How to apply uniaxial stress?

Message

zrfssh · #1 Post by **zrfssh** » Thu Aug 02, 2012 6:16 am

In VASP, one can set PSTRESS to get a hydrostatic pressure. Is that possible to let VASP automatically relax to a certain uniaxial normal stress, e.g. Stress_x=0, Stress_y=0, while Stress_z=50GPa?

Thanks

icemoonhe · #2 Post by **icemoonhe** » Thu Aug 16, 2012 5:14 am

the uniaxial stress could be applied by changing the length of the z axis. but I have other questions: the OUTCAR file gives the stress on different directions :
Direction X Y Z XY YZ ZX

in kB -90.92 -90.92 -169.98 0.0 0.0 0.0
the x and y directions also give non zero stresses, and this can understand because when the z direction have the strain and neglect the possion effect the x and y direction also have strain. the questions are:
(1) how to release the strains in the x and y directions.
(2) the stress values calculated by vasp.4.6 (ISIF=2 IBRON=1) in the OUTCAR file can be reliable?
(3) if the new version or specific program (like NEB the vasp.neb should be compiled) should be used or compiled.
(4)how to compare the above stress in z direction with the stress calculated by the force on the fixed atoms? why they are different? which is more reliable?

thank you

<span class='smallblacktext'>[ Edited Thu Aug 16 2012, 04:11PM ]</span>

maartendft · #3 Post by **maartendft** » Tue Aug 21, 2012 8:23 pm

1) Adapt the lengths of x and y cell vectors until you reach zero stress in those directions. You will have to do this manually or write a script for it. You can use the Young's modulus of your material to get a good initial guess for the strain adaption.
2) Yes, but use high enough PW cutoff.
3) ?
4) By relaxing the atoms, the stress tensor changes due to changing atomic interactions. If you expect the material you're modeling to "relax" in real life, you could choose to relax the ions in VASP.