Synthesis and Properties of the Ba2PrWO6 Double Perovskite

We report details on the synthesis and properties of barium praseodymium tungstate, Ba2PrWO6, a double perovskite that has not been synthesized before. Room-temperature (RT) powder X-ray diffraction identified the most probable space group (SG) as monoclinic I2/m, but it was only slightly distorted from the cubic structure. X-ray photoelectron spectroscopy confirmed that the initial (postsynthesis) material contained praseodymium in both 3+ and 4+ charge states. The former (Pr3+) disappeared after exposure to UV light at RT. Photoluminescence studies of Pr3+ revealed that Ba2PrWO6 is an insulator with a band gap exceeding 4.93 eV. Pressure-dependent Raman spectroscopy excluded the possibility of a phase transition up to 20 GPa; however, measurements between 8 and 873 K signified that there might be a change toward the lower symmetry SG below 200 K. Electron paramagnetic resonance spectra revealed the presence of interstitial oxygen which acts as a deep electron trap.


INTRODUCTION
−4 They have the typical ABX 3 perovskite structure but with lattice constants doubled along all the three dimensions, creating a compound of the general A 2 BB′X 6 formula.In this structure, the larger A cation, typically an alkali or alkaline earth metal, is 12-fold coordinated with X anions and occupies the void between corner-sharing BX 6 and B′X 6 octahedra.The alternating B and B′ occupancy in the adjacent octahedra is frequently referred to as a rock-salt-type arrangement.The smaller B, B′ cations are mostly different transition metal (TM) ions or combinations of rare-earth and TM ions, respectively.The X anions are usually halogens or chalcogens, but the most numerous are oxides.Depending on the kind of incorporated ions, DPs can exhibit diverse electrical or magnetic properties.−9 Those containing paramagnetic ions can exhibit long-range magnetic order: 10,11 ranging from antiferromagnetic, 12,13 through ferrimagnetic, 3,14 to ferromagnetic. 3 Spin glass behavior is also frequently encountered. 15,16−29 Recently, we have added a new DP, Ba 2 CeWO 6 (BCW), 30 to the about thousand DP oxides listed in ref 1, which were synthesized up to 2015.Now, we add another new DP to the list�Ba 2 PrWO 6 (BPW).The replacement of Ce by Pr should improve the luminescent properties of the material and make it more suitable for prospective applications as a UV to visible light downconverter.Although the theoretically predicted compound's structure has been reported in the Materials Project database (entry mp-1519191) up to date, there are no published experimental data, neither on the crystal structure nor on other properties, implying that the material has not been synthesized before.
This work focuses on the structural and luminescent properties of Ba 2 PrWO 6 , especially the charge-transfer processes occurring under UV illumination.Various X-ray techniques were applied, such as powder X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), and X-ray absorption spectroscopy (XAS), to establish the space group (SG) and charge states of the constituting ions, as well as eventual impurity phases.The results of complementary for the extra charge, W 18 O 49 was included by controlled preheating of WO 2 .
The initial substrates used for solid-state reaction synthesis were BaCO 3 (99.9%,Strem Chemicals) preheated to remove water, WO 2 (99.9%,Merck), and Pr 6 O 11 (99.99%,Strem Chemicals) taken in a weight ratio of 1.135:0.621:0.49g (per 2 g of the final product) to ensure Ba, Pr, and W cations stoichiometry.However, XRD measurements revealed that the nominally 99.99% pure Pr 6 O 11 phase contained 6% of Pr 3 O 5 .Prior to synthesis, WO 2 was heated in Ar for 5 h at 650 °C.XRD performed immediately afterward identified that in the intermediate product, only 24% of tungsten was in the W 18 O 49 phase (ICDS 05-0393 card).All substrates were then mixed in an agate mortar and compressed to a pellet with a hydraulic press at about 10 MPa.
Ba 2 PrWO 6 was synthesized by a solid-state reaction in a quasi-inert Ar/H 2 (999−995:1−5 mL) gas mixture.This atmosphere was essential because any surplus of oxygen redirected the reaction toward other tungstate phases.The synthesis occurred in three stages.In the first stage, the pellet was heated at 1000 °C for 10 h in a constant-flow, tube furnace under a gas pressure of 0.7 atm in order to release CO 2 .In the next stage, the temperature was increased to 1100 °C, while the duration time and gas pressure remained the same.In the last stage, the material was heated at 1100 °C for 12 h in a chamber pressurized up to ∼200 atm in order to improve the ionic migration and fusion rate. 36Between these stages, all pellets were ground into a fine powder and compressed back into pellets.To avoid carbon contamination, corundum crucibles were used.

RESULTS AND DISCUSSION
3.1.Macro-and SEM Microphotographs.SEM micrographs and photographs of the obtained material are shown in Figure 1.The lighter-colored outer rim of the pine-green BPW pellet was found to contain significantly more impurity phases than the interior; therefore, it was scrapped off and discarded prior to further investigations.The material consists of ca.micron-sized, polygon-shaped crystallite grains embedded in finer powder.The average size of 20 randomly chosen grains was about 855 ± 91 nm (with a spread from 515 to 992 nm).The full width at half-maximum (fwhm) of XRD peaks can give more information on the crystallite size, according to the Debye−Scherrer equation. 37Although for grains larger than 200 nm the method is considered to be less accurate, it is more adequate than estimates based on particular, very small areas visualized in SEM.Based on shape constants (K ∼ 0.88−0.93)for the three dominant peaks at 2Θ = 110.3,85.4, and 69.8 and their fwhm's, the determined average size is 914 ± 7 nm.
When heated in the air, BPW slowly decomposes to an orange powder starting from the surface (Figure S1).The mechanism of oxidation can vary but mainly involves the formation of BaWO 4 , as discussed in the Supporting Information.

Powder XRD.
The powder XRD pattern of the synthesized product is shown in Figure 2a.The best Rietveld refined fit to the experimental pattern was obtained for 95.71%

Inorganic Chemistry
Ba 2 PrWO 6 with the monoclinic I2/m SG, similar to Ba 2 BiYO 6 (ICSD 65555). 38The structure is shown in Figure 2b.The unit cell parameters are a = 6.0219 structure similar to that of SrLa 2 WO 7 (JCPDS 049-0353) and Pr 2 O 3 (0.32%). 39 possible alternative to the I2/m SG is R-3, with the following unit cell parameters: a = b = 6.0219(4)Å, c = 14.7509(8)Å, and V = 463.26(1)Å 3 , for which the error factor is only slightly greater (χ 2 = 42.2 for I2/m and 43.1 for R-3).The refined fit to the same XRD experimental data and the derived structure are presented in Figure S2.For the sake of further discussion, the I2/m SG is chosen; however, to confirm this assignment unambiguously, neutron powder diffraction would be necessary.
The background-corrected Rietveld refinement parameters for the alternative R-3 SG are given in Table S1 while those for I2/m are presented in Table 2.The atomic coordinates and occupancy factors are collected in Table 3 (and Table S2 for R-3).
3.3RT.XPS Data.XPS is a standard technique to determine the atomic content of solids.In contrast to XRD, which detects only crystalline phases, in XPS, all ions are detected, including the ones at interstitial sites or residing in amorphous phases.Another advantage of XPS is the distinction of valence states of the constituting ions.The latter can be, however, treated only as an indication since the measurement is performed under ionizing radiation, which significantly changes the charge-state occupancies.We employed this technique to determine the total atomic content in the synthesis product.The X-ray photoelectron spectrum of the Pr 3d 5/2 and 3d 3/2 spin−orbit split core levels (M 5 and M 4 edges, respectively) is shown in Figure 3a.The rich structure can be deconvoluted into three sets of doublets related to Pr 4+ , denoted as v, v″, and v‴ in the d 5/2 component and u, u″, and u‴ in the d 3/2 component�green lines in Figure 3a.−42 We follow here the peak notation and assignment of Konysheva and Kuznetsov. 42The same peaks are detected in the X-ray absorption spectrum of BPW (Figure S3).
In contrast to praseodymium, the XPS spectrum of tungsten 4f O 2,3 edges shown in Figure 3b contains only two spin−orbit split doublets, denoted by "1" and "2" for the 4f 7/2 and 4f 5/2 core states, respectively.The intense peaks at binding energies (BE) of 33.1 eV (4f 7/2 ) and 35.4 eV (4f 5/2 ), depicted by the blue line, are assigned to W 4+ , while those at 34.6 and 36.9 eV (green line) to W 5+ . 43,44No signal related to W 6+ was detected (expected at BE = 35.6−35.7 eV), 43,44 which indicates that the charge state occurring in the initial substrate was reduced during synthesis, most probably due to electron transfer from some of the Pr 3+ ions.
The oxygen K-edge photoelectron spectrum is shown in Figure 4a.It consists of a very broad, intense peak at BE = 530.2eV and a more than twice narrower, lower intensity peak at 532.6 eV.Since there are different surroundings of oxygen in the BPW lattice depending on the charge states of praseodymium and tungsten ions, we assign the broad peak to the substitutional oxygen and the weaker one to interstitial oxygen (O i ).The presence of O i − was confirmed by EPR experiments (Section 3.4).The barium M 4,5 -edge XPS spectrum presented in Figure 4b shows a strong spin−orbit split 3d 5/2 and 3d 3/2 doublet, located at 783.8 and 795.7 eV, respectively, which originates from the main BPW phase.The weak shoulders located at 780.5 and 793.4 eV most likely stem from the BPWO phase.The peak positions do not differ much from typical values. 45,46he peak positions, fwhm, and determined percentage atomic contents for all XPS-analyzed species are collected in Table 4, together with the atomic fractions normalized to tungsten content.For comparison, the atomic fractions summed for all crystalline phases detected by XRD are also given.The latter data are not much different from those obtained by XPS (also indicating a low contribution of amorphous phases), except that significantly more oxygen is detected by XPS.This supports our assignment of the 532.6 eV O 1s XPS peak to interstitial oxygen.The numbers given in Figures 3 and 4 pertain only to the initial, relative charge distribution under ionizing radiation and cannot accurately determine the true number of constituting ion states, especially considering the influence of the other phenomena, like charge transfer, described in the next section.
3.4.Charge-Transfer Processes.PL and EPR are the easiest experimental techniques to monitor changes in the charge states of constituting ions that occur after near-UV illumination.Especially, the latter technique can provide information on the presence of additional paramagnetic defects, which can take part in charge-transfer processes.In this subsection, we seek answers to the following questions: (i) what is the mechanism of electron release from Pr 3+ ?and (ii)   on which ions, apart from W 5+ , are the released electrons captured?
The PL and PLE spectra collected at RT are shown in Figure 5a.The excitation (for PL) and detection (for PLE) wavelengths are given in the legend.As an excitation source, a xenon lamp was used.Under 252 nm excitation, the luminescence is dominated by sharp lines stemming from the characteristic intra 4f-shell transitions of Pr 3+ , 47,48 superimposed on a weak, broad band attributed tentatively to WO x charge-transfer transitions. 49The energy level diagram of Pr 3+ , together with the assignment of the observed transitions, is presented in Figure 5c.The PL excitation spectrum monitored for the strongest Pr 3+ emission line at 485 nm shows a sharp line at 252 nm, related to intrashell 3 H 4 → 1 S 0 transition, 50 while the spectrum detected for wavelengths between the sharp Pr 3+ lines, e.g., 578 nm, shows a broad absorption band at about 360 nm.Excitation within this range does not lead to Pr 3+ emission.
At RT, the Pr 3+ PL intensity decreases strongly with the irradiation time.Already after 25 min of exposure to UV light, the luminescence is almost completely bleached, as shown in Figure 5d.In contrast, at 10 K, it remains practically stable.The results of LT PL measurements presented in Figure 5b indicate that the bleaching efficiency depends exponentially on the temperature.The deactivation energy determined from the loss of integrated intensity vs inverse temperature is equal to 7.7 meV (see the inset in Figure 5b) and corresponds to the thermal ionization energy of an electron from the 1 S 0 excited state.The LT PL spectrum recorded without a 435 nm cutoff filter is shown in Figure S4.
The EPR spectrum of BPW recorded at 3 K is presented in Figure 6a.Since Pr 3+ ions with 4f 2 electron configuration cannot be detected in our conventional EPR setup, 51 we expected to detect mainly an anisotropic (C 2h local symmetry) powder spectrum of Pr 4+ (4f 1 ) 52,53 ions, similar to that of the isoelectronic Ce 3+ observed previously in BCW, 30 which would be visible only at cryogenic temperatures.Instead, the spectrum is dominated by an isotropic paramagnetic signal that persists up to RT (red line in Figure 6a).The lack of fine structure and the g-factor of 2.12 indicate an acceptor-type defect with spin S = 1/2.The obvious candidate is interstitial oxygen in the −1 charge state (O i − ).Substitutional O s − can be excluded due to lower (C s or C 1 ) site symmetry.After UV irradiation at RT, the signal intensity is drastically reduced (compare red and black lines in Figure 6b), which indicates that a part of the electrons ionized from Pr 3+ is captured by O . The assignment of the defect to interstitial oxygen is further corroborated by the fact that annealing in a reducing atmosphere containing H 2 completely removes the EPR signal, as shown by the blue line in Figure 6b.
The lack of an EPR signal from Pr 4+ ions, despite their high concentration, is probably an effect of antiferromagnetic ordering at LTs since most of the A 2 BB′O 6 perovskites containing a single paramagnetic B site cation are reported to be antiferromagnetic (Vasala and Karppinen 1 and references therein).In BCW, most of the ceria were in the diamagnetic 4+ charge state (4f 0 ); 30 thus, the distance between paramagnetic Ce 3+ ions was too large for superexchange interaction to be effective and the compound remained paramagnetic even at 3 K.
The photoelectron spectra of Pr 3d and W 4f core levels presented in Figure 7 show that after exposure to a broadspectrum UV light source, all of the tungsten and practically all praseodymium ions (except a trace fraction) occur in the 4+ charge state.The fact that both W 5+ and O i − capture electrons thermally ionized from the 1 S 0 excited state of Pr 3+ proves that the charge-transfer process occurs via the conduction band.It   S5); however, it is well known that DFT drastically underestimates band gaps.The situation changes further only for tungsten if the sample is carelessly exposed to a more intense, laser light from UV−C spectrum without any protective atmosphere.Special care must be taken since even O i is prone to react if the sample is luminously over-stimulated�more about this is in Supporting Information Section 3.4 page S8.

Raman and FTIR Spectroscopies.
Raman and FTIR experiments were performed as an attempt to clear up doubts about the proper choice of the BPW SG (I2/m or R-3).The spectra recorded under ambient conditions are compared in Figure 8.The former was recorded on a single-crystalline grain, hoping to minimize the peak widths.The deconvoluted peak positions, together with literature-based mode assignments, are summarized in Table 5.As can be seen, the number of detected Raman peaks exceeds the 12 modes predicted by group theory for perfect I2/m (7A 1g + 5B g ) as well as R-3 (4A g + 4 1 E g + 4 2 E g ) SGs (for the total number of predicted phonon modes and their irreducible representations, see Table S3).This may be due to different Pr 3+/4+ /W 4+,5+ ion pairings and the presence of point defects.Unfortunately, not only XRD patterns but also Raman data cannot unambiguously distinguish between I2/m and R-3 SGs.Also, the FTIR spectrum does not give much support since the signals are quite broad (fwhm ∼ 64 cm −1 and cut off below 400 cm −1 . To shed more light on the proper SG choice, Raman studies at non-ambient conditions were performed, in order to search for possible phase transitions.These usually follow a specific SG sequence, which can indirectly point to the most probable initial SG. 1,60,61Raman spectra recorded at LTs (increasing from 8 K to RT) are presented in Figure 9a, and the temperature dependence of the peak positions is given in Figure 9b.It can be seen that the 8 K spectrum contains 7 peaks more than the RT one.
Four (out of seven) of the additional peaks disappear above 150 K, and their positions change with higher temperature coefficients than those of the other three, which remain detectable until 200 K.The peak positions and temperature coefficients are collected in Table 6.The greater number of observed Raman modes may indicate a phase transition to a group with a lower symmetry (possibly monoclinic P2 1 /n).It is tempting to associate the apparent symmetry lowering either with induced magnetic ordering or with increasing octahedral tilt, 61 however, some of the extra modes are observed under residual Ar pressure at RT, as shown in Figure 10.
RT Raman spectra under compression and decompression cycles are presented in Figure 10a,b, respectively, while the peak positions vs increasing and decreasing pressure are depicted in Figure 10c,d.Typical linear blue shifts of the Raman modes are observed with increasing pressure, but no phase transition occurs up to 20 GPa.The pressure coefficients are given in Table 6.Some peaks are so weak that their pressure dependence cannot be followed in the whole range� this concerns especially the debatable ones related to PrO 2−x defects (see Figure S6).Phase transitions at higher pressures cannot be excluded, but up to 20 GPa, the material is quite stable as hysteresis during pressure release is not higher than 2 GPa, comparable to BCW from our previous work. 30.6.Material Stability at High Temperatures.The stability of BPW at high temperatures is important from the point of view of possible applications, e.g., as a down-converter in solar panels.Therefore, XRD and Raman studies were performed up to 600 °C in both air and inert N 2 gas.The   results of differential scanning calorimetry (DSC), thermogravimetric analysis (TGA), and heat capacity measurements performed in a similar temperature range are shown in Figure S7.
Temperature-dependent Raman spectra and XRD patterns of BPW are presented in Figure 11.When heated in a protective N 2 atmosphere (Figure 11a,b), no significant changes in the spectra were observed up to 873 K.In the Raman spectra (Figure 11a), a barely noticeable red shift can be seen with some intensity increase of PrO 2−x modes above 600 K.The XRD peaks (Figure 11b) also shift slowly with increasing temperature, indicating lattice expansion and changing tilt angle, yet no structural transition occurs.In air, the situation is totally different.Above 500 K, the Raman spectra as well as the XRD patterns visibly change (Figure 11c,d).In Raman spectra, modes related to Ba x WO 3+x and PrO 2−x species (marked by asterisks and hashtags in Figure 11c, respectively) gradually appear, indicating decomposition Asterisks mark modes that could also be associated with PrO 2−x impurities according to theoretical calculations. 59

CONCLUSIONS
The newly synthesized ordered barium praseodymiumtungstate double-perovskite Ba 2 PrWO 6 (BPW) is reported.XRD studies have shown that the crystal structure of BPW (∼96% pure) is better described by the SG (a 0 b − b − ) I2/m with a = 6.0219(3)Å, b = 6.0218(3)Å, c = 8.5167(3) Å, and slightly tilted β angle ∼90.01(0)°(R wp ∼ 17.3%) than by R-3 with a = b = 6.0219(4)Å and c = 14.7509(8)Å (R wp ∼ 17.5%).The greater number of Raman modes obtained at LTs than at RT could point to the possible onset of a transition to a lower symmetry (P2 1 /n) phase.This would favor the I2/m SG assignment, according to group theory considerations, since an R-3�I2/m phase transition would not increase the number of modes (Figure S8 and Table S4).Still, this assignment is not really unambiguous.With increasing temperature, the Raman lines considerably broaden, and some of them can no longer be detected.CIF structure details for both SGs are provided in Supporting Information Table S5.
XPS investigations have shown that the freshly acquired material contains Pr 3+ and Pr 4+ ions as well as W 5+ and W 4+ .The estimated total atomic content makes the material stoichiometrically closer to Ba 2.09 Pr 1.17 WO 7.24 than XRDdetermined Ba 1.96 Pr 1.05 WO 6.05 (almost ideal Ba 2 PrWO 6 ).However, one has to bear in mind that in the former formula, the presence of impurity phases (∼4% BaPr 2 WO 7 and ∼0.3% Pr 2 O 3 ) as well as interstitial oxygen ions (unambiguously detected by EPR) is taken into account.
The Pr 3+ content was shown by XPS to disappear after 25 min of broad-band UV−C illumination (λ ≤ 252 nm) at RT, accompanied by a total change of the W charge state from 5+ to 4+.PL investigations have shown that the bleaching of Pr 3+ emission also occurs under intra 4f-shell excitation due to thermal ionization of the 1 S 0 excited state which lies only 7.7 meV below the BPW conduction band edge.Moreover, EPR studies have shown that electrons ionized from Pr 3+ are also captured by deep interstitial oxygen traps, thus proving that the charge-transfer phenomenon proceeds via the conduction band.
As Raman and XRD spectra show, BPW is stable up to at least 870 K in an inert atmosphere but oxidizes when heated in air above 550 K, decomposing to BaWO 4 and PrO 2−x .This was confirmed by DSC and TGA measurements.Additionally, considering the presence of interstitial oxygen which can also contribute to gradual Pr and W-oxidation (while exposed to potent ionizing radiation), it makes the material hardly suitable for most applications without a protective coating.

Figure 1 .
Figure 1.Upper panel: SEM micrographs of BPW were obtained for two different magnifications.The scale is given in the graphs.Lower panel: photographs of the obtained pellets: whole (left) and cut through (right).

Figure 2 .
Figure 2. (a) XRD pattern of the synthesis product: dots�experimental data and black line�Rietveld refined fit assuming ∼96% of the Ba 2 PrWO 6 I2/m phase, ∼4% of BaPr 2 WO 7 (BPWO), and ∼0.3% of Pr 2 O 3 .Blue line: difference between experimental and calculated data.Green bars denote the Bragg positions for the three phases.(b) Projection of the unit cell on the ab plane (left) and the bc plane (right).Green, yellow, gray, and red spheres denote Ba, Pr, W, and O ions, respectively.

Figure 3 .
Figure 3. X-ray photoelectron spectra of the Pr 3d (a) and W 4f (b) core states.Red lines are fits to experimental data (black lines) taking into account two charge states for each ion (blue and green lines) and the background (cyan line).

Figure 5 .
Figure 5. (a) PL and PLE spectra of BPW at RT.The excitation (for PL) and detection (for PLE) wavelengths are given in the legend.The spectra are shifted vertically for clarity.(b) Selected temperature-dependent spectra under 232 nm excitation of the OPO laser.The inset shows the Arrhenius plot of the integrated PL intensity loss, I(10 K) − I(T), normalized to the initial value at 10 K. (c) Energy level diagram of Pr 3+ .(d) RT PL after 25 min exposure to UV illumination (λ ≤ 252 nm).

Figure 6 .
Figure 6.(a) EPR spectra of BPW at 3 and 296 K (black and red lines, respectively).(b) RT EPR spectra before UV irradiation (red line), after 25 min of UV irradiation (black line), and after subsequent annealing in reducing Ar/H 2 atmosphere (blue line).

Figure 7 .
Figure 7. XPS spectra of Pr 3d (a) and W 4f (b) core states recorded after 25 min of exposure to UV illumination.

Figure 8 .
Figure 8. Raman (black line, left axis) and FTIR (red line, right axis) spectra of BPW recorded under ambient conditions.No signals were detected above 1100 cm −1 (up to 4000 cm −1 ).

Figure 10 .
Figure 10.Raman spectra collected under HP by using consecutive compression (a) and decompression (b) cycles.Peak positions (symbols) vs increasing (c) and decreasing (d) pressure.Solid lines in (c) are linear fits and in (d) are guides for the eye.

Figure 11 .
Figure 11.Temperature-dependent Raman spectra (a,c) and powder XRD (b,d) patterns recorded when heated in an inert N 2 atmosphere (a,b) and in air (c,d).

Figure 12 .
Figure 12.Temperature dependencies of lattice parameters: (a) a and b (black and red symbols, respectively), (b) c, and (c) β tilt angle.(d) Temperature dependence of the unit cell volume.Solid lines are linear fits.

Table 1 .
Classic Goldschmidt (t) and Modified (τ) Tolerance Factors Predicting the Existence of Ba 2 PrWO 6 Depending on the Pr and W Charge States a

Table 2 .
Rietveld Reliability Factors for the Diffractogram Are Given in Figure2a

Table 3 .
Atomic Coordinates and Occupancy Factors for the I2/m SG

Table 4 .
Summary of XPS Investigations of Ionic Core Levels of BPW: Peak Positions, FWHM, Percentage Atomic Contents, and Ratios (Normalized to W) Compared to Those Determined from XRD for Detected Crystalline Phases allows us, moreover, to place the energy level of the 3 H 4 ground state of Pr 3+ about 4.93 eV below the minimum of the BPW conduction band.This value is considerably larger than the band gap obtained from density functional theory (DFT) calculations (Figure

Table 5 .
Experimental Raman and FTIR Data Obtained at Ambient Conditions, with Literature-Based Phonon Assignments 30,54−58a -site lattice translations of mixed B g and 1 E g / 2 E g origin 435 in-plane and out-of-plane bending MO 6 bands of B u /E u origin down to 100 cm −1 99.5 514.5 ν sym stretch of MO 6 octahedra of A u origin 163.1 639 ν asym stretch of MO 6 octahedra of A u origin 179.8 in-plane σ ρ and out-of-plane σ τ bending in MO 6 octahedra of mixed B g and 1 E g / 2 E g origin sc and out-of-plane σ ω bending in MO 6 of mixed B g and 1 E g origin asym and ν sym oxygen stretches of A 1g or A g symmetry for various Wyckoff sites not only partially occupied but also hosting B-site Pr/W ions with different charges a