IRC calculations

[cml]

Hello all,

I am glad to have an IRC module now available in Turbomole (6.5). However, I have not yet figured out how to influence the length of the steps on the IRC path.

I have optimized the transition structure of an organometallic reaction (65 atoms, 1 imag. frequency at 254cm^-1). Using default settings "DRC -i", the geometry does change almost unnotably and the energy change is less than 0.1 mH after the first 10 steps of the IRC run. With other QC programs (e.g. Gaussian, MOPAC), I am used to obtain nice minimum energy paths with IRC. I have played around with the parameter 'initial step length' ("-t XXX"), but with no success.

Has anybody of you been able to obtain reasonable paths with the IRC script of Turbomole 6.5 ?

Best wishes,

Christian

[Arnim]

Hi Christian,

the DRC script integrates the reaction path over time (using the frog module), not over space (as done in other implementations). The only way to make larger or smaller steps is by changing the damping factor for the atomic velocities.

When using the -i (or -d 0) option, the velocities are set to zero in each step. This ensures, that the reaction path lies closely on the IRC, but you will need a lot of steps. Even hundreds might be needed to fully reach the minima.

When the DRC is started without damping, the atoms move faster, thus bigger steps are taken, but you might encounter the problem, that other vibrational modes than the reaction mode are heated up, leading to an overall increase of energy.

There is the additional option to switch on an heuristic controller of the kinetic energy, which tries to adjust the damping factor on the fly. I had the best experience with the call: DRC -d 0.0 -f. However, the script is quite new and I would be very interested to learn about the experience of other users to be able to improve DRC further.

Cheers,

Arnim