Molcas Forum

mabuch

Member

Registered: 2020-10-10

Posts: 2

excited states - diatomic molecule

Dear All,

I'm trying to calculate ground and excited states (X1Sigma and in particular repulsive curve 3Sigma; as well as others) of OH- ion. My input is:

&GATEWAY
 coord
 2
 bohr
 O 0.0 0.0 0.0
 H 2.0 0.0 0.0
  basis=ANO-L-VQZP
&SEWARD
&RASSCF
 charge=-1
 spin=1
 inactive= 1 0 0 0
 ras2= 5 2 0 2 (or something similar)
&CASPT2
 multistate= 2 1 2 (or multistate = 1 1 at first).

1. The ground state is as I expected (with spin=1,multistate= 1 1).

2. I have tried what I will get with spin=3, multistate= 1 1 - it looks repulsive but should be steeper and smoother.

3. Finally when I tried multistate= 2 1 2 there is "Colossal value detected in SIGMA" error.

Q1a. Is there anything I could do to improve the results? Where I can find some guidelines for setting ras2 in case of diatomic molecules, especially excited states? Perhaps I should use other method but CASPT2 seemed most appropriate, I need full potential curves (from 1bohr to 12 bohrs for example). (I must admit that some keywords such as ciroot are quite misterious for me now)
Q1b. I have set spin=3 for another state perhaps the repulsive one, there is also low lying 3Pi state so I was not sure which one I will get - it seems I got the repulsive one (can I read out term symbol form output). can I improve something expecially for this desired 3Simga state.
Q2. How to fix input to avoid "Colossal value detected in SIGMA" and other possible errors in calculation of excited states (I used multistate= 2 1 2).
Q3. Can I somehow control which excited state I'am calculating (or can I translate molecular term symbol into openmolcas input)?

I plan to deal usually with ions so keyword "charge" is obligatory for me.

Is there some workshop/webinar planned?

I am sorry for several questions at once but I'am trying to learn to do a specific job - calculation of potential energy curves (ground and excited) of diatomic molecules (usually ions) and in the best case how to control which curve I'm calculating.

Regards,
Marcin

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,079

Re: excited states - diatomic molecule

• Start with CASSCF only (not CASPT2) to analyze your wavefunctions. Relative energies may be wrong by a lot, but qualitative curves should be OK, smoother and with one (or more) less point of failure.

• Multistate CASPT2 is useless if you don't have several CASSCF states (with same spin and space symmetry), i.e. the CIRoot keyword. If "Multistate = 2 1 2" didn't stop with user input error, it should have.

• CIRoot is not that difficult. If you want n states (same spin and space symmetry), just write "CIRoot = n n 1".

• You're doing calculations with symmetry. If you don't say anything, it will by default be "Symmetry = 1" (fully symmetric), probably Sigma in your case. If you want any other space symmetry, you have to specify it in the RASSCF input. Which one is right depends on the order of the symmetry operators, you should be able to figure it out from the symmetry information printed by GATEWAY.

• "Charge" is never obligatory, you can specify the number of active electrons (NActEl). The number of active electrons plus twice the number of inactive orbitals is the total number of electrons... which gives the charge.

mabuch

Member

Registered: 2020-10-10

Posts: 2

Re: excited states - diatomic molecule

Thank you for help. Now using the followig input I recovered 3 PECs of OH-.

&GATEWAY
 coord
 2
 bohr
 O 0.0 0.0 0.0
 H 2.0 0.0 0.0
  basis=ANO-L-VQZP
&SEWARD
&RASSCF
 charge=-1
 spin=1
 inactive= 0 0 0 0
 ras2= 8 2 1 2
 ciroot= 5 5 1
 symmetry= 1
&CASPT2
 multistate= 1 1

1. As I understood changing spin, symmetry and using multistate I should be able to get all the curves (as I did so far)?
2. Can I figure out term symbol from output file? (I understand that looking for term should be easier a posteriori - in output than a priori - in input)
3. Does number of states in "CIRoot = n n 1" works the more the better or there are some rules?
4. Could you elaborate on analyzing wavefunction? What should I check in output?
5. I have also tried to calculate ground state curve of NH (3Sigma-) and I have recovered the curve but absolute energies do not agree with data in the publication I'm comparing with ("we used the
CASSCF method followed by icMRCI approach [48,49] with Davidson modification (icMRCI + Q) [50,51] to calculate all the PECs") - should I take something into account calculating molelules such as NH which is similar and I in fact expected very good results (the same molecular orbitals but less electrons)?

Supplement - my NH energies are in agreement with other publication. What can I do to have reference quality energies on such molecules as NH, OH-, OH+, CH, CH+, C2... ?

Last edited by mabuch (2020-10-16 14:53:15)