If I know crystal structure (lattice parameters, atomic positions) from experiments(rietveld refinement data) and I generate the structure file using this data, should I optimized (computationally) the structure for further calculation of different properties or should I directly carry on with scf step and eliminate relaxation.
You have to optimize it ideally, as it depends on many parameters such as Ecut and kpoints values of pseudo-potentials. Moreover, lattice constants also get the change to have global minima. For example, if you use LDA you will have a lattice constant near to experimental, and if you GGA it will slightly overestimate it. In my opinion, one should optimize structure even if you switch pseud potentials besides cif file or XRD refined parameters.
Thank you @narayan som that clarified most of my doubt although what could be the reason of huge change in optimized lattice parameter for ultrasoft and norm conserving pseudo potential after proper convergence have been achieved?
- Badges
- Science topic
- Similar topics
- Filing