Volume Optimization for tetragonal, rhombohedral and orthorhombic lattices

[poorvasingh]

Dear VASP admin and VASP users,
I intend to do volume optimization for different lattices e.g. rhombohedral, tetragonal and orthorhombic etc. I have read VASP manual and various papers on how to do the same, but not very confident about the method. I searched through various posts in forum, but still not clear.
I want to plot energy versus volume from which I wish to evaluate enthalpy vs pressure graph (fitting the curve to Birch Murnaghan equation of states):

Kindly suggest the correct method. I could understand the following after reading various papers:

1. For tetragonal or hexagonal (with a, c as two internal parameters to be optimized)
a.Find optimum c/a ratio (find the c/a from experiments, vary c/a along + and - , and at each c/a value, relax the structure i.e. use ISIF=2, followed by ISIF=7.

b. Vary both c, a values to generate different sets of volume such that for each volume data set the c/a remains fixed at the c/a ratio obtained from step a. above. At each date set (of volume), relax the structure by using ISIF =4 and and obtain the energy data point.

c. Now fit the curve (Volume vs energy obtained from step b.) to Birch Murnaghan equation of states.

2. Rhombohedral (a and aplha as internal parameters)

a. Find optimum aplha (vary aplha, such that at each alpha data point, parameter a remains fixed (may be taken from experiment). Then at each alpha data point, relax structure by using ISIF=2, followed by ISIF=7.

b. Use this optimized alpha and generate different sets of parameter a so that we generate different sets of volume. Then at each data set of volume, relax the structure using ISIF =4 (i.e. cell shape and ions) and obtained the energy data point.

c. Now fit the curve (Volume vs energy obtained from step b.) to Birch Murnaghan equation of states.

3. Since, I am not clear about tetragonal and rhombohdral, I am highly doubtful about the method suitable for vol optimization of orthorhombic latttice (since it involves 3 internal parameters, a, b, c).

Can someone please help me in this regard.
Thank You.

[admin]

I would suggest to
1) first do an automatic fit (ISIF=3, sufficiently high ENCUT, PREC=Accurate, EDIFF=1e-6 or even higher, EDIFFG = -0.01), in steps
of 3-5 ionic steps (NSW) each, please delete the WAVECAR files after each run
2) THEN you can do a search around the minima obtained by the automatic fit by varying the free parameters slightly