The thermal considerations of QuantumWise

[perfetti]

Dear everyone,
        I am working on a case that CNT-Bi's conductance properties VS temperature. We optimized the same structure at different temperatures and then calculate their transmission spectrum. However, the structure after the optimization looks ridiculous as the metal atoms go far away from the carbon nanotube. And the transmission coefficient has no particular trend, as it goes down from 300K to 403K, but goes up from 403K to 423K.
       I am not sure if the QuantumWise software considered the temperature effect on properties, and to which extent?  Maybe it's not a proper way to try to pry the temperature effect on this software?
       On the other side, my device may have problem and any picky advice is very welcomed.
       Thanks very much!

[Anders Blom]

Not enough information available. Method used, DFT?

It's probably not really a suitable way to study the temperature effect as the temperature-dependent of the conductance is more due to the interaction of phonons and electrons than the difference in conductance due to some random fluctuation of the atom position with temperature. One could do an MD simulation and take snapshots at different times, do statistics, and the compare the average at different temperatures, but it seems that's not how you do it. In fact, I'm not sure I understand how you take the temperature into account at all in the optimization (are you doing MD?), hence the question about method etc.

[perfetti]

Thanks Dr. Blom.
I used DFT and that's the way I felt a little bit unproper.
About how I did it, I just set the temperature of each calculations at different values, and then calculate the corresponding result.

On the other hand, if you suggest MD, does QuantumWise considered phonons?
And if I want to do research not only on the thermal properties, but also on the mechanical properties, which method is the suitable way?

Thanks very much.

Not enough information available. Method used, DFT?

It's probably not really a suitable way to study the temperature effect as the temperature-dependent of the conductance is more due to the interaction of phonons and electrons than the difference in conductance due to some random fluctuation of the atom position with temperature. One could do an MD simulation and take snapshots at different times, do statistics, and the compare the average at different temperatures, but it seems that's not how you do it. In fact, I'm not sure I understand how you take the temperature into account at all in the optimization (are you doing MD?), hence the question about method etc.

[Anders Blom]

So in essence you are describing the effect of hot electrons, which usually is very small. In the tutorial I referred to, the main effect is observed when the electrodes is at different temperatures, but also the thermionic effect is larger for larger tunnel distances.

ATK 13.8 can compute phonon frequencies and phonon transport, but electron-phonon interaction is not yet included.

In theory you can do DFT-MD but it's extremely time-consuming, so the sensible choice for mechnical properties is surely classical MD. ATK 13.8 introduces many new functionalities in this area, such as the Tersoff potential, NPT, and a few other things.

[perfetti]

Thanks Dr. Blom.
Maybe I will switch to other MD software as a backup plan to this kind of research.
thank you so much.

So in essence you are describing the effect of hot electrons, which usually is very small. In the tutorial I referred to, the main effect is observed when the electrodes is at different temperatures, but also the thermionic effect is larger for larger tunnel distances.

ATK 13.8 can compute phonon frequencies and phonon transport, but electron-phonon interaction is not yet included.

In theory you can do DFT-MD but it's extremely time-consuming, so the sensible choice for mechnical properties is surely classical MD. ATK 13.8 introduces many new functionalities in this area, such as the Tersoff potential, NPT, and a few other things.

[Anders Blom]

ATK 13.8 will also have quite a lot of new MD functionality, but if you are really looking for electron-phonon coupling it's not really a matter of MD, but rather this particular functionality added to a DFT or tight-binding code, and that's hard to find. We're working on it, but it will take a while yet.

I know some people have tried to run MD and then compute the transmission at various snapshots in time. This would be one way to include some lattice temperature effect, at least in an approximate way. But it would still be coherent elastic transport. This could be done with ATK today, although as mentioned it's quite heavy (you can do the MD with classical potentials, possibly, if you have a potential for the materials of interest).