Molcas Forum

lijingbai2009

Member

Registered: 2019-08-28

Posts: 29

Read Hessian from previous frequency calculation for IRC

Hi all,

I would like to ask if it is possible to read the Hessian matrix for IRC calculation from a previous frequency calculation (use &mckinley). The problem is when I performed a TS optimization, I can use the RunFile, which has the updated Hessian matrix and rename it as OldRun for the IRC calculation, but if I add &mckinley to calculate the exact Hessian matrix after the TS optimization, the program said no IRC vector in the OldRun.

I think it would be better if someone can tell how to prepare the reaction vectors based on the frequency calculation. I saw the vibrational modes saved in the freq.molden file, but I do not know how to convert it to the reaction vector, such as the unit conversion.

Another question is can I do IRC for higher-order saddle point? Because I wanted to use an exact Hessian matrix to guide the path on a selected vibrational mode among three imaginary frequencies of a Franck-Condon structure. I read somewhere in the forum that one can manually displace the structure as the vibrational mode suggested and do MEP but it would be really helpful if I can select the mode and direction of the MEP.

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: Read Hessian from previous frequency calculation for IRC

The easiest is to write down the reaction vector in the &SLAPAF input ("REACtionVector" keyword). The units are irrelevant, the format is free, and the accuracy is not much significant either (except that it should not be orthogonal to the "real" vector). Just take the desired normal mode from the output or molden file and write the numbers, e.g.:

  Reaction vector
   -0.90712  0.17215 -0.10279
    0.03250  0.00382 -0.00224
   -0.05880 -0.01740  0.01026
   -0.00494 -0.00997  0.00598
    0.03815  0.00388 -0.00230

You can start an IRC from any point, but you should understand what it does: it simply "perturbs" the initial geometry along the given reaction vector and runs a MEP, and then perturbs it in the opposite direction and runs another MEP. It does not "follow" a particular mode, although depending on the surface features, accuracy, symmetry, etc., you might be able to keep the structure on a "ridge" for some time.