Hello,
I am wondering if performing SCF stability analysis on a triplet structure using the “INTERNAL_STABILITY TRUE” keyword would provide any different results from running a TD-DFT calculation for the structure to see if there are any negative excitations.
I believe that both methods should work. (I don’t have an unstable triplet input lying around but for an unstable doublet sample job, both methods detect the same negative eigenvalue for the orbital Hessian.) One advantage of INTERNAL_STABILITY in this respect is that in case of an unstable solution, the stability job will automatically displace (perturb) the orbitals in the direction of largest curvature, so that you can use SCF_GUESS = READ in a subsequent job in order to try to converge a stable solution. The CIS/TD-DFT code doesn’t do that, it just reports excitation energies.