Mechanisms for High Selectivity in the Hydrodeoxygenation of 5-Hydroxymethylfurfural over PtCo Nanocrystals

Article abstract of DOI:10.1021/acscatal.6b00750

Carbon-supported, Pt and PtCo nanocrystals (NCs) with controlled size and composition were synthesized and examined for hydrodeoxygenation (HDO) of 5-hydroxymethylfurfural (HMF). Experiments in a continuous flow reactor with 1-propanol solvent, at 120 to 160 °C and 33 bar H₂, demonstrated that reaction is sequential on both Pt and PtCo alloys, with 2,5-dimethylfuran (DMF) formed as an intermediate product. However, the reaction of DMF is greatly suppressed on the alloys, such that a Pt₃Co₂ catalyst achieved DMF yields as high as 98%. XRD and XAS data indicate that the Pt₃Co₂ catalyst consists of a Pt-rich core and a Co oxide surface monolayer whose structure differs substantially from that of bulk Co oxide. Density functional theory (DFT) calculations reveal that the oxide monolayer interacts weakly with the furan ring to prevent side reactions, including overhydrogenation and ring opening, while providing sites for effective HDO to the desired product, DMF. We demonstrate that control over metal nanoparticle size and composition, along with operating conditions, is crucial to achieving good performance and stability. Implications of this mechanism for other reactions and catalysts are discussed.

Full text of DOI:10.1021/acscatal.6b00750

Subscriber access provided by ORTA DOGU TEKNIK UNIVERSITESI KUTUPHANESI
Article
Mechanisms for High Selectivity in Hydrodeoxygenation
of 5-Hydroxymethylfurfural over PtCo Nanocrystals
Jing Luo, Hongseok Yun, Alexander V. Mironenko, Konstantinos A. Goulas,
Jennifer D. Lee, Matteo Monai, Cong Wang, Vassili Vorotnikov, Christopher
B. Murray, Dionisios G. Vlachos, Paolo Fornasiero, and Raymond J. Gorte
ACS Catal., Just Accepted Manuscript • Publication Date (Web): 20 May 2016
Downloaded from http://pubs.acs.org on May 20, 2016
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
ACS Catalysis is published by the American Chemical Society. 1155 Sixteenth Street
N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
Mechanisms for High Selectivity in Hydrodeoxygenation of 5ꢀ
Hydroxymethylfurfural over PtCo Nanocrystals
a,†
b,†
c,†
c
Jing Luo , Hongseok Yun , Alexander V. Mironenko , Konstantinos Goulas , Jennifer D.
b
d
a
c
b
Lee , Matteo Monai , Cong Wang , Vassili Vorotnikov , Christopher B. Murray , Dionisios G.
c
d
a,*
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Vlachos , Paolo Fornasiero and Raymond J. Gorte
a
Department of Chemical & Biomolecular Engineering, University of Pennsylvania
Philadelphia, PA 19104, United States
b
Department of Chemistry, University of Pennsylvania
Philadelphia, PA 19104, United States
Department of Chemical & Biomolecular Engineering, University of Delaware
c
Newark, DE 19716, United States
Department of Chemical and Pharmaceutical Sciences, University of Trieste
d
Trieste, 34127, Italy
Abstract
Carbonꢀsupported, Pt and PtCo nanocrystals (NCs) with controlled size and composition
were synthesized and examined for hydrodeoxygenation (HDO) of 5ꢀhydroxymethylfurfural
(HMF). Experiments in a continuous flow reactor with 1ꢀpropanol solvent, at 120 to 160°C and
3
3 bar H , demonstrated that reaction is sequential on both Pt and PtCo alloys, with 2,5ꢀ
2
dimethylfuran (DMF) formed as an intermediate product. However, the reaction of DMF is
greatly suppressed on the alloys, such that a Pt Co catalyst achieved DMF yields as high as 98%.
3
2
XRD and XAS data indicate that the Pt Co catalyst consists of a Ptꢀrich core and a Co oxide
3
2
surface monolayer whose structure differs substantially from that of bulk Co oxide. Density
functional theory (DFT) calculations reveal that the oxide monolayer interacts weakly with the
furan ring to prevent side reactions, including overꢀhydrogenation and ring opening, while
providing sites for effective HDO to the desired product, DMF. We demonstrate that control over
metal nanoparticle size and composition, along with operating conditions, is crucial to achieving
good performance and stability. Implications of this mechanism for other reactions and catalysts
are discussed.
Keywords: 5ꢀhydroxymethylfurfural, hydrodeoxygenation, 2,5ꢀdimethyl furan, PtCo nanocrystal,
bimetallic catalyst
†
*
These authors contribute equally.
Corresponding Author: Raymond J. Gorte
Email: gorte@seas.upenn.edu
ACS Paragon Plus Environment

ACS Catalysis
Page 2 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
INTRODUCTION
ꢀHydroxymethylfurfural (HMF) is a key platform chemical in biomass conversion. It
is readily obtained by the acidꢀcatalyzed dehydration of Cꢀ6 sugars (e.g, fructose, but its high
1
,2
5
3
–5
degree of functionality precludes its direct use as a fuel. One promising approach to stabilize
HMF is the selective hydrodeoxygenation to form 2,5ꢀdimethylfuran (DMF). A relatively small
amount of hydrogen is consumed in the formation of DMF and DMF can be used directly as fuel
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
additive due to its high energy density and an octane rating of 119. DMF can also be converted
7
to pꢀxylene via a DielsꢀAlder reaction with ethylene.
The reaction of HMF to DMF has been studied extensively over various metal and metalꢀ
8–10
alloy catalysts
but selectivity to DMF over ringꢀopened (e.g. 2ꢀhexanone, 2,5ꢀhexanedione)
and ringꢀhydrogenated (e.g. 2,5ꢀdimethyl tetrahydrofuran) products is often poor. Understanding
what is required for a good catalyst is made more difficult by the large variances in yields that
1
1,12
have been reported by different groups for materials of similar composition.
Some of these
variances appear to be due the sensitivity of results to the type of reactor that is used in the rate
1
3
measurements. Even so, bimetallic catalysts appear to be more selective than their pureꢀmetal
1
4–16
17
analogs.
For example, Schüth and coꢀworkers reported DMF yields as high as 98% on Ptꢀ
Co nanoparticles which were encapsulated in hollow carbon spheres.
Some of us have recently reported that the reaction of HMF to DMF is sequential on Pt
18
and many other metals , with HMF reacting selectively to DMF, but DMF then going on to
form secondary products over the same catalysts. Based on these mechanistic considerations, the
yield of DMF is determined by the relative rates of DMF formation and DMF consumption and
the most selective catalysts should show low activity for reaction of DMF. An implication of this
picture is that selective catalysts cannot consist of a mixture of selective and nonselective
components, since even a relatively small fraction of nonselective material could convert DMF
1
8
to other products. For the case in which two metals are nonselective (e.g. Pt and Co) but their
alloy is selective, this means that the composition of the metal particles must be uniform.
The synthesis of highly uniform Pt and PtCo nanocrystals (NCs) have demonstrated via
19,20
solvothermal methods.
The solvothermal method can readily control the size, morphology,
and composition of NCs and these parameters are strongly correlated to their catalytic
2
0,21
properties.
For example, Huang et al. reported that small amounts of molybdenum doping
into Pt Ni NCs can enhance its specific activity for the oxygen reduction reaction by 81 fold
3
ACS Paragon Plus Environment

Page 3 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
22
compared with the activity of a commercial Pt/C catalyst. Therefore, this synthetic approach
enables us to prepare and test catalysts whose properties are well controlled, allowing us to
understand the effect of various parameters on catalytic activities.
Here, we report the HDO of HMF into DMF catalyzed by Pt, Pt Co, and Pt Co NCs
3
3
2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
using a continuous flow reactor. DMF yields as high as 98% were achieved with Pt Co alloys
3
2
because this alloy shows a very low reactivity towards DMF. Furthermore, the bimetallic
catalysts synthesized using the solvothermal method are stable and superior to alloy catalysts
prepared by traditional wet impregnation. Characterization of the PtꢀCo NCs reveals they have a
special structure consisting of a monolayer of surface oxide on a metallic core. Calculations
using Density functional theory (DFT) rationalize the stability of this structure and indicate that
the oxide prevents side reactions while providing catalytic sites for effective conversion of HMF
to DMF. Computations reveal a radicalꢀmediated reaction mechanism, which some of us have
2
3
found to be critical for selective HDO on oxides. Structural characterization and DFT
calculations confirm that controlling the bimetallic composition is essential for preparing a good
catalyst.
METHODS
Catalyst synthesis and characterization
Nearly monodisperse Pt, Pt Co, and Pt Co NCs were synthesized by using or modifying
3
3
2
1
9,20
the reported methods.
(See Supporting Information Section S1 for further details) To prepare
Pt NCs, platinum (II) acetylacetonate (acac) dissolved in 20 mL of benzyl ether was decomposed
at 300 °C in the presence of oleic acid, oleylamine, and trioctylphosphine under nitrogen. For
platinum cobalt bimetallic NC synthesis, the hot injection technique of dicobalt octacarbonyl
(
Co (CO) ) is employed. In the synthesis of Pt Co, the reaction mixture of Pt(acac) , 1ꢀ
2 8 3 2
adamantane carboxylic acid, 1,2ꢀhexadecanediol (HDD), and hexadecylamine were heated under
nitrogen and Co (CO) was injected into the reaction mixture at 170 °C. The reaction mixture is
2
8
heated up to 230 °C. After 40 mins, the reaction mixture was cooled down to room temperature.
24
In this case, HDD was used to reduce Pt, which is called “polyol process” . On the other hand,
for the synthesis of Pt Co , hexadecylamine was replaced by oleylamine and HDD was not
3
2
added. After the injection of Co (CO) at 170 °C, the reaction vessel was heated to 300 °C and
2
8
2
5
kept for 30 mins. As oleylamine is a relatively milder reducing agent than HDD, this reaction
ACS Paragon Plus Environment

ACS Catalysis
Page 4 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
resulted in relatively cobalt rich NCs compared to the reaction done by the polyol process. From
the Transmission Electron Microscope (TEM) images of NCs in Figure 1, the average diameters
of NCs were 2.4 nm, 3.2 nm, and 3.7 nm with less than 8% of size distribution for Pt, Pt Co, and
3
Pt Co NCs, respectively. A superlattice structure for each composition of NCs was observed,
3
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
indicating the high monodispersity of these NCs. Wide angle xꢀray scattering data show that all
the NCs possess faceꢀcentered cubic (fcc) crystal structure as displayed in Figure 1d. The
compositions of NCs were confirmed by Inductively Coupled Plasma Optical Emission
Spectrometry (ICPꢀOES).
The NCs are dispersed into carbon support to prepare 10 weight % of metals on carbon.
(Supplementary information for further details about the preparation procedure) As the colloidal
NCs are covered by organic ligands, such as oleic acid and oleylamine, for colloidal stability, the
catalytic reactions are inhibited due to the limited access of reactants to the binding sites on the
NC surface. Therefore, the surface ligands should be removed to maximize their available
surface area. For this purpose, the catalysts were gone through the two steps of surface treatment.
First, the 300 mg of 10 wt % metal/C catalysts were exposed to O plasma for 15 mins. After the
2
2
6
plasma treatment, the NC surface was further cleaned by rapid thermal annealing (RTA).
Figure S1 shows TEM images of asꢀdeposited and surface ligand removed NCs on carbon
support. After the plasma and thermal treatment, no significant change in size and morphology is
observed in all the cases of NCs, which proves the reliability of our surface treatment technique.
Also, from wide angle xꢀray scattering study (Figure S2), it is confirmed that the phase of Pt Co
3
and Pt Co is not transformed from disordered fcc to ordered fct structure. Therefore, we can
3
2
exclude the potential effects of crystal structure ordering on the catalytic activity and attribute
27
any difference to the compositional effect.
The metal dispersions were determined volumetrically using CO adsorption uptakes at
room temperature on Al O supported catalysts. CO chemisorption experiments were performed
2
3
on alumina supported NCs in order to avoid sorption problems related to active carbon
28
o
supports. 0.1 g Catalyst was loaded into a sample cell, evacuated at 400 C and reduced for 30
o
o
minutes at 250 C or 400 C in 200 Torr of H . Finally, the sample was evacuated and cooled to
2
room temperature. CO uptakes were then determined by adding small pulses of CO until a rise in
the pressure of the sample cell was detected. In determining the NCs dispersion, we assumed an
2
9
adsorption stoichiometry of one CO molecule per surface Pt. Chemisorption of CO on Co was
ACS Paragon Plus Environment

Page 5 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
not included in the calculation of the dispersion because control experiments on Co/Al O3
2
catalysts prepared by impregnation method reduced at 400°C did not show any CO adsorption.
Liquid-phase HDO with H in a flow reactor
2
The threeꢀphase reactions were carried out in a continuous flow reactor that has been
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
13,18
described in detail elsewhere.
The reactor was a stainlessꢀsteel tube, 20ꢀcm long, with a 4.6ꢀ
mm ID. In most cases, 0.05 g of catalyst was packed into the middle portion of the tube and held
in place by glass wool; however, conversions and selectivities were dependent only on reactor
space time, independent of reactor catalyst loading. An inert glass tube was placed in the
stainlessꢀsteel tube, downstream from the catalyst, in order to reduce the empty volume of the
reactor and to prevent the catalyst bed from being pushed out of the heated zone by the reactant
flow. The liquid feed, either a mixture of 1.0 g HMF (99%, SigmaꢀAldrich) or 0.76 g DMF (99%,
SigmaꢀAldrich) in 100 mL of 1ꢀpropanol (Fisher Scientific), was introduced into the reactor by
an HPLC pump (Series I+, Scientific Systems Inc.), which was also used to measure the total
pressure in the reactor. (The reaction of DMF was performed with 0.76 g of DMF in 100 ml of 1ꢀ
propanol to maintain the same molar concentration as that used in the HMF rate measurements.)
The H (UHP grade, Airgas) flow to the reactor was controlled by varying the pressure drop
2
across 8 feet of 0.002ꢀinch ID capillary tubing (Valco Instruments, Inc.). A bubble meter at the
reactor exit was used periodically to check the H flow rates. The liquid flow rates could be
2
varied from 0.02 to 0.2 ml/min, while the H flow rates were 2 to 20 mL/min (STP). In order to
2
vary the space velocity in the reactor at fixed experimental conditions, the ratio of liquid and gas
flow rates was kept constant. The pressure within the reactor was controlled by a backꢀpressure
regulator (KPB series, Swagelok) placed at the reactor exit.
Prior to rate measurements, the catalyst was pretreated at 250ºC in 1 bar of flowing H for
2
3
0 min. Because the products left the reactor at room temperature, product analysis was carried
out using a syringe to inject the liquid effluent into a GCꢀMS (QPꢀ5000, Shimadzu), equipped
with a capillary column (HPꢀInnowax, Agilent Technologies). The gasꢀphase products were also
examined but, under the conditions of these experiments, were found to consist of only H and
2
solvent vapor. Product selectivities were quantified using standard solutions with known
concentrations of HMF, DMF, DMTHF, 2ꢀhexanone, 2ꢀhexanol, and 2,5ꢀhexandione (all
purchased from Sigma Aldrich). For quantification of other furanꢀbased, intermediate products,
the GC sensitivity was assumed to be the same as that for HMF. For openꢀring, ether products,
ACS Paragon Plus Environment

ACS Catalysis
Page 6 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the GC sensitivity was assumed to be the same as 2ꢀhexanone or 2,5ꢀhexandione. The carbon
balance from GC analysis was always better than 95%.
The typical time for an experiment was 4 h. For each sample injection, the GC analysis
required 30 min. Most of the data reported here were chosen from the second or third
measurement (40 to 60 min after starting the reaction). For the conditions of this study, no
reaction was observed in the absence of a catalyst in the flow reactor.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
X-ray Absorption Measurements
Xꢀray absorption measurements were performed at the beamline 5BMꢀD at the Advanced
Photon Source, Argonne National Laboratory. Catalyst samples were diluted with boron nitride
o
and pressed into pellets in a sixꢀwell sample holder. The catalysts were reduced at 250 C (preꢀ
o
treatment temperature) and 400 C (higher than the pretreatment temperature) for 1 h (ramp rate
o
ꢀ1
ꢀ1
5 C min ) under atmospheric pressure in a 40 ml min hydrogen flow prior to measurement at
ambient temperature under hydrogen. Data processing was done using the program Athena of the
Demeter suite. Fitting of the EXAFS oscillations was done using the Artemis program of the
same suite. For the Pt L3 edge, oscillations were fitted using the Fourier transform (FT) from
ꢀ1
ꢀ1
wavenumbers k = 3 Å to k = 12 Å and R = 1.6 Å to 3.3 Å while for the Co K edge,
ꢀ1
ꢀ1
oscillations were fitted using a FT from wavenumbers k = 3 Å to k = 12 Å , and R = 1.4 Å to R
3.2 Å. Amplitude reduction factors were calculated using the EXAFS spectrum of the Pt or Co
=
foils and assuming a coordination number (CN) equal to 12. A single DebyeꢀWaller factor was
used for all scattering paths.
Density functional theory (DFT) calculations
Spinꢀpolarized density functional theory calculations have been carried out under the
30–32
generalized gradient approximation using VASP software
. KohnꢀSham eigenstates have
been expanded in a plane wave basis set with a kinetic energy cutoff of 400 eV. Sampling of the
3
3
first Brillouin zone has been carried out according to the MonkhurstꢀPack 3x3x1 kꢀpoint mesh.
The initial magnetic moment of Co atoms has been set to 2.0 Bohrꢀmagnetons. Exchange,
34,35
correlation, and dispersion effects have been approximated with a PBEꢀD3
functional.
Threshold value for maximum atomic forces has been set to 0.05 eV/Å. Transition states have
3
6–39
been identified via a climbingꢀimage nudged elastic band (CINEB) and/or a dimer
method
with the forces’ tolerance set to 0.1 eV/Å. A honeycomb Co O on a Ptꢀrich metallic core has
3
2
been modeled as a Co O /Pt(111) surface using a (4x4) supercell with a honeycomb Co O
2
3
2
3
ACS Paragon Plus Environment

Page 7 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
structure placed on top of three Pt layers (bottom two were fixed).We used a finite difference
method to calculate selected vibrational frequencies. Throughout the paper, DFT energies in a
vacuum are reported, unless stated otherwise.
RESULTS
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
HDO Studies
Previous studies on carbonꢀsupported Pt and Co catalysts have shown that
1
8
hydrodeoxygenation of HMF to DMF is a series reaction , proceeding as indicated in Scheme 1.
The HMF (A) first reacts to a group of partially hydrogenated intermediate compounds (B),
including 2ꢀpropoxymethylꢀ5ꢀfuranmethanol (etherꢀfurfuryl alcohol, or EFA), 2ꢀpropoxymethylꢀ
5
ꢀmethylfuran (etherꢀmethyl furan, or EMF), 5ꢀmethyl furfural (MF), 2ꢀhydroxylmethylꢀ5ꢀ
methyl furan (HMMF), 2,5ꢀbis(hydroxymethyl)furan (BHMF), and 2,5ꢀbis(propoxymethyl)furan
BEF). These intermediate compounds can all be converted to DMF (C), which in turn reacts to
overꢀhydrogenated products (D), dimethyl tetrahydrofuran (DMTHF), 2ꢀhexanone, 2ꢀhexanol,
,5ꢀhexanedione, and their etherification derivatives, 1ꢀpropoxyꢀ1ꢀmethylꢀpentane (2ꢀ
(
2
propoxyhexane) and 1,4ꢀdipropoxyꢀ1,4ꢀdimethylꢀbutane (2,5ꢀdipropoxyhexane).
Figure 2a, which is a plot of the conversion and product yield as a function of space time
on the 10ꢀwt% Pt/C catalyst prepared from Pt NCs, shows that Scheme 1 also applies to reaction
of HMF in this case. These experiments were carried out at 120°C, a significantly lower
temperature than that used in a previous study with Pt catalysts prepared by infiltration of metal
salts, because the NCs catalysts were much more active. Even for the shortest space time and at
this low temperature, the HMF conversion was greater than 65%. Initially, partially
hydrogenated products, B, were formed in the highest yields, but these declined steadily with
increasing space time. DMF yield initially increased, then decreased, providing strong evidence
that DMF is an intermediate product in a series reaction. The maximum yield was approximately
41%. Overꢀhydrogenated compounds, D, only formed at higher space times, indicating that they
are not primary products. The fact that their formation follows the consumption of DMF strongly
suggests they are formed from DMF. A detailed analysis of the partially hydrogenated (B) and
overꢀhydrogenated (D) compounds is given in the Supplemental Information, Table S1 and
1
8
Figure S3a. The HDO of HMF over Co/C has been reported in a previous study . The products
formed as a function of space time were similar to that formed over Pt catalysts. Co itself is not
selective to DMF, due to the formation of the overꢀhydrogenated product, 2,5ꢀhexanedione, at
ACS Paragon Plus Environment

ACS Catalysis
Page 8 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
high space times.Similar experiments were performed on the 10 wtꢀ% PtꢀCo, NCs catalysts, with
Pt:Co ratios of 3:1 and 3:2. Data at 120°C and 33 bar are shown in Figure 2b and 2c. For a given
space time, the HMF conversions over the Pt Co in Figure 2b were slightly lower than that
3
obtained on the pure Pt NCs; but the initial products were the same partially hydrogenated
compounds, B (see Table S1), with these again being converted to DMF at a similar rate.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
However, on the Pt Co catalyst, the DMF yield continued to increase, to a value of 75%; and
3
only relatively small quantities of overꢀhydrogenated compounds, D, were formed at the largest
space times. The activity of the Pt Co catalyst was noticeably lower than that of the other two
3
2
samples, and 100% conversion of HMF was achieved only at the highest space time. Because of
the lower activity, the DMF yield was still increasing at the highest space time and the
production of Dꢀgroup compounds was negligible.
Due to the lower rates on the PtꢀCo catalysts, additional reaction measurements were
performed at 160°C and 33 bar in order determine the evolution of products, with results shown
in Figure 3. As shown in Figure 3a, the HMF conversion was nearly 90% on the Pt Co sample,
3
even at the lowest space time. The Bꢀproducts again decline steadily with time but DMF yield
went through a maximum of about 75% at this temperature, with overꢀhydrogenated products
being produced from the DMF. However, results for the Pt Co sample in Figure 3b show the
3
2
DMF yields continuing to increase, up to 98%.
The Pt Co sample was also very stable compared to the Pt catalyst. Figure 4 shows the
3
2
HMF conversion and DMF yield for the two catalysts as a function of time at 160°C and a space
of 1.0 g·min/mL. The PtꢀCo catalysts has no observable deactivation or change in selectivity for
a period of at least 14 h. By contrast, the Pt/C deactivated rapidly under these same conditions. It
should be noted that Pt/C catalyst was highly active under these conditions, so that the low initial
yield is due to the “overꢀreaction” of DMF to overꢀhydrogenated (D) compounds. The increasing
yield with short times results from the lower catalyst activity. In addition to the stability against
4
0
coking, the alloy catalyst is also more thermally stable. Figure S1 has shows TEM images for
Pt and PtCo catalysts after 5 hours reaction at 160°C. The average size of Pt NCs increased and
some aggregates formed, while the size and shape of Pt Co NCs remained unchanged. In Figure
3
2
S4 and Table S2, statistical particle size analysis is presented. Pt NCs underwent morphological
changes, with the average particle size increasing from 2.4 nm to 3.5 nm and a significant size
distribution change from 7.3 % to 19 %. By contrast, Pt Co NCs showed almost no
3
2
ACS Paragon Plus Environment

Page 9 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
morphological change. The average size of the particles were unaffected and the size distribution
increased only slightly from 7.3 % to 10 %.
The results in Figure 2 and Figure 3 indicate that Co alloying with Pt has a modest effect
on HDO rates for HMF but strongly suppresses reactions of DMF. To investigate this in more
detail, we examined the reaction of DMF on the same three catalysts, with conversions and
product distributions shown in Figure 5. The reactions were again carried out at 33 bar and either
120°C (for Pt) or 160°C (for the PtꢀCo alloys), using 1ꢀpropanol solutions with the same DMF
molarity as that used in the HMF experiments. Because water is formed in the reaction of HMF,
these experiments do not perfectly mimic the sequential reaction found in HMF; however, some
water was also formed in the DMF reaction measurements by dehydration of the alcohols to
form diꢀpropyl ether.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
As shown in Figure 5a, DMF is converted rapidly on Pt/C, even at 120°C. The main
products are the openꢀring ketones and ethers, the products that were also formed at high space
times for the reaction of HMF (see Figure S3 in SI). The reaction of DMF on the PtꢀCo catalysts
was carried out at 160°C because of their lower activities. Even at this higher temperature, DMF
conversions on the Pt Co sample were lower than that observed on the Pt catalyst, although still
3
significant. The products on the Pt Co catalyst were essentially the same as that observed on Pt.
3
However, the conversion of DMF on the Pt Co sample at 160°C, shown in Figure 5c, was very
3
2
low for all space times, reaching a value of only 10% at a space time of 1.0 g·min/mL.
Characterization
In order to understand the role of Co for high HMFꢀtoꢀDMF selectivity and elucidate the
nature of the active site, a combination of microscopic, spectroscopic, and computational tools
were employed, as discussed below. First, the compositions of all the NCs samples were checked
by Inductively Coupled Plasma/Mass Spectroscopy (ICPꢀMS) and found to agree with the
quoted compositions. Second, the NCs were of uniform size, with an FCC structure, as shown by
the TEM and wide angle xꢀray scattering results in Figure 1. The uniformity of the particle sizes
is demonstrated by the particles packing in regular hexagonal lattices. Wide angle xꢀray
scattering data (Figure 1d) show that the (220) peak shifts from 67.6° for Pt NCs to 68.2° and
6
8.4° for Pt Co and Pt Co , indicating the replacement of Pt by Co in the lattice structure. The
3 3 2
lattice constants of the NCs, determined from the position of the (220) peak on the xꢀray
scattering patterns, are 3.92, 3.87, and 3.87 Å for Pt, Pt Co, and Pt Co NCs, respectively. Using
3
3
2
ACS Paragon Plus Environment

ACS Catalysis
Page 10 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Vegard’s law, the bulk Co fraction in the Pt Co and Pt Co NCs is estimated to be 13.4 mol.%.
3
3
2
The fact that the bulk Co fraction is lower than that of the alloy stoichiometry is an initial
indication of Co segregation.
The local environments of the Pt and Co atoms were further investigated using XꢀRay
Absorption Spectroscopy (XAS). The Pt L3 nearꢀedge spectra of Pt Co samples reduced in H at
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
2
2
o
2
50 and 400 C are shown in Figure 6a, together with the Co K nearꢀedge spectra (Figure 6b). On
the Pt edge, the white line and edge positions of the alloy coincide with those of the Pt foil for
both reduction temperatures, demonstrating that the Pt is fully reduced in all cases. However, the
Extended XꢀRay Absorption Fine Structure (EXAFS) on the Pt edge suggests that there is
surface segregation of Co in the PtꢀCo alloy particles, as the Pt:Co ratio in the Pt coordination
sphere (3.1 ± 0.6) is greater than the nominal. Considering the fact that the cobalt precursor was
injected at 170 °C in the synthesis, it is reasonable that the PtꢀCo alloy NCs had cobalt rich shell,
because platinum precursor can be decomposed and nucleated at lower temperature, which
possibly formed the core of NCs. The fitting parameters for these spectra are reported in Table
o
S3, columns 3ꢀ5. The Co edge EXAFS was not fitted for the sample reduced at 250 C, as the
variables required to completely describe the structure are greater than the number of available
independent data points. After reduction at 400°C, the Pt:Co ratio in the Pt coordination sphere is
closer to the nominal (2.1 ± 0.4), consistent with at least partial reverse Co segregation to the
bulk. A similar reverse segregation phenomenon has been reported previously for PtNi
4
1
nanoparticles .
The XꢀRay Absorption Near Edge Spectra (XANES) of the Co K edge, reported in
Figure 6b, provides further information on the nature of the alloy NCs. First, the spectra indicate
that Co remains partially oxidized, even after reduction at 400°C. Using a linear combination of
spectra from CoO and Co standards to fit the results for the alloy catalyst, the average Co
oxidation states after reduction at 250°C and 400 °C were 1.2 (60% CoO) and 0.72 (36% CoO),
4
2
respectively. Surface oxygen is known to induce Co surface segregation in a PtꢀCo alloy . In
43
turn, Co forms a surface monolayer oxide with properties distinct from those of the bulk CoO .
44
As step sites constitute ca. 30% of all surface sites for 3ꢀ4 nm nanoparticles and tend to stay
4
5
more oxidized than terraces , we assume 30% of surface sites to be composed of Co in the +2
oxidation state (see Section S2 for model details). Following an analogous Fe O /Pt(111)
3
2
ACS Paragon Plus Environment

Page 11 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
46
structure observed using STM , we consider the rest of the monolayer surface oxide to be Co O
3
2
with a honeycomb structure on a PtꢀCo metallic core as a nanoparticle model (Figure 7a).
Table 1compares XAS/XRD results with geometric estimates of the alloy core/oxide
shell spherical nanoparticle model. Overall, the model exhibits excellent agreement with our
experimental data, particularly given the approximations invoked. We predict an average Co
oxidation state of 1.18, close to the experimental estimated value of 1.20. The low Co oxidation
state is consistent with an O:Co atomic ratio of less than 1 on the majority of surface sites. In
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4
3
contrast, a previously observed surface oxide with CoO stoichiometry would yield an average
oxidation state >1.5, different from the XANES results. The predicted Co content in the bulk
alloy (14.4 mol. %) is close to the Vegard’s law estimation from XRD (13.4 mol. %) The
experimental and geometric PtꢀCo and PtꢀPt CNs agree well. The XAS data demonstrate that
Pt Co NCs, reduced at 250°C, consist of a Ptꢀrich core with the majority of Co segregated to the
3
2
surface, forming a CoO surface oxide shell. The Co O honeycomb monolayer as a dominant
x
3
2
surface structure is consistent with XAS results.
Table 1. Average coordination numbers and Co oxidation states for Pt Co NCs reduced at
3
2
o
2
50 C, determined by Xꢀray absorption spectroscopy and estimated using a spherical core/shell
NC model with planes covered by a Co O surface oxide monolayer (70%) and step sites
3
2
covered by CoO (30%). CN stands for “coordination number”.
XAS/XRD data Geometric Model
1
Co oxidation state
PtꢀCo CN
1.20
1.18
2.92
8.91
14.4
1
2.88 ± 0.44
1
PtꢀPt CN
8.98 ± 0.57
2
Co content in the bulk alloy, %
13.4
1
2
XAS data regression; XRD analysis (Vegard’s law)
Further evidence for Co O surface oxide formation comes from CO chemisorption
3
2
measurements on NCs supported on γꢀAl O , shown in Table S4. After reduction at 250°C, CO
2
3
adsorption on the Pt Co NCs is negligible, consistent with the Pt atoms being covered. After
3
2
4
00°C reduction, CO chemisorption is comparable to what was observed with Pt NCs, due to
reverse segregation of Co to the bulk. DFT results indicate that CO interacts weakly with
ACS Paragon Plus Environment

ACS Catalysis
Page 12 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
46
Co O /Pt(111) compared to Pt(111) (ca ꢀ0.7 vs. ~ꢀ2 eV binding energy , respectively),
3
2
consistent with the lack of CO adsorption. A similar in magnitude CO binding energy was
correlated with no CO adsorption in XPS measurements on Fe O /Pt(111) honeycomb structure .
4
6
3
2
The importance of a Co O overlayer structure is further supported from DFT calculations,
3
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
discussed next.
Theoretical Insights into the Reaction Mechanism, Catalyst Composition Effects, and
Catalyst Stability
DFT calculations were performed in order to understand the mechanism of the HDO
reaction, catalyst stability issues, and the differences among the three catalysts (Pt, Pt Co, and
3
Pt Co ). Regarding the reaction mechanism on the Pt Co catalyst, the calculations showed that
3
2
3
2
the Co O honeycomb monolayer supported on the Pt rich core is capable of catalyzing key
3
2
reaction steps involved in the HMF to DMF conversion. The overall reaction is assumed to
proceed via the following steps: 1) H dissociation, 2) C=O hydrogenation, and 3) selective HDO
2
with concomitant oxygen removal from the surface in the form of water. These steps and their
activation barriers are reported in Table S5. It is worth noting that we did not consider the EFA
(see Table S1), one of the B intermediates, in our calculations. The ether intermediates are
formed by etherification with solvent molecules, and including them would greatly complicate
the calculations. Moreover, the HDO reaction is not significantly affected by formation of the
13
ether products, as shown in a previous study. The H dissociation step can occur via several
2
homolytic and heterolytic dissociation paths (Table S5). The calculations indicate that homolytic
splitting of a weakly physisorbed H molecule (ꢀ0.1 eV binding energy) over a single Co atom is
2
most energetically favorable (a 0.7 eV reaction barrier). The final state (0.3 eV more stable than
gaseous H ) entails both H atoms bound to Co and Pt atoms in bridging configurations (Figure
2
S5, reaction 1, transition state).
Prior to the C=O hydrogenation step, HMF weakly adsorbs on the surface (ꢀ0.8 eV BE);
hydrogenation of the C=O carbonyl group exhibits a low reaction barrier when it is weakly
47,48
bound to the surface.
Specifically, a concerted addition of two H atoms to HMF occurs with
a 0.8 eV barrier, yielding BHMF. The overall hydrogenation is exothermic (reaction energy of ꢀ
.6 eV). BHMF can either desorb (0.9 eV desorption energy) or undergo HDO, ultimately
forming DMF.
0
ACS Paragon Plus Environment

Page 13 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
The subsequent HDO mechanism of BHMF on the Ptꢀrich particle with Co O coating is
3
2
radical in nature, and is depicted in Figure 8 (Table S5, reactions 7ꢀ10). Some of us have found a
similar mechanism of furfural HDO to form 2ꢀmethyl furan on a Ru/RuO catalyst to be
2
2
3
consistent with a range of experimental and computational data. BHMF undergoes CꢀO bond
scission on a honeycomb edge site consisting of two Co atoms (Figure S5), forming a loosely
bound radical and an OH group, with a reaction energy of +0.9 eV and a barrier of 1.2 eV. Next,
a hydrogen atom transfers from the OH to the radical, yielding HMMF and a chemisorbed
oxygen atom. CꢀO scission occurs similarly on the second hydroxymethyl group (not shown),
forming DMF as the final product. The chemisorbed O atom (+1.7 eV binding energy with
respect to H and H O) reacts rapidly with H (a ꢀ1.2 eV exothermic dissociative adsorption with
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
2
2
a 0.3 eV barrier) to form coꢀadsorbed OH and H that subsequently recombine with a 0.2 eV
barrier (ꢀ0.9 eV reaction energy) to form water. Finally, water desorbs with a +0.4 eV energy to
complete the catalytic cycle.
HMF, BHMF, HMMF, and DMF weakly interact with the Co O surface (~ ꢀ0.8 eV BE,
3
2
4
9
dominated by dispersion forces , as opposed to chemisorption on metal atoms with BE of the
5
0
order of ꢀ2 eV ), largely retaining a gaseousꢀlike molecular geometry (Figure S5). The absence
of covalent bonding of the ring with the metal surface is key to rationalizing the high selectivity
of the catalyst, because opening of the furanic ring and decarbonylation require strong
2
3
chemisorption of the furan ring in a flat geometry, with partial sp ꢀsp reꢀhybridization of ring
5
1–56
carbon atoms.
Lack of covalent bonding between the ring with the Co O oxide protects the
3 2
ring from further side reactions and explains the low reactivity of DMF. The Co O surface layer
3
2
is capable of catalyzing CꢀO bond hydrogenolysis in HMF that leads to selective production of
DMF.
In order to assess the catalyst stability at a H ꢀrich environment, we calculated the rate of
2
initiation of Co O reduction via vacancy formation, with the details presented in Section S3.
3
2
o
Under experimental HDO conditions (160 C, 33 bar H ), the vacancy formation rate is a factor
2
o
of 2 lower than under in situ XAS conditions (250 C, 1 bar H ). Furthermore, the vacancy, once
2
formed, is easily reoxidized by BHMFꢀtoꢀHMMF reactions. This analysis provides a
rationalization as to why the Co O surface oxide is stable in a reducing reaction environment.
3
2
In contrast to the highly selective, oxideꢀcovered Pt Co catalyst, Pt carries out facile
3
2
hydrogenation of the C=O group to BHMF, but dehydroxylates BHMF to form DMF slowly (see
ACS Paragon Plus Environment

ACS Catalysis
Page 14 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Section S4). Furthermore, the DMF ring interacts strongly with Pt, promoting ring hydrogenation
and ring opening with barriers which are lower than that of the dehydroxylation reaction (see
Section S4). The computed barriers for HDO of HMF are comparable to the barriers for ring
hydrogenation and ring opening of DMF, consistent with the observation that selectivity to DMF
is modest.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Pt Co NCs exhibit catalytic properties intermediate between Pt and Pt Co . Unlike the
3
3
2
Pt Co NCs, there are not enough Co atoms to completely cover the surface with an oxide
3
2
monolayer (Table S6); in this catalyst, the surface is predicted to consist of 1/2 Co oxide and
1/2 Pt atoms. We believe that this significant difference in structure exposes Pt patches on the
~
surface (Figure 7b). The presence of Pt sites at the surface leads to the partial nonꢀselectivity of
the Pt Co catalyst.
3
DISCUSSION
The development of better catalysts for HDO of HMF requires an understanding of the
1
3,18
reaction mechanism. First, it is important to recognize that the reaction is sequential.
The
poor selectivity that is observed with many metals is due to the fact that they further catalyze
reactions of DMF, the desired product. While it is required that a catalyst has good activity for
HDO of HMF, a selective catalyst must also be a poor catalyst for reactions of DMF. The
sequential nature of the reaction also makes it essential that no part of the catalyst is nonselective.
For reactions in which both the desired and side products form in parallel, having a small
percentage of the catalyst surface showing a lower selectivity will not dramatically change the
overall selectivity. With a sequential reaction, the nonselective part of the catalyst can have a
much more dramatic effect. This has important consequences for alloy catalysts. While the
catalyst based on Pt Co NCs has the necessary properties to achieve very high selectivities,
3
2
alloy catalysts prepared by conventional impregnation methods will not be so compositionally
uniform. Both Pt and Co are individually are nonselective because they are active for reactions of
DMF, so that any metal in the catalyst which is not in the form of an alloy will be nonselective.
(
Note: This is indeed observed. A Pt Co catalyst prepared by incipient wetness of the metal salts
3 2
showed poor control of particle size and composition (Figure S6 in SI) and a maximum
selectivity of less than 80% (Figure S7 of SI).)
ACS Paragon Plus Environment

Page 15 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
17
In this context, it is interesting to consider the work from Schüth and coꢀworkers , who
first reported extremely high selectivities for HDO of HMF with PtCo alloys. In their case, the
highest selectivities were achieved when the metal particles were encapsulated in porous carbon
spheres. We suggest that those carbon spheres were important for achieving a high compositional
uniformity in the particles.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Control of the metal composition is essential in order to achieve good HDO selectivity.
From the HDO studies, we learned that Pt Co was not as selective as Pt Co . The reason for this
3
3
2
is the difference in particle surface structure. Pt Co does not have enough Co atoms to
3
completely cover the surface. The incomplete coverage of Co oxide leads to the presence of
uncovered Pt sites, and therefore causes partial nonꢀselectivity of the Pt Co catalyst. While it
3
may be possible to increase the Co:Pt ratio further, a catalyst with excess Co is likely to be
1
8
unselective since monometallic Co catalysts are unselective. Catalyst stability is equally
important to activity and our results suggest that there is a direct correlation between stability
and selectivity. The most serious and rapid deactivation in our experiments was due to coking
which must be caused by further reaction of overhydrogenated products, such as the 2,5ꢀ
hexanedione. The PtꢀCo alloy catalyst also seems to be more tolerant against sintering, possibly
as a result of the coreꢀshell structure.
While the catalyst based on Pt Co NCs exhibits excellent properties, it would obviously be
3
2
beneficial to find a selective catalyst that did not require a precious metal. Investigations on the
gasꢀphase reaction of furfural have shown that baseꢀmetal alloys can exhibit good selectivity for
1
4
the HDO reaction to methylfuran over the decarbonylation reaction to form furan. Also, recent
work by some of us demonstrated that there are many similarities between the reactions of
furfural and HMF, with differences due mainly to the operating conditions typically employed
5
7
for these reactions. Therefore, it will be interesting to examine some of the compositions used
in the furfural reactions for HDO of HMF.
CONCLUSIONS
High selectivity of DMF from liquidꢀphase HDO of HMF with H can be achieved over a
2
wellꢀcontrolled PtꢀCo /C catalyst. Particularly, over Pt Co catalyst, 98% of DMF yield was
3
2
obtained with the optimized reaction temperature and space time. Recognizing the sequential
nature of the HMF HDO reaction is the key for catalystꢀdevelopment strategies. Noble metals
ACS Paragon Plus Environment

ACS Catalysis
Page 16 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
interact strongly with the ring to promote side reactions. The fundamental principle for the
superior performance of Pt Co is that the bimetallic alloy forms a monolayer oxide on the
3
2
surface of the metallic core that interacts weakly with the furan ring to prevent overꢀ
hydrogenation and ring opening of DMF to secondary byꢀproducts while forming active sites to
carry out the HDO process. In this regard, composition control is crucial to cover the entire
surface with an oxide layer and avoid exposed metallic patches that can promote side reactions.
Given that HDO is commonly employed in biomass upgrade, the learnings from this study could
be used for the upgrade of other molecules, including bioꢀoil.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
SUPPORTING INFORMATION
Product distributions of partiallyꢀhydrogenated and overꢀhydrogenated compounds,
particle size distributions of NCs, EXAFS fitting parameters, metal dispersions, DFT energies
and barriers, honeycomb structure parameters, oxide vacancy formation kinetic, TEM analysis of
NCs size change before and after reaction, HDO elementary reaction steps, impregnated PtCo
HDO performance, reaction network and energy barriers for HDO over Pt, preparation method
for NCs.
ACKNOWLEDGEMENTS
We acknowledge support from the Catalysis Center for Energy Innovation, an Energy
Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of
Basic Energy Sciences under Award no. DEꢀSC0001004 and useful discussions with Prof.
Ayman Karim.
REFERENCES
(1) RománꢀLeshkov, Y.; Barrett, C. J.; Liu, Z. Y.; Dumesic, J. A. Nature 2007, 447, 982ꢀ985.
(
(
(
(
2) Nakagawa, Y.; Tamura, M.; Tomishige, K. ACS Catal. 2013, 3, 2655ꢀ2668.
3) Wang, T.; Nolte, M. W.; Shanks, B. H. Green Chem 2014, 16, 548ꢀ572.
4) Dashtban, M.; Gilbert, A.; Fatehi, P. RSC Adv 2014, 4, 2037ꢀ2050.
5) Torres, A. I.; Daoutidis, P.; Tsapatsis, M. Energy Environ. Sci. 2010, 3, 1560ꢀ1572.
(6) Chen, G.; Shen, Y.; Zhang, Q.; Yao, M.; Zheng, Z.; Liu, H. Energy 2013, 54, 333ꢀ342.
ACS Paragon Plus Environment

Page 17 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
(
7) Williams, C. L.; Chang, C.ꢀC.; Do, P.; Nikbin, N.; Caratzoulas, S.; Vlachos, D. G.; Lobo,
R. F.; Fan, W.; Dauenhauer, P. J. ACS Catal. 2012, 2, 935ꢀ939.
(8) Chidambaram, M.; Bell, A. T. Green Chem. 2010, 12, 1253ꢀ1262.
(9) Thananatthanachon, T.; Rauchfuss, T. B. Angew. Chem. Int. Ed. 2010, 49, 6616ꢀ6618.
(10) Jae, J.; Zheng, W.; Lobo, R. F.; Vlachos, D. G. ChemSusChem 2013, 6, 1158ꢀ1162.
(11) Hu, L.; Tang, X.; Xu, J.; Wu, Z.; Lin, L.; Liu, S. Ind. Eng. Chem. Res. 2014, 53, 3056ꢀ
3064.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(12) Basudeb Saha; Christine M. Bohn; Mahdi M. AbuꢀOmar. ChemSusChem 2014, 7, 3095ꢀ
101.
13) Luo, J.; ArroyoꢀRamírez, L.; Gorte, R. J.; Tzoulaki, D.; Vlachos, D. G. AIChE J. 2015, 61,
90ꢀ597.
(14) Sitthisa, S.; An, W.; Resasco, D. E. J. Catal. 2011, 284, 90ꢀ101.
3
(
5
(15) Myint, M.; Yan, Y.; Chen, J. G. J. Phys. Chem. C 2014, 118, 11340ꢀ11349.
(16) Lee, J.; Kim, Y. T.; Huber, G. W. Green Chem. 2014, 16, 708ꢀ718.
(17) Wang, G.ꢀH.; Hilgert, J.; Richter, F. H.; Wang, F.; Bongard, H.ꢀJ.; Spliethoff, B.;
Weidenthaler, C.; Schüth, F. Nat. Mater. 2014, 13, 293ꢀ300.
(
18) Luo, J.; ArroyoꢀRamírez, L.; Wei, J.; Yun, H.; Murray, C. B.; Gorte, R. J. Appl. Catal.
Gen. 2015, 508, 86ꢀ93.
(
19) Shevchenko, E. V; Talapin, D. V; Rogach, A. L.; Kornowski, A.; Haase, M.; Weller, H. J.
Am. Chem. Soc. 2002, 124, 11480ꢀ11485.
(20) Cargnello, M.; DoanꢀNguyen, V. V. T.; Gordon, T. R.; Diaz, R. E.; Stach, E. a; Gorte, R.
J.; Fornasiero, P.; Murray, C. B. Science 2013, 341, 771ꢀ773.
(21) Gordon, T. R.; Cargnello, M.; Paik, T.; Mangolini, F.; Weber, R. T.; Fornasiero, P.;
Murray, C. B. J. Am. Chem. Soc. 2012, 134, 6751ꢀ6761.
(22) Huang, X.; Cao, L.; Chen, Y.; Zhu, E.; Lin, Z.; Li, M.; Yan, A.; Zettl, A.; Wang, Y. M.;
Duan, X.; Mueller, T. Science 2015, 348, 1230ꢀ1234.
(23) Mironenko, A. V.; Vlachos, D. G. Under Review.
(24) Fievet, F.; Lagier, J.; Blin, B.; Beaudoin, B.; Figlarz, M. Solid State Ion. 1989, 32ꢀ33, 198ꢀ
2
15.
(25) Yu, Y.; Yang, W.; Sun, X.; Zhu, W.; Li, X. Z.; Sellmyer, D. J.; Sun, S. Nano Lett. 2014,
4, 2778ꢀ2782.
1
ACS Paragon Plus Environment

ACS Catalysis
Page 18 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
26) Cargnello, M.; Chen, C.; Diroll, B. T.; DoanꢀNguyen, V. V. T.; Gorte, R. J.; Murray, C. B.
J. Am. Chem. Soc. 2015,137, 6906ꢀ6911.
(27) Wang, D.; Xin, H. L.; Hovden, R.; Wang, H.; Yu, Y.; Muller, D. a.; DiSalvo, F. J.; Abruña,
H. D. Nat. Mater. 2012, 12, 81ꢀ87.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(28) Tengco, J. M. M.; LugoꢀJosé, Y. K.; Monnier, J. R.; Regalbuto, J. R. Catal. Today 2015,
246, 9ꢀ14.
(29) Scholten, J. J. F.; Pijpers, A. P.; Hustings, A. M. L. Catal. Rev. Sci. Eng. 1985, 27, 151ꢀ
206.
(
(
(
30) Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47, 558ꢀ561.
31) Kresse, G.; Furthmüller, J. Comput. Mater. Sci. 1996, 6, 15ꢀ50.
32) Kresse, G.; Furthmüller, J. Phys. Rev. B 1996, 54, 11169ꢀ11186.
(33) Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188ꢀ5192.
(34) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865ꢀ3868.
(35) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys. 2010, 132, 154104ꢀ154104.
(36) Henkelman, G.; Jonsson, H. J. Chem. Phys. 2000, 113, 9978ꢀ9978.
(37) Henkelman, G.; Uberuaga, B. P.; Jónsson, H. J. Chem. Phys. 2000, 113, 9901ꢀ9901.
(38) Sheppard, D.; Terrell, R.; Henkelman, G. J. Chem. Phys. 2008, 128, 134106ꢀ134106.
(39) Sheppard, D.; Henkelman, G. J. Comput. Chem. 2011, 32, 1769ꢀ1771.
(40) Ng, Y. H.; Ikeda, S.; Harada, T.; Sakata, T.; Mori, H.; Takaoka, A.; Matsumura, M.
Langmuir 2008, 24, 6307ꢀ6312.
(41) Lonergan, W. W.; Vlachos, D. G.; Chen, J. G. J. Catal. 2010, 271, 239ꢀ250.
(42) Menning, C. A.; Chen, J. G. J. Chem. Phys. 2008, 128, 164703ꢀ164703.
(43) De Santis, M.; Buchsbaum, A.; Varga, P.; Schmid, M. Phys. Rev. B 2011, 84, 125430ꢀ
125430.
(44) Blaylock, D. W.; Zhu, Y.ꢀA.; Green, W. H. Top. Catal. 2011, 54, 828ꢀ844.
(
45) Zhu, Z.; Tao, F.; Zheng, F.; Chang, R.; Li, Y.; Heinke, L.; Liu, Z.; Salmeron, M.;
Somorjai, G. A. Nano Lett. 2012, 12, 1491ꢀ1497.
(
46) Knudsen, J.; Merte, L. R.; Grabow, L. C.; Eichhorn, F. M.; Porsgaard, S.; Zeuthen, H.;
Vang, R. T.; Lægsgaard, E.; Mavrikakis, M.; Besenbacher, F. Surf. Sci. 2010, 604, 11ꢀ20.
(47) Tuokko, S.; Pihko, P. M.; Honkala, K. Angew. Chem. Int. Ed. 2016, 55, 1670ꢀ1674.
(48) Loffreda, D.; Delbecq, F.; Vigné, F.; Sautet, P. J. Am. Chem. Soc. 2006, 128, 1316ꢀ1323.
ACS Paragon Plus Environment

Page 19 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
(
49) Caratzoulas, S.; Vlachos, D. G. Carbohydr. Res. 2011, 346, 664ꢀ672.
50) Mironenko, A. V.; Gilkey, M. J.; Panagiotopoulou, P.; Facas, G.; Vlachos, D. G.; Xu, B. J.
Phys. Chem. C 2015, 119, 6075ꢀ6085.
(
(51) Wang, S.; Vorotnikov, V.; Vlachos, D. G. Acs Catal. 2014, 5, 104ꢀ112.
(52) Vorotnikov, V.; Mpourmpakis, G.; Vlachos, D. G. Acs Catal. 2012, 2, 2496ꢀ2504.
(53) Wang, S.; Vorotnikov, V.; Vlachos, D. G. Green Chem. 2014, 16, 736ꢀ747.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(54) Pang, S. H.; Schoenbaum, C. A.; Schwartz, D. K.; Medlin, J. W. Nat. Commun. 2013, 4,
1-6.
(
55) Pang, S. H.; Schoenbaum, C. A.; Schwartz, D. K.; Medlin, J. W. Acs Catal. 2014, 4, 3123ꢀ
131.
56) Schoenbaum, C. A.; Schwartz, D. K.; Medlin, J. W. Acc. Chem. Res. 2014, 47, 1438ꢀ1445.
3
(
(57) Luo, J.; Monai, M.; Yun, H.; ArroyoꢀRamírez, L.; Wang, C.; Murray, C. B.; Fornasiero, P.;
Gorte, R. J. Catal. Lett. 2016, 146, 711ꢀ717.
ACS Paragon Plus Environment

ACS Catalysis
Page 20 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1. Reaction network for HMF hydrodeoxygenation using alcohols as solvent.
ACS Paragon Plus Environment

Page 21 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. The TEM images of a) Pt NCs, b) Pt Co NCs, and c) Pt Co NCs, after removing the
3
3
2
solvent. d) Wide angle xꢀray scattering data of Pt (red), Pt Co (green), and Pt Co (blue). The
3
3
2
scale bars in the insets represent 2 nm.
ACS Paragon Plus Environment

ACS Catalysis
Page 22 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(a)
(b)
(
c)
Figure 2. Conversion and product distribution for the HDO reaction of HMF over (a) 10ꢀwt%
Pt/C, (b) 10ꢀwt% Pt Co/C, (c) 10ꢀwt% Pt Co /C, as a function of reactor space time. Reaction
3
3
2
conditions: 33 bar and 120 ºC. (ꢀ) HMF conversion, (ꢁ) product group B, (ꢀ) DMF, (ꢁ)
product group D
ACS Paragon Plus Environment

Page 23 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(a)
(b)
Figure 3. Conversion and product distribution for the HDO reaction of HMF over (a) 10ꢀwt%
Pt Co/C, (b) 10ꢀwt% Pt Co /C, as a function of reactor space time. Reaction conditions: 33 bar
3
3
2
and 160 ºC. (ꢀ) HMF conversion, (ꢁ) product group B, (ꢀ) DMF, (ꢁ) product group D
ACS Paragon Plus Environment

ACS Catalysis
Page 24 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. Time on stream measurements for HMF hydrodeoxygenation. Reaction conditions: 33
bar, 160 ºC, W/F 1.0 g·min/mL. (ꢁ) HMF conversion over 10ꢀwt% Pt Co /C, (ꢂ) DMF yield
3
2
over 10ꢀwt% Pt Co /C, (ꢀ) HMF conversion over 10ꢀwt% Pt/C, (ꢃ) DMF yield over 10ꢀwt%
3
2
Pt/C
ACS Paragon Plus Environment

Page 25 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(a)
(b)
(c)
Figure 5. Conversion and product distribution for the reaction of DMF as a function of space
time at 33 bar: (a) 10ꢀwt% Pt/C at 120 ºC; (b) 10ꢀwt% Pt Co/C at 160 ºC, (c) 10ꢀwt% Pt Co /C
3
3
2
at 160 ºC. (ꢀ) DMF conversion, (ꢂ) DMTHF, (ꢃ) 2ꢀhexanone, 2ꢀhexanol and 2ꢀ
propoxyhexane, (ꢂ) 2,5ꢀhexandione, 2,5ꢀdipropoxyhexane, () hexane
ACS Paragon Plus Environment

ACS Catalysis
Page 26 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 6. (a) Pt L3 edge XANES. Pt foil(red), Pt Co reduced at 250 ºC (blue) and Pt Co
2
3
2
3
reduced at 400 ºC (black). (b) Co K edge XANES. Co foil standard (black), CoO standard (red),
Pt Co reduced at 250 ºC (yellow) and Pt Co reduced at 400 ºC (green).
3
2
3
2
ACS Paragon Plus Environment

Page 27 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 7. (a) Pt Co nanocrystal model involving an alloy core (88% Pt, 12% Co based on XRD)
3
2
covered with a Co O surface oxide monolayer with a honeycomb structure; (b) and (c)
3
2
correspond to Pt and Pt Co NC models.
3
ACS Paragon Plus Environment

ACS Catalysis
Page 28 of 29
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. Reaction mechanism of BHMF hydrodeoxygenation to HMMF on the Co O /Pt(111)
3
2
surface. DFT reaction barriers (energies) are given in eV. The inset depicts a portion of a
Co O /Pt(111) surface. Two Co atoms participating in CꢀO bond activation are encircled with a
3
2
white ellipsoid.
ACS Paragon Plus Environment

Page 29 of 29
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table of Contents Graphic
ACS Paragon Plus Environment