[molpro-user] Calculating molecular oxygen excited state

[Lorenzo Lodi]

Dear MOLPRO forum,
I would like to be given some advice on an MRCI calculation of the 
second excited state of the oxygen molecule.
Here's a minimal introduction.
The three lowest electronic state of molecular oxygen are:

ground state: spin-triplet Sigma minus gerade; spectroscopic label X
first excited: spin-singlet Delta gerade; spectroscopic label a
second excited: spin-singlet Sigma plus gerade; spectroscopic label b

Fixing the energy of X to zero and quoting the values reported in the 
book by Herzberg and Huber state "a" has energy 7921 cm-1 and state "b" 
  13196 cm-1.

I would like to compute the potential energy curve for the "b" state 
using MRCI and the full-valence complete active space (MOLPRO's default 
choice).

I attach a minimal version of the input I'm using. It'll run RHF,CASSCF 
and MRCI for the ground state e then for the excited state and print the 
  vertical excitation energies in cm-1. It runs less than 30 seconds on 
a normal PC.
Using MOLPRO 2006.1 I get as output:

RHF: 13730.24945731 cm-1
CASSCF: 11949.09628260 cm-1
MRCI: 7840.49255049 cm-1

Looking at these energies it would seem that MRCI is converging to the 
"a" state instead of the desired "b". Is this possible?
I also noted a couple of other things:

1) If in the excited-state CASSCF run I do not specify "WF, NELEC=16, 
SYMMETRY=1, SPIN=0" (but still specifying LQUANT,0) I get slightly 
different CASSCF/MRCI energies:
CASSCF: 11940.35248204 cm-1
MRCI: 7831.22300238 cm-1

This seems to be because, if WF,16,1,0 is not specified explicitly, a 
state-averaged calculation is performed (but with which states?).


2) If "LQUANT, 0" is not specified CASSCF converges to
CASSCF: 7723.21965528 cm-1 ,
presumably a component of the "a" state. In this case it does not make a 
difference if "WF, NELEC=16, SYMMETRY=1, SPIN=0" is specified or not.

My main question is: in MRCI converging to the a state, as looking at 
the energies seem to indicate?


Thank you for your attention,

Lorenzo
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: O2.input
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20090924/0bf17eb4/attachment.ksh>