High entropy alloys energy simulation

[naoya_taniguchi]

Dear VASP experts,

I am working on understanding high entropy alloys as hydrogen storage materials in VASP calculations.
My research target is Mg0.10Ti0.30V0.25Zr0.10Nb0.25. While conducting this research, I encountered cases where energy calculations could or could not be performed depending on the amount of hydrogen.
I would like to understand the reason for this difference. Could you let me know if you find the cause?

I would also appreciate it if you could let me know whether ISPIN=2 should be set for this calculation.

I will attach the calculation files below.

traceH30.zip

traceH60.zip

Naoya

[pedro_melo]

Dear naoya_taniguchi,

Your traceH30 structure seems hard to converge, but it looks like that the free energy is slowly reaching the point of convergence. Did you try following the suggestion written in the OUTCAR and restart your calculation using a lower EDIFF or copying the CONTCAR file into the POSCAR?

Regarding your question on whether or not ISPIN should be set to 2, I am not an expert on such systems. Since you seem to have an already converged structure (traceH60) I would suggest that you compute its bandstructure with ISPIN=1 and then ISPIN=2 and check if there are significant changes to the bandgap (if the system has one) and the overall shape of the bands.

Kind regards,
Pedro

[naoya_taniguchi]

Dear Pedro

Thank you for reply.

Following the suggestion in the OUTCAR, I copied the CONTCAR to POSCAR and tried recalculating. As you can see from the attached file, the calculation barely progressed before it stopped.
This kind of situation often occurs during other calculations as well. If there is any way to prevent this, please let me know.
For now, I'll try running the same calculation again, and I'll share the results as soon as it's finished.

Kind regards,
Naoya

traceH30(CONTCAR→POSCAR).zip

[pedro_melo]

Dear Naoya,

Your new OUTCAR does not show a reason for why the calculation stopped. There are several reasons as to why this can happen on systems with several atoms, so I cannot advise here without more information. Do you have the standard output for this job?

From my past experience I would advise to run first a non-polarised calculation (ISPIN=1) with softer parameters for the convergence criteria. Only afterwards I would go for ISPIN=2, which should start from a better structure already.

On another note: You have a warning in your OUTCAR that you do not set the tag MAGMOM. Do you happen to have any information on the magnetic behaviour of the system? By default VASP sets the starting magnetic moments of each atom to 1, but this might be inconsistent with your system. If the starting point is really far away from the ground state, then it will be very hard to converge your calculation.

Kind regards,
Pedro

[pedro_melo]

Dear Naoya,

Another point: it might also help to run an initial calculation with ALGO=FAST. While the resulting structure should be checked against a more precise calculation, it can provide a better starting point for subsequent jobs.

Kind regards,
Pedro

[naoya_taniguchi]

Dear Pedro

I apologize for the delayed response.
After repeating the same calculation about four times, it finally converged. I'll share the calculation results from that time.

I have now changed the research focus to MgTiVCrNb and am performing similar calculations. In this calculation, I will first run it with ISPIN=1. If it does not converge, I plan to recalculate with ISPIN=2 and ICHARG=1. I will let you know once the results are available.

Kind regards,
Naoya

H30-convergence.zip