optimum K-points for convergence

[tara]

Does optimum  K-points  for a device structure depend on the electrode bias  we apply?? I observed that for higher drain bias  ,i require higher value k points in x-diection(i have periodic boundary condition in x direction)  for slater koster  calculation convergence??

[Umberto Martinez]

Indeed, having more k-points may also increase convergence besides increasing accuracy.
so yes, if at high bias you have convergence problems, having more k-points may (or may not) help.

[abhi005]

sir , i am trying to simulate a cntfet with drain voltage 1.5v and gate voltage from -2 to +2 .geometry is converged for gate voltage .2 to 1v but did not converge for other voltages .
what should be the k-point sampling? i took 1,1,300, is it correct?
cntfet is 16 A long with gate of 12 A.

[Umberto Martinez]

it is not wrong especially if it converges.
However, there may be something else to try first.

your gate region is also probably too big compared to  the central region.
increase or decrease the central region or gate size, respectively.

with other voltages do your mean negative voltages? how do you calculate this points?
It is highly recommended to restart from 0 bias state when you switch to negative biases.

[abhi005]

thanks for your reply sir,
yes sir it is not converging for negative voltages.
i first calculate transmission spectrum for 0 bias then i make a loop to calculate the transmission spectrum for other bias and with the help of these transmission spectrum i can calculate curves for current.Is this the correct way to calculate current?
i also increase the electrode size and electron temperature to 2000, because electron temperature has its effect on convergence.
am I doing correct changes?