This	document	is	the	Accepted	Manuscript	version	of	a	Published	Work	that	appeared	in	final	form	in	The	Journal	of	Physical	Chemistry	C,	
copyright	©	American	Chemical	Society	aBer	peer	review	and	technical	ediDng	by	the	publisher.	To	access	the	final	edited	and	published	
Letter
work	see	hFps://pubs.acs.org/doi/abs/10.1021/acs.jpcleF.8b00536	
Cite This: J. Phys. Chem. Lett. 2018, 9, 2568−2573

pubs.acs.org/JPCL

One Oxygen Vacancy, Two Charge States: Characterization of
Reduced α‑MoO3(010) through Theoretical Methods
Marcos Rellán-Piñeiro and Núria López*
Institute of Chemical Research of Catalonia, ICIQ, and The Barcelona Institute of Science and Technology, BIST, Av. Països Catalans
16, 43007 Tarragona, Spain

Downloaded via UNIV ROVIRA I VIRGILI on April 15, 2019 at 10:24:37 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S
* Supporting Information

ABSTRACT: Molybdenum oxides are finding increasing applications that rely on their
redox character. For the most common polymorph, α-MoO3, oxygen vacancy formation
leaves two electrons on the surface that can be stored as small polarons. Detailed density
functional theory calculations that properly account for the self-interaction term, Ueff = 3.5
eV, show that the vacancy generates two different configurations: either two Mo5+ centers
(Mo5+□ and Mo5+O) or a single double-reduced Mo4+. These states are separated by
0.22 eV with a barrier for interconversion of 0.33 eV, and thus both are populated at
catalytic temperatures, as shown by first-principles molecular dynamics. At higher
reduction levels, vacancies can only be accumulated along a preferential direction and the
energy difference between the 2×Mo5+ and Mo4+ configurations is reduced. These results
point out the need for a revision of the experimental assignments based on our
characterization that includes charges, vibrational frequencies, and XPS signatures.

T

he properties of oxides are dominated by defects,
commonly oxygen vacancies.1,2 Vacancies affect the
electronic structure of materials, as the two electrons left
upon O removal can be (i) a localized pair at the position of the
anion (like for MgO);3 (ii) separated and localized at the
surrounding cations like for TiO2 and CeO2 effectively leaving
two M(n-1)+;4,5 (iii) partially delocalized between a group of
metals, like for In2O3.6,7 Recently, molybdenum oxides have
attracted a lot of attention due to their large number of new
physical and chemical applications as electro- and chromic
devices, in photo-, electro-, and thermal catalysis, in solar cells,
and as light-emitting diodes, photodetectors/sensors, batteries,
pseudocapacitors, thermoelectric, and ferroelectric materials.8
All of these applications are based on the versatile geometric
(with several structure forms)9,10 and electronic structure
linked to the redox character that can be fine-tuned by the
addition of cations11,12 or hydrogen13−15 or by oxygen
removal,16 which leads to the formation of Magnéli phases.
In the most common polymorph, α-MoO3, Mo centers are in
the oxidation state Mo6+, and surrounded by oxygen atoms in
octahedral environments with a band gap between 3.0 and 3.2
eV.17−19 α-MoO3 has an orthorhombic unit cell with Pbnm
symmetry, where distorted MoO6 octahedra are ordered in
bilayers along the xz plane; see Figure 1. van der Waals
interactions hold the layered structure packed. There are three
nonequivalent oxygen atoms: (i) three coordinated oxygens,
O3c, that form a symmetrical bridge along the x axis between
two Mo centers of the same layer and interact with a sublayer
Mo; (ii) asymmetric oxygens, Ob, that bridge two Mo centers
along the z axis; and (iii) terminal oxygens, Ot, double bonded
to only one Mo center present in the interlayer region. Oxygen
depletion severely affects the electronic structure of the
© 2018 American Chemical Society

Figure 1. Structures of α-MoO3(010) pristine surface, left, and local
structure of the two possible Ot vacancies, right. (a) O3c, (b) Ob, and
(c) Ot. Mo: green and O: red.

material, changing its very large work function (∼6.9 eV) to
lower values ∼6.5 eV for a 10% oxygen-deficient material,
introducing donor gap states.20 Upon reduction, the photoemission spectra present additional features: At low vacancy
concentrations, these features are assigned to the formation of
Mo5+ centers, and after a certain concentration, Mo4+ features
also appear.20−22 Electron paramagnetic resonance (EPR)
studies also indicate the formation of structural point defects
of Mo 5+ close to oxygen vacancies, which at higher
concentrations lead to the formation of other centers associated
with extended shear structures.23 Raman vibrational24 spectroscopy points out the fingerprints for 996, 823 (MO
Received: February 18, 2018
Accepted: April 27, 2018
Published: April 27, 2018
2568

DOI: 10.1021/acs.jpclett.8b00536
J. Phys. Chem. Lett. 2018, 9, 2568−2573

The Journal of Physical Chemistry Letters

Letter
−1

thermochemical data are not available, computed values for
an acid−base and two redox reactions with the hybrid
functional HSE06 were taken as reference. The optimized
value for Ueff is 3.5 eV (Figure S1). This value also provides the
best match in a separate test against the gold standard in
molecular chemistry, the CCSD(T) method (Figure S2).
Either the terminal or the two oxygens in the plane surface
could be the first ones to be removed. Vacancy (V) formation
energies for the reaction MoO3 → MoO3−x + 1/2 O2(g) at the
Ot position are shown in Figure 2. For the Ot−Ob pair it was

asymmetric and symmetric stretchings), and 667 cm
(asymmetric O−Mo−O). Additional Raman features at 1004
and 1008 cm−1 appear at low reduction levels,25 which
disappear at longer reduction times or by hydrogen adsorption
and electron beam treatment.26,27
Understanding the electronic properties of α-MoO3 from
first-principles28 has been hampered by the presence of the
layered structure, the semiconductor character of the material,
and the reducibility of the cation. Hybrid methods have been
applied, for instance, to evaluate the vacancy formation energy
in the bulk, which is endothermic by 3.68 eV within HSE06.29
However, the use of these functionals remains unpractical when
exploring complex reaction networks on the material.
Alternatively, density functional theory (DFT) generalized
gradient approximations (GGAs) have been employed. In αMoO3 studies with pure GGA functionals,30−33 electron
localization is not found, and thus isolated oxidation states
cannot be assigned. The empirical Ueff parameter needs to be
incorporated to achieve localization.34 In the literature,
scattered U values have been used. Bulk calculations with Ueff
= 5.0 eV were employed to analyze the preferential oxygen
eliminated. Ot was identified as the easiest to remove, but the
nature of the charge state was not described.35 With a value of
Ueff = 6.3 eV, vacancy formation at Ot position in the bulk
leads to a Mo4+ center, while the removal of the Ob or O3c
lattice oxygen ends up in the formation of two Mo5+
centers.27,36 A recent study with Ueff = 6.0 eV suggested the
formation of two different states upon reduction, but no
characterization was performed, and low vacancy formation
energies, ∼1.50 eV, were reported.37
In surface models, a value of ∼6 eV was obtained by fitting to
hybrids calculations on cluster structure, and it was used by
Chen,38 Asta,39 and Willock.40 The latter reported the
formation of a Mo4+ center. In turn, Metiu41,42 and Bell43,44
proposed a Ueff value to match the experimental enthalpy of the
reaction MoO3 + H2 → MoO2 + H2O, but they employed very
different values, 2 and 8.6 eV, respectively. Metiu found no
localization upon Ot removal and two Mo5+ when the vacancy
was formed at the O3c position. Therefore, the origin of the
different oxidation states observed in experiments cannot be
assigned from the previous computational studies.
In this work, we reoptimized the Ueff value to describe the
reactivity on α-MoO3(010) accurately following a set of
benchmarks that include the comparison to hybrid functionals
and CCSD(T) references; see Section S1. Then, we employed
PBE+U-D2 to identify the true nature of vacancies in terms of
oxidation state of the cations and the corresponding
spectroscopic fingerprints.
The bulk PBE-D2 calculated lattice parameters are a = 3.921,
b = 14.480, and c = 3.708 Å, in agreement with the
experimental values: a = 3.963, b = 13.855, and c = 3.696
Å.45 Because of the layered packing through dispersion forces
along the y axis, the (010) surface is the easiest to cleave. The
calculated surface energy was 0.01 eV·Å−2 within PBE-D2. The
density of states (DOS) of the pristine material shows a small
band gap of 1.99 eV with the valence band formed by O(2p)
and the conduction band by Mo(4d). The Bader charge
analysis of the Mo centers provides a value of 2.63 |e−|, which is
well below the formal charge.
Because reduced α-MoO3 requires the use of PBE+U,40,44
the Ueff value was optimized with two criteria: an electronic
term given by electron localization and a thermodynamic one
based on reaction energies; see Section S1. Since good

Figure 2. (a) VOt vacancy formation energy according to the reaction
MoO3 → MoO3−x + 1/2 O2(g) calculated on the (2 × 2) and (3 × 3)
supercells at PBE+U and HSE06 levels. (b) VOt vacancy formation
energy as a function of oxygen depletion for 2×Mo5+ and Mo4+ states
at PBE+U, Ueff = 3.5 eV.

found that the calculations within PBE result in similar
geometries (the remaining oxygen stays at an intermediate
position between that of Ot and Ob) and similar formation
energies.31,32,40 However, in our PBE+U calculations, the
converged VOt geometry keeps the Ob in the surface plane,
whereas all of our attempts to converge a VOb resulted in a
displacement of the Ot to the Ob position and the convergence
in the same Ot vacancy geometry. Therefore, no vacancies
appear at Ob positions on the (010) surface. In turn, the
formation of Ot vacancies is endothermic, between 2.61 and
2.87 eV, whereas that of O3c requires an additional 1.80 eV; see
Tables S1 and S2. The large difference in energy indicates that
vacancies are preferentially formed at Ot positions (VOt).
VOt can be present in two different electronic configurations.37 The ground state has a vacancy formation energy of
2.61 eV (3 × 3 supercell). In the local geometry of this
configuration the undercoordinated Mo (Mocus) cation shortens the long Mo−Ob and the Mo−O3c sublayer bonds by 0.308
and 0.417 Å, whereas its neighboring Mo (still six-fold
coordinated) increases Mo−Ob distances by 0.244 and 0.283
Å; see Figure 3. The charge of these two centers differs from
that of the pristine surface, 2.48 and 2.49 |e−| values are
retrieved. The DOS in Figure 3 shows how one electron is
located in a dorbital of each reduced Mo, and the d-orbital of
the undercoordinated Mo is lower in energy. The electron
magnetization is slightly higher in the undercoordinated center:
0.96 versus 0.83 μB. Therefore, this configuration corresponds
to Mo5+(□)−O−Mo5+(O), so both Mo, where the vacancy
is created (□), and the neighboring cation, have a 5+ character.
A second configuration with a vacancy formation energy of
2.84 eV has also been obtained. In it, the Mocus−O bond
distances with the oxygens of the surface layer are similar to
those of the regular surface, but the Mo atom sinks and
reinforces its coordination to O3c, and the distance is shortened
by 0.309 Å; see Figure 3. The metal charge is 2.24 |e−|. The
DOS clearly shows the localization of the two electrons, with
2569

DOI: 10.1021/acs.jpclett.8b00536
J. Phys. Chem. Lett. 2018, 9, 2568−2573

The Journal of Physical Chemistry Letters

Letter

Figure 3. (a) Projected density of states, (b) local geometry with Mo−O bond distances, in angstroms, and (c) fingerprints for the clean surface and
the two different states corresponding to the VOt. Magnetization, μ, in μB; Bader charges, q, in |e−|; XPS displacement in eV; and frequencies, υ in
cm−1.

the same spin, in two different d orbitals of the Mo4+ center,
according to the calculated magnetization of 1.70 μB. Therefore,
this structure is assigned to the formation of a single doublereduced Mo4+(□) center.
The formation of two different electronic states can be
understood as follows. The ground-state 2×Mo5+ can
disproportionate to Mo6+ and Mo4+ centers because the
relative energies to increase the oxidation state of Mo5+
(toward Mo6+) and reduce it (toward Mo4+) cancel each
other out. The higher electrostatic interaction between the Mo
centers and the surrounding oxygens in the 2×Mo5+ is
responsible for its larger stability; see Section S3.
Benchmarks performed with the hybrid functional at this
concentration (1/9) and the PBE+U geometries lead to a
slightly higher vacancy formation energy range: 2.82 and 3.18
eV. The reoptimized PBE+U (Ueff = 6.3 eV) energies are much
lower, 1.72 and 1.88 eV, but even at these geometries the
HSE06 values are 2.93 and 3.30 eV, thus highlighting the
critical choice of the Ueff when reproducing the formation
energies. In summary, the ground state can be identified as a
bipolaron, where two electrons are localized on two
neighboring Mo centers (Mo5+), whereas the metastable state
corresponds to a single dipolaron at the same center (see
Figure 4a) and the energy difference is well-represented by PBE
+U, Ueff = 3.5 eV.
We have inspected how concentration affects the VOt vacancy
formation energies (Figure 2b). 2×Mo5+-like defects are
increasingly more costly (with a linear dependence on
concentration), whereas Mo4+ ones are independent of the
concentration until coverage reaches 0.5. Moreover, vacancies
cannot be accumulated through the [100] direction for both
configurations; see Table S1. The origin of this vacancy
alignment is at the core of the relaxations observed when the
vacancy is formed. When two neighboring vacancies are aligned
along [100], the shared O3c sinks toward the subsurface Mo
atom (reduces its bond length by 0.1 Å). This ends up

Figure 4. (a) Energy profile for the electron hopping for the
conversion 2×Mo5+ → Mo4+ on the (2 × 2) and (3 × 3) supercells
and calculated spin density for the two surface states at 0.01e−Å−3. (b)
Surface Mo local magnetization as a function of time for the molecular
dynamics run at 300 °C and θVOt = 0.25 ML (atom labels shown in the
scheme). (c) Calculated XPS at different vacancy coverages, θVOt, at
350 °C. Green Mo6+, red Mo5+, and blue Mo4+.

increasing the vacancy formation energy. Vacancies would then
accumulate on the perpendicular direction, generating crystal
shears that induce the formation of Magnéli phases.
Next, we analyzed the interconversion between the two
vacancy configurations. α-MoO3 is known to generate small
polarons upon electron addition,36,39 which are highly mobile,
with reported barriers of 0.35 eV in the plane and 0.50 eV
between bilayers.36,39 The CI-NEB results in a barrier for the
2570

DOI: 10.1021/acs.jpclett.8b00536
J. Phys. Chem. Lett. 2018, 9, 2568−2573

The Journal of Physical Chemistry Letters

Letter

conversion between oxidation states 2×Mo ↔ Mo of ∼0.3
eV, higher within HSE06, 0.40−0.60 eV, but still easy to
surmount at reaction temperatures. The barrier is low enough
to ensure that the two states are in equilibrium even at room
temperature. Therefore, we performed first-principles molecular dynamics (MD) simulations to illustrate the dynamic
nature of the electronic states in VOt. The local magnetization
of the surface centers is shown in Figure 4b along the 10 ps
trajectory at 300 °C. The main features are that the Mocus is
preferentially Mo5+ but also samples Mo4+ states and the
second polaron is localized either preferentially along the [001]
direction or in zigzag but not in the nearest neighbor along the
[100] direction. At the temperatures at which α-MoO3catalyzed reactions take place (∼300 °C), the relative
Boltzmann population at low reduction levels (θVOt = 0.25
ML) is ∼93% of Mo5+ and ∼7% of the Mo4+ configuration;
from our MD trajectory, these values are 95 and 5%. This is
remarkable because experiments have shown that materials
reduced at 350 °C and quantified by the XPS fingerprints
provide 93:7 Mo5+:Mo4+.21
The experimental fingerprints of oxygen-deficient α-MoO3
can then be reanalyzed considering the dual nature of the
vacancies. For instance, the Raman fingerprint of α-MoO3 has
three characteristic bands: MoOt, Mo−Ob and Mo−O3c
stretchings at 996, 823, and 667 cm−1, respectively.46,47 We
have calculated the vibrational modes of the different
coordination spheres of Mo centers. The reference Mo=Ot is
974 cm−1, which increases by 15 cm−1 for the hexacoordinated
Mo5+ in the Mo5+(□)−O−Mo5+(O), in agreement with the
bands over 1000 cm−1 identified in the experiments.25 In
addition, the stretching frequency for Mo−Ob bond is reduced
by 35−40 cm−1 for undercoordinated cations, and, more
importantly, by 100 cm−1 for the octahedral Mo5+ and therefore
could be a fingerprint for this Mo5+(O) center. The
frequency for the Mo−O3c stretching increases for the
undercoordinated cations: 10 and 36 cm−1 for Mo4+ and
Mo5+, respectively. Thus this mode can be used to discriminate
different states. Complementary, experimental XPS21,22 on
reduced MoO3 shows the peaks for Mo6+, Mo5+, and Mo4+
surface centers. For Mo5+ the peaks are displaced around −1.80
eV and for Mo4+ −3.00 eV. We calculated the theoretical XPS
displacement and obtained values of −0.94 and −1.29 eV for
Mo5+ and Mo4+, respectively. As shown in Figure 4c, increasing
vacancy concentrations would have a dramatic effect on the
shape of the XPS spectra.
The ability of the Mo centers to reach different oxidation
states can have implications in the chemistry of the oxide. To
analyze this we have calculated the adsorption for the two key
species in the Formox (methanol to formaldehyde) process.30
While methanol adsorbs through the hydroxyl to the Mocus by
−1.55 eV in both Mo4+ and 2×Mo5+; formaldehyde bonds to
both Mocus and an Ob by −1.60 and −1.83 eV for the Mo5+ and
the Mo4+ centers, respectively (see Figure S4). The single-site
adsorption by lone-pair donation of methanol is rather
independent of the oxidation state of the metal atom. In
contrast, the covalent bifunctional coordination of formaldehyde perturbs the structure (for 2×Mo5+ the Mocus−Ob−
Mo angle changes from 160 to 155° upon adsorption but does
not affect Mo4+), and thus the adsorption energies for the
electronic configurations are different.
In summary, a reaction-thermodynamics-adapted PBE+U
scheme has been employed to describe the nature of vacancies
in α-MoO3(010). The removal of terminal oxygens generates
5+

4+

two different configurations. The ground-state configuration is
constituted by two Mo5+ polarons that is separated from the
metastable configuration with a single Mo4+ by 0.2 eV. The
interconversion is possible through polaron hopping, and at
relevant catalytic temperatures the relative populations are 93:7.
Therefore, the complex spectroscopic patterns observed for this
material might be due to the different population of ground and
metastable states that correspond to a unique type of atom
deficiency. This might have implications in the electronic and
transport properties, Magnéli phase formation, and chemical
properties because the adsorption from the different states can
be different. This phenomenon will appear for oxides in which
the cations can adopt multiple low oxidation states.

■

COMPUTATIONAL METHODS
The Vienna Ab Initio Simulation Package (VASP), version
5.3.3, was used to perform the DFT calculations on slabs
models.48,49 PBE50 with Grimme’s semiempiric D251 and U
approximation34 (PBE+U-D2) and BEEF-vdW52 that contains
vdW-DF253 nonlocal correlation energy were used in the
benchmark. In addition, single points with the hybrid HSE06
functional were performed at PBE+U geometries with Ueff = 0,
3.5, and 7.0 eV.54,55 The core electrons were described by the
projector-augmented wave (PAW) pseudopotentials56 (with 14
valence electrons for Mo atoms), and valence ones were
expanded in plane waves with a kinetic cutoff energy of 450 eV.
Spin polarization was taken into account in all cases.
The bulk was optimized for the PBE functional with a 11 × 3
× 11 k-point mesh and a cutoff energy of 600 eV and employed
in all calculations. The (010) surface containing two bilayers
was cleaved, and benchmark calculations were performed on a
(2 × 2) supercell with a 5 × 5 × 1 k-point sampling. Vacancy
formation was studied on a (2 × 2), (3 × 3), (2 × 3), and (3 ×
2) supercell with a 3 × 3 × 1, 5 × 5 × 1, 5 × 3 × 1, and 3 × 5 ×
1 k-point sampling, respectively. HSE06 single-point calculations were performed with 3 × 3 × 1 k-point mesh for the (2
× 2) supercell and Γ-point sampling for (3 × 3) supercell.
During the optimizations, the upper bilayer and adsorbates
were relaxed, except where stated otherwise. A vacuum gap of
15 Å was added to prevent the interaction between slabs
together with the dipole correction.57 Complementary, the firstprinciples MD run was performed in the Born−Oppenheimer
approximation on a (2 × 2) supercell with one vacancy defect
at 300 °C during 10 ps with 2 ps of equilibration. The most
relevant structures have been added to the ioChem-BD
database58 and can be consulted in the following link.59

■

ASSOCIATED CONTENT

* Supporting Information
S

The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.jpclett.8b00536.
Ueff fitting, detailed vacancy formation energies, electrostatic model, adsorption study (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: nlopez@iciq.es.
ORCID

Núria López: 0000-0001-9150-5941
Notes

The authors declare no competing financial interest.
2571

DOI: 10.1021/acs.jpclett.8b00536
J. Phys. Chem. Lett. 2018, 9, 2568−2573

The Journal of Physical Chemistry Letters

Letter

■

ACKNOWLEDGMENTS
́
We thank the Spanish Ministerio de Economia y Competitividad (MINECO) for financial support (CTQ 2015-68770-R,
Severo Ochoa Excellence Accreditation 2014−2018 SEV-20130319, and Severo Ochoa predoctoral grant SVP-2014-068237).
In addition, we thank BSC-RES for providing computational
resources. We thank Drs. Març Capdevila-Cortada and Hanne
al
Falsig for valuable suggestions.

■

(18) Kröer, M.; Hamwi, S.; Meyer, J.; Riedl, T.; Kowalsky, W.;
g
Kahn, A. P-type Doping of Organic Wide Band Gap Materials by
Transition Metal Oxides: A Case-study on Molybdenum Trioxide.
Org. Org. Electron. 2009, 10, 932−938.
(19) Sian, T. S.; Reddy, G. B. Optical, Structural and Photoelectron
Spectroscopic Studies on Amorphous and Crystalline Molybdenum
Oxide Thin Films. Sol. Energy Mater. Sol. Cells 2004, 82, 375−386.
(20) Greiner, M. T.; Chai, L.; Helander, M. G.; Tang, W.-M.; Lu, Z.H. Transition Metal Oxide Work Functions: The Influence of Cation
Oxidation State and Oxygen Vacancies. Adv. Funct. Mater. 2012, 22,
4557−4568.
(21) Zacharopoulou, V.; Vasiliadou, E. S.; Lemonidou, A. A.
Exploring the Reaction Pathways of Bioglycerol Hydrodeoxygenation
to Propene over Molybdena-based Catalysts. ChemSusChem 2018, 11,
264−275.
(22) Shetty, M.; Murugappan, K.; Prasomsri, T.; Green, W. H.;
Román-Leshkov, Y. Reactivity and Stability Investigation of Supported
Molybdenum Oxide Catalysts for The Hydrodeoxygenation (HDO)
of m-Cresol. J. Catal. 2015, 331, 86−97.
(23) Labanowska, M. Paramagnetic Defects in MoO3−Revisited.
Phys. Chem. Chem. Phys. 1999, 1, 5385−5392.
(24) Py, M. A.; Maschke, K. Intra- and Interlayer Contributions to
the Lattice Vibrations in MoO3. Physica B+C 1981, 105, 370−374.
(25) Mestl, G.; Verbruggen, N. F. D.; Bosch, E.; Knöinger, H.
z
Mechanically Activated MoO3. 5. Redox Behavior. Langmuir 1996, 12,
2961−2968.
(26) Vasilopoulou, M.; Douvas, A. M.; Georgiadou, D. G.; Palilis, L.
C.; Kennou, S.; Sygellou, L.; Soultati, A.; Kostis, I.; Papadimitropoulos,
G.; Davazoglou, D.; Argitis, P. The Influence of Hydrogenation and
Oxygen Vacancies on Molybdenum Oxides Work Function and Gap
States for Application in Organic Optoelectronics. J. Am. Chem. Soc.
2012, 134, 16178−16187.
(27) Hanson, E. D.; Lajaunie, L.; Hao, S.; Myers, B. D.; Shi, F.;
Murthy, A. A.; Wolverton, C.; Arenal, R.; Dravid, V. P. Systematic
Study of Oxygen Vacancy Tunable Transport Properties of Few-Layer
MoO3−x Enabled by Vapor-Based Synthesis. Adv. Funct. Mater. 2017 ,
27, 1605380.
(28) Scanlon, D. O.; Watson, G. W.; Payne, D. J.; Atkinson, G. R.;
Egdell, R. G.; Law, D. S. L. Theoretical and Experimental Study of the
Electronic Structures of MoO3 and MoO2. J. Phys. Chem. C 2010, 114,
4636−4645.
(29) Akande, S. O.; Chroneos, A.; Vasilopoulou, M.; Kennou, S.;
Schwingenschlöl, U. Vacancy Formation in MoO3: Hybrid Density
g
Functional Theory and Photoemission Experiments. J. Mater. Chem. C
2016, 4, 9526−9531.
(30) Rellán-Piñeiro, M.; López, N. The Active Molybdenum Oxide
Phase in the Methanol Oxidation to Formaldehyde (Formox Process):
A DFT Study. ChemSusChem 2015, 8, 2231−2239.
(31) Mei, D.; Karim, A. M.; Wang, Y. Density Functional Theory
Study of Acetaldehyde Hydrodeoxygenation on MoO3. J. Phys. Chem.
C 2011, 115, 8155−8164.
(32) Choksi, T.; Greeley, J. Partial Oxidation of Methanol on
MoO3(010): A DFT and Microkinetic Study. ACS Catal. 2016, 6,
7260−7277.
(33) Shetty, M.; Buesser, B.; Román-Leshkov, Y.; Green, W. H.
Computational Investigation on Hydrodeoxygenation (HDO) of
Acetone to Propylene on α-MoO3(010) Surface. J. Phys. Chem. C
2017 , 121, 17848−17855.
(34) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.;
Sutton, A. P. Electron-energy-loss Spectra and the Structural Stability
of Nickel Oxide: An LSDA+U study. Phys. Rev. B: Condens. Matter
Mater. Phys. 1998, 57, 1505−1509.
(35) Inzani, K.; Grande, T.; Vullum-Bruer, F.; Selbach, S. M. A van
der Waals Density Functional Study of MoO3 and Its Oxygen
Vacancies. J. Phys. Chem. C 2016, 120, 8959−8968.
(36) Tahini, H. A.; Tan, X.; Lou, S. N.; Scott, J.; Amal, R.; Ng, Y. H.;
Smith, S. C. Mobile Polaronic States in α-MoO3: An ab Initio
Investigation of the Role of Oxygen Vacancies and Alkali Ions. ACS
Appl. Mater. Interfaces 2016, 8, 10911−10917.

REFERENCES

(1) Pacchioni, G. Oxygen Vacancy: The Invisible Agent on Oxide
Surfaces. ChemPhysChem 2003, 4, 1041−1047.
(2) Henrich, V. E.; Cox, P. A. The Surface Science of Metal Oxides;
Cambridge University Press: Cambridge, U.K., 1996.
(3) Carrasco, J.; López, N.; Illas, F. First Principles Analysis of the
Stability and Diffusion of Oxygen Vacancies in Metal Oxides. Phys. Rev.
Lett. 2004, 93, 225502.
(4) Kowalski, P. M.; Camellone, M. F.; Nair, N. N.; Meyer, B.; Marx,
D. Charge Localization Dynamics Induced by Oxygen Vacancies on
the TiO2(110) Surface. Phys. Rev. Lett. 2010, 105, 146405.
(5) Ganduglia-Pirovano, M. V.; Da Silva, J. L. F.; Sauer, J. DensityFunctional Calculations of the Structure of Near-Surface Oxygen
Vacancies and Electron Localization on CeO2(111). Phys. Rev. Lett.
2009, 102, 026101.
(6) Walsh, A.; Da Silva, J. L. F.; Wei, S.-H.; Kö
rber, C.; Klein, A.;
Piper, L. F. J.; DeMasi, A.; Smith, K. E.; Panaccione, G.; Torelli, P.;
Payne, D. J.; Bourlange, A.; Egdell, R. G. Nature of the Band Gap
In2O3 Revealed by First-Principles Calculations and X-Ray Spectroscopy. Phys. Rev. Lett. 2008, 100, 167402.
(7) Albani, D.; Capdevila-Cortada, M.; Vilé, G.; Mitchell, S.; Martin,
O.; López, N.; Pérez-Ramírez, J. Semihydrogenation of Acetylene on
Indium Oxide: Proposed Single-Ensemble Catalysis. Angew. Chem.
2017 , 129, 10895−10900.
(8) de Castro, I. A.; Datta, R. S.; Ou, J. Z.; Castellanos-Gomez, A.;
Sriram, S.; Daeneke, T.; Kalantar-zadeh, K. Molybdenum Oxides −
From Fundamentals to Functionality. Adv. Mater. 2017 , 29 (40),
1701619.
(9) Greenwood, N. N.; Earnshaw, A. 23-Chromium, Molybdenum
and Tungsten. In Chemistry of the Elements, 2nd ed.; ButterworthHeinemann: Oxford, U.K., 1997; pp 1002−1039.
(10) Goodenough, J. B. Proc. Climax 4th Intern. Conf. Chemistry and
Uses of Molybdenum; Climax Molybdenum Co.: London, 1982.
(11) Lee, S.-H.; Kim, Y.-H.; Deshpande, R.; Parilla, P. A.; Whitney,
E.; Gillaspie, D. T.; Jones, K. M.; Mahan, A. H.; Zhang, S.; Dillon, A.
C. Reversible Lithium-Ion Insertion in Molybdenum Oxide Nanoparticles. Adv. Mater. 2008, 20, 3627−3632.
(12) Hu, X.; Zhang, W.; Liu, X.; Mei, Y.; Huang, Y. Nanostructured
Mo-based electrode materials for electrochemical energy storage.
Chem. Soc. Rev. 2015, 44, 2376−2404.
(13) Cheng, H.; Wen, M.; Ma, X.; Kuwahara, Y.; Mori, K.; Dai, Y.;
Huang, B.; Yamashita, H. Hydrogen Doped Metal Oxide Semiconductors with Exceptional and Tunable Localized Surface Plasmon
Resonances. J. Am. Chem. Soc. 2016, 138, 9316−9324.
(14) Xie, W.; Su, M.; Zheng, Z.; Wang, Y.; Gong, L.; Xie, F.; Zhang,
W.; Luo, Z.; Luo, J.; Liu, P.; Xu, N.; Deng, S.; Chen, H.; Chen, J.
Nanoscale Insights into the Hydrogenation Process of Layered αMoO3. ACS Nano 2016, 10, 1662−1670.
(15) Borgschulte, A.; Sambalova, O.; Delmelle, R.; Jenatsch, S.; Hany,
R.; Nüsch, F. Hydrogen Reduction of Molybdenum Oxide at Room
e
Temperature. Sci. Rep. 2017 , 7, 40761.
(16) Magneli, A.; Blomberg-Hansson, B.; Kihlborg, L.; Sundkvist, G.
Studies on Molybdenum and Molybdenum Wolfram Oxides of the
Homologous Series MenO3n‑1. Acta Chem. Scand. 1955, 9, 1382−1390.
(17) Bouzidi, A.; Benramdane, N.; Tabet-Derraz, H.; Mathieu, C.;
Khelifa, B.; Desfeux, R. Effect of Substrate Temperature on the
Structural and Optical Properties of MoO3 Thin Films Prepared by
Spray Pyrolysis Technique. Mater. Sci. Eng., B 2003, 97, 5−8.
2572

DOI: 10.1021/acs.jpclett.8b00536
J. Phys. Chem. Lett. 2018, 9, 2568−2573

The Journal of Physical Chemistry Letters

Letter

(37) Kim, H. S.; Cook, J. B.; Lin, H.; Ko, J. S.; Tolbert, S. H.;
Ozolins, V.; Dunn, B. Oxygen Vacancies Enhance Pseudocapacitive
Charge Storage Properties of MoO3−x. Nat. Mater. 2017 , 16, 454.
(38) Lei, Y.-H.; Chen, Z.-X. DFT+U Study of Properties of MoO3
and Hydrogen Adsorption on MoO3(010). J. Phys. Chem. C 2012, 116,
25757−25764.
(39) Ding, H.; Lin, H.; Sadigh, B.; Zhou, F.; Ozoliņ̌V.; Asta, M.
s,
Computational Investigation of Electron Small Polarons in α-MoO3. J.
Phys. Chem. C 2014, 118, 15565−15572.
(40) Coquet, R.; Willock, D. J. The (010) surface of α-MoO3, a DFT
+ U study. Phys. Chem. Chem. Phys. 2005, 7, 3819−3828.
(41) Agarwal, V.; Metiu, H. Energy of Oxygen-Vacancy Formation
on Oxide Surfaces: Role of the Spatial Distribution. J. Phys. Chem. C
2016, 120, 2320−2323.
(42) Agarwal, V.; Metiu, H. Oxygen Vacancy Formation on α-MoO3
Slabs and Ribbons. J. Phys. Chem. C 2016, 120, 19252−19264.
(43) Lutfalla, S.; Shapovalov, V.; Bell, A. T. Calibration of the DFT/
GGA+U Method for Determination of Reduction Energies for
Transition and Rare Earth Metal Oxides of Ti, V, Mo, and Ce. J.
Chem. Theory Comput. 2011, 7, 2218−2223.
(44) Getsoian, A. B.; Bell, A. T. The Influence of Functionals on
Density Functional Theory Calculations of the Properties of Reducible
Transition Metal Oxide Catalysts. J. Phys. Chem. C 2013, 117, 25562−
25578.
(45) Kihlborg, L. Least-squares Refinement of the Crystal Structure
of Mo Trioxide. Ark. Kemi 1963, 21, 357−64.
(46) Camacho-López, M. A.; Escobar-Alarcón, L.; Picquart, M.;
Arroyo, R.; Córdoba, G.; Haro-Poniatowski, E. Micro-Raman Study of
the m-MoO2 to α-MoO3 Transformation Induced by cw-laser
Irradiation. Opt. Mater. 2011, 33, 480−484.
(47) Dieterle, M.; Weinberg, G.; Mestl, G. Raman spectroscopy of
molybdenum oxides Part I. Structural Characterization of Oxygen
Defects in MoO3‑x by DR UV/VIS, Raman Spectroscopy and X-ray
Diffraction. Phys. Chem. Chem. Phys. 2002, 4, 812−821.
(48) Kresse, G.; Furthmü J. Efficient Iterative Schemes for ab
ller,
initio Total-energy Calculations Using a Plane-wave Basis Set. Phys.
Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169−11186.
(49) Kresse, G.; Furthmü J. Efficiency of ab-initio Total Energy
ller,
Calculations for Metals and Semiconductors Using a Plane-wave Basis
Set. Comput. Mater. Sci. 1996, 6, 15−50.
(50) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient
Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868.
(51) Grimme, S. Semiempirical GGA-type Density Functional
Constructed with a Long-range Dispersion Correction. J. Comput.
Chem. 2006, 27, 1787−1799.
(52) Wellendorff, J.; Lundgaard, K. T.; Møgelhøj, A.; Petzold, V.;
Landis, D. D.; Nørskov, J. K.; Bligaard, T.; Jacobsen, K. W. Density
Functionals for Surface Science: Exchange-correlation Model Development with Bayesian Error Estimation. Phys. Rev. B: Condens. Matter
Mater. Phys. 2012, 85, 235149.
(53) Lee, K.; Murray, É. D.; Kong, L.; Lundqvist, B. I.; Langreth, D.
C. Higher-accuracy van der Waals Density Functional. Phys. Rev. B:
Condens. Matter Mater. Phys. 2010, 82, 081101.
(54) Heyd, J.; Scuseria, G. E.; Ernzerhof, M. Hybrid Functionals
Based on a Screened Coulomb Potential. J. Chem. Phys. 2003, 118,
8207−8215.
(55) Paier, J.; Marsman, M.; Hummer, K.; Kresse, G.; Gerber, I. C.;
Á ngyán, J. G. Erratum: “Screened Hybrid Density Functionals Applied
to Solids” [J. Chem. Phys. 124, 154709 (2006)]. J. Chem. Phys. 2006,
125, 249901.
(56) Kresse, G.; Joubert, D. From Ultrasoft Pseudopotentials to the
Projector Augmented-wave Method. Phys. Rev. B: Condens. Matter
Mater. Phys. 1999, 59, 1758−1775.
(57) Makov, G.; Payne, M. C. Periodic Boundary Conditions in ab
initio Calculations. Phys. Rev. B: Condens. Matter Mater. Phys. 1995, 51,
4014−4022.
(58) Á lvarez-Moreno, M.; de Graaf, C.; López, N.; Maseras, F.;
Poblet, J. M.; Bo, C. Managing the Computational Chemistry Big Data

Problem: The ioChem-BD Platform. J. Chem. Inf. Model. 2015, 55 (1),
95−103.
(59) Rellán-Piñeiro, M.; López, N. Data Set for One Oxygen Vacancy,
Two Charge States: Characterization of Reduced α-MoO3(010) through
Theoretical Methods. DOI: 10.19061/iochem-bd-1-59.

2573

DOI: 10.1021/acs.jpclett.8b00536
J. Phys. Chem. Lett. 2018, 9, 2568−2573