# How to extract lattice constants a and c from relaxed hexagonal structure using VASP?

Taking a bulk WS$$_2$$ (bilayer) structure and relaxing it with ISIF=3 and IVDW=10 (DFT-D2 method) using the VASP code. Then opening this structure with VESTA software. Are the shown $$a$$ and $$c$$ lattice constants the optimized ones? Or is there something else that must be done to extract them? Here is the CONTCAR of the relaxed structure:

W2S4-D2
1.00000000000000
3.1894907296110424   -0.0000000015017362    0.0000000000000000
-1.5947448663004180    2.7621798956603310   -0.0000000000000000
-0.0000000000000000   -0.0000000000000000   12.1495560778044602
W    S
2     4
Direct
0.3333330000000032  0.6666669999999968  0.2500000000000000
0.6666669999999968  0.3333330000000032  0.7500000000000000
0.3333330000000032  0.6666669999999968  0.6212727769502864
0.6666669999999968  0.3333330000000032  0.1212727769502864
0.3333330000000032  0.6666669999999968  0.8787272230497136
0.6666669999999968  0.3333330000000032  0.3787272230497136

0.00000000E+00  0.00000000E+00  0.00000000E+00
0.00000000E+00  0.00000000E+00  0.00000000E+00
0.00000000E+00  0.00000000E+00  0.00000000E+00
0.00000000E+00  0.00000000E+00  0.00000000E+00
0.00000000E+00  0.00000000E+00  0.00000000E+00
0.00000000E+00  0.00000000E+00  0.00000000E+00

• What do you mean by bulk WS2 bilayer? Is the calculation for bulk or bilayer? and there is no way of telling if the cell is relaxed unless you post some of the data you obtained. Jul 28 '20 at 16:31
• I don't see the POSCAR Jul 28 '20 at 19:22
• Looks like your calculation is for bulk. Verify if that's really what you intend to do. To check if the cell is relaxed, you can look at the forces printed in the output file (OUTCAR I think, I'm not completely conversant with VASP). But as long as the forces are less than your defined threshold, you should be in good shape. Jul 28 '20 at 20:13
• It might be viewable on more browsers, if you click "code" in the edit window, and then copy and paste it. The text will automatically be color coded, see here for an example of what a POSCAR looks like on this site: mattermodeling.stackexchange.com/q/1410/5 Jul 28 '20 at 20:33
• The simplest relaxation technique is for ISIF =3 which relaxes both the cell and the atoms. If this is what you intended to do and you did it correctly, then you should obtain the relaxed lattice constants. Jul 29 '20 at 6:19

## 1 Answer

The CONTCAR file contains the final position and velocities from a VASP run. You can find out more information from this vasp wiki page or from VASP manual. Please note, the 3rd 4th and 5th lines are the information about lattice constants a,b and c respectively. If you open this in VESTA the values shown for a, b and c are the magnitude of the corresponding lattice vectors. If you are getting $$\alpha=90^{\circ}, \beta=90^{\circ}$$ and $$\gamma=120^{\circ}$$, then you are with a relaxed hexagonal lattice.
From the CONTCAR it seems to me that the relaxed cell parameters are not of the hexagonal unit cell, as I can see a very small fraction on the y component of the a vector. For a perfectly hexagonal unit cell it must be zero.
You can try another run of ISIF=3 with a higher MP K-points grids(Assuming you are using optimised $$\Gamma$$-centered K-points). This may take you to the perfectly relaxed unit cell for a hexagonal lattice.

• Thank you sir for your useful answer I'll conduct another run with Γ-centered 12*12*3 k points mesh, but please allow me to ask you more question: Isn't sufficient to conduct just one ISIF=3 run? What are the different methods to ensure that I have a relaxed structure? Jul 31 '20 at 12:51
• Check the forces on individual atoms in the OUTCAR file. IF the forces are less than the convergence criterion of your necessity (specified with EDIFFG tag), your calculation is complete. There is no need to reoptimize. Aug 1 '20 at 9:29
• Thanks for the answers. They helped me.
– skm
May 28 at 15:01