Catalytic conversion of syngas into C2+ oxygenates over Rh/SiO2-based catalysts: The remarkable effect of hydroxyls on the SiO2

Article abstract of DOI:10.1016/j.molcata.2012.10.022

A series of Rh-Mn-Li catalysts supported on SiO₂ prepared by different methods were tested for the synthesis of C₂₊ oxygenates via CO hydrogenation. The catalysts were comprehensively characterized by N ₂ adsorption-desorption, XRD, in situ FT-IR, CO-TPD, TPSR, and H ₂-TPR. It was found that the Rh-Mn-Li catalyst supported on the SiO₂ synthesized by the St?ber method exhibited the highest CO conversion and selectivity toward C₂₊ oxygenates compared with other catalysts. The investigation based on the catalytic performance and characterizations of the catalysts suggests that the hydroxyl-metal interaction over the catalysts supported on different SiO₂ results in different desorption behavior of the adsorbed CO. The specific interaction between the active components and weakly H-bonded hydroxyls on the SiO₂ prepared by the St?ber method promoted the transformation from Rh ⁺(CO)₂ to Rh-CO and facilitated the desorption/reactivity of adsorbed CO as a function of temperature in the presence of hydrogen, which were proposed to be the crucial factors for obtaining higher CO conversion and C₂₊ oxygenates selectivity.

Full text of DOI:10.1016/j.molcata.2012.10.022

Journal of Molecular Catalysis A: Chemical 367 (2013) 38–45
Contents lists available at SciVerse ScienceDirect
Journal of Molecular Catalysis A: Chemical
journal homepage: www.elsevier.com/locate/molcata
Catalytic conversion of syngas into C2+ oxygenates over Rh/SiO2-based catalysts:
The remarkable effect of hydroxyls on the SiO2
Jun Yua,b, Dongsen Maob,∗, Lupeng Hanb, Qiangsheng Guob, Guanzhong Lua,b,∗∗
a Key Laboratory for Advanced Material and Research Institute of Industrial Catalysis, East China University of Science and Technology, Shanghai 200237, PR China
b Research Institute of Applied Catalysis, School of Chemical and Environmental Engineering, Shanghai Institute of Technology, Shanghai 201418, PR China
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 23 June 2012
Received in revised form 12 October 2012
Accepted 18 October 2012
Available online 10 November 2012
A series of Rh–Mn–Li catalysts supported on SiO2 prepared by different methods were tested for the
synthesis of C2+ oxygenates via CO hydrogenation. The catalysts were comprehensively characterized by
N2 adsorption–desorption, XRD, in situ FT-IR, CO-TPD, TPSR, and H2-TPR. It was found that the Rh–Mn–Li
catalyst supported on the SiO2 synthesized by the Stöber method exhibited the highest CO conversion
and selectivity toward C2+ oxygenates compared with other catalysts. The investigation based on the
catalytic performance and characterizations of the catalysts suggests that the hydroxyl–metal interaction
over the catalysts supported on different SiO2 results in different desorption behavior of the adsorbed
CO. The specific interaction between the active components and weakly H-bonded hydroxyls on the SiO2
prepared by the Stöber method promoted the transformation from Rh+(CO)2 to Rh–CO and facilitated the
desorption/reactivity of adsorbed CO as a function of temperature in the presence of hydrogen, which
were proposed to be the crucial factors for obtaining higher CO conversion and C2+ oxygenates selectivity.
Keywords:
SiO2
Rh-based catalyst
CO hydrogenation
C2+ oxygenates
FT-IR
© 2012 Elsevier B.V. All rights reserved.
1. Introduction
different properties of SiO2 have a considerable influence on the
Rh-based catalysts, which finally leads to disparities in the cat-
The synthesis of ethanol and other C2 oxygenates (e.g., acetalde-
hyde and acetic acid) from syngas (CO + H2), which was derived
from coal or biomass, has attracted much attention in recent years
because it can decrease the global dependence on petroleum [1–4].
In spite of the continuous amount of research on this catalytic con-
version route, no commercial process exists due to the relatively
low selectivity and yield of C2 oxygenates using most known cata-
lysts [1]. Therefore, it is essential to develop more effective catalysts
for this catalytic process.
Rh-based catalysts have been known to be the most efficient
ones in the synthesis of C2 oxygenates from CO hydrogenation
due to their unique CO adsorption behavior [5–12]. The influence
of various supports on the catalytic activity of Rh-based cata-
lysts has been widely investigated [13–19]. SiO2 has been the
most frequently used support, since the Rh-based catalysts sup-
ported on it usually exhibit moderate activity and good selectivity
toward C2 oxygenates from CO hydrogenation. Furthermore, the
alytic performance [20–24]. For example, Chen et al. [21] observed
that the crystallite size distribution was narrower and Rh parti-
cles were more homogeneously dispersed when 14–20 mesh silica
was used instead of 20–40 mesh as a support for Rh–Mn–Li cata-
lyst. As a result, the space time yield (STY) and selectivity of C2+
oxygenates were significantly increased from 338.6 g/(kg h) and
49.2% to 618.4 g/(kg h) and 54.6%, respectively (T = 300 ◦C, P = 3 MPa,
SV = 12,500 h−1). Treatment of silica support with nC1–C5 alcohols
also resulted in an enhancement of the STY of C2 oxygenates by
10–30% depending on the specific alcohol [22]. This enhancement
was attributed to the improved Rh dispersion and the increased
ratio of Rh+/Rh0 sites on the catalyst surface, which facilitated the
CO insertion reaction. Moreover, it is also considered that surface
hydroxyl groups on silica play an important role in the structural
rearrangement of the metal aggregates and the change in the oxi-
dation state of the metal. Solymosi et al. [23] proposed that Rh+
was most probably formed via an oxidation of the Rh0 clusters
by the surface OH groups of SiO2. Basu et al. [24] reported that
specific OH groups on Al2O3 and SiO2 are consumed as CO inter-
acts with supported Rh crystallites to produce atomically dispersed
Rh+(CO)2. All these results showed the great significance of the sur-
face properties of SiO2 for the Rh-based catalysts, in particular the
role of surface hydroxyl groups. Nevertheless, to the best of our
knowledge, there have been no further reports of the effect of dif-
ferent surface hydroxyl groups over silica supports on the catalytic
∗
Corresponding author. Tel.: +86 21 6087 7221; fax: +86 21 6087 7231.
Corresponding author at: Research Institute of Applied Catalysis, School
∗∗
of Chemical and Environmental Engineering, Shanghai Institute of Technology,
Shanghai 201418, PR China. Tel.: +86 21 6087 7221; fax: +86 21 6087 7231.
E-mail addresses: dsmao1106@yahoo.com.cn, dsmao@sit.edu.cn (D. Mao),
gzhlu@ecust.edu.cn (G. Lu).
1381-1169/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.molcata.2012.10.022

J. Yu et al. / Journal of Molecular Catalysis A: Chemical 367 (2013) 38–45
39
activity of Rh for CO hydrogenation. And, it is quite possible that the
influences of the different kinds of hydroxyl groups on the catalytic
activity of Rh-based catalysts may be different.
detector). The conversion of CO was calculated based on the frac-
tion of CO that formed carbon-containing products according to:
%Conversion = (ꢀniMi/MCO) × 100, where ni is the number of car-
bon atom in product i, Mi is the percentage of product i detected,
and MCO is the percentage of carbon monoxide in the syngas feed.
The selectivity of a certain product was calculated based on car-
bon efficiency using the formula niCi/ꢀniCi, where ni and Ci are the
carbon atom number and molar concentration of the ith product,
respectively.
In this work, in order to study the role of silica surface hydroxyl
groups in the synthesis of C2 oxygenates during CO hydrogena-
tion, different SiO2 were employed in serving as the supports for
Rh-based catalysts. Here we took a well known recipe of multi-
component promoted Rh catalyst including Mn and Li [21], as a
part of screening for catalysts with a high activity and selectivity
of C2 oxygenates. The catalysts were extensively characterized and
evaluated for the conversion of syngas to oxygenates. Especially,
the correlation between the surface hydroxyl groups on SiO2 and
catalytic performance of the Rh/SiO2-based catalysts was discussed
in detail for the first time.
2.3. Catalyst characterization
The specific surface area (SBET), pore volume (Vp), and pore
diameter (Dp) of the samples were obtained using N2 adsorption
at −196 ◦C in a Micromeritics ASAP 2020 M+C adsorption appa-
ratus. Prior to N2 adsorption, the samples were degassed under a
vacuum of 10−1 Pa for 10 h at 200 ◦C.
2. Experimental
The X-ray powder diffraction (XRD) spectra of the samples were
obtained on a PANalytical X’Pert diffractometer operating with Ni
␤-filtered Cu K␣ (ꢁ = 0.15418 nm) radiation at 40 kV and 40 mA.
Two theta angles ranged from 10◦ to 85◦ with a scanning rate of 6
per minute. Transmission electron microscopy (TEM) images were
obtained on JEM-2100 instrument operated at 200 kV.
The amount of hydrogen adsorption of various catalysts was cal-
culated on the basis of H2-TPD profiles. For H2-TPD measurements,
the catalyst (0.1 g) was reduced in situ for 2 h at 400 ◦C in 10% H2/N2,
and then was held at 400 ◦C for another 30 min before being cooled
down to room temperature (RT) in He flow. The next step was H2
adsorption at RT for 0.5 h, and then the gas was swept again with
He for 3 h. Subsequently, the sample was heated in a flowing He
stream (50 mL/min) up to ∼500 ◦C at a rate of 10 ◦C/min, while the
desorbed species was detected with a TCD detector. The uptake
of H2 was used to calculate Rh metal dispersion and particle size,
assuming that each surface metal atom adsorbs one H atom, i.e.
H/Rhsurface = 1.
CO adsorption was studied using a Nicolet 6700 FT-IR spectrom-
eter equipped with a DRIFT (diffuse reflectance infrared Fourier
transform) cell with CaF2 windows. The sample in the cell was pre-
treated in 10% H2/N2 at 400 ◦C for 2 h, and then the temperature was
dropped to RT. After the cell was outgassed in vacuum to <10−3 Pa,
the background was recorded. Following by introducing CO into the
IR cell (pCO = 8.0 × 103 Pa), the IR spectrum of CO adsorbed on the
catalyst was recorded. Then the 10% H2/N2 was introduced again,
and the IR spectrum of CO adsorbed was recorded at different tem-
peratures. The concentration of CO was higher than 99.97%, and it
was pretreated by dehydration and deoxygenization before being
used. The spectral resolution was 4 cm−1 and the number of scans
was 64.
2.1. Catalyst preparation
SiO2 was prepared by the Stöber method [25] as follows. The
mixture solution of 21 mL tetraethylorthosilicate (TEOS) (99.5%,
SCRC) and 50 mL anhydrous ethanol (99.7%, SCRC) was added
slowly into the solution of 76 mL NH3·H2O (26 vol.%, SCRC) and
200 mL anhydrous ethanol. Then, this synthesis solution was aged
for 4 h and separated centrifugally at 7000 rpm. Finally, the col-
lected product was washed with de-ionized water for three times
and dried at 70 ◦C for 12 h, which was denoted as SiO2(SB). SiO2
synthesized by the sol–gel method was denoted as SiO2(SG). In a
typical synthesis, TEOS anhydrous ethanol solution of 40 vol.% was
dropped into the solution of 5 mL water, 4 g citric acid (99.5%, SCRC),
and 25 mL anhydrous ethanol, under vigorous stirring at 40 ◦C to
obtain a gel. The formed gel was dried at 90 ◦C for 24 h to obtain
powder. The commercial SiO2 from the Qingdao Ocean Desiccant
Co, PR China, was denoted as SiO2(CM). In order to remove the sur-
face impurities, SiO2(CM) was boiled in de-ionized water for 24 h
and dried at 90 ◦C for 12 h. Before being used, all of the SiO2 were
calcined in static air at 500 ◦C for 6 h.
RhCl3 hydrate (Rh ∼ 36 wt.%, Fluka), Mn(NO3)2·6H2O (99.99%,
SCRC), Li2CO3 (99.5%, SCRC), and SiO2 mentioned above were
used in catalyst preparations. Catalysts were prepared by co-
impregnation to incipient wetness of silica (1.0 g) with an aqueous
solution of RhCl3 hydrate and aqueous solutions of precursors of
the promoters, followed by drying at 90 ◦C for 4 h, and then at
120 ◦C overnight before being calcined in air at 350 ◦C for 4 h. For
all the catalysts, Rh loading was 1.5 wt.% and the weight ratio
of Rh:Mn:Li = 1.5:1.5:0.07. The obtained catalysts are denoted as
RML/SiO2(SB), RML/SiO2(SG) and RML/SiO2(CM), respectively.
CO temperature-programmed desorption (TPD) or H2
temperature-programmed reduction (TPR) measurements were
carried out in a quartz microreactor. For TPD measurements, the
catalyst (0.1 g) was reduced in situ for 2 h at 400 ◦C in 10% H2/N2,
followed by flushing with a He flow for 30 min at the same temper-
ature before cooling down to RT. The next step was CO adsorption
at RT for 30 min, and then the gas was swept again with He for
3 h. Subsequently, the sample was heated in a flowing He stream
(50 mL/min) up to 750 ◦C at a rate of 10 ◦C/min with a quadruple
mass spectrometer (QMS, Balzers OmniStar 200) as the detector to
monitor the desorbed species. For TPR measurements, 0.1 g of the
as-prepared sample was first pretreated at 350 ◦C in 20% O2/N2 for
1 h prior to a TPR measurement. During the TPR experiment, 10%
H2/N2 was used at 50 mL/min and the temperature was ramped
from RT to 500 ◦C at 10 ◦C/min while the effluent gas was analyzed
with a TCD.
2.2. Reaction
CO hydrogenation was performed in a fixed-bed micro-reactor
with length ∼ 350 mm and internal diameter ∼ 5 mm. The catalyst
(0.3 g) diluted with inert ␣-alumina (1.2 g) was loaded between
quartz wool and axially centered in the reactor tube, with the
temperature monitored by a thermocouple close to the cata-
lyst bed. Prior to reaction, the catalyst was heated to 400 ◦C
(heating rate ∼ 3 ◦C/min) and reduced with 10% H2/N2 (total flow
rate = 50 mL/min) for 2 h at atmospheric pressure. The catalyst was
then cooled down to 300 ◦C and the reaction started as gas flow was
switched to a H2/CO mixture (molar ratio of H2/CO = 2, total flow
rate = 50 mL/min) at 3 MPa. All post-reactor lines and valves were
heated to 150 ◦C to prevent product condensation. The products
were analyzed on-line (Agilent GC 6820) using a HP-PLOT/Q column
(30 m, 0.32 mm ID) with detection with an FID (flame ionization
detector) and a TDX-01 column with a TCD (thermal conductivity
The temperature-programmed surface reaction (TPSR) experi-
ments were carried out as follows: after the catalyst was reduced

40
J. Yu et al. / Journal of Molecular Catalysis A: Chemical 367 (2013) 38–45
Table 1
CO hydrogenation performance on different catalysts.
Catalyst
CO conv. (C%)
TOFa (s−1
)
Selectivity of productsb (C%)
STY(C2+ Oxy)
(g/(kg h))
CO2
CH4
MeOH
AcH
EtOH
C2+ Oxyc
C2+ HCd
RML/SiO2(SB)
RML/SiO2(SG)
RML/SiO2(CM)
8.2
2.2
6.5
0.068
0.012
0.028
5.3
16.1
17.5
10.3
14.4
11.8
1.7
11.6
5.9
35.5
7.7
15.7
16.5
25.2
18.1
59.1
42.4
37.2
23.6
15.5
27.6
146.8
30.2
80.8
Reaction conditions: T = 300 ◦C, P = 3 MPa, catalyst: 0.3 g, and flow rate = 50 mL/min (H2/CO = 2), data taken after 15 h when steady state reached. Experimental error: 5%.
a
TOF based on CO conversion and H2 chemisorption.
Based on carbon efficiency, carbon selectivity = niCi/ꢀniCi.
C2+ Oxy denotes oxygenates containing two and more carbon atoms.
C2+ HC denotes hydrocarbons containing two and more carbon atoms.
b
c
d
at 400 ◦C in 10% H2/N2 for 2 h, it was cooled down to RT and CO was
introduced for adsorption for 0.5 h; afterwards, the H2/N2 mixture
was swept again, and the temperature was increased at the rate
of 10 ◦C/min with a quadruple mass spectrometer as the detec-
tor to monitor the signals of CO (m/z = 28), CO2 (m/z = 44), and CH4
(m/z = 15).
et al. [25] and Hsu et al. [27]. Furthermore, for the pore distribution,
SiO2(SB) and SiO2(CM) have a similar pore in the range of 8–9 nm,
and SiO2(SG) has a smaller pore of 2.5 nm. Upon being loaded with
metal components, there was a decrease in the surface area and the
pore volume for all the catalysts. In comparison, those of SiO2(SG)
decreased much heavily. It is supposed that SiO2(SG) has a large
surface area before loaded, and its pore size is just in the range
of 2–3 nm, which is similar with the Rh particle size supported on
SiO2(SG) shown in Table 2. So, the larger decrease in the surface area
and the pore volume of the RML/SiO2(SG) catalyst is most likely due
to pore blocking by metal particles.
3. Results and discussion
3.1. Catalytic activities
X-ray diffraction (XRD) patterns (not shown) of silica and the
prepared Rh–Mn–Li catalysts showed no discernable peaks related
to any crystalline phase, indicating that the metal particles are
highly dispersed on the SiO2 supports due to the small content,
which is consistent with the results of the dispersion and metal
particle size of the catalysts shown in Table 2.
The TEM micrographs of the catalysts (not shown) indicate that
the Rh particles are highly dispersed with spherical shapes, and
mainly distribute in 1–5 nm. The result is consistent with those
obtained by XRD and H2 chemisorption (Table 2).
Based on the above results, it was found that although the dis-
persion of Rh differs among different catalysts, all of the Rh particle
size on these catalysts fell in the range of 1–4 nm which is a suit-
able size for CO hydrogenation [5]. It has been reported that the
TOF values increased with the increase in the Rh particle size over
the Rh/Al2O3 catalyst prepared by the microemulsion technology
[28]. In our case, however, the trend in TOF values of CO conversion
over the catalysts does not follow that of metal particle size. On the
other hand, no noticeable difference in morphology of the parti-
cles was observed by TEM measurements. These results suggest
that there must be other factors rather than the size and mor-
phology of Rh particles that affect the intrinsic site-activity (TOF)
of the catalyst [26,29]. But it is still confusing that the catalyst
of RML/SiO2(SB) with the smallest BET surface area and poorest
dispersion showed the highest activity and selectivity to C2+ oxy-
genates. Generally, most researches suggested that a support with
a high surface area favors the dispersion of metal species, further
improving the catalytic conversion of syngas to oxygenates [30].
However, some different opinions were also proposed. For instance,
Table 1 shows the catalytic activities of the Rh–Mn–Li/SiO2
catalysts for CO hydrogenation at 300 ◦C. It can be seen that
the activity and turnover frequency (TOF) of CO conversion
over the catalysts under the same conditions decreased in
a sequence: RML/SiO2(SB) > RML/SiO2(CM) > RML/SiO2(SG). With
respect to selectivity, the highest C2+ oxygenates selectivity was
obtained on RML/SiO2(SB) catalyst. The selectivity of CO2 was sim-
ilar on SiO2(SG) and SiO2(CM) supported catalysts, which was much
higher than that on RML/SiO2(SB) catalyst. Interestingly, formation
of the undesirable by-product methane appeared to be suppressed
over the RML/SiO2(SB) catalyst, on which the methane selectivity
was the lowest. Generally speaking, a higher C2+ oxygenates selec-
tivity would be obtained at a lower CO conversion over the same
catalyst [26]. Taking this fact into account, the C2+ oxygenates selec-
tivity for the RML/SiO2(SB) catalyst would be even more higher than
those for the RML/SiO2(CM) and RML/SiO2(SG), if compared at the
same CO conversion. Furthermore, it is noteworthy that the yield
of the C2+ oxygenates increased remarkably from 30.2 g/(kg h) over
RML/SiO2(SG) to 146.8 g/(kg h) over RML/SiO2(SB).
3.2. Structural and textural properties
N2 adsorption–desorption was carried out to characterize the
textural properties of the supports and the corresponding sup-
ported catalysts. As shown in Table 2, the BET surface area of
SiO2(SG) and SiO2(CM) are 584 and 329 m2/g, respectively. How-
ever, an extremely small BET surface area was obtained on SiO2(SB).
The value of it is only 11 m2/g; similar surface areas of silica pre-
pared by the Stöber method have also been reported by Szekeres
Table 2
Specific surface areas (SBET), pore volume (Vp) and pore diameter (Dp) from N2 adsorption and H2 chemisorption.
Samples
SBET (m2/g)
Vp (cm3/g)
Dp (nm)
H2 chemisorption
H2ads (␮mol/g)
Dispersiona (%)
Particle size (nm)
SiO2(SB)
RML/SiO2(SB)
SiO2(SG)
RML/SiO2(SG)
SiO2(CM)
RML/SiO2(CM)
11
10
584
376
329
321
0.027
0.026
0.207
0.176
0.877
0.828
8.3
7.9
2.5
2.3
9.2
9.0
–
25
–
38
–
–
34
–
52
–
–
3.3
–
2.2
–
49
67
1.7
a
Assuming H/Rhsurface = 1; experimental error: 5%.

J. Yu et al. / Journal of Molecular Catalysis A: Chemical 367 (2013) 38–45
41
3668
3735
2073
2034
2104
3465
a
RML/SiO2(SB)
RML/SiO2(SG)
2064
2030
2098
b
2067
2030
RML/SiO2(CM)
1830
c
2200
2100
2000
1900
1800
3800
3600
3400
3200
3000
2800
Wavenumber (cm-1)
Wavenumber (cm-1)
Fig. 2. FT-IR spectra of CO chemisorbed at 30 ◦C for 30 min.
Fig. 1. FT-IR spectra of (a) SiO2(SB) (straight) and RML/SiO2(SB) (dash), (b) SiO2(SG)
(straight) and RML/SiO2(SG) (dash) and (c) SiO2(CM) (straight) and RML/SiO2(CM)
(dash) in N2 flow at 300 ◦C.
interact with the H-bonded hydroxyls during the course of cata-
lyst preparation, according to the decrease in the percentage of
H-bonded hydroxyls on SiO2 after loading silver. By combining the
activities of catalysts and the above discussion, it can be inferred
that the different catalytic performances of the Rh–Mn–Li cata-
lysts supported on various SiO2 might be related to the different
hydroxyl–metal interactions.
Fan et al. [8] found that the high surface area is not the decisive
factor for the dispersion of metal species in the carbon-supported
Rh-based catalyst. In fact, it is suggested that the combination of
the nanochannels and graphitic structure play an important role
in promoting CO hydrogenation in addition to the metal disper-
sion in these multi-component promoted Rh catalysts. Moreover,
Chen et al. [31] studied the effect of SBA-15 on Rh/Mn-supported
catalysts on CO hydrogenation, and found that the catalytic per-
formance depended on the different chemical properties of SBA-15
modified by different promoters rather than the surface areas. In
this work, we also infer that there are some other properties rather
than the surface area of silica, such as the role of surface hydroxyl
groups on silica mentioned by many researchers [32–36], which
affect the catalytic performance of the Rh–Mn–Li/SiO2 catalysts.
The spectra of CO adsorbed on the in situ reduced catalysts at
30 ◦C for 30 min are given in Fig. 2. The IR spectrum was mainly
composed by a doublet at ∼2100 and ∼2030 cm−1 and a band
at around 2067 cm−1. The doublet can be assigned to the sym-
metric and asymmetric carbonyl stretching of the gem-dicarbonyl
Rh+(CO)2[CO(gdc)] and the 2067 cm−1 band can be attributed to
the linear adsorbed CO [CO(l)] [39]. Meanwhile, the band at around
1830 cm−1 is assigned to the bridge bonded CO [CO(b)] [6]. It is
widely accepted that the CO(l) species can be formed on the Rh0
sites and CO(gdc) be on the Rh+ sites which may be highly dis-
persed [40,41]. It can be seen from Fig. 2 and Table 3 that both
the intensity of adsorbed CO and the ratio of dicarbonyl versus
linear CO (Rh+/Rh0) on the catalysts decreased in the following
order: RML/SiO2(CM) > RML/SiO2(SG) > RML/SiO2(SB). This situa-
tion is similar to the reported view that a higher metal dispersion
and ratio of Rh+/Rh0 sites were achieved with lower Si–OH den-
sity [42]. Moreover, it can be seen that the position of adsorbed CO
for RML/SiO2(SB) catalyst shifted significantly to higher frequency
compared with that for other two catalysts, while the position of
CO(l) for RML/SiO2(SG) catalyst was the lowest among the three
catalysts, suggesting that the Rh–CO bond strength was different
caused by the different hydroxyl–metal interaction. In fact, many
researchers also considered that hydroxyl groups on oxide supports
can be involved in the structural rearrangement of the metal aggre-
gates and in changes in the oxidation state of the metal [24], which
can affect the CO adsorption.
3.3. FT-IR study
The IR spectra of SiO2 and the corresponding catalysts in N2
atmosphere at 300 ◦C are presented in Fig. 1. In these IR spec-
tra, there are three main absorption regions: the first band at
3735 cm−1, arises from the absorption of isolated –OH groups; the
second region with peaks at 3670–3630 cm−1 are assigned to the
weakly H-bonded OH groups; and the third region in the range
3500–2750 cm−1 with a maximum at 3465 cm−1 originates from
the absorption of H2O and strongly H-bonded OH groups [24,37].
As shown in Fig. 1, there are different intensities of the hydroxyl
groups on the different silica surface. More amount of weakly
H-bonded hydroxyls covers on the surface of SiO2(SB), whereas
the surface of SiO2(SG) exhibits more amounts of strongly H-
bonded hydroxyls. Compared with the supports of SiO2(SB) and
SiO2(SG), the amount of surface hydroxyl groups on SiO2(CM) is
much lower. Furthermore, the intensity of the hydroxyl groups on
SiO2 changed obviously after supporting metal components. It can
be seen that the weakly H-bonded hydroxyls over RML/SiO2(SB)
decreased and the isolated hydroxyls increased, indicating there
is a metal–hydroxyl interaction. A similar variation of the strongly
H-bonded hydroxyl groups was observed on RML/SiO2(SG). How-
ever, for the RML/SiO2(CM) catalyst, there is nearly no change in an
absorption intensity of hydroxyl groups, which can be attributed
mainly to a weak interaction between the metal components and
hydroxyls.
In an attempt to relate the CO absorption behavior on different
catalysts to their catalytic performances for CO hydrogenation, a
series of infrared spectra of the in situ reduced catalysts after CO
adsorption at 30 ◦C for 30 min followed by heating with 10% H2/N2
Table 3
The ratio of peak area for the adsorbed CO species.
Catalyst
RML/SiO2(SB)
1.5
RML/SiO2(SG)
1.7
RML/SiO2(CM)
2.3
CO(gdc)/CO(l)a
a
In fact, a similar result was also observed by Qu et al. on Ag/SiO2
catalysts [38]. It is suggested that the silver species could tend to
CO(gdc)/CO(l) denotes the peak area ratio of CO(gdc) versus CO(l) in Fig. 2;
experimental error: 5%.

42
J. Yu et al. / Journal of Molecular Catalysis A: Chemical 367 (2013) 38–45
decreased in the range of 200–300 ◦C. Although the spectra for
A
2104
2073
the desorption behavior of adsorbed CO on RML/SiO2(CM) cata-
lyst was similar to that on RML/SiO2(SB), the CO(gdc) band did
not completely disappear until the temperature reached 200 ◦C.
According to the observation that the intensity of CO(l) band would
increase followed by the attenuation of CO(gdc) band, it is pro-
posed that the Rh+ could be reduced to Rh0 by H2 on RML/SiO2(SB)
and RML/SiO2(CM) catalysts. For the RML/SiO2(SG) catalyst, it can
be seen differently that the absorption band of CO(l) strongly
decreased firstly, while the CO(gdc) band was more stable in H2
atmosphere even if at a higher temperature. Moreover, the thermal
stability of the adsorbed CO species on the catalysts as a function of
temperature were different, following the order as shown in Fig. 3:
RML/SiO2(SG) > RML/SiO2(CM) > RML/SiO2(SB).
2034
0.2
2060
1830
30 0C
100 0C
200 0C
300 0C
Based on the above IR results, it is considered that the differ-
ent surface properties of SiO2 resulted in significant changes in CO
adsorption species and in the intensities of CO adsorption peaks
on different catalysts as a function of temperature in the pres-
ence of hydrogen: the CO(gdc) species on RML/SiO2(SB) catalyst
can be easily transformed to CO(l) species, which was consumed
quickly in the reaction temperature region (200–300 ◦C); however,
the CO(gdc) species on RML/SiO2(SG) catalyst was stable in the H2
atmosphere, either desorbed or reacted slowly even if in the reac-
tion temperature region. Since it has been proved that the CO(gdc)
species can be formed from CO(l) species by assisting of specific
OH groups and can also be reversed to CO(l) species by H2 addi-
tion [24], we further speculate that the interaction between weakly
H-bonded OH groups on SiO2(SB) and Rh particles weakens the
Rh–CO bond strength, which is favorable for the formation of lin-
ear adsorbed CO transformed from Rh+(CO)2 species and facilitates
the desorption/reactivity of the adsorbed CO. Thus, high CO con-
version was obtained. The negative result came out because of the
interaction between strongly H-bonded OH groups on SiO2(SG) and
Rh particles.
2200
2100
2098
2000
1900
1800
Wavenumber (cm-1)
B
2064
2030
0.2
1800
1800
30 0C
100 0C
200 0C
300 0C
2200
2100
2098
2000
1900
Wavenumber (cm-1)
3.4. CO-TPD and TPSR
The desorption profiles of CO (m/z = 28) in Fig. 4(a) show that
a similar associative desorption of CO took place on each catalyst,
suggesting comparable strength of interaction of CO with respec-
tive catalyst surfaces. Based on the TPD results of adsorbed CO on Rh
particle [43], the thermal stability of various types of adsorbed CO
on Rh of the highly dispersed Rh–Mn–Li/SiO2 catalyst is in the fol-
lowing order: CO(b) > CO(gdc) > CO(l). Therefore, the peak at around
105 ◦C should be corresponded to CO(l) species. The peak at 160 ◦C
may be assigned to CO(gdc) species, while the peak exceeded 250 ◦C
may be assigned to the desorption of CO(b) species or other forms
[43]. Moreover, it is obvious that the desorption amount of CO(l)
and CO(gdc) is consistent with the intensity of adsorbed CO dis-
cussed in Section 3.3.
C
2030
2067
0.2
2060
1815
30 0C
100 0C
200 0C
300 0C
Both of the signals of CO2 (m/z = 44) and H2 (m/z = 2) were mon-
itored simultaneously during the CO-TPD experiment. The results
are depicted in Fig. 4(b). CO2 was observed at ∼160 ◦C on each cat-
alyst, and this peak is at the same temperature at which CO(gdc)
species desorbed. It is suggested that CO(gdc) was desorbed in
two different modes: (a) an associative desorption and (b) after
being dissociated these species further react to form CO2. Based
on the amounts of CO2 desorption at ∼160 ◦C, it can be seen that
the CO(gdc) species on RML/SiO2(SB) catalyst was dissociated more
easily than that on the other two catalysts, which may also be inter-
preted by the change of the Rh–CO bond strength caused by the
specific hydroxyl–metal interaction.
2200
2100
2000
1900
1800
Wavenumber (cm-1)
Fig. 3. FT-IR spectra of CO chemisorbed at 30 ◦C for 30 min (dash line) followed by
heating with 10% H2/N2 (the straight lines): (a) RML/SiO2(SB); (b) RML/SiO2(SG) and
(c) RML/SiO2 (CM).
are given in Fig. 3. As observed, the CO(gdc) band on RML/SiO2(SB)
catalyst was significantly attenuated by heating to only 30 ◦C, and
it totally disappeared followed by increase in the intensity of
CO(l) species at 100 ◦C. During the following heating process, the
position of CO(l) shifted to lower frequency, which can be inter-
preted by the decrease in surface coverage [11]. Its intensity fast
When the desorption temperature exceeded 250 ◦C, the desorp-
tion peaks of CO2 on the catalysts became broader and more
complicated, and a corresponding H2 peak was observed. It is simi-
lar to the Rh catalysts supported on Al2O3 and TiO2 where the CO2 is

J. Yu et al. / Journal of Molecular Catalysis A: Chemical 367 (2013) 38–45
43
a
330
160
105
RML/SiO (SB)
2
RML/SiO (SG)
2
RML/SiO (CM)
2
100
200
300
400
500
600
Temperature (oC)
100
200
300
400
500
600
Temperature (oC)
OH +CO
1/2H +CO
2
2
(S)
b
(ads)
320
160
CO
2
RML/SiO (SB)
2
H
2
320
CO
2
RML/SiO (SG)
2
H
2
400
400
100
200
300
400
500
600
Temperature (oC)
CO
2
H
RML/SiO (CM)
2
2
100
200
300
500
600
Temperature (oC)
Fig. 4. Desorption profiles of CO (a) and CO2 and H2 (b) from CO-TPD of
RML/SiO2(SB), RML/SiO2(SG) and RML/SiO2(CM), after CO adsorption at room tem-
perature.
suggested to be formed by the reaction of surface hydroxyl groups
with CO species strongly adsorbed to metallic surface [44,45]:
CO(ad) + OH(s) → CO2 + (1/2)H2
As shown in Fig. 4(b), the high-temperature CO2 peak on
RML/SiO2(CM) was located at 400 ◦C, which is 80 ◦C higher than
that on the catalysts of RML/SiO2(SB) and RML/SiO2(SG) (at 320 ◦C),
suggesting that the surface hydroxyls on it could more difficultly
interact with the strongly adsorbed CO due to the weak interaction
between Rh particles and surface hydroxyls on SiO2(CM).
100
200
300
400
500
600
Temperature (oC)
Fig. 5. CO, CO2 and CH4 desorption from TPSR of RML/SiO2(SB) (a), RML/SiO2(SG) (b)
and RML/SiO2(CM) (c), following continuous CO adsorption at room temperature.
Fig. 5 displays the TPSR profiles of the samples. It can be seen
that there were two methane peaks in the profile, one was at
255 ◦C and the other was at around 520 ◦C. Although the peak
location was almost the same (at 255 ◦C), the area of the peak
for each catalyst was different: RML/SiO2(SG) had the strongest
intensity of CH4 peak; on the contrary, the smallest peak of CH4
was shown on RML/SiO2(SB). According to the viewpoint of Chen
et al. [31], this observation indicates that the number of active
sites responsible for methane production decreased in the follow-
ing order: RML/SiO2(SG) > RML/SiO2(CM) > RML/SiO2(SB), which is
in line with the CH4 selectivity over the catalysts. In addition, the
broad peak centered at 520 ◦C might be due to the hydrogenation
of carbon species formed at lower temperatures and transformed
to a less active form [43].
By combining the results of CO-TPD with TPSR, it is obvious
that CH4 is formed at the expense of the strongly adsorbed CO
species, and most of them come from the CO(b) species and part
of adsorbed CO could transform to CO2 at a high temperature
(above 250 ◦C) in the absence of H2. The desorption peak of CO
below 250 ◦C in TPSR was also different from CO-TPD; only a
broad peak between 100 ◦C and 250 ◦C left on these catalysts. Com-
pared with the catalysts of RML/SiO2(SB) and RML/SiO2(CM), the
biggest peak at ∼160 ◦C attributed to the desorption of CO(gdc)
species was kept on RML/SiO2(SG) catalyst, which is consistent
with the desorption behavior of the adsorbed CO in the presence
of H2 observed by IR. This result also proved that H2 spillover
might occur on the catalysts at the low temperature [46] and the

44
J. Yu et al. / Journal of Molecular Catalysis A: Chemical 367 (2013) 38–45
4. Conclusions
150
130
The direct synthesis of ethanol and other oxygenates from CO
hydrogenation over Rh–Mn–Li/SiO2 catalysts depended greatly on
the properties of SiO2. Different activity and selectivity of C2+
oxygenates were obtained on different SiO2-supported Rh–Mn–Li
catalysts. Results from FT-IR confirmed that various surface
hydroxyl groups exist on the supports, and interact differently with
the metal particles. Based on the result of H2-TPR, it is further
proved that the original hydroxyl groups on the supports have dif-
ferent interactions with metal particles, which finally affects the
Rh–Mn interaction. The IR spectra of CO adsorption indicated that
the linear Rh–CO is the more active adsorbed species for CO hydro-
genation. In addition, the results suggest that the effect of weakly
H-bonded hydroxyl groups is favorable to transform CO(gdc) to
CO(l) species and facilitate the desorption/reactivity of the CO(l)
species, finally resulting in high CO conversion and selectivity of
C2+ oxygenates over the RML/SiO2(SB) catalyst. Overall, the differ-
ent Rh–Mn interaction intensity, which is caused by the reaction of
metal (Rh and Mn) with different hydroxyl groups on the surface
of various SiO2 supports, plays a crucial role in deciding the activity
and selectivity of the Rh–Mn–Li/SiO2 catalyst for CO hydrogenation.
200
RML/SiO (SB)
2
190
RML/SiO (SG)
2
RML/SiO (CM)
2
100
200
300
400
500
Temperature (0C)
Fig. 6. TPR profiles of the catalysts.
adsorbed CO with different bond strength caused by the various
hydroxyl–metal interactions could be desorbed by different behav-
iors. Nevertheless, one contradiction should be noted here that
the highest amount of CH4 was formed on RML/SiO2(SG) which
had the least amount of CO(b) species. It may be concluded that
the boosting of hydrogenation at a certain temperature (255 ◦C)
would cause the adsorbed CO to dissociate and form CH4, if the
adsorbed CO still adsorbs on the metallic surface at that tempera-
ture.
Acknowledgments
The authors gratefully acknowledge financial support from the
Science and Technology Commission of Shanghai Municipality
(08520513600), Leading Academic Discipline Project of Shanghai
Education Committee (J51503) and Shanghai Institute of Tech-
nology (KJ2011-02). The linguistic revisions of the manuscript
provided by the anonymous reviewers are also gratefully acknowl-
edged.
3.5. H2-TPR
Fig. 6 shows the temperature programmed reduction (TPR) pro-
files for all the catalysts. There were three distinct peaks of H2
consumption in the TPR profile of RML/SiO2(SB) catalyst. Accord-
ing to the previous results, the high temperature peak centered
at 200 ◦C is ascribed to the reduction of MnO2 [10,44]. The peaks
at 130 ◦C and 150 ◦C are ascribed to the reduction of Rh2O3 not
intimately contacting with Mn species (denoted as Rh(I)) and of
Rh2O3 intimately contacting with Mn species (denoted as Rh(II)),
respectively [47,48]. The sample of RML/SiO2(CM) also had three
peaks which are similar to RML/SiO2(SB), but the reduction peak
of Rh(II) shifted to a lower temperature and the reduction peak
of MnO2 shifted to a higher temperature compared with that
of RML/SiO2(SB). Moreover, the sample of RML/SiO2(SG) only
showed a wide peak centered at 190 ◦C, which pointed to a strong
interaction between Rh2O3 and Mn species. According to the
viewpoint proposed by Chen et al. [49], it is suggested that a
moderate Rh–Mn interaction is favorable for production of C2+ oxy-
genates.
References
[1] V. Subramani, S.K. Gangwal, Energy Fuels 22 (2008) 814–839.
[2] J.J. Spivey, A. Egbebi, Chem. Soc. Rev. 36 (2007) 1514–1528.
[3] J.R. Rostrup-Nielsen, Science 308 (2005) 1421–1422.
[4] D.H. Mei, R. Rousseau, S.M. Kathmann, V.A. Glezakou, M.H. Engelhard, W.L.
Jiang, C.M. Wang, M.A. Gerber, J.F. White, D.J. Stevens, J. Catal. 271 (2010)
325–342.
[5] T. Hanaoka, H. Arakawa, T. Matsuzaki, Y. Sugi, K. Kanno, Y. Abe, Catal. Today 58
(2000) 271–280.
[6] A.C. Yang, C.W. Garland, J. Phys. Chem. 61 (1957) 1504–1512.
[7] J.P. Hindermann, G.J. Hutchings, A. Kiennemann, Catal. Rev.-Sci. Eng. 35 (1993)
1–127.
[8] Z.L. Fan, W. Chen, X.L. Pan, X.H. Bao, Catal. Today 147 (2009) 86–93.
[9] S.S.C. Chuang, R.W. Stevens Jr., R. Khatri, Top. Catal. 32 (2005) 225–232.
[10] H.Y. Luo, P.Z. Lin, S.B. Xie, H.W. Zhou, C.H. Xu, S.Y. Huang, L.W. Lin, D.B. Liang,
P.L. Yin, Q. Xin, J. Mol. Catal. A: Chem. 122 (1997) 115–123.
[11] X. Mo, J. Gao, J.G. Goodwin Jr., Catal. Today 147 (2009) 139–149.
[12] J. Gao, X. Mo, A.C. Chien, W. Torres, J.G. Goodwin Jr., J. Catal. 262 (2009) 119–126.
[13] X. Mo, J. Gao, N. Umnajkaseam, J.G. Goodwin Jr., J. Catal. 267 (2009) 167–176.
[14] N.D. Subramanian, J. Gao, X. Mo, J.G. Goodwin Jr., W. Torres, J.J. Spivey, J. Catal.
272 (2010) 204–209.
On basis of the H2-TPR, FT-IR, and catalytic test results, we
propose the following model to explain the different catalytic
performances of Rh–Mn–Li/SiO2 via different hydroxyl–metal
interactions. It is conceivable that SiO2(CM) has less hydroxyl
groups to interact with the metal species, resulting in a weak
Rh–Mn interaction on RML/SiO2(CM) catalyst. Compared with the
SiO2(CM) supported catalyst, the interaction between weakly H-
bonded hydroxyl groups on SiO2(SB) and metal particles could
achieve a moderate contact between Rh and Mn, which is con-
ducive to the transformation of CO(gdc), and finally resulting in
the high CO conversion and selectivity of C2+ oxygenates. In con-
trast, the effect of strongly H-bonded associated hydroxyl groups on
RML/SiO2(SG) enhances the Rh–Mn interaction, which suppresses
the transformation of CO(gdc), finally inhibits the CO conversion
and formation of C2+ oxygenates.
[15] M. Ojeda, M.L. Granados, S. Rojas, P. Terreros, F.J. García- García, J.L.G. Fierro,
Appl. Catal. A: Gen. 261 (2004) 47–55.
[16] Z.L. Zhang, A. Kladi, X.E. Verykios, J. Catal. 156 (1995) 37–50.
[17] J.A. Anderson, M.M. Khader, J. Mol. Catal. A: Chem. 105 (1996) 175–183.
[18] J.C. Lavalley, J. Saussey, J. Lamotte, R. Breault, J.P. Hindermann, A. Kiennemann,
J. Phys. Chem. 94 (1990) 5941–5947.
[19] R. Burch, M.J. Hayes, J. Catal. 165 (1997) 249–261.
[20] S. Ho, Y. Su, J. Catal. 168 (1997) 51–59.
[21] W.M. Chen, Y.J. Ding, D.H. Jiang, Z.D. Pan, H.Y. Luo, Catal. Lett. 104 (2005)
177–180.
[22] D.H. Jiang, Y.J. Ding, Z.D. Pan, W.M. Chen, H.Y. Luo, Catal. Lett. 121 (2008)
241–246.
[23] F. Solymosi, I. Tombacz, M. Kocsis, J. Catal. 75 (1982) 78–93.
[24] P. Basu, D. Panayotov, J.T. Yates Jr., J. Am. Chem. Soc. 110 (1988) 2074–2081.
[25] M. Szekeres, O. Kamalin, P.G. Grobet, R.A. Schoonheydt, K. Wostyn, K. Clays,
A. Persoons, I. Dkénáy, Colloids Surf. A: Physicochem. Eng. Aspects 227 (2003)
77–83.
[26] S. Ito, S. Ishiguro, K. Nagashima, K. Kunimori, Catal. Lett. 55 (1998) 197–199.
[27] W.P. Hsu, R.C. Yu, E. Matijevic, J. Colloid Interface Sci. 156 (1993) 56–65.

J. Yu et al. / Journal of Molecular Catalysis A: Chemical 367 (2013) 38–45
45
[28] M. Ojeda, S. Rojas, M. Boutonnet, F.J. Pérez- Alonso, F.J. García- García, J.L.G.
[39] S.D. Worley, G.A. Mattson, R. Caudlll, J. Phys. Chem. 87 (1983) 1671–1673.
[40] R.R. Cavanagh, J.T. Yates Jr., J. Chem. Phys. 74 (1981) 4150–4155.
Fierro, Appl. Catal. A: Gen. 274 (2004) 33–41.
[29] T. Tago, T. Hanaoka, P. Dhupatemiya, H. Hayashi, M. Kishida, K. Wakabayashi,
Catal. Lett. 64 (2000) 27–31.
[41] C.A. Rice, S.D. Worley, C.W. Curtis, J.A. Guin, A.R. Tarrer, J. Chem. Phys. 74 (1981)
6487–6497.
[30] L. Guzzi, Z. Schay, K. Matusek, I. Bogyay, Appl. Catal. 22 (1986) 289–309.
[31] G.C. Chen, C.Y. Guo, X.H. Zhang, Z.J. Huang, G.Q. Yuan, Fuel Process Technol. 92
(2011) 456–461.
[32] A. Brenner, D.A. Hucul, J. Catal. 61 (1980) 216–222.
[33] D.A. Hucul, A. Brenner, J. Phys. Chem. 85 (1981) 496–498.
[34] A.K. Smith, F. Hugues, A. Theolier, J.M. Basset, R. Ugo, G.M. Zanderighi, J.L. Bilhon,
V. Bilhou-Bougnal, W.F. Graydon, Inorg. Chem. 18 (1979) 3104–3112.
[35] F. Solymosi, M. Pasztor, J. Phys. Chem. 89 (1985) 4789–4793.
[36] F. Solymosi, M. Pasztor, J. Phys. Chem. 90 (1986) 5312–5317.
[37] J.B. Peri, J. Phys. Chem. 70 (1966) 2937–2945.
[42] S.L. Guo, M. Arai, Y. Nishiyama, Appl. Catal. 65 (1990) 31–44.
[43] Y. Wang, H.Y. Luo, D.B. Liang, X.H. Bao, J. Catal. 196 (2000) 46–55.
[44] T. Ioannides, X. Verykios, J. Catal. 140 (1993) 353–369.
[45] A. Egbebi, V. Schwartz, S.H. Overbury, J.J. Spivey, Catal. Today 149 (2010) 91–97.
[46] Y. Wang, Z. Song, D. Ma, H.Y. Luo, D.B. Liang, X.H. Bao, J. Mol. Catal. A: Chem.
149 (1999) 51–61.
[47] D.H. Jiang, Y.J. Ding, Z.D. Pan, X.M. Li, G.P. Jiao, J.W. Li, W.M. Chen, H.Y. Luo, Appl.
Catal. A: Gen. 331 (2007) 70–77.
[48] H.M. Yin, Y.J. Ding, H.Y. Luo, H.J. Zhu, D.P. He, J.M. Xiong, L.W. Lin, Appl. Catal.
A: Gen. 243 (2003) 155–164.
[38] Z.P. Qu, W.X. Huang, S.T. Zhou, H. Zheng, X.M. Liu, M.J. Cheng, X.H. Bao, J. Catal.
234 (2005) 33–36.
[49] W.M. Chen, Y.J. Ding, H.Y. Luo, L. Yan, T. Wang, Z.D. Pan, H.J. Zhu, Chin. J. Appl.
Chem. 22 (2005) 470–474.