Sites for methane activation on lithium-doped magnesium oxide surfaces

Article abstract of DOI:10.1002/anie.201310632

Density functional calculations yield energy barriers for H abstraction by oxygen radical sites in Li-doped MgO that are much smaller (12±6 kJmol^-1) than the barriers inferred from different experimental studies (80-160 kJmol^-1). This raises further doubts that the Li ⁺O^.- site is the active site as postulated by Lunsford. From temperature-programmed oxidative coupling reactions of methane (OCM), we conclude that the same sites are responsible for the activation of CH ₄ on both Li-doped MgO and pure MgO catalysts. For a MgO catalyst prepared by sol-gel synthesis, the activity proved to be very different in the initial phase of the OCM reaction and in the steady state. This was accompanied by substantial morphological changes and restructuring of the terminations as transmission electron microscopy revealed. Further calculations on cluster models showed that CH₄ binds heterolytically on Mg²⁺O ^2- sites at steps and corners, and that the homolytic release of methyl radicals into the gas phase will happen only in the presence of O ₂.

Full text of DOI:10.1002/anie.201310632

.
Angewandte
Communications
DOI: 10.1002/anie.201310632
ꢀ
C H Activation on MgO
Sites for Methane Activation on Lithium-Doped Magnesium Oxide
Surfaces**
Karolina Kwapien, Joachim Paier, Joachim Sauer,* Michael Geske, Ulyana Zavyalova,
Raimund Horn,* Pierre Schwach, Annette Trunschke,* and Robert Schlçgl
Dedicated to the MPI fꢀr Kohlenforschung on the occasion of its centenary
Abstract: Density functional calculations yield energy barriers
for H abstraction by oxygen radical sites in Li-doped MgO that
are much smaller (12 ꢁ 6 kJmolꢀ1) than the barriers inferred
from different experimental studies (80–160 kJmolꢀ1). This
raises further doubts that the Li+OCꢀ site is the active site as
postulated by Lunsford. From temperature-programmed oxi-
dative coupling reactions of methane (OCM), we conclude that
the same sites are responsible for the activation of CH4 on both
Li-doped MgO and pure MgO catalysts. For a MgO catalyst
prepared by sol–gel synthesis, the activity proved to be very
different in the initial phase of the OCM reaction and in the
steady state. This was accompanied by substantial morpho-
logical changes and restructuring of the terminations as
transmission electron microscopy revealed. Further calcula-
tions on cluster models showed that CH4 binds heterolytically
on Mg2+O2ꢀ sites at steps and corners, and that the homolytic
release of methyl radicals into the gas phase will happen only in
the presence of O2.
in chemical industry to natural gas, there is renewed interest
in the formation of higher hydrocarbons, for example by
oxidative coupling of methane (OCM):[6]
2 CH4 þ O2 ! C2H4 þ 2 H2O
ð1Þ
The simplest catalysts for this reaction, among a large
number of complex solid oxides, is Li-doped MgO.[7] Early,
Lunsford proposed that the active sites are OCꢀ radicals
neighbored to Li+, with Li+OCꢀ formally replacing Mg2+O2ꢀ,[8]
ꢀ
and that the C H bond is activated by homolytic splitting
involving hydrogen atom transfer to the OCꢀ sites:[9]
ꢀ
þ
Cꢀ
C
H3CꢀH þ ½O LiþꢂMgO ! H3C þ ½HO Li ꢂMgO
ð2Þ
However, there is also evidence that the Li+OCꢀ site may
ꢀ
not be the active site and that the C H bond may be
heterolytically split,[6] as Lunsford already mentioned in his
1995 review.[7] Recently, crucial ENDOR experiments showed
that in none of the powder catalysts that were run under an
OCM atmosphere Li+Oꢀ centers could be found,[10,11]
although they were detectable in Li-doped MgO single
crystals prepared by arc fusion of MgO/Li2CO3.[10] Instead,
by careful multi-method characterization,[11,12] Li addition was
found to lead to restructuring of the MgO surface exposing
steps and corner sites and high-index crystallographic surfaces
alien to pristine MgO. Studies of thin MgO films by surface
science techniques reached the same conclusion.[10]
T
aylorꢀs active site concept[1] has stimulated catalysis
research over almost a century, but it took many decades
until surface science identified low-coordinated atoms at step
edges as active sites of metal catalysts.[2] Subsequently, the
complex nature of active sites at supported metal[3] or metal
oxide catalysts[4,5] has been revealed by combined experi-
mental and computational studies. With the raw material shift
[*] Dr. K. Kwapien, Dr. J. Paier, Prof. Dr. J. Sauer
Institut fꢀr Chemie, Humboldt-Universitꢁt
Unter den Linden 6, 10099 Berlin (Germany)
E-mail: sek.qc@chemie.hu-berlin.de
Herein, we provide theoretical and further experimental
evidence that the Li+OCꢀ site is not the active site and
conclude that the activity of OCM catalysts is connected with
morphological features of the crystallites that form under
reaction conditions and depend on the synthesis process.
Lunsford already points to a discrepancy[7] between the
measured apparent activation energy for the formation of
methyl radicals (96 ꢁ 8 kJmolꢀ1)[13] and quantum chemical
calculations. In 2005 Catlow et al. used density functional
theory and periodic models to calculate the energy barrier of
H abstraction by an OCꢀ site at the (001) surface of Li-doped
MgO.[14] The barrier obtained, 74 kJmolꢀ1, is more or less in
apparent agreement with the above value and reported
barriers of 85 kJmolꢀ1 (CH4/CD4 isotope exchange)[15] and
90 kJmolꢀ1 (C2 hydrocarbon formation).[16] Microkinetic
simulations yielded significantly higher values, namely
147 kJmolꢀ1.[17] More recently, Li-doped MgO catalysts have
been found unstable under reaction conditions, and after 24 h
Dr. M. Geske, Dr. U. Zavyalova, Prof. Dr. R. Horn, P. Schwach,
Dr. A. Trunschke, Prof. Dr. R. Schlçgl
Fritz-Haber-Institut der Max-Planck-Gesellschaft
Faradayweg 4-6, 14195 Berlin (Germany)
Prof. Dr. R. Horn
Current address: Institut fꢀr Chemische Reaktionstechnik
Technische Universitꢁt Hamburg-Harburg
Eißendorfer Strasse 38, 21073 Hamburg (Germany)
[**] This work has been supported by Deutsche Forschungsgemein-
schaft within the Cluster of Excellence “Unifying Concepts in
Catalysis”. We thank Wiebke Frandsen for recording of MgO images
by transmission electron microscopy. K.K. thanks the International
Max Planck Research School “Complex Surfaces in Materials
Science” for a fellowship. J.S. is grateful for a Miller Visiting
Professorship at Berkeley, during which parts of this paper have
been written.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201310632.
8774
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 8774 –8778

Angewandte
Chemie
time on stream barriers between 89 and 160 kJmolꢀ1 have
been measured, depending on the synthesis method.[18]
We report quantum chemical calculations that go beyond
limitations of previous work as far as models for the active
site, localization method of the transition structure and
accuracy of the quantum chemical approach are concerned.
Our predicted energy barriers, between 7 ꢁ 6 and 27 ꢁ
6 kJmolꢀ1, are surprisingly low compared to the experimental
results, which suggests that the Li+OCꢀ site is not the active
site.
Support for this conclusion comes from temperature
programmed reaction experiments, which show that on both
Li-doped MgO and pure MgO catalysts conversion of CH4
and O2 starts at about 4108C and formation of C2 species at
about 5408C. The difference is that in this initial phase of the
reaction Li-MgO is far more active and selective in forming C2
coupling products than pure MgO. We conclude that the
reaction pathways are the same for both materials and that
the same active sites are present. The role of Li-doping is
increasing the number of active sites, which is most likely due
to changes of the morphology connected with the formation
of a larger number of low-coordinate O2ꢀ ions at edges,
corners, and kinks.[10–12]
Figure 1. From top to bottom: Cluster model, constraint cluster model
For further kinetic studies, a pure MgO model catalyst has
been synthesized by a sol–gel process and characterized by
transmission electron microscopy (TEM). The catalyst activ-
ity was very different in the initial phase of the OCM reaction
and in the steady state, and this was accompanied by
substantial morphological changes. In the steady state, the
apparent activation barrier for CH4 conversion is 133 ꢁ
2 kJmolꢀ1, within the 80 to 160 kJmolꢀ1 range inferred
before for Li-doped MgO.[15–18]
embedded in a periodic array of point charges, and periodic model of
an Li+OCꢀ site on the MgO(001) terrace. Left: active site structure,
right: transition structure for H abstraction from CH4. O red, Mg black,
Li green, C yellow, H gray, spin density blue.
structure, the spin localizes at the surface oxygen ion above
the lithium ion for all models. In the initial state, the
unconstrained model shows delocalization over two addi-
tional oxygen ions. This is rectified when embedding the
cluster in a periodic array of point charges and a finite number
of full ion pseudopotentials, which also lowers the apparent
barrier by 20 kJmolꢀ1 (Table 1, terrace). The periodic model
barrier is only 14 kJmolꢀ1 lower than the embedded cluster
barrier and 12 kJmolꢀ1 of this difference are due to relaxing
a larger number of atoms in the periodic model. This we
gather from the single point embedded cluster result at the
structure obtained for the periodic model.
To further explore possible sites for CH4 activation on Li-
free MgO catalysts, we have examined the interaction of CH4
with morphological defects (steps, corners) by DFT. The
ꢀ
calculations showed that the C H bond adds heterolytically
onto an Mg2+O2ꢀ ion pair, but that homolytic release of
methyl radicals in the gas phase is only likely in the presence
of O2.
Figure 1 shows both cluster and periodic models adopted
for the Li+OCꢀ site at the MgO (001) surface terrace. The
LiO(MgO)8 cluster has the topology of a two-layer cut-out
from the surface. In the constraint cluster model we limit
structure relaxation to the Li+ and OCꢀ ions in the center of the
cluster, whereas the positions of all other atoms are fixed at
the positions they have in the MgO bulk. In our four-layer
slab model, all of the ion positions are relaxed except those in
the two bottom layers. The B3LYP hybrid density func-
tional[19,20] has been used, which reproduces within 6 kJmolꢀ1
We conclude that for Li+OCꢀ terrace sites, the activation
barrier is as low as 12 ꢁ 6 kJmolꢀ1 (Table 1). The much higher
Table 1: Apparent energy barriers (kJmolꢀ1, B3LYP functional) for
hydrogen abstraction from methane at Li+OCꢀ sites at the MgO(001)
terrace and corner sites.
Model
Terrace
Corner 1
Corner 2
ꢀ
CCSD(T) coupled cluster results of C H bond splitting
+
barriers for CH4 at OCꢀ sites of (MgO)nC clusters.[21,22]
cluster, free
61.3
57.3
18.9
22.8
21.7
22.4
cluster, constraint[a]
cluster, embedded[a,b]
periodic model
+dispersion
CCSD(T) is considered to be chemically accurate. For gas-
phase metal oxide clusters featuring radical sites, B3LYP has
also been shown to correctly predict for which clusters
hydrogen abstraction can be observed by mass spectrometry
and for which not.[21,23]
Figure 1 shows the geometric structures of the active site
and the transition state (for bond distances, see the Support-
ing Information) as well as the spin densities. In the transition
41.2 [29.3][c]
26.8 (35.1)[d]
12.3 (26.8)[d]
29.8
29.5
(23.7)[d]
(21.8)[d]
(52.2)[d]
(42.1)[d]
[a] Only the surface O ion and the Li ion beneath are allowed to move.
[b] Periodic electrostatic embedded cluster model.[24] [c] Single-point
calculation at the B3LYP structure obtained for the periodic model.
[d] Single-point hybrid B3LYP(cluster):PBE(periodic) calculations at the
PBE structure obtained for the periodic model.
Angew. Chem. Int. Ed. 2014, 53, 8774 –8778
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8775

.
Angewandte
Communications
previous estimates (74 kJmolꢀ1)[14] we ascribe to neglect of
dispersion and an incomplete optimization for the transition
structure in that study. For different morphological positions,
for example, at corners, depending on the model the barriers
may be 5 kJmolꢀ1 lower (“corner 1”) or 15 kJmolꢀ1 higher
(“corner 2”) than for terrace sites (Table 1, last row, values in
brackets).
Temperature-programmed reaction measurements were
carried out on 150 mg of pure MgO and 5 wt% Li-doped
MgO, both prepared by gel combustion synthesis as described
previously.[11] Briefly, glycerol as fuel was mixed with aqueous
solutions of LiNO3 and Mg(NO3)2. After water was removed
by evaporation, a highly energetic combustible gel was
obtained in which Li+ and Mg2+ ions were molecularly
dispersed. Upon ignition, the gel combusted vigorously to the
oxides followed by rapid thermal quenching. However,
despite rapid combustion and rapid thermal quenching,
which may lead to structures far from equilibrium, it was
not possible to detect Li+Oꢀ defects in any material obtained
this way. As discussed in detail in Ref. [11], CW-EPR,
ENDOR and DR-UV/Vis, all established diagnostic methods
for Li+Oꢀ defects,[10,25] were applied, but no method could
detect Li+Oꢀ defects in any of our materials. Using optical
spectroscopy morphological defects, that is, low-coordinate
O
2ꢀ ions at edges, corners, and kinks of the MgO surface have
been identified that arise from the volatilization of the Li
component being initially present in the MgO sample as solid
solution.
The pure MgO and the Li-doped MgO catalysts were
subjected to a temperature ramp of 3 Kminꢀ1 in a plug flow
reactor using an OCM mixture consisting of 10 mLminꢀ1
CH4, 1.25 mLminꢀ1 O2, and 2 mLminꢀ1 Ar as internal
standard. Figure 2 shows the results for CH4, O2, C2H4, and
C2H6; the corresponding data for CO and CO2 formation are
given in the Supporting Information. On both catalysts, O2
and CH4 conversion into CO2 commences at about 4108C
with CO2 as the far dominant oxidation product. Arrhenius
plots on O2 consumption (see the Supporting Information)
give very similar apparent activation energies of 205 and
184 kJmolꢀ1 on pure MgO and Li-MgO, respectively, indicat-
ing a similar mechanism of initial O2 activation on both
materials. CO, C2H6, and C2H4 formation begins on both
materials at 5408C, but Li-MgO is far more active and
selective in forming C2 coupling products than pure MgO.
TEM investigations of MgO particles obtained by a sol–
gel process show substantial morphology changes during time
on-stream. Figure 3 shows a TEM image of the MgO catalyst
used for the data in Table 2 after 6 h and after 230 h time on-
stream. Careful transfer into the TEM under exclusion of air
reveals the restructuring of the termination, losing the (100)
Figure 2. Flow rates of CH4, O2 (top), and C2H6, C2H4 (bottom) in
temperature-programmed oxidative methane coupling on pure MgO
and 5 wt% Li-doped MgO.
Figure 3. TEM images of pure MgO after 6 h (left) and after reaching
steady state (230 h, right).
orientation from the fresh MgO for a rough termination
structure. In both cases, substantial numbers of non-terrace
sites are present that expose Mg sites with lower coordination
than fivefold and may serve as active sites. The diffuse
termination of the 230 h used sample indicates the presence of
additional terminating species, such as OH groups that may
block many of such sites for methane adsorption.
Table 2: Performance of the Li-free MgO catalyst in the oxidative
coupling of methane.
Rate
Rate
X(CH4)
S(C2)
ꢀ1
ꢀ2
[mmolsꢀ1 gcat
]
[mmolsꢀ1 mcat
]
[%][a]
[%][a]
initial state
final state
214.4
8.57
5.56
1.26
26.04
4.70
29.84
13.85
Concomitantly with the morphological changes, the
catalyst activity was very different in the initial phase of the
OCM reaction and the steady state (Table 2). CH4 conversion
[a] X: conversion, S(C2): selectivity to ethane and ethylene.
8776 www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 8774 –8778

Angewandte
Chemie
and C2 selectivity changed from 26 and 30%, respectively, to 5
and 14%, respectively. This is connected with substantial
restructuring as the rate per catalyst weight reduces by factor
of 25, whereas the rate per surface area reduces by a factor of
4.4 only.
OCM process have already been discussed.[7] Our model
calculations are indeed strong evidence that heterolytic
chemisorption of CH4 on MgO in the presence of O2 becomes
energetically feasible on morphological defects such as steps
or corners and plays a significant role in the OCM process.
Currently, we are further investigating possible mechanisms
based on this idea. Whereas quantum chemical calculations
on realistic models as presented here can provide reliable
information on individual reaction steps, comparison with
experimental kinetic data requires micro-kinetic simulations
that use these data as input.
As a model for morphological defects (steps, corners) on
Li-free MgO catalysts we adopt the (MgO)9 cluster. Figure 4
Our experiments (see also Ref. [31])in combination with
theoretical studies suggest a different role of non-reducible
oxides as catalysts on methane activation. Their role is not to
provide and receive back electrons as with reducible oxide
catalysts, but merely to stay inert with its own electronic
system and just bring together the reactants allowing
exchanging redox equivalents directly between themselves.
Such a function is expressed in the designation “catalyst”
meaning the bringing together of reactants.
Figure 4. Reaction energy diagram for chemisorption of CH4 onto
ꢀ
corner/edge sites of a Mg9O9 cluster showing C H bond addition on
an Mg2+O2ꢀ pair. B3LYP energies are given in kJmolꢀ1, for energies
including zero-point vibrational contributions see Supporting Informa-
tion. EC encounter complex, TS transition state, IN intermediate;
colors as in Figure 1.
Received: December 7, 2013
Revised: February 19, 2014
Published online: April 23, 2014
ꢀ
Keywords: active sites · C H activation ·
density functional calculations · Li-doped MgO ·
magnesium oxide
.
shows the energy diagram for the reaction with CH4.
ꢀ
Chemisorption occurs by heterolytic addition of the C H
bond onto an Mg2+O2ꢀ pair:
½Mg2þO2ꢀꢂMgO þ HꢀCH3! ½HOꢀðMgꢀCH3ÞþꢂMgO
[1] H. S. Taylor, Proc. R. Soc. London Ser. A 1925, 108, 105 – 111.
[2] T. Zambelli, J. Wintterlin, J. Trost, G. Ertl, Science 1996, 273,
1688 – 1690.
ð3Þ
ꢀ
Cꢀþ
Cꢀ
CH3
! ½HO Mg ꢂMgO
þ
[3] M. Behrens, F. Studt, I. Kasatkin, S. Kꢁhl, M. Hꢂvecker, F.
Abild-Pedersen, S. Zander, F. Girgsdies, P. Kurr, B.-L. Kniep, M.
Tovar, R. W. Fischer, J. K. Nørskov, R. Schlçgl, Science 2012,
336, 893 – 897.
[4] M. V. Ganduglia-Pirovano, C. Popa, J. Sauer, H. L. Abbott, A.
Uhl, M. Baron, D. Stacchiola, O. Bondarchuk, S. Shaikhutdinov,
H.-J. Freund, J. Am. Chem. Soc. 2010, 132, 2345 – 2349.
[5] K. Amakawa, L. Sun, C. Guo, M. Hꢂvecker, P. Kube, I. E. Wachs,
S. Lwin, A. I. Frenkel, A. Patlolla, K. Hermann, R. Schlçgl, A.
Trunschke, Angew. Chem. 2013, 125, 13796 – 13800; Angew.
Chem. Int. Ed. 2013, 52, 13553 – 13557.
[6] E. V. Kondratenko, M. Baerns, Handbook of Heterogeneous
Catalysis, Vol. 6 (Ed.: H. K. G. Ertl, F. Schꢁth, J. Weitkamp),
Wiley-VCH, Weinheim, 2008.
[7] J. H. Lunsford, Angew. Chem. 1995, 107, 1059 – 1070; Angew.
Chem. Int. Ed. Engl. 1995, 34, 970 – 980.
The surface species formed are an OH group (by
protonation of O2ꢀ) and a Grignard type Mg-methylate (by
addition of CH3 to Mg2+). Such reactions have been
ꢀ
discussed before,[26] for example, for low-coordinate sites on
g-alumina.[27] The barrier for this slightly endothermic reac-
tion is at the low side of typical barriers for OCM reactions,
but releasing a methyl radical to the gas phase is barrierless
but needs as much as 228 kJmolꢀ1. However, the unpaired
electron formed in this reaction facilitates O2 chemisorption
as a superoxide species:
ꢀ
Cꢀþ
Cꢀ
½HO Mg ꢂMgO þ O2 ! ðO2 Þ½HOꢀMg2þꢂMgO
ð4Þ
This reaction is very exothermic (ꢀ191 kJmolꢀ1,[28] so that the
[8] D. J. Driscoll, W. Martir, J. X. Wang, J. H. Lunsford, J. Am.
Chem. Soc. 1985, 107, 58 – 63.
overall process:
[9] T. Ito, J. Wang, C. H. Lin, J. H. Lunsford, J. Am. Chem. Soc. 1985,
107, 5062 – 5068.
Cꢀ
C
½Mg2þO2ꢀꢂMgO þ HꢀCH3 þ O2 ! ðO2 Þ½HOꢀMg2þꢂMgO þ CH3 ð5Þ
[10] P. Myrach, N. Nilius, S. V. Levchenko, A. Gonchar, T. Risse, K. P.
Dinse, L. A. Boatner, W. Frandsen, R. Horn, H. J. Freund, R.
Schlçgl, M. Scheffler, ChemCatChem 2010, 2, 854 – 862.
[11] U. Zavyalova, G. Weinberg, W. Frandsen, F. Girgsdies, T. Risse,
K. P. Dinse, R. Schloegl, R. Horn, ChemCatChem 2011, 3, 1779 –
1788.
[12] U. Zavyalova, M. Geske, R. Horn, G. Weinberg, W. Frandsen, M.
Schuster, R. Schlçgl, ChemCatChem 2011, 3, 949 – 959.
[13] M. Xu, C. Shi, X. Yang, M. P. Rosynek, J. H. Lunsford, J. Phys.
Chem. 1992, 96, 6395 – 6398.
becomes almost thermal neutral (DE = 37 kJmolꢀ1). There is,
of course, an entropy penalty, as two gas-phase species are
converted into one, but reaction (5) also yields a superoxo
surface species that is more reactive than O2 in the gas phase.
Future studies should also consider possible reactions of
dioxygen with the Mg-methylate, which may also lead to
methyl radicals and surface peroxo species.[29,30] Pros and cons
of reaction (5) as a possible source of methyl radicals in the
Angew. Chem. Int. Ed. 2014, 53, 8774 –8778
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8777

.
Angewandte
Communications
[14] C. R. A. Catlow, S. A. French, A. A. Sokol, J. M. Thomas, Philos.
Trans. R. Soc. London Ser. A 2005, 363, 913 – 936.
[15] V. S. Muzykantov, A. A. Shestov, H. Ehwald, Catal. Today 1995,
24, 243 – 244.
[16] M. Y. Sinev, V. Y. Bychkov, V. N. Korchak, O. V. Krylov, Catal.
Today 1990, 6, 543 – 549.
[17] J. Sun, J. W. Thybaut, G. B. Marin, Catal. Today 2008, 137, 90 –
102.
[18] S. Arndt, U. Simon, S. Heitz, A. Berthold, B. Beck, O. Gçrke,
J. D. Epping, T. Otremba, Y. Aksu, E. Irran, G. Laugel, M.
Driess, H. Schubert, R. Schomꢂcker, Top. Catal. 2011, 54, 1266 –
1285.
[23] S. Feyel, J. Dçbler, R. Hçckendorf, M. K. Beyer, J. Sauer, H.
Schwarz, Angew. Chem. 2008, 120, 1972 – 1976; Angew. Chem.
Int. Ed. 2008, 47, 1946 – 1950.
[24] A. M. Burow, M. Sierka, J. Dçbler, J. Sauer, J. Chem. Phys. 2009,
130, 174710.
[25] M. M. Abraham, Y. Chen, L. A. Boatner, R. W. Reynolds, Phys.
Rev. Lett. 1976, 37, 849 – 852.
[26] C. Copꢃret, Chem. Rev. 2010, 110, 656 – 680.
[27] R. Wischert, P. Laurent, C. Copꢃret, F. Delbecq, P. Sautet, J. Am.
Chem. Soc. 2012, 134, 14430 – 14449.
[28] D. Ricci, G. Pacchioni, P. V. Sushko, A. L. Shluger, Surf. Sci.
2003, 542, 293 – 306.
[19] A. D. Becke, J. Chem. Phys. 1993, 98, 5648 – 5652.
[20] C. Lee, W. Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785 – 789.
[21] K. Kwapien, M. Sierka, J. Dçbler, J. Sauer, ChemCatChem 2010,
2, 819 – 826.
[29] A. G. Davies, B. P. Roberts, Acc. Chem. Res. 1972, 5, 387 – 392.
[30] P. J. Bailey, R. A. Coxall, C. M. Dick, S. Fabre, L. C. Henderson,
C. Herber, S. T. Liddle, D. LoroÇo-Gonzꢄlez, A. Parkin, S.
Parsons, Chem. Eur. J. 2003, 9, 4820 – 4828.
[22] K. Kwapien, Active Sites for Methane Activation in MgO and
Li-doped MgO, Doctoral thesis, Humboldt University, 2011.
[31] P. Schwach, W. Frandsen, M. Willinger, R. Schlçgl, A. Trunschke,
in preparation 2013.
8778
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 8774 –8778