What about open- versus closed-shell calculations?

Hi Vladimir,

I am sorry. I don't seem to get your point.

But here's the thing, I am doing quantum espresso calculation for flake the molecule-like system. I am using martyna-tuckerman and spin-polarized calculations. I do not know the default whether it is open-shell or closed one. Could you give me some specific pointers?

Thank you Vladimir A. Basiuk

I am guessing that that the 15-atom system has unpaired electrons, and the 16-atom system does not. This would imply open- (slower convergence) and closed-shell calculations, respectively.

hi Vladimir,

I wrote the wrong thing. If you're saying that 16 is more fast than 15, the fact says that 16 is not converging while 15 does. Therefore if this is like you said, then the closed shell must be slower than open shell ones. Is that true? Still, 14 atoms (most likely open shell) have slower convergence (probably not converge) if we compare them to 15 atoms. This is weird.

Thanks a lot,

Valerius Evan

Can you show the structures (as an image attached)?

Hi Valerius

Do you use pseudo potentials ?

Does it depends strongly on the choice if yes ?

Also the network for FFT is critical

JLuc

And also is your mechanical configuration stable ?

Generally before calculation of electronic potentials, you have to make a preliminary optimizing calculation, varying slightly the inter atomic distances to find the energy minimum.

Hi Vladimir,

These are the images that you asked

Hi Jean-Luc Yves Babigeon

I don't know if the optimization calculation you mean is relax calculation in a self-consistent way, such as I did through Quantum ESPRESSO software. Also, I quite believe my configuration is a bit stable theoretically, although they might not have real implementation in experiment, I presume, since it is too small. And yes, I used pseudopotentials, I used one of them called C.pbe-n-kjpaw_psl.1.0.0.UPF, with rel = 0, since I wanted the non-relativistic system, though, from SSSP accuracy. I was asked to vary the cutoff radii and I try to find the convergence for every configuration, though. And since my system is graphene flake, a molecule-like solid system, or simply molecule system; I believe that cell parameters is quite unimportant if I use correction to isolate the system. Although I might still need to optimize it quite a bit.

Do these clusters have H atoms to fill dangling bonds? Otherwise they are unrealistic. Especially the one with 15 atoms.

taken from your question :

"...In a certain configuration 15 atoms are much easier to converge than 15 atoms..."

Not sure to understand : You mean 15 and 15 or 16 and 15 ?

From your 2 images, the symetry group should not be identical.

JLuc

Hi Vladimir A. Basiuk

Well, since I do relax calculation they can make their own dangling bonds I guess, or reconstruction perhaps, and also my supervisor did not ask me to calculate those with edge termination. And I pretty much know this situation is unrealistic, but I am just trying to find band gap trend and convergence by adding atom by atom. That's what I am trying to pull.

And for the notice, the computational power in my place kinda very low even when I do this, it is already so heavy and not fast. What will happen I wonder if I do about hundreds of atoms..

I found a paper from Kalita that graphene growth in CVD using CrO2 gives irregular shape somehow. Since this is just a model and not quite realistic, I probably will save it from there. Except there is a miracle. But no I just want to finish this job.

Other than that I am still unsure of your opinion about the open and close shell in this case, sir. Would you please explain more?

And hello to Jean-Luc Yves Babigeon

I am pretty sure I have mentioned above that I typed wrong. It's a typo. It is 15 with 16 not 15 with 15. Otherwise my question does not make sense too.

Thanks a lot

I think you have to learn the basics of computational chemistry.

Yes sir, Vladimir A. Basiuk , with all honor, no one pointed me the right way. So, what do you suppose me to do specifically? Since the computational chemistry basics still sounds too general to me.