I would like to ask for a bit of advice in choosing a "good" active space (CAS) in my CASPT2 calculations for the following reaction:

TiCl4 TiCl3 + Cl

My goal is to obtain an accurate potential energy surface for this reaction (there is no transition state). I did an initial molecular orbital (MO) analysis by hand using group theory and then I confirmed the results by running a simple HF calculations with the double-zeta basis set (vdz). In the equilibrium TiCl4 geometry I obtained the following MOs ordered with an increasing energy:

Occupied:

MOs, bonding type Ti AOs           Cl AOs

a1 (sigma bonding, Ti: 4s      and Cl: 3s,3p)

t2  (sigma bonding, Ti: 3d,4p and Cl: 3p)

e   (pi bonding,        Ti: 3d      and Cl: 3p)

t2  (pi bonding,        Ti: 4p,3d and Cl: 3p)

t1  (non bonding,                           Cl: 3p) HOMO

Virtuals:

e*   (pi) LUMO

t2*  (pi)

a1* (sigma)

t2*  (sigma)

I am now just wondering which orbitals I should set as active in my MCSCF calculations. I am planning to run a lot of single point CASPT2 jobs, so if possible I would prefer the CAS which is not too big. I am using MOLPRO for all my computations.

I would be very grateful for any advice.

Daniel