Download Theoretical study of charge transfer interactions in methanol

Surface Science 504 (2002) 235–243
www.elsevier.com/locate/susc
Theoretical study of charge transfer interactions in
methanol adsorbed on magnesium oxide
M.M. Branda
a,*
, J.E. Peralta b, N.J. Castellani a, R.H. Contreras
b
a
b
Departamento de Fısica, Universidad Nacional del Sur, Av. Alem 1253, 8000 Bahıa Blanca, Argentina
Departamento de Fısica, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires, Buenos Aires, Argentina
Received 7 September 2001; accepted for publication 26 November 2001
Abstract
The adsorption of methanol on a magnesium oxide (1 0 0) surface was analysed using an appropriate cluster for
modeling the substrate. Results obtained employing the Hartree–Fock approach with different Gaussian basis sets were
compared. Several aspects of the adsorption mechanism were analysed using the natural bond orbitals method. It was
found that a H-bond type interaction is present between a surface oxygen anion and the hydrogen belonging to the
methanol hydroxyl group. This interaction produced a weakening of the O–H methanol bond. Ó 2002 Published by
Elsevier Science B.V.
Keywords: Ab initio quantum chemical methods and calculations; Catalysis; Physical adsorption; Magnesium oxides; Alcohols
1. Introduction
Catalysts are used to increase the rate of a given
chemical reaction. Despite the fact that the catalyst does not enter into the overall stoichiometric
balance, its accelerating effect is generally due to
the creation of a reaction pathway, usually multistep, characterized by a lower activation energy.
The appropriate selection of catalyst and reaction
conditions makes possible to lead the reaction
along a selected pathway to obtain the required
product. The selectivity of a given catalyst fundamentally depends on the type of intermediate
*
Corresponding author. Fax: +54-291-4595142.
E-mail address: [email protected] (M.M. Branda).
formed. About this matter, oxides show acidic and
basic sites in the surface, making possible the
formulation of catalytic reaction pathways for a
large variety of chemical reactions, including hydrogenations, dehydrogenations and oxidations
[1]. They constitute an interesting alternative to the
widely applied metallic catalysts, mainly because
the spatial configuration of ions on the surface can
present a great deal of situations, depending on the
crystallographic plane exposed.
Oxides employed in the chemical and petrochemical industry comprize several compounds
[1,2] such as single metal oxides (AlO3 , ZnO), alkaline or alkaline earth oxides (LiO, MgO), transition metal oxides (WO3 , NiO, MoO3 ), noble
metal oxides (CuO) and non-metal oxides (SiO2 ,
SnO2 ). Oxides can be simple or binary (NiMoO4 ,
BiPO4 ). Moreover, the transition metal oxides can
0039-6028/02/$ - see front matter Ó 2002 Published by Elsevier Science B.V.
PII: S 0 0 3 9 - 6 0 2 8 ( 0 2 ) 0 1 1 0 2 - 0
236
M.M. Branda et al. / Surface Science 504 (2002) 235–243
be stoichiometric or non-stoichiometric. As in
oxides the ions have different surface concentrations, acidic or basic powers and local atomic environments, different intermediate complexes can be
produced. Two types of oxygen species have been
identified at the surface of oxides, with electro
2
philic (O
2 , O ) or nucleophilic (O ) character.
The former are a consequence of the adsorption
of O2 molecules. The latter are oxygen species
coming from the solid lattice. When they participate in the catalytic process, the oxide becomes
reduced and its regeneration is possible by reacting
with O2 .
The study of the catalytic properties for a given
catalyst is facilitated by considering the adsorption, dissociative or not, of certain very simple
molecules, because they can be directly related to
the existence of intermediate complexes. So the
Fischer–Tropsch reaction as well as the olefin hydrogenation both imply the dissociative adsorption of H2 and the activation of the C–O bond or
the olefin p bond. MgO, together with L2 O3 , ZnO
and ZrO2 are important catalysts for olefin hydrogenation [1,3,4]. The oxidation of saturated
hydrocarbons seems to proceed via the heterolytic
activation of r C–H bonds, with the subsequent
loss of the hydrogen atom and the anchorage of
the alkyl group at the cations or anions. In the first
case, the reaction continues by a-elimination to
produce aldehydes or by b-elimination, olefins. In
the second situation, the intermediate is relatively
stable, attaining complete oxidation. Li-doped MgO
and rare-earth oxides at high temperature are very
active catalysts for the oxidative coupling of methane to give ethane from the methyl groups [5].
Other catalyst examples for oxidation of saturated
hydrocarbons are the transition metal oxides such
as MoO3 [6]. The dehydrogenation of alcohols can
proceed by dehydration (Al2 O3 ) [7] or by the activation and rupture of the O–H bond (MgO,
ZnO, CuO, Cr2 O3 ) [8]. In the last case, an open
question is which is the role played by the nucleophilic oxygen alone or the combined acidic–basic
pair for making active the O–H bond.
As it is apparent from the above presentation,
MgO can be viewed as a very good prototype of an
oxide with different and important catalytic applications for modeling purposes. In order to un-
derstand the role played by the surface acidic or
basic sites of MgO on these reactions, several
calculations based on quantum mechanics have
been performed in the past, considering the adsorption of H2 [9], CO [10–13], H2 O [14–16], NO2
[17], H2 S [18,19], CH4 [9,20,21], NH3 [22] and O2
[23] molecules on clusters or slabs. MgO exhibits a
relatively simple crystallographic rock-salt structure and its electronic structure implies only s–p
electrons, making accessible the theoretical
ab initio study of the interaction of molecules with
the surface, contrary to the case of transition metal
oxides.
The adsorption and thermal decomposition of
methanol on MgO has been studied by thermal
desorption spectroscopy (TDS) [24], infrared spectroscopy (IR) [25] and nuclear magnetic resonance
(NMR) [26] techniques, indicating the presence of
at least four different species. Three of them are
directly involved with such acidic–basic pairs. Specifically, the IR study shows that two species,
named as I and II, are weakly adsorbed, disappearing after evacuation [25]. Species I is a physisorbed methanol molecule. Species II is a methanol
molecule linked by a hydrogen bond to the acidic–
basic pair. The last species requires a stronger
evacuation. Moreover, it was suggested that the
different frequencies corresponding to the OH
stretching could be associated with an ensemble of
different atomic configurations for the acidic–basic
pairs. The remaining two species observed at greater
desorption temperatures, named as III and IV, are
dissociated molecules [25]. Species III corresponds
to a methoxy group linked to a magnesium ion and
an adjacent hydrogen atom forming a hydroxyl
group with an exposed oxygen ion of the surface.
Finally, species IV has been assigned to a surface
methyl group linked to an oxygen ion and a hydroxyl group residing on a Mg ion. Only species
III has been observed in a NMR study performed
at room temperatures [26].
No theoretical efforts have been directed previously to study the adsorption of methanol on this
oxide. In the present work, this process is studied
by self-consistent molecular orbital calculations
with the aim to answer the different questions
opened around the dehydrogenation of methanol,
which finalizes by producing formaldehyde.
M.M. Branda et al. / Surface Science 504 (2002) 235–243
237
2. Computational details
Calculations were carried out in a model system
containing a methanol molecule adsorbed on the
(1 0 0) face of MgO represented by a cluster of 12
oxygen atoms and 12 magnesium atoms, i.e., with
the (MgO)12 formula. Two conformations of the
adsorbed methanol on the MgO cluster were
considered, i.e., eclipsed and staggered rotamers.
In Fig. 1 these two conformers and in Fig. 2 the
CH3 OH molecule as staggered rotamer adsorbed
on (MgO)12 are displayed. The adsorbed methanol
geometry was fully optimized using the Hartree–
Fock approach employing three different Gaussian
basis sets, namely A (3-21G), B (6-31G) and C
(locally dense 6-31G). In the locally dense C basis
set, the following polarization functions were added
for atoms involved in the adsorption process: dtype orbitals for O and Mg atoms and p orbitals
for the hydroxyl H atom. The cluster geometry
was taken from Ref. [27] where the Mg–O distance
. It was not allowed to relax in the opis 2.106 A
timization process. Electron delocalization interactions that take place in the methanol molecule
and between the methanol and the MgO cluster
were studied employing the natural bond orbitals
(NBO) [28,29] approach. Within this approach,
localized orbitals with occupation numbers close
to 2 correspond either to core, bonds and/or lone
pairs, localized orbitals with occupation numbers
Fig. 1. Two conformations of the methanol were considered,
i.e., eclipsed and staggered rotamers.
Fig. 2. Cluster of 12 oxygen atoms and 12 magnesium atoms
representing the substrate for the adsorption of methanol.
notably smaller than 1, to antibonds and Rydberg
orbitals. Their respective occupation numbers are
given by the density-matrix element as calculated
in the NBO basis. Since the Fock matrix is not
diagonal in the NBO basis, it is also possible to
evaluate, by means of second order perturbation
theory, the delocalization energy, DEð2Þ , associated
to the charge delocalization from a highly occupied orbital to an almost unoccupied orbital [28,
29]. All calculations reported in this work were
carried out with the Gaussian 98 package of programs [30].
The use of a cluster to represent the interaction
of small molecule with an oxide is justified by the
fact that normally a limited number of atoms of
the substrate participates in the adsorption process. The so-called ‘‘cluster size effects’’ refer to the
long-range interactions, absent in the cluster, but
present in the semi-infinite and periodic system.
The importance of these effects (finite Madelung
potential, dangling bonds of final atoms, for example) have been studied for MgO in the literature
and different ways have been proposed to remedy
the problem. One of them considers clusters whose
size increases until an acceptable degree of convergence is obtained [18,22]. Another possibility is
to use a relatively small cluster as (MgO)5 and to
238
M.M. Branda et al. / Surface Science 504 (2002) 235–243
mimic the influence of the rest of the solid by
embedding this cluster in an array of model potentials [10,21,31]. It has been observed that, for
MgO, the use of a cluster (MgO)n with n 20–25
units and the appropriate geometry is sufficient to
give adsorption energies close to those obtained
for periodic bidimensional systems [19]. The importance of employing high quality basis for small
clusters with n ¼ 4, 8 has been remarked in Ref.
[14]. In a recent work, Snyder and coworkers [12]
have made an account of the cluster size effect
for CO on MgO. On the other hand, when we are
more interested to attain a qualitative description
of our system, small clusters with n 9–15 can be
used, as it was demonstrated for the adsorption of
CO on surface defects of MgO [11].
In order to study the effect of the cluster finite
size on the electronic properties of adsorbed methanol, the following procedure was adopted. Four
different cluster sizes were considered as models
to perform NBO analyses of charge transfer interactions between adsorbate and substrate, their
formulae being (MgO)5 , (MgO)12 , (MgO)18 and
(MgO)30 (see Figs. 2 and 3). All the four clusters
used in the representation of magnesium oxide
surface have been constructed in such way that
always a central MgO pair is present. While the
Fig. 3. (MgO)5 , (MgO)18 and (MgO)30 clusters employed for
analysing the cluster size effect on NBO parameters. The
(MgO)12 cluster is shown in Fig. 2.
(MgO)18 cluster allows us to study the effect of a
third layer on the adsorption behaviour of this pair,
the (MgO)30 cluster includes interactions of this
pair with next nearest neighbours of the exposed
surface. The adsorbed geometry was optimized for
the second model (see Fig. 2) and then it was kept
fixed when performing NBO calculations on the
other three models. All these calculations were
carried out employing the C basis set. Our (MgO)18
and (MgO)30 clusters are similar in size to the
Mg18 O17 and Mg26 O25 clusters of Ref. [18], respectively. In that work, the Mg18 O17 and Mg26 O25
clusters gave a reasonable representation of the
molecule–oxide interaction. Despite that the geometries are different in these two approaches, we
think that, as the (MgO)18 and (MgO)30 clusters
are in the same size range as those of Ref. [18], here
the size effect should not be dominant. Additionally, we notice that the geometry parameters obtained in Ref. [18] for the (MgO)9 cluster are very
similar to those corresponding to the Mg26 O25
cluster (only 1% difference). This gives confidence
to our procedure of optimizing the geometry with
the (MgO)12 cluster.
3. Results and discussion
First we consider the effect of cluster size on the
electronic properties of adsorbed methanol. For
that purpose, the main NBO parameters characterizing the adsorption of methanol on MgO: the
oxygen, hydrogen and methanol net charges, the
charge transfer interactions from oxygen lone
pairs and the (O–Ha ) bond occupation number,
have been tested. The comparison of these NBO
parameters for clusters of different size is shown in
Table 1. Notice that the small methanol net charge
(0:053 to 0:042) shows an oscillatory behaviour, the values for the (MgO)5 and (MgO)30
clusters being somewhat lower in magnitude than
those calculated for (MgO)12 and (MgO)18 . On the
other hand, the Oa and Ha net charges for the
adsorbate seem stabilized near 0.85 and 0.51,
respectively, for the three greatest clusters. The
(O–H) bond occupation number shows a similar
tendency, stabilizing close to 0.044. The lone pair
transfer from oxygen to the (O–Ha ) bond shows
M.M. Branda et al. / Surface Science 504 (2002) 235–243
239
Table 1
Effect of cluster size on selected NBO parameters of adsorbed methanola staggered conformer with C basis set
Q(CH3 OH)b
RnðOs Þ ! ðO–Ha Þ c
RnðOa Þ ! n ðMgÞc
Occ:ðO–Ha Þ d
QðOa Þe
QðHa Þe
(MgO)5
(MgO)12
(MgO)18
(MgO)30
0.0352
24.50
0.06
0.0369
0.8213
0.4872
0.0413
22.99
0.16
0.0446
0.8555
0.5093
0.0420
23.42
0.21
0.0456
0.8635
0.5118
0.0371
20.83
0.20
0.0438
0.8480
0.5110
a
Subscript ‘a’ stands for adsorbate, while subscript ‘s’ stands for substrate.
Total charge transferred from the substrate to the adsorbate (in a.u.).
c
‘n’ stands for a lone pair and ‘R ’ for both oxygen lone pairs.
d
NBO occupation number.
e
Atomic NBO charge.
b
an oscillation, nevertheless the calculated values for
the three greatest clusters are all three close to 23.0.
The transfer to the lone pair n of magnesium
exhibits a rather monotonic variation, the greater
jump corresponding to going from (MgO)5 to
(MgO)12 . As it is evident from all these results, the
(MgO)12 cluster retains the main properties of the
greater clusters, (MgO)18 and (MgO)30 .
Relative energies for the staggered and eclipsed
conformers of isolated as well as adsorbed methanol are shown in Table 2. In both cases, for calculations carried out with the A and B basis sets,
the preferential conformer is the staggered one.
When employing the C basis set, no local minimum was found corresponding to the adsorbed
eclipsed conformer. If the optimization procedure
begins with the adsorbed molecule as the eclipsed
conformer, then the molecule rotates ending as the
staggered conformer.
It is worth noting that using the A and B basis
sets the calculated staggered–eclipsed energy difference is notably reduced upon adsorption. On
intuitive grounds, it can be expected that the
staggered conformer is the preferential one owing
to the intramolecular stabilizing charge transfer
interaction from the (C–H1 ) bond to the (Oa –Ha )
antibond that takes place. It suggests that upon
adsorption there is a significant charge transfer
from the substrate to the adsorbate (Oa –Ha ) antibond, destabilizing the staggered conformation
between the Oa –Ha and C–H1 bonds.
Eclipsed and staggered adsorption energies,
Eads , as calculated with the A, B and C basis sets
are displayed in Table 3, where it is observed
that Eads ðeclipsedÞ > Eads ðstaggeredÞ, although, as
shown in Table 2, the adsorbed staggered conformer is more stable than the eclipsed one. This
suggests that in the eclipsed rotamer the (Oa –Ha )
antibond is a better electron acceptor than that of
the staggered rotamer favouring the charge transfer interaction in the adsorption process.
Geometrical data of optimized structures of
isolated staggered methanol rotamer and their
adsorbed species are displayed in Table 4. The
following features of data shown in that table are
noticeable. A and B basis set calculations yield a
Table 2
Relative energies for the methanol staggered and eclipsed conformers of the isolated and adsorbed species as calculated with
the A, B and C basis sets
Table 3
Adsorption energies, Eads (kcal/mol) for staggered and eclipsed
methanol conformers as calculated with the A, B and C basis
setsa
Basis set
DEiso a
DEads b
Basis set
Staggered
Eclipsed
A
B
C
1.48
1.17
1.25
0.91
0.74
–
A
B
C
30.24
11.08
8.69
30.80
11.52
–
a
b
DEiso ¼ Eiso ðeclipsedÞ Eiso ðstaggeredÞ ðin kcal=molÞ.
DEads ¼ Eads ðeclipsedÞ Eads ðstaggeredÞ ðin kcal=molÞ.
a
Eads ¼ EðMgO clusterÞ þ EðmethanolÞ EðMgO cluster &
methanolÞ:
240
M.M. Branda et al. / Surface Science 504 (2002) 235–243
Table 4
Selected geometrical data from optimized geometries with the A, B and C basis sets, for the methanol staggered rotamer when isolated
and when adsorbed on magnesium oxidea
A
Isolated
dðHa Os Þ
dðOa MgÞ
dðOa –Ha Þ
dðC–Oa Þ
dðC–H1 Þ
dðC–H2;3 Þ
C–Oa –Ha
Oa –C–H1
Oa –C–H2;3
–
–
B
Adsorbed
1.727
2.139
0.966
0.988
1.441
1.432
1.079
1.079
1.085
1.083
110.3
119.5
106.3
106.6
112.2
111.3
a
; angles are
Bond lengths and distances are given in A
C
Isolated
Adsorbed
Isolated
Adsorbed
–
–
1.981
2.271
0.958
1.425
1.077
1.082
118.4
106.6
111.2
–
–
1.877
3.245
0.955
1.417
1.080
1.085
110.4
107.4
111.7
0.950
1.430
1.077
1.083
113.4
106.2
111.6
0.943
1.416
1.080
1.085
109.2
107.0
111.9
given in degrees.
methanol O atom and a cluster Mg atom distance,
Oa Mg, shorter than the sum of their van der
Waals radii, suggesting that there is a strong attraction between them. That Oa Mg distance is
close to the Os Mg bond length taken from the
literature [27] to build up the cluster that simulates
the substrate. This seems to indicate that these
modest basis sets overemphasize the role played by
the electrostatic interaction in the Oa Mg contact, yielding this strong interaction a chemisorption character to the association between methanol
and substrate. However, when the same calculation is carried out with the C basis set, which is of
better quality than the former two, it yields the
longer than the other
Oa Mgs distance 1 A
two, suggesting that such a strong attraction is
only an artifact introduced when using the modest
A and B basis sets.
On the other hand, the calculated Oa –Ha Os
distance with any of these basis sets is typical of a
strong hydrogen bond interaction. The important
strength of this interaction can also be assessed
by noting the lengthening effect on the Oa –Ha
bond length that the adsorption process causes,
which is typical for an important charge transfer interaction [32]. It is interesting to note that
this lengthening effect is stronger for the eclipsed
than for the staggered rotamer, supporting the
assumption made above about the larger electron
acceptor ability of the (Oa –Ha ) antibond which is
placed syn to a methyl C–H bond. In other terms,
the electron acceptor ability of the (Oa –Ha ) anti-
bond is inhibited when it is placed anti to a C–H
bond.
Comparing other isolated methanol geometrical
parameters with the corresponding ones in the
adsorbed species, important changes are observed
for calculations carried out using the A and B basis
set. However, these changes are very small for the
staggered rotamer when the optimization procedure is carried out with the C basis set. This is
further evidence that the nature of the adsorbate–
substrate interaction is not properly described with
the modest A and B basis sets.
In Table 5, several NBO parameters that characterize the adsorbate–substrate charge transfer
interactions are displayed. In this table, only the
data corresponding to the staggered conformer of
methanol are exhibited. These parameters provide
a quantitative description of the charge transfer
interaction between the substrate and the adsorbate commented above. Regardless of the basis
set used to carry out the calculations, an important negative charge is transferred from the substrate to the adsorbed methanol molecule. The
charge transfer associated with the hydrogen
bond between the adsorbate and the substrate,
RnðOs Þ ! ðOa –Ha Þ , depends strongly on the quality of the basis set employed. Besides, the strength
of this interaction parallels the occupation number
of the (Oa –Ha ) antibond. The decrease of the
nðOa Þ ! n ðMgÞ when increasing the quality of
the basis set is noticeable, becoming almost negligible for calculations carried out with the C basis
M.M. Branda et al. / Surface Science 504 (2002) 235–243
241
Table 5
Selected NBO parameters corresponding to adsorbate–substrate interactions (staggered conformer of methanol)a
A
QðCH3 OHÞb
RnðOs Þ ! ðOa –Ha Þ c
RnðOa Þ ! n ðMgÞc
Occ:ðOa –Ha ÞðadsÞ d
Occ:ðOa –Ha ÞðisoÞ d
QðOa ÞðadsÞ e
QðOa ÞðisoÞ e
QðHa ÞðadsÞ e
QðHa ÞðisoÞ e
0.0479
53.05
21.72
0.0907
0.0043
0.8628
0.7287
0.4858
0.4431
B
C
0.0153
12.37
13.31
0.0323
0.0054
0.8775
0.7987
0.5231
0.4798
0.0413
22.99
0.16
0.0446
0.0049
0.8555
0.7917
0.5093
0.4831
a
Charges are given in a.u.; interaction energies are given in kcal/mol. ‘a’ stands for adsorbate, ‘s’ stands for substrate, ‘ads’ means
adsorbed and ‘iso’ means isolated.
b
Total charge transferred from the substrate to the adsorbate.
c
‘n’ stands for a lone pair and ‘R’ for both oxygen lone pairs.
d
NBO occupation number.
e
Atomic NBO charge.
set. These last two trends are in line with the
lengthening of the Oa Mg distance commented
above when discussing geometrical data. It should
be noted that the polarity of the Oa –Ha bond increases notably in the adsorption process, isolated
to adsorbed. The change in the NBO charge for Oa
and Ha (Table 5) is indicative of this increase.
It is worth commenting that M€
ulliken overlap
populations (OPs) involving pairs of atoms belonging to the MgO substrate (OPs between 0.08
and 0.26) are noticeably smaller than those between pairs of bonded atoms in methanol (OPs
between 0.26 and 0.80), which is indicative of the
important ionic character within the bonds of the
substrate [27]. Taking into account that the bonds
involved in the linking of methanol to the MgO
surface are also relatively small (calculated OPs are
0.06 and 0.10), the Oa Mg bond is also of ionic
character and the Ha Os bond corresponds to
an H-bond interaction with no covalent character.
These observations are independent of the basis set
considered. The calculated OPðHa ; Os Þ value with
the C basis set is slightly larger than that calculated
with the B basis set, in agreement with a shorter
Ha Os distance obtained with the addition of
polarization functions.
On the other hand, the OPðOa ; Ha Þ decreases
significantly (by 20%, independent of the basis)
after the methanol molecule is adsorbed. This variation implies a weaker Oa –Ha bond owing to the
fact that the H atom is linked with two oxygen
atoms. This result is in agreement with the stretching of the Oa –Ha bond, as it was outlined before. A
similar observation is valid for the methanol C–Oa
bond, after the molecule is adsorbed, although in
this case the OP decrease is less pronounced
(around 13% for the B basis and 7% for the C
basis).
Recently, several theoretical studies of the adsorptive and reactive properties of MgO have been
performed considering the interaction of a small
molecule and the surface of this oxide. Particularly, the H2 O and H2 S molecules can be considered chemically similar to methanol molecule
because they have the possibility to polarize its OH
or SH groups, when adsorbed on MgO. So a
comparison with the adsorption of these molecules could give us an alternative source of information. The published results concerning the
adsorption of H2 S on MgO(1 0 0), representing the
MgO surface by a cluster [18] as well as by a slab
[19], show that the interaction of this molecule is
due mainly to a S–Mg bond, the S–H bonds being
oriented outwards the MgO surface. On the other
hand, from our calculations we conclude that the
H atom of OH group in methanol molecule is
oriented towards the surface, establishing a Hbond with an O ion of MgO. In both cases, the
adsorbed molecule is linked to the surface through
a relatively weak bond, with an adsorption energy
242
M.M. Branda et al. / Surface Science 504 (2002) 235–243
of 6–9 kcal/mol for SH2 and 9 kcal/mol for
CH3 OH, without relevant net charge transfers
with the surface atoms. The results quoted for
H2 O in Ref. [14], performed with the cluster approach and with the H2 O molecule at internal
constrained geometry, are more relevant due to the
presence of OH groups. In this case, the two H
atoms are directed to respective O ions of MgO
and the O atom resides over a Mg ion of MgO.
The calculated adsorption energy and molecule–
surface distance values result to be 5 kcal/mol
lower than in our calculations,
greater and 0.3 A
respectively, probably due to that the presence of
two H bonds instead of one, as is in the case of
methanol.
Recently, TPD spectroscopy results concerning
the methanol adsorption on MgO(1 0 0)/Mo(1 0 0)
have been published [33]. They show that methanol adsorbed on the more favourable sites of this
system has an adsorption energy value of 17.4
kcal/mol. On the other hand, if correlation effects
based on the perturbative Møller–Plesset method
at second order (MP2) [34] are included in our
approach, an adsorption energy equal to 13.8 kcal/
mol is attained. This result is only 20% lower than
the experimental value. Similar corrections for slab
calculations of H2 O on MgO have been cited in the
literature [35].
As it was mentioned above, the present results
indicate that the methanol molecule can be adsorbed in a non-dissociative way on MgO(1 0 0)
making a hydrogen bond with an O ion of the
oxide, through its hydroxyl group. This kind of
adsorption corresponds with the proposed species
II of adsorbed methanol in Ref. [25]. The IR spectra
for this species reveal that the mOH frequency of O–
H bond shifts from 3800 cm1 , for free methanol,
to a band localized in the range 3110–3332 cm1 .
This result is in agreement with the weakening of
O–H bond associated with the formation of the
hydrogen bond, as it was deduced in the above
discussed orbital population analysis. Because this
adsorption is accompanied by a weakening of O–
H bond, it would be possible that the dissociative
form of methanol, identified in Ref. [25] as species
III, could be achieved after surmounting a certain
activation barrier. Actually, our calculations show
that the separated fragments are not stable when
they reside on neighbouring penta-coordinated
sites, giving as a final product the species II. This
result conducts us to conclude that the (1 0 0)
surface is not able to activate the O–H bond of
methanol. Then, other sites must be considered,
specifically with lower coordination number. In
the past, the activity of MgO catalysts has been
related to the presence of multiple exposed faces
and surface defects, associating the coordination
number of surface ions to the appearance or not of
a specific reaction [1]. This idea is in agreement
with previous observations reported in other theoretical works. So in Ref. [11] it was shown that the
CO molecule adsorbs more strongly on sites of
lower coordination. Similar results have been
found for the adsorption of H2 O [15]. The dissociation of this molecule has been studied on threecoordinated sites of MgO [14]. In the recent works
[18,19] related to the adsorption of H2 S, HS and
S2 on MgO, it was concluded, after looking at the
results for HS and S2 , that the three-coordinated
sites show a greater dissociative effect than the
penta-coordinated ones. We also notice that, in
these works, the adsorption energy of H2 S exhibits
an increment of 3 kcal/mol when the coordination
decreases from 5 to 3.
Recently, experimental results obtained from
electron spectroscopies and TPD have revealed
that methanol adsorbed on a MgO(1 0 0) surface
without defects show no dissociation [33]. Moreover, another work, where the acid–base properties of MgO(1 0 0) films and of a monocrystalline
MgO(1 0 0) surface were studied with surface
spectroscopies, has demonstrated that the methanol dissociation is directly related with the density
of surface defects [36]. A sputtered and subsequently annealed MgO surface loses its dissociative ability. In this way, we have a clear evidence of
the above commented relationship between the
coordination number of the adsorption site and
the activation of molecular bonds.
4. Conclusions
The main conclusions of the ab initio calculations performed in this work can be summarized as
follows:
M.M. Branda et al. / Surface Science 504 (2002) 235–243
1. It is observed that, for both the isolated and adsorbed methanol molecules, the staggered conformation is energetically favourable instead
of the eclipsed one. The difference in energy between the staggered and eclipsed conformations
reduces in the adsorption process.
2. The intermolecular distances and the adsorption energies are indicative of a hydrogen bond
interaction between the methanol OH group
and the surface O2 anion. In addition to this
fact, a Oa Mgs attractive interaction takes
place. The calculated magnitude of this interaction is weaker when polarization functions are
added to the basis set employed. Magnesium
oxide behaves as a basic catalyst when methanol is adsorbed. There are electron drifts between the substrate and the adsorbate in both
directions resulting in a polarized methanol.
3. The lengthening of the Oa –Ha bond in the adsorption process can be related with the greater
(O–H) NBO occupation number, showing a
weakening of the O–H bond. In addition, a polarization of the Oa –Ha is found.
Acknowledgement
Financial support from CONICET, UBACyT,
UNS and ANPCyT (# 604) is gratefully acknowledged.
References
[1] K. Tanabe, M. Misono, Y. Ono, H. Hattori, in: B.
Delmon, J.T. Yates (Eds.), New Solid Acids and Bases.
Their Catalytic Properties, Studies in Surface Science and
Catalysis, vol. 51, Kodansha & Elsevier, Tokyo & Amsterdam, 1989.
[2] J. Haber, in: J.P. Bonelle, B. Delmon, E. Derouane (Eds.),
Surface Properties and Catalysis by Non-Metals, D. Reidel
Publishing Company, Dordrecht, 1982.
[3] R.J. Kokes, A.L. Dent, Adv. Catal. 22 (1972) 1.
[4] H. Hattori, Y. Tanaka, K. Tanabe, J. Am. Chem. Soc. 98
(1976) 4652.
[5] T. Ito, J.-X. Wang, J.H. Lundsford, J. Am. Chem. Soc. 107
(1985) 5062.
[6] M.R. Smith, U.S. Ozkan, J. Catal. 141 (1993) 124.
[7] H. Niiyama, E. Echigoya, Bull. Chem. Soc. Jpn. 44 (1971)
1739.
[8] L. Nodek, J. Sedlacek, J. Catal. 40 (1975) 34.
243
[9] J.L. Anchell, K. Morokuma, J. Chem. Phys. 99 (1993)
6004.
[10] J.A. Mejıas, A.M. Marquez, J. Fernandez Sanz, M.
Fernandez Garcia, J.M. Ricart, C. Sousa, F. Illas, Surf.
Sci. 327 (1995) 59.
[11] R. Soave, G. Pacchioni, Chem. Phys. Lett. 320 (2000) 345.
[12] J.A. Snyder, D.R. Alfonso, J.E. Jaffe, Z. Lin, A.C. Hess,
M. Gutowski, J. Phys. Chem. B 104 (2000) 4717.
[13] C. Minot, M.A. Van Hove, J.-P. Biberian, Surf. Sci. 346
(1996) 283.
[14] J.L. Anchell, A.C. Hess, J. Phys. Chem. 100 (1996) 18317.
[15] A.L. Almeida, J.B.L. Martins, C.A. Taft, E. Longo, W.A.
Lester Jr., Int. J. Quant. Chem. 71 (1999) 153.
[16] B. Ahlswede, T. Homann, K. Jug, Surf. Sci. 445 (2000) 49.
[17] J.A. Rodriguez, T. Jirsak, S. Sambasivan, D. Fischer, A.
Maiti, J. Chem. Phys. 112 (22) (2000) 9929.
[18] J.A. Rodriguez, T. Jirsak, S. Chaturvedi, J. Chem. Phys.
111 (17) (1999) 8077.
[19] J.A. Rodriguez, A. Maiti, J. Phys. Chem. B 104 (2000) 3630.
[20] A.M. Ferrari, S. Huber, H. Kn€
ozinger, K.M. Neyman, N.
R€
osch, J. Phys. Chem. B 102 (1998) 4548.
[21] K. Todnem, K.J. Børve, M. Nygren, Surf. Sci. 421 (1999)
296.
[22] Y. Nakajima, D.J. Doren, J. Chem. Phys. 105 (17) (1996)
7753.
[23] L.N. Kantorovich, M.J. Gillan, Surf. Sci. 374 (1997) 373.
[24] D.C. Foyt, J.M. White, J. Catal. 47 (1977) 260.
[25] A.J. Tench, D. Giles, J.F.J. Kibblewhite, Trans. Faraday
Soc. 67 (1971) 854.
[26] S.H.C. Liang, I.D. Gay, Langmuir 1 (1985) 593.
[27] A.F. Wells, Structural Inorganic Chemistry, fifth ed.,
Clarendon Press, Oxford, 1984, 1004 pp.
[28] A.E. Reed, L.A. Curtiss, F. Weinhold, Chem. Rev. 88
(1988) 899.
[29] E.D. Glendening, A.E. Reed, J.E. Carpenter, F. Weinhold,
NBO version 3.1.
[30] M.J. Frisch, G.W. Trucks, H.B. Schlegel, P.M.W. Gill,
B.G. Johnson, M.A. Robb, J.R. Cheeseman, T. Keith,
G.A. Petersson, J.A. Montgomery, K. Raghavachari,
M.A. Al-Laham, V.G. Zakrzewski, J.V. Ortiz, J.B. Foresman, J. Cioslowski, B.B. Stefanov, A. Nanayakkara, M.
Challacombe, C.Y. Peng, P.Y. Ayala, W. Chen, M.W.
Wong, J.L. Andres, E.S. Replogle, R. Gomperts, R.L.
Martin, D.J. Fox, J.S. Binkley, D.J. Defrees, J. Baker, J.P.
Stewart. M. Head-Gordon, C. Gonzalez, J.A. Pople,
Gaussian 98, Revision D.4, Gaussian, Inc., Pittsburgh,
PA, 1998.
[31] H. Kobayashi, M. Yamaguchi, T. Ito, J. Phys. Chem. 94
(1990) 7206.
[32] J.E. Peralta, M.C. Ruiz de Az
ua, R.H. Contreras, J. Mol.
Struct. (Theochem.) 491 (1999) 23.
[33] J. G€
unster, G. Liu, J. Stultz, S. Krischok, D.W. Goodman,
J. Phys. Chem. B 104 (2000) 5738.
[34] R. Krishnan, J.A. Pople, Int. J. Quant. Chem. 14 (1978) 91.
[35] C.A. Scamehorn, N.M. Harrison, M.I. McCarthy, J.
Chem. Phys. 101 (2) (1994) 1547.
[36] X.D. Peng, M.A. Barteau, Langmuir 7 (1991) 1426.