Are you check the broadening parameter for PDOS?

Hadi Jabbar Alagealy Please I used the default 0.0 but in my scf and nscf I used 0.01. Could be the reason?

Dennis Boakye I think the k-mesh you need to select carefully. Second, the step size of energy in the estimation of the density of states is also important. Note that k-mesh fo SCF and DOS need to take as much as you can take (Energy conversed values of k-mesh), and dense enough total number of k-points in the irreducible Brillouin zone is required to set while plotting the band dispersion. Could you let us know the DFT code you are using (pseudopotential or full potential), the broadening parameter is very crucial, especially, the density of states near the band edge? So cross-check with the k-mesh, branding parameter. In addition to that, I hope the energy cut-off is taken sufficiently good enough to avoid the artefact results which might be possible to obtain from DFT codes. Good luck.

Saurabh Singh I selected a list of k-points in the G-M-K-G region in the first brillouin zone. I took two hundred list of k-points using Xcrysden. For the pseudopotential I am using Ultrasoft pseudopotentials. The broadening parameter is 0.01 (degauss). The converged cutoff energy is 50 Ry. A similar calculation was done for Vanadium doping as shown in the attached figure.

Dennis Boakye Thank you for the detailed information. Now I understand the situation much better way. By taking a look near chemical potential, we can say the system is half-metallic, which is mainly confirmed by the band dispersion plot (with 200 total k-point). Now the issue still remains that DOS for down spin have the finite density of states. So this trouble could be possible due to the broadening parameter. No need to worry much, as for the case of V doped the DOS and band dispersion seems to be consistent with each other. isn't it. Now lets discuss on your privious issue where you are getting the finite density of states for down spin while band dispersion corresponding to this minority channel is clearly showing the semiconductor nature. i would recommend two options (i) play with braodneing parameter. (ii) since Cu, V, and other transition metal elements have d-orbital electron, in which onsite Coulomb interaction between d orbital electrons need to be taken care. You can try with DFT + U, where again you need to play with U parameter. But, this is just a fitting, not solving the descrepancy between band dispersion and DOS plot. So trying (i) option is worth. Finally, it would be nice to try with all electron based code. I am not aware about how big the suppercell you have, and how many number of atoms. If you wish, and if it is manageable, i can try with Wien2k code. You can take my support to figure out the problem, untill some expert person let you have the easy and smart solution. Good luck

Saurabh Singh Thank you for your response. I really do appreciate it. I have read about DFT+U so will figure it out. How do I play with the broadening parameter, please? Do you mean changing it?

Dennis Boakye Yes, vary the U value. I am not pretty much sure it will work. However, you can try. To save time just make one or two calculations with U = 3eV and 6 eV, a rough step to see anything work and make sense. Related to the broadening parameter, yes, the magnitude can be changed while making DOS calculations. I suggest you check the recommended range of broadening parameters for the case of semiconductors, if the things work within that it's fine. Else, it will be again a type of fitting to find the result of your interest by some parameter. I hope you will find some way to solve the issue. Please let us know also, it will save others time and computational resources, especially, dealing with supercell calculations. Good luck

the discussion of Saurabh Singh is very best

Hadi Jabbar Alagealy Saurabh Singh Thank you very much. Would do just that