Mixed Basis Sets for SOC

[LUXiu]

Hi,

I have recently been running SOC calculations for materials with different types of relativistic basis sets for each element. For instance, NdI2, in which iodine has Stuttgart, but Neodymium only has Columbus based on the manual. Here are my two concerns:

1. Is running SOC with different relativistic basis sets for each element likely to produce reliable results?

2. In the CRYSTAL23 manual, page 91:
   "As for the Columbus and Stuttgart-Cologne ECP libraries, molecular basis sets are available online in clickable periodic tables at: http://people.clarkson.edu/~pchristi/reps.html"
   The link is not working. So far, I made the basis set manually from the paper. But it would be useful if users had access to it.

Btw, thanks for running the forum!

[Jacques]

Dear LUXiu,

For 1, one can mix basis sets and pseudopotentials with SOC, as with a regular SCF calculation, with the usual caveats (e.g. small basis set on some elements and large on other elements can lead to basis set superposition errors, etc...).

For 2, Philipp Christiansen's page, containing the valence basis sets, appears to be temporarily down. But, one can also access the basis sets from the paper, as you have correctly done, or from basis set exchange https://www.basissetexchange.org, under the name "CRENBL"/"CRENBS".

Alternatively, for Neodymium, (although this pseudopotential is not part of the internal library), a pseudopotential with spin-orbit operator is available from this paper from the Stuttgart-Cologne group https://pubs.aip.org/aip/jcp/article-abstract/90/3/1730/95906/Energy-adjusted-ab-initio-pseudopotentials-for-the. Associated valence basis sets are available on the CRYSTAL website https://www.crystal.unito.it/Basis_Sets/neodymium.html.

[LUXiu]

Thanks Jacques! These are very informative.
The information for specific basis sets is helpful, I will try them.