Dear All,
I am looking for some advice in choosing a method that would avoid the mentioned conical intersection. Here is my problem:
I am trying to build a potential energy surface for TiCl3 + Cl reaction. The idea is to find energies at different locations of the two interacting fragments (keeping TiCl3 structure fixed at its asymptotic geometry). I am mostly interested in the region where TiCl3 and Cl are 3-6 angstroms apart (at different angles as well). As a first pass I decided to use MS-CASPT2 (multi state) method with the active space consisting of 6 electrons and 4 orbitals (3p Cl orbitals and radical orbital on TiCl3) where orbitals were optimised in a state-averaged way (3 states depending on which p orbital of Cl atom is used) in order to avoid root-flipping problems. However, I have noticed that at about 3.8 angstrom there is a conical intersection. Two excited states cross the ground state energy curve. I was then wondering if there exist any other method that is not so expensive that would give me correct energy curves? I am using Molpro for all calculations. I would be very grateful for any advice.
To my mind question is incorrect: if there is CI, you can not avoid it (or if you do - you have wrong method).
As for cheaper methods, look here for comparison of different approaches:
Article Shape of Multireference, Equation-of-Motion Coupled-Cluster,...
Dear Oleg,
Many thanks for such a quick response and clarification. The CI subject is completely new to me and I was a little bit confused. Given your answer and articles that I have just read I see my mistake now.
Best regards!
I think it is a good idea to first try a CASSCF approach. CASPT2 is outstanding at getting good values of excitation energies and other things that needs a full CI estimate, but if you are interested in the topology of the potential energy surface you don't need to add the second order perturbation just yet. And whilst state averaging is good for a balanced approach (and needed when you do the PT2) it will not do you any favours if you are interested in the more 'intimate' parts of the surface, such as where intersections are etc.
I would also suggest that there are some more orbitals that are in fact active (so partially occupied) along your reaction coordinate ? Maybe there are some metal-ligand charge transfer states kicking about? Perhaps some cool Jahn-Teller conicals? (I've only fiddled with organic molecules so haven't drawn a good MO diagram in years.. :))
Mike Robb, Bearpark, Yarkony , Martin Paterson... one of those guys might have a good paper. Although they are probably biased to problems that suit Gaussian use...
Anyways... All the best!
Dear Therese,
Thanks for good tips! However, I believe that the CASPT2 is a must in this particular case. I am going to use the resulting PES to compute rate constant, where I need accurate description of the interaction energies between the fragments.
Best reards!
- Badges
- Science topic
- Similar topics
- Mathematics