Band structure calculation of Si_G structure's

[AJMALGHAN M]

I am working on Graphene Silicon Structure. I have 4 layer of atoms which is of about 36 A height. But i donot want to use all the atoms for band structure calculation  which will hang my cluster.

So how can run the program without hanging the system? May be lesser atoms , but i can prove theoretically till which layer Band structure will not change? How to prove that mathematically?

Also what about using Projection list? may be only s and p orbitals which are the outermost shell?


But while am using the bands_above_fermi_level, Its showing error.

Please someone help me on this. I am struck at my work. I am uploading my program with this.

[zh]

I am working on Graphene Silicon Structure. I have 4 layer of atoms which is of about 36 A height. But i donot want to use all the atoms for band structure calculation  which will hang my cluster.

So how can run the program without hanging the system? May be lesser atoms , but i can prove theoretically till which layer Band structure will not change? How to prove that mathematically?

This depends on the ordering of the atomic layers. If the atomic layers are stacked in the ABCD.... ordering, the decrease of the atomic layers in the unit cell will result in a different system, and hence the obtained band structure will change dramatically.  If the atomic layers are stacked in the ABAB... ordering, the use of two layers could be sufficient in some cases.

Also what about using Projection list? may be only s and p orbitals which are the outermost shell?

For the carbon and silicon atoms, the projection of s and p orbital is is reasonable.

But while am using the bands_above_fermi_level, Its showing error.

It may include too many unoccupied bands. You may consider only tens of bands above the Fermi level, instead of all unoccupied bands above the Fermi level 

[AJMALGHAN M]

Thank you sir for detailed reply,

i am using Si(111)_Graphene structure. system is periodic in x and y directions. in z direction i have 4 layer of silicon atoms. on top of Si(111), i gave a gap of 2.3 A which eventually minimised to 1.9 A ( which is bond length of SI-G).

But problem is can remove all 3 Si layers since no. of atoms in system is in the range of 2000 ( Now after 3 days of running , system is struck. I ma running in 20 core processor with 34gb ram in total)

I am using extended huckel calculation. Please tell one best way to reduce the simulation time.

1 more problem is while i am using  projection list, i cannot use  bands_above_fermi_level command. its showing error. Is there any restriction like that?

[zh]

"bands_above_fermi_level " is available in the "DensityOfStates()" or "Bandstructure()" objects, not in the "ProjectionList()" object.

Your simulated system is actually very large. As you said, it contains 2000 atoms. The tuning of computational parameters is not the feasible way to reduce the computing time. Alternatively, you have to increase the number of used CUP cores.

[AJMALGHAN M]

Can you help me with K point sampling ( presently using 9,9,1) and basis set suitable for the situation. i Think it may also cause the struck for running.

 I am using 10 CPU cores because of computational constraints.

[zh]

Because the supercell size (i.e., vector_a and vector_b)  in your simulated system is large, you can use a smaller k-mesh  or even a single Gamma point (1 x 1 x 1) in the k-point sampling.