[DFTB-Plus-User] QUASINANO parameter set

Bálint Aradi balint.aradi at bccms.uni-bremen.de
Fri Oct 18 17:50:18 CEST 2013

Dear Abraham,

> I am curious about a paper that came out this August and its effect on this
> community. The paper, published in ACS, is here:
> http://pubs.acs.org/doi/abs/10.1021/ct4004959 which claims a
> parametrization scheme for the electronic part of the DFTB method that
> covers the periodic table. Agreement with DFT-PBE in electronic structures
> is generally rather good across a broad range of homo-and-heteronuclear
> compounds that may be found in the supporting info of the paper (
> http://pubs.acs.org/doi/suppl/10.1021/ct4004959).

There are a few issues to consider here:

* First of all, the generated sets only contain the electronic part of
the interaction, so that only the spectrum for a given geometry can be
calculated, but no total energy, consequently no forces and no
relaxation. If you need only that, you could use empirical tight
binding, which people routinely do with ten thousands of atoms.

* Second, the generated electronic parameter are fitted on band
structures only. While band structures are a big help in finding those
parameters, we have quite ambivalent experience with sets only fitted on
those. Especially if you wish to create parameters which describes
bulks, surfaces and clusters at the same time up to a reasonable,
according my experience so far, it is not necessary helpful to take the
electronic parameters with the best BS fit.

* Last, but not least, having a good band structure for a monoatomic
system does not necessary yield reasonable band structures for binary
systems, so sometimes may be you want to optimize some of your
parameters on something else.

> In the article it states that the parameter set is available in the ADF
> code. I was wondering if there were any plans to include this
> QUASINANO2013.1 set compatible with the DFTB/DFTB+ license, either in the
> short-term, after more work is done to produce repulsive potentials which
> are forthcoming, or if this is strictly a commercial project. The webpage
> here: http://www.scm.com/quasinano/ probably indicates the latter, but I
> was curious anyway.

Yes, ADF is a commercial code. Theoretically, we could take the values
from the paper and generate electronic SK-files with that, but currently
I am missing manpower for that. But I agree, that despite my scepticism
about the usefulness of those parameters, it could have some sense to
offer them to the DFTB+ community, so we'll definitely think about a
solution in the near future.

  Best regards,


Dr. Bálint Aradi
Bremen Center for Computational Materials Science, University of Bremen

More information about the DFTB-Plus-User mailing list