Relaxed PES for Excited States

I am so excited to test Q-Chem’s amazing features for excited state chemistry but, to my surprise, I stumbled upon an error message while trying to perform an excited state PES. The message makes it clear that I can’t do that with Q-Chem v6.0:

Checking the input file for inconsistencies... 
         PES scan currently implemented only for the ground state.
         Please remove CIS_STATE_DERIV from $rem.
         The Q-Chem input file has failed to pass inspection

Since it is possible that I am doing something wrong, my question is: is, at all, possible to do relaxed PES scans for excited states using Q-Chem?
Below is the input that I put together:

   JOBTYPE  pes_scan
   BASIS  =  def2-tzvp
   GUI  =  2
   METHOD  =  wB97XD
   CIS_N_ROOTS         5
   CIS_SINGLETS        true
   CIS_TRIPLETS        false
   CIS_STATE_DERIV     1   Lowest TDDFT state
   STATE_FOLLOW        true
   XC_GRID             3
   THRESH              12

   tors 19 18 13 12 0 360 8

   Theory      CPCM
   Method      SWIG
   Solver      Inversion
   Radii       Bondi


It is not currently a feature that is enabled, as the error message suggests. In part that’s because JOBTYPE=PES_SCAN is a “relaxed” scan that optimizes other degrees of freedom at each step. Excited-state geometry optimization is not always black-box, due to state-switching.

Thanks for the info. I was hoping that the root/state following implemented in Q-Chem could be more robust than ORCA’s.

It’s not a bug, it’s a feature. In electronic structure theory, we compute adiabatic (Born-Oppenheimer) eigenstates of the electronic Hamiltonian, and those states may not have uniform diabatic labels (e.g., pi-pi*, n-pi*, etc.) across the potential energy surface. CIS_STATE_DERIV specifies the index of the adiabatic state, so with CIS_STATE_DERIV=1 and singlet symmetry, you optimize on S1 regardless of what diabatic character that state may have. If the diabatic character changes, this can lead to abrupt changes in the forces that hamper the optimization. So we could probably turn this feature on for excited states (and if you are a developer you can probably figure out how to do this in the code), but without a lot of additional tools or checks, it is likely to generate a lot of problems for users and thus a lot of questions that aren’t related to the software but rather the physics of the problem.

1 Like