[molpro-user] Symmetry breaking problem

Denis DUFLOT denis.duflot at univ-lille1.fr
Wed Apr 8 13:15:16 BST 2009

Dear Molpro users,

I want to calculate the polarisability of Cl2 at varying R distances, at 
CASSCF then MRCI level.
For alpha_zz, you need to apply an electric field on the molecular axis z
and break the symmetry from D2h to C2v. Before doing that, I got strange 
results while calculating the CASSCF energy
of the ground state of Cl2 in the C2v symmetry and without any field. 
The input is as follows (with avtz it takes a few minutes) :

 do i=1,#rcl

The results are the following :

       R         ECAS             MOMZ
     3.00   -918.8764455   -8.87898467D-03
     3.10   -918.9201909   -1.01054493D-02
     3.20   -918.9532175   -1.05214172D-02
     3.30   -918.9779788   -2.69249371D-05
     3.40   -918.9960462   -8.80644811D-07
     3.50   -919.0086897   -7.12186770D-09
     3.60   -919.0170423   -1.45478461D-11
     8.00   -918.9601474    1.08097581D-10
     9.00   -918.9601854   -1.43403170D-12
    10.00   -918.9602606   -1.76912890D-11
    15.00   -918.9603767    3.09126457D-10
    20.00   -918.9603921    1.26118706D-08

At short R (<3.4], the CASSCF dipole moment is really different from 
zero which is clearly unphysical.
Moreover the energy is LOWER than the value obtained in D2h. A R = 3.0 :
D2h: E = -918.8752201
C2v: E = -918.8764455
The explanation is clear when inspecting the MO's : at short R, the 2 1s 
, for example, are partially localised on
each atom and the u/g symmetry is broken, as in a heteronuclear 
molecule. When R increases, this localisation
progressively disappears and u/g is restored. It is also interesting to 
note that the RHF MO's are OK.

(I am still with 2006.1 version and I tried on 2 different 

This is puzzling since I do not expect CASSCF to behave like this. I 
asked a colleague to check with
molcas : whatever the symmetry (D2h, C2v or even C1) the energy at R = 
3.0 is -918.8752201 ie the D2h value.

I tried a few tricks (doing a CASSCF on the di-anion and use the MO's to 
start the neutral, trying to begin
in D2h then switching to C2v - this seems not possible - and even 
beginning with R = 20.0). I never managed
to get the "right" (D2h) results. So my questions are :

- as far as I know the CASSCF result should be the same whatever the 
symmetry. So is this a "bug" in Molpro?
(the non-zero dipole moment is especially annoying in the polarisability 

- maybe there are some options or keywords I do not know about to obtain 
the correct behavior?

- or maybe I miss something?

Any ideas/suggestions are welcome...
D. Duflot

Dr Denis Duflot
Maître de Conférences/Assistant Professor
Université Lille1
Laboratoire de Physique des Lasers, Atomes et Molécules
UMR CNRS 8523 Bâtiment P5
F-59655 Villeneuve d'Ascq Cedex
e-mail : Denis.Duflot at univ-lille1.fr
Tél : (33) 3 20 43 49 80
Fax : (33) 3 20 43 40 84

More information about the Molpro-user mailing list