Molcas Forum

dommett110

Member

Registered: 2017-03-01

Posts: 16

Excited States with CASPT2 in Solvent (PCM)

Hi there,

I would like to compute vertical excitations in acetone for a particular system - for now I will use acrolein as an example. In the manual (http://www.molcas.org/documentation/man … 0000000000) there are instructions for how to do this,
but I am having some problems following the guide.

As I understand, first a caspt2 calculation in solvent must be done on the ground state:

&GATEWAY
  Title=  Acrolein  molecule
  coord  =  acrolein.xyz
  basis  =  6-31G*
  group  =  c1
  RF-input
  PCM-model
  solvent
  water
&SEWARD;  &SCF
&RASSCF
  LumOrb
  Spin=  1;  Nactel=  6  0  0;  Inactive=  12;  Ras2=  5
&CASPT2
RFPErt

Then, "in the same working directory", I must perform an excited state calculation. If I want to caclulate three states, would the following input work?

&GATEWAY
  Title=  Acrolein  molecule
  coord  =  acrolein.xyz
  basis  =  6-31G*
  group  =  c1
  RF-input
  PCM-model
  solvent
  water
&SEWARD;  &SCF
&RASSCF
  LumOrb
  Spin=  1;  Nactel=  6  0  0;  Inactive=  12;  Ras2=  5
  NONEquilibrium 
 CIRoot
 3  3 1 
&CASPT2
Multistate= 3 1 2 3
RFpert
RFPErt
&RASSI
EJOB

What I dont follow is how the excited state uses the information from the ground state wavefunction in this second job? Would my inputs be correct in this case? Any help would be much appreciated

Thanks

Michael

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: Excited States with CASPT2 in Solvent (PCM)

I think you would need to perform individual calculations for each state with RFRoot, and MS-CASPT2 and RASSI would not be working options (because we have states obtained with different external fields).

The ground-state information should be in the $Project.RunFile file, which is used through the NonEq keyword. The relevant information is the "slow" polarization of the solvent, while the "fast" polarization is computed for each electronic state.

dommett110

Member

Registered: 2017-03-01

Posts: 16

Re: Excited States with CASPT2 in Solvent (PCM)

Hi Ignacio,

Thank you for your response, it is very helpful. I have been running a few tests. To calculate the first vertical excited state for acrolein in water with CASPT2, would the following be correct?

1. Calculate the ground state

&GATEWAY
  coord  =  acrolein.xyz
  basis  =  6-31G*
  group  =  c1
  RF-input
  PCM-model
  solvent
  water
  end of rf-input
&SEWARD;  &SCF
&RASSCF
  LumOrb
  Spin=  1;  Nactel=  6  0  0;  Inactive=  12;  Ras2=  5
  CiRoot
  2 2 1
  RFROot=1
&CASPT2
RFPErt
>>COPY $WorkDir/$Project.RunFile $CurrDir/RUNFILE2

2. In the same directory, calculate the excited state response:

&GATEWAY
  coord  =  acrolein.xyz
  basis  =  6-31G*
  group  =  c1
  RF-input
  PCM-model
  solvent
  water
  end of rf-input
&SEWARD;  &SCF
&RASSCF
  LumOrb
  Spin=  1;  Nactel=  6  0  0;  Inactive=  12;  Ras2=  5
  CiRoot
  2 2 1
  NONEquilibrium
&CASPT2
RFPErt

Does this seem the correct strategy to you? In the second calculation I have removed the RFRoot keyword. Furthermore, would reading saving the orbitals from the 1st step and reading them as input orbtials for the second RASSCF calculation be a better approach?

Once again, thanks for your help it is much appreciated

Michael

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: Excited States with CASPT2 in Solvent (PCM)

I think you still "need" RFRoot in the second calculation (except that it may be that by default it takes the highest state, which is what you want in this case, but check the output to be sure). However, the main issue I think is that the second calculation needs to read the "slow" component which I believe is stored in the RunFile, and GATEWAY will overwrite the previous one. I would just skip GATEWAY, SEWARD and SCF in the second calculation. Using starting orbitals from a converged calculation is always a good idea, it usually saves times and problems.