absolute value of I-V

[atk_user]

Dear all,

I'm testing the I-V calculations for Al(100) / Al(100) / Al(100) very simple structure.
If I consider the supercell of 2x2 along the x and y direction, the current values are increased.
The current values are depends on the size of the model structures? That means the absolute current values are meaningless?

How can I define the absolute current values?

Also,

The direct comparison of I-V data with different stoichiometric system in central (scattering) region (Al2O3, AlO etc.; total number of atoms are different) is possible?

[Anders Blom]

There are two possible effects to consider. First, if you double the surface area you should get twice as many conducting channels, and thus 2x the current. So, in essence you are computing the current per unit area of the structure.

Now, it may happen that you get a slightly different factor than 2, if you keep the k-point sampling when you double the cell, since you now will have a more accurate calculation.

[atk_user]

Thank you for your reply. I have a further question on this matter. 

I found the experimental value of resistivity of aluminum (Al) is 2.709 x 10^-8 ohm . m (@ 298 K)
The converted current values from above resistivity are about 10^-2 A. (I attached my model system. Calculate: PBE functional, SZP basis sets)
However, the calculated values are about 10^-4 A. It is lower by two orders of magnitude than experimental results.

These comparison are make sense? I'm not sure the accuracy of the calculated results.
How could I validate my calculated I-V data? Is there any need to calibrate the current values?

[Anders Blom]

You can not calculate resistivity this way, so the numbers cannot be compared at all. Resistivity is measure of the diffusive transport of electrons, over large distances where phonon and other dissipative scattering mechanisms dominate. With the NEGF method, you are instead computing the coherent tunnel resistance, which in principle is zero for a perfect crystal, so in fact such calculations are in a sense meaningless. ATK is rather designed to simulate cases where you are tunneling from a metal into a semiconducting region (can be a molecule, a bulk semiconductor, or just anything that breaks the periodicity of the perfect crystal of the electrodes, or an interface between two metals/semiconductors, whatever). And so, you compute not the conductivity but the conductance, which does not take dissipative scattering into account.