I was asked to post the replies I was given for the calculation of Cu^2+ A-tensor,
so here is the last reply I was sent by Erik van Lenthe.

> The ADF program will not produce any reasonable results in this case.
> This is a bug in the ADF program.
> Only if the atom is spherical symmetric ADF can calculate the
> correct result, for example for the neutral Cu (d10s1).
> You can get a more reliable result for the isotropic
> A-tensor if you are using the ADF keyword: allpoints
> But in your case of (Cu)^(2+) (d9) ADF can not
> not calculate the anisotropy correctly.
> At least I do not know of a way to do this.
>
> One of the problems is that you are using atomic symmetry
> (implicitely, because yo do not specify the symmetry).
> The ADF program will then make an average-of-configuration
> calculation by dividing the 4 beta d-electrons
> with fractional occupation numbers over the 5 degenerate
> d-orbitals d_xy, d_xz, d_yz, d_x2-y2, d_z2, each occupied
> with 0.8 electrons. This will result in a spherical density
> and spin-density, and an isotropic A-tensor.
> If you are not using the keyword: allpoints,
> ADF gives the wrong result,in which it looks
> like there is an anisotropy in the A-tensor.
> If there is some relation to the correct anisotropy this is only because of
> some coincidence.
>
> Best regards,
> Erik van Lenthe
Received on 2002-05-24 12:37:41