Suppression of CO by-production in steam reforming of methanol by addition of zinc oxide to silica-supported copper catalyst

Article abstract of DOI:10.1016/j.jcat.2009.09.026

Methanol steam reforming to hydrogen and carbon dioxide is catalyzed over 30 wt% Cu/SiO₂ prepared by a sol-gel method. The addition of zinc ions to the starting material of the catalyst results in the suppression of CO formation in the reaction

Full text of DOI:10.1016/j.jcat.2009.09.026

Journal of Catalysis 268 (2009) 282–289
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Suppression of CO by-production in steam reforming of methanol by addition
of zinc oxide to silica-supported copper catalyst
b
Yasuyuki Matsumura a, , Hideomi Ishibe
*
a National Institute of Advanced Industrial Science and Technology (AIST) Kansai Center, Midorigaoka, Ikeda, Osaka 563-8577, Japan
b Nippon Seisen Co. Ltd., Hirakata Plant, Ikenomiya, Hirakata, Osaka 573-8522, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 30 June 2009
Revised 25 September 2009
Accepted 26 September 2009
Available online 27 October 2009
Methanol steam reforming to hydrogen and carbon dioxide is catalyzed over 30 wt% Cu/SiO2 prepared by
a solÀgel method. The addition of zinc ions to the starting material of the catalyst results in the suppres-
sion of CO formation in the reaction. Zinc oxide is highly dispersed in the catalyst even if the content is as
high as 10 wt%. In the catalyst reduced with hydrogen, large Cu particles coexist with fine Cu particles
which are oxidized to Cu2O during the reaction. It is estimated that zinc oxide interacts with these Cu
particles and stabilizes Cu+ species formed during the reaction.
Keywords:
Methanol steam reforming
Ó 2009 Elsevier Inc. All rights reserved.
Cu/SiO2
Modification of zinc oxide
Hydrogen production
Carbon monoxide
TPR
XRD
TEM
XPS
EXAFS
1. Introduction
We have investigated a silica-supported copper catalyst pre-
pared by the solÀgel method [4]. The CO selectivity is significantly
Methanol is a potential feedstock of hydrogen processors for
fuel cells especially for mobile applications such as electric vehicles
because hydrogen can be easily produced by the steam reforming
of methanol (CH3OH + H2O ? 3H2 + CO2) over copper catalysts at
200À300 °C [1]. Polymer electrolyte fuel cells (PEFCs) are generally
employed for the portable devices by virtue of their compactness
and mild operation conditions. However, the hydrogen fuel must
not contain carbon monoxide above 10 ppm due to the poisoning
of the catalytic anode [2]. Although the methanol steam reforming
is fairly selective, by-production of carbon monoxide is not negligi-
ble over copper catalysts such as Cu/ZnO/Al2O3 and additional
reactors are required in the fuel processor system for the removal
of by-produced carbon monoxide [3]. Selective oxidation of carbon
monoxide is often carried out to remove carbon monoxide to a le-
vel less than 10 ppm [2], but an excessive amount of oxygen is re-
quired in the process and it consumes hydrogen. Thus, suppression
of CO by-production in methanol steam reforming is important to
increase the efficiency of the hydrogen processors.
lower than a commercial Cu/ZnO/Al2O3 in the reaction at 250 °C.
This is probably because Cu+ species is stabilized by the interaction
with the silica support under the reaction condition; nevertheless,
the selectivity increases at 300 °C to the same level as Cu/ZnO/
Al2O3. In order to improve the performance of the Cu/SiO2 catalyst,
we have modified zinc oxide to it and mainly tested the activity to
methanol steam reforming at 300 °C. It was reported that addition
of zinc oxide to a Cu/SiO2 catalyst increases the activity to metha-
nol synthesis from carbon dioxide and hydrogen [5]. Here, it is
shown that the CO selectivity decreases significantly by the modi-
fication of zinc oxide.
2. Experimental
A silica-supported copper catalyst (Cu/SiO2) was prepared by
the hydrolysis and polymerization of a mixture of tetramethoxy
silane (Tokyo Chemical Industry), copper nitrate (Wako, S-grade),
water, and methanol at room temperature for 1 day. After drying
in air at 120 °C it was heated in air for 5 h at 500 °C for removal
of NOÀ3 anions and residual organic compounds. Zinc was
introduced to Cu/SiO2 by the addition of zinc nitrate (Wako,
S-grade) to the starting materials. The catalysts (n-ZnO/Cu/SiO2,
* Corresponding author. Fax: +81 72 751 9623.
E-mail address: yasu-matsumura@aist.go.jp (Y. Matsumura).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.09.026

Y. Matsumura, H. Ishibe / Journal of Catalysis 268 (2009) 282–289
283
n represents the ZnO content in wt%) were also heated in air for 5 h
at 500 °C. The Cu content was fixed at 30 wt% in the reduced form
because Cu/SiO2 with this content produced fairly low CO selectiv-
ity at 300 °C [4]. A commercial Cu/ZnO/Al2O3 catalyst (Süd-Chemie,
MDC-3) was used as a reference. The catalyst was reported to be
durable in the methanol steam reforming at 350 °C [6].
range of 30À150 nmÀ1. The analysis was performed with a pro-
gram of ‘‘REX2000” supplied by Rigaku.
The surface of the catalyst reduced with hydrogen at 250 °C for
1 h was oxidized by the decomposition of N2O at 50 °C [9]. The
amount of N2 produced was measured with a TCD.
Catalytic tests were performed in a fixed-bed continuous-flow
reactor operated under atmospheric pressure. A catalyst (0.50 g,
granules of 10À20 mesh) was usually placed in a tube reactor
made of stainless steel (i.d., 7 mm) with quartz wool plugs. After
the pre-reduction in a stream of 15-vol% hydrogen diluted with ar-
gon (3.5 dm3 hÀ1) at 250 °C or 300 °C for 1 h, a mixture of metha-
nol, steam, and argon (1.0/1.5/1.0 in molar ratio) was introduced
with a flow rate of 10.6 dm3 hÀ1 at the same temperature as in
the reduction. The effluent gas was dried with a cold trap at ca.
À50 °C, and analyzed with an on-stream gas chromatograph (Shi-
madzu GC-8A; activated carbon, 2 m; Ar carrier) equipped with a
thermal conductivity detector (TCD). The methanol conversion
was determined from the material balance of the reactant and
the products, and the error was within 5%. No formation of formal-
dehyde, methane, or methyl formate was observed.
Temperature-programed reduction (TPR) of the catalyst (0.50 g)
was carried out in a stream of 15-vol% hydrogen diluted with argon
at a flow rate of 3.5 dm3 hÀ1. Temperature of the sample bed was
risen linearly at a rate of 200 °C hÀ1. The hydrogen consumption
was monitored by analyzing the hydrogen concentration with
the on-stream GC every 3 min. No complete hydrogen consump-
tion was observed in the reduction.
3. Results
3.1. Methanol steam reforming over ZnO-modified Cu/SiO2
Steam reforming of methanol was carried out over the copper
catalysts. Hydrogen and carbon dioxide were almost selectively
produced at 300 °C. Addition of zinc oxide significantly lessened
the CO selectivity of Cu/SiO2 and discernibly increased the catalytic
activity. The activities at 6 h-on-stream are shown in Fig. 1. When
double the amount of Cu/SiO2 (1.0 g) was charged in the reactor,
the CO selectivity increased to 2.5% with 88% methanol conversion
at 300 °C. The CO selectivity leveled off at the ZnO content of 5 wt%.
The activities at 250 °C were mostly parallel to those at 300 °C and
the CO selectivity decreased by the addition of zinc oxide. The
activity of 5-ZnO/Cu/SiO2 at 300 °C was almost stable within 6 h-
on-stream, and the CO selectivity was about a half of that with
Cu/ZnO/Al2O3 (Fig. 2). When the amount of 5-ZnO/Cu/SiO2 was
Powder XRD (X-ray diffraction) patterns of the catalysts were
recorded in air at room temperature with an MAC Science MP6XCE
diffractometer using nickel-filtered Cu K
a radiation. The Rietveld
refinement of the patterns was carried out using the Rietveld soft-
ware RIETAN-2000 [7]. To obtain the quantitative phase abun-
dance, the XRD data of 30À70° in two theta were modeled with
monoclinic CuO (space group, C2/c) and cubic Cu (Fm-3 m) using
a pseudo-Voigt function. The factor of Rwp was always less than 5%.
The BET surface areas of the catalysts were determined from the
isotherms of nitrogen physisorption.
The structure of 10-ZnO/Cu/SiO2 after reduction with hydrogen
at 250 °C for 1 h was evaluated by transmission electron micros-
copy (TEM) and high-angle annular dark field imaging in scanning
transmission electron microscopy (HAADF/STEM) with energy dis-
persed X-ray Spectroscopic (EDS) analysis using an FEI Tecnai G2
F20 Twin with an EDX detecting unit (EDAX Inc.) at the accelera-
tion voltage of 200 kV. The sample was introduced to the micro-
scope at room temperature in air.
Fig. 1. Catalytic activity of Cu/SiO2 and ZnO/Cu/SiO2 in methanol steam reforming
after 6 h.
X-ray photoelectron spectra (XPS) were recorded at room tem-
perature with a JEOL JPS-9010MX spectrometer (Mg Ka). The cat-
alysts used in the reaction were taken out from the reactor after
cooling under an argon stream for 12 h or longer and mounted in
air to a sample holder. A sample reduced with hydrogen at
250 °C for 1 h was mounted to a sample holder under Ar atmo-
sphere and measured in situ. Binding energies were corrected by
the reference of the C 1s line at 284.6 eV. The surface atomic con-
centrations of Cu, Zn, Si, and O were calculated from the peak areas
using the atomic relative sensitivity factors (ARSF) of Cu 2p3/2 (13),
Zn 2p3/2 (17), Si 2p (1.0), and O 1s (2.4) determined by measuring
standard materials of copper and zinc plates, silicon wafer, and
quartz glass plate. The concentration analysis using ARSF is quan-
titative and the uncertainty is less than 2% [8].
The profile of EXAFS (extended X-ray absorption fine structure)
for 10-ZnO/Cu/SiO2 reduced with hydrogen at 250 °C for 1 h was
taken at room temperature in transmission mode for Zn K-edge
at beam-line BL14B2 of SPring-8. The sample was handled in air
and vacuum-sealed with polyethylene films. The Fourier transfor-
mation was performed on k3-weighted EXAFS oscillations in the
Fig. 2. Comparison of the catalytic activity of 5-ZnO/Cu/SiO2 to that of a commercial
Cu/ZnO/Al2O3 at 300 °C.

284
Y. Matsumura, H. Ishibe / Journal of Catalysis 268 (2009) 282–289
1.0 g, the CO selectivity was increased to 1.0% with 100% methanol
conversion at 300 °C.
3.2. TPR of copper catalysts
Reduction of the copper catalysts started at ca. 150 °C in TPR
experiments regardless of the ZnO content (Fig. 3). Two peaks were
distinguished in the profile of Cu/SiO2. The peaks seemed to merge
in the profile for ZnO/Cu/SiO2. However, the shape of the profile
showed the presence of a shoulder peak and the profile can be
deconvoluted into two Gaussian peaks (Table 1). The molar
amounts of hydrogen consumption corresponded with the Cu con-
tent in the catalysts (4.7 mmol gÀ1), showing complete reduction
of CuO in the catalysts. In a TPR experiment, Cu/SiO2 was reduced
up to 210 °C and quenched under an argon stream. The H2 con-
sumption was 2.2 mmol gÀ1. In the case of 10-ZnO/Cu/SiO2 re-
duced up to 210 °C, the H2 consumption was 3.7 mmol gÀ1
.
3.3. XRD patterns for copper catalysts
Only the XRD peaks attributed to CuO were recorded in the
pattern of Cu/SiO2 as prepared (Fig. 4a) [10]. The mean crystal-
lite size of CuO was determined as 24 nm from the line broaden-
ing of the peak at 48.8° in 2h using Scherrer equation [11]. Small
peaks at 43.4° and 50.5° being assigned to Cu(1 1 1) and Cu(200)
[10], respectively, appeared with the peaks for CuO in the XRD
pattern of Cu/SiO2 reduced up to 210 °C in the TPR (Fig. 4b).
The mean crystallite size of Cu was 34 nm from the line broad-
ening of Cu(1 1 1) and the size of CuO was 26 nm. The molar
fraction of Cu/(Cu + CuO) determined by the Rietveld analysis
was 33%. The peaks for CuO were completely diminished by
the reduction at 250 °C for 1 h and the crystallite size of Cu
was 34 nm (Fig. 4c).
Fig. 4. XRD patterns of Cu/SiO2 and 10-ZnO/Cu/SiO2. (a) Cu/SiO2 as prepared, (b)
reduced with hydrogen up to 210 °C, (c) reduced at 250 °C, (d) 10-ZnO/Cu/SiO2 as
prepared, (e) reduced up to 210 °C, and (f) reduced at 250 °C.
was 21 nm. The XRD peaks for Cu metal appeared with those for
CuO in the pattern of the sample reduced up to 210 °C (Fig. 4e).
The mean crystallite sizes of Cu and CuO were 25 nm and 23 nm,
respectively. The molar fractions of Cu/(Cu + CuO) determined by
The XRD pattern for 10-ZnO/Cu/SiO2 as prepared was similar
to that of Cu/SiO2 (Fig. 4d), and the mean crystallite size of CuO
Fig. 5. XRD patterns of Cu/SiO2 and ZnO/Cu/SiO2 after the reaction at 300 °C for 6 h.
(a) Cu/SiO2, (b) 2-ZnO/Cu/SiO2, (c) 5-ZnO/Cu/SiO2, and (d) 10-ZnO/Cu/SiO2.
Fig. 3. Profiles of TPR for Cu/SiO2 and ZnO/Cu/SiO2 with a heating rate of 200 °C hÀ1
.
Table 1
The amounts of hydrogen consumption in the TPR experiment.
Catalyst
Peak 1
Peak 2
Temperature (°C)
H2 consumption (mmol gÀ1
)
Temperature (°C)
H2 consumption (mmol gÀ1
)
Cu/SiO2
197
205
207
185
3.2
3.7
3.8
3.2
231
233
230
212
1.5
0.8
0.9
1.3
2-ZnO/Cu/SiO2
5-ZnO/Cu/SiO2
10-ZnO/Cu/SiO2

Y. Matsumura, H. Ishibe / Journal of Catalysis 268 (2009) 282–289
285
the Rietveld analysis was 53%. Only the peaks attributed to Cu
3.4. TEM and HAADF/STEM of ZnO-modified Cu/SiO2
metal were recorded in the XRD pattern for the sample reduced
at 250 °C for 1 h (Fig. 4f). The mean crystallite size of Cu was
25 nm.
Dark portions were seen in a TEM image of an aggregate of 10-
ZnO/Cu/SiO2 (Fig. 6a), while the edges of the aggregate were in the
amorphous shape (Fig. 6b). In the HAADF/STEM image which
shows the same area as in Fig. 6a, bright and round particles with
the size of 11À13 nm were mainly present (Fig. 6c), while some
particles seemed to bond. Since the intensity of the HAADF/STEM
image relates to the square of atomic number of the detected
atoms, i.e., Z2, the bright grains should consist of copper. This is
supported by the EDX analysis (Fig. 6d); Cu was mainly detected
at the bright points 1À3 by EDS analysis and Si was mainly present
at the dark points 4 and 5 (Fig. 6d). Zinc was also detected with Cu
at the points 2 and 3, but no particles were identified as zinc oxide.
In the amorphous portion (Fig. 7a), small and dark spots with the
size of ca. 1 nm were seen (some spots are indicated with arrows
in Fig. 7b). The HAADF/STEM image (Fig. 7c) was obtained at a
higher sensitivity than that for Fig. 6c. In an amorphous portion,
Si was mainly present while Cu and Zn were also detected (Fig. 7d).
A broad peak at 36.7° assigned to Cu2O (1 1 1) was recorded
with the peaks for Cu metal in the XRD pattern of Cu/SiO2 after
the reaction at 300 °C for 6 h (Fig. 5a) [10]. Addition of zinc
oxide to Cu/SiO2 resulted in a discernible line broadening of
the peak at 43.4° attributed to Cu metal (Figs. 5b–d). The mean
crystallite size of Cu metal for Cu/SiO2 was 39 nm, and it was
decreased to 29À33 nm by the addition of zinc oxide (Table 2).
No significant change in the weak peak for Cu2O was observed
by the addition. The mean crystallite size was 3À4 nm regardless
of the samples.
Table 2
Mean crystallite sizes and BET surface areas of the catalysts after methanol steam
reforming at 300 °C for 6 h.
Catalyst
Crystallite size (nm)
BET surface area (m2 gÀ1
)
Cu2O
Cu
3.5. XPS of the copper catalysts
Cu/SiO2
Cu/SiO2
2-ZnO/Cu/SiO2
5-ZnO/Cu/SiO2
10-ZnO/Cu/SiO2
10-ZnO/Cu/SiO2
4
39
34
33
33
29
25
291
360
273
239
244
277
a
À
The surfaces of Cu/SiO2 and ZnO/Cu/SiO2 after the reaction at
300 °C were characterized by XPS. The binding energies of Cu
2p3/2 were 931.6À932.6 eV (Fig. 8), whereas the energies for Cu
metal and Cu2O were 932.7 and 932.4 eV, respectively [12]. In or-
der to identify the electronic state, the Auger line of Cu L3VV was
recorded and the kinetic energies were 917.0À918.0 eV (Fig. 9).
4
3
4
À
a
a
Just after reduction with hydrogen at 250 °C for 1 h.
Fig. 6. TEM and HAADF/STEM images and EDS analysis of 10-ZnO/Cu/SiO2 reduced with hydrogen at 250 °C for 1 h. (a) TEM image, (b) partial enlargement of the TEM image,
(c) HAADF/STEM image, and (d) EDS spectra on points 1À5.

286
Y. Matsumura, H. Ishibe / Journal of Catalysis 268 (2009) 282–289
Fig. 7. TEM and HAADF/STEM images and EDS analysis of amorphous portion in 10-ZnO/Cu/SiO2 reduced with hydrogen at 250 °C for 1 h. (a) TEM image, (b) partial
enlargement of the TEM image, (c) HAADF/STEM image, and (d) EDS spectra on points 1À3.
Fig. 8. XPS of Cu 2p region for Cu/SiO2 and ZnO/Cu/SiO2 after the reaction at 300 °C
for 6 h. (a) Cu/SiO2, (b) 2-ZnO/Cu/SiO2, (c) 5-ZnO/Cu/SiO2, (d) 10-ZnO/Cu/SiO2, and
(e) 10-ZnO/Cu/SiO2 reduced with hydrogen at 300 °C for 1 h.
The energies for Cu metal and Cu2O are 917.8 and 916.8 eV, respec-
Fig. 9. Auger line of Cu L3VV for Cu/SiO2 and ZnO/Cu/SiO2 after the reaction at
300 °C for 6 h. (a) Cu/SiO2, (b) 2-ZnO/Cu/SiO2, (c) 5-ZnO/Cu/SiO2, (d) 10-ZnO/Cu/
SiO2, and (e) 10-ZnO/Cu/SiO2 reduced with hydrogen at 300 °C for 1 h.
tively. The
a values (Auger parameter + photon energy) were
1849.5À1850.0 eV being close to 1849.4 eV for Cu2O (Table 3)
[12]. The value for the metallic Cu is 1851.3 eV [12].
The
a
values of Cu/SiO2 and 10-ZnO/CuSiO2 reduced at 250 °C
hydrogen at 300 °C for 1 h was handled in air, the a value was
with hydrogen were both 1850.3 eV where the samples were han-
1850.5 eV, showing that the handling in air did not affect the XPS
dled in situ (see Table 3). Although 10-ZnO/Cu/SiO2 reduced with
measurement seriously.

Y. Matsumura, H. Ishibe / Journal of Catalysis 268 (2009) 282–289
287
The binding energies of Zn 2p3/2 for the samples containing zinc
were 1022.7 0.2 eV (Fig. 10), and the energies of O 1s and Si 2p
were 533.0 0.2 eV and 103.3 0.1 eV, respectively, regardless of
the samples. The surface atomic concentrations of Cu, Zn, and Si
are also given in Table 3.
3.6. BET surface area of the catalyst after reaction
The BET surface area of Cu/SiO2 after the reaction at 300 °C was
291 m2 gÀ1 and it was decreased to 239À273 m2 gÀ1 by the addi-
tion of zinc oxide (see Table 2). The areas of Cu/SiO2 and 10-ZnO/
Cu/SiO2 just after the reduction with hydrogen at 250 °C for 1 h
were 360 m2 gÀ1 and 277 m2 gÀ1, respectively.
3.7. Structural analyses of zinc in ZnO-modified Cu/SiO2 by EXAFS
In order to analyze the atomic structure of zinc, EXAFS (ex-
tended X-ray absorption fine structure) analysis was performed
for 10-ZnO/Cu/SiO2 reduced at 250 °C for 1 h. The Fourier trans-
form showed the presence of a peak at 0.16 nm (phase shift uncor-
rected) attributed to ZnÀO interaction (Fig. 11). The distance was
slightly shorter than that of a ZnO standard sample, but no peak
attributed to ZnÀZn interaction was present in the profile for 10-
ZnO/Cu/SiO2.
Fig. 10. XPS of Zn 2p region for ZnO/Cu/SiO2 after the reaction at 300 °C for 6 h. (a)
2-ZnO/Cu/SiO2, (b) 5-ZnO/Cu/SiO2, and (c) 10-ZnO/Cu/SiO2.
3.8. Surface oxidation of ZnO-modified Cu/SiO2 with N2O
The surface amounts of copper on Cu/SiO2 and 10-ZnO/Cu/SiO2
reduced with hydrogen at 250 °C were evaluated by the surface
oxidation with N2O. The quantities of N2O decomposed were
0.039 mmol gÀ1 and 0.049 mmol gÀ1, respectively. The Cu surface
amounts were 0.078 mmol gÀ1 and 0.098 mmol gÀ1, respectively,
on the basis of the stoichiometry, that is, 2Cu + N2O ? Cu2O + N2
[9].
4. Discussion
4.1. Morphology of ZnO-modified Cu/SiO2
Fig. 11. Fourier transforms of k3-weighted Zn K-edge EXAFS for (a) 10-ZnO/Cu/SiO2
reduced with hydrogen at 250 °C for 1 h and (b) ZnO.
The XRD pattern of Cu/SiO2 reduced with hydrogen at 250 °C
shows the presence of large Cu particles with the mean crystallite
size of 34 nm. In the TPR of Cu/SiO2 up to 210 °C, 48% of CuO in the
sample is reduced, but the reducibility of the large CuO particles
determined by XRD (33%) is considerably smaller than that. This
is probably due to the co-existence of fine CuO particles that are
undetectable by XRD, as discussed in the previous paper [4]. A sin-
gle TPR peak at around 190 °C was observed with 10 wt% Cu/SiO2
containing fine CuO particles only [4], suggesting that the first peak
is not due to reduction of Cu2+ to Cu+. Hence, the TPR peak decon-
voluted at the lower temperature (peak 1) and it can be attributed
to the reduction of the fine CuO particles (see Table 1).
While the TPR peaks for 10-ZnO/Cu/SiO2 are not separated
clearly (see Fig. 3d), the reducibility up to 210 °C is evaluated as
82% by TPR and significantly higher than the reducibility deter-
mined by XRD (53%). Hence, the fine CuO particles undetectable
by XRD should be also present in the unreduced sample and they
must be responsible for the peak 1 in the TPR (see Table 1). Actu-
ally, copper is detected in the amorphous portion mainly com-
prised of silica despite the lack of presence of large Cu particles
in the TEM and STEM (see Fig. 7). The dark spots with the size of
Table 3
Summary of the XPS data for the catalysts after methanol steam reforming at 300 °C for 6 h.
Catalyst
Cu 2p3/2 (eV)
Cu L3VV (eV)
a
(eV)
Surface atomic composition (%)
Cu
Zn
Si
Cu/SiO2
Cu/SiO2
931.7
932.5
931.6
932.6
932.4
932.3
932.1
917.8
917.8
918.0
917.0
917.6
918.0
918.4
1849.5
1850.3
1849.6
1849.6
1850.0
1850.3
1850.5
5
2
6
7
4
2
4
0
0
1
2
6
3
6
28
28
27
18
17
28
29
a
2-ZnO/Cu/SiO2
5-ZnO/Cu/SiO2
10-ZnO/Cu/SiO2
10-ZnO/Cu/SiO2
10-ZnO/Cu/SiO2
a
b
a
Just after reduction with hydrogen at 250 °C for 1 h (in situ).
Just after reduction with hydrogen at 300 °C for 1 h (ex situ).
b

288
Y. Matsumura, H. Ishibe / Journal of Catalysis 268 (2009) 282–289
ca. 1 nm in Fig. 7b may be the fine Cu particles because the silica
particle with this size is less distinguishable with TEM.
4.2. Effect of ZnO on the catalytic performance of Cu/SiO2
Since the H2 consumption in the TPR corresponds well with the
Cu content of the catalyst, the quantity of the fine Cu particles is
estimated from the H2 consumption for the peak 1 (see Table 1).
Assuming that the fine particles in 10-ZnO/Cu/SiO2 reduced at
250 °C are spherical with a diameter of 1 nm and that all the par-
ticles are on the surface, the Cu surface amount of the fine particles
is estimated to be 7 mmol gÀ1 (Table 4); here, the site density of Cu
is 0.032 mmol mÀ2 on the basis of the atomic radius. The amount
largely exceeds that determined by the N2O adsorption
(0.078 mmol gÀ1), showing that the major part of the fine particles
is buried in the support. In the cases of silica-supported catalysts
prepared by the sol–gel technique, the major part of metallic par-
ticles is often encapsulated in the support [13,14]. The surface of
metallic copper catalysts is often oxidized to Cu2+ in air [15], but
no such oxidation takes place with Cu/SiO2 and ZnO/Cu/SiO2 pre-
sumably because of the encapsulation in the silica support. Silicon
is always detected in the EDS analysis on the surface of 10-ZnO/Cu/
SiO2 (see Figs. 6d and 7d), indicating that most of the Cu particles
are covered with the silica support. Although Cu particles sinter
easily at high temperatures, the surface Cu concentration of 10-
ZnO/Cu/SiO2 after the reduction at 300 °C is significantly higher
than that after the reduction at 250 °C (see Table 3). Hence, a part
of the fine Cu particles encapsulated in the support is supposed to
come out of the support in the process of reduction at 300 °C.
No peaks attributed to zinc oxide are present in the XRD pat-
terns of ZnO/Cu/SiO2 (see Figs. 4 and 5). The Fourier transform of
the EXAFS for 10-ZnO/Cu/SiO2 shows lack of ZnÀZn interaction
(see Fig. 11a). The spectrum is very similar to that for ZnO/SiO2
prepared by a solÀgel method and can be assigned as zinc oxide
species such as oligomers/clusters consisting of ZnO4 tetrahedra
[16]. The binding energy of Zn 2p3/2 for ZnO/Cu/SiO2 is significantly
higher than those for ZnO (1022.1 eV) and Zn2SiO4 (1021.7 eV)
[11]. Since the binding energy usually depends on the valence,
the zinc ions in ZnO/Cu/SiO2 should be electron-deficient in com-
parison with those in ZnO and it probably reflects the local struc-
ture of the zinc oxide species.
The presence of Cu2O is detected in the XRD patterns for Cu/
SiO2 and ZnO/Cu/SiO2 after the reaction at 300 °C (see Fig. 5). Since
no peak attributed to Cu2O is present in the patterns for the sam-
ples after the reduction with hydrogen at 250 °C (see Fig. 4), the
formation of Cu2O is not due to the handling of the samples in
air. That is, the Cu2O species are formed during the reaction. The
small crystallite size strongly suggests partial oxidation of the fine
Cu particles undetectable by XRD (see Table 2).
The quantity of the large Cu particles can be estimated from the
H2 consumption for the peak 2 in TPR (see Table 1). Assuming that
all the large Cu particles in Cu/SiO2 are spherical with a diameter of
39 nm and all the particles are on the surface, the Cu surface
amount on the large Cu particles will be 0.05 mmol gÀ1 (see Table
4). On the other hand, the Cu surface amount of Cu/SiO2 after the
reaction is calculated as 0.6 mmol gÀ1 (see Table 4). The amount
is evaluated from the BET surface area and the surface atomic con-
centrations because the formation of Cu2O during the reaction pre-
vents correct evaluation using N2O adsorption. The Cu surface
amount overwhelms the amount on the large particles, even if
the former amount determined from BET and XPS is overestimated.
Thus, the Cu atoms on the large particles are minor species and the
information of the Cu 2p3/2 and Cu L3VV in the XPS for the catalyst
after the reaction is considered to originate mainly from the fine
particles. As can be seen in Fig. 6c, the shape and the size of the
large Cu particles are not uniform. The mean crystallite size of
10-ZnO/Cu/SiO2 reduced at 250 °C is 25 nm. On the other hand,
the main particle size observed in Fig. 6c is a half of the mean crys-
tallite size, while some particles seem to bond together. If the par-
ticle size of Cu/SiO2 is a half of the crystallite size, the Cu surface
amount on the large particles will be 0.10 mmol gÀ1. The value is
still small in comparison with the Cu surface amount. In addition,
the actual surface amount of the large particles is supposed to be
considerably lower than the estimated value because the majority
of the large particles are probably encapsulated in the support.
Since the
a value obtained from the XPS suggests that the major
surface species are present as Cu+ (see Table 3), the fine Cu parti-
cles are oxidized to Cu2O during the reaction. This is in harmony
with the observation from the XRD measurement. In the cases of
2-ZnO/Cu/SiO2 and 5-ZnO/Cu/SiO2 the Cu surface amounts also
overwhelm those for the large Cu particles (see Table 4), suggesting
that the Cu atoms on the large particles are minor species.
Carbon monoxide is considered as a secondary product of the
steam reforming of methanol over Cu/ZnO/Al2O3 by the reverse
water gas-shift (WGS) reaction (CO2 + H2 ? CO + H2O) [17,18]. In
this mechanism CO production is enhanced at a high methanol
conversion. The CO selectivity with Cu/SiO2 increased from 1.2%
to 2.5% at a low space velocity at 300 °C, where the conversion also
increased from 62% to 88%. In the case of 5-ZnO/Cu/SiO2, the selec-
tivity of 0.6% at the conversion of 75% increased to 1.0% at the 100%
conversion. Decomposition of methanol is a possible mechanism of
CO by-production [19–21], but the CO selectivity will be indepen-
dent of the conversion. Hence, it is estimated that the CO formation
on Cu/SiO2 and ZnO/Cu/SiO2 is mainly caused by the reverse WGS
reaction.
The amount of surface Cu atoms on the catalyst can be roughly
estimated from the BET surface area and the surface atomic com-
position determined by XPS, assuming the site densities of Cu+ as
0.058 mmol mÀ2, Zn2+ as 0.077 mmol mÀ2, Si4+ as 0.35 mmol mÀ2
,
and O2À as 0.030 mmol mÀ2 on the basis of the ionic radii [4]. Both
the values for Cu/SiO2 and 10-ZnO/Cu/SiO2 reduced with hydrogen
at 250 °C for 1 h are calculated as 0.2 mmol gÀ1 (see Table 4). Since
the amounts determined by the N2O adsorption are
0.078 mmol gÀ1 and 0.098 mmol gÀ1, respectively, the values ob-
tained from BET and XPS are probably overestimated.
Table 4
Estimation of Cu surface amount of the catalysts after methanol steam reforming at
300 °C for 6 h.
Catalyst
Cu surface amounta (mmol gÀ1
)
Cu amount on particle
surfaceb (mmol gÀ1
)
Cu
Fine
Large
A mechanism for the reverse WGS has been proposed on copper
catalysts, that is, CO2 + 2Cu ? CO + Cu2O and H2 + Cu2O ? H2O + 2-
Cu [22], indicating that stabilization of Cu+ species such as Cu2O
suppresses the reverse WGS and causes low CO selectivity. Ritz-
kopf et al. reported that a zirconia-supported copper catalyst gives
low selectivity to carbon monoxide in methanol steam reforming
and the copper surface is oxidized to Cu+ state during the reaction
[23]. Hence, Cu/SiO2 is advantageous in CO suppression because
Cu+ species are dominant on the surface [4]. The CO selectivity of
Cu/SiO2 decreases significantly by addition of zinc oxide as small
Cu/SiO2
Cu/SiO2
2-ZnO/Cu/SiO2
5-ZnO/Cu/SiO2
10-ZnO/Cu/SiO2
10-ZnO/Cu/SiO2
0.6
0.2
0.7
0.6
0.4
0.2
2
0.05
0.06
0.04
0.04
0.07
0.08
c
À
2
3
2
7d
c
a
b
c
Evaluated from the surface atomic concentrations and BET surface area.
Evaluated from the crystallite size in Table 2 and the quantity in Table 1.
Just after reduction with hydrogen at 250 °C for 1 h.
d
The particle size is assumed to be 1 nm.

Y. Matsumura, H. Ishibe / Journal of Catalysis 268 (2009) 282–289
289
as 2 wt% despite an increase in the methanol conversion (see
5. Conclusions
Fig. 1), suggesting that the reverse WGS is suppressed by the addi-
tion of zinc oxide. The CO selectivity with Cu/SiO2 at the conversion
of 75% is expected to be ca. 2%. The selectivity with 5-ZnO/Cu/SiO2
is only 0.6% at the same conversion, showing that the addition of
ZnO to Cu/SiO2 is effective for the suppression of CO by-production
at 300 °C.
Since the addition of zinc oxide to the catalyst does not greatly
change the properties of copper particles as shown in Figs. 3, 5 and
8, the basic properties of the active sites remain after the ZnO mod-
ification. Nakamura et al. showed that co-existence of zinc oxide
with silica-supported copper increases oxygen coverage of copper
surface, suggesting formation of Cu+ÀOÀZn2+ species [24,25]. Peak
broadening in the Auger line of Cu L3VV is observed with the sam-
ples containing zinc oxide (see Fig. 9), and it is probably due to the
interaction between the fine Cu particles and zinc oxide which is
highly dispersed as evidenced by the EXAFS analysis (see Fig. 11).
The EDS analysis on the portion of amorphous silica, where the fine
Cu particles should be present, shows the presence of zinc with
copper (see Fig. 7d). Thus, it is estimated that the interaction stabi-
lizes Cu+ species during the reaction even at 300 °C and prevents
CO formation from hydrogen and carbon dioxide.
Addition of zinc oxide to 30 wt% Cu/SiO2 prepared by a solÀgel
method significantly suppresses CO by-production in the steam
reforming of methanol to hydrogen and carbon dioxide at 300 °C.
Zinc oxide is highly dispersed in the catalyst and can be assigned
as the species such as oligomers/clusters consisting of ZnO4 tetra-
hedra. In the reduced form of Cu/SiO2 modified/unmodified with
zinc oxide, large Cu particles with the mean crystallite size at
around 30 nm coexist with fine Cu particles which cannot be de-
tected by XRD. The fine Cu particles are oxidized to Cu2O during
the reaction. The presence of zinc oxide probably stabilizes the
Cu+ species on the surface and prevents the reduction to metallic
Cu in the reaction. Zinc oxide is also present on the surface of
the large Cu particles and may stabilize Cu+ species by forming
Cu+ÀOÀZn2+ species. The presence of zinc oxide does not affect
the activity of the Cu sites significantly.
References
[1] A. Qi, B. Peppley, K. Karan, Fuel Process. Technol. 88 (2007) 3.
[2] P. Marques, N.F.P. Ribeiro, M. Schmal, D.A.G. Aranda, M.M.V.M. Souza, J. Power
Sources 158 (2006) 504.
The activity of the large Cu particles cannot be negligible as dis-
cussed in the previous paper [4], although the amount of Cu atoms
exposed on the surface is estimated to be small. Since the Cu sur-
face amount on the large particles is significantly smaller than that
on the fine particles, the detection of the surface state of the large
particles is difficult by XPS. However, the EDS analysis shows that
Zn is present on the large Cu particles (e.g., points 2 and 3 in
Fig. 6c). Hence, zinc oxide also modifies the surface of large Cu par-
ticles and may stabilize Cu+ species by forming Cu+ÀOÀZn2+ spe-
cies [24,25]. Günter et al. reported that dissolution of zinc atoms
up to 4 mol% in Cu lattice under the reduction of Cu/ZnO with
hydrogen at 250 °C [26].
The methanol conversion with Cu/SiO2 at 300 °C increases dis-
cernibly with the addition of zinc oxide up to 5 wt% (see Fig. 1).
However, the turn-over frequencies (TOF) are calculated as
0.08 sÀ1 for Cu/SiO2, 0.07 sÀ1 for 2-ZnO/Cu/SiO2, and 0.09 sÀ1 for
5-ZnO/Cu/SiO2 on the basis of the quantity of methanol reacted
and the Cu surface amount in Table 4. Since the conversions are
in a narrow range (62À75%) where no saturation takes place, the
catalytic activity relates approximately to the conversion. Thus,
zinc oxide does not significantly activate the copper sites and the
increase in the catalytic activity is probably caused by indirect ef-
fects of zinc oxide. In the case of 10-ZnO/Cu/SiO2 the TOF is
0.14 sÀ1. Since the Cu surface amount of the large particles on
10-ZnO/Cu/SiO2 is appreciably high in comparison with the other
catalysts (see Table 4), the larger contribution of the large Cu par-
ticles may increase the conversion [4].
[3] C. Song, in: N. Brandon, D. Thompsett (Eds.), Fuel Cells Compendium, Elsevier,
Oxford, 2005, p. 53.
[4] Y. Matsumura, H. Ishibe, Appl. Catal. B 86 (2009) 114.
[5] E.K. Poels, D.S. Brands, Appl. Catal. A 191 (2000) 83.
[6] Y.-M. Lin, M.-H. Rei, Catal. Today 67 (2001) 77.
[7] F. Izumi, T. Ikeda, Mater. Sci. Forum 321–324 (2000) 198.
[8] ISO 18118, 2004.
[9] J.W. Evans, M.S. Wainright, A.J. Bridgewater, D.J. Young, Appl. Catal. 7 (1983)
75.
[10] JCPDS files, 40836, 50667, and 410254.
[11] C. Hammond, The Basics of Crystallography and Diffraction, Oxford University
Press, New York, 1997.
[12] C.D. Wagner, in: D. Briggs, M.P. Seah (Eds.), Practical Surface Analysis, Auger
and X-ray Photoelectron Spectroscopy, second ed., vol. 1, John Wiley & Sons,
Inc., New York, 1990, p. 595.
[13] Z. Wang, Q. Liu, J. Yu, T. Wu, G. Wang, Appl. Catal. A 239 (2003) 87.
[14] Y. Matsumura, N. Tode, T. Yazawa, M. Haruta, J. Mol. Catal. A 99 (1995) 183.
[15] Y. Matsumura, H. Ishibe, Appl. Catal. B 91 (2009) 524.
[16] H. Yoshida, T. Shimizu, C. Murata, T. Hattori, J. Catal. 220 (2003) 226.
[17] J. Agrell, H. Birgersson, M. Boutonnet, J. Power Sources 106 (2002) 249.
[18] H. Purnama, T. Ressler, R.E. Jentoft, H. Soerijanto, R. Schlögl, R. Schomäcker,
Appl. Catal. A 259 (2004) 83.
[19] Y. Choi, H.G. Stenger, Appl. Catal. B 38 (2002) 259.
[20] B.A. Peppley, J.C. Amphlett, L.M. Kearns, R.F. Mann, Appl. Catal. A 179 (1999)
21.
[21] A. Mastalir, B. Frank, A. Szizybalski, H. Soerijanto, A. Deshpande, M.
Niederberger, R. Schomäcker, R. Schlögl, T. Ressler, J. Catal. 230 (2005) 464.
[22] M.J.L. Ginés, A.J. Marchi, C.R. Apesteguía, Appl. Catal. A 154 (1997) 155.
´
[23] I. Ritzkopf, S. Vukojevic, C. Weidenthaler, J.-D. Grunwaldt, F. Schüth, Appl.
Catal. A 302 (2006) 215.
[24] J. Nakamura, T. Uchijima, Y. Kanai, T. Fujitani, Catal. Today 28 (1996) 223.
[25] I. Nakamura, T. Fujitani, T. Uchijima, J. Nakamura, Surf. Sci. 400 (1998) 387.
[26] M.M. Günter, T. Ressler, R.E. Jentoft, B. Bems, J. Catal. 203 (2001) 133.