A dianionic allene bound to vanadium(iii) enables interconversion between vanadatetrahedranes, vanadacyclobutadienes and deprotiovanadacyclobutadienes

Abstract

Herein we disclose a novel route to redox-active vanadacyclobutadiene (VCBD) and vanadatetrahedrane (VT_d) complexes, which circumvents a vanadium alkylidyne precursor ([V^V] CR). [V^V] VCBD salts and VT_d's are prepared through an unusual, non-classical oxidative addition reaction via the addition of electrophiles to a low-spin [V^III] deprotiovanadacyclobutadiene (dVCBD). The microscopic reverse reaction, a reductive elimination, reverts the [V^V] VCBD and VT_d scaffolds back to the low-spin [V^III] dVCBD species via the addition of a Brønsted base. The interconversion of the [V^V] VCBD salts and VT_d's is mediated through anion exchange, which promotes a change in the spin-state and geometry of the organometallic species, highlighting the redox active nature of the allene ligand (C₃) bound to vanadium (V). Single electron reduction of the [V^V] VCBD or VT_d with cobaltocene (CoCp₂) led to the formation of neutral and radical-based [V^IV] VCBDs. Oxidation of the [V^IV] VCBDs with ferrocenium based salts ([FeCp₂][X], X = BArF₂₀⁻, BArF₂₄⁻, OTf⁻) or chloride-delivering oxidants (i.e. triphenylmethyl chloride (Ph₃CCl) or lead(ii) chloride (PbCl₂)) regenerates the discrete [V^V] VCBD salts or VT_d's, respectively. Cyclic voltammetry studies reveal quasi-reversible one electron redox couples, while X-band electron paramagnetic resonance (EPR) spectroscopic studies confirms the presence of a paramagnetic [V^IV], d¹ system in the VCBD. Computational analysis of the dVCBD, VT_d, and VCBD complexes affords detailed insight into the structure and bonding of this unusual class of molecules and delineates the role of the anion in their interconversion. Further examination of the reaction of dVCBD with electrophiles corroborates the nonclassical character of the oxidative addition sequence involving metal–ligand cooperation, with ligand-centered redox activity, rather than a formal two-electron oxidation at the metal center. The present work demonstrates how these rare dVCBD, VCBD, and VT_d scaffolds can all be interconverted via reversible C–C bond formation and splitting pathways using the right combination of anion exchange, redox, and Brønsted acid–base chemistry.

Without a vanadium alkylidyne, and using a dianionic allene bound to vanadium, we demonstrate how vanadatetrahedranes and vanadacyclobutadienes can be accessed and how these species can interconvert.

Introduction

Electrophile addition procedure to construct MCBDs and MT_d's

Traditionally, metallacyclobutadienes (MCBDs) and metallatetrahedranes (MT_d's) are synthesized via two approaches; electrophilic addition (alternatively called oxidative addition) reactions (Scheme 1A and B) or [2 + 2]-cycloaddition reactions (Scheme 1C–E). Chronologically, the electrophilic addition pathway was the original pathway to uncover MT_d and MCBD scaffolds. The discovery of the MT_d can be traced back to the studies of Gowling and Kettle, wherein the reaction of a cyclopropenium salt, triphenylcyclopropenium bromide [C₃Ph₃][Br], with nickel tetracarbonyl (Ni(CO)₄), gave way to the proposed dimer complex, [(Br)(CO)Ni(η³-C₃-PhCC(Ph)CPh)]₂.¹ Tuggle and Weaver were able to structurally characterize a monomeric form of a Ni-based MT_d from the addition of [C₃Ph₃][Cl] in pyridine (C₅H₅N, py) to Ni(CO)₄ to afford, (Cl)(py)₂Ni(η³-C₃-PhCC(Ph)CPh).^2–4 The first MCBD complex was isolated via the addition of triphenylcyclopropenium tetrafluoroborate [C₃Ph₃][BF₄] to Vaska's [Ir^I] complex, trans-Ir(Cl)(CO)(PMe₃)₂, to yield the iridacycle [Ir^III] complex, [trans-(Cl)(CO)(PMe₃)₂Ir(κ²-C,C-PhCC(Ph)CPh)][BF₄].^5,6 Importantly, the identity of the counter anion for the oxidative addition reaction plays a significant role in the geometry of the final product. In short, smaller anions bind preferentially to the metal center enforcing a MT_d geometry, while larger anions do not bind to the metal center leading to the formation of the MCBD geometry. Following in their footsteps, numerous researchers have identified an array of MT_d^2–4,7–34 and MCBD^5,6,35–39 complexes through this electrophilic addition pathway.

Scheme 1. Previous synthetic routes to access MCBD and MT_d scaffolds.

Moreover, electrophilic addition reactions conducted on functionalized metallacyclobutenes have led to the formation of MCBD complexes (Scheme 1B). This procedure was first carried out by Wojcicki and coworkers, whereupon the addition of triethyloxonium hexafluorophosphate ([Et₃O][PF₆]), to the anionic rhenium metallacyclobutenone, [Na][(CO)₄Re(κ²-C,C-MeCC(CO₂Me)C O)] led to the formation of the neutral MCBD, [(CO)₄Re(κ²-C,C-MeCC(CO₂Me)COEt)].⁴⁰ Further reports on electrophile addition reactions on metallacyclobutenones^41–48 and other related substrates^34,49–53 has led to the formation of MCBD complexes with heteroatoms substituents on the MCBD ring.

Metathesis procedure to construct MCBDs and MT_d's

In contrast to electrophilic addition pathways, the isolation of MCBD and MT_d complexes via an alkyne metathesis procedure was first uncovered by Schrock and coworkers in 1982 (ref. 54) and 1984,⁵⁵ respectively (Scheme 1C). The first report of a MCBD involved the addition of 2-butyne (MeC CMe) to the tungsten alkylidyne complex, (dme)(Cl)₃W C^tBu (where dme = 1,2-dimethoxyethane), resulting in the formation of the tungstacyclobutadiene, (dme)(Cl)₃W(κ²-C,C-^tBuCC(Me)CMe).⁵⁴ To promote rearrangement of the planar ring in the MCBD to the MT_d geometry, the addition of donor ligands such as trimethylphosphine (PMe₃) or tetramethylethylenediamine ((CH₃)₂NCH₂CH₂N(CH₃)₂, tmeda) to the MCBD complexes, (Cp)(Cl)₂W(κ²-C,C-^tBuCC(Me)C^tBu) (where Cp = η⁵-cyclopentadienyl) or the latter (Cl)₃W(κ²-C,C-^tBuCC(Me)CMe) resulted in (PMe₃)(Cp)(Cl)₂W(η³-C₃-^tBuCC(Me)C^tBu)⁵⁶ or (tmeda)(Cl)₃W(η³-C₃-^tBuCC(Me)CMe),⁵⁵ respectively. Following these preliminary reports other research groups have identified MCBD^57–101 and MT_d^{55,56,58,66,102–106} by taking advantage of the [2 + 2]-cycloaddition of an alkyne across a metal alkylidyne (M CR).

Moreover, MCBDs can also be constructed from the reductive splitting of internal alkynes (RC CR) with metal–metal triple bonded complexes (L_nM ML_n).^70,107–111 An isolated MCBD from this pathway (Schemes 1D) was first identified by Schrock and coworkers,¹¹¹ from the reaction of W₂(O-2,6-Me₂C₆H₃)₆ with three equivalents of MeC CMe or 3-hexyne (EtC CEt) to form the tungstacyclobutadiene complexes (O-2,6-Me₂C₆H₃)₃W(κ²-C,C-RCC(R)CR) (where R = Me or Et). In the same vein, MCBD can be prepared via the nitrile–alkyne cross metathesis (NACM) reaction (Scheme 1E). This procedure was identified by Johnson and coworkers who observed the formation of the MCBD (OCH(CF₃)₂)₃W(κ²-C,C-EtCC(Et)CEt) from the reaction of the tungsten nitride (OCH(CF₃)₂)₃W N with EtC CEt, resulting in the loss of propionitrile (EtC N).¹¹² The NACM reaction has been observed for other metal nitrides (L_nM N) on W¹¹³ and Mo^114–119 complexes, and Fe¹²⁰ complexes in the gas phase. Recently, Fürstner and coworkers have proposed another route to form MCBDs from M CR, through an unprecedented stepwise process rather than the traditional [2 + 2]-cycloaddition pathway.¹²¹

Aborted alkyne metathesis leads to formation of dMCBDs

In contrast, alkyne metathesis involving terminal alkynes (HC CR) with M CR often results in a different outcome. The most kinetically feasible [2 + 2]-cycloaddition results in a MCBD with the proton at the beta-carbon (C_β) position of the MCBD. However, the latter species can undergo an intramolecular deprotonation event to generate a deprotiometallacyclobutadiene (dMCBD) concurrently with demetallation of a ligand (HL, Scheme 1C), which was generally referred to as an “aborted” alkyne metathesis reaction. Such a highly unsaturated dMCBD scaffold was first introduced by Schrock and co-workers over 40 years ago,¹²² upon the reaction of the tungsten alkylidyne complex, (Cp)(Cl)₂W C^tBu, with tert-butyl acetylene (^tBuC CH), to generate the dMCBD, (Cp)(Cl)W(η^3-tBuCCC^tBu). The authors noted that the dMCBD species was not an active catalyst for alkyne metathesis, but instead furnished polyalkynes when using excess alkyne.^63,123,124 Since the initial discovery of the dMCBD from an aborted alkyne metathesis reaction, few reports have focused their attention on the dMCBD scaffold itself. In one exception, Mortreux and coworkers speculated that the dMCBD scaffold was critical in the formation of polyalkynes since this scaffold could rearrange to a carbene ligand.¹²³ Until quite recently, dMCBDs were only reserved for Mo^{63,116,125,126} and W^122,127 complexes since their preparation required a high-valent M CR as a precursor.

[V^V] CR are an extremely rare organometallic species which have only been isolated through two synthetic pathways.^{101,128–130} In a previous study we isolated to our knowledge the first vanadacyclobutadiene (VCBD) and deprotiovanadacyclobutadiene (dVCBD) complexes (Scheme 2A) from the reaction of a [V^V] C^tBu scaffold with terminal alkynes (HC CR).¹⁰⁰ Notably, our work established that dVCBD scaffolds were key intermediates in the formation of cyclic-polyphenylacetylene (c-PPA), indeed confirming the original proposal by Mortreux and coworkers¹²³ and recently highlighted by Cui and Jia.¹³¹ Identifying that the VCBD, and more specifically, the dVCBD were critical in accessing c-PPA, we investigated an alternative route to these scaffolds that would circumvent altogether the need for the challenging preparative [V^V] C^tBu fragment. The synthesis of this species typically involves a [V^III] bis-neopentyl precursor,^128,132 followed by multiple steps including oxidatively-induced α-hydrogen abstractions, anion exchange, alkylations, and a ligand deprotonation (overall yield ∼10%, 7 steps). For this reason, we turned our attention to a more direct route to VCBD scaffolds by alkylation or protonation of dVCBD that could be construed directly by salt elimination using dilithio salts of 1,3-disubstituted allenes.

Scheme 2. Top: Previous work involving dMCBD and MCBD scaffolds. Bottom: This work involving the synthesis of VCBD and VT_d from a dVCBD, and their respective interconversions.

Salt metathesis procedure to access dMCBDs

We were inspired by the seminal work by Beweries, Reiß, and coworkers (Scheme 2B) who reported the assembly of unsaturated four-membered rings on [Ti^IV]^133,134 and [Zr^IV]^134–136 complexes via the salt metathesis of the corresponding metallocene dihalide precursors (L_nMCl₂; where L_n = ansa-cyclopentadienyl), with a dilithioallene reagent (Li₂C₃R₂) (where R = Ph or SiMe₃).^134,137 These unsaturated four-membered rings have been referred to as 1-metallacyclobuta-(2,3)-dienes (MCB-2,3-D) in that they exhibit distinct structural and electronic features when compared to the more traditional dMCBDs of [W^VI] and [Mo^VI]. Namely, these MCB-2,3-D exhibit electron localization on the allene scaffold (C₃) on the metallacycle ring, whereas the dMCBD exhibits a more delocalized resonance, in where the scaffold contains a higher formal negative charge. These features are also reflected in the reactivity of these species, showing mostly classical metallacycle insertion reactivity with electrophiles having carbonyl and carbodiimide motifs.^133,135,136 Irrespective of these electronic differences, the MCB-2,3-D scaffold could be a segue to a dMCBD by virtue of resonance and by taking advantage of a more electron rich metal center donating into the MCB-2,3-D framework.

Due to the streamlined and modular approach through the salt metathesis of a L_nMX₂ and the Li₂C₃R₂, we developed a simpler route to metallacyclic complexes of [Ti^III] and [V^III] metal centers supported by the BDI⁻ ligand scaffold (BDI⁻ = [ArNC(CH₃)]₂CH, Ar = 2,6-ⁱPr₂C₆H₃), through a similar salt elimination procedure, thus avoiding the need to prepare a synthetically cumbersome [V^V] CR (Scheme 2C).¹³⁸ Moreover, these metallacycles were also found to be active catalysts for the production of c-PPA, akin to how dMCBD were proposed to operate.^100,123 In comparison to the prepared MCB-2,3-D reported by Beweries and coworkers,^133–136 the metallacycles supported by the BDI⁻ ligand scaffold using more electron rich metal ions [Ti^III] and [V^III] more closely resemble dMCBD rather than MCB-2,3-D when judging from single crystal X-ray diffraction (scXRD) studies, specifically in the case of the [V^III] complex.¹³⁸ However, reactivity studies of the [V^III] analogue highlights another dominant resonance form, the MCB-2,3-D, as this species can further undergo a retro [2 + 2]-cycloaddition to access a transient [V^III] alkylidene alkynyl complex, {(BDI)V C(SiMe₃)C C(SiMe₃)} (Scheme 2D). The rearrangement to this transient species was trapped via oxidation with nitrous oxide (N₂O) to yield the isolable [V^V] alkylidene oxo complex,¹³⁸ (BDI)V(O) C(SiMe₃)C C(SiMe₃) (Scheme 2D). Alternatively, rearrangement to a [V^III] alkylidene alkynyl complex was also proposed based on an intramolecular [2 + 2] cross metathesis between the alkylidene ligand and the imine motif of the BDI⁻ scaffold. This transformation was promoted by heat or via the addition of a Lewis base (4-dimethylaminopyridine, ((CH₃)₂NC₅H₄N, DMAP) to form the vanadium imido monoazabutadieneallenyl complex, (ArNC(CH₃)CHC(CH₃)C(SiMe₃)CC(SiMe₃))V NAr that was inactive for alkyne polymerization.¹³⁹

Due to the expedited synthesis of these 3d dMCBD scaffolds we were interested in probing the reactivity of this electron rich scaffold by performing alkylation and protonation reactions. Since dMCBD are generally formed from the deprotonation of the β-C of the corresponding MCBD framework, one could anticipate that the microscopic reverse, protonation of dMCBD to the MCBD framework, could also be feasible. Herein, we report the reactivity of a dVCBD, (BDI)V(η³-Me₃SiCCCSiMe₃) (1),¹³⁸ with electrophiles (Scheme 2E). We demonstrate reversible C–C bond formation by means of assembling VCBD salts and VT_d's, without the need of a [V^V] CR motif, and report their respective redox activity as well as Brønsted acid–base chemistry. From a readily accessible dVCBD scaffold we show interconversion between VCBD and VT_d enabled by Brønsted acid–base, anion exchange or redox chemistries. Computational studies of the dVCBD, VT_d, and VCBD complexes provide insight into the structure and bonding of this unusual class of organometallic compounds and elucidate the role of anions in their interconversion. Furthermore, analysis of the reactions of the dVCBD complexes with electrophiles confirms the unusual nonclassical nature of this oxidative addition sequence.

Results and discussion

Revisiting the electronic structure of complex 1

Before exploring the reactivity of 1 with electrophiles we re-examined the nuclear magnetic resonance (NMR) spectroscopic features and its original assignment of such species as a high-spin (HS) d² complex. Specifically, we were perplexed why the previously isolated and related dVCBD, (BDI)V(η^3-tBuCCCMes)¹⁰⁰ (dVCBD₁) was diamagnetic and exhibited sharp resonances in its NMR spectra, which allowed for solution phase assignment of nearly all its components using a combination of multinuclear and multidimensional NMR spectroscopic experiments. Complex 1 on the other hand exhibited much broader spectral features in the ¹H NMR spectrum at 300 K (Fig. S11). To probe whether this broadened phenomenon was the consequence of fluctuation in solution as opposed to paramagnetism, a variable temperature VT-NMR spectroscopic experiment was carried out using a sample of 1 in deuterated toluene (C₇D₈) from 297 K to 243 K (Fig. S4). Since 1 is amenable to an intramolecular rearrangement involving its C₃ scaffold and the imine residue of the BDI⁻ supporting ligand at elevated temperatures, we avoided temperatures above 297 K.¹³⁹ Accordingly, lowering the temperature to 273 K resulted in the coalescence of resonances corresponding to the 2,6-(diisopropyl)phenyl substituents of the BDI⁻ scaffold (broad resonance at δ 1.03, Fig. S4) and the trimethylsilyl (SiMe₃) groups of the metallacycle (broad resonance at δ 0.32, Fig. S4). Using a modified Eyring equation outlined by Gunther,¹⁴⁰ the ΔG^‡ for the transition was estimated to be 51.1 kJ mol⁻¹, which is essentially barrierless at room temperature. Upon lowering the temperature of the spectrometer to 243 K, a sharpening of resonances was observed (Fig. S4), suggesting the room-temperature broadening of resonances in 1 to be dynamic in nature. This fluxional temperature-dependent behavior may be due to the reduction in the steric profile for 1 when compared to dVCBD₁.¹⁰⁰ The rigid and more sterically encumbering mesityl group in the latter most likely inhibits interconversion of an exo- and endo-isomerization of the complex in solution when compared to the more spherical and symmetric SiMe₃ groups in 1, as these would be able to freely rotate (Scheme S3). At 243 K nearly all the components of complex 1 could be identified (Fig. S2 and S3) using a combination of NMR spectroscopic experiments (Fig. S1 and S3–S11). For instance, the broad resonance feature corresponding to the equivalent SiMe₃ groups at δ 0.32 can be resolved into two sharper features at δ 0.66 and 0.09 at 243 K. Additionally, the broad resonance at δ 1.03 can also be resolved at 243 K, clearly separating into resonances at δ 1.58, 1.26, 0.76, and 0.34, which correspond to the diastereotopic CH₃ on the ⁱPr residues of the BDI⁻ framework (Fig. S3; H_e, H_g, H_h, and H_j). Moreover, two new resonances resolved at 243 K at δ 4.13 and 0.19, which correspond to the methine CH protons on the ⁱPr residues (Fig. S3; H_d and H_k). The large difference in the chemical shift of these two resonances is due to the inhibited rotation of the diisopropylphenyl (Dipp) substituent on the BDI⁻ ligand at reduced temperatures. Thus, the methine CH protons then occupy a position above and below the plane of the V metal center (Scheme S4), which have different chemical environments and in turn lead to disparate chemical shifts, which is also apparent in the chemical shifts for the SiMe₃ groups on the dVCBD scaffold. This trend in the chemical shifts of the methine CH protons on the Dipp scaffold is observed for the other VCBD salts reported herein (vide infra). Moreover, the ⁵¹V NMR spectrum of complex 1 at 243 K shows a single broad downfield resonance at δ 1185 (Δν_1/2 = 2051 Hz) (Fig. S10), comparable to dVCBD₁ which exhibits a resonance at δ 1353 (Δν_1/2 = 2320 Hz). The ²⁹Si NMR experiment conducted at 243 K revealed no silicon (Si) resonances, likely due to the low abundance of ²⁹Si (natural abundance = 4.7%) and the Si proximity to the quadrupolar ⁵¹V nucleus, which we have found to broaden heteronuclear resonances in close proximity to that nucleus.^139,141 Thus, a ¹H–²⁹Si NMR experiment (Fig. S6) was conducted, which showed a correlation between the H resonances at δ 0.66 and 0.09 to Si resonances observed at δ −16.9, −16.7. Additional 2D NMR experiments at 243 K (¹H–¹H COSY; Fig. S7, ¹H–¹³C HSQC; Fig. S8, ¹H–¹³C HMBC; Fig. S9) have also allowed us to confirm complex 1 to be a diamagnetic (low-spin (LS), [V^III], d²) complex exhibiting fluxional behavior at 300 K. Our proposed configuration is further supported by experimental (Fig. S13) and computed (Fig. S131) electronic absorption spectra. In addition to these experimental considerations, we have conducted a detailed computational analysis of the structure and bonding of this complex. The results of this study are discussed as part of the theoretical section of this manuscript (vide infra).

Synthesis and characterization of a VT_d scaffold

Stemming from the molecular orbital picture for the transient dVCBD, {(BDI)V(η^3-tBuCCCPh)},¹⁰⁰ the highest occupied molecular orbital (HOMO) was found to exhibit electron density on the C_β with the orbital lobe oriented away from V and outwards from the dVCBD scaffold. Based on this premise, we surmised that the C_β in such dVCBD scaffolds could possess nucleophilic character and thus decided to probe this in reactions with electrophiles. Earlier studies by us using the closely related heterocumulene (dBDI)V(η³-AdNCC^tBu) (where dBDI²⁻ = ArNC(CH₃)C(H)C(CH₂)NAr, Ar = 2,6-ⁱPr₂C₆H₃) have found that Brønsted acid–base chemistry results in chelotropic and spin-state rearrangements in such scaffold if the anion coordinates to V (e.g., OTf⁻).¹⁴² As a result, we explored a suitable proton source that would contain a coordinating anion to maximize its interaction with V.

Accordingly, when complex 1 was dissolved in toluene (C₇H₈) and solid pyridinium chloride ([PyH][Cl]) was added to the stirring solution at 293 K, a color change was observed from a red-orange to a dark violet color within minutes after the addition of the Brønsted acid. After 30 minutes the solution was dried under reduced pressure, extracted into toluene, filtered, concentrated, and cooled to 238 K overnight, which allowed dark violet crystals to precipitate from the concentrated solution in a 66% yield. This material was determined to be the η³-C₃-vanadatetrahedrane (BDI)V(η³-C₃-Me₃SiCC(H)CSiMe₃)(Cl) (1-H(Cl)) (Scheme 3) using a combination of scXRD and NMR spectroscopic studies. When stored as crystalline solid at 238 K, complex 1-H(Cl) is stable, but in solution the complex gradually decomposes over several hours at 297 K. scXRD studies of 1-H(Cl) confirmed the η³-C₃-cyclopropyl motif, bound in an η³ fashion to V (Fig. 1). Salient structural features of 1-H(Cl) are outlined in Table 1 and the table inset in Fig. 1. In the solid-state structure of 1-H(Cl), the V–C bond distances in 1-H(Cl) are 2.021(2), 2.075(2), and 1.990(2) Å, which fall in the range of short V–C single bonds. Cloke and coworkers previously identified a pseudo MT_d on V, where the cyclopropyl ring is partially bound to V to form the complex, (CO)₄V(η³-C₃-^tBuCC(^tBu)C^tBu)⁷ (VT_d1, Fig. 2 and Table 1). The V–C bond distances in the latter complex are, 2.056(9), 2.025(8), and 2.416(9) Å, with the two former distances agreeing nicely with 1-H(Cl). Additionally, when comparing the bond distances of 1-H(Cl) with Mo based MTd's (Fig. 2 and Table 1), 1-H(Cl) displays comparable bond lengths to the alkyne metathesis catalyst, [SiP]Mo(η³-C₃-EtCC(Et)CEt) (where SiP = C₆H₃(C₆H₄SiPh₂O)₃ (MoT_d1),¹⁰⁴ which has an average Mo–C bond length of 2.062(5) Å. Whereas the complex (CO)₂(Cp)Mo(η³-C₃-PhCC(Ph)CPh) (MoT_d2)²⁷ prepared from cyclopropenium oxidative addition, exhibits an average Mo–C bond length of 2.216(3) Å. Moreover, the M-centroid (central point of the cyclopropyl ring) distance in 1-H(Cl) is 1.844 Å, which is significantly shorter than VT_d1 (2.075 Å), MoT_d2 (2.05 Å), and slightly shorter than MoT_d1 (ref. 104) (1.884 Å). However, one must treat this comparison of the metrical parameters with caution since they depend on the identity of the metal center and coordination number imposed by the supporting ligand framework. The C–C bond distances in 1-H(Cl) (1.455(3), 1.453(3), 1.488(3) Å) also compare well to the previously isolated MT_d's and correspond to short C–C single bonds. The average acute bond angles in 1-H(Cl) tetrahedrane core (∠C–C–C: 60.0(1)°, ∠C–V–C: 42.31(8)°, and ∠V–C–C: 68.85(1)°) are well in line with values found for the MT_d core of [SiP]Mo(η³-C₃-ⁿBuCC(ⁿBu)CⁿBu)⁶⁶ (∠C–C–C: 60.0°, ∠C–Mo–C: 41.1°, and ∠Mo–C–C: 69.47°)).

Scheme 3. Synthesis of the dVCBD in 1 and protonation to form the low-spin [V^III] or [V^V] VT_d in 1-H(Cl), or the [V^V] VCBD in [1-H][BArF₂₄].

Fig. 1. scXRD structure of 1-H(Cl) (left). scXRD structures are illustrated with ellipsoids shown at the 50% probability level at 100 K. Aryl group fragments on the BDI⁻ ligand are shown in wireframe while hydrogen atoms (except on the BDI⁻ ligand and on the η³-C₃ VT_d) are omitted for clarity. A truncated scXRD of 1-H(Cl) highlighting the η³-C₃-Me₃SiCC(H)CSiMe₃ core (top right). V–C bond distances shown in angstrom (embedded table, bottom right).

Table 1. Selected metrical parameters obtained from scXRD studies of 1-H(Cl) in comparison to other MT_d's in the literature. Bond distances are reported in Å and angles in °.

Complex	M–C (Å)	C–C (Å)	M–C3 centroid (Å)	∠C–C–C (°)	∠C–M–C (°)
1-H(Cl)	2.021(2)	1.455(3)	1.844	59.1(1)	42.58(8)
	2.075(2)	1.453(3)		61.6(1)	41.82(8)
	1.990(2)	1.488(3)		59.3(1)	42.54(8)
MoTd1 (ref. 66)	2.058(5)	1.458(7)	1.884	60.3(3)	41.43(19)
	2.064(5)	1.458(6)		59.4(3)	41.02(19)
	2.065(5)	1.445(7)		60.3(4)	41.36(18)
MoTd2 (ref. 27)	2.215(3)	1.454(4)	2.05	61.1(2)	38.8(1)
	2.165(3)	1.413(5)		58.3(2)	36.7(1)
	2.270(3)	1.448(5)		60.6(2)	38.0(1)
VTd1 (ref. 7)	2.416(9)	1.380(11)	2.075	75.5(6)	49.5(3)
	2.056(9)	1.412(12)		53.1(5)	34.8(3)
	2.025(8)	1.710(11)		51.4(5)	35.7(3)

Fig. 2. Illustration of the MTd's discussed in Table 1.

In addition to crystallographic studies, NMR spectroscopic analysis in deuterated toluene (C₇D₈) further confirmed the connectivity of 1-H(Cl) in solution phase. A ¹H NMR spectrum recorded at 300 K (Fig. S16) showed broad resonances in the aliphatic regions, which were previously observed for 1. Therefore, VT-NMR studies were conducted to elucidate these broad features (Fig. S17). Additionally the VT-NMR study allowed us to estimate the energy barrier (ΔG^‡) for this transition,¹⁴⁰ which was determined to be 54.3 kJ mol⁻¹, which is quite similar to the ΔG^‡ value for 1 (51.1 kJ mol⁻¹). Cooling a sample of 1-H(Cl) to 250 K exhibited sharp signals in the aliphatic region allowing for the full characterization of 1-H(Cl) in the ¹H NMR spectrum (Fig. S19 and S20). A diagnostic singlet observed at δ 4.86 was determined to be the γ-CH methine of the BDI⁻ ligand (Fig. S20). This was confirmed by a ¹H–¹³C HSQC NMR experiment (Fig. S23), which showed a one bond correlation between this resonance and the C trace at δ 102.8. Moreover, the other sharp singlet resonance at δ 3.47 was assigned to the CH fragment on the η³-C₃-cyclopropyl ring. The evidence for this assignment derives from the aforementioned ¹H–¹³C HSQC spectrum (Fig. S23) and the ¹H–¹³C HMBC NMR experiments (Fig. S24) which showed one bond correlation to a carbon resonance at δ 128.3 (C_β) and two bond correlation to carbon resonances at δ 92.0 (C_α) and 113.8 (C_α′). Other salient features of this molecule include the trimethylsilyl (SiMe₃) groups which resonate at δ 0.51 and −0.35 in the ¹H NMR spectrum (Fig. S20). Akin to the characterization of 1, it was difficult to identify the ²⁹Si resonances of 1-H(Cl) through traditional ²⁹Si NMR techniques. Thus, a ¹H–²⁹Si HMBC NMR spectrum (Fig. S25) was collected on 1-H(Cl) which shows two Si resonances at δ 0.6 and −3.7 that exhibit two-bond correlation to the H resonances attributed to the SiMe₃ groups at δ 0.51, −0.35, and three-bond correlation to the CH fragment on the η³-C₃-cyclopropyl ring at δ 3.47. A ⁵¹V NMR spectroscopic experiment (Fig. S26) conducted on 1-H(Cl) at 300 K revealed a single resonance at a highly downfield shift of δ 2941 (Δν_1/2 = 1241 Hz). This resonance is much more deshielded with respect to precursor 1, δ 1185 (Δν_1/2 = 2070 Hz). Cumulatively, this MT_d scaffold around V significantly deshields the ⁵¹V spectral resonance.¹⁴¹ To our surprise, the electronic absorption spectrum (Fig. S28) for 1-H(Cl) shows two features at 669 (ε = 700 M⁻¹ cm⁻¹) and 535 nm (ε = 659 M⁻¹ cm⁻¹), that we attribute to d–d transitions. These transitions suggest that 1-H(Cl) could arguably be also considered a low-spin [V^III] complex instead of a high-valent [V^V] complex. Schrock and coworkers also discussed the ambiguity of the formal oxidation-state of the MT_d complex, (PMe₃)(Cp)(Cl)₂W(η³-C₃-^tBuCC(Me)C^tBu),⁵⁶ where the complex could either contain a [W^VI] core and a C₃R₃³⁻ ligand or a [W^IV] core and a C₃R₃⁻ (i.e. η³-cyclopropenyl) ligand, which have also been described in subsequent reports on W based MT_d's.^55,102 Thus, we can envision a similar process taking place for the VT_d1-H(Cl), where the complex could exhibit either a [V^V] core and a C₃R₃³⁻ ligand or a [V^III] core and a C₃R₃⁻ ligand. In addition, time-dependent density functional theory (TD-DFT) (Fig. S134) and a charge-transfer spectrum (CTS) analysis (Fig. S135) reveal four absorption bands above 500 nm with its main contributions from electron charge transfer from V to the C₃ unit and vice versa followed by charge redistribution at V.

Synthesis and characterization of a [V^V] VCBD with a β-hydrogen

To probe whether the identity of the anion plays a role in the formation of the VT_d in 1-H(Cl), we decided to test the reactivity of 1 with a proton source containing a weakly coordinating anion to minimize disruption of the V coordination sphere. Accordingly, one equivalent of Brookhart's acid, [H(OEt₂)₂][BArF₂₄]¹⁴³ (BArF₂₄ = B(3,5-(CF₃)₂C₆H₃)₄), was added as a solid to a stirring pentane solution of 1 at 238 K. This reaction resulted in a homogeneous red-orange solution giving away to a heterogeneous mixture after 30 minutes, which contained a red precipitate that was filtered, washed and identified to be the VCBD salt complex, [(BDI)V(κ²-C,C-Me₃SiCC(H)CSiMe₃)][BArF₂₄] ([1-H][BArF₂₄]), in an 86% isolated yield (Scheme 3). Red single crystals of [1-H][BArF₂₄] can be obtained from a concentrated THF solution layered with pentane at 238 K after 12 hours. A scXRD study confirmed the assignment of a discrete salt having a VCBD scaffold with a β-hydrogen (Fig. 3).

Fig. 3. scXRD structure of [1-H][BArF₂₄] (left). scXRD structures are illustrated with ellipsoids shown at the 50% probability level at 100 K. The BArF₂₄⁻ counter anion and aryl fragments in the structure are shown in wireframe while hydrogen atoms (except on the BDI⁻ ligand and on the κ²-C,C-vanadacycle) are omitted for clarity. Truncated scXRD structure of [1-H][BArF₂₄] highlighting the V(κ²-C,C-Me₃SiCC(H)CSiMe₃) core (top right). V–C bond distances shown in Å (embedded table, bottom right).

Salient structural features of [1-H][BArF₂₄] are listed in Table 2 with the VCBD scaffold being comparable to the neutral and previously reported complex, (dBDI)V(κ²-C,C-^tBuCC(H)C^tBu)¹⁰⁰ (VCBD₁) (Fig. 4 and Table 2), whereas the structural parameters of [1-H][BArF₂₄] differ from 1 quite significantly. Specifically, the V–C_β transannular distance increases from 1.971(2) Å in 1 to 2.024(2) Å in [1-H][BArF₂₄]. Additionally, the C_α/α′–C_β bond distances increase when comparing 1 (1.384(2) and 1.372(3) Å) to [1-H][BArF₂₄] (1.424(3) and 1.447(3) Å). Moreover, the V–C_α/α′ bond distances slightly decrease upon going from 1 (1.872(3) and 1.907(1) Å) to [1-H][BArF₂₄] (1.848(2) and 1.810(2) Å). Another distinct trend that was observed when moving from the dVCBD scaffold in 1 to the VCBD in [1-H][BArF₂₄] is the variation in the ∠C_α–C_β–C_α′ bond angle. While the structure of 1 shows a more obtuse angle at 131.90(2)°, the VCBD scaffold in [1-H][BArF₂₄] has a more acute angle at 122.2(2)°. All these structural changes from the dVCBD complex 1 to the VCBD scaffold in [1-H][BArF₂₄] suggests that the latter can be described as a more electron delocalized system, whereas in 1 the electron density resides predominantly on the C₃ fragment (see Section 2.11 of the SI for details). These trends are consistent with the reports from Schrock and coworkers who observed elongated distances of the W–C_β and C_α/α′–C_β bonds, truncated distances of the W–C_α/α′, and a reduction in the ∠C_α–C_β–C_α′ bond angle, when comparing the dMCBD scaffold in (Cp)(Cl)W(η^3-tBuCCC^tBu)¹²² with the MCBD scaffold in (Cl)₃W(κ²-C,C-^tBuCC(Me)CMe).⁵⁴

Table 2. Salient structure features in the vanadacycles reported in this study as well as comparison to previously reported vanadacycles dVCBD₁,¹⁰⁰VCBD₁,¹⁰⁰ and VCBD₂.¹⁰¹ Bond distances are reported in Å and angles in °.

Complex	V–Cα	V–Cα′	V–Cβ	Cα–Cβ	Cβ–Cα′	∠Cα–Cβ–Cα′	τ 4
dVCBD1 (ref. 100)	1.860(1)	1.908(1)	1.967(1)	1.376(2)	1.378(2)	131.7(1)	0.79
VCBD1 (ref. 100)	1.891(7)	1.788(9)	2.004(8)	1.410(1)	1.464(1)	124.2(7)	0.83
1 (ref. 138)	1.872(3)	1.907(1)	1.971(2)	1.384(2)	1.372(3)	131.90(2)	0.84
[1-H][BArF24]	1.848(2)	1.810(2)	2.024(2)	1.424(3)	1.447(3)	122.2(2)	0.82
1-H	1.887(2)	1.865(2)	2.069(2)	1.434(3)	1.431(3)	123.4(2)	0.80
[1-Me][OTf]	1.800(4)	1.836(5)	2.050(4)	1.459(7)	1.415(4)	119.6(7)	0.83
1-Me	1.871(2)	1.867(2)	2.105(2)	1.439(2)	1.437(2)	120.4(2)	0.83
[1-Me][BArF20]	1.839(2)	1.807(2)	2.055(2)	1.433(3)	1.449(3)	119.7(2)	0.80
VCBD2 (ref. 101)	1.821(2)	1.841(2)	2.058(2)	1.440(3)	1.438(3)	120.11(2)	0.82

Fig. 4. Illustrations of the dVCBD's and VCBD's discussed in Tables 2 and 3.

Owing to the diamagnetic nature of [1-H][BArF₂₄], extensive NMR spectroscopic studies were undertaken, and key spectroscopic features are outlined in Table 3. Notably, the ¹H–¹³C HSQC NMR spectrum (Fig. S34) shows one bond correlation to the resonance for the β-H at δ 5.34 to the sp² hybridized C resonance at δ 94.7. This spectral feature is in good agreement with VCBD₁ (ref. 100) (Fig. 4 and Table 3), which exhibits a similar correlation (¹H: δ 5.52 and ¹³C: δ 96.4). Interestingly, this resonance feature does not agree well with 1-H(Cl), which exhibits a ¹H–¹³C correlation at ¹H: δ 3.47 and ¹³C: δ 128.26. This difference can be attributed to the difference in the coordination environment of the C₃ fragment around V. Moreover, the ⁵¹V spectral resonance of [1-H][BArF₂₄] (δ 513, Δν_1/2 = 55 Hz) (Fig. S36) differs quite significantly from those of 1 (δ 1185, Δν_1/2 = 2051 Hz) and 1-H(Cl) (δ 2941, Δν_1/2 = 1241 Hz) (Table 3), which again supports our ongoing hypothesis that the geometry around V significantly influences the NMR spectral resonance features.¹⁴¹ On the other hand, the difference in the ⁵¹V chemical shift in comparison with the structurally similar complex VCBD₁ (δ 251.6, Δν_1/2 = 45 Hz) can be assigned to the different electronic features of the dBDI²⁻versus the BDI⁻ ligand frameworks. As like 1-H(Cl), no ²⁹Si NMR spectral resonances in complex [1-H][BArF₂₄] were observed at 297 K even when conducting ²⁹Si INEPT, and ¹H–²⁹Si HMBC experiments implying this complex to exhibit dynamic behavior in solution akin to 1. However, ¹H–²⁹Si HMBC NMR spectra collected at 223 K (Fig. S39) did exhibit two cross-resonances at (¹H, ²⁹Si) δ 0.51, −1.9 and −0.13, −1.5; which correlates to the two SiMe₃ resonances in the ¹H NMR spectral trace (Fig. S29). Interestingly, a cross-resonance for the β-H of the VCBD at δ 5.34 to the silicon of the trimethylsilyl fragments was not observed in [1-H][BArF₂₄], which was apparent for the VT_d complex, 1-H(Cl) (vide supra). This feature can be attributed to the difference in the coordination environment of the C₃ fragment around V when comparing the planar VCBD metallacycle versus the tetrahedral VT_d fragment. Unlike the VT_d in 1-H(Cl), we were unable to observe the resonances for the C_α in [1-H][BArF₂₄], due to the lack of any protons directly bonded to it, which in turn increases the relaxation time for those carbons, as well as its proximity to a quadrupolar ⁵¹V nuclei.¹⁴¹ We have previously observed this phenomenon when analyzing the complex (ArNC(Me)C(H)C(Me)C(SiMe₃)CC(SiMe₃))V NAr (Ar = 2,6-ⁱPr₂C₆H₃), where extensive NMR studies did not reveal C resonances for the C_α and C_β carbons proximal to V.¹³⁹ Thus the geometry around V plays a substantial role in the relaxation of nuclei proximal to the V nucleus. Lastly, the electronic absorption spectrum of [1-H][BArF₂₄] (Fig. S41 and S42) lacks any d–d absorption bands above 600 nm (unlike in 1-H(Cl), vide supra), thus suggesting this species to be more in accord with a [V^V] ion supported by a formally trianionic [Me₃SiC–CH CSiMe₃]³⁻ ligand. Indeed, this result is further corroborated by TD-DFT studies and CTS analysis (vide infra, Fig. S136 and S137).¹⁴⁴

Table 3. Salient NMR spectroscopic features in the vanadacycles reported in this study as well as comparison to previously reported vanadacycles dVCBD₁,¹⁰⁰ and VCBD₁,¹⁰⁰ with chemicals shifts (δ) and Δν_1/2 in (Hz) reported for each nuclei.

Complex	51V (Δν1/2)	29Si	VCBD 1H(13Cβ)	γ-CH, BDI 1H(13C)
dVCBD1 (ref. 100)	1353 (2320)	—	—	5.85 (100.80)
VCBD1 (ref. 100)	252 (45)	—	5.52 (96.38)	5.64 (105.90)
1	1185 (2051)	−16.86, −16.71	—	5.55 (96.70)
1-H(Cl)	2941 (1241)	0.63, −3.66	3.47 (128.26)	4.86 (102.84)
1-Me(Cl)	3258 (861)	1.41, −5.22	1.16 (135.98)	4.81 (103.70)
[1-H][BArF24]	513 (55)	−1.51, −1.91	5.34 (94.68)	6.78 (103.26)
[1-Me][OTf]	617 (58)	−3.27, −5.22	4.18 (169.73)	4.37 (91.35)
[1-Me][BArF20]	631 (54)	−2.88, −4.44	4.17 (168.60)	6.68 (103.32)

Synthesis of [V^V] VCBD salts without β-hydrogens

In addition to protonation of 1, we also explored other electrophilic species as well as the role of the anion. Treatment of 1 with one equivalent or an excess of trimethylsilyl chloride (Me₃SiCl) or trimethylsilyl triflate (Me₃SiOTf) resulted in no reaction. However, when the electrophile methyl triflate (MeOTf) was added as a neat liquid to a stirring solution of 1, an orange precipitate quickly formed in a 55% isolated yield, which was determined to be the [V^V] VCBD salt, [(BDI)V(κ²-C,C-Me₃SiCC(CH₃)CSiMe₃)][OTf] ([1-Me][OTf]) (Scheme 4). Single crystals of [1-Me][OTf] were grown from a concentrated THF solution layered with pentane, and a scXRD study (Fig. 5) showed a distorted tetrahedral [V^V] center with the triflate anion (OTf⁻) outside of the coordination sphere of V (τ₄ = 0.85, Fig. 5, top). However, the OTf⁻ is within proximity of the γ-CH methine (2.255 Å) of the BDI⁻ ligand, implying a hydrogen bonding interaction. Metrical parameters of [1-Me][OTf] outlined in Table 2 exhibit overall similar structural features to [1-H][BArF₂₄]. Specifically, the average V–C_α/α′ bond distance in [1-Me][OTf] (1.818(5) Å) are in the same range as in [1-H][BArF₂₄] (1.829(2) Å), whereas the transannular V–C_β bond distance in [1-Me][OTf] (2.050(4) Å) is slightly longer than in [1-H][BArF₂₄] (2.024(2) Å). Moreover, the ∠C_α–C_β–C_α′ bond angle differ only marginally ([1-Me][OTf]: 119.6(7)°; [1-H][BArF₂₄]: 122.2(2)°), and the average C_α/α′–C_β bond length in [1-Me][OTf] (1.435(7) Å) and [1-H][BArF₂₄] (1.437(3) Å) do not differ statistically, resulting in structurally similar VCBD scaffolds. Moreover, when comparing [1-Me][OTf] to a recently characterized VCBD without a β-hydrogen, (Ph₃SiO)(Me₃SiCH₂)V(κ²-C,C-Me₃SiCC(CH₃)C(4-(NMe₂)C₆H₄))¹⁰¹ (VCBD₂, Table 2 and Fig. 4), we observe similar structural parameters, with the most notable being the transannular V–C_β bond distance of VCBD₂ (2.058(2) Å), which is statically similar to [1-Me][OTf] (2.050(4) Å).

Scheme 4. Synthesis of the VCBD scaffold from methylation of dVCBD in 1 to produce [1-Me][OTf] as well as anion exchange to replace OTf⁻ with BArF₂₀⁻ by forming a more stable salt [1-Me][BArF₂₀].

Fig. 5. Top: scXRD structure of [1-Me][OTf] (left) and truncated scXRD structure of [1-Me][OTf] (right) highlighting the V(κ²-C,C-Me₃SiCC(Me)CSiMe₃) core and triflate counterion. Bottom: scXRD structure of [1-Me][BArF₂₀] (left) and truncated scXRD structure of [1-Me][BArF₂₀] (right) highlighting the V(κ²-C,C-Me₃SiCC(Me)CSiMe₃) core. scXRD structures are illustrated with ellipsoids shown at the 30% (for [1-Me][OTf]) and the 50% probability level (for [1-Me][BArF₂₀]) at 100 K. Counter anion (in [1-Me][BArF20]) and aryl fragments in the BDI⁻ ligand in the structures are shown in wireframe while hydrogen atoms (except on the BDI⁻ ligand) are omitted for clarity. For [1-Me][OTf], interaction between the triflate oxygen and γ-H of BDI⁻ is denoted with a black dotted line. V–C bond distances shown in angstroms (embedded tables).

In addition to scXRD studies, complex [1-Me][OTf] was probed via NMR spectroscopic studies (Table 3). Comparison of NMR spectroscopic features to [1-H][BArF₂₄] (vide supra) reveals that complex [1-Me][OTf] exhibits a highly deshielded ⁵¹V resonance at δ 617 (Δν_1/2 = 58 Hz) (Fig. S60), which matches quite well with VCBD₂,¹⁰¹ whose resonance was reported at δ 490. Additionally, a more deshielded ¹³C NMR (Fig. S55) resonance for the C_β in the planar VCBD ring at δ 169.7 (C_β resonance in [1-H][BArF₂₄], δ 94.7, vide supra), was confirmed via a ¹H–¹³C HMBC experiment (Fig. S59). This unique discrepancy may in fact relate to the transannular V–C_β bond distance which slightly elongates upon moving from [1-H][BArF₂₄] to [1-Me][OTf]. This interesting feature highlights the possible electron density in the transannular V–C_β bond, which can not only affect the distance between the atoms but also the chemical shift of the two atoms as well. Moreover, the electronic absorption spectrum for [1-Me][OTf] shows no evidence for d–d transitions (Fig. S63, S64 and for computational data Fig. S140 and S141), thus corroborating this VCBD system to most likely contain a high-valent [V^V] ion, akin to what was observed for [1-H][BArF₂₄]. The other salient spectroscopic feature in [1-Me][OTf] is the γ-CH resonance for the methine component of the BDI⁻ ligand, found at δ 4.37 (δ 91.4) in the ¹H–¹³C HSQC spectra (Fig. S57). These resonances are both shielded when compared to [1-H][BArF₂₄]. However, we suspect that this shielding phenomenon results from the hydrogen bonding being retained in solution with one oxygen atom of the OTf⁻ anion, and as implied by a combination of scXRD analysis and computational studies (vide supra).

Interestingly and qualitatively, solutions of complex [1-Me][OTf] decompose over several hours at 297 K, indicated by the color change from orange to green. Given the more congested environment in the VCBD scaffold in [1-Me][OTf], we speculated if the OTf⁻ anion could be playing a role in its decomposition. For this reason, we conducted an anion exchange with sodium tetrakis(pentafluorophenyl)borate,¹⁴⁵ [Na][BArF₂₀] (BArF₂₀ = B(C₆F₅)₄). Accordingly, a THF solution of [1-Me][OTf] was added to a THF solution containing [Na][BArF₂₀], and after work up of the reaction mixture, a new VCBD salt was isolated in 89% yield and identified to be the discrete salt [(BDI)V(κ²-C,C-Me₃SiCC(CH₃)CSiMe₃)][BArF₂₀] ([1-Me][BArF₂₀]) (Scheme 4). Unlike [1-Me][OTf], compound [1-Me][BArF₂₀] is notably more stable in solution at 297 K. A scXRD study of [1-Me][BArF₂₀] (Fig. 5, bottom) confirmed the formation of a discrete salt having a β-Me substituted VCBD framework, with the former being anticipated given the weakly-coordinating role of the anions OTf⁻ and BArF₂₀⁻. Salient structural parameters as well as some important NMR spectroscopic features are outlined in Tables 2 and 3, respectively. The structural features when comparing the VCBD [1-Me][OTf] to [1-Me][BArF₂₀] do not differ significantly. However, their ¹H and ¹³C NMR spectra do reveal some subtle differences. Although the Me group on the β-C in the VCBD scaffold in [1-Me][OTf] and [1-Me][BArF₂₀] have relatively similar chemical shifts (Table 3), the methine γ-CH on the BDI⁻ fragment is more diagnostic and sensitive to its chemical environment. Specifically, the latter chemical shift in [1-Me][BArF₂₀] (δ 6.68) more closely resembles that of complex [1-H][BArF₂₄] (δ 6.78) than that of [1-Me][OTf] (δ 4.37), which is in agreement with our original hypothesis involving the closer ion pairing in [1-Me][OTf] when compared to the borate salts (Table 3). This contrast also suggests that the anion–cation interaction in [1-Me][BArF₂₀] and [1-H][BArF₂₄] should be relatively weak (if any), in both solid state and solution phases, whereas [1-Me][OTf] retains some pairing of the F₃CSO₃⁻–HC_BDI in both phases. Although complex [1-Me][BArF₂₀] is stable in solution over several hours, these borate salts do gradually transform over several days when judged by the color change to green. While the electronic absorption spectrum of [1-Me][BArF₂₀] shows no features in the visible window (Fig. S88 and S89), storing the solution over a few days under inert atmosphere results in formation of a new species that exhibits an electronic absorption ∼650 nm suggesting a d–d transition and most likely the formation a species containing a [V^IV] ion (vide infra).

Synthesis and characterization of [V^IV] VCBD complexes from reduction of [V^V] VCBD complexes

Owing to the high-oxidation state of the [V^V] ion in VCBD salts and VT_d reported earlier, we suspected that these species could be transforming in solution to a new [V^IV] species when judged by their electronic absorption spectra (vide supra). To interrogate the redox behavior of the [V^V] VCBD species, we examined the chemical reduction of [1-H][BArF₂₄] and [1-Me][X] (where X = OTf⁻, BArF₂₀⁻) with cobaltocene (CoCp₂). Accordingly, adding a THF solution of the reductant to a THF solution of [1-H][BArF₂₄] or [1-Me][X] resulted in an immediate color change from red or orange to green. Work-up of the reaction mixture via extraction with pentane allowed for the separation of the cobaltocenium salts produced in the reactions ([CoCp₂][X]; X = OTf⁻, BArF₂₄⁻, and BArF₂₀⁻), which was identified by ¹H NMR spectroscopy.^146–148 Subsequent cooling of the alkane extract to 238 K yielded green crystals of neutral [V^IV] VCBD complexes in yields ranging from 60–90% (Scheme 5). These new complexes were identified as (BDI)V(κ²-C,C-Me₃SiCC(H)CSiMe₃) (1-H) and (BDI)V(κ²-C,C-Me₃SiCC(CH₃)CSiMe₃) (1-Me), respectively, using scXRD studies (Fig. 6) and a variety of spectroscopic methods. Complexes 1-H and 1-Me could be also prepared directly from 1via addition of Brookhart's acid or MeX (X = I⁻, OTf⁻) and subsequent treatment with CoCp₂, respectively.

Scheme 5. Synthesis of [V^IV] VCBD scaffolds [1-H] and [1-Me], from one electron reduction of [V^V] VCBD salts with CoCp₂. The process is chemically reversible since oxidation of [V^IV] with [FeCp₂][X] (X = BArF₂₄⁻, OTf⁻, and BArF₂₀⁻) reagents results in reformation of the VCBD salts [1-H][BArF₂₄], [1-Me][OTf], and [1-Me][BArF₂₀].

Fig. 6. Top: scXRD structure of 1-H (left) and truncated scXRD structure of 1-H (right) highlighting the VCBD core. Bottom: scXRD structure of 1-Me (left) and truncated scXRD structure of 1-Me (right) highlighting the VCBD core. scXRD structures are illustrated with ellipsoids shown at the 50% probability level at 100 K. Aryl group fragments in the structure are shown in wireframe while hydrogen atoms (except on the BDI⁻ ligand and on the κ²-C,C-vanadacycles) are omitted for clarity. V–C bond distances shown in angstroms (embedded tables).

The scXRD study on single crystals of 1-H and 1-Me (Fig. 6) shows [V^IV] ions and VCBD motifs confined to distorted tetrahedral geometries (τ₄ = 0.80 (for 1-H) and 0.83 (for 1-Me)). Salient structural features are listed in Table 2, with the most notable change being a considerable elongation of the transannular V–C_β bond distance (2.069(2) Å for 1-H and 2.105(2) Å for 1-Me) upon reduction of the [V^V] salts ([1-H][BArF₂₄] (2.024(2) Å), [1-Me][OTf] (2.050(4) Å), and [1-Me][BArF₂₀] (2.055(2) Å) and when compared to 1 (1.971(2) Å). Additionally, there is a subtle elongation of the transannular V–C_β bond from the protonated VCBDs (1-H and [1-H][BArF₂₄]) to the methylated VCBDs (1-Me, [1-Me][OTf], and [1-Me][BArF₂₀]). The ¹H NMR spectra of 1-H (Fig. S43) and 1-Me (Fig. S66) also showed broadened resonances outside the common δ 0–8 range, suggestive of a paramagnet, while a solution magnetic study using the Evans' method was in accord with a doublet ground state (µ_eff = 1.92µ_B (1-H) and 1.86µ_B (1-Me); 297 K, C₆D₆). Further evidence for these complexes having an unpaired electron derives from their electronic absorption spectra collected in THF solution which showed one absorption band that we attribute to a d–d transition at 660 nm (ε = 671 M⁻¹ cm⁻¹, for 1-Me, Fig. S68) and at 662 nm (ε = 199 M⁻¹ cm⁻¹, for 1-H, Fig. S45), respectively. TD-DFT analysis shows that these features can be understood as an overlap of multiple absorption bands. Specifically, calculated spectra revealed six absorption bands above 500 nm in 1-H (Fig. S138) and seven in 1-Me (Fig. S142). The CTS analysis of these two species clearly shows that the main contributions of these bands are electron redistributions at V, followed by charge transfer processes from V to the ligand, while the BDI⁻ ligand contributes slightly less than the C₃ scaffold (R = H, 1-H, Fig. S139; R = Me, 1-Me, Fig. S143). In both cases, the lower wavelength absorption bands are dominated by ligand-to-ligand charge transfer from the C₃ scaffold to the BDI⁻ ligand. Additionally, the electron redistributions in the BDI⁻ ligand and the [V^IV] to BDI⁻ ligand charge transfer are significant contributing processes.

Electron paramagnetic resonance spectroscopic (EPR) studies on [V^IV] VCBD systems

Having established that complexes 1-H and 1-Me were [V^IV] d¹ systems, we conducted EPR spectroscopic studies. The continuous wave (CW) X-band EPR spectra of 1-H and 1-Me were recorded in 1 mM solutions in toluene at 293 K (Fig. 7, top) and as a frozen glass at 95 K (Fig. 7, bottom). The EPR spectra of these neutral species at 293 K show an isotropic signal resulting from hyperfine coupling (A_iso = 4.89 mT, 1-H; 10.44 mT, 1-Me) of the unpaired electron with the ⁵¹V nucleus (I = 7/2, 99.75% nat. abundance) and a g_iso value of 1.97 (1-H) and 1.98 (1-Me) in accord with a [V^IV]-centered radical. At 95 K, the spectrum for 1-Me shows axial symmetry with g_‖ = 1.94 and g_⊥ = 1.99. In contrast, 1-H shows a rhombic spectrum with g₁ = 1.95, g₂ = 2.00, and g₃ = 2.005. Both spectra display well-resolved hyperfine coupling to a ⁵¹V nucleus (A₁ = 166 × 10⁻⁴ cm⁻¹ (18.3 mT), A₂ = 65.0 × 10⁻⁴ cm⁻¹ (6.96 mT) A₃ = 55.0 × 10⁻⁴ cm⁻¹ (5.89 mT) for 1-H; A_‖ = 165 × 10⁻⁴ cm⁻¹ (17.9 mT), A_⊥ = 57.0 × 10⁻⁴ cm⁻¹ (6.18 mT) for 1-Me). The EPR spectral data collected for the [V^IV] VCBD's at 293 K, closely resemble previously reported four-coordinate [V^IV] alkylidene complexes supported by the BDI⁻ ligand scaffold; [(BDI)V CH^tBu(THF)][BPh₄]¹³² (g_iso = 1.982, A_iso = 8.8 mT) and (BDI)V CH^tBu(I)¹³² (g_iso = 1.976, A_iso = 6.4 mT).

Fig. 7. Top, CW X-band EPR spectrum of 1-H (left) and 1-Me (right) recorded as a 1 mM solution in toluene at 293 K. Bottom, frozen spectra of 1-H (left) and 1-Me (right) recorded as a 1 mM solution in toluene at 95 K. Black trace is experimental data and the red trace is its simulated spectrum. The asterisk in the bottom left figure corresponds to the super hyperfine coupling of the ¹H nucleus, which could not be simulated.

Electrochemical and chemical redox studies involving VCBD complexes

Cyclic voltammetry (CV) experiments were performed on complexes 1-H (Fig. 8A and S47) and 1-Me (Fig. 8C and S71) to further understand the [V^V]/[V^IV] redox couple in these VCBDs. The respective CVs highlight two events; one irreversible cathodic event (−2.35 V, 1-H; −3.02 V, 1-Me, referenced to the [FeCp₂]^+/0 at 0.00 V, Fig. S47 and S71) and one feature which appears quasi reversible (−1.08 V for 1-H, Fig. 8A; −0.98 V for 1-Me, Fig. 8C). We have assigned the quasi-reversible event at more positive potentials to the [V^V]/[V^IV] redox couple while the more negative potential represents the irreversible [V^IV]/[V^III] reduction event. From the CV data it is apparent that the VCBD scaffold with a Me group on the β-C position, 1-Me, renders the [V^IV] center easier to oxidize but harder to reduce when compared to the β-H substituted derivative 1-H. To probe the reversibility of the [V^IV]/[V^V] redox couple we conducted a scan rate dependence experiment, followed by Randles–Sevcik analysis for complexes 1-H (Fig. 8B) and 1-Me (Fig. 8D), with the latter being more reversible based on the least squares fit.¹⁴⁹

Fig. 8. Cyclic voltammetry (CV) studies of 1-H (A) and 1-Me (C) in THF (1.8 mM) with supporting electrolyte [ⁿBu₄N][PF₆] (0.20 M) referenced to the FeCp₂^0/+ redox couple at 0.00 V, and scan rate = 100 mV s⁻¹ and Randles–Sevcik plots showing the reversibility of the anodic waves for 1-H (B) and 1-Me (D).

We then turned our attention to chemical redox studies involving VCBD complexes (Scheme 5). To further probe the anodic quasi reversibility of 1-H and 1-Me, we decided to turn to one-electron oxidants with more positive potentials > −0.90 V such as ferrocenium ([FeCp₂]⁺) salts. Accordingly, when a THF solution of 1-H was added to a THF solution containing ferrocenium tetrakis(3,5-trifluoromethyl)phenylborate,¹⁵⁰ [FeCp₂][BArF₂₄], an immediate color change from green to orange was observed, and subsequent work-up of the mixture resulted in the isolation of [1-H][BArF₂₄] in an 83% yield, which was validated spectroscopically by comparison to the independently prepared sample (vide infra). The cogeneration of FeCp₂ was identified via¹H NMR spectroscopic studies of the reaction mixture. In the case of 1-Me, addition of [FeCp₂][X] (X = OTf⁻, BArF₂₀⁻)^150,151 and subsequent work-up yielded the complexes [1-Me][OTf] (76% yield) and [1-Me][BArF₂₀] (91% yield), which were also confirmed spectroscopically (Scheme 5). We found that the one-electron process is chemically reversible, since reductions of [1-H][BArF₂₄], [1-Me][OTf], or [1-Me][BArF₂₀] with a reductant having a potential < −1.1 V, such as CoCp₂, resulted in formation of the prototypical green colored solutions from which neutral [V^IV] complexes 1-H and 1-Me could be isolated in high yield (>90%) by crystallization from toluene (Scheme 5). After extraction of the neutral [V^IV] complexes with pentane, the less soluble and diamagnetic cobaltocenium salts were identified by ¹H NMR analysis.^146–148

Rearrangements among the dVCBD, VCBD, and VT_d scaffolds

We have established that the addition of an electrophile (H⁺, Me⁺) to 1 resulted in conversion of the [V^III] dVCBD scaffold into a [V^V] VCBD salt when the stabilizing anion was weakly coordinating, as in the case of VCBD salts [1-H][BArF₂₄] and [1-Me][X] (X = OTf⁻, BArF₂₀⁻). The stability in these salts greatly depended on the weakly coordinating nature of the anion, since these can slowly transform to neutral [V^IV] VCBD complexes 1-H and 1-Me along with side products, which to this moment we have been unable to identify. When we utilize an electrophile with a coordinating anion, such as [PyH][Cl], the [V^III] dVCBD scaffold in 1 converts into the VT_d1-H(Cl). Given this information, we first explored the microscopic reverse of forming 1-H(Cl), namely, dehydrochlorination to reform 1. Indeed, reaction of 1-H(Cl) with a strong base, neopentyl lithium (LiNp, where Np = CH₂^tBu), resulted in the rapid extrusion of neopentane (NpH) and quantitative formation of 1 (Scheme 6). One could view the conversion 1-H(Cl) → 1 as an α-H abstraction step resulting in reformation of the transannular V–C_β bond via the build-up of charge in the dVCBD. The formation of 1 most likely proceeds via transmetallation to produce a transient tetrahedrane-neopentyl species {(BDI)V(η³-C₃-Me₃SiCC(H)CSiMe₃)(Np)} {1-H(Np)} (Scheme 6, top left), which could be understood as a low-spin [V^III] or a high-valent [V^V] species. We propose that {1-H(Np)}, would then undergo an α-H abstraction to lose NpH, which would lead to the scission of the Me₃SiC–CSiMe₃ bond in the VT_d. Inclusion of a carbanion in the C₃ framework induces a rearrangement to planar form to most likely diffuse and delocalize the formal 4− charge on the scaffold. Specifically, the dVCBD scaffold in 1 can relieve the build-up of charge among few atoms by reductively eliminating in the form of a dVCBD having a low-spin [V^III] center. In other words, the planar geometry of the dVCBD allows for better modularity between different redox states by taking advantage of a redox-active planar C₃ ligand. Hence, the conversion of 1 → 1-H(Cl), and its microscopic reverse, can be readily accomplished using Brønsted acid and base chemistry, respectively. Since coordination of the chloride is critical in formation of the VT_d, we explored removal of the halide via a one electron reduction of 1-H(Cl) to probe if rearrangement to the planar geometry could be promoted by a redox event. Accordingly, treatment of 1-H(Cl) with CoCp₂ resulted in smooth conversion of the dark red solution to the characteristic green color of a [V^IV] ion (vide supra). After separation of the insoluble [CoCp₂][Cl] salt,¹⁵² complex 1-H could be readily isolated from the reaction in ∼88% yield. We propose the conversion of 1-H(Cl) → 1-H to likely involve a transient [V^II] VT_d intermediate (Scheme 6, top right), which ring-opens to form the [V^IV] VCBD, 1-H (possibly through the oxidative addition of the Me₃SiC–CSiMe₃ bond). Complex 1-H can be also generated directly from 1 in a 76% yield by sequential addition of [PyH][Cl] and CoCp₂ (Scheme 6). Oxidation of 1-H with triphenylmethyl chloride (Ph₃CCl) cleanly reforms the VT_d scaffold in 1-H(Cl) in a 78% yield (Scheme 6).

Scheme 6. Interconversion of protonated metallacycles 1-H(Cl), 1-H and [1-H][BArF₂₄] and dVCBD (1) through redox and Brønsted acid–base chemistry. Proposed intermediates are displayed as grayed-out structures.

Based on the computational results inspecting the role of the anion in VT_d formation (vide infra), we surmised that formation of the methylated analog of 1-H(Cl), 1-Me(Cl), should also be feasible from the [V^IV] VCBD, 1-Me, via simultaneous single-electron oxidation and delivery of a chloride. Consequently, when 1-Me was treated with either Ph₃CCl or half an equivalent of lead(ii) chloride (PbCl₂), the green solution of the [V^IV] changed to a dark red akin to that of 1-H(Cl). In the case of PbCl₂ as the oxidant, a new dark red compound could be isolated after separation of the precipitated Pb⁰ metal, which was identified as the methylated VT_d, (BDI)V(η³-C₃-Me₃SiCC(CH₃)CSiMe₃)(Cl) (1-Me(Cl)), through both scXRD and multinuclear NMR spectroscopic measurements (Scheme 7 and Fig. 9).

Scheme 7. Synthesis of the VT_d scaffold from oxidation of the VCBD in 1-Me to produce 1-Me(Cl).

Fig. 9. scXRD structure of 1-Me(Cl) (left). scXRD structures are illustrated with ellipsoids shown at the 50% probability level at 100 K. Aryl group fragments on the BDI⁻ ligand are shown in wireframe while hydrogen atoms (except on the BDI⁻ ligand) are omitted for clarity. A truncated scXRD of 1-Me(Cl) highlighting the η³-C₃-Me₃SiCC(CH₃)CSiMe₃ core (top right). V–C bond distances shown in angstroms (embedded table).

In contrast to the oxidation of Ph₃CCl, which results in formation of Gomberg's dimer,¹⁵³ the use of PbCl₂ as an oxidant allows for cleaner separation of the organometallic product. Structurally, 1-H(Cl) and 1-Me(Cl) are strikingly similar, with the only notable difference being the replacement of the β-H with a methyl group. This is also reflected in the ¹H NMR spectral data, with the signal corresponding to the β-H at δ 3.47 in 1-H(Cl) being absent in 1-Me(Cl). A new resonance arises at δ 1.16, which we attribute to the β-Me protons (Fig. S94). At 300 K two inequivalent SiMe₃ resonances are observed, indicating a lack of C₂ symmetry in solution at room temperature, unlike 1 and 1-H(Cl). The highly deshielded ⁵¹V resonance of 1-H(Cl) shifts another 300 ppm downfield in 1-Me(Cl), where the resonance arises at δ 3258 (Δν = 861 Hz) (Fig. S101). Similar to 1-H(Cl), 1-Me(Cl) exhibits two low energy bands in the electronic absorption spectrum at 562 (ε = 700 M⁻¹ cm⁻¹) and 474 nm (ε = 659 M⁻¹ cm⁻¹) (Fig. S103), again supporting the assignment of the tetrahedranes as low-spin [V^III] systems, rather than high-valent [V^V].

Schrock and Lee have observed the transformation of a MCBD to a MT_d with group 6 MCBDs.^{56,66,102,104} Schrock and coworkers have found coordination of a Lewis base to promote rearrangement to a MT_d, whereas Lewis base dissociation reverses the process.^56,102 Computational studies by Fürstner, Neese, and coworkers have found the MT_d to be thermodynamically more favorable than the MCBD.⁷⁹ This pattern seems to be also operative herein since replacement of the chloride in 1-H(Cl) with sodium tetrakis[(3,5-trifluoromethyl)phenyl]borate, [Na][BArF₂₄], in fluorobenzene results in clean formation of the [V^V] VCBD complex [1-H][BArF₂₄] in 79% yield. This process is reversible, since treatment of the MCBD salts [1-H][BArF₂₄] and [1-Me][OTf] with lithium chloride (LiCl) results in rapid reversion back to the VT_d compounds 1-H(Cl) and 1-Me(Cl) (Scheme 6). Deprotonation of the metallacyclic β-H in [1-H][BArF₂₄] with sodium bis(trimethylsilyl)amide, (NaN(SiMe₃)₂), results in conversion back to 1 in a modest 63% yield (Scheme 6). Although interconversion between MCBD and MT_d scaffolds have been previously shown, the use of redox chemistry to promote their rearrangement, as well as Brønsted acid–base chemistry to interconvert dMCBD, MT_d, and MCBD scaffolds was not realized until this work.

Computational analysis of structure and bonding in vanadatetrahedrane and vanadacyclobutadiene complexes

Stemming from the extensive experimental observations regarding the transformations of the central dVCBD complex 1 discussed in the preceding sections, a comprehensive computational investigation was undertaken to address several key questions. Specifically, we sought to (i) determine the preferred spin state of complex 1, (ii) elucidate the nature of the bonding within the four-membered vanadacycles, (iii) rationalize the observed anion dependence in the formation of the VT_d complexes 1-H(Cl) and 1-Me(Cl), (iv) assess the presence of anion–cation van der Waals interactions in [1-Me][OTf], and (v) clarify the electronic pathways associated with H⁺ and Me⁺ addition to the dVCBD framework of 1.

We first carried out a benchmark study, considering four different combinations of methods and basis sets. The gas phase structures of compound 1-H were calculated and selected structural parameters were compared with the experimental scXRD data, clearly showing that the pure density functional PBE^154,155 in combination with an empirical dispersion correction D3 (ref. 156 and 157) and the triple ζ basis set def2-TZVP¹⁵⁸ provided the best agreement with the experiment (notation: PBE-D3/def2-TZVP; Table S1).

(i) Spin state of complex 1

The analysis of ten selected structural parameters of gas phase structures of compound 1 in low-spin (singlet) and high-spin (triplet) configurations at the PBE-D3/def2-TZVP level of theory showed that the agreement of all values with experimental data is greater for the low-spin configuration, which agrees with the revisited NMR analysis (vide supra). Furthermore, these calculations showed that at this level, the adiabatic singlet–triplet gap is ΔE_S–T(PBE) = −33.9 kJ mol⁻¹, indicating a low-spin ground state. To calculate the energy of the singlet–triplet gap in complex 1 with higher accuracy, single-point energies were calculated using the coupled cluster singles and doubles (CCSD) with perturbative triples correction method (CCSD(T)/def2-TZVP//PBE-D3/def2-TZVP).^159–163 From these energies the adiabatic singlet–triplet gap was computed to be ΔE_S–T/ad.(CCSD(T)) = −16.8 kJ mol⁻¹ and the vertical ΔE_S–T/ver.(CCSD(T)) = −70.5 kJ mol⁻¹. Disregarding the adiabatic energy difference between the two spin states and using the Boltzmann distribution theorem, a low population of 0.1% of the triplet state at 293.15 K can be estimated. At 243.15 K, the triplet population is only 0.02%. It can thus be concluded that the low population of the triplet state is not responsible for the broadening of the NMR spectral signals. Consequently, all experimental and calculated findings are in excellent agreement with a ground state low spin configuration in dVCBD complex 1. It should be noted that the electronic situation in this complex can likely be better described using multi-reference calculations as an open-shell singlet state. However, this type of analysis would go beyond the scope of this study and will be part of a more extensive comparison of the structure and bonding of this and related metallacyclic complexes.

(ii) Bonding in four-membered vanadacycles

The present bonding analyses are based on single-point calculations performed using the optimised geometries (PBE) and the corresponding hybrid functional PBE0 (ref. 154 and 164) (notation PBE0-D3/def2-TZVP//PBE-D3/W06/def2-TZVP). Given the poor anion–cation interactions in [1-H][BArF₂₀] and [1-Me][BArF₂₄], as well as minimal interaction of the OTf⁻ with the VC₃ scaffold in [1-Me][OTf], only V fragments [1-H]⁺ and [1-Me]⁺ were considered in our calculations. As for the calculations of 1, the optimized geometry for the singlet state agrees better with the experimental data than the optimized geometry in the triplet state. Therefore, only the singlet state geometry and calculations were considered and adapted for further calculations. To examine the topological bonding in the VC₃ scaffold in these VCBDs in more detail, we first performed a set of quantum theory of atoms in molecules (QT-AIM)^165–167 analyses (Fig. 10, left). Interestingly, a bond path (brown line), a bond critical point (blue dot), and two ring critical points (orange dot), are only observed for the V–C_β interaction in complex 1, whereas this interaction is absent in all other VCBDs, similar to other group 4 dMCBDs.^133–136 This suggests that the dVCBD scaffold in 1 possesses an electronic structure that is more in line with the dMCBDs on group 6 metal centers,^{63,116,122,127,168} rather than that of MCB-2,3-Ds isolated thus far for high-valent group 4 metals.^133–136 The lower V–C_β Wiberg Bond Indices (WBI) in the VCBDs (WBI: 0.71–0.82) compared to 1.16 for 1 lend further support to the result obtained from the QT-AIM analysis (Table S3). It is worth noting that in the complexes analyzed here, all V–C_α/α′ WBIs are higher (e.g., 1.45/1.54 in 1) than in the previously investigated group 4 MCB-2,3-Ds. This supports a more delocalized metal-carbon double bond character in the VC₃ systems presented here. In line with the latter, WBIs for the C_α/α′–C_β bonds in the VC₃ units range from 1.65 to 1.45, lower compared to values in group 4 MCB-2,3-Ds of more than 1.9.^133–136 These findings are corroborated by Electron Localization Function (ELF) analysis (Fig. 10, right). Specifically, a significant amount of electron density is confined around C_β in 1 (Fig. 10, top right, orange area around C77) and pointing away from V, indicating the presence of a lone pair of electrons at C_β. Interestingly, the electronic structure of the C₃ scaffold is better matched to that of bent allenes, such as the previously reported carbodiimide complex Cp₂Ti(κ²-N,N-^tBuNCN^tBu)¹⁶⁹ or alternatively “carbodicarbenes”^170,171 (Fig. S128). This further explains the observed reactivity of 1 towards electrophiles, addressing this increased electron density at the C_β, and in good agreement with our initial hypothesis.

Fig. 10. Left: Contour plots of the Laplacian of the electron density ∇²r of complexes in the VC₃ plane. Dashed lines indicate negative (local charge concentration), solid lines indicate positive values (local charge depletion). The Laplacian plot is overlaid with the molecular graph from QT-AIM analysis and Wiberg bond indices (italic small numbers). Brown lines indicate bond paths, brown, blue dots correspond to bond critical points, light brown dots indicate ring critical points. Density from PBE0-D3/def2-TZVP//PBE-D3/def2-TZVP calculation. Right: ELF plots of complexes in the VC₃ ring plane, where areas in red correspond to high electron density and areas in blue correspond to low electron density.

(iii) Anion-dependence in the formation of VTd scaffold in 1-H(Cl) and 1-Me(Cl)

As detailed above, formation of the VT_d complex only occurs when a formal “HCl carrier” is added to 1, 1-H/1-Me is treated with an oxidizing chloride source, or 1-H[BArF₂₄]/1-Me[OTf] is treated with LiCl. To determine the influence of the anion on the observed reactivity we first evaluated the formation of 1-H(Cl) and 1-Me(Cl), as well as the hypothetical complexes {1-H(OTf)} and {1-Me(OTf)}, through a computational analysis (PBE-D3/def2-TZVP). In the case of chloride complexes, a two-step process could be identified for the formation of the respective VT_d compounds from their respective VCBD complexes in which the chloride ion first enters the coordination sphere of V (Scheme S12). This is followed by an exothermic C–C coupling event of two carbon atoms of the former VCBD unit to form the VT_d fragment (Δ_RG^θ(Int3) = −74.5 kJ mol⁻¹ for 1-H(Cl) and −62.2 kJ mol⁻¹ for 1-Me(Cl) (Table S20)). In both cases, the latter step is rate-determining with energy barriers of ΔG^#(TS2) = 77.3 kJ mol⁻¹ for 1-H(Cl) and 55.0 kJ mol⁻¹ for 1-Me(Cl) (Table S20). For the corresponding triflate complexes {1-H(OTf)} and {1-Me(OTf)} (Scheme S13), only the transition state that corresponds to the C–C coupling event could be located. These had considerably higher kinetic barriers than for the chloride species (ΔG^#(TS1) = 117.8 kJ mol⁻¹ for {1-H(OTf)} and 119.4 kJ mol⁻¹ for {1-Me(OTf)}). These results along with the subsequent and almost thermoneutral formation of the corresponding VT_d complexes {1-H(OTf)} and {1-Me(OTf)}, illustrate the pronounced influence of the anion for the VT_d formation reaction (Δ_RG^θ(Int2) = −6.4 kJ mol⁻¹ for {1-H(OTf)} and −8.4 kJ mol⁻¹ for {1-Me(OTf)}) (Table S21).

(iv) Anion–cation van der Waals interaction in [1-Me][OTf]

Evaluation of interactions between the cationic VCBD fragments and the corresponding anions in the isolated compounds [1-H][BArF₂₄], [1-Me][OTf] and [1-Me][BArF₂₀] was probed using an interaction region identifier (IRI)¹⁷² analysis based on single point calculations on the scXRD structures (PBE0-D3/def2-TZVP//scXRD; SI, Section 2.10). This analysis reveals areas with attractive van der Waals (vdW, green) interactions between the anion and cation pairs in all three cases. However, only in the case of [1-Me][OTf] was the size of the anion such that a direct vdW interaction could be observed with the γ-CH unit in the backbone of the BDI⁻ ligand used here (Fig. 11) and consistent with the structural and spectroscopic data (vide supra).

Fig. 11. Plot of the IRI of the isolated complex [1-Me][OTf] with focus on the C–H anion interaction (circled in red). Color code: blue – chemical bond, green – vdW interactions, red – repulsions and steric effects in aryl rings.

(v) Addition of electrophiles H⁺ or Me⁺ to β-C atom in dVCBD

To further elucidate the reactivity of the [V^III] dVCBD scaffold as a nucleophile in 1, resulting in the formation of [V^V] VCBD salts [1-H]⁺ and [1-Me]⁺, we investigated this process in detail. Intrinsic Bond Orbitals (IBO) in the VC₃ unit of selected representatives were analyzed (PBE0/def2-SVPD//PBE-D3/W06/def2-TZVP) to illustrate the orbital contributions to bonding and reactivity in the dVCBD scaffold in 1. The IBOs for compounds 1, [1-H]⁺ and [1-Me]⁺ are rather similar (Table S4–S7) with two classical V–C_α/α′ and C_α/α′–C_β σ-bonds in the ring plane. The π-component is described by two IBOs each, which are composed of one V d-orbital and two p-orbitals on C_α–C_β and C_α′–C_β respectively. This picture reflects the experimentally determined bond lengths and suggests a delocalized double bond system. The V–C_β bond in 1 is composed of an in-plane d-orbital at V and a p-orbital at C_β, consistent with the increased electron density on the C_β atom observed in ELF analysis (Fig. 10, right). A corresponding IBO was not found in the cationic fragments [1-H]⁺ (Table S5) and [1-Me]⁺ (Table S7), and instead the more classical σ-bond to the H in [1-H]⁺ or CH₃ in [1-Me]⁺ is found. Building on this result, we investigated the electron flow during the formation of the latter two species. In relaxed potential energy scans the C_β–H and C_β–Me distances were systematically extended in cations [1-H]⁺ (Fig. S129) and [1-Me]⁺ (Fig. 12), respectively. Our calculations demonstrate that heterolytic cleavage of the C_β–H and C_β–Me bonds to H⁺ and Me⁺ traverse through a barrierless step. By virtue of microscopic reversibility: nucleophilic attack of the formal C_β lone pair to the Me⁺ is also barrierless but with high energy gain, which explains the experimentally observed immediate reaction. Furthermore, the IBO analysis of these reaction profiles 1 → [1-H]⁺ and 1 → [1-Me]⁺ reveal the origin of the C_β–H and C_β–Me bonding electrons from the former V–C_β bond during the nucleophilic attack on H⁺ and CH₃⁺. In more conventional terms, this reaction can be understood as a non-classical oxidative addition or an electrophilic addition reaction, with the primary coordination environment of the V ion being relatively unperturbed. As a result, the addition of an electrophile such as H⁺ or CH₃⁺ results in the net oxidation of the low-spin [V^III] ion in 1 to a high-valent [V^V] in the VCBD species [1-H][BArF₂₄], [1-Me][OTf], and [1-Me][BArF₂₄]. More generally, the oxidative addition observed in the dVCBD system represents a genuine case of ligand-centered oxidative addition (LC-OA). In contrast to conventional metal–ligand cooperative (MLC) processes, where bond activation occurs partially at the metal center, the bond formation in our system takes place exclusively at the C₃ scaffold. Computational analysis reveals that the dVCBD's C₃ ligand undergoes direct interaction with the substrate, forming new covalent bonds, while the metal center experiences a formal increase in oxidation state without participating directly in the bond-making event. This mechanistic scenario clearly differentiates our system from MLC-type pathways, where the metal actively mediates bond formation and substrate activation, as exemplified by Milstein-type pincer complexes.¹⁷³ Instead, the dVCBD scaffold functions as a redox-active, electronically responsive ligand, effectively serving as the primary site of substrate activation. In this respect, the present system bears conceptual resemblance to complexes supported by redox-active ligands, in which ligand-centered redox processes accompany bond activation events.^174,175 However, in contrast to these systems, where metal–ligand cooperativity typically involves shared redox participation, the reactivity observed here is essentially ligand-centered. Consequently, the metal acts primarily as an electron reservoir, undergoing formal oxidation only as consequence of such ligand-based transformation. Overall, this example highlights a distinct mechanistic class in organometallic chemistry, where oxidative addition is fully ligand-centered, offering new avenues for the design of redox-active scaffolds that mediate two-electron transformations without direct metal involvement.

Fig. 12. Relaxed potential energy profile for the stepwise elongation of the C_β–CH₃ distance in [1-Me]⁺ along with corresponding selected IBOs representing the flow of electrons.

Conclusions

In this report we disclosed a novel route to unusual cationic [V^V] and neutral [V^IV] vanadacyclobutadiene (VCBD) complexes, as well as entry to vanadatetrahedranes (VT_d). This was accomplished through the addition of various electrophiles to the deprotiovanadacyclobutadiene (dVCBD) 1, thus circumventing the need to access such scaffolds via [2 + 2]-cycloaddition of the alkyne across vanadium alkylidyne ([V^V] CR) species,^{100,101,128–130} which are quite rare and synthetically challenging to attain. By virtue of Brønsted acid–base chemistry we promote reversible C–C bond formation via the interconversion between a VT_d and the dVCBD. Isomerization between [V^V] VCBD and VT_d frameworks was also accomplished reversibly by exchange of coordinating and weakly-coordinating anions. Additionally, we have found that the VCBD scaffold exhibits a [V^V]/[V^IV] redox couple with minimal structural reorganization, if the anion of the oxidant does not coordinate to the vanadium metal center. In contrast, reduction of the VT_d promotes the cleavage of a C–C bond to form the [V^IV] VCBD motif whereas oxidation with a chloride source reverses the process. The reversibility of this [V^V]/[V^IV] redox couple in the VCBD systems was probed through chemical reversibility experiments and electrochemical studies. Structural studies of these complexes highlighted the tetrahedral nature of the VT_d and the planar ring in the VCBDs. In addition, multinuclear NMR spectroscopic analysis of the [V^V] VCBD and [V^V] complexes confirmed the existence of the complexes in solution. The [V^V] VCBD undergoes spontaneous conversion to the [V^IV] VCBD, and these radicals were further corroborated by electron paramagnetic resonance (EPR) studies which confirmed the formation of a metal centric S = 1/2 system. The conversion of a dVCBD to a VCBD salt or a VT_dvia the addition of electrophiles can be best understood as a secondary coordination sphere reaction involving an unusual ligand-centered non-classical oxidative addition or an electrophilic addition reaction with electron flow from the metal to the ligand. Conversely, the reverse reaction constitutes a non-classical reductive elimination or a nucleophilic abstraction reaction.

Author contributions

John B. Russell: writing – review and editing, writing – original draft, methodology, investigation, formal analysis. Dylan P. Grigas: review & editing, methodology, investigation and formal analysis. Eva Körber: methodology, investigation, formal analysis. Alexandra M. Bacon: investigation, formal analysis. Michael R. Gau: investigation, formal analysis. Karsten Meyer: writing – review and editing, supervision, funding acquisition. Fabian Reiß: writing – review and editing, writing – original draft, methodology, investigation, formal analysis. Torsten Beweries: writing – review and editing, supervision, funding acquisition. Daniel J. Mindiola: writing – review and editing, supervision, funding acquisition. All authors have given approval to the final version of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

CCDC 2379001–2379005 (1-H[BArF24], 1-H, 1-Me[OTf], 1-Me and 1-Me[BArF20]), 2446159 (1-H(Cl)) and 2522894 (1-Me(Cl)) contain the supplementary crystallographic data for this paper.^176a–g

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information: experimental details as well as spectral data. See DOI: https://doi.org/10.1039/d6sc05321a.