In ATK, is it possible to extract the quantitative value for self-energy?

[Jin-Kyu Choi]

Dear all,

I am a student ATK-user and interested in interactions between bulk-surface (electrodes) and molecules as well as, of course, transport in molecular junctions (usually two probe systems).

As I studied, transport properties through a molecule in molecular junctions (molecule sandwiched in two electrodes) is quantitatively characterized by the electrical quantities, i.e. current and conductance, which are theoretically calculated from transmission coefficients (transmission spectrum along with energy) based on nonequilibrium Green's function (NEGF) formalism.

Furthermore, to calculate the transmission coefficient, 'Self-Energy' term which stands for interactions between electrode and molecule should be considered, as I know. And such sequential considerations are processed in ATK package.

Thus, in order to fully understand physical meanings and theoretical background for calculated results, I would like to want to know and extract the 'Self-Energy' value.

What I want to ask is:
1. In ATK, is it possible to extract the 'Self-Energy' value and how to do it?
2. To only consider interactions between electrode-surface and molecule, it is possible to calculate 'Self-energy' for only one electrode and molecule system not two probe system (i.e. namely one probe system)?

Please give some comments for my questions and what I understand if there is incorrect explaining.

Thank you.

[Anders Blom]

The self energy is not just a value, it's a very complicated matrix. Extracting it would technically be possible (although it's not supported currently) but would hardly provide you with any useful information.

One-probe systems are not implemented yet, it's an interesting idea for the future that we have looked into, but so far have not had the resources to pursue.

[Jin-Kyu Choi]

Thank you for your kind reply 

In a given circumstance, still I would want to investigate about the interaction between electrode and molecule by addressing the 'Self-Energy' concept.

For the 'Self-Energy', as I studied, its real and imaginary parts are related to shift and broadening of the molecular orbital of the junctioned molecule in a given molecular junction compared to its isolated (or free) state, respectively.

Thus, I think, in the current state in which I couldn't access the 'Self-Energy' matrix of complicated form in ATK package as you explained,
it could be a simple and effective way to theoretically study the electronic interaction between electrode and molecule: comparison of molecular orbitals of a molecule in the isolated and sandwiched states.

So, I would like to ask that "is there any way to calculate molecular orbital (orbital energy) for a free molecule in ATK package to compare the results with those of molecular junction?"

Actually in recent, in order to take account for it, I have been trying to co-use Gaussian package which was used to calculate the molecular orbital for a isolated molecule. But, due to their fundamental difference, it is difficult to find or conclude meaningful results.

Please give a help or comment on my idea.

Thank you.

[Anders Blom]

Yes, this is possible. You can do a projected spectrum. If the setup is a device configuration (you can try to make larger vacuum gap to the right surface to mimic the "one-probe" configuration) then if you compute the MolecularSpectrum projected onto the molecule, it will give you the spectrum of the hybridized molecule. If you compute the DeviceDensityOfStates you can also see the width of the levels, provided that you can make a clear mapping.

[Jin-Kyu Choi]

I would like to thank you for your valuable idea. 

I will try to do it very soon.

Thank you once again.