Using SF-TDDFT single-point energies for TS optimized at DFT level

Questions
#1

Hi,

I have a question regarding transition state (TS) calculations using the SF-TDDFT method in Q-Chem. I understand that analytic Hessians are not available for SF-TDDFT.

In my current workflow, I have:

  • Optimized all reaction intermediates using SF-TDDFT (analytic gradients available), and
  • Optimized the transition state at the DFT level, followed by a frequency calculation.

My question is: Is it acceptable to perform only a single-point SF-TDDFT calculation on the DFT-optimized TS geometry and use these energies together with the SF-TDDFT–optimized intermediates to construct the potential energy surface (PES)?

In other words, can a PES be considered reliable if the intermediates are optimized at the SF-TDDFT level, while the TS geometry comes from DFT and is characterized at the SF-TDDFT level only via single-point calculations?

Any guidance or best-practice recommendations would be greatly appreciated.

#2

What is “acceptable” is really in the eyes of the beholder but there’s plenty of precedent for optimizing geometries of stationary points (local minima and/or transition states) at one level of theory, then performing single-point energy calculations at a different level of theory. There’s even a nomenclature for it: Method1/Basis1//Method2/Basis2.

That said, it sounds like maybe you’re thinking of optimizing different stationary points at different levels of theory? If so, that feels inconsistent to me.

Note that the newest versions of Q-Chem have MPI-parallelized finite-difference routines so depending on your system size, the lack of analytic gradients might no longer make something intractable.