From owner-chemistry@ccl.net Sat May 21 02:47:00 2011 From: "Stefan Grimme grimmes%x%uni-muenster.de" To: CCL Subject: CCL: DFT-D for DZP basis sets Message-Id: <-44725-110521015028-2071-4yVHu+cSPQDrW797eniqQw,server.ccl.net> X-Original-From: "Stefan Grimme" Date: Sat, 21 May 2011 01:50:26 -0400 Sent to CCL by: "Stefan Grimme" [grimmes[a]uni-muenster.de] >Anybody know if DFT-D dispersion parameters for def2_SV(P) have been >published, or indeed if its sensible/possible to allow for short range >attraction due to BSSE with the DFT-D approach and small basis sets. Dear Richard, the answer regarding def2-SV(P) is no but we tested in the DFT-D3 paper (see supporting info there) already a re-parametrization for TZVPP to implicitly account for BSSE and thus avoiding CP corrections or intramolecular BSSE. This works in practice and should also be possible for SVP (but with an expected degradation of accuracy because of lower basis set quality). In DFT-D3 the C8/R^8 term is responsible for the short/middle-range behavior and the adjustment of the correction to the short-range behavior of the under-lying functional (and basis set). So what you asked for is possible and makes sense. There is only one scaling parameter (s8) for this that can be adjusted by a trivial fitting to a common set of inter-molecular interactions (e.g. S22 or an extended version of it). One could also adjust the entire correction (two parameters or three for DFT-D3 with Becke-Johnson damping). One final remark regarding this topic: the above approach is practical but theoretically not fully satisfactory because the distance dependence of BSSE and dispersion (exp vs R^-N) are different. In typical applications this will be buried under the typical errors of the functionals but one should keep this in mind. Cheers! Stefan Stefan Grimme grimmes^^uni-muenster.de