Article

Cite This: J. Phys. Chem. C 2017, 121, 24809-24815 pubs.acs.org/JPCC

Assessing Oxygen Vacancies in Bismuth Oxide through EELS

Measurements and DFT Simulations
Pau Torruella,*,†,‡ Catalina Coll,†,‡ Gemma Martín,†,‡ Lluís López-Conesa,†,‡,§ María Vila,∥
Carlos Díaz-Guerra,∥ María Varela,∥,⊥ María Luisa Ruiz-González,# Javier Piqueras,∥ Francesca Peiró,†,‡
and Sònia Estradé†,‡
†
LENS-MIND, Departament d’Enginyeries: Electrònica, ‡Institute of Nanoscience and Nanotechnology (IN2UB), and §TEM-MAT,
CCiT, Universitat de Barcelona, Barcelona 08028, Spain
∥
Departamento de Física de Materiales, Facultad de Ciencias Físicas, ⊥Instituto de Magnetismo Aplicado, Facultad de Ciencias Físicas,
and #Departamento de Química Inorgánica, Facultad de Químicas, Universidad Complutense de Madrid, Madrid 28040, Spain
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

*
S Supporting Information
Downloaded via UNIV OF PESHAWAR on June 29, 2022 at 14:46:05 (UTC).

ABSTRACT: Pioneering electron energy loss spectroscopy

(EELS) measurements of α-Bi2O3 are performed on three
samples obtained through different synthesis methods.
Experimental low-loss and core-loss EELS spectra are acquired.
By combining them with detailed structural characterization
and Density Functional Theory (DFT) simulations, we are
able to detect and evaluate the presence of oxygen vacancies in
the samples. This type of information has not been accessed
previously from EELS data in bismuth oxide, because high-
resolution EELS spectra or how vacancies reflect in Bi2O3
spectra were unreported. This novel measurement is further
validated through comparison with photoluminescence data.
Therefore, the technique has the ability to probe oxygen
vacancies in Bi2O3 at an unprecedented resolution, which might allow solving material science and technological issues related to
this material.

■ INTRODUCTION
Bismuth oxide is a polymorph with four main crystallographic
phases: the monoclinic α-phase, which is stable at low
temperature; the tetragonal β and face-centered cubic γ phases,
which are metastable; and the δ-phase, which is stable at high
temperatures.1,2 It is a widely used material in the state-of-the-
art of many technological fields such as gas sensors,3,4 optical
devices,5,6 and solid oxide fuel cells.7,8 However, the functional
properties of Bi2O3 are strongly dependent on its crystalline
phase. As an example, the conduction mechanism of α-Bi2O3 is
mainly electronic, whereas for δ-Bi2O3 it is mainly ionic.9,10
Moreover, with the advent of nanotechnology, a huge
miscellany of Bi2O3 nanostructures is being reported, from
nanocoatings to nanowires, nanotubes, or hollow nanoparticles,
each with its own applications and virtues.11−15 Therefore,
when tailoring these complex nanostructures, a way to assess
Bi2O3 phase, defects, and other structural properties with high
resolution is mandatory. In this context, the use of a high-
resolution transmission electron microscope (TEM) provides a
great tool to evaluate crystalline properties at the nanoscale.16
Additionally, when coupled with electron energy loss spectros-
copy (EELS), it can also probe chemical and electronic
properties of a given material at high resolution. In particular,
the band structure of a material is related to low-loss EELS
spectra, while information on the oxidation state, coordination,
or vacancy presence of a given element in the sample can be
obtained from core-loss spectra by analyzing its energy loss
near edge structure (ELNES) features.17 The latter properties
are of particular importance when dealing with bismuth oxide,
because its high ionic conductivity is discussed in terms of
oxygen vacancy ordering, generation, and transport.9,10,18,19
However, obtaining this information through the ELNES fine
structure is not straightforward and requires previous knowl-
edge on how microscopic properties of the given material alter
the spectra. This can only be obtained from reference samples
or simulations.20,21 Surprisingly, reported EELS measurements
of Bi2O3 only use low-loss spectra to determine band gap
energy,19 and none, to the best of our knowledge, presents any
spectrum from the core-loss region.
The present work aims to bridge this gap of knowledge and
provide Bi2O3 core-loss EELS data, while linking experimental
and simulation EELS results to microscopic properties of three
different Bi2O3 samples, bismuth oxide nanowires synthesized
by thermal evaporation,22 commercial Bi2O3 powder, and a
Received: June 27, 2017
Revised: September 29, 2017
Published: October 24, 2017

© 2017 American Chemical Society 24809 DOI: 10.1021/acs.jpcc.7b06310

J. Phys. Chem. C 2017, 121, 24809−24815
The Journal of Physical Chemistry C
Figure 1. Optical characterization. (A) Normalized Raman spectra representative of the three kinds of Bi2O3 samples investigated. (B) Normalized
PL spectra from the three Bi2O3 investigated samples.
Bi2O3 ceramic23 obtained in the form of a sintered pellet from
this powder. The crystalline structure of the samples has been
fully characterized through scanning-TEM (STEM) and Raman
spectroscopy. Careful EELS measurements have been per-
formed in low-loss and core-loss regimes. To understand the
EELS features from the obtained spectra, the measurements are
supported with density functional theory (DFT) simulations
that can calculate the band structure and the ELNES features of
a given crystalline structure.24 Finally, photoluminescence (PL)
measurements are performed, and the results are compared to
the conclusions obtained from the novel EELS measurements.
■
METHODS
Commercial Bi2O3 powders (99.999%) were purchased from
Sigma-Aldrich. To obtain the Bi2O3 ceramics, these powders
were compacted under compressive load, forming disk-shaped
samples of about 1 mm thickness and 7 mm diameter. These
samples were then annealed in air at 750 °C for 10 h.23 Bi2O3
nanowires were grown mixing 80 wt % Bi powder (Goodfellow,
99.997%) and 20 wt % Er2O3 powder (STREM, 99.9%). This
mixture was compressed to form disk-shaped pellets, which
were then annealed at 800 °C for 4 h in a horizontal tube
furnace under Ar flow. This treatment led to the growth of a
high density of α-Bi2O3 nanowires.22 It must be pointed out
that Er was not incorporated into the nanowires.25 Its role is to
alter the oxidation kinetics, favoring the formation of a high
density of nanostructures.
Micro-Raman and micro-PL measurements were carried out
at room temperature in a Horiba Jovin-Ybon LabRAM HR800
system. To collect Raman spectra, the samples were excited by
a 633 nm He−Ne laser on an Olympus BX 41 confocal
microscope with a 100× objective. A 325 nm He−Cd laser and
a 40× objective were used for PL measurements.
Scanning TEM (STEM) high angle annular dark field
(HAADF) and EELS measurements from the samples were
acquired in a JEOL ARM200cF TEM operated at 200 kV and
equipped with a cold field emission fun (c-FEG) and a GATAN
Quantum GIF EEL spectrometer. A Gatan double tilt cryo-
holder (model 636) was used for the ceramic sample to
minimize electron beam damage, with a working temperature of
97 K. The collection and convergence semiangles were 27.78
and 35 mrad, respectively, throughout all of the EELS
measurements. These measurements were performed using
the dual-EELS acquisition mode, which allows acquiring both
24810
Article
low-loss and core-loss EELS spectra simultaneously. The latter
were focused in the 520−560 eV energy loss range, were the O
K edge can be observed. The dual acquisition enables aligning
core-loss spectra with the zero-loss peak, therefore making the
absolute energy-loss position of different features of the peaks
reliable. Single spectra for each sample were obtained by
summing larger spectrum images after zero-loss peak alignment.
These spectrum images were acquired with atomic resolution,
and the corresponding coacquired HAADF images (Figure S1)
allow identification of the zone axis of the crystal during the
acquisition as well as confirm the lack of beam damage. They
were also acquired so that there were no carbon contributions
either from the TEM grid or from contamination.
Theoretical calculations of the electronic structure of α-Bi2O3
were carried out using the linearized augmented plane wave
(LAPW) method based on DFT, implemented on the
WIEN2k26,27 ab initio simulation package. The DFT calculation
was performed considering the local density approximation28,29
(LDA). The atomistic input model was built according to the
crystallographic information on the unit cell (UC) from α-
Bi2O3 structure as reported in ref 30, which has a P21c space
group (5.8486, 8.1661, and 7.5097 Å lattice parameters and
beta angle of 113°, giving five nonequivalent atomic positions).
The cutoff energy of the simulation was set at −11.9 Ry,
defining the following valence and semicore electrons: 2s2 and
2p4 for oxygen and 4f14, 5d10, 6p3, and 6s2 for bismuth. DFT
self-consistency cycles were performed until the total energy
was converged to 10−5 eV and the residual forces on atoms
were below 0.01 eV/Å with 1000 k-points. The plane-wave
basis set cutoff was set as RKmax = 7.
■
RESULTS
Normalized Raman spectra representative of the Bi 2O 3
powders, ceramics, and nanowires are shown in Figure 1A.
No significant differences, in terms of peak positions and
bandwidths, were found in Raman spectra of the different
samples. Raman bands appear peaked at 84, 119, 139, 152, 184,
210, 280, 313, 411, 448, and 526 cm−1, which are all
characteristic of the α-Bi2O3 phase.31−33
PL measurements, Figure 1B, show three narrow, well-
resolved emission bands, peaked at 1.86, 1.95, and 2.05 eV.
Besides those, two weaker and broader emissions, respectively
centered near 2.10 and 2.27 eV, can be also appreciated. The
PL intensity varies strongly among the specimens. The Bi2O3
DOI: 10.1021/acs.jpcc.7b06310
J. Phys. Chem. C 2017, 121, 24809−24815
The Journal of Physical Chemistry C Article

Figure 2. STEM-HAADF imaging. (A) Low magnification STEM-HAADF image of a group of Bi2O3 nanocrystals from the reference powder. (B)
Atomic resolution STEM-HAADF image of a Bi2O3 nanocrystal seen along the [001]α-Bi2O3 zone axis. (C) Low magnification STEM-HAADF
image of the FIB-prepared Bi2O3 pellet. The top bright layer seen in the image is due to the preparation of the sample. (D) Atomic resolution
STEM-HAADF image from the pellet sample seen along the [001] zone axis of the α-phase. (E) Low magnification TEM image of the nanowire
sample. (F) STEM-HAADF image of an α-phase nanowire along the [010] zone axis of the α-phase.

ceramic shows the maximum PL intensity, which halves for the
nanowires, and is 3 times weaker in the powder. In addition, the
relative intensity of the 2.10 and 2.27 eV bands varies for each
sample, with the ceramic showing the highest 2.27 eV relative
emission and the commercial powder the lowest. Moreover, the
peak positions of the above-mentioned three narrow emissions
observed centered between 1.86 and 2.05 eV in PL spectra of
powders and nanowires appear shifted 0.3 eV toward higher
energies in the ceramic. Additionally, PL spectra from the latter
sample show a weak and broad new emission at about 3.0 eV as
well as an intense band centered near 1.72 eV.
The commercial powder sample was prepared for TEM
observation by the standard dilution in hexane and drop
pouring onto a TEM holey carbon grid. The TEM structural
analysis showed that it was comprised of crystals with lateral
24811
sizes varying from 50 to 500 nm, depicted in Figure 2A, which
consisted of Bi2O3 in the α-phase, seen in the [001] zone axis in
Figure 2B. As for the pellet, the sample was prepared by the
standard FIB lift-out technique.34 TEM characterization
evidenced that it consisted of large (several micrometers) α-
phase Bi2O3 crystals (Figure 2C,D). The Bi2O3 nanowires were
prepared by solution and sonication in hexane, followed by
pouring a drop of the solution onto a TEM holey carbon grid.
As can be seen in the low-magnification TEM image (Figure
2E), they were found to form hierarchical structures and also
consisted mainly of bismuth oxide in the α-phase (Figure 2F).
Sparse, smaller nanowires of the metastable β-Bi2O3 nanowires
were also observed, but as Raman measurements show, they are
not statistically relevant.
DOI: 10.1021/acs.jpcc.7b06310
J. Phys. Chem. C 2017, 121, 24809−24815
The Journal of Physical Chemistry C
The EELS measurements at 200 kV for each sample are
shown in Figure 3. Clear differences between samples appear in
Figure 3. Experimental EELS. Core-loss and low-loss from the three
investigated samples. Note that each spectrum has been divided by the
intensity at its maximum. The main peaks of the spectra have been
marked and labeled: a1 = 532.5 eV, b1 = 534.6 eV, c1 = 540.2 eV, d1 =
545.7 eV, a2 = 11.0 eV, b2 = 18.7 eV, c2 = 29.5 eV.
Figure 4. DFT simulations of EEL spectra. (A) Simulated O K edge spectrum for α-Bi2O3 (plotted in black). The contributions from each
nonequivalent oxygen in the unit cell are also plotted. An inset with the α-Bi2O3 unit cell seen along the [010] zone axis shows the position of each
of these oxygens in blue, red, and green (the Bi atoms are shown in purple). The energy positions of the spectral features identified in the
experimental spectra are also shown as vertical lines. (B) Simulated core-loss and (C) low-loss spectra for α-Bi2O3 and for α-Bi2O3 with one oxygen
less for every two unit cells.
24812
Article
the spectra. To discuss these differences, the main features of
the spectra have been labeled. The core-loss spectra show a
main peak at 534.6 eV (b1), which is at the same position for all
samples. The onset of this peak shows a shoulder at 532.5 eV
(a1) that varies in height when comparing the commercial
powder sample with the other two, which were thermally
treated. The a1 intensity goes from being almost 25% lower
than the b1 peak in the commercial powder to being equal in
the nanowire and the sintered pellet. Moreover, the powder
shows some additional peaks in the continuum region of the
spectrum at 540.2 eV (c1) and 545.7 eV (d1).
To understand information from the EEL spectra, an initial
DFT simulation for the standard unit cell was carried out.
Importantly, the LDA approximation used in the calculations is
known to not reliably calculate absolute onset energies for
EELS edges.35 Because of this, a shift of 1.2 eV will be applied
to all simulations. This value has been chosen so that the b1
peak energy of the experimental spectrum of the commercial
powder matches the maximum of the simulated edge from the
unit cell, de facto taking the commercial powder as a reference
sample. The simulated O K edge from the unit cell is shown in
Figure 4A, where the energies of the experimental features
(a1,b1,c1,d1) are also labeled. An excellent match with the
experimental spectrum from the reference Bi2O3 powder can be
appreciated: the simulation shows all of the features of the
experimental spectra, with only small differences in the position
of the c1 peak (at a lower energy in the simulation) and the d1
peak (at a higher energy). Remarkably, this is the first reported
experimental spectrum of α-Bi2O3, with additional confirmation
from the simulation, and will be made available at the EELS
DataBase service.36 In contrast, the a1 shoulder is much more
intense in the pellet and nanowire spectra, as well as both
showing a steeper onset than the reference sample. The good
match between the powder sample and the UC simulation,
while small discrepancies show up for nanowire and pellet,
points to significant changes between these samples.
DOI: 10.1021/acs.jpcc.7b06310
J. Phys. Chem. C 2017, 121, 24809−24815
The Journal of Physical Chemistry C Article

An important fact to consider is that in the UC of α-Bi2O3

there are 6 oxygen atoms located at three different non-
equivalent crystallographic positions (two in each one), each
one seeing a different electronic environment in the structure.
These positions will be referred as OI for x = 0.2337a, y =
0.4533b, z = 0.1266c, and equivalent positions; OII for x =
0.265a, y = 0.0294b, and z = 0.0115c; and OIII for x = 0.7783a, y
= 0.3037b, and z = 0.2080c, where a, b, and c are the unit cell
parameters. The contributions of oxygen atoms in OI, OII, and
OIII can also be extracted from the simulated UC, and are
shown in Figure 4A. The total spectra can be evaluated as the
sum of these contributions weighted by their multiplicity (the
number of atoms in each position and equivalents, two for all in
this case). In Figure 4A, it is clear that these contributions are
different. In fact, the most notable difference is also the height
of the a1 shoulder, OII having the lowest, and the b1 peak
distance respect to the edge onset, OII having the highest. This
fact hints that the difference in the experimental spectra might
be due to an imbalance in the occupation of the three oxygen Figure 5. DFT simulations of oxygen vacancies. O K energy shift for
each of the DFT simulations.
sites. To further investigate this, simulations of unit cells with
oxygen vacancies were undertaken.

A few considerations must be taken into account when
removing an atom from a unit cell in a DFT simulation. Even
though a unit cell with one missing oxygen atom might seem a
rather small defect in a big crystal, the simulations are
performed with periodic boundary conditions, which means
actually an infinite crystal with one less oxygen in each unit cell
is being simulated. Because of this, if the vacancy proportion
wants to be kept low, a large “supercell” consisting of multiple
unit cells in which only one oxygen vacancy is being introduced
must be simulated. Three additional DFT simulations were
performed to evaluate the oxygen vacancy concentration impact
on EELS spectra: a 2 × 1 × 1 “supercell”, which will be referred
to as V simulation, with one oxygen atom missing, a 3 × 1 × 1
“supercell” with one vacancy, and a 3 × 1 × 1 “supercell” with
two vacancies. Considering that the unit cell of Bi2O3 has 12
oxygen atoms in total, vacancy concentrations are 1/24 =
4.17%, 1/36 = 2.78%, and 2/36 = 5.56%, respectively. The
crystallographic position in which the vacancy is formed must
be carefully considered. The vacancy formation energy for each
position was calculated yielding similar values for each one
within the reliability of the simulation. Because of this,
vacancies in the OII site were chosen, as their contribution
can account for the observed discrepancies between the
experimental spectra of the powder and the nanowire/pellet
sample. The same offset of 1.2 eV applied to the UC simulation
is applied to the vacancy simulations calculated spectra.
UC and V simulations were compared both in the core-loss
(Figure 4B) and in the low-loss regime (Figure 4C).
Remarkably, no major differences appear in the low-loss
spectra, apart from a small peak at 1.9 eV. Similarly, the
experimental low-loss spectra barely change between each
sample, and unfortunately at 1.9 eV the zero-loss peak overlaps
with the loss function of the samples too much to see any peak
of this kind in experimental data. Yet the most interesting
feature in this comparison is the appearance of a 1.3 eV shift to
lower energies of the O K edge in the V simulation. This seems
very reasonable if one bears in mind that this shift has also been
seen in EELS spectra of other transition metals, such as Fe, Co,
or Mn, when they are reduced.37−40 By comparing the O K
EELS edge position of all of the simulations (Figure 5), it is
clear that the energy shift is proportional to the vacancy
concentration.
24813
■
DISCUSSION
As far as the cationic structure is concerned, the samples show
no significant differences, all of them being mainly α-Bi2O3.
However, the experimental EEL spectra show clear differences,
and their origin must be elucidated.
At first, one could think that the origin of these small
differences could be found in the orientation dependence of
EEL spectra41 because the samples were not in the same zone
axis during the acquisition. To explore this possibility, UC
simulations for the relevant zone axis were carried out. As can
be seen in Figure S2, they show no significant differences in our
experimental conditions, so this factor can be ruled out.
Additionally, because of the known dependence on oxygen
vacancies of bismuth oxide properties, it is reasonable to
suspect that they might have a role in these measurements. To
perform a rigorous analysis of the core-loss spectra from the
nanowire and pellet sample in terms of oxygen vacancies, each
configuration of missing oxygen atoms for increasingly large
supercells should be simulated. This approach is not feasible in
terms of computational time. In this regard, it must be kept in
mind that bismuth is one of the heaviest stable elements on the
periodic table, with 83 electrons, 29 of which have been treated
rigorously within the DFT formalism (orbitals 4f to 6s as stated
in the Methods), and therefore costly to simulate. However, on
the basis of the knowledge that the previous simulations have
yielded, that vacancies shift the O K edge, and that b1, a1
features vary for different symmetry sites, the following simple
model can be inferred: Iexp(ΔE) = k1IOI(ΔE − s) + k2IOII(ΔE −
s) + k3IOIII(ΔE − s). Properly normalized, the ki parameters
represent occupancies of the different oxygen sites, and the
parameter s accounts for the energy shift that the V simulation
demonstrates. This model can now be used to fit the
experimental data and give some grounds for discussion on
the presence and nature of oxygen vacancies on the material.
Importantly, it must be kept in mind that the positions of the
simulated oxygen spectra have been calibrated by using the
commercial powder sample, and therefore any property
calculated by this method will be relative to the commercial
powder. However, this seems reasonable because it is the purest
form in which we can obtain this material.
DOI: 10.1021/acs.jpcc.7b06310
J. Phys. Chem. C 2017, 121, 24809−24815
The Journal of Physical Chemistry C
The model was implemented in Python using a Scipy curve
fitting routine,42 and the results of the fit with the experimental
are summarized in Figure 6. The obtained ki and s parameters
Figure 6. EELS fitting. Fitting of the core-loss experimental spectra
from each α-Bi2O3 sample with the weighted contributions of each
nonequivalent oxygen atom and an energy shift.
(detailed in the Supporting Information) show that the
occupancy of OI and OIII barely changes in the three samples.
Nonetheless, a small shift of approximately 0.45 eV for the
nanowire sample and 0.58 eV for the pellet is introduced, and
the occupation OII drops to zero for these samples.
Furthermore, from Figure 5, the linear relation between O K
edge energy shift and vacancy concentration can be calculated
by a linear regression, yielding an r-value of 0.97 (details in the
Supporting Information). From the regression values, the O K
shift value of the nanowire and pellet sample can be translated
into approximately 1.2% of vacant oxygen sites in the nanowires
and 1.6% in the pellet sample, relative to the powder. These
results would indicate that, although the commercial powder
has a certain density of vacancies, they are uniformly distributed
in all crystallographic positions, while the ceramics and
nanowires have a higher concentration of vacancies preferably
located at the OII site.
The findings are in good agreement with PL results. Previous
luminescence studies of α-Bi2O3 ceramic samples annealed in
different atmospheres had shown that a 2.10 eV band is related
to oxygen vacancies.23 Other works reported that the emission
in the low-energy spectral range (≈below 2.1 eV) can also be
attributed to oxygen vacancies that form defect donor states,43
and that the higher is the emission intensity in this spectral
range, the greater is the vacancy density.44 Hence, the higher
PL intensity observed in ceramics and nanowires, as compared
to that measured in powder samples, as well as the higher
relative intensity of the PL emissions in the low-energy spectral
range, support the fact that our EELS results reveal a higher
concentration of oxygen vacancies in the studied α-Bi2O3
nanowires and ceramics.
24814
Article
■ CONCLUSIONS
Structural characterization and PL measurements have been
carried out. Core-loss EELS spectra of Bi2O3 have been
reported for the first time for three different α-Bi2O3 samples.
These measurements have been interpreted with the help of
DFT simulations of the α-Bi2O3 unit cell and of super cells
containing oxygen vacancies. The DFT simulations establish
the relation between oxygen vacancy concentration and O K
edge onset. The EELS spectra analysis confirms an increase in
oxygen vacancies in the nanowires and the pellet sample,
occupying preferably the OII crystallographic site. Remarkably,
the PL results are in good agreement with these findings.
Therefore, the current work allows ELNES analysis to be an
efficient tool to assess oxygen vacancies in Bi2O3, extracting a
type of information that was not possible to access previously
from EEL spectra.
■
*
ASSOCIATED CONTENT
S Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.jpcc.7b06310.
HAADF images coacquired from the spectra in Figure 3;
Bi2O3 O K edge orientation dependence simulations;
vacancy formation energies obtained from the DFT
calculations; and fitting description and obtained
parameters from Figures 5 and 6 (PDF)
■ AUTHOR INFORMATION
Corresponding Author
*E-mail: ptorruella@el.ub.edu.
ORCID
Pau Torruella: 0000-0002-6864-4000
Catalina Coll: 0000-0002-3035-2555
Lluís López-Conesa: 0000-0003-1456-079X
Notes
The authors declare no competing financial interest.
■ ACKNOWLEDGMENTS
This work has been carried out in the frame of the Spanish
research projects MAT2013-41506-P, MAT2016-79455-P,
MAT2013-48628-R, MAT2015-65274-R, MAT2015-066888-
C3-3-R, CSD2009-0013, and FIS2013-46159-C3-3-P. It was
also supported by the Spanish Agencia Estatal de Investigación
(AEI) and the European Regional Development Fund (ERDF).
Catalan Government support from the SGR2014-672 and
2014-SGR-1015 projects is also acknowledged. Financial
support from the ERC PoC MAGTOOLS is acknowledged.
Measurements were performed in the Centro Nacional de
Microscopiá Electrónica (CNME) at the Universidad Complu-
tense de Madrid (UCM) and also at the Centres Cientifico- ́
Tècnics of the Universitat de Barcelona (CCIT-UB). The DFT
simulations were carried out in the LUSITANIA II super-
computer located in Cáceres, a node of the Spanish
supercomputing net (Red Española de Supercomputación-
RES).
■ REFERENCES
(1) Harwig, H. A.; Gerards, A. G. The Polymorphism of Bismuth
Sesquioxide. Thermochim. Acta 1979, 28, 121−131.
(2) Harwig, H. A. On Structure of Bismuth Sesquioxide: The A, B, Γ
and Δ-Phase. Z. Anorg. Allg. Chem. 1978, 444, 151−166.
DOI: 10.1021/acs.jpcc.7b06310
J. Phys. Chem. C 2017, 121, 24809−24815
The Journal of Physical Chemistry C
(3) Gou, X.; Li, R.; Wang, G.; Chen, Z.; Wexler, D. Room-
Temperature Solution Synthesis of Bi2O3 Nanowires for Gas Sensing
Application. Nanotechnology 2009, 20 (49), 495501.
(4) Cabot, A.; Marsal, A.; Arbiol, J.; Morante, J. R. Bi2O3 as a
Selective Sensing Material for NO Detection. Sens. Actuators, B 2004,
99 (1), 74−89.
(5) Vila, M.; Díaz-guerra, C.; Lorenz, K.; Piqueras, J.; Alves, E.;
Nappini, S.; Magnano, E. Structural and Luminescence Properties of
Eu and Er Implanted Bi 2 O 3 Nanowires for Optoelectronic
Applications. J. Mater. Chem. C 2013, 1, 7920−7929.
(6) Eberl, J.; Kisch, H. Visible Light Photo-Oxidations in the
Presence of A-Bi2O3. Photochem. Photobiol. Sci. 2008, 7, 1400.
(7) Wachsman, E. D.; Lee, K. T. Lowering the Temperature of Solid
Oxide Fuel Cells. Science (Washington, DC, U. S.) 2011, 334, 935.
(8) Zhong, G. H.; Wang, J. L.; Zeng, Z. Ionic Transport Properties in
Doped Δ-Bi2O3. J. Phys.: Conf. Ser. 2006, 29, 1−5.
(9) Takahashi, T.; Iwahara, H.; Arao, T. High Oxide Ion Conduction
in Sintered Oxides of the System Bi2O3-Y2O3. J. Appl. Electrochem.
1975, 5, 187−195.
(10) Shuk, P.; Wiemhöfer, H. D.; Guth, U.; Göpel, W.; Greenblatt,
M. Oxide Ion Conducting Solid Electrolytes Based on Bi2O3. Solid
State Ionics 1996, 89 (1996), 179−196.
(11) Niu, K. Y.; Park, J.; Zheng, H.; Alivisatos, a. P. Revealing
Bismuth Oxide Hollow Nanoparticle Formation by the Kirkendall
Effect. Nano Lett. 2013, 13 (11), 5715−5719.
(12) Kumari, L.; Lin, J.; Ma, Y. One-Dimensional Bi2O3 Nanohooks:
Synthesis, Characterization and Optical Properties. J. Phys.: Condens.
Matter 2007, 19, 406204.
(13) Qiu, Y.; Liu, D.; Yang, J.; Yang, S. Controlled Synthesis of
Bismuth Oxide Nanowires by an Oxidative Metal Vapor Transport
Deposition Technique. Adv. Mater. 2006, 18 (19), 2604−2608.
(14) Sarma, B.; Jurovitzki, A. L.; Smith, Y. R.; Mohanty, S. K.; Misra,
M. Redox-Induced Enhancement in Interfacial Capacitance of the
Titania Nanotube/bismuth Oxide Composite Electrode. ACS Appl.
Mater. Interfaces 2013, 5 (5), 1688−1697.
(15) Mahmoud, W. E.; Al-Ghamdi, A. A. Synthesis and Properties of
Bismuth Oxide Nanoshell Coated Polyaniline Nanoparticles for
Promising Photovoltaic Properties. Polym. Adv. Technol. 2011, 22
(6), 877−881.
(16) Williams, D. B.; Carter, C. B. Transmission Electron Microscopy:
A Textbook for Materials Science; Springer: New York, 2009; Vols. V1−
V4.
(17) Egerton, R. F. Electron Energy-Loss Spectroscopy in the Electron
Microscope, 3rd ed.; Springer: New York, 2011.
(18) Boyapati, S.; Wachsman, E. D.; Jiang, N. Effect of Oxygen
Sublattice Ordering on Interstitial Transport Mechanism and
Conductivity Activation Energies in Phase-Stabilized Cubic Bismuth
Oxides. Solid State Ionics 2001, 140, 149−160.
(19) Shuk, P.; Wiemhöfer, H. D.; Göpel, W. Electronic Properties of
Bi2O3 Based Solid Electrolytes. Z. Anorg. Allg. Chem. 1997, 623, 892−
896.
(20) Yedra, L.; Xuriguera, E.; Estrader, M.; López-ortega, A.; Baró,
M. D.; Nogués, J.; Roldan, M.; Varela, M.; Estradé, S.; Peiró, F. Oxide
Wizard: An EELS Application to Characterize the White Lines of
Transition Metal Edges. Microsc. Microanal. 2014, 20, 698−705.
(21) Varela, M.; Oxley, M. P.; Luo, W.; Tao, J.; Watanabe, M.;
Lupini, a. R.; Pantelides, S. T.; Pennycook, S. J. Atomic-Resolution
Imaging of Oxidation States in Manganites. Phys. Rev. B: Condens.
Matter Mater. Phys. 2009, 79 (8), 1−14.
(22) Vila, M.; Diaz-Guerra, C.; Piqueras, J.; Lopez-Conesa, L.;
Estrade, S.; Peiro, F. Growth, Structure, Luminescence and Mechanical
Resonance of Bi2O3 Nano- and Microwires. CrystEngComm 2015, 17
(1), 132−139.
(23) Vila, M.; Díaz-Guerra, C.; Piqueras, J. Luminescence and Raman
Study of A-Bi2O3 Ceramics. Mater. Chem. Phys. 2012, 133 (1), 559−
564.
(24) Hébert, C. Practical Aspects of Running the WIEN2k Code for
Electron Spectroscopy. Micron 2007, 38, 12−28.
24815
Article
(25) Vila, M.; Díaz-Guerra, C.; Piqueras, J. Laser Irradiation-Induced
A to Δ Phase Transformation in Bi2O3 Ceramics and Nanowires. Appl.
Phys. Lett. 2012, 101 (7), 071905.
(26) Blaha, P.; Schwarz, K.; Madsen, G.; Kvasicka, D.; Luitz, J.
WIEN2K, an Augmented Plane Wave + Local Orbital Program for
Calculating Crystal Properties; Technische Universität Wien: Germany,
2001.
(27) Blaha, P.; Schwarz, K.; Sorantin, P. Full-Potential, Linearized
Augmented Plane Wave Programs for Crystalline Systems. Comput.
Phys. Commun. 1990, 59 (2), 399−415.
(28) Hohenberg, P.; Kohn, W. Inhomogenous Electron Gas. Phys.
Rev. 1964, 136 (3), 864−871.
(29) Kohn, W.; Sham, L. J. Self-Consistent Equations Including
Exchange and Correlation Effects. Phys. Rev. 1965, 140 (4), 1133−
1139.
(30) Kalinchenko, F. V.; Borzenkova, M. P.; Novoselova, A. V. The
BiF3-Bi2O3 System. Russ. J. Inorg. Chem. 1981, 26, 118−120.
(31) Betsch, R. J.; White, W. B. Vibrational Spectra of Bismuth Oxide
and the Sillenite-Structure Bismuth Oxide Derivatives. Spectrochim.
Acta Part A Mol. Spectrosc. 1978, 34 (5), 505−514.
(32) Denisov, V. N.; Ivlev, A. N.; Lipin, A. S.; Mavrin, B. N.; Orlov,
V. G. Raman Spectra and Lattice Dynamics of Single-Crystal A-Bi2O3.
J. Phys.: Condens. Matter 1997, 9 (23), 4967−4978.
(33) Narang, S. N.; Patel, N. D.; Kartha, V. B. Infrared and Raman
Spectral Studies and Normal Modes of A-Bi2O3. J. Mol. Struct. 1994,
327 (2−3), 221−235.
(34) Giannuzzi, L. A.; Kempshall, B. W.; Schwarz, S. M.; Lomness, J.
K.; Prenitzer, B. I.; Stevie, F. A. FIB Lift-Out Specimen Preparation
Techniques. Introduction to Focused Ion Beams; Kluwer Academic
Publishers: Boston, MA, 2005; pp 201−228.
(35) Hamann, D. R.; Muller, D. A. Absolute and Approximate
Calculations of Electron-Energy-Loss Spectroscopy Edge Thresholds.
Phys. Rev. Lett. 2002, 89 (12), 126404−4.
(36) Ewels, P.; Sikora, T.; Serin, V.; Ewels, C. P.; Lajaunie, L. A
Complete Overhaul of the Electron Energy-Loss Spectroscopy and X-
Ray Absorption Spectroscopy Database: Eelsdb.eu. Microsc. Microanal.
2016, 22, 717−724.
(37) Schmid, H. K.; Mader, W. Oxidation States of Mn and Fe in
Various Compound Oxide Systems. Micron 2006, 37 (5), 426−432.
(38) López-Ortega, A.; Roca, A. G.; Torruella, P.; Petrecca, M.;
Estradé, S.; Peiró, F.; Puntes, V.; Nogués, J. Galvanic Replacement
onto Complex Metal-Oxide Nanoparticles: Impact of Water or Other
Oxidizers in the Formation of Either Fully Dense Onion-like or
Multicomponent Hollow MnOx/FeOx Structures. Chem. Mater. 2016,
28 (21), 8025−8031.
(39) Wang, Z.; Bentley, J.; Evans, N. Valence State Mapping of
Cobalt and Manganese Using near-Edge Fine Structures. Micron 2000,
31 (4), 355−362.
(40) Torruella, P.; Arenal, R.; de la Peña, F.; Saghi, Z.; Yedra, L.;
Eljarrat, A.; López-Conesa, L.; Estrader, M.; López-Ortega, A.; Salazar-
Alvarez, G.; et al. 3D Visualization of the Iron Oxidation State in FeO/
Fe3O4 Core−Shell Nanocubes from Electron Energy Loss Tomog-
raphy. Nano Lett. 2016, 16, 5068−5073.
(41) Schattschneider, P.; Jouffrey, B.; Hebert, C. Orientation
Dependence of Ionization Edges in EELS. Ultramicroscopy 2001, 86,
343−353.
(42) Jones, E.; Oliphant, T.; Peterson, P.; et al. {SciPy}: Open Source
Scientific Tools for {Python}, 2001.
(43) Schmidt, S.; Kubaski, E. T.; Volanti, D. P.; Sequinel, T.; Bezzon,
V. D. N.; Beltrán, A.; Tebcherani, S. M.; Varela, J. A. Effect of
Pressure-Assisted Heat Treatment on Photoluminescence Emission of
A-Bi2O3 Needles. Inorg. Chem. 2015, 54 (21), 10184−10191.
(44) Wu, Y.; Lu, G. The Roles of Density-Tunable Surface Oxygen
Vacancy over Bouquet-like Bi2O3 in Enhancing Photocatalytic Activity.
Phys. Chem. Chem. Phys. 2014, 16 (9), 4165−4175.
DOI: 10.1021/acs.jpcc.7b06310
J. Phys. Chem. C 2017, 121, 24809−24815