Seeking advice on improving slow growth calculations for proton transport

[nhi_phan]

Dear everyone,

I am currently using the slow growth approach to calculate the free energy barrier for proton transport.

To begin, I performed simulations using low accuracy criteria. Below is the INCAR I used for the initial run:
PREC = Normal
ISTART = 0
ICHARG = 2
GGA = PE
LREAL = auto
ALGO = Fast
ENCUT = 300
EDIFF = 1e-4
ISYM = 0
ISPIN = 2
NELM = 300
NELMIN = 5
LWAVE = .FALSE.
LCHARG = .FALSE.
IDIPOL = 3
LDIPOL = .TRUE.
IVDW = 11
ISIF = 2

MDALGO = 2
SMASS = 0
TEBEG = 300
NWRITE = 0
NBLOCK = 10
NSW = 5000
IBRION = 0
POTIM = 1
LBLUEOUT = .TRUE.
INCREM = 0.0008
RANDOM_SEED = 301137787 0 0

Then, I increased the accuracy with ENCUT = 450 eV and PREC = Accurate. However, the results from both settings show considerable fluctuations in forces and free energy, and the curves do not appear smooth or reliable.

Here is my results.

test_CV.png

I have also tried decreasing the INCREM value to 0.0004 and 0.0006, but unfortunately, this did not lead to significant improvements in the smoothness or shape of the free energy curves.

Would anyone have suggestions on how I can improve the results?
Since I’m new to this method, I would really appreciate any advice or suggestions.
Is there anything I might be missing in terms of setup or parameters that could help stabilize the results?

Any insights or experiences you can share would be greatly appreciated.
Thank you so much in advance for your help!

Best regards,

[pedro_melo]

Dear nhi_phan,

I am not an expert on proton transport, so could you explain to me what kind of results do you intend to obtain with your calculations?

Regarding the difference between results when using PREC=Normal vs PREC=Accurate and different cutoffs: how are you choosing these values? Do you take them from the ENMAX in your POTCAR? Also, it is hard to compare two different calculations where both the precision and the cutoff are changed. Did you try with the same precision and different cutoff, and also with the same cutoff but with different precision?

Also, I have to ask if you ran a previous ground-state calculation to determine the precision level that ensures that your results are converged, as well as the number of k-points, energy cutoffs, etc.

Kind regards,
Pedro