Exercise: Comparing Fe Polymorphs
Overview
Now it's time to try some VASP calculations on your own. Our goal in this task will be to study bulk Fe. Specifically, we will use VASP to determine which is more stable: iron in the cubic phase or hexagonal phase. Try to do this yourself!
In order to complete this goal, you will need to:
Create initial guess structures for both phases of bulk Fe.
Relax both the atomic positions and cell shape/volume for each material to ensure they are at their local minimum energy configuration.
Compare their energies to determine which is more stable and by how much.
I strongly encourage you to try this exercise yourself first! But when you get stuck or if you want you to compare your results, check out the walkthrough below.
Tasks
Fetch Initial Structures
Go to the Materials Project and download structures
mp-13 (i.e. cubic Fe) and mp-136 (i.e. hexagonal Fe).
Normally, you would either construct the structure yourself or download the experimental crystal structure from a data repository. Here, we will use the Materials Project as a way to acquaint yourself with this powerful resource. The Materials Project contains DFT-computed data, so it actually answers our question already, but you can confirm for yourself now!
Input File Setup
Make the relevant VASP input files (i.e.
INCAR , KPOINTS , POSCAR , POTCAR , submission script) for both materials, referring to ✍️Input Files for help. Some questions to get you critically thinking about how to approach this are included below:INCAR
What value of ENCUT should you pick? Strictly speaking, you should run a "static" (i.e. single-point calculation) and do a convergence test to determine an appropriate value.
In practice, you can later use often try ~520 eV as a good starting point.
Warning: If you do try to converge ENCUT, do not do so while
LASPH set to True. It won't work.Picking the exchange-correlation functional is one of the most important parts of DFT. Here, let's use PBE to be roughly compatible with the Materials Project. What do you have to set in the INCAR to use PBE?
You want to run a geometry optimization. Which flags should you set to ensure that this is done?
Iron is known to be magnetic in its ground-state. You will need to run the calculation with spin-polarization. Which flags should you set? And what might be an appropriate guess for the initial magnetic moments?
KPOINTS
How many k-points should you use? Strictly speaking, you should run a "static" (i.e. single-point calculation) and do a convergence test to determine an appropriate value.
In practice, you can often try a density of ~100 normalized by volume via the
kpoints_by_vol function as a good starting point.Will you need to use the
vasp_std or vasp_gam executable given your choice of k-points?POSCAR
You'll get this from the Materials Project, as described above. Is it a primitive cell though? It might be worth checking so that you don't waste unecessary compute!
POTCAR
There are multiple types of pseudopotentials for Fe. What does VASP recommend as a good default? What does
Fe vs. Fe_pv vs. Fe_sv mean anyway?Results Analysis
When you successfully ran your calculations:
What are the converged lattice constants for each material?
What are their relative energies? Note: you will have to normalize them on a per-atom basis so that they are directly comparable!
Which phase is more stable?
How do your results compare to the Materials Project?