CCL:G: NBO 3.1 in gaussian : d-elements

[Jon Wright <jon~!~gate.sinica.edu.tw>]

 Sent to CCL by: Jon Wright [jon::gate.sinica.edu.tw]
 Karine,
Have you tried a slightly older version of Gaussian? Only last week I
 emailed them beacause one of our students has been getting very
 different numbers for Zn in an NBO analysis between g03e1 and g03d2,
 perhaps you are seeing the same problem he has.
I have emailed Gaussian and they are looking into it, perhaps you can
 also email them with your numbers so that if it is really a problem
 with
 g03e1 as our seems to be they know and can check and they might make it
 a higher priority.
 Jon
>  Sent to CCL by: "Karine  Costuas" [kcostuas^-^univ-rennes1.fr]
>  Dear CCLers,
>  We recently bought the GENNBO 5.0. We are currently studying a problem of
>  reactivity in Fe containing complexes with Gaussian. We were surprized to have
>  really different NBO analysis between the NBO 3.1 included in Gaussian 03, and
>  the NBO 5.0 (post-treatment of the file47 obtained with archive keyword).
>  Here are the atomic charges we obtained for two characteristic atoms:
>  Gaussian 03 E.01 (B3LYP, LANL2DZ + pol)
>  NBO 3.1: Fe -1.010, C -0.218
>  NBO 5.0: Fe -0.167, C -0.158
>  Gaussian 03 E.01 (B3LYP, AHLRICHS DZ all electrons) (same atoms, same geometry)
>  NBO 3.1: Fe -0.648, C -0.182
>  NBO 5.0: Fe +0.414, C -0.38
>  We are facing the same problems for ruthenium systems. I must confess I have
>  difficulties to understand what's going on. Does anyone know if NBO 3.1 takes
>  correctly into accound d orbitals. Moreover, does anyone already face such a
>  change in charges just by modifying the atomic basis set?
>  Thanks in advance, sincerely,
>  Karine Costuas
>  CNRS researcher, University of Rennes 1
>  France (research in danger here!)