Hydrogenation of Carbon Dioxide over K-Promoted FeCo Bimetallic Catalysts Prepared from Mixed Metal Oxalates

Article abstract of DOI:10.1002/cctc.201601337

The hydrogenation of carbon dioxide over K-promoted FeCo bimetallic catalysts prepared by sequential oxalate decomposition and carburization of FeCo with CO was studied in a fixed-bed reactor at 240 °C and 1.2 MPa. The initial CO₂ conversion was found to be dependent on K loading, whereas both unpromoted and K-promoted FeCo catalysts (except 90Fe10Co3.0K) exhibited similar levels of CO₂ conversion after a few hours of time on stream. A decarburization study on freshly activated and used FeCo suggests that potassium increases the stability of iron carbides and graphitic carbon under a reducing atmosphere. Also, K addition tends to decrease the hydrogenation function of FeCo bimetallic catalysts and, thus, controls product selectivity. Under similar CO₂ conversions, potassium enhanced acetic acid formation while suppressing ethanol production, which indicates that a common intermediate might be responsible for the changes observed with C₂ oxygenates.

Full text of DOI:10.1002/cctc.201601337

DOI: 10.1002/cctc.201601337
Full Papers
Hydrogenation of Carbon Dioxide over K-Promoted FeCo
Bimetallic Catalysts Prepared from Mixed Metal Oxalates
Muthu Kumaran Gnanamani,[a] Hussein H. Hamdeh,[b] Gary Jacobs,[a] Wilson D. Shafer,[a]
Shelley D. Hopps,[a] Gerald A. Thomas,[a] and Burtron H. Davis*[a]
The hydrogenation of carbon dioxide over K-promoted FeCo
bimetallic catalysts prepared by sequential oxalate decomposi-
tion and carburization of FeCo with CO was studied in a fixed-
bed reactor at 2408C and 1.2 MPa. The initial CO2 conversion
was found to be dependent on K loading, whereas both un-
promoted and K-promoted FeCo catalysts (except
90Fe10Co3.0K) exhibited similar levels of CO2 conversion after
a few hours of time on stream. A decarburization study on
freshly activated and used FeCo suggests that potassium in-
creases the stability of iron carbides and graphitic carbon
under a reducing atmosphere. Also, K addition tends to de-
crease the hydrogenation function of FeCo bimetallic catalysts
and, thus, controls product selectivity. Under similar CO2 con-
versions, potassium enhanced acetic acid formation while sup-
pressing ethanol production, which indicates that a common
intermediate might be responsible for the changes observed
with C2 oxygenates.
Introduction
The increase in CO2 emissions with industrialization and the re-
sulting greenhouse effect have given rise to ever-increasing
concerns about the disposal of CO2. The chemical transforma-
tion of CO2 into fuels and chemicals is a topic of interest for
many.[1–5] Hydrogenation of carbon dioxide to form higher hy-
drocarbons is one way of utilizing CO2.[6–8] Various catalysts
based on Fe,[9] Co,[10] Ni,[11] and Ru[12] metals were tested for
this reaction, and all catalysts, except for those based on iron,
produced solely methane.[13] It seems likely that the transfor-
mation of carbon dioxide into higher hydrocarbons proceeds
mainly by reverse water–gas shift (RWGS), which makes CO
from CO2, with the carbon monoxide that is produced reacting
subsequently by Fischer–Tropsch (FT) synthesis to produce
mostly hydrocarbons. Fe-based FT catalysts possess WGS/
RGWS activity under typical FT-synthesis conditions, which
makes iron a suitable candidate for CO2 conversion.[14]
catalysts yielded a small amount of oil from CO2 hydrogena-
tion; however, the authors found only methane from an alkali-
promoted, Kieselguhr-supported FT catalyst. In another study,
Cubeiro et al.[23] obtained higher olefin selectivity for C2–C4 hy-
drocarbons with K-promoted Fe catalysts obtained by thermal
decomposition of iron citrate. Fierro and co-workers[24] have
shown that alkali addition promotes long-chain hydrocarbon
products for both CO and CO2 hydrogenation processes. The
authors concluded that CO2 conversion was favored by alkali
addition, especially addition of potassium, as a result of the
promotion of the water–gas shift reaction.
In an early study, Stowe and Russell[25] used cobalt, iron, and
some of their alloys as catalysts for the hydrogenation of
carbon dioxide. Satthawong et al.[26] performed CO2 hydroge-
nation over K-promoted alumina-supported FeCo bimetallic
catalysts, and these authors found that the selectivity for light
olefins increased with K addition. They correlated this with
weaker adsorption of hydrogen on the metal surfaces. In
a recent study, we showed that FeCo bimetallic catalysts offer
advantages for CO2 hydrogenation over monometallic catalysts
(Fe or Co) for the production of higher C2 + hydrocarbons.[27]
In a continuation of this vein of research, in this work, the
effect of potassium on FeCo-catalyzed CO2 hydrogenation was
explored. The FeCo bimetallic catalyst was prepared by decom-
position of metal oxalates and carburization with CO. The ef-
fects of K content on the carburization rate of FeCo and cata-
lyst performance parameters (that is, stability, activity, and se-
lectivity) for CO2 hydrogenation were investigated.
Iron FT catalysts are often promoted by alkali metals such as
potassium[15,16] and structural promoters, for example, SiO2 or
Al2O3,[17] as well as reduction promoters such as copper;[18] pro-
motion effects include higher conversion and C5 + selectivity
for CO hydrogenation. Similarly, a promoter effect of potassium
for CO2 hydrogenation has also been reported.[19–21] An early
work from Kꢀster[22] showed that alkali-promoted Fe and Co
[a] Dr. M. K. Gnanamani, Dr. G. Jacobs, W. D. Shafer, S. D. Hopps,
Dr. G. A. Thomas, Prof. B. H. Davis
Center for Applied Energy Research
University of Kentucky
2540 Research Park Drive, Lexington, Kentucky 40511 (USA)
Fax: (+1)859-257-0302
E-mail: burtron.davis@uky.edu
Results and Discussion
[b] Prof. H. H. Hamdeh
Department of Physics
Wichita State University
Wichita, Kansas 67260 (USA)
The detailed elemental analyses of K-promoted FeCo bimetallic
oxalates are shown in Table 1. The ratios of Co/Fe and K/(Fe+
ChemCatChem 2017, 9, 1 – 11
1
ꢁ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
is epsilon carbide (e-Fe2.2C), with the remaining 22% of iron in
the magnetite form. The concentration of Hꢂgg carbide was
found to decrease with the increase in K content in the cata-
lyst, whereas the epsilon carbide content increased slightly.
The magnetite phase slightly increased with increasing K con-
tent. Cubeiro et al.[23] obtained a similar trend over thermally
decomposed iron citrate promoted with potassium. The au-
thors observed a slight decrease in iron carbide content with
an increase in K loading.
Table 1. Elemental analysis by ICP-OES.
Catalyst
Elemental composition [wt%] Co/Fe[a]
K/(Fe+Co)[a]
Fe
Co
K
90Fe10Co0K
86.2
10.3
9.7
10.0
9.8
0.0
0.3
0.7
1.2
1.6
0.11 (0.11) 0.0 (0)
90Fe10Co0.5K 81.4
90Fe10Co1.0K 74.6
90Fe10Co2.0K 77.9
90Fe10Co3.0K 79.5
0.11 (0.11) 0.005 (0.005)
0.13 (0.11) 0.012 (0.01)
0.12 (0.11) 0.020 (0.02)
0.10 (0.11) 0.026 (0.03)
8.7
[a] Theoretical ratios given in parenthesis.
The X-ray diffraction patterns for the freshly carburized cata-
lysts are shown in Figure 2. Diffraction patterns corresponding
to iron carbide and magnetite phases were observed for all
catalysts. The presence of the magnetite phase is slightly dom-
inant with higher K contents. The XRD pattern corresponding
to the low-temperature iron carbide is usually either composed
of very small crystallites or is very distorted, which results in
the XRD patterns looking broader in nature. Similarly, in this
case, FeCo bimetallic oxalates decomposed to ferrites, which
Co) for various catalysts obtained by ICP-OES match very well
with those of the theoretical values according to the formula
of Fe0.9Co0.1(C2O4)·2H2O.
Reference [28] provides details about the Mçssbauer fitting
procedure. Mçssbauer spectroscopy results (Table 2 and
Figure 1) for the unpromoted catalyst (90Fe10Co0K) indicate
that 56% of iron is present as Hꢂgg carbide (c-Fe5C2) and 22%
Table 2. Phase compositions of Fe for freshly carburized and used cata-
lysts. Results obtained by using Mçssbauer spectroscopy measured at T=
ꢀ2538C.
Catalyst
Fresh [% Fe][a]
Used [% Fe][a]
Fe3O4 c-Fe5C2 e’-Fe2.2
C
Fe3O4 c-Fe5C2 e’-Fe2.2C
90Fe10Co0K
90Fe10Co0.5K 22
90Fe10Co1K
90Fe10Co2K
90Fe10Co3K
22
56
51
44
46
44
22
27
26
26
25
39
36
24
28
32
45
39
44
42
34
16
25
32
30
34
30
28
31
[a] Uncertainty in these numbers varies by 3–5%.
Figure 1. Mçssbauer spectra of fresh (left), and used (right) 90Fe10Co(x)K
catalysts (x=0.0, 0.5, 1.0, 2.0, and 3.0) performed at T=ꢀ2538C after cata-
lyst pretreatment in a CO flow at T=2508C and P=1.2 MPa The fitted
curves are shown as solid lines: black, total spectra; red, Fe3O4; blue, c-
Fe5C2; green, e’-Fe2.2C.
Figure 2. XRD patterns of freshly activated samples of 90Fe10Co(x)K in CO at
T=2508C and 0.10 MPa pressure (top) and of used catalysts (bottom).
&
ChemCatChem 2017, 9, 1 – 11
www.chemcatchem.org
2
ꢁ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
then carburized in CO at T=2508C; notably, this temperature
is approximately 208C less than the typical carburization tem-
perature of a-Fe2O3, which is 2708C. The peaks corresponding
to the iron carbide phase appear very broad and are not very
intense.
(a), amorphous (b), iron carbides (g), and graphitic (d)) based
on the temperature at which the species becomes hydrogenat-
ed. The authors suggested that the atomic carbon species ex-
hibited higher FT activity, whereas amorphous carbon species
showed only moderate activity; finally, semiordered graphitic
carbon exhibited low reactivity. In our case, potassium shifts
the low-temperature peak from T=415 to 5298C, along with
a significant drop in the intensities. At the same time, multiple
peaks appear in the higher temperature region, which indi-
cates that the strength of these carbides could vary depending
upon the nature of the carbides and graphitic carbon present,
as well as their interaction with potassium.
The carbon deposits and carbides formed under our activa-
tion conditions were investigated by temperature-pro-
grammed decarburization. This method gives information
about the relative amount and stability of various surface
carbon species formed after carburization and during the reac-
tion. Figure 3 shows the decarburization profiles of the unpro-
moted and K-promoted 90Fe10Co catalysts after activation
(top) and after the reaction (bottom). The decarburization pro-
file for the unpromoted carburized catalyst has two peaks; one
is centered at T=4158C and another at T=4758C. K addition
was found to shift the peak maxima toward higher tempera-
ture, and the intensities corresponding to the low-temperature
peaks decreased. This indicates that potassium increases the
stability of surface carbides formed on the catalyst. Eliason and
Bartholomew[29] investigated various carbonaceous surface spe-
cies and bulk iron carbides formed under FT-synthesis condi-
tions by using temperature-programmed hydrogenation. The
authors classified mainly four different carbon species (carbidic
HRTEM images of the freshly carburized/passivated samples
of unpromoted and K-promoted FeCo bimetallic catalysts are
shown in Figure 4. TEM images at lower magnification (Fig-
ure 4A–C) shows iron particles uniformly distributed for all
three catalysts (90Fe10Co0K, 90Fe10Co1K, and 90Fe10Co3K)
with the particle diameter of iron varying from 20 to 30 nm. At
higher magnification, all three catalysts displayed Hꢂgg carbide
surrounded by carbon fringes. The thickness of the carbon
fringes increased with increasing K content in 90Fe10Co. At
the same time, metallic Co or FeCo bimetallic compounds
were observed to have imperfect crystalline structures (that is,
short-range order). This shows that potassium increases addi-
tional carbon deposition over Fe FT catalysts in the form of
semicrystalline carbon, which could control the stability of the
Fe catalysts during the reaction.
The conversions of CO2 and H2 over unpromoted and vari-
ous K-promoted FeCo bimetallic catalysts are shown in
Figure 5. The initial CO2 conversions were higher in the case of
90Fe10Co0K and 90Fe10Co0.5K catalysts and then dropped
with further increases in K content (1.0, 2.0, and 3.0). The varia-
tions in CO2 conversion diminished with time on stream,
except in the case of 90Fe10Co3.0K. Regarding H2 conversions,
both 90Fe10Co0K and 90Fe10Co0.5K catalysts showed similar
activity, and the conversions of H2 for these two catalysts were
higher than those with the other K-containing FeCo bimetallic
catalysts. The H2 conversions of the various catalysts followed
in the order: 90Fe10Co0K=90Fe10Co0.5K>90Fe10Co1.0K>
90Fe10Co2.0K>90Fe10Co3.0K. The trend indicates that potas-
sium improves CO2 conversion of FeCo bimetallic catalysts
only to a certain extent and, at the same time, potassium di-
minishes the hydrogenation activity of the catalyst.
The FT product distributions (CO free) obtained from the
CO2 hydrogenation reaction over unpromoted and various K-
promoted FeCo bimetallic catalysts are shown in Figure 6. The
selectivity to methane decreased with addition of potassium
to the FeCo bimetallic catalyst. For 90Fe10Co0K, methane se-
lectivity remained at approximately 62%, whereas with
90Fe10Co0.5K, the selectivity decreased to 50%, with further
decreases to approximately 38% for 90Fe10Co1.0K, 25% for
90Fe10Co2.0K, and 22% for 90Fe10Co3.0K. Furthermore, C2 +
hydrocarbon selectivity was found to increase with increasing
K content. The selectivity to oxygenates increased proportion-
ally with the K content in the catalyst. Interestingly, the selec-
tivity to oxygenates for the highly K-loaded catalyst
(90Fe10Co3.0K) was three times higher (ꢁ15–17 C%, CO free)
Figure 3. Temperature-programmed decarburization profiles of various fresh-
ly carburized (top) and used (bottom) catalysts.
ChemCatChem 2017, 9, 1 – 11
www.chemcatchem.org
3
ꢁ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Figure 4. HRTEM images of the carburized sample of A) 90Fe10Co0K, B) 90Fe10Co1K, and C) 90Fe10Co3K. Images of the respective catalyst particles (D, E, F)
taken at a higher magnification.
In the present work, hydrogenation of CO2 on Fe may proceed
than that for the unpromoted catalyst (ꢁ5 C%, CO free). The
by reverse water–gas shift. The presence of cobalt also influen-
alkene/(alkene+alkane) fractions for C3 and C4 hydrocarbon
ces the product selectivity of the FeCo bimetallic catalyst to
products plotted against time on stream for various catalysts
some extent, especially in terms of light hydrocarbon forma-
are shown in Figure 7. The promoting effect of potassium in
tion. The product selectivity was compared for unpromoted
the 90Fe10Co0.5K catalyst is manifested in the C3 and C4
and K-promoted FeCo bimetallic catalysts at a similar level of
alkene ratios, which are twice the value of those with the un-
CO2 conversion, as indicated in Table 3. The selectivity to C1 de-
promoted catalyst. In the case of the C3 and C4 hydrocarbon
creased from 54.3 to 11.5 (C, mol%) upon an increase in the K/
fraction, the alkene ratios for 90Fe10Co1.0K were double those
(Fe+Co) ratio from 0 to 0.03. On the other hand, the lower hy-
of 90Fe10Co0.5K. However, further increases in K loading did
drocarbon (C2–C4) selectivity for the 90Fe10Co0.5K catalyst was
not result in any further changes to the alkene selectivity.
higher than that with 90Fe10Co0K; however, further increases
Some authors have reported that olefins are formed as a pri-
in potassium caused a drop in the selectivity to C2–C4 hydro-
mary product over paraffins and undergo hydrogenation in
carbons. We could only manage to quantify C5 + hydrocarbons
a secondary step. In our system, the hydrogenation activity of
for 90Fe10Co3.0K; however, their presence with other catalysts
the catalysts decreased with increasing K content in the cata-
tested in this series cannot be ruled out because they may be
lyst and, therefore, olefins formed in the reaction did not un-
present but below the detection limit. Surprisingly, CO selectiv-
dergo further hydrogenation to form higher olefins. Similar
ity for the catalysts increased from 15.1 to 55.7 (C, mol%) with
conclusions were drawn by others.[30]
increasing K/(Fe+Co) mole ratios from 0 to 0.03. Some authors
Hydrogenation of CO2 involves two different pathways:
reported that the selectivity for CO decreases with increasing K
direct hydrogenation [Eq. (1)] and reverse water–gas shift
content for CO2 hydrogenation over alumina-supported Fe cat-
[Eq. (2)], whereby CO produced is hydrogenated through the
alysts.[31] However, Martinelli et al.[32] obtained higher selectivity
FT reaction.
for CO with increasing K loading for the hydrogenation of CO2
over traditional Fe–Zn–Cu–K catalysts. Riedel et al.[33] indicated
n CO2 þ 3n H2 ! ðCH2Þn þ 2n H2O
CO2 þ H2 $ CO þ H2O
ð1Þ
ð2aÞ
ð2bÞ
that potassium yields higher CO selectivity for CO2 hydrogena-
tion over Fe/Al2O3 catalysts because of an enhancement in CO
shift activity. In the present work, though the conversion of
CO2 obtained remains more or less the same (except with
90Fe10Co3.0K) for catalysts with different K content, the hydro-
CO þ 2 H2 ! ðCH2Þn þ H2O
&
ChemCatChem 2017, 9, 1 – 11
www.chemcatchem.org
4
ꢁ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Figure 5. Conversions of CO2 and H2 over carburized samples of
90Fe10Co(x)K with varying K content (x=0, 0.5, 1.0, 2.0, and 3.0). Reaction
conditions: T=2408C, P=1.2 MPa H2/CO2 =3.0, space velocity
(SV)=1.5 nLhꢀ1 gꢀ1 of metal oxalates.
genation activities varied from 17.6 to 9.6% with increasing K/
(Fe+Co) ratios from 0 to 0.02. This suggests that potassium
suppresses the hydrogenation function of the catalyst, proba-
bly by depositing more carbon on the surface of the iron, as
evidenced from H2 temperature-programmed decarburization.
Also, if hydrogen tends to donate electrons to iron on adsorp-
tion, the presence of potassium would be expected to weaken
the iron–hydrogen bond and this could contribute to suppres-
sion of the dissociation of the H2 molecule. Therefore, potassi-
um can function in two ways on iron with regard to CO and
CO2 activation. It could facilitate the dissociative adsorption of
CO and CO2 and, thereby, increase carbon deposition on the
iron. On the other hand, the basic nature of potassium may
decrease the adsorption strength of H2 on the surface of the
iron.
Figure 6. Product distributions obtained for the hydrogenation of carbon di-
oxide over carburized samples of 90Fe10Co(x)K with varying K content
(x=0, 0.5, 1.0, 2.0, and 3.0). Reaction conditions: T=2408C, P=1.2 MPa, H2/
CO2 =3.0, SV=1.5 nLhꢀ1 gꢀ1 of metal oxalates.
gen conversion but then leveled off with further increases in
hydrogen conversion. These findings suggest that acetic acid,
acetaldehyde, and ethanol may originate from the same inter-
mediate (Scheme 1). Davis[34] documented the formation of
surface acetate species over an Fe catalyst, in which CO inser-
tion onto MꢀOCH3 species may be responsible for the acetate
formation. In that perspective, CO2 can only initiate chain
growth and does not take part in chain propagation, as shown
in Scheme 2.[35]
The rates of formation of C2 oxygenates (ethanol, acetalde-
hyde, and acetic acid) plotted against conversion rate of H2 for
FeCo bimetallic catalysts at three different K loadings are
shown in Figure 8. The rate of formation of acetic acid de-
creased with increasing hydrogen conversion, and acetalde-
hyde formation followed a similar trend. The rate of ethanol
formed from CO2 initially increased with an increase in hydro-
ChemCatChem 2017, 9, 1 – 11
www.chemcatchem.org
5
ꢁ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Figure 8. Rates of C2 oxygenates (ethanol, acetic acid, and acetaldehyde)
formed from CO2 as a function of hydrogenation rates of CO2 for carburized
K-promoted FeCo bimetallic catalysts. Reaction conditions: T=2408C,
P=1.2 MPa, H2/CO2 =3.0, SV=1.5 nLhꢀ1 gꢀ1 metal oxalates. The values in pa-
rentheses indicate the K/(Fe+Co) ratio.
On the other hand, Dry[36] proposed two pathways for acetic
acid formation from CO or CO2. In the first, CO2 terminates FT
chain growth, as shown in Scheme 3A. In another pathway, al-
dehyde, alcohol, or acid formation depends on whether H2 or
H2O is involved in chain termination (Scheme 3B).
The hydrogenation activity of the Fe catalyst likely plays an
important role in controlling the formation of oxygenates, par-
ticularly acids from CO or CO2. In the current context, a signifi-
cant drop in the conversion of hydrogen was observed with K
addition. This is consistent with our earlier proposed mecha-
nism on FeCo bimetallic catalysts,[27] in which potassium may
take part in the catalytic cycle for the formation of acetic acid
from CO2.
Figure 7. Olefin-to-paraffin ratios of C3 and C4 hydrocarbons obtained with
time on stream for the hydrogenation of carbon dioxide over carburized
samples of 90Fe10Co(x)K with varying K content (x=0, 0.5, 1.0, 2.0, and 3.0).
Reaction conditions: T=2408C, P=1.2 MPa, H2/CO2 =3.0, SV=1.5 nLhꢀ1 gꢀ1
metal oxalates.
Many have agreed that the deactivation of iron occurs
under Fischer–Tropsch synthesis conditions, mainly as a result
of the loss of active catalyst surface area because of accumula-
tion of carbonaceous species on iron carbide surfaces and/or
the loss of iron carbides (e-Fe2.2C, c-Fe5C2, q-Fe3C) through oxi-
dation. It has been reported that losses in the catalytic activity
of Fe catalysts for CO2 hydrogenation occur mainly by oxida-
tion of the active iron carbide phase to magnetite.[37] In the
current work, we examined the used Fe catalysts (after passiva-
tion under flowing 1% O2 in N2 at room temperature for a few
hours) by XRD, Mçssbauer spectroscopy, and decarburization
techniques. The XRD patterns of the used catalysts are shown
in Figure 2 (bottom). Diffraction lines corresponding to the
magnetite phase were observed very clearly for the unpromot-
ed catalyst. On the other hand, the diffraction lines for K-pro-
moted FeCo catalysts were less pronounced.
Table 3. Hydrogenation of carbon dioxide for carburized samples of
90Fe10Co(x)K (x=0, 0.5, 1.0, 2.0, and 3.0).[a]
K/(Fe+Co) Molar ratio
0
0.005
0.01
0.02
0.03
77.0
TOS [h]
conversions [%]:
68.6
118.7
139.5
95.0
carbon dioxide
hydrogen
14.5
17.6
14.2
16.0
14.5
12.4
14.0
9.6
8.0
10.5
Product distribution [C, mol%]:
C1
C2–C4
C5 +
CO
CH3OH
C2H5OH
CH3CHO
CH3CO2H
C3H7OH
CH3CH2CO2H
54.3
26.7
–
15.1
2.7
1.1
–
–
–
–
36.5
32.6
–
26.2
1.0
3.7
–
–
–
–
20.3
26.9
–
45.5
0.7
5.9
0.3
0.4
–
13.0
22.7
–
57.9
0.1
2.9
0.6
2.4
–
11.5
19.1
7.6
55.7
0.2
2.6
0.7
2.2
–
The Mçssbauer spectrum of the used unpromoted and K-
promoted FeCo catalysts are shown in Figure 1 (right). The cor-
responding quantitative estimation of the various Fe carbide
phases present in each sample is shown in Table 2. After reac-
tion, the percentage of magnetite formed had increased from
22 to 39% for the unpromoted FeCo catalyst and the c and e-
iron carbides had decreased from 56 to 45% and 22 to 16%,
–
0.4
0.4
[a] Reaction conditions: T=2408C, P=1.2 MPa, H2/CO2 =3.0,
SV=1.5 nLhꢀ1 gꢀ1 of metal oxalates.
&
ChemCatChem 2017, 9, 1 – 11
www.chemcatchem.org
6
ꢁ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
between the fresh and used catalyst was the e-Fe
carbide (Fe2.2C) content. In the used catalyst, the
amount of e-Fe carbide increased with increasing K
content, whereas the e-Fe2.2C content for the freshly
activated catalysts remained more or less the same;
this indicates that potassium enhanced the transfor-
mation of highly active c-Fe5C2 carbide into less
active e-Fe2.2C carbide. A similar result was reported
by our group with CO hydrogenation over precipi-
tated Fe catalysts.[38] The origin of this transforma-
tion is not well understood.
The decarburization of used Fe catalysts was per-
formed to know what kinds of carbon species re-
mained in the used catalyst and to make an attempt
to elucidate their individual roles in the reaction.
Bartholomew and co-workers[29,39] made a detailed
study of the temperature-programmed hydrogena-
tion of carbonaceous species on Fe FT catalysts
during steady-state synthesis. The authors suggested
that the deactivation of Fe catalysts in FT synthesis
under the conditions of their study could be ex-
plained by two factors: first, the transformation of
high-activity Hꢂgg carbide to less active epsilon car-
bide, and second, the formation of highly ordered
graphite on the surface of the catalyst. As shown in
Figure 3, in the current work, most of the low-tem-
perature carbon species (T=325–5008C) were con-
sumed by the reaction and only those catalysts with
moderate to higher levels of potassium had signifi-
cant levels of amorphous and atomic carbon species.
Moreover, the used unpromoted FeCo catalyst ex-
hibited carbon species for which the reduction tem-
perature was greater than 5508C, which indicates
Scheme 1. A plausible pathway for the formation of acetic acid, acetaldehyde, and etha-
nol from acetate intermediate on Fe catalysts.
either that atomic carbon or Hꢂgg carbide was stabilized in
the presence of hydrocarbon products or that amorphous
moderately ordered graphitic carbon formed during the course
of the reaction.
The present data reveal that potassium protects iron car-
bides (active form of iron for CO and CO2 hydrogenation) from
oxidation to magnetite (not a significantly active phase for the
reaction) during the hydrogenation of carbon dioxide. Howev-
er, there is an optimum K loading for the Fe catalysts, in this
case, 1.0 K, to achieve an optimal balance between activity and
stability for the conversion of CO2 to higher hydrocarbons. For
instance, the 90Fe10Co3.0K catalyst displayed greater stability
in preserving the active Hꢂgg carbide phase of iron during the
reaction and, at the same time, the content of the less active
epsilon iron carbide (e-Fe2.2C) increases as well. Therefore, the
catalyst activity declined over time, as shown in Figure 5. It
seems that the unpromoted FeCo bimetallic catalyst suffers
from deactivation, likely as a result of the loss of the active
Hꢂgg carbide phase to magnetite during the reaction, whereas
highly loaded K-promoted FeCo catalysts (such as
90Fe10C03.0K) exhibit lower initial activity and experience
more rapid deactivation, probably because of the lower hydro-
genation activity caused by the deposition of highly ordered
graphite on the Fe surface.
Scheme 2. Formation of surface acetate species over Fe catalyst through CO
insertion onto MꢀOCH3 species.[34]
respectively. Similar results were observed with 90Fe10Co0.5K.
On the other hand, all other catalysts (90Fe10Co1.0K,
90Fe10Co2.0K, and 90Fe10Co3.0K) showed that either the per-
centage of magnetite remains the same or decreases slightly
before and after reaction. Another noticeable difference found
ChemCatChem 2017, 9, 1 – 11
www.chemcatchem.org
7
ꢁ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
with a decrease in the hydrogenation function of
the Fe catalyst, likely prevents ethanol formation
from the intermediate.
Experimental Section
Catalyst preparation
Iron(II) sulfate heptahydrate 98%, cobalt(II) nitrate hexa-
hydrate 98%, potassium nitrate (99.9%), and oxalic acid
dihydrate 99% were purchased from Sigma–Aldrich Co.
and used without further purification. The preparation
of
cobalt-containing
iron
bimetallic
oxalates
(90Fe:10Co) is described elsewhere.[27] Typically, oxalic
acid (37.821 g, 0.3 mol) was dissolved in deionized
water (250 mL) with constant stirring at T=358C. In
a separate flask, iron sulfate (50.549 g, 0.18 mol) and
cobalt nitrate (5.939 g, 0.02 mol) were dissolved in de-
ionized water (750 mL) at T=358C. The metal-contain-
ing solution was added dropwise to the aqueous oxalic
acid. Precipitation was virtually instantaneous, and the
mixture was vigorously stirred at T=358C for 1 h. The
solution was then allowed to cool to room temperature
with continued stirring. After removal of the superna-
tant, the precipitated metal oxalate complexes were fil-
tered and washed three times with deionized water
(1 L). The resulting solid was dried at T=1008C in air in
an oven for 24 h. Potassium was added to obtain K/
Scheme 3. A) Acetic acid formation from CO or CO2 over Fe catalyst proposed by Dry.[35]
Conclusions
((Fe+Co) atomic ratios of 0.005, 0.01, 0.02, and 0.03 by using
a slurry-phase impregnation technique. The catalyst was then dried
overnight at T=1008C.
The hydrogenation of carbon dioxide over K-promoted FeCo
bimetallic catalysts was investigated by using a fixed-bed reac-
tor at moderate temperature and pressure. The carburized
form of the Fe catalyst was obtained from an oxalate precursor
after decomposition and carburization with CO. The decarburi-
zation of the freshly activated FeCo catalysts suggests that po-
tassium decreases the low-temperature surface carbon species
and increases bulk iron carbides and graphitic carbon formed
on the Fe surface during activation. The addition of potassium
to FeCo up to 1.0 K was found to enhance the activity and sta-
bility of iron carbide for the hydrogenation of carbon dioxide.
At the same time, the hydrogenation function of the catalyst is
suppressed with increasing K levels in the catalyst, which re-
sults in a higher selectivity to CO (that is, the product of the
RWGS reaction) and oxygenates. Mçssbauer spectroscopy anal-
ysis of the freshly carburized FeCo bimetallic catalysts suggests
that the unpromoted FeCo bimetallic catalyst oxidizes to
a greater extent than the K-promoted catalyst. The selectivity
to C2–C4 hydrocarbons increases from 26.7 to 32.6% with an
increase in the K level from 0 to 0.05 [90Fe10Co(x)K], whereas
further increases in potassium were found to decrease the se-
lectivity to 26.9, 22.7, and 19.1% for 0.1 K, 0.2 K, and 0.3 K, re-
spectively. As expected, the olefin ratios for C3 and C4 hydro-
carbons increased monotonically with increases in K level in
the catalyst. The selectivity to oxygenates for FeCo bimetallic
catalysts increases (to a maximum of 18 C%, CO free) with in-
creases in the K content. Potassium shifts the product distribu-
tion of C2 oxygenates from CO2 towards acetic acid. A synergis-
tic effect by potassium (that is, possible involvement in the
catalytic cycle, as proposed in our previous work[27]), along
Catalyst activation
The FeCo bimetallic oxalates (2.5 g) with and without K were de-
composed in flowing nitrogen (100 sccm) at T=4008C for 2 h and
subsequently treated with CO (10%CO in N2) at T=2508C for 24 h.
The resulting FeCo bimetallic carbides were tested for CO2 hydro-
genation or passivated under a flow of 1% O2/N2 at T=258C for
4 h for the purpose of characterization.
Characterization
Elemental analysis
The compositions of FeCo bimetallic oxalate complexes were de-
termined by inductively coupled plasma optical emission spectros-
copy (ICP-OES) with a Varian 720-ES analyzer. The materials were
dissolved in a perchloric/nitric acid mixture, and the emission spec-
tra of dissolved species (Fe, Co, and K) were compared to those of
a series of standard solutions of known concentrations.
X-Ray diffraction (XRD)
Powder X-ray diffractograms of the freshly CO-activated and used
catalyst samples were recorded after passivation in 1% O2 in He at
room temperature by using a Philips X’Pert diffractometer with
monochromatic CuKa radiation (l=1.5418). XRD scans were taken
over a 2q range of 10–908. The scanning step was 0.01, the scan
speed was 0.0025 sꢀ1, and the scan time was 4 s.
&
ChemCatChem 2017, 9, 1 – 11
www.chemcatchem.org
8
ꢁ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
was analyzed by using an SRI (Torrance, CA) GC-TCD-8610C appara-
tus with a 6’ Poropak-Q stainless steel packed column. A 5973N
MSD instrument coupled to a 6890 gas chromatograph from Agi-
lent was employed for qualitative analysis of various oxygenated
compounds. The conversion and selectivity reaction parameters
are defined as in Equations (3) and 4, in which nCO and nCO are
the numbers of moles of CO2 fed and not-consumed, respectively.
The selectivity is defined as the percentage of moles of CO2 con-
sumed to form a particular Cn product (hydrocarbon, CO, or oxy-
genate), normalized by the amount of CO2 consumed.
Temperature-programmed decarburization
The carburized/passivated catalyst (about 0.2 g) was positioned be-
tween a bed of quartz wool, and an H2:He (1:3) gas mixture was in-
troduced at 100 ccminꢀ1. The temperature of the catalyst bed was
increased to 8508C with a ramp of 2.38Cminꢀ1. The evolved prod-
uct was methane, and it was continuously detected by a flame ion-
ization detector (SRI-GC), which gave the temperature-pro-
grammed decarburization profile. A similar procedure was followed
to that described above to study the decarburization of the used
catalysts for the purpose of comparison.
2 in
2 out
n
inꢀn
CO2
CO2 out
conversion ¼ 100 ꢃ
selectivity ¼ 100 ꢃ
ð3Þ
ð4Þ
n
CO2 in
High-resolution transmission electron microscopy (HRTEM)
nproduct out carbon number
inꢀn
In-depth analysis of individual particles of unpromoted and K-pro-
moted FeCo bimetallic catalysts in carburized form was obtained
with a field-emission analytical transmission electron microscope
(JEOL JEM-2010F) operated at an accelerating voltage of 200 kV.
HRTEM images were recorded under optimal focus conditions at
typical magnifications of 100–500 K. The electron beam had
a point-to-point resolution of 0.5 nm. Gatan Digital Micrograph
software was used for image processing. Samples were prepared
on lacey-carbon copper grids and dispersed as powders.
n
CO2
CO2 out
Acknowledgements
This research was conducted with financial support from the
Commonwealth of Kentucky.
Keywords: bimetallic
hydrogenation · cobalt · iron · potassium
catalysts
·
carbon
dioxide
·
55Fe Mçssbauer spectroscopy
55Fe Mçssbauer spectra were collected in transmission mode with
a standard constant-acceleration spectrometer (MS-1200, Ranger
Scientific). A radiation source of 50 mCi 57Co in Rh matrix was used,
and the spectra were obtained with a gas detector. To transport
each sample for Mçssbauer spectroscopy, each sample was careful-
ly fixed into wax inside a glove box after activation. All samples
were investigated at low temperature (ꢀ2538C) by using a closed-
[1] G. A. Olah, Angew. Chem. Int. Ed. 2005, 44, 2636–2639; Angew. Chem.
2005, 117, 2692–2696.
[2] G. Centi, S. Perathoner, Catal. Today 2009, 148, 191–205.
[3] J. J. Spivey, A. Egbebi, Chem. Soc. Rev. 2007, 36, 1514–1528.
[4] V. Subramani, S. K. Gangwal, Energy Fuels 2008, 22, 814–839.
[5] Y. Q. Zhang, G. Jacobs, D. E. Sparks, M. E. Dry, B. H. Davis, Catal. Today
2002, 71, 411–418.
[6] D. E. Dorner, D. R. Hardy, F. W. Williams, B. H. Davis, H. D. Willauer, Energy
Fuels 2009, 23, 4190–4195.
[7] P. S. Sai Prasad, J. W. Bae, K. W. Jun, K. W. Lee, Catal. Surv. Asia 2008, 12,
170–183.
cycle refrigerator, typically over a velocity range of ꢂ10 mmsꢀ1
.
Structure analysis of each sample was preformed by least-squares
fitting of the Mçssbauer spectra to a summation of hyperfine sex-
tets. Details about the least-squares fitting procedures are de-
scribed elsewhere.[28]
[8] R. Satthawong, N. Koizumi, C. Song, P. Prasassarakich, Top. Catal. 2014,
57, 588–594.
[9] M. K. Gnanamani, G. Jacobs, H. H. Hamdeh, W. D. Shafer, B. H. Davis,
Catal. Today 2013, 207, 50–56.
CO2 hydrogenation reaction
[10] G. D. Weatherbee, C. H. Bartholomew, J. Catal. 1984, 87, 352–362.
[11] J. L. Falconer, A. E. Zagli, J. Catal. 1980, 62, 280–285.
[12] F. Wang, S. He, H. Chen, B. Wang, L. Zheng, M. Wei, D. G. Evans, X.
Duan, J. Am. Chem. Soc. 2016, 138, 6298–6305.
[13] T. Riedel, G. Schaub, Top. Catal. 2003, 26, 145–156.
[14] S. Saeidi, N. A. S. Amin, M. R. Rahimpour, J. CO2 Util. 2014, 5, 66–81.
[15] M. E. Dry, T. Shingles, L. J. Boshoff, G. J. Oosthuizen, J. Catal. 1969, 15,
190–199.
[16] H. Zhang, H. Ma, H. Zhang, W. Ying, D. Fang, Catal. Lett. 2012, 142,
131–137.
[17] M. E. Dry in Catalysis: Science and Technolog, Vol. 1 (Eds: J. R. Anderson,
M. Boudart), Springer-Verlag, Germany, 1981, 159–255.
[18] V. R. R. Pendyala, G. Jacobs, M. K. Gnanamani, Y. Hu, A. MacLennan, B. H.
Davis, Appl. Catal. A 2015, 495, 45–53.
[19] P. H. Choi, K.-W. Jun, S.-J. Lee, M.-J. Choi, K.-W. Lee, Catal. Lett. 1996, 40,
115 – 118.
[20] N. Fischer, R. Henkel, B. Hettel, M. Iglesias, G. Schaub, M. Claeys, Catal.
Lett. 2016, 146, 509–517.
[21] Z. You, W. Deng, Q. Zhang, Y. Wang, Chin. J. Catal. 2013, 34, 956–963.
[22] H. Kuster, Brennst.-Chem. 1936, 17, 221–228.
[23] M. L. Cubeiro, H. Morales, M. R. Goldwasser, M. J. P. -Zurita, F. G. -Jime-
nez, React. Kinet. Catal. Lett. 2000, 69, 259–264.
The CO2 hydrogenation reaction was performed by using a fixed-
bed reactor (stainless steel with a length of 17 cm and an inside di-
ameter of 1.6 cm). The reaction was performed with no internal or
external heat and mass-transfer limitations.[27] Typically, the dried
metal oxalate (2.5 g; after activation, the weight of the catalyst
sample was 1.2 g; 60–100 mm) was diluted with powdered glass
beads (2.5 g) in the size range 40–100 mm and loaded on to
a quartz wool plug in the reactor. The temperature of the catalyst
bed was monitored by placing a K-type thermocouple in the
middle of the catalyst bed. After activation, a gas mixture, contain-
ing 71.5% H2 and 28.5% CO2 at 2.0 nLhꢀ1 gꢀ1 catalyst, was fed into
the reactor, which was maintained at a temperature of 2408C and
a pressure of 0.92 MPa. Brooks mass-flow controllers were used to
control the flow rates of H2 and CO2. The conversions of CO2 and
H2 were obtained by gas chromatography (GC) analysis (micro-GC
equipment with a thermal conductivity detector) of the exit-gas
mixture from the reactor. The reaction products were collected in
two traps maintained at different temperatures: a hot trap (T=
1508C) and a cold trap (T=58C). The products were separated into
different fractions for quantification. The liquid products con-
densed in the hot trap were analyzed by using an HP 7890 GC in-
strument with DB-5 capillary column, whereas the aqueous phase
[24] T. H. Herranz, S. Rojas, F. J. P. Alonso, M. Ojeda, P. Ferreros, J. L. G. Fierro,
Appl. Catal. A 2006, 311, 66–75.
[25] R. A. Stowe, W. W. Russell, J. Am. Chem. Soc. 1954, 76, 319–323.
ChemCatChem 2017, 9, 1 – 11
www.chemcatchem.org
9
ꢁ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
[26] R. Satthawong, N. Koizumi, C. Song, P. Prasassarakich, J. CO2 Util. 2013,
3–4, 102–106.
[27] M. K. Gnanamani, G. Jacobs, H. H. Hamdeh, W. D. Shafer, F. Liu, S. D.
Hopps, G. A. Thomas, B. H. Davis, ACS Catal. 2016, 6, 913–927.
[28] A. Sarkar, G. Jacobs, Y. Ji, H. H. Hamdeh, B. H. Davis, Catal. Lett. 2008,
121, 1–11.
[29] S. A. Eliason, C. H. Bartholomew in Catalyst Deactivation 1997 in Studies
in Surf. Sci. and Catalysis, Vol. 111 (Eds.: C. H. Bartholomew, G. A.
Fuentes), Elsevier, Amsterdam, 1997, pp. 517–526.
[30] D. B. Bukur, X. Lang, A. Akgerman, Z. Feng, Ind. Eng. Chem. Res. 1997,
36, 2580–2587.
[33] T. Riedel, M. Claeys, H. Claeys, H. Schulz, G. Schaub, S.-S. Nam, K.-W. Jun,
M.-J. Choi, G. Kishan, K.-W. Lee, Appl. Catal. A 1999, 186, 201–213.
[34] B. H. Davis, ACS Div. Fuel Chem. Prep. 2000, 45, 129–133.
[35] B. H. Davis, Fuel Process. Technol. 2001, 71, 157–166.
[36] M. E. Dry, Catal. Today 1990, 6, 183–206.
[37] T. Riedel, G. Schaub, Ind. Eng. Chem. Res. 2001, 40, 1355–1363.
[38] V. R. R. Pendyala, U. M. Graham, G. Jacobs, H. H. Hamdeh, B. H. Davis,
Catal. Lett. 2014, 144, 1704–1716.
[39] J. Xu, C. H. Bartholomew, J. Phys. Chem. B 2005, 109, 2392–2403.
Manuscript received: October 20, 2016
Revised: January 19, 2017
[31] J. Wang, Z. You, Q. Zhang, W. Deng, Y. Wang, Catal. Today 2013, 215,
186–193.
[32] M. Martinelli, C. G. Visconti, L. Lietti, P. Forzatti, Catal. Today 2014, 228,
77–88.
Accepted Article published: January 19, 2017
Final Article published: && &&, 0000
&
ChemCatChem 2017, 9, 1 – 11
www.chemcatchem.org
10
ꢁ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

FULL PAPERS
Promoting effect of potassium: In the
hydrogenation of CO2 over K-promoted
FeCo bimetallic catalysts, initial CO2 con-
version was dependent on K loading,
whereas most catalysts exhibited similar
conversion levels after a few hours. De-
carburization suggests that potassium
stabilizes iron carbides and decreases
the hydrogenation function of the cata-
lysts; thus, it controls product selectivity.
Potassium enhanced acetic acid forma-
tion and suppressed ethanol produc-
tion.
M. K. Gnanamani, H. H. Hamdeh,
G. Jacobs, W. D. Shafer, S. D. Hopps,
G. A. Thomas, B. H. Davis*
&& – &&
Hydrogenation of Carbon Dioxide
over K-Promoted FeCo Bimetallic
Catalysts Prepared from Mixed Metal
Oxalates
ChemCatChem 2017, 9, 1 – 11
www.chemcatchem.org
11
ꢁ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ