Dear Valeria,

I think that there is an error with respect to the appropriate usage of the pseudopotential for Cs. It should be defined as it is given below, so just give it a try:

------------------------------------------

6-31G*

****

Cs 0

SDD

****

Cs 0

MWB46

------------------------------------------

Since the Cs atom is not anymore in the range H-Kr for which the ECP is defined , you have to use the effective core potential in combination with the quasi-relativistic Wood-Borsig-core potentials with 46 electrons (look here: http://www.gaussian.com/g_tech/g_ur/k_pseudo.htm). This has to be defined seperately.

Hope this helps!

Dear Okan,

thanks for the answer. I tried and unfortunately I got again the same error . Actually, I did several tests before. First  I tried LANL2DZ pseudo potential and I got the same error. Then, I perform Hirshfeld population analysis. The calculations ended with normal termination in both the cases (LANL2DZ and SDD potentials). So I think is something related to the NBO method and not to the pseudo potential.

I forgot to say that I performed the calculation with Gaussian 09 and 03. 

Additionally, you could try: modification of "pop=nbo" to "pop=nboread" in the input line and at the end of the file you should include "$NBO E2PERT $END". Let me know if this works for you. Otherwise, I suspect that the nbo package is not compiled properly...

Thanks again for your help. Unfortunately, using "pop=nboread" and "$NBO E2PERT $END" did not help because I got the same error. Anyway, I was able to get normal termination using a different all electron basis set.

Dear Valeria Butera,

Finally, how did you solve this problem? I am also facing the same @ to Valeria Butera,

Hi Valeria and S. Jayapandi,

I ran into this problem also..... the LANL2DZ basis set only has s and p orbitals for group 1 and 2 elements. The NBO analysis will work if you use a basis set that also includes d orbitals for the heavier elements (such as Def2TZVP).