ATK 12.2.2 is released to give users access to one new feature (charged electrodes), a performance and quality update to another one (complex band structure), and a few small bug fixes.
For more details on ATK 12.2, see the original 12.2 release letter and the ATK 12.2.1 release notes.
ATK 12.2.2 introduces the possibility to use charged electrodes. It was already earlier been possible to simulate charged bulk cell, but the electrode calculators didn't pick up the keyword. Now they do, and this means you can add a positive or negative charge to the electrodes to simulate doping.
Doping can of course be simulated by just replacing e.g. a Si atom by P, but the resulting doping concentration becomes unreasonably high. In order to obtain realistic doping level one would have to use a very large simulation cell, which wouldn't be tractable for DFT and perhaps not even tight-binding. The effect of doping is to shift the Fermi level, such that the thermal occupation probability of electrons in the conduction band states (or, for acceptors, holes in the valence band) increases. The Fermi level shift is also the effect obtained when the a bulk cell is charged, and thus a substitute method to simulate doping is to add a small charge to the electrode cells. Since the charge can be specified as any floating point number, you can effectively add 0.001 or 0.01 electrons (or holes) to the electrode cell, which brings the doping concentration - and the Fermi level shift - to realistic levels.
Complex band structure
- The capabiulity of ATK to compute the complex band structure of a material, in a particular crystal direction, has been around for a while, but the new implementation introduced in ATK 12.2.2 is faster and doesn't involve any internal repetition of the cell, and thus there is no zone folding of the results which therefore are easier to interpret. We will soon publish a tutorial showing how to use this feature.
Krylov self energy method
The Krylov method for computing the electrode self energies is very fast, and unique to ATK. It is however an approximation, and sometimes it fails - in particular when there are no propagating states at a particular energy. ATK will then (as of version 12.2.0) revert to the Direct method, which also since 12.2 is the default method. It was however noticed that when this occurred, the Krylov results for higher energies could come out wrong. This is now fixed.
Since the Krylov method can be up to 10x faster than the Direct method, it is definitely well worth trying it. If you don't see problems with negative transmission (which can occur, due to the approximations made), then you can trust the results, but to be absolutley sure it's a good idea to try e.g. the zero-bias spectrum for a smaller number of k-points and energy points, just to see that Direct and Krylov agree for your particular system.
Minor other updates
- ExpressionSelect now (properly) supports operators "<=", ">=" and "!=". Update to version 1.2 via the AddOn Manager, if you don't want to install 12.2.2.
- Fixed a small issue with Hotbit on-site parameter file parsing.