Li vs. Zn substitution in Li₁₇Si₄ – Li_17–ε–δZn_εSi₄ connecting the structures of Li₂₁Si₅ and Li₁₇Si₄

1 Introduction

Nowadays lithium-ion batteries (LIBs) are widely used as rechargeable power cell for electronic devices. Lithium intercalation of the anode is one of the important processes in lithium ion batteries. Due to its rather high theoretical lithium capacity, silicon is one of the promising candidates for anode materials in LIBs. Silicon anodes exhibit a more than one order of magnitude larger capacity (4200 mA h g⁻¹ for Li₁₇Si₄) than currently used graphite anodes (372 mA h g⁻¹) [1], [2].

Several phases are observed at the lithium-rich side of the binary phase diagram. These are Li₁₇Si₄, Li₂₁Si₅, the high-temperature phase Li_16.44Si₄ and meta-stable Li₁₅Si₄. None adapts to a composition appropriate for an electron-precise Zintl phase. For a valence-precise compound with isolated Si atoms a ratio of Li to Si of 4:1 is expected according to the formal charge transfer (Li⁺)₄(Si⁴⁻) or Li₁₆Si₄. Most intriguing is meta-stable Li₁₅Si₄ which is, according to the Zintl formalism, short by one cation to charge balance 16 negative charges of four Si⁴⁻. However, meta-stable Li₁₅Si₄ transforms to thermodynamically stable phases by substituting Li by more electron rich electropositive elements. For Li₁₅Si₄ we found a stabilization through Li by Mg, Zn, and Al substitution and formation of Li₁₄MgSi₄, Li_14.05Zn_0.95Si₄ and Li_14.25Al_0.75Si₄, respectively [3], [4].

Recently Li₁₇Si₄ has been established as the lithium-richest phase in course of a detailed investigation of the Li-rich side of the binary phase diagram and allowed for a very accurate structure determination [5]. This revealed significant differences to previously reported Li₂₁Si₅ [6] to which it is very close in composition (Li₁₇Si₄=Li_4.25Si, Li₂₁Si₅=Li_4.20Si). Interestingly, no substantial phase width is reported for any of these phases.

In this context we also investigated the partial substitution of Li by Zn in Li₁₇Si₄ and report here on the particular site preference of the Zn atoms.

2 Experimental section

3 Results and discussion

Single crystals of Li_17−ε−δZn_εSi₄ were obtained through Li flux syntheses by using the atomic ratio Zn:Si of 1:3 and Li removal at high temperature by isothermal centrifugation. Diffraction intensities were indexed in space group F4̅3m which is non-centrosymmetric, and therefore inversion twinning of two enantiomorphic species was considered for the structure refinement of a single crystal. All Li atoms could be refined with anisotropic displacement parameters. In the unit cell there are 80 silicon atoms located at 24g, 24f and two 16e sites, and 340 Li atoms located at 13 independent atomic sites (four 48h, 24g, 24f, six 16e, and 4a). Only the Li13 site (4a) shows a significant statistically mixed Li/Zn occupation (see Table 2). Careful inspection of the residual electron density revealed some density in the vicinity of Li7 (16i). Applying a split model for this position (Li7A and Li7B) led to unreasonably short Li7B–Li13/Zn13 distances of 1.366(9) Å. Consequently the statistically mixed Li/Zn position which is not fully occupied remains empty in correlation with the occupation of Li7B (Fig. 1). The total s.o.f. for the Li13/Zn13 site was restricted to 0.832 (Table 2). The only short interatomic distance of 2.23(1) Å remains between two Li7B atoms (Fig. 1e). The Li7A tetrahedron is occupied by a statistical Li/Zn mixture. The distances between the atoms Li13/Zn13 and Li7A are 2.388(3) Å (Fig. 1d) and thus slightly shorter than those reported for Li₁₇Si₄ (2.395(2) Å) [5]. This small difference can be explained by the fact that Zn has a smaller metallic radius than Li (1.57 Å for Li and 1.37 Å for Zn) [14]. The distances between the Li7A atoms are similar to those observed in Li₁₇Si₄ (3.900(4) Å comparing to 3.9101(16) Å). One could speculate that the 4a site is occupied only by Zn atoms, but this would lead to an even bigger vacancy at 4a and does not correlate with the split of the Li7A–Li7B atoms.

Fig. 1:

Coordination of the atomic site 4a and split model for the atom Li7. (a) Coordination of the atomic site 4a; (b) the situation with Li7A atoms; (c) the situation with Li7B atoms; (d)–(e) distances between the 4a site and the Li7A or Li7B atoms.

The crystallographically determined composition is Li_16.942(1)Zn_0.025(1)Si₄ where the Zn content is at the level of an impurity. However, 0.12 at.% of Zn in the compound does appear as a significant value on the respective atomic site and qualitative EDX measurements confirmed the presence of Zn in the compound (see Figure S2, Supporting Information available online), but it was not possible to determine the Zn amount precisely by this method. The anisotropic displacement parameters for the Li13/Zn13 site are about 2–3 times smaller than for other Li sites (Table 3) which also supports the idea of the presence of the heavier Zn atom at this site. The results are in line with the observation that Li₁₇Ge₄ also incorporates Zn under the formation of Li_17−εZn_εGe₄ (ε=0.0052(12)) [15], with a statistical Li/Zn mixture at the 4a atomic site (around 2.6% of Zn comparing to 12.4% for Li_17−ε−δZn_εSi₄). However, this site was fully occupied in the Ge compound, and a splitting of the neighboring atoms around the 4a site has not been observed.

Currently Li₁₇Si₄ is the Li-richest phase in the binary Li-Si phase diagram, and it has been shown that its crystal structure is closely related to another Li-rich phase, Li₂₁Si₅ [5]. In both compounds there are four sites of higher symmetry 4a (0, 0, 0), 4b (1/2, 1/2, 1/2), 4c (1/4, 1/4, 1/4), and 4d (3/4, 3/4, 3/4), which are all unoccupied for Li₂₁Si₅, but in which exclusively 4a is occupied in the model for Li₁₇Si₄. These special sites have a characteristic surrounding of neighboring atoms (Fig. 2a–d). The first shell consists of 4 Li atoms which form a tetrahedron. The next shell contains either 4 Si atoms or 4 Li atoms which in both cases form tetrahedra. These two tetrahedra interpenetrate each other to give a star, which is enclosed by an octahedron of either Li or Si atoms.

Fig. 2:

Atomic shells with atom labels around the special positions in space group F4̅3d: (a) around the Li/Zn-filled site 4a (0, 0, 0); (b) around the unoccupied 4b site (1/2, 1/2, 1/2); (c) around the unoccupied site 4c (1/4, 1/4, 1/4); (d) around the unoccupied site 4d (3/4,3/4,3/4). (e) The NaTl-type arrangement of the Si polyhedra around the special positions 4a, 4b, 4c, and 4d (marked in red, blue, green, and orange, respectively).

As mentioned above, the site 4a in Li_17−ε−δZn_εSi₄ is occupied by a statistical Li/Zn mixture, and the positions are not fully occupied. The first shell of Li atoms around 4a (Fig. 2) is presented by tetrahedra formed by Li7A and Li7B atoms. The split model and the under-occupancy of 4a lead to a rather straightforward model: when the Li7B atom site is occupied, the central 4a site remains empty as in Nesper’s model for Li₂₁Si₅. When the Li7A atom site is occupied, the 4a site is statistically occupied by Li/Zn. This reflects the Li₁₇Si₄ model and is analogous to the situation in Li_17−εZn_εGe₄.

4 Conclusion

After the undisputable structure determination of Li₁₇Si₄ some years ago, the question arises whether the closely related Li₂₁Si₅ exists as well. The difference between the two structures originates from the occupation of the specific site 4a. Whereas Li₂₁Si₅, which an occupation of the 4a site with Li was not found in recent detailed investigations of the Li-rich side of the Li-Si phase diagram, we found now that in Li_17−ε−δZn_εSi₄ the site 4a is occupied by a mixture of Li and Zn and the occupation correlates with the occupation of neighboring atom sites. Thus the new ternary silicide Li_17−σ−εZn_εSi₄ shows the characteristics of Li₂₁Si₅ and can be considered as the missing link between the structures of Li₂₁Si₅ and Li₁₇Si₄.

5 Supporting information

Indexing of the X-ray powder pattern collected for Li_17−ε−δZn_εSi₄ and data of EDX measurements on single crystals of Li_17−ε−δZn_εSi₄ are given as supplementary material available online.