As per suggested i optimize molecule in semi emipirical Am1,pm3 set and reoptimize it in dft . And directly optimize from gaussian input file . But optimize geometrical parameter and mulliken charge were different for two cases.Is it okay to have different result?
In the final DFT optimisation with your desired large BS, add the keyword integral=ultrafine . Compare the final total electronic energies of your molecule at the optimised geometries to verify which one is a true absolute energy minimum. In principle, this should display no imaginary frequency. A freq calculation is therefore mandatory to discriminate between your optimised structures.
Adding a bit of technical background to Massimiliano Arca 's already good answer:
if a calculation is run with coarse convergence criteria, the algorithm may determine it has reached a minimum because the derivatives are below a threshold. In the best case you're close to the real minimum, but if the potential surface is very flat, the derivative values may be small over a large range of coordinates and you will have an equally large geometric range at which you calculation will converge. If your structure has "converged" to some place in the surrounding of the minimum, a frequency calculation as suggested by Prof. Arca will reveal this because in a true minimum, you will only get real, positive frequencies, while for a false minimum, you will get at oeast one imaginary frequency.
- Badges
- Science topic