The geometry optimization has converged fully and stationary state was found, but the frequency calculations have not fully converged.
I found the reason for this in the following link.
http://gaussian.com/faq3/
Other than resuming from the .chk file of the failed run (as suggested in the above page), Can we eliminate this problem from happening by using CalcFc or calcall keyword?
OK I see. You listed in the screen shot two results, the one has still NO for max displacement, was that the frequency calculation? or the other way around?
as you will have read from the Gaussian.com reference one can have an energy minimum (local or absolute). This does not mean the convergence is good enough for a frequency calculations, even though it seems from you data it is OK.
I have performed really very very many Gaussian calculations in my life, but the last one is a while ago. I seem to remember to have seen this before what you describe. The calculational procedure starts anew when it is asked to do the frequency calculation, and the results for the criteria is not always the same as before.
Therefore, as you wrote the frequency calculation is not fully converged, please have a look how many imaginary frequencies you have (so at the top of the frequency list).
Are you sure you are looking for an absolute minimum, or could it be a local minimum or saddle point?
Yeah, the one which has NO for the displacement in Frequency calculations.
I am looking for the global minimum in order the get the enthalpy of the compound (I am evaluating thermodynamic feasibility of a particular reaction partway)
The compound did not have any imaginary frequencies sir. according to what I see in the output file, the only problem is with the non convergence of the frequencies.
That's not a problem - the FREQ calculation rechecks the convergences, but I don't think it uses all of the previous geometries and gradients and approximate second derivatives as the OPT step does - if it was truly far off, maybe there's an issue with the quality of the optimization, but this only 0.000018 bohr or radians off from the set convergence criterion...not a problem -