natural charge TURBOMOLE vs Gaussian16

[t-iwasa]

Dear all,

I'm calculating [IrAu12]3+ cluster under Ih symmetry, where Ir is at the center of Icosahedral Au12. I use B3LYP_Gaussian/def-SV(P) level to optimize and get natural charges, but there are large difference between TURBOMOLE 7.8 result and Gaussian16 result.

Turbomole generate the charge of Ir about -1.5, while G16 generates -3.0. Mulliken population analysis shows similar results in which Ir has negative charge of about -4.5.

I tested with and without RI approx and def vs def2 basis sets, but still Turbomole gives the Ir charge of about -1.5.

I'm just wondering how should I take these differences. Is it known trend or any chance of bug in either program?

Thanks in advance.

[t-iwasa]

Dear all,
the problem solved by using the lates version of NBO7 in Gaussian16 as pop=nbo7.

The above result of larger negative charge was calculated using pop=nbo, in which NBO version 3 is used in the Gaussian.

Best,
Takeshi