SurfaceConfiguration not converged properly

[sukhito teh]

Dear developers and users,
I am trying to calculate work function of TiN using green function method, however the calculation failed to converge nicely at 111 surface, i.e. energy is converged but dQ at central region is still large (~0.15).  The work function obtained seems reasonable, but for other surface calculation, I often obtained dQ that is smaller than 0.01, so I am not sure if I should concern with such value of dQ.

I had also tried different approaches including increases contour points, higher mesh, reducing number of history steps, increase electrode length, central length , but the dQ remained large.

Thank you for your attention

[Anders Blom]

There does not seem to be anything immediately wrong with the calculation, and it sounds like you have carefully checked the important parameters. In particular it's reassuring that the surface atoms converge to the same relative electron distribution (about 0.284e transferred between Ti and N) as in the bulk electrode. So I would say your result is likely a reflection of the fact that the 111 surface is quite different from 100, which is also why the work function is a lot larger.

Now, there are still things to verify. The 0.14e on the surface is located in the ghost atoms, so perhaps you need a few more layers of ghost atoms, esp. if the ghost atoms are not as populated in the 100 case. Also, it is quite crucial whether you have N, Ti, or mixed termination of the surface which should be covered by the investigations, incl. relaxation of the surface atoms (if you didn't already).

I suppose you already studied the slab calculations of TiN, like http://dx.doi.org/10.1109/ACCESS.2020.3017726. I think you have a chance of showing a better approach!

[sukhito teh]

Thank you for your reply! I l tried the calculation with more layers of ghost atoms and the result remains similar, anyway I am glad to hear that such results doesn't necessary  means  unreliable calculations. I did check and compare results with different terminations, with/without surface relaxation, as you said, they can indeed make a big difference.