On active-site heterogeneity in pyrolyzed carbon-supported iron porphyrin catalysts for the electrochemical reduction of oxygen: An in situ moì?ssbauer study

Article abstract of DOI:10.1021/jp0266087

Some FeTPP-Cl/carbon electrocatalysts, heat-treated at temperatures up to 800 ?°C, have been studied with cyclic voltammetry, x-ray photoelectron spectroscopy, extended X-ray absorption fine structure, and in situ Moì?ssbauer spectroscopy. It appears that

Full text of DOI:10.1021/jp0266087

J. Phys. Chem. B 2002, 106, 12993-13001
12993
On Active-Site Heterogeneity in Pyrolyzed Carbon-Supported Iron Porphyrin Catalysts for
the Electrochemical Reduction of Oxygen: An In Situ Mo1ssbauer Study
A. L. Bouwkamp-Wijnoltz, W. Visscher, and J. A. R. van Veen
Schuit Catalysis Institute, EindhoVen UniVersity of Technology, P. O. Box 513, EindhoVen, The Netherlands
E. Boellaard and A. M. van der Kraan*
Interfacultair Reactor Instituut, Delft UniVersity of Technology, Mekelweg 15, Delft, The Netherlands
S. C. Tang
Shell DeVCo, Westhollow Technology Centre, Houston, Texas
ReceiVed: July 25, 2002; In Final Form: September 30, 2002
Some FeTPP-Cl/carbon electrocatalysts, heat-treated at temperatures up to 800 °C, have been studied with
cyclic voltammetry, x-ray photoelectron spectroscopy, extended X-ray absorption fine structure, and in situ
Mo¨ssbauer spectroscopy. It appears that the heat treatment induces considerable site heterogeneity in electronic
terms, although structurally the Fe-N4 moiety seems persistent. The data indicate that only part of these
Fe-N4 sites contributes to the activity for the electrochemical reduction of O2 but that they operate according
to the well-known redox mechanism.
Introduction
mechanism,13-15 in that a TM needs to be in its two valent state
to be able to accept an oxygen molecule. This mechanism is
not universally admitted, though,16 and indeed, it is worrisome
that in cyclic voltammetry (CV) only a fraction of the macro-
cyclics present seem to be electroactive and that, especially in
nonpyrolyzed catalysts, the apparent Fe(2+)/Fe(3+) redox
potential is so low that it is difficult to see how it can be active
at the much higher O2 reduction potentials. We have tried to
shed some light on this problem through in situ Mo¨ssbauer
spectroscopy, and the results obtained will be discussed herein.
This paper, then, presents ex situ and in situ 57Fe Mo¨ssbauer
absorption spectroscopy data, supported by some X-ray absorp-
tion spectroscopy and X-ray photoelectron spectroscopy (XPS)
measurements, on FeTPP-Cl adsorbed on high surface area
carbon black and heat-treated at various temperaturessthese data
indicate (i) that the majority of Fe present is electroactive,
although for a large part the electron transfer kinetics may be
very slow, (ii) that something similar to the original Fe-N4
moiety continues to be the site of the O2 reduction activity, even
after a heat treatment at temperatures as high as 800 °C, but
(iii) that the central Fe ion exists in a variety of electronic states,
and (iv) that the Fe(2+) state is in fact accessible at O2 reduction
potentials.
Carbon-supported transition metal (TM) N4 macrocyclics are
well-known catalysts for the electrochemical reduction of
oxygen, with possible application in a direct methanol fuel cell
with an acidic electrolyte since they are not poisoned by
methanol. For this purpose, the TM complexes are adsorbed
onto a support like carbon black or activated carbon and heat-
treated (at rather elevated temperatures, 500-800 °C) in an inert
atmosphere to increase the activity, selectivity, and stability of
the catalysts. The structure of the active site generated by such
a heat treatment has been the subject of some debate, and there
appears to be a growing consensus that at the optimum pyrolysis
temperatures the metal chelate has lost its aromatic ring structure
but has retained its Me-N4 core, which constitutes the active
site.1-7 However, other ideas continue to be canvassed; indeed,
recently, Bae et al.,8 in a fine in situ Fe-K edge X-ray
absorption near-edge (XANES) study, adduced further evidence
for their view that it is a TM in a non-N4 environment that is
the active site for oxygen reduction. We think that this view is
mistaken9 and would like to present some further evidence here,
arguing in particular that all non-Fe-N4 species that may be
generated in the pyrolysis step can be removed from the catalyst
without impairing its electrocatalytic activity. The nature of the
active sites generated at still higher temperatures and/or in the
presence of NH3 is not addressed here.10,11
Experimental Section
Another point that Bae et al.8 bring up is the site heterogeneity
in heat-treated carbon-supported iron porphyrins. It is indeed
the case that usually the active site population is seen as pretty
homogeneous, even in heat-treated catalysts, but this turns out
to be completely unwarranted. This point was also addressed
in A.L.B.-W.’s Ph. D. project by means of Mo¨ssbauer spec-
troscopy,12 and we would like to present the pertinent data in
this paper.
Undoped FeTPP-Cl was obtained from Aldrich. 57Fe-
enriched FeTPP-Cl was prepared using a method adapted from
refs 17-19 as follows:
First, tetraphenylporphyrin, H2TPP, was prepared applying
the Adler method using refluxing propionic acid.20 The chlorin
impurity was removed using 2,3-dichloro-5,6-dicyano-benzo-
quinone,21 and the H2TPP was purified by dry column chro-
matography, using alumina and CH2Cl2 as eluent, followed by
precipitation from the concentrated eluted solution through the
addition of methanol. Then, 33.5 mg of 57Fe was dissolved in
It seems natural to concieve of the reduction of oxygen on
these carbon-supported TM chelates in terms of a redox
10.1021/jp0266087 CCC: $22.00 © 2002 American Chemical Society
Published on Web 11/20/2002

12994 J. Phys. Chem. B, Vol. 106, No. 50, 2002
Bouwkamp-Wijnoltz et al.
12 mL of refluxing acetic acid and 1 mL of acetic acid
anhydride, under argon. Still under argon, the freshly made Fe-
(2+)-acetate solution was added to a solution of 372 mg of
H2TPP and 140 mg of NaCl in 30 mL of CH2Cl2. The reaction
mixture was brought to reflux, and argon was replaced with
oxygen. After 1 h, the solvent was removed, and the resulting
solid was dissolved in 250 mL of CH2Cl2; 200 mL of
concentrated HCl was added, and the mixture was stirred
magnetically for 1 h. The solvent was again removed, and the
product was dissolved in CHCl3, to which hot methanol was
added. After the product was cooled, dark crystals of 57FeTPP-
Cl formed. These were filtered off and dried at 125 °C/2 Torr;
yield, 410 mg (93%).
Vulcan XC-72R was chosen as carbon support. Catalysts were
prepared by loading 7 wt % FeTPP-Cl, corresponding to about
80-90% of a monolayer (assuming a flat adsorption geometry),
on the Vulcan support. This was achieved by dissolving the
appropriate amount of chelate in CHCl3, suspending the carbon
in this solution, and adding hexane slowly, under magnetic
stirring, until adsorption was completed (as judged from the
total disappearance of color from the solution).
Heat treatment was carried out in a down-flow reactor in an
argon atmosphere, deoxygenated using a Cu tower at 125 °C.
The argon flow rate was 25 mL/min. A heating rate of 10 °C/
min was applied and a dwell time at the final temperature of 2
h. The sample was cooled to room temperature in argon as well.
Figure 1. In situ Mo¨ssbauer cell.
In some cases, a heat-treated catalyst was “acid-washed”, i.e.,
treated with 0.5 M H2SO4 at 80 °C for 10-15 min, followed
by extensive rinsing with water. The Fe content before and after
this acid wash was determined by atomic absorption spectros-
copy (AAS) after total destruction of the sample (by heating
50 mg of catalyst, 2.5 mL of concentrated nitric acid, and 7.5
mL of concentrated sulfuric acid for 2 h at 150 °C in a Teflon
cup in a metal holder). Nitrogen elemental analysis was
performed at TNO Zeist and Mikroanalytisches Labor Pascher
(Remagen-Bandorf, Germany).
All electrochemical measurements were performed in con-
ventional three electrode setups, as described previously.15,22
Activity measurements employed the floating electrode tech-
nique23 in which Teflon-bonded catalysts (20 wt % Teflon, ex
Fluon GP-1 from Du Pont, in the final dry electrode) applied
on an 80 mesh gold screen were positioned horizontally such
that it just contacted the (oxygen-saturated) electrolyte (details
in ref 22). CV was mainly performed on Leit C electrodes,
employing a gold rotating disk electrode with a hollow cavity,
about 2 mm in depth. This cavity was filled with “Leit C”, a
conducting carbon cement obtained from Neubauer Chemicals,
to which the catalyst was applied by dipping the electrode with
fresh paste in a small amount of catalyst, followed by drying,
wetting with ethanol, and thorough rinsing with water (details
in ref 15). The AC voltammetry with such electrodes was carried
out at the University of Twente, employing the equipment
described in ref 24.
Figure 2. Experimental setup during Mo¨ssbauer experiments.
platinum in the form of a coarse mesh screen. The cell contained
two electrolyte compartments. Laminar flow of the electrolyte
was achieved by blocking one of the gas out flows, and in this
way, complete penetration of the electrode with electrolyte was
accomplished. If the electrolyte was not flowed through every
now and then during argon saturation, traces of oxygen remained
inside the working electrode. Two platinum counter electrodes
were used to ensure a reasonable current distribution. The
counter electrodes were interconnected. For the potential-
dependent measurements, the potential was scanned with 1 mV/s
from the open-circuit potential (OCP) to the desired potential.
The Mo¨ssbauer experimental setup is given in Figure 2. For
the potential-dependent measurements, the cell was quickly
cooled to liquid nitrogen temperature (77 K). As soon as the
electrolyte resistance reached a high value, indicated by the
Both ex situ and in situ Mo¨ssbauer experiments were
performed using the 57Fe-enriched FeTPP-Cl material. For the
in situ experiments, an electrochemical cell as shown in Figure
1 was used. It allowed experiments in oxygen- and argon-
saturated solutions. A mercury-sulfate electrode (MSE) was
used as a reference electrode. Its compartment was connected
to the cell via an agar-agar salt bridge. The electrolyte was
0.5 M H2SO4. The electrolyte layer was minimized by two thin
silicon spacers. The working electrode was made of Teflon-
sintered catalyst (100 mg of catalyst, 16 wt % Teflon, cured at
the standard temperature of 325 °C). The current collector was

Pyrolyzed Carbon-Supported Iron Porphyrin Catalysts
J. Phys. Chem. B, Vol. 106, No. 50, 2002 12995
TABLE 1: Fe-K Edge EXAFS Data for FeTPP-Cl/Vulcan
by Fitting in the R-Space until a Satisfactory Fit Is Obtained
for the k1-Weighted Fourier Transform Functionsa
first shell Fe-N
∆σ2
4.2 2.047 0.00214 0.07 8.2 2.971 0.00007 15.1
second shell Fe-C
catalyst
fresh
N
R
∆E
N
R
∆σ2
∆E
HT 325 3.3 2.056 0.00359 -0.07 8 3.044 0.00046 7.6
(4.0b) (2.044) (0.00712) (1.31)
(3.025) (0.000) (9.8)
HT 700 4.1 1.925 0.0100 11.6
a N ) coordination number ((20%), R ) interatomic distance ((0.02
Å), ∆σ2 ) Debye-Waller factor (in Å2 ( 20%), and ∆E (in eV (
20%). b As the value for N of the first coordination shell is fixed to
4.0 in the case of sample HT 325, the deduced values are presented
between brackets.
temperature, the Co-N4 moiety remains substantially intact;3
indeed, it appears that when Co ions and a suitable N donor
(e.g., 2,5-dimethylpyrrole) are added to the carbon support and
it is heated at the optimum temperature of 700 °C, the Co-N4
configuration forms spontaneously.22 At still higher tempera-
tures, decomposition of also this central part of the macrocyclic
is observed, with a concomitant loss in activity.3 To verify if
this picture also pertains in the case of FeTPP-Cl/carbon, three
samples (fresh, HT 325, and HT 700) were examined with XAS.
The analysis results of the iron K edge EXAFS data for these
three FeTPP-Cl/Vulcan samples are summarized in Table 1.
The Fe-N and Fe-C coordinations analyzed for the fresh
sample are consistent with the FeTPP-Cl crystal data,27 strongly
suggesting that the porphyrin adsorbs intact. On the other hand,
it is curious that there is no sign of the Cl ligand, the more so
as Cl is clearly present according to XPS, and the 57Fe
Mo¨ssbauer spectrum of the fresh catalyst is identical to that of
crystalline FeTPP-Cl (vide infra). Perhaps some variation in
the Fe-Cl interatomic distance is induced upon adsorption,
sufficient to make it disappear from the EXAFS spectrum.
After a heat treatment at 325 °Csthe curing temperature of
Teflon-bonded electrodessthe structure around the Fe has
changed only slightly, indicating that although largely intact
the porphyrin is starting to interact with the subjacent carbon
more strongly than before. This phenomenon is probably due
to redispersion of initially formed porphyrin crystallites, leading
to a more monolayer-like active phase (see below). The first
shell coordination number does appear to be a bit low at 3.3,
but constraining it to 4.0 does not visibly impair the fit (Table
1). The optimum heat treatment temperature of 700 °C obviously
has a much stronger impact. The EXAFS results indicate by
the large values of ∆σ2 ) 0.01 and ∆E ) 11.6 an increase in
static and/or dynamic disorder of the local structure of the iron
site. This increase in disorder of the local structure is most
probably due to the removal of the periphery of the macrocyclic
ligand (structure beyond the first shell could not be determined).
Nevertheless, the first shell still appears to be Fe-N4, but the
interatomic distance has contracted to 1.93 Å, becoming similar
to that observed for Co-N4, 1.90 Å, indicating that where Fe
sits above the porphyrin plane in intact FeTPP-Cl, it has moved
in-plane upon the destruction of the periphery of the macrocyclic
ligand. However, given the large disorder, it cannot be absolutely
excluded that other Fe sites than Fe-N4 exist in the HT 700
catalyst. Whether some Fe support bond has been formed, as
hypothesized earlier,28 must also remain moot.
Figure 3. Oxygen reduction at 7 wt % FeTPP-Cl/Vulcan heat-treated
at different temperatures. Floating electrodes, 0.5 M H2SO4, 1 mV/
sec.
overload signal of the potentiostat, the latter was turned off and
the leads were disconnected from the cell. This procedure was
taken from Fierro et al.25 who opined that the situation at the
electrode remained intact this way. Because of the position of
the cell, outside the cryostat, the temperature could not be
maintained at 77 K, but it slowly rose to about 130 K, which
then, however, remained rather constant during the experiment.
Room temperature EXAFS (extended X-ray absorption fine
structure) measurements were performed at Station 9.3 of the
Daresbury Synchrotron Radiation Source operating at an energy
of 2.0 GeV and an average current of 250 mA. The data were
taken in the fluorescent mode using a Si(220) double crystal
monochromator, and for iron, the K edge at 7.11 keV was used.
The data were analyzed with the XDAP package, version 2.2.2.26
Some XPS measurements were carried out at Shell Research
and Technology Centre, Amsterdam, making use of a Kratos
XSAM 800 instrument.
Results and Discussion
Overall Structure of the Active Site. The oxygen reduction
activity of FeTPP-Cl/Vulcan, heat-treated at various temper-
atures, is shown in Figure 3. As expected, the activity increases
with heat treatment temperature up to 700 °C, but heating at
800 °C does not lead to a further increase in activitysrather, a
decrease is observed. When these catalysts are measured with
a rotating ring-disk electrode (RRDE) setup, it is observed12
that the selectivity to H2O2 remains approximately constant, in
agreement with ref 6, suggesting that no major changes in active
site configuration occur.
If, then, the Fe-N4 moiety remains largely intact upon a heat
treatment at 700 °C, we should expect that hardly any Fe would
be lost upon acid washing, and in fact, only ∼15% of the Fe is
lost after the H2SO4 treatment at 80 °C. When, however, we do
the heat treatment at 800 °C, the picture changes dramatically.
In the case of CoTPP/carbon electrocatalysts, previous
EXAFS data concerning the effect of the pyrolysis on the local
Co structure show that up to and including the optimum
(optimum in terms of observed catalytic activity) pyrolysis

12996 J. Phys. Chem. B, Vol. 106, No. 50, 2002
Bouwkamp-Wijnoltz et al.
Figure 4. Result of the acid treatment, ex situ Mo¨ssbauer spectra,
300 K.
Figure 5. Ex situ Mo¨ssbauer spectra of 7 wt % 57FeTPP-Cl/Vulcan
heat-treated at different temperatures, 300 K.
Now, 40% of the Fe can be dissolved from the surface, and the
question is whether this Fe is involved in some way in active
site formation as Bae et al.8 claim or whether it represents an
inactive fraction. From the ex situ Mo¨ssbauer spectra in Figures
4 and 5, we deduce that (i) after a heat treatment at 800 °C the
spectrum is dominated by a species, a magnetic iron oxide phase
(presumably formed from metallic Fe, which should be the
primary pyrolysis product upon chelate destruction, through air
exposure), that is completely absent after the 700 °C treatment
and (ii) an acid wash removes this oxidic species nearly
completely, to leave a Mo¨ssbauer spectrum that closely re-
sembles the one obtained after 700 °C heating, which on the
basis of the EXAFS results, corresponds to some Fe-N4
species.
As the oxygen reduction activities of FeTPP-Cl/Vulcan HT
800 and its acid-washed counterpart turn out to be virtually
identical,12 one cannot but conclude that the Fe species that form
upon destruction of the porphyrin ligand are lost to catalysis,
which is dependent on the intact cores of the original chelate.
An overview of the ex situ Mo¨ssbauer spectra is given in
Figure 5. As mentioned above, the spectrum of the fresh catalyst
is identical to that of the FeTPP-Cl powder itself, which in
turn is in accordance with published data.18,19,21 A (apparent)
singlet is observed, due to spin-spin relaxation phenomena that
result in magnetic broadening.29 Treating FeTPP-Cl/Vulcan at
low temperatures (225, 325 °C) already changes its spectrum,
which indicates that even in this temperature range the environ-
ment of the iron center is already affected. The change is
probably connected with the redispersion of the porphyrin (see
discussion in the next section). After the 325 °C treatment, the
original FeTPP-Cl spectrum still accounts for ∼25% of the
observed spectrum. Increasing the pyrolysis temperature results
in spectra with broad lines due to a variety of doublets. The
HT 500 and HT 700 spectra, which are somewhat different from
those obtained previously for on the face of it comparable
samples28,30 for unknown reasons, will be further discussed
below. On the basis of the broad lines, it is already clear,
however, that a lot of heterogeneity has been induced at the
iron sites. This means that while the overall structure of the
iron site remains close to the original Fe-N4 moiety (on the
basis of the EXAFS results), there must be substantial local
variation in how this unit is attached to the subjacent carbon
carrier.
Cyclic Voltammetry (CV). The evolution of the redox
properties of FeTPP-Cl/Vulcan as the heat treatment temper-
ature increases was at first studied with CV. The CVs of the
fresh and HT 325 catalysts are compared in Figure 6. Both show
the same redox transition at 0.2 V vs RHE, which has been
assigned to the Fe(2+)/Fe(3+) transition of adsorbed FeTPP-
Cl.24 Although the peak position does not change, the signal is
much stronger for the HT 325 sample. This is due, we believe,
to redispersion of the chelate, implying that our preparation
method of forced adsorption in this case has led to small
crystallites: on one hand, it is observed in XPS that the N signal
for the HT 325 sample is stronger than for the fresh one (0.76
vs 0.30 at %), which can only be rationalized by assuming that
redispersion of initially formed crystallites has occurred; on the
other hand, while a physical mixture of FeTPP-Cl and Vulcan

Pyrolyzed Carbon-Supported Iron Porphyrin Catalysts
J. Phys. Chem. B, Vol. 106, No. 50, 2002 12997
Figure 6. Cyclic voltammograms of fresh and HT 325 7 wt % FeTPP-
Cl/Vulcan. Leit C electrodes, argon-saturated 0.5 M H2SO4, 100 mV/
sec.
as such does not show a redox transition, after a heat treatment
at 325 °C a (small) redox peak becomes visible. In terms of the
effectiveness of the pyrolysis process, this redispersion is a good
thing, as superactive site formation starts from the chelate in
monolayer form.9,17,29
It has been observed before that only a fraction of the chelate
adsorbed on a carbon carrier turns out to show up in CV.8,25,32
(That is, the charge corresponding to the redox transition seen
in CV is much less than the charge expected if all chelate
molecules present in the electrode would contribute to the
signal.) In the present case, it is only of the order of 10%. The
reason behind this relative invisibility is unknownsperhaps it
is related to inhomogeneity in the carbon support, inasmuch as
it has been shown that electron transfer rates are very slow for
the basal planes of graphite (HOPG) and can be greatly
improved by the introduction of surface defects.32 As will be
shown below in the paragraph of in situ Mo¨ssbauer experiments,
a much larger fraction is electroactive than appears from the
CV. For the moment, however, we will take the visible fraction
as representative.
The cyclic voltammograms obtained for FeTPP-Cl/Vulcan
after various heat treatment temperatures, from 325 to 700 °C,
are shown in Figure 7. It appears that the position of the redox
transition slowly moves to higher potentials with increasing HT
temperature and that its intensity continuously decreases, to
disappear in the HT 700 sample. The latter could simply be
due to peak broadening, preventing the redox current from rising
above the background charging current, and so, we turned to
AC voltammetry. As Figure 8 shows, the evolution of the redox
transition(s) can now be followed till the highest HT temper-
atures. The original Fe(2+)/Fe(3+) transition slowly moves from
about 0.2 V vs RHE in the fresh catalyst to about 0.4 V vs
RHE in the HT 700 one; after an 800 °C treatment, however,
it can no longer be clearly distinguished. Beginning with the
HT 500 sample, a second redox transition is in evidence, at
about 0.9 V vs RHE, which appears to broaden quite substan-
tially and to shift to lower potentials in the HT 700 and 800
samples. At present, we cannot definitely assign this transition,
but we tend to think that it may be due to a ligand transition;
in the intact porphyrin, the TPP/TPP+• transition occurs at about
1.2 V vs RHE, and the partial destruction of the porphyrin
ligand, with concomitant bonding of the pyrrolic part to the
subjacent carbon, as schematized in Figure 9, may just move it
to lower potentials. It is noted that a similar redox transition
occurs in the CV of CoTPP/Vulcan HT 500 but not for the
corresponding H2TPP sample, which however is thermally less
Figure 7. Cyclic voltammograms of 7 wt % FeTPP-Cl/Vulcan heat-
treated at different temperatures. Leit C electrodes, argon-saturated 0.5
M H2SO4, 100 mV/s.
stable. By the way, Figure 9 should be taken as indicative only;
we do not want to imply that all Fe-N4 sites retain square
symmetry, and indeed, we do not have any precise information
on how exactly the pyrrolic fragments bind to the carbon carrier.
That there is ample local variation of the precise geometry
follows from the EXAFS data (vide supra).
In sum, these voltammetric data indicate that the Fe(2+)/Fe-
(3+) transition occurs at much lower potentials (0.2-0.4 V vs
RHE) than where the oxygen reduction activity starts to be
measurable (0.6-0.8 V vs RHE), which would imply that the
availability of Fe(2+) sites at O2 reduction potentials is extremely
low. Still, on the assumption that such Fe(2+) sites are highly
active for O2 reduction, this shift toward higher potentials with
increasing heat treatment temperature could explain the increas-
ing activity of the FeTPP-Cl/Vulcan catalysts, in accordance
with the model we proposed earlier.28
In Situ Mo1ssbauer Experiments. Three FeTPP-Cl/Vulcan
samples have been studied with in situ Mo¨ssbauer spectroscopy,
namely, fresh (effectively HT 325, since we are using Teflon-
bonded electrodes that have been cured at 325 °C), HT 500,
and HT 700. The spectra obtained, whether in the dry state, at
open circuit, or under potential control, could be analyzed with
four doublets (Lorentzian line shapes) in the HT 500 and HT
700 cases, but the fresh electrode resisted. It turns out that part
of the iron in the fresh electrode is still characterized by
relaxation phenomena like in the untreated FeTPP-Cl, preclud-
ing full analysis of these spectra. The results of all fits have
been collected in Table 2. The four doublets can be qualitatively
characterized as follows:
(i) Doublet I is characterized by a high isomeric shift and a
high quadrupole splitting. This combination is unknown for iron
porphyrins,34 so we are dealing here with a yet unknown Fe-
(2+)-N4 species, or, more likely, with an Fe(1+) species.

12998 J. Phys. Chem. B, Vol. 106, No. 50, 2002
Bouwkamp-Wijnoltz et al.
Figure 8. As in Figure 7, AC voltammetry data, Leit C electrodes, nitrogen-saturated 0.5 M H2SO4, 100 mV/s, 50 mV amplitude, 150 Hz.
Figure 9. Visualization of the reaction of the porphyrin with the carbon during heat treatment.
(ii) Doublet II may originate from a high-spin Fe(3+) or a
low-spin Fe(2+). Because the relative contribution of doublet
II increases upon cathodic polarization, it is likely that it
originates from an Fe(2+) species. Also, it has been observed
for iron porphyrins that the quadrupole splitting increases when
the iron moves out-of-plane,34,35 and such is presumably the
case here (cf. the long Fe-N distance observed in EXAFS).
(iii) Doublet III is difficult to assign, since again it is not
possible to discriminate between a high-spin Fe(3+) and a low-
spin Fe (2+). However, on the basis of the data generally found
for iron porphyrins,34 a high-spin Fe(3+) seems most likely.
(iv) Doublet IV has a very low isomer shift, which may be
indicative of Fe(4+).
apart from the Fe-N4 ones. Recently, the Dodelet group has
made a case for an FeN2 site,36 and although we do not have
any proof of its existence in our heat-treated materials, we cannot
exclude it either.
Upon comparing the Mo¨ssbauer spectra of the electrodes
before and after contacting them with argon-saturated electrolyte,
essentially no differences are noticed (see first three entries in
Table 2 (HT 500 and HT 700), implying that at open circuit
the distribution over the Fe sites is not changed to any great
extent with respect to the dry catalyst. That Bae et al.8 came to
a different conclusion is due to the fact that they studied HT
800 material, in which part of the central Fe-N4 species is
destroyed giving rise to Fe species that are soluble in the
electrolytesin which case, obviously the Mo¨ssbauer spectrum
is going to be different in the dry and wet states. The present
results indicate that as long as the Fe-N4 moiety is preserved
(i.e., up to HT 700) no important changes occur when contacting
an FeTPP-Cl/Vulcan catalyst with acid electrolyte.
Although definitive assignments are difficult to make, it does
seem clear that the Fe species in our electrodes exist in a limited
variety of reasonably well-defined electronic states. Neverthe-
less, the apparent coexistence of Fe(1+) through Fe(4+) species
is surprising and may indicate that there are some other sites

Pyrolyzed Carbon-Supported Iron Porphyrin Catalysts
J. Phys. Chem. B, Vol. 106, No. 50, 2002 12999
TABLE 2: Results of Lorentz Fits for Several In Situ Mo1ssbauer Spectraa
δ*
∆
Γ
spectr.
contr. (%)
(mm s-1
)
(mm s-1
)
(mm s-1
)
HT 500
dry electrode
300 K
II
II
III
IV
1.20
0.69
0.61
0.20
2.43
1.85
0.87
2.27
0.73
0.67
0.60
0.30
5
15
79
1
electrode in 0.5 M H2SO4 OCP, argon
300 K
II
II
III
IV
1.11
0.75
0.62
0.20
2.62
1.85
0.85
2.24
0.92
0.69
0.61
0.34
6
15
76
3
electrode in 0.5 M H2SO4 OCP, argon
130 K
I
II
III
IV
1.54
0.92
0.70
0.29
3.33
2.15
0.87
2.40
0.65
0.80
0.60
0.56
3
20
71
6
electrode in 0.5 M H2SO4 at 0.55 V, argon
130 K
I
II
III
IV
1.58
0.97
0.71
0.31
3.27
2.20
0.87
2.39
0.56
0.74
0.59
0.71
6
21
63
10
electrode in 0.5 M H2SO4 at -0.05 V, argon
130 K
I
I,II
II
1.62
1.24
0.89
0.72
3.21
3.37
2.19
0.85
0.39
0.42
0.83
0.53
13
6
48
33
III
HT 700
dry electrode
300 K
I
II
III
IV
1.41
0.80
0.63
0.13
3.14
2.40
0.96
2.38
0.23
0.96
0.64
0.41
2
26
67
5
electrode in 0.5 M H2SO4
300 K
I
II
III
IV
1.36
0.87
0.63
0.20
3.03
2.36
0.91
2.45
0.28
0.80
0.63
0.84
3
21
63
13
electrode in 0.5 M H2SO4 OCP, argon
130 K
I
II
III
IV
1.61
0.97
0.73
0.20
3.17
2.34
0.88
2.40
0.31
0.96
0.58
0.68
3
33
50
14
electrode in 0.5 M H2SO4 at 0.63 V, oxygen
130 K
I
II
III
IV
1.63
1.03
0.73
0.27
3.07
2.23
0.82
2.30
0.60
0.70
0.56
0.60
6
25
48
21
electrode in 0.5 M H2SO4 at -0.05 V, argon
130 K
I
I,II
II
1.63
1.08
1.01
0.77
3.19
3.72
2.28
0.91
0.38
0.68
0.81
0.58
16
19
43
22
III
a The double indication I, II means that this contribution is somehow between I and II.
To find out what fraction of the Fe sites is actually
electroactive, the electrodes were polarized to -0.05 V vs RHE
and cooled to 130 K, and their Mo¨ssbauer spectra were
determined. In Figure 10, they are compared with the spectra
obtained at the OCP, and the fit results for the HT 500 and HT
700 electrodes can be found in Table 2, last entries. An example
of the deconvolution of the Mo¨ssbauer spectra into four doublets
is shown in Figure 11. Clearly, a large fraction of the Fe sites
is affected by the potential changesapproximately two-thirds,
which is similar to the result obtained by Fierro et al. 25 for an
untreated (FeTPP)2O/carbon electrode. Although we could not
completely analyze the fresh electrode spectrum, doublet I is
readily apparent in it. As to why only a relatively small part of
these Fe sites show up in the CV, we can only speculate, but
site heterogeneity is what comes most readily to mind (however,
see below).
oxygen) samples are given in Table 2. It appears that there is
relatively little change with respect to the OCP data, which leads
us to the following remarks. (i) These Mo¨ssbauer results are
consistent with our interpretation of the AC voltammetry (vide
supra), since only the low-potential redox transition, which
always occurs below 0.6 V, was ascribed to Fe species; in the
process, the assignment of the high-potential transition to a
ligand-based redox couple is strengthened. (ii) The presence of
oxygen does not leave a strong imprint on the distribution of
Fe over its various states, and perhaps, none should have been
expected since oxygen reduction is first-order in oxygen,14
implying a low surface coverage with reaction intermediates.
Do the present in situ Mo¨ssbauer data throw any light on the
question why the HT 700 sample is so much more active than
the HT 500 one (cf. Figure 3)? It has been our view14 that the
increased activity of heat-treated Fe-porphyrin/carbon catalysts
is due to an increase of the redox potential of the central Fe
ion, mainly entailing a higher number of active Fe(2+) sites at
oxygen reduction potentials. A much-increased presence of Fe-
The Mo¨ssbauer spectra obtained at about 0.6 V vs RHE, two
under oxygen and one under argon, are shown in Figure 12.
The fit results for the HT 500 (under argon) and HT 700 (under

13000 J. Phys. Chem. B, Vol. 106, No. 50, 2002
Bouwkamp-Wijnoltz et al.
Figure 10. In situ Mo¨ssbauer spectra of fresh, HT 500, and HT 700
7 wt % 57FeTPP-Cl/Vulcan, 0.5 M H2SO4, 130 K, E vs RHE.
Figure 12. In situ Mo¨ssbauer spectra of fresh, HT 500, and HT 700
7 wt % 57FeTPP-Cl/Vulcan, 0.5 M H2SO4, 130 K. (a) Fresh, 0.55 V
vs RHE, oxygen. (b) HT 500, 0.55 V vs RHE, argon. (c) HT 700, 0.63
V vs RHE, oxygen.
which we reported for FeTPP a long time ago.14 On this
interpretation, CV invisibility is connected with extremely
sluggish electron transfer rates, also preventing the sites
concerned from making their mark in the electrocatalysis.
Conclusions
The data presented in this paper on (heat-treated) FeTPP-
Cl/Vulcan would appear to lead to the following conclusions.
(I) The EXAFS results are consistent with the continued
presence of the central Fe-N4 moiety up to temperatures of
700 °C, the optimum pyrolysis temperature as far as the oxygen
reduction activity is concerned, in line with our previous EXAFS
study of (heat-treated) CoTPP/carbon.3 It is only at 800 °C that
evidence for its partial destruction is obtained, but this leads to
species that do not contribute to the O2 reduction activity as
they can be acid washed out of the catalyst without deterioration
of the observed activity. (II) The CV data, combined with XPS,
indicate that a heat treatment at 325 °C improves the contact
between the chelate and the carbon carrier through redispersion
of the initially present FeTPP-Cl crystallites. Still, only about
10% of the FeTPP-Cl applied contributes to the redox transition
occurring at 0.2 V vs RHE. According to AC voltammetry, the
position of this redox transition shifts to higher potentials with
increasing heat treatment temperature, up to about 0.4 V vs RHE
at 700 °C. After it is heated at 800 °C, however, it is no longer
clearly distinguishable. (III) The in situ Mo¨ssbauer data on the
(57Fe-enriched) HT 500 and HT 700 samples show (i) that they
are quite heterogeneous, necessitating four doublets to get a
Figure 11. Deconvolution of the in situ Mo¨ssbauer spectrum of HT
700 7 wt % 57FeTPP-Cl/Vulcan, 0.5 M H2SO4, 130 K, -0.05 V vs
RHE.
(2+) sites in the HT 700 electrode is not readily apparent from
the data collected in Table 2. Still, there is certainly Fe(2+)
present at oxygen reduction potentials and the fraction of Fe-
(2+) is relatively larger for the HT 700 catalyst (compare the
results for the HT 500 and HT 700 electrodes in H2SO4 at 300
and 130 K (at OCV)). Maybe then, after all, it is only the fraction
of Fe sites visible in CV that contributes to the oxygen reduction
activity, which in the present case would mean that most of the
porphyrin incorporated in the electrode is wasted; this may be
related to the pronounced dependency of the specific activity
(mA/mg chelate) on the loading (mg chelate/g support), starting
out high at very low loadings and decreasing subsequently,

Pyrolyzed Carbon-Supported Iron Porphyrin Catalysts
J. Phys. Chem. B, Vol. 106, No. 50, 2002 13001
(8) Bae, I. T.; Tryk, D. A.; Scherson, D. A. J. Phys. Chem. B 1998,
102, 4114.
(9) van Veen, J. A. R.; Colijn, H. A.; van Baar, J. F. Electrochim.
Acta 1988, 33, 801.
(10) Lefe`vre, M.; Dodelet, J. P.; Bertrand, P. J. Phys. Chem. B 2000,
104, 11238.
(11) Bron, M.; Fiechter, S.; Hilgendorff, M.; Bogdanoff, P. J. Appl.
Electrochem. 2002, 32, 211.
(12) Wijnoltz, A. L. Ph.D. Thesis, ISBN 90 386 0384 3, Eindhoven,
1995.
(13) Beck, F. J. Appl. Electrochem. 1977, 7, 239.
(14) van Veen, J. A. R.; van Baar, J. F.; Kroese, C. J.; Coolegem, J. G.
F.; de Wit, N.; Colijn, H. A. Ber. Bunsen-Ges. Phys. Chem. 1981, 85, 693.
(15) Bouwkamp-Wijnoltz, A. L.; Visscher, W.; van Veen, J. A. R.
Electrochim. Acta 1998, 43, 3141.
(16) Radyushkina, K. A.; Levina, O. A.; Tarasevich, M. R. Elek-
trokhimiya 1994, 30, 991.
(17) Widelov, A.; Larsson, R. Electrochim. Acta 1992, 37, 187.
(18) Torrens, M. A.; Straub, D. K.; Ebstein, L. M. J. Am. Chem. Soc.
1972, 94, 4160.
reasonable fit; (ii) that contacting the electrode with electrolyte
does not change its spectrum, indicating that the mere presence
of 0.5 M H2SO4 does not modify the structure of the Fe sites;
(iii) that polarization to low potential (-0.05 V vs RHE) does
change the spectra quite considerably, and the fits, although
qualitative, indicate that the majority (on the order of two-thirds)
of the Fe ions has changed valence, which would imply that a
lot of the Fe that is CV invisible still responds to a potential
change, either because their redox potentials are distributed over
a broad range or because their electron transfer kinetics are too
sluggish; and (iv) that the differences observed between the HT
500 and the HT 700 electrodes at O2 reduction potentials (about
0.6 V vs RHE) are too small to readily explain their difference
in O2 reduction activity, although the suspected active Fe(II)
site is somewhat more abundant for the HT 700 material.
The picture that emerges from this is that there is no reason
to abandon the redox model of O2 electrocatalysis by TM
chelates, in which the availability of Me(II)-N4 sites at O2
reduction potentials is an important parameter,12 but we admit-
tedly have a complication in the present case of Vulcan-
supported FeTPP-Cl, in that only a fraction of the available
Fe-N4 sites appears to be involved in the catalysis. It is
tempting to identify this fraction with what is visible in
voltammetry, since it shows the shift in Fe(2+)/Fe(3+) redox
potential upon pyrolysis, that is an essential feature of the redox
model. We realize, however, that further work is necessary to
actually prove all of this.
(19) Torrens, M. A.; Straub, D. K.; Ebstein, L. M. J. Am. Chem. Soc.
1972, 94, 4162.
(20) Adler, A. D.; Longo, F. R.; Finarelli, J. D.; Goldmacher, J.; Assour,
J.; Korsokoff, L. J. Org. Chem. 1967, 32, 476.
(21) (a) Barnett, G. H.; Hudson, M. F.; Smith, K. M. Tetrahedron Lett.
1973, 30, 2887. (b) van der Made, A. W. Ph.D. Thesis, Utrecht, 1988.
(22) Bouwkamp-Wijnoltz, A. L.; Visscher, W.; van Veen, J. A. R.; Tang,
S. C. Electrochim. Acta 1999, 45, 379.
(23) Giner, J.; Parry, J. M.; Smith, S.; Turchan, M. J. Electrochem. Soc.
1969, 116, 1692.
(24) van den Ham, D.; Hinnen, C.; Magner, G.; Savy, M. J. Phys. Chem.
1987, 91, 4743.
(25) Fierro, C. A.; Mohan, M.; Scherson, D. A. Langmuir 1990, 6, 1338.
(26) Vaarkamp, M.; Linders, J. C.; Koningsberger, D. C. Physica B 1995,
208/209, 159.
(27) (a) Hoard, J. L.; Cohen, G. H.; Glick, M. D. J. Am. Chem. Soc.
1967, 89, 1992. (b) Scheidt, W. R.; Finnegan, M. Acta Crystallogr. C 1989,
45, 1214.
(28) van Veen, J. A. R.; van Baar, J. F.; Kroese, C. J. J. Chem. Soc.,
Faraday Trans. 1 1981, 77, 2827.
(29) Maricondi, C.; Straub, D. K.; Epstein, L. M. J. Am. Chem. Soc.
1972, 94, 4157.
(30) Blomquist, J.; Lang, H.; Larsson, R.; Widelov, A. J. Chem. Soc.,
Acknowledgment. We thank Drs. B. Palys and D. van den
Ham+ (Universtity of Twente) for their great help in the AC
voltammetry and Mr. G. Mulder (Shell Amsterdam) for the XPS
measurements.
References and Notes
Faraday Trans. 1992, 88, 2007.
(1) van Veen, J. A. R.; van Baar, J. F. ReV. Inorg. Chem. 1982, 4,
293.
(2) Ohms, D.; Gupta, S.; Tryk, D. A.; Yeager, E.; Wiesener, K. Z.
Phys. Chem. Leipzig 1990, 271, 451.
(3) van Wingerden, B.; van Veen, J. A. R.; Mensch, C. T. J. J. Chem.
Soc., Faraday Trans. 1 1988, 84, 65.
(4) Lalande, G.; Cote, R.; Guay, D.; Dodelet, J. P.; Weng, L. T.;
Bertrand, P. Electrochim. Acta 1997, 42, 1379.
(5) Gouerec, P.; Savy, M.; Riga, J. Electrochim. Acta 1998, 43, 743.
(6) Gojkovic, S. L.; Gupta, S.; Savinell, R. F. Electrochim. Acta 1999,
45, 889.
(31) Scherson, D. A.; Gupta, S. L.; Fierro, C.; Yeager, E. B.; Kordesch,
M. E.; Eldridge, J.; Hoffman, R. W.; Blue, J. Electrochim. Acta 1983, 9,
1205.
(32) Ouyang, J.; Shigehara, K.; Yamada, A.; Anson, F. C. J. Electroanal.
Chem. 1991, 297, 489.
(33) Kneten Kline, K.; McDermott, M. T.; McCreery, R. L. J. Phys.
Chem. 1994, 98, 5314.
(34) Debrunner, P. G. Phys. Bioinorg. Chem. Ser. 1989, 4, 137-234.
(35) Collman, J. P.; Gagne, R. R.; Reed, C. A.; Halbert, T. R.; Lang,
G.; Robinson, W. T. J. Am. Chem. Soc. 1975, 97, 1427.
(36) Lefe`vre, M.; Dodelet, J. P.; Bertrand, P. J. Phys. Chem. B 2002,
106, 8705.
(7) Sun, G.-Q.; Wang, J.-T.; Gupta, S.; Savinell, R. F. J. Appl.
Electrochem. 2001, 31, 1025.