Question About Cluster Expansion and Delithiated Structures

icet
1

Hello,

I have been running trials with ICET and was wondering how to simulate a delithiated structure and build a convex hull. For example, in the latest tutorial, they model the MOVC compound with vacancies by using Be as a placeholder for vacancies.

My questions are:

  1. To generate a database with a set of structures for a lithium compound, let’s say LixCo2, where I vary x from 0 to 1, how should I proceed? Should I simply remove lithium atoms, or should I use another atomic symbol to represent vacancies? I am unsure about the correct approach.

  2. After generating the structures, should I relax each cell and store the energy in an ASE database? Would the reference structure be LiXCo2 in this case?

Thank you very much for your help.

2
  1. When using cluster expansion w icet you need to have a dummy atom-type on the lattice-sites where vacancies are.
  2. Yes typically one uses relaxed energies when training cluster-expansions. When running the relaxation you do need to remove the dummy atom type from your structures. You dont have to store the output in an ase database, but it might be convenient. I dont know what you mean with reference structure.

I’d recommend looking and following the tutorials on icet homepage Tutorial — icet documentation