Self-supported PdxBi catalysts for the electrooxidation of glycerol in alkaline media

Article abstract of DOI:10.1021/ja412429f

Highly active self-supported Pd_xBi catalysts are synthesized by the sacrificial support method. Self-supported Pd_xBi catalysts have a porous nanostructured morphology with high surface areas (in the range from 75 to 100 m² g^-1), making Pd_xBi a state-of-the-art catalyst. Pd₄Bi displays the highest activity toward glycerol oxidation. In situ Fourier transform infrared spectroscopy highlights the unique catalytic behavior of self-supported Pd_xBi materials due to their particular structure and morphology. The confinement of reactants and intermediates in pores acting as nanoreactors is responsible for the high selectivity as a function of the electrode potential: aldehyde and ketone at low potentials, hydroxypyruvate at moderate potentials, and CO₂ at high potentials. Moreover, the selectivity depends on the electrode history: it is different for the positive potential scan direction than for the reverse direction, where the catalyst becomes selective toward the production of carboxylates.

Full text of DOI:10.1021/ja412429f

Article
pubs.acs.org/JACS
Self-Supported PdxBi Catalysts for the Electrooxidation of Glycerol in
Alkaline Media
Anna Zalineeva,† Alexey Serov,‡ Monica Padilla,‡ Ulises Martinez,‡ Kateryna Artyushkova,‡
Steve Baranton,† Christophe Coutanceau,*,† and Plamen B. Atanassov*,‡
̀
†Universite
Poitiers Cedex 9, France
́
de Poitiers, IC2MP, UMR CNRS 7285, “Catalysis and Non-conventional Media” group, 4 rue Michel Brunet, 86073
‡Department of Chemical and Nuclear Engineering and Center for Emerging Energy Technologies, University of New Mexico,
Albuquerque, New Mexico 87131, United States
ABSTRACT: Highly active self-supported PdxBi catalysts are synthesized by the sacrificial
support method. Self-supported PdxBi catalysts have a porous nanostructured morphology with
high surface areas (in the range from 75 to 100 m2 g−1), making PdxBi a state-of-the-art catalyst.
Pd4Bi displays the highest activity toward glycerol oxidation. In situ Fourier transform infrared
spectroscopy highlights the unique catalytic behavior of self-supported PdxBi materials due to
their particular structure and morphology. The confinement of reactants and intermediates in
pores acting as nanoreactors is responsible for the high selectivity as a function of the electrode
potential: aldehyde and ketone at low potentials, hydroxypyruvate at moderate potentials, and
CO2 at high potentials. Moreover, the selectivity depends on the electrode history: it is different
for the positive potential scan direction than for the reverse direction, where the catalyst becomes
selective toward the production of carboxylates.
second metal with a promotional effect.20,21 In the case of
alcohol electro-oxidation in alkaline media, it has been shown
INTRODUCTION
■
Direct alcohol fuel cells (DAFCs) have attracted more and
that interactions between palladium and bismuth modified the
more interest in the recent years1,2 as an alternative to
electrocatalytic properties,22−25 although the kind of inter-
hydrogen fed fuel cells. Alcohols are liquids, and therefore, they
actions is not yet fully understood. In particular, the bismuth
have high volumetric and gravimetric energy densities, their
deposition mode and the resulting composition of the Bi
storage is easy, and their distribution does not need new
deposit, the interaction with the substrate, and so forth have an
infrastructures. Among alcohols, glycerol has been proposed as
impact on the activity and selectivity of the Pd modified
a convenient fuel because it has a biorenewable character,3 and
this compound is a nonvalued byproduct of the biofuel
catalysts.26,27
The morphology and structure of the catalytic material can
industry. For energy production the complete oxidation of
also influence its catalytic behavior. Here, the synthesis by the
glycerol into CO2 requires the breaking of the C−C bonds,
sacrificial support method (SSM)28−30 of hierarchically
which is rather difficult to perform at the low working
structured materials, consisting in self-supported, porous, and
temperatures of DAFCs and glycerol oxidation reaction
produces a large number of reaction products.4,5 However, it
nanostructured PdxBi catalysts with different Pd/Bi atomic
ratios and with high surface area, is reported. The materials are
characterized by BET, SEM, TEM, EDS, and XPS to assess
their microstructure, morphology and composition, and by
has been proposed that all C3 oxidized species from glycerol
are valuable fine chemicals.6,7
Most of alcohol electro-oxidation studies have been
electrochemical methods to obtain information about their
performed in acidic media. The majority of anode electro-
surface. The activity of the catalysts toward glycerol oxidation
catalysts developed for DAFCs working with a proton exchange
membrane (namely, Nafion) used platinum based catalysts.8,9
Platinum is known to be the most active material for the
in alkaline medium was evaluated by cyclic voltammetry. In situ
FTIR spectroscopy is a very convenient method for
dissociative adsorption of alcohols.10,11 In contrary, in alkaline
determining the reaction products formed at the electrode/
electrolyte interface, particularly because the assignment of the
media, Pt-free catalysts can be used in both electrodes,
cathode12,13 and anode.14,15 Palladium based nanoalloys have
indeed shown high activity and stability in alkaline media, and
absorption bands was confirmed by chromatography analysis of
the reaction products after chronoamperometry measure-
ments.7,24 This technique was thus used for studying the
can be considered as possible economical substitutes of Pt for
alcohol oxidation.16−19
selectivity of the glycerol oxidation reaction as a function of the
The improvement of the catalytic activity and selectivity of
palladium electrocatalysts is generally realized by increasing the
active surface area of the catalyst and by the addition of a
Received: December 11, 2013
Published: February 18, 2014
© 2014 American Chemical Society
3937
dx.doi.org/10.1021/ja412429f | J. Am. Chem. Soc. 2014, 136, 3937−3945

Journal of the American Chemical Society
Article
electrode potential. Results allow a tentative proposal of the
relationship between structure−composition and electrocata-
lytic behavior for these materials.
EXPERIMENTAL SECTION
■
Synthesis of Self-Supported PdxBi Catalysts. A series of
catalysts with various Pd:Bi ratios were prepared by the SSM, which
was developed at University of New Mexico.28−30 A known amount of
silica Cab-O-Sil EH-5 (surface area ∼ 400 m−2 g−1) was dispersed in
water with an ultrasonic probe. Then, the appropriate amounts of
metal precursors Pd(NO3)2·xH2O (metal content = 40 wt %) and
Bi(NO3)3 (metal content = 42.98 wt %) from Sigma-Aldrich were
added to the silica suspension. Total loading of metals on silica was
calculated to be 13 wt %. The silica/precursor mixture was allowed to
dry overnight. The composite materials were reduced under hydrogen
atmosphere (7% H2) at 300 °C for 2 h. After reduction, silica template
was removed by etching in 7 M KOH solution and then abundantly
washed with water until neutral pH was achieved. The nominal Pd to
Bi atomic ratios were selected as 6:1, 4:1, and 2:1. The catalysts were
denominated as Pd, Pd6Bi, Pd4Bi, and Pd2Bi.
Physicochemical Characterization. The catalysts were compre-
hensively characterized by scanning electron microscopy (SEM;
Hitachi S-5200 Nano SEM instrument with an accelerating voltage
of 10 keV), transmission electron microscopy (TEM; JEOL 2010
TEM instrument with an accelerating voltage of 200 keV), energy
dispersion spectroscopy (EDS), X-ray photoelectron spectroscopy
(XPS; Kratos Ultra DLD spectrometer), and nitrogen adsorption (N2-
BET method using Micrometrics 2360 Gemini analyzer). SEM and
TEM provided information about the morphology of samples and size
of nanoparticles, while EDS and XPS were used to determine the
composition of the samples for comparison with the expected
composition. XPS was also used to determine the oxidation states
and the eventual interactions between different elements in the
materials.
Electrochemical Characterization. The inks for cyclic voltam-
metry (CV) experiments were prepared by dispersing 5 mg of catalyst
powder with 925 μL of a water and isopropanol alcohol (4:1) mixture
and 75 μL of Nafion (0.5 wt % from DuPont). Homogeneity of the
inks was achieved by means of sonication using an ultrasound probe.
Then, 10 μL of the mixture was applied onto a 0.2472 cm2 geometric
surface area glassy carbon disk, for a catalyst loading of 0.2 mg cm−2.
The electrochemical analysis of the synthesized material was
performed using a Pine Instrument Company electrochemical analysis
system. The cyclic voltammetry experiments were performed using a
disk electrode rotated at 1300 rpm (rpm). The electrolyte was 1 M
KOH saturated with N2 at room temperature. A platinum wire and a
Hg/HgO electrode were used as counter and reference electrodes,
respectively, although all potentials are quoted versus the reversible
hydrogen electrode (RHE). Electrocatalytic measurements were
realized in 1 M KOH + 0.1 M glycerol solution at a scan rate of
0.02 V s−1. The voltammograms were recorded in the range of
potentials from 0.0 to 1.4 V vs RHE. The catalysts were cycled through
the potential range several times until stable voltammograms were
recorded. To assess the stability of materials, chronoamperometry was
performed in the presence of 0.1 M glycerol for 5000 s at a potential of
0.7 V vs RHE.
Figure 1. (a,b) SEM and (c,d) TEM micrographs for a self-supported
Pd4Bi sample.
that the presence of beads of metal delimiting pores will
increase the active surface area and further will increase the
catalyst utilization. This is confirmed by all catalysts presenting
a well-developed 3D structure and very high surface areas
determined by N2-BET method (between 75 and 100 m2 g−1)
despite moderately high temperatures for the synthesis. TEM
images in Figure 1c and d also clearly show that the metal beads
are composed of agglomerated nanoparticles with size of ca. 5
nm. In summary, the self-supported PdxBi materials prepared
by SSM consist of beads of agglomerated nanoparticles with a
diameter of ca. 5 nm delimiting pores with sizes from 30 to 100
nm.
Cyclic voltammograms in supporting electrolyte are
presented in Figure 2. For the pure Pd material, a typical
voltammogram of polycrystalline palladium is obtained, with
the adsorption/absorption region of hydrogen at low
potentials31 (between 0.3 and −0.1 V) in the negative going
scan, the hydrogen desorption peak in the positive going scan
between −0.1 and 0.7 V, followed by the Pd surface oxide
region from 0.724,32 to 1.4 V and in the reverse scan the
reduction peak of Pd oxide species at ca. 0.6 V, according to the
following equation:
Pd + 2OH− = PdO + H2O + 2e−
(1)
Under the present experimental conditions, the electro-
chemical surface area (ECSA) of the pure self-supported Pd
catalyst can be calculated using the method described by Grden
́
et al.,32 and it was estimated at ca. 20 m2 gPd−1. This value is
four to five times lower than that determined by BET,
indicating the presence of microporosity nonaccessible to
hydrated species. But the ESCA is of the same order than that
In situ Fourier transform infrared (FTIR) spectroscopy (IRRAS -
Nicolet 6700 FT-IR spectrometer with MCT detector) was used to
gain insight into the mechanism of the glycerol electro-oxidation in
alkaline media on the self-supported Pd, Pd4Bi, and Pd2Bi catalysts.
The experimental method is described elsewhere.30
obtained by Simoes et al. with carbon supported Pd
̃
nanoparticles of 4.0 nm mean diameter,33 which is remarkable
considering that the metal is not dispersed on a high specific
surface area substrate such as carbon Vulcan XC72.
RESULTS
The presence of bismuth clearly affects the shape of the
voltammograms by limiting the hydrogen adsorption/absorp-
tion processes on palladium at low potentials, as previously
observed,34 and changing the oxidation processes at high
potentials. The hydrogen adsorption/absorption processes have
almost completely disappeared for the self-supported Pd6Bi and
■
The sacrificial support method leads to self-supported
bimetallic PdxBi materials with a spongelike structure as
shown in the SEM image of Figure 1a. At higher SEM
magnification, Figure 1b shows clearly that unsupported Pd4Bi
material displays nanostructured morphology. It is expected
3938
dx.doi.org/10.1021/ja412429f | J. Am. Chem. Soc. 2014, 136, 3937−3945

Journal of the American Chemical Society
Article
Figure 3. Detailed XPS core level spectra of Pd 3d and Bi 4f orbitals
on synthesized catalysts (a,b) Pd6Bi; (c,d) Pd4Bi; (e,f) Pd2Bi.
shapes of XPS spectra are obtained depending on the
composition of catalyst. In PdxBi materials, the 3d 5/2 core
level spectra of palladium were better fitted with four
symmetrical peaks at ca. 335, 336, 337, and 338 eV, assigned
to Pd, Pd(OH)x, PdO, and PdO2 species, respectively, as well as
the 4f 7/2 core level spectra of bismuth fitted with four peaks at
ca. 157, 158, 159, and 160 eV, which were attributed to Bi,
Bi(OH)3, BiOOH, and Bi2O3 species, respectively. The
spectrum analysis is based on data obtained from the
“Handbook of X-ray Photoelectron Spectroscopy”36 and from
previous studies by Casella and Contursi22 on bismuth adatoms
interacting with Pd surface, and also on the reasonable
assumption that higher oxidation states of species lead to
XPS peaks located at higher binding energies. Table 1
summarizes the main results of XPS analysis, as well as those
from EDS characterization. The EDS data indicate a Pd
enrichment of catalysts, whereas XPS indicated a Bi enrichment
of the surface of the catalysts, but results are relatively close to
the nominal atomic ratios. The EDS microanalyses carried out
in several different zones of the nanostructured catalysts
indicated a relatively good homogeneity of chemical
composition in catalysts.
The electroactivity of the PdxBi catalysts toward glycerol
oxidation in alkaline medium was examined by cyclic
voltammetry and compared with that of a pure Pd material.
At a scan rate of 20 mV s−1 (Figure 4a,b), an oxidation peak
appears in the positive scan direction, starting from ca. 0.6 V for
pure Pd material and Birich one (Pd2Bi) and from ca. 0.5 V for
Pd6Bi and Pd4Bi materials. The promotion effect of bismuth for
glycerol oxidation is evidenced by the shift of the oxidation
potential onset by ca. 100 mV toward lower potentials and by
the increase of the oxidation current densities in the potential
range from 0.5 to 1.2 V vs RHE. Moreover, the presence of
bismuth leads to a shift of the current drop toward higher
potentials, due to the slower kinetics of Pd surface oxide
formation on PdxBi than on Pd, which will delay the surface
deactivation by oxide species coverage. This clearly demon-
Figure 2. Cyclic voltammograms of (a) Pd and Pd6Bi self-supported
catalysts and (b) PdxBi self-supported catalysts, recorded in 0.1 M
KOH N2-saturated electrolyte (v = 0.020 V s−1; T = 25 °C).
Pd4Bi catalysts, indicating a high coverage of the palladium
surface by bismuth atoms and strong interactions between
bismuth and palladium atoms. A positive current peak related
to a surface oxidation process appears with a maximum
intensity located close to 0.9−0.95 V.
According to the potential-pH diagram of bismuth35 in an
aqueous medium, the Bi2O3 bismuth oxide phase exists as its
hydrated form Bi(OH)3, which is insoluble in alkaline solutions,
and this latter hydrated specie can be formed as soon as 0.48 V
according to the following reaction:
2Bi + 6OH− = 2Bi(OH)3 + 6e−
(2)
In the negative going potential scan, a sharp single reduction
current peak is observable between 0.55 and 0.6 V for the Pd6Bi
and Pd4Bi catalysts with a higher intensity than for the pure Pd
material. This reduction peak is attributed to the reduction of
the surface oxides, including bismuth oxides22 formed during
the positive going potential scan. This suggests that bismuth
redox process in Pd6Bi and Pd4Bi catalysts is related to the
palladium redox process, and this is evidence that electronic
interaction between both metals occurs. This strong interaction
avoids the possibility to determine the ECSA of the Bi
containing catalyst, conversely to pure Pd material, but it can
reasonably be stated that it is lower than that of the pure Pd
catalyst. For higher bismuth content, several oxidation/
reduction peaks appeared which are due to several redox
reactions of bismuth species. The reduction of oxidized Bi
surface species strongly interacting with palladium occurs at
higher potentials than that of oxidized Bi surface species weakly
or not interacting with palladium.23
The XPS core level spectra of Pd 3d and Bi 4f are presented
in Figure 3 for the three different PdxBi materials. Different
3939
dx.doi.org/10.1021/ja412429f | J. Am. Chem. Soc. 2014, 136, 3937−3945

Journal of the American Chemical Society
Article
respectively). The cyclic voltammograms are very similar to
those recorded at 20 mV s−1, except for the current drops at
high potentials which are shifted toward higher potentials when
the scan rate is increased. This phenomenon of catalyst surface
deactivation can be attributed to the slow kinetics of Pd surface
oxide formation, which will delay the surface coverage by oxide
species. No strong rate dependence in cyclic voltammograms
were observed in terms of shapes and achieved peak current
densities, which seems to indicate that the limitation of the
reaction kinetics by diffusion of species from the bulk
electrolyte toward the catalytic sites is low and that the
confinement of reactant in the catalyst pores is responsible of
the reaction kinetics. The behavior of the bismuth rich Pd2Bi
catalyst is different from those of the other ones: the peak
current density decreases with the increase of the potential scan
rate. This can be explained considering that, due to high surface
coverage of the palladium surface by bismuth species, the
reaction kinetics is limited in this case by the reactant
adsorption process.
Table 1. Nominal Atomic Ratios, Atomic Ratios Determined
from EDS Measurements, and Data from XPS
Characterization (atomic ratios, binding energy values of
XPS peaks related to Pd 3d and Bi 4f, and relative ratios of
corresponding Pd and Bi species) of Self-Supported PdxBi
Materials
nominal atom %
86/14 (6:1)
80/20 (4:1)
84.5/15.5
66/33 (2:1)
72.4/27.6
EDS μ-analysis
atom %
A1
87.1/12.9
A2
A3
87.2/12.8
86.8/13.2
85/15
70.9/29.1
70.8/29.2
71.5/28.5
61/39
EDS atom %
XPS atom %
87/13
75/25
84.5/15.5
73.5/26.5
BE
(eV)
BE
(eV)
BE
(eV)
a
a
a
atom %
atom %
atom %
XPS Pd species atom %
Pd0
335.2
336
30.7
25.4
26.9
17
335.2
335.9
336.9
338.1
35
335.3
336
25
Pd(OH)x
PdO
24
23.5
36.5
15
336.9
337.8
29.5
11.5
337.1
338
Chronoamperometry curves were recorded for 5000 s at 0.6
and 0.7 V in the presence of 0.1 M glycerol on different PdxBi
catalysts (Figure 5a and b, respectively) and compared with
PdO2
XPS Bi species atom %
Bi0
157.4
158.4
158.8
159.8
18.6
43.9
23.3
14.2
157.4
158.3
158.8
160
23
157.4
158.5
159
10
Bi(OH)3
BiOOH
Bi2O3
40
53.5
26
24.5
12.5
159.8
10.5
a
Binding energy.
Figure 5. Chronoamperometry recorded for the glycerol electro-
oxidation on a self-supported Pd and PdxBi catalysts at (a) 0.6 V vs
RHE and (b) 0.7 V vs RHE, in 1 M KOH + 0.1 M glycerol solution (T
= 25 °C).
Figure 4. Polarization curves recorded for the oxidation of 0.1 M
glycerol in 1 M KOH electrolyte on self-supported PdxBi catalysts at
different scan rates: (a,b) 0.005 V s−1, (c,d) 0.020 V s−1, and (e,f)
0.100 V s−1 (T = 25 °C).
those recorded on the pure Pd catalyst. At 0.6 V vs RHE, the
currents recorded on both Pd and Pd2Bi catalysts drop rapidly
toward a value close to zero, whereas on the Pd6Bi and Pd4Bi
−1
strates an increased the reaction rate of glycerol oxidation in the
presence of bismuth. Under these quasi stationary experimental
conditions, the following order in activity was found: Pd4Bi >
Pd6Bi > Pd2Bi > Pd.
catalyst the currents first decrease and stabilize at ca. 10 A gPd
and ca. 20 A gPd−1, respectively, the Pd4Bi material being the
most active. At 0.7 V vs RHE, the Pd4Bi material still remains
the most active catalyst. After the initial current density of
several hundreds of A gPd−1, a dramatic drop of current to ca. 70
Cyclic voltammetries were also performed at lower and
−1
higher scan rates (5 mV s−1 and 100 mV s−1 in Figure 4a and c,
A gPd occurs in a few seconds, likely related to capacitive
3940
dx.doi.org/10.1021/ja412429f | J. Am. Chem. Soc. 2014, 136, 3937−3945

Journal of the American Chemical Society
Article
current decay. Then a slow decrease and the stabilization of the
activity at ca. 60 A gPd occur, which is consistent with cyclic
lower potentials in the positive scan direction, assigned to the
formation of hydroxypyruvate,41,42 arises simultaneously with
an absorption peak at ca. 2341 cm−1 typical of interfacial CO2
production.11,40,43 At last, over the 0.9−0.6 V range in the
negative scan direction, the peak of DHA at ca. 1335 cm−1
becomes smaller than in the positive scan direction, whereas a
new absorption band at ca. 1305 cm−1 corresponding to
carboxylate groups accompanies the one of hydroxypyruvate.
Astonishingly, no absorption band located between ca. 1800
cm−1 and ca. 2100 cm−1, which would have corresponded to
bridge bonded or linear bonded CO on the noble metal
surface,44,45 has been observed over the whole studied potential
range.
−1
voltammetry (Figure 4). This behavior can be attributed to the
formation of adsorbed species from glycerol at the surface of
the catalyst leading to decreased active surface area and
decreased glycerol oxidation activity until equilibrium between
adsorption of species from glycerol and desorption of reaction
products is reached. These results not only highlight the higher
activity of Bi containing catalysts, but also the increased stability
of the Pd4Bi material.
The in situ infrared spectra recorded on the Pd4Bi catalyst for
glycerol oxidation in alkaline medium are presented in Figure 6.
For comparison, the in situ infrared spectra recorded on the
pure Pd and Birich Pd2Bi catalysts for glycerol oxidation in
alkaline medium are presented in Figure 7. In the case of pure
Pd catalyst, only the absorption band corresponding to
carboxylate groups7,24 at ca. 1306 and 1380 cm−1 are present,
whereas in the case of the Pd2Bi catalyst the absorption bands
related to carboxylates at ca. 1305 and 1380 cm−1, and to
hydroxylpyruvate at ca. 1350 cm−1 are observed, with different
relative intensity than for the Pd4Bi. A small absorption band
can also be seen at low potentials at ca. 1335 cm−1
corresponding to the formation of DHA. No absorption peak
at ca. 2343 cm−1 has been observed with both catalysts,
indicating that no CO2 is produced.
DISCUSSION
■
The beneficial effect of modifying palladium nanostructure by
bismuth for the oxidation of alcohol,25,46−48 and particularly
glycerol,24,26 was already pointed out. But, here it is shown that
in the case of self-supported PdxBi materials an optimal
bismuth/palladium atomic ratio of 1 to 4 leads to enhanced
activity. Simoes et al. observed that in the case of carbon
̃
supported PdxBi1−x, the increase of the bismuth content in the
catalyst formulation toward values higher than 10 atom % did
not change the activity of the catalyst for glycerol electro-
oxidation. On the basis of TEM, EDX, XRD, and electro-
chemical measurements, it was proposed that bismuth
deposited on the carbon support formed clusters inactive for
the glycerol oxidation reaction. The electrochemical character-
ization of PdxBi materials in supporting electrolyte (Figure 2)
indicated that the intensity and related charge in the reduction
peak at ca. 0.55 V becomes lower as the bismuth content in the
material increases; the Pd and strongly interacting Pd/Bi
surfaces, responsible for this reduction peak, are then smaller at
high bismuth content, whereas the surface of bismuth weakly or
not interacting with Pd is higher as evidenced by the reduction
peak at ca. 0.2 V vs RHE. This observation can be confronted
with the EDS and XPS data in Table 1 showing an enrichment
of the material surface by bismuth as the Bi/Pd atomic ratio is
increased. As the bismuth content is increased in the catalyst,
islands or clusters of bismuth form at the palladium surface,
decreasing the Pd surface available for glycerol adsorption and
the interface between Pd surface and Bi surface structures,
where strong interaction between both metals occurs. The
order in the catalytic activity (Pd4Bi > Pd6B1 > Pd2Bi > Pd) is
then due to the surface concentration of Bi. It has been
proposed that the activity depended on the balance between (i)
an inhibiting effect due to blocking of active sites and a catalytic
effect due to enhanced adsorption of OHads on sites adjacent to
Bi,49 (ii) the amount and composition of strongly chemisorbed
species and the course of electro-oxidation mechanism,50,51 and
Figure 6. In situ FTIR spectra recorded for the glycerol electro-
oxidation on a self-supported Pd4Bi catalyst from 0.6 V vs RHE to 1.2
V vs RHE and to 0.6 V vs RHE in 1 M KOH + 0.1 M glycerol solution
(scan rate 0.001 V s−1, T = 25 °C).
The part of the spectra in the range from 1400 cm−1 to 2000
cm−1 was removed because the presence of water leads to IR
absorption bands in this wavenumber region which super-
impose with absorption bands of other species (such as
carbonyl)37,38 and make interpretation impossible. In the IR
fingerprint region, from 1000 cm to 1400 cm−1, several
absorption bands can be observed. In previous works,7,24 the
assignment of the infrared absorption bands was confirmed by
HPLC analysis of the reaction products after chronoamperom-
etry experiments carried out at potentials where IR absorbed
bands appeared. The first one at ca. 1070 cm−1 corresponds to
CO stretching of aldehyde39,40 and is observable over the whole
potential range in the positive as well as in the negative scan
directions. In the positive scan direction, the absorption band at
1070 cm−1 is accompanied from 0.6 V to ca. 0.9 V by two bands
located at ca. 1100 cm−1 and at ca. 1335 cm−1, related to CO
stretching of alcohol groups and to dihydroxyacetone (DHA),41
respectively. For potentials between 0.9 and 1.2 V, in both
positive and negative scan directions, a well define absorption
peak at ca. 1350 cm−1, which only appeared as a shoulder for
3941
dx.doi.org/10.1021/ja412429f | J. Am. Chem. Soc. 2014, 136, 3937−3945

Journal of the American Chemical Society
Article
self-supported Pd4Bi material, in terms of selectivity is very
interesting and unique. The first important observation is its
high selectivity at low overpotentials (from 0.6 V vs RHE to ca.
0.8 V vs RHE) toward low oxidized compounds, that is,
glyceraldehyde (GAl) and dihydroxyaceone (DHA), as
evidenced by the absorption bands at 1070 and 1335 cm−1,
respectively, appearing in this potential range. This indicates
that the increase of activity induced by bismuth does not imply
the so-called bifunctional mechanism53 where it is proposed
that surface species from alcohol adsorption are oxidatively
removed thanks to the presence of surface OH species from
water adsorption following a Langmuir−Hinshelwood mecha-
nism. Indeed, the formation of aldehyde and ketone groups
from alcohols does not need the addition of extra oxygen
atoms, so that the enhancement of activation is due to the
increased turnover frequency of reactants on the catalytic
surface, possibly induced by a different adsorption mode of
glycerol because Pd adsorption site dilution by bismuth atoms.
Adzic also proposed that Bi adatoms prevents the adsorption
and the formation of the strongly bonded intermediates which
occupied multiple surface sites.52 In acidic media, Koper et al.
observed that the presence of bismuth salt in the electrolyte
lowered the onset potential of oxidation on a Pt/C electrode
and enhanced the turnover frequency by forming a bismuth-
related active site on the surface poised for secondary alcohol
oxidation.54
As soon as the electrode potential is increased, the formation
of hydroxypyruvate (HyP) occurs as evidenced by the shoulder
appearing at 1350 cm−1 from ca. 0.8 V vs RHE. In this case, the
formation of the carboxylate function implies the addition of
extra oxygen atoms and, therefore, the bifunctional mechanism.
It is worth noting that, in previous works on classical carbon
supported Pd and PdBi based nanocatalysts,7,24 the formation
of GAl and DHA was observed by in situ FTIR measurement
and HPLC analyses, whereas the formation of HyP was not
detected on such catalysts, conversely to what is observed here.
Such a compound was only detected with a pure Au/C
nanocatalyst.42 It was proposed that an equilibrium between
both DHA and GAl isomers existed.24 In the case of high
potentials, the oxidation reaction kinetics of a primary alcohol
group toward carboxylate group becomes significantly faster
than the isomerization reaction kinetics of DHA into GAl and
HyP can be formed in such amount sufficient for its detection.
The discrepancies between results obtained with carbon
supported catalysts and self-supported materials come from
the particular morphology and structure of the self-supported
Pd4Bi which are responsible of this unique behavior: the pores
of size in the range from ca. 30 to 100 nm work as confined
nanoreactors leading to such selectivity.
For higher potentials, the production of CO2 evidenced by
the sharp absorption band at ca. 2343 cm−1, starts just after the
detection of hydroxypyruvate whereas the absorption band
assigned to DHA tends to disappear. Here, the breaking of the
C−C bond has obviously occurred, and it is likely that in
addition to CO2, other C1 and C2 species are also formed,
which are not detected. Indeed, the absorption peak at ca. 1240
cm−1 assigned to glycolate and at 1330 cm−1 assigned to oxalate
are not present in Figure 6a. The absorption band a ca. 1400
cm−1 assigned to carbonate and at ca. 1577 cm−1 assigned to
glycolate cannot be observed with the IRRAS method, because
of the presence of the interfacial water absorption band which
prevent the accurate determination of the other absorption
bands. Again, from previous work on carbon supported
Figure 7. In situ FTIR spectra recorded for the glycerol electro-
oxidation on (a) Pd and (b) Pd2Bi self-supported catalysts from 0.6 V
vs RHE to 1.2 V vs RHE in 1 M KOH + 0.1 M glycerol solution (scan
rate 0.001 V s−1, T = 25 °C).
(iii) the change in adsorbed species implied by the Pd surface
atoms dilution by Bi surface atoms and strength of
adsorption.52 Note that all these reasons also stand for
explaining the higher stability of the self-supported Pd4Bi
materials toward glycerol oxidation shown in Figure 5.
Beyond the activity, it is also known that the modification of
platinum group metals surface by foreign atoms influences the
selectivity of the alcohol oxidation reaction. From studies on
the electro-oxidation of alcohol containing more than one
carbon atom on platinum modified by adatoms, Petrii and co-
workers50,51 concluded that the presence of adatoms led to
significant effects on the amount and course of electro-
oxidation of strongly bonded species and that it was also
possible that they affect the composition of chemisorbed
species (and therefore the selectivity of the catalyst). In
addition, the behavior of the most active catalyst, that is, the
3942
dx.doi.org/10.1021/ja412429f | J. Am. Chem. Soc. 2014, 136, 3937−3945

Journal of the American Chemical Society
Article
nanostructured PdxBiy/C catalyst, no production of CO2 was
observed, and it was proposed that the dilution of Pd atoms by
Bi atoms led to change the adsorption of glycerol, avoiding the
breaking of the C−C bond. Here, the confined nanoreactor as
proposed before can again be invoked to explain this ability to
break the C−C bond at high potentials with HyP as an
intermediate, according to following reactions:
surface by bismuth dilutes Pd surface atoms, avoiding the
absorption mode leading to C−C bond breaking, and further to
CO2 formation.
Considering the use of the Pd4Bi catalyst as anode material in
an alkaline electrolysis cell, low glycerol conversion and
hydrogen production rate would occur at low overvoltage,
but with production of high value added ketone/aldehyde
products of great interest for pharmaceutical, polymer, food,
etc., industries, whereas high glycerol conversion into CO2 and
hydrogen production rate would occur at higher overvoltages,
according to the following equations:
CH2OH−CO−COO− + 2OH−
→ CH2OH−COO− + CO2 + H2O + 2e−
(1)
at low electrolysis cell voltages (low anode overpotentials):
and/or
CH2OH−CO−COO− + 7OH−
→ COO−−COO− + CO2 + 5H2O + 6e−
CH2OH−CHOH−CH2OH
→ CH2OH−CO−CH2OH + H2
(4)
(2)
and
and/or
CH2OH−CO−COO− + 7OH− → 3CO2 + 5H2O + 8e−
CH2OH−CHOH−CH2OH
→ CH2OH−CHOH−CHO + H2
(3)
(5)
The slow diffusion of products from the pores of the catalysts
toward the bulk electrolyte leads to increased residence time in
the catalytic layer, so that overoxidation of hydroxypyruvate
toward CO2 can occur. This may explain the absence of CO
absorption bands although CO2 is formed, by considering that
this step related to the presence of hydroxypyruvate does not
involve the formation of strongly adsorbed CO, making its
further oxidation to CO2 easier as it was proposed elsewhere in
the case of ethanol electro-oxidation involving acetaldehyde
intermediate at PtSn catalyst.11,55 The formation of CO2 can be
explained by the local pH decrease; the depletion of hydroxyls
species in the nanopores is a consequence of their consumption
in the several electrochemical steps.
In the case of the pure Pd material, the in situ FTIR study
only showed the formation of carboxylate species; the
absorption bands at ca. 1335 and 1350 cm−1 assigned to
DHA and hydroxypyruvate, respectively, were not detected.
This can be explained by considering that the oxidation of
glycerol occurs according to the bifunctional mechanism
involving the formation of hydroxyl species at the Pd surface
and reaction with adsorbed species from glycerol to give
directly carboxylate species. Then, the absence of CO2 at higher
potentials is directly due to the fact that as soon as the Pd
surface is oxidized, the catalyst loses its activity as evidenced by
the decrease of the oxidation current in voltammograms of
Figure 4. The difference in behavior between both catalysts,
pure Pd and Pd4Bi, is the delay for the formation of activated
oxygenated species at the catalyst surface allowing the
Langmuir−Hinshelwood mechanism to occur. Bismuth may
play the role of oxygen pump protecting Pd surface from
oxidation, as it was proposed to explain the better activity of
PtM catalysts toward oxygen reduction reaction.56 For Pd2Bi,
the same mechanism as for Pd4Bi stands for low and
intermediate potentials as the absorption band at 1350 cm−1
assigned to hydroxypyruvate formation is observed; this
confirms the important role of bismuth not only toward
catalytic activity but also selectivity. However, formation of
carboxylate is also confirmed by the absorption bands at ca.
1306 and 1380 cm−1 while no CO2 was detected at higher
potentials, as for pure Pd catalyst. Again, this is related to the
drop of the catalytic activity at high potentials as evidenced in
Figure 4. But the reason is different than that invoked for the
pure Pd catalyst. Here, the high coverage of the palladium
at high electrolysis cell voltages (high anode overpotentials)
CH2OH−CHOH−CH2OH + H2O
→ CH2OH−CO−COOH + 3H2
(6)
and
CH2OH−CHOH−CH2OH + 3H2O → 3CO2 + 7H2
(7)
If we consider that the oxidation of glycerol into CO2 could
also leads to the formation of other C1 and C2 species
(although not detected), the number of hydrogen molecules
produced per oxidized glycerol molecule should range between
3 and 7. Moreover, in all cases, the energy of hydrogen
production would be at least twice to three times lower than in
the case of water electrolysis, as it only depends on the
electrode potential.57 For water electrolysis, the onset potential
for the anodic oxygen evolution reaction is higher than 1.23 V
vs RHE (the standard water oxidation potential), whereas that
for the glycerol oxidation at a Pd4Bi catalyst is ca. 0.5 V vs RHE.
At last, it is worth noting a new absorption peak assigned to
carboxylate group a ca. 1305 cm−1 which appears below 0.95 V
in the negative potential scan direction. In this scan direction,
the catalyst surface is first covered with oxide species at high
potentials which can transform into hydroxide surface species
as the electrode potential is lowered. These species are able to
provide the extra atoms of oxygen needed for the oxidation of
glycerol or adsorbed species from glycerol into carboxylate
species, according to the bifunctional mechanism. In contrary,
in the positive scan direction, palladium surface is most likely
under reduced form at low electrode potentials and surface
hydroxide species can only be formed for potentials higher than
0.7−0.8 V vs RHE, explaining the formation of aldehyde/
ketone species at low potentials and hydroxypyruvate at higher
potentials.
CONCLUSION
■
A series of highly active PdxBi catalysts were synthesized by the
combination of the sacrificial support method and a thermal
reduction process. Self-supported PdxBi catalysts have a porous
morphology corresponding to beads of agglomerated 5 nm
nanoparticles delimiting pores of 30−100 nm diameters. Such a
3943
dx.doi.org/10.1021/ja412429f | J. Am. Chem. Soc. 2014, 136, 3937−3945

Journal of the American Chemical Society
Article
structure leads to high surface areas, in the range of 75−100 m2
g−1, making PdxBi state-of-the-art catalysts. All self-supported
PdxBi displayed higher activity toward glycerol oxidation,
confirming the promotion effect of bismuth, but the Pd4Bi,
which was found to be the most active. The most important
highlights are the following: (i) The particular morphology and
structure of the self-supported Pd4Bi material leads to the
formation of pores acting as nanoreactors. (ii) The confinement
of reactants, intermediates, and products in the nanoreactors is
responsible for the unique behavior of Pd4Bi material in terms
of selectivity for the glycerol electrooxidation as a function of
the electrode potential. (iii) The selectivity is dependent on the
electrode history, being different for the positive potential scan
direction than for the reverse direction. (iv) In the positive
potential scan direction, the catalyst is highly selective toward
aldehyde and ketone at low electrode potentials. (v) At high
potentials, again thanks to its particular morphology, the
catalyst is able to break the C−C bond and to perform the
complete oxidation of glycerol into CO2 through hydroxypyr-
uvate intermediate. (vi) At last, in the negative potential scan
direction, the catalyst becomes selective toward the production
of carboxylates.
(9) Calvillo, L.; Celorrio, V.; Moliner, R.; Garcia, A. B.; Camea
Lazaro, M. J. Electrochim. Acta 2013, 102, 19−27.
(10) Hogarth, M. P.; Ralph, T. R. Platinum Met. Rev. 2002, 46, 146−
164.
(11) Leg
Electrochim. Acta 2005, 50, 5118−5125.
(12) Lefevre, M.; Proietti, E.; Jaouen, F.; Dodelet, J.-P. Science 2009,
324, 71−74.
́
n, I.;
́
er, J. -M.; Rousseau, S.; Coutanceau, C.; Hahn, F.; Lamy, C.
̀
(13) Lu, S.; Pan, J.; Huang, A.; Zhuang, L.; Lu, J. Proc. Natl. Acad. Sci.
U.S.A. 2008, 105, 20611−20614.
(14) Marchionni, A.; Bevilacqua, M.; Bianchini, C.; Chen, Y. X.;
Filippi, J.; Forniasero, P.; Lavacchi, A.; Miller, H.; Wang, L.; Vizza, F.
ChemSusChem 2013, 6, 518−528.
(15) Ilie, A.; Simoes, M.; Baranton, S.; Coutanceau, C.; Martemianov,
̃
S. J. Power Sources 2011, 196, 4965−4971.
(16) Xu, C.; Cheng, L.; Shen, P.; Liu, Y. Electrochem. Commun. 2007,
9, 997−1001.
(17) Bianchini, C.; Shen, P. K. Chem. Rev. 2009, 109, 4183−4206.
(18) Tian, N.; Zhou, Z.-Y.; Yu, N.-F.; Wang, L.-Y.; Sun, S.-G. J. Am.
Chem. Soc. 2010, 132, 7580−7581.
(19) Ye, K.-H.; Zhou, S.-A.; Zhu, X.-C.; Xu, C.-W.; Shen, P. K.
Electrochim. Acta 2013, 90, 108−111.
́
(20) Leger, J.-M.; Coutanceau, C.; Lamy, C. Electrocatalysis for the
direct alcohol fuel cell. In Fuel Cell Catalysis: a surface science approach;
Koper, M., Ed.; Wiley-VCH: Weinheim, 2009; pp 337−367.
This work represents an important demonstration that the
combination of the control of the catalyst composition/
morphology/structure and of the electrode potential allows
orientating selectively the oxidation of alcohols, particularly
glycerol, toward a given valuable product for electrosynthesis
application, or toward CO2 production for energy conversion
applications (hydrogen production).
́
(21) Rousseau, S.; Coutanceau, C.; Lamy, C.; Leger, J.-M. J. Power
Sources 2006, 158, 18−24.
(22) Casella, I. G.; Contursi, M. Electrochim. Acta 2006, 52, 649−657.
(23) Demarconnay, L.; Brimaud, S.; Coutanceau, C.; Leger, J.-M. J.
́
Electroanal. Chem. 2007, 601, 169−180.
(24) Simoes, M.; Baranton, S.; Coutanceau, C. Appl. Catal., B 2011,
̃
110, 40−49.
(25) Cai, J.; Huang, Y.; Guo, Y. Electrochim. Acta 2013, 99, 22−29.
(26) Zalineeva, A.; Baranton, S.; Coutanceau, C. Electrochem.
Commun. 2013, 34, 335−338.
AUTHOR INFORMATION
■
Corresponding Authors
christophe.coutanceau@univ-poitiers.fr
plamen@unm.edu
(27) Zalineeva, A.; Simoes, M.; Baranton, S.; Coutanceau, C. Int. J.
̃
Hydrogen Energy 2013, submitted.
(28) Serov, A.; Robson, M. H.; Artyushkova, K.; Atanassov, P. Appl.
Catal., B 2012, 127, 300−306.
Notes
The authors declare no competing financial interest.
(29) Serov, A.; Robson, M. H.; Halevi, B.; Artyushkova, K.;
Atanassov, P. Electrochem. Commun. 2012, 22, 53−56.
(30) Serov, A.; Robson, M. H.; Smolnik, M.; Atanassov, P.
Electrochim. Acta 2013, 109, 433−439.
ACKNOWLEDGMENTS
■
A.Z. thanks the Center for Emerging Energy Technologies,
Farris Engineering Center (UNM) for assistance and technical
support, and the county council of Poitou-Charentes (France)
for financial support. UNM portion of this work was funded in
part by DOE-BES EPSCoR Implementation Award: “Materials
for Energy Conversion”.
(31) Conway, B. E.; Jerkiewicz, G. J. Electroanal. Chem. 1993, 357,
47−66.
́
(32) Grden, M.; Łukaszewski, M.; Jerkiewicz, G.; Czerwinski, A.
Electrochim. Acta 2008, 53, 7583−7598.
(33) Simoes, M.; Baranton, S.; Coutanceau, C. J. Phys. Chem. C 2009,
̃
113, 13369−13376.
(34) Simoes, M.; Baranton, S.; Coutanceau, C. Electrochim. Acta
̃
2010, 56, 580−591.
REFERENCES
■
́ ́ ̀
(35) Van Muyder, J.; Pourbaix, M. Atlas d’equilibres electrochimiques a
(1) Lamy, C.; Coutanceau C. Electrocatalysis of alcohol oxidation at
platinum group metals. In Catalysis for Alcohol Fuelled Direct Oxidation
Fuel Cells; Liang, Z.-X., Zhao T. S. Eds.; RSC Energy and Science
Series No. 6; Royal Society of Chemistry: Cambridge, 2012; pp 1−70.
(2) Lai, S. C. S.; Koper, M. T. M. Phys. Chem. Chem. Phys. 2009, 11,
10446−10456.
(3) Serov, A.; Kwak, C. Appl. Catal., B 2010, 97, 1−12.
(4) Bambagioni, V.; Bevilacqua, M.; Filippi, J.; Marchionni, A.;
Marchionni, A.; Moneti, S.; Vizza, F.; Bianchini, C. Chim. Oggi 2010,
28, VII−X.
25 °C; Gauthier-Villars & Cie: Paris, 1963; pp 533−539.
(36) Wagner, C. D., Riggs, W. M., Davis, L. E., Moulder, J. F.,
Mouilenberg, G. E., Eds. Handbook of Xray Photoelectron Spectroscopy;
Perkin Elmer Corporation: Eden Prairie, MN, 1978; p 1.
(37) Xia, X. H.; Liess, H. -D.; Iwasita, T. J. Electroanal. Chem. 1997,
437, 233−240.
(38) Vigier, F.; Coutanceau, C.; Hahn, F.; Belgsir, E. M.; Lamy, C. J.
Electroanal. Chem. 2004, 563, 81−89.
(39) Falase, A.; Main, M.; Garcia, K.; Serov, A.; Lau, C.; Atanassov, P.
Electrochim. Acta 2012, 66, 295−301.
(5) Bambagioni, V.; Bianchini, C.; Marchionni, A.; Filippi, J.; Vizza,
F.; Teddy, J.; Serp, P.; Zhiani, M. J. Power Sources 2009, 190, 241−251.
(6) Behr, A.; Eilting, J.; Irawadi, K.; Leschinski, J.; Lindner, F. Green
Chem. 2008, 10, 13−30.
(40) Wang, L.; Meng, H.; Shen, P. K.; Bianchini, C.; Vizza, F.; Wei,
Z. Phys. Chem. Chem. Phys. 2011, 13, 2667−2673.
(41) Pouchert, C. J. The Aldrich Library of Infrared Spectra, 3rd ed.;
Aldrich Chemical Company, Inc.: Milwaukee, WI, 1981; p 1.
(7) Simoes, M.; Baranton, S.; Coutanceau, C. ChemSusChem 2012, 5,
̃
(42) Simoes, M.; Baranton, S.; Coutanceau, C. Appl. Catal., B 2010,
2106−2124.
̃
(8) Chen, A.; Holt-Hindle, P. Chem. Rev. 2010, 110, 3767−3804.
93, 354−362.
3944
dx.doi.org/10.1021/ja412429f | J. Am. Chem. Soc. 2014, 136, 3937−3945

Journal of the American Chemical Society
Article
(43) Dubau, L.; Hahn, F.; Coutanceau, C.; Leg
Electroanal. Chem. 2003, 554−555, 407−415.
(44) Kunimatsu, K. J. Electroanal. Chem. 1982, 140, 205−210.
(45) Beden, B.; Hahn, F.; Juanto, S.; Lamy, C.; Leger, J.-M. J.
Electroanal. Chem. 1987, 225, 215−225.
(46) Motoo, S.; Watanabe, M. J. Electroanal. Chem. 1979, 98, 203−
211.
(47) Beden, B.; Kadirgan, F.; Lamy, C.; Leg
Chem. 1982, 142, 171−190.
́
er, J. -M.; Lamy, C. J.
́
́
er, J.-M. J. Electroanal.
(48) Casado Rivera, E.; Volpe, D. J.; Alden, L.; Lind, C.; Downie, C.;
VazquezAlvarez, T.; Angelo, A.; DiSalvo, F. J.; Abruna, H. J. Am. Chem.
Soc. 2004, 126, 4043−4049.
(49) Schmidt, T. J.; Behm, R. J.; Grgur, B. N.; Markovic, N. M.; Ross,
P. N., Jr. Langmuir 2000, 16, 8159−8166.
(50) Podlovchenko, B. I.; Petrii, O. A.; Frumkin, A. N.; Lal, H. J.
Electronal. Chem. 1966, 11, 12−25.
(51) Smirnova, N. W.; Petrii, O. A.; Grzejdziak, A. J. Electroanal.
Chem. 1988, 251, 73−87.
(52) Adzic, R. R. Electrocatalysis on surfaces modified by foreign
metal adatoms. In Advances in Electrochemistry and Electrochemical
Engineering; Gerisher, H., Tobias, C. W., Eds.; Wiley- Interscience:
New York, 1984; Vol. 13, pp 159−260.
(53) Watanabe, M.; Motoo, S. J. Electroanal. Chem. 1975, 60, 275−
283.
(54) Kwon, Y.; Birdja, Y.; Spanos, I.; Rodriguez, P.; Koper, M. T. M.
ACS Catal. 2012, 2, 759−764.
́
(55) Coutanceau, C.; Brimaud, S.; Dubau, L.; Lamy, C.; Leger, J.-M.;
Rousseau, S.; Vigier, F. Electrochim. Acta 2008, 53, 6865−6880.
(56) Shukla, A.; Neergat, M.; Parthasarathi, B.; Jayaram, V.; Hegde,
M. S. J. Electroanal. Chem. 2011, 504, 111−119.
(57) Lamy, C.; Jaubert, T.; Baranton, S.; Coutanceau, C. J. Power
Sources 2014, 245, 927−936.
3945
dx.doi.org/10.1021/ja412429f | J. Am. Chem. Soc. 2014, 136, 3937−3945