The AMS2020.103 subrelease contains an important bug fix for analytic frequencies with PBE, see below for more details. Other improvements and bug fixes in COSMO-RS, AMSjobs, and other components are included in this subrelease as well, so we suggest updating to this subrelease.
A bug in the analytic Hessian for unrestricted PBE calculations was introduced in ADF2006. Thanks to Marcel Swart for spotting and fixing this bug. It could affect frequencies for spin-polarized calculations, while free energy corrections are usually hardly affected (see further below).
There is no need to worry or investigate further in any of the following cases:
– you used numerical frequencies or used the LibXC keyword or you did not calculate frequencies at all (geometries are not affected)
– you used another xc functional (e.g. LDA, BP, BLYP, mGGAs, hybrids, etc.)
– your molecule was not spin-polarised (even if you used unrestricted)
– the conclusions of your publication did not sensitively depend on the precise value of one or more frequencies
More specifically, the bug only affects analytical frequency calculations ran without libxc, for unrestricted calculations with GGAs using PBE (PBE, OPBE, RPBE, revPBE, and S12g). It has been fixed in the most recent subrelease, AMS2020.103, which is now available on our default download page. For previous versions a workaround is to use numerical frequencies or LibXC.
To find out how this bug may have affected your calculations, we have tested on 27 spin-polarized systems of different types and sizes.
The changes in the Gibbs free energy are typically less than 0.1 kcal / mol and only for Cr-superoxo it was larger than 0.5 kcal/mol
In 20 out of 27 cases the largest frequency change was less than 10 cm-1. The 7 cases with larger frequency shifts:
Cr2 198-169 (broken spin-symmetry anti-ferromagnetic)
C4H9N 229-209 543-506 589-576 (+1 charge, spin on N, NH2 group)
Cr-superoxo 49-22 128-113 165-148 197-162 355-334 700-678 (-117)-(-141) (at TS)
FeO_ana (-130)-(-144) (at TS)
A case of a larger Fe-oxo compound was reported to us where the change in frequency was less than 1 cm-1.
Please contact us at email@example.com in case you have any questions.