Charge surface calculation using NELECT

[tanay11]

Hello, everyone,

I am trying to do a charged surface calculation for Ruthenium by providing one extra electron on the surface. I have got 14 atoms of Ruthenium in my POSCAR. and to provide charge on the whole surface I am using the tag "NELECT". I am running this calculation on Compute Canada cluster using VASP/6.1.2. The problem I am facing is that the calculation gets stopped after running for 30 mins without any error message in the OUTCAR file. I submit the JOB using JOB manager named "SLURM", which gives me a message - "Slurm Job_id=22586384 Name=vasp-test Failed, Run time 00:27:55, FAILED". But when I open the output files I could not see any error message or I could not comprehend why the job is not running.
I am attaching the OUTCAR, INCAR, and slurm-output files for reference. Can someone suggest to me why this happening? Is this related to "NELECT" tag or I am not using the right tags?

Kpoints - 5*5*2

Thanks in advance.

[fabien_tran1]

Hi,

Your calculation did not crash or finish with an error, but stopped because the maximum number of ionic steps (NSW=20 in INCAR) has been reached (grep for "Iteration" in OUTCAR). You can see that the atoms are still moving (the atoms positions are indicated below "POSITION"). To let the system to reach the equilibrium geometry, you need to increase NSW to a larger value, let's say 100. Note that the file CONTCAR can be used as a new POSCAR when restarting the calculation.

[tanay11]

@fabien_tran1
Thank you for the suggestion. I tried your suggestion, but it did not completely solve the problem. As you suggested, I increased the ionic step to 100, but the job still gets terminated with "Slurm Job_id=22713968 Name=vasp-test Failed, Run time 05:37:29, FAILED" almost the same message, just the time for stopping has increased from 30 mins to 5 hrs.

[fabien_tran1]

As I wrote previously, the atoms were still moving after 30 ionic steps. At each ionic step, the positions of atoms are written in OUTCAR below "POSITION". Have a look at them to see if the equilibrium geometry has still not be reached after 100 steps.

[tanay11]

Thanks @fabien_tran1
I figured out that problem. As I am moving forward with the calculations. I came across the post :

https://www.vasp.at/forum/viewtopic.ph ... ECT#p19854

And I want to do much similar to the paper (https://arxiv.org/abs/2005.03378) mentioned by @martin.schlipf. So I want to know whether this feature as mentioned in the previous post has been released in the latest version of vasp or not. If it is released then how to use that.

I would be really blessed to have any help in this regard.
Thanks and regards

[fabien_tran1]

Hi,

Actually, calculations on charged slabs are problematic and require a special treatment. The paper from da Silva et al. addresses this and their SCPC method can be used with VASP 6.2.0 onwards (so, unfortunately not with 6.1.2 that you are using). However, note that in 6.2.0 and 6.2.1 there is a bug in src/scpc.F, which means that this file has to be replaced by the one provided on https://github.com/aradi/SCPC-Method. This replacement will not be necessary for the upcoming release 6.3 of VASP.

So, to use the SCPC method, you need to compile the SCPC code from github, use VASP 6.2.0 or a more recent version (and replace src/scpc.F for 6.2.x) that has to be compiled with -DSCPC.