Changes induced by transition metal oxides in Pt nanoparticles unveil the effects of electronic properties on oxygen reduction activity

Article abstract of DOI:10.1039/c8ta10642h

Although the relevance of electronic effects in the electrocatalysis of the oxygen reduction reaction has been recognized, the impossibility of separating the effects of composition and particle size for Pt-based materials has hindered establishing clear activity-property relationships. Herein, we report a systematic study based on induced changes via the interactions of pure Pt nanoparticles with transition metal oxide/carbon supports (Pt/MO_x/C catalysts, MO_x = CeO₂, SnO₂, TiO₂, ZrO₂ and WO₃). A thorough analysis of aberration-corrected HR-STEM images demonstrated that Pt particles are similar in size and shape for all catalysts, while the direct probing of electronic properties by in situ X-ray absorption spectroscopy evidenced charge transfer between Pt and the supports. This approach allowed ascribing the changes in electrocatalytic activity for oxygen reduction solely to the variations in the electronic vacancy of the Pt 5d band resulting from the interactions between the metal nanoparticles and the supports containing different transition metal oxides. Oxygen reduction was studied in acid and in alkaline solutions, and linear correlations between the kinetic current densities and the Pt 5d band vacancy of pure Pt nanoparticles were found in both media. Possible first steps of the reduction of oxygen are discussed to explain the trends observed. The results, evidencing that enhanced ORR activity on Pt particles is promoted by a lower 5d band vacancy in acid solutions and by a higher one in alkaline medium, provide new insights on the fundamental aspects of oxygen reduction, and open up new possibilities to develop catalysts with enhanced activity for fuel cell cathodes by tuning their electronic properties.

Full text of DOI:10.1039/c8ta10642h

Journal of
Materials Chemistry A
View Article Online
View Journal
PAPER
Changes induced by transition metal oxides in Pt
nanoparticles unveil the effects of electronic
properties on oxygen reduction activity†
Cite this: DOI: 10.1039/c8ta10642h
b
c
Felipe B. Ometto,a Emilia A. Carbonio,
E´rico Teixeira-Neto
‡
a
*
and Hebe M. Villullas
Although the relevance of electronic effects in the electrocatalysis of the oxygen reduction reaction has
been recognized, the impossibility of separating the effects of composition and particle size for Pt-
based materials has hindered establishing clear activity–property relationships. Herein, we report
a systematic study based on induced changes via the interactions of pure Pt nanoparticles with
transition metal oxide/carbon supports (Pt/MOx/C catalysts, MOx ¼ CeO2, SnO2, TiO2, ZrO2 and
WO3). A thorough analysis of aberration-corrected HR-STEM images demonstrated that Pt particles
are similar in size and shape for all catalysts, while the direct probing of electronic properties by in
situ X-ray absorption spectroscopy evidenced charge transfer between Pt and the supports. This
approach allowed ascribing the changes in electrocatalytic activity for oxygen reduction solely to the
variations in the electronic vacancy of the Pt 5d band resulting from the interactions between the
metal nanoparticles and the supports containing different transition metal oxides. Oxygen reduction
was studied in acid and in alkaline solutions, and linear correlations between the kinetic current
densities and the Pt 5d band vacancy of pure Pt nanoparticles were found in both media. Possible
first steps of the reduction of oxygen are discussed to explain the trends observed. The results,
evidencing that enhanced ORR activity on Pt particles is promoted by a lower 5d band vacancy in
acid solutions and by a higher one in alkaline medium, provide new insights on the fundamental
aspects of oxygen reduction, and open up new possibilities to develop catalysts with enhanced
activity for fuel cell cathodes by tuning their electronic properties.
Received 5th November 2018
Accepted 18th December 2018
DOI: 10.1039/c8ta10642h
rsc.li/materials-a
Pt-based systems, such as PtCo,1–5 PtFe,3,5–7 PtNi,1,5,8,9 PtCu,8,10,11
PtCr,5,12–14 and PtV,5,14–16 that generally exhibit enhanced ORR
Introduction
activity as compared with Pt/C. Over the years, the ORR activity
enhancement promoted by the addition of a second metal has
been attributed to more favorable Pt–Pt nearest-neighbor
distances,17 to structural and electronic properties,18,19 and to
a reduction of coverage of the Pt surface by OH species.19–22 The
particle size, shape, structure and composition of Pt-based
bimetallic catalysts also inuence considerably the ORR
activity.23 The analysis of the individual effects of those different
properties is, however, a long-standing difficulty because, more
oen than not, they are interdependent for nanoparticles. For
instance, varying the composition of PtFe nanoparticles of
similar sizes not only alters the lattice parameter and, thus, the
Pt–Pt distance but also promotes changes in the electronic
vacancy of the Pt 5d band,24 while variations in size and alloying
produced by heat treatments also affect the electronic proper-
ties of Pt.6,16 Improvements in ORR activity were also reported
for Pt nanoparticles on different supports, such as oxides25–28
and carbon-oxide hybrids.20,29–34 Strong metal–support interac-
tions (SMSI)35 inducing electronic effects were invoked to
explain the effects of supports on ORR activity,29,33 even though
The oxygen reduction reaction (ORR) is one of the most relevant
electrochemical processes in energy conversion. Despite the
progress made regarding the performance of low temperature
fuel cells, the ORR, which takes place at the cell cathode, has
remained a cause of large voltage losses that result from its very
sluggish kinetics even on Pt, which is the most active metal for
this reaction in acid solutions. Research aiming to develop
high-performance ORR catalysts has been quite intense over the
last few decades, and many papers were devoted to the study of
a
˜
Sao Paulo State University (UNESP), Institute of Chemistry, Rua Prof. Francisco Degni,
˜
55 – 14800-060 Araraquara, Sao Paulo, Brazil. E-mail: mercedes.villullas@unesp.br
b
¨
Helmholtz-Zentrum Berlin fur Materialien und Energie, BESSY-II, Albert-Einstein-
Straße 15, 12489, Berlin, Germany
cBrazilian Nanotechnology National Laboratory (LNNano), Brazilian Center for
˜
Research in Energy and Materials (CNPEM), Campinas, Sao Paulo, Brazil
† Electronic supplementary information (ESI) available. See DOI:
10.1039/c8ta10642h
‡ Present address: University of Campinas (Unicamp), Institute of Chemistry.
´
˜
Inomat – Bloco J, Cidade Universitaria, 13083-970 Campinas, Sao Paulo, Brazil,
CP 3154.
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
compressive strain36 and alloy formation37 were also proposed.
In most cases, however, the enhanced ORR activity promoted by
different supports is attributed to electronic effects although,
more oen than not, on a merely speculative basis because
experimental data on electronic properties are seldom reported.
Recent reviews on the ORR on Pt-based catalysts38–43 and on
support effects are available.44–46
Oxygen reduction is a complex process and no agreement
has been reached regarding the details of the reaction mecha-
nism and the rate-determining step. Nonetheless, a consensus
seems to have been reached regarding the two main pathways,47
the direct 4-electron reduction to water and the peroxide
pathway:
Experimental
Preparation of catalysts
Pt nanoparticles were synthesized in dioctyl ether by a modied
polyol process4,53 by reducing the metal precursor with 1,2-
hexadecanediol in the presence of commercial nanopowders of
transition metal oxides. Concisely, 80.7 mg of platinum(II) ace-
tylacetonate (Sigma-Aldrich, 97%) and 16.1 mg of 1,2-hex-
adecanediol (Sigma-Aldrich, technical grade, 90%) were
dissolved in 10 mL of dioctyl ether (Sigma-Aldrich, 99%) and
kept constantly stirred under an Ar atmosphere. The solution
was then heated up to 110 ꢁC and then 32.0 mg of nanopowder
of the transition metal oxide dispersed in 5 mL of dioctyl ether
was added. Aer 5 minutes, 273 mL of oleic acid (Sigma-Aldrich,
technical grade, 90%) and 394 mL of oleylamine (Sigma-Aldrich,
technical grade, 70%) were added. Aer the addition of the
capping agents, the temperature was kept at 110 ꢁC for 15
minutes and then it was raised until reux (about 298 ꢁC). Aer
30 min under reux to complete the reduction of Pt, the heating
was stopped and the system was le to cool to room tempera-
ture. The solid (Pt/MOx) was occulated adding ethanol and
separated from the reaction medium by centrifugation (9000
rpm). The Pt/MOx material was then redispersed in a hexane +
ethanol mixture and centrifuged again. These last steps (redis-
persion and centrifugation) were repeated until obtaining
a colorless supernatant. In sequence, a dispersion of the Pt/MOx
powder in hexane was mixed with a dispersion of carbon
powder (Vulcan XC-72, Cabot Corp.) in isopropanol. This
mixture was kept under stirring overnight. Then, the obtained
Pt/MOx/C catalyst was separated by ltration and washed thor-
oughly with ethanol, acetone and water. Subsequently, the Pt/
MOx/C material was dispersed in ethanol another time, ltered
and washed again with ethanol, acetone and water. This
procedure was repeated twice. Finally, the catalyst was dried in
O2 + 4H+ + 4eꢀ / 2H2O (direct pathway)
O2 + 2H+ + 2eꢀ / H2O2 (peroxide pathway)
(1)
(2)
The H2O2 formed in reaction (2) can be further reduced to
water or disproportionate:
H2O2 + 2H+ + 2eꢀ / 2H2O
2H2O2 / 2H2O + O2
(3)
(4)
The equivalent reactions in alkaline solutions are:
O2 + 2H2O + 4eꢀ / 4OHꢀ (direct pathway)
(5)
(6)
(7)
(8)
ꢀ
O2 + H2O + 2eꢀ / HO2 + OHꢀ (peroxide pathway)
HO2 + H2O + 2eꢀ / 3OHꢀ
ꢀ
2HO2 / 2OHꢀ + O2
ꢀ
The experimental data gathered over several decades suggest
that the ORR on Pt and Pt-based alloys takes place through
parallel direct and peroxide pathways, with the reaction going
mainly through the 4-electron route, in both media. Several
series-parallel mechanisms were proposed.48–52
ꢁ
an oven at 80 C for 2 h.
All Pt/MOx/C catalysts were prepared with 20 wt% Pt loading
and with a transition metal oxide to carbon proportion of
20 : 80 wt%. The particle sizes of commercial nanopowders of
transition metal oxides (Sigma-Aldrich) were <50 nm for CeO2
and TiO2 and <100 nm for ZrO2, SnO2 and WO3.
The fundamental challenge of developing highly active
ORR catalysts is still strongly dependent on advances
regarding property–activity correlations. Clearly, changes in
the electronic properties of an active metal might modify the
strength with which different species adsorb on its surface
and, in turn, alter the catalytic activity of that metal for elec-
trochemical reactions involving adsorption of reactants and/
or intermediates, as in the ORR. Unfortunately, experimental
measurements for direct assessment of electronic properties
are still rare. Herein, we present a systematic assessment of
the effects of the electronic properties of pure Pt nanoparticles
on the ORR activity by tuning the 5d band electronic vacancy
by incorporating transition metal oxides into the support. It is
demonstrated, for the rst time, that the ORR kinetically
controlled current densities on Pt nanoparticles of similar size
and shape, i.e., in the absence of composition and particle size
and/or shape changes, vary linearly with the Pt 5d band elec-
tronic occupancy.
Catalyst characterization
XRD measurements were done with Rigaku Ultima IV equip-
˚
ment, model Ru200B using Cu Ka radiation (1.5406 A) and
recorded at 0.5 degrees per min for 2q values between 20ꢁ and
100ꢁ.
The catalyst morphology was investigated by the acquisition
of high-resolution images using an aberration corrected trans-
mission electron microscope FEI Titan Themis Cubed 60-300
operating at 300 kV. The images were acquired in scanning
mode (HR-STEM). The samples were prepared by directly
applying the catalyst powder onto standard Cu TEM grids.
XPS spectra were collected with a K-ALPHA surface analysis
spectrometer (Thermo Scientic, Inc.) using monochromatic Al
Ka (1486.7 eV) as the photon source, in constant energy analyzer
mode with an energy pass of 30 eV and energy steps of 0.1 eV. A
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2019

View Article Online
Paper
Journal of Materials Chemistry A
sample of Au/C was also analyzed to estimate the instrumental
broadening, which was later used for the deconvolution of spectra
of all other samples. In all cases, a Shirley background was sub-
tracted and spectra were corrected using the C 1s core-line (284.5
eV) as the reference. For deconvolution, a Lorentzian function was
used for all non-metallic components and a Doniach-Sunjic
function (i.e., asymmetric)54,55 was used for all metallic compo-
nents. For the tting procedure the asymmetry parameter, aDS,
was le to vary between 0.18 and 0.22, based on reported values
for Pt.56,57 The full width at half maximum (fwhm) for each
component was determined for the Pt/C material and the values
obtained were used for all the Pt/MOx/C catalysts.
Results and discussion
Catalyst properties
X-ray diffraction analysis showed for the Pt/C catalyst the
signals typical of the face centered cubic (fcc) structure of Pt
(JCPDS 4-802) at 2q about 40, 46, 67 and 81ꢁ as well as the one
characteristic of the carbon support at 2q around 25ꢁ. For Pt/
MOx/C catalysts, several of those signals overlap with those of
the transition metal oxide. However, comparison of the
diffraction patterns of Pt/MOx/C catalysts and Pt/C showed that
the 2q position could be compared with minimal interference of
the oxide signals for at least one Pt diffraction signal, as illus-
trated in Fig. 1 for Pt/CeO2/C. In this case, it is seen that no
signals of CeO2 appear in the region of the diffraction peak of
the [111] planes of Pt (2q about 40ꢁ). Comparisons of diffraction
patterns like that of Fig. 1 are shown for Pt/TiO2/C, Pt/SnO2/C,
Pt/ZrO2/C and Pt/WO3/C in Fig. S1.† From the positions of the Pt
diffraction peaks less affected by the signals (at 2q about 40, 46,
or 67ꢁ depending on the catalyst) it is apparent that the Pt
signals appear at the same 2q values, allowing to rule out the
formation of the PtM alloy.37 Additionally, these comparisons
suggest that Pt nanoparticles have similar crystallite sizes.
Thus, the average crystallite diameter (D) was estimated for
Pt/C using the Scherrer eqn:
In situ X-ray absorption spectroscopy (XAS) experiments were
carried out around the Pt L3 edge (11 564.25 eV, 2p3/2 / 5d
transitions) at the XAFS1 beamline of the Brazilian Synchrotron
Light Laboratory (LNLS). Measurements were carried out using
the catalyst as the working electrode in
a spectroelec-
trochemical cell similar to that described in the literature.58 For
that, a mixture of 30 mg of catalyst powder and 400 mL of
Naon® solution (Aldrich, 5 wt% in a mixture of alcohols and
water) was pressed on a carbon cloth. All working electrodes
were prepared with a Pt loading of 6 mg cmꢀ2. A Pt mesh and
a reversible hydrogen electrode were employed as auxiliary and
reference electrodes, respectively. Spectra were collected while
the catalysts were kept polarized at a constant potential (0.55,
0.80 and 0.90 V vs. RHE) in acid (0.5 M H2SO4) and in alkaline
(0.1 M KOH) solutions. Normalization of spectra was done with
the program Athena of the Demeter soware package,59 which
was also used to determine the Pt L3 edge energy of all raw
spectra from the zero crossing of the second derivative.
0:9l
D ¼
(9)
u cos q
˚
where l is the X-ray wavelength (1.54056 A) and q is the angle of
the Bragg reection. The crystallite mean size was found to be
2.5 nm.
The images obtained by HR-STEM showed that for all cata-
lysts, most of the Pt nanoparticles are deposited on the transi-
tion metal oxide forming small aggregates, as shown for
Pt/ZrO2/C and Pt/CeO2/C in Fig. 2. As can be seen, Pt nano-
particles seem to have similar sizes. In general, some Pt parti-
cles are also seen on the carbon support for all materials.
Images of other catalysts are shown in Fig. S2.†
The average particle diameter was estimated from a size
distribution histogram for Pt/C and it was found to be 3.1 ꢂ
0.4 nm (Fig. S3†). Close examination and comparison of several
Electrochemical measurements
A conventional three-compartment electrochemical cell was
used for all the experiments. Cyclic voltammetry (CV) experi-
ments were carried out in the potential range of 0.05 V to 1.0 V
in argon-saturated solution (0.5 M H2SO4 and 0.1 M KOH) at
a scan rate of 50 mV sꢀ1. The electrocatalytic activity for the ORR
was studied in O2-saturated solution using the rotating ring
disk electrode (RRDE) technique. The RRDE electrode consisted
of a glassy carbon disk (0.247 cm2) and a platinum ring (0.187
cm2) and had a collection efficiency of 0.37 (Pine Research
Instrumentation).
The catalysts were used in the form of an ultrathin layer,
which was obtained by depositing 13 mL of catalyst ink on
a glassy carbon disk. The catalyst ink was prepared with 2.9 mg
of catalyst powder, 1.0 mL of isopropanol and 15 mL of Naon®
solution. The metal load on the working electrode was 28 mg
cmꢀ2. A platinized platinum wire was used as the counter
electrode and potentials were measured against a reversible
hydrogen reference electrode. The ORR polarization curves were
recorded in the potential range of 0.1 V to 1.0 V at different
rotation rates and a sweep rate of 5 mV sꢀ1, while the hydrogen
peroxide formed during the reduction of oxygen was monitored
by its oxidation current on the Pt ring, which was kept polarized
at 1.2 V. All electrochemical measurements were carried out at
room temperature.
Fig. 1 Comparison of diffraction patterns of Pt/CeO2/C with those of
Pt/C and CeO2.
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
Fig. 2 HR-STEM images obtained in the bright field for Pt/ZrO2/C (top)
and Pt/CeO2/C (bottom).
Fig. 3 HR-STEM images of Pt nanoparticles. Pt/C (top) and Pt/ZrO2/C
(bottom).
images taken for each material revealed that Pt nanoparticles
have a similar shape and size (about 3 nm) for all the Pt/MOx/C
catalysts and Pt/C, as exemplied by Fig. 3, which shows Pt
particles on Pt/C and on the carbon of the Pt/ZrO2/C catalyst.
The XPS Pt 4f spectra of all materials were tted using one
component corresponding to Pt0 and two components corre-
sponding to oxidized states (Ptn+). The tted curves are depicted
in Fig. 4. The shis of the Ptn+ components with respect to Pt0
published for Pt/TiO2, and Pt/M–TiO2 (M ¼ V, Cr, and Nb)62 that
showed a lowering of binding energy as compared with Pt/C. On
the other hand, the amount of Ptn+ species is slightly smaller for
the Pt/MOx/C catalysts than for Pt/C.
Cyclic voltammetry of Pt/MOx/C catalysts
are 1 eV and 1.7 eV. These components have been attributed in The CV curves recorded at 50 mV sꢀ1 in 0.5 M H2SO4 solution
the literature to Pt2+ and Pt4+ species, respectively.60
are shown in Fig. 5a, which evidences that curves for Pt/MOx/C
Binding energies, asymmetry parameters (aDS) and the and Pt/C are alike. The electrochemically active area (EAA) of Pt
calculated atomic percentages of Pt0 and Ptn+ species are given was calculated from the charge of oxidation of adsorbed
in Table 1. It is interesting to note that the binding energy of all hydrogen (0.210 mC cmꢀ2), which was obtained by integrating
Pt species does not change for the different transition metal the anodic current up to ꢃ0.35 V, aer accounting for the
oxide supports. These results are in good agreement with data double-layer contribution. The obtained values (ca. 2 cm2,
reported for Pt particles on hybrid C–MOx supports,61 but differ which corresponds to approximately 8.1 cm2 of Pt per cm2 of
from those reported for Pt/TiO2/C and Pt/WO3/C29 and those electrode geometric area) indicate that Pt EAAs for Pt/MOx/C
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2019

View Article Online
Paper
Journal of Materials Chemistry A
Fig. 4 Fitting of the Pt 4f photoemission spectra of Pt/C and Pt/MOx/C
catalysts. Components are Pt0 (violet) and Ptn+ (green and pink).
Table 1 Results obtained from fitting the Pt 4f photoemission spectra.
The values in parentheses correspond to the value obtained for the
asymmetry parameter (aDS)
Pt 4f7/2 BE/eV
and (aDS)
Fig. 5 (a) CV curves recorded for Pt/MOx/C and Pt/C catalysts in 0.5 M
H2SO4 at 50 mV sꢀ1. (b) Pt EAA values calculated from the charge of
hydrogen desorption for all catalysts.
Catalyst
Pt/C
Specie
At%
71.3 (0.2)
72.2
Pt0
75.9
24.1
Ptn+
currents of hydrogen peroxide oxidation were recorded at
different rotation rates for each catalyst (Fig. S4†). A shi of the
ORR polarization curve towards higher potentials is clearly seen
in Fig. 6 for Pt/SnO2/C and Pt/CeO2/C. The behavior of a RDE
with its surface modied by nanoparticles was modelled by
Masa et al.63 that showed that a positive half-wave potential
(E1/2) shi could result not only from a catalytic improvement
(better intrinsic activity) but also from a higher electroactive
surface area. For the Pt/MOx/C catalysts studied in this work,
the Pt EAA is nearly the same (Fig. 5b) and, therefore, the E1/2
shis of ca. 20 and 40 mV observed, respectively, for Pt/SnO2/C
and Pt/CeO2/C as compared with Pt/C evidence enhanced elec-
trocatalytic activity.
73.9
71.2 (0.2)
72.2
73.9
71.2 (0.19)
72.2
73.8
71.1 (0.21)
72.1
73.8
71.1 (0.21)
72.1
Pt/SnO2/C
Pt/TiO2/C
Pt/ZrO2/C
Pt/CeO2/C
Pt/WO3/C
Pt0
83.0
17.0
Ptn+
Pt0
82.7
17.3
Ptn+
Pt0
80.8
19.2
Ptn+
Pt0
82.4
17.6
Ptn+
73.9
71.1 (0.21)
72.1
Pt0
84.4
15.6
Ptn+
74.0
The number of electrons transferred in the potential region
of mixed kinetic-diffusion control can be estimated using the
Koutecky–Levich equation:
catalysts and for Pt/C are similar (within less than 10%), as
shown in Fig. 5b. These results are in good agreement with Pt
particles having similar sizes for all materials, as evidenced by
analysis of HR-TEM images.
1
1
1
¼
þ
(10)
i
ik iL
where i is the measured current, ik the kinetic current and iL is
the diffusion-controlled or limiting current given by the Levich
equation
Oxygen reduction in acid medium
A comparison of ORR polarization curves for Pt/MOx/C and Pt/C
catalysts is shown in Fig. 6. RRO polarization curves and the
iL ¼ 0.62nFAD2/3 ꢀ1/6
n
Cou1/2
(11)
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
As seen in Fig. 6, for all catalysts, the H2O2 yield is smaller
than 1% between 0.9 and 0.7 V and it stays below 5% even in the
potential region where the ORR current is diffusion-controlled.
The number of electrons calculated from the RRDE currents is
shown as a function of potential in Fig. 7, which clearly
evidences that the RRO proceeds predominantly through the 4-
electron pathway.
The kinetic currents were calculated from the measured and
diffusion limited values using eqn (10). In order to compare
intrinsic activities, current densities were obtained by normal-
izing by the Pt EAA. Fig. 8a and b show the ORR kinetic current
densities at 0.90 V and 0.8 V, respectively. At both potentials,
higher values of kinetic current density are obtained for the
ORR on Pt/CeO2/C and Pt/SnO2/C catalysts while the values for
Pt/WO3/C and Pt/C are similar.
Fig. 9 shows the mass-transport corrected Tafel plots for Pt/
CeO2/C and Pt/ZrO2/C, which exhibit slopes of ꢀ60 mV decꢀ1
and ꢀ120 mV decꢀ1 in the regions of low and high currents,
respectively. This slope variation is well known for poly-
crystalline Pt,47 Pt low index surfaces,66,67 and carbon-supported
Pt and Pt alloys.6,68 Although a Tafel slope change would be
expected if the rate-determining step changes, there is
a consensus that this is not the case for the ORR on Pt and the
change is attributed to the potential dependence of the coverage
of adsorbed oxygen-containing species.47,66 Similar Tafel plots
were obtained for other Pt/MOx/C materials (Fig. S6†). Thus,
considering that all Tafel plots are alike, the ORR rate-
determining step is likely to be the same on all catalysts.
Fig. 6 ORR polarization curves measured at the disk electrode in O2-
saturated 0.5 M H2SO4 solution and H2O2 yield. Scan rate: 5 mV sꢀ1
.
Rotation rate: 2500 rpm. Currents normalized by the geometric area of
the disk electrode (0.247 cm2).
where n is the number of electrons transferred in the reaction, A
is the disk electrode area, D is the diffusion coefficient of the
reacting species, n is the kinematic viscosity of the electrolyte,
Co is the bulk concentration of electroactive species, u is the
rotation rate in rad sꢀ1, and F is the Faraday constant. In the
region of mixed kinetic-diffusion control, the so called Kou-
tecky–Levich plots (iꢀ1 against uꢀ1/2) showed parallel lines with
slopes close to the expected value for 4 electrons, in good
agreement with the literature for Pt materials.47 Some Kou-
tecky–Levich plots are shown in Fig. S5.†
Electronic properties
Normalized in situ XAS spectra taken around the Pt L3 edge in
0.5 M H2SO4 solution are shown in Fig. 10a. In general, spectra
show that the intensity of the absorption peak varies for the
different Pt/MOx/C catalysts. In order to compare the electronic
properties on a more quantitative basis, the XANES (X-ray
Absorption Near Edge Structure) region of the spectrum was
tted combining a Lorentzian curve with an arc tangent curve,
as proposed by Shukla et al.,7 to account for transitions to
bound states and to the continuum, respectively. To make
comparisons of the Pt 5d band electronic vacancy, the Lor-
entzian curve tted is then integrated (a larger electronic
vacancy leads to higher absorption and, consequently, larger
The hydrogen peroxide yield (% H2O2) was calculated with
eqn (12).
ꢀ
ꢁ
2ir=N
H2O2ð%Þ ¼
ꢄ 100%
(12)
id þ ir=N
where id is the ORR current measured at the disk, ir is the
current of oxidation of H2O2 measured at the ring electrode, and
N is the RRDE collection efficiency (0.37). Another advantage of
the RRDE as compared with the RDE is that it allows obtaining
the number of electrons (n) as a function of potential directly
from the measured currents at the disk and ring electrodes
using eqn (13):64,65
ꢀ
ꢁ
4id
n ¼
(13)
id þ ir=N
Thus, the peroxide yield and the number of electrons are
related as:64,65
ꢀ
ꢁ
4 ꢀ n
H2O2ð%Þ ¼ 100 ꢄ
(14)
Fig. 7 Number of electrons calculated from RRDE currents measured
in 0.5 M H2SO4 solution as a function of potential.
2
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2019

View Article Online
Paper
Journal of Materials Chemistry A
Fig. 9 Tafel plots for Pt/CeO2/C and Pt/ZrO2/C catalysts. The lines
correspond to ꢀ60 mV decꢀ1 and ꢀ120 mV decꢀ1. Kinetic current
densities obtained by normalizing by the Pt EAA.
absorption (Fig. S7†). Altogether, Fig. 9 reveals that the elec-
tronic vacancy of the Pt 5d band is lower for Pt/CeO2/C and Pt/
SnO2/C and higher for Pt/TiO2/C and Pt/ZrO2/C, as compared to
Pt/C.18,69 Thus, there is charge transfer to the Pt 5d orbital for
SnO2/C and CeO2/C supports, and from the Pt 5d orbital to TiO2/
C and ZrO2/C.
It is noteworthy that direct probe of the Pt 5d band electronic
vacancy by XAS evidences that changes induced by the inter-
action of the Pt particles with the MOx/C supports promote
a shi in the Fermi level of Pt that results in new electronic
states (vacant or lled). Comparison with XPS data discussed
above, however, clearly demonstrates that not always these
electronic changes are enough to promote variations of binding
energies, in agreement with previously published data.61
Electronic effects in oxygen reduction
Data depicted in Fig. 8 demonstrate that the transition metal
oxide incorporated into the support inuences the ORR activity
of Pt nanoparticles, while results shown in Fig. 10 indicate that
there is also an effect on the electronic occupancy of the Pt 5d
band. Several papers have attributed the changes in ORR
activity on Pt-based catalysts, such as PtFe, PtNi and PtCo, to the
variation of electronic properties with the second metal
content.19,24 Although ORR results showed a correlation with the
electronic properties, the variation of composition of the cata-
lysts also involved changes in the lattice parameters that, in
turn, could alter the adsorption of the O2 molecule. For
instance, for a set of Pt alloy catalysts Mukerjee et al.18 observed
a volcano-type dependence of the logarithm of the ORR current
density at 0.90 V with the Pt–Pt distance as well as with the Pt d-
band vacancy of Pt, concluding that the catalysts at the top of
the curve have the best combination of electronic properties
and geometric parameters. For Pt-based catalysts, Toda et al.19
observed a volcano-type variation of the ORR kinetic current
density with the atomic percentage of Fe, Co and Ni, which was
interpreted in terms of changes of the d-band vacancy with
Fig. 8 (a) Kinetic current density of the ORR at the different Pt/MOx/C
and Pt/C catalysts at 0.90 V. (b) Results at 0.80 V.
values of the integral of the Lorentzian curve18,69). The results
obtained from treating the XANES region of spectra collected at
three different applied potentials are depicted in Fig. 10b.
For all catalysts, the growth of the integral of the tted Lor-
entzian curve as the applied potential is raised indicates an
increase in the Pt 5d band vacancy, in agreement with previous
literature reports for carbon supported Pt18,70,71 and Pt alloys.16,72
This electronic vacancy increase with applied potential has been
interpreted as due to the effect of oxygenated species adsorbed
on the Pt surface.18 We must point out that XAS data for the Pt/
WO3/C catalyst were not included in Fig. 10 because the prox-
imity of the W L2 absorption signal prevents treatment. As
observed for other catalysts, the rise of applied potential
promotes an increase in the absorption intensity around the Pt
L3 that evidences an increase in the Pt 5d band vacancy. In
contrast, no signicant changes were seen for the W L2
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
Fig. 11 Kinetic current densities of oxygen reduction at (a) 0.90 V and
(b) 0.80 V against the values of the integral of the Lorentzian curve
adjusted to the in situ XAS spectra measured at those potentials in
0.5 M H2SO4. Larger values of integral of the Lorentzian curve corre-
spond to a higher Pt 5d band electronic vacancy. Pearson's coeffi-
cients of linear fit: 0.976 (0.90 V) and 0.986 (0.80 V).
Fig. 10 (a) Normalized in situ XAS spectra of Pt/MOx/C and Pt/C
catalysts obtained in 0.5 M H2SO4 at 0.90 V. Inset: region of maximum
intensity enlarged. (b) Values of the integral of the Lorentzian curve
adjusted to the XAS spectra measured at different applied potentials.
Pt/ZrO2/C (green), Pt/TiO2/C (light blue), Pt/C (gray), Pt/SnO2/C (blue),
and Pt/CeO2/C (red).
The results reported for Pt/C and Pt–Co/C catalysts showed
a linear correlation between the ORR activities at 0.9 V
determined from measurements in a PEM fuel cell at 80 ꢁC
and the electronic properties assessed from the Pt L3 XANES
region,72 but the particle size and Pt–Pt distance (alloying) in
composition. Keeping in mind that surface as well as electronic the catalysts studied had a simultaneous variation. The
properties are relevant in electrocatalysis,73 the variation of correlation found for Pt/C and Pt–Co/C catalysts evidenced
properties other than the electronic vacancy of the Pt 5d band is that lowering the Pt 5d band electronic vacancy resulted in
likely to be the cause of opposite explanations for ORR higher ORR currents, which were attributed to a reduced OH
enhanced activity, which was attributed, at least partially, to adsorption on Pt sites.72 Also, Ho et al.28 observed a signi-
a higher19,24 and to a lower15,72 Pt d-band vacancy.
cantly enhanced ORR activity for Pt on the Ti0.7Mo0.3O2
Thus, it is pertinent to emphasize again that Pt/MOx/C support at the time that from XAS measurements at the Pt L3
catalysts studied in this work have the same loading of pure Pt edge concluded that this support produced a decrease in the
nanoparticles with similar sizes and shapes, minimizing Pt 5d band vacancy. Thus, it must be highlighted here that
variations of surface properties, and that the Pt EAA is nearly results shown in Fig. 11 show, for the rst time, how the ORR
the same. The kinetic current densities of oxygen reduction at activity on pure Pt nanoparticles depends on the 5d band
0.80 V and 0.90 V show a good linear correlation with the electronic vacancy.
values of the integral of the Lorentzian curve adjusted to the
There are different views regarding the rate-determining step
in situ XAS spectra measured at those potentials, as illustrated of the ORR. Some authors propose that the reaction begins with
in Fig. 11. The change in kinetic current density is about the O2 adsorption (O2 / O2(ads)).47 The rate-determining step
twofold at both potentials, revealing that a decrease in the could be the breaking of the O–O bond. In this case, a strong
electronic vacancy of the Pt 5d band enhances the ORR interaction with the surface would promote a weakening of the
activity.
bond and favor its scission. Sawyer and Seo74 proposed in 1977
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2019

View Article Online
Paper
Journal of Materials Chemistry A
ꢀ
that the rst step in the reduction of oxygen would be the one- indicative of adsorbed O2 in acid solution and thatꢀwas inter-
electron reduction to the superoxide anion
preted as likely due to a rapid protonation of O2 causing
a considerꢀably shorter lifetime. In the present case, the forma-
tion of O2 as cause of the inuence of the d-band electronic
vacancy on the ORR kinetic current density depicted in Fig. 11
seems, therefore, unlikely.
Thus, we should consider simultaneous electron and proton
transfer. In the case of a less vacant Pt 5d band, a weaker
interaction of adsorbed O2 could favor a rate increase of the
reaction. This hypothesis seems compatible with the drop in
activity with the increase in the Pt 5d band vacancy depicted in
Fig. 11.
O2 + eꢀ / O2
(15)
ꢀ
According to the authors' analysis of the experimental
evidence available at that time, the formation of the superoxide
anion would be reversible and independent of the electrode
material. In other words, it would be an outer-sphere electron
transfer to an O2 molecule. A superoxide adsorbed state was also
considered as well as the possibility of an equilibrium of
ꢀ
adsorbed and in solution O2 , particularly in alkaline media.75
Some authors have proposed that the reaction might be initi-
ated by the formation of superoxide (eqn (15)) and it has been
considered that the ORR rate-determining step in the low and
high current density regions would be the rst electron transfer
with formation of superoxide.66 Some evidence of the formation
of superoxide was obtained more than 30 years ago, for
example, by analyzing solutions with an electron spin reso-
nance spectrometer, with the ORR taking place in a ow-type
In summary, our data for Pt/MOx/C catalysts having an
identical loading of similar pure Pt nanoparticles evidence quite
clearly that the ORR takes place with formation of small
amounts of H2O2. Tafel plots are alike suggesting that the
reaction goes mainly through the same pathways on Pt/C and
Pt/MOx/C catalysts. Electronic changes are induced by the
interactions of the metal particles and hybrid supports,
differing only in the transition metal oxide used, and have
a signicant effect on the activity for the ORR on Pt
nanoparticles.
ꢀ
thin layer cell.75 Spectroscopic evidence of adsorbed O2 was
reported later by Shao et al.76 Another possibility oen consid-
ered is the concomitant electron and proton transfer
O2(ads) + H+ + eꢀ / HO2
(16)
Oxygen reduction in alkaline solution
Measurements similar to those described above were also
carried out in 0.1 M KOH solution and RRDE data are shown for
all catalysts in Fig. S8.† The number of electrons transferred
calculated from Koutecky–Levich plots for the different catalysts
was found to be in the range of 3.7 to 4. Some Koutecky–Levich
plots are depicted in Fig. S9.† A comparison of ORR polarization
curves and H2O2 yields is shown in Fig. S10.† Fig. 12 depicts the
kinetically controlled current at 0.90 V calculated with eqn (10).
Comparison of values with those of Fig. 8a evidences that the
catalysts are signicantly more active in alkaline solution than
in acid medium. On the other hand, differently than in an acid
environment, in alkaline solution all Pt/MOx/C catalysts are
The ORR data presented above evidence that the reaction
takes place mainly through the direct 4-electron reduction of O2
to H2O, with the peroxide pathway occurring to a small extent.
In addition, Fig. 10 demonstrates that the ORR activity is higher
for materials with a higher electronic occupancy of the Pt 5d
band. Different O2 adsorption energies were considered the
cause of the different ORR activities of Pt low index surfaces,66
attributing the higher activity to a strong interaction of O2 with
the surface and the resulting lowering of the activation energy
for dissociative adsorption. From that viewpoint, a stronger
interaction of O2 with Pt favoring the O–O bond breaking seems
improbable as the cause of the enhancement of ORR activity on
Pt/MOx/C catalysts exhibiting a higher d-band electronic occu-
pancy. A more likely scenario would be that a less vacant d-band
resulting in an inhibition of OH adsorption would contribute to
keep Pt sites available for O2 adsorption. However, that being
the case, to consider that the bond breaking is the rate-
determining step it would be necessary to assume that the
same changes in the d-band vacancy responsible for the varia-
tions of OH coverage would have only a minor effect on the
strength of the interaction of O2 with Pt. This hypothesis seems
questionable.
ꢀ
Now, let us consider the formation of O2 as the rate-
determining step. First, we can rule out an outer-sphere reac-
tion, which is not inuꢀenced by the electronic properties of the
catalyst. Adsorbed O2 was detected for the ORR on Pt in
alkaline solution by surface-enhanced infrared reection
absorption spectroscopy (SEIRAS).76 The formation of adsorbed
ꢀ
O2 from an adsorbed species was inferred from the depen-
dence on the potential of the band ascribed to the stretching
mode of the O–O bond.76 These authors observed no signal
Fig. 12 Kinetic current densities of the ORR at 0.90 V in 0.1 M KOH.
Kinetic current densities obtained by normalizing by the Pt EAA.
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
more active than Pt/C. The hydrogen peroxide yields in alkaline
solution are somewhat higher than in acid medium. Nonethe-
less, for all catalysts the % H2O2 remains below 3% above
0.80 V, i.e., until the beginning of the diffusion-controlled
region. Consequently, the number of electrons (Fig. S11†) is
very close to 4 in the region of interest for application (between
0.8 and 1.0 V) and higher than 3.75 in the limiting current
region.
As observed in acid solution, the Tafel plots for Pt/C and Pt/
MOx/C catalysts are alike, as illustrated in Fig. 13, but the slopes
are signicantly smaller. The Tafel slope is ꢀ30 mV decꢀ1 in the
region of very low currents and at higher currents it is ꢀ60 mV
decꢀ1. Slopes of ꢀ30 mV decꢀ1 changing to higher ones
(ranging from ꢀ60 to ꢀ80 mV decꢀ1) were reported for other
systems in alkaline solution.47 The results obtained suggest that
the mechanism of the ORR in alkaline solution is the same for
all catalysts although it seems to be different from the one in
acid medium.
Fig. 14 Kinetic current densities of oxygen reduction at 0.90 V against
the values of the integral of the Lorentzian curve adjusted to the in situ
XAS spectra measured at those potentials in 0.1 M KOH. Pearson's
coefficient of linear fit: 0.969.
also leads to an outer-sphere electron transfer mechanism.
According to these authors, for Pt in alkaline media the outer-
sphere charge transfer would take place only in the region of
oxide formation and depending on the coverage of OH both
inner- and outer-sphere mechanisms coexist for the ORR. Based
on theoretical calculations for the ORR in alkaline solutions on
Au and Ag surfaces, Ignaczak et al.78 concluded that the more
The in situ XAS measurements around the Pt L3 edge per-
formed in 0.1 M KOH solution showed that for all Pt/MOx/C
catalysts the d-band vacancy is larger than that for Pt/C (Fig. S12†),
contrasting with data taken in acid solution (Fig. 10b). The trend
observed in alkaline solution is likely due to dissimilarities in the
surface density of OH species adsorbed on the different transition
metal oxides associated with diverse values of zeta potential. In
principle, OH species adsorbed on the oxide surface could favor
a drawing of electronic density from the oxide itself that, in turn,
would promote a charge transfer from the Pt particles to the
oxide. The relationship between the ORR kinetically controlled
current density and the d-band vacancy is linear, as shown in
Fig. 14, and reveals that the catalytic activity gets better as the Pt
5d band vacancy increases.
ꢀ
favorable path for the formation of adsorbed O2 from O2 is
ꢀ
through an outer-sphere charge transfer, and that O2 species
adsorbed and in solution coexist at the equilibrium potential.
However, extending the analysis to Pt(111) these researchers
suggested that the adsorption of O2 (O2ꢀ/ O2(ads)) and forma-
ꢀ
tion of adsorbed O2 (O2(ads) + eꢀ / O2 (ads)) could take place
simultaneously, and that the latter is probably more favorable
than the outer-sphere reaction.78
As discussed above, adsorbed O2ꢀ was detected for the ORR
on Pt in alkaline solution, which appeared to be formed from
adsorbed O2.76 For the ORR in alkaline media, Ramaswamy and
Mukerjee77 suggested that in alkaline media the adsorption of
OH species not only blocks the sites for the adsorption of O2 but
Data depicted in Fig. 14 show an enhancement of ORR
activity as the Pt 5d band vacancy increases. The donation of 2p
electrons from adsorbed O2 (or adsorbed O2ꢀ) to Pt would make
the interaction stronger. As mentioned above, a stronger
interaction would weaken the O–O bond easing its breaking and
resulting in a higher reaction rate.66
Altogether, the results obtained in this work show that
taking advantage of the changes induced by the interactions
between the metal and transition metal oxides incorporated in
supports it is possible to explore the effects of the electronic
properties of pure Pt nanoparticles on the ORR activity. In the
absence of composition and size and/or shape variations, it was
demonstrated that the electrocatalysis of the ORR on Pt parti-
cles is favored by a lower 5d band vacancy in acid solutions and
by a higher one in alkaline medium. The new understandings
shed some light on the fundamental aspects of the electro-
catalysis of the ORR on Pt particles and give new bases for
developing catalysts with enhanced activity for fuel cell
cathodes.
Conclusions
Fig. 13 Tafel plots for Pt/SnO2/C (blue) and Pt/C (black). The lines
correspond to ꢀ30 mV decꢀ1 and ꢀ60 mV decꢀ1. Kinetic current
densities obtained by normalizing by the Pt EAA.
The properties of pure Pt nanoparticles with similar sizes and
shapes were explored taking advantage of the changes induced
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2019

View Article Online
Paper
Journal of Materials Chemistry A
˜
by the interactions between the metal and different transition granted by Coordenaçao de Aperfeiçoamento de Pessoal de
´
metal oxides in the support, in the systematic study of Pt/MOx/C Nıvel Superior (CAPES).
catalysts having the same Pt loading (20 wt%). These materials
were characterized by several techniques. XPS analysis showed
that the binding energy of Pt species does not change for the
References
different transition metal oxide supports. In situ XAS measure-
ments demonstrated that the interactions of the Pt particles and
MOx/C supports induced changes in the Pt 5d band vacancy. In
acid solutions, data show that the ORR takes place predomi-
nately through the direct 4-electron pathway for all Pt/MOx/C
catalysts. Tafel plots are alike for all materials, with slopes of
ꢀ60 mV decꢀ1 and ꢀ120 mV decꢀ1 in the regions of low and
high currents, respectively. The Pt 5d band electronic vacancy
variations have a signicant impact on the ORR activity. The
kinetic currents evaluated at 0.90 and 0.80 V show a linear
decrease as the Pt 5d band vacancy increases. In alkaline solu-
tion, ORR Tafel plots are also alike for all Pt/MOx/C catalysts,
but slopes at low and high currents are, respectively, ꢀ30 mV
decꢀ1 and ꢀ60 mV decꢀ1. The dependence of the ORR kinetic
current is also linear, but in contrast with results for acid
solution, the enhancement of activity follows an increase in the
Pt 5d band vacancy. The effect of band vacancies on the ORR
kinetic current density observed in acid solution is compatible
with simultaneous electron and proton transfer upon oxygen
adsorption being the rate-controlling step. In alkaline solution,
1 U. A. Paulus, A. Wokaun, G. G. Scherer, T. J. Schmidt,
V. Stamenkovic, V. Radmilovic, N. M. Markovic and
P. N. Ross, J. Phys. Chem. B, 2002, 106, 4181–4191.
2 J. R. C. Salgado, E. Antolini and E. R. Gonzalez, J.
Electrochem. Soc., 2004, 151, A2143–A2149.
3 L. Xiong and A. Manthiram, Electrochim. Acta, 2005, 50,
2323–2329.
4 E. I. Santiago, L. C. Varanda and H. M. Villullas, J. Phys.
Chem. C, 2007, 111, 3146–3151.
5 H. Yano, M. Kataoka, H. Yamashita, H. Uchida and
M. Watanabe, Langmuir, 2007, 23, 6438–6445.
6 A. R. Malheiro, J. Perez and H. M. Villullas, J. Electrochem.
Soc., 2009, 156, B51–B58.
7 A. K. Shukla, R. K. Raman, N. A. Choudhury, K. R. Priolkar,
P. R. Sarode, S. Emura and R. Kumashiro, J. Electroanal.
Chem., 2004, 563, 181–190.
8 L. Xiong and A. Manthiram, J. Electrochem. Soc., 2005, 152,
A697–A703.
9 E. Antolini, J. R. C. Salgado and E. R. Gonzalez, J. Electroanal.
Chem., 2005, 580, 145–154.
the linear increase in catalytic activity with the Pt d-band 10 E. A. Carbonio, F. Colmati, E. G. Ciapina, M. E. Pereira and
vacancy can be ascribed to the stronger interaction of adsor-
E. R. Gonzalez, J. Braz. Chem. Soc., 2010, 21, 590–602.
bed O2 (or O2ꢀ) with the surface promoting the breaking of the 11 M. Oezaslan, F. Hasche and P. Strasser, J. Electrochem. Soc.,
O–O bond. Overall, the results presented unveiled the relevance
2012, 159, B444–B454.
of the electronic properties in the electrocatalysis of the ORR on 12 H. Yang, N. Alonso-Vante, J. M. Leger and C. Lamy, J. Phys.
Pt nanoparticles. Chem. B, 2004, 108, 1938–1947.
The results reported contribute not just to increase our 13 R. C. Koffi, C. Coutanceau, E. Garnier, J. M. Leger and
current understanding of the fundamental aspects of the ORR C. Lamy, Electrochim. Acta, 2005, 50, 4117–4127.
but also provide new bases for developing efficient catalysts for 14 F. H. B. Lima, M. J. Giz and E. A. Ticianelli, J. Braz. Chem.
fuel cell cathodes by tuning their electronic properties. In this
Soc., 2005, 16, 328–336.
context, exploring the effects on the Pt d-band vacancy of 15 L. G. R. A. Santos, K. S. Freitas and E. A. Ticianelli,
supports containing other transition metal oxides as well as
other compounds, such as carbides, seems highly desirable.
Electrochim. Acta, 2009, 54, 5246–5251.
16 R. Gentil and H. M. Villullas, J. Solid State Electrochem., 2016,
20, 1119–1129.
17 V. Jalan and E. J. Taylor, J. Electrochem. Soc., 1983, 130, 2299–
2301.
Conflicts of interest
18 S. Mukerjee, S. Srinivasan, M. P. Soriaga and J. McBreen, J.
Electrochem. Soc., 1995, 142, 1409–1422.
19 T. Toda, H. Igarashi, H. Uchida and M. Watanabe, J.
Electrochem. Soc., 1999, 146, 3750–3756.
20 K. Sasaki, L. Zhang and R. R. Adzic, Phys. Chem. Chem. Phys.,
2008, 10, 159–167.
There are no conicts to declare.
Acknowledgements
˜
The authors acknowledge the Brazilian agencies Fundaçao de
`
˜
Amparo a Pesquisa do Estado de Sao Paulo (FAPESP) (2014/ 21 V. Stamenkovic, T. J. Schmidt, P. N. Ross and
12255-6) and Conselho Nacional de Desenvolvimento N. M. Markovic, J. Phys. Chem. B, 2002, 106, 11970–11979.
Cientıco e Tecnologico (CNPq), (407143/2013-0) for nancial 22 V. S. Murthi, R. C. Urian and S. Mukerjee, J. Phys. Chem. B,
support. We also acknowledge the Brazilian Synchrotron Light 2004, 108, 11011–11023.
Laboratory (LNLS) for the access that allowed X-ray absorption 23 Y. J. Wang, N. N. Zhao, B. Z. Fang, H. Li, X. T. T. Bi and
experiments, and the Brazilian Nanotechnology National H. J. Wang, Chem. Rev., 2015, 115, 3433–3467.
Laboratory (LNNano) for XPS measurements and for the use of 24 A. R. Malheiro, J. Perez and H. M. Villullas, J. Power Sources,
the FEI Titan Themis 60-300 microscope. The XAFS1 beamline 2010, 195, 7255–7258.
staff is acknowledged for the assistance during the X-ray 25 P. J. Kulesza, B. Grzybowska, M. A. Malik and
´
´
absorption measurements. FBO acknowledges the fellowship
M. T. Galkowski, J. Electrochem. Soc., 1997, 144, 1911–1917.
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
26 G. C. Abha Bharti, J. Power Sources, 2017, 363, 413–421.
51 A. J. Appleby and M. Savy, J. Electroanal. Chem., 1978, 92, 15–
27 K. Kakinuma, Y. Chino, Y. Senoo, M. Uchida, T. Kamino,
30.
H. Uchida, S. Deki and M. Watanabe, Electrochim. Acta, 52 R. W. Zurilla, R. K. Sen and E. Yeager, J. Electrochem. Soc.,
2013, 110, 316–324. 1978, 125, 1103–1109.
28 V. T. T. Ho, C. J. Pan, J. Rick, W. N. Su and B. J. Hwang, J. Am. 53 S. H. Sun, C. B. Murray, D. Weller, L. Folks and A. Moser,
Chem. Soc., 2011, 133, 11716–11724. Science, 2000, 287, 1989–1992.
29 A. Lewera, L. Timperman, A. Roguska and N. Alonso-Vante, J. 54 S. Doniach and M. Sunjic, J. Phys. C: Solid State Phys., 1970, 3,
Phys. Chem. C, 2011, 115, 20153–20159. 285–291.
30 K. Fugane, T. Mori, D. R. Ou, P. Yan, F. Ye, H. Yoshikawa and 55 S. Hufner, Photoelectron Spectroscopy. Principles and
J. Drennan, Langmuir, 2012, 28, 16692–16700. Applications, Springer Verlag, Berlin, 2003.
31 T. Masuda, H. Fukumitsu, K. Fugane, H. Togasaki, 56 J. W. Keister, J. E. Rowe, J. J. Kolodziej and T. E. Madey, J. Vac.
D. Matsumura, K. Tamura, Y. Nishihata, H. Yoshikawa, Sci. Technol. B, 2000, 18, 2174–2178.
K. Kobayashi, T. Mori and K. Uosaki, J. Phys. Chem. C, 57 S. Hufner, G. K. Wertheim and J. H. Wernick, Solid State
2012, 116, 10098–10102. Commun., 1975, 17, 417–422.
32 D. Takimoto, C. Chauvin and W. Sugimoto, Electrochem. 58 J. McBreen, W. E. Ogrady, K. I. Pandya, R. W. Hoffman and
Commun., 2013, 33, 123–126. D. E. Sayers, Langmuir, 1987, 3, 428–433.
33 B. R. Camacho, C. Morais, M. A. Valenzuela and N. Alonso- 59 B. Ravel and M. Newville, J. Synchrotron Radiat., 2005, 12,
Vante, Catal. Today, 2013, 202, 36–43. 537–541.
34 B. Ruiz-Camacho, M. A. Valenzuela, R. G. Gonzalez-Huerta, 60 L. K. Ono, J. R. Croy, H. Heinrich and B. Roldan Cuenya, J.
¨
¨
K. Suarez-Alcantara, S. E. Canton and F. Pola-Albores, Int.
J. Hydrogen Energy, 2013, 38, 12648–12656.
35 S. J. Tauster, S. C. Fung, R. T. K. Baker and J. A. Horsley,
Science, 1981, 211, 1121–1125.
Phys. Chem. C, 2011, 115, 16856–16866.
61 D. R. M. Godoi, H. M. Villullas, F.-C. Zhu, Y.-X. Jiang,
S.-G. Sun, J. Guo, L. Sun and R. Chen, J. Power Sources,
2016, 311, 81–90.
36 J. H. Kim, S. Chang and Y. T. Kim, Appl. Catal., B, 2014, 158, 62 J. H. Kim, G. Kwon, H. Lim, C. Zhu, H. You and Y. T. Kim, J.
112–118. Power Sources, 2016, 320, 188–195.
37 L. Timperman, A. Lewera, W. Vogel and N. Alonso-Vante, 63 J. Masa, C. Batchelor-McAuley, W. Schuhmann and
Electrochem. Commun., 2010, 12, 1772–1775. R. G. Compton, Nano Res., 2014, 7, 71–78.
38 M. H. Shao, Q. W. Chang, J. P. Dodelet and R. Chenitz, Chem. 64 E. Claude, T. Addou, J. M. Latour and P. Aldebert, J. Appl.
Rev., 2016, 116, 3594–3657. Electrochem., 1998, 28, 57–64.
39 S. Sui, X. Y. Wang, X. T. Zhou, Y. H. Su, S. Riffatc and 65 M. Lefevre and J. P. Dodelet, Electrochim. Acta, 2003, 48,
C. J. Liu, J. Mater. Chem. A, 2017, 5, 1808–1825. 2749–2760.
40 J. Stacy, Y. N. Regmi, B. Leonard and M. H. Fan, Renewable 66 N. M. Markovic, R. R. Adzic, B. D. Cahan and E. B. Yeager, J.
Sustainable Energy Rev., 2017, 69, 401–414. Electroanal. Chem., 1994, 377, 249–259.
41 C. L. Zhang, X. C. Shen, Y. B. Pan and Z. M. Peng, Front. 67 J. Perez, H. M. Villullas and E. R. Gonzalez, J. Electroanal.
Energy, 2017, 11, 268–285. Chem., 1997, 435, 179–187.
42 M. Escudero-Escribano, K. D. Jensen and A. W. Jensen, Curr. 68 U. A. Paulus, T. J. Schmidt, H. A. Gasteiger and R. J. Behm, J.
Opin. Electrochem., 2018, 8, 135–146. Electroanal. Chem., 2001, 495, 134–145.
43 M. Kiani, J. Zhang, Y. Luo, C. P. Jiang, J. L. Fan, G. Wang, 69 S. Mukerjee and J. McBreen, J. Electrochem. Soc., 1999, 146,
´
´
ˇ
J. W. Chen and R. L. Wang, J. Energy Chem., 2018, 27,
600–606.
1124–1139.
70 M. E. Herron, S. E. Doyle, S. Pizzini, K. J. Roberts,
J. Robinson, G. Hards and F. C. Walsh, J. Electroanal.
Chem., 1992, 324, 243–258.
44 Y. Luo and N. Alonso-Vante, Electrochim. Acta, 2015, 179,
108–118.
45 L. Du, Y. Y. Shao, J. M. Sun, G. P. Yin, J. Liu and Y. Wang, 71 B. J. Hwang, Y. W. Tsai, J. F. Lee, P. Borthen and
Nano Energy, 2016, 29, 314–322. H. H. Strehblow, J. Synchrotron Radiat., 2001, 8, 484–486.
46 Y. J. Xue, S. S. Sun, Q. Wang, Z. H. Dong and Z. P. Liu, J. 72 E. Antolini, J. R. C. Salgado, M. J. Giz and E. R. Gonzalez, Int.
Mater. Chem. A, 2018, 6, 10595–10626. J. Hydrogen Energy, 2005, 30, 1213–1220.
47 M. R. Tarasevich, A. Sadkowski and E. Yeager, in 73 A. S. Bandarenka and M. T. M. Koper, J. Catal., 2013, 308, 11–
Comprehensive Treatise of Electrochemistry, ed. B. Conway, 24.
J. M. Bockris, E. Yeager, S. M. Khan and R. White, Springer 74 D. T. Sawyer and E. T. Seo, Inorg. Chem., 1977, 16, 499–501.
US, 1983, ch. 6, pp. 301–398. 75 E. Yeager, Electrochim. Acta, 1984, 29, 1527–1537.
48 A. Damjanovic, M. A. Genshaw and J. O. Bockris, J. Chem. 76 M.-H. Shao, P. Liu and R. R. Adzic, J. Am. Chem. Soc., 2006,
Phys., 1966, 45, 4057–4059. 128, 7408–7409.
49 V. S. Bagotskii, M. R. Tarasevich and V. Y. Filinovskii, Sov. 77 N. Ramaswamy and S. Mukerjee, Adv. Phys. Chem., 2012,
Electrochem., 1969, 5, 1158–1161. 491604, DOI: 10.1155/2012/491604.
50 H. S. Wroblowa, Y. C. Pan and G. Razumney, J. Electroanal. 78 A. Ignaczak, R. Nazmutdinov, A. Goduljan, L. M. D. Pinto,
Chem., 1976, 69, 195–201.
F. Juarez, P. Quaino, E. Santos and W. Schmickler, Nano
Energy, 2016, 26, 558–564.
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2019