ROTATIONAL BARRIERS OF 1,1'-BINAPHTHYLS - A COMPUTATIONAL STUDY

The activation energies for rotation about the sigma-bonds of 1,1'-bin aphthyl (1) and 2,2'-dibromo-l,l'-binaphthyl (2) have been computed wi th MNDO, AM1, and PM3 and of 2,2'-dilithio-l,l'-binaphthyl (3).2EDA (e thylenediamine) (3a) with MNDO. All methods find that 1 should racemiz e preferably through the anti path, in agreement with previous force f ield calculations. The PM3 rotational barrier for 1 (23.1 kcal/mol) ma tches the experimental value (22.5 kcal/mol) best; the ground-state bo nd lengths correspond well with the X-ray data. We developed a procedu re which evaluates the distortion and the steric repulsion effects in the transition structures roughly. In 1, distortion effects (e.g., rin g deformation) account for about 2/3 of the activation energy. On the basis of the rotational behavior of 1, previous authors have only cons idered the anti racemization mechanism to be viable for 2,2'-dimethyl- l,l'-binaphthyl (4). In contrast, we found that in 2 (methyl is about the same size as bromine) the syn pathway is favored substantially ove r the anti route by 15.1 (MNDO), 20.6 (AM1), and 26.3 (PM3) kcal/mol. For 2, PM3 again yields the lowest rotational barrier (30.3 kcal/mol) but the AM1 value (38.4 kcal/mol) is in better agreement with an earli er estimate for 4 (37-40 kcal/mol). The transition structures (TS) rel ated to 2 are even more strongly dominated (75-93 %) by distortion eff ects than those for l. Two energetically almost degenerate energy mini ma are computed with MNDO for 3a: one with the lithiums symmetrically doubly bridging the markedly twisted naphthyl rings (twist angle: 42.5 -degrees) and the other with each lithium closely coordinated to the c ontiguous pi-system (twist angle: 122.0-degrees). Despite the size of the Li.ethylenediamine 2,2'-substituents in 3a, the anti racemization pathway is preferred by 6.4 kcal/mol with an unusually close Li ... H contact (1.79 angstrom). However, the syn-TS is 6.3 kcal/mol lower in energy with Li.EDA (3a) instead of hydrogens (1) in the 2,2'-positions . Thus, the syn-TS of 3a profits from electrostatic stabilization thro ugh lithium double bridging. Upon further rotation, the lithium atoms swap their counterions. To correct for the known overestimation of the Li-C bond strength by MNDO, we compared its performance on a model sy stem (1,4-dilithio-1,3-butadiene (6)) with the MP2/6-31G//6-31G* resu lts. Deviations between the semiempirical and the ab initio geometries of four isomers of 6 suggest that the magnitude of the MNDO overestim ation are 8 kcal/mol for Li-C and 4 kcal/mol for Li-H interactions. Wh en these corrections are applied, the activation energy of 3a (anti-TS : 22.1 kcal/mol) should be close to the experimental estimate for 2,2' -dilithio-l,l'-binaphthyl (3) in solution (18.4 kcal/mol).