quantum dot

[nikila]

Dear all,
I am new to VASp. Can anyone please tell me wether we can calculate the band structure of few non periodic atoms. IF yes then how the POSCAR file and kpoints to be generated.

Thanks in advance

[admin]

Bands and band structure is a property of the periodic structure. Atoms and molecules exhibit separated energy levels, not bands.
On the other hand one can draw DOS of molecules. Just insert a molecule into a box (e.g. 10x10x10A) and use some smearing (ISMEAR=0.2) to get broadening of energy levels.

[vistawanted]

The easiest way to make un-periodic model is big box method. Some other methods can also make isolate, such as Makov-Payne, Martyna-Tuckerman methods. But I do not have idea whether these methods are supported by VASP.
After all, if the quantum dot work is wanted, the model should be very huge in ab-initio area, since the normal quantum dot's size is at least several nano-meters, which is much largger than lattice parameters.

[SKM]

hi all,
though it is very old, thread, now i am thinking of similar work on simulating qunatum dots in VASP.
So, i am also thinking if i want a QD in VASP, it will be a huge size and thousands of atoms.
So not possible in VASP as per this thread final reply?

but my problem is only dealing with core-shell structure of QDs. So my work is related a small portion of the surface say 10x10x10 angstroms slice and z-direction may be added with more vacuum. so its like a 2D model but not actually 2D material.

after this i want study, atomic vacancy effects in the structure. is this sounds logical to do with VASP?
any expert opinion, please?

regards.

[martin.schlipf]

This obviously depends a lot on the particular system and level of theory that you intend to use. With decent computational hardware it should be possible to simulate ~1000 atoms with VASP. Keep in mind that the vacuum region will also increase the computational cost, so a 1000 atom molecule is much more expansive than a crystal with the same number of atoms.
But the general approach would be anyways to start with a smallish system of 20-50 atoms that kind of resembles what you want to do. This gives you an idea of the computational effort, because you know that doubling the system size will lead to an increase of cost by a factor of 4 to 8. This way you can assess the feasibility without spending a large amount of resources initially.

[SKM]

Hi Martin

Thanks for the quick reply.
1. Good information on 'vacuum space also increases computational cost'. I didn't know this. So, as a thumb rule, if i have the atoms occupied thickness of say X angstroms, then adding the vacuum above it around 10 to 15 Angstroms, would be enough? or still any computational cost related aspects are involved in this too?

2. coming to the system i mentioned above of 10x10x10 is not necessary, i guess, as per your reply. The point is the core/shell type structure is something like a coating on a substrate. So its not like attaching a ligands at specific places on QD. in that case i can take a a further small portion (say even 3x3x3) of slice should do, to study the effects of vacancy creation. But before i do that way, kindly let me know the reply on:- if i take such portion, will it affect the Periodic Boundary Condition (PBC) aspect of VASP? As it may depend on how exactly the PBC as implemented in VASP means? I mean, is it like the PBC is assumed a perpetual extension in all directions without boundaries? because the QD will be around say 5-6 nm (50-60 Angstrom) size. So, will the features vary, if we take small slice compared with full QD is simulated? [not sure if i put my query correctly]

Regards

[martin.schlipf]

@1: You need to converge your vacuum distance. But there are some tools in VASP to make this faster. But as an order of magnitude 10 Angstrom seems reasonable.

@2: VASP assumes an infinitely extended crystal, so when you calculated confined systems with VASP, there are two reasonable limits: The system is so small that you can describe all atoms in a box or the system is so large that the effects at the surface can be neglected compared to the extended behavior. Then you would do the confinement as a postprocessing step.
The difficult to impossible size of systems are the one where you have an extend of 10~100nm along one dimension and the effects of the boundary cannot be neglected.