CCL: SUMMARY of IRC in Gaussian 98 stops at false minimum

Dear Colleagues,
 I would like to thank for all answers to my question
     "IRC in Gaussian 98 stops at false minimum".
 Final solution was tightening of the convergence criteria for the
 optimization in Gaussian by means of IOP(1/7) and updating force constants
 always after some points.
 The responses I received are below.
 Best regards,
 Petr Toman
 *******************************************************************************
 ORIGINAL MESSAGE:
 Date: Tue, 2 Mar 2004 13:18:40 +0100 (CET)
 From: Petr Toman <ccl.-at-.linus.imcm.cas.cz>
 To: CHEMISTRY.-at-.ccl.net
 Subject: CCL:IRC in Gaussian 98 stops at false minimum
 Dear Colleagues,
 I would like to calculate the reaction path from a transition structure of
 a merocyanine molecule (37 atoms).
 Transition point was verified by a Freq calculation, imaginary frequency
 is 43 cm-1.
 The problem is that IRC calculation stops after very few points reporting
 "Minimum found on this side of the potential".
 However, usual geometry optimization (OPT) starting at this point finds
 the actual minimum far away.
 My question is what is the criterion for a minimum in Gaussian IRC
 calculation?
 Thank you for your help.
 Best regards,
 Petr Toman
 *******************************************************************************
 Date: Tue, 2 Mar 2004 08:03:50 -0500 (EST)
 From: Binh Hoa Bui <bbui.-at-.sunchem.chem.uga.edu>
 To: Petr Toman <ccl.-at-.linus.imcm.cas.cz>
 Subject: Re: CCL:IRC in Gaussian 98 stops at false minimum
 Hi Petr,
   Did you check the nature of the 43 cm-1 vib. mode? The structure may be
 a different local minimum with a rotor as indicated by the mode. You did
 not find the right TS, yet, i think.
 Sincerely,
 binh
 *******************************************************************************
 Date: Tue, 2 Mar 2004 08:25:46 -0600
 From: Phil Hultin <hultin.-at-.cc.UManitoba.CA>
 To: Petr Toman <ccl.-at-.linus.imcm.cas.cz>
 Subject: RE: IRC in Gaussian 98 stops at false minimum
 You may have run into this problem because the force constants need
 updating.  Unless you specify CalcAll, Gaussian will use the estimated
 force
 constants from the TS and extrapolate them to each point along the IRC.
 Often this works ok, but sometimes you will get into a region where this
 extrapolation is not good.
 We did not find it falling into a minimum (it may actually be a real
 minimum
 on your potential surface, by the way, just not the one you want) but we
 had
 IRC jobs fail after various numbers of points.  As I recall, the solution
 was to do the rather slow CalcAll.  It may also help to reduce the step
 size, because you may be jumping too far along the predicted IRC path.
 Dr. Philip G. Hultin
 Associate Department Head and
 Associate Professor of Chemistry
 University of Manitoba
 Winnipeg, MB, Canada R3T 2N2
 (vox) 204-474-9814
 (fax) 204-474-7608
 mailto:hultin.-at-.cc.umanitoba.ca
 http://www.umanitoba.ca/chemistry/
 *******************************************************************************
 Date: Tue, 2 Mar 2004 08:27:36 -0800
 From: Stephen Bowlus <chezbowlus.-at-.goldrush.com>
 To: Petr Toman <ccl.-at-.linus.imcm.cas.cz>
 Subject: Re: CCL:IRC in Gaussian 98 stops at false minimum
 I don't use Gaussian, but it looks like you have located a TS that has
 nothing to do with the reaction you are studying.
 1.  Usually, a true reaction TS will give an imaginary frequency which
 is reasonably large - a couple hundred wave numbers.
 2.  Small imaginary frequencies are "trivial" wrt the reaction, but
 may
 be "real" in describing a motion like the TS for an internal rotation
 of a methyl group (the same types of negative roots will creep into the
 calculations of ground state starting materials and products).
 3.  Does the same thing happen if you run the IRC in the opposite
 direction?
 4.  For refining a TS, I have had to use very gentle methods like
 eigenvector following or Bartel's method.
 sb
 *******************************************************************************
 Date: Thu, 4 Mar 2004 18:34:15 -0500 (EST)
 From: Cust. Service Doug <gaussian.com!csd.-at-.gaussian.com>
 Reply-To: gaussian.com!help%gaussian.com.-at-.gaussian.com
 To: Petr Toman
 <uunet!linus.imcm.cas.cz!ccl%gaussian.com.-at-.uunet.uu.net>
 Subject: Re: CCL:IRC in Gaussian 98 stops at false minimum
   Dr. Toman,
    The stopping criteria is a difficulty in this type of search.  The
 algorithm
 is built around a fixed step so it does not aim to actually zero the full
 gradient.  Rather it looks at variation in the energy and tries to check
 for change in sign on the gradient in case it steps past the minimum.
 Both
 of these can be tricked too easily but nothing better has been developed.
    Try using a smaller step and see if that lets it follow the IRC better.
 Note that it will generally stop far from the minimum but after computing
 the requested number of points along the line, 6 on each side by default.
 --
   Douglas J. Fox
   Technical Support
   Gaussian, Inc.
   help.-at-.gaussian.com
 *******************************************************************************