There is not clear answer to that as this might depend on a variety of factors.
The so called ´ chemical accuracy´ is defined around 50 meV and many pseudopotentials are aimed at that. But it applies to single atom, therefore for bulk structures this might be not true. Other approximations, for example pbe for solids, might improve on that. Hybrid pseudopotentials aims at better overall accuracy but they depend on fit parameters and are considerably more computationally expensive.
My thumb rule in general is that, if two structures have an energy difference below 50 meV they might require further analysis—- for example making convergence parameters more strict or changing pseudopotential to test if the difference remains the same.
I hope this helps,
Roberto
Dear Lt. Dr. Viswanathan Ekambaram,
Actually, if the total energy per cell is too similar, it is better to introduce other parameters to distinguish them than more computational cost energy calculation. In addition to the cohesive energy of the cell, you can use lattice parameters, bulk modulus, young modulus, density, electric constant properties, electronic properties, and others.
Sincerely,
- Badges
- Science topic
- Similar topics
- Filing