Solvent feedstock effect: The insights into the deactivation mechanism of Cu/SiO2 catalysts for hydrogenation of dimethyl oxalate to ethylene glycol

Article abstract of DOI:10.1039/c3cc40570b

The variation of the supports on the Cu/SiO₂ catalyst plays an important role in the catalytic performance for hydrogenation of dimethyl oxalate. The loss of silica in the form of tetramethoxysilane from the support under the reaction conditions is responsible for the deactivation of the Cu/SiO₂ catalyst.

Full text of DOI:10.1039/c3cc40570b

ChemComm
View Article Online
View Journal | View Issue
COMMUNICATION
Solvent feedstock effect: the insights into the
deactivation mechanism of Cu/SiO2 catalysts for
hydrogenation of dimethyl oxalate to ethylene glycol†
Cite this: Chem. Commun., 2013,
49, 5195
Received 23rd January 2013,
Accepted 17th April 2013
Chao Wen, Yuanyuan Cui, Wei-Lin Dai,* Songhai Xie and Kangnian Fan
DOI: 10.1039/c3cc40570b
www.rsc.org/chemcomm
The variation of the supports on the Cu/SiO2 catalyst plays an stable in the DMO–ethanol feedstock than in the DMO–methanol
important role in the catalytic performance for hydrogenation of feedstock. They suggested that the CO splitting from methanol
dimethyl oxalate. The loss of silica in the form of tetramethoxy- and the oxidation of alcohols over Cu/SiO2 could be the main
silane from the support under the reaction conditions is responsible causes for the aggregation of copper species. However, based on
for the deactivation of the Cu/SiO2 catalyst.
our present work, the variation of the silica supports in the DMO–
methanol solution should not be overlooked and the tiny silica
erosion under reaction conditions which further results in the
agglomeration of copper particles could be another critical factor
for the poor stability of the Cu/SiO2 catalyst.
The synthesis of ethylene glycol (EG) from syngas is more
economic and energy efficient compared with the petrochemical
route for EG production.1 This synthesis process contains two
steps. The first step for production of dimethyl oxalate (DMO)
has been scaled up to industrial levels with a capacity of 10000
tons per year in 2010.2 The hydrogenation of DMO to EG is
attracting significant interest from both academic and industrial
researchers. Cu supported silica catalysts that exhibit excellent
catalytic activity and high selectivity to EG have been extensively
studied.3 However, these well-established silica-supported,
copper-based catalysts still have a fatal problem, the short
lifespan, which severely blocks their industrial application.
Numerous studies have been devoted to improve the poor
catalytic performance and short lifespan of the Cu/SiO2 catalysts.
Modification with boron on the surface of the Cu/SiO2 catalysts
shows great enhancement in the catalytic stability on account of
maintaining the balance of Cu0 and Cu+ distribution and the
In our present work, a Cu/SiO2 catalyst with 15 wt% copper
loading is prepared by the ammonia-evaporation method with-
out any modification. The catalytic performance for vapor-phase
hydrogenation of DMO is studied in the DMO–MeOH stream
and DMO–EtOH stream containing 15 wt% DMO in methanol
and ethanol solution respectively. To further investigate the
solvent feedstock effect, the catalyst is also exposed to the pure
methanol stream alone under identical reaction conditions.
The long-time catalytic performance for vapor-phase hydroge-
nation of DMO is shown in Fig. 1. It is clear that the Cu/SiO2
catalyst is more stable in the DMO–EtOH stream, even after 300 h
of time on stream. Both the catalytic activity and selectivity remain
unchanged. However, obvious deactivation occurs on the Cu/SiO2
catalyst in the DMO–MeOH stream after running for 140 h.
Conversion on the liquid hourly space velocity (LHSV) is also
conducted to evaluate the catalytic behaviour (Fig. S1, ESI†), and
the DMO conversion on LHSV in the DMO–MeOH and DMO–
EtOH stream is almost the same in the first 8 h, however, a severe
drop in the conversion could be observed at higher LHSV on the
Cu/SiO2 catalyst running in the DMO–MeOH stream for 100 h,
whereas the conversion in the DMO–EtOH stream running for 8 h
and 100 h seems to be unchanged. These results indicate that
some of the active sites are definitely lost during the long-time
catalytic testing in the DMO–MeOH stream.
retarding surface transmigration of copper nanoparticles.4
A
kind of Cu/SiO2/Cordierite monolithic catalyst was also studied
by Ma et al.,5 and their results showed that the regular channel
structure and the uniform distribution of flow of the monolithic
catalyst were responsible for the improved thermal stability.
The modulation of interaction between copper metal parti-
cles and the silica supports is generally used to promote copper
dispersion or decrease the copper particle size, however, the
deactivation mechanism of the Cu/SiO2 catalyst in the DMO to
EG process seems to be still undiscovered. Recently, Lin et al.6
studied the effect of the solvent on copper particles and found
that the Cu/SiO2 catalyst without any modification was more
The TEM images of the reduced and spent catalysts are shown
in Fig. 2. Obvious aggregation of the copper particles could be
observed on the Cu/SiO2 catalysts after reaction. The much bigger
copper particle size (ca. 7.9 nm) of the catalyst running in the
DMO–MeOH stream for 300 h than that of the catalyst running in
Department of Chemistry and Shanghai Key Laboratory Molecular Catalysis and
Innovative Materials, Fudan University, Shanghai 200433, P. R. China.
E-mail: wldai@fudan.edu.cn; Fax: +86-21-55665572; Tel: +86-21-55664678
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c3cc40570b the DMO–EtOH stream (ca. 6.0 nm) could be observed, indicating
c
This journal is The Royal Society of Chemistry 2013
Chem. Commun., 2013, 49, 5195--5197 5195

View Article Online
Communication
ChemComm
the catalyst spent in the DMO–EtOH stream showed the micro-
porous structure with a pore width of about 1.4 nm, however, in
the catalyst after reaction in the DMO–MeOH stream, the pore
distribution peak showed an obvious shift to 1.6 nm, meanwhile,
the peak intensity decreased too (Fig. S3, ESI†). The increase in the
microporous width and the decrease in the intensity of micropore
distribution demonstrate that the microporous structures in the
Cu/SiO2 catalyst were destroyed during the reaction in the DMO–
MeOH stream and the support structures of the catalysts are
definitely changed. Our observations are consistent with Kim’s
studies, in which the higher KOH modification temperature
resulted in a higher degree of damage to the SWNTs and the
migration of micropore distribution to the larger pore width.7
In order to find out how methanol impacts on the silica
supports, the product solution is carefully analyzed using
GC-MS. Surprisingly, a tiny amount of tetramethoxysilane (TMOS)
is detected (Fig. S3, ESI†). It is important to note that the TMOS is
ever present in the DMO–MeOH stream product yet the amount is
so small and usually goes undetected by the usual GC measure-
ment. This result indicates that the TMOS, which is widely used as
a silica source to synthesize SiO2,8 could be generated from
methanol and silica during the hydrogenation process. Thus
some of the silica in the Cu/SiO2 catalyst could be leached away
in the form of TMOS, which may further lead to the aggregation of
copper species. For a closer look, the mass loss rate and the
accumulative amount of SiO2 are shown in Fig. 3. When the
Cu/SiO2 catalyst is exposed to the methanol stream under
the reaction conditions, the silica loss rate is slow at the beginning
and becomes higher gradually with the time on stream. Impor-
tantly, after 150 h of running in the DMO–methanol stream,
0.05 g of silica (ca. 6.5 wt% of the support) is leached away from
the catalyst, furthermore, after 300 h of time on stream, 0.14 g of
the support in the catalysts is lost. The silica loss rate for the
Cu/SiO2 catalyst in the DMO–methanol stream is higher than that
Fig. 1 Catalytic performance of the Cu/SiO2 catalysts in the DMO–MeOH
stream and DMO–EtOH stream. Reaction conditions: PH = 3.0 MPa, T = 473 K,
2
H2/DMO = 150 (mol/mol). LHSV = 0.3 hÀ1
.
Fig. 2 TEM images of the reduced Cu/SiO2 (A); Cu/SiO2 after reaction in DMO–
MeOH stream for 300 h (B) and in DMO–EtOH stream for 300 h (C); copper
particle size distribution for Cu/SiO2 after reaction in DMO–MeOH stream for
300 h (D) and in DMO–EtOH stream for 300 h (E).
that the ethanol solvent could better prevent the copper agglo-
meration during the reaction process compared with the methanol
solvent. Interestingly, clear spherical silica particles could be
observed in the freshly reduced and DMO–EtOH stream spent
catalysts, whereas these spheres in the spent catalyst after 300 h
reaction in the DMO–MeOH stream seem to be destroyed with
an irregular shape, suggesting that the silica structure must be
changed under the methanol containing stream.
The Cu/SiO2 catalysts present a kind of mesoporous structure
with type IV nitrogen sorption isotherms, and the mesopore
distributions seem to be constant even when the Cu/SiO2 catalysts
are exposed to different solvent streams (Fig. S2, ESI†). However,
certain amounts of the micropores are also detected which could
provide more refined information for the variation of the catalyst
structure during the reaction process. Both the fresh Cu/SiO2 and
Fig. 3 Mass loss rate and the accumulative loss amount of SiO2 of the Cu/SiO2
in DMO–MeOH stream or MeOH stream. Reaction conditions: PH = 3.0 MPa,
2
T = 473 K, H2/DMO = 150 (mol/mol). LHSV = 0.3 hÀ1
.
c
5196 Chem. Commun., 2013, 49, 5195--5197
This journal is The Royal Society of Chemistry 2013

View Article Online
ChemComm
Communication
in the pure methanol stream. Based on Ono’s research,9 the silica using ethanol as solvent could avoid the silica erosion effect caused
can react with dimethyl carbonate, which could be thought as the by methanol, and the higher content of well dispersed copper
methoxy group source, to synthesize TMOS via gas-phase reaction particles could be afforded compared with the catalyst spent in the
over the alkali hydroxide supported catalyst. Thus, in the vapor- DMO–MeOH stream even after 300 h of time on stream.
phase hydrogenation of DMO, the methoxy group derived from
Although the solvent feedstock effect on the hydrogenation of
the DMO molecule, which contains double ester groups, could DMO has been carefully studied by Lin and co workers,6 their
also be thought as the methoxyl group donor and help the research mainly focussed on the aggregation of the copper
formation of TMOS. Particularly, in the methanol solvent, the particles caused by the CO splitting and the oxidation of alcohols
transesterification effect between DMO and methanol may on the copper surface. Based on the present study, the variation
enhance the concentration of CH3OÀ and leads to a higher TMOS of the support under the methanol steam, which further leads to
formation rate.10 Also, the reaction between SiO2 and methanol to a decrease in the specific surface area and copper surface area,
generate TMOS under both supercritical and gas-phase conditions undoubtedly has significant influence on the deactivation of the
was evidenced by Chibiryaev et al.11 However, in the case of the Cu/SiO2 in the DMO–MeOH stream. The loss of the silica in the
DMO–EtOH stream, neither TMOS nor tetraethoxysilane (TEOS) form of TMOS evidenced by the GC-MS could be considered as a
was detected in the DMO–EtOH stream product mixture. crucial factor for the copper aggregation.
Although methanol is a product of DMO hydrogenation and the
In summary, the poor catalytic performance of the Cu/SiO2
formation of TMOS seems to be possible, the concentration of catalyst in hydrogenation of DMO with methanol as solvent is
CH3OÀ seems to be limited and both the generation amount and disclosed. The Cu/SiO2 catalyst is more stable in the DMO–EtOH
the rate of TMOS seem to be negligible. Chibiryaev et al. also stream than in the DMO–MeOH one. The loss of the silica from the
confirmed the impossibility of formation of TEOS from silica and Cu/SiO2 in the form of TMOS plays an important role in the decrease
ethanol.11 In addition, it can be roughly estimated that the in copper surface area and the aggregation of copper particle size,
synthesis of TEOS via silica and ethanol requires more energy which will lead to poor stability. As far as we know, this is the first
compared with the synthesis of TMOS via silica and methanol by report to study the solvent effect on the Cu/SiO2 catalyst. Our study
the bond enthalpy calculations.12 Meanwhile, a study on the illustrates the deactivation mechanism of the Cu/SiO2 catalyst and
thermal decomposition of TMOS and TEOS by Lin et al.13 shows the erosion of the silica support, thus has extremely important
that TMOS is 2400 times more stable than TEOS at 800 K. Thus, practical significance in its large-scale industrial application.
TMOS is more favourably generated under the hydrogenation of
DMO, and the Cu/SiO2 catalyst is more stable in the DMO–EtOH (Grant No. 2012CB224804), NNSFC (Project 21173052, 20973042),
stream than in the DMO–MeOH one. and the Science & Technology Commission of Shanghai Munici-
We thank the Major State Basic Resource Development Program
The physicochemical properties of the catalysts are shown in pality (08DZ2270500) for financial support.
Table S1 (ESI†). An obvious increase in the Cu loading could be
Notes and references
observed in the catalyst after reaction in the DMO–MeOH stream,
confirming again the loss of silica species during the reaction
process. The decrease in the specific surface area and copper
dispersion could be observed from the spent Cu/SiO2 catalyst. It is
interesting to find that the catalyst spent in the DMO–EtOH
stream shows a lower decreasing value than that in the DMO–
MeOH stream. The variation of the supports usually causes the
changes in the specific surface area, which could further induce
copper aggregation and result in the small copper surface area,
the latter is considered as the crucial factor for the poor activity in
the hydrogenation of DMO.14 The increased copper particle size
(7.2 nm) after 300 h reaction in the DMO–MeOH stream as well as
the catalytic activity loss should be ascribed to the erosion of the
SiO2 support in the form of TMOS. To further study the influence
of the variation of the supports on the reduction behaviour of the
Cu/SiO2 catalyst, the TPR measurement was conducted and the
1 H. Yue, Y. Zhao, X. Ma and J. Gong, Chem. Soc. Rev., 2012, 41, 4218.
2 X. Gao, Y. Zhao, S. Wang, Y. Yin, B. Wang and X. Ma, Chem. Eng. Sci.,
2012, 66, 3513.
3 (a) L. F. Chen, P. J. Guo, M. H. Qiao, S. R. Yan, H. X. Li, W. Shen,
H. L. Xu and K. N. Fan, J. Catal., 2008, 257, 172; (b) C. Wen, Y. Cui,
A. Yin, W. Dai and K. Fan, ChemCatChem, 2013, 1, 138; (c) A. Yin,
X. Guo, W. Dai and K. Fan, J. Phys. Chem. C, 2009, 113, 11003;
(d) L. Zhao, Y. Zhao, S. Wang, H. Yue, B. Wang, J. Lv and X. Ma, Ind.
Eng. Chem. Res., 2012, 51, 13935.
4 Z. He, H. Lin, P. He and Y. Yuan, J. Catal., 2011, 277, 54.
5 H. Yue, Y. Zhao, L. Zhao, J. Lv, S. Wang, J. Gong and X. Ma, AlChE,
2012, 58, 2798.
6 J. Lin, X. Zhao, Y. Cui, H. Zhang and D. Liao, Chem. Commun., 2012,
48, 1177.
7 S. M. Lee, S. H. Park, S. C Lee and H. J. Kim, Chem. Phys. Lett., 2006,
432, 518–522.
8 Y. Zhang and J. Huang, J. Mater. Chem., 2011, 21, 7161.
9 (a) E. Suzuki, M. Akiyama and Y. Ono, J. Chem. Soc., Chem. Commun.,
1992, 2, 136; (b) Y. Ono, M. Akiyama and E. Suzuki, Chem. Mater.,
1993, 5, 442.
results are shown in Fig. S6 (ESI†). The fresh Cu/SiO2 catalyst 10 N. W. Cant, S. P. Tonner, D. L. Trimm and M. S. Wainwright,
J. Catal., 1985, 91, 197–207.
11 A. M. Chibiryaev, I. V. Kozhevnikov and O. N. Martyanov, Appl.
shows only one peak at 525 K assigned to the well-dispersed
copper oxide,3a which is consistent with the XRD results. The
Catal., A, 2013, 456, 159–167.
Cu/SiO2 catalyst spent in the DMO–MeOH stream also shows a 12 (a) S. J. Blanksby and G. B. Ellison, Acc. Chem. Res., 2003, 36, 255;
(b) B. S. Jursic, J. Chem. Soc., Perkin Trans. 2, 1999, 369;
single reduction peak, but the reduction peak shifts to a higher
(c) S. Venimadhavan, K. Amarnath, N. G. Harvey, J. P. Cheng and
temperature of ca. 575 K, indicating that the crystalline copper
E. M. Arnett, J. Am. Chem. Soc., 1992, 114, 221.
species with large particle size are dominant.3a For the Cu/SiO2 13 (a) J. C. S. Chu, J. Breslin, N. S. Wang and M. C. Lin, Mater. Lett.,
1991, 12, 179; (b) J. C. S. Chu, R. Soller, M. C. Lin and C. F. Melius,
J. Phys. Chem., 1995, 99, 663.
14 S. Zhao, H. Yue, Y. Zhao, B. Wang, Y. Geng, J. Lv, S. Wang, J. Gong
catalyst spent in the DMO–EtOH stream, there are two distinct
peaks which are assigned to the well-dispersed and crystalline
copper species respectively. These results strongly suggest that
and X. Ma, J. Catal., 2013, 297, 142.
c
This journal is The Royal Society of Chemistry 2013
Chem. Commun., 2013, 49, 5195--5197 5197