wzkchem5 发表于 2023-5-5 01:18
SMD is the best one for single point energies. For geometry optimization, IEFPCM is normally used; ...

Thanks

rpestana94 发表于 2023-5-4 15:05
Thanks, about the implicit solvent model I don't know which I should choose, gaussian shows IEFPCM ...

SMD is the best one for single point energies. For geometry optimization, IEFPCM is normally used; although it is theoretically less accurate than SMD, the influence of the error on the geometries is small, and IEFPCM is more numerically stable, i.e. the geometry convergence is easier and the frequency calculation is less likely to give false imaginary frequencies. I don't know, though, whether this is due to the theories of IEFPCM and SMD, or due to their specific implementation in Gaussian.

sobereva 发表于 2023-5-4 03:36
This combination is appropriate. Using PBE will worse the result.
If you hope to obtain a better re ...

Thanks

wzkchem5 Published on 2023-5-4 03:33
PBE is almost never better than B3LYP for organic molecules. In general, B3LYP is good for organic m ...

Thanks, about the implicit solvent model I don't know which I should choose, gaussian shows IEFPCM, SMD, SCI-PCM and CPCM, there is one that you recommend?

This combination is appropriate. Using PBE will worse the result.
If you hope to obtain a better result, you can use ma-def2-TZVP for single point calculation.

PBE is almost never better than B3LYP for organic molecules. In general, B3LYP is good for organic molecules, while PBE is good for periodic systems (especially metals and alloys). When self-interaction error is non-negligible (for example when there is a delocalized radical or an extremely large conjugating system), B3LYP may fail, but in this case PBE is even worse, since it has not HF exchange, and one should use e.g. M06-2X or wB97XD instead.
Your functional and basis set are fine, but you must include the solvent in the calculations, since this is a heavily charged molecule. In the gas phase, both the basicity of the phosphate and the acidity of the protonated amino group are greatly enhanced, so that the former will abstract a proton from the latter during geometry optimization, and the optimized molecule will no longer reflect the situation in aqueous phase. Another two minor things are: (1) you should do a frequency calculation after the geometry optimization converges, to ensure that there are no imaginary frequencies; (2) I would recommend using 6-311+g(d,p) for the optimization as well; this does not necessarily improve the results very much, but since your molecule is small, it does not slow down your calculation very much either.