Polyhedral [M2B5] Metallaborane Clusters and Derivatives: An Overview of Their Structural Features and Chemical Bonding

A large number of metallaborane clusters and their derivatives with various structural arrangements are known. Among them, M2B5 clusters and derivatives constitute a significant class. Transition metals present in these species span from group 4 to group 7. Their structure can vary from oblatonido, oblatoarachno, to arachno type open structures. Many of these clusters appear to be hypoelectronic and are often considered as ‘rule breakers’ with respect to the classical Wade–Mingos electron counting rules. This is due to their unique highly oblate (flattened) deltahedral structures featuring a cross-cluster M−M interaction. Many theoretical calculations were performed to elucidate their electronic structure and chemical bonding properties. In this review, the synthesis, structure, and electronic aspects of the transition metal M2B5 clusters known in the literature are discussed. The chosen examples illustrate how, in synergy with experiments, computational results can provide additional valuable information to better understand the electronic properties and electronic requirements which govern their architecture and thermodynamic stability.


Introduction
Chemists know how the electronic structure and geometry of molecules are strongly imbricated. In essence, the electron count governs the shape of molecules because it relates to the closed-shell principle [2]. This principle states that, in general, any molecule is (thermodynamically and kinetically) stable when a significant energy gap separates the fully occupied molecular orbitals (MO) (generally the bonding or nonbonding ones) from the unoccupied MOs (generally the antibonding ones). Based on this, different electron counting rules have been derived, such as the octet and the 18-electron rules which are the simplest and the more widespread ones that many main group and transition metal compounds conform to. These rules are of pivotal importance, providing a rudimentary means to describe the chemical bonding in terms of localized two-center−two-electron (2c−2e) bonds. However, these electron counting rules are not universal and often have a limited scope of application. For instance, the octet and 18-electron rules become insufficient to describe the bonding in systems that have three-dimensional (3-D) structures where extensive delocalization occurs. Hence, new electron counting rules and models have then emerged to rationalize these compounds by the use of multicenter bonding descriptions, of which the three-center−two-electron (3c−2e) model is the simplest. This is particularly the case of many boron-containing 3-D molecules. Such systems containing boron obey the closed-shell principle, DFT calculations suggested that these oblatocloso arrangements are indeed very stable due to an intricate mutual interaction of the bimetallic fragment and the borane cage [25].
To get more precise information about the shape and the electron count of these 'rule-breakers', theoretical computations are generally needed [1,12]. These calculations are useful to illustrate and discuss: (i) about the shape itself, (ii) which isomer is energetically preferred, (iii) about their bonding, electronic structure, electronic properties, etc. They are generally of the density functional theory (DFT) type, often complemented with a cornucopia of modern theoretical tools (quantum theory of atoms in molecules (QTAIM) analysis [30], Laplacian of the electron density [31], electron localization function (ELF) [32][33][34], natural bond orbital (NBO) analysis [35], etc.) which can provide deep insights into the nature of the chemical bonding of these anomalous molecules. Here, in this review, we illustrate the chemical bonding in some of the dimetallaborane compounds, especially those species having a core made of two metal atoms and a few boron atoms.
During the 1990s, the chemistry of metallaborane clusters, which can be viewed as borane clusters with one or more vertices replaced by a transition metal fragment, observed a flourishing development [36][37][38][39][40]. Their chemical and physical properties make them promising candidates for use in different areas such as catalysis [41][42][43][44][45] or materials science etc. [46][47][48][49][50]. With the impetus of Fehlner who started to practice a fruitful synthetic method employing monoborane reagents and cyclopentadienyl metal chlorides [51,52], a large set of new metallaboranes have been characterized. Among them, many dimetallaboranes of early to late transition metals were synthesized and characterized with different size and structure. These dimetallaboranes mainly vary from five-vertex M2B3 clusters to seven-vertex M2B5 clusters and beyond [53][54][55][56][57]. The transition metals in the M2Bn clusters range from group 4 to group 9 with different skeletal electron donations. The structural arrangement of these clusters strongly varies according to their electron count. One of the important and largely explored cluster systems in metallaborane chemistry is the M2B5 clusters which can adopt different skeletal structures. The most largely encountered ones adopt a flattened hexagonal bipyramidal structure with a missing vertex and are hypoelectronic with only 6 seps (I, Figure 1). Due to their flattened and open structure, they are conveniently called oblatonido species with the two metal atoms on opposite sides in the flattened direction so that the metal-metal distances are short enough for direct metal-metal bonding interaction [25]. The examples of this kind of M2B5 species mostly range among the early transition metals such as group 5 in [(CpV)2B5H11] [58] and [(Cp*Ta)2B5H11] [59], group 6 in [(Cp*M)2B5H9] (M = Cr, Mo, W) [60][61][62][63][64][65], and group 7 in [(Cp*Re)2B5Cl5H2] [24] (Cp = C5H5, Cp* = C5Me5). Due to their unique structure, thorough theoretical studies were performed by us and others in order to get some insight into their electronic structure and chemical bonding [66][67][68][69][70]. Later, many heterometallic [63,[71][72][73][74] and chalcogenated derivatives of oblatonido-M2B5 [75][76][77][78] were reported in the literature, which are isostructural as well as isoelectronic. In all these derivatives, one vertex in the parent cluster is replaced by an isolobal metal fragment or a chalcogen atom. Syntheses of most of these derivatives were obtained from the tendency of the unsaturated hypoelectronic clusters to react with metal or chalcogen fragments, which can readily donate further electrons to the cluster framework. A few other structural arrangements with M2B5 composition are also known but far less common (II-IV, Figure 1). Their structures vary from 7-sep oblatoarachno to 9-sep open M2B5 structures. Types II [59,79,80] and III [81] adopt oblatoarachno geometries, which can be generated from the same  A few other structural arrangements with M 2 B 5 composition are also known but far less common (II-IV, Figure 1). Their structures vary from 7-sep oblatoarachno to 9-sep open M 2 B 5 structures. Types II [59,79,80] and III [81] adopt oblatoarachno geometries, which can be generated from the same parent structure. The number of representatives of these structural types is comparatively very limited. The last type of structure (IV, Figure 1) differs strongly from the others and is extremely rare, only known with group 4 metals, Zr and Hf [82,83]. In this review, we focus on the fundamental aspects, i.e., the synthesis, structure, electronic properties, and chemical bonding of these different structural types (I-IV) of M 2 B 5 dimetallaborane clusters and their derivatives.  [61]. These methods signify the high yield targeted synthesis of clusters of 2 and 3. Interestingly, the reaction condition for the formation of 2 and 3 indicates that the addition of the fifth boron requires more forcing conditions in the case of chromium. On the contrary, a clean one-step synthesis occurs for Mo 2 B 5 cluster 3. The reaction of [Cp*MoCl 4 ] or [Cp*MoCl 2 ] 2 with BH 3 ·thf also led to the formation of 3 [62,63]. Somewhat later, 3 has also been synthesized and structurally characterized by Girolami's group from the recation of [Cp*MoCl 2 ] 2 with [NaB 3 H 8 ] [64]. Hence, it indicates 3 as a thermodynamically favorable product as it forms by any combination of the borane source with the molybdenum cyclopentadienyl precursor. The tungsten analog was reported a bit later, in 1999, in Fehlner's group [65] [24]. This compound represents the first example of a dimetallaborane with fully chlorinated boron atoms. The synthesis of 5 is very similar to the synthesis of 2 with borane addition. Later, we reported the Ta analog [(Cp*Ta) 2 B 5 H 11 ], 6 [59]. This compound was synthesized from the reaction of [Cp*TaCl 4 ] with a six-fold excess of LiBH 4 followed by BH 3 ·thf in toluene, along with a few other tantalaboranes which include the substituted analog, [(Cp*Ta) 2 B 5 H 10 (C 6 H 4 CH 3 )], 7 [59]. Apart from 7, a few more substituted derivatives of 6 are known from the C-H activation of arenes and heteroarenes by 6 (these compounds are not described in this review) [84]. The last entry in this series is [(CpV) 2 B 5 H 11 ], 8, which was synthesized from a similar reaction as that of Ta system (Scheme 1) [58].

Structure and Electron Count
The molecular structure of clusters 1-8 was confirmed by single-crystal X-ray diffraction studies. The structure of these clusters was first considered as a bicapped closo-trigonal bipyramid where the M1-M2-B3 triangle is the trigonal plane, B2 and B4 are the apical atoms and two M-M-B planes (M1-M2-B2 and M1-M2-B4) are capped by B1 and B5 atoms ( Figure 2). The core of these molecules is rather highly symmetric (C 2v symmetry). Thus, the metal atoms as well as B1 and B5, and B2 and B4 are symmetry related (symmetry-related atoms are labeled with the same color in Figure 2). Accordingly, experimental 11 B NMR spectra of 1-8 exhibit three different types of boron resonances in a 2:2:1 ratio (Table 1). According to this structural description, an M-M bond is assumed in all of these clusters in agreement with the M-M bond distances measured experimentally (Table 1). Though, the cluster structure of 1-8 is similar, the number and position of the bridging hydrogen atoms varies in different Molecules 2020, 25, 3179 5 of 21 metal analogs to reach their electron count requirement. The differences in 1 H chemical shifts of the M-H-B protons in 1-8 are negligible, but there is a significant difference in the 11 B chemical shifts of the three types of boron atoms in the M 2 B 5 core on moving from first-to third-row transition metals (Table 1).

Structure and Electron Count
The molecular structure of clusters 1-8 was confirmed by single-crystal X-ray diffraction studies. The structure of these clusters was first considered as a bicapped closo-trigonal bipyramid where the M1-M2-B3 triangle is the trigonal plane, B2 and B4 are the apical atoms and two M-M-B planes (M1-M2-B2 and M1-M2-B4) are capped by B1 and B5 atoms ( Figure 2). The core of these molecules is rather highly symmetric (C2v symmetry). Thus, the metal atoms as well as B1 and B5, and B2 and B4 are symmetry related (symmetry-related atoms are labeled with the same color in Figure 2). Accordingly, experimental 11 B NMR spectra of 1-8 exhibit three different types of boron resonances in a 2:2:1 ratio (Table 1). According to this structural description, an M-M bond is assumed in all of these clusters in agreement with the M-M bond distances measured experimentally (Table 1). Though, the cluster structure of 1-8 is similar, the number and position of the bridging hydrogen atoms varies in different metal analogs to reach their electron count requirement. The differences in 1 H chemical shifts of the M-H-B protons in 1-8 are negligible, but there is a significant difference in the 11  the three types of boron atoms in the M2B5 core on moving from first-to third-row transition metals (Table 1).  As stated earlier, bicapped closo-trigonal bipyramidal species 1-8 are characterized by 6 seps or 46 cves. This sep count is expected considering that, according to the capping principle which states that capping a face should not generally modify the favored number of seps [85,86]. For example, a m-capped-n-vertex closo structure (m = 2, n = 7) will have (n − m) + 1, i.e., 6 skeletal electron pairs. The observed 46-cve count is theoretically obtained taking into account the principle of polyhedral  As stated earlier, bicapped closo-trigonal bipyramidal species 1-8 are characterized by 6 seps or 46 cves. This sep count is expected considering that, according to the capping principle which states that capping a face should not generally modify the favored number of seps [85,86]. For example, a m-capped-n-vertex closo structure (m = 2, n = 7) will have (n − m) + 1, i.e., 6 skeletal electron pairs. The observed 46-cve count is theoretically obtained taking into account the principle of polyhedral condensation (or fusion) which stipulates that the total cve count of fused clusters is equal to the sum of the cve of the parent polyhedra minus the cve of the shared unit (atom, pair of atoms, triangle, etc.) [8][9][10]87]. Thus, the bicapped closo-trigonal bipyramidal clusters 1-8 ( Figure 2) can then be viewed as the fusion of an M 2 B 3 closo-trigonal bipyramid (42 cves) with two M 2 B 2 tetrahedra (2 × 40) through two M 2 B triangular faces (-{2 × 38}). This results to a cve count of 46 (42 + 80 − 76), equal to the cve for the corresponding molecular formula.
Alternatively, the structure of these species can also be described as a derivative of a triple-decker sandwich structure in which the middle deck is made by the B 5 plane and the Cp* ligands act as outer decks, somewhat flattened along the M-M axis [25]. In other words, clusters 1-8 adopt a structure which can be derived from a (flattened) closo-hexagonal bipyramid by removal of an equatorial vertex ( Figure 3). These species identified as oblatonido metallaboranes are considered as hypoelectronic compounds with only 6, i.e., n − 1 seps instead of 9 (n + 2) seps which would be expected according to the Wade-Mingos rules. As stated earlier, bicapped closo-trigonal bipyramidal species 1-8 are characterized by 6 seps or 46 cves. This sep count is expected considering that, according to the capping principle which states that capping a face should not generally modify the favored number of seps [85,86]. For example, a m-capped-n-vertex closo structure (m = 2, n = 7) will have (n − m) + 1, i.e., 6 skeletal electron pairs. The observed 46-cve count is theoretically obtained taking into account the principle of polyhedral condensation (or fusion) which stipulates that the total cve count of fused clusters is equal to the sum of the cve of the parent polyhedra minus the cve of the shared unit (atom, pair of atoms, triangle, etc.) [8][9][10]87]. Thus, the bicapped closo-trigonal bipyramidal clusters 1-8 ( Figure 2) can then be viewed as the fusion of an M2B3 closo-trigonal bipyramid (42 cves) with two M2B2 tetrahedra (2 × 40) through two M2B triangular faces (-{2 × 38}). This results to a cve count of 46 (42 + 80 − 76), equal to the cve for the corresponding molecular formula.
Alternatively, the structure of these species can also be described as a derivative of a tripledecker sandwich structure in which the middle deck is made by the B5 plane and the Cp* ligands act as outer decks, somewhat flattened along the M-M axis [25]. In other words, clusters 1-8 adopt a structure which can be derived from a (flattened) closo-hexagonal bipyramid by removal of an equatorial vertex ( Figure 3). These species identified as oblatonido metallaboranes are considered as hypoelectronic compounds with only 6, i.e., n − 1 seps instead of 9 (n + 2) seps which would be expected according to the Wade-Mingos rules.

Electronic Structure
These dimetallaboranes were theoretically studied much earlier to get a better understanding about their chemical bonding. Semi-empirical Fenske-Hall molecular orbital calculations were first performed on Cp analogs (2' and 3') which used to tentatively analyze the electronic factors underlying in the short M-M bond and other different structural features in clusters 2 and 3 [63]. A comparison was made with the Cp analog of the reactive starting precursor, [(Cp*Cr) 2 B 4 H 8 ]. Results indicated that, as expected, cluster 2' was more thermodynamically stable (larger energy gap between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) and showed stronger Cr-Cr bonding interaction (larger Mulliken overlap population) than in the starting species.
Subsequently, a detailed study was performed at the DFT level of theory on the rhenium species [(Cp*Re) 2 B 5 Cl 5 H 2 ], 5 [24]. The primary aim was to locate the Re−H−B bridging hydrogens more precisely which were identified by NMR spectroscopy but not by X-ray crystallography. Two possible isomers corresponding to cis or trans arrangements of the two bridging hydrogens on the open M1B1M2B5 face ( Figure 2) were considered and computed. Similar Re-Re distances were computed in both isomers, comparable to that experimentally measured (2.7641(3) Å). Computed Re−H and B−H bond lengths marginally differ from one isomer to the other. Moreover, the two isomers are nearly isoenergetic. This is consistent with fluxional behavior of 5 observed in solution. The calculated NMR chemical shifts were also very similar, but values of the trans-isomer were closer to the experimental values.
Later, King and coworkers studied various possible geometries for [(CpM) 2 B 5 H 9 ] clusters with early and late transition metals of the second and third rows including the experimentally known species 2 and 3 [68]. Different structures were predicted for each metal, depending upon their electronic requirement, which in most cases, could be related to the Wade-Mingos electron counting rules. Gratifyingly, the bicapped trigonal bipyramid structure of 2' and 3' (Cp analogs of 2 and 3) were found largely thermodynamically more stable than all other possible geometries, consistent with the experimental observation [68].
At the same time, the structural, electronic, and NMR properties of some dimetallaboranes with varied stoichiometry and structure, including the group 6 oblatonido-clusters 2-4, were studied to test the strength of different DFT methods in analyzing their properties [69]. In clusters 2-4, computations indicated that the 11 B NMR chemical shifts of the boron atoms attached directly to the metal atoms (B2, B4, and B3, Figure 2) show a large systematic shift to higher field, whereas the chemical shifts at upper field for the boron atoms connected to the metal atoms via M−H−B bridged bonds (B1 and B5, Figure 2) hardly deviate from Cr to Mo to W species. The nature of the M-M interaction in clusters 2-6 and 8 was further probed using Cp-analog models employing the electron density and its derivatives within the framework of the QTAIM method and the aid of bonding indicators [70]. Results revealed that both through-space and through-bond (via boron atoms of the ring) interactions account for a substantial metal-metal interaction with a covalent bond order around 1.

Synthesis
The isolobal concept mentioned earlier is of value in theoretically generating new mixed main-group-transition-metal clusters [1]. In other words new species where one or more p-block boron fragments are substituted by d-block metal moieties in metallaborane compounds must be possible. Indeed, the first metalla-derivative of oblatonido-M 2 B 5 clusters, [(Cp*Cr) 2 B 4 H 8 Fe(CO) 3 ], 9 was reported by Fehlner and colls. from the room temperature reaction of the unsaturated cluster [(Cp*Cr) 2 B 4 H 8 ] with Fe 2 (CO) 9 [71]. Similarly, they showed later that the reaction with Co 2 (CO) 8

Structure and Electron Count
Molecules M2M'B4 9-19 have a similar oblatonido-hexagonal bipyramidal core to that of the parent M2B5 clusters in which one {BH} unit has been substituted by isolobal metal fragments. In

Electronic Structure
Most of these M 2 M'B 4 species were theoretically studied. A fragment analysis employing Fenske-Hall molecular orbital calculations was first carried out on Cp analogs of 9 [71] and 10 [63] to show that strong bonding interactions are occurring between the M' organometallic fragment and the rest of the molecule. Later, a DFT computational analysis was performed on models 11'-13', Cp analogs of 11-13 [72]. A comparison of 11' with its M 2 B 5 parent 3 indicated that the replacement of a {BH 2 } vertex in the open face of 3 by a {Cp(Mo(CO) 2 } unit in 11' slightly stabilizes the LUMO and destabilizes the HOMO, resulting in a reduced HOMO-LUMO energy gap. A qualitative MO analysis shows that both the HOMO and HOMO-1 mostly account for a strong (Cp)Mo-Mo(Cp) bonding character (Figure 4), similar to that in the parent cluster 3. This strong Mo-Mo interaction is supported by an NBO analysis which shows a Mo apical -Mo apical Wiberg bond index (WBI) [88,89] 1.5 times higher than the corresponding Mo apical -Mo equatorial ones [72]. Though 12' slightly differs from 11' due to the absence of bridging hydrogen interaction with the W(CO) 5 fragment, a molecular orbital analysis shows very similar bonding properties [72]. HOMO, HOMO-1, and LUMO comparable to those in 11' are computed for 12' and 13' (Figure 4). The HOMO and HOMO-1 show an important σ-bonding character between the metal atoms while the LUMO shows some δ-bonding character. In 12', an acceptor-donor interaction between the W(CO) 5 metal fragment and the rest of the cluster is reflected by a NBO charge analysis which shows a rather high negative charge on W (−1.57) with respect to Mo atoms (−0.87 and −0.84).  Theoretical calculations were also performed on 19', the Cp analog of 19 [74]. An MO analysis showed that this time, the HOMO is mainly localized on the equatorial Ru metal atom and to a lesser extent on the apical W atoms with a strong Ru−W σ-bonding character and some W-W δ* character ( Figure 5). The LUMO possesses some W−W δ-bonding character. W−W δ* and δ character for the HOMO and LUMO, respectively, was also observed for the parent molecule 4' (Figure 5). Theoretical calculations were also performed on 19', the Cp analog of 19 [74]. An MO analysis showed that this time, the HOMO is mainly localized on the equatorial Ru metal atom and to a lesser extent on the apical W atoms with a strong Ru−W σ-bonding character and some W-W δ* character ( Figure 5). The LUMO possesses some W−W δ-bonding character. W−W δ* and δ character for the HOMO and LUMO, respectively, was also observed for the parent molecule 4' (Figure 5).

Synthesis
Similarly to the heterometallic M2M'B4 derivatives, a few chalcogen derivatives have also been synthesized in our laboratory where a {BH} unit is replaced by a chalcogen E atom (E = S, Se and Te). These M2EB4 derivatives are only reported with Mo as the metal. Many methods were employed so far for the synthesis of these compounds (Scheme 3). The sulfur derivative, 20, was first synthesized from the reaction of the in situ intermediate generated from the Cp*MoCl4 and LiBH4 reaction with the 2-mercaptobenzothiazol [76]. The Se analog, 21-Ph, was synthesized from a molybdaborane intermediate reacting with the Ph2Se2 reagent [77]. In this structure, one of the terminal hydrogen atoms on boron is substituted by a phenyl group. Finally, the Te-derivatives 22-4Cl and 22-3Cl were synthesized from a similar reaction with Te powder [75,78]. The same method with other chalcogen powders yielded the S and Se analogs as well (Scheme 3) [78].

Synthesis
Similarly to the heterometallic M 2 M'B 4 derivatives, a few chalcogen derivatives have also been synthesized in our laboratory where a {BH} unit is replaced by a chalcogen E atom (E = S, Se and Te). These M 2 EB 4 derivatives are only reported with Mo as the metal. Many methods were employed so far for the synthesis of these compounds (Scheme 3). The sulfur derivative, 20, was first synthesized from the reaction of the in situ intermediate generated from the Cp*MoCl 4 and LiBH 4 reaction with the 2-mercaptobenzothiazol [76]. The Se analog, 21-Ph, was synthesized from a molybdaborane intermediate reacting with the Ph 2 Se 2 reagent [77]. In this structure, one of the terminal hydrogen atoms on boron is substituted by a phenyl group. Finally, the Te-derivatives 22-4Cl and 22-3Cl were synthesized from a similar reaction with Te powder [75,78]. The same method with other chalcogen powders yielded the S and Se analogs as well (Scheme 3) [78].

Structure and Electron Count
The structure of all analogs 20-22 is isostructural to the parent molecule 3, with one of the BH 3 units on the open face replaced by an isolobal chalcogen atom. Hence, these species can also be viewed as oblatonido-Mo 2 EB 4 clusters with electron counts of 6 seps or 46 cves as the parent Mo 2 B 5 species 3. Compound 22-3Cl was only characterized spectroscopically. All the others were characterized crystalographically also. The Mo-Mo, average Mo-B, and B-B distances in these clusters (Table 3) are comparable with those measured for 3. In more detail, the Mo-Mo distance in the Te analog 22-4Cl is slightly elongated but slightly shortened in the S and Se analogs 20 and 21. All of these complexes show four 11 B NMR resonances corresponding to the four unique boron atoms in 1:1:1:1 ratio ( Table 3). The most up-field chemical shift is assigned to the boron atom attached to the chalcogen atom. The value increases upfield in the order Te < Se < S. The other boron atoms are highly deshielded.

Structure and Electron Count
The structure of all analogs 20-22 is isostructural to the parent molecule 3, with one of the BH3 units on the open face replaced by an isolobal chalcogen atom. Hence, these species can also be viewed as oblatonido-Mo2EB4 clusters with electron counts of 6 seps or 46 cves as the parent Mo2B5 species 3. Compound 22-3Cl was only characterized spectroscopically. All the others were characterized crystalographically also. The Mo-Mo, average Mo-B, and B-B distances in these clusters (Table 3) are comparable with those measured for 3. In more detail, the Mo-Mo distance in the Te analog 22-4Cl is slightly elongated but slightly shortened in the S and Se analogs 20 and 21. All of these complexes show four 11 B NMR resonances corresponding to the four unique boron atoms in 1:1:1:1 ratio ( Table 3). The most up-field chemical shift is assigned to the boron atom attached to the chalcogen atom. The value increases upfield in the order Te < Se < S. The other boron atoms are highly deshielded.

Electronic Structure
DFT calculations were carried out to analyze the electronic structure of the Cp-analog models 20'-22' [66,78,90]. The molecular orbital study shows a significant destabilization of their HOMO with respect to that of model 3' (Cp analog of 3) (Figure 6), that suggests a higher reactivity as experimentally observed. In addition, moving from S to Te, there is a gradual decrease in the HOMO-LUMO energy gap. The destabilization of the HOMO is attributed to the introduction of the π-donor chalcogen atoms into the cluster cages.

Electronic Structure
DFT calculations were carried out to analyze the electronic structure of the Cp-analog models 20'-22' [66,78,90]. The molecular orbital study shows a significant destabilization of their HOMO with respect to that of model 3' (Cp analog of 3) (Figure 6), that suggests a higher reactivity as experimentally observed. In addition, moving from S to Te, there is a gradual decrease in the HOMO-LUMO energy gap. The destabilization of the HOMO is attributed to the introduction of the π-donor chalcogen atoms into the cluster cages.

Synthesis
As detailed in the beginning, some metallaboranes of M2B5 composition adopt geometrical arrangements which differ from that of type I. The first oblatoarachno-M2B5 compound of type II (Figure 1

Synthesis
As detailed in the beginning, some metallaboranes of M2B5 composition adopt geometrical arrangements which differ from that of type I. The first oblatoarachno-M2B5 compound of type II (Figure 1), [(Cp*TaCl)2B5H11], 23, was synthesized by our group from the same reaction which yielded clusters 6 and 7 (Scheme 4) [59]. Later, compound 23 and its other halogenated derivatives, 24 and 25, were synthesized by the halogenation of [(Cp*Ta)2B5H11], 6, with CH2Cl2, CH2Br2, and I2 [79]. The monochlorinated derivative [(Cp*TaCl)2B5H10Cl], 26, was also isolated from similar reaction conditions [80].  [81]. Though this synthetic method is similar to the synthesis of 21-Ph, here, the reaction yielded a new structural type rather than forming a derivative cluster.

Structure and Electron Count
Clusters 23-26 were characterized by X-ray crystallography. The average B-B distance is marginally shorter, whereas average Ta-B distances are longer in 23-26 (Table 4) than the corresponding ones in 6. The Ta−Ta distance in 23-26 (ca. 3.23 Å) is substantially longer than that in 6 (2.93 Å). This is too long for a single M−M bond but also too short to have no interaction between the two metal centers. The 11 B NMR spectroscopy of 23-25 shows four resonances in a 2:1:1:1 ratio as the boron atoms B4 and B5 are symmetry related (Figure 7). In 26, due to the lowering of symmetry, each boron atom becomes unique, and five different chemical shifts are observed.

Structure and Electron Count
Clusters 23-26 were characterized by X-ray crystallography. The average B-B distance is marginally shorter, whereas average Ta-B distances are longer in 23-26 (Table 4) than the corresponding ones in 6. The Ta−Ta distance in 23-26 (ca. 3.23 Å) is substantially longer than that in 6 (2.93 Å). This is too long for a single M−M bond but also too short to have no interaction between the two metal centers. The 11 B NMR spectroscopy of 23-25 shows four resonances in a 2:1:1:1 ratio as the boron atoms B4 and B5 are symmetry related (Figure 7). In 26, due to the lowering of symmetry, each boron atom becomes unique, and five different chemical shifts are observed.

Structure and Electron Count
Clusters 23-26 were characterized by X-ray crystallography. The average B-B distance is marginally shorter, whereas average Ta-B distances are longer in 23-26 (Table 4) than the corresponding ones in 6. The Ta−Ta distance in 23-26 (ca. 3.23 Å) is substantially longer than that in 6 (2.93 Å). This is too long for a single M−M bond but also too short to have no interaction between the two metal centers. The 11 B NMR spectroscopy of 23-25 shows four resonances in a 2:1:1:1 ratio as the boron atoms B4 and B5 are symmetry related (Figure 7). In 26, due to the lowering of symmetry, each boron atom becomes unique, and five different chemical shifts are observed.

Metallaborane sep d [M-M] (Å) d avg [M-B] (Å) d avg [B-B] (Å) 11 B NMR (ppm)
[(Cp*TaCl) 2  The core structure for clusters 23-26 was initially considered as a nido-structure derived from a closo-dodecahedron as shown in Figure 7 [59]. But this is not in agreement with the usual Wade-Mingos rules since these clusters possess only 7 seps ([2 × (−1) (Cp*TaX) + 5 × 2 (BH) + 6 × 1 (bridging H)]/2). Alternatively, these clusters may be seen as 7-vertex oblatoarachno-M 2 B 5 species, the structure of which can be derived by the removal of two equatorial boron atoms from a 9-vertex oblatocloso-M 2 B 7 cluster (Figure 8. This makes these clusters hypoelectronic, with fewer valence electrons than generally observed for a canonical arachno-structure of the same nuclearity. But according to a specific electron counting rule proposed by King [25], n-vertex oblatoarachno-clusters should have a non-Wadean electron count of n skeletal electron pairs which matches well with the 7-sep count of 23-26. In 27, the W−W bond length (2.85 Å) bridged by a TePh group is slightly longer than in 4 (2.81 Å) [65]. Its structure can also be viewed in different ways. With a rather short W−W bond length, it can be first considered as a capped octahedron in which the W2B4 octahedron (W1, W2, B2, and B3 as the base, and B1 and B4 as the apical vertices) is capped on the W1-W2-B4 face by a BH3 unit (B5, Figure 9), similar to the core structure found for [(Cp*M)3B4H4] (M = Co, Rh, and Ir) [41,91,92]. In other words, the structure of 27 results from the fusion of an octahedron with a tetrahedron through a triangular face. According to the condensation principle  Alternatively, 27 can also be considered as 7-sep hypoelectronic cluster [25] having an M2B5 oblatoarachno-structure which can be derived from a 9-vertex closo heptagonal bipyramid with two diamond-square-diamond (dsd) rearrangements and removal of two vertices (Figure 9). Interestingly, the core structure of 23 and its derivatives 24-26, and that of 27 can be seen as structural In 27, the W−W bond length (2.85 Å) bridged by a TePh group is slightly longer than in 4 (2.81 Å) [65]. Its structure can also be viewed in different ways. With a rather short W−W bond length, it can be first considered as a capped octahedron in which the W 2 B 4 octahedron (W1, W2, B2, and B3 as the base, and B1 and B4 as the apical vertices) is capped on the W1-W2-B4 face by a BH 3 unit (B5, Figure 9), similar to the core structure found for [(Cp*M) 3  Alternatively, 27 can also be considered as 7-sep hypoelectronic cluster [25] having an M 2 B 5 oblatoarachno-structure which can be derived from a 9-vertex closo heptagonal bipyramid with two diamond-square-diamond (dsd) rearrangements and removal of two vertices (Figure 9). Interestingly, the core structure of 23 and its derivatives 24-26, and that of 27 can be seen as structural isomeric forms (Figure 8).

Electronic Structure
The electronic structure of all the halogenated Ta-species 23-25 was extensively studied [79]. Pertinent results are reported in Table 5. The HOMO−LUMO gaps of 23-25 are around 2 eV which is slightly smaller than that for the parent cluster 6 which is 2.3 eV. The energy trend is in accord with the electronegativity of the halogens (Cl > Br >I). Table 5. Calculated bond length parameters (Å), Ta−Ta bond multiplicity, 11 B chemical shifts (ppm), HOMO and LUMO energy (eV), and HOMO-LUMO gaps (ΔE, eV) for 23-25.

Electronic Structure
The electronic structure of all the halogenated Ta-species 23-25 was extensively studied [79]. Pertinent results are reported in Table 5. The HOMO−LUMO gaps of 23-25 are around 2 eV which is slightly smaller than that for the parent cluster 6 which is 2.3 eV. The energy trend is in accord with the electronegativity of the halogens (Cl > Br >I). Some of the information about the nature of the bonding between the Ta atoms in clusters 23-25 was obtained by an analysis of the electron localization function (ELF) [32][33][34]. ELF is directly related to the electron pair probability density, and its graphical representation contributes to the understanding of electron localization/delocalization and consequently to the degree of metal−metal interaction. Two-dimensional electron density distribution plots are depicted in Figure 10   Some of the information about the nature of the bonding between the Ta atoms in clusters 23-25 was obtained by an analysis of the electron localization function (ELF) [32][33][34]. ELF is directly related to the electron pair probability density, and its graphical representation contributes to the understanding of electron localization/delocalization and consequently to the degree of metal−metal interaction. Two-dimensional electron density distribution plots are depicted in Figure 10  DFT calculations were also performed on 27', the Cp analog of 27 [81]. A HOMO-LUMO energy gap smaller than that of the parent cluster 4' was computed. The computed WBI values indicate strong W−W and W−Te bonding interactions. Finally, the topology analysis of the Laplacian, −∇ 2 ρ(r), of the total electron density, identifies 3c−2e bonds in the B1−B2−B3 and B2−B3−B4 triangles of the cluster cage (see Figure 9 for atom numbering). DFT calculations were also performed on 27', the Cp analog of 27 [81]. A HOMO-LUMO energy gap smaller than that of the parent cluster 4' was computed. The computed WBI values indicate strong W−W and W−Te bonding interactions. Finally, the topology analysis of the Laplacian, −∇ 2 ρ(r), of the total electron density, identifies 3c−2e bonds in the B1−B2−B3 and B2−B3−B4 triangles of the cluster cage (see Figure 9 for atom numbering).

Synthesis
The last M 2 B 5 structural type IV (Figure 1) is encountered only for two representatives so far.

Structure and Electron Count
The structures of 28 and 29, obtained from X-ray crystallography, show a highly open architecture with no metal−metal interaction in contrast to I-III (Figure 1). The arrangement of atoms in 28 and 29 resembles that of a substituted cyclopropane molecule with a central B 3 ring. This B 3 ring is similar to the [B 3 H 8 ] − structural matrix encountered in many borane compounds [96][97][98][99][100][101][102]. Pertinent bond length parameters are given in Table 6. 11 B NMR spectroscopy shows three boron chemical shifts with a 2:2:1 ratio. Similar boron atoms are marked in the same color in Figure 11 (right). 28 and 29 possess 9 seps ([2 × 2 (Cp 2 M) + 1 × 3 (B) + 2 × 2 (BH) + 2 × 1 (BH 2 ) + 5 × 1 (µ-H) = 18]/2) but their structure cannot be generated from a single structure. It can be generated however, from different ways such as the fusion of a six-vertex arachno M 2 B 4 fragment analogous to B 6 H 12 , with an MB 2 triangle, as shown in Figure 11.

Electronic Structure
The electronic structure of 28 was elucidated [82]. The optimized structure and the computed 11 B NMR chemical shifts are in a very good agreement with the experimental data. Interestingly, an NBO analysis indicated a close relationship of the central B 3 ring of 28 and the borane analogs NaB 3 H 8 and B 3 H 9 . The topological analysis of the Laplacian of the electron density also shows similar topological features for these three molecules, as shown in Figure 12.

Conclusions
Fehlner pioneered the development of metallaborane chemistry a few decades ago. Subsequently, we and others made a large number of metallaborane clusters and their derivatives with various structural arrangements. Among them, M2B5 clusters and derivatives constitute an important class of those compounds. The transition metals present in these M2B5 species span from group 4 to group 7, particularly with group 6 elements. Their structure varies from oblatonido-, oblatoarachno-, to arachno-type open structures. The most open cluster architecture is observed for a couple of examples with group 4 early-transition metals. Closer M2B5 geometries such as oblatonido and oblatoarachno structures with M−M interactions dominate with group 5 and 6 metals. For group 7 metals, only one oblatonido-M2B5 structure is known with Re. Many of these species appear to be hypoelectronic clusters and are often considered as 'rule breakers' with respect to the classical Wade-Mingos electron rules. This is due to their unique highly oblate (flattened) deltahedral structures featuring a cross-cluster M−M interaction. Many theoretical calculations were performed to elucidate their electronic structure and chemical bonding properties. More specifically, a lot of theoretical works were devoted to the understanding of the nature of the cross-cluster M−M interaction in these species.
In this review, we have focused on the fundamental aspects, i.e., the synthesis, structure, electronic properties, and chemical bonding of the transition metal M2B5 clusters known in the literature. Examples which have been discussed have illustrated how, in synergy with experiments, computational results can provide additional valuable information to provide further insights about the electronic properties and electronic requirements which govern their architecture and thermodynamic stability. The field of metallaborane chemistry still remains a fertile ground of investigation that continues to create interest not only for experimentalists, but also among theoriticians, concerning viable predictions, particularly the synthesis of new metallaborane clusters of diverse stoichiometry with late-transition metals or mixed early-and late-transition metals.

Conclusions
Fehlner pioneered the development of metallaborane chemistry a few decades ago. Subsequently, we and others made a large number of metallaborane clusters and their derivatives with various structural arrangements. Among them, M 2 B 5 clusters and derivatives constitute an important class of those compounds. The transition metals present in these M 2 B 5 species span from group 4 to group 7, particularly with group 6 elements. Their structure varies from oblatonido-, oblatoarachno-, to arachno-type open structures. The most open cluster architecture is observed for a couple of examples with group 4 early-transition metals. Closer M 2 B 5 geometries such as oblatonido and oblatoarachno structures with M−M interactions dominate with group 5 and 6 metals. For group 7 metals, only one oblatonido-M 2 B 5 structure is known with Re. Many of these species appear to be hypoelectronic clusters and are often considered as 'rule breakers' with respect to the classical Wade-Mingos electron rules. This is due to their unique highly oblate (flattened) deltahedral structures featuring a cross-cluster M−M interaction. Many theoretical calculations were performed to elucidate their electronic structure and chemical bonding properties. More specifically, a lot of theoretical works were devoted to the understanding of the nature of the cross-cluster M−M interaction in these species.
In this review, we have focused on the fundamental aspects, i.e., the synthesis, structure, electronic properties, and chemical bonding of the transition metal M 2 B 5 clusters known in the literature. Examples which have been discussed have illustrated how, in synergy with experiments, computational results can provide additional valuable information to provide further insights about the electronic properties and electronic requirements which govern their architecture and thermodynamic stability. The field of metallaborane chemistry still remains a fertile ground of investigation that continues to create interest not only for experimentalists, but also among theoriticians, concerning viable predictions, particularly the synthesis of new metallaborane clusters of diverse stoichiometry with late-transition metals or mixed early-and late-transition metals.