On the preparation of molybdenum carbide-supported metal catalysts

Article abstract of DOI:10.1016/j.jcat.2012.02.012

Carbide-supported metal catalysts have been demonstrated to be highly active for a number of reactions including water gas shift and electrochemical hydrogen evolution. Their activity appears to be a consequence of intimate interactions between the metal and support. These interactions as well as the structures and compositions depend on the conditions used during preparation. The objective of research described in this paper was to define key aspects regarding the adsorption and reaction of metal precursor complexes with unpassivated, high surface area Mo₂C powders during wet impregnation. The rate of metal deposition and the final state of the metal were functions of the nature of the precursor (i.e., counter ion), electrostatic interactions, and red-ox reactions with the Mo₂C surface. The deposition rate decreased in the following order: H₂PtCl₆ ～ Pd(NH ₃)₄(NO₃)₂ ～ CuCl₂ ～ Cu(NO₃)₂ > Co(NO₃)₂ ～ Ni(NO₃)₂ ? CoCl₂ ～ NiCl₂ ～ FeCl₂ ～ Fe(NO₃)₃. The observed adsorption and red-ox behavior suggest that supporting metals onto native carbide surfaces could result in properties that are significantly different from those for metals supported on oxidized carbide surfaces.

Full text of DOI:10.1016/j.jcat.2012.02.012

Journal of Catalysis 289 (2012) 210–217
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
On the preparation of molybdenum carbide-supported metal catalysts
⇑
Joshua A. Schaidle, Neil M. Schweitzer, Olabode T. Ajenifujah, Levi T. Thompson
Department of Chemical Engineering and Hydrogen Energy Technology Laboratory, University of Michigan, Ann Arbor, MI 48109-2136, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Carbide-supported metal catalysts have been demonstrated to be highly active for a number of reactions
including water gas shift and electrochemical hydrogen evolution. Their activity appears to be a conse-
quence of intimate interactions between the metal and support. These interactions as well as the struc-
tures and compositions depend on the conditions used during preparation. The objective of research
described in this paper was to define key aspects regarding the adsorption and reaction of metal precur-
sor complexes with unpassivated, high surface area Mo2C powders during wet impregnation. The rate of
metal deposition and the final state of the metal were functions of the nature of the precursor (i.e., coun-
ter ion), electrostatic interactions, and red–ox reactions with the Mo2C surface. The deposition rate
decreased in the following order: H2PtCl6 ꢀ Pd(NH3)4(NO3)2 ꢀ CuCl2 ꢀ Cu(NO3)2 > Co(NO3)2 ꢀ Ni(NO3)2
ꢁ CoCl2 ꢀ NiCl2 ꢀ FeCl2 ꢀ Fe(NO3)3. The observed adsorption and red–ox behavior suggest that support-
ing metals onto native carbide surfaces could result in properties that are significantly different from
those for metals supported on oxidized carbide surfaces.
Received 26 November 2011
Revised 15 February 2012
Accepted 16 February 2012
Available online 27 March 2012
Keywords:
Molybdenum carbide
Wet impregnation
Metal deposition
Ó 2012 Elsevier Inc. All rights reserved.
1. Introduction
forms prior to deposition of the metal. Like other highly reduced
nanoscale materials, high surface area early transition metal car-
Most heterogeneous catalysts consist of an ‘‘active species’’ dis-
persed on a high surface area support. The active species directly
influences the catalytic rates and selectivities, while the support
typically affects the size and morphology of the active species do-
mains [1]. For some systems, the support can also modify elec-
tronic properties of the active species and/or participate directly
in the reaction as a co-catalyst. Oxides such as alumina and silica
are commonly used as supports for heterogeneous catalysts; how-
ever, new types of supports are being investigated to drive old and
new reactions more efficiently. Early transition metal carbides are
promising candidates for use as catalyst supports [2–5]. These
materials can be produced in high surface form [6] and have cata-
lytic properties that can resemble those of platinum group metals
[6,7]. Carbide-supported metal catalysts have been reported to be
active for reactions including Fischer–Tropsch Synthesis [8,9], hyd-
rodenitrogenation and hydrodesulfurization [10], methanol steam
reforming [2,5] and water gas shift [3,4]. These materials are also
active for electrocatalytic reactions [11–16]. For example, Espositio
et al. [11] reported that a monolayer of Pt supported on WC exhib-
ited hydrogen evolution rates comparable with that for bulk Pt.
A number of methods have been used to prepare carbide-sup-
ported metal catalysts including thermal evaporation in a vacuum
environment [11], atomic layer deposition [13], and wet impregna-
tion [8–10]. In most cases, the carbides were in their passivated
bides are often passivated before exposure to air to avoid deep oxi-
dation of the materials. While the resulting materials can be highly
active, they are best described as metals supported on oxycarbide
or oxide surfaces. Setthapun et al. [2] and Schweitzer et al. [4] used
a wet impregnation method to deposit metals directly onto high
surface area carbide supports. The resulting materials tended to
be more active than those prepared using passivated supports
[17]. The deposition of Pt directly onto a native Mo2C surface pro-
duced Pt/Mo2C catalysts with thin ‘‘raft-like’’ Pt particles and very
high Pt dispersions [4]. Water gas shift activities for these catalysts
exceeded those for oxide supported Pt catalysts by in some cases
more than an order of magnitude. The high activities were attrib-
uted to a high density of active sites at the interface between the
raft-like Pt particles and Mo2C support. Motivated by observations
like this, we have been examining the nature of interactions be-
tween late transition metals and early transition metal carbide
surfaces.
The objective of the research described in this paper was to
investigate the kinetics and mechanism(s) involved in the deposi-
tion of metals on native Mo2C surfaces during wet impregnation
with aqueous metal precursor solutions. Like other types of heter-
ogenous processes [18], the preparation of supported catalysts in-
volves multiple steps including transport of the metal ion/complex
to the support surface, adsorption of the ion/complex, and possibly
reaction at the support surface. There are a number of factors that
could affect interactions of the metal precursors with the support
surfaces: the nature of the support (e.g., point of zero charge),
⇑
Corresponding author.
E-mail address: ltt@umich.edu (L.T. Thompson).
0021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.02.012

J.A. Schaidle et al. / Journal of Catalysis 289 (2012) 210–217
211
solution pH, and chemistry of the metal precursors including their
ionic strengths [19]. Brunelle was among the first to report that
electrostatic interactions can play a role in the adsorption of metal
complexes on supports [19]. Electrostatic adsorption is governed
by double layer theory [20] and has been investigated in depth
by Regalbuto et al. [21–23]. According to the electrostatic model,
the solution pH, the charge of the metal complex in solution, and
the point of zero charge (PZC, pH at which the surface of the mate-
rial has a net zero charge) of the support are the key parameters
affecting metal adsorption [19,21–23]. Electrostatic binding does
not, however, account for the adsorption of all metals on supports
[20,24], so more complex theories have been developed. These in-
clude the triple layer theory, specific site adsorption, and ligand ex-
change mechanisms [20].
In addition to transport and adsorption, the metal precursor
could react with the support surface. These reactions can produce
mixed molecular species that remain on the surface of the support
[20]. One example is the formation of layered nickel silicate struc-
tures during the adsorption of nickel(II) ammine complexes on sil-
ica [25,26]. The metal precursors can also be involved in red–ox
reactions with the support surface. The synthesis of bimetallic me-
tal particles is often accomplished via the direct red–ox reaction of
a parent metal and a second metal in oxidized form [24,30–31].
Since transition metal carbides behave like metals (band structure
of carbides is metallic-like, that is, the d-band of the metal is situ-
ated around the Fermi level and there is no bandgap) [27–29], red–
ox reactions could be involved in interactions between the metals
and carbide surfaces.
In this paper, we describe the preparation of Mo2C-supported
Pt, Pd, Cu, Co, Ni, and Fe catalysts using a wet impregnation meth-
od. Special precautions were taken to preserve the native carbide
surface chemistry and structure by avoiding exposure of the mate-
rials to air; this allowed the metal precursors to interact directly
with Mo2C. Uptake of the metal was monitored as a function of
time, and the resulting materials were characterized using X-ray
diffraction (XRD), X-ray absorption spectroscopy (XAS), and X-ray
photoelectron spectroscopy (XPS). The results were used to define
key aspects of the mechanisms governing the deposition of metals
onto Mo2C and could help establish a scientific basis for the prep-
aration of other carbide-supported metal catalysts.
area of ꢀ120 m2/g) assuming complete deposition of the precursor.
The nominal metal loading would be 1.76 l
mol metal per m2. The
solution was purged with Ar gas for more than 15 min in order to re-
move any dissolved oxygen. The freshly synthesized Mo carbide
powder was then transferred, under an Ar blanket, to the solution
as it was continuously purged. For temporal analysis of the amount
deposited, the support powder remained in the precursor solution
for up to 45 h with continuous Ar purging and samples of the solu-
tion were extracted for elemental analysis via inductively coupled
plasma-optical emission spectroscopy (ICP-OES). Finally, the result-
ing powders were transferred back into the quartz reactor tube, un-
der an Ar blanket gas.
Some of the solid materials were dried in a He stream flowing at
20 mL/min at room temperature over night. The resulting materi-
als were then stored in a glove box until analyzed by XAS. These
materials are referred to as ‘‘dried.’’ Some of the solid materials
were dried at 110 °C for 2 h in H2 and then reduced in H2 at
450 °C for 4 h. After reduction, the resulting materials were also
transferred into a glove box for subsequent characterization. These
materials are referred to as ‘‘reduced.’’ Materials not intended for
XAS analysis were passivated in 1% O2/He flowing at 20 mL/min
at room temperature for 6 h after the synthesis.
2.2. Materials characterization
The PZC of the Mo carbide was determined according to the pro-
cedure described by Park and Regalbuto [32]. Eighteen milliliters of
deionized water was purged with Ar, and the solution pH was ad-
justed to the desired value using either concentrated HCl or
NH4OH. After the pH stabilized, the unpassivated Mo carbide pow-
der was transferred to the solution under an Ar blanket. The solu-
tion was continuously stirred, and the pH of the solution was
recorded as
a function of time. The surface loading was
ꢀ5000 m2/L, similar to the value used in Park and Regalbuto [32].
The metal loading was determined using one of two procedures.
The first involved direct characterization of the powder product.
Approximately 15 mg of catalyst was dissolved in 3 mL of aqua re-
gia solution (75 vol.% HCl and 25 vol.% HNO3). The solution was al-
lowed to stabilize overnight. One milliliter of this solution was
diluted with 13 mL of deionized water then analyzed using ICP-
OES with a Varian 710-ES ICP Optical Emission Spectrometer. The
second method involved analysis of the precursor solutions after
contact with the support. For this method, 1 mL of the solution
was diluted with 13 mL of deionized water then analyzed using
ICP-OES. For each sample, the relative intensities were compared
with those for standards of known concentrations.
2. Materials and methods
2.1. Catalyst preparation
The Mo carbide supports were synthesized using a temperature
programmed reaction procedure [3,4]. Ammonium paramolybdate
((NH4)6Mo7O24ꢂ4H2O; 81–83% as MoO3; Alpha Æsar) was crushed
and sieved to retain particles with sizes between 125 and
X-ray diffraction analysis was performed using a Rigaku Mini-
flex Diffractometer with Cu
Ka radiation and a Ni filter
(k = 1.540 Å). The 2h range (10° < 2h < 90°) was scanned at a rate
of 5°/min with a 0.02° step size. X-ray absorption spectroscopy
was performed at the Argonne National Laboratory Advanced Pho-
250
lm. The powder (ꢀ1.3 g) was then supported in a tubular
quartz reactor with quartz wool and placed in a vertical furnace.
The powder was exposed to H2 flowing at 400 mL/min and heated
to 350 °C at 5 °C/min and held at this temperature for 12 h. The gas
was then changed to 15% CH4/H2 flowing at 375 mL/min, and the
temperature was increased at 2.67 °C/min to 590 °C and held at
this temperature for 2 h. The resulting materials were then cooled
to room temperature in 15% CH4/H2. A single batch produced
ꢀ0.75 g of Mo carbide.
ton Source beamline MR-CAT 10-ID-B. Spectra at the Fe K
(7112 eV), Ni K (8333 eV), Cu K (8979 eV), Pd K (24,350 eV), and
Pt L3 (11,564 eV) edges were collected in fluorescence mode using
Ar ion chambers for detection. Materials for XAS characterization
were prepared in a glove box to ensure that the materials were
completely isolated from air so that neither the metal nor Mo car-
bide support was oxidized.
The XAS results were analyzed using free online software Athe-
na utilizing standard analysis procedures. The background was re-
moved by fitting the pre-edge data to a linear function and
subtracting out the result. The curves were then normalized by fit-
ting the post-edge data to a third-order polynomial and dividing
the data by the extrapolated value at Eo, the energy at the step
edge. Finally, the background of the post-edge data was removed
using a cubic spline function. The R-space data were plotted by first
Metals were deposited onto the supports via a wet impregnation
method. Precursors containing different metals and ligands/counter
ions were used. First, an appropriate amount of Pd(NH3)4(NO3)2,
H2PtCl6, Cu(NO3)2, CuCl2, Co(NO3)2, CoCl2, Ni(NO3)2, NiCl2, Fe(NO3)3,
or FeCl2 was dissolved in 70 mL of deionized water to produce solu-
tions containing 2.27 mmol metal per liter. This metal concentration
was chosen to achieve ꢀ10% coverage of the Mo2C surface (surface

212
J.A. Schaidle et al. / Journal of Catalysis 289 (2012) 210–217
converting the post-edge data into k-space with a k2 weighting.
Then, the k-space data were converted into R-space using a Fourier
transform.
Some of the Mo2C-supported metal catalysts were character-
ized using XPS to determine the chlorine concentration on the sur-
face. These experiments were performed using a Kratos Axis Ultra
temperatures on a carbide-modified Mo(110) surface resulting in
the formation of surface hydroxyl species [37]. Similar results have
been reported by others using experimental [38] and computa-
tional methods [39]. Based on the electrostatic adsorption theory,
the surface of the hydroxylated support becomes protonated and
has a net positive charge when the pH is below the PZC and will
adsorb anionic species [19,21–23,32]. At a pH greater than the
PZC, the surface becomes de-protonated and has a net negative
charge and will adsorb cationic species.
The relative metal loadings achieved after 20 h of contact be-
tween the precursor solutions and Mo2C are listed in Table 1. These
loadings were determined from elemental analysis of the Mo2C-
supported catalysts and are defined as:
X-ray photoelectron spectrometer with an Al anode (K radiation
a
at 1486.6 eV) operating at 10 mA and 14 kV. The spectra were ana-
lyzed using CasaXPS, a commercially available XPS analysis pro-
gram. The adventitious carbon peak centered at 284.8 eV was
used to reference the other binding energies. The peak areas were
normalized using the appropriate atomic sensitivity factors. This
allowed comparison of the relative atomic fractions of each species
on the catalyst surfaces.
Measured metal wt:%
Nominal metal wt:%
Relative Metal Loading ¼
ð1Þ
3. Results
Note that the solution pH, which is also listed in Table 1, was not
adjusted; it was established solely by the metal precursor. The Fe
precursors, CoCl2 and NiCl2 did not achieve more than 15% relative
metal loading, whereas all of the other precursors achieved greater
than 90% relative metal loadings.
Temporal variations in the solution metal concentrations for
H2PtCl6, Pd(NH3)4(NO3)2, CuCl2, Cu(NO3)2, Co(NO3)2, and Ni(NO3)2
are illustrated in Fig. 3 and were fit to equations of the following
form [40]:
The diffraction pattern for the Mo2C support (Fig. 1) contained
peaks for b-Mo2C [33] and a-MoC1ꢃx [34]. The relative peak areas
suggested similar amounts of each. These results are consistent
with previous reports using the same synthesis conditions [3].
Although multiple carbide phases were observed, the Mo:C ratio
was close to 2 so we will refer to this material as ‘Mo2C’ throughout
the paper. No peaks were observed for MoO2 [35] or MoO3 [36],
indicating that the synthesis conditions achieved complete bulk
carburization.
According to Park and Regalbuto [32], the flat region of the
curve shown in Fig. 2 corresponds to the PZC, which was approxi-
mately pH 5 for the Mo2C support. As is the case for oxides, the
Mo2C surface should be hydroxylated in the aqueous solution.
Hwu and co-workers reported that H2O dissociated at low
dCi
dt
ꢃ
¼ kCni
ð2Þ
Table 1
Relative metal loadings, initial pH values, and anticipated ionic species for the various
metal precursors.
Precursor
Relative metal
loadinga (%)
Initial precursor
solution pH
Ionic
speciesb
PtCl26ꢃ
H2PtCl6
100
2.5
7.3
2þ
Pd(NH3)4(NO3)2 100
PdðNH3Þ4
Cu2+
Cu2+
Co2+
CuCl2
Cu(NO3)2
CoCl2
Co(NO3)2
NiCl2
Ni(NO3)2
FeCl2
100
100
9
91
11
91
3
5.0
5.1
6.2
5.7
5.9
5.7
4.7
2.9
Co2+
Ni2+
Ni2+
Fe2+
Fe(NO3)3
14
Fe3+
a
Relative metal loading defined in Eq. (1) and determined using ICP-OES after
20 h of loading.
b
Ionic species refers to the metal ion/complex present in the precursor. In an
aqueous solution, the ionic species may be hydrolyzed to form other complexes.
Fig. 1. X-ray diffraction patterns for (a) Mo2C and peak positions for polycrystalline
(b) b-Mo2C [33] and (c) a-MoC1–x [34].
Fig. 2. Initial pH as a function of the final pH for aqueous solutions containing Mo2C
Fig. 3. Normalized metal concentrations (C/C0) in the precursor solutions as a
function of time. The metal concentrations were determined using ICP-OES.
at a surface loading of 5000 m2/L.

J.A. Schaidle et al. / Journal of Catalysis 289 (2012) 210–217
213
Table 2
Orders and rate constants for various metal precursors derived from temporal variations in the metal concentrations (see Eq. (2)).
Precursor
Zero order
First order
k (sꢃ1
Second order
k (mol Lꢃ1 sꢃ1
)
R2
)
R2
k (L molꢃ1 sꢃ1
)
R2
H2PtCl6
Pd(NH3)4(NO3)2
CuCl2
Cu(NO3)2
Co(NO3)2
Ni(NO3)2
2.3 ꢄ 10ꢃ7
1.3 ꢄ 10ꢃ7
1.4 ꢄ 10ꢃ7
1.1 ꢄ 10ꢃ7
8.2 ꢄ 10ꢃ8
6.8 ꢄ 10ꢃ8
0.76
0.75
0.69
0.68
0.88
0.80
6.5 ꢄ 10ꢃ4
3.7 ꢄ 10ꢃ4
4.7 ꢄ 10ꢃ4
3.8 ꢄ 10ꢃ4
9.9 ꢄ 10ꢃ5
9.7 ꢄ 10ꢃ5
0.90
0.99
0.99
0.98
0.96
0.86
1.1
1.9
5.2
3.5
0.89
0.73
0.57
0.74
0.97
0.99
1.1 ꢄ 10ꢃ1
1.1 ꢄ 10ꢃ1
The bold values correspond to rate orders that best fit the adsorption data for a given precursor.
where k is the reaction rate constant, t is time, Ci is the concentra-
tion of metal in solution, and n is the reaction order (n = 0, 1, or 2).
The fit parameters are listed in Table 2. Trends for the other precur-
sors were not fit with Eq. (2) because the metal precursor concen-
trations only decreased slightly, making it difficult to fit the data.
Uptake trends for the precursors were adequately described by
the first-order equation although fits for the Co(NO3)2 and Ni(NO3)2
precursors were slightly better when using the second-order equa-
tion. Second-order kinetics would be consistent with two reactant
species being involved in the rate limiting step. The difference in
the reaction order suggests a different mechanism for these two
metal groups. Based on the first-order rate constants, the rate of up-
take decreased in the following order:
for H2PtCl6 which had a chlorine concentration of 13.62 mmol/L.
These results suggest that Clꢃ did not inhibit adsorption of the pre-
cursor. Chlorine was, however, present on the Pt/Mo2C and Cu/
Mo2C catalysts, and the Cl/Mo atomic ratios for these materials
were 0.165 and 0.045, respectively.
Diffraction patterns for the Mo2C-supported metal catalysts
(Fig. 5) resembled those for the Mo2C support. There were no
clearly discernable peaks for the supported metals, indicating that
dimensions for the crystallites, if present, were below the detection
limit of the diffractometer.
X-ray absorption spectroscopy was used to characterize three
different samples:
(i) Solution: The metal precursor dissolved in deaerated deion-
ized water prior to adding Mo2C.
H2PtCl6 ꢀ PdðNH3Þ4ðNO3Þ2 ꢀ CuCl2 ꢀ CuðNO3Þ2 > CoðNO3Þ2
ꢀ NiðNO3Þ2 ꢁ CoCl2 ꢀ NiCl2 ꢀ FeCl2 ꢀ FeðNO3Þ3
(ii) Dried: After exposing the precursor solution to Mo2C, the
resulting material was removed from the solution without
exposure to air and dried in He at room temperature.
(iii) Reduced: Instead of being dried in an inert atmosphere as
described in (ii), the material was removed from the precursor
solution without exposure to air and reduced in H2 at 450 °C.
Results summarized in Table 1 suggest that counter ions (Clꢃ or
NOꢃ3 ) affected interactions of the Co and Ni metal precursors with
Mo2C. The Co and Ni chloride precursors achieved ꢀ10% relative
metal loading while the nitrate precursors achieved greater than
90%. Assuming that the cations were the same for both solutions,
the results suggest that the Clꢃ ion inhibited precursor adsorption.
X-ray photoelectron spectroscopy was used to measure the chlorine
contents for the dried materials. The materials were not reduced
prior to the measurements to ensure that any Cl that may have been
adsorbed on the Mo2C surface was not removed. The Cl 2p spectra
for these materials are shown in Fig. 4. No clearly discernable Cl 2p
peaks were observed for the Co/Mo2C, Ni/Mo2C, and Fe/Mo2C cata-
lysts, materials that had the lowest relative metal loadings of the
materials prepared using metal chloride precursors. Chlorine
concentrations in the precursor solutions were 4.54 mmol/L, except
Spectra for the dried and reduced materials were collected without
exposure to air.
Fig. 6 shows the X-ray absorption near edge structure (XANES)
spectra for the solution, dried, and reduced samples produced
using the H2PtCl6, Pd(NH3)4(NO3)2, CuCl2, NiCl2, and FeCl2 precur-
sors. In addition, spectra for metal foil (zero-valent) references
are included. The spectrum for the H2PtCl6 in solution had a strong
white line indicating that the Pt was in an oxidized state (formal
oxidation state of +4). For Pt supported on Mo2C (Fig. 6a), spectra
Fig. 4. Chlorine 2p XPS spectra for the Pt/Mo2C, Cu/Mo2C, Co/Mo2C, Ni/Mo2C, and
Fe/Mo2C catalysts synthesized using chlorine precursors. After depositing the
metals, the samples were dried in 1% O2/He at room temperature.
Fig. 5. X-ray diffraction patterns for the Mo2C-supported metal catalysts.

214
J.A. Schaidle et al. / Journal of Catalysis 289 (2012) 210–217
Fig. 6. X-ray absorption near edge structure (XANES) spectra for the precursor solutions, dried, and reduced samples as well as metal foil references (zero-valent) for (a)
H2PtCl6 at Pt L3 edge, (b) Pd(NH3)4(NO3)2 at Pd K edge, (c) CuCl2 at Cu K edge, (d) NiCl2 at Ni K edge, and (e) FeCl2 at Fe K edge.
for the dried and reduced materials resembled the spectrum for the
Pt foil. These results indicated that the Mo2C support was capable
of reducing the [PtCl6]2ꢃ precursor complex directly to zero-valent
Pt during the adsorption process. This type of response is not typ-
ically observed for oxide supports such as Al2O3, where the metal is
often in an oxidized state upon adsorption and remains complexed
with counter ions [41–44]. Results for the Pd/Mo2C (Fig. 6b) and
Cu/Mo2C (Fig. 6c) catalysts were similar to those for the Pt/Mo2C
catalyst with spectra for the dried and reduced materials resem-
bling those for the metal foil references. Conversely, spectra for
the dried Ni/Mo2C (Fig. 6d) and Fe/Mo2C (Fig. 6e) materials closely
resembled those for the precursor solutions. Only after reduction at
450 °C in H2 did spectra for the Ni/Mo2C and Fe/Mo2C materials
resemble those for the foils. In both cases, the dried materials
exhibited large white line peaks while the reduced materials did
not possess distinct peaks. These results indicated that the Mo2C
support was able to reduce the Pt, Pd, and Cu precursors to
zero-valent metals during the wet impregnation process; however,
Mo2C was unable to reduce the Ni or Fe precursors.
4. Discussion
Results presented in this paper provide important insights
regarding the mechanistic steps involved in the deposition of met-
als onto Mo2C supports including transport of the metal ions/com-
plexes to the surface, adsorption of the ions/complexes, and
reaction at the surface.
Based on the nature of the precursor metal salts, speciation cal-
culations [45] and investigations of the metal complexes in aque-
ous solutions [43,46–47], species in the aqueous precursor
solutions were most likely ionic. When the surface is charged,
these species could be attracted to or repulsed by electrostatic
forces [19–23]. For a material like Mo2C with a PZC of pH 5, the

J.A. Schaidle et al. / Journal of Catalysis 289 (2012) 210–217
215
surface should electrostatically attract cations in solutions with pH
>5. In solutions with a pH <5, the Mo2C surface should attract an-
ions. Regalbuto et al. [21–23] provided a further refinement sug-
gesting that electrostatic interactions have the strongest effects
on metal adsorption at pH values significantly far away from the
PZC. For example, the maximum uptake for chloroplatinic acid over
unoxidized carbon xerogels (PZC ꢀ9.4) was observed at approxi-
mately pH 3 [23]. For the purposes of this paper, we assumed that
strong electrostatic interactions will be important during precursor
adsorption when the solution pH is less than 4 or greater than 6.
For 4 < pH < 6, electrostatic interactions might be weak or
insignificant.
support on contact at room temperature via a red–ox reaction
(see Fig. 6). This behavior could reconcile their high metal loadings
and is very different from that typically observed for oxide sup-
ports. The probability for occurrence of these red–ox reactions
can be determined from the relevant thermodynamic properties.
The Gibb’s free energy for electrochemical reactions is given by:
D
G ¼ ꢃnFðEred ꢃ EoxÞ
ð3Þ
where n is the number of electrons transferred, F is faraday’s
constant, Ered is the potential of the reduction half-cell reaction,
and Eox is the potential of the oxidation half-cell reaction [24].
If Ered > Eox, the Gibb’s free energy of the overall reaction is negative
and the reaction is favorable.
Fig. 7 shows the relative metal loadings and initial solution pH
for different metal precursors. The solution pH was not monitored
during the metal adsorption due to the difficulty in ensuring the
absence of oxygen in the solution. The surface loading of
ꢀ1200 m2/L was similar to that used by Lambert et al. when inves-
tigating the adsorption of Pt on carbon xerogels [23]. They observed
minimal pH buffering, thus we expected that pH buffering would be
negligible under the conditions employed and that the initial
solution pH would be representative of the final solution pH. For
Mo2C, the pH range within the dotted lines (4 < pH < 6) indicates
where electrostatic interactions may be weak or insignificant. With
pH values higher and lower than those indicated by the dotted
lines, solutions of the H2PtCl6, Pd(NH3)4(NO3)2, and FeCl3 precursors
are predicted to be affected by strong electrostatic effects. The pH
values for the H2PtCl6 and Pd(NH3)4(NO3)2 could explain the high
relative loadings for the Pt/Mo2C and Pd/Mo2C catalysts. In particu-
lar, for the H2PtCl6 solution, the pH lies significantly below the PZC
of Mo2C and the net positive charge on the Mo2C surface would
attract the [PtCl6]2ꢃ ions. For Pd(NH3)4(NO3)2, the pH lies above
the PZC of Mo2C and the net negative charge on the Mo2C surface
would attract the [Pd(NH3)4]2+. With a pH below the PZC, the
Mo2C surface would have a positive charge and repel the Fe3+ ions,
possibly accounting for the low relative loading for the Fe/Mo2C
catalyst. The Fe3+ ions may have hydrolyzed in the aqueous
solutions [47]; nevertheless, the resulting complexes are typically
cationic and thus would be repelled by the surface. The attractive
electrostatic interactions would likely provide a driving force for
transport of these metal complexes to the Mo2C surface.
Fig. 8 displays the relative metal loadings as a function of the
standard reduction potentials for the metal ion complexes ex-
pected to be in the precursor solutions. Except for Ni(NO3)2 and
Co(NO3)2, high relative metal loadings were achieved for precur-
sors that had standard reduction potentials higher than 0.3 V.
The XANES for the Pt, Pd, and Cu precursors (Fig. 6) indicated that
the reduction product was zero-valent metal, although, the iden-
tity of the reductant and oxidation product remains unknown.
While standard potentials and thermodynamic data are not avail-
able for Mo2C, it is useful to consider reactions involving the oxida-
tion of Mo (note that Mo in Mo2C is most often considered to be
mono- or di-valent) and C:
Mo ! Mo3þ þ 3eꢃ E0 ¼ 0:20 V
ð4Þ
ð5Þ
ð6Þ
Mo þ 4H2O ! MoO2 þ 4Hþ þ 4eꢃ E0 ¼ 0:15 V
CðsÞ þ H2O ! COðgÞ þ 2Hþ þ 2eꢃ E0 ¼ ꢃ0:52 V
The Mo containing oxidation products could remain on the carbide
surface or dissolve into the aqueous solution. In addition to achiev-
ing the highest relative loadings, the Pt, Pd, and Cu precursors dem-
onstrated the fastest rates of metal adsorption, as characterized by
the first-order rate constants listed in Table 2. Reduction potentials
for the Co and Ni precursors were much lower than those for the Pt,
Pd, and Cu precursors suggesting a correlation between the adsorp-
tion rate and driving force for the surface reaction.
Weak attractive interactions between the surface and precursor
ions could explain the high relative loadings observed for the
Co(NO3)2 and Ni(NO3)2 derived materials; however, for the CuCl2
and Cu(NO3)2 precursors, electrostatic interactions cannot account
for the results. For the Cu/Mo2C catalysts as well as the Pt/Mo2C
and Pd/Mo2C catalysts, the precursors appeared to react with the
Loadings for the chloride precursors other than CuCl2 were
much lower than those for the nitrate precursors suggesting that
the nature of the counter ion played a role. In particular, the Co
and Ni nitrate precursors yielded relative loadings >90% while
loadings for the chloride precursors were <11%. This does not ap-
pear to be due to Cl blocking adsorption sites on the Mo2C surface
as XPS (Fig. 4) indicated that there was very little chlorine on the
Fig. 7. Relative metal loadings and initial pH values for precursor solutions as a
function of metal precursor. The ‘‘PZC window’’ for Mo2C (4 < pH < 6) is given by the
red dotted lines.
Fig. 8. Relative metal loadings and standard reduction potentials foraqueous
solutions of the various metal precursors.

216
J.A. Schaidle et al. / Journal of Catalysis 289 (2012) 210–217
Table 3
Potential ion/ionic species, and factors influencing the adsorption and reaction of various metal precursors onto Mo2C.
Precursor
Ionic speciesa
Adsorption/reaction factors
Electrostatic attraction/repulsionb
Red–oxc
Additional factors
PtCl26ꢃ
H2PtCl6
SA
SA
O
O
2þ
Pd(NH3)4(NO3)2
PdðNH3Þ4
CuCl2
Cu(NO3)2
CoCl2
Co(NO3)2
NiCl2
Ni(NO3)2
FeCl2
Cu2+
Cu2+
Co2+
Co2+
Ni2+
Ni2+
Fe2+
Fe3+
–
–
O
–
–
–
NO
–
NO
–
WA
WA
WA
WA
–
Chloride Inhibition
Chloride Inhibition
Chloride Inhibition
Fe(NO3)3
SR
a
Ionic species refers to the metal ion/complex present in the precursor. In an aqueous solution, the ionic species may be hydrolyzed to form
other complexes.
b
SA refers to Strong Attraction, WA to Weak Attraction, and SR to Strong Repulsion.
O refers to Observed red–ox behavior using XAS, and NO stands for Not Observed.
c
Mo2C after deposition of NiCl2, CoCl2 or FeCl2. Moreover, the Cl sur-
face concentration was the highest for the metals with the highest
relative metal loadings (H2PtCl6 and CuCl2). We propose two expla-
nations for these results. It is possible that Clꢃ complexed with me-
tal ions in solution and altered their red–ox potentials [24]. For
example, the reduction of Pd2+ to Pd metal has a standard reduc-
tion potential of 0.99 V, whereas the reduction of Pd2+ present in
the form of PdCl24ꢃ has a standard reduction potential of 0.62 V
[24]. For Co and Ni, complexation with Cl could decrease the stan-
dard reduction potential, making reaction with the Mo2C surface
species unfavorable. It is also possible that the NOꢃ3 counter ion
facilitated adsorption of the Ni and Co precursors by partially oxi-
dizing the Mo2C surface so that they could participate in red–ox
reactions. This would be consistent with second-order kinetics; re-
call that the second-order equation fit data for the Ni and Co pre-
cursors slightly better than first order while the Pt, Pd, and Cu
precursors followed first-order kinetics (Table 2). Nevertheless,
further research is required to fully understand the effect of ligand
type on the relative metal loadings and rates of adsorption.
Laboratory at the University of Michigan. The XPS was performed
in the Electron Microbeam Analysis Laboratory at the University
of Michigan (NSF DMR-0420785), and the XAS was performed at
the Advanced Photon Source at Argonne National Laboratory with
the assistance of the members of the Materials Research Collabora-
tive Access Team (supported by the Department of Energy and the
MRCAT member institutions).
References
[1] C.H. Bartholomew, R.J. Farrauto, Fundamentals of Industrial Catalytic
Processes, second ed., John Wiley and Sons, 2006.
[2] W. Setthapun, S.K. Bej, L.T. Thompson, Top. Catal. 49 (2008) 73–80.
[3] J.A. Schaidle, A.C. Lausche, L.T. Thompson, J. Catal. 272 (2010) 235–245.
[4] N.M. Schweitzer, J.A. Schaidle, O.K. Ezekoye, X. Pan, S. Linic, L.T. Thompson, J.
Am. Chem. Soc. 133 (2011) 2378–2381.
[5] A.C. Lausche, J.A. Schaidle, L.T. Thompson, Appl. Catal. A 401 (2011) 29–36.
[6] S.T. Oyama, Catal. Today 15 (1992) 179–200.
[7] R.B. Levy, M. Boudart, Science 181 (1973) 547–549.
[8] A. Griboval-Constant, J.-M. Giraudon, G. Leclercq, L. Leclercq, Appl. Catal. A 260
(2004) 35–45.
[9] A. Griboval-Constant, J.-M. Giraudon, I. Twagishema, G. Leclercq, M.E. Rivas, J.
Alvarez, M.J. Perez-Zurita, M.R. Goldwasser, J. Mol. Catal. A: Chem. 259 (2006)
187–196.
[10] M. Lewandowski, A. Szymanska-Kolasa, P. Da Costa, C. Sayag, Catal. Today 119
(2007) 31–34.
[11] D.V. Espositio, S.T. Hunt, A.L. Stottlemyer, K.D. Dobson, B.E. McCandless, R.W.
Birkmire, J.G. Chen, Angew. Chem. 49 (2010) 9859–9862.
[12] D.J. Ham, C. Pak, G.H. Bae, S. Han, K. Kwon, S.-A. Jin, H. Chang, S.H. Choi, J.S. Lee,
Chem. Commun. 47 (2011) 5792–5794.
[13] I.J. Hsu, D.A. Hansgen, B.E. McCandless, B.G. Willis, J.G. Chen, J. Phys. Chem. C
115 (2011) 3709–3715.
[14] M. Nie, H. Tang, Z. Wei, S.P. Jiang, P.K. Shen, Electrochem. Commun. 9 (2007)
2375–2379.
[15] R. Ganesan, J. Dong, J.S. Lee, Electrochem. Commun. 9 (2007) 2576–2579.
[16] E.C. Weigert, S. Arisetty, S.G. Advani, A.K. Prasad, J.G. Chen, J. New Mat,
Electrochem. Syst. 11 (2008) 243–251.
[17] J. Patt, Ph.D. Thesis, University of Michigan, 2003.
[18] Satterfield, Heterogeneous Catalysis in Industrial Practice, second ed.,
McGraw-Hill, 1991.
[19] J.P. Brunelle, Pure Appl. Chem. 50 (1978) 1211–1229.
[20] J.-F. Lambert, M. Che, J. Mol. Catal. A: Chem. 162 (2000) 5–18.
[21] K.B. Agashe, J.R. Regalbuto, J. Colloid Interface Sci. 185 (1997) 174–189.
[22] J.R. Regalbuto, A. Navada, S. Shadid, M.L. Bricker, Q. Chen, J. Catal. 184 (1999)
335–348.
5. Conclusions
The preparation of Mo2C-supported Pt, Pd, Cu, Co, Ni, and Fe cat-
alysts via wet impregnation has been investigated. The genesis of
these materials has been explained in terms of adsorption of the me-
tal precursors and surface reactions. Interactions that appeared to
have the most significant influence on the supported catalysts are
summarized in Table 3. The strength of these interactions as well
as the nature of the metal precursor (i.e., counter ion) affected the
rate of deposition. Based on first-order rate constants, the rate of
deposition decreased in the following order: H2PtCl6 ꢀ Pd(NH3)4
(NO3)2 ꢀ CuCl2 ꢀ Cu(NO3)2 > Co(NO3)2 ꢀ Ni(NO3)2 ꢁ CoCl2 ꢀ NiCl2
ꢀ FeCl2 ꢀ Fe(NO3)3. Strong interactions between the Pt, Pd, and Cu
precursors and the native Mo2C surface caused by electrostatic inter-
actions and/or red–ox reactions could be the cause for the high metal
dispersions observed for these catalysts. For example, Schweitzer
et al. reported that strong interactions between H2PtCl6 and Mo2C re-
sulted in the formation of raft-like particles [4]; characterization of
the particle sizes/shapes for the other Mo2C-supported metal cata-
lysts is ongoing. Given that for most systems the precursor does
not react with the support during initial stages of preparation, the
red–ox behavior observed for the carbide-supported materials could
result in catalysts with significantly different properties than those
for catalysts produced from the oxidized (i.e., passivated) carbides.
[23] S. Lambert, N. Job, L. D’Souza, M.F.R. Pereira, R. Pirard, B. Heinrichs, J.L.
Figueiredo, J.-P. Pirard, J.R. Regalbuto, J. Catal. 261 (2009) 23–33.
[24] J. Barbier, in: G. Ertl, H. Knozinger, J. Weitkamp (Eds.), Preparation of Solid
Catalysts, Wiley-VCH, 1999.
[25] O. Clause, M. Kermarec, L. Bonneviot, F. Villain, M. Che, J. Am. Chem. Soc. 114
(1992) 4709–4717.
[26] M. Kermarec, J.Y. Carriat, P. Burattin, M. Che, A. Decarreau, J. Phys. Chem. 98
(1994) 12008–12017.
[27] J.R. Kitchin, J.K. Norskov, M.A. Barteau, J.G. Chen, Catal. Today 105 (2005) 66–
73.
[28] C. Ruberto, A. Vojvodic, B.I. Lundqvist, Surf. Science 600 (2006) 1612–
1618.
[29] A. Vojvodic, A. Hellman, C. Ruberto, B.I. Lundqvist, Phys. Rev. Lett. 103 (2009)
146103.
Acknowledgments
This work was supported by funding from the National Science
Foundation (CBET-0933239), and the Hydrogen Energy Technology

J.A. Schaidle et al. / Journal of Catalysis 289 (2012) 210–217
217
[30] J. Barbier, J.C. Menezo, C. Montassier, J. Naja, Catal. Lett. 14 (1992) 37–43.
[31] C. Montassier, J.C. Menezo, J. Naja, J. Barbier, J.M. Dominguez, P. Sarrazin, B.
Didillon, J. Mol. Catal. 91 (1994) 107–117.
[32] J. Park, J.R. Regalbuto, J. Colloid Interface Sci. 175 (1995) 239–252.
[33] S. Nagakura, M. Kikuchi, S. Oketani, Acta Crystallogr. 21 (1966) 1009.
[34] J.C. Schuster, H. Nowotny, Monatsh. Chem. 110 (1979) 321.
[35] M. Ghedira, C. Do-Dinh, M. Marezio, J. Mercier, J. Solid State Chem. 59 (1985)
159.
[36] J.D. Hanawalt, H.W. Rinn, L.K. Frevel, Anal. Chem. 10 (1938) 457.
[37] H.H. Hwu, J.G. Chen, Surf. Sci. 536 (2003) 75–87.
[38] S.V. Didziulis, P. Frantz, S.S. Perry, O. El-bjeirami, S. Imaduddin, P.B. Merrill, J.
Phys. Chem. B 103 (1999) 11129–11140.
[40] H.S. Fogler, Elements of Chemical Reaction Engineering, third ed., Prentice Hall,
New Jersey, 1999. pp. 235–237.
[41] B. Shelimov, J.-F. Lambert, M. Che, B. Didillon, J. Catal. 185 (1999) 462.
[42] B.N. Shelimov, J.-F. Lambert, M. Che, B. Didillon, J. Mol. Catal. A: Chem. 158
(2000) 91.
[43] W.A. Spieker, J. Liu, J.T. Miller, A.J. Kropf, J.R. Regalbuto, Appl. Catal. A 232
(2002) 219.
[44] W.A. Spieker, J. Liu, X. Hao, J.T. Miller, A.J. Kropf, J.R. Regalbuto, Appl. Catal. A
243 (2003) 53.
[45] N.M. Schweitzer, Ph.D. Thesis, University of Michigan, 2010.
[46] J.D. Cuppett, S.E. Duncan, A.M. Dietrich, Chem. Senses 31 (2006) 689.
[47] H. Hellman, R.S. Laitinen, L. Kaila, J. Jalonen, V. Hietapelto, J. Jokela, A. Sarpola, J.
Ramo, J. Mass Spectrom. 41 (2006) 1421–1429.
[39] H. Tominaga, M. Nagai, J. Phys. Chem. B 109 (2005) 20415–20423.