Structure and composition peculiarities and spectral-luminescent properties of colorless and pink Bi₄Ge₃O₁₂ scintillation crystals

2.1 Objects and methods

The colorless Bi₄Ge₃O₁₂ single crystal has been grown by the Czochralski method. Additional high temperature treatment (900–950 °C) of colorless sample results in obtaining the colored (pink) BGO sample. The single-crystal and powder X-ray diffraction were performed at room temperature with the Enraf-Nonius CAD-4 and HZG-4 diffractometers, respectively. The EXAFS/XANES spectroscopy of the powdered samples was carried out at room temperature at the Structural Materials Science beamline of the Kurchatov synchrotron radiation source. The transmission spectra were measured with a spectrometer PerkinElmer Lambda 900 and the luminescence and excitation spectra were recorded with an Edinburgh FLSP920 spectrofluorometer.

2.2 Synthesis procedure

The colorless optically-qualitative Bi₄Ge₃O₁₂ single crystal, BGO(C), of 16 mm in diameter and 65 mm in length has been grown by the Czochralski method (Fig. 1a, b). It was free from cracks, bubbles, inclusions of second phases, and scattering centers and it was found to be transparent in the range of 0.35–6.5 µm, which has been detected using a Carry-5000 spectrophotometer. Under optimal conditions, the bulk crystallization rate of the crystal, grown along the [1 1 1] direction, was not exceed 0.5 cm³/h, the axial temperature gradients in the growth and annealing zones were 80–90 °C/cm and 10–15 °C/cm, respectively.

Close

Fig. 1

Photo of colorless BGO(C) crystal (a), control of scattering centers in BGO crystal (b), and colorless BGO(C) and pink BGO(P) samples cut for investigations (c).

Export to PPT

The colored (pink) samples, BGO(P) (Fig. 1c), were obtained by additional high temperature treatment (900–950 °C) of plates of 1.5–2.0 mm in thickness, cut from the colorless boule, in graphite crucibles at a pressure of 10⁻² Torr and using stannous oxalate SnC₂O₄ in an amount of several milligrams as a reducing agent. The annealing time was varied from 6 to 10 h.

2.3 Single crystal X-ray diffraction

The single-crystal X-ray diffraction (SCXRD) analysis of the BGO(C) and BGO(P) microcrystals of ∼0.2 × 0.2 × 0.2 mm³ in size was carried out with a Enraf-Nonius CAD-4 single-crystal diffractometer at room temperature (AgK_α, graphite monochromator, ω- scan mode). To reduce the error associated with absorption, the SCXRD data were collected over the entire Ewald sphere. The preliminary SCXRD data processing was carried out using the WinGX pack (Farrugia, 1999). The atomic coordinates, anisotropic displacements parameters of all atoms, and occupancies of cation and oxygen sites were refined using the SHELXL-97 software package (Sheldrick, 2008), taking into account the atomic scattering curves for neutral atoms, with the semi-empirical (azimuthal scan) (North et al., 1968) or empirical (Walker and Stuart, 1983) correction of absorption.

The structural parameters were refined in several steps. Initially, the positional and thermal parameters were simultaneously refined in the isotropic and anisotropic approximations. Then the refinement of thermal parameters together with the occupancy of octahedral site was performed with the fixed occupancies of tetrahedral (Ge) and oxygen sites. Thereafter, the occupancy of tetrahedral site was refined with the fixed occupancy of oxygen site, and vice versa. Due to the well-known correlation between thermal parameters and site occupancies we used the strategy of crystal structure refinement developed by our team for both present objects and other complex oxides and described in detail (Kaurova et al., 2010; Kuz’micheva, 2016). After each refinement step, the residual electron density, thermal parameters, and interatomic distances were analyzed. The actual compositions taking into account the electroneutrality, the correct values of the thermal parameters, the lowest values of the R factors, and the absence of residual electron-density peaks serve as criteria for the accuracy of the structure refinement and the correctness of the determination of the composition.

2.4 Powder X-ray diffraction

The powder X-ray diffraction analysis (XRPD) of ground samples (∼100 mg) was carried out at room temperature on an HZG-4 X-ray powder diffractometer.

2.5 EXAFS/XANES spectroscopy

The EXAFS/XANES spectra of powdered samples (∼100 mg), taken from the same part of crystals as for SCXRD experiment, were measured at room temperature at the Structural Materials Science beamline of the Kurchatov synchrotron radiation source (Trofimova et al., 2013). The samples were ground to a powder and evenly applied on the adhesive tape, having a small X-ray absorption coefficient. The sample thickness was chosen experimentally in order to provide a unit edge step, i.e. to make an absorption coefficient change approximately by one in natural log scale (or e times in linear scale) during the energy scan of near-edge range, but not to allow the absorption coefficient itself exceed 4 in log scale. The energy scans were performed using a Si(2 2 0) crystal monochromator with an energy resolution of ΔE/E ∼ 2 × 10⁻⁴. XAFS spectra were collected at the Bi and Ge K edges in a transmission mode, placing the sample between two ionization chambers connected to picoammeter (Keithley), which also served as a voltage source. The intensity of the monochromatic beam incident on the sample and passing through it was measured in an air ionization chamber and a chamber filled with pure Ar up to atmospheric pressure, respectively.

The standard processing of the experimental spectra was performed using the IFEFFIT program package (Ravel and Newville, 2005). The character of the atom's immediate environment was studied by analyzing the radial distribution function φ(r), obtained by Fourier transform of the k³·χ(k) function during the experiment, where the multiplication by k³ was used to compensate the attenuation of the oscillations of the X-ray absorption coefficient with distance from the absorption edge.

The Fourier transforms of EXAFS oscillations were extracted in the range of the photoelectron wavenumbers (k) from 2 to 14 Å⁻¹ at the Ge K-edge and from 2 to 12.5 Å⁻¹ at the Bi L₃-edge and modelled in the range of the interatomic distances (R, Å) from 1.2 to 3.0 Å at the Bi L₃-edge, and from 1.4 to 3.6 Å at the Ge K-edge.

2.6 Optical and luminescence spectroscopy

The transmission spectra were measured in the wavelength range from 0.3 to 2 µm on a spectrometer Lambda 900 (PerkinElmer) with the scan step 1 nm. Measurements of luminescence and excitation spectra of the samples were carried out using an optical spectrofluorimeter FLSP920 (Edinburgh Instruments) with a 250 W xenon lamp operating in the wavelength range 240–850 nm. The spectral width of excitation line Δλ was ∼10 nm. The luminescence in the spectral region 240–1650 nm was detected by photomultipliers in a perpendicular configuration. The spectral resolution of the luminescence spectra obtained was 1 nm. Glass filters were used to eliminate scattered excitation radiation.

3.1 SCXRD analysis

Fig. 2 shows the diffraction patterns of the BGO powders made of the single crystals. Their indexing confirms that the BGO samples under investigation are isostructural and belong to the eulytine structure (JCPDS №89-1419). A shift for the reflections at high 2θ angles evidences most likely the difference in the unit cell parameters. In turn, the difference in the unit cell parameters found for BGO(C) and BGO(P), both for powdered samples and microcrystals (Fig. 3), evidences the difference in their actual compositions.

Close

Fig. 2

Diffraction patterns of BGO powders made of the single crystals (λ = 0.6889 Å). Box gives the reflection profile for BGO(C) and BGO(P).

Export to PPT

Close

Fig. 3

The unit cell parameters for BGO(C) and BGO(P) samples according to SCXRD (microcrystals) and XRPD (powdered crystals).

Export to PPT

Tables 1 and 2 show the experimental parameters and the results of refinement of BGO(C) and BGO(P) crystal structures, and Table 3 shows their refined actual compositions, namely, (Bi_3.994□_0.006)(Ge_2.980(48)Bi_0.020)O_12.00 – for the colorless BGO(C) crystal and (Bi_3.987(45)□_0.013)(Ge_2.988(35)Bi_0.012)(O_11.98(5)□_0.02) – for the pink BGO(P) crystal (vacancies are denoted by squares, □).

As can be seen from the refined compositions, both crystals contain defects (Table 3). The unit cell parameters of these crystals are less than those found for the defect-free Bi₃Ge₄O₁₂ crystals, grown by the Czochralski and hydrothermal methods and investigated by the neutron diffraction including the refinement of site occupancies (Radaev et al., 1990). Analysis of the compositions of microcrystals (without standard deviations) allowed to reveal: a small deficiency of oxygen sites (V_O^•n) in the BGO(P) structure and the absence of O vacancies in the BGO(C) crystal; vacancies in the Bi sites (V_Bi′^m) in both structures (greater for the BGO(P) microcrystal); the presence of Bi in tetrahedral Ge sites, i.e. antisite defect Bi_Ge′ (with the assumption that Bi has formal charge 3+) for both crystals (greater for the BGO(C) sample); the higher Bi content in the BGO(C) microcrystal compared with the BGO(P) one.

The distorted BiO₆ polyhedron with the different cation-anion distances (Bi-3O2 > Bi-3O1) is a characteristic feature of the Bi₃Ge₄O₁₂ crystal structure (Fig. 4a). The distortion is due to the lone active pair of electrons (E-pair), the location of which is detected by the analysis of the residual electron density (x = y = z = 0.9389) in the structure of BGO(C) sample (Fig. 4b).

Close

Fig. 4

The BiO₆ polyhedron (a), the BiO₆ polyhedron with an active lone pair of electrons (b), the GeO₄ polyhedron (c), the connection between BiO₆ and GeO₄ polyhedra (d, e), and the unit cell of Bi₄Ge₃O₁₂ (f).

Export to PPT

According to the calculation of the degree of distortion (δ) of the BiO₆ polyhedron using the formula δ = ΣΔd_i²/(CN − 1) (Δd is a difference in distances between the vertices of a distorted and ideal coordination polyhedron; the i is changed from 1 to CN), its value in the structure of BGO(P) sample (δ = 1.534) is found to be lower than that in the BGO(C) structure (δ = 1.855). It is due to the presence of O vacancies (V_O^•n) in the structure of BGO(P) sample, which reduce the action of the active E-pair and contribute to the symmetrization of polyhedron.

The GeO₄ tetrahedra have equal cation-anion distances (Fig. 4c) with different O-Ge-O bond angles (Table 2). The distorted BiO₆ octahedron is connected with the vertices of the GeO₄ tetrahedra by the O2 atoms, the most distant from Bi atoms (Fig. 4d), and with other BiO₆ polyhedron by an edge (Fig. 4e). The GeO₄ and BiO₆ polyhedra are connected with each other by their vertices to form the chains (Fig. 4f) due to the connection between the vertices of three GeO₄ tetrahedra, and four Bi polyhedra, and the edges of two BiO₆ polyhedra. The coordination number of oxygen atoms is three (CNO = 3): the O atoms are coordinated by two Bi atoms and one Ge atom.

The refined compositions of Bi crystallographic site correlate with the interatomic distances in the BiO₆ polyhedron (Table 2): the average Bi-O distance in the structure of BGO(C) sample with less concentration of defects in the Bi site is higher than the same distance in the structure of BGO(P) sample. However, the cation-anion distance in the GeO₄ tetrahedron is slightly lower in the structure of BGO(C) crystal compared with the same distance in the structure of pink BGO(P) one with less Bi content in the tetrahedral sites (rBi > rGe, r – radius of ion). Such a behavior of the interatomic distances can be explained by the mutual influence of polyhedra on each other (a decrease in the Bi-O2 interatomic distance in the BiO₆ polyhedron increases the Ge-O interatomic distance in the GeO₄ polyhedron), as it was observed in the structure of defect-free Bi₃Ge₄O₁₂ crystal with increasing temperature (Radaev et al., 1990).

There is also another explanation for such a behavior of the interatomic distances in the BGO structures. The presence of the Bi³⁺ions in the tetrahedral sites is highly problematic due to the large cation size (rBi³⁺ ∼ 0.85 Å). At the same time, the presence of the Bi⁵⁺ions (rBi⁵⁺ ∼ 0.60 Å) in the tetrahedral sites, especially together with the other ions with the smaller sizes (Craig and Stephenson, 1975), is justified in terms of the crystallochemical point of view. According to the oxygen radius (R ∼ 1.38 Å), the atoms in the tetrahedral sites may have dimensions r ∼ 0.27 ÷ ∼0.57 Å. On the other hand, based on the well-known Magnus-Goldschmidt rule, the tetrahedral coordination is stable in the range of r/R from 0.225 to 0.414 (r - cation radius, R - anion radius) for preferably ionic crystals. It follows that the cations with the r ∼ 0.31 ÷ ∼0.57 Å may be located in the center of tetrahedron, which is consistent with the above-mentioned limit (r ∼ 0.27 ÷ ∼0.57 Å), taking into account the limitations for such calculations (Kuz’micheva, 2016).

3.2 XRPD analysis

The full-profile refinement of structures of BGO single crystals (Fig. 1c) ground to a powder is performed using a MRIA program (Zlokazov and Chernyshev, 1992). Due to the strong correlation between the structural parameters (atomic coordinates, atomic displacements, and occupancies), the site occupancies of all the atoms have not been refined. However, when the atomic displacements for O atoms and the site occupancies for all atoms are fixed to be B_iso(O) = 1.00 and p = 1, respectively, the refined atomic displacements for the Bi and Ge atoms have the following values: B_iso(Bi) = 0.34(2) Ų and B_iso(Ge) = 0.51(3) Ų - for the BGO(C) sample, and B_iso(Bi) = 0.23(2) Ų and B_iso(Ge) = 0.14(3) Ų - for the BGO(P) sample. In this case, it can be concluded that the greater Bi content (N_Bi = 83, N is an atomic number in the Periodic Table) is in the Ge site (N_Ge = 32) of BGO(C) structure compared with the BGO(P) one, which is consistent with the results of SCXRD analysis (Table 3).

3.3 EXAFS/XANES spectroscopy

According to the XRPD data obtained for the BGO single crystals ground to a powder, an average size of crystallites for both samples can be estimated as hundreds of nanometers, hence, the content of surface atoms is sufficiently low. Therefore, it can be assumed that the change in the amplitude of EXAFS oscillations, which determine the intensity of peaks on the curves of quasi-radial electron density distribution and the average values of coordination numbers (CN) obtained as a result of the simulation of EXAFS spectra, are not related to a dispersity and due only to the differences between crystal structures of pink and colorless BGO crystals.

The XANES spectra measured for the BGO(C) and BGO(P) samples at both edges are almost identical (Fig. 5). The small shift of ∼0.5 eV in the absorption edge towards lower energies, found for the BGO(P) sample, could mean a decrease in the average formal charge (FC) of Bi (the method does not allow to distinguish similar atoms in different sites) (Fig. 5). However, this shift is below the energy resolution of monochromator: ΔE/E ∼ 2 × 10⁻⁴ gives ΔE ≈ 2.6 eV at E ≈ 13 keV; repeated measurements do not lead to reproducible shift values but indicate that for the BGO(P) sample the absorption jump lies in the lower energy region. If only it is not caused by inaccuracy of measurements or sample preparation procedures, such a decrease can be explained by a different distribution of Bi atoms over two crystallographic sites with different FCs, Bi⁵⁺and Bi³⁺. The Bi⁵⁺content is found to be lower in the pink BGO(P) crystal compared with the colorless BGO(C) one. Based on the XANES curves at the Bi L₃-edge (Fig. 5a), the greater content of the Bi³⁺ions in the BiO₆ sites and lower content of the Bi⁵⁺ions in the GeO₄ sites was found for the BGO(P) sample compared with the BGO(C) one. Moreover, due to the fact that there are only Bi³⁺ions in the distorted octahedron (due to its crystallochemical properties), the Bi ions with FC > 3 occupy tetrahedral site together with the Ge ions, which is consistent with the interatomic distances in the tetrahedron (Table 2).

Close

Fig. 5

The XANES spectra, measured at the Bi L₃-edge (a) and Ge K-edge (b) for BGO(C) and BGO(P) crystals.

Export to PPT

A similar result was obtained for the phase Bi₂₄Mn₂O₄₀ with the sillenite structure (sp. gr. I23; the Bi³⁺ions with the active lone pair of electrons are located in semi-octahedral sites and the Mn⁵⁺ions are in the tetrahedra) (Mel’nikova et al., 2011): the presence of Bi cations in an amount up to 10% was revealed in the tetrahedral sites using the XANES/EXAFS spectroscopy. The EXAFS spectrum of this sample at the Bi L₃-edge is well reproduced in the event that one additional interatomic distance Bi-O ∼ 1.90 Å is added to a description model together with three initial Bi-O distances (2.07–2.56 Å). In the XANES spectra at the Bi L₃-edge, a slight shift towards higher energies was revealed for the Bi₂₄Mn₂O_40-δ sample with respect to the spectrum of defect-free Bi³⁺₂₄Si⁴⁺_2.00(4)O₄₀ sample, which is due to the appearance of Bi atoms in the tetrahedral sites, i.e. Bi₂₄(Bi,Mn)₂O_40-δ. It should be noted that the average FC of Bi in the Bi₂₄(Bi,Mn)₂O_40-δ is higher than that in the Bi³⁺₂₄Si⁴⁺_2.00(4)O₄₀. This fact indicates that the FC of Bi in the tetrahedral site of Bi₂₄(Bi,Mn)₂O_40-δ structure is higher than 3+, i.e. Bi^3+δ. The XANES/EXAFS results together with the results of X-ray analysis of sillenite-family single crystals show that the concentration of these ions in the tetrahedral sites should not exceed ∼0.25 FU (FU is a formula units) (Mel’nikova et al., 2014). It should be noted that the possibility of the presence of Bi³⁺ions together with the Bi⁵⁺ions in the tetrahedral sites of sillenite structure has been described in several works, in particular (Craig and Stephenson, 1975; Harwig, 1978).

Based on the above-mentioned XANES data, the peak on the curve of quasi-radial electron density distribution at the Bi edge, corresponding to the Bi-O distance, is to be more intense in the structure of pink BGO(P) crystal than that in the colorless BGO(C) one. In the curve at the Ge K-edge (Fig. 5b), on the other hand, the changes in oxygen peak are to be minor, as all Ge atoms are located in the tetrahedra in the structures of both samples. Similarly, an increase in the Bi content in the special tetrahedral Ge sites should lead to an increase in the intensity peaks for the Bi-Bi (at the Bi L₃-edge) and Ge-Bi (at the Ge K-edge) distances. In turn, the intensity of Bi-Ge and Ge-Ge peaks should be decreased due to the fact that the portion of heavy Bi atoms in the Ge sites is reduced. All the above-mentioned changes are correlated with the EXAFS data (Fig. 6).

Close

Fig. 6

EXAFS Fourier transforms, measured at the Bi L₃-edge (a) and Ge K-edge (b) for BGO(C) and BGO(P) crystals. The first peak corresponds to the oxygen environment of both atoms (single scattering path for photoelectrons M-O, M = Bi, Ge), the second and the third peaks (for Bi) correspond to M-Bi and M-Ge. The arrows show the behavior of the spectra intensity.

Export to PPT

There are challenges in the quantitative interpretation of EXAFS at the Bi L₃-edge, caused by the Bi location in two different crystallographic sites and asymmetry of the local environment of Bi in the Ge sites (Fig. 6a). Coordination environment of Bi in the GeO₄ includes three close O atoms (Bi-O_close ∼ 2.15 Å) and three more distant O atoms (Bi-O_far ∼ 2.60 Å). Close and distant O atoms are to be described by two independent scattering paths. Hence, a model with three independent scattering paths is to be used for a correct description of the most intense Bi-O peak. The location of Bi atoms in the Ge sites should not be taken into account due to the fact that the typical values of accuracy make up more than 10% (when determining the CN of atom according to the EXAFS data). In this case, only two scattering paths, corresponding to the octahedral oxygen environment of Bi atoms, should be included into the model. The parameters of this model are shown in Table 4. The CNs for close and distant O atoms were equated to simplify the model according to reasons related to large size of the particles (see above). Giant values of the Debye parameter for distant O atoms are not caused by any errors in measurement or simulation: the forced reduction of these values (for example, by equating to the Debye parameters made for close O atoms) leads to an increase in the R-factor (reliability factor) up to a few percent, and an independent determination of CN leads to incorrect values, which are significantly greater than the bulk values for a perfect crystal. Such high values of the Debye parameter can be easily explained by the fact that the distant O atoms, which form the Bi environment, are, at the same time, close neighbors of the Ge atoms. In case of the partial substitution of Ge atoms by the Bi ones (antisite point defects, Bi_Ge′), these O atoms should be shifted significantly. It follows that an increase in the Debye parameter is not associated with the thermal fluctuations of atoms, but it is probably due to the point defects in the crystal structure. A small value of the Debye parameter, defined for the same atoms from the EXAFS spectrum at the Ge edge, confirms this assumption (Table 5).

A decrease in the content of Bi⁵⁺ions in the BGO(P) crystal compared with the BGO(C) one (Table 3) can be explained by the growth of BGO(P) crystal in a reducing atmosphere: $${\text{Sn}}^{2+}{\text{C}}_2{\text{O}}_4\to {\text{Sn}}^{2+}\text{O}+\left({\text{Bi}}_{3.994\square 0.006}\right)\left[{\text{Ge}}_{2.980\left(48\right)}{\left({\text{Bi}}_{1-y}^{3+}{\text{Bi}}_y^{5+}\right)}_{0.020}\right]{\text{O}}_{12.00}\left(\text{BGO}\right(\text{C}\left)\right)\to {\text{Sn}}^{4+}{\text{O}}_2+\left({\text{Bi}}_{3.987\left(45\right)\square 0.013}\right)\left[{\text{Ge}}_{2.988\left(35\right)}{\left({\text{Bi}}_{1-z}^{3+}{\text{Bi}}_z^{5+}\right)}_{0.012}\right]\left({\text{O}}_{11.98\left(5\right)\square 0.02}\right)\left(\text{BGO}\right(\text{P}\left)\right)\text{with}\text{y}>\text{z.}$$

It should be noted that the refinement of the BGO(P) crystal structure, taking into account that the Sn ions can partially occupy tetrahedral or octahedral sites, resulted in incorrect thermal parameters.

Thus, an investigation of the statistical crystal structure of colorless BGO(C) and colored (pink) BGO(P) crystals by the full-profile method (powdered crystals) and X-ray diffraction analysis (microcrystals), as well as an investigation of the local structure of crystals by the X-ray absorption spectroscopy, allows to obtain a general picture of the defect formation in the samples under investigation: 0 → V_Bi′^m + Bi_Ge′ + Bi_Bi^•p - for BGO(C) crystal and 0 → V_Bi′^m+Bi_Ge′ + Bi_Bi^•p + V_O^•n + ne′ or 0 → V_Bi′^m + Bi_Ge′ + Bi_Bi^•p + (V_O^•n, ne′)^× - for BGO(P) crystal taking into account the color center (V_O^•n,ne′)^×. The similar change in color from colorless to pink, found for the Gd₃Ga₅O₁₂ crystals with the garnet structure, was also caused by the presence of the color center (V_O^••,2e′)^× (F-center) (Kuz'micheva et al., 1988).

3.4 Optical and luminescent properties

The colorless BGO crystals are transparent in the spectral range of 0.35–6.5 µm. The BGO(P) crystal, having specific pink color, is characterized by a series of absorption bands in the visible range at 365, 455, and 518 nm and by weak absorption bands at 960, 1230, and 1800 nm (Fig. 7).

Close

Fig. 7

Absorption spectra of the colored BGO(P) crystal after the vacuum annealing of the colorless BGO(C) crystal in the graphite crucible at 900 and 950 °C.

Export to PPT

A luminescence in the visible region at 480–500 nm and its absence in the infrared region have been revealed for the colorless BGO crystals (Atroschenko et al., 1998). For the pink-colored BGO(P) sample, treated in vacuum in a graphite crucible, two luminescence peaks, the broad one at 1000–1100 nm and the narrow one at 1500 nm, were detected (Fig. 8). The excitation spectrum of the luminescence band at 1030 nm consists of the narrow band peaked at 370 nm, whereas one of the luminescence band at 1500 nm has the broad band at 450 nm. It should be noted that the presented excitation spectra consist of the overlapping bands. As a consequence, both the luminescence bands appeared at 400-nm excitation (Fig. 8). Also, when excitation wavelength was varied, both the luminescence bands were also observed whereas their luminescence intensity ratio changed. The occurrence of infrared luminescence at 1000–1100 and 1500 nm was observed only for the colored (pink) BGO(P) samples.

Close

Fig. 8

Luminescence spectrum excited at 400 nm (a) and excitation spectra of 1030 (b) and 1500-nm (c) luminescence of the colored BGO(P) crystal obtained by vacuum annealing of the colorless BGO(C) crystal in the graphite crucible at 900 °C.

Export to PPT