Production of hydrogen by catalytic reforming of dimethoxymethane over bifunctional catalysts

Article abstract of DOI:10.1016/j.jcat.2007.03.013

Reforming of hydrocarbons and oxygenated hydrocarbons is a potential way to supply H₂ for portable and household fuel cell applications. Of course, avoiding the use of toxic fuels is desirable. The extremely low toxicity of dimethoxymethane (DMM) may make it a preferred fuel for portable and household H₂ sources. We found that DMM can be effectively reformed to hydrogen on specially designed complex catalysts with bifunctional characters. The complex catalysts consisted of a traditional Cu-ZnO/γ-Al₂O₃ for the reforming of methanol and an acidic component for the hydrolysis of DMM. The acidity of the acidic component was found to be essential for the complex catalyst. A significant amount of dimethyl ether (DME) was produced when a strong solid acid such as H-ZSM-5 was used with the Cu-ZnO/γ-Al₂O₃ for the reforming of DMM. On the other hand, γ-Al₂O₃ exhibited low activity for the hydrolysis of DMM, resulting in the low efficiency of the complex catalyst. Tests showed that the acidic carbon nanofibers (H-CNF) seemed suitable as the acidic component in the complex catalysts. In fact, the complex catalysts Cu-ZnO/γ-Al₂O₃-H-CNF were found to exhibit excellent performance for the reforming of DMM. The rate of H₂ production from the reforming of DMM on the Cu-ZnO/γ-Al₂O₃-H-CNF could be as good as that from the reforming of methanol over the traditional Cu-ZnO/γ-Al₂O₃, whereas the reforming of DMM over the Cu-ZnO/γ-Al₂O₃ without an acidic component exhibited low activity and produced a significant amount of DME. Mechanistic studies showed that DME was produced from DMM on the Cu surface when no additional acidic component was present.

Full text of DOI:10.1016/j.jcat.2007.03.013

Journal of Catalysis 248 (2007) 101–110
www.elsevier.com/locate/jcat
Production of hydrogen by catalytic reforming
of dimethoxymethane over bifunctional catalysts
Yuchuan Fu, Jianyi Shen ∗
Laboratory of Mesoscopic Chemistry, School of Chemistry and Chemical Engineering, Nanjing University, Nanjing 210093, China
Received 8 December 2006; revised 11 March 2007; accepted 12 March 2007
Abstract
Reforming of hydrocarbons and oxygenated hydrocarbons is a potential way to supply H for portable and household fuel cell applications.
2
Of course, avoiding the use of toxic fuels is desirable. The extremely low toxicity of dimethoxymethane (DMM) may make it a preferred fuel
for portable and household H sources. We found that DMM can be effectively reformed to hydrogen on specially designed complex catalysts
2
with bifunctional characters. The complex catalysts consisted of a traditional Cu–ZnO/γ -Al O for the reforming of methanol and an acidic
2
3
component for the hydrolysis of DMM. The acidity of the acidic component was found to be essential for the complex catalyst. A significant
amount of dimethyl ether (DME) was produced when a strong solid acid such as H-ZSM-5 was used with the Cu–ZnO/γ -Al O for the reforming
2
3
of DMM. On the other hand, γ -Al O exhibited low activity for the hydrolysis of DMM, resulting in the low efficiency of the complex catalyst.
2
3
Tests showed that the acidic carbon nanofibers (H-CNF) seemed suitable as the acidic component in the complex catalysts. In fact, the complex
catalysts Cu–ZnO/γ -Al O –H-CNF were found to exhibit excellent performance for the reforming of DMM. The rate of H production from
2
3
2
the reforming of DMM on the Cu–ZnO/γ -Al O –H-CNF could be as good as that from the reforming of methanol over the traditional Cu–ZnO/
2
3
γ -Al O , whereas the reforming of DMM over the Cu–ZnO/γ -Al O without an acidic component exhibited low activity and produced a signif-
2
3
2 3
icant amount of DME. Mechanistic studies showed that DME was produced from DMM on the Cu surface when no additional acidic component
was present.
© 2007 Elsevier Inc. All rights reserved.
Keywords: Acidic carbon nanofibers; Copper catalyst; Dimethoxymethane; Hydrogen production; Steam reforming
1. Introduction
much more difficult than that of methanol and involves more
challenges [6,7]. The production of hydrogen from catalytic
reforming of biomass-derived compounds [2,8] and dimethyl
ether (DME) [9–11] also has been reported in recent years.
Generally, the production of H2 for fuel cells from a carbon-
containing compound involves reforming of the fuel, followed
by the water–gas shift reaction (WGSR) and preferential ox-
idation (PROX) of CO in H2 rich gas to reduce CO in the
gas stream to an extent that can be tolerated by the anode
of a proton-exchange membrane fuel cell (PEMFC). Because
WGSR is exothermic, a low reaction temperature (e.g., 400–
500 K) favors equilibrium and thus the low concentration of
CO. However, the activity of the present WGSR catalysts is low
at low temperatures [12,13], and recent efforts have been ded-
icated to exploring this issue [13–15]. The PROX process was
developed to further reduce the concentration of CO in H2-rich
gas [5,16]. PROX requires the injection of O2 into the stream of
The production of H2 through reforming of hydrocarbons
and oxygenated organic compounds for fuel cells has been a
common research focus in academia and industry for decades
[1–3]. In particular, reforming of methanol and gasoline has
received much attention. Methanol is easily reformed, and its
use has been demonstrated by Daimler Chrysler in a prototype
NECAR3 vehicle [4]. The disadvantage of using methanol as
a fuel is that it is poisonous. This problem is even more se-
rious when methanol is used in a microreformer for portable
H2 sources [5]. The advantage of using gasoline as a fuel for
the production of H2 is the existing infrastructure for the dis-
tribution of gasoline; however, the reforming of gasoline is
*
Corresponding author. Fax: +86 25 83317761.
E-mail address: jyshen@nju.edu.cn (J. Shen).
0021-9517/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2007.03.013

102
Y. Fu, J. Shen / Journal of Catalysis 248 (2007) 101–110
H2 to oxidize CO to CO2. The O2 injected is usually in excess
of the stoichiometric amount for CO oxidation, and some frac-
tion of H2 is consumed [17,18]. Recent reports from Dumesic
et al. [19,20] suggested a process for the use of reformate gas
containing H2 and CO for PEMFC. They used an intermedi-
ate (a reducible polyoxometalate compound [POM]) in the fuel
cell; the POM was reduced by CO, and the reduced POM was
readily oxidized on the anode to generate electricity. Through
this process, the WGSR reaction is bypassed, and the CO can be
used directly as an additional source of energy when the stream
of H2 is purified [20].
through the treatment with concentrated nitric acid and found
that they were quite stable below 573 K. In the present work,
we compared H-ZSM-5, γ -Al2O3, and acidic CNFs as acidic
components for the hydrolysis of DMM and found that the sur-
face acidity (nature and strength) of the acidic component is the
essential factor in the effective reforming of DMM on complex
catalysts.
2. Experimental
2.1. Preparation of catalysts
Dimethoxymethane (DMM), or methylal, has extremely low
toxicity [21] and may be produced in a large scale [22]. Re-
cently, it was reported that DMM can be produced by the
direct oxidation of methanol [23,24], which will be more ef-
ficient if the process is industrialized. DMM is usually used
as a solvent in pharmaceutical and perfume applications, as a
methoxymethylation reagent in organic synthesis [25], and as
an intermediate for the production of concentrated formalde-
hyde [26]. DMM is generally an environmentally benign chem-
ical, being easily degraded in air with a lifetime of about 2 days,
without the formation of organic peroxide intermediate and
photochemical pollution [21].
Recently, we found that DMM can be effectively reformed to
produce H2 over Cu–ZnO/Al2O3-NbP complex catalysts [27].
The reforming of DMM by water can be considered to com-
prise reactions (1)–(3)—that is, the hydrolysis of DMM to form
methanol and formaldehyde, which are then further reformed to
produce H2 and CO2—although the actual pathways might be
more complicated:
The bifunctional catalysts for the reforming of DMM were
composed of a CuZnAl catalyst and an acidic component. The
CuZnAl catalyst was prepared according to previous patents
[38,39]. During preparation, precipitates were formed through
three steps: (1) addition of an aqueous solution containing
Cu(NO3)2 and Zn(NO3)2 into a solution of NaHCO3 up to
pH 7.0, to obtain a slurry (denoted as A); (2) addition of di-
lute solution of ammonia into the aqueous solution of Al(NO3)3
up to pH 7.0, to obtain a slurry of Al(OH)3 (denoted as B);
and (3) mixing of the slurries A and B to form a mixture that
was stirred vigorously for 15 min and aged for 1 h. After wash-
ing, the precipitate was dried at 378 K for 12 h and calcined at
623 K for 3 h. The CuZnAl catalyst prepared contained 63%
Cu, 21% Zn, and 16% Al in moles. The catalyst was pressed,
crushed, and sieved to granules of 40–60 mesh for further use.
H-ZSM-5, γ -Al2O3, and acidic CNF (H-CNF) were used
as the acidic components for the hydrolysis of DMM. The
H-ZSM-5 (SiO2/Al2O3 = 25, 450 m2 g−1) and γ -Al2O3
(290 m2 g−1) were commercial products and were calcined
in air at 573 K for 3 h before the catalytic tests. The CNF
was prepared by the decomposition of propylene on an unsup-
ported Cu–Ni (4:6 by weight) catalyst following the procedure
described by Shen et al. [40]. The H-CNF was obtained by
treating the CNF in concentrated nitric acid (63 wt%). Specifi-
cally, 30 mL of concentrated nitric acid was added per 1 g CNF,
and the mixture was refluxed for 30 min. Then the H-CNF was
filtered, washed thoroughly with deionized water, and dried at
393 K for 12 h.
CH3OCH2OCH3 + H2O → 2CH3OH + CH2O,
CH3OH + H2O → 3H2 + CO2,
CH2O + H2O → 2H2 + CO2.
(1)
(2)
(3)
The hydrolysis of DMM [Eq. (1)] requires an acidic catalyst,
whereas the reforming of methanol and formaldehyde [Eqs. (2)
and (3)] usually uses Cu–ZnO/γ -Al2O3 catalyst (designated
CuZnAl below) [28,29]. Thus, an effective catalyst, compris-
ing acidic and CuZnAl components, must be developed for the
reforming of DMM.
Zeolites and γ -Al2O3 are two types of widely used solid
acids. Typically, a zeolite has mainly Brønsted acid sites on
its framework [30], whereas γ -Al2O3 exhibits mainly Lewis
acidity [31]. The application of carbon nanofibers (CNFs) as
catalyst supports has received extensive attention over the last
10 years [32–34]. It has been found that the catalytic behavior
of nickel crystallites is altered dramatically when the metal is
dispersed on CNFs compared with on activated carbon (AC)
and γ -alumina [32]. Dong et al. [35] found that the addi-
tion of carbon nanotubes (CNTs) into the CuZnAl significantly
enhances the catalytic activity for the synthesis of methanol.
CNFs are usually oxidized by concentrated nitric acid or in a
mixture of concentrated nitric and sulfuric acids before being
used as catalyst supports. Such treatments result in the forma-
tion of large amounts of carboxylic and phenol groups, which
are acidic [36]. Toebes et al. [37] studied the thermal stabil-
ity of the acidic oxygen-containing groups in CNFs formed
The complex catalysts containing H-ZSM-5 or γ -Al2O3
for the reforming of DMM were prepared by mixing with
the CuZnAl. The resulting mixture was then ground, pressed,
and sieved to granules of 40–60 mesh. The complex cata-
lyst CuZnAl–H-CNF containing 12.5% H-CNF was prepared
by adding the H-CNF into the solution of Al(NO3)3 before
slurry B was formed for the preparation of the CuZnAl catalyst,
whereas the complex catalysts containing 20 and 40% H-CNF
were obtained by mixing in appropriate amounts of CuZnAl
and H-CNF.
2.2. Characterization of catalysts
X-ray diffraction (XRD) patterns were collected in ambi-
ent atmosphere by an X-ray diffractometer (X’TRA, ARL Co.,
Switzerland) with CuKα radiation (λ = 1.5418 Å). The 2θ

Y. Fu, J. Shen / Journal of Catalysis 248 (2007) 101–110
103
scans covered the range 10◦–70◦. The applied voltage and cur-
rent were 40 kV and 40 mA, respectively. The reduced copper
catalysts were passivated by flowing N2 containing about 0.1%
O2 at room temperature for 12 h. Scherrer’s equation was used
for estimating the crystallite size of Cu in the catalysts accord-
ing to the broadening of the XRD peak at 2θ = 43.472◦.
Transmission electron microscopy (TEM) images were ob-
tained using a JEM2100 instrument operated at 200 kV. Sam-
ples were prepared by suspending the catalysts in ethanol under
ultrasonic vibration. Some drops of a suspension were brought
onto a carbon film on a copper grid. The particle sizes were ob-
tained by counting 50–100 particles in each sample. Scanning
electron micrography (SEM) was done with a Hitachi S800 mi-
croscope at an accelerating voltage of 15 kV.
the TCD was connected to a TDX-01 column for the analysis of
methane, COx, and N2. The amount of H2 produced was usually
calculated according to the conversion of DMM and the com-
position of other products and was checked using another gas
chromatograph with a TDX-01 column. (This gas chromato-
graph also can be used for the analysis of formaldehyde when
switched to a Porapak-N column when necessary.) The reform-
ing of methanol was also performed for comparison.
The selectivity to H2 for the reforming of DMM was de-
fined as the total mol of H2 actually produced divided by the
mol of H2 theoretically produced according to the reaction
CH3OCH2OCH3 + 4H2O = 3CO2 + 8H2 [27]. The selectivity
to a carbonaceous product is defined as the amount of carbon in
this product divided by the total amount of carbon in converted
DMM. The byproducts found were CH4, CH3OH, and DME.
The surface areas of the catalysts were determined by N2
adsorption at 77.3 K using the BET method. The active sur-
face areas of copper in the catalysts were determined using the
technique of adsorptive decomposition of N2O at 363 K accord-
ing to Chinchen et al. [41]. The surface areas of copper were
calculated according to the amount of N2 produced and the sto-
ichiometric reaction N2O + 2Cu = Cu2O + N2.
3. Results and discussion
3.1. Characterization of catalysts
Fig. 1 shows the SEM images of the as-prepared (CNF) and
nitric acid-treated (H-CNF) CNFs. The particles in these sam-
ples had an average diameter of about 300 nm and surface areas
of 175 m2 g−1 for CNF and 169 m2 g−1 for H-CNF. Treatment
with nitric acid did not seem to cause any significant changes
in morphology and surface area, but it did rupture some fibers.
In addition, the XRD data (not shown here) revealed that the
graphite phase of the CNF was not changed and no metal im-
purities (Cu or Ni) were detected in the CNF after the treatment
with nitric acid.
Fig. 2 shows the XRD patterns of the CuZnAl and CuZnAl–
12.5% H-CNF catalysts calcined at 623 K. The phases de-
tected in the CuZnAl were mainly CuO and some ZnCO3 and
Cu6Al2(OH)16CO3. Incorporation of H-CNF seemed to retard
the decomposition of precursors of CuO, because substantial
amounts of Cu2(OH)2CO3 was detected in the CuZnAl–12.5%
H-CNF sample. This might be due to the decomposition of car-
boxylic groups on the surface of H-CNF, which released CO2
and inhibited the decomposition of Cu2(OH)2CO3. The aver-
age particle sizes of CuO in the CuZnAl and CuZnAl–12.5%
H-CNF catalysts were estimated as 11 and 10 nm, respectively,
according to Scherrer’s method. The XRD patterns of the cata-
lysts after reduction are shown in Fig. 3. Both samples exhibited
essentially the same features of metallic copper and ZnO after
reduction. In addition, the average sizes of metallic copper par-
ticles as determined by Scherrer’s method were almost the same
(13 nm) for the two catalysts.
Temperature-programmed reduction (TPR) was performed
using a quartz U-tube reactor loaded with about 50 mg of
sample. A mixture of N2 and H2 (5.13% H2 by volume) was
used and the flow rate was maintained at 40 mL min−1. The
hydrogen consumption was monitored using a thermal conduc-
tivity detector (TCD). The temperature was raised from 303 to
1173 K at a programmed rate of 5 K min−1
.
Microcalorimetric measurements of ammonia adsorption
were performed to determine the surface acidity of the sam-
ples at 423 K. A Setaram C-80 calorimeter was connected
to a volumetric system equipped with a Baratron capacitance
manometer for the pressure measurement and gas handling.
Before the microcalorimetric adsorption of ammonia, HZSM-5
and γ -Al2O3 were activated at 673 K in 500 Torr O2, followed
by evacuation at the same temperature for 1 h, whereas the CNF
samples were directly outgassed at 533 K for 1 h. After ther-
mal equilibrium was reached for a sample in vacuum at 423 K,
microcalorimetric data were collected with probe molecules of
NH3 dosed sequentially. Elemental analysis was performed us-
ing an inductively coupled plasma method.
2.3. Tests of catalytic activity
The catalytic reactions for the hydrolysis and reforming of
DMM were performed in a fixed-bed microreactor. The DMM
and H2O were introduced to the reaction zone by bubbling N2
(99.999%) through a glass saturator filled with DMM (Aldrich,
99%) maintained at 273 K and a glass saturator filled with H2O
maintained at 333 K, respectively. The feed composition was
maintained at N2:H2O:DMM = 24:5:1 (v/v). The catalysts con-
taining copper were prereduced at 533 K (raised from room
temperature at a rate of 1 K min−1) for 3 h in 12% H2/N2 at a
flow rate of 50 mL min−1. The products were analyzed by an
online gas chromatograph equipped with an FID and a TCD.
The FID was connected to a Porapak N column for the sep-
aration of methanol, DMM and other organic compounds, and
The number of surface copper atoms was found to be
−1
−1
557 µmol g
for CuZnAl and 483 µmol g
for CuZnAl–
cat
cat
12.5% H-CNF by the adsorptive decomposition of N2O. The
dispersion of metallic Cu (7.8 and 7.9%), as well as the crystal-
lite size of the copper particles (13.4 and 13.2 nm), were found
to be almost the same for the two samples.
Fig. 4 compares the TEM images of the pure CuZnAl cata-
lyst with that of the complex CuZnAl–12.5% H-CNF catalyst
before and after reduction. The particles seen in the images of
the nonreduced catalysts are irregular, with an average size of

104
Y. Fu, J. Shen / Journal of Catalysis 248 (2007) 101–110
Fig. 1. SEM images of carbon nanofibers: (a) as-prepared and (b) refluxed in 63% HNO for 30 min.
3
to the surface of carbon fiber in Fig. 4b might be the mix-
ture of CuO and Cu2(OH)2CO3 before reduction (see Fig. 2b).
Figs. 4aꢀ and 4bꢀ present TEM images of the two catalysts af-
ter reduction. The spherical particles seen in Fig. 4aꢀ might be
the 10- to 20-nm particles of metallic copper, consistent with
those determined according to the XRD peak width and Scher-
rer’s equation, as well as the N2O adsorption. A similar result
was found for the CuZnAl–12.5% H-CNF catalyst after reduc-
tion, with 10- to 20-nm metallic copper particles attached to the
surface of carbon fiber clearly visible (Fig. 4bꢀ).
The results of microcalorimetric adsorption of ammonia are
presented in Fig. 5. The coverage of ammonia is usually defined
as the amount of ammonia adsorbed with the differential heat
>40 kJmol−1. The as-prepared CNF was nonacidic, because
Fig. 2. X-ray diffraction patterns of (a) CuZnAl and (b) CuZnAl–12.5% H-CNF
catalysts after calcination at 623 K.
the associated heat of adsorption of NH3 was <10 kJmol−1
.
Treatment with nitric acid greatly increased its surface acidity.
The initial heat and NH3 coverage for H-CNF were found to be
about 150 kJ mol−1 and 330 µmol g−1, respectively. This result
is consistent with that reported by Toebes et al. [37] for the ni-
tric acid-treated CNF [37]. The number of acid sites on the sur-
face of our H-CNF as titrated by NH3 was 1.2 nm−2. Although
the surface acidity of the H-CNF was decreased on reduction
at 533 K in 12% H2/N2 (a condition used for the reduction of
CuZnAl), it still had substantial surface acidity with an initial
heat of about 125 kJ mol−1 and a coverage of 260 µmol g−1 for
the adsorption of ammonia. The surface acidity of H-ZSM-5
was strong with the initial heat of about 180 kJ mol−1 and
coverage of about 700 µmol g−1 for the adsorption of NH3.
The initial heat and coverage for ammonia adsorption on the
γ -Al2O3 were about 220 kJ mol−1 and 400 µmol g−1, respec-
tively. The surface acidity of the γ -Al2O3 was weaker than that
of H-ZSM-5, but stronger than that of H-CNF in terms of the
microcalorimetric adsorption of ammonia.
Fig. 3. X-ray diffraction patterns of (a) CuZnAl and (b) CuZnAl–12.5% H-CNF
catalysts after reduction at 533 K in 12% H /N for 3 h and passivation at room
2
2
temperature in about 0.1% O /N for 12 h.
2
2
The results of TPR for the CuZnAl and CuZnAl–12.5%
H-CNF catalysts are compared in Fig. 6. The two samples
exhibited similar TPR profiles with three peaks each. These
peaks may indicate the reduction of different copper precur-
sors to metallic copper in the catalysts. The addition of H-CNF
about 15 nm, although the phases of the particles might be dif-
ferent for different catalysts. The particles in the CuZnAl before
reduction (Fig. 4a) may be mainly CuO with some ZnCO3 and
Cu6Al2(OH)16CO3, as revealed by XRD. The particles attached

Y. Fu, J. Shen / Journal of Catalysis 248 (2007) 101–110
105
ꢀ
ꢀ
Fig. 4. TEM images for CuZnAl before (a) and after (a ) the reduction, as well as for CuZnAl–12.5% H-CNF before (b) and after (b ) the reduction. The reduction
was performed at 533 K in 12% H /N for 3 h, followed by passivation at room temperature in about 0.1% O /N for 12 h.
2
2
2
2
Fig. 5. Differential heat versus coverage for NH adsorption at 423 K over
3
Fig. 6. Temperature-programmed reduction profiles of CuZnAl and
CuZnAl–12.5% H-CNF catalysts.
H-ZSM-5, γ -Al O , CNF, H-CNF, and H-CNF reduced at 533 K in 12%
2
3
H /N for 3 h.
2
2
of surface area of the catalysts, active surface area of copper,
and dispersion and particle size of copper were essentially the
same. Thus, the addition of H-CNF did not seem to affect the
main physical properties of the CuZnAl catalyst.
seemed to promote the reduction of copper species, because
each TPR peak shifted to the lower temperature for CuZnAl–
12.5% H-CNF compared with CuZnAl. This result is consis-
tent with the findings of Dong et al. [35], who reported that
the addition of CNT lowered the reduction temperatures of the
copper-based catalysts for methanol synthesis.
3.2. Hydrolysis of DMM
Table 1 summarizes some physical properties of the CuZnAl
and CuZnAl–12.5% H-CNF catalysts. The data in Table 1 show
that the main physical properties of the two catalysts in terms
Table 2 presents the results for the hydrolysis of DMM over
H-ZSM-5, γ -Al2O3, and H-CNF. The conversion of DMM on

106
Y. Fu, J. Shen / Journal of Catalysis 248 (2007) 101–110
Table 1
Physical properties of the CuZnAl and CuZnAl–12.5% H-CNF catalysts
a
Catalyst
S
wt%
of Cu
Copper on the surface
Dispersion
of Cu (%)
d
d
copper
(nm)
copper
b
BET
2
c
−1
−1
−1
)
cat
2
(m g
)
(nm)
(m g
)
(µmol g
cat
CuZnAl
CuZnAl–12.5% H-CNF
99
101
45.2
38.9
22.8
19.8
557
483
7.8
7.9
13.4
13.2
13.1
13.3
a
Determined by using the technique of adsorptive decomposition of N O.
Particle size of metallic copper as calculated according to the dispersion of copper.
Particle size of metallic copper as calculated according to the Sherrer’s equation.
2
b
c
Table 2
a
Hydrolysis of DMM over H-ZSM-5, γ -Al O and H-CNF
2
3
b
Catalyst
Temp.
(K)
GHSV
DMM
conv. (%)
Selectivity (%)
−1
−1
)
(ml g
h
5
Methanol
DME
cat
H-ZSM-5
473
493
473
493
513
473
493
513
2.0 × 10
2.0 × 10
2.3 × 10
2.3 × 10
2.3 × 10
9.0 × 10
9.0 × 10
9.0 × 10
98.1
99
2.9
60.6
45.9
99.3
96.2
97.7
98.4
98.3
98.9
39.4
54.1
0.7
3.8
2.3
1.6
1.7
1.1
5
4
4
4
4
4
4
γ -Al O
2
3
8.3
15.6
93.4
98.2
98.3
c
H-CNF
a
Feed composition: N /H O/DMM = 24/5/1 (v/v).
2
2
b
The selectivity to methanol and DME was normalized, and other products
Fig. 7. Effect of reduction of H-CNF at different temperatures in 12% H /N
2
2
such as formaldehyde, methyl formate and trace of methane were not included.
on the hydrolysis of DMM. Reaction conditions: N /H O/DMM = 24/5/1 (v/v)
2
2
c
4
−1 −1
h .
Prior to the hydrolysis test, the H-CNF was reduced at 533 K in 12% H /N
and GHSV = 9.0 × 10 mlg
2
2
for 3 h, which was the same condition for the reduction of CuZnAl–H-CNF for
the reforming of DMM.
microcalorimetric adsorption of ammonia showed that the H-
CNF exhibited fairly strong surface acidity with the initial heat
of 150 kJ mol−1 and coverage of 330 µmol g−1 for the adsorp-
tion of NH3. The reduction in 12% H2/N2 at 533 K decreased
the surface acidity (Fig. 5) and thus decreased the activity for
the hydrolysis of DMM. The surface acidity of CNF would be
further decreased if reduction were done at higher temperatures.
Accordingly, the activity of H-CNF for the hydrolysis of DMM
decreased with increasing temperature of reduction. The data in
Fig. 7 show that when the hydrolysis of DMM was carried out
at 473 K, the conversion of DMM over the H-CNF was about
93% when reduced at 533 K and 76% when reduced at 673 K.
These results indicate that the acidic functional groups on the
surface of H-CNF might be responsible for the hydrolysis of
DMM. Although the conversion of DMM was lower on the H-
CNF reduced at 673 K compared with the nonreduced H-CNF
for the hydrolysis of DMM at 473 K, the activity of the reduced
H-CNF was still much higher than that of γ -Al2O3. For ex-
ample, the rate of DMM conversion over the H-CNF reduced
at 673 K was much higher than that over the γ -Al2O3 for the
hydrolysis of DMM (about 100 times higher when reduced at
473 K and 20 times higher when reduced at 513 K). The conver-
sion of DMM was about 96% on the H-CNF reduced at 673 K
for the hydrolysis of DMM at 533 K. Stable surface acidity of
H-CNF is the primary requirement for a stable complex catalyst
CuZnAl–H-CNF for the reforming of DMM.
H-ZSM-5 was high (>98%) with a substantial amount of DME
as the hydrolysis product at 473 and 493 K. Apparently, the
methanol produced from the hydrolysis of DMM was further
converted to DME over this strong solid acid. Thus, H-ZSM-5
does not seem to be suitable as an acidic component for the
reforming of DMM. In contrast, the γ -Al2O3 exhibited low
conversion of DMM up to 513 K (15.6%), although the selec-
tivity to methanol was high for the hydrolysis of DMM. Thus,
a high fraction of γ -Al2O3 would be required for the complex
catalyst, which would lower the efficiency of DMM reforming.
Data in Table 2 show that the H-CNF exhibited high activity
with high selectivity to methanol for the hydrolysis of DMM.
For example, both the conversion of DMM and selectivity to
methanol were >98% on the H-CNF at 493 K with space ve-
locity four times greater than that for the hydrolysis of DMM
on γ -Al2O3. These results clearly indicate that the choice of
solid acid with proper surface acidity might be the essential
factor in determining the catalytic behavior of the complex cat-
alysts for the reforming of DMM to produce H2. It should be
mentioned that before the hydrolysis test, the H-CNF was re-
duced at 533 K in 12% H2/N2 for 3 h, the same condition for
the reduction of CuZnAl and CuZnAl–H-CNF. The stability of
surface acidity of H-CNF resistant to the reduction by H2 is im-
portant, because the reforming of DMM is performed under the
reducing atmosphere.
The microcalorimetric results show that the surface acidity
of γ -Al2O3 was stronger than that of H-CNF in terms of the
initial heat and uptake for the adsorption of NH3. However, the
activity of γ -Al2O3 was much lower than that of H-CNF sam-
The effect of reduction of H-CNF by H2 on the hydrolysis of
DMM was further investigated; the results are shown in Fig. 7.
The activity of H-CNF was high for the hydrolysis of DMM; the
conversion of DMM was 100% over H-CNF at 473 K. In fact,

Y. Fu, J. Shen / Journal of Catalysis 248 (2007) 101–110
107
Table 3
a
Reforming of DMM over the CuZnAl and complex catalysts
b
Catalyst
DMM
Selectivity to carbonaceous products (%)
Temp.
(K)
Conv.
(%)
H
(%)
sel.
CO
CO
CH OH
3
DME
CH
4
2
2
c
c
c
c
CuZnAl
493
513
533
553
493
513
473
493
513
533
553
473
493
513
533
2.8
6.9
14.8
30.2
100
87.1
84.5
81.2
77.9
62.0
78.0
79.9
94.0
96.2
95.3
93.7
93.8
98.3
97.5
96.5
n.d.
n.d.
n.d.
n.d.
0.2
88.6
86.5
83.6
80.8
65.9
80.9
82.2
94.7
96.0
91.5
87.6
94.5
98.4
94.9
92.3
2.9
1.3
0.8
0.5
1.2
1.0
7.7
1.5
0.9
0.0
0.0
3.2
0.4
0.0
0.0
8.2
11.6
14.7
17.4
32.4
17.2
9.8
3.7
2.0
1.9
2.0
0.3
0.6
0.8
1.3
0.0
0.9
0.3
0.1
0.1
0.1
0.1
0.1
0.1
0.0
0.0
CuZnAl–H-ZSM-5
100
5.6
1.2
c
c
CuZnAl–γ -Al O
n.d.
n.d.
1.0
2
3
30.2
79.5
99.9
100
37.8
99.6
100
6.5
10.3
c
CuZnAl–H-CNF
n.d.
0.3
4.3
6.9
2.2
0.9
0.8
0.7
100
−1
−1
h .
cat
a
3
Conditions for DMM reforming: N /H O/DMM = 24/5/1 (v/v) and GHSV = 4.55 × 10 mlg
2
2
b
c
The content of HZSM-5, γ -Al O , and H-CNF in the complex catalysts was 25, 33, and 12.5%, respectively.
2
3
Not detectable.
ples for the hydrolysis of DMM. This might be due to the fact
that the acid sites on γ -Al2O3 are mainly Lewis type [31], and
the strong Lewis acid sites might be converted into weak Brøn-
sted sites when H2O was adsorbed during the reaction of DMM
hydrolysis [42,43]. On the other hand, this phenomenon might
reflect the nature of surface acidity. The acid sites on H-CNF
are known to be the surface carboxylic and phenol groups [36],
which are Brønsted type in nature. In addition, niobium phos-
phate exhibited mainly Brønsted acidity and high activity for
the hydrolysis of DMM [27]. Thus, the solid acids with fairly
strong Brønsted acidity seemed to be a suitable acidic com-
ponent for the reforming of DMM to produce H2. In contrast,
strong Brønsted solid acids, such as H-ZSM-5, and Lewis solid
acids, such as γ -Al2O3, did not function properly in this role.
the reforming of DMM over the CuZnAl–H-ZSM-5 at temper-
atures above 513 K.
The complex catalyst CuZnAl–γ -Al2O3 exhibited much
higher activity than the pure CuZnAl for the reforming of
DMM. In addition, the selectivity to H2 was high for the re-
forming of DMM over the CuZnAl–γ -Al2O3. However, the
conversion of DMM was low at the temperatures below 513 K,
due to the low activity of γ -Al2O3 for the hydrolysis of DMM.
The conversion of DMM increased with increasing reaction
temperature and achieved 100% at 533 K. It is interesting to
note that the selectivity to H2 over the CuZnAl–γ -Al2O3 passed
through a maximum with the increase of reaction temperature
from 473 to 553 K. At 473 K, γ -Al2O3 exhibited low activ-
ity for the hydrolysis of DMM; thus, some of the DMM was
directly converted to DME on the surface of CuZnAl. The ac-
tivity of γ -Al2O3 for the hydrolysis of DMM increased with
increasing reaction temperature, leading to increased selectivity
to H2 for the reforming of DMM on complex catalyst CuZnAl–
γ -Al2O3. However, the selectivity to H2 would decrease with
further increases in the reaction temperature, due to the equilib-
rium limitation of the WGSR.
The CuZnAl–12.5% H-CNF catalyst exhibited high activity
and selectivity to H2 for the reforming of DMM even at rela-
tively low reaction temperatures. For example, the conversion
of DMM over the CuZnAl–12.5% H-CNF was nearly 100%
at 493 K, 40 K lower than the reaction temperature for the
same conversion of DMM over the CuZnAl–γ -Al2O3. Appar-
ently, the high efficiency of the CuZnAl–12.5% H-CNF for the
reforming of DMM results from the high activity and selec-
tivity of the H-CNF for the hydrolysis of DMM to methanol
and formaldehyde. High temperature would cause the sintering
of copper crystallite and accelerate the deactivation of CuZnAl
catalyst [44]; thus, the low reaction temperature enhances the
durability of the complex catalysts. Moreover, a low reaction
temperature favors a low concentration of CO in the refor-
3.3. Reforming of DMM
Table 3 presents the results for the reforming of DMM over
CuZnAl as well as the complex catalysts CuZnAl–HZSM-5,
CuZnAl–γ -Al2O3, and CuZnAl–H-CNF. For comparison, the
complex catalysts comprised 25% HZSM-5, 33% γ -Al2O3, and
12.5% H-CNF. The CuZnAl catalyst exhibited low activity for
the reforming of DMM and produced a significant amount of
DME as a byproduct, probably due to the lack of acidic sites in
the catalyst (we address this issue further in Section 3.4).
The conversion of DMM was greatly enhanced on the addi-
tion of any of the acidic components. It seems that the catalytic
behavior of the complex catalysts for the reforming of DMM
is closely related to the catalytic behavior of the corresponding
acid components for the hydrolysis of DMM. The activity of
CuZnAl–H-ZSM-5 for the reforming of DMM was high, with
100% conversion of DMM at 473 K, but the selectivity to H2
was low due to the formation of substantial amount of DME. In
addition, some light hydrocarbons were also produced during

108
Y. Fu, J. Shen / Journal of Catalysis 248 (2007) 101–110
Table 4
Comparison of the rate of hydrogen production from the reforming of DMM and methanol at 513 K
Catalyst
Fuel
GHSV
Conv.
(%)
Rate of H production
Selectivity (%)
2
−1
−1
)
−1
−1
−1 −1 c
s )
(ml g
h
(ml g
h
)
(H Cu
2
H
CO
CO
CH OH
3
DME
cat
2
2
cat
a
a
4
CuZnAl
CH OH
3
2.6 × 10
2.6 × 10
2.1 × 10
2.1 × 10
2.1 × 10
94.2
91.6
45.9
77.3
99.7
5230
5090
2500
4130
5200
0.12
0.13
0.06
0.11
0.19
99.8
99.8
97.4
95.6
93.0
0.7
0.6
n.d.
n.d.
0.6
99.3
99.4
97.7
96.1
93.3
–
–
1.1
3.6
6.1
0.0
0.0
1.2
0.3
0.0
4
4
4
4
CuZnAl–12.5% H-CNF
CuZnAl–12.5% H-CNF
CuZnAl–20% H-CNF
CuZnAl–40% H-CNF
a
CH OH
3
b
d
d
DMM
DMM
DMM
b
b
Feed conditions for methanol reforming: N /H O/CH OH = 11.8/1.25/1 (v/v).
2
2
3
b
c
d
Feed conditions for DMM reforming: N /H O/DMM = 24/5/1 (v/v).
2
2
Molecules of H per atom of surface Cu per second.
2
Not detectable.
mate gas, because it is determined mainly by the equilibrium
of WGSR.
and
CH3O* + * → CH2O* + H*.
(5)
Table 4 shows the results of reforming of methanol and
DMM over the CuZnAl and CuZnAl–H-CNF with varying
H-CNF content at 513 K. The reforming of methanol over the
CuZnAl and CuZnAl–12.5% H-CNF was essentially the same
in terms of conversion of methanol, rate of hydrogen produc-
tion, and selectivity to H2, implying that the addition of H-CNF
did not affect the catalytic behavior of CuZnAl for the reform-
ing of methanol.
Because the hydrolysis of 1 mol of DMM produced 2 mol of
methanol and 1 mol of formaldehyde (with formaldehyde re-
formed much faster than methanol), the rate of H2 production
from the reforming of DMM over CuZnAl–H-CNF must be
33% greater than that from the reforming of methanol over
CuZnAl as long as the hydrolysis of DMM is sufficiently rapid
on acid sites. Thus, if a better acidic component than H-CNF
could be developed and used with CuZnAl for the reforming
of DMM, then the mass ratio of CuZnAl in the complex cata-
lyst could be higher, and a higher rate of H2 production might
be achieved. In this case, H2 production from the reforming of
DMM on the complex catalyst might be more effective than that
from the reforming of methanol on CuZnAl.
The concentration of CO in the reformate gas produced from
the reforming of methanol is determined mainly by the equilib-
rium of the WGSR [27]. Similarly, the CO concentration in the
reformate gas from the reforming of DMM is close to the equi-
librium value of WGSR when the conversion of DMM is close
to 100%, as calculated according to the data in Table 3 for the
CuZnAl–H-CNF at temperatures above 513 K.
The data in Table 4 also show the effect of the amount of
H-CNF in CuZnAl–H-CNF on the reforming of DMM. The
conversion of DMM was much lower than that of methanol
over CuZnAl–12.5% H-CNF at 513 K with similar space ve-
locity. The conversion of DMM increased with an increasing
amount of H-CNF in CuZnAl–H-CNF for the reforming of
DMM. The conversion of DMM was found to be 77.3% for
the complex catalyst containing 20% H-CNF and 99.7% for
that containing 40% H-CNF. The rate of H2 production was
also increased accordingly, reaching about 5200 ml g−ca1t h−1 for
the reforming of DMM over CuZnAl–40% H-CNF at 513 K,
almost the same as that from the reforming of methanol over
CuZnAl at−t1he same temperature with similar space velocity
(5230 ml gcat h
−1), with conversions of methanol and DMM
3.4. Mechanistic aspect of DME formation on CuZnAl
reaching 94.2 and 99.7%, respectively.
This result demonstrates that the reforming of DMM over
the complex CuZnAl–H-CNF catalysts could be as effective
as that of methanol over CuZnAl. In addition, when the activ-
ity is calculated based on the number of copper sites on the
surface, the rate of H2 production was much higher for the re-
forming of DMM on CuZnAl–40% H-CNF (0.19 molecules of
H2 per atom of surface Cu per second) than that of methanol
on CuZnAl (0.12 molecules of H2 per atom of surface Cu per
second).
It is generally known that the reforming of methanol in-
volves the dissociative adsorption of methanol to form the sur-
face methoxy groups [Eq. (4)] [29,45,46], with the abstraction
of an additional H from the methoxy group to form adsorbed
formaldehyde [Eq. (5)] as the rate-determining step [29,47].
This implies that the following steps that could be taken as the
reforming of formaldehyde were relatively fast.
The data in Table 3 show that the reforming of DMM
over CuZnAl exhibited low activity and produced a significant
amount of DME. One apparent reason for the low activity in
the CuZnAl catalyst was a lack of acidity for the hydrolysis
of DMM. Thus, understanding why DMM was converted into
DME over CuZnAl is of importance. Because the reforming of
methanol over CuZnAl produces mainly H2 with little DME,
there must be a pathway for the formation of DME from DMM
over CuZnAl. The degradation of DMM over the copper sur-
face might involve the following pathways:
CH3OCH2OCH3 + MO + M → MOCH3 + CH3OCH2OM,
(6)
MOCH2OCH3 + MO + M → MOCH3 + MOCH2OM, (7)
MOCH2OCH3 + MOCH3 → MOCH2OM + CH3OCH3, (8)
CH3OH + 2* → CH3O* + H*
(4)
MOCH2OM → 2M + H2 + CO2,
(9)

Y. Fu, J. Shen / Journal of Catalysis 248 (2007) 101–110
109
could be as good as that of the traditional method of reform-
ing methanol on the CuZnAl catalyst. The traditional CuZnAl
catalyst alone exhibited low activity for the reforming of DMM
and produced a significant amount of DME on the copper sur-
face. The addition of an acidic component with proper nature
and strength is essential for the reforming of DMM. H-ZSM-5
is a strong solid acid that is not suitable for the hydrolysis of
DMM due to the formation of a significant amount of dimethyl
ether, and the Lewis solid acid γ -Al2O3 is not sufficiently ac-
tive for the hydrolysis of DMM.
Our findings demonstrate that H-CNF may be a good acidic
component. The H-CNF has stable Brønsted acidity that ef-
fectively catalyzes the hydrolysis of DMM. Accordingly, the
addition of H-CNF into CuZnAl formed complex catalysts
CuZnAl–H-CNF, which catalyzed the reforming of DMM with
high activity and selectivity to H2. Because DMM is a non-
poisonous and environmentally benign compound, it may be a
more appropriate fuel for portable and household H2 sources
for fuel cells.
Fig. 8. Selectivity to DME over unsupported Cu and ZnO/γ -Al O for the
2
3
reforming of DMM. Reaction conditions: N /H O/DMM = 24/5/1 (v/v) and
2
2
−1
3
−1
.
GHSV = 4.55 × 10 mlg
h
cat
2MOCH3 → MO + M + CH3OCH3,
MO + M + H2O → 2MOH,
and
(10)
(11)
Acknowledgments
MOCH3 + MOH → MO + M + CH3OH,
(12)
where M denotes the Cu surface. It has been established that
adsorbed oxygen species exist on the surface of copper when
H2O is present in the reaction system [46,48]. Both DMM and
methanol could be dissociatively adsorbed on Cu to produce
surface methoxy groups. Although the surface methoxy groups
can be combined to form DME as Eq. (10), this step might be
a minor reaction for the formation of DME, because surface
methoxy groups might be readily converted into H2 and CO2 on
CuZnAl [29]. In fact, the reforming of methanol produced little
DME despite the fact that the adsorption of methanol produced
large amounts of surface methoxy groups [29]. Therefore, there
must be pathways from DMM to DME over the Cu surface that
are not via two methoxy groups. Equation (8) might be such a
step in the formation of DME from DMM on Cu.
To confirm that DME was produced from DMM on Cu rather
than on ZnO/γ -Al2O3, we prepared an unsupported Cu sam-
ple (23 m2 g−1 before reduction) and a ZnO/γ -Al2O3 sample
(142 m2 g−1) for the reforming of DMM. Fig. 8 shows that
little DME was formed over the ZnO/γ -Al2O3, whereas the se-
lectivity to DME was as high as 18% over the unsupported Cu
at 513 K. These results clearly demonstrate the existence of a
pathway for the formation of DME from DMM on the Cu sur-
face.
This work was supported by the Ministry of Science and
Technology of China (Grants 2004DFB02900 and 2005CB22-
1403) and the National Science Foundation of China (Grant
20373023).
References
[1] A. Qi, S. Wang, G. Fu, C. Ni, D. Wu, Appl. Catal. A 281 (2005) 233.
[2] R.D. Cortright, R.R. Davda, J.A. Dumesic, Nature 418 (2002) 964.
[3] J.C. Amphlett, R.F. Mann, B.A. Peppley, P.R. Roberge, A. Rodrigues, J.P.
Salvador, J. Power Sources 71 (1998) 179.
[4] F. Panik, J. Power Sources 71 (1998) 36.
[5] C. Song, Catal. Today 77 (2002) 17.
[6] D.J. Moon, K. Sreekumar, S.D. Lee, B.G. Lee, H.S. Kim, Appl. Catal.
A 215 (2001) 1.
[7] C. Song, K.M. Reddy, Appl. Catal. A 176 (1999) 1.
[8] G.A. Deluga, J.R. Salge, L.D. Schmidt, X.E. Verykios, Science 303 (2004)
993.
[9] V.V. Galvita, G.L. Semin, V.D. Belyaev, T.M. Yurieva, V.A. Sobyanin,
Appl. Catal. A 216 (2001) 85.
[10] Y. Tanaka, R. Kikuchi, T. Takeguchi, K. Eguchi, Appl. Catal. B 57 (2005)
211.
[11] T.A. Semelsberger, K.C. Ott, R.L. Borup, H.L. Greene, Appl. Catal. A 309
(2006) 210.
[12] Y. Tanaka, T. Utaka, R. Kikuchi, K. Sasaki, K. Eguchi, Appl. Catal. A 238
(2003) 11.
4. Conclusion
[13] Q. Fu, W. Deng, H. Saltsburg, M. Flytzani-Stephanopoulos, Appl. Catal.
B 56 (2005) 57.
This work has demonstrated that DMM can be effectively
reformed to hydrogen on specially designed complex catalysts
with bifunctional character, comprising an acidic component
and a traditional CuZnAl catalyst. The basic chemistry for the
reforming of DMM involves the hydrolysis of DMM on acidic
sites to form methanol and formaldehyde, which are then re-
formed to produce H2 and CO2. The rate of H2 production
through the reforming of DMM over the complex catalysts
[14] N. Schumacher, A. Boisen, S. Dahl, A.A. Gokhale, S. Kandoi, L.C.
Grabow, J.A. Dumesic, M. Mavrikakis, I. Chorkendorff, J. Catal. 229
(2005) 265.
[15] T. Tabakova, F. Boccuzzi, M. Manzol, J.W. Sobczak, V. Idakiev, D. An-
dreeva, Appl. Catal. B 49 (2004) 73.
[16] M.M. Schubert, A. Venugopal, M.J. Kahlich, V. Plzak, R.J. Behm,
J. Catal. 222 (2004) 32.
[17] F. Mariño, C. Descorme, D. Duprez, Appl. Catal. B 54 (2004) 59.

110
Y. Fu, J. Shen / Journal of Catalysis 248 (2007) 101–110
[18] O. Pozdnyakova, D. Teschner, A. Wootsch, J. Kröhnert, B. Steinhauer,
H. Sauer, L. Toth, F.C. Jentoft, A. Knop-Gericke, Z. Paál, R. Schlögl,
J. Catal. 237 (2006) 1.
[19] W.B. Kim, T. Voitl, G.J. Rodriguez-Rivera, S.T. Evans, J.A. Dumesic,
Angew. Chem. Int. Ed. 44 (2005) 778.
[20] W.B. Kim, T. Voitl, G.J. Rodriguez-Rivera, J.A. Dumesic, Science 305
(2004) 1280.
[21] Lambiotte and Cie, online publications (http://www.lambiotte.com/
methylal.html).
[22] S. Satoh, Y. Tanigawa, US Patent 6 379 507, 2002.
[23] Y. Yuan, T. Shido, Y. Iwasawa, Chem. Commun. (2000) 1421.
[24] H. Liu, N. Bayat, E. Iglesia, Angew. Chem. Int. Ed. 42 (2003) 5072.
[25] K. Fuji, S. Nakano, E. Fujita, Synthesis (1975) 276.
[26] J. Masamoto, T. Iwaisako, M. Chohno, M. Kawamura, J. Ohtake, K. Mat-
suzaki, J. Appl. Polym. Sci. 50 (1993) 1299.
[34] G.L. Bezemer, P.B. Radstake, V. Koot, A.J. van Dillen, J.W. Geus, K.P. de
Jong, J. Catal. 237 (2006) 291.
[35] X. Dong, H.-B. Zhang , G.-D. Lin, Y.-Z. Yuan, K.R. Tsai, Catal. Lett. 85
(2003) 237.
[36] T.G. Ros, A.J. van Dillen, J.W. Geus, D.C. Koningsberger, Chem. Eur. J. 8
(2002) 1151.
[37] M.L. Toebes, J.M.P. van Heeswijk, J.H. Bitter, A.J. van Dillen, K.P.
de Jong, Carbon 42 (2004) 307.
[38] C. Hong, Z. Zhang, J. Cao, D. Qiu, Y. Li, Y. Fu, CN Patent 1 356 169,
2003.
[39] C. Hong, Z. Zhang, J. Cao, D. Qiu, Y. Li, Y. Fu, CN Patent 1 356 166,
2004.
[40] J. Shen, Q. Sun, H. Zhang, CN Patent 1 562 468, 2006.
[41] G.C. Chinchen, C.M. Hay, H.D. Vanderwell, K.C. Waugh, J. Catal. 103
(1987) 79.
[27] Q. Sun, A. Auroux, J. Shen, J. Catal. 244 (2006) 1.
[28] J.-P. Shen, C. Song, Catal. Today 77 (2002) 89.
[29] B.A. Peppley, J.C. Amphlett, L.M. Kearns, R.F. Mann, Appl. Catal. A 179
(1999) 31.
[30] T. Barzetti, E. Selli, D. Moscotti, L. Forni, J. Chem. Soc. Faraday Trans. 92
(1996) 1401.
[31] J. Shen, M.J. Lochhead, K.L. Bray, Y. Chen, J.A. Dumesic, J. Phys.
Chem. 99 (1995) 2384.
[32] A. Chambers, T. Nemes, N.M. Rodriguez, R.T.K. Baker, J. Phys. Chem.
B 102 (1998) 2251.
[42] M. Li, J. Shen, J. Catal. 205 (2002) 248.
[43] T. Takeguchi, K.-I. Yanagisawa, T. Inui, M. Inoue, Appl. Catal. A 192
(2000) 201.
[44] X.-M. Liu, G.Q. Lu, Z.-F. Yan, J. Beltramini, Ind. Eng. Chem. Res. 42
(2003) 6518.
[45] I.E. Wachs, R.J. Madix, J. Catal. 53 (1978) 208.
[46] I.A. Fisher, A.T. Bell, J. Catal. 184 (1999) 357.
[47] C.J. Jiang, D.L. Trimm, M.S. Wainwright, N.W. Cant, Appl. Catal. A 97
(1993) 145.
[48] G.C. Chinchen, M.S. Spencer, K.C. Waugh, D.A. Whan, J. Chem. Soc.
Faraday Trans. I 83 (1987) 2193.
[33] A. Dandekar, R.T.K. Baker, M.A. Vannice, J. Catal. 183 (1999) 131.