Isobutane dehydrogenation over metal (Fe, Co, and Ni) oxide and sulfide catalysts: Reactivity and reaction mechanism

Article abstract of DOI:10.1002/cctc.201402173

Silica-supported metal oxide and sulfide catalysts of Fe, Co, and Ni were evaluated comparatively for the catalytic dehydrogenation of isobutane. The results of the activity test and temperature-programmed reduction of hydrogen characterization indicate that metal oxides, except Fe₂O₃, are easily reduced to metal ensembles, which are extremely active for alkane hydrogenolysis and lead to the formation of a considerable amount of methane and coke. However, the dehydrogenation performance was significantly improved after sulfidation treatment. The introduction of sulfur affects the catalysts in two ways: one is the geometric effect, which dilutes the aggregated metallic species and reduces hydrogenolysis activity, and the other is the electronic effect, which facilitates the desorption of olefin and increases the product selectivity. Moreover, the reaction mechanism is explored by using the proposed model of the interaction between isobutane and sulfide catalysts. Finally, sulfur loss and partial coke deposition are determined to be the main reasons for catalyst deactivation.

Full text of DOI:10.1002/cctc.201402173

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201402173
Isobutane Dehydrogenation over Metal (Fe, Co, and Ni)
Oxide and Sulfide Catalysts: Reactivity and Reaction
Mechanism
Guowei Wang,[a] Chuancheng Gao,[b] Xiaolin Zhu,[a] Yanan Sun,[a] Chunyi Li,*[a] and
Honghong Shan[a]
Silica-supported metal oxide and sulfide catalysts of Fe, Co,
and Ni were evaluated comparatively for the catalytic dehydro-
genation of isobutane. The results of the activity test and tem-
perature-programmed reduction of hydrogen characterization
indicate that metal oxides, except Fe2O3, are easily reduced to
metal ensembles, which are extremely active for alkane hydro-
genolysis and lead to the formation of a considerable amount
of methane and coke. However, the dehydrogenation per-
formance was significantly improved after sulfidation treat-
ment. The introduction of sulfur affects the catalysts in two
ways: one is the geometric effect, which dilutes the aggregat-
ed metallic species and reduces hydrogenolysis activity, and
the other is the electronic effect, which facilitates the desorp-
tion of olefin and increases the product selectivity. Moreover,
the reaction mechanism is explored by using the proposed
model of the interaction between isobutane and sulfide cata-
lysts. Finally, sulfur loss and partial coke deposition are deter-
mined to be the main reasons for catalyst deactivation.
Introduction
The dehydrogenation of light alkanes has attracted increased
attention as an efficient route to produce alkenes from low-
cost saturated hydrocarbons.[1,2] A number of commercial cata-
lytic dehydrogenation processes have been developed for the
production of propene and isobutene, which are widely used
in the synthesis of polymers and gasoline-octane boosting
additives.
to increased formation of COx and inferior selectivity toward al-
kenes.[22,23] Consequently, it is difficult to make a significant
breakthrough in this route. Therefore, the development of
a cost-effective and environmentally friendly catalyst system
demonstrating an excellent dehydrogenation performance in
the absence of oxidants is important.[24]
Because CÀH bond activation is regarded as the rate-deter-
mining step in alkane dehydrogenation,[25,26] the mechanism of
CÀH bond activation of the alkane and the interaction of the
catalyst should be determined primarily. However, owing to
the complexity of heterogeneous catalytic reactions, the activa-
tion of alkanes with transition-metal ions in the gas phase has
been investigated.[27–37] It has been found that CÀC bond inser-
tion is more favorable on thermochemical grounds than CÀH
bond insertion over Ni, Co, and Fe ions.[31,37]
Pt- and CrOx-based catalysts are generally used in these
commercial dehydrogenation processes. To increase alkene se-
lectivity and decrease coke formation, appropriate promoters,
such as Sn, K, Ca, and Li,[3–6] are introduced into these catalysts.
Although the catalytic performance is relatively satisfactory
after the modification, the high cost of Pt and the hazardous
effects of Cr6+ have unavoidably limited the use of these
catalysts.
The oxidative dehydrogenation of alkanes, an alternative
route to catalytic dehydrogenation, has been studied widely
owing to its thermodynamic advantages.[7,8] Previous studies
have been focused on supported vanadium oxide[9–12] and mo-
lybdenum oxide[13–16] catalysts. Although many attempts have
been made to modify the catalyst formula,[17,18] and to select
mild oxidants,[19–21] oxidation reactions inevitably occur, leading
We have previously found that the introduction of sulfur
into oxide catalysts of Ni and Fe inhibits cracking reactions ef-
fectively and increases dehydrogenation activity significant-
ly.[38,39] In addition to the aforementioned catalysts, a series of
metal sulfide catalysts demonstrating an excellent dehydrogen-
ation performance, even better than that of industrial Pt- and
CrOx-based catalysts, has been discovered.[40] Given that sup-
ported MoS2,[41] CoS,[42] and NiS[43] catalysts are widely used in
the hydrogenation reaction, and that hydrogenation and dehy-
drogenation are reverse reactions, the excellent dehydrogena-
tion performance of these sulfides seems to be reasonable.
However, supported sulfides of Cu, Mn, and Fe, rarely used in
hydrogenation reactions, also demonstrate excellent per-
formance. Therefore, to better understand the active sites for
dehydrogenation, the reaction behavior and mechanisms of
sulfide catalysts should be investigated in detail.
[a] Dr. G. Wang, Dr. X. Zhu, Dr. Y. Sun, Prof. C. Li, Prof. H. Shan
State Key Laboratory of Heavy Oil Processing
China University of Petroleum
Qingdao 266580 (P.R. China)
Fax: (+86)532-86981718
E-mail: chyli@upc.edu.cn
chyli_upc@126.com
[b] Prof. C. Gao
Shandong Hengyuan Petrochemical Company Ltd.
Dezhou 251500 (P.R. China)
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 2305 – 2314 2305

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Herein, to explore the reasons for the remarkable
improvement in the dehydrogenation performance
after the introduction of sulfur, silica-supported oxide
catalysts of Fe, Co, and Ni with and without sulfida-
tion were chosen as representative samples and eval-
uated for the dehydrogenation of isobutane. More-
over, the dehydrogenation reaction mechanisms and
the origin of catalyst deactivation over metal sulfides
were systematically studied.
Table 1. Physicochemical properties and activity test results of silica-supported oxide
and sulfide catalysts.
[a]
[b]
Sample
SBET
[m2 gÀ1
Vp
[cm3 gÀ1
Dp
Cisobutane Sisobutene
TOF[c] Coke content[d]
[hÀ1
[wt%]
]
]
[nm] [wt%]
[wt%]
]
13NiO/SiO2
13NiS/SiO2
13Co3O4/SiO2 225
13CoS/SiO2 211
13Fe2O3/SiO2 239
13FeS/SiO2 236
236
234
0.88
0.74
0.86
0.71
0.88
0.84
8.5
9.5
8.7
9.7
8.4
9.5
91.1
67.0
14.1
71.1
13.3
69.2
7.6
87.6
25.3
87.0
43.5
86.6
0.26 5.5
2.35 0.1
0.15 0.6
2.51 0.1
0.22 0.4
2.16 0.1
[a] Conversion of isobutane obtained in the first 10 min during the activity test; [b] Se-
lectivity toward isobutene obtained in the first 10 min during the activity test; [c] TOF
calculated as the amount of formed isobutene per hour divided by the amount of the
active metal atom (determined through H2 chemisorption) supported on the catalyst;
[d] Weight content of coke on the spent catalyst (coke content was determined after
2 h reaction for oxide catalysts and after 5 h for sulfide catalysts).
Results and Discussion
Structural characteristics of metal oxide and sulfide
catalysts
The XRD patterns of silica-supported oxide catalysts
of Fe, Co, and Ni (with a loading of 13 wt%), along
with those of the corresponding sulfide catalysts, are shown in
Figure 1. The diffraction peaks representative of Fe2O3, Co3O4,
and NiO are observed for oxide catalysts after calcination. As
for the sulfide catalysts, newly emerged diffraction peaks char-
acteristic of the cubic structure of FeS, CoS, and NiS have been
detected, indicating that sulfidation treatment can efficiently
convert oxides on the catalysts to the relevant sulfides.
In addition, BET surface areas and pore-structure properties
of both metal oxide and sulfide catalysts were measured to
clarify whether sulfidation treatment could have destructive ef-
fects on the pore channels of the catalysts; the results are
listed in Table 1. These results indicate that the catalyst surface
area and pore volume decreased slightly after sulfidation and
the pore diameter increased to some extent, suggesting that
sulfidation treatment caused no significant damage to the cat-
alyst structure.
a pulse-mode microreactor equipped with an online mass
spectrometer; the obtained mass spectra are shown in
Figure 2. As for the 13Fe2O3/SiO2 catalyst, an increasing trend
of the signal intensity was detected for isobutane (m/z=58),
isobutene (m/z=56), and hydrogen (m/z=2) in several initial
pulses. The signal intensity demonstrated no significant
change with the increase in pulse number. Moreover, the re-
sults of the activity test in the first 10 min are summarized in
Table 1. Although the ability to activate isobutane was relative-
ly poor for the Fe-based catalyst, the selectivity toward isobu-
tene was the highest among the three catalysts.
Compared with the relatively stable signal intensity ob-
served for the 13Fe2O3/SiO2 catalyst, the signal intensity for
13Co3O4/SiO2 and 13NiO/SiO2 varied significantly with pulse
number. The signal intensities of hydrogen and methane
(m/z=16) first increased and then decreased. In contrast, the
signals corresponding to isobutane and isobutene presented
an opposite variation trend; that is, the minimum signal inten-
sities of isobutane and isobutene were achieved when the sig-
nals corresponding to hydrogen and methane reached the
maximum, implying that with the increase in pulse number,
the reaction first transformed from dehydrogenation into
cracking and then reverted to
Performance of metal oxide catalysts
To detect the variation trend of products more accurately
during the reaction, isobutane dehydrogenation was per-
formed over silica-supported metal oxide catalysts with
dehydrogenation.
For the 13Co3O4/SiO2 and
13NiO/SiO2 catalysts, the signal
intensities of both hydrogen and
methane began to increase at
the same pulse number. Howev-
er, as for the 13NiO/SiO2 catalyst,
the signal intensities of hydro-
gen and methane first increased
at a faster rate and then de-
creased at a slower rate. More-
over, consistent with the activity
results summarized in Table 1,
no isobutene was detected in
the mass spectra of the Ni-based
catalyst, indicating that the
Figure 1. XRD patterns of silica-supported Fe-, Co-, and Ni-based catalysts.
cracking ability of NiO/SiO2 was
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 2305 – 2314 2306

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
lysts may be attributed to the
changes in surface species.
During the catalysis of metal
oxide on the fresh catalyst, iso-
butene was generated initially
through the dehydrogenation
reaction in spite of the low activ-
ity. Subsequently, with the re-
duction of oxides in isobutane
atmosphere, aggregated metal
ensembles were formed in situ
and played a significant role in
cracking reactions, confirmed
earlier in the dehydrogenation
of isobutane over the NiO/
MgAl2O4 catalyst.[38] However,
Figure 2. Mass spectra of the pulse-mode isobutane dehydrogenation over silica-supported metal oxide catalysts.
much higher than that of Co3O4/SiO2. In addition, the coke
content in the first 10 min was the highest over the 13NiO/
SiO2 catalyst.
the active centers for cracking were gradually covered by the
large amount of coke generated, which results in weakened
cracking reactions; this observation was consistent with the re-
ported result[44] that the carbonaceous species formed on the
Ni particles led to a low cracking activity of the Ni/MoO3 cata-
lyst.
To better understand the differences in the dehydrogenation
performance of the 13Fe2O3/SiO2, 13Co3O4/SiO2, and 13NiO/
SiO2 catalysts, temperature-programmed reduction of hydro-
gen (H2-TPR) experiments were performed. As shown in
Figure 3, the reducibility of the three catalysts decreased in the
Formation of methane over metal oxide catalysts
Given the considerable amount of methane generated during
isobutane dehydrogenation over Fe-, Co-, and Ni-based cata-
lysts, the route for the formation of methane should be deter-
mined. Notably, metallic Ni facilitated CÀC bond scission in al-
kanes,[31,37,45] and Coughlin et al.[46,47] reported that large en-
sembles of group VIII metals were active for the hydrogenoly-
sis reaction. Moreover, Resasco et al.[48] believed that aggregat-
ed metallic Ni catalyzed the hydrogenolysis reaction of
isobutane and produced large amounts of methane and coke;
this finding was consistent with our results. Considering that
hydrogen has been introduced into the reactor in the work of
Resasco et al.,[48] the question to be answered is whether
methane is formed through the hydrogenolysis reaction with-
out the introduction of hydrogen.
Figure 3. H2-TPR profiles of silica-supported metal oxide catalysts.
To answer the above question, the pulse-mode isobutane
dehydrogenation over silica-supported metallic Ni catalysts
with and without exterior hydrogen was performed with
a homemade apparatus (Scheme 1); the method is described
in the Experimental Section. In the absence of exterior hydro-
gen (Figure 4a), with the increase in pulse number, the signal
intensity of methane (m/z=16) demonstrated an increasing
trend owing to the increase in the signal intensity of hydrogen
(m/z=2). In addition, the signals corresponding to isobutane
(m/z=58) and isobutene (m/z=56) were detected, even
though their intensities were too weak to be observed in the
original proportion. In contrast, in the presence of exterior hy-
drogen (Figure 4b), the intensities of hydrogen and methane
reached the maximum and remained unchanged with the in-
crease in pulse number. At the same time, the signals for iso-
butane and isobutene could not be detected in the pulse-
mode spectra, which indicated that isobutane was completely
following order: 13Co3O4/SiO2 >13NiO/SiO2 >13Fe2O3/SiO2,
which indicated that the reduction of Fe2O3 species was the
most difficult. The reduction temperatures of 13Co3O4/SiO2
and 13NiO/SiO2 were approximately 360 and 4008C, respec-
tively, which suggested that Co3O4 and NiO on the support
could easily be reduced below the reaction temperature of
5608C (see the dotted line in Figure 3). However, Fe2O3 species
could not be reduced completely to the metallic state at that
temperature. Therefore, the steady performance of the
13Fe2O3/SiO2 catalyst can be attributed to its relatively stable
redox properties, that is, it is difficult to reduce to a metallic
species at the reaction temperature. In addition, the cracking
ability of the three metal-based catalysts decreased in the fol-
lowing order: Ni>Co>Fe.
According to the above results, the transformation of the re-
action with time on stream over Co3O4/SiO2 and NiO/SiO2 cata-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 2305 – 2314 2307

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
fined as the moles of isobutene produced per mole of active
metal atoms per hour, and the number of active metal atoms
was determined through H2 pulse chemisorption. The TOF
values of metal sulfide catalysts were approximately ten times
larger than those of the corresponding oxide catalysts
(Table 1), suggesting that the dehydrogenation performance
was improved remarkably after sulfidation treatment.
As listed in Table 1, the selectivity toward isobutene in-
creased after sulfidation treatment, which was more than
85 wt% for all metal sulfide catalysts. Moreover, the catalytic
activity was maintained at a relatively high level. Isobutane
conversion reached up to 71 wt% for the 13CoS/SiO2 catalyst
and decreased in the following order: 13CoS/SiO2 >13FeS/
SiO2 >13NiS/SiO2. In addition, the coke content decreased sig-
nificantly after sulfidation, especially for the Ni-based catalyst
(Table 1). However, the excellent dehydrogenation performance
was not maintained and both the conversion of isobutane and
the selectivity toward isobutene decreased as the reaction pro-
ceeded (Figure 5). Such a phenomenon reveals possible loss
and consumption of the active metal sulfide during the
reaction.
Scheme 1. Representation of the apparatus used to investigate the hydroge-
nolysis behavior of the catalysts.
To characterize the catalyst deactivation rate, a deactivation
index, defined as the ratio of isobutane conversion at different
reaction times to the initial isobutane conversion, is used; that
is, a lower value represents much faster deactivation. The deac-
tivation index decreased in the following order: 13CoS/SiO2 >
13FeS/SiO2 >13NiS/SiO2 (Figure 6), which is consistent with
the order of isobutane conversion. These results suggested
that the catalyst with higher activity deactivated much faster
and was probably caused by the accelerated consumption of
the active metal sulfide and increased coke deposition on the
catalyst. In brief, an excellent dehydrogenation performance
could be achieved at the initial reaction time for the three sul-
fide catalysts, although minor differences existed.
Figure 4. Mass spectra of the pulse-mode tests over silica-supported metallic
Ni catalyst a) without and b) with exterior hydrogen.
Effect of the loading amount of metal sulfide catalysts
converted to methane through the hydrogenolysis reaction in
the presence of exterior hydrogen.
To investigate the effect of the active-component loading on
the dehydrogenation performance of metal sulfide catalysts,
sulfided NiO/SiO2 catalysts with different NiO loadings (0.5, 1,
5, 9, 17, and 21 wt%) were prepared and evaluated. As shown
in Figure 7, isobutane conversion increased gradually with the
In conclusion, the large amount of methane generated
during the catalysis of Fe-, Co-, and Ni-based catalysts resulted
from the hydrogenolysis reaction of isobutane. The hydroge-
nolysis reaction occurs even
without the introduction of hy-
drogen.
Performance of metal sulfide
catalysts
The catalytic performance of Fe-,
Co-, and Ni-based sulfide cata-
lysts in isobutane dehydrogena-
tion was also evaluated under
the same reaction conditions,
and the results are presented in
Table 1 and Figure 5. Herein,
turnover frequency (TOF) is de- Figure 5. Catalytic performance of silica-supported metal sulfide catalysts for isobutane dehydrogenation.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 2305 – 2314 2308

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
creased at the same time. During this process, the formation
rate of isobutene decreased slightly at first and then increased
gradually and accelerated at a low loading amount. On de-
creasing the NiO loading to 0.5 wt% the TOF value increased
to 12.38. Thus, both the activity of a single Ni atom and the
utilization efficiency of Ni species on the catalyst increased
with the decrease in NiO loading.
As discussed above, a relatively high conversion ability of
isobutane can still be obtained over supported NiS catalysts
even with a low Ni loading. However, the deficiency of the
active phase in bulk leads to a sharply decreasing catalytic
activity.
Figure 6. Deactivation index of silica-supported metal sulfide catalysts for
isobutane dehydrogenation.
Origin of catalyst deactivation
As discussed above, although an
excellent dehydrogenation per-
formance can be obtained for
metal sulfide catalysts, the cata-
lyst stability still needs to be im-
proved. Therefore, to explore ap-
propriate measures to recover
the catalytic activity, the origin
of the rapid catalyst deactivation
should be investigated.
Phase analysis was performed
for the three sulfide catalysts
after a 5 h reaction time, and
their XRD patterns are shown in
Figure 1c. For the spent 13FeS/
SiO2 catalyst, the intensity of dif-
fraction peaks characteristic of
FeS decreased as compared with
those for the fresh sulfide cata-
lyst. For the other two spent sul-
fide catalysts, Co9S8 and Ni3S2
Figure 7. Catalytic performance of sulfided NiO/SiO2 catalysts with different NiO loadings for isobutane
dehydrogenation.
emerged instead of the original
CoS and NiS, respectively, which
increase in the loading amount and reached the maximum of
Table 2. Surface Ni densities, Ni dispersions, and particle sizes of various
NiO/SiO2 catalysts after reduction at 773 K and TOF values of sulfided
NiO/SiO2 catalysts in isobutane dehydrogenation.
74.8 wt% at a NiO loading of 17 wt%. With further increase in
the loading amount, a minor decrease in isobutane conversion
was observed. Notably, isobutane conversion up to 54.3 wt%
could still be obtained with an NiO loading as low as 0.5 wt%.
However, the low amount of active species on the catalyst sur-
face was consumed quickly; thus, the activity decreased sharp-
ly with time on stream and only 19.5% of the initial activity
was reserved after 1 h reaction. In contrast, at higher loadings,
both the conversion of isobutane and the selectivity toward
isobutene decreased at a much slower rate with time on
stream. This decrease was probably caused by faster replenish-
ment of the consumed surface active sites by bulk NiS species.
Ni dispersion and TOF values (defined as above) of sulfided
NiO/SiO2 catalysts with different NiO loadings are listed in
Table 2. With the decrease in NiO loading, both surface Ni den-
sity and particle size decreased gradually and Ni dispersion in-
[b]
Sample[a]
Ni surface density
Ni dispersion[b]
[%]
dNi
TOF[c]
[hÀ1
[atomNi nmÀ2
]
[nm]
]
0.5NiO/SiO2
1NiO/SiO2
5NiO/SiO2
9NiO/SiO2
13NiO/SiO2
17NiO/SiO2
21NiO/SiO2
0.13
0.28
1.47
2.85
4.44
6.44
9.06
71.21
60.49
39.14
25.78
19.19
12.59
7.62
1.21
1.43
2.21
3.35
4.51
6.87
11.34
12.38
8.60
2.85
2.47
2.35
3.11
3.76
[a] Loading amount refers to the weight content of NiO in the as-pre-
pared catalysts; [b] Determined through H2 chemisorption. Ni dispersion
refers to the molar ratio of surface Ni and bulk Ni; the latter was obtained
through Ni(NO3)2·6H2O impregnation; [c] TOF calculated as the amount
of formed isobutene per hour divided by the amount of the active Ni
atom (determined through H2 chemisorption) supported on the catalyst.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 2305 – 2314 2309

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
suggested the loss of sulfur during the reaction. Although the
pathways of sulfur loss were different for the three sulfide cat-
alysts, the loss of sulfur probably caused the deactivation of
the catalyst.
conversion decreased to some extent in the initial reaction
period and then the catalytic performance remained relatively
steady, with an isobutene yield greater than 50 wt% for all
three samples in the continuous 7 h reaction. This finding indi-
cated that high catalytic performance for dehydrogenation can
be maintained with online replenishment of sulfur, which
again demonstrated the importance of sulfur.
An online mass spectrometer was also used to detect the re-
leased gases during isobutane dehydrogenation over the
13NiS/SiO2 catalyst. In addition to the dehydrogenation prod-
uct, a signal corresponding to H2S (m/z=34) was detected
(data not shown). This observation confirms the loss of sulfur
during the reaction as well as suggesting that it was lost
mainly in the form of H2S, which is consistent with our previ-
ous report of isobutane dehydrogenation over Ni/MgAl2O4
catalysts.[38]
In addition to sulfur loss, coke deposition inevitably occurred
and possibly contributed to catalyst deactivation. The tempera-
ture-programmed oxidation by oxygen (O2-TPO) test of the
spent NiS/SiO2 catalyst was performed to investigate the coke
deposited during the reaction; meanwhile, an online mass
spectrometer was used to detect the released CO2 (m/z=44),
and the result is presented in Figure 10a. The appearance of
two distinct peaks (327 and 4648C) corresponding to CO2 for-
mation during the coke-burning process indicated that at least
To supplement sulfur and recover the activity of the spent
sulfide catalysts, a sulfidation treatment was performed for an-
other 3 h after the 5 h reaction, and a total of five sulfidation–
reaction cycles were performed. The initial conversion of isobu-
tane and selectivity toward isobutene obtained in different
cycles over the three sulfide catalysts are shown in Figure 8.
Notably, the catalytic performance could be completely recov-
ered after each sulfidation, demonstrating that sulfidation
treatment is an appropriate approach to maintain the activity
of metal sulfide catalysts. Moreover, isobutane dehydrogena-
tion over metal sulfide catalysts, along with a co-feed of H2S
(2 mLminÀ1), was performed. As shown in Figure 9, isobutane
Figure 10. Mass spectra obtained during O2-TPO and H2-TPR of different NiS/
SiO2 catalysts.
Figure 8. Catalytic performance of metal sulfide catalysts in different sulfida-
tion–reaction cycles.
two types of coke species exist-
ed on the catalyst.
Full recovery of the activity of
the catalyst through sulfidation
not only indicated that sulfur
loss was one reason for the cata-
lyst deactivation, but also meant
that coke deposition could prob-
ably be eliminated through sulfi-
dation with H2S/H2. To investi-
gate the above speculation, the
H2-TPR test of the spent NiS/SiO2
catalyst was performed, and the
corresponding MS spectra are
Figure 9. Catalytic performance of metal sulfide catalysts along with a co-feed of diluted H2S.
shown in Figure 10b. The ap-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 2305 – 2314 2310

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
pearance of methane (m/z=16) during the H2-TPR process
suggested that the presence of hydrogen eliminated the coke
deposition to some extent. After the H2-TPR test, the O2-TPO
test of the spent catalyst was performed, and the correspond-
ing mass spectra are shown in Figure 10c. Compared with the
spent catalyst without the H2-TPR process (Figure 10a), the
high-temperature peak in the mass spectra was generally re-
tained; however, the low-temperature peak disappeared. These
results revealed that the deposition of coke species burnt at
low temperature (below 4008C) was probably the other cause
of catalyst deactivation whereas coke species burnt at high
temperature had little effect on the catalytic activity.
In contrast, the bonding strength between the adsorbed in-
termediate and the catalyst surface also affected the reactivity
significantly. The weak bonding interaction facilitated the de-
sorption of the intermediate and reduced the reactivity where-
as the strong bonding interaction led to the strengthened
carbon–metal bond and weakened carbon–carbon bond in al-
kanes,[60] finally resulting in an aggravated hydrogenolysis reac-
tion and formation of methane. The adsorption amounts of
isobutane and isobutene over reduced oxide catalysts (i.e.,
metal catalysts) and sulfided catalysts are listed in Table 3. As
Table 3. Comparison of amounts of isobutane and isobutene adsorbed
over reduced oxide catalysts and sulfided catalysts at 3008C.
Reaction mechanisms
À1
Sample
Amount adsorbed [mLg
]
cat
Compared with the corresponding oxide catalysts, the remark-
able improvement in the dehydrogenation performance of
metal sulfide catalysts is surely not only caused by the modifi-
cation of physicochemical properties (adsorption–desorption
and redox behaviors, etc.), but can also be fundamentally at-
tributed to the emergence of metal sulfides, which have been
speculated to be the active sites of sulfide catalysts for dehy-
drogenation.[38]
Isobutane
Isobutene
13NiO/SiO2
reduced
sulfided
13Co3O4/SiO2
reduced
sulfided
13Fe2O3/SiO2
reduced
sulfided
1.18
0.59
3.75
0.23
0.72
0.65
1.38
0.25
0.29
0.62
0.29
0.26
According to the literature, the hydrogenolysis reaction of
alkanes is structure sensitive[49–51] and requires a large amount
of aggregated metal sites to constitute the active phase.[48] It is
also believed that the hydrogenolysis activity is closely related
to the size of metal particles.[52–56] Dalmon and Martin[54] report-
ed that at least nineteen neighboring Ni atoms on the catalyst
surface are necessary for the hydrogenolysis of butanes. Over
Fe-, Co-, and Ni-based catalysts, the aggregated metallic sites
formed in situ led to the generation of a large amount of
methane owing to the hydrogenolysis reaction. Moreover, with
the decrease in loading amount, the size of metal particles de-
creased and thus the activity of the hydrogenolysis reaction
also decreased.[38] Nevertheless, the dehydrogenation reaction
seems to be structure insensitive, which is reflected by the
high initial isobutane conversion over the sulfided Ni-based
catalyst with NiO loading as low as 0.5 wt%. Moreover, Biloen
et al.[57] reported that the dehydrogenation reaction can be cat-
alyzed even by a single metal-atom site.
for metal catalysts, the adsorption amounts of both isobutane
and isobutene decreased in the following order: Ni>Co>Fe,
which was consistent with the order of cracking ability. There-
fore, this result explained the high ability of the Ni-based cata-
lyst for activating isobutane and the corresponding high isobu-
tane conversion. Meanwhile, given the low adsorption amount
of isobutene over the Fe-based catalyst, the formed isobutene
could desorb from the catalyst promptly and the high selectivi-
ty toward isobutene could also be understood. Compared with
metal catalysts, the adsorption amount of isobutane over sul-
fide catalysts was moderate for all three cases. However, the
adsorption amount of isobutene decreased significantly after
sulfidation treatment, especially for Ni- and Co-based catalysts.
Such a phenomenon indicated that the introduction of sulfur
facilitated the desorption of olefin; that is, the interaction be-
tween surface metal atoms and adsorbed alkene molecules
was weakened by the presence of neighboring sulfur atoms.
One possible reason is that a higher electron density of surface
atoms, caused by the electron donation of introduced sulfur,
led to more repulsive interactions with olefins. That is, the in-
troduction of sulfur probably adjusted the electronic properties
of Ni atoms, altered the nature of surface species, and de-
creased the adsorption heat of olefin as well as the activation
energy of its desorption, consequently leading to an increased
selectivity toward isobutene.
The activity results indicated that the introduction of sulfur
species played a significant role in the dehydrogenation reac-
tion, and thus it is important to clarify how the sulfur species
works. We have reported previously that the introduction of
sulfur facilitated the dispersion of Ni particles on the MgAl2O4
support.[38] Moreover, the existence of the metal–sulfur bond in
metal sulfide catalysts weakened the interaction between
metal atoms. Therefore, the effect of sulfur introduction can be
explained in terms of the geometric effect, which dilutes the
aggregated Ni sites on the catalyst surface and breaks large
metallic Ni ensembles active for hydrogenolysis. Consequently,
the occurrence of hydrogenolysis and coking reactions is re-
duced. Such an effect is similar to that of Sn introduction into
Pt-based catalysts; Sn acts as a spacer, isolates Pt particles into
smaller ensembles, reduces the deactivation rate, and increases
the selectivity for dehydrogenation reactions.[58,59]
The interaction model between the isobutane molecule and
the catalyst surface before and after sulfidation is demonstrat-
ed in Scheme 2. As for metal oxide catalysts, the oxides on the
support are readily reduced to metal atoms during the reac-
tion. A strong interaction between carbon atoms in isobutane
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 2305 – 2314 2311

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 11. Intermediate species in the proposed catalytic cycle.
Scheme 2. Interaction model between the isobutane molecule and the cata-
lyst surface.
meanwhile, the active center is regenerated. If the NiÀS bond
weakens, H2S is generated and released, and the Ni center
with one less bonded sulfur atom (3, Figure 11) demonstrates
much lower dehydrogenation activity (Scheme 3, red arrows).
The catalytic activity of the spent catalyst can be recovered by
sulfur replenishment through sulfidation with H2S/H2
(Scheme 3, blue arrows).
molecules and the metal atoms weakens and breaks carbon–
carbon bonds in isobutane and thus leads to the generation of
a significant amount of methane. However, with the introduc-
tion of sulfur, the aggregated metallic species are diluted and
separated, further reducing the possibility for the formation of
the adjacent carbon–metal bonds in an isobutane molecule. In
addition, the hydrogen atoms in isobutane molecules bond
with sulfur atoms neighboring metallic species on the catalyst
surface and thus the carbon–hydrogen bonds weaken and
break. Subsequently, isobutene and hydrogen are released to
the gas phase as the final products.
On the basis of the catalytic cycle proposed above, a patent
has been filed for a circulating fluidized-bed unit with a reactor
for dehydrogenation, a regenerator for coke burning, and a sul-
fidation section for catalyst activation.[64,65]
Conclusions
To better understand the active sites, the reaction mecha-
nism of isobutane dehydrogenation over metal sulfide cata-
lysts has been explored further. For example, in the NiS/SiO2
catalyst, NiS has a hexagonal structure (JCPDS no. 01-075-
0613) as indicated by the XRD patterns. Moreover, for the hy-
drogenation reaction catalyzed by metal sulfide catalysts, the
active sites are generally believed to be sulfur vacancies, that
is, the coordinatively unsaturated sites.[61–63] Therefore, the
active Ni atom is speculated to be bonded with three sulfur
atoms. Accordingly, a catalytic cycle of sulfide catalysts, includ-
ing reaction, deactivation, and regeneration, is proposed,
which is illustrated in Scheme 3. An isobutane molecule diffus-
es to the catalyst surface and dissociatively adsorbs on the co-
ordinatively unsaturated Ni sites, forming an intermediate spe-
cies (1, Figure 11) with a Ni atom bonded to the tertiary butyl
group and a sulfur atom bonded to the left hydrogen atom.
Through b-hydrogen transfer, a hydrogen molecule is formed
and desorbs to the gas phase, leaving a labile intermediate (2,
Figure 11) bonded to a p-bonded alkene. Then, an isobutene
molecule is evolved and released from the catalyst surface;
Silica-supported metal oxide and sulfide catalysts of Fe, Co,
and Ni have been evaluated comparatively for the catalytic de-
hydrogenation of isobutane. The combined results of the activ-
ity test and catalyst characterization indicate that metal oxides,
except Fe2O3 on silica, are easily reduced to metal ensembles
active for alkane hydrogenolysis, which results in the formation
of a large amount of methane and coke.
Metal sulfide catalysts have been prepared by sulfonation of
the corresponding oxide catalysts with H2S/H2. After the intro-
duction of sulfur, the hydrogenolysis reaction was inhibited ef-
fectively and the dehydrogenation performance was improved
significantly. The selectivity toward isobutene of all the tested
catalyst samples could reach up to 85 wt%, and isobutane
conversion over the Co-based catalyst increased to 71 wt%. In
general, the promoting effects of sulfur introduction can be
understood in two ways: Firstly, the geometric effect dilutes
the aggregated metallic species and inhibits the CÀC bond
rupture; secondly the electronic effect facilitates the desorp-
tion of olefins and increases the product selectivity.
Furthermore, the mechanisms of dehydrogenation reaction,
catalyst deactivation, and regen-
eration have been explored by
using the speculated model of
the interaction between isobu-
tane molecules and sulfide cata-
lysts. Finally, sulfur loss in the
form of H2S and the deposition
of coke species burnt at low
temperature were determined to
be the main reasons for catalyst
deactivation.
Scheme 3. Proposed catalytic cycle (including reaction, deactivation, and regeneration) of the NiS/SiO2 catalyst for
isobutane dehydrogenation.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 2305 – 2314 2312

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
after the catalyst reduction process hydrogen was fed to the cata-
lyst bed together with isobutane without switching to helium in
the pulse-mode test.
Experimental Section
Catalyst preparation
Silica-supported oxide catalysts of Fe, Co, and Ni were prepared
through an incipient wetness impregnation of mesoporous silica
(80–180 mm) with aqueous solutions of Fe(NO3)3·9H2O, Co-
(NO3)2·6H2O, and Ni(NO3)2·6H2O, respectively. The prepared cata-
lysts were dried in air at 1208C overnight and then calcined at
5608C in air for 2 h, and the as-prepared samples were labeled as
aMOx/SiO2, in which a was the nominal MOx content in wt% and
M represented Fe, Co, or Ni. Subsequently, the supported metal
oxide catalysts were sulfided in a flow of 10 vol% H2S/H2 (flow
rate: 30 mLminÀ1) at 5608C for 3 h and the corresponding metal
sulfide catalysts were obtained, labeled as aMSx/SiO2.
Catalytic activity test
Catalytic activity tests were performed in a conventional fixed-bed
microreactor at 5608C under atmospheric pressure, with 4 g of the
catalyst loaded into the reactor. To remove adsorbed oxygen and
moisture from the catalyst, a flow of nitrogen was introduced into
the reactor before the reaction. During the test, a mixture of isobu-
tane and nitrogen (molar ratio: 1:6; total flow rate: 14 mLminÀ1
)
was fed into the reactor. The compositions of the reactant and the
effluent gaseous product were determined with a Bruker 450 gas
chromatograph equipped with three detectors, which included
a flame ionization detector to analyze the composition of hydro-
carbons and two thermal conductivity detectors to measure the
contents of H2, N2, CO, and CO2. In addition, the amount of coke
deposited on the catalyst after the reaction was evaluated by ana-
lyzing the amount of CO2 generated during in situ coke combus-
tion with the chromatogram. Mass balances of all tests were above
95 wt%.
Catalyst characterization
Nitrogen adsorption–desorption measurements were performed
with a Quadrasorb SI instrument (Quantachrome Corporation, USA)
at liquid nitrogen temperature to determine the BET specific sur-
face area and pore properties of the catalyst samples. To ensure
complete removal of the adsorbed moisture, all samples were
evacuated at 3008C and 1.0ꢁ10À3 kPa for 4 h before the measure-
ment. The XRD measurements of the samples was performed with
an X’Pert PRO MPD diffractometer system (Philips, The Nether-
lands) to analyze the crystalline structure of the catalysts by using
CuKa radiation at 40 kV and 40 mA. Moreover, the H2-TPR experi-
ment was performed to investigate the redox behavior of the sam-
ples, and detailed operational methods have been provided in our
published work.[24]
Acknowledgements
This work was financially supported by the National Natural Sci-
ence Foundation of China (no. U1362201) and the National 973
Program of China (no. 2012CB215006).
Keywords: dehydrogenation · isobutane · metal oxides · metal
sulfides · supported catalysts
Metal dispersion and particle size were measured through H2 pulse
chemisorption with a FINESORB-3010 chemisorption analyzer (Fine-
tec Instruments, China). The catalyst (ꢀ0.3 g) was reduced in situ
under a flow of 10 vol% H2/N2 (flow rate: 30 mLminÀ1) at 773 K for
1 h and then cooled to 1008C in argon atmosphere. Then, H2/N2
was pulsed into the catalyst bed to determine metal dispersion.
[1] B. M. Weckhuysen, I. E. Wachs, R. A. Schoonheydt, Chem. Rev. 1996, 96,
3327–3350.
[2] B. M. Weckhuysen, R. A. Schoonheydt, Catal. Today 1999, 51, 223–232.
[3] P. Sun, G. Siddiqi, M. Chi, A. T. Bell, J. Catal. 2010, 274, 192–199.
[4] E. Rombi, M. G. Cutrufello, V. Solinas, S. De Rossi, G. Ferraris, A. Pistone,
Appl. Catal. A 2003, 251, 255–266.
[5] G. Neri, A. Pistone, S. De Rossi, E. Rombi, C. Milone, S. Galvagno, Appl.
Catal. A 2004, 260, 75–86.
[6] M. L. Casella, G. J. Siri, G. F. Santori, O. A. Ferretti, M. M. Ramꢂrez-Corre-
dores, Langmuir 2000, 16, 5639–5643.
[7] M. D. Argyle, K. Chen, A. T. Bell, E. Iglesia, J. Catal. 2002, 208, 139–149.
[8] K. Chen, S. Xie, A. T. Bell, E. Iglesia, J. Catal. 2000, 195, 244–252.
[9] J. Le Bars, A. Auroux, M. Forissier, J. C. Vedrine, J. Catal. 1996, 162, 250–
259.
[10] B. Schimmoeller, Y. Jiang, S. E. Pratsinis, A. Baiker, J. Catal. 2010, 274,
64–75.
To compare adsorption amounts of isobutane or isobutene over
metal catalysts and the corresponding sulfide catalysts, pulse
chemisorption tests were performed using the same equipment
used for H2 chemisorption. The catalyst (ꢀ0.5 g) was loaded into
a quartz tube reactor. After the prereduction of oxide catalysts and
pretreatment of sulfide catalysts, isobutane or isobutene
(0.803 mL) was pulsed into the catalyst bed with a flow of nitrogen
(flow rate: 30 mLminÀ1) at 3008C. An online GAS-100Q quadrupole
mass spectrometer (AVI GmbH, Switzerland) was used to detect
signals of the released hydrocarbons.
To investigate the hydrogenolysis behavior of metal catalysts, the
FINESORB-3010 chemisorption analyzer was adopted (a schematic
of the apparatus is shown in Scheme 1). For each test, the catalyst
(ꢀ0.1 g) was loaded into a quartz tube reactor and preheated in
a flow of helium (flow rate: 30 mLminÀ1) at 2008C for 0.5 h. Subse-
quently, with a four-way valve, the helium flow was switched to
a mixture of 10 vol% H2/N2 (flow rate: 30 mLminÀ1) to reduce the
metal oxides on the support to metallic species, with the tempera-
ture increased from 200 to 5608C (heating rate: 108CminÀ1). After
the reduction process, the four-way valve was switched back to
the helium flow to purge the catalyst bed. Finally, isobutane was
injected into the pulse-mode reactor with a six-way valve and the
gas released from the reactor was analyzed with an online GAS-
100Q quadrupole mass spectrometer. Moreover, to investigate the
hydrogenolysis reaction with the introduction of exterior hydrogen,
[11] A. Khodakov, J. Yang, S. Su, E. Iglesia, A. T. Bell, J. Catal. 1998, 177, 343–
351.
[12] L. Cheng, G. A. Ferguson, S. A. Zygmunt, L. A. Curtiss, J. Catal. 2013,
302, 31–36.
[13] M. Abello, M. Gomez, O. Ferretti, Appl. Catal. A 2001, 207, 421–431.
[14] N. Haddad, E. Bordes-Richard, A. Barama, Catal. Today 2009, 142, 215–
219.
[15] K. Chen, S. Xie, E. Iglesia, A. T. Bell, J. Catal. 2000, 189, 421–430.
[16] M. Høj, T. Kessler, P. Beato, A. D. Jensen, J.-D. Grunwaldt, Appl. Catal. A
2014, 472, 29–38.
[17] W. Liu, S. Y. Lai, H. Dai, S. Wang, H. Sun, C. T. Au, Catal. Today 2008, 131,
450–456.
[18] A. Dejoz, J. M. Lꢃpez Nieto, F. Mꢄrquez, M. I. Vꢄzquez, Appl. Catal. A
1999, 180, 83–94.
[19] M. Chen, J. Xu, Y. Cao, H. Y. He, K. N. Fan, J. H. Zhuang, J. Catal. 2010,
272, 101–108.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 2305 – 2314 2313

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[20] M. Chen, J. Xu, F.-Z. Su, Y.-M. Liu, Y. Cao, H.-Y. He, K.-N. Fan, J. Catal.
2008, 256, 293–300.
[43] L. Zhang, J. C. Yu, M. Mo, L. Wu, Q. Li, K. W. Kwong, J. Am. Chem. Soc.
2004, 126, 8116–8117.
[21] F. Dury, M. A. Centeno, E. M. Gaigneaux, P. Ruiz, Appl. Catal. A 2003, 247,
231–246.
[44] S. David Jackson, G. J. Kelly, G. Webb, Phys. Chem. Chem. Phys. 1999, 1,
2581–2587.
[22] J. Ding, Z. Qin, X. Li, G. Wang, J. Wang, J. Mol. Catal. A 2010, 315, 221–
225.
[23] S. Sokolov, M. Stoyanova, U. Rodemerck, D. Linke, E. V. Kondratenko, J.
Catal. 2012, 293, 67–75.
[24] G. Wang, C. Li, H. Shan, W. Wu, Ind. Eng. Chem. Res. 2013, 52, 13297–
13304.
[45] C. R. Carrera Cerritos, F. R. Ramꢂrez, A. A. F. Alvarado, M. n. J. M. Rosales,
V. T. Garcꢂa, G. I. R. Esquivel, Ind. Eng. Chem. Res. 2011, 50, 2576–2584.
[46] R. W. Coughlin, K. Kawakami, A. Hasan, J. Catal. 1984, 88, 150–162.
[47] R. W. Coughlin, A. Hasan, K. Kawakami, J. Catal. 1984, 88, 163–176.
[48] D. E. Resasco, B. K. Marcus, C. S. Huang, V. A. Durante, J. Catal. 1994,
146, 40–55.
[25] K. Chen, E. Iglesia, A. T. Bell, J. Catal. 2000, 192, 197–203.
[26] K. Chen, A. Khodakov, J. Yang, A. T. Bell, E. Iglesia, J. Catal. 1999, 186,
325–333.
[27] L. Halle, R. Houriet, M. M. Kappes, R. H. Staley, J. Beauchamp, J. Am.
Chem. Soc. 1982, 104, 6293–6297.
[28] G. Byrd, R. Burnier, B. Freiser, J. Am. Chem. Soc. 1982, 104, 3565–3569.
[29] M. Polyakov, M. W. E. van den Berg, T. Hanft, M. Poisot, W. Bensch, M.
Muhler, W. Grꢅnert, J. Catal. 2008, 256, 126–136.
[30] R. B. Freas, J. E. Campana, J. Am. Chem. Soc. 1986, 108, 4659–4661.
[31] R. Georgiadis, E. R. Fisher, P. Armentrout, J. Am. Chem. Soc. 1989, 111,
4251–4262.
[49] J. Z. Zhang, Y.-T. Tsai, K. L. Sangkaewwattana, J. G. Goodwin, Jr., J. Catal.
2011, 280, 89–95.
[50] G. S. Ranhotra, A. T. Bell, J. A. Reimer, J. Catal. 1987, 108, 40–49.
[51] R. J. Lobo-Lapidus, B. C. Gates, J. Catal. 2009, 268, 89–99.
[52] M. F. Guilleux, J. A. Dalmon, G. A. Martin, J. Catal. 1980, 62, 235–242.
[53] I. Alstrup, U. E. Petersen, J. R. Rostrup-Nielsen, J. Catal. 2000, 191, 401–
408.
[54] J. A. Dalmon, G. A. Martin, J. Catal. 1980, 66, 214–221.
[55] J. L. Carter, J. A. Cusumano, J. H. Sinfelt, J. Phys. Chem. 1966, 70, 2257–
2263.
[56] G. A. Martin, J. Catal. 1979, 60, 345–355.
[32] P. A. Van Koppen, J. Brodbelt-Lustig, M. T. Bowers, D. V. Dearden, J. Beau-
champ, E. R. Fisher, P. Armentrout, J. Am. Chem. Soc. 1990, 112, 5663–
5665.
[33] M. R. A. Blomberg, P. E. M. Siegbahn, U. Nagashima, J. Wennerberg, J.
Am. Chem. Soc. 1991, 113, 424–433.
[34] R. H. Schultz, J. Elkind, P. Armentrout, J. Am. Chem. Soc. 1988, 110, 411–
423.
[35] P. A. van Koppen, M. T. Bowers, C. L. Haynes, P. Armentrout, J. Am. Chem.
Soc. 1998, 120, 5704–5712.
[57] P. Biloen, F. M. Dautzenberg, W. M. H. Sachtler, J. Catal. 1977, 50, 77–86.
[58] O. A. Bariꢆs, A. Holmen, E. A. Blekkan, J. Catal. 1996, 158, 1–12.
[59] B. M. Nagaraja, C.-H. Shin, K.-D. Jung, Appl. Catal. A 2013, 467, 211–223.
[60] J. H. Sinfelt, J. Carter, D. Yates, J. Catal. 1972, 24, 283–296.
[61] P. G. Moses, B. Hinnemann, H. Topsøe, J. K. Nørskov, J. Catal. 2009, 268,
201–208.
[62] T. C. Ho, J. M. McConnachie, J. Catal. 2011, 277, 117–122.
[63] H. Wang, E. Iglesia, J. Catal. 2010, 273, 245–256.
[64] C. Li, G. Wang, N. Sun, Y. Sun, (China University of Petroleum), C.N.
201310014789.1, 2013.
[36] P. A. M. Van Koppen, P. Kemper, M. T. Bowers, J. Am. Chem. Soc. 1992,
114, 1083–1084.
[65] C. Li, G. Wang, N. Sun, Y. Sun, (China University of Petroleum), C.N.
201210536414.7, 2012.
[37] D. Jacobson, B. Freiser, J. Am. Chem. Soc. 1983, 105, 5197–5206.
[38] G. Wang, Z. Meng, J. Liu, C. Li, H. Shan, ACS Catal. 2013, 3, 2992–3001.
[39] Y.-n. Sun, L. Tao, T. You, C. Li, H. Shan, Chem. Eng. J. 2014, 244, 145–151.
[40] G. Wang, C. Li, H. Shan, ACS Catal. 2014, 4, 1139–1143.
[41] S. E. Skrabalak, K. S. Suslick, J. Am. Chem. Soc. 2005, 127, 9990–9991.
[42] N. A. Dhas, A. Ekhtiarzadeh, K. S. Suslick, J. Am. Chem. Soc. 2001, 123,
8310–8316.
Received: March 27, 2014
Published online on July 7, 2014
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 2305 – 2314 2314