Dimethyl carbonate production via the oxidative carbonylation of methanol over Cu/SiO2 catalysts prepared via molecular precursor grafting and chemical vapor deposition approaches

Article abstract of DOI:10.1016/j.jcat.2004.10.001

The influence of catalyst synthesis method and Cu source on the activity and selectivity of Cu/SiO₂ catalysts for the gas-phase oxidative carbonylation of methanol to dimethyl carbonate (DMC) is reported. [CuOSi(O ^tBu)₃]₄, [CuO ^tBu]₄, and CuCl were used as precursors to produce highly dispersed silica-supported copper. XANES and EXAFS characterization prior to reaction (but after thermal treatment under He) showed that Cu in the catalysts prepared with CuCl and [CuOSi(O ^tBu)₃]₄ was present primarily as isolated Cu(I) species, whereas [CuO ^tBu]₄ produced 1-nm Cu particles. During the catalytic reaction, the Cu in catalysts prepared from CuCl and [CuOSi(O ^tBu)₃]₄ formed highly dispersed CuO moieties, whereas the Cu in catalysts prepared from [CuO ^tBu]₄ formed a cuprous oxide layer over a Cu(0) core. For comparison, poorly dispersed Cu on silica was prepared via traditional incipient wetness impregnation with Cu(NO₃)₂. It was found that activity for DMC formation increased with increasing Cu dispersion. The selectivity for DMC formation (relative to CO) decreased with decreasing Cu dispersion when the original state of the Cu was Cu(0) directly preceding reaction conditions.

Full text of DOI:10.1016/j.jcat.2004.10.001

Journal of Catalysis 230 (2005) 14–27
www.elsevier.com/locate/jcat
Dimethyl carbonate production via the oxidative carbonylation
of methanol over Cu/SiO catalysts prepared via molecular
2
precursor grafting and chemical vapor deposition approaches
a
b,c
a,∗
b,c
Ian J. Drake , Kyle L. Fujdala , Alexis T. Bell , T. Don Tilley
a
Department of Chemical Engineering, University of California, Berkeley, CA 94720-1462, USA
b
Department of Chemistry, University of California, Berkeley, Berkeley, CA 94720, USA
c
Chemical Sciences Division, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720, USA
Received 27 July 2004; revised 27 September 2004; accepted 1 October 2004
Available online 28 December 2004
Abstract
The influence of catalyst synthesis method and Cu source on the activity and selectivity of Cu/SiO catalysts for the gas-phase oxidative
2
t
t
carbonylation of methanol to dimethyl carbonate (DMC) is reported. [CuOSi(O Bu) ] , [CuO Bu] , and CuCl were used as precursors to
3
4
4
produce highly dispersed silica-supported copper. XANES and EXAFS characterization prior to reaction (but after thermal treatment under
t
He) showed that Cu in the catalysts prepared with CuCl and [CuOSi(O Bu) ] was present primarily as isolated Cu(I) species, whereas
3
4
t
t
[
CuO Bu] produced 1-nm Cu particles. During the catalytic reaction, the Cu in catalysts prepared from CuCl and [CuOSi(O Bu) ]
4
3
4
t
formed highly dispersed CuO moieties, whereas the Cu in catalysts prepared from [CuO Bu] formed a cuprous oxide layer over a Cu(0)
4
core. For comparison, poorly dispersed Cu on silica was prepared via traditional incipient wetness impregnation with Cu(NO ) . It was found
3
2
that activity for DMC formation increased with increasing Cu dispersion. The selectivity for DMC formation (relative to CO) decreased with
decreasing Cu dispersion when the original state of the Cu was Cu(0) directly preceding reaction conditions.

2004 Elsevier Inc. All rights reserved.
Keywords: DMC synthesis; Oxidative carbonylation; Cu EXAFS; Cu K-edge XANES; Cl K-edge XANES; Molecular precursor
1
. Introduction
These include cycloaddition of CO2 to epoxides with the
use of titanosilicate molecular sieves [4], various transes-
terification methods [5,6], methanol carboxylation over zir-
conia [7–9], and electrochemical oxidative carbonylation of
methanol [10,11]. From a thermodynamic standpoint, the
oxidative carbonylation of methanol remains the most favor-
able reaction [1]. To overcome problems with catalyst sepa-
ration inherent in liquid-phase processes, recent interest has
focused on vapor-phase oxidative carbonylation of methanol
over supported copper-based catalysts [12–27]. CuCl and
CuCl2 supported on activated carbon have been evaluated
by a number of groups because of the inherent similar-
ity of such catalysts to those used in the liquid-phase sys-
tem [12–18]. CuCl, CuCl2, and bimetallic PdCl2-CuCl2 de-
posited on mesoporous silica supports (HMS silica, MCM-
41, and SBA-15) have also been evaluated [20–22]. Addi-
Dimethyl carbonate (DMC) has considerable potential as
both a chemical intermediate and a fuel additive [1]. It can
also be used as a precursor for carbonic acid derivatives and
as a methylating agent [2]. Because of its high oxygen con-
tent, DMC has been proposed as a replacement for methyl
tert-butyl ether (MTBE) as a fuel additive [1]. DMC is cur-
rently made via the oxidative carbonylation of methanol on
unsupported cuprous chloride suspended in a slurry reac-
tor [3]. Since this process suffers from catalyst deactivation,
equipment corrosion, and difficulties in product separation,
a number of alternative approaches have been investigated.
*
Corresponding author. Fax: +1 510 642 4778.
E-mail address: bell@cchem.berkeley.edu (A.T. Bell).
0021-9517/$ – see front matter  2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2004.10.001

I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
15
tional work has been reported on increasing the activity and
selectivity of the above-mentioned catalysts by the incorpo-
ration of quaternary ammonium salts such as tetrabutylam-
monium bromide (TBAB) and tetraelthylammonium chlo-
ride (TEAC) [22,23]. Finally, Cu-exchanged X and Y zeo-
lites, which are nearly free of chloride ions, have been found
to be active for DMC synthesis [24–27]. Although these
studies demonstrate the importance of Cu as an essential el-
ement, the nature of the active site required to achieve high
activity and selectivity is not well understood. In particular,
there is little understanding of the oxidation state of Cu and
the dispersion of the Cu-containing species under reaction
conditions. The aim of the present work was to investigate
the performance of well-characterized catalysts prepared by
the dispersion of Cu on the mesoporous silica SBA-15. [Cu-
of the Cu deposition methods have been described previ-
ously [36,37]. After preparation, each catalyst was heated in
He at 473 K to remove all organic matter from the samples
t
t
prepared from either [CuOSi(O Bu)3]4 or [CuO Bu]4, and
for consistency for the samples prepared from CuCl. Cata-
t
t
lysts prepared from CuOSi(O Bu)3]4, [CuO Bu]4, or CuCl
t
are designated as CuOSi/SBA(x.x), CuO Bu/SBA(x.x), and
CuCl/SBA(x.x), respectively, where x.x indicates the weight
loading of Cu determined after He pretreatment by ICP.
Designations without the x.x suffix refer to all catalysts of
a single Cu precursor. A 5.0 wt% Cu on SBA-15 catalyst
was also prepared via aqueous impregnation of Cu(NO3)2.
The freshly impregnated material was dried under ambient
conditions, oven dried at 393 K for 5 h, and then calcined
in air at 773 K for 2 h. Following calcinations to remove
the nitrogen-containing species, the catalyst was reduced in
CO at 773 K for 2 h. The reduced catalyst is designated as
Cu/SBA(5.0).
t
t
OSi(O Bu)3]4, [CuO Bu]4, and CuCl were used as Cu pre-
cursors, since all three were found to produce well-dispersed
Cu moieties on SBA-15. XANES and EXAFS were used in
situ to follow changes in the oxidation state and structure of
the supported Cu.
2.3. Cu K-edge X-ray absorption spectroscopy (XAS)
EXAFS and XANES data were acquired at the Stan-
ford Synchrotron Radiation Laboratory (SSRL) at Stanford
University and at the National Synchrotron Light Source
2
. Experimental
(
NSLS) at Brookhaven National Laboratory. Cu K-edge ab-
2
.1. General
sorption measurements were performed on beamline 6-2 at
SSRL and on beamline X11A at NSLS. The premonochro-
mator vertical aperture of the beams was set to 0.5 mm for
improved resolution, defining an energy resolution of 1.8 eV
at the Cu K-edge. The monochromator was detuned 20–30%
at 400 eV above the Cu K-edge. Cu metal foil (7 µm) was
used for energy calibration.
All standards and test samples were handled under an in-
ert atmosphere (N2) in a drybox, with the exception of CuO
and CuCl2, which were handled in air. A Quantachrome
Autosorb-One was used for N2 porosimetry measurements.
The Brunauer–Emmet–Teller (BET) method [28] was used
to determine surface areas, and the Barrett–Joyner–Halenda
Each sample was pressed into a rectangular pellet (0.43×
(
BJH) method [29] was used to obtain pore size distribu-
1
.86 cm) and loaded into an in situ cell for transmission ex-
tions. The Cu and Cl (where appropriate) contents of the
materials studied were determined by Galbraith Laborato-
ries (Knoxville, TN, USA) with ionization-coupled plasma
periments [38]. A sufficient quantity of each sample was
used to give a calculated absorption length (µmρx) of 2
−6
[
39–42]. The cell was evacuated to 10 Torr, and the sam-
(
ICP) methods. Si(OEt)4, Cu2O, CuO, Cu(NO3)2, CuCl,
ple holder was maintained at 77 K during the experiment.
Intensities of the beam were measured over a 900-eV range.
The EXAFS region was taken from 30 eV to 12k in 0.05k
steps and from 12k to 16k in 0.07k steps holding for 2 and
CuCl2, and methanol (MeOH) were purchased from Aldrich
and used as received. Chlorotriphenylsilane, ClSi(C6H5)3,
was purchased from Gelest and used as received. Cs2CuCl4
was prepared by known literature procedures [30,31]. The
mesoporous silica support, SBA-15, was prepared accord-
ing to literature procedures [32] and characterized by PXRD
3
s, respectively. Ionization chambers (N2 filled) were used
to measure the incident (I0) and sample transmitted (I1)
fluxes. A third detector (I2) was used to measure the flux
through a (7-µm thick) Cu foil internal standard.
2
−1
and N2 porosimetry (surface area, 894 m g ; pore vol-
3
−1
ume, 1.13 cm g ; average pore diameter, 77 Å). The
hydroxyl group concentration of the SBA-15 was deter-
mined to be 1.3(1) OH nm , via reaction of SBA-15 with
A portable flow manifold was used to treat the catalysts
on-site. Each sample was first treated in He (99.999%) at a
−2
3
−1
flow rate of 60 cm min . The temperature was increased
Mg(C6H5CH2)2 ·2THF and quantification of the toluene
−1
at 10 K min to 573 K and kept isothermal for 2 h to fa-
1
evolved by H NMR spectroscopy [33].
cilitate removal of residual organic species from the grafted
materials. This protocol has been shown to be sufficient to
remove the organic species. For consistency, this pretreat-
ment was also applied to all CuCl/SBA(x.x) samples. Fol-
2
2
.2. Catalyst preparation
Cu was dispersed onto SBA-15 either by room-tempera-
lowing pretreatment in He, reaction gases (CO, O , MeOH)
t
t
ture grafting of [CuOSi(O Bu)3]4 [34] or [CuO Bu]4 [35],
or by reaction of CuCl vapor at elevated temperature. Details
were introduced into the cell. A MeOH/O /CO/He mixture
2
(4.0/1.0/9.0/19.3) was fed to the cell at a nominal total flow

16
I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
3
−1
rate of 20 cm min . The feed gas was passed over the cat-
alyst for 45 min at 363 K. After this step, the temperature
the energy is on the order of tens of meV [47]. The CuCl
edge did not change over the course of the experiment.
−1
was increased at 10 K min to 403 K and maintained at this
temperature for 1 h. During in situ XANES measurements,
scans were taken every 4–6 min.
2.5. XANES analyses
Cu and Cl K-edge XANES data were analyzed with the
WinXAS (v. 2.3) software package [48,49]. The energy was
calibrated with the use of the Cu K-edge of the Cu foil,
which was taken as 8980 eV. Pre-edge absorptions due to
the background and detector were subtracted with the use of
a linear fit to the data in the range of −200 to −50 eV, rel-
ative to the sample edge energy (E0). Each spectrum was
normalized by a constant determined by the average absorp-
tion in the range of 100 to 300 eV relative to E0. The edge
energy of each sample and reference was taken at the first
inflection point beyond any pre-edge peaks.
Prior to analysis of the Cl K-edge data, a linear fit was
made to the data in the range of −120 to −20 eV rela-
tive to E0, and this line was subtracted from the data. The
spectrum of each sample was normalized by a constant de-
termined by the average absorption in the range of 2840–
2
.4. Cl K-edge X-ray absorption measurements
Cl K-edge XANES measurements were performed on
beamline 9.3.1 of the Advanced Light Source (ALS) at
the Lawrence Berkeley National Laboratory. The beamline
has a double-crystal monochromator and was equipped with
Si(111) crystals. Focusing optics were present before and
after the monochromator. A focused beam spot of approx-
imately 2 mm was measured at the sample. The premono-
chromator vertical slits were adjusted to define the resolution
and flux. The pre-edge feature of CuCl2 was monitored at
four different vertical slit separations to optimize the en-
ergy resolution. A resolving power of approximately 4000
(
0
E/ꢀE) was achieved, which corresponds to approximately
.5 eV resolution at the Cl K-edge. The beamline deliv-
2
860 eV. Energies reported for the inflection point were de-
11
−1
ers more than 10 photons s between 2500 and 5000 eV.
Because the flux drops precipitously above 5000 eV, no de-
tuning of the monochromator was necessary, since the first
allowed harmonic occurs above 8000 eV and has a negligi-
ble flux. The ALS storage ring operates at 1.9 GeV, and data
were taken at a current between 380 and 200 mA.
termined by taking the highest energy maximum in the first
derivative of the data in the rising edge region. A pseudo-
Voigt line shape set to a fixed 50:50 ratio of Lorentzian–
Gaussian contributions was used to model the pre-edge fea-
tures [43]. The intensity of the pre-edge feature is reported as
the peak height multiplied by the full width at half-maximum
Samples were loaded on 1 cm × 1 cm plates and loaded
(
FWHM). A detailed error analysis was not pursued.
into the sample chamber of the end station, which was op-
−7
erated at 10 Torr. No windows isolated the end station
from the beamline in normal operation. A silicon photodi-
ode (Hamamatsu model 3584-02) detector could be maneu-
vered at a 45 angle relative to the incident radiation within
2
2
.6. EXAFS analyses
Extraction and fitting of the EXAFS function, χ(k), were
◦
done by standard methods with the aid of the UWXAFS
50–52] suite of software programs. We subtracted a back-
–5 mm of the sample face to measure X-ray and visible flu-
[
orescence. An electrometer (Keithley 6517A) was used to
amplify the photodiode current measurement.
ground function by taking spline points between k values
−1
of 1.5 and 15 Å . An Rbkg value of 1.0 was chosen. Non-
Cl K-edge XANES data were collected for standards
phase-corrected Fourier transforms (FT) were performed on
(CuCl, CuCl2, Cs2CuCl4, and ClSi(C6H5)3), a sample of
3
3
the k -weighted χ(k) functions. FT k χ(k) data are plotted
without phase correction unless explicitly identified as such.
The k range of the transform varied between kmin = 2.0–
newly prepared CuCl/SBA(2.9), and the same material af-
ter it had been He pretreated and then exposed to reaction
conditions. XANES data were also obtained for SBA-15 that
had been impregnated with HCl (from a 0.12 M HCl(aq) so-
lution) and then dried at 393 K for 2 h before heating in a
vacuum oven at 423 K for 2 h. A thin pellet of each sam-
ple (∼ 15 mg) was pressed and applied to the adhesive side
of pre-cut Kapton tape, which was subsequently mounted on
the sample holder. All samples were exposed to ambient con-
ditions for ∼ 5 min during transfer from a vial to the sample
holder of the end station. Three scans of each standard and
five scans of each test sample were measured for an adequate
signal-to-noise ratio of the data. The energy was calibrated
from the Cl K-edge of Cs2CuCl4 [43–46]. The maximum of
the first edge-region feature in the spectrum was assigned to
−
1
−1
3
.0 Å and kmax = 11.5–14.0 Å for all spectra. The kmin
and kmax values were chosen at node points to help mini-
mize spectral broadening. All spectra were transformed with
−1
a Hanning window function with a windowsill (dk) of 1 Å
centered on the chosen nodal position.
2
3
S was extracted by fitting the first peak in FT k χ(k)
0
for Cu foil, with the use of the theoretical values of F (k)
j
and φ (k) determined by the FEFF8.2 code [53,54]. The fit
j
was done in back FT k-space and in R-space. A value of
2
0
S = 0.865 was obtained in both refinements, and this value
was used in all subsequent analyses.
All fits of samples and standards were performed on the
3
real and imaginary parts of the FT k χ(k) data. The k and R
2
820.20 eV. Scans were made between 2700 and 2923 eV
ranges of all of the sample fits and the number of indepen-
dent points available to minimize the sum of squares of the
with a 0.1-eV step in the edge region. The reproducibility of

I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
17
Table 1
ionization detector (FID) and by a packed column (Alltech,
Haysep DB packing) connected to a thermal conductivity
detector (TCD). Oxygenated compounds (MeOH, methyl
formate, dimethoxymethane, dimethyl ether, and formalde-
hyde) were detected with the FID, whereas fixed gases (CO,
O2, CO2) were detected with the TCD. Complete GC prod-
uct analysis run time was ca. 30 min. Three points were
taken at each temperature or residence time and checked for
reproducibility.
k-Space and r-space range for fitting
a
Catalyst
k
kmax
(Å
ꢀk
R
Rmax
(Å)
ꢀR N
ind
min
min
−
1
−1
(Å
)
)
(Å)
CuCl/SBA(2.9)
2.2
2.9
11.0
12.3
12.8
8.8 1.2
9.4 1.2
9.8 1.1
3.0
3.0
3.0
1.8 12.1
1.8 12.8
1.9 13.9
CuOSi/SBA(3.6)
t
CuO Bu/SBA(3.4) 3.0
a
Number of independent data points (N ) in fit calculated by N
=
ind
ind
2(ꢀk)(ꢀR)/π + 2.
difference between the model and χ(k) data are listed in Ta-
ble 1. Two to three shells were fit simultaneously, allowing
σ, Rj , and Nj to be adjusted for each Fourier component.
A single edge shift correction (ꢀE0) was refined for each fit,
unless reported otherwise. The quality of a particular fit was
3. Results
3.1. General
2
evaluated by use of the reduced chi-square method (χ ) [55].
The surface area and pore volume of freshly prepared Cu-
OSi/SBA(3.6) (678 m g and 0.88 cm g ) were lower
than those of SBA-15 (894 m g and 1.13 cm g ). How-
ever, after heating at 673 K under flowing N (to remove
ν
2
−1 −1
3
Once a best fit was found, the ꢀ-factor was recorded for
presentation of the fits. This factor gives a sum-of-square
measure of the fractional misfit [55]. More extensive details
of the EXAFS analysis performed in this work, including
details of the phase and amplitude functions used in the fits,
can be found elsewhere [36,37].
2
−1
3
−1
2
the organic species), the surface area and pore volume of
2
−1
3
−1
CuOSi/SBA(3.6) increased (800 m g and 1.01 cm g ),
indicating that pore filling by the bulky molecular precur-
sor is responsible for the initial decreases. Similar behavior
t
2
.7. Studies of catalytic activity and selectivity
was observed for CuOSi/SBA(5.2), CuO Bu/SBA(3.4), and
t
CuO Bu/SBA(4.3) [56].
3
Catalyst (150 mg) (typically 1.2 cm ) was loaded into a
Reaction of SBA-15 with varying amounts of CuCl at
873 K produced a light yellow powder for all Cu load-
ings. The surface area and pore volume of freshly prepared
CuCl/SBA(2.9) (590 m g and 0.76 cm g ) were lower
than the values found for the SBA-15 support (894 m g
and 1.13 cm g ), indicative of pore collapse during the
catalyst preparation. Helium pretreatment and exposure to
reaction conditions did not affect the surface area or pore
volume of any catalysts studied.
1
0-mm-diameter tubular flow reactor made of Pyrex. A glass
frit fused inside the reactor was used to support the cata-
lyst. The reactor was heated in a tubular furnace. The reactor
thermocouple and a pressure transducer, multiple solenoid
valves, and all system thermocouples were interfaced to a
personal computer via an analog-to-digital/digital-to-analog
board (NI PCI-6052E). A LabVIEW program was devel-
oped to monitor and control the reactor temperature and the
pressure in the reactor and to switch gases. Each catalyst
was pretreated at 573 K for 1 h in a stream of high-purity
He (99.999%). For the catalytic experiment, CO (99.9%,
2
−1
3
−1
2
−1
3
−1
Table 2 lists the actual and targeted Cu contents for each
t
of the catalysts. It is evident that using [CuO Bu] or [Cu-
4
t
OSi(O Bu) ] as the Cu source provides Cu loadings close
3 4
0
.1% Ar), O2 (20.0%, 80.0% He), and He (99.999%) were
to the targeted values. In contrast, using CuCl led to di-
vergence of the targeted and actual Cu loadings with in-
creasing Cu loading. These data indicate that Cu deposition
used together with methanol (MeOH), which was intro-
duced by passage of the He carrier gas through a satura-
tor maintained at 306 K. CO, O2, and the MeOH-saturated
He stream were mixed in a mixing volume filled with 1.0-
mm quartz beads before they were passed into the reactor.
A MeOH/O2/CO/He mixture (4.0/1.0/9.0/19.3) was fed
t
using the low-temperature grafting of [CuO Bu] and [Cu-
4
t
OSi(O Bu) ] is much more efficient when compared with
3 4
high-temperature CVD using CuCl. For example, only a Cu
loading of 2.92% could be achieved despite the use of suffi-
cient CuCl for a Cu loading of 7%. Furthermore, use of CuCl
introduced Cl into the final catalyst (between 0.1 and 0.35%
as the Cu loading increased from 1.1 to 2.9%).
3
−1
to the reactor at a nominal total flow rate of 20 cm min
,
3
−1
which was varied between 10 and 40 cm min to study the
effect of residence time. The feed gas was passed over the
catalyst for 45 min at 363 K. After this step, the temperature
−1
was increased at 10 K min and held constant in 10 K incre-
ments between 383 and 423 K. Following reaction at 423 K,
the reactor was cooled to 393 K to check the reproducibil-
ity of catalyst performance. All lines were heated to avoid
condensation of methanol and products. The product stream
was analyzed by a gas chromatograph (GC) (Agilent 6890)
equipped with a capillary column (Alltech, AT aquawax;
polyethylene glycol stationary phase) connected to a flame
3.2. Catalytic performance
The activities of CuCl/SBA(2.9), CuOSi/SBA(3.6), and
t
CuO Bu/SBA(3.4) were compared at a constant reactor resi-
dence time of 3.5 s for temperatures between 383 and 423 K.
The only products observed were DMC, methyl formate
(MF), and CO . Fig. 1 presents the DMC, MF, and CO
2
2
activities of each catalyst reported as an apparent turnover

18
I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
Table 2
Materials analysis
Catalyst ID
Preparation method
CuCl sublimation
Cu
Targeted Cu
(wt%)
Before reaction
Cl (wt%)
After reaction
(wt%)
2
Cu/Cl (atom)
Cu (nm )
Cl (wt%)
CuCl/SBA(1.1)
CuCl/SBA(2.2)
CuCl/SBA(2.9)
1.14
2.17
2.92
1.5
3.5
7.0
0.08
0.18
0.35
7.8
6.7
4.7
0.14
0.26
0.35
0.10
0.18
0.33
CuOSi/SBA(1.4)
CuOSi/SBA(3.6)
CuOSi/SBA(5.2)
Non-aqueous grafting
Non-aqueous grafting
1.44
3.61
5.21
1.5
3.5
5.0
–
–
–
–
–
–
0.17
0.43
0.62
t
CuO Bu/SBA(1.4)
1.37
3.35
4.31
1.5
3.5
5.0
–
–
–
–
–
–
0.16
0.40
0.51
t
CuO Bu/SBA(3.4)
t
CuO Bu/SBA(4.3)
Fig. 1. Effect of precursor composition on catalyst activity. Catalysts are compared at a constant residence time of 3.5 s. Rates of (a) DMC, (b) MF, and (c)
3
−1
.
CO production. Reaction conditions: MeOH/O /CO/He gas mixture (4.0/1.0/9.0/19.3) was fed to the reactor at a nominal total flow rate of 20 cm min
2
2
frequency (TOF) normalized to the total number of moles
of Cu reported from data in Table 2. The DMC activity de-
creases in the order CuCl/SBA(2.9) > CuOSi/SBA(3.6) >
OSi/SBA(3.6) approached those of CuCl/SBA(2.9) with in-
creasing temperature. CuO Bu/SBA(3.4) was the least se-
lective catalyst for DMC formation from both methanol and
CO.
t
t
CuO Bu/SBA(3.4), whereas the MF activity decreases in
t
the order CuCl/SBA(2.9) > CuO Bu/SBA(3.4) > CuOSi/
Table 3 lists the apparent activation energies (Eapp) based
on the data in Fig. 1. The values of Eapp for DMC for
SBA(3.6). The rate of CO2 production decreases in the order
t
CuO Bu/SBA(3.4) > CuCl/SBA(2.9) > CuOSi/SBA(3.6).
t
CuCl/SBA(2.9), CuOSi/SBA(3.6), and CuO Bu/SBA(3.4)
DMC selectivity is reported on the basis of methanol con-
sumption (SDMC/MeOH) and on the basis of CO consump-
tion (SDMC/CO). The CO2 formed is attributed exclusively
to CO oxidation, since over the temperature range investi-
gated no evidence was found for the combustion of methanol
or products derived from methanol. CuCl/SBA(2.9) exhibits
the highest values of SDMC/MeOH and SDMC/CO at each tem-
perature. The values of SDMC/MeOH and SDMC/CO for Cu-
are 30, 34, and 25 kJ/mol, respectively. Whereas Eapp for
DMC does not vary significantly, larger differences are seen
in Eapp for MF and CO2. The values of Eapp for MF for
t
CuCl/SBA(2.9), CuOSi/SBA(3.6), and CuO Bu/SBA(3.4)
are 102, 80, and 90 kJ/mol, respectively. Last, the values
of Eapp for CO2 for CuCl/SBA(2.9), CuOSi/SBA(3.6), and
t
CuO Bu/SBA(3.4) are 72, 53, and 71 kJ/mol, respectively.

I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
19
t
The effects of Cu weight loading were investigated at
03 K for a constant residence time of 3.5 s. Turnover fre-
Three CuO Bu/SBA catalysts were also studied, with Cu
loadings of 1.4, 3.4, and 4.3 wt%. The DMC activities of
these catalysts based upon the total Cu content of each cata-
lyst are very similar to those of CuOSi/SBA, and both sets of
catalyst show a modest decrease in DMC TOF with increas-
ing Cu loading. The MF TOFs calculated on a total Cu basis
4
quencies and DMC selectivities are presented in Table 4. As
the Cu loading for CuCl/SBA increases from 1.1 to 2.9 wt%,
the DMC turnover frequency calculated on a total copper ba-
−1
sis stays nearly constant (3.5 s ). Over the same range of
Cu loadings, the TOFs for CO2 and MF formation remain
essentially constant. DMC selectivities are shown for resi-
dence times of 3.5 and 0 s. The latter value was obtained
by the extrapolation of data taken as a function of residence
time and represents the intrinsic selectivity exclusive of sec-
t
are also very similar for the CuOSi/SBA and CuO Bu/SBA
catalysts. A significant difference between all of the cata-
lysts is that the CO2 TOF is much higher for CuO Bu/SBA
than for either CuOSi/SBA or CuCl/SBA. Consistent with
t
ondary reactions. The values of So
and So
DMC/CO
DMC/MeOH
for CuCl/SBA show little dependence on Cu loading. In all
cases, the intrinsic selectivity for DMC from methanol is
very high, reaching a maximum value of 90%, whereas the
selectivity to DMC from CO is only slightly higher than
5
0%. Similar patterns are observed in the activity and se-
lectivity of CuOSi/SBA for Cu weight loadings of 1.4, 3.6,
and 5.2 wt%. For these catalysts, the intrinsic selectivity for
DMC from methanol does not exceed 83%, but the intrinsic
selectivity for DMC from CO is again near 50%.
Table 3
Apparent activation energies for CuCl/SBA(2.9), CuOSi/SBA(3.6), and
t
CuO Bu/SBA(3.4)
Catalyst
Eapp
DMC
MF
CO
2
(kJ/mol)
(kJ/mol)
(kJ/mol)
a
CuCl/SBA(2.9)
CuOSi/SBA(3.6)
CuO Bu/SBA(3.4)
30 ± 2
34 ± 2
25 ± 1
102± 5
80± 1
80± 4
72 ± 3
53 ± 5
71 ± 5
Fig. 2. Effect of catalyst composition on catalyst selectivity. Catalysts
are compared at a constant residence time of 3.5 s. (a) S
,
t
DMC/MeOH
(
b) S . Reaction conditions: MeOH/O /CO/He gas mixture
DMC/CO
2
a
ꢀ
t
95
error calculated using Arrhenius plots at three to four different
(4.0/1.0/9.0/19.3) was fed to the reactor at a nominal total flow rate of
3
−1
.
residence times.
20 cm min
Table 4
Performance comparison for all catalysts as a function of copper loading at 403 K
a
b
o
c
d
o
e
f
−1
−1) (×105)
Catalyst
S
S
S
S
TOF
(mol
mol
s
DMC/CO
DMC/MeOH
product
product
Cu
DMC/CO
DMC/MeOH
DMC
MF
CO
2
CuCl/SBA(1.1)
CuCl/SBA(2.2)
CuCl/SBA(2.9)
48.8
47.6
43.8
52.0
58.7
55.4
80.1
69.7
66.5
90.0
85.5
83.3
3.5
3.5
3.3
0.9
1.5
1.6
3.6
3.8
4.2
CuOSi/SBA(1.4)
CuOSi/SBA(3.6)
CuOSi/SBA(5.2)
38.2
36.2
34.5
44.2
49.5
49.5
71.4
62.1
52.3
82.9
79.4
72.7
2.1
1.7
1.5
0.8
1.1
1.4
3.4
3.0
2.8
t
CuO Bu/SBA(1.4)
17.9
18.3
14.5
22.3
–
18.3
65.9
51.8
51.9
80.5
–
67.0
2.0
1.5
1.3
1.1
1.4
1.2
9.3
6.5
7.8
t
g
CuO Bu/SBA(3.4)
t
CuO Bu/SBA(4.3)
Cu/SBA(5.0)
a
7.3
10.0
37.3
52.0
0.8
1.5
11.0
Reaction temperature, 403 K; CO, 0.27 atm; MeOH, 0.12 atm; O , 0.03 atm; residence time, 3.5 s.
2
b
c
S
= [DMC]/([DMC] + [CO ]).
DMC/CO
2
o
S
is the selectivity extrapolated to zero residence time.
DMC/CO
d
e
S
= [DMC]/([DMC] + [MF]).
DMC/MeOH
o
S
is the selectivity extrapolated to zero residence time.
DMC/MeOH
f
TOF is reported on the basis of total Cu.
g
o
o
S
and S
are not reported because the residence time was not varied.
DMC/MeOH
DMC/CO

20
I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
3
Fig. 3. Fourier transformed k χ(k) (non-phase corrected) for (a) CuCl/SBA(2.9), CuCl/SBA(2.2), and CuCl/SBA(1.1); (b) CuOSi/SBA(5.2) and Cu-
t
t
OSi/SBA(3.6); (c) CuO Bu/SBA(4.3) and CuO Bu/SBA(3.4). Catalysts were pretreated in He at 573 K prior to EXAFS analysis.
the above observations, the values of So
are signif-
be assigned to Cu–O backscattering, suggesting that the
DMC/CO
t
dominant sites are still Cu(I) species associated with O
icantly lower for CuO Bu/SBA catalysts when compared
with those for the other two sets of catalysts; however, the
3
[
35–37]. The FT k χ(k) spectra of CuOSi/SBA(3.6) and
values of So
remain relatively high, although they
CuOSi/SBA(5.2) also exhibit peaks at 2.11, 3.34, 4.1, and
.7 Å (associated with Cu metal), but they are signifi-
cantly smaller in magnitude than those observed in the
DMC/MeOH
4
are still lower than those for CuCl/SBA and CuOSi/SBA.
Cu/SBA(5.0), prepared via the incipient wetness impreg-
nation of SBA-15 with the use of Cu(NO3)2, followed by
reduction in CO at 773 K for 2 h, was also studied. This cat-
alyst was the most active for MF and CO2 production and
the least active for DMC formation. Consequently, this cat-
alyst was the least selective for DMC production from CO
consumption, approaching only 10.0% at zero conversion.
t
t
spectra of CuO Bu/SBA(3.4) and CuO Bu/SBA(4.3). Sug-
gesting the presence of only minor amounts of Cu metal.
3
The FT k χ(k) spectra of CuCl/SBA(2.9), CuCl/SBA(2.2),
and CuCl/SBA(1.1) after He pretreatment at 573 K for 1 h
(Fig. 3a) are quite similar to those of CuOSi/SBA(3.6) and
CuOSi/SBA(5.2), suggesting that the primary sites contain
Cu(I) species associated with O atoms, and that only a
small fraction of Cu is in a metallic state [36]. The spec-
tra presented in Figs. 3a and b are well represented by
Likewise, So
was only 52.0% at zero conversion.
DMC/MeOH
3
3
.3. Catalyst characterization
3
a linear combination of the experimental k χ(k) and FT
3
t
.3.1. EXAFS of pretreated catalysts
k χ(k) data for He-treated (573 K) CuO Bu/SBA(3.4) with
freshly prepared (not shown, refer to [37]) CuCl/SBA(2.9),
CuCl/SBA(2.2), and CuCl/SBA(1.1) and CuOSi/SBA(5.2),
CuOSi/SBA(3.6), and CuOSi/SBA(1.4), respectively, which
contain only isolated Cu(I) species [36,37]. Table 5 shows
the fraction of Cu present as small crystallites following He
pretreatment at 573 K for each of the catalysts. Also listed
is the fraction of the Cu available for catalysis. It is apparent
that more than 91% of the Cu present on CuCl/SBA and Cu-
OSi/SBA is available for catalysis, but that only 58% of the
3
FT k χ(k) spectra of CuCl/SBA(2.9), CuCl/SBA(2.2),
CuCl/SBA(1.1),
CuOSi/SBA(5.2),
CuOSi/SBA(3.6),
t
t
CuO Bu/SBA(4.3), and CuO Bu/SBA(3.4) taken after He
pretreatment at 573 K are shown in Fig. 3. Detailed discus-
sion of these spectra has previously been presented [36];
hence only a brief summary is presented here. The FT
3
t
t
k χ(k) spectra for both CuO Bu/SBA(3.4) and CuO Bu/
SBA(4.3) (Fig. 3c) have characteristics resembling those
of Cu foil (not shown). The peaks at 2.23, 3.4, 4.1, and
t
4
.7 Å (non-phase-corrected) represent the first four co-
Cu is available for catalysis on CuO Bu/SBA.
ordination shells of Cu in Cu metal. Combined with the
absence of any significant Fourier component that can be
assigned to Cu–O backscattering, this suggests that all Cu
has been reduced to Cu metal. A complete FEFF simu-
Fig. 4 shows normalized absorption and first-derivative
spectra for selected standards. Edge energies (E0) obtained
from the XANES spectra of the standards are listed in Ta-
ble 6. The values of E0 for Cu(0) (based on Cu foil), Cu(I)
(based on Cu2O), and Cu(II) (based on CuO) are 8980.0 eV,
8980.8 eV, and 8984.5 (eV), respectively. Although E0 for
Cu2O is very close to that of metallic Cu, it lies within
the range of values reported for Cu(I) compounds, 8980.8
and 8982.8 eV [37]. In situ XANES of CuCl/SBA(2.9),
t
t
lation of CuO Bu/SBA(3.4) and CuO Bu/SBA(4.3) was
3
performed, assuming that the observed FT k χ(k) spec-
tra represent the average scattering of a single Cu atom
in a 7-Å-diameter cuboctahedron [36,37]. Close agreement
between simulation and experiment suggests that Cu in
the as-prepared material undergoes reduction and sinter-
t
CuSiO/SBA(3.6), and CuO Bu/BA(3.4) were recorded to
3
ing during He pretreatment. By contrast, the FT k χ(k)
determine the change in oxidation state of each catalyst upon
exposure of the He-pretreated catalyst to the feed gas mix-
ture at 363 K. The temporal changes in the normalized ab-
spectra of CuOSi/SBA(3.6) and CuOSi/SBA(5.2) (Fig. 3b)
show a prominent Fourier component at 1.47 Å that can

I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
21
Table 5
Table 6
Fraction of Cu present as metallic Cu crystallites after He pretreatment and
the dispersion of Cu determined by EXAFS analysis
XANES edge-energies for samples under reaction conditions compared to
standards
a
b
c
Sample
X
X
Dispersion (%)
Sample
Formal
valency
E0 (eV)
Fresh
Metal
t
CuO Bu/SBA(3.4)
0.000
0.000
1.000
1.000
58.1
58.1
First
inflection
Second
inflection
t
a
b
CuO Bu/SBA(4.3)
CuOSi/SBA(3.6)
CuOSi/SBA(5.2)
0.880
0.785
0.120
0.215
94.9
91.0
Cu foil
0
1
1
1
1
2
8980.0
8982.1
8981.5
8982.8
8980.8
8984.5
–
–
–
–
–
–
t
c
[CuOSi[O Bu]3]4
t
c
[
CuO Bu]
4
CuCl/SBA(1.1)
CuCl/SBA(2.2)
CuCl/SBA(2.9)
0.980
0.926
0.893
0.020
0.074
0.107
99.2
96.9
95.5
c
CuCl
Cu O
2
CuO
d
Cu/SBA(5.0)
a
0.000
1.000
5
CuCl/SBA(2.9), t = 0 min
CuCl/SBA(2.9), t = 30 min
1
–
8981.5
8981.5
–
Fraction of total Cu present as Cu metal in the freshly prepared sample.
Fraction of total Cu present as Cu metal after He pretreatment. Particles
8985.4
b
CuOSi/SBA(3.6), t = 0 min
CuOSi/SBA(3.6), t = 30 min
1
–
8982.0
8981.5
–
8986
of Cu are taken to be 55 atom cluster (fcc with corners removed) [36].
c
Fraction of Cu sites in contact with the gas phase.
Estimated dispersion determined by Scherrer analysis of the Cu(111)
t
d
CuO Bu/SBA(3.4), t = 0 min
0
1
8980.4
8981.3
–
–
t
CuO Bu/SBA(3.4), t = 30 min
diffraction peak [61]. The particles are estimated to be 300 Å in diameter.
a
Based on the first inflection point corresponding to a shoulder of peak
on the rising absorption edge.
b
Based on the second inflection point corresponding to a shoulder or peak
on the rising absorption edge.
c
Refs. [36,37].
Following exposure of the catalyst to the reaction feed
gas mixture, changes occurred in the spectrum of each
catalyst at 363 K. The normalized first-derivative spectra
of CuCl/SBA(2.9) shows a decrease in the peak centered
at 8981.5 eV (Cu(I)) and an increase in a second peak
centered at 8985.4 eV (Cu(II)). Similarly, the normalized
first-derivative spectrum of CuOSi/SBA(3.6) shows a cor-
responding decrease in the first peak centered at 8981.5 eV
and an increase in a second peak centered at 8986.0 eV. Fi-
t
nally, CuO Bu/SBA(3.4) shows a 1.0-eV shift in the first
inflection point with an increase in the derivative intensity.
With time, the peak at 8981.3 decreases slightly, and only
a very small peak at 8986.0 eV is observed. The first of
these peaks is clearly identified with Cu(I) species, whereas
the second is probably due to a small concentration of
Cu(II) species. The in situ spectra of CuCl/SBA(2.9) and
CuSiO/SBA(3.6) exhibit edge energies at 8981.4 eV and
8985.4 eV, and 8982.0 eV and 8986.4 eV, respectively. The
Fig. 4. Cu K-edge XANES spectra of standards. (a) Normalized absorption
0
spectra of Cu (Cu foil), Cu(I) (Cu O), and Cu(II) (CuO) standards. (b)
2
First derivative spectra of (a).
lower of the two values of E lies in the range character-
0
istic of Cu(I), whereas the higher value of E0 is charac-
teristic of Cu(II). The fraction of Cu present as Cu(I) in-
sorption and first-derivative spectra of CuCl/SBA(2.9), Cu-
SiO/SBA(3.6), and CuO Bu/BA(3.4) over a 30-min period
are shown in Fig. 5, and the measured edge features are
listed in Table 6. At t = 0 min, the absorption spectra of
CuCl/SBA(2.9) and CuOSi/SBA(3.6) exhibit a single inflec-
tion point, which occurs at 8981.5 for CuCl/SBA(2.9) and at
t
creases in the order CuOSi/SBA(3.6) ∼ CuCl/SBA(2.9) <
t
CuO Bu/SBA(3.4). With increasing reaction temperature
the fraction of Cu(I) in CuCl/SBA(2.9) increased slightly
over that in CuOSi/SBA(3.6).
As reported in Table 2, all of the CuCl/SBA catalysts con-
tained residual Cl species. The nature of the Cl on the surface
of CuCl/SBA(2.9) was investigated by Cl K-edge XANES
(Fig. 6), since this technique is known to be a direct probe
of Cu–Cl bonding [43–45,57–59]. After treatment in reac-
tion conditions, CuCl/SBA(2.9) exhibits a pre-edge feature
at 2822.5 eV, a shoulder/edge peak at 2823.8 eV, and a ris-
ing edge inflection point at 2825.3 eV. Although the first of
8
982.0 for CuOSi/SBA(3.6), corresponding to the Cu(I) ox-
idation state. There is no indication of an inflection point at
ca. 8985 eV corresponding to Cu(II). The absorption spec-
t
trum of CuO Bu/SBA(3.4) also exhibits a single inflection
point at t = 0 min. However, this inflection point at 8980.4
corresponds to the Cu(0) state, consistent with the EXAFS
analysis above.

22
I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
Fig. 5. Normalized absorption and first derivative spectra of transient change in XANES spectra: CuCl/SBA(2.9) (a,d), CuOSi/SBA(3.6) (b,e), and
t
CuO Bu/SBA(3.4) (c,f) during first contact of reaction gas flow at 363 K. MeOH/O /CO/He gas mixture (4.0/1.0/9.0/19.3) was fed to the in situ cell
2
3
−1
.
at a nominal total flow rate of 20 cm min
Fig. 6. Cl K-edge XANES spectra of standards and catalysts. Normalized (a) absorption and (b) first derivative spectra. A four point smoothing is shown
for the first derivative spectrum of CuCl/SBA(2.9) to accentuate pre-edge inflection point indicated by arrow. (c) Deconvolution of the normalized absorption
spectrum of CuCl/SBA(2.9) after reaction conditions at 403 K. The absorption edge was simulated by a pseudo-Voigt fit.
these features is 1.6 eV above that for CuCl2 (2820.9 eV), it
lies well within the range of pre-edge energies reported pre-
viously for tetrahedral metal chlorides [56]. Association of
Cl with a Cu in a formal +2 state is consistent with the Cu
XANES spectrum of CuCl/SBA(2.9) taken after treatment
in reaction conditions. The shoulder/edge peak at 2823.8 eV
may be due to Cl atoms bonded to the support via Si–Cl
bonds. This interpretation is suggested by the similarity of

I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
23
the position of this peak to that observed for HCl-treated
SBA-15 (2825.1 eV) and ClSi(C6H5)3 (2824.1 eV). All rel-
evant spectral features are summarized in Table 6.
suggests that there is no long-range ordering and, hence,
that the CuO species are small and amorphous. The Cu–
Cu and Cu–O coordination numbers were larger for Cu-
OSi/SBA(3.6) (2.3(8) and 3.4(2), respectively) compared
3
3
In situ FT k χ(k) and k χ(k) spectra for CuCl/SBA(2.9),
t
CuOSi/SBA(3.6), and CuO Bu/SBA(3.4) are shown in
Fig. 7 together with their respective fits in R-space. The
fitted EXAFS parameters for specific atom pairs are given
in Table 7. CuCl/SBA(2.9) and CuOSi/SBA(3.6) exhibit
similar spectra; both having a non-phase-corrected Fourier
component centered at 1.53 Å attributed to oxygen backscat-
tering [36] and a second, non-phase-corrected Fourier com-
ponent centered at 2.63 Å. To identify the backscatterer
responsible for this latter peak, a phase and amplitude cor-
rection was applied according to Eq. (1). In the equation
k2
ꢀ
F (R) = k3w(k)χ(k)e
ꢁ
−φ(k) ir
e dk,
(1)
k1
w(k) is the Hanning apodization function, φ(k) is the phase
function, and all other terms are as defined previously. Phase
and amplitude functions associated with Si and Cu backscat-
tering were considered. Based upon the Lee and Beni crite-
ria [60] for determination of a specific backscattering pair,
the second Fourier component could be assigned to Cu–
Cu backscattering. The Cu–Cu phase-corrected spectrum for
CuCl/SBA(2.9) after reaction conditions at 403 K is shown
in Fig. 8. The Fourier component is symmetrical, and the
peak in the magnitude corresponds to the peak in the imag-
inary part of the magnitude function, proving its proper
identification as a Cu–Cu backscattering contribution. This
technique was also applied to CuOSi/SBA(3.6) with identi-
cal results.
The bond distances determined for CuCl/SBA(2.9) and
CuOSi/SBA(3.6) by fitting of the Cu–O and Cu–Cu shells to
theoretical scattering pairs (RCu–O ∼ 1.94 Å, RCu–Cu ∼ 2.93
Å, respectively) agree closely with the bond distances char-
acteristic of crystalline CuO (RCu–O ∼ 1.95 Å, RCu–Cu ∼
3
Fig. 7. Fourier transformed k χ(k) and the fitting results for the catalysts
after reaction conditions of 20% CO, 2% O and 10% MeOH at 403 K:
a) CuO Bu/SBA(3.4), (b) CuOSi/SBA(3.6), (c) CuCl/SBA(2.1). The fits
are indicated by open circles. The inset of each respective spectrum shows
the corresponding k χ(k) used for the Fourier transform.
2
t
(
3
2
.90 Å). The absence of Fourier components beyond 3 Å
Table 7
Cl K-edge spectral analysis
a
Sample
Pre-edge energy
eV)
Normalized pre-edge
intensity (eV)
Shoulder/edge-peak
inflection point (eV)
Rising edge inflection
point (eV)
Coordination
charge (η)
d
e
f
g
(
CuCl
2
2820.9
2820.2
2821.6
–
1.47
0.51
0.05
–
–
0.11
0.48
2824.7
2824.3
–
2824.1
2825.1
2823.9
2823.8
2825.0
2825.4
2825.0
2826.6
2826.6
2825.1
2825.3
−0.65
−0.52
−0.65
−0.13
−0.13
−0.62
−0.56
Cs CuCl
2
4
b
c
CuCl
ClSi(C H )
SBA-15 (0.12 M HCl)
CuCl/SBA(2.9) Fresh
CuCl/SBA(2.9) Rxn
6
5 3
–
2821.7
2822.5
a
Maximum in the first edge-region feature in the spectrum.
Not expected since Cu(I) is d . Suggests slight Cu(II) contamination.
A pseudo-Voigts fit with a fixed 50:50 ratio of Lorentzian–Gaussian contribution was used to reproduce the spectral features of CuCl/SBA(2.9) Rxn.
Integrated pre-edge intensity. Intensity equals the height multiplied by the full-width-at-half-maximum. Pre-edge modeled as pseudo-Voigt line shape with
b
c
d
10
a fixed 50:50 ratio of Lorentzian–Gaussian contribution.
e
Inflection point corresponding to shoulder or peak on rising absorption edge.
f
Highest energy maximum in the first derivative of the data in the rising edge region.
g
−1
Coordination charge (η) = −916.66 + 0.32425 eV (Rising edge Inflection Point) [43].

24
I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
Fig. 9. XRD pattern of Cu/SBA(5.0) after a MeOH/O /CO/He gas mixture
2
(
4.0/1.0/9.0/19.3) was fed to the reactor at 403 K.
3
.3.2. Powder XRD of the catalyst prepared by incipient
wetness impregnation
Cu/SBA(5.0) was examined by XRD both before and af-
ter exposure to reaction conditions. Following reduction in
CO at 573 K, the XRD pattern of Cu/SBA(5.0) exhibits a
strong (111) diffraction peak of Cu metal. Following reac-
tion conditions of 20% CO, 2% O2, and 10% MeOH at
4
03 K, the XRD pattern of Cu/SBA(5.0) continues to exhibit
a (111) diffraction peak of Cu metal (Fig. 9). By Scherrer
analysis [61], an upper estimate [62] of the particle size was
found to be 300 Å.
Fig. 8. Cu–Cu phase corrected spectrum of CuCl/SBA(2.9) after reaction
conditions at 403 K. Phase correction identifies second nearest neighbor as
Cu–Cu scattering component.
4
. Discussion
The DMC synthesis activities of various Cu-based cat-
with CuCl/SBA(2.9) (1.6(3) and 2.8(3), respectively); how-
ever, they were lower than that expected for bulk CuO
alysts are compared in Table 9. The most active catalyst
reported in this communication, CuCl/SBA(2.9), is simi-
lar in activity to Cu/X-zeolite [25] and Cu/Y-zeolite [25]
at the same temperature and comparable feed composi-
tions. CuCl/SBA(2.9) is nearly three times more active on
a total Cu basis compared with CuCl/MCM-41 [21] and
(
4 O nearest neighbors at 1.95 Å, 4 and 8 Cu nearest
neighbors at 2.90 and 3.08 Å, respectively). By contrast,
t
CuO Bu/SBA(3.4) has a noticeable peak at ∼ 2.1 Å and re-
solvable peaks above 3 Å. Fitting this spectrum reveals a
Cu–O contribution and two different Cu–Cu scattering con-
tributions. The Cu–Cu bond distance of 2.54 Å is character-
istic of copper metal, and the Cu–O and second Cu–Cu bond
distances are more similar to Cu2O.
similar in activity to CuCl /a.c. [15]. It is noted, how-
2
ever, that CuCl/MCM-41 and CuCl /a.c. were tested at CO
2
and MeOH partial pressures higher than those used to test
CuCl/SBA(2.9). Increasing the CO and MeOH partial pres-
Table 8
Fitting results for samples after reaction conditions at 403 K
a
b
2 c
2
0 d
e
Catalyst
Shell
CN
R
(Å)
σ
(Å )
E
(eV)
ꢀ-factor
0.0199
CuCl/SBA(2.9)
Cu–O
Cu–Cu
2.8 (3)
1.6 (3)
1.94 (1)
2.92 (1)
0.006 (1)
0.011
6.1 (1.4)
f
CuOSi/SBA(3.6)
Cu–O
Cu–Cu
3.4 (2)
2.3 (8)
1.933 (5)
2.93 (1)
0.0050 (7)
0.011 (2)
4.3 (8)
0.0086
0.0171
t
CuO Bu/SBA(3.4)
Cu–O
Cu–Cu
Cu–Cu
2.3 (2)
0.8 (1)
1.9 (4)
1.879 (9)
2.54 (1)
2.97 (1)
0.004 (1)
3.6 (1.6)
f
0.008
f
0.012
a
Coordination number.
Fitted radial distance.
Debye–Waller factor.
Energy reference shift.
b
c
d
e
ꢀ
-factor defined in Experimental section.
Fixed value.
f

I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
25
Table 9
DMC productivity comparison to conventional and literature catalysts for vapor phase oxidative carbonylation of methanol
a
Catalyst
Cu content
T
CO
(atm)
O
(atm)
MeOH
(atm)
TOF
(mol
Reference
2
DMC
DMC
−
s
Cu
1
−1) (×105)
(wt%)
(K)
mol
CuCl/SBA(2.9)
2.9
4.6
6.3
30.0
7.3
403.0
393.0
403.0
403.0
403.0
385.0
0.27
7.60
4.40
0.40
0.46
7.30
0.03
0.50
0.50
0.08
0.06
0.40
0.12
1.90
0.30
0.20
0.26
3.60
3.30
3.10
0.85
2.00
6.80
17.00
This work
[15]
[21]
[24]
[25]
CuCl /a.c.
2
CuCl/MCM-41
Cu/X-Zeolite
Cu/Y-Zeolite
Cu/Y-Zeolite/TEAC 12.7
[23]
a
TOF reported on a total Cu basis. Where appropriate, the reported TOF valve was converted to common units based on details in the experimental section
of the listed reference.
sures used to evaluate CuCl/SBA(2.9) should have improved
its activity, since the kinetics of DMC formation are nearly
first order in CO partial pressure [24]. TEAC (tetraethylam-
monium chloride)-impregnated Cu/Y zeolite is over 5 times
more active than CuCl/SBA(2.9) [23]. However, this catalyst
was also tested at CO and MeOH partial pressures higher
than those used to evaluate CuCl/SBA(2.9). Hence, a part
of the higher activity of TEAC-promoted Cu/Y is due to the
difference in reaction conditions.
was similarly passivated by reaction conditions, as shown for
CuO Bu/SBA(3.4) under reaction conditions by EXAFS.
t
As noted in Figs. 1 and 2 and Table 3, the DMC activity
of CuCl/SBA(2.9) is roughly a factor of 2 higher than that
of CuOSi/SBA(3.6) on a total Cu basis, but the two cata-
lysts exhibit comparable activities for MF and CO2. Another
striking similarity between these two catalysts is in the val-
ues of So
and So
. These are somewhat larger
DMC/MeOH
DMC/CO
for CuCl/SBA(2.9) versus CuOSi/SBA(3.6) at 403 K, but
they become identical at 433 K. The higher DMC activity
of CuCl/SBA(2.9) may be due to the promoting effects of
Cl on this catalyst. As discussed above, Cl XANES analyses
suggest that a portion of the residual Cl in CuCl/SBA(2.9)
is associated with Cu. Exactly how Cl would enhance the
DMC activity is not understood at this time. Since Cl pro-
motion is known to enhance the DMC synthesis activity of
Cu/Y zeolites [23], an effort was made to determine whether
an increase in Cl content would further improve the activity
of CuCl/SBA(2.9). CH3Cl was introduced into the reaction
mixture as 500-µl pulses. It was determined, however, that at
the reaction temperature used CH3Cl did not dissociate to re-
The results presented in Fig. 3 and Table 5 demonstrate
that, following He pretreatment at 573 K, the state of Cu
dispersion depends more upon the Cu precursor than on
the weight loading of Cu below 5.2 wt%. Both atomically
dispersed Cu(I) cations and small crystallites of metal-
lic Cu are observed, with the dispersion of Cu decreasing
t
in the order CuCl/SBA ∼ CuOSi/SBA ꢂ CuO Bu/SBA
ꢂ Cu/SBA(5.0). In all cases, the dispersion of Cu on
SBA-15 catalysts prepared by nonaqueous grafting or high-
temperature CVD is significantly higher than that which can
be achieved by incipient wetness impregnation.
During the oxidative carbonylation of methanol (0.26 atm
CO, 0.03 atm O2, 0.12 atm MeOH, balance He, 403 K),
XANES and EXAFS analyses demonstrated that the Cu
dispersed on SBA-15 undergoes significant rearrangement.
−
lease Cl. Impregnation of the catalysts with Cl -containing
solutions was not attempted, since it was anticipated that
the state of Cu might change during drying and calcina-
tion.
3
The FT k χ(k) for CuCl/SBA(2.9) and CuOSi/SBA(3.6)
It is also notable that the intrinsic DMC selectivity from
CO for both catalysts is roughly 50%. This suggests that
the synthesis of DMC occurs with the overall stoichiome-
try shown in reaction (2) and implies that CO2 is produced
concurrently with DMC. A mechanism for such a process
has been identified for DMC synthesis from Cu(I) cations
exchanged into a zeolite [63]:
are similar in appearance (see Fig. 7). For both cata-
lysts, the bond distances determined by fitting Cu–O and
Cu–Cu shells to theoretical standards agree closely with
those characteristic of crystalline CuO; however, the ab-
sence of Fourier components beyond 3 Å suggests that
there is no long-range ordering and, hence, that the CuO
is not crystalline. Consistent with this interpretation, the
Cu–Cu and Cu–O coordination numbers are lower than
those expected for bulk CuO. The XANES analyses of
CuCl/SBA(2.9) and CuOSi/SBA(3.6) show evidence for
Cu(I) and Cu(II) species. The Cu(I) cations are presumed to
lie on the surface of the CuO structures. Helium-pretreated
2CH OH + CO + O = (CH O) CO + CO ,
3
2
3
2
2
(2)
(3)
(CH3O)2CO + H2O = 2CH3OH + CO2.
It is also known that CO can be produced by the hydroly-
2
sis of DMC, as shown in reaction (3). To establish whether
t
CuO Bu/SBA(3.4) also undergoes oxidation upon exposure
this reaction might account for CO formation during DMC
2
to reaction conditions, but in this case EXAFS and XANES
analyses suggest that the small Cu particles become covered
with a layer of Cu2O. In the case of Cu/SBA(5.0), diffraction
peaks associated with Cu2O were not observed. Neverthe-
less, we believe an overlayer of the large metal particles
synthesis, an experiment was performed in which a flow con-
taining 0.01 atm DMC and 0.002 atm H O was fed over
2
CuCl/SBA(2.9) at 403 K. Assuming first-order kinetics in
DMC and H O, the apparent rate constant for the forward re-
2
5
5
−1
−1
−1
min .
action was found to be 1.1×10 cm mol mol
Cu
DMC

26
I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
Although this reaction is rapid, at the concentrations of
DMC and H2O produced during DMC synthesis the rate
of DMC hydrolysis is calculated to be too low to account
for the additional CO2 observed. Thus, it appears that the
correct stoichiometric reaction for DMC synthesis is reac-
tion (2).
In conclusion, it was found that:
1. Highly dispersed Cu(I) cations on very small particles
of noncrystalline CuO promote S
, S
,
DMC/CO
DMC/MeOH
and activity. Although chlorine is not a requirement for
DMC production, its presence appears to enhance DMC
selectivity.
t
The DMC and MF activities of CuO Bu/SBA(3.4) are
nearly identical to those of CuOSi/SBA(3.6) on a total Cu
basis, as is the DMC selectivity from methanol. The signif-
icant difference between these catalysts is the much lower
DMC selectivity from CO for CuO Bu/SBA(3.4). Selectiv-
ity lower than 50% suggests that CO2 is produced not only
2. Reduced Cu (in Cu(I) oxide aggregates) promotes in-
creased activity to side reactions producing CO2.
t
Acknowledgments
as a by-product of DMC synthesis, but also via CO combus-
The authors thank S.H. Choi, J. Bronkema, S. Mukhopad-
hyay, N. Stephenson, B. Wood, Q.T. Liu, M. Zeralla, and
Andreas Hyden for their assistance in collecting XAS data at
SSRL and NSLS. Dr. F. Schlachter is also recognized for his
guidance and support at BL 9.3.1 at the ALS. Portions of this
research were carried out at the Stanford Synchrotron Radia-
tion Laboratory, a national user facility operated by Stanford
University on behalf of the US Department of Energy, Of-
fice of Basic Energy Sciences. Research carried out at the
National Synchrotron Light Source, Brookhaven National
Laboratory, is supported by the US Department of Energy,
Division of Materials Sciences and Division of Chemical
Sciences, under Contract No. DE-AC02-98CH10886. This
work was supported by the Methane Conversion Coopera-
tive funded by BP and by the Department of Energy under
Contract No. DE-AC03-76SF00098.
tion. The low value of So
for CuO Bu/SBA(3.4) is
t
DMC/CO
attributed to the presence of Cu2O in conjunction with Cu.
This interpretation is supported by literature reports that
show CO combustion activity increases with Cu2O for-
mation on Cu metal [64–66]. Consistent with this, Ta-
ble 4 shows that when DMC synthesis is carried out on
Cu/SBA(5.0) prepared by aqueous impregnation, the DMC
selectivity from CO is very low on a total Cu basis, at the
same time that the DMC selectivity from MeOH is compa-
rable to that obtained with CuCl/SBA(2.9).
Thus, the results of the present study indicate that to
achieve high DMC activity and selectivity from both MeOH
and CO, it is desirable to maintain a high dispersion of Cu
and to avoid the formation of Cu crystallites. The preferred
active sites for DMC synthesis by oxidative carbonylation
appear to be Cu(I) cations dispersed on very small particles
of noncrystalline CuO. The activity of these sites is enhanced
by the presence of small amounts of Cu–Cl species.
References
[
[
[
1] M. A Pacheco, C.L. Marshall, Energy and Fuels 11 (1997) 2.
2] P. Tundo, M. Selva, Acc. Chem. Res. 35 (2002) 706.
3] U. Romano, R. Tesel, M.M. Mauri, P. Rebora, Ind. Eng. Chem. Prod.
Res. Dev. 19 (1980) 396.
5
. Conclusions
t
t
[
CuOSi(O Bu)3]4, [CuO Bu]4, and CuCl were used as
[
[
4] R. Srivastava, D. Srinivas, P. Ratnasamy, Catal. Let. 91 (2003) 133.
5] T. Wei, M. Wang, W. Wei, Y. Sun, B. Zhong, Green Chemistry 5 (2003)
343.
precursors to produce mesoporous silica-supported copper
catalysts. XANES and EXAFS characterization prior to re-
action showed that Cu in the catalysts prepared with the
use of CuCl and [CuOSi(O Bu)3]4 was present as isolated
Cu(I) cations, whereas use of [CuO Bu]4 produced 1-nm
[6] B.M. Bhanage, S. Fujita, Y. Ikushima, M. Arai, Green Chemistry 5
t
(2003) 429.
[
[
7] S. Xie, A.T. Bell, Catal. Let. 70 (2000) 137.
8] K.T. Jung, A.T. Bell, J. Catal. 204 (2001) 339.
t
Cu particles. The state of Cu dispersion depends more upon
the Cu precursor than on the weight loading of Cu. Fol-
lowing reaction conditions, EXAFS analysis reveals that
CuCl/SBA(2.9) and CuOSi/SBA(3.6) have very similar lo-
cal Cu environments. Residual Cl is present in CuCl/SBA,
and XANES analysis shows that at least some of the Cl
is bound to Cu. During reaction, Cu in catalysts prepared
[9] K.T. Jung, A.T. Bell, Top. Catal. 20 (2002) 97.
[
[
10] I. Yamanaka, A. Funakawa, K. Otsuka, Chem. Let. (2002) 448.
11] G. Filardo, A. Galia, F. Rivetti, O. Scialdone, G. Silvestri, Elec-
trochem. Acta 42 (1997) 1961.
[
12] G.L. Curnutt, A.D. Harley, in: Oxygen Complexes and Oxygen Acti-
vation by Transition Metals, Plenum, New York, 1988, pp. 215–232.
[13] G.L. Curnutt, US Patent, 4,625,044 (1986).
14] G.L. Curnutt, US Patent, 5,004,827 (1991).
[15] M.S. Han, B.G. Lee, I. Suh, H.S. Kim, B.S. Ahn, S.I. Hong, J. Mol.
Catal. A. 170 (2001) 225.
16] M.S. Han, B.G. Lee, I. Suh, B.S. Ahn, H.S. Kim, D.J. Moon, S.I.
[
t
from CuCl and [CuOSi(O Bu)3]4 formed highly dispersed
CuO moieties, whereas the Cu in the catalyst prepared from
[
t
[
CuO Bu]4 formed a cuprous oxide layer over a Cu(0)
Hong, J. Mol. Catal. A. 170 (2003) 137.
core.
High DMC activity correlated with Cu dispersion. The
values of So and So
decreased with decreas-
ing Cu dispersion when the original state of the Cu was
Cu(0).
[17] K. Tomishige, T. Sakaihori, S. Sakai, K. Fujimoto, Appl. Catal. A. 181
1999) 95.
(
[18] H. Itoh, Y. Watanabe, K. Mori, H. Umino, Green Chemistry 5 (2003)
DMC/MeOH
DMC/CO
558.
[19] R.X. Jiang, S.F. Wang, X.Q. Zhao, Y.J. Wang, C.F. Zhang, Appl. Catal.
A. 238 (2003) 131.

I.J. Drake et al. / Journal of Catalysis 230 (2005) 14–27
27
[20] Y. Cao, J.C. Hu, P. Yang, W.L. Dai, K. Fan, Chem. Commun. (2003)
[46] The exact energy position of the pre-edge peak of Cs CuCl used in
2 4
9
08.
the Cl K-edge calibration of Refs. [31,32] is believed to be based on
the S K-edge of 2472.0 eV discussed in Ref. [33].
[
21] Z. Li, K. Xie, R.C.T. Slade, Appl. Catal. A. 205 (2001) 85.
[
22] P. Yang, Y. Cao, J.C. Hu, W.L. Dai, K. Fan, Appl. Catal. A 241 (2003)
[47] Personal communication with Fred Schlachter. The rotary encoder was
installed as part of recent upgrade to the BL 9.3.1 of the ALS.
3
63.
23] D.C. Molzahn, M.E. Jones, G.E. Hartwell, J. Puga, US Patent
,387,708.
[
[48] T. Ressler, http://ourworld.compuserve.com/homepages/t_ressler/.
[49] T.J. Ressler, Synchrotron Radiat. 5 (1998) 118.
[50] E.A. Stern, M. Newville, B. Ravel, Y. Yacoby, D. Haskel, Physica
B 208–209 (1995) 117.
5
[
[
[
[
[
[
24] S.A. Anderson, T.W. Root, J. Catal. 217 (2003) 396.
25] S.T. King, J. Catal. 161 (1996) 530.
26] S.T. King, Catal. Today 33 (1997) 173.
27] S.T. King, M.E. Jones, M.M. Olken, US Patent 5,391,803 (1995).
28] S. Brunauer, P.H. Emmett, E. Teller, J. Am. Chem. Soc. 60 (1938) 309.
29] E.P. Barrett, L.G. Joyner, P.P. Hakebda, J. Am. Chem. Soc. 73 (1951)
[
[
[
[
51] M.J. Newville, Synchrotron Rad. 8 (2001) 322.
52] IFEFFIT manual: http://cars9.uchicago.edu/ifeffit/.
53] A.L. Ankudinov, J.J. Rehr, Phys. Rev. B 56 (1997) R1712.
54] A.L. Ankudinov, C. Bouldin, J.J. Rehr, J. Sims, H. Hung, Phys. Rev.
B 65 (2002) 104107.
373.
[
[
[
30] J.A. McGinnety, J. Am. Chem. Soc. 94 (1972) 8406.
31] M.J. Sharnoff, Chem. Phys. 42 (1965) 3383.
32] D. Zhao, J. Feng, Q. Huo, N. Melosh, G.H. Fredrickson, B.H.
Chmelka, G.D. Stucky, Science 279 (1998) 548.
2
[55] This statistical value is related to the standard definition of χ but is
normalized to the number of degrees of freedom in the fit. The details
of this calculation have been described elsewhere; M. Newville, B.
Ravel, D. Haskel, J.J. Rehr, E.A. Stern, Y. Yacoby, Physica B 208–
209 (1995) 154.
[
[
33] K.L. Fujdala, T.D. Tilley, J. Am. Chem. Soc. 123 (2001) 10133.
34] K.W. Terry, C.G. Lugmair, P.K. Gantzel, T.D. Tilley, Chem. Mater. 8
2
−1
(1996) 274.
[56] For example, uncalcined: CuOSi/SBA(3.5), 678 m g
and
;
3
−1
2
−1
3
−1
[
[
35] T. Tsuda, T. Hashimoto, T. Saegusa, J. Am. Chem. Soc. 94 (1972) 658.
36] I. Drake, K.L. Fujdala, S. Baxamusa, T.D. Tilley, A.T. Bell, J. Phys.
Chem. B, in press.
0.88 cm g ; CuOSi/SBA(5.0), 567 m g
and 0.80 cm g
t
2
−1
3
−1
CuO Bu/SBA(5.0), 745 m g and 0.98 cm g . After heating: Cu-
2
−1
3
−1
OSi/SBA(3.5), 800 m g and 1.01 cm g ; CuOSi/SBA(5.0), 782
2
−1 −1 −1
3
t
2
[37] K.L. Fujdala, I. Drake, T.D. Tilley, A.T. Bell, J. Am. Chem. Soc. 126
m g and 1.04 cm g ; and CuO Bu/SBA(5.0), 813 m g and
3
−1
.
(
2004) 10864.
38] R.E. Jentoft, S.E. Deutsch, G.C. Gates, Rev. Sci. Instrum. 67 (1996)
111.
1.09 cm g
[
[57] T. Glaser, B. Hedman, K.O. Hodgson, E. Solomon, Acc. Chem.
2
Res. 33 (2000) 859.
[39] E.A. Stern, K. Kim, Phys. Rev. B. 23 (1981) 378.
[
[
[
[
[
58] C. Sugiura, T. Suzuki, J. Chem. Phys. 75 (1981) 4357.
59] C. Sugiura, J. Chem. Phys. 58 (1973) 5444.
[40] D.C. Koningsberger, R. Prins, X-ray Absorption, Wiley, New York,
1988.
60] P.A. Lee, G. Beni, Phys. Rev. B 15 (1977) 2862.
61] P. Scherrer, Nachr. Ges. Wiss. Göttinger (1918) 96.
62] S. Calvin, M.M. Miller, R. Goswami, S.-F. Cheng, S.P. Mulvaney, L.J.
Whitman, V.G. Harris, J. Appl. Phys. 94 (2003) 778.
63] B. Peters, A.T. Bell, Manuscript in preparation.
64] A. Knop-Gericke, M. Hävecker, Th. Schedel-Niedrig, R. Schlögl, Top.
Catal. 15 (2001) 27.
[
[
[
[
[
41] D.C. Koningsberger, B.L. Mojet, G.E. van Dorssen, D.E. Ramaker,
Top. Catal. 10 (2000) 143.
42] B.K. Teo, EXAFS: Basic Principles and Data-analysis, Springer, New
York, 1986.
43] S.E. Shadle, B. Hedman, K.O. Hodgson, E.I. Solomon, Inorg.
Chem. 33 (1994) 4235.
[
[
44] B. Hedman, K.O. Hodgson, E.I. Solomon, J. Am. Chem. Soc. 112
[65] A. Knop-Gericke, M. Hävecker, Th. Schedel-Niedrig, R. Schlögl, Top.
Catal. 10 (2000) 187.
(1990) 1543.
45] B. Hedman, P. Frank, S.F. Gheller, A.L. Roe, W.E. Newton, K.O.
Hodgson, J. Am. Chem. Soc. 110 (1988) 3798.
[66] G.G. Jernigan, G.A. Somorjai, J. Catal. 147 (1994) 567.