Dangling bonds at electrode unit cell edges of 2-probe system

[Susan]

When we build a 2-probe system, we will have a finite scattering region and 2 semi-infinite electrodes. We can’t specify all the atoms of the electrodes, instead we usually only give the unit cells of the electrodes. At the outer edges of these electrode unit cells, there will be some dangling bonds.

Will these dangling bonds make troubles or affect the convergence/results? Can anyone tell me how ATK deals with these dangling bonds? Is there any recommended literatures which talks about this?

Thank you in advance!

Susan

[zh]

The periodicity is kept along the a and b directions of a two-probe system. If the electrodes consist of crystalline solid, there will be no dangling bonds. If the electrodes consist of 1 dimensional nanowires (e.g., silicon nanowire), you may need to add hydrogen atoms to passivate the dangling bonds. 

[Susan]

Hi zh，

Thank you for your illustration. What i mean by "dangling bonds" here are not really dangling bonds since the electrodes are periodic in c-direction. But at the interface of two neighbouring unit cells, there might be some bonds accross the interface, and if we focus only on one unit cell, these bonds accrossing a unitcell interface are the "dangling bonds" I referred here.

Will these bonds make troubles in the calculation? If we need to passivate these bonds with H, do we have to do this manually when we set up the configuration or we just leave it and ATK will do it during the calculation?

Thank you!
Susan

[Anders Blom]

The electrodes are always periodic in A and B, so if there is no vacuum gap, then the bonds are not dangling just because they appear to go outside the unit cell. They join with the next repeated copy of the cell. You can check this by setting the visual repetition to 2 or 3 in the Viewer or Builder.

Dangling bonds only appear when an atom doesn't have enough neighbors to create bonds for all its valence electrons. Dangling bonds are actually not visualized in VNL (because there are not atoms to bond to), so the way to detect if you have any is to count the number of shown bonds, and compare to the known valence. So, if you for instance make a graphene nanoribbon, the edge atoms will only have 2 bonds (2 neighbor C atoms), while all atoms in the middle of the ribbon have 3. So unless you add hydrogen on the edge, there will be dangling invisible bond.

But, as mentioned, if you see bonds sticking out of the cell, these are by definition not dangling but real bonds.

[nori]

there might be some bonds accross the interface,

If so, they are not dangling bonds because the unit cell interacts physically with the adjacent cell.

So you don't need to worry about it.
Instead, if you add the passivation with hydrogen, your calculation will result in wrong one.