The Reductive Activation of CO₂ Across a Ti=Ti Double Bond: Synthetic, Structural, and Mechanistic Studies

Abstract

The reactivity of the bis(pentalene)dititanium double-sandwich compound Ti₂Pn^†₂ (1) (Pn^† = 1,4-{SiⁱPr₃}₂C₈H₄) with CO₂ is investigated in detail using spectroscopic, X-ray crystallographic, and computational studies. When the CO₂ reaction is performed at −78 °C, the 1:1 adduct 4 is formed, and low-temperature spectroscopic measurements are consistent with a CO₂ molecule bound symmetrically to the two Ti centers in a μ:η²,η² binding mode, a structure also indicated by theory. Upon warming to room temperature the coordinated CO₂ is quantitatively reduced over a period of minutes to give the bis(oxo)-bridged dimer 2 and the dicarbonyl complex 3. In situ NMR studies indicated that this decomposition proceeds in a stepwise process via monooxo (5) and monocarbonyl (7) double-sandwich complexes, which have been independently synthesized and structurally characterized. 5 is thermally unstable with respect to a μ-O dimer in which the Ti–Ti bond has been cleaved and one pentalene ligand binds in an η⁸ fashion to each of the formally Ti^III centers. The molecular structure of 7 shows a “side-on” bound carbonyl ligand. Bonding of the double-sandwich species Ti₂Pn₂ (Pn = C₈H₆) to other fragments has been investigated by density functional theory calculations and fragment analysis, providing insight into the CO₂ reaction pathway consistent with the experimentally observed intermediates. A key step in the proposed mechanism is disproportionation of a mono(oxo) di-Ti^III species to yield di-Ti^II and di-Ti^IV products. 1 forms a structurally characterized, thermally stable CS₂ adduct 8 that shows symmetrical binding to the Ti₂ unit and supports the formulation of 4. The reaction of 1 with COS forms a thermally unstable complex 9 that undergoes scission to give mono(μ-S) mono(CO) species 10. Ph₃PS is an effective sulfur transfer agent for 1, enabling the synthesis of mono(μ-S) complex 11 with a double-sandwich structure and bis(μ-S) dimer 12 in which the Ti–Ti bond has been cleaved.

Introduction

In light of increasing global energy demands and environmental concerns over anthropogenic climate change,¹ there is considerable interest in the utilization of CO₂ as a renewable C₁ feedstock for industrially important chemicals.²⁻⁴ One of the most promising methods is the use of transition-metal catalysts to bring about selective and energy-efficient CO₂ incorporation into organic compounds or reduction to CO, which can then be converted into liquid fuels via Fischer–Tropsch chemistry.⁵

The difficulty in achieving CO₂ reduction lies in the thermodynamic stability of the CO₂ molecule and the kinetic barrier to its activation. As a result, highly reducing oxophilic metal complexes are typically employed. Stoichiometric CO₂ activation has been achieved using well-defined complexes of the main-group and transition metals⁶⁻¹⁰ as well as metal-free systems based on frustrated Lewis pairs (FLPs).¹¹⁻¹³ Lower-oxidation-state f-element complexes are also being increasingly explored in this context: for example, Evans and co-workers have shown that samarium(II) metallocenes facilitate the reductive coupling of CO₂ to form oxalate [O₂CCO₂]^2– 14,15 and the reductive disproportionation of COS to [S₂CO]^2– and CO.¹⁵ More recently, organouranium(III) complexes have attracted interest because of their significant capacity for CO₂ activation.¹⁶⁻¹⁸ For example, trivalent uranium metallocenes U(Cp^R)₃ reductively activate CO₂ (for Cp^R = η⁵-C₅H₄{SiMe₃}) and COS (for Cp^R = η⁵-C₅H₄Me) to generate μ-oxo/μ-sulfido products with concomitant release of CO.^19,20 Density functional theory (DFT) studies by Maron and co-workers have provided insight into the mechanism of reductive CO₂ and COS activation at Sm(II)²¹ and U(III)^22,23 centers, implicating in each case a bimetallic μ-CO₂ complex in which carbon dioxide was doubly reduced. The possibility of a concerted pathway involving a monometallic CO₂ intermediate with a [CO₂]^•– radical anion ligand was ruled out on the basis of the observed products. Meyer and co-workers reported the first example of a linearly coordinated [CO₂]^•– moiety in a U(III) complex supported by a stabilizing macrocyclic ligand, thus illustrating the importance of ligand architecture in reactivity.²⁴ Recent studies by Cloke and co-workers with the mixed-sandwich U(III) complexes (COT′)UCp^Me4R{THF}_x (COT′ = η⁸-1,4-{SiMe₃}₂C₈H₆; R = Me, Et, ⁱPr, ^tBu) have shown that the steric environment around the metal center plays a key role in guiding the possible reductive transformation pathways of CO₂, i.e., reductive coupling, disproportionation, or deoxygenation.²⁵ Nonetheless, better understanding of the mechanisms of these reductive transformations is required, with a view to tuning the structure of the active complex or the reaction conditions to give more useful product outcomes and assessing the potential for catalytic turnover.²⁶

Divalent titanium sandwich complexes are an underdeveloped series of compounds in this regard, despite the rich and varied reaction chemistry that “titanocenes” have shown with a variety of small molecules.^27,28 Alt et al.²⁹ reported the synthesis of the titanium CO₂ adduct Cp₂Ti(CO₂)(PMe₃) via simple ligand substitution of Cp₂Ti(PMe₃)₂, and a number of acetylene complexes of Cp₂Ti have shown CO₂ insertion into the Ti–C bond of the titanacyclopropene ring.³⁰⁻³³ In contrast, reaction of the permethyltitanocene complex Cp*₂Ti(η²-Me₃SiC₂SiMe₃) with CO₂ yields, after elimination of the silylalkyne, a mixture of the dimeric carbonate (Cp*₂Ti)₂(CO₃) and the carbonyl complex Cp*₂Ti(CO)₂.^34,35 Reductive disproportionation of CO₂ has been achieved by Cp₂Ti(CO)₂ to generate CO and the carbonate-bridged tetranuclear complex [(Cp₂Ti)₂(CO₃)]₂.³⁶ Titanium(III) complexes have also shown efficacy in reductive transformations of CO₂ to yield oxo,³⁶ carbonate,³⁶⁻³⁹ and oxalate⁴⁰ products, and the relative cheapness and low toxicity of this metal makes it attractive for fundamental reactivity studies.

Multiple bonds between metal atoms have fascinated chemists for over 50 years,⁴¹ and M–M bonded complexes have also shown ability to facilitate the reductive activation of CO₂. Recent examples include systems featuring reduced Mg–Mg,⁴² Cr–Cr,⁴³ and Pd–Pd cores⁴⁴ and a highly polar heterobimetallic Zr–Co complex recently reported by Thomas and co-workers.⁴⁵

Our interest in M–M interactions is focused on the aromatic ligand pentalene (Pn, C₈H₆), which is related to Cp by the edge-sharing ring fusion of two cyclopentadienyl fragments and shows a unique variety of coordination modes in its organometallic complexes.⁴⁶ In particular, the ability of Pn to bind two metal centers with η⁵ hapticity has proven ideal for the synthesis of bimetallic analogues of the ubiquitious bis(Cp)metallocenes, so-called “double-sandwich” complexes.⁴⁷⁻⁵³ We have recently reported the synthesis and characterization of the titanium double-sandwich complex Ti₂(μ:η⁵,η⁵-Pn^†)₂ (1) using the silylated pentalene ligand 1,4-{SiⁱPr₃}₂C₈H₄ (Pn^†). In addition to providing a rare example of a Ti–Ti multiple bond, 1 exhibits a solid-state structure unique among the M₂Pn₂ complexes in which the pentalene ligands are nonparallel and tilted around the Ti₂ core.⁵⁴ DFT studies on the model system Ti₂Pn₂ calculated an M–M bond order of 2 and an optimized geometry with C_2v molecular symmetry, providing a relatively open structure in which the frontier orbitals are sterically accessible.

The combination of a low-valent group IV metal and a metal–metal double-bonded system led us to explore the reactivity of 1 with small molecules. We recently communicated the room-temperature reaction of 1 with 1 equiv of CO₂ to give the bis(oxo)-bridged dimer [(η⁸-Pn^†)Ti(μ-O)]₂ (2), in which the Ti–Ti bond has been cleaved, and the dicarbonyl complex Ti₂(μ:η⁵,η⁵-Pn^†)₂(CO)₂ (3), which was independently synthesized from 1 with 2 equiv of CO.⁵⁵ Herein we explore the scope of this reductive transformation of CO₂, investigating its mechanistic pathway by the identification of key intermediates using spectroscopic and computational techniques. Furthermore, by the use of isoelectronic CS₂ and unsymmetrical COS as model molecules, we report the isolation and structural characterization of stable analogues for the proposed reaction intermediates, supporting the theoretical predictions. In the interests of brevity, an in-depth description of the bonding in these molecules and other related compounds is provided in the companion paper.⁵⁶

Results

Reaction of CO₂ with Ti₂Pn^†₂

The reaction of 1 with 1 equiv of either ¹²CO₂ or ¹³CO₂ was carried out in methylcyclohexane-d₁₄ (MeCy-d₁₄) at −78 °C (Scheme 1). This resulted in a color change from deep red to dark green, which accompanied quantitative conversion of 1 to an intermediate species, 4, that was observed by low-temperature spectroscopic measurements but had not been previously detected in a room-temperature reaction. The ¹H NMR spectrum of 4 at −30 °C showed four sharp doublets in the aromatic region, consistent with a diamagnetic complex that exhibits C₂ molecular symmetry on the NMR time scale, while the ¹³C{¹H} spectrum showed a singlet at 219 ppm with no further labeled ¹³C signals. In situ IR studies of 4 at −65 °C showed distinct bands at 1678 and 1236 cm^–1, which shifted to 1637 and 1217 cm^–1, respectively, in the isotopically labeled [¹³C]4. These IR bands are assigned to asymmetric and symmetric ν(OCO) stretches, respectively, of the CO₂ ligand by analogy with the monomeric complex Cp₂Ti(CO₂)(PMe₃),²⁹ which was studied in-depth by Mascetti et al.⁵⁷ and shows a C-coordinated bonding mode. Compound 4 could not be isolated because of its susceptibility to further reaction.

Scheme 1. Reactivity of 1 with CO₂ (R = SiⁱPr₃).

When 4 was allowed to warm from −78 °C to ambient temperature under dynamic vacuum, the red mono(oxo) complex 5 was isolated (Scheme 1). 5 could also be obtained by the slow addition of 1 equiv of N₂O to 1 via Toepler line. The ¹H NMR spectrum of 5 is again consistent with a diamagnetic complex with C₂ symmetry, and elemental analysis and mass spectrometry measurements support the proposed formulation. The molecular structure determined by single-crystal X-ray diffraction (XRD) reveals a “dimetallaepoxide” structure with an oxo ligand symmetrically bridging two formally Ti(III) centers (Figure 1). The Pn^† ligands are less tightly bound to the Ti₂ core of 5 relative to 1 and show the largest average Ti–C and Ti–centroid distances among all of the double-sandwich derivatives structurally characterized herein (Table 1). Interestingly, the ring slippage parameter (Δ = 0.15) and the twist angle between the Pn^† ligands and the dimetal unit (26.6(5)°) also have the largest values among all of these structures (see Figure 2 for definitions of these parameters).

Figure 1.

Displacement ellipsoid plot (30% probability) of 5 with hydrogen atoms and ⁱPr groups omitted for clarity. Primed atoms are generated by symmetry.

Table 1. Selected Distances (Å), Angles (deg), and Values of the Parameters Defined in Figure 2 for 5, 7, 8, 11, and 12 (Ct Denotes the η⁵ Centroid of a Pn Ring).

parameter	5	7	8	11	12
Ti–Ti	2.3991(7)	2.4047(5)	2.4432(10)	2.4880(8)	3.1402(6)
Ti–Cta	2.133(10)	2.058(9)	2.104(13)	2.1088(13)	1.9470(18)
Ct–Ti–Cta	139.722(12)	146.41(5)	138.02(9)	137.11(5)	57.09(9)
C53–Tiproximala	–	1.975(9)	2.238(4)	–	–
Ti–O/Sa	1.8607(15)	2.352(6)	2.497(16)	2.3728(8)	2.323(2)
C53–O/Sa	–	1.211(11)	1.657(4)	–	–
Δa	0.150(19)	0.005(7)	0.114(4)	0.135(2)	–
twist angle	26.6(5)	20.1(8)	20.8(2)	20.3(2)	21.1(5)
fold anglea	0.15(13)	8.7(4)	4.6(2)	3.86(8)	17.30(12)

^aAverage values.

Figure 2.

Definition of the ring slippage, twist angle, and fold angle parameters for pentalene structures.

Complex 5 is thermally sensitive, and allowing a C₆D₆ solution to stand for 12 h at room temperature led to the precipitation of green crystals of 6. X-ray crystallography established the connectivity in 6 as a μ-O dimer in which the Ti–Ti bond has been cleaved and one pentalene ligand binds in an η⁸ fashion to each of the formally Ti(III) centers (see Scheme 1 and Figure S1 in the Supporting Information). However, the low quality of the X-ray data precludes meaningful discussion of the metric parameters. In the ¹H NMR spectrum of 6 in THF-d₈ solution, only broad signals were observed, and the effective magnetic moment determined by the Evans method^58,59 was 1.73μ_B per Ti, a value consistent with one unpaired electron per metal center. Elemental analysis and mass spectrometry measurements were consistent with the proposed formulation for 6.

A monocarbonyl complex, 7, was prepared by the addition of 1 equiv of CO to a solution of 1 at −78 °C via Toepler line (Scheme 2). ¹H NMR spectroscopy showed that the reaction is quantitative, and red crystals of 7 were isolated from pentane at −35 °C in 62% yield. The solid-state structure of 7 (Figure 3) shows the CO ligand bound “side-on”, asymmetrically bridging the two Ti centers. The proximal Ti atom lies closer to the CO ligand in 7 than in the dicarbonyl complex 3, and an additional interaction between the CO and the distal Ti atom is clearly in effect in the former complex (Ti1–C53a = 2.245(10) Å, Ti1–O1a = 2.294(5) Å), causing the O–C53–Ti_proximal angle to deviate considerably from linearity (Table 1).

Scheme 2. Reactivity of 1 with CO (R = SiⁱPr₃).

Figure 3.

Displacement ellipsoid plot (30% probability) of 7 with hydrogen atoms and ⁱPr groups omitted for clarity.

The ¹H NMR spectrum of 7 features eight doublet resonances in the aromatic region (see Figure S3 in the Supporting Information), consistent with an unsymmetrical structure. The ¹H NMR spectrum of the isotopically labeled product [¹³C]7 differs from that of [¹²C]7 in that the Pn^† ring signal at 7.69 ppm appears as a triplet (J = 3.2 Hz) with an ¹H–¹³C (HMBC) correlation to the isotopically enriched μ-¹³CO. Inspection of the ²⁹Si{¹H} NMR spectrum, which consists of four singlets for [¹²C]7, reveals splitting of one signal into a doublet (J = 1.5 Hz) for [¹³C]7 (see Figure S4 in the Supporting Information). These features are attributed to through-space scalar couplings between the carbonyl carbon and the hydrogen and silicon in the proximal “wing-side” position of the Pn^† ligands above and below the Ti₂CO plane (Figure 4). The average distance between the C_CO and these H and Si atoms in the solid-state structure are 2.807(8) and 3.846(11) Å respectively. The solid-state IR spectrum of 7 exhibits a low ν(CO) stretching frequency of 1655 cm^–1, which shifts to 1616 cm^–1 upon ¹³C isotopic substitution, consistent with a side-on-bound carbonyl ligand.

Figure 4.

Through-space coupling between CO and Pn^† wing-side substituents.

Kinetic Studies

Experimentally, the reaction of 1 with CO₂ at low temperatures (between −90 and −30 °C) led to the quantitative formation of CO₂ adduct 4 (vide supra). When solutions of 4 were allowed to reach room temperature (in the absence of a dynamic vacuum) a color change from green to brown was observed, and the coordinated CO₂ molecule was reduced quantitatively over a period of minutes to give bis(oxo)-bridged dimer 2 and the dicarbonyl complex 3. ¹H NMR spectra recorded during the course of this reaction showed additional transient species, including (as the major components) the mono(oxo) compound 5 and monocarbonyl complex 7, identified by comparison with the spectra of independently synthesized samples.

The rate of decomposition of 4 was such that it could be conveniently followed by ¹H NMR spectroscopy. In a typical experiment, a 0.0295 M solution of 1 in MeCy-d₁₄ at −78 °C (in the presence of ferrocene, with which 1 does not react, as an internal standard) was treated with 1 equiv of ¹³CO₂, and the resulting mixture was shaken over 30 min until complete conversion to dark-green 4 was observed. The reaction mixture was then placed in an NMR probe held at 20 °C and allowed to equilibrate thermally. The conversion of reactants to products was measured over time and showed first-order exponential decay of 4 (Figure 5). The rate constant for this process was therefore determined at seven different temperatures in the range 9–40 °C (see Table S3 in the Supporting Information), and the relevant Eyring plot is shown in Figure 6. The derived activation parameters are ΔH^⧧ = 19.6 ± 1.0 kcal mol^–1, ΔS^⧧ = −9 ± 3 e.u., and ΔG_298 K^⧧ = 22.4 ± 1.5 kcal mol^–1.

Figure 5.

Concentration vs time curves for the decomposition of 4 to give oxo products 5 and 2 at 20 °C, as monitored by ¹H NMR spectroscopy.

Figure 6.

Eyring plot for the decomposition of 4 in MeCy-d₁₄ (R² = 0.990).

Reaction of CS₂ and COS with Ti₂Pn^†₂

Heteroallenes CS₂ and COS are commonly used to model the reactivity of CO₂.^60,61 Treatment of a pentane solution of 1 with CS₂ at −35 °C gave a dark-green solution upon warming to room temperature. Subsequent workup and cooling to −35 °C produced a red microcrystalline solid in 62% yield, which was identified by spectroscopic and analytical methods as 8 (Scheme 3). Crystals suitable for X-ray studies were grown from an Et₂O solution, and the molecular structure (Figure 7) reveals a bent CS₂ molecule in a μ:η²,η² binding mode between the two Ti centers.

Scheme 3. Reaction of 1 with CS₂ (R = SiⁱPr₃).

Figure 7.

Displacement ellipsoid plots (30% probability) of 8, 11, and 12 with hydrogen atoms and ⁱPr groups omitted for clarity. Primed atoms are generated by symmetry.

The CS₂ ligand in 8 exhibits a S–C–S angle of 137.4(3)°, while the C–S bond lengths (av 1.657(4) Å) are longer than the C–S bond length in free CS₂ (1.554 Å).⁶² The double-sandwich structure remains intact, with an elongated Ti–Ti distance (2.4432(10) Å) relative to 1 (2.399(2) Å)⁵⁴ and a smaller angle between the pentalene rings and the Ti₂ unit (av Ct–Ti–Ct = 138.02(9)°), giving a more open sandwich structure (Table 1). The ring slippage parameters (Figure 2) for each metal (Δ_Ti1 = 0.145, 0.092; Δ_Ti2 = 0.138, 0.082) indicate differential bonding to each side of the pentalene ligand.

The solid-state structure of 8 provides supporting evidence for the doubly bridging CO₂ ligand in 4. However, 8 is stable at room temperature and does not decompose even upon heating to 100 °C, in stark contrast to the CO₂ adduct 4, which decomposes above −30 °C to give the μ-oxo complex with concomitant liberation of carbon monoxide. This may be explained by the relative stability of CO compared with CS (Δ_fH_298 K^° = −26.4 and 66.1 kcal mol^–1, respectively)^63,64 and the driving force for Ti–O bond formation (Δ_bondH_298 K = 160.7 ± 2.2 kcal mol^–1) relative to Ti–S (Δ_bondH_298 K^° = 99.9 ± 0.7 kcal mol^–1).⁶⁵ The notable difference in the reactivity of CS₂ versus CO₂ with 1 prompted a study with COS.

The reaction of 1 with 1 equiv of COS at −78 °C in MeCy-d₁₄ produced a metastable adduct, 9, that shows an ¹H NMR spectrum consistent with the loss of C₂ molecular symmetry, as opposed to those of the C₂-symmetric species 4 and 8. The ¹³C{¹H} NMR spectrum of 9 shows a singlet at 282 ppm, a value intermediate between those of the CO₂ and CS₂ analogues (219 and 356 ppm respectively). In situ IR spectroscopy at −65 °C showed a band at 1498 cm^–1 which is tentatively assigned to the ν(CO) stretch of coordinated COS. However, the COS adduct 9 proved to be more thermally unstable than the CO₂ adduct 4, and the coordinated molecule underwent scission to mono(μ-S) mono(CO) species 10 in 41% conversion as determined by ¹H NMR spectroscopy at −40 °C. Intermediate 10 was independently prepared by the reaction of mono(μ-S) complex 11 with 1 equiv of CO; 11 itself was readily synthesized by the reaction of 1 with Ph₃PS (Scheme 4) and fully characterized by spectroscopic and analytical methods (vide infra).

Scheme 4. Reactivity of 1 with COS and Ph₃PS (R = SiⁱPr₃).

The synthesis of compound 11 was achieved by the reaction of 1 with 1 equiv of Ph₃PS in toluene (Scheme 4), and following workup 11 was isolated as red crystals in 52% yield. The proposed formulation of 11 is consistent with ¹H, ¹³C, and ²⁹Si NMR and elemental analysis data and was ultimately confirmed by a single-crystal XRD study. The molecular structure of 11 (Figure 7) shows the sulfide ligand bridging the two Ti centers in a three-membered ring, akin to mono(μ-oxo) complex 5. The Ti–Ti bond is considerably longer in 11 than in 5 (2.4880(8) and 2.3991(7) Å, respectively), and the larger bridging chalcogen atom is bonded at a greater distance from the Ti₂ core (av Ti–S = 2.3728(8); cf. av Ti–O = 1.8607(15) Å) with a more acute Ti–E–Ti angle (62.68(4)° and 80.28(8)°, respectively) (see Table 1). However, the smaller values of the Ti–Ct distance, ring slippage (Δ), and twist angle (Figure 2) for 11 reflect the fact that the Pn^† ligands are more tightly bound compared with 5, and this may explain the relative stability of this mono(μ-S) complex.

This triangular M–S–M structural motif is common among binuclear complexes with thiolato bridges but is unique in titanocene monosulfide chemistry. For example, [Cp*₂Ti]₂(μ-S) is a green paramagnetic complex with a Ti···Ti separation of 4.7069(7) Å and a Ti–S–Ti angle of 174.37(4)°.⁶⁶ The lack of a bonding (or an antiferromagentic coupling) interaction between the two d¹ centers in [Cp*₂Ti]₂(μ-S) gives rise to an electronic triplet state. In contrast, both 11 and 10 are diamagnetic.

The bis(μ-sulfide) complex 12 was independently synthesized by the reaction of 1 with 2 equiv of Ph₃PS in toluene (Scheme 4) and isolated in 80% yield. NMR spectroscopy, mass spectrometry, and elemental analysis confirmed the identity and purity of the product. Single crystals suitable for XRD analysis were grown from a saturated Et₂O solution at −35 °C, and its molecular structure is depicted in Figure 7, with important structural parameters listed in Table 1. Complex 12 is highly symmetrical; the solid-state structure shows a Ti₂S₂ unit that makes a regular ring with a perfectly planar diamond core (Ti1–S1–Ti1′–S1′ torsion angle = 0°). However, there is significant asymmetry in the Ti–S bond lengths (2.2238(16) and 2.4227(16) Å). The Ti–S–Ti angle is relatively acute (84.92(6)°), yet the Ti···Ti distance is considerably longer than in the bis(μ-oxo) congener (2.7376(6) Å).⁵⁵ The relative orientation of the η⁸-pentalene ligands (as defined by the angles between the two bridgehead C–C vectors) is 21.1(5)°.

Calculated Structures

Theoretical calculations were carried out on model systems with Pn (C₈H₆) as the ligand, mostly using two different functional/basis set combinations as embedded in two codes (ADF, BP/TZP; Gaussian, B3LYP/SDD). Selected calculated structural parameters are listed in Table 2 (in this article, experimentally observed/isolated compounds are labeled with Arabic numerals and their computational equivalents with the corresponding Roman numerals). In addition to the characterized compounds reported herein, structures were optimized for a possible intermediate and transition state in the transformation of IV.

Table 2. Selected Calculated Structural Parameters (Å, deg) for Optimized Structures^a.

structure	Ti–Ti	Ti–Ct	Ct–Ti–Ct	Ti–C	Ti–O/S	C–O/S	Ti–O/S–Ti
Ti2Pn2 (I)54	2.37, 2.31	2.00, 2.03	153, 158	–		–	–
PnTiO2TiPn (II)	2.74, 2.75	2.00, 2.02	56, 56	–	1.87, 1.85	–	95, 96
Ti2Pn2(CO)2 (III)	2.42, 2.42	2.05, 2.05	144, 144	2.08, 2.08		1.17, 1.17	–
			142, 144			1.17, 1.18
Ti2Pn2CO2 (IV)	2.41, 2.40	2.07, 2.10	141, 141	2.18, 2.14	2.27, 2.25	1.26, 1.29	–
Ti2Pn2O (V)	2.38, 2.36	2.13, 2.14	139, 140	–	1.87, 1.85	–	79, 79
PnTiOTiPn (S = 1) (VIt)	3.40, 3.69	1.96, 1.99	57, 57	–	1.86, 1.85	–	133, 180
PnTiOTiPn (S = 0) (VIs)	2.88, 2.80	1.96, 1.99	57, 57	–	1.85, 1.83	–	103, 100
Ti2Pn2CO (VII)	2.38, 2.36	2.06, 2.08	143, 143	2.04, 2.02	2.35, 2.26	1.21, 1.25	–
Ti2Pn2CS2 (VIII)	2.43, 2.41	2.10, 2.11	138, 138	2.27. 2.24	2.54, 2.58	1.67, 1.72	–
Ti2Pn2COS (IX)	2.41, 2.40	2.09, 2.11 (S)	138, 140 (S)	2.25, 2.20 (S)	2.63, 2.69 (S)	1.68, 1.73 (S)	–
		2.08, 2.09 (O)	140, 141 (O)	2.19, 2.17 (O)	2.19, 2.15 (O)	1.26, 1.29 (O)
Ti2Pn2S(CO) (X)	2.48, 2.47	2.12, 2.13	135, 136	2.08, 2.07	2.30, 2.31	1.16, 1.18	61, 61
		2.11, 2.13	136, 136		2.54, 2.57
Ti2Pn2S (XI)	2.44, 2.42	2.11, 2.12	137, 140	–	2.37, 2.39	–	62, 61
PnTiS2TiPn (XII)	3.13, 3.19	1.98, 1.99	57, 57	–	2.33, 2.33	–	84, 86
Ti2Pn2O(CO) (Int1)	2.46, 2.43	2.18, 2.17	135, 137	2.08, 2.07	1.76, 1.74	1.16, 1.17	–
		2.11, 2.13	138, 138		2.10, 2.07
Ti2Pn2OCO (TS1)	2.51, 2.50	2.11, 2.13	136, 136	2.16, 2.14	1.96, 1.93	1.36, 1.40	–
		2.09, 2.10	143, 142	2.75, 2.63	3.16, 3.09	1.23, 1.25

^aADF (BP/TZP) values are given in normal text, and Gaussian (B3LYP/SDD) values are given in italics.

Discussion

Structure of Ti₂Pn^†₂CO₂

Geometry optimization of the model system Ti₂Pn₂CO₂ (replacement of the silyl groups by H enhances the molecular symmetry) gave structure IV with C_2v symmetry and showing a bent CO₂ group bound symmetrically to the two Ti atoms (Figure 8). The calculated asymmetric and symmetric stretching modes for the CO₂ group are 1669 and 1214 cm^–1, respectively, which are in good agreement with the experimental results. Thus, we conclude that the high-symmetry NMR spectrum is a result of a symmetrical structure rather than a fast exchange process.

Figure 8.

Optimized geometries of IV and VIII.

Carbon dioxide has been observed to bind to dinuclear metal sites with three different bonding modes (Figure 9).^10,67−69 The proposed μ:η²,η²- coordination geometry has only recently been structurally authenticated by X-ray crystallography in an N-heterocyclic carbene nickel(0) dimer, [(IPr)Ni]₂(μ-CO)(μ:η²,η²-CO₂) (IPr = 1,3-bis(2,6-diisopropylphenyl)imidazol-2-ylidene).⁷⁰

Figure 9.

Crystallographically characterized bonding modes of CO₂ in bimetallic complexes.

Support for the symmetrical binding of CO₂ is provided by the authenticated structure of the sulfur analogue. Geometry optimization of Ti₂Pn₂CS₂ also resulted in a structure VIII of C_2v symmetry (Figure 8), in good agreement with the experimental stucture of 8. The solution IR spectrum of 8 shows a band at 1104 cm^–1 which, compared with the asymmetric stretch of free CS₂ (1535 cm^–1),⁷¹ is consistent with a sulfur-bound CS₂ moiety. This is in fair agreement with the stretching vibration for the model system VIII (1079 cm^–1) calculated by DFT methods.⁵⁶ Precedent for this type of CS₂ coordination involving both C=S π bonds is limited to a single example of a dinuclear Cu(I) complex recently reported by Limberg and co-workers.⁷² However, in this complex the mean plane of the CS₂ atoms is twisted with respect to the M–C–M mean plane by 49.1°, whereas in 8 these planes are nearly parallel (3.1°).

Structure of Ti₂Pn^†₂CO

The side-on binding of CO in 7 is sufficiently unusual to deserve comment. Compound 7 is, to the best of our knowledge, the first example of a “semibridging” carbonyl ligand between two titanium centers, and heterometallic examples involving titanium carbonyls are very rare. A weak semibridging carbonyl–gold interaction was inferred in the anionic gold–titanium complex [K(15-crown-5)₂][Ti(CO)₆(AuPEt₃)] on the basis of X-ray structural parameters and a difference between the solid-state and solution IR ν(CO) stretching frequencies.⁷³ Semibridging carbonyls have also been observed in the aforementioned Ti/M heterobimetallics Cp₂Ti(μ-PEt₃)(μ-η¹:η²-OC)M(CO)Cp (M = Mo, W),⁷⁴ Cp₂Ti(μ-C{R}CH₂)(μ-η¹:η²-OC)W(CO)Cp,^75,76 and Cp₂Ti(μ-CR)(μ-η¹:η²-OC)W(CO)Cp (R = p-Tol),⁷⁷ which show substantial reduction in the C–O bond order compared with typical terminal carbonyl moieties, as evidenced by the C–O bond distance and ν(CO).

The geometric function Ω = exp[d(M_distal–C)/d(M_distal–O)] measures the extent of the interactions of the distal metal atom with the C and O ends of the CO ligand.⁷⁸ This function can be used to differentiate between C- and O-bonded metal carbonyls in “end-on” (Σ-CO) and “side-on” (Π-CO) disposition. The average value for 7 (Ω = 2.69) lies comfortably in the range for a Π-CO (typically 2.2 < Ω < 3.3), and comparable values have been found for Ti/M heterobimetallics with semibridging CO ligands: Cp₂Ti(μ-C{p-Tol}CH₂)(μ-η¹:η²-OC)W(CO)Cp (Ω = 2.65)^75,76 and Cp₂Ti(μ-C{p-Tol})(μ-η¹:η²-OC)W(CO)Cp (Ω = 2.63).⁷⁷ The C–O bond distance in 7 (av 1.211(11) Å) is lengthened with respect to that in free CO (1.128 Å)⁷⁹ to an extent that is comparable to the Σ-CO complexes Cp₂Ti(THF)(μ-OC)Mo(CO)₂Cp (1.201(8) Å)⁸⁰ and Cp*₂Ti(Me)(μ-OC)Mo(CO)₂Cp (1.212(5) Å),⁸¹ in which the titanocene moiety behaves as a Lewis acid, coordinating to an oxygen atom of the Lewis basic [CpMo(CO)₃]⁻ fragment.

Geometry optimization of Ti₂Pn₂CO resulted in a structure with C_s symmetry (VII) in which the carbonyl ligand is bound side-on to the Ti dimer (Figure 10). A double bond is retained between the two Ti atoms in VII,⁵⁶ and the calculated Ti–Ti distance (2.38, 2.36 Å) is in reasonable agreement with the X-ray data (2.4047(5) Å). The wavenumber for the CO stretch in VII was calculated as 1644 cm^–1, which is in the low end of the range for symmetric bridging carbonyls (1850–1600 cm^–1)⁸² but in good agreement with the experimental value of 1655 cm^–1 in 7. The disfavoring of a symmetrically bridged CO is attributable to the absence of a high-lying orbital of suitable symmetry for back-donation in 1.⁵⁶

Figure 10.

Optimized geometry of VII.

Thermal Transformations of Ti₂Pn^†₂CO₂

The fate of 4 upon warming depends on whether the solution is sealed or under dynamic vacuum. Removing CO from the reaction leads to the formation of 5, which undergoes further reaction to give its isomer 6. A possible intermediate, undetected as yet experimentally, in the decomposition of Ti₂Pn^†₂CO₂ (4) is Ti₂Pn^†₂(μ-O)CO, in which a CO bond has broken, the detached O bridges the two Ti atoms, and the CO ligand formed is bonded to one of the Ti atoms. Geometry optimization of Ti₂Pn₂(μ-O)CO gave a local minimum for such a species, Int1 (Figure 11), in which the Ti–Ti distance (2.46, 2.43 Å) is still indicative of metal–metal bonding. The bridging O is placed asymmetrically, further from the Ti to which the CO is coordinated. The search for a transition state between IV and Int1 yielded structure TS1 shown in Figure 11, with geometric parameters given in Table 2; the transition state also possesses C_s symmetry. Both a Ti–C and a Ti–O bond are breaking and the remaining Ti–O bond has shortened, but the new Ti–O bond to the bridging O has not yet formed.

Figure 11.

Optimized geometries of Int1 and TS1.

Loss of CO from Ti₂Pn₂O(CO) (Int1) leads to a mono(oxo) product. Maintenance of the double-sandwich structure of the Ti₂Pn₂ fragment leads to a local minimum with C_2v symmetry, structure V. Two other structures were found with η⁸ coordination of Pn to Ti, one with a triplet state, VIt, and the other with a singlet state, VIs (Figure 12). The structures found for the triplet state by the two methods differed in the Ti–O–Ti angle: the ADF (BP/TZP) calculations optimized to a bent angle, while the Gaussian (B3LYP/SDD) calculations gave a linear Ti–O–Ti unit. Similar structures with η⁸ coordination were found for the singlet state by the two methods (Table 2). The energies of all three structures are close, and which structure is the most stable is method- and temperature-dependent (Table 3): ADF (BP/TZP) has the double-sandwich structure as the most stable, while Gaussian (B3LYP/SDD) estimates the SCF energy of the sandwich structure to be the lowest but on the basis of the free energy at 298 K predicts the triplet η⁸-coordinated structure to be the most stable. This is in agreement with experiment, as the double-sandwich structure 5 is known to convert to the triplet state 6 at room temperature. The large ring slippage and twist angle found in the solid-state structure of 5 may be construed as a “snapshot” of the Pn^† rings as they move across the Ti₂ core from a bridging (μ-η⁵,η⁵) bonding mode in 5 toward a capping (η⁸) mode in the structural isomer 6.

Figure 12.

Optimized geometries of the mono(oxo) species V, VIt, and VIs.

Table 3. Ti–O–Ti Angles (deg) and Relative Energies (kcal mol^–1) of Structures Found for Ti₂Pn₂O^56,a.

structure	Ti–O–Ti	ΔE(SCF)	ΔH298 K°	ΔG298 K°
V	79, 79	0, 0	0, 0	0, 2
VIt	133, 180	12, 4	13, 3	8, 0
VIs	103, 100	12, 19	13, 21	10, 20

^aADF (BP/TZP) values are given in normal text, and Gaussian (B3LYP/SDD) values are given in italics.

The free energy profile found for the binding of CO₂ to I and the transformation of IV when CO is removed is given in Figure 13 (the reaction pathway of V to form VIt was too complex to model). Binding of CO₂ to I is strongly favored, but the activation energy calculated for the decomposition of IV (22, 13 kcal mol^–1) is reasonably accessible at room temperature.

Figure 13.

Free energies for binding of CO₂ to I and the decomposition of IV under dynamic vacuum. ADF (BP/TZP) values are given in red and Gaussian (B3LYP/SDD) values in black.

The decomposition of 4 to 6 via 5 when CO is removed and the detection of 5 when CO is present strongly suggest that a similar route is followed when the decomposition proceeds further in the presence of CO (see Scheme 1). DFT calculations were used to examine the possible pathways for the formation of 2 and 3 (eqs 1–5 below; in the chemical formulas, the Pn ligands have been omitted; numbers in Roman numerals refer to the optimized, calculated structures in Table 2; numbers at the right are ADF, Gaussian reaction free energies).

1

2

3

4

5

Consideration of eqs 1 and 2 suggests that VI is oxophilic and attacks either V or the proposed oxocarbonyl intermediate (Int1) to form the bis(oxo) complex II. The more favorable values found for the reaction shown in eq 2 indicate that the presence of CO helps drive the disproportionation, explaining the lack of further decomposition when CO is removed from the reaction. Binding of a further molecule of CO to form the dicarbonyl III (eq 4) is favorable but not as exothermic as forming the monocarbonyl (eq 3). The overall decomposition is found to be thermodynamically favorable (eq 5). That the reaction appears to be first order suggests that the rate-determining step is the initial rupture of the bound CO₂ to form the proposed intermediate, rather than any subsequent disproportionation process, which would result in a second-order or more complex rate expression. Figure 14 shows the energies of intermediates involved in a possible disproportionation route. The experimental activation energy (ΔG_298 K^⧧ = 22.4 ± 1.5 kcal mol^–1) is in good agreement with the ADF (BP/TZP)-calculated activation energy for the decomposition of IV (22 kcal mol^–1; vide supra).

Figure 14.

Calculated free energies of intermediates in the disproportionation of Ti₂Pn₂O(CO) (Int1). Zero represents the energy of Ti₂Pn₂CO₂ (IV). ADF (BP/TZP) values are shown in red and Gaussian (B3LYP/SDD) values in black.

In terms of the final products of the reaction of I with CO₂, geometry optimization of Ti₂Pn₂(CO)₂ (III) by both ADF (BP/TZP) and Gaussian (B3LYP/SDD) gave a structure of C_s symmetry only slightly displaced from C_2v symmetry (Figure 15), which agrees well with that found experimentally. The calculated wavenumbers for the antisymmetric and symmetric CO stretches in III are 1947, 1899 cm^–1 and 1878, 1810 cm^–1, respectively, which are higher than the ν(CO) values found experimentally in 3 (1910 and 1987 cm^–1). The optimized structure of II had C_2h symmetry (Figure 15) with eclipsed pentalene rings and the Ti₂O₂ plane at an angle of 22° with respect to the pentalene bridging carbons.

Figure 15.

Optimized geometries of II and III.

Preliminary trials on an NMR scale to observe the proposed mono(μ-oxo) mono(CO) intermediate by reaction of [¹³C]7 with 1 equiv of N₂O or 5 with 1 equiv of ¹³CO were inconclusive. The ¹³C{¹H} NMR spectra were complex in each case, with the dicarbonyl complex [¹³C]3 identified as the major labeled product after warming to room temperature. ¹H NMR spectra confirmed the presence of bis(oxo) 2 in each case, which is a thermodynamic sink in the reaction of 1 with CO₂ because of the formation of strong Ti–O bonds.

Reaction of COS with Ti₂Pn^†₂

Support for the proposed intermediate comes from the reaction sequence observed when COS is added to 1. The first formed species, 9, shows a CO stretch at 1498 cm^–1, which is in good agreement with the stretching vibration at 1487 cm^–1 for model system IX (Figure 16) calculated by DFT methods.⁵⁶

Figure 16.

Optimized geometries of IX, X, and XI.

The free energy of the reaction of I with COS is intermediate between that of CS₂ (−52 kcal mol^–1) and CO₂ (−37 kcal mol^–1) (eq 6).

6

The formation of X, from IX, is calculated to be slightly favorable (eq 7) with an activation energy of 32, 21 kcal mol^–1. The value found for the equivalent reaction to Int1 is 22, 13 kcal mol^–1. Loss of CO from X is unfavorable (eq 8) and more so than from Int1 (5, 11 kcal mol^–1).

7

8

The mono(μ-S) mono(CO) intermediate 10 was too unstable to be isolated in the solid state but was characterized by spectroscopic methods in solution. In situ IR spectroscopy showed a broad band at 2011 cm^–1, in the expected region for a terminal CO ligand, that shifted to 1966 cm^–1 in the isotopically labeled [¹³C]11. The calculated structure X (Figure 16) has a CO vibration at 1937 cm^–1. The discrepancy here is rather large. The ¹H NMR spectrum of 10 consists of four doublets in the aromatic region, indicative of chemically equivalent Pn^† rings on the NMR time scale. This can be attributed to a fluxional process involving intramolecular exchange of CO between the two Ti centers that results in an averaged NMR structure of C₂ symmetry.

In the presence of excess of COS, green solutions of 10 decompose within hours at room temperature to give a brown solution (Scheme 4). ¹H NMR spectroscopy of the brown solution showed near-quantitative conversion to the bis(μ-S)-bridged dimer [(η⁸-Pn^†)Ti(μ-S)]₂ (12) as the sole Ti-containing product, in contrast to the decomposition of CO₂ adduct 4, which also gives dicarbonyl complex 3. The related oxo species, Ti₂Pn₂O(CO) (Int1), was implicated in the DFT-modeled pathway of the CO₂ reaction but was not observed experimentally.

The free energy of formation of VIII is calculated as −53 kcal mol^–1, which is significantly greater than that of IV. The energy to form equivalent decomposition products is unfavorable (eq 9), and we were unable to identify computationally an intermediate equivalent to Int1 where a C–S bond had been ruptured and bound CS formed; instead, the structure reverted to an alternative η² mode of CS binding.

9

Conclusions

Complex 1 shows unique reactivity among bis(pentalene)dimetal double-sandwich complexes, including multiple adduct formation with carbon monoxide and reductive activation of carbon dioxide. Mechanistic studies show that 1 binds CO₂ at low temperatures, but this adduct 4 is thermally unstable and upon warming to room temperature, reaction of the coordinated CO₂ molecule occurs. We have gained significant insight into the mechanism for this reaction using synthetic, spectroscopic, and computational methods to identify key intermediates. Thermally driven cleavage of the bound CO₂ occurs to give oxo and carbonyl fragments, and exposure of 4 to dynamic vacuum removes CO, allowing the isolation of a diamagnetic mono(oxo) complex, 5. The latter shows a strained three-membered “dimetallaepoxide” motif in its molecular structure and is unstable at room temperature with respect to the paramagnetic, bridging mono(oxo) complex 6, in which the Ti^III centers are better stabilized, electronically and sterically, with η⁸-bound pentalene ligands.

The reactivity of 1 with the heteroallenes CS₂ and COS provides useful insight in comparison with that observed for CO₂. The thermally stable CS₂ adduct 8 shows symmetrical binding to the Ti₂ unit, providing supporting evidence for the doubly bridging CO₂ ligand in 4. 1 displays greater, and different, reactivity toward COS, but some parallels can be made with the CO₂ reaction. The asymmetric COS molecule undergoes scission at low temperatures, forming the mono(μ-S) mono(CO) species 10, which has sufficient stability for spectroscopic characterization. A mono(μ-O) mono(CO) counterpart was implicated in the CO₂ reaction pathway but was not observed experimentally.

Experimental Section

General Procedures

All manipulations were carried out using standard Schlenk techniques under Ar or in an MBraun glovebox under N₂ or Ar. All glassware was dried at 160 °C overnight prior to use. Solvents were purified by predrying over sodium wire and then distilled over Na (toluene), K (methylcyclohexane), or Na–K alloy (Et₂O and pentane) under a N₂ atmosphere. Dried solvents were collected, degassed, and stored over argon in K-mirrored ampules. Deuterated solvents were degassed by three freeze–pump–thaw cycles, dried by refluxing over K for 3 days, vacuum-distilled into ampules, and stored under N₂. The gases used were of very high purity—CO (99.999%) and CO₂ (99.99%) were supplied by Union Carbide, COS (>97.5%) was supplied by Aldrich, N₂O (purity >99.998%) was supplied by Fluka—and all gases were added via Toepler line. Isotopically enriched gases, ¹³CO (99.7%) and ¹³CO₂ (99%), were supplied by Euriso-top and Cambridge Isotopes, respectively, and added via Toepler pump. Triphenylphosphine sulfide (98%) was purchased from Aldrich and used as received. The compounds Ti₂Pn^†₂ (1),⁵⁴ Ti₂Pn^†₂O₂ (2), and Ti₂Pn^†₂(CO)₂ (3)⁵⁵ were prepared according to published procedures. NMR spectra were recorded on a Varian VNMRS 400 spectrometer (¹H, 399.5 MHz; ¹³C{¹H}, 100.25 MHz; ²⁹Si{¹H}, 79.4 MHz). The ¹H and ¹³C spectra were referenced internally to the residual protic solvent (¹H) or the signals of the solvent (¹³C). ²⁹Si{¹H} NMR spectra were referenced externally relative to SiMe₄. IR spectra were recorded between NaCl plates using a PerkinElmer Spectrum One FTIR instrument or a Mettler-Toledo ReactIR system featuring an IR probe inside a gas-tight cell attached to a Toepler pump. Mass spectra were recorded using a VG Autospec Fisons instrument (EI at 70 eV). Elemental analyses for were carried out at the Elemental Analysis Service, London Metropolitan University.

Syntheses

Ti₂(μ:η⁵,η⁵-Pn^†)₂CO₂ (4)

A J. Young NMR tube was charged with 1 (16 mg, 0.017 mmol) and methylcyclohexane-d₁₄ (0.5 mL), and the solution was cooled to −78 °C. The headspace was evacuated, and CO₂ (0.017 mmol) was admitted. The tube was briefly shaken, and a color change from deep red to dark green was observed. NMR spectra were immediately measured, with the probe precooled to −30 °C. Alternatively, using ¹³CO₂ in the method above yielded the labeled product [¹³C]4. NMR yield: quantitative with respect to 1. ¹H NMR (methylcyclohexane-d₁₄, 399.5 MHz, 243 K): δ_H 5.98 (2H, d, ³J_HH = 3.6 Hz, Pn H), 5.83 (2H, d, ³J_HH = 2.9 Hz, Pn H), 5.66 (2H, d, ³J_HH = 2.8 Hz, Pn H), 5.37 (2H, d, ³J_HH = 3.6 Hz, Pn H), 0.69 (6H, dt, ³J_HH = 7.4, 14.9 Hz, ⁱPr CH), 0.57 (6H, dt, ³J_HH = 7.4, 14.7 Hz, ⁱPr CH), 0.34 (18H, d, ³J_HH = 7.3 Hz, ⁱPr CH₃), 0.31 (18H, d, ³J_HH = 7.5 Hz, ⁱPr CH₃), 0.20 (18H, d, ³J_HH = 7.4 Hz, ⁱPr CH₃), 0.16 (18H, d, ³J_HH = 7.4 Hz, ⁱPr CH₃). ¹³C{¹H} NMR (methylcyclohexane-d₁₄, 100.5 MHz, 243 K, selected data): δ_C 219 (s, CO₂). ²⁹Si{¹H} NMR (methylcyclohexane-d₁₄, 79.4 MHz, 243 K): δ_Si 2.12, 1.88. IR (methylcyclohexane, −65 °C): 4 1678 (ν_asym(OCO)), 1236 (ν_sym(OCO)) cm^–1; [¹³C]4 1637 (ν_asym(O¹³CO)), 1217 (ν_sym(O¹³CO)) cm^–1. EI-MS and microanalysis data were not obtained for 4 because of its susceptibility to further reaction above 0 °C.

Ti₂(μ:η⁵,η⁵-Pn^†)₂(μ-O) (5)

Method A: A J. Young NMR tube was charged with 1 (17 mg, 0.018 mmol) and C₆D₆ (0.6 mL), and the solution was frozen at −78 °C. The headspace was evacuated, and N₂O (0.016 mmol) was admitted. The tube was warmed to room temperature, and a color change to bright red was observed as the solution thawed. NMR yield: 68% with respect to 1. Method B: An ampule charged with 1 (78.5 mg, 0.0848 mmol) and methylcyclohexane (5 mL) was cooled to −78 °C, and the headspace was evacuated. CO₂ gas (1 atm) was admitted, resulting in a color change from deep red to dark green. The ampule was placed under reduced pressure and then removed from the −78 °C bath. The solvent was slowly removed in vacuo as the reaction mixture warmed to room temperature, giving a crude red residue. Recrystallization from toluene (1 mL) at −35 °C afforded 5 as a red microcrystalline solid. Total yield: 66.3 mg (83% with respect to 1). Subsequent recrystallization from a concentrated toluene/pentane solution at −35 °C afforded X-ray-quality crystals. ¹H NMR (C₆D₆, 399.5 MHz, 303 K, selected data): δ_H 7.86 (2H, d, ³J_HH= 3.3 Hz, Pn H), 6.78 (2H, d, ³J_HH = 3.2 Hz, Pn H), 6.45 (2H, d, ³J_HH= 3.2 Hz, Pn H), 6.39 (2H, d, ³J_HH = 3.2 Hz, Pn H), 1.55 (6H, m, ⁱPr CH), 1.26 (6H, m, ⁱPr CH), 1.13 (18H, d, ³J_HH= 7.2 Hz, ⁱPr CH₃), 1.10 (18H, d, ³J_HH= 7.4 Hz, ⁱPr CH₃), 1.01 (18H, d, ³J_HH= 7.4 Hz, ⁱPr CH₃), 0.96 (18H, d, ³J_HH= 7.4 Hz, ⁱPr CH₃). ¹³C{¹H} NMR (C₆D₆, 100.5 MHz, 303 K, selected data): δ_C 134.2 (Pn C), 133.4 (Pn C), 132.6 (Pn C), 114.9 (Pn C), 111.3 (Pn C), 108.1 (Pn C), 102.2 (Pn C), 99.47 (Pn C), 19.94 (ⁱPr CH), 19.89 (ⁱPr CH), 19.39 (ⁱPr CH₃), 19.35 (ⁱPr CH₃), 14.11 (ⁱPr CH), 12.97 (ⁱPr CH). ²⁹Si{¹H} NMR (C₆D₆, 303 K): δ_Si 2.39, 1.01. EI-MS: m/z 941 (100%) [M]⁺. Anal. Found (Calcd for C₅₂H₉₂OSi₄Ti₂): C, 66.21 (66.35); H, 9.70 (9.85) %.

[(η⁸-Pn^†)Ti]₂(μ-O) (6)

Storage of a C₆D₆ solution of 5 (6.6 μmol) overnight at room temperature afforded green crystals of 6, which were isolated by decantation, washed with cold pentane (1 mL, −35 °C) and dried in vacuo. Total yield: 2.0 mg (32% with respect to 5). ¹H NMR (toluene-d₈, 399.5 MHz, 303 K): δ_H 4.09 (br, Δν_1/2 = 138 Hz, Pn H), 4.09 (br, Δν_1/2 = 138 Hz, Pn H), 1.77 (br, Δν_1/2 = 110 Hz, ⁱPr CH), 1.13 (br, Δν_1/2 = 106 Hz, ⁱPr CH₃). ¹³C and ²⁹Si NMR resonances were not observed because of the paramagnetic nature of 6. EI-MS: m/z 941 (100%) [M]⁺. Anal. Found (Calcd for C₅₂H₉₂OSi₄Ti₂): C, 66.17 (66.35); H, 9.75 (9.85) %. Magnetic susceptibility (Evans method, THF-d₈, 303 K): μ_eff = 2.45μ_B per molecule = 1.73μ_B per Ti.

Ti₂(μ:η⁵,η⁵-Pn^†)₂(μ:η²-CO) (7)

An ampule charged with 1 (50 mg, 0.054 mmol) and toluene (0.5 mL) was cooled to −78 °C, and the headspace was evacuated. CO gas (0.9 molar equiv, 0.050 mmol) was admitted slowly to the mixture with vigorous stirring, resulting in a color change from deep red to purple. After the mixture was warmed to room temperature, the volatiles were removed in vacuo to afford a crude purple solid that was recrystallized from pentane (1 mL) at −35 °C. Total yield: 32 mg (62% with respect to 1). Alternatively, exposing a methylcyclohexane-d₁₄ solution of 1 to 0.9 equiv of ¹³CO via Toepler pump yielded the labeled product [¹³C]7. ¹H NMR (methylcyclohexane-d₁₄, 399.5 MHz, 303 K): δ_H 7.69 (1H, apparent t, ³J_HH = 3.2 Hz, Pn H), 7.10 (1H, d, ³J_HH = 3.2 Hz, Pn H), 6.62 (1H, d, ³J_HH = 3.5 Hz, Pn H), 6.34 (1H, d, ³J_HH = 3.5 Hz, Pn H), 6.30 (2H, d, ³J_HH = 3.2 Hz, Pn H), 6.22 (2H, d, ³J_HH = 3.6 Hz, Pn H), 5.74 (1H, d, ³J_HH = 3.5 Hz, Pn H), 5.62 (1H, d, ³J_HH = 3.2 Hz, Pn H), 1.62–1.25 (12H, overlapping m, ⁱPr CH), 1.25–0.83 (72H, overlapping m, ⁱPr CH₃). ¹³C{¹H} NMR (methylcyclohexane-d₁₄, 100.5 MHz, 303 K): δ_C 295.1 (s, CO), 136.8 (Pn C), 132.5 (Pn C), 132.4 (Pn C), 129.1 (Pn C), 127.7 (Pn C), 125.1 (Pn C), 124.9 (Pn C), 115.1 (Pn C), 108.2 (Pn C), 104.7 (Pn C), 103.2 (Pn C), 101.5 (Pn C), 99.75 (Pn C), 95.08 (Pn C), 93.16 (Pn C), 89.87 (Pn C), 20.94 (ⁱPr CH₃), 20.84 (ⁱPr CH₃), 20.79 (ⁱPr CH₃), 20.64 (ⁱPr CH₃), 20.62 (ⁱPr CH₃), 20.37 (ⁱPr CH₃), 20.36 (ⁱPr CH₃), 20.13 (ⁱPr CH₃), 15.14 (ⁱPr CH), 15.12 (ⁱPr CH), 14.45 (ⁱPr CH), 14.22 (ⁱPr CH). ²⁹Si{¹H} NMR (methylcyclohexane-d₁₄, 79.4 MHz, 303 K): δ_Si 3.40 (s), 2.88 (d, J_SiC = 1.2 Hz), 2.40 (s), 2.22 (s). EI-MS: m/z 924 (100%) [M – CO]⁺, 882 (10%) [M – CO – ⁱPr]⁺. Anal. Found (Calcd for C₅₃H₉₂OSi₄Ti₂): C, 66.78 (66.77); H, 9.81 (9.73) %. IR (NaCl): 7 1655 (br, ν(CO)); [¹³C]7 1616 (br, ν(¹³CO)) cm^–1.

Ti₂(μ:η⁵,η⁵-Pn^†)₂(μ:η²,η²-CS₂) (8)

To a solution of 1 (58 mg, 0.058 mmol) in pentane (2 mL) at −35 °C was added CS₂ (14 μL, 0.23 mmol) via microsyringe. The resultant dark-green mixture was allowed to warm to room temperature and stir for 3 h. Filtration followed by cooling to −35 °C afforded red microcrystals that were isolated by decantation and dried in vacuo. Total yield: 36 mg (62% with respect to 1). Recrystallization from a saturated Et₂O solution at −35 °C afforded dark-red crystals suitable for X-ray diffraction studies. Alternatively, recrystallization from a saturated SiMe₄ solution at −35 °C afforded red crystals with different unit cell parameters (see Table S1 in the Supporting Information). ¹H NMR (toluene-d₈, 399.5 MHz, 303 K): δ_H 7.30 (2H, ³J_HH = 3.0 Hz, Pn H), 6.35 (2H, ³J_HH = 3.0 Hz, Pn H), 6.25 (2H, ³J_HH = 3.5 Hz, Pn H), 5.85 (2H, ³J_HH = 3.5 Hz, Pn H), 1.48 (6H, m, ⁱPr CH), 1.42 (6H, m, ⁱPr CH), 1.15 (18H, d, ³J_HH = 7.4 Hz, ⁱPr CH₃), 1.13 (18H, d, ³J_HH = 7.5 Hz, ⁱPr CH₃), 1.05 (18H, d, ³J_HH = 7.4 Hz, ⁱPr CH₃), 1.01 (18H, d, ³J_HH = 7.5 Hz, ⁱPr CH₃). ¹³C{¹H} NMR (toluene-d₈, 100.5 MHz, 303 K): δ_C 355.5 (CS₂), 133.4 (Pn C), 132.4 (Pn C), 127.1 (Pn C), 107.6 (Pn C), 105.5 (Pn C), 95.1 (Pn C), 20.37 (ⁱPr CH₃), 20.21 (ⁱPr CH₃), 19.82 (ⁱPr CH₃), 14.73 (ⁱPr CH), 12.93 (ⁱPr CH). ²⁹Si{¹H} NMR (toluene-d₈, 79.4 MHz, 303 K): δ_Si 7.65, 7.39. EI-MS: m/z 1002 (65%) [M]⁺, 960 (30%) [M – ⁱPr]⁺. Anal. Found (Calcd for C₅₃H₉₂S₂Si₄Ti₂): C, 63.49 (63.56); H, 9.26 (9.26) %. IR (methylcyclohexane, 26 °C): 1104 (ν_asym(SCS)) cm^–1.

Ti₂(μ:η⁵,η⁵-Pn^†)₂COS (9)

A J. Young NMR tube was charged with 1 (20 mg, 0.022 mmol) and methylcyclohexane-d₁₄ (0.6 mL), and the solution was cooled to −78 °C. The headspace was evacuated, and COS (0.048 mmol) was admitted. The tube was briefly shaken, and a color change from deep red to green was observed. NMR spectra were immediately measured, with the probe precooled to −30 °C. NMR yield: 45% with respect to 1. ¹H NMR (methylcyclohexane-d₁₄, 399.5 MHz, 243 K, selected data): δ_H 7.00 (1H, d, ³J_HH = 3.0 Hz, Pn H), 6.96 (1H, d, ³J_HH = 3.0 Hz, Pn H), 6.72 (1H, d, ³J_HH = 3.7 Hz, Pn H), 6.45 (1H, d, ³J_HH = 3.0 Hz, Pn H), 6.43 (1H, d, ³J_HH = 3.6 Hz, Pn H), 6.31 (1H, d, ³J_HH = 3.7 Hz, Pn H), 6.09 (1H, d, ³J_HH = 3.7 Hz, Pn H), 1.79–1.44 (12H, overlapping m, ⁱPr CH), 1.34–1.02 (72H, overlapping m, ⁱPr CH₃). One Pn H signal was not resolved because of coincidental overlap with signals of 12. ¹³C{¹H} NMR (methylcyclohexane-d₁₄, 100.5 MHz, 243 K, selected data): δ_C 283 (s, COS). ²⁹Si{¹H} NMR (methylcyclohexane-d₁₄, 79.4 MHz, 243 K, selected data): δ_Si 2.48, 2.07, 1.73, 1.08. IR (NaCl): 1493 (ν_asym(SCO)) cm^–1. IR (methylcyclohexane, −65 °C): 1498 (ν_asym(SCO)) cm^–1. EI-MS and microanalysis data were not obtained for 9 because of its susceptibility to further reaction above −40 °C.

Ti₂(μ:η⁵,η⁵-Pn^†)₂S(CO) (10)

Method A: A J. Young NMR tube charged with 1 (16 mg, 0.017 mmol) and toluene-d₈ (0.65 mL) at −78 °C was degassed, and COS gas (0.015 mmol) was slowly admitted. The reaction mixture was briefly shaken, resulting in a color change from deep red to green, and the mixture was allowed to warm to room temperature over a period of 2 h. NMR yield: 47% with respect to 1. Method B: An ampule charged with 11 (0.028 mmol) and toluene-d₈ (0.7 mL) was cooled to −78 °C, and the headspace was evacuated. ¹³CO gas (0.025 mmol) was slowly admitted to the stirred mixture, resulting in a color change from red to green. The solution was warmed to room temperature and then decanted into a J. Young tube, and [¹³C]10 was identified by its NMR and IR spectra. NMR yield: 75% with respect to 1. Alternatively, exposing a solution of 11 to 0.9 molar equiv of CO via Toepler pump yielded primarily the unlabeled product 10. ¹H NMR (toluene-d₈, 399.5 MHz, 303 K, selected data): δ_H 7.81 (2H, d, ³J_HH = 2.9 Hz, Pn H), 6.09 (br, Δν_1/2 = 13 Hz, Pn H), 5.75 (2H, d, ³J_HH = 2.9 Hz, Pn H), 5.38 (2H, d, ³J_HH = 2.7 Hz, Pn H), 1.62–1.24 (12H, overlapping m, ⁱPr CH), 1.22–0.96 (72H, overlapping m, ⁱPr CH₃). ¹³C{¹H} NMR (toluene-d₈, 100.5 MHz, 303 K, selected data): δ_C 247 (br, Δν_1/2 = 110 Hz, CO). ²⁹Si{¹H} NMR (toluene-d₈, 79.4 MHz, 303 K, selected data): δ_Si 2.56, 1.46. IR (methylcyclohexane, −65 °C): 10 2011 (br, ν(CO)); [¹³C]10 1966 (br, ν(¹³CO)) cm^–1. EI-MS and microanalysis data were not obtained for 10 because of its susceptibility to further reaction above −40 °C.

Ti₂(μ:η⁵,η⁵-Pn^†)₂(μ-S) (11)

A solution of Ph₃PS (56 mg, 0.19 mmol) in toluene (3 mL) was added dropwise to a stirred solution of 1 (177 mg, 0.191 mmol) in pentane (3 mL) at −35 °C. The resultant red mixture was allowed to warm to room temperature, stirred for 30 min, and then stripped to dryness. A red residue was obtained, which was placed under high vacuum (ca. 10^–6 mbar) and heated at 75 °C for 1 h and then 100 °C for 2 h to ensure complete sublimation of a white crystalline solid, presumably Ph₃P. The residue was extracted with pentane (3 × 2 mL), filtered, and allowed to concentrate to ca. 3 mL. Cooling of this solution to −35 °C afforded red crystals of 11 suitable for X-ray diffraction studies. Total yield: 95 mg (52% with respect to 1). ¹H NMR (C₆D₆, 399.5 MHz, 303 K): δ_H 9.25 (2H, d, ³J_HH= 3.4 Hz, Pn H), 7.06 (2H, d, ³J_HH = 3.2 Hz, Pn H), 6.10 (2H, d, ³J_HH= 3.1 Hz, Pn H), 5.96 (2H, d, ³J_HH = 3.4 Hz, Pn H), 2.01 (6H, m, ⁱPr CH), 1.35 (6H, m, ⁱPr CH), 1.17 (18H, d, ³J_HH= 7.5 Hz, ⁱPr CH₃), 1.13 (18H, d, ³J_HH= 7.3 Hz, ⁱPr CH₃), 0.98 (18H, d, ³J_HH= 7.5 Hz, ⁱPr CH₃), 0.60 (18H, d, ³J_HH= 7.5 Hz, ⁱPr CH₃). ¹³C{¹H} NMR (C₆D₆, 100.5 MHz, 303 K): δ_C 141.7 (Pn C), 132.5 (Pn C), 130.0 (Pn C), 123.0 (Pn C), 118.6 (Pn C), 115.3 (Pn C), 113.2 (Pn C), 98.53 (Pn C), 19.50 (ⁱPr CH₃), 19.41 (ⁱPr CH₃), 19.14 (ⁱPr CH₃), 18.89 (ⁱPr CH₃), 13.72 (ⁱPr CH), 13.15 (ⁱPr CH). ²⁹Si{¹H} NMR (C₆D₆, 79.4 MHz, 303 K): δ_Si 2.20, 1.82. EI-MS: m/z 989 (60%) [M + S]⁺, 956–959 (principal peak 957, 20%) [M]⁺, 877 (100%) [M – Ti – S]⁺. Anal. Found (Calcd for C₅₂H₉₂SSi₄Ti₂): C, 65.33 (65.23); H, 9.61 (9.69) %.

[(η⁸-Pn^†)Ti(μ-S)]₂ (12)

Method A: An ampule containing a solution of 1 (141 mg, 0.152 mmol) in benzene (2 mL), frozen at −35 °C, was degassed and fitted with a balloon containing COS gas (1 atm). The solution was allowed to warm to room temperature and, shortly after thawing, appeared green in color. After 30 min of stirring at room temperature, a color change to red-brown was observed. The reaction mixture was frozen and lyophilized, and a spectroscopically pure red-brown solid was isolated. Total yield: 145 mg (96% with respect to 1). Method B: To an ampule containing 1 (100 mg, 0.108 mmol) and Ph₃PS (64 mg, 0.216 mmol) was added toluene (2 mL). After 4 h of stirring at room temperature, the red-brown mixture was stripped to dryness, extracted with Et₂O (2 × 5 mL), and filtered, and the solvent was removed in vacuo. A brown solid was obtained, which was placed under high vacuum (ca. 10^–6 mbar) and heated at 60 °C for 2.5 h, 75 °C for 1 h, and finally 100 °C for 30 min to ensure complete sublimation of a white crystalline solid, presumably Ph₃P. The residue was extracted with Et₂O (2 × 5 mL), filtered, and allowed to concentrate to ca. 4 mL. Cooling of this solution to −35 °C afforded red crystals suitable for X-ray diffraction studies. Total yield: 86 mg (80% with respect to 1). ¹H NMR (C₆D₆, 399.5 MHz, 303 K): δ_H 6.59 (4H, d, ³J_HH = 3.3 Hz, Pn H), 5.92 (4H, d, ³J_HH = 3.2 Hz, Pn H), 1.31 (12H, m, ⁱPr CH), 1.19 (36H, d, ³J_HH = 7.3 Hz, ⁱPr CH₃), 1.14 (36H, d, ³J_HH = 7.3 Hz, ⁱPr CH₃). ¹³C{¹H} NMR (C₆D₆, 100.5 MHz, 303 K): δ_C 148.9 (Pn C), 130.0 (Pn C), 115.2 (Pn C), 113.9 (Pn C), 19.66 (ⁱPr CH₃), 19.50 (ⁱPr CH₃), 12.31 (ⁱPr CH). ²⁹Si{¹H} NMR (C₆D₆, 79.4 MHz, 303 K): δ_Si −0.15. EI-MS: m/z 989 (100%) [M]⁺, 946 (50%) [M – ⁱPr]⁺. Anal. Found (Calcd for C₅₂H₉₂S₂Si₄Ti₂): C, 63.20 (63.12); H, 9.44 (9.37) %.

Kinetic Studies

Decomposition rates of 4 at seven temperatures between 9 to 40 °C were measured by the same method: A J. Young NMR tube was charged with 0.50 mL of a stock solution containing 1 (0.0295 M, 0.015 mmol) and ferrocene (0.0935 M) in methylcyclohexane-d₁₄, and the solution was cooled to −78 °C. The headspace was evacuated, and ¹³CO₂ (0.015 mmol) was admitted. The tube was briefly shaken, and a color change from deep red to dark green was observed. After 30 min the tube was removed from the −78 °C bath, placed in a pre-equilibrated VT-NMR probe, and shimmed as thermal equilibrium was attained (10 min). To avoid any complications due to different relaxation times for the resonances, spectra were recorded using one 90° pulse. ¹H spectra were acquired periodically (33.6 s repetition time), and the time of each spectrum was recorded along with the data by the internal clock on the spectrometer. A baseline correction was applied, and the concentrations of 4 were determined by integration over the same region of each spectrum relative to the ferrocene resonance. The temperature in the probe was calibrated by measuring the peak separation of a sample of neat methanol.⁸³ Concentration versus time data were analyzed by fitting the data for 4 to a first-order exponential decay. Plots of ln([4]) versus time and the derived rate constants are included in the Supporting Information. The activation parameters were calculated from a linear fit of the Eyring plot, ln(k_obs/T) versus 1/T. The error in the determination of the activation parameters was calculated according to Girolami and co-workers.⁸⁴

Crystallographic Details

Single-crystal XRD data for 7, 8 (SiMe₄), 11, and 12 were collected by the U.K. National Crystallography Service (NCS)⁸⁵ at the University of Southampton on a Rigaku FR-E+ Ultra High Flux diffractometer (λ_Mo Kα) equipped with VariMax VHF optics and a Saturn 724+ CCD area detector. The data were collected at 100 or 150 K using an Oxford Cryosystems Cobra low-temperature device. Data collected by the NCS were processed using CrystalClear-SM Expert 3.1 b18,⁸⁶ and unit cell parameters were refined against all data. Single-crystal XRD data for 5 and 8 (Et₂O) were collected at the University of Sussex on a Enraf-Nonius CAD4 diffractometer with graphite-monochromatized (λ_Mo Kα) radiation or an Agilent Technologies Xcalibur Gemini Ultra diffractometer (λ_Mo Kα or λ_Cu Kα source) equipped with an Eos CCD area detector. The data were collected at 173 K using an Oxford Cryosystems Cobra low-temperature device. Data were processed using KappaCCD software or CrysAlisPro (version 1.171.36.32),⁸⁷ and unit cell parameters were refined against all data. An empirical absorption correction was carried out using the Multi-Scan program.⁸⁸ All of the structures were solved using SHELXL-2013,⁸⁹ DIRDIF-2008,⁹⁰ or SUPERFLIP⁹¹ and refined on F_o² by full-matrix least-squares refinements using SHELXL-2013.⁸⁹ Solutions and refinements were performed using the OLEX2⁹² or WinGX⁹³ packages and software packages within. All non-hydrogen atoms were refined with anisotropic displacement parameters. All hydrogen atoms were refined using a riding model. Disordered solvent molecules in 8 (Et₂O) and 12 could not be modeled properly; therefore, this disorder was treated using the SQUEEZE⁹⁴ routine in PLATON.⁹⁵ Disordered solvent molecules in 11 were treated using the solvent mask routine in OLEX2,⁹² which resulted in improved refinement indices relative to SQUEEZE, so these data have been included.

Computational Details

DFT calculations were carried out using two computational methods. One employed the Amsterdam Density Functional package (version ADF2012.01).⁹⁶ The Slater-type orbital (STO) basis sets were of triple-ζ quality augmented with one polarization function (ADF basis TZP). Core electrons were frozen (C 1s; Ti 2p) in the model of the electronic configuration for each atom. The local density approximation (LDA) of Vosko, Wilk, and Nusair (VWN)⁹⁷ was used together with the exchange–correlation corrections of Becke and Perdew (BP86).^98,99 The other used Gaussian 09, revision A.02,¹⁰⁰ with the B3LYP functional and SDD basis set. In both sets of calculations, tight optimization conditions were used and frequency calculations were used to confirm stationary points.

Supporting Information Available

Additional NMR spectroscopic, X-ray crystallographic, and kinetic data; a text file of all computed molecule Cartesian coordinates in .xyz format for convenient visualization; crystallographic data in CIF format. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.organomet.5b00315.

The authors declare no competing financial interest.