2.1 Superionic conductors

In the first example, we reacted the simple lithium or sodium salts Li₄[SnS₄] or Na₄[SnS₄] with P₂S₅. By choosing a suitable ratio of the reactants, the lithium superion conductor Li₁₀SnP₂S₁₀ or the sodium superion conductor Na₁₁Sn₂PS₁₂ was obtained [10]. Both belong to the best-known sulfide-based solid-state ion conductors, with ion conductivities on the order of magnitude of liquid electrolytes [11]. In contrast to the liquid materials, the solid-state ion conductors do not bear the risk of inflammation. However, the use of such materials still requires optimization of electrode contacting and the electrochemical stability, which is currently actively investigated worldwide. Somewhat better Li⁺ ion conductivities were obtained for the homolog Li₁₀GeP₂S₁₀ [12], but this compound is more expensive (by a factor of 2–3), owing to the presence of Ge instead of Sn. The price of the sodium conductor would be again lower, in accordance with the larger abundance of sodium versus lithium in the earth crust. The crystal structures of the said compounds and their conductivities are illustrated in Fig. 2.

As mentioned above, although the ion conductivity properties of such quaternary salts are excellent, their electrochemical stability is limited and they may undergo chemical reactions with electrode materials, which currently inhibit their use in commercially available batteries. One possible way to overcome this drawback, for instance, is the developing taylor-made interphases (solid electrolyte interphases) between the electrode and the electrolyte [13].

2.2 Supersphere cages

The second example addresses reactions in ionic liquids. These are—as per definition—salts that melt (markedly) below 100 °C, many of them even below or around room temperature, as the combination of cations (mostly organic) and anions (mostly weakly coordinating) does not allow for a sufficiently large lattice energy of the crystal structure [14]. Such melts have very good solvent properties, yet without the disadvantageous properties of common solvents, such as low boiling points and high vapor pressure, which causes the release of (toxic) vapor. Moreover, many ionic liquids are readily synthesized and may be easily varied by the choice of cations and anions. This is why ionic liquids are often referred to as “designer solvents” that enable, for instance, the synthesis of new solid compounds at relatively low temperatures between room temperature and moderately elevated temperatures (≤200 °C), whereas solid materials have been traditionally accessed via high-temperature approaches (≥700 °C).

We use ionic liquids for the synthesis of novel chalcogenidometalates from a diversity of chalcogenidometalate salts based on [E¹⁴E¹⁶₄]⁴⁻ tetrahedra [15] or their molecular extension into an adamantane-type [E¹⁴₄E¹⁶₁₀]⁴⁻ anion [16]. Our studies also indicated the possibility for a dimensional reduction of three-dimensional solids [17], such as the known phase K₂Sn₂Se₅ [18]. By choosing suitable reaction conditions in terms of the nature of the ionic liquid, the nature and amount of an auxiliary amine, and the temperature, it is possible to produce salts of a supersphere anion, [Ge₂₄Sn₃₆Se₁₃₆]²⁴⁻. It is the largest known main group element polyanion, which has a spherical ball-type structure surrounding a large inner void space of more than 800 Å³ [16]. The reactant K₄[Ge₄Se₁₀]·3H₂O is reacted with SnCl₄·5H₂O in different ionic liquids of the general type [C_n(C₁)C₁Im]X, with C_n representing an alkyl chain with n = 4, 6, 8, or 10 carbon atoms, C₁ representing methyl groups on the imidazolium (Im) cation, and X representing a 1:1 mixture of Cl⁻ and [BF₄]⁻ anions. In the presence of a small amount (0.1 mol %) of the auxiliary amine 1,3-dimethylmorpholine, large red crystals of salts of the supersphere anion are selectively observed, whereas larger amounts of the auxiliary serve to form different, non-molecular chalcogenidometalate structures. Depending on the nature of the ionic liquid cation, the ball-shaped anions are arranged differently in the crystal structures, with the cations located in between them—in some cases also occupying the void space. The compounds are capable of trapping small molecules like I₂ and to heterolytically cleave the I–I bond. Fig. 3 illustrates the synthesis of this compound, its crystal structure, and the behavior towards a solution of I₂ in cyclohexane. Besides these hints toward molecular activation, the anion and the corresponding crystal structures have attracted attention by their mere aesthetics.

2.3 White-light generators

As briefly outlined above, another extension of the basic concept is the replacement of one of the chalcogenide ligands by an organic group. Small organotetrel chalcogenide clusters of the general formula [(RE¹⁴)₄E¹⁶₆] or [(RE¹⁴)₃E¹⁶₄]⁺ are obtained by a reaction of organotetrel trihalides RSnX₃ (X = Cl, Br) with A₂E¹⁶ (A = alkali metal, SiMe₃). Dependent on the nature of the organic group R, the structure of the [(RE¹⁴)₄E¹⁶₆] clusters is either adamantane-type (thus related to the previous example) or “double-decker”–type, with two coplanar (E¹⁴₂E¹⁶₂) four rings that are bridged by two E¹⁶ atoms [19]. For R comprising a donor atom (O, N), back-coordination to the E¹⁴ occurs, which leads to the preferred formation of the double-decker–type motif with five-coordinate E¹⁴ atoms, whereas organic ligands without donor atoms stabilize the adamantane-type architecture with four-coordinate E¹⁴ atoms [20]. The inorganic cores of these small clusters may be further modified by a reaction with transition metal compounds; if the organic ligands bear reactive groups, an extension of the ligand shell with complementarily functionalized organic compounds is also possible [21]. However, even the small clusters themselves show uncommon properties, which were observed only recently.

As indicated in Fig. 4 (left), the reaction of an organotin trichloride with a styryl ligand affords an air-stable amorphous powder. Its composition was determined by means of mass spectrometry as [(StySn)₄S₆] (Sty = styryl, C₆H₄CHCH₂; Fig. 4, right). As the molecular structure within amorphous compounds cannot be determined by X-ray diffraction, quantum chemistry was applied, indicating the cluster to adopt the adamantane-type structure (Fig. 4, center). This structure type lacks molecular inversion symmetry, which prompted us to investigate potential nonlinear optical properties. A typical nonlinear phenomenon is second harmonic generation, in which monochromatic light is converted into light with exactly doubled frequency (i.e., with half of the original wavelength) upon transmission of a material. Although this was what we expected to find, [(StySn)₄S₆] does not exhibit second harmonic generation properties. Instead, and most notably, the amorphous powder is capable of transforming infrared laser light from an inexpensive continuous wave (CW) GaAs laser diode into white laser light without loss of the directionality, as illustrated in Fig. 5 [22]. In the meantime, further amorphous compounds comprising adamantane-based clusters [(RE¹⁴)₄E¹⁶₆] (R = phenyl, styryl; E¹⁴ = Si, Ge, Sn; E¹⁶ = S, Se) were identified that show the same characteristics [23].

Supercontinuum generation processes in general have been already described for many systems, however only upon use of very expensive pulsed lasers, and not necessarily along with directionality. This indicates that the new findings may revolutionize the generation of directed white-light, which may be useful for projection or medical purposes. The exact mechanism behind this new type of supercontinuum generation is not known so far, but apparently it is dependent on missing long-range order in the material and the presence of aromatic ligands for electron delocalization.

2.4 Multimetallic clusters

The fourth and final type of compound is taken from the mentioned reactions of the P₄ analogs (E¹⁴₂E¹⁵₂)²⁻ or (E¹³E¹⁵₃)²⁻, with formally charged (E¹⁴)⁻ or (E¹³)²⁻ atoms being isoelectronic with neutral (E¹⁵) atoms. We note that, for the elemental combination of E¹⁴ and E¹⁵, such anions also represent heavier homologs of a hypothetical tetrahedral dimer of cyanide (CN)⁻, being isoelectronic with the lighter P₄ homolog N₂. The chemistry behind these studies belongs to the contemporary chemistry of Zintl ions, which themselves have been known for more than a century, but the post-synthesis reactions of which have only recently been investigated. Besides organic derivatization and linkage of clusters [24] or their oxidative degeneration to form oligomers, polymers, opal structures, or, finally, new elemental modifications [25], a third direction of how to modify Zintl anions is the formation of heterometallic clusters upon reactions with transition metal compounds [9]. In most cases, the obtained clusters represent so-called “intermetalloid” clusters, that is, main group atom cages that embed one (or several) interstitial transition metal atom(s). Given suitable electron counts, such systems may be also viewed as superatoms [26]. Intermetalloid clusters are very reactive and do not only readily undergo oxidation, they may also deprotonate (basic) solvents and they further seem to support C–C coupling reactions [27]. This is one of the reasons, why we are interested to find out more about these heterometallic molecules, besides very fundamental objectives, such as their intriguing structures and uncommon bonding situations.

The concept of using binary anions as precursors has two significant consequences. First, it enables the introduction of many different binary combinations of group 13–15 elements into such systems, as the (E¹⁴₂E¹⁵₂)²⁻ or (E¹³E¹⁵₃)²⁻ anions have been synthesized and isolated as their [K(crypt-222)]⁺ salts (crypt = 4,7,13,16,21,24-hexaoxa-1,10-diazabicyclo[8.8.8]-hexacosane, C₁₈H₃₆N₂O₆) for many elemental combinations, namely Ga/Bi, In/Bi, Tl/Bi; Si/P, Ge/P, Ge/As, Ge/Sb, Sn/Sb, Sn/Bi, Pb/Sb, Pb/Bi [3]. Second, the structural variety is naturally larger than upon use of homoatomic systems, in which the adjustment of negative charges to fit with the requirements of the structure and crystallization cannot be done by simply choosing between (E¹⁴)⁻, (E¹³)²⁻, or (E¹⁵) atoms, respectively, for a specific atomic position. Consequently, a large variety of ternary clusters with tunable electronic properties have been accessed.

Fig. 6 gives an overview of the clusters that were obtained with binary Zintl anions of the named elemental combinations and different types of transition metal atoms [3a,3b,3f,4,27,28]. Some elemental combinations do not lead to the formation of small tetrahedral anions but to larger molecules instead, because of a distinct mismatch of atomic sizes. An example is (Ge₄Bi₁₄)⁴⁻, which shows an element segregation on the molecular level, in accordance with the immiscibility of Ge and Bi solids [29].

The bonding situation and structures of most of the shown clusters can be rationalized by one of two well-known concepts. Electron-precise clusters, like the binary tetrahedral precursors and the aforementioned (Ge₄Bi₁₄)⁴⁻ anion, or ternary clusters enclosing early transition metal or lanthanide/actinide ions, comply with the pseudo-element concept by Zintl [30a], Klemm [30b], and Busmann [30c]. Here, the number of bonds per atom follow the 8-N rule, with N being the number of valence electrons plus the formal (negative) atomic charge of the respective atom. An electron-precise interpretation also applies for ternary clusters with interstitial Co²⁻ ions or Ni atoms that can be regarded from the viewpoint of fulfilling the 18-electron rule at the transition metal atom. Pseudo-element situations are also found in ligand-like adducts of otherwise unchanged homoatomic or binary Zintl anions to transition metal ions like Cu⁺, Zn²⁺, or Cd²⁺. Electron-deficient clusters with their typical deltahedral shape are formed with interstitial Cu⁺ or Zn²⁺ ions and are in accordance with Wade–Mingos rules [31]. In this regard, ternary transition metal–main group element clusters [9d] do not differ from their binary transition metal–main group element cluster cousins [9a,9b,9c]. However, in both the binary and ternary intermetalloid/heterometallic clusters, there are examples known that cannot be understood in terms of any of these concepts. Examples of corresponding binary clusters are [M@Ge₁₀]³⁻ (M = Fe, Co) [32] and the paramagnetic [Ru@Ge₁₂]³⁻ anion [33]. These clusters feature three-bonded Ge atoms within non-deltahedral 10-atom or 12-atom cages that, however, cannot be simply regarded as comprising pseudo-group 15 atoms. Instead, the anions have to be viewed as entities comprising delocalized cluster orbitals (with the unpaired electron of the latter example residing in a Ge₁₂-derived molecular orbital). This finding is in contrast to a ternary (and diamagnetic) analog of the Ru/Ge cluster, [Ta@Ge₈³⁻ = [Ta⁵⁺@(Ge⁻)₈As₄]³⁻ [3a], in which the pseudo-element concept is perfectly fulfilled. At first glance, the binary cluster [U@Bi₁₂]³⁻ [3h], in which a doughnut-shaped Bi₁₂ ring surrounds a U atom is a relatively simple case, consisting of an interstitial U³⁺ ion and a macrocyclic (Bi₁₂)⁶⁻ ligand. However, the magnetic behavior and the reproduction of structural features by means of DFT calculations indicated [U⁴⁺@(Bi₁₂)⁷⁻]³⁻ to be the correct formulation, including a highly unexpected paramagnetic (Bi₁₂)^7−• unit. Another extraordinary example is the ternary cluster anion [Pd₃@Sn₈Bi₈]⁴⁻ [28e], which accords to neither an electron-precise model nor a typical Wade–Mingos-type electron-deficient description. Again, the help of sophisticated quantum chemistry was required to show that a neutral Pd₃ ring is enclosed within an uncommon (Sn₈Bi₆)⁴⁻ shell.

Among the known cases of ternary intermetalloid/heterometallic cluster anions (Fig. 6), one recognizes the predominance of certain structural motifs for certain types of interstitial atoms, which indicates the latter to act as a template during the formation of the clusters. This is obvious for the 12-atom, 13-atom, or 14-atom cages that are formed with interstitial lanthanide, actinide, or group five metal ions, for which a clear correlation between cage size and ionic radii of the inner atom is observed [28b,28i]. Hence, we asked ourselves how such clusters might form around the d-block or f-block metal atoms. On the basis of the isolation of an intermediate cluster, [Ta@Ge₆As₄]³⁻, which crystallized upon the extraction of a quaternary K/Ta/Ge/As solid mixture, and the observation of the known Ge₁₀⁴⁻ anion in the reaction solutions, we proposed the following formation cascade to occur at the formation of such nondeltahedral clusters (Fig. 7) [3a].

As observed very recently, some of the clusters seem to be in the position to activate small molecules. As monitored by electrospray ionization mass spectrometry, the cluster [{Ru(cod)}Tl₂Bi₆]²⁻ (cod = cycloocta-1,5-diene) can form an O₂ adduct upon release of the cod ligand. Quantum chemical calculations indicated the activation of an O₂ molecule attached to the Ru atoms, as can be seen from the significant elongation of the O–O bond (Fig. 8) [28g]. Further studies regarding the potential use of such species for redox catalysis are underway.