Effects of Al2O3 support modifications on MoO x and VOx catalysts for dimethyl ether oxidation to formaldehyde

Article abstract of DOI:10.1039/b302776g

Dispersed two-dimensional MoO_x and VO_x oligomers on Al₂O₃ and SnO_x-modified Al₂O ₃ supports were examined for selective dimethyl ether (DME) oxidation to HCHO and their structure and reduction rates in H₂ were determined using Raman and X-ray near edge absorption spectroscopies (XANES), respectively. Modifying Al₂O₃ supports with SnO _x or other reducible oxides (ZrO_x, CeO_x and FeO_x) led to MoO_x domains with higher rates for catalytic DME oxidation and for reduction in H₂, while maintaining the high HCHO selectivity observed on MoO_x/Al₂O₃ catalysts. This appears to reflect the higher reactivity of lattice oxygen atoms as Mo-O-M acquires more reducible M cations. On Al₂O ₃ modified with SnO_x species at near monolayer coverages (5.5 Sn nm^-2) DME oxidation turnover rates (per Mo-atom) were approximately three times greater than on unmodified Al₂O ₃ samples containing predominately polymolybdate domains (～7 Mo nm^-2). The rates of DME oxidation and of reduction by H₂ increased in parallel with increasing Sn surface density. HCHO selectivities decreased slightly with increasing Sn surface density, but they were significantly higher than on MoO_x domains supported on bulk crystalline SnO₂. The use of more reducible VO_x domains instead of MoO_x also led to higher DME oxidation rates (per V or Mo atom) without significant changes in HCHO selectivity and to effects of Al ₂O₃ modification by SnO_x similar to those observed on MoO_x-based catalysts. Al₂O₃ supports with higher surface area led to catalytic materials with similar rates per V or Mo atom and similar HCHO selectivities for a given surface density (～7 V or Mo nm^-2), because of the prevalence of accessible two-dimensional oligomeric domains of the active oxides on both Al ₂O₃ supports at these surface densities. Higher surface area Al₂O₃ supports, however, led to proportionately higher rates per catalyst mass, as a result of the larger number of active domains that can be accommodated at higher surface areas. These studies provide a rationale for the design of more efficient catalysts for selective DME oxidation to HCHO and illustrate the significant catalytic productivity improvements available from support modifications in oxidation catalysts.

Full text of DOI:10.1039/b302776g

View Article Online / Journal Homepage / Table of Contents for this issue
Effects of Al O support modifications on MoO and VO catalysts
2
3
x
x
for dimethyl ether oxidation to formaldehyde
Haichao Liu, Patricia Cheung and Enrique Iglesia*
Department of Chemical Engineering, University of California at Berkeley, Berkeley, CA 94720,
USA. E-mail: iglesia@cchem.berkeley.edu; Fax: (510) 642-4778; Tel: (510) 642-9673
Received 10th March 2003, Accepted 21st July 2003
First published as an Advance Article on the web 31st July 2003
Dispersed two-dimensional MoO
examined for selective dimethyl ether (DME) oxidation to HCHO and their structure and reduction rates in
were determined using Raman and X-ray near edge absorption spectroscopies (XANES), respectively.
Modifying Al supports with SnO or other reducible oxides (ZrO , CeO and FeO ) led to MoO domains
with higher rates for catalytic DME oxidation and for reduction in H , while maintaining the high HCHO
selectivity observed on MoO /Al catalysts. This appears to reflect the higher reactivity of lattice oxygen
atoms as Mo–O–M acquires more reducible M cations. On Al O modified with SnO species at near
x x 2 3 x 2 3
and VO oligomers on Al O and SnO -modified Al O supports were
H
2
2
O
3
x
x
x
x
x
2
x
2 3
O
2
3
x
À2
monolayer coverages (5.5 Sn nm ) DME oxidation turnover rates (per Mo-atom) were approximately three
times greater than on unmodified Al O samples containing predominately polymolybdate domains (ꢀ7 Mo
2
3
À2
2
nm ). The rates of DME oxidation and of reduction by H increased in parallel with increasing Sn surface
density. HCHO selectivities decreased slightly with increasing Sn surface density, but they were significantly
higher than on MoO domains supported on bulk crystalline SnO . The use of more reducible VO domains
instead of MoO also led to higher DME oxidation rates (per V or Mo atom) without significant changes in
x
2
x
x
HCHO selectivity and to effects of Al
catalysts. Al O supports with higher surface area led to catalytic materials with similar rates per V or Mo atom
2 3 x x
O modification by SnO similar to those observed on MoO -based
2
3
À2
and similar HCHO selectivities for a given surface density (ꢀ7 V or Mo nm ), because of the prevalence of
accessible two-dimensional oligomeric domains of the active oxides on both Al O supports at these surface
3
2
2 3
densities. Higher surface area Al O supports, however, led to proportionately higher rates per catalyst mass, as
a result of the larger number of active domains that can be accommodated at higher surface areas. These studies
provide a rationale for the design of more efficient catalysts for selective DME oxidation to HCHO and
illustrate the significant catalytic productivity improvements available from support modifications in oxidation
catalysts.
1
1,13
x 2
for both DME oxidation and MoO reduction with H .
1
. Introduction
Kinetic studies have probed the primary and secondary reac-
tions responsible for the observed HCHO selectivities during
DME oxidation, which depend also on the domain size and
on the nature of the underlying supports. Less reducible sup-
ports with higher Lewis acidity led to the weaker binding of
3
Formaldehyde (HCHO) is produced via methanol (CH OH)
oxidation and widely used as an intermediate in the synthe-
sis of many chemicals. Recent advances in dimethyl ether
1
(
DME an attractive precursor for HCHO and for other chemi-
DME, CH OCH ) synthesis from H /CO reactants make
3 3 2
6
+
DME-derived intermediates and of HCHO on Mo
which favor HCHO desorption and discourage HCHO read-
sorption and secondary reactions to form CO and methyl
sites,
2
cals currently produced from CH OH. The selective oxida-
–4
3
tion of dimethyl ether on catalysts based on dispersed MoO
and VO structures provides an alternate and selective route
x
x
1
1,13
formate.
These significant effects of active oxide reducibility and of
support identity on turnover rates and HCHO selectivities
x
5
for HCHO synthesis at low temperatures.
Recently, we have shown that small MoO domains consist-
x
ing predominately of two-dimensional polymolybdate struc-
tures supported on Al O , ZrO and SnO2 catalyze DME
oxidation with high primary HCHO selectivities (80–98%
led us to tailor the reduction properties of active MoO
domains by modifying Al O supports, which exhibited
x
2
3
2
2
3
low turnover rates and high HCHO selectivities, with a surface
coating of a more reducible oxide before anchoring MoOx
domains. We report here a detailed study of DME conversion
to HCHO on MoO and VO domains supported on Al O
5
HCHO, CH OH-free basis) and high reaction rates at much
3
6
–10
lower temperatures than previously reported.
domains supported on SnO showed the highest DME oxidation
MoO
x
2
x
x
2
3
turnover rates, but relatively low HCHO selectivities, while simi-
lar MoO domains supported on Al O gave very high HCHO
modified with SnO
x
, ZrO
x
x x
, CeO and FeO overlayers, and
also of the structure and reduction properties of the active
oxide domains. We also explore additional improvements in
catalytic reaction rates by coating Al O supports of higher
surface areas. In doing so, this study provides useful details
for the rational design of selective catalysts for the conversion
of DME to HCHO and also reports catalytic materials with
unprecedented reactivity and selectivity for this reaction.
x
2
3
5
,11
selectivities, but lower oxidation turnover rates.
DME oxidation to HCHO occurs via redox cycles invol-
2
3
1
2
ving lattice oxygen atoms. DME oxidation turnover rates
increase in parallel with increasing rates of the stoichiometric
x 2
reduction of MoO domains using H as the reductant; lar-
ger domains and more reducible supports led to higher rates
DOI: 10.1039/b302776g
Phys. Chem. Chem. Phys., 2003, 5, 3795–3800
3795
This journal is # The Owner Societies 2003

View Article Online
2
. Experimental
A Si (111) crystal monochromator was used and detuned by
0% in order to eliminate the harmonics. The spectra were
3
2
.1 Synthesis of catalytic materials
measured in transmission mode using 5 eV increments in the
pre-edge region (19.905–19.990 keV), 0.50 eV increments in
SnO -modified Al O supports (SnO –Al O and ZrO –
Al O ) were prepared by incipient wetness impregnation of
x
2
3
x 2 3 x
˚
À1
the near-edge region (19.990–20.033 keV) and 0.04 A in
the fine structure region (20.033–20.200 keV). Each XANES
spectrum consists of a single scan at the energy increments.
Energies were calibrated by placing the first inflection point
of a Mo foil held in the beam path at its reported absorption
energy (19.999 keV). Spectra were analyzed using WinXAS
2
3
2
(Degussa, AG, 101 m g , or Alcoa, HiQ31, 196 m
À1
2
Al
g
2
O
3
À1
; designated as Al O (A) and Al O (B), respectively) with
2 3 2 3
an isopropanol solution of Sn(i-C
metal basis) at 298 K. Impregnated samples were kept in dry
for 5 h at 298 K, and then dried at 393 K in ambient air
overnight and treated in flowing dry air (Airgas, zero grade)
at 673 K for 3 h. ZrO , CeO and FeO -modified Al sup-
ports (ZrO –Al O , CeO –Al O and FeO –Al O ) were
3 7 4
H O) (Alfa Aesar, 98%
N
2
1
5
(
Version 1.2). Background subtraction was carried out using
a linear fit of the pre-edge region and a cubic spline for the
x
x
x
2 3
O
4
+
post-edge region. The fraction of the Mo present as Mo
was determined from the 19.990–20.180 keV spectral region
using linear superimposition methods and the spectra of
MoO and of each unreduced catalyst sample.
x
2
3
x
2
3
x
2
3
prepared by incipient wetness impregnation of Al
2
O
3
(A)
) with aqueous solutions of
(Aldrich, 99.99%), Ce(NO (Aldrich, 99.99%)
O (Aldrich, 99.99%), respectively, at 298
1
6
2
À1
g
(
Degussa, AG, 101 m
ZrO(NO
and Fe(NO
2
3
)
2
3 4
)
3
)
3
Á9H
2
K for 5 h in ambient air. Impregnated samples were then dried
at 393 K in ambient air overnight and treated in flowing dry air
2
.3 Catalytic reactions of dimethyl ether
Dimethyl ether oxidation reaction rates and selectivities were
measured at 513 K in a packed-bed quartz flow reactor. Cata-
lyst samples (0.15–0.30 g) were diluted with acid-washed
quartz powder (ꢀ1 g) in order to prevent bed temperature gra-
(
Airgas, zero grade) at 673 K for 3 h. SnO
hydrolysis of an aqueous tin(IV) chloride pentahydrate (98%,
OH (14.8 M,
2
was prepared by
Alfa Aesar) solution at a pH of ꢀ7 using NH
4
Fisher Scientific). The precipitates were washed with deionized
water until the effluent was free of Cl ions, as detected by
3
À1
dients and treated in flowing 20% O –He (0.67 cm s ) for 1.5
2
h at 773 K before catalytic measurements. The reactant mix-
ture consisted of 80 kPa DME (99.5%, Praxair), 18 kPa O2
and 2 kPa N (2 kPa) (Praxair, Certified O –N mixture).
2 2 2
Homogeneous reactions were detected in empty reactors only
above 590 K.
The reactants and products in the effluent stream were ana-
lyzed by on-line gas chromatography (Hewlett-Packard 6890
GC) using a methyl silicone capillary column (HP-1; 30 m,
AgNO
dry air (Airgas, zero grade) at 773 K for 3 h.
Supported MoO catalysts were prepared by incipient wet-
ness impregnation of these supports with aqueous (NH ) -
3
addition. The resulting solids were treated in flowing
x
4
6
Mo
were similarly prepared using aqueous ammonium metavana-
date [NH VO ] (Aldrich, 99%) solutions containing oxalic acid
Mallinckrodt, analytical grade; NH VO –oxalic acid (0.5 M)).
7 x
O24 (Aldrich, 99%) solutions. Supported VO catalysts
4
3
(
4
3
0
(
.25 mm, 0.25 mm film) and a Porapak Q packed column
80–100 mesh, 1.82 m, 3.18 mm) connected to flame ionization
All samples were dried at 393 K in ambient air overnight after
impregnation and treated in flowing dry air (Airgas, zero
grade) at 773 K for 3 h. The Mo or V surface density for all
supported samples is reported as Mo nm or V nm , based
on the Mo or V content determined from the concentration of
the impregnating solution and the BET total surface area for
each sample.
and thermal conductivity detectors, respectively. Methanol,
formaldehyde (HCHO), methyl formate (MF), CO, CO2 ,
H O, and trace amount of dimethoxymethane (DMM) were
2
À2
À2
the only products detected.
Dimethyl ether conversions were varied by changing the
reactant space velocity and kept below 10% in all experiments.
DME reaction rates and product selectivities were extrapo-
lated to zero residence time in order to obtain the correspond-
ing primary rates and selectivities. In view of the available
2
.2 Structural characterization
Surface areas were measured using N at its normal boiling
2
3 3
pathways for DME–CH OH interconversion and for CH OH
point (Autosorb-1; Quantachrome) and BET analysis meth-
ods. Raman spectra were measured at 298 K in ambient air
using a HoloLab 5000 Raman spectrometer (Kaiser Optical)
and a frequency-doubled Nd:YAG laser at a wavelength of
conversion to HCHO, rates and selectivities are reported here
on a methanol-free basis.
11,13
5
32 nm. Samples were pressed into self-supporting thin wafers
3
. Results and discussion
and placed on a rotary stage within a quartz cell. The samples
were rotated at 16 Hz in order to avoid structural damage from
local laser heating.
11,13
Previous studies
cibility of active oxide domains on the rate and selectivity
of DME oxidation reactions led us to prepare and evaluate
on the effects of support and of the redu-
2 3
Al O supports modified with more reducible oxides, in an
2
.2 Reducibility of supported MoO
x
samples in H
2
attempt to increase reaction rates without the loss of HCHO
selectivity that previously accompanied the use of more redu-
cible bulk oxides as supports. Our initial approach involved
The rates of stoichiometric reduction of MoO
x
2
in H were
measured using in situ Mo-K edge X-ray absorption near-edge
spectroscopy (XANES). XANES spectra were measured using
beamline 4-1 at the Stanford Synchrotron Radiation Labora-
tory. The electron storage ring was operated at 3.0 GeV with
a beam current of 82 mA. Samples were diluted with Al O in
chemical modifications of Al
SnOx species; SnOx was chosen because it led to very high
activity and reducibility for MoO domains, albeit with signi-
ficant selectivity to undesired CO (CO + CO ) by-products in
2 3
O surfaces with well-dispersed
x
2
3
x
2
1
1
order to maintain a constant concentration of Mo absorbers (5
wt%) in all samples and then pressed and sieved to retain 0.18–
previous studies.
Fig. 1 shows primary DME conversion rates and primary
selectivities to HCHO, methyl formate (MF), and CO
x
0
capillary (1.0 mm OD, 0.1 mm wall thickness) held horizon-
.25 mm particles. These particles were placed in a quartz
(CO + CO ) on samples with Mo surface densities of ꢀ7 Mo
2
1
tally in a heated chamber. The sample was treated at 773
4
À2
nm as the SnO surface density varies from zero to 11.2
Sn nm on Al O (A). Al O (A) and SnO -modified Al O (A)
x
3
À1
À2
K for 1 h in flowing 20% O –He (ꢀ0.1 cm s ; Airgas, certi-
2
2
3
2
3
x
2
3
fied mixture), cooled to ambient temperature in He, and heated
;
Matheson UHP, certified mixture).
did not catalyze DME reactions at these conditions in the
absence of active MoO species. Primary HCHO selectivities
À1
3
À1
to 823 K at 0.167 K s in flowing 20% H –Ar (ꢀ0.1 cm s
2
x
remained nearly unchanged on MoO
x
/Al
2
O
3
(A) (ꢀ98%)
3
796
Phys. Chem. Chem. Phys., 2003, 5, 3795–3800

View Article Online
6
rates to measure the rate of incipient reduction of Mo spe-
+
1
1
cies to defect oxides. In-situ X-ray absorption spectroscopy
near the Mo-K edge (XANES), however, is an element-specific
technique that allows direct measurements of the extent of
6
+
x
reduction of MoO domains. The concentrations of Mo
4
+
and Mo were estimated by linear superimposition methods
using the starting spectrum and the spectrum for crystalline
2
MoO as principal components.
Fig. 2 shows the Mo-K near-edge spectra for MoO
nm ) supported on Al
x
(7.1 Mo
(A) modified with a monolayer of
/SnO –Al (A)) during contact
À2
2
O
3
À2
SnO
with 20% H
MoO at ambient conditions. The pre-edge feature weakened
2
(5.5 Sn nm ; MoO
x
2
2 3
O
2
–Ar at 298, 623, 723 and 823 K, and for crystalline
2
with increasing temperature and the spectral features
approached those typical of crystalline MoO . Above 623 K,
2
the spectra are accurately described by linear combinations
of the spectra for the initial sample (at 298 K) and for crystal-
line MoO , without any spectral contribution from crystalline
2
MoO or any other species.
3
4
+
Fig. 3 shows the fraction of the Mo atoms present as Mo
as the sample temperature increases in 20% H –Ar for MoO
Fig. 1 Primary DME conversion rates and primary selectivities to
HCHO, methyl formate (MF) and CO (CO + CO ) as a function of
Sn surface density on MoO /SnO –Al (A) catalysts at Mo surface
densities of ꢀ7.0 Mo nm (513 K; 80 kPa DME, 18 kPa O and 2
kPa N ).
2
x
À2
x
2
domains (ꢀ7.0 Mo nm ) on Al
2 3
O (A) modified with various
x
À2
x
2
O
3
4+
amounts of SnO . At each temperature, the Mo fraction
x
2
increased with increasing Sn surface density and it was highest
on bulk crystalline SnO . Thus, we conclude that the replace-
2
2
x 2 3
ment of Mo–O–Al prevalent in MoO /Al O with Mo–O–Sn
within this Sn surface density range. Primary MF (ꢀ2%) selec-
linkages leads to more reactive oxygen atoms and to more
reducible Mo cations, as also inferred from the effects of
SnO modification on catalytic DME oxidation rates (Fig. 1
tivities were very low and CO
samples.
x
was not detected on these
2
Raman and X-ray absorption spectra, as well as the
absence of X-ray diffraction lines (not shown here), showed
and Table 1). These data provide additional evidence for the
mechanistic connection between the catalytic and the redox
1
1,13
that MoO
polymolybdate domains on these samples. As a result, most
MoO species reside at surfaces and they are accessible for
catalytic DME reactions; therefore measured reaction rates
per Mo) become turnover rates and reflect the reactivity of
x
species exist predominately as two-dimensional
properties of oxide domains in DME oxidation reaction,
as also found for other reactions involving redox cycles, such
1
6,17
x
as oxidative dehydrogenation of alkanes
hols.
and of alco-
These findings are consistent with the involvement
1
8,19
(
of lattice oxygen atoms and with kinetically-relevant steps
involving C–H bond activation in adsorbed methoxide species,
which require transition states leading to electron donation to
metal centers during catalytic cycles leading to HCHO synth-
exposed surfaces on polymolybdate domains. These turnover
À1
rates increased from 4.6 to 12.8 mol (g-atom Mo h) as Sn
surface densities on Al O (A) increased from 0 to 11.2 Sn
2
3
À2
12
nm . These rates (per Mo) remained below those measured
on polymolybdate domains supported on bulk crystalline
esis from DME.
DME oxidation reactions occur via parallel and sequential
À1
13
steps shown in Scheme 1; these steps include primary DME
SnO
2
(32.6 mol (g-atom Mo h) , Fig. 1, Table 1). Thus, it
appears that the formation of more reducible MoOx species
as SnO increasingly covers Al supports leads in turn to
faster redox cycles during DME oxidation turnovers to
reactions to form CH OH (k ), HCHO (k ), methyl formate
3
0
1
x
2
O
3
(MF) (k
mary HCHO products to form MF (k ) and COx (k5).
2 x 3
) and CO (k ), as well as secondary reactions of pri-
1
3
4
1
2
HCHO. The formation of Mo–O–Sn linkages between dis-
persed MoO domains and SnO –Al O (A) or SnO surfaces
HCHO selectivities depend on two rate constant ratios, k
1
/
x
x
2
3
2
(k + k ) and k /((k + k )C ), where k is the reaction rate
Ao
2
3
1
4
5
i
1
3
leads to an apparent increase in electron density at lattice
oxygen atoms compared with that in Mo–O–Al structures
constant for reaction i and CAo the inlet DME concentration.
Values of k /(k + k ) reflect primary DME conversion rates
1
2
3
prevalent on pure Al
2
O
firmed by the parallel increase in the rate of stoichiometric
3
surfaces. This proposal was con-
to HCHO (k ) relative to those for MF (k ) and CO (k3)
1 2 x
formation; thus, they provide an alternate measure of
primary HCHO selectivities. In contrast, k /((k + k )CAo)
reduction of MoO
Sn surface density.
x
domains in H
2
observed with increasing
1
4
5
1
ratios reflect primary HCHO formation rates (k ) relative to
In SnO
x
MoO and SnO
x
-containing samples, some concurrent reduction of
species prevents the use of H consumption
the rates for secondary HCHO reactions to form MF (k4)
and CO (k ); this kinetic parameter reflects the relative
x
2
x
5
Table 1 Primary DME conversion rates, selectivities and relative rate constants for MoO
and on Al
x
domains supported on unmodified Al
(513 K; 80.0 kPa DME, 18 kPa O and 2 kPa N )
2 3 2
O (A) and SnO
2
O
3
(A) modified with near one monolayer SnO , ZrO
x
x
, CeO
x
and FeO
x
2
2
Mo surface
density/
Primary DME
reaction rate/
mol (g-atom Mo h)
Primary selectivity (%)
À2
À1
Catalyst (MoO
x
wt%)
Mo nm
HCHO
MF
CO
x
k
1
/(k
2
+ k
3
)
k
1
/((k
4
+ k
5
)CAo
)
MoO
MoO
MoO
MoO
MoO
MoO
x
x
x
x
x
x
/Al
2
O
3
(A) (15.0%)
(5.9%)
7.0
6.3
7.1
6.8
6.6
6.9
4.6
32.6
12.2
8.7
98.1
70.4
97.7
98.6
98.8
99.7
1.9
12.5
2.3
0
17.2
45
2.4
0.35
0.08
0.31
0.27
0.29
0.29
/SnO
/SnO
2
x
–Al
–Al
–Al
–Al
2
O
3
(A) (15.0%)
(A) (15.0%)
0
0
0
0
43
70
/ZrO
/CeO
/FeO
x
2
O
3
1.4
1.2
x
2
O
3
(A) (13.4%)
(A) (14.9%)
6.8
6.2
82
332
x
2
O
3
0.3
Phys. Chem. Chem. Phys., 2003, 5, 3795–3800
3797

View Article Online
Scheme 1 Primary and secondary reaction pathways for dimethyl
ether conversion on MoO -based catalysts.
x
losses that prevail as SnO
surface densities. Chemical modifications of Al
2 3
monolayer coverages of ZrO , CeO and Fe O
2
crystallites form at higher Sn
with near
also led to
2 3
O
2
2
higher DME conversion rates compared with MoO
(
mary or secondary HCHO selectivities (see k /((k + k )CAo
1 4 5
x
/Al
2
O
3
À2
ꢀ7 Mo nm ) (Table 1), without significant changes in pri-
x
Fig. 2 X-Ray absorption spectra near the Mo K edge for MoO /
À2
À2
SnO
20% H
at ambient temperature.
2
–Al
2
O
3
(A) (7.1 Mo nm ; 5.5 Sn nm ) after treatment in H
2
values in Table 1).
(
2
–Ar) at 298, 623, 723 and 823 K and for crystalline MoO
2
Clearly, the catalytic function of polymolybdate domains
depends on the chemical identity and the reduction properties
of the support surfaces to which they are atomically connected.
These effects appear to be reasonably general and to apply also
tendency of a given catalyst to form and convert HCHO.
Higher values of k /(k + k ) and k /((k + k )CAo) lead to
1
2
3
1
4
5
to oxidation reactions of DME on VO domains. Polyvana-
x
higher HCHO selectivities at a given DME conversion and
the latter becomes increasingly important as DME conversion
increases in determining HCHO yields.
date domains are more reducible than polymolydbate domains
on a given support surface, and modifications of Al O sup-
x
ports with SnO (5.5 Sn nm ) lead to higher DME oxidation
1
6
2
3
À2
Fig. 4 shows k
1
/(k
2
+ k
3
) and k
1
/((k
4
+ k
a function of Sn surface density. The k /(k + k ) ratios
5
)CAo) values as
rates (per active metal atom). Table 2 shows that DME oxida-
tion rates (per active metal atom) on VO /Al O (A)and VO /
1
2
3
x
2
3
x
remained nearly constant as Sn surface density increased from
zero to 11.2 Sn nm , consistent with the previously discus-
SnO –Al O (A) were 1.5 and 1.4 times greater than for MoO
x 2 3 x
À2
domains on each respective support. Primary HCHO selectiv-
ities and k /(k + k ) ratios, as well as k /((k + k )CAo) ratios
sed constant primary HCHO selectivites (Fig. 1). The k
(k + k )CAo) ratios, however, initially remained similar to
those on pure Al
1
/
1
2
3
1
4
5
(
4
5
on these VOx catalysts were similar to those measured on
MoO domains for a given support. As for MoO domains,
2
O
3
supports, but then decreased for Sn sur-
x
x
À2
face densities greater than 5.5 Sn nm . These Sn surface
density effects indicate that near-monolayer SnO
2 3 x
chemical modifications of Al O (A) by a monolayer of SnO
x
coverages
(110) planes) on Al O provide a
led to an increase in the DME conversion rate by a factor of
2.4 on VO samples. For WO domains, which are less reduci-
À2
(
ꢀ5.0 Sn nm , for SnO
2 2 3
x
x
1
7
compromise between the higher reactivity of MoO
x
domains
supported on SnO -modified Al and the HCHO yield
ble than MoOx and VOx domains, DME conversion pro-
ducts were not detected on either WO /Al O (A) (7.6 W
nm ) or WO /SnO –Al O (A) (7.8 W nm ) at reaction
x
2 3
O
x
2
À2
3
À2
x
x
2
3
temperatures between 513 and 553 K.
Fig. 3 MoO
MoO and initial spectra as a function of treatment temperature in
0% H –Ar for MoO domains supported on SnO -modified Al (A)
2
fraction measured from linear superimposition of
2
2
2
x
x
2
O
3
Fig. 4 Rate constant ratios k
function of Sn surface density on MoO
1
/(k
2
+ k
3
) and k
x
/SnO –Al
1
/((k
4
+ k
(A) catalysts at
5
)CAo) as a
À2
with Sn surface densities of 2.8 (˘), 5.5 (S) and 11.2 (N) Sn nm , and
x
À2
2
O
3
on unmodified Al
density (6.3–7.1 Mo nm ).
2
O
3
(A) (/) and SnO
2
(L) at similar Mo surface
Mo surface densities of ꢀ7.0 Mo nm (513 K; 80 kPa DME, 18
kPa O and 2 kPa N ).
À2
2
2
3
798
Phys. Chem. Chem. Phys., 2003, 5, 3795–3800

View Article Online
Table 2 Primary DME conversion rates, selectivities and relative rate constants on MoO
À2
x
, VO
x
and WO
x
domains at near one monolayer surface
density supported on unmodified and SnO -modified Al
x
O
2 3
(A) (5.5 Sn nm ) (513 K; 80.0 kPa DME, 18 kPa O
2
and 2 kPa N
2
)
Primary DME
M surface density/ reaction rate/
Primary selectivity (%)
À2
À1
Catalyst (MO
x
wt%)
Metal nm
mol (g-atom Mo h)
HCHO
MF
CO
x
k
1
/(k
2
+ k
3
)
1 4 5
k /((k + k )CAo)
MoO
VO /Al
WO /Al
MoO /SnO
VO /SnO –Al
WO /SnO –Al
x
/Al
2
O
3
(A) (15.0%)
(A) (10.0%)
(A) (18.0%)
7.0
8.0
7.6
4.6
6.8
0
98.1
99.5
–
1.9
0.5
–
0
0
–
0
0
–
45
199
–
0.35
0.29
–
x
2 3
O
x
2
O
3
x
x
2
–Al O
3
(A) (15.0%) 7.1
12.2
16.3
0
97.7
97.7
–
2.3
2.3
–
43
43
–
0.31
0.27
–
x
x
2
O
3
(A) (10.1%)
7.9
7.8
x
x
2 3
O (A) (16.8%)
Next, we explored whether practical rates per catalyst mass
or volume) can be improved by using modified Al O sup-
lar Raman features, and two bands were detected at ꢀ965
À1 À1
(
cm and ꢀ912 cm , showing the two-dimensional MoOx
oligomers present on the support surfaces at the similar Mo
surface densities.
2
3
ports with higher surface area and anchoring well-dispersed
MoO or VO domains at surface densities corresponding to
near monolayer coverages. In effect, we attempt here to simply
x
x
Table 3 shows primary DME reaction rates and HCHO
selectivities, as well as the two kinetic parameters responsible
for selectivities [k /(k + k ) and k /((k + k )CAo)] at 513 K
1 2 3 1 4 5
increase the number of active sites, while maintaining the sup-
À2
port composition (SnO
x
–Al
2
O
3
; 5.5 Sn nm ) and the active
À2
À2
oxide surface densities (ꢀ7 Mo nm or ꢀ8 V nm ) that
led to highest site reactivities on modified alumina supports
with lower surface area. We examined the structure and cata-
on these four MoO catalyst samples. On MoO domains sup-
x
x
ported on pure Al
2
O
3
, DME reaction rates per gram of cata-
lyst increased from 4.7 to 9.4 mmol (g-cat h) as the sample
À1
2
À1
lytic properties of four samples: MoO
x
and VO
x
supported on
Al O (B) and SnO –Al O (B), both of which have higher
surface area increased from 90.0 to 174.9 m g , indicating
a proportional increase in catalyst productivity with increasing
surface area. Reaction rates per Mo-atom (4.7 vs. 5.1 mmol (g-
2
3
x
2
3
2
À1
, respectively) than the
surface areas (196 and 181 m
corresponding pure and SnO -modified Al
discussed above (Tables 3 and 4).
g
À1
x
2
O
3
(A) materials
atom Mo h) ), primary HCHO selectivities (98.1 vs. 96.0%)
and k
0.31) ratios were very similar on these two samples; these
results confirm that the dispersion and structure of MoO
are unaffected by the surface area of the Al support, as long
as Mo surface densities are kept at similar values. Similarly,
1 2 3 1 4 5
/(k + k ) (45 vs. 31) and k /((k + k )CAo) (0.35 vs.
À1
Fig. 5 shows Raman spectra in the 500–1100 cm range for
–Al
MoO
and MoO
sities (6.4–7.1 Mo nm ). MoO
x
/Al
2
O
3
(B), MoO
x
/Al
2
O
3
(A), MoO
x
/SnO
x
2
O
3
(B)
x
x
/SnO –Al (A) samples with similar surface den-
x
2
O
3
2 3
O
À2
À2
)
x
/Al
2
O
3
(B) (6.4 Mo nm
À1
showed a broad band at ꢀ965 cm and a shoulder at ꢀ912
DME conversion rates per gram of catalyst on MoO
–Al
(7.1 Mo nm ) were nearly proportional to their surface area
and no significant changes in primary or secondary HCHO
selectivities were detected (Table 3).
x
/
(A)
À1
À2
cm , which are assigned to terminal Mo=O and bridging
Mo–O–Mo stretching modes in two-dimensional oligomeric
SnO
2
O
(B) (6.4 Mo nm ) and MoO
/SnO –Al
x
O
3
x
3
x
2
À2
1
6,20
MoO
x
, respectively.
This sample also showed three sharper
bands at 674, 825 and 1003 cm , characteristic of crystalline
À1
2
MoO3 . The strong MoO bands in this sample suggest the
0
VO -based catalysts showed similar trends. Table 4 shows
x
3
presence of some crystalline MoO
3
, but Raman scattering
cross-sections are 10–10 times greater for MoO3 than for
that DME reaction rates increased with increasing support sur-
face area for VO species supported at similar surface densities
3
x
2
1,22
À2
dispersed MoO
x
species.
Thus, we conclude from the
on Al
2
O
3
[VO
x
/Al
2
O
3
(B) (7.8 V nm ) and VO
x
/Al
2
O
3
(A)
(8.0 V nm )] and SnO –Al O [VO /SnO –Al O (B) (7.5 V
À1
À2
observed relative intensities for the 965 and the 1003 cm
bands that this sample contains predominately two-dimen-
sional MoOx oligomers. This conclusion was confirmed by
the absence of MoO X-ray diffraction lines and by the lack
3
of MoO features in the X-ray absorption fine structure spec-
3
x
2
3
x
x
2
3
À2
À2
nm ) and VO
x
/SnO
x
–Al
2
O
3
(A) (7.9 V nm )] supports. Site
reactivities (DME oxidation rates per V-atom) and HCHO
selectivity parameters were unaffected by changes in the sur-
face area of the support; we note also that the site reactivity
tra for this sample. These Raman features are similar to those
observed in MoO /Al O (A) (7.0 Mo nm ), which has a
enhancements introduced by SnO
on the higher surface area Al O supports. Thus, catalyst
2
x
overlayers are maintained
À2
x
2
3
3
lower surface area, but similar Mo surface densities as
MoO /Al O (B) (6.4 Mo nm ); thus, MoOx domains are
productivities can be significantly improved by exploiting
the beneficial effects of SnO overlayers and of well-dispersed
À2
x
2
3
x
structurally similar on the two Al
surface densities are kept relatively constant. The two SnO -
2
O
3
supports, as long as
polymolybdate domains on alumina supports with higher
surface areas, while maintaining near monolayer surface densi-
x
À2
modified samples, MoO
and MoO /SnO –Al O (A) (7.1 Mo nm ), also showed simi-
x
x
/SnO –Al
2
O
3
(B) (6.4 Mo nm
)
x x x
ties for both the SnO modifier and the active MoO or VO
domains.
À2
x
x
2
3
x
Table 3 Surface area effects on primary DME conversion rates, HCHO selectivities and relative rate constants for MoO domains at near one
monolayer surface density supported on unmodified and SnO
x
-modified Al
2
O
3
(ꢀ5.5 Sn nmÀ2) (513 K; 80.0 kPa DME, 18 kPa O
2
and 2 kPa N
2
)
BET surface Mo surface Primary DME
2
Primary DME Primary
reaction rate/mol HCHO
area/m
g-cat
density/
Mo nm
reaction rate/
mmol (g-cat h)
À2
À1
À1
Support (MoO
x
wt%)
(g-atom Mo h)
selectivity (%)
k
1
/(k
2
+ k
3
)
1 4 5
k /((k + k )CAo
)
Al
Al
2
O
O
3
(A) (15.0%)
(B) (26.8%)
90.0
174.7
87.9
7.0
6.4
7.1
6.4
4.7
9.4
4.6
5.1
98.1
95.9
97.7
98.2
45
23
41
55
0.35
0.32
0.31
0.41
2
3
SnO
SnO
2
–Al
–Al
2
O
O
3
(A) (15.0%)
12.6
18.4
12.1
11.4
2
2
3
(B) (22.9%) 150.3
Phys. Chem. Chem. Phys., 2003, 5, 3795–3800
3799

View Article Online
Table 4 Surface area effects on primary DME conversion rates, HCHO selectivities and relative rate constants for VO
x
domains at near one
)
monolayer surface density supported on unmodified and SnO -modified Al
x
2
O
3
(ꢀ5.5 Sn nmÀ2) (513 K; 80.0 kPa DME, 18 kPa O
2
and 2 kPa N
2
BET surface V surface Primary DME
2
Primary DME
reaction rate/
mol (g-atom V h)
Primary
HCHO
selectivity (%)
area/m
(g-cat)
density/
À2
reaction rate/
mmol (g-cat h)
À1
À1
À1
Support (V
O
2 5
wt%)
V nm
k
1
/(k
2
+ k
3
)
1 4 5
k /((k + k )CAo
)
Al
Al
2
O
O
3
(A) (10.0%)
(B) (23.2%)
83.0
195.9
84.3
8.0
7.8
7.9
7.5
6.7
13.5
16.2
25.3
6.8
5.9
99.5
97.1
97.7
98.0
199
34
0.29
0.30
0.27
0.28
2
3
SnO
SnO
2
–Al
–Al
2
O
O
3
(A) (10.1%)
16.3
15.3
43
52
2
2
3
(B) (16.8%) 149.2
4
. Conclusions
domains of the active oxides on both Al
surface densities. Higher surface area Al O supports lead to
2 3
O supports at these
2
3
Improvements of the reactivity of the supported MoO
domains, without significant loss of the selective properties
of the Al -supported MoO catalyst for DME conversion
to HCHO, were achieved by surface modification of Al
supports with reducible SnO , ZrO , CeO and FeO , as a
result of the increase in the reducibility of the MoO domains
on the modified Al supports. SnO is the best modifier
among these oxides. The reducibility and reactivity of the sup-
ported MoO domains on SnO –Al supports increase in
parallel with increasing the Sn surface density, and a mono-
layer coverage of SnO species provides the best compromise
x
proportionately higher rates per catalyst mass, as a result of
the larger number of active domains that can be accommo-
dated at higher surface areas.
2
O
3
x
2
O
3
x
x
x
x
x
Acknowledgements
2
O
3
x
This study was supported by BP as part of the Methane Con-
version Cooperative Research Program at the University of
California at Berkeley. The authors also acknowledge helpful
technical discussions with Drs Theo Fleisch and John Collins
of BP. The X-ray absorption data were collected at Stanford
Synchrotron Radiation Laboratory (SSRL), which is operated
by Department of Energy (DOE), Office of Basic Energy
Sciences under contract DE-AC03-76SF00515.
x
x
2 3
O
x
between the reactivity and the selectivity properties of the sup-
domains. At 5.5 Sn nm , near one monolayer
À2
ported MoO
x
coverage, the DME oxidation rate (per Mo atom) is about
three times greater for MoO /SnO –Al than the rate for
unmodified MoO /Al O at similar Mo surface densities of
x
x
2 3
O
x
2
3
À2
ꢀ
7.0 Mo nm . Replacing MoO species by more reducible
x
VO species also leads to higher DME oxidation rates (per
x
References
Mo or V atom) without significant changes in HCHO selecti-
vity (as reflected by the k /(k + k ) and k /((k + k )CAo)
1
2
3
1
4
5
1
2
3
J. M. Tatibouet, Appl. Catal., A, 1997, 148, 213.
ratios) and to effects of Al
to those observed on MoO -based catalysts. Al O supports
2
O
3
modification by SnO
x
similar
J.-L. Li, X.-G. Zhang and T. Inui, Appl. Catal. A, 1996, 147, 23.
T. H. Fleisch, A. Basu, M. J. Gradassi and J. G. Masin, Stud.
Surf. Sci. Catal., 1997, 107, 117.
x
2
3
with higher surface area led to catalytic materials with similar
rates per V or Mo atom and similar HCHO selectivities for a
4
T. Shikada, Y. Ohno, T. Ogawa, M. Ono, M. Mizuguchi, K.
Tomura and K. Fujimoto, Stud. Surf. Sci. Catal., 1998, 119, 515.
H. Liu and E. Iglesia, J. Catal., 2002, 208, 1.
G. P. Hagen and M. J. Spangler, US Pat., 6,265,528, 2001.
G. P. Hagen and M. J. Spangler, US Pat., 6,350,919, 2002.
R. M. Lewis, R. C. Ryan and L. H. Slaugh, US Pat. 4,442,307,
À2
given surface density (ꢀ7 V or Mo nm ), because of the pre-
5
6
7
8
valent presence of accessible two-dimensional oligomeric
1
984.
R. M. Lewis, R. C. Ryan and L. H. Slaugh, US Pat., 4,439,624,
984.
R. M. Lewis, R. C. Ryan and L. H. Slaugh, US Pat., 4,435,602,
984.
9
0
1
1
1
1
1
1
1
2
3
H. Liu, P. Cheung and E. Iglesia, J. Catal., 2003, 217, 222.
H. Liu, P. Cheung and E. Iglesia, unpublished results.
H. Liu, P. Cheung and E. Iglesia, J. Phys. Chem. B, 2003, 107,
4
118.
W. Li, G. D. Meitzner, R. W. Borry III and E. Iglesia, J. Catal.,
000, 191, 373.
1
4
2
1
1
5
6
T. Ressler, WinXAS 97, version 1.2, 1998.
K. Chen, S. Xie, A. T. Bell and E. Iglesia, J. Catal., 2001,
1
98, 232.
1
1
1
7
8
9
K. Chen, A. T. Bell and E. Iglesia, J. Catal., 2002, 209, 35.
G. Deo and I. E. Wachs, J. Catal., 1994, 146, 323.
W. Zhang, A. Desikan and S. T. Oyama, J. Phys. Chem., 1995, 99,
1
4 468.
G. Mestl and T. K. K. Srinivasan, Catal. Rev.-Sci. Eng., 1998,
8, 451.
R. Radhakrishnan, C. Reed, S. T. Oyama, M. Semen, J. N.
Kondo, K. Domen, Y. Ohminami and K. Asakura, J. Phys.
Chem. B, 2001, 105, 8519.
2
0
1
3
2
Fig. 5 Raman spectra for (a) MoO
c) MoO /SnO –Al (B) and (d) MoO
with surface densities of 6.4–7.1 Mo nm at ambient conditions.
x
/Al
2
O
3
(B), (b) MoO
x 2 3
/Al O (A),
(
x
x
2
O
3
x
/SnO –Al (A) catalysts
x
2
O
3
22 C. C. Williams, J. G. Ekerdt, J. M. Jehng, F. D. Hardcastle and
I. E. Wachs, J. Phys. Chem., 1991, 95, 9791.
À2
3
800
Phys. Chem. Chem. Phys., 2003, 5, 3795–3800