device k-point convergence

[karma]

for a device configuration we calculate the k-points for the electrodes and then the centeral region ...it was mentioned in a forum that since the centeral region is non periodic the k-point in the c direction is automatically kept as 1
but when we calculate the configuration as a device then  how do we set the k-points for the device  along  a,b & c  ...also in the tutorial although the convergence is at a much lower value for electrode at  11 along c the value of 51 is taken.

[Jess Wellendorff]

The simple answer is that you do not really need to think about such details. You just need to select one single set of k-points (kA, kB, kC) in the VNL Scripter, then the k-point sampling for both electrode calculations and for the device calculation are automatically set up correctly. The more advanced answer is that 3 calculations are needed (electrode #1, electrode #2, device NEGF), so 3 k-point grids are set up in the ATK Python script. But they are all identical. Example:

Code

left_electrode_k_point_sampling = MonkhorstPackGrid(
    na=5,
    nb=5,
    nc=165,
    )
left_electrode_numerical_accuracy_parameters = NumericalAccuracyParameters(
    k_point_sampling=left_electrode_k_point_sampling,
    density_mesh_cutoff=140.0*Hartree,
    )

right_electrode_k_point_sampling = MonkhorstPackGrid(
    na=5,
    nb=5,
    nc=165,
    )
right_electrode_numerical_accuracy_parameters = NumericalAccuracyParameters(
    k_point_sampling=right_electrode_k_point_sampling,
    density_mesh_cutoff=140.0*Hartree,
    )

device_k_point_sampling = MonkhorstPackGrid(
    na=5,
    nb=5,
    nc=165,
    )
device_numerical_accuracy_parameters = NumericalAccuracyParameters(
    k_point_sampling=device_k_point_sampling,
    density_mesh_cutoff=140.0*Hartree,
    )