Chemistry of Lewis Acid−Base Pairs on Oxide Surfaces

advertisement
Feature Article
pubs.acs.org/JPCC
Chemistry of Lewis Acid−Base Pairs on Oxide Surfaces
Horia Metiu,* Steeve Chrétien, Zhenpeng Hu,† Bo Li,‡ and XiaoYing Sun
Department of Chemistry and Biochemistry, University of California, Santa Barbara, Santa Barbara, California 93106-9510,
United States
ABSTRACT: We examine a large number of DFT calculations regarding the chemistry of oxide surfaces and show that
their qualitative conclusions can be predicted by using a few
rules derived from the Lewis acid−base properties of the
species involved. (1) The presence of a Lewis acid on an oxide
surface increases substantially the binding energy of a Lewis
base. (2) If an oxide has certain properties because it is a Lewis
base, these properties can be suppressed by adsorbing a Lewis
acid on the surface. (3) The presence of a Lewis base on an
oxide surface diminishes the binding energy of another base, as compared to the binding energy on the same surface with no base
on it. These rules also hold if the words “acid” and “base” are exchanged. We show that these rules apply to a large number of
systems which seem to have no relationship to each other and which are important for catalysis by oxides.
1. INTRODUCTION
Oxides are useful catalysts, and the chemisorption of molecules
on oxide surfaces has been the subject of numerous density
functional calculations. In this article, we show that the results
of many such calculations can be derived from a few rules
involving Lewis acid−base pairs. These rules appear to be general
and to have predictive power. Moreover, they apply to chemical
systems that seem to have no connection to each other.
Since there are numerous definitions of a Lewis acid or base,
we specify the definition used here. A molecule whose electron
charge increases during a reaction is a Lewis acid; the one that
loses electrons is a Lewis base. Whether a molecule is an acid or
a base depends on its reaction partner; a molecule can act as an
acid with one partner and as a base with another partner.
Nevertheless, some molecules have such a strong propensity to
be an acid that it is safe to assume that they are acid in their
interaction with most partners; the same goes for some
molecules that have a strong tendency to be bases.
When it was proposed,1 this definition was hard to use since
it was not possible to determine the electronic charge
redistribution during a reaction. Quantum chemistry now
provides a variety of methods for determining such charge
rearrangements. Among them we favor the one proposed by
Bader,2−4 and we use it throughout this article. In a reaction,
the Lewis acid gains Bader charge, and the Lewis base loses it.
In a few instances, we say that A is a Lewis base and B is a
Lewis acid if the energy of the highest occupied molecular
orbital (HOMO) of A is larger than that of the lowest unoccupied
molecular orbital (LUMO) of B. The basicity of a molecule seems
to be particularly strong when it has a singly occupied molecular
orbital (SOMO); such orbitals tend to have higher energy because
they do not benefit from the effect of “electron pairing”.
In the rest of this article, we show that the results of a large
number of calculations, on a great variety of systems, can be
predicted by the following propensity rules.
© 2012 American Chemical Society
1. (a) The coadsorption of a Lewis acid A with a Lewis base
B, on an oxide surface S that is neither a Lewis acid nor a
Lewis base, results in a very large “attractive” interaction
energy between A and B. By this we mean that the
energy of coadsorbing A and B is much larger than the
adsorption energy of A alone plus the adsorption energy
of B alone. Another way to phrase this is to say that the
presence of a Lewis acid on a surface enhances substantially the binding energy of a Lewis base (compared
to the binding energy when the acid is absent). (b) This
rule also applies to the adsorption of a molecule AB
which dissociates to form adsorbed A and B. The energy
of the dissociative adsorption is largest when the binding
sites of A and B are such that A is a Lewis base and B is a
Lewis acid. (c) If A is a Lewis base and the adsorbed
molecule B has several isomers, then the presence of A
will induce B to form (when it binds to the surface) the
isomer that is a Lewis acid.
All these statements remain true if one interchanges
the words acid and base.
2. If a surface has certain chemical properties because it is a
Lewis acid, these properties are modified substantially by
adsorbing a Lewis base on the surface, as if the base
neutralizes the acid. The same thing happens if the
surface is a base and the adsorbate is an acid.
3. The interaction energy between a Lewis acid and a Lewis
base, coadsorbed on an oxide surface that is neither an
acid nor a base, depends on the strength of the base and
of the acid.
4. If an oxide surface is modified to be a Lewis acid, then a
molecule that is a Lewis acid adsorbs more weakly on the
Received: February 9, 2012
Revised: March 16, 2012
Published: March 20, 2012
10439
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
The Journal of Physical Chemistry C
Feature Article
when the functional is changed, large energies will remain large,
and the small ones will stay small.
modified oxide than on the unmodified one. A similar
rule is true for bases.
These rules are different manifestations of a strong
interaction between a Lewis acid−base pair adsorbed on an
oxide surface. Some chemists may not be surprised by this.
However, we show here that in the case of oxide surfaces this
interaction: (a) is surprisingly strong (of the order of 1 eV or
more); (b) takes place through the oxide, as opposed to a bond
formation between the acid and the base; (c) is universal (we
found no exception); and (d) explains a large number of
computational results for systems that seem to have no
relationship with each other.
These qualitative rules generalize results of many calculations
of the chemical properties of adsorbates on oxide surfaces, by
correlating them with the Lewis acid−base properties of the
participants. When we say that they are general we mean that
calculations show that they work for many different systems,
and we see no reason why they should not work for systems
that have not been studied yet by DFT. It is however possible
that as more examples are examined the rules will have to be
amended, qualified, or extended.
Many of the examples of Lewis acid−base pair interactions
given here interpret results of calculations already published in
the literature or of new calculations performed in our group.
The presentation in this article reverses the path used in the
discovery of the rules: we postulate the rules and use them to
“predict” the results of the calculations. We use this pedagogical
device to emphasize our belief that the rules have predictive
power.
3. PRESENCE OF A LEWIS BASE ON A SURFACE
INCREASES SUBSTANTIALLY THE BINDING
ENERGY OF A LEWIS ACID
We start by examining the results of calculations in which a
Lewis acid is adsorbed on the surface of TiO2(110) on which a
Lewis base has been preadsorbed. The bases examined so far
are a H atom adsorbed to make a hydroxyl on the bridging
oxygen row,14,15 an adsorbed alkali atom (Li, or Na, or K), or
the Au clusters Au3, Au5, and Au7.15 Calculations of Bader
charges have shown that all these compounds donate electrons
when they adsorb on rutile.16−19 Therefore, they are Lewis
bases with respect to the rutile surface. Our rules predict that
the presence of any one of these bases will increase the binding
energy of a Lewis acid (as compared to the case when no base
is present on the surface). The Lewis acids we examine here are
a Au atom and an O2 molecule.
The calculated binding energy of a Au atom to a clean,
stoichiometric (i.e., no oxygen vacancies) rutile surface is
0.45 eV, and the Bader charge on the adsorbed Au atom is
+0.05 electron.16 When a base is present on the surface, the
binding energy of Au to the surface changes (from 0.45 eV on
clean TiO2) to a value between 1.2 and 1.37 eV (depending on
the base); the charge on the adsorbed Au atom15 is between
−0.35 and −0.47 electron (depending on the base). Note that
these are changes in the binding energy of Au to the TiO2
surface (not to the base).
The binding energy of the Au atom to the rutile surface and
the Bader charge on it are about the same regardless of which of
the above bases is present on the surface. This suggests that it is
the electron donation by the base, not its specific chemical
nature, that causes this increase in the binding energy. This
assumption was tested by performing calculations in which a Au
atom is adsorbed on a rutile slab that contains an extra electron
but no adsorbed Lewis base.15 The presence of an additional
electron turns the rutile slab into a Lewis base, and according to
the rules advocated here this base should bind more strongly the
Au atom (a Lewis acid) than the slab with no additional electron.
The calculations show that indeed it does. The binding energy of
Au on this charged rutile surface is 1.21 eV, and the charge on
the adsorbed Au atom is −0.40 electron. These numbers are
remarkably close to those obtained when the slab is electrically
neutral (i.e., no additional electron in the slab) but has a Lewis
base adsorbed on it. To a good approximation, it is the basicity
that matters, not the specific chemical.
The Au clusters Au3, Au5, and Au7 have an odd number of
electrons, and the presence of an unpaired electron in the
SOMO tends to make them Lewis bases with respect to
TiO2(110). Calculations show that they donate electrons when
they bind to the surface.16,17,20 According to our rules, their
presence on the surface should enhance the binding energy of a
Au atom to the oxide (not to the Au clusters), and they do. On
the other hand, the Au2, Au4, and Au6 clusters are not Lewis
bases or acids when they are adsorbed on rutile16,17 (they do
not exchange electrons with the surface). If the rules we
propose here are correct, the presence of these clusters on the
surface should have no effect on the adsorption energy of the
Au atom to the oxide (not to the Au clusters). Calculations15
show that indeed this is the case. The effect of the Au clusters
with an odd number of electrons on the adsorption energy of a
Au atom has nothing to do with the fact that the clusters
2. COMPUTATIONAL ASPECTS
We compare here results of many calculations: some are old,
and some have not been published before. Some of the older
calculations used GGA, which is known to give some artifacts in
the electronic structure of the oxides of transition metals or
rare-earth metals. The more recent calculations either used5
GGA+U or showed that, for some oxides (e.g., La2O3), using
the Hubbard U correction makes a minor difference in the
results. When we use results obtained with GGA, we are careful
not to include in the present discussion those aspects that are
suspected of being artifacts of the method. For calculations that
used GGA+U, we have used only results that are not changed
qualitatively by changing the value of U. Recently, Nolan6−10
and Watson11,12 and Sanz13 have advocated the use of a
Hubbard correction for oxygen. Having two adjustable
parameters, this method seems to improve the description of
doped or reduced oxides. We found, however, that using two
Hubbard parameters does not affect the qualitative conclusions
reached by using the other DFT methods.
All calculations examined here have been performed on slabs
that mimic large surfaces. They used the accepted standards of
the field regarding the magnitude of the energy cut off, the
thickness of the slabs, and the convergence criteria for
geometry and energy optimization. Most calculations used
spin-polarized DFT to find the lowest energy spin state.
In this article, we present qualitative rules that predict large
energy changes. The magnitude of the energies cited here will
depend on the functional used in the DFT calculations and on
the value of the U parameter. However, the qualitative rules
that we propose are likely to be valid for all functionals because
they are based on large energy differences. We assume that,
10440
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
The Journal of Physical Chemistry C
Feature Article
Br−O)/La2O3 is −0.40 eV (not the 0.2 eV estimated above by
assuming that the two Br atoms do not interact). (Br−La, Br−O)/
La2O3 represents a La2O3 surface on which one Br atom is bonded
to La and the other is bonded to O. The Bader charge on the Br
atom in the Br−La group is −0.71 electron, and this atom is a
strong acid. The Bader charge on the Br atom in the Br−O group
is 0.21 electron, and this atom is a base. Having a Br atom in the
acid Br−La group “directs” the second Br atom to bind to the
basic site Br−O. The resulting acid−base interaction lowers
substantially the energy of dissociative adsorption of Br2.
We have observed similar behavior when Br2 or Cl2
dissociates on CeO2(111).26 We discuss here only the results
obtained for Br2 since those for Cl2 are similar. A Br atom can
adsorb on ceria either on an O atom, to form a compound
denoted (Br−O)/CeO2(111), or on a Ce atom, to form the
compound denoted (Br−Ce)/CeO2(111). The reaction 1/2Br2 +
CeO2(111) → (Br−O)/CeO2(111), in which one Br atom is
adsorbed on the surface, is endothermic, and its energy is +0.23 eV;
the adsorbed Br atom in (Br−O)/CeO2(111) loses +0.26 electron
when it makes the Br−O bond; this Br atom is a Lewis base
with respect to CeO2. The reaction 1/2Br2 + CeO2(111) →
(Br−Ce)/CeO2(111), to form one Br atom adsorbed on a Ce site,
is also endothermic, and the reaction energy is +0.19 eV; this
Br atom gains 0.28 electron when it binds to Ce, and it is a Lewis
acid. Note that no matter where a single Br atom binds the
reaction is uphill. Naively, one would think that if the binding of
one Br atom is exothermic the binding of two of them would be
even more so. However, one of our rules predicts that the
dissociative adsorption of Br2 should be exothermic if one Br atom
binds to O and the other to Ce because this will form a Lewis
acid−base pair on the surface; this should stabilize the dissociative
adsorption. Indeed, it does: the energy of the dissociative adsorption of Br2, to form (Br−O, Br−Ce)/CeO2(111), is −0.27 eV.
The energy to dissociate Br2 and form two Br−O groups or two
Br−Ce groups is much higher, and it is positive (a metastable
state).
We can also use our rules to predict qualitatively the behavior
of HCl dissociation on CeO2(111). The H atom will bind to an
oxygen atom since making a hydroxyl is most often preferred to
making a hydride. Where will the Cl atom bind to provide the
most stable structure when HCl dissociates? Our rule predicts
that because H binding to an oxygen atom on the surface of an
oxide is a base the presence of the hydroxyl will induce the Cl
atom to function as an acid and stabilize the dissociation of HCl
by an acid−base interaction. Indeed, we find that formation of
(H−O, Cl−Ce)/CeO 2 (111) (acid−base pair) is more
exothermic than the formation of (H−O, Cl−O)/CeO2(111)
(base−base pair) by 0.85 eV. When the Cl atom is alone on the
surface and it binds to a Ce atom, it has a Bader charge of −0.33
electron. However, when it is part of (H−O, Cl−Ce)/
CeO2(111), the Cl in the Cl−Ce group acts as a strong Lewis
acid and has a Bader charge of −0.64 electron. Pairing it with H
(a Lewis base) on the surface increases the charge on the Cl (the
Lewis acid) by 0.31 electron and the dissociative adsorption
energy by a substantial amount. A very similar behavior is seen25
for the dissociative chemisorption of HBr on La2O3.
Ultrathin films of oxides supported on metals27−29 provide
another example in which manipulating the Lewis acid−base
properties of a system changes the chemistry of adsorbed
species. Pacchioni et al.30 predicted that Au atoms adsorbed on
an ultrathin MgO film on Ag(100) will draw charge from the
Ag substrate. In this process, the metallic Ag works as a base,
and the Au atom is an acid. This prediction was confirmed by
contain gold; the important feature is that the clusters which
affect the binding of the Au atom are Lewis bases, and the ones
that have no effect are not.
In the examples discussed above, all bases we studied modify
the binding of a Lewis acid which is able to take one electron
(i.e., Au) by roughly the same amount. Similarly, all bases
change the binding energy of oxygen by the same amount. We
do not have enough examples to postulate that this “invariance”
in the binding energy of the Lewis acid with the nature of the
base is a general rule.
If the observations listed above are explained by the acid−
base interaction, similar results should be obtained for other
acids with the same bases. Calculations show15 that this is true
when O2 (a Lewis acid) is coadsorbed on TiO2(110) with one
of the bases listed above. The binding energy of O2 to rutile is
0.15 eV, and the molecule is unable to gain electron charge from
the rutile surface,15 in spite of the fact that O2 has a large
electronegativity; stoichiometric rutile is a poor base. Since DFT
does not account for the van der Waals interactions, the
magnitude of the binding energy is larger than the calculated
one. However, what matters here is that the number is small,
which is in agreement with experiment and other calculations.21−24
The presence on the surface of any of the bases mentioned
above changes the binding energy of oxygen from 0.15 (when
no base is present) to ∼1.00 eV. This is the binding energy to
the rutile surface, not to the base. The charge on the
coadsorbed O2 is ∼ −0.5 electron (instead of zero, when no
base is present). In agreement with our rule, the presence of a
base on the surface allows oxygen to act as a Lewis acid, and
this causes a substantial increase in its binding energy to the
surface. The binding energy of O2 is roughly the same for all
bases, including a slab with an additional electron and no
adsorbate on its surface.
This behavior is not confined to rutile or to the acids and the
bases mentioned above. A more interesting and somewhat
surprising behavior is observed when a halogen or a halogen
hydride is adsorbed on La2O3 or on CeO2.
The reaction 1/2Br2 + La2O3(001) → (Br−La)/La2O3, in
which one Br atom binds to three La atoms, is endothermic,25
and the reaction energy is 0.10 eV. The symbol (Br−La)/La2O3
is used to indicate that the Br atom formed by dissociation
binds to La not to O. This Br atom is bonded to the surface,
but the state is metastable with respect to the formation of
1/2Br2 in the gas. The Bader charge on the Br atom in the Br−La
group is −0.40; this atom is a Lewis acid with respect to La2O3.
Br−La is the lowest-energy binding site of a Br atom on La2O3.
On the basis of what we just said about the adsorption of
one Br atom, one is tempted to guess that the dissociative
adsorption of Br2 to form two adsorbed Br atoms ought to be
even more endothermic than the adsorption of one Br atom
and that both Br atoms formed by dissociation will form a Br−
La complex. If we assume that there is no interaction between
the two Br atoms formed by the dissociation of Br2, the
dissociative adsorption energy ought to be 0.2 eV.
However, according to our rules, the dissociative adsorption
energy of Br2 (to form two adsorbed Br atoms) will benefit
substantially if one atom binds to a surface site on which it is a
Lewis acid and the other binds to a site where it is a Lewis base.
Calculations show25 that this is what happens. The lowestenergy state for the dissociative adsorption of Br2 is the one in
which one Br atoms binds to La (to form the Br−La complex)
and the other Br atom binds to an oxygen atom (to form a Br−O
group). The energy of the reaction Br2 + La2O3 → (Br−La,
10441
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
The Journal of Physical Chemistry C
Feature Article
experiments in Freund’s group.31 The ability of the metal
support to act as a base depends on its work function. When
the work function is smaller, the LUMO of the metal is higher,
and the metal is a stronger base. Therefore, our rules predict
that the binding energy of a Au atom or an O2 molecule (which
are both Lewis acids) to an ultrathin MgO layer supported on a
metal should increase as the work function of the metal
decreases. Calculations by Hellman et al.32,33 show that this is
indeed the case. The binding energy of oxygen to a MgO film
supported on Ag(100) is 0.82 eV, and it is 0.2 eV for MgO.32
The binding energy of a Au atom on an ultrathin film of MgO
supported on Mo, Ag, Pd, Au, or Pt is higher than on MgO, and
it increases as the work function decreases (i.e., base strength
increases).33
A similar example is provided by calculations regarding the
effect of two Au strips present on the rutile surface and used as
Lewis bases.15 One strip has nine atoms in the supercell (and is
denoted here Au9), while the other has ten (Au10). We
emphasize that Au9 and Au10 are not gold clusters but infinite
strips of Au along one lattice vector of the surface. The Au10
strip is obtained from Au9 by adding one Au atom on top of the
Au9 strip. This addition does not cause a substantial change in
the structure of the Au atoms originally in the Au9 strip, but the
chemical properties of Au10 are very different from those of
Au9. The question is how these strips affect the binding of a Au
atom or of an O2 molecule to the rutile surface (not to the strip
but to the oxide surface not covered by the strip). Both strips
affect equally the binding of a Au atom to the rutile surface: the
binding energies are 1.12 eV (on Au9) and 1.18 eV (on Au10).
Given the errors in DFT, these numbers should be considered
equal. The binding energy of O2 changes dramatically, from
1.11 eV for Au9 to 1.96 eV for Au10. Adding one Au atom on
top of Au9 makes it a stronger Lewis base, but this change can
be “detected” only by coadsorbing O2, which is a stronger acid
than Au. An examination of the density of states shows that the
HOMO of Au10 has substantially higher energy than the
HOMO of Au9, and this explains why Au10 is a stronger base
than Au9. Surprisingly, the difference in the binding energy of
O2 on the two surfaces (one with Au9 and the other with Au10)
is roughly equal to the difference in the HOMO energies of Au9
and Au10. It is prudent to think of this equality as a coincidence
rather than a rule.
It is, however, reasonable to generalize these two observations. The acidity of Au is “neutralized” readily, and
increasing the strength of the base beyond a certain value has
no effect on the adsorption energy of Au. On the other hand,
O2, which is a stronger acid, is not neutralized as readily, and its
binding energy can be increased by increasing the strength of
the base.
We have already mentioned that the binding energy of Au on
an ultrathin MgO film supported on a metal increases as the
work function of the metal decreases.33 A smaller work function
means a higher HOMO energy and hence a stronger basicity.
Again, increasing the strength of the base results in a larger
binding energy for the acid.
4. EFFECT OF THE ACID−BASE STRENGTH
It is possible to state that a Lewis acid (or base) is stronger than
another Lewis acid (or base), and there are many methods for
establishing acidity (basicity) scales. Here we use the term
strength in a qualitative and limited sense. Rather than try to
define it precisely, we clarify it by the examples given below.
One definition of a Lewis acid or base makes use of orbital
energies. Qualitatively we can say that A is a stronger Lewis acid
than B if the LUMO of A has lower energy than that of B.
Similarly, A is a stronger base than B if the HOMO of A has
higher energy than that of B. A deficiency of this definition is
that HOMO and LUMO energies are not observable quantities
(only the excitation energies are), but numerous examples have
shown that orbital energies are useful for predicting trends in
chemical behavior.34,35
One difference between Au and O2 is that O2 has two
unpaired electrons, located in two different SOMOs, while the
Au atom has only one. It is therefore qualitatively reasonable to
say that O2 is a stronger acid than Au since it can use two orbitals
for “housing” electrons. How will Au and O2 behave if they are
adsorbed on oxide surfaces having different base strength?
We have already mentioned that the binding energies of Au
and O2 to TiO2(110) are substantially increased if a hydroxyl
(a Lewis base) is present on the surface. What happens if we
make the surface a stronger base by having two hydroxyls per
supercell, next to each other, on the bridging-oxygen row? It
turns out that the binding energy of Au to the surface with two
hydroxyls is the same as that obtained when only one hydroxyl
is present.15 Au is a weak acid, and it behaves as if it is only able
to make use of one electron; the existence of an additional
electron, provided by the second OH, makes no difference.
One hydroxyl already “neutralizes” the acidity of Au; increasing
the basicity (by adding an extra hydroxyl) no longer affects the
behavior of Au. In contrast, adding a second hydroxyl changes
the binding energy of oxygen from 0.93 eV (when coadsorbed
with one hydroxyl) to 1.94 eV (when coadsorbed with two
hydroxyls). The Bader charge on O2 changes from −0.47 electron (for one hydroxyl) to −0.91 electron (for two hydroxyls).
Making the surface more basic increases the binding energy of a
strong acid.
A similar behavior is observed if we add electrons to the
rutile slab to turn it into a base. The binding energy of Au is
the same whether we add one electron or two electrons, but
the binding energy and the Bader charge of O2 depend on the
number of electrons: adding two electrons has about the same
effect as having two hydroxyls. Again, roughly speaking, the
effect of the base depends on its ability to donate electrons and
not on other chemical details.
5. STRUCTURAL EFFECTS OF PAIRING A LEWIS ACID
WITH A LEWIS BASE
We have already mentioned that the lowest-energy Au4 cluster
does not exchange electrons with rutile when it is adsorbed on
its surface. However, Au4 has many structural isomers,15 which
differ from each other through the arrangement of the Au
atoms in the cluster and/or through their location on the
surface. One of them, which we denote Au4*, is less stable than
Au4: the energy of the rutile with Au4* on it is 0.53 eV higher
than the energy of rutile with Au4 adsorbed on it. Bader charge
calculations show that Au4* is a Lewis base with respect to
rutile (i.e., it transfers electrons to the surface). Given the big
difference between the energies of these two adsorbed isomers,
practically no Au4* will be present on the surface if the isomers
of Au4 are in thermodynamic equilibrium.
Calculations15 show that the binding energy of O2 to the
rutile surface (not to the Au cluster) is not affected by the
preadsorption of Au4 on the surface. This is consistent with our
rules: Au4 is not a Lewis base, so it will not affect the binding of
a Lewis acid (the O2). On the other hand, our rules suggest that
adsorbed O2 (a Lewis acid) would interact strongly with Au4*
10442
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
The Journal of Physical Chemistry C
Feature Article
It is striking that the binding energy of O2 to the rutile
surface having an oxygen vacancy is 1.87 eV, and the Bader
charge is 0.87 electron; the binding energy of O2 to a rutile slab
having two electrons is 2.10 eV, and the Bader charge is 0.95
electron; and the binding energy of O2 to a rutile slab having
two hydroxyls on the bridging oxygen row is 1.94 eV, and the
Bader charge is 0.91 electron. To a rough approximation, these
three basic surfaces have the same effect because they all
provide two electrons to the O2 acid.
An interesting situation occurs in the case of a Cl vacancy in
LaOCl.70 Removing a Cl atom from the surface layer leaves
behind one unpaired electron (not two as in the case of an
oxygen vacancy). Nevertheless, the surface with a missing Cl
atom is a strong base, and because of this it adsorbs O2
(a strong Lewis acid) readily, even on sites away from the
vacancy. If the surface does not have a Cl vacancy, it will not
adsorb O2. This is similar to the behavior of an oxygen vacancy
except that the oxygen vacancy is a stronger base since it has
two electrons to donate.
It is possible to create a Cl vacancy on LaOCl by a spillover
process: the Cl atom is removed from its normal lattice site,
and it is adsorbed on the surface, rather than being taken away
in the gas (as 1/2Cl2). This process is endoergic, but on some
of the faces of LaOCl the energy required is small: for example,
the spillover energy for LaOCl(110) is 0.10 eV; the largest
spillover energy is ∼2 eV, for LaOCl(001). In contrast, the
energy to make a Cl vacancy and form 1/2Cl2 in gas is between
2.76 and 4.23 eV, depending on the crystal face. It is clear that
it is easier to make Cl vacancies by spillover than by removing
Cl to form 1/2 Cl2 in the gas. The reason for this is the Lewis
acid−base interaction. The adsorbed Cl atom, formed by
spillover, is a Lewis acid, and the Cl vacancy is a Lewis base.
According to the rules proposed here, forming this Lewis acid−
base pair should benefit from an acid−base interaction: the acid
and the base neutralize each other, and this leads to a gain in
energy. This is why it requires much less energy to make a
vacancy by spillover than by removing Cl from the surface to
form 1/2Cl2 in the gas. A consequence of this neutralization is
that the Cl vacancy made by spillover is no longer a base; the
electron left behind in the Cl vacancy (the base) is used by the
spilled Cl atom to bind to the surface. According to our rules,
the surface formed by a Cl spillover is not a base (it has been
neutralized by the spilled Cl atom), and it should adsorb O2 (a
Lewis acid) weakly. Calculations show that this is the case. In
addition, our rules predict that if the Cl vacancy is made by
removing the Cl atom in the gas, then the defective surface is a
base and will adsorb O2 strongly (on the vacancy site or on
another site on the surface). Calculations show that this is
indeed the case.
(a Lewis base). Calculations show that this is the case: the
energy of the rutile slab on which O2 is coadsorbed with Au4* is
lower by 0.67 eV than that of the rutile slab on which O2 is
coadsorbed with Au4. The presence of the Lewis acid (i.e., O2)
stabilizes that Au cluster that is a Lewis base (namely Au4*). In
the absence of oxygen, the most stable four-atom Au cluster on
the surface is Au4, but when O2 is adsorbed on the rutile
surface, the most stable cluster is Au4*. The presence of the
Lewis acid on the surface forces the four-atom Au cluster to
change its structure to become a Lewis base. The same behavior
has been observed for the Au6 clusters coadsorbed with O2.
In an experiment in which a surface covered with four-atom
Au clusters is exposed to oxygen, the situation is more
complicated because O2 reacts with small Au clusters.36
The rules proposed here can be applied to the general case
when an adsorbed molecule has two isomers A and B whose
binding energy to an oxide surface does not differ by a very
large amount (no more than ∼0.4 eV). If isomer A is a Lewis
acid, coadsorption with a Lewis base will force the molecule to
adopt the structure A. If the isomer B is a Lewis base,
coadsorption with a Lewis acid will force the molecule to have
the structure B.
6. OXYGEN VACANCIES AS LEWIS BASES
Removing an oxygen atom from an oxide surface (to make an
oxygen vacancy) leaves behind two unpaired electrons. The
properties of the oxygen vacancies and the fate of the unpaired
electrons left behind have been the subject of many
calculations12,19,24,37−62 and some experiments.62−64 Oxygen
vacancies are of interest because the energy of vacancy
formation is an indirect measure of the ability of the oxide to
work as an oxidant in catalytic oxidation taking place by a
Mars−van Krevelen mechanism.65−67
There are differences in the behavior of oxygen vacancies
depending on whether the oxide is reducible or not. For the
purpose of the present work, we say that the oxide A is reducible
when the cation in it is able to form more than one stable oxide.
For example, CeO2 and TiO2 are reducible because Ce2O3 and
Ti2O3 are stable; MgO or CaO are irreducible.
Because the formation of an oxygen vacancy produces two
unpaired electrons, a surface with an oxygen vacancy on it is a
Lewis base. It will therefore tend, according to our rules, to
interact strongly with a Lewis acid. This means that if a Lewis
acid is present on the surface of an oxide the energy of making
an oxygen vacancy is considerably lowered by the acid−base
interaction. Another way to say this is that the presence of
oxygen vacancies on the surface increases the binding energy of
a Lewis acid (to the site vacated by the oxygen or to another
site on the surface).
Given the fact that the vacancy site in the surface of an
irreducible oxide is electron-rich, it is not a surprise that
electrophilic compounds such as O2, X2 (X is a halogen), or a
Au atom (all these are Lewis acids) will bind readily to the site
vacated by oxygen.24,68,69 More interesting is the fact that the
presence of an oxygen vacancy affects the binding energy of a
Lewis acid away from the vacancy site. One example is provided
by a TiO2(110) surface that has an oxygen vacancy in the
bridging oxygen row. The presence of the vacancy increases
the binding energy of a Au atom (to the surface, not to the
vacancy) to ∼1.2 eV as compared to 0.45 eV on the surface
without vacancy. This effect is present whether the oxygen
vacancy is in the first or the second oxygen layer.
7. LOW-VALENCE DOPANTS TURN THE OXIDE
SURFACE INTO A LEWIS ACID
There is much work, both experimental71 and computational,6,8−10,38,46,47,52,54,72−85 that explores the possibility that
the catalytic activity of an oxide can be improved by
substitutional doping (a small fraction of the cations of the
oxide are replaced with different cations). In this section, we
point out that one can turn an oxide surface into a Lewis acid if
one substitutes the cation of the host oxide with a lowervalence cation46,77,86,87 (which we call a low-valence dopant
(LVD)). For example, we could replace a La atom in La2O3
with a Mg atom or a Zn atom. Similarly, a cerium atom in CeO2
can be replaced with a La or a Y atom, etc. When making such a
10443
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
The Journal of Physical Chemistry C
Feature Article
substitution, we create an electron “deficit” in the oxide. In the
first example, the La atom being replaced provided (formally)
three electrons to the oxide, and its replacement (Mg or Zn)
provides only two. The substitution creates a “need” for an extra
electron, and this turns the doped surface into a Lewis acid.
The notion that the oxide doped with a LVD is a Lewis acid
is supported by the analysis of the density of states. It has been
found46,77,86,87 that doping creates a hole in the oxide. The
LUMO is shifted, by doping, from the bottom of the valence
band to a state of lower energy. Therefore, doping increases the
ability of the oxide to accept electrons, making the system a
stronger acid. According to our rules, a base binds substantially
more strongly to the surface of the doped oxide than to the
surface of the undoped one.
The rule that a Lewis base binds more strongly to the surface
of an oxide doped with a LVD works whether the oxide is
reducible or not. However, the details of the mechanism of
charge exchange between the acid and the base can be different
on reducible oxides than on the irreducible ones. The reducible
oxides are more complicated because their cations are Lewis
acids (see Section 9): in the presence of a base they can be
reduced from Ce4+ to Ce3+ or from Ti4+ to Ti3+. Therefore, a
reducible oxide doped with a LVD has two acidic sites: the hole
created by the presence of the LVD and the reducible cation
(e.g., Ce4+ or Ti4+). When we adsorb a base on such a surface,
the two acids compete for the electron donated by the base. It
is not clear a priori whether the electron donated by the base
fills the hole or reduces the cation. In an irreducible oxide
doped with a LVD, the base always fills the hole since the
cations cannot be reduced.
(1) Since making an oxygen vacancy is equivalent to turning
the surface of the oxide into a Lewis base, our rules predict that
making an oxygen vacancy, which is a base, requires
substantially less energy on an oxide doped with a LVD than
on the undoped oxide. This happens because the LVD turns
the surface into a Lewis acid, and the formation of the vacancy
creates a Lewis base. The energy of making a vacancy is
lowered by the large Lewis acid−base pair interaction. This
“prediction” has been verified, without exception, in all
calculations we are aware of,9,38,46,52,72,73,80,87−91 and we give
here only a few examples to illustrate the magnitude of the
changes of energy of vacancy formation caused by the acid−
base interaction. The energy of the reaction Ox → Oxv +
1/2O2(g) (here Ox is an oxide and Oxv is the same oxide with
an oxygen vacancy in the top layer of the supercell) is defined
here as the energy of oxygen-vacancy formation. Making a
vacancy on the surface of La2O3(001) requires 6.44 eV.46 Doping
with Cu reduces this number to 0.52 eV, doping with Zn to
2.01 eV, and doping with Mg to 2.59 eV. We have observed
similar effects when doping CaO or ZnO with alkali79 or TiO2
or CeO2 with Au.73,80 In all cases that have been examined,8,9,25,38,46,47,52,54,72−85 the reduction in the vacancy formation
energy, due to the presence of the dopant, is very large.
This effect is relevant to oxidation, or oxidative dehydrogenation reactions catalyzed by oxides, that proceeds through the
Mars−van Krevelen (MvK) mechanism.65−67 In this mechanism, the reductant is oxidized by the oxygen atoms from the
surface of the oxide (not by adsorbed oxygen). The easier it is
to make a vacancy, the better oxidant the surface is.73,80 This
implies that doping an oxide surface with a LVD makes it a
better oxidant. This is not to say that the best oxidant is the
best oxidation catalyst. In the MvK mechanism, the gas-phase
oxygen reoxidizes the surface reduced by the reactant (e.g., CO
or CH4).73,80 If the oxygen is too easy to remove, it is difficult
to put it back. In designing an oxidation catalyst, by doping an
oxide, one should aim for the middle ground: the dopant
should make it easy to remove an oxygen atom, but not too
easy. We call this a “moderation principle”.
(2) If doping an oxide with a LVD turns it into a Lewis acid,
then doping should increase substantially the binding energy of
any Lewis base, not just the formation of an oxygen vacancy.
Calculations show that this is indeed the case.73,77,80,86 This is
true of CO adsorption, which is a Lewis base, whose binding
energy to the oxygen atom next to a LVD is substantially larger
than the binding energy to the undoped oxide.54,72−74,80,88,92,93
The binding of a H atom (a Lewis base) provides another
example.77,86,94 The energy of the reaction 1/2H2 + Ox → H/Ox
(here H/Ox is the oxide with a H adsorbed to make a
hydroxyl) is 1.53 eV if Ox = La2O3 and −2.35 eV if Ox =
MgLa2O3; it is −1.24 eV if Ox = CeO2 and −2.94 eV if Ox =
LaCeO2 (here MgLa2O3 is La2O3 doped with Mg, and LaCeO2
has a similar meaning). This large shift in the binding energy
has been observed for all surfaces doped with LVDs for which
we have calculations. A very similar shift is observed for the
energy of the reaction CH3(g) + Ox → CH3/Ox since CH3 is a
Lewis base. The energy of the reaction is −4.56 eV if Ox is
LaCeO2(111) and −2.71 eV if Ox is CeO2; it is 0.56 eV if Ox =
La2O3 and −3.26 eV if Ox = MgLa2O3(001).
It is interesting to see how doping with a LVD affects the
reaction CH4(g) + Ox → (H−O, CH3−O)/Ox, which is the
dissociative adsorption of methane to form a hydroxyl and a
methoxy group on the surface. CH3 and H are both Lewis
bases. The dissociative adsorption energy is −1.31 eV if Ox =
CeO2(111) and −2.93 eV if Ox = LaCeO2(111). The La-doped
ceria surface is a Lewis acid, and doping does help the
dissociation of CH4 because the reaction creates two Lewis bases
(H and CH3). However, the change in the adsorption energy
caused by doping with the LVD is not as large as one would
expect from knowing the binding energy of H alone and of CH3
alone (which are very large). This happens because one of the
bases neutralizes the acidity of the surface, and the second base
no longer benefits from a Lewis acid−base interaction.77,86
The calculations show that increasing the Lewis acidity of an
oxide surface, by doping an oxide with a LVD, increases the
energy of the dissociative adsorption of methane38,47,78,86 and
lowers the activation energy38,47,78 for breaking the C−H bond.
This is important because breaking this bond is the rate-limiting
step for methane activation, a process that is becoming more
important due to recent changes in the technology of methane
extraction, which dramatically increased the supply of natural gas
and caused its price to collapse. Converting methane to high
value chemicals has become a priority for the chemical industry.
The chemistry of the reducible oxides doped with LVDs is
complicated by the fact that the cations are acidic and the
dopant-induced acidity is additional to that already present in
the oxide. If the oxide is reducible, a base binds strongly to the
surface and donates electronic charge to fill the hole created by
the LVD. In an irreducible oxide, the electron charge donated
by the base may either fill the hole or reduce the cation. Several
of Nolan’s papers have examined how these two mechanisms
compete.6,8,9,52,72 The rule we propose here, that the acidity of
the surface enhances the binding energy of a base, is valid
regardless of the details of the mechanism of charge exchange.
10444
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
The Journal of Physical Chemistry C
Feature Article
8. CHEMICAL COMPENSATION EFFECT IS EQUIVALENT
TO THE NEUTRALIZATION OF AN ACID BY A BASE
In a previous work,77 we have proposed a chemical compensation rule, which works as follows. An oxide surface doped
with a LVD lowers the energy of oxygen-vacancy formation,
increases the binding energy of H and CH3 to surface oxygen,
and makes the dissociative adsorption of CH4 (with H and CH3
bonded to surface oxygen) more exothermic than on the
undoped oxide. The effect of the dopant on the energy of
these processes is very substantial, of the order of several
electronvolts.
According to the rule proposed here, we can neutralize the
effect of a Lewis acid by adsorbing on the surface a Lewis base.
Calculations show77 that such neutralization takes place and has
a large effect on the chemistry of the surface. Consider, as an
example, what happens when we adsorb a H atom (a base) on
La-doped CeO2 (an acid) to form the system we denote
H/LaCeO2. The energy to make an oxygen vacancy is 3.00 eV
on CeO 2(111), 1.36 eV on LaCeO 2, and 2.99 eV on
H/LaCeO2(111). The energy of oxygen-vacancy formation on
H/LaCeO2 is essentially the same as that on CeO2; the
presence of the adsorbed hydrogen cancels the effect of the La
dopant. Similar behavior was found for Mg-doped La2O3 or
CaO doped with alkali. However, the complete compensation
observed in the case of H/LaCeO2 is an exception. For
example, the energy of oxygen vacancy formation is 5.77 eV
on CaO(001), 3.13 eV on NaCaO(001), and 5.33 eV on
H/NaCaO(001). The compensation does not return the energy
of vacancy formation exactly to the value on the undoped oxide.
If this compensation effect is caused by the acid−base
interaction, it should be observed for other bases besides the
hydroxyl formed by adsorbing H. This is indeed the case: the
adsorption of CH3, instead of H, has the same effect. Moreover,
if we do not adsorb a Lewis base but put instead an extra
electron in the slab, we observe the same compensation77 of the
acidity of the doped surface.
base, and the cations are the acid) depends on the location of
the polarons and can be as large as 1 eV.40,105 Thus, the
formation of a base on the surface (i.e., an oxygen vacancy) is
facilitated by the acidic character of the cations in the oxide.
The fact that Ce ions are Lewis acids has interesting
consequences when ceria is doped with a high-valence dopant
(HVD). A HVD is a dopant that has a higher valence than the
cation it substitutes. For example, pentavalent Ta is a HVD
when it dopes ceria. A HVD is a Lewis base because it has an
excess of valence electrons, when compared to the cation it
substitutes. According to our rules, we expect that the HVD is
affected strongly when it dopes a reducible oxide. Such an
influence has been observed in calculations.87 If CeO2 is doped
with Nb or Ta, these pentavalent dopants lose an electron,
which fills an f-orbital localized on a Ce ion. The formal valence
of this Ce ion is 3+. Similarly, the hexavalent dopants W or Mo
lose two electrons to reduce two cerium ions. The energy of
these systems is lower than the energy that would be obtained if
the formation of Ce3+ was prevented.
This acid−base neutralization has an interesting consequence. Doping ceria with Mo, W, Ta, Nb, Ru, Pt, or Zr has
roughly the same effect on the energy of oxygen vacancy
formation.87 This is unexpected since the Zr, Ta, and Nb form
very stable oxides, and one would think that they would make
strong bonds with the neighboring oxygen atoms in the doped
ceria. Moreover, Ru and Pt oxides are much less stable than
ceria, and one would expect that they would bind the
neighboring oxygen atoms more weakly than Zr, Ta, or Nb.
This is not what is observed: the acid−base neutralization
process mentioned above makes all dopants tetravalent, and
because of this, all these dopants have a similar effect on the
neighboring oxygen atoms.
Further evidence for the acidity of the reducible oxides is
provided by calculations that show that a Cu atom106 or a Au
atom107 adsorbed on ceria acts as a base which donates an
electron to reduce one cerium cation.
It is important to note that to capture the acid behavior of
reducible oxides it is important to use GGA+U. The ability of
an adsorbed atom to function as a Lewis base and reduce a
cation in the oxide is predicated on the ability of the functional
to create states localized on the cation.
9. REDUCIBLE OXIDES ARE WEAK LEWIS ACIDS
Many experiments with ceria or titania have shown14,64,95−103
that, unless special precautions are taken, Ce3+ and Ti3+ are
present in TiO2 and CeO2. Something always reduces some of
the Ce4+ or Ti4+ cations. Moreover, it was observed that the
formation of oxygen vacancies or the presence of hydroxyls on
the surface increases the number of the reduced cations.
Because the vacancies and the hydroxyls are Lewis bases, this
observation suggests that the cations in these two oxides act as
Lewis acids. According to our rules, this means that the
adsorption of a base on the surface is facilitated (i.e., the binding
energy is larger) by the acidity of the cations in the oxide.
Recent calculations have helped clarify the interaction
between oxygen vacancies and acidic properties of the reducible
cations in ceria6,12,58,104 and rutile.40,105 These calculations have
shown that the two unpaired electrons created when an oxygen
atom is removed (to create an oxygen vacancy) become
localized on two Ce or two Ti atoms, reducing them from a
formal charge of 4+ to 3+. The reduction of the cations is
accompanied by a displacement of the oxygen atoms from their
normal position (i.e., in the stoichiometric oxide) to form a
polaron. According to our rules, the formation of an oxygen
vacancy, which is a Lewis base, must be aided by the Lewis
acidity of the cations being reduced by the unpaired electrons.
The magnitude of the acid−base interaction (the vacancy is the
10. HIGH-VALENCE DOPANTS ARE LEWIS BASES
High-valence dopants in an irreducible oxide have a different
behavior than in a reducible one. To illustrate the difference, we
contrast the effect of Ta dopant in ZrO2 and in CeO2. In ceria,
a Ta dopant donates an electron87 to the oxide to reduce a Ce4+
to Ce3+. Such reduction does not occur when Ta dopes ZrO2
since Zr2O3 does not exist, and there is no driving force to
create Zr3+. When it replaces a Zr atom in ZrO2, a Ta atom can
use only four of its five valence electrons. The “unused”
electron stays on the Ta dopant, turning it into a strong Lewis
base. We have found this to be true for many HVDs in the
irreducible oxides CaO, La2O3, MgO, ZnO, and NiO.
According to our rules, a HVD is a Lewis base, and its
presence will make it more difficult to remove an oxygen atom
from the surface. Making an oxygen vacancy creates a base on a
surface that is already a base (the oxide doped with HVD), and
this requires more energy than making the same base (the
vacancy) on the undoped oxide (which is neither an acid nor a
base). Calculations show that this effect is very small on oxides
that bind the oxygen very strongly (e.g., CaO, MgO, or La2O3),
but it is substantial on less stable oxides such as ZnO79 or NiO.
10445
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
The Journal of Physical Chemistry C
Feature Article
The energy of making an oxygen vacancy in NiO(011) is
3.28 eV, and it is 4.13 eV for Nb-doped NiO, 4.25 eV for
Ti-doped NiO, and 4.6 eV for Al-doped NiO. A similar increase
in the oxygen vacancy formation energy occurs for all HVDs in
ZnO.79 By strengthening the bond of the oxygen atoms to the
oxide, these dopants hinder oxidation by a Mars−van Krevelen
mechanism.
We are interested in doped oxides because of their potential
of being good catalysts for oxidation reactions. Under catalytic
conditions, the doped oxide will be in contact with gas-phase
oxygen. Because the HDV is a Lewis base, our rules indicate
that the dopant will adsorb strongly Lewis acids such as O2. It is
therefore necessary to examine the chemical properties of this
system. We denote the group formed by O2 adsorption on the
dopant by DO2 where D is the dopant. Similarly, we use DO
for the dopant with an oxygen atom (captured from the gas, not
from the oxide) bound to it.
We anticipate that oxides with a DO2 or a DO dopant will
fall into two groups. (1) If the dopant is not a very strong base,
it adsorbs O2 (or X2), but it does not manage to neutralize its
acidity completely. Because of this, the adsorbed O2 is capable
of reacting with Lewis bases. (2) If the dopant is a very strong
base, it neutralized the (Lewis acid) O2 completely, so it will no
longer react with Lewis bases. In this case, it is profitable to
think that DO2 is a substitutional dopant. Since the O2 ties
down the “extra electrons” in the dopant, the DO2 behaves as if
it is a LVD; the surface becomes acidic, with the attendant
change in surface chemistry.
We discuss these limiting cases separately.
10.1. HVDs Adsorb Oxygen and Activate It. If the HVD
is a weak base, it will adsorb O2, but it will not neutralize
completely its acidity. Therefore, O2 will react readily with
bases such as CO, H, or CH3. This “prediction” has been
confirmed by several calculations. If ZnO is doped with Ti (or
Al), the oxygen molecule adsorbs on the dopant.76 The Bader
charge on O2 increases, and the bond length of the adsorbed
molecule is larger than that of the gas-phase molecule. The
chemical properties of this adsorbed species are closer to those
of O2− than to those of O2. DFT calculations76 show that the
O2 adsorbed on the HVD reacts readily with CO, initiating a
sequence of reactions that result in the CO oxidation to CO2.
This mechanism is the antipode of the Mars−van Krevelen
mechanism: the oxygen in the oxidized species comes from the
adsorbed molecule not from the surface of the oxide. This has
been confirmed by C16O oxidation experiments,76 catalyzed by Tidoped ZnO, which used gaseous 18O2 and produced C18O16O.
This mechanism is also active for Zr-doped CaO(001) and
La-doped CaO(001), which adsorb O2 from the gas phase with
the adsorption energy of −2.67 and −2.66 eV, respectively. The
binding energy is so large because of the acid−base stabilization
proposed here. The bond length of O2 chemisorbed on the Zr
dopant is 1.35 Å, while the bond length of the gas phase O2 is
1.20 Å; this suggests the formation of a chemically active,
negatively charged O2 species. Methane reacts with the
adsorbed O2 to form Zr−O−H and Zr−O−CH3; the two
oxygen atoms present in these species are from the O2
chemisorbed on Zr (Figure 1). The energy of this dissociative
adsorption reaction is −0.84 eV. One can understand this
behavior in terms of the acid−base properties of the system.
The dopant is not a strong enough base to neutralize completely the O2 acid. Therefore, the O2 in the DO2 group is still
sufficiently acid to bind well the basic fragments H and CH3.
Figure 1. (a) Side view of the structure formed by the dissociative
adsorption of methane on O2 adsorbed on Zr-doped CaO(001). The
dissociation fragments bind to the oxygen atoms of the O2 molecule
adsorbed on Zr. Ca is green; oxygen is red; Zr is blue; C is gray; and H
is white. The dissociation is exothermic, and the reaction energy is
−0.84 eV. (b) Structure in (a) seen from above.
This additional binding energy in the final state helps lower the
energy of the dissociative adsorption of CH4.
Since a HVD is a Lewis base, it should react readily with
other Lewis acids, besides O2. We found that to be the case for
Br2 adsorption on Zr-doped La2O3(001). The gas-phase Br2
(a Lewis acid) adsorbs and dissociates on Zr (a Lewis base), and
the reaction energy is −2.67 eV. This is much larger than the
adsorption of Br2 on La2O3 (which is neither acid nor base) or
on Mg-doped La2O3 (which is acid). Both Br atoms gain a large
Bader charge, 0.68 electron and 0.49 electron, respectively,
indicating that the adsorption is an acid−base reaction.
10.2. HVD Adsorbs Oxygen and Becomes a Different
Dopant. The adsorption of O2 on the dopant ties down some
of the electrons that made the dopant a base. How is the oxide
interaction with the dopant modified when the DO2 group is
formed? To explore this question, it is simplest to consider that
O2 adsorption changed the dopant from D to DO2. Since the
oxygen uses some of the “extra electrons” in D, DO2 has fewer
electrons to affect the host oxide than D. How does this affect
surface chemistry?
This question was posed in an article108 that examined how
rutile TiO2(110) doped with V, Cr, Mo, W, or Mn is modified
by the presence of gas-phase oxygen. A thermodynamic
equilibrium calculation showed that, at the oxygen pressure
10446
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
The Journal of Physical Chemistry C
Feature Article
and surface temperatures relevant to catalysis, these dopants
(except Mn) will have an oxygen atom adsorbed on them.
Therefore, if oxygen is present in the gas, the dopants are Mn,
CrO, MoO, WO, or VO. This is not surprising. In its most
stable oxide, Mn is tetravalent, and one can conjecture that
when it replaces another tetravalent cation (Ti in the case of
TiO2) the Mn is neither an acid nor a base. Therefore, it is
unlikely to bind O strongly. Since V is pentavalent and W and
Mo are hexavalent, they will act as bases and will bind oxygen
well. In the case of Cr, there is some ambiguity: CrO3, CrO,
and Cr2O3 have roughly the same stability, and it is difficult to
predict a priori whether Cr is an acid or a base when it replaces
a Ti atom in TiO2.
To test what kind of dopant DO is (D = Cr, V, W, Mo), Kim
et al.108 calculated the energy ΔEv for making an oxygen
vacancy in the surface of the oxide near the dopant. If the
dopants are V, Cr, Mo, or W, the value of ΔEv is 3.37, 2.67,
3.49, or 3.61 eV, respectively. All “naked” dopants act as HVDs
(they increase the energy of vacancy formation) except Cr
which acts more like a low-valence dopant. The behavior of the
MO dopants is markedly different. The values of ΔEv for the
rutile doped with VO, CrO, MoO, and WO are 1.53, 1.73, 1.71,
and 1.97 eV, respectively, which is a substantial decrease as
compared to the surface doped with V, Cr, Mo, or W or with
the undoped surface. All these dopants behave like LVDs, that
is, like Lewis acids. The O atom, which is a strong acid,
overcomes the basicity of the “naked dopants” D, so that DO
becomes acidic, and the surface of the host oxide changes
accordingly.
These older calculations were performed with GGA, and the
results are questionable. It is likely that a GGA+U calculation
would allow V, Cr, Mo, and W to donate electrons to reduce
the Ti atoms from Ti4+ to Ti3+. In other words, within GGA+U
the rutile cations have some acidity, and they will tend to
partially neutralize the basicity of the HVDs V, Cr, Mo, or W.
However, possible doubts in the reliability of the qualitative
results obtained by GGA are allayed by two attenuating
circumstance. First, the change in ΔEv when the dopant
changes from D to DO is very large, and this provides a margin
of safety. It is likely that GGA+U will also find that the presence
of DO lowers ΔEv. Moreover, the cations in TiO2 are weak
acids: given a choice between neutralizing Ti or forming a bond
with the oxygen, it is likely that the extra electrons that make D
a base will choose the latter.
A similar behavior has been observed for Zr-doped CaO and
La2O3. The formation of ZrO2 does activate the surface oxygen
atoms near the dopant. When this system reacts with CH4, the
alkane dissociates, and it can either form Zr−O−CH3 and Zr−
O−H (Figure 1) or form Zr−O−CH3 and OH (Figure 2), with
the oxygen in OH being the surface oxygen. The latter is
activated by the presence of ZrO2. The energies for these two
dissociative adsorption processes are close to each other (−0.84
and −0.91 eV, respectively)
A more dramatic example of this effect is provided by Nbdoped NiO(011). The Nb dopant, which has five valence
electrons, replaces a divalent Ni atom. Therefore, Nb is a very
strong base. The adsorption energy of O2 on the Nb dopant is
−8.40 eV. This energy is so low that no reaction involving the
oxygen atoms in the NbO2 group will result in gas-phase
products (the Sabatier principle). This means that the exposure
of the Nb-doped NiO(011) surface to oxygen, during an
oxidation reaction, will cause each Nb atom in the surface to
bind two oxygen atoms. If any catalysis is to take place on this
Figure 2. (a) Side view of the structure formed by the dissociation of
methane on Zr-doped CaO(001). Unlike Figure 1, here the H atom
formed by dissociation binds to an oxygen atom from the surface. The
CH3 radical forms a methoxide with one of the oxygen atoms adsorbed
on the Zr dopant. Ca is green; oxygen is red; Zr is blue; C is gray; and
hydrogen is white. The dissociation is exothermic, and the reaction
energy is −0.91 eV. (b) Structure in (a) seen from above.
surface, the catalyst is NiO(001) doped with NbO2. Our rules
can help guess the behavior of this system. Formally, the
adsorption of O2 on Nb brings to the NbO2 four electrons
(O is divalent) so that (formally) the NbO2 group has only one
available valence electron. Since this “monovalent” NbO2 group
replaces a divalent Ni atom, it ought to act as a low-valence
dopant and turn the surface into a Lewis acid. If this is true,
then according to our rules, the presence of NbO2 dopant will
weaken the binding energy of the surface oxygen atoms around
the dopant and make it easier to make oxygen vacancies.
The DFT calculations show that this prediction is correct. In
Figure 3 we show the NiO(011) surface doped with NbO2. We
marked two of the oxygen atoms as A and B. Making an oxygen
vacancy by removing the atom A requires 2.31 eV, and making
a vacancy by removing B costs 2.67 eV. This is substantially
lower than the energy required for making an oxygen vacancy
on the undoped surface, which is 3.28 eV. The NbO2 dopant
acts indeed as a LVD (it is acid), and it facilitates the formation
of oxygen vacancies (which are basic). This surface has another
property typical of an oxide doped with a LVD: the oxygen
atoms near the NbO2 group help the dissociative adsorption of
ethane. Among many possible structures for the dissociation
10447
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
The Journal of Physical Chemistry C
Feature Article
atom, such as Pt, or Au, or Cu, donates electrons to reduce the
oxide’s cation. (g) High-valence dopants are Lewis bases, and they
adsorb strongly Lewis acids such as O2 or X2 (X is a halogen). If
the basicity of the dopant is not excessive, the acidity of the
adsorbed oxygen or Br2 is not completely neutralized, and they are
able to react with Lewis bases. If the dopant is very basic, the O2
molecule (or X2) adsorbed on it is bound too strongly, and it is
not chemically active. However, in some cases the group formed
by the dopant with O2 adsorbed on it acts as a low-valence dopant
and activates surface oxygen.
These rules seem general (we found no exceptions); the
effects predicted by them are large; and they seem to apply any
time a chemical process involves a Lewis acid and a base,
regardless of the chemical nature of the pair.
Figure 3. NiO(011) doped with NbO2 which was obtained by
replacing a surface Ni atom with Nb and then adsorbing an O2
molecule on Nb. The NbO2 group replaces a Ni atom, and it acts as a
LVD (it is a Lewis acid). Because of this, the oxygen atoms marked A
and B are easier to remove, to make oxygen vacancies, than in the case
of the undoped oxide. Ni is purple; the oxygen in the top layer is red;
the oxygen in the second layer is yellow; the oxygen adsorbed on Nb is
dark blue; and Nb is light blue.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: metiu@chem.ucsb.edu.
fragments, several involve the binding of H or CH3CH2 to the
oxygen atoms in the surface of NiO, which are activated by the
presence of the NbO2 dopant. The energy for this dissociative
adsorption reaction, for various final fragment configurations,
varies between −0.06 and −1.06 eV. This is lower than the
dissociation energy on NiO. Doping with NbO2 favors the
breaking of the C−H bond, which is typical of an oxide doped
with a LVD.
Present Address
†
School of Physics, Nankai University, Tianjin 300 071,
People’s Republic of China.
‡
Institute of Metal Research, Chinese Academy of Sciences
(IMR CAS), Shenyang 100 016, People’s Republic of China.
Notes
The authors declare no competing financial interest.
Biographies
11. SUMMARY
We have shown that many trends related to chemical processes
relevant to catalytic chemistry of oxide surfaces can be
predicted (or rationalized) by a few rules pertaining to the
Lewis acid−base properties of the chemicals involved.
These rules provide guidance on possible ways of manipulating
oxide surface chemistry. (a) The binding energy of a Lewis acid
to an oxide surface can be increased substantially if the surface
is modified to become a Lewis base. This modification can be
achieved by preadsorbing on the surface electron donors.
Conversely, the binding energy of a Lewis base can be increased
by modifying the surface to become a Lewis acid, which can be
achieved by doping with low-valence dopants. (b) It is possible
to define, qualitatively, an acid or a base strength, and the acid−
base interaction increases as the strength of the acid or the base
is increased. (c) If a surface is modified to be a Lewis acid, to
confer on it certain properties, these properties can be
substantially suppressed by adsorbing on the surface a Lewis
base. The system behaves as if the base neutralizes the acid.
(d) In some cases, an adsorbed species can have two isomers,
one that is a base and the other that is an acid. If that species is
coadsorbed with a Lewis acid, it will take the structure that is a
base. If it is coadsorbed with a Lewis base, it will take the
structure in which it is an acid. (e) Oxides doped with lowvalence dopants are Lewis acids. Because of this, they lower the
energy of oxygen vacancy formation, increase the binding
energy of various bases (e.g., H, CH3, CO) to the surface of the
doped oxide, and make the dissociative adsorption an alkane
RH more exothermic (because the fragments R and H
produced by dissociation are Lewis bases and the surface is a
Lewis acid). (f) Reducible oxides are Lewis acids, and this
acidity lowers the energy of oxygen vacancy formation and
increases the binding energy of single metal atoms. The vacancy
is a strong base, and the unpaired electrons produced when the
oxygen atom is removed are transferred to the cations of the oxide.
This reduces the energy to form the vacancy. An adsorbed metal
Horia Metiu graduated from the Polytechnic Institute in Bucharest in
1961, received his Ph.D. from MIT in 1974 working with John Ross
and Robert Silbey, and then was a postdoctoral fellow at the University
of Chicago with Karl Freed. He has been a professor in the
Department of Chemistry at the University of California, Santa
Barbara since 1976. He is currently involved in experimental and
computational investigations of catalysis, electrocatalysis, and photoelectrocatalysis.
Steeve Chrétien is currently an assistant research specialist at the
University of California, Santa Barbara. He received his M.Sc. in 1998
and his Ph.D. in 2002 in chemistry at the Université de Montréal
under the supervision of Prof. Dennis R. Salahub. His thesis research
focused on a mechanistic study of the cyclotrimerization of acetylene
to benzene catalyzed by iron clusters in the gas phase using density
functional theory (DFT). His current research focuses on using DFT
to study the catalytic activity of doped oxides and metallic clusters
supported on oxides.
Zhenpeng Hu received his B.S. Degree in Applied Chemistry from the
University of Science and Technology of China in 2002 and earned his
Ph.D. degree in Condensed Matter Physics there in 2008 under the
supervision of Prof. Jianguo Hou and Prof. Jinlong Yang. From 2008
to 2011, he did postdoctoral research with Prof. Horia Metiu and Prof.
Eric McFarland at the University of California, Santa Barbara. In 2012
he joined the faculty of Nankai University.
Bo Li received his M.S. in Physical Chemistry from Jilin University in
2004 and his Ph.D. in Physics in 2009 from the theory department of
the Fritz-Haber-Institüt under the supervision of Prof. Matthias
Scheffler and Prof. Angelos Michaelides. Then he went to the
University of California, Santa Barbara and joined Prof. Horia Metiu’s
group as a Postdoc. Currently, he is a staff scientist in the Institute of
Metal Research, Chinese Academy of Sciences.
XiaoYing Sun received her B.E. in Chemistry from Beihua University,
China, in 2001. She received her Ph.D. in Physical Chemistry from
Jilin University in 2006. Then she went to Technische Universität of
10448
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
The Journal of Physical Chemistry C
Feature Article
(31) Sterrer, M.; Risse, T.; Martinez Pozzoni, U.; Giordano, L.;
Heyde, M.; Rust, H.-P.; Pacchioni, G.; Freund, H.-J. Phys. Rev. Lett.
2007, 98, 096107.
(32) Hellman, A.; Klacar, S.; Grönbeck, H. J. Am. Chem. Soc. 2009,
131, 16636−16637.
(33) Frondelius, P.; Hellman, A.; Honkala, K.; Häkkinen, H.;
Grönbeck, H. Phys. Rev. B 2008, 78, 085426.
(34) Chrétien, S.; Gordon, M. S.; Metiu, H. J. Chem. Phys. 2004, 121,
3756−3766.
(35) Chrétien, S.; Gordon, M. S.; Metiu, H. J. Chem. Phys. 2004, 121,
9931−9937.
(36) Chrétien, S.; Buratto, S. K.; Metiu, H. Curr. Opin. Solid State
Mater. Sci. 2007, 11, 62−75.
(37) Cui, X.; Wang, B.; Wang, Z.; Huang, T.; Zhao, Y.; Yang, J.; Hou,
J. G. J. Chem. Phys. 2008, 129, 044703.
(38) Mayernick, A. D.; Janik, M. J. J. Catal. 2011, 278, 16−25.
(39) Cheng, H. Z.; Selloni, A. Phys. Rev. B 2009, 79, 092101.
(40) Chrétien, S.; Metiu, H. J. Phys. Chem. C 2011, 115, 4696−4705.
(41) Dholabhai, P. P.; Adams, J. B.; Crozier, P.; Sharma, R. J. Chem.
Phys. 2010, 132, 094104.
(42) Esch, F.; Fabris, S.; Zhou, L.; Montini, T.; Africh, C.; Fornasiero,
P.; Comelli, G.; Rosei, R. Science 2005, 309, 752−755.
(43) Finazzi, E.; Di Valentin, C.; Pacchioni, G.; Selloni, A. J. Chem.
Phys. 2008, 129, 154113.
(44) Herschend, B.; Baudin, M.; Hermansson, K. Surf. Sci. 2005, 599,
173−186.
(45) Khan, S.; Oldman, R. J.; Cora, F.; Catlow, C. R. A.; French,
S. A.; Axon, S. A. Phys. Chem. Chem. Phys. 2006, 8, 5207−5222.
(46) Li, B.; Metiu, H. J. Phys. Chem. C 2010, 114, 12234−12244.
(47) Mayernick, A. D.; Janik, M. J. J. Phys. Chem. C 2008, 112,
14955−14964.
(48) Minato, T.; Sainoo, Y.; Kim, Y.; Kato, H. S.; Aika, K. I.; Kawai,
M.; Zhao, J.; Petek, H.; Huang, T.; He, W.; Wang, B.; Wang, Z.; Zhao,
Y.; Yang, J.; Hou, J. G. J. Chem. Phys. 2009, 130, 124502.
(49) Morgan, B. J.; Watson, G. W. Surf. Sci. 2007, 601, 5034−5041.
(50) Morgan, B. J.; Watson, G. W. J. Phys. Chem. C 2009, 113, 7322−
7328.
(51) Morgan, B. J.; Watson, G. W. J. Phys. Chem. C 2010, 114, 2321−
2328.
(52) Nolan, M. J. Mater. Chem. 2011, 21, 9160−9168.
(53) Nolan, M.; Parker, S. C.; Watson, G. W. Surf. Sci. 2005, 595,
223−232.
(54) Nolan, M.; Soto Verdugo, V.; Metiu, H. Surf. Sci. 2008, 602,
2734−2742.
(55) Wu, X. Y.; Selloni, A.; Nayak, S. K. J. Chem. Phys. 2004, 120,
4512−4516.
(56) Yang, Z. X.; Yu, X. H.; Lu, Z. S.; Li, S. F.; Hermansson, K. Phys.
Lett. A 2009, 373, 2786−2792.
(57) Zhang, C. J.; Michaelides, A.; King, D. A.; Jenkins, S. J. Phys. Rev.
B 2009, 79, 075433.
(58) Ganduglia-Pirovano, M. V.; Da Silva, J. L. F.; Sauer, J. Phys. Rev.
Lett. 2009, 102, 026101.
(59) Ganduglia-Pirovano, M. V.; Hofmann, A.; Sauer, J. Surf. Sci. Rep.
2007, 62, 219−270.
(60) Deskins, N. A.; Rousseau, R.; Dupuis, M. J. Phys. Chem. C 2011,
115, 7562−7572.
(61) Du, Y. G.; Deskins, N. A.; Zhang, Z. R.; Dohnalek, Z.; Dupuis,
M.; Lyubinetsky, I. Phys. Chem. Chem. Phys. 2010, 12, 6337−6344.
(62) Jerratsch, J. F.; Shao, X.; Nilius, N.; Freund, H. J.; Popa, C.;
Ganduglia-Pirovano, M. V.; Burow, A. M.; Sauer, J. Phys. Rev. Lett.
2011, 106, 246801.
(63) Shao, X.; Jerratsch, J.-F.; Nilius, N.; Freund, H.-J. Phys. Chem.
Chem. Phys. 2011, 13, 12646−12651.
(64) Papageorgiou, A. C.; Beglitis, N. S.; Pang, C. L.; Teobaldi, G.;
Cabailh, G.; Chen, Q.; Fisher, A. J.; Hofer, W. A.; Thornton, G. Proc.
Natl. Acad. Sci. U.S.A. 2010, 107, 2391−2396.
(65) Mars, P.; van Krevelen, D. W. Chem. Eng. Sci. Spec. Suppl. 1954,
3, 41−59.
Berlin as a postdoc. She is currently working as a postdoc in Prof.
Horia Metiu’s group at the University of California, Santa Barbara.
■
ACKNOWLEDGMENTS
This research was supported in part by the University of California
Lab Fees Program (09-LR-08-116809), the U.S. Department of
Energy (DE-FG02-89ER140048), and the Air Force Office of
Scientific Research (FA9550-09-1-0333), and by the National
Science Foundation through TeraGrid resources provided by
Ranger@TACC under grant number TG-ASC090080. We
acknowledge support from the Center for Scientific Computing
from the CNSI, MRL: an NSF MRSEC (DMR-1121053) and
NSF CNS-0960316 and Hewlett-Packard. Use of the Center for
Nanoscale Materials was supported by the U.S. Department of
Energy, Office of Science, Office of Basic Energy Sciences, under
Contract No. DE-AC02-06CH11357.
■
REFERENCES
(1) Lewis, G. N. J. Franklin Inst. 1938, 226, 293−313.
(2) Bader, R. Atoms in Molecules: A Quantum Theory; Clarendon:
Oxford, 1994.
(3) Henkelman, G.; Arnaldsson, A.; Jonsson, H. Comput. Mater. Sci.
2006, 36, 354−360.
(4) Sanville, E.; Kenny, S. D.; Smith, R.; Henkelman, G. J. Comput.
Chem. 2007, 28, 899−908.
(5) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.;
Sutton, A. P. Phys. Rev. B 1998, 57, 1505−1509.
(6) Nolan, M. Chem. Phys. Lett. 2010, 499, 126−130.
(7) Iwaszuk, A.; Nolan, M. J. Phys. Chem. C 2011, 115, 12995−
13007.
(8) Nolan, M. J. Phys. Chem. C 2011, 115, 6671−6681.
(9) Yeriskin, I.; Nolan, M. J. Phys.: Condens. Matter 2010, 22, 135004.
(10) Nolan, M. Chem. Phys. Lett. 2010, 492, 115−118.
(11) Nolan, M.; Watson, G. W. J. Chem. Phys. 2006, 125, 144701−
144706.
(12) Nolan, M.; Grigoleit, S.; Sayle, D. C.; Parker, S. C.; Watson, G. W.
Surf. Sci. 2005, 576, 217−229.
(13) Plata, J. J.; Marquez, A. M.; Sanz, J. F. J. Chem. Phys. 2012, 136,
041101.
(14) Liu, L. M.; Crawford, P.; Hu, P. Prog. Surf. Sci. 2009, 84, 155−
176.
(15) Chrétien, S.; Metiu, H. J. Chem. Phys. 2008, 128, 044714.
(16) Chrétien, S.; Metiu, H. J. Chem. Phys. 2007, 127, 084704.
(17) Chrétien, S.; Metiu, H. J. Chem. Phys. 2007, 126, 104701.
(18) Chrétien, S.; Metiu, H. J. Chem. Phys. 2007, 127, 244708.
(19) Stausholm-Møller, J.; Kristoffersen, H. H.; Hinnemann, B.;
Madsen, G. K. H.; Hammer, B. J. Chem. Phys. 2010, 133, 144708.
(20) Molina, L. M.; Rasmussen, M. D.; Hammer, B. J. Chem. Phys.
2004, 120, 7673−7680.
(21) Rasmussen, M. D.; Molina, L. M.; Hammer, B. J. Chem. Phys.
2004, 120, 988−997.
(22) Lisenbigler, A.; Lu, G.; Yates, J. T. J. Phys. Chem. 1996, 100,
6631.
(23) Henderson, M. A.; Epling, W. S.; Perkins, C. L.; Peden, C. H. F.;
Diebold, U. J. Phys. Chem. B 1999, 103, 5328−5337.
(24) Wu, X. Y.; Selloni, A.; Lazzeri, M.; Nayak, S. K. Phys. Rev. B
2003, 68, 241402.
(25) Li, B.; Metiu, H. J. Phys. Chem. C 2012, 116, 4137−4148.
(26) Hu, Z.; Metiu, H. J. Phys. Chem. C 2012, 116, 6664−6671.
(27) Freund, H. J. Chem.−Eur. J. 2010, 16, 9384−9397.
(28) Freund, H.-J.; Pacchioni, G. Chem. Soc. Rev. 2008, 37, 2224−
2242.
(29) Nilius, N.; Risse, T.; Schauermann, S.; Shaikhutdinov, S.;
Sterrer, M.; Freund, H. J. Top. Catal. 2011, 54, 4−12.
(30) Pacchioni, G.; Giordano, L.; Baistrocchi, M. Phys. Rev. Lett.
2005, 94, 226104.
10449
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
The Journal of Physical Chemistry C
Feature Article
(66) Doornkamp, C.; Ponec, V. J. Mol. Catal. A: Chem. 2000, 162,
19−32.
(67) Vannice, M. A. Catal. Today 2007, 123, 18−22.
(68) Wendt, S.; Schaub, R.; Matthiesen, J.; Vestergaard, E. K.;
Wahlstrom, E.; Rasmussen, M. D.; Thostrup, P.; Molina, L. M.;
Laegsgaard, E.; Stensgaard, I.; Hammer, B.; Besenbacher, F. Surf. Sci.
2005, 598, 226−245.
(69) Chrétien, S.; Metiu, H. J. Chem. Phys. 2008, 129, 0747705.
(70) Chrétien, S.; Metiu, H. J. Phys. Chem. C 2012, 116, 681−691.
(71) Hegde, M. S.; Madras, G.; Patil, K. C. Acc. Chem. Res. 2009, 42,
704−712.
(72) Yeriskin, I.; Nolan, M. J. Chem. Phys. 2009, 131, 244702.
(73) Shapovalov, V.; Metiu, H. J. Catal. 2007, 245, 205−214.
(74) Nolan, M. J. Phys. Chem. C 2009, 113, 2425−2432.
(75) Pala, R. G. S.; Metiu, H. J. Catal. 2008, 254, 325−331.
(76) Pala, R. G. S.; Tang, W.; Sushchikh, M. M.; Park, J.-N.; Forman,
A. J.; Wu, G.; Kleiman-Shwarsctein, A.; Zhang, J.; McFarland, E. W.;
Metiu, H. J. Catal. 2009, 266, 50−58.
(77) Hu, Z.; Li, B.; Sun, X.; Metiu, H. J. Phys. Chem. C 2011, 115,
3065−3074.
(78) Tang, W.; Hu, Z.; Wang, M.; Stucky, G. D.; Metiu, H.;
McFarland, E. W. J. Catal. 2010, 273, 125−137.
(79) Pala, R. G. S.; Metiu, H. J. Phys. Chem. C 2007, 111, 8617−8622.
(80) Chrétien, S.; Metiu, H. Catal. Lett. 2006, 107, 143−147.
(81) Barrio, L.; Kubacka, A.; Zhou, G.; Estrella, M.; Martinez-Arias,
A.; Hanson, J. C.; Fernandez-Garcia, M.; Rodriguez, J. A. J. Phys. Chem.
C 2010, 114, 12689−12697.
(82) Wang, X.; Rodriguez, J. A.; Hanson, J. C.; Gamarra, D.;
Martinez-Arias, A.; Fernandez-Garcia, M. J. Phys. Chem. B 2005, 109,
19595−19603.
(83) Rodriguez, J. A.; Wang, X. Q.; Hanson, J. C.; Liu, G.; Iglesias-Juez,
A.; Fernandez-Garcia, M. J. Chem. Phys. 2003, 119, 5659−5669.
(84) Rodriguez, J. A.; Hanson, J. C.; Kim, J.-Y.; Liu, G.; Iglesias-Juez,
A.; Fernandez-Garcia, M. J. Phys. Chem. B 2003, 107, 3535−3543.
(85) Carrettin, S.; Hao, Y.; Aguilar-Guerrero, V.; Gates, B. C.;
Trasobares, S.; Calvino, J. J.; Corma, A. Chem.−Eur. J. 2007, 13,
7771−7779.
(86) Li, B.; Metiu, H. J. Phys. Chem. C 2011, 115, 18239−18246.
(87) Hu, Z.; Metiu, H. J. Phys. Chem. C 2011, 115, 17898−17909.
(88) Chen, H.-T.; Chang, J.-G. J. Phys. Chem. C 2011, 115, 14745−
14753.
(89) Yang, Z. X.; Ma, D. W.; Yu, X. H.; Hermansson, K. Eur. Phys. J.
B 2010, 77, 373−380.
(90) Yang, Z. X.; He, B. L.; Lu, Z. S.; Hermansson, K. J. Phys. Chem.
C 2010, 114, 4486−4494.
(91) Wang, X. Q.; Shen, M. Q.; Wang, J.; Fabris, S. J. Phys. Chem. C
2010, 114, 10221−10228.
(92) Camellone, M. F.; Fabris, S. J. Am. Chem. Soc. 2009, 131,
10473−10483.
(93) Yang, Z.; Fu, Z.; Zhang, Y.; Wu, R. Catal. Lett. 2011, 141, 78−
82.
(94) Vicario, G.; Balducci, G.; Fabris, S.; de Gironcoli, S.; Baroni, S.
J. Phys. Chem. B 2006, 110, 19380−19385.
(95) Diebold, U. Surf. Sci. Rep. 2003, 48, 53−229.
(96) Diebold, U.; Li, S. C.; Schmid, M. Annu. Rev. Phys. Chem. 2010,
61, 129−148.
(97) Gorte, R. J. AIChE J. 2010, 56, 1126−1135.
(98) Trovarelli, A. Catal. Rev.: Sci. Eng. 1996, 38, 439−520.
(99) Trovarelli, A. Catalysis by Ceria and Related Materials; Imperial
College Press: London, 2002.
(100) Pang, C. L.; Lindsay, R.; Thornton, G. Chem. Soc. Rev. 2008,
37, 2328−2353.
(101) Yim, C. M.; Pang, C. L.; Thornton, G. Phys. Rev. Lett. 2010,
104, 036806.
(102) Yim, C. M.; Pang, C. L.; Thornton, G. Phys. Rev. Lett. 2010,
104, 259704.
(103) Pang, C. L.; Thornton, G. Surf. Sci. 2009, 603, 3255−3261.
(104) Li, H. Y.; Wang, H. F.; Gong, X. Q.; Guo, Y. L.; Guo, Y.; Lu,
G. Z.; Hu, P. Phys. Rev. B 2009, 79, 193401.
(105) Deskins, N. A.; Rousseau, R.; Dupuis, M. J. Phys. Chem. C
2010, 114, 5891−5897.
(106) Szabova, L.; Camellone, M. F.; Huang, M.; Matolin, V.; Fabris, S.
J. Chem. Phys. 2010, 133, 234705.
(107) Hernandez, N. C.; Grau-Crespo, R.; de Leeuw, N. H.; Sanz, J. F.
Phys. Chem. Chem. Phys. 2009, 11, 5246−5252.
(108) Kim, H. Y.; Lee, H. M.; Pala, R. G. S.; Shapovalov, V.; Metiu, H. J.
Phys. Chem. C 2008, 112, 12398−12408.
10450
dx.doi.org/10.1021/jp301341t | J. Phys. Chem. C 2012, 116, 10439−10450
Download