From chemistry-request@ccl.net Mon Jun 6 18:32:55 2005 Received: from mail.chem.tamu.edu (mail.chem.tamu.edu [165.91.176.8]) by server.ccl.net (8.13.1/8.13.1) with ESMTP id j56MWoU1003152 for ; Mon, 6 Jun 2005 18:32:50 -0400 Received: by mail.chem.tamu.edu (CommuniGate Pro PIPE 4.2.8) with PIPE id 2948114; Mon, 06 Jun 2005 16:32:50 -0500 Received: from [83.78.17.203] (account singleton[at]mail.chem.tamu.edu) by mail.chem.tamu.edu (CommuniGate Pro WebUser 4.2.8) with HTTP id 2948121 for chemistry[at]ccl.net; Mon, 06 Jun 2005 16:32:35 -0500 From: "Daniel A. Singleton" Subject: CCL:conflict between energy and imaginary frequency To: chemistry[at]ccl.net X-Mailer: CommuniGate Pro WebUser Interface v.4.2.8 Date: Mon, 06 Jun 2005 16:32:35 -0500 Message-ID: In-Reply-To: References: MIME-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1"; format="flowed" Content-Transfer-Encoding: 8bit X-Chemistry_Department-MailScanner-Information: Chemistry_Helpdesk X-Chemistry_Department-MailScanner: Found to be clean X-MailScanner-From: singleton[at]mail.chem.tamu.edu X-Spam-Status: No, score=0.0 required=5.0 tests=none autolearn=failed version=3.0.3 X-Spam-Checker-Version: SpamAssassin 3.0.3 (2005-04-27) on server.ccl.net Dear Karla and Xav, If the energies that Karla is talking about include zero point energy, then this is a common issue. I have seen the potential energy + zpe for a transition structure exceed that of a close-by minimum at least a couple of dozen times (out of several thousand calculations). The issue is that the zpe for the minimum has one extra contributing normal mode compared to the transition structure, and this adds to its zpe. In reality, the zpe is being wrongly calculated by the harmonic approximation in these cases. However, under any circumstances, one should not take too seriously a minimum that is in such an exceedingly shallow well. I don't see how BSSE can cause what was observed. I suppose an integration grid issue might, in principle, though I've never ran into that with an ultrafine grid. Finally, make sure all of the cartesian forces on the 'optimzed' structures are small - in my experience for simple structures, about 1% of optimizations in reduntant internal coordinates lead to structures that are not really fully optimized, even with opt=verytight. Dan Singleton Dear, there are two possible explainations. The first one, as you mentioned is the numerical noise due to the grid of integration. The cure is to use a "better" grid! Is it possible? The second possibility is intramolecular BSSE. Look at the work of Franck Jensen Chem. Phys. Lett. 261 (1996) 633. Yours. ...Xav Pr. Xavier Assfeld Xavier.Assfeld[at]cbt.uhp-nancy.fr Chimie et Biochimie theoriques T: (33) 3 83 68 43 82 Faculte des Sciences F: (33) 3 83 68 43 71 54506 Vandoeuvre, France http://www.lctn.uhp-nancy.fr -----Message d'origine----- De : Computational Chemistry List [mailto:chemistry-request[at]ccl.net]De la part de Karla Tersago Envoye : lundi 6 juin 2005 11:35 A : chemistry[at]ccl.net Objet : CCL:conflict between energy and imaginary frequency Dear CCL'ers, When I perform a geometry optimization, with frequency calculation of a certain molecule with Gaussian03, I have the following problem: Cs symmetry: imaginary frequency of -20 cm-1, so this is transition state structure. C1 symmetry: no imaginary frequencies, so it is a minimum on the potential energy surface. It's lowest frequency is 27 cm-1. I checked the normal modes of both structures and the imaginary frequency of the Cs form and the lowest frequency of the C1 form correspond to the same normal mode. The problem is that energy of the optimized structured shows that the Cs form has a lower energy (it is stabilized by 0.02 kcal/mol) than the C1 form. How can it be that a transition state has a lower energy than a minimum? The calculations were done with B3LYP/6-311+G* opt=verytight int=ultrafine freq As the data conflict, I thought it might be a problem with some thresholds, so I tried to set better thresholds with: Iop(3/27=15) and Iop(3/29=15) I also tried iop(3/32=x) in combination with the above set options, but for x=3,4,5 the scf did not converged (even not if I put MaxCycle on 500). Does anybody have any suggestions on how I can solve this? Best regards, Karla -- ---------------------------------------------------------------------------- - Karla Tersago Structural Chemistry Group, Department of Chemistry University of Antwerp, Universiteitsplein 1, B-2610 Antwerp, Belgium E-mail: karla.tersago[at]ua.ac.be /Phone:+32(03)8202366/Fax:+32(03)8202310 ---------------------------------------------------------------------------- -