Sr₇N₂Sn₃: a layered antiperovskite-type nitride stannide containing zigzag chains of Sn₄ polyanions

1 Introduction

A wide variety of oxides having the perovskite-type structure (ABC ₃), or related structures with components containing blocks of perovskite structure units, have been synthesized, and the properties and applications of these materials have been examined [1]. Recently, compounds having the perovskite-type structure in which the cation and anion sites are exchanged have been synthesized and characterized [2, 3]. As an example, there are 118 records in the inorganic crystal structure data base of ideal antiperovskite compounds with the space group Pm 3 ‾ m containing nitrogen atoms at the octahedral B site [4]. Layered (modular) perovskite oxides such as Dion–Jacobson, Ruddlesden–Popper, and Aurivillius phases are organized with two-dimensional infinite slabs sliced along the [100] direction of the ideal perovskite structure and separated from one another by cations or cationic motifs [1]. Slabs cut parallel to the ideal perovskite (110) plane are stacked in phases related to Ca₂NbO₇.

In contrast, there have been few reports on layered antiperovskite compounds. (PrCa₃N)Bi₂ has been shown to have the antistructure to that of K₂NiF₄, which is a Ruddlesden–Popper phase [5]. The crystal structure of Sr₁₁Ge₄N₆ contains layers of the nitridogermanate group [GeN₂]⁴⁻ and a distorted antiprism layer of [Sr_8/2Ge]⁴⁺ sandwiched between slabs of double antiperovskite layers cut in the perovskite [100] direction [6]. Some subnitrides and nitride-related compounds containing N³⁻ at the centers of Sr/Ba octahedra as also occurs in antiperovskite subnitrides, and incorporating nitridometallate and Zintl polyanions, have been prepared by the Na flux method [7]. The present paper reports the preparation, crystal structure, and electronic structure of a new strontium nitride stannide having a layered antiperovskite structure.

2 Experimental

All experimental manipulations described herein were carried out in an Ar-filled glove box (MBRAUN, O₂, H₂O < 1 ppm). Sn powder (99.9%, grain size ≤45 μm, Fujifilm Wako Pure Chemical Co.), Sr (99.99%, Aldrich), NaN₃ (99.9%, Tokyo Chemical Industry Co., Ltd.), and Na (99.95%, Nippon Soda Co. Ltd.) were used to synthesize compounds in the Sr–Sn–N system. In these experiments, 0.046 g of Sn, 0.068 g of Sr, 0.030 g of NaN₃, and 0.036 g of Na (for a Sn:Sr:NaN₃:Na molar ratio of 0.5:1.0:0.6:2.0, Sample I) or 0.033 g of Sn, 0.099 g of Sr, 0.022 g of NaN₃, and 0.066 g of Na (Sn:Sr:NaN₃:Na molar ratio of 0.5:2.0:0.6:2.0, Sample II) were placed in boron nitride crucibles (99.5%; inner diameter, 6.5 mm; depth, 18 mm; Showa Denko K.K.), that were then sealed in a stainless steel (SUS 316) container (inner diameter, 10.7 mm; depth, 80 mm). The container was subsequently heated to T = 1073 K for 4 h, held at this temperature for 2 h, and then cooled to 573 K for 96 h. Following this, the heater power was turned off and the sample was allowed to cool naturally in the furnace. The cooled container was opened in the glove box and the specimen was transferred to a glass container and then heated under a reduced pressure of approximately 0.1 Pa at 573 K for 12 h to evaporate residual Na. Single crystals were extracted from the samples while observing the specimens using an optical microscope attached to the glove box.

The chemical composition of each single crystal was analyzed using electron probe microanalysis (EPMA) and wavelength dispersive X-ray spectrometry (WDX) (A-8200, JEOL). The single crystals were fixed in glass capillaries using grease after which the capillaries were sealed on a hot Pt filament. Single crystal X-ray diffraction (XRD) data were collected using a single-crystal diffractometer (D8 QUEST, Bruker AXS) using MoKα radiation and employing the Apex3 software package [8]. Numerical and multi-scan absorption corrections were performed using the program Sadabs [9]. Refinement of the atomic coordinates and displacement parameters was performed with the Shelxl-2014 program [10]. The Vesta software [11] was employed to visualize the crystal structures.

CCDC 095786 (Sr₇N₂Sn₃), 2095787 (SrSn) and 2095839 (Sr₃N_0.77Sn) contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The joint ICSD and CCDC database via www.ccdc.cam.ac.uk/data_request/cif.

To provide an insight into the electronic structure of the ternary nitride stannide, density-functional-based methods were employed by means of the projector augmented wave (PAW) method [12] as implemented in the Vienna ab initio simulation package (VASP) [13], [14], [15], [16], [17]. Prior to the bonding analyses, full structural optimizations which included both lattice parameters as well as atomic positions were accomplished utilizing the aforementioned approach. Correlation and exchange in all computations were described by the generalized gradient approximation of Perdew, Burke, and Ernzerhof (GGA-PBE) [18], while the energy cut-off of the plane wave basis set was 500 eV. A 7 × 4 × 10 k-points set was used to sample the first Brillouin zone, and all computations were expected to be converged as the energy difference between two iterative steps fell below 10⁻⁸ (and 10⁻⁶) eV per cell of the electronic (and ionic) relaxations.

A chemical bonding analysis was completed based on an examination of the Mulliken and Löwdin charges [19] as well as the projected crystal orbital Hamilton populations (pCOHP) [20] for Sr₇N₂Sn₃. The Mulliken and Löwdin charges were obtained by subtracting the gross population of a given atom from its number of valence electrons, while the Hamilton populations [21, 22] are determined by weighting the off-site entries of the densities-of-states with the respective Hamilton matrix elements. The constructions of the pCOHPs and the computations of the Mulliken and Löwdin charges require the use of local basis sets, whose nature is in stark contrast to that of the delocalized one of the plane-wave basis sets. Therefore, the results of the plane-wave-based computations had to be projected into a local representation using transfer matrices as well as all-electron Slater-type orbitals. These transformations were accomplished by means of the Local-Orbital Basis Suite Towards Electronic-Structure Reconstruction (LOBSTER) code [23, 24]. The DOS and pCOHP curves of Sr₇N₂Sn₃ were visualized using the wxDragon code [25], while the program Diamond4 [26] was employed to create the representation showing the distributions of the Mulliken and Löwdin charges in the crystal structure of the ternary nitride stannide. In addition to the pCOHP and Mulliken as well as Löwdin population analyses, the electron localization function (ELF) [27, 28] of Sr₇N₂Sn₃ was also computed by means of the aforementioned PAW method and analyzed as well as visualized utilizing the program Vesta [11].

3 Results and discussion

Sample I is mainly comprised of aggregates of black metallic platelet single crystals with a maximum length along the sides of 500 μm and a thickness of 20–50 μm. Figure 1 shows a scanning electron microscopic (SEM) image taken for a single crystal of Sample I with the EPMA apparatus. The EPMA data obtained from the platelet single crystal indicated that the specimen was made of 58.1% Sr, 36.7% Sn, 2.6% N, and 1.9% O (for a total 99.3 mass%), while the Na concentration was less than 0.01%. The oxygen in this specimen was likely present in a surface oxide or hydroxide layer generated while transferring the sample into the EPMA apparatus under ambient air. The Sr:Sn molar ratio in this crystal was calculated to be approximately 7:3, which agreed with the formula of Sr₇N₂Sn₃ determined from the subsequent single crystal XRD analysis. In addition to the platelet single crystals of Sr₇N₂Sn₃, black granular single crystals of SrSn with a maximum size of approximately 60 μm were also found in Sample I. The orthorhombic lattice parameters of this compound were a = 5.0658(3), b = 12.0252(6), and c = 4.4945(2) Å, consistent with the reported values (a = 5.064, b = 12.04, c = 4.494 Å [29]; a = 5.033(6), b = 12.00(2), c = 4.493(3) Å [30]; a = 5.045(5), b = 12.040(1), c = 4.495(5) Å [31]). Sample II was found to comprise granular single crystals of the antiperovskite Sr₃N _x Sn less than 70 μm in diameter. The cubic lattice parameter of these crystals was a = 5.24180(10) Å, similar to the value of a = 5.2351(5) Å reported for a powder sample of Sr₃N_0.74Sn [32].

Figure 1:

Scanning electron microscopic (SEM) image of a Sr₇N₂Sn₃ single crystal.

All XRD reflections from a platelet single crystal of Sr₇N₂Sn₃ could be indexed using the orthorhombic lattice parameters a = 10.4082(2), b = 18.0737(4), and c = 7.43390(10) Å, and the possible space groups determined from systematic extinctions were P2na and Pmna. A crystal structure model for the Pmna space group was generated using the Intrinsic Phasing module of the Apex3 program package and the structure was refined to wR2 and R1 (all data) values of 3.98 and 1.80%, respectively. The results of this structure analysis are summarized in Table 1, and the refined atomic coordinates, anisotropic displacement parameters, and selected interatomic distances are provided in Tables 2, 3, and 4, respectively. The results for the SrSn and Sr₃N _x Sn single crystals obtained in the present study are also included in these tables.

The N1 and N2 occupancies in Sr₇N₂Sn₃ were refined to be 1.034(10) and 0.991(11), and these occupancies were fixed at 1.0 in the final refinement. The nitrogen content, x, in Sr₃N _x Sn determined based on the N1 occupancy refinement was 0.767(17), which was close to the value reported for Sr₃N_0.74Sn powder [32]. A lattice parameter of a = 10.4708(7) Å (twice that for the antiperovskite-type structure due to nitrogen defect ordering) has been reported for Sr₃N_0.59Sn prepared using a Li flux [33]. One-half of this value, 5.2354 Å, agreed with that of Sr₃N_0.74Sn (5.2351(5) Å), and no superstructure reflections corresponding to the nitrogen defect ordering were observed in the single-crystal XRD images of the present study.

The crystal structure of Sr₇N₂Sn₃ consists of four layers of corner-sharing NSr₆ octahedra separated by [Sn₄] four-atom chains as shown in Figure 2(a) and (b). The plane of the four-layer slabs corresponds to the (110) plane of the ideal antiperovskite structure. This structure can be expressed as an inverse structure related to that of Ca₂Nb₂O₇, in which slabs made of four perovskite layers [Ca₂Nb₄O₁₄] are separated by two layers of Ca atoms (Figure 2(c) and (d)) [34]. For this reason, Sr₇N₂Sn₃ can also be expressed as [Sn₄][Sn₂N₄Sr₁₄], following the representation of Ca₂[Ca₂Nb₄O₁₄]. However, the [Sn₂N₄Sr₁₄] layers are stacked in the b-axis direction, while the [Ca₂Nb₄O₁₄] layers are stacked in the a-axis direction with a c/4 shift (Figure 2(b) and (d)). The results of single crystal XRD experiments indicate that the plate surface of the single crystal shown in Figure 1 equated to the slab surface perpendicular to the b axis, while the step lines observed on the crystal were in the a-axis direction.

Figure 2:

Projections of the crystal structure of Sr₇N₂Sn₃ on (a) the crystallograpic bc plane (0.0 ≤ a < 0.5) and (b) the ab plane, and for Ca₂Nb₂O₇ [34] on (c) the ab plane (0.0 ≤ c < 0.5) and (d) the ac plane.

Figure 3(a) shows the antiperovskite region of Sr₇N₂Sn₃, in which Sn3 is at the A site surrounded by 12 Sr atoms, and there are two N1-centered Sr1₂Sr2₂Sr4₂ octahedra and six N2-centered Sr1₂Sr3₂Sr5₂ octahedra. The Sn3‒Sr interatomic distances range from 3.6029(6) to 3.8173(5) Å and the average distance of 3.717 Å is close to the Sn1‒Sr1 distance of 3.70651(8) Å determined for antiperovskite Sr₃N_0.77Sn (Table 4). N1‒Sr and N2‒Sr distances are in the ranges of 2.4726(18)–2.6451(19) and 2.60580(12)–2.6637(19) Å, respectively, and the N1-centered Sr octahedra are more distorted than the N2-centered ones. The average N1‒Sr and N2‒Sr distances are 2.576 and 2.626 Å, respectively, and the average values of 2.601 Å is a little shorter than the N1‒Sr1 distance of Sr₃N_0.77Sn (2.62090(6) Å).

Figure 3:

Atomic arrangements in crystals of Sr₇N₂Sn₃ (a) in the anti-perovskite slab and (b) around Sn3 and the polyanion Sn2‒Sn1‒Sn1‒Sn2.

Each Sn1 in the [Sn₄] four-membered zigzag chain is surrounded by nine Sr atoms. The coordination environment of each Sn2 is similar to that of Sn3 in the antiperovskite unit except that one of the 12 Sr atoms at the C site is replaced with a Sn atom at Sn1. The Sn2‒Sr distances are in the range from 3.4753(4) to 3.9084(5) Å, and the average distance of 3.717 Å is longer than the averages of the Sn1‒Sr (3.689 Å) and Sn3‒Sr (3.681 Å) distances, but close to the Sn1‒Sr1 distance of Sr₃N_0.77Sn (3.70607(11) Å). The shortest value of 3.4753(4) Å among the Sn2‒Sr distances is in the range of Sn1‒Sr1 distances (3.4267(4)–3.6197(3) Å, av. 3.498 Å) observed for sevenfold coordinated Sn atoms in SrSn (Table 4).

The Sn1‒Sn1 and Sn1‒Sn2 bond lengths in the [Sn₄] four-membered chains (that is, the Sn2‒Sn1‒Sn1‒Sn2 chains) were determined to be 3.0932(5) and 3.1118(6) Å, respectively, and the ∠Sn2‒Sn1‒Sn1 bond angle was 104.187(16)°. The bond lengths are comparable to the Sn‒Sn distances in metallic β-Sn (3.02192(10)–3.1810(3) Å [35]).

Similar [Sn₄] four-membered zigzag chains have been reported in the structure of Ca₇Sn₆ [36], in which the Sn1‒Sn1 and Sn1‒Sn2 bond lengths of [Sn₄] are 2.8987(6) and 2.9238(6) Å, respectively. These are shorter than the lengths found in Sr₇N₂Sn₃, although a ∠Sn2‒Sn1‒Sn1 angle of 104.353(16)° in Ca₇Sn₆ is consistent with that in Sr₇N₂Sn₃. [Sn₂] dumbbells are also contained in Ca₇Sn₆. Assuming Sn‒Sn single bonds, formal charges of −6 and −10 are expected for [Sn₂] and [Sn₄] based on the (8−N) rule [37]. In this model, two electrons are unaccounted for because 14 electrons are provided from the seven electropositive Ca atoms in Ca₇Sn₆. Siggelkow et al. proposed the resonance hybrid structures [Ge–Ge]⁶⁻ +  [ Ge − ⃛ Ge − ⃛ Ge − ⃛ Ge ] 8 −  ↔ [Ge=Ge]⁴⁻ + [Ge–Ge–Ge–Ge]¹⁰⁻, for Ca₇Ge₆, Sr₇Ge₆, and Ba₇Ge₆, all of which are isostructural with Ca₇Sn₆. In these structures, the positive and negative formal charges are balanced by the introduction of π-bonding character between the Ge atoms. Since the Ge–Ge bond lengths decrease in the order of Ca₇Ge₆ (2.529 Å) to Sr₇Ge₆ (2.528 Å) to Ba₇Ge₆ (2.502 Å) in the case of the [Ge₂] dumbbells and increase in the order of Ca₇Ge₆ (2.526, 2.571 Å) to Sr₇Ge₆ (2.545, 2.592 Å) to Ba₇Ge₆ (2.547, 2.594 Å) for the [Ge₄] zigzag chains, the resonance is shifted to the end of these two series of structures with increasing atomic number of the alkaline earth elements. The average Sn1‒Sn1 and Sn1‒Sn2 bond lengths in the [Sn₄] four-membered chains in Ca₇Sn₆ (2.911 Å) are longer than the Sn‒Sn distance of 2.80991(5) Å in α-Sn having the diamond structure [38], but comparable to the Sn‒Sn single bond distance in infinite zigzag chains, [ Sn 1 ] 1 ∞ in CaSn (2.918(11) Å) [39], and longer than the Sn–Sn length in the [Sn₂] dumbbells (2.8735(7) Å) in Ca₇Sn₆. Thus, Ca₇Sn₆ could be expressed as (Ca₇)¹⁴⁺[Sn₂]⁴⁻[Sn₄]¹⁰⁻.

The formal charge of Sn atoms at Sn3 in the antiperovskite slab of Sr₇N₂Sn₃ is −4, so that the charge of the [Sn₄] four-membered chain must be −8 to meet the requirement of charge neutrality ({(Sr₁₄)²⁸⁺(N₄)¹²⁻(Sn₂)⁸⁻}[Sn₄]⁸⁻). However, the Sn1‒Sn1 single bond length obtained for the [ Sn 1 ] 2 − 1 ∞ zigzag chain in SrSn (Table 4) is 2.9420(4) Å and thus shorter than the Sn1‒Sn1 and Sn1‒Sn2 lengths (3.0932(5) and 3.1118(6) Å, av. 3.1025 Å) in Sr₇N₂Sn₃. Therefore, a formal charge of [Sn₄]⁸⁻ in conjunction with π bonding between Sn atoms are not supported by these bond lengths.

Assuming that each [Sn₄] chain has single Sn‒Sn bonds (even though the bond lengths are longer than the expected values of approximately 2.9420(4) Å), the formal positive and negative charges could be balanced by introducing oxygen as {(Sr₁₄)²⁸⁺ (N₂)⁶⁻(O₂)⁴⁻(Sn₂)⁸⁻}[Sn₄]¹⁰⁻. The wR2 (all data) factors obtained from crystal structure refinements with the Sr₇(N_0.5O_0.5)₂Sn₃, Sr₇(N1O2)Sn₃, and Sr₇(O1N2)Sn₃ models were 4.32, 4.58, and 4.26%, respectively, all of which were larger than the value for Sr₇N₂Sn₃ (3.98%).

Gärtner and Korber have pointed out that the bond lengths in Zintl polyanions are determined by the configuration and the valence electron numbers of the constituent anions as well as the size of the electropositive cations around the polyanions [40]. In fact, SrSn is isostructural with CaSn, and the Sn‒Sn distance in the infinite zigzag chain, [ Sn 1 ] 2 − 1 ∞ in SrSn (2.9420(4) Å) is longer than that in CaSn (2.918(11) Å). Thus, the Sn‒Sn bond lengths might also be affected by the coordination environment around the polyanions. The Sn1 atom in the zigzag chain in SrSn is surrounded by seven Sr atoms, while Sn1 and Sn2 of the [Sn₄] four-membered zigzag chains in Sr₇N₂Sn₃ are surrounded by nine and eleven Sr atoms, respectively, and these Sr atoms constitute the antiperovskite slabs in the structure (Figures 2 and 3). Therefore, [Sn₄] chains with −8 formal charges might be present in the structure, although the Sn–Sn bond lengths (3.0932(5) and 3.1118(6) Å) are longer than the single bond length of 2.9420(4) Å observed in SrSn. To answer this question, we followed up our experimental work with an examination of the electronic structure of Sr₇N₂Sn₃.

An examination of the DOS curves for Sr₇N₂Sn₃ (Figure 4) reveals that the states in the energy regions near the Fermi level, E _F, largely originate from the Sn-5p and N-2p atomic orbitals with minor contributions from the Sr-5s atomic orbitals. Because the states related to the latter type of atomic orbitals are mainly located above E _F, it can be concluded that the strontium atoms are oxidized, and, hence, act as valence electron donors in Sr₇N₂Sn₃. A closer inspection of the DOS in the energy regions near E _F indicates that the Fermi level is located below a broad pseudogap in the DOS of Sr₇N₂Sn₃, which should lead to metallic conductivity. Furthermore, it is remarkable that E _F is located below and not in the pseudogap, whereas the location of a Fermi level in a pseudogap is typically related to an electronically favorable situation for a given solid-state compound [41], [42], [43], [44], [45]. This circumstance can be explained based on an examination of the –pCOHP curves for Sr₇N₂Sn₃ (Figure 4): as the −pCOHP of all interactions change from bonding to antibonding states below E _F, the populations of further states would lead to a decrease of the bond energy and, accordingly, an electronically less favorable situation [41].

Figure 4:

(a)–(c) Net as well as atom- and orbital-projected densities-of-states (DOS) curves of Sr₇N₂Sn₃: the orbital-projected DOS correspond to those states providing the largest contributions to the DOS in the energy regions near the Fermi level, E _F; (d) projected crystal orbital Hamilton populations (pCOHP) of diverse interactions in Sr₇N₂Sn₃: the cumulative −IpCOHP per cell values and their percentages to the net bonding capacities have been included using the font color of a respective sort of interaction.

To analyze the nature of bonding in Sr₇N₂Sn₃ in more detail, we also projected the diverse cumulative −IpCOHP per cell values, i.e. the sums of all −IpCOHP per bond values of a given sort of interaction within a unit cell, as percentages to the net bonding capacities – a procedure that has been described in detail elsewhere [46]. The comparison of the percentages to net bonding capacities for the diverse interactions reveals that the largest contributions arise from the Sr–N and Sr–Sn interactions. In this connection it is remarkable that the percentages of Sn–Sn interactions are nearly half as great as those of the Sr–N contacts, while the amount of the latter separations per cell (48) is much higher than that of the homoatomic interactions (6). This is because the Sr−N interactions correspond to much smaller −IpCOHP per bond values (<−IpCOHP per bond> = 0.517 eV) than the Sn–Sn interactions (<−IpCOHP per bond> = 1.937 eV), which also correspond to higher −IpCOHP per bond values than those of the heteroatomic Sr–Sn interactions (<−IpCOHP per bond> = 0.302 eV). How can this finding be explained? As the analysis of the DOS curves for Sr₇N₂Sn₃ clearly indicates that the strontium atoms act as valence electron donors, the Sr–Sn and Sr–N interactions should be depicted as rather polar bonds, whose nature translates into a less bonding character relative to that of the Sn–Sn interactions.

The suggested presence of rather polar Sr–Sn and Sr–N bonds in Sr₇N₂Sn₃ is also backed by Mulliken and Löwdin (Figure 5) population analyses, which underline the role of strontium as a valence-electron donor for the tin and nitrogen atoms. In this connection it is remarkable that the Sn1, Sn2, and Sn3 atoms differ in their respective Mulliken and Löwdin charges: the Sn3 atoms are most reduced among all the different Sn atoms, while the Sn2 atoms are more reduced than the Sn1 atoms. This result indicates that the Sn1 and Sn2 atoms correspond to different electronic configurations which could translate into a structural configuration as expected for a [Sn₄]⁸⁻ polyanion. This conclusion is further backed by an examination of the electron localization function (ELF; Figure 5), as localization domains for high values of the localization parameter (η > 0.8) are located around the Sn1 and Sn2 atoms and could be regarded as Sn1 and Sn2 lone pairs. In addition, the relatively large −IpCOHP per bond values corresponding to the separations between the tin atoms also point to the presence of Sn–Sn interactions with a strong bonding character as expected for covalent bonds; yet, the antibonding character of the Sn–Sn interactions at E _F weakens these homoatomic bonds and, ultimately, translates into enlarged Sn–Sn distances – a circumstance that has also been encountered by previous research on tin-containing compounds [47]. In this connection it should also be noted that the computed Mulliken and Löwdin charges are evidently smaller than the charges predicted by the Zintl–Klemm treatment. This outcome means that the Sr–Sn and Sr–N bonds are not solely constructed of closed-shell species with opposite electric charges as expected for ionic bonds, but that also certain covalent contributions must be present in the heteroatomic interactions in derogation from the Zintl–Klemm ideal.

Figure 5:

(a) Averaged Mulliken and Löwdin (in parentheses) charges in Sr₇N₂Sn₃: the diverse types of strontium polyhedra enclosing the tin and nitrogen atoms are shown in the separate representations; (b) representation of the electron localization function (ELF) for a Sr₇N₂Sn₃ unit cell in the bc plane at x = 0.5; (c) ELF isosurface with η = 0.8: sectional views of localization domains which correspond to values of the ELF larger than 0.8 have been included.

4 Conclusions

Herein, we have reported on the crystal and electronic structure of the ternary compound Sr₇N₂Sn₃, which was obtained from reactions of strontium and tin in the presence of a sodium flux and N₂ generated in the thermal decomposition of sodium azide. The crystal structure of this compound can be related to the inverse structure of the previously reported Ca₂Nb₂O₇, and it is composed of antiperovskite-fashioned slabs separated by four-membered tin units, [Sn₄]. Although for Sr₇N₂Sn₃ the Zintl–Klemm formalism predicts an electron-precise valence electron distribution according to the formula (Sr²⁺)₁₄(N³⁻)₄(Sn⁴⁻)₂([Sn₄]⁸⁻), there are certain structural peculiarities that cannot be explained by applying the aforementioned formalism. From a first principles-based examination of the electronic structure, it is clear that there is a significant valence electron transfer from the strontium to the tin and nitrogen atoms within these heteroatomic contacts, while the bonding nature of the Sn–Sn interactions should be depicted as covalent – a picture that is in full agreement with the predictions based on the Zintl–Klemm treatment. And yet, there are also certain differences between the predictions of the Zintl–Klemm treatment and the results of the first principles-based bonding analyses, which allow us to understand the aforementioned structural peculiarities.