Direct synthesis of furfuryl alcohol from furfural: Catalytic performance of monometallic and bimetallic Mo and Ru phosphides

Article abstract of DOI:10.1039/c9cy00705a

The catalytic properties of monometallic and bimetallic Ru and Mo phosphides were evaluated for their ability to selectively hydrogenate furfural to furfuryl alcohol. Monometallic MoP showed high selectivity (98%) towards furfuryl alcohol, while RuP and Ru₂P exhibited lower selectivity at comparable conversion. Bimetallic promotional effects were observed with Ru_1.0Mo_1.0P, as the pseudo-first order reaction rate constant for furfural hydrogenation to furfuryl alcohol, k₁, was at least 5× higher than MoP, RuP, and Ru₂P, while maintaining a 99% selectivity. Composition-directed catalytic studies of Ru_xMo_2-xP (0.8 < x < 1.2) provided evidence that Ru rich compositions positively influence k₁, but not the selectivity. The rate constant ratio k₁/(k₂ + k₃) for furfuryl alcohol production compared to methyl furan (k₂) and tetrahyrofurfuryl alcohol (k₃) followed the trend of Ru_1.0Mo_1.0P > Ru_1.2Mo_0.8P > MoP > Ru_0.8Mo_1.2P > RuP > Ru₂P. Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was used to examine the configuration of adsorbed furfural on the synthesized catalysts, but the results were inconclusive and no correlation could be found with the selectivity due to the possible IR inactive surface modes with furfural adsorption. However, gas phase density functional theory calculations suggested the x = 1.0 material in Ru_xMo_2-xP (0.8 < x < 1.2) had the most favorable furfural adsorption energy. Experimentally, we also observed that the solvent greatly influenced both the conversion and selectivity, where isopropanol provided the highest selectivity to furfuryl alcohol. Finally, recycling experiments showed a 12% decrease in k₁ after 3 cycles without any regeneration, but the activity could be fully recovered through a re-reduction step.

Full text of DOI:10.1039/c9cy00705a

View Article Online
View Journal
Catalysis
Science &
Technology
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: Y. Bonita, V. Jain,
F. Geng, T. O'Connell, W. Wilson, N. Rai and J. C. Hicks, Catal. Sci. Technol., 2019, DOI:
10.1039/C9CY00705A.
This is an Accepted Manuscript, which has been through the
Volume 6 Number 1 7 January 2016 Pages 1–308
Royal Society of Chemistry peer review process and has been
accepted for publication.
Catalysis
Science &
Technology
Accepted Manuscripts are published online shortly after
www.rsc.org/catalysis
acceptance, before technical editing, formatting and proof reading.
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the
author guidelines.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
ISSN 2044-4753
standard Terms & Conditions and the ethical guidelines, outlined
in our author and reviewer resource centre, still apply. In no
event shall the Royal Society of Chemistry be held responsible
PAPER
Qingzhu Zhang et al.
Catalytic mechanism of C–F bond cleavage: insights from QM/MM
analysis of fluoroacetate dehalogenase
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.
rsc.li/catalysis

Please do not adjust margins
Catalysis Science & Technology
Page 1 of 14
View Article Online
DOI: 10.1039/C9CY00705A
ARTICLE
Direct Synthesis of Furfuryl Alcohol from Furfural: Catalytic
Performance of Monometallic and Bimetallic Mo and Ru
Phosphides
Received 00th January 20xx,
Accepted 00th January 20xx
Yolanda Bonitaa, Varsha Jainb, Feiyang Genga, Timothy P. O’Connella, Woodrow N. Wilsonb, Neeraj
DOI: 10.1039/x0xx00000x
Raib, and Jason C. Hicksa*
The catalytic properties of monometallic and bimetallic Ru and Mo phosphides were evaluated for their ability to
selectively hydrogenate furfural to furfuryl alcohol. Monometallic MoP showed high selectivity (98%) towards furfuryl
alcohol, while RuP and Ru2P exhibited lower selectivity at comparable conversion. Bimetallic promotional effects were
observed with Ru1.0Mo1.0P, as the pseudo-first order reaction rate constant for furfural hydrogenation to furfuryl alcohol,
k1, was at least 5x higher than MoP, RuP, and Ru2P, while maintaining a 99% selectivity. Composition-directed catalytic
studies of RuxMo2-xP (0.8 < x < 1.2) provided evidence that Ru rich compositions positively influence k1, but not the
selectivity. The rate constant ratio (k1/k2+k3) for furfuryl alcohol production compared to methyl furan (k2) and
tetrahyrofurfuryl alcohol (k3) followed the trend of Ru1.0Mo1.0P > Ru1.2Mo0.8P > MoP > Ru0.8Mo1.2P > RuP > Ru2P. Diffuse
reflectance infrared Fourier transform spectroscopy (DRIFTS) was used to examine the configuration of adsorbed furfural
on the synthesized catalysts, but the results were inconclusive and no correlation could be found with the selectivity due
to the IR inactive surface modes possible with furfural adsorption. However, gas phase density functional theory
calculations suggested the x = 1 material in RuxMo2-xP (0.8 < x < 1.2) had the most favorable furfural adsorption energy.
Experimentally, we also observed that the solvent greatly influenced both the conversion and selectivity, where
isopropanol provided the highest selectivities to furfuryl alcohol. Finally, recycling experiments showed a 12% decrease in
k1 after 3 cycles without any regeneration, but the activity could be fully recovered through a re-reduction step.
electronic effects from well-distributed metal atoms or charge
transfer between the metals contributed to variations in the
catalytic performance for FAL hydrogenation.9, 10 As a result, a
Introduction
Furfuryl alcohol (FOL) is an important chemical intermediate
with a variety of applications in resins, lubricants, fragrances,
flavorings, and lysine production.1-4 In current furfuryl alcohol
production, Cu-chromite (CuCr2O4) is used as a catalyst to
selectively hydrogenate furfural (FAL), a platform molecule
that is also derived from lignocellulosic biomass.5, 6 Although
Cu-chromite achieves high selectivity to FOL (>99%) at
complete conversion, two primary issues have been identified
with this catalytic system: (1) environmental concerns exist
with the disposal of spent Cu-chromite catalysts,5 and (2) Cu-
chromite suffers from rapid deactivation due to coke
formation or via a change in the Cu oxidation state during
reaction.5, 7, 8
shift in the product distribution was observed where
methylfuran (MF) was the dominant product.9, 10 Noble metal-
based catalysts such as Pt, Rh, Pd, and Ru have also been
reported as highly active catalysts for FAL hydrogenation.15
Nevertheless, the main hydrogenation product observed was
tetrahydrofurfuryl alcohol (THFA), which was formed via FOL
ring hydrogenation. Efforts to tune the selectivity of noble
metal catalysts have included pairing the noble metal with a
less reactive metal such as Sn, where the incorporation of Sn
can either improve the FAL selectivity when paired with Ru or
increase the reactivity when paired with Pt.16, 17 Other metals
such as Mo, Mn, and Fe have also been shown to enhance the
selectivity to FAL in Pt-based materials.18
A variety of alternative Cu-based catalysts have been studied
in significant detail, including Cu/C, Cu/ZnO, Cu-Ni, Cu-Co, and
Cu-Fe.9-14 In bimetallic Cu-Fe and Cu-Ni, structural and
Noble metal based phosphides such as RuxP, Rh2P, and Pd3P
were able to deoxygenate biomass-derived molecules,19-21
with MoP and Ni2P as potential catalysts for vapor phase
furfural hydrogenation to form methyl furan.22 We have
recently reported the low temperature hydrogenation ability
of noble metal-based bimetallic Ru1.0Mo1.0P for hydrogenation
of various aromatic functionalities such as benzaldehyde to
benzyl alcohol.23, 24 The Lewis acidic nature of Ru1.0Mo1.0P was
responsible for the reduction of the aldehyde and is a
reasonable catalyst for the selective production of FOL from
FAL.24
a. Department of Chemical and Biomolecular Engineering, University of Notre
Dame, Notre Dame, Indiana 46556, United States. E-mail: jhicks3@nd.edu
b. Dave C. Swalm School of Chemical Engineering and Center for Advanced Vehicular
Systems, Mississippi State University, Mississippi State, Mississippi 39762, United
States
Electronic Supplementary Information (ESI) available: [details of any
supplementary information available should be included here]. See
DOI: 10.1039/x0xx00000x
Please do not adjust margins

Please do not adjust margins
Catalysis Science & Technology
Page 2 of 14
ARTICLE
Journal Name
Herein, we report a detailed comparison of monometallic
MoP, Mo3P, RuP, Ru2P phosphides and bimetallic RuxMo2-x
region was deconvoluted into the reduced Mo 3d5/2 and 3d3/2 pair.
View Article Online
P
The contribution from the oxidized speciesDcOanI: 1b0e.10o3b9s/eCr9vCeYd0i0n70th5Ae
satellite peak that was deconvoluted into the oxidized Mo 3d5/2 and
3d3/2 pair. The low amount of oxidized species observed was
attributed to the brief exposure to air during sample transfer or
possibly an oxidized layer that formed during passivation. The P 2p
region was deconvoluted into the reduced 2p3/2 and 2p1/2 doublet
and the oxidized 2p3/2 and 2p1/2 doublet at higher binding energies.
The Ru 3p region was deconvoluted into the reduced 3p3/2 and 3p1/2
doublet as well as its oxidized contributions. The Ru 3p region was
used to determine the surface composition of the materials since
the Ru 3d region overlaps with the C 1s peak.
phosphides for FAL hydrogenation. A series of experiments
and simulations were used to determine the following: (1) if
transition metal phosphides were capable of performing
selective hydrogenation of FAL to FOL, (2) if an enhancement
in catalytic activity and selectivity is observed with bimetallic
materials, and (3) the catalytic consequences of changing the
metal ratio in RuxMo2-xP. We close this manuscript with
recycle studies of high performing bimetallic catalysts.
Experimental
X-ray absorption spectroscopy (XAS) experiments were performed
at Argonne National Laboratory in the Advanced Photon Source
sector 10 Insertion Device (ID) line of the Material Research
Collaborative Access Team (MRCAT). Boron nitride was used to
dilute the samples to obtain a sufficient signal. Approximately 10
mg of the mixture was pressed into a pellet in the sample holder.
The sample was placed inside a sealed pretreatment cell, and the
samples were pretreated at 450 °C under 4% of H2 in He for 1 hr.
After the cell was cooled down, the cell was sealed and placed for
analysis. The measurements were obtained under transmission
mode with a 105 photon flux per second. WinXAS 3.1 was used to
analyze the data.
Materials
Citric acid monohydrate (Amresco, 99%), (NH4)6Mo7O24·4H2O (Alfa
Aesar, 99%), RuCl3·xH2O (Oakwood Chemicals, 90%), (NH4)2HPO4
(Amresco, 98.6%), furfural (Sigma Aldrich, 99.5%), furfuryl alcohol
(TCI Chemicals), 2-methyl furan (TCI Chemicals, 99%),
tetrahydrofurfuryl alcohol (Alfa Aesar, 99%), 2-propanol (J.T. Baker,
99%), 5% Ru/Al2O3 (Riogen).
All of the gases are purchased from Airgas: H2 (99.999%), N2
(99.999%), 1%O2/He, 30%CO/He, 2%NH3/He.
Material synthesis
All materials were synthesized using a temperature programmed
reduction (TPR) method described in our previous work and
included in the supporting information.23, 25-27
A Bruker Vertex 70 Fourier transform infrared (FTIR) spectrometer
equipped with a Harrick Praying Mantis diffuse reflectance high
temperature reaction cell (ZnSe windows) accessory was used to
study the interaction of furfural with the catalyst surface. In a
typical experiment, the catalyst sample (20 mg) was loaded in the
sample cup and pretreated with 30 mL/min of H2 at 400°C for 2 h
followed by 30 mL/min of N2 at 400°C for 1 hour. After the
pretreatment, the sample was cooled down to room temperature,
and a background measurement was performed. Furfural was
loaded into a glass bubbler with 30 mL of N2 flow used as the carrier
gas to saturate the sample for 30 mins at room temperature. The
saturated samples were then purged with N2 for 30 mins. The IR
spectra were recorded over 1000 scans with a 1 cm-1 resolution
with a liquid N2 cooled mercury cadmium telluride (MCT) detector.
Catalyst characterization
The synthesized materials were characterized using various
methods. A Bruker powder x-ray diffractometer (XRD) with a Cu-Kα
source was used to confirm the crystal structures of the materials.
Rietveld refinement was performed using FullProf Suite software. A
Si (111) standard was added, and the corresponding peak was
shifted to 28.44° 2θ. The Brunauer-Emmett-Teller (BET) surface
area was measured with a Quantachrome Nova 2200e with N2 at 77
K. A Micromeritics Chemisorb 2750 unit equipped with a thermal
conductivity detector (TCD) was used to quantify the CO-accessible
sites. For CO-pulse chemisorption experiments, the samples were
pretreated under 20 mL/min of H2 flow at 400°C with a 10°C/min
ramp rate for 2 h followed by 20 mL/min of He flow at the same
temperature for 1.5 h and 20 mL/min of H2 flow at 100°C for 1 h to
prevent polycarbonyl formation.28 The CO-pulse was performed at
35°C with 0.1 mL injections of 30% CO in He. The amount of CO
chemisorbed was quantified using an external calibration curve.
The desorption was performed using a 10°C/min ramp rate to a
final temperature of 450°C. Elemental analysis was done with a
Perkin Elmer Optima 8000 Inductively Coupled Plasma (ICP) –
Optical Emission Spectroscopy (OES) system and quantified using an
external calibration.
Catalyst testing
The catalytic testing was performed in a 300 mL Parr batch reactor.
Based on the CO chemisorption experiments, the reactor was
loaded with the same amount of catalyst in each experiment (0.45
μmol of CO sites for each catalyst). The reaction media consisted of
40 mL of 0.1 M FAL in isopropanol with dioxane added as an
internal standard unless otherwise specified. The reactor was
charged with H2 to the pressure of interest. Using a 1/16” sampler
tube installed inside the reactor, 250 μL samples were taken at
various times during the reaction. The line was flushed between
each sampling in order to prevent any cross contamination with
previous sampling times. The reaction results were analyzed using
gas chromatography – mass spectrometry (GCMS, Agilent 5975-
7890) and quantified with an external calibration curve created for
each reactant and product. The carbon balance for all points was
>95%. The conversion and selectivity were calculated according to
X-Ray Photoelectron Spectroscopy (XPS) was used to determine the
binding energies of the various catalysts. All sample preparations
occurred in a N2 glove box, and the exposure to air during sample
transfer was minimized. Elemental Ni powder was physically mixed
into the sample as a standard where the Ni 2p3/2 peak was shifted
to 852.60 eV. Asymmetric peak fitting was applied in the
deconvolution of XP spectra peaks.29-31 The XP spectra in the Mo 3d
2 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 3 of 14
Catalysis Science & Technology
Please do not adjust margins
Journal Name
ARTICLE
(112) plane to expose an atomic layer that has all three types
Equation (1) and (2) respectively where XFAL is the FAL conversion,
View Article Online
of atoms in each catalytic surface. We applied the same
DOI: 10.1039/C9CY00705A
SFOL is selectivity to FOL, and Cj is the concentration of the species, j.
푪푭푨푳, 풐풖풕
approach to the (210) plane for comparison between the two
surfaces. The number of phosphorus atoms was maintained
constant (60 atoms) for all the calculations, whereas the
number of Ru and Mo atoms was governed by the
stoichiometry.
푿푭푨푳 = ퟏ ―
(1)
푪푭푨푳, 풊풏
푪푭푶푳
푺푭푶푳
=
(2)
푪푭푶푳 + 푪푴푭 + 푪푻푯푭푨
The conversion over time was fitted into a pseudo-1st order batch
reactor equation (Equation 4) that was obtained from Equation 3 to
extract the reaction rate constant, k1, for furfural hydrogenation to The slab used for simulations consists of three layers in
RuxMo2-xP with a supercell size P (2 X 2) with a 15 Å thick
vacuum layer in the z-direction (perpendicular to the surface)
to minimize interactions between neighboring image slabs. For
all calculations, the bottom most layer was fixed to represent
the corresponding crystal structure obtained from DFT,
whereas all other atoms were allowed to relax.
furfuryl alcohol.
풅푪푭푨푳
= ― 풌ퟏ 푪푭푨푳
풅풕
(3)
(
)
― 퐥퐧 ퟏ ― 푿푭푨푳 = 풌ퟏ 풕
(4)
Subsequent reactions with 0.10 M furfuryl alcohol as the substrate
were performed to extract the rate constant of MF and
tetrahydrofurfuryl alcohol (THFA), denoted as k2 and k3,
respectively. Equation 5 was derived for parallel 1st order reactions
in a batch reactor. Accordingly, a plot of ln (1-XFOL) against time
resulted in a slope of k2 + k3.
The adsorption energies (denoted as EAD) were calculated
according to Eqn. (9), wherein EADSORBATE+SURFACE is defined as
the total energy of species adsorbed on the surface; ESURFACE is
defined the total energy of surface; and EADSORBATE is defined as
the energy of the adsorbed species on the surface in the gas
phase.
( )
푭푶푳 = 푪푭푨푳,풊풏퐞퐱퐩 ( ― 풌ퟐ + 풌ퟑ 풕)
푪
(5)
The time-dependent CMF and CTHFA values in Equation 6 and 7 were
then derived. The ratio between CMF and CTHFA resulted in Equation
8, which was plotted to obtain k2/k3. Using these equations, k2 and
EAD = EADSORBATE+SURFACE −ESURFACE −EADSORBATE
(9)
k3 were determined.
Result and Discussion
Catalytic evaluation of monometallic phosphides
풌ퟐ 푪푭푶푳,풊풏
(
(
)
)
푪푴푭
=
ퟏ ― 퐞퐱퐩 ( ― 풌ퟐ + 풌ퟑ 풕)
(6)
(7)
(8)
풌ퟐ + 풌ퟑ
풌ퟑ 푪푭푶푳,풊풏
(
(
)
)
푪푻푯푭푨
=
ퟏ ― 퐞퐱퐩 ( ― 풌ퟐ + 풌ퟑ 풕)
풌ퟐ + 풌ퟑ
Four unsupported monometallic phosphides MoP, Mo3P, RuP,
and Ru2P were synthesized using the TPR method described in
detail in the supporting information. The crystal structures
were confirmed using XRD (Figure S1 and Figure S2).
풌ퟐ
푪푴푭
=
푪푻푯푭푨
풌ퟑ
Computational details
We performed plane wave periodic density functional theory
Table
1 summarizes the textural properties of all the
phosphide catalysts used in this study as well as results from
CO titrations (NCO). The SBET was obtained using N2
physisorption, and the values ranged from 6.1 m2/g – 12 m2/g
for all materials except for MoP (32 m2/g), which are similar to
other reports.49-51 The SBET of Mo3P was lower than MoP due to
higher reduction temperature needed to synthesize Mo3P
(800°C). It is possible to vary the amount of citric acid added
during synthesis to increase the surface area of the resulting
materials.52 Supporting the phosphides can improve the
surface area; however, multiple phases can be formed, and the
33
(DFT)32,
calculations using Vienna ab initio simulation
package (VASP.5.4.4).34-37 We used the optB88-vdW functional,
which accounts for nonlocal, long-range van der Waals
interactions.38, The core electrons are described with the
39
projector augmented wave (PAW) method40 to solve the Kohn-
Sham equations.41, 42 The energy cut-off was taken as 450 eV
to ensure high precision. Total energies were calculated using
a first-order Methfessel−Paxton smearing function with a
width of 0.1 eV, and the total energy was extrapolated to 0 K.43
Optimizations were carried out until the net forces acting on
atoms were smaller than 0.03 eV Å−1, using a total energy
convergence of 1x10−5 eV. The calculations were carried out
with spin polarization. For the integration of the Brillouin zone
(BZ), we used a Γ-point sampling in all calculations.44 The
partial charges on various species were derived using a Bader
charge analysis.45-48
various phosphide phases may add complexity to the
evaluation of their catalytic performance.53,
54
CO-pulse
chemisorption was used to quantify the number of CO
adsorption sites for each of the phosphide catalysts. The CO-
titration results provided similar values ranging between 20-28
μmol/g on all materials, and these values were used to
standardize the experiments as well as normalize the reaction
rates in some cases.
Crystal structures of three RuxMo2-xP catalysts were optimized
in three-dimensional periodic boundary conditions based on
the cell parameters determined via XRD patterns of these
catalysts. Optimized unit cell parameters in these catalysts
showed a deviation of ≤ 1 % in comparison with those
experimentally determined (Table S1). Using these optimized
cell parameters, the model phosphide surface was constructed
by implementing the standard slab approach, in which a slab
of finite thickness was cut out of the RuxMo2-xP crystal at the
Table 1. Textural properties of Mo-based and Ru-based
Catalysts
MoP
SBET (m2/g)
NCO (μmol/g)
32
6.1
12
28
25
23
Mo3P
RuP
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 3
Please do not adjust margins

Please do not adjust margins
Catalysis Science & Technology
Page 4 of 14
ARTICLE
Journal Name
Ru2P
10
9.2
7.2
11
24
21
20
25
provided a suitable kinetic region to compare catalytic
View Article Online
DOI: 10.1039/C9CY00705A
materials.59
Ru0.8Mo1.2
Ru1.0Mo1.0
Ru1.2Mo0.8
P
P
P
In addition to the H2 pressure study, the initial concentration
of FAL was varied to confirm the pseudo-1st order trend
initially observed. With a 4.2 MPa H2 pressure, five initial
concentrations of FAL in IPA (0.10M, 0.15M, 0.18M, and 0.25
M) were studied in a batch reactor at 100°C with real-time
sampling of the reaction progress. The reaction results are
plotted as -ln (1-X) versus time in Figure S5 for all initial FAL
concentrations. The linear trend observed in the plot further
indicated pseudo-1st order behavior in FAL for all initial
concentrations. Moreover, the k-values were extracted from
the slopes to yield k1 of 0.176 h-1, 0.177 h-1, 0.171 h-1, and
0.174 h-1 for initial concentration of 0.10 M, 0.15 M, 0.18 M,
and 0.25 M, respectively. At higher FAL concentration (i.e.,
0.50 M), the solubility of H2 decreases and new reaction
conditions to satisfy the pseudo-1st order model with excess H2
in solution are requried.60 Therefore, to fulfill the pseudo-1st
order assumptions, an initial concentration of 0.10 M for FAL is
used throughout the experiments. At lower concentrations
(0.01 M – 0.05 M), lower than unity order dependency was
observed as shown in other reports for FAL hydrogenation
over a Pt/C catalyst.55 The rate expression can therefore be
simplified to Equation 3, with k1 representing the lumped,
pseudo-1st order reaction rate constant.
Scheme 1 represents the reaction pathways observed in this
study. In the first hydrogenation step, FAL is converted into
FOL with a rate constant denoted as k1. To measure this rate
constant, the monometallic catalysts were tested in a batch
reactor with 0.10 M FAL in isopropanol at 100°C and 4.2 MPa.
Using the CO chemisorption values, the catalyst loading in
each experiment was maintained at 0.45 μmol of CO accessible
sites for all experiments based on the invariance observed
between the consumption rate of furfural with respect to the
catalyst amount (Figure S3) and prior literature suggesting CO
titrated sites as the active sites for FAL hydrogenation.5,
16
Samples were taken throughout the reaction, and the
conversion was plotted according to Equation 4 to obtain a
pseudo-1st order rate constant, k1.
Background experiments were also performed to quantify the
contribution of H2 transfer from isopropanol (solvent). After 24
h of reaction time at 4.2 MPa of N2 pressure for proper
comparison, the furfural conversion was 1.1% with only
isopropanol as the H2 source. Therefore, these contributions
were considered negligible in this study. After linearization of
Equation 10, the various catalysts were studied and compared
(Figure 1). The rate constant, k1, was measured as 0.030 h-1,
0.027 h-1, and 0.029 h-1 for RuP, Ru2P, and MoP, respectively.
These rate constants were similar in magnitude for all
monometallic materials. Mo3P was also tested for this
reaction, but it was inactive at temperatures tested below
125°C, which is likely due to unfavorable adsorption of the
reactant as also noted by Xiao et al.61 At much higher
temperatures (>300°C), however, Mo3P was active with ~80%
selectivity to 2-(isopropoxy)methyl furan from the
etherification reaction between FAL and isopropanol. This
result is a possible indication of the weak (or lack of) BrØnsted
acidity of the phosphides, resulting in the production of the
ether product.24, 62-64
Scheme 1. Reaction pathway of FAL hydrogenation to produce
FOL, MF, and THFA
Several reaction mechanisms based on
a
Langmuir-
Hinshelwood model have been proposed for FAL
hydrogenation involving adsorption of both H2 and FAL.11, 55-58
From these mechanisms, several rate laws has been derived in
the literatures.11, 56, 57 In each of these cases, dependence on
both H2 and FAL concentrations complicate the rate
measurements.
To simplify the kinetic expression, the reaction was operated
at the appropriate pressure such that the rate depended only
on FAL and not H2. To confirm that the operating H2 pressure
was in the 0th order region, several experiments were
performed at 100°C with different partial pressures of H2 while
keeping the total pressure of 4.2 MPa using N2 as the
balancing gas. The results are plotted in Figure S4 as -ln (1-X)
against reaction time, where X is the FAL conversion. The
reaction constant, k1, was obtained from the slope of the plot
in Figure S4 for 2.1 MPa, 3.4 MPa, 3.8 MPa, and 4.2 MPa H2
partial pressure. The linear dependency between -ln (1-X) and
reaction time observed in the plot implied a pseudo-1st order
dependency in FAL concentration could be assumed.
Moreover, the kinetic profiles and k1 were statistically
invariant according to the Dixon q-test (Equation S1) in
experiments between 3.4 – 4.2 MPa H2 partial pressure, which
4 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 5 of 14
Catalysis Science & Technology
Please do not adjust margins
Journal Name
ARTICLE
Figure 1. The difference in rate constant k1 between monometallic is higher than RuP (461.8 eV). Both of these binVdieinwgArteicnleeOrgnliienes
RuP (green – R2 = 0.976), Ru2P (blue – R2 = 0.984 ), and MoP (red – showed slightly oxidized Ru atoms since the binding energy shift of
DOI: 10.1039/C9CY00705A
R2 = 0.987)
Ru 3p3/2 in Ru/Al2O3 was measured to be 461.6 eV. Similarly, the P
atom in Ru2P is more anionic than in RuP.
All active catalysts also showed a high selectivity to FOL
production. At a conversion of 15%, the selectivity towards X-ray absorption spectroscopy experiments were also performed
FOL was 98 ± 0.8% for MoP, while the selectivity decreased for the various monometallic phosphides. From the XANES results in
slightly to 97 ± 1.2% and 94 ± 1.3% for RuP and Ru2P, Table S4 and Figure S7, the Mo in MoP (20,001.25 eV) and Mo3P
respectively, with MF observed as the other by-product. It was (20,000.61 eV) are slightly oxidized compared to the Mo-foil
apparent that MoP showed the highest selectivity towards (20,000.00 eV) with Mo3P being less oxidized, which agrees with the
FOL, while the highest rate constant was observed with RuP.
XPS measurements. The Ru K-edge absorption energy for Ru2P was
determined to be 22,117.42 eV, which is also considered oxidized
Subsequently, FOL was used as a reactant to determine k2 and k3, compared to the reference Ru foil (22,117.2 eV). Based on these
which correspond to rate constants to produce MF and THFA, results, the Mo and Ru are electron donators in the monometallic
respectively (Scheme 1). The constants were determined with batch Ru and Mo phosphides, while P acts as an electron acceptor.
reaction experiments using 0.10 M FOL in isopropanol as a reactant.
Equation 5 and Equation 8 were used to extract k2 and k3 values As shown in Table S6, partitioning of charge density of surface
simultaneously. These values were then used to model the atoms according to the Bader scheme reveals depletion of
concentration of FOL, MF, and THFA with all active monometallic electronic charge density for Mo in MoP (+0.52 eV) and Mo3P
catalysts. The model was plotted with the raw data for MoP, RuP, (+0.91 eV) as well as Ru in RuP (+0.11 eV) and Ru2P (+0.13 eV).
and Ru2P in Figure 2a-c. From these plots, it could be seen that the Meanwhile, the Bader charge of the P atoms were -0.52 eV, -0.91
model fits the raw well. The k-values are compiled in Table S3 for eV, -0.11 eV, and -0.13 eV for MoP, Mo3P, RuP, and Ru2P. The Bader
comparison. The rate constants for MF formation (k2) were 0.0035 charge calculations were performed on the most dominant facet
h-1, 0.0013 h-1, and 0.0002 h-1 for MoP, RuP, and Ru2P, respectively. observed in XRD, which were the (101), (321), (211), and (211) for
The rate constants for THFA (k3) were 0.0185 h-1 and 0.1752 h-1 for MoP, Mo3P, RuP, and Ru2P, respectively. These calculations also
RuP and Ru2P, respectively, where no THFA was observed with confirm the electron transfer in monometallic Ru and Mo
MoP. From these measurements, Ru2P displayed the highest k3 phosphides, where Ru and Mo donate electrons to P. Taken
value, which suggested high preference for THFA production. High together, XPS, XANES, and Bader charge calculations suggest that
selectivities to THFA from FOL hydrogenation have also been the Ru and Mo in monometallic phosphides possess Lewis acidic
observed in the literature with Ru/TiO2 along with other noble character (e.g., Ruδ+ and Moδ+), which can serve as the reactant
66
metal catalysts (Pd, Rh) and Ni boride catalysts.65, It has been adsorption site in metal phosphides.69
noted that phosphides and borides share a similar metal-metalloid
interaction (M-B and M-P) as well as B-B and P-P interactions.67 It is Catalytic evaluation of bimetallic Ru1.0Mo1.0
P
plausible that the formation of THFA observed in both phosphides
and borides is due to the existence of similar sites (i.e., Mδ+ or Pδ-).15 Our previous studies have provided evidence of catalytic
enhancements with bimetallic phosphides in comparison to
The binding energy shifts of the monometallic Ru and Mo their monometallic counterparts.27 Due to the high selectivity
phosphides were measured with XPS to provide the relative of MoP to FOL and the higher activity of RuP, we first
oxidation of the materials (Table S4). The Mo 3d5/2 binding energy synthesized a bimetallic Ru1.0Mo1.0P catalyst (Table 1). The k1,
shift was determined as 228.2 eV and 227.5 eV for MoP and Mo3P, k2, and k3 values for Ru1.0Mo1.0P were determined similarly to
which were more oxidized when compared to Mo0 (i.e. 226.7 eV). the methods described in the previous section at 100°C and
Conversely, the P 2p3/2 binding energy shift was found to be 129.5 4.2 MPa H2. The measured k1 for Ru1.0Mo1.0P was 0.176 h-1,
eV and 129.1 eV for MoP and Mo3P, respectively. These values were which was >6 times higher than the monometallic MoP, RuP,
in the range of negatively charged P (i.e. 130.0 – 130.9 eV).68
similar observation was also observed with the monometallic Ru improvement due to inherent bimetallic effects associated
phosphides. The binding energy shift of Ru 3p3/2 in Ru2P (462.0 eV) with the Ru1.0Mo1.0 catalyst. The k2 and k3 values of
A
and Ru2P.
This result showcased the reaction rate
P
Figure 2. Concentration profiles for FOL (black filled circles), MF (red filled squares), and THFA (blue open squares) with (a) MoP,
(b), RuP, and (c) Ru2P determined experimentally (circles and squares) and fitted to the predicted reaction model (solid line).
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 5
Please do not adjust margins

Please do not adjust margins
Catalysis Science & Technology
Page 6 of 14
ARTICLE
Journal Name
Ru1.0Mo1.0P were 0.0050 h-1 and 0.0039 h-1, respectively. However, these barriers were not extracted froVmiewaAdrtdiclietiOonnlinael
DOI: 10.1039/C9CY00705A
Although Ru1.0Mo1.0P produced both MF and THFA, the experiments because the yields of MF were not sufficient
production rates were low.
enough to provide justification for using higher temperatures
to control the reaction selectivities. However, the effective
activation barrier (EA,1) was obtained during the temperature
sweep experiments for the initial reduction of FAL. The
activation barrier was ~51 kJ/mol, which was similar to those
reported for the commercial Cu-chromite (46 kJ/mol) and
Cu/SiO2 (50 kJ/mol).5, 57
20
15
10
5
0
P
o
P
P
P
2
u
0
u
.
R
M
1
o
R
M
0
.
1
u
R
Figure 3. The ratio of k1/(k2+k3)) for MoP, Ru1.0Mo1.0P, RuP, and
Ru2P
The highest FOL selectivity was observed with Ru1.0Mo1.0
P
followed by MoP > RuP > Ru2P, respectively. It was evident
that the addition of P weakened the hydrogenation ability of
Ru in the various Ru-based phosphides. Similar observations
were seen in a P-doped Ru(0001) surface where the apparent
charge transfer between Ru and P atoms reduced the electron
back donation from Ru to the reactant.19 Moreover, the
addition of Mo shifted the hydrogenation preference from
aromatics to the reduction of the carbonyl in FAL. Pairing
noble metals with an oxophilic metal has been reported to
create bifunctional materials capable of deoxygenating
biomass-derived compounds.70 The noble metal provides a H2
splitting site while the oxophilic metal binds the O in the
reactant.70 Similarly, it is possible that bimetallic Ru1.0Mo1.0
was more active compared to its monometallic counterparts
due to its multifunctional behavior.
P
Figure 4. (a) FOL selectivity vs. conversion with Ru1.0Mo1.0P at
75°C (blue), 100°C (black), 125°C (green), and (b) linearized
rate data plotted vs. time for Ru1.0Mo1.0P at 75°C (black – R2 =
0.950), 100°C (blue – R2 = 0.981), 112°C (red – R2 = 0.910),
125°C (green – R2 = 0.911) with an Arrhenius plot for
Ru1.0Mo1.0P (inset)
A series of experiments from ambient temperature to 150°C
were performed at 4.2 MPa H2 with Ru1.0Mo1.0P to study the
effects of temperature on the product selectivity (Figure 4).
The catalyst was active as low as 20°C with 4% furfural
conversion observed after 15 hr (not shown). Figure 4 depicts
Compositional variation of bimetallic RuxMo2-xP catalysts
the selectivity and conversion variation for Ru1.0Mo1.0
P
The kinetic studies provided evidence for catalytic
enhancements with bimetallic Ru1.0Mo1.0P . Therefore, the
composition of Ru, Mo, and P was varied in the solid solution
to determine if the bimetallic effect could be further enhanced
and to determine the potential causes for the observations.
Moreover, the change in material composition influences the
electronic properties of the bimetallic phosphides as shown for
FexMo2-xP (0.8 < x <1.5), where the catalytic selectivity and
activity for phenol hydrodeoxygenation were both influenced
between 75°C – 125°C. At 75°C and 100°C, FOL is the dominant
product with >99% selectivity. As the conversion increased to
60%, the selectivity at 100°C dropped to 97% due to the
further reaction of FOL on the catalyst surface to produce MF.
At 125°C the selectivity to FOL dropped from 90% at low
conversion to 84% at 75% conversion. The decrease in
selectivity with the increase in reaction temperature is related
to the effective activation barriers for each of the reactions.
6 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 7 of 14
Catalysis Science & Technology
Please do not adjust margins
Journal Name
ARTICLE
by the bulk and surface compositions.25 Therefore, in an effort production
to control the product selectivity, the effect of material mechanism.62, 63
through
the
Meerwein-Pondorf-Verley
View Article Online
DOI: 10.1039/C9CY00705A
In addition to k1, both k2 and k3 were also determined by using
the same method as described previously (Figure 6a). As
Figure 5. XRD patterns of RuxMo2-xP for (a) x = 0.8 (b) x = 1.0 (c)
x = 1.2 and (d) Ru1.0Mo1.0P reference pattern (PDF 04-015-
7732). (e) The lattice parameters calculated experimentally
(black) and computationally (red).
composition in RuxMo2-xP was explored.
Three compositions of RuxMo2-xP were synthesized (x = 0.8,
1.0, and 1.2) with all of the resulting solid solutions
maintaining an orthorhombic lattice. The ratio was verified
using both ICP-OES and XPS (Table S2). The XRD patterns of
RuxMo2-xP are presented in Figure 5. A Si (111) standard was
added to each of the samples prior to the XRD measurement
to eliminate peak shifts from the instrument. The Si (111) is
marked with an asterisk (*) in Figure 5 and was shifted to
28.44° 2θ. Rietveld refinement was used to calculate the
lattice parameter, a, and they were compared to the
optimized, calculated lattice parameter. Figure 5e represents
similarities between the experimental lattice parameter (black)
and computationally calculated lattice parameter (red), which
are within 1% error. The b and c cell dimensions are reported
in Table S1, which are also in stong agreement.
Figure 6. (a) Psuedo-1st order dependency of FAL with RuxMo2-x
P
where x = 0.8 (blue – R2 = 0.964) and x = 1.0 (black – R2 = 0.981), x =
1.2 (red – R2 = 0.985), and 5% Ru/Al2O3 (grey – R2 = 0.935) and (b)
the ratio of k1/(k2+k3) for bimetallic RuxMo2-xP and Ru/Al2O3
depicted in Scheme 1, the parallel reaction pathway of FOL
hydrogenation can yield both MF and THFA. However, THFA
was not observed with the Mo-rich (x = 0.8) catalyst. Figure 7a-
c shows the concentration profiles for the reactant and
product species observed using RuxMo2-xP where x = 0.8, 1.0,
and 1.2, respectively for both the measured data (circles and
squares) and predicted reaction model using k-values obtained
from Equations 5-7 (lines). The model provided an excellent fit
to the experimental data in the time and concentration range
collected. It is important to note that the FOL hydrogenation
step is significantly slower with all of the catalysts compared to
the first FAL hydrogenation step.
From FAL hydrogenation experiments, the k1 values were
0.078 h-1, 0.176 h-1, and 0.258 h-1 for RuxMo2-xP with x = 0.8,
1.0, and 1.2, respectively (Figure 6a).
The catalytic
performance of commercial Ru/Al2O3 was also analyzed to
provide a comparison between a supported metal catalyst
with the unsupported metal phosphides. The measured
reaction rate constant for Ru/Al2O3 was 0.074 h-1, which was 2-
3 times higher than the monometallic phosphides, but it was
about two times lower than bimetallic Ru1.0Mo1.0P. The FOL
selectivity was not quantified due to the presence of many
side products including 2-(isopropoxy)methyl furan as a result
of the interaction with the solvent (isopropanol). This product
was previously observed using a Ru/C catalyst for FAL
hydrogenation, and it could be an intermediate to FOL
The FOL production rate constant in RuxMo2-xP was examined
based on the ratio between k1 and (k2 + k3). The ratio is plotted
in Figure 6b with respect to x in RuxMo2-xP. Although
Ru1.2Mo0.8P has the highest k1 value, its k2 and k3 were higher
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 7
Please do not adjust margins

Please do not adjust margins
Catalysis Science & Technology
Page 8 of 14
ARTICLE
Journal Name
compared to the other bimetallic phosphides. The ratio facet was chosen as a representative facet for the adsorption
View Article Online
DOI: 10.1039/C9CY00705A
between k1 and k2 + k3 is the highest for Ru1.0Mo1.0P followed calculations. Average partial atomic charges calculated using
by Ru1.2Mo0.8P and Ru0.8Mo1.2P. It is possible that the high FOL Bader analysis on these three catalytic surfaces are presented
selectivity is related to the metal-metal dispersion on the in Table S5. The Lewis acidic nature of Ru1.0Mo1.0P has been
metal surface, which could explain why Ru1.0Mo1.0P showed probed by pyridine adsorption using diffuse reflectance
the highest FOL selectivity.9, 10
infrared Fourier transform spectroscopy (DRIFTS) to show
evidence of the presence of Lewis acid and no strong evidence
of BrØnsted acidity.24 Since Mo is the most electron deficient
atom, Moδ+ is the dominant Lewis acid site in RuxMo2-xP.
Interestingly, the Mo atom in Ru1.0Mo1.0P contains the highest
positive charge in the bulk and on the surface.
Theoretical and experimental surface and adsorption studies of
bimetallic RuxMo2-x
P
XPS was used to probe the surface oxidation on the bimetallic
metal phosphides (Table 2). The peak deconvolution was
conducted according to the methods described in the
experimental section and plotted in Figure S6. In the bimetallic
RuMo phosphides, the Ru 3p3/2 was slightly negative when
compared to Ru0 at 416.6 eV. Meanwhile, the Mo was slightly
positive in comparison with Mo0 (226.8 eV). The binding
energies of the phosphorus atoms were lower than P0 (130.9
eV), suggesting anionic surface P species.68 The XANES K-edge
absorption energy of Ru and Mo are 22,116.96 eV and
20,0001.29 eV, respectively, suggesting an anionic Ru and
The Bader charge for Ru in RuxMo2-xP was negative to show
that Ru is an electron acceptor in the Ru1.0Mo1.0
P system
(Table 2). Although the negative charge on Ru is non-intuitive,
Mulliken charge analysis in Ru1.0Mo1.0P also showed that Ru
can act as an electron acceptor.74 Moreover, similar effects
have been reported in bimetallic Ce-doped Co phosphides,
where the Co becomes negative with Ce addition.75 However,
caution should be applied when comparing Bader charge
calculations and experimentally obtained results due to the
complicated nature of the metallic states and the simplified
oxidized Mo in Ru1.0Mo1.0
P
similar to the XPS result. The
77
system applied with the Bader charge calculation.76,
anionic nature of Ru is in contrast with a previous report
where the Ru in Ru1.0Mo1.0P was reported positive.69 This is
due to the difference in reference point used in the data
analysis, which was verified herein by using a Ru/Al2O3
reference rather than literature values. Overall, these
observation agreed with literature reports on phosphides
where charge sharing is observed in metal phosphides evident
by a slightly positive species (Mo) complemented by partially
negative species (Ru, P).23, 31, 71-73 However, as shown in Table
2, the binding energy differences for the various bimetallic
compositions were too small to draw strong correlations
between these materials. The same observation was reported
in (Ni1-xM’)2P (M = Cr, Fe, Co) where the binding energy shift is
around 0.1 eV resolution for different Ni to M’ ratio.71
Nevertheless, the XPS, XANES, and Bader charge analysis
indicated the same electron flow from Mo to Ru and P in
bimetallic Ru1.0Mo1.0P.
Computationally, the average positive partial charge on the
Mo atoms on the catalytic surfaces followed the order:
Ru1.2Mo0.8P (+0.54 |e|) < Ru0.8Mo1.2P (+0.62 |e|) < Ru1.0Mo1.0
P
(+0.70 |e|), which suggested more average positive charge in
Ru1.0Mo1.0P. This greater charge transfer (from Mo to Ru and P
atoms) could potentially lead to stronger binding with the
electron rich carbonyl O in furfural with Ru1.0Mo1.0P . It is
important to note that the surface charge could also be
correlated with the selectivity, where more positive charge on
Mo leads to higher FOL selectivity.
Computational calculations were performed to determine the
surface and bulk charges in the different RuxMo2-xP catalysts
on the (112) surface. For comparison, additional calculations
were performed on the Ru1.0Mo1.0P (210) facet, which resulted
in nearly identical Bader charges for Ru, Mo, and P (Table S5).
The adsorption energies of furfural on the Ru1.0Mo1.0P (112)
and (210) facets were calculated to be -3.06 eV and -3.08 eV,
respectively. These small differences would not lead to
significantly different adsorption results. Therefore, the (112)
Other reports have shown that the furfural orientation plays
an important role in the reaction selectivity.57, 78-82 The two
binding modes that have been suggested to be most favorable
in furfural adsorption are η1(O) and η2(C-O) (Figure S8).57, 80
The η1(O) surface configuration was suggested from an FTIR
experiment using a Cu catalyst where the O of the carbonyl
interacts with the surface to preferably produce FOL.57
Meanwhile, the η2(C-O) mode was suggested in group VIII
Figure 7. The concentration profiles of FOL (black), MF (red) and THFA (blue) obtained experimentally (circles and
squares) and through model fits (straight line) for RuxMo2-xP where (a) x = 0.8, (b) x = 1.0, (c) x =1.2
8 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 9 of 14
Catalysis Science & Technology
Please do not adjust margins
Journal Name
ARTICLE
catalysts through DFT calculations where both of the C and O furfural (Figure 8a-c) and cis-furfural (Figure 8d-f) on
View Article Online
DOI: 10.1039/C9CY00705A
from the C=O carbonyl of FAL interact with the surface leading Ru0.8Mo1.2P, Ru1.0Mo1.0P, and Ru1.2Mo0.8P were evaluated. The
to the formation of MF and furan.80 The η1(O) surface adsorption energies were found to be ~0.07 – 0.31 eV higher
interaction was studied experimentally with DRIFTS and for trans-furfural, which suggested more favorable adsorption
theoretically with DFT calculations in this work. Meanwhile the of trans-furfural. The result was in an agreement with the
η2(C-O) interaction was not observed,83-85 but it was observed previous theoretical studies.88, 89 The center of mass (COM)
in other work using high resolution electron energy loss was calculated for FAL and RuxMo2-xP (Table S7). Based on the
(HREEL) spectroscopy.86 The adsorption mode was of interest DFT calculation, the COM distance to the surface was
as a potential descriptor for the product selectivity distribution calculated to be 1.73 Å, 1.68 Å, and 3.51 Å for Ru0.8Mo1.2P,
in FAL hydrogenation. Furthermore, others have reported the Ru1.0Mo1.0P, and Ru1.2Mo0.8P respectively. The COM was found
incorporation of Cu into Pd decreases the amount of to be the shortest in Ru1.0Mo1.0P. Additionally, the Mo-O
decarbonylation product due to the shift from η2(C-O) in Pd to distance was calculated to be the shortest on Ru1.0Mo1.0P (1.68
η1(O) in Pd-Cu.87 For the same reason, the incorporation of Mo Å), Ru0.8Mo1.2P (1.80 Å), and Ru1.2Mo0.8P (1.87 Å), respectively,
in RuxMo2-xP may provide a FOL selectivity shift.
which suggested that the carbonyl O would preferably interact
with the more oxophilic metal (i.e. Mo). The calculation also
DRIFTS was therefore performed on the synthesized suggested that Mo-O interaction might be crucial in selective
phosphide catalysts after adsorption of furfural vapor. The FAL hydrogenation to FOL as the shortest Mo-O distance was
DRIFTS experiment was based on previous report that showed found in the most selective catalyst, Ru1.0Mo1.0P.
the existence of η1(O) adsorption mode in IR spectroscopy.57
The catalyst surface was saturated with a stream of N2 that The adsorption energies for trans-furfural were calculated as -
passed through a FAL bubbler at 200°C for 50 mins. The 2.52 eV, -3.06 eV, and -2.12 eV for Ru0.8Mo1.2P, Ru1.0Mo1.0P,
saturation point was indicated by the existence of two peaks and Ru1.2Mo0.8P, respectively, and their corresponding binding
at 1720 cm-1 from gas phase furfural C=O stretching vibration orientation was represented in Figure 8a-c. The adsorption
and 1670 cm-1 from C-O stretching band in the adsorbed FAL energy results were in a similar range with the reported
species as can be seen Figure S9.57
furfural adsorption on supported and unsupported metal
catalysts.9, 79, 88-92 For example, the reported FAL adsorption
The adsorbed C-O vibration mode was observed at a lower energy on Pt(111), Pt(211), and Pt55 surface were -1.36 eV, -
wavenumber from the gas phase FAL due to a weakened C-O 1.95 eV, and -2.52 eV respectively.79 Similarly, the adsorption
bond resulting from the C=O adsorption on the surface energy on Pd(111) was found to be -1.83.89 Interestingly, the
through a η1(O)-type of interaction. This result indicated that reported FAL adsorption energy on Cu(111) was in the range of
the same carbonyl configuration in Ru1.0Mo1.0P was observed -0.05 to -0.17 eV, which was significantly lower.57 This might
with the Cu-catalyst. Interestingly, the adsorbed C-O vibration be due to less interaction between FAL and the surface since
mode can also be observed in the DRIFTS experiments for all the tilted configuration was favored in Cu(111). It can be seen
monometallic MoP, RuP, Ru2P as well as bimetallic Ru0.8Mo1.2
P
from Figure 8a-c, the FAL molecule interacted with the catalyst
and Ru1.2Mo0.8P (Figure S9) yet these catalysts have different surface not only through the carbonyl O but also through the
selectivities for furfural hydrogenation. Therefore, the DRIFTS carbonyl C and furan ring. This type of interaction has been
peak at ~1670 cm-1 could not be correlated with the selectivity suggested to follow the FAL η2(C-O) adsorption mode to
with MoP, RuP, Ru2P, Ru0.8Mo1.2P, Ru1.0Mo1.0P, and produce MF.80 Since the calculation was done in the gas phase,
Ru1.2Mo0.8P.
the DFT result was in agreement with the experimental result
at higher temperature where other hydrogenation products
DFT calculations in the gas phase were performed to were observed. Additionally, high MF selectivity in the gas
benchmark the adsorption energies of FAL on the bimetallic phase reactions have been observed in other works using MoP
RuxMo2-xP to the literature. The adsorption energy of trans- and Ni2P.22, 93
Table 2. Binding energy shift, surface, and bulk Bader charges of RuxMo2-xP on the (112) facet and binding energy shift from XPS.
Binding energy (eV)
Ru Mo
Surface energy (eV)
Ru Mo
Bulk energy (eV)
Ru Mo P
P
P
Ru0.8Mo1.2
Ru1.0Mo1.0
Ru1.2Mo0.8
P
P
P
461.5 228.2 129.0 -0.21 +0.62 -0.41 -0.15 +0.20 -0.19
461.5 228.2 129.1 -0.23 +0.70 -0.46 -0.18 +0.70 -0.18
461.4 228.1 129.2 -0.29 +0.54 -0.25 -0.13 +0.67 -0.11
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 9
Please do not adjust margins

Please do not adjust margins
Catalysis Science & Technology
Page 10 of 14
View Article Online
DOI: 10.1039/C9CY00705A
ARTICLE
and 65% in hexanes. The result showed that solvent selection
can greatly influence the selectivity to FOL. As noted earlier,
background experiments using isopropanol as a H2-donor
resulted in negligible conversion. The solvent effect on the
surface configuration and the elucidation of reaction
mechanisms are ongoing projects in our research groups.
Catalyst recyclability
Catalyst stability is one of the major problems with copper
Figure 9. The effect of solvents towards FAL conversion (filled)
and FOL selectivity (empty). The data were recorded after 3 h.
chromite, as it was reported to decrease by 40% in 4 h.8 The
deactivation has been attributed to carbonaceous formations,
leaching, and sintering.7, 8, 95-98 Therefore, a stability study was
Figure 8. Adsorption of (i) trans- and (ii) cis-furfural on the
(112) facets for (a) Ru0.8Mo1.2P, (b) Ru1.0Mo1.0P, and (c)
Ru1.2Mo0.8P catalyst in the horizontal mode. The atom
colors are purple for Ru, blue for Mo, green for P, grey for
C, silver for H, and red for O.
performed with Ru1.0Mo1.0
P
through recycling experiments.
The stability of Ru1.0Mo1.0
P
was tested with three recycling
experiments. Each of the experiments was performed using
25.0 mg of catalyst for 6 hr at 100°C and 4.2 MPa. The reaction
rate constant, k1, was determined from each of the runs as
well as the FAL conversion and FOL selectivity after 6 hrs. The
results are presented in Figure 10. Based on the recycling
experiments, the conversion at 6 h decreased slightly from
69% to 65% on the second cycle and finally to 61% on the final
cycle. The reaction rate constant also decreased from 0.176 h-1
on the first cycle to 0.171 h-1 and 0.165 h-1 for the second and
third cycle, respectively. Meanwhile the selectivity towards
FOL remained high at >97% for all cycles. The deactivation
seen from each cycle could be due to surface oxidation during
a drying step between the cycles. To overcome this issue, re-
reduction at 650°C for 2 h under 160 mL/min of H2 was
However, the DFT calculation could not represent the
experimental result since the calculation was done in the gas
phase while the reaction was performed in a condensed
phase. The discrepancy could be due to the absence of other
surface species in the DFT calculation such as solvents.
Multiple studies have suggested solvents influenced the FAL
94
conversion and FOL selectivity.16,
Therefore, toluene and
hexanes were used as solvents for FAL hydrogenation at the
same reaction condition of 100°C and 4.2 MPa with 0.10 M
starting concentration. According to the reaction results
(Figure 9), the conversion decreased significantly in toluene
and hexanes, which was consistent with previous
observations. The selectivity towards FOL also decreased conducted on the materials recovered from the third cycle.
significantly from >99% with isopropanol to 88% in toluene
The re-reduced material was then tested at the same reaction
Please do not adjust margins

Page 11 of 14
Catalysis Science & Technology
Please do not adjust margins
Journal Name
ARTICLE
condition by loading the same catalyst weight. In the fourth
cycle, the conversion was recovered back to 71% with a
selectivity of 96%.
This work was supported in part by funding fromVtiehweArNticaletiOonnlinael
Science Foundation through the CAREER award program
DOI: 10.1039/C9CY00705A
(CBET-1351609 and CBET- 1752036), Defense University
Research Instrumentation Program under AFOSR Award No.
FA9550-17-1-0376, Patrick and Jana Eilers Fellowship from ND
Energy, and Center of Environmental Science and Technology
(CEST) Fellowship. We thank CEST for the use ICP-OES and
Material Characterization Facility (MCF) at the University of
Notre Dame for the use of XRD and XPS. The computation
portion of this work used resources at the High Performance
Computing Collaboratory and the National Energy Research
Scientific Computing Center (NERSC), a U.S. Department of
Energy Office of Science User Facility operated under Contract
No. DE-AC02-05CH11231 at the Mississippi State University.
We acknowledge VESTA,[93] CANVAS, and VMD for use of
their software to create images of the crystal structures
contained in this manuscript. The XANES portion of the work
used resources from U.S. Department of Energy (DOE) Office
of Science User Facility operated for the DOE Office of Science
by Argonne National Laboratory under contract No. DE-AC02-
06CH11357. We thank Dr. Ce Yang and Prof. Jeffrey T. Miller at
Purdue University, as well as Dr. Hui Li and Prof. Christopher
Paolucci at University of Notre Dame for assistance at the
beamline.
Figure 10. Conversion (empty) and FOL selectivity (shaded) of
Ru1.0Mo1.0P for FAL hydrogenation at 4.2 MPa and 100°C in
recycling experiment. The dotted line signifies re-reduction.
Conclusions
The hydrogenation of FAL was successfully demonstrated over
a series of monometallic phosphides (MoP, RuP, Ru2P) and
bimetallic phosphides (RuxMo2-xP for x = 0.8, 1.0, and 1.2).
Bimetallic Ru1.0Mo1.0P and monometallic MoP exhibited high
selectivity (99%) towards FOL production in FAL hydrogenation
with minimal production of MF. The catalytic results also
showed that a bimetallic effect was observed through an
increase in the reaction rate of furfural with bimetallic RuxMo2-
xP compared to its monometallic analogues. Kinetic evaluation
determined an activation energy (EA) of 51.1 kJ/mol for
Ru1.0Mo1.0P, which was comparable to Cu-based catalysts
reported in the literature. The ratio between k1/k2+k3 was
found to be the highest with Ru1.0Mo1.0P > Ru1.2Mo0.8P > MoP >
Notes and references
1.
2.
3.
4.
X. D. Li, P. Jia and T. F. Wang, ACS Catal., 2016, 6, 7621-
7640.
J. C. Serrano-Ruiz, R. Luque and A. Sepulveda-Escribano,
Chem. Soc. Rev., 2011, 40, 5266-5281.
R. Mariscal, P. Maireles-Torres, M. Ojeda, I. Sadaba and
M. L. Granados, Energ. Environ. Sci., 2016, 9, 1144-1189.
K. Bauer, J. Panten and H. Surburg, Common fragrance
and flavor materials : preparation, properties and uses,
Weinheim : Wiley-VCH, 2016.
Ru0.8Mo1.2
P
> RuP > Ru2P, which suggested that the
production of FOL per reaction rate was more favored with 5.
Ru1.0Mo1.0P. DFT calculations in the gas phase showed the
R. Rao, A. Dandekar, R. T. K. Baker and M. A. Vannice, J.
Catal., 1997, 171, 406-419.
R. Connor, K. Folkers and H. Adkins, J. Am. Chem. Soc.,
1931, 53, 2012-2012.
H. B. Zhang, Y. Lei, A. J. Kropf, G. H. Zhang, J. W. Elam, J. T.
Miller, F. Sollberger, F. Ribeiro, M. C. Akatay, E. A. Stach, J.
A. Dumesic and C. L. Marshall, J. Catal., 2014, 317, 284-
292.
D. X. Liu, D. Zemlyanov, T. P. Wu, R. J. Lobo-Lapidus, J. A.
Dumesic, J. T. Miller and C. L. Marshall, J. Catal., 2013,
299, 336-345.
B. Seemala, C. M. Cai, R. Kumar, C. E. Wyman and P.
Christopher, ACS Sustain. Chem. Eng., 2018, 6, 2152-2161.
H. Sheng and R. F. Lobo, Chemcatchem, 2016, 8, 3402-
3408.
S. Srivastava, G. C. Jadeja and J. Parikh, Chem. Eng. Res.
Des., 2018, 132, 313-324.
S. Srivastava, N. Solanki, P. Mohanty, K. A. Shah, J. K.
Parikh and A. K. Dalai, Catal. Lett., 2015, 145, 816-823.
C. P. Jimenez-Gomez, J. A. Cecilia, D. Duran-Martin, R.
Moreno-Tost, J. Santamaria-Gonzalez, J. Merida-Robles, R.
6.
most favorable adsorption on Ru1.0Mo1.0P > Ru0.8Mo1.2P >
Ru1.2Mo0.8P which was consistent with the FOL selectivity
trend. Additionally, the use of isopropanol was found to
positively influence the selectivity to FOL, while non-polar
solvents (toluene and hexanes) decreased the conversion and
selectivity. Lastly, we have shown that Ru1.0Mo1.0P can be
recovered to its original rate through re-reduction. Overall,
this study provides strong support for the use of bimetallic
phosphides for furfural hydrogenation to furfuryl alcohol as
well as other selective hydrogenation reactions.
7.
8.
9.
10.
11.
12.
13.
Conflicts of interest
There are no conflicts to declare
Acknowledgements
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 11
Please do not adjust margins

Please do not adjust margins
Catalysis Science & Technology
Page 12 of 14
ARTICLE
Journal Name
Mariscal and P. Maireles-Torres, J. Catal., 2016, 336, 107- 47.
115.
W. Tang, E. Sanville and G. Henkelman, J. Phys.-Condens.
View Article Online
Mat., 2009, 21, 084204.
DOI: 10.1039/C9CY00705A
14.
15.
K. Yan and A. C. Chen, Fuel, 2014, 115, 101-108.
N. Merat, C. Godawa and A. Gaset, J Chem Technol Biot,
1990, 48, 145-159.
A. B. Merlo, V. Vetere, J. F. Ruggera and M. L. Casella,
Catal. Commun., 2009, 10, 1665-1669.
J. J. Musci, A. B. Merlo and M. L. Casella, Catal. Today,
2017, 296, 43-50.
A. O'Driscoll, T. Curtin, W. Y. Hernandez, P. Van Der Voort
48.
49.
50.
51.
E. Sanville, S. D. Kenny, R. Smith and G. Henkelman, J.
Comput. Chem., 2007, 28, 899-908.
C. Stinner, R. Prins and T. Weber, J. Catal., 2000, 191, 438-
444.
V. M. L. Whiffen and K. J. Smith, Energ. Fuel, 2010, 24,
4728-4737.
R. Wang and K. J. Smith, Catal. Lett., 2014, 144, 1594-
1601.
16.
17.
18.
and J. J. Leahy, Org. Process Res. Dev., 2016, 20, 1917- 52.
1929.
V. M. L. Whiffen, K. J. Smith and S. K. Straus, Appl. Catal.
A, 2012, 419, 111-125.
19.
20.
A. Sampath, S. A. Chang and D. W. Flaherty, J. Phys. Chem. 53.
S. T. Oyama, J. Catal., 2003, 216, 343-352.
S. J. Sawhill, K. A. Layman, D. R. Van Wyk, M. H.
Engelhard, C. Wang and M. E. Bussell, J. Catal., 2005, 231,
300-313.
P. D. Vaidya and V. V. Mahajani, Ind. Eng. Chem. Res.,
2003, 42, 3881-3885.
R. S. Rao, R. T. K. Baker and M. A. Vannice, Catal. Lett.,
1999, 60, 51-57.
S. Sitthisa, T. Sooknoi, Y. G. Ma, P. B. Balbuena and D. E.
Resasco, J. Catal., 2011, 277, 1-13.
S. H. Pang and J. W. Medlin, ACS Catal., 2011, 1, 1272-
1283.
R. B. Dean and W. J. Dixon, Anal. Chem., 1951, 23, 636-
638.
S. Jaatinen, J. Touronen, R. Karinen, P. Uusi-Kyyny and V.
Alopaeus, J. Chem. Thermodyn., 2017, 112, 1-6.
P. Xiao, M. A. Sk, L. Thia, X. M. Ge, R. J. Lim, J. Y. Wang, K.
H. Lim and X. Wang, Energ. Environ. Sci., 2014, 7, 2624-
2629.
P. Panagiotopoulou and D. G. Vlachos, Appl. Catal. A,
2014, 480, 17-24.
C, 2018, 122, 23600-23609.
54.
S. E. Habas, F. G. Baddour, D. A. Ruddy, C. P. Nash, J.
Wang, M. Pan, J. E. Hensley and J. A. Schaidle, Chem.
Mater., 2015, 27, 7580-7592.
R. H. Bowker, M. C. Smith, M. L. Pease, K. M. Slenkamp, L.
Kovarik and M. E. Bussell, ACS Catal., 2011, 1, 917-922.
C. P. Jiménez-Gómez, J. A. Cecilia, R. Moreno-Tost and P.
Maireles-Torres, Chemcatchem, 2017, 9, 2881-2889.
Y. Bonita and J. C. Hicks, J. Phys. Chem. C, 2018, 122,
13322-13332.
Y. Bonita, T. P. O'Connell, H. E. Miller and J. C. Hicks, Ind.
Eng. Chem. Res., 2019, 58, 3650-3658.
D. J. Rensel, J. Kim, V. Jain, Y. Bonita, N. Rai and J. C. Hicks,
Catal. Sci. Technol., 2017, 7, 1857-1867.
55.
56.
57.
58.
59.
60.
61.
21.
22.
23.
24.
25.
26.
27.
28.
D. J. Rensel, J. Kim, Y. Bonita and J. C. Hicks, Appl. Catal. A,
2016, 524, 85-93.
D. J. Rensel, S. Rouvimov, M. E. Gin and J. C. Hicks, J.
Catal., 2013, 305, 256-263.
G. Corro and R. Gomez, React. Kinet. Catal. Lett., 1979, 12, 62.
145-150.
29.
30.
31.
S. Doniach and M. Sunjic, J. Phys. C, 1970, 3, 285-&.
D. J. Morgan, Surf. Interface Anal., 2015, 47, 1072-1079.
P. E. R. Blanchard, A. P. Grosvenor, R. G. Cavell and A. 64.
63.
M. Koehle and R. F. Lobo, Catal. Sci. Technol., 2016, 6,
3018-3026.
J. Luo, J. Yu, R. J. Gorte, E. Mahmoud, D. G. Vlachos and
M. A. Smith, Catal. Sci. Technol., 2014, 4, 3074-3081.
M. A. Tike and V. V. Mahajani, Ind. Eng. Chem. Res., 2007,
46, 3275-3282.
Y. Song, W. Li, M. Zhang and K. Tao, Front. Chem. Sci. Eng.,
2007, 1, 151-154.
S. Carenco, D. Portehault, C. Boissiere, N. Mezailles and C.
Sanchez, Chem. Rev., 2013, 113, 7981-8065.
M. Pelavin, D. N. Hendrickson, J. M. Hollander and W. L.
Jolly, J. Phys. Chem., 1970, 74.
V. Jain, Y. Bonita, A. Brown, A. Taconi, J. C. Hicks and N.
Rai, Catal. Sci. Technol., 2018, 8, 4083-4096.
A. M. Robinson, J. E. Hensley and J. W. Medlin, ACS Catal.,
2016, 6, 5026-5043.
P. E. R. Blanchard, A. P. Grosvenor, R. G. Cavell and A.
Mar, J. Mater. Chem., 2009, 19, 6015-6022.
A. P. Grosvenor, R. G. Cavell and A. Mar, J. Solid State.
Chem., 2007, 180, 2702-2712.
Mar, Chem. Mater., 2008, 20, 7081-7088.
32.
33.
34.
35.
A. D. Becke, J. Chem. Phys., 2014, 140.
R. G. Parr, Horizons of Quantum Chemistry, 1980, 5-15.
G. Kresse and J. Hafner, Phys. Rev. B., 1993, 47, 558-561.
G. Kresse and J. Hafner, Phys. Rev. B., 1994, 49, 14251-
14269.
G. Kresse and J. Furthmüller, Phys. Rev. B., 1996, 54,
11169-11186.
G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6,
15-50.
J. Klimeš, D. R. Bowler and A. Michaelides, Phys. Rev. B.,
2011, 83.
J. Klimeš, D. R. Bowler and A. Michaelides, J. Phys.-
Condens. Mat., 2009, 22, 022201.
P. E. Blöchl, Phys. Rev. B., 1994, 50, 17953-17979.
G. Kresse and D. Joubert, Phys. Rev. B., 1999, 59, 1758- 72.
1775.
65.
66.
67.
68.
69.
70.
71.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
P. J. Stephens, F. J. Devlin, C. F. Chabalowski and M. J. 73.
Frisch, J. Phys. Chem., 1994, 98, 11623-11627.
M. Methfessel and A. T. Paxton, Phys. Rev. B., 1989, 40, 74.
3616-3621.
A. Genova and M. Pavanello, J. Phys.-Condens. Mat., 75.
2015, 27.
A. P. Grosvenor, S. D. Wik, R. G. Cavell and A. Mar, Inorg.
Chem., 2005, 44, 8988-8998.
W. Wong-Ng, W. Y. Ching, Y.-N. Xu, J. A. Kaduk, I. Shirotani
and L. Swartzendruber, Phys. Rev. B., 2003, 67, 144523.
W. Gao, M. Yan, H.-Y. Cheung, Z. Xia, X. Zhou, Y. Qin, C.-Y.
Wong, J. C. Ho, C.-R. Chang and Y. Qu, Nano Energy, 2017,
38, 290-296.
G. Henkelman, A. Arnaldsson and H. Jónsson, Comput.
Mater. Sci., 2006, 36, 354-360.
76.
A. Walsh, A. A. Sokol, J. Buckeridge, D. O. Scanlon and C.
R. A. Catlow, J. Phys. Chem. Lett., 2017, 8, 2074-2075.
M. Yu and D. R. Trinkle, J. Chem. Phys., 2011, 134, 064111.
12 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 13 of 14
Catalysis Science & Technology
Please do not adjust margins
Journal Name
ARTICLE
77.
78.
79.
80.
81.
82.
83.
A. Walsh, A. A. Sokol, J. Buckeridge, D. O. Scanlon and C.
View Article Online
DOI: 10.1039/C9CY00705A
R. A. Catlow, Nat. Mater., 2018, 17, 958-964.
S. H. Pang, C. A. Schoenbaum, D. K. Schwartz and J. W.
Medlin, Nat. Commun., 2013, 4.
Q. X. Cai, J. G. Wang, Y. G. Wang and D. H. Mei, Aiche J.,
2015, 61, 3812-3824.
S. Sitthisa, W. An and D. E. Resasco, J. Catal., 2011, 284,
90-101.
S. Sitthisa and D. E. Resasco, Catal. Lett., 2011, 141, 784-
791.
M.-d. Qian, J.-l. Xue, S.-j. Xia, Z.-m. Ni, J.-h. Jiang and Y.-y.
Cao, J. Fuel Chem. Technol., 2017, 45, 34-42.
R. G. Greenler, D. R. Snider, D. Witt and R. S. Sorbello,
Surf. Sci., 1982, 118, 415-428.
84.
85.
R. G. Greenler, J. Chem. Phys., 1966, 44, 310-315.
H. A. Pearce and N. Sheppard, Surf. Sci., 1976, 59, 205-
217.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
J. L. Davis and M. A. Barteau, Surf. Sci., 1990, 235, 235-
248.
S. Sitthisa, T. Pham, T. Prasomsri, T. Sooknoi, R. G.
Mallinson and D. E. Resasco, J. Catal., 2011, 280, 17-27.
Y. Shi, Y. L. Zhu, Y. Yang, Y. W. Li and H. J. Jiao, ACS Catal.,
2015, 5, 4020-4032.
B. Liu, L. Cheng, L. Curtiss and J. Greeley, Surf. Sci., 2014,
622, 51-59.
S. Wang, V. Vorotnikov, J. E. Sutton and D. G. Vlachos, ACS
Catal., 2014, 4, 604-612.
V. Vorotnikov, G. Mpourmpakis and D. G. Vlachos, ACS
Catal., 2012, 2, 2496-2504.
W. Wan, G. R. Jenness, K. Xiong, D. G. Vlachos and J. G.
Chen, Chemcatchem, 2017, 9, 1701-1707.
Y. Z. Wang, F. Liu, H. Han, L. F. Xiao and W. Wu,
Chemistryselect, 2018, 3, 7926-7933.
X. Chen, L. Zhang, B. Zhang, X. Guo and X. Mu, Sci. Rep.,
2016, 6, 28558.
B. J. O'Neill, D. H. K. Jackson, A. J. Crisci, C. A. Farberow, F.
Y. Shi, A. C. Alba-Rubio, J. L. Lu, P. J. Dietrich, X. K. Gu, C. L.
Marshall, P. C. Stair, J. W. Elam, J. T. Miller, F. H. Ribeiro,
P. M. Voyles, J. Greeley, M. Mavrikakis, S. L. Scott, T. F.
Kuech and J. A. Dumesic, Angew. Chem. Int. Edit., 2013,
52, 13808-13812.
96.
97.
98.
M. V. Twigg and M. S. Spencer, Appl. Catal. A, 2001, 212,
161-174.
J. G. M. Bremner and R. K. F. Keeys, J. Chem. Soc., 1947,
DOI: DOI 10.1039/jr9470001068, 1068-1080.
G. Seo and H. Chon, J. Catal., 1981, 67, 424-429.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 13
Please do not adjust margins

Catalysis Science & Technology
Page 14 of 14
View Article Online
DOI: 10.1039/C9CY00705A
70x45mm (149 x 149 DPI)