Potassium-Promoted Molybdenum Carbide as a Highly Active and Selective Catalyst for CO2 Conversion to CO

Article abstract of DOI:10.1002/cssc.201700412

The high concentration of CO₂ bound in seawater represents a significant opportunity to extract and use this CO₂ as a C₁ feedstock for synthetic fuels. Using an existing process, CO₂ and H₂ can be con

Full text of DOI:10.1002/cssc.201700412

Accepted Article
Title: Potassium-promoted molybdenum carbide as a highly active and
selective catalyst for CO2 conversion to CO
Authors: Marc D Porosoff, Jeffrey W Baldwin, Xi Peng, Giannis
Mpourmpakis, and Heather D Willauer
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemSusChem 10.1002/cssc.201700412
Link to VoR: http://dx.doi.org/10.1002/cssc.201700412
A Journal of
www.chemsuschem.org

ChemSusChem
10.1002/cssc.201700412
FULL PAPER
Potassium-promoted molybdenum carbide as a highly active and
selective catalyst for CO conversion to CO
2
[a]
[b]
[d]
[d]
[c]
Marc D. Porosoff , Jeffrey W. Baldwin , Xi Peng , Giannis Mpourmpakis and Heather D. Willauer *
Abstract: The high concentration of CO
represents a significant opportunity to extract and use this CO
feedstock for synthetic fuels. Through an existing process
patented by the U.S. Navy, CO and H can be concurrently
extracted from seawater, then catalytically reacted to produce
synthetic fuel. Hydrogenating CO directly into liquid hydrocarbons
2
bound in seawater
byproduct, negatively influencing catalytic activity and product
[
10-11]
2
as a
selectivity.
Conversely, Co-based catalysts are water
and modifying an Fe catalyst with Co improves
catalytic performance and selectivity towards C + hydrocarbon
Improvements have also been made to Fe-based
catalysts by adding Cu, which enhances CO -FT activity and
[
12]
C
1
tolerant
2
2
2
[
13-14]
products.
2
2
[
14]
is exceptionally difficult, but by first identifying a catalyst for selective
CO production via reverse water-gas shift (RWGS), CO can then be
hydrogenated to fuel through Fischer-Tropsch (FT) synthesis.
Results of this study demonstrate that potassium-promoted
selectivity.
Although there are promising catalysts for CO
structure-property relationships that control activity and
2
-FT, the
[
15]
selectivity to intermediate hydrocarbons are not well studied.
Furthermore, because of the complexity of CO -FT, the
molybdenum carbide supported on γ-Al
O
2 3
2
(K-Mo C/γ-Al
2 3
O ) is a
2
low-cost, stable and highly selective catalyst for RWGS over a wide
range of conversion. These findings are supported by X-ray
diffraction (XRD), scanning electron microscopy (SEM) with energy
dispersive X-ray spectroscopy (EDS), X-ray photoelectron
alternative route of feeding CO produced via reverse water-gas
shift (RWGS) into a FT reactor must also be considered. For
industrial RWGS, operating temperatures are very high, typically
at or above 600 °C at 2.8 MPa, over ZnO/γ-Al
4
catalysts. Because methane (CH )
2
O
3
and
is
spectroscopy (XPS) and density functional theory (DFT) calculations. ZnO/Cr
2
O
3
thermodynamically favored below 600 °C, these catalysts
require high temperatures to selectively produce CO, which
[
16-17]
results in substantial deactivation.
To make fuel synthesis
Introduction
from CO viable, a low-cost and stable RWGS catalyst is first
2
required, which can achieve high selectivity to CO over a wide
range of conversion and operating temperatures.
-
1
2
The high concentration of CO in seawater, ca. 100 mg L ,
represents a significant opportunity to extract and use this CO
as a C
process patented by the U.S. Navy, CO
2
Recently, Pt-based catalysts have been investigated for
1
feedstock for synthetic fuels. Through an existing
[18-19]
RWGS,
but they are expensive, and thus, unviable for an
conversion process. As an alternative,
transition metal carbides (TMCs) are low-cost, with similar
[
1]
2
and H
2
can be
industrial scale CO
2
concurrently extracted from seawater and used as reactants for
direct Fischer-Tropsch from CO (CO -FT) to produce valuable
oxygenates, specialty chemicals and intermediate hydrocarbons
2
2
[20-21]
electronic properties to precious metals.
theory (DFT) calculations over the TMC, molybdenum carbide
Mo C) demonstrate that Mo-terminated Mo has many
properties similar to transition metals including Ru, Fe, Co and
Density functional
[
2-3]
(
C
2
-C
renewable, the entire process can be considered CO
The most commonly used catalysts for CO -FT are slight
6
) for synthetic fuel.
If the energy input is nuclear or
(
2
2
C
[
4]
2
-neutral.
2
[22]
Ni, all of which are active for CO
2
conversion.
DFT
variations of Fe and Co-based Fischer-Tropsch (FT) catalysts,
which show promise, but are not specifically designed for the
calculations by Shi et al. further illustrate that CO
2
dissociation
(
(
CO
CO
2
→ CO + O) is more favorable than CO
+ H → HCOO or COOH) over Mo C, suggesting high CO
2
Reactor experiments over unsupported-Mo C
2
hydrogenation
[5-7]
CO
achieves a CO
-C + hydrocarbons of 62.4% at a gas hourly space velocity
GHSV) of 0.0015 L g s , but the mechanism is poorly
2
reactant.
The current optimal catalyst, K-Mn-Fe/γ-Al
2 3
O ,
2
2
2
conversion of 41.4% and a selectivity towards
[23]
selectivity.
C
(
2
5
powder catalysts for RWGS at 300 °C and 0.1 MPa show 8.7%
-1
-1
[24]
conversion and 93.9% selectivity towards CO, confirming the
[
8]
understood, making catalyst improvements challenging. There
is some consensus that an Fe carbide formed during the
reaction is the catalytically active phase; however, reports also
state that Fe catalysts are poisoned by water, an unavoidable
DFT calculations. Another study over Mo
reports high activity and CO selectivity at 600 °C.
intrinsic activity of Mo C originates from CO binding in a bent
configuration, leading to spontaneous breakage of a C=O bond,
2
C nanowires also
[25]
The high
[
9]
2
2
[
26]
leaving CO and O bound to the surface.
The CO can desorb
C) is restored to
from the surface, while the oxy-carbide (O-Mo
2
[a]
[b]
[c]
Dr. Marc D. Porosoff
NRC Postdoctoral Research Associate
Dr. Jeffrey W. Baldwin
Acoustics Division
Dr. Heather D. Willauer*
[
24]
the active carbide through hydrogenation.
Mo C can also be modified with metal nanoparticles (Cu,
Co, Ni), which influence the product selectivity, leading to MeOH
2
[27]
[28-29]
with Cu,
Because modifying Mo
the selectivity between MeOH, C
be possible to modify Mo C to selectively produce even more
C
2
+ hydrocarbons with Co and CH
C with a metal promoter can further tune
+ hydrocarbons or CH , it may
4
with Ni.
Materials Science and Technology Division
Naval Research Laboratory
2
4555 Overlook Avenue, Washington DC, 20375
2
4
E-mail: heather.willauer@nrl.navy.mil
Prof. Giannis Mpourmpakis, Xi Peng
Department of Chemical and Petroleum Engineering
University of Pittsburgh
2
[d]
CO across a wide range of conversions and temperatures.
Experimental and theoretical studies suggest that potassium (K)
promoters increase the binding energy, and therefore, reactivity
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201700412
FULL PAPER
of CO
CO.
2
, thereby promoting C=O bond scission and formation of
2 3 2 2 3
Al O catalysts. These peaks are not present in Mo C/γ-Al O ,
suggesting that the addition of K promotes the formation of a
metallic Mo phase.
[
30-31]
In this paper, kinetic experiments and characterization
tools have been combined with DFT calculations to probe the
catalytic properties of K-promoted Mo C and understand the
reaction mechanisms of CO dissociation. Flow reactor results
indicate that K-Mo C/γ-Al is a highly active and stable RWGS
catalyst exhibiting high selectivity towards CO over a range of
operating conditions, with the presence of K promoting CO
dissociation to CO. These findings are supported by X-ray
diffraction (XRD), scanning electron microscopy (SEM) with
energy dispersive X-ray spectroscopy (EDS), X-ray
photoelectron spectroscopy (XPS) measurements and DFT
calculations.
To better identify the structure of K-Mo
images with EDS mapping of the reduced catalyst is shown in
Fig. 2. Overall, the morphology and particle size of K-Mo C/γ-
Al (Fig. 2a) and Mo C/γ-Al (Fig. S2) appear to be similar.
The EDS maps, however, show that the distribution of Mo over
each catalyst is notably different. The EDS map of the Mo C/γ-
Al catalyst, found in the SI, indicates that molybdenum is
evenly distributed over the γ-Al support. On K-Mo C/γ-Al
2 2 3
C/γ-Al O , SEM
2
2
2
2
2
O
3
2
O
3
2
2 3
O
2
2
2 3
O
O
2 3
2
2 3
O
in Fig. 2b-d, there is both (1) a large degree of segregation
between Mo (blue) and Al-rich (red) areas and (2) K (green) is
preferentially found in the Mo-rich areas, which suggests K
directly affects the electronic properties of the active Mo
phase.
2
C
Results and Discussion
To experimentally determine the effect of K addition on Mo
based supported catalysts, K-Mo C and the corresponding Mo
Mo and K-Mo control catalysts, all supported on γ-Al , were
2
C-
2
2
C,
2
O
3
synthesized through an evaporation-deposition procedure, as
outlined in the experimental section. XRD measurements over
the reduced catalysts, shown in Fig. 1, indicate that each of the
syntheized catalysts contain a combination of MoO
2
[JCPDS 32-
C [JCPDS 35-787] and metallic Mo [JCPDS 42-
120]. Each of these phases is assigned to the synthesized
671], β-Mo
2
1
catalysts by comparing the XRD spectra with the standard
database for specific bulk Mo phases.
2 2 3
Figure 2. (a) SEM micrograph for the reduced 2 wt% K-Mo C/γ-Al O , with (*)
and (▲) representing Mo rich and Al rich phases, respectively. EDS maps are
included of (b) Al in red, (c) Mo in blue and (d) K in green.
The difference in distribution of molybdenum between the two
catalysts can be explained by polymerization of a layer of
2 3 2 2 3
molybdenum oxide over the γ-Al O support in the Mo C/γ-Al O
catalyst, as seen in the SEM image in the SI, which may occur
[32]
during the calcination step of synthesis.
particles are simply too small to be resolved by SEM and must
be imaged through other techniques. On the K-Mo C/γ-Al
Alternatively, the Mo
2
2 3
O
Figure 1. XRD spectra of the reduced Mo
and 2 wt% K-Mo/γ-Al with the γ-Al
correspond to: γ-Al (♦), MoO (*), β-Mo
2
C/γ-Al
support as a reference. Symbols
2
C (•) and metallic Mo (Δ).
2 3 2 2 3
O , 2 wt% K-Mo C/γ-Al O
catalyst, there are clearly large K-Mo aggregates seen in Fig. 2,
which could be a result of the K promoter lowering the affinity of
Mo for the γ-Al O support.
2 3
2
O
3
2 3
O
2
O
3
2
Regardless of the differences in catalyst particle size and
morphology, there is no significant difference in catalytic activity
between the two samples. As seen in Fig. 3a, the conversion of
XRD measurements of the Mo-based catalysts indicate that
Mo
Mo
2
C/γ-Al
C and MoO
2
O
3
and 2 wt% K-Mo contain a mixture of β-
2
C/γ-Al
2
O
3
2
2
supported on γ-Al O . All supported Mo-based
2 3
Mo
activity of the two catalysts is comparable, the addition of 2 wt%
K to Mo C/γ-Al significantly improves the selectivity towards
CO. The effect of K addition becomes more apparent in Fig. 3b,
where CO selectivity of Mo C/γ-Al and 2 wt% K-Mo C/γ-Al
is reported across a range of conversions, by adjusting the gas
2 2 3 2 2 3
C/γ-Al O and K-Mo C/γ-Al O is similar. Although the
catalysts exhibit large peaks at 45.8° and 66.6°, from the γ-Al
2 3
O
support, and no identifiable peaks for MoO [JCPDS 35-609] are
3
2
2 3
O
present in any of the samples. Closer inspection of the XRD
spectra reveals the presence of a phase assigned to metallic Mo
at 40.5°, 58.7° and 73.7° on the K-Mo C/γ-Al O and K-Mo/γ-
2 2 3
2
2
O
3
2
2 3
O
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201700412
FULL PAPER
hourly space velocity (GHSV). Clearly, there is a strong
promotional effect from the addition of K, which leads to high CO
selectivity (~95%) from 6 to 23% conversion. The equilibrium
Uncarburized Mo/γ-Al
also included in Table 1 to clarify the role of the metallic Mo
identified in K-Mo C/γ-Al in the XRD measurements in Fig. 1.
ratio. The Mo/γ-Al and K-Mo/γ-Al control catalysts are reduced
ex situ in pure H at 600 °C prior to reaction to form metallic Mo.
The pre-reduction step ensures the high activity and CO
selectivity of the Mo C-based catalysts originate from the Mo
carbide phase, and not metallic Mo. Clearly, as seen in Table 1,
the Mo carbides, synthesized with CH , are more active than the
2 3 2 3
O and 2 wt% K-Mo/γ-Al O catalysts are
2
2 3
O
conversion for RWGS is 22.8% at 300 °C with a 3:1 H
2
:CO
2
2
O
3
2 3
O
Furthermore, as seen in Fig. 3a, the addition of the K promoter
decreases the deactivation percentage from 11.7% to 7.3% after
2
68 h on stream, an improvement in catalytic stability.
A
2
summary of the catalytic performance for both catalysts under
typical RWGS conditions can be found in Table 1.
4
Table 1 also includes K-Mo
loading to demonstrate the effect of K on catalytic performance.
The 1 wt% K-Mo C/γ-Al has a slightly higher CO yield than 2
wt% K-Mo C/γ-Al , but with increased methane production,
which wastes valuable H and requires a separation step before
2
C/γ-Al
2
O
3
with 1 – 3 wt% K
corresponding uncarburized catalysts, indicating that the
carburization step is necessary for high catalytic activity and that
2
2
O
3
the metallic Mo phase in K-Mo
responsible for the high performance.
2
C/γ-Al
2
O
3
is not solely
2
2 3
O
2
FT. Furthermore, as K loading increases, there is a drop in
catalytic activity, likely from the blocking of active sites. This
relationship between K loading and CO yield is not linearly
dependant on temperature, as clearly shown by Fig. S4 and
Table S1 for RWGS at 450 °C. At the higher temperature, the 3
Table 1. Summary of catalyst performance data for RWGS experiments at
-
1
-1
300 °C, 2.1 MPa and 3:1 H
2 2
:CO ratio at GHSV of 0.00084 L g s .
Deactivation percentage calculated for experiments at higher GHSV of 0.0021
-
1
-1
L g s , shown in Fig. 3a. RWGS catalysts from literature are also included
as a basis of comparison with the current K-Mo
2
C/γ-Al
O
2 3
catalyst.
Carbon Based
Selectivity / %
wt% K-Mo
selectivity, which outperforms the 2 wt% K-Mo
industrial ZnO/γ-Al and ZnO/Cr catalysts.
2 2 3
C/γ-Al O achieves 40.5% conversion and 98.2% CO
Conversion
/ %
CO Yield Deactivation
Catalyst
C/γ-Al
2
16]
O
3
and
A
/ %
/ %
2
CO
CH
4
C
2
-C
4
[
2
O
3
2 3
O
Mo
/γ-Al
2
C
2
1
9.9
73.5
19.7
12.4
6.8
6.6
14.6
11.7
comparison of the current catalyst with other relevant catalysts is
also included in Table 1 and the SI to demonstrate the high
O
3
1%K-
Mo
2
C
2
24.3
81.0
95.9
98.5
19.7
17.3
7.4
N/A
7.3
selectivity of K-Mo
studies are required to fully understand the effect of K loading on
K-Mo C/γ-Al for RWGS at different temperatures.
2
2 3
C/γ-Al O . Further reactor and spectroscopic
/γ-Al
O
3
3
2%K-
2
2 3
O
Mo
2
C
18.1
7.5
2.7
1.5
1.4
0
/γ-Al
2
O
3%K-
Mo
2
C
N/A
/
γ-Al
2
O
3
3
Mo
7
.1
.5
82.5
96.9
34
15.3
3.1
2.2
0
5.9
6.3
5.8
N/A
N/A
N/A
/γ-Al
2
O
2
%K-Mo
6
2 3
/γ-Al O
Mo C /
2
[
29] 17
N/A
N/A
250 °C
2
Cu-Mo C
/
229]50 °C
13
40
78
97
N/A
N/A
N/A
N/A
N/A
N/A
5.2
N/A
N/A
N/A
[
2 3
Pd/Al O
/
323]60 °C
N/A
34.2
N/A
32.1
[
Mo/γ-
Al
00 °C
2
O /
3
[
34]
6
The difference in CO yield between Mo
2
C/γ-Al
2
O
3
2
and K-Mo C/γ-
Al O is further explained by measuring the apparent activation
2
3
barrier for CO formation under RWGS conditions at 5 different
temperatures between 270 and 330 °C. For activation barrier
measurements, the GHSV is adjusted to operate at lower
conversions to ensure measurements are within the reaction
limited region. As seen calculated in Fig. 4, the activation
-1
barriers for CO formation are 14.0 and 11.4 kcal mol for
Mo C/γ-Al and K-Mo C/γ-Al , respectively. The lower
activation barrier for K-Mo C/γ-Al confirms the addition of 2
wt% K facilitates CO formation through a decrease of the
apparent activation barrier.
2
2
O
3
2
2 3
O
Figure 3. Performance of Mo
Al (hollow symbols) during RWGS. (a) Conversion (■/□) and selectivity to
CO (●/○) versus time on stream, and (b) CO (●/○) and CH (▲/Δ) selectivity
versus conversion. Experiments in (a) conducted at a GHSV of 0.0021 L g
2 2 3 2
C/γ-Al O (solid symbols) and 2 wt% K-Mo C/γ-
2
2 3
O
2 3
O
4
-
1
-
s
1
-1 -1
and range from 0.00084 to 0.0042 L g
s
in (b). All reactor studies are
under a 3:1 H :CO reactant ratio at 300 °C and 2.1 MPa.
2
2
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201700412
FULL PAPER
3
+
5+
carbides), represented by Mo and Mo on the catalyst surface.
It is also clear from the XPS measurements that metallic Mo is
present in K-Mo
active phase. From XRD measurements in Fig. 1, 2 wt% K-
Mo/γ-Al exclusively contains metallic Mo and γ-Al features,
but as seen in Table 1, the catalyst is one-third as active as 2
wt% K-Mo C/γ-Al under the same reaction conditions.
2 2 3
C/γ-Al O , but it alone is not the catalytically
2
O
3
2 3
O
2
2 3
O
Table 2. Composition of deconvoluted spectra for Mo
Mo C/γ-Al before and after RWGS reaction indicated in Table 1 with 3:1
:CO ratio at 300 °C and 2.1 MPa.
2 2 3
C/γ-Al O and 2 wt% K-
2
2 3
O
Figure 4. Activation barrier plots represented as ln(CO yield) versus 1/T for
H
2
2
2 2 3 2 2 3
(+) Mo C/γ-Al O and (X) 2 wt% K-Mo C/γ-Al O . The values of CO yield are
Mo Composition
calculated by averaging data points between 15 – 17 h on stream. The
dashed line, representing maximum thermodynamic CO yield, is included as a
reference.
0
2+
3+
4+
5+
6+
Catalyst
2 3
Mo C/γ-Al O
Condition Mo Mo
Mo
Mo
Mo
Mo
Fresh
Spent
Fresh
Spent
0
0
0
0
2.4
32.3 2.3
63.1
2
0.2
15.7 27.6 16.8 39.8
20.6 11.3 48.8
12.7 6.5
To further elucidate the effect of K on catalyst activity and
2 2 3
K-Mo C/γ-Al O
16.3 10.7 15.0 16.4 22.5 19.1
selectivity, XPS measurements of the Mo
found in Fig. 5. Fig. 5a contains the spectra of both the as-
prepared (fresh) and after reaction (spent) Mo C-based catalysts. Although the XPS results in Fig. 5 show both Mo
As evidenced by the figure, there is a shift towards lower binding oxides, the exact nature of the active phase is difficult to confirm
energies for both catalysts after reaction. On 2 wt% K-Mo C/γ-
Al
2
C-based catalysts are
2
2
C and Mo-
2
because the XPS measurements in this study are ex situ.
2
3
O , there is an additional large peak at 227.6 eV after reaction. During transfer of the catalyst samples from the reactor into the
From the deconvoluted XPS fits in Fig. 5b, the peak is from the
appearance of Mo , likely metallic Mo, which is supported by the
XPS, there is likely some degree of catalyst oxidation. As
shown in Table 2, there is a much larger amount of Mo and
0
2+
additional peaks seen on 2 wt% K-Mo
measurements in Fig. 1.
2
C/γ-Al
2
O
3
in the XRD
other reduced phases in K-Mo
2 2 3 2
C/γ-Al O relative to Mo C/γ-
Al , which suggest reduced Mo phases and oxy-carbides may
2
O
3
be critical for high catalyst stability and selectivity towards CO.
To fully reveal the nature of the active phase, further in situ
studies are required over K-Mo
To better understand the reaction mechanism of CO
dissociation and the electronic effects of K addition on Mo C-
based catalysts, DFT calculations were performed on pristine
and K-promoted β-Mo C(001) surfaces. The DFT results
indicate that CO chemisorption and activation are
thermodynamically feasible on the Mo-terminated β-Mo C(001)
surface (CO activates on hollow position, and can further
2 2 3
C/γ-Al O .
2
2
2
2
2
2
dissociate), as shown in Fig. 6, whereas, it does not chemisorb
on the C-terminated surface (not shown). Similar to that on the
pristine surface, CO does not chemisorb on the K-promoted C-
2
terminated surface, but can be activated and dissociated on the
K-promoted Mo-terminated surface. By directly comparing the
BE of CO on pristine and K-promoted surfaces, it becomes
2
Figure 5. Mo3d spectra (solid line) of (a) Mo
Al before and after RWGS reaction with fits (dashed line) and (b)
deconvoluted Mo3d spectra of spent Mo
2 2 3 2
C/γ-Al O and 2 wt% K-Mo C/γ-
2 3
O
clear that the presence of K enhances the binding energies of
2
C/γ-Al
O
2 3
2
and K-Mo C/γ-Al
2
O .
3
-1
both the physisorbed (from -0.5 to -6.5 kcal mol ) and
6+
5
4+
3+
Contributions include Mo , Mo , Mo and Mo for both samples, while spent
-1
chemisorbed states (from -31.4 to -46.4 kcal mol for activated
2+
0
2
K-Mo C/γ-Al
O
2 3
also includes Mo and Mo .
-1
CO
2
and from -74.8 to -80.9 kcal mol for dissociated CO
2
,
respectively, as shown in the intermediate states of Fig. 6 and
Fig. S6 in the SI).
The quantitative composition of each catalyst from the
deconvoluted XPS spectra before and after reaction can be
found in Table 2. Each deconvoluted spectra is fit with the
These adsorption energy changes in the presence of K are
[36-37]
due to electronic effects. Bader charge analysis
suggests
0
following values obtained from a report by Oshikawa et al, Mo :
that K (Bader charge = 0.82 e) loses almost one electron, which
is transferred to the surface of Mo C. As a result, the presence
of a positive point charge (K-cation) increases the dipole-dipole
interaction during CO physisorption, thus increasing the
physisorption energy. Concurrently, the partially negatively
charged Mo C surface facilitates the activation of CO (by
transforming the molecule from a linear to bent configuration),
2
+
3+
4+
2
27.6 eV, Mo : 228.2 eV, Mo : 228.8 eV, Mo : 229.9 eV,
2
5
+
6+
[35]
Mo : 231.8 eV and Mo : 233.1 eV.
, the K-Mo C/γ-Al sample is further reduced than Mo
Al after reaction. The facile reduction of K-Mo C leads to the
decrease of MoO
formation of larger amounts of Mo
Clearly, as seen in Table
2
2
2
O
3
2
C/γ-
2
2
O
3
2
6
+
4+
3
2
(Mo ) and MoO (Mo ) phases and
2
2
2
+
2
2
C (Mo ) and O-Mo C (oxy-
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201700412
FULL PAPER
and the adsorption of electrophilic species, such as atomic
temperature programmed desorption (TPD) also confirmed that
[
38]
oxygen.
The presence of K also affects the CO
2
dissociation
the presence of K on Mo
2
C surface increases the CO binding
[40]
kinetics through reduction of the barrier for CO
2
dissociation, as
energy thereby promoting its dissociation.
shown in Fig. 6 (CI-NEB calculations appear in the SI).
Interestingly, our DFT calculations reveal that the
desorption energy of CO from the surface with coadsorbed O*
BE = -0.5
from CO
sensitive to the presence of K. The desorption energy of CO is
almost isoenergetic for the K-promoted and pure Mo C(100)
surfaces (51.2 and 51.0 kcal mol , respectively). As a result,
the higher selectivity towards CO formation in the presence of K
can be attributed to electronic effects (surface charging) causing
2
dissociation (see final state in Fig. 6), seems to be less
BE = -14.6
BE = -31.4
2
-1
BE = -77.1
enhancement in CO
chemisorption) and reduction of the CO
Furthermore, it should be noted that the CO
2
adsorption (both physisorption and
dissociation barriers.
dissociation step
BE = -6.5
2
2
forms an oxy-carbide, as indicated by the high exothermicity of
the last reaction step in Fig. 6. These findings agree with the
BE = -32.4
BE = -46.4
XPS data in Fig. 5 and a previous study of Mo
which uses in situ X-ray absorption near edge spectroscopy
XANES) and ambient pressure-XPS to show that the active
2
C for RWGS,
BE = -80.9
(
Figure 6. CO
2
dissociation profile on Mo-terminated, pristine (black) and K-
[24, 41]
phase of Mo
Furthermore, the lack of CH
over K-Mo C/γ-Al is likely a result of less molecular hydrogen
2
C is an oxy-carbide under reaction conditions.
-
1
promoted (red), β-Mo
2
C(001) surfaces (energy in kcal mol ). Color code:
4
and other hydrogenated products
grey-C, cyan-Mo, red-O, purple-K.
2
2 3
O
available on the K-promoted surface.
2
For the surface without K (black profile), CO is first activated
-1
without a barrier (BE = -31.4 kcal mol ), and then dissociates to
-1
CO and O, with a barrier of 16.8 kcal mol . On the K-promoted,
Mo-terminated surface (red profile), CO is first barrierlessly
activated to a chemisorbed state (BE = -46.4 kcal mol ), and
Conclusions
2
-1
2 2 3
By modifying Mo C/γ-Al O with a K promoter, the CO selectivity
-1
then dissociates to CO and O with a barrier of 14.0 kcal mol ,
and yield increase significantly, and can approach the maximum
thermodynamic yield for RWGS under the appropriate reaction
conditions. Addition of K also improves the catalyst stability,
with only 7.3% deactivation after 68 h on stream. Catalyst
characterization by SEM with EDS clearly shows that K is
preferably found in Mo-rich regions, while Mo is more evenly
-
1
2
.8 kcal mol lower than that of the pristine β-Mo
In both cases, all of the steps are exothermic with respect to the
initial, CO physisorbed state. By considering the decrease in
activation energy for CO dissociation with the increase in CO
physisorption energy on the K-promoted, Mo-terminated β-
Mo C(001) surface in Fig. 6, it is apparent that K promotion
2
C(001) surface.
2
2
2
2
distributed in Mo
2
C/γ-Al
2
O
3
.
2 2 3
Furthermore, K-Mo C/γ-Al O
facilitates the formation of CO. These DFT-calculated reaction
barriers are in excellent agreement with those obtained
experimentally, as shown in Table 3.
maintains the Mo in a reduced and active state as evidenced by
XPS measurements. These experimental results are supported
by DFT calculations, which show enhanced CO
reduced CO dissociation barriers on the K-promoted, compared
to the pristine, Mo-terminated β-Mo C(001) surfaces. Notably,
2
adsorption and
2
Table 3. Comparison of theoretical activation barrier (Ea) calculated by DFT
2
and experimentally determined apparent activation barrier (Eaapp) for CO
-1
the DFT calculations predict a 2.8 kcal mol lower activation
barrier for CO formation upon K addition, which is in excellent
agreement with the experimentally measured difference of 2.6
2 2
formation from CO over pristine and K-promoted Mo C-based catalysts.
Theoretical
Ea / kcal mol
Experimental
Eaapp / kcal mol
-
1
-1
-1
kcal mol . The findings of this paper show that K-Mo
is a highly selective catalyst for producing CO from CO
2
C/γ-Al
2 3
O
Pristine (No K) Mo
2
C
16.8
14.0
14.0
11.4
2
and has
K-Promoted Mo
2
C
the potential to be used as a commercial RWGS catalyst for the
first step of synthesizing jet fuel from seawater for the U.S. Navy.
Future studies will focus on large-scale reactions and in situ
measurements to better understand and characterize the
In addition to the lower activation barrier for CO
over K-promoted β-Mo
lower CO uptake of K-Mo
2
dissociation
C(001), the higher CO selectivity and
2
-1
2
-
C/γ-Al
2
O
3
(1.7 μmol g ) relative to
(21.4 μmol g ) might indicate that addition of K
decreases the CO binding energy, allowing CO to desorb before
it is further hydrogenated into CH or other hydrocarbons.
However, DFT studies in the literature report that the addition of
2 2 3
catalytically active phase of K-Mo C/γ-Al O .
1
2 2 3
Mo C/γ-Al O
4
Experimental Section
K to Mo
CO.
2
C increases the binding energy and dissociation of
Previous surface science experiments using high
Catalyst Preparation. Potassium-promoted molybdenum carbide,
supported on gamma alumina (K-Mo C/γ-Al ) was synthesized by co-
impregnation of KNO3 and (NH )6Mo O •4H O precursors onto a γ-
[
31, 39]
2
2 3
O
resolution electron energy loss spectroscopy (HREELS) and
4
7
24
2
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201700412
FULL PAPER
Al O
2 3
support by the evaporation-deposition method. All chemicals were
Scientific Kα) equipped with a monochromatic Al Kα source and a
purchased from Sigma-Aldrich. In brief, the precursors were dissolved in
deionized water at the concentrations required to obtain molar ratios of
180° double focusing hemispherical analyzer with 128-channel detector.
The nominal XPS spot size and analyzer field of view were 400 μm .
2
1
/4/15 K/Mo/γ-Al
2
O
3
, which translated to 2 wt% potassium (K) and 20
support. For studies with other potassium
Settings for data collection were: Pass Energy = 20 eV, dwell time = 100
msec, energy step size = 0.1 eV over 35 averaged scans.
wt% Mo loading on the γ-Al
O
2 3
loadings, the amount of K precursor was adjusted to yield 1 – 5 wt% K.
Aqueous solutions of the metal precursors were added to a beaker of γ-
Charge compensation was necessary. Calibration of the instrument was
confirmed with the Cu2p3/2 peak of a Cu plate at 933 eV.
Al
2
O
3
and dried overnight under stirring at 60 °C, then calcined in air
Density Functional Theory (DFT) Calculations. DFT calculations
were performed using the Vienna Ab initio Simulation Package
overnight at 350 °C.
[
43]
The K-Mo/γ-Al
O
2 3
catalyst was then carburized in a 21% CH
C/γ-Al in a procedure
After the first 1.5 h, the CH was shut off and
the carbide was cooled to room temperature in H . At room temperature,
the catalyst was passivated in 1% O in N for several hours. The
Mo C/γ-Al catalyst was prepared by the same method with 20% Mo
4
in H
2
(VASP).
functional
Perdew–Burke–Ernzerhof (PBE) exchange–correlation
was employed within the generalized gradient
[
44-45]
mixture at 600 °C for 2.5 h to form the K-Mo
2
2
O
3
[
42]
adopted from Hou et al.
4
approximation (GGA) using the plane-wave projector augmented wave
[
46]
2
(PAW) pseudo-potentials.
2
The slab model of the β-Mo C(001) surface
2
2
had interchanging layers of C and Mo atoms, and therefore both C-
terminated and Mo-terminated surfaces were inspected. The slab model
2
2 3
O
loading, but without the inclusion of potassium. Metallic Mo-based
catalysts were used as controls and synthesized through the same
procedure as above, but without CH during the carburization step.
4
2
of the β-Mo C(001) surface (both terminations) contained 4 layers, with a
total of 16 C and 32 Mo atoms in one unit cell (using a 2x2 supercell with
size 10.50×12.15×14.55 Å, with a vacuum space of 10 Å). For K-
Pulse CO Chemisorption. Pulse CO chemisorption was
performed using an Autochem II Chemisorption Analyzer (Micrometrics)
to compare the number of active sites in each catalyst. Approximately
promoted β-Mo
2
C(001), one K atom was placed on the top layer. The
[
26]
kinetic energy cutoff was set to 415 eV,
the convergence criteria to
-5
-1
10 eV for the total electronic energy and 0.01 eV Å for the forces
acting on atoms and the k-point mesh was 5×5×1 k-point grid, generated
200 mg of catalyst was added into a U-shaped quartz reactor and
[
47]
reduced in 10% H in Ar at 300 °C for 2.5 h. The catalyst was cooled to
2
by Monkhorst-Pack scheme.
The Climbing Image-Nudged Elastic
Band (CI-NEB) method has been employed to locate transition states
in the CO dissociation pathway. During geometry optimization, the top
two layers as well as the CO molecules were fully relaxed, while the
[
48]
room temperature in He before pulsing 10% CO in He. A thermal
conductivity detector was used to analyze the amount of CO adsorbed by
each catalyst, which provides a means compare the number of active
sites.
X-Ray Diffraction (XRD). X-ray diffraction (XRD) measurements
of the reduced catalysts were performed on a Bruker D8 Advance
diffractometer using Cu Kα radiation, operated at 40 kV and 40 mA over
the range of 2θ = 20–80° with a 0.02° step size and 4 sec per step at
room temperature.
Scanning Electron Microscopy (SEM). Scanning electron
microscopy (SEM) measurements of the reduced catalysts were
performed with a Carl Zeiss Supra 55 Schottky thermal field emitting
microscope with a 30 μm aperture at 5–15 kV accelerating voltage. For
EDS measurements, an Oxford Instruments X-Max, 80 mm silicon drift
detector, at an accelerating voltage of 5 kV and working distance of 8.5
mm was inserted into the SEM chamber. The detector was calibrated
with a copper standard. Data was collected on 2048 channels for
2
2
bottom two layers were fixed in their bulk positions. Vibrational
frequencies on the adsorbates were performed to verify local minima
(absence of imaginary mode) and transition states (presence of
imaginary mode).
Two different CO
considered: horizontal and perpendicular to the surface. In addition,
different sites, top and hollow, were taken into consideration in CO
adsorption. The binding energies of different CO adsorption
configurations are given in the Supporting Information (SI). The binding
energy (BE) is calculated as: BE (adsorbate) = E(surface + adsorbate) –
E(surface) – E(adsorbate), where E(surface + adsorbate) is the total
electronic energy of the surface with the adsorbed CO
corresponding energy of the clean surface (without any adsorbate), and
E(adsorbate) is that of the CO molecule.
2
adsorption configurations on the surfaces were
2
2
2
, E(surface) is the
2
500,000 counts using a process time setting of 4 with pulse pile up
correction. The quantitative analysis was unnormalized using the
supplied 5 kV quantitative standard database.
Acknowledgements
Reactor Studies. To measure catalyst performance shown in Table
1
, 500 mg of catalyst was loaded into a 0.25 in stainless steel reactor and
reduced under 50 sccm H
for 2.5 h at 300 °C and 0.5 MPa. After
reduction, the reactor was isolated and the bypass pressurized to 2.1
MPa with 6.3 sccm CO , 18.9 sccm H and 5.0 sccm N , for a GHSV of
:CO ratio of 3:1. At 2.1 MPa, concentration of
the reactants in the bypass was recorded as a baseline and gases were
flowed into the reactor. Reactions were run for 22 h at 300 °C and
concentrations of reactants and products were analyzed by online gas
chromatography equipped with a flame ionization detector (FID) and 2
thermal conductivity detectors (TCD). The concentration of each gas-
phase species was calibrated by correlating the peak area of the pure
compound to its concentration in a calibration gas standard.
2
This work was supported by the Office of Naval Research both
directly and through the U.S. Naval Research Laboratory (NRL)
and OPNAV N45. This research was performed while Marc D.
Porosoff held a NRC Research Associateship Award at NRL.
Computational support has been provided by the center for
Simulation and Modeling (SAM) at the University of Pittsburgh.
X.P. and G.M. would like to thank Dr. Mudit Dixit for fruitful
discussions.
2
2
2
-1
-1
0.00084 L g
s
and H
2
2
2
Keywords: molybdenum carbide • CO conversion •
For calculating apparent activation energies, the conversion of CO
2
heterogeneous catalysis • reverse water-gas shift • potassium
promoter
was measured at 5 different temperatures from 270 °C to 330 °C, while
adjusting the GHSV to maintain conversion well below the equilibrium for
RWGS. For selectivity versus conversion and stability experiments,
GHSV was also adjusted by varying catalyst mass and reactant flow
[1] F. DiMascio, D. R. Hardy, M. K. Lewis, H. D. Willauer, F. Williams, (Ed.:
-1
-1
S. o. t. N. USA), US9303323 B2, 2016.
rates to maintain GHSV between 0.00084 and 0.0042 L g s .
X-Ray Photoelectron Spectroscopy (XPS). X-ray photoelectron
spectroscopy (XPS) measurements were performed over the fresh and
[
2] W. Wang, S. Wang, X. Ma, J. Gong, Chem. Soc. Rev. 2011, 40, 3703-
727.
3] G. Centi, S. Perathoner, Catal. Today 2009, 148, 191-205.
3
[
spent catalysts using
a
commercial XPS system (Thermo
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201700412
FULL PAPER
[
[
[
4] H. D. Willauer, D. R. Hardy, K. R. Schultz, F. W. Williams, J. Renew. and
Sustain. Energ. 2012, 4, 033111.
5] P. Kaiser, R. B. Unde, C. Kern, A. Jess, Chem-Ing-Tech 2013, 85, 489-
[25] J. Gao, Y. Wu, C. Jia, Z. Zhong, F. Gao, Y. Yang, B. Liu, Catal. Comm.
2016, 84, 147-150.
[26] S. Posada-Perez, F. Vines, P. J. Ramirez, A. B. Vidal, J. A. Rodriguez, F.
Illas, Phys. Chem. Chem. Phys. 2014, 16, 14912-14921.
[27] S. Posada-Perez, P. J. Ramirez, R. A. Gutierrez, D. J. Stacchiola, F.
Vines, P. Liu, F. Illas, J. A. Rodriguez, Catal. Sci. Technol. 2016, 6,
6766-6777.
499.
6] D. Chakrabarti, A. de Klerk, V. Prasad, M. K. Gnanamani, W. D. Shafer,
G. Jacobs, D. E. Sparks, B. H. Davis, Ind. Eng. Chem. Res. 2015, 54,
1189-1196.
[
[
[
[
[
[
7] R. W. Dorner, D. R. Hardy, F. W. Williams, H. D. Willauer, Energ.
[28] A. Griboval-Constant, J. M. Giraudon, G. Leclercq, L. Leclercq, Appl.
Catal. A-Gen. 2004, 260, 35-45.
[29] W. Xu, P. J. Ramírez, D. Stacchiola, J. L. Brito, J. A. Rodriguez, Catal.
Lett. 2015, 145, 1365-1373.
[30] F. Solymosi, L. Bugyi, Catal. Lett. 2000, 66, 227-230.
[31] C. Pistonesi, M. E. Pronsato, L. Bugyi, A. Juan, Catal. Today 2012, 181,
102-107.
[32] V. S. Marakatti, D. Mumbaraddi, G. V. Shanbhag, A. B. Halgeri, S. P.
Maradur, RSC Adv. 2015, 5, 93452-93462.
[33] D. J. Pettigrew, D. L. Trimm, N. W. Cant, Catal. Lett. 1994, 28, 313-319.
[34] A. G. Kharaji, A. Shariati, M. Ostadi, J. Nanosci. Nanotechnol. 2014, 14,
6841-6847.
[35] K. Oshikawa, M. Nagai, S. Omi, J. Phys. Chem. B 2001, 105, 9124-9131.
[36] G. Henkelman, A. Arnaldsson, H. Jónsson, Comp. Mater. Sci. 2006, 36,
354-360.
Environ. Sci. 2010, 3, 884-890.
8] R. W. Dorner, D. R. Hardy, F. W. Williams, H. D. Willauer, Appl. Catal.
A-Gen. 2010, 373, 112-121.
9] S. C. Lee, J. S. Kim, W. C. Shin, M. J. Choi, S. J. Choung, J. Mol. Catal.
A-Chem. 2009, 301, 98-105.
10] T. Riedel, M. Claeys, H. Schulz, G. Schaub, S. S. Nam, K. W. Jun, M. J.
Choi, G. Kishan, K. W. Lee, Appl. Catal. A-Gen. 1999, 186, 201-213.
11] H. D. Willauer, R. Ananth, M. T. Olsen, D. M. Drab, D. R. Hardy, F. W.
Williams, J. CO2 Util. 2013, 3–4, 56-64.
12] H. Schulz, M. Claeys, S. Harms, in Studies in Surface Science and
Catalysis, Vol. Volume 107 (Eds.: R. L. E. C. P. N. J. H. S. M. de Pontes,
M. S. Scurrell), Elsevier, 1997, pp. 193-200.
[13] R. Satthawong, N. Koizumi, C. S. Song, P. Prasassarakich, Catal. Today
2
015, 251, 34-40.
14] R. Satthawong, N. Koizumi, C. Song, P. Prasassarakich, Top. Catal.
014, 57, 588-594.
[
[37] W. Tang, E. Sanville, G. Henkelman, J. Phys. Condens. Matter 2009, 21,
2
084204.
[
[
15] M. D. Porosoff, B. Yan, J. G. Chen, Energ. Environ. Sci. 2016, 9, 62-73.
16] O.-S. Joo, K.-D. Jung, I. Moon, A. Y. Rozovskii, G. I. Lin, S.-H. Han, S.-J.
Uhm, Ind. Eng. Chem. Res. 1999, 38, 1808-1812.
[38] N. Austin, B. Butina, G. Mpourmpakis, Prog. Mater. Sci. Mat. Int. 2016,
26, 487-492.
[39] C. Pistonesi, M. E. Pronsato, L. Bugyi, A. Juan, J. Phys. Chem. C 2012,
[
17] S.-W. Park, O.-S. Joo, K.-D. Jung, H. Kim, S.-H. Han, Korean J. Chem.
Eng. 2000, 17, 719-722.
116, 24573-24581.
[40] L. Bugyi, A. Oszkó, F. Solymosi, Surf. Sci. 2000, 461, 177-190.
[41] F. Solymosi, R. Németh, Catal. Lett. 1999, 62, 197-200.
[42] R. Hou, K. Chang, J. G. Chen, T. Wang, Top. Catal. 2015, 58, 240-246.
[43] G. Kresse, J. Furthmüller, Phys. Rev. B 1996, 54, 11169-11186.
[44] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865-
3868.
[18] S. Kattel, W. Yu, X. Yang, B. Yan, Y. Huang, W. Wan, P. Liu, J. G. Chen,
Angew. Chem. Int. Edit. 2016, 128, 8100-8105.
[19] M. D. Porosoff, J. G. Chen, J. Catal. 2013, 301, 30-37.
[20] R. B. Levy, M. Boudart, Science 1973, 181, 547-549.
[21] M. D. Porosoff, S. Kattel, W. Li, P. Liu, J. G. Chen, Chem. Comm. 2015,
5
1, 6988-6991.
[45] S. Grimme, J. Comput. Chem. 2006, 27, 1787-1799.
[46] P. E. Blöchl, Phys. Rev. B 1994, 50, 17953-17979.
[47] J. D. Pack, H. J. Monkhorst, Phys. Rev. B 1977, 16, 1748-1749.
[48] D. Sheppard, R. Terrell, G. Henkelman, J. Chem. Phys. 2008, 128,
134106.
[
[
[
22] A. J. Medford, A. Vojvodic, F. Studt, F. Abild-Pedersen, J. K. Nørskov, J.
Catal. 2012, 290, 108-117.
23] Y. Shi, Y. Yang, Y.-W. Li, H. Jiao, Appl. Catal. A-Gen. 2016, 524, 223-
236.
24] M. D. Porosoff, X. Yang, J. A. Boscoboinik, J. G. Chen, Angew. Chem.
Int. Edit. 2014, 53, 6705-6709.
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201700412
FULL PAPER
Entry for the Table of Contents (Please choose one layout)
Layout 1:
FULL PAPER
Modifying a molybdenum carbide
catalyst supported on gamma-alumina
with potassium (K) improves
Marc D. Porosoff, Jeffrey W. Baldwin, Xi
Peng, Giannis Mpourmpakis and
Heather D. Willauer*
selectivity towards CO and catalyst
stability during reverse water-gas shift.
Over a broad conversion range, the
selectivity to CO is maintained near
Page No. – Page No.
Potassium-promoted molybdenum
carbide as a highly active and selective
catalyst for CO conversion to CO
2
95%.
This article is protected by copyright. All rights reserved.