Reduction of Propanoic Acid over Pd-Promoted Supported WOx Catalysts

Article abstract of DOI:10.1002/cctc.201901302

Silica-, titania-, and zirconia-supported tungsten oxide catalysts were synthesized by wetness impregnation techniques. When promoted with Pd, these materials catalyzed the reduction of propanoic acid to 1-propanol at 433 K with a selectivity of up to 92 % (13.5 % conversion) in atmospheric pressure of H₂. Over Pd-promoted WO_x/TiO_2, the observed orders of reaction were 0.2 in H₂ and 0.7 in propanoic acid, and the apparent activation energy was 54 kJ mol^?1. In situ X-ray absorption spectroscopy of Pd-promoted WO_x/SiO₂ revealed a slight reduction of the W from +6 to an average oxidation state of about +5 during H₂ treatment above 473 K. In situ infrared spectroscopy indicated the catalyst surface was covered mostly by propanoate species during reaction. For comparison, supported phosphotungstic acid was also evaluated as a catalyst under identical conditions, but the resulting high acidity of the catalyst was deleterious to alcohol selectivity.

Full text of DOI:10.1002/cctc.201901302

Accepted Article
Title: Reduction of Propanoic Acid over Pd-promoted Supported WOx
Catalysts
Authors: James D Kammert, Gordon Brezicki, Naomi Miyake, Eli
Stavitski, and Robert Davis
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemCatChem 10.1002/cctc.201901302
Link to VoR: http://dx.doi.org/10.1002/cctc.201901302
A Journal of
www.chemcatchem.org

10.1002/cctc.201901302
ChemCatChem
Reduction of Propanoic Acid over
Pd-promoted Supported WOx Catalysts
Dr. James D. Kammerta, Gordon Brezickia, Naomi Miyakea, Dr. Eli Stavitskib,
and Prof. Robert J. Davisa,*
a) University of Virginia Department of Chemical Engineering, 102 Engineer’s Way,
PO Box 400741, Charlottesville, VA 22904-4741, United States
b) National Synchrotron Light Source II, Brookhaven National Lab, Upton, NY
11973, United States
* To whom correspondence should be addressed: rjd4f@virginia.edu
1
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
Abstract
Silica-, titania-, and zirconia-supported tungsten oxide catalysts were synthesized by wetness
impregnation techniques. When promoted with Pd, these materials catalyzed the reduction of
propanoic acid to 1-propanol at 433 K with a selectivity of up to 92% (13.5% conversion) in
atmospheric pressure of H2. Over Pd-promoted WOx/TiO2, the observed orders of reaction were
0.2 in H2 and 0.7 in propanoic acid, and the apparent activation energy was 54 kJ mol-1. In situ X-
ray absorption spectroscopy of Pd-promoted WOx/SiO2 revealed a slight reduction of the W from
+6 to an average oxidation state of about +5 during H2 treatment above 473 K. In situ infrared
spectroscopy indicated the catalyst surface was covered mostly by propanoate species during
reaction. For comparison, supported phosphotungstic acid was also evaluated as a catalyst under
identical conditions, but the resulting high acidity of the catalyst was deleterious to alcohol
selectivity.
Graphical abstract:
Promoting supported WOx with Pd results in catalytic activity for
the selective (up to 92%) reduction of carboxylic acids to alcohols.
Palladium-promoted tungsten oxides were deposited on a variety
of supports and characterized using steady-state conversion of
propanoic acid, X-ray absorption spectroscopy, and diffuse
reflectance infrared Fourier transform spectroscopy.
Keywords: Carboxylic acids, infrared spectroscopy, reduction, tungsten, X-ray absorption
spectroscopy
2
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
1
Introduction
World demand for oil-derived products, especially chemicals, is expected to grow rapidly
in the coming years, as emerging economies develop an increased desire for plastics.[1] While many
current technologies still use fossil resources to produce chemicals, a significant effort is underway
to develop cost-effective ways to convert biomass to commodity chemicals. Biomass has a similar
effective hydrogen-to-carbon ratio as many value-added chemicals, and offers the potential to
produce these chemicals at a lower carbon cost in efficient processes.[2] A study performed by the
Department of Energy in 2004 found that many of the top building block chemicals that could be
produced from biomass are carboxylic acids.[3] Many of these building blocks continue to show
promise as potential platform molecules, including short chain carboxylic acids such as lactic acid,
succinic acid, 3-hydroxypropanoic acid, and levulinic acid, which can be subsequently upgraded
through reduction of their carboxylic acid functionality.[4] As a result, selective reduction of
carboxylic acids to produce aldehydes and alcohols is of significant importance to the development
of cost-effective technologies for biomass valorization.
Copper-based oxide spinels (such as copper chromite) are well-known catalysts for the
reduction of esters (and carboxylic acids) to their corresponding alcohols.[5] Dissociation of the
ester into alkoxy and acyl species followed by hydrogenation of the acyl species determined the
rate of alcohol formation in this process over copper chromite and supported copper.[6–8] Other
works studied the oxides of Re,[9,10] Sn,[11] and Mo[12] promoted by platinum group metals and
found they are effective catalysts for the direct catalytic reduction of carboxylic acids. A recent
review by Pritchard et al. highlights some of the major recent findings concerning heterogeneous
reduction of carboxylic acids.[13] In general, a combination of a reducible metal and an oxophilic
metal produces an active catalyst when the two metals are placed in close proximity on a support.
3
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
This combination provides adsorbed H atoms to the active site on the metal oxide species, which
results in the reduction of the carboxylic acid to an aldehyde intermediate that is subsequently
hydrogenated to form the corresponding alcohol. The ability of Pd-promoted supported tungsten
oxides, however, to catalyze selective reduction of carboxylic acids to alcohols has not been
thoroughly assessed.
Tungsten oxide has a variety of uses as a heterogeneous catalyst, and it is well-known to
produce Brønsted acid sites of varying acid strength and site density, depending on the support
upon which it is dispersed. Yamaguchi et al. found that WOx supported on TiO2, Al2O3, and ZrO2
resulted in the formation of higher densities of acidic sites compared to unsupported WOx or WOx
supported on SiO2 and MgO.[14] The resulting acidic sites were catalytically active for 1-butene
disproportionation at 373 K. Hino and Arata found that controlled synthesis could produce
WOx/ZrO2 with superacid properties capable of catalyzing butane isomerization at temperatures
as low as 303 K.[15] There is good agreement in the literature that the rate of catalytic isomerization
reactions per tungsten atom over WOx/ZrO2 is maximized at loadings slightly above the theoretical
monolayer coverage of WOx on the ZrO2.[16–18] Similar trends in reactivity can be found on
supports other than zirconia, where dispersion of the WOx phase appears to play a critical role.[19]
Barton et al. proposed this trend in activity resulted from the formation of reducible polytungstate
clusters at optimal intermediate WOx surface loadings that could stabilize Brønsted acidic
protons.[20] Lower loadings of WOx resulted in irreducible monotungstate species that were less
active in the isomerization reactions, while higher loadings resulted in the formation of inactive
WO3 crystallites. The addition of platinum allowed the slight reduction of WOx polytungstate
clusters to occur at mild temperatures (350 K), and resulted in turnover frequencies on the order
of 102 s-1 for n-heptane isomerization (473 K, 650 kPa H2, 100 kPa n-heptane).[21] Characterization
4
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
of acidic sites by ex situ and in situ titration with pyridine, ammonia, or sterically hindered
pyridine found conflicting trends in Brønsted acid site density with relation to catalytic
activity.[17,18,22] Some studies reported a maximum in Brønsted acidity per W atom at intermediate
loadings, while others have found that Brønsted acidity per W atom, as well as Lewis acidity per
W atom, decreased continuously with increasing W loading. On the other hand, Zr-stabilized,
distorted WO3 crystallites have also been suggested to play a key role in catalysis, for example, in
the acid-catalyzed dehydration of methanol to dimethyl ether.[23] Evidently, the nature of
catalytically active sites present in supported WOx catalysts is not completely understood.
Heteropolyanions derived from tungsten, such as phosphotungstic acid (H3PW12O40), also
have many uses as heterogeneous acid catalysts.[24] Phosphotungstic acid (PTA) consists of a
tetrahedrally coordinated phosphorus atom surrounded by 12 tungsten(VI) oxide octahedra,
forming a well-defined Keggin unit (Figure 1). The resulting compound can stabilize up to three
Brønsted acidic protons, which can be used to catalyze alkylation, condensation, and
dehydration reactions, among many other acid-catalyzed reactions. More information about the
properties and uses of tungsten heteropolyanions can be found in reviews by Kozhevnikov24 and
Misono.[25] Recent studies have also investigated PTA as a catalyst for conversion of biomass.
Examples include the hydrolysis-hydrogenation of cellulose in the presence of Ru/C,[26] and the
esterification of triethylene glycol and methacrylic acid.[27]
5
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
Figure 1: The Keggin structure of phosphotungstic acid, consisting of 12 octahedral tungsten
(VI) oxide species surrounding a tetrahedrally coordinated phosphorus atom. The central
oxygens connect the central phosphorus atom to the surrounding tungsten octahedra, while
bridging oxygens are capable of stabilizing Brønsted-acidic protons.[28]
Few studies have explored tungstates as catalysts for the reduction of carboxylic acids to
alcohols. In 1989, tungstate species were identified as a component of the enzymatic reduction of
carboxylic acids to aldehydes in Clostridium thermoaceticum.[29] Several years later, a patent by
Kitson and Williams claimed a mixture of Pt group metals and Re, Mo, and W for the conversion
of carboxylic acids to alcohols and esters, but only gave details of the synthesis of several Pd-
alloy-promoted Re catalysts.[30] In 1997, Pestman et al. screened a number of metal oxide catalysts
for the reduction of acetic acid and reported that WO3 produced a small amount (17% selectivity)
of acetaldehyde in H2 at 723 K, while the majority product was acetone.[31] In 2013, a patent for a
catalyst comprised of tungsten trioxide WO3, an alkaline earth metal promoter, and a Sn-promoted
6
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
transition metal component was claimed by Jevtic et al. for the reduction of carboxylic acids to
alcohols.[32] The catalyst could also be promoted by a wide range of transition metals including
Pd, but no metal was specifically preferred. Recent studies by Gosselink and Hollak found that
bulk WO3 and W2C performed hydrodeoxygenation of stearic acid, tristearin, and methyl stearate
to C17 and C18 hydrocarbon products at 523 K, but few oxygenate products were observed under
the conditions studied.[33,34] Palladium-promoted WOx/Al2O3 has demonstrated activity for the
hydrodeoxygenation of guaiacol,[35] and Pt-promoted WOx/Al2O3 and WOx/ZrO2 were active in the
hydrogenolysis of glycerol to 1,3-propanediol and could be promoted with SiO2 addition.[36,37]
These previous findings suggest that Pd- and Pt-promoted WOx materials have the potential to
catalyze the selective reduction of carboxylic acids to alcohols.
In this work, we investigated the catalytic activity of SiO2-, TiO2-, and ZrO2-supported Pd-
promoted WOx and compared them to supported phosphotungstic acid (PTA) in the gas-phase
reduction of propanoic acid. Biomass conversion processes are often carried out in the liquid phase
and are applied to molecules that may sometimes have multiple functional groups (e.g. lactic,
succinic, 3-hydroxypropionic, and levulinic acids mentioned earlier). To generate insights
specifically about the role of the metal-oxide and metal-support interactions, however, a simplified
gas phase system was chosen to study the reduction of a short-chain fatty acid as a surrogate for
longer chain free fatty acids that can be produced by fermentation of sugars by engineered
microorganisms.[38] The chosen system was expected to be free from complicating effects of
solvent interaction with the active site or leaching of the active components (such as
phosphotungstic acid, which is soluble in polar media) and did not require extreme H2 pressures
to operate. The adsorption of H2 and CO was performed to estimate the dispersion of Pd and
explore the effect of spillover on H2 uptake by WOx. Temperature-programmed reduction and in
7
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
situ X-ray absorption spectroscopy were used to follow changes in structure of the WOx before
and after treatment of the catalysts in flowing H2, which were correlated to catalyst activity. In
situ diffuse reflectance infrared Fourier transform spectroscopy provided insights on the species
present on the tungsten oxide and support surfaces.
2
Experimental Methods
2.1
Catalyst Synthesis
Silica-supported (Fuji Silysia Chemical Ltd, 511 m2 g-1, 75-150 µm particle size) 8 wt %
W (W/SiO2) and 1 wt % Pd, 8 wt % W (PdW/SiO2) catalysts were synthesized by the incipient
wetness impregnation method. The desired loading of tungsten was introduced as ammonium
metatungstate hydrate (Aldrich, 99.99%) to an aqueous solution equal to the pore volume of the
SiO2. The solution was added to the support until the point of incipient wetness. The mixture was
dried at 383 K in ambient air, then heated in flowing air to 773 K over 3 h and held at 773 K for 3
h in flowing air. This process was repeated for the desired loading of Pd (using aqueous
tetraaminepalladium(II) nitrate, 10 wt%, Aldrich). Titania- and zirconia- supported 8 wt % W
(W/TiO2, W/ZrO2) and 1 wt % Pd, 8 wt % W (PdW/TiO2, PdW/ZrO2) catalysts were synthesized
by first adding 5 g of either TiO2 (Degussa P25, 47 m2 g-1) or ZrO(OH)2 (MEL Chemicals, 420 m2
g-1) to 80 cm3 of distilled, deionized water (16.8 MΩ), and then dissolving the desired loading of
tungsten as ammonium metatungstate hydrate into this suspension. The suspension was
dehydrated in a rotary evaporator at 353 K, and then dried at 383 K overnight in ambient air. The
dried material was then heated in flowing air to 773 K over 3 hours, held in flowing air at 773 K
for 3 h, cooled, and sized to obtain a particle size of between 55 and 180 µm in diameter. The
desired loading of Pd was then added as an aqueous solution of tetraaminepalladium(II) nitrate by
the incipient wetness procedure outlined above, subjected to the same 773 K treatment in air, and
8
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
then re-sized to obtain particles between 55 and 180 µm in diameter. Phosphotungstic acid hydrate
(Sigma Aldrich, reagent grade) and Pd-promoted phosphotungstic acid (1 wt % Pd, 8 wt % W)
catalysts supported on SiO2 (PTA/SiO2 and PdPTA/SiO2) and TiO2 (PTA/TiO2 and PdPTA/TiO2)
were also synthesized using the incipient wetness impregnation method. To avoid decomposition
of the phosphotungstic acid (PTA) supported on SiO2 and TiO2, Pd was added first by the incipient
wetness impregnation procedure for Pd addition described earlier, and then an aqueous solution
of PTA was added until the point of incipient wetness to the support. The supported PTA was
dried at 383 K in ambient air overnight prior to use.
2.2
Catalytic Reactions
The effects of catalyst pretreatment and reaction conditions on the reduction of propanoic
acid (Sigma Aldrich, 99.5%) in H2 (Praxair, 5.0) and D2 (Praxair, 2.7) were studied in the gas phase
using a packed-bed downward flow reactor. Propanoic acid was supplied from a stainless-steel
vapor saturator equipped with a thermocouple to measure the saturator temperature. The reactor
consisted of a stainless-steel tube with inner diameter of 0.46 cm, enclosed in a cylindrical
aluminum jacket 3.8 cm in diameter, which was heated externally. The reaction temperature was
measured by a thermocouple inserted from the top of the reactor tube into the middle of the catalyst
bed, which was supported in the center of the reactor tube on a plug of glass wool. Prior to catalytic
testing, all lines and saturators leading to the reactor were purged with N2 for at least 10 min. The
reduction of propanoic acid (0-0.5 kPa) was studied between 403 and 433 K, at a total pressure of
0.1 MPa composed of H2 (0-0.1 MPa) in a balance of N2. The components in the product stream
were separated using an on-line SRI 8610C gas chromatograph (GC) equipped with an MXT-
WAX (0.53 mm i.d., 1 µm film thickness, 30 m) column and quantified using a flame-ionization
detector (FID).
9
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
The fractional conversion (fi) and product selectivity of propanoic acid (i) to products (j)
were obtained using equations (1) and (2):
∑
푀
푗
푓 =
(1)
(2)
푖
∑
푀푖 + 푀
푗
푀
푗
푆 =
푗
∑
푀
푗
where Mj is defined as the moles of carbon product j. Hydrocarbon products having carbon
numbers below C6 could not be separated using the MXT-WAX column, and therefore were
grouped together as light hydrocarbons, or LHC, for analysis.
2.3
Catalyst Characterization
The chemisorption of H2 (Praxair, 5.0) and CO (Praxair, 3.0) was performed using a
Micromeritics ASAP 2020 instrument. Prior to analysis, each sample was heated to 473 K in
flowing H2 at 10 K min-1, then held at 473 K for 1 h under flowing H2. The samples were
subsequently evacuated at 473 K for 3 h, before cooling to 373 K or 308 K to perform the H2 or
CO chemisorption, respectively. Dihydrogen or carbon monoxide was dosed over the range of
0.13 to 67 kPa. The reported chemisorption values were obtained by extrapolating the probe
molecule uptake in the high pressure linear region of the isotherm to zero pressure, which was
considered the saturated monolayer coverage.
In situ X-ray absorption spectroscopy (XAS) at the tungsten LI and LIII edges and Pd K
edge was carried out on beamline 8-ID of the National Synchrotron Light Source II (NSLS-II) at
Brookhaven National Laboratory in Upton, New York, operated at 3.0 GeV and with a typical
current of ca. 325 mA. All experiments were performed using the Si(111) double crystal
monochromator with uncoated Si and Pt-coated higher harmonic rejection mirrors, for W and Pd
edges respectively, and a spot size of 0.4 mm. Samples were prepared for analysis by packing the
10
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
desired loading of catalyst into a 6 in (0.25 in OD) stainless-steel tube outfitted with Kapton
windows and gas tight fittings to allow for sample heating and gas treatment. Samples were held
in place in the stainless-steel tubes by supporting the catalyst bed on either side with a plug of
shredded graphite felt (AvCarb Felt G200). Prior to treatment, each of the individually fed reactor
tubes was flushed with He at 20 cm3 min-1 for 10 min to remove any residual O2 in the tubes.
Temperature-programmed reduction studied at the W L edges was carried out by initiating the
flow of H2 at 20 cm3 min-1, waiting for 5 min to allow replacement of the He atmosphere with H2,
and then ramping the temperature at 3 K min-1 from ambient temperature to 673 K. Samples were
then cooled in He to room temperature. At the Pd K edge, samples were heated at 3 K min-1 from
ambient temperature to 473 K. After holding the temperature at 473 K for 45 min, the samples
were heated to 673 K at 10 K min-1, then cooled to 373 K to collect spectra.
Spectra obtained from X-ray absorption spectroscopy were processed and analyzed using
the Demeter software package, written by Bruce Ravel.[39] Prior to analysis, X-ray absorption near-
edge structure (XANES) and X-ray absorption fine structure (EXAFS) were background corrected
and normalized uniformly at each edge using Athena, an XAS analysis program that is part of the
Demeter software package. The oxidation state of the W component of each sample was estimated
using the measured position of the W LI and LIII edges. The edge position at the W LI edge was
arbitrarily determined at 0.75 times the normalized absorbance height to minimize interference
with the presence of the observed pre-edge feature of the WVI oxide, while the W LIII-edge position
was arbitrarily determined at a step height of unity in the normalized absorbance spectrum to
account for the position of the white line. Local coordination structure and EXAFS were analyzed
at the W LIII and Pd K edges. Powders of WO2 (Alfa Aesar, 99.9%), WO3 (Aldrich, 99.995%), and
ammonium metatungstate hydrate were used as standards at the W LI and LIII edges, while Pd foil
11
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
(Goodfellow, 12.5 μm, 99.95%) and PdO (Aldrich, 99.999%) were used as standards at the Pd K
edge. The WO3 and Pd foil standards were used as energy references at their respective absorption
edges. Powder standards were prepared by grinding each powder thoroughly, mixing with boron
nitride (Aldrich, 98%), and then compressing the mixture into a self-supporting wafer using a
hydraulic press. The wafers were then affixed to Kapton tape and placed in the beam path for
measurement.
In situ infrared spectra of propanoic acid adsorbed on PdW/TiO2 and W/TiO2 were
obtained using diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS). Spectra
were recorded using a Bruker Vertex 70 FTIR spectrometer equipped with a liquid-nitrogen-
cooled MCT detector and a Harrick Praying Mantis in situ reactor cell with 5 mm ZnSe windows.
Spectra were collected every 20 min, by averaging 800 scans at a resolution of 2 cm-1. The inlet
gases were purified using an OMI purifier (Supelco) upstream of a stainless-steel propanoic acid
vapor saturator with a bypass line leading to the in situ cell. Prior to data collection, samples were
ramped to 473 K in flowing H2, and then cooled to 413 K in H2 for analysis. Each sample was
exposed to 0.3 kPa of propanoic acid in 0.1 MPa H2 flowing at 15 cm3 min-1 until the sample
reached a steady state (approximately 120 min). The flow was then switched to bypass the in situ
cell, thus allowing the cell to be purged with pure H2. Again, spectra were recorded until the
sample reached the new steady state under purging conditions. All resulting spectra were
normalized to a reference spectrum taken prior to exposure of the sample to propanoic acid at 413
K in H2, and converted to Kubelka-Munk units using the Kubelka-Munk equation.[40]
3
Results
3.1
Influence of Catalyst Composition and Pretreatment on Propanoic Acid Reduction
The catalytic reduction of 0.5 kPa propanoic acid in 0.1 MPa H2 at 433 K was studied over
12
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
Pd-promoted W oxide and phosphotungstic acid supported on SiO2, TiO2, and ZrO2. The steady-
state rate and product distribution of propanoic acid reduction are shown for each catalyst studied
(fresh and after pretreatment in H2 at 673 K) in Figure 2. Tungsten oxide and phosphotungstic
acid without a Pd promoter were completely inactive for the conversion of propanoic acid
Figure 2: Steady-state rate and product distribution of propanoic acid (0.5 kPa) reduction over
SiO2-, ZrO2-, and TiO2-supported Pd-promoted W-based catalysts in 0.1 MPa H2 at 433 K. Bars
correspond to product selectivity on the left axis, while light blue circles correspond to the rate
of propanoic acid conversion on the right axis. Values above the bars indicate conversion of
acid. NP indicates no in situ H2 pretreatment, while 673 K indicates a 1 h pretreatment in situ
in flowing H2 at 673 K prior to reaction.
13
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
under the conditions studied. Whereas fresh PdW supported on ZrO2 was active for propanoic acid
conversion (PdW/ZrO2 NP in Figure 2), pretreatment of the catalyst for 1 h in H2 at 673 K nearly
doubled the observed the rate of propanoic acid conversion and increased the selectivity towards
oxygenated products (propanal and 1-propanol). Fresh PdW/TiO2 was more active than
PdW/ZrO2, and more selective to oxygenated products, although a small amount of 1-
propoxypropane was also formed as a product. Figure 2 also shows that pretreatment of the
PdW/TiO2 in H2 at 673 K, compared to a non-pretreated (NP) catalyst, did not have any significant
effect on the rate of propanoic acid conversion, but selectivity to light hydrocarbons was reduced
by the treatment, leading to a 1-propanol selectivity of 92% at 13.5% conversion of the acid. Fresh
SiO2-supported PdW demonstrated the highest rate of propanoic acid reduction of the supported
WOx catalysts under the conditions studied, and was 87% selective to 1-propanol at 9.9% acid
conversion. Pretreatment of the PdW/SiO2 catalyst in H2 at 673 K resulted in a loss of nearly half
of the activity of the fresh catalyst (Figure 2).
Although fresh catalysts prepared with phosphotungstic acid were the most active for
propanoic acid conversion, they were 25-50% less selective to 1-propanol (at conversions of 26 to
36%) and produced more light hydrocarbon products. At the conversion levels reported in Figure
2, fresh PdPTA/TiO2 exhibited 5 times higher selectivity to alcohols than PdPTA/SiO2, while
retaining the improved activity of the PTA. On TiO2-supported catalysts, the formation of 1-
propoxypropane was also observed. Formation of ethers has been previously observed using TiO2-
supported W oxide and phosphotungstic acid catalysts.[41,42] As treatment at 673 K is known to
decompose PTA,[43] the loss of activity in the PTA catalysts after 673 K treatment in flowing H2 is
not surprising.
14
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
3.2
Influence of Reaction Conditions
The rate of reduction of carboxylic acids over fresh PdW/TiO2 and PdPTA/TiO2 slowly
decreased over time, with PdPTA/TiO2 deactivating faster than PdW/TiO2. The PdW/TiO2 lost
18% of its activity over 15.75 h time on stream, while the PdPTA/TiO2 lost 40% of its activity
over 21.5 h on stream. The rate of propanoic acid reduction versus inverse flowrate is plotted in
Figure 3(a). Differential conversion that could be used to evaluate a rate was not achieved until
conversion was below 5% over PdPTA/TiO2, and below 3% over PdW/TiO2. The selectivity of
propanoic acid conversion towards 1-propanol decreased in favor of propanal as conversion was
lowered. This influence of conversion on selectivity is plotted in Figure 4 and is consistent with
a two-step alcohol formation reaction proceeding through an aldehyde intermediate, which has
been reported previously.[44] Under differential conditions, the influence of temperature on the rate
of reaction was measured and is summarized in Figure 3(b). The apparent activation energy for
propanoic acid conversion was 54 kJ mol-1 over PdW/TiO2 (Figure 3(b)) and the orders of reaction
with respect to propanoic acid pressure and H2 pressure were 0.7 and 0.2, respectively, as
determined from the results reported in Figure 3(c) and (d). Similar behavior was observed over
the PdPTA/TiO2, shown in Figure S.1 in the supplemental information.
The influence of H2 pressure on product selectivity is shown in the supplemental
information in Figure S.2. While changing the H2 pressure did not have a strong effect on the rate
of propanoic acid reduction, lowering the H2 pressure first leads to an increase in 1-propanol
selectivity, followed by decreasing selectivity as the H2 pressure is lowered further. Moreover, the
rate of propanoic acid conversion became immeasurably small when H2 was completely removed
from the reactant stream. An inverse kinetic isotope effect of 0.91 was observed when the H2 in
15
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
the reactant stream was replaced by D2 (Table 1) during the reduction of propanoic acid by
PdW/TiO2.
16
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
Figure 3: (a) Influence of flow rate on conversion, (b) influence of temperature on rate of
propanoic acid conversion, (c) influence of propanoic acid pressure on rate of conversion, and
(d) influence of H2 pressure on rate of conversion during propanoic acid reaction using fresh
PdW/TiO2. All reactions were carried out at 413 K in 0.5 kPa propanoic acid and 0.1 MPa H2
unless otherwise specified. Dotted lines have been inserted to highlight observed trends.
17
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
Figure 4: Selectivity versus conversion for the reduction of propanoic acid in 0.5 kPa propanoic
acid and 0.1 MPa H2 at 413 K over (a) PdW/TiO2 and (b) PdPTA/TiO2. Solid lines are linear
interpolation between data points.
Table 1: Comparison of rate of reduction of 0.5 kPa propanoic acid in 0.1 MPa H2 or D2 at 413 K over PdW/TiO2.
Rate of reduction
Product selectivity
1-Propanol Propanal LHC 1-Propoxypropane
Overall isotope effect
(rateH/rateD)
Catalyst
H2/D2 Conversion [mol gcat-1 h-1]
H2
D2
6.7%
7.4%
4.5×10-4
4.9×10-4
0.74
0.74
0.07
0.06
0.07 0.11
0.07 0.12
PdW/TiO2
0.91
18
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
3.3
Adsorption of H2 and CO
The adsorption of H2 and CO on Pd-promoted WOx and PTA revealed differences in H2
uptake by the TiO2 and SiO2 supported catalysts (Table 2). The calculated Pd dispersions of
PdW/TiO2 and PdPTA/TiO2, assuming a ratio of 1.0 H/Pdsurface for TiO2 supported Pd, were 62%
and 58%, respectively. Assuming a ratio of 0.53 CO/Pdsurface,
[45] the calculated Pd dispersions of
PdW/TiO2 and PdPTA/TiO2 were 69% and 71%, respectively. In contrast, the calculated Pd
dispersions of PdW/SiO2 and PdPTA/SiO2, assuming a ratio of 1.0 H/Pdsurface for SiO2 supported
Pd, were 36% and 32%, respectively. Assuming a ratio of 0.75 CO/Pdsurface for the silica-supported
catalysts,[46] the calculated Pd dispersions of PdW/SiO2 and PdPTA/SiO2 were 25% and 40%,
respectively. Regardless of probe molecule, the dispersion of the Pd component is high and
corresponds to metal particle sizes ranging from 2 to 4 nm.
Table 2: Total chemisorption uptake of H2 and CO on Pd-promoted WOx catalysts.
-1
Chemisorption [µmol gcat
]
Sample
H2 (373 K)
CO (308 K)
H/Pd[a] (373 K) CO/Pd[a] (308 K)
PdW/TiO2
PdPTA/TiO2
PdW/SiO2
PdPTA/SiO2
33
31
19
17
39
40
20
32
0.62
0.58
0.36
0.32
0.37
0.38
0.19
0.30
(a) Calculated using nominal weight loading of 1 wt % Pd
3.4 X-Ray Absorption Spectroscopy
The X-ray absorption near edge structure (XANES) of PdW/SiO2 and W/SiO2 was
investigated at the W LI and LIII edges during in situ H2 temperature programmed reduction (TPR)
to determine the role of pretreatment on the electronic structure and coordination environment of
the W. As reported earlier, the treatment of PdW/SiO2 at 673 K in H2 results in a loss of nearly half
of its catalytic activity (Figure 2). The W LI-edge energy of the PdW/SiO2 and W/SiO2 samples,
measured at a step height of 0.75 in the normalized spectrum, was followed as the samples were
19
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
heated in flowing H2 up to 673 K (Figure S.5). On exposure of the samples to H2, the W LI-edge
energy of the PdW/SiO2 and W/SiO2 energy remained fairly constant until 570 K. Above 570 K,
the PdW/SiO2 and W/SiO2 LI-edge positions started to shift to lower energies, and continued to
shift monotonically until the end of the experiment. As shown in Figure 5, the W LI-edge position
of PdW/SiO2 and W/SiO2 both shifted 1.2 eV to lower energy over the course of the H2-TPR
experiment. The pre-edge features of both catalysts also shifted to slightly lower energy and
increased in intensity during the H2 treatment, which is consistent with an increased distortion of
the octahedral W center in WOx.[47]
20
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
Figure 5: (a) The W LI-edge XANES of PdW/SiO2 before pretreatment (short dash) and at 673
K in 0.1 MPa flowing H2 (solid) compared to the WO2 (long dash) and WO3 (dash-dot) standard
spectra. (b) The W LI-edge XANES of W/SiO2 before pretreatment (short dash) and at 673 K in
0.1 MPa flowing H2 (solid) compared to the WO2 (long dash) and WO3 (dash-dot) standard
spectra.
21
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
Figure 6: The W LIII-edge XANES spectra obtained during in situ H2-TPR of PdW/SiO2 in 6%
H2/He at 298 K, 440 K, and 670 K.
The W LIII-edge XANES were also investigated during H2-TPR, and representative spectra
of PdW/SiO2 are shown in Figure 6. During the H2-TPR, the whiteline feature was initially
asymmetrical, indicative of a slightly distorted octahedral structure typical of clusters of WVI,
including ammonium metatungstate and bulk WO3 (standards are shown in supplemental
information Figure S.6(a)). Upon heating to 440 K, however, the width of the white line
decreased, indicating a decrease in the splitting of the W 5-d orbitals.[47] This behavior has been
observed previously in WVI oxides and heteropolyacids, and is attributed to an increase of the
distortion of the octahedral coordination geometry of O surrounding W.[47–49] When the
temperature was increased from 440 K to 670 K, the W LIII edge position also shifted to lower
energy, suggesting that WVI was reduced to an average oxidation state between WIV and WVI. Prior
to the H2-TPR, the W LIII-edge position of both W/SiO2 and PdW/SiO2 was 10205.8 eV. After the
22
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
TPR, the edge positions shifted to 10205.2 eV (W/SiO2) and 10204.9 eV (PdW/SiO2). The edge
positions of the WO3 and WO2 standards were 10205.7 and 10204.7 eV, respectively, suggesting
that WOx in both catalysts had been reduced to an average oxidation state between WVI and WIV.
Evidently, the Pd-promoted W/SiO2 was reduced to a lower average oxidation state than the
unpromoted W/SiO2 after the TPR. This reduction behavior is illustrated in the supplemental
information (Figure S.6(b)), and supports the results obtained at the W LI edge. Analysis of the W
LIII-edge EXAFS (shown in Figure S.7) from 298 K to 440 K does not indicate any significant
change in the coordination number of the first shell W-O scattering path or the W-O scattering
path length (Table S.1). These results are consistent with minimal reduction of the W (which
remained in an oxidation state near WVI) in the PdW/SiO2 under the conditions studied.
Analysis of the Pd K-edge XANES revealed the as-synthesized catalysts contained only
PdO (Figure S.9). Qualitative analysis of the Pd K-edge EXAFS prior to reduction found similar
Pd-O first shell scattering paths in both samples and the PdO standard. The second shell Pd-Pd
scattering paths were less pronounced, however, resulting from a higher dispersion of Pd in the
supported PdO particles on the as-synthesized catalysts. Treatment in H2 at elevated temperature
reduced the PdO on both PdW/SiO2 and PdPTA/SiO2 to metallic Pd (Figure S.10). Fitting of the
EXAFS resulted in Pd-Pd first-shell coordination numbers of 8.5 and 7.4 in PdW/SiO2 and
PdPTA/SiO2, respectively, after treating the catalysts in H2 at 673 K. These first-shell coordination
numbers (Table S.2) are less than the value of 12 expected from bulk Pd, which confirms that Pd
in both catalysts was highly dispersed (60-80%), with an approximate particle diameter of between
1 and 2 nm, assuming a spherical particle geometry.[50]
23
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
3.5
Diffuse Reflectance Infrared Fourier Transform Spectroscopy
Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was used to probe the
surface species present under reaction conditions on the PdW/TiO2 and W/TiO2. When propanoic
acid in H2 was introduced to PdW/TiO2 and W/TiO2, similar diffuse reflectance spectra were
obtained, which are plotted in Figure 7. In the C-H stretching region, three weak peaks were
visible at 2985, 2949, and 2892 cm-1. These features are shifted to lower wavenumber than the
three main visible C-H stretching modes of vapor phase propanoic acid at 2993, 2954, and 2900
cm-1, indicating chemisorption of the propanoic acid on the PdW/TiO2 surface. A very weak
feature in the PdW/TiO2 spectrum in the presence of propanoic acid was observed at 1965 cm-1,
suggesting a small amount of strongly bound CO was present. In the carbonyl stretching region,
vibrational modes at 1791, 1772, and 1730 cm-1 are consistent with vibrations of vapor phase acid
monomers and dimers that have been observed in acetic acid.[51,52] Strong bands at 1661 and 1663
cm-1 cm were observed on PdW/TiO2 and W/TiO2 respectively, which were not observed during
the adsorption of propanoic acid on TiO2 or Pd/TiO2. This mode was assigned to an asymmetrical
O–C=O vibration characteristic of the binding of propanoic acid to the W component of the
catalyst. The broad band at 1511 cm-1 is assigned to bridging bidentate adsorption of propanoic
acid on TiO2, while the sharp feature at 1470 cm-1 is assigned to a C-H deformation mode in
adsorbed propanoic acid.[52] Features at 1435 cm-1 and 1418 cm-1 are assigned to the symmetrical
O–C=O vibrations of bidentate bridging propanoate on TiO2 and adsorption of propanoate on W,
respectively, and the feature at 1380 cm-1 is assigned to a C-H bending mode in adsorbed propanoic
acid.[52,53] These assignments are summarized in Table 3.
When the spectra collected in flowing H2 after propanoic acid treatment are subtracted
from the spectra obtained in flowing H2 and propanoic acid, the vapor-phase propanoic acid
24
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
features are clearly visible (Figure 7(a,c)), as is a carbonyl stretching mode near 1680 cm-1 (Figure
7(b,d)). Carbonyl stretching modes of similar values have previously been reported during
acetaldehyde adsorption on Pd/CeO2[54] and Pt/SiO2 promoted by Na,[55] and acetic acid reduction
on Pt/TiO2,[56] and were attributed to the formation of acetyl surface species. The difference spectra
also revealed a decrease in the intensity of the C-H stretching modes at higher wavenumber than
the species observed on both catalyst surfaces in the in situ or H2 purging spectra, indicating the
desorption of more weakly bound propanoic acid species during H2 purging.
Figure 7: Infrared spectra recorded at a sample temperature of 413 K showing PdW/TiO2 and
W/TiO2 in the (a, c, respectively) C-H stretching region and (b, d, respectively) carbonyl and
backbone deformation region during treatment in 0.3 kPa propanoic acid in 0.1 MPa H2 and
25
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
after treatment in 0.1 MPa H2 (purging). Difference spectra were obtained by subtracting the
spectra taken during H2 purging from the spectra obtained during propanoic acid treatment.
26
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
Table 3: Assignment of vibrational modes observed during reduction of 0.3 kPa propanoic acid in 0.1 MPa H2 at 413 K over
PdW/TiO2 and W/TiO2.
Adsorbent Conditions
PdW/TiO2 0.3 kPa propanoic ν(OH)
Vibrational mode Frequency [cm-1] Assignment
3500-3100 broad Propanoic acid, water (physisorbed)
or W/TiO2 acid, 0.1 MPa H2,
413 K
νa(CH3)
νa(CH2)
νs(CH3)
νs(CH2)
ν(C=O)
2985
2949
2930-2900 w
2892
Propanoic acid, bidentate[56,57]
Propanoic acid, bidentate
Propanoic acid, bidentate
Propanoic acid, bidentate
1791,1772,1732 Propanoic acid, (vapor, monomer and dimer)
1680
Propanoyl[54–56]
1661
1520
1470
1436, 1418
1380
Propanoic acid, WOx
Propanoic acid, bidentate[53,57], TiO2
Propanoic acid
νa(O—C=O)
δ(C—H)
νs(O—C=O)
δ(C—H)
Propanoic acid[53]
Propanoic acid
This article is protected b2y7copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
4
Discussion
4.1
Catalytic Activity of Pd-promoted WOx and PTA
Promotion of SiO2-, TiO2-, and ZrO2-supported WOx and PTA with Pd produced active
catalysts for the conversion of carboxylic acids that reduced propanoic acid at a higher rate per
mass of catalyst than Pd-promoted Re catalysts previously studied under similar conditions. For
example, Pd-promoted WOx (8 wt% W) and Pd-promoted ReOx (8 wt% Re) supported on the same
-1
SiO2 exhibited a rate of propanoic acid conversion of 1.5∙10-3 and 3.0∙10-4 mol gcat h-1,
respectively, at 433 K, 0.5 kPa acid and 0.1 MPa H2.[58] Prior work found that carboxylic acid
reduction is catalyzed by Pt/TiO2 at the Pt-TiO2 interface, wherein Pt was believed to supply H to
active sites on the TiO2 surface in a bifunctional surface reaction mechanism.[56,59] More recent
studies of carboxylic acid reduction catalyzed by Re and Pd-promoted Re (which are much more
active than the Pt/TiO2 system) have struggled to demonstrate that the cooperation of metal and
metal oxide sites is still required to produce a catalytically active site.[44,60–63] Studies suggest that
“intimate contact” between Pd and Re must be achieved to produce an active catalyst,[64] but as Re
is often reduced to a mixture of oxidized and metallic species under reaction conditions, the active
state of the Re is obscured. In this study, the combination of metallic Pd and W oxide produces an
active catalyst for carboxylic acid reduction when the two components are placed on the same
support. This result suggests that a bifunctional mechanism utilizing a metal (here Pd) to dissociate
H2, which makes adsorbed H species available to the catalytic active site (likely via H spillover)
for carboxylic acid reduction on the metal oxide (here WOx), is responsible for the observed
activity , as has been previously proposed in the Pd-promoted Re system.[58]
The observed activity depended strongly on both support and pretreatment of the catalyst,
which are believed to have a significant effect on the type, strength, and quantity of acid sites
28
This article is protected by copyright. All rights reserved.

10.1002/cctc.201901302
ChemCatChem
present on the WOx surface. Supports such as TiO2 and ZrO2, which form Brønsted acid sites when
loaded with dispersed WOx species,[14,17,18] produced catalysts that were not as active for the
reduction of propanoic acid as fresh SiO2-supported WOx. When Pd-promoted PTA was supported
on SiO2 and TiO2, higher activity for the reduction of propanoic acid was observed relative to Pd-
promoted WOx on SiO2 or TiO2. Despite their higher activity, PdPTA catalysts were only half as
selective for the formation of oxygenated products, and the observed rate of formation of
oxygenated products over PdPTA was still comparable to the rate of oxygenate formation observed
over the PdW catalysts on SiO2 and TiO2. In light of this observation, we propose that higher
surface densities of Brønsted acid sites are detrimental and lead to less selective reduction of the
carboxylic acid functionality, as observed for PdW supported on ZrO2 and the supported
heteropolyacids. Presumably, Brønsted acid sites catalyze unselective dehydration steps that lead
to hydrocarbons and ethers. Yamaguchi et al. studied the isomerization of cyclopropane over
similarly made WOx catalysts and found that WOx/ZrO2 was more active for the isomerization
reaction, while WOx/TiO2 was less active and WOx/SiO2 demonstrated no activity. This trend
followed the previously reported acid site strength of each catalyst, where WOx/ZrO2 and
WOx/TiO2 are claimed to have stronger acid sites, and WOx/SiO2 had weaker acid sites.[14]
However, the trend in reactivity observed here was the opposite. Apparently, supports that are
expected to generate weaker Brønsted acid sites when combined with WOx can be correlated with
higher activity for the reduction of carboxylic acids.
Strong Lewis acid sites may also catalyze the C-O bond breaking of weakly adsorbed
carboxylic acids. A positive reaction order in propanoic acid was observed, which contrasted
previously studied Re and Pd-promoted Re systems wherein the rate was nearly zero order in
acid.[44,58,61] Under similar conditions, Pd-promoted Re was found to reside in an average oxidation
29
This article is protected by copyright. All rights reserved.