Electroreduction of oxygen on carbon-supported gold catalysts

Article abstract of DOI:10.1016/j.electacta.2009.08.001

The electrochemical reduction of oxygen was studied on Au/C catalysts (20 and 30 wt%) in 0.5 M H₂SO₄ and 0.1 M KOH solutions using the rotating disk electrode (RDE) method. The thickness of the Au/C-Nafion layers was varied between 1

Full text of DOI:10.1016/j.electacta.2009.08.001

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Electrochimica Acta 54 (2009) 7483–7489
Contents lists available at ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
Electroreduction of oxygen on carbon-supported gold catalysts
Heiki Eriksona, Gea Jürmanna, Ave Sarapuua, Robert J. Potterb, Kaido Tammeveskia,∗
a Institute of Chemistry, University of Tartu, Jakobi 2, 51014 Tartu, Estonia
b Johnson Matthey Technology Centre, Blount’s Court, Sonning Common, Reading RG4 9NH, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 6 April 2009
Received in revised form 3 August 2009
Accepted 4 August 2009
Available online 11 August 2009
The electrochemical reduction of oxygen was studied on Au/C catalysts (20 and 30 wt%) in 0.5 M H2SO4
and 0.1 M KOH solutions using the rotating disk electrode (RDE) method. The thickness of the Au/C–Nafion
layers was varied between 1.5 and 10 ␮m. The specific activity of Au was independent of catalyst loading
in both solutions, indicating that the transport of reactants through the catalyst layer does not limit the
process of oxygen reduction under these conditions. The mass activity of 20 wt% Au/C catalysts was higher
due to smaller particle size. The number of electrons involved in the reaction and the Tafel slopes were
found; the values of these parameters are similar to that of bulk polycrystalline gold and indicate that
the mechanism of O2 reduction is not affected by carbon support or the catalyst configuration.
© 2009 Elsevier Ltd. All rights reserved.
Keywords:
Oxygen reduction
Electrocatalysis
Kinetic parameters
Au nanoparticles
High-area carbon
1. Introduction
tion of peroxide to OH− in a limited potential region at low
overpotentials [30–37]. By varying the conditions of Au electrode-
The electroreduction of oxygen is one of the most extensively
studied reactions in electrochemistry, as it finds application in
many technologies, most importantly, in fuel cells and metal-air
batteries [1,2]. The most active electrocatalyst for this reaction
is platinum, but due to its high cost and scarcity, investigations
are directed also to other materials. Gold electrodes show modest
activity towards O2 reduction in acidic solutions, but in the alkaline
media it is greatly increased, especially on Au(1 0 0) single crystal
face [2,3].
In the last decade gold nanoparticles (AuNPs) have gained
attention for unique catalytic properties for several reactions,
for example, low temperature CO oxidation [4,5]. O2 reduction
has been studied on Au nanoparticles supported on bulk carbon
electrodes [6–27] prepared by various methods, such as elec-
trodeposition [10–19], vacuum evaporation and sputter deposition
[20–22,24–27]. The reduction of O2 has been also investigated on
colloidal AuNPs attached to self-assembled monolayers [28,29].
The morphology of the nanostructured Au has a large influence
on its electrocatalytic activity towards O2 reduction, because this
reaction is highly structure sensitive [2,3]. In acidic solution, the
activity increases in the sequence of Au(1 1 1) < Au(1 1 0) < Au(1 0 0)
[2,3]. In alkaline solution, Au(1 0 0) has much higher activity than
the other low-index surfaces and it supports the further reduc-
position, electrodes of different morphology can be prepared, and
their electrocatalytic activity also varies [10,14–16]. There are
also other methods for preparing the AuNPs of well-defined mor-
phology, for example, gold nanorods with only (1 1 1) and (1 1 0)
surface domains [6,7,9] and cubic Au nanoparticles with high
amount of (1 0 0) sites [8,9] have been synthesised in water-in-oil
microemulsion in the presence of different additives. Au nanopar-
ticles electrodeposited on an organic template partially inhibit the
4e− O2 reduction pathway compared with bulk Au electrodes [18].
The oxygen reduction reaction in alkaline solution can serve as an
indirect means for characterising the crystallographic orientation
of such structures [6–9,14–16]. In some cases, the nanostructured
Au electrodes have shown increased electrocatalytic activity as
compared to bulk Au [13,17,20,21,38,39], but the origin of this
effect remains unclear. Recently, the increase of kinetic current
density of O2 reduction with decreasing AuNP size in alkaline
solution was noted [40]. In acid solutions, however, the specific
activity of vacuum-deposited Au films decreased as the size of
AuNPs decreased below ∼3 nm [25]. The Au-coated nanoparticles of
other metals (Fe, Co, Ni and Pb) exhibited higher oxygen reduction
activity in acid solutions than bulk Au [41].
For practical electrocatalysts, optimisation of the catalyst layer
thickness is crucial. It is desirable to keep the catalyst loading as
low as possible without performance loss. The RDE method is con-
venient for studying the carbon-supported Pt electrocatalysts in
the form of “ink” electrodes and the reproducibility of the mea-
surements is satisfactory [42]. Several reports deal with thickness
∗
Corresponding author. Tel.: +372 7375168; fax: +372 7375181.
E-mail address: kaido@chem.ut.ee (K. Tammeveski).
0013-4686/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2009.08.001

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
7484
H. Erikson et al. / Electrochimica Acta 54 (2009) 7483–7489
on the electrodes was varied from 5.6 to 57 ␮g cm−2, which corre-
sponds to approximate thickness of the catalyst layer from 1.5 to
10 ␮m. The electrodes were left to dry overnight at room temper-
ature.
effects of Pt/C catalysts for oxygen reduction [43–46]. However, if
the catalyst film on the electrode is rather thick, then O2 is reduced
on all catalyst particles only at low current densities, while at higher
overpotentials, just the outermost layer of the film is active [42].
The studies of the electrodes with catalyst gradient confirm that
the catalyst particles near the solution interface are utilized more
efficiently [47]. The macro-homogeneous model has been used for
describing the behaviour of the compact catalyst layer on RDE [43],
whereas for gas diffusion electrodes, the agglomerate model gives
better results [44]. This model takes into account gas diffusion,
ohmicdropsand interfacial kinetics withinthethinlayer andallows
calculating the kinetic parameters [43]. Attachment of the catalyst
powder on the electrode via a very thin Nafion film minimizes the
effect of oxygen diffusion and the mathematical modelling is not
necessary [48].
Jiang and Yi have found that the activity of the Pt/C films
increases with increasing the catalyst film thickness. As an expla-
nation to this effect they proposed increased utilization efficiency
of the catalyst that is caused by the enhanced diffusion of
adsorbed reagent species on Pt particles for thicker films, where
the particle–particle distance is shorter [45]. They concluded that
a film thickness of 2–4 ␮m is required for reasonable activity [45].
There are only a few reports dealing with oxygen reduction on
practical Au/C catalysts. Zhong et al. have prepared Au and AuPt
nanoparticles with diameter (d) of 3–5 nm supported on carbon
black and studied the reduction of oxygen in acid and alkaline solu-
tion. They found that both 2e− and 4e− reduction of oxygen takes
place on Au/C catalysts [49,50]. Similarly, the value of the number
of electrons transferred per O2 molecule n = 3.5–4 was determined
by Lobyntseva et al. in alkaline solution [51]. In their study, polyte-
trafluoroethylene was used as a binder in preparing Au/C catalyst
layers. Bron studied the reduction of oxygen in acid media on Au/C
catalysts and concluded that the surface specific activity does not
depend on the Au particle size or on the type of the supporting car-
bon black [52]. The electrocatalytic properties of AuNPs modified
carbon nanotubes in acid solution have also been studied [53,54].
The objective of the present work was to study the effect of the
thickness of Au/C catalyst layer on the kinetics of oxygen reduction
in acid and alkaline solutions. Thin layer rotating disk electrode
method was used for electrochemical testing of the catalysts. To
our knowledge, this is the first work in which the dependence of
O2 reduction activity on the Au/C catalyst layer thickness has been
systematically investigated.
2.2. Electrochemical measurements
Oxygen reduction was studied in 0.5 M H2SO4 and in 0.1 M KOH
employing a rotating disk electrode (RDE). The solutions were pre-
pared from 96% H2SO4 (Suprapur, Merck) or potassium hydroxide
pellets (pro analysis, Merck) and Milli-Q (Millipore) water; these
were saturated with pure O2 (99.999%, AGA) or deaerated with Ar
gas (99.999%, AGA).
An EDI101 rotator and a CTV101 speed control unit (Radiome-
ter, Copenhagen) were used for the RDE experiments. A saturated
calomel electrode (SCE) was employed as a reference and all the
potentials are referred to this electrode. The experiments were
carried out in a three-electrode glass cell. Pt wire served as a
counter electrode and the counter electrode compartment was sep-
arated from the main cell compartment by a glass frit. Potential
was applied with an Autolab potentiostat/galvanostat PGSTAT30
(Eco Chemie B.V., The Netherlands) and the experiments were
controlled with General Purpose Electrochemical System (GPES)
software. All experiments were carried out at room temperature
(23 1 ◦C).
2.3. Characterisation of the Au/C catalysts structure and
morphology
Transmission electron microscopy (TEM) measurements were
performed on a Tecnai F20 instrument at 200 kV accelerating
voltage. Images were acquired in scanning transmission electron
microscopy (STEM) mode using a high angle annular dark field
(HAADF) detector. For preparation of the TEM specimens, a small
portion of catalyst powder was dispersed directly onto perforated
carbon film copper grid.
X-ray diffraction (XRD) data was acquired with a Bruker AXS D8
diffractometer using Ni-filtered Cu K␣ radiation at the tube voltage
of 40 kV and 40 mA current. A Lynxeye PSD detector was employed
at the 2ꢀ scan range of 10–130◦ and 0.02◦ step size. The average
grain size of the Au nanoparticles was calculated using Rietveld
analysis.
3. Results and discussion
2. Experimental
3.1. Surface characterisation of the Au/C catalysts
2.1. Electrode preparation
The representative TEM micrographs for two Au/C catalysts
employed are presented in Fig. 1. The particle size analysis was
conducted on 264 and 123 particles for 20 and 30 wt% catalysts,
respectively. A rather wide size distribution with the maximum
frequency at about 10 nm for 20 wt% catalyst and 14 nm for 30 wt%
catalyst was obtained. The average particle size was 11.0 1.7 nm
for 20 wt% catalyst and 14.0 1.7 nm for 30 wt% catalyst.
For comparison, the XRD method was used for the estimation
of Au particle size. The XRD pattern of Au/C catalysts is shown in
Fig. 2. Four major peaks with 2ꢀ values of 38.2◦, 44.4◦, 64.6◦ and
77.6◦ corresponding to the (1 1 1), (2 0 0), (2 2 0) and (3 1 1) planes of
the bulk Au, respectively, were observed, which can be assigned to
the Au face-centered cubic structure. The broad peak at about 25◦ is
due to the graphitic regions of carbon support. The Rietveld analysis
was carried out to calculate the crystallite size of Au particles and
the average values of d = 16.8 nm and d = 26.3 nm were obtained
for 20 and 30 wt% Au/C, respectively. These values are larger than
those obtained from TEM images. This is most probably due to the
Au/C catalysts (20 and 30 wt% Au nanoparticles supported on
Vulcan XC-72R carbon black) were supplied by Johnson Matthey.
The catalysts were prepared by a proprietary process involving the
base hydrolysis of an aqueous solution of hydrogen tetrachloroau-
ric acid (HAuCl4). Carbon powder (Vulcan XC-72R, product of Cabot
Corporation, Inc.) was used for comparison. The specific surface
area of the carbon powder used was 250 m2 g−1 according to the
producer. Catalyst suspensions were prepared by mixing 4.7 mg of
catalyst powder, 20 ␮l of 5 wt% solution of Nafion (Aldrich) and
a known volume of ethanol, followed by sonication for 15 min.
Glassy carbon (GC) electrodes with a geometric area of 0.196 cm2
were prepared by mounting GC disks (GC-20SS, Tokai Carbon) into
Teflon holders. The surface of the GC electrodes was polished to
a mirror finish with 1.0 ␮m alumina slurry (Buehler) and after
that the electrodes were sonicated in Milli-Q (Millipore) water
for 5 min. A 5 ␮l aliquot of the catalyst suspension was dropped
onto the GC disk using a micropipette. The resulting Au loading

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
H. Erikson et al. / Electrochimica Acta 54 (2009) 7483–7489
7485
Fig. 1. High angle annular dark field TEM images (a and b) and Au particle size distribution (c and d) of Au/C catalysts. (a and c) 20 wt% catalyst, (b and d) 30 wt% catalyst.
oxide reduction peak and using a value of 400 ␮C cm−2 for the
reduction of an oxide monolayer [55]. This area increased linearly
with increasing the catalyst loading, which indicates that the acces-
sibility of the Au particles is not affected by changing the layer
thickness.
wide size distribution of Au particles, as the small number of large
Au particles contains a significant fraction of all gold and therefore
have a higher contribution to the XRD response.
3.2. Cyclic voltammetry (CV)
Prior to O2 reduction measurements, the thin layer Au/C elec-
trodes were electrochemically pre-treated and characterised by
cyclic voltammetry in Ar-saturated 0.5 M H2SO4 by scanning the
potential for 20 cycles between −0.3 and 1.45 V at a scan rate of
100 mV s−1. The stable CV curves obtained are presented in Fig. 3.
The anodic peak at E > 1.1 V corresponds to the formation of Au
surface oxides and the cathodic peak at ca 0.89 V to the reduction
of these oxides [55]. The large background current is mainly due
to the supporting carbon black and it increases proportionally to
the catalyst loading. The current increase at E > 1.3 V is apparently
caused by the oxidation of carbon surface. The pair of peaks cen-
tered at ca 0.3 V is related to the oxidation and reduction of some
functional groups on the carbon surface and it has been suggested
that these may be quinone-type species [56]. These peaks increase
during the first potential cycles as a result of the oxidation of the
carbon surface.
The electroactive surface area of gold (Ar) was determined from
the stable cyclic voltammograms by charge integration under the
Fig. 2. XRD pattern for (a) 20 wt% Au/C catalyst and (b) 30 wt% Au/C catalyst.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
7486
H. Erikson et al. / Electrochimica Acta 54 (2009) 7483–7489
Fig. 4. RDE voltammetry curves for O2 reduction on 10 ␮m Au/C (20 wt%) in O2-
saturated 0.5 M H2SO4. ꢁ = 10 mV s−1. Inset shows the Koutecky–Levich plots for O2
reduction derived from the RDE data at various potentials: (ꢀ) 0 V; (᭹) −0.05 V; (ꢁ)
−0.1 V; (ꢂ) −0.2 V and (ꢃ) −0.3 V.
catalyst layer thickness. This is in accordance to previous results
obtained for Au thin films [26,27] and Au/C catalysts [52].
The electrocatalytic activity of Au/C electrodes towards oxygen
reduction was almost proportionally increasing with increasing the
catalyst loading and the half-wave potential (E1/2) shifted positively
(Table 1). The j–E curves for 20 wt% Au/C electrodes of different Au
loadings are presented in Fig. 5. The specific activity (SA) of the
electrodes was calculated from
Ik
SA =
(2)
Ar
where Ik is the kinetic current and Ar is the real surface area of gold;
the SA values at E = 0 V vs. SCE are given in Table 1 and in Fig. 6. At
this potential, the supporting carbon powder is inactive towards
oxygen reduction (Fig. 5) and the reduction occurs only at Au parti-
cles. It is evident that there is no large dependence of the SA value
on the catalyst loading. In addition, the specific activities of 20 wt%
Au/C and 30 wt% Au/C catalysts are not significantly different. The
mass activities (MA) were determined from
Fig. 3. Cyclic voltammograms for (a) 20 wt% Au/C catalyst and (b) 30 wt% Au/C cat-
alyst in Ar-saturated 0.5 M H2SO4. Catalyst layer thickness: (1) 10 ␮m; (2) 6 ␮m; (3)
3 ␮m and (4) 1.5 ␮m. ꢁ = 100 mV s−1
.
3.3. Oxygen reduction in 0.5 M H2SO4
Both gold and carbon are rather inactive catalysts for oxygen
reduction in acid solution. For all Au/C electrodes studied, single-
wave oxygen reduction polarisation curves with no well-defined
current plateau were obtained using the rotating disk electrode
method, and a typical set of current–potential curves registered
at various electrode rotation rates for a 20 wt% catalyst is given
in Fig. 4. The Koutecky–Levich (K–L) equation was employed for
analysing the RDE data [57]:
Ik
MA =
(3)
mAu
where Ik is the kinetic current and mAu is the mass of gold in the
catalyst layer, calculated theoretically from the mass of the catalyst
applied onto the electrode. The mass activity was smaller for 30 wt%
1
j
1
jk
1
jd
1
1
=
+
= −
−
(1)
nFkCOb
0.62nFD2O/3
ꢁ
−1/6COb ω1/2
2
2
2
where j is the measured current density, jk and jd are the kinetic
and diffusion-limited current densities, respectively, n is the num-
ber of electrons transferred per O2 molecule, k is the rate constant
for O2 reduction, F is the Faraday constant (96 485 C mol−1), ω is
the rotation rate, COb is the concentration of oxygen in the bulk
2
(1.13 × 10−6 mol cm−3 [58]), DO is the diffusion coefficient of oxy-
gen (1.8 × 10−5 cm2 s−1) [58] an2d ꢁ is the kinematic viscosity of the
solution (0.01 cm2 s−1) [59]. The corresponding K–L plots are pre-
sented in the inset of Fig. 4. From the slopes of the K–L plots, the
number of electrons transferred per O2 molecule (n) was found. The
values of n were close to 2 at the foot of the polarisation curve and
therefore, H2O2 is the final reduction product at these potentials. n
gradually increases at more negative potentials, as a result of fur-
ther reduction of H2O2. At E = −0.3 V the n value slightly depends on
the catalyst loading, increasing from ca 2.5 to 3 with increasing the
Fig. 5. RDE voltammetry curves for O2 reduction on 20 wt% Au/C catalyst in O2-
saturated 0.5 M H2SO4. Catalyst layer thickness: (ꢁ) 1.5 ␮m; (ꢃ) 3 ␮m; (᭹) 6 ␮m
and (ꢀ) 10 ␮m. (×) 3 ␮m layer of carbon powder; (ꢂ) bulk Au. ω = 1900 rpm;
ꢁ = 10 mV s−1
.