Enhanced alcohol production over binary Mo/Co carbide catalysts in syngas conversion

Article abstract of DOI:10.1016/j.jcat.2020.09.003

Recently metal carbide catalysts have attracted attention as alternatives to pure metals for the conversion of syngas to higher oxygenates, a process which could enable the sustainable production of fuels, polymers, and chemicals. Although Mo and Co carbides have both shown promise for higher oxygenate production, they have not achieved the requisite activity and selectivity for practical implementation. In this work we synthesize and characterize a binary Mo and Co carbide catalyst that exhibits improved activity and oxygenate selectivity relative to either pure metal carbide. We apply a combination of advanced electron microscopy and X-ray diffraction to show that the binary Mo/Co carbide catalyst forms uniformly mixed amorphous nanoparticles. Through in situ X-ray absorption spectroscopy studies, we determine that the structure of the mixed metal carbide catalyst under reaction conditions consists of both carbidic and bimetallic components. By testing the catalytic properties of a series of Mo/Co carbide catalysts prepared by different synthesis methods, we find that the Mo and Co sites must be in close contact to achieve improved syngas conversion to higher alcohols. Through in situ DRIFTS measurements, both Mo and Co atoms at the surface of the catalyst are identified as adsorption sites for reactive species.

Full text of DOI:10.1016/j.jcat.2020.09.003

Journal Pre-proofs
Enhanced Alcohol Production over Binary Mo/Co Carbide Catalysts in Syn-
gas Conversion
Arun S. Asundi, Adam S. Hoffman, Miaofang Chi, Sindhu S. Nathan, Alexey
Boubnov, Jiyun Hong, Simon R. Bare, Stacey F. Bent
PII:
S0021-9517(20)30368-7
https://doi.org/10.1016/j.jcat.2020.09.003
YJCAT 13891
DOI:
Reference:
To appear in:
Journal of Catalysis
Received Date:
Revised Date:
Accepted Date:
17 July 2020
1 September 2020
2 September 2020
Please cite this article as: A.S. Asundi, A.S. Hoffman, M. Chi, S.S. Nathan, A. Boubnov, J. Hong, S.R. Bare, S.F.
Bent, Enhanced Alcohol Production over Binary Mo/Co Carbide Catalysts in Syngas Conversion, Journal of
Catalysis (2020), doi: https://doi.org/10.1016/j.jcat.2020.09.003
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2020 Published by Elsevier Inc.

Enhanced Alcohol Production over Binary Mo/Co Carbide Catalysts in Syngas Conversion
Arun S. Asundi1, Adam S. Hoffman2, Miaofang Chi3, Sindhu S. Nathan1, Alexey Boubnov2,†,
Jiyun Hong2, Simon R. Bare2, Stacey F. Bent1
1Department of Chemical Engineering, Stanford University, Stanford, California, 94305, USA
2SSRL, SLAC National Accelerator Laboratory, Menlo Park, California, 94025, USA
3CNMS, Oak Ridge National Laboratory, Oak Ridge, Tennessee, 37830, USA
†Current Address: Karlsruhe Institute of Technology, Karlsruhe, Germany
Abstract
Recently metal carbide catalysts have attracted attention as alternatives to pure metals for the
conversion of syngas to higher oxygenates, a process which could enable the sustainable
production of fuels, polymers, and chemicals. Although Mo and Co carbides have both shown
promise for higher oxygenate production, they have not achieved the requisite activity and
selectivity for practical implementation. In this work we synthesize and characterize a binary Mo
and Co carbide catalyst that exhibits improved activity and oxygenate selectivity relative to
either pure metal carbide. We apply a combination of advanced electron microscopy and X-ray
diffraction to show that the binary Mo/Co carbide catalyst forms uniformly mixed amorphous
nanoparticles. Through in situ X-ray absorption spectroscopy studies, we determine that the
structure of the mixed metal carbide catalyst under reaction conditions consists of both carbidic
and bimetallic components. By testing the catalytic properties of a series of Mo/Co carbide
Page 1 of 46

catalysts prepared by different synthesis methods, we find that the Mo and Co sites must be in
close contact to achieve improved syngas conversion to higher alcohols. Through in situ DRIFTS
measurements, both Mo and Co atoms at the surface of the catalyst are identified as adsorption
sites for reactive species.
1. Introduction
Non-renewable resources like coal, natural gas, and petroleum are widely used in the
production of many common materials, including fuels, chemicals, and polymers.1–3 However, as
these non-renewable carbon sources are rapidly depleted and the environmental impacts of their
use become increasingly severe, it has become necessary to develop sustainable methods of
producing fuels and high value chemicals. One promising strategy is the catalytic conversion of
synthesis gas (syngas, CO + H2) to higher oxygenates, which can be directly used as fuel
additives or further refined to more valuable products.4–6 Because syngas can be produced from
non-renewable carbon sources, as well as from biomass and CO2, the conversion of syngas to
value-added products is well-suited to facilitate the transition to the utilization of sustainable
resources.7,8
The most significant challenge in the catalytic production of valuable chemicals from
syngas is achieving high selectivity towards higher oxygenates. This stems from the two possible
forms of CO adsorption on a catalyst surface. On one hand, CO can undergo dissociative
adsorption on the catalyst surface, in which the carbon-oxygen bond is broken. Subsequent
hydrogenation of the adsorbed carbon and coupling of alkyl species result in the formation of
hydrocarbon products, as is characteristic of Fischer-Tropsch catalysts. On the other hand, CO
can associatively adsorb on the catalyst surface and be hydrogenated to form methanol. Most
Page 2 of 46

catalysts nearly exclusively support only one form of CO adsorption, and consequently exhibit
high selectivity towards either hydrocarbons or methanol.9 However, the generally accepted
mechanism for higher oxygenate formation involves a CO insertion step, in which a CO
molecule inserts into an alkyl fragment, yielding an oxygenate with multiple carbon atoms.4,7
This requires a dual-function catalyst surface that supports both dissociative and non-dissociative
CO adsorption. Because of the bifunctional surface site requirement, the development of a high
performance catalyst for higher oxygenate synthesis necessitates the selection of multiple active
materials, optimization of active site structure, and tuning of surface electronic properties in
order to direct reaction pathways towards the desired products.
The best performing catalysts that have been developed are promoted rhodium materials,
owing to the moderate intrinsic selectivity of Rh towards higher oxygenates.10–12 However,
because Rh is extremely scarce and very expensive,13 the practical implementation of higher
alcohol synthesis requires the development of less expensive catalytically active materials. To
this end, metal carbides have emerged as a promising alternative to pure metal catalysts.14 Metal
carbides have many similar properties to metals, including close-packed structure and electrical
conductivity,15,16 making them good alternatives to metal catalysts traditionally used in syngas
conversion. Moreover, the surface electronic properties of a metal carbide differ from those of
the corresponding metal due to hybridization between the metal d and carbon 2p orbitals;17 thus,
an inexpensive metal that shows poor performance for higher oxygenate synthesis can potentially
be transformed into a highly selective metal carbide catalyst.
Previous studies have shown that both molybdenum and cobalt carbides exhibit intrinsic
selectivity towards higher oxygenate synthesis from syngas.18–22 Pei et al.23 and Singh et al.24
demonstrated that Co/Co2C catalysts possess relatively high selectivity towards methanol and
Page 3 of 46

higher alcohols compared to pure Co, which is a known Fischer-Tropsch catalyst.25 Both studies
agreed that that the catalyst surface supported different types of CO adsorption sites, thus
enabling CO insertion and higher oxygenate formation. Christensen et al. compared the syngas
conversion properties of Mo, W, and Nb carbides and concluded that molybdenum carbide
showed the highest activity and selectivity towards alcohols.26 The performance of Mo carbide
catalysts has also been improved by the addition of alkali and transition metals. In particular,
several studies27–31 have demonstrated that over Mo-based catalysts, including metallic Mo,
Mo2C, and MoS2, the alcohol yield could be improved by adding Co. Despite the evidence for
enhanced higher alcohol production from mixed Mo and Co carbide catalysts, there are still
many questions regarding the structure of the catalyst under reaction conditions. Different
catalyst phases have been observed in the Co-promoted Mo-based catalysts described in
literature, including Mo2C, Co2C, Co3Mo3C, and metallic Mo and Co. In addition, active sites
and the nature of the Mo/Co interaction that gives rise to improved catalyst performance have
not been determined. Xiang et al.28 suggested several possible roles of Co in promoting Mo2C,
including the formation of Co3Mo3C, the formation of Co2C, or the electronic modification of
Mo surface sites. On the other hand, Yin et al.29 demonstrated enhanced alcohol production over
Co-modified Mo2C catalysts, but observed no Mo/Co interactions through spectroscopy or
diffraction that could explain the improved catalytic behavior. Furthermore, Al-Megren et al.30
and Christensen et al.31 utilized Co-modified Mo-based catalysts, but focused their studies on the
effects of support and feed gas composition, and did not discuss Mo/Co interactions. Finally, in
all these studies, K or Rb were utilized as promoters, which resulted in improved catalyst
performance but convoluted the Mo/Co interactions with the effects of the alkali metal promoter.
Page 4 of 46

In this work we seek to better understand how the structure and active sites of a binary
Mo/Co carbide catalyst result in enhanced syngas conversion to higher alcohols. Measurements
of the Mo/Co carbide catalytic reactivity are consistent with results in the literature that mixed
Mo/Co carbide catalysts possess enhanced activity and higher oxygenate selectivity compared to
either pure metal carbide. Interestingly, the pure Mo carbide catalyst forms the -MoC1-x phase,
훼
which shares similar catalytic properties to the more commonly studied -Mo C phase. Using ex
훽
2
situ and in situ spectroscopic, scattering, and microscopic characterization methods, we
demonstrate that the binary metal carbide catalysts form uniformly mixed, amorphous
nanoparticles. Furthermore, in situ diffuse reflectance infrared Fourier transform spectroscopy
(DRIFTS) measurements of CO adsorbed on the catalysts reveal the presence of both Mo and Co
binding sites on the surface of the mixed Mo/Co carbide. Finally, we compare the performance
of mixed Mo/Co carbide catalysts prepared by different synthesis methods to determine the
necessary conditions required to achieve improvements in activity and oxygenate selectivity. We
show that improved catalyst performance only occurs when the Mo and Co carbide are in close
contact, and as a result, we conclude that the mixture of Mo and Co active sites on the Mo/Co
carbide surface are favorable for oxygenate formation.
2. Experimental Methods
2.1 Catalyst Synthesis
Silica gel (Davisil grade 643, Sigma-Aldrich) was used as the support material for all
catalysts. To remove the effects of impurities in the silica support,32 the catalysts were prepared
on silica gel that had been modified with the following pretreatments. To remove alkali and
transition metal impurities from the support, the silica was washed in 2 M nitric acid at 80 ºC for
2 h. The powder was rinsed with Milli-Q water until a neutral pH was obtained and subsequently
Page 5 of 46

dried in air at 120 ºC for 24 h.* Finally, the SiO2 powder was calcined in air at 750 ºC for 4 h to
ensure that the subsequent high temperature annealing steps during catalyst synthesis and
activation would not induce structural changes in the support.
Mo and Co were added to the catalyst by incipient wetness impregnation (IWI) of
ammonium heptamolybdate tetrahydrate (AHT, (NH4)6Mo7O24·4H2O, Sigma-Aldrich) and
cobalt(II) nitrate hexahydrate (CNH, Co(NO3)2·6H2O, Sigma-Aldrich). Catalysts were prepared
both by co-impregnation, in which the AHT and CNH were mixed in the impregnation solution,
and by sequential impregnation, in which the addition of AHT and CNH occurred in series. For
all Mo impregnation processes, hexamethylenetetramine (HMT, (CH2)6N4, Sigma-Aldrich) was
also added to the impregnation solution as a carbon source to carburize the Mo with a 9:1 molar
ratio of HMT to AHT, as reported elsewhere.33,34 The total weight loading of metal was fixed at
10% for all catalysts and the ratio of Mo:Co was controlled by varying the amounts of each
metal precursor used in the synthesis. Prior to IWI, the SiO2 powder was held under vacuum for
2 h to evacuate the pores. Appropriate masses of AHT and HMT were dissolved in 35%
NH4OH/H2O (v/v). Similarly an appropriate mass of CNH was dissolved in milliQ water. The
Mo and Co precursor solutions were mixed and quickly added dropwise to the SiO2 under
constant stirring. The powder was dried in air for 24 h, then calcined in N2 at 700 ºC for 2 h. The
catalysts resulting from this procedure are denoted “as-prepared”.
To differentiate the catalysts the following nomenclature will be used. Catalysts prepared
by co-impregnation of the Mo and Co, in which the two precursors were simultaneously
* In the absence of impurities the Co2C phase formed in the 100Co catalyst was found to be unstable and
decomposed to metallic Co under reaction conditions. Measurements of syngas conversion reactivity for the other
Mo/Co carbide catalysts prepared by co-impregnation revealed no effect of impurities on catalytic performance
(Figure S1). To ensure formation of the Co2C phase, the 100Co catalyst studied in this work was prepared on
unwashed SiO2, whereas all other catalysts were prepared on the washed SiO2. The effect of impurities on the
formation and stability of Co2C is the focus of current studies.
Page 6 of 46

introduced to the powder, are termed “XMo/YCo” to denote that the 10 wt% metal consists of X
atomic percent Mo and Y atomic percent Co. For example, 50Mo/50Co refers to a 10 wt% metal
catalyst with a 50:50 atomic ratio of Mo:Co. Catalysts prepared by sequential impregnation are
termed “50Mo/50Co seq imp” and “50Co/50Mo seq imp”, with the metal impregnation
occurring in the order listed. Finally, a physical mixture of the pure Mo and Co carbide catalysts
was also tested. This mixture is denoted “100Mo + 100Co phys mix”, and the masses of the
powders used to create the physical mixture were chosen to yield a 50:50 atomic ratio of Mo:Co.
2.2 Syngas Conversion Testing
Catalytic performance was measured in an Altamira Benchcat 4000 HP packed bed
reactor. In a typical experiment, 40 mg of catalyst powder was diluted with 60 mg silica gel and
loaded into a 1/4 in. outer diameter stainless steel tube with a chemically inert coating
(Silcolloy). The powder was packed between two plugs of quartz wool and supported by a bed of
glass beads. The first catalyst pretreatment was a reduction step, in which the catalysts were
exposed to 50 sccm H2 (Airgas, 5.0 purity) at 450 ºC for 2 h. The sample was then cooled to 250
ºC and the reactor pressure was increased to 20 bar. In order to convert Co to Co2C, the catalysts
then underwent an additional carburization pretreatment, in which the samples were exposed to
50 sccm CO (Airgas, 5.0 purity) for 12 h at 250 ºC and 20 bar.35 In order to remove metal
carbonyls, the CO was first passed through alumina pellets at 300 ºC and a room temperature bed
of carbon black (Vulcan XC-72). Following these pretreatment steps, the syngas conversion
reactivity of the catalyst was measured at 250 ºC, 20 bar with a feed of 10 sccm H2 and 5 sccm
CO for 24 h. The temperature, pressure, and gas flow rates were identical across all catalysts and
chosen to ensure a low CO conversion so the intrinsic reactivity of the catalyst could be studied.
Reaction products were analyzed by an Agilent 7890B gas chromatograph-mass spectrometer
Page 7 of 46

and quantified by a flame-ionization detector. The percentage of CO conversion was calculated
by dividing the total carbon content in all products by the CO concentration in the feed. Using
the CO flow rate and catalyst mass, a normalized reaction rate was determined from the percent
CO conversion. Carbon-weighted selectivities were calculated by dividing the total number of
CO molecules converted to a given product by the total number of CO molecules converted to all
products.
2.3 Catalyst Reduction and Ex Situ Characterization
Temperature-programmed reduction (TPR) experiments were performed in an Altamira
Hybrid Chemisorption-Reactor system. 50 mg of catalyst powder was loaded into a quartz U-
shaped tube and held in place by two quartz plugs. A thermocouple was loaded into the quartz
tube and placed next to the catalyst bed to monitor the temperature. Catalysts were heated to 300
ºC in 25 sccm He to remove excess water. After cooling to 40 ºC, the TPR experiment was
performed in 10% H2/90% Ar with a ramp rate of 10 ºC/min. The effluent gas was analyzed by a
thermal conductivity detector (TCD) to monitor the catalyst reduction.
X-ray diffraction (XRD) was performed on a Bruker D8 Venture diffractometer using a
Mo K source (17.45 keV, = 0.7107 Å). Powders were loaded into 0.034 inch OD polyimide
훼
휆
capillary for measurement. Diffraction patterns were collected by a 2D Photon 100 detector.
Transmission electron microscopy (TEM) was performed on an FEI Tecnai G2 F20 X-
TWIN TEM system operated at 200 kV. Samples were prepared by sonicating 1 mg of powder in
1 mL ethanol for 5 minutes, then drop-casting on lacey carbon/copper TEM grids (Ted-Pella).
Aberration corrected Scanning-TEM (STEM) imaging and energy dispersive X-ray spectroscopy
(STEM-EDS) elemental mapping was performed at 200 kV using a JEOL NEOARM microscope
equipped with dual SDD detectors. To minimize carbon contamination, powders were heated to
Page 8 of 46

100 ºC in N2 for 12 hours prior to imaging. Particle compositions were calculated from the Co-K
and Mo-L peaks in the EDS spectra integrated over each nanoparticle.
X-ray photoelectron spectroscopy (XPS) measurements were performed on a PHI
VersaProbe III spectrometer with Al K radiation (1486.6 eV). Pass energies of 224.0 and 55.0
훼
eV were used for survey and high-resolutions scans, respectively. Samples were prepared by
pressing powders into carbon tape, affixed to a Si wafer. Dual electron and ion neutralization
were used during measurement to mitigate charging effects and a spot size of 200 µm was used.
All binding energies were normalized to the position of the Si 2p peak at 103.5 eV.
2.4 In Situ Catalyst Characterization
Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was performed in
a Praying Mantis DRIFTS system (Harrick). Samples were loaded into a high temperature and
pressure reaction cell with ZnSe windows. Gas flow rates were controlled using EL-Flow series
mass flow controllers (Bronkhorst). CO was passed through a bed of alumina pellets at 300 ºC to
decompose iron and nickel carbonyl impurities. Prior to measurement, catalysts were reduced at
380 ºC (ramp rate of 10 ºC/min) in 20 sccm H2 at 1 atm for 1 hr. After cooling to room
temperature, background spectra were measured in 20 sccm N2 at 1 atm. The feed gas was then
switched to 50 sccm CO and the reaction cell was pressurized to 10 bar with pressure monitored
using a pressure transducer (Setra). After exposure of the catalyst to 10 bar CO for 15 minutes to
allow the surface to saturate, the feed gas was switched to 10 sccm N2 and the pressure was
reduced to 1 atm to purge excess gas phase CO. IR spectra were collected throughout the sample
treatment process using a Bruker Vertex 70 spectrometer with a liquid nitrogen-cooled mercury-
cadmium-telluride (MCT) detector. Each spectrum was measured by averaging 200 scans over
Page 9 of 46

the range 600 – 4400 cm-1. The spectra presented in this work were measured after the pressure
had reached 1 atm and most of the gas phase CO had been purged from the cell.
X-ray absorption spectroscopy (XAS) was performed on beamline 4-1 at the Stanford
Synchrotron Radiation Lightsource. Samples were packed in a 1 mm OD quartz glass tube with
wall thickness 20 µm between two plugs of quartz wool. The capillary was loaded into a custom-
built high temperature and pressure reactor described previously.36 Gas flow rates were
controlled using mass flow controllers (Brooks) and metal carbonyls were removed from the CO
using a Nanochem Metal-X purifier (Matheson). Pressure was controlled with a back-pressure
regulator (Brooks). Temperature was controlled with a Eurotherm PID controller and monitored
with a K-type thermocouple in the catalyst bed. During syngas exposure, the effluent gas was
analyzed by a mass-spectrometer (Hiden QGA) to ensure the reaction was taking place.
X-ray absorption near edge structure (XANES) and extended X-ray absorption fine
structure (EXAFS) measurements were performed at the Mo K-edge (20000 eV) and Co K-edge
(7709 eV). Incident photon energy was set by step-scanning the Si (220) (crystal orientation
=
휑
0) double-crystal monochromator. XAS data was collected in transmission mode using ionization
chambers (Ar used at Mo edge, N2 used at Co edge) with the sample bed oriented at 45º relative
to the beam. Spectra of the Mo and Co foils were simultaneously measured for energy
calibration. Four replicates of each EXAFS spectrum were collected and merged to ensure a high
signal-to-noise ratio.
A full in situ XAS experiment consisted of the same catalyst pretreatment and reaction
condition exposures that were performed in the catalytic performance tests. First, EXAFS
measurements were performed with the catalyst at room temperature and under 10 sccm He flow
(as-prepared state). Catalysts were then reduced in 10 sccm H2 at 450 ºC with a ramp rate of 10
Page 10 of 46

ºC/min. XANES scans were collected every 7 min during the TPR process to monitor the state of
the catalyst, and the reduction was judged to be complete when sequential XANES scans were
identical. Following reduction, the temperature was reduced to 250 ºC in H2 and EXAFS
measurements were taken (reduced state). The reactor was then pressurized to 20 bar in He, and
once operating pressure had been reached, the feed gas was switched to 10 sccm CO at 250 ºC to
carburize the catalyst. XANES measurements were collected every 7 minutes during this
carburization pretreatment step until the catalyst fully carburized as determined by consecutive
XANES scans being identical (~180 min at Mo edge, ~70 min at Co edge). The structure of the
catalyst was then measured by EXAFS (carburized state). Finally, the feed gas was switched to
syngas reaction conditions, 10 sccm H2 + 5 sccm CO at 250 ºC and 20 bar, and XANES scans
were collected to monitor the evolution of the catalyst structure during reaction. Spectra were
collected for at least 90 minutes at each edge energy to provide enough time for structural
changes during reaction to occur. EXAFS spectra were then taken under reaction conditions to
determine the final structure of the catalyst (reaction state). The 50Mo/50Co catalyst was taken
through the full experiment, with spectra collected at both the Mo and Co K-edges.
Measurements were performed on 100Mo and 100Co catalysts through the carburization step.
The 75Mo/25Co and 25Mo/75Co catalysts were studied up to the reduced state at both the Mo
and Co K-edges.
The XAS data was processed using the Athena software of the Demeter package.37 The
replicate XAS spectra were aligned, merged, energy calibrated, and normalized. Linear
combination fitting was attempted using MoO3, MoO2, Mo2C, and bcc Mo foil as standards for
the Mo XANES, and using CoO, Co(NO3)2, Co3O4, Co2C and fcc Co foil as standards for the Co
XANES.
Page 11 of 46

Modeling of the EXAFS data was performed in the Artemis software of the Demeter
package.37 Models were built from absorber-backscatter scattering paths simulated from bulk
2
material crystallographic data using FEFF6. S0 was determined by modeling the Mo and Co foil
EXAFS and found to be 0.94 and 0.77 for Mo and Co, respectively. Models of the catalyst
EXAFS were optimized in R-space using k1, k2, and k3 weightings, with all models obeying the
Nyquist criterion.38 Only the EXAFS spectra of the catalysts in the carburized and reaction states
were modeled, because models of the catalyst structures under other conditions required more
parameters than were available in the data. Fourier transformation was performed on EXAFS
function (k = 3 – 11 Å-1 for Mo; k = 3 – 10 Å-1 for Co) and fitting was performed on an R-range
of 1.3 – 3 Å for 100Mo and 50Mo/50Co at the Mo edge, 1 – 4 Å for 100Co, and 1 – 3 Å for
50Mo/50Co at the Co edge. Scattering paths generated from MoC, bcc Mo, and Co7Mo6 were
used to model the Mo edge EXAFS, and paths generated from Co2C and Co7Mo6 were used to
model the Co edge EXAFS.
3 Results and Discussion
3.1 Catalyst Performance
Page 12 of 46

Figure 1: Steady state syngas conversion activity and selectivity of Mo/Co carbide catalysts.
All reactions were carried out at 250 ºC, 20 bar, 15 sccm total flow rate (2:1 H2:CO). C2+ HC
represents higher hydrocarbons and C2+ oxy represents higher oxygenates.
Table 1: Steady State Syngas Conversion Performance of Mo/Co Carbide Catalysts
Carbon Selectivity (%)
% CO
Reaction Rate
Catalyst
100Mo
75Mo/25Co
50Mo/50Co
25Mo/75Co
100Co
CH4
26 ± 3
25 ± 3
24 ± 2
32 ± 5
44 ± 0.3
26
C2+ HC
40 ± 4
29 ± 2
30 ± 3
34 ± 4
21 ± 0.3
28
CO2
20 ± 6
4 ± 1
5 ± 2
4 ± 1
11 ± 10
3
MeOH
10 ± 2
27 ± 6
25 ± 7
17 ± 2
7 ± 4
C2+ oxy Converted (µmol CO/g*s)
4 ± 1
15 ± 2
16 ± 2
13 ± 4
16 ± 6
18
1.1 ± 0.3
2.6 ± 0.5
2.5 ± 1.4
1.9 ± 1.2
0.9 ± 0.03
1
1.3 ± 0.3
3.2 ± 0.7
2.5 ± 0.3
2.5 ± 0.6
0.7 ± 0.1
3.7
50Mo/50Co
seq imp
25
50Co/50Mo
seq imp
100Mo + 100
Co phys mix
36
45
34
30
4
9
15
12
11
3
3
3.6
0.3
0.6
Measurements performed at 250 ºC, 20 bar syngas, 15 sccm total flow (H2:CO = 2:1). C2+ HC
represents higher hydrocarbons and C2+ oxy represents higher oxygenates. Uncertainties shown
are standard deviations across three batches of sample.
Figure 1 and Table 1 show the syngas conversion activity and selectivity for Mo/Co
carbide catalysts. While the pure metal carbides, 100Mo and 100Co, both show some selectivity
towards methanol and higher oxygenates, consistent with results reported in literature,18,24,26 the
binary Mo/Co carbide catalysts prepared by co-impregnation show several improvements in
performance. First, the selectivities towards higher oxygenates (C2+ oxy) are significantly higher
in the mixed Mo/Co carbide catalysts compared to the 100Mo catalyst and are comparable to the
100Co catalyst. The mixed carbide catalysts also show higher methanol selectivities than either
pure metal carbide. This suggests a greater surface coverage of non-dissociated CO, which can
be hydrogenated to methanol or undergo CO insertion to form C2+ oxygenates. Second, the
binary Mo/Co carbide catalysts show slight decreases in hydrocarbon selectivity and
significantly decreased CO2 production. This is also consistent with a decrease in CO
dissociation on the Mo/Co carbide surface, which leads to less surface coverage of oxygen, and
Page 13 of 46

thus less CO2 formation either through the Boudouard reaction or via water formation and
subsequent water-gas shift. These changes in hydrocarbon and oxygenate selectivity are most
significant when some Co is added to the Mo carbide, as seen in comparing the 100Mo and
75Mo/25Co catalysts. With decreasing Mo:Co ratio, the oxygenate selectivity decreases and the
methane selectivity increases, trending towards the performance of the pure Co carbide catalyst.
Third, the binary Mo/Co carbide catalysts have higher CO conversion rates compared to the pure
metal carbides. The addition of Co to the Mo carbide initially causes a 2.5-fold increase in
activity, as seen in the 75Mo/25Co catalyst, and the activity decreases as the Co content in the
catalyst increases.
To understand the types of active sites in the binary metal carbide catalysts, the catalytic
reactivity was compared between several Mo/Co carbide catalysts with a 1:1 atomic ratio of
Mo:Co, prepared by different methods. Figure 1 and Table 1 show the catalytic performance of
Mo/Co carbide catalysts prepared by sequential impregnation, both with the Mo deposited before
the Co, and in the opposite order. The catalytic performance of a physical mixture of the pure
Mo and Co carbide catalysts was also tested, with the amounts of each catalyst selected to yield a
1:1 atomic ratio of Mo:Co. The physical mixture of the 100Mo and 100Co catalysts shows no
enhancement in catalytic performance, with activity and selectivity comparable to each
individual metal carbide. In contrast, both catalysts prepared by sequential impregnation show
similar increases in activity and oxygenate selectivity that were observed in the binary catalysts
prepared by co-impregnation. The differences in selectivity between the two sequentially
impregnated catalysts may result from particle size effects, as will be discussed in section 3.3.
These results suggest that the two metal carbides must be in close contact in order to achieve the
improved catalytic performance. Thus, it is suggested that the active sites for higher oxygenate
Page 14 of 46

formation are present on the mixed Mo and Co carbide surface, with no gas phase intermediates
that transport from one metal carbide to the other. Furthermore, the similarity in observed
catalytic performance between the co-impregnated and sequentially impregnated catalysts
suggests that these catalysts have similar morphologies under reaction conditions. This
morphology will be characterized and discussed in the next sections.
3.2 Characterization of As-prepared and Reduced Catalysts
Figure 2: XRD patterns of Mo/Co carbide catalysts in (a) as-prepared, (b) post reduction, and
(c) post carburization conditions. Standard diffraction patterns for MoC1-x, fcc Co, and Co2C
are plotted for reference. The patterns have been offset vertically for clarity.
The formation of Mo and Co carbides was achieved in two steps. The first step occurred
during incipient wetness impregnation of the Mo and Co. Hexamethylenetetramine was added to
the impregnation solution as a carbon source and the catalyst was calcined to form the Mo
carbide as has been reported in the literature.33 The calcined catalysts are denoted “as-prepared”.
Following reduction of the catalyst, the second carburization step was performed to ensure
formation of Co carbide. Conversion of metallic Co to Co carbide was achieved via a high
pressure CO exposure.24,35 Attempts were made to carburize the Mo through this high pressure
CO exposure rather than using hexamethylenetetramine, but no Mo carbide formation was
observed. The state of the catalyst following the high pressure CO exposure will be denoted
“carburized”. This second carburization step was performed immediately prior to syngas
Page 15 of 46

reaction, and thus was performed in the syngas conversion reactor. In order to characterize the
phases present in the catalyst, the structure of the material in each stage of the catalyst lifetime
was measured, as discussed in sections 3.2 – 3.4.
The crystalline phases present in the as-prepared catalysts were characterized by XRD
(Figure 2a). The 100Mo catalyst shows diffraction peaks consistent with molybdenum carbide in
the -MoC1-x phase (PDF No. 00-015-0457). This phase has a rock-salt structure and can
훼
potentially contain vacancies at the carbon sites.39,40 The slight shifts in diffraction peak positions
in the 100Mo catalyst relative to the MoC1-x standard may result from a difference in carbon
vacancy content or strain in the MoC1-x lattice due to the small nanoparticle size.41 As the Co
content increases, the intensities of the MoC1-x peaks decrease and new peaks attributed to fcc Co
emerge (PDF No. 00-015-0806). In particular the reflection at q
3.1 Å-1 grows more intense
≈
and a peak at q
3.6 Å-1 emerges, both of which are consistent with fcc Co. We note, however,
≈
that these peaks are broad and weak in intensity, indicating that the majority of Co species are
either highly dispersed or not crystalline, and thus cannot be fully characterized by XRD. The
presence of MoC1-x without cobalt carbide in the as-prepared catalyst suggests that the Mo
carbide phase is stable under high temperature N2 conditions, as has been reported previously,33
whereas the cobalt(II) nitrate decomposes under these synthesis conditions to form metallic Co
and dispersed Co species. The 100Co catalyst in the as-prepared state shows peaks that can all be
assigned to a Co3O4 phase (PDF No. 00-042-1467). This is unexpected for the decomposition of
cobalt(II) nitrate in N2,42 and it is possible that a small leak resulted in exposure of the sample to
air during synthesis and oxidized the Co to Co3O4.43 However, because the catalyst is reduced at
450 ºC to fcc Co prior to carburization and reaction, the presence of Co3O4 in the as-prepared
100Co catalyst should not affect any of the characterization or catalytic reactivity results.
Page 16 of 46

TPR experiments were performed to determine the effect of reduction pretreatment on the
catalyst structure and oxidation state. Reduction profiles are shown in Figure S2. The peaks in
the reduction profile can be ascribed to the reaction of H2 with surface oxide species to form
H2O, but may also result from the decomposition of surface carbide species to form CH4. The
profile for the 100Co catalyst shows two peaks, corresponding to the two step reduction of
Co3O4 to CoO and then to Co. The 100Mo and Mo/Co carbide catalysts show reduction peaks
centered at 380-450 ºC.
The oxidation states of Co and Mo during the reduction process were studied by in situ
TPR XANES. For the 75Mo/25Co, 50Mo/50Co, and 25Mo/75Co catalysts, the Co K-edge TPR
XANES profiles show a reduction in the white line intensity and a decrease in edge energy
consistent with reduction of the catalysts (Figure S3).44–46 The presence of isosbestic points in
the Co XANES scans indicates a 1-to-1 transformation between the initial and final catalyst
structures. The chemical structure of the catalyst was characterized by linear combination fitting
of the Co XANES of the as-prepared and reduced catalysts to bulk standards (Figure S4, Table
S1). The features in the XANES spectra of the as-prepared catalyst are similar to those of CoO,
and in the reduced catalysts, the structures are similar to fcc metallic Co. We attribute the
discrepancies between the measured XANES data and the fits to either the presence of Mo,
which may affect the structure of the Co phases, and/or the phases in the catalysts being present
as nanoparticles, which will exhibit weaker intensity oscillations in the XANES spectra
compared to bulk standards.47 The TPR XANES spectra for the 100Co catalyst shows a multi-
step reduction process (Figure S3d). Fitting the XANES spectra to Co3O4, CoO, and fcc Co foil
standards reveals the two-step reduction process of Co3O4 to CoO to Co (Figure S5), consistent
with the reduction profile for this catalyst.
Page 17 of 46

The Mo K-edge TPR XANES profiles for the catalysts are shown in Figure S6. A
decrease in edge energy is observed for all catalysts during the TPR process, indicating a
reduction in the oxidation state of Mo.48 The lack of isosbestic points in the TPR XANES spectra
indicates that the reduction occurs through a multi-step process, which is consistent with the
presence of multiple reduction peaks observed in the TPR profiles. The Mo XANES spectra for
the 75Mo/25Co, 50Mo/50Co, and 25Mo/75Co as-prepared catalysts show a pre-edge peak,
which is characteristic of Mo6+ species.49 However, once reduction is complete, the pre-edge
peak is no longer present, and the XANES spectra of all catalysts have a relatively similar shape.
This suggests that the structure of the Mo is qualitatively similar across all reduced catalysts. The
results of these reduction studies are consistent with the removal of a passivating surface oxide
encapsulating the catalyst nanoparticles. With the exception of the 100Co catalyst, the as-
prepared catalysts consist of MoC1-x and metallic Co, which are passivated by a surface oxide
layer upon exposure to air. To confirm the presence of MoC1-x and fcc Co in the reduced
catalysts, ex situ XRD measurements of the reduced samples were performed (Figure 2b). The
diffraction peaks attributed to MoC1-x are present in the reduced catalysts, demonstrating the
stability of the Mo carbide phase during the reduction process. With increasing Co content, the
reflection at q
3.1 Å-1 grows in intensity, consistent with the presence of fcc Co. Thus, we
≈
find that the reduction treatment at 450 ºC is appropriate to remove the surface oxides, while
leaving the underlying MoC1-x and metallic Co intact.
Page 18 of 46

Figure 3: In situ XANES spectra of reduced Mo/Co carbide catalysts at (a) Mo K-edge and
(b) Co K-edge. Reduction was performed at 450 ºC, 1 atm, 10 sccm H2, and measurements
were performed at 250 ºC, 1 atm, 10 sccm H2. Spectra are offset for clarity.
To characterize structural differences between the catalysts, the XANES spectra of the
reduced catalysts were compared. The Mo and Co K-edge positions of the reduced catalysts are
shown in Table S2 along with Mo and Co oxide and carbide standards for comparison. The Mo
edge energies are very close to those of Mo metal and the Mo2C standard. This is consistent with
the presence of a Mo carbide, which is expected to have metal-like properties.15 The Co edge
energies are also very similar to that of the Co foil, in agreement with the formation of metallic
Co in the reduced catalyst. The reduced catalyst XANES spectra are plotted in Figure 3. The
variations in the positions and intensity of the XANES features observed between the catalysts
may result from the differences in bonding environments of a typical Mo or Co atom as the
catalyst composition is changed. There is, however, a general trend that the oscillations in the
XANES spectra for catalysts with intermediate Mo/Co compositions are broadened and weaker
in intensity than those of the 100Mo or 100Co samples. The weakened intensity of the XANES
oscillations can result from either a smaller nanoparticle size or a decrease in crystallinity of the
Page 19 of 46

mixed metal carbide catalysts.50 Each of these possibilities will be discussed further in the next
section.
3.3 Ex Situ Characterization of Carburized Catalysts
Following reduction, the catalysts underwent a carburization step, before being exposed
to syngas reaction conditions. The carburized catalysts were characterized in order to determine
the structure and species present at the start of the syngas reaction. TEM images of the
carburized catalysts are shown in Figure S7. In all catalysts, nanoparticles are formed and the
average diameter of the nanoparticles increases with Co content (size range 3.4 – 8.5 nm
diameter). TEM images of the sequentially impregnated catalysts show that depositing Co before
Mo produces larger nanoparticles than the reverse order (Figure S8).
The variations in particle size across these catalysts may have an effect on reactivity. In
particular, from a purely geometric perspective with all other factors unchanged, the total
catalytic surface area per gram of material is expected to be inversely proportional to the
nanoparticle diameter. To quantify the effects of nanoparticle size on catalyst reactivity, area-
normalized activities were computed for each catalyst, assuming spherical nanoparticles with a
homogeneous distribution of Mo and Co (Table S3). The trends in the area-normalized catalyst
activities are very similar to those presented in Figure 1 and Table 1, with the mixed Mo/Co
carbide catalysts showing a factor of 2-3 increase in activity compared to either pure metal
carbide. Thus the primary factor driving the improved syngas conversion activity in the mixed
metal carbides is likely the variation in Mo:Co ratio, rather than the total exposed catalytic
surface area.
There may, however, be some effects of catalyst nanoparticle size on selectivity. For both
the co-impregnation and sequential impregnation synthesis methods, the catalysts with larger
Page 20 of 46

particle sizes showed greater selectivity towards hydrocarbons rather than oxygenates. Such
trends in catalyst performance likely result from variations in ratios of different types of surface
sites as a function of particle size and have been reported extensively in the literature.51–54 Thus,
although the differences in catalyst composition undoubtedly impact selectivity, these results
suggest that variations in nanoparticle size may also be driving the observed changes in product
selectivity.
Page 21 of 46

Figure 4: TEM images and STEM-EDS elemental maps of Mo/Co carbide catalysts post-
carburization. Catalysts are (a) 75Mo/25Co, (b) 50Mo/50Co, (c) 25Mo/75Co, (d) 50Mo/50Co
seq imp, and (e) 50Co/50Mo seq imp. Maps show Co (red), Mo (green), and overlay map
along with Si (blue).
To characterize the distribution of Mo and Co atoms in the binary metal carbides, high
resolution STEM-EDS elemental mapping was performed on the carburized catalysts. In the
Page 22 of 46

catalysts prepared by co-impregnation, the elemental maps reveal that the Mo and Co atoms are
uniformly mixed within each nanoparticle (Figure 4a-c). This is expected because the
impregnation solution contained a mixture of the metal precursors, and the results show that no
significant phase separation between the species occurs during the reduction and carburization
steps. However, somewhat surprisingly, the Mo and Co atoms in the catalysts prepared by
sequential impregnation are also uniformly mixed (Figure 4d-e), suggesting that a mixed
structure is more stable than one where the metals are phase segregated. This conclusion is in
agreement with the reactivity results, which showed that the co-impregnated and sequentially
impregnated catalysts have similar catalytic properties, implying that they have similar
structures. The Co and Mo EDS signals from Figure 4 were integrated over each nanoparticle for
all the catalysts and the resulting Mo fractions are shown in Figure 5. The average atomic
fraction of Mo in each nanoparticle agrees well with the nominal value expected from the
synthesis. We note some inhomogeneity in the Mo:Co ratio for each sample. Despite the
variations in particle composition, the presence of both Mo and Co atoms in all the nanoparticles
agrees well with the catalytic results showing that a close contact between Mo and Co is
necessary to achieve improved performance. Finally, STEM-EDS mapping of carbon was also
attempted to assess the spatial and elemental distribution of the metal carbides. Although efforts
were made to clean the catalyst surface prior to measurement, the carbon maps for most of the
catalysts were dominated by carbon contamination, making it impossible to distinguish carbon in
the metal carbide from that of the contamination. A carbon map with sufficiently low
background signal could only be performed for the 25Mo/75Co catalyst (Figure S9). In this
sample the location of carbon coincides with that of Co and Mo, consistent with the formation of
metal carbides.
Page 23 of 46

Figure 5: Box-whisker plots of nanoparticle compositions. Compositions were quantified
from STEM-EDS spectra integrated over each nanoparticle. From bottom to top, horizontal
lines in each box-whisker plot indicate minimum, first quartile, median, third quartile, and
maximum in particle composition. + sign indicates the mean particle composition. Expected
Mo concentrations are shown as horizontal lines for 75Mo/25Co, 50Mo/50Co, and
25Mo/75Co compositions.
The phases present in the carburized catalysts were probed by XRD (Figure 2c). The
100Mo catalyst shows peaks consistent with MoC1-x, as expected from the previous XRD
measurements collected post-reduction. The structure of the 100Co catalyst is Co2C (PDF No.
04-004-4639), indicating that the Co metal is carburized by the high temperature and high
pressure CO treatment. XRD patterns of the binary Mo/Co carbide catalysts show some peaks
that are attributable to both MoC1-x and Co2C. In particular, as the Co content increases, the peak
at q
3 Å-1 becomes more intense, aligning with the strongest peak in Co2C. In the 25Mo/75Co
≈
catalyst, additional structure can be observed in the peak from 2.8-3.2 Å-1, resulting from the
three peaks in the Co2C diffraction pattern in this range. The XRD patterns of the sequentially
impregnated catalysts also show MoC1-x features and an increased intensity of the q
3 Å-1
≈
reflection compared to pure MoC1-x that can be attributed to Co2C (Figure S10). However, in all
of the binary metal carbide catalysts the diffraction peaks are very broad and low in intensity.
This can stem either from the small size of the nanoparticles, or from a decrease in crystallinity
Page 24 of 46

in the mixed metal carbides. Because the TEM images discussed above show that the
nanoparticles are not only large enough to be detected by XRD (> 3 nm), but also increase in size
with Co content, we attribute the weak, broad peaks to poor crystallinity and conclude that the
structure of the binary metal carbide catalysts lack the long-range order that is present in the pure
metal carbides. This conclusion is consistent with the XANES spectra of the reduced catalysts,
which also showed a decrease in oscillation intensity in the mixed metal carbides as a result of
the amorphous structure.
Finally, XPS measurements were performed on the carburized catalysts to characterize
the Mo and Co oxidation states present at the surface. XPS survey scans of the catalysts show
Mo and Co peaks from the nanoparticles, as well as Si from the SiO2 support (Figure S11). The
oxygen peaks arise primarily from the SiO2 support, but may also have contributions from
surface oxidation of the nanoparticles. Additionally, C peaks can be observed, but likely result
from adventitious carbon as well as the carbon tape used to affix samples for the measurement.
The atomic percentages of each element, along with the fraction of Mo relative to Mo+Co are
shown in Table S4. The atomic fraction of Mo follows the expected decreasing trend as Co
content increases. The minor discrepancies between the expected and nominal Mo concentrations
likely result from the very low atomic percentages measured for Mo and Co, approaching the
lower detection limit of the instrument. High resolution scans of the Mo 3d and Co 2p regions of
the XPS spectrum are shown in Figure S12. For a single Mo oxidation state, the Mo 3d region is
expected to have 2 peaks with a separation of 3.15 eV, resulting from the 3d5/2 and 3d3/2 spin-
orbit splitting. The Mo XPS spectra for the catalysts show multiple peaks, indicating a mixture of
oxidation states (Figure S12a). The low oxidation state Mo can be assigned to Mo carbide,20,55 in
which the Mo is expected to have metal-like characteristics. The higher oxidation state Mo likely
Page 25 of 46

results from surface oxidation that occurred when transporting the sample from the reactor to the
XPS spectrometer. The Co 2p region of the XPS spectrum also shows a mixture of oxidation
states (Figure S12b). The Co0 peak is assigned to cobalt carbide, which has nearly identical
binding energy to that of metallic Co.56,57 An additional Co2+ peak can be observed, which likely
arises from air exposure resulting in surface oxidation to CoO when transferring samples to the
XPS spectrometer. Together, the ex situ spectroscopic and structural characterization of the
carburized samples reveals that the metal carbide catalysts form homogeneous, uniformly mixed
nanoparticles, which have the expected Mo:Co ratios and an amorphous structure.
3.4 In Situ Characterization of Catalysts under Carburization and Reaction Conditions
Figure 6: In situ Mo K-edge XANES for (a) 100Mo and (b) 50Mo/50Co catalysts.
Measurements are shown for the catalyst in as-prepared (25 ºC, 1 atm, 10 sccm He), reduced
(250 ºC, 1 atm, 10 sccm H2), carburized (250 ºC, 20 bar, 10 sccm CO), and reaction (250 ºC,
20 bar, 10 sccm H2, 5 sccm CO) conditions. XANES spectra for bulk bcc Mo and Mo2C
standards are shown. (c) Magnitude (black) and imaginary component (blue) of the EXAFS
Fourier transform are shown. Measured data are shown as points and fits are shown as solid
lines. Catalyst spectra were fit in an R range of 1.3-3 Å. Spectra are offset for clarity.
Corresponding k-space EXAFS and table of EXAFS fitting results are shown in Figure S14
and Table S5.
In situ XAS experiments were performed on the 100Mo, 100Co and 50Mo/50Co
catalysts to confirm the presence of the carbide phases observed in ex situ measurements and
determine if the structure is stable under reaction conditions. After reduction, the time evolution
Page 26 of 46

of the Mo K-edge XANES for the 100Mo catalyst under carburization conditions show some
slight changes, but similar overall structure, consistent with the presence of MoC1-x in the
carburized catalyst (Figure S13a). The in situ XANES spectra of the 100Mo catalyst in the as-
prepared, reduced, and carburized states are all similar, demonstrating the stability of MoC1-x
throughout these pretreatment conditions (Figure 6a). We note that the XANES spectra of the
catalyst show different features from both the bcc Mo and Mo2C standards. This is expected
because the structure of the MoC1-x phase present in the catalyst differs from the structures of
these standards. To further understand the in situ structure of the carburized 100Mo catalyst,
detailed modeling of the EXAFS spectrum was performed (Figures 6c, S14, Table S5). The
model consists of a first shell Mo-C scattering path, and a second shell Mo-Mo scattering paths.
In order to obtain a satisfactory fit, Mo-Mo scattering paths from both MoC and bcc Mo needed
to be included in the EXAFS model. We note that the contributions of the Mo-Mo paths from
MoC and bcc Mo are out of phase with one another (Figure S15), and as a result, the magnitude
of the Mo-Mo scattering peak in the Fourier transform is low and the uncertainties on the Mo-
Mo coordination numbers are large. However, the catalyst is expected to be primarily Mo
carbide based on the MoC1-x phase measured by XRD and the lack of change in the XANES
spectra throughout the reduction and carburization processes. In addition, we note that the
coordination numbers are slightly lower than would be expected for bulk MoC (6 for Mo-C and
12 for Mo-Mo). This is expected for the EXAFS model of the 3-4 nm nanoparticles observed by
TEM, since the surface atoms are undercoordinated.58 Because the XRD and XANES
measurements indicate that the catalyst structure is primarily MoC1-x, we suggest that the Mo-Mo
scattering path from bcc Mo in the model of the catalyst EXAFS represents a distortion in the
MoC1-x structure rather than the presence of a metallic Mo phase. This assertion is consistent
Page 27 of 46

with the catalyst XRD patterns, which show shifts in the 100Mo diffraction peak positions
relative to the MoC1-x standard pattern and no detectable metallic Mo phase. The distortion in the
Mo-Mo bond length may arise from carbon vacancies in the MoC1-x structure or from strain
induced by the small nanoparticle sizes. Thus the EXAFS modeling of the 100Mo catalyst
reveals a Mo carbide structure, but with some distortions from a perfectly crystalline MoC1-x
phase.
Figure 7: In situ Co K-edge XANES for (a) 100Co and (b) 50Mo/50Co catalysts.
Measurements are shown for the catalyst in as-prepared (25 ºC, 1 atm, 10 sccm He), reduced
(250 ºC, 1 atm, 10 sccm H2), carburized (250 ºC, 20 bar, 10 sccm CO), and reaction (250 ºC,
20 bar, 10 sccm H2, 5 sccm CO) conditions. XANES spectra for bulk fcc Co and Co2C
standards are shown. (c) Magnitude (black) and imaginary component (blue) of the EXAFS
Fourier transform are shown. Measured data are shown as points and fits are shown as solid
lines. Catalyst spectra were fit in an R range of 1-4 Å for 100Co and 1-3 Å for 50Mo/50Co.
Spectra are offset for clarity. Corresponding k-space EXAFS and table of EXAFS fitting
results are shown in Figure S16 and Table S6.
Following reduction, XANES evolution of the 100Co catalyst during carburization shows
a 1-to-1 transformation from fcc Co to Co2C, indicated by the presence of isosbestic points in the
XANES spectra (Figures 7a, S16a). The EXAFS of the carburized 100Co catalyst can be fit well
with a model consisting of the first three shells of bulk Co2C (Figures 7c, S17, Table S6). When
performing the fit, the coordination numbers were fixed at bulk values for Co2C, which is
reasonable for the relatively large nanoparticle sizes in the 100Co catalyst observed by TEM.
Page 28 of 46

The formation of a Co2C phase in the 100Co catalyst is consistent with ex situ XRD
measurements of the carburized catalyst and prior studies of Co-based catalysts.
The 50Mo/50Co catalyst shows slight changes in both Mo and Co K-edge XANES
spectra during carburization (Figures S13b, S16b). However, no changes in catalyst structure
were observed from the carburized state throughout syngas reaction conditions (Figure 6, 7,
S18). The lack of structural change in the catalyst during reaction was verified by ex situ XRD
measurements, in which the patterns for the post-carburization and post-reaction (“spent”)
catalysts are identical (Figure S19). Because of this, the EXAFS spectra for the 50Mo/50Co
catalyst after carburization and under steady state reaction conditions were modeled and fit
simultaneously. The model for the Mo K-edge EXAFS of the 50Mo/50Co catalyst consisted of
three paths: a Mo-C scattering path generated from bulk MoC, and Mo-Mo and Mo-Co
scattering paths from the bimetallic Co7Mo6 (Figures 6b, S14, S20 Table S5). The coordination
number of Mo-C is 2.9 ± 0.3, which is 48% of the value expected for bulk MoC. This suggests
that approximately half of the Mo atoms are in a carbide-like structure. The remaining Mo in the
50Mo/50Co catalyst is in a metallic structure, indicated by the presence of metallic Mo-Mo and
Mo-Co scattering paths in the model and the relatively low Mo-C coordination number. The
EXAFS fit yielded similar coordination numbers for the Mo-Mo and Mo-Co paths, showing that
the average Mo atom is expected to have both Mo and Co neighbors. This result is consistent
with the well-mixed nanoparticles observed by STEM-EDS elemental mapping.
The Co K-edge EXAFS model for the carburized and reaction condition 50Mo/50Co
catalyst was achieved using four scattering paths: Co-C and Co-Co from Co2C, and Co-Mo and
Co-Co from Co7Mo6 (Figures 7c, S17, Table S6). As with the Mo EXAFS model, a satisfactory
fit to the Co EXAFS data could only be achieved by including both carbidic and metallic
Page 29 of 46