QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: Sylvan on July 3, 2018, 14:21

Title: Relaxation of Bismuth
Post by: Sylvan on July 3, 2018, 14:21

So, I'm trying to relax H-passivated bismuth nanowires in ATK, however doing DFT-based relaxation is quite time-consuming. Thus I've been trying to perform some ForceField based calculations, at least to get a first approximation before doing DFT iterations.

There are however no TremoloX/Brenner/EMT parameters available for bismuth and bismuth compounds. Since we can add custom potentials now, I've tried using a naive Lennard-Jones potential for Bi-Bi and Bi-H potentials. This fails and results in a massive spaghetti ball, as seen in the attached picture.

Here is my question then, for people who are more familiar with molecular dynamics: is there some more appropriate pseudopotential that would be suited for quick-but-accurate relaxation of Bi nanowires ?

Title: Re: Relaxation of Bismuth
Post by: Ulrik G. Vej-Hansen on July 3, 2018, 15:57

Have you tried fitting the LJ potentials against DFT results? Not sure how much it will really help though. If the nanowire is constructed from a relaxed bulk Bi crystal, it might make sense to neglect relaxation entirely. This depends on the purpose of your calculations of course.

Title: Re: Relaxation of Bismuth
Post by: Sylvan on July 3, 2018, 16:17

Thanks for the suggestion, I'll try fitting it.
I am trying to characterize surface states in those nanowires, so I do need to perform a relaxation.

Title: Re: Relaxation of Bismuth
Post by: Ulrik G. Vej-Hansen on July 4, 2018, 10:43

In that case, I would recommend that you try to use DFT with some low-accuracy settings, at least as a benchmark for the LJ potential.