7
$\begingroup$

I am new to CASSCF. I want to draw a PES for the ground state and the first excited sate of methyl amine for learning purpose. I want to verify if I am correct using the correct approach.

  1. Do a RHF/STO-3G opt pop=full ... orbital calculation.
  2. Use the checkpoint file obtained in step 1 and use CASSCF (provided I know the active space) to optimize the structure again and then use the scan keyword to map the PES by increasing the N-H bond distance in a different Gaussian job.
  3. Take the optimized geometry of ground state obtained using CASCF and then use NROOT=2 and opt to get the the optimized geometry of 1st excited state and then scan to get the PES, similar to what was done in step 3.

Any correction would be greatly appreciated.

$\endgroup$
5
$\begingroup$

Yes in principle that is a correct approach, but you should be aware of a few pitfalls here and there.

The basis set you chose for your optimisation might not be large enough. I assume you are not going to use STO-3G for your CASSCF calculation. It is better if you run your initial optimisation with the same basis set that you are using for the CASSCF calculation. You will most likely be closer to the minimum and will safe a few cycles when optimising the molecular structure on the larger level of theory.
Picking the right active space might not be trivial. You may have to run a couple of single points to find the right balance of what you must include and what you may neglect.
The SCAN produces a rigid scan of the molecule, i.e. the backbone of the structure does not react with the induced change. That might not lead to the correct behaviour. More sophisticated (and more costly) is a relaxed scan via the ModRedundant keyword.

Be aware, that CASSCF calculations are quite resource demanding and you probably won't get results overnight.

$\endgroup$
  • $\begingroup$ Thanks for the reply. I am well aware of SCAN and ModRedundant keywords. I will also use another basis set which will yield values matching with the experimental ones. However, the problem is choosing the right active space. Can you refer any book or article explaining this in detail and maybe a few examples,if possible? It would be really helpful. $\endgroup$ – Chemist Oct 30 '17 at 15:02
  • 2
    $\begingroup$ @Chemist figuring that out is by some considered an art form. Unfortunately I don't really know a concise guide for that. Maybe you can find some help from one of the books in our resource thread. I'm sure I have written a few words on this site, too, but I'm on mobile and can't find it right now. You can always ask more questions here, when you have exhausted your other resources. You can also find me and other computational chemists in the periodic table (Chemistry Chat), feel free to drop by. $\endgroup$ – Martin - マーチン Oct 30 '17 at 15:37
  • 3
    $\begingroup$ @Chemist Multiconfigurational Quantum Chemistry by Björn Roos et al g.co/kgs/hp86Uo ... also, Gaussian is by far not the nicest CAS program around. $\endgroup$ – pH13 - Yet another Philipp Oct 30 '17 at 18:14
  • $\begingroup$ To expand on Martin's comment about STO-3G, I know it's usually my first "test" basis as well, but the virtual space may be too poor to ever converge. You can almost certainly afford def2-SV(P) or pc-0. If those don't work, def2-SVP or pc-1. These are inexpensive but much better basis sets, and not so much larger that you can't count MOs by hand. $\endgroup$ – pentavalentcarbon Oct 31 '17 at 12:37
  • 2
    $\begingroup$ You also have to be careful if you scan over a large range of geometries. It is very common for states of different symmetry to cross, so that e.g what was the first excited state when the molecule is planar is not the excited state when it is twisted. $\endgroup$ – Eric Brown Nov 2 '17 at 18:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.