Effect of pressure-assisted heat treatment on photoluminescence emission of α-Bi 2 O 3 needles

Materials with high photoluminescence (PL) intensity can potentially be used in optical and electronic devices. Although the PL properties of bismuth (III) oxide with a monoclinic crystal structure (α-Bi 2 O 3 ) have been explored in the last few years, methods of increasing PL emission intensity and information relating PL emission to structural defects are scarce. This research evaluated the effect of a pressure-assisted heat treatment (PAHT) on the PL properties of α-Bi 2 O 3 with a needle morphology, which was synthesized by a microwave-assisted hydrothermal (MAH) method. The pressure-assisted heat treatment caused an angular increase between the [BiO 6 ]–[BiO 6 ] clusters of α-Bi 2 O 3 , resulting in a significant increase in the PL emission intensity. The Raman and XPS spectra also showed that the α-Bi 2 O 3 PL emissions in the low-energy region (below about 2.1 eV) are attributed to oxygen vacancies that form defect donor states. The experimental results are in good agreement with first-principles total-energy calculations that were carried out within the periodic density-functional-theory (DFT) framework.


INTRODUCTION
2] In spite of the great technological interest in Bi2O3, some of its optical and electronic properties are not well known.One of the problems associated with Bi2O3 studies is the existence of four main polymorphic phases that exhibit distinct properties: α (monoclinic), β (tetragonal), γ (body-centered cubic), and δ (face-centered cubic).The stable phase at room temperature is α-Bi2O3, while the high-temperature phase is δ-Bi2O3 that stabilizes around 730ºC and melts at 824ºC.The γ-Bi2O3 and β-Bi2O3 phases form at temperatures below 639°C and 650°C, respectively.The formation of β-Bi2O3 depends on impurities and oxide reaction conditions.3] For example, Harwig and Gerards 4 showed the formation γ and β phases using cooling rates of 0.1 and 33ºC min -1 .They also provide a detailed study on the formation of metastable phases of Bi2O3 at different cooling rates.
Depending on the method of preparation, α-Bi2O3 can contain traces of a secondary phase, which can contribute to the bulk properties of the oxide because its polymorphs have a distinct nature.In some cases, the presence of a secondary phase impedes adequate control of the properties of Bi2O3 and makes the technological application of this oxide difficult. 5Therefore, the evaluation of the phases that are present and their possible transitions are important topics related to the potential uses of this material. 6Furthermore, the combination of X-ray powder diffraction (XRD) and the Rietveld Method (RM) 7 allows evaluation of the possible structural variations in α-Bi2O3.
1] The PL spectra of α-Bi2O3 microcrystals generally show a band at 2.8 eV attributed to a recombination between valence and conduction bands, and a band at 2.98 eV, which corresponds to surface state interactions.α-Bi2O3 nanoparticles show a band of PL emission at 3.12 eV. 6,92] Unlike for other oxides of technological interest, information regarding PL emission in α-Bi2O3 related to native defects or other types of defects is still scarce. 13Furthermore, low-energy PL emissions (below about 2.1 eV) are explained in terms of oxygen defects and bismuth oxidation states. 6,11,14Materials with increased PL properties are of interest, and obtaining a higher PL emission at a low synthesis temperature is important to decrease costs in several electronic and optical devices.
Pressure-assisted heat treatment (PAHT) was successfully used to prepare calcium copper titanate (CaCu3Ti4O12, CCTO) 15 and zinc oxide (ZnO) thin films. 16In CCTO films, PAHT caused a decrease in the band gap, moving the PL emission to a lower energy region. 15milarly, in ZnO films, PAHT promoted a significant change in the level of defects, shifting the PL emission from green to orange-yellow. 16HT caused significant changes in the PL properties of CCTO 15 and ZnO 16 thin films, and it is hypothesized that a similar influence could be found in bulk α-Bi2O3.Based on these facts, this study aims to investigate the effect of PAHT on the PL properties of α-Bi2O3 needles synthesized by the microwave-assisted hydrothermal (MAH) method

Microwave-assisted hydrothermal (MAH) synthesis of -Bi2O3
The bismuth solution was prepared by dissolving 2.19 g of Bi(NO3)35H2O in 40 mL HNO3 with continuous stirring.After complete dissolution, 2 mL of PEG-300 were added and the solution was stirred for 10 min.Then, a KOH solution (6 molL -1 ) was added dropwise with stirring until pH = 14.The solution was stirred for 5 min and the formation of a gelatinous white precipitate was observed.The solution and the precipitate were transferred to a polytetrafluoroethylene reactor, and were placed in the MAH system for 30 min at 80ºC with a heating rate of 10ºCmin -1 .After cooling the reactor to room temperature, the formation of a yellow precipitate was observed.The resulting material was washed six times with distilled water until pH ~7.After centrifugation, the powder was dried in petri dishes at 70ºC for 24 h.Finally, the powder was calcined at 400ºC for 1 h at a heating rate of 10ºCmin -1 to eliminate any trace of PEG-300, which could be adsorbed on the samples.Samples obtained by this route were labeled α-Bi2O3-MAH.

Pressure-assisted heat treatment (PAHT)
Samples of α-Bi2O3-MAH were heat treated under an air pressure of 2 MPa at 120ºC for 48 h to evaluate the effect of PAHT on the PL properties of α-Bi2O3.Samples that underwent the pressure-assisted heat treatment were labeled α-Bi2O3-PAHT.

Sample characterization
Powder samples of α-Bi2O3-MAH and α-Bi2O3-PAHT were characterized by X-ray diffraction (XRD) at room temperature using a Rigaku Ultima IV diffractometer with CuKα radiation, a goniometer radius of 185 mm, a curved graphite monochromator, and a scintillation counter operating in the reflection mode with the Bragg-Brentano geometry.The divergence, scattering, and receiving slits were 0.5º, 0.5º, and 0.3 mm, respectively.Data were collected over the 2θ measuring range from 15° to 90° in the continuous scan mode (1°/min).
Refinements were carried out by the Rietveld method using Fundamental Parameters (FP) implemented in the software Topas Academic V. 5. 17 FPs were used to describe the peak profile by executing the convolutions of functions in reference to the contribution of wavelength distribution, instrumental parameters of the geometry used, and sample physical properties.A silicon standard sample was used to obtain the instrumental broadening, which was corrected using FPs. 18The background was adjusted using a Chebyshev polynomial function implemented in the software with four terms for correction.Lattice parameters and crystallite size were refined, and a spherical harmonics preferred orientation correction with eight terms was applied.
Malmros 19 reported the crystalline structure used in these refinements.The values of the crystallite size were calculated using a macro implemented in the software Topas Academic V. 5; 17 the macro considers the integrated peak intensity to estimate the mean value of the crystallite size.After the refinements, CIF files were created for each sample to simulate material structures in the software Diamond V. 3.2e. 20In addition, interatomic distances with uncertainties were calculated using the program Bond_Str. 21rphological features of the samples were analyzed by field emission scanning electron microscopy (FESEM) using a Jeol 7500F instrument.Samples were dispersed in isopropyl alcohol with the assistance of an ultrasound bath.A drop of this suspension was deposited on a conductive silicon substrate, and the sample was dried at room temperature before FESEM analysis.
PL measurements of α-Bi2O3-MAH and α-Bi2O3-PAHT were performed with a Thermo Jarrel Ash Monospec (27 cm) monochromator coupled to an R955 Hamamatsu photomultiplier using a krypton ion laser (Coherent Innova 200) operating at a wavelength of 350 nm and 60 mW output power. Deconvolution analyses of PL spectra were performed using the Peakfit Program (version 4.05); 24 a Gaussian function was used to adjust the PL bands to determine peak positions and their corresponding areas.The least number of Gaussian emissions necessary to obtain the best fit in plots was used, and each parameter was fitted for each spectrum independently.The contribution of each PL band was estimated in terms of its area that was expressed in percentage.
Ultraviolet-visible (UV-vis) absorption spectra were obtained using a Varian Cary 50 spectrophotometer to determine the band gap of α-Bi2O3-MAH and α-Bi2O3-PAHT; the band gap was estimated using the Kubelka-Munk method. 25All these characterizations were performed at room temperature.
Raman spectroscopy was used to estimate short and medium range order, crystal potential fluctuations, and local atomic arrangement of samples, and as an auxiliary technique to investigate changes in crystal symmetry that can be caused by PAHT.Raman spectra were obtained by a spectrometer equipped with a Kimmon set of mirrors.Measurements were obtained for solid samples at room temperature using a He-Cd laser with a wavelength of 514 nm.Scans were obtained over the range of 100 to 600 cm -1 .
The chemical states of the elements in α-Bi2O3-MAH and α-Bi2O3-PAHT were characterized by X-Ray Photoelectron Spectroscopy (XPS) using a UNI-SPECS UHV System commercial spectrometer at a pressure below 5  10 -7 Pa.The Mg Kα line (energy of 1253.6 eV) was used as the ionization source and the pass energy of the analyzer was set to 10 eV.The inelastic background of the high-resolution spectra Bi 4f, C 1s, and O 1s orbitals was subtracted using the Shirley method. 26The spectral binding-energy scale was corrected using the hydrocarbon component fixed at 285.0 eV.The width at half height varied between 1.2 and 2.1 eV, and the peak position were determined with a precision of ± 0.1 eV.

Theoretical calculation
First-principles total-energy calculations were carried out within the periodic densityfunctional-theory (DFT) framework using CRYSTAL14 program package. 27The Kohn-Sham equations have been solved by means of exchange-correlation functionals in the generalized gradient approximation (GGA) developed for solids by Perdew, Burke and Ernzerhof (PBESol). 28The Bi and O centers have been described in the scheme [PS]-41G* and 6-31G*, respectively, where [PS] stands for the non-relativistic effective core pseudo-potential.Both the [PS]-41G* and 6-31G* basis sets can be found at http://www.crystal.unito.it/basis-sets.php.We have optimized geometrical parameters and internal positions at both ambient pressure and 2 MPa (EXTPRESS option).

RESULTS AND DISCUSSION
Both α-Bi2O3-MAH and α-Bi2O3-PAHT samples showed a pure α-Bi2O3 phase, as determined from the XRD data and Rietveld plots that are depicted in Figure 1.Table 1 shows the values of the lattice parameters and atomic positions obtained after the refinements.This table also presents the agreement factors of the refinements.Although these numeric criteria are slightly high, Young 29 reported that it is also necessary to consider graphical criteria.Consequently, a careful evaluation of Figures 1(a  The comparison between the results of the two samples shows a slight variation in atomic position, especially for the oxygen sites.This variation is probably because of the synthesis method used for sample α-Bi2O3-MAH.PAHT virtually caused no alteration in the lattice parameters of the samples; the cell volume was the same for both samples.In addition, another parameter that showed no variation was the crystallite size because both samples showed the same order of magnitude.Crystallite size (nm) 137 (5)  117(4) *Beq = temperature factor.Obs: The temperature factors for oxygen atoms were constrained during the refinement following the values presented by Malmros. 19he Diamond software 20 allowed simulation of the material structure, the coordination of the Bi and O atoms, and measurement of the angle between the coordination polyhedra.The simulations showed changes in the relative distance of separation between atoms when both α-Bi2O3-MAH and α-Bi2O3-PAHT are compared (Table 2), although the coordination of the polyhedra had been preserved.α-Bi2O3-PAHT shows that the relative angles between clusters [BiO6]-[BiO6] were greater than in α-Bi2O3-MAH (Figure 2).       Figure 6 shows the UV-vis absorption spectra of α-Bi2O3-MAH and α-Bi2O3-PAHT, and an estimate of the band gap (Egap) for direct transitions, which was calculated using the Kubelka-Munk 25 method.The deconvolution shown in Figure 5 reveals that the PL emissions with the highest energy levels occurred around 3.05 eV and 2.85 eV (violet region) for α-Bi2O3-MAH and α-Bi2O3-PAHT, respectively.These emissions can be related to the direct transitions between the conduction and valence bands since the band gap energies for both samples were estimated as 2.89 and 2.88 eV.0][11] The band gap of α-Bi2O3-MAH estimated by Kubelka-Munk is slightly lower than the corresponding value for the direct transitions found in PL deconvolution; other works 6,11 reported a similar effect.Furthermore, PAHT promoted an increase in the contribution of the spectral range related to the direct transitions between the conduction and valence bands.Trivalent bismuth ions in inorganic compounds exhibit PL properties related to their 6s 2 levels.
3 In α-Bi2O3-MAH, the bands associated with the blue-green emission corresponded to 65% of the total emission (29% was attributed to the blue region  2.77 eV  while 36% was attributed to the green region  2.50 eV).α-Bi2O3-PAHT showed an increase in PL in the blue-green spectral range to 72%, and an increase in the band relative to the blue component; in this case, 53% was related to the blue region (2.60 eV) and 19% to the green region (2.28 eV).

Figures 4 and 5 also show PL emission in the low-energy region, and in this case, PAHT
showed a significant decrease in the emission of this region related to the total PL emission.α-Bi2O3-MAH has 19% of its emission at 2.15 eV and 7% at 1.89 eV that correspond to 26% of the total emission, while α-Bi2O3-PAHT has 7% of its emission at 1.96 eV and 2% at 1.77 eV, which correspond to only 9% of the total PL emission of α-Bi2O3-PAHT.The luminescence in the low-energy spectral range is normally attributed to the 2 P3/2 -2 P1/2 transition of Bi 2+ ions. 8,10wever, recent works 11,30 reported that the emission in the low-energy spectral range is attributed to the structural defects of oxygen (oxygen vacancies).Wu and Lu 30 also associated the PL emission at this spectral range to oxygen vacancies and stated that the higher the emission intensity at this spectral range the greater the vacancy density.Therefore, in order to determine the cause for the PL emission at this spectral range, it is necessary to verify the existence of Bi 2+ ions on the material surface.In the present work, the presence or absence of Bi 2+ ions in samples was tested using Raman and XPS spectra.
Raman characteristics of α-Bi2O3 are well known. 31For the α-Bi2O3 monoclinic structure with space group P21/c and 4 formula units or 20 atoms per unit cell, analyses predict 30 active Raman modes (15Ag + 15Bg), but only a fraction of these modes is generally observed. 11The Raman spectra in Figure 7 show ten prominent Raman bands for both samples at 117, 138, 150, 183, 210, 284, 313, 410, 445, and 530 cm -1 , which are in agreement with bands found in literature for α-Bi2O3. 6,11,32The band at 117 cm -1 is attributed to the vibration mode of bismuth while bands at 138 and 150 cm -1 refer to Bi-O scattering.Furthermore, all other bands were attributed to oxygen vibrations, which are broader than bands observed at 117, 138, and 150 cm - 1 , and demonstrate a strong anharmonicity of oxygen vibration modes.Bands related to neither another Bi2O3 phase nor Bi 2+ were detected for both α-Bi2O3-MAH and α-Bi2O3-PAHT.
A detailed comparison between the Raman spectra of α-Bi2O3-MAH and α-Bi2O3-PAHT in the low frequency range shows no relevant spectral changes.On the other hand, for frequencies higher than 284 cm -1 , α-Bi2O3-PAHT showed an increase in the spectrum intensity compared to α-Bi2O3-MAH.These different intensities can be associated to changes in the electronic density of the crystal, which also modifies the PL properties.The effect of intensity change in the Raman spectra has already been observed in other materials.For example, a change in the intensity of the Raman spectra was observed for CaTiO3 perovskite-based materials obtained by MAH, 33 and for irregular cube-like (Ca1−xCux)TiO3 microcrystals also synthesized by MAH. 34In both studies, the authors showed that a change in the cluster angles promoted variations in intensities of the Raman spectra, which were attributed to changes in the average electronic density of the crystal, and could modify the PL properties.4] In the present work, the increase in Raman spectra intensity was caused by an increase in the average electronic density, which  and Bi4f5/2 with binding energies of 159 and 164 eV, respectively.In addition, the spin orbit splitting with these binding energies values is a characteristic of Bi 3+ ions in α-Bi2O3. 39e XPS profiles of O1s levels, Figures 8(e) and (f), show a dominant peak at 529.7 eV, whose binding energy is a characteristic of oxygen in Bi2O3 (structural oxygen). The hydroxyl groups could originate when the samples are exposed to air or could be adsorbed during the MAH synthesis in an alkaline medium.
Raman spectroscopy and XPS did not detect any signal indicating the presence of Bi 2+ ions in α-Bi2O3-MAH and α-Bi2O3-PAHT.These results suggests the PL emission in the low energy region (below about 2.1 eV) is due to oxygen vacancies that form defect donor states, and this PL comprises a lot of defect donor states to valence band emissions. 3The oxygen vacancy states may contain centers of bismuth in other oxidation states, including Bi 2+ , which can emit PL from red to blue emission. 41The present work shows that emission in the low energy region (below about 2.1 eV) can be attributed to oxygen vacancies that were coherently reduced after PAHT.
These results are in good agreement with reports from literature that states the PL emission at this region is explained in terms of oxygen defects and bismuth oxidation states. 6,11,14However, XPS results presented in the work of Vila et.al. 6 showed the presence Bi 2+ on sample surface while in the present work Raman and XPS results showed no signal of Bi 2+ .Upon comparing the PL emission of α-Bi2O3-MAH and α-Bi2O3-PAHT (Figure 4), the latter showed an increase in the PL intensity.The pressure-assisted heat treatment of α-Bi2O3-MAH needles caused an increase in the angle between [BiO6]-[BiO6] clusters, which promoted a rise in the electronic density of the material, which was considered as the cause for the increase in PL emission intensity.On the other hand, the contribution of PL emission in the low-energy range to the total PL emission is considerably greater in α-Bi2O3-MAH than in α-Bi2O3-PAHT.This is mainly because α-Bi2O3-PAHT suffered the pressure-assisted heat treatment under air atmosphere (rich in oxygen).Consequently, under higher oxygen pressure, the formation of oxygen vacancies that form the defect donor states was in disadvantage.clusters that were greater in α-Bi2O3-PAHT than in α-Bi2O3-MAH.Rietveld refinements already showed the difference in cluster angles of samples (Figure 2).In addition, the change in cluster angle can directly change the polarization of O 2p states, and, consequently, their interaction with Bi 6p states.As a result, the change in the angles between [BiO6]-[BiO6] clusters promote an increase in the electronic density of crystal and the increase in the PL intensity.These findings are in good agreement with the discussion on Raman spectra (Figure 7), that shows an increase in the electronic density of α-Bi2O3-PAHT.

CONCLUSIONS
The pressure-assisted heat treatment of α-Bi2O3-MAH needles caused an intrinsic disorder characterized by an increase in the angle between the [BiO6]-[BiO6] clusters.This angular increase promoted a significant rise in the electronic density of the material, which was considered as the cause for the increase in PL emission intensity in the blue region.
Consequently, the origin of the structure with higher electronic density and the increase in PL ) and (b) indicates an adequate adjustment between the observed and calculated profiles.

Table 1 .
Refined structural parameters and the respective agreement factors of the refinement.

Figure 6 .
Figure 6.UV-vis absorption spectra and estimation of the band gap.
was promoted by the change in the angles between clusters [BiO6]-[BiO6].Consequently, the intensity increase in the Raman spectra observed in α-Bi2O3-PAHT is explained by a structural disorder caused by the increase in the relative angles between clusters [BiO6]-[BiO6].Results of three experimental and theoretical studies 35-37 on the structural properties of α-Bi2O3 under high pressures, which also demonstrated that pressure promotes a distortion in clusters [BiO6]-[BiO6], supports the results presented in this work.The chemical states of the elements in the α-Bi2O3-MAH and α-Bi2O3-PAHT samples were also characterized by XPS, and Figure 8 depicts these results.Figures 8(a) and (b) show the XPS spectra obtained for α-Bi2O3-MAH and α-Bi2O3-PAHT, respectively.Figures 8(c) and (d) show the Bi4f levels, while Figures 8(e) and (f) show the O1s levels for both samples.

Figure 9 .
Figure 9. Electronic band structures and partial electron density of states for (a) α-Bi2O3MAH Theoretical results showed that O 2p states have an important role in photoelectron transfers in both samples.For both α-Bi2O3-MAH and α-Bi2O3-PAHT, the top of VB is mainly composed of O 2p states, which are predominantly influenced by changes on axial oxygen atoms of[BiO6]  octahedra.The axial oxygen atoms are directly involved in the angles between [BiO6]-[BiO6] photoelectron transfers in both samples.The change in the angle between the [BiO6]-[BiO6]

Table 2 .
Interatomic distances (Å) with uncertainties between Bi and O atoms in the α-Bi2O3