Hi Valerius,

You may model a large supercell of graphene with edge atoms functionalized/passivated by H or suitable groups depending on the problem of interest. In QE, periodicity may be turned off (technically, it still does periodic calculations) by keeping this supercell in a large cubic cell. Make sure, there is no notable interactions present between the two repeating units (~10 Ang vacuum) along any given direction. In principle, this approach may be used to study the edge effects. But, the calculations will be computationally expensive, due to the large size of the supercell. When you have such finite structures, it is better use other software that uses localized basis sets than QE or VASP. You may use SIESTA, CRYSTAL, TURBOMOLE or even Gaussian for QM calculations or can use DFT-B / other semi-empirical force field based methods.

regards,

Nijam

Hi Nijam,

Thanks a lot for your kind answer. Pardon me since I would like to ask more questions.

I just wanted to know any reference that I can use for the distance between periodic xy image and also the type of calculations that I should probably run. Whether I should use hexagonal unit cell or cubic or trigonal for "vc-relax" or "relax". Should I perhaps run the calculation more than 1 input parameter for convergence? And if I guess correctly, then, if I input my atomic position with distance about ~10A for each periodic image, then for each unit cell there must be a distance ~5A from the edge of the atomic system position to the edge of the unit cell, right?

Also, how should I know that my band calculation will not give flat (really flat) band structure? Since I tried to increase the supercell size while keeping the atomic position a bit far from the edge of the cell which gave me totally flat bands.

And also, how do I insert bulk effect in my input parameter? Should I perhaps study about bilayer, trilayer, and few layer graphene to know about the interlayer spacing between them and then doing vc-relax calculation?

Finally, I wish to know about how computationally expensive the calculation is. Which means, if I run using cluster, how long should the calculation probably take? And also, how many variations of input 'nat' should I give usually to report the calculation thoroughly?

Thank you

Hi Valerius,

In my understanding, the best approach is to perform some calculations on simple unit cell of graphene (2 atom unit cell, 2D sheet) and benzene (not really a graphene QD. But, helps to understand if the calculation is correct). It takes practice to know how to adjust the cell parameters.

For finite systems, you only need to do relax. For bulk and 2D systems, vc-relax (sometime, relax only along 2D) is recommended. The details can be found elsewhere in the internet.

I would run a few calculations to know what happens to the band structures, and how long does it take for the calculations to converge.

best,

Nijam