A Mononuclear and High-Spin Tetrahedral Ti^II Complex

Abstract

A high-spin, mononuclear Ti^II complex, [(Tp^tBu,Me)TiCl] [Tp^tBu,Me– = hydridotris(3-tert-butyl-5-methylpyrazol-1-yl)borate], confined to a tetrahedral ligand-field environment, has been prepared by reduction of the precursor [(Tp^tBu,Me)TiCl₂] with KC₈. Complex [(Tp^tBu,Me)TiCl] has a ³A₂ ground state (assuming C_3v symmetry based on structural studies), established via a combination of high-frequency and -field electron paramagnetic resonance (HFEPR) spectroscopy, solution and solid-state magnetic studies, Ti K-edge X-ray absorption spectroscopy (XAS), and both density functional theory and ab initio (complete-active-space self-consistent-field, CASSCF) calculations. The formally and physically defined Ti^II complex readily binds tetrahydrofuran (THF) to form the paramagnetic adduct [(Tp^tBu,Me)TiCl(THF)], which is impervious to N₂ binding. However, in the absence of THF, the Ti^II complex captures N₂ to produce the diamagnetic complex [(Tp^tBu,Me)TiCl]₂(η¹,η¹;μ₂-N₂), with a linear Ti=N=N=Ti topology, established by single-crystal X-ray diffraction. The N₂ complex was characterized using XAS as well as IR and Raman spectroscopies, thus establishing this complex to possess two Ti^III centers covalently bridged by an N₂^2– unit. A π acid such as CNAd (Ad = 1-adamantyl) coordinates to [(Tp^tBu,Me)TiCl] without inducing spin pairing of the d electrons, thereby forming a unique high-spin and five-coordinate Ti^II complex, namely, [(Tp^tBu,Me)TiCl(CNAd)]. The reducing power of the coordinatively unsaturated Ti^II-containing [(Τp^tBu,Me)TiCl] species, quantified by electrochemistry, provides access to a family of mononuclear Ti^IV complexes of the type [(Tp^tBu,Me)Ti=E(Cl)] (with E^2– = NSiMe₃, N₂CPh₂, O, and NH) by virtue of atom- or group-transfer reactions using various small molecules such as N₃SiMe₃, N₂CPh₂, N₂O, and the bicyclic amine 2,3:5,6-dibenzo-7-azabicyclo[2.2.1]hepta-2,5-diene.

Short abstract

Tetrahedral Ti^II complexes with high-spin d² electronic configurations have hitherto eluded isolation. We show that treatment of mer-[TiCl₃(THF)₃] with [Tl(Tp^tBu,Me)] and KC₈ affords such a complex: [(Tp^tBu,Me)TiCl]. The electronic structure of [(Tp^tBu,Me)TiCl] was explored by SQUID magnetometry, electrochemistry, X-ray absorption spectroscopy, and high-frequency and -field electron paramagnetic resonance as well as theoretical studies. Tetrahedral [(Tp^tBu,Me)TiCl] activates various small molecules including N₂, N₂O, a strained amine, THF, and an isonitrile, benefiting from the low coordination number and reducing power of the Ti^II ion.

Introduction

Divalent Ti compounds have been widely used in C–C bond coupling reactions, such as the Kulinkovich or McMurry type reactions,¹ and recently in catalytic C–N bond coupling reactions by Tonks.² However, the coordination chemistry of isolable Ti^II is largely restricted to octahedral systems of the types [TiX₂(L)] {X^– = Cl, CH₃, BH₄, and OPh and L = 2 dmpe [1,2-bis(dimethylphosphino)ethane];³ X^– = Cl and L = 2 TMEDA (tetramethylethylenediamine), 4 pyridines;⁴ X₂^2– = porphyrin,⁵ L = 2 THF (tetrahydrofuran), 2 phosphine (-oxide)} and [TiΤp₂] [Tp^– = hydridotris(pyrazolyl)borate].⁶ Other strategies to stabilize the large and highly reducing Ti^II ion are to coordinatively saturate it with soft or π acids, such as N₂,⁷ CO,⁸ isonitriles,⁹ bipyridine,¹⁰ 1,10-phenanthroline,¹¹ alkynes,¹² olefins,¹³ cyclooctatetraene,¹⁴ phosphines,¹⁵ or bulky Cp^–-based¹⁶ ligands. Common to these strategies is the reliance on maximizing the coordination number, and, notably, low-coordinate Ti^II complexes continue to elude isolation. In particular, we found that no example of a four-coordinate, high-spin Ti^II d² complex exists. Despite the aforementioned cases all being formally categorized as Ti^II, the strongly π-accepting nature of one or more of the ligands means that the metal does not typically behave as a diradical or powerful π base. Perhaps only in the case of trans-[TiCl₂(TMEDA)₂]^4a,4b could one argue that the Ti^II ion represents a bona fide example of a high-spin d² configuration, given the more innocent nature of the chloride and chelating diamine ligands. However, this complex is highly unstable in solution, undergoing dinuclearization or disproportionation reactions.¹⁷ More recently, Lin, Tilley, Ye, and co-workers as well as Deng and co-workers have used sterically bulky, cyclic (alkyl)(amino)carbene (cAAC) ligands to stabilize C₂-symmetric Ti¹⁸ and Hf¹⁹ centers of the type [(cAAC)₂MCl₂], wherein a combination of π-accepting properties of the ligand and/or low symmetry electronically favors closed-shell ground states.

Given the propensity for mononuclear Ti^II to disproportionate, especially if bearing halide or pseudohalide ligands, we decided to reinvestigate its chemistry with the ubiquitous²⁰ Tp^– (scorpionate) ligand. Lee and co-workers established that two Tp^– ligands can be substituted onto Ti^II, with each being κ³ (vide supra).⁶ However, the use of more sterically encumbering groups on the pyrazoles should allow one to isolate and stabilize tetrahedral geometries with divalent ions, which has been observed for most of the 3d metals. In fact, Petrov and co-workers recently reported the synthesis of a tetrahedral divalent V^II complex, [(Tp^tBu₂)VCl] [Tp^tBu₂– = hydridotris(3,5-di(tert-butyl)pyrazol-1-yl)borate²¹], via reduction of the V^III precursor [(Tp^tBu₂)VCl₂].²² Inspired by the work of Lee, Deng, Lin, Tilley, Ye, and Petrov, we have now isolated and fully characterized the only example of a high-spin, tetrahedral complex of Ti^II, by using a relatively weak-field ligand, in this case Tp^tBu,Me– [Tp^tBu,Me– = hydridotris(3-tert-butyl-5-methylpyrazol-1-yl)borate]. If one excludes the hypothetical compound [Tp^RScCl], this Ti complex would essentially complete the divalent tetrahedral transition-metal 3d series [Tp^RMCl] [Tp^R– = a sterically demanding κ³-tris(pyrazolyl)borate; M = Ti, V,²² Cr,²³ Mn,²⁴ Fe,²⁵ Co,²⁶ Ni,²⁷ Cu,²⁸ and Zn²⁹]. Herein, we provide a reliable synthetic route to a mononuclear, high-spin Ti^II ion confined to a tetrahedral geometry and provide complete spectroscopic and magnetic data for this unique system, including high-frequency and -field electron paramagnetic resonance (HFEPR) and X-ray absorption spectroscopy (XAS). This panel of characterization is complemented by theoretical, electrochemical, and reactivity studies. We demonstrate how the Ti^II ion readily binds THF (reversibly) as well as N₂. In fact, coordination even of a strong σ base/π acid, such as CNAd (Ad = 1-adamantyl) to the Ti^II ion does not perturb the high-spin nature of the metal ion, despite forming a rare example of a five-coordinate isocyanide adduct, [(Tp^tBu,Me)TiCl(CNAd)]. Last, we showcase how the Ti^II ion can be oxidized to Ti^IV to form platforms of the type [(Tp^tBu,Me)Ti=E(Cl)] (E^2– = O, NSiMe₃, N₂CPh₂, and NH).

Results and Discussion

Synthesis of Mononuclear Ti^III Complexes with a Sterically Encumbering Trispyrazolylborate Ligand

Lee and co-workers reported how [TiCl₂(TMEDA)₂] could be cleanly transmetalated with 2 equiv of [KTp] to afford the octahedral Ti^II complex [TiTp₂]. Intuitively, one would expect this reaction to proceed via the monomeric intermediate [TpTiCl]. Surprisingly, we found that no examples of mono-Tp-based Ti^II complexes exist, despite the V^II complex [(Tp^tBu₂)VCl] having been recently reported by Petrov and co-workers.²² An initial investigation of salt metathesis between mer-[TiCl₃(THF)₃] and [KTp^tBu₂] led to isolation of the Ti^III complex [(κ¹,κ¹,η²-Tp^tBu₂)TiCl₂], albeit in low (∼10%) yield and with the presence of impurities (Scheme 1). Single-crystal X-ray diffraction (XRD; Figure 1) studies revealed a highly distorted system where the Tp^tBu₂– ligand coordinates in an unusual κ¹,κ¹,η² mode, which has been observed only in coordination polymers³⁰ and for large metal ions that favor high coordination numbers, such as Ba^II,³¹ Sm^III,³² Yb^II,³³ and U^III.³⁴ Here, the severe skewing of one pyrazolyl arm results in an almost side-on interaction between both of its N atoms and the metal ion. A moderate degree of elongation for one of the B–N bonds [B1–N2, 1.597(3) Å] compared to the other two [B1–N4 and B1–N6, 1.562(3) and 1.542(2) Å, respectively] can be observed. Consequently, the N–N distances in each ring reflect a subtle degree of activation of the pyrazolyl groups, with the side-on-bonded one possessing a longer distance [1.403(2) Å] compared to the other two [1.382(2) and 1.385(2) Å]. Distortion of one of the sterically encumbered pyrazolyl arms in [(κ¹,κ¹,η²-Tp^tBu₂)TiCl₂] implies the possibility of decomposition reactions, involving B–N and N–N bond activation, especially in subsequent reduction chemistry, leading to the large and strongly reducing Ti^II ion. Not surprisingly, attempts to prepare pure samples of [(κ¹,κ¹,η²-Tp^tBu₂)TiCl₂] invariably resulted in impure bulk material containing traces of other species, including free pyrazole. Given the larger ionic radius of Ti^II (100 pm) versus V^II (93 pm), we decided to reduce some of the steric congestion in Tp^tBu₂– by replacing the tert-butyl group in the 5 position of each pyrazolyl arm with a methyl group. Theopold and co-workers have popularized this ligand scaffold with various 3d metals, including the early-transition-metal Cr in the formal oxidation states I–V.^23,35

Scheme 1. Synthesis of [(κ¹,κ¹,η²-Tp^tBu₂)TiCl₂].

Figure 1.

Two views of the solid-state structure of [(κ¹,κ¹,η²-Tp^tBu₂)TiCl₂], showing distortion of one pyrazolyl arm of the Tp^tBu₂– ligand. Thermal ellipsoids are at 50% probability. Solvent and H atoms (except on B) are omitted.

Following a modified recipe to prepare [(κ¹,κ¹,η²-Tp^tBu₂)TiCl₂], treatment of mer-[TiCl₃(THF)₃] with 1 equiv of [Tl(Tp^tBu,Me)] in toluene over 13 h at 70 °C resulted in the smooth formation of [(Tp^tBu,Me)TiCl₂] (1), which was isolated in 72% yield as blue crystals (Scheme 2). The use of [Tl(Tp^tBu,Me)] as opposed to the K⁺ salt improves the overall isolated yield because of the lower content of residual pyrazole that remains from synthesis of the alkali salt. X- and Q-band EPR spectroscopic studies of 1 in toluene glass afford typical S = ¹/₂ rhombic spectra, indicative of low symmetry around the Ti^III ion (X band, g = [1.973, 1.923, 1.848], 12 K; Q-band, g = [1.973, 1.924, 1.848], 2 K; Figure 2). Likewise, solution Evans (μ_eff = 1.70 μ_B, 27 °C, C₆D₆), IR (ν_BH = 2569 cm^–1), and UV–vis (d–d, 648 nm; ε = 35 M^–1 cm^–1) spectral studies are all in accordance with a monomeric d¹ species. While the starting material, mer-[TiCl₃(THF)₃], shows a UV–vis absorption spectrum with two transitions due to a Jahn–Teller-distorted excited E_g state (668 and 763 nm, ε = 7 and 4 M^–1 cm^–1, respectively), complex 1 shows a single transition at 648 nm with a 5-fold increase in intensity in accord with the lower symmetry about the Ti^III ion (Figure S51). XRD studies also confirmed a five-coordinate monomer more in line with a square-pyramidal geometry at the Ti^III ion (τ₅ = 0.27;³⁶Figure 3). Moreover, no isomeric species resulting from borotropic shifts³⁷ have been observed for 1, and significant twisting of the pyrazolyl arms is no longer as noticeable when judged by the B–N and N–N metrical parameters. The view down the Ti–B–H axis (Figure 3, right) clearly shows how twisting of the pyrazole arms in 1 is minimized compared to [(κ¹,κ¹,η²-Tp^tBu₂)TiCl₂].

Scheme 2. Synthesis of Precursor 2via Reduction of 1, Reactivity Studies Involving THF, N₂, CNAd, and Various Oxidants (N₂CPh₂, N₃SiMe₃, and N₂O), as Well as the Bicyclic Amine Hdbabh to Form the Parent Imide 7.

Figure 2.

X-band (derivative, left) and Q-band (absorption, right) EPR spectra of 1 (black curves, experimental; red curves, simulated). Key experimental (black text) and S = ¹/₂ powder pattern simulation (red text) parameters are given (W = Gaussian line widths, hwhm). The Q-band spectrum exhibits an absorption, rather than first derivative, line shape due to rapid passage effects.³⁸

Figure 3.

Solid-state structure of 1 from two different views. Thermal ellipsoids are at 50% probability. H atoms (except on B) are omitted.

Synthesis of a High-Spin, Tetrahedral Ti^II Complex and Exploration of Its Electronic Structure

A cyclic voltammogram of 1 revealed an irreversible one-electron anodic process with an anodic peak at 0.44 V (referenced to FeCp₂^0/+ as 0.0 V), while a reversible one-electron cathodic process could be observed at −1.85 V (Figure 4). In Lee’s [Tp₂Ti]PF₆ complex, reversible one-electron oxidation and reduction processes occur at 1.20 and −1.28 V versus SCE, i.e., at 0.65 and −1.83 V versus FeCp₂^0/+.³⁹ These features indicate that reduction potentials are largely conserved between five-coordinate 1 and pseudooctahedral [Tp₂Ti]PF₆, and, importantly, both Ti^III complexes provide access to the corresponding Ti^II ions in a reversible manner. In the case of 1, the nonreversible electrochemical features are likely due to significant structural distortion upon reduction, whereas the irreversible anodic feature might be due to the instability of a hypothetical four-coordinate [1]⁺ species, especially in the presence of electrolyte. Encouraged by the facile electrochemical reduction chemistry, we conducted a chemical reduction of 1 under Ar with KC₈ in THF over 10 min, which resulted in an immediate color change from blue to intense green. Subsequent workup of the reaction mixture resulted in the formation of [(Tp^tBu,Me)TiCl] (2) as light-green crystals in a respectable yield (78%; ν_BH = 2554 cm^–1; Scheme 2). XRD studies revealed a slightly distorted tetrahedral Ti^II ion in each of the two crystallographically independent molecules (Table 1 and Figure 5, top). The chlorido ligand and Ti–B vector in the conformers of 2 range from being aligned in a nearly linear fashion (2-lin; τ₄ = 0.75 and τ_δ = 0.73⁴⁰) to being significantly bent (2-ben; τ₄ = 0.69 and τ_δ = 0.57) with Cl–Ti–B = 174.6(1)° and 162.9(1)°, respectively. This tilt of the chlorido ligand has notable consequences on the solid-state electronic structure of 2 (Supporting Information, section 13). Figure 5 (bottom) shows the 3-fold axis along the Cl–Ti–B vector. As expected for the larger Ti^II ion, the average Ti–Cl distances in 2 exceed those in 1 (2.386 vs 2.307 Å), whereas the Ti–N distances remain practically unchanged (2.169 vs 2.178 Å). In fact, the Ti–Cl bonds in 2 fall within the longest 7% of crystallographically characterized Ti–Cl distances in the CSD (version 2020.1), reflecting the large size of the Ti^II ion in 2. Solution (Evans, μ_eff = 2.72 μ_B, 27 °C, C₆D₆) and direct-current (dc) SQUID (μ_eff = 2.61 μ_B, 27 °C; Table 2) magnetization measurements of 2 are consistent with a high-spin Ti^II d² ion. Specifically, the SQUID magnetization measurements of 2 in the temperature range 2–300 K are consistent with an S = 1 species (for two independently isolated samples). Below 10 K, the magnetic moment drastically decreases to 2.24 μ_B (at 2 K), as shown in Figure 6, due to zero-field splitting (ZFS, vide infra). Variable-temperature and -field (VTVF) magnetization measurements between 1 and 5 T were also measured and simulated using a g_avg value of 1.85 (Figure 6, inset, and Table 2). Being confined to a tetrahedral coordination geometry, complex 2 shows ligand-field transitions at lower energies than those in 1, extending into the near-IR (NIR) region of the UV–vis spectrum (λ = 391, 701, 857, 889, and 925 nm; ε = 1500, 180, 340, 320, and 230 M^–1 cm^–1, respectively; Figure S52).

Figure 4.

Scan-rate dependence for cyclic voltammograms of 1 under N₂ in 0.1 M [ⁿBu₄N][PF₆] in THF.

Table 1. Geometric Parameters for the Two Crystallographically Independent Molecules of Complex 2 (Pca2₁).

geometry	τ4/τδ	B–Ti–Cl(deg)	Ti–Cl (Å)	Ti–Navg (Å)
2-lin	0.75/0.73	174.6(1)	2.375(2)	2.159
2-ben	0.69/0.57	162.9(1)	2.397(2)	2.180

Figure 5.

Solid-state structures of the two crystallographically independent molecules in 2 (left, 2-lin; right, 2-ben). Thermal ellipsoids are at 50% probability. H atoms (except on B) are omitted.

Table 2. Electronic Structure Parameters for Complex 2 Extracted with SQUID Magnetometry and HFEPR Spectroscopy.

SQUID
	D (cm–1)	E/D	gavg
dc, 1 T	–5.08	0	1.85
VTVF	–4.86	0	1.85

HFEPR
species	D (cm–1)	E (cm–1)	E/D	gx	gy	gz
2-pos	+1.55(1)	+0.38(1)	0.25	1.87(1)	1.89(1)	1.92(2)
2-neg	–5.91(2)	–0.31(2)	0.05	1.91(2)	1.94(2)	1.7(1)

Figure 6.

Solid-state dc and VTVF SQUID (inset) magnetization studies of 2 at 1–5 T from 2 to 300 K.

Given the unique geometry of 2 and the even number of unpaired electrons resulting in an integer electronic spin, we carried out HFEPR spectroscopic measurements of powdered samples of 2 at low temperatures (Figure 7, top, 216 GHz; spectra at additional frequencies are shown in Figures S59–S62). The spectra are characteristic powder patterns for a triplet spin state (S = 1) and reveal the presence of two independent S = 1 species. The number and intensity of turning points in the spectra could be readily simulated at various frequencies using two sets of spin-Hamiltonian parameters (Table 2). These parameters were obtained from 2D maps of field versus frequency (Figure 7, bottom), with a tunable-frequency methodology applied.⁴¹ The sign of D was obtained from the simulation of single-frequency spectra such as those shown in Figure 7 and was found to be negative for one species, hence designated as 2-neg and positive for the other, hence 2-pos (E was assigned the same sign as that of D by convention). The spectroscopically determined D values compare well with the average value of the same parameter extracted from SQUID magnetometry (−5.08 cm^–1; Table 2), and the presence of two crystallographically independent molecules in the asymmetric unit of the crystal structure of 2 readily explains the two S = 1 species observed by HFEPR. Apparently, the sign and magnitude of D is highly sensitive to the geometry of the Ti^II ion in 2. Ligand-field-theory (LFT) analysis using the angular overlap model provides an explanation for 2-neg and 2-pos having different magnitudes and signs of ZFS (Supporting Information, section 13). Conformer 2-lin, which more closely approximates trigonal (C_3v) symmetry, is assigned to 2-pos, whereas the more distorted conformer 2-ben is assigned to 2-neg.

Figure 7.

Top: Solid-state HFEPR spectrum of 2 at 4.5 K and 216 GHz (black trace). The colored traces are simulations using spin-Hamiltonian parameters (Table 2) of 2-neg (bottom) and 2-pos (top); red traces are simulations using positive D; blue traces are negative D values. Bottom: 2D plot of turning points in the powder spectra of 2 as a function of the frequency, marked as squares: full squares, 2-neg; empty squares, 2-pos. The lines are simulated with red color representing the turning points with the external magnetic field, B, parallel to the x axis of the zfs tensor, blue: B || y, and black: B || z.

The dependence of D on molecular structure was further explored by quantum-chemistry computational means. The g and D tensors were calculated for 2 by employing hybrid density functional theory (DFT) calculations using the B3LYP hybrid density functional with the CP(PPP) basis set on the Ti atom and the ZORA-def2-TZVP(-f) basis set on all other atoms; the calculations were based on the atomic coordinates from the crystal structures. The two conformers of 2 (cf. Table 1) each gave a unique set of parameters, in accord with the appearance of two spin systems in the HFEPR data. Notably, as with classical LFT (Supporting Information, section 13), the calculations reproduce the oppositely signed D values, although the magnitudes of D and E/D poorly match the experimental data (Table 3), which was also the case using LFT. In light of this, complete-active-space self-consistent-field (CASSCF) calculations with the strongly contracted N-electron valence perturbation theory 2 (NEVPT2⁴²) dynamical correlation correction were undertaken. Multiconfigurational approaches, including CASSCF/NEVPT2, predict ZFS for transition-metal complexes more accurately than DFT.⁴³ This is partially attributable to multiconfigurational methods giving a more accurate prediction of d–d excited-state energies, which are essential to accurately determining the spin–orbit coupling (SOC) contribution to the ZFS. The present calculations used the effective Hamiltonian theory and included both contributions from SOC and spin–spin coupling (SSC);⁴⁴ the SSC contribution is not directly accounted for in the classical LFT model.

Table 3. Calculated Spin-Hamiltonian Parameters for Conformers of Complex 2.

geometry	τ4	method	g1	g2	g3	D (cm–1)	E/D
2-lin/2-pos	0.75	DFT	1.966	1.971	1.973	–5.58	0.029
		CASSCF/NEVPT2a	1.792	1.829	1.916	+2.87	0.19
2-ben/2-neg	0.69	DFT	1.953	1.970	1.979	+7.59	0.020
		CASSCF/NEVPT2a	1.668	1.888	1.940	–7.77	0.068

^aCalculated using CAS(2,5) with an effective Hamiltonian SOC contribution and including a SSC contribution. CASSCF calculations are state-averaged over all d–d states of the d² configuration.

The CASSCF/NEVPT2 calculations were based on the quasi-restricted orbitals (QROs^43d) generated by the B3LYP calculation. The active spaces comprised two electrons and five orbitals [CAS(5,2)] and were averaged over all singlet (15) and triplet (10) ligand-field states expected for a d² ion in C_3v symmetry. State energies and ground-state configurations are included in the Supporting Information. The spin-Hamiltonian parameters (g, D, and E/D) calculated using this approach conform better with experiment than values calculated using DFT (Table 3). Importantly, species 2-neg observed by HFEPR can be plausibly assigned to conformer 2-ben, while 2-pos can be assigned to conformer 2-lin, exactly as proposed from LFT.

Reactivity of Complex 2 toward N₂, THF, and an Isocyanide

Transient Ti^II species are known to activate N₂, but reports of isolable and well-defined Ti^II species that can activate this inert gas remain quite rare.^8c,45 Potential synthetic applications of this activation chemistry include reductive splitting of N₂.⁴⁶ In order to prepare complex 2, manipulations must be carried out under a strictly pure Ar atmosphere, and solvents must be sparged thoroughly in order to avoid reactivity with residual N₂. Indeed, we found that reduction of 1 with KC₈ in THF under a N₂ (as opposed to Ar) atmosphere, and with workup carried out essentially identically with that for 2, resulted in the clean formation of a new diamagnetic product. Moreover, exposure of 2 in a pentane solution to an atmosphere of N₂ afforded the same species, quantitatively, over the course of minutes as a dark-green crystalline material. We should stress that, even in an Ar-filled glovebox, complex 2 gradually converts, in solution, to this new diamagnetic species, making it a potent scavenger of N₂! By contrast, crystalline 2 shows no signs of conversion into this new species when stored under N₂ over at least 24 h, reflecting the energetic penalties associated with structural changes in the solid state. The ¹H and ¹³C NMR, in addition to IR (ν_BH = 2559 cm^–1), spectral data of this new species feature a single pyrazolyl chemical environment, indicating that the new five-coordinate species undergoes rapid Berry pseudorotations in solution. XRD studies established formation of the bridged N₂ species [(Tp^tBu,Me)TiCl]₂(η¹,η¹;μ₂-N₂) (3) having a linear Ti=N=N=Ti topology (Scheme 2 and Figure 8). The elongated N–N bond of 1.207(4) Å with respect to free N₂ (1.0977 Å),⁴⁷ the short Ti–N distances [1.805(3) and 1.812(3) Å], and the diamagnetic nature hints to this motif being somewhere in the N₂^2– range. The two Cl ligands are oriented with dihedral angles of 85.43(7)–115.98(5)° (one Cl is disordered), consistent with complex 3 adopting both gauche and eclipsed conformations in the solid state. Dark-green 3 absorbs strongly across the entire UV–vis region with a maximum at 652 nm (ε = 580 M^–1 cm^–1; Figure S55), reflecting strong interaction between the bridging N₂ ligand and Ti centers.

Figure 8.

Solid-state structure of 3. Thermal ellipsoids are at 50% probability. Solvent and H atoms (except on B) are omitted. Disorder in two ^tBu groups and Cl1′ is not shown.

Returning to the Ti^III/Ti^II redox couple from the electrochemical studies of 1, it would be interesting to establish whether the identity of the reduced species might correspond to 2 or 3. During the synthesis of 2 and 3, we noted that the reduction of 1 with KC₈ in THF, in both cases, afforded an intense-green solution, the color of which matches neither the light-green color of 2 nor the dark-green color of 3. In both syntheses, removal of the volatiles produces an intense-green film, and the addition of pentane induces crystallization of 2 (under Ar) and 3 (under N₂) over the course of minutes. Surprisingly, an initial attempt at generating the ¹⁵N isotopologue of 3 by exposing 2 to ¹⁵N₂ in THF over 1 h resulted in the isolation of only unconverted 2. Likewise, a complementary experiment involving the reduction of 1 with KC₈ in THF under N₂, followed by removal of the volatiles in vacuo and subsequent workup under Ar, afforded a mixture of 2 and 3 in a 70:30% proportion. To better understand the apparent inhibition of N₂ binding to 2 in THF, we turned to UV–vis spectroscopic studies. Upon going from toluene to a THF solution, absorption from 2 at 391 nm (1500 M^–1 cm^–1) undergoes a slight redshift to 412 nm (380 M^–1 cm^–1). More importantly, the intense absorptions extending into the NIR region (λ = 701, 857, 889, and 925 nm and ε = 180, 340, 320, and 230 M^–1 cm^–1, respectively) disappear altogether and are replaced by a weaker and broad absorption at 606 nm (38 M^–1 cm^–1; Figure S53). Thus, titration of toluene solutions of 2 with THF (0–0.70 M) indicates that an equilibrium takes place between 2 and the Lewis base with an association constant in the range of 5–8 M^–1 (Figure 9, left). In neat THF (12.3 M), this corresponds to 98–99% of 2 being present as the five-coordinate [(Tp^tBu,Me)TiCl(THF)] (2-THF; Scheme 2 and Figure 9, right). Eventually, we isolated crystalline 2-THF (in the strict absence of N₂) by prolonged storage of 2 in pentane containing low concentrations of THF (−35 °C). XRD studies verify that 2-THF is a rare example of a five-coordinate, mononuclear, and high-spin Ti^II ion (τ₅ = 0.46). Moreover, the Ti–Cl [2.414(1) Å] and Ti–O_THF [2.251(2) Å] distances in 2-THF fall in the longest decile of crystallographically characterized Ti–Cl and Ti–O_THF distances in the CSD (version 2020.1), again reflecting the size of the Ti^II ion in 2-THF. Importantly, the reactivity and spectroscopic and structural data provide evidence that the reduction of 1 to ultimately produce 2 or 3 proceeds with 2-THF as a common intermediate (Scheme 2). It is apparent from our studies that THF blocks the binding site for N₂ activation. This finding is quite influential because it suggests that reductions of metal ions, often conducted in polar solvents such as THF, might actually disfavor binding of N₂ if the association constant of THF is large. In addition to this feature, the lability of the THF ligand is central for the Ti^II ion in subsequently attaining the tetrahedral geometry that is characteristic of complex 2.

Figure 9.

Left: UV–vis absorption spectra of 2 (1.6 mM) in toluene with increasing concentrations of THF. Right: Solid-state structure of 2-THF. Thermal ellipsoids are at 50% probability. Solvent and H atoms (except on B) are omitted.

Exposing complex 2 to isotopically enriched ¹⁵N₂ in pentane afforded [(Tp^tBu,Me)TiCl]₂(η¹,η¹;μ₂-¹⁵N₂) (3-¹⁵N), and collecting ¹⁵N NMR spectroscopic data revealed a downfield chemical shift at 309 ppm [vs NH₃(l) referenced at 0.0 ppm; top right in Figure 10]. The ¹⁵N NMR data indicate a reduced N₂ ligand; known [{L_nTi}₂(η¹,η¹;μ₂-¹⁵N₂)] complexes display ¹⁵N resonances in the range of 322–558 ppm versus NH₃(l).^7h,7i,7k,7l A comparison of the IR spectra of 3-¹⁵N and 3 allows the identification of low-energy Ti–N stretching frequencies (¹⁴N/¹⁵N at 769/749 cm^–1; Figure 10, top left), which gives a ratio (1.027) in good agreement with that calculated within the harmonic oscillator approximation for a ⁴⁸Ti–¹⁴N/⁴⁸Ti–¹⁵N isotopic system (calcd 1.0269). Resonance Raman spectroscopy obtained by excitation at 405 nm reveals a strongly red-shifted N–N vibration (¹⁴N/¹⁵N at 1401.1/1346.3 cm^–1; Figure 10, bottom), indicating a highly reduced N₂ unit consistent with a N₂^2– ligand (informative N–N Raman stretching frequencies;⁴⁸ N₂H₄, 1076 cm^–1; trans-N₂H₂, 1529 cm^–1). With these data, 3 can be formally construed as two Ti^III ions bridged by a N₂^2– ligand.

Figure 10.

Top left: IR spectral data for 3 (black) and 3-¹⁵N (red) along with a difference spectrum (blue). Top right: ¹⁵N NMR spectrum of 3-¹⁵N [with an external standard, AdC¹⁵N, referenced at 244 ppm vs NH₃(l) referenced at 0.0 ppm⁴⁹]. Bottom: Resonance Raman (405 nm excitation) data for 3 (black) and 3-¹⁵N (red) along with a difference spectrum (blue).

We also investigated the reactivity of 2 with a π acid, such as the isocyanide CNAd, to probe whether this reaction would yield a low-spin complex. Accordingly, treatment of 2 with CNAd resulted in clean formation of the adduct [(Tp^tBu,Me)TiCl(CNAd)] (2-CNAd) in 66% yield as a dark-maroon crystalline material. The ¹H NMR spectrum of 2-CNAd covers a ∼50 ppm chemical shift range (−8 to +40 ppm), similar to the range found for 2 (−2 to +46 ppm). Likewise, the magnetic moment extracted with Evans’ method (μ_eff = 2.72 μ_B, 28 °C, C₆D₆) attests to 2-CNAd still being high spin, consistent with the persistence of a Ti^II ion. Akin to 2-THF, an XRD study reaffirmed the connectivity in 2-CNAd, showing a rare example of a five-coordinate high-spin Ti^II complex (Figure 11). The linear Ti–C–N geometry [174.2(1)°] in 2-CNAd and the modest redshift of the C≡N stretching frequency versus free CNAd (2096 vs 2124 cm^–1; Figure 11) reflect a moderate degree of π-back-bonding, in line with the high-spin nature of 2-CNAd. However, the isonitrile ligand is not entirely spectroscopically innocent because the UV–vis spectrum of 2-CNAd displays much more intense charge-transfer bands (λ < 450 nm and ε > 3000 M^–1 cm^–1; Figure S56) than complexes 2 and 2-THF.

Figure 11.

Left: Solid-state structure of 2-CNAd. Thermal ellipsoids are at 50% probability. Solvent and H atoms (except on B) are omitted. Right: IR spectral data of 2-CNAd (black) and free CNAd (gray).

XAS Spectra of Complexes 1–3 and 2-CNAd

Ti K-edge XAS spectra were obtained for compounds 1–3 and the titanium(II) isocyanide adduct, 2-CNAd (Scheme 2 and Figure 12). The rising-edge energies obtained from the first derivatives of the spectra are presented in Table 4. Compounds 1 and 3 exhibit inflection points at 4974.5 and 4973.9 eV, respectively, while 2 and 2-CNAd have inflection points at 4971.6 and 4971.1 eV, respectively. The ca. 3 eV gap between the pairs of rising-edge energies accords with the assignment of 2 and 2-CNAd as bona fide Ti^II species, whereas 1 and 3 are physically defined Ti^III species. The XAS spectra also display prominent preedge absorption features, which were assigned with aid from time-dependent density functional theory (TDDFT) calculations (vide infra).

Figure 12.

Ti K-edge XAS data for complexes 1 (blue), 2 (black), 2-CNAd (gray), and 3 (red).

Table 4. Ti K-Edge XAS Preedge and Rising-Edge Energies for 1–3 and 2-CNAd.

complex	preedge energy (eV)	rising-edge energy (eV)
1	4968.0	4974.5
2	4966.8	4971.6
2-CNAd	4966.6	4971.1
3	4967.3	4973.9

Electronic Structure Calculations

Hybrid DFT calculations using the B3LYP hybrid density functional with the CP(PPP) basis set on the Ti atom and the ZORA-def2-TZVP(-f) basis set on all other atoms were carried out using atomic coordinates from the crystal structures. The calculations were used as starting points for TDDFT calculations of the Ti K-edge XAS data. Calculated excitation energies correlated strongly and linearly with the experimental data (Figure S66; R² = 0.97). The linear fit was used to shift the calculated energies such that the calculated and experimental spectra overlap (Figures 13 and 14). Spectra for 2 were calculated for both independent conformers in the crystal structure (2-lin and 2-ben; Table 1) but were found to be effectively superimposable.

Figure 13.

Overlap of the experimental and TDDFT-calculated [B3LYP and CP(PPP) on the Ti atom and ZORA-def2-TZVP(-f) on all other atoms] Ti K-edge XAS spectra of 2 (geometry for conformer 2-lin) with acceptor MOs for preedge transitions plotted at an isovalue of 0.03 au. Calculated data have been energy-shifted following the correlation given in Figure S66.

Figure 14.

Overlap of the experimental and TDDFT-calculated [B3LYP and CP(PPP) on the Ti atom and ZORA-def2-TZVP(-f) on all other atoms] Ti K-edge XAS spectra of 3 with acceptor MOs for preedge transitions plotted at an isovalue of 0.03 au. Calculated data have been energy-shifted following the correlation given in Figure S66.

Assignments of the preedge features in the Ti K-edge XAS data of 2 and 3 were facilitated using molecular orbital (MO) diagrams produced by the initial single-point DFT calculations. Complex 2 exhibits a 2 + 1 + 2 d–d splitting diagram, as would be expected for an effectively C_3v system (Figure 15). The two d electrons reside within an effectively degenerate (e) level comprised of MOs of ca. 80% d character in accordance with the physical Ti^II d² assignment suggested by the XAS data. Preedge excitations thus comprise several quadrupole-allowed Ti 1s → 3d excitations that gain intensity due to 4p mixing in the C_3v environment.

Figure 15.

Truncated MO diagram depicting the ligand field for 2 (geometry for conformer 2-lin) using QROs generated following a spin-unrestricted (UKS) DFT calculation using the B3LYP hybrid density functional with the CP(PPP) basis set on the Ti atom and the ZORA-def2-TZVP(-f) basis set on all other atoms. Orbitals are plotted at an isovalue of 0.03 au.

The MO diagram of 3 (Figure 16) resulting from a spin-restricted hybrid DFT calculation depicts four electrons in two pseudodegenerate MOs of ∼50% Ti 3d and ∼30% N 2p parentages. An alternative electronic structure description for 3 involves the antiferromagnetic coupling of two Ti^III ions through the N₂ bridge. To explore these bonding scenarios, we carried out broken-symmetry (BS1,1) calculations on 3. However, unrestricted corresponding orbital analysis reveals that no pairs of α and β spin orbitals have low overlap (S < 0.75). Two pairs of spin orbitals have intermediate overlap (S ∼ 0.8); these are π-bonding combinations between two Ti 3d and two N 2p orbitals, which are effectively delocalized over the entire (Ti₂N₂)⁴⁺ fragment. Moreover, the calculated coupling constant⁵⁰ is J ∼ −5000 cm^–1, an excessively large magnitude to be considered magnetic exchange. Together, these values suggest that considering 3 as a pair of Ti^III centers antiferromagnetically coupled through the bridging N₂ fragment is inappropriate. Instead, the crystallographic, spectroscopic, and computational data for 3 are in line with a Ti^III–N₂^2––Ti^III core having two covalent bonds between N₂ and two Ti.

Figure 16.

Truncated MO diagram for 3 generated with a spin-restricted hybrid DFT calculation employing the B3LYP hybrid density functional, the CP(PPP) basis set on the Ti atom, and the ZORA-def2-TZVP(-f) basis set on all other atoms. Orbitals in black have large contributions from Ti 3d, while orbitals in gray have small (<10%) contributions from Ti 3d. Orbitals are plotted at an isovalue of 0.03 au.

Reactivity of Complexes 2 and 3 toward Small Molecules

Complex 2 is stable in the solid state at room temperature and can be stored over months without showing signs of decomposition. In a toluene or benzene solution, 2 is stable over several hours at room temperature, after which it degrades to a myriad of unrecognizable diamagnetic products (full degradation after 20 h). By contrast, 2 is stable for at least 24 h in a THF solution at room temperature (by forming 2-THF), reflecting the increase in stability achieved upon attainment of a five-coordinate geometry (vide supra). Complex 2 displays rich chemistry; for example, it rapidly reacts with N₃SiMe₃ to form the imide [(Tp^tBu,Me)Ti≡NSiMe₃(Cl)] (4) in 80% isolated yield along with N₂ extrusion (Scheme 2). Complex 4 is diamagnetic and has been characterized by a combination of IR and ¹H and ¹³C NMR spectra in addition to solid-state structural studies (Figure 17, top left). ²⁹Si NMR spectroscopy reveals an upfield resonance (−13.5 ppm, vs SiMe₄ referenced at 0.0 ppm), indicating that the “(Tp^tBu,Me)Ti≡N^–(Cl)” fragment in 4 is more electron-donating than a methyl group. The structure shows short Ti–N distances for the imide ligand [1.692(1)–1.700(2) Å] in accordance with metal–ligand multiple-bond character, whereas the Ti–N–Si angles [158.94(10)–160.24(9)°] indicate that the imide N is best described as an sp-hybridized motif. In contrast, when 2 is treated with N₂CPh₂, N₂ is not expelled, and instead of the hypothetical carbene “[(Tp^tBu,Me)Ti=CPh₂(Cl)]”, the diamagnetic diazoalkane adduct [(Tp^tBu,Me)Ti≡NNCPh₂(Cl)] (5) is isolated in 91% yield. Neither thermolysis (100 °C, overnight) nor photolysis (full spectrum from a Xe lamp, 1 h) promotes N₂ extrusion from 5. Thus, 5 behaves like Herberhold’s prototypical [Cp₂Ti(N₂CPh₂)(PMe₃)] complex⁵¹ but shows marked contrast to the more reactive [Cp₂Ti(diazoalkane)] systems studied by Bergman, Andersen, and Chirik.⁵² While 4 shows C_s symmetry in solution (two sets of pyrazole resonances), 5 displays a symmetry similar to that of 3 in solution, based on multinuclear NMR spectral studies, reflecting the larger separation between the CPh₂ and ^tBu groups in 5 compared to the SiMe₃ and ^tBu groups in 4. A solid-state structure confirmed retention of the N₂ unit in 5, as well as the presence of a short Ti–N multiple bond [1.722(1) Å; Figure 17, top right]. Bond distances within the diazoalkane ligand in 5 [C–N, 1.297(2) Å; N–N, 1.327(2) Å] correspond reasonably well with bond distances in the hydrazone, Ph₂CNNH₂ [C–N, 1.286(2) Å; N–N, 1.371(3) Å].⁵³ This hints at a dianionic diazoalkane ligand; notably, free diaryldiazoalkanes display much shorter N–N bond distances (1.12–1.16 Å).⁵⁴

Figure 17.

Top: Solid-state structures of 4 (two polymorphs with two and three crystallographically independent molecules, respectively) and 5. Bottom: Solid-state structures of 6 and 7 (two crystallographically independent molecules). Thermal ellipsoids are at 50% probability. Solvent and H atoms (except on B and N) are omitted.

We also explored the reactivity of the high-spin, tetrahedral Ti^II ion in 2 with other small molecules. N₂O reacts immediately upon contact with a benzene solution of 2 to form the terminal oxo [(Tp^tBu,Me)Ti≡O(Cl)] (6) along with 3 (Scheme 2), due to the concomitant release of N₂ and trapping by 2 equiv of the azophile 2. The greater affinity of 2 for N₂ than N₂O most likely results in capture of the former. However, one could also argue for a bimolecular mechanism⁵⁵ involving a N₂O complex,⁵⁶ “[(Tp^tBu,Me)Ti(N₂O)(Cl)]”, reacting with 2. Regardless, the reaction mixture progresses over 18 h to eventually form the oxo complex in quantitative spectroscopic yield, and workup results in colorless crystals of 6 in 75% isolated yield (Scheme 2). Independently, treating 3 with 1 atm of N₂O cleanly produces 6, thus rendering the former species a masked form of Ti^II ion.^8e,57 The Ti≡O bond in 6 is remarkably short [1.621(1) Å; Figure 17, bottom left], possibly in line with triple-bond character; this is also manifested in the ¹H NMR spectrum of 6, where the Me and ^tBu groups display broad resonances because of restricted rotation (e.g., via a Berry pseudorotation) around the central Ti≡O unit. This feature might contribute to the more restricted rotation in 4 compared to 3 and 5, but it is noteworthy that the ¹H NMR spectrum of 7 (a titanium imide akin to 4, vide infra) shows only a single pyrazole environment.

The bicyclic amine 2,3:5,6-dibenzo-7-azabicyclo[2.2.1]hepta-2,5-diene (Hdbabh)⁵⁸ delivers a parent imide moiety to the Ti^II ion of 2 by forming [(Tp^tBu,Me)Ti≡NH(Cl)] (7) in near-quantitative spectroscopic yield (Scheme 2). Separation of 7 from the side product anthracene can be achieved via filtration of the reaction mixture through activated charcoal, providing yellow crystals of 7 in 32% isolated yield. The ¹H–¹⁵N HSQC NMR spectrum of 7 unambiguously identifies the parent imide group through a cross-peak between a ¹H triplet (4.97 ppm) and a downfield ¹⁵N nucleus (458 ppm; Figure S36). The same conclusion can be derived from the IR spectrum of 7, which exhibits a strong NH stretching vibration (3296 cm^–1). Whereas terminal titanium–oxo ligands are not uncommon, terminal parent imides are quite a rare motif for group 4 transition metals.⁵⁹ A solid-state structural study of 7 (Figure 17, bottom right) confirms the terminal parent imide motif and a geometry similar to that of 4 and 6. The imide H for both independent molecules was located in the difference Fourier map, and the Ti–N bond distances [1.674(4)–1.680(4) Å] fall within the range for terminal titanium imide complexes [1.627(8)–1.747(2) Å].^59a−59d The transfer of a parent imide group to low-valent Ti, thereby forming a mononuclear group 4 transition-metal imide, is unprecedented. In a few instances, this type of reaction has been reported for group 5 metals, via NH group transfer from 2-methylaziridine to V^III and Ta^III ions.⁶⁰ Moreover, NH transfer reactions from aziridines or Hdbabh are known to generate transient imide functionalities, which are implicated in the formation of amide and nitride complexes of Ti, W, and Fe.⁶¹ The few known examples of titanium complexes having the parent imide group have been reported by several routes: deprotonation of NH₃,^59a protonation of a nitride salt,^59d or via a nitridyl radical that effects H-atom abstraction^59b,59c Along these lines, an alternative, nitridyl-mediated, route to 7 consists of treatment of the Ti^III precursor 1 with NaN₃ in THF. However, this reaction suffers from the sacrificial use of the [Tp^tBu,MeTi] fragment as a H-atom source as well as the partial substitution of chloride for azide, thus giving mixtures of the corresponding azide and chloride imide complexes [(Tp^tBu,Me)Ti≡NH(X)] [X^– = N₃ or Cl (7)] as well as other poorly defined decomposition products. The azide imide complex was identified via the characteristic ¹H NMR resonance from the NH group (triplet at 3.58 ppm; Figure S33).

With the availability of structural and vibrational data for the new complexes 1, 2, adducts of 2, and the derived products (3–7), some general conclusions about their geometries and electronic structure can be inferred (Table 5). With 1 as the exception (τ₅ = 0.27), the values of τ₅ fall closely around 0.5, such that complexes 2-THF, 2-CNAd, and 3–7 do not lend themselves to a classification as being either trigonal-bipyramidal or square-pyramidal. The divalent nature of 2-THF and 2-CNAd can be directly seen from the long Ti–E and Ti–Cl distances compared to complexes 3–7, both reflecting the single-bond character of the Ti–E bonds and the large size of the Ti^II ion. Complexes 2-THF, 2-CNAd, and 3–7 underline the strong tendency for complex 2 to undergo reactions that render the Ti ion five-coordinate. As the reverse to this, complexes 4 and 7 might, in principle, eliminate small molecules (Me₃SiCl and HCl) to generate a neutral titanium nitride functionality, but this mode of reactivity is not operational for the (Tp^tBu,Me)Ti≡E(Cl) scaffold. Finally, inspection of the molecular structures of all complexes reported herein shows that the Ti^II complexes have Ti–Cl and B–H vectors oriented in an antiparallel fashion, whereas the Ti^IV complexes have the same vectors aligned nearly perpendicular. Indeed, the average B–Ti–Cl angles correlate well with the average Ti–B separations, as shown in Figure 18, revealing that all complexes group into three distinct regions when judged by their formal oxidation states. The link between metric variation and the formal oxidation state likely reflects the increasing ionic radius encountered upon going from Ti^IV to Ti^II; thus, a reduced Ti center tends to slide out of the Tp^tBu,Me– ligand, accounting for the increase in Ti–B separations and the straightening of B–Ti–Cl angles. Importantly, complex 3 falls between the Ti^II and Ti^III regions in Figure 18, augmenting support for the assignment of 3 as consisting of two Ti^III centers covalently bound through an N₂^2– ligand.

Table 5. Geometric Parameters and Vibrational Spectroscopic Data for Five-Coordinate Complexes 1, 2-THF, 2-CNAd, and 3–7.

complex	E	τ5	Ti–Cl (Å)	Ti–E (Å)	νBH (cm–1)
1		0.27	2.2816(6)–2.3310(6)		2569
2-THF	O	0.46	2.414(1)	2.251(2)
2-CNAd	C	0.54	2.4123(5)	2.207(1)	2557
3	N	0.46–0.50	2.317(2)–2.360(2)	1.805(3)–1.812(3)	2559
4	N	0.46–0.59	2.3440(6)–2.3567(6)	1.692(1)–1.700(2)	2542
5	N	0.48	2.3486(4)	1.722(1)	2545
6	O	0.54	2.3468(6)	1.621(1)	2554
7	N	0.46–0.50	2.340(1)–2.348(1)	1.674(4)–1.680(4)	2543

Figure 18.

Average Ti–B distances versus Cl–Ti–B angles for the Ti complexes reported herein, representing formal oxidation states II–IV.

Conclusions

In summary, we have demonstrated the use of a sterically encumbering tris(pyrazolyl)borate ligand, Tp^tBu,Me–, to generate the Ti^III precursor, 1. Tuning of the steric profile of the Tp ligand (replacement of the 5-^tBu group with Me) proved essential for obtaining a Ti^III precursor with sufficient purity and stability for subsequent chemistry. Reduction of 1 under Ar furnishes a new geometry with a unique electronic structure: a stable tetrahedral Ti^II ion, 2, having two unpaired electrons. The Ti K-edge XAS spectrum of 2 is shifted to lower energy relative to structurally similar complexes bearing formally Ti^III centers, consistent with 2 having a more reduced metal center. Complex 2 exhibits a ³A₂ electronic ground state because of its confinement to a tetrahedral coordination geometry, as demonstrated by the magnetic moments extracted from SQUID magnetometric studies; the spin-only d² behavior of 2 persists to temperatures as low as 10 K, whereupon ZFS results in a drastically lowered magnetic moment. HFEPR spectroscopic studies corroborate that 2 is a bona fide example of an S = 1 Ti^II ion via the observation of two spin triplet systems (ZFS parameters: D = −5.91 and +1.55 cm^–1, respectively). The origin of two S = 1 systems can be traced to the two crystallographically independent molecules (conformers) in 2, which display geometric distortions around the tetrahedral Ti^II ion, e.g., B–Ti–Cl angles spanning 162.9(1)–174.6(1)°. This assertion is supported by classical LFT and by CASSCF/NEVPT2 calculations, which closely reproduce these ZFS parameters and depict excited-state manifolds consistent with d² ions in an effective C_3v environment. Complex 2 is a potent scavenger of N₂, thus producing the dinuclear diamagnetic complex 3, where the N₂ unit coordinates in a linear fashion between two Ti centers. The use of ¹⁵N₂ to generate isotopically labeled 3-¹⁵N₂ confirms atmospheric N₂ as the source of the bridging N₂ ligand. Labeling studies further identify a N–N stretching frequency by Raman spectroscopy (¹⁴N/¹⁵N at 1401.1/1346.3 cm^–1), a Ti–N stretching frequency by IR vibrational spectroscopy (¹⁴N/¹⁵N at 769/749 cm^–1), and a ¹⁵N NMR resonance at 309 ppm. Ti K-edge XAS supported by TDDFT calculations accord with these complexes bearing physically defined Ti^III d¹ centers. Taken together, these results show that the bonding in 3 can be viewed as a covalent interaction between two Ti centers and N₂, producing a Ti^III–N₂^2––Ti^III core. A combination of electrochemical, reactivity, spectroscopic, and structural studies demonstrate that THF coordinates reversibly to 2 with a fairly high association constant on the order of 5–8 M^–1; the so-formed 2-THF adduct is impervious to N₂ binding, implying the intermediacy of a four-coordinate geometry in reactions, where 1 is reduced to generate 2 as well as 3. Even when using a strong-field ligand, such as CNAd, complex 2 forms a five-coordinate adduct, 2-CNAd, while preserving the high-spin nature of the Ti^II ion, owing to only a moderate degree of π-back-bonding, as seen from the linear Ti–C–N geometry [174.2(1)°] and slight redshift of ν(C≡N) (28 cm^–1). The availability of an open coordination site as well as two energetically high-lying d electrons renders tetrahedral 2 a versatile precursor to a family of five-coordinate Ti^IV complexes of the type [(Tp^tBu,Me)Ti≡E(Cl)], with E^2– = O, NH, NSiMe₃, and N₂CPh₂. These products result from O-atom transfer from N₂O and NH group transfer from the bicyclic amine Hdbabh to generate respectively the corresponding terminal oxide and parent terminal imide complexes. While N₃SiMe₃ reacts with 2 under N₂ extrusion, N₂CPh₂ retains an intact N=N=C spine up to 100 °C.

We are presently scrutinizing the unique electronic structure of additional tetrahedral Ti^II complexes, as well as examining their potential for generating unusual metal–ligand multiple bonds.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.inorgchem.0c02586.

• NMR, XRD, cyclic voltammetry, SQUID, X- and Q-band EPR, HFEPR, Ti K-edge XAS, and computational data (PDF)

Accession Codes

CCDC 2005728–2005738 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge viawww.ccdc.cam.ac.uk/data_request/cif, or by emailing data_request@ccdc.cam.ac.uk, or by contacting The Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 336033.

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.