A. Syntheses

We prepared four types of powder specimens with different starting compositions of [Li:Eu:Ta] = [0.965:0.035:1] (sample LETO), [Li:Eu:Ta:Ti] = [0.974:0.026:0.75:0.25] (sample LETTO), [Li:Sm:Ta:Ti] = [0.969:0.031:0.89:0.11] (sample LSTTO), and [Li:Sm:Mg:Ta:Ti] = [0.954:0.030:0.016:0.89:0.11] (sample LSMTTO) in atomic ratio from the chemicals of Li₂CO₃, Eu₂O₃, Sm₂O₃, MgO, Ta₂O₅, and TiO₂. Each of the well-mixed chemicals was heated at 1273 K for 3 h, and then successively at 1423 K for 24 h (samples LETO and LETTO) and for 15 h (samples LSTTO and LSMTTO), and finally cooled to ambient temperature. We finely ground the densely sintered pellets and obtained the fine powder specimens suitable for XRPD. The former two samples (LETO and LSTTO) were prepared under exactly the same experimental conditions as those of Li(Ta_1−x Ti _x )O_3−x/2:Eu³⁺ phosphors examined for PL properties in a previous study (Nakano et al., Reference Nakano, Ozono, Hayashi and Fujihara2012).

B. Characterization

The XRPD intensities in the 2θ range of 10.0°–149.0° (CuKα ₁) with 15 559 total data points were collected on a diffractometer in the Bragg–Brentano geometry (X'Pert PRO Alpha-1, PANalytical B.V., Almelo, The Netherlands). The X-ray generator was operated at 45 kV and 40 mA. We used a computer program VESTA (Momma and Izumi, Reference Momma and Izumi2011) to visualize the structural models, equidensity isosurfaces of EDDs and 2D EDD maps. Distortion parameters for the coordination polyhedra were determined using a computer program IVTON (Balic-Zunic and Vickovic, Reference Balic-Zunic and Vickovic1996). Excitation and emission spectra were obtained for the (Li_1−y Mg _y )(Ta_0.89Ti_0.11)O_{2.945 +  y/2}:Sm³⁺ phosphors by a PL spectrometer (model FP-6500, JASCO International Co., Ltd., Tokyo, Japan). We measured the PL intensities on the sintered sample surface to eliminate the effects of powder particle size and degree of crystallinity.

A. Crystal structure refinements and electron-density distributions

The sample LETO was composed of both Eu³⁺-doped lithium tantalate and a small amount of EuTaO₄ (Krylov and Strelina, Reference Krylov and Strelina1963). We successfully indexed the XRPD peaks, corresponding to the doped lithium tantalate, with a hexagonal unit cell. Initial structural parameters were taken from those of (Li_0.977Eu³⁺ _0.023)(Ta_0.89Ti_0.11)O_2.968 (space group R3c) determined by Uchida et al. (Reference Uchida, Suehiro, Asaka, Nakano and Fukuda2013). There is one (Li, Eu) site and one Ta site, both of which are located at the Wyckoff position 6a, and one O site at 18b in the hexagonal unit cell (Z = 6). We refined the structural parameters of all atoms using the Rietveld method on a computer program RIETAN-FP (Izumi and Momma, Reference Izumi and Momma2007). We quantitatively determined the phase composition of the sample using the phase-analysis method based on Brindley's procedure (Brindley, Reference Brindley1945), the subroutine of which was implemented in the program RIETAN-FP. The phase composition was, under the condition of effective particle radii being 5.00 µm, found to be 99.90 mol% Eu³⁺-doped lithium tantalate and 0.10 mol% EuTaO₄ (Table I). We determined the chemical formula of the former by subtracting the EuTaO₄ component from the bulk chemical composition. Because the bulk atomic ratio of sample LETO was [Li:Eu:Ta] = [0.965:0.035:1], the chemical formula should be (Li_0.925Eu³⁺ _0.025)TaO₃ (x = 0). The Rietveld refinement result was satisfactory due to the relatively low reliability (R) indices (Young, Reference Young and Young1993) of R _wp = 10.14%, S( = R _wp/R _e) = 1.29%, R _p = 7.24%, R _B = 2.84%, and R_F  = 1.40%. We summarized the crystal data in Table II, and the final atomic positional parameters and isotropic atomic displacement parameters (ADPs) in Table III, where the figures in parentheses of these tables indicate estimated standard uncertainties.

We visualized the 3D EDDs with 108 × 108 × 276 voxels in the unit cell, the spatial resolution of which is ~0.005 nm, using the MPF method on the computer programs Dysnomia (Momma et al., Reference Momma, Ikeda, Belik and Izumi2013) and RIETAN-FP to confirm the validity of the structural model (Figure S1(a) in Supplemental Data). The one REMEDY cycle further decreased the R _B- and R_F -indices to 1.63 and 0.80%, respectively (R _wp = 10.00%, S = 1.27, and R _p = 6.89%). The reduction of these R-indices, which must be induced by slight improvements of EDDs, implies that the EDDs more clearly demonstrate the structural details as compared with the ball-and-stick model in Figure 2(a). The XRPD patterns that are obtained by observation and calculation for the final MPF, together with their difference, are plotted in Figure S2(a) (see Supplemental Data).

Figure 2. (Color online) (a) Part of the crystal structure viewed along [110] of (Li_0.925Eu³⁺ _0.025)TaO₃ in sample LETO. Yellow and magenta bicolor balls are for Li (yellow) and Eu (magenta) sites. Blue balls are for Ta sites. (b) Part of the crystal structure viewed along [110] of (Li_0.968Eu³⁺ _0.032)(Ta_0.81Ti_0.19)O_2.937 in sample LETTO. Yellow and magenta bicolor balls are for Li (yellow) and Eu (magenta) sites. Because the occupancy of the oxygen site is less than unity, the O atoms are depicted as red circle graphs for occupancies. Blue and cyan bicolor balls are for Ta (blue) and Ti (cyan) sites. (c) Part of the crystal structure viewed along [110] of (Li_0.967Sm³⁺ _0.033)(Ta_0.89Ti_0.11)O_2.978 in sample LSTTO. Yellow and magenta bicolor balls are for Li (yellow) and Sm (magenta) sites. Because the occupancy of the oxygen site is less than unity, the O atoms are depicted as red circle graphs for occupancies. Blue and cyan bicolor balls are for Ta (blue) and Ti (cyan) sites. (d) Part of the crystal structure viewed along [110] of (Li_0.950Sm³⁺ _0.033Mg_0.017)(Ta_0.89Ti_0.11)O_2.987 in sample LSMTTO. Yellow and magenta bicolor balls are for Li (yellow) and (Sm, Mg) (magenta) sites. Because the occupancy of the oxygen site is less than unity, the O atoms are depicted as red circle graphs for occupancies. Blue and cyan bicolor balls are for Ta (blue) and Ti (cyan) sites.

The other samples of LETTO, LSTTO, and LSMTTO were composed of doped lithium titanotantalate and small amounts of impurities of Li₃TaO₄ (Agulyanskii et al., Reference Agulyanskii, Bessonova, Kuznetsov and Kalinnikov1986) and/or Li₄Ti₅O₁₂ (Leonidov et al., Reference Leonidov, Leonidova, Perelyaeva, Samigullina, Kovyazina and Patrakeev2003). The analytical processes for these samples were very similar to that of LETO aforementioned. We determined the phase compositions (Table I) and subsequently determined the chemical formulae of doped lithium titanotantalate compounds from calculation by subtracting the impurity components from the bulk chemical compositions. We obtained the satisfactory R indices for all of the Rietveld refinements (Table SI in Supplemental Data), the refined structural models of which are shown in Figures S1(b)–S1(d) (see Supplemental Data). The crystal data of doped lithium titanotantalate compounds are given in Tables IV, V and VI, and the final atomic positional parameters and isotropic ADPs are given in Tables VII, VIII and IX, where the figures in parentheses of these tables indicate estimated standard uncertainties. We applied the MPF method and further decreased the R indices after two REMEDY cycles (Table SI in Supplemental Data). Observed, calculated, and difference XRPD patterns for the final MPF are plotted in Figures S2(b)–S2(d) (see Supplemental Data).

Figures 2(a)–2(d) show parts of the refined crystal structures of doped lithium tantalate phosphors, the chemical formulas of which are (Li_0.925Eu³⁺ _0.025)TaO₃ (x = 0) in sample LETO, (Li_0.968Eu³⁺ _0.032)(Ta_0.81Ti_0.19)O_2.937 (x = 0.19) in sample LETTO, (Li_0.967Sm³⁺ _0.033)(Ta_0.89Ti_0.11)O_2.978 (y = 0) in sample LSTTO, and (Li_0.950Sm³⁺ _0.033Mg_0.017)(Ta_0.89Ti_0.11)O_2.987 (y = 0.017) in sample LSMTTO. We confirmed that all of them are isostructural with (Li_0.977Eu³⁺ _0.023)(Ta_0.89Ti_0.11)O_2.968 (x = 0.11) (Uchida et al., Reference Uchida, Suehiro, Asaka, Nakano and Fukuda2013). The individual equidensity isosurfaces of 3D EDDs were in accord with the arrangements of atoms. The 2D EDD maps show that the positions of (Li, Eu, Sm, Mg) and (Ta, Ti) sites are successfully disclosed by the EDDs (Figure 3). The EDD voxel data have several local maximum, which correspond to the coordinates of atoms. We determined the amounts of deviations between the maxima positions and corresponding atomic coordinates to find that they were necessarily less than 0.003 nm, which is within the resolution limit of the 3D EDDs. Thus, we concluded that the refined structural models would satisfactorily represent the corresponding crystal structures.

Figure 3. (Color online) (a) Bird's eye view of electron densities determined by MPF of the (Li, Eu) and Ta atoms on the (110) plane of (Li_0.925Eu³⁺ _0.025)TaO₃ in sample LETO. (b) Bird's eye view of electron densities determined by MPF of the (Li, Eu) and (Ta, Ti) atoms on the (110) plane of (Li_0.968Eu³⁺ _0.032)(Ta_0.81Ti_0.19)O_2.937 in sample LETTO. (c) Bird's eye view of electron densities determined by MPF of the (Li, Sm) and (Ta, Ti) atoms on the (110) plane of (Li_0.967Sm³⁺ _0.033)(Ta_0.89Ti_0.11)O_2.978 in sample LSTTO. (d) Bird's eye view of electron densities determined by MPF of the (Li, Sm, Mg) and (Ta, Ti) atoms on the (110) plane of (Li_0.950Sm³⁺ _0.033Mg_0.017)(Ta_0.89Ti_0.11)O_2.987 in sample LSMTTO.

B. Structure comparison and photoluminescence property

The coordination elements of each structure are the [(Ta, Ti)O₆] octahedron and [(Li, Eu, Sm, Mg)O₁₂] polyhedron. Selected interatomic distances and their standard deviations are listed in Tables SII–SV (see Supplemental Data). In comparing the polyhedral distortion parameters of [(Ta, Ti)O₆] octahedra (Table X) among the four isomorphous structures, together with those of (Li_0.977Eu³⁺ _0.023)(Ta_0.89Ti_0.11)O_2.968 (Uchida et al., Reference Uchida, Suehiro, Asaka, Nakano and Fukuda2013), we found that the V _S/V _P-values (i.e., the ratios of the volumes of the circumscribed sphere and the polyhedron) ranged between 3.129 and 3.157, and hence they were close to the relevant value of regular octahedron ( = 3.1416) (Makovicky and Balic-Zunic, Reference Makovicky and Balic-Zunic1998). Thus, the [(Ta, Ti)O₆] polyhedra were only slightly distorted from the regular octahedra. The values for distortion parameters (Δ r _S, V _S, σ, and V _P) were almost the same among the five phosphors, hence we concluded that the [(Ta, Ti)O₆] octahedra were all comparable with one another. With [(Li, Eu, Sm, Mg)O₁₂] polyhedra, the centroid-to-(Li, Eu, Sm, Mg) distances (Δ-values) widely varied from 0.004 to 0.047 among the five compounds, although the other polyhedral parameter values (r _S, V _S, σ, and V _P) were almost unchanged (Table X).

Here, we compare the eccentricity vectors Δ , which are defined from the centroids to the (Li, Eu, Sm, Mg) sites of the [(Li, Eu, Sm, Mg)O₁₂] polyhedra, to indicate more clearly the features of the crystal structures of Li(Ta_1−x Ti _x )O_3−x/2:Eu³⁺ and (Li_1−y Mg _y )(Ta_0.89Ti_0.11)O_{2.945 +  y/2}:Sm³⁺ by the directions of Δ as well as their magnitudes | Δ | (Uchida et al., Reference Uchida, Suehiro, Asaka, Nakano and Fukuda2013). With (Li_0.977Eu³⁺ _0.023)(Ta_0.89Ti_0.11)O_2.968 (x = 0.11) (Figure 1), the eccentricity vector Δ is in the [ $00\overline 1 $ ] direction and | Δ | = 0.047 nm. On the other hand, the Δ vectors for (Li_0.925Eu³⁺ _0.025)TaO₃ (x = 0) (Figure 2(a)) and (Li_0.968Eu³⁺ _0.032)(Ta_0.81Ti_0.19)O_2.937 (x = 0.19) (Figure 2(b)) are in the [001] direction with the smaller magnitudes of 0.019 and 0.004 nm, respectively. Thus, the marked difference among the three phosphors was clarified as the degree of eccentricity along the c-axis of the (Li, Eu) position in [(Li, Eu)O₁₂]. Because the concentrations of Eu in the (Li, Eu) sites are relatively low and nearly the same among the three compounds with x = 0, 0.11, and 0.19, the replacement of Ti for Ta would mainly induce the displacement of the (Li, Eu) site along the c-axis, without changing the outer shapes of both [(Li, Eu)O₁₂] and [(Ta, Ti)O₆] polyhedra. The PL intensities determined by Nakano et al. (Reference Nakano, Ozono, Hayashi and Fujihara2012) were highest for the Li(Ta_1−x Ti _x )O_3−x/2:Eu³⁺ phosphor with x = 0.11, followed by those with x = 0.19 and x = 0 (Figure 4). This strongly suggests that the displacement of the Eu³⁺ position from the centroid of [(Li, Eu)O₁₂] polyhedra contributes to the highly enhanced hypersensitive ⁵D₀–⁷F₂ transition in Eu³⁺ when excited by near-UV light (Nakano et al., Reference Nakano, Ozono, Hayashi and Fujihara2012). The slight change of coordination environment of Eu³⁺ ion would markedly affect the f-f transition process, which would eventually cause an increase of emission intensity.

Figure 4. (Color online) Relationship between normalized photoluminescence (PL) intensity and centroid-to-cation distance (Δ-value) of [(Li, Eu)O₁₂] polyhedra with Ti/(Ti + Ta) ratio (=x) for Li(Ta_1−x Ti _x )O_3−x/2:Eu³⁺. The PL intensity data are from Nakano et al. (Reference Nakano, Ozono, Hayashi and Fujihara2012). The Δ-value with x = 0.11 is from Uchida et al. (Reference Uchida, Suehiro, Asaka, Nakano and Fukuda2013).

With (Li_0.967Sm³⁺ _0.033)(Ta_0.89Ti_0.11)O_2.978 (y = 0), the eccentricity vector Δ is in the [001] direction with | Δ | = 0.006 nm, whereas the Δ vector of (Li_0.950Sm³⁺ _0.033Mg_0.017)(Ta_0.89Ti_0.11)O_2.987 (y = 0.017) is in the opposite direction with the larger | Δ |-value of 0.046 nm (Figures 2(c) and 2(d)). Accordingly, the prominent difference between the two compounds was clarified as the degree of eccentricity along the c-axis of the (Li, Sm, Mg) position in [(Li, Sm, Mg)O₁₂]. The concentrations of Sm³⁺ in the Li sites of (Li_1−y Mg _y )(Ta_0.89Ti_0.11)O_{2.945 +  y/2}:Sm³⁺ are exactly the same between the two compounds with y = 0 and y = 0.017, hence the replacement of Mg for Li would mainly induce the displacement of the (Li, Sm, Mg) site along the c-axis, keeping both [(Li, Sm, Mg)O₁₂] and [(Ta, Ti)O₆] polyhedra nearly undistorted. Each of the PL spectra in the range 550–680 nm, recorded at the excitation wavelength 410 nm, is caused by the ⁴G_5/2 → ⁶H_J transitions of Sm³⁺ ion (Sakirzanovas et al., Reference Sakirzanovas, Katelnikovas, Bettentrup, Kareiva and Jüstel2011; Dillip et al., Reference Dillip, Kumar, Raju and Dhoble2013) and consists of three groups of narrow band emission in the range of 550–585 (J = 5/2), 585–625 (J = 7/2), and 630–680 (J = 9/2) (Figure 5). Among them, the most intense band is located at ~607 nm because of the ⁴G_5/2 → ⁶H_7/2 transition. The relevant PL intensities were ~1.6 times higher for (Li_0.950Sm³⁺ _0.033Mg_0.017)(Ta_0.89Ti_0.11)O_2.987 (y = 0.017) than for (Li_0.967Sm³⁺ _0.033)(Ta_0.89Ti_0.11)O_2.978 (y = 0), which strongly suggests that the displacement of the Sm³⁺ position from the centroid of [(Li, Sm, Mg)O₁₂] polyhedra contributes to the higher PL efficiency.

Figure 5. (Color online) Photoluminescence (PL) emission spectra of (Li_0.967Sm³⁺ _0.033)(Ta_0.89Ti_0.11)O_2.978 in sample LSTTO and (Li_0.950Sm³⁺ _0.033Mg_0.017)(Ta_0.89Ti_0.11)O_2.987 in sample LSMTTO, showing the higher PL intensity for the latter phosphor (λ _ex = 410 nm).

We have confirmed that the crystal structures of doped lithium tantalate phosphors are flexible with respect to the substitution of Eu³⁺, Sm³⁺, and Mg for Li and that of Ti for Ta. A more detailed discussion on the PL properties of Li(Ta_1−x Ti _x )O_3−x/2:Eu³⁺ and (Li_1−y Mg _y )(Ta_0.89Ti_0.11)O_{2.945 +  y/2}:Sm³⁺ would be made based on the comparison among the electronic states of these compounds, which might be determined by, for example, density functional theory. The reliable atomic coordinates, which are essential for the theoretical calculations, are available at the moment for the doped lithium tantalate phosphors.