Evidence for carbon–hydrogen–titanium interactions: synthesis and crystal structures of the agostic alkyls [TiCl3(Me2PCH2CH2PMe2)R](R = Et or Me)

The compounds [TiCl3(dmpe)R](R = Me, CH2D, or Et; dmpe = Me2PCH2CH2PMe2) have been prepared. An X-ray crystal structure determination of [TiCl3(dmpe)Et] shows that the Ti–C–C angle of the Ti–Et moiety is 86.3(6)°. Variable-temperature n.m.r. studies show the molecule to be fluxional at room temperature. At –90 °C the n.m.r. spectra correspond to the structure found in the crystal and J(C–H) for the ethyl CH2 hydrogens is 150.2 Hz. It is concluded that the ethyl group is bonded by a [graphic omitted] system. The crystal structure of [TiCl3(dmpe)Me] has been determined from X-ray and neutron diffraction data. The data show the methyl group to be distorted such that it is tilted with respect to the Ti–C vector and one hydrogen atom closely approaches the Ti centre giving an angle for Ti–C–H of 93.5(2)°(neutron data). Variabletemperature n.m.r. studies show that [TiCl3(dmpe)R](R = CH3 or CH2D) are fluxional but provide no evidence for differentiation of the C–H moiety of the methyl group.

The dmpe ligand in (2) is disordered at C(5) and C (6). A similar disorder in (3) is not obvious but the short C(5)-C (6) bond [ 1.485 (9) A] in (3X) contrasts with the more normal value of 1.535( 1 ) A in (3N) and is indicative of some disorder at room temperature. Comparison of the dmpe ligands in Figure 2 shows how the disorder in (2) and to a lesser extent in (3X) increases the apparent symmetry of the dmpe ligand. The dimensions of the dmpe ligands are otherwise unexceptional. In both (2) and (3) the Ti-P( 1 ) bond trans to CI(1) is not significantly different in length from Ti-P(2) frans to the alkyl group, but the Ti-P bonds found here (mean 2.579 A) are rather shorter than those observed in the 16- 13 The equatorial Ti-CI( 1) bond length in (3X) and (3N) is the same or very slightly shorter than the axial bond lengths, but in (2) it is exceptionally long, 2.408(3) A.
The most interesting aspects of these structures concern the alkyl groups. In contrast to the 16-and 18-electron metal ethyls where M-C-C angles in the range 108-126" have been the Ti-C(1)-C(2) angle in (2) is 86.3(6)O and indicates clearly that there is a severe distortion with the result that the methyl group C(2) is drawn towards the titanium centre. There is no evidence from the structure that the unusual disposition of the ethyl group arises from intermolecular interactions. The ethyl hydrogen atoms are clearly resolved. The H(23) atom of the methyl group is located 2.22 ( 10) A from the metal. There are other examples where hydrogen atoms of formally saturated C-H systems closely approach a metal centre." H(23) is 0.17 A out of the P(l),P(2),C(l),Cl(l) plane. The ethyl C(l)-C(2) bond length of 1.467 A is rather shorter than the average C-C bond length, 1.541 A, in metal ethyls.' 1.14 The Ti-C( 1 ) distance C2.15 l(9) A] is in the expected range for Ti-C bond lengths. The molecule represents a step along the reaction co-ordinate towards p-elimination to form the ethylene-hydride cation and a chloride anion, with a shortening of C(l)-C(2) towards the ethylenic C-C bond length, a substantial Ti . H(23) interaction, and a lengthening of Ti-C1( 1) prior to expulsion as the anion.
The Ti-C( 1) bond length to the methyl group in (  It appears that in the X-ray study of (3X) the position of H( 11) is subject to appreciable error, well beyond the leastsquares derived e.s.d. [angle Ti<( 1)-H( 1 1) differs by more than 70 between (3X) and (3N)l. Ibers16 has noted the difficulties sometimes experienced in properly locating hydrogen atoms in X-ray work; these difficulties stem from the fact that peaks due to hydrogen atoms are typically near the noise level of the electron-density maps. However the discrepancy may be due to an asymmetric distribution of electron density about the hydrogen nucleus so that the centroid of the electron density does not coincide with the nuclear position. Thus, neutron data may be required to accurately and unambiguously locate hydrogen nuclei.

N.M.R.
Studies.-The ' H n.m.r. spectrum of (2) at -90 "C may be assigned in a straightforward manner. N.m.r. data and assignments are given in Table 2. The 31P n.m.r. spectrum at -70 "C indicates the two phosphorus nuclei to have different chemical shifts. The I3C n.m.r. spectrum at -60 "C shows there to be two pairs of inequivalent PMe, groups and two different CH, groups. These data strongly suggest that the structure of (2) in solution at -90 "C is the same as that derived from the crystal structure determination. Variable-temperature n.m.r. data show that (2) undergoes fluxional processes at room temperature. As the temperature is raised, two dominant changes are seen in the ' H n.m.r. spectrum. The chemical shift of the band assigned to the TiCH, group moves towards that of the band assigned to the methyl of TiEt. At -30 "C, the CH, band also begins to move toward that of the CH, group. At CQ. 0 "C, the two signals superimpose (at 2.65 p.p.m.). As the temperature is raised to 40 "C, the two signals have 'passed through' each other and at higher temperatures move to 2.76 (Ti-CH,) and 2.57 p.p.m. (CH,). At this temperature the Ti-CH, band is a seven-line multiplet, and that for the CH, group is a complex, partly resolved eight-line multiplet. On cooling the original spectrum returns unchanged. Variable-temperature chemical shift data for the Ti-C,H, hydrogens are given in Table 3.
The bands assignable to the PCH, and PCH, groups of (2) also shift with varying temperature. At -90 "C there are two multiplets at 2. 29  View Article Online very broad and are close to coalescence and at 20 "C they begin to sharpen. At 40 "C the signals are well resolved doublets at 2.24 (2PCH,) and 1.66 p.p.m. (4PCH,). Thus the PCH, groups become equivalent.
The ',C n.m.r. at 0 "C shows a quartet of doublets at 13.91 p.p.m. due to the four equivalent PCH, carbon nuclei split successively by their three protons and only one phosphorus nucleus. At 20 "C the ,'P n.m.r. spectrum shows only a single very broad resonance and the phosphorus nuclei have become equivalent. The equivalence of the two CH,PMe, groups presumably arises from rapid intramolecular interchange of the ethyl group from the low-temperature equatorial site to an axial position, oiu a 'turnstile' mechanism.
The value of J(C-H) at -60 "C for the TiCH, group is 150.2 Hz and this lies within the range expected for an sp2-hybridised CH, group rather than sp3. J(C-H) for the CH, group (of TiEt) is 126.8 Hz. The high value of J(C-H) for the CH, group, which is close to that for sp2 C-H, can be correlated with the acute Ti-C-C angle at the CH, group. The value of J(C-H) of 126.8 Hz for the CH, group is consistent with the expected value for J(C-H),, for an agostic ethyl group. 15 n.m.r. spectrum of (2) at 0 "C shows a broad triplet at 85.16 p.p.m. due to the 2-carbon of the Ti-C,H, group coupled to two protons. The J(C-H) value at 147.3 Hz is the same, within experimental error, as at -60 "C. There is also a broad quartet at 5.76 p.p.m. due to the P-carbon of the Ti-C2H, ethyl group. The value of J(C-H) at 0 "C (125.7 Hz) is close to that at -60 "C. These data suggest that the agostic ethyl group is still present at 0°C. Finally, the change of chemical shift with temperature of the Ti-C,H, hydrogens may be ascribed to the changing equilibrium concentrations of isomers with the isomers involved presumably being those where the ethyl group lies in the Tip, plane (equatorial) or normal to it (axial). There may also be an equilibrium between the agostic ethyl structure and an isomer of (2) with a normal Ti-C,H, group.
The 'H n.m.r. spectrum of (3) in CD2CI, at -90 "C shows a 1 : 2: 1 triplet at 2.29 p.p.m. assignable to the TiCH, hydrogens coupled to two apparently equivalent ,'P nuclei. There are two complex unresolved multiplets at 2.19 and 2.03 p.p.m. due to two inequivalent PCH, groups; and two sets ofdoublets at 1.61 and 1.42 p.p.m. due to two inequivalent sets of PCH, groups. The I3C n.m.r. spectrum of (3) in CD,CI, at -90 "C shows a doublet of quartets at 87.52 p.p.m. assignable to the Ti-CH, carbon nucleus coupled to three hydrogens and one phosphorus nucleus. In contrast to the observation of a binomial triplet for the Ti-C,H, group in the 'H n.m.r. spectrum, the lowtemperature ( -70 "C) 31P-('H} n.m.r. spectrum shows the presence of two inequivalent phosphorus nuclei at 28.14 and 26.72 p.p.m. We conclude that the magnitudes of J(P-H) for the Ti-C,H, group are fortuitously coincident and that the structure of (3) at low temperature corresponds to that found in the crystal. On warming from -90 "C the 'H n.m.r. spectrum steadily changes over the temperature range -90 to 35 "C, the triplet resonance for the Ti-CH? group moves from 2.29 to 2.19 p.p.m. The triplet is due to the TiCH, resonance coupled to two apparently equivalent phosphorus nuclei. This triplet is superimposed on a doublet at 2.15 p.p.m. assignable to two equivalent sets of PCH, protons. There is a doublet at 1.59 p.p.m. assignable to four equivalent sets of PCH, protons. Thus the four PCH, groups become chemically equivalent at the higher temperature limit, presumably via a turnstile mechanism involving the Ti-CH, group and the chlorine ligands, as proposed for (2). Rapid inversion of the PCH,CH,P backbone is also presumably taking place at this temperature, although this is not a requirement imposed by the spectrum as long as the methyl group occupies all four remaining octahedral available sites in the course of its dynamic behaviour. The * P-{ 'H) n.m.r.
spectrum at 10 "C shows a single broad resonance at 28.5 p.p.m.

The '
The broad nature of this signal presumably is due to its not having fully sharpened following coalescence. The deuteriomethyl analogue of (3), [TiCl,(dmpe)(CH,D)] (3D) was prepared using Zn(CH,D),. The data for (3D) correspond closely to those of (3). The I3C n.m.r. spectrum provided the value for the 13C-'H coupling constant of the methyl group of 130.5 Hz; this is closely similar to that for the C-methyl group of The compound [TiCl,(dmpe)Me] was treated with a large excess of KH in toluene at room temperature. After 3 d the solution had changed from blood-red to an intense orange colour. Reddish crystals (30%) were obtained and a crystal structure determination showed them to be [TiCl,(dmpe),] (4). During our work on (4) we discovered that the compound had been prepared independently,' ' and so we discontinued our study. (3).

Discussion
The distortions observed in the compounds (2) and (3)  It is quite new to think in terms of alkyl groups in transition metal-alkyl compounds acting as bidentate ligands, which formally donate three electrons to the metal centre. In order to emphasise this relatively recently recognised but probably quite widespread mode of bonding between saturated hydrocarbon groups and transition metal centres, the term agostic* has been introduced to describe the M-H-C(saturated) bond.' Consideration of the bonding in titanium-methyl derivatives akin to (3) has been given by Eisenstein and Jean19 and Morokuma and co-workers. 20 Eisenstein and Jean adopt a description somewhat analogous to that proposed for the distortions in tantalum-alkylidene groups" whereby there is overlap between the Ti-CH, p orbital and a low-lying Ti d orbital. The consequential tilting of the methyl group does not involve direct Ti-H interaction and the local geometry of the CH, fragment is not significantly perturbed from C,, in agreement with the observed geometry. (pz = pyrazolyl)." We prefer to think of the close approach of the C-H bond of ethyl compounds to a transition metal as arising from a M-H-C bonding interaction.
The intermolecular oxidative-addition of P-C-H groups of metal alkyls giving the metal-olefin-hydride compound is a classic reaction of metal alkyls. In the ethyl compound (2) it appears as if the ethyl group models a stage about half-way along the reaction co-ordinate for a p-elimination reaction to form the titanium-ethylene-hydride complex, [TiH(dmpe)(q-C,H,)CI,]. However, this compound would not be expected to be stable since the d o titanium centre cannot formally backdonate electrons to the ethylene ligand, as required in the conventional description of the metal-olefin bond. It might therefore have been expected that these distorted alkyls would only be observed when the metal centre had both an electron count less than 18 and a d o configuration.
However, the agostic M-H-C structure can be preferred to the 18-electron polyene-hydride alternative' even when the metal centre has suitable d" electrons, i.e. even for n > 2, the equilibrium (2) can lie to the right-hand side. The neutron diffraction data show that the distortion of the methyl group may be described as a tilting of a methyl group about the normal to the Ti,C(l),H(l 1) plane. The H-C-H angles are all normal and, more importantly, the three C-H ---distances are the same within experimental error ( Table 1 ) . Neutron diffraction data on other agostic C-H hydrogens show that there is a lengthening of the C-H bond. 15 In all previous cases where both neutron diffraction data and appropriate n.m.r. data, including partial deuteriation, have been obtained on compounds with agostic C-H groups, the presence of the agostic hydrogen is indicated by the n.m.r. data. The lengthening of the C-H(agostic) bond is necessary for the detection of the agostic hydrogen by the Shapley experiment28 since lengthening of the C-H(agostic) bond would cause a significant change in the C-H stretching frequency compared to those of the two normal C-H bonds of the agostic methyl group. Different C-H force constants for C-H groups which undergo fast exchange on the n.m.r. time-scale are required for a difference in the zero-point energy differences between H and D in the agostic bonds relative to terminal bonds. This changes the relative residence times of the agostic and normal hydrogens in the bridge position for the n.m.r. experiment. We conclude that since the three C-H distances of the Ti-CH, group of (3) are the same the C-H bond strengths will be closely similar. The closely similar n.m.r. data for (3) and (3D) can be understood in this way.
The averaged value of J(C-H) for the agostic methyl group of (2) is 126.8 Hz. This is comparable with J(C-H),, found for other methyl groups with agostic hydrogens.' It therefore seems that the C-H(agostic) of (2) will be found to be longer than the other two C-H bonds. This in turn means that the C-H-Ti interaction in (2) would be stronger than for (3). Elementary ring-strain arguments predict that the fourmembered ring in (2) would be preferred to a three-membered ring in (2) involving an agostic a-hydrogen.
The compounds [TiCl,(dmpe)R] have, at first sight, a 12electron count, but the presence of the M-H-C bridge raises this to 14 electrons. (A simple procedure for counting electrons in two-electron, three-centre bridging hydrogen systems has been discussed elsewhere. between the partial migration of the hydrogen in the agostic-ethyl structure and the migration of an alkyl group to co-ordinated olefin is that this description is evocative of the alkyl-migration (Cossee) mechanism for Ziegler-Natta cataly-s~s .~' The fact that the fragment Ti(dmpe)CI, can interact with both an x-hydrogen of (3) and a P-hydrogen of (2) suggests that the stereoelectronic properties of a metal centre which favour pinteractions also favour z-hydrogen interactions. Thus, whilst the correlation between P-agostic-alkyl interactions and lowenergy alkyl-to-ethylene migration provides an elegant basis for support for the Cossee mechanism, the involvement of xhydrogen interactions,' although less compelling, cannot be excluded. toluene were dried by refluxing over potassium metal under dinitrogen. Light petroleum (b.p. 3 0 -4 0 and 40-60 "C) was dried over sodium-potassium alloy. Solvents were distilled immediately prior to use.
Dichloromethane was dried by refluxing over CaH, and stored over molecular sieves (type 4A) for prolonged periods. N.m.r. spectra were obtained on either JNM-PMX, Brucker R32 (90 MHz), and 300 MHz or 400 MHz Brucker instruments.
Spectra were calibrated using solvent peaks as internal standards. All chemical shifts are given in p.p.m. relative to G(SiMe,) = 0 p.p.m. Coupling constants (J) are in Hz.

Triehloro(ethyl)titanium.-
The procedure previously described' was employed with small modifications. Tetraethyl-lead (7.13 g, 0.022 mol) in a Schlenk vessel (250 cm3) equipped with a magnetic stirrer was attached to a vacuum manifold, cooled to -196 "C and evacuated. Light petroleum (50 cm', b.p. 40-60°C) was then distilled in vacuo into this reaction vessel, followed by TiCl, (8.39 g, 0.044 mol). The cooled mixture was evacuated for a few minutes and then allowed to warm gradually to room temperature (r.t.). On melting, contact between the reagents gave rise to a deep red-brown colouration and a thick brown precipitate. The mixture was stirred rapidly for 1 h after it had reached r.t. It was then filtered to give a deep red filtrate which was cooled to -78 "C. After 1 h a mass of purple-black crystals precipitated. The supernatant solution was discarded by decantation and the residual solid was redissolved in light petroleum (1 5 cm3, 40-60 "C). The solution was cooled to -78 "C giving purple crystals which were collected and stored at -80 "C.
Tr iehloro( me thy l )titan ium.-T he published p r oced u re8 was used with some modifications. ZnMe, (2.80 g, 0.029 mol) was placed in a Schlenk vessel (150 cm3) equipped with a magnetic stirrer bar. The vessel was attached directly to a vacuum manifold, cooled to -196 "C and evacuated. 2-Methylbutane (25 cm3) was then distilled in vacuo into this vessel followed by TiCl, (10.62 g, 0.056 rnol). On melting, contact between the reagents gave rise to an initial red-brown colouration which dissipated as the mixture was allowed to warm gradually to r.t. to give an orange-yellow solution containing a finely divided white suspension of ZnC1,. The mixture was stirred rapidly for 1 h after reaching room temperature, filtered, and the filtrate was cooled to -78 "C. A solid purple-black crystalline crust formed over 1-2 h which was then isolated by decantation. The solid was recrystallised twice from 2-methylbutane and then dried in vacuo at -78 "C for several hours. Yield 4.69 g, ca. 50%. The compound was stored at -78 "C under argon.  (1.983 g, 1.1 mmol) in light petroleum (25 cm3, b.p. 40-60"C) was stirred rapidly at   in toluene (50 cm3) was treated with an excess of a slurry of potassium hydride (ca. 1.0 g) in light petroleum (50 cm3, b.p. 40--6O"C). The mixture was stirred at r.t. for 3 d and the solution changed from blood-red to an intense orange. It was then filtered and reduced in volume gradually, under reduced pressure, to 2 cm3, and small orange-red crystals formed. Light petroleum (50 cm3, b.p. 40-60 "C) was added to the mixture to complete dissolution. The solution was slowly cooled to -25 "C and after 3 d red crystals precipitated. These were recrystallised from light petroleum (b.p. 40-60 "C). Yield 0.020 g, 30%.
Structure Analysis.-The very air-sensitive dark red prismatic crystals of (2) and (3) were mounted in glass capillary tubes under an argon atmosphere. The crystal data are given in Table  4.
X-Ray measurements were made on an Enraf-Nonius CAD-4 diffractometer at Oxford. For and electron-density methods and refined first isotropically then anisotropically by full-matrix least-squares methods. For (2) inspection of the electron-density maps, bond lengths, and thermal parameters during the refinement process indicated that C(5) and C(6) of the dmpe backbone were disordered. C(6) was successfully resolved on two sites with 50% occupancy, but C(5) was found to be best represented by a single atom site with very large thermal parameters. Within the chelate ring P-C was constrained to 1.83(3) and C-C to 1.54(3) The dmpe hydrogen atoms were not found but those of the ethyl group were located from the difference electron-density syntheses and their positional parameters refined with fixed isotropic thermal parameters. The refinement converged at R 0.056, R' = 0.066 (unit weights), max. (shift/e.s.d.) <0.01. The final difference electron-density map had no major feature except a peak of 0.55 e A-3 very near to C(5). For (3) the refinement proceeded smoothly first with isotropic then anisotropic temperature parameters to R 0.058, when all hydrogen atoms were located from the difference-electron density synthesis. Those bonded to C( I), the most prominent, were refined freely with isotropic temperature factors and the rest had C-H constrained to 0.95(2) %( and a fixed common isotropic temperature factor. The final refinement, which converged at R 0.040, R' 0.038 (unit weights), max. (shift/e.s.d.) < 0.01, included an empirical extinction parameter. The maximum peak height on the final difference- View Article Online ray analysis. In the final cycles of least-squares, which included a secondary extinction correction, all atoms were refined with anisotropic temperature factors. Scattering lengths were obtained from Koester3' and were not refined as variables. The refinement converged at R(F) 0 For each analysis the final atomic parameters are presented in Table 5. All the X-ray analysis calculations were carried out at Oxford on the inhouse VAX 11/750 computer using the Oxford CRYSTALS program suite. 38 The neutron analysis was performed at the Brookhaven and Argonne National Laboratories.