Systematic DFT Studies on Binary Pseudo‐tetrahedral Zintl Anions: Relative Stabilities and Reactivities towards Protons, Trimethylsilyl Groups, and Iron Complex Fragments

Abstract Binary pseudo‐tetrahedral Zintl anions composed of (semi)metal atoms of the p‐block elements have proven to be excellent starting materials for the synthesis of a variety of heterometallic and intermetalloid transition metal–main group metal cluster anions. However, only ten of the theoretically possible 48 anions have been experimentally accessed to date as isolable salts. This brings up the question whether the other species are generally not achievable, or whether synthetic chemists just have not succeeded in their preparation so far. To contribute to a possible answer to this question, global minimum structures were calculated for all anions of the type (TrTt3)5−, (TrPn3)2−, and (Tt2Pn2)2−, comprising elements of periods 3 to 6 (Tr: triel, Al⋅⋅⋅Tl; Tt: tetrel, Si⋅⋅⋅Pb; Pn: pnictogen, P⋅⋅⋅Bi). By analyzing the computational results, a concept was developed to predict which of the yet missing anions should be synthesizable and why. Additionally, the results of an electrophilic attack by protons or trimethylsilyl groups or a nucleophilic attack by transition metal complex fragments are described. The latter yields butterfly‐like structures that can be viewed as a new form of adaptable tridentate chelating ligands.


Introduction
One century after the discovery [1] and the first structural characterization [2] of Zintl anions, their chemistry has become remarkably diverse. [3] However,t here are still challenges that inorganic chemists are facing while seekingf or novel Zintl cluster compositions, structures, and eventually properties.
During the past two decades, it was shown that Zintl anions and their respective salts are excellent starting materials for the generation and isolation of compounds with heterometallic and intermetalloid clustera nions. While the use of homoatomic Zintl anionsy ields bimetallic clusters, reactions of binary Zintl anions, several of which have been knownt oe xist with atoms of groups 13 and 14, 13 and 15, or 14 and 15, usually lead to the formation of trimetallic clusters. It was shown that the largerd egree of freedom upon using binarya nions as precursors is reflected in al argev ariety of new clusters tructures and bondingm odes.
In this context, binaryp seudo-tetrahedral anions that are isoelectronicw ith P 4 or As 4 are of particular interest. They show ah igh reactivity in cluster synthesis, but exhibit al ower overall charget han homoatomicZ intl anions Tt 4 4À (Tt:S i, Ge, Sn, Pb) fore lemental combinationsi nvolving groups 13 and 15 or 14 and 15 in the anions (TrPn 3 ) 2À and (Tt 2 Pn 2 ) 2À (Tr: triel, Tt: tetrel, Pn:p nictogen).H owever,n ot all of these elemental combinationsc ould be experimentally accessed to date. Some of them, like the combination of Ge and Bi, yieldedo ther anionsl ike (Ge 4 Bi 14 ) 4À , [4] so such combinations may systematically be unsuitablef or this pseudo-tetrahedral arrangement. To understand these findings,a nd to investigate the relative stabilitieso ft hese anions for predicting possible extensions of the known collection,w ep erformed extensive and systematic computational studies on binary pseudo-tetrahedral anions with the general formulae (TrTt 3 ) 5À ,( TrPn 3 ) 2À ,a nd (Tt 2 Pn 2 ) 2À .
Another contribution to contemporary Zintl anion chemistry is charger eduction by (element-)organic ligand decoration, as an alternative or addition to the admixture of neutral atoms. [5] So, in as econd step, we studied the reactivity andp ossible functionalization of the pseudo-tetrahedrala nions with protons, trimethylsilylg roups, or organometallic substituents.W e applied density functional theory (DFT) methods (vide infra)t o simultaneously optimize geometricand electronic structures.
The results presented herein provide ad eeper understanding of binary pseudo-tetrahedral Zintl anions. Being valence isoelectronic to P 4 in white phosphorous and As 4 in yellow arsenic,t hese species can eitherb ed escribed by the pseudo-element concept, also referred to as Zintl-Klemm-Busmann concept, [6] or as an ido-type clustera ccordingt ot he Wade-Mingos rules. [7] Since most of the anions are intrinsically disordered in the crystal structureso ft he according salts, this study mighta dditionally be helpfulf or the interpretation of experi-   2À ,m)(TlP 3 ) 2À ,n)(TlAs 3 ) 2À ,o)(TlSb 3 ) 2À ,p)(TlBi 3 ) 2À (Al: yellow,G a: light orange, In:o range, Tl:brown, P: light green,As: bright green,Sb: green,B i: dark green).B ond lengthsa re given in pm.
Mulliken analyses, [10] natural population analyses, [11] as well as population analyses based on occupation numbers [12] were performed to gain furtheri nsighti nto the electronic properties of the anions. It was shown that the negative charges are always delocalized over all four cluster atoms-in stark contrast to the formal charge assignment done by meanso ft he pseudo-element concept, according to which group 15 atoms are neutral, group 14 atoms are charged À1, and group 13 atoms are charged À2. This even holds for the compounds with the highest differences in electronegativity,l ike (InP 3 ) 2À , (TlP 3 ) 2À ,o r( Pb 2 P 2 ) 2À .S till, the highest partial charges, and thus the highest electron densities, are located at the more electronegative sites, indicating that the formalism of the pseudo-element concept is oversimplifying the matter,w hile the trend is still correct.The resultso ft he population analyses also showed that the distribution of the partial chargesp lays no significant role for the stabilities of these anions. Figure 4i llustrates the frontierm olecular orbitals (MOs) of the three types of anions. The MOs of (TrTt 3 ) 5À and (TrPn 3 ) 2À look qualitatively the same, whichi sd ue to their common1 :3 elementr atio. That is except foraf ew group 13/14 anions, where the HOMOÀ1a nd the HOMOÀ2, as well as the LUMO and the LUMO + 1, are interchanged (see Figure S1 in the Supporting Information). The latter has, however,o nly am inor influence on the observed substitution patterns (vide infra). The lowest unoccupied molecular orbital (LUMO) of the (TrTt 3 ) 5À type anions is always located at the triangular tetrel base. The highest occupied molecular orbital (HOMO) is doubly degenerate and extends along the heteroatomic bonds. The main con-tribution to HOMOÀ1s tems from the group 13 atom, hence essentially representing its lone pair.T he doubly degenerate HOMOÀ2o nly shows contributions of the three group 14 or group 15 atoms. As an alternative, the HOMOÀ2a nd the LUMO mayb ev iewed as representing the bonding( occupied) ea nd antibonding (unoccupied) a 1 combination of the inplane tangential p-type atomico rbitals (p-AOs) of ah ypothetical "Tt 3 2À "a nion, hence the 3-center-4-electron( 3c4e) s-type bond of this species, which is isoelectronic with the (C 3 H 3 ) + cation). HOMOÀ1a nd HOMO are based on the bonding( and occupied) a 1 andt he antibonding (andu noccupied) er epresentation of the orthogonal p-AOs, hence the 2p aromatic system of the said species. The latter overlap effectively with the p-AOs of ah ypothetical" Tr 3À "a nion,w hicha dds another 4 p-electrons to the 10 electrons in 5h ighest occupied MOs of the resulting anion. As ac onsequence of the combination of "Tt 3 2À "a nd "Tr 3À ", the binary anions become electron-precise with only minor multi-center bonding (vide infra). For the C 2v - Figure 3. Calculated minimum structures in C 2v symmetry of all possible (Tt 2 Pn 2 ) 2À type anions: a) (Si 2 P 2 ) 2À ,b)(Si 2 As 2 ) 2À ,c)(Si 2 Sb 2 ) 2À ,d)(Si 2 Bi 2 ) 2À , e) (Ge 2 P 2 ) 2À ,f)(Ge 2 As 2 ) 2À ,g)(Ge 2 Sb 2 ) 2À ,h )(Ge 2 Bi 3 ) 2À ,i )(Sn 2 P 2 ) 2À ,j)(Sn 2 As 2 ) 2À , k) (Sn 2 Sb 2 ) 2À ,l )(Sn 2 Bi 2 ) 2À ,m )(Pb 2 P 2 ) 2À ,n)(Pb 2 As 2 ) 2À ,o)(Pb 2 Sb 2 ) 2À ,p)(Pb 2 Bi 2 ) 2À (Si:t urquoise, Ge:sky blue, Sn:blue, Pb:dark blue, P: light green, As:bright green,S b: green, Bi:d ark green). Bond lengths are given in pm.  2À ,asar epresentative for the anions with a1:3 element ratio and of b) (Sn 2 As 2 ) 2À as ar epresentative for the anions with a1:1 element ratio (In:orange,B i: darkg reen,Sn: blue, As:bright green;contour values: AE 0.05 a. u.). It mustb en oted that the orientationo ft he (Sn 2 As 2 ) 2À anion is differentf rom the orientation in Figure 2. symmetric (Tt 2 Pn 2 ) 2À type anions, the LUMO extendsa long the PnÀPn bond, the HOMO along the TtÀTt bond and the HOMOÀ1, as well as the HOMOÀ2, alongt he heteroatomic bonds.
To further investigate the bondings ituation within these clusters,w ecalculated localized molecular orbitals (LMOs) according to Boys' method. [13] This is exemplified in Figure5 for the bonds in (InBi 3 ) 2À and (Sn 2 As 2 ) 2À .S ince we were able to localize the MOs, the overall bonding in this polyhedra can be viewed as being based on regular 2-center-2-electron (2c2e) bonds (in agreement with the pseudo-element concept, rendering Wade-Mingosr ules less appropriate), with the main contributions comingf rom the p-AOs. This is in good agreement with earlier studies. [14] For both clusters, we clearly observe ap olarization of the heteroatomic bonds. Further examination via Paboon, however,a lso indicated additional weak multi-center interactions, which become smallerw ith increasing atomicnumber.
We would like to note here that our findings are in agreement with the alternative way of discussing such clusters as superatoms [15] and according to the Jellium model. [16] As shown for the homoatomic2 0o r4 0v alence electron closedshell species[ Si 4 ] 4À or [Si 9 ] 4À and their analogues, [17] and also for heteroatomic superatoms like the monomeric unit of {[CuSn 5 Sb 3 ] 2À } 2 , [18] the charge is naturallyd elocalized over the whole clusteri ns uch cases, and only the total electron count matters.
From the resultsd iscussed so far,o ne cannotd educew hy some of the binary pseudo-tetrahedral Zintl anionsc ould be synthesized and some could not, or why someo ft he anions seem to be more stable than others. While (Sn 2 Bi 2 ) 2À has been synthesized for the first time 35 years ago and in several different salts since then, [8a,g,h] "(Ge 2 Bi 2 ) 2À "r emains unknown to date. Another example is (GaBi 3 ) 2À ,w hichc an be synthesized, [8e] but readily decomposes and disproportionates into Ga 0 and Bi n qÀ polyanions. [19] Besidest hat, the general claim of heteroatomic bonds to be energetically favored is challenged in somec ases by their destabilization due to big differences of the covalent radii, [20] as impressively shown on the example of the large polyanion (Ge 4 Bi 14 ) 4À ,with strictly separated element types. [4] All of the anions that have been known to date exhibit ratios of the covalentr adii (Q cr )b etween 0.8 and 1.1, e.g., Q cr (Ga:Bi) = 0.82, Q cr (Tl:Bi) = 0.98 or Q cr (Ge:P) = 1.12 (vide infra). Given that the different atomics izes are critical, Q cr should have an effect on the strength of the respective heteroatomic bonds within the pseudo-tetrahedral architecture. To corroborate this, we calculated shared electron numbers( SEN) of the heteroatomic bonds, relative to the homoatomic TtÀTt bonds in (TrTt 3 ) 5À and relative to the PnÀPn bonds in (TrPn 3 ) 2À and (Tt 2 Pn 2 ) 2À ,r espectively,( SEN rel ). The results that mayb et aken as am easure of the relative stabilityo ft he binary anionsa re given in Ta ble 1.
Ta ble 1s hows two trends for the (TrTt 3 ) 5À and (TrPn 3 ) 2À type anions:(a) The heteroatomic bonds become weaker for heavier triel atoms, as expected. (b) The opposite is the case as the tetrel or pnictogen atoms become heavier,t hat is, the GaÀBi bond is stronger than the TlÀBi bond, despite as eemingly unfavorablev alue of Q cr .A tf irst glance,t his seems to contradict the conclusions drawn above, accordingt ow hich the difference of the covalent radiis hould be as small as possible. An explanation for this discrepancy can be derived from the molecular structures. Longer TtÀTt or PnÀPn bonds allow ac loser approacho ft he triel atoms towards the center of the trigonal bases,s imilar to ac lose-packed lattice:aGa atom fits better in the space between three Bi atoms, than the Tl atom does, leadingt oabetter overlap of the involved orbitals and to the formation of stronger bonds (see Figure 4). Anions with Q cr smallert han the ideal value of 1.00 should therefore be more stable than those with Q cr > 1.00.
This, however,d oes not explain why (GaBi 3 ) 2À is actually less stable than (InBi 3 ) 2À according to the experiments,w hilei t should be more stable accordingt othe data in Ta ble 2. We obviously have two competing factorsh ere-oneb eing ex-  [a] It must be notedt hat sharede lectronnumbersa re not to be mistaken as an actual absolute measure for the bondstrength; they just serve to illustrate trends.
plained above.T he second factor seems to be that the anion formationo ri ts stabilityi sg enerally hampered if the deviation from the ideal value of Q cr = 1.00 becomes too large. In addition, charge distributiona nd the size of the anionsi nc omparison to that of the cation of choice will definitely play ar ole in affectingt he lattice energy and thus the formation of ar espective isolable salt. As ar esult, anions of the type (TrTt 3 ) 5À and (TrPn 3 ) 2À with Q cr close to 1.00 should be most stable, whereby deviations towards larger values are less tolerable than deviations towards smaller values. As mentioned above,t he (Tt 2 Pn 2 ) 2À type anionsc an formally be viewed as being composed of two homoatomic dumbbells. The lack of as imilars tabilizing effect as for the (TrTt 3 ) 5À and (TrPn 3 ) 2À type anions resultsi nt hese anionsb eing much more sensitivet od eviations from Q cr = 1.00 towards smaller values. This is shown by the fact that (GaBi 3 ) 2À with Q cr = 0.82 wass uccessfully synthesized, whereas (Ge 2 Bi 2 ) 2À with Q cr = 0.81 is unknown ands eems to be systematically inaccessible. In contrast, deviations towards larger ratios (Q cr > 1.00) seem to be less problematic, cf. the experimentallyo bserved (Ge 2 P 2 ) 2À anion (Q cr = 1.12);s till, the anion undergoes ar e-organization in solution to form al arger Zintl anion in the course of severald ays, indicating its metastability.
All of the considerations above lead us to the conclusion that af ew more anions at least should be accessible. Ta ble 2 lists all 48 anionsa long with their respective Q cr values. Known anions are italicized and highlighted in bold and dark grey, while anions that we predict to be synthesizablea re highlighted in bold.
In conclusion, the discovery or development of suitable synthetic methods seems to be criticalf or accessing the yet missing anions-notably,i tt ook more than 80 years from the prediction of Bi 7 3À and Bi 11 3À polyanions to their isolation as salts. [19,21] Reactivities and substitution patterns Because of the high anionic charges of À2a nd À5, respectively,s alts of the pseudo-tetrahedral binary anionsa re more difficult to handle in common (organic)s olvents than speciesw ith lower or no charge. We therefore performed extensive studies on possible electrophilics ubstitution with protons and trimethylsilyl (TMS) groups, and on possible nucleophilic substitutions with organometallic substituents, in order to reduce the cluster charge, and therebym odifyp roperties like solubility and reactivity.A sn one of these attempts provedp ossible for the herein discussed binary anions in experimental work so far, we aimed at examining the effects of substituents on the molecular structures by theoretical work. We limited our efforts to structures with two substituents, neutral (TrPn 3 R 2 ), (Tt 2 Pn 2 R 2 ), as well as anionic (TrTt 3 R 2 ) 3À and (Tt 2 Pn 2 R 2 ) 2À (in case of organometallic substituents).
While first theoretical studies of as ingly protonated P 4 tetrahedrons uggested the proton to be located at an apex of the molecule, [22] protonation of the tetrahedral edges was predicted to be energetically favored later on. [23] This substitution pattern was recently verified experimentally,a nd by meansofn ew DFT and coupled-cluster calculations, for the first known protonated variants of tetrahedral 20 valence electron species, [P 4 (m-H)] + [24a] and [Si 4 (m-H)] 3À , [24b] as well as ap rotonated unit used as ligand to ZnPh 2 in [(m-H)(h 2 -Ge 4 )ZnPh 2 ] 3À . [24c] Additionally,S cheschkewitz and co-workers recently reported (Si 5 R 4 ) 2À . This anion can be interpreteda sa nS i 4 4À tetrahedron with two substituents (thus reducingt he overall charge) and an additional SiR 2 moietya cting as an electrophile and bridging one of the SiÀSi bonds of the underlying tetrahedral structure motif. [24d] In our theoretical study,w ea dded two protons to the anions (in order to compensate for all chargeso ft he (TrPn 3 ) 2À and (Tt 2 Pn 2 ) 2À species), and explored all possible protonation sites of the resultings pecies.
Anions of the type (TrTt 3 ) 5À and (TrPn 3 ) 2À show the same preferred protonation pattern. As expected, the hydrogen atoms bridget wo heteroatomic bonds by involving the doubly degenerate HOMO (see Figure 4) in the energetically favored isomers, resulting in C s -symmetric molecules. The bridge is most asymmetricf or the two lightesth omologues, (AlSi 3 H 2 ) 3À and (AlP 3 H 2 ), where the Ha toms are much closer to the Al atoms.
For the (Tt 2 Pn 2 ) 2À type anions, bridgingo ft he TtÀTt bond and one of the heteroatomic bonds is most favorable, which is realizedb yi nvolvement of HOMO and HOMOÀ1. In all cases, the (m-H)-bridged bonds are significantly elongated, by 8-10 % for all three cluster types (see Figure 6f or (InBi 3 H 2 )a nd (Sn 2 As 2 H 2 )a se xamples). This is in perfect agreement with the recente xperimental findings for the isoelectronic speciesm entioned above.
The presence of 3-center-2-electron( 3c2e) bonds upon m-Hbridging is supported by corresponding SEN values, and the 3c2e bonds becomes tronger for values of Q cr close to 1.00. They are hence the weakest( on average)f or clusters (TrPn 3 H 2 ). Ta ble 3l ists corresponding data for (InBi 3 H 2 )a nd (Sn 2 As 2 H 2 )a s examples. NPAa nd Mulliken analyses illustrate the impact of the protonation on the electronics tructures. While the negative charge was relativelye venly distributed over all four atoms in the naked anions, the three atoms that are involved in the bonds to the two hydrogen atoms (e.g.,I n1, Bi1, and Bi2 in Figure 6a) have positive partial chargesn ow.I nt urn, the unsubstituted (semi)metala tom (e.g.,B i3 in Figure 6a)a nd both hydrogen atoms are partially negatively charged in the overall neutral cluster.I na ccordance with their larger electronegativity as compared to any of the p-block( semi)metals,t he hydrogen atoms thus undergo an umpolungt owards ah ydridic charac-ter.A si llustrated in Figure 6, 3c2e bonds involving heteroatomic tetrahedral edges are slightly polarized towards the more electronegative (semi)metala tom. For the series of anions (TrTt 3 H 2 ) 3À ,t he negative charge is delocalized over all atoms. Yet, the largest electron density is also localized at the Ha toms and the (unsubstituted) tetrel atom.
Geometry optimizations with the Ha toms being forced into ap ositiono ver the trigonal faces wered one to study stability trends. The molecular structuresr elaxed into local minimat hat were significantly higher in energy than the structures exhibiting edge-bridging, as illustrated in Figure 7f or the series (InPn 3 H 2 )w ith Pn:P ,A s, Sb, Bi. The energy differences with respect to the globalm inimum decreases in the order Pn:P> As > Sb > Bi. Hence, the stabilizationo ft he edge-bridged isomer is more significant for underlying pseudo-tetrahedra with larger size differences of the (semi)metala toms, thus larger Q cr in the given series.
In case of the clusters (Tt 2 Pn 2 H 2 ), HOMOÀ2c ontributes to the bond, which hasastabilizing effect. Nevertheless, the difference between the globala nd the local minimum is still large, e.g.,97kJmol À1 for (Sn 2 As 2 H 2 ).
While protonating the binary pseudo-tetrahedral anions always yields clusters with bridged edges,t he pictureb ecomes more complex for the (hypothetical) addition of trimethylsilyl (TMS) groups.A fter the geometry optimizations, we find three different (dominant) substitution patterns for clusters {TrTt 3 (SiMe 3 ) 2 } 3À ,{ TrPn 3 (SiMe 3 ) 2 }a nd {Tt 2 Pn 2 (SiMe 3 ) 2 }, respectively.T hese follow ac ontinuous trend from ap reference of edgebridging to ap reference of terminal bonding, which can be put down to the anions' different tendency to form efficient 3c2e bonds.
The derivatization of homoatomic, tetrahedral main group compounds with four alkylsilyl groups yielding tetrahedranelike structures was previously reported. [25] The steric demando f four alkylsilyl groups resultsi nt he preference of terminal Figure 6. a) Calculated minimum structure of (InBi 3 H 2 )a nd illustration of one of the two LMOs representing the two 3c2e bonds; b) Calculated minimum structureo f(Sn 2 As 2 H 2 )and LMOs representing the two 3c2e bonds (In: orange, As:bright green,B i: dark green,Sn: blue, H: grey;c ontour values: AE 0.05 a. u.). It must be noted that the orientation of the (Sn 2 As 2 H 2 )molecule is different from the orientation of the parent( Sn 2 As 2 ) 2À anion in Figure 2.  bondingo vere dge-bridging. In {Tl 4 (C{SiMe 3 }) 4 }, for instance, the alkylsilyl groups are tilted sideways. [25d] This, however,i s more likely due to the steric demando ft he substituents, than being an indication for at endency to forming3 c2e bonds, as we see them in the clusters studied herein (vide infra). Anions of thet ype {TrTt 3 (SiMe 3 ) 2 } 3À prefer edge-bridging under formation of two 3c2e bonds. Notably,o nly here, the above mentioned interchanging of HOMOÀ1a nd HOMOÀ2i n anions of the type( TrTt 3 ) 5À plays ar ole:i nc ases, in which these MOs are interchanged,t he bridging involves oneh eteroatomic bond and one of the opposing homoatomicb onds (see Figure 8f or {GaGe 3 (SiMe 3 ) 2 } 3À as an example), with the exception of {InGe 3 (SiMe 3 ) 2 } 3À ,i nw hich two homoatomic bonds are bridged.T he other order of MOs, however,l eads to bridging of two heteroatomic bonds instead (see Figure S2). The relative energy difference between both isomer types is smallest for anions comprising heaviest atoms, such as {TlPb 3 (SiMe 3 ) 2 } 3À . In all cases, addition of TMS groupsc auses elongation of the involved bonds,b y4 -10 %. This elongation becomes less prominentf or values of Q cr % 1.00, irrespective of the observed substitution pattern.
Due to the larger differences of the electronegativity of the involved atoms, clusters of the type {TrPn 3 (SiMe 3 ) 2 }t end the least to forming 3c2e bonds, like discussed for the protonated species. Here, the preferred substitution pattern includes two terminal bonds, with one TMS group bonded to the triel atom, and the other one bonded to one of the pnictogen atoms (see Figure 9f or {InBi 3 (SiMe 3 ) 2 }a sa ne xample). The heteroatomic bonds in the 4-vertex units are elongated by up to 10 %( most distinctly for heaviest atoms, while lighter homologues show no or only slight elongations). In this conformation, the two TMS groups are the furthest apart of all examples discussed herein.W er ecognizea ne xception from this pattern only for {InP 3 (SiMe 3 ) 2 }a nd for the sub-series including the heaviestt riel thallium, {TlPn 3 (SiMe 3 ) 2 }( with Pn:P ,A s, Sb;s ee Figure S3), where the TMS groups are bondedt ot wo pnictogena toms. We ascribe this to the fact that both InÀSi and TlÀSi bonds are significantly disfavored in comparison with PnÀSi bonds, owing to the larger differences in covalent radii.
The silylation pattern for clusters of the type {Tt 2 Pn 2 (SiMe 3 ) 2 } finally represents ac ombinationo ft he aforementioned cases: Here, we find the TtÀTt bond to be bridged, thereby involving the HOMO of the naked anion under formationo fa3c2e bond. The TtÀTt bond lengths are again elongated by up to 10 %, depending on the respective Pn atoms, with the largest effect observed for the heaviest Pn atoms, again.T he second TMS group is bonded as at erminal substituent to one of the Tt atoms, or to one of the Pn atoms, depending on the similarity of the covalent radiio ft he respectiveT to rP na toms and the Si atoms of the terminal TMS group (Si TMS ;s ee Figure10 for {Sn 2 As 2 (SiMe 3 ) 2 }a sa ne xample). In the two heaviest homologues, {Pb 2 Sb 2 (SiMe 3 ) 2 }a nd {Pb 2 Bi 2 (SiMe 3 ) 2 }, an additional bond is formedb etween one of the Pb atoms and the Si TMS atom (see Figure S4). For this class of clusters, silicon atoms cause differences again:a ll species of the sub-series {Si 2 Pn 2 (SiMe 3 ) 2 }( with Pn:P ,A s, Sb, Bi)p refer another substitution pattern. Here, both TMS groups are bonded to the two Si atoms of the former pseudo-tetrahedron,t hereby forming a chain-like Si 4 moiety (SiÀSi:222-236pm, see Figure S5).
In summary,t he attachment of two protons to pseudo-tetrahedral, binary anionso fp -block (semi)metals leads to mostly hydridic substituents. The substitution patternsf or substitutions with TMS groups are different buta tt he same time characteristic for the chosen combination of main group elements, with some exceptionsf or subseries involving the lightest(Si)o r heaviest(Tl) congener(s) of the involved group(s).
The calculations discussed so far addressed the electrophilic attack of binarya nions. While the mono-protonation of Si 4 4À to form [Si 4 (m 2 -H)] 3À was reported, [24c] this has been the only example to date involving H + as an electrophile, and ac orresponding result remains elusive for any othert etrahedral Zintl anions. The attachment of electrophilic transition metal complex fragments like (MesCu) + ,( ZnPh) + ,o rZ n 2 + to a4 -vertex anion has also been unknown for binary anionsi nvolving atomsf rom different main groups, but it was reported for tet-

Chemistry-A European Journal
Full Paper doi.org /10.1002/chem.202001379 rahedral anions of the type Tt 4 4À (includingS i/Ge mixtures). [24c, 26, 27] Binary anionsh ave so far shown to readily undergo cluster fragmentation and re-arrangement instead, which seems to be induced or catalyzed by the transition metal atoms, thereby yieldingo ther beautiful heterometallic and intermetalloid cluster structures. [27] To the best of our knowledge, an ucleophilic attack towards any Zintl anion has not yetb een reported until today, [27] which soundsr easonable at first glance, owing to the presence of negative charges on the surface of the anionic molecules. However,i tw as previously shown that P 4 or As 4 can be activated via nucleophilica ttack by transition metal complexes to form butterfly-like moieties. Thef irst compound reported to emerge from such reactions was[ {Cp''Fe(CO) 2 } 2 (m:h 2:2 -P 4 )] (Cp'': h 5 -C 5 H 3 tBu 2 ). [28] More recently,S cheera nd co-workers used [{Cp'''Fe(CO) 2 } 2 (m:h 2:2 -Pn 4 )] (Pn:P ,A s; Cp''': h 5 -C 5 H 2 tBu 3 )t os how that such moieties can subsequently act as chelating ligands for Lewis acidic species, like the cationic complex fragment [Cu(NCMe)] + . [29] Inspired by this work, and beyondt he background that the binary Zintl anionsp ossesst he same electron count and very similar frontier orbitals,w et ried to expand this concept to (hypothetic)s pecies with binary cluster cores, [{CpFe(CO) 2 } 2 (m:h 2:2 -Tt 2 Pn 2 )] 2À (Tt:S i, Ge, Sn, Pb;P n: P, As), hence based on pseudo-tetrahedral species, in which two of the Po rA sa toms were replaced with tetrel atoms.T or educe the computational effort, we used the smaller Fp substituent (Fp:C pFe(CO) 2 · ;C p: h 5 -C 5 H 5 )a nd performed geometry optimizationsf or the resulting molecules.
Since the LUMO of the naked anions expands alongt he re-spectiveP n-Pn edge, an ucleophilic attack addresses the Pn atoms under cleavage of said edge. The Fp substituents then form terminalF e ÀPn bonds, resulting in the desired butterflylike structures. Our hypothesis was that the formal replacement of two of the pnictogen atoms by tetrel atoms would notably influence the electronic situation at the pnictogen atoms in the bridgehead positions. We expected to observe heterometallic chelating ligands with tunable properties at the pnictogena toms, generally suitable for the tailored coordination of various Lewis acids.
The optimized structure of [{CpFe(CO) 2 } 2 (Ge 2 P 2 )] 2À ,a sa ne xample of the resulting type of anionic molecules, is shown in Figure 11.I nc ontrast to the results obtained by the Scheer group, where the Cp'''g roups are turned sideways and in opposite directions with respect to the "butterfly" orientation, both of the Cp ligands are orientatedi nt he same way,y et away from the open "butterfly" edge. This difference is most likely due to the smaller steric demando ft he Fp moieties in the calculated species. Ta ble 4s ummarizes relevant structural and electronic data.
The Pn···Pn distances become larger with increasing atomic number of the tetrel atoms. This is accompanied by smaller SEN values, which can be viewed as av ery rough approximation of the trend of the remaininge lectron density between the two Po rA sa toms. The dihedral angles stay relatively constant, between 888 and 908 on average. NPAa nd Mulliken analysesf urther showedt hat the pnictogen atoms are partially negativelyc harged, while the tetrel atoms exhibit ap ositive partial charge. In the homoatomic analogues,[ {CpFe(CO) 2 } 2 (P 4 )] and [{CpFe(CO) 2 } 2 (As 4 )],the partial charge distribution is exactly the opposite (hence questioning the equivalence of the "nucleophilic attack" in these two cases). The negative partial charge at the bridgehead sites increases as the difference in electronegativityb etween the two main group elements gets larger. Hence, also the softness of the respective pnictogen (donor) atom,a ccordingt ot he Pearson concept, increases. Therefore, these anions shouldb eless suitable as ligands for electrophiles than the homoatomic reference clusters, whichm ay explain the lack of experimental evidence so far.
To check this hypothesis, we added a[ Cu(NCMe)] + fragment to the butterfly-shaped moieties in silico, and performed geometry optimizationsf or the anions [Cu(NC-Me)(Tt 2 Pn 2 {CpFe(CO) 2 } 2 )] À (Tt:S i, Ge, Sn, Pb;P n: P, As). Indeed, the results are slightly different from the experimental findings for the homoatomic phosphorous or arsenic analogues, where the [Cu(NCMe)] + fragment is pointing away from the P 4 or As 4 unit in an orientation perpendicular to two of the Pn-Pn edges.H ere, we alwaysf ind the [Cu(NCMe)] + fragment to be tilted sideways, thus forming an additional TtÀCu bond. Population analyses showedt he strengtho ft hese bonds to be roughlyo ft he same order of magnitude as for the other PnÀ Fe and the PnÀCu bonds, thus corroborating the Lewis-basic character of the respective chelating ligand. Further back-don- Figure 11. Calculated minimum structure of [{CpFe(CO) 2 } 2 (m:h 2:2 -Ge 2 P 2 )] 2À as at ypical example for moleculeso ft he type [{CpFe(CO) 2 } 2 (m:h 2:2 -Tt 2 Pn 2 )] 2À (Ge:sky blue, P: light green,F e: dark yellow,O:red, C: black, Ha toms are omitted for clarity). ation from the transition metal complex fragment into the LUMO of the ligand was not observed. The calculated minimum structure of [Cu(NCMe)(Ge 2 P 2 {CpFe(CO) 2 } 2 )] À ,a sa ne xample of the whole series, is displayed in Figure 12, along with a cutout of the illustration of its HOMO. Figure 12 also shows that the newly formed TtÀCu bond results from an interaction of the Cu atom's d z 2 atomico rbital with HOMOÀ4o ft he chelating [{CpFe(CO) 2 } 2 (Ge 2 P 2 )] 2À moiety.T hisH OMOÀ4i sm ainly located at the tetrel atoms and between them (see also Figure S6), thus rendering the formation of said TtÀCu bond energetically favorable.
Ac omparison of the absolute energies shows that the hypothetical reactions of the respective butterfly-like anionsw ith a [Cu(NCMe)] + fragment are highly exoenergetic( see Table S13). Furthermore, we found the P-containing speciess lightly favored compared to the As-containing analogues.T his can again be explained by the Pearson concept and is in agreement with our previousresults.
To verify the observedc onformation to be preferred, we forced the [Cu(NCMe)] + fragment into aperpendicular position by symmetry (C 2v ), and calculated the absolute energies of the corresponding isomersf or all elemental combinations studied in this section. We found these structures to be local minima on the potentialh ypersurface within the given symmetry restrictions. Their total energies are between 22 kJ mol À1 and 82 kJ mol À1 above the total energies of the isomers with tilted [Cu(NCMe)] + moieties, depending on the elemental combination of the underlying binary "butterfly" core. In the global minimum structures, the coordination sphere around the Cu atom is thus not trigonal planar as in the P 4 -based and As 4based structures, but strongly distorted "tetrahedral". Hence, the [{CpFe(CO) 2 } 2 (m:h 2:2 -Tt 2 Pn 2 )] 2À type anionsd on ot only act as bidentate, but as tridentate chelating ligands. This and the increased softness of the respective pnictogen atoms, due to their highern egative partial charge, suggest that these anions are more suitable as ligandsf or softer Lewis acids that tend to tetra-coordination, such as Hg 2 + ,P t 2 + or Ag + ,w hich will be studied in future work.

Conclusions
In summary,w epresented calculated globalm inimum structures for all binary pseudo-tetrahedral Zintl anionso ft he type (TrTt 3 ) 5À ,( TrPn 3 ) 2À ,a nd (Tt 2 Pn 2 ) 2À ,c omposed of p-block( semi)metals.W ed escribed structural trends, and found possible answers to the question,w hy some of these cluster anionss eem to be systematicallye lusive in experimental work. At the same time, our findings allow to predict that some of the yet not isolated speciesshould be generally accessible.
Furthermore, we studied the effect of substitution with protons or trimethylsilylg roups,a nd we discussed the behavioro f these anions upon substitution with nucleophilesa nd their possible applicability as (tridentate) chelating ligands for Lewis-acidic transition metal cations.
The findings presented herein might be of help for synthetic chemists (includingo urselves) and their approaches towards more of thesefascinating compounds.

Experimental Section
Computational details:A ll calculations were undertaken by means of the program system TURBOMOLE, [30] applying the TPSS functional [31] and def2-TZVP basis sets [32] with the corresponding auxiliary bases [33] and effective core potentials (ECPs) at In, Tl, Sn, Pb, Sb, and Bi. [34] The electronic structures were investigated by Mulliken [10] and natural population analyses (NPA), [11] as well as by population analyses based on occupation numbers (Paboon) [12] implemented in TURBOMOLE. COSMO, the conductor-like screening model, [35] was used to compensate the negative charges (standard values, e = 1). Localized molecular orbitals were obtained via Boys' method. [13] The verification of the minima structures was done by analysis of the force constants. [36] For more details, see the Supporting Information.