Transition state searching

[avsfan92]

Hello,

I am currently working on finding a transition state for a proposed reaction but I'm having trouble with the transition state optimizer -- while my guess structure has an imaginary mode corresponding to the motion of interest, the optimizer has not been effective. The gradient doesn't approach convergence (but it is not clearly oscillating either), the energies are pretty flat across geometry cycles but steadily decrease. In the end I have a structure more similar to the products than anything else, with energy ~100 kJ/mol lower than my guess structure, and no imaginary mode.

I've tried re-calcuating the hessian at every 5 cycles as well as decreasing the step size (max trust radius = 0.1, maybe this is still too big?) but it doesn't seem to have helped. Is there any other options which might help the optimizer reach the desired point?

Thank you,
Cara

[avsfan92]

I think I have solved my issue -- for anybody else who is having trouble with the transition state optimizer and looking for help on this forum, here is what seemed to have helped in my experience:

1. The optimizer appears to play nicer with Cartesian coordinates than internal coordinates. This is implied in "turbomole tutorial", but I don't think it's stated in the documentation or anywhere else that I can find. In a calculation with an otherwise identical control file, a cartesian optimization converged quite quickly and painlessly.
2. The keyword `keeptmode` is useful. It will ensure that the optimizer always follows the eigenvalue of the same index that you set for itvc.

Cara