[molpro-user] Problem building a PES of a collision of a proton with a diatomic molecule when angle is different than 0⁰
Muammar El Khatib
muammarelkhatib at gmail.com
Tue Mar 20 15:37:39 GMT 2012
I'm trying to build the potential energy surface of the collision of proton with
a diatomic molecules which symmetry is X¹∑g in its ground state. By using do
loops, I have been able to approach a proton to the diatomic molecule with an
angle of collision of 10⁰:
X-X / } 10⁰
I am using CAS-SCF (MULTI) and then MRCI, with a CAS(8,10). The MULTI wf
converges without problems along all the X-H distances set in the loop up to the
one I already calculated the geometry optimization for knowing its value at the
equilibrium with the same methodology. But then, when I proceed to do the
breaking of the bond of the diatomic molecule:
X----X / } 10⁰
At the very second step in the loop (every step being a change in the distance
of 0.1Å), the calculation stops with the next message:
?Error: MCSCF not converged. This error exit can be avoided using the
IGNORE_ERROR option on the ORBITAL directive.
I know the problem is because the X¹∑g symmetry is being broken to an excited
state, a triplet (by analyzing the CAS occupancies). And IMHO, the use of
IGNORE_ERROR shouldn't be an option given that it'll lead to meaningless results
as it is also stated in the manual. Thus, I'd be really glad if somebody could
help me or give me an idea for solving this. I have been thinking in maybe:
1) Do a state averaged MCSCF for the breaking of the X-X bond. (?)
2) Project the Λ state , being X¹∑g, by using the lquant card. (?)
3) Just change the wave function of this part to represent a triplet state
instead of the same WF used for the X¹∑g. (?)
But I'm not quite sure how those ideas I have can affect the energies and
reliability of the PES I'm trying to build.
PS. Collision when angle is 0⁰ does not have any issues.
Muammar El Khatib.
Linux user: 403107.
Key fingerprint = 90B8 BFC4 4A75 B881 39A3 1440 30EB 403B 1270 29F1
http://muammar.me | http://proyectociencia.org
: :' :
More information about the Molpro-user