Synthesis, Structure and Bonding of the Tungstaboranes [Cp*W(CO) 2 B 3 H 8 ] and [(Cp*W) 3 (CO) 2 B 4 H 7 ]

: The structure and bonding of two novel tungstaboranes which were synthesized using diverse synthetic methods are described. (i) The room-temperature photolysis of [Cp*W(CO) 3 Me] with [BH 3 · SMe 2 ] led to the isolation of the hydrogen-rich tungstaborane [Cp*W(CO) 2 B 3 H 8 ] ( 1 ). Its geometry consists of an arachno butterﬂy core similar to tetraborane(10) and obeys the Wade-Mingos electron counting rules ( n vertices, n + 3 skeletal electron pairs (seps)). (ii) Further, the tungstaborane [(Cp*W) 3 ( µ -H) 2 ( µ 3 -H)( µ -CO) 2 B 4 H 4 ] ( 4 ) was isolated by thermolysis reaction of a tungsten intermediate, obtained by low temperature reaction of [Cp*WCl 4 ] and [LiBH 4 · THF] with [Cr(CO) 5 · THF]. Compound 4 which seems to have formed by replacement of a BH unit in [(Cp*W) 2 B 5 H 9 ] by the isoelectronic fragment {Cp*W(CO) 2 }, adopts an oblato - nido hexagonal-bipyramidal core ( n vertices, n –1 seps). Both compounds were characterized using multinuclear NMR, IR spectroscopy, mass spectrometry as well as single crystal X-ray diffraction analysis. In addition, density functional theory (DFT) calculations were performed in order to elucidate their bonding and electronic structures.

Metallaboranes were initially synthesized either through thermolysis or basic metathesis reactions involving pre-existing polyborane anions and transition-metal halides. Subsequently, it was shown that inclusion of monoboranes (such as [LiBH 4 ·THF], [BH 3 ·THF], or [BH 3 ·SMe 2 ]) in mono-cyclopentadienyl metal chlorides ([Cp*MCl] n ) represents a highly effective approach for producing metallaboranes that incorporate transition metals spanning from groups 5 to 9. Both of these methodologies have been employed to synthesize several group 6 metallaboranes. For example, Pett et al. isolated the first tungstaborane [(PMe 2 Ph) 3 H 2 WB 9 H 13 ] by the direct reaction of [WH 6 (PMe 2 Ph) 3 ] with the arachno-[B 9 H I4 ] − anion [29]. Similarly, Girolami and co-workers synthesized the homoleptic transition metal  2 , from the grinding of CrCl 3 with NaB 3 H 8 , which turned out to be an excellent precursor for chemical vapor deposition to metal diboride [30]. Fehlner 3 ·THF] which is a noteworthy example using the later approach [31]. Although metallaboranes have been synthesized from various transition metals across the periodic table, utilizing various synthetic strategies and conditions, synthesis of tungstaboranes is relatively less encountered. With regards to tungstaboranes, the clusters with smaller number of boron atoms are interesting both in terms of geometry as well as cluster build up chemistry. After it was discovered that the reaction of the hydrogen-rich [(Cp*WH 3 [36]. Therefore, in an attempt to synthesize some higher-vertex tungstaborane clusters, we have explored the reactivity of Cp*WL n (L n = (CO) 3 Me or Cl 4 ) in different conditions that led to the isolation of two different tungstaboranes with one and three W metal atoms, respectively. In this article, we report the structures and bonding of these two metallaboranes with the heaviest metal of group 6 triad. Furthermore, we conducted density functional theory (DFT) calculations to understand their bonding and electronic structures.

Results and Discussion
With the purpose of making higher nuclearity group 6 clusters, we performed the reaction of [Cp*M(CO) 3 Me] (M = W, Mo) with [BH 3 ·SMe 2 ] in photolytic conditions that led to the formation of air and moisture sensitive yellow solid products 1 and 2 in 27% and 31% yield, respectively (Scheme 1). Note that some low-yielded products were also formed in these reactions which could not be isolated. Compounds 1 and 2 were characterized using different spectroscopic techniques. Compound 2 was characterized as [Cp*Mo(CO) 2 B 3 H 8 ] [37] which was reported earlier from the reaction of [Cp*Mo(CO) 3 Me] with [BH 3 ·THF] in similar conditions. Detailed spectroscopic and structural characterization of compound 1 is given below.  [36]. Therefore, in an attempt to synthesize some higher-vertex tungstaborane clusters, we have explored the reactivity of Cp*WLn (Ln = (CO)3Me or Cl4) in different conditions that led to the isolation of two different tungstaboranes with one and three W metal atoms, respectively. In this article, we report the structures and bonding of these two metallaboranes with the heaviest metal of group 6 triad. Furthermore, we conducted density functional theory (DFT) calculations to understand their bonding and electronic structures.

Photolysis of [Cp*M(CO)3Me] (M = W, Mo) with [BH3·SMe2]
With the purpose of making higher nuclearity group 6 clusters, we performed the reaction of [Cp*M(CO)3Me] (M = W, Mo) with [BH3·SMe2] in photolytic conditions that led to the formation of air and moisture sensitive yellow solid products 1 and 2 in 27% and 31% yield, respectively (Scheme 1). Note that some low-yielded products were also formed in these reactions which could not be isolated. Compounds 1 and 2 were characterized using different spectroscopic techniques. Compound 2 was characterized as [Cp*Mo(CO)2B3H8] [37] which was reported earlier from the reaction of [Cp*Mo(CO)3Me] with [BH3·THF] in similar conditions. Detailed spectroscopic and structural characterization of compound 1 is given below. The molecular structure of 1 was determined using single-crystal X-ray diffraction analysis, which confirms the structural inferences made on the basis of spectroscopic data. Figure 1 depicts the molecular structure of 1, which exhibits a WB 3 butterfly core geometry that closely resembles that of [B 4 H 10 ] where one of the wing-tip is replaced by a {Cp*W(CO) 2 } fragment. The dihedral angle between W1−B1−B2 and B1−B2−B3 plane is 121.54 • and is comparable with that of related arachno-octahedral metallaboranes, shown in Table 1 [30,[37][38][39][40][41][42]. The average W−B and B−B bond lengths of 2.511(5) Å and 1.767 Å, respectively are in the expected region for such molecules [30,[37][38][39][40][41][42]. From the 11 B{ 1 H} NMR spectrum, the chemical shift at δ = 1.8 ppm corresponds to atom B3 and that at δ = −41.9 ppm corresponds to atoms B1 and B2. Further, a 1 H-11 B{ 1 H} HSQC (Heteronuclear Single Quantum Coherence) experiment was performed that confirmed a correlation between the bridging W−H−B hydrogen atoms and the boron atoms B1 and B2. of Cp* and CO ligands. The IR spectrum shows peaks that supported the existence of terminal carbonyl ligands and terminal B-H groups. However, a clear explanation eluded us until the single-crystal X-ray diffraction analysis of 1 was performed. The molecular structure of 1 was determined using single-crystal X-ray diffraction analysis, which confirms the structural inferences made on the basis of spectroscopic data. Figure 1 depicts the molecular structure of 1, which exhibits a WB3 butterfly core geometry that closely resembles that of [B4H10] where one of the wing-tip is replaced by a {Cp*W(CO)2} fragment. The dihedral angle between W1−B1−B2 and B1−B2−B3 plane is 121.54° and is comparable with that of related arachno-octahedral metallaboranes, shown in Table 1 [30,[37][38][39][40][41][42]. The average W−B and B−B bond lengths of 2.511(5) Å and 1.767 Å, respectively are in the expected region for such molecules [30,[37][38][39][40][41][42]. From the 11 B{ 1 H} NMR spectrum, the chemical shift at δ = 1.8 ppm corresponds to atom B3 and that at δ = −41.9 ppm corresponds to atoms B1 and B2. Further, a 1 H-11 B{ 1 H} HSQC (Heteronuclear Single Quantum Coherence) experiment was performed that confirmed a correlation between the bridging W−H−B hydrogen atoms and the boron atoms B1 and B2. Based on Wade's rules [15] and the molecular formula, it can be concluded that [Cp*W(CO)2B3H8], 1, adopts an arachno-octahedral butterfly geometry (n = 4 vertices) with seven (n + 3 = 7) skeletal electron pairs (seps) [{Cp*W(CO)2} 3 × 1 + (BH) 2 × 2 + (BH2) 3 × 1 + (µ-H) 1 × 4 = 14 electrons, i.e., seven seps]. It is structurally and electronically analogous to the molybdaborane congener (2) reported earlier by us [37] and its geometry matches with other arachno-type butterfly clusters reported earlier ( Table 1). Note that the Cp analogue of 1 was first reported by Gaines in 1978, generated from the photolysis reaction of [CpW(CO)3Cl] in presence of [(CH3)4NB3H8] [43]. However, its structure was established only based on preliminary spectroscopic studies. Based on Wade's rules [15] and the molecular formula, it can be concluded that [Cp*W(CO) 2 B 3 H 8 ], 1, adopts an arachno-octahedral butterfly geometry (n = 4 vertices) with seven (n + 3 = 7) skeletal electron pairs (seps) [{Cp*W(CO) 2 } 3 × 1 + (BH) 2 × 2 + (BH 2 ) 3 × 1 + (µ-H) 1 × 4 = 14 electrons, i.e., seven seps]. It is structurally and electronically analogous to the molybdaborane congener (2) reported earlier by us [37] and its geometry matches with other arachno-type butterfly clusters reported earlier ( Table 1). Note that the Cp analogue of 1 was first reported by Gaines in 1978, generated from the photolysis reaction of [CpW(CO) 3 Cl] in presence of [(CH 3 ) 4 NB 3 H 8 ] [43]. However, its structure was established only based on preliminary spectroscopic studies.
To gain further insight into the bonding and electronic properties, computational density functional theory (DFT) studies were undertaken on model 1 (Cp analogue used to mimic 1) at the B3LYP/def2-TZVP level of theory (see Section 3.3 for computational details). The optimized core structure of 1 is in a good agreement with the solid-state X-ray structure determination (see Table S1). The calculated 11 B NMR chemical shifts are also in well accordance with the values experimentally measured for cluster 1 (see Table S2). Subsequently, DFT studies were also carried out for the Mo and Cr analogues 2 (Cp analogue of 2) and 3 (Cp analogue of Cp*Cr(CO) 2 B 3 H 8 ), respectively, for a comparative bonding analysis. A molecular orbital (MO) analysis suggests that both the HOMO and LUMO predominantly show a bonding character between tungsten and carbonyl ligands. As expected, a similar kind of interaction occurs for 2 and 3 . Interestingly, the energy of the HOMOs decreases as we move from W to Mo to Cr, whereas, that of the LUMOs get more stabilized in the case of 2 than that of 3 when compared with 1 . The result is that the HOMO-LUMO gap of these group 6 metal-B 3 H 8 clusters decrease while moving down the triad (Figure 2a). In this connection, a natural bond order (NBO) analysis suggested a decreasing trend of negative natural charge on the metal center, while moving from 3 to 1 . Furthermore, an important σ-bonding interaction was observed in HOMO−8 that involves d-orbitals of the metal and p-orbitals of the boron atoms ( Figure 2b). Likewise, as shown in Figure 2c, π-type overlap of boron p orbitals generate B−B π-bonding interaction delocalized throughout the triangular B 3 framework. All the above findings seemingly justify the stability of the butterfly core in these species. Table 1.
Selected structural parameters and NMR chemical shifts of selected metaloctahydrotetraborane complexes.

Metal-B 3 H 8 Complexes
Spectroscopic Parameters (ppm) Structural Parameters Ref. 11          creasing trend of negative natural charge on the metal center, while moving from 3′ to 1′. Furthermore, an important σ-bonding interaction was observed in HOMO−8 that involves d-orbitals of the metal and p-orbitals of the boron atoms ( Figure 2b). Likewise, as shown in Figure 2c, π-type overlap of boron p orbitals generate B−B π-bonding interaction delocalized throughout the triangular B3 framework. All the above findings seemingly justify the stability of the butterfly core in these species.

Reactivity of [Cp*WCl4] with [Cr(CO)5·THF]
The reactivity of [Cp*WCl4] with various boranes has been explored over several decades and has led to the formation of different tungstaborane clusters [33,34,44,45]. Thus, with an objective to get higher-vertex hybrid-clusters, we further explored the reactivity of [Cp*WCl4] with an excess of

Reactivity of [Cp*WCl 4 ] with [Cr(CO) 5 ·THF]
The reactivity of [Cp*WCl 4 ] with various boranes has been explored over several decades and has led to the formation of different tungstaborane clusters [33,34,44,45]. Thus, with an objective to get higher-vertex hybrid-clusters, we further explored the reactivity of [Cp*WCl 4 ] with an excess of [LiBH 4 ·THF] followed by treating the solution with [Cr(CO) 5  Thin-layer chromatography (TLC) was used to separate and purify the products. Deta spectroscopic and structural characterisation of cluster 4 is described below. In addition, the 1 H NMR spectrum of 4 shows three distinct peaks for Cp* protons at δ = 1.95, 2.01 and 2.11 ppm in a 1:1:1 ratio. Further, 13 C NMR also confirmed the presence of three types of Cp*. In addition, there are three peaks in the upfield region of 1 H NMR at δ = −9.09, −9.36 and −12.86 ppm, indicating three types of bridging hydrogen atoms, for example, W−H−W and W−H−B protons. IR spectroscopy indicated the presence of bridging carbonyl ligands and terminal B−H groups. Although mass spectrometric data along with all the other spectroscopic data provided some guidelines about the structure of 4, the core geometry was truly established when the single crystal X-ray diffraction analysis was carried out (Figure 3). For that, a suitable crystal of 4, grown by the slow evaporation of hexane/dichloromethane solution at −5 • C was used.

Scheme 2. Synthesis of cluster 4.
The mass spectrum of 4 showed a molecular ion peak at m/z = 1063 which co sponds to [C32H52B4O2W3] + . The 11 B{ 1 H} NMR spectrum shows peaks of equal intensi δ = 25.7, 36.2, 69.1 and 73.8 ppm which confirm the presence of four boron atoms. In a tion, the 1 H NMR spectrum of 4 shows three distinct peaks for Cp* protons at δ = 1.95, and 2.11 ppm in a 1:1:1 ratio. Further, 13 C NMR also confirmed the presence of three t of Cp*. In addition, there are three peaks in the upfield region of 1 H NMR at δ = − −9.36 and −12.86 ppm, indicating three types of bridging hydrogen atoms, for exam W−H−W and W−H−B protons. IR spectroscopy indicated the presence of bridging bonyl ligands and terminal B−H groups. Although mass spectrometric data along wit the other spectroscopic data provided some guidelines about the structure of 4, the geometry was truly established when the single crystal X-ray diffraction analysis was ried out (Figure 3). For that, a suitable crystal of 4, grown by the slow evaporation of ane/dichloromethane solution at −5 °C was used.
In order to investigate the electronic properties of 4, DFT calculations of the Cp analogue of 4 (4 ) was carried out at the B3LYP/def2-TZVP level of theory. An MO analysis indicates that both the HOMO and LUMO are mostly non-bonding metal d-type orbitals with almost insignificant bonding interactions between the metal and boron atoms. Indeed, as shown in Figure 4a, the p orbitals of all four boron centers overlap to generate low energy-lying σ-bonding interactions in a delocalized fashion resulting in a region of charge concentration along the B1-B2, B2-B3 and B3−B4 bonds. This also shows high covalent character, which is also reflected by the average electron density [ρ r ] and energy density [H(r)] values (Table S3). Although any bonding interaction along the cross-cluster W2−W3 bond in the MO analysis is not straightforward, an NBO analysis showed indeed some bonding interaction. The NBO analysis suggested also the existence of bonding interactions along the W1−W2 and W1−W3 bonds. Furthermore, a substantial Wiberg bond indices value of 0.738 suggests the presence of a strong bonding interaction along the W2−W3 bond as compared to the W1−W3 and W1−W2 bonds (0.545 and 0.391 respectively). Therefore, based on these theoretical results, it is reasonable to assume that the cross-cluster W-W bond is even stronger as compared to the other W−W bonds (Table S1).
In order to investigate the electronic properties of 4, DFT calculations of the Cp analogue of 4 (4′) was carried out at the B3LYP/def2-TZVP level of theory. An MO analysis indicates that both the HOMO and LUMO are mostly non-bonding metal d-type orbitals with almost insignificant bonding interactions between the metal and boron atoms. Indeed, as shown in Figure 4a, the p orbitals of all four boron centers overlap to generate low energy-lying σ-bonding interactions in a delocalized fashion resulting in a region of charge concentration along the B1-B2, B2-B3 and B3−B4 bonds. This also shows high covalent character, which is also reflected by the average electron density [ρr] and energy density [H(r)] values (Table S3). Although any bonding interaction along the cross-cluster W2−W3 bond in the MO analysis is not straightforward, an NBO analysis showed indeed some bonding interaction. The NBO analysis suggested also the existence of bonding interactions along the W1−W2 and W1−W3 bonds. Furthermore, a substantial Wiberg bond indices value of 0.738 suggests the presence of a strong bonding interaction along the W2−W3 bond as compared to the W1−W3 and W1−W2 bonds (0.545 and 0.391 respectively). Therefore, based on these theoretical results, it is reasonable to assume that the cross-cluster W-W bond is even stronger as compared to the other W−W bonds (Table S1).

General Methods and Instrumentation
All experiments were carried out utilizing standard Schlenk and glove-box techniques within an argon atmosphere. Prior to usage, solvents were distilled under an argon environment. Other chemicals such as Cp*H, [BH 3 .SMe 2 ], [Cr(CO) 6 ], and MeI (Aldrich) were used as received. [Cp*W(CO) 3 Me] [49] and [Cp*WCl 4 ] [50] were prepared according to a literature procedure. Thin-layer chromatography was carried on 250-µm diameter aluminum supported silica gel TLC plates (MERCK TLC Plates). The NMR spectra were recorded on a 400 and 500 MHz Bruker FT-NMR spectrometer. The Infraref spectra (IR) were recorded using a JASCO FT/IR -4100 spectrometer. The mass spectra were recorded on an Agilent Technologies 6545-Q-TOF LC/MS spectrometer. The experimental photoreactions discussed in this study were performed using a Luzchem LZC-4 V photoreactor under irradiation conditions ranging from 254 to 350 nm. Synthesis of 1: A solution of [Cp*W(CO) 3 Me] (0.05 g, 0.15 mmol) and [BH 3 ·SMe 2 ] (0.7 mL, 7 mmol) in THF (12 mL) was prepared in a flame-dried Schlenk tube. The reaction mixture was then exposed to irradiation at room temperature for a duration of 2 h. Following the reaction, the volatile components were removed under vacuum, and the remaining residue was extracted into hexane and passed through Celite. The solvent was subsequently evaporated, and the resulting residue underwent chromatographic purification using silica gel TLC plates. Elution with a mixture of hexane and CH 2 Cl 2 (95:5 v/v) led to the isolation of arachno-1 as a yellow solid (0.016 g, 27% yield).

Single Crystal X-ray Diffraction Analysis
Slow evaporation of a hexane−CH 2 Cl 2 solution of compound 1 and 4 allowed us to grow crystals which are suitable for XRD analysis. Table 2 provides crystallographic and structural refinement data for 1 and 4. A Bruker D8 VENTURE diffractometer with PHOTON II detector at 296(2) K was used to collect and integrate crystal diffraction data of 1. Further, a Bruker APEX-II CCD diffractometer was used to collect as well as integrate the crystal data of 4. A dual-space algorithm using SHELXT program [51] was used to solve the structures and refinement was achieved utilizing full-matrix least-squares methods based on F 2 using SHELXL-97 software [52]. Olex2-1.3 [53] was used to draw molecular structures. The refinement of non-hydrogen atoms was aided by anisotropic displacement parameters (ADPs), while the introduction of hydrogen atoms, except for bridging hydrogens, was achieved through analysis of Fourier difference maps. H atoms were finally included in their calculated positions and treated as 'riding' on their parent atom with constrained thermal parameters. Crystallographic data can be acquired for free from the Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif, accessed on 31 May 2023.

Computational Details
Geometry optimizations of all model clusters were carried out using the Gaussian 09 software package [54]. In order to reduce the computational efforts, the computations were carried out with the Cp ligands instead of Cp* (for optimized model clusters, see Figures S14-S17). The Becke-Lee-Yang-Parr (B3LYP) hybrid functional [55], along with the def2-TZVP basis set [56] from the EMSL (Environmental Molecular Sciences Laboratory) Basis Set Exchange Library [57] for all atoms. Optimizations were carried out without any solvent effects or symmetry constraints. To ensure that the optimized geometries represented genuine minima on the potential energy hypersurface, frequency calculations were conducted for all structures. These calculations were used to examine the nature of the stationary states and verify that there were no imaginary frequencies present. Using the optimized geometries at the B3LYP/def2-TZVP level, NMR chemical shifts were calculated using the gauge-including atomic orbitals (GIAOs) method [58][59][60]. The 11 B NMR chemical shifts were calculated relative to B 2 H 6 , with a B3LYP shielding constant of 83.6 ppm. To convert these values to the standard [BF 3 ·OEt 2 ] scale, the computed values were adjusted by adding 16.6 ppm (the experimental δ ( 11 B) value of B 2 H 6 ). The internal standard used for calculating 1H NMR chemical shifts was TMS (SiMe 4 ). Orbital graphics and optimized structure plots were generated using Chemcraft [61]. To obtain the Wiberg bond indices (WBI) [62], a natural bond orbital (NBO) analysis [62] was conducted.

Conclusions
The present work describes the isolation and structural characterization of a monoand a tri-metallic tungstaborane clusters generated from two different metal complex precursors, namely [Cp*W(CO) 3 3 ] core, cluster 1 obeys the Wade-Mingos electron counting rules (n vertices, n + 3 seps), whereas species 4 is an oblato-nido hexagonalbipyramidal cluster (n vertices, n − 1 seps).