a paradox about Conductance of Si and Cu

[spin]

Dear all

I calculated the conductance of Si and Cu atoms in equal conditions, but the conductance of Si was greater than Si. The electrodes were Gold linear electrodes.
Cu atom is metal with a great conductivity but Si is semiconductor or semi metal, then why the G value for Si is higher?

[zh]

Is the bulk silicon (diamond structure) coupled with the electrode? or other structure of silicon? It is better to state clearly the exact structure of Cu and Si coupled with the electrodes.

The electronic structure of  single atom is quite different from those of bulk form.

[spin]

I located just a Si atom and a Cu atom between electrodes, then I tried different distances. I set the system in the contact mode in minimum distance between Gold and related atom, then the distances were increased symmetrically in both side. I tested the system for about 20 different cases and in all of the Si conductance was higher than Cu!!! 

[spin]

I have attached the related system for Si, electrode length is 1.8 and mesh cut off=100, K-points=100 and other default parameters.
you can change the central region width for example from 8 to 10 and then with the same conditions make the computations again for Cu, you'll see Si conductivity would be greater!!! Please kindly someone do it, I want to know that this error is inside me or not.

tnx

[Anders Blom]

This structure is too simplistic. Your electrodes should, first of all, be at least 3 or 4 atoms long, which in turn will demand more atoms in the central region (see http://quantumwise.com/publications/tutorials/mini-tutorials/99 for details on this).

Moreover, as already pointed out, a Cu atom is not necessarily more conductive than a Si atom. It all depends on the bonding an level alignment to the electrode band structure.

[spin]

Thank you so much for your nice response. And another question: do you think there are any difference between conductance results of a defined system in 2008 version and 2011 version?

[Nordland]

It is also very important to understand there is a difference between a Cu atom and Cu crystal. And the same goes for Silicon.

Take my favorite element carbon. As a single atom it has a huge homo-lumo gap. As a atomic wire it can be metallic. As graphene it is semi-metallic, as nanotubes it can be semiconducting.

So in short the behavior of materials can rarely be derived from a single atom, as the interaction with other atoms is most of the physics.

[Anders Blom]

About your question regarding versions 2008.x and 11.x, there are differences but they are not enormous. Above all 11.x provides more stability which makes it easier to converge calculations, contains more features, and is somewhat faster. But although we did fix some bugs and improved the transport algorithm, I would say that obtained results from 2008.x calculations are as correct as those from 11.x provided they are converged properly in all aspects (like screening length, k-points, etc). There are a few exceptions to this, like forces (in particular under bias) which were not entirely correct in 2008.x and earlier.

[spin]

It is also very important to understand there is a difference between a Cu atom and Cu crystal. And the same goes for Silicon.

Take my favorite element carbon. As a single atom it has a huge homo-lumo gap. As a atomic wire it can be metallic. As graphene it is semi-metallic, as nanotubes it can be semiconducting.

So in short the behavior of materials can rarely be derived from a single atom, as the interaction with other atoms is most of the physics.

It means that the roles may be different for atomic structures when we have only one atom, so my question: when we say that Silicon is semiconductor, it means that bulk silicon is semiconductor? and may be the single atom is not... is it true? 

[Nordland]

Exactly. In daily terms when talking about silicon, then it is often about diamond silicon which is a semi-conductor.

[Anders Blom]

Whereas many Si nanowires are found to be metallic

[spin]

tnx Anders and Nordland for you good responses.