A Radical Tandem Reaction with Homolytic Cleavage of a TiO Bond

A radical merry-go-round. An electron is transferred from a titanocene(III) complex to the substrate and finally back to the catalyst in a novel atom-economical tandem reaction. Complex structures can be readily accessed (see scheme). The unprecedented mechanism involving a homolytic cleavage of a TiO bond is supported by DFT calculations. Over the last years radicals have been used increasingly in multistep syntheses due to the mild reaction conditions, high functional group tolerance, and broad accessibility of interesting structures, often obtained in sequential transformations.1 In this context we wish to report our first results on a novel radical tandem reaction1b featuring an unprecedented formal homolytic substitution reaction at a Ti-O bond for the formation of tetrahydrofurans. The planned sequence is shown in Scheme 1. The initial step is based on the titanocene-mediated opening of 1 described by Nugent and RajanBabu,2 which we have developed into a catalytic reaction,3 to give the radicals cis- and trans-2. General concept of the titanocene-catalyzed tandem reactions with tetrahydrofuran formation. The second and conceptually novel step of our tandem reaction constitutes the attack of the tertiary radical 2 on the Ti-O bond. Mechanistically, this can be viewed as a homolytic substitution of the [Cp2TiCl] radical.4 Our work therefore introduces metal–oxygen bonds as a very useful class of radical traps. As a consequence of this homolytic substitution the redox catalyst [Cp2TiCl] is regenerated. To the best of our knowledge, this concept of a catalytic redox isomerization is unknown in the literature. Because the second cyclization is counterintuitive, a supposedly strong Ti-O bond is cleaved and a C-O bond formed, we analyzed the transformation of model system 4 by density functional theory (DFT) calculations as shown in Scheme 2.5 As expected1c, 6 both the ring opening of 4 (−8.3 kcal mol−1) and the 5-exo cyclization of 5 (−12.2 kcal mol−1 for trans-6; −10.6 kcal mol−1 for cis-6) are exothermic. Despite this thermodynamic preference for trans-6 the formation of cis-6 should be kinetically favored according to the Beckwith–Houk rules6 and our own calculations on related systems. A detailed analysis of the transition-state structures of titanocene-mediated cyclizations will be published in due course. Intermediates in the DFT model calculation of the reaction of epoxide 4. The relative energies in brackets are given in kcal mol−1 and refer to the separated reactants (4 and [TiCp2Cl] radical); bond lengths are given in Å.5 The key steps of our tandem reaction, formation of cis-7 and trans-7 complexed to [Cp2TiCl], were also found to be exothermic by −12.2 and −4.6 kcal mol−1. Thus, breaking of the TiO bond is predicted to be possible! The low energy of formation of trans-7 compared to that of cis-7 can be attributed to the strain in the resulting trans-fused bicyclo[3.3.0] system. The product complexes [7⋅TiCp2Cl] shown in Scheme 2 are slightly more stable (−31.1 and −25.1 kcal mol−1) than the separated molecules. Since entropy effects favor the dissociation of the product complex, ΔG will most likely be negative and the catalyst will be regenerated as desired for efficient catalysis. The relative energies of the transition-state structure, 6, and 7 were also confirmed by MP2 calculations. The transition-state structure for the formation of [cis-7⋅Cp2TiCl] is shown in Figure 1. This final ring closure has a low barrier (+11.4 kcal mol−1) and should therefore be viable even at low temperatures. The transition state exhibits similar TiO and OC bond lengths. This indicates that a homolytic, concerted substitution reaction4 is taking place which resembles the SN2 reaction with anionic nucleophiles. The calculated spin densities give a clear indication that the radical center is shifted from carbon to the metal (C +0.37, O −0.05, Ti +0.70). Transition-state structure for the ring closure of cis-6 to cis-7⋅TiCp2Cl (bond lengths in Å). These findings suggest that only cis-3 will be preparatively accessible for both kinetic and thermodynamic reasons. With substrates similar to 1 selectivities of 85:15 to 90:10 in favor of the cis isomer required here were typically observed2a,2d, 3a,3b,3e in accordance with the Beckwith–Houk rules.1c, 6 Epoxy olefin 1, which is easily prepared, should therefore constitute a useful starting point for preparative investigations. We have chosen the tertiary radical 2 as the key intermediate for the tandem reaction because it is relatively inert towards hydrogen-atom abstraction or reduction by a second equivalent of the titanium(III) reagent to yield an organometallic intermediate.2, 3 However, in the absence of other pathways this reduction has been observed.7 This relative persistence under reducing conditions through exclusion of competing radical pathways should promote the desired tandem reaction. Our results with 1 are summarized in Table 1. The reaction works well in THF or ethyl acetate in the presence of 10 mol % [Cp2TiCl2] as the precatalyst and Mn or Zn dust (2.0–0.2 equiv) as the reductant, producing cis-3 in up to 66 % yield. No trans-3 was formed as expected from the calculations. It made little difference whether Mn or Zn was used (entries 1–4). The experiments suggest that a low stationary concentration of the titanium(III) reagent under the highly reducing conditions is essential for better yields due to the deceleration of the undesired reductive trapping of the tertiary radical. Thus, the stoichiometric use of [Cp2TiCl] (entry 6) results in a reduced yield. The use of ethyl acetate (entry 4) is beneficial with Zn because the reduction of titanium(IV) is slowed down markedly. Entry Catalyst Conditions Yield [%] 1 [Cp2TiCl2] (10 mol %) THF, Zn (2 equiv), Coll⋅HCl (2.5 equiv) 57[b] 2 [Cp2TiCl2] (10 mol %) THF, Mn (2 equiv), Coll⋅HCl (2.5 equiv) 51[b] 3 [Cp2TiCl2] (10 mol %) THF, Mn (0.2 equiv), Coll⋅HCl (0.5 equiv) 66[c] 4 [Cp2TiCl2] (10 mol %) EA, Zn (0.2 equiv), Coll⋅HCl (0.5 equiv) 63[d] 5 [Cp2TiCl2] (10 mol %) THF, Mn (0.2 equiv), Coll⋅HCl (0.5 equiv) 66[e] 6 [Cp2TiCl2] (100 mol %) THF, Zn (2 equiv) 50[b] 7 [CpTiCl3] (10 mol %) THF, Mn (0.2 equiv), Coll⋅HCl (0.5 equiv) 48[f] The reaction of trans-2 yielded products of simple reductive cyclization, and addition of collidine hydrochloride (0.5 equiv, not optimized) was needed to regenerate the catalyst by the usual protonation of Ti-O and C-O bonds.2a,2b,2d Entry 6 demonstrates that collidine hydrochloride is not necessary for the reaction to occur. Without reductant no product was formed. Thus, the transformation did indeed proceed under titanium(III) catalysis. The reaction could be accelerated by heating to 70 °C without affecting the yield (entry 5). The observation that [CpTiCl3] led to distinctly inferior results (entry 7) lends support to our mechanistic proposal. Here the Ti-O bond should be stronger and hence more difficult to break because of the higher Lewis acidity of titanium and the reduced steric interaction of the alkoxide with only one bulky cyclopentadienyl ligand. The reaction is not only catalytic but also atom-economical, a feature rarely displayed by radical reactions.9 We also tested SmI2, a powerful and frequently employed electron-transfer reagent,10 CrCl2,11 and VCl2,12 but none of the desired product was obtained. Some other tandem reactions are summarized in Table 2. The reaction works well for substrates with trisubstituted double bonds (entries 1–4). The spiro tricyclic compounds 9, 11, and 13 should be of special interest as there are no other simple approaches for these structurally complex units. The bicyclo[4.3.0] system 17 (entry 5) demonstrates the usefulness of our titanocene-catalyzed protocol. Usually 6-exo cyclizations are distinctly less efficient than the corresponding 5-exo cyclizations in contrast to our system.1 The reaction also works well for substrates containing 1,2-disubstituted double bonds (entries 6 and 7). The diastereoconvergent synthesis of 21 creates a tricyclic system with complete selectivity in just three steps from commercially available starting materials and amply demonstrates the potential of our reaction for the synthesis of complex molecules. Entry Substrate Product Yield [%] 1 -1 -1 61[a] 2 -1 -1 62[b] 3 -1 -1 62[a] 4 -1 -1 64[b] 5 -1 -1 66[b], d.r.=88:12 6 -1 -1 60[c], d.r.=80:20 7 -1 -1 63[c], d.r.>98:2 Although all the experimental evidence and the theoretical results support the proposed radical pathway, we also investigated the possibility of cationic pathways.13, 14 However, exposing 1 to MnCl2, ZnCl2, camphorsulfonic acid, [Cp2TiCl2], and 2,4,6-collidine hydrochloride did not lead to 3, whereas reaction with BF3⋅Et2O lead to extensive decomposition. In the case of MeAlCl213 lactone 22 was the only product (52 %) obtained besides recovered 1 (23 %) as shown in Scheme 3. Thus, nucleophilic substitution is unlikely under our conditions. The oxidation of radical cis-2 to give a cation by a radical–polar crossover mechanism14 and THF ring closure by an SN1 reaction were rendered unlikely by the other two examples shown in Scheme 3. In both cases hardly any (<15 % as judged by NMR analysis of the crude mixture) of the desired product was formed. The phenyl and the acetoxy group should promote the putative oxidation by stabilizing the cation by mesomeric effects. Besides, this mechanism requires the presence of oxidants. This is unlikely under our reducing conditions because excess metal powder is always present. Experimental evidence against cationic intermediates in the reaction of 1. Dedicated to Professor Manfred T. Reetz on the occasion of his 60th birthday