Methane coupling over magnesium oxide: How doping can work

Article abstract of DOI:10.1002/anie.201305470

Electronic doping of magnesium oxide catalysts has an effect on the oxidative coupling of methane. Highly active sites can be created by co-modification of MgO with iron and gold in ppm quantities. Copyright

Full text of DOI:10.1002/anie.201305470

Angewandte
Chemie
DOI: 10.1002/anie.201305470
À
C H activation
Methane Coupling over Magnesium Oxide: How Doping Can Work**
Pierre Schwach, Marc Georg Willinger, Annette Trunschke,* and Robert Schlçgl
Dedicated to Professor Helmut Schwarz on the occasion of his 70th birthday
The functionalization of methane remains a challenging
target from an academic as well as an industrial point of
In the present work, we put the concept to test with
powder catalysts working at high temperature (T= 1023 K).
We synthesized doped magnesium oxide as it has been
investigated frequently in OCM.[7] Pure magnesium oxide is
deactivated at this temperature quite rapidly and reaches
a stationary state after a few minutes to several hours
depending on the applied contact time and the initial
nanostructure of the magnesium oxide. In the stationary
state, a low, but constant yield of the coupling products ethane
and ethene is obtained over smooth, rounded MgO parti-
cles.[8]
We introduced Fe in ppm quantities into MgO. The
synthesis of Fe-doped polycrystalline magnesium oxide in
which the Fe dopant is homogeneously distributed over the
entire bulk is, however, quite challenging and requires highly
sensitive analytical techniques for verification. The issue is
distributing the dopant throughout the bulk of MgO in such
a way that no precipitates or segregated nanostructured
dopant phases occur under the drastic reaction conditions.
Only then can the validity of the electronic doping concept
put forward by Nilius and Freund et al. be tested without
interference from other catalytic actions of secondary phases.
The presence of Fe atoms on the surface may introduce
additional redox chemistry into the activation mechanism of
methane, but, even in an ideal solid solution, terminal Fe
atoms cannot be avoided. Therefore, the Fe-MgO catalyst was
modified by subsequent adsorption of highly dispersed gold
on the surface. An Au-MgO catalyst[9] was included in the
study for reference.
The catalysts were synthesized by hydrothermal treatment
of MgO in the presence of aqueous solutions of FeSO4,
HAuCl4, and a mixture of the two solutions in a microwave
autoclave at 483 K and 10 bar for 3 h, followed by annealing
in flowing Ar at 1173 K for 3 h. In order to remove potentially
segregated transition metal and metal oxide particles, the
obtained solids were treated with aqua regia and nitric acid,
respectively, and annealed again in Ar at 1123 K for 3 h. X-ray
diffraction reveals that the lattice constants of the three doped
catalysts do not vary significantly, but the size of the
coherently scattering domains (reported from full-pattern
X-ray diffraction (XRD) analysis as the volume-weighted
mean column length based on integral breadth (LVol-IB))
differs. The largest crystalline domains are found for Au-Fe-
MgO, which is also reflected in the lowest specific surface area
(Table 1). Structural investigation by transmission electron
microscopy (TEM) reveals typical small aggregates consisting
of cube-shaped particles that are intergrown and connected
mainly along shared faces (Figure 1). The domain size
determined by XRD is similar for Fe-MgO and Au-MgO
(Table 1) and ranges between 5 and 100 nm according to
view.[1] New concepts for the catalytic activation of C H
À
bonds are needed[1a] to overcome the current limitations in
selectivity, which hamper the broad application of methane
coupling for the production of olefins from sustainable
resources like natural gas and organic waste. Among the
various inorganic materials that have been evaluated as
heterogeneous catalysts for the oxidative coupling of methane
(OCM), alkaline earth oxides doped with alkali elements or
transition metal ions have received particular attention.[2]
High reaction temperatures (973–1273 K) are needed. How-
À
ever, the high temperature is not required for C H activation,
which may aided by coordinatively unsaturated sites already
at low temperature,[3] but rather for recovery of an active
catalyst surface free of adsorbed hydroxides and carbonates.
Under the harsh reaction conditions, oxide catalysts like Li-
MgO undergo fast deactivation due to sintering promoted by
water, which is an unavoidable reaction product.[4] Oxygen
and magnesium vacancies (VO’’, VMg’’) are involved in the
sintering of MgO by facilitating reconstructions due to the
enhanced diffusion of lattice ions, which is fast in any case at
such high temperatures.[5] On the other hand, vacancies may
have an impact on activity and selectivity in catalysis.
However, point defects have not detected so far over MgO
under realistic working conditions in OCM.
In a recent study, Nilius, Freund, and co-workers et al.
provide evidence that strongly bound O2À species, precursors
of dissociatively adsorbed O2, are formed on highly ordered
CaO films doped with Mo2+ ions. The results indicate that
molecular activation on doped oxides does not require any
surface structural defects.[6] Accordingly, it is suggested that
the activation of methane on smooth surfaces of transition-
metal-doped wide-band-gap oxides may involve such acti-
vated oxygen species.
[*] P. Schwach, Dr. M. Willinger, Dr. A. Trunschke, Prof. Dr. R. Schlçgl
Abteilung Anorganische Chemie
Fritz-Haber-Institut der Max-Planck-Gesellschaft
Faradayweg 4–6, 12489 Berlin (Germany)
E-mail: trunschke@fhi-berlin.mpg.de
Homepage: http://www.fhi-berlin.mpg.de/acnew/welcome.epl
[**] We thank M. Hashagen, F. Rybicki, Dr. F. Girgsdies, Dr. M.
Eichelbaum, and Dr. O. Timpe for experimental support and
scientific discussions. This work was conducted in the framework of
the COE “UniCat” (www.unicat.tu-berlin.de) of the German Science
Foundation.
Supporting information for this article (including detailed infor-
mation on catalyst synthesis, characterization, and catalytic tests) is
available on the WWW under http://dx.doi.org/10.1002/anie.
201305470.
Angew. Chem. Int. Ed. 2013, 52, 11381 –11384
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
11381

.
Angewandte
Communications
with respect to the coordination environment of Fe3+ (d5)
ions, and potentially appearing Fe2+ (d6) ions. The spectra of
the gold-containing catalysts exhibit the characteristic surface
plasmon mode around 520 nm (19230 cmÀ1) that is ascribed
to gold nanoparticles.[11] The broad appearance of the band
suggests that Au occurs in different particle sizes ranging from
atomic dimensions to Au particles with a diameter of about
10 nm.[12] Indeed, very rare and scarcely distributed isotropic
gold nanoparticles can be found on the surface of the MgO
particles by TEM.
Table 1: General characteristics of the doped MgO catalysts and
normalized formation rates of coupling products ethane and ethene in
OCM of methane.
Catalyst
Fe-MgO
Au-Fe-MgO
Au-MgO
Batch ID[a]
cFe[b] [ppm]
a[c] [ꢀ]
14696
654
4.214(1)
46(1)
32.0
5.12
0.16
6%
14593
402
4.213(1)
67(1)
19.9
13.55
0.68
7%
14588
–
4.214(1)
47(1)
25.6
0.68
0.03
–
L
VolÀIB[c] [nm]
A[d] [m2 gÀ1
]
rC2[e] [mmolsÀ1 gÀ1
]
rC2[e] [mmolsÀ1 m2]
Temperature-programmed reduction (TPR) and oxida-
tion (TPO) cycles were performed to analyze the redox
properties of the Fe-containing MgO catalysts (Figures S3
and S4). In the first run, the catalysts were heated in inert gas
and evolution of hydrogen was observed in a temperature
range between 573 and 673 K and under isothermal con-
ditions at 1073 K; the results indicated the formation of point
defects in the bulk of Fe-MgO and Au-Fe-MgO during the
thermal pretreatment.[13] The hydrogen-consumption profile
of the first TPR run differs from that of the second and third
runs, which were performed in each case after intermediate
temperature-programmed oxidation. In contrast, the second
and third TPR profiles are identical, indicating the high
reversibility, stability, and absence of segregation processes
after initial stabilization during the first cycle. This applies to
Fe-MgO (Figure S3) as well as to Au-Fe-MgO (Figure S4),
indicating that the catalysts are comparable, in particular,
because the bulk oxide properties, which were sensed by
redox probing, are quite similar and in agreement with the
XRD results. The first TPR run reveals a dopant-induced
difference between the two catalysts. In addition to the two
high-temperature peaks around 810–840 K and 1050–1060 K,
Fe-MgO exhibits an additional hydrogen-consumption peak
at low temperature (513 K) arising from easily reducible Fe
species. The peak does not appear again in subsequent cycles
indicating the dissolution of the species into the bulk where it
is protected from reoxidation. The amount of consumed
hydrogen proves that only a minor fraction of the low-level
iron doping is susceptible to reduction on the surface (Table 1
and Table S1); this indicates the homogeneous distribution of
the dopant within the bulk of magnesium oxide. The fraction
of reducible iron is initially higher in Fe-MgO than in Au-Fe-
MgO, indicating that surface iron species are shielded by
topping gold species.
The low yield of coupling products in OCM as presented
in Figure 2 was deliberately measured at short contact times
to prevent full oxygen conversion and thus to allow a mean-
ingful kinetic comparison. Gold seems to block the active sites
on the surface of MgO since Au-MgO shows only negligible
activity. This is ascribed to the propensity of gold to adsorb at
step edges of the MgO surface and thus cover the active sites
related to edges.[14] Fe-MgO is more active (Figures S5 and
S6) and selective (Figure S7) resulting in higher yield. The
behavior of Fe-MgO at the beginning of the experiment is
attributed to the presence of surface redox-active iron species,
in accordance with the observations made by TPR. Further
increase in methane conversion and selectivity is achieved by
codoping with Fe and Au. The synergistic effect of the two
transition metal additives becomes even more apparent by
Redox-active Fe[f]
[a] Catalyst ID for clear identification of the batch. [b] Measured by
atomic absorption spectroscopy (AAS). [c] Lattice constant of MgO
determined by XRD. [d] Specific surface area calculated by applying the
BET equation. [e] Formation rate of the coupling products ethane and
ethane measured after 4 h time on stream; reaction conditions:
T=1023 K, W/F=0.033 gsmLÀ1, CH4/O2/N2 =3:1:1. [f] Percentage of
Fe that is accessible at the surface as estimated by temperature-
programmed reduction (see Table S1).
Figure 1. a) HRTEM image showing the stepped surface of Au-Fe-
MgO. b) HAADF STEM image with characteristic bright contrast at
steps and edges due to decoration with heavy atoms. c) Columns of
heavy atoms can also be seen in thin regions in HRTEM.
TEM (Figure 1). The slightly lower specific surface area of
Au-MgO indicates that its surface is smoother than that of Fe-
MgO.
The UV/Vis spectra of Fe-MgO and Au-Fe-MgO are
dominated by an intense absorption in the range of the ligand-
to-metal charge-transfer (LMCT) bands (Figure S1). The
absorption maximum near 288 nm (34723 cmÀ1) is attributed
to isolated Fe3+ ions with octahedral coordination, in other
words, on Mg2+ lattice positions.[10] EPR spectroscopy con-
firms the occurrence of Fe3+ in cubic symmetry sites in MgO
(Figure S2). Contributions of oligomeric species and iron
oxide nanoparticles to the UV/Vis spectra cannot be excluded
due to the complex fine structure and the extended tail of the
bands towards decreasing energy. However, no aggregates of
iron were detected by TEM, indicating high Fe dispersion.
Since the Fe concentration is low, the weak absorption due to
d–d transitions in the range between 330 nm (30000 cmÀ1
)
and 1000 nm (10000 cmÀ1) gives featureless, very weak, and
broad bands that do not allow unambiguous interpretation
11382 www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 11381 –11384

Angewandte
Chemie
Figure 2. Yield of ethane and ethene in the oxidative coupling of
methane as a function of time on stream (TOS) at T=1023 K;
W/F=0.0167 gsmLÀ1; CH4/O2/N2 =3:1:1.
Figure 3. a) HAADF STEM and b) TEM image with localized contrast
variations due to defects caused by bulk doping as indicated by
arrows. c) HRTEM image of doped MgO particles; the lattice rotation
due to strain is shown in (d).
comparison of the rate of C2 formation normalized to the
specific surface area of the catalyst (Table 1). After a forma-
tion period with increasing activity, Au-Fe-MgO shows stable
activity at the time scale of the present experiment. This
behavior is surprising and novel for alkaline earth metal oxide
based catalysts.
active sites as indicated clearly by the temporal evolution of
the catalytic activity shown in Figure 2 and Figures S5–S7.
The inverse trend for the codoped system as compared to the
other systems points to the formation of active sites for
oxygen activation in Au-Fe-MgO during time on stream,
whereas the other systems are deactivated through the loss of
monatomic step sites following surface transformation caused
by the reaction products water and CO2. The stable minimal
activity of the Au-MgO system marks the intrinsic activity of
active sites in the present MgO catalyst that are not associated
with step edges. Compared to this activity (see Table 1) the
electronic doping by Fe increases the rate of C2 formation by
over an order of magnitude.
For the classical Au-MgO system it was found that small
doping levels produce metallic and some chemically active
gold species, which decorate steps and thus reduce the activity
of the parent MgO exactly as found in the present study.[9]
Only at much higher doping levels were catalytic effects
described and associated with Au particles and defect
formation in MgO. In the present study two types of gold
were evidenced based on the catalytic results: In single doped
Au-MgO only the step-decorating poisoning effect was found.
Through the presence of subsurface iron species in the
codoped system the edge-decorating effect was massively
overruled by the beneficial effect of creating structurally
stable novel active sites at terraces of the MgO without
introducing vacancies that may destabilize the system at
longer time on stream. This second gold species apparently
arises from a significant gold–support interaction stemming
from the strain in the MgO due to iron doping. The possible
effects of isotropic unstrained gold particles are negligible in
the present study due to their very low abundance as
a consequence of the synthesis strategy. Electronic doping
of MgO terraces is also achieved through iron dissolution
only, but with significant lower effectiveness. It is tempting to
conclude that the codoping creates highly active sites for
An explanation may be provided by electron microscopy.
High-angle angular dark-field scanning transmission electron
microscopy (HAADF STEM), which is sensitive to variations
in atomic weight can be used to locate the heavier dopant
atoms in the MgO matrix. As can be seen in Figure 1, corners
and edges are characterized by brighter contrast, indicating
the presence of either Au or Fe. This is confirmed by HRTEM
images recorded from thin regions of MgO crystals, where
rows of strongly scattering atoms are detected at surface steps
even after catalytic testing (Figure 1c and Figure S8). These
atoms will suppress the action of steps as active sites and thus
poison the Au-MgO catalyst, which does not have sites caused
by electronic doping. In addition, investigation of the bulk
structure by HAADF STEM and TEM reveals the presence
of local strain, causing particular contrast variations such as
indicated in Figure 3a,b. Analysis of the lattice fringes[15]
shows doping-induced strain and lattice rotation, as visualized
in Figure 3c,d. Due to the low concentration, energy-dis-
persive X-ray (EDX) elemental analysis is not sensitive
enough to clarify whether the defects are caused by gold or
iron incorporation into the MgO lattice.
The sintering behavior of Au-Fe-MgO is similar to that of
undoped MgO. Initially, the surface of the cube-shaped
particles exposes stepped and atomically flat planes
(Figure 1). Steps are usually in the range of half (one atom)
and single unit cell height (Figure 1a). During the catalytic
reaction, the abundance of such small steps decreases and
larger steps form. Despite these sintering phenomena, the
catalytic activity of the Au-Fe-MgO catalyst shows appreci-
able stability. This is a clear indication that the lasting activity
is not due to the conversion of methane at steps but rather at
sites located at terraces. Doping has changed the nature of the
Angew. Chem. Int. Ed. 2013, 52, 11381 –11384
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 11383

.
Angewandte
Communications
oxygen activation to give a peroxy species,[16] whereas iron
doping ends up with sites creating oxo species that are less
active in creating methyl radicals for OCM.
[1] a) H. Schwarz, Angew. Chem. 2011, 123, 10276; Angew. Chem.
Int. Ed. 2011, 50, 10096; b) C. Hammond, S. Conrad, I. Hermans,
ChemSusChem 2012, 5, 1668.
The purpose of the present work was to elucidate, much
along the demand of Hutchings,[17] how electronic promoters
may change the reaction pathway of the OCM reaction.
Stimulated through the clear-cut model observation of Nilius,
Freund et al. we synthesized a polycrystalline form of a model
compound exhibiting the same subsurface doping as evidently
achieved in the their model system. By carefully avoiding the
formation of nanoparticles of the transition metal oxide we
observed the predicted doping effect without interference
from total oxidation of the methane molecule.
It was not our intention to explore the potential of this
electronic effect with respect to selectivity and yield. We
rather wanted to demonstrate by a catalytic experiment that
an electronic doping effect predicted by chemical physics
exists for alkali earth oxides and affects oxygen activation. An
additional benefit is the amplification of this doping effect by
gold as a result of its specific interaction with MgO at iron-
modified sites. As the doping procedure is reproducible and
the levels of doping species are low, it is conceivable that the
concept of homogeneous electronic doping of alkali earth
oxides may find application for the stabilization of realistic
systems.
[2] U. Zavyalova, M. Holena, R. Schlçgl, M. Baerns, ChemCatChem
2011, 3, 1935.
[3] T. Ito, T. Watanabe, T. Tashiro, K. Toi, J. Chem. Soc. Faraday
Trans. 1 1989, 85, 2381.
[4] a) U. Zavyalova, M. Geske, R. Horn, G. Weinberg, W. Frandsen,
M. Schuster, R. Schlçgl, ChemCatChem 2011, 3, 949 – 959; b) U.
Zavyalova, G. Weinberg, W. Frandsen, F. Girgsdies, T. Risse,
K. P. Dinse, R. Schloegl, R. Horn, ChemCatChem 2011, 3, 1779 –
1788.
[5] D. Beruto, A. W. Searcy, R. Botter, M. Giordani, J. Phys. Chem.
1993, 97, 9201 – 9205.
[6] Y. Cui, N. Nilius, X. Shao, M. Baldofski, J. Sauer, H.-J. Freund,
Angew. Chem. 2013, DOI: 10.1002/ange.201305119; Angew.
Chem. Int. Ed. 2013, DOI: 10.1002/anie.201305119.
[7] J. H. Lunsford, Angew. Chem. 1995, 107, 1059 – 1070; Angew.
Chem. Int. Ed. Engl. 1995, 34, 970 – 980.
[8] J. S. J. Hargreaves, G. J. Hutchings, R. W. Joyner, C. J. Kiely, J.
Catal. 1992, 135, 576 – 595.
[9] K. Blick, T. D. Mitrelias, J. S. J. Hargreaves, G. J. Hutchings,
R. W. Joyner, C. J. Kiely, F. E. Wagner, Catal. Lett. 1998, 50, 211 –
218.
[10] G. Lehmann, Z. Phys. Chem. Neue Folge 1970, 72, 279 – 297.
[11] C. F. Bohren, D. R. Huffman, Absorption and Scattering of Light
by Small Particles, Wiley, New York, 1983.
[12] K. C. Grabar, R. G. Freeman, M. B. Hommer, M. J. Natan, Anal.
Chem. 1995, 67, 735 – 743.
Received: June 25, 2013
Published online: September 17, 2013
[13] B. V. King, F. Freund, Phys. Rev. B 1984, 29, 5814 – 5824.
[14] J. S. J. Hargreaves, G. J. Hutchings, R. W. Joyner, C. J. Kiely,
Catal. Today 1992, 13, 401 – 407.
[15] M. J. Hꢀtch, E. Snoeck, R. Kilaas, Ultramicroscopy 1998, 74, 131.
[16] P. Landon, P. J. Collier, A. J. Papworth, C. J. Kiely, G. J. Hutch-
ings, Chem. Commun. 2002, 0, 2058.
Keywords: alkaline earth oxides · defects · doping ·
heterogeneous catalysis · oxidative coupling
.
[17] G. J. Hutchings, Catal. Lett. 2001, 75, 1.
11384 www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 11381 –11384