DOI: 10.1002/chem.

200900487

Theoretical Study on the Role of Surface Basicity and Lewis Acidity on the
Etherification of Glycerol over Alkaline Earth Metal Oxides

Monica Calatayud,*[a] Agnieszka M. Ruppert,[b, c] and Bert M. Weckhuysen*[b]

Abstract: Alkaline earth metal oxides
(MO) are catalytically active in the
etherification of glycerol. Density
Functional Theory (DFT) calculations
have been used to examine the reactiv-
ity of glycerol with MO surfaces with
M = Mg, Ca, Sr or Ba. More specifical-
ly, the optimum glycerol adsorption
mode and the strength of glycerol in-
teraction with regular MO (001) sur-
ACHTUNGREfaces and a stepped CaO surface have
been investigated and involves the in-
teraction with acid–base surface sites.
The basicity of lattice oxygen atoms is
correlated with the adsorption energy:
BaO (3.02 eV) > SrO (2.85 eV) >
CaO (2.05 eV) > MgO (1.35 eV).
The interactions have an exothermic
character, that is, the more basic the al-
kaline earth metal oxide, the more exo-
thermic is the adsorption process and
the higher the dissociation extent.
Thus, the dissociation of glycerol in-
creases in the order: MgO (not dissoci-
ated) < CaO (partially dissociated <
SrO (partially dissociated) < BaO
(completely dissociated). The presence
of defects is found to play a key role in
the mechanism: glycerol interaction
with a stepped CaO surface presents
the highest adsorption energy
(3.78 eV), and the molecule is found
to dissociate at the step. The calculated
structural parameters are found to be
Keywords: alkaline earth metals ·
basicity · density functional calcula-
tions · glycerol · heterogeneous
catalysis · renewable resources
in good agreement with experimental
data on catalyst reactivity. Moreover,
the earlier postulated reaction mecha-
nism, which also involves the addition-
al involvement of Lewis acid sites
proved to be feasible for CaO and SrO
regular surfaces, and for the stepped
CaO surface. It was found that for
these oxides one of the most favored
adsorption modes involves a non-disso-
ciative adsorption of one hydroxyl
group of glycerol, which as a result be-
comes a better leaving group. There-
fore, theoretical evidence was found
for the possible direct involvement of
Lewis acid sites in the catalytic etherifi-
cation of bio-derived alcohols, such as
glycerol, as it is anticipated that these
observations can be extended to sugar
alcohols as well.

Introduction

The use of renewable resources, forced by diminishing and

insecure fossil fuels supplies, is nowadays stimulated by gov-
[a] Dr. M. Calatayud
Laboratoire de Chimie Thorique UMR 7616, CNRS
ernmental bodies all over the world. Biomass is considered
Universit P. M. Curie-Paris 06 as highly important since it currently represents the only re-
4 Pl. Jussieu case 137 75252 Paris (France) newable resource for the production of liquid transportation
Fax: (+ 33) 144-27-41-17 fuels. The European Union has set an overall target for 2010
E-mail: calatayu@lct.jussieu.fr
of 5.75 % share for biofuels.[1] Glycerol is a byproduct of
[b] Dr. A. M. Ruppert, Prof. Dr. B. M. Weckhuysen
bioACHTUNGREdiesel production and for every 1000 kg of fatty acid
Inorganic Chemistry and Catalysis group
Department of Chemistry, Faculty of Science esters around 100 kg of glycerol is formed. Although glycer-
Utrecht University, Sorbonnelaan 16 ol itself has a lot of commercial applications, for example, in
3584 CA Utrecht (The Netherlands) cosmetic and pharmaceutical industries, new attractive ap-
Fax: (+ 31) 30-251-1027 plications should be introduced to improve the economics of
E-mail: b.m.weckhuysen@uu.nl
the biodiesel process.[2] Furthermore, glycerol is regarded as
[c] Dr. A. M. Ruppert
Institute of General and Ecological Chemistry
a potential platform molecule since it can be directly pro-
Technical University of Lodz duced from sugars or sugar alcohols, which are one of the
90-924 Łdź, ul. Żeromskiego 116 (Poland) cornerstones of future biorefinery schemes.

10864  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 10864 – 10870

The catalytic valorization of glycerol has been the subject
of several review articles.[3–5] Direct use of crude glycerol as
a cheap, attractive starting material has also been investigat-
ed.[6, 7] One possible route of glycerol valorization is the syn-
thesis of di-, tri- and/or polyglycerol. Di- and triglycerol and
their derivatives can be used as polymer additives, lubri-
cants, plasticizers, emulsifiers, stabilizers and dispersants.
Glycerol etherification is catalyzed by both acids[8] and
bases.[9] Considerable attention in the literature has been
given to both mesoporous materials and base-modified zeo-
lites investigated as catalysts for this reaction. Worth men-
tioning here is the early work of Clacens et al.[10–12] with the
broad scope of examined materials, and also a more recent
paper about base-type zeolites doped with Cs.[13] As recently
shown, the acid–base properties of metal oxides largely de-
termine their catalytic activity. For example, our group has
shown that alkaline earth metal oxides (MgO, CaO, SrO
and BaO) are promising solid catalysts and glycerol conver-
sion increases with increasing catalyst basicity. Furthermore,
Lewis acidity seems also to play an important role since dif-
ferently prepared CaO materials show different glycerol
etherification activity, which even may surpass that of
BaO.[9]
Theoretical calculations on the nature of acid–base prop-
erties of alkaline earth metal oxides have already been car-
ried out. Adsorption on oxide surfaces, including MgO, has
been investigated by Calatayud et al.[14] Pacchioni et al.[15]
have examined the basic character of alkaline earth oxides
in great detail. In this work, the different reactivities of
MgO and CaO have been explained by their electrostatic
nature. In those cases the oxygen ion is stabilized on the sur-
face of oxides by the Madelung potential of the ionic crystal
(the electrostatic potential generated by the charge distribu-
tion of the crystal). Based on ab initio cluster calculations
they found that MgO and CaO surfaces possess different
electrostatic (Madelung) potentials. This potential is smaller
for CaO than for MgO, which results in a more diffuse
charge distribution around the oxygen atoms in CaO and in
a lower cost for the charge transfer from the donor orbital
of the surface anion. In other words, the cation–anion dis-
tances in CaO in comparison to MgO are larger and the Ma-
delung potential is smaller. All this leads to a higher basicity
and reactivity for CaO.
More recently, the Madelung potential of alkaline earth
metal oxides was studied in relation to the adsorption of CO
on defect-free surfaces of MgO, CaO and SrO. Halim[16]
found that the strength of adsorption decreases in the fol-
lowing order: SrO > CaO > MgO. This corresponds to cat-
ionic site distances dACHTUNGRE(O–M), which are the shortest for MgO
and the longest for SrO. This is in line with the cation size,
which changes in the same order. It is also known that the
metal oxides with larger metal cations are more basic and
thus more active towards acidic species. This ability seems
to be determined by the differences of crystal potentials as-
sociated with the large differences in lattice parameters and
by the difference in the size of the cation. Density functional
calculations have also been used to study the interaction of
Chem. Eur. J. 2009, 15, 10864 – 10870  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
FULL PAPER
MgO, CaO and SrO with water and methanol.[17, 18] In the
case of MgO, it was found that water maintains its molecular
character and dissociation was not observed at room tem-
perature on the (001) surface. This is due to the large barri-
ers involved in the separation of proton-hydroxyl struc-
ture.[19] At the same time it was found that defects can in-
crease the activity and dissociation of water can occur on
those places.[20]
To the best of our knowledge, the interaction of glycerol
with the surfaces, including alkaline earth metal oxides, has
never been investigated from a theoretical point of view.
Only the structure of glycerol solvated in water has been op-
timized with DFT calculations by Sethu Raman et al.[21] Our
work aims to provide a deeper understanding of the interac-
tion of glycerol, as a representative example of a biomass-
derived polyalcohol, with the surfaces of alkaline earth
metal oxides. For this purpose, we have investigated: a) The
reactivity of glycerol with MgO, CaO, SrO and BaO; b) the
optimum glycerol adsorption mode and strength of interac-
tion with the surface; and c) the influence of surface steps
on the interaction of glycerol. These findings will be com-
pared with experimental data on the glycerol etherification
reaction.[9]
Results and Discussion
Surface models of alkaline earth metal oxides and their in-
teraction with glycerol: We intended to remain as close as
possible to the experimental conditions used for the etherifi-
cation of glycerol where freshly prepared oxides have been
used.[9] As a consequence, only the lattice oxygen and cation
sites have been taken into account in the surface models of
the different alkaline earth metal oxides. Firstly, we have
performed the modeling of the interaction of glycerol mole-
cule with the optimized, rocksalt-type surfaces of the alka-
line earth metal oxides. The reactivity towards glycerol for
MO with M = Mg, Ca, Sr or Ba has been compared. The op-
timum glycerol adsorption mode and the strength of glycerol
interaction with the surface were investigated.
The most stable surface for the MO is (001). The MO
p p
(001) surface was modeled by a 2  2 unit cell five layers
thick, as illustrated in Figure 1. The uppermost three layers
were allowed to relax together with the adsorbate, the two
bottom layers were kept fixed to optimized bulk positions.
In this unit cell four M and four O atoms are exposed on
the surface. One glycerol molecule is considered (coverage
q = 1/4 molecules/surface M), which is a reasonable choice
considering the high coverage expected in real conditions
where the catalyst is put in a pure glycerol reaction medium.
The carbon atoms of glycerol are labeled as 1, 2 and 3 in
Figure 1. Carbons 1 and 3 are terminal, while carbon 2 is the
middle one. A vacuum of at least 10  is ensured in all the
adsorption systems to prevent interaction between succes-
sive layers. Table 1 displays the details of the models used.
The interaction is calculated as the difference between the
final adsorption system total energy and the gas-phase mole-
www.chemeurj.org 10865
 B. M. Weckhuysen, M. Calatayud and A. M. Ruppert

Table 1. Selected geometrical parameters used in the DFT calculations

assessing the interaction between glycerol and different alkaline earth
metal oxides.
MgO CaO SrO BaO
dshort[a] [] 3.00 3.41 3.67 3.96
dlong[a] [] 4.25 4.82 5.20 5.60
p p
2  2 area [2] 36.15 46.49 54.00 62.69
q [molecules nm2] 2.77 2.15 1.85 1.60
[a] M–M distance, see Figure 1.

Figure 2. The best adsorption modes for glycerol on MgO (001). In pa-
rentheses is the adsorption energy; distances in ; carbon atoms are indi-
cated by numbers 1, 2 and 3.

Figure 1. Top view (left) and side view (right) of the slab used in the cal-
culations, together with the glycerol molecule (top right). The unit cells
p p
1  1 and 2  2 as well as the shortest M–M distances are displayed.
M: metal sites, O: oxygen sites.

cule plus bare slab energies; the more negative the values,
the more favorable the interaction between glycerol and the
alkaline earth metal oxide.
The adsorption geometries of the glycerol molecule in dif-
ferent orientations on the MO surfaces have been calculat-
ed: On top of the metal sites (top), bridging two M sites
(bri), completely protonated (3 H) or partially deprotonated
(2 H, 1 H and 0 H). We have considered that different hy- Figure 3. The best adsorption modes for glycerol on CaO (001). In paren-
droxyl groups of the glycerol molecule might bind to the theses is the adsorption energy; distances in ; carbon atoms are indicat-
surface: one (coming from carbon 1 or 2), two (coming from ed by numbers 1, 2 and 3.
carbon atoms 1, 2 or 1, 3), or three (coming from carbon
atoms 1, 2 and 3), perpendicular or flat to the surface. Table 2. Adsorption energy Eads obtained for all the metal oxide surfaces
More than 50 geometries have been tested and in what studied, negative values indicate exothermicity. The protonation of the
follows we present a summary of the results obtained. glycerol molecule in the most favorable mode is reported, together with
the average M–Ogly bond length dACHTUNGRE(M–Ogly).
The preferred adsorption modes of glycerol onto MgO,
CaO, SrO and BaO are shown in Figures 2–5. In all cases, it MgO CaO SrO BaO CaO step

was observed that the glycerol–surface interaction has elec- Figure 2 3 4 5 8

Eads [eV] 1.36 2.05 2.85 3.52 3.78
trostatic nature, involving acid–base surface pairs. Glycerol
1.31 2.02 2.75 3.00 3.69
oxygen atoms interact with surface M2 + cations and the hy- glycerol protonation full partial partial none partial
droxyl groups interact via hydrogen bonding with surface dACHTUNGRE(M–Ogly) [] 2.0–2.3 2.4–2.6 2.5–2.7 2.6–2.9 2.3–2.5
basic O2 sites. Glycerol exhibits dissociation of its hydroxyl
groups depending on the surface, with formation of Ogly/
M2 + surf and H + gly/O2surf bonds. Although some features is a top parallel molecule where the middle hydroxyl
were similar, in particular cases different interaction modes group is dissociated (top-2-2 H, 1.31 eV). Two MgOgly
have been observed for each alkaline earth metal oxide. bonds are formed due to electrostatic interactions be-
Table 2 summarizes the adsorption data obtained for all the tween the surface and the adsorbate, while the third hy-
surfaces studied. droxyl group is stabilized by hydrogen bonding to lattice
oxygen. The glycerol molecule fits the pattern of the sur-
a) MgO (Figure 2): The most favorable adsorption energies face Mg sites. The acid–base character of Mg and O is
are in the range of 1.30 eV. They correspond to glycer- not strong enough to dissociate glycerol to a large
ol located on top of Mg sites, parallel to the surface extent. This is observed during the optimization process.
(top-123-3 H, 1.36 eV). Isoenergetic with this structure Starting from dissociated glycerol, it recombines with the

10866 www.chemeurj.org  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 10864 – 10870
Alkaline Earth Metal Oxides
FULL PAPER
protons and leads to the non-deprotonated molecule. A
similar behavior has been reported in the literature for
the interaction between water or methanol and
MgO.[17, 19]
b) CaO (Figure 3): The most exothermic adsorption ener-
gies are about 2 eV. The glycerol molecule adsorbs flat
on the surface, completely deprotonated, its oxygen
atoms bridging two Ca surface atoms each (bri-123-0 H,
2.05 eV). Partial deprotonation of two hydroxyl groups
is isoenergetic (bri-12-1 H, 2.02 eV). There is no prefer-
ence for 1,2 or 1,3 deprotonation. As a result of the dis-
sociative adsorption, surface hydroxyl groups appear in
interaction with the glycerol molecule forming hydrogen
bonds.
Figure 5. The best adsorption modes for glycerol on BaO (001). In paren-
theses is the adsorption energy; distances in ; carbon atoms are indicat-
ed by numbers 1, 2 and 3. The rumpling of some Ba sites is indicated
with thick arrows.

Glycerol interaction with a stepped CaO surface: In the

Figure 4. The best adsorption modes for glycerol on SrO (001). In paren-
theses is the adsorption energy; distances in ; carbon atoms are indicat-
ed by numbers 1, 2 and 3. The rumpling of some Sr sites is indicated with
thick arrows.
c) SrO (Figure 4): The most favorable adsorption energy is
2.85 eV corresponding to a deprotonated glycerol mol-
ecule (bri-123-0 H). The glycerol oxygen sites adsorb on
a strontium site bridging neighboring sites. The struc-
tures starting from top positions evolve to the bridging
sites. Partial protonation of glycerol is slightly (bri-123-
1 H, 2.75 eV) or significantly less favorable (bri-12-1 H,
2.06 eV, not shown). The rumpling of some Sr sites is
also observed.
d) BaO (Figure 5): The most favorable adsorption energy is
3.52 eV (bri-123-0 H). Glycerol is completely deproton-
ated, which corresponds to a highly basic material. The
three glycerol oxygen atoms bind to the same Ba site as
well as to the neighboring cations. This is due to the
large Ba–Ba distance as compared to the previous struc-
tures (Table 1). The structures with partial deprotonation
(bri-12-1H 3.01 eV, bri-13-1H 2.70 eV) are less fa-
vored although their adsorption energies are higher than
for the previous materials. The structure with deproton-
ated hydroxyls in position 1,2 is slightly preferred to the
corresponding 1,3 by 0.31 eV. The rumpling of some Ba
sites is also observed.
second part of our study we concentrated on the behavior of
CaO. From our experimental results[9] it is known that CaO
possesses both basicity and Lewis acidity, leading to very
high glycerol etherification activity. It was shown as well
that CaO prepared in the right manner might be as active as
more basic materials, such as BaO. It is also known that de-
fects on the surface can change the activity due to a higher
number of coordinatively unsaturated atoms.[22] Further-
more, interactions of MgO with water or methanol have al-
ready been calculated[23, 24] and it was proven that local sur-
face defects, like corners, kinks and steps, can significantly
influence the adsorption of water.[23]
Therefore, we have modeled a stepped surface of CaO
represented by the Miller indices (501), as shown in
Figure 6. The slab has dimensions of 12.29  9.64  11.32 
Figure 6. Top view (left) and side view (right) of the slab used as model
for the stepped CaO surface.
and contains 100 atoms. The uppermost layers are allowed
to relax and the bottom layers are kept fixed to their bulk
positions. A total of 28 structures have been calculated. The
interaction of glycerol with steps in different geometries as
well as on the terrace region has been tested. Glycerol ad-
sorption by a different number of hydroxyl groups with sur-
face metal sites have been explored: two OH groups (12 or
13), or three OH groups (123). Full and partial deprotona-

Chem. Eur. J. 2009, 15, 10864 – 10870  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 10867
 B. M. Weckhuysen, M. Calatayud and A. M. Ruppert

tion has been considered (0 H, 1 H respectively). A selection

of the initial interaction models is shown in Figure 7.
Table 3 summarizes the calculated adsorption energies for
glycerol for the most representative structures, which are
shown in Figure 8.

Figure 7. Selected initial geometries for glycerol interacting with the step-
ped CaO (501) surface.
Figure 8. Selected optimized structures for glycerol interacting with the
stepped CaO (501) surface. The most favorable geometries involve inter-
Table 3. Adsorption energy for the most stable systems on a stepped action with the step.
CaO (501) slab, calculated as the difference between the adsorption
system and the sum of bare surface plus gas-phase glycerol total energy;
negative values indicate exothermicity. Labels indicate the model (see mode. This supports the presence of defects as the basis of
Figure 7) and the protonation of the glycerol fragment (1H: partially pro-
higher reactivity found in irregular materials.
tonated, 0H: deprotonated).
The adsorption takes place by the interaction of glycerol
Structure Eads [eV]
oxygen sites with surface Ca2 + sites forming strong CaOgly
123 C 1H 3.78
bonds. Indeed, the interaction is of electrostatic character
0H 3.69
123 B 0H 3.66 and proceeds as an acid–base reaction. Lewis acid lattice
1H 3.19 Ca2 + sites interact with basic glycerol Ogly sites. In parallel,
12 D 1H 2.88 the glycerol hydroxyl groups deprotonate through another
0H 2.55 acid-base reaction. The acidic proton binds to the lattice
13A 1H 2.84
oxygen Olatt site behaving as a basic site. Thus, the acid–base
12 B 1H 2.61
123 D 0H 2.16 pairs present on the slab are necessary to stabilize glycerol
123 D 1H 2.11 on the metal oxide surface.
Summarizing, the interaction of glycerol with regular or
defective CaO surfaces involves its partial or complete de-
protonation. It was found that at least two hydroxyl groups
Analysis of the results shows that the most favorable dissociate on CaO-forming lattice hydroxyl groups, whereas
structures involve interaction with the step. Adsorption en- in some cases full deprotonation is observed. There is no
ergies can be divided into three groups. The most exother- preference for terminal or bridging glycerol deprotonation.
mic modes, above 3.60 eV, correspond to glycerol adsorb- Such deprotonation is related to the basicity of the material,
ing as an extension of the step. The oxygen sites are parallel the lattice oxygen sites being the basic sites. The oxygen
to the surface facing the step atoms (123-C and 123-B), sites lying in the step are the most basic and the formation
while the high exothermicity corresponds to the formation of step hydroxyl groups is expected to be preferred. For the
of a maximum number of Ogly–Mlatt bonds. The second interaction of glycerol with the stepped CaO surface it can
group presents adsorption energies in the range 2.88 to be concluded that:
2.55 eV. Geometries are associated to the interaction of
glycerol on the border of the step (12-D) or tilted to the ter- a) Glycerol interacts directly with the step. Adsorption en-
race (13-A or 12-B). A third group considers the adsorption ergies are exothermic and are between 3.78 (step) and
of glycerol on the terrace region (123-D). This geometry is 2.16 eV (terrace).
only 2.16 eV exothermic. These results clearly show that b) Glycerol deprotonates to a high extent. Formation of
the interaction of glycerol with the step is extremely favora- surface hydroxyl groups takes place; and
ble as compared to the terrace sites; that is, more than c) Glycerol adsorption takes place on surface acid/base
1.5 eV in the calculated adsorption energy for the best pairs forming Ogly/Ca2 + and H + /Olatt bonds.

10868 www.chemeurj.org  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 10864 – 10870
Alkaline Earth Metal Oxides
These results support the highest reactivity of stepped CaO
surfaces.
Role of Lewis acidity in glycerol etherification: Although
the theoretical work described so far concerns the adsorp-
tion of a single glycerol molecule on the surface of alkaline
earth metal oxide surfaces, we will try to refer to some ob-
servations where two glycerol molecules are activated by
the metal oxide. In our previous study[9] the differences in
activity of differently prepared CaO materials could not be
explained solely by their basic strength and we postulated
the role of Lewis acid sites. More specifically, it was argued
that the leaving of a hydroxyl group (normally a very bad
leaving group) could be facilitated by a Lewis acid site at
the CaO surface.
In some of the favored adsorption modes on regular CaO
(bri-12-1H and bri-13-1H) one of the OH groups is not dis-
sociated even though its oxygen atom is adsorbed on a Ca
ion. It opens the possibility that in the conditions of glycerol
etherification this Ca ion, as a Lewis acid site, indeed acti-
vates the hydroxyl group as a leaving group. The adsorption
energy for this mode is only 0.03 eV lower than that of the
most favorable one. Therefore, its contribution must be sig-
nificant enough to considerably shift the reaction equilibri-
um towards the products. It would explain our observation
that the CaO material with the highest Lewis acidity was
the most active catalyst of glycerol etherification.
Similarly to CaO, in one adsorption mode of SrO (bri-
123-1H) one of the glycerol hydroxyl groups is adsorbed via
an oxygen atom, but not deprotonated (Figure 5). However,
this situation seems to have similar potential of enhancing
the glycerol etherification reaction, as the adsorption energy
difference between this mode and the most favorable one is
slightly higher (0.1 eV) than for CaO (0.03 eV). Finally, the
most favorable geometry for glycerol on the stepped CaO
surface (123-C-1H) also presents a protonated OH group
pointing to the same mechanism.
The Lewis acid site-involving glycerol etherification mech-
anism[9] may still take place over SrO surfaces, however, it is
probably less likely to occur than in the case of some CaO
materials. Finally, for BaO surfaces where total deprotona-
tion of glycerol is strongly favored, Lewis acidity seems to
play no significant role. In other words, its strongest basicity
solely accounts for the high glycerol etherification activity
and the reaction proceeds in spite of the bad leaving group
(HO).
Conclusions
Glycerol adsorption on MO (001) surfaces of alkaline earth
metal oxides involves the interaction with acid–base surface
sites. The basicity of lattice oxygen is correlated to the ad-
sorption energy and follows the order BaO (3.52 eV) >
SrO (2.85 eV) > CaO (2.05 eV) > MgO (1.36 eV).
The more basic the character of the oxide, the more exo-
thermic is the energy and the higher the dissociation extent
Chem. Eur. J. 2009, 15, 10864 – 10870  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
FULL PAPER
of glycerol. Thus, glycerol dissociation increases in the
series: MgO (not dissociated) < CaO (partially dissociated)
< SrO (partially dissociated) < BaO (completely dissociat-
ed). The dissociation of glycerol forms surface hydroxyl
groups that interact by hydrogen bonds with glycerol. The
geometry of adsorption is determined by the topology of the
surface, that is, the lattice parameters and the basicity, to-
gether with the interaction with the surface hydroxyl groups.
The M–Ogly distances correlate with the crystal lattice pa-
rameter. Regarding the adsorption modes, MgO shows pref-
erence for top adsorption sites, while CaO, SrO and BaO
stabilize bridging modes between two surface metal sites,
due to the larger M–Olatt distances. The presence of steps in
CaO enhances the interaction with glycerol (3.78 eV) so
the final system is more reactive than BaO. The molecule
adsorbs at the step sites and is partially deprotonated.
Overall, the theoretical results obtained correspond nicely
with our experimental data on glycerol etherification[9] that
the strength of basic sites largely determines the glycerol
etherification activity. It confirms that deprotonation of a
glycerol hydroxyl group is a key factor in the reaction. The
empirical glycerol conversion and the calculated tendency of
glycerol deprotonation follow the same pattern; that is,
MgO < CaO < SrO < BaO. MgO, which we proved to be
not basic enough to effectively deprotonate glycerol, is
indeed the least active. During glycerol deprotonation hy-
droxyl groups are formed on the surface and the highest
number of those groups is observed on BaO. This could be
well connected with the partial hydration of the metal oxide,
which was observed during the reaction - the highest ten-
dency being observed for BaO.[9] Finally, the theoretical cal-
culations reveal that for glycerol adsorption on CaO and
SrO surfaces some of the favored modes display one glycer-
ol hydroxyl group being adsorbed in a non-dissociative way
on the Lewis acid metal cation. This finding supports our
earlier postulated mechanism of glycerol etherification
where an OH group is activated for leaving by a Lewis acid
site during the nucleophilic attack of another glycerol mole-
cule.
Computational Details
The theoretical calculations have been performed with the periodic ab
initio VASP code.ACHTUNGRE[25–27] The PW91 functional was used together with a
plane wave basis set (cutoff 400 eV). The core electrons have been re-
placed by PAW generated pseudopotentials,ACHTUNGRE[28, 29] according to Table 4.
Table 4. Details of the theoretical calculations used.
Valence electrons
Mg 2p6 3s2
Ca 3p6 4s2
Sr 4s2 4p6 5s2
Ba 5s2 5p6 6s2
O 2s2 2p4
C 2s2 2p2
H 1s1
www.chemeurj.org 10869
 B. M. Weckhuysen, M. Calatayud and A. M. Ruppert

The distance between k points in the reciprocal space was ~ 0.02 1. Nieddu, R. J. M. Klein Gebbink, B. M. Weckhuysen, ChemSusChem,
This methodology was tested for selected bulk structures leading to ex- 2008, 1, 193.
cellent agreement with previous experimental and theoretical results. [8] J. M. Clacens, Y. Pouilloux, J. Barrault, Appl. Catal. A 2002, 227,
This is illustrated in Table 5 for the rocksalt MO structures with M = Mg, 181.
Ca, Sr and Ba. The unit cell was fully optimized. [9] A. M. Ruppert, J. D. Meeldijk, B. W. M. Kuipers, B. H. Ern, B. M.
Weckhuysen, Chem. Eur. J. 2008, 14, 2016.
[10] J. M. Clacens, Y. Pouilloux, J. Barrault, Stud. Surf. Sci. Catal. 2000,
Table 5. Experimental and theoretical details for modeled surfaces. 143, 687.
Oxide Space Cell parameters in  Crystallographic [11] J. M. Clacens, Y. Pouilloux, J. Barrault, C. Linares, M. Goldwasser,
surface group calculated experimental structure Stud. Surf. Sci. Catal. 1998, 18, 895.
[12] J. Barrault, J. M. Clacens, Y. Pouilloux, Top. Catal. 2004, 27, 137.
MgO 225 a = 4.25 a = 4.21 rocksalt
[13] Y. K. Krisnandi, R. Eckelt, M. Schneider, A. Martin, M. Richter,
CaO 225 a = 4.82 a = 4.80 rocksalt
ChemSusChem 2008, 1, 835.
SrO 225 a = 5.20 a = 5.20 rocksalt
[14] M. Calatayud, A. Markovits, C. Minot, Catal. Today 2004, 89, 269 –
BaO 225 a = 5.60 a = 5.52 rocksalt
278.
[15] G. Pacchioni, J. J. M. Ricart, F. Illas, J. Am. Chem. Soc. 1994, 116,
10152.
Acknowledgements [16] W. S. A. Halim, Appl. Surf. Sci. 2007, 253, 8974.
[17] I. D. Gay, N. M. Harrison, Surf. Sci. 2005, 591, 13.
The authors thank ACTS-ASPECT and COST WD36 for financial sup- [18] W. S. A. Halim, A. S. Shalabi, Appl. Surf. Sci. 2004, 221, 53.
port. [19] J. Carrasco, F. Illas, N. Lopez, Phys. Rev. Lett. 2008, 100, 016101.
[20] W. Langel, M. Parrinello, J. Chem. Phys. 1995, 103, 2340.
[21] M. Sethu Raman, V. Ponnuswamy, P. Kolandaivel, K. Perumal, J.
Mol. Liq. 2008, 142, 10.
[1] J.-P. Lange in Catalysis for Renewables: From Feedstock to Energy [22] R. Licona, J. F. Rivas-Silva, Int. J. Quantum Chem. 2005, 104, 919.
Production (Eds.: G. Centi, R. van Santen), Wiley-VCH, Weinheim, [23] C. Chizallet, G. Costentin, M. Che, F. Delbecq, P. Sautet, J. Phys.
2007, p. 21. Chem. B 2006, 110, 15878.
[2] T. S. Viinikainen, R. S. Karinen, A. O. I. Krause in Catalysis for Re- [24] D. Costa, C. Chizallet, B. Ealet, J. Goniakowski, F. Finocchi, J.
newables: From Feedstock to Energy Production (Eds.: G. Centi, R. Chem. Phys. 2006, 125, 054702.
van Santen), Wiley-VCH, Weinheim, 2007, p. 209. [25] G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558.
[3] Y. Zheng, X. Chen, Y. Shen, Chem. Rev. 2008, 108, 5253. [26] G. Kresse, J. Hafner, Phys. Rev. B 1994, 49, 14251.
[4] M. Pagliaro, R. Ciriminna, H. Kimura, M. Rossi, C. Della Pina, [27] G. Kresse, J. Furthmller, Phys. Rev. B 1996, 54, 11169.
Angew. Chem. 2007, 46, 4434; Angew. Chem. Int. Ed. 2007, 119, [28] G. Kresse, J. Hafner, J. Phys. Condens. Matter 1994, 6 8245.
4516. [29] G. Kresse, D. Joubert, Phys. Rev. B 1999, 59, 1758.
[5] F. Jrme, Y. Pouilloux, J. Barrault, ChemSusChem 2008, 1, 586.
[6] Y. Gu, J. Barrault, F. Jrme, Adv. Synth. Catal. 2008, 350, 2007. Received: February 22, 2009
[7] a) R. Palkovits, I. Nieddu, C. A. Kruithof, R. J. M. Klein Gebbink, Revised: April 24, 2009
B. M. Weckhuysen, Chem. Eur. J. 2008, 14, 8995; b) R. Palkovits, I. Published online: September 16, 2009

10870 www.chemeurj.org  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 10864 – 10870