When optimizing by DFT medium-sized organic molecules containing aromatic CF3 groups (for example, adducts involving takemoto's catalyst), the frequency calculation on optimized structures have always small negative frequencies involving CF3 groups (small rotations), no matter which program (gaussian or orca), which functional (b3lyp, m06-2x, pbeh-3c or b97-3c), which grid (ultrafine, grid6), which base (double or triple-zeta, with or without diffuse or polarization terms) or convergence criteria. The only solution is to manually rotate the groups a bit and launch another optimization (with smaller step) and another frequency calculation, over and over (sometimes I finally find a true minimum after 3 or 4 optimization/frequency cycles, sometimes after 7 or 8). This procedure is quite time-consuming, considering the number of frequency calculations involved.
Anyone has ever faced a similar problem? Are there any tips or tricks able to solve (or mitigate) this problem?
thanks
