# Using all built-in basis sets for heavy atoms/molecules

According to this list, Karlsruhe basis sets, the first two (STO-3G and 3-21G) of Pople basis sets and some of the “Others” basis sets are able to be used for calculations involved with fifth and higher period elements. Is there any way to use other basis sets for these elements?

The question you ask in the topic and the question you ask in your post aren’t quite the same, so please clarify the question if this doesn’t answer it:

Try to use a basis set with an element to see if Psi has it. If you get an error, Psi doesn’t have it. But if you can download a basis set in Psi4 format from the basis set exchange, you can have Psi use it as described in the documentation.

Interfacing Psi with the basis set exchange directly, so this manual download is unnecessary, is a future project we’d like to, but we don’t have a definite timeline for it.

You are right. I forgot to mention a thing I came across. Like you said, I tried some basis sets. I tried 6-31G and 6-31G** (or 6-31gss) in a calculation for H2 and it worked. But it didn’t work that kind of heavy elements. So I think psi4 currently doesn’t let me use specific basis sets for heavy elements. I use at least these two basis sets for fifth and higher period elements. Further, it would be better If I could use any basis set in the list.

Do you have a question? It isn’t clear from your last post.

6-31G and 6-31G** are not available for heavy elements. The Pople basis sets are also somewhat obsolete, and contain errors: 6-311G is supposed to be triple zeta but is actually a double zeta basis according to 1989 and 2006 results from H. F. Schaefer.

For an alternative, the Karlsruhe def2 basis sets are available for the almost the whole periodic table (1 <= Z <= 86, from H to Rn), and employ effective core potentials for heavy elements to account for relativistic effects as well as reduce the size of the basis set. [Unfortunately ECP support is a bit flaky in Psi4 at the moment.]

The split-valence basis, def2-SV, is roughly similar in size to 6-31G. Adding polarization, leading to def2-SVP for polarization functions on all atoms or def2-SV(P) if you exclude the hydrogens, you get a basis that is similar in size to 6-31G** or 6-31G*, respectively.

If you want more accurate results, you have the triple-zeta polarized valence def2-TZVP basis set, as well as the quadruple-zeta polarized valence def2-QZVP basis sets. These work really well for HF and DFT. If you want to run post-HF calculations, you need to use an additional set of polarization functions, this gives the def2-TZVPP and def2-QZVPP basis sets. Diffuse functions are available in def2-SVPD, def2-TZVPD, and def2-QZVPD.

1 Like