Hi everybody, I am a new user wien2k 14. I am trying to optimize the electronic structure of ternary alloy (with 8 atom in the unit cell) but I am not sure about the good k points number.
Thank you.
It depends on the electronic structure of your system. If there is no partially occupied states (i.e. your system is insulating) then you can use a fairly small number of k-points. On the other hand, if some of the bands are partially occupied (i.e. the system is metallic), then you usually need a larger number of k-points. 'How many more' depends on the size of the system and complexity of the band structure at the Fermi level. In such cases, you may need to do a few test calculations with different sets of k-points to find the optimal k-number. Note that in some cases (mainly cubic systems), shifting the origin of k-mesh can dramatically reduce the number of k-points.
It depends on the electronic structure of your system. If there is no partially occupied states (i.e. your system is insulating) then you can use a fairly small number of k-points. On the other hand, if some of the bands are partially occupied (i.e. the system is metallic), then you usually need a larger number of k-points. 'How many more' depends on the size of the system and complexity of the band structure at the Fermi level. In such cases, you may need to do a few test calculations with different sets of k-points to find the optimal k-number. Note that in some cases (mainly cubic systems), shifting the origin of k-mesh can dramatically reduce the number of k-points.
Dear Amine,
Start to run scf calculations by using 100 k-points. then increase the number of k-points in steps of 50 or 100 and run scf calculations. When the value of band gap will not change after increasing k-points then your k-mesh is said to be optimized. It is necessary to optimize k-points for your crystal because the time duration of different calculations depends on the number of k-points. And your calculation will become more time-costly for more k-points.
There is no variational relationship between kpoint numbers and energy. It means increasing the number of kpoints does not necessarily improve the optimization results. The practical way to find the best grid for your calculation is starting from a very low density grid, runing test calculations, increasing the kpoints numbers gradually, then by plotting energy vs. kpoints and volume vs. kpoints you can find the best grid. When energy and volume converge with respect to kpoints variation, is the optimal point.
In general, the number of k-points required reduces as you go for bigger super-cells or as the lattice constant increases. But, you must check convergence of GAP or energy from the scf files.
Start with a very low value of k-points ( say 100), a medium value ( 1000) and highest (10000).
See the variation in the above values after running scf calculations and then to fine-tune it, pick atleast 5 -6 k-points around the one with highest variation.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Atomic, Molecular and Optical Physics
- Atoms