For states of different multiplicity (e.g., S_0 and T_1), the Gaussian documentation is not very clear on how one goes about computing the spin-orbit coupling matrix element. Consider for instance a benzene molecule which has a ground singlet state (S_0) and the lowest lying excited state is a triplet (T_1). When I perform a state averaged CASSCF calculation, the computed spin-orbit coupling elements are between S_0 and S_1 (i.e., states of the same multiplicity) rather than states of different multiplicity. A state-averaged calculation with a triplet state (NRoot = 1) in attempt to couple with S_0 throws an error. However, from the paper (East, Allan LL, and Edward C. Lim. "Naphthalene dimer: Electronic states, excimers, and triplet decay." The Journal of Chemical Physics 113.20 (2000): 8981-8994.), it seems that it is possible to do spin-orbit coupling calculations for states of different multiplicity in Gaussian98. I am currently using Gaussian16, and would be very glad if anyone has encountered this problem before or knows about the syntax used to request such a calculation.

Thanks!