Ni-Co/Al2O3 Bimetallic Catalysts for CH4 Steam Reforming: Elucidating the Role of Co for Improving Coke Resistance

Article abstract of DOI:10.1002/cctc.201402695

A series of supported Ni-Co/γ-Al₂O₃ bimetallic catalysts with a fixed 12% Ni loading but different Co contents were prepared by using the coimpregnation method and investigated for methane steam reforming. The addition of Co can sign

Full text of DOI:10.1002/cctc.201402695

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201402695
Ni–Co/Al2O3 Bimetallic Catalysts for CH4 Steam Reforming:
Elucidating the Role of Co for Improving Coke Resistance
Xiaojuan You,[a] Xiang Wang,*[a] Youhe Ma,[a] Jianjun Liu,[a] Wenming Liu,[a] Xianglan Xu,[a]
Honggen Peng,*[a] Changqing Li,[b] Wufeng Zhou,[b] Ping Yuan,[b] and Xiaohong Chen[b]
A series of supported Ni–Co/g-Al2O3 bimetallic catalysts with
a fixed 12% Ni loading but different Co contents were pre-
pared by using the coimpregnation method and investigated
for methane steam reforming. The addition of Co can signifi-
cantly improve the coke resistance and the reaction stability of
Ni/Al2O3 at a mild loss of the reforming activity. XPS and TEM
results prove the existence of strong interaction between Ni
and Co species. XRD and high-angle annular dark-field scan-
ning transmission electron microscopy mapping results of the
reduced catalysts provide direct evidence for surface Ni–Co
alloy formation upon Co addition onto Ni/Al2O3, which can
block part of the active low coordinated Ni sites and lower the
metal dispersion, thus effectively suppressing coking and im-
proving the reaction stability in comparison with the unmodi-
fied Ni/Al2O3 catalyst.
Introduction
It is well-known that H2 is a potential green-energy source
with little pollution, and could be a long-term energy option
for human being in the future. With the rapid development of
industry and economy, the environment pollution becomes
more and more severe, therefore, the demand for clean
energy, such as hydrogen and solar energy, is growing very
fast.[1,2] Hydrogen, as one of the important green-energy sour-
ces, might play a key role in the 21st century. Presently, meth-
ane steam reforming (MSR) is still the major and the most fea-
sible route for large-scale industrial hydrogen production,[2] be-
cause of the abundant availability of natural gas and the rela-
tively low cost compared with other methods. The main reac-
tions involved in this process are described here:[3]
MSR is usually performed on the supported group VIII
metals, such as Fe, Co, Ni, Pt, Pd, Ru, and Rh. Owing to its low
price and high activity, Ni supported on different supports,
such as g-Al2O3, a-Al2O3, MgO, CaAl2O4, or MgAl2O4, is most
commonly used.[8–16] Although other group VIII metals are also
reported to be active for the reaction, they have some draw-
backs, for instance, iron is oxidized too rapidly to lose activity,
cobalt is not able to withstand the high steam partial pres-
sures, and the precious metals have generally limited source
and too high costs. Therefore, to obtain a Ni-based catalyst
with high activity and stability for large-scale H2 production as
an energy source, it is still necessary to design and develop
catalysts with potent resistance to coking and active-site sin-
tering, which are believed to be the two major reasons leading
to the catalyst deactivation.[17–19] One of the commonly used
solutions is the addition of basic metal oxides as promoters,
such as MgO, CaO, and La2O3, CeO2.[20–23] Another method, the
addition of a second active metal, has also been reported to
be effective to improve the performance of the catalysts. For
example, the bimetallic Ni–Co catalysts have been reported to
show improved properties for some reactions, such as CO hy-
drogenation,[24–26] CH4/CO2 reaction,[27–31] methane partial oxida-
tion,[32] ethanol and acetic acid steam reforming.[33,34] Zhang
et al.[30,31] investigated Ni/M/Al/Mg/O catalysts (M=Co, Fe, Cu,
or Mn) and found that Ni–Co bimetallic catalysts have a superi-
or performance for CO2 reforming of CH4 in terms of activity
and stability to other Ni/M combination. In a 2000 h stability
test, the Ni–Co catalyst displayed a very stable performance
with very low carbon deposition. The excellent performance of
the Ni–Co bimetallic catalyst could be closely related to its
high metal dispersion, strong metal–support interaction, and
formation of a stable solid solution. Chen et al.[27] investigated
the effect of the Co–Ni ratio on the activity and stability of Co–
Ni bimetallic aerogel catalysts for methane oxy-CO2 reforming,
CH4 þ H2O ¼ CO þ 3 H2 DHq ¼ 206:2 kJ molꢀ1
ð1Þ
ð2Þ
298
CO þ H2O ¼ CO2 þ H2
DHq ¼ ꢀ41:2 kJ molꢀ1
298
Owing to the strong endothermic property of reaction (1), a re-
former generally operates at high temperatures (800–
11008C)[4–6] and water is usually fed in excess to reduce coke
formation and prolong the lifespan of the catalysts. For indus-
trial-scale H2 production, the commonly used steam-to-meth-
ane ratios are approximately 2.5–3.0.[7]
[a] X. You, Prof. Dr. X. Wang, Y. Ma, J. Liu, Dr. W. Liu, Dr. X. Xu, Dr. H. Peng
College of Chemistry
Nanchang University
Nanchang, Jiangxi 330031 (P.R. China)
Tel : (+86)15979149877
E-mail: xwang23@ncu.edu.cn
penghonggen@ncu.edu.cn
[b] C. Li, W. Zhou, P. Yuan, X. Chen
Jiangxi Golden Century Advanced Materials Co.Ltd.
Nanchang, Jiangxi 330013 (P.R. China)
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 3377 – 3386 3377

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
and found that the Co/Ni ratio has a strong influence on the
catalytic performance, which was attributed to the formation
of uniform metal alloy on the catalysts after reduction. In an-
other study,[35] the authors also investigated the support and
alloy effects on the activity and product selectivity for ethanol
steam reforming over supported Ni–Co catalysts. The authors
suggested that the support plays a very important role for the
reaction. Yu et al.[36] revealed that a Ni–Co alloy was formed in
the bimetallic catalysts. The interaction between Ni and Co im-
proved the metal particle dispersion and resulted in smaller Ni
size, thus increasing the catalytic activity and coke resistance.
Studying CH4 dry reforming on Co–Ni/TiO2 catalysts, Aiwa et al.
detected the formation of Ni–Co alloy on the catalyst surface,
which significantly improves the coke resistance of the cata-
lysts.[37] However, differently from the above-mentioned work
by Zhang et al. and Yu et al., they found that the addition of
Co decreased the dispersion of Ni active sites. Despite these
conflict observations, it is commonly concluded that the addi-
tion Co onto Ni-based catalysts can improve the reaction per-
formance, especially for coke resistance. However, to the best
of our knowledge, the application of supported bimetallic Ni–
Co catalysts, especially Ni–Co/g-Al2O3, for MSR has been rarely
profiled.[38] Furthermore, the influence of the synergetic inter-
action between metallic Ni and Co on the reaction per-
formance deserves further elucidation.
Figure 1. Reaction performance of the catalysts for MSR. Reaction condi-
tions: p=0.1 MPa, GHSV=18000 mLgꢀ1 hꢀ1, and H2O/CH4 =2:1.
coke formation on this sample could be suppressed thorough-
ly, which will be discussed in the following section. In addition,
compared with that of the unmodified Ni/Al2O3, its activity de-
crease is not significant. Notably, though 1%Co–12%Ni/Al2O3
has a high activity similar to that of the unmodified sample,
stability tests indicate that methane conversion on it still drops
gradually, implying the Co amount is not enough to suppress
the carbon deposition completely, and slow deactivation still
occurs.
Aiming to solve these problems and to develop catalysts
with potential industrial application, we prepared a series of
supported Ni–Co/g-Al2O3 bimetallic catalysts with different Co
contents but fixed Ni loading and investigated them for MSR.
The results demonstrated that Co addition onto Ni/Al2O3 cata-
lysts leads to the formation of surface Ni–Co alloys, as evi-
denced by XRD analysis and high-angle annular dark-field
scanning transmission electron microscopy (HAADF-STEM)
mapping results, which effectively suppresses coke formation
and improves the reaction stability in comparison with the un-
modified Ni-based catalyst.
The comparison of the reaction stability of 7%Co–12%Ni/
Al2O3 with that of 12%Ni/Al2O3 and 12%Co/Al2O3 is shown in
Figure 2. Under the stringent condition adopted in this study,
12%Co/Al2O3 exhibits low activity and very fast deactivation. In
comparison, 12%Ni/Al2O3 is much more active than 12%Co/
Al2O3 but its methane conversion decreases evidently from the
initial 95% to 80% after 80 h, because of the presence of coke
formation. However, 7%Co–12%Ni/Al2O3 displays very stable
reforming activity, as testified by the stable methane conver-
sion remaining at 95% during the 180 h test. Obviously, al-
though supported Ni or Co alone is not a good catalyst for the
Results
Activity evaluation and stability test for Co-Ni/Al2O3
catalysts with different Co contents
Our previous work has demonstrated that 12% Ni is the opti-
mal loading for the g-Al2O3 used in this study.[39] As an endeav-
or to analyze the Co effects and develop better catalysts, this
Ni loading was used for all the catalysts but the Co contents
were varied. As shown in Figure 1, at temperatures below
8008C, the unmodified catalyst exhibits the highest activity.
The addition of 1% Co has little negative effect on the activity,
but further increasing the Co contents degrades the activity
evidently. The activity sequence is as follows: 12%Ni/Al2O3
ꢁ1%Co–12%Ni/Al2O3 >7%Co–12%Ni/Al2O3 ꢁ3%Co–12%Ni/
Al2O3 >12%Co–12%Ni/Al2O3 ꢁ15%Co–12%Ni/Al2O3. At 8008C,
the same methane conversion was achieved on all the cata-
lysts.
Figure 2. Comparison of the reaction stability of 7%Co–12%Ni/Al2O3 with
that of 12%Ni/Al2O3 and 12%Co/Al2O3 for MSR. Reaction conditions:
p=0.1 MPa, GHSV=18000 mLgꢀ1 hꢀ1, and H2O/CH4 =2:1, T=8008C.
To clarify the modification effects of Co on Ni/Al2O3 catalysts,
7%Co–12%Ni/Al2O3 was selected for further study, because
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 3377 – 3386 3378

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
reaction, the combination of the two metals can result into Ni–
Co bimetallic catalysts with superior performance.
12%Ni/Al2O3 is 0 mggcat.ꢀ1 hꢀ1, suggesting the addition of
7%Co suppresses the coke formation completely. Apparently,
the addition of Co can improve the coking resistance of Ni/
Al2O3 without significant loss of its activity.
Thermogravimetric analysis differential scanning calorime-
try and SEM investigation for the coke resistance of
7%Co–12%Ni/Al2O3
The SEM images of the fresh and used 7%Co–12%Ni/Al2O3
and 12%Ni/Al2O3 are shown in Figure 4. Compared with the
clean surface of fresh 12%Ni/Al2O3 shown in Figure 4(a), whisk-
er-like carbon deposits are clearly present on the surface of
the catalyst used for 80 h, as shown in Figure 4(b). The results
To elucidate the inherent reasons leading to the superior per-
formance of the supported bimetallic Ni–Co catalysts, the
coking amount of the spent catalysts after long-term stability
tests was analyzed by thermogravimetric analysis differential
scanning calorimetry (TGA–DSC) techniques (Figure 3). For
12%Ni/Al2O3, as shown in Figure 3(A), a significant weight loss
Figure 4. SEM images of the catalysts. a) Fresh 12%Ni/Al2O3, b) spent
12%Ni/Al2O3, c) fresh 7%Co–12%Ni/Al2O3, d) spent 7%Co–12%Ni/Al2O3.
Scale bars: a) 300 nm, b) 500 nm, c) 300 nm, d) 300 nm.
are consistent with the previous findings on hydrocarbon dry
reforming over Ni-based catalysts.[41–44] In contrast, for 7%Co–
12%Ni/Al2O3, even after 180 h reaction time, no carbon deposit
can be observed, as shown by the images of the fresh and the
used catalysts in Figure 4(c) and (d), respectively. These SEM
results agree very well with the coke formation rates obtained
from TGA–DSC analysis and thus provide additional evidence
that the addition of Co can effectively reduce and suppress
coking on the supported Ni catalysts for CH4 steam reforming.
N2–BET, XRD, and H2 temperature programmed reduction
analysis of the fresh catalysts
The specific surface areas of the fresh samples are listed in
Table 1. g-Al2O3 support has a surface area of 137 m2 gꢀ1. The
addition of 12% Ni or Co drops its surface area to approxi-
mately 110 m2 gꢀ1. The addition of Co and Ni together further
decreases the surface areas of the prepared bimetallic cata-
lysts. In addition, the more amount of Co is added, Co, the
lower the surface area becomes. It is therefore suggested that
the decrease of the specific surface areas is one of the reasons
for the decreased reforming activity of the Co-modified sam-
ples at low temperatures.
Figure 3. A) TGA and B) DSC profiles of catalysts after the long-term stability
test.
stage with a weight loss of 22% is observed at 500–6508C,
which is accompanied by a strong exothermic peak ascribed
to coke combustion at the same temperature region in Fig-
ure 3(B).[40] For 7%Co–12%Ni/Al2O3, interestingly, no clear
weight loss stage on its TGA profile and no exothermic peak
on its DSC curve can be observed in the same temperature
region. Based on the first 80 h reaction results, 12%Ni/Al2O3
has a coking rate of 2.7 mggcat.ꢀ1 hꢀ1, and that for 7%Co–
XRD and H2 temperature programmed reduction (H2–TPR)
analyses were used together to identify the phase composi-
tions of the fresh catalysts, with the XRD patterns shown in
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 3377 – 3386 3379

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
H2–TPR profile in Figure 6 of this sample reveals only a single
reduction peak at 8278C, which is attributed to the reduction
of spinel NiAl2O4.[45] Clearly, NiO reacted thoroughly with the g-
Al2O3 support during the high-temperature calcination, there-
fore, no free NiO was present in this sample. For 12%Co/Al2O3,
the H2–TPR profile reveals two groups of reduction peaks. The
first group of peaks at 400 and 4718C are assigned to the step-
wise reduction of Co3O4,[46,47] as testified by our quantification
results, for which the H2 consumption amount of the 4718C to
4008C peak is at the stoichiometric ratio of 3:1. The second
group peak at 7958C is assigned to the reduction of spinel
CoAl2O4.[35] As both Co3O4 and CoAl2O4 have spinel structure
and nearly the same XRD diffraction features, all the peaks of
12%Co/Al2O3 can be assigned to the overlapping diffraction
from both, except for the small peak at 67.148, which is the
strongest diffraction peak of the g-Al2O3 support. Apparently,
Co3O4 and CoAl2O4 coexist in this sample but the amount of
Co3O4 is relatively larger, as evidenced by H2–TPR results.
Table 1. Specific surface area (BET) and phase composition of the fresh
catalysts.
Sample
Surface area
Phase composition analyzed by
XRD and H2–TPR
[m2 gꢀ1
]
12%Ni/Al2O3
109
106
99
94
81
78
113
137
NiAl2O4
1%Co–12%Ni/Al2O3
3%Co–12%Ni/Al2O3
7%Co–12%Ni/Al2O3
12%Co–12%Ni/Al2O3
15%Co–12%Ni/Al2O3
12%Co/Al2O3
NiAl2O4, NiO, Co3O4, CoAl2O4
NiAl2O4, NiO, Co3O4, CoAl2O4
NiAl2O4, NiO, Co3O4, CoAl2O4
NiAl2O4, NiO, Co3O4, CoAl2O4
NiAl2O4, NiO, Co3O4, CoAl2O4
Co3O4, CoAl2O4
g-Al2O3 support
g-Al2O3
Figure 5 and H2–TPR profiles in Figure 6. The phase composi-
tions of the catalysts are also listed in Table 1.
For the unmodified 12%Ni/Al2O3, all the diffraction peaks
can be assigned to spinel NiAl2O4 (Figure 5). In addition, the
For all of the Ni–Co bimetallic catalysts, with the increasing
of Co amount, a diffraction peak at 19.148, which can be safely
assigned to spinel NiAl2O4, becomes smaller. Concurrently, a dif-
fraction peak at 43.048, which is assigned to NiO, appears in
1%Co–12%Ni/Al2O3 and eventually becomes larger with the
increasing amount of Co. This indicates that the addition of Co
can impede the formation of spinel NiAl2O4 effectively, thus
keeping part of NiO free. Notably, a diffraction peaks at 31.608
becomes stronger with the addition amount of Co, which is
obviously related to Co species. Moreover, all spinel NiAl2O4,
Co3O4, and CoAl2O4 exhibit diffraction peaks at approximately
37.02, 45.30, 60.00, and 65.848, therefore, for these peaks of
the catalysts, though they change slightly with the addition of
Co, it is difficult to draw any definite conclusion based on
them. However, it is still evident that with the addition amount
of Co, the intensity of these peaks becomes weaker. Therefore,
they could reflect the decrease of the amount of NiAl2O4 in the
catalysts. For clarification, the peak at 65.848 is enlarged and
shown beside the major figure. Clearly, besides the intensity
change, the peak also shifts to a smaller angle, and eventually
close to that of the peak of 12%Co/Al2O3.
Figure 5. XRD patterns of the fresh catalysts. a) 12%Ni/Al2O3, b) 1%Co–
12%Ni/Al2O3, c) 3%Co–12%Ni/Al2O3, d) 7%Co–12%Ni/Al2O3, e) 12%Co–
12%Ni/Al2O3, f) 15%Co–12%Ni/Al2O3, g) 12%Co/Al2O3, h) g-Al2O3. Peaks char-
acteristic for the respective phase compositions are indicated. The box
shows an enlargement of the peak around 65.848.
The H2–TPR results in Figure 6 also demonstrate that with
the increasing Co amount, the reduction peak of spinel NiAl2O4
becomes smaller, and that of NiO (350–4008C) becomes larger,
but both shift to lower temperature region. In comparison, the
reduction peak at approximately 470–5008C, which is assigned
to the reduction of Co3O4 in different chemical environment,
becomes larger. As 12%Co/Al2O3 displays a reduction peak of
CoAl2O4 at 7958C, it is more reasonable to assign the high-
temperature peak of all the bimetallic Ni–Co catalysts to the
overlapping reduction of both NiAl2O4 and CoAl2O4.
In summary, H2–TPR provides extra evidence to prove that
the addition of Co can impede the formation of NiAl2O4, thus
keeping part of NiO free on the catalyst surface, which may fa-
cilitate the formation of Ni–Co alloy during the reduction pro-
cess. The formed alloy is believed to be important to increase
the coke resistance of the Co-modified catalysts.[27,30] However,
it was previously reported that the presence of Ni in the crystal
matrix of NiAl2O4 can hinder the growth of Ni particle size
Figure 6. H2–TPR results of fresh catalysts with different Co contents.
a) 12%Ni/Al2O3, b) 1%Co–12%Ni/Al2O3, c) 3%Co–12%Ni/Al2O3, d) 7%Co–
12%Ni/Al2O3, e) 12%Co–12%Ni/Al2O3, f) 15%Co–12%Ni/Al2O3, g) 12%Co/
Al2O3.The straight lines illustrate the shift in the peak maxima.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 3377 – 3386 3380

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
during the pre-reduction for Ni catalysts prior to reaction, thus
achieving catalysts with high activity.[48] The decrease of the
amount of NiAl2O4 in Ni–Co bimetallic catalysts could be nega-
tive for the activity, and might be one of the reasons account-
ing for their lowered reforming activity at low temperatures.
However, the peak shifts to 782.4 eV in 7%Co–12%Ni/g-Al2O3.
Compared with the two monometallic Ni and Co samples, the
combination of Ni and Co in the bimetallic 7%Co–12%Ni/g-
Al2O3 obviously migrates the Ni2p3/2 peak to lower binding
energy, but the Co2p3/2 peak to higher binding energy, sug-
gesting that electron migration occurs between Ni and Co.
This further confirms the close interaction between the two
metals in the bimetallic catalysts, which may facilitate the Ni–
Co alloy formation on the catalyst surface during
reduction.[31,37]
X-ray photoelectron spectroscopy analysis of the fresh
catalyst
X-ray photoelectron spectroscopy (XPS) analysis on the fresh
catalysts was performed to obtain more insight into the sur-
face properties of the bimetallic Ni–Co catalysts, with the re-
sults shown in Figure 7. For 12%Ni/Al2O3 catalyst, the binding
XRD and TEM studies on the reduced Co–Ni/Al2O3 catalysts
XRD patterns of the reduced catalysts with different Co con-
tents are shown in Figure 8. As all the catalysts were reduced
at 8008C, metallic Ni and Co are completely formed. The re-
duced 12%Ni/Al2O3 and 12%Co/Al2O3 display basically the
same XRD diffraction patterns, indicating the diffraction of
metallic Ni and Co are overlapped. Therefore, it is impossible
Figure 7. Comparison of A) Ni2p3/2 and B) Co2p3/2 XPS binding energies of
the fresh 7%Co–12%Ni/Al2O3 with those of 12%Ni/Al2O3 and 12%Co/Al2O3
catalysts, respectively.
energy of the Ni2p3/2 is 857.2 eV, which is assigned to the Ni2+
cations in NiAl2O4.[45] After Co addition, this peak shifts to
857.0 eV in 7%Co–12%Ni/g-Al2O3, which could be affected by
the presence of NiO and Co species. For 12%Co/Al2O3 catalyst,
the binding energy of the Co2p3/2 is 782.0 eV, which is ascribed
to the Co2+ cations in both spinel Co3O4 and CoAl2O4.[7,49,50]
Figure 8. XRD patterns of the reduced catalysts. a) 12%Ni/Al2O3, b) 1%Co–
12%Ni/Al2O3, c) 3%Co–12%Ni/Al2O3, d) 7%Co–12%Ni/Al2O3, e) 12%Co–
12%Ni/Al2O3, f) 15%Co–12%Ni/Al2O3, g) 12%Co/Al2O3. A) 2q range 10–908;
B) 2q range 42.0–45.58. Peaks characteristic for the respective phase compo-
sitions are indicated.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 3377 – 3386 3381

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
presence of two groups of large
particles. Thus the results con-
firm that the added Co has
strong interaction with Ni, which
affects the chemical environ-
ment of both elements.
Table 2. Texture and morphology properties of the reduced catalysts.
Sample
Ni crystallite
size [nm][a]
Ni particle
size [nm][b]
Ni/Co, hkl (111)
2q [8]
Ni/Co, hkl (200)
d [ꢁ]
2q [8]
d [ꢁ]
12%Ni/Al2O3
25.5
29.6
31.1
28.1
29.4
31.8
31.7
25.4
30.3
31.4
27.0
29.9
31.4
–
44.44
44.42
44.40
44.38
44.36
44.34
44.27
2.0371
2.0378
2.0386
2.0395
2.0400
2.0413
2.0457
51.80
51.72
51.70
51.68
51.66
51.60
51.56
1.7610
1.7635
1.7660
1.7673
1.7679
1.7698
1.7710
1%Co–12%Ni/Al2O3
3%Co–12%Ni/Al2O3
7%Co–12%Ni/Al2O3
12%Co–12%Ni/Al2O3
15%Co–12%Ni/Al2O3
12%Co/Al2O3
H2 adsorption–desorption
measurements of the reduced
Co–Ni/Al2O3 catalysts
[a] Crystallite size calculated from XRD diffraction peak hkl (111). [b] Particle size measured by TEM.
In Table 3, the metal dispersion
results on the reduced samples
to distinguish metallic Ni and Co from each other for those bi-
metallic samples. The crystallite sizes of the Ni/Co metals esti-
mated using Scherrer’s equation are listed in Table 2. The crys-
tallite size of pure Ni0 for the unmodified 12%Ni/Al2O3 is
25.5 nm, and that of pure Co0 for 12%Co/Al2O3 is 31.7 nm. For
all of the bimetallic samples, the crystallite sizes of the Ni/Co
metals ranges from 28.1 to 31.8 nm, which are larger than that
of pure Ni0 and close to that of pure Co0. It was reported previ-
ously that the formation of alloy generally resulted in the
growth of metal crystallite size.[37]
measured by H2 adsorption–desorption are listed. As all the
samples were calcined and reduced at 8008C, it is not surpris-
ing to find that they have generally low metal dispersion.
12%Ni/Al2O3 catalyst has the highest metal dispersion, and
Table 3. Metal dispersion of the catalysts after reduction.
Catalyst
Metal dispersion [%]
12%Ni/Al2O3
9.2
3.6
2.1
1.5
0.8
0.6
0.8
1%Co–12%Ni/Al2O3
3%Co–12%Ni/Al2O3
7%Co–12%Ni/Al2O3
12%Co–12%Ni/Al2O3
15%Co–12%Ni/Al2O3
12%Co/Al2O3
To further clarify this, the strongest diffraction peak, the
(111) peak between 44 and 458 of each sample was enlarged
and plotted separately in Figure 8(B). The 2q peak of pure Ni0
is at 44.448, whereas that of pure Co0 is at 44.278 (Table 2). Al-
though the difference is very marginal, it is still possible to dif-
ferentiate the peaks of the monometallic catalysts. However,
for all bimetallic catalysts, only a uniform diffraction peak can
be observed for the coexisted Ni0 and Co0 in the sample,
which shifts to lower 2q value gradually with the increasing of
Co content, similar to what was reported on NimCon/cordierite
catalysts.[51] Additionally, the 2q values of (200) peaks of the
samples were also carefully measured and listed in Table 2, and
a similar shift was observed. This peak shift is typical for alloy
formation, suggesting that Ni–Co alloy is formed in the bimet-
allic catalysts during the reduction.[37]
12%Co/Al2O3 catalyst has the lowest one among all of the
samples. Although the intrinsic property of a metal could be
the major reason to affect its activity, the metal dispersion can
also influence its activity. Therefore, it is easy to understand
that 12%Ni/Al2O3 exhibit very high initial steam reforming ac-
tivity, but the activity of 12%Co/Al2O3 is extremely low, as
shown in Figure 2. The addition of Co onto the samples appa-
rently decreases the Ni/Co alloy dispersion. In addition, with
the increasing of Co amount, the alloy dispersion keeps on de-
creasing. With the increasing of the Co content to15%, the
alloy dispersion on the sample is even slightly lower than that
of pure 12%Co/Al2O3. It is commonly accepted that the oxida-
tion activity of a supported metal catalysts is remarkably affect-
ed by the exposed metal surface. Regularly, on the same sup-
port, the higher the metal dispersion is, the higher is the activi-
ty. Therefore, herein, metal dispersion is clearly another param-
eter to determine the reforming activity, because the activities
shown in Figure 1 and Figure 2 are almost proportional to the
metal dispersion extent.
The metal-particle morphology and size distribution of the
reduced catalysts were investigated by TEM technique, with
the images shown in Figure 9. The mean particle size of metal-
lic Ni/Co measured by TEM for each sample is also listed in
Table 2. For 12%Ni/Al2O3 catalyst, the mean Ni0 particle size is
approximately 25.4 nm (Figure 9a). Notably, regular TEM tech-
nique and image cannot distinguish Ni0 and Co0, therefore, the
particles measured on the bimetallic sample surface actually
consist of both metallic Ni and Co, which are Ni–Co alloy parti-
cles formed on the surface, as testified by XRD results. In line
with the XRD results, the Ni–Co alloy particle sizes measured
by TEM are larger than that of pure Ni0 but close to that of
Co0. In addition, all of the catalysts have similar crystallite and
particle sizes, indicating the absence of the secondary aggre-
gation of the Ni–Co alloy microcrystals. Regarding the Ni/Co
distribution on the particle surface, the monometallic Ni
sample exhibits monomodal distribution, indicating the only
presence of one group of large particles. In comparison, all the
bimetallic samples display bimodal distribution, revealing the
HAADF–STEM mapping study on the reduced
7%Co–12%Ni/Al2O3
It is well-known that scanning transmission electron microsco-
py with energy-dispersive X-ray spectroscopy (STEM–EDX)
mapping can provide clear image for element distributions. As
an endeavor to further clarify the formation of Ni–Co alloys on
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 3377 – 3386 3382

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 10. HAADF–STEM mapping of reduced 7%Co–12%Ni/Al2O3 catalyst.
a) Overview, b) Ni mapping image, c) Co mapping image. The boxes in (a)
indicate the mapping zones. Scale bars: a) 100 nm, b) 50 nm, c) 50 nm
study, was thus subjected to HAADF–STEM mapping study. As
shown in Figure 10(a), the mapping zone is labelled by a “+1”
box. The two particles in the selected area have a mean size of
approximately 30 nm, respectively. The Ni and Co mapping
images are displayed in Figure 10(b) and (c). The strong signals
of both particles, which have the same shape, testify clearly
that the two selected particles in the mapping zone consist of
both Ni0 and Co0, which provides direct and additional evi-
dence to the XRD results that there is strong interaction be-
tween Ni0 and Co0 resulting in the formation of Ni–Co alloy on
the surface of the reduced bimetallic catalysts.
Discussion
Co-modified Ni-based catalysts have been extensively investi-
gated for hydrogen production reactions, such as CH4 dry re-
forming,[28,31,52] toluene and acetic acid steam reforming,[53] and
tar cracking from biomass pyrolysis.[51] It has been commonly
concluded that the Ni–Co bimetallic catalysts have significantly
improved reaction performances compared to the correspond-
ing monometallic Ni-based catalysts, especially for coke resist-
ance, a critical issue for all the reforming reactions and cata-
lysts. Studying CH4 dry reforming, Zhang et al. ascribed the
promotional effects to the improved Ni dispersion, the strong
metal–support interaction, and the formation of solid solution
by Co addition.[30,31] Yu et al. also found higher metal disper-
sion by Co addition.[36] In addition, Ni–Co alloy formation was
also observed by them and several other groups,[27,30,32] which
produces synergism effect between metallic Ni and Co, and
Figure 9. TEM images of the reduced catalysts with different Co contents.
a) 12%Ni/Al2O3, b) 1%Co–12%Ni/Al2O3, c) 3%Co–12%Ni/Al2O3, d) 7%Co–
12%Ni/Al2O3, e) 12%Co–12%Ni/Al2O3, f) 15%Co–12%Ni/Al2O3.
Scale bars: a–f) 100 nm.
the surface of the reduced bimetallic catalysts, the fully re-
duced 7%Co–12%Ni/Al2O3 catalyst, a typical sample in this
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 3377 – 3386 3383

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
improves the coke resistance of the resulted catalysts. Howev-
er, in the case of Ni–Co/TiO2 catalysts for methane dry reform-
ing, although Aika et al. also found the formation of Ni–Co
alloy, they observed decreased Ni dispersion by Co addition,[37]
which is in line with what found in this study. Despite the com-
monly accepted fact that Ni–Co bimetallic catalysts are good
catalysts for various reforming reactions, there still exist some
unclear issues for the promotional effects of Co on Ni-based
catalysts, which deserve further study. In addition, to the best
of our knowledge, Ni-Co bimetallic catalysts have rarely been
introduced into MSR, especially supported on Al2O3, an impor-
tant reaction currently employed for large-scale industrial hy-
drogen production. With the purpose to clarify the positive ef-
fects of Co on Ni, and eventually develop better catalysts with
potent coke resistance and improved stability for natural-gas
steam reforming, Ni–Co/Al2O3 bimetallic catalysts were investi-
gated for MSR in this study.
unmodified catalysts. However, after 10 h reaction time, evi-
dent deactivation occurred to the unmodified 12%Ni/Al2O3
catalyst, owing to severe coke formation. Based on the infor-
mation provided in this study, further research on this catalyst
system, such as tuning the Co/Ni ratios, optimizing the prepa-
ration condition and selecting better supports, could achieve
active and stable catalysts for large-scale industrial application
for natural-gas steam reforming.
Conclusions
A series of supported bimetallic Ni–Co/g-Al2O3 catalysts with
fixed 12% Ni loading but with different Co contents were pre-
pared by the coimpregnation method and investigated for CH4
steam reforming. The addition of Co can effectively improve
the coke resistance of Ni/Al2O3 and the reaction stability at
a reasonable loss of the reforming activity at lower tempera-
tures. At 8008C, the regular operation temperature for industri-
al-scale natural-gas steam reforming, the Co-modified catalysts
exhibit the same activity as the unmodified catalyst. XRD anal-
yses and high-angle annular dark-field scanning transmission
electron microscopy mapping of the reduced catalysts provid-
ed direct evidence for the observation that the addition of Co
leads to the formation of surface Ni–Co alloys, which play a crit-
ical role in suppressing the coke formation of the bimetallic
Ni–Co catalysts. However, the formation of Ni–Co alloy could
also block part of the low-coordinated active Ni sites and de-
crease the metal dispersion, which is believed to be the major
reason accounting for the lowered reforming activity of the
Co-modified catalysts at lower temperatures. Based on the in-
formation from this work, improved natural-gas steam reform-
ing catalysts with superior coke resistance and applicable activ-
ity could be designed and developed for large-scale industrial
application.
As a result, although the addition of Co decreased the re-
forming activity at low temperatures, the coke resistance as
well as the reaction stability were improved significantly. Previ-
ously, it was found that the addition of Sn into Ni-based cata-
lysts formed Ni–Sn alloys on the catalyst surface.[39,54,55] As
a result, the activity of the Ni sites can be degraded, as evi-
denced by the improved activation energy of CꢀH bonds. The
activity degradation was ascribed to the geometric and elec-
tronic changes of Ni sites induced by Sn, for example, Sn could
block the low-coordinate Ni sites on Ni particles, which are ac-
tually the active sites for CꢀH activation, the rate-determining
step for CH4 reforming. Studying Au-modified Ni catalysts for
steam reforming, Norskov et al. also found the formation of
surface Au-Ni alloy, which improves the coke resistance of the
bimetallic catalyst significantly but at the expense of the activi-
ty.[56] As with Sn-modified catalysts, the formation of Au–Ni
alloy, as concluded by the authors, decreases the activity of
the active sites. In the study herein, Ni–Co alloy formation was
observed, as directly testified by XRD and TEM analyses and
STEM–EDX mapping and indirectly evidenced by XPS and H2–
TPR results. As a consequence, the active low-coordinate Ni
sites could be blocked by alloy formation, as what occurred in
the cases of Sn and Au modification. Therfore, compared with
the activity of the unmodified Ni/Al2O3 catalysts, the reforming
activity at low temperatures was decreased by the addition of
Co. In addition, Co addition also increased the crystallite and
particle size of the metals, and decreased the metal dispersion
and the specific surface areas of the bimetallic catalysts, which
could also account for the lowered reforming activity. There-
fore, the drop of the activity at low temperatures by Co modifi-
cation, as shown in Figure 1, is reasonable.
Experimental Section
Catalyst preparation
The bimetallic Ni–Co catalysts supported on spherical g-Al2O3 (2–
3 mm) were prepared by the impregnation method, with
Ni(NO3)2·6H2O and Co(NO3)2·6H2O as the precursors. g-Al2O3 (8008C
4 h calcined) support was added into the premixed aqueous solu-
tion of Ni(NO3)2 and Co(NO3)2 and stirred constantly for 24 h. After-
wards, the solution was heated at 808C until all the water was
evaporated, then the mixture was further dried at 1108C overnight
and subsequently calcined in air atmosphere at 8008C for 4 h. The
catalysts were denoted by m%Co–12%Ni/Al2O3, because the Ni
loading was always kept at 12 wt.%, but the Co loading varied
from 1 wt.% to 15 wt.%. For comparison, 12%Ni/Al2O3 and
12%Co/Al2O3 were also prepared with impregnation method.
As shown in Figure 3 and 4, with the addition of 7% Co, the
severe coking of the original Ni/Al2O3 can be completely sup-
pressed. Therefore, no deactivation occurred to 7%Co–12%Ni/
Al2O3 during the 180 h test under the stringent condition
(Figure 2). Although all the Co-modified catalysts have lower
activity than the unmodified 12%Ni/Al2O3 below 8008C, at
8008C, the general temperature for industrial operation for
natural-gas steam reforming for syngas production, the initial
methane conversion is the same for both the Co-modified and
Catalyst characterization
TGA–DSC was used to monitor the coke deposition amount of the
spent catalysts. The experiments were performed with catalyst
amounts of approximately 10 mg on a TA Q600 instrument with
a ramping rate of 108Cminꢀ1 from 25 to 8008C in an air flow of
100 mLminꢀ1
.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 3377 – 3386 3384

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
XRD method was used to analyze the phase composition of the
freshly calcined and reduced catalysts. The experiments were per-
formed on a Bruker AXS D8Focus diffractometer instrument oper-
ating at 40 kV and 40 mA with a Cu target and Ka-ray irradiation.
Scans were collected over a range of 2q degree from 108 to 908
with a step of 0.038sꢀ1. The calculated experimental error for 2q
measurement of the peaks was ꢂ0.018.
Acknowledgements
This work was supported by the Chinese Natural Science Founda-
tion (21263015, 21203088),the Education Department of Jiangxi
Province (KJLD14005, GJJ14205), and the Natural Science Foun-
dation of Jiangxi Province (20142BAB213013), which is greatly
acknowledged by the authors.
H2–TPR technique was used to study the effect of Co on the redox
property of NiO species in the catalysts. The experiments were per-
formed on a FINESORB 3010C instrument in a 10% H2/Ar mixture
gas flow, with the temperature increased from room temperature
to 9008C at a rate of 108Cminꢀ1. Generally, a catalyst amount of
10 mg was used for the test. A thermal conductivity detector (TCD)
was used to monitor the H2 consumption, and high-purity CuO
(99.99%) was employed as a standard sample to quantify the H2
consumption.
Keywords: alloys · cobalt · hydrogen · nickel · scanning probe
microscopy
[1] M. G. Schultz, Science 2003, 302, 624–627.
[2] R. Kothari, D. Buddhi, R. L. Sawhney, Renewable Sustainable Energy Rev.
2008, 12, 553–563.
[3] F. Qi, S. Howard, F. S. Maria, Science 2003, 301, 935–938.
[4] S. Ayabe, H. Omoto, T. Utaka, R. Kikuchi, Appl. Catal. A 2003, 241, 261–
269.
[5] Y. Matsumura, T. Nakamori, Appl. Catal. A 2004, 258, 107–114.
[6] R. R. N. Jens, S. Jens, K. N. Jens, Adv. Catal. 2002, 47, 65–139.
[7] S. B. Wang, Energy Fuels 1996, 10, 896–904.
[8] A. Arpornwichanop, M. Wasuleewan, Int. J. Hydrogen Energy 2011, 25,
929–934.
[9] S. C. Baek, J. W. Bae, J. Y. Cheon, K. W. Jun, K. Y. Lee, Catal. Lett. 2011,
141, 224–234.
[10] N. de Miguel, J. Manzanedo, P. L. Arias, Chem. Eng. Technol. 2012, 35,
2195–2203.
[11] Y. S. Jung, W. L. Yoon, Y. W. Rhee, Y. S. Seo, Int. J. Hydrogen Energy 2012,
37, 9340–9350.
[12] I. H. Son, S. J. Lee, A. Soon, H. S. Roh, H. Lee, Appl. Catal. B 2013, 134–
135, 103–109.
[13] Y. L. Zhao, Y. K. Lu, L. P. Chang, W. R. Bao, J. Fuel Chem. Technol. 2010,
38, 218–222.
[14] X. Du, D. Zhang, L. Shi, R. Gao, J. Zhang, J. Phys. Chem. C 2012, 116,
10009–10016.
[15] X. Du, D. Zhang, R. Gao, L. Huang, L. Shi, J. Zhang, Chem. Commun.
2013, 49, 6770–6772.
[16] T. Xie, L. Shi, J. Zhang, D. Zhang, Chem. Commun. 2014, 50, 7250–7253.
[17] D. L. Trimm, Catal. Today 1997, 37, 233–238.
[18] C. J. Liu, J. Y. Ye, J. J. Jiang, Y. X. Pan, ChemCatChem 2011, 3, 529–541.
[19] B. C. Zhang, X. L. Tang, Y. Li, Y. D. Xu, W. J. Shen, Int. J. Hydrogen Energy
2007, 32, 2367–2373.
[20] Y. P. Li, T. J. Wang, C. Z. Wu, Y. Gao, X. H. Zhang, Ind. Eng. Chem. Res.
2010, 49, 3176–3183.
[21] Z. Y. Hou, X. M. Zheng, T. Yashima, React. Kinet. Catal. Lett. 2005, 84,
229–235.
H2 adsorption–desorption was measured with a Micromeritics
ASAP2020 system to analyze the Ni/Co metal dispersion on differ-
ent supports. Typically, a catalyst amount of 50 mg was used for
the test. The sample was reduced first in a 30 mLminꢀ1 10% H2/Ar
gas mixture flow at 8008C for 120 min and then purged by
a 30 mLminꢀ1 ultra-high purity He flow at RT for 30 min before H2
adsorption in a 20 mLminꢀ1 10% H2/Ar flow. Afterwards, the
sample was purged again in a 30 mLminꢀ1 ultra-high purity He
flow for 30 min to remove any physically adsorbed H2. After get-
ting a stable baseline, the temperature was increased to 8008C
with a heating rate of 108Cminꢀ1. A H2 desorption peak was gen-
erally detected for each sample at approximately 1008C, and the
relative H2 desorption amount of each sample was monitored with
a TCD. For the quantification of the metal dispersion, a standard
0.5% Pt/Al2O3 calibration sample provided by Micromeritics was
used.
SEM images of the catalysts were taken on a Hitachi S-4800 field-
emission scanning electron microscope. The TEM images were
taken on a Tecnai F30 transmission electron microscope, and
HAADF–STEM mapping analysis was performed on Oxford Instru-
ments X-Max equipped with TEM F30.
XPS experiments were performed on a RBD upgraded PHI-5000C
ESCA system by using a single MgKa X-ray source operating at
250 W and 14 kV. The spectra were obtained at ambient tempera-
ture with an ultrahigh vacuum. The binding energies were calibrat-
ed by using the C1s peak of graphite at 284.6 eV as a reference.
[22] J. Li, D. Wang, G. D. Zhou, Y. X. Xue, C. Li, T. X. Cheng, Ind. Eng. Chem.
Res. 2011, 50, 10955–10961.
[23] Y. J. Bang, J. G. Seo, Int. J. Hydrogen Energy 2011, 36, 8307–8315.
[24] H. Woo, R. Srinivasan, L. Rice, R. De Angelis, P. Reucroft, J. Mol. Catal.
1990, 59, 83–95.
[25] I. Tatsumi, E. Koichi, A. Hiromichi, Appl. Catal. 1987, 30, 225–238.
[26] P. C. Das, N. C. Pradhan, A. K. Dalai, N. N. Bakshi, Catal. Lett. 2004, 98,
153–160.
[27] L. Chen, Q. S. Zhu, R. F. Wu, Int. J. Hydrogen Energy 2011, 36, 2128–
2136.
Catalyst activity evaluation
The catalysts were tested in a fixed-bed microreactor. Prior to reac-
tions, generally a catalyst amount of 400 mg was reduced in situ in
a 10% H2/Ar flow at 8008C for 2 h. Afterwards, deionized water
was pumped into a preheating chamber controlled at 2008C to
evaporate it completely into steam, which was then mixed with
CH4 (99.99%) flow at the stoichiometric H2O/CH4 ratio of 2:1. The
flow rate of pure CH4 was 40 mLminꢀ1.The system pressure was
0.1 MPa, and the gas hourly space velocity (GHSV) was
10800 mLgꢀ1 hꢀ1. The outlet gas was cooled by ice water to trap
the excess water vapor in the flow, and analyzed on-line on
a GC9310 gas chromatograph equipped with a TDX-01 column
and a TCD detector, using Ar as the carrier gas to monitor the
amount of H2, CO, CO2, and CH4.
[28] D. S. Josꢂ-Alonso, M. J. Illꢃn-Gꢄmez, M. C. Roman-Martꢅnez, Int. J. Hydro-
gen Energy 2013, 38, 2230–2239.
[29] P. Djinovic, Appl. Catal. B 2012, 125, 259–270.
[30] J. G. Zhang, H. Wang, A. K. Dalai, Appl. Catal. A 2008, 339, 121–129.
[31] J. Zhang, H. Wang, A. K. Dalai, J. Catal. 2007, 249, 300–310.
[32] A. C. W. Koh, L. W. Chen, W. K. Leong, B. F. G. Johnson, Int. J. Hydrogen
Energy 2007, 32, 725–730.
[33] I. Rossetti, J. Lassoa, E. Finocchiob, G. Ramisb, V. Nichelec, Appl. Catal. A
2014, 477, 42–53.
[34] X. Hu, G. X. Lu, J. Mol. Catal. A 2007, 261, 43–48.
[35] L. W. Chen, C. K. S. Choong, Z. Y. Zhong, L. Huang, Z. Wang, J. Y. Lin, Int.
J. Hydrogen Energy 2012, 37, 16321–16332.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 3377 – 3386 3385

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[36] Y. Yu, G. Q. Jin, Y. Y. Wang, X. Y. Guo, Catal. Commun. 2013, 31, 5–10.
[37] K. Takanabe, K. Nagaoka, K. Nariai, K. C. Aika, J. Catal. 2005, 232, 268–
275.
[38] D. Harshini, Y. Kwon, J. Han, S. P. Yoon, S. W. Nam, T.-H. Lim, Korean J.
Chem. Eng. 2010, 27, 480–486.
[48] N. Muradov, F. Smith, A. T-Raissi, Int. J. Hydrogen Energy 2008, 33, 2023–
2035.
[49] S. Rana, S. Ram, Mater. Sci. Technol. 2005, 21, 243–249.
[50] S. Y. Foo, C. C. Kui, T. H. Nguyen, A. A. Adesoji, Int. J. Hydrogen Energy
2012, 37, 17019–17026.
[39] J. J. Liu, H. G. Peng, W. M. Liu, X. L. Xu, X. Wang, ChemCatChem 2014, 6,
2095–2104.
[51] M. Lu, P. M. Lv, Z. H. Yuan, H. W. Li, Renewable Energy 2013, 60, 522–
528.
[40] K. Tomishige, Y. Himeno, Y. Matsuo, Y. Yoshinaga, K. Fujimoto, Ind. Eng.
Chem. Res. 2000, 39, 1891–1897.
[41] X. Du, D. S. Zhang, L. Y. Shi, R. H. Gao, J. P. Zhang, Nanoscale 2013, 5,
2659–2663.
[42] D. L. Trimm, Catal. Rev. 1977, 16, 155–189.
[43] C. Wei, G. F. Zhao, Q. S. Xue, Appl. Catal. B 2013, 136–137, 260–268.
[44] J. S. Choi, K. I. Moon, Y. G. Kim, J. S. Lee, C. H. Kimb, D. L. Trimm, Catal.
Lett. 1998, 52, 43–47.
[52] Z. Hou, T. Yashimab, Catal. Lett. 2003, 89, 193–197.
[53] S. Bona, P. Guillꢂn, J. G. Alcalde, L. Garcꢅa, R. Bilbao, Chem. Eng. J. 2008,
137, 587–597.
[54] E. Nikolla, A. Holewinski, J. Schwank, S. Linic, J. Am. Chem. Soc. 2006,
128, 11354–11355.
[55] E. Nikolla, J. Schwank, S. Linic, J. Am. Chem. Soc. 2009, 131, 2747–2754.
[56] F. Besenbacher, I. Chorkendorff, B. S. Clausen, B. Hammer, A. M. Molen-
broek, J. K. Nørskov, Science 1998, 279, 1913–1915.
[45] J. K. Xu, W. Zhou, Z. J. Li, J. H. Wang, J. X. Ma, Int. J. Hydrogen Energy
2009, 34, 6646–6654.
[46] X. L. Xu, H. Han, J. J. Liu, W. M. Liu, W. L. Li, X. Wang, J. Rare Earths 2014,
32, 159–169.
Received: August 31, 2014
Revised: September 18, 2014
Published online on October 30, 2014
[47] Y. Lou, L. Wang, Z. Y. Zhao, Y. H. Zhang, Z. G. Zhang, G. Z. Lu, Y. Guo,
Y. L. Guo, Appl. Catal. B 2014, 146, 43–49.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 3377 – 3386 3386