Molcas Forum

Lukhmanul Hakeem k

Member

Registered: 2024-02-05

Posts: 45

Clarification on Vertical and Adiabatic Excitation Energy Calculations

Dear Sir,

I hope you are doing well. I am new to OpenMolcas, and I am performing excited-state calculations. While going through the manual and performing some test calculations, I encountered a few doubts that I would be very grateful if you could clarify.

I am studying the excited-state behavior of a molecule, and my workflow is as follows:

Ground-State Optimization and Single-Point Calculation
I optimized the ground state using the 6-31+G** basis set and then performed a single-point energy calculation at the ANO-L-VDZP level using the following setup:

&GATEWAY
  Symmetry = Z
  Basis set
  C.ANO-L-VDZP.
  ...
&SEWARD
&SCF
  OCCUPIED = 47 10
  ITERATIONS = 40
&RASSCF
  Symmetry = 1
  Spin = 1
  NACTEL = 12 0 0
  INACTIVE = 47 4
  RAS2 = 1 11
  LumOrb
&GRID_IT
  All

The total energy obtained for the ground state (S₀) was –761.54368664 a.u.

Vertical Excitation Calculation
Using the optimized ground-state geometry, I performed a state-averaged CASSCF calculation (3 roots with equal weights) using:

&GATEWAY
  Symmetry = Z
  Basis set
  C.ANO-L-VDZP.
  ...
&SEWARD
&SCF
  OCCUPIED = 47 10
  ITERATIONS = 40
&RASSCF
  Symmetry = 1
  Spin = 1
  NACTEL = 12 0 0
  INACTIVE = 47 4
  RAS2 = 1 11
  CIROOT = 3 3 1
  LumOrb

The obtained energies were:

Root 1: –761.53732837 a.u.

Root 2: –761.38803548 a.u.

My first doubt:
To calculate the vertical excitation energy of the first excited state (Root 2), which reference energy should I use? (Root2-which energy)
(a) the ground-state single-root energy (–761.54368664 a.u.), or
(b) the ground-state energy from the state-averaged calculation (Root 1 = –761.53732837 a.u.)?

I performed the SA-CASSCF because the second and third excited states (Roots 2 and 3) are close in energy, and I plan to apply CASPT2 correction afterward. Which CASPT2 method should I use — MS-CASPT2 or SS-CASPT2?

Adiabatic Excitation Energy and CASPT2 Correction
I optimized the geometry for the first excited state (Root 2) using the 6-31+G** basis set and then performed a state-averaged CASSCF calculation with CIROOT = 3 3 1 and RLXRoot = 2.
Next, I performed a CASPT2 calculation on this structure with the following input:

&GATEWAY
  Symmetry = Z
  Basis set
  C.ANO-L-VDZP.
  ...
&SEWARD
&SCF
  OCCUPIED = 47 10
  ITERATIONS = 40
&RASSCF
  Symmetry = 1
  Spin = 1
  NACTEL = 12 0 0
  INACTIVE = 47 4
  RAS2 = 1 11
  CIROOT = 3 3 1
  LumOrb
&CASPT2
  LROOT = 2

Since I was focusing on Root 2, I used single-state CASPT2 (SS-CASPT2).

My second doubt:
As the SA-CASSCF was performed with three roots and the Root 2 and Root 3 energies are close, should I instead use multi-state CASPT2 (MS-CASPT2) for more accurate treatment of state interaction? Or is it still reasonable to apply SS-CASPT2 in this case?

I would be very thankful for your guidance on these points, as I am still familiarizing myself with the best practices for excited-state calculations in OpenMolcas.

Thank you very much for your time and help.

Kind regards,
Lukhmanul Hakeem K.

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,165

Re: Clarification on Vertical and Adiabatic Excitation Energy Calculations

In general, when you compare energies, you want them all to be of the same "quality". The single-state CASSCF has orbitals optimized for the ground state. The state-averaged CASSCF has orbitals that are not optimized for any of the 3 lowest states, but an average of them, hopefully the orbitals will be equally good/bad for all states. So it's better to compare the energies of excited and ground state all from the SA calculation.

If there are states close in energy, you probably want MS-CASPT2 (or rather XMS- or RMS-). Again, if you're comparing energies (e.g. adiabatic vs. diabatic excitations), you usually want to perform all calculations with the same method and settings, as far as possible.