Active sites on oxide surfaces: ZnO-catalyzed synthesis of methanol from CO and H2

Article abstract of DOI:10.1002/anie.200462374

(Chemical Equation Presented) Is this place taken? A mechanism has been proposed for the formation of methanol from CO and H₂ on ZnO surfaces in which CO is adsorbed at oxygen vacancies on the heterogeneous catalyst (see picture: C red, H white, O gray, Zn black). The active sites are blocked when CO₂ is added to the gas mixture.

Full text of DOI:10.1002/anie.200462374

Communications
Heterogeneous Catalysis
Active Sites on Oxide Surfaces: ZnO-Catalyzed
Synthesis of Methanol from CO and H2**
Melanie Kurtz, Jennifer Strunk, Olaf Hinrichsen,*
Martin Muhler, Karin Fink,* Bernd Meyer, and
Christof Wꢀll
Chemical processes on metal oxide surfaces have been of
great interest for a long time, mostly due to their relevance in
the field of heterogeneous catalysis. Since “perfect” oxide
surfaces usually show only little or no activity, it is widely
believed that the high reactivity of oxide powders results from
the presence of a small number of active sites.[1] Most of the
models proposed for these active sites are based on oxygen
vacancies or defects. There is literature consensus on the role
of these active sites and the particular importance of oxygen
vacancies. However, only a few case studies have been
described in which for a specific reaction the geometric
structure of an active site is not only specified, but also
confirmed by theoretical calculations and validated by kinetic
studies conducted in a reactor under conditions close to those
applied industrially.
We have conducted such a study for the synthesis of
methanol over ZnO powders. In a recently published
theoretical paper,[2] an active site and a reaction mechanism
were proposed but without a comparison to kinetic data. It
was assumed that a Cu-free model for the active site is
sufficient to explain the formation of methanol from a CO2/H2
mixture on an industrially applied Cu/ZnO/Al2O3 catalyst. In
our extensive analysis we find—for an active site similar to
that proposed—an energetically more favorable reaction
pathway, in which it is not carbon dioxide acting as carbon
source, as was assumed so far, but carbon monoxide, which is
favored for thermodynamic reasons. On the ZnO-supported
Cu-containing powder catalysts used at present for the
industrial synthesis of methanol, another reaction mechanism
with carbon dioxide as the carbon source is occurring. This
was a point of controversy for a long time, but the mechanism
[*] Dr. M. Kurtz, Dipl.-Chem. J. Strunk, Prof. Dr.-Ing. O. Hinrichsen,
Prof. Dr. M. Muhler
Lehrstuhl fꢀr Technische Chemie
Ruhr-Universitꢁt Bochum, 44780 Bochum (Germany)
Fax: (+49)234-32-14115
E-mail: o.hinrichsen@techem.ruhr-uni-bochum.de
Dr. K. Fink, Dr. B. Meyer
Lehrstuhl fꢀr Theoretische Chemie
Ruhr-Universitꢁt Bochum, 44780 Bochum (Germany)
Fax: (+49)234-32-14045
E-mail: karin.fink@ruhr-uni-bochum.de
Prof. Dr. C. Wꢂll
Lehrstuhl fꢀr Physikalische Chemie I
Ruhr-Universitꢁt Bochum, 44780 Bochum (Germany)
[**] This research was supported by the German Science Foundation
within the framework of the Collaborative Research Center 558
“Metal–Substrate Interactions in Heterogeneous Catalysis”.
2790
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200462374
Angew. Chem. Int. Ed. 2005, 44, 2790 –2794

Angewandte
Chemie
was finally clarified by experiments using isotope-labeled
reactants.[3–7]
concentration was constant over time, 1% CO2 was added to
the synthesis gas. The concentrations of CO and H2 were kept
constant, and only the N2 content was adjusted. After the
system had again reached a steady-state CH3OH content, the
CO2 fraction in the inlet flow was increased to 2, 3, 4, 6, and
8%. The activity plot thus determined experimentally
(Figure 2) clearly shows that even an addition of CO2 as low
One of the major problems in the analysis of chemical
processes on such active sites is their typically rather low
concentration on the surface. Thus, the systematic analysis of
precise data by the well-established methods of surface
physics and chemistry is impossible or at least highly
problematic. For example, it is not possible to determine
experimentally the binding energy of CO on defect sites with
reliability. Just recently, reliable information available con-
cerning binding energies on perfect ZnO surfaces was
reported.[8–11] In the present study, we have therefore chosen
an approach in which we combine precise quantum chemical
ab initio calculations—which had been validated beforehand
by calculations on perfect systems and by a comparison with
experimental data[12]—with detailed measurements under
steady-state conditions close to those applied in industry.
Based on these measurements and the quantum chemical
results, we propose a reaction mechanism in which a CO
molecule is bound in an unusual adsorption geometry and
transformed into a formyl species.
Figure 2. The activity of the coprecipitated ZnO/Al2O3 catalyst as a
The kinetic and characterization studies of several differ-
ently prepared ZnO powder samples presented in ref. [13]
showed that methanol synthesis on ZnO is a structure-
sensitive reaction; polar surfaces are distinguished by their
high turnover frequencies. Figure 1 compares reactions per-
function of the fraction of CO2 in the inlet flow. Experimental condi-
tions: V=100 NmLminꢀ1, T=583 K, P=4.1 MPa, wcat. =1.00 g.
˙
as 1% results in a strong decrease of the CH3OH yield.
Further addition of CO2 continuously lowers the CH3OH
yield, but the effect is less pronounced. CO2 therefore seems
to act on the ZnO/Al2O3 catalyst as a catalyst poison. Since
the catalytic activity of the ZnO/Al2O3 catalyst depends on
the preset CO/CO2 ratio, the working state of the active site
under industrial conditions is of particular importance.
Even small amounts of CO2 in the synthesis gas markedly
lower the formation of CH3OH on the ZnO/Al2O3 catalyst.
This observation strongly supports the hypothesis that the
active site of the ZnO/Al2O3 catalyst is an oxygen vacancy. In
this case the number of active sites on the catalyst would
increase in a highly reducing atmosphere (no CO2, high level
of CO). Supplementary measurements confirmed the forma-
tion of small amounts of CO2 when the catalyst was exposed
to a gas mixture with a high concentration of CO. These
findings can be explained by a direct removal of oxygen from
the ZnO lattice by CO, leading to the formation of CO2.
Inversely, addition of CO2 to the synthesis gas would cause an
oxidation of the vacancies and catalyst activity would drop.
To support these mechanistic ideas we performed quan-
tum chemical ab initio calculations to determine the geo-
metric structure and the energies of different intermediates
for the reaction of CO with 2H2 over ZnO catalysts. An
“embedded cluster” approach was used in which the sur-
roundings of the active site are described by a cluster treated
explicitly with quantum chemical methods while the cluster
itself is embedded in an extended point-charge field.
Figure 1. Comparison of the activity of the binary and ternary catalysts.
formed with a Cu-free binary ZnO/Al2O3 catalyst and a
ternary Cu catalyst under near-industrial reaction conditions.
The latter catalyst is applied technically in large-scale, low-
temperature methanol synthesis. Under similar conditions it
is much more active than the binary ZnO/Al2O3 catalyst,
which is highly temperature-stable. When a CO/H2 mixture is
used, the space–time yield (STY) is already on the same order
of magnitude as when the temperature is 100 K lower. The
difference in the STYs is even more pronounced when gas
mixture with another composition is employed. A synthesis
gas containing CO2 results in a drastic increase in activity for
the ternary system but a decrease for the binary system.
The inhibiting effect of CO2 was proven by the following
experimental sequence: a gas mixture containing about 72%
H2, 10% CO, and 18% N2 was passed over the ZnO/Al2O3
catalyst at a flow rate of 100 NmLminꢀ1. When the CH3OH
As in the work of French et al.,[2] the polar O-terminated
¯
ZnO(0001) surface was chosen as starting point for our
calculations. To build a model for the active center we first
investigated the atomic structure and chemical composition
¯
of the (0001) surface under catalytic conditions. For a long
¯
time it was believed that the (0001) surface of ZnO is
Angew. Chem. Int. Ed. 2005, 44, 2790 –2794
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2791

Communications
unreconstructed and exhibits an ideal “truncated bulk”
¯
structure. Because of the polar character of the (0001) surface
inevitably partly occupied surface bands would be present in
this case, leading to a vanishing band gap.[14,15] However, in
the last years it was shown by experimental and theoretical
work[11,12,16] that the (0001) surface is stabilized by the
¯
formation of O vacancies or the adsorption of hydrogen,
depending on the experimental conditions. The driving force
behind these structural changes is the avoidance of partly
¯
filled surface bands. Assuming that the ZnO(0001) surface is
Figure 3. The ab initio cluster consists of the three Zn atoms of the
first Zn coordination shell (black, big) next to the defect site (square),
the nine O atoms of the first O coordination shell (six from the first
and three from the second layer, gray), three O atoms from the third
layer that are a short distance (4.567 ꢃ) from the defect site, six further
O atoms from the first layer (gray), all H atoms (white) bound to the
cluster O atoms, and of all Zn atoms neighboring the cluster oxygen
atoms (black, small).
in thermodynamic equilibrium with H2 and O2 in the gas
phase, Meyer[16] determined a surface phase diagram by using
DFT slab calculations. At pressures and temperatures typi-
cally employed in methanol synthesis the most stable
structure was found to be a surface with a hydrogen-atom
surface coverage of 50% where all surface bands are filled.
The point-charge field used for the embedding of the ab
initio cluster was generated by using the optimized structure
¯
of the hydroxylated ZnO(0001) surface from ref. [16]. The
ically during the reaction: for the gas-phase reaction CO +
2H2!CH3OH, by 64.7 kJmolꢀ1 (for experimental data see
ref. [22]).
Figure 4 shows the geometries and the relative energies of
the most important steps in the reaction mechanism for the
synthesis of CH3OH from CO and H2. The binding energies
for the optimized structures are given in Table 1.
The combination of the results of the quantum chemical
calculations and the kinetic measurements suggests the
following model for the overall reaction: Over ZnO catalysts,
methanol is formed mainly from adsorbed CO. At the active
site CO is adsorbed perpendicularly to the surface and can be
bound either through the C or the O atom. The binding
energy for the C-bound configuration, 17.1 kJmolꢀ1, is almost
the same as for the O-bound adsorption within the accuracy
of the calculations. Therefore, both situations should be
present under catalytic conditions. CO adsorbed through the
oxygen atom is more favorable for the subsequent hydro-
genation steps. Hence, this configuration was considered in
our calculations for the next steps. The calculated binding
active site with the O vacancy was obtained by removing an
entire H2O molecule from the surface in accordance with the
tendency to avoid occupied surface states (Figure 3). As the
first step, the adsorption geometries for different adsorbed
intermediates were determined.[17] All atoms of the adsor-
bates and the positions of the three Zn ions together with the
six surface O atoms (including the H atoms bound to these O
atoms) next to the vacancy were relaxed. For the optimized
structures the adsorption energies were calculated with the
Bochum program package,[18–20] taking into account the
dynamic correlation within the multiconfiguration coupled
electron pair approximation (MCCEPA),[20] an approximate
coupled-cluster method. We used TZVPP basis sets for the
adsorbate molecules and the Zn and O ions of the first
coordination shell of the active site. The basis set super-
position error for the binding energy was corrected as
proposed by Boys and Bernadi.[21] To compare the energies
of the different intermediates it is important to correct all
energies for zero-point vibrations, since they change dramat-
Figure 4. Potential energy diagram and adsorption geometries for formation of methanol from CO and H2 catalyzed by ZnO. Cluster: Zn black, O
gray, H white; Adsorbate: C black, O gray, H white.
2792
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 2790 –2794

Angewandte
Chemie
Table 1: Calculated energies [kJmolꢀ1] of the intermediates relative to the energy of the starting
compounds.
formation of methanol is signifi-
cantly slower. In contrast to the CO
mechanism via a formyl species,
the reaction path with CO2 takes
place preferentially under these
reaction conditions because the
formate intermediate is more
stable than formyl. In our opinion,
the activity of the Cu-containing
catalyst is determined by the strong
interaction between Cu and
ZnO1ꢀx (x is equal to the number
of O vacancies). The hydrogena-
Adsorbate(s)
Gas phase
DE(SCF)
DE(MCCEPA)
NPE [a]
E incl. NPE
dO [ꢃ][b]
1
2
3
4
5
6
7
CO+2H2
2H2
H2
0.0
ꢀ12.0
3.4
ꢀ91.2
ꢀ320.0
ꢀ193.0
ꢀ115.1
0.0
ꢀ15.0
3.1
ꢀ105.8
ꢀ344.0
ꢀ226.1
ꢀ143.7
66.2
66.2
61.5
0.0
ꢀ15.0
ꢀ1.6
OC
0.90
0.22
0.53
0.12
0.79
CHOꢀ, H+
CH2O
H2
96.0
ꢀ76.0
ꢀ307.6
ꢀ158.1
ꢀ75.8
CH3Oꢀ, H+
CH3OH
102.6
134.2
134.2
CH3OH
[a] NPE: calculated zero-point energy. [b] dO: distance of the O atoms from the position of the defect site.
energy for CO at the active site is much smaller than the CO2
adsorption energy of 69.0 kJmolꢀ1 obtained by French et al.[2]
Therefore, CO2 in the synthesis gas can possibly block the
active site for the CO reaction.
tion of CO2 at the Cu-ZnO1ꢀx interface is favored over the
conversion of CO.[34] For definite proof that methanol is
formed by these two different reaction paths, spectroscopic
analysis of the intermediates, calculation of transition states,
and comparison of the results obtained by microkinetic
modeling based on quantum chemical calculations with the
data of kinetic measurements will be necessary.
We expect the formation of a formyl species to be the next
reaction step.[24–27] It is well known that H2 adsorbs dissocia-
tively at the O vacancies of the polar surfaces. Therefore, we
assume that formyl is formed by the transfer of Hꢀ to the
carbon atom of the CO, adsorbed through the oxygen atom, at
the active site, while a neighboring oxygen atom is protonated
at the same time. In the energy diagram in Figure 4, the
formyl species has about the same energy as the starting
situation as a result of two compensating effects: The
adsorption of the negatively charged formyl species is
strongly stabilized in the Madelung field of the ZnO environ-
ment and leads in combination with the proton affinity of the
cluster to strong binding at the active site. On the other hand,
breaking the CO triple bond and dissociating the H2 molecule
cost a lot of energy.
Experimental Section
The catalysts were synthesized by the coprecipitation method. The
hydroxycarbonate precursors were prepared by precipitation from
metal nitrate solutions with sodium carbonate at pH 7. The concen-
trations of the nitrate solutions was adjusted to a nominal catalyst
composition of 85 mol% ZnO and 15 mol% Al2O3 for the ZnO/
Al2O3 catalyst. The resulting precipitate was aged in the mother
liquor under continuous stirring for 2 h. The aged precipitate was
filtered and washed with deionized water (6 ꢀ 20 mL). Drying was
performed overnight at 393 K followed by calcination in synthetic air
(V= 10 NmLminꢀ1) increasing the temperature from room temper-
˙
ature to 593 K at a heating rate of 2 Kminꢀ1 and maintaining the
temperature for 3 h. The calcined precursor was pressed and sieved,
yielding a grain fraction of 250–355 mm. The preparation of the
ternary Cu/ZnO/Al2O3 catalyst with a nominal catalyst composition
of 70 mol% Cu, 15 mol% ZnO, and 15 mol% Al2O3 followed the
same procedure.
Since the adsorption of CO on ZnO is weak,[12,29] we
assume that the formation of formyl is the rate-determining
step in methanol synthesis. The following stepwise hydro-
genation of the formyl species with dissociatively adsorbed
hydrogen leads to the methoxy species, which is also observed
on ZnO in studies on intermediates.[27,30–32] The formyl species
abstracts the proton of a neighboring OH group, rearranges to
formaldehyde, and then reacts with a further Hꢀ to give the
strongly bound methoxy species. The methoxy ion is adsorbed
such that the negatively charged oxygen atom is located at the
defect site. In the last hydrogenation step, a proton of a
neighboring OH group is transferred to the methoxy species
forming methanol. The methanol molecule desorbs from the
surface, the O defect is reestablished, and the catalytic cycle is
completed. The catalyst is regenerated continuously, since in
the presence of CO and H2 at high temperatures small
amounts of CO2 and H2O are always formed, leading to the
formation of new active sites. This might be the reason why
ZnO is very stable under reaction conditions and suitable as a
catalyst for CO hydrogenation under industrial conditions.
Adding CO2 to the CO/H2 gas mixture leads to a change in
the reaction mechanism in the ZnO-catalyzed methanol
synthesis. In analogy to the mechanism described in ref. [2]
for ZnO or to the redox mechanism postulated for Cu-
containing catalysts,[33] the reaction path for the hydrogena-
tion of CO2 includes the formation of a formate and a
methoxy species as intermediates. In the presence of CO2, the
For the catalytic experiments, 1.00 g of the calcined precursor
(250–355 mm) was used in a fixed-bed single-pass tubular reactor (bed
length ca. 2.1 cm). For pretreatment, the sample was heated at a
constant rate of 1 Kminꢀ1 to 448 K and held at this temperature
overnight. A mixture of 2% H2 in He (V= 60 NmLminꢀ1) was chosen
˙
as the pretreatment gas. Then the reactor temperature was increased
at a constant rate of 1 Kminꢀ1 to 513 K and held at this temperature
for 30 min. Then the pretreatment gas mixture was replaced with pure
hydrogen (V= 60 NmLminꢀ1). The space-time yield (STY) was
˙
determined at 4.1 MPa and 453–583 K using synthesis gas mixtures
*
containing H2, CO, CO2, and N2 with the following composition: (
)
*
73.1% H2, 10.8% CO, 16.1% N2; ( ) 72.3% H2, 9.9% CO, 4.5%
&
CO2, 13.3% N2; (&) 73.9% H2 9.6% CO, 16.5% N2; ( ) 73.2% H2,
9.5% CO, 4.6% CO2, 12.7% N2.
Received: October 20, 2004
Revised: January 18, 2005
Published online: April 4, 2005
Keywords: heterogeneous catalysis · kinetics ·
.
methanol synthesis · quantum chemistry · zinc oxide
Angew. Chem. Int. Ed. 2005, 44, 2790 –2794
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2793

Communications
[27] C. Chauvin, J. Saussey, J. C. Lavalley, G. Djega- Mariadassou,
Appl. Catal. 1986, 25, 59.
[28] S. A. French, A. A. Sokol, S. T. Bromley, C. R. A. Catlow, P.
Sherwood, Top. Catal. 2003, 24, 161.
[29] H. H. Kung, Catal. Rev. Sci. Eng. 1980, 22, 235.
[30] A. Ueno, T. Onishi, K. Tamaru, Trans. Faraday Soc. 1971, 67,
3585.
[31] L. Chan, G. L. Griffin, Surf. Sci. 1985, 155, 387.
[32] S. Akhter, W. H. Cheng, K. Lui, H. H. Kung, J. Catal. 1984, 85,
437.
[1] V. E. Henrich, P. A. Cox, The Surface Science of Metal Oxides,
Cambridge University Press, 1996.
[2] S. A. French, A. A. Sokol, S. T. Bromley, C. R. A. Catlow, S. C.
Rogers, F. King, P. Sherwood, Angew. Chem. 2001, 113, 4569;
Angew. Chem. Int. Ed. 2001, 40, 4437.
[3] B. Denise, R. P. A. Sneeden, CHEMTECH 1982, 108.
[4] Y. B. Kagan, L. G. Liberov, E. V. Slivinskii, S. Loktev, G. I. Lin,
A. Y. Rozovskii, A. N. Bashikov, Dokl. Akad. Nauk SSSR 1975,
221, 1093.
[33] J. B. Hansen in Handbook of Heterogeneous Catalysis, Vol. 4
(Eds.: G. Ertl, H. Knꢁzinger, J. Weitkamp), Wiley-VCH,
Weinheim, 1997, p. 1856 – 1876, and references therein.
[34] M. Kurtz, N. Bauer, H. Wilmer, O. Hinrichsen, M. Muhler,
Chem. Ing. Tech. 2004, 76, 42; M. Kurtz, N. Bauer, H. Wilmer, O.
Hinrichsen, M. Muhler, Chem. Eng. Technol. 2004, 27, 1146.
[5] G. H. Graaf, E. J. Stamhuis, A. A. C. M. Beenackers, Chem. Eng.
Sci. 1988, 43, 3185.
[6] K. Klier, V. Chatikavanij, R. G. Herman, G. W. Simmons, J.
Catal. 1982, 74, 343.
[7] G. Liu, D. Willcox, M. Garland, H. H. Kung, J. Catal. 1984, 90,
139.
[8] T. Becker, C. Boas, U. Burghaus, C. Wꢁll, Phys. Rev. B 2000, 61,
4536.
[9] T. Becker, C. Boas, U. Burghaus, C. Wꢁll, J. Vac. Sci. Technol.
2000, 18, 1089.
[10] T. Becker, M. Kunat, C. Boas, U. Burghaus, C. Wꢁll, J. Chem.
Phys. 2000, 113, 6334.
[11] M. Kunat, S. Gil Girol, T. Becker, U. Burghaus, C. Wꢁll, Phys.
Rev. B 2002, 66, 081402.
[12] V. Staemmler, K. Fink, B. Meyer, D. Marx, M. Kunat, S. Gil
Girol, U. Burghaus, C. Wꢁll, Phys. Rev. Lett. 2003, 90, 106102.
[13] H. Wilmer, M. Kurtz, K. V. Klementiev, O. P. Tkachenko, W.
Grꢂnert, O. Hinrichsen, A. Birkner, S. Rabe, K. Merz, M. Driess,
C. Wꢁll, M. Muhler, Phys. Chem. Chem. Phys. 2003, 5, 4736.
[14] A. Wander, F. Schedin, P. Steadman, A. Norris, R. McGrath, T. S.
Turner, G. Thornton, N. M. Harrison, Phys. Rev. Lett. 2001, 86,
3811.
[15] B. Meyer, D. Marx, Phys. Rev. B 2003, 67, 035403.
[16] B. Meyer, Phys. Rev. B 2004, 69, 045416.
[17] The geometry optimizations were performed at the Hartree—
Fock level with the program package TURBOMOLE: R.
Ahlrichs, M. Bꢃr, M. Hꢃser, H. Horn, C. Kꢁlmel, Chem. Phys.
Lett. 1989, 162, 165.
[18] V. Staemmler, Theor. Chim. Acta 1977, 45, 89.
[19] U. Meier, V. Staemmler, Theor. Chim. Acta 1989, 76, 95.
[20] R. Fink, V. Staemmler, Theor. Chim. Acta 1993, 87, 129.
[21] S. F. Boys, F. Bernadi, Mol. Phys. 1970, 19, 553.
[22] T. Shimanouchi, Tables of Molecular Vibrational Frequencies
1972, National Bureau of Standards.
[23] To obtaine the zero-point energies for the different reaction
steps, the zero-point vibrations of the adsorbates and the
molecules remaining in the gas phase were calculated in the
harmonic approximation using B3LYP and a 6-31G* basis set:
Gaussian98 (RevisionA.11), M. J. Frisch, G. W. Trucks, H. B.
Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G.
Zakrzewski, J. A. Montgomery, R. E. Stratmann, J. C. Burant, S.
Dapprich, J. M. Millam, A. D. Daniels, K. N. Kudin, M. C. Strain,
O. Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi, B.
Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochterski, G. A.
Petersson, P. Y. Ayala, Q. Cui, K. Morokuma, D. K. Malick,
A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. Cioslowski,
J. V. Ortiz, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I.
Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A.
Al-Laham, C. Y. Peng, A. Nanayakkara, C. Gonzalez, M.
Challacombe, P. M. W. Gill, B. G. Johnson, W. Chen, M. W.
Wong, J. L. Andres, M. Head-Gordon, E. S. Replogle, J. A.
Pople, Gaussian, Inc., Pittsburgh, PA, 1998.
[24] J. C. Lavalley, J. Saussey, T. Rais, J. Mol. Catal. 1982, 17, 289.
[25] J. Saussey, J. C. Lavalley, J. Lamotte, T. Rais, J. Chem. Soc. Chem.
Commun. 1982, 17, 278.
[26] J. Saussey, J. C. Lavalley, T. Rais, A. Chakor-Alami, J. P.
Hindermann, A. Kinnemann, J. Mol. Catal. 1984, 26, 159.
2794
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 2790 –2794