ABINIT Discussion Forums

Dear Abinit user,

I want to compute the energy of ion Fe2+ using PAW got from PAW database of Abinit. However I can not converge the acell. This is my input



and output


I guess that my problem comes from using spin polarise. How can I solve this problem and get the energy of Fe2+

Thank you.

Hello,

You have a system which is not neutral so the charge contained in the unit cell interacts with the charge of its periodically repeated images. This is a well known problem concerning the description of isolated charged systems within periodic boundary conditions. You have to increase your cell a lot more and you will for sure get a convergence (it can be seen from your results that going from 10 10 10 to 15 15 15 changes the energy by about 0.1 Ha and going from 15 15 15 to 20 20 20 changes the energy by about 0.05 Ha so there seems to be a beginning of convergence I would say). Another possibility would be to add an a posteriori correction to the energy obtained for each size of your unit cell. For isolated systems with a net charge such as atoms or molecules, this is quite easy. For example you can use the Makov and Payne correction (Makov, G. & Payne, M. C. Periodic boundary conditions in ab initio calculations Phys. Rev. B, American Physical Society, 1995, 51, 4014-4022). The energy with and without the correction are going asymptotically to the same value (hopefully !) but with the correction it should be much faster.

David