From chemistry-request@www.ccl.net Thu Feb 25 06:03:36 1999 Received: from harfang.CC.UMontreal.CA (harfang.CC.UMontreal.CA [132.204.2.102]) by www.ccl.net (8.8.3/8.8.6/OSC/CCL 1.0) with ESMTP id GAA28535 Thu, 25 Feb 1999 06:03:36 -0500 (EST) Received: from esi21.ESI.UMontreal.CA (esi21.ESI.UMontreal.CA [132.204.2.91]) by harfang.CC.UMontreal.CA (8.8.8/8.8.8) with ESMTP id GAA23357 for ; Thu, 25 Feb 1999 06:03:38 -0500 (EST) Received: from localhost (proynov@localhost) by esi21.ESI.UMontreal.CA with SMTP id GAA04707 (8.8.7/IDA-1.6 for ); Thu, 25 Feb 1999 06:03:37 -0500 (EST) X-Authentication-Warning: esi21.ESI.UMontreal.CA: proynov owned process doing -bs Date: Thu, 25 Feb 1999 06:03:37 -0500 (EST) From: Proynov Emil X-Sender: proynov@esi21.ESI.UMontreal.CA To: chemistry@www.ccl.net Subject: CCL: Comment on DFT & Hydrogen bonds Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Dear netters, On Thu, 28 Jan 1999, Didier MATHIEU wrote: > Hi, > > Many people asked me to summarize the replies to my question concerning > the performance of DFT, and especially early functionals such as > Becke-Perdew86, with regard to the description of the energetic of > H-bonds. > Clearly, everybody agree that hybrid functionals are better than GGA > functionals, the latter being themselves much better than LDA. Recently, > the team by Salahub has developped a LAP functional which performs well. > Somebody also pointed to the need for extended basis sets for > convergence. ---------------- I would like just to add here some references about the LAP functionals mentioned above: I would recommend examining particularly refs. [4-6] enclosed below, from which it can be envisaged that the hybrid methods may not alwas be a panacea as far as intra-molecular hydrogen bonds concern. Moreover, while the hybrid methods seem to be indeed more demanding with respect to basis set convergence (keeping in mind that in Gaussian the basis set optimization is (post) HF-based not DFT-based ), "pure DFT codes" like deMon-KS3, deMol, NUMOL, ADF, DGauss, etc. use basis sets that are specially optimized and adapted to DFT numerics (which leads to less convergence demands and smaller BSSE errors on average). Adamo and Barone have recently found for example that including p,d, and f functions in the Gaussian G6-311 basis set influences substantially the magnitude of the activation barrier for internal proton transfer in malonaldehyde with B3LYP and other hybrid functionals as implemented in Gaussian 98. I double-checked whether a similar basis set sensitivity exists with the PLAP3 functional within the deMon-KS3 code and I did not find a shift larger than 0.1 kcal/mol for this particular case? I also find that the old good GGA BP is not that bad for systems like H2O dimer and HF dimer as far as energetics concern (it just gives a bit shorter O...H and F...H distances). references: [1]. E.I. Proynov and D.R. Salahub, Phys.Rev.B v.49, 7874 (1994) ; Erratum, ibid. v.57, 12616 (1998). [2] E. Proynov, A. Vela and D. R. Salahub, Chem.Phys.Lett. vol.230, 419 (1994); Erratum, ibid., vol 234, 462 (1995). [3] E.I. Proynov, E. Ruiz, A. Vela, and D.R. Salahub, Int. J. Quant. Chem.(Symp.) vol 29, 61 (1995); [4] E. I. Proynov, S. Sirois and D. R. Salahub, Int. J. Quant. Chem.vol.64, 427, (1997). [5] S. Sirois, E. Proynov, D. Nguyen and D.R. Salahub, J.Chem.Phys. v107, 6770 (1997). [6] H. Guo, S. Sirois, E. Proynov, and D.R. Salahub, in: "Density Functional Theory and Its Applications to Hydrogen Bonded Systems" Theoretical treatment of Hydrogen Bondings}, H; Hadzi ed. John Willey (1997). Emil Proynov +-----------------------------+ | Dr Emil Proynov | | proynov@chimie.umontreal.ca | | eproynov@scf.fundp.ac.be | +-----------------------------+