ABINIT Discussion Forums

Hello to all,

I'm currently trying to run some PAW calculations on graphane using the JTH PAW setups but running into the problem of large overlap between the respective PAW spheres of carbon and hydrogen.

I have compiled atompaw and used the inputs given at the end of the pseudopotential files on the Abinit web page to generate new PAW potentials simply decreasing the radii in hydrogen and carbon by 0.2.

In carbon I changed rpaw, rshape and rveff to 1.3 from 1.5 and correspondingly all four r_c matching radii.

For hydrogen i changed rpaw from 0.9 to 0.7, rshape from 0.8 to 0.6 but left rveff at 0.35. Again I lowered all three r_c matching radii from 0.9 to 0.7.

I have newer generated my own PAW potentials so I wanted to ask whether this is a valid procedure (a first test on the band gap seems fine) and how I could best test the validity of these "new" setups.

As I mentioned I'm really not an expert on PAW and your help would be greatly appreciated.

Thank you and have a pleasant rest o f the day,
Lukas

marsoner wrote:how I could best test the validity of these "new" setups.

Go to http://users.wfu.edu/natalie/papers/pwp ... odictable/ and click on either "Small core datasets" or "Large core datasets". Then pick an element, e.g., Si, and click on "Si data". The reported lattice constant and bulk modulus are from accurate LAPW calculations obtained in the LDA parametrization. If you get the same results with your generated PAW dataset, then it is accurate for most ground state calculations.