PC GAMESS/Firefly-related discussion club

Learn how to ask questions correctly

Andrei V Scherbinin
scherb@classic.chem.msu.su

>Today, I found a strange problem. ZEFF(1) values are not read. The attached
>file is an example. In this input file, ZEFF(1)=1.0, 1435.0 . But, in the output
> file, the following message appears.

>      EFFECTIVE NUCLEAR CHARGES ARE
>         1     1.000   2    35.000
>Why?

The problem seems to arise because you write both statements ZEFTYP=3-21G and ZEFF(1)=1.0, 1435.0
in the $TRANST group. When I omit the first statement according to

$TRANST ZEFF(1)=1.0, 1435.0  MS=0 $END

the input is correct:

    EFFECTIVE NUCLEAR CHARGES ARE
   1     1.000   2  1435.000


>In addition, the Zeff value of Br, 1435, is from Koseki et al. paper (1995).
>This value was determinded for SBK(d,p) basis function. But this basis function
> can be not used in the job of METHOD=ZEFF, since this basis function is not
> in SPD type. Hence, I use DZV in the input file. Thus, DZV is inconsistent
> with Zeff=1435 for Br. Is this OK?

Yes, Zeff=1435 is inconsistent with DZV.

In ECP calculations, so huge value of ZEFF is required to compensate for the
lack of core electrons. In fact, core electrons mainly contribute to the relativism,
not the valence ones, because the closer an electron to the nucleus, the bigger
is its velocity! (recall also 1/r^3 dependence of the SO operator).

However, DZV is an all-electron basis set, with all core shells present,
hence nothing is to compensate for, and the ZEFF value which is optimal for DZV
basis should not differ significantly from the true nuclear charge value 35.

Btw, you can manually split the L-shells in the SBK basis set so as to input
S- and P-functions separately. Then there will be no problem to use them in the
SO calculation.



Regards,
Andrei