Li₂Pt₃Se₄: a new lithium platinum selenide with jaguéite-type crystal structure by multianvil high-pressure/high-temperature synthesis

2.1 High-pressure/high-temperature syntheses

Starting materials for the multianvil high-pressure/high-temperature synthesis of Li₂Pt₃Se₄, according to Eq. 1, was a stoichiometric mixture of Li₃N (purity >99.4%, Alfa Aesar), platinum sponge (purity >99.8%, Strem Chemicals, Inc.), and selenium powder (purity >99.9%, Fluka). Similarly, for syntheses of the isostructural compounds A₂Pd₃Se₄ (A=Ag, Cu), according to Eq. 2, stoichiometric mixtures of silver powder (purity >99.9%, Strem Chemicals, Inc.) or copper powder (purity >99%, Sigma Aldrich) in combination with palladium powder (purity >99.95%, Strem Chemicals, Inc.) and selenium powder (purity >99.9%, Fluka) were used.

The mixtures of the starting materials were milled and loaded into 18/11-assembly crucibles made of hexagonal boron nitride (HeBoSint^® P100, Henze BNP GmbH, Kempten, Germany). Because of the air and moisture sensitivity of lithium nitride required for the synthesis of Li₂Pt₃Se₄, it was essential to prepare the assembly under inert gas atmosphere. To compress the walker module of the press up to the desired pressure of 8 GPa, 210 min were necessary. After reaching the synthesis pressure the samples were heated to 1200°C within 15 min, kept constant for 10 min and were gently cooled down for the next 2 h to 500°C, followed by quenching the samples to room temperature. This annealing process under pressure can improve the crystallinity of the samples. After finishing the decompression process of the press, the samples could be easily separated from the surrounding assembly materials and no reaction with the crucible material was observed. Further information about the technique and the construction of the different assemblies can be found in numerous references [18], [19], [20], [21]. The polycrystalline samples were stable in air and appeared silvery with metallic lustre. By milling, they turned into dark gray powders.

2.2 X-ray diffraction and data collections

Characterization of the polycrystalline high-pressure samples was performed by powder diffraction on a stoe Stadi P diffractometer with (111)-curved Ge-monochromatized MoK_α1 radiation (λ=70.93 pm) in transmission geometry. The powdered samples were mounted between acetate films and fixed with high-vacuum grease. The diffraction intensities were collected by a Dectris Mythen2 1K microstrip detector with 1280 strips.

Based on the parameters derived from the single-crystal structure model, Rietveld refinement of Li₂Pt₃Se₄ was done with the Diffrac^plus-Topas^® 4.2 software package (Bruker AXS, Karlsruhe, Germany). Instrument contributions were taken into account by using a measured instrument function for reflection profiles. Peak shapes were modeled using modified Thompson–Cox–Hastings pseudo-Voigt profiles [22], [23]. The background was fitted with Chebychev polynomials up to the 12^th order. The lattice parameters derived from the Rietveld refinement agree well with those obtained from the single-crystal data (see Table 1). Figure 1 shows the result of the Rietveld refinement of Li₂Pt₃Se₄. Elemental platinum [26], PtSe₂ [27], and Pt₅Se₄ [28] were always present in different amounts as by-products. The bulk samples of Ag₂Pd₃Se₄ and Cu₂Pd₃Se₄ from high-pressure experiments indicated a low degree of crystallinity, and powder refinements resulted in no precise lattice parameters. Nevertheless, some single crystals suitable for X-ray diffraction were identified inside the bulk samples.

Fig. 1:

XRD pattern (MoK_α1 radiation) of Li₂Pt₃Se₄, simultaneously refined with Pt [26], PtSe₂ [27], and Pt₅Se₄ [28] (R_exp=1.65, R_wp=2.57, R_p=1.85, and GOF=1.56).

By mechanical fragmentation, several irregularly shaped silvery crystals were isolated from the crushed samples treated under high-pressure/high-temperature conditions. For better handling, the crystal fragments were embedded in perfluoropolyalkylether (viscosity 1800 cSt) and fixed on thin glass fibers with high-vacuum grease. The intensity data collections of the Cu₂Pd₃Se₄ and Li₂Pt₃Se₄ crystals were carried out on a Nonius Kappa-CCD diffractometer with graphite-monochromatized MoK_α radiation (λ=71.07 pm) at room temperature. The program Scalepack [24] was used to correct the intensity data for absorption based on equivalent and redundant intensities. For the Ag₂Pd₃Se₄ single-crystal data collection, a Bruker D8 Quest diffractometer with a Photon 100 detector system and an Incoatec Microfocus source generator (multi layered optics-monochromatized MoK_α radiation, λ=71.07 pm) was used. Collection strategies, concerning ω and ϕ scans, were optimized with the Apex-2 program package [25], resulting in data sets of complete reciprocal spheres up to high angles with high completeness. Reflection intensities were integrated with the program Saint [25] using a narrow-frame algorithm and corrected for absorption effects with the program Sadabs [25], based on the semi-empirical multi-scan approach. Table 1 summarizes the experimental details.

3.1 Structure refinements

For all three compounds the systematic extinctions h0l with l≠2n, 0k0 with k≠2n, and 00l with l≠2n were evident and led to the space group P2₁/c (no. 14). The initial positional parameters were deduced from an automatic interpretation of Direct Methods with Shelxs-2013 [29], and the following full-matrix least-squares refinements on F² were performed with Shelxl-2013 [30], [31]. In the case of Li₂Pt₃Se₄, the positional parameters of the isostructural compound Li₂Pt₃S₄ [7] were used as starting values for the structure refinement. To verify the correct compositions of Li₂Pt₃Se₄, Cu₂Pd₃Se₄, and Ag₂Pd₃Se₄, all occupancy parameters were refined in separate series of least-squares cycles. All sites were fully occupied within two standard deviations. Furthermore, all sites were refined with anisotropic displacement parameters, but only for Li₂Pt₃Se₄ the final Fourier synthesis did not reveal any significant residual peaks (see Table 1). The positional parameters as starting values for the refinement of Ag₂Pd₃Se₄ were derived from Cu₂Pd₃Se₃ with the Cu⁺ position taken for Ag⁺. A refinement of a possible mixed occupation of Pd sites with Cu or Ag resulted in higher R values even if, in the case of Ag₂Pd₃Se₄, there are only slight differences in the scattering factors of Ag⁺ and Pd²⁺. Moreover, geometrical aspects, which will be discussed later, are in favor of a clear distinction between Pd²⁺ sites on the one hand and Cu⁺ or Ag⁺ sites on the other.

The refinements of the synthetic mineral phases Cu₂Pd₃Se₄ and Ag₂Pd₃Se₄ exhibited considerable residual peaks in distances of 0.65(5) Å from the Cu1, Ag1, and Pd1 sites. As a consequence, a split-atom model of the Cu1 and Pd1 sites in Cu₂Pd₃Se₄ as well as of the Ag1 and Pd1 sites in Ag₂Pd₃Se₄, was evident in separate difference-Fourier syntheses. The additional split sites were refined with isotropic displacement parameters. Due to high correlation factors, refinements with anisotropic displacement parameters were not reasonable. The splitting of the sites in Ag₂Pd₃Se₄ was about twice as pronounced as in Cu₂Pd₃Se₄. About 10% of the Ag1 and Pd1 atoms of Ag₂Pd₃Se₄ occupy additional split sites, whereas in Cu₂Pd₃Se₄ about 5% of Cu1 and Pd1 are distributed over two positions. Without refinement of the split positions, the R values of Ag₂Pd₃Se₄ and Cu₂Pd₃Se₄ are R1=0.0685/wR2=0.1704 and R1=0.0654/wR2=0.1441, respectively. A refinement including the split positions resulted in considerably better R values of R1=0.0325/wR2=0.0721 for Ag₂Pd₃Se₄ and R1=0.0460/wR2=0.1000 for Cu₂Pd₃Se₄. Final Fourier syntheses revealed negligible residual peaks in distances of 0.70(2) Å next to Se1 and Se2. A refinement of other split positions was neglected. Non-merohedral twinning as a possible reason for the structural disorder was checked carefully but no evidence could be found. Small epitaxially grown crystals, which could be indexed with Cell_Now [25], are in our opinion not the reason for the observed structural disorder as their scattering intensity contributed only about 1% to the total scattering of the main component. Details of the single-crystal structure measurements are shown in Table 1, and the positional parameters (Table 2), anisotropic displacement parameters (Table 3), interatomic distances, and angles (Tables 4 and 5) are listed additionally.

Further details of the crystal structure investigations may be obtained from Fachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany (fax: +49-7247-808-666; e-mail: crysdata@fiz-karlsruhe.de, http://www.fiz-karlsruhe.de/request_for_deposited_data.html) on quoting the deposition numbers CSD-431596 (Li₂Pt₃Se₄), CSD-431597 (Ag₂Pd₃Se₄), and CSD-431598 (Cu₂Pd₃Se₄).

3.2 Crystal chemistry

As already mentioned, the lithium platinum selenide, Li₂Pt₃Se₄ presented here, which was synthesized under high-pressure/high-temperature conditions of 8 GPa and 1200°C, crystallizes isostructurally to the corresponding sulfides Li₂M₃S₄ (M=Pd, Pt) [7], and to the minerals chrisstanleyite (Ag₂Pd₃Se₄) and jaguéite (Cu₂Pd₃Se₄) [3]. Synthesis attempts for the missing compound Li₂Pd₃Se₄ have failed so far, resulting in binary lithium selenide and various palladium selenides. Normal-pressure experiments in sealed silica ampoules were also performed without success. According to Vymazalová [4], the mineral chrisstanleyite is only stable at temperatures below 430°C. Our experiments to synthesize Li₂M₃X₄ (M=Pd, Pt, X=S, Se) compounds exceeded this temperature significantly. Repetitions of the experiments at low temperatures are still pending. In Table 6 the lattice parameters of the known isostructural compounds are summarized.

The crystal structure of Li₂Pt₃Se₄ is built up from two distinct Pt sites (Pt1 on a general position; Pt2 on the special position at 0, 0, 0), two general selenium sites, and one general lithium site (see Table 2). Both Pt atoms are in a nearly square planar Se coordination with Pt–Se distances ranging from 244.8 to 247.5 pm. There are distinct differences between the two Pt sites, however. The atoms Pt2 occur in single PtSe₄ units, whereas the atoms Pt1 form pairs of square planar PtSe₄ units paired via a common Se2–Se2 edge and resulting in Pt1–Pt1 distances of 406.1(1) pm (see Fig. 2). These two units (single and paired PtSe₄ units) build up a three-dimensional open network structure by linking the units through common corners. As a result of the arrangement of the PtSe₄ units, channels are created, in which the lithium atoms are located. Figure 2 shows the crystal structure of Li₂Pt₃Se₄ viewed along the crystallographic a and c axes, respectively. For a more detailed description of the crystal structure, the reader is referred to the references [3], [7]. A comparison of Li₂Pt₃Se₄ and the synthetic mineral phases chrisstanleyite (Ag₂Pd₃Se₄) and jaguéite (Cu₂Pd₃Se₄) is given below.

Fig. 2:

Crystal structures of Li₂Pt₃Se₄ as viewed along the crystallographic a (top) and c axes (down). Lithium, platinum and selenium are drawn as dark purple, light gray and orange circles, respectively. Relevant PtSe₄ polyhedra are emphasized.

As already mentioned in the introduction, Paar [1], [2] described chrisstanleyite and jaguéite as commonly occurring, partially intergrown minerals, with a mixed (Ag, Cu) occupancy. Twenty percent of the silver atoms in chrisstanleyite are substituted by copper and 5% of the copper atoms in jaguéite are replaced by silver resulting in the corresponding compounds Ag_1.6Cu_0.4Pd₃Se₄ (chrisstanleyite) and Cu_1.9Ag_0.1Pd₃Se₄ (jaguéite). In the case of chrisstanleyite, Topa refined the mixed occupancy with equal isotropic displacement factors, but with free positional parameters [3] creating additional sites. For natural jaguéite no mixed occupancies were refined by Topa. The appearance of additional sites in this structure type is rather a stabilizing effect of an optimized network of metal-metal bonds than an effect of differences in the ionic radii of silver (Ag⁺ C.N. 6: 1.29 Å [32]) and copper (Cu⁺ C.N. 6: 0.91 Å [32]). This can be concluded from the refinements of the pure synthetic minerals presented here which show a comparable splitting of the copper and silver sites, without an occupation of the split sites with different atom types. Furthermore, for the Pd1 sites of the synthetic minerals a refinement of hitherto not described split positions at distances of 0.53 Å (Ag₂Pd₃Se₄) and 0.64 Å (Cu₂Pd₃Se₄) resulted in significantly better R values. Overall, the effect of layer splitting is more pronounced in Ag₂Pd₃Se₄ (Ag1a/Ag1b: 0.90(1)/0.10(1); Pd1a/Pd1b: 0.92(1)/0.08(1)) than in Cu₂Pd₃Se₄ (Cu1a/Cu1b: 0.94(2)/0.06(2); Pd1a/Pd1b: 0.95(1)/0.05(1)). Most likely, copper and silver occur strictly in the oxidation state +1 in these compounds. The occurrence of Cu²⁺ (or Ag²⁺) ions in a mixed occupation of the Pd1a/Pd1b sites as reason for the structural disorder is rather unlikely, but cannot be completely excluded. As already mentioned above, the R values worsen when the disorder is taken into accont. Moreover, the d⁸ Pd²⁺ ions gain a significantly better stabilization in the square planar ligand field than d⁹-configured Cu²⁺ (or Ag²⁺) ions would get. Magnetic measurements to confirm the exclusive +1 oxidation state of copper and silver have not been attempted as yet. Unfortunately, up to now we were not able to optimize the synthesis of Cu₂Pd₃Se₄ in such a way as to obtain pure products without side phases, which is a premise for meaningful magnetic measurements.

The replacement of Cu⁺/Ag⁺ by lithium (Li⁺ C.N.6: 0.90 Å [32]) in Li₂Pt₃Se₄ and the isostructural compounds Li₂Pt₃S₄ and Li₂Pd₃S₄ [7] leads to the absence of split positions. Figure 3 shows the stabilizing metal–metal bond system of this structure type mentioned above, consisting of a nearly collinear sequence of four metal–metal bonds with distances of 268.8–283.9 pm (Li₂Pt₃Se₄). The Pt1–Li1–Pt2–Li1–Pt1 sequences are interconnected via Pt1–Li1 bonds to zig-zag layers parallel to the bc plane. A comparable situation can be found for the isostructural compounds Li₂Pd₃S₄ and Li₂Pt₃S₄ with Li–Pd and Li–Pt contacts varying from 260.5–283.1 pm and 259.7–285.0 pm, respectively [7]. In Fig. 4, a cutout of the collinear metal–metal sequence demonstrates the different bonding situations in Li₂Pt₃Se₄ and Ag₂Pd₃Se₄. All contacts shorter than 301 pm are drawn as bonds. The metal–metal distances in Li₂Pt₃Se₄ are less different and the contacts interconnecting the collinear sequences are much shorter than in Ag₂Pd₃Se₄. With implementation of the additional split sites (split atoms are labeled with appendix a and b in Fig. 4) in Ag₂Pd₃Se₄, two shorter Ag1b–Pd1a and Ag1b–Pd1b contacts of 227 pm and 275 pm become prominent, whereas the Ag1b–Pd2 contact is elongated up to 344 pm. Therefore, the Pd2 atoms remain isolated and the collinear metal-metal sequence changes into an alternating Ag1–Pd1 zig-zag sequence in the direction along the c axis. For Cu₂Pd₃Se₄ (not shown) the situation is comparable.

Fig. 3:

Crystal structure of Li₂Pt₃Se₄ as viewed along the a (top) and c axes (down) with Li–Pt metal–metal bonds drawn in a collinear sequence Pd2–Li–Pd1–Li–Pd2. Lithium, platinum and selenium are drawn as dark purple, light gray and orange circles, respectively.

Fig. 4:

Collinear sequence of Pt1-Li1-Pt2-Li1-Pt1 in Li₂Pt₃Se₄ (top) and corresponding sequence in Ag₂Pd₃Se₄ (down) with Pd1 and Ag1 split positions. Atom distances shorter than 301 pm are drawn as bonds. All distances are given in pm.

Differences between the lithium-containing compounds and the mineral phases regarding the coordination of lithium, silver and copper are evident. Topa [3] described the coordination polyhedra of copper and silver as strongly elongated tetrahedra with Cu–Se distances of 244.4–285.7 pm for jaguéite, and Ag–Se distances of 259.0–294.8 pm for chrisstanleyite. The lithium atoms in Li₂M₃X₄ (M=Pd, Pt, X=S, Se) are coordinated by six chalcogenide atoms in a distorted trigonal prismatic coordination geometry. The distances range from 249.8–288.5 pm for Li₂Pd₃S₄, 247.9–288.6 pm for Li₂Pt₃S₄, and 262.3–292.5 pm for Li₂Pt₃Se₄ (see Table 4). In Fig. 5 the lithium, silver and copper coordination polyhedra are shown with relevant distances. Li₂Pt₃Se₄ exhibits two short Li–Se contacts (262.3 pm, 272.2 pm) and four longer (284.2–292.5 pm) ones (see Table 4). Nevertheless, the distances are more uniform than in Ag₂Pd₃Se₄ and Cu₂Pd₃Se₄. Between the shortest and the longest contacts in Li₂Pt₃Se₄, a difference of 30 pm occurs. Considering the same distance criterion for the compounds Ag₂Pd₃Se₄ and Cu₂Pd₃Se₄, only two Se atoms appear inside the coordination sphere. For a tetrahedral coordination of Ag1a and Cu1a, the minimum spread of the four Ag–Se and Cu–Se distances are 35 pm and 39 pm, respectively. Significantly more pronounced is the difference between the shortest and longest contacts, assuming a trigonal prismatic coordination of Ag (61 pm) and Cu (77 pm) as shown for Li₂M₃X₄ (M=Pd, Pt, X=S, Se). As a result of the split positions, the effect increases even more, and the coordination of Ag and Cu moves towards a square planar coordination sphere.

Fig. 5:

The distorted trigonal prismatic selenium coordination polyhedra of the Li, Ag, and Cu atoms in Li₂Pt₃Se₄ (top), Ag₂Pd₃Se₄ (middle), and Cu₂Pd₃Se₄ (down) are shown, respectively. All distances are given in pm.