Computation-guided development of Au-catalyzed cycloisomerizations proceeding via 1,2-Si or 1,2-H migrations: regiodivergent synthesis of silylfurans.

A novel highly efficient regiodivergent Au-catalyzed cycloisomerization of allenyl and homopropargylic ketones into synthetically valuable 2- and 3-silylfurans has been designed with the aid of DFT calculations. This cascade transformation features 1,2-Si or 1,2-H migrations in a common Au-carbene intermediate. Both experimental and computational results clearly indicate that the 1,2-Si migration is kinetically favored over the 1,2-shifts of H, alkyl, and aryl groups in the beta-Si-substituted Au-carbenes. In addition, experimental results on the Au(I)-catalyzed cycloisomerization of homopropargylic ketones demonstrated that counterion and solvent effects could reverse the above migratory preference. The DFT calculations provided a rationale for this 1,2-migration regiodivergency. Thus, in the case of Ph(3)PAuSbF(6), DFT-simulated reaction proceeds through the initial propargyl-allenyl isomerization followed by the cyclization into the Au-carbene intermediate with the exclusive formation of 1,2-Si migration products and solvent effects cannot affect this regioselectivity. However, in the case of a TfO(-) counterion, reaction occurs via the initial 5-endo-dig cyclization to give a cyclic furyl-Au intermediate. In the case of nonpolar solvents, subsequent ipso-protiodeauration of the latter is kinetically more favorable than the generation of the common Au-carbene intermediate and leads to the formation of formal 1,2-H migration products. In contrast, when polar solvent is employed in this DFT-simulated reaction, beta-to-Au protonation of the furyl-Au species to give a Au-carbene intermediate competes with the ipso-protiodeauration. Subsequent dissociation of the triflate ligand in this carbene in polar media due to efficient solvation of charged intermediates facilitates formation of the 1,2-Si shift products. The above results of the DFT calculations were validated by the experimental data. The present study demonstrates that DFT calculations could efficiently support experimental results, providing guidance for rational design of new catalytic transformations.