Oxidation mechanism of formic acid on the bismuth adatom-modified Pt(111) surface

Article abstract of DOI:10.1021/ja505943h

In order to improve catalytic processes, elucidation of reaction mechanisms is essential. Here, supported by a combination of experimental and computational results, the oxidation mechanism of formic acid on Pt(111) electrodes modified by the incorporatio

Full text of DOI:10.1021/ja505943h

Communication
pubs.acs.org/JACS
Oxidation Mechanism of Formic Acid on the Bismuth Adatom-
Modified Pt(111) Surface
†
‡,§
†
,†
́
Juan Victor Perales-Rondon, Adolfo Ferre-Vilaplana, Juan M. Feliu, and Enrique Herrero*
†
‡
§
Instituto de Electroquímica, Universidad de Alicante, Apdo. 99, E-03080 Alicante, Spain
Instituto Tecnologico de Informatica, Ciudad Politecnica de la Innovacion, Camino de Vera s/n, E-46022 Valencia, Spain
Departamento de Sistemas Informaticos y Computacion, Escuela Politecnica Superior de Alcoy, Universidad Politecnica de Valencia,
Plaza Ferrandiz y Carbonell s/n, E-03801 Alcoy, Spain
S Supporting Information
́
́
́
́
́
́
́
́
́
*
thus is difficult to oxidize, and through an active intermediate,
2
,7
ABSTRACT: In order to improve catalytic processes,
elucidation of reaction mechanisms is essential. Here,
supported by a combination of experimental and computa-
tional results, the oxidation mechanism of formic acid on
Pt(111) electrodes modified by the incorporation of
bismuth adatoms is revealed. In the proposed model,
formic acid is first physisorbed on bismuth and then
deprotonated and chemisorbed in formate form, also on
bismuth, from which configuration the C−H bond is
which is the desired route. For this route, the nature of the
8
−13
intermediate is still under discussion.
Electrocatalytic
reaction rates depend on the surface structure of the electrodes.
So, Pt(111) electrodes exhibit low activity through the active
intermediate and negligible CO formation, whereas Pt(100)
14,15
surfaces show the highest activity for both routes.
Moreover, to increase the reaction rate through the direct
route, pure platinum electrodes have been modified incorporat-
ing other species. Bismuth has given rise to very promising
cleaved, on a neighbor Pt site, yielding CO . It was found
16−18
2
results.
The addition of Bi adatoms to pristine Pt(111)
computationally that the activation energy for the C−H
bond cleavage step is negligible, which was also verified
experimentally.
single-crystal electrodes diminishes the overpotentials and
increases the current density up to 30−40 times that measured
15
for the unmodified surface. Such improvement is clearly
related to the deposition of Bi on the terraces, since it reaches
the maximum for coverages close to the saturation value. For
low and moderate Bi coverages, only isolated or <1 nm islands
have been seen by scanning tunneling microscopy, which
supports the random distribution model of adatoms on the
leaner energy based on the oxidation of small organic
molecules (i.e., formic acid, methanol, ethanol, etc.) in
fuel cells requires efficient electrocatalysts capable of operating
at low overpotentials and high current densities. To guide the
development of optimal catalysts, the oxidation mechanisms of
the fuels have to be fully understood. The oxidation of formic
C
19
surface. Also, statistical models suggest that a Bi−Pt ensemble
would be the reactive site, although the role played by each
atom in the considered oxidation process has not yet been
acid to CO , involving the exchange of only two electrons and
20
2
described.
the cleavage of a single C−H bond, is probably the simplest
It has been pointed out that adsorbed formate plays an
important role in the process of oxidation of formic acid on
pure platinum electrodes. This species has been detected by
1
,2
among the processes considered. In order to activate the
aforementioned C−H bond cleavage, platinum surfaces are
particularly effective, giving rise to overpotentials lower than
those observed in the oxidation of small organic molecules on
others metals. However, even on pure platinum, the measured
oxidation currents, at low overpotentials, are still low. To
increase the activity of the catalyst, modifications of platinum
with other elements have been proposed. Intermetallic PtBi
nanoparticles have shown an excellent performance, where high
9−13
21
FTIR
and voltammetry, showing that the onset for the
oxidation process coincides with that for the adsorption of
21
formate. Also, density functional theory calculations indicate
that adsorbed formate is a key part of the considered oxidation
22,23
process.
More specifically, it has been suggested that
adsorbed formate facilitates the adsorption of a new formic acid
3
−6
(or formate) molecule with the C−H bond directed toward the
currents at very low overpotentials have been recorded.
24
surface. From such a configuration, the activation energy for
the cleavage of the C−H bond would be low (ca. 0.5 eV),
Nevertheless, the role played by Bi, or any other modifier, in
the considered oxidation process is still unknown. In this work,
using a combination of DFT calculations and experimental
results on single-crystal electrodes, the electro-oxidation
mechanism of formic acid on the bismuth adatom-modified
Pt(111) surface is completely elucidated. Note that single-
crystal electrodes allow direct comparison between computa-
tional and experimental results.
yielding CO as the final product. To adsorb the formate−
2
formic dimer, a Pt ensemble with several Pt atoms is required.
Thus, the initial hypothesis could be that bismuth facilitates the
adsorption of formate on Pt sites.
To detect adsorption of formate on single-crystal electrodes,
21
fast-scan voltammetry can be used. During formic acid
The electrooxidation of formic acid on pure platinum
electrodes can evolve following two main routes: through
CO, which remains strongly adsorbed on the electrode, and
Received: June 13, 2014
©
XXXX American Chemical Society
A
dx.doi.org/10.1021/ja505943h | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
oxidation, voltammetric currents come from two different
processes: the oxidation process itself and adsorption processes
corroborates that these processes are due to the adsorption of
25
the formate anion. As the amount of bismuth coverage
increases, the onset for the formate anion adsorption is
(
mainly hydrogen and anions) on the electrode surface. At low
26
scan rates, oxidation currents are much higher than adsorption
currents. However, adsorption currents are proportional to the
scan rate, whereas oxidation currents are much less affected by
displaced toward higher potential values (see Figure S5). In
parallel to the appearance of the new signal between 0.4 and 0.6
V, those signals corresponding to the adsorption of OH above
0.7 V disappear, and a small shift in the bismuth redox peaks is
1
8
the scan rate. By using high scan rates, adsorption currents
21
can dominate the voltammetric profile. For a Bi adatom-
modified Pt(111) electrode with a coverage of 0.25, which is
very close to the maximum activity, both types of experiments
observed. All these results indicate that, for θ = 0.25, formate
Bi
is present on platinum sites only above 0.5 V, whereas on the
21
unmodified Pt(111) electrode it can be detected at 0.35 V.
(low and high scan rate voltammetry) are shown in Figure 1.
On the other hand, the onset potential for formic acid oxidation
diminishes as the bismuth coverage increases. Thus, it is clear
that the catalytic effect of bismuth is not linked to the
adsorption of formate on Pt sites.
In order to gain insight into the mechanisms explaining the
higher performance of bismuth modified electrodes, DFT
calculations, modeling different aspects of the investigated
process, were carried out. Adsorption of bismuth on fcc and
hcp sites of the Pt(111) surface was found to be favorable by
3
.20 and 3.17 eV, respectively. The adsorption of bismuth
causes a redistribution of electron density on the bare surface.
The results corresponding to fcc sites are displayed in Figure 2;
Figure 1. Voltammetric profiles of Bi−Pt(111) electrode with θ
=
Bi
0
.25 in 0.5 M HClO + 0.05 M HCOOH (full line) and 0.5 M HClO
4
4
(
dashed lines) at (A) 0.05 and (B) 50 V/s.
The same kinds of experimental results are shown in the
Supporting Information, Figures S1−S3, for the bare electrode
and two additional coverages. As can be seen by comparison of
the voltammetries in presence and absence of formic acid, at 50
mV/s (Figures 1A and S1A−S3A), adsorption currents are
much smaller than those corresponding to formic acid
oxidation, the shape of the voltammogram indicating an
irreversible oxidation process that is inhibited at high potentials.
At high bismuth coverages and above 0.5 V, the incoming
molecules are immediately oxidized on the electrode.
Figure 2. Electrostatic potential [Ha/e] mapped on electron iso-
density surface for a density value ρ = 0.01 e/Å for a Bi adatom
adsorbed on the Pt(111) surface.
3
However, at 50 V/s (Figures 1B and S1B−S3B), adsorption
currents have increased 3 orders of magnitude, making
adsorption the dominant process recorded in the voltammetry.
In presence of formic acid, the voltammograms are practically
symmetrical through the potential axis, which indicates that
currents are mainly due to adsorption processes, and thus, the
contribution from oxidation processes can be discarded.
Comparison with the voltammograms measured in the absence
of formic acid enables us to establish the formate adsorption
range on platinum. In perchloric acid solutions, a hydrogen
adsorption region appears at potentials below 0.3 V, whereas
bismuth redox give rise to a pair of peaks between 0.6 and 0.7
V, in a similar way to what is found at 50 mV/s (Figures 1B and
S4). Above 0.7 V, the signals corresponding to OH adsorption
on Pt sites can be detected. In presence of formic acid at
the ones corresponding to hcp sites are virtually identical. From
Figure 2, it can be inferred that an excess of positive charge
would be concentrated on the adatom, while the compensating
negative charge would be distributed among the platinum
atoms close to the adatom. Thus, bismuth has cationic
character when adsorbed on Pt(111). Here we propose that
the formation of the identified cationic site determines the
oxidation process of formic acid on the Bi−Pt(111) surface,
driving its mechanism.
It was found that physisorption of hydrated formic acid near
the adatom is favorable by ca. 0.26 eV. In its lower energy
configuration, the carbonyl group is directed toward the adatom
(Figure S6), with changes in the bonds geometry being
observed. In such a configuration, the O−H distance of the
carboxylic group increases, whereas the distance to the
corresponding hydrogen bond diminishes as compared to the
distances calculated in absence of surface. Adatoms favoring
physisorption of formic acid in the C−H down configuration
5
0 V/s, some changes can be observed in the voltammograms.
First, depending on the coverage, a signal corresponding to
formate adsorption can be seen between 0.4 and 0.6 V in the
positive scan direction (Figure S5). In Figure 1B, since bismuth
coverage is high, the amount of free Pt sites is small, and
therefore, the total amount of adsorbed formate on platinum is
also small. However, in Figure S5, the signal corresponding to
formate adsorption diminishes as the coverage increases. For
bismuth-modified surfaces, the voltammograms in the presence
of formic acid at 50 V/s are almost identical to those measured
in sulfuric acid solutions in the absence of formic acid, which
27
have been previously found, though the evolution of formic
acid on the adatom site was not studied, and only the generic
argument that the configuration facilitates the cleavage of the
C−H bond is provided. Here, the relevance of the role played
by the adatom during the oxidation process is established, and
the complete oxidation mechanism is revealed.
B
dx.doi.org/10.1021/ja505943h | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
From the described physisorbed configuration, the deproto-
nation of the formic acid molecule would be assisted by the fact
that, from this location, formate was found to be favorably
chemisorbed by 0.39 eV on the bismuth adatom without
barrier. Therefore, the two processes of formic acid
deprotonation and formate chemisorption on bismuth would
take place simultaneously. The above-described physisorption-
and formate chemisorption-related observations suggest that
were experimentally estimated. To do so, voltammetric profiles
for the oxidation process were recorded at different temper-
atures (Figure 4A). Since CO formation on these electrodes is
the pK of formic acid near the surface would be lower than that
a
in the bulk. Physisorption, deprotonation, and formate
chemisorption would all be driven by the positive charge
located on the adatom.
Once formate chemisorption on the adatom has taken place,
it was assumed that the electron in excess exits toward the
circuit. So, the evolution of the HCOO fragment bonded to the
bismuth modified platinum surface was analyzed. Running
quantum mechanical molecular dynamics simulations, it was
found that, transiting among different configurations whose
energies fluctuate within the error of the calculations, the
chemisorbed HCOO fragment evolves in a rotating process for
which the Bi−O and O−C bonds play the role of rotation axis,
until a sufficiently favorable C−H down configuration is
reached. From this configuration the C−H bond is cleaved to
yield CO and an adsorbed H atom (Figure 3). The activation
2
Figure 4. (A) Voltammetric profile of a Bi−Pt(111) electrode with θ
Bi
=
0.25 in 0.5 M H SO + 0.1 M HCOOH at 0.05 V/s at different
2 4
temperatures. (B) Measured activation energy for different Bi
coverages θ : (▼) 0.00, (■) 0.10, (●) 0.22, and (▲) 0.28.
Bi
1
5,17
negligible,
the measured currents correspond to the direct
route. So, the logarithm of the current density at constant
potential can be charted vs T− in an Arrhenius plot to
determine the apparent activation energy (Figure S7).
Activation energies vs electrode potentials for different bismuth
coverages are displayed in Figure 4B. For comparison, the
activation energy for the unmodified Pt(111) electrode is also
included. For the bare electrode, the activation energy was
determined from the intrinsic activity of the electrode through
the active intermediate reaction measured at different temper-
1
1
4,15
atures.
In Figure 4B, as bismuth coverage increases, a significant
diminution in the activation energy of the process can be
observed. For the electrode with the highest activity, the
diminution is around 20 kJ/mol with respect to that
corresponding to the unmodified Pt(111) electrode. It should
be borne in mind that experimentally estimated activation
energies are apparent activation energies, which are a
combination of the activation energies of all the different
steps involved in a mechanism, whereas computationally
estimated activation energies correspond to a single step in a
process. Additionally, in the case of the bismuth-modified
Pt(111) electrodes, the measured activation energy is an
average of the activation energies for all the different kinds of
reactive sites on the surface. For the investigated surfaces, two
different kinds of active sites are considered: the Pt−Bi
ensemble of Figure 3, and the Pt ensemble on which the
reaction occurs through the formate−formic dimer. If the
activation energy for the Pt−Bi ensemble is much lower than
that for the Pt sites, it is clear that the measured activation
energy should diminish when increasing bismuth coverage, as
found experimentally. For the bismuth coverage with the
Figure 3. (A) HCOO fragment adsorbed on the Bi−Pt(111) surface
and (B) final products in the oxidation of formic acid.
energy estimated from the calculations for the C−H bond
cleavage step was negligible, presenting a favorable energy of
1
.21 eV. Comparison with previous results on unmodified
24
Pt(111) electrodes shows a significant diminution in
activation energy for the C−H bond cleavage step, from 0.5
to 0 eV.
In summary, the oxidation reaction model proposed here
draws a bifunctional catalyst in which the bismuth adatom plays
a determinant role in the formic acid adsorption and evolution
to chemisorbed formate, whereas an adjacent Pt site is
responsible for the cleavage of the C−H bond, which is in
agreement with previous analysis suggesting that the catalytic
2
0
site is a Bi−Pt ensemble.
To corroborate the proposed reaction mechanism, activation
energies for different bismuth coverages on Pt(111) electrodes
C
dx.doi.org/10.1021/ja505943h | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
highest catalytic activity (θ = 0.28), the measured activation
Notes
Bi
energy is not zero because of the contribution of other previous
steps and also from some Pt ensembles still accessible on the
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
surface. At 0.45 V, the measured activation energy for θ = 0.28
Bi
■
is 21 kJ/mol, whereas on the unmodified surface it is 42 kJ/
mol. From these data, the contribution of the different steps/
processes to the apparent activation energy at this bismuth
coverage can be estimated. Since each bismuth adatom blocks
three platinum sites, at the considered coverage, the number of
free Pt sites is 0.16. This implies that the contribution of these
sites to the total energy would be 0.16 × 42 = 6.7 kJ/mol. The
remaining 14.3 kJ/mol (or 0.15 eV) should arise, then, from the
previous steps for the reaction on the Bi−Pt ensemble,
probably from the deprotonation of the formic acid close to
the Bi adatom to give rise to formate, which will adsorb
immediately on the Bi adatom.
This work has been financially supported by the MINECO
(Spain) (project CTQ2013-44083-P) and Generalitat Valenci-
ana (project PROMETEOII/2014/013).
REFERENCES
■
(
1) Parsons, R.; Vandernoot, T. J. Electroanal. Chem. 1988, 257, 9.
(2) Koper, M. T. M.; Lai, S. C. S.; Herrero, E. In Fuel cell catalysis, a
surface science approach; Koper, M. T. M., Ed.; John Wiley & Sons,
Inc.: Hoboken, NJ, 2009; p 159.
(
3) Roychowdhury, C.; Matsumoto, F.; Zeldovich, V. B.; Warren, S.
C.; Mutolo, P. F.; Ballesteros, M.; Wiesner, U.; Abruna, H. D.; Disalvo,
F. J. Chem. Mater. 2006, 18, 3365.
4) Roychowdhury, C.; Matsumoto, F.; Mutolo, P. F.; Abruna, H. D.;
DiSalvo, F. J. Chem. Mater. 2005, 17, 5871.
5) Casado-Rivera, E.; Gal, Z.; Angelo, A. C. D.; Lind, C.; DiSalvo, F.
̃
(
Finally, it should be stressed that the reaction mechanism
suggested by our DFT calculations is in very good agreement
with a broad set of experimental data. First, unlike the Pt(111)
(
J.; Abruna, H. D. ChemPhysChem 2003, 4, 193.
12
electrodes, for which currents increase with the pH, the
(6) Blasini, D. R.; Rochefort, D.; Fachini, E.; Alden, L. R.; DiSalvo, F.
J.; Cabrera, C. R.; Abruna, H. D. Surf. Sci. 2006, 600, 2670.
(7) Feliu, J. M.; Herrero, E. In Handbook of fuel cellsfundamentals,
technology and applications; Vielstich, W., Gasteiger, H., Lamm, A.,
Eds.; John Wiley & Sons, Ltd.: Chichester, 2003; Vol. 2.
currents measured on the bismuth-modified Pt(111) electrodes
18
do not depend on the pH for values between 0 and 2, which
implies that the formic acid molecule is the reactant species at
these pH values. As has been shown, a formic acid molecule
(
8) Samjeske, G.; Miki, A.; Ye, S.; Osawa, M. J. Phys. Chem. B 2006,
10, 16559.
9) Chen, Y. X.; Ye, S.; Heinen, M.; Jusys, Z.; Osawa, M.; Behm, R. J.
J. Phys. Chem. B 2006, 110, 9534.
10) Cuesta, A.; Cabello, G.; Gutierrez, C.; Osawa, M. Phys. Chem.
close to the Bi adatom would have a lower pK , which would
a
1
(
facilitate its adsorption as formate on the bismuth adatom;
therefore, the presence of a formate molecule close to the
adatom is not required. Second, the activation energy
diminishes with the bismuth coverage. Third, the proposed
reaction mechanism helps to explain why the onset of the
reaction is between 0.2 and 0.3 V for the different bismuth
coverages. In the final step, adsorbed hydrogen is formed,
(
Chem. Phys. 2011, 13, 20091.
(11) Osawa, M.; Komatsu, K.; Samjeske, G.; Uchida, T.; Ikeshoji, T.;
Cuesta, A.; Gutierrez, C. Angew. Chem., Int. Ed. 2011, 50, 1159.
(12) Joo, J.; Uchida, T.; Cuesta, A.; Koper, M. T. M.; Osawa, M. J.
Am. Chem. Soc. 2013, 135, 9991.
“poisoning” the active site. In order to reactivate the active site,
(
13) Chen, Y. X.; Heinen, M.; Jusys, Z.; Behm, R. J. Langmuir 2006,
2, 10399.
14) Grozovski, V.; Climent, V.; Herrero, E.; Feliu, J. M.
ChemPhysChem 2009, 10, 1922.
15) Grozovski, V.; Climent, V.; Herrero, E.; Feliu, J. M. Phys. Chem.
Chem. Phys. 2010, 12, 8822.
16) Clavilier, J.; Fernandez-Vega, A.; Feliu, J. M.; Aldaz, A. J.
Electroanal. Chem. 1989, 258, 89.
(17) Herrero, E.; Fernandez-Vega, A.; Feliu, J. M.; Aldaz, A. J.
Electroanal. Chem. 1993, 350, 73.
(18) Macia, M. D.; Herrero, E.; Feliu, J. M. J. Electroanal. Chem.
003, 554, 25.
hydrogen should be desorbed. Thus, a steady current for formic
acid oxidation is possible only at potentials at which hydrogen
is easily desorbed from the Pt sites, that is, where hydrogen
equilibrium coverage on the surface is low. On the unmodified
Pt(111) surface, hydrogen coverage is low above 0.30 V, and
this potential value diminishes as the bismuth coverage
increases (see Figure S4). Thus, the onset potential for the
reaction shifts accordingly. And, fourth, the proposed
mechanism also explain why the activation energy diminishes
from 0.2 to 0.5 V. Hydrogen desorption also contributes to the
measured activation energy, since it can be considered also the
first step in the reaction at potentials where hydrogen is
adsorbed. As the potential is made more positive, this step is
faster (i.e., with a lower activation energy), which results in a
diminution of the overall activation energy. As a conclusion, it
can be verified that the computational results presented here
are in very good agreement with a broad set of experimental
data, thus explaining them, which rigorously supports the
proposed reaction model.
2
(
(
(
́
́
́
2
(19) Kim, J.; Rhee, C. K. Electrochem. Commun. 2010, 12, 1731.
(20) Leiva, E.; Iwasita, T.; Herrero, E.; Feliu, J. M. Langmuir 1997,
1
(
3, 6287.
21) Grozovski, V.; Vidal-Iglesias, F. J.; Herrero, E.; Feliu, J. M.
ChemPhysChem 2011, 12, 1641.
22) Neurock, M.; Janik, M.; Wieckowski, A. Faraday Discuss. 2009,
(
140, 363.
(23) Gao, W.; Keith, J. A.; Anton, J.; Jacob, T. J. Am. Chem. Soc. 2010,
132, 18377.
(
(
24) Wang, H.-F.; Liu, Z.-P. J. Phys. Chem. C 2009, 113, 17502.
25) Clavilier, J.; Feliu, J. M.; Aldaz, A. J. Electroanal. Chem. 1988,
ASSOCIATED CONTENT
■
2
43, 419.
*
S
Supporting Information
(26) Climent, V.; Herrero, E.; Feliu, J. M. Electrochem. Commun.
Experimental details, computational methods, and additional
figures. This material is available free of charge via the Internet
at http://pubs.acs.org.
2001, 3, 590.
27) Peng, B.; Wang, H.-F.; Liu, Z.-P.; Cai, W.-B. J. Phys. Chem. C
010, 114, 3102.
(
2
AUTHOR INFORMATION
■
Corresponding Author
herrero@ua.es
D
dx.doi.org/10.1021/ja505943h | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX