Site requirements and elementary steps in dimethyl ether carbonylation catalyzed by acidic zeolites

Article abstract of DOI:10.1016/j.jcat.2006.09.020

Steady-state, transient, and isotopic-exchange studies of dimethyl ether (DME) carbonylation, combined with adsorption and desorption studies of probe molecules and infrared (IR) spectroscopy, were used to identify methyl and acetyl groups as surface intermediates within specific elementary steps involved in the synthesis of methyl acetate from DME-CO mixtures with >99% selectivity on H-zeolites. Carbonylation rates increased linearly with CO pressures but did not depend on DME pressures, suggesting that the addition of CO to CH₃ groups present at saturation coverage controls catalytic carbonylation rates. These reactions lead to acetyl groups that subsequently react with DME to form methyl acetate (423-463 K; >99% selectivity) and regenerate methyl intermediates, consistent with kinetic studies of CO reactions with CH₃ groups previously formed from DME and with kinetic and IR studies of DME reactions with acetyl groups formed by stoichiometric reactions of acetic anhydride. These studies show that CO reacts with DME-derived intermediates bound on zeolitic Al sites from the gas phase or via weakly held CO species adsorbed non-competitively with CH₃ groups. These reactions, in contrast with similar reactions of methanol, occur under anhydrous conditions and avoid the formation of water, which strongly inhibits carbonylation reactions.

Full text of DOI:10.1016/j.jcat.2006.09.020

Journal of Catalysis 245 (2007) 110–123
www.elsevier.com/locate/jcat
Site requirements and elementary steps in dimethyl ether carbonylation
catalyzed by acidic zeolites
a
a
b
b
a,∗
Patricia Cheung , Aditya Bhan , Glenn J. Sunley , David J. Law , Enrique Iglesia
a
Department of Chemical Engineering, University of California at Berkeley, Berkeley, CA 94720, USA
b
BP Chemicals, Hull Research and Technology Centre, Saltend, Hull HU12 8DS, UK
Received 24 July 2006; revised 18 September 2006; accepted 25 September 2006
Available online 27 October 2006
Abstract
Steady-state, transient, and isotopic-exchange studies of dimethyl ether (DME) carbonylation, combined with adsorption and desorption studies
of probe molecules and infrared (IR) spectroscopy, were used to identify methyl and acetyl groups as surface intermediates within specific
elementary steps involved in the synthesis of methyl acetate from DME–CO mixtures with >99% selectivity on H-zeolites. Carbonylation rates
increased linearly with CO pressures but did not depend on DME pressures, suggesting that the addition of CO to CH groups present at saturation
3
coverage controls catalytic carbonylation rates. These reactions lead to acetyl groups that subsequently react with DME to form methyl acetate
(
423–463 K; >99% selectivity) and regenerate methyl intermediates, consistent with kinetic studies of CO reactions with CH groups previously
3
formed from DME and with kinetic and IR studies of DME reactions with acetyl groups formed by stoichiometric reactions of acetic anhydride.
These studies show that CO reacts with DME-derived intermediates bound on zeolitic Al sites from the gas phase or via weakly held CO species
adsorbed non-competitively with CH groups. These reactions, in contrast with similar reactions of methanol, occur under anhydrous conditions
3
and avoid the formation of water, which strongly inhibits carbonylation reactions.
©
2006 Elsevier Inc. All rights reserved.
Keywords: Carbonylation; Acid; Zeolite; Mordenite; Dimethyl ether; Methanol; Carbon monoxide; Methyl acetate; Acetic acid; Carboxylic acid
1
. Introduction
carboxylic acids via Koch-type pathways, which involve CO
insertion into C–O bonds within tertiary surface-bound alkox-
ides and subsequent hydrolysis of the bound acetyl interme-
diates formed [10–16]. Fujimoto and co-workers first reported
methanol carbonylation to acetic acid on zeolites and proposed
an intermediate role of surface methyl groups [17]. Similar
reactions of methanol and dimethyl ether (DME) were later
reported on acidic zeolites and polyoxometallate clusters, but
with significant homologation side reactions and catalyst de-
activation [18–23]. A preliminary note from our group showed
that H-mordenite (H-MOR) and H-ferrierite (H-FER) catalyzed
DME carbonylation to methyl acetate with stable rates and
Methanol carbonylation accounts for ∼60% of acetic acid
TM
production worldwide [1–5]. The Monsanto and BP Cativa
processes use homogeneous Rh or Ir complexes and iodide co-
catalysts to carbonylate methanol (30–60 bar, 423–473 K) [1].
The Acetica process uses an immobilized Rh–carbonyl com-
plex grafted onto a pyridine-containing solid resin and iodide
co-catalysts to improve throughput and catalyst recovery [6].
Practical methanol carbonylation catalysts without costly noble
metals and corrosive iodide co-catalysis remain unavailable.
Carbonylation of alkenes and alkanols to carboxylic acids
via Koch-type reactions [7] is catalyzed by strong acids with-
out metal co-catalysts [8,9]. Acidic zeolites and sulfated zir-
conia catalyze the carbonylation of alkanols and alkenes to
>
99% selectivity at 423–463 K after an initial induction pe-
riod, during which acidic protons were replaced by methyl
groups and the water co-produced was removed [24]. The rate
of DME carbonylation was much higher than for similar reac-
tions of CH3OH, at least in part because H2O, formed in paral-
*
lel CH OH dehydration reactions, inhibits carbonylation steps.
Corresponding author. Fax: +1 510 219 0142.
3
E-mail address: iglesia@cchem.berkeley.edu (E. Iglesia).
Methyl acetate synthesis rates did not depend on DME pressure,
0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.09.020

P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
111
but increased linearly with CO pressure up to ∼1 MPa. This ki-
Table 1
2
7
Elemental analysis and extra-framework Al content from Al MAS NMR of
zeolite samples
netic response is consistent with a sequence of elementary steps
involving the formation of surface methyl groups at Brønsted
acid sites initially via direct DME reactions with protons and
then via chain transfer reactions of DME with surface acetyls
formed via rate-determining CO insertion into C–O bonds in
methyl groups at the catalytic steady state [24].
Zeolite name
Source
Si/Al
Na/Al
AlEF (%)
ICP
ICP
HMOR_8.9
HMOR_10.1
HMOR_9.8
HMOR_9.5
HMOR_6.0
Tosoh
8.9
10.1
9.8
9.5
6.0
46
9.1
9.0
8.7
–
–
–
–
20.7
18.6
19.5
19.9
21.0
–
–
–
–
6.6
∼1
–
–
–
Sudchemie
Zeolyst
Zeolyst
Zeolyst
Zeolyst
Zeolyst
Zeolyst
Zeolyst
Zeolyst
Zeolyst
Zeolyst
Zeolyst
Zeolyst
Zeolyst
AlSi-Penta
Here we provide evidence for these intermediates and el-
ementary steps and for their consistency with the measured
kinetic effects of DME, CO, and H2O using infrared (IR) spec-
troscopy, isotopic tracer and kinetic effects, and transient reac-
tion studies. We also report that H-MOR samples with similar
Si/Al ratio and extra-framework Al content give different car-
<0.002
HMOR_46
–
NaMOR_0.17
NaMOR_0.27
NaMOR_0.41
NaMOR_0.55
NaMOR_0.90
HMOR-Ox_15.3
HMOR-Ox_14.2
HMOR-Ox_16.5
HFER_34.5
0.17
0.27
0.41
0.55
0.90
–
9.0
9.9
+
bonylation turnover rates (per H ); methyl acetate synthesis
15.3
14.2
16.5
34.5
12.2
rates on these H-MOR samples do not correlate with the num-
ber of CO binding sites measured from low-temperature CO
adsorption uptakes, suggesting that the specific siting and struc-
ture of acidic Al–OH groups is critical to stabilize transition
states required for kinetically relevant CO insertion steps. In the
context of these results, we discuss various possible structures
for such active sites, specifically those involved in formation of
the C–C bond in methyl acetate synthesis.
–
–
–
–
3.5
15.4
HMFI_12.2
−
1
Praxair) for 3 h at 773 K (0.0167 K s ) to remove residual
+
+
organics and to convert NH cations to H .
4
27
Al magic-angle spinning (MAS) NMR spectra were mea-
sured with a Bruker AV-500 spectrometer using a 4-mm MAS
broadband probe. The spectra were referenced to octahedral
2
. Experimental
3+
Al cations in 1 M aluminum nitrate solutions. Acid forms
of the zeolites were kept for ꢀ12 h in a desiccator containing
aqueous 1 M NaCl (∼96% relative humidity), because these
hydration protocols sharpen 27Al NMR lines by weakening
quadrupole interactions [27]. The amount of extra-framework
Al, measured from these 27Al NMR; the Si, Al, and Na con-
tents (Galbraith Laboratories, ICP-OES); and the nomenclature
used are reported in Table 1.
2
.1. Catalyst preparation
Na-MOR (Si/Al ∼ 6.5, Zeolyst) was converted to NH4-
MOR via four sequential exchanges of Na-MOR (10 g) with
1
M NH4NO3 (0.2 L) at 353 K for 12 h. After each exchange,
the NH4-MOR was washed with 0.2 L of deionized water and
isolated by filtration. After the final exchange, the sample was
treated overnight at 393 K in ambient air and then in flowing
dry air (zero grade, Praxair) for 3 h at 773 K (0.167 K s ).
Chemically dealuminated HMOR samples were prepared by
refluxing HMOR (Si/Al ∼ 10, Zeolyst) in a 1.5 M aqueous
solution of oxalic acid dihydrate (99%, Fluka) (catalyst:oxalic
acid solution = 1:5, v/v) at 313, 323, or 343 K for 3 h before
washing with (0.2 L) deionized water, filtering, and drying at
−
1
2.2. Steady-state catalytic reactions of DME–CO and
DME–CO–H O mixtures
2
DME carbonylation rates and selectivities were measured in
a packed-bed stainless steel reactor (8.1 mm i.d., 9.5 mm o.d.)
held within a three-zone resistively heated furnace (Applied
Test Systems, 3210 series). Temperature was measured using
an axial multipoint thermocouple contained within a stainless
steel thermowell (1.6 mm o.d.). Catalyst samples (0.2–0.6 g,
125–250 µm particle diameter) were treated in flowing dry air
3
93 K in ambient air overnight; this process was carried out
to remove detrital Al species, but it also removed some frame-
work Al atoms [25,26]. The samples were subsequently treated
at 773 K (0.0167 K s ) in flowing dry air (zero grade, Praxair)
for 3 h. This treatment was carried out to explore the effects of
Al content on DME carbonylation rates.
−
1
3
−1 −1
g , zero grade, Praxair) for 2 h at 773 K
(∼1.67 cm s
−
1
at a heating rate of 0.167 K s
He (∼3.33 cm s
before cooling in flowing
−1 −1
g , UHP, Praxair) to reaction tempera-
3
NH4-MOR (Si/Al ∼ 10, Zeolyst, ∼14 g) was exchanged
with Na using 0.5 L aqueous NaNO3 (99%, EMD Chemicals,
tures (420–513 K) and introducing DME (99.5%, Praxair), 95%
CO/Ar (UHP, Praxair), or 2% DME/5% Ar/93% CO (99.5%
DME, UHP Ar/CO, Praxair) reactant mixtures. The reactant
mixtures were dried before use by passing through a CaH2
(0.5 g, 99.99%, Aldrich) bed held at ambient temperature. Wa-
ter was introduced into the reactant mixture after the CaH2
bed using a syringe pump (Cole-Parmer, model 100 series)
in experiments designed to explore kinetic inhibition of car-
bonylation reactions by water. The reactor effluent was brought
via transfer lines held at 423–473 K into a mass spectrometer
(MKS Spectra Minilab) or into a gas chromatograph (Agilent
0
.014–2.44 M) at 353 K for 12 h and then washed in 2 L of
deionized water and treated at 393 K overnight in ambient air
and then in flowing dry air (zero grade, Praxair) for 3 h at 773 K
−1
(
0.167 K s ). This treatment was carried out to explore the
+
+
effects of replacing H with Na on the rate of DME carbony-
lation on residual protons.
NH4-MOR (Si/Al ∼ 10, Zeolyst), H-MOR (Si/Al ∼ 44.5,
Zeolyst), H-MOR (Si/Al ∼ 10, Sudchemie), H-MOR (Si/Al ∼
9
.5, Tosoh), and NH4-MFI (Si/Al ∼ 12.5, AlSi-Penta Zeolithe
GmbH) samples were treated in flowing dry air (zero grade,

112
P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
6
0
890) equipped with a methyl-siloxane column (HP-1, 50 m ×
.32 mm × 1.05 µm) connected to a flame ionization detector
× 1.05 µm) using mass spectrometry (Agilent 5973) and flame
ionization detection. 12C and C isotopomer distributions in
DME and methyl acetate were measured from ion yields (DME:
46, 47, and 48 amu; methyl acetate: 43, 44, 45, 74, 75, and
76 amu) using deconvolution methods and mass fragmentation
patterns for unlabeled molecules [29]. Samples were treated in
13
and a Porapak Q column (80–100 mesh, 12 ft. × 1/8 in.) con-
nected to a thermal conductivity detector. These apparatus and
treatment protocols were also used for the studies described in
Sections 2.3–2.5.
3
−1 −1
g , zero grade, Praxair) at 773 K
flowing dry air (3.33 cm s
−
1
2
.3. Transient reaction studies involving DME–CO and CO
(0.0167 K s ) for 2 h and then cooled to reaction temperatures
before introducing reactants.
cycling
12
12
CH3O CH3 (99.5%, Praxair) and He (UHP, Praxair) mix-
Samples were sequentially contacted with DME–CO reac-
tures were recirculated over the catalysts held at 438 K for 1
h, followed by evacuation and He purge for ∼3.25 h before
the samples were exposed to isotopic mixtures to eliminate in-
tants and pure CO streams in transient studies designed to
probe the nature of carbonylation reactive intermediates and el-
ementary steps. Catalysts were treated in flowing dry air and
contacted first with DME–CO mixtures (930 kPa CO, 20 kPa
DME, 50 kPa Ar, 438 K) (as described in Section 2.2) for
12
12
duction periods. A mixture of CH3O CH3 (99.5%, Praxair)
13
13
13
and CH3O CH3 (99 at% C, Isotec) isotopomers was con-
tacted with CO (UHP, Praxair) on HMOR_9.8 to probe the
12
∼
5 h to obtain steady-state rates; the system was then brought
reversibility of C–O cleavage in DME during carbonylation
reactions. CH3OCH3 and CD3OCD3 (99 at%, Isotec) were re-
acted with CO either separately or as an equimolar mixture to
probe the reversibility of C–H bond dissociation steps and the
involvement of C–H bonds in kinetically relevant steps during
DME carbonylation.
−
1
−1 −1
to ambient pressure and treated in He (3.33 cm
s g ,
UHP, Praxair) until DME levels in the effluent (measured by
mass spectrometry) were below 0.02%. The system pressure
was then increased to 1 MPa in He before the samples were
exposed to either a 95% CO/Ar (UHP, Praxair) or pure CO
(
99.99%, Praxair) flow at 1 MPa and 438 K for various time
intervals, after which DME–CO mixtures were reintroduced
at 1 MPa total pressure (930 kPa CO, 20 kPa DME) and
2
.6. IR spectroscopic studies of DME, acetic anhydride, and
carbon monoxide adsorption
4
38 K. The transient evolution of methyl acetate (43 amu) and
DME (43 and 45 amu) was measured by on-line mass spec-
trometry with a time resolution of 10 s during these experi-
ments.
IR spectra were measured with 2 cm 1 resolution on self-
supporting wafers (∼20–40 mg) held within a quartz vacuum
cell with NaCl windows using a Nicolet NEXUS 670 IR spec-
trometer equipped with a Hg–Cd–Te (MCT) detector in the
−
2
.4. Acetic anhydride reactions with Brønsted acid sites
−
1
4
000–400 cm
frequency region. Samples were treated in
3
−1
HMOR_9.8 samples (∼0.5 g) treated in dry flowing air
flowing dry air (∼1.67 cm s , zero grade, Praxair) at 723 K
for 1 h, evacuated at 723 K for 2 h using a diffusion pump
(<0.01 Pa dynamic vacuum; Edwards E02) and cooled to 438 K
in vacuum before samples were contacted with DME (99.5%,
Praxair) or acetic anhydride (99%, EMD Chemicals) for 0.25 h.
Samples were treated similarly before cooling to 123 K (using
a constant flow of liquid N2) and exposing them to CO (UHP,
Praxair) at 123 K. IR spectra were collected for 120 s after each
CO dose without intervening evacuation.
(
(
as described in Section 2.2) were exposed to flowing He
3
−1 −1
g , UHP, Praxair) saturated with acetic anhy-
3.33 cm s
dride (99%, EMD Chemicals; 273 K, 0.11 kPa) and held at
4
38 K for ∼3.5 h. The reactor effluent was analyzed by gas
chromatography and mass spectrometry (as described in Sec-
tion 2.2) to determine the identity and the number of molecules
evolved in reactions of acetic anhydride molecules with zeolitic
protons. After contact with acetic anhydride, samples were ex-
posed to flowing He (3.33 cm s
remove physisorbed molecules before introducing DME/CO/Ar
3
−1 −1
g
) for ∼2 h at 438 K to
−1 −1
g ; 2/93/5 kPa; 438 K) and monitoring
2
.7. CO adsorption and temperature-programmed desorption
3
mixtures (3.33 cm s
studies
the products formed by mass spectrometry and gas chromatog-
raphy.
Samples (0.3–0.7 g, 125–250 µm pellet diameter) were
3
−1 −1
g , zero grade, Prax-
treated in flowing dry air (∼6.67 cm s
2
.5. Isotopic exchange and kinetic isotope effects
−
1
air) at 773 K for 1 h before cooling (at 0.167 K s ) to 253 K
using liquid nitrogen. A mixture of 1% CO/He (UHP, Praxair;
Isotopic-exchange experiments were carried out in a gradi-
3
3
−1 −1
g ) was introduced for 0.5–0.75 h on sam-
entless recirculating batch reactor (590 cm ) as described previ-
∼1.67 cm s
3
−1
)
ously [28]. The reactant stream was circulated (∼3.33 cm s
ples held at 253 K before flushing CO(g) with flowing He
3
−1 −1
g , UHP, Praxair) for 1.5 h to remove weakly
over the catalyst bed (∼0.5 g; 125–250 µm particle diam-
(∼1.67 cm s
eter) using a graphite gear micropump (Micropump, model
adsorbed CO. The temperature was then increased to 523 K
(0.167 K s ) and held at 523 K for 360 s. The concentration of
3
−1
1
82–000). Gas samples (1 cm ) were extracted from the circu-
lating gas stream and analyzed by gas chromatography (Agilent
890; HP-1 methyl siloxane capillary column; 50 m × 0.32 mm
CO (28 amu) in the He stream was monitored continuously by
mass spectrometry (MKS Orion Compact).
6

P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
113
2
.8. UV–visible spectroscopy of Co-exchanged zeolites
Co-exchanged HMOR_9.8 and HMFI_12.2 catalysts were
prepared by contacting zeolites with 0.1 M Co(NO ) ·6H O
3
2
2
solutions (Sigma Aldrich, CAS 10026-22-9, 98+%) for 24 h at
53 K. Samples were filtered, washed with 0.5 L of deionized
water, dried in ambient air overnight, and treated in flowing dry
3
−
1
air (zero grade, Praxair) at 773 K (0.0167 K s ) for 3 h. The re-
sulting Co/Al ratios measured by elemental analysis (Galbraith
Laboratories, ICP-OES) were 0.32 for MOR_9.8 and 0.16 for
MFI_12.2.
UV–vis spectra were measured using a Cary 400 Bio spec-
trophotometer (Varian) and an in situ flow cell (Harrick) with
a diffuse reflectance accessory. MgO powders treated in He
at 300 K (Praxair, UHP, dried on a 13X molecular sieve)
were used as reference. Reflectance (R) data were converted to
pseudo-absorbance [F(R)] using the Schuster–Kubelka–Munk
formalism. Cobalt-exchanged zeolite samples were treated in
Fig. 1. Methyl acetate (",!) and methanol (Q,P) synthesis rates without DME
pretreatment (solid symbols) and with DME pretreatment (DME
/Al = 1,
dosed
7.2 ks He purge, 438 K, open symbols) [HMOR_9.8, 930 kPa CO, 20 kPa DME,
38 K].
−
1
He flow by heating to 373 K (0.083 K s and hold for 1 h), then
4
−
1
to 523 K (0.083 K s and hold for 1 h), and finally to 773 K
(
−1
0.083 K s and hold for 1 h). Samples were then cooled to am-
bient temperatures (∼303 K), and the spectra were measured
in He flow. Linear combination methods and peak positions
and relative absorption coefficients reported previously [30–32]
were used to measure the relative distribution of Co cations
among α, β, and γ sites in HMOR_9.8 and HMFI_12.2. This
method provides a semiquantitative measure of Al pair site oc-
cupation in zeolites, because the siting of Co cations depends on
the method of Co incorporation, the dehydration procedure, the
presence of co-cations, and the total Co concentration [30–32].
3
. Results and discussion
3
.1. Steady-state DME carbonylation rates, kinetics, and
mechanism
Fig. 1 shows methyl acetate and methanol formation rates on
Fig. 2. Methyl acetate (") and methanol (P) synthesis rates in (A) DME–CO
mixtures (26 kPa DME, 123 kPa CO) and in (B) DME–CO–H O mixtures
HMOR_9.8 at 438 K as a function of reaction time after treating
samples in flowing dry air at 773 K before exposure to DME–
CO reactants (930 kPa CO, 20 kPa DME). Steady-state car-
bonylation rates and methyl acetate selectivities (>99%) were
achieved after ∼4 h. Methyl acetate synthesis rates increased
initially with reaction time, whereas methanol synthesis rates
decreased concurrently during this induction period (Fig. 1).
This induction period was not observed when samples were ex-
posed to DME pulses (DMEdosed/Al = 1) at 438 K and then
flushed with He for 2 h before contact with DME–CO reactants.
DME reacted with acidic protons in H-MOR and HMFI_12.2
2
(
26 kPa DME, 123 kPa CO, and 1.1 kPa H O) [HMOR_9.8, 423 K].
2
The low methanol selectivities (<1%) measured at steady state
indicate that essentially anhydrous conditions prevail during
steady-state DME carbonylation. Induction periods disappeared
when catalysts are exposed to DME before reaction, because
exchange sites become saturated with methyl groups and the
water formed is removed from the catalyst bed before catalysis.
Water formed during saturation of exchange sites with CH3
groups inhibits carbonylation rates. Fig. 2 shows that water
(1.1 kPa) added to DME–CO reactants (26 kPa DME, 123 kPa
CO, 423 K, HMOR_9.8) decreased methyl acetate formation
(
DMEdosed/Al ꢁ 1) to form methyl groups and water; the
DME/Al uptake stoichiometry was 0.5 ± 0.05 and 0.51 on
these samples [24]. DME molecules first react with an acidic
proton to form a methanol molecule and a methyl group; the
methanol thus formed can then react with another Brønsted
acid site to form H2O(g) and a second methyl group [24]. Dur-
ing steady-state DME–CO reactions, some CH3OH molecules
leave the reactor before they react with another proton, because
of competing reactions of DME reactants with such protons.
−
1
−1
rates from 0.034 to 0.0024 mol (g-atom Al)
h and in-
creased methanol formation rates by a factor of 14; no acetic
acid or other products were detected with or without added wa-
ter. Methanol and methyl acetate synthesis rates returned grad-
ually (∼4 h) to their initial values after the added H2O was
removed. This reactivation period resembled those observed
during initial exposure to DME–CO reactants.

114
P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
Scheme 1. Proposed elementary steps for carbonylation of dimethyl ether on acidic zeolites.
Methyl acetate synthesis rates were proportional to CO pres-
sure (0–930 kPa) and independent of DME pressure (0.8–
6.8 kPa), both under anhydrous conditions and when H O
3. Quasi-equilibrated reaction of DME with acidic protons to
form methanol and a chemisorbed methyl group at zeolite-
exchange sites:
6
2
(
0.5 kPa) was added [24]. Thus, active sites remained satu-
K3
CH3OCH3 + H ⇐⇒ CH3OH + *bCH3.
*b
rated with CH3 groups even in the presence of H2O, because
methyl acetate synthesis rates remained independent of DME
pressure at H2O levels that strongly inhibit carbonylation rates.
The first-order dependence on CO (up to ∼1 MPa) and the zero-
order kinetics in DME require non-competitive binding of DME
and CO or Eley–Rideal-type reactions involving gas-phase CO
species with bound DME-derived intermediates at saturation
coverage. Hence, water may inhibit reaction rates either by ad-
sorbing competitively with CO or by decreasing the reactivity
of surface methyl groups toward CO molecules. The linear de-
pendence of carbonylation rates on CO reflects the very low
occupancy of any CO binding sites prevalent during catalysis;
thus, the presence or involvement of chemisorbed CO species
cannot be confirmed by in situ spectroscopic methods. As a
result, the identity of CO-binding centers remains unclear at
this time, but our experimental observations are consistent with
the sequence of elementary steps detailed below (and shown
in Scheme 1). In what follows, we provide evidence for this
mechanistic proposal and discuss, in this context, possible car-
bonylation sites.
4
. Quasi-equilibrated reaction of methanol with acidic pro-
tons to form water and a second chemisorbed methyl group at
zeolite-exchange sites:
K4
CH3OH + H ⇐⇒ H2O + *bCH3.
*b
5
. Irreversible reaction of activated CO with a methyl group to
form an acetyl group at zeolite-exchange sites:
k5
*aCO + *bCH3 −→ *bCOCH3 + *a.
6. Reaction of DME with an acetyl group to form methyl acetate
and to re-form a methyl group at zeolite-exchange sites:
k6
CH OCH + COCH −→ CH COOCH + *bCH3.
*b
3
3
3
3
3
7
. Quasi-equilibrated CH3 exchange between DME and surface
CH3 groups:
K7
*b
*b
*a
CH OCH + CH ⇐⇒ CH + CH OCH .
1
. Quasi-equilibrated adsorption of CO on a binding site ( ):
3
3
3
3
3
3
Two distinct catalytic sites are involved in these pathways:
K1
CO + ⇐⇒ *aCO.
*a
*b
zeolite-exchange sites ( ) that stabilize acidic protons and
*a
methyl and acetyl groups, and CO-binding centers ( ) that
non-competitively bind DME and CO. The quasi-equilibrium
assumption for CO, H2O, and DME adsorption, and the pseudo-
steady-state approximation for all reactive intermediates gave a
rate equation for methyl acetate synthesis [Eq. (1); derivation in
2
. Quasi-equilibrated (competitive) H2O adsorption onto this
CO binding site:
K2
H2O + ⇐⇒ *aH2O.
*a

P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
115
Appendix A] of the form
3.2. In situ IR spectroscopic studies of DME reactions
K1k5PCOCT,a
IR spectra were collected after a sequence of DME pulses
(DME/Al = 1.2 cumulative total) on HMOR_9.8 held within a
closed system (Fig. 3). The intensity of acidic O–H stretches
rCH COOCH =
3
3
1
+ K1PCO + K2PH O
2
ꢀ
ꢁ
1
−
1
×
,
(1)
(3610 cm , Fig. 3a) decreased sharply with the number of
DME pulses, whereas bands for C–H stretches concurrently
became more intense (Fig. 3b). Evacuation (<0.01 Pa) after
dosing DME (DME/Al = 1.2) restored some of the initial O–H
band intensity (∼20%, Figs. 3c and 3d) but did not influence the
PH
O
1 5
CO T,a
K k P C
2
1
+
+
k6PDME(1+K1PCO+K2PH2O)
K4PCH OH
3
+
in which turnover rates (per H ) are not normalized by the con-
centration of CO binding sites, CT,a, because of their uncertain
identity.
−1
−1
strong band at 2978 cm or the weaker bands at 2868 cm
In these elementary steps, induction periods reflect the initial
slow replacement of protons with methyl groups via reactions
with DME. These steps initiate a propagation cycle that ulti-
−1
and 2855 cm , even after 1 h at 438 K. The bands at 2978
−
1
and 2868 cm correspond to antisymmetric and symmetric
C–H stretches in methyl groups, respectively. Such bands were
detected also after exposing H-MFI to DME or methanol at
73–523 K [33,34]. The weak band at 2855 cm was assigned
to methyl groups at silanols present at external H-MFI surfaces
SiOCH3) [33]. The bands and shoulders at 3011, 2971, 2947,
mately avoids H O formation by re-forming CH groups via
2
3
−
1
methoxylation of acetyl intermediates. When CH3 groups are
the most abundant reactive intermediates (MARI) and the con-
centration of surface-bound CO is low [or reactions occur di-
rectly with unbound CO(g)], these steps lead to a carbonylation
rate equation consistent with rate data. Methyl acetate synthe-
sis rates depend only on CO pressure and on the number of CO
4
(
−
1
and 2844 cm were previously assigned to DME molecules
hydrogen-bonded to acidic O–H groups [33]; these bands dis-
appeared on evacuation at 438 K for ∼0.4 h. CH groups and
3
binding sites (CT,a), or CH binding sites if reaction occurs with
hydrogen-bonded DME coexist during contact with DME, be-
3
CO(g), in step 5 [Eq. (2)]:
cause any H O formed (in steps 3 and 4) remains within the
2
closed system until evacuation and can displace CH3 (the re-
rCH COOCH = K1k5PCOCT,a.
(2)
verse of step 4) during evacuation to re-form some OH groups.
Molecular simulations have confirmed the persistent nature of
CH3 groups in HZSM-5 during methanol-to-olefin (MTO) reac-
3
3
If CO(g) reacted directly with MARI CH3 groups to form
acetyl groups, then methyl acetate synthesis rates [Eq. (3), per
13
tions [35], and C MAS NMR has detected them after contact-
+
H ] would show a kinetic dependence identical to that for re-
ing methanol with H-Y, HZSM-5, H-SAPO-34, and H-ZSM-
actions with any CO bound present at low fractional coverages
1
1 at 423–473 K [36–38]. These IR studies, taken together
[
Eq. (2)]:
with the DMEadsorbed/Al adsorption stoichiometry of ∼0.5 on
HMOR_6.0, HMOR_9.8, and HMFI_12.2 [24], provide spec-
troscopic and stoichiometric evidence for the prevalence and
relevance of CH3 groups at Brønsted acid sites during DME
carbonylation reactions.
k5PCO
r
=
≈ k P
CO
.
(3)
CH3COOCH3
5
PH
O
2
k5PCO
k6PDME
1
+
+
K4PCH OH
3
If CO(g) reacted directly, then H2O could inhibit rates ei-
ther by decreasing the rate of CO with CH3 or by altering the
concentration or reactivity of the latter. If CH3 concentrations
were influenced, however, H2O would adsorb competitively
with DME-derived intermediates and carbonylation rates would
become dependent on DME pressure as protons become the
3
.3. Isotopic evidence for the identity, kinetic relevance, and
reversibility of carbonylation elementary steps
Experimental protocols for 12C–13C mixed isotope stud-
ies initially involved exposing the HMOR_9.8 sample to
MARI at high H O concentrations. In such instances, Eq. (3)
12
12
2
3 3
CH O CH –He mixtures (Section 2.5), to avoid reactions
would become sensitive to both H2O and DME [Eq. (4)], in
contradiction to the experimental evidence provided herein:
of methyl groups with any H2O formed and retained during the
induction period. These reactions would form methanol [36]
and lead to isotopic scrambling via elementary steps unavail-
able during anhydrous DME carbonylation at the steady state.
These DME pretreatment protocols also eliminate induction pe-
riods, required to form methyl groups, so that steady-state car-
1
3
/2 1/2
1/2
DME
K
K
k5P
PCO
4
r
=
.
(4)
CH3COOCH3
1
H2O
/2
P
12
12
Carbonylation rates remained independent of DME pressure
even when H2O was added and markedly inhibited such rates
3 3
bonylation occurs immediately on contact with CH O CH –
13
13
12
CH O CH – CO mixtures and exclusively via the relevant
3
3
[
24]; thus, H2O does not inhibit rates by replacing CH3 groups
propagation steps 5 and 6 in Scheme 1.
+
with H , and H2O inhibition must reflect either a decrease
in CH3 reactivity in carbonylation reactions or the competi-
tive binding of CO and H2O on sites that bind reactive CO
species. Below we provide spectroscopic and isotopic evidence
to confirm the identity of proposed surface intermediates and of
elementary steps in Scheme 1 (steps 5–7).
Figs. 4a and 4b show the concentrations of DME and methyl
acetate isotopomers with increasing reaction time. Methyl
groups scramble to form 12CH O CH more than 70 times
13
3
3
faster than they convert to methyl acetate; consistent with
this, isotopomers reached their binomial distribution at reaction
times leading to only ∼1% DME conversion to methyl acetate

116
P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
Fig. 3. Infrared spectra of HMOR_9.8 at 438 K (treated at 723 K in air within the infrared cell and cooled to 438 K in vacuum) of (a) O–H and (b) C–H stretching
regions during exposure to DME (DME/Al = 0–1.2) and (c) O–H and (d) C–H stretching regions following exposure to DME (DME/Al = 1.2) and evacuation for
0–5.4 ks.
Fig. 4. Mole fraction of (a) DME and (b) methyl acetate isotopomers (solid symbols) and calculated binomial isotopomer distributions (dotted lines) during reactions
1
2
12
13
13
12
12
12
12
12
13
13
of CH O CH , CH O CH , and CO following pretreatment at 438 K in 4 kPa CH O CH for 1 h [2.35 kPa CH O CH , 2.03 kPa CH O CH ,
9
3
3
3
3
3
3
3
3
3
3
1
2
7.4 kPa CO, 0.5 g HMOR_9.8, 438 K].
(
∼2 ks). Rapid CH3 scrambling reflects rapid methoxylation
by theoretical calculations on chabazite [39] and may provide a
source of methyl groups for CO insertion steps, as for alkylation
in superacids [40] and methanol homologation on ZSM-5 [35].
of surface methyl groups by DME (step 7) or fast reversible
formation of trimethyloxonium ions; the latter were suggested

P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
117
These data also suggest that methoxylation of acetyls (step 6)
may be fast and kinetically inconsequential. However, fast
methyl scrambling prevents us from determining whether CO
Table 2
Initial carbonylation rate calculated from transient reactions of methyl groups
with CO and steady-state methyl acetate synthesis rates [transient: 1000 kPa
CO, 438 K; steady-state: 20 kPa DME, 930 kPa CO, 438 K]
insertion occurs before or after its dissociation from methyl ac-
etate isotopomers. No 13C (<1% detection limit) was detected
Zeolite
Initial carbonylation rate
Steady-state methyl
acetate synthesis rate
−
1
−1
h ]
[mol (g-atom Al)
in CO(g) or in the carbonyl group of methyl acetate (from
−
1
−1
h ]
[mol (g-atom Al)
1
2
12
COCH3 mass fragments) during reactions of CH3O CH3–
*
13
13
12
HMOR_9.8
HMOR_8.9
HMOR_16.5
HMFI_12.2
0.42
0.82
0.55
0.32
0.027
CH3O CH3– CO mixtures, indicating that carbonyl groups
in methyl acetate are derived only from CO(g).
0.19
0.14
Not detected
12
Kinetic isotope effects from methyl acetate synthesis rates
with CH3OCH3–CO and CD3OCD3–CO mixtures were very
small (kH/kD = 1.06) at 438 K on HMOR_9.8, suggesting that
C–H bonds are not cleaved or formed in kinetically relevant
steps. In addition, H-D scrambling in DME or methyl acetate
did not occur during reactions of CH3OCH3–CD3OCD3–CO
mixtures at 438 K, except to form CH3OCD3 as a result of fast
methyl scrambling.
*
950 kPa CO.
restores the methyl groups consumed to form acetyl intermedi-
ates, completing the catalytic cycle. The replacement of DME–
CO mixtures with pure CO led to an immediate decrease in
methyl acetate synthesis rates to undetectable values, indicat-
ing that any acetyls present during catalysis desorbed only after
methoxylation by DME. The replacement of DME–He (in-
stead of DME–CO) mixtures with pure CO before re-exposure
to DME–He at 438 K gave undetectable rates [<0.01 mol (g-
3
.4. Transient reactions of DME–CO and CO cycling
CO reactions with surface methyl groups to form acetyl
−
1
−1
atom Al)
h ] after a spike in methyl acetate formation on
species were proposed as the sole kinetically relevant step in
DME carbonylation to methyl acetate on HMOR_9.8 [24]. The
abrupt replacement of DME–CO mixtures with pure CO for
various time intervals led to the formation of precursors to
methyl acetate; however, these precursors desorbed only after
pure DME or DME–CO reactants were reintroduced. Subtract-
ing steady-state rates from “excess” methyl acetate formation
rates and integrating over time during contact with DME–CO
reactants led to an estimate of the number of methyl acetate pre-
cursors formed (per Al) during previous exposure of preformed
methyl groups to CO(g). These reflect, in turn, the number of
stranded acetyls formed, which increased monotonically with
increasing time of exposure to CO(g) (Fig. 5). Acetyls desorb
rapidly (as methyl acetate) on contact with DME or DME–
CO mixtures via reactions with DME (step 6), in a step that
re-exposure of samples to DME/He mixtures; the absence of
CO in the gas phase precludes the continuing replacement of
the acetyls used to form methyl acetate during this initial con-
tact, and rates become undetectable as acetyls are depleted
(step 6).
The initial rate of CO–CH3 reactions can be measured from
the number of excess methyl acetate molecules formed (per Al)
during transient experiments by determining the initial slope of
excess methyl acetate concentrations as a function of CO expo-
sure time (Table 2). These rates reflect the dynamics of CO reac-
tions on a surface initially saturated with methyl groups, which
reflects, in view of the measured steady-state kinetic depen-
dence, surface and kinetically relevant steps during carbonyla-
tion catalysis. Reaction rates decrease with CO exposure time,
as methyl groups are depleted in reactions forming CH3CO
species. Fig. 5 shows that CO reactions with CH3 groups are
faster on HMOR_9.8 than on HMOR_8.9, HMOR-Ox_16.5, or
HMFI_12.2, as is also found in the case of steady-state carbony-
lation, indicating that HMOR_9.8 is a superior catalyst because
it carbonylates methyl groups more efficiently than the other
samples.
The rates (per Al) of CO reactions with CH3 groups are
∼2 times lower than for steady-state DME carbonylation (Ta-
ble 2) except for HMFI_12.2, on which we could not detect
excess methyl acetate after exposure to CO. These differences
between steady-state and transient carbonylation rates may re-
flect nonuniform reactivity of methyl species, depending on
their location within channels or on their position relative to
other CH3 or CH3CO groups, or the carbonylation of trimethy-
loxonium ions during steady-state catalysis instead of methyl
groups during these transient experiments. The striking paral-
lel between transient and steady-state rates led us to conclude,
however, that CH3CO formation via CO reactions with sur-
face CH3 is indeed the kinetically relevant elementary step in
DME carbonylation on all samples, as was also indicated by
the measured dependence of steady-state rates on DME and
Fig. 5. Excess methyl acetate formed per Al during reaction with DME–CO
mixtures [930 kPa CO, 20 kPa DME, 50 kPa Ar, 438 K] immediately fol-
lowing exposure to pure CO [1000 kPa CO (950 kPa CO for HMOR_9.8)]
for varying flush times at 438 K on HMOR_9.8 (F, A), HMOR_8.9 (Q, B),
HMOR-Ox_16.5 (2, C), and HMFI_12.2 (", D); values determined by first-
(—) and second-order (- - -) rate dependences on methyl group concentration.

118
P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
Fig. 6. Infrared spectra of the (a) O–H and (b) C=O regions following pretreatment in air at 773 K for 1 h and cooling to 438 K (A), acetic anhydride addition
followed by evacuation (B), and following DME addition (DME/Al > 1) to acetic anhydride-treated samples for 7.2 ks and evacuation (C) [HMOR_9.8, 438 K].
CO pressures. The differences in catalytic carbonylation rates
among the various HMOR samples do not reflect differences
in the number of available CH3 groups, because samples with
similar Al content (HMOR_9.8 and HMOR_8.9) and CH3/Al
stoichiometries gave different steady-state carbonylation rates.
These differences may also reflect a range in reactivity of CH3
groups (for reactions with CO) depending on their location (in
pockets or channels in MOR) or on the number and proximity
of Al centers, CH3 groups, or acetyls; differences in the reten-
tion of water among zeolite structures may also contribute to
these reactivity trends, in light of the strong kinetic inhibition
by water.
3
.5. Formation of surface acetyls and their reaction with
dimethyl ether
We have proposed that acetyl groups formed by dissocia-
tive adsorption of acetic anhydride (or by carbonylation of
Fig. 7. The rate of methyl acetate upon re-introduction of DME–CO reactants
93 kPa CO, 2 kPa DME, 5 kPa Ar, 438 K) on HMOR_9.8 after exposure to
acetic anhydride–He (0.11 kPa acetic anhydride, 100 kPa He) at 438 K for
3.5h.
CH groups) react with DME to form methyl acetate mole-
3
(
cules (step 6, Scheme 1). We find, in accordance with previous
studies [41], that acetic anhydride reacts below 400 K with
O–H groups in H-MFI to form CH3CO groups. The IR band
∼
−
1
for acidic OH groups (3610 cm ) in HMOR_9.8 (Fig. 6a)
weakened on exposure to acetic anhydride (dose: 1.5 per Al) at
ously formed acetyls. The concurrent formation of acetic acid
or acetic anhydride was not observed. These data provide spec-
troscopic and stoichiometric evidence for CH3 and CH3CO
groups at Brønsted acid sites and for the high reactivity of acetyl
groups with DME to form methyl acetate and adsorbed methyl
species (step 6).
−
1
4
38 K for 0.25 h; two bands appeared at 1705 and 1755 cm
(
Fig. 6b), assigned to C=O stretches in acetyls (CH3CO) [41].
But this O–H band did not disappear on exposure to acetic an-
hydride, because it cannot access Brønsted acid sites within side
pockets in MOR. The O–H and acetyl bands were unchanged
during subsequent evacuation, but disappeared on contact with
3.6. Assessment of CO binding sites from IR spectra and
desorption dynamics of chemisorbed CO
DME at 438 K (Fig. 6b). DME reacts with these CH CO groups
3
to form methyl acetate and surface methyl groups; the latter
−
1
were evident from their C–H stretches at 2978 and 2868 cm
The nature of CO reactions with CH groups and the CO
3
(
not shown).
binding sites required remain unclear, but marked differences
HMOR_9.8 exposed to acetic anhydride (0.11 kPa) at
38 K (Section 2.4) led to methyl acetate synthesis rates
in carbonylation reactivity among catalysts with similar Si/Al
ratio (Fig. 8) appear to reflect kinetic nonuniformity of ac-
4
temporarily higher than steady-state rates (Fig. 7) on subse-
quent contact with DME/CO/Ar mixtures (2/93/5 kPa; 438 K;
tive sites involved in CO–CH reactions (Section 3.4). Extra-
framework Al atoms can act as Lewis acid centers that bind CO
[42–44], but carbonylation rates did not vary in a systematic
3
3
−1 −1
s ), as a result of fast methoxylation of previ-
3
.33 cm g

P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
119
Fig. 8. Arrhenius plot of methyl acetate synthesis rates on (a) H-MOR (Si/Al ∼ 10) from various sources: HMOR_9.5 (F), HMOR_9.8 (1), HMOR_8.9 ("),
HMOR_10.1 (P) and (b) on HMOR_9.8 refluxed in 1.5 M oxalic acid for 3 h (b): no oxalic acid treatment (1), 313 K (2), 343 K (P), and 323 K (") [930 kPa CO,
20 kPa DME].
Fig. 9. Infrared spectra of CO doses adsorbed at 123 K on (a) HMOR_9.8 (CO/Al = 0.04–0.19) and (b) HMOR_9.8 (CO/Al = 0.10–0.21) following water
pre-adsorption (0.042 H O/Al) at 298 K and degassing for 0.5 h before cooling to 123 K.
2
manner with the number of extra-framework Al atoms mea-
sured by 27Al MAS NMR (Table 1). Here we examine IR fea-
tures of CO adsorbed at low temperatures and CO temperature-
programmed desorption to explore the nature and number of
potential CO binding sites as well as their catalytic conse-
quences.
0.042) inhibited CO adsorption at 123 K on L1 and L2 sites, but
−1
did not influence CO interactions with OH groups (2173 cm
)
−
1
or with channel walls (2137 cm ) (Fig. 9b). The saturation in-
tensities of the L and L bands are shown in Fig. 10 for HMOR
1
2
samples with similar Si/Al ratios (∼10). Methyl acetate synthe-
sis rates did not vary monotonically with L or L intensities
1
2
CO interacted only with Lewis acid centers on HMOR on
(from peak areas normalized by those for Si–O–Si overtones at
2110–1760 cm ) (Fig. 11a).
−
1
initial exposure, leading to C=O stretches at 2224 (L ) and
1
−1
2
196 cm (L2) (Fig. 9a). The L1 band was assigned to CO
The total number of CO binding sites was measured from the
amount of CO adsorbed at 253 K and CO pressures leading to
saturation of L1 and L2 sites. CO/Al ratios increased from 0.01
in the parent mordenite (HMOR_9.8) to 0.03–0.04 in samples
dealuminated by treatment with oxalic acid (Table 3). However,
carbonylation rates did not vary in any systematic manner with
the measured number of CO binding sites (Fig. 11b), suggest-
ing that strongly bound CO on Lewis acid sites is not involved
in carbonylation reactions. Carbonylation rates are proportional
to CO pressure (up to >1 MPa), indicating that any CO bound
strongly enough to be detectable in spectroscopic studies is un-
likely to be involved in kinetically relevant reactions of CO.
3+
on strong Lewis acid sites (e.g., extra-framework Al ) in
H-MOR [44] and SiO2–Al2O3 [45]. L2 sites correspond to
penta-coordinated Al3 centers that are more weakly acidic
than L1 sites [46]. All Lewis acid centers became saturated
before CO interacted with acidic O–H groups via H-bonding
+
−1
(
ν
= 2173 cm ) or with channel walls via van der Waals
C=O
−
1
−1
interactions (νC=O = 2137 cm ). The band at 2173 cm
−
1
strengthened as the band for acidic O–H groups (3610 cm
)
−
1
shifted to lower frequencies and broadened (∼3300 cm
,
not shown), consistent with interactions mediated by hydrogen
bonding [44,47,48]. Water pre-adsorbed at 298 K (H O/Al =
2

120
P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
3
.7. Studies of H-MOR with different Brønsted acid site
act earlier during stoichiometric carbonylation of previously
concentrations
formed CH3 groups, perhaps for reasons unrelated to their rela-
tive proximity. Although the Al or CH3 proximity requirements
remain unclear at this time, these studies unequivocally show
that Brønsted acid sites are required for carbonylation.
The requirement for Brønsted sites was explored by varying
the number of H in H-MOR by changing Na/Al or Si/Al ratios
+
and by chemical dealumination. Methyl acetate synthesis rates
3
.8. Requirement for CO binding on a separate site for
+
+
(
per available H ) increased monotonically with increasing H
carbonylation of CH3
+
concentration (H /Al) and proximity (Fig. 12), a finding in-
consistent with the sole involvement of H species of uniform
+
The formation of C–C bonds via CO insertion into O–CH3
bonds in adsorbed methyl groups involves CO-derived species
that do not compete with DME-derived CH3 groups for bind-
ing sites. These data do not allow us to discern among three
possible mechanistic hypotheses:
reactivity in stabilizing reactive intermediates required for ki-
netically relevant carbonylation steps. The reactivity of methyl
or trimethyloxonium groups at Brønsted acid sites appears to
depend on the number and identity of vicinal sites or on the
location of such sites in channel or side pockets.
The possibility that proximity among CH3 groups influences
(
i) Unbound CO(g) reacts directly with surface CH3 groups.
their reactivity was explored by describing the trends in Fig. 5
(
ii) CO reacts after binding (non-competitively) on a site al-
2
with rates that depend on (CH ) or (CH ) . Carbonylation rates
3
3
ready containing a CH group.
3
depend slightly more strongly than linearly on (CH3), but these
trends probably merely indicate that more reactive CH3 re-
(iii) CO reacts after binding on a site vicinal to a CH group
3
(on which CH does not bind).
3
Proposal (i) seems inconsistent with the effect of zeolites
+
+
structure and H density (Fig. 12) on turnover rates (per H ),
+
because all H in all zeolites saturate with CH3 via reaction
with DME. H2O inhibition effects also cannot be reconciled
with this proposal, because surfaces remain saturated with CH3
groups and carbonylation remains zero-order in DME even in
Table 3
Concentration of CO bound per Al atom at 253 K from CO temperature-
programmed desorption studies
Zeolite
CO/Al
HMOR_9.8
HMOR_9.5
0.021
0.0092
0.016
0.0033
0.038
0.029
0.013
HMOR_10.1
HMOR_8.9
HMOR-Ox_14.2
HMOR-Ox_16.5
HMFI_12.2
Fig. 10. Infrared spectra of CO doses adsorbed at 123 K following saturation
−
1
of 2224 and 2196 cm bands with CO on HMOR_9.8 (A), HMOR_8.9 (B),
HMOR_9.5 (C), and HMOR_10.1 (D).
−
1
Fig. 11. Methyl acetate synthesis rates per Al atom as functions of (a) saturated CO adsorption infrared band areas of peaks centered at 2224 cm (L , Q) and at
1
− −1
196 cm (L , !) (spectra collected at 123 K and normalized per Si–O–Si overtone areas located at 2110–1760 cm ) and (b) CO/Al adsorption ratios determined
2
1
2
by CO temperature-programmed desorption of CO pre-adsorbed at 253 K [20 kPa DME, 930 kPa CO, 438 K].

P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
121
+
The observed increase in rate with increasing H content in
MOR samples suggests that proximity of surface CH3 groups
(
or trimethyloxonium) species increases carbonylation rates,
perhaps because of a vicinal CH3 group or merely because of
the consequent implicit presence of a neighboring Al, leading
to a more reactive CH3 group. For random Al site distribu-
tions [50], this cannot explain differences in reactivity among
−
1
−1
−1
MOR (∼1 [mol g-atom Al h ]), FER (∼0.1 [mol g-atom
−
1
−1
−1
Al h ]), and MFI (∼0.01 [mol g-atom Al h ]) with sim-
ilar Si/Al ratios. All zeolites give DMEadsorbed/Al ratios of
0
.5 ± 0.05; thus the proximity of CH3 groups could not dif-
fer for random Al siting. The relative distribution of α, β, and
γ sites from the UV–visible spectra of Co-exchanged materials
(
using methods described in Section 2.8) shows a predomi-
nance of β sites (>60%) in HMOR_9.8 and MFI_12.2 samples
and no specific preference for α or γ sites for MOR or MFI
materials [30–32], yet carbonylation rates on HMOR_9.8 and
HMFI_12.2 catalysts differ by ∼100, indicating that rates do
not change in parallel with the number of specific types of sites
+
Fig. 12. Methyl acetate synthesis rates per available H on Na-exchanged
samples (HMOR_9.8, NaMOR_0.27, NaMOR_0.41, NaMOR_0.55,
NaMOR_0.90) (") on samples with different Si/Al ratios (HMOR_6.0,
HMOR_9.8, HMOR_46) (2) and on chemically dealuminated MOR_9.8
samples with different Al contents (Q) [20 kPa DME, 930 kPa CO, 438 K].
(
α, β, γ ). Alternatively, differences in reactivity may represent
the presence of H2O concentrations that strongly inhibit car-
bonylation rates.
heterogeneity in CH groups based on steric confinement in mi-
cropores for MOR samples with low Si/Al ratios versus CH3
3
A mechanism involving competitive adsorption of CO and
DME on the same type of site would lead to negative DME
reaction orders, in contradiction with the measured zero-order
kinetics in DME. Thus, CO bound competitively onto Brønsted
acid sites cannot be the active species in carbonylation reac-
tions. CO bound onto framework Al with methyl groups already
bound would remove this inconsistency, as long as it occurred
non-competitively (proposal (ii)); water inhibition of reactions
groups accessible via mesopores in dealuminated MOR materi-
als with higher Si/Al ratios or via larger three-dimensional pore
structures in MFI and USY. These specific requirements remain
the subject of intense interest and additional studies within our
group.
4
. Conclusion
of CO with CH species would then reflect coordination of wa-
ter to Al species vicinal to CH3 groups, which interferes with
the docking of active CO species.
3
DME carbonylation occurs with high selectivity (∼99%)
on acidic mordenites at low temperatures (423–463 K) with
first- and zero-order kinetic dependencies on CO and DME
pressures, respectively. Induction periods are removed by pre-
treatment in DME because DME reacts initially with acidic
protons to form methyl groups and methanol and/or water;
the latter byproducts reversibly inhibit CO adsorption or re-
actions of CO with methyl groups to form acetyl intermedi-
ates in kinetically relevant steps. DME dissociative adsorption
occurs reversibly to form surface methyl groups, and surface
acetyl groups react with DME to re-form surface methyl groups
during steps that lead to methyl acetate formation. Acidic
mordenites with similar Si/Al ratios catalyze methyl acetate
synthesis at varying rates. The concentrations of CO bind-
ing sites as measured by low-temperature IR and TPD studies
in these materials did not correlate with DME carbonylation
rates, and their low levels (∼0.01 per Al) cannot justify the
∼55% excess methyl acetate/Al formed in transient reactions
of DME and CO. Methyl acetate synthesis rates per Al in-
crease with increasing proton concentration, suggesting that
aluminum pair sites occupied by either methyl groups of Brøn-
sted sites may be required for CO reactions with CH3 groups
or that CH3 groups in sterically confined environments of low-
silica mordenites may have higher reactivity for carbonyla-
tion.
Proposal (iii) would require CO binding on sites other than
(e.g., Lewis acid sites that bind CO at low temperatures).
Such sites may exist as extra-framework Al [49], but DME
carbonylation rates (per total Al) did not change systemati-
cally with their concentration (Section 3.6; Figs. 10 and 11).
+
H
CO binds on L -type Lewis acid sites with a heat of adsorp-
1
tion of ∼130 kJ/mol [44], which would lead to saturation of
sites with CO surfaces during catalysis at 423 K and 1 MPa
CO and thus to zero-order CO kinetic dependences. Simi-
lar arguments would render any CO species with detectable
spectral features an unlikely candidate as a reactive interme-
diate. The ability of most and perhaps all CH3 groups to
form acetyl groups (Section 3.4; Fig. 5) indicates that sites
that bind reactive centers are not minority species, because
the reactivity of most CH3 groups would require mobility of
acetyls or CH3 to place them near such minority sites. We
have ruled out such mobility by showing that physical mix-
tures of HMOR_9.8 and HMFI_12.2 (DMEadsorbed/Altotal
.5) have similar carbonylation rates (per Al in HMOR_9.8)
=
0
during transient CH3-CO reactions; thus, sites in MOR cannot
be used to carbonylate CH3 groups on HMFI_12.2 sites located
1 µm away, as would have been expected for mobile acetyls
>
or methyls.

122
P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
Acknowledgments
The rate of methyl acetate formation is
ꢂ
ꢃ
∗
b
rCH COOCH = k6 COCH3 PDME,
The authors thank BP for its Methane Conversion Coopera-
tive Research Program at the University of California at Berke-
ley and graciously acknowledge Drs. Theo Fleisch, Sander Gae-
mers, and Ben Gracey of BP for providing helpful technical
discussions and supplying Sudchemie, Tosoh, and Zeolyst mor-
denite samples.
3
3
which, after substitution, becomes
K1k5PCOCT,a
= 1
+ K1PCO + K2PH O
2
ꢀ
ꢁ
1
×
.
PH
O
1 5
CO T,a
K k P C
2
1
+
+
k PDME(1+K P +K PH
Supplementary material
K P
4
O
)
CH OH
6
1
CO
2
3
2
In the presence of few surface-bound CO and the absence of
H2O, the rate of methyl acetate synthesis becomes
Supplementary material for this article may be found on Sci-
enceDirect, in the online version.
Please visit DOI: 10.1016/j.jcat.2006.09.020.
rCH COOCH = K1k5PCOCT,a.
3
3
Appendix A. Derivation for the kinetic equations
described in Section 3.1
References
The concentrations of *aCO and *aH2O are solved using
quasi-equilibrium approximations in terms of ( ), the concen-
tration of unspecified CO binding sites,
*a
[1] G.J. Sunley, D.J. Watson, Catal. Today 58 (2000) 293.
[2] M.J. Howard, M.D. Jones, M.S. Roberts, S.A. Taylor, Catal. Today 18
1993) 325.
(
ꢂ
ꢃ
[
3] P.M. Maitlis, A. Haynes, G.J. Sunley, M.J. Howard, J. Chem. Soc. Dalton
Trans. (1996) 2187.
∗
a
∗a
CO = K1PCO
[
[
4] F.E. Paulik, J.F. Roth, Chem. Commun. (1968) 1578.
5] J.F. Roth, J.H. Craddock, A. Herschman, F.E. Paulik, Chemtech (1971)
and
ꢂ
ꢃ
∗
a
600.
H2O = K2PH O
.
[6] N. Yoneda, Y. Shiroto, K. Hamato, S. Asaoka, T. Maejima, US 5334755,
a
CT,a is the sum of ( ) and the concentration of the species
1994, Chiyoda.
bound to *
a
[7] H. Koch, Fette Seifen Anstrichmittel 59 (1957) 493.
,
[
[
8] A. Bagno, J. Bukala, G.A. Olah, J. Org. Chem. 55 (1990) 4284.
9] H. Bahrmann, New Syntheses with Carbon Monoxide, Springer-Verlag,
Berlin, 1980, p. 372.
ꢂ ꢃ
ꢂ
ꢂ
ꢃ
∗
a
∗a
∗a
CT,a =
+
ꢂ ꢃ
∗a
=
2
[10] M.V. Luzgin, V.N. Romannikov, A.G. Stepanov, K.I. Zamaraev, J. Am.
Chem. Soc. 118 (1996) 10890.
11] M.V. Luzgin, A.G. Stepanov, A. Sassi, J. Sommer, Chem. Eur. J. 6 (2000)
2368.
12] A.G. Stepanov, M.V. Luzgin, A.V. Krasnoslobodtsev, V.P. Shmachkova,
N.S. Kotsarenko, Angew. Chem. Int. Ed. 39 (2000) 3658.
[13] A.G. Stepanov, M.V. Luzgin, V.N. Romannikov, V.N. Sidelnikov, K.I.
Zamaraev, J. Catal. 164 (1996) 411.
a
The concentration of CO becomes
[
ꢂ
ꢃ
∗
a
CO =
.
[
1
2
b
*b
The concentration of H in terms of ( CH3) is determined by
the quasi-equilibrium approximation,
[
14] A.G. Stepanov, M.V. Luzgin, V.N. Romannikov, K.I. Zamaraev, J. Am.
∗
b
Chem. Soc. 117 (1995) 3615.
ꢂ
ꢃ
P
( CH )
∗
b
H2O 3
H =
.
[15] T. Li, N. Tsumori, Y. Souma, Q. Xu, Chem. Commun. (2003) 2070.
[16] Q. Xu, S. Inoue, N. Tsumori, H. Mori, M. Kameda, M. Tanaka, M. Fuji-
wara, Y. Souma, J. Mol. Catal. A Chem. 170 (2001) 147.
K4PCH OH
3
The pseudo-steady-state approximation is used to solve for
*
b
[17] K. Fujimoto, T. Shikada, K. Omata, H. Tominaga, Chem. Lett. (1984)
(
COCH3),
2047.
∗
b
ꢂ
ꢃꢂ
ꢃ
ꢂ∗b
ꢃ
[
18] B. Ellis, M.J. Howard, R.W. Joyner, K.N. Reddy, M.B. Padley, W.J. Smith,
d( COCH3)
dt
∗
∗b
=
k5 aCO
CH3 − k6 COCH3 PDME ≈ 0,
Stud. Surf. Sci. Catal. 101 (1996) 771.
[19] W.J. Smith, United Kingdom 5420345, 1995, BP Chemicals Limited.
[20] A. Sardesai, S. Lee, T. Tartamella, Energ. Source 24 (2002) 301.
[21] G.G. Volkova, L.M. Plyasova, A.N. Salanov, G.N. Kustova, T.M. Yurieva,
V.A. Likholobov, Catal. Lett. 80 (2002) 175.
∗
a
∗b
ꢂ
ꢃ
k ( CO)( CH )
∗
b
5
3
COCH3 =
,
k6PDME
and the sum of all species bound at Brønsted acid sites becomes
[
22] G.G. Volkova, L.M. Plyasova, L.N. Shkuratova, A.A. Budneva, E.A. Pauk-
shtis, M.N. Timofeeva, V.A. Likholobov, Stud. Surf. Sci. Catal. 147 (2004)
ꢂ
ꢃ
ꢂ
ꢃ
ꢂ
ꢃ
∗
b
∗b
∗b
CT,b =
CH3 + H + COCH3
403.
ꢄ
ꢅ
∗
a
ꢂ
ꢃ
P
H2O
k ( CO)
[23] R.W. Wegman, J. Chem. Soc. Chem. Commun. (1994) 947.
[24] P. Cheung, A. Bhan, G.J. Sunley, E. Iglesia, Angew. Chem. Int. Ed. 45
(2006) 1617.
∗
bCH3 1 +
5
=
+
.
K4PCH OH
k6PDME
3
[25] M.R. Apelian, A.S. Fung, G.J. Kennedy, T.F. Degnan, J. Phys. Chem. 100
The acetyl concentration is given by
(1996) 16577.
ꢀ
ꢁ
∗
a
ꢂ∗
ꢃ
k ( CO)
C
[26] M. Muller, G. Harvey, R. Prins, Micropor. Mesopor. Mater. 34 (2000) 135.
[27] A.P.M. Kentgens, K.F.M.G.J. Scholle, W.S. Veeman, J. Phys. Chem. 87
(1983) 4357.
b
5
T,b
COCH3 =
.
PH
O
∗a
k6PDME
2
k5( CO)
k6PDME
1
+
+
K4PCH OH
3

P. Cheung et al. / Journal of Catalysis 245 (2007) 110–123
123
[
28] E. Iglesia, J. Baumgartner, G.L. Price, K.D. Rose, J.L. Robbins,
[39] R. Shah, J.D. Gale, M.C. Payne, J. Phys. Chem. B 101 (1997) 4787.
[40] N. Hartz, G. Rasul, G.A. Olah, J. Am. Chem. Soc. 115 (1993) 1277.
J. Catal. 125 (1990) 95.
[
[
[
29] G.L. Price, E. Iglesia, Ind. Eng. Chem. Res. 28 (1989) 839.
30] J. Dedecek, B. Wichterlova, J. Phys. Chem. B 103 (1999) 1462.
31] J. Dedecek, D. Kaucky, B. Wichterlova, O. Gonsiorova, Phys. Chem.
Chem. Phys. 4 (2002) 5406.
[41] O. Kresnawahjuesa, R.J. Gorte, D. White, J. Mol. Catal. A Chem. 208
(2004) 175.
[42] A. Zecchina, C.O. Arean, Chem. Soc. Rev. 25 (1996) 187.
[43] J.A. Lercher, C. Grundling, G. Eder-Mirth, Catal. Today 27 (1996) 353.
[44] L. Benco, T. Bucko, J. Hafner, H. Toulhoat, J. Phys. Chem. B 108 (2004)
13656.
[32] L. Drozdova, R. Prins, J. Dedecek, Z. Sobalik, B. Wichterlova, J. Phys.
Chem. B 106 (2002) 2240.
[
[
33] T.R. Forester, R.F. Howe, J. Am. Chem. Soc. 109 (1987) 5076.
34] S.M. Campbell, X.Z. Jiang, R.F. Howe, Micropor. Mesopor. Mater. 29
[45] M.I. Zaki, H. Knozinger, Spectrochim. Acta A 43 (1987) 1455.
[46] H. Knozinger, P. Ratnasamy, Catal. Rev. Sci. Eng. 17 (1978) 31.
[47] S. Bordiga, C. Lamberti, F. Geobaldo, A. Zecchina, G.T. Palomino, C.O.
Arean, Langmuir 11 (1995) 527.
[48] M. Maache, A. Janin, J.C. Lavalley, E. Benazzi, Zeolites 15 (1995) 507.
[49] M.A. Makarova, A.E. Wilson, B.J. van Liemt, C. Mesters, A.W. de Winter,
C. Williams, J. Catal. 172 (1997) 170.
(1999) 91.
[35] D. Lesthaege, V. Van Speybroeck, G.B. Marin, M. Waroquier, Chem.
Phys. Lett. 417 (2006) 309.
[36] W. Wang, A. Buchholz, M. Seiler, M. Hunger, J. Am. Chem. Soc. 125
(2003) 15260.
[
[
37] W. Wang, M. Seiler, M. Hunger, J. Phys. Chem. B 105 (2001) 12553.
38] I.I. Ivanova, A. Corma, J. Phys. Chem. B 101 (1997) 547.
[50] B.R. Goodman, K.C. Hass, W.F. Schneider, J.B. Adams, Catal. Lett. 68
(2000) 85.