My Community

Hello people,

I'm doing some calculations of cyclical molecules on metal surfaces. Relaxations are rather slow because if there is a force on one atom, the displacement has to 'propagate' through the molecule. This process is not very efficient. It's even worse if the molecule needs to rotate. On top of this, the energy gradients for translating and rotating the molecules are quite shallow, so the calculations go on for ages while yielding only modest, slow energy minimization. I haven't even tried doing some ab initio MD, I expect that to be completely hopeless.

What would probably help a lot is if a molecule could be treated as a firm object, i.e. if the molecule translated or rotated as a whole, rather than atoms being displaced individually.

I expect that I'm not the first VASP user to run into this. Has anyone ever implemented something for this in their private version of the code? Or would the VASP group find the idea worth implementing?

Sure, you are not the only one. I have a project of simulating a film of molecules on a surface. This is quite different from yours, I am writing a new rooutine for this project.

However, in your case, you could simply do:
1. A scf with VASP.
2. Get forces acting on the molecule. Write a (separate) routine that can move the molecule as whole. Generate a new POSCAR
3. Rerun VASP again
Repeat the process until you get what you want.

Good luck.

This is good info to know zoowe, I have been wondering how to treat molecules on surfaces correctly for quite some time and this clears it up for me, thanks!

Thanks zoowe. What you say wouldn't require digging into the VASP source. Quite practical, nice tip.