Thulium nickel/lithium distannide, TmNi1−xLixSn2 (x = 0.035)

The quaternary thulium nickel/lithium distannide, TmNi1−xLixSn2 (x = 0.035), crystallizes in the orthorhombic LuNiSn2 structure type. The asymmetric unit contains three Tm sites, six Sn sites, two Ni sites and one Ni/Li site [relative occupancies = 0.895 (8):0.185 (8)]. Site symmetries are .m. for all atoms. The 17-, 18- and 19-vertex distorted pseudo-Frank–Kasper polyhedra are typical for all Tm atoms. Four Sn atoms are enclosed in a 12-vertex deformed cubooctahedron, and another Sn atom is enclosed in a pentagonal prism with three added atoms. A tricapped trigonal prism is typical for a further Sn atom. The coordination number for all Ni atoms and Ni/Li statistical mixtures is 12 (fourcapped trigonal prism [Ni/LiTm5Sn5]). Tm atoms form the base of a prism and Ni/Li atoms are at the centres of the side faces of an [SnTm6Ni/Li3] prism. These isolated prisms are implemented into three-dimensional-nets built out of Sn atoms. Electronic structure calculations using TB-LMTO-ASA suggest that the Tm and Ni/Li atoms form positively charged n[TmNi/Li]m+ polycations which compensate the negative charge of 2n[Sn]m− polyanions. Analysis of the interatomic distances and electronic structure calculations indicate the dominance of a metallic type of bonding.

Financial support from the Ministry of Education and Science of Ukraine is gratefully acknowledged.

Comment
The RETSn 2 and RET x Sn2 (x<1) type metallic compounds where RE is a rare-earth element (Gd-Lu) and T is a delectron element crystallize in different orthorhombic crystal structures LuNiSn 2 (space group Pnma) and CeNiSi 2 -type (space group Cmcm) respectively. In the ternary RELiSn 2 compounds lithium atoms occupy the same crystallographic position as the atoms of transition metal in the original CeNiSi 2 structure type (Pavlyuk et al., 1989a). Previous structural studies of the four-component alloys from TbLiSn 2 -TbZnSn 2 sections indicate the existence of TbLi 1-x Zn x Sn 2 limited solid solution (Stetskiv et al., 2012). X-ray single-crystal study showed that the TbLi 1-x Zn x Sn 2 solid solution was formed by the partial substitution of lithium atoms by zinc atoms in 4c site. The ability of lithium atoms to partially substitute the atoms of transition metals was previously observed by us while studying solid solutions RELi x Cu 2-x Si 2 and RELi x Cu 2-x Ge 2 (Pavlyuk et al., 1993). The ordered substitution of transition metals by lithium is observed for Tm 2.22 Co 6 Sn 20 and TmLi 2 Co 6 Sn 20 stannides (Stetskiv et al., 2013). The ability of lithium atoms to occupy the same crystallographic position as the atoms of transition metal was observed previously while studying compounds RELiGe with the ZrNiAl-type (Pavlyuk et al., 1991 andPavlyuk &Bodak, 1992a), RE 3 Li 2 Ge 3 with Hf 3 Ni 2 Si 3 -type (Pavlyuk & Bodak, 1992b) and Yb 5 Li 4 Ge 4 with Nb 5 Cu 4 Si 4 -type (Pavlyuk et al., 1989b).
The four-component phase TmNi 1-x Li x Sn 2 with low content of lithium from the TmLiSn 2 -TmNiSn 2 section was detected by us during the systematic study of alloys of Tm-Ni-Li-Sn system. Selected single-crystal data show that the title compound crystallizes with the orthorhombic space group Pnma as a LuNiSn 2 -type (Komarovskaya et al., 1983). The projection of the unit cell and coordination polyhedra of the atoms are shown in Fig. 1. The Tm atoms are enclosed in 17-, 18-and 19-vertex distorted pseudo Frank-Kasper polyhedra. The coordination polyhedron of Sn4, Sn7, Sn8 and Sn9 atoms is 12-vertex distorted cubooctahedron. The Sn5 is enclosed in pentagonal prism with three added atoms. The tricapped trigonal prism is typical for Sn6 atom. The environment of the Ni atoms and Ni/Li statistical mixture is a fourcapped trigonal prism and a coordination number equals 10 (Tm 5 Sn 5 ).
The distribution of nickel/lithium and thulium atoms in three-dimensional-nets built of Sn atoms is shown in Fig. 2

Experimental
Thulium, nickel, lithium and tin, all with a nominal purity more than 99.9 wt. %, were used as starting elements. First, the pieces of the pure metals with a stoichiometry Tm 25 Ni 20 Li 5 Sn 50 were pressed into pellet, enclosed in tantalum crucible and placed in a resistance furnace with a thermocouple controller. Heating rate from room temperature to 670 K was equal to 5 K per minute. At this temperature the alloy was kept over 2 d and then the temperature was increased from 670 to 1070 K over 1 h. Then, the alloy was annealed at this temperature for 8 h and slowly cooled down to room temperature. After the melting and annealing procedures, the total weight loss was less than 2%. Small good quality single-crystal of the title compound was isolated from the alloy.
The synthesized alloy is practically single-phase. Therefore, in order to confirm the accuracy of the compositions, the density of the alloy was determined using the volumetric method. The measured density is 9.08 (5) Mg m -3 , and these values differ by less than 1% from the densities calculated from the X-ray data. For the TmNiSn 2 ternary phase density is 9,19 Mg m -3 (Komarovskaya et al., 1983).

Refinement
The structure of the title phase was solved by direct methods after the analytical absorption correction. In the first stage of the refinement, the thermal displacement parameter of Ni12 atom was considerably different from those of other Ni sites, suggesting that this position is partially occupied by the lithium atom. In the final refinement cycles all atoms were successfully refined with anisotropic displacement parameters.  The projection of the unit cell and coordination polyhedra of the atoms.

Figure 2
The distribution of thulium and nickel/lithium atoms in three-dimensional-nets built of the Sn atoms.  Refinement. Refinement of F 2 against ALL reflections. The weighted R-factor wR and goodness of fit S are based on F 2 , conventional R-factors R are based on F, with F set to zero for negative F 2 . The threshold expression of F 2 > σ(F 2 ) is used only for calculating R-factors(gt) etc. and is not relevant to the choice of reflections for refinement. R-factors based on F 2 are statistically about twice as large as those based on F, and R-factors based on ALL data will be even larger.