In a normal DFT electronic structure calculation it is essential to specify the spatial location of every atomic nucleus. The program then uses this information to construct the potential field felt by the electrons. (In an all electron calculation, a Coulomb potential based on the atomic number Z is centered at the position of the nucleus. In a pseudopotential calculation, a modified potential that accounts for the nuclear charge and the core electrons is centered at the position of the nucleus.) There can be no fractional occupancy of a site because either a nucleus is there or it is not. The canonical procedure to handle fractional occupancy is to create a supercell in which a fraction of the nominally equivalent sites is occupied. For example, if the occupancy of site X is 1/N, a supercell is created by combining N unit cells. In the supercell, an atom is placed at one of the X sites, while the other N-1 X sites are left vacant. Obviously this imparts more long-range order that may actually be present in the real material. Another problem is that for occupancies that are not simple whole-number ratios, the supercell would have to be unworkably large. (To model 53% occupancy, one would not create a 1x1x100 supercell of 100 primitive cells and occupy 53 of them. It would be more reasonable to create a 1x1x2 supercell and occupy 1 of the 2 sites, assuming that 50% occupancy is a good approximation of 53%.)

Mr. Kumar

Try this software cif2cell https://sourceforge.net/projects/cif2cell/.

CIF2Cell is a tool to generate the geometrical setup for various electronic structure codes from a CIF (Crystallographic Information Framework) file.

I tried with the cif2cell, burai and winmos (window based gui) to convert the cif to QE input file but all have removed the occupancies, and Ca and Ti atoms and its wyckoff. Thus the remaining species again the same as Bi, Fe and O in the input file. Since I am doping BiFeO3 with 20% Ca and Ti at Bi and Fe site respectively. Thus according to structure analysis part the wycoff of Ca and Ti will be the same as Bi and Fe but the occupancy will be 0.2 (20%) for Ca and Ti and Bi and Fe will have 0.8 (80%). So the problem arises as after converting the CIF file into quantum espresso input file the Ca and Ti not get considered in the input file and its get removed and only remains Bi, Fe and O atomic parameters. So I want to know how to force the quantum espresso input so that it can considered the occupancies and doped atoms. Or How to do the DFT for doped compound where the occupancies of the atoms is not 1 (100%).

I'm not a QE user so I may be wrong, but I think QE handles fractional atomic occupancies by explicitly mixing pseudopotentials in the virtual crystal approximation (VCA) to create a new atomic species, and then in your main input you have a whole one of that.

E.g. if you want 0.2 Ca and 0.8 Bi, you use the "virtual.x" program to make a new chemical species (whose pseudopotential is literally 0.2*Ca psp + 0.8*Bi psp) and then put 1 of those in your crystal (no fractional occupancy, that's included via the pseudopotential).

Hope that helps,

Phil Hasnip

Thanks Philip for your suggestion I will try it surely

In a normal DFT electronic structure calculation it is essential to specify the spatial location of every atomic nucleus. The program then uses this information to construct the potential field felt by the electrons. (In an all electron calculation, a Coulomb potential based on the atomic number Z is centered at the position of the nucleus. In a pseudopotential calculation, a modified potential that accounts for the nuclear charge and the core electrons is centered at the position of the nucleus.) There can be no fractional occupancy of a site because either a nucleus is there or it is not. The canonical procedure to handle fractional occupancy is to create a supercell in which a fraction of the nominally equivalent sites is occupied. For example, if the occupancy of site X is 1/N, a supercell is created by combining N unit cells. In the supercell, an atom is placed at one of the X sites, while the other N-1 X sites are left vacant. Obviously this imparts more long-range order that may actually be present in the real material. Another problem is that for occupancies that are not simple whole-number ratios, the supercell would have to be unworkably large. (To model 53% occupancy, one would not create a 1x1x100 supercell of 100 primitive cells and occupy 53 of them. It would be more reasonable to create a 1x1x2 supercell and occupy 1 of the 2 sites, assuming that 50% occupancy is a good approximation of 53%.)

Thanks Sir, This the exact and best answer of my question that I got here. Thanks Sir Karl Sohlberg.

'glad to help.