Furthermore, the mechanistic details were investigated with the aid of density functional theory (DFT) calculations at the level of B3LYP-D3/SMD (solvent = DMF)/SDD&def2-TZVP//B3LYP-D3/SMD (solvent = DMF)/LANL2DZ+
p&6–31G(d) (see Supporting information for more computational details). As shown in
Fig. 1, reduction of NiBr
2 complex
CP1 to Ni
0 complex
CP2′ with Zn is highly endergonic by 34.8 kcal/mol, being significantly less favored than the reduction to Ni(Ⅰ) complex
CP2. This result is consistent with our control experiment which excluded a Ni
0-Ni
Ⅱ catalytic process (
Scheme 4b). Very recently, the experimental mechanistic study from Li also indicates that the reduction of NiI
2 precatalyst by Zn generates Ni
Ⅰ species rather than Ni
0 species [
20]. From
CP2, oxidative addition of PhI goes through
TS1 to afford Ni
Ⅲ complex
CP3 which is then again reduced to Ni
Ⅰ−pH complex
CP4 by Zn. Thereafter, sulfonium cation
2a+ can oxidize the Ni
Ⅰ species
via C-substitution transition state
TS2-C or S-substitution transition state
TS2-S. For
2a+,
TS2-S is slightly favored over
TS2-C by 1.4 kcal/mol and generates a Ni
Ⅱ complex
CP6-S that contains an alkyl radical (Fig. S1 in Supporting information for the plot of Mulliken spin density). The bonding of the alkyl radical to Ni
Ⅱ center forms the more stable Ni
Ⅲ complex
CP6-C, from which C(sp
3)−C(sp
2) reductive elimination can occur
via TS3. The resulting Ni
Ⅰ complex
CP7 undergoes oxidative addition with another PhI
via TS4, followed by reduction with Zn to
CP4 to restart the catalytic cycle. According to the calculated energy profile, the oxidation of Ni
Ⅰ−pH complex
CP4 determines the regioselectivity for asymmetric sulfonium salts (
Scheme 2,
3aa−
3ac). Taking the starting material of product
3aa as the model substrate, the S-substitution transition states (
TS2-SMe and
TS2-SH) also have lower energies than the C-substitution transition states (
TS2-CMe and
TS2-CH). This phenomenon possibly results from the fact that the benzene ring on the S atom is closer to the pyridine rings of ligand
L8 in
TS2-SMe and
TS2-SH as reflected by the minimum distance of their geometric centers (
Fig. 1 and Fig. S1). Thus a stronger
π-π stacking effect is expected in
TS2-SMe and
TS2-SH (
Fig. 1 and Fig. S1 for plots of noncovalent interaction). On the other hand,
TS2-SMe is more stable than
TS2-SH by 3.3 kcal/mol, supporting the superiority of C−S bond cleavage at the alkyl-substituted C-terminal. It was found that remarkable Mulliken spin density spreads on the cleaving carbon in
TS2-SMe and
TS2-SH with the absolute values of 0.495 and 0.561, respectively, contributing to the lower energy of
TS2-SMe that owns a more stable carbon radical.