Mid gap states

[Dipankar Saha]

Is this possible, for any perfect, relaxed struct. (which is a 1D ribbon actuallly)__
occurrence of mid gap states (in Tr_spectrm.) depending on the bias (between R and L electrodes) ....? [I mean....can their position, amplitude etc. vary...for diff. values of bias voltage??]

Or, they are merely spurious states coming due to the numerical calculation..!!

Best_
Dipankar

[Anders Blom]

As has been repeatedly and repeatedly and repeatedly pointed out on this Forum, for any perfect, periodic structure, running a finite bias calculation is typically not a good idea. The models in ATK are designed to handle the case of scattering - if there is no scattering at all, you can just as well do a linear response extrapolation from the zero-bias transmission curve.

[Dipankar Saha]

In ATK...while finding transmission spectra...do you explicitly calculate the dispersion??

Thanks & Regards_
Dipankar

[Jess Wellendorff]

The electronic transmission spectrum is calculated from the retarded Green's function for the central region and from broadening functions for the left and right electrodes. See bottom of
http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/chap.negf.html#sect1.negf.transmission.

[Dipankar Saha]

For a 2D system.....if I take 1 x N x 100 k-points(x,y,z) ...the Fermi level varies significantly with the "N"  ...!!!
1)
Is this the only way...that I have to vary the "N" till still the chemical potential of the electrodes stop varying?!! 
Since the density matrices are calculated separately...for the three regions (left,center,right)....therefore, how important is the k-point selection for say "y" and "z" directions ...such that the equilibrium is maintained b/w the electrodes and the channel??

2)
Which one is a better a choice...for say, MoS₂ sheet....;  1x15x15 or, 1x16x16 ? Because for such a 2D material...the important point of the BZ.... that I will look for...is not the "Gamma"...rather say "K" ....!!!

Thanks and Regards_
Dipankar

[Jess Wellendorff]

1) This should not worry you. The absolute value of the fermi level does not really matter. What matters are the eigenvalues of electronic states as compared to the fermi level. When your fermi level depends on the kpoint sampling, the vacuum level (work function) will change in a similar manner. However, when matching density matrices for electrodes and the central region, the kpoint sampling in the electrodes should be dense along the transport direction (c, z). This is why k_c = 100 is ATK default.

2) I have no definite answer for this. Converging whatever property you are interested in wrt. k-points would be my choice.

[Dipankar Saha]

Thanks a lot Jess.....!!!
_____________

Converging whatever property you are interested in wrt. k-points would be my choice.

convergence...considering  a certain property is fine... ; but what I wanted to know...... that's somewhat related to the odd and the even no.  ........
Is there any specialty in choosing the no.   .... like 1x8x8 ?? Besides, for such 2D materials.....why it is often a multiple of '3' ??

Regards__
Dipankar

[Jess Wellendorff]

Sometimes one can make clever choices of kpoint grids that utilizes the symmetries of the Brillouin zone to get a minimum of irreducible kpoints while still having a reasonable overall sampling. Choice depends on the shape of the Brillouin zone.

[Anders Blom]

The K point in graphene is at 1/3,1/3,0 in the Brillouin zone, that's why a sampling which is a multiple of 3 can catch this point, which is crucial for good accuracy.

[Dipankar Saha]

Okay... thank you Anders  ...!!!