Hi all, anyone have any good tricks to share for locating transition states that are really tricky to converge.
I'm running B3LYP with gridsize m5, scfconv 7
Before optimizing the TS I run a full frequency calculation with aoforce, and the correct normal mode is displayed by a single imaginary frequency for a quite reasonable geometry (often from a distance scan)
then I've tried both with statpt default settings, and with hessfrq 1
I've also tried to reduce the step size in optimization by reducing tradius and radmax, also in combination with hessfrq 1.
Still the optimization fails to find a pure transition state, while the optimization sometimes claim to find convergence, a frequency calculation afterwards shows more than one imaginary frequency, and often the first one has the wrong normal mode.
