j)

I
“I

h
A

.1 an! .1. .7

“mi-J

13.

my .
knmw

53:

.
.2... 3%
‘JUJL‘V. ‘13,!
‘ dang“
SK
1..

 

...,
.20....nHy , .

 

ll
I'll

 

 

WW? LIBRARY
Michigan State
5933,37,,” University

 

 

This is to certify that the
dissertation entitled

FABRICATION, CHARACTERIZATION, AND
ELECTROCATALYTIC ACTIVITY OF METAL/DIAMOND
COMPOSITE ELECTRODES

presented by
Jian Wang

has been accepted towards fulﬁllment
of the requirements for the

Ph.D. degree in Chemistry

4d ’1‘ I417 $22210?)

major Professor’s Signature
12/20/02

Date

MSU is an Afﬁnnative Action/Equal Opportunity Institution

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

 

lﬂA

 

cp-<

0 6 7006
b

“ii tfﬁ

 

-——£H3¥-2-¥-fﬁﬁﬁ
UN 1 2085

 

 

W

’93 \ﬂ 2 & 2M7

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/DatoDue.p65—p.15

 

FABRICATION, CHARACTERIZATION, AND
ELECTROCATALYTIC ACTIVITY OF METAL/DIAMOND
COMPOSITE ELECTRODES

By

Jian Wang

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Chemistry

2002

ABSTRACT

FABRICATION, CHARACTERIZATION, AND ELECTROCATALYTIC ACTIVITY
OF NIETAUDIAMOND COMPOSITE ELECTRODES

By

Jian Wang

This dissertation is devoted to a new research area of diamond electrochemistry:
application of boron-doped diamond thin-film electrodes in the ﬁeld of electrocatalysis.
The main focus is on the fabrication and characterization of Pt/diamond composite
electrodes and on the investigation of their electrocatalytic activities.

Dimensionally stable Pt/diamond composite electrodes are fabricated by a
sequential diamond groth Pt deposition/ diamond growth procedure. In this multistep
process, a continuous boron-doped diamond thin film is ﬁrst deposited on a conductive
substrate (e.g. Pt, Si or W). DC magnetron sputtering or electrodeposition is next used to
deposit a discontinuous layer of Pt particles on the surface. The Pt-coated diamond ﬁlm is
then subjected to a short, secondary diamond growth in order to anchor the metal
particles into the diamond surface microstructure.

The resulting Pt/diamond composite ﬁlms are characterized by SEM, EDX, AFM,
AES, XRD, SIMS, Raman spectroscopy, and cyclic voltammetry. The metal particles
range in diameter from 10 to 500 nm. In the electrodeposition approach, the number of
exposed metal particles, particle size, and distribution can be controlled, to some extent,
by adjusting the electrodeposition and secondary diamond growth conditions. The

second, short diamond deposition serves to entrap many of the metal particles into the

diamond rnicrostructure by surrounding their base. Consequently, there is little catalyst
detachment or particle aggregation during use under extreme conditions. The catalytic
activity of the composite electrodes is extremely stable, as no microstructural alterations
or activity losses were observed during a 2 h anodic polarization in 85% H3PO4 at 0.1
Aocm'2 and 170°C.

The electrode’s catalytic activity is evaluated using the oxygen reduction reaction
and methanol oxidation reaction, at room temperature in acidic media. The catalytic
activities of the composite electrode are found to be comparable to those for a
polycrystalline and carbon-supported Pt catalyst. The results suggest that there is little
alteration of the physicochemical properties of the Pt catalyst dun'ng secondary diamond
growth. The diamond ﬁlm exhibits metal-like electrical conductivity as both a support
material and a current collector.

As part of the groundwork for my research, copper electrodeposition is
investigated in order to gain a fundamental understanding of metal nucleation/growth on
diamond surface. The electrodeposition of copper is strongly influenced by the electronic
properties (i.e., doping level) of the diamond ﬁlm. An increase in the boron-doping level
signiﬁcantly increases the number density of nucleation sites and alters the nucleation
mechanism. The size of the metal deposits, their spatial distribution and number density
are potentially to be controlled by adjusting the applied overpotential, deposition time

and electrical conductivity (i.e., boron doping level) of the diamond ﬁlm.

ACKNOWLEDGEMENTS

I am appreciative of all those who have helped along the way. Foremost, I would
like to thank my advisor, Dr. Greg M. Swain, for his long-term patience, constant
support, and never-ending encouragement during my entire graduate education. Greg has
been a great mentor. Not only has he expertly guided my scientific development, but his
extra personal touch has also helped me over many rough spots during the past ﬁve years.

I am also grateful to my former advisory committee members at Utah State
University, Dr. Robert Brown, Dr. Stephen Bialkowski, Dr. John Hubbard, and Dr. T. C.
Shen. Their constant guidance and encouragement have played a critical role at the early
stage of my research. I would also like to thank my committee members at Michigan
State University, Dr. Simon Garrett, Dr. Thomas Pinnavaia, and Dr. Jes Asmussen, for
reading my thesis and providing helpful suggestions.

The deepest gratitude goes to my parents, in—laws, my sisters and brother-in-laws
for their unreserved support, love, and trust in these years. Special thanks are given to my
wife, Ning Yu. She has sacriﬁced so much, particularly during my two-year stay here at
Michigan, so that I could finally make it through my graduate study. It is her love,
understanding, and support that make it possible for me to overcome the obstacles I faced
during the past ﬁve years.

Many people I have met in the Swain research group have been a source of
support at one time or another during my graduate career. Special thanks go to three
visiting scientists: Dr. Jerzy Strojek taught me most of the experimental skills when I

joined the group. Dr. Jerzy Zak and Dr. Miles Koppang were other mentors to me. Many

iv

thought-provoking discussions with them expanded my knowledge of both experimental
and theoretical aspects of this research. I would also like to thank the past and present
Swain group members: Mike, Jishou, Qingyun, Matt, Maggie, Prema, Jason B, Jason S,
Shannon, JinWoo, Show, Grace, and Gloria. Thank you for all the friendships I have
acquired, which have made my graduate life a really enjoyable one.

Finally, I gratefully acknowledge the Ofﬁce of Science, Department of Energy for
the ﬁnancial support, Utah State University and Michigan State University for the
Graduate Teaching and Research Assistantships, and Seagate Technology for the 2002

Summer Internship.

TABLE OF CONTENTS

 

LIST OF TABLES ................................................................................. viii
LIST OF FIGURES .................................................................................. x
CHAPTER 1 INTRODUCTION - - - - -- __ - _______ - ..... .-l
1.1 Electrocatalysis and Direct Methanol Fuel Cells ................................................... 1
1.2 Carbon Support Materials ..................................................................................... 4
1.3 The Problem ......................................................................................................... 5
1.4 Diamond Electrochemistry .................................................................................... 8
1.5 The Motivations .................................................................................................. 17
1.6 Outline of the Dissertation .................................................................................. 18
References -- -- _ -- - - - - - - - - - -- ........... _ - -21

 

CHAPTER 2 ELECTROCHEMICAL AND IN SITU ATOMIC FORCE
MICROSCOPIC STUDIES OF COPPER DEPOSITION ON BORON-DOPED

 

DIAMOND THIN- FILM ELECTRODES - ............ - - _ 25
2.1 Introduction ........................................................................................................ 25
2.2 Experimental ...................................................................................................... 27
2.3 Theories of Metal Nucleation and Grth ........................................................... 32
2.4 Results ................................................................................................................ 36

2.4.1 Electrochemical Study of Copper Deposition ............................................... 36
2.4.2 ECAFM Study of Copper Deposition ........................................................... 50

2.5 Discussion .......................................................................................................... 63
2.6 Conclusions ........................................................................................................ 68
References--_---.- -- - -_ -- - _ - ........................ - 70

 

CHAPTER 3 FABRICATION AND CHARACTERIZATION OF
DMENSIONALLY STABLE lPT/DIAMOND COMPOSITE ELECTRODES FOR

ELECTROCATALYSIS — THE MAGNETRON SPU'ITERIN G APPROACH ...73

3.1 Introduction ........................................................................................................ 73
3.2 Experimental ...................................................................................................... 75
3.3 Results and Discussion ....................................................................................... 78
3.3.1 Structural Characterization ........................................................................... 78
3.3.2 Electrochemical Characterization ................................................................. 85

3.4 Conclusions ........................................................................................................ 94
References--. -- .......... _ .......................................... - ....... -- - 95

 

CHAPTER 4 FABRICATION AND CHARACTERIZATION OF
DIMENSIONALLY STABLE PT/DIAMON D COMPOSITE ELECTRODES FOR

ELECTROCATALYSIS —— THE ELECTRODEPOSITION APPROACH ............ 97
4.1 Introduction ........................................................................................................ 97
4.2 Experimental ...................................................................................................... 99
4.3 Results and Discussion ..................................................................................... 103

vi

4.3.1 Electrodeposition of Pt on Diamond Thin Films ......................................... 103

4.3.2 Secondary Diamond Growth ...................................................................... 110
4.3.3 Stability of the Pt/Diamond Composite Electrode ....................................... 121

4.4 Conclusions ....................................................................................................... 124
References - - - - - - ..... -- -- - - - 125

 

CHAPTER 5 OXYGEN REDUCTION REACTION ON PT/DIAMOND
COMPOSITE ELECTRODES — A ROTATING DISK VOLTAMNIETRIC

 

 

STUDY... - -- - - - - -- - 127
5.1 Introduction ...................................................................................................... 127
5.2 Mechanism and Kinetics of OR ...................................................................... 129
5.3 Experimental .................................................................................................... 131
5.4 Results and Discussion ..................................................................................... 133

5.4.1 Characterization of the Diamond Film on Mo Disk .................................... 133
5.4.2 ORR at Stationary Pt/Diamond Composite Electrodes ................................ 139
5.4.3 OR at Pt/Diamond Composite RDEs ....................................................... 143

5.5 Conclusions ....................................................................................................... 156
References.....-..-.-- -- -- - -- -- - - ........ - ...... 158

CHAPTER 6 METHANOL OXIDATION REACTION ON METAL/DIAMOND

 

COMPOSITE ELECTRODES-- - - -- - - ------------- - - 160
6.1 Introduction ...................................................................................................... 160
6.2 Mechanism and Kinetics of MOR ..................................................................... 162
6.3 Experimental .................................................................................................... 165
6.4 Results and Discussion ..................................................................................... 166

6.4.1 Measurement of the Open Circuit Potential ................................................ 166
6.4.2 MOR at Pt/Diamond Composite Electrodes ................................................ 168
6.4.2.1 Cyclic Voltammetric Studies of MOR ................................................ 168
6.4.2.2 Chronoamperometric Studies of MOR ................................................. 174

6.4.3 MOR at a Bimetallic Catalyst ..................................................................... 179

6.5 Conclusions ...................................................................................................... 183
References--- - - -- - -- - - ...... - . 184

 

vii

Table 2.1.

Table 2.2.

Table 2.3.

Table 2.4.

Table 2.5.

Table 2.6.

Table 4.1.

Table 5.1.

Table 5.2.

Table 5.3.

Table 5.4.

Table 6.1.

Table 6.2.

Table 6.3.

LIST OF TABLES

Summary of the diamond thin-ﬁlm deposition parameters. ........................... 28

CV data for copper electrodeposition on different electrodes in 1 mM

CuSO4 + 50 mM H2SO4. .............................................................................. 38
Current-time transient data for copper electrodeposition on LBD. ................ 44
Current-time transient data for copper electrodeposition on I-IBD. ................ 44
Current-time transient data for copper electrodeposition on HOPG ............... 45
Current-time transient data for copper electrodeposition on GC. ................... 45

Pt particle size, distribution, hydrogen desorption charge and
roughness factor (RF) for the Pt/diamond composite electrodes before

and alter the secondary diamond growth. ................................................... 108
Electrochemically active surface area and roughness factor for the
Pt/diamond composite RDEs with different Pt loadings .............................. 138

Open circuit potentials (25 0C) for the Pt/diamond composite RDEs in
different acid electrolytes. .......................................................................... 140

Comparison of the kinetics parameters for OR at the Pt/diamond
composite electrodes with the literature data reported for single crystal,
polycrystalline, and carbon-supported Pt at room temperature. ................... 152

Comparison of the speciﬁc activity (SA) and mass activity (MA) of the
Pt/diamond composite electrode for ORR with the literature data
reported for other carbon-supported Pt electrodes ....................................... 155

Electrochemically active surface area and roughness factor for the

polycrystalline Pt and Pt/diamond composite electrodes with different
Pt loadings. ................................................................................................ 166

Open circuit potentials for Pt/diamond composite electrodes in 0.1 M
HClO4 with different methanol concentrations. .......................................... 167

Cyclic voltammetric data for the MOR in 0.1 M HClO4 at Pt/diamond
composite electrodes with different Pt loadings. ......................................... 172

viii

Table 6.4. Summary of the Chronoamperometric i-t data for the oxidation of
methanol at a Pt/diamond composite electrode in 0.1 M HClO4 as a
function of the methanol concentration ....................................................... 179

ix

LIST OF FIGURES

Figure 1.1. (A) Schematic diagram of a direct methanol fuel cell working in an
acidic medium. 03) Schematic diagram of a proton exchange
membrane ﬁle] cell. Highly-dispersed catalyst (small dots) is
supported on carbon black particles (big dots). ............................................... 3

Figure 1.2. Diamond ﬁlm morphologies as a function of methane-to-hydrogen
source gas ratio used during microwave plasma CVD. The surface area
is 10 x 10 umz. Z-axis range is 6 um. Hydrogen ﬂow rate is 200 sccm.
Growth time is 10 h ...................................................................................... 10

Figure 1.3. The effect of anodic polarization on the electrode morphology. Both
the diamond ﬁlm and glassy carbon electrodes were subjected to
polarization at 2 V vs. Ag/AgCl in 0.1 M HClO4. The in situ AFM
images were recorded for polarization times 2, 10, and 15 min.,
respectively. ................................................................................................. 14

Figure 2.1. Diagram of the single-compartment, glass electrochemical cell. (a) Cu
or Al metal current collecting back plate, (b) working electrode, (c)
Viton O-ring, (d) input for nitrogen purge gas, (e) carbon rod or Pt
counter electrode, and (0 reference electrode inside a glass capillary
tube with a cracked tip. ................................................................................ 30

Figure 2.2. Diagram of the ECAF M ﬂuid cell. (a) photodetector, (b) He-Ne laser,
(0) auxiliary electrode, ((1) reference electrode, (e) tip, (t) solution
inlet, (g) solution outlet, (h) working electrode, (i) O-ring, and (j)
piezoscanner. ............................................................................................... 3 1

Figure 2.3. (A) Schematic representation of the grth of the difﬁrsion zones an
their eventual overlap. The arrows indicate the directions of the
diffusional ﬁeld during the grth of the nuclei. (B) Typical current-
time transient recorded during the potential step after background
correction. .................................................................................................... 34

Figure 2.4. Cyclic voltammetric i-E curves for copper deposition on (A) GC, (B)
HBD, (C) LBD, and (D) HOPG in 1 mM CuSO4 + 50 mM H2804.
Scan rate = 20 mV/s. .................................................................................... 37

Figure 2.5. Potentiostatic current-time transients for copper deposition on LBD in
1 mM CuSO4 + 50 mM H2SO4 at the overpotentials indicated in mV. .......... 40

Figure 2.6. Comparison of experimental current-time transients (dash curves) for
LBD (see Figure 2.5) with the theoretical transients corresponding to
instantaneous (solid curve a) and progressive (solid curve b)

nucleation at different overpotentials: (A) —410, (B) -460, (C) —5 10,
(D) -610. (E) 660, and (F) —710 mV. ............................................................. 42

Figure 2.7. Potentiostatic current-time transients for copper deposition on HBD in
1 mM CuSO4 + 50 mM H2804 at the overpotentials indicated in mV. .......... 43

Figure 2.8. Comparison of experimental current-time transients (dashed curves)
for HBD (see Figure 2.7) with the theoretical transients corresponding
to instantaneous (solid curve a) and progressive (solid curve b)
nucleation at overpotentials of (A) —410 and (B) —610 mV. ........................... 43

Figure 2.9. Comparison of experimental current-time transients (dashed curves)
with the theoretical transients according to Equation 2.8 for copper
deposition on LBD at overpotentials of (A) -410 and (B) —610 mV .............. 46

Figure 2.10. Plot of the logarithm of the number density of active sites vs.
overpotential, constructed from the data given in Table 2.3 and 2.4. ............. 46

Figure 2.11. AFM images (force mode, in air, 5 x 5 pm) of (A) LBD, (B) HBD,
(C) HOPG, and (D) GC. .................................................................................. 51

Figure 2.12. ECAFM images (force mode, 10 x 10 um) collected for copper
deposition on LBD at overpotentials of (B) 390 mV, (C) -460, (D) —
560, (E) —710, and (F) 390 mV for 30 5 each. The solution is lmM
CuSO4 + 50 mM H2804. (a)-(i) represent individual copper crystallite.
The diagram of the sequential potential steps is shown in (A). The
triangular markers indicate the starting time of each imaging. ....................... 54

Figure 2.13. ECAFM images (height mode, 2.5 x 2.5 pm) collected for copper
deposition on LBD in lmM CuSO4 + 50 mM H2804 (A) before and
(B-E) after the potential step to —460 mV. The deposition times are (B)
30, (C) 60, (D) 120, and (E) 180 s, respectively. Corresponding cross
sections, through the top view images shown in the middle column, are
indicated by the straight lines. The horizontal distance between the two
black cursors represents the lateral dimension of the crystallite. .................... 55

Figure 2.14. ECAFM images (force mode, 4 x 4 pm) collected for copper
deposition on LBD in lmM CuSO4 + 50 mM H2804 (B) before and
(GP) after the potential step to —710 mV. The deposition times are (C)
30, (D) 60, (E) 120, and (F) 180 s, respectively. (a)-(f) represent
individual copper crystallite. The diagram of the potential step is
shown in (A). The triangular markers indicate the starting time of each
imaging. ........................................................................................................... 57

Figure 2.15. ECAFM images (force mode, 5 x 5 pm) collected for copper
deposition on HBD at overpotentials of (B) 390 mV, (C) —460, (D) —-

xi

560, (E) —710, and (F) 390 mV for 30 3 each. The solution is lmM
CuSO4+ 50 mM H2804. The diagram of the sequential potential steps
is shown in (A). The triangular markers indicate the starting time of
each imaging. .......................................................................................

Figure 2.16. ECAF M images (force mode, 3.4 x 3.4 um) collected for copper
deposition on HBD in lmM CuSO4+ 50 mM H2804 (B) before and
(C-F) after the potential step to —460 mV. The deposition times are (C)
30, (D) 60, (E) 120, and (D) 180 s. (a)—(e) represent individual copper
crystallite, respectively. The diagram of the potential step is shown in
(A). The triangular markers indicate the starting time of each imaging.

Figure 2.17. Proposed band structure for the diamond/electrolyte interface. Eva,
Egg and vac are the energies at the bottom of the conduction band, at
the top of the valance band and in the vacuum. The potentials are
quoted versus SCE. The band positions are determined using a ﬂat-
band potential of 0.45 V vs. SCE and a band gap of 5.45 eV. ...............

Figure 3.1. Fabrication process for the Pt/diamond composite electrodes. .............

Figure 3.2. SEM images of composite diamond electrodes fabricated (A) with
and (B) without Pt. ...............................................................................

Figure 3.4. Raman spectra for diamond composite electrodes fabricated with and
without Pt. Excitation = 514.5 nm. Integration time = 10 s. The laser

power density at the sample was estimated to be ca. 150 kW/cmz. ........

Figure 3.5. Dynamic SIMS proﬁles (positive ion mode) for diamond composite
electrodes prepared (A) with and (B) without Pt. ..................................

Figure 3.6. Cyclic voltammetric i-E curves for a Pt/diamond composite electrode

........ 61

........ 62

........ 64

........ 75

........ 78

........ 81

........ 84

in (A) 0.1 M H00; and (B) l M H2SO4. Scan rate = 50 mV/s. .................... 86

Figure 3.7. Cyclic voltammetric i-E curves in (A) 0.1 M H00; and (B) 1 M
H2SO4 for a composite electrode prepared without Pt. Scan rate = 50

mV/s. ........................................................................................................... 87

Figure 3.8. Cyclic voltammetric i-E curves in 0.1 M HClO4 for a Pt/diamond
composite electrode before and after 1000 potential cycles between —

400 and 1500 mV vs. Ag/AgCl. Scan rate = 50 mV/s. The maximum
anodic and or cathodic current density during the cycling was ~1

mA/cm2 ........................................................................................................ 89

Figure 3.9. Electrochemical AFM images for a Pt/diamond composite electrode
in 0.1 M HClO4 (A) before and (B) after 1000 potential cycles
between —400 and 1500 mV vs. Ag/AgCl. Scan rate = 50 mV/s. The 2-

xii

axis is 4 nN full-scale. The images were acquired at 350 mV vs.
Ag/AgCl. ..................................................................................................... 91

Figure 3.10. Cyclic voltammetric i-E curves for boron-doped diamond composite
electrodes with and without deposited with Pt in 0.6 M CH3OH + 0.1
M HClO4. Scan rate = 50 mV/s. ................................................................... 92

Figure 3.11. Cyclic voltammetric i-E curves for diamond composite electrodes,
with and without Pt, in Oz-saturated 0.1 M HClO4. Scan rate = 50
mV/s. ........................................................................................................... 93

Figure 4.1. (A) Successive potentiodynamic i-E curves for a boron-doped
diamond thin ﬁlm in lmM KthC15 + 0.1M HClO4. Scan rate =
50mV/s. (B) AF M force-mode image (air) of the diamond thin ﬁlm

after 25 potential cycles .............................................................................. 104

Figure 4.2. AF M force-mode images (air) of boron-doped diamond thin ﬁlms
after galvanostatic deposition of Pt from a solution of 1 mM KthC15 +
0.1M HClO4_ The deposition times are (A) 100, (B) 200, (C), 300 and
(D) 400 s, respectively. .............................................................................. 107

Figure 4.3. Cyclic voltammetric i-E curves in 0.1 M HClO4 for a boron-doped
diamond thin ﬁlm electrodeposited with Pt, during 1000 potential
cycles between -400 and 1500 mV vs. Ag/AgCl at 50 mV/s. Dashed
arrows show the decrease in current response with scan number. The
maximum anodic and/or cathodic current density during the cycling
was ca. 1 mA/cmz. ..................................................................................... 109

Figure 4.4. AFM force-mode images (air) of Pt coated diamond thin ﬁlms (as
indicated in Figure 4.2) after a 3 h secondary diamond growth. The
corresponding Pt deposition times are (A) 100, (B) 200, (C), 300 and

(D) 400 s, respectively. .............................................................................. 111
Figure 4.5. Energy dispersive x-ray analysis spectrum for a Pt metal particle

shown in the SEM image (inset) ................................................................. 112
Figure 4.6. XRD pattern of a Pt/diamond composite electrode. .................................. 113

Figure 4.7. Cyclic voltammetric i-E curves for a Pt-coated diamond thin ﬁlm in
0.1M HClO4 before (solid line) and after (dashed line) a 3 h of
secondary diamond growth. The Pt deposition time was 2003. Scan
rate = 50 mV/s. .......................................................................................... 116

Figure 4.8. SEM images of a Pt/diamond composite electrode (A) before and (B)
after etching in aqua regia. The Pt deposition time was 200 s. .................... 1 18

xiii

Figure 4.9. Cross-sectional SEM images of an acid-etched Pt/diamond composite
electrode. (A) Secondary electron imaging mode. (B) Back—scattered
electron imaging mode. The secondary diamond grth time was 3 h.
Arrow 1-3 show three Pt particles covered by the diamond ﬁlm. Arrow
4 shows a cavity after dissolving Pt. ........................................................... 120

Figure 4.10. Cyclic voltammetric i-E curves for a Pt/diamond composite
electrode in 0.1 M HC104 before (dashed line) and after two l-h
polarizations (solid lines) in 85 wt % H3PO4 at 170 °C and 0.1 A-cm’z.
The Pt deposition time was 200 s. .............................................................. 123

Figure 4.11. AFM images (air) of the Pt/diamond composite electrode (A) before
and (B) after 2-h anodic polarization in 85 wt % H3PO4 at 170 0C and
0.1 A-cm'z. ................................................................................................. 123

Figure 5.1. Schematic diagram of a rotating disk Pt/diamond composite electrode:
a) a top and side view showing the direction of mass transport; b) a
view of the electrode mounted in Teﬂon; c) a side view of the
composite disk electrode architecture. ........................................................ 132

Figure 5.2. SEM image of a boron-doped diamond thin ﬁlm deposited on a Mo
disk. The diamond deposition time was 15 h and methane-to-hydrogen
ratio was 0.35% .......................................................................................... 134

Figure 5.3. Raman spectrum for a boron-doped diamond thin ﬁlm deposited on a
Mo disk. Excitation wavelength = 532 nm. Incident power density =
500 chm'z. Integration time = 10 s. ......................................................... 135

Figure 5.4. Cyclic voltammetric i-E curves for a boron-doped diamond thin ﬁlm
deposited on a Mo disk in 0.1 M HCIO4. Scan rate = 100 mV/s .................. 137

Figure 5.5. Cyclic voltammetric i-E curves for Pt/diamond composite electrodes
in 0.1 M HClO4. The Pt deposition times were 200, 300, and 400 s,
respectively. Scan rate = 50 mV/s. ............................................................. 137

Figure 5.6. Cyclic voltammetric i-E curves for (A) RDE #1 and (B)
polycrystalline Pt foil electrode in Nz-purged (dashed line) and 02-
saturated (solid line) 0.] M HClO4, Scan rate 50 mV/s. Active surface
area of the Pt foil electrode is 0.43 cm2 (geometric area is 0.2 cmz). ........... 141

Figure 5.7. Linear potential sweep voltammetric i-E curves for RDE #1 in Oz-
saturated 0.1 M HClO4 at different scan rates. Inset shows the plot of

peak potential vs. logarithm of scan rate. .................................................... 142

Figure 5.8. Cyclic voltammetric i-E curves for RDE #3 in (A) Ng-purged and (B)
02- saturated 0.1 M HC104. The dashed line is the curve for oxygen

xiv

reduction aﬁer background correction. Scan rate = 20 mV/s. Rotation
rate = 1000 rpm. ......................................................................................... 144

Figure 5.9. Linear sweep voltammetric i-E curves for RDE#3 in Oz-saturated 0.1
M HClO4 as a function of the rotation rate. Scan rate = 20 mV/s. ............... 146

Figure 5.10. Koutechy-Levich plots for the oxygen reduction reaction in 0.1 M
HClO4. Data taken from the polarization curves shown in Figure 5.9. ........ 148

Figure 5.11. Tafel plots for the oxygen reduction reaction in 0.1 M HClO4. Data
taken from the polarization curves shown in Figure 5.9. ............................. 149

Figure 5.12. Tafel plots for the oxygen reduction reaction at RDE #3 in 0.1 M (A)
HClO4, (B) H2804 and (C) H3PO4. Scan rate = 20 mV/s. Rotation rate
= 2500 rpm. ............................................................................................... 151

Figure 5.13. Tafel plots for the oxygen reduction reaction in 0.1 M HClO4 at
different Pt/diamond RDEs. Scan rate = 20 mV/s. Rotation rate =
2500 rpm .................................................................................................... 153

Figure 6.1. Cyclic voltammetric i-E curves for (A) Pt/diamond composite
electrode and (B) polycrystalline Pt foil electrode in 0.1 M HClO4 +
0.2 M CH30H (solid line) and 0.1 M HClO4 (dashed line). Scan rate =
50 mV/s. Active surface area of the composite electrode is 0.44 cm2
(Pt deposition time was 200 s) .................................................................... 169

Figure 6.2. Cyclic voltammetric i-E curves for Pt/diamond composite electrodes
in 0.1 M HClO4 + 0.6 M CH3OH. Scan rate = 50 mV/s. ............................. 172

Figure 6.3. Cyclic voltammetric i-E curves for the electrooxidation of 0.6 M
CH30H in different 0.1 M acid electrolytes at a Pt/diamond composite
electrode. Active surface area of the composite electrode is 0.44 cm2
(Pt deposition time was 200 s). Scan rate = 50 mV/s. ................................. 173

Figure 6.4. Current-time transients for a Pt/diamond composite electrode in 0.1 M
HClO4 + 0.6 M CH30H at different electrode potentials. Active
surface area of the composite electrode is 0.60 cmz. ................................... 175

Figure 6.5. Current-time transients for a Pt/diamond composite electrode
obtained after a potential step to 0.6 V vs. Ag/AgCl in 0.1 M HClO4
containing different concentrations of methanol. Active surface area of
the composite electrode is 0.60 cmz. ........................................................... 177

Figure 6.6. Log i vs. log c plot at 0.6 V vs. Ag/AgCl. Data taken from the
current-time transients shown in Figure 6.5. ............................................... 177

XV

Figure 6.7. AFM image (10 x 10 um) of a Pt/Ru/diamond composite electrode.
The deposition time is 300 s. ...................................................................... 180

Figure 6.8. Cyclic voltammetric i-E curves for the Pt/Ru/diamond composite

electrodes in (A) 0.1 M H00. and (B) 0.1M HClO4 + 0.2 M CH3OH.
The electrodeposition time is 300 3. Scan rate = 50 mV/s. .......................... 181

xvi

Chapter 1

Introduction

1.1 Electrocatalysis and Direct Methanol Fuel Cells

Electrocatalysis represents the combination of two important subdisciplines of

‘ The term, which first

physical chemistry, namely, electrochemistry and catalysis.
appeared as early as 1963 in a paper by Grubb,2 can be deﬁned as the study of
heterogeneous catalytic reactions that involve reactant and product species transfening
electrons through an electrolyte/catalyst interface.3 An electrocatalyst is a conducting
chemical entity that enhances the rate of an electrochemical reaction without being either
consumed or generated in the process. Since an electrocatalyst is part of the electrode,

sufficient electrical conductivity and electrochemical stability of the material at the

operating electrode potential are essential.

The electrochemical technology mostly influenced by advances in electrocatalysis
is electrochemical energy conversion, especially fuel cells. A fuel cell is a device that
converts the chemical energy of a fuel directly into electrical energy via electrocatalytic
reactions at an anode and cathode. The fuel cell is a nineteenth century invention.
However, no application of fuel cells was reported until the early 1960s. Recently, rising
concerns about the energy shortage and the environmental consequences of fossil fuel use
have led to a renewed interest in the fuel cell technology. For example, fuel cells are
promising candidates as power sources for electric vehicles. Unlike the case of Camot-

limited thermal engines, the free energy change in fuel cell reactions can be converted to

electrical energy (i.e., work) with a theoretical conversion efficiency of over 80%. In
addition, the low emission, lack of noise and low cooling requirement are additional

attributes.

There are several types of fuel cells under study and they are generally classiﬁed
according to the type of fuel and electrolyte, operating temperature, or direct or indirect
utilization of fuel. Hydrogen and methanol are two favored fuels. Hydrogen is the most
electroactive fuel for fuel cells operating at low and intermediate temperatures. Methanol
can be considered as a hydrogen carrier in a fuel cell. Conventionally, methanol has been
reformed to produce hydrogen. Methanol, however, can be fed directly in to the fuel cell
without the intermediate step of reforming. In this case, the fuel cell is called a direct

methanol fuel cell (DMFC).

In a DMFC, methanol is oxidized at the anode and oxygen is reduced at the
cathode (Figure 1.1A). The two electrodes are separated by an ionic conducting
electrolyte, either an aqueous solution or a proton exchange membrane. The direct

electrooxidation of methanol in acid media is given by the following reaction4

CH3OH +H20 —-) C02 + 6H+ + 6c
E°=0.016Vvs. SHE (1.1)

in which C02 and protons are formed as products. The protons migrate through the

electrolyte and combine with oxygen to produce water at the interface. The

electroreduction of oxygen (from air) can be written as

02 + 4H1 + 4e' ——) 21120
E“ = 1.229 v vs. SHE (1.2)

   
  

   
 

       
  
 

Anode

Electronlc conductor
+ Catalyst
C02 + 6H" + 6e'

Cathode

Electronic conductor
+ Catalyst

6e' + 6H“ + 3/20;

Electrolyte I
Ionic conductor I

 
  
 

CH30H

   

02

'0' H20

 

 

C02 3H20
Anode ' Cathode B
/ Vent Vent /
‘ ‘ ' ' ' ' Membrane Carbon
upport
Catalyst

 

 

I Cathode

Anode I
FGBd, 02

Feed, CH30H

 

Figure 1.1. (A) Schematic diagram of a direct methanol fuel cell working in an acidic
medium. (B) Schematic diagram of a proton exchange membrane fuel cell. Highly-

dispersed catalyst (small dots) is supported on carbon black particles (big dots).5

The thermodynamic cell voltage of a DMFC is 1.21 V. However, the practical
energy conversion efﬁciency is rather low due to the slow kinetics of both the anodic and
cathodic reactions, even with the state-of-the-art electrocatalysts. The catalysts routinely
used in low temperature fuel cells are precious metals and their alloys, such as Pt and
PtRu.6’7 Recently, a number of other catalyst systems have been investigated for methanol

oxidation, including the tertiary and higher Pt-based catalyst alloys —— PtRuOs,8 PtRuSn,9

PtRuW,'° PtRuSnW‘ ”2 and PtRuOsIr. '3

The slow kinetics, scarcity, and expense of noble-metal-based catalysts are
driving forces for ﬁnding ways to improve the catalytic activity. A high level of activity
is achieved by using ﬁnely dispersed catalyst, i.e. catalyst with a high speciﬁc surface
area supported on a high surface-area, electrically conducting material (Figure 1.1B).
Carbons and metal oxides are the most favored support materials.14 Boron carbide,
tantalum boride, and silicides of tungsten and titanium have been also reported as catalyst

supports due to their high corrosion resistance in aggressive chemical environments.”16

1.2 Carbon Support Materials

The primary role of the catalyst support is to provide a high surface area over
which small metallic particles can be dispersed and stabilized.'7 The porous support
should also allow facile mass transport of reactants and products to and from the active
sites. Several properties of the support are critical: porosity, pore size distribution, crush
strength, surface chemistry, and microstructural and morphological stability.‘7

The materials that appear to possess the best combination of these characteristics

are carbons. Carbonaceous materials offer microscopic to macroscopic porosity, which

can be readily tailored by controlling the manufacturing conditions, yielding the required
pore-size distributions. They also exhibit a chemical inertness, with no metal-support
interactions. Additionally, these materials can be burnt off, providing an economical way
to recover the precious metal. For electrochemical reactions, the attractive properties of
carbon supports are the electrical conductivity and good chemical and electrochemical
stability.

Typical carbon supports include carbon black, graphite, and graphitized carbon
black. These materials all contain spz-bonded carbon, but differ signiﬁcantly from one
another in terms of the particle size, porosity, and microstructure. The ideal graphite
structure consists of layers of carbon atoms arranged in fused hexagonal rings with an
interlayer spacing of 3.354 A. The parameters of importance are the coherence lengths of
the graphite crystallites along a-axis (14,) parallel and c-axis (LC) perpendicular to the
layer planes. Graphitic allotropes of carbon differ physically from each other in terms of
the graphite crystallite dimensions. For example, highly oriented pyrolytic graphite has

an 1.‘ larger than 10 um, whereas L, for graphite powders can be smaller than 100A.l8

Carbon blacks are considered to be amorphous carbons, consisting of randomly oriented
small graphitic crystallites with values for L, and Lc in the range of 15 to 25 A,
respectively.19 Carbon blacks are the most-often-used support materials and are available
in a variety of forms. By controlling the manufacturing conditions, carbon blacks can be
produced with particle sizes varying from less than 50 A to greater than 3000 A, leading

to a surface areas ranging from 1000 to 10 mZ/g.19

1.3 The Problem

Although spz-bonded carbons have some attractive properties as an electrocatalyst
support, a significant limitation of these materials is their susceptibility to corrosion,
particularly under the prevailing working conditions of a fuel cell. The corrosion could
either be catastrophic causing electrode failure (e.g., C02 evolution) or more subtle
involving microstructural and morphological alteration that lead to catalyst detachment.
Support corrosion may influence the overall electrode performance in several ways. First,
electrochemical corrosion of carbon is favored at the cathode operating potential, i.e.,
0.6-1.2 V. Carbon oxidation seems to establish a mixed potential at the cathode, thus
limiting the cathode working potential. Second, carbon surface oxides formed during the
electrochemical oxidation increase the electrical resistance of the carbon support.20 Third,
The gasiﬁcation product, CO, is a poison to Pt-based catalysts, causing decreased
catalytic activity. Fourth, limited carbon corrosion causes surface rnicrostructure changes
that affect the carbon-metal particle contact, extent of aggregation, the accessibility of the
catalyst particles within the support pores (so-called hidden catalyst effect). Fifth, severe
carbon corrosion is destructive to the electrode bulk, causing catastrophic mechanical
failure.

The importance of support stability has long been recognized. The investigation
of electrochemical oxidation of carbon, particularly under the fuel cell operating
conditions, can be traced back to the early 19603. Extensive studies were performed by

1.,20 Kinoshita and Bett,21 Stonehart22 and, Gruver.23 These studies provided

Binder et a
important insight into the mechanistic and kinetic aspects of the carbon corrosion

process. Carbon corrosion in acidic media occurs by at least two anodic reaction

pathways involving the evolution of gaseous products and the formation of surface

oxides. These reactions can be written as '9

C + 2H20 ———> C02 + 4H+ + 4e'
E“ = 0.207 v vs. SHE at 25°C (1.3)
and
C + HzO-——> C0+2H'+ 2e'
E" = 0.518 v vs. SHE at 25°C (1.4)

Both reactions are thermodynamically favorable, but, in general, kinetically slow. The
reaction rate strongly depends on the electrode potential, electrode rnicrostructure,
electrolyte composition, and pH. The rate is also affected at a given potential by other
factors such as the temperature and vapor pressure. In general, high temperature, high
pressure, and high operating potentials result in increased rates of corrosion.

Several approaches have been reported to increase the stability (i.e., corrosion
resistance) of carbon support materials. The fact that carbon corrosion generally initiates
at surface defect sites and crystallite plane edge sites resulted in the use of heat treatment
(graphitization) to reduce the number of surface active sites and thus reduce the corrosion
rate?"25 Doping carbon with boron or silicon has been also proposed to enhance the
corrosion resistance, as well as to introduce trap sites on the surface that inhibit the
catalyst aggregation.26

Considerable efforts have been directed toward the development of alternative
support materials. Meilbuhr27 reported that boron carbide might be a useful support for
the cathode in the alkaline (KOH) fuel cell. Mckee's'“5 observed that boron carbide,

tantalum boride, and silicides of tungsten and titanium were corrosion-resistant in 85%

H3P04 (150 0C) and 6 N H2804 (850 OC) and, since these materials have high electrically
conductivity, proposed that they might function as electrocatalyst supports. However, the

viability of these materials in fuel cells has not been demonstrated.

1.4 Diamond Electrochemistry

Unlike graphite and amorphous carbons, diamond is a carbon allotrope that only
recently has begun to be examined as an electrode material for electrochemical
technologies. Diamond has long been known for its unique mechanical, thermal, optical,
and electrical properties.” and "mews ”mi“ For example, diamond is extremely hard, inert
to corrosive reagents, resistant to radiation damage, and optical transparent in the visible
and infrared spectral regions of the electromagnetic spectrum. Historically, the material
has been used as an abrasive and as an optical window. Diamond is a wide band gap (5.5
eV) semiconductor with a conduction band close to the vacuum level. Therefore, negative
electron affinity is observed for hydrogen-terminated diamond surfaces, making it an
excellent choice for field emission devices.29 In addition, diamond is an attractive
material for high-temperature, high-frequency and high-power device applications
because of its high thermal conductivity (20 Wcm'l k'l), low dielectric constant (5.5),30
high breakdown field (106-107 Vern"), and high electron and hole mobility (2000 cmZV'
‘s'1 and 1800 cmZV’ls'l).3‘

The R&D activities drastically expanded when chemical vapor deposition (CVD)
of diamond became a reality in the early 19705. CVD affords the possibility of producing
diamond thin ﬁlms on a variety of substrates with relatively low cost. To date, hot-

ﬁlament and microwave-assisted CVD are the most popular deposition protocols, with a

CH4/H2 mixture routinely used as the source gas. Owing to the synthetic nature of
diamond growth, it is possible to manipulate the physical and chemical properties of
diamond by proper control over the gas composition, system pressure, substrate
temperature and reactor design. For example, microcrystalline diamond ﬁlms, with
typical crystallite size range from 1 to 10 um, can be readily deposited from a hydrogen-
rich hydrocarbon environment, whereas nanocrystalline diamond with crystallite sizes in
the 10 - 50 nm range, is deposited from a hydrogen-poor environment.32 The methane
concentration in the source gas mixture also has a signiﬁcant impact on diamond
nucleation, crystal size, and ﬁlm composition. Figure 1.2 shows a series of atomic force
microscope (AFM) images of ﬁlms deposited with different CH4/H2 ratios. Clearly, as the
CH4/H2 ratio increases so does the secondary growth rate leading to smaller crystallite
size. Also, the films become thicker, more opaque and have higher incidence of non-
diamond impurity and other defects with increased CH4/H2 ratio. Most importantly, CVD
allows for in situ doping of a variety of impurities such as boron, phosphorus, and sulfur
into the diamond lattice, rendering diamond p-type or n-type semiconductor properties.33
Boron is the most prominent dopant because of its small covalent radius, and can be
easily incorporated into substitutional sites within the diamond without causing lattice
distortion. Boron concentrations in excess of 102'cm’3 are possible.34

Diamond thin ﬁlms with sufficient electrical conductivity are attractive electrode
materials. The use of diamond in electrochemistry is a relatively new ﬁeld of research
that has begun to blossom in recent years.35'38‘ 3"" references "mm The ﬁrst paper was
published in 1983 by Iwaki et al.,39 but the field was actually initiated by the seminal

paper from Pleskov et al..40 Boron-doped, hydrogen terminated, polycrystalline diamond

Na

6000 0

6000.0

 

6000.0

 

3% 5%

Figure 1.2. Diamond ﬁlm morphologies as a function of methane-to-hydrogen source gas

ratio used during microwave plasma CVD. Hydrogen flow rate is 200 sccm. Growth time

is 10 h.

thin ﬁlms possess a number of important and practical electrochemical properties,
unequivocally distinguishing them from other commonly used spz-bonded carbon
electrodes, such as glassy carbon (GC), pyrolytic graphite, and carbon paste. These
properties are: (i) controllable electrical conductivity via doping; (ii) low and stable
voltammetric and amperometric background current; (iii) a wide working potential
window in aqueous and nonaqueous media; (iv) reversible to quasi-reversible electron
transfer kinetics for several inorganic redox systems without conventional pretreatment;
(v) superb morphological and microstructural stability at extreme anodic and cathodic
potentials; (vi) weak adsorption of polar molecules from aqueous solution, like
anthraquinone-2,6-disulfonate;41 (vii) long-term response stability; and (viii) optical
transparency in the UV/vis and IR regions of the electromagnetic spectrum. The

attractive properties of diamond have been exploited a variety of electrochemical tech-

42-46 47-49

nologies including electroanalysis, electrocatalysis, spectroelectrochemistry,50

bioelectrochemistry, and electrochemical-based toxic waste-detection and
remediation.5 1‘52

The as-deposited diamond surfaces are typically hydrogen-terminated.
Chemisorbed hydrogen is critical for stabilizing the structure and prohibiting phase
transformations (i.e., graphitization) during growth. The most reproducible electro-
chemical properties of diamond are observed for the nonpolar and hydrophobic
hydrogen-terminated surface. Hydrogen-terminated diamond lacks of surface sites for
adsorption and stabilization of reaction intermediates. Therefore, electrochemical

reactions that involve adsorbed intermediates or radicals can be strongly hindered on

diamond electrode.53 A good example of this is the water discharge reaction that involves

ll

the oxygen and hydrogen evolution reactions, both of which involve multistep processes
and formation of reaction intermediates. The suppressed kinetics for both reactions on
diamond result in a wide working potential window (3 — 4 V) in aqueous media. This is
an attractive feature because it allows electrochemical reactions to be monitored that are
screened on other spz-bonded electrodes. For example, zinc electrodeposition (E0 = -0.76
V vs. SHE) can be studied at diamond without severe interference from hydrogen
evolution. At the anodic end, aliphatic polyamine oxidation46 and halides evolution“55
has been reported. The nonpolar surface shows weak adsorption of polar molecules,
leading to improved resistance to electrode deactivation and fouling. The hydrogen-
terrnination also partially contributes to the low double layer capacitance of diamond

electrodes, 1 to 8 uF/cmz, compared to 30—40 uF/cm2 for glassy carbon. This affords

enhanced signal-to-background ratios in electroanalysis due to the low background
current and capacitance.

The most striking feature of diamond is its superb corrosion resistance and
electrochemical stability. The diamond stability during anodic polarization in acidic
ﬂuoride, alkaline, acidic chloride, and neutral chloride media has been investigated. For
example, Swain56 compared the corrosion resistance of boron-doped diamond, HOPG
and glassy carbon (GC) during potential cycling experiments in an acidic ﬂuoride
solution at 50 0C. Based on voltammetric, Raman spectroscopic, and capacitance
measurements, the morphology and rnicrostructure of diamond were comparatively
unchanged by the polarization while severe damage was observed for both HOPG and
GC. Katsuki57 found that ozone could be generated in H,SO, at boron-doped diamond thin
ﬁlms using a current density from 1 to 10 Acm'z. No structural damage of diamond such

as pitting, grain toughening, or film delamination, were observed. Chen et al.55 carried

12

out a comparative study of the corrosion resistance of diamond and HOPG for chlorine
generation at high anodic current densities. Polycrystalline diamond underwent no gross
morphological or rrricrostructural changes at anodic current densities of 0.1 A/cm2 for 12
plus hours in an acidic chloride media [E = 3-4 V vs. saturated calomel electrode (SCE)].
HOPG catastrophically corroded within 30 5 under the same conditions.

Figure 1.3 shows a comparison of the surface morphologies of diamond and GC
electrode during anodic polarization in an acidic medium as measured by in situ AFM. It
can be seen diamond undergoes no morphological change during polarization, whereas
the GC surface is severely roughened due to the alteration in the surface microstructure
and the formation of surface oxide layer. The diamond surface is easily converted to an
oxygen-termination during the anodic polarization, but the oxidation is restricted to the
surface. The structural and chemical robustness of diamond results from its high atomic
density and strong, directional covalent bonding, which inhibit bulk oxidation under all
but the most severe conditions (e. g., oxygen plasma treatment). The morphological and
microstructural stability contrasts with the severe bulk damage that occurs for many Sp2_
nonded carbon electrodes, which is caused by a combination of intercalation, oxidation,
and gasiﬁcation reactions. Therefore, diamond offers signiﬁcant improvements in
corrosion resistance and dimensional stability compared with more commonly used spz-
bonded carbon supports, and can be used in aggressive chemical environments at high
temperature and current density, including electrosynthesis, energy conversion devices,

and electrochemical—based toxic waste remediation.

l3

Diamond Glassy
Film Carbon

   

Figure 1.3. The effect of anodic polarization on the electrode morphology. Both the
diamond ﬁlm and glassy carbon electrodes were subjected to polarization at 2 V vs.
Ag/AgCl in 0.1 M HClO4. The in situ AFM images were recorded for polarization times

2, 10, and 15 min., respectively.

14

CVD diamond films are most often polycrystalline because the surface free
energy of diamond is high such that the 2D growth of diamond on a foreign substrate is
difﬁcult. Furthermore, the covalent C-C bonds in diamond are so rigid that diamond
nuclei, individually formed on a substrate, are very unlikely to coalesce without forming
grain boundaries.58 Certainly, polycrystalline thin ﬁlms have been the most studied form
of diamond electrode, thus far. The electron transfer kinetics at diamond electrodes are
inﬂuenced by several factors: (i) the dopant type and concentration, (ii) morphological
features, such as grain boundaries and extended and point defects, (iii) the non-diamond
or amorphous carbon impurity content, (iv) the primary crystallographic orientation, and
(v) the surface termination (H, F or 0, etc). The degree to which any of these factors
inﬂuence the electrode response strongly depends on the electrode reaction mechanism
for a particular redox analyte.

The boron doping level and uniformity are major factors governing the electrical

3

conductivity of diamond ﬁlms. Doping levels on the order of 1017 to 1018 cm' result in

35-405" At this level,

diamond thin films that exhibit semiconducting electronic properties.
conduction occurs mainly through holes in the valence band formed the thermal
promotion of electrons to empty states of the substitutional boron. The activation energy
is rather high (0.37 eV), therefore, only about 0.2 % of the boron sites are ionized at room
temperature.33 At higher doping levels, conduction occurs by nearest neighbor and
variable range hopping of holes between ionized sites. This is accompanied by a drop in
mobility.34 As very high doping level, 1019 cm'3 or greater, the overlapping wave

functions of nearby acceptor centers leads to band formation, and impurity band

conduction takes place.33 The activation energy for conduction decreases with doping

15

level as impurity band conduction dominates, and at sufficiently high boron
concentrations, the impurity band merges with the valence band leading to metallic
conduction. For highly doped films, activation energies as low as 0.002 eV have been
reported due to band formation.33 The boron doping level is also one of the major factors
governing the electrode kinetics. Boron doping introduces electronic states within the
band gap which mediate the charge transfer across the diamond/electrolyte interface. The
density of these states increases with the doping level.

Morphological defects and grain boundaries also affect the electrical conductivity
of diamond in a complex manner. On one hand, grain boundary conduction is known to
be dominant in undoped and low doped polycrystalline diamond films.60 Grain
boundaries are a source of disordered spz-bonded carbon, along which variable range
hopping occurs.61 On the other hand, the electrical conduction is inhibited by grain
boundaries and other defects at high doping levels because of a reduction in the electron
and hole mobilities.62 Defects can serve as discrete sites for electron transfer or can
simply affect the electronic properties of the material by increasing the density of states.
In addition, the grain boundaries could contain nondiamond carbon impurity phases
giving rise to additional midgap electronic states and reactive sites.53 Nondiamond
impurity is a major factor that affects the electrochemical performance of diamond in
terms of the working potential window and background current magnitude, electrode
stability, and adsorption.63 Compared with the diamond phase, the nondiamond carbon
phase is highly susceptible to oxidation, which introduces oxygen-containing functional

groups onto the electrode surface."4

16

1.5 The Motivations

We believe that polycrystalline diamond possesses properties ideally suited for an
active material support and current collector for batteries, and an electrocatalyst support
for fuel cells and electrosynthesis. The material possesses superior morphological
stability and corrosion resistance, compared to conventional sp2 carbon support materials,
being able to withstand current densities on the order of 1 A-cm'2 for days, in both acidic
and alkaline media, without any evidence of structural degradation. The material is
chemically inert allowing for its use at elevated temperatures in oxidizing or reducing
environments without loss of properties. The thermal conductivity of the material is near
that of copper at room temperature, such that diamond might be useful for thermal
management purposes. The rough, polycrystalline morphology might also serve to anchor
adsorbed metal catalyst particles or active material, leading to improved electrode
performance. Metal catalyst particles well anchored into the diamond matrix, rather than
just physically sitting on the surface, possess much improved electrode stability and
durability. Some of the problems observed with spZ-bonded carbon supported catalyst,
such as catalyst mobility, aggregation and detachment, particularly when operated at
elevated temperature, may be alleviated by incorporating the catalyst particles into the
diamond matrix. In addition, the diamond surface is H-terminated so some of the site
blocking effects often experienced with 0-terminated carbon supports may be minimized.
Surface carbon-oxygen functionalities terminating the nearby edge-plane sites of the spz-
bonded carbon supports can provide site blocking 0H moieties that diminish the Pt

catalyst activity.

17

The synthetic nature of the diamond film growth process, along with the strongly
reducing environment (i.e., high ﬂux of atomic hydrogen), allows for the possibility of
incorporating foreign metal particles during the deposition. There are several possible
ways one might consider incorporating metal impurities into the growing diamond film.
We have attempted to incorporate Pt nanoparticles into a diamond matrix by ﬁrst
growing a conductive diamond ﬁlm, then depositing dispersed Pt particles on top of the
conducting diamond, and followed by a secondary diamond film growth around the base
of the metal particles. Our ideas for this manner of metal inclusion were stimulated by

8 showed that highly

work performed by two groups. First, Tachibana and coworkers5
textured diamond ﬁlms can be grown on Pt(111) substrates. Highly oriented and
spontaneously coalesced diamond (111) faces can be grown in so-called “Shintani”
growth process. During the course of our electrochemical investigations of this material,
we observed that electroactive Pt nanoparticles were serendipitously incorporated into the
surface region of the ﬁlm. This observation led us to try and control the incorporation of

65 successfully incorporated

such catalyst particles. Second, Callstrom and coworkers
nanoscale metal clusters into an spz-bonded carbon matrix. Speciﬁcally, Pt was
incorporated into a “glassy carbon” matrix by low temperature pyrolysis of polymeric
precursors.

We expected that the incorporation of metal particles into diamond surface

rnicrostructure would result in a conductive, dimensionally stable electrode material

containing metal particles of controlled composition, size, and catalytic activity.

1.6 Outline of the Dissertation

l8

This dissertation is devoted to a new research area of diamond electrochemistry:
application of boron-doped diamond thin—film electrodes in the ﬁeld of electrocatalysis.
The main focus is on the fabrication and characterization of Pt/diamond composite
electrodes and on the investigation of their electrocatalytic activity.

If diamond is going to be effective catalyst support, a fundamental understanding
of how metals nucleate, grow, and adhere to the surface is needed. Chapter 2 describes
detailed studies of copper electrodeposition on boron-doped diamond thin-ﬁlm
electrodes, in comparison with the spz-bonded carbon electrodes. A unique aspect of the
work is the use of in situ atomic force microscopy to investigate the shape, spatial
distribution, and the nucleation/growth mechanism of the metal deposits. The copper
deposition depends strongly on the electronic and chemical properties of the electrode.
Hydrogen-terminated diamond exhibits poor interaction with the metal deposits. The size,
number density, and spatial distribution of metal deposits are controlled by adjusting the
applied overpotential, deposition time, and the electrical conductivity (boron-doping
level) of the diamond ﬁlm.

The methodology for fabricating the metal/diamond composite electrodes is
explored in Chapter 3. Our initial attempts to prepare the Pt/diamond composite
electrodes by a sequential diamond growth/Pt magnetron sputtering/diamond growth
fabrication procedure are described. Nanometer-sized Pt particles can be incorporated
into the diamond surface rnicrostructure. The dispersed Pt particles are stabilized by the
growth of a thin ﬁlm of diamond around their base and are in good electrical
communication with the current collecting substrate through the boron-doped diamond

matrix. Chapter 4 describes a similar multistep fabrication procedure, the difference

19

being that the dispersion of Pt onto the diamond surface is accomplished by
electrodeposition. The size and distribution of Pt particles can be controlled by adjusting
the electrodeposition and secondary diamond growth conditions. The composite material
is extremely stable with unchanging catalytic activity even after 2 h of anodic
polarization in aggressive chemical environments.

It is of speciﬁc interest to probe the electrocatalytic activity of the composite
electrodes for two important fuel cell electrode reactions: oxygen reduction (cathode) and
methanol oxidation (anode). Chapter 5 describes a study of the oxygen reduction reaction
kinetics at Pt/diamond composite electrodes. The study focused on the use of rotating
disk voltammetry to determine the kinetic parameters for the reaction as a function of the
Pt loading and electrolyte composition. Chapter 6 describes a study of the methanol
oxidation reaction kinetics at Pt/diamond composite electrodes. The kinetic parameters
were determined as a function of the Pt loading, methanol concentration, and catalyst
composition. The catalytic activity of the Pt/diamond composite electrodes is comparable
to that for polycrystalline Pt. Efforts to prepare the Pt/Ru/diamond composites were
made. Preliminary results indicated the electrocatalytic activity for methanol oxidation

reaction is greatly enhanced when modifying Pt with Ru.

20

References:

10.

11.

12.

13.

14.

15.

K. Christmann, in Electrocatalysis, J. Lipkowski and P. N. Ross, Editors, Wiley-
VCH, New York, 1 ( 1998).

W. Grubb, Nature, 198, 883 (1963).

A. J. Appleby, in Comprehensive Treatise of Electrochemistry, B. E. Conway, J.
0. M. Bockris, E. Yeager, S. U. M. Khan and R. E. White, Editors, Plenum Press,
New York, 7, 173 (1983).

C. Lamy and J. M. Leger, in Interfacial Electrochemistry: Theory, Experiment,
and Applications, A. Wieckowski, Editor, Marcel Dekker, New York, 885 (1999).

C. Lamy, J. M. Leger, and S. Srinivasan, in Modern Aspects of Electrochemistry,
R. E. White, Plenum, New York, 34, 53 (2001).

M. Watanabe and S. Motoo, J. Electroanal. Chem. Interfacial Electrochem, 60,
259 (1975).

M. Watanabe and S. Motoo, J. Electroanal. Chem. Interfacial Electrochem, 60,
267 (1975).

K. L. Ley, R. Liu, C. Pu, Q. Fan, N. Leyarovska, C. Segre, and E. S. Smotkin, J.
Electrochem. Soc., 144, 1543 (1997).

A. Aramata and M. Masuda, J. Electrochem. Soc., 138, 1949 (1991).
C. He, H. R. Kunz, and J. M. Fenton, J. Electrochem. Soc., 144, 970 (1997).

A. S. Arico, Z. Poltarzewski, H. Kim, A. Morana, N. Giordano, and V. Antonucci,
J. Power Sources, 55, 159 (1995).

A. S. Arico, P. Creti, N. Giordano, V. Antonucci, P. L. Antonucci, and A.
Chuvilin, J. Appl. Electrochem, 26, 959 (1996).

E. Reddington, A. Sapienza, B. Gurrau, R. Viswanathanm, S. Sarangapani, E. S.
Smotkin, and T. E. Mallouk, Science, 280, 1735 (1998).

K. Kinoshita and P. Stonehart, in Modern Aspects of Electrochemistry, R. E.
White, Plenum, New York, 12, 53 (1977).

L. W. Niedrach and D. W. Mckee, in Proceedings of the 21st Annual Power

Sources Conference, May 16-18, 1967, PSC Publications Committee, Red Bank,
NJ, 6.

21

l6.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

W. T. Grubb and D. W. Mckee, Nature, 210, 192 (1966).

E. Auer, A. Freund, J. Pietsch, and T. Tacke, Appl. Catalysis A: General, 173,
259 (1998).

R. L. McCreery, in Interfacial Electrochemistry: Theory, Experiment, and
Applications, A. Wieckowski, Editor, Marcel Dekker, New York, 631 (1999).

K. Kinoshita, Carbon: Electrochemical and Physicochemical Properties, John
Wiley & Sons, New York, 22 (1988).

H. Binder, A. Kohling, K. Richter, and G. Sandstede, Electrochim. Acta, 9, 255
(1964).

K. Kinoshita and J. A. S. Bett, Carbon, 11, 237 (1973).

P. Stonehart, Carbon, 22, 423 (1984).

G. A. Gruver, J. Electrochem. Soc., 225, 1719 (1978).

J. B. Donnet, P. Ehrburger, and A. Voet, Carbon, 10, 737 (1972).

Z. Wei, H. Guo, and Z. Tang, Journal of Power Sources, 62, 233 (1996).

P. Stonehart and J. P. MacDonald, Stability of Acid Fuel Cell Cathode Materials,
Research Project RP 1200-2, Final report for Jan.l, 1978 to May 31, 1984,
Prepared for the Electric Power Research Institute, Palo Alto, CA, 1985.

S. G. Meilbuhr, Nature, 210, 409 (1966).

L. S. G. Plano, in Diamond: Electronic Properties and Applications, D. R. Kania,
Kluwer Academic Publishers, Norwell, MA, 62 (1995).

B. B. Pate, in Diamond: Electronic Properties and Applications, L. S. Pan and D.
R. Kania, Editors, Kluwer Academic Publishers, Norwell, MA, 45 (1995).

C. Chen and S. H. Chen, Diamond and Related Materials, 4, 451 (1995).

V. K. Bazhenov, I. M. Viokulin, and A. G. Goonar, Sov. Phys. Semicond, 19, 829
(1985).

D. M. Gruen, Annual Review of Materials Science, 29, 211 (1999).
8. Han, L. S. Pan, and D. R. Kania, in Diamond: Electronic Properties and

Applications, L. S. Pan and D. R. Kania, Editors, Kluwer Academic Publishers,
Norwell, MA, 260 (1995).

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

45.

46.

47.

48.

49.

50.

51.

K. Nishimura, K. Das, and G. J. T., J. Appl. Phys, 69, 3142 (1991).

Y. V. Pleskov, Russian Chemical Reviews, 68, 381 (1999).

R. Tenne and C. Levy-Clement, Israel Journal of Chemistry, 38, 57 (1998).

G. M. Swain, A. B. Anderson, and J. C. Angus, MRS Bulletin, 23, 56 (1998).
A. Fujishima and T. N. Rao, Diamond and Related Materials, 10, 1799 (2001).

M. Iwaki, S. Sato, K. Takahashi, and H. Sakairi, Nucl. Instrum. Methods, 209,
1129 (1983).

Y. V. Pleskov, A. Y. Sakharova, M. D. Krotova, L. L. Builov, and B. V. Spitsyn,
J. Electroanal. Chem. Interfacial Electrochem, 228, 19 (1987).

J. Xu, Q. Chen, and G. M. Swain, Anal. Chem, 70, 3146 (1998).
J. Xu and G. M. Swain, Anal. Chem, 70, 1502 (1998).

A. Manivannan, M. S. Seehra, D. A. Tryk, and A. Fujishima, Anal. Lett., 35, 355
(2002).

M. C. Granger, J. Xu, J. W. Strojek, and G. M. Swain, Anal. Chim. Acta, 397, 145
(1999).

B. V. Sarada, T. N. Rao, D. A. Tryk, and A. Fujishima, Proceedings -
Electrochemical Society, 99-32, 502 (2000).

M. A. Witek and G. M. Swain, Anal. Chim. Acta, 440, 119 (2001).

J. Wang, G. M. Swain, T. Tachibana, and K. Kobashi, Proceedings -
Electrochemical Society, 99-32, 428 (2000).

J. Wang, G. M. Swain, T. Tachibana, and K. Kobashi, Electrochem. Solid-State
Lett., 3, 286 (2000).

J. Wang, G. M. Swain, T. Tachibana, and K. Kobashi, J. New Mater.
Electrochem. Sys., 3, 75 (2000).

J. K. Zak, J. E. Butler, and G. M. Swain, Anal. Chem, 73, 908 (2001).

C. Terashima, T. N. Rao, B. V. Sarada, D. A. Tryk, and A. Fujishima, Anal.
Chem, 74, 895 (2002).

23

52.

53.

54.

55.

56.

S7.

58.

59.

60.

61.

62.

63.

64.

65.

T. N. Rao, B. H. Loo, B. V. Sarada, C. Terashima, and A. Fujishima, Anal.
Chem, 74, 1578 (2002).

M. C. Granger, M. Witek, J. Xu, J. Wang, M. Hupert, A. Hanks, M. D. Koppang,
J. E. Butler, 0. Lucazeau, M. Mermoux, J. W. Strojek, and G. M. Swain, Anal.
Chem, 72, 3793 (2000).

N. Vinokur, B. Miller, Y. Avyigal, and R. Kalish, J. Electrochem. Soc., 143, L238
(1996).

Q. Chen, M. C. Granger, T. E. Lister, and G. M. Swain, J. Electrochem. Soc., 144,
3806 (1997).

G. M. Swain, J. Electrochem. Soc., 141, 3382 (1994).

N. Katsuki, S. Wakita, Y. Nishiki, T. Shimamune, Y. Akiba, and M. Iida, Jpn. J.
Appl. Phys, Part 2, 36, L260 (1997).

T. Tachibana, Y. Yokota, K. Miyata, T. Onishi, K. Kobashi, M. Tartutani, Y.
Takai, R. Shimizu, and Y. Shintani, Phys. Rev. B, 56, 15967 (1997).

C. Reuben, E. Galun, H. Cohen, R. Tenne, R. Kalish, Y. Muraki, K. Hashimoto,
A. Fujishima, and J. Butler, J. Electroanal. Chem, 396, 233 (1995).

B. Fiegl, R. Kuhnert, M. Ben-Chorin, and F. Koch, Appl. Phys. Lett., 65, 371
(1994).

Y. Muto, T. Sugino, J. Shirafuji, and K. Kobashi, Appl. Phys. Lett., 59, 843
(1991).

G. A. Sokolina, A. A. Botev, L. L. Builov, S. V. Banysekov, O. I. Lazareva, and
A. F. Belyanin, Sov. Phys. Semicond, 24, 105 (1990).

J. Bennett and G. M. Swain, unpublished results.

H. B. Martin, A. Argoitia, J. C. Angus, and U. Landau, J. Electrochem. Soc., 146,
2959 (1999).

M. R. Callstrom, T. X. Neenan, R. L. McCreery, and D. C. Alsmeyer, J. Am.
Chem. Soc., 112, 4955 (1990).

24

Chapter 2

Electrochemical and in situ Atomic Force
Microscopic Studies of Copper Deposition on
Boron-Doped Diamond Thin-Film Electrodes

2.1 Introduction

There has been a growing interest over the years in the fundamental
understanding of the electrochemical properties of boron-doped diamond thin ﬁlms”, and
“menus "mi" Such interest has resulted from the promising applications of diamond in the
field of electroanalysis, electrosynthesis, electrocatalysis, and electrochernical-based
toxic waste remediation. Diamond is also an interesting electrode material for
fundamental research, including the study of electron-transfer processes at the
electrode/electrolyte interface. In contrast to many other carbon-based (e. g. glassy carbon
(GC) and highly oriented pyrolytic graphite (HOPG)) and metal electrodes, diamond does
not form a significant interfacial (oxide) layer when brought to contact with an aqueous
solution. Moreover, it does not undergo microstructure alteration over a wide potential
range.

A signiﬁcant motivation of previous and current research is the need to develop a
predictive understanding of the structure-reactivity relationship of boron-doped diamond
thin-ﬁlm electrodes. We have established a multistep electrochemical characterization
protocol to assess diamond electrode responsivenesss"8 Our goal is to correlate the

physical, chemical, and electronic properties of the material with its electrochemical

responsiveness. Several electrochemical probes are routinely employed to characterize

25

the electrodes, including Fe(CN)6'3/'4, Ru(NH3)6+3/+2, methyl viologen (MV+2’+/O), Few”,
catechols, and oxygen reduction. These redox systems are used because each is sensitive
to different physicochemical properties, and their formal potentials span a wide potential
range (~ +1.1 to —0.8 V vs. SHE).7

Metal electrodeposition is another type of surface “sensitive” charge-transfer
process that can be used to probe the reactivity of diamond electrodes. The original metal
deposition work on diamond was conducted by Awada et al., in which they demonstrated
that Pt, Pb, and Hg were readily deposited on a diamond surface.9 Moreover, metal
deposition on diamond electrodes is of great practical signiﬁcance because (i)
electrodeposition allows for fabrication of various passive and active electrical contacts
on the diamond surface, as metallization processes will be indispensable for establishing
electronic devices based on diamond technology;3 (ii) it is the groundwork for developing
diamond-based anodic stripping voltammetric detection of heavy metal ions; and (iii)
surface roughness, corrosion resistance and chemical stability of polycrystalline diamond
ﬁlms combined with the catalytic activity of metal adlayers afford the possibility of
developing an advanced class of highly sensitive and stable electocatalytic surfaces. For
these reasons, electrodeposition of Ag,'0 Hg,”ll Pt,12 Pb13 and Cum‘16 on boron-doped
diamond electrodes has been recently reported.

We present herein the first detailed investigation of copper electrodeposition on
diamond electrodes using a combination of electrochemical and microscopic techniques.
The study was aimed at elucidating (i) the mechanism of metal nucleation on diamond,
(ii) the structure-reactivity relationship at diamond by examining metal nucleation and

growth, and (iii) the capability of diamond serving as a support/host for electrocatalytic

26

metal particles. Copper electrodeposition was examined on diamond films with different
boron-doping levels and the results were compared with those obtained for GC and
HOPG. The diamond ﬁlms used had a boron doping level ca. 1019 and 102' cm'3, which
are referred to as low boron-doped diamond (LBD) and highly boron-doped diamond
(HBD), respectively. The nucleation mechanism (progressive vs. instantaneous
nucleation) was studied by examining initial stages of copper deposition with cyclic
voltammetry (CV) and chronoamperometry. Electrochemical atomic force microscope
(ECAFM) was used to in situ monitor the new phase formation on diamond during
deposition. The location and shape of the metal deposits provide important information
regarding the spatial reactivity (grain boundary vs. grain reactivity) of the diamond
electrodes. The nucleation mechanism revealed by ECAFM was in good agreement with

the electrochemical data.

2.2 Experimental

Electrode preparation. Boron-doped diamond thin ﬁlms were deposited on p-
Si(100) substrates using microwave-assisted chemical vapor deposition (CVD) (Kobe
Steel Ltd., Japan). A summary of the growth conditions is shown in Table 2.1. The
substrates were 0.1 cm thick and 1 cm2 in area. The source gas used was a CH4/H2
mixture with a CH; concentration of 0.2-0.6 %. The gas pressure was 50 torr, and the
microwave power was 0.8-1 kW. The substrate temperature was about 875 0C. The boron
dopant source was Bsz and the bom-doping was controlled by adjusting the volumetric
ratio of B2H6 in the source gas. The boron dopant concentrations were estimated to be

approximately 1019 to 1021 B/cm3 (in the ﬁlm), based on the secondary ion mass

27

spectrometry analysis of other films deposited using similar conditions. The apparent in-

plane ﬁlm resistivities ranged from 0.1 to 0.01 52cm, and were measured using a

tungsten four-point probe with the diamond ﬁlm attached to the conducting Si substrate.

Table 2.1. Summary of the diamond thin—ﬁlm deposition parameters.

 

Substrate p-Si (100)
CH4/H2 (v/v) ratio 0.2-0.6 %
Microwave power 0.8-1 kW
Deposition pressure 50 torr
Deposition time 20 hrs
Substrate temperature 875 °C
Boron dopant source B2I16
Doping level (in ﬁlm) 1019-102] cm"3

 

Glassy carbon (1 cm2, GC-30, Tokai Ltd., Japan) was prepared by polishing on
separate felt pads with successively smaller grades of alumina powder slurried in
ultrapure water (1.0, 0.3, and 0.05pm). The electrode was rinsed thoroughly with and
ultrasonicated for 5 min. in ultrapure water after each polishing step. A fresh basal plane
of HOPG (Advanced Ceramics Corp., USA) was exposed by cleaving the topmost layers
with scotch tape. This cleaving process leaves a variable amount of adventitious defects
on the surface.

Electrochemical measurements. The cyclic voltammetric and chronoampero-
metric measurements were performed in a single compartment glass cell using a CYSY-
2000 digital potentiostat (Cypress Systems, Inc., Lawrence, KS). Figure 2.1 shows a

diagram of the cell. A commercial saturated calomel electrode (SCE) was used as the

28

reference and a large area carbon rod served as the counter electrode. The diamond film
electrodes were pressed against the bottom of the glass cell with the ﬂuid being contained
by a Viton 0-ring. The geometric area exposed was 0.2 cm2 and all currents were
normalized to this area. The large area counter electrode was positioned parallel to the
working electrode. The reference electrode was positioned near the working electrode
using a glass capillary that was ﬁlled with 50 mM H2S04. The end of the capillary was
cracked to allow for ion conduction. Electrical connection was made to the diamond in
one of three ways with each method generally providing similar results: (i) scratching the
backside of the Si substrate with a diamond scribe and then coating the area with graphite
before contacting the A] current collecting backplate, (ii) same as above but coating the
area with Ag paste, or (iii) same as above but running a bead of Ag paste from the top
edge of the diamond ﬁlm to the backside of the Si substrate. The Al plate was polished
clean prior to use. Care must be taken to avoid solution leakage which can cause
formation of salt deposits on the backside of the Si substrate and impede good contact
with the current collecting plate. Sometimes, an In/Ga eutectic was used to make ohmic
contact to the backside of the scratched Si substrate. All measurements were made at
room temperature with solution deoxygenation by a nitrogen purge gas. The diamond
ﬁlm working electrodes were pretreated, once mounted in the cell, by a thorough rinsing
with ultrapure water (>17 MQ-cm, Bamstead Nanopure), a 20 min. soak in distilled
isopropanol, followed by another thorough rinsing with ultrapure water. Isopropanol can
be an effective cleaning agent for spz-bonded carbon electrodes.13 A three-step cleaning
procedure was used for all glassware: washing in a KOH/methanol bath, washing in a

liquid detergent/water bath (Alconox, Inc.) and rinsing with ultrapure water.

29

 

\/
>s

|\/O

 

 

 

 

l2 ,
a g r

 

 

 

 

 

if

 

Figure 2.1. Diagram of the single-compartment, glass electrochemical cell. (a) Cu or Al
metal current collecting back plate, (b) working electrode, (c) Viton O-ring, ((1) input for
nitrogen purge gas, (e) carbon rod or Pt counter electrode, and (f) reference electrode

inside a glass capillary tube with a cracked tip.

ECAFM measurements. The ECAFM images were acquired using N anoscope II
atomic force microscope (Digital Instruments Inc., Santa Barbara, CA) equipped with a
standard ECAFM ﬂuid cell. Figure 2.2 shows the diagram of the ﬂuid cell. The cell has
tapered channels for allowing the ﬂow of the electrolyte and also the insertion of

auxiliary and reference electrodes. Both were platinum wires. The ﬂuid cell was sealed

30

by pressing the working electrode against the bottom of the cell, isolated by a Viton O-
ring. The exposed surface area of the electrode was 0.38 cmz. A pyramidal S13N4 450 tip,
mounted on a gold-coated 100nm v-shaped silicon nitride cantilever (0.38 N/m spring
constant), was used for imaging. The tip was held in an optically transparent holder in the

ﬂuid cell. The AFM was operated in the contact mode.

 

 

 

 

 

 

 

 

 

Figure 2.2. Diagram of the ECAFM ﬂuid cell. (a) photodetector, (b) He-Ne laser, (c)
auxiliary electrode, (d) reference electrode, (e) tip, (f) solution inlet, (g) solution outlet,

(h) working electrode, (1) O-ring, and (j) piezoscanner.

ECAFM images were obtained in either the height mode, in which the tip height

was adjusted via a feedback mechanism so as to maintain the force constant, or force

mode in which the tip height was kept constant so as to measure the force. A potentiostat

31

was used to perform the potential cycling or potential step measurements. In ECAFM, all
potentials were recorded against the Pt quasi-reference electrode. The potential difference
between Pt and SCE electrode was ca. 0.3 V in 50 mM H2804. For convenience, all
potentials are reported, here after, converted to the SCE reference scale. All the images

were recorded with 400 x 400 pixels. All measurements were made at room temperature.
Chemicals. The CuSO4-5H20 was reagent grade quality (Fisher Scientific) and

the H2804 acid electrolyte was ultrahigh purity grade (99.999%) (Aldrich). Ultrapure

water (>17MQ-cm) from a Bamstead E-pure system was used to prepare the solutions,

clean the glassware and rinse all electrodes.

2.3 Theories of Metal Nucleation and Growth

The early stages of electrochemical phase transformations are usually associated
with 3-D nucleation process followed by diffusion-controlled growth of the mature
nuclei.l7 These processes are generally studied by chronoamperometry, in which the
electrode potential is stepped to a value cathodic enough to cause diffusion-controlled
metal deposition, and then the potentiostatic current-time transient is recorded. Reliable
information regarding the number density of nuclei on the surface, nucleation rate, and
nucleation mechanism can be obtained from the analysis of the current transient.

The nucleation/growth during the early stages of deposition is illustrated in Figure
3.3A.l7 A number of hemispherical nuclei are randomly formed on the electrode surface
at short times after the potential step. Because of their small size and isolation, the growth
is described in terms of localized spherical diffusion to individual microelectrodes. This

corresponds to the rising part of the current transient (see Figure 3.3B). As the radius of

32

the growth center increases and the hemispherical diffusion zones of various nuclei begin
to overlap, the hemispherical diffusion gives way to linear diffusion to a planar electrode
surface.17 Consequently, the current reaches a maximum and then decays at a t'”2 rate as
predicted from the Cottrell equation. Such a 3D nucleation/growth mechanism has been
described in several theoretical models.”18

In the approach of Scharifker and Hills,17 nucleation is classified as either
“instantaneous”, in which rapid nucleation takes place on a ﬁx number of active sites, or
“progressive”, in which the number of nuclei increases with time and the number of

active sites available for nucleation are considered to be virtually infinite.

For instantaneous nucleation, the current density is given by

, nFDmc
1:—

1/2 1/2
71 t

[1 — exp(-N7d<Dt)] (2.1)

where D is the diffusion coefﬁcient, c is the bulk concentration of metal ions, nF is the

molar charge transferred during electrodeposition, and k is
k = (8722M / p)“2 (2.2)
where M and p are the molecular weight and density of the deposited metal, respectively.
For progressive nucleation, the current density is given by

._ nFDWc

t...
1/2 1/2
72 t

[1 - exp(—AN0721<'Dt2 /2)] (2.3)
and

- 4 .12
k = 3487th / ,o) (2.4)

where No is the number density of active sites over the substrate surface, A is the

nucleation rate constant per site.

33

 

 

 

 

 

 

 

 
  

 

 

 

 

 

 

\‘\\\\\‘\\\_\\\\11\ 1 .111

 

 

 

’
“---- I
ail---- ’
I
..-—u I
“’0’- r’
I
’1
, I
' /
I ”
4' I’
I I
/’ \///’/
I I
’I ‘\I”/’
,, III!
’a II,
I
,, \\‘\01“.ll"[l'/
Q
h
. B

Current density

 

 

 

 

 

Time

Figure 2.3. (A) Schematic representation of the growth of the diffusion zones an their
eventual overlap. The arrows indicate the directions of the diffusional ﬁeld during the
growth of the nuclei. (B) Typical current—time transient recorded during the potential step

after background correction.

34

For convenient analysis, the transients are generally presented in dimensionless
form by plotting (i/iM)2 vs. t/tM, where iM and tM are the current and the time at the

maximum, as in Figure 238. They are

tri=weir)"ital-12564011
til“2254(5)"ll'expl'mlélll

for instantaneous and progressive nucleation, respectively.
In the approach of Scharifker and Mostany,18 it is not necessary to classify the
nucleation process as instantaneous or progressive, since these cases appear as the

extremes of a more general mechanism. The entire current transient can be expressed as

1/2 _ _
i = "FTIZU—f—[r — exp{— N07sz[t - (1 6"? M M (2.7)
The dimensionless form is

._ . l ...i-._+a[1exii)i)%llll
[ l U {._...[-.+all-exi‘§llll

where x = NottthM, and or = NottkD/A. In this approach, the A and N0 can be calculated

 

Qlk

 

(2.8)

1M

separately.
In this work, we analyze the experimental results according to both approaches.

The nucleation mechanism of copper on diamond surface is illustrated by the ﬁrst

35

approach, and the potential dependence of the nucleation rate and number density of

active sites are determined by the second approach.

2.4 Results

2.4.1 Electrochemical Study of Copper Deposition

Cyclic voltammetry. The electrodeposition of copper on a low (LBD) and highly
(HBD) boron-doped diamond, as well as HOPG and GC, was initially studied by cyclic
voltammetry. Figure 2.4 shows the voltammetric i-E curves recorded in Nz-purged 1 mM
CuSO4 + 50 mM H2804, The potential sweep was initiated at 0.4 V and scanned in the
negative direction at a scan rate of 20 mV/s. Each i-E curve represents the first sweep.
The overall reaction is'9

Cu2+ + 2e <—> Cu E“ = 0.098 v vs. SCE. (2.9)
A summary of the voltammetric data is shown in Table 2.2.

The voltammograms exhibit the characteristic features of nucleation and growth
phenomena, particularly for LBD and HOPG, i.e., the large negative shift of the reduction
peak potential and the current crossover during the reverse scan. The equilibrium
potential, Ecq, for Reaction (2.9) in 1 mM CuSO4, is ca. 10 mV vs. SCE, as calculated
from the Nemst equation. Therefore, the overpotential needed to initiate the formation of
the bulk metal phase ranges from —110 to —240 mV, and increases in the order GC <
HBD < LBD < HOPG. No current crossovers, however, were observed in the i-E curves
for CC and HBD. Rather, the cathodic process most likely proceeds in two steps. In both

cases, discernible cathodic current starts at ca. 70 mV, leading to the formation of a

36

shoulder peak prior to the steep rise of the current. The shoulder peak is presumably due
to the reduction of trace amount of Cu+. The reaction can be written as 20

Cu+ + e 9 Cu E“ = 0.28 V vs. SCE. (2.10)

which occurs at more positive potential.

 

j—Vﬁ ' f ‘v V ' v v r t v v v ' v v v ' v v V

 

 

H
8 . .
i-i
I‘='l B i
U - i

 

 

 

 

 

 

Potential (mV vs. SCE)

Figure 2.4. Cyclic voltammetric i-E curves for copper deposition on (A) GC, (B) HBD,

(C) LBD, and (D) HOPG in 1 mM CuSO4 + 50 mM H2804. Scan rate = 20 mV/s.

Table 2.2. CV data for copper electrodeposition on different electrodes in 1 mM CuSO4

 

 

+ 50 mM H2804.

Electrode Onset ipred Epred Ep"x AEp Qox/Qrcd

reduction (uA) (mV) (mV) (mV)

potential 3

(mV)

GC -100 -57.6 -132 26 158 0.94
HBD -138 -55.2 -180 36 216 0.90
LBD -185 -50.3 -255 31 286 0.73
HOPG -230 -46.2 -320 43 363 0.85

 

a. Onset reduction potentials for Reaction 2.9.

The voltammetric data for all four electrodes are shown in Table 2.2. The onset
reduction potential for copper deposition is greatest for HOPG and least for GC. The two
diamond ﬁlms have values in between. The reduction peak current for each electrode
increased linearly with the square root of scan rate (r2 = 0.998) over the range from 20 to
200 mV/s, suggestive of a diffusion-controlled process. In contrast, the oxidation peak
current increased rather linearly with the scan rate (r2 = 0.992), indicative of a surface-
conﬁned process. Very broad oxidation peaks are observed for both diamond surfaces in
contrast to that for GC, which are most likely attributed to the heterogeneity of both the
diamond surface structure and electronic properties. Similar results were obtained for Ag
and Hg deposition on diamond.10 In addition to the broad stripping peaks, the anodic

currents were observable at potentials positive of the equilibrium value. Integration of the

38

areas under both anodic and cathodic peaks, after background subtraction, yields the
oxidation and reduction charge. As shown in Table 2.2, the anodic stripping charge is ca.
10-25% lower than the cathodic deposition charge, indicating that the copper deposits do
not completely strip during the anodic scan, at least during the time scale of the
measurement. Indeed, some residual copper particles were detected on the LBD surface
in the ECAFM study, as described below.

It should be noted that copper deposits weakly adhere to the diamond surface, a
phenomenon that has been observed in the case of Pb,9 Ag and Hg'0 electrodeposition on
diamond. The deposits can be easily dislodged by the turbulence of solution mixing. The
weak deposit-substrate interaction is attributed to the low activity of hydrogen—terminated
diamond surface.l4

Chronoamperometry. Single-step Chronoamperometric measurements were
carried out to investigate the metal nucleation/growth process in greater detail. Figure 2.5
presents a set of current—time transients for copper deposition on LBD from 1 mM CuSO4
+ 50 mM H2804, The measurements were performed by stepping the potential from 0.5
V, where copper was not deposited, to different ﬁnal potentials, Er, sufficiently cathodic
to induce nucleation and subsequent crystallite growth. Note the potentials shown in
Figure 2.5 are overpotentials, t] = Ef — Eeq. The working electrode was poised at a
positive potential of 0.6 V for 5 rrrin after each step in order to completely oxidize the
copper deposits. This led to a good reproducibility of consecutive measurements
performed under the same conditions.

The current increases immediately after the application of the potential step and

decreases at very short times, as the electrode double layer fully charges. The current then

39

increases as a result of the formation of stable nuclei and growth of the metal deposits. As
the individual hemispherical diffusion zones of the growing crystallites coalesce, the
current passes a maximum, iM, then decays at a I'm rate predicted from Cottrell equation.

The values of 1M and tie], the time at which iM occurs, are listed in Table 2.3.

 

 

 

 

 

 

 

 

3.5 - . - . - . . . e .
30’ -710 J
'E .
9 .
E»
‘2
Q)
'5
‘8'
t:
3
o
0.0...........
0 50 100 150 200 250 300
Time (ms)

Figure 2.5. Potentiostatic current-time transients for copper deposition on LBD in 1 mM

CuSO4 + 50 mM H2804 at the overpotentials indicated in mV.

Analysis of the current transients and classifying the nucleation process. The
current-time transients were initially analyzed according to Schariﬂ<er and Hills’
approach, i.e., to classify the nucleation process as “instantaneous” or “progressive”.

Figure 2.6 shows some examples of the experimental data (dashed curves) in comparison

40

with the theoretical curves (solid curve a for instantaneous and b for progressive
nucleation). The theoretical curves are constructed by plotting (i/iM)2 vs. t/tM according
with Equation 2.5 and 2.6. The actual experimental data without background correction
are also presented (dashed line) in the dimensionless plots. The ﬁt for any of the
experimental curves with entire the instantaneous or progressive curves is not perfect, but
the data do tend to more closely match one lirrriting case or the other at all potentials. The
deviation from the limiting situations is a typical phenomenon for heterogeneous
nucleation on a finite number of active sites on the surface.21 The nucleation of copper on
LBD appears to follow an instantaneous nucleation mechanism at low overpotentials (ml
< 610 mV), and a progressive mechanism at high overpotentials, (ml > 610 mV).

The Chronoamperometric measurements were also performed on HBD, GC and
HOPG. The current-time transients for copper deposition on HBD are shown in Figure
2.7. The iM increases and tM decreases with increased boron doping level (i.e., the
conductivity of the diamond film). The comparisons of the experimental current-time
transients with the theoretical curves are shown in Figure 2.8. The experimental data
deviate from both limiting cases, but follow closely the response predicted for
progressive nucleation at both low (-410 mV) and high overpotentials (-610 mV).

Analysis of the current transients without classifying the nucleation process. In
Schariﬂ<er and Mostany’s approach, it is not necessary to classify the nucleation process
as one of the two limiting cases. The parameters A and No can be estimated separately
from a single point of the experimental transients, such as the current maximum, or by

nonlinear ﬁtting of the entire current-time transients to the theoretical expression

41

(MM)2

(01,02

(i/iM)2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1.2 12 - 1
1.0 ’ a ' 1'0 L ,oo 'Wmm-wu
“MW...” '
as b “A 0.8 ' /a
0.6 - / SE 0.6 i ,
La
04 P1 04 .... .
0.2 . " A . 02 . B .
0.0 ‘ . ‘ 0.0 . . ‘
0.0 0.5 1.0 1.5 20 0.0 0.5 1.0 1.5 20
I/tM tltM
1.2 - 1 1.2 - -
1.0 b 1.0 '
0.8) I «3‘ “i ,2”
x" 2 /
0.6 . /o' S 0.6 ' [In
t' V I
0.4 . r " “4' /
0.2- "
0.2 ’ C '0 D
x
0.0 ‘ -
o.“ n A A
0.0 0.5 1.0 1.5 2.0 ”'0 1'0 1'5 7'"
t/tM tltM
1.2 - g 1.2 -
1.0 . 1.0»
..\‘“ "I
0.8 l “\‘.‘.\ r (“A 0.8 ’ I
. "x. 2
0.6 » s 0-6'
,I’ v '1
.1 ,.
0.4 i I, 0.4 . /
/’ '
03 ’ . E 0.2 r F
0.0 ‘ . 0.0 - '
0.0 05 1.0 1.5 2.0 0.0 1 o 1.5 2.0
t/tM mm

Figure 2.6. Comparison of experimental current-time transients (dash curves) for LBD

(see Figure 2.5) with the theoretical transients corresponding to instantaneous (solid

curve a) and progressive (solid curve b) nucleation at different overpotentials: (A) —410,

(B) -460, (C) —510, (D) -610, (E) -660, and (F) —710 mV.

42

 

 

Current density (mA-cm'z)

 

 

 

' 0 50 100 150 200 250 300
Time (ms)

Figure 2.7. Potentiostatic current-time transients for copper deposition on HBD in 1 mM

CuSO4 + 50 mM H2804 at the overpotentials indicated in mV.

 

 

 

 

 

 

 

 

 

1.2 . v . 1.2
1.0 1
0"}

0": 0-8 I /" 6:3
S 0.6 ' I" \‘ x‘ S
v I, v

0.4 l /

/
0.2 VI A
0.0 ‘ ‘ . ‘ ‘
0.0 0 5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
t/tM t/tM

Figure 2.8. Comparison of experimental current-time transients (dashed curves) for HBD
(see Figure 2.7) with the theoretical transients corresponding to instantaneous (solid

curve a) and progressive (solid curve b) nucleation at overpotentials of (A) —410 and (B)

—610 mV.

43

Table 2.3. Current—time transient data for copper electrodeposition on LBD.

 

 

.11 1M tM thM "’1 /a 10'7 NO A 10‘7 N...
(mV) (mA-cm‘z) (S) (cm'z) (s‘) (cm'z)
360 0.913 0.226 0.7890 0.82 26.9 —
410 1.075 0.165 0.7953 1.17 31.9 —
460 1.232 0.128 0.8027 1.58 37.1 —
510 1.375 0.105 0.8081 2.02 41.7 —
560 1.660 0.073 0.8155 3.06 53.4 4.28
610 1.961 0.054 0.8285 4.63 59.2 5.54
660 2.567 0.033 0.8463 9.15 71.7 8.57
710 2.970 0.025 0.8620 14.3 80.7 11.4

 

Table 2.4. Current-time transient data for copper electrodeposition on HBD.

 

 

-n 1,. IM thM ‘4 la 10'8 NO A 10'8 Ns
(mV) (mA-cm‘Z) (s) (cm‘z) rs") (mi)
360 1.280 0.127 0.8007 0.14 37.9 0.24
410 1.742 0.074 0.8297 0.30 42.2 0.38
460 2.382 0.042 0.8498 0.66 53.0 0.63
510 2.896 0.029 0.8608 1.13 60.2 0.87
560 3.481 0.020 0.8706 1.94 65.3 1.19
610 4.250 0.014 0.8807 3.70 69.0 1.69
660 5.167 0.010 0.8848 6.47 79.7 2.40

 

Table 2.5. Current-time transient data for copper electrodeposition on HOPG.

 

 

 

 

 

-n 1,, tM thM V“ /a 10'7 No A 10” Ns
(mV) (mAms) (s) (cm'Z) (3") (mi)
360 0.557 0.652 0.7491 0.18 20.8 —
410 0.795 0.340 0.7729 0.40 22.7 -—
460 1.098 0.190 0.7980 0.83 26.7 —
510 1.415 0.120 0.8172 1.51 31.8 ——
560 1.884 0.070 0.8324 2.94 42.7 3.75
610 2.201 0.052 0.8383 4.20 52.0 4.94
660 2.555 0.039 0.8410 5.79 66.4 6.56
710 2.927 0.030 0.8450 7.89 80.4 8.42

Table 2.6. Current-time transient data for copper electrodeposition on GC.

41 iM tM thM V” /a 10'8 N0 A 10'8 N,
(mV) (mA-cm’z) (S) (cm’z) (s‘) (cm’z)
360 2.680 0.044 0.8589 0.59 61.4 0.64
410 3.127 0.033 0.8679 0.94 65.2 0.83
460 3.545 0.026 0.8734 1.36 70.7 1.04
510 4.070 0.020 0.8794 2.09 72.5 1.30
560 4.570 0.016 0.8832 3.02 77.5 1.62
610 5.103 0.013 0.8855 4.13 86.1 1.99
660 5.877 0.010 0.8888 6.54 90.8 2.58

 

45

 

 

 

 

 

 

 

 

1.2 . . - - -
1.0 » B
6‘A 0.8 ’ i
'33 ...5. 0,
s 5
V 0.4 >
0.2 i “a...“ .
l. MN.”’°'»..
0.0 . - a
0.0 2 0 4 0 6.0 8 0 10 0

 

Figure 2.9. Comparison of experimental current—time transients (dashed curves) with the
theoretical transients according to Equation 2.8 for copper deposition on LBD at

overpotentials of (A) —410 and (B) —610 mV.

 

 

 

 

21.0
i
20 0 O 0
. HBD ,
' A
19.0 . LBD °
’3 t o A
z A
V 18.0 ' ° ‘
5 . A .
o A
17.0 ' A
A
A
16.0 'A
15.0 m n - 1 - A - L - n - A A A A
350 400 450 500 550 600 650 700 750
'11 (mV)

Figure 2.10. Plot of the logarithm of the number density of active sites vs. overpotential,

constructed from the data given in Table 2.3 and 2.4.

46

(Equation 2.7). In this section, we have estimated the values of A and No from the
experimental current maximum of the transients by solving a system of equations that are

derived from Equation 2.7'8

. 1/2
ln[l—iM—ti’—]+x-a(l—e_"m)=0 (2.11)
a
ln[1+2x(1—e""’a)]-x+o:(l—e""“) =0 (2.12)

where a = nfDV’c/lt V’. The relationship between the two approaches is as follows. There
are two limiting cases of the general mechanism proposed in the second approach, small
or and large 01, corresponding to the two extreme cases described in the first approach.
For 01 —> 0 (instantaneous nucleation), thMV’ /a = 0.7153, x z 1.2564; for or —> oo
(progressive nucleation), thM” /a .- 0.9034, x = 2.161801 V’.

The diffusion coefﬁcient of Cu”, D, is needed to interpret the data. The diffusion

coefﬁcient can be determined from the decay portion of the current-time transient using

112
i= IIFC(2] (2.13)
7a

The estimated value of D in 1 mM CuSOa + 50 mM H2804 is (2.5 i 0.5) ><10‘5 cmzs",

the Cottrell equation,

higher than that expected for typical diffusion coefﬁcient of Cu2+ in aqueous solution,22
i.e., 7.8 x1045 cmzs'l. It is important to note that it is impossible to solve the Equation

2.11 and 2.12 using the literature reported value. Moreover, the value of D estimated

from the current transient is in good agreement with that calculated from the cyclic

47

voltammetric study. Therefore, the value of 2.5 x10'5 cmzs'l is used in this work to

calculate the kinetic parameters for copper electrodeposition.

The current-time transient data and calculated results for copper electrodeposition
on LBD, HBD, HOPG and GC are listed in Table 2.3 - 2.6, respectively. The accuracy of
the calculation was studied. Figure 2.9 shows the comparison of the experimental data
with the theoretical current transients for copper deposition on LBD at both low and high
overpotential. The theoretical curves were constructed according to Equation 2.8 using
the calculated values of A and N0 listed in Table 2.3. It is clear that the Schariﬂrer and
Mostany’s approach adequately describes the copper deposition process on these

substrates.

In all cases, the reaction rate, reﬂected by the current maximum, iM, increases
with applied overpotential, whereas tM decreases. In other words, the time period required
for the coalesce of the hemispherical diffusion zone of each mature nuclei, which serves
as the growth center, decreases with overpotential, implying an increase in the number
density of growth centers. The observation is consistent with the fact that the number
density of sites, No, that are active toward copper nucleation, increases with
overpotential. Figure 2.9 shows the dependence of No for copper deposition on diamond
electrodes with overpotential. A rather linear relationship (r2 = 0.997) between the ln(N0)
and 11 is observed for copper nucleation on HBD within the overpotential range from —
360 to —660 mV. However, the plot seems to exhibit two distinct slopes on LBD. At high
overpotentials, the increase of N0 with overpotential is comparable to that on HBD,

whereas at low overpotentials the active sites toward copper nucleation are restricted and

48

the potential-dependence of N0 diminishes. The value of thMV’la, varying between 0.7
and 0.9, indicates that the copper nucleation does not follow the extreme “instantaneous”
or “progressive” nucleation mechanism, which is typical for heterogeneous nucleation on
a ﬁnite number of active sites on surface. However, the value increase with overpotential
and the nucleation approaches the progressive limit on each substrate at high
overpotentials.

The kinetics of copper deposition also strongly depend on the nature of the
substrate. The lowest number density of active sites is observed on HOPG because of the
low density of electrochemically active surface defects. Both No and A increase in the
order of HOPG < LBD < HBD < GC. The current maximum is increased by a factor of
two while the boron doping level of the diamond film increases from 1019 to 1021 cm'3.
Furthermore, the number density of active sites for copper deposition on I-IBD is nearly
one-order of magnitude higher than that on LBD at the same overpotential. However, the
values of No in all cases are much smaller than the atomic density of the substrate (ca.
1015 cm'z), which indicates that the restricted density of active sites on the surface may
constitute a severe limitation for the formation of copper nuclei.

For the case of progressive nucleation, the steady-state nucleation rate over the
entire surface can be expressed as the product of the nucleation rate per active site, A, and
the number density of active sites, No. As the nucleation rate decreases continuously due

to the decline of the No available for nucleation, the saturation number density of nuclei,

attained at long times, is given by‘7

N. =( AN“ J (2.14)

 

49

The meanings of k’ and D are the same as in Section 6.3.

In the case of 3D nucleation/growth, the growth centers develop into isolated 3D
crystallites. Therefore, the saturate number density of nuclei actually reﬂects the number
density of the isolated metal crystallites distributed on the surface. The estimated values
of NS for copper deposition on different substrates are also listed in Table 2.3-2.6. The NS,

ranging from 107 to 108 cm“, increases with overpotential.

2.4.2 ECAFM Study of Copper Deposition

Electrochemical studies of the mechanism and kinetics of metal deposition are
macroscopic in nature. They provide integral measurements over the whole surface.
However, local information concerning the structure of the substrate and the deposit is
needed if one wishes to build up a more rigorous model of the process occurring at the
microscopic level. Scanning probe microscopy (SPM) has proven to be a useful aid in
understanding the fundamentals of electrodeposition at the atomic level.23'27 The ECAFM
study of copper deposition on diamond thin-ﬁlm electrodes is presented in this section.
The study has a focus on the measurements of potential-dependent number density of
metal particles, their spatial distribution, and the nucleation mechanism.

Surface morphology of the carbon electrodes. The carbon electrodes employed
in this work differ from one another in terms of the surface microstructure, chemical
composition and electronic properties. The surface morphology of the electrodes is
illustrated in Figure 2.11 (A-D). The diamond surface is polycrystalline consisting of
well faceted microcrystallites and numerous grain boundaries. The diamond crystals

range in diameter from 0.5 to 3 pm. The diamond films exhibit signiﬁcant surface

50

roughness, ca. 1.4 i 0.5 um. Grain boundaries, formed from the coalescence of adjacent
grains, are the major defects in polycrystalline ﬁlms.28 There are also extended defects
observed on the grains. Both the grain boundaries and surface defects may serve as
active sites for metal deposition. As expected, the structure and morphology is not

appreciably altered by boron doping.29

 

Figure 2.11. AFM images (force mode, in air, 5 x 5 urn) of (A) LBD, (B) HBD, (C)

HOPG, and (D) GC.

51

HOPG is an ordered spz-bonded carbon material, in which the graphite sheets are
oriented on a macroscopic (i.e. centimeter) scale.30 As seen in Figure 2.11C, freshly
cleaved HOPG surface is characterized by atomically ﬂat domains (basal planes) and
differently structured defects (edge planes). Cleavage edges (lines shown in Figure
2.11C) are the most often encountered defects on HOPG, whereas some pits and crevices
are also observed. GC has graphitic domains limited to 5 — 10 nm, and is isotropic at
larger distance scales.30 As seen in Figure 2.11D, the freshly polished GC surface has a “
grainy” texture due to the carbon microparticle layer that forms during the abrasive action
of the mechanical polishing.3| The mean roughness over the 0.5 x 0.5 pm2 area, is 45 i 6
nm. This apparent roughness is very near that of the 50 nm diameter alumina grit used in
the ﬁnal polishing step.

It has been well established that the electronic properties of the “edge” planes, or
defects on HOPG surface, differ drastically from the basal planes. The edge planes
provide sites with higher density of “localized” electronic states, favoring facile electron
transfer.30 For example, the electron-transfer rate for Fe(CN)6'3M and Ru(NH3)6+3'+2 is
approximately four orders of magnitude faster on edge plane than on the basal plane.30
Therefore, the electrochemical properties of the spz-bonded carbon materials are strongly
dependent on the fraction of edge plane exposed at the surface. GC, generally with a
fractional coverage of edge plane near 100%, exhibits metallic properties for a variety of
systems. In contrast, HOPG, with a typical coverage less than 1%, behaves more like a
semimetal.30

Copper electrodeposition on LBD. The copper deposition on LBD from lmM

Cu804 + 50 mM H2804 was investigated by ECAFM. The copper deposition, as a

52

function of the applied overpotential, was first examined. As illustrated in Figure 2.12A,
the electrode potential was sequentially stepped to different overpotentials (i.e., 390,
-460, -560, -710, and 390 mV). A series of images (Figure 2.12B-F) were collected in
situ at 30 s after the application of each potential step. It should be noted that each AFM
image was collected by scanning the tip from the bottom to the top, and it generally took
20 s to complete each imaging. Therefore, each image corresponds to a 20-8 segment of
the deposition time. In other words, the copper deposits shown in the upper part of the
image are deposited for longer period of time than the ones at the bottom. The triangular
markers in Figure 2.12A indicate the starting time of each imaging.

Figure 2.123 shows the AFM image of LBD surface after the electrode potential
was stepped to an overpotential of 390 mV. As expected, no copper deposition takes
place. The image represents part of a bare diamond surface, which is identical with that
obtained at open circuit potential. After the potential was stepped from an overpotential
of 390 mV to —460 mV, a number of copper crystallites were generated, as can be seen in
Figure 2120 The copper crystallites have sizes ranging from 20 nm to 500 nm. It is
clear that copper deposits are not uniformly distributed over the surface. The copper
nucleation most likely initiates from a limited number of active sites which are
predominantly located at the grain boundaries or defects in the diamond surface. As the
potential was stepped to higher overpotentials (-560 and —710 mV), the increase in
surface coverage of copper deposits is evident (Figure 2.12D and E). On one hand, the
initially deposited copper crystallites seem to serve as the growth centers (e.g., Crystallite
a) and they increase in size with time at a given overpotential. 0n the other hand, new

copper crystallites were formed (e.g., Crystallite d and e in Figure 2.12D, and Crystallite

53

11 (mV)

li§§§.§§§

 

 

s
E
i
S
E

 

F

 

Figure 2.12. ECAFM images (force mode, 10 X 10 um) collected for copper deposition
1 on LBD at overpotentials of (B) 390 mV, (C) —460, (D) —560, (E) —710, and (F) 390 mV
for 30 s each. The solution is 1mM CuSOa + 50 mM H2804. (a)-(i) represent individual
copper crystallite. The diagram of the sequential potential steps is shown in (A). The

triangular markers indicate the starting time of each imaging.

54

Figure 2.13. ECAFM images (height mode, 2.5 X 2.5 urn) collected for copper
deposition on LBD in 1mM CuSOa + 50 mM H2S04 (A) before and (BE) after the
potential step to —460 mV. The deposition times are (B) 30, (C) 60, (D) 120, and (E) 180
s, respectively. Corresponding cross sections, through the top view images shown in the
middle column, are indicated by the straight lines. The horizontal distance between the

two black cursors represents the lateral dimension of the crystallite.

55

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

56

Tl (mV)

gli§§§c§§§

 

 

0 30 60 ”1201501802102”
Time (s)

 

 

Figure 2.14. ECAFM images (force mode, 4 x 4 11m) collected for copper deposition on
LBD in 1mM CuSO4 + 50 mM H2S04 (B) before and (C-F) after the potential step to —
710 mV. The deposition times are (C) 30, (D) 60, (E) 120, and (F) 180 s, respectively.
(a)—(f) represent individual copper crystallite. The diagram of the potential step is shown

in (A). The triangular markers indicate the starting time of each imaging.

57

f and g in Figure 2.12E). The increase of number density of deposits with overpotential
was evident in the Chronoamperometric studies. Furthermore, the number density of
copper deposits, estimated by manually counting the particles in the AFM image, is about
107-108 cm'z, in good agreement with the theoretical calculations (see Table 2.3). It is
also noticeable that some copper crystallites (e.g. Crystallite b and c) disappeared as the
potential was stepped to the overpotential of —560 mV. This is presumably due to the
occasional displacement of the microparticles by the AFM tip since the copper deposits
poorly adhere to the diamond surface. After an overpotential of 390 mV was applied, the
anodic stripping of copper deposits took place. Figure 2.12F shows an AFM image
recorded after 30 8. Clearly, most of metal deposits are gone and the image is almost
identical to that shown in Figure 2.12B, except for some residue copper crystallites (e.g.,
Crystallite h and i). The result confirms the incomplete stripping of the copper deposits
from LBD surface, a phenomenon that was observed in the electrochemical studies.

The copper deposition, at a given overpotential as a function of deposition time,
was next exarrrined. In the following experiments, the electrode potential was stepped to a
ﬁxed cathodic overpotential and the gradual development of the copper deposits was
monitored. Figure 2.13 presents a series of images taken before and after the electrode
potential was stepped to a low overpotential (i.e., —460 mV). Figure 2.13A shows the
diamond surface prior to the copper deposition. The region consists of three well-faceted
diamond microcrystallites and a “deep valley” (grain boundary) between two adjacent
grains. As expected, the defective grain boundary provides active sites toward copper
deposition, and a copper crystallite (indicated by the arrow in Figure 2138), was formed

after application of the potential step. The diameter of the crystallite increases with time.

58

The growth is apparently three-dimensional. The cross section analysis provides a more
quantitative assessment of the 3D growth process, which is also shown in Figure 2.13.

The copper crystallite has a lateral dimension of ca. 1.3 um and a height of ca. 0.2 pm

after 3 min growth. It’s evident that no other copper crystallites were formed in the
course of the deposition while the existing one increased in size, which suggests an
instantaneous nucleation mechanism. This is in good agreement with the electrochemical
measurements.

A series of AFM images (Figure 2.14) were also collected after the electrode
potential was stepped to a relatively high overpotential (i.e., -710 mV). As seen in Figure
2.14C, a number of copper crystallites (e.g., Crystallite a, b and c) were formed at the
grain boundaries upon application of the potential step. The further development of the
crystallites is shown in Figure 2.14D-F. In contrast to that observed at low overpotential,
while the existing ones increased gradually in size, some new copper crystallites were
formed (e.g., Crystallite d in Figure 2.14D, and Crystallite e and f in Figure 2.14E). It is
likely that continuous activation of surface sites toward copper nucleation/growth occurs
at high cathodic overpotential on LBD, thus progressive nucleation is favored. This is
also consistent with the electrochemical results derived from the current-time transient
measurements.

Copper deposition on HBD. The copper deposition on I-IBD was also
investigated by ECAFM. Figure 2.15 presents a series of images taken for copper
deposition at different cathodic overpotentials. As revealed in Figure 2.15A, the chosen
diamond surface region is free of any copper deposits at an overpotential of 390 mV.

Copper nucleation/growth takes place upon stepping potential to an overpotential of —460

59

mV. The estimated number density of copper crystallites, ca. 108 cm'z, falls into the range
predicted by the theoretical calculation (see Table 2.4). The number density of copper
deposits increases with overpotential as new copper crystallites are continuously formed
(Figure 2.15D). At —710 mV, the significant overlap of adjacent copper crystallites and
the nucleation/ growth over the existing copper deposits take place.

Some interesting features for copper deposition on HBD are noticeable. Copper
deposits not only initiate at the grain boundaries, but also decorate the grain surfaces
(indicated by the arrows in Figure 2.15D). At the same overpotential, the number density
of copper deposits on HBD is much higher than that on LBD. This suggests that the
number density of surface active sites toward copper nucleation increases with boron
doping level. Complete stripping of copper is observed upon stepping potential back to
the overpotential of 390 mV. As can be seen in Figure 2.15F, the surface was returned to
a state identical to that before the copper deposition (Figure 2.153).

Boron doping level can also affect the copper nucleation mechanism. The copper
deposition, at a given overpotential as a function of time, was examined. Shown in Figure
2.16 are AFM images taken at the same time sequence as that in Figure 2.13, at an
overpotential of -460 mV. Copper deposition on I-IBD follows a progressive nucleation
mechanism, which is in contrast to the observation on LBD. As evidenced by Figure
2.16E, some new copper crystallites (e. g., Crystallite c, d and e) were formed at the grain

boundaries as the electrodeposition proceeded.

60

 

n(mV)
li§§§.§§§

 

    

110 100 240 300
Time (s)

F
Figure 2.15. ECAFM images (force mode, 5 x 5 pm) collected for copper deposition on
HBD at overpotentials of (B) 390 mV, (C) —460, (D) —560, (E) —710, and (F) 390 mV for
30 5 each. The solution is 1mM CuSOa + 50 mM H2S04. The diagram of the sequential

potential steps is shown in (A). The triangular markers indicate the starting time of each

imaging.

61

11 (mV)

ii§§§.§£§

 

    

 

ONGONINISOIMZWM

Time (s)

Figure 2.16. ECAFM images (force mode, 3.4 X 3.4 pm) collected for copper deposition
on HBD in 1mM CuSOa + 50 mM H2804 (B) before and (GP) after the potential step to
—460 mV. The deposition times are (C) 30, (D) 60, (E) 120, and (D) 180 s. (a)-(e)
represent individual copper crystallite, respectively. The diagram of the potential step is

shown in (A). The triangular markers indicate the starting time of each imaging.

62

2.5 Discussion

Metal electrodeposition taking place at an electrode—solution interface involves
two basic processes: the discharge of solution ions and the formation of new phase on the
surface. Therefore, both the electronic properties of the substrate and the chemical
natures of the surface can affect the electrodeposition process. Diamond is widegap
semiconductor. There are two important mechanisms for electron transfer between a
semiconductor and a redox couple in solution: direct electron exchange with conduction
band or valance band, and exchange mediated by rrridgap electronic states.

Figure 2.17 shows the schematic interfacial energy diagram for a polycrystalline
diamond electrode, with boron doped at a level of 10'9 cm'3 or greater.7 In the case of
copper deposition, direct electron exchange is unlikely to occur because the standard
reduction potential of Cu”!0 falls into the band gap region. Therefore, the midgap-state-
mediated electron transfer dominates. As a consequence, the copper deposition kinetics
are governed by the density of midgap electronic states. Boron-doped, polycrystalline
diamond electrodes deviate from ideal p-type semiconductor behavior because of a high
density of midgap electronic states. Granger et 81. has proposed that the midgap density
of states results from at least four factors: (i) boron-doping level, (ii) lattice hydrogen
content, (iii) inherent grain boundaries and other defects in the polycrystalline film, and
(iv) nondiamond carbon impurity phases at the surface.7

The boron doping level and uniformity are major factors governing the electrical
conductivity of diamond ﬁlms, hence the electron transfer kinetics at the
diamond/electrolyte interface. Boron doped diamond can possess semiconductor to

semimetal properties, depending on boron doping level. At boron doping on the order of

63

1019 cm”3 or greater, impurity band conduction dominates. At sufficiently high boron
concentration, the impurity band tends to emerge with the valence band leading to
metallic conduction. Diamond is also host for a variety of extended defects such as
stacking faults, microtwins, dislocations, nondiamond carbon impurities, and grain
boundaries. The defects could serve as discrete sites for electron transfer or could simply

affect the electronic properties of the material by increasing the density of states.7
EVAC (—4.84V)
Egg (-4.94V) :::l

Localized
Electronic

  

States
0 V
0.098 V Cu+2’°
Boron Dopant
Band
EVB (0.516V)

Figure 2.17. Proposed band structure for the diamond/electrolyte interface. Evg, Eon and
EVAC are the energies at the bottom of the conduction band, at the top of the valance band
and in the vacuum. The potentials are quoted versus SCE. The band positions are

determined using a ﬂat-band potential of 0.45 V vs. SCE and a band gap of 5.45 eV.

The observations for copper deposition at diamond electrode can be correlated to

its semimetal-semiconductor nature. Distinct copper nucleation overpotentials are

observed. The overpotential for HBD is greater than for CC by some 40 mV, whereas,
the overpotential for LBD is greater by some 80 mV. The overpotential is inﬂuenced by
two processes: the overpotential related to electron transfer (m), and the overpotential
related to crystallization (11c). It is likely that n, dominates in the case of copper
nucleation on LBD. Diamond electrodes with a low boron concentration have a relatively
low density of states within the midgap. Therefore, high activation energy is required for
electron hoping between these electronic states. Furthermore, electrons supplied by the
electrode to the solution metal ions need to tunnel through a space charge layer. A
potential drop within the electrode bulk might take place. Since the width of the space-
charge layer decreases and density of states increases with boron doping, r], is expected to
decrease as the boron-doping level increases.

The overpotential related to crystallization, no, is inﬂuenced by the substrate since
crystallization involves the incorporation of metal adatoms into the surface
rnicrostructure. According to the euqation32

AG(N) = -Nze|t7| + <l>(N) (2. 15)
the Gibbs free energy of crystal formation is reduced as the overpotential increases (i.e,
the -Nze|r]| term). 011 the other hand, part of the energy is consumed by the new phase
formation (i.e, the <I>(N) term). <I>(N) denotes the total energy mainly associated with the

formation of new interfaces: crystal/solution and crystal/substrate. It can be written as32

<I>(N)=Zo,A,+Aj.(oj—ﬁ) (2.16)

where o, and A, are the specific surface energy and surface area of respective

crystallographic face i confining the crystal form. Gj and Ajrk are the speciﬁc surface

65

energy and surface area of crystal face contacting with the substrate. [3 is the specific

adhesion energy between the crystal and the substrate.

Thereby, the smaller <I>(N), the smaller overpotential required to initiate the
nucleation. Strong metal-substrate interaction (i.e., large [3) leads to a reduced total

surface energy associated with the new phase formation. Therefore, small crystallization
overpotential is required. Hydrogen-terminated diamond surface is nonpolar and
chemically inert, implying a weak interaction between the deposits and the substrate.
Therefore, a relatively higher 11,, is expected. In contrast, GC surface is covered with a
variety of oxygen-containing functionalities. Strong interaction between metal and
surface oxides has been reported.33

Surface sites associated with localized electronic states (surface states), mediating
electron transfer at the diamond/electrolyte interface, may serve as the active sites toward
c0pper nucleation. The population of surface states vary drastically upon polarization of
the electrodes in different potentials.3 Therefore, increases in number density of active
sites with increased cathodic overpotentials are observed on both diamond electrodes.
However, the trend for LBD is dirrrinished at low overpotentials. It is presumably due to
the high activation energy of the boron sites and the non-uniform distribution of the
surface states. At potentials below a threshold (i.e, ca. —610 mV), most of the localized
states remain inactivated and may not participate in the charge transfer at the
electrode/electrolyte interface. Accordingly, copper nucleation is restricted to very few
active sites on the surface which become exhausted at an early stage in the process,
nucleation thus approaches the instantaneous limit. Most of the surface sites are activated

at high overpotentials. In this case, progressive nucleation takes place.

66

The low density of states and the non-uniformity in the energy distribution might
lead to a heterogeneous electronic structure on the LBD surface. It is our assumption that
the surface of LBD has regions of high electrical conductivity isolated from each other by
regions of lower conductivity. Allongue34 has studied metal deposition on
semiconductors, also suggesting that the density of charge carriers on the semiconductor
surface is small and they must be considered as individual reactants. The fact that copper
nuclei formed elusively at the intercrystalline grain boundaries on LBD suggests the grain
boundaries are the most active regions toward copper deposition. Indeed, electrically,
grain boundaries introduce states within the rrridgap which trap charge carriers and
mediate the charge transfer across the diamond/electrolyte interface.35 Chemically,
nondiamond carbon impurities tend to concentrate at the grain boundaries,”3 making them
susceptible to oxygen attack in the aqueous solution.37 As discussed above, the formation
of oxides might reduce the copper nucleation overpotential.

Incomplete copper stripping is observed at the anodic polarization on LBD, which
could be also related to the heterogeneous electronic structure. Copper nuclei always
initiate at the electrically active sites and then the crystals are gradually spread over to
less or non-active sites. During the anodic process, copper deposits which are connected
to the active sites can dissolve but dissolution of the apparently poorly contacted rest of
the copper deposits does not occur until enough anodic potential applied. According to a
“undercutting” model postulated by Adzic et al,38 the faster dissolution of the part of
deposit closer to the surface might occasionally cuts off the contact of the rest part of the

deposit with the electrode, leaving them as residues on the surface.

67

The density of states within the midgap induced by boron doping increases with
the doping level.3 The number density of active sites increases with increased boron-
doping level, in consistent with the increased density of surface states. Moreover, the
active sites are no longer restricted at the grain boundaries as evidenced by the formation
of copper deposits on the diamond grain surfaces. In other words, both the grains and
grain boundaries are equally electrochemically active toward the copper deposition on
HBD. The observation is consistent with our recent studies of single crystal diamond

139

ﬁlm. Chen et a investigated the electrochemical behavior of diamond (100) electrode.

Relatively small AEps are observed for Ru(NH3)6‘“2/+3 and MV+2M0 which have formal
potentials well into the band gap region. Since the grain boundaries and defects are
negligible at this electrode, the results suggest that the density of states on the single
crystal surface, induced by boron doping (> 1019 cm'3), are sufficient high to support

facile electron transfer at the diamond/electrolyte interface.

2.6 Conclusions

The copper electrodeposition on boron-doped diamond thin-film electrodes was
investigated using cyclic voltammetry, chronoamperometry, and in situ atomic force
microscopy, in comparison with that on spz-bonded carbon materials, i.e., GC and
HOPG. It is shown that the copper nucleation/growth depends strongly on the electrical
and chemical natures of the substrate.

Distinct nucleation overpotential is observed for copper deposition on diamond
electrodes, presumably due to the chemical inertness of the diamond surface and/or

sluggish electron transfer at the diamond/electrolyte interface. The overpotential

68

decreases with increased boron doping level. A potential-dependent nucleation
mechanism was observed on low boron-doped diamond electrode. Copper nucleation at
relatively low cathodic overpotentials is restricted to few active sites, which are
exclusively located at grain boundaries and other surface defects, thus an instantaneous
nucleation mechanism follows. The density of states increases with overpotential, and the
progressive nucleation occurs at high overpotentials. Incomplete anodic stripping of
copper was observed, presumably due to the heterogeneity of the surface electronic
structure.

The number density of active sites toward copper nucleation increases by a factor
of 10 when the boron doping is increased from 1019 to 1021cm’3. In contrast to low boron-
doped diamond electrode, the surface active sites are no longer restricted at grain
boundaries. The diamond grains are also electrochemically active at high boron doping
level, and a progressive nucleation mechanism is obtained over a wide potential range.

in situ microscopic strudies shows that the copper electrodeposition on boron-
doped diamond film electrode follows a 3D nucleation/growth mechanism. It is apparent
that the size of the metal deposits, number density and spatial distribution are potentially
to be controlled by adjusting the applied overpotential, deposition time and electrical

conductivity (i.e., boron doping level) of the diamond film.

69

References:

10.

ll.

12.

l3.

14.

15.

16.

17.

J. Xu, M. C. Granger, Q. Chen, J. W. Strojek, T. E. Lister, and G. M. Swain, Anal.
Chem, 69, 591A (1997).

G. M. Swain, A. B. Anderson, and J. C. Angus, MRS Bulletin, 23, 56 (1998).
R. Tenne and C. Levy-Clement, Israel Journal of Chemistry, 38, 57 (1998).
Y. V. Pleskov, Russian Chemical Reviews, 68, 381 (1999).

S. Alehashem, F. Chambers, J. W. Strojek, G. M. Swain, and R. Ramesham, Anal.
Chem, 67, 2812 (1995).

M. C. Granger and G. M. Swain, J. Electrochem. Soc., 146, 4551 (1999).

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

M. C. Granger, J. Xu, J. W. Strojek, and G. M. Swain, Anal. Chim. Acta, 397, 145
(1999).

M. Awada, J. W. Strojek, and G. M. Swain, J. Electrochem. Soc., 142, L42
(1995).

N. Vinokur, B. Miller, Y. Avyigal, and R. Kalish, J. Electrochem. Soc., 146, 125
(1999).

A. Manivannan, M. S. Seehra, D. A. Tryk, and A. Fujishima, Anal. Lett., 35, 355
(2002).

0. Enea, B. Riedo, and G. Dietler, Nano Letters, 2, 241 (2002).

C. Prado, S. J. Wilkins, F. Marken, and R. G. Compton, Electroanalysis, 14, 262
(2002).

S. Nakabayashi, D. A. Tryk, A. Fujishima, and N. Ohta, Chem. Phys. Lett., 300,
409 (1999).

F. Bouarrrrane, A. Tadjeddine, R. Tenne, J. E. Butler, R. Kalish, and C. Levy-
Clement, J. Phys. Chem. B, 102, 134 (1998).

J. Zak and M. Kolodziej-Sadlok, Electrochim. Acta, 45, 2803 (2000).

B. Scharifker and G. Hills, Electrochim. Acta, 28, 879 (1983).

70

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

B. Schariﬂ<er and J. Mostany, J. Electroanal. Chem, 177, 13 (1984).

A. J. Bard and L. R. Faulkner, Electrochemical Methods, Fundamentals and
Applications, John Wiley & Sons, Inc, New York, 1980.

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

M. Y. Abyaneh and M. Fleischmann, Electrochim. Acta, 27, 1513 (1982).
T. I. Quickenden and Q. Xu, J. Electrochem. Soc., 143, 1248 (1996).
N. Batina, T. Will, and D. M. Kolb, Faraday Discuss, 94, 93 (1992).

R. J. Nichols, W. Beckmann, H. Meyer, N. Batina, and D. M. Kolb, J.
Electroanal. Chem, 330, 381 (1992).

Z. Chen, J. Li, and E. K. Wang, J. Electroanal. Chem, 373, 83 (1994).
N. Kimizuka and K. Itaya, Faraday Discuss, 94, 117 (1992).
J. Y. Josefowicz, L. Xie, and G. C. Fanington, J. Phys. Chem, 97, 11995 (1993).

L. S. Pan and D. R. Kania, Diamond: Electronic Properties and Applications,
Kluwer Academic Publishers, Norwell, MA, (1995).

R. Locher, J. Wagner, F. Fuchs, and R. Kalish, Diamond Relat. Mater., 4, 678
(1995).

R. L. McCreery, in Interfacial Electrochemistry: Theory, Experiment, and
Applications, A. Wieckowski, Editor, Marcel Dekker, New York, 631 (1999).

B. Kazee, D. E. Weisshaar, and T. Kuwana, Anal. Chem, 57, 2736 (1985).

E. Budevki, G. Staikov, and W. J. Lorenz, Electrochemcial Phase Formation and
Growth, VCH, New York, 1996.

K. kinoshita and P. Stonehart, in Modern Aspects of Electrochemistry, J. O. M.
Bockris and B. E. Conway, Editors, Plenum Press, New York, 183 (1977).

P. Allongue, S. Blonkowski, and E. Souteyrand, Electrochim. Acta, 37, 781
(1992).

N. Vinokur, B. Miller, Y. Avyi gal, and R. Kalish, Electrochem. Solid-State Lett.,
1, 265 (1998).

71

J

W. Zhu, in Diamond: Electronic Properties and Applications, D. R. Kania,
Kluwer Academic Publishers, Norwell, MA, 173 (1995).

H. B. Martin, A. Argoitia, J. C. Angus, and U. Landau, J. Electrochem. Soc., 146,
2959 (1999).

G. D. Adzic, A. R. Despic, and D. M. Drazic, J. Electroanal. Chem, 241, 353
(1988).

Q. Chen, J. Wang, G. M. Swain, I. Sakaguchi, M. Nishitani-Gamo, and T. Ando,
unpulished results.

Chapter 3

Fabrication and Characterization of
Dimensionally Stable Pt/Diamond Composite
Electrodes for Electrocatalysis — The
Magnetron Sputtering Approach

3.1 Introduction

The use of nanometer-sized noble metal particles, supported on high surface area
carbon materials, is common practice in heterogeneous electrocatalysis. Their catalytic
activity essentially depends on the physical and chemical nature of both the catalyst and
the support material.“ and ”memes mm“ While spz-bonded carbon supports suffer severe
oxidation/corrosion in aggressive chemical environments and at elevated temperature,
sp3-bonded conductive diamond thin films are extremely stable in such environments.
The materials possess several important properties: controllable electrical conductivity,
chemical inertness, morphological stability, and corrosion resistance,5'7 all of which make
the material an ideal candidate for a catalyst support/host.

The importance of the catalyst/electrode preparation method is well recognized.
The preparation method affects the crystallite size, composition, and surface
crystallographic orientation of the catalyst, factors that strongly inﬂuence the catalytic
activity.1 Common methods, for the preparation of high surface-area catalysts supported
on porous substrates (e.g., spz-bonded carbons), include impregnation, ion exchange,
colloidal adsorption, electrodeposition, and physical vapor deposition techniques.

Clearly, most of the methods might not be applicable for preparing supported catalyst

73

particles on diamond ﬁlm because of the geometric nature of the material and its inert
surface.

In the search for alternate strategies for the preparation of diamond-film-
supported electrocatalysts, magnetron sputtering was a method investigated for
depositing the catalytic metal particles. This method has been successfully used, as
reported in several studies,8‘9 to deposit nanoparticles of catalyst on a variety of
substrates. In addition, magnetron sputtering, coupled with annealing, has been employed
for depositing metal adlayers on diamond films for the fabrication of ohmic contacts'o‘ll

In this chapter, we report on our initial attempts to prepare Pt/diamond composite
electrodes by sequential diamond growth! Pt magnetron sputtering/ diamond growth
fabrication procedure. In this multistep process, a continuous boron-doped diamond thin

film was first deposited on a conductive substrate (e. g. Pt foil). The ﬁlm thickness was 2-

5 pm. Next, DC magnetron sputtering was used to deposit a discontinuous layer of Pt on

the surface. The Pt-coated diamond ﬁlm was then subjected to a short, secondary
diamond growth in order to anchor the metal particles into the surface microstructure.12
The resulting Pt/diamond composite ﬁlms were characterized by scanning electron
microscopy (SEM), energy dispersive x-ray analysis (EDX), atomic force microscopy
(AFM), auger electron spectroscopy (AES), Raman spectroscopy, powder x-ray
diffraction (XRD), secondary ion mass spectrometry (SIMS), and cyclic voltammetry.
The results demonstrated that 10-500 nm Pt particles are stably anchored into the surface
region of a boron-doped diamond thin film, and the particles are in good electronic
communication with the current collecting substrate. The catalyst particles were active

for the UPD of hydrogen, reduction of oxygen, and oxidation of methanol.

74

3.2 Experimental

 

 

. ’~ . -.E . n+1; l .
3v .,;.;..',\'i‘ 41.-v“ T Slngle crystal
Pt

Diamond ﬁlm
growth

 

Pt metal adlayer

Secondary
diamond film
growth

 

 

 

 

Figure 3.1. Fabrication process for the Pt/diamond composite electrodes.

Preparation of Pt/diamond composite electrodes. The multistep fabrication
procedure for the composite electrodes is shown diagrammatically in Figure 3.1. A
boron-doped diamond thin ﬁlm was deposited on Pt (100) substrate using microwave-
assisted chemical vapor deposition (CVD). The boron-doped ﬁlm was deposited for 12 h
using a CM; ratio of 0.33 %, a gas pressure of 60 torr, and a substrate temperature of

875 °C. The ﬁlm thickness at this stage was ca. 3 pm. The boron doping was

75

accomplished by controlling the B2H6 introduced with the source gas mixture (~O.l%

3. The diamond growth was then

B/C). The doping level was in the range of ~ 1019cm'
stopped, the substrate was removed from the chamber, and the diamond film was coated
with a discontinuous layer of Pt. The Pt layer was deposited by DC magnetron sputtering.
SEM images revealed the Pt metal incompletely covered the diamond surface. The Pt-
coated ﬁlms were then placed back into the CVD reactor and a second diamond
deposition was performed, as described above, for 3 h. The extra film thickness was less
than 1 pm. The diamond ﬁlm deposited on Pt(100) substrate without any sputtered Pt
catalyst was prepared using the same 2-step procedure.

Film characterization. SEM was performed with a Hitachi S-SOO field-emission
microscope. Images were recorded using both secondary and backscattered electrons at
an accelerating voltage of 20 kV. Some of the images were also obtained at the Materials
Characterization Center (University of Puerto Rico, San Juan, PR) using a JEOL 5800LV
microscope with an EDX, x-ray ﬂuorescence CDU leap detector. AFM (both in situ and
ex situ) was performed with a Nanoscope II instrument (Digital Instruments, Santa
Barbara, CA). The instrument was equipped with a 12 um scanning head and was
operated in the contact mode. A standard ﬂuid cell, constructed with Kel-F and equipped
with a gold-coated spring clip and a pyramidal Si3N4 45° tip mounted on a cantilever (100
um legs, 0.38 N/m spring constant), was used to acquire the images in situ. The tip was
held in an optically transparent holder as part of the fluid cell. The cell was sealed with an
acid-resistant ethylene-propylene-diethylene o-n'ng. The geometric area of the exposed

electrode was 0.38 cmz. Both the auxiliary and quasi-reference electrodes were Pt wires.

76

Raman spectroscopy was performed in a 1800 backscattered collection geometry
using 1.25 m spectrograph (Instruments SA) consisting of an air-cooled 100 mW Ar+
laser, a 1.25m spectrometer (1800 grooves/mm holographic grating), a 1482E MicraMate
confocal microscope assembly, and a Spectrum One 2000 x 800 CCD detector. Spectra
were collected with an incident power density of ca. 150 kW/cm2 (0.010 W and 5 urn
spot size). The spectrometer was calibrated with either a piece of single crystal Si (lattice
phonon mode = 519.0 cm") or a Type [1a diamond standard (lattice phonon mode =
1332.5 cm").

XRD was performed using a Rigaku RINT1000 diffractometer in a 2-theta/theta
geometry. The x-ray source was Cu (1.5407 A). The x rays were generated at 200 mA,
with a accelerating voltage of 50 kV. The scan rate was 4 degrees/min.

Positive ion SIMS data were acquired with a Camera IMSSF instrument using 02+
at 8kV and 100 nA (Surface Analysis Laboratory, University of Utah). The dynamic
SIMS proﬁles were collected over an 85 um2 area with an apparent sputter rate of 0.82
nm/sec. The total sampling depth was estimated to be approximately 1.8 pm.

The cyclic voltammetry was performed with an CYSY-1090 digital potentiostat
(Cypress Systems Inc., Lawence, KS) using a single compartment glass cell (see Chapter
2) . The exposed geometric area of the working electrode was 0.2 cm2. A platinum wire
served as the auxiliary electrode and a Ag/AgCl electrode (in sat. KCl) was used as the
reference. The H2S04 and H00: acid electrolytes were ultrahigh purity grade (99.999%
Aldrich) and the methanol was reagent grade quality (Fisher Scientiﬁc). All glassware

was cleaned in a KOH/methanol bath prior to use.

77

3.3 Results and Discussion

3.3.1 Structural Characterization

 

Figure 3.2. SEM images of composite diamond electrodes fabricated (A) with and (B)

without Pt.

Figure 3.2A and B show SEM images of boron-doped diamond thin ﬁlms with
and without deposited Pt, respectively. Both images reveal a well-facted, polycrystalline
morphology with octahedral, cubic, and cubo-octahedral microcrystallites. The larger
crystallites are approximately 2 pm in diameter at the base, and roughly 1 pm in height.
There are numerous grain boundaries, extended defects and re—entrant grooves on the
crystallites. The ﬁlm deposited with Pt clearly shows the presence of pseudo-spherical
metal particles dispersed over the surface. The larger particles have diameters between

100 and 500 nm. Unfortunately, these deposits are too large for a practical catalytic

78

electrode. The metal deposits exist at both defect sites on the grains, and in the grain
boundaries. As will be shown below, the deposits are strongly anchored to the surface
because of the secondary diamond ﬁlm growth around the base.

Figure 3.3 shows a higher magnification top-view, 5 x 5 pm2 AFM image (force
mode, air) of Pt/diamond composite electrode which reveals the dispersed metal deposits.
The larger deposits are located almost exclusively at grain boundaries and other extended
defects. There are also many smaller deposits dispersed on the facets. These smaller

deposits have diameters ranging from 10 to 50 nm.

   

Figure 3.3. AFM force-mode image (contact mode, air) of a Pt/diamond composite

electrode. The z-axis is 40 nN full-scale.

Energy dispersive x-ray analysis (EDX) data, combined with SEM, and AES

results conﬁrmed that the spherical particles were Pt. In the EDX microanalysis

79

spectrum, the C Kor (0.277 KeV), O Kor (0.525 KeV), and Pt Mor (2.05 KeV) and Lor
(9.44 KeV) emissions were observed. The atomic percentages of each element present
were ca. 95, 2 and 3, respectively. These values are semi-quantitative because there is
typically a l to 5% error in the determination of the lighter elements and a ca. 0.5% error
in the determination of the heavier elements by this method. The estimated Pt/C and O/C
atomic ratios were 0.033 and 0.024, respectively. For comparison, AES analysis of a
larger region on the surface indicated a PVC ratio of 0.071. The approximately equal
amounts of Pt and O is not surprising because this electrode was used extensively in
electrochemical measurements prior to making the EDX measurements. Consequently,
the surface of the Pt particles and the local diamond support would be expected to
possess some Chemisorbed oxygen.

Raman spectra for composite electrodes, fabricated with and without the
incorporated Pt, are shown in Figure 3.4. The spectrum for the metal-containing electrode
shows an intense peak at 1336.8 cm", which is attributed to the ﬁrst-order diamond
phonon mode. The peak position is shifted some 5 cm'1 toward higher energy due to the
compressive stress within the ﬁlm because of (i) platinum’s higher thermal expansion
coefﬁcient and (ii) the incorporation of the Pt particles.“14 The ﬁrst-order diamond line
intensity and position are sensitive to factors such as the crystallite size, optical density,
ﬁlm quality, and net stress.15 The intrinsic stress (compressive or tensile) in a ﬁlm can be
determined by the shift of the diamond line from the 1332.5 cm'1 position expected for
stress-free, phase-pure diamond.”16 The linewidth (FWHM) of the peak is 9.2 cm“,
which is larger than the 2-3 cm'1 linewidth observed for single crystal diamond. The

linewidth is a measure of the crystalline perfection, being inversely related to the phonon

80

 

2% V V r V v I v I v V ‘r v v ' V V ' v V

 

without
15000»
3 .
§ 10000 . .
¢
0
5000 '

a
................

 

 

 

 

 

Raman Shift (cm'l)

Figure 3.4. Raman spectra for diamond composite electrodes fabricated with and without
Pt. Excitation = 514.5 nm. Integration time = 10 s. The laser power density at the

sample was estimated to be ca. 150 kW/cmz.

lifetime. There is a weak scattering intensity at 1589.9 cm", which is attributable to non-
diamond spz-bonded carbon impurity domains. The peak position is such that these non-
diamond impurities appear graphitic in nature, and probably exist at the interface between
the Pt substrate and the diamond. TEM analysis revealed a 100-300 nm region at the
diamond/substrate interface that consists of graphitic carbon.14 The ratio of the 1589 to
1337 cm'1 peak areas is 0.006. It should be noted that the scattering cross section
coefficient for graphite (a model for the spz-bonded carbon impurity) is 50 times larger

than that for diamond using visible excitation, thus the spectra reveal the ﬁlm is largely

81

diamond in character.17 The spectrum for the composite electrode without Pt shows an
intense diamond line at 1335.7 cm" with a linewidth of 7.3 cm". The diamond line is
much more intense for this film because the optical density is much lower (i.e., sampling
volume of the excitation light is larger) without the Pt particles. A peak at 1587.3 cm",
associated with scattering from graphitic carbon impurities, is also observed. The ratio of
the 1587 to 1336 cm'1 peak areas is 0.002. The larger linewidth for the composite
electrode with Pt reﬂects a higher degree of crystalline imperfection due to the
incorporated metal deposits.

The diamond films are clearly well faceted with different exposed primary
crystallographic orientations. Powder XRD was used to determine the crystallographic
orientations of the diamond as well as Pt particles. The major diamond reﬂections were
observed at 43.9 degrees (1 l l-orientation), 75.5 degrees (220—orientation), and 91.7 (311-
orientation). Reﬂections associated with the Pt (either the Pt particles or the Pt substrate)
were observed at 39.5 degrees (Ill-orientation), 46.3 degrees (200-orientation), 67.5
degrees (220-orientation), 81.5 degrees (31 l-orientation), 86.0 degrees (222-orientation),
108.8 degrees (400-orientation), and 123.0 degrees (420-orientation).1 There was also
some weak diffraction intensity detected at 26.5 degrees associated with the graphitic
carbon impurities, probably located either at the diamond/Pt substrate interface. This
peak intensity was about 5% as intense as the diamond (111) line at 43.9 degrees. These
data confirm the crystalline nature of the diamond, the Pt nanoparticles, and the Pt
substrate.

Figure 3.5A and B show dynamic SIMS profiles (positive ion mode) for several

of the major ions sputtered from composite electrodes deposited with and without Pt,

82

respectively. The major ion profiles are shown for C (m/z 12), B (m/z 11), O (m/z 16), N
(m/z 14), Si (m/z 29), and Pt (m/z 195). The proﬁle for the composite electrode
containing Pt shows significantly increased levels of B, O, Si, and particularly Pt, in the
near-surface region. The Pt relative ion intensity is about 5 orders of magnitude larger at
the surface than in the diamond film bulk. The Pt does appear to be distributed in the
near—surface region of the film because 1000 seconds of sputtering time corresponds to
about 820 nm, and the microcrystallite height is approximately 1 pm. It should be
mentioned that Pt particles trapped in the diamond matrix, away from the Pt substrate,
have been observed by TEM.14 Most interesting is the significant increase in the boron
signal near the surface, concomitant with the increase in Pt. The surface B signal is some
2 to 3 orders of magnitude larger at the surface than in the bulk. This is a very interesting
observation and may result from the formation of a “Pt-boride” complex at some of the
Pt sites. The SIMS proﬁle for the composite electrodes deposited without Pt reveals the
same major ion profiles, but the signal for Pt is near the baseline level at all depths. Very
interestingly, the B signal decreases in the near surface region, from that of the bulk, and
then increases at the surface. The reason for this ﬂuctuation in the B concentration is
unknown at present. The other elements have constant ion intensity proﬁles from the
surface into the bulk. The N concentration at the surface slightly decrease from that

observed in the bulk.

83

 

>

 

12(C)

x; 11 (B)
W 16(0)

 

* H 14 (N)
. "ﬁ 29 (SI)

 

n n.n‘.95.(".9. . . -

 

5m) IHWO 15M) 20M) 25M) :umo

 

V rv'vvvv' Y Y

 

12(C)

Relative secondary ion intensity
c

V 7"

 

 

 

 

16(0)
11 (a)

14 (N)
29 (SI)

 

 

A A A I A A A A l J A A A I A A A A 1 A A {g'n ‘A A A A

0 5M) ‘NWO ISMD 20M) 25M) le0

Time (sec)
Figure 3.5. Dynamic SIMS profiles (positive ion mode) for diamond composite

electrodes prepared (A) with and (B) without Pt.

84

3.3.2 Electrochemical Characterization

The as-deposited Pt/diamond composite electrode was initially conditioned in 0.1
M HClO4 by potential cycling between —300 mV and 1500 mV vs. Ag/AgCl. Figure 3.6A
shows a background cyclic voltammetric i-E curve in 0.1 M HClO4 for a Pt/diamond
composite electrode obtained after 50 cycles. Figure 3.6B shows a similar curve in 1 M
H2804. The potential sweep rate for both was 50 mV/s and the sweeps were initiated at 0
mV and scanned in the positive direction. The voltammetric features are all characteristic
of polycrystalline Pt with the Pt oxide formation and oxygen evolution at potentials
positive of 900 mV, the reduction of Pt oxide at ca. 400 mV, the adsorption and
desorption of hydrogen between 0 and -250 mV, and the double layer region between ca.
100 and 250 mV. Integration of the oxidation charge associated with the desorption of

hydrogen between 100 and -250 mV yields a value of 120 uC/cmz. This charge can be

used to calculate the active Pt metal loading of 242 ng/cm2 (diamond electrode geometric

'8‘” These features demonstrate the

area), assuming a l x l Hth surface structure.
composite electrode contains active Pt particles at the surface that are in good electronic
communication with the current collecting substrate, through the boron-doped film.
Figure 3.7A and B show background cyclic voltammetric i-E curves in 0.1 M
HClO4 (ﬁrst scan) and l M H2804 for a composite electrode deposited without
incorporated Pt, respectively. These voltammograms show no evidence for Pt. These

results indicate that the diamond ﬁlm completely covers the Pt substrate, and that the

substrate does not directly inﬂuence the electrochemical response of the electrodes.

85

 

 

 

 

 

 

 

Current (uA)

 

 

 

 

 

 

 

L L A A A

-400 0 400 800 1200 1600

 

Potential (mV vs. Ag/AgCl)

Figure 3.6. Cyclic voltammetric i-E curves for a Pt/diamond composite electrode in (A)

0.1 M HClO4 and (B) l M H2S04. Scan rate = 50 mV/s.

86

100 ﬁ-f

 

75'
50-

25-

 

 

 

700 1100 1500 1900

 

Current (uA)
's
O
§
0)
8

75. B

50'

25’
> —>

 

 

 

A A 4 A A A A A 4

A

0 400 800 1200 1600 2000

 

 

Potential (mV vs. Ag/AgCl)

Figure 3.7. Cyclic voltammetric i-E curves in (A) 0.1 M H00; and (B) l M H2S04 for a

composite electrode prepared without Pt. Scan rate = 50 mV/s.

87

There are no pinholes or voids in the diamond film. In fact, the curves are rather
featureless between the solvent electrolysis potentials; that is the oxidation of water to
oxygen at 1700 mV and the reduction of water to hydrogen at —700 mV, yielding a
working potential window of 2.4 V. This window is smaller than the 3 to 4 V typically

d,7‘20 and probably results from the

observed for high quality, polycrystalline diamon
defects and the graphitic impurities present in the ﬁlm, formed as a result of the growth
on the Pt substrate (see evidence for this in the Raman data).'4 There is some anodic
charge passed at ca. 1400 mV, just prior to the onset of oxygen evolution. This is likely
due to the oxidation of the surface at exposed non-diamond carbon impurity sites.

It is our supposition that polycrystalline diamond could function well as a catalyst
support. The fact that the Pt particles are anchored into the diamond matrix could leads to
a high degree of electrode stability and durability.”22 Some of the problems, such as
catalyst mobility, aggregation and detachment, that can occur with spZ-carbon-supported
catalysts, particular when operated at elevated temperature in aggressive chemical
environments, may be alleviated by incorporating the catalyst particles into the diamond
surface microstructure. The stability of the a Pt/diamond composite electrode was
investigated at room temperature in 0.1 M HClO4 during long-term potential cycling
between —400 mV and 1500 mV vs. Ag/AgCl. One thousand cycles were performed at a
potential scan rate of 50 mV/s. Figure 3.8 shows cyclic voltammetric i-E curves before
and after the cycling. The maximum current density achieved at the anodic and cathodic

potential limits was ca. 1 mA/cmz. The open circuit potential before cycling was 454 mV

vs. Ag/AgCl and increased only slightly to 467 mV after cycling. This minimal change in

88

 

200

    

150 '

  

_— Before cycling

100 ' ............... After cycling

  

50-

 

-50 -

Current (uA)
G

-100 -

 

-150 -

 

 

 

l

 

499100 l ‘ 0‘ ‘ ‘400‘ ‘ ‘800‘ ‘ ‘1200‘ i .1600

Potential (mV vs. Ag/AgCl)

Figure 3.8. Cyclic voltammetric i-E curves in 0.1 M HClO4 for a Pt/diamond composite
electrode before and after 1000 potential cycles between —400 and 1500 mV vs.
Ag/AgCl. Scan rate = 50 mV/s. The maximum anodic and or cathodic current density

during the cycling was ~l mA/cmz.

potential indicates the absence of significant alterations in the physical and chemical
properties of the surface. The charge associated with the oxidative desorption of adsorbed
hydrogen was measured before and after cycling. This charge is a good measure of the
active Pt exposedzo'z' The charge before cycling was 175 uC/cm2 (353 ng/cmz) and after
cycling increased to 210 itC/cm2 (424 ng/cmz). The loading values reﬂect the mass of the
electrochemically active Pt normalized to the diamond film geometric area. The increase

indicates that metal dispersions were not lost during the cycling, but rather the active

89

catalyst area increased. This increase is attributable to (i) the surface roughening that
occurs during the potential cycling through the Pt-oxide/oxygen and hydrogen evolution
regimes, and (ii) the cleaning effect brought about by the oxygen and hydrogen gas
evolution (i.e., removal of hydrocarbon contaminants or Pt-boride complexes). The
hydrogen desorption charge after the extended cycling is much larger than the 120

p.C/cm2 reported in Figure 3.6. This is because the electrode was used in numerous

electrochemical measurements between the time the data in Figures 3.6 and 3.8 were
obtained, such that significant cleaning occurred.

Further evidence for the electrode stability and durability comes from
electrochemical AFM investigations of the composite electrode during long-term
potential cycling between the same potential limits as measured in the previous figure.
Figure 3.9A shows a 5 x 5 1.1m2 image of the electrode surface in 0.1 M HClO4 prior to
any cycling. The well faceted diamond morphology can be seen along with the Pt metal
dispersions. The metal particles are distributed in the grain boundaries and at defects on
the grains. The deposits range in diameter from about 100 to 500 nm. Figure 3.9B shows
a 5 x 5 urn2 image of the same region of the electrode after one thousand potential cycles.
No alterations in the diamond morphology or in the size, distribution and location of the
metal deposits are seen. These observations support the notion that the meta] deposits are
strongly anchored into the diamond surface rnicrostructure by virtue of the secondary
diamond growth, and that the electrode structure is very stable.

The electrocatalytic activity of the Pt/diamond composite electrode for the

electrooxidation of methanol was examined. Figure 3.10 shows cyclic voltammetric i-E

90

 

Figure 3.9. Electrochemical AFM images for a Pt/diamond composite electrode in 0.1 M
HC104 (A) before and (B) after 1000 potential cycles between —400 and 1500 mV vs.
Ag/AgCl. Scan rate = 50 mV/s. The z-axis is 4 nN full-scale. The images were acquired

at 350 mV vs. Ag/AgCl.

curves for electrodes, with and without incorporated Pt, in 0.6 M CH3OH + 0.1 M
HClO4. Methanol is considered to be a useful fuel for fuel cells, and Pt is known to
catalyze the oxidation of methanol.”25 The oxidation reaction is complex, the oxidation
current depends on the crystallographic orientation of the Pt electrode, and several
different products can form (e.g., carbon monoxide, carbon dioxide, formaldehyde,
etc.).23'25 No attempt was made to identify the product(s) in the present work. Comparing
the two voltammograms, one can see that the Pt-containing electrode is electroactive for
methanol oxidation while the bare diamond electrode is not. The methanol oxidation
current at the Pt/diamond composite electrode commences at 400 mV on the forward

sweep and reaches a maximum current density of ca. 0.9 mA- cm'2 (geometric area of

91

diamond ﬁlm) at 650 mV. The current then decreases on the forward sweep at potentials
positive of 650 mV as some of the surface sites, previously available for coordinating the
methanol reactant, become blocked by the forming Pt-oxides. Upon scan reversal, a
second oxidation maximum is observed at ca. 500 mV as the Pt-oxides are reductively
removed making sites available again for coordinating methanol. A more detailed
discussion of the methanol electrooxidation at Pt/diamond composite electrodes is
provided in Chapter 6. Clearly, the bare diamond surface shows little activity for

methanol oxidation.

 

  
 
    

 

 

 

 

 

200 - - - . . - - ,
* _with
150» Pt
I ........... Without
A . Pt
3 100
H ’ 4
ﬂ .
2
r... 50' .
5 .
U |
0- .
'5““““ee44---..-,

Potential (mV vs. Ag/AgCl)
Figure 3.10. Cyclic voltammetric i—E curves for boron-doped diamond composite
electrodes with and without deposited with Pt in 0.6 M CH3OH + 0.1 M HClO4. Scan rate

= 50 mV/s.
The electrocatalytic activity of the Pt/diamond composite electrode for oxygen

reduction was also examined. The oxygen reduction reaction is the anodic reaction in fuel

92

cells and platinum is the best-known electrocatalyst?“28

The potential sweep was
initiated at 600 mV and scanned in the positive direction at a scan rate of 50 mV/s.
During the cathodic scan, the reduction current commences at ca. 600 mV and reaches a
maximum of 215 “A at ca. 280 mV. It should be noted that the cathodic Current is a sum
of the oxygen reduction current, the current due to Pt oxide reduction, and the double-
layer charging current. Clearly, the oxygen reduction current at 250 mV is due solely to
the activity of the Pt catalyst. This is evident by comparing the cyclic voltammetric i-E
curve for the diamond composite electrode without incorporated Pt. This response of the
bare electrode is featureless from -400 mV to 1500 mV with no reduction current at 250
mV. Diamond is known to have a large overpotential for oxygen reduction in both

acidic29 and basic media.30 Detailed kinetic data for oxygen reduction at Pt/diamond

composite electrodes are reported in Chapter 5.

 

 

 

 

 

 

100 - - - . + - - ﬂ - - - . - - - . - - -

50 - ..

0 . ‘--— _ -=:—> —- U .......... -..-’8‘:

‘—
A .
i -50 ~ .
V I
*’ .100 - , .
5 , wrth
l-i _—
‘5 .150 - / Pt .
U ' \ ........... without '
-250 k ‘ # ‘ ‘ ‘ ‘ g k A - l - A A A - - -
-500 -100 300 700 1100 1500

Potential (mV vs. Ag/AgCI)

Figure 3.11. Cyclic voltammetric i-E curves for diamond composite electrodes, with and

without Pt, in Oz-saturated 0.1 M HClO4. Scan rate = 50 mV/s.

93

3.4 Conclusions

A new electrocatalytic composite electrode was prepared and characterized — a
boron-doped diamond thin ﬁlm containing incorporated Pt nanoparticles. The composite
electrodes were fabricated by a multistep procedure (1. diamond growth, 2. metal
deposition, 3. secondary diamond growth), in which the Pt catalyst was deposited by DC
magnetron sputtering. The composite electrodes were characterized by scanning electron
microscopy, energy dispersive x-ray analysis, atomic force microscopy, auger electron
spectroscopy, Raman spectroscopy, powder x-ray diffraction, secondary ion mass
spectrometry, and cyclic voltammetry. The dimensionally stable, corrosion-resistant, and
electrically conductive composite electrode consists of well-faceted diamond
microcrystallites with dispersed Pt particles incorporated into the surface microstructure.
These particles range widely in diameter from 10 to 500 nm, and practically speaking, are
too large for real world application. Nevertheless, the Pt nanoparticles are strongly
anchored into the surface region because of their location at the intercrystalline grain
boundaries, and because the diamond ﬁlm appears to grow around their base. Most
interestingly, the Pt nanoparticles at the ﬁlm surface are in electronic communication
with the current collecting Si or Pt substrate through the boron-doped diamond matrix,
and they are electroactive for the UPD of hydrogen, the reduction of oxygen, and the
oxidation of methanol. The dispersed Pt particles in the diamond rnicrostructure are
extremely stable as no decrease in the activity was observed after 1000 potential cycles

between the oxygen and hydrogen evolution regimes (~l-6 mA~cm'2) in 0.1 M HClO4.

94

References:

10.

ll.

12.

13.

14.

15.

l6.

17.

K. Kinoshita and P. Stonehart, in Modern Aspects of Electrochemistry, J. O. M.
Brokris, B. E. Conway and R. E. White, Editors, Plenum, New York, 12, 183
(1977).

K. Kinoshita, in Modern Aspects of Electrochemistry, J. O. M. Brokris, B. E.
Conway and R. E. White, Editors, Plenum, New York, 14, 53 (1982).

T. D. Jarvi and E. M. Stuve, in Electrocatalysis, J. Lipkowski and P. N. Ross,
Editors, Wiley-VCH, New York, 75 (1998).

C. Lamy, J. M. Leger, and S. Srinivasan, in Modern Aspects of Electrochemistry,
J. O. M. Brokris, B. E. Conway and R. E. White, Editors, Plenum, New York, 34,
53 (2001).

J. Xu, M. C. Granger, Q. Chen, J. W. Strojek, T. E. Lister, and G. M. Swain, Anal.
Chem, 69, 591A (1997).

R. Tenne and C. Levy-Clement, Israel Journal of Chemistry, 38, 57 (1998)
G. M. Swain, A. B. Anderson, and J. C. Angus, MRS Bulletin, 23, 56 (1998).
A. B. Anderson, Electrochim. Acta, 44, 1441 (1998).

A. Berthet, A. L. Thomann, F. J. Cadete Santos Aires, M. Brun, C. Deranlot, J. C.
Bertolini, J. P. Rozenbaum, P. Brault, and P. Andreazza, J. Catal., 190, 49 (2000).

P. E. Viljoen, E. S. Lambers, and P. H. Holloway, J. Vac. Sci. Technol., B12,
1997 (1994).

L. L. Moazed, J. R. Zeidler, and M. J. Taylor, J. Appl. Phys., 68, 2246 (1990).
C. Wang and T. Ito, Appl. Phys. Lett., 75, 1920 (1999).
L. Bergman and R. J. Nemanich, J. Appl. Phys., 78, 6709 (1995).

T. Tachibana, Y. Yokota, K. Miyata, T. Onishi, K. Kobashi, M. Tartunati, Y.
Takai, R. Shimizu, and Y. Shitani, Phys. Rev. B, 56, 15967 (1997).

W. Yarbrough and R. J. Messier, Science, 247, 688 (1990).
J. R. Dennison, M. Holtz, and G. M. Swain, Spectroscopy, 11, 38 (1996).

D. S. Knight and W. B. White, J. Mater. Res., 4, 385 (1989).

95

 

10.

l!

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

J. Clavilier, J. P. Ganon, and M. Petit, J. Electroanal. Chem, 265, 231 (1985).
W. Chrzanowski, H. Kim, and A. Wieckowski, Catalysis Lett., 50, 69 (1998).
J. Xu, Q. Chen, and G. M. Swain, Anal. Chem, 70, 3146 (1998).

N. L. Pocard, D. C. Alsmeyer, R. L. McCreery, T. X. Neenan, and M. R.
Callstrom, J. Am. Chem. Soc., 114, 769 (1992).

M. R. Callstrom, T. X. Neenan, R. L. McCreery, and D. C. Alsmeyer, J. Am.
Chem. Soc., 112, 4955 (1990).

W. Chrzanowaki and A. Wieckowski, Langmuir, 14, 1967 (1998).

W. Chrzanowaki and A. Wieckowski, in Interfacial Electrochemistry: Theory,
Experiment, and Applications, A. Wieckowski, Editor, Marcel Dekker, New
York, 937 (1999).

W. Chrzanowski, H. Kim, G. Tremiliosi-Filho, A. Wieckowski, B. Grzybowska,
and P. Kulesza, J. New Mater. Electrochem. Sys., 1, 31 (1998).

N. M. Markovic, H. A. Gasteiger, and P. N. Ross, Jr., J. Phys. Chem, 99, 3411
(1995).

N. Markovic, H. Gasteiger, and P. N. Ross, J. Electrochem. Soc., 144, 1591
(1997).

J. Perez, E. R. Gonzalez, and E. A. Ticianelli, Electrochim. Acta, 44, 1329 (1998).

T. Yano, E. Popa, D. A. Tryk, K. Hashimoto, and A. Fujishima, J. Electrochem.
Soc., 146, 1081 (1999).

T. Yano, D. A. Tryk, K. Hashimoto, and A. Fujishima, J. Electrochem. Soc., 145,
1870 (1998).

96

4.1

lab

C0

an
1116

C01

Chapter 4

Fabrication and Characterization of
Dimensionally Stable Pt/Diamond Composite
Electrodes for Electrocatalysis — The
Electrodeposition Approach

4.1 Introduction

A new type of carbon electrocatalyst support/host is under development in our
laboratory; electrically conducting diamond thin ﬁlms."2 The use of electrically
conducting microcrystalline and nanocrystalline diamond electrodes in electrochemistry
is a relatively new field of research.3’8 The properties of this new electrode material make
it ideally suited for electrochemical applications, particularly demanding ones (i.e.,
complex matrix, high current density, and potential, high temperature, extremes in pH,

etc.).

The preparation technique is crucial in order to improve both the catalytic activity
and reduce the cost of the diamond film supported noble metal catalyst. In the preceding
chapter, we demonstrated that nanometer-sized Pt particles could be incorporated into the
surface rnicrostructure of boron-doped diamond thin ﬁlms via a multistep fabrication
procedure. The dispersed Pt particles were achieved by magnetron sputtering deposition
and further stabilized by the growth of a thin film of diamond around their base. While
the catalyst particles are stably anchored, the magnetron sputtering approach affords poor

control over the particle size and distribution.

97

cl

The preparation of supported catalyst nanoparticles using electrochemical
methods has recently attracted much attention.“14 Compared with methods involving
evaporation of metal in vacuum, electrodeposition is a relatively simple, cost-efficient,
and versatile technique that allows the preparation of large bulk samples with high purity,
low porosity, and enhanced thermal stability.12 Furthermore, the control of the
nanomaterial properties, e.g., grain size, distribution, and crystallite shape, is possible
through manipulation of the deposition conditions. There is another nice feature of the
electrochemical procedure with respect to electrocatalysis. Metal particles are exclusively
deposited on the active sites of the support material where both facile electron transfer
and electrolyte accessibility are possible. Hence, the incorporation of unusable noble
metal catalyst, a problem often encountered during the preparation of current state-of—the-
art electrodes, is avoided, and the overall catalyst cost is substantially reduced.

The control over the particle size and distribution via an electrodeposition

 

approach is apparent in the electrochemical version of the Gibbs-Thomson equation: ”"5
2 V
r, = 0 "' (4.1)
zelnl

in which r, is the size of critical nucleus (grains with r > n. are stable), 0 the specific
surface energy, Vm the atomic volume in the crystal lattice, z the number of electrons, and
n the overpotential. One can see that high overpotential is necessary to maintain a high
nucleation rate in order to achieve small particles. However, the nucleation rate is
strongly limited by the rapid buildup of a depletion layer in the vicinity of the electrode at
high overpotential. Therefore, either forced convection in constant current/potential mode
or pulsed current/potential should be applied in order to minimize the concentration

polarization.

98

using
cont
lopi
fror
con
git

\N'l

In this chapter, we report on the fabrication of Pt/diamond composite electrodes
using an electrodeposition approach. The same multistep procedure for fabricating the
composite electrode was employed. The dispersion of Pt nanoparticles onto the boron-
doped diamond thin ﬁlm surface was accomplished by galvanostatic electrodeposition
from a dilute PtCl62‘ solution. The size and distribution of the metal particles can be
controlled, to some extent, by adjusting the electrodeposition and secondary diamond
growth conditions. Potentiodynamic deposition of the metal was also investigated and
will be brieﬂy discussed herein, but was found to be inferior to galvanostatic deposition.
The morphology, microstructure, and catalytic activity of the composite electrodes were
examined by scanning electron microscopy (SEM), atomic force microscopy (AFM),
powder x-ray diffraction (XRD), Raman spectroscopy, and cyclic voltammetry (CV),
respectively. The resulting Pt/diamond composite electrodes have metal catalyst particles
that are strongly anchored into the surface rnicrostructure, and are extremely stable with
unchanging catalytic activity even after 2 h of anodic polarization in 85% H3PO4 at

170°C and 0.1 A-cm'z. An acid etching experiment conﬁrmed that the secondary diamond

deposition results in film growth around the base of the metal deposits, stably anchoring

them into the surface microstructure.

4.2 Experimental

Preparation of Pt/diamond composite electrodes. Diamond thin ﬁlms were
deposited by microwave-assisted chemical vapor deposition (1.5 kW, 2.54 GHz, ASTeX,
Inc., Lowell, MA) on highly conducting p-Si (100) substrates (~10'3 Q-cm). The
substrates (0.1 cm thick x 1 cm2 in area, Virginia Semiconductor, Inc.) were pretreated by

rinsing in toluene, methylene chloride, acetone, isopropanol and methanol. They were

99

then etched in concentrated HF for 60 s, rinsed with ultrapure water and air dried. The
cleaned substrates were then sonicated in a diamond powder/acetone slurry (0.1 pm

diam., GE Superabrasives, Worthington, OH) for 20 min. followed by a rinse with clean
acetone. The sonication scratches the surface and leaves “seed” crystals (>108
particles/cmz), both of which serve as nucleation sites during film growth. The seeded
substrates were then placed in the CVD reactor and the system pressure reduced to about
10 mTorr. Ultrahigh-purity (99.999%) methane and hydrogen were used as the source
gases. The hydrogen ﬂow rate was 200 sccm and the methane ﬂow rate was 0.70 sccm
during the deposition. Prior to introducing methane, the substrates were heated to the
growth temperature in a hydrogen plasma for 10 min. The plasma was formed with a
system pressure of 40 Torr and a power of 1000 W. The substrate temperature was
approximately 850 °C, as measured using a disappearing ﬁlament optical pyrometer. The
films were doped using solid-state boron dopant sources: a boron diffusion source (GS-
126, BoronPlusTM, Techniglas, Inc., Perrysburg, OH) and a small piece of boron nitride
(Goodfellow, Ltd., England). Boron dopant concentrations were estimated to be
approximately 1019 to 1020 B/cm3 (300~3000 ppm B/C), based on boron nuclear reaction
analysis of other films deposited using similar conditions. The apparent in-plane ﬁlm
resistivities ranged from 0.1 to 0.01 Q-cm, and were measured using a tungsten four-
point probe (0.1 cm probe spacing) with the diamond film attached to the conducting Si
substrate.

The fabrication of the Pt/diamond composite electrode follows the same stepwise
procedure as described in the previous chapter, except that the substrate was the Si (100)

and the Pt was deposited electrochemically. A boron-doped diamond thin ﬁlm was

100

initially deposited onto a Si(100) substrate for 12 h. The thickness of the film was about

4-6 um. The diamond growth was then stopped and the substrate cooled to less than 300

0C in the presence of atomic hydrogen. After completely cooling to room temperature,
the coated substrate was removed from the reactor. Metal particles were then
electrodeposited onto the surface, galvanostatically, from 1 mM KthCl6 + 0.1 M HClO4
at 0.5 mA/cmz. Both diffusion and convection inﬂuenced the deposition as nitrogen was
slowly bubbled through the solution during the process. This current, and all others in the
text unless otherwise stated, are normalized to the geometric area of the electrode (0.2
cmz). Four diamond ﬁlms from the same batch were employed using Pt deposition times
of 100, 200, 300, and 400 8 each. Some experiments involved potentiodynamically
depositing the metal particles from the same solution while scanning the potential
between 0.8 and —0.3 V vs. Ag/AgCl at 50 mV/s. The Pt-coated ﬁlm was then placed
back in the reactor and boron-doped diamond deposited for additional 1-3 h, using the
same conditions as described above. During this step, the diamond ﬁlm grows around the
base of the particles, anchoring and stabilizing them within the surface microstructure.
The secondary diamond film thickness is less than 1 urn.

Film characterization. The atomic force microscopy (AFM) was performed with
a Nanoscope II instrument (Digital Instruments, Santa Barbara, CA) in the contact mode.
Pyramidal Si3N4 tips, mounted on gold cantilevers (200 um legs, 0.38 N/m spring
constant), were used to acquire topographical images in air. Raman spectra were obtained
at room temperature with a Chromex RAMAN 2000 system (Chromex Inc.,
Albuquerque, NM) consisting of a diode-pumped, frequency-doubled CW Nd:YAG laser

(500 mW at 532 nm, COHERENT), a Chromex 500is spectrometer (f/4, 600 grooves/mm

101

holographic grating), and a thermoelectrically-cooled 1024 x 256 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 diam. spot size).

Electrochemical measurements. The Pt electrodeposition and cyclic
voltammetry were performed in a single-compartment glass cell using a CYSY-2000
computer-controlled potentiostat (Cypress Systems, Inc., Lawrence, KS). A Ag/AgCl (in
sat. KCl) electrode was used as the reference and a large-area carbon rod served as the
counter electrode. The Pt/diamond composite working electrode was pressed against the
bottom of the glass cell using an Al plate current collector with the ﬂuid being contained
by a Viton® o-ring. Silver paste was applied to the center portion of the backside of the Si
substrate, after cleaning for electrical connection. The electrode’s geometric area was 0.2
cm2, and all currents are normalized to this area. The diamond ﬁlm working electrode
was cleaned, once mounted in the cell, by a thorough rinse with ultrapure water (>17
MSZ-cm, Bamstead E-Pure), a 20-min. soak in distilled isopropanol, followed by another
thorough rinse with ultrapure water. The electrolyte was deoxygenated with nitrogen
(99.9%, BOC Gases) for 20 min. prior to any electrochemical measurements, and the
solution was blanketed with the gas during the measurements.

Chemicals. All chemicals were reagent-grade quality or better, and used without
additional purification. The HClO4 (Aldrich) was ultrahigh purity (99.999%). The HCl
(Aldrich), HNO3 (Aldrich), H3PO4 (Aldrich), and KthC16 (Aldrich) were all of reagent
grade quality. All solutions were prepared with ultrapure water from a Bamstead E-Pure

puriﬁcation system (>17 MQ-cm).

102

4.3 Results and Discussion

4.3.1 Electrodeposition of Pt on Diamond Thin Films

Both potentiodynamic and galvanostatic deposition of Pt were investigated in
order to examine the inﬂuence of the deposition method and conditions upon the
morphology, size, and distribution of the metal deposits.

Potentiodynamic deposition. Figure 4.1A shows cyclic voltammetric i-E curves
obtained during the metal deposition in 1mM KthC16 + 0.1M HClO4. A total twenty-ﬁve
potential sweeps were made and the i-E curves for the 15‘, 5‘", 10'”, 15th, 20th and 25th
scans are presented. The potential sweep was initiated at 800 mV and scanned in the
negative direction to -300 mV at a scan rate of 50 mV/s. During the ﬁrst cycle (innermost
curve), the cathodic current for the metal deposition begins to gradually increase at ca.
350 mV vs. Ag/AgCl, with the signiﬁcant current onset at 0 mV. Based on the redox
reaction,l6

PtCl42' + 2e' (—> Pt + 4cr E" = 0.56 v vs. Ag/AgCl (4.2)
this represents an overpotential of over 200 mV for the deposition of Pt on diamond. The
result is in agreement with the observation of copper electrodeposition. The large
overpotential is attributed, in a very general way, to the inertness of hydrogen-terminated
diamond surface. The metal deposition likely proceeds by two irreversible reactions,
corresponding to sequential two-electron transfers (PtCloz' ——> PtCl42' —~> Pt), as has been
proposed for Pt electrodeposition on carbon ﬁber electrodes.'7 The reactions occur at
potentials sufﬁciently close such that each step was not well resolved for diamond (50
mV/s). The reduction peak at -240 mV is attributed to the adsorption of hydrogen on the

electrodeposited Pt particles, a reaction that is occurring simultaneously with the metal

103

 

 

30-
E 0'
2’

5 -3o-

3-

III

:1

U -60-
-90
-120

 

 

 

-400 -200 0 200 400 600 800 1000 1200

150

 

Figure 4.1. (A) Successive potentiodynamic i-E curves for a boron-doped diamond thin
ﬁlm in 1mM KthC16 + 0.1M HClO4. Scan rate = 50mV/s. (B) AFM force-mode image

(air) of the diamond thin film after 25 potential cycles.

deposition. The current crossover seen during the reverse sweep is a characteristic feature
of metal nucleation and growth.‘8 The voltammetric currents grow with cycle number
(the 25th scan is the outermost curve) due to the progressive increase in metal particle size
and distribution. All voltammograms, after the initial one, show the characteristic current
features for Pt: Pt oxide formation (700 - 1200 mV), oxygen evolution (1200 mV), Pt
oxide reduction (500 mV), hydrogen adsorption and desorption (0 — -300 mV), and
hydrogen evolution (-300 mV).

Figure 4.1B shows a top-view, 10 x 10 um2 AFM image (force mode, air) of the
diamond surface after 25 potential cycles. A well-faceted, polycrystalline film
morphology is observed with cubic, octahedral, and cubo—octahedral microcrystallites.
Also present are the intercrystalline grain boundaries where the microcrystallites join
together. The Pt deposits are not uniformly distributed over the surface but rather are
located almost exclusively at grain boundaries and other visible extended defects (see
arrows). The particle diameters range from 0.3 to 2 pm with a distribution density of ca. 4
x 107 cm'2 (geometric area). After 25 potential cycles, most of the surface is still void of
Pt and many of the particles have diameters up to ca. 2 um. These observations, and
others, suggest that the metal deposition preferentially initiates, under these conditions, at
the grain boundaries, or other surface defects during the ﬁrst cathodic scan, with a limited
number of Pt nuclei formed. 3D growth of the metal particles from these nuclei then
dominates the successive potential sweeps.

Galvanostatic deposition. It is clear the potentiodynamic deposition produced
particles with large diameters and poor distribution. A recent study has demonstrated that

constant current, as well as pulsed current, deposition can be used to successfully prepare

105

highly dispersed Pt nanoparticles on porous carbon substrates.14 Galvanostatic deposition
of Pt was carried out in 1mM KthCl6 + 0.1M HClO4 with a constant current density of

0.5 mA-cm'z. Forced convection (N2 bubbling) was employed in order to maintain the

electrode potential between from -100 to -300 mV vs. Ag/AgCl during the deposition.
Figure 4.2A-D shows top-view, 10 X 10 um2 AFM images (force mode, air) of diamond
surfaces after deposition for 100, 200, 300, and 400 s, respectively. The nominal particle
size, variances, and distribution density for each deposition time are summarized in Table
4.1.

The metal particles decorate the entire surface with much better distribution than
was achieved with the potentiodynamic method. The nominal particle size is also smaller.
Both are desirable trends for a catalytic electrode. After 100 s of deposition (Figure
4.2A), distinguishable Pt particles are uniformly dispersed over the diamond surface
decorating both the microcrystallite facets and the grain boundaries. The particle size
ranges from 20-200 nm with an average particle distribution of 0.5 x 109 cm'z. The
distribution was determined by manually counting the particles in AFM images from
multiple sites on the surface. Slightly higher distribution is observed after 200 s of
deposition (Figure 4.2B). The particle size ranges from 10 — 250 nm and the distribution

increases to 0.8 x 109 cm'z. A further increase in the distribution is seen after 300 s of

deposition (Figure 4.2C). The particle size ranges from 50 - 300 nm and the distribution

is 1.0 x 109 cm’z. Finally, somewhat larger particles and a higher distribution are observed

after 400 s of deposition (Figure 4.2D). The particle size ranges from 60 - 500 nm and the

distribution is 1.2 x 109 cm’z. The image features are consistent with a progressive

nucleation and 3D growth mechanism,’8 in which new Pt nuclei form progressively as a

106

function of time while the existing nuclei increase in size. Nuclei form on both the grain
boundaries and the facet surfaces. This indicates that both regions are electrochemically

active and support the ﬂow of current.

15)

 

Figure 4.2. AFM force—mode images (air) of boron-doped diamond thin ﬁlms after
galvanostatic deposition of Pt from a solution of 1 mM KthC16 + 0.1M HClO4. The

deposition times are (A) 100, (B) 200, (C), 300 and (D) 400 s, respectively.

107

 

5mm iv Nod v.0 H Wm gméo we ma: cod H NA oomém oov

q; Qm Ed md H ow Ommém 5m 3; 3.0 H o._ oomém 8m

odm Em omd 5o H ﬁn cemém w.m owd vod H wd OmNA: com

 

Q? 2 mmd m._ H o.m ommém Wm Ed Ed H We oomém o3
meson: menace
ewseo ﬁnesse: 2.5 ounce messes Es
COSEOmoQ bis—ea onm scant—Omen— htmcoa oﬁm
GE ”2 Ewen? 85.5.55 22:8 ..3 Swan»: essences 225m 3
amor— 05:.
ooatzm Hod< vacuum :OSBOQOQ

 

538w 95:56 beep—08m 05 Sam was 283 monotone—o

0:89:00 veep—«6P— 2: a8 E5 883 mwocswne use emano cosh—camp comp—U»; .cousnEmG .35. 22:3 5 A6 035,—.

108

 

Current (uA)

 

 

 

 

.300 A A 1 A A 1 A A A A L 1 A A
-400 0 400 800 1200 1600

Potential (mV vs. Ag/AgCl)

Figure 4.3. Cyclic voltammetric i-E curves in 0.1 M HClO4 for a boron-doped diamond
thin ﬁlm electrodeposited with Pt, during 1000 potential cycles between -400 and 1500
mV vs. Ag/AgCl at 50 mV/s. Dashed arrows show the decrease in current response with
scan number. The maximum anodic and/or cathodic current density during the cycling

was ca. 1 mA/cmz.

Stability of the electrodeposited Pt particles. Weak adhesion of electro-

'9~2° which is

deposited metal particles to the diamond surface is a common observation,
presumably attributed to the lack of strong interaction between the deposits and the non-
polar, hydrogen-terminated diamond surface. The stability of the electrodeposited Pt
particles was investigated, immersing the coated ﬁlm in 0.1 M HClO, and applying a

series of potential cycles between -400 and 1500 mV vs. Ag/AgCl. These potentials were

extreme enough to result in the cyclic generation of oxygen and hydrogen gas. One

109

thousand cycles were performed at a scan rate of 50 mV/s. Figure 4.3 shows a series of
cyclic voltammetric i-E curves recorded during the cycling period. There is a progressive
loss of the Pt features with cycle number. The hydrogen desorption charge prior to
cycling is 800 uC/cm2 and decreases signiﬁcantly to 25 uC/cm2 after cycling. This
represents a decrease of over 96%. The loss of electrodeposited Pt results from the

turbulence of the gas evolution, which physically dislodges the weakly adhering deposits.

4.3.2 Secondary Diamond Growth

Structural characterization. The secondary diamond growth is necessary to
stabilize the electrodeposited Pt particles by anchoring them into the surface
microstructure. The secondary growth is accomplished by placing the Pt-coated ﬁlm back
into the CVD reactor and depositing boron-doped diamond for a short period of time,
such that the ﬁlm does not completely cover the metal particles but only surrounds their
base. Figure 4.4A-D shows 5 x 5 umz AFM images (force mode, air) of Pt-coated ﬁlms
after 3 additional hours of diamond growth. The images are of the same Pt-coated ﬁlms
shown in Figure 4.2. Numerous particles are randomly distributed over the surface on
both the microcrystallite facets and in the grain boundaries, particularly for 200, 300, and
400 s electrodeposition times. The nominal particle size is slightly increased by the
secondary diamond growth, ranging from 30 - 500 nm, but the distribution is lower. The
process of entrappin g the particles with diamond reduces the total active Pt exposed. This
is seen by comparing the hydrogen desorption charge in Table 4.1. There is a 30-37%

loss in Pt activity, as measured in the hydrogen desorption voltammetric charge between

110

0 and —300 mV. This is consistent with the secondary diamond growth around the base of

the metal deposits. The loss is further discussed below.

 

Figure 4.4. AFM force-mode images (air) of Pt coated diamond thin ﬁlms (as indicated
in Figure 4.2) after a 3 h secondary diamond growth. The corresponding Pt deposition

times are (A) 100, (B) 200, (C), 300 and (D) 400 s, respectively.

111

 

Pt

 

 

 

PtPt
111

0.000 20.480

 

 

 

Figure 4.5. Energy dispersive x—ray analysis spectrum for a Pt metal particle shown in

the SEM image (inset).

SEM and semiquantitative x-ray ﬂuorescence measurements confirmed the
deposits are Pt. Figure 4.5 shows an SEM image of a composite diamond film and the
corresponding energy dispersive x-ray analysis spectrum taken from one of the metal
deposits on the film surface. The deposits occupy sites on both the facet surfaces and in
the grain boundaries. As stated above, this general observation indicates that both the
grains and the grain boundaries are electrically conductive with a low nominal resistance
to the ﬂow of current. It is most probable that film defects serve as the nucleation sites for
metal deposition. The C Kor (0.277 KeV), O K0t (0.525 KeV) and Pt M01 (2.05 KeV), Let

(9.44 KeV) emissions are observed in the EDX spectrum. The estimated atomic

112

 

percentage of each element is 94, 2, and 4, respectively, yielding PVC and O/C atomic
ratios of 0.04 and 0.02.

The x-ray diffraction spectrum (Cu source) of a Pt/diamond composite electrode
is presented in Figure 4.6, revealing the primary crystallographic orientations of the Pt
particles and the diamond microcrystallites. The Pt electrocatalyst displayed the
characteristic diffraction peaks of the Pt fcc structure, i.e., peaks at 40.2 degrees (111-
orientation), 46.6 degrees (200-orientation), and 67.8 degrees (220-orientation). It is clear
that the diamond (111) orientation dominates in the ﬁlm, with strong diffraction intensity
observed at 44.2 degrees. These data conﬁrm the crystalline nature of the diamond and

the incorporated Pt particles.

 

 

2000 ' ' ' I ' ' ' l ' ' ' I v - v t v v v y w v v
a 1
5
'5
1500 - g
E
.9. 1
£5 a
g
c 1000 '
U A
i:
C
soo- E s
5: E

 

 

 

 

30 40 50 60 70 80 90
2 0 (Degrees)

Figure 4.6. XRD pattern of a Pt/diamond composite electrode.

113

f

Basic electrochemical properties. Figure 4.7 shows the typical cyclic
voltammetric i-E curves for the Pt/diamond composite electrode in Nz-purged 0.1 M
HClO4, before and after the secondary diamond growth. The potential sweep was
initiated at 200 mV and scanned in the positive direction at a scan rate of 50 mV/s. All
the characteristic current features of polycrystalline Pt are present. Two reversible
hydrogen adsorption/desorption peaks are evident in the potential region from 0 and —300
mV.

Integration of the area under these peaks can be used to estimate the
electrochemically active surface area (or real surface area) of the Pt particles using the

equation21

 

S = Q’" (4.3)

in which S, is the real surface area, Q,,. is the charge associated with the formation of a
monolayer of adsorbates, e is the electron charge, and d,,, is the surface metal atom
density.

In the case for Pt, d,,, takes a value of 1.3 x 1015 cm'z, then

S,(cm2)= Qm(1“C) _. (4.4)
210(,uC-cm ')

 

From the knowledge of S,, the roughness factor p can be calculated as

 

=’ 4.5
PS ()

where 5,, is the geometric area of the diamond electrode.
The results are presented in Table 4.1. Both the hydrogen desorption charge and

the roughness factor increase with the deposition time (i.e., Pt loading). Assuming a

114

100% current efficiency, the mass of Pt deposited is calculated to be 25.2, 50.3, 75.8 and
100.1 ug-cm'2 for the 100, 200, 300 and 400 s deposition times, respectively. Thus, the
hydrogen desorption charge per mass of Pt, after the secondary diamond growth, is

calculated to be 12.6, 11.1, 10.7, and 9.2 C-g", respectively. The decreasing ratio reﬂects

the increasing nominal metal particle size.

The voltammetric curve for the Pt-coated diamond film, prior to the secondary
growth (solid line), was obtained after the first scan, and the shape remained unchanged
during 10 subsequent scans. However, the voltammetric curve for the Pt-coated diamond
ﬁlm, after the secondary diamond growth (dashed line), had a poor ﬁrst scan, especially
in the hydrogen adsorption/desorption region. In fact, many scans (100 cycles) were
required for the characteristic Pt features to develop. This observation is consistent with
the removal of surface contamination formed during the diamond deposition. It is highly
probable that some of the Pt surfaces are initially covered with adsorbed carbon and/or
hydrocarbon moieties, and these contaminants block surface sites for the hydrogen
adsorption. The contaminants are effectively removed by potential cycling between the
oxygen and hydrogen evolution regimes‘. Hence, potential cycling is the normal protocol
for cleaning and activating the Pt surface after the secondary diamond film growth. The
potential cycling (SO-100 cycles) is performed in 0.1M HClO4 at scan rate of 50 mV/s
between —400 mV and 1500 mV. The voltammogram shown for the Pt-diamond
composite electrode is for the 100‘h cycle. The ﬂat oxide-formation region at potentials
between 700 and 1200 mV indicates that the oxidation of adsorbed contaminants is not

occurring to any appreciable extent.

115

 

150 r

100 -

50-

 

 

Current (uA)

 

-150 ' '

 

 

 

-200 A A A A A A A A A A A A A A A A A A A
-400 0 400 800 1200 1600

Potential (mV vs. Ag/AgCl)

Figure 4.7. Cyclic voltammetric i-E curves for a Pt-coated diamond thin film in 0.1M
HClO4 before (solid line) and after (dashed line) a 3 h of secondary diamond growth. The

Pt deposition time was 2003. Scan rate = 50 mV/s.

There are, however, some notable differences between the two voltammograms.
The hydrogen adsorption/desorption peaks, the Pt oxide formation/reduction waves, and
the double-layer charging current between 100 and 250 mV are reduced after the
secondary diamond growth, indicating a reduction in the active Pt surface area. As
indicated in Table 4.1, an average of 34% of the active surface atoms are lost, based on
the reduction in the hydrogen desorption charge.

Mechanistic aspects of secondary diamond growth. Several possible causes for

the reduced Pt activity were systematically investigated. One possibility was particle

116

aggregation on the diamond surface at the deposition temperature of ca. 850 0C. Another
possibility was gasification or sputtering of the Pt by the atomic hydrogen present in the
plasma. An experiment was performed to examine these two possibilities, in which
another set of Pt-coated diamond films, similar to those shown in Figure 4.2, were placed
into the CVD reactor. The electrodes were then exposed to the conditions used for the
secondary diamond growth except that no CH4 was added. The absence of CH; did not
signiﬁcantly affect the substrate temperature, at least as determined by an optical
pyrometer. Again the morphology was examined by AFM and the active surface area was
determined from the hydrogen adsorption/desorption voltammetric charge in 0.1M
HClO4 before and after a 3-h hydrogen plasma treatment. No signiﬁcant change in either
the particle size or voltammetric charge for hydrogen desorption was observed (AFM
images not shown here), at least for Pt particles with diameters from 20 to 300 nm.
Therefore, the loss in active Pt was not due to particle aggregation, gasification or
sputtering. Voltammetric measurements indicated that only 2-4%, not 30-37%, of the
active Pt was lost during the 3-h hydrogen plasma treatment. The melting and smoothing
of small particles may contribute to this slight activity loss. Although the melting
temperature of bulk Pt is 1772 0C, nanometer-sized Pt particles have a much lower
melting temperature, as indicated by the size-dependent melting temperature theory. 22 As
the size of the particle decreases, an increased proportion of atoms occupy surface or
interfacial sites. These atoms are more loosely bound than bulk atoms, which facilitates
the melting.23 Pt particles with diameters of less than 2 nm have a melting point less than
850 0C, based on the application of the theory. It should also be noted that segregated Pt

particles, rather than a continuous film, can be formed on different substrates by high

117

temperature annealing in vacuum.”25 The break-up of an initially continuous or quasi-
continuous Pt ﬁlm into individual Pt aggregates on a quartz substrate has been achieved

by vacuum heat treatment at 600 K. 24

’ ‘\ fl.
.4”
. ..- ...

   

Figure 4.8. SEM images of a Pt/diamond composite electrode (A) before and (B) after

etching in aqua regia. The Pt deposition time was 200 s.

Confirmation that the secondary diamond growth causes the loss of active Pt by
covering the base of the metal deposits was next addressed. In the experiment, a
Pt/diamond composite electrode was etched in hot aqua regia (3 HCl : 1 HNO; (v/v)) to
dissolve the Pt particles. The diamond lattice is quite stable in this solution and undergoes
no morphological alteration. SEM images of the composite electrode, before and after the
acid etching, are presented in Figure 4.8A and B. Pt particles with diameters of 50 - 300
nm are seen on the surface prior to the chemical treatment (Fig. 4.8A). Numerous pits are
observed in the diamond film after treatment. These pits are the voids left after dissolving
the metal, and have diameters of approximately the same range. This observation

conﬁrms that the diamond film does surround the base of the metal particles entrapping

118

and anchoring them in the surface microstructure. The depth of the pits depends on the
secondary diamond growth time.

The secondary diamond growth can also completely cover some of the small Pt
particles, once the thickness of the ﬁlm exceeds the height of the particles. This
postulation has been conﬁrmed by the cross-section analysis using SEM. Figure 4.9A
shows a cross sectional image of an acid-etched Pt/diamond composite electrode. The
corresponding back scattering SEM image is shown in Figure 4.9B. Elements with higher
atomic number yield brighter images than lighter elements in the back-scattered electron
imaging mode. Three bright spots are observed in Figure 4.9B, which corresponds to
three metal particles indicated as l, 2, and 3 in Figure 4.9A. EDX analysis further
confirmed the particles are Pt. The fact that these particles were not dissolved during the
acid-etching procedure indicates they are completely covered by the corrosion—resistant
diamond ﬁlm.

The thickness of the secondary diamond film, and thus the growth rate, can be
estimated from the vertical distance between the film surface and the bottom of the
particles or the pit (see arrow 4). The growth rate, ca. 0.04-0.1 limb", is far less than the
growth rate, ca. 0.3 rim-h", generally observed for diamond growth on Si in our reactors.
Mechanistic studies are in progress focused on understanding the inﬂuence of the foreign
metal particles on the diamond nucleation and growth. It is supposed that the above
observations are attributable to at least two reasons. First, the diamond nucleation and
growth rate, in the presence of Pt, is rather low. The assessment is based on observations

made by Belton et al.,26 and Tachibana et a].27 during the study of diamond growth on Pt.

119

 

 

Bottom
edge

 

B lum

 

Figure 4.9. Cross-sectional SEM images of an acid-etched Pt/diamond composite
electrode. (A) Secondary electron imaging mode. (B) Back-scattered electron imaging
mode. The secondary diamond growth time was 3 h. Arrow 1-3 show three Pt particles

covered by the diamond film. Arrow 4 shows a cavity after dissolving Pt.

120

Pt could catalyze the conversion of diamond to graphite, a process that competes with the
diamond growth. The early stage of diamond growth likely involves the formation of
thin-layer of graphitic carbon on the Pt surface, followed by etching of surface graphite
and formation of adsorbed hydrocarbon deposits as the growth continues. Therefore, the
fact that sufﬁcient time is needed to etch the graphitic species, leads to an apparent low
growth rate during the early stage. Second, an “undergrowth” mechanism may be
involved. As demonstrated in the electrochemical study, the hydrogen-terminated
diamond surface lacks anchoring sites to stabilize the electrodeposited Pt particles.
Therefore, the secondary diamond film could be deposited beneath the particles, as well

as surrounding the base.

4.3.3 Stability of the PtlDiamond Composite Electrode

The dimensional stability of the Pt/diamond composite electrodes was examined
during a 2-h exposure to 85% H3PO4 acid at 170 0C and 0.1A/cm2. Electrochemical and
AFM measurements revealed no loss in Pt activity and no degradation of the diamond
rnicrostructure after 2 h of electrolysis.

Figure 4.10 shows cyclic voltammetric i-E curves for a Pt/diamond composite
electrode in 0.1 M HClO4 before and after two l-h periods of anodic polarization. The
curve for the electrode prior to the polarization (dashed line) reveals the presence of Pt
with the expected features. After the two l—h polarizations, the voltammetric features are
unchanged and revealing that there is no loss of catalyst activity due to degradation of the
diamond rnicrostructure and morphology. Irnportantly, there is no loss in the charge

associated with hydrogen ion adsorption and desorption. Such loss would be expected if

121

the Pt catalyst particles were detached from the surface due to an oxidizing and corroding
diamond support. In fact, the charge associated with the hydrogen ion adsorption actually
increases after the electrolysis. The cathodic charge is 355 uC-cm'2 before and increases
to 420 and 455 jiC-cm'2 after the two l-h polarizations, respectively. The increased
charge is attributed to minor surface cleaning and crystallographic changes in the deposits
that occur during the vigorous gas evolution. One type of minor cleaning that is possible
is the oxidative removal of residual carbon deposits formed during the secondary
diamond deposition. These deposits do not affect the stability of the metal particles but,
rather, inﬂuence their surface activity toward faradaic electron transfer processes. There
is no significant change in the particle size and coverage after polarization, at least as
revealed by AFM. Some representative images are shown below. The most signiﬁcant
change in the voltammograms is the reduced overpotential for oxygen evolution after the
polarizations. The current associated with the Pt-oxide reduction increases after the
electrolysis and the current maximum shifts to slightly more negative potentials. There is
also a minor decrease in the overpotential for hydrogen evolution after the polarization.
Figure 4.11A and B show ex situ AFM images of the Pt/diamond composite
electrode before and after the two l-h polarizations. A well faceted, polycrystalline
morphology is observed before and after electrolysis. The crystallites are randomly
oriented with hemispherical Pt dispersions decorating both the facets and grain
boundaries. Clearly, there is no evidence of any severe morphological or microstructural
damage such as film delamination, grain roughening, or pitting. The similarity of the

image features before and after polarization is consistent with the voltemmetric data.

122

 

 

 

 

 

_200 A A A A
-400 0 400 800 1200 1600

Figure 4.10. Cyclic voltammetric i-E curves for a Pt/diamond composite electrode in 0.1
M HClO4 before (dashed line) and after two l-h polarizations (solid lines) in 85 wt %

H3PO4 at 170 0C and 0.1 A-cm‘z. The Pt deposition time was 200 s.

 

Figure 4.11. AFM images (air) of the Pt/diamond composite electrode (A) before and (B)

after 2-h anodic polarization in 85 wt % H3P04 at 170 0C and 0.1 A-cm'z.

123

4.4 Conclusions

We report on a new, dimensional stable Pt/diamond composite electrode that is
fabricated by using a multistep procedure of diamond deposition, Pt electrochemical
deposition, and diamond deposition. Dispersed Pt particles are galvanostatically
deposited onto a boron-doped polycrystalline diamond thin-ﬁlm surface and entrapped
within the dimensionally stable rnicrostructure by a subsequent diamond deposition. The
second, short deposition serves to anchor many of the metal particles into the diamond
rnicrostructure by surrounding their base. The number of exposed particles, particle size,
and distribution can be controlled, to some extent, by adjusting the galvanostatic
deposition time and secondary diamond growth conditions. Particle sizes are in the range
of 30 — 500 nm with a distribution of ~ 109 cm’z. This nominal particle size is still too
large for a real catalytic electrode, and work is ongoing to reduce the particle size to the
10 nm range. About 34% of the active Pt surface is lost after the secondary diamond
deposition due to a combination of complete coverage of some of the smaller particles
and partial coverage of the base of the larger particles. The catalytic activity of the
composite electrodes is extremely stable, as no microstructural alterations or activity

losses were observed during a 2 h anodic polarization in 85% H3PO4 at 0.1 A~cm'2 and

170°C.

124

References:

10.

11.

12.

13.

14.

15.

16.

17.

J. Wang, G. M. Swain, T. Tachibana, and K. Kobashi, Electrochem. Solid State
Lett., 3, 286 (2000).

J. Wang, G. M. Swain, T. Tachibana, and K. Kobashi, J. New Mater.
Electrochem. Sys., 3, 75 (2000).

J. S. Xu, M. C. Granger, Q. Y. Chen, J. W. Strojek, T. E. Lister, and G. M. Swain,
Anal. Chem, 69, A591 (1997).

G. M. Swain, A. B. Anderson, and J. C. Angus, Mrs. Bulletin, 23, 56 (1998).
G. M. Swain, J. Electrochem. Soc., 141, 3382 (1994).

R. Tenne and C. Levy-Clement, Israel J. Chem, 38, 57 ( 1998).

Y. V. Pleskov, Russian Chemical Reviews, 68, 381 (1999).

T. Yano, D. A. Tryk, K. Hashimoto, and A. Fujishima, J. Electrochem. Soc., 145,
1870 (1998).

J. V. Zoval, J. Lee, S. Gorer, and R. M. Penner, J. Phys. Chem. B, 102, 1166
(1998).

J. V. Zoval, R. M. Sriger, P. R. Biemacki, and R. M. Penner, J. Phys. Chem, 100,
837 (1996).

H. Natter and R. Hempelmann, J. Phys. Chem, 100, 19525 (1996).
H. Natter, M. Schmelzer, and R. Hempelmann, J. Mater. Res., 13, 1186 (1998).
M. W. Verbrugge, J. Electroanal. Chem, 141, 46 (1994).

M. S. Lofﬂer, B. Grob, H. Natter, R. Hempelmann, T. Krajewski, and J. Divisek,
Phys. Chem. Chem. Phys., 3, 333 (2001).

E. Budevki, G. Staikov, and W. J. Lorenz, Electrochemcial Phase Formation and
Growth, VCH, New York, 1996.

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

N. Georgolios, D. J annakoudakis, and P. Karabinas, J. Electroanal. Chem, 264,
235 (1989).

125

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

B. Schariﬂ<er and G. Hills, Electrochim. Acta, 28, 879 (1983).

S. Nakabayashi, D. A. Tryk, A. Fujishima, and N. Ohta, Chem. Phys. Lett., 300,
409 (1999).

N. Vinokur, B. Miller, Y. Avyi gal, and R. Kalish, J. Electrochem. Soc., 146, 125
( 1999).

J. M. D. Rodriguez, J. A. H. Melian, and J. P. Pena, J. Chem. Ed., 77, 1195
(2000).

K. F. Peters, J. B. Cohen, and Y. W. Chung, Phys. Rev. B, 57, 13430 (1998).

M. Zhang, M. Y. Efremov, F. Schiettekatte, E. A. Olson, A. T. Kwan, S. L. Lai,
T. Wisleder, J. E. Greene, and L. H. Allen, Phys. Rev. B, 62, 10548 (2000).

W. Romanowski and R. Lamber, Thin Solid Films, 127, 129 (1985).
E.I.Altman and R. J. Gorte, Surf Sci., 172, 71 (1986).
D. Belton and S. Schmei g, Surf. Sci., 233, 131 (1990).

T. Tachibana, Y. Yokota, K. Miyata, T. Onishi, K. Kobashi, M. Tartutani, Y.
Takai, R. Shimizu, and Y. Shintani, Phys. Rev. B, 56, 15967 (1997).

126

Chapter 5

Oxygen Reduction Reaction on PtlDiamond
Composite Electrodes — A Rotating Disk
Voltammetric Study

5.1 Introduction

The oxygen reduction reaction (ORR) is one of the most widely studied
electrochemical reactions due to its considerable importance in several ﬁelds of research,
including metal corrosion, electrolysis, and electrochemical energy conversion and
storage."4 Despite the considerable attention paid by electrochemists during the past three
decades, the ORR remains an important ﬁeld of research due to the complex reaction
kinetics and the need for improved electrocatalysts.4 The reaction kinetics and
mechanism are known to be inﬂuenced by several factors, including the type of cathode
material (e.g., electrocatalyst), supporting electrolyte, and pH. For example, the reaction
may involve a 2-electron pathway, in which oxygen is reduced to peroxide. This is the
primary reaction on most carbon, gold, mercury, and transition-metal oxide electrodes.2
The reaction may also involve a 4-electron pathway, in which oxygen is directly reduced
to water. This is the primary reaction on noble metals (e.g. Pt, Pd, Ag), metal oxide, and
some transition-metal macrocyclics.2 The 4—electron pathway is preferred for many
applications, such as fuel cells, because it leads to a high power density and a non-

polluting final product (H20).

127

Platinum dispersed on high-surface-area carbon supports is still the best
electrocatalyst employed in low temperature proton exchange membrane fuel cells (PEM—
FC) for oxygen reduction. The ORR on Pt proceeds principally through the 4-electron
pathway with a thermodynamic electrode potential of 1.23 V vs. SHE in acid solution
(pH = 0).5 However, the electrode kinetics are rather slow at high potentials because of
Chemisorbed species on Pt surface (PtO and PtOH). The surface oxides inhibit the OR
and reduce the operating potential of the electrode to ca. 0.8 V vs. SHE.6 Since the
overpotential for the anodic reaction is negligible when pure hydrogen is used as the fuel,
the slow ORR kinetics is the major factor limiting the overall energy conversion
efficiency in the PEM-FC.7 Another limitation is the susceptibility of the carbon support
material to electrochemical corrosion. The oxygen cathode is exposed to oxidizing
conditions, which can cause damaging corrosion of carbon, particularly at elevated
temperatures.8 Electrochemical corrosion can limit the cathode potential and produce
microstructural degradation and surface chemical changes, which cause signiﬁcant loss in
the catalytic activity.8

As described in Chapters 3 and 4, conductive diamond thin films are an advanced
material that serves as a superb electrocatalyst support/host. Nanometer-sized Pt particles
can be incorporated into the surface microstructure of boron-doped microcrystalline
diamond thin ﬁlms via a multistep diamond deposition/ Pt deposition/ diamond
deposition procedure. The dispersed Pt particles are stabilized by the growth of a thin-
ﬁlm of diamond around their base and are in good electrical communication with the
current collecting substrate through the boron-doped diamond base. The resulting

dimensionally stable electrode can withstand high current densities (2 0.1 Acm'z), both

128

anodically and cathodically, without any morphological and microstructural degradation,
even under demanding conditions (e.g., 85% H3PO4 and 170 °C).

In this chapter, results from the study of the ORR at Pt/diamond composite
electrodes in acidic media using rotating disk voltammetry are presented. Rotating disk
voltammetry is commonly used to study this reaction, because it allows for the separation
of the mass-transfer effects from the electron transfer kinetics, and provides a
measurement of the true catalytic activity.5 The composite films were deposited on
molybdenum disks (5mm diameter) rather than Si wafers because of the hardness of the
metal, which allows for the easy assembly as a rotating disk electrode (RDE). The open
circuit potential was measured in oxygen-saturated 0.1 M H3PO4, H2804, and HCIO4 at
room temperature. The kinetic parameters (e. g. Tafel slope and exchange current density)
were determined as a function of the electrolyte composition and Pt loading in the same

acid solutions.

5.2 Mechanism and Kinetics of OR

Mechanism. The ORR on Pt in acid solution follows a complicated reaction
mechanism which includes a number of elementary steps involving different reaction
intermediates. A simplified reaction scheme is given below.4 The ORR proceeds through
either a “direct” 4-electron pathway, in which oxygen is reduced to water with a rate
constant, k], without any intermediate formation, or a “series” 4-electron pathway, in
which oxygen is first reduced to HzOz,ads with a rate constant, kg, and adsorbed peroxyl

species are further reduced to water with a rate constant, k3, before they desorb.

129

k1

 

4e’, 4H”

[(2 k3 '

 

 

02,8 02,8d8 23., 2H+ > ”202,“: 29., 2”!» > 2H20

It is widely accepted that these two pathways generally occur simultaneously.4
However, the reaction rate is rather structure sensitive. Whether splitting of the O-O bond
occurs before the peroxide formation is strongly dependent on the arrangement of the
atoms on the electrode surface and anion adsorption. For example, strongly adsorbed Cl'
and Br' can block the Pt sites needed for the dissociative chemisorption of Oz, favoring
the “series” reaction pathway.9'lo

In either case, the following reaction steps take place4

(02)sol + Pt '9 Pt'(02)ads (51)
Pt-(02)sol + H+ + e' —) Pt-(H02)ads (5.2)
Pt-(HOz)so. + 3H+ + 3e' —+ Pt + 2HzO (5.3)

Kinetics. The rate-determining step is the first charge transfer to 0; (Reaction

5.2) with the approximate rate expression beinglo

 

'ﬂFmepFJ—rgﬂ) (5.4)

i=anC 1—@ ‘ex
0=( M) p( RT RT

in which n is the number of electrons, k is the rate constant, C01 is the concentration of

Oz in the solution, Gad is the total surface coverage by adsorbed species, x is the number

130

of Pt sites occupied by the adsorbed ion, B and y are the symmetry factors (assumed to be
'/2), and ram is the parameter characterizing the rate of change of apparent standard Gibbs
free energy of adsorption with the surface coverage by adsorbing species. In deriving
equation 5.4, it is assumed that the reaction intermediates (e.g., Oz'ms, H021“, and H02’
.ads, etc) are adsorbed at low coverage, i.e. they are not a significant part of Gad.
Therefore, the kinetics of the ORR are largely determined either by free platinum sites
available for adsorption of Oz (l-Oad term in Equation 5.4) and the Gibbs free energy of
adsorption of reaction intermediates by adsorbing anions and/or surface oxides (rOad term
in Equation 5.4). In other words, the case of r = 0 is synonymous with Langmuir
adsorption conditions (i.e., no interaction takes place between the adsorbed molecules),
whereas for r at O Temkin adsorption conditions are obtained (i.e., the Gibbs energy of
adsorption of the reaction intermediates is dependent on the surface coverage of

adsorbates).

5.3 Experimental

Electrode preparation. Diamond thin films were deposited on polycrystalline
Mo disks following the same procedure as described in Chapter 4. The substrates (2 mm
thick x 5 mm in diameter, Aldrich, Inc.) were polished first with 800-grit sandpaper, then
successively with A1203 water suspensions (particle size 1.0, 0.3, 0.05 pm), and cleaned
ultrasonically in ultrapure water for 5 min after each polishing step. The substrates were
rinsed in toluene, methylene chloride, acetone, isopropanol, and methanol. The cleaned

substrates were then sonicated in a diamond powder/acetone slurry (0.1 pm diam., GE

131

Superabrasives, Worthington, OH) for 20 min followed by a rinse with clean acetone.
The seeded substrates were then placed in the CVD reactor, allowing for a 15-h diamond
deposition with the hydrogen ﬂow rate of 200 sccm and methane of 0.70 sccm. The
Pt/diamond composite electrodes were then fabricated following the same procedure

described in Chapter 5 with the Pt deposition time ranging from 100 to 500 s.

 

a) (l7 b) Teﬂon

/ Teﬂon

 

 

 

 

 

 

 

/
Co per .
Disk
NJ) electrode
/ Disk
electrode
* Pt
\ I c)
‘\ /' n or n Dir-damond
'_ Streamlines -* m ' (2 growth)
‘\ /' .-. .1... . ;;..'\ Diamond
(1“ growth)

      

Mo

4”;

    

'41:

 

 

 

Figure 5.1. Schematic diagram of a rotating disk Pt/diamond composite electrode: a) a
top and side view showing the direction of mass transport; b) a view of the electrode

mounted in Teﬂon; c) a side view of the composite disk electrode architecture.

The schematic diagram of a rotating disk Pt/diamond composite electrode is

shown in Figure 5.1. The electrode was mounted into the disk position of an insulating

132

mantle. The geometric area of the electrode is 0.196 cmz. A piece of copper rod was
soldered to the backside of the Mo disk substrate in order to reduce the contact resistance.
The RDE was controlled by a Metrohm 628—10 RDE controller (Metrohm, Switzerland)
with a rotation rate from 500 to 3000 rpm.

Instruments and chemicals. All the electrochemical measurements were
performed in conventional three-compartment electrochemical cell with a CHI 650A
digital potentiostat (CH Instruments, Austin, TX). A Ag/AgCl electrode served as
reference electrode. In order to avoid the contamination of the electrolyte, a glass
junction containing the electrolyte was used to connect the reference electrode to the cell.
For the ORR experiments, the solution was saturated with oxygen gas (99.9%, BOC
Gases) by bubbling for 30 min. prior to any electrochemical measurement. All solutions
were prepared with ultrapure water from a Bamstead E-Pure puriﬁcation system (>17

MSZ-cm). The HClO4, H2804, and H3PO4 were ultrahigh purity grade (99.999% Aldrich).

All glassware was cleaned in a KOH/methanol bath prior to use.

5.4 Results and Discussion

5.4.1 Characterization of the Diamond Film on Mo Disk

Prior to performing the Pt electrodeposition, structural and electrochemical
characterization of the diamond ﬁlms was conducted. An important consideration is the
uniformity of the diamond coating as pinholes and small crevices can provide pathways
for solution to permeate through the ﬁlm and reach the underlying substrate. In such
cases, electroactive substrates will directly participate in the electrochemical reactions,

and the potential-dependent current will be enhanced.

133

Figure 5.2 shows a typical SEM image of the polycrystalline diamond ﬁlm
deposited on Mo disk substrate. The image reveals a well—faceted ﬁlm composed of
octahedral, cubic, and cubo-octahedral microcrystallites, grain boundaries, some
secondary growths (i.e., smaller growth features on top of the large facets and grain
boundaries), and growth hillocks. The grain size is in the range of several hundreds of
nanometers to 211m. The grains are heavily twinned. The diamond (111) orientation is
dominant in this ﬁlm, as inferred from the x-ray diffraction measurement. The diamond
ﬁlm is continuous over the substrate with no pinholes and crevices observed. Diamond
growth on the molybdenum substrate is preceded by the formation of an intermediate
layer of MoC which is electrically conductive. The conducting interlayer ensures good

electrical and mechanical connection between the Mo substrate and the diamond coating.

 

Figure 5.2. SEM image of a boron-doped diamond thin film deposited on a Mo disk. The

diamond deposition time was 15 h and methane-to-hydrogen ratio was 0.35%.

134

 

Intensity (arb. units)

 

 

JL-

\

 

 

‘ A m l A

1100 1200 1300 1400 1500 1600 1700 1800 1900

 

Raman shift (cm'l)

Figure 5.3. Raman spectrum for a boron-doped diamond thin ﬁlm deposited on a Mo
disk. Excitation wavelength = 532 nm. Incident power density = 500 chm'z. Integration

time = 10 s.

A characteristic Raman spectrum for the diamond film is presented in Figure 5.3.
The ﬁrst-order diamond phonon line is observed at 1337.2 cm‘l and has a full width at
half maximum (FWHM) of 10.1 cm". The peak position is shifted some 5 cm'1 toward
higher wave numbers, indicating the presence of compressive stress within the film.
Compressive stress is expected because of the difference of the thermal expansion
coefﬁcient for the film and substrate. The FWHM is much larger than the value of 2-3
cm'1 that is routinely observed for single crystal diamond. The increased width partially

results from stress heterogeneities and stress induced splitting of the triplet zero phonon

135

line into a singlet and a doublet.ll The FWHM is also a measure of the crystalline
perfection, and to a first approximation, is inversely related to the phonon lifetime (i.e.,
defect density). Another main feature in the spectrum is the broad band in the range of
the 1450-1650 cm'l that is assigned to the nondiamond carbon impurity in the ﬁlm.
However, the level of this impurity is very low considering that the peak intensity is weak
and the scattering cross-section for graphite (a model for the .nondiamond carbon
impurity) is ca. 50 times larger than that for diamond.

Electrochemical methods of analysis are quite sensitive probes of the
physicochemical properties of diamond electrode. Cyclic voltammetry can be used to
ascertain information about the diamond ﬁlm quality, as the background current and the
working potential window are highly sensitive to the presence of the electroactive
substrate material, as well as the nondiamond carbon impurity. Figure 5.4 shows a typical
cyclic voltammetric i-E curve for a diamond ﬁlm deposited on a Mo disk in 0.1 M
HClO4. The i-E curve is ﬂat and featureless over a wide potential range (3.4 V at i 50
11A). The scans are stable with multiple sweeps and no background features associated
with Mo are evident, indicating that there is no penetration of the electrolyte solution
through the grain boundaries or defects to the reactive substrate. In other words, the
molybdenum substrate is completely coated with the diamond film. However, the
background current for diamond films deposited on Mo is 3 11A at 250 mV, slightly
larger than the current for the films deposited on Si. The slightly larger current is most
likely due to some nondiamond carbon impurities at the grain boundaries and defect sites,

consonant with the observations obtained in the Raman spectrum.

136

 

 

Current (uA)

 

 

 

. 4——
-10 .
.20 - .
-30 -
_40 .
-50 A A A L A A A A L A A A A
-l400 -900 -400 100 600 1100 1600 2100

Potential (mV vs. Ag/AgCl)

Figure 5.4. Cyclic voltammetric i-E curves for a boron-doped diamond thin film

deposited on a Mo disk in 0.1 M HClO4. Scan rate = 100 mV/s.

 

200;—Rr)1«:in 4:

 

Current (11A)

 

 

 

 

-0.30 0.00 0.30 0.60 0.90 1.20 1.50

Potential (V vs. Ag/AgCl)

Figure 5.5. Cyclic voltammetric i-E curves for Pt/diamond composite electrodes in 0.1 M
HClO4. The Pt deposition times were 200, 300, and 400 s, respectively. Scan rate = 50

mV/s.

137

Table 5.1. Electrochemically active surface area and roughness factor for the

Pt/diamond composite RDEs with different Pt loadings.

 

 

Electrode Pt Maximum Hydrogen Roughness Active
deposition Pt loadinga adsorption/ factor surface
time (l.lg/Cm2)b desorption charge area
(s) (mC/cmz)b (cmz)
RDE #1 200 50.3 0.50 2.4 0.48
RDE #2 300 75.8 0.71 3.4 0.68
RDE #3 400 101.1 0.84 4.0 0.80

 

a. Assuming 100% current efﬁciency for the 4-electron transfer reaction (Pt4+ —> Pt).

b. Referred to the geometric surface area of the electrode (0.196 cmz).

Based on these observations, it can be concluded that the diamond film is
continuous over the surface, and the morphology and microstructure are stable such that
solvent/electrolyte does not permeate the film and attack the electroactive substrate. The
Pt/diamond composite rotating disk electrodes were next fabricated. Figure 5.5 presents
the cyclic voltammetric i-E curves for three Pt/diamond composite RDEs with varying Pt
loadings. The curves were recorded after 100 cycles in 0.1 M HClO4 at a scan rate of 50
mV/s. Potential cycling between the hydrogen and oxygen evolution region (-0.3 V to 1.5
V) is an efﬁcient way to clean the Pt surface of residual amorphous carbon and other
impurities deposited during the secondary diamond growth. The characteristic
electrochemical features of clean platinum (poly) gradually develop with potential

cycling. As expected, the charge for the hydrogen adsorption/desorption (between —O.25

138

 

V and 0.1 V), Pt oxide formation (at ca. 0.65 V) and Pt oxide reduction (at ca. 0.45 V) all
increase with increasing Pt loading. The double layer charging current observed between
0.1 and 0.25 V also increases due to the increase in Pt surface area as the loading
increases. The hydrogen adsorption/desorption charge was used to estimate the active Pt

surface area and the surface roughness factor. The results are presented in Table 5.1.

5.4.2 ORR at Stationary PtlDiamond Composite Electrodes
Measurement of the open circuit potential. The open circuit potential (OCP) is
often used as an indicator of the electrocatalytic activity of an electrode. After the
extended potential cycling, the open circuit potential was measured in oxygen-saturated
0.1 M acid solutions at room temperature. An immediate decay of the OCP was observed
in all the solutions before the potential reached a steady-state value. Steady-state OCPs in
the different acid solutions, for the electrodes varying in Pt loading, are presented in
Table 5.2. The OCP is, on average, ca. 0.74 V vs. Ag/AgCl (0.94 V vs. SHE) for all the
electrodes in 0.1 M HClO4, similar to values reported for polycrystalline Pt and Pt coated

with Naﬁon®.'2 The equilibrium potential is 0.972 V vs. Ag/AgCl for the four-electron

transfer ORR in acid solution (pH = l) at room temperature.5 Therefore, the Pt/diamond
composite electrodes attain an overpotential loss of ca. 0.23 V in 0.1 M HClO4,
corresponding to a 24% efficiency loss. The overpotential losses have been attributed
mainly to the mixed potential that is established at the electrodes.6 The mixed potential is
due to the cathodic process, i.e., slow oxygen reduction and the competing anodic

process, such as the platinum oxide formation and/or impurity oxidation.6

139

Results in Table 5.2 also indicate that the OCPs are independent of the Pt loading
in a particular acid solution. However, the OCP values do depend on the electrolyte type.
The nominal OCP is ca. 10 mV more negative in 0.1M H2804 and ca. 25 mV more
negative in 0.1 M H3PO4. The negative shift is attributed to the anion adsorption effect.‘3
The specific adsorption of these anions increases in the order HzPO4' > S04 2' (or HSO4) '

> ClO4'. The specific adsorption of anions blocks the active sites needed for the

chemisorption of oxygen, resulting in more sluggish electrode kinetics.

Table 5.2. Open circuit potentials (25 0C) for the Pt/diamond composite RDEs in

different acid electrolytes.

 

 

Electrode 0.1 M HClO4 0.1 M H2804 0.1 M H3PO4
RDE#I 0.742 0.727 0.720
RDE#2 0.748 0.735 0.721
RDE#3 0.744 0.738 0.718

 

Note: Values are V vs. Ag/AgCl.

Voltammetric studies of ORR. The ORR at Pt/diamond composite RDEs was
initially studied by cyclic voltammetry in quiescent solution. Figure 5.6A (solid line)
shows a typical cyclic voltammetric i-E curve for ORR on a Pt/diamond composite
electrode. For comparison, the voltammetric response for a polycrystalline Pt foil
electrode is shown in Figure 5.6B. The experiments were carried out in Oz-saturated 0.1
M HCIO4 at room temperature. The potential sweep initiated at 1.2 V, whereupon the Pt

surface is covered with a thin oxide ﬁlm, scanned to —0.25 V, and then scanned back to

140

 

100 f , - , - . - , - , - . -

 

 

A

 

 

A L

-0.30 -0.10 0.10 0.30 0.50 0.70 0.90 1.10 1.30

 

 

1% v i v ' V ' v ' v ' v V v ' v

Current (uA)

 

 

 

B .

 

 

A I A 1

 

-4“) A A A A A A A A A A A
-0.30 -0.10 0.10 0.30 0.50 0.70 0.90 1.10 1.30

Potential (V vs. Ag/AgCl)

Figure 5.6. Cyclic voltammetric i-E curves for (A) RDE #1 and (B) polycrystalline Pt
foil electrode in Nz-purged (dashed line) and Oz-saturated (solid line) 0.1 M HClO4, Scan

rate 50 mV/s. Active surface area of the Pt foil electrode is 0.43 cm2 (geometric area is

0.2 cmz).

141

1.2 V. The response for the Pt/diamond composite electrode resembles that for the clean
Pt foil. During the cathodic scan direction, discernible reduction current commences at
ca. 0.72 V, close to the open circuit potential in this electrolyte. A dramatic increase in
the cathodic current is observed at potentials below 0.55 V, accompanied with the
reductive removal of the Pt oxide, in agreement with the fast ORR kinetics on oxide-free
Pt surface. The current reaches a maximum of 0.33 mA at ca. 0.45 V, and decays rapidly
on the low potential side of the peak due to the depletion of oxygen at the electrode
surface (diffusion controlled). Evidently, oxygen reduction continuously contributes to
the total current, even during the reverse scan, until the Pt oxides begin to form on the

electrode surface.

 

100 - - - - f v

 

    

Current (11A)
1'9
8

 

 

 

 

 

'300 ' v (V/s) ‘ ‘ l
0'05 Slope: \‘.\‘

.400 - 0.10 \ 400' ~64 mV/dec ‘. ' 1
0.20
0.30 330 ‘ * ‘ g‘ ‘#

-500 - - . - . - . - . ‘
0.40 14 “log (1006 02

-600 - 4 ‘

 

0.00 0.20 0.40 0.60 0.80 1.00 1.20

Potential (V vs. Ag/AgCl)
Figure 5.7. Linear potential sweep voltammetric i-E curves for RDE #1 in Oz-saturated
0.1 M HClO4 at different scan rates. Inset shows the plot of peak potential vs. logarithm

of scan rate.

142

The rate of ORR on oxide-free Pt surface is limited by the mass transport of
dissolved oxygen to the electrode surface. The dependence of the oxygen reduction on
the scan rate, v, was examined. Figure 5.7 shows the linear potential sweep voltammetric
i-E curves for a Pt/diamond composite electrode in Oz-saturated 0.1 M HClO4. The peak
current, ip, increases linearly with v”2 when the scan rate changes from 50 to 400 mV/s,
indicative of a semi-infinite linear diffusion controlled process. However, ip rather
approaches a proportionality with v at scan rates higher than 400 mV/s. This is expected
as the reaction shifts from being mass transport limited to control by the surface
adsorption process.5

Along with the increase in peak current, the peak potential, Ep, shifts positively
with a scan rate increase. The relationship between the peak potential and the scan rate

for an irreversible reaction is14
b b
EP=EW+§ 1.04-log—D~—210gkf+logv (5.5)

in which Em is the half-wave potential, b is the Tafel slope, D is the diffusion coefficient,
and kf is the rate constant. Therefore, the Tafel slope can be determined by plotting the
peak potential vs. the logarithm of scan rate, which is shown as the inset in Figure 5.7.
The estimated Tafel slope is —128 mV/dec, close to the value of —l20 mV/dec that is
expected for ORR on oxide—free Pt surface under Langmuirian conditions.15 It is also in

agreement with the results obtained from the RDE studies describe below.

5.4.3 ORR at PtlDiamond Composite RDEs

143

Polarization curves forvORR. The ORR at Pt/diamond composite RDEs was
initially studied by cyclic voltammetry. Figure 5.8 shows the cyclic voltammetric i-E
curve for a composite electrode (RDE #3). The response was recorded in Oz-saturated 0.1
M HClO4 using a rotating rate of 1000 rpm. For comparison, the voltammetric response
in Nz-purged electrolyte is also shown. Prior to each measurement, the electrode potential
was held at 1.2 V for 2 min., allowing for the formation of well-deﬁned Pt oxide film.
The potential scan was then initiated at this potential and scanned in the negative
direction. Similar to the voltammetric response in quiescent solution, drastic increase in

cathodic current was not observed until the reductive removal of the Pt oxide at a

 

2% V V ' V V I V V ' V V r V ‘Vi

 

Current (11A)

 

 

 

 

-800 ee44*+--‘--L--
-0.30 0.00 0.30 0.60 0.90 1.20

Potential (V vs. Ag/AgCl)

Figure 5.8. Cyclic voltammetric i-E curves for RDE #3 in (A) Nz-purged and (B) 02-
saturated 0.1 M HClO4. The dashed line is the curve for oxygen reduction after

background correction. Scan rate = 20 mV/s. Rotation rate = 1000 rpm.

144

potential of ca. 0.55 V. The oxygen reduction is then under combined kinetic and
diffusion control at potentials between 0.30 and 0.55 V. As the electrode potential
approaches the potential region where clean Pt is formed, a limiting current of ca. 0.7 mA
is observed, indicating the ORR is controlled by mass transport.

A notable feature in the i-E curve is that the ORR kinetics are different depending
on potential sweep direction. The onset potential for ORR is more negative by about 0.1
V during the forward sweep than during reverse sweep. This hysteresis is also a
characteristic feature of single crystal Pt,16 Pt supported on high-surface-area carbon,‘7

and a Pt electrode coated with Naﬁon®.18 The i-E curve recorded in the negative sweep

reﬂects the onset of OR on a partially oxidized Pt surface while during the reverse
sweep the ORR occurs on a surface that is essentially oxide-free. Recent studies have
suggested that the Pt oxides impede the ORR in two ways: oxides block the active Pt sites
and thus hinder the chemisorption of oxygen, and/or increase the adsorption energy of the
reaction intermediates which are formed during the OR on bare Pt sites.10

The current magnitude of the background i-E curve does not change with rotation
rate because the hydrogen adsorption/desorption, Pt oxidation/reduction are all surface-
controlled reactions. The well deﬁned shape of the voltammogram also indicates that the
electrolyte is pure, without mass-transfer controlled impurity adsorption process taking
place. The hydrogen adsorption is not signiﬁcantly affected by the presence of dissolved
oxygen. Rather, the adsorbed hydrogen suppresses the ORR, and even alters the reaction
pathway.16 Since the current recorded in Oz-saturated electrolyte is a sum of the oxygen

reduction current, the current due to Pt oxide formation and reduction, and double—layer

145

charging current, the i-E curve has been corrected for the background current, which is
shown as the dashed cuvee in Figure 5.8.

A series of background-corrected linear sweep voltammetric i-E curves recorded
at different rotation rates is shown in Figure 5.9. All the curves were recorded during the
cathodic potential sweep direction. Well-deﬁned, limiting current plateaus are seen,
particularly at low rotation rates. At high rotation rates, the potential region for combined
kinetic-diffusion control is markedly extended. For a ﬁrst-order reaction mechanism, the
limiting current is proportional to the concentration of dissolved oxygen and the square

root of the rotation rate, as predicted by the Levich equationS

 

 

 

 

 

 

 

 

 

100 - i f T - 1 - . - . - . - .
-100 - i
’ (ii/rpm ‘
-300 1' 1
2 ’ 500 ‘
3: $00? /
H
5 1000 l
t -700 1
3
U 1500 -
-900 2 001'—
.1100 250°
11.1
-1300 A A A m A A A A A A A A A A

 

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Potential (V vs. Ag/AgCl)

Figure 5.9. Linear sweep voltammetric i-E curves for RDE#3 in Oz-saturated 0.1 M

HClO4 as a function of the rotation rate. Scan rate = 20 mV/s.

146

iL = 0.62nFADZ’3v'“°a)‘/2C (5.6)
or in short form
iL = Ba)“2 (57)
where n is the number of electrons transferred per Oz molecule, F is the Faraday’s
constant, A the electrode surface area, C the concentration of dissolved Oz in solution, D
the diffusion coefﬁcient of dissolved 02, v the viscosity of the solution, and 0) the
rotation rate of the RDE. B is the so-called Levich slope. If the rotation rate is given in
revolutions per minute and C in moles per liter, B, given in mA- rpm'l’z, can be expressed
as
B = 0.20nFAD2’3v ""‘C (5.8)
For this study, the theoretical B value can be calculated for n = 4, A = 0.2 cm2, using the
literature data19 for v = 0.0087cm2 - s", and CD?"3 = 7.6 x 10 "0 mol- cm’m- s'm, yielding
a value of 2.5 x 10'2 mA- rpm'l’z.
As mentioned above, the ORR is under mixed kinetic-diffusion control over a
wide potential range. Extracting the kinetically-controlled current at different potentials

can be made using the Koutecky-Levich equations,

1 1 1 1 1 l
T=.—+.— 0’ T=.——+ 1,.
l '11" 1L l lkin B“) b

 

(5.9)

where i is the total current, iL is the limiting current, and im represents the current in the
absence of any mass transfer effects, which gives the order of absolute kinetic activity of

the surface. Plotting i'l vs. 00'”2 yields a y-axis intercept equal to ik~,,.'l and slope equal to

B.1

147

Koutecky-Levich plots for ORR on a Pt/diamond composite RDE in 0.1 M HClO4
are presented in Figure 5.10. These plots were constructed from the current values at
0.46, 0.40, 0.30, and 0 V in the polarization curves shown in Figure 5.9. All the plots are
straight lines with the same slope. Similar results were also obtained for other electrodes
and acid electrolytes. The y-intercept is 0.54 mA'1 for 0.46 V, 0.21 for 0.40 V, 0.07 for
0.30 V and 0.02 for O V, respectively. The corresponding slopes are 40.3 mA-l- rpmm,
40.8, 41.1 and 41.2, respectively. The B values, 2.43 - 2.48 x 10‘2 mA- rpm'm, are in

agreement with the theoretical value (2.5 x 10'2 mA- rpm'm), suggesting that the ORR on

the Pt/diamond composite electrode involves a 4-electron reduction pathway.

 

 

 

 

2.50 - . - . - . - . - ......r'
........ 15' .
E N ............ .....o:
2.00 - O 0.46 ...._.,...::g:::::.
D 0.40 9 ............... 1
A 0.30 .6 .- ... .........
L 1.50 . 0 0.0 a ............ w ":33; ..
E ....O a ....... 23:»
v ........... 3...“, ....
... 1.00 . ............. p 13’".
:_ ....-,.-" -°"'....:,..::§I.o
0'50 i .......... 3:221:23: °°°°°°
..... : :12: ---- J
0.00.. ..... k 4 ‘ ‘ ‘ ‘ - . ‘
0.00 0.01 0.02 0.03 0.04 005
(04/2 (rpm)'m

Figure 5.10. Koutechy-Levich plots for the oxygen reduction reaction in 0.1 M HClO4.

Data taken from the polarization curves shown in Figure 5.9.

148

Tafel plots. The kinetic parameters, the exchange current densities and Tafel

slopes, were obtained from mass-transfer—corrected Tafel plots (log i vs. 11). The
overpotential for the oxygen reduction, 11, was determined by subtracting the equilibrium

potential, Eeq, from the applied potential, Eapp. The kinetically-controlled currents can be
calculated using Equation 5.9, ikm = iLi/(iL-i). The current densities are normalized to the
2

real electrochemically active surface area of the composite electrode (0.80 cm, as

estimated from the roughness factor and geometric area),

 

 

 

 

 

-002 ' ‘ ' U ' ' ' I ' ' ‘ I ' ' ' t ﬁ ' ' v V r j
. 1
"mm—59 mV/dec
'0'3 . .."cg: N . l
-0.4 -
3 '05 ’ -125 mV/dec ‘
C
-0.6 r
.0.7 A A A A A A A A A A A A A A A A 4 A A o A A

-5.5 -4.5 -3.5 -2.5 -1.5 -0.5 0.5

log (iLi/(iL-i» (A-eni")
Figure 5.11. Tafel plots for the oxygen reduction reaction in 0.1 M HClO4. Data taken

from the polarization curves shown in Figure 5.9.

Figure 5.11 shows the set of Tafel plots constructed from the polarization curves

presented in Figure 5.9. The plots for different rotation rates overlap and two distinct

149

Tafel slopes are observed. The break in the Tafel plot, which occurs at an overpotential
between -0.45 and -0.50 V, has been previously reported and interpreted as a change in

20'” At low overpotentials, the Pt surface is

the oxygen chemisorption mechanism.
completely or partially covered with oxide film. A low Tafel slope of -60 mV/dec (-
2.3RT/F) is accounted for by assuming oxygen chemisorption under Temkin conditions
in the potential range of oxide formation. At high overpotentials, the coverage of
adsorbed oxygen-containing species becomes negligible, and a high Tafel slope of -120
mV/dec (-2 x 2.3RT/F) corresponds to a potential regime where oxygen reduction is
under Langmuirian conditions. The kinetic data from each region were independently
analyzed and the exchange current density corresponding to each Tafel slope was

calculated by extrapolating the potential to r] = 0, i.e. the equilibrium potential. As

illustrated in Table 5.3, a Tafel slope of —59 mV/dec at low overpotentials and —125
mV/dec at high overpotentials is seen for ORR in 0.1 M HClO4. The exchange current
densities, in, are 1.9 x 10''0 and 3.1 x 10'7 A-cm'z, respectively.

Effect of anion adsorption. There are several ways in which the electrolyte can
inﬂuence the ORR pathway and kinetics. These include changes in the concentration and
activity coefﬁcients of the reactants and intermediates, the transition states, solution pH,
and the anion adsorption. 4 A well—known phenomenon for smooth Pt surfaces is the
speciﬁc adsorption of anions, such as sulfate/bisulfate and phosphate/biphosphate, which
competes with the oxygen chemisorption and retards the reaction kinetics.22 Moreover,
Chemisorbed anions can alter the adsorption energy of the reaction intermediates and thus
the reaction pathway.”22 The effect of anion adsorption on the OR at Pt/diamond

composite electrode was examined. Polarization curves were also recorded in 0.1 M

150

HzSO4 and H3PO4. The corresponding Tafel plots are shown in Figure 5.12, and the
kinetics parameters are summarized in Table 5.3. The overpotential for ORR increases,
meaning that the electrode reaction kinetics decrease, in the order of HClO.i < H2SO4 <
H3PO4. At high overpotentials where the Pt surface is void of oxide, Tafel slopes of —132

and —136 mV/dec, and exchange current densities of 18 and 3.6 x 10'8 A-cm'2 are

observed in H2SO4 and H3PO4. At low overpotentials, the exchange current densities are
a factor of 30-90 lower in these two acids. Clearly, the speciﬁc adsorption of
sulfate/bisulfate and phosphate/biphosphate anions inhibits the reaction kinetics, but does
not affect the reaction pathway for ORR since similar Tafel slopes are observed for these

three acids. These results are consistent with reports in the literature. 23

 

.0.2 V V v I v v v I v V V v v v v I
-0.3 - «

-0.4 -

n(V)

-0.5 -

 

 

 

 

-5.5 -4.5 -3.5 -2.5 -1.5 -0.5

log (iLi/(iL-i» (A-cm'z)
Figure 5.12. Tafel plots for the oxygen reduction reaction at RDE #3 in 0.1 M (A)

HClO4, (B) H2804 and (C) H3PO4. Scan rate = 20 mV/s. Rotation rate = 2500 rpm.

151

Table 5.3. Comparison of the kinetics parameters for OR at the Pt/diamond composite

electrodes with the literature data reported for single crystal, polycrystalline, and carbon-

supported Pt at room temperature.

 

 

Electrocatalyst Electrolyte Tafel slope in Reference
(-mV/-dec) (A-cm'z)

Pt(l 1 1) 1 M st04 64.7 2.2 x 10'10 23

Pt(100) 1 M st04 64.5 4.6 x 10‘10 23

Pt(110) 1 M H2804 64.6 3.4 x 10"" 23

Pt 0.1 M HClO4 6O ~1 x 10"” 1

Pt 0.1 M st04 70 2 x 10'“ 24

Pt 98% H3PO4 99 7.76 x 10'l2 25

Pt 0.05 M 112504 145 3 x 10*3 26

Pt/C 1.0 M NaOH 55 1 x 10'8 27

Pt/C 100% H3P04 90 (3-7) x 10'8 28

Pt/diamond 0.1 M HC104 59 1.9 x 10‘10 This work
125 3.1 x 10‘7

Pt/diamond 0.1 M st04 62 6.3 x 10'H This work
132 1.8 x 10'7

Pt/diamond 0.1 M H3PO4 63 2.2 x 10'” This work
136 3.2 x 10'8

 

A comparison of the kinetic parameters for ORR at the Pt/diamond composite
electrodes with the literature data reported for single crystal, polycrystalline, and carbon-
supported Pt is presented in Table 5.3."15‘24'28 The Tafel slopes and exchange current
densities for the Pt/diamond composite electrodes in these acid electrolytes are

comparable to those reported for polycrystalline Pt. The results indicate that the Pt

152

electrocatalyst is not contaminated with the carbon residual from the secondary diamond
growth. In addition, the possible contact resistance between the catalyst and the
supporting material is negligible. In other words, the diamond matrix exhibits metal-like

electrical conductivity as both the support material and the current collector.

 

n(V)

-0.5 '

1

-0.6

 

 

 

 

.0.7 A A A A A A A A A A A 4 A A A J A A A A A A A
-5.5 -4.5 -3.5 -2.5 -1.5 -0.5 0.5

log (iii/(inn) (A-cm")
Figure 5.13. Tafel plots for the oxygen reduction reaction in 0.1 M HClO4 at different

Pt/diamond RDEs. Scan rate = 20 mV/s. Rotation rate = 2500 rpm.

Effect of Pt loading. The effect of Pt loading on the ORR at the Pt/diamond
composite electrodes was also examined in the three acid electrolytes. Polarization curves
for the three composite RDEs (see Table 5.1) were recorded in 0.1 M H00; and the
corresponding Tafel plots are shown in Figure 5.13. The kinetically—controlled current

densities are normalized to the electrochemically active surface area of the RDEs (0.48,

153

0.68 and 0.80 cmz, respectively). The plots overlap at both low and high overpotentials,
meaning that there is no significant inﬂuence of Pt loading on the reaction kinetics.

There are two additional parameters that reﬂect the electrocatalyst activity for the
ORR; speciﬁc activity (SA) and mass activity (MA). These terms are deﬁned by the
following relationships3

SA(A/cm2 Pt) = , mm" (5.10)
active surface area of Pt

 

and

MA(A/ gPt) = M (5.11)
mass of Pt

The specific activity provides a measure of the electrocatalytic activity of Pt atoms in the
particle surface. The mass activity has more practical implications because the cost of the
electrode depends on the amount of platinum used. The relationship between the SA and
MA can be expressed as

MA=SAXS (5.12)
where S is the specific surface area deﬁned as the real surface area of unit mass of Pt
(cmzlg).

Therefore, the Tafel plots in Figure 5.12 and 5.13 essentially reﬂect the
dependence of the SA of the Pt/diamond electrocatalyst on the electrolyte type and
catalyst loading because all the currents are normalized to the real surface area of Pt. It is
a common practice to compare the SA and MA for ORR by measuring the current at 0.9
V vs. SHE (corresponding to an overpotential of ca. —0.33 V), where inﬂuences of the
mass transport are negligible.7'29 The SA and MA of the Pt/diamond composite electrodes

in the three acid electrolytes are listed in Table 5.4, along with the literature data reported

154

for other carbon-supported Pt electrocatalysts.7’29'32 Both SA and MA decrease in the

order of HClO4 > H2804 > H3PO4_ The SA in 0.1 M HClO4 is ca. 55 uA/cmz, about two

orders of magnitude larger than in H3PO4_

Table 5.4. Comparison of the specific activity (SA) and mass activity (MA) of the

Pt/diamond composite electrode for ORR with the literature data reported for other

carbon-supported Pt electrodes.

 

 

Electrocatalyst Temperature Electrolyte SA“ MAa Reference
(“0 (uA-cm‘h (A-g“)

Pt/C 70 20% HZSO4 0.4-14 0.1-9.1 30
Pt/graphite 25 1 M HZSO4 40-70 4-25 29
Pt/Vulcan 50 1 M H2S04 20 — 32
Pt/Vulcan 60 0.5 M HClO4 65 — 7
Pt/Vulcan 60 0.5M H2SO4 35 — 7
Pt/diamondb 25 0.1 M H00; 55 2.2 This work
Pt/diamond 25 0.1 M H2804 19.3 0.77 This work
Pt/diamond 25 0.1 M H3PO4 0.72 0.03 This work

 

a. Activities measured at 0.9 V vs. SHE (overpotential = — 0.33 V).

b. Composite electrode with a Pt deposition time of 400 s.

As seen in Figure 5.13, the SA of the Pt/diamond composite electrode for OR is

independent on the Pt loading. The value, 55 i 6 uAtch, is rather comparable to those

reported for carbon-supported Pt. The MA of the electrode slightly decreases from 2.6 to

2.2 A-g'l while the apparent Pt loading increases from 50.3 to 101.1 g-cm'z. These results

are consistent with those reported in the literature and can be interpreted in terms of

155

 

structure sensitivity of the ORR. Recent studies of ORR on low-index Pt single-crystal
surfaces indicated that the reaction rate varies with crystallographic orientation. In HClO4
the specific activity increases in the order Pt(100) < Pt(110) z Pt(111). Since small Pt
particles are generally treated as an assembly of the low index facets, the OR on the
particles may resemble the behavior of Pt single crystals and the overall activity varies
with varying the relative ratio of the low index facets exposed on the particle surface. The
XRD measurements reveal that the Pt particles dispersed in the diamond matrix have a
predominant fraction of (111) facets. Furthermore, the relative ratio of the three low
index facets remains nearly constant while the Pt loading increased from 50.3 to 101.1
gem2 with Pt particle size ranging from 40 to 500 nm. Therefore, the observed specific
activity is expected to be independent on the Pt loading. This is consistent with the
studies by Zeliger”, Bett et a134 , Kunz and Gruver”, and Vogel and Baris36, which show
that the speciﬁc activity of Pt particles larger than 3 nm for oxygen reduction is
independent of the particle size and even approximately the same as smooth Pt. However,
the MA decreases with Pt loading because the particle size slightly increases, thus the

speciﬁc surface area decreases, with increased Pt loading.

5.5 Conclusions

Electrochemical measurements of the oxygen reduction kinetics at the Pt/diamond
composite electrodes were presented. The results can be summarized as follows:
1. The mechanism of oxygen reduction at the Pt/diamond composite electrode

follows that proposed for polycrystalline and carbon-supported Pt catalysts.

156

Two distinct Tafel slopes were obtained, with the value consistent with —
2.3RT/F and —2><2.3RT/F, respectively.

. The exchange current density for the ORR in the fuel cell operative potential
region (low overpotential region) is ca. 10'10 A-cm'2 in 0.1 M HClOa, two
order of magnitude higher than in 0.1 M H3PO4.

. The speciﬁc activity of the composite electrodes is 55 i 6 uA-cm'2 at 0.9 V
vs. SHE in 0.1 M HClO4, comparable to those reported for carbon-supported
Pt. The effect of the anion adsorption on the specific activity and mass activity
are significant. But the specific activity is independent on the Pt loading.

. The comparable catalytic activity of the compOsite electrode to the smooth Pt
indicates that contamination of the Pt particles by the carbon residues from the
secondary diamond growth is negligible. The diamond ﬁlm exhibits metal-like
electrical conductivity as both a support material and current collector.

. Further work will focus on the optimization of diamond growth conditions
and catalyst dispersion procedure in order to reduce the particles size and

improve the reactivity of this new type of catalyst.

157

References:

10.

11.

12.

13.

14.

15.

16.

17.

A. Damjanovic and V. Brusic, Electrochim. Acta, 12, 1171 (1967).
E. Yeager, Electrochim. Acta, 29, 1527 (1984).

K. Kinoshita, Electrochemical Oxygen Technology, John Wiley & Sons, Inc.,
(1992).

R. Adzic, in Electrocatalysis, J. Lipkowski and P. N. Ross, Editors, Wiley-VCH,
New York, 197 (1998).

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

A. Parthasarathy, S. Srinivasan, A. J. Appleby, and C. R. Martin, J. Electroanal.
Chem, 339, 101 (1992).

U. A. Paulus, T. J. Schmidt, H. A. Gasteiger, and R. J. Behm, J. Electroanal.
Chem, 495, 134 (2001).

A. J. Appleby, Corrosion, 43, 398 (1987).

V. Stamenkovic, N. M. Markovic, and P. N. Ross, J. Electroanal. Chem, 500, 44
(2001).

N. M. Markovic, H. A. Gasteiger, B. N. Grgur, and P. N. Ross, J. Electroanal.
Chem, 467, 157 (1999).

Y. Kaenel, J. Stieger, J. Michler, and E. Blank, J. Appl. Phys., 61, 1726 (1997).

A. Parthasarathy, S. Srinivasan, and A. J. Appleby, J. Electrochem. Soc., 139,
2530 (1992).

P. N. Ross and P. C. Andricacos, J. Electroanal. Chem, 154, 205 (1983).
E. Gileadi, Electrode Kinetics, VCH, New York, 173 (1993).

A. Damjanovic, M. A. Genshaw, and J. O. M. Bockris, J. Electrochem. Soc., 114,
466 (1967).

N. M. Markovic, H. A. Gasteiger, and P. N. Ross, Jr., J. Phys. Chem, 99, 3411
(1995).

S. L. Gojkovic, S. K. Zecevic, and R. F. Savinell, J. Electrochem. Soc., 145, 3713
(1998).

158

 

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

S. L. Gojkovic, S. K. Zecevic, and R. F. Savinell, J. Electrochem. Soc., 145, 3713
(1998).

N. M. Markovioc, R. R. Adzic, B. D. Cahan, and E. B. Yeager, J. Electroanal.
Chem, 377, 249 (1994).

A. Damjanovic and V. Brusic, Electrochim. Acta, 12, 615 (1967).
A. Damjanovic and M. A. Genshaw, Eletrochim. Acta, 12, 1281 (1970).
P. N. Ross and P. C. Andricacos, J. Electroanal. Chem, 154, 205 (1983).

N. Markovic, H. Gasteiger, and P. N. Ross, J. Electrochem. Soc., 144, 1591
(1997).

F. E. Kadiri, R. Faure, and R. Durand, J. Electroanal. Chem, 177 (1991).

M. A. Enayetullah, T. D. Devilbiss, and J. O. M. Bockris, J. Electrochem. Soc.,
136, 3369 (1989).

P. Fischer and J. Heitbaum, J. Electroanal. Chem, 112, 231 (1980).

J. Perez, A. Tanaka, E. R. Gonzalez, and E. A. Ticianelli, J. Electrochem. Soc.,
141, 431 (1994).

H. R. Kunz and G. A. Gruver, Electrochim. Acta, 23, 219 (1978).
J. A. Poirier and G. E. Stoner, J. Electrochem. Soc., 141, 425 (1994).

K. F. Blurton, P. Greenburg, H. G. Oswin, and D. R. Rutt, J. Electrochem. Soc.,
119, 559 (1972).

J. A. Poirier and G. E. Stoner, J. Electrochem. Soc., 142, 1127 (1995).
J. Bett and J. Lundquist, Electrochim. Acta, 18, 343 (1973).
H. Zeliger, J. Electrochem. Soc., 114, 144 (1967).

J. Bett, J. Lundquist, E. Washington, and P. Stonehart, Electrochim. Acta, 18, 343
(1973).

H. R. Kunz and G. A. Gruver, J. Electrochem. Soc., 122, 1279 (1975).

W. M. Vogal and J. M. Baris, Electrochim. Acta, 22, 1259 (1977).

159

ﬁ-

Chapter 6

Methanol Oxidation Reaction on Metal/
Diamond Composite Electrodes

6.1 Introduction

During the past three decades, the methanol oxidation reaction (MOR) has been
investigated intensively because of its importance as a feedstock in the direct methanol
fuel cell (DMFC)."5 There has recently been a resurgence in DMFC R&D in view of it’s
possible use in transportation applications, i.e. electric vehicles.6 Methanol is one type of
liquid fuel offering several advantages over others with respect to the storage, safe
handling and rapid refueling. Its use eliminates the need for a bulky, heavy, and complex
fuel processor, which is estimated to have about the same weight and volume as the fuel
cell stack. Methanol also has a high energy, gravimetrically and volumetrically. One mL
of methanol can deliver about 2.5 Wh of electricity even at 50% efficiency.

There are, however, at least two research challenges that need to be solved in
order for the large-scale application of DMFCs to emerge: the sluggish oxidation reaction
kinetics at the anode (methanol) and high catalyst cost. Pt-based materials are, to date, the
only acceptable catalysts for methanol oxidation. Dissociative chemisorption of methanol
can proceed quite rapidly on the platinum surface, particularly at potentials about 0.2 V
vs. RHE, forming Chemisorbed species such as PtCOH and PtCO. The further conversion
of these adsorbates to form the final product, COz, requires attack on the carbon by
oxygen-containing species that are generated from the dissociative chemisorption of

water.7 However, this process is rather slow at electrode potentials below ca. 0.4 V vs.

160

 

RHE since pure Pt does not efficiently adsorb water over this potential range.7
Consequently, the accumulation of CO and other organic residues takes place, which can
block the reactive sites and thus plague the overall reaction. In order to overcome the
poisoning problem, many efforts have been made to develop binary or ternary catalysts of
Pt alloyed with more oxophilic elements, such as Ru8’9, Sn”), and Mo”, etc. These
promoters for methanol oxidation are believed to act primarily either by providing
“active” oxygen adducts or by increasing the susceptibility of the carbon-containing
intermediates to oxygen-donor attack.

Considering the catalyst cost, noble metals are essentially utilized as nanometer-
sized particles dispersed onto a porous supporting material, in order to achieve the
highest possible speciﬁc surface area (i.e., surface area to mass ratio). The small particle
size can also modify the catalytic activity.12 Some research has been undertaken to
examine possible effects of the particle size, but the findings appear to be controversial.
Yahikozawa et a1. studied Pt deposited on glassy carbon surface by vacuum evaporation
and found that the specific activity for methanol electrooxidation increased with
increasing Pt particle size from 3.8 to 5.3 nm, approaching the speciﬁc activity of smooth
Pt.l3 Kabbabi et al. used Pt supported on Vulcan carbon (E-TEK) with the average Pt
particle size from 1.8 to 25 nm. They reported a sharp increase in speciﬁc activity with
increasing Pt particle size from 1.8 to 6 nm, followed by a nearly constant specific
activity up to 25 nm.'4 Gojkovic et al. reported a maximum specific activity for a Pt
particle size of 3.5 nm supported on Black Pearls carbon (E-TEK).15 However, Watanabe
et a1. observed no particle size effect, even for a particle size as small as 1.4 nm.

Therefore, they concluded that increasing the platinum dispersion on carbon is an

161

interesting possibility for achieving higher catalytic activity with low amounts of
platinum.”

In Chapter 5, it was shown that the dimensionally stable Pt/diamond composite
electrodes exhibit comparable electrocatalytic activity to polycrystalline Pt for the ORR
—— the cathode reaction in DMFCs,. It is of immediate interest to probe the catalytic
activity of the composite electrodes toward MOR — the anode reaction in DMFCs. In
this chapter, results from studies of MOR at metal/diamond composite electrodes are
presented, speciﬁcally data from cyclic voltammetric and Chronoamperometric
measurements. The purpose of the work was to compare the reaction kinetics of Pt
supported on diamond with Pt supported on other spz-bonded carbon supports. Efforts to
prepare the Pt/Ru/diamond composites were also made. Preliminary results presented
below indicate that the electrocatalytic activity for MOR was greatly enhanced by the

bimetallic catalyst.

6.2 Mechanism and Kinetics of MOR

Mechanism. Beden et al. proposed a complete and widely accepted mechanism
for the electrooxidation of methanol at Pt in acidic media.5 The main reaction pathway is
the formation of C02 via CO as an intermediate, which includes the adsorption of
methanol, dehydrogenation of adsorbed methanol, formation of PtCO, and reaction of
PtCO with surface bonded hydroxyl radicals, PtOH. The chemical reactions for these
steps are written as follows:

The ﬁrst step of the electrooxidation reaction is the dissociative chemisorption of

methanol, leading to several partially hydrogenated species and CO.

162

(CH3OH)S°1+ Pt—> Pt-(CH3OH)2..1s (6.1)

Pt-(CH3OH)ads —-) Pt-(CH20H)ads + H+ + e' (6.2)
Pt-(CHzOHLms —) Pt-(CHOH),.ds + H+ + e' (6.3)
Pt-(CHOH)ads —) Pt-(CHO)ads + H+ + e‘ (6.4)

After the Reaction 6.4, the formyl-like species is spontaneously dissociated on
platinum according to the reaction:
Pt-(CHO)ads —) Pt-(CO)ads + H+ + e' (6.5)

Oxidation of adsorbed intermediates requires the dissociation of water to form

OH species:
Pt + H2O ——> Pt-(OH)ads + H+ + e' (6.6)
Pt-(CHO)Ms + Pt-(OH),,ds ———) 2Pt + CO2 + 2H+ + 2e' (6.7)
or Pt-(CHO)Ms + Pt-(OH)ads ———> Pt + Pt-(COOH)ads + 11" + e' (6.8)
Pt-(COOH).ds ———> Pt + CO2 + H+ + e' (6.9)

Adsorbed CO can be oxidized through the reactions:

ll’t-(CO),....s + Pt-(OH),,ds ——> 2Pt + CO2 + H+ + e' (6.10)

or Pt-(CO),,ds + Pt-(OH),,ds ——9 Pt + Pt-(COOHMs (6.11)

This mechanism takes into account the formation of all the products: CO2,
formaldehyde, and formic acid.

At a bimetallic catalyst, the adsorbed OHMs is formed at lower potentials at the
promoter site:

M + H2O ————> M-(OH),¢, + H+ + e' (6.12)

Then, the oxidation reaction can take place

163

Pt-(CHO).ds + M-(OH),,,1s —> Pt + M + CO2 + 2H+ + 2e' (6.13)
Adsorbed CO can be oxidized at lower potentials through the reactions:
Pt-(CO),ds + M-(OH)ads ——> Pt -1- M + CO2 + H" + e’ (6.14)

or Pt-(CO),ds + M-(()H)..ds ———) M + Pt-(COOHMds (6.15)

Kinetics. The rate-determining step (rds) of MOR on platinum is the oxidation of
adsorbed COMS with adsorbed hydroxyl species (Reaction 6.10).5 The current density for

the reaction can be obtained from the following equation

on FE
i = anO 0 ex "" 6.16
res 0H p[ RT ] ( )

 

where 0,6, and 00H are the coverages of adsorbed methanol residues and hydroxyl
groups, n and n,ds are the number of electrons involved in the overall reaction and in the
rate—determining step, k is the rate constant, F the Faraday constant, a the transfer

coefficient, and E the electrode potential. Clearly, the oxidation rate at constant potential
depends on the surface concentration of the adsorbed species, rather than the bulk
concentration. As proposed by Bagotzky and Vassilyev, the adsorption of the reaction

intermediates is governed by a Temkin-type isotherm: '7

0,“. = const + -—1-- In C (6.17)

where f is the inhomogeneity factor, and c is the bulk concentration of methanol.

164

 

6.3 Experimental

Electrode preparation. The composite electrodes were fabricated via the same
stepwise procedure described in Chapter 4, except for the co-deposition of Pt and Ru in
the case of the bimetallic catalyst preparation. The co-deposition of Pt and Ru was carried
out in N2-purged 0.1 M HClO4 containing 1 mM K2PtCl6 and RuCl3, with concentrations
ranging from 0.2 to 1mM. Deposition was performed galvanostatically at 0.5 mA/cm2
and the deposition time ranged from 100—700 s. A polycrystalline Pt foil electrode was.
used for comparison measurements. This Pt surface was pretreated by polishing in
successively small alumina/water slurries (particle size 1.0, 0.3, 0.05 11m), and cleaning
by ultrasonication in ultrapure water for 5 min after each polishing step. The active
surface area and roughness factor for all these electrodes, determined from the hydrogen
adsorption/desorption charge, are presented in Table 6.1.

Electrode cleaning procedure. A clean Pt surface is necessary in order to obtain
valid and reproducible results for the MOR. Two cleaning procedures were employed
according to the type of experiment and solution. In pure electrolyte, a potential cycling
procedure was employed. The electrode was subjected to 50 potential cycles between —
0.4 and 1.5 mV vs. Ag/AgCl at a sweep rate of 50 mV/s. The final electrode potential
was ﬁxed at 0.2 V vs. Ag/AgCl. In solutions containing methanol, a pulse cleaning
procedure was employed for complete oxidative removal of Chemisorbed methanol
residues. The electrode was subjected to the following potential pulse sequences: 2 s at —
0.3 V vs. Ag/AgCl, 2 s at 1.4 V, 2 s at —0.3 V, 2 s at 1.4 V, and 10 ms at —0.1 V. After

each potential step or sweep measurement, in which methanol was oxidized, the pulse

sequence given above was reapplied.

165

 

Table 6.1. Electrochemically active surface area and roughness factor for the

polycrystalline Pt and Pt/diamond composite electrodes with different Pt loadings.

 

Electrode Pt deposition Hydrogen adsorption/ Roughness Active

 

time desorption charge factor surface
(s) (mC/cmz) area
(cmZ)
E1 100 0.25 1.2 0.24
E2 200 0.46 2.2 0.44
E3 300 0.63 3.0 0.60
Pt foil — 0.40 1.9 0.38

 

Instruments and chemicals. All the electrochemical measurements were
performed with a CHI 650A digital potentiostat (CH Instruments, Austin, TX). All the
potentials were recorded vs. a commercial Ag/AgCl reference electrode (E0 = 0.197 V vs.
SHE). All solutions were prepared with ultrapure water from a Bamstead E-Pure
puriﬁcation system (>17 MQ-cm). The HC104, H2S04, and H3PO4 were ultrahigh purity

grade (99.999% Aldrich) and the K2PtC16, RuCl3 (Aldrich) and CH3OH (Fisher

Scientific) were reagent grade quality. All glassware was cleaned in a KOH/methanol

bath prior to use.

6.4 Results and Discussion

6.4.1 Measurement of the Open Circuit Potential

166

Table 6.2. Open circuit potentials for Pt/diamond composite electrodes in 0.1 M HC104

with different methanol concentrations.

 

 

Electrode Methanol Concentration
(M)
0.2 0.6 1.0 1.5 2.0
El 0.240 0.221 0.210 0.182 0.169
E2 0.232 0.212 0.200 0.179 0.168
E3 0.230 0.200 0.193 0.178 0.170

 

Note: Values are in V vs. Ag/AgCl .

Open circuit potential (OCP) measurements are often used as an indicator of the
electrocatalytic activity of an electrode. The OCPs for the Pt/diamond composite
electrodes with different platinum loadings, were measured in N2-purged 0.1 M HClO4 as
a function of the CH3OH concentration. Prior to each measurement, the electrodes were
subjected to the potential cycling cleaning procedure, followed by exposure to CH3OH
solution for 10 min. The OCP slowly increased (more positive) during the initial stage of
the measurement before reaching a steady state after about 5 nrin. Steady-state OCP
values are presented in Table 6.2. The OCP is, on average, about 0.20 V vs. Ag/AgCl,
indicating a ca. 0.4 V positive shift with respect to the equilibrium potential of methanol
oxidation (-0.180 V vs. Ag/AgCl).6 The gradual positive shift in OCP is presumably due
to the dissociative chemisorption of CH30H to form CO. The OCP decreases with the
increase in the CH3OH concentration, consistent with the trend predicted by the Nemst

equaﬁon.

167

 

6.4.2 MOR at PtlDiamond Composite Electrodes

6.4.2.1 Cyclic Voltammetric Studies of MOR

The MOR kinetics were initially studied by cyclic voltammetry. Figure 6.1A
(solid line) presents a typical cyclic voltammetric i-E curve for MOR at a Pt/diamond
composite electrode. For comparison, the voltammetric response for a polycrystalline Pt
foil electrode is shown in Figure 6.1B. The experiments were carried out at room
temperature in N2-purged 0.1 M HClO4 containing 0.2 M CH3OH. The potential sweep
was initiated at 0 mV and scanned in the positive direction at a scan rate of 50 mV/s. The
background i-E curves for both electrodes are shown as dashed lines.

The response for the Pt/diamond composite electrode resembles that for the clean
Pt foil. The voltammogram displays peaks that are also characteristic of MOR on
supported Pt catalysts in acidic solution.15 During the forward or anodic sweep,
discernible faradaic current begins at ca. 0.25 V, corresponding to the dehydrogenation
reactions (Reaction 6.1—4). The current is low due to surface poisoning by the rapid
formation of Chemisorbed organic residues, mainly CO. As the applied potential
becomes more positive, dissociative chemisorption of water takes place (Reaction 6.6).
Accordingly, there is a sharp rise in the current at E > 0.4 V, which is attributed to the
oxidative removal of the adsorbed organic residues (Reaction 6.7-10). An oxidation peak

X

is observed at 0.65 V with a peak current, iplo , of 420 11A. The oxidation charge is
greater than that predicted for the oxidation of a monolayer of adsorbed CO, so in

addition to the oxidation of adsorbed CO, oxidation of additional CH3OH molecules is

occurring.

168

 

400 - 1

 

 

 

-0.30 0.00 0.30 0.60 0.90 1.20 1.50

 

Current (11A)

400} B
300;
200b
100: 4!

0 .-

 

-100

 

 

 

-200 A A A 4A A A A A A A A A A A A A A
-0.30 0.00 0.30 0.60 0.90 1.20 1.50

Potential (V vs. Ag/AgCl)

Figure 6.1. Cyclic voltammetric i-E curves for (A) Pt/diamond composite electrode and
(B) polycrystalline Pt foil electrode in 0.1 M HClO4 + 0.2 M CH30H (solid line) and 0.1
M HClO4 (dashed line). Scan rate = 50 mV/s. Active surface area of the composite

electrode is 0.44 cm2 (Pt deposition time was 200 s).

169

The decline in current above 0.65 V reﬂects the inhibition of the MOR by surface oxide
formation at these potentials. However, some catalytically active surface oxides are
formed again at potentials greater than 0.85 V, accounting for the second anodic peak
near 1.2 V (Reaction 6.10). The oxygen evolution reaction takes place at potentials
greater than 1.3 V. The surface is inactive toward CH3OH oxidation on the reverse
sweep until the reduction of the surface oxides at 0.55 V (i.e., formation of bare Pt). An

oxidation peak is then observed at ca. 0.42 V with a peak current, ip2°", of 225 11A. At

potentials below 0.2 V, the active surface sites become blocked by CO formation from
the dehydrogenation of Chemisorbed CH3OH. Consequently, hydrogen adsorption/
desorption is significantly suppressed.

The specific activity (SA) was determined by normalizing the current to the
electrochemically active surface area. At 0.6 V vs. Ag/AgCl, the speciﬁc activity for the
composite electrode (0.75 mA-cm’z) is smaller than that for polycrystalline Pt (1.03
mA-cm'z). One possible explanation for this difference is that the activity is structure
sensitive. Recent work with single crystal Pt surfaces has shown that, in acidic media, the
rate of CH3OH oxidation increases in the order (111) < (100) << (110).18 The
polycrystalline Pt and Pt particle surface can be treated as an ensemble of these low-index
facets; therefore, their activity is dependent on the fraction of these facets exposed. XRD
measurements indicated that the (110) orientation is dominant on the polycrystalline Pt
foil, whereas the (111) and (100) orientations are exposed on the Pt particles. The latter
two crystallographic orientations are less active than the (111) orientation for the MOR.

Effect of Pt loading. The possible effect of platinum loading on the MOR

kinetics was studied by cyclic voltammetry. Figure 6.2 presents the cyclic voltammetric

170

i-E curves recorded for the first potential scan in N2-purged 0.1 M HClO4 containing 0.6
M CH3OH. The voltammetric data are summarized in Table 6.3. As seen in Table 6.1, the
electrochemically active surface area of the composite electrodes increases with
increasing the apparent Pt loading (i.e., increasing deposition time). Accordingly, the
peak currents, in both positive and negative scan, follow the same trend. It was also
observed that the peak potentials shifted positively with increasing Pt loading due to the
increased surface coverage of reactants.

The comparison of the specific activity was made by normalizing the currents
obtained at 0.4 V and 0.6 V during the positive scan, with respect to the active Pt surface
area. There is no signiﬁcant change in the specific activity with platinum loading. As
discussed earlier, methanol oxidation is a structure-sensitive reaction. The unchanging
speciﬁc activity implies that the surface structure (crystallographic orientation) of the Pt
particles, dispersed onto the diamond, does not change with loading. XRD measurements
conﬁrmed that the relative ratio of the low index facets remains nearly constant with
varying the Pt loading. For the same reason, the Pt loading also does not considerably
inﬂuence the onset potential for methanol oxidation.

Effect of anion adsorption. The effect of anion adsorption on the MOR kinetics
was examined in 0.1 M HClO4, H2SO4 and H3PO4. Figure 6.3 presents the cyclic
voltammetric i-E curves for the oxidation of 0.6 M methanol in these media. The main
features of the three voltamograms are very similar in potential. However, the anion
effect on the peak current is clearly evident. The peak current during the anodic potential

scan is ca. 0.68 mA in HClO4, which is approximately three times higher than in H2SO4

171

 

1200

 

 

 

 

 

 

 

1°00- “““ E‘ r 1
D ooooooooooooooooooo E2
r 1
800 - E3 1 .
A D
a 600 b. / ”I “1
V ,1 Il ‘.
H 1 I | I .
= j I, ’ l 1
3 400 " " "
3.. ' I 1 '1
= ’ Il ‘ 1
I I I ‘
O , ’I ll \ “ $1 ,
200 - ’l’ ’I l \ ‘s o ’l
. I’l/ I” /\ \ ~“' I'I”
,x” ."l’ \ \ . ‘3’
O 1- . A 9 \ ----- v
-200 A A 1 A A 1 A A A A A A A A 4 1
.0020 0000 0020 0040 0060

0.80 1.00 1.20

Potential (V vs. Ag/AgCl)

Figure 6.2. Cyclic voltammetric i-E curves for Pt/diamond composite electrodes in 0.1 M

HClO4 + 0.6 M CH3OH. Scan rate = 50 mV/s.

Table 6.3. Cyclic voltammetric data for the MOR in 0.1 M HClO4 at Pt/diamond

composite electrodes with different Pt loadings.

 

 

Electrode Eonset Eplo" ipio" Ep2°x ip2o" SA SA
(V) (V) (mA) (V) (mA) at Ea = 0.4V at E, = 0.6V
(mA-cm’z) (mA-cm'z)
E1 0.23 0.66 0.33 0.52 0.24 0.09 1.05
E2 0.21 0.71 0.68 0.56 0.58 0.08 0.97
E3 0.22 0.74 1.09 0.57 0.98 0.09 0.97

 

172

and four times than in H3PO4, The current reduction is due to the fact that
sulfate/bisulfate, and phosphate ions are more strongly Chemisorbed than are perchlorate
ions, and thus, compete with methanol molecules for surface sites leading to a lower

oxidation current.

 

 

 

 

 

800 ~ - . . - . - . ff - . . . , . . , , T
600 -
3 400» .
E .
Q)
g .
U 200 ’ A
0 .
-200 ‘ ‘ + l A L g+ 1 A A 1 A

-0.20 0.00 0.20 0.40 0.60 0.80 1.00 1.20

Potential (V vs. Ag/AgCl)

Figure 6.3 Cyclic voltammetric i-E curves for the electrooxidation of 0.6 M CH3OH in
different 0.1 M acid electrolytes at a Pt/diamond composite electrode. Active surface area
of the composite electrode is 0.44 cm2 (Pt deposition time was 200 3). Scan rate = 50

mV/s.

173

 

Another interesting feature is observed at low potentials, indicative of a quite
different anion effect. At potential less than 0.5 V during the anodic sweep, the highest
oxidation current is seen for H3PO4. Since only Reactions 6.1-4 take place over this
potential range, the observation may suggest that phosphate anion adsorption somehow
hinders CO poisoning, which allows for higher rate of methanol
adsorption/dehydrogenation. A similar observation was made by Herrero et al. during the
study of methanol oxidation on low index Pt surfaces.18 At low potentials,
sulfate/bisulfate adsorption inhibits the methanol oxidation as usual, probably resulting
from the enhanced CO formation due to strong lateral CO-bisulfate interaction. This

effect was recently conﬁrmed by Ogasawara et al.19

6.4.2.2 Chronoamperometric Studies of MOR

Cyclic voltammetry is useful for studying the overall reaction kinetics of
electrocatalytic reactions, but steady-state Chronoamperometric methods offer a more
straightforward means to extract detailed mechanistic information about the methanol
oxidation process. Two types of Chronoamperometric measurements were performed.
Analyzing the current decays allows for the Tafel slope, reaction order, and rate—
determining step to be determined.

Effect of electrode potential. The current-time transients recorded at different
electrode potentials for a Pt/diamond composite electrode immersed in Nz-purged 0.6 M
CH3OH + 0.1 M HClO4 are presented in Figure 6.4. Prior to each measurement, the
electrode was conditioned by the potential pulse cleaning procedure (see experimental
section). The potential step was applied from 0 V to the electrode potentials ranging from

0.45 to 0.70 V (refer to Fig. 7.4). The current rises initially due to a combination of

174

 

double-layer charging and methanol/adsorbed CO oxidation at a clean electrode surface.
The current then decreases as the double layer fully charges and as some sites become

blocked by adsorbed organic residues. Possible adsorbates are COH and CO which come

 

 

 

 

 

 

 

 

 

 

 

500 ~ . . - - . - - . - - . f -
\ 1
400 -
a 300 J
v l E (V vs. Ag/AgCl)
H 4
g 200 . .
1.. 1 Ar 0.65 1
5 1 0.70 1
100 : 0.60 ‘
0.55 ‘
’ t 0.50 ‘
0 f 0.45 f
' 1
.100 A L l A A l A A j A A 1 A A
0.0 5.0 10.0 15.0 20.0 25.0

Time (min.)
Figure 6.4. Current-time transients for a Pt/diamond composite electrode in 0.1 M HClO4
+ 0.6 M CH3OH at different electrode potentials. Active surface area of the composite

electrode is 0.60 cmz.

from the serial reaction pathway. The current-eventually reaches a near steady-state value
that depends on the electrode potential. At 0.45 V, the steady-state current at 20 min. is

ca. 4 11A. Increasing the electrode potential causes a rise in the reaction rate, consistent

with the cyclic voltammetric results. The steady-state current at 0.65 V is ca. 120 11A,

175

approximately 30 times higher than at 0.45 V. However, further increase in the electrode
potential (e.g. 0.7 V) results in a drop in the current, corresponding to the formation of
reaction-inhibiting surface oxides.

Effect of methanol concentration. The dependence of the oxidation current on
the methanol concentration was investigated by the potential-step method. Figure 6.5
shows a set of current-time transients obtained after the electrode potential was stepped
from 0 to 0.6 V vs. Ag/AgCl for a Pt/diamond composite electrode in Nz-purged 0.1 M
HC104 with methanol concentrations from 0.2 M to 2.0 M. The total oxidation charge
and steady-state current density, obtained at 20 min. are listed Table 6.4. The rate of
methanol electrooxidation increases with increasing methanol concentration. Evidently,
the dependence of the electrooxidation rate upon the bulk concentration is of a fractional
order, as even though the methanol concentration is increased by a factor of 10, the
oxidation current density in 2 M methanol is only about four times higher than in 0.2 M.

The fractional order of the dependence of the reaction rate upon the bulk
concentration of methanol can be interpreted by the Temkin-type methanol adsorption
isotherm. It is well known that the rate of methanol electrooxidation depends on the
surface concentration rather than bulk concentration of methanol (Equation 6.16). The
coverage of methanol residues, however, increases with the bulk concentration in a
logarithmic manner (Equation 6.17). Furthermore, the rate of methanol oxidation is
controlled by a balance between the adsorbed methanol residues and hydroxyl species, as
seen from Equation 6.16. The increase of the coverage of adsorbed methanol residues
may block some of the electrode sites, preventing adsorption of hydroxyl species, and

therefore, reduce the overall reaction rate.

176

 

 

 

 

 

 

 

 

 

700 ' ' v r * I V *v v v v
600 4
500

a 400 1
v 300 Methanol ‘
E ' Concentration '
Q) 200 1' 2-0 M "

t . 1.5 M

= 100 _ _ 1.0 M

0.6 M

U . 0.2 M
0 - .
. 1

.1m A A n A A n A g A A A A A A
0.0 5.0 10.0 15.0 20.0 25.0

 

Time (min.)
Figure 6.5 Current-time transients for a Pt/diamond composite electrode obtained after a
potential step to 0.6 V vs. Ag/AgCl in 0.1 M HClO4 containing different concentrations

of methanol. Active surface area of the composite electrode is 0.60 cmz.

 

 

 

 

 

'304 ' r f I - 1 v v r v
.5" -3.6 - .
E .
9 , ,
<
a -3.8 - «
ED 1
-4.0 - «
1 1
4.2 A A A A A 1 L L A n A
-0.80 -0.60 .040 .0.20 0.00 0.20 0.40
log (c/M)

Figure 6.6 Log i vs. log 0 plot at 0.6 V vs. Ag/AgCl. Data taken from the current-time

transients shown in Figure 6.5.

177

Nevertheless, the apparent reaction order can be determined by using the
following equations. The overall methanol elecrtooxidation current density can be
expressed as20

i: anC'" (6-18)

Where n is the number of electrons, F is the Faraday constant, k is the reaction constant,
C is the bulk concentration of methanol, and m is the reaction order, then

logi= logan+mlogC (6.19)~
and the slope of the log i vs. log C at a constant potential gives a apparent reaction order
(m) of methanol electro-oxidation reaction with respect to methanol concentration. Using
the steady-state current densities, the plot for methanol oxidation is presented in Figure
6.6. A reaction order of 0.6 is estimated, which is similar to that reported by Bagotaky et
al.17 for smooth and platinized Pt electrode.

One of the figures of merit for catalytic activity is the turnover number (tn)
defined as the number of the molecules converted per surface site per second. For
oxidative heterogeneous catalysis, the turnover number should be in the range from 10 to
100 to make a given material an effective catalyst. However, to date, the turnover number
for MOR at the noble metal catalyst in the operational fuel cell is ca. 0.06, still far below
the needed performance level. The following formula can be used to calculate the

turnover number (tn) from the current density:2|

(6.20)

 

nF-l.3><10'5

I molecules _ i(mA - cm‘z) ~ 6.02x1020
s - site

178

There are 1.3 x 1015 platinum sites per 1 cm2 of the real platinum surface area, and n = 6
for the methanol oxidation reaction yielding C02 as the product. Collection of the

constants yields the formula
tn = 0.8 - i(mA - cm'z) (6.21)
Calculated turnover numbers are listed in Table 6.4. The turnover number increases with

methanol concentration as expected.

 

Table 6.4 Summary of the Chronoamperometric i-t data for the oxidation of methanol at

 

 

 

a Pt/diamond composite electrode in 0.1 M HClO4 as a function of the methanol -,_-
concentration. ' .-
Methanol concentration 0.2 0.6 1.0 l .5 2.0
(M)
Charge density 0.20 0.30 0.45 0.50 0.55
(C-cm'z)

Steady state 0.08 0.15 0.22 0.26 0.32

current density
(mA- cm'z)
Turnover number 0.06 0.12 0.18 0.21 0.26

 

Note: Data recorded at 20 min. after a potential step from 0 to 0.6 V vs. Ag/AgCl. All

measurements WCI‘C made at room temperature.

6.4.3 MOR at a Bimetallic Catalyst
The best catalyst for methanol oxidation to date, is a Pt/Ru bimetallic catalyst

22-25

dispersed on a carbon support or directly on a Nafion PEM membrane.26 The

mechanism proposed in Section 6.2 by which the alloy improves the electrocatalytic

179

 

activity of pure Pt is a simple bifunctional one. The atoms of different metals act
independently and perform different functions. Methanol adsorption and decomposition
takes place on Pt, while the Ru atoms provide preferred sites for binding OHads. In this
section, some preliminary results for fabricating and testing Pt/Ru/diamond composite

electrodes are presented.

4.0 nm

2.0 nm

0.0 nm

 

Figure 6.7 AFM image (10 x 10 um) of a Pt/Ru/diamond composite electrode. The
deposition time is 300 s.

Figure 6.7 shows a top-view, 10 X 10 pm2 AFM image of Pt/Ru/ composite
electrodes. The co—deposition of Pt and Ru was performed in 0.1 M HC104 containing 1
mM K2PtC16 + 0.5 mM RuCl3 with a deposition time of 300 s. Post metal deposition,
diamond film growth was applied for 2 h to entrap the metal particles. Numerous
particles are randomly distributed on the microcrystallite facets and in the grain
boundaries. The particle size ranges from 50 to 300 nm with a distribution of about 5
x108/cm2. Energy dispersive x—ray (EDX) analysis confirmed the presence of Ru. The

estimated atomic ratio of Pt/Ru was 89:11.

180

 

 

 

200--.--.--,.ﬁ,

 

150 ’ Ptos9Ru011/dlam0nd

’ ................... Pt/diamond :5
100 -

50L

A A LA A I A A j

 

 

 

 

 

 

 

' -0.40 0.00 0.40 0.80 1.20 1.60

 

 

Current (11A)
§

 

600-.-.-f-.v
D

 

500: ................. . Wdiamond f1.“ .

   

 

 

 

-1 m L l A A A l A A A A A A A
-0.20 0.00 0.20 0.40 0.60 0.80 1.00 1.20

Potential (V vs. Ag/AgCl)

Figure 6.8 Cyclic voltammetric i-E curves for the Pt/Ru/diamond composite electrodes
in (A) 0.1 M HClO4 and (B) 0.1M HClO4 + 0.2 M CH3OH. The electrodeposition time is

300 5. Scan rate = 50 mV/s.

181

The as-deposited Pt/Ru/diamond composite electrode was initially activated by
potential cycling in Nz-purged 0.1 M HClO4. The cycling was limited to between -0.3
and 1.0 V in order to avoid oxidizing the Ru. The cyclic voltammetric i-E curve for the
composite electrode is shown in Figure 6.8A. For comparison, the voltammetric response
for a Pt/diamond composite electrode, fabricated with the same Pt deposition time of 300
s, is shown as a dashed line. For the Pt/Ru/diamond composite electrode, there are
noticeable changes in the voltammetric features. The hydrogen adsorption/desorption
peaks are suppressed over the potential range from -0.3 to 0 V, and the charge in the
double layer region, 0 to 0.15 V, increases. Discemible faradaic current commences at ca.
0.15 V, presumably due to the dissociation of water to form O-containing species at Ru
(Reaction 6.12). The onset potential for dissocative chemisorption of water on Ru is
about 0.2 V negative of that for Pt because Ru is more oxophilic than Pt.27 It is
reasonable to suppose that this facile dissociation of water on Ru plays an important role
in the promotion of methanol oxidation on Pt/Ru catalysts.28

Cyclic voltammetric i-E curves for the electrodes in 0.1 M HClO4 containing 0.6
M CH3OH are presented in Figure 6.88. During the anodic potential sweep, the methanol
oxidation current commences at potential as low as 0.1 V, and starts to increase rapidly at
0.2 V. Note that two peaks are observed near 0.48 and 0.62 V, respectively. Since
dissociation of water on Ru starts at 0.15 V, the former peak may correspond to the
methanol electrooxidation promoted by the adsorbed OH species. The latter one might be
attributed to the methanol electrooxidation on Pt sites because it appears at the same

potential as that for Pt/diamond electrode. The most notable feature is the drastically

182

 

 

enhanced catalytic activity of the alloy catalyst at low potential. At 0.4 V, the oxidation

current on the alloy catalyst is about 6 times higher than that for Pt.

6.5 Conclusions

A kinetic study of the methanol electrooxidation reaction (MOR) on the
metal/diamond composite electrodes was presented in this chapter. The results can be
summarized as follows:

1. The mechanism of methanol electrooxidation on the metal/diamond composite
electrode follows that proposed for polycrystalline Pt and supported Pt
catalyst. The speciﬁc activity of the supported Pt catalyst on diamond is
comparable to that on high surface area carbons.

2. Anion adsorption affects the rate of methanol oxidation. Specifically,
adsorbed anions compete with methanol molecules for adsorption sites
resulting in a much smaller oxidation currents. Specifically adsorbed anions
may also block the adsorption of poison species, minimizing the poisoning
effect.

3. The reaction order is about 0.6 over the methanol concentration range from
0.2 to 2 M.

4. Pt/Ru alloy particles can be incorporated into the diamond surface micro-
structure. The bimetallic catalyst shows enhanced catalytic activity toward

methanol oxidation at low potential region.

183

References:

10.

11.

12.

l3.

14.

15.

A. Hammett, in lnterfacial Electrochemistry: Theory, Experiment, and
Applications, A. Wieckowski, Editors, Dekker, New York, 843 (1999).

T. D. Jarvi and E. M. Stuve, in Electrocatalysis, J. Lipkowski and P. N. Ross,
Editors, Wiley-VCH, New York, 75 (1998).

B. Beden, J. M. Leger, and C. Lamy, in Modern Aspects of Electrochemistry, J. O.
M. Bokris, B. E. Conway and R. E. White, Editors, Plenum, New York, 22, 97
(1992).

S. G. Sun, in Electrocatalysis, J. Lipkowski and P. N. Ross, Editors, Wiley-VCH,
New York, 243 (1998).

C. Lamy, J. M. Leger, and S. Srinivasan, in Modern Aspects of Electrochemistry,
J. O. M. Brokris, B. E. Conway and R. E. White, Plenum, New York, 34, 53
(2001).

C. Lamy and J. M. Leger, in Interfacial Electrochemistry: Theory, Experiment,
and Applications, A. Wieckowski, Editor, Marcel Dekker, New York, 885 (1999).

B. Gurau, A. B0, Y. Yi, H. W. Lei, S. Suh, R. Liu, S. R. Bare, and E. S. Smotkin,
Preprints of Symposia - American Chemical Society, Division of Fuel Chemistry,

46, 691 (2001).

H. A. Gasteiger, N. M. Markovic, and P. N. Ross, Jr., J. Phys. Chem, 99, 8290
(1995).

W. Chrzanowaki and A. Wieckowski, Langmuir, 14, 1967 (1998).

H. A. Gasteiger, N. Markovic, and P. N. Ross, J. Phys. Chem, 99, 8945 (1995).
B. Grgur, N. Markovic, and P. N. Ross, J. Phys. Chem. B, 102, 2494 (1998).

S. Mukerjee and J. McBreen, J. Electroanal. Chem, 448, 163 (1998).

K. Yakikozava, Y. Fujii, Y. Matsuda, K. Nishimura, and Y. Takasu, Electrochim.
Acta, 36, 973 (1991).

A. Kabbabi, F. Gloaguen, F. Andolfatto, and R. Durand, J. Electroanal. Chem,
373, 251 (1994).

S. L. Gojkovic and T. R. Vidakovic, Electrochimica Acta, 47, 633 (2001).

184

 

 

16.

17.

18.

19.

20.

21.

23.

24.

25.

26.

27.

28.

M. Watanabe, S. Saegusa, and P. Stonehart, J. Electroanal. Chem, 271, 213
(1989).

V. S. Bagotzky and Y. B. Vassilyev, Electrochim. Acta, 12, 1323 (1967).
E. Herrero, K. Franaszczuk, and A. Wieckowski, J. Phys. Chem, 98, 5074 (1994).
H. Ogasawara and M. Ito, Chem. Phys. Lett., 245, 304 (1995).

A. J. Bard and L. R. Faulkner, Electrochemical Methods, Fundamentals and
Applications, John Wiley & Sons, Inc, New York, (2001).

W. Chrzanowaki and A. Wieckowski, in Interfacial Electrochemistry: Theory,
Experiment, and Applications, A. Wieckowski, Editor, Marcel Dekker, New
York, 937 (1999).

 

O. Savadogo and M. Lacroix, New Materials for Fuel Cell and Modern Battery
Systems 11, Proceedings of the International Symposium on New Materials for
Fuel Cell and Modern Battery Systems, 2nd, Montreal, July 6-10, 1997, 872
(1997).

H. Hoster, T. Iwasita, H. Baumgartner, and W. Vielstich, Physical Chemistry
Chemical Physics, 3, 337 (2001).

W. Chrzanowski, H. Kim, G. Tremiliosi-Filho, A. Wieckowski, B. Grzybowska,
and P. Kulesza, Journal of New Materials for Electrochemical Systems, 1, 31
(1998).

W. Chrzanowski and A. Wieckowski, Langmuir, 14, 1967 (1998).

X. M. Ren, M. S. Wilson, and S. Gottesfeld, J. Electrochem. Soc., 143, L12
(1996).

A. J. Applyby, Surf. Sci., 27, 225 (1971).

A. Wieckowski, H. Kim, I. R. de Moraes, Q. Lu, and A. Crown, Book of
Abstracts, 218th ACS National Meeting, New Orleans, Aug. 22—26, 1999.

185