RELAXATION EXPLOSION

Vasp transition state theory tools

Moderator: moderators

Post Reply
gitosr
Posts: 5
Joined: Fri Mar 25, 2022 5:51 pm

RELAXATION EXPLOSION

Post by gitosr »

I am studying a W grain boundary, which contains 456 atoms. In particular, I am trying to converge a configuration in which I have a vacancy between 1) 2 He (in white) and 2) an interstitial atom. As you can see in the file attached, in the first few steps, the surrounding atoms seem to 'explode' slightly, is this normal, and is there a way to prevent them from exploding when two atoms are 'close enough' and the forces are high? I am using IBRION=2, ISIF=0. Also, as it's a big and relative 'complex' system, I am doing the relaxation in two steps, first with a not very demanding convergence criterion (EDIFFG=-0.03). Any kind of help or recommendation is welcome!

Best regards,
Jorge
Attachments
xdatcar.mp4
(681.35 KiB) Downloaded 264 times
graeme
Site Admin
Posts: 2253
Joined: Tue Apr 26, 2005 4:25 am
Contact:

Re: RELAXATION EXPLOSION

Post by graeme »

It's pretty common to have high initial forces which can lead to unstable optimization. I recommend using a conservative optimizer in this case, such as IBRION=3 POTIM=0.01. You will likely just need about 10-20 iterations for the forces to drop to about 1 eV/Ang. Then you can go back to whatever optimizer you prefer for the rest of the optimization.
Rebeccarosan
Posts: 1
Joined: Fri Apr 07, 2023 6:06 am

Re: RELAXATION EXPLOSION

Post by Rebeccarosan »

Hello,
I think I am late for the wright the comment but I am just want to say that It is normal to observe some atomic relaxation or 'explosion' when you introduce a vacancy and an interstitial atom in a W grain boundary system. The atomic relaxation occurs due to the presence of high forces between the atoms, which can cause the surrounding atoms to move and adjust their positions to accommodate the new configuration. However, it is important to note that excessive relaxation or 'explosion' can lead to an incorrect final configuration.

To prevent excessive atomic relaxation, you can try reducing the convergence criteria gradually during the relaxation process, starting from a higher value and gradually decreasing it to achieve a more accurate final configuration. Additionally, you may consider using a more advanced optimization algorithm, such as the conjugate gradient (CG) method or the quasi-Newton method, which can help to minimize the number of steps required for convergence.

Finally, it may be helpful to check the convergence behavior of the system by monitoring the total energy and force convergence during the relaxation process. If the convergence criteria are too loose, it may result in an incorrect final configuration. Therefore, it is important to ensure that the convergence criteria are set appropriately for the system under study.

https://mcdvoice.me/
Last edited by Rebeccarosan on Sat Apr 08, 2023 3:47 am, edited 1 time in total.
gitosr
Posts: 5
Joined: Fri Mar 25, 2022 5:51 pm

Re: RELAXATION EXPLOSION

Post by gitosr »

Thank you very much to both of you, Henkelman and Rebecca. Finally, I was able to solve the problem by following Henkelman's advice, that is, by first using the Damped Molecular Dynamics method (IBRION=3) for about 15 steps. This allowed the atoms to gradually relax avoiding explosions until they reached a state where the forces were lower, and the Conjugate Gradient (IBRION=2) method could be run perfectly. I have been using this methodology for the past month and have had very good results.
Post Reply