fourth, the charge passed during potentiostatic polarization, as well as the increase in
surface oxide coverage was always larger for GC than for graphite. Moreover, there were
no noticeable changes in the GC microstructure after polarization at most potentials and
temperatures. These results provide evidence for the so-called order/disorder mechanism
where structurally disordered carbons corrode more severely because of oxidation and
gasiﬁcation of the exposed edge plane. Preliminary results for the potentiostatic
polarization (0.8 to 1.4 V vs. Ag/AgCl at 25, 50, and 80 °C) of a carbon black commonly
used in fuel cells, Vulcan XC-72, show that the charge passed is not sensitive to cell
temperature, but changes as a ﬁrnction of the applied potential. Oxidation of Vulcan XC-
72 seemed to play a major role in the corrosion, with evidence for possible gasiﬁcation at
higher potentials (1.4 V). The charge recorded as a function of polarization potential and
temperature under similar conditions for a Pt-loaded (20 wt. %) Vulcan XC-72 showed
similar behavior. Signiﬁcant Pt surface area was lost (>30%) after polarization at 1.2 V
and all temperatures. This could possibly be due to corrosion of the carbon support and
subsequent loss of electrocatalyst, as well as electrocatalyst corrosion. Changes in the
cyclic voltammetric i-E curve, including a decrease in the current for Pt-oxide formation,
as well as for hydrogen adsorption/desorption on Pt were noted after polarization at 50
and 80 °C. The curves were also characterized by signiﬁcant electrode resistance after

1.2 and 1.4 V at 80 °C, likely because of severe support gasiﬁcation.

55

3.1 Introduction

The electrochemical corrosion of spZ-bonded carbon electrocatalyst supports (e. g.,
carbon black) in the phosphoric acid fuel cell environment (PAF C, 96% H3PO4, >100 °C)
is well studied."8 Although it was originally thought that corrosion of these same
materials was not a problem at the low operating temperature (~80 °C) and potential
(~O.7 V vs. SHE) of the PEMFC, it has recently been recognized that some of the
performance degradation of PEMFCs is due to corrosion of the spz-bonded carbon
electrocatalyst support when the cell is subject to potentials above the normal operating
voltage.9’l3 Kangasniemi and co-workers studied the electrochemical surface oxidation
of a common carbon electrocatalyst support in fuel cells, Vulcan XC-72, during
potentiostatic polarization from 0.8 to 1.2 V vs. SHE for 16, 60, and 120 h, without the Pt
electrocatalyst.10 Most of the experiments in this report were done at 25 °C, although
polarization carried out at 65 °C was also brieﬂy discussed.‘0 Using cyclic voltammetry,
X-ray photoelectron spectroscopy, and thermal gravimetric analysis coupled with on-line
mass spectrometry, they provided evidence for carbon oxidation at potentials 21.0 V vs.
SHE at 25 °C and 20.8 V vs. SHE at 65 °C.10 Roen and co-workers studied the inﬂuence
of Pt on carbon corrosion, utilizing cyclic voltammetric measurements with on-line mass
spectrometry.11 In this study, CO2 generation increased as the Pt mass fraction increased
(0, 10, and 39%).11 In addition, studies at 30, 50, and 70 °C showed a lower apparent
activation energy for gasiﬁcation of Vulcan XC-72 with Pt as compared to the bare

carbon.11

Dahn and co-workers have studied the thermal degradation of two carbon
blacks, Vulcan XC-72 and BP2000, with and without Pt, in dry and humidiﬁed air, and at

high temperatures (125 to 195 °C).];13 In these studies, it was shown that increasing

56

mass fraction of Pt, humidity, and higher carbon surface areas all increased the carbon
corrosion.12’l3

Although these reports have presented some useful data about the conditions that
cause support corrosion, they do not investigate the fundamental structural and
microstructural causes of carbon corrosion. The early PAF C corrosion work showed that
carbon black corrosion proceeds through a two-step process, which includes oxidation of
carbon edge plane and defect sites and gasiﬁcation of surface oxides to CO2.5 Corrosion
of the structurally disordered regions occurs ﬁrst, but the more graphitic regions are also
susceptible to corrosion during prolonged exposure harsh conditions.4 It is unclear if this
same process, the so-called order/disorder mechanism, occurs in the PEMFC. The extent
of degradation (i.e., surface oxidation or carbon gasiﬁcation) is also unclear. Elucidation
of the degradation mechanism and the effect that electrocatalyst support degradation has
on the activity of the electrocatalyst is an important task, not only to better understand the
conditions that cause support degradation, but also to enable the systematic development
an advanced material that will exhibit superior stability in a PEMFC.

An extensive amount of research has been published about the electrochemical
corrosion of spZ-bonded carbons such as highly-oriented pyrolytic graphite (HOPG) and
glassy carbon (GC). It would be advantageous to utilize this large body of literature and
draw parallels between the carbon electrochemical literature and the electrochemical
corrosion of carbon in PEMFCs to more fully elucidate the carbon degradation
mechanism. It is well known in the electrochemical community that the electrochemical

nature of spz-bonded carbons, including the electron-transfer rate and dimensional

stability, depends strongly on microstructure, surface roughness, and the presence of

57

physi- and chemisorbed species.”18 McCreery and co-workers systematically studied
the electrochemical reaction kinetics for common redox analytes on spZ-bonded carbons
of varying microstructure (i.e., differences in edge to basal plane ratio) and electrode
pretreatment (e. g., laser activation and polishing) to elucidate the relationship between
carbon type and electrochemical reactivity and stability.ls '17’19'21 Of speciﬁc importance
to the fuel cell community, two reports paired electrochemical methods with Raman
spectroscopic investigations to yield a wealth of information about the dependence of the
electron-transfer rate on the presence of carbon edge planew’21 Carbon electrodes with a
higher edge plane density (e. g., GC with respect to HOPG) exhibited orders of magnitude
higher heterogeneous electron-transfer rate constants for redox systems such as
Fe(CN)(,3 74221 McCreery and co-workers also showed that the electron-transfer rate for
electrodes that did not have a signiﬁcant edge plane density (e.g., HOPG) could be
increased by fracturing the graphite plane via electrochemical pretreatment (i.e., holding
the electrode at anodic potentials) or by laser activation.19 In this study, parallel Raman
spectroscopic investigations revealed that both laser activation and electrochemical
pretreatment at potentials >1.6 V vs. Ag/AgCl increased the edge plane density of
HOPG.19

Detailed studies of the HOPG degradation mechanism during electrochemical
pretreatment using Raman spectroscopy, atomic force microscopy, optical microscopy,
and scanning electron microscopy have shown that intercalation plays a signiﬁcant role in

damage of the spz-bonded carbon lattice.”22

From Raman spectroscopic analysis of
short (2 min) potential steps up to 2.0 V vs. the sodium saturated calomel electrode,

Alsmeyer and McCreery showed that (1) anion intercalation always preceded lattice

58

damage and (2) intercalation did not necessarily result in lattice damage.20 For example,
there was no increase in the edge plane density (i.e., lattice damage) for intercalation in 1
M NaClO4, while potential excursions in l M HClO4, HNO3, and H2SO4 resulted in

lattice damage.20

Damage of the graphite lattice is thought to occur because of strain
caused by the formation of oxides after intercalation.20 From this, a general mechanism
of corrosion and oxidation for graphitic carbons can be proposed, which is shown in
Figure 3.1. This mechanism includes: (i) anion intercalation between the graphite layers
which causes distortion in the graphene layer d002 spacing, (ii) exfoliation of the layers

creating new edge plane density, (iii) oxidation of new edge plane sites, and (iv)

gasiﬁcation of surface oxides to CO and CO2.

““‘ “ ﬁMLWﬁ-EHWKWW I AW‘Y:WPZ*;FV ‘ /

 

 

 

‘l-NY-Hz'.v— r. - w .- ,
-
- «psi-2‘5“:viﬂzrirtﬁitﬁﬂw Intercalation a—ssamwmw Exfoliation AtﬁkH'S-ﬁ'i'ﬂtliimﬁi‘ﬁy
Int CFC a] ant ,..~'.':-.’«$"11ﬁf'é'kﬁw‘ﬁ'mfclﬁr —+ _mamrem-siw ——’ Mf'ﬁtﬁﬁiwiftiﬁffmﬁ,
A»~&:~Nam:arme9' 4...... ".9-.- . . - . ..., 7 .11:;""....-.‘."'“;',"
Oxrdatron at
CO/COZ new edge plane
ﬁw ’ -. ,geuﬂhﬁms‘h‘iv '- ' /
M.,-www- GaSIﬁcatlon mim-
.mmcssmmzmmer “Qaggmgrygtﬁzmlﬁéiﬁv
,.aztzz::2ssiezraia35wney Macrame harm‘s-35:31:13?

Figure 3.1 Schematic of the oxidation of graphitic carbons including intercalation and
exfoliation of the graphite lattice, oxidation of new exposed edge plane, and gasiﬁcation
(corrosion) to CO and CO2.

This chapter presents a ﬁrst step towards elucidation of the mechanism of carbon
corrosion in PEMFCs. In this work, the electrochemical (cyclic voltammetry) and

spectroscopic (Raman) changes for two structurally well-characterized spZ-bonded carbon

59

powders, graphite (structurally well-ordered) and GC (structurally disordered), were
compared under different PEMFC-relevant potential and temperature conditions.
Comparisons are drawn from the well-known electrochemical and Raman spectroscopic
literature of spz-bonded carbons studied by McCreery and co-workers to propose a
degradation mechanism. Data were also collected for a structurally disordered spz-
bonded carbon commonly used in PEMFCs, Vulcan XC-72 (with and without Pt
electrocatalyst), under similar potential and temperature conditions. Preliminarily results

from these experiments will be discussed.

3.2 Results and Discussion

3. 2.1 Anodic Polarization of Graphite and Glassy Carbon (GC) Powders

Chronoamperometric i-t curves for graphite and GC powder electrodes recorded
in 0.1 M HClO4 during potentiostatic polarization at 1.0 and 1.6 V vs. Ag/AgCl (25 °C)
are shown in Figure 3.2. The curves are plotted against current (A) and log current (B) to
more clearly show the differences for the carbons at different polarization potentials. All
currents, unless otherwise noted, are normalized to the apparent electrode area in contact
with the solution, as described in Chapter 2. 1.0 and 1.6 V were selected as an initial
comparison because 1.0 V vs. Ag/AgCl is equal to ca. 1.2 V vs. SHE (the PEMFC E°)
and 1.6 V vs. Ag/AgCl represents an extreme potential value. The current density for all
curves increases sharply at short times due to charging of the electrical double layer
(nonfaradaic), followed by an initial rapid decrease. The current recorded for graphite
and GC at 1.0 V quickly reaches a relatively steady-state value of ca. 0.02 and 0.07

uA/cmz, respectively. In contrast, the current for graphite and GC at 1.6 V increases over

60

the 1 h period, aﬂer the initial decay of nonfaradaic current. Importantly, the current for
GC is signiﬁcantly higher during polarization at 1.6 V than that for graphite at the same
potential.

Current recorded in a Chronoamperometric experiment, in the absence of an
electroactive surface or solution species (i.e., nonfaradaic), is expected to decay with e't
behavior, quickly reaching a non-zero, steady-state value. The magnitude of this steady-
state current depends on experimental variables such as the electrode capacitance. Thus,
the increase in current for graphite and GC electrodes anodically polarized at 1.6 V
relative to the current recorded during polarization at 1.0 V may provide evidence for
faradaic electrochemical reactions occurring at the electrode surface. Surface
electrochemical reactions such as oxidation of the carbon edge plane, carbon gasiﬁcation
to CO2, and intercalation of solvent molecules are common in acidic solutions at these
potentials.2‘6‘”'2°‘23 However, it is also possible that the increase in current is a result of
delamination of the polytetraﬂuoroethylene binder which would expose additional
electrode area to the solution. In fact, it is difﬁcult to determine the nature of the current
in a chronoarnperometric experiment solely from the i-t proﬁle. Other electrochemical
(cyclic voltammetry) and spectroscopic (Raman) experiments are necessary in the case of
these carbon powders to conﬁrm surface oxidation and microstructural degradation,

respectively, and will be presented later.

61

450

 

   
  

 

 

 

 

 

 

 

    
  

 

 

 

 

    

 

 

 

A '
NE 375
0
2 300
:3.
V225
4..)
G
g 150
:3 .
U 75 Graphrte- 1.6V
!‘/_____——1
0 Graphite and GC - 1.0 V
0 500 1000 1500 2000 2500 3000 3500
Tlme (s)
A 3.0
NS 2.5 GC-1.6V
O 2-0 Graphite-1.6V
2 :5 . I
1 t
V 1'0i
4.;
g 0.5?
t: 0.0-
: .
U -0.5
CD -1.0
3 4.5% Gra bite-1.0V
_20 . r . r I l . r . r . r . 1
0 500 1000 1500 2000 2500 3000 3500

Time (s)

Figure 3.2 Amperometric i-t curves for graphite and GC powders during a 1 h
polarization at 1.0 and 1.6 V vs. Ag/AgCl in 0.1 M HClO4 at 25 °C. The top curve (A) is
a plot of current and the bottom curve (B) is a plot of the logarithm of the current.

62

Integration of the i-t curve yields the charge passed during polarization, which is a
sum of the charge to form the electric double layer, the charge for formation of surface
carbon-oxygen functionalities, and the charge due to gasiﬁcation of the carbon. Thus,
this value can provide a gauge of the relative reactivity of graphite and GC at the
different potentials. The charge recorded for a faradaic electrochemical reaction is
proportional to the number of moles electrolyzed according to Equation 3.1:

Q=nFN 3.1

where Q is the charge passed (Coulombs), n is the number of electrons transferred in the
reaction, F is the Faraday constant, and N is the number of moles electrolyzed. Table 3.1
presents the charge density passed during anodic polarization of the two powders at 1.0
and 1.6 V. The charge density at 1.0 V is 0.14 and 0.15 mC/cm2 for graphite and GC,
respectively. A signiﬁcant increase and difference in the charge density arises at 1.6 V,
as values of 460 and 880 mC/cm2 are seen for graphite and GC, respectively. The larger
charge for GC is likely a reﬂection of a greater degree of oxidative reactivity and is
expected due to the greater microstructural disorder. Many early works on the
electrochemical corrosion of carbon blacks in phosphoric acid at high temperatures (>100
°C) found a correlation between the d-spacing, or graphitic nature, of the carbon and the
extent of electrochemical oxidationm’6 In this so-called order/disorder mechanism, less-
ordered regions of the material oxidize and/or degrade ﬁrst, followed by reaction of the
more graphitic regions during prolonged exposure to harsh conditions.4 Considering the
higher proportion of exposed edge plane density for GC with respect to graphite, the

difference in charge passed provides evidence for the order/disorder mechanism.

63

Table 3.1 Charge passed during the l h potentiostatic polarization of graphite and
GC powders at 1.0 and 1.6 V vs. Ag/AgCl (25 °C).

 

 

Garb“ Type (v giggcr) Chlhﬁ/grfigiﬁy
Graphite 1.0 0.14
Glassy Carbon 1.0 0.15
Graphite 1 .6 460
Glassy Carbon 1.6 880

 

The charge density passed during a 1 h potentiostatic polarization in 0.1 M HClO4
for graphite and GC powder electrodes at 1.0, 1.2, 1.4, and 1.6 V (25, 50, and 80 °C) are
presented in Table 3.2 and 3.3, respectively. Potentiostatic polarization was also
completed at 0.8 V. However, no detectable changes in surface oxidation or
microstructure were observed so these data are omitted for clarity. A few important
trends can be seen in the charge density measurements. First, the charge passed during
polarization increases as the polarization potential is increased for both graphite and GC.
If oxidation and gasiﬁcation of the carbon surface is taking place, the extent of these
processes would be expected to grow at higher potentials. Second, the charge passed
increases at any given potential as the temperature is raised. This is important
considering that fuel cells operate at higher temperatures (~80 °C) for maximum
efﬁciency. Third, the charge passed for GC is always higher than that for graphite,
except for polarization at 1.6 V and 80 °C. The increased reactivity of GC compared to
that of graphite is again attributed to the higher edge plane density of GC, which are the
active sites for oxidation and subsequent structural degradation. GC electrodes

anodically polarized at 1.6 V at and 80 °C, as well as at higher potentials at all

64

temperatures, failed (i.e., began to break apart). The charge passed for graphite at
polarization potentials >1 .6 V continued to increase as the potential and temperature were
increased up to 2.2 V (the upper potential limit studied). We suppose that carbon
gasiﬁcation becomes the major degradation process at high potentials and temperatures
for GC, causing the electrode to fail. Support gasiﬁcation would be very detrimental to
fuel cell operating efﬁciency and lifetime. In contrast, the major process for the graphite
electrodes is probably anion intercalation and fracturing of the graphite lattice which
exposes additional surface area for oxidation. These processes are expected to increase
as the potential increases and likely cause the large increases in charge density passed at
1.6 V for both materials. Future work will focus on analyses of the gases evolved (e. g.,
CO and CO2) as a function of potential and temperature to more fully characterize the
carbon gasiﬁcation as well as in situ Raman spectroscopic investigations to more fully

characterize anion intercalation and lattice damage.

Table 3.2 Charge passed (mC/cmz) during the 1 h potentiostatic polarization of
graphite powder ﬁ'om 1.0 to 1.6 V vs. Ag/AgCl at 25, 50, and 80 °C.

 

 

Potential 0 o o
(Vvs. A g/ AgCl) 25 C 50 C 80 C
1.0 0.14 0.25 0.61

1.2 0.84 0.82 4.1

1.4 6.5 13 20

1.6 460 1300 4200

 

65

Table 3.3 Charge passed (mC/cmz) during the 1 h potentiostatic polarization of GC
powder from 1.0 to 1.6 V vs. Ag/AgCl at 25, 50, and 80 °C.

 

 

Potential 0 o o
(Vvs. A g/ AgCl) 25 C 50 C 80 C
1.0 0.15 0.32 0.82

1.2 1.1 1.1 4.2

1.4 9.1 15 40

1.6 880 2600 3000

 

3.2.2 Evaluation of the Extent of Surface Oxidation in Terms of
E lectroactive Oxide Formation

Cyclic voltammetry was used to assess the correlation between the charge passed
during anodic polarization and the extent of surface oxidation of the two powders.
Background cyclic voltammograms of spz-bonded carbons are characterized by faradaic
current arising from the electroactive quinone/hydroquinone surface functionality
terminating the edge plane and defect siteszms The magnitude of this current can be
used as a measure of the relative extent of surface oxidation of a carbon during anodic
polarization. Background cyclic voltammograms for graphite and GC powder in 0.1 M
HClO4 at 0.1 V/s before and after polarization at 1.0 V for 1 h are shown in Figure 3.3A
and B, respectively. The background cyclic voltammograms for both powder types
recorded before anodic polarization have a nonfaradaic background current of
approximately 2 to 3 uA/cmz. Faradaic current due to the electroactive
quinone/hydroquinone surface carbon-oxygen functionality is also present, as indicated
by the arrows, centered ca. 350 mV for both carbon types. Aﬂer polarization, the cyclic

voltammetric faradaic and nonfaradaic current increased for both carbons, with much

66

larger changes noted after polarization at higher potentials (e. g., 1.6 V) and temperatures.
Nonfaradaic, or capacitive, current in an electrochemical measurement is a result of
charging of the electric double layer and can be described by Equation 3.2:

it = ACv (32)
where ic is the capacitive background current (A), A is the electrode area (cmz), C is the
electrode capacitance (F/cmz), and v is the scan rate (V/s). Because the electrode
capacitance and scan rate do not change during a measurement, changes in the
nonfaradaic background cyclic voltammetric current noted here results from increases in
the electrode surface area exposed to solution. This increase could result from
microstructural changes (e. g., fracturing of the graphite basal plane to expose additional
edge plane density), corrosion (i.e., carbon gasiﬁcation), or delamination of the binder.
Microstructural changes to graphite and GC as a result of anodic polarization were
probed using Raman spectroscopy and will be discussed in the following section.

There are also increases in the faradaic current for oxidation and reduction of the
quinone/hydroquinone surface oxygen ﬁmctionality after anodic polarization at potentials
as low as 1.0 V and 25 °C. This provides evidence for the oxidation of the electrode
surface. In fact, comparison of the faradaic current density for this surface redox reaction
provides a measure of the extent of oxidation, speciﬁcally the electroactive oxygen
content, after polarization. Equation 3.3 was used to assess the increase in the surface

oxide content for graphite and GC after polarization:
A] : jf,after _ jf,before (3'3)

where AJ is the change in current density (uA/cmz), j f after is the faradaic current

density (uA/cmz) of the reduction peak aﬂer polarization , and j f before is the faradaic

67

current density (uA/cmz) of the reduction peak before polarization. As shown in Table
3.4 (graphite) and 3.5 (GC), AJ follows the same trend as the measured polarization
charge, increasing with the applied potential and temperature. There is also a large
increase in AJ for both graphite and GC at 1.6 V. The similar trends between the charge
passed and electrode oxidation provides evidence that at least some of the charge passed
during anodic polarization, even at the thermodynamic PEMFC operating potential, is
due to oxidation of the graphite and GC surfaces. Irnportantly, AJ for GC is always
higher than that for graphite, especially at 80 °C. Considering the order/disorder
mechanism proposed for carbon blacks and the relationships found between edge plane

4,19-21,26

density and electrochemical reactivity of carbon, the higher edge plane density of

GC leads to a signiﬁcantly higher proportion of surface oxides.

 

 

      

8
6 - —Before 1.0 V ' —Before 1.0 V
NA ~ ~—-Aﬁerl.0V ,’7\\ NA 67“"Aﬁ¢Tl-OY-’7“-\
E 4- a 4L - ........... -
O . o .
32 i2
v 0: v 0:
a . a -2
(D 0)
i:‘2 54
5-4 U -6
6 '8»
m 10

 

 

 

 

 

 

0.2 A 0.4 0.8 1.0

0.0 0.6
Potential (V vs. Ag/AgCl)

0.2 0.4

0 O 0.6 0.8
Potential (V vs. Ag/AgCl)

1.0

Figure 3.3 Background cyclic voltammetric i-E curves for (A) graphite and (B) GC in
0.1 M HClO4 at 0.1 V/s before and aﬂer polarization at 1 V vs. Ag/AgCl and 25 °C for 1
h.

68

Table 3.4 Change in current density (AJ , uA/cmz) for surface oxide reduction

before and after polarization on graphite powder as a function of the polarization
potential and temperature in 0.1 M HClO4 for 1 h.

 

Potential

 

(Vvs. A g/ AgCl) 25 C 50 C 80°C
1.0 0.38 0.53 1.8
1.2 0.95 1.5 4.7
1.4 3.9 9.0 14
1.6 89 230 460

 

Table 3.5 Change in current density (AJ , uA/cmz) for surface oxide reduction

before and after polarization on GC powder as a function of the polarization potential and
temperature in 0.1 M HClO4 for 1 h.

 

 

(v £92K; c1) 25 °C 50 °C 80 °c
1.0 0.67 0.76 9.0
1.2 2.2 3.5 10
1.4 11 25 62
1.6 590 1900 2100

 

3. 2.3 Evaluation of Microstructural Changes

Raman spectroscopy is a useful tool to characterize spz-bonded carbon materials
because the Raman spectral features are dependent on microstructural characteristics such
as La, Lc, and the interplanar d002 spacing.27 A number of reports have been published on
the Raman spectra of spZ-bonded carbons to determine the origin of the spectral
features.”31 The primary Raman line at 1582 cm'l, as shown in Figure 3.4 for graphite
and GC before anodic polarization, is a result of the E28 vibrational mode which consists

of the in-plane stretching of the carbon atoms within the graphite lattice. The origin of a

69

second ﬁrst-order mode that arises at 1360 cm'1 (the so-called disorder, or D band) in
carbons with La>1000 A was the subject of debate for some time.27 It is now generally
accepted that this mode is associated with edge plane density in carbons; the intensity of
the D band is inversely proportional to La.31 Small D band scattering intensity can be
seen in the Raman spectrum for graphite before anodic polarization in Figure 3.4, while
that for GC is much more intense because of the higher proportion of edge plane density.
McCreery and co-workers used the carbon D to E2g ratio (D/E2g) to track the edge
plane density and "disorder" (with respect to a graphite single crystal), correlating carbon
structural elements with the electrochemical behavior of carbon electrodes.19'2"26
Irnportantly, by studying various pretreatment procedures such as laser irradiation and
anodic polarization on planar HOPG electrodes, the relationship between D/E2g and

corrosion (i.e., damage of the graphite lattice) became more clear.'9’2'

For example,
evidence was presented suggesting that both oxidative (anodic polarization) and non-
oxidative (laser irradiation) treatment resulted in increases in D/E2g, caused by fracturing
of the graphite lattice.21 Although surface oxide formation was not required to note
differences in D/E28 (i.e., during laser irradiation), oxidation of HOPG caused by anodic
polarization caused strain on the carbon lattice and resulted in lattice damage, increasing
the amount of exposed edge plane.21 Important spatial information was also gained
which showed that anodization at potentials, such as 1.85 V, caused heterogeneous
changes in the edge plane density, while higher potentials led to a more uniform disorder
across the HOPG surface.19 This ﬁnding supports a nucleation and growth process of the
order/disorder mechanism for carbon corrosion at potentials such as 1.85 V, where

disordered or defect sites may serve as nuclei for fracturing of the graphite planes.19

70

 

Glassy Carbon

Intensity (a.u.)

 

 

Graphite

 

 

'--

| l l l I i A 4 J l l

1000 1200 1400 1600 1800 .2000

Raman Shift (cm'l)

Figure 3.4 Raman spectra for graphite and GC before anodic polarization showing
the 1328 mode (~1580 cm“) and D band (~1360 cm").

. L l

 

To characterize microstructural changes in carbon powder electrodes and gain a
better understanding of the degradation mechanism of carbons in the PEMFC, Raman
spectra were taken at 6 different locations on the graphite and GC powder electrode
surfaces before and after anodic polarization in 0.1 M HClO4 (1.0 to 1.6 V at 25, 50, and
80 °C). Although these data alone cannot verify microstructural degradation because of
the large standard deviation in the measurements, when considered with the other
characterization data presented in this chapter, they provide evidence that the carbon
microstructure is affected by polarization. Average D/E2g and standard deviation from
these measurements are presented in Table 3.6 (graphite) and Table 3.7 (GC). The ﬁrst
row in each column corresponds to electrodes that were subjected to 0.1 M HClO4 at each

respective temperature for 1 h, with no applied potential. These samples will be referred

71

to as the controls. Comparison of D/E2g on the control samples shows that the
temperature and acidic environment alone did not cause microstructural changes in either
graphite or GC.

A few trends can be seen in Table 3.6 for graphite powder electrodes. First, at 25
°C, D/E2g decreases slightly after polarization at all potentials with respect to that of the
control. The graphite powder used in these experiments was not pristine HOPG, which
has little to no defect density (i.e., no 1360 cm'1 scattering intensity). Rather, the Raman
spectrum of graphite powder before polarization, shown in Figure 3.4, has small
scattering intensity ~1360 cm'l. As discussed above, the order/disorder mechanism
predicts that disordered regions with a high fraction of edge plane density will be more
reactive towards electrochemical corrosion. Thus, the decrease in D/E2g is likely a result
of selective corrosion of disordered regions (i.e., regions with higher edge plane density),
which are more susceptible to microstructural degradation than the more well-ordered
regions. This same trend is seen for graphite powders subject to anodic polarization at
1.0 and 1.2 V and 50 °C. However, at 1.4 and 1.6 V and 50 0C, D/E2g increases,
suggesting that at these potentials and temperature, well-ordered portions of the graphite
plane are beginning to fracture, perhaps due to anion intercalation, exposing additional
edge plane density. It is interesting to note the correlation between the large increase in
surface oxygen content for graphite at 50 °C that occurs at 1.4 V (Table 3.4) and the
increase in D/E2g under these same conditions. This suggests that oxidation of the
graphite lattice induces strain which results in lattice damage, as reported by Alsmeyer
and McCreery.20 Table 3.4 also shows that the oxidation of graphite is higher at all

potentials during anodic polarization at 80 °C, than at 25 and 50 °C. This ﬁnding

72

correlates well with the noticeable increase in D/E2g as the potential is increased at 80 °C.
Irnportantly, the standard deviation for D/E2g was usually quite high, ranging from ca. 6
to over 100%. This large deviation suggests that the graphite lattice is damaged
heterogeneously across the electrode surface by a nucleation and growth process. This

ﬁnding agrees well with work on the anodic polarization of planar HOPG electrodes.l9

Table 3.6 Ratio (n=6) of the D/E2g Raman modes for graphite powder electrodes at
different polarization potentials and temperatures.

 

Potential

 

(V vs. A g/ AgCl) 25 °C 50 °C 80 °C
None 0.15 :t 0.03 0.14 :l: 0.03 0.15 i 0.01
1.0 0.11 :l: 0.04 0.14 :l: 0.05 0.2 d: 0.1
1.2 0.1 :l: 0.1 0.08 :l: 0.05 0.23 :l: 0.07
1.4 0.08 i 0.09 0.14 :t 0.02 0.25 :l: 0.02
1.6 0.11 i 0.06 0.24 d: 0.03 0.3 :l: 0.1

 

In contrast to the trends in the graphite Raman spectra with potential and
temperature, there is relatively little change in D/E2g for GC until the electrode material is
subject to potentiostatic polarization of 1.6 V at 50 and 80 °C. Because GC does not have
large, well-ordered regions of basal plane like graphite, anion intercalation and fracturing
of the lattice are not expected. Rather, the high proportion of exposed edge plane is
oxidized and gasiﬁed, without increases in the edge plane density that are seen for

graphite.

73

Table 3.7 Ratio (n=6) of the D/E2g Raman modes for GC powder electrodes at
different polarization potentials and temperatures.

 

 

(v hill/ma) 25 °C 50 °C 80 °C
Nmm 11101 10101 12101
L0 12101 13102 11101
L2 12101 11101 10101
L4 12102 12102 121007
16 12101 161009 16101

 

Several important results from the electrochemical (cyclic voltammetric) and
spectroscopic (Raman) data presented above may provide insight on the degradation
mechanisms of spZ-bonded carbon electrocatalyst supports (e.g., carbon blacks). First,
increases in the surface oxide content at low temperature (25 °C) and potential (1.0 V)
were caused by oxidation of exposed edge plane for both graphite and GC. Second, as
the potential and temperature were raised, there was evidence for increases in the exposed
edge plane density for graphite, likely because of intercalation and fracturing of the
graphite lattice. The heterogeneity of this corrosion is indicative of a nucleation and
growth process, where disordered regions and defects serve as active sites for
electrochemical corrosion, while other, more structurally ordered regions do not corrode.
Third, increases in the graphite edge plane density (i.e., increased D/E2g) correlated well
with changes in the surface carbon-oxide content, suggesting that fracturing of the
graphite lattice, perhaps as a result of intercalation, exposed additional carbon edge plane.
Fourth, the charge passed during potentiostatic polarization, as well as the increase in

surface oxide coverage was always larger for GC than for graphite, as long as the GC

74

electrode did not fail (e. g., 1.6 V at 80 °C). Moreover, there were no noticeable changes
in the GC microstructure at most potentials and temperatures. These results provide
evidence for the so-called order/disorder mechanism where structurally disordered
carbons corrode more severely because of oxidation and gasiﬁcation of the exposed edge
plane. Although this result is not surprising, it has important implications in terms of the
stability of current carbon electrocatalyst supports in fuel cells (e. g., Vulcan XC-72) and

the directions for development of an advanced support material.

3.3 Anodic Polarization of Carbon Black, Vulcan XC- 72

Potentiostatic polarization was also carried out on Vulcan XC-72, a high-surface-
area (~250 mz/g) carbon black commonly used in fuel cells, followed by subsequent
electrochemical characterization using cyclic voltammetry. These measurements were
done with both bare and Pt-loaded (20 wt. %) Vulcan XC-72 and may serve as a
preliminary comparison of the corrosion behavior of a structurally disordered carbon
material and the effect that the corrosion has on the Pt electrocatalyst. However, more
detailed experiments should be carried out to more fully evaluate the behavior of Vulcan
XC-72 exposed to potentiostatic polarization. All of the data presented in this section
were from single measurements such that statistical reproducibility of the data cannot be
evaluated.

The charge passed during potentiostatic polarization of bare Vulcan XC-72
powder from 0.8 to 1.4 V at 25, 50, and 80 °C is presented in Figure 3.5. Surprisingly,
the charge density passed for these samples was much lower than that for graphite and

GC powders. We expected the charge for Vulcan XC-72 to be higher than that for

75

graphite or GC because of a higher edge plane density and suppose that this disparity
may be a function of the manner in which the data were normalized. The surface area,
calculated from the background cyclic voltammetric current before polarization, was two
orders of magnitude higher for Vulcan XC-72 than for graphite or GC. For this
normalization method, it is difﬁcult to numerically compare the charge for carbons with
signiﬁcantly different surface areas, but the trends are still important. It is clear ﬁom
Figure 3.5 that the charge passed increases as the polarization potential is increased. It is
unclear at this point why the temperature does not have a signiﬁcant affect on the
recorded charge. It should be noted that, aside from polarization at 1.4 V, the charge
passed did increase slightly at each potential as the temperature was increased, although
the increase was not nearly as large as seen for graphite or GC. We suspect that the
electrodes exposed to potential polarization at 1.4 V at 50 and 80 °C may have begun to
fail due to extensive carbon gasiﬁcation or binder delamination under such harsh
conditions, as discussed for GC at potentials above 1.6 V. Data were not collected for
potentiostatic polarization at 1.6 V because the Vulcan XC-72 electrodes were not stable

at this potential.

76

2)

m
N
o

 

1.8 l -D- 25 °C
-v— 50 0C
—o— 80 °C

ty(mC/c
' 'S'SQETSZ

.O .9
ex 00
l

h

:- O/

.699
ONA

-

 

 

Charge Densi

L l A L

 

00.8 . 1.0 I . 1.2 1.4 A
Polarizat1on Potentral (V vs. Ag/AgCl)

Figure 3.5 Charge density vs. polarization potential for Vulcan XC-72 as a function
of the temperature.

Evaluation of Vulcan XC-72 electrode oxidation, presented in Figure 3.6, was
done by analysis of the change in the faradaic reduction current associated with surface
carbon-oxygen functional groups (AJ ), as discussed above for graphite and GC. There
is evidence for carbon oxidation at 0.8 V at all temperatures, while oxidation was not
noted for graphite or GC until 1.0 V. The extent of Vulcan XC-72 oxidation increases
signiﬁcantly for polarizations at 80 °C until 1.4 V, where it decreases. This deviation is
attributed to electrode failure as a result of carbon gasiﬁcation or binder delamination.
The electrochemical oxidation follows a more linear trend than the charge measurements,
suggesting that carbon oxidation plays a more signiﬁcant role in the corrosion current as
the potential is increased. Although there is not much of a difference in A] between 25

and 50 °C, AJ increases for polarization at 80 °C, especially for 1.0 and 1.2 V.

77

Kangasniemi and co-workers found evidence for surface oxidation of bare Vulcan XR-72
after potentiostatic polarization at 1.0 V vs. SHE and 25 °C (~0.8 V vs. Ag/AgCl), as well
as at 0.8 V vs. SHE and 65 °C, (~1.0 V vs. Ag/AgCl) for 16 h.10 However, the extent of
oxidation at the higher temperature was not quantiﬁed such that the extent of increase in

surface oxidation cannot be compared.

 

2-0 : --1:1— 25 0C
1.8 _- —v— 50 °C
1.6: —o— 80 °C 0
1.4 - / O
1.2 ' O

1.0
0.8 L
0.6
0.4
0.2
0.0
-02

' l

I I V V '

AJ (uA/cmz)

I 1 I I

 

 

1L

 

0.8 1.0 1.2 1.4

Polarization Potential (V vs. Ag/AgCl)

Figure 3.6 Change in faradaic current density for surface oxide reduction of Vulcan
XC-72 before and after polarization as a function of the polarization potential and
temperature in 0.1 M HClO4 for l h.

Carbon corrosion in a PEMFC environment has been well-established”l3 In
addition to the importance of understanding the corrosion mechanism so that an advanced
support can be developed, it is also important to understand the inﬂuence of carbon

corrosion on electrocatalyst activity. As a ﬁrst step towards this goal, Pt-loaded (20 wt.

78

%) Vulcan XC-72 electrodes were polarized ﬁom 0.8 to 1.4 V at 25, 50, and 80 °C.
Figure 3.7 is a representative background cyclic voltammetric i-E curve of a Pt-loaded
(20 wt. %) Vulcan XC-72 electrode in 0.1 M HClO4 at 0.005 V/s (cycle 10).
Voltammetric features consistent with those expected for polycrystalline Pt, including
anodic current from Pt-oxide formation between ca. 0.6 and 1.0 V, a cathodic Pt-oxide
reduction peak at 0.5 V, and anodic and cathodic current resulting from hydrogen
adsorption and desorption, respectively, on Pt between ca. 0.1 V and -0.2 V were
observed after ca. 3 cycles. These features became more well-deﬁned as the electrode
was cycled from -0.2 to 1.0 V vs. Ag/AgCl at 0.05 mV/s, however, the background cyclic
voltammetric i-E signal (i.e., peak shape and magnitude) for the Pt-loaded (20 wt. %)
Vulcan XC-72 electrodes was constant after ca. 10 cycles. The nonfaradaic background
charging current observed at all potentials is a combination of capacitive and
pseudocapacitive current arising from both the carbon and the Pt metal nanoparticles.
Because these signals cannot be easily or reliably separated, the surface area of carbon
exposed to solution cannot be determined from the voltarnmogram. However, the Pt
electrocatalyst surface area, to which the charge density data presented below was
normalized, can be calculated by integrating the cyclic voltammetric charge (anodic
cycle) between 0.15 to -0.2 V vs. Ag/AgCl, which is associated with H desorption, as
shown in the shaded region of Figure 3.7. This calculation assumes a lelPt surface
interaction and a surface charge of 210 uC/cm2 for hydrogen adsorption on an atomically
ﬂat Pt(111) surface.”33

The charge density passed during potentiostatic polarization, normalized to the Pt

electrocatalyst surface area, is plotted as a function of potential and temperature in Figure

79

3.8. Considering the difference in normalization procedures (i.e., normalization to carbon
surface area vs. normalization to Pt surface area), the charge density passed during
potentiostatic polarization cannot be directly (i.e., quantitatively) compared to that of the
bare carbon powder. However, the trend of increasing charge passed as the potential is
raised is similar to that of the bare Vulcan XC-72 powder. There also seems to be little
change in charge density with temperature, although the charge passed at 80 °C is higher
for polarization at 0.8, 1.0, and 1.2 V, than that at 25 or 50 °C. The difference in
temperature dependence for Pt-loaded Vulcan XC-72 (Figure 3.8) compared to bare
Vulcan XC-72 (Figure 3.5) may be a result of increased corrosion in the presence of Pt
that has been reported by others.ll Electrodes polarized at 1.4 V at 50 and 80 °C visibly

failed, presumably due to gasiﬁcation of the carbon support or binder delamination.

 

.O
O
p...
O
l

    
 

_O
O
O
LA
I

. /\QAnodic

(mA/cmz)

-0.005

rﬁ

-0.010

I

Current

-0.015

 

 

L l l l A L A l A I

-0.2 0.0 0.2 0.4 0.8 1.0

Potential (V vs. Ag/AgCl)

 

Figure 3.7 Background cyclic voltammetric i-E curve for Pt-loaded (20 wt. %)
Vulcan XC-72 in 0.1 M HClO4 at 0.005 V/s. Current is normalized to the estimated Pt
surface area.

80

 

 

 

A
N
E 25 - —o—25°c D
O o
\ —v-—50 C
LE) 20-- —0—80°C
v i g
3315 e
6.1-4 .
‘é’
é)10-
°) 5 " :/
OD . O/D
H 4:1,”;
C3 0_ a
:5:
U

1—

 

0.8 1.0 1.2 1.4

Polarization Potential (V vs. Ag/AgCl)

Figure 3.8 Charge density vs. polarization potential as a ﬁrnction of temperature for
Pt-loaded (20 wt. %) Vulcan XC-72 as a function of the temperature.

Background cyclic voltammetric i-E curves recorded before and after
potentiostatic polarization of Pt-loaded (20 wt. %) Vulcan XC-72 from 0.8 to 1.4 V at 25
°C recorded in 0.1 M HClO4 at 0.005 V/s are shown in Figure 3.9. Although there is a
slight downward shift of the current below ca. 0.2 V after polarization at 1.4 V,
presumably due to the presence of oxygen in the solution, there is little change in the
curve shape after polarization. At 50 and 80 °C, the Pt-oxide reduction peak and the
hydrogen adsorption/desorption peaks become less well-deﬁned at 1.2 and 1.4 V. In
addition, resistive behavior, as indicated by a sloping curve, is observed after
potentiostatic polarization at 80 °C, particularly at 1.2 and 1.4 V. To assess whether
potentiostatic polarization affected the electrocatalytic activity, the surface area of Pt was

compared before and after polarization. Reduction of Pt surface area in these

81

experiments could be due to corrosion of both carbon and Pt, but is a useful gauge of the

effect of potential and temperature on the electrocatalytic activity. Table 3.8 presents the

percentage of Pt surface area lost (% Lost) as a result of potentiostatic polarization

according to Equation 3.4

APt,before - APt,after
A

 

%LOS, =( )* 100 (3.4)

Pt,before

where A Pt,befor e is the area (cm2) of Pt before polarization and A Pt, after is the area

(cm2) of Pt after polarization, calculated from the charge of the hydrogen desorption
reaction as described above. The active Pt surface area decreased at all potentials and
temperatures, except for that at 0.8 V and 25 0C. It is likely that additional Pt surface area
was exposed during this 1 h polarization, possibly as a result of cleaning of the Pt surface

34

or because of particle roughening, as reported by others. The % Lost increased as a

function of both potential and temperature, except for potentiostatic polarization at 80 °C,
where it seems the Pt area was rather constant at 0.8 and 1.0 V. However, it should be
noted that, as mentioned above, the cyclic voltammetric i-E curves for electrodes exposed
to polarization at 80 °C exhibited resistive behavior aﬂer polarization (i.e., sloping curve),
such that determination of the Pt surface area was difﬁcult. In fact, a reliable

measurement could not be done for A Pt after on the 1.4 V, 80 °C sample. The decrease in

Pt surface area would have detrimental consequences for a fuel cell, especially at higher
potentials, such as 1.2 or 1.4 V, where a large amount (>30%) of the Pt surface area was
lost at all temperatures. Future work should include methods to more fully characterize
the causes and consequences of support corrosion with Pt, in addition to the studying the

loss of electrocatalyst surface area in more detail.

82

 

0.010

T

NA 0.005

     

 

 

0 0.000 -
E0005 -
V _
El -0.010 - Before
<1) _ - - - After 0.8 v
50015 - '°"' :2“ 1(2):],
0 _ er .
-0.020 _ _After 1.4 v

-0.2 0.0 0.2 0.4 0.6 0.8 1.0

Potential (V vs. Ag/AgCl)

Figure 3.9 Background cyclic voltammetric i-E curves for Pt-loaded (20 wt. %)
Vulcan XC-72 in 0.1 M HClO4 at 0.005 V/s before and after polarization from 0.8 to 1.4
V vs. Ag/AgCl at 25 °C. Current is normalized to the estimated Pt surface area before
polarization.

 

Table 3.8 Percent Pt lost (% Lost) as a result of potentiostatic polarization of Pt-

loaded (20 wt. %) Vulcan XC-72 in 0.1 M HClO4 at different polarization potentials and
temperatures.

 

 

(v v2 91;}: c1) 25 °C 50 °C 80 °c
0.8 -6.04 9.11 5.4
1.0 4.3 13.6 3.4
1.2 31.5 32.3 58.6
1.4 46.7 88.9 --

 

83

3.4 Conclusions

Potentiostatic polarization of two structurally well-characterized carbons, graphite
and GC provided evidence for an order/disorder mechanism of carbon
degradation/corrosion under PEMFC conditions. The extent of surface oxidation
increased for both materials as potential and temperature increased. Microstructural
changes evaluated by Raman spectroscopy indicated that intercalation followed by lattice
damage may be the prevalent mechanism for graphite corrosion, while oxidation of
existing edge plane density is the major process for degradation of GC. The corrosion
charge recorded during anodic polarization of a carbon black commonly used in fuel
cells, Vulcan XC-72, increased as the potential increased, but temperature had little
effect. Cyclic voltammetric analysis provided evidence for oxidation of Vulcan XC-72 at
0.8 V and 25 °C, the extent of which increased somewhat linearly as the potential and
temperature were raised. Initial results for the potentiostatic polarization of Pt-loaded (20
wt. %) Vulcan XC-72 showed electrocatalyst surface area was lost during at all potentials
and temperatures, except for 0.8 V and 25 0C. Background cyclic voltammograms after

polarization showed evidence for loss of Pt as well as for increased electrode resistance.

84

3L4!

(1)

(2)

(3)

(4)

(5)
(6)
(7)

(8)

(9)

(10)

(11)

(12)
(13)

(14)

(15)

IQeyﬂereamcuas

Antonucci, P. L.; Passalacqua, E.; Pino, L.; Giordano, N., Proceedings of the 2nd
Int. Conf. on Engineering Materials - Bologna-Modena, Italy 1988; 527-535.

Antonucci, P. L.; Romeo, P.; Minutoli, M.; Alderucci, E.; Giordano, N. Carbon
1988, 26, 197-203.

Appleby, A. J. Corrosion (Houston) 1987, 43, 398-408.

Donnet, J .-B.; Voet, A. Carbon Black: Physics, Chemistry, and Elastomer
Reinforcement; Marcel Dekker: New York, 1976.

Kinoshita, K.; Bett, J. Carbon 1973, 11, 237-247.
Stonehart, P. Carbon 1984, 22, 423-431.
Voet, A.; Donnet, J.-B.; Marsh, P. A.; Schultz, J. Carbon 1966, 4, 155-157.

Gruver, G. A.; Pascoe, R. F.; Kunz, H. R. J. Electrochem. Soc. 1980, 127, 1219-
1224.

Mathias, M. F.; Makharia, R.; Gasteiger, H. A.; Conley, J. J .; Fuller, T. J .;
Gittleman, C. J .; Kocha, S. S.; Miller, D. P.; Mittelsteadt, C. K.; Xie, T.; Yan, S.
G.; Yu, P. T. Interface 2005, 14, 24-35. '

Kangasniemi, K. H.; Condit, D. A.; Jarvi, T. D. J. Electrochem. Soc. 2004, 151,
E125-E132.

Roen, L. M.; Paik, C. H.; Jarvi, T. D. Electrochem. Solid-State Lett. 2004, 7, A19-
A22.

Stevens, D. A.; Dahn, J. R. Carbon 2005, 43, 179-188.

Stevens, D. A.; Hicks, M. T.; Haugen, G. M.; Dahn, J. R. J. Electrochem. Soc.
2005, 12, A2309-A2315.

McCreery, R. L. In Interfacial Electrochemistry; Wieckowski, A., Ed.; Marcel
Dekker: New York, 1999, pp 631-647.

McDermott, M. T.; McDermott, C. A.; McCreery, R. L. Anal. Chem. 1993, 65,
937-944.

85

(l6) Kneten, K. R.; McCreery, R. L. Anal. Chem. 1992, 64, 2518-2524.

(17) Kneten Cline, K.; McDermott, M. T.; McCreery, R. L. J. Phys. Chem. 1994, 98,
5314-5319.

(18) Kinoshita, K. Carbon: Electrochemical and Physicochemical Properties; John
Wiley & Sons: New York, 1988.

(19) Bowling, R. J .; Packard, R. T.; McCreery, R. L. J. Am. Chem. Soc. 1989, 111,
1217-1223.

(20) Alsmeyer, D. C.; McCreery, R. L. Anal. Chem. 1992, 64, 1528-1533.
(21) Bowling, R. J .; Packard, R. T.; McCreery, R. L. Langmuir 1989, 5, 683-688.

(22) Goss, C. A.; Brumﬁeld, J. C.; Irene, E. A.; Murray, R. W. Anal. Chem. 1993, 65,
1378-1389.

(23) Gewirth, A. A.; Bard, A. J. J. Phys. Chem. 1988, 92, 5563-5566.
(24) Fagan, D. T.; Hu, l.-F.; Kuwana, T. Anal. Chem. 1985, 57, 2759-2763.

(25) McCreery, R. L. In Neuromethods; Boulton, A., Baker, G., Adams, R. N., Eds;
Humana Press, Inc., 1995; Vol. 27, pp 1-26.

(26) Alsmeyer, Y. W.; McCreery, R. L. Langmuir 1991, 7, 23 70-2375.

(27) Wang, Y.; Alsmeyer, D. C.; McCreery, R. L. Chem. Mater. 1990, 2, 557-563.
(28) Dresselhaus, M. S.; Dresselhaus, G. Adv. Phys. 1981, 30, 139-326.

(29) Nemanich, R. J.; Sonin, S. A. Phys. Rev. B 1979, 20, 392-401.

(30) Nathan, M. 1.; Smith, J. E. J .; Tu, K. N. J. Appl. Phys. 1974, 45, 2370.

(31) Tunistra, F.; Koenig, J. L. Chem. Phys. 1970, 53, 1126-1130.

(32) Essalik, A.; Amouzegar, K.; Savadogo, O. J. Appl. Electrochem. 1995, 25 , 404-
407.

86

(33) Perez, J .; Tanaka, A. A.; Gonzalez, E. R.; Ticianelli, E. A. J. Electrochem. Soc.
1994, 141, 431-436.

(34) Wang, J .; Swain, G. M.; Tachibana, T.; Kobashi, K. J. New Mater. Elect. Syst.
2000, 3, 75-82.

87

Chapter 4

The Electrochemical Performance of Diamond
T hin-F ilm Electrodes from Different Commercial
Sources

Abstract

The electrochemical properties of two commercial (Condias and Sumitomo)
boron-doped diamond thin-ﬁlm electrodes were compared with two types of boron-doped
diamond thin ﬁlm deposited in our laboratory (microcrystalline and nanocrystalline).
Scanning electron microscopy and Raman spectroscopy were used to characterize the
electrode morphology and microstructure, respectively. Cyclic voltammetry was used to
study the electrochemical response, with ﬁve different redox systems serving as probes
(Fe(CN)63'/4', Ru(NH3)63+/2+, IrCl62'B', 4-methylcatechol, and Fey/2+). The response for
the different systems was quite reproducible from electrode type-to-type and from ﬁlm-
to-ﬁlm for a given type. For all ﬁve redox systems, the forward reaction peak current
varied linearly with the square root of the scan rate (vm), indicative of electrode-reaction

kinetics controlled by mass transport (semi-inﬁnite linear difﬁrsion) of the reactant. The

apparent heterogeneous electron transfer rate constant, kgpp, for all ﬁve redox systems

was determined from AEp-v experimental data, according to the method described by

l

Nicholson. The rate constant was also veriﬁed through digital simulation (DigiSim®

3.03) of the voltammetric i-E curves at different scan rates. Good ﬁts between the

experimental and simulated voltammograms were found for v up to 50 V/s.' k 3,, values

88

of 0.05 to 0.5 cm/s were observed for Fe(CN)63'/4', Ru(NH3)63+/2+, and IrCl62'l3' without

extensive electrode pretreatment (e.g., polishing). Lower kgpp values of 104 to 10’6 cm/s

were found for 4-methylcatechol and Fey/2+.

The voltammetric responses for
Fe(CN)63‘/4' and Ru(NH3)(,3+/2+ were also examined at all four electrode types at two
different solution pH values (1.90 and 7.35). The AEp, if", and if" values for the two
systems were relatively unaffected by solution pH because hydro gen-terminated diamond
surfaces contain few, if any, ionizable carbon-oxygen functionalities (e.g., carboxylic

acid, pKa ~4.5). This is in contrast to the typical behavior of oxygenated, spz-bonded

carbon electrodes, such as glassy carbon.

89

4.1 Introduction

Chemical vapor deposited (CVD) diamond electrodes have attracted considerable
interest in recent years due to their versatility and excellent electrochemical properties.”12
The material can be used in electroanalysis to provide low detection limits for analytes
with superb precision and stability, for high-current density electrolysis (1-10 A/cmz) in
aggressive solution environments without any microstructural or morphological
degradation, as a corrosion resistant electrocatalyst support, and as an optically
transparent electrode (OTE) for spectroelectrochenrical measurements.”17 The research
presented in this dissertation is focused on the development of a more stable and robust
electrocatalyst support material for fuel cells using high surface area, electrically
conducting diamond. The ﬁrst step toward development of this new material was to
establish that the diamond electrodes sold commercially have reproducible physical,
chemical, and electrochemical properties, similar to those of laboratory-grown ﬁlms.

At least four commercial units now market electrically conducting diamond thin-
ﬁhn electrodes. Because the speciﬁc substrate pretreatment and ﬁlm deposition
conditions are typically not disclosed, these products are presumably grown using
different methods and process variables. As discussed in Chapter 1, variables such as
substrate pretreatment prior to growth, source gas composition, grth temperature, and
chamber pressure can all affect the resultant ﬁlm properties, thereby affecting the
electrochemical behavior. Several factors inﬂuence the diamond thin-ﬁlm electrode
response (i.e., electrode reaction kinetics): (i) potential-dependent density of electronic
states, which is affected by the doping type, level, and distribution, (ii) surface chemistry,

(iii) morphology and microstructure, (iv) defect density, (v) nondiamond carbon impurity

90

content, and (vi) double layer structrrre.3""’7’9‘13 The extent to which any of these factors
inﬂuence the electrode response depends greatly on the reaction mechanism for the
particular redox system for both sp2- and sp3-bonded carbons.7’9" 3’18'24

In this chapter, a comparison of the physical, chemical, and electrochemical
properties of two commercially-available (Condias and Sumitomo) boron-doped diamond
thin-ﬁlm electrodes is made with two types of diamond produced in our laboratory
(microcrystalline, MSUMicro, and nanocrystalline, MSUNano). Electrode morphology
and microstructure were characterized using scanning electron microscopy (SEM) and
Raman spectroscopy. Cyclic voltammetry was used to study the electrochemical
response, with ﬁve different redox systems serving as probes: Fe(CN)63'/4', Ru(NH3)63+’2+,
IrC162'/3', 4-methylcatechol, and Fey/2+. Additionally, the voltammetric responses for
F e(CN)(,3 "4' and Ru(NH3)62+/3+ were recorded at different pH (1.90 and 7.35), in order to
demonstrate the absence of electrostatic double-layer effects. Such effects are commonly

observed for spz—bonded carbon electrodes, like glassy carbon, because of the presence of

ionizable surface carbon-oxygen functional groups (e. g., carboxylic acid functionalities).

4.2 Results and Discussion

4. 2.1 Material Characterization

Scanning electron micrographs for MSUMicro, Condias, Sumitomo, and
MSUNano boron-doped diamond thin-ﬁlm electrodes are presented in Figure 4.1A-D,
respectively. All ﬁlms are polycrystalline and uniformly coat the substrate with no
visible cracks, pinholes, or voids. The MSUMicro, Condias, and Sumitomo ﬁlms contain

well-faceted crystallites with a nominal diameter of ca. 1, 5 and 2 um, respectively. The

91

crystallites of the MSUMicro, Condias, and Sumitomo ﬁlms possess signiﬁcant
triangular (111) faceting with an abundance of multiply twinned crystals (e. g., decahedral
shapes with ﬁve-fold symmetry). Twins are a common defect in diamond polycrystalline
diamond, particularly ones with a more octahedral morphology, resulting when two or
more intergrown crystals form in a symmetric fashion. The Condias ﬁlm has a large
number of re-entrant grooves (i.e., comers) and stacking faults. The re-entrant corners
have a higher growth rate than other regions because chemisorbed grth precursors
(e.g., methyl radicals) in the grooves are more stable against H attack than moieties
attached to the more open, planar surface.25 The MSUNano ﬁlm, in contrast, possesses a
much smoother texture with feature sizes on the order of ~100 nm.9 Previously reported
TEM data revealed that these features consist of aggregates of randomly-oriented grains

9

of diamond with a nominal diameter of ca. 15 nm. The nanocrystalline ﬁlm is grown

under conditions of high rates of re-nucleation, which leads to the small grain size and
smooth texture.”32

Raman spectroscopy is an excellent tool for probing the microstructure of carbon
materials. For diamond, the Raman spectral features, including band positions, line
widths, and relative peak intensities, are quite sensitive to the bulk crystal structure,
boron-doping level, internal stress (intrinsic and thermal), defect density, and presence of
spz-bonded nondiamond carbon irnpurity.'8’33'43 As shown in Figure 4.2, spectral analysis
revealed that the ﬁrst-order phonon line for diamond, expected at 1332 cm", is most
intense, narrow, and symmetric for the Condias ﬁhn. In contrast, this line is down-

shifted to ca. 1304 cm], attenuated, broadened, and asymmetric, with an upward shift in

scattering intensity on the high-wavenumber side, for the Sumitomo and MSUMicro

92

ﬁlms. This so-called Fano shiﬁ is induced by quantum-mechanical interference between
the discrete phonon state and an electronic continuum that occurs particularly with heavy

18,39,41

boron doping. Two additional peaks emerge near 1200 and 500 cm'1 for the

Sumitomo and MSUMicro ﬁlms which are also consistent with heavy boron doping.18’38'
40 It is important to note that minimal scattering intensity is seen in the 1500-1600 cm"1
region, which is associated with the presence of nondiamond, spZ-bonded carbon.

Considering that the scattering efﬁciency for graphite is about 50 times that of diamond,

the nondiamond impurity content in these ﬁlms is low.36

 

Figure 4.1 Scanning electron micrographs of (A) MSUMicro, (B) Condias, (C)
Sumitomo, and (D) MSUNano diamond thin ﬁlms.

93

The Rarnan spectrum of the MSUNano ﬁlm is characterized by broad peaks at
1150, 1333, 1470, and 1550 cm], consistent with Raman spectra previously reported for
high-quality nanocrystalline diamond.9'35’37’4244 In nanocrystalline diamond, small
clusters of sp3-bonded diamond grains are connected by grain boundary carbon in an spz-
bonded conﬁguration.9’26’32 The one-phonon diamond line at 1332 cm'1 is obscured by a
large background signal and is considerably broader than for the microcrystalline ﬁlms,
140 versus 10 cm}. Two different explanations exist for the line-broadening in
nanocrystalline diamond ﬁlms. The ﬁrst, so-called conﬁnement model, states that the
smaller the domain size, the larger the range of phonon modes (with different q vectors
and energies) that are allowed to participate in the Raman process.37 In other words, the
line width results from the spread in phonon energy. The second attributes broadening to
phonon scattering by impurities and defects (i.e., grain boundaries)” The scattering
event shortens the lifetime of the phonons and thus broadens the Raman line.

The peak at 1150 cm'1 is often used as a signature for high-quality nanocrystalline
diamond.35 Prawer and coworkers, through the study of clean nanocrystalline diamond

particles (~5 nrn diam), have attributed this peak to a surface phonon mode of
diamond.42 On the other hand, Ferrari and Robertson assert that this peak is associated
with sp2 -bonded carbon, speciﬁcally transpolyactelylene segments at the grain
boundaries.43 Their assignment of spz- rather than sp3-bonded carbon, as has oﬁen been
proposed, is based on the observations that the peak position changes with excitation
energy, the peak intensity decreases with increasing excitation energy, and the peak is
always accompanied by another peak at ~1450 cm'l, which behaves similarly with

excitation energy.35 Recent work by Birrell and co-workers conﬁrmed this ﬁnding."’4 In

94

their study, the amount of hydrogen in a nanocrystalline diamond deposition plasma was
varied, which resulted in transitions from nanocrystalline to more microcrystalline
morphologies at high hydrogen content (~10-20%).44 These microstructural changes
correlated well with the presence of carbon-hydrogen bonds (i.e., amorphous carbon) in

the nanocrystalline diamond grain boundaries."’4

 

MSUNano

   
 

I I 1 I n I n l J

 

 

 

  
 

. E
Sumitomo -1 :
A 1208cm :
=7 \ : 13040111"1
«5 4/
v . 2. 1
b 1 . 1 . L 1 1 1 1 .l 1 . 1 L 1 M
o I
o; Condras
Ci <—1332cm1
8
C‘:
H

  

 

 

D
-———-—-4-—--

I
I r I

J g l 1 I 1 I

400 600 800 1000 1200 1400 1600 1800 2000

l L I

 

 

 

Raman Shift (cm'1

Figure 4.2 Raman spectra for MSUMicro, Condias, Sumitomo, and MSUNano
diamond thin ﬁhns.

95

4. 2.2 Electrochemical Characterization

Background cyclic voltammetric i-E curves can be very informative about the
boron-doped diamond ﬁlm quality (i.e., presence of nondiamond carbon impurity at the
surface) and electrical conductivity. Figure 4.3 shows cyclic voltammetric i-E curves for
MSUMicro, Condias, Sumitomo, and MSUNano, as well as that for glassy carbon (same
geometric area in each case) in 0.1 M HClO4. The voltammetric curves for the four
diamond types are mostly ﬂat and featureless with a low background current that results
from (i) a reduced pseudocapacitance because of the absence of redox active and or
ionizable surface carbon-oxygen functional groups and (ii) a reduced capacitance due to a
slightly lower internal charge carrier concentration. In contrast, the background current
for glassy carbon is approximately a factor of 10 higher with faradaic current evident due
to a redox couple (pseudocapacitance) centered at ca. 400 mV (acidic pH). This faradaic
current is associated with redox-active surface carbon-oxygen functional groups
terminating the edge plane and defect sites (e.g., quinone/hydroquinone).45’46 These
peaks are typically absent in the background cyclic voltammograms for high-quality,
hydrogen-terminated boron-doped diamond because the surface is devoid of such
functional groups. However, close inspection of the background cyclic voltammograms
for the Condias and Sumitomo ﬁlms shows very weak peaks centered at ca. 500 mV. At
this point, it is unclear if these peaks are associated with surface carbon-oxygen
functional groups.

Cyclic voltammetry was used to study the electrode response for Fe(CN)63'/4',

Ru(NH3)63+’2+, IrClbz'B', 4-MC, and Fey/2+. The effects of a number of variables

including surface morphology, substrate pretreatment, and surface termination on the

96

electrode response of spz-bonded carbon and sp3-bonded diamond for these analytes has
been studied extensively.7’9’l3 ’19’24 It is important to note that the extent to which any of
these factors inﬂuence the electrode response depends on the mechanistic aspects for the

Particular redox analyteoml 3.19-24

 

 

 

 

 

 

 

            

 

 

 

8
6 "' I»
P ’ ’ ’ ’ I ‘
4 - ’ , - ------------- ;
A 1- ’ ’ a I :
<2 2 - ...........
i I”, -— Jammw
a. O lT-ﬂrj_2'_f;ILILJLT-lrlI.‘..'.‘.:_?;:.‘.:.::_‘..'.:.'—-.a-IA:-:::’- 7
H QW ﬂ.-o' ’ I I l
a 2 1’ , """""
" "' I ’ v -
<1) . ,I __________ , , , ,
t: -4 —: -----
:5 1 ,’ ----- Glassy Carbon
I O
U '6 'i :’ —— MSUMicro
-81 x -------- Condias
I O
10 t ,’ ------- Sum1tomo
—12 1 I I L 4 I 1
-0. 4 -0. 2 0. O 0. 4 0. 8 1.0

Potential (mV) vs. Ag/AgCl

Figure 4.3 Cyclic voltammetric i-E curves for all four diamond types in 0.1 M HClO4
at 0.1 V/s (geometric area = 0.2 cmz).

Heterogeneous electron-transfer reactions can be broadly categorized as two
types. The ﬁrst, so-called outer-sphere reactions, proceed by simple mass transport of the
analyte to the electrode-solution interface, with the electrode serving solely as a source or

sink for electrons. The electrode kinetics for this type of reaction are relatively

97

insensitive to factors such as the surface chemistry and microstructure, but very sensitive
to the density of electronic states near the formal potential of the reaction. The second,
so-called inner-sphere reactions, includes those that occur via some speciﬁc interaction
with the electrode surface. For example, reactions that proceed through an adsorbed state
or involve some speciﬁc surface site would be categorized as inner-sphere
electrochemical reactions. The electrode kinetics for this type of reaction tend to be quite
sensitive to surface chemistry and microstructure as well as the density of electronic
states near the formal potential. On boron-doped diamond electrodes, Ru(NH3)63+’2+,
methyl viologen, Irle" 3 ', Ru(bpy)33+/2+, and chlorpromazine proceed by an outer-sphere
electron-transfer pathway with the electrode kinetics being relatively insensitive to the

7,9,13,47,48
d.

physicochemical properties of diamon Apparent heterogeneous electron-

transfer rate constants, kgpp, between 0.01 and 0.2 cm/s are commonly observed for

conducting polycrystalline ﬁlms (both microcrystalline or nanocrystalline) without
extensive pretreatment.7’9'l3’47’48 In contrast, extensive pretreatment is required to activate
spZ-bonded carbon and metal electrodes for charge-transfer.

Fe(CN)63'/4' proceeds through a more inner-sphere electron-transfer pathway on

boron-doped diamond, with the electrode kinetics being highly sensitive to the diamond

7,9,13,47-50

surface termination. Speciﬁcally, the reaction appears to proceed through a

speciﬁc surface site that is blocked by oxygen. kgpp values from 0.01 to 0.1 cm/s are

commonly observed for clean, hydrogen-terminated ﬁlms, but the rate constants decrease

7,9,l 3,47-50 F63+l2+

by over two orders of magnitude for oxygen-terminated ﬁlms. is also of

0

the inner-sphere type with kapp

being very sensitive to the surface carbon-oxygen

functionalities on spz-bonded carbon electrodes, speciﬁcally carbonyl functional

98

groups.5| In this system, carbonyl groups serve to lower the large energy banier for

3+/2+

reorganizing the solvation sphere around the Fe redox center following electron

transfer, apparently by forming a bridging complex with the reactant at the electrode

51

surface. These functional groups are absent on hydrogen-terminated diamond and

0

app is low, in the range of 10'4 to 10’6 crr1/s.7’9’13’47'48 The

presumably this is the reason k

more complicated organic systems, like dopamine, 4-MC, and 4-tert-butylcatechol, are

all of the inner-sphere type with kgpp from 104 to 10‘6 cm/s at hydrogen-terminated

diamond. It is thought that the slow kinetics for these polar aromatic analytes result from
weak adsorption on the diamond surfacezom’48 Strong adsorption of these compounds on
spZ-bonded carbon electrodes results from 7r-7r interactions between the molecule and the

electrode. Diamond electrodes contain little n-electron density and any that exists is

highly localized in grain boundaries. Recent work by our group has shown that AEp for
dopamine decreases with increasing levels of nondiamond spz-bonded carbon impurity.52
Further work is needed to fully understand the structure-function relationships for such
inner-sphere redox systems at diamond.

Cyclic voltammetric i-E curves for all ﬁve redox systems (0.1 and 1 V/s data) are
presented for all four diamond types in Figure 4.4A-E. It is evident that the
electrochemical response for all four diamond types is similar toward Fe(CN)63'/4',
Ru(NH3)63+’2+, IrCl62’/3', and 4MC, with the largest response variance seen for Fey/2+.
The electrode response, in terms of 21Ep and ip”, was very reproducible from ﬁlm type-to-
type and from ﬁhn-to-ﬁlrn for a give type. The response reproducibility for our
MSUMicro and MSUNano ﬁlms is routinely 2-4% from ﬁlrn-to-ﬁlm. Additionally, over

the years we have studied boron-doped diamond ﬁlms from various well-regarded

99

laboratories around the world (e. g., the US. Naval Research Laboratory and the National
Institute for Materials Science in Japan) and have found the response for these hydrogen-

terrninated ﬁlms is very similar to the response for the ﬁhns reported herein.

3.74-

9

In general, all four ﬁlm types exhibit relatively low AEp values for Fe(CN)6
Ru(NH3)63+/2+, IrC162'/3’, with larger values for 4-MC and Fey/2+. In all cases, the forward

l/2

reaction peak currents varied linearly with v , indicating that the electrode kinetics are

controlled by mass transfer (semi-inﬁnite linear difﬁrsion) of the reactant to the electrode

3-74-

9

surface. All four diamond thin-ﬁlm types exhibit active responses for Fe(CN)6
Ru(NH3)63+/2+ and IrCl62'B' without any conventional pretreatment. Time-consuming
pretreatment is almost always required to activate spz-bonded carbon and metal
electrodes for electron-transfer. Pretreatment oﬁen improves: the electrode kinetics for
redox analytes, but can cause signiﬁcant variability in electrode surface microstructure
and chemistry depending on the pretreatment procedure. These differences can lead to
signiﬁcant response variability for the inner-sphere systems.53 The larger 21Ep values for
4-MC and Fey/2+ are not attributable to an unclean surface, but rather, result from the

nature of the reaction mechanism at the non-polar, hydrogen-terminated diamond surface.

100

 

 

 

 

 

 

Current (uA)

r r 1:3993‘ “NW
ooédnooroeomeo o o o o o o
.,.,.,. . . .

l-|m.‘r-
u.n.n.v‘-nvw-“
-ﬂ.”.

1 I / o3.:r ------------- .
N 0.1V/S -6~ ‘ \t 0] v/S

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. -9l....--.-..11
-0.2 0.0 0.2 0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.l 0.2
Potential (V vs. Ag/AgCl) Potential (V vs. Ag/AgCl)
30 60
20’ 40’
10_ 20E D /\-1
o A 0'
A .10 "
3 -20 E.” 1 I‘ll/S
*" 6 ,1 I:
g 4 , . 0
U .(2) __ -' _, .. 5-- - -. V .2 1 , /- O -2111...
0.2 0.3 0.4 0.5 0.6 0.7 0.8 09 1.0 1.1 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0
P01311113] (V VS- Ag/AgCl) Potential (V vs. Ag/AgCl)
........ MSUNano A
_— MSUMicro é
Condias g
— - - Sumitomo o

 

 

0.1V/s
-04 0.2 00 0.2 0.4 0.6 08 1.0 1.2

Potential (V vs. Ag/AgCl)

l . l

 

 

Figure 4.4 Cyclic voltammetric i-E curves for all four diamond types in (A) 0.1 mM
Fe(CN),3"4', (B) 0.1 mM RugNH36 36) 3”“ and (C) 0.1 mM 1:016 "3' all in 1 M KCl, and (D)
4-methylcatechol and (E) Fe 1’2 in 20. l M HClO4. Curves for scan rates of 0.1 and 1 V/s
are shown (geometric area= 0. 2 cm 2).

101

A summary of the cyclic voltammetric data for all the diamond thin-ﬁlm types is

presented in Table 4.1. 21Ep and ipox/ip'ed values are presented for a single scan rate. kgpp

was calculated from the variation of 21Ep with scan rate, as described in Chapter 2. kgpp

values are listed, as well as the range of scan rates over which they were determined.

Finally, some comparison rate constant data for glassy carbon are given.

For Fe(CN)63'/4', Ru(NH3)63+/2+, and IrCl62'l3', 21Ep is approximately the same for

%M-

all four electrode types excluding Fe(CN)6 at the Sumitomo electrode, as 21Ep is 174

mV rather than ca. 80 mV. For Ru(NH3)63+/2+ and IrC162'/3', relatively low 21Ep values are

344-

seen even for moderately fast scan rates of 20-30 V/s, while for Fe(CN)6 , relatively

low 21Ep values are observed only at scan rates below 2 V/s. The ipox/ip'ed peak current

ratio for all three redox systems is near 1, as expected. Standard deviations in kgpp of

less than 10% are seen and, importantly, the deviation does not depend on the scan rate.

This indicates that the AEp values, hence kgpp, are not signiﬁcantly affected by

uncompensated ohmic resistance (both the solution and electrode) within the cell. kgpp

for Ru(NH3)63+/2+ and IrC162'/3‘ is in the low 10'1 cm/s range, which is close to the highest

values reported for freshly activated glassy carbon.19'24‘47’48 kg”, for Fe(CN)63'/4' is a

factor of 5 lower in the mid 10'2 cm/s range. The electrode kinetics for Fe(CN)63'/4' are a
little more sluggish at the Sumitomo electrode, possibly due to low levels of surface
oxygen. Surface carbon-oxygen functionalities are known to inhibit the electrode
kinetics for Fe(CN)63'/4' at diamond.”50 Although the presence of such functional groups

3+0+

was not experimentally veriﬁed, the Fe data presented below give credibility to this

supposition.

102

Table 4.1 Cyclic voltammetric and heterogeneous electron-transfer rate constant
data for ﬁve redox systems at MSUMicro, Condias, Sumitomo, and MSUNano boron-
doped diamond thin-ﬁlm electrodes.“

 

 

 

 

 

 

15821.? p” 5““R2“g° .. 1:3,,
Analyte Electrode Rate at 0.1 for kapp k app (cm/s) (cm/s) for
Ms» V" (Ws) ' GC”
MSUMicro 77 (2) 1.00 0.710 0.066 1 0.005
“((3306 341 Condias 83 (2) 1.01 0.35.0 0.056 1 0.005 0.10 1 0.02
a " 9-5 Sumitomo 174 (2) 1.01 01.20 0.0077 1 0.0006
MSUNano 81 (2) 1.01 1.03.5 0.053 1 0.004
MSUMicro 88 (20) 1.01 50.29 0.107 1 0.009
Rummy: 392+ Condias 90(1) 0.94 0.31.5 0.0213 1 0.0007 0.24 1 0.07
0‘ = 9-5 Sumitomo 97 (20) 0.95 6.0-20 0.082 1 0.007
MSUNano 88 (20) .93 40.30 0.102 1 0.008
MSUMicro 81 (30) 1.09 1550 f 0.19 1 0.04
1,06 143- Condias 80 (30) 1.08 12-30 0.27 5: 0.05 0.50 1 0.06
0‘ = 0-5 Sumitomo 85 (30) 1.08 15-30 0.18 1 0.03
MSUNano 89 (30) 1.09 1540 0.14 1 0.02
MSUMicro 360 (0.1) 1.81 01-10 1.0 x 10‘5
4-MC Condias 473 (0.1) 1.61 01—10 1.0 x 10'6
a = 0-45 Sumitomo 510 (0.1) 1.58 0.1-1.0 5.0 x 10'7 ......
MSUNano 365 (0.1) 1.66 01-10 8.5 x 10:6
MSUMicro 393 (0.1) 1.00 01.10 3.0 x 10“
Fey/2+ Condias 683 (0.1) 0.87 01-10 2.0 x 10'5 (2.5 1 0.3)
0‘ = 0-65 Sumitomo 614 (0.1) 1.08 01-10 2.0 x 10“ x 10-3
MSUNano 529 (0.1) 1.21 01-10 8.0 x 10'5

 

1‘ The supporting electrolyte for 0.1 mM solutions of Fe(CN)63'/4', Ru(NH3)63+/2+, and
1106243" was 1 M KCl, while 0.1 mM solutions of 4-MC and Few2+ were prepared in 0.1
M HClO4. Heterogeneous electron-transfer rate constants for Fe(CN)63'/4', Ru(NH3)63+/2+,
and IrCléz'B' were determined from AEp-v trends, while the rate constants for 4-MC and
Fey/2+, as well as all or values, were determined by digital simulations (DigiSim 3.03).

b GC heterogeneous electron-transfer rate constants from ref 44.

103

Consistent with previous observations, kgpp for 4-MC and Fey/2+ is several

orders of magnitude lower than the values for the other redox systems.7’9"3 As mentioned
above, these two redox systems proceed through an inner-sphere mechanism and thus the
kinetics of the reaction are very sensitive to the physicochemical properties of the
diamond thin-ﬁlm electrode. The largest rate constant for 4-MC is observed for the
MSUMicro electrode (10'5 cm/s), while the lowest (10'7 cm/s) is seen for the Sumitomo
electrode. We suppose that the relatively slow kinetics for this redox system on diamond
result from the lack of adsorption on the non-polar, hydrogen-terminated surface.
Research with glassy carbon has shown that the electrode kinetics for 4-MC are relatively
insensitive to the presence of surface oxides, and that adsorption to the surface is
necessary for rapid electron transfer.24 For example, Hunt-DuVall and McCreery
demonstrated that removing surface impurities to increase adsorption sites on glassy
carbon resulted in an increase in the electron-transfer rate for catechols.24 Their work
demonstrates the vital role adsorption plays in the electron-transfer process, effects that
may involve changes in the reorganization energy. Anything that inhibits the adsorption
(e. g., chemical modiﬁcation) of catechols also inhibits the electrode kinetics. As
mentioned above, related work in our group has shown that the adsorption of dopamine
on high-quality, hydrogen-terminated diamond can be promoted by the introduction of
nondiamond spz-bonded carbon impurity, which can be controllably incorporated during

the deposition.52

At this preliminary stage, there appears to be a direct correlation
between a reduction in 21Ep and an increase in the surface coverage and the nondiamond

carbon content.

104

3+/2+

The rate constant for Fe is in the 10“1 to 10'5 cm/s range for all four electrode

types. One of the largest kgpp is observed for the Sumitomo diamond thin ﬁlm, which

also exhibits the smallest kgpp for Fe(CN)63'/4'. As mentioned, surface oxygen inhibits

the Fe(CN)63'/4' and promotes the Few2+ electrode kinetics on boron-doped diamond thin-
ﬁlrn electrodes. Considering this, it is possible that this diamond thin-ﬁlm type has some

unintentional surface carbon-oxygen functional groups, likely of the carbonyl type.

The k 3P}? values were additionally veriﬁed by digital simulation (DigiSim® 3.03).

All simulated cyclic voltammetric i-E curves consisted of only faradaic current for the

redox system. kgpp values and transfer coefﬁcients, a, were varied to match the

simulated 21Ep and voltammetric i-E curve shape with the experimental curves. In all

0

cases for the experimentally determined kapp, the difference between the experimental

and simulated 21Ep values was 9 mV or less. Figure 4.5A and B show two examples of
simulated and experimental i-E curves for F e(CN):53 44' (0.1 V/s) at the Sumitomo
electrode and IrClbz'B' (50 V/s) at the MSUMicro electrode. The agreement of the 21Ep
and shape of the experimental and simulated voltammetric curves is excellent, even with
the absence of background current, which is most prominent for the higher scan rate. a
values of 0.5 for Fe(CN)63'/4', Ru(NH3)63+’2+, and IrCléz'B‘, 0.45 for 4-MC and 0.65 for

Few2+ were determined, consistent with previously reported values for diamond thin

ﬁlm.7

105

7.5
60'
4.5’
30’
1.5-
00'
-1.5
-30
-45.
-6.01A
-03 -02 -01 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Potential (V) vs. Ag/AgCl

 

—— Simulation
-------- Experimental

 

Current (uA)

 

 

 

r "

 

400

 

—— Simulation
300 - -------- Experimental

—
o

. -

‘°-_----"

200

I

100

 

Pr
0
o
0'
t
1

 

-100'

.....
-—-
----
su.
.
......
.
0.
‘.
___
~
.
s
.
s

-200 -

Current (uA)

-300

 

 

 

0.72 L0:3 . 0:4 4 0:5 ‘ 0:6 1 0:7 - 0:8 I 0i9 J 110 n 1.1
Potential (V) vs. Ag/AgCl

Figure 4.5 Experimental and simulated (DigiSim® 3.03) cyclic voltarmnetric i-E
curves for (A) 0.1 mM Fe(CN)63'/4' in l M KCl at the Sumitomo diamond electrode (0.1
V/s) and (B) 0.1 mM IrCl62'l3' in 1 M KCl at the MSUMicro diamond electrode (50 V/s).

The experimental voltammograms contain both the background and faradaic currents
while the simulated voltammograms contain just the faradaic current.

106

4. 2.3 pH Effects on Voltammetric Response

Oxides of varying type and coverage exist on spz-bonded carbon electrode
surfaces, terminating the edge plane and defect sites of the microstructure. They form
naturally during exposure to the atmosphere (air/H2O), and are also introduced during
pretreatment (e.g., mechanical polishing or anodic polarization). Surface oxides
inﬂuence the chemical and electrochemical properties of the carbon and they can have a
profound effect on the electric double-layer structure and the extent of analyte adsorption.
Carboxylic acid functional groups are common on oxygen-containing glassy carbon
surfaces with a pKa near 4545’54 Therefore, depending on the solution pH, these groups
can either be protonated (uncharged) or deprotonated (negatively charged). Highly
charged redox systems can be affected by the presence of these ionizable functional
groups (e. g., excess surface charge). For example, AEp for an anionic redox system, such
as F e(CN)(,3 74' or IrC162'/3', becomes strongly pH dependent, increasing signiﬁcantly when
these functional groups are deprotonated. On the other hand, 21Ep for a cationic redox

system, such as Ru(NH3)63+m, is reduced when these functional groups are deprotonated.

For instance, Deakin et al. observed that kgpp for Fe(CN)63’/4' or IrC162'/3‘ decreases with

increasing pH from 2 to 8, while kgpp for Ru(NH3)63+/2+ increases.54 The larger 21Ep for

Fe(CN)63'/4' and smaller value for Ru(NH3)63+/2+, when there is excess negative surface
charge, reﬂects more sluggish and more rapid electrode kinetics, respectively, because of
electrostatic repulsion and attraction (e.g., electric double-layer effects). Electrostatic
repulsion of Fe(CN)63'/4’ leads to a greater electron trmneling distance and more sluggish

electrode kinetics. The opposite trend holds true for Ru(NH3)63+/2+.

107

The response for charged redox analytes at hydrogen-terminated diamond thin-
ﬁlm electrodes should be unaffected by solution pH because of the absence of ionizable

and

surface oxides. 6344'

To test this, the cyclic voltammetric responses for Fe(CN)
Ru(NH3):53’+/2+ were examined at pH 1.90 and 7.35 — below and above the expected pKa of
the carboxylic acid ﬁrnctional groups. Table 4.2 presents a summary of the results for all

four electrode types. Generally speaking, there is minimal effect of solution pH on the

AEP and k0

app for Fe(CN)63'/4' and

values for these two redox systems. kgpp

Ru(NH3)(,3+’2+ differs by a factor of 2, or less, at both pH values, at all four diamond thin-

ﬁlm types, consistent with a low oxide, hydrogen-terminated electrode surface.

Table 4.2 Cyclic voltammetric and heterogeneous electron-transfer rate constant
data for Fe(CN)63'/4' and Ru(NH3)(,3+/2+ at MSUMicro, Condias, Sumitomo, and
MSUNano boron-doped diamond thin-ﬁhn electrodes, as a function of the solution pH.21

 

 

 

Scan Range
Analyte Electrode 102454 and 2118, Range (mV) kgpp (cm/S)
kapp (V/s)
MSUMicro 0.7-5.0 69-83 66-76 0.072 1 0.004 0.112 1 0.01
Fe(CN), 344- Condias 01-10 68-89 67-82 0.026 1 0.002 0.035 1 0.002
9 = 0‘5 Sumitomo 0.3-2.0 77-103 77-126 0.024 1 0.001 0.0141 1 0.0001
MSUNano 0.5-50 70-92 69-86 0.055 1 0.003 0.062 1 0.002
MSUMicro 15-39 78-97 82-99 0.13 1 0.01 0.104 1 0.004
Ru(NH:). 392+ Condias 01-10 76-113 79.115 0.011 1 0.001 0.011 1 0.001
a = 0-5 Sumitomo 3015 77-100 77-96 0.065 1 0.002 0.071 1 0.002
MSUNano 3020 73-95 71-88 0.083 1 0.002 0.103 1 0.004

 

‘ The supporting electrolyte for 0.1 mM Fe(CN)63'/4' and Ru(NH3)63'+/2+ was 1 M
phosphate buffer at pH 7.35 and 1.90. Heterogeneous electron-transfer rate constants
were determined from AEp-v trends and veriﬁed by digital simulation (Di giSim 3.03).

108

4.3 Conclusions

The electrochemical response of two commercial (Condias and Sumitomo) boron-
doped diamond thin-ﬁlm electrodes was compared with that of two types of boron-doped
diamond thin ﬁlm deposited in our laboratory (microcrystalline and nanocrystalline).
The electrode response for all four electrode types, at least for the redox systems probed,
was very reproducible from type-to-type and from ﬁlm-to-ﬁlrn for a given type. The
MSUMicro, Condias and Sumitomo ﬁlms had similar textures, with nominal grain sizes
of 1, 5 and 2 pm, respectively. The MSUNano ﬁlm was much smoother, with 100 nm
aggregates of diamond grains. The Raman spectral features for each ﬁlm type were
consistent with high-quality (i.e., low spz-bonded carbon impurity content) boron-doped
diamond. All the diamond electrodes exhibited a low and stable background current,

about a factor of 10 lower than comparably sized glassy carbon. Apparent heterogeneous

electron-transfer rate constants, k0 for Fe(CN)63"4', Ru(NH3)63+/2+, and IrClém' were

app,
in the mid 10'2 to low 10'1 cm/s range without extensive electrode pretreatment. kgpp for
4-MC and Fey/2+ was much lower, in the range of 10'5 to 10'7 cm/s. k0 for Fe(CN)63'

“PP
’4' and Ru(NH3)63+/2+, at all four diamond types, was relatively independent of the solution
pH. Surface carbon-oxygen functionalities, particularly those that are ionizable (e.g.,
carboxylic acid, pKa ~4.5), can strongly inﬂuence the carbon electrode response. Such
effects are routinely observed for glassy carbon, as the surface normally contains carbon-
oxygen functionalities. Hydrogen-terminated diamond surfaces contain few, if any,
ionizable carbon-oxygen functionalities; therefore, these pH effects are not observed.

These results conﬁrm that high-quality boron-doped diamond with reproducible

109

electrochemical behavior is available commercially, such that production of diamond
electrodes as advanced electrocatalyst supports is possible once a suitable geometry is

established.

110

4L4!

(1)
(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

Iszﬂareurcuas

Nicholson, R. S. Anal. Chem. 1965, 37, 1351-1355.

Li, L.-F.; Totir, D.; Miller, B.; Chottiner, G.; Argoitia, A.; Angus, J. C.; Scherson,
D. J. Am. Chem. Soc. 1997, 119, 7875-7876.

Xu, J .; Granger, M. C.; Chen, Q.; Strojek, J. W.; Lister, T. E.; Swain, G. M. Anal.
Chem. 1997, 69, 591A-597A.

Swain, G. M.; Anderson, A. B.; Angus, J. C. MRS Bulletin 1998, 23, 56-60.
Chen, Q.; Swain, G. M. Langmuir 1998, 14, 7017-7026.
Xu, J .; Chen, Q.; Swain, G. M. Anal. Chem. 1998, 70, 3146-3154.

Granger, M. C.; Witek, M.; Xu, J .; Wang, J.; Hupert, M.; Hanks, A.; Koppang, M.
D.; Butler, J. E.; Lucazeau, G.; Mermoux, M.; Strojek, J. W.; Swain, G. M. Anal.
Chem. 2000, 72, 3793-3804.

Rao, T. N.; Yagi, I.; Miwa, T.; Tryk, D. A.; Fujishima, A. Anal. Chem. 1999, 71,
2506-2511.

Show, Y.; Witek, M. A.; Sonthalia, P.; Swain, G. M. Chem. Mater. 2003, 15, 879-
888.

Spataru, N.; Sarada, B. V.; Popa, E.; Tryk, D. A.; Fujishima, A. Anal. Chem.
2001, 73, 514-519.

Rao, T. N.; Ivandini, T. A.; Terashima, C.; Sarada, B. V.; Fujishima, A. New
Diam. Front. C. Tech. 2003, 13, 79-88.

Pleskov, Y. Russ. Chem. Rev. 1999, 68, 381-392.

Hupert, M.; Muck, A.; Wang, J .; Stotter, J .; Cvackova, Z.; Haymond, S.; Show,
Y.; Swain, G. M. Diam. Relat. Mater. 2003, 12, 1940-1949.

Stotter, J .; Zak, J .; Behler, Z.; Show, Y.; Swain, G. M. Anal. Chem. 2002, 74,
5924-5930.

Martin, H. B.; Morrison, P. W. J. Electrochem. Solid St. 2001, 4, E17-E20.

111

(16)

(17)

(18)

(19)

(20)

(21)
(22)
(23)
(24)

(25)

(26)

(27)

(28)

(29)

(30)

(31)

Haymond, S.; Zak, J .; Show, Y.; Butler, J. E.; Babcock, G. T.; Swain, G. M. Anal.
Chim. Acta. 2003, 500, 137-144.

Stotter, J .; Haymond, S.; Zak, J.; Show, Y.; Cvackova, Z.; Swain, G. M. Interface
2003, 12, 33-38.

Ager, J. W. I.; Walukiewicz, W.; McCluskey, M.; Plano, M. A.; Landstrass, M. 1.
Appl. Phys. Lett. 1995, 66, 616-618.

Rice, R. J .; Pontikos, N. M.; McCreery, R. L. J. Am. Chem. Soc. 1989, 112, 4617-
4662.

Yang, H.-H.; McCreery, R. L. Anal. Chem. 1999, 71 , 4081-4087.

Chen, P.; Fryling, M. A.; McCreery, R. L. Anal. Chem. 1995, 67, 3115-3122.
Chen, P.; McCreery, R. L. Anal. Chem. 1996, 68, 3958-3965.

Kneten, K. R.; McCreery, R. L. Anal. Chem. 1992, 64, 2518-2524.

DuVall, S. H.; McCreery, R. L. J. Am. Chem. Soc. 2000, 122, 6759-6764.

Angus, J. C.; Sunkara, M.; Shahaida, S. R.; Glass, J. T. J. Mater. Res. 1992, 7,
3001-3009.

Gruen, D. M. MRS Bulletin 1998, 23, 32-35.

Zhao, D.; Gruen, D. M.; Qin, L. C.; McCauley, T. G.; Krauss, A. R. J. Appl. Phys.
1998, 84, 1981-1989.

Jiao, S.; Sumant, A.; Kirk, M. A.; Gruen, D. M.; Krauss, A. R.; Auciello, O. J.
Appl. Phys. 2001, 90, 118-122.

McCauley, T. G.; Gruen, D. M.; Krauss, A. R. Appl. Phys. Lett. 1998, 73, 1646-
1648.

Bhattacharyya, S.; Auciello, O.; Birrell, J .; Carlisle, J. A.; Curtiss, L. A.; Goyette,
A. N.; Gruen, D. M.; Krauss, A. R.; Schlueter, J .; Sumant, A.; Zapol, P. Appl.
Phys. Lett. 2001, 79, 1441-1443.

Chen, Q.; Gruen, D. M.; Krauss, A. R.; Corrigan, T. D.; Witek, M.; Swain, G. M.
J. Electrochem. Soc. 2001, 148, E44-E51.

112

(32)

(33)

(34)

(35)

(36)
(37)

(38)

(39)

(40)

(41)

(42)

(43)

(44)

(45)

(46)

(47)

Gruen, D. M.; Krauss, A. R.; Zuiker, C. D.; Csencsits, R.; Terrninello, L. T.;
Carlisle, J. A.; J irnenez, 1.; Sutherland, D. G. J .; Shu, D. K.; Tong, W.; Himpsel,
F. Appl. Phys. Lett. 1996, 68, 1640-1642.

Dennison, J. R.; Holz, M. W.; Swain, G. M. Spectroscopy 1996, 11, 38-45.

Mermoux, M.; Marcus, B.; Swain, G. M.; Butler, J. E. J. Phys. Chem. B 2002,
106, 10816-10827.

Nemanich, R. J .; Glass, J. T.; Lucovsky, G.; Shroder, R. E. J. Vac. Sci. T echnol.
1988, A6, 1783-1787.

Knight, D. S.; White, W. B. J. Mater. Res. 1989, 4, 385-393.
Bergman, L.; Nemanich, R. J. J. Appl. Phys. 1995, 78, 6709-6719.

Gonon, P.; Gheeraert, E.; Deneuville, A.; Fontaine, F .; Abello, L.; Lucazeau, G. J.
Appl. Phys. 1995, 78, 7059-7062.

Ushizawa, K.; Watanabe, K.; Ando, T.; Sakaguchi, 1.; Nishitani-Gamo, M.; Sato,
Y.; Kanda, H. Diam. Relat. Mater. 1998, 7, 1719-1722.

Pruvost, F.; Bustarret, E.; Deneuville, A. Diam. Relat. Mater. 2000, 9, 295-299.
Pruvost, F .; Deneuville, A. Diamond Relat. Mater. 2001, 10, 531-535.

Prawer, S.; Nugent, K. W.; Jarnieson, D. N.; Orwa, J. O.; Bursill, L. A.; Peng, J.
L. Chem. Phys. Lett. 2000, 332, 93-97.

Ferrari, A. C.; Robertson, J. Phys. Rev. B. Condens. Matter Mater. Phys. 2001,
63,1214051-1214054.

Birrell, J .; Gerbi, J. E.; Auciello, 0.; Gibson, J. M.; Johnson, J .; Carlisle, J. A.
Diam. Relat. Mater. 2005, 14, 86-92.

McCreery, R. L. In Neuromethods; Adams, R. N., Ed.; Humana Press, 1995; Vol.
27, pp 1-26.

Fagan, D. T.; Hu, I.-F.; Kuwana, T. Anal. Chem. 1985, 5 7, 2759-2763.

Swain, G. M. In Electroanalytical Chemistry; Rubinstein, 1., Ed.; Marcel Dekker,
2003; Vol. 22, pp 181-277.

113

(48)

(49)

(50)

(51)

(52)

(53)

(54)

Swain, G. M. In Thin Film Diamond 11; Ristein, J ., Ed.; Elsevier Academic Press:
Boston, 2004; Vol. 77, pp 121-148.

Yagi, 1.; Notsu, H.; Kondo, T.; Tryk, D. A.; Fujishima, A. J. Electroanal. Chem.
1999, 473, 179-185.

Granger, M. C.; Swain, G. M. J. Electrochem. Soc. 1999, 146, 4551-4558.

McDermott, C. A.; Kneten, K. R.; McCreery, R. L. J. Electrochem. Soc. 1993,
140, 2593-2599.

Bennett, J. A.; Wang, J .; Show, Y.; Swain, G. M. J. Electrochem. Soc. 2004, 151,
E306-E313.

Hu, I.-F.; Karweik, D. H.; Kuwana, T. J. Electroanal. Chem. 1985, 188, 59-72.

Deakin, M. R.; Stutts, K. J .; Wightrnan, R. M. J. Electroanal. Chem. 1985, 182,
1 13-122.

114

Chapter 5

Preparation and Characterization of Boron-Doped
Diamond Powder: A Possible Dimensionally Stable
Electrocatalyst Support Material

Abstract

Electrically conducting diamond powder was prepared by coating insulating
diamond powder (8-12 pm diam, ~2 mz/g) with a thin boron-doped diamond layer using
microwave plasma-assisted chemical vapor deposition. Deposition times from 1 to 6 h
were evaluated. Scanning electron microscopy revealed that the diamond powder
particles became more faceted with secondary growths developing at longer deposition
times. Fusion of neighboring particles was also observed with increasing grth time.
The ﬁrst-order diamond phonon line appeared in the Rarnan spectrum at ca. 1331 cm'1
for deposition times up to 4 h, and was downshifted to as low as 1317 cm’1 for some
particles after the 6 h growth. Electrical resistance measurements of the bulk powder
conﬁrmed that a conductive diamond overlayer formed, as the conductivity increased
with deposition time from near zero (insulating, <10'5 S/cm) for the uncoated powder to
1.5 S/cm after the 6 h growth. Ohmic behavior was seen in current-voltage curves
recorded for the 4 h powder between :th V. Cyclic voltammetric i-E curves for
F e(CN)(,3 44‘ and Ru(NH3)(,3+/2+ were recorded to evaluate the electrochemical properties of
the conductive powder when mixed with a 10 wt. % polytetraﬂuoroethylene (PTFE)

binder. At scan rates between 10 and 500 mV/s, 21Ep for Fe(CN)63'/4' ranged from 140 to

350 mV while AEP for Ru(NH3)63+/2+ was slightly larger. The large AEP reﬂect 1) ohmic

115

resistance within the powder/binder electrode because of poor particle-particle
connectivity between the hard, jagged powder particles and 2) signiﬁcant hydrophobicity
(i.e., low electrode wetting) at a binder concentration of 10 wt. % PTFE. (AEp for
Ru(NH3)63+/2+ decreased signiﬁcantly by reducing the binder to 5 wt. %). Anodic
polarization at 1.6 V vs. Ag/AgCl and 25 °C for l h was performed to compare the
morphological and microstructural stability of the conductive diamond, graphite, and
glassy carbon (GC) powders. The total charge passed during polarization was largest for
the GC powder (0.88 C/cmz) and smallest for conductive diamond powder (0.18 C/cmz).
SEM images taken of conductive diamond powder after polarization showed no evidence

for degradation while signiﬁcant morphological changes were seen for GC.

116

5.1 Introduction

In Chapter 1, the criteria for an ideal polymer electrolyte membrane fuel cell
(PEMFC) electrocatalyst support material were outlined. Among other important
requirements, high surface area (~100 mz/g) supports are necessary to maximize
electrocatalyst dispersion and utilization, while a gas permeable support is needed to
allow diffusion of the reactants to the triple phase boundary. Typical spz-bonded carbon-
based supports in a powder form, such as Vulcan XC-72, satisfy both of these
requirements, but lack long-term stability in PEMFCs. Planar boron-doped diamond
ﬁlms are neither high-surface-area nor gas permeable, but exhibit excellent dimensional
stability in harsh environments."2 Thus, we believe development of high-surface-area,
gas permeable, and corrosion resistant diamond electrode architectures will result in a
new electrocatalyst support material for PEMFCs. Our ﬁrst step toward development of
an advanced electrocatalyst support was the preparation and characterization of
electrically conducting diamond powder.

High-surface-area diamond powder (so-called nanodiamond) is produced in large
quantities worldwide via the detonation of explosives. This nanodiamond powder has a
nominal particle size of ca. 5-10 nm and a speciﬁc surface area of 200-600 mz/g.3
Although it is available commercially and is relatively inexpensive ($1 to S/g),
nanodiamond powder is not electrically conducting, and thus, cannot be used as an
electrocatalyst support. Diamond powder of lower surface area, used as an abrasive, is
also available commercially. This diamond powder is produced by high pressure, high

temperature synthesis and typically has particle size ranging from ca. 0.1 to 100 um and a

speciﬁc surface area of ca. 5 mZ/g, or lower. However, the commercially-available

117

abrasive powders are not electrically conducting. To overcome this limitation and to
investigate the potential of conducting diamond powder as an electrocatalyst support, we
produced electrically conducting diamond powder by coating the diamond abrasive with
a thin layer of boron-doped diamond via microwave plasma-assisted chemical vapor
deposition (PACVD).

This chapter describes the initial phase of our advanced carbon support
development in which the ﬁrst-known electrically conducting diamond powder was
prepared and characterized. The powder was formed by coating insulating abrasive
diamond powder with a boron-doped diamond overlayer, as depicted in Figure 5.1.
Conductive powders were prepared using deposition times from 1 to 6 h. Although the
surface area of this new material (~2 mz/ g) is lower than that desired for an ideal support,
it is a useful starting point for making electrically conducting diamond powder and
evaluating its physical, electrical, and electrochemical properties. The powder
morphology and microstructure were characterized before and after conductive diamond
growth by scanning electron microscopy (SEM) and Raman spectroscopy. X-ray
diffraction (XRD) analysis was used to verify the crystallinity of the coated powder.
Bulk electrical measurements were performed to determine the ohmic resistance of the
packed powder. Electrodes for electrochemical testing were prepared by mixing the
conductive powder with a p'olytetraﬂuoroethylene (PTFE) binder, casting a thin layer of
the slurry on glass, and curing in an oven. The electrochemical properties of the powder
electrode were investigated by cyclic voltammetry using two redox systems: F e(CN)63 '/ 4'

and Ru(NH3)63+/2+. Finally, potentiostatic measurements were performed in strong acid

to evaluate the degradation resistance of the conducting diamond during anodic

118

polarization. Comparison measurements were made with graphite and glassy carbon

(GC) powders.

 

 

' ’
Figure 5.1 Deposition scheme for coating insulating diamond powder with a
conductive, boron-doped diamond overlayer showing both full and partial grain coverage

as well as fusion of neighboring grains. The shaded areas represent the conducting
diamond overlayer.

5.2 Results and Discussion

5. 2. 1 Material Characterization and Electrical Measurements

According to the scheme in Figure 5.1, the insulating powder is either partially or
fully covered with an electrically conductinglayer of boron-doped diamond. Scanning
electron micrographs of the uncoated (acid-washed) and coated powders aﬂer deposition
times of l, 2, and 4 h are shown in Figure 5.2A-D. Although it is difﬁcult to distinguish
differences in particle size, changes in the powder morphology are apparent as a function
of the deposition time. The uncoated powder particles in Figure 5.2A are irregularly
shaped and characterized by jagged edges with little faceting. The particle diameters are
in the range of ~8-12 um. Figure 5.2B shows the powder after a 1 h deposition. The

particle edges become smoother and more well-deﬁned faceting develops. Many of the

119

particle surfaces consist of multiple grooves along the edges of the triangular facets
which may be sites of preferential nucleation. For example, carbon atoms may nucleate
at these sites with growth continuing laterally across the surface. Pits are also present on
the facet surfaces and may result from incomplete coalescence of grth layers across

the surface, which begin at different nucleation sites.

 

Figure 5.2 Scanning electron micrographs of (A) bare diamond powder and
conductive diamond powder aﬁer (B) 1, (C) 2, and (D) 4 h coating times.

The most obvious morphological change after diamond deposition is the fusion of

neighboring particles, as indicated by the arrow in Figure 5.2B. During the 2 h

120

deposition, more neighboring particles fused together forming clusters of 4 or 5 grains, as
seen in Figure 5.2C. These clusters become larger with longer deposition times. Clusters
of ~8 to 10 diamond particles formed during the 4 h deposition and even larger clusters
formed during the 6 h period. Ideal growth conditions would produce individually coated
diamond particles without the loss of surface area that occurs when neighboring particles
fuse together. On the other hand, the particle fusion likely imparts a greater degree of
mechanical stability and electrical conductivity to the powder. For this deposition
method, shorter growth times of 2 or 4 h seem appropriate to form the conductive
diamond overlayer while minimizing the extent of particle fusion.

Figure 5.3 presents an X-ray diffraction (XRD) spectrum of a 4 h grth powder,
which is representative of spectra collected for all growth times. The XRD spectrum
reveals highly crystalline diamond with no scattering for spz-bonded carbon impurity
(graphitic or amorphous carbon) at all the deposition times (e. g., no d002 graphite peak at
26.5 degrees).4 Intense peaks at 43.9, 75.3, and 91.5 degrees are present that correspond
to the (111), (220), and (311) crystallographic orientations of cubic diamond (ASTM 6-

0675). The broad peak at 69.5 degrees is from the Si substrate (ASTM 27-1402).

121

 

(111)

(220)

L 1L Ah (311)

L 1 I 1 I 1 I 1 I L I

20 30 4O 50 60 70 80 90 100

2 Theta (Degrees)

Figure 5.3 X-ray diffraction spectrum of a 4 h growth diamond powder sample.

Intensity (a.u.)

 

 

 

 

 

Raman spectroscopy was also used to evaluate the microstructure of the coated
diamond powders. Figure 5.4 presents Raman spectra for the diamond powders after
deposition times of 0 (acid washed powder), 1, 2, 4, and 6 h. The ﬁrst-order diamond
phonon line, expected at 1332 cm], is present in the spectra for the uncoated and coated
powders after the 1, 2, and 4 h periods. The linewidth (FWHM) remains relatively
constant, ca. 8 cm'l, for all deposition times. For comparison, the measured linewidth for
an HPHT reference diamond sample was 7 cm". To a ﬁrst approximation, the linewidth
is inversely related to the phonon lifetime and is a measure of the defect density. Thus,
the F WHM data suggest that the defect density in the diamond overlayer does not change

much from that of the uncoated powder particles. However, a major difference in the

122

Rarnan spectrum is seen for the 6 h deposition as the ﬁrst-order diamond phonon line
downshifts by an average of ca. 6 cm'l. In addition, there is a large deviation in the line
position for different powder particles in a 6 h growth sample. For example, the diamond
line position for about half the particles sampled was 1331 cm", while the peak was
downshifted for the other half to as low as 1317 cm". The diamond line position is
known to shift with stress; the degree and direction of which are measures of stress
magnitude and type (i.e., tensile or compressive).5'7 Particle fusion in the 6 h grth may
cause local tensile stress that is reﬂected in the downshiﬂ in the one-phonon line. In
addition to changes in the diamond line position at long deposition times, the spectra also
show that the scattering intensity from nondiamond carbon impurity, centered between
1550 and 1580 cm'l, increases with growth time. Some scattering intensity is seen for all
the deposition times, but is most intense for the 6 h growth. The nondiamond carbon is

not pure graphite but likely consists of a mixture of sp2 and sp3-bonded carbon.

In order to verify that the particles were coated with an electrically conducting
overlayer, bulk powder electrical resistance measurements were made and the data are
presented as conductivities in Table 5.1. These data provide conclusive evidence that the
particles are either partially or fully coated with a conducting overlayer and that a high
level of connectivity exists when the powder particles are packed together. Several
factors inﬂuence the electrical conductivity of the packed powder including (i) the
electronic properties of the overlayer (charge carrier concentration and mobility) that are
governed primarily by the boron-doping level, (ii) the extent and conformity of the
conductive diamond overlayer, (iii) the particle-particle connectivity, and (iv) the

nondiamond carbon impurity that might form during deposition. As seen in Table 5.1,

123

the uncoated powder has poor conductivity (resistance >40 MO). The powder
conductivity after the l h deposition increases to 0.027 S/cm for the as-deposited (i.e., no
post-grth cleaning treatment) powder. The powder conductivity increases further for

the as-grown 4 and 6 h depositions to 1.4 and 1.5 S/cm, respectively.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LA 6 h Growth
6 J 4 h Growth
=3 - - ~ ~~ s 11 ‘ r ‘ “
C6 1 . 1 . 1 3 1 1 . 1 1 1 . 1 . 1 . 1 .
v J 2 h Growth
9% I 1 I 1 i I 1 L 1 I I 1 I 1 l 1 I 1
5: i 1 h Growth
CD
1-1 a
q I 1 I 1 I : 1 I 1 L 1 I 1 I 1 I 1 I 1
"1 J No Growth

 

 

 

1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000

Raman Shift (cm-1)

Figure 5.4 Raman spectra of bare diamond powder and conductive diamond powder
after coating times of 1, 2, 4, and 6 h.

124

Table 5.1 Bulk conductivity values of graphite and CVD diamond powder before
(as-grown) and after (cleaned) washing in warm aqua regia (30 min) and warm 30%
H202 (30 min).

 

 

Sample Conductance (S/cm)
Graphite powder 1.3
Acid washed diamond powder 0.0

Conductive diamond

Powder As-grown Cleaned
1 h growth 0.027 4.0 x 10'5
2 h growth 0.015 0.010
4 h grth 1.4 0.29
6 h growth 1.5 0.30

 

As indicated by the Raman data, nondiamond carbon impurity forms during the
deposition and this will have an inﬂuence on the overlayer conductivity. In order to
assess the contribution of nondiamond carbon, the conductivity of the powders was also
measured after acid washing (i.e., "cleaned"), which has been shown to remove
nondiamond carbon and other impurities.8 Scanning electron micrographs of a 4 h
conductive diamond powder before and after acid washing are presented in Figure 5.5A
and B, respectively. There was no evidence of morphological damage to the diamond
powder as a result of acid washing, however, the conductivity of the samples changed.
For example, Table 5.1 shows that the conductivity of the 1 h powder decreases by 3
orders of magnitude after cleaning, which is consistent with its original conductivity
being largely due to the presence of the nondiamond carbon impurity on the surface.

After acid washing, the conductivity for the 4 and 6 h growths decreases by only a factor

125

of 5-7. Therefore, their conductivity is primarily due to the boron-doped overlayer with

much less contribution from nondiamond carbon impurity.

 

Figure 5.5 Scanning electron micrographs of a 4 h growth conductive diamond
powder (A) as-grown and (B) after acid washing.

The data in Table 5.1 were obtained from a simple resistance measurement using
an ohmmeter. However, the conductivity of the diamond powder alter a 4 h deposition
and the graphite powder were also evaluated by measuring i-V curves between $10 V.
Figure 5.6 shows these data, in addition to that of a boron-doped diamond (BDD) thin-
ﬁlm electrode, for comparison. Ohmic behavior (linear curve) was observed for both
powder types and for BDD thin ﬁlm over the entire voltage range with linear regression
correlation coefﬁcients greater than 0.999. The conductivities (I/R) of the conductive
diamond and graphite powders were calculated from the curve slopes (V = iR) and were
in agreement with the resistance values obtained by the ohmmeter, 18 and 6 9,

respectively. These results indicate that the boron-doped powder has high electrical

126

conductivity exhibiting ohmic behavior between $10 V and that good particle-particle

contact is achieved when packed as a dry powder.

 

100 ’ —--— Conductive Diamond Powder f6

- —o— Graphite Powder
—v— BDD Thin Film

NU’I\I
MOUI

 

 

 

I1I1ILJ1I I111I1I I11

-10-8-6-4-20246810

Applied Voltage (V)

Figure 5.6 I-V curves for a 4 h growth conductive diamond powder, graphite powder,
and a boron-doped diamond thin ﬁlm over a $10 V range.

 

5. 2.2 Electrochemical Characterization

Figure 5.7A shows background cyclic voltammetric i-E curves in 0.1 M HClO4
for the diamond powder electrodes bound with PTFE (10 wt. %) and mounted on glass
slides. Curves are shown for powder samples coated with a conductive diamond
overlayer for different periods of time. All the powders were used as-grown and not

cleaned by acid washing. The curve for the uncoated diamond powder (not shown)

127

exhibited zero current at all potentials due to its electrically insulating nature. It is
important to remember that for sufﬁcient electrical conduction through the powder
electrode, the particles must be conforrnally coated with a conductive diamond overlayer
and be in good contact with one another. The network of conductive particles must also
be in good electrical contact with the current collector (e.g., copper foil). If poor
electrical connectivity exists within the powder network, then the electrical resistance
will be high and this resistance will affect the i-E curve by distorting the shape and
increasing the AEP value for a redox test system. The curves in Figure 5.7A appear
relatively ﬂat and featureless within the working potential window for all deposition
times. Close inspection of the curve shapes, however, reveals that small oxidation and
reduction peaks develop near ~350 mV, as shown in Figure 5.7B, particularly for the 4
and 6 h powders. These peaks are commonly seen for spz-bonded carbon electrodes with
a signiﬁcant fraction of exposed edge plane (e.g., GC) where redox-active surface
carbon-oxygen functionalities can form (e.g., quinones).9 Since such peaks are not
normally seen with high quality, hydrogen-terminated diamond thin ﬁlms, we attribute
these peaks to carbon-oxygen functional groups that form at reactive sites on the
nondiamond carbon impurity. The largest background current in this region is seen for
the 6 h growth. These electrochemical observations are consistent with the Raman data
presented above which revealed increasing nondiamond carbon impurity content for the 4

and 6 h powders.

128

 

 

 

  

 

 

 

 

 

 

 

 

 

2.5 L —2 hGrowth
A 2.01, ~~4hGrowth A
NE 15~_----6hGrowth
0 1.0}
2 0.5 17
3; 0.0:
*5 -05-
0) -10"
i: -15"
:3 ' .
U -20
-25
_30 1 1 1 1 1 1 1 1 g 1 1 1
-03 0.0 0.3 0.6 0.9 1.2 1.5
Potential (V vs. Ag/AgCl)
1.0
0 8» —— 2 hGrowth
A ' {—— 4hGrowth B
E 0.6---- “Growth/u-
1 0.2
v
5:: .
-1.0 h 1 I 1 l 1 I 1
00 0.2 0.4 0.6 08

1 Potential (V vs. Ag/AgCl) .

Figure 5.7 (A) Background cyclic voltammetric i-E curves for conductive diamond
powder electrodes (2, 4, and 6 h deposition times) in 0.1 M HClO4 at 100 mV/s. (B)
Enlarged view of an 800 mV region. Current is normalized to the diamond powder
surface area exposed to solution.

129

In general, the working potential window (potential range between
solvent/electrolyte breakdown) for all the grth times is less than what is commonly
observed for high-quality diamond thin ﬁlms.8’1°'l3 Such diamond thin-ﬁlm electrodes
usually exhibit a ﬂat and featureless voltammetric response over a potential window
greater than 3 V in acidic media. The smaller potential window for the conducting
powders is also consistent with the presence of nondiamond carbon impurity on the
powder surface. A small amount of this impurity is sufﬁcient to cause a reduced
potential window, as oxygen and hydrogen evolution are known to occur with lower
overpotential on spz-bonded carbon electrodes, such as GC.ll In fact, research with
diamond thin-ﬁlm electrodes containing differing levels of nondiamond carbon impurity
has revealed that as the impurity increases, the working potential window decreases.14

The electrochemical response of the conductive diamond powder (with 10 wt. %

-/-
34 and

PTFE binder) was also probed by cyclic voltammetry using Fe(CN)6
Ru(NH3)63+’2+. Both of these redox systems are regularly used to study the
electrochemical response of a variety of carbon electrodes, including boron-doped
diamond thin ﬁlm and spz-bonded carbon, such as planar GC, as discussed in Chapter
4.8’10’n’J5'2| The cyclic voltammetric AEp values and curve shapes for electrochemical
redox systems are inﬂuenced by a number of variables, including the electron-transfer
kinetics and ohmic resistance effects. For example, the density of electronic states in the
electrode material near the formal potential of the redox system (i.e., the electrical
conductivity) affects the electron-transfer kinetics. Regardless of the mechanism (i.e.,

inner- or outer-sphere), high rates of electron-transfer can only occur when the electrode

has a sufﬁcient density of states available at the energy (i.e., potential) required to accept

130

electrons from donor states of the analyte or to donate electrons to acceptor states of the
analyte.18 Previous work has shown that highly boron-doped diamond thin ﬁlm possess a
high density of electronic states and supports relatively rapid electron-transfer rates for
Fe(CN)63'/4‘ and Ru(NH3)63+/2+.8’1°"3’20'23 Ohmic resistance (solution or electrode) can
also signiﬁcantly affect curve shape, increasing AEp and decreasing the peak current (ip).
Cyclic voltammetric i-E curves for Fe(CN)63”4' (10 to 500 mV/s) for a 4 h as-
grown diamond powder electrode are shown in Figure 5.8A. Oxidation and reduction
peaks are seen for Fe(CN)63'/4' in Figure 5.8A, although the waveform is distorted and
AEp is large, ranging from 143 mV at 10 mV/s to 348 mV at 500 mV/s. Table 5.2 lists
AEp recorded for Fe(CN)63'/4' for a 4 h conductive diamond powder, graphite powder, GC
powder, and boron-doped diamond thin-ﬁlm electrode at 10, 100, and 500 mV/s. As
shown in this table, AEP for Fe(CN)63'/4' for the GC and graphite powder electrodes
ranged from 128 to 213 mV, and 102 to 137 mV, respectively, at the same scan rates.
The AEp values for all the powders are larger than the typical values seen for this redox
system at diamond thin ﬁlm or a planar GC electrode; AEp values of ~60-7O mV at 100
mV/s are typical for high quality microcrystalline and nanocrystalline boron-doped
diamond thin-ﬁlm electrodes and ﬁ'eshly polished GC.8’13 In addition to the large AEp

values reported here for Fe(CN)53'/4'

, it was generally observed the AEp for a given
diamond powder electrode increased with usage, apparently due to lessened connectivity
between the powder particles. In other words, rinsing the cell between solutions, as well
as the turbulence caused by degassing the solution may have weakened the bound particle

matrix. It should be noted that there was no visual evidence that the electrodes

delaminated or that particles were dislodged from the bound electrode.

131

Table 5.2 Cyclic voltammetric AEP values (mV) for 1 mM Fe(CN)63'/4' at conductive
diamond powder, graphite powder, GC powder, and a planar microcrystalline boron-
doped diamond thin-ﬁlm electrodes at three different scan rates.

 

 

Scan Rate Conductive Graphite GC Microcrystalline
(mV/s) Diamond Powder Powder Powder Thin Film
10 140 102 128 60
100 240 137 213 62
500 350 174 329 67

 

As mentioned, the voltammetric curve shapes can be signiﬁcantly affected by
ohmic resistance either within the solution or the electrode, in addition to electron-
transfer rate or density of states effects. Based on what is known about 21Ep values at
diamond thin-ﬁlm electrodes and the bulk resistance measurements presented earlier
which indicated that a well-packed coated diamond powder possesses relatively high
electrical conductivity, the large AEP values are attributed mainly to ohmic resistance
effects. More speciﬁcally, poor particle-particle connectivity within the electrode bound
with PTFE limits the electrical conductivity. The connectivity problems could be due to
(i) rough particle morphology which would limit the contact area or (ii) uneven
distribution of binder though the powder network. SEM images revealed that the
diamond particles are irregularly shaped with sharp edges, which may create challenges
in achieving good connectivity when a binder is used. To test this hypothesis, a
composite electrode was made, consisting of 4 h grth diamond powder mixed with 20
wt. % graphite powder and the 10 wt. % PTFE binder. If the rough diamond particle
shape (i.e., low contact area) is responsible for the limited connectivity, the ﬂatter,

compressible graphite particles should serve to improve the particle-particle contact,

132

thereby increasing the electrical connectivity. AEp for Fe(CN)63"4’

at this composite
electrode was 86 mV at 10 mV/s, much less than the value for the diamond powder
electrode without any added graphite powder. While it is possible that the reduction in
AEp is inﬂuenced by more facile electron-transfer kinetics at the graphite powder surface,
the effect is likely more one of reduced ohmic resistance due to better particle-particle
connectivity. This is based on the fact that AEP for the composite electrode was lower
than that for the graphite powder alone.

Figure 5.8B shows a plot of the oxidation peak current for Fe(CN)63”4' versus the

1’ 2. The peak current increases linearly (r2>0.99) up to 100 mV/s indicative of a

scan rate
mass-transport limited response (semi-inﬁnite linear diffusion). However, at scan rates
above 100 mV/s, the faradaic current response levels off. SuCh behavior is not observed
for a boron-doped diamond thin ﬁlm, as the oxidation peak current increases
proportionally over the entire scan rate range. In general, current ﬂow through an
electrode is limited by the electrical conductivity, speciﬁcally the charge carrier
concentration and carrier mobility. If the carrier concentration or mobility is limited, the
current response may level off as the potential scan rate or analyte concentration is
increased (i.e., as more current is passed). The bulk electrical conductivity measurements
showed that when the coated powder particles are well packed, relatively high
conductivity is seen. This suggests a highly conducting boron-doped diamond layer is
formed over most of the powder particle surfaces and that good particle-particle contact
can be established so as to not severely restrict the carrier mobility. For example, in the

i—V curve measurements between ilO V, tens to hundreds of mA ﬂowed through the

packed powder, much higher currents than those ﬂowing during the electrochemical

133

measurement. Therefore, we attribute the limited faradaic current response for
Fe(CN)63'/4' at high scan rates to a limited carrier mobility within the conductive diamond
powder due to poor particle-particle connectivity when bound with PTFE, rather than to

low electrical conductivity of the diamond overlayer.

 

  
  
   
 

 

. n 500 mV/s
75 f 300 mV/s
NA 60 - 100 mV/s
E 45 - 7O mV/s
O 30 L 50 mV/s
2 15 *_ 30 mV/s
3 0:
a-b -15 -
5 -3oL
E .45 3
-50 -
U -75 C- A

 

 

 

90 _ 1 1 1 1 1 1 1 1
-0.2 0.0 0.2 0.4 0.6 0.8

Potential (V vs. Ag/AgCl)

 

O\ \l
O O
I ' l '

w .5
O O
r 1 '

l B
3 i 6 ‘ 9'12‘15'18‘21‘24
1/2 1/2
Scan Rate (mV )

 

 

 

Oxidation Current (uA/cmz)
N Ur
O C

Figure 5.8 (A) Cyclic voltammetric i-E curves for a conductive diamond powder
electrode (4 h coating time) in 1 mM Fe(CN)63'/4' at scan rates between 10 and 500 mV/s

and (B) a plot of the oxidation peak current versus the scan ratem.

134

More severe distortion of the cyclic voltammetric i-E curves are seen for
Ru(NH3)63+fz+ (10 to 300 mV/s) for a 4 h as-grown diamond power electrode in Figure
5.9A. AEP for Ru(NH3)(r,3+/2+ is high even at low scan rates, ranging from 220 mV at 10
mV/s to 530 mV at 300 mV/s. AEp values of 112 and 91 mV were observed in

comparison measurements of Ru(NH3)63+/2+

with GC and graphite powder electrodes,
respectively, at 10 mV/s. In addition to issues with the ohmic resistance of the diamond
powder electrodes discussed above, limited electrode wettability may also inﬂuence the
45,, recorded for highly charged analytes such as Fe(CN)63'/4' and Ru(NH3),3*’2*. After
diamond growth, the conductive diamond powder is hydrogen terminated. In addition,
the powder is bound with a non-polar polymer binder, rendering the electrode quite
hydrophobic. To investigate the contribution of electrode hydrophobicity on the cyclic
voltammetric curve shape, an electrode was prepared using 4 h conductive diamond
powder and only 5 wt. % PTFE, instead of the 10 wt. % used to fabricate all other
electrodes. Cyclic voltammetric i-E curves recorded for Ru(NH3)63+/2+ on this electrode
are presented in Figure 5.9B. AEp is signiﬁcantly reduced, to 117 mV at 10 mV/s and
167 mV at 300 mV/s. These values are comparable to those recorded for graphite and
GC powder with 10 wt. % binder, electrode materials that are expected to be more
hydrophilic than hydrogen terminated diamond. Thus, it seems that both ohmic
resistance as a result of poor particle-particle connectivity and electrode hydrophobicity
inﬂuence the cyclic voltammetric response for conductive diamond powder when bound
with PT FE. Future work will focus on new approaches to fabricate the powder electrodes
that will improve the particle-particle connectivity, as well as on the use of different

binders to improve electrode wettability.

135

 

  
  
 
 
   
 
  
 
 
  

ON
C
I ' I

 

ll 300 mV/s
A 45: 100 mV/s
NE 30: 70mV/s
15 l. 50 mV/s
O ’ 30 mV/s
2 0‘
r 10 mV/s
1 -15 :-
v
4—0 -30_~
0’ -45 -
E -60 -
5 -75 I
-90 - A
_105r11111411111

 

 

 

-O.6 -O.5 -O.4 -O.3 -O.2 -0.1 0.0 0.1 0.2

Potential (V vs. Ag/AgCl)

 

150- I

_ l 300 mV/s

100 __ 100 mV/s
_ 7O mV/s
50 .. 50 mV/s
. 3O mV/s
10 mV/s

 

Current (uA/cmz)

.L. .1.
Ur o
o o

I

 

 

-200 i- I I I I L I I I
-O.6 -05 -0.4 .03 oz -o.1 0.0 0.1 0.2

Potential (V vs. Ag/AgCl)

 

Figure 5.9 Cyclic voltammetric i-E curves for 1 mM Ru(NH3)63+/2+ for a conductive
diamond powder electrode (4 h coating time) at scan rates between 10 and 300 mV/s with
(A) 10 wt. % PTFE binder and (B) 5 wt. % PTFE binder.

136

5. 2.3 Morphological and Microstructural Stability

In order to evaluate the morphological and microstructural stability of the
conductive diamond powder, the electrodes were anodically polarized at 1.6 V vs.
Ag/AgCl and 25 °C for 1 h. Comparison measurements were made with graphite and GC
powders. High potentials such as this are possible for short duration in fuel cells during
idle operation and can affect the long-term stability and operational lifetime of the
device.24 Figure 5.10 shows the Chronoamperometric i-t curves recorded for each
electrode material, normalized to the electrode area calculated from the cyclic
voltammetric background current as described in Chapter 2. The total charge passed (i.e.,
integrated i-t curve) is largest for the GC powder (0.88 C/cmz), followed by graphite
(0.46 C/cmz) and the conductive diamond powder (0.18 C/cmz). For GC, the current
rises and then decreases quickly after the potential step as the electric double layer forms.
This is followed by a progressive increase in current over time, which is likely caused by
microstructural degradation of GC leading to a continuously increasing exposed surface
area. The i-t curve for the graphite powder follows the same trend, but the current
increase over time is smaller, consistent with less microstructural alterations. This result
is not surprising considering that the degradation of carbon materials is known to be
related to the degree of microstructural disorder, as presented in Chapter 3. More
disordered carbon materials have a higher fraction of exposed edge plane, resulting in
more extensive oxidation and corrosion. Clearly, the lowest charge density is passed for
the diamond powder consistent with a more microstructurally stable material. There is,

however, an increase in current over time.

137

Figure 5.11 shows scanning electron micrographs of the conductive diamond
powder (A) before and (B) after the anodic polarization at 1.6 V vs. Ag/AgCl and 25 0C.
Before polarization, the PTFE binder covers some of the particles with a cauliﬂower-like
texture. After polarization, there appears to be much less PTFE covering the grains,
although upon close inspection, the polymer still exists between the particles. Magniﬁed
images of individual diamond particles showed the same diamond grain morphology and
texture before and alter anodic polarization. There was no visible morphological or
microstructural damage to the conductive diamond powder aﬁer polarization under these
conditions. Therefore, it is concluded that the increase in the current seen for diamond
was caused by exposure of more surface area over time as the PTFE binder delaminated
from some of the grains. In contrast, SEM images for GC (C) before and (D) after anodic
polarization in Figure 5.11 reveal severe morphological and microstructural degradation
of some particles. Degradation of this nature in fuel cells would be catastrophic to the
electrode performance because electrocatalyst particles from these areas would be lost
and the electrical resistance would increase. The graphite powder particles are ﬂat,
irregularly shaped, and not uniform in size. Therefore, SEM images of graphite powder
are not presented because the morphology of the particles was such that the degradation

was not easily detected.

138

LII
O
O

 

 

     
 

 

 

A
N
E 400 r
O Glassy Carbon Powder
2:1 300 - \
v
E 200 F
0 .
/
E 100 L/L/
U W
0 _ Conductive Diamond Powder
0 600 1200 1800 2400 3000 3600

Figure 5.10 Chronoamperometric i-t curves for the 4 h conductive diamond powder,
graphite powder, and glassy carbon powder electrodes recorded during potentiostatic

Time (s)

polarization at 1.6 V vs. Ag/AgCl and 25 °C for 1 h.

 

Figure 5.11

0.1 M HC104.

Scanning electron micrographs of a conductive diamond-coated diamond
powder electrode (4 h coating time) (A) before and (B) after and GC powder electrode
(C) before and (D) after constant potential polarization of 1.6 V vs. Ag/AgCl for 1 h in

139

5.3 Conclusions

The physical, electrical, and electrochemical properties of a boron-doped diamond
powder have been described. Even though the surface area and conductivity are lower
than that desired for an ideal fuel cell support, these powders represent a ﬁrst step toward
the desired values. The conductive powder was formed by coating a boron-doped
overlayer on insulating diamond powder via chemical vapor deposition. SEM, Raman
spectroscopy, electrical measurements, and electrochemical tests were used to
characterize the material. The results conﬁrm that a boron-doped diamond overlayer was
grown on the powder surface with co-deposition of some nondiamond carbon impurity.
Stress was also introduced into the powder at long deposition times (6 h) due to the
particle fusion. Electrical resistance measurements proved that the powders became
electrically conducting aﬁer being coated with the boron-doped diamond overlayer. The
nondiamond carbon impurity was also shown to contribute some to the conductivity, but
its effect was much less for the 4 and 6 h deposition times. i—V curves for the packed, dry
powder were ohmic between :tIO V. Electrochemical tests revealed the presence of the
nondiamond carbon at the surface of the powder and that the powder electrodes (with
binder) were responsive to Fe(CN)63'/4' and Ru(NH3)63+/2+. The cyclic voltammetric
curves were, however, distorted by ohmic resistance effects attributed to poor particle-
particle connectivity within the binder and poor wettability. Amperometric i-t curves for
the polarization of conductive diamond, graphite, and GC powders at 1.6 V vs. Ag/AgCl
and 25 °C showed that the lowest anodic charge was passed for diamond. SEM images of

the diamond powder before and after polarization showed no evidence of morphological

140

degradation or corrosion. In contrast, signiﬁcant morphological changes were found for

the GC powder.

141

5.4

(1)

(2)

(3)

(4)

(5)
(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

References
Swain, G. M. J. Electrochem. Soc. 1994, 141, 3382-3393.

Chen, Q.; Granger, M. C.; Lister, T. E.; Swain, G. M. J. Electrochem. Soc. 1997,
144, 3806-3812.

Dolrnatov, V. Y. Russ. Chem. Rev. 2001, 70, 607-626.

Kinoshita, K. Carbon: Electrochemical and Physicochemical Properties; John
Wiley & Sons: New York, 1988.

Bergman, L.; Nemanich, R. J. J. Appl. Phys. 1995, 78, 6709-6719.

Mermoux, M.; Marcus, B.; Swain, G. M.; Butler, J. E. J. Phys. Chem. B 2002,
106, 10816-10827.

Yoshikawa, M.; Katagiri, G.; Ishida, H.; Ishitani, A. Appl. Phys. Lett. 1989, 55,
2608-2610.

Granger, M. C.; Witek, M.; Xu, J .; Wang, J .; Hupert, M.; Hanks, A.; Koppang, M.
D.; Butler, J. E.; Lucazeau, G.; Mermoux, M.; Strojek, J. W.; Swain, G. M. Anal.
Chem. 2000, 72, 3793-3804.

Fagan, D. T.; Hu, I.-F.; Kuwana, T. Anal. Chem. 1985, 5 7, 2759-2763.

Show, Y.; Witek, M. A.; Sonthalia, P.; Swain, G. M. Chem. Mater. 2003, 15, 879-
888.

Witek, M.; Wang, J .; Stotter, J .; Hupert, M.; Haymond, S.; Sonthalia, P.; Swain,
G. M.; Zak, J.; Chen, Q.; Gruen, D. M.; Butler, J. E.; Kobashi, K.; Tachibana, T.
J. Wide Bandgap Mater. 2001, 8, 171-189.

Wang, J .; Swain, G. M.; Mermoux, M.; Lucazeau, G.; Zak, J .; Strojek, J. W. New
Diam. Front. C. Tech. 1999, 9, 317-343.

Fischer, A. E.; Show, Y.; Swain, G. M. Anal. Chem. 2004, 76, 2553-2560.

Bennett, J. A.; Wang, J .; Show, Y.; Swain, G. M. J. Electrochem. Soc. 2004, I51,
E306-E3l3.

Kneten, K. R.; McCreery, R. L. Anal. Chem. 1992, 64, 2518-2524.

142

(16)

(17)

(18)

(19)

(20)

(21)

(22)

(23)

(24)

Rice, R. J .; Pontikos, N. M.; McCreery, R. L. J. Am. Chem. Soc. 1989, 112, 4617-
4622.

Chen, P.; Fryling, M. A.; McCreery, R. L. Anal. Chem. 1995, 6 7, 3115-3122.

Bard, A. J .; Faulkner, L. R. Electrochemical Methods: Fundamentals and
Applications, Second ed.; John Wiley & Sons, Inc.: New York, 2001.

Gerischer, H. Electrochim. Acta. 1990, 35, 1677-1699.

Hupert, M.; Muck, A.; Wang, J .; Stotter, J .; Cvackova, Z.; Haymond, S.; Show,
Y.; Swain, G. M. Diam. Rel. Mater. 2003, 12, 1940-1949.

Granger, M. C.; Swain, G. M. J. Electrochem. Soc. 1999, 146, 4551-4558.

Yagi, I.; Notsu, H.; Kondo, T.; Tryk, D. A.; Fujishima, A. J. Electroanal. Chem.
1999, 473, 173-178.

Swain, G. M. In Electroanalytical Chemistry; Bard, A. J., Rubinstein, I., Eds;
Marcel Dekker: New York, 2003; Vol. 22, pp 181-277.

Report of the Basic Energy Sciences Workshop on Hydrogen Production, Storage,
and Use, Argonne National Laboratory: Argonne, IL, 2003.

143

Chapter 6

Preparation and Characterization of Boron-Doped
Diamond Paper with Pt Nanoparticles

Abstract

Nanocrystalline boron-doped diamond (BND) has been successfully coated on
carbon paper (TCP) via microwave plasma-assisted chemical vapor deposition. Scanning
electron microscopy revealed that ﬁbers within the ﬁrst ~100 pm of the TCP surface
were fully coated with nanocrystalline diamond (ﬁlm thickness ~l um). X-ray
diffraction and Raman spectroscopy conﬁrmed that TCP had been coated with a
crystalline diamond overlayer. Two different (A and B) currents were used to
electrodeposit Pt nanoparticles on bare and diamond-coated TCP using a pulsed
galvanostatic method (Pt/TCP and Pt/BND). Under condition A (0.146 mA/cm2 or
0.0047 C/cmz), large, widely dispersed Pt particles were deposited on TCP (Pt/TCP-A).
Under these same conditions, nanoparticles with a diameter of 177 :t: 39 nm and a density
of ~1.2 x 109 particles/cm2 were deposited on BND (Pt/BND-A). Under condition B
(1.25 mA/cm2 or 0.040 C/cmz) Pt nanoparticles formed on both TCP (Pt/TCP-B, 233 :L-
89 mm diam and ~2.7 x 108 particles/cmz) and BND (Pt/BND-B). The Pt particles on
Pt/BND-B were so dense and closely spaced that an accurate measure of the particle size
and density could not be obtained. The composite electrode surface enhancement factors
(SEF, Pt surface area/electrode surface area) were 0.033, 0.50, 3.3, and 1.7 for Pt/TCP-A,
Pt/BND-A, Pt/TCP—B, and Pt/BND-B, respectively. The speciﬁc Pt surface areas (Spt,

mZ/g) were 1.4, 21, 16, and 8.2 for Pt/TCP-A, Pt/BND-A, Pt/TCP-B, and Pt/BND-B,

144

respectively. The oxygen reduction reaction was used to probe the electrocatalytic
activity of the composite electrodes. The most active electrode was Pt/BND-A with a
half-wave potential, Ep/z, of 0.583 V vs. Ag/AgCl and a limiting current of ca. -1000
uA/SEF. The stability of the bare TCP and BND substrates and the composite Pt/TCP
and Pt/BND electrodes were studied using potentiostatic polarization at 1.4 and 1.6 V vs.
Ag/AgCl (25 °C) in 0.1 M HClO4. The BND electrode exhibited superior morphological
and microstructural stability over TCP at 1.6 V. Evidence was found for dissolution and
redeposition of Pt after polarization of the composite electrodes at 1.4 V, particularly for

Pt/TCP.

145

6.1 Introduction

As discussed in Chapter 1, the hydrogen oxidation and oxygen reduction reactions
occur on an electrocatalyst (e.g., Pt), which is supported on high-surface-area carbon
within the catalyst layer of the PEMFC. The catalyst layer is sandwiched between two
pieces of carbon paper that make up part of the gas diffusion layers (GDLs), though
which the reactant gases are distributed evenly into the reactive zone. The ﬁrst part of
our work to develop a dimensionally stable electrocatalyst support material, presented in
Chapter 5, focused on the development and characterization of electrically conducting
diamond powder as an alternative to Vulcan XC-72 in the catalyst layer. Although the
diamond powder exhibited greater dimensional stability with respect to graphite and
glassy carbon powders, ohmic resistance of the electrode when using
polytetraﬂuoroethylene as a binder limited evaluation of its properties as an
electrocatalyst support. An alternative approach addressed in this chapter focuses on the
development of a binderless diamond material that is based on the morphology of the
porous GDL.

PEMFC GDLs are typically made of a spz-bonded carbon paper backing layer
(BET ~9 mZ/g), a porous material composed of interconnected carbon ﬁbers. To be
effective in the fuel cell, carbon paper GDL materials must be electrically conductive, gas
permeable, and thin (~01 to 0.3 mm), with high mechanical strength. The
electrochemical stability requirements for carbon paper in the GDL are not as demanding
as those for carbons in the catalyst layer, mainly because the electrocatalytic activity of

the fuel cell is not dependent on GDL stability. Thus, corrosion of carbon paper when

146

 

used in the GDL is not as detrimental to fuel cell operating efﬁciency and lifetime as is
corrosion of the electrocatalyst support material.

In this second approach to develop high-surface-area diamond as an
electrocatalyst support material, we propose to remove the catalyst layer completely and
use the porous carbon paper substrate directly as the electrocatalyst support. Used in this
manner, the carbon paper stability requirements are similar to those of the catalyst layer
itself, discussed in Chapter 1. We showed in Chapters 3 and 5 that spz-bonded carbons
corrode under high potentials (1.2 to 1.5 V vs. SHE) present during fuel cell start-up,
shut-down, and idle operation. Thus, to increase the corrosion resistance of the porous
carbon paper for this application, we have coated the individual carbon ﬁbers of a carbon
paper GDL with a nanocrystalline boron-doped diamond overlayer. Used in this way, the
diamond coating may not only impart a higher degree of microstructural stability and
corrosion resistance to the paper, but the improved support stability may also translate
into improved electrocatalytic activity and stability of the metal-coated electrode toward
the oxygen reduction reaction.

A number reports have been published recently that discuss the deposition of
insulatingl’2 and electrically conducting (i.e., boron-doped)“5 microcrystalline diamond
on carbon ﬁbers for a variety of applications. There area a few drawbacks to this method,
however, in terms of the development of a material for fuel cell electrodes. First, high
atomic hydrogen concentration in microcrystalline diamond deposition plasmas results in
signiﬁcant gasiﬁcation of the underlying spZ-bonded carbon ﬁber. This alters the carbon
paper morphology and increases its brittleness, resulting in decreased mechanical

strength. Second, coating the carbon ﬁber with large (~l um), rough microcrystalline

147

diamond grains would likely decrease the porosity and surface area of the carbon paper.
Coating the carbon paper with nanocrystalline boron-doped diamond, however, can
potentially overcome both of these issues. The gas composition for nanocrystalline
diamond grth has little to no atomic hydrogen, which would should reduce the effects
of carbon ﬁber gasiﬁcation. Moreover, the smooth texture and small grain size (~10-15
nm) of a nanocrystalline diamond overcoat would maintain the original carbon paper
morphology and porosity.

In this chapter, the preparation and characterization of nanocrystalline boron-
doped diamond deposited on carbon paper (BND) is described. Although the surface
area of this material (BET ~9 mz/g) is lower than that needed for a viable electrocatalyst
support, it provides a useful starting point for evaluating the properties of a high-surface-
area boron-doped diamond electrocatalyst support material. Scanning electron
microscopy, X-ray diffraction, and Raman spectroscopy were used to characterize the
morphology, crystallinity, and microstructure of the bare (TCP) and diamond-coated
carbon paper. BND and TCP were decorated with Pt nanoparticles by pulsed
galvanostatic deposition. Cyclic voltammetry and chronoamperometry were used to
evaluate the dimensional stability and corrosion resistance of the bare and diamond-
coated carbon paper subjected to potentiostatic polarization, as described in Chapter 3.
The electrocatalytic activity of the Pt/carbon composite electrodes towards the oxygen
reduction reaction was studied via linear sweep voltammetry in oxygen-saturated 0.1 M

HClO4 before and aﬁer potentiostatic polarization.

148

6.2 Results and Discussion

6.2.] Nanocrystalline Diamond Deposition on Carbon Paper
6.2.1.1 Morphology and Microstructure

Scanning electron micrographs of the bare (TCP) and diamond-coated carbon
paper (BND) are presented in Figure 6.1A-D. As seen in Figure 6.1A and B, TCP is
composed of interconnected polyacrylonitrile ﬁbers that form a porous ﬁber network.
The individual ﬁbers are about 5 pm in diameter and each ﬁber has striations along the
longitudinal axis. Aﬁer nanocrystalline diamond deposition, all ﬁbers within the ﬁrst
~100 pm of the carbon paper surface are fully coated with no evidence of cracks,
pinholes, or voids. As showniin Figure 6.1C, BND has a relatively smooth texture with
randomly-oriented diamond grains. The linear striations of the TCP substrate have been
completely covered by the diamond. The cross-sectional image of a coated carbon ﬁber,
presented in Figure 6.1D, reveals that the diamond coating completely covers the ﬁber
with a thickness of ~1 pm. There was little change in the diameter of the carbon ﬁber, as
seen in the center of the cross-section, suggesting that the small amount of atomic

hydrogen present during growth did not signiﬁcantly gasify the TCP substrate.

149

 

Figure 6.1 Scanning electron micrographs of (A) bare carbon paper, (B) an individual
bare carbon ﬁber within the paper, (C) boron-doped nanocrystalline diamond deposited
on the carbon paper, and (D) a cross-section of a nanocrystalline diamond-coated carbon
ﬁber.

X-ray diffraction (XRD) spectra for TCP and BND from 29 of 10 to 100 degrees
are presented in Figure 6.2. The insets show enlarged views of the spectra between 40
and 100 degrees. The XRD spectrum for TCP is characterized by the reﬂections
expected for spZ-bonded carbon at ~26.5, 43.2, 54.4, 77.7, and 86.9 degrees, which are
assigned to the (002), (100), (004), (110), and (006) crystallographic orientations of
graphite, respectively.7 These reﬂections are denoted in Figure 6.2 with a "G". The XRD

spectrum for BND also has reﬂections for cubic diamond (denoted with a "D") at 43.9,

150

75.3, and 91.5 degrees, which are assigned to the (111), (220), and (311) crystallographic
orientations (ASTM 6-0675). The diamond spectrum also exhibits the graphite

reﬂections seen for TCP. Since XRD is a bulk technique and only TCP ﬁbers within the
top ~100 pm of the paper were coated with diamond, it is not surprising that spZ-bonded

carbon reﬂections are also seen for the diamond-coated material.

 

 

G (002) BND

G (100)G (004) G (006)

 

 

 

 

 

 

 

f? D (1 11) D (3‘ 1)
:3 / D (220) /
O \
G3 )| 1 . . . .
V 40 30 6404 70 80 90 100
g G (002) TCP
L1 G (004)
3 G (100)
E G (006)

 

 

 

 

G 11
/ (Q /

40 5‘0 60 7‘0 80 90 100
. .4

I 1 I 1 I 1 I 1 I 1 I 1 I

10 20 3O 4 50 6O 7O 80 90 100

0
Two Theta (Degrees)

Figure 6.2 X-ray diffraction spectra of bare (TCP) and boron-doped nanocrystalline
diamond-coated carbon paper (BND).

 

 

 

 

As discussed in previous chapters, Raman spectroscopy is a useful tool for
studying sp2- and sp3-bonded carbons because it is sensitive to both short and long-range
microstructural order. The Raman spectrum for TCP presented in Figure 6.3 contains

two peaks at 1582 cm'1 and 1360 cm], which correspond to the E2g and D modes of

151

amorphous spZ-bonded carbon, respectively, as discussed in Chapter 38’12 Brieﬂy, the
presence and intensity of the D band suggests that TCP has some microstructural disorder
(i.e., some fraction of exposed edge plane density) with respect to crystalline graphite.
The Raman spectrum for BND is characterized by broad peaks at 1150, 1333, 1470, and
1550 cm], consistent with that for high-quality, nanocrystalline boron-doped diamond.13 '
'7 The origin of these peaks was discussed in detail in Chapter 4. The spectrum for BND
was similar at different points on the electrode surface with no evidence of spz-bonded

carbon.

 

 

 

 

 

BND
”T
:1
cc
v
3? J J A . TCP
m
:
<1)
.11.)
s:
b—t
1000 1100 1200 1300 1400 1500 1600 1700 1800

Raman Shift (cm'l)

Figure 6.3 Raman spectra for the bare (TCP) and boron-doped nanocrystalline
diamond-coated carbon paper (BND). '

152

6.2.1.2 Electrochemical Characterization

Background cyclic voltammetric i-E curves for the TCP and BND electrodes in
0.1 M HClO4 are presented in Figure 6.4. The current for each electrode was stable with
potential cycling between -0.4 and 1.0 V vs. Ag/AgCl (~40 cycles) and is normalized to
the solution-exposed electrode area. It is important to note that although the geometric
surface areas of TCP and BND were similar, the electrode surface area of TCP exposed
to the solution, which was calculated from the background cyclic voltammogram (see
Chapter 2), was smaller than that of BND calculated in the same manner. Although the
reason for this is not clear at this time, it may be due, in part, to electrode wettability
differences. It seems that only the ﬁrst few layers of carbon ﬁbers in the TCP electrode
were contacted by the solution, while the solution penetrated more deeply into the BND
electrode. Future work will include strategies to maximize electrode wettability so that
the electrode area exposed to solution is more consistent for both materials.

The voltammogram for TCP is characterized by a non-faradaic (i.e., capacitive)
background current that is about two times larger than that of BND, after normalization
to the exposed electrode surface area. As discussed in Chapter 4, the voltammetric
background current is typically smaller for sp3-bonded diamond compared to spz-bonded
carbon electrodes because of a lower capacitance and less pseudocapacitance from
electroactive surface carbon-oxygen functionalities. Faradaic current at ca. 400 mV can
be seen for the TCP electrode, which is associated with electroactive surface oxygen
functional groups at edge plane and defect sitesm’19 Faradaic current is also observed for
the BND electrode in this region, although the current magnitude is much lower. It is

likely that the background current for BND is inﬂuenced by both the diamond-coated and

153

the uncoated ﬁbers in the interior of the carbon paper. Thus, the faradaic current ca. 400
mV for BND is likely caused by electroactive surface-oxygen functional groups on the
spz-bonded TCP. The cathodic current at ~-O.4 V vs. Ag/AgCI arises from oxygen
reduction on both TCP and BND. Although the solution was degassed with N2 for 30
min prior to all electrochemical measurements, it is possible that oxygen was still present
within the porous electrode. Hydrogen evolution for TCP and BND does not occur until

ca. -1 V vs. Ag/AgCl.

 

 

 

 

 

 

 

 

15
g 5: _______ "II E...
0
:3. - I J‘
:j -5 ------ u, -- -------- ’l
a ”
t1
:3 -15
U 20 _BND
- --—rc1>

 

 

 

€04 ‘ 0.2 - 0:0 ‘ 0:2 . 0T4 l 0:6 1 0:8 ‘ 1.0
Potential (V vs. Ag/AgCl)

Figure 6.4 Background cyclic voltammetric i-E curves for bare (TCP) and boron-
doped nanocrystalline diamond-coated carbon paper (BND) in 0.1 M HClO4 at 100 mV/s.
The current is normalized to the electrode area estimated to be in contact with the
solution.

154

6. 2.2 Electrodeposition of Pt on T CP and BND

Electrodeposition of Pt onto planar boron-doped diamond thin-ﬁlm electrodes has
been accomplished using potentiodynamic, galvanostatic, and pulsed galvanostatic
depositionmll Pulsed galvanostatic deposition produced Pt nanoparticles with nominal
diameters that were an order of magnitude smaller than those deposited by constant
potential or current methods.20 Although these Pt/diamond composite electrodes are
active for the oxygen reduction reaction, the low surface area of the diamond support
renders them impractical for use in fuel cells. Moreover, the planar electrodes are not
porous and thus, cannot serve as gas diffusion electrodes. Carbon paper, currently used
as a gas difﬁrsion layer in fuel cells is higher in surface area (BET ~9 mz/g) than the
planar ﬁlms and is gas permeable. The diamond-coated carbon paper is also both gas
permeable and higher in surface area than the planar ﬁlms (BET ~8.5 mz/g). Although
the surface area of this material is not as high as that required for fuel cells (~100 mZ/g),
its fabrication and availability is an important step toward our long-term goal of
developing dimensionally stable, high-surface-area, gas diffusion Pt/diamond composite
electrodes.

Two different pulsed galvanostatic current/charge conditions (denoted as A and
B) were used to electrodeposit Pt on the TCP and BND gas difﬁrsion electrodes, as
shown in Table 6.1. To compare the resultant Pt nanoparticles for a given deposition
condition, it is important that the same current density be applied to each electrode.
Thus, accurate determination of the electrode surface area (TCP or BND) is essential. As
discussed above, there were large differences in the TCP and BND surface areas exposed

to solution because of possible differences in electrode wettability. Although this surface

155

area may change slightly throughout the course of the electrochemical characterization
(i.e., increased electrode wetting), it is the most accurate means of comparing the
electrodes. Electrode geometric area reﬂects the surface area of the entire electrode,
including portions that are not exposed to solution (i.e., portions on which Pt cannot be
deposited). Thus, normalization to the electrode geometric area will not provide an
accurate comparison of regions on which Pt was deposited. However, comparison of the
actual surface area (calculated from the mass of the electrode and the BET surface area)
to the solution-exposed surface area is useful to provide some insight into the electrode
wettability. Table 6.1 presents the solution-exposed surface area (used for all
normalization) that was calculated from the background cyclic voltammetric i-E curve, as
well as the geometric surface area that was calculated from the mass of the electrode and
the electrode BET surface area.

All Pt/carbon composite electrodes (Pt/TCP and Pt/BND) were exposed to 1 s
pulses with a 25% duty cycle (i.e., 200 ms on-time and 800 ms off-time). First, 2.5 mA
(a total of 0.080 C) was applied to a BND electrode (condition A) to electrodeposit Pt.
The normalized current and charge were calculated to be 0.146 mA/cm2 (a total of 0.0047
C/cmz), based on the BND solution-exposed surface area. The same normalized current
and charge was then applied to a TCP electrode to deposit Pt. These conditions were
chosen as a starting point because the total current and charge applied to the BND
electrode were similar to values used in previous work for Pt deposition on diamond thin
ﬁhn via pulsed galvanostatic deposition.20 Second, 1.25 mA/cm2 (a total of 0.040 C/cmz)
was applied to both electrode types (condition B). On planar microcrystalline and

nanocrystalline boron-doped diamond thin-ﬁlm electrodes, this current density resulted in

156

the most active Pt/diamond composite electrode in terms of particle diameter, coverage,

and speciﬁc Pt surface area (mZ/g).20

Table 6.1 Two different (A and B) current and charge conditions for the pulsed
galvanostatic deposition of Pt metal on carbon paper (TCP) and boron-doped
nanocrystalline diamond deposited on carbon paper (BND) including the solution-
exposed and actual surface area (SA).

 

 

 

current Charge S°'““°“' Actual SA ATOtliaeld ATOtlileld
Condition Density Density Electrode Exposed 2 pp pp
(mA/cm’) (C/cmz) SA (cm’) (cm) Current Charge
(mA) (mC)
TCP-A 0.452 650 0.066 0.002
A 0.146 0.0047
BND-A 16.9 884 2.5 0.08
TCP-B 1.32 703 1.65 0.05
B 1.25 0.040
BND-B 7.64 935 9.56 0.30

 

Scanning electron micrographs of the Pt/carbon composite electrodes, Pt/TCP-A,
Pt/BND-A, Pt/TCP-B, and Pt/BND-B, deposited under the conditions A and B of Table
6.1, are presented in Figure 6.5. At 0.146 mA/cm2 or 0.0047 C/cmz, rather large Pt
deposits were formed (circled areas) on Pt/TCP-A that were widely dispersed across the
electrode surface. Under these same conditions, Pt metal nanoparticles were deposited on
the Pt/BND-A electrode with a particle diameter of 177 i 39 nm and density of ~1.2 x
109 particles/cmz. The particles were unifonnly dispersed across the electrode surface
and Pt metal was observed up to a depth of ~100 pm (as far as down in the sample as the
electron beam could be focused). At 1.25 mA/cm2 or 0.040 C/cmz, Pt nanoparticles were
deposited onto both Pt/TCP-B and Pt/BND-B, as shown in Figure 6.5. Deposition of Pt

nanoparticles on Pt/TCP-B was. limited to the outermost ﬁbers, presumably because only

157

these ﬁbers were wetted by the solution. Particle analysis revealed that the Pt/TCP-B
nanoparticles were 233 :1: 89 nm, with a density of ~2.7 x 108 particles/cmz. Under the
same conditions, Pt nanoparticles smaller in size, with a much higher density were
deposited onto Pt/BND-B. Although the particle size, coverage, and density look similar
for Pt/BND-A and Pt/BND-B on the scale of the SEM presented, Pt/BND-B
nanoparticles had a much higher particle density, such that there was little open space
between particles. In fact, the particles were so closely spaced that accurate particle
analysis could not be completed. The Pt particle size and density on BND under both

conditions A and B were superior to that deposited on TCP under condition B.

Filter—1 "i ‘

l’t/IJND-‘B-

 

Figure 6.5 Scanning electron micrographs of Pt/carbon composite electrodes
(Pt/TCP-A, Pt/BND—A, Pt/TCP-B, and Pt/BND—B under two different pulsed
galvanostatic deposition conditions A (0.146 mA/cm or 0.0047 C/cmz) and B (1.25
mA/cm2 or 0.040 C/cmz).

158

Background cyclic voltammetric i-E curves of Pt/TCP-A and Pt/BND-A in 0.1 M
HClO4 at 50 mV/s are presented in Figure 6.6A. The corresponding curves for Pt/TCP-B
and Pt/BND-B are shown in Figure 6.6B. The voltammograms were stable after ca. 10
cycles with cycle number 15 is shown here. All currents are normalized to the solution-
exposed area of the original electrode (TCP or BND), presented in Table 6.1. The curves
for Pt/BND-A, Pt/TCP-B, and Pt/BND-B exhibit voltammetric features associated with
clean, polycrystalline Pt. In contrast, the curve for Pt/TCP-A seems to resemble TCP
without Pt metal. Close inspection of the Pt/TCP-A voltammogram does reveal the
characteristic Pt voltammetric features. However, the signal for this electrode is
dominated by current from the original TCP substrate because the amount of Pt is low.
These cyclic voltammetric results, coupled with the SEM images presented in Figure 6.5,
show that Pt can be electrodeposited on both TCP and BND, that the Pt nanoparticles are
in electrical contact with the carbon substrate, and that the composite electrode is
electrochemically active.

From the curves in Figure 6.6B, it seems that the active Pt surface area is larger
for Pt/TCP-B than for Pt/BND-B. It should be reemphasized, however, that these curves
were normalized to the solution-exposed carbon electrode surface area (TCP or BND)
that were presented in Table 6.1. Considering the signiﬁcant difference in the TCP and
BND surface area, a more accurate analysis would be to normalize the Pt surface area to
the electrode surface area to give the dimensionless surface enhancement factor (SEF),22

as shown in Equation 6.1

SEF =

 

SA
P’ (6.1)
SAC

159

where SA Pt is the electrochemically-active Pt surface area (cm2) and SAC is the carbon
(TCP or BND) substrate surface area (cmz) exposed to solution. SA Pt can be estimated

from the cyclic voltammetric charge associated with the adsorption (cathodic) or
desorption (anodic) of H+ between 0.15 and -0.02 V vs. Ag/AgCl, assuming a 1H21Pt
surface interaction and a charge of 210 uC/cm2 for hydrogen adsorption on a ﬂat Pt (11 l)
surface.23’24 Pt surface areas for TCP-A, BND-A, TCP-B, and BND-B, as well as the
SEFs are presented in Table 6.2. For condition A, 0.146 mA/cm2 or 0.0047 C/cmz, the Pt
SEF for BND (0.50) is a factor of 10 larger than that for TCP (0.033). In contrast, the Pt

SEF for TCP (3.3) and BND (1.7) is more similar under condition B, 1.25 mA/cm2 or

0.040 C/cmz.

Table 6.2 Electrode and Pt surface areas (SA) calculated for the Pt/carbon composite
electrodes (Pt/TCP-A, Pt/BND-A, Pt/TCP-B, and Pt/BND-B) with the metal phase
formed by pulsed galvanostatic conditions A (0.146 mA/cm2 or 0.0047 C/cmz) and B
(1.25 mA/cm2 or 0.040 C/cmz) with the corresponding surface enhancement factor, SEF.

 

 

 

Electrode E'ecgg‘lf SA 2:15? SEF
Pt/TCP-A 0.452 0.015 0.033
Pt/BND-A 16.9 8.3 0.50
Pt/TCP-B 1.32 4.3 3.3
Pt/BND-B 7.64 13 1.7

 

160

 

N
C

1'6
o

 

Current (uA/cmz)

 

 

 

-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Potential (V vs. Ag/AgCl)

 

 

400
. — Pt/BND-B

200 L - - - Pt/TCP-B /,

 

1'9

0

O
l

 

Current (uA/cmz)
is

1 l
-600 t ' B
-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Potential (V vs. Ag/AgCl)

Figure 6.6 Cyclic voltammetric i-E curves of Pt/carbon composite electrodes Pt/TCP-
A, Pt/BND-A, Pt/TCP-B and Pt/BND-B deposited by pulsed galvanostatic deposition (A
= 0.146 mA/cmz or 0.0047 C/crrr2 and B = 1.25 mA/cmz or 0.040 C/cmz) in 0.1 M Helo4
at 50 mV/s. Currents are normalized to the carbon electrode surface area.

1

 

 

 

161

 

The speciﬁc surface area of Pt (Spt, mZ/gpt) in fuel cell electrodes is an important
parameter that affects both the efﬁciency and cost. SP1 values of ~100 mZ/g are typical in
PEMFCs.25 The mass of Pt deposited (i.e., loading) can be estimated by assuming a
deposition (i.e., coulometric) efﬁciency of 100% for the 4 electron reduction of Pt4+ to Pt0
according to Reaction 6.1:

PtC162' + 4c «- Pt° + cor (6.1)
The area is determined from the voltammetric charge for H” adsorption or desorption.
This estimated value of Spt really represents a lower limit on TCP and BND as the actual
loading is likely higher because the efﬁciency is probably not 100%. Estimated SP1
values for the pulsed galvanostatic deposition of Pt on TCP and BND electrodes, Pt/TCP-
A, Pt/BND-A, Pt/TCP-B, and Pt/BND-B, are presented in Table 6.3. The largest Sp.
value (21 mz/g) was calculated for Pt/BND-A. Although this is about 5 times less than
Sp! in fuel cells, this lower limit may be improved by further optimization of the
electrodeposition (i.e., reduction of particle size, increase of particle density, and increase
of Pt activity).
Table 6.3 Estimated speciﬁc surface area of Pt on Pt/TCP-A, Pt/BND-A, Pt/TCP-B,
and Pt/BND-B formed using pulsed galvanostatic deposition conditions A (0.146

mA/cm2 or 0.0047 C/cmz) and B (1.25 mA/cm2 or 0.040 C/cmz) assuming 100%
deposition efﬁciency.

 

 

 

Electrode Speﬁf;12::27gr)face
Pt/TCP-A 1 .4
Pt/BND-A 21
Pt/T C P-B 16
Pt/BND-B 3. 2

 

162

To be viable for fuel cell use, the Pt/carbon composite electrodes must exhibit
high activity for the oxygen reduction reaction (ORR), as discussed in Chapter 1. SEF-
normalized linear sweep voltammetric i—E curves (ﬁrst scan) of Pt/TCP-A, Pt/BND-A,
Pt/TCP-B, and Pt/BND-B in oxygen-saturated (30 min 02 bubble) 0.1 M HClO4 at 1
mV/s are shown in Figures 6.7A and B. The oxygen reduction current recorded for each
electrode is only associated with the reduction occurring at Pt sites because both sp2- and
sp3-bonded electrodes are known to have a large overpotentials for this reaction in acidic
media. Sigrnoidal curve shapes are generally observed for the OR on bare Pt

microelectrodes, indicative of pseudo-steady-state behavior.”29

However, when the
concentration of oxygen at the electrode surface is higher than in bulk solution, such as
when Pt is coated with a Naﬁon® ﬁlm that preconcentrates oxygen, a peak is observed in
the OR curve.29 This peak corresponds to the reduction of a high concentration of
oxygen and is followed by the typical pseudo-steady-state response as solution oxygen is
reduced.29 Thus, we suppose that the peaks in these curves are a result of oxygen
preconcentration within the electrode during the 30 min 02 bubble. It should be noted
that a second scan results in the expected pseudo-steady-state behavior without a large
reduction peak. Irnportantly, the onset and half-wave (Hp/2) potentials for oxygen
reduction, as well as the reduction charge, do not change.

By; for the OR (V vs. Ag/AgCl and SHE) as well as the approximate limiting
peak current (i.e., pseudo-steady-state), normalized to both SEF (11A) and mass of Pt
(mA/mgpt), are presented in Table 6.4. As discussed in Chapter 1, the standard potential

for oxygen reduction is 1.23 V vs. SHE, or ca. 1.0 V vs. Ag/AgCl. If ohmic resistance

effects are minimal, as in these experiments, cathodic shifts from this standard potential

163

are indicative of sluggish electron transfer kinetics. Although the kinetics of the OR
reaction were not studied in detail in these experiments, the shift in Ep/z from 0.616 mV
vs. Ag/AgCl for Pt/BND-B to to 0.567 mV vs. Ag/AgCl for Pt/TCP-B indicates that the
electron-transfer rate may be enhanced on the BND electrode. As shown in Table 6.4,
there is also a large difference in the SEF and Pt mass-normalized ORR limiting currents
(i,,-,,,) for the electrodes. Most notably, the Pt/BND-A electrode had a limiting current of
ca. -1000 uA/SEF (-11.6 mA/mgpt), while a current for the Pt/TCP-A electrode,

deposited under similar conditions, could not be evaluated.

Table 6.4 Peak position (Em) and approximate limiting current (in-m) for the oxygen
reduction reaction on Pt/TCP-A, Pt/BND-A, Pt/TCP-B, and Pt/BND-B in oxygen-
saturated 0.1 M HClO4 at 1 mV/s normalized to the surface enhancement factor (inn, SEF,
uA/SEF) and to the mass of Pt (inm, mass, mA/mgpt).

 

 

 

Ep/z Ep/z i11m, SEF illm, mass
Electmde (v vs Ag/AgCl) (v vs SHE) (uA/SEF) (mA/mgpt)
Pt/TCP-A -- -- -- --
Pt/BND-A 0.583 0.780 -1000 -11.6
Pt/TCP-B 0.567 0.764 -130 -270
Pt/BND-B 0.616 0.813 -250 -15.1

 

164

500
0
-500
-1000
-1500
-2000
-2500
-3000
-3500
-4000
-4500

Current (uA/SEF)

0

-200
-300

Current (MA/SEF)

-700

0.0 0.2

 

 

4".--
'
1—
..
sh~"'"‘
’
‘—
’—

' 1

 

 

I r1 I rlﬁ r V

-

r —Pt/BND-A
' - - —Pt/TCP—A

 

 

A. l. L

A

 

 

0.4 0.6

0.8

Potential (V vs. Ag/AgCl)

1.0

 

-100

-400
-500
-600 -

 

 

_ — Pt/BND-B

 

_ - - -Pt/TCP-B

B

 

 

0.0 0.2 0.4 0.6

0.8

Potential (V vs. Ag/AgCl)

1.0

Figure 6.7 Linear sweep voltammetric i-E curves for the oxygen reduction reaction at
Pt/TCP-A, Pt/BND-A, Pt/TCP-B, and Pt/BND-B in oxygen-saturated 0.1 M HClO4 at 1

mV/s. Current is normalized to the surface enhancement factor (SEF).

6. 2.3 Evaluation of WT CP and Pt/BND Stability

The microstructural stability of the bare and Pt/carbon composite electrodes was

evaluated before and after anodic polarization, as presented for the carbon powder

165

electrodes in Chapter 3. First, Chronoamperometric i—t curves were recorded for metal-
free TCP and BND held at 1.4 and 1.6 V vs. Ag/AgCl and 25 °C in 0.1 M HClO4 for 1 h.
Polarizations could not be completed at 80 °C because the insulating material was not
stable in HClO4 at this high temperature. Scanning electron micrographs of BND and
TCP polarized at 1.6 V are shown in Figure 6.8A and B, respectively. Although no
morphological changes were noted after polarization at 1.4 V, severe morphological
damage was observed for TCP after polarization at 1.6 V. As shown in Figure 6.8B,
individual carbon ﬁbers fractured and separated from the electrode. In contrast, no
morphological damage was seen for BND after polarization at 1.6 V.

The charge passed during polarization, change in the faradaic current associated
with electroactive surface-oxide functional groups (AI), and the Raman D/Ezg ratios
before and after polarization are presented in Table 6.5. The charge density passed
during the 1 h polarization for BND at both potentials was lower than that for TCP under
the same conditions. As discussed in Chapter 3, the charge recorded during anodic
polarization is associated with reactions at the electrode surface such as oxidation of edge
plane and defect sites and gasiﬁcation of carbon to C02, as well as electric double layer
formation. In addition, some charge may arise from the oxygen evolution reaction,
particularly at the higher polarization potentials. The lower charge for BND is consistent
with the material being less reactive under these conditions and more corrosion-resistant.
Despite the lower reactivity of BND, it seems that the surface oxide content of both
materials increased similarly during polarization at 1.4 V. It should be reemphasized,
however, that some fraction of the spz-bonded carbon within the BND electrode is likely

exposed to the solution because of incomplete diamond coverage through the interior of

166

the electrode. It is possible that the charge recorded during polarization, as well as the
increase in the fraction of surface oxides, is due to oxidation and/or corrosion of the

exposed spZ-bonded portions of the electrode.

 

Figure 6.8 Scanning electron micrographs of (A) BND and (B) TCP alter
potentiostatic polarization at 1.6 V vs. Ag/AgCl at 25 °C for 1 h.

167

Table 6.5 Charge density (mC/cmz), change in faradaic current density for
electroactive surface-oxygen functional groups (Al), and Raman D/Ezg ratio for bare
carbon paper (TCP) and nanocrystalline diamond deposited on carbon paper (BND) after
polarization at different potentials in 0.1 M HClO4 at 25 °C for l h.

 

 

 

 

Polarization Charge AJ
Electrode Potential Density 2 DIE;g
(V vs. Ag/AgCl) (mC/cmz) (”Al cm )

TCP Untreated -- -- 0.30 d: 0.08
TCP 1.4 3.75 4.15 0.20 :t 0.09
BND 1.4 0.347 3.23 --
TCP 1.6 87.3 51.2 0.4 :1: 0.1
BND 1.6 33.7 14.1 --

 

As discussed in Chapter 3, an increase in the D/Ezg ratio is indicative of increases
in the microstructural disorder (i.e., edge plane density) of spZ-bonded carbons. As
shown in Table 6.5, the average D/Ezg ratio for TCP decreased slightly after anodic
polarization at 1.4 V, from 0.30 at 0.08 to 0.20 :1: 0.09. According to the order/disorder
mechanism presented in Chapter 3, more disordered regions of the carbon will corrode
ﬁrst, leaving a more ordered material under certain conditions. After polarization at 1.6
V, D/Ezg increased to 0.4 i 0.1, indicative of increases in the carbon edge plane density.
This is not surprising considering the severe morphological damage noted for the TCP
electrode at 1.6 V. In contrast, there were no microstructural changes observed for BND
as shown in the Raman spectra recorded before and after potentiostatic polarization,

presented in Figure 6.9.

168

 

Before Polarization

l 1 I 41 l I E l 1 l i l 1

After Polarization

 

Intensity (a.u.)

 

 

900 J 1050 I 1200 I 13501 1500‘ 1650 .11800 I 1950
Raman Shift (cm )

Figure 6.9 Raman spectra of BND before and after potentiostatic polarization at 1.6
V vs. Ag/AgCl in 0.1 M HClO4 at 25 °C for l h.

 

In addition to the stability requirements of the support material, the stability of the
Pt electrocatalyst is also an important concern for maintaining PEMFC efﬁciency and
operational lifetime. The stability of the Pt electrocatalyst over time was ﬁrst evaluated
under conditions similar to those used in phosphoric acid fuel cells (PAF C) and has been
attributed to: (1) Pt dissolution/reprecipitation (i.e., an Ostwald ripening process) (ii) Pt
nanoparticle migration and coalescence, and (iii) Pt agglomeration resulting from
corrosion of the spZ-bonded carbon support material.3°’3l’ and "‘femccs them“ In systems
where corrosion of the support was not an issue, there was no consensus in these early
studies on the mechanism for loss of electrocatalyst surface area. In a recent publication,

Ferreira and co-workers studied the mechanism of Pt surface area loss in a PEMFC.31

169

This report provides a detailed discussion of what is known about Pt instability in PAF Cs
and PEMFCs, as well as a number of relevant references.“ In PEMFCs, evidence has
been found that loss of Pt electrocatalyst surface area occurs by both Ostwald ripening
and by solubilization of Pt (i.e., oxidation), subsequent migration of Pt ions within the
ionomer near the cathode, and ﬁnally, reduction to Pt metal by H2 crossover from the
anode.31

The stability of the Pt/carbon composite electrodes deposited under condition B,
as discussed above, was evaluated by potentiostatic polarization at 1.4 V vs. Ag/AgCl
and 25 °C in 0.1 M HClO4 for l h. Figure 6.10 shows scanning electron micrographs of
Pt/TCP-B and Pt/BND-B before and after anodic polarization. Pt nanoparticles decorate
only the ﬁrst few layers of Pt/BND-B before polarization. However, after the 1 h
polarization, Pt nanoparticles can be easily seen at least ~100 um into the electrode,
which suggests that Pt was dissolved and redeposited during polarization (i.e., Ostwald
ripening). Changes are more difﬁcult to note for the Pt/BND-B electrode because Pt
decorated ﬁbers deep within the electrode before polarization. The Pt particle density on
Pt/BND-B after polarization seems to be slightly less than that before polarization on the
outermost ﬁbers, suggesting that Pt dissolution and redeposition also occurred on

Pt/BND-B.

170

l’t/'l‘(‘I’-IS l’t/l'(‘l'-B
lit-fort- After

101m _ ' . 20pm

I i 1 I Before‘ ‘ , After '

 

Figure 6.10 Scanning electron micrographs of Pt/TCP-B and Pt/BND-B before and
alter potentiostatic polarization at 1.4 V vs. Ag/AgCl and 25 °C in 0.1 M HClO4 for 1 h.

Background cyclic voltammetric i-E curves recorded in 0.1 M HClO4 at 50 mV/s
before and after anodic polarization of Pt/TCP-B and Pt/BND-B at 1.4 V vs. Ag/AgCl are
shown in Figure 6.11. The electrochemically-active Pt surface area, calculated from the
hydrogen adsorption charge, is presented in Table 6.6. The curves for Pt/TCP-B and
Pt/BND-B look similar before and after polarization and all electrodes exhibit the
characteristic Pt signal. Although it seems that the peak current for the Pt-oxide
reduction (~450 mV vs. Ag/AgCl) is larger after polarization for Pt/TCP-B, the entire

curve is shifted negative below ~600 mV vs. Ag/AgCl, presumably because there is

171

oxygen present in the electrode after anodic polarization. The electrochemically-active Pt
surface area does not change for either electrode after polarization (the measured values

were always within 5 %).

 

400 -
300'-
2001

1001 1 -
() L
-100
-200
-300
-400 ,,
-500
-600?
-700 r . . .
-0.4 -0.2 0

 

 

Current (uA/cmz)

' I

 

 

I I 1 I J J 1 I

.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Potential (V vs. Ag/AgCl)

 

— Pt/BND-B Before Polarization

— - - Pt/BND-B TCP After Polarization
-——-—- Pt/TCP-B Before Polarization

- - - Pt/TCP-B After Polarization

Figure 6.11 Cyclic voltammetric i-E curves for Pth‘CP-B and Pt/BND-B in 0.1 M
HClO4, recorded at 50 mV/s normalized to the electrode surface enhancement factor
(SEF), before and after a 1 h polarization at 1.4 V vs. Ag/AgCl at 25 °C.

Table 6.6 Electrochemically-active Pt surface area (Pt/TCP-B and Pt/BND-B) before
and after anodic polarization at 1.4 V vs Ag/AgCl and 25 °C in 0.1 M HClO4 for l h.

 

Pt Surface Area (cmz) Pt Surface Area (cmi)

 

Electrode Before Polarization After Polarization
Pt/TCP-B 3.90 3.92
Pt/BND-B 12.2 12.4

 

172

Linear sweep voltammetric i-E curves (ﬁrst scan) for Pt/TCP-B and Pt/BND-B in
oxygen-saturated 0.1 M HClO4 at 1 mV/s, before and after anodic polarization at 1.4 V
vs. Ag/AgCl, are presented in Figure 6.12. As shown in Table 6.7, the half-wave
potential (RP/2) and limiting currents (inm) normalized to the electrode surface
enhancement factor are quite similar before and after polarization, suggesting that Pt
mass was not lost as a result of polarization. The only signiﬁcant difference in the OR
alter polarization is the shape of the curves, likely a result of differences in the oxygen

concentration within the electrode, as discussed above.

 

L

 

-600 .

Current (MA/SEF)

 

 

 

A

 

0.0 l 0.3 0:6 A 0:9
Potential (V vs. Ag/AgCl)

— Pt/BND-B Before Polarization
- - - Pt/BND-B After Polarization
-——— Pt/TCP-B Before Polarization
- - - PthCP-B After Polarization

Figure 6.12 Linear sweep voltammetric i-E curves for the oxygen reduction reaction at
Pt/TCP-B and Pt/BND-B in oxygen-saturated 0.1 M HClO4 before and after polarization
at 1.4 V vs. Ag/AgCl at 25 °C in 0.1 M HClO4 for 1 h. Curves were recorded at 1 mV/s
and are normalized to the electrode surface enhancement factor (SEF).

173

Table 6.7 Half-peak position (Hp/2) and current (tum) for the oxygen reduction
reaction on Pt/TCP-B and Pt/BND-B in oxygen-saturated 0.1 M HClO4 after anodic
polarization at 1.4 V vs. Ag/AgCl and 25 °C in 0.1 M HClO4 for l h. The scan rate was 1
mV/s and the current was normalized to the surface enhancement factor (tum, SEF,
pA/SEF).

 

 

illm, SEF inm, SEF
Ep/Z (V) Epn (V)
Electrode Before After (“:18“) (”Q/5E“
Polarization Polarization e.ore. . er .
Polarization Polarization
Pt/TCP-B 0.567 0.602 -130 -160
Pt/BND-B 0.616 0.630 -250 -280

 

6.3 Conclusions

Carbon paper has been successﬁrlly coated with nanocrystalline boron-doped
diamond and studied as an electrocatalyst support material. Scanning electron
microscopy, XRD, and Raman spectroscopy conﬁrmed that the carbon ﬁbers were fully
coated, within the ﬁrst ~100 pm of the TCP surface, with high-quality, nanocrystalline
boron-doped diamond. Pulsed galvanostatic deposition of Pt nanoparticles on TCP and
BND was carried out at two different current densities, A (0.146 mA/cm2 or 0.0047
C/cmz) and B (1.25 mA/cm2 or 0.040 C/cmz). Under the best of these two conditions for
BND (condition A), 21 m2/g Pt (lower limit estimate) was deposited and an oxygen
reduction current of ca. -1000 uA/SEF (-11.6 mA/mgpt) was seen, whereas only 1.4 mz/g
Pt was deposited on TCP with no well-deﬁned oxygen-reduction peak. BND exhibited
superior structural stability and corrosion resistance during anodic polarization at 1.6 V
with no morphological or microstructural damage. In contrast, severe morphological
damage was observed for TCP. Anodic polarization of Pt/TCP-B and Pt/BND-B at 1.4 V

provided evidence for an Otswald ripening mechanism where the Pt was dissolved and

174

redeposited. There were no changes in the morphology of the electrocatalyst supports,

Pt/TCP-B or Pt/BND-B, after the polarization.

175

(£41

(1)

(2)

(3)

(4)

(5)

(6)
(7)

(8)
(9)
(10)
(11)
(12)

(13)

(14)

(15)

(16)

lieyinnewtcuas

May, P. W.; Rego, C. A.; Ashfold, M. N. R.; Rosser, K. N.; Lu, G.; Walsh, T. D.;
Holt, L.; Everitt, N. M.; Partridge, P. G. Diam. Rel. Mater. 1995, 4, 794-797.

Shah, S. I.; Waite, M. M. J. Vac. Sci. T echnol. A 1995, 13, 1624-1627.

Ahneida, E. C.; Diniz, A. V.; Trava-Airoldi, V. J .; Ferreira, N. G. Thin Solid
Films 2005, 485, 241-246.

Almeida, E. C.; Trava-Airoldi, V. J .; F erreira, N. G.; Rosolen, J. M. Diam. Rel.
Mater. 2005, 14, 1673-1677.

Almeida, E. C.; Diniz, A. V.; Rosolen, J. M.; Trava-Airoldi, V. J .; Ferreira, N. G.
Diam. Rel. Mater. 2005, 14, 679-684.

Ferreira, N. G.; Silva, L. L. G.; Corat, E. J. Diam. Rel. Mater. 2002, 11, 657-661.

Ungar, T.; Gubicza, J .; Ribarik, G.; Pantea, C.; Zerda, T. W. Carbon 2002, 40,
929-937.

Tunistra, F.; Koenig, J. L. Chem. Phys. 1970, 53, 1126-1130.

Wang, Y.; Alsmeyer, D. C.; McCreery, R. L. Chem. Mater. 1990, 2, 557-563.
Nemanich, R. J .; Sonin, S. A. Phys. Rev. B 1979, 20, 392-401.

Nathan, M. 1.; Smith, J. E. J .; Tu, K. N. J. Appl. Phys. 1974, 45, 2370.
Dresselhaus, M. S.; Dresselhaus, G. Adv. Phys. 1981, 30, 139-326.

Show, Y.; Witek, M. A.; Sonthalia, P.; Swain, G. M. Chem. Mater. 2003, I5, 879-
888.

Bergman, L.; Nemanich, R. J. J. Appl. Phys. 1995, 78, 6709-6719.

Ferrari, A. C.; Robertson, J. Phys. Rev. B. Condens. Matter Mater. Phys. 2001,
63,1214051-1214054.

Nemanich, R. J .; Glass, J. T.; Lucovsky, G.; Shroder, R. E. J. Vac. Sci. T echnol.
1988,A6, 1783-1787.

176

(17)

(18)

(19)

(20)

(21)

(22)

(23)

(24)

(25)

(26)

(27)

(28)

(29)

(30)

(31)

Prawer, S.; Nugent, K. W.; Jamieson, D. N.; Orwa, J. O.; Bursill, L. A.; Peng, J.
L. Chem. Phys. Lett. 2000, 332, 93-97.

McCreery, R. L. In Neuromethods; Boulton, A., Baker, G., Adams, R. N., Eds;
Humana Press, 1995; Vol. 27, pp 1-26.

Fagan, D. T.; Hu, I.-F.; Kuwana, T. Anal. Chem. 1985, 5 7, 2759-2763.

Bennett, J. A.; Show, Y.; Wang, S.; Swain, G. M. J. Electrochem. Soc. 2005, 152,
E184-El92.

Wang, J .; Swain, G. M. J. Electrochem. Soc. 2003, 150, E24-E32.

Stevens, D. A.; Hicks, M. T.; Haugen, G. M.; Dahn, J. R. J. Electrochem. Soc.
2005, 152, A2309-A2315.

Perez, J .; Tanaka, A. A.; Gonzalez, E. R.; Ticianelli, E. A. J. Electrochem. Soc.
1994, 141, 431-436.

Essalik, A.; Amouzegar, K.; Savadogo, o. J. Appl. Electrochem. 1995,25, 404-
407.

Mathias, M. F .; Makharia, R.; Gasteiger, H. A.; Conley, J. J.; Fuller, T. J .;
Gittleman, C. J .; Kocha, S. S.; Miller, D. P.; Mittelsteadt, C. K.; Xie, T.; Yan, S.
G.; Yu, P. T. Interface 2005, 14, 24-35.

Parthasarathy, A.; Srinivasan, S.; Appleby, A. J .; Martin, C. R. J. Electrochem.
Soc. 1992, 139, 2856-2862.

Parthasarathy, A.; Srinivasan, S.; Appleby, A. J .; Martin, C. R. J. Electrochem.
Soc. 1992, 139, 2530-2537.

Lawson, D. R.; Whitely, L. D.; Martin, C. R.; Szentirmay, M. N.; Song, J. 1. 0R
at Naﬁon 1988, 135, 2247-2253.

Gottesfeld, S.; Raistrick, I. D.; Srinivasan, S. J. Electrochem. Soc. 1987, 134,
1455-1462.

Antolini, E. J. Mater. Sci. 2003, 38, 2995-3005.

Ferreira, P. J .; la 0', G. J .; Shao-Hom, Y.; Morgan, D.; Makharia, R.; Kocha, S.
S.; Gasteiger, H. A. J. Electrochem. Soc. 2005, 152, A2256-A2271.

177

Chapter 7

Future Research Directions

Rising energy demands coupled with a limited supply of fossil fuels has lead to a
signiﬁcant amount of research and development into alternative energy technologies. For
example, the polymer electrolyte membrane fuel cell (PEMFC), an energy conversion
device that operates on hydrogen fuel, has received a great deal of attention in recent
years because it has a theoretical operating efﬁciency of 100% with no harmful
emissions. However, current PEMFC technologies have a limited operating lifetime and
reduced electrical efﬁciency because of problems with the durability and performance of
electrode materials under certain PEMFC operating ' conditions. Speciﬁcally,
microstructural degradation and corrosion reactions of the spZ-bonded carbon
electrocatalyst support under high potentials (1.2 to 1.5 V vs. SHE) during start-up and
shut-down, as well as idle operation, is a problem that directly affects PEMFC efﬁciency.
Although the degradation mechanism is not well understood, microstructural damage of
the carbon support has been shown to lead to loss of electrocatalytic activity through
pathways such as electrocatalyst aggregation and, in extreme cases, complete electrode
failure. The research presented in this dissertation focused on the study of spz-bonded
carbon degradation mechanisms and the development of a new, dimensionally-stable
electrocatalyst support material based on electrically conducting diamond. In this

chapter, future work in these two areas is discussed.

178

Chapter 3: Study of the Degradation Mechanisms of spZ-Bonded Carbon
Powders During Electrochemical Polarization

The morphology and microstructure of two structurally well-characterized spz-

bonded carbon powders, graphite and glassy carbon (GC), were studied under applied

potential (0.8 to 1.6 V vs. Ag/AgCl) at 25, 50, and 80 °C. Oxidation of graphite and GC,

coupled with heterogeneous microstructural changes across a graphite electrode surface,

provided evidence for the order/disorder mechanism of carbon corrosion, with a

nucleation and growth process for graphite. Preliminary comparisons were made to a

carbon black commonly used in fuel cells, Vulcan XC-72, with and without Pt

electrocatalyst.

1)

2)

3)

These measurements were carried out ex situ after short times (1 h) such that no
temporal information or conclusions about long-term degradation could be
gained. Future work should include in situ Raman spectroscopic measurements
over longer time periods to gain a better understanding about the nature of
degradation.

Characterization of carbon oxidation and gasiﬁcation to CO and CO2 should be
done using on-line mass spectrometry. This type of study, coupled with the in
situ microstructural information, will lead to a better understanding of the
processes that occur during carbon degradation.

Membrane electrode assemblies composed of Vulcan XC-72 with and without Pt
electrocatalyst should be studied in a fuel cell at different potentials, temperatures,

and time periods. The effect of the Pt electrocatalyst on carbon corrosion, as well

179

_ gm

 

as the consequences of electrocatalyst support corrosion on fuel cell efﬁciency

should be characterized.

Chapter 5: Preparation and Characterization of Boron-Doped Diamond
Powder: A Possible Dim ensionally Stable Electrocatalyst Support Material

and

Chapter 6: Preparation and Characterization of Boron-Doped Diamond
Paper with Pt Nan oparticles

Two high-surface-area electrocatalyst support materials based on electrically conducting
diamond were synthesized.
1) Future work with conducting diamond powder should include strategies to
improve the electrochemical response of the bound electrode.

- The electrode wetting should be improved by testing different types of
binders and by varying the powder/binder ratio.

- Electrodes fabricated using large (8-12 1.1m diam) conductive diamond
powder particles mixed with a small percentage (e.g., 10-20 wt. %) of
smaller (e.g., 0-2 pm) conductive diamond powder particles may reduce
electrode ohmic resistance by improving the particle-particle contact.

- Composite electrodes made by mixing conducting diamond powder with
ﬂatter, more compressible powder such as hydrogenated graphite may
help to reduce the electrode resistance while still maintaining a signiﬁcant

portion of the corrosion resistance.

180

- Hot-pressing the powder into a Naﬁon membrane, as is done with carbon
black during MEA fabrication, may reduce the contact resistance. In
addition, this electrode geometry could be used directly in a fuel cell.

2) The powder surface area should be increased.

- Characterization of detonation nanodiamond powder with different boron
doping strategies should continue. This includes conductivity
measurements and material characterization such as Raman spectroscopy
to assess the nondiamond carbon impurity content.

- Effective coating strategies should be developed to coat small diameter
(nm-sized) abrasive diamond powder with an electrically conducting
layer.

3) The diamond powder should be tested as an electrocatalyst support material.

- Strategies should be developed to deposit Pt on conducting diamond
powder so that its electrocatalytic activity can be addressed. These may
include chemical or electrochemical deposition. Durability and
performance are key issues.

- Methods to anchor the Pt electrocatalyst particles to the diamond support
should be developed that result in a stable support-catalyst interaction, but
do not signiﬁcantly increase the area of the support or decrease the area of

the electrocatalyst.

181

4) The Pt/carbon paper composite electrodes should be further developed as an
electrocatalytic electrode.

- A method to accurately assess the carbon paper surface area should be
developed so that increases in electrode wetting during electrochemical
characterization do not inﬂuence the normalized current or charge. This
may include strategies to increase electrode wetting at the outset by
soaking the carbon paper in a variety of solvents from non-polar to polar.

- Pt electrodeposition on diamond-coated carbon paper should be optimized
in terms of small, uniform particle diameter and high particle density to
increase electrocatalytic activity of the Pt/carbon paper electrodes.

- Carbon paper and diamond-coated carbon paper should be hot-pressed
with a Naﬁon® membrane to form a membrane electrode assembly and
tested in a fuel cell to address the electrocatalytic activity of the

electrodes, as well as their stability over time.

182

JilllllLlllllllll/llllll

31