Molecular electronics optimization.

[Aadhityan]

For single molecule devices, Is it necessary to check if the scf is converged or not for all the bias(for an example take 0 V to-5 V (increasing by 0.5V))?

[Petr Khomyakov]

If you want to get consistently accurate results for the entire IV curve, you should converge the scf calculations for each bias voltage applied.

[Aadhityan]

Till a particular voltage. I got scf convergence(say 1.5 volt). If I increase further the scf didn't converged. what interpretation we can give?

Then also shall I change some input parameter like k-pont sampling,mesh cutoff etc.?

Also If I change I think without bias or less than that (say less than 1.5 V) the system may get converged. In those case what shall i do?

[Petr Khomyakov]

If you have a convergence problem for higher voltage calculations, you may do IV calculations by increasing bias voltage in small steps (it should perhaps be smaller than 0.5 V adopted in your calculations), using the lower bias converged calculation as the initial state for the higher bias calculation. Take a look at the section on "Finite-bias calculations" in the ATK transport tutorial http://docs.quantumwise.com/tutorials/atk_transport_calculations/atk_transport_calculations.html.

Note that for some systems applying high voltages does not make sense, as the system may not stand high currents; this is something to figure out by consulting the corresponding experimental data in the literature. 

[Aadhityan]

Thanks for your replies Petr Khomyakov