Oxygen electrocatalysis on thin porous coating rotating platinum electrodes Joelma Perez, E.R. Gonzalez *, E.A. Ticianelli Instituto de Quı´mica de Sa ˜o Carlos, CP 780, 13560-970 Sa ˜o Carlos, SP, Brazil Received 12 June 1998 Abstract This work discusses the electrocatalysis of the oxygen reduction reaction on platinum on carbon thin porous coating rotating disk electrodes in alkaline and acid media. The electrochemical techniques considered are cyclic voltammetry, steady state polarization and impedance spectroscopy. Both the dc and ac polarization results are analyzed and simulated with the thin film/flooded agglomerate model. Cyclic voltammetry allows an inspection of the crystal facet structure of the Pt particles and the determination of eective surface areas which are lower than those determined from X-ray diraction techniques or transmission electron microscopy. The dc polarization and impedance results show clearly a duplication of the Tafel slope due to structural eects on the porous electrode. Experiments in alkaline solutions show that in this medium there is an important contribution from the carbon to the kinetics of the reaction, which is negligible in acid medium. The eect of Pt particle size on the oxygen reduction electrocatalysis in both electrolytes is correlated with the predominant facets of the platinum crystallites. # 1998 Elsevier Science Ltd. All rights reserved. Keywords: Oxygen reduction; Electrocatalysis; Dispersed catalysts; Thin porous coating electrodes; Electrode modeling 1. List of symbols b Tafel slope of the ORR in the catalyst (V dec 1 ) c * solubility of oxygen (mol cm 3 ) c a dimensionless capacitance for the agglomer- ate network C dl double layer capacitance (F) D a diusion coecient of oxygen in the agglomerate (cm 2 s 1 ) D f diusion coecient of oxygen in the thin film (cm 2 s 1 ) E potential (V) E 0 standard potential (V) F Faraday constant (C/mol) i current density (A cm 2 ) i 0 exchange current density (nFk 0 c * ) (A cm 2 ) i d limiting current density (A) j 1 p J reduced current density (i/i 0 ) k 0 eective rate constant for ORR at reversible potential (cm s 1 ) L y thickness of the agglomerate (cm) n apparent number of electrons transferred R s resistance outside the TPC (solution resist- ance) (O) R lf low frequency resistance (O) R 1 , R 2 dimensionless resistances in the agglomerate network Z impedance in the agglomerate (O) Z electrode total impedance (O) f electrode rotation speed (rpm) Electrochimica Acta 44 (1998) 1329–1339 0013-4686/98/$ - see front matter # 1998 Elsevier Science Ltd. All rights reserved. PII: S0013-4686(98)00255-2 PERGAMON * Corresponding author. Tel.: +55-16-274-9208; Fax: +55- 16-274-9205.
11
Embed
Oxygen electrocatalysis on thin porous coating rotating platinum electrodes
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Oxygen electrocatalysis on thin porous coating rotatingplatinum electrodes
Joelma Perez, E.R. Gonzalez *, E.A. Ticianelli
Instituto de QuõÂmica de SaÄo Carlos, CP 780, 13560-970 SaÄo Carlos, SP, Brazil
Received 12 June 1998
Abstract
This work discusses the electrocatalysis of the oxygen reduction reaction on platinum on carbon thin porous
coating rotating disk electrodes in alkaline and acid media. The electrochemical techniques considered are cyclicvoltammetry, steady state polarization and impedance spectroscopy. Both the dc and ac polarization results areanalyzed and simulated with the thin ®lm/¯ooded agglomerate model. Cyclic voltammetry allows an inspection ofthe crystal facet structure of the Pt particles and the determination of e�ective surface areas which are lower than
those determined from X-ray di�raction techniques or transmission electron microscopy. The dc polarization andimpedance results show clearly a duplication of the Tafel slope due to structural e�ects on the porous electrode.Experiments in alkaline solutions show that in this medium there is an important contribution from the carbon to
the kinetics of the reaction, which is negligible in acid medium. The e�ect of Pt particle size on the oxygen reductionelectrocatalysis in both electrolytes is correlated with the predominant facets of the platinum crystallites. # 1998Elsevier Science Ltd. All rights reserved.
Z overpotential (E 0ÿE)m reduced overpotential (Z/b)f dc potential-independent agglomerate di�u-
sion parameter (k 0Ly/Da)1/2
u kinematic viscosity (cm2 sÿ1)C ac potential dependent agglomerate di�u-
sion parametery angular frequency (rad sÿ1)oa characteristic agglomerate frequency (rad
sÿ1)of characteristic thin-®lm frequency (rad sÿ1)
2. Introduction
Dispersed catalysts of the type used in gas di�usionelectrodes have become key materials in electrochemi-cal technology. In spite of the work carried out to
date, it is well known that there are several di�cultiesin interpreting experimental results, both dc and ac, ongas di�usion electrodes. To overcome those di�cultiesseveral theoretical models have been considered, one of
the most successful being the thin ®lm/¯ooded agglom-erate model [1]. The thin porous coating rotating diskelectrode (TPC/RDE) technique has proved to be very
useful to study and evaluate dispersed catalysts of thetype used in gas di�usion electrodes for fuel cells [2±4].The advantage of the TPC/RDE is that it is an exper-
imental setup conceptually similar to a ¯oodedagglomerate, given by the electrode structure, in con-tact with a thin ®lm determined by the rotation of the
electrode.This work presents studies on the electrocatalysis of
the oxygen reduction reaction (ORR) on platinum oncarbon TPC/RDE in alkaline and acid media. The ma-
terials studied are dispersed platinum on carbon (Pt/C)with di�erent values of the ratio Pt/C. The electroche-mical techniques considered are cyclic voltammetry,
steady state polarization and impedance spectroscopy.Both the dc and ac results are analyzed and simulatedwith the thin ®lm/¯ooded agglomerate model.
3. Experimental
The rotating electrode was made by adapting aPTFE cylinder with a 0.5 cm diameter cavity to a
PINE rotator system. A graphite rod was forced intothe cavity leaving a recess of 0.015 cm which was ®lledwith the catalyst material. This was prepared by mix-
ing the catalyst powders (Vulcan XC-72 carbon, 10±80w/w Pt/carbon and pure platinum (E-TEK, USA))with a dilute suspension (12% w/w) of a Te¯on emul-
sion (DuPont TM30). A conventional one compart-
ment electrochemical glass cell with a Luggin capillary
was used in the electrochemical experiments. A large
area platinized platinum foil served as the counter elec-
trode and a saturated calomel electrode (SCE) or a re-
versible hydrogen reference electrode (RHE) were
employed in the alkaline and acid electrolytes, respect-
ively. Except when mentioned, all potentials are
Fig. 1. Cyclic voltammograms of TPC electrodes prepared
with di�erent Pt/C catalysts in N2-saturated 1.0 M NaOH sol-
utions. v=10 mV sÿ1. Voltammograms were recorded after
cycling the electrode until a steady state was reached.
J. Perez et al. / Electrochimica Acta 44 (1998) 1329±13391330
referred to the RHE. All the experiments were carried
out in 1.0 M NaOH or 0.5 M H2SO4 solutions, pre-
pared from high purity reagents (Merck) and water
distilled and puri®ed in a Milll-Q (Millipore) system.
The electrolytes were saturated with puri®ed N2 or O2
gases. All the experiments were conducted at room
temperature (0258C).Cyclic voltammetry was performed with a PARC
Model 273 potentiostat. The ac impedance and rotat-
ing disk polarization measurements were carried outwith a Solartron 1286/1250 impedance system. The
electrode rotation speed was controlled by aAFMSRE/ASR Pine Instrument system. The oxygenreduction polarization potentials for the ac impedance
measurements were set point-by-point in the potentio-static mode. The ac perturbation was a 5 mV peak-to-peak sinusoidal signal, with the frequency varying
from 0.005 Hz to 10 kHz. In all cases, the dc steadystate currents at each electrode potential were recordedprior to the ac measurement.
Transmission electron microscopy (TEM) analysesof the Pt/C catalysts were conducted using a ZeizzEM912 equipment. X-ray di�raction (XRD) analysiswere made employing a X-URD, Carl Zeiss X-ray sys-
tem. The crystallite size of the Pt particles was calcu-lated from the full width at half-maximum of the (111)peak at 2y=39.88 [5]. Calculation of the Pt area from
these data was made using the standardprocedure [5, 6].
4. Results
4.1. Cyclic voltammetry
Cyclic voltammetry on the TPC/RDE allows the de-
termination of the e�ective electrochemical surfaceareas of the electrodes and an insight on the crystallinestructure of the catalysts. Figs. 1 and 2 present some
of the cyclic voltammograms obtained for the di�erentPt/C electrocatalysts in the alkaline and acid electro-lytes respectively. In these plots the currents were
divided by the mass of platinum in the TPC layer inorder to allow an easy comparison of the results. Avery good resolution of the platinum features is foundin the cyclic voltammograms. The active surface areas
per gram of platinum were evaluated using the hydro-gen adsorption/desorption peaks [7] and the results arepresented in Table 1, which also includes the mean
values obtained with TEM and XRD.
4.2. Steady state polarization
Fig. 3 shows the steady state oxygen reductionpolarization curves for several electrode rotation ratesobtained in 1.0 M NaOH solution, on the electrode
prepared with the 20 wt% Pt/C catalyst. The expectedlinear behavior of the di�usional limiting currents as afunction of the square root of the electrode rotation [8]
was observed for all catalysts and electrolytes, exceptfor pure carbon in acid media which is not active forthe ORR.
In order to understand the characteristics of the dcpolarization responses, the results were analyzed interms of the thin ®lm-¯ooded agglomerate model [1]
Fig. 2. Cyclic voltammograms of TPC electrodes prepared
with di�erent Pt/C catalysts in N2-saturated 0.5 M H2SO4 sol-
utions. v=10 mV sÿ1. Voltammograms were recorded after
cycling the electrode until a steady state was reached.
J. Perez et al. / Electrochimica Acta 44 (1998) 1329±1339 1331
from which it is derived that the reduced current den-
sity (J) passing from the agglomerate/thin ®lm into the
solution is given by the equation:
J � i=i0
� exp�m��tanh�f exp�m=2���=�f exp�m=2��1� G exp�m��tanh�f exp�m=2���=�f exp�m=2�� , �1�
where i is the actual current density and i0 is the
exchange current density. This expression contains par-ameters related to the current generation and the struc-
tural parameters of the system: the reduced
overpotential (m), the thickness (d) and di�usion of
oxygen in the thin ®lm (G), and the thickness (Ly) anddi�usion of oxygen in the agglomerate (f).Substitution of the expressions for m, f and G in
Eq. (1) gives the theoretical dependence of the current
density as a function of the electrode potential and the
several physicochemical and electrochemical par-
ameters of the system: Da, Df , Ly, d, c*, i0, b and n.
Parametrization of the theoretical equation was per-
formed using the depth value of the electrode cavity
(Ly=0.015 cm) and tabulated values for the di�usion
coe�cient and solubility of oxygen in the electrolytes
(here it was assumed Da=Df ).
The details of the entire ®tting procedure were pre-
sented elsewhere [3, 9]. Brie¯y, the method involved the
following steps: (i) Starting with standard values for
the other model parameters, i0 and b are adjusted with
the data at low overpotentials where the e�ects of fand G are negligible [4]. This is made taking into
account that the Tafel slope for the ORR is dependent
on the coverage of oxygen on the platinum surface (see
below). (ii) The values of n and d are obtained with
the data at high overpotentials where the dc polariz-
ation curve reaches the limiting current. The procedure
involved the iteration with the RDE convective-di�u-
sion equation [8, 10]
d � 0:62nFD2=3f uÿ1=6f 1=2c* �2�
until convergence of the n and d values. (iii) With the
values of i0, b, n and d already adjusted, the e�ect of
the agglomerate is considered (with Ly=0.015 cm,
Da3Df ) at intermediate overpotentials. (iv) Steps i to
Fig. 3. Steady-state polarization curves for di�erent electrode
rotation rates for oxygen reduction on 20% Pt/C TPC/RDE
in 1.0 M NaOH solution. T=258C.
Fig. 4. Experimental (symbols) and ®tted (Eq. (1), solid lines)
Tafel plots for oxygen reduction on di�erent Pt/C catalysts in
1.0 M NaOH solution f=2500 rpm.
Fig. 5. Experimental (symbols) and ®tted (Eq. (1), solid lines)
Tafel plots for oxygen reduction on di�erent Pt/C catalysts in
0.5 M H2SO4 solution. f=2500 rpm.
Table 1
Average particle size obtained from transmission electron mi-
croscopy and X-ray di�raction techniques. Active platinum
surface areas obtained from TEM/XRD and cyclic voltam-
metric (CV) experiments in alkaline and acid solutions. The
thickness of the TPC layer is 0.015 cm
Pt/C (%) Particle
size (nm)
Active area m2/g Pt
TEM/XRD CV/NaOH CV/H2SO4
10 2.4 114 45 65
20 2.8 100 52 58
30 3.8 74 29 31
40 4.5 62 28 30
60 9 31 13 15
80 16.5 17 7 7
J. Perez et al. / Electrochimica Acta 44 (1998) 1329±13391332
iii are repeated starting with the latest set of ®tting
parameters until convergence of all values.Figs. 4 and 5 illustrate the e�ect of the catalyst com-
position ratio on the theoretical and experimental
polarization curves for a constant rotation speed of theelectrode ( f=2500 rpm) for both alkaline and acidelectrolytes. Tables 2 and 3 present the values of i0, band n resulting from the ®ttings in both electrolytes. In
the alkaline medium only one set of i0 and b values isfound for all electrode potentials for each catalyst,while in the acid medium the values of i0 and b were
di�erent in the regions of low and high overpotentials.
4.3. Impedance measurements
The interpretation of the shape and magnitude ofthe impedance responses of the porous electrode with
respect to the ORR as a function of potential wasmade following the expressions derived for the thin®lm/¯ooded agglomerate model [1]
p �1� Gem�C2=f2em��tanh C=C��tanh �����������������
j�o=o f�p
=� �����������������j�o=o f�
p �
��ÿ1, �4�
o f � Df=d2, o a � Da=L
2y, �5�
ca � 1:216
�fem=2�2�1� 0:56�fem=2�2�1=2 , �6�
R1 � fem=2
tanh fem=2, �7�
R2 ��1
2
�sec h2fem=2 � 1
R1
�ÿ1�ÿ R1, �8�
and
C � �f2em � j�o=o a��1=2: �9�It should be emphasized that, while in the steady
state TPC/RDE measurements only the oxygen con-centration at the boundary of the porous electrode andthe external electrolyte can be controlled with the ro-
tation speed, in the ac experiment the concentrationinside the porous structure can be modulated depend-ing on the ac frequency.
The ®tting procedure for the ac method followedessentially the same guidelines as for the dc case. Thevalue of the solution resistance is obtained from thehigh frequency region of the impedance spectra and the
capacity of the double layer by adjusting the frequen-cies used in the ®tting with those used experimentally.Examples of the ®ttings of Eqs. (3)±(9) to the exper-
imental ac impedance results of the ORR on the TPC/RDE made with 40 wt% Pt/C are presented in Figs. 6
and 7 for several electrode potentials in the alkalineand acid electrolytes, respectively. In all cases the elec-
Table 2
Kinetic parameters obtained from the ®ttings of the dc results
with Eq. (1) for oxygen reduction on the various Pt/C cata-
lysts in 1.0 M NaOH at 258C
Pt/C i0(A cmÿ2)
b
(V decÿ1)
n
0 0.9�10ÿ13 0.036 1.8
10 1.2�10ÿ9 0.045 2.8
20 1.5�10ÿ8 0.05 3.6
30 5.5�10ÿ8 0.05 3.7
40 8.0�10ÿ8 0.056 3.8
60 1.1�10ÿ7 0.057 3.8
80 1.5�10ÿ7 0.057 3.8
100 8.2�10ÿ7 0.06 4
Ly=0.015 cm; d=0.0011 cm; Da=Df =1.65� 10ÿ5 cm2
sÿ1, c *=1.1�10ÿ5 g mol cmÿ3.
Table 3
Kinetic parameters obtained from the ®ttings of the dc results
with Eq. (1) for oxygen reduction on several Pt/C catalysts in
0.5 M H2SO4 at 258C
Pt/C (%) Low current density High current density n
J. Perez et al. / Electrochimica Acta 44 (1998) 1329±1339 1333
trode was rotating at 2500 rpm. A summary of thevalues of i0, b, n and C dl resulting from the analyses onboth electrolytes is presented in Tables 4 and 5.
In Figs. 6 and 7 it is observed that the agreement
between the experimental results and the theoretical
lines is quite satisfactory, specially at the higher over-
Fig. 7. Impedance spectra at several potentials for oxygen re-
duction reaction on 40% Pt/C in 0.5 M H2SO4: (a) (.) 0.900,(w) 0.875, (Q) 0.850, (q) 0.825 and (R) 0.800 V vs. RHE; (b)
(r) 0.750; (R) 0.725; (w) 0.700; (.) 0.675 and (q) 0.650 V vs.
RHE. The frequency range is 0.01 Hz (®rst low frequency
point) to 10 kHz (10 points/dec). (± � ±) ®tted (Eqs. (3) and
(4)) impedance spectra. f=2500 rpm, T=258C.
Table 5
Kinetic parameters and double layer capacity obtained from
the ®ttings of ac results with Eqs. (3) and (4) for oxygen re-
duction on several Pt/C catalysts in 0.5 M H2SO4 at 258C
Pt/c
(%)
Low current density High current density n C dl
i0(A cmÿ2)
b
(V decÿ1)
i0(A cmÿ2)
b
(V decÿ1)
10 2.6�10ÿ10 0.056 1.1� 10ÿ5 0.120 4 0.016
20 3.8�10ÿ10 0.057 3.5� 10ÿ5 0.120 4 0.028
30 1.6�10ÿ9 0.057 1.0� 10ÿ5 0.120 4 0.031
40 1.7�10ÿ9 0.057 2.6� 10ÿ5 0.120 4 0.045
60 5.6�10ÿ9 0.060 2.8� 10ÿ5 0.120 4 0.060
80 2.6�10ÿ9 0.060 2.7� 10ÿ5 0.120 4 0.060
100 5.0�10ÿ9 0.060 2.6� 10ÿ5 0.120 4 0.070
Ly=0.015 cm; d=0.0010 cm; Da=Df =1.80� 10ÿ5 cm2
sÿ1, c *=1.1� 10ÿ5 g mol cmÿ3, Rs=2 O.
Table 4
Kinetic parameters and double layer capacity obtained from
the ®ttings of the ac results with Eqs. (3) and (4) for oxygen re-
duction on the various Pt/C catalysts in 1.0 M NaOH at 258C
Pt/C
(%)
i0(A cmÿ2)
b
(V decÿ1)
n C dl
(F)
0 1.4� 10ÿ13 0.036 1.8 0.009
10 1.3� 10ÿ9 0.045 2.8 0.023
20 1.3� 10ÿ8 0.05 3.6 0.040
30 1.7� 10ÿ8 0.05 3.7 0.050
40 1.4� 10ÿ8 0.057 3.8 0.056
60 2.3� 10ÿ7 0.06 4 0.068
80 2.3� 10ÿ7 0.06 4 0.071
100 2.4� 10ÿ7 0.06 4 0.120
Ly=0.015 cm; d=0.0011 cm; Da=Df =1.65� 10ÿ5 cm2
sÿ1, c *=1.1�10ÿ5 g mol cmÿ1, Rs=4 O.
Fig. 6. Impedance spectra at several potentials for oxygen re-
duction on 40% Pt/C in 1.0 M NaOH: (a) (.) 1.020, (w)
0.995, (Q) 0.970, (q) 0.945 and (R) 0.920 V vs. RHE; (b) (W)
0.870, (r) 0.845, (.) 0.820, (w) 0.795 and (Q) 0.770 V vs.
RHE. Frequency range is 0.01 Hz (®rst low frequency point)
to 5 kHz (15 points/dec). (± � ±) ®tted (Eqs. (3) and (4))
impedance spectra. f=2500 rpm, T=258C.
J. Perez et al. / Electrochimica Acta 44 (1998) 1329±13391334
potentials. The higher deviations observed at low fre-
quencies and low overpotentials can be explained by
the presence of some drift in the experimental responsecaused by the presence of impurities in the high surface
area carbon substrate. At low overpotentials thephenomenon should be more apparent because the far-
adaic currents for oxygen reduction are much smallerthan the capacitive currents involved in double layer
charging. In Figs. 6 and 7 it is also seen that the mag-nitude of both the theoretical and the experimental
arcs ®rst decreases and then increases with the decrease
of the electrode potential. When the kinetics of theORR is the only factor considered previous studies [2±
4] have shown that the diameter of the impedance arcsdecreases exponentially with the decrease of the elec-
trode potential. However, for higher overpotentials thesize of the arc increases and those same studies have
shown that this is due to the di�usion of oxygen in thethin ®lm.
The impedance and the steady state polarization data
were compared in terms of mass transport in the exter-nal electrolyte corrected Tafel diagrams. The exper-
imental dc data were plotted as E vs. log[(i� id)/(idÿ i)], where id is the convective-di�usion limiting cur-
rent density [8]. The theoretical diagrams corrected forthe thin ®lm di�usion e�ect by taking G 4 0, were
obtained by plotting E vs. log i (Eq. (1), where
m=(E 0ÿE)/b) and i= i0J) for the dc case and E vs.log(1/R lf) (Eq. (4)) for the ac case, where R lf (low fre-
quency resistance) corresponds to the diameter of theimpedance arc, that is the value of Z when o 4 0.
Examples of these results are presented in Figs. 8 and 9.
The speci®c e�ect of the Pt particle size on the elec-trocatalysis of ORR at low overpotentials was ana-
lyzed in terms of the values of 1/R lf calculated for aconstant electrode potential (E=0.9 V vs. RHE in
acid media and E=1.02 V vs. RHE in alkalinemedia), maintaining the electrochemical parameters i0,
b, n and C dl the same as those used in the ®ttings, andconsidering two situations: (1) G(d) and f(Ly) tendingto zero; (2) G(d) tending to zero, while maintainingf(Ly) the same as that used in the ®ttings. Figs. 10±12
present the results of the Pt activity as a function ofparticle size. Figs. 10 and 12 present the speci®c ac-tivity (1/R lf per unit surface area of Pt catalyst,
obtained by TEM/XRD) for the acid and alkalinemedia, respectively. Fig. 11 presents the mass activity(1/R lf per unit mass of Pt catalyst) for the alkaline
medium.
5. Discussion
From the results in Table 1 it is observed that the cv
surface area per gram of platinum presents a tendency
Fig. 8. Mass-transport corrected Tafel plots for oxygen re-
duction on 40% Pt/C in 1.0 M NaOH. (± � ±) log(1/R lf) vs.
E; (w) log[(Id� I)/(IdÿI)] vs. E; (full line) ®tted plot using
Eq. (1) with G 4 0. f=2500 rpm.
Fig. 9. Mass-transport corrected Tafel plots for oxygen re-
duction on 40% Pt/C in 0.5 M H2SO4. (± � ±) log(1/R lf) vs.
E; (q) log[(Id� I)/(IdÿI)] vs. E; (full line) ®tted plot using
Eq. (1) with G4 0. f=2500 rpm.
Fig. 10. Dependence of the speci®c activity for oxygen re-
duction on Pt particle size in 0.5 M H2SO4 solution: (.)Ly=0.0 cm and (r) Ly=0.015 cm.
J. Perez et al. / Electrochimica Acta 44 (1998) 1329±1339 1335
to decrease with the increase of the Pt/C ratio. This
e�ect can be attributed to an increase of the Pt particle
size with the increase of Pt/C ratio, as observed from
the TEM and XRD measurements. It is also seen that
the speci®c areas obtained from cv in alkaline or acid
media are smaller than those determined with the
other techniques. The smaller values of the areas
obtained by cv compared with TEM/XRD, should be
explained mainly by a low catalyst utilization and/or
the wetting characteristics related to the presence of
PTFE in the TPC layer. Other contributing factors
may be: (i) the presence of impurities in the Pt/C par-
ticles, leading to a partial blocking of the hydrogen
adsorption/desorption process in the Pt catalyst, (ii)
the fact that a polycrystalline structure equivalent to a
smooth surface (which correspond to a hydrogen
adsorption/desorption charge of 0.210 mC cmÿ2) is
assumed for the Pt crystallites (see below) and (iii) the
fact that in a TEM or XRD calculations a spherical
structure is assumed for the Pt particles and that the
total area of Pt loaded on the electrode is considered
in the calculation, therefore including the area partially
blocked as a consequence of anchoring the Pt on the
carbon substrate and the area of Pt which is not in
electrical contact with the current collector.
From the voltammograms in Figs. 1 and 2 it is seen
that in both, the acid and alkaline electrolytes, there
are some di�erences on the features of the hydrogen
desorption process, when the responses of the low and
high Pt/C ratio catalysts are compared. In the sulfuric
acid solution the middle hydrogen oxidation peak, cor-
responding to the desorption of the hydrogen atoms
from the edge and corner sites of the platinum
crystallites [5, 11], is particularly pronounced for the
smaller platinum particles. This phenomenon was also
observed in other works using the same electrolyte and
has been attributed to the fact that small platinum par-
ticles show an increased amount of edge and corner
sites in the crystals, which is a consequence of the
decrease of the atomic surface fraction (ASF) of the
(111) and (100) faces [5, 11, 12].
In the results of Figs. 1 and 2 it is seen that the
background current (corresponding to the desorption
of hydrogen from the (111) face [13, 14], over which
the hydrogen desorption peaks on the (110) and (100)
sites are superimposed, is particularly large for small
particles in the acid electrolyte. This is consistent with
the fact that for the smaller particles (2±4 nm), the
ASF of Pt atoms in the (111) face is higher than those
for the other orientations [11].
In both, alkaline and acid electrolytes, the dimin-
ution of particle size 16±2 nm (Table 1) leads to a
decrease of the more anodic hydrogen desorption peak
which corresponds to the desorption of hydrogen
atoms from the (100) face [11, 13]. As pointed above,
this is a consequence of a decrease in the ASF of Pt
atoms in that face with the decrease of particle size.
The ®rst peak in the voltammograms includes the
contribution of hydrogen desorption from the (110)
face. The (110) sites are likely to form on particles that
do not contain the exact number of atoms to produce
a complete cubo-octahedral structure, which was pro-
posed by Kinoshita [11] as a model of small platinum
crystallites. These sites correspond to a maximum
mass-average fraction of 15% in Pt particles with a di-
ameter of about 2 nm [11, 12], decreases for larger par-
ticles and becomes almost zero for particles with
diameter above ca. 8 nm [11, 12]. However, the results
in Figs. 1 and 2 show that the e�ect of the (110) face
predominates and the ®rst hydrogen desorption peak
increases as a function of particle size. This fact indi-
cates that a reconstruction process takes place during
the potential cycling, as observed for single
crystals [13, 14] when the upper potential limit is raised
above10.8 V vs. RHE.
Fig. 11. (Q) Dependence of the mass activity for oxygen re-
duction on Pt particle size in 1.0 M NaOH solution. (- - -)
AVF (111) as a function of particle size from Ref. [11].
Fig. 12. (Q) Dependence of the speci®c activity for oxygen re-
duction on Pt particle size in 1.0 M NaOH solution. (- - -)
ASF (111) as a function of particle size from Ref. [11], solid
line: mass activity (solid line, Fig. 11) divided by the Pt
speci®c area.
J. Perez et al. / Electrochimica Acta 44 (1998) 1329±13391336
Tables 2±5 show that the values of i0, b and n
obtained from either the dc or the ac simulations aretotally consistent with each other. In Tables 2±5 it isalso seen that the C dl, and i0 values increase with the
increase of the Pt/C weight ratio in both electrolytes.This is due to the increase in active area in the electro-des resulting from the increase of Pt loading on the
thin porous coating layer with the increase of Pt/Cratio.
It is well known that the ORR can proceed via a 4-electron pathway to give water as the ®nal product ora 2-electron pathway to form peroxide ions, depending
on the electrode material [15±18]. Smooth electrodesformed by polycrystalline gold, graphite and carbon
catalyze the reaction through the two electronmechanism [19±29], while gold single crystal withorientation (100), silver, platinum and several metals
of the platinum group promote the four electronpathway [15±18, 20, 30±35].For smooth polycrystalline platinum electrodes, two
Tafel regions, with slopes of 60 and 120 mV for lowand high overpotentials, respectively, have been
observed for the ORR in acid or alkaline solutions.The existence of two slopes is explained in terms of thecoverage of the electrode surface by adsorbed oxygen,
which follows a Temkin isotherm at low overpotentialsand a Langmuir isotherm at higher overpotentials [31±33]. It is interesting to note that the same character-
istics of the Tafel slope have been found for platinumsingle crystals in sulfuric acid [14].
Tables 2 and 4 show that in alkaline solution theTafel slopes obtained from the ®ttings (Eqs. (1) and(4)) for the several Pt/C catalysts increase from 36 to
60 mV decÿ1 as the Pt weight ratio increases from 0to 100% It is also seen that the number of electrons
involved in the ORR changes from a value close totwo to a value equal to four. These results show thatin this medium there is an important contribution
from the carbon to the kinetics of the reaction, onwhich the ORR takes place by a two electron re-duction process, even in the presence of small amounts
of Pt. In contrast, the results in Tables 3 and 5 showthat in acid medium, where the contribution from the
carbon to the ORR is negligible, the simulated valuesof b and n are practically the same for all catalysts andpresent the expected values for single crystal or poly-
crystalline platinum.In Figs. 8 and 9 it is observed that both the dc and
the ac results lead to Tafel plots that show a doubling
of the slope in alkaline medium and a duplication ®rstand then a quadruplication of the slope in acid med-
ium. The duplication phenomenon takes place at po-tentials which are not compatible with those alreadyobserved many times on smooth platinum surfaces and
explained on the basis of the oxidation state of thesurface [31±33]. Simulation of the Tafel plots with the
theoretical model shows clearly that the duplication in
alkaline medium and the ®rst duplication in acid sol-ution are due to structural e�ects in the porous elec-trode especially associated with the thickness of the
¯ooded agglomerate zones (Ly).On the basis of the above picture it is seen that the
structural characteristics of the TPC layer are respon-sible for a strong e�ect on the polarization behavior ofthe electrode. To compensate for this e�ect, the value
of R lf was calculated at low overpotentials, where oxy-gen adsorption follows the Temkin isotherm, taking Gand f tending to zero for all catalysts. In this way, the
calculated values of 1/R lf are equivalent to those of alarge area ¯at plate electrode composed of Pt/C with-
out di�usion limitations or structural contributions.Thus, the value of 1/R lf corrected in this way, which isproportional to the current generated by the electro-
chemical reaction, is very adequate to discuss thespeci®c e�ect of the Pt particle size on the electrocata-lysis of the ORR.
This problem has been treated by several authors,for the reaction occurring on gas di�usion electrodes
in acid media, under several di�erent experimentalconditions [5, 11±13, 36]. The analyses are usually madein terms of the speci®c activity (current at 0.9 V vs.
RHE divided by the platinum active area) and themass activity (current at E=0.9 V vs. RHE dividedby the platinum loading). In general there is some dis-
agreement on the results and conclusions, and thecauses were recently analyzed by Tamizhmani et al. [5]
and Markovic et al. [13], who attributed the discrepan-cies to the anion adsorption characteristics of thedi�erent acid electrolytes used in di�erent studies. On
the other hand, it is highly probable that in some ofthese works the reason for the discrepancy is related tothe uncompensated e�ects introduced by the electrode
structure, as those mentioned above for the TPC sys-tem.
The results in Fig. 10 show that in the sulfuric acidelectrolyte the speci®c activity of the platinum par-ticles, calculated without compensation for the struc-
tural e�ects of the active layer, is practicallyindependent of Pt particle size. On the other hand the
values obtained after compensation for the structurale�ects present a marked dependence on particle size.These results show that, at least for the TPC electro-
des, the structural e�ects are very important in deter-mining the activity pro®les.The behavior of the structurally corrected speci®c
activity (Fig. 10) as a function of the Pt particle size insulfuric acid has also been studied by other authors
using di�erent kinds of active layers [5, 13, 36]. Theshape of the speci®c activity curve has been explainedin terms of the shape of the curve representing the
ASF made up of (100) facets as a function of particlesize [11, 13], because the pro®les of both curves present
J. Perez et al. / Electrochimica Acta 44 (1998) 1329±1339 1337
exactly the same tendency. This proposal has been
made under the assumption that only the (100) and
(111) facets are present in the Pt crystallites [11] and
that only the atoms on the (100) orientation are active
for the ORR [13]. However, the results of the present
work clearly show that the (110) facet is also present
and thus its contribution to the ORR
electrocatalysis [13, 14] can not be disregarded.
Figs. 11 and 12 present the structurally corrected
mass and speci®c activities, respectively, in the NaOH
electrolyte plotted as a function of Pt particle size. In
this electrolyte the (111) face is the more active for the
ORR [13]. So, the mass activity should present the
same pro®le as the atomic volume fraction (AVF)
made up of (111) facets, as represented by the dashed
line in Fig. 11. Analogously, the speci®c activity should
present the same pro®le as that of the ASF made up
of (111) facets (dashed line in Fig. 12).
In Fig. 11 it is seen that, although the shape of the
mass activity and the AVF lines are similar, the pos-
itions of the maxima do not coincide. It is observed
that the maximum of the AVF appears for particle
sizes smaller than those employed in the experimental
investigation. Based on this fact it is expected that, in
the range of 2.5±5.0 nm, the trend on the experimental
curve would be a monotonic decrease of the mass ac-
tivity with the increase in particle size. The deviation
from this behavior can be attributed to the fact that
for the smaller Pt/C ratio catalysts (smaller Pt particle
size) there is a strong participation of the carbon in
the catalysis of the ORR (a 2eÿ process), which is
apparent until the Pt/C ratio reaches 30/40 wt% (3.5/
4.5 nm). Thus the increase in the mass activity in the
region of small particle sizes, can be understood as an
increase of the platinum contribution to the catalysis
of ORR (a 4eÿ process). Beyond 5 nm, the decay in
the mass activity is strongly governed by the decrease
of the AVF of the Pt crystallite (111) facets.
The comparison of the speci®c activity and the ASF
lines in Fig. 12 is more complex, because of the dis-
persion in the experimental data. To facilitate the
analysis, the speci®c activities of the particles were also
calculated from the mass activity line (solid line in
Fig. 11), taking the speci®c activity as the mass activity
divided by the Pt speci®c area (m2 gÿ1). This result
corresponds to the solid line in Fig. 12. It is seen that
there is a good agreement between the two sets of ex-
perimental data, indicating that the pro®le is indeed
realistic, although it does not follow the same trend as
the ASF lines. As observed from cyclic voltammetry
and the ORR polarization data, the causes for the dis-
crepancy are probably related to the reconstruction of
the crystallite facets and/or to the contribution of the
carbon particles to the catalysis of the ORR.
Acknowledgements
The authors wish to thank Fundac° aÄ o de Amparo aÁ
Pesquisa do Estado de SaÄ o Paulo (FAPESP), ConselhoNacional de Desenvolvimento Cientõ fõ co e Tecnolo gico
(CNPq), Fundac° aÄ o Coordenac° aÄ o de Aperfeic° oamentode Pessoal de NõÂ vel Superior (CAPES) andFinanciadora de Estudos e Projetos (FINEP), Brazil,
for ®nancial assistance.
References
[1] T.E. Springer, I.D. Raistrick, J. Electrochem. Soc. 136
(1989) 1594.
[2] A.A. Tanaka, C. Fierro, D. Scherson, E.B. Yeager, J.
Phys. Chem. 91 (1987) 379.
[3] J. Perez, A.A. Tanaka, E.R. Gonzalez, E.A. Ticianelli, J.
Electrochem. Soc. 141 (1994) 431.
[4] M.L. Calegaro, J. Perez, A.A. Tanaka, E.A. Ticianelli,
E.R. Gonzalez, Denki Kagaku 6 (1996) 436.
[5] G. Tamizhmani, J.P. Dodelet, D. Guay, J. Electrochem.
Soc. 143 (1996) 18.
[6] E.A. Ticianelli, J.G. Beery, S. Srinivasan, J. Appl.
Electrochem. 21 (1991) 597.
[7] M. Watanabe, K. Makita, H. Usami, S. Motoo, J.