From oles@kjemi.uio.no Sun May 15 07:54:31 1994 Received: from pat.uio.no for oles@kjemi.uio.no by www.ccl.net (8.6.4/930601.1506) id HAA26048; Sun, 15 May 1994 07:23:31 -0400 From: Received: from ulrik.uio.no by pat.uio.no with local-SMTP (PP) id <18963-0@pat.uio.no>; Sun, 15 May 1994 13:23:28 +0200 Received: by kelvin ; Sun, 15 May 1994 13:23:27 +0200 Date: Sun, 15 May 1994 13:23:27 +0200 Message-Id: <9405151123.AA27952@kelvin> To: chemistry@ccl.net Subject: DFT Failures Cc: oles@kjemi.uio.no Hello everybody, I have got several answers to my question about failures of DFT, and my thanks go to everybody who responded. My original posting was: >Hello everybody, > >I'm giving a seminar on chemical applications of DFT. In picking >cases to present, I face a dilemma: There are enough success stories >to choose from, but for the sake of balance I need to illustrate >some weak spots as well. Rumour has it that transition state energy >barriers in bond-breaking reactions often are underestimated by DFT, >and some of my own results point in the same direction. I would >appreciate any information, references, thoughts, etc. on this or >other shortcomings of DFT. Please reply by Email; I'll summarize >to the net if I'm asked. > > Thanks in advance, > > Ole Swang >From the replies, I am left with the impression that the failures are few and diffuse when it comes to numerical answers. The disappearance of a 8.5 kcal/mol transition state is reported, as are failure in describing van der Waals interactions. The present impossibility of improving the functional in a systematic way, as one can improve correlation descriptions in ab initio methods, is a major drawback. In light of the good numerical performance, I am tempted to say *the* major drawback. Here, then, are the replies. Again, thanks to everybody! +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ >From cramer@maroon.tc.umn.edu Sun May 15 12:57:01 1994 > > I would certainly appreciate receiving a summary of your DFT failures. >This is a bit tricky, of course. Our own experience with DFT has been fairly >positive (if you're still looking for success stories, perhaps you'd be >interested in Chem. Phys. Lett. 218 (1994) 387-394), but the right functional >is critical. Regrettably, there still does not seem to be a systematic way to >ensure that your functional is chemically useful. > >Chris >-- > >Christopher J. Cramer >University of Minnesota >Department of Chemistry >207 Pleasant St. SE >Minneapolis, MN 55455-0431 >(612) 624-0859 >cramer@maroon.tc.umn.edu > +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > > >From ken@ichem.nci.nih.gov Sun May 15 12:57:13 1994 >From: ken@ichem.nci.nih.gov (Kenneth Ray Korzekwa) >Subject: dft problems > >Ole- We have been using gaussian dft for hydrogen abstraction reactions. >I tend to get inappropriate minima similar to that shown in Zeigler's JACS 114, >10890, '92 paper when no nonlocal correcctions were used. I get these for some >reactions using the various gradient corrected methods. I can reproduce his >data when extensive basis sets are used (6-311G**), but other reactions fail, >even with a 6-311++G** basis set. I would very much appreciate hearing about >the answers you receive from the net. Thanks in advance. -Ken Korzekwa >ken@ichem.nci.nih.gov > > > +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > > >From markz@chem.duke.edu Sun May 15 12:57:33 1994 >From: markz@chem.duke.edu (Mark A. Zottola) >Subject: failures in DFT > >Ole, > >As a general rule DFT is truly poor in describing weak molecular interactions >such as hydrogen bonded systems. I have been working on some very unique >hydrogen bonded systems. I have been using standard hartree-fock methods >with and without the use of perturbation theory to account for correlation >effects. I havewhat I feel to be reasonable structures for the system >using HF type methods. When I switch to high level DFT type calculations, >I find that the geometries change radically. >TO flesh this out, I am looking at hydrogen to hydrogen contacts. Typical >hartree-fock methods show the two protons to approach within >1.8 to 1.9 Angstroms (at the mp2/6-31g** level). If I switch to a >corresponding basis set with DFT I find that the H to H contacts shrink to >nearly 1.40 Angstroms. Whether or not you believe in light atom to light atom >contacts, the 1.4 value is clearly unreasonable for whatever camp you're in. >I know this is sketchy, and I'd be willing to answer questions and send you >a preprint of our work (which does NOT contain any mention of the DFT results...). >Good luck in getting "failiures"... >-mark > ********************************************** > * * > * Mark A. Zottola * > * markz@chem.duke.edu * > * Department of Chemistry * > * Duke University * > * * > * 'The fault, dear Brutus, lies not with * > * ourselves, rather within our CPU's...' * > * (with due apologies to W.S.) * > * * > ********************************************** > +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > > >From suter@physik.unizh.ch Sun May 15 12:57:37 1994 >From: Hans-Ueli Suter >Subject: DFT against the thruth > >Dear Dr. Swang, > >you have asked about some of the failure of DFT >theory. Some example which shows this quite >good is the DFT calculation of the >isotropic hfcc or Fermi contact term. However, >this is also problematic for standard ab initio >methods. Okay, look first at the atoms >(fortunate there exist a paper: >J. Kong et al., Chem.Phys.Lett. 217 (1994) 25) >There in table 1 the sign and factors away >>from experimental and MR--CI results may be >seen. I once calculated similar things >but I was not so sure to publish it. >The other examples are so that it >heavenly depends on the electronic structure >of the system, well the direct contribution >is overestimated against an underestimation of the` >indirect contribution (direct : value of the >unoccupied orbital). However, in these >cases DFT is still a method which is superior >to UHF and UMP2. I plan to publish some >comparision to DFT and MR-CI, CCSD for the >isotropic hfcc. I can send you a preprint if >you wish. Hmm, may be that one should cite >that as H.U. Suter et al., to be rejected. > > greetings, Hans Ulrich Suter > +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > > >From WILLSD@conrad.appstate.edu Sun May 15 12:57:52 1994 > >In regard to your recent post to CCL on DFT, I have a couple of items: >1) Please do summarize your responses to CCL. >2) I recently attended a gaussian 92 workshop in North Carolina, USA. >AS part of this John Pople spoke about the DFT methods that are >available in g92 (and other codes also). While he did relate several >sucess stories, his major complaint was that unlike Hartree-Fock and >the various post HF methods, (mpx, ci, qci, cc....) if the DFT >methods (gradient corrected exchange and correlation) do not give >the accuracy that you need, there are no systematic methods available >to improve the accuracy. With post HF methods you know what to do >to improve things, even though you may not be able to afford the >required computing. >Looking forward to your summary, >Steve Williams > +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > >From jdurant@mephisto.ca.sandia.gov Sun May 15 12:58:01 1994 >From: jdurant@mephisto.ca.sandia.gov (Joe Durant) >Subject: Re: CCL:DFT: Failure stories > >Hello! > >As soon as I get more time... I have been looking at some transition states >using DFT, principly using the BLYP functional. Using that functional the >surfaces tend to be too attractive, so that transition states are located >too far out, and barriers are too low. The worst case I have found is >for O-HCl, where the ~8.5 kcal/mole barrier goes away, and the surface >becomes overall attractive. Interestingly enough, you can see that the >reaction "occurs" at roughly the right O-H distance; the surface falls >smoothly til O-H~1.2 A, at which point the HCl bond ruptures, and the >surface drops off rapidly. I have started looking at TS's using the >Becke "Half and Half" + LYP functional... it looks better, but I haven't >had time to do enough to say more. > >Needless to say I am quite interested in a summary (or just a forwarding) >of the replies you get on this topic. > >Joe >- -- > >###################################################################### ># Joe Durant voice: (510) 294-3343 # ># Mail Stop 9055 FAX: (510) 294-2276 # ># Sandia National Laboratories jdurant@ca.sandia.gov # ># Livermore, CA 94551 # >###################################################################### > +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > >From jdurant@mephisto.ca.sandia.gov Sun May 15 12:58:21 1994 >From: jdurant@mephisto.ca.sandia.gov (Joe Durant) >Subject: Re: CCL:DFT: Failure stories > >Hello Ole, > >> what you wrote is just the kind of information I wanted! > >Great! > >As to your questions: > >1) The 8.5 kcal/mole barrier is from work by Koizumi, Schatz and Gordon >(JCP 95, 6421, 1991) where they tweaked an ab initio PES to reproduce the >experimental rate at 298 K. The experimental Ea is somewhere in the >6.5 kcal/mole range. I haven't done the conversion from Ea to E0... I >had originally thought that the reaction would have alot of tunneling, but >the Arrhenius plot seems quite reasonable. Oh, yes... the 8.5 kcal/mole >number is the classical barrier, the ZPE corrected value is 6.6 kcal/mole. > >2) Koizumi, et al used MP2/6-31G** to create their surface. I used G2 and >G2Q (G2 with QCISD/6-311G** geometries and frequencies) (JCP 98, 8031, 1993) >Of course the G2 runs have HF, MP2, MP4 and QCISD(T) pieces in them; if you >want output files I can mail them to you. > >3) Yes, the overall reaction is as you assumed: O + HCl = OH + Cl > >4) The transition state at HF is linear, but as soon as correlation is >turned on it bends. It is also bent in DFT. > >Hope this helps. > >Joe > >- -- > >###################################################################### ># Joe Durant voice: (510) 294-3343 # ># Mail Stop 9055 FAX: (510) 294-2276 # ># Sandia National Laboratories jdurant@ca.sandia.gov # ># Livermore, CA 94551 # >###################################################################### From ipcakc@vigyan.iisc.ernet.in Sun May 15 11:54:34 1994 Received: from relay1.UU.NET for ipcakc@vigyan.iisc.ernet.in by www.ccl.net (8.6.4/930601.1506) id LAA26614; Sun, 15 May 1994 11:13:39 -0400 From: Received: from iisc.ernet.in (via iisc.iisc.ernet.in) by relay1.UU.NET with SMTP (5.61/UUNET-internet-primary) id AAwpzw06963; Sun, 15 May 94 11:13:33 -0400 Received: by iisc.ernet.in (ERNET-IISc/AT&T UNIX SYS V.3.2) id AA15280; Sun, 15 May 94 20:33:54 EDT Received: by vigyan.iisc.ernet.in (smail2.3) id AA24865; 15 May 94 20:24:46 EST (Sun) To: CHEMISTRY@ccl.net Subject: two A atom Date: 15 May 94 20:24:46 EST (Sun) Message-Id: <9405152024.AA24865@vigyan.iisc.ernet.in> This is in response to the Anthony Scott's question---- When two atoms A are kept at a large distance the vib. freq. is is zero, and so also its force constant. The RT term per vib. deg. of freedom arises due to the presence of two sqare terms in the hamiltonian of the oscillator as P**2/2m + 1/2 K X**2 Since in this case K=0, only one term will exist. So according to principle of equipartition of energy there will be only 1/2RT for this vibration. Hence the total enthalpy will be equal to 6/2RT which is same as that of the two separated atoms. It can be tested with more than two atomic sysyem keeping then in linear or non linear form. Debasis Sengupta Indian Institute of Science, Bangalore ipcakc@ipc.iisc.ernet.in From mailer-daemon@csv.warwick.ac.uk Sat May 14 09:54:16 1994 Received: from violet.csv.warwick.ac.uk for mailer-daemon@csv.warwick.ac.uk by www.ccl.net (8.6.4/930601.1506) id JAA21715; Sat, 14 May 1994 09:27:26 -0400 Message-Id: <24831.199405141327@violet.csv.warwick.ac.uk> Received: from chem.warwick.ac.uk by violet.csv.warwick.ac.uk with SMTP id OAA24831; Sat, 14 May 1994 14:27:22 +0100 Received: by chemhp.chem.warwick.ac.uk (1.37.109.4/16.2) id AA10007; Sat, 14 May 94 14:36:46 +0100 From: Craig Wilson Subject: Molecular viewing software To: chemistry@ccl.net Date: Sat, 14 May 94 14:36:42 BST Reply-To: msrge@csv.warwick.ac.uk Mailer: Elm [revision: 70.85] Dear CCL Netters, I would very much like to obtain a good piece of software for displaying molecular structures from my ab initio calculations. Ideally, the program should be able to display structures in ball-and-stick fashion, with options for appending the bond lengths and angles, as well as atomic symbols to the structure, so that a self-explanatory picture is created. The program should be able to create postscript files as well. I have access to UNIX and X-Windows on HP 9000, Sun OS3 or OS4 and Silicon Graphics platforms, and access to MS-Windows on a 486PC. I would be very interested to hear your opinions on what are the best commercial and shareware packages in the area. I will summarize a list of responses to the net, of course. Thanks very much in anticipation of your help, Craig Wilson Postgraduate Student at the University of Warwick, COVENTRY, UK. From eslone@mason1.gmu.edu Sun May 15 19:54:40 1994 Received: from mason1.gmu.edu for eslone@mason1.gmu.edu by www.ccl.net (8.6.4/930601.1506) id TAA29420; Sun, 15 May 1994 19:07:35 -0400 Received: by mason1.gmu.edu (5.65/DEC-Ultrix/4.3/GMUv3) id AA00593; Sun, 15 May 1994 19:07:33 -0400 Message-Id: <9405152307.AA00593@mason1.gmu.edu> Subject: Upload of Molgen To: chemistry@ccl.net Date: Sun, 15 May 1994 19:07:33 -0400 (EDT) From: "J. Eric Slone" X-Mailer: ELM [version 2.4 PL23] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 762 I have just uploaded a runnable version of Molgen into the ccl.net archives. It is currently in /incoming/molgen, but should be moved shortly to /pub/chemistry/software/MOLGEN-demo. Eric ________________________________________________________________________________ J. Eric Slone Scientific Consulting Services Serving Government & Industry Since 1982 Internet: eslone@mason1.gmu.edu Compuserve: 73757,2776 "True science teaches, above all, to Fax: (703) 751-6639 doubt, and to be ignorant." Voice: (703) 461-7078 Miguel do Unamuno ________________________________________________________________________________ From sgsrini@u.washington.edu Sun May 15 20:54:42 1994 Received: from hardy.u.washington.edu for sgsrini@u.washington.edu by www.ccl.net (8.6.4/930601.1506) id UAA29938; Sun, 15 May 1994 20:21:25 -0400 Received: by hardy.u.washington.edu (5.65+UW94.4/UW-NDC Revision: 2.30 ) id AA00378; Sun, 15 May 94 17:21:24 -0700 X-Sender: sgsrini@hardy.u.washington.edu Date: Sun, 15 May 1994 17:21:23 -0700 (PDT) From: Srinivasan Srivilliputhur Subject: CCL: FREE ENERGY Computations in Magnetic Materials (FERRITES). To: chemistry@ccl.net Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Hi Friends- I'd like to know the internet site(s) containing the source code for the Free-Energy Computations in Ferrite Based Magnetic Materials. My ex-advisor in India wants this info. I'd very much appreciate if you could e-mail me this info. Thanks a lot, -Srini ------------------------------------------------------------------------------- S. G. Srinivasan email : sgsrini@u.washington.edu Graduate Student Dept. Of Mater. Sci. & Engg. University Of Washington Phone : (206)-633-1757 (H) Seattle, WA 98195 (206)-685-3851 (W) ------------------------------------------------------------------------------- From grechtst@pepvax.pepperdine.edu Sun May 15 22:54:42 1994 Received: from pepvax.pepperdine.edu for grechtst@pepvax.pepperdine.edu by www.ccl.net (8.6.4/930601.1506) id WAA00777; Sun, 15 May 1994 22:42:49 -0400 Received: by pepvax.pepperdine.edu (5.65c/IDA-1.4.4/Pep-5.1) id AA23296 for chemistry@ccl.net; Sun, 15 May 1994 19:44:56 -0700 From: Gregory Rechtsteiner Message-Id: <199405160244.AA23296@pepvax.pepperdine.edu> Subject: A Research Question To: chemistry@ccl.net Date: Sun, 15 May 94 19:44:56 PDT X-Mailer: ELM [version 2.3 PL11] Hello: I recently posted a similar document to sci.chem and sci.chem.organometals. Sorry for any duplication. My question is a follows: Has anyone synthesized the complex Co(II)-4-4'-4''-4'''-tetrasulfophthalo- cyanine? I tried following the synthesis reported by Weber et. al. in Inorg. Chem. 1965, however, the synthesis did not work. I am not sure why I could not synthesize the compound, but I do think that some of the problems came >from errors due to the vagueness of the synthesis steps concering the reprecipiation process. Can anyone give me a more detailed desription of the synthesis? Or can anyone give me a different method for the synthesis? Thank you, gregory rechtsteiner -- -------------------------------------------------------------------------------- Gregory A. Rechtsteiner Pepperdine University Research Assistant 24255 PCH # 572 Fax: 310.456.4314 (work) Malibu, CA. 90263 grechtst@pepvax.pepperdine.edu / grechtst@netcom.com --------------------------------------------------------------------------------