Structural requirements and reaction pathways in dimethyl ether combustion catalyzed by supported Pt clusters

Article abstract of DOI:10.1021/ja073712z

The identity and reversibility of the elementary steps required for catalytic combustion of dimethyl ether (DME) on Pt clusters were determined by combining isotopic and kinetic analyses with density functional theory estimates of reaction energies and activation barriers to probe the lowest energy paths. Reaction rates are limited by C-H bond activation in DME molecules adsorbed on surfaces of Pt clusters containing chemisorbed oxygen atoms at near-saturation coverages. Reaction energies and activation barriers for C-H bond activation in DME to form methoxymethyl and hydroxyl surface intermediates show that this step is more favorable than the activation of C-O bonds to form two methoxides, consistent with measured rates and kinetic isotope effects. This kinetic preference is driven by the greater stability of the CH₃OCH ₂* and OH* intermediates relative to chemisorbed methoxides. Experimental activation barriers on Pt clusters agree with density functional theory (DFT)-derived barriers on oxygen-covered Pt(111). Measured DME turnover rates increased with increasing DME pressure, but decreased as the O₂ pressure increased, because vacancies (*) on Pt surfaces nearly saturated with chemisorbed oxygen are required for DME chemisorption. DFT calculations show that although these surface vacancies are required, higher oxygen coverages lead to lower C-H activation barriers, because the basicity of oxygen adatoms increases with coverage and they become more effective in hydrogen abstraction from DME. Water inhibits reaction rates via quasi-equilibrated adsorption on vacancy sites, consistent with DFT results indicating that water binds more strongly than DME on vacancies. These conclusions are consistent with the measured kinetic response of combustion rates to DME, O₂, and H₂O, with H/D kinetic isotope effects, and with the absence of isotopic scrambling in reactants containing isotopic mixtures of ¹⁸O₂-¹⁶O₂ or ¹²CH₃O¹²CH₃-¹³CH ₃O¹³-CH₃. Turnover rates increased with Pt cluster size, because small clusters, with more coordinatively unsaturated surface atoms, bind oxygen atoms more strongly than larger clusters and exhibit lower steady-state vacancy concentrations and a consequently smaller number of adsorbed DME intermediates involved in kinetically relevant steps. These effects of cluster size and metal-oxygen bond energies on reactivity are ubiquitous in oxidation reactions requiring vacancies on surfaces nearly saturated with intermediates derived from O₂.

Full text of DOI:10.1021/ja073712z

Published on Web 10/04/2007
Structural Requirements and Reaction Pathways in Dimethyl
Ether Combustion Catalyzed by Supported Pt Clusters
Akio Ishikawa, Matthew Neurock,* and Enrique Iglesia*,†
†
,‡
Contribution from the Department of Chemical Engineering, UniVersity of California, Berkeley,
California 94720, and Departments of Chemical Engineering and of Chemistry,
UniVersity of Virginia, CharlottesVille, Virginia 22904
Received June 8, 2007; E-mail: iglesia@berkeley.edu (E.I.); mn4n@virginia.edu (M.N.)
Abstract: The identity and reversibility of the elementary steps required for catalytic combustion of dimethyl
ether (DME) on Pt clusters were determined by combining isotopic and kinetic analyses with density
functional theory estimates of reaction energies and activation barriers to probe the lowest energy paths.
Reaction rates are limited by C-H bond activation in DME molecules adsorbed on surfaces of Pt clusters
containing chemisorbed oxygen atoms at near-saturation coverages. Reaction energies and activation
barriers for C-H bond activation in DME to form methoxymethyl and hydroxyl surface intermediates show
that this step is more favorable than the activation of C-O bonds to form two methoxides, consistent with
measured rates and kinetic isotope effects. This kinetic preference is driven by the greater stability of the
CH
on Pt clusters agree with density functional theory (DFT)-derived barriers on oxygen-covered Pt(111).
Measured DME turnover rates increased with increasing DME pressure, but decreased as the O pressure
3 2
OCH * and OH* intermediates relative to chemisorbed methoxides. Experimental activation barriers
2
increased, because vacancies (*) on Pt surfaces nearly saturated with chemisorbed oxygen are required
for DME chemisorption. DFT calculations show that although these surface vacancies are required, higher
oxygen coverages lead to lower C-H activation barriers, because the basicity of oxygen adatoms increases
with coverage and they become more effective in hydrogen abstraction from DME. Water inhibits reaction
rates via quasi-equilibrated adsorption on vacancy sites, consistent with DFT results indicating that water
binds more strongly than DME on vacancies. These conclusions are consistent with the measured kinetic
response of combustion rates to DME, O
of isotopic scrambling in reactants containing isotopic mixtures of
CH . Turnover rates increased with Pt cluster size, because small clusters, with more coordinatively
2 2
, and H O, with H/D kinetic isotope effects, and with the absence
18
16
12
12
13
13
O
2
- O
2
or CH
3
O CH
3
3
- CH O -
3
unsaturated surface atoms, bind oxygen atoms more strongly than larger clusters and exhibit lower steady-
state vacancy concentrations and a consequently smaller number of adsorbed DME intermediates involved
in kinetically relevant steps. These effects of cluster size and metal-oxygen bond energies on reactivity
are ubiquitous in oxidation reactions requiring vacancies on surfaces nearly saturated with intermediates
2
derived from O .
1
. Introduction
its homogeneous combustion pathways via radical intermedi-
8
,9
ates. Here, we explore the catalytic combustion of DME
because of its potential use in power generation and radiant
heating devices and its ability to combust with low emissions
of NOx, CO, and organic volatile compounds.
The first study of catalytic DME combustion reported that
the specific reactivity of noble metals (per exposed metal surface
atom) increased in the sequence Ru > Pt > Ir > Pd > Rh,10
but such reactivity trends were measured on clusters of different
dispersions [Ru (0.06), Pt (0.41), Ir (0.76), Pd (0.23), and Rh
(0.46)]. As a result, the trends may reflect the ubiquitous effects
of cluster size on surface reactivity, as reported for many
Dimethyl ether (DME) can be produced via dehydration of
1-3
methanol or via direct conversion of synthesis gas.
DME is
2
relatively inert, noncorrosive, and nontoxic and resembles
liquefied petroleum gas (LPG) in its physical properties. As a
result, it has emerged as a convenient energy carrier for
distributed power generation, space heating, and diesel replace-
ment in developing economies. Recent studies have addressed
4
-7
steam reforming of DME on supported metal catalysts
and
†
‡
University of California.
University of Virginia.
(
(
1) Li, J. -L.; Zhang, X. -G.; Inui, T. Appl. Catal., A 1996, 147, 23.
2) Fleisch, T. H.; Basu, A.; Gradassi, M. J.; Masin, J. G. Stud. Surf. Sci. Catal.
997, 107, 117.
1
(7) Faungnawakij, K.; Tanaka, Y.; Shimoda, N.; Fukunaga, T.; Kawashima,
S.; Kikuchi, R.; Eguchi, K. Appl. Catal., A 2006, 304, 40.
(8) Curran, H. J.; Fisher, S. L.; Dryer, F. L. Int. J. Chem. Kinet. 2000, 32,
741.
(9) Curran, H. J.; Pitz, W. J.; Westbrook, C. K.; Dagaut, P.; Boettner, J. -C.;
Cathonnet, M. Int. J. Chem. Kinet. 1997, 30, 229.
(10) Solymosi, F.; Cser e´ nyi, J.; Ov a´ ri, L. Catal. Lett. 1997, 44, 89.
(
(
3) Fleisch, T. H.; Sills, R. A. Stud. Surf. Sci. Catal. 2004, 147, 31.
4) Galvita, V. V.; Semin, G. L.; Belyaev, V. D.; Yurieva, T. M.; Sobyanin,
V. A. Appl. Catal., A 2001, 216, 85.
(
(
5) Takeishi, K.; Suzuki, H. Appl. Catal., A 2004, 260, 111.
6) Nishiguchi, T.; Oka, K.; Matsumoto, T.; Kanai, H.; Utani, K.; Imamura,
S. Appl. Catal., A 2006, 301, 66.
10.1021/ja073712z CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 13201-13212
9
13201

A R T I C L E S
Ishikawa et al.
catalytic reactions,11-14 instead of intrinsic reactivity differences
among these metals. In methane combustion, for example, the
requirement for oxygen atom and vacancy site (O*-*) pairs
and the effects of cluster size on oxygen binding energy lead
to marked effects of cluster size on turnover rates.1
Table 1. Pt Dispersions, Average Crystallite Diameters, and DME
Combustion Turnover Rates (Extrapolated to Zero Residence
Time and Conversion)a
Pt
crystallite
diamb (nm)
DME turnover
ratec
catalyst
dispersion
5-18
0
0
.8 wt % Pt/Al2O3
.9 wt % Pt/Al2O3
0.99
0.75
0.50
0.52
0.44
0.42
0.36
0.26
1.2
1.6
2.3
2.2
2.5
2.7
3.2
4.4
0.11
0.20
0.40
0.33
0.35
0.55
0.63
1.18
We provide here kinetic, isotopic, and theoretical evidence
for the identity and kinetic relevance of the elementary steps
involved in DME combustion on Pt clusters over a wide range
of dispersions (0.26-0.99; ∼1-4 nm diameter) supported on
γ-Al2O3 and ZrO2. These details are essential for a rigorous
description of intrinsic reactivity and for the practical imple-
mentation of catalytic DME combustion within complex hy-
drodynamic environments that exhibit ubiquitous temperature
and concentration gradients. This study provides evidence for
the relevance of C-H bond activation steps in molecularly
adsorbed DME and for the role of vacancies within chemisorbed
oxygen layers in the catalysis of such steps. Turnover rates
decreased with increasing Pt dispersion because of the con-
comitant increase in the coordinative unsaturation of exposed
metal atoms and in the binding energy of chemisorbed oxygen
atoms with decreasing cluster size, which led in turn to a smaller
number of vacancies on nearly saturated surfaces.
2.3 wt % Pt/Al O3
2
0
1
1
1
.9 wt % Pt/Al2O3
.1 wt % Pt/ZrO2
.0 wt % Pt/ZrO2
.0 wt % Pt/ZrO2
1.0 wt % Pt/ZrO2
a
Conditions: 473 K, 2 kPa of DME, 20 kPa of O2, balance He.
Estimated from fractional Pt dispersion assuming hemispherical clusters.
Units of mol of DME/(mol of surface Pt s).
b
c
(Quantachrome Corp.). H
2
adsorption isotherms were measured at 3-50
at 673 K
kPa of H after passivated samples were treated in pure H
2
2
for 1 h and then in a dynamic vacuum at 673 K for 1 h. A back-
sorption isotherm was then measured after evacuation at 313 K for 0.5
h to determine uptakes of weakly bound hydrogen. Pt dispersions were
measured from the difference between these two isotherms extrapolated
2
to zero H pressure by assuming a 1:1 H:Pt adsorption stoichiometry.
Average cluster diameters were estimated from these dispersion values
by assuming hemispherical clusters and using the density of bulk Pt
metal21 (21.45 g cm ). Dispersions and cluster sizes are reported in
Table 1. X-ray absorption near-edge spectroscopy (XANES) did not
detect bulk oxidation of Pt clusters during DME combustion catalysis
at 473 K (Supporting Information); Pt clusters are therefore denoted
as Pt metal.
2. Experimental Methods
-3
2
.1. Catalyst Synthesis and Characterization. γ-Al
2
O
3
(193 m2
3
O (Sasol North America Inc., lot
g- ) was prepared by treating Al
1
2
3
-1 -1
no. C1643) in flowing dry air (Praxair, 99.99%, 0.8 cm g
23 K (0.083 K s ) for 5 h. ZrO
as described previously.
s
) at
-1
2
-1
9
2
(monoclinic, 34 m g ) was prepared
precursors obtained by precipitation
1
9,20
ZrO
2
2.2. Steady-State Catalytic DME Combustion Reactions. Catalytic
rates were measured on catalysts (5 mg, 250-425 µm particle size)
2 2
of ZrOCl ·8H O (Aldrich, CAS no. 13520-92-8) were treated in flowing
2
-1
dry air at 1023 K for 5 h and at 923 K for 5 h. The crystal structure
diluted within pellets with inert supports (SiO , 280 m g , Chro-
2
and surface area of the supports were characterized by X-ray diffraction
matographic Silica Media, CAS no. 112926-00-8) to avoid temperature
and concentration gradients and to ensure strict kinetic control. The
2
(XRD; Siemens, D 500, Cu KR radiation) and N adsorption at 77 K
using an Autosorb 6 system (Quantachrome, Inc), respectively.
catalysts (50-100 mg) diluted with SiO (2.5-3.8 g) at dilution ratios
2
Supported Pt catalysts were prepared by incipient wetness impregna-
of SiO
2
:catalyst ) 25:1, 50:1, and 75:1 were pressed at ∼10 MPa and
tion of γ-Al O
2 3
and ZrO
2
2 6 2
with aqueous H PtCl ·6H O solutions (Aldrich,
sieved to retain 106-250 or 250-425 µm particles, and then a portion
CAS no. 16941-12-1). Impregnated samples were treated in ambient
(catalyst content 5 mg) of the sieved sample was placed within a quartz
tube (8 mm i.d.). The SiO diluent was washed with 3 M nitric acid
2
3
-1 -1
air at 393 K and then in flowing dry air (0.7 cm g
s
) at 823 K
(Praxair,
-1
(
9
0.083 K s ) for 5 h. The samples were then treated in pure H
2
(EMD Chemicals Inc., CAS no. 7697-37-2, ACS reagent) and treated
in dry air at 923 K for 5 h before use. Temperatures were measured
with a type K thermocouple enclosed within a sheath in contact with
the catalyst bed. Temperatures were within (0.5 K of the average bed
temperature at all bed positions.
3
-1 -1
-1
9.999%, 0.8 cm g
s
) at 723 K (0.083 K s ) for 2 h and in 1%
3
-1
s
-1) at ambient temperature for 1 h to
O
2
/He (Praxair, 0.4 cm g
passivate them before exposure to ambient air and use in rate and
chemisorption measurements. Pt contents were measured by ICP-OES
(
Galbraith Laboratories). Pt/Al
ZrO (1.0 and 1.1 wt %) samples were prepared as shown in Table 1.
Portions of 0.9 wt % Pt/Al and 1.0 wt % Pt/ZrO samples were
then treated again in (i) flowing dry air (0.7 cm g s-1) at different
temperatures (823-923 K) and then in H at 723 K for 2 h or (ii)
flowing H at temperatures above 723 K (773-923 K) to vary their
metal dispersion; these samples were also passivated in 1% O /He
2
O
3
(0.8, 0.9, and 2.3 wt %) and Pt/
Reactant mixtures consisted of premixed gases containing 2% DME
2
and 20% O with He as the diluent (Praxair certified mixture) or of
2
2
O
3
2
separate streams consisting of 15% DME in He (Praxair certified
mixture), 90% O2 in N2 (Praxair certified mixture), and pure He
(99.999%, Praxair), which were mixed by independent control of their
respective flow rates. The effects of CO and H O products on DME
3
-1
2
2
2
2
2
combustion rates were measured by introducing 50% CO /Ar (Matheson
2
before exposure to ambient air and subsequent use for chemisorption
and catalytic measurements.
certified mixture) or by vaporizing liquid H O (doubly distilled,
deionized water; syringe pump, Cole Parmer, 60061 series) into flowing
2
Pt dispersions were measured from volumetric uptakes of strongly
DME-O reactant streams. Heated lines (373-423 K) were used to
2
chemisorbed H
2
at 313 K using a Quantasorb chemisorption analyzer
transfer all streams into the reactor and into an Agilent 6890 gas
chromatograph equipped with a methyl silicone capillary column (HP-
1, 25 m × 0.32 mm × 1.05 µm) connected to a flame ionization detector
and a Porapak Q packed column (80-100 mesh, 12 ft × 1/8 in.)
connected to a thermal conductivity detector.
(
(
(
11) Valeden, M.; Lai, X.; Goodman, D. W. Science 1998, 281, 1647.
12) Haruta, M. Catal. Today 1997, 36, 153.
13) Markovic, N.; Gasteiger, H.; Ross, P.N. J. Electrochem. Soc. 1997, 144,
1
591.
(
(
(
(
(
14) Saib, A. M.; Claeys, M.; Van Steen, E. Catal. Today 2002, 71, 395.
2
.3. Isotopic Exchange Rates and Kinetic Isotopic Effects. Kinetic
15) Bethke, K. A.; Kung, H. H. J. Catal. 1997, 172, 93.
12
16) Wei, J.; Iglesia, E. J. Phys. Chem. B 2004, 108, 4094.
isotope effects (CH
3
OCH
3
-CD
3
OCD
3
18
) and isotopic exchange rates ( -
17) Hicks, R. F.; Young, M. L.; Lee, R. G. J. Catal. 1990, 122, 295.
18) Fujimoto, K.; Ribeiro, F. H.; Avalos-Borja, M.; Iglesia, E. J. Catal. 1998,
12
13
13
or 16O
CH
3
O CH
3
- CH
3
O CH
3
2
2
- O ) were measured on a 1.0 wt
1
79, 431.
(
19) Wei, J.; Iglesia, E. J. Phys. Chem. B 2004, 108, 7253.
(21) David, R. L. Handbook of Chemstry and Physics, 87th ed.; CRC Press:
(20) Barton, D. G. Ph.D. Dissertation, University of California at Berkeley, 1998.
Boca Raton, FL, 2006.
13202 J. AM. CHEM. SOC.
9
VOL. 129, NO. 43, 2007

Catalytic DME Combustion on Supported Pt Clusters
A R T I C L E S
%
Pt/ZrO
batch reactor. Reactants and products were circulated at 1.7 cm s
over catalyst samples (1-15 mg, diluted with SiO 1:50) using a
2
(0.42 dispersion) catalyst using a gradientless recirculating
3
-1
2
graphite gear micropump. Chemical and isotopic compositions were
measured by direct sampling into a gas chromatograph equipped with
mass-selective and flame ionization detectors (Agilent 6890 GC-MS
system). Catalyst samples were treated in H
2
(Praxair, 99.999%, 0.7
) at 673 K for 1 h before all measurements.
OCH (Matheson, 99.5%), CD OCD (Isotec, chemical purity
99.0%), 13CH
(Isotec, chemical purity >99.0%), 50%
Ar (Praxair, 99.99%),
3
-1 -1
cm g
s
CH
3
3
3
3
13
16
>
3
O CH
3
2
O /
1
6
18
O
2
(Praxair, 99.99%),
2
O (Isotec, chemical
1
8
purity >99 atom % O), and He (Praxair, balance to give a 110 kPa
total pressure) were used as reagents, and Ar (3 kPa) was introduced
as an internal standard. Trace impurities (<1.0%) in CD
CH O CH
3 3
3
OCD
led to catalyst deactivation and required further purifica-
tion of these reactants by trapping impurities at 77-273 K.
.4. Computational Methods. Ab initio density functional theory
DFT) quantum chemical calculations were carried out to determine
3
and
13
13
Figure 1. DME conversion as a function of temperature during (b) catalysis
on 1.0 wt % Pt/ZrO (5 mg, 0.26 dispersion) and (0) gas-phase reaction
2
(without a catalyst) and (2) CO2 formation rate as a function of temperature
2
for gas-phase reaction [2 kPa of DME, 20 kPa of O2, balance He, DME
(
-
6
-1
inlet molar flow rate 1.37 × 10 mol s ].
adsorption and reaction energies and activation barriers for various
elementary steps using periodic plane wave density functional theory
as implemented in the Vienna Ab Initio Simulation Program (VASP).2
The PW91 functional form of the generalized gradient approximation
was used to calculate nonlocal corrections to the exchange and
homogeneous DME combustion pathways. These homogeneous
DME oxidation and pyrolysis radical chain reactions become
detectable only above 523 K and predominantly formed CO,
2-24
8
,30
CO2, CH4, HCHO, HCOOH, and H2O. Homogeneous DME
pyrolysis and oxidation rates initially increased, then decreased,
and ultimately again increased with increasing temperature, as
2
5
correlation energies. The electron-ion interactions were described
2
6
using ultrasoft pseudopotentials with a plane wave cutoff energy of
2
7
3
96.0 eV. A 3 × 3 × 1 Monkhorst-Pack k-point mesh was used to
8
also previously reported; these trends reflect changes in the
sample the first Brillouin zone. The Pt surface was treated as an ideal
Pt(111) substrate comprised of four metal layers separated by 15 Å of
vacuum. The bottom two layers of the slab were held fixed at their
bulk lattice positions while the top two layers were allowed to relax.
A 3 × 3 unit cell consisting of 9 metal atoms per layer (36 total Pt
atoms) was used to simulate the adsorption of DME and its initial
reaction paths over Pt(111) surfaces with different coverages of
chemisorbed oxygen. The structures as well as the first two layers of
the Pt surface were optimized such that the total free energies and the
concentration of reactive OH radicals with increasing temper-
ature. Homogeneous pathways formed HCHO, CO2, CH3OH,
and HCOOH below 973 K in our experiments; CO and H O
2
2
became the only products above 1073 K. Homogeneous com-
bustion reactions required much higher temperatures than cata-
lytic combustion on Pt-based catalysts and led to negligible con-
tributions to the catalytic rates reported here at 423-523 K.
Turnover rates for highly exothermic reactions, such as DME
combustion
-
4
band structure energies converged to within 1 × 10 eV. Subsequent
calculations were carried out on three metal layer substrates and 2 ×
2
unit cells to test the level of accuracy of the energy estimates. Reaction
CH OCH + 3O f
3
3
2
-1
energies varied by less than 3 kJ mol , indicating that the calculations
reported on the larger system provide an acceptable level of accuracy.
The activation barriers were all determined using the nudged elastic
band climbing (NEB-CI) method developed by Jonsson and described
in detail elsewhere.
-1 21
2CO
+ 3H O (∆H°298 ) -1460.4 kJ mol ) (1)
2
2
28
must be measured under strict kinetic control without ubiquitous
transport artifacts that lead to temperature and concentration
gradients within pellets or packed beds. These artifacts can be
detected and removed only by intrapellet and bed dilution of
catalysts with inert solids. Figure 2 shows DME combustion
turnover rates as a function of residence time at 473 K on 1.0
wt % Pt/ZrO2 (0.26 dispersion) for different pellet diameters
and extents of dilution. At low dilution levels (25:1), local
temperatures exceed those measured and rates become artifi-
cially high as a result of these artifacts. Higher dilution ratios
eliminated these artifacts and led to reaction rates that did not
depend on pellet diameters (105-250 vs 250-425 µm) or on
additional intrapellet dilution (50:1 vs 75:1) (Figure 2). DME
combustion turnover rates decreased with increasing residence
time because of the combined effects of reactant depletion (3-
14% DME conversion) and inhibition by products, as discussed
below. All rates reported henceforth were measured at the
highest dilution level and extrapolated to zero reactant residence
times, except for those measured with H2O and CO2 added to
reactants, which are reported at the prevailing DME conversion
levels.
2
9
3. Results and Discussion
3
.1. DME Catalytic Oxidation on Supported Pt Metal
Clusters at Low Temperature (423-523 K). The catalytic
combustion of DME on Pt-based catalysts forms only CO2 and
H2O. Figure 1 shows DME conversion with and without catalyst
(1.0 wt % Pt/ZrO2, 0.26 dispersion) as a function of reaction
temperature (2 kPa of DME, 20 kPa of O2, He balance). DME
combustion products (CO2, H2O) became detectable at ∼423
K on Pt/ZrO2. Figure 1 also shows CO2 formation rates from
(
(
(
(
22) Kresse, G.; Hafner, J. Phys. ReV. B 1993, 47, 558.
23) Kresse, G.; Furthm u¨ ller, J. Comput. Mater. Sci. 1996, 6, 15.
24) Kresse, G.; Furthm u¨ ller, J. Phys. ReV. B 1996, 54, 11169.
25) Perdew, P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M.
R.; Singh, D. J.; Fiolhais, C. Phys. ReV. B 1992, 46, 6671.
26) Vanderbilt, D. Phys. ReV. B 1990, 41, 7892.
(
(
(
27) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188.
28) Henkelman, G.; Uberuaga, B. P.; Jonsson, H. J. Chem. Phys. 2000, 113,
9
901.
(
29) The nudged elastic band is first used to optimize the forces along a set of
fixed points chosen here to be equally spaced along the reaction path
between the reactant and product. The climbing algorithm is subsequently
used to isolate the point of highest energy. This method increases the energy
of the structures along the reaction path to more quickly converge upon
the transition state.
(30) Rosado-Reyes, C. M.; Francisco, J. S.; Szente, J. J.; Maricq, M. M.;
Østergaard, L. F. J. Phys. Chem. A 2005, 109, 10940.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 43, 2007 13203

A R T I C L E S
Ishikawa et al.
Figure 4. Kinetic isotope effects for DME combustion on 1.0 wt % Pt/
ZrO2 catalysts (5-10 mg, 0.42 dispersion, 250-450 µm, 50:1) at various
temperatures and reactant pressures [(0) 10 kPa, (2) 20 kPa, and (O) 50
kPa of O2 and 2 kPa of CH3OCH3 or CD3OCD3].
Figure 2. DME combustion turnover rates as a function of residence time
on 1.0 wt % Pt/ZrO2 (5 mg, 0.26 dispersion) at 473 K for various pellet
diameters and SiO2:catalyst intrapellet dilution ratios: (2) 250-450 µm,
2
5
5:1; (b) 250-450 µm, 50:1; (0) 250-450 µm, 75:1; (O) 106-250 µm,
response of DME turnover rates to DME and O2 pressures may
reflect the requirement for vacancy sites (*) on oxygen-covered
Pt surfaces for DME activation and also the role of DME as
the reductant required to form such sites within catalytic redox
0:1 (2 kPa of DME, 20 kPa of O2, balance He).
18
cycles. Fujimoto et al. proposed that CH4 combustion pathways
on PdOx crystallites involve the activation of CH4 on site pairs
consisting of oxygen atoms (O*) and oxygen vacancies (*). The
availability of such O*-* site pairs depends on reactant and
product (H2O, CO2) pressures and on the size of active PdOx
clusters. The effect of DME and O2 partial pressures on the
surface concentration of O*-* site pairs leads to the measured
dependence of turnover rates on DME and O2 partial pressures,
as we discuss below. DME and O2 kinetic orders of 2.0 ( 0.2
and -1.0 ( 0.3, respectively, were measured by regression
(dashed curves) of the data in Figure 3 to a simple power law
model. The dashed curves in Figure 3 show that DME turnover
rates can be accurately described by a rate equation of the form
Figure 3. Effects of DME (a) and O2 (b) partial pressures on DME
combustion turnover rates on 1.0 wt % Pt/ZrO2 (5 mg, 0.42 dispersion,
2
50-450 µm, 50:1) at 473 K [(O) 10 kPa and (2) 20 kPa of O2 in (a) and
(b) 2 kPa and (0) 4 kPa of DME in (b)].
2
]-1
3
.2. Kinetic Dependence of Dimethyl Ether Combustion
r ) k[DME] [O
2
(2)
Rate on Reactant Partial Pressures. The kinetic response of
DME combustion rates to DME and O2 partial pressures was
measured on Pt/ZrO2 (1.0 wt %, 0.42 dispersion) catalysts at
3.3. CH3OCH3-CD3OCD3 Kinetic Isotope Effects. Kinetic
isotope effects (rH/rD) were measured from DME combustion
rates (extrapolated to zero conversion) using CH3OCH3-O2 or
CD3OCD3-O2 reactants (2 kPa of DME) at several temperatures
(473, 483, 493, and 503 K) and O2 partial pressures (10, 20,
and 50 kPa) using a recirculating gradientless batch reactor.
Measured kinetic isotope effects (KIEs) are shown in Figure 4.
These KIE values are significantly larger than unity, indicating
that the kinetically relevant steps involve the activation of C-H
bonds, and not C-O bonds, in DME-derived intermediates.
These KIE values did not depend on O2 pressures, indicating
that the relative importance of any denominator terms in the
more complex equations derived below did not vary with O2
pressure. Similar KIE values were reported previously for C-H
bond activation in methoxide species (CH3O*) involved in
kinetically relevant steps during methanol oxidation on MoO3-
supported (533 K, KIE ) 4.2) and RuOx/TiO2 (393 K, KIE )
4
73 K. Figure 3a shows the effects of DME pressure on
combustion turnover rates at two different O2 pressures (10 and
0 kPa). DME turnover rates increased faster than linearly with
2
increasing DME pressure. The effects of O2 on DME combus-
tion turnover rates are shown in Figure 3b at two different DME
pressures (2 and 4 kPa). At both DME pressures (2 and 4 kPa),
combustion rates on Pt catalysts decreased with increasing O2
pressure at low O2 pressures (<20 kPa), but became nearly
independent of O2 pressure as O2 pressures increased.
Adsorption of O2 on reduced Pt clusters (Pt/Al2O3) leads to
a nearly saturated monolayer of chemisorbed oxygen at 300
3
1
K. The high activation energy for recombinative desorption
of oxygen from nearly saturated Pt surfaces [∼175 ( 10 kJ
-
1
31
32
mol on 2.9 wt % Pt/Al2O3 (0.85 dispersion); Pt(110), ∼199
-
1
(
10 kJ mol ] prevents equilibration of dissociative chemi-
3
4-36
sorption steps at temperatures below ∼800 K. Reductants such
as CO and H2, however, can remove chemisorbed oxygen atoms
from Pt surfaces at nearly ambient temperatures.33 The kinetic
2.5) catalysts.
DME oxidation to formaldehyde on MoO3/
Al2O3 catalysts also involves methoxide species formed via
37
dissociative adsorption of DME, but C-O bond cleavage
(
31) Bourne, A.; Bianchi, D. J. Catal. 2002, 209, 114.
(34) Holstein, W. L.; Machiels, C. J. J. Catal. 1996, 162, 118.
(35) Machiels, C. J.; Sleight, A. W. J. Catal. 1982, 76, 238.
(36) Liu, H.; Iglesia, E. J. Phys. Chem. B 2005, 109, 2155.
(37) Cheung, P.; Liu, H.; Iglesia, E. J. Phys. Chem. B 2004, 108, 18650.
(
32) Wartnaby, C. E.; Stuck, A.; Yeo, Y. Y.; King, D. A. J. Chem. Phys. 1995,
1
02, 1855.
(33) Bourne, A.; Bianchi, D. J. Catal. 2004, 220, 4.
1
3204 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007
9

Catalytic DME Combustion on Supported Pt Clusters
A R T I C L E S
1
2
13
Figure 5.
CH3O CH3 mole fractions in CH3OCH3 reactants during
Figure 6. 16O O mole fractions in O2 reactants during reactions of CH316-
18
1
2
12
13
13
reactions of CH3O CH3- CH3O CH3 and O2 on 1.0 wt % Pt/ZrO2 (5
16
18
OCH3 and equimolar O2- O2 on 1.0 wt % Pt/ZrO2 (5 mg, 0.42 dispersion,
12
12
mg, 0.42 dispersion, 250-450 µm, 50:1) at 473 K [1 kPa of CH3O -
16
2
50-450 µm, 50:1) at 473 K [2 kPa of DME, 10 kPa of O2, 10 kPa of
1
3
13
CH3, 1 kPa of CH3O CH3, 20 kPa of O2, 3 kPa of Ar, balance He].
18
O2, 3 kPa of Ar, balance He].
Scheme 1. Reaction Pathways for the Oxidation of DME on
Supported Pt Clusters, Including the Alternate Assumptions of (i)
Methoxide Formation via DME Dissociative Adsorption and (ii) No
Methoxide Formation
leading to methoxide formation is the kinetically relevant step,
and this reaction shows very weak kinetic isotope effects. Thus,
it seems plausible at first to attribute the measured KIE values
for DME oxidation to the relative rates of hydrogen abstraction
from CH3O* and CD3O* surface species. It is also possible that
these isotope effects reflect the kinetically relevant activation
of C-H bonds in molecularly adsorbed DME, as proposed from
vibrational spectra and desorption dynamics on Rh surfaces at
subambient temperatures (250-300 K)38 and also suggested by
the difficult C-O bond activation in DME oxidation to
formaldehyde on MoOx-based catalysts.37 These two possible
pathways cannot be discerned from these data because they can
lead to similar rate equations; thus, we require a more rigorous
analysis of the elementary steps involved in DME activation
and of the isotopic exchange data that we present below.
3
.4. Isotopic Evidence for the Irreversible Character of
12
12
DME and O2 Dissociation Steps. Reactions of CH3O CH3-
CH3O CH3-O2 mixtures were used to probe the rates of
1
3
13
recombinative desorption of DME-derived methoxide intermedi-
ates. Figure 5 shows the distribution of DME isotopomers during
DME combustion on Pt/ZrO2 (1.0 wt %, 0.42 dispersion) at
13
4
73 K in the recirculating batch reactor. Mixed 12CH3O CH3
binative desorption of O2 on Pt single-crystal surfaces (∼199
isotopomers were not detected at any DME chemical conversion
level (0-50%), indicating that the formation of methoxide
species via C-O cleavage in molecularly adsorbed DME is
irreversible or that the C-O bond cleavage ultimately required
for the formation of CO and CO2 in DME combustion reactions
on Pt occurs only after all kinetically relevant steps.
-1 30
(
10 kJ mol ), which would preclude adsorption-desorption
equilibration at 423-523 K.
.5. Plausible Elementary Steps Involved in Dimethyl
3
Ether Activation and Combustion. The lack of methyl
exchange between labeled and unlabeled DME (section 3.4)
indicates that methoxide formation via DME dissociation is
either irreversible or kinetically irrelevant. We consider next
the kinetic implications of (1) an irreversible DME dissociation
step followed by H-abstraction and (2) an alternate route via
H-abstraction from undissociated molecular DME and consider
these steps as kinetically relevant. We examine here the
sequence of elementary steps shown in Scheme 1, which
includes these two routes for C-H bond activation: (i) C-H
activation from methoxide species formed after irreversible
DME dissociation (steps 1.2 and 1.3) and (ii) C-H activation
in molecularly adsorbed DME (step 1.4). In both routes, surface
vacancies (*) form via quasi-equilibrated recombination of two
hydroxyl species (OH*) to give water (step 1.5) and the
The reversibility of dissociative O2 chemisorption steps was
probed by measuring O O formation rates during reactions
16
18
16
16
18
of CH3 OCH3 with equimolar O2- O2 mixtures (2 kPa of
DME, 10 kPa of 16O2, 10 kPa of 18O2). Quasi-equilibrated
dissociation steps would lead to binomial isotopomer distribu-
tions (50% O O). Mixed isotopomers were not detected at
73 or 503 K over a wide range of DME chemical conversions
0-50%, Figure 6). These data show that O2 dissociation steps
are irreversible; O O detection limits indicate that recombi-
native desorption rates are >200 times lower than DME
chemical conversion rates at 473 K. These conclusions are
consistent with the strong binding of chemisorbed oxygen on
1
6
18
4
(
1
6
18
3
9
Pt surfaces and with the high activation energies for recom-
(
39) van Santen, R. A.; Neurock, M. Molecular Heterogeneous Catalysis; Wiley-
(38) Bugyi, L.; Solymosi, F. Surf. Sci. 1997, 385, 365.
VCH Verlag GmbH & Co. KgaA: Weinheim, Germany,2006.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 43, 2007 13205

A R T I C L E S
Ishikawa et al.
sequence is completed via a set of kinetically irrelevant steps
that ultimately lead to the desorption of CO2 (step 1.6). DME
absorbs on surface vacancies (*) in a quasi-equilibrated adsorp-
tion-desorption step (step 1.1). This is supported by DFT
calculations reported below in section 3.6 which show that the
bond between DME and the Pt(111) surface is rather weak.
Surface O* species are replenished via irreversible dissociative
chemisorption of O2 coreactants (step 1.7). These two deriva-
tions and the measured KIE values (4.2-3.2 at 473-503 K)
indicate that hydrogen abstractions from methoxide species (step
adsorption (step 1.1, K1) and irreversible DME dissociation (step
1.2, k2), but lacks any kinetic or thermodynamic parameters for
elementary steps involving C-H bond activation (step 1.3, k3).
Thus, the sequence involving C-H activation of chemisorbed
methoxide intermediates (steps 1.2 and 1.3) cannot be respon-
sible for DME combustion on Pt cluster surfaces. In contrast,
eq 6 includes the rate constant for H-abstraction from molecu-
larly adsorbed DME (k4). It is, therefore, consistent with
measured KIE values (4.2-3.2 at 473-503 K). Equation 6
indicates that KIE values for DME combustion depend on the
ratio of rate constants given by
1
.3, k3) and from molecularly adsorbed DME (step 1.4, k4) are
the respective kinetically relevant steps in these two possible
catalytic sequences. If the respective rate equations derived from
the sequence of elementary steps do not contain k3 or k4
parameters, those steps become kinetically irrelevant; as a result,
rate equations without k3 or k4 and the mechanisms giving rise
to such rate equations are inconsistent with the measured KIE
values.
Pseudo-steady-state concentrations for all adsorbed species
in Scheme 1 lead to complex kinetic rate expressions for the
cases of (i) C-H abstraction from methoxide species (eq 3)
and (ii) C-H activation from molecularly adsorbed DME (eq
2
(K1,Hk4,H) k
7,D
KIE )
(7)
2
(
K1,Dk4,D) k
7,H
H-abstraction (step 1.4, k4) from adsorbed DME is expected
to show a normal H-D kinetic isotope effect, while equilibrated
DME adsorption (step 1.1, K1) and reoxidation of the vacancy
sites (step 1.7, k7) would give small or no isotope effects,
respectively. We note also that the constant KIE values measured
as a function of O2 are consistent with the presence of a single
dominant term in the denominator of eq 4, as indicated by its
simplified form given by eq 6 used to describe the data in Figure
4
). Their full algebraic derivations are included in Appendices
1
and 2 in the Supporting Information, and only the final
3
.
equations are listed here.
Figure 7 shows parity plots for measured DME turnover rates
Figure 3) and those calculated from eq 6. The kinetic
Each term in the denominator of these equations represents
the concentration of the indicated adsorbed intermediate, relative
to the concentration of vacancy sites. Equations 3 and 4 become
eqs 5 and 6,
(
parameters derived from this regression analysis are reported
in Table 2. Turnover rates calculated from this kinetic expression
agree with measured rates. A comparison of methanol and DME
combustion rates, discussed next, confirms that DME dissocia-
tion and the subsequent activation of C-H bonds in the
methoxide species formed are not responsible for measured
DME combustion rates on Pt cluster surfaces.
If the methoxide intermediates shown to mediate methanol
oxidation catalysis were also kinetically relevant in DME
oxidation, the rates of these two reactions would be similar as
these methoxide species become the MASIs. Methanol and
DME oxidation rates on Pt/ZrO2 (1.0 wt %, 0.42 dispersion)
are shown in Figure 8 as CO2 formation rates. Methanol
oxidation formed formaldehyde, methyl formate (MF), and CO2
and H2O at lower temperatures (313-423 K) than DME
combustion to form CO2 and H2O (473-523 K). These
methanol oxidation rates are similar to those reported previously
2
2
2
9
K k [DME]
1 2
r )
(5)
(6)
k [O ]
7
2
2
2
2
4
.5K k [DME]
1 4
r )
k [O ]
7
2
respectively, when O* species are the most abundant surface
intermediates (MASIs). Both rate equations are consistent with
the kinetic data in Figure 3 (eq 2); we have found no other
sequence of elementary steps or combinations of steady-state
and MASI assumptions consistent with these measured kinetic
data.
Equation 5, corresponding to activation of methoxide species
formed via DME dissociation, however, predicts very weak
H-D isotope effects arising from equilibrated DME molecular
4
0-42
on Pt-based catalysts.
MF forms via HCHO reactions with
methoxide intermediates. We note that (i) methanol oxidation
36
13206 J. AM. CHEM. SOC.
9
VOL. 129, NO. 43, 2007

Catalytic DME Combustion on Supported Pt Clusters
A R T I C L E S
Table 3. DFT-Calculated Adsorption of DME and Reactive
Intermediates Bound to Pt(111)
oxygen
coverage
(ML)
binding
oxygen
coverage
(ML)
binding
energy
(kJ mol-1)
energy
(kJ mol-1)
adsorbate
adsorbate
DME (CH3OCH3)
0
-21.0 CH3OCH2
-19.9
0
-182.4
1
4
2
0
1
4
2
/9
/9
/3
1/9
4/9
2/3
1/3
0
-12.3
-107.8
-103.1
-365.3
-221.2
-11.0
-144.6 oxygen (O)
-108.5 hydroxyl (OH)
methoxy (CH3O)
/9
/9
/3
1/9
4/9
-77.4
-189.1
-179.0
2/3
Figure 7. Parity plots for the calculated DME turnover rates using kinetic
rate expression (eqs 6 and 10) and DME turnover rates as shown in Figures
These results, as well as the absence of HCHO and MF products
during DME oxidation, indicate that DME activation on
supported Pt clusters does not involve methoxide intermediates.
The activation of terminal C-H bonds in DME controls reaction
rates, consistent with DFT calculations described next.
3
and 14 [DME effects (Figure 3a), (9) 10 kPa of O2, (0) 20 kPa of O2;
O2 effects (Figure 3b), (4) 2 kPa of DME, (2) 4 kPa of DME; H2O effects
Figure 14), (]) 5 kPa of H2O].
(
Table 2. Parameters for the Kinetic Model for Figure 7
3.6. Ab Initio Calculations of DME Combustion Elemen-
(
K
1
k
4
)2/k
mol/(mol of
7
1 4 5
K k K
tary Steps on Pt(111) Surfaces at Various Oxygen Coverages.
The initial C-H and C-O bond activation steps involved in
DME oxidation were examined by carrying out ab initio density
functional theoretical calculations on idealized Pt(111) surfaces.
While the (111) surface is an incomplete representation of the
active surfaces on Pt clusters, it provides a convenient starting
point for a systematic assessment of the energetics that control
the elementary processes of adsorption and desorption and the
surface reactivity for the specific activation paths that govern
reaction selectivities. The data reported below (section 3.9) also
show that larger Pt clusters are much more active than smaller
clusters, making the closest packed (111) surface the most
relevant model for comparisons among alternate elementary
steps. DFT calculations are used herein to determine the reaction
energies and activation barriers for both the C-O and C-H
activation paths (paths 2 and 4, respectively, in Scheme 1)
involved in the conversion of DME to two surface methoxide
[
(mol/(mol of
rate expression
surface Pt s kPa)]
surface Pt s)]
a
2
2
2
r ) 4.5K1 k4 [DME] /k7[O2]
0.35 ( 0.012
b
r ) K1k4K5[DME]/[H2O]
0.82 ( 0.042
a
DME turnover rates without addition of H2O, eq 6. b DME turnover
rates with addition of H2O, eq 10.
(CH3O*) species or to the methoxymethyl (CH3OCH2*) and
hydroxyl (OH*) surface products, respectively.
The binding energies for the key intermediates involved in
the initial DME combustion paths were calculated first on the
ideal Pt(111) surface in the absence of oxygen; the results are
summarized in Table 3. The results indicate that the DME
molecule is weakly bound on Pt(111) (-19.9 kJ mol ),
consistent with the reversibility assumed in the kinetic analysis
described above. DME adsorption involves electron donation
from the lone pair in its oxygen atom to the Pt surface. The
adsorbed species lies tilted from the normal to the Pt(111)
surface. As the oxygen coverages increased to 4/9 monolayer
Figure 8. Arrhenius plot for DME (9) and methanol (O) combustion on
1
.0 wt % Pt/ZrO2 (5 mg, 0.42 dispersion, 250-450 µm, 50:1) [2 kPa of
DME or 2 kPa of methanol, 20 kPa of O2, balance He].
-
1
occurs at nearly ambient temperatures (vs ∼423 K for DME
combustion in Figure 1), (ii) CO2 formation rates from methanol
are much faster than CO2 formation rates from DME on Pt/
ZrO2 catalysts (Figure 8), (iii) all methoxide intermediates
formed during CH3OH combustion reactions are converted to
CO2 at ∼413 K (data not shown), and (iv) DME combustion
did not form HCHO or MF products. Specifically, the C-H
activation of the methoxide intermediates and CH3OH complete
oxidation to CO2 and H2O occur above 413 K. The kinetically
relevant step is therefore much slower in the DME combustion
reactions, apparently because the activation of C-O bonds to
form methoxides is unfavorable and methoxides do not form
before the activation of terminal C-H bonds in adsorbed DME.
(
ML) and then to 2/3 ML, the DME binding energy decreased
-1
to -12.3 and -11.0 kJ mol , respectively. The adsorbed DME
molecules at the higher coverage adopt an orientation that is
nearly perpendicular to the surface to minimize repulsive
interactions with neighboring chemisorbed oxygen atoms.
DME can initially react with oxygen to form either two
methoxide surface intermediates or adsorbed methoxymethyl
and hydroxyl intermediates. The calculated structures for these
intermediates indicate that the methoxide intermediate prefer-
entially adsorbs atop a Pt surface atom, especially as oxygen
surface coverage increases. The binding energy for the meth-
oxide intermediate on Pt(111) in the absence of coadsorbed
(
40) Hinz, A.; Larsson, P. -O.; Andersson, A. Catal. Lett. 2002, 78, 177.
41) Wu, J. C. S.; Fan, W. -C.; Lin, C. -A. Ind. Eng. Chem. Res. 2003, 42,
(
3
225.
(42) Chantaravitoon, P.; Chavadej, S.; Schwank, J. Chem. Eng. J. 2004, 97,
1
61.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 43, 2007 13207

A R T I C L E S
Ishikawa et al.
Table 4. DFT-Calculated Reaction Energies for the C-O and C-H Activation of Dimethyl Ether with Oxygen in the Gas Phase and over
Pt(111) and Oxygen-Covered Pt(111) Surfaces
1/9 ML O surface coverage
2/9 ML O surface coverage
4/9 ML O surface coverage
2/3 ML O surface coverage
reaction
gas phase
E
a
∆E
rxn
E
a
∆E
rxn
E
a
∆E
rxn
E
a
∆E
rxn
DME(g) + O(g) f CH
3
OC H
2
(g) + OH(g)
-32.8
-51.8
DME(g) + O(g) f 2CH
3
O( g)
* + OH*
DME* + O* f CH
3
OC H
O*
2
+114.8 -79.2 (-51.1)
+234.0 -7.0 (+43.9)
+108.2
-102.0
+100.9 -104.0 (-88.4)
-9.2
+99.0 -107.0 (-108.4)
+221.0 -14.0
DME* + O* f 2CH
3
-1
oxygen was calculated to be -144.6 kJ mol . The methoxym-
ethyl intermediate adsorbs in a di-σ configuration via the
formation of σ bonds between Pt and the CH2 carbon and Pt
and oxygen. The interaction between the surface and the CH2
group is much stronger than with the oxygen atom. It takes on
the form of a substituted methyl surface intermediate. The
binding energy of the methoxymethyl intermediate on Pt(111)
-
1
was calculated to be -182.4 kJ mol . The surface hydroxyl
intermediate binds to atop sites with a binding energy of -212.2
-
1
kJ mol , while atomic oxygen binds at 3-fold fcc sites on the
-
1
surface with a binding energy of -365.3 kJ mol .
The binding energies for all of these potential intermediates
increased in the order
CH OCH * < CH O* < CH OCH * < OH* < O*
3
3
3
3
2
on both Pt(111) and oxygen-covered Pt(111) surfaces. The
calculated binding energies are reported in Table 3 as a function
of the coverage of chemisorbed oxygen. The binding energies
for all these species decreased as the oxygen coverage increased.
The most strongly bound intermediates showed the greatest
decrease in their binding energies as oxygen coverage increased.
The same binding energy rankings hold, however, at all oxygen
coverages examined here.
The overall reaction energies for various proposed elementary
steps were first estimated using a thermochemical cycle that
neglects intermolecular interactions among adsorbed surface
species:
Figure 9. DFT-calculated structures and reaction energies for the C-H
and C-O bond activation paths of DME over Pt(111) at 1/9 ML oxygen
coverage.
for C-H over C-O activation, therefore, solely reflects the
much greater stability of the adsorbed products formed in the
C-H bond activation path. The resulting overall reaction
energies from a Born-cycle analysis were calculated to be -51.1
-1
kJ mol for DME to form the methoxymethyl and hydroxyl
-1
surface intermediates and +43.9 kJ mol for DME to form
two methoxide intermediates (see Table 4). These calculations
support the mechanistic conclusions reached on the basis of both
the kinetic and isotopic analyses and described in sections 3.2
and 3.3. In contrast with CH3OH, DME reacts via the activation
of C-H bonds in molecularly adsorbed species on O*-covered
Pt surfaces.
CH OCH * + O* f
3
3
-
1
2
CH O* (∆E ) +43.9 kJ mol ) (8)
3 rxn
CH OCH * + O* f
3
3
-
1
CH OCH * + OH* (∆E ) -51.1 kJ mol ) (9)
3
2
rxn
These reactions were probed in more detail by specifically
allowing reactants to form products within a given unit cell, a
strategy that accounts for lateral through-space and through-
surface interactions among reactant and product states. The
overall reaction energies as well as the activation barriers for
the two proposed DME oxidation paths were first calculated at
low oxygen surface coverage, 1/9 ML. The corresponding
structures for the reactant and product state are shown in Figure
These estimates use the adsorption energies reported in Table
3
4
together with the gas-phase reaction energies reported in Table
. The C-H bond cleavage step that forms methoxymethyl
intermediates (CH3OCH2*) from molecularly adsorbed DME
-
1
gives more favorable reaction energies (by 95 kJ mol ) than
C-O bond cleavage to form two chemisorbed methoxide
species. The reaction energy for DME to react entirely in the
gas phase with a reactive atomic oxygen atom to form two
9
. The corresponding reaction energies are now slightly more
-
1
-
1
exothermic at -79.2 and -7.0 kJ mol for DME to form the
CH3OCH2*/OH* and CH3O*/CH3O* surface intermediates as
the result of these lateral interactions. These interactions
selectively weaken reactant over product states, thus making
both reactions more favorable than reported above from ideal
thermodynamic cycles. The calculated activation barrier for
DME reactions with chemisorbed oxygen to form methoxym-
methoxide radicals is 19 kJ mol more favorable than the
corresponding homogeneous reaction from DME and atomic
O to form the methoxymethyl and hydroxyl radicals. This energy
difference, however, is quite small and compensated in surface
reactions by the larger combined heats of adsorption of CH3-
-
1
-1
OCH2* (-182.4 kJ mol ) and OH* (-221.2 kJ mol ), which
are much more favorable than for two CH3O* species (-289.2
-
1
-1
-1
kJ mol total, -144.6 kJ mol for each methoxide species).
ethyl and OH intermediates was calculated to be 114.8 kJ mol ,
which is much lower than the 234.0 kJ mol activation barrier
-
1
The reactant state for both paths is the same. The selectivity
13208 J. AM. CHEM. SOC.
9
VOL. 129, NO. 43, 2007

Catalytic DME Combustion on Supported Pt Clusters
A R T I C L E S
Figure 11. DFT-calculated structures and reaction energies for the C-H
and C-O bond activation paths of DME over Pt(111) at 2/3 ML oxygen
coverage.
These reaction paths on surfaces with low oxygen coverages
may not accurately reflect the reactivity of surfaces nearly
saturated with chemisorbed oxygen, which prevail during
catalytic DME combustion. Thus, reaction energies were also
calculated for C-O and C-H bond activation at 2/9, 4/9, and
2
/3 ML of adsorbed oxygen. We present the optimized product
Figure 10. Isolated transition states for (a) the C-H activation of DME
to form methoxymethyl and hydroxyl intermediates and (b) the C-O
activation of DME to form two methoxide intermediates.
structures for the 2/3 ML of oxygen in Figure 11 simply to
illustrate the availability of sites at these higher coverages. The
resulting activation and overall reaction energies studied are
reported in Table 4 for each coverage. These results indicate
that higher oxygen coverages lead to more favorable activation
of both C-H and C-O bonds. The overall reaction energies
for C-H bond activation decreased in the order -79.2, -102.0,
-104.0, and -108.0 kJ mol as the oxygen coverage increased
from 1/9 to 2/9, 4/9, and 2/3 ML, respectively. The activation
barriers concurrently decreased in the order 114.8 (1/9 ML of
O), 108.2 (2/9 ML of O), 100.9 (4/9 ML of O), and 99.0 (2/3
ML of O) kJ mol as the oxygen coverage increased. There is
a clear correlation between the activation barriers and the
exothermicity of the reaction (Figure 12), as expected from
ubiquitous Brønsted-Evans-Polanyi (BEP) relations. We have
reported similar relationships for methane and methanol activa-
calculated for DME and chemisorbed oxygen to form two
surface methoxide intermediates.
The transition-state structures isolated for both C-H bond
and C-O bond activation paths to form the CH3OCH2*/OH*
and CH3O*/CH3O* products are shown in Figure 10. The
transition state for C-H activation shows the incipient stretching
of the C-H bond and the interaction of the H-atom with one
of the chemisorbed oxygen species. The metal atom next to
that binding DME also has a bond to an oxygen atom, but it is
still accessible because this oxygen resides at an fcc site on the
other side of the chemisorbed DME. The availability of this
neighboring Pt atom helps to stabilize both the carbon and
hydrogen atoms as the C-H bond elongates and facilitates the
required charge transfer. The basic oxygen atom ultimately binds
with the abstracted H-atom, which acquires a more positive
charge as the C-H bond is broken. The calculated barrier here
is very similar to that found for the oxygen-assisted activation
-1
-
1
44
tion, both of which also involve the activation of primary C-H
bonds.
As the oxygen coverage increased, the transition-state struc-
tures remained effectively unchanged. The differences in
coverage resulted predominantly in changes in the local
electronic structure of the surface. As the oxygen surface
coverage increased, the activating oxygen atom in the transition
state became more weakly bound and thus more basic. This
enhanced its ability to abstract hydrogen from adsorbed DME.
The activation of the C-H bond of DME with oxygen can be
compared with the similar step in methane reactions, because
DME can be considered to be a substituted methane molecule.
The transition states for the activation of DME and methane
both involve stretching of a C-H bond and the formation of
an O-H bond and are expected to give similar slopes in BEP
-
1 43
of methane on Pt, which is 97 kJ mol . The transition state
to form the two methoxide species shown in Figure 10b involves
the cleavage of the oxygen-methyl bond and the formation of
a new oxygen-methyl bond. There is significant steric repulsion
between the methyl group and the surface-bound oxygen atoms
in the transition state. In addition, the metal surface does not
directly participate in C-O bond activation. The CH3 group in
DME transfers directly from its oxygen to the coadsorbed
oxygen without the formation of intermediate metal-carbon
bonds. These two factors ultimately lead to the high activation
barrier required to activate C-O bonds of DME.
(43) Janik, M. J.; Neurock, M. Unpublished results.
(44) Buda, C.; Neurock, M. Unpublished results.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 43, 2007 13209

A R T I C L E S
Ishikawa et al.
on 1.0 wt % Pt/ZrO2 (2.7 nm), and 88 kJ mol- on 1.0 wt %
Pt/ZrO2 (3.2 nm). The larger clusters are likely to contain greater
contributions from 111 facets and resemble more closely the
1
(111) surface used in theoretical estimates.
DFT-derived barriers can be compared with measured ap-
parent activation energies by noting that the latter reflect the
temperature dependence for the combined kinetic and thermo-
dynamic constants in eq 6. The apparent activation energies
measured reflect the activation barrier for initial C-H bond
activation (E1), the adsorption enthalpy for DME molecules
(
-∆H1), and the activation barrier for O2 dissociation (E7) (Eapp
2E4 + 2∆H1 - E7). DFT estimates for the C-H bond
activation barrier and the enthalpy of adsorption of DME at
)
-
1
high coverage were found to be +99 and -11 kJ mol ,
respectively. The activation barrier for the activation of O2 over
a 2/3 ML of oxygen-covered Pt(111) surface was calculated
-
1
from DFT to be 61 kJ mol . The apparent activation energy
calculated from these DFT-derived constants is 115 kJ mol ,
Figure 12. DFT-calculated Brønsted-Evans-Polanyi relationships for the
C-H activation of DME by chemisorbed oxygen on Pt(111). The calculated
BEP relationship for the C-H activation of methane by chemisorbed oxygen
on Pt(111) is also shown for comparison.
-1
in reasonable agreement with the range of experimental values
-
1
reported here on supported Pt clusters (88-104 kJ mol ).
.7. Dimethyl Ether Combustion Reaction Pathways. The
3
relations. The two will differ, however, because the oxygen in
DME ultimately interacts with the surface, thus resulting in a
more exothermic reaction than for methane. There is also some
interaction of the oxygen and the surface in the transition state
for DME, but not in the case of methane. This would lead to
slightly lower barriers and to a smaller BEP slope, consistent
with the results shown in Figure 12.
irreversible nature of DME and O2 dissociation is consistent
with the proposed elementary steps in Scheme 1. The proposed
sequence of elementary steps (Scheme 1) for DME oxidation
on Pt clusters does not involve methoxide species (step 1.2) or
subsequent H-abstraction from these species (step 1.3). The
catalytic sequence involves instead the kinetically relevant C-H
bond activation (step 1.4) in adsorbed DME on O*-* site pairs
and resembles that for CH4 combustion on supported PdOx
clusters. The rate expression (eq 6) obtained from this sequence
of proposed elementary steps is consistent with measured DME
combustion turnover rates (Figure 3) when O* species are the
MASIs. Measured KIE values, the absence of reversible DME
dissociation steps, and DFT-calculated activation barriers and
reaction energies indicate that the reaction proceeds via H-
abstraction from molecularly adsorbed DME in irreversible steps
(step 1.4), with the C-O bond activation step occurring in
subsequent steps and without any detectable kinetic relevance.
The C-O activation path to form two methoxide intermedi-
18
ates also becomes slightly more exothermic as the coverage
-
1
increased from 1/9 ML of oxygen (-7.0 kJ mol ) to 4/9 ML
-
1
of oxygen (-9.2 kJ mol ) to 2/3 ML of oxygen (-14 kJ
-
1
-1
mol ). The barrier decreased from 234 to 221 kJ mol as the
oxygen coverage increased from 1/9 to 2/3 ML. We have
explored only the two extremes of low and high oxygen
coverages here because barriers for C-O activation were nearly
twice those for the more favorable C-H activation in all cases
examined.
At higher oxygen coverages, both C-H and C-O bond
activation steps become more favorable. The DFT results
showed that vacant sites are required to adsorb DME. This is
consistent with the relevant elementary steps suggested by the
kinetic and isotopic analyses described in sections 3.2 and 3.3
and with the proposed involvement of O*-* site pairs in
kinetically relevant steps. As oxygen coverages increase, the
C-H bond activation steps become increasingly favored over
competing C-O activation steps to form methoxide species,
which do not appear to be involved in DME combustion on Pt
surfaces. All the theoretical results presented here indicate that
reactions of DME on the oxygen-covered Pt surfaces proceed
via the preferential activation of the C-H bond over the C-O
bond in molecularly adsorbed DME.
The density of oxygen vacancy sites (*) required for DME
adsorption (step 1.1) depends on DME and O2 concentrations
(eq 4). Also, the concentration of these O*-* pairs depends
on the quasi-equilibrated adsorption-desorption of H2O and
CO2 as their concentrations increase with increasing DME
conversion. As a result, the number of available O*-* site pairs
decreases as O2, H2O, and CO2 concentrations increase and as
DME concentrations decrease. Equation 6 is also consistent with
the decrease in DME combustion turnover rates as a function
of residence time (Figure 2), indicating that depletion of the
reactant is the main reason for the observed decrease of DME
combustion turnover rates with increasing residence time. The
products inhibit DME combustion turnover rates by changing
the concentration of O*-* pair sites; these inhibition effects
are discussed in the next section.
The elementary-step activation energies derived from DFT
can be compared with those measured from experimental DME
combustion rates extrapolated to zero conversion to exclude
product inhibition effects on intrinsic rates and to measure
barriers corresponding to those for kinetically relevant steps.
As discussed in section 3.9, measured apparent activation
energies vary with Pt cluster size. Specifically, measured barriers
3.8. Inhibition of Dimethyl Ether Combustion by Reaction
Products. The effects of CO2 and H2O products on combustion
rates were measured by adding CO2 and H2O to DME-O2
reactant streams. These rates were measured at low DME
3
-1 -1
conversions (<1.5%) and a similar flow rate (330 cm g s ).
Figure 13 shows that combustion turnover rates decreased with
increasing H2O concentration, while rates were essentially
are 104 kJ mol on 0.8 wt % Pt/Al2O3 (1.2 nm), 95 kJ mol-1
-
1
13210 J. AM. CHEM. SOC.
9
VOL. 129, NO. 43, 2007

Catalytic DME Combustion on Supported Pt Clusters
A R T I C L E S
Figure 13. Effects of H2O (a) and CO2 (b) partial pressures on DME
combustion turnover rates on 1.0 wt % Pt/ZrO2 (5 mg, 0.42 dispersion,
2
50-450 µm, 50:1) at 473 K [(9) 10 kPa and (O) 50 kPa of O2 and 2 kPa
Figure 15. DME combustion turnover rates and calculated rate constants
of DME].
2
-1
((K1k4) k7 ) as a function of Pt cluster diameter for ZrO2 and γ-Al2O3
supports [473 K, 2 kPa of DME, 20 kPa of O2, balance He, (0) 1.0 wt %
Pt/ZrO2, (b) 1.0 wt % Pt/Al2O3, ([) 2.3 wt % Pt/Al2O3]. Rate constants
were calculated by using eq 6.
H2O products readsorb to give significant coverages by chemi-
sorbed OH* species.
Figure 7 also shows the parity plots for DME turnover rates
calculated using the kinetic rate expression (eq 10) and measured
DME turnover rates (Figure 14). It is observed that the reaction
rates calculated with the kinetic expressions derived from our
model are also in good agreement with measured DME turnover
rates (Figure 14). The effects of DME, O2, H2O, and CO2
concentrations on turnover rate are consistent with rate-
determining C-H bond activation (from adsorbed DME species)
on site pairs consisting of a Pt-O species and an oxygen
vacancy.
Figure 14. Effects of DME (a) and O2 (b) partial pressures on DME
combustion turnover rates with 5 kPa of H2O (open symbols) and without
H2O (closed symbols) [1.0 wt % Pt/ZrO2 (5 mg, 0.42 dispersion, 250-450
µm, 50:1), 473 K, 20 kPa of O2 in (a) and 2 kPa of DME in (b)].
unaffected by CO2 concentration. Inhibition by water reflects
the reversible nature of hydroxyl recombination steps (Scheme
The availability of vacancy sites on PdO surfaces during CH4
combustion depends sensitively on oxygen and hydroxyl binding
x
1
, step 1.5), which leads to a decrease in the number of vacancy
energies, which are strongly influenced, in turn, by the size of
1
5-18
(*) and oxygen (O*) species, both of which are required for
PdO clusters.
The similar mechanistic requirements for
x
DME activation (Scheme 1, step 1.4). The proposed rate
equation (eq 4) simplifies to the following equation when OH*
species are the MASIs:
these two reactions suggest that similar sensitivity to structure
would be apparent in catalytic DME combustion. Such a
hypothesis is examined in the next section, in which dispersion
effects on DME turnover rates are reported for the first time.
K k K [DME]
1
4
5
3
.9. Cluster Size and Support Effects on DME Combus-
tion Turnover Rates. The effects of Pt cluster size and support
Al2O3, ZrO2) on DME turnover rates are shown in Table 1
r )
(10)
[
H O]
2
(
Turnover rates then become linear in DME pressure and
independent of O2 and decrease with increasing H2O concentra-
tion. Figure 14 shows the effects of DME and O2 partial pressure
on combustion turnover rates in the presence of H2O (5 kPa).
For comparison, the effects of DME (20 kPa of O2) and O2 (2
kPa of DME) in the absence of H2O (Figure 3) are also shown
in Figure 14. Without H2O, DME turnover rates were propor-
and Figure 15. DME combustion turnover rates increased with
increasing Pt cluster size (Figure 15), indicating that surfaces
of large Pt clusters are more effective than those of smaller
clusters in catalyzing the kinetically relevant steps required for
DME combustion. For a given dispersion, DME combustion
turnover rates were similar on Pt/Al2O3 and Pt/ZrO2 for samples
with relatively high dispersions (0.26-0.99) and lower reactiv-
ity. For the most active catalysts, with the lowest dispersion
2
-1
tional to [DME] and [O2] , consistent with eq 6 (O* ) MASI).
In contrast, rates increased linearly with increasing DME
concentration and were almost zeroth-order in O2 in the presence
of water, consistent with the trends predicted by eq 10. The
kinetic response of DME combustion rates to DME, O2, and
H2O concentrations is similar to that for CH4 combustion rates
(
0.02-0.26), Al2O3 supports gave more reactive catalysts than
45
ZrO2 supports. As reported previously, these support effects
reflect bifunctional pathways in which acid sites on γ-Al2O3
catalyze the hydrolysis of DME to form CH3OH using H2O
formed via DME combustion on active Pt clusters. The higher
H2O concentrations at the higher DME conversions prevalent
on larger and more active Pt clusters led to stronger contributions
from these bifunctional pathways and to the detection of the
1
0
-1
18
(
[CH4], [O2] , and [H2O] ) on supported PdOx clusters. DME
combustion on supported Pt clusters and CH4 combustion on
supported Pd clusters18 both involve the activation of DME and
CH4, respectively, on site pairs consisting of oxygen atoms (O*)
and oxygen vacancies (*), the density of which decreases as
(45) Ishikawa, A.; Iglesia, E. Chem. Commun. 2007, 2992.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 43, 2007 13211

A R T I C L E S
Ishikawa et al.
consequent bifunctional support effects on Pt catalysts with
lower dispersions.
DME to produce two methoxide surface species. The absence
of recombinative desorption of DME-derived intermediates and
O* species was also confirmed by measurement of isotopic
exchange rates. The derived rate expression from the proposed
sequence of elementary steps was consistent with kinetic
response of DME turnover rates to DME, O2, H2O, and CO2
concentrations. The density of O*-* site pairs required for
kinetically relevant H-abstraction steps from DME adsorbates
was strongly affected by reactant and product concentrations
and by the Pt cluster size. Theory indicates that the higher
oxygen surface coverages increase the basicity of the adsorbed
oxygen, thus making it more active in activating the C-H bonds
of DME. Vacant sites are necessary to allow DME to bind to
the surface. The observed effects of Pt crystallite size are related
to changes in the density and stability of vacancies on the Pt
surface and in the reactivity of surface oxygen atoms for the
kinetically relevant C-H activation step.
Similar effects of metal cluster size were reported for CH4
combustion turnover rates on supported PdOx18 and Pt clusters.46
Smaller metal clusters exhibit a higher density of coordinately
unsaturated sites, which bind oxygen more strongly than larger
4
7
clusters predominantly exposing low-index surfaces. Figure
2
-1
1
5 also shows the effective rate constant, keffect ((K1k4) k7 ),
calculated using the kinetic rate expression (eq 6) and DME
turnover rates measured at various residence times. The effective
rate constant, keffect, increased with increasing Pt cluster size.
This indicates that turnover rates increase as metal clusters
become larger, apparently because of a decrease in the binding
energy of chemisorbed oxygen atoms increasing both the activity
of O* for C-H bond activation steps (Scheme 1, step 1.4) and
the availability of uncovered metal atoms (*) required for
molecular chemisorption of DME (Scheme 1, step 1.1).
Therefore, Pt dispersion (cluster size) strongly influences the
number of O*-* site pairs available for kinetically relevant
H-abstraction steps from DME adsorbates.
Acknowledgment. This study was supported by BP as part
of the Methane Conversion Cooperative Research Program at
the University of California at Berkeley under the technical
direction of Dr. Theo Fleisch. We acknowledge Drs. Aditya
Bhan and Manuel Ojeda, Ms. Cathy Chin, Mr. Josef Macht,
and Mr. Brian Weiss of the University of California at Berkeley
for helpful technical discussions and their careful review of this
manuscript. We also kindly acknowledge the computational
resources provided by the Environmental Molecular Sciences
Laboratory at Pacific Northwest National Laboratory.
4. Conclusions
Isotopic measurements, kinetic dependences, and ab initio
density functional theoretical calculations led to a consistent
mechanistic picture of DME combustion reactions on supported
Pt catalysts, involving a kinetically relevant step requiring C-H
bond activation from molecular DME on a site pair consisting
of an adsorbed oxygen atom and a vacancy site on Pt surfaces.
Isotopic measurements as well as DFT results indicate that the
initial combustion of DME proceeds via the C-H activation
over O*-* site pairs to form methoxymethyl and hydroxyl
surface intermediates rather than through C-O activation in
Supporting Information Available: In situ evolution of Pt
clusters during DME combustion reactions obtained from X-ray
absorption spectra (Figure S1) and Appendices 1 and 2. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(
46) Li, M.; Chin, Y. -H.; Wei, J.; Iglesia, E. Unpublished results.
(
47) Xu, Y.; Shelton, W. A.; Schneider, W. F. J. Phys Chem. B 2006, 110,
JA073712Z
1
6591.
13212 J. AM. CHEM. SOC.
9
VOL. 129, NO. 43, 2007