A synchrotron X-ray study of triclinic LiCa₂Nb₃O₁₀ with perovskite-type slabs

The triclinic superstructure of a small crystal of LiCa₂Nb₃O₁₀, lithium dicalcium triniobium decaoxide, has been investigated by synchrotron X-ray diffraction. The unit cell is an almost rectangular parallelepiped, although there is a 0.245° offset from orthogonality for β. The structure essentially belongs to a homologous series of Li[Na_n−3Ca₂Nb_nO_3n+1] with n = 3, where the moiety in square brackets has a perovskite-type slab structure. The superstructure has a doubled unit-cell volume with respect to the tetragonal aristotype. The NbO₆ octahedra are rotated about axes parallel to [110] by approximately 10°. Adjacent slabs are connected by Li atoms and are geo­metrically related by 4₂ pseudosymmetry lying parallel to c. There are twice as many sites as Li atoms, providing a variation of population at these Li sites.

S1. Comment

Recently, the layered perovskite-type compounds, ACa₂Nb₃O₁₀, with A = Li, Na, K, Rb or Cs, have attracted attention because they possess properties such as superconductivity (Fukuoka et al., 1997; Takano et al., 1997), luminescence (Bizeto et al., 2000) and photocatalytic behaviour for overall water splitting (Takata et al., 1998). These layered perovskites belong to a homologous series of general formula A[Na_n-3Ca₂Nb_nO_3n+1], with n = 3. In this notation, the moiety in square brackets denotes the perovskite-type slab, with n corresponding to the number of NbO₆ octahedra across the slab. The A atom outside the brackets is located at the interlayer and joins adjacent slabs.

Although the aristotype structure of ACa₂Nb₃O₁₀ has tetragonal symmetry, detailed analyses based on single-crystal X-ray diffraction suggest lower symmetries. For example, CsCa₂Nb₃O₁₀ is orthorhombic, space group Pnam (Dion et al., 1984), and KCa₂Nb₃O₁₀ is orthorhombic, space group Cmcm (Fukuoka et al., 2000). A single-crystal structure analysis of LiCa₂Nb₃O₁₀, (I), has not been undertaken prior to this work, mainly due to the small size of the crystals, the presence of twinning and poor crystallinity. However, in the present study, several crystals grown in Li₂SO₄ flux were found to be tolerable for single-crystal diffraction using synchrotron radiation. The triclinic structure of LiCa₂Nb₃O₁₀ has thus been determined at the Photon Factory, High Energy Accelerator Research Organization, Tsukuba, and the results are presented here. \sch

In the structure of (I), there are two crystallographically independent slabs, composed of Nb, Ca and O atoms, labelled with even and odd numbers, respectively. These slabs are stacked alternately along c, as shown in Figs. 1 and 2. The Ca atom has twelvefold coordination. The corner-linked NbO₆ octahedra are rotated relative to each other by approximately 10°. The dominant component of the rotation occurs about axes parallel to [110]. There is also a trace rotation component about [001] for the Nb1O₆, Nb2O₆, Nb5O₆ and Nb6O₆ octahedra. The dominant rotation axes align parallel within a slab, while they are flipped by 90° in adjacent slabs, as shown in Fig. 2. As a result, the slabs are geometrically related by 4₂ pseudo symmetry along c, if the 0.245° offset from orthogonality is neglected for β. This P1 structure, in which two similar layers are rotated relative to each other by 90° around their normal, is also found in K₂Cr₂O₇ (Brandon & Brown, 1968).

Displacement of the Nb atom from the centre of the coordination octahedron along c is notable for Nb3, Nb4, Nb7 and Nb8 on the outer side of the slab, while none is observed for Nb1, Nb2, Nb5 and Nb6 at the inversion centres. The electric dipoles induced by off-centre displacements of the former group of Nb atoms cancel each other within each slab.

Adjacent slabs are joined together by Li atoms located at the interlayer, as shown in Fig. 3. They have essentially tetrahedral coordination, although the rotation of NbO₆ octahedra appends to Li1 and Li2 another Li—O bond which is shorter than two of the tetrahedral ones. These extra Li—O bonds exist along directions nearly parallel to c. The O atoms surrounding the Li atom tetrahedrally occupy diagonal corners of a pseudo cube, with dimensions of approximately 2.74 Å along a and b, and 1.48 Å along c. These cubes are significantly compressed along c, by 1.26 Å. This is partly due to the rotation of the NbO₆ octahedra in the perovskite-type slab.

There are twice as many sites as Li atoms, providing a variation of population at these Li sites. The fivefold Li1 and Li2 sites have populations approximately twice as large as the fourfold Li3 and Li4 sites. The excess number of Li sites may enable Li diffusion along the interlayers in the crystal. From symmetry constraints, there is only one crystallographically independent interlayer structure composed of Li atoms in the unit cell.

Depending on which Li sites are occupied, the O atoms near the interlayer could be displaced one way or the other from their refined average positions. This would result in local changes of the rotation angles of the NbO₆ octahedra, an effect that would be transmitted across the slab or at least to the octahedron in the centre of the slab. The large anisotropy of the atomic displacement parameters of several O atoms therefore probably represents static disorder rather than thermal motion. The 2:1 ratio in the occupancies of the Li sites is suggestive of a possible long-range modulation with a tripled c axis periodicity. However, the present X-ray study using both area and counter detectors revealed no such modulation along c. It is possible that a longer annealing time would allow such a modulation to develop in the present crystal.

Dion et al. (1981) reported a tetragonal unit cell of a = 7.720 (7) and c = 28.331 (3) Å for (I). If a tetragonal aristotype with unit cell vectors of a₀, b₀ and c₀ is considered, the structure of Dion et al. (1981) can be expressed as a ≈ 2a₀, b ≈ 2b₀ and c ≈ c₀, and the present structure as a ≈ a₀ + b₀, b ≈ a₀ - b₀ and c ≈ c₀. The unit-cell volume of the present structure is twice as large as the aristotype and half as large as that of Dion et al. No superstructure reflections indicative of the latter structure were observed in the present study.

It is notable that the c length of the present crystal is approximately 7% (1.778 Å) shorter than that reported for the crystal of Dion et al. (1981). This suggests the possible existence of two phases with the same or slightly different compositions. A similar phenomenon is also observed for flux-grown Na[NaCa₂Nb₄O₁₃] crystals belonging to the homologous series with n = 4 (Chiba et al., 1999).

The present crystal of (I) has triclinic P1 symmetry with monoclinic cell dimensions. Such apparent symmetry enhancement is also found in another layered perovskite-type compound, Bi₄Ti₃O₁₂, which has a monoclinic symmetry with orthorhombic cell dimensions (Rae et al., 1990).

S2. Experimental

Single crystals of (I) were grown by the flux method using Li₂SO₄ as flux. Reagent grade chemicals of CaCO₃, Nb₂O₅ and Li₂SO₄ were weighed and mixed together to form a nominal composition of 1 mol% LiCa₂Nb₃O₁₀. A platinum crucible containing a mixture of approximately 5 g was heated to 1373 K at the rate of 45 K h^-1 in an air atmosphere, kept at 1373 K for 2 h, cooled to 723 K at 5 K h^-1 and then discharged. The flux was rinsed out with water at 353 K. Thin crystals with a maximum size of 0.2 × 0.2 × 0.1 mm were identified as LiCa₂Nb₃O₁₀ by chemical analysis using an energy dispersive X-ray spectrometer and X-ray diffraction. An additional product with hexagonal faces was identified to be Li₃NbO₄.

S3. Refinement

A preliminary examination of the crystals was carried out using an Imaging Plate diffractometer (Rigaku Rapid). A crystal of approximately 60 × 46 × 25 µm along the [110], [110] and [001] directions, respectively, was chosen from a number of crystal fragments. No clear indication of twinning was observed for the crystal. Monoclinic distortion was suggested from the preliminary examination and confirmed by further study using synchrotron radiation. Diffraction data were obtained using a horizontal-type four-circle diffractometer at beamline 14 A, Photon Factory (Satow & Iitaka, 1989). Vertically polarized X-rays of λ = 0.75008 (2) Å were obtained using a Si double crystal monochromator. The wavelength was calibrated using a spherically ground Si standard crystal. Although most reflections were quite weak at high angles about 2θ > 65°, several parent structure reflections and their equivalents still had sharp profiles in the region. Cell dimensions were thus determined using 15 of those reflections. An eight channel avalanche photodiode detector was used for photon counting (Kishimoto et al., 1998). The aperture of the detector was approximately 6 mrad. The detection efficiency was approximately 75% at this wavelength. Since the dynamic range of the detector exceeds 10⁸ c.p.s., no attenuator nor absorber was used for data collection. The recorded maximum count rate for the present crystal was approximately 1.4 × 10⁶ c.p.s. for the 020 reflection. The half width at half maximum was approximately 0.13° in ω for strong reflections at low angles. Absorption coefficients were taken from Sasaki (1990). The unit cell has a monoclinic shape. As listed in the Table, the determination of cell dimensions assuming a triclinic cell revealed a significant though small offset for β, but no significant offsets from orthogonality for α and γ. The intensity statistics of the parity groups of the measured reflections suggested that there were no systematic absences for hkl except for 0k0 with k odd. Accordingly the possibility of glide planes or centred lattices was discarded. For example, the existence of five reflections with I > 3σ(I) for h0l with h odd eliminated the a glide plane perpendicular to b in the monoclinic cell with the b axis unique. Therefore, the crystal was assumed to have monoclinic P12₁1 symmetry until the final stage of refinement. After the structure was solved with P12₁1, the possible existence of extra pseudosymmetries, for example, a fourfold axis along c and mirror planes perpendicular to a or b, was suggested by the programs BUNYIP (Hester & Hall, 1996) and PLATON (Spek, 2001). Most of these extra symmetries are the remains of the tetragonal aristotype with I4/mmm symmetry, and are incompatible with the rotation scheme of corner-linked NbO₆ octahedra in the perovskite-type slab. These symmetries disappeared when the match tolerance was reduced to 0.2 Å in BUNYIP. However, it was difficult to eliminate the possibility of inversion centres completely, because only the Li and several O atom positions were displaced more than three s.u.s from the theoretical centrosymmetric positions. Since the monoclinic space groups containing mirrors can be excluded by the geometrical problems they would involve for octahedral rotation, there are no monoclinic centrosymmetric space groups in accord with the observed extinction rules for reflections. Therefore, triclinic P1 symmetry was tested finally and the results were compared with those of P12₁1. The R factor assuming P12₁1 was 0.062 for 187 parameters and 2037 independent reflections, employing anisotropic atomic displacement parameters (ADPs) for Nb and Ca, and isotropic ADPs for O and Li atoms. The corresponding R factor assuming P1 was 0.073 for 194 parameters and 3750 independent reflections. A trial to refine the Li population parameters jointly with the other parameters was unsuccessful in the P12₁1 model. Therefore, the Li populations were fixed at values which gave moderate isotropic ADPs. No such problems occurred in the P1 model. The P12₁1 refinement gave an R factor of 0.042 for 291 parameters using anisotropic ADPs for Nb, Ca and O atoms, and isotropic ADPs for Li. Although the R factor was smaller than that obtained for P1 as given in the Table, it was difficult to ascertain the plausibility of the P12₁1 model because of the following reservations: (1) the apparent superiority of the R factor is considerably reduced when one considers that the number of reflections used is almost halved for P12₁1 while the number of parameters is approximately the same; (2) the anisotropic ADPs of several O atoms converged at non-positive definite values; and (3) population parameters of Li did not converge when refined together with the Li isotropic ADPs. On the other hand, the P1 refinement converged smoothly, employing anisotropic ADPs for Nb, Ca and O atoms and isotropic ADPs for Li, refined together with the population parameters of Li. Although the O12 atom ADP is a prolate spheroid with a max/min ratio of 4.2, the ADPs for all O atoms were much improved compared with the results obtained for the P12₁1 model. Thus, the space group P1 was adopted for the present crystal. The major reason for the removal of the twofold screw symmetries lies in the interlayer structure, i.e. the positional shifts of Li atoms and their distribution among the available sites. In the final difference Fourier maps, a negative difference electron density of -4.03 e Å^-3 was found to be 0.24 Å away from Ca2.

Computing details

Data collection: DIFF14A (Vaalsta & Hester, 1997); cell refinement: Xtal3.4 (Hall et al., 1995); data reduction: Xtal3.4; program(s) used to solve structure: SHELXS97 (Sheldrick, 1997); program(s) used to refine structure: SHELXL97 (Sheldrick, 1997); molecular graphics: ORTEP (Johnson, 1970) and ATOMS (Dowty, 1995).

Figures

	Fig. 1. The structure of (I) with displacement ellipsoids drawn at the 90% probability level.
	Fig. 2. The polyhedral view of (I) along [110]. Displacement ellipsoids of Li and Ca are drawn at the 90% probability level.
	Fig. 3. The interlayer moiety near the plane at z = 1/4. Displacement ellipsoids of Li and Nb are drawn at the 90% probability level.

(I)

Crystal data

Ca2LiNb3O10	Z = 4
Mr = 525.83	F(000) = 984
Triclinic, P1	Dx = 4.379 Mg m^−3
a = 5.4809 (3) Å	Synchrotron radiation, λ = 0.75008 (2) Å
b = 5.4804 (3) Å	Cell parameters from 24 reflections
c = 26.5533 (16) Å	θ = 32.5–50.7°
α = 89.999 (4)°	µ = 5.57 mm^−1
β = 90.245 (4)°	T = 293 K
γ = 89.999 (5)°	Irregular, colourless
V = 797.59 (8) Å^3	0.08 × 0.05 × 0.03 mm

Data collection

KEK BL14A four-circle diffractometer	Rint = 0.021
Silicon monochromator	θmax = 32.5°, θmin = 0.8°
ω/2θ scans	h = −7→7
Absorption correction: analytical (de Meulenaer & Tompa, 1965)	k = 0→7
Tmin = 0.598, Tmax = 0.885	l = −38→38
5607 measured reflections	8 standard reflections every 200 reflections
4904 independent reflections	intensity decay: none
3750 reflections with I > 2σ(I)

Refinement

Refinement on F^2	1 restraint
Least-squares matrix: full	Primary atom site location: structure-invariant direct methods
R[F^2 > 2σ(F^2)] = 0.051	Secondary atom site location: difference Fourier map
wR(F^2) = 0.169	w = 1/[σ^2(Fo^2) + (0.101P)^2 + 5.14P] where P = (Fo^2 + 2Fc^2)/3
S = 0.94	(Δ/σ)max = 0.002
4904 reflections	Δρmax = 2.13 e Å^−3
297 parameters	Δρmin = −4.03 e Å^−3

Special details

Geometry. All e.s.d.'s (except the e.s.d. on the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell e.s.d.'s are taken into account individually in the estimation of e.s.d.'s in distances, angles and torsion angles; correlations between e.s.d.'s in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell e.s.d.'s is used for estimating e.s.d.'s involving l.s. planes.

Refinement. Refinement of F² against ALL reflections. The weighted R-factor wR and goodness of fit S are based on F², conventional R-factors R are based on F, with F set to zero for negative F². The threshold expression of F² > σ(F²) is used only for calculating R-factors(gt) etc. and is not relevant to the choice of reflections for refinement. R-factors based on F² are statistically about twice as large as those based on F, and R- factors based on ALL data will be even larger.

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Ų)

	x	y	z	Uiso*/Ueq	Occ. (<1)
Nb1	0	0	0	0.01101 (14)
Nb2	0	0.5	0.5	0.01100 (14)
Nb3	0.99467 (6)	0.99965 (6)	0.154237 (13)	0.01070 (13)
Nb4	0.99469 (6)	0.50045 (6)	0.654254 (13)	0.01071 (13)
Nb5	0.5	0.5	0	0.01103 (14)
Nb6	0.5	0	0.5	0.01107 (14)
Nb7	0.50344 (6)	0.50842 (6)	0.154250 (13)	0.01073 (13)
Nb8	0.50337 (6)	0.99167 (6)	0.654242 (13)	0.01078 (13)
Ca1	0.00721 (19)	0.49439 (18)	0.08093 (4)	0.0262 (3)
Ca2	0.00730 (19)	0.00563 (18)	0.58089 (4)	0.0261 (3)
Ca3	0.49521 (19)	0.00654 (18)	0.07870 (4)	0.0266 (3)
Ca4	0.49512 (19)	0.49343 (18)	0.57871 (4)	0.0264 (3)
Li1	0.747 (2)	0.760 (2)	0.2606 (5)	0.019 (3)*	0.64 (3)
Li2	0.247 (2)	0.259 (2)	0.2403 (5)	0.027 (4)*	0.70 (4)
Li3	0.252 (4)	0.752 (4)	0.2487 (9)	0.019 (6)*	0.35 (3)
Li4	0.755 (4)	0.252 (4)	0.2485 (9)	0.014 (6)*	0.31 (3)
O1	0.2455 (7)	0.7526 (6)	0.14114 (19)	0.0341 (9)
O2	0.7545 (6)	0.2539 (6)	0.35880 (19)	0.0331 (9)
O3	0.7453 (5)	0.7570 (5)	0.13183 (11)	0.0129 (5)
O4	0.2547 (5)	0.2568 (5)	0.36811 (11)	0.0135 (5)
O5	0.7485 (7)	0.2570 (6)	0.14111 (19)	0.0332 (9)
O6	0.7494 (6)	0.2433 (6)	0.64121 (19)	0.0329 (9)
O7	0.2490 (6)	0.2536 (5)	0.15727 (12)	0.0187 (6)
O8	0.7512 (6)	0.7544 (5)	0.34269 (12)	0.0181 (6)
O9	0.7998 (5)	0.7000 (5)	0.01381 (12)	0.0162 (6)
O10	0.1993 (5)	0.1999 (5)	0.48638 (11)	0.0154 (5)
O11	0.6950 (6)	0.1952 (6)	0.0001 (2)	0.0401 (11)
O12	0.6961 (6)	0.3045 (6)	0.5001 (2)	0.0399 (11)
O13	0.5335 (7)	0.5408 (7)	0.22112 (12)	0.0274 (8)
O14	0.4669 (7)	0.0399 (7)	0.27870 (12)	0.0279 (8)
O15	0.9636 (7)	0.9703 (7)	0.22111 (12)	0.0269 (8)
O16	0.9649 (7)	0.5295 (7)	0.72114 (12)	0.0265 (8)
O17	0.4467 (8)	0.4599 (8)	0.07152 (13)	0.0361 (10)
O18	0.5525 (8)	0.9605 (8)	0.42844 (13)	0.0364 (10)
O19	0.0428 (8)	0.0541 (8)	0.07147 (13)	0.0373 (10)
O20	0.0415 (8)	0.4460 (8)	0.57145 (13)	0.0371 (10)

Atomic displacement parameters (Ų)

	U^11	U^22	U^33	U^12	U^13	U^23
Nb1	0.0109 (3)	0.0117 (3)	0.0105 (3)	0.00033 (17)	0.00056 (18)	0.00010 (18)
Nb2	0.0110 (3)	0.0114 (3)	0.0106 (3)	−0.00187 (17)	0.00051 (18)	0.00087 (18)
Nb3	0.0093 (2)	0.00845 (19)	0.0144 (2)	−0.00092 (13)	0.00060 (14)	0.00083 (14)
Nb4	0.0092 (2)	0.00857 (19)	0.0144 (2)	−0.00065 (13)	0.00061 (14)	0.00017 (14)
Nb5	0.0123 (3)	0.0102 (2)	0.0106 (3)	0.00033 (17)	0.00090 (18)	0.00070 (18)
Nb6	0.0124 (3)	0.0101 (2)	0.0108 (3)	−0.00175 (17)	0.00097 (18)	0.00029 (18)
Nb7	0.0097 (2)	0.00812 (19)	0.0143 (2)	−0.00105 (13)	0.00014 (14)	0.00047 (13)
Nb8	0.0096 (2)	0.00816 (19)	0.0146 (2)	−0.00060 (13)	0.00016 (14)	0.00065 (13)
Ca1	0.0250 (5)	0.0237 (5)	0.0299 (6)	−0.0038 (4)	−0.0043 (4)	0.0051 (4)
Ca2	0.0248 (5)	0.0236 (5)	0.0299 (6)	0.0028 (4)	−0.0047 (4)	−0.0046 (4)
Ca3	0.0272 (6)	0.0251 (5)	0.0276 (6)	−0.0075 (4)	−0.0061 (4)	0.0074 (4)
Ca4	0.0263 (6)	0.0254 (5)	0.0273 (6)	0.0065 (4)	−0.0066 (4)	−0.0070 (4)
O1	0.0169 (16)	0.0153 (15)	0.070 (3)	0.0044 (12)	−0.0020 (16)	−0.0006 (16)
O2	0.0135 (15)	0.0129 (14)	0.073 (3)	−0.0071 (11)	0.0008 (16)	0.0010 (15)
O3	0.0136 (12)	0.0111 (11)	0.0140 (13)	−0.0063 (9)	0.0001 (10)	0.0002 (9)
O4	0.0135 (12)	0.0125 (12)	0.0146 (13)	0.0039 (9)	−0.0011 (10)	−0.0003 (9)
O5	0.0175 (16)	0.0144 (15)	0.068 (3)	0.0040 (12)	0.0024 (16)	0.0019 (15)
O6	0.0145 (15)	0.0114 (14)	0.073 (3)	−0.0066 (11)	0.0030 (16)	−0.0017 (15)
O7	0.0195 (15)	0.0161 (13)	0.0205 (15)	−0.0127 (11)	−0.0009 (11)	0.0016 (11)
O8	0.0195 (14)	0.0162 (13)	0.0185 (15)	0.0071 (10)	0.0011 (11)	0.0001 (10)
O9	0.0145 (13)	0.0132 (12)	0.0208 (14)	−0.0074 (10)	−0.0043 (10)	0.0067 (10)
O10	0.0149 (13)	0.0133 (12)	0.0180 (14)	0.0032 (9)	−0.0041 (10)	−0.0040 (10)
O11	0.0105 (14)	0.0112 (13)	0.099 (4)	0.0019 (11)	0.0010 (17)	0.0007 (17)
O12	0.0107 (14)	0.0106 (13)	0.099 (4)	−0.0050 (11)	0.0013 (17)	0.0013 (16)
O13	0.038 (2)	0.0354 (18)	0.0087 (13)	−0.0204 (15)	0.0007 (12)	0.0017 (12)
O14	0.039 (2)	0.0357 (18)	0.0093 (13)	0.0190 (15)	0.0006 (13)	0.0006 (12)
O15	0.038 (2)	0.0348 (18)	0.0082 (13)	−0.0201 (15)	0.0004 (12)	0.0003 (12)
O16	0.0353 (19)	0.0362 (18)	0.0080 (13)	0.0175 (14)	−0.0003 (12)	0.0001 (12)
O17	0.050 (2)	0.051 (2)	0.0076 (14)	−0.0335 (19)	0.0018 (14)	−0.0010 (13)
O18	0.049 (2)	0.052 (2)	0.0077 (14)	0.0296 (19)	0.0051 (14)	0.0010 (14)
O19	0.055 (3)	0.048 (2)	0.0084 (14)	−0.034 (2)	0.0006 (15)	−0.0016 (14)
O20	0.054 (3)	0.049 (2)	0.0082 (14)	0.031 (2)	0.0008 (15)	0.0029 (14)

Geometric parameters (Å, º)

Nb1—O19^i	1.934 (3)	Ca1—O17	3.087 (5)
Nb1—O19	1.934 (3)	Ca1—O11	3.193 (4)
Nb1—O11	1.985 (3)	Ca1—O11	3.196 (4)
Nb1—O11	1.985 (3)	Ca2—O10	2.393 (3)
Nb1—O9	2.011 (3)	Ca2—O18	2.434 (5)
Nb1—O9	2.011 (3)	Ca2—O20	2.434 (4)
Nb2—O20	1.933 (3)	Ca2—O4	2.446 (3)
Nb2—O20	1.933 (3)	Ca2—O2	2.505 (4)
Nb2—O12	1.981 (3)	Ca2—O6	2.506 (4)
Nb2—O12	1.981 (3)	Ca2—O8	2.753 (3)
Nb2—O10	2.008 (3)	Ca2—O10	2.926 (3)
Nb2—O10	2.008 (3)	Ca2—O18	3.083 (5)
Nb3—O15	1.792 (3)	Ca2—O20	3.083 (4)
Nb3—O1	1.962 (4)	Ca2—O12	3.188 (4)
Nb3—O7	1.971 (3)	Ca2—O12	3.192 (4)
Nb3—O5	1.982 (4)	Ca3—O3	2.392 (3)
Nb3—O3	1.996 (3)	Ca3—O19	2.500 (5)
Nb3—O19	2.235 (4)	Ca3—O17	2.506 (4)
Nb4—O16	1.791 (3)	Ca3—O5	2.557 (4)
Nb4—O2	1.956 (3)	Ca3—O1	2.565 (4)
Nb4—O8	1.974 (3)	Ca3—O11	2.578 (5)
Nb4—O6	1.977 (3)	Ca3—O11	2.582 (5)
Nb4—O4	1.996 (3)	Ca3—O7	2.834 (3)
Nb4—O20	2.235 (4)	Ca3—O9	2.933 (3)
Nb5—O17	1.935 (3)	Ca3—O17	3.014 (4)
Nb5—O17	1.935 (3)	Ca3—O19	3.020 (5)
Nb5—O11	1.983 (3)	Ca3—O9	3.347 (3)
Nb5—O11	1.983 (3)	Ca4—O4	2.394 (3)
Nb5—O9	2.008 (3)	Ca4—O20	2.506 (5)
Nb5—O9	2.008 (3)	Ca4—O18	2.508 (4)
Nb6—O18	1.935 (3)	Ca4—O6	2.560 (4)
Nb6—O18	1.935 (3)	Ca4—O2	2.562 (4)
Nb6—O12	1.985 (3)	Ca4—O12	2.581 (5)
Nb6—O12	1.985 (3)	Ca4—O12	2.586 (5)
Nb6—O10	2.010 (3)	Ca4—O8	2.837 (3)
Nb6—O10	2.010 (3)	Ca4—O10	2.940 (3)
Nb7—O13	1.791 (3)	Ca4—O18	3.010 (4)
Nb7—O5	1.957 (4)	Ca4—O20	3.013 (5)
Nb7—O7	1.975 (3)	Ca4—O10	3.346 (3)
Nb7—O1	1.976 (4)	Li1—O15	1.962 (12)
Nb7—O3	1.994 (3)	Li1—O13	1.975 (12)
Nb7—O17	2.233 (4)	Li1—O8	2.180 (14)
Nb8—O14	1.796 (3)	Li1—O14	2.224 (12)
Nb8—O6	1.961 (3)	Li1—O16	2.289 (12)
Nb8—O8	1.973 (3)	Li2—O16	1.937 (12)
Nb8—O2	1.981 (3)	Li2—O14	1.981 (12)
Nb8—O4	1.993 (3)	Li2—O7	2.205 (14)
Nb8—O18	2.231 (4)	Li2—O13	2.262 (12)
Ca1—O9	2.392 (3)	Li2—O15	2.274 (12)
Ca1—O17	2.430 (5)	Li3—O13	2.07 (2)
Ca1—O19	2.434 (4)	Li3—O16	2.11 (2)
Ca1—O3	2.444 (3)	Li3—O15	2.11 (2)
Ca1—O1	2.499 (4)	Li3—O14	2.12 (2)
Ca1—O5	2.505 (4)	Li4—O15	2.06 (2)
Ca1—O7	2.754 (3)	Li4—O16	2.10 (2)
Ca1—O9	2.933 (3)	Li4—O14	2.12 (2)
Ca1—O19	3.084 (4)	Li4—O13	2.12 (2)
O15—Li1—O13	115.6 (7)	O7—Li2—O15	76.9 (4)
O15—Li1—O8	122.6 (6)	O13—Li2—O15	154.0 (7)
O15—Li1—O14	97.5 (5)	O13—Li3—O16	98.5 (9)
O15—Li1—O16	95.9 (5)	O13—Li3—O15	138.9 (12)
O13—Li1—O8	121.7 (6)	O13—Li3—O14	97.8 (9)
O13—Li1—O14	97.3 (5)	O13—Li3—O1	69.6 (7)
O13—Li1—O16	95.6 (5)	O16—Li3—O15	97.1 (9)
O8—Li1—O14	78.3 (4)	O16—Li3—O14	135.6 (12)
O8—Li1—O16	76.9 (4)	O16—Li3—O1	112.1 (9)
O8—Li1—Li4	97.6 (7)	O15—Li3—O14	97.0 (9)
O8—Li1—Li3	97.0 (7)	O15—Li3—O1	69.3 (7)
O8—Li1—Li3	96.1 (6)	O14—Li3—O1	112.3 (9)
O8—Li1—Li4	95.8 (6)	O15—Li4—O16	99.0 (9)
O14—Li1—O16	155.3 (7)	O15—Li4—O14	97.9 (9)
O16—Li2—O14	117.1 (7)	O15—Li4—O13	139.2 (12)
O16—Li2—O7	122.8 (6)	O15—Li4—O5	70.0 (7)
O16—Li2—O13	97.4 (5)	O16—Li4—O14	135.2 (12)
O16—Li2—O15	97.1 (5)	O16—Li4—O13	97.0 (9)
O14—Li2—O7	120.1 (6)	O16—Li4—O5	112.6 (9)
O14—Li2—O13	96.2 (5)	O14—Li4—O13	96.5 (9)
O14—Li2—O15	96.2 (5)	O14—Li4—O5	112.2 (9)
O7—Li2—O13	77.2 (4)	O13—Li4—O5	69.2 (7)

Symmetry code: (i) −x, −y, −z.