Can ATK simulate the relation between temperature and transport property(I-V)?

[frank]

Hello! I am trying to study the effect of temperature(a true physical temperature) on transport properties of graphene.
And, I know that the "temperature"  in ATK is not a true physical  temperature by browsing related questions. How can I solve this problem?
Thank you!

[Anders Blom]

There are two ways to consider this. You can include electron-phonon interaction in the calculations (as in http://quantumwise.com/publications/tutorials/item/837-inelastic-transmission) or you can perform an MD simulation at finite temperature, and compute the I-V for some snapshots in the trajectory, and then make a statistical average.

[frank]

Thanks for your reply! It seems that the ATK 2015 has not been released,so I have to choose the second way(MD simulation).
But I am a beginner and is not familiar with MD simulation.Is there some tutorials?

[Anders Blom]

We have an excellent tutorial for general MD simulations for beginners:
http://quantumwise.com/publications/tutorials/item/732-basic-molecular-dynamics-tutorial
What particularly nice about it is that it focuses on using MD practically, not the underlying theory (which, honestly, you mostly need to worry about only if you will write your own MD code ).

[frank]

I have read the tutorial and know about MD simulation.But I have another question that what is the relationship between MD simulation and I-V.
MD offers different types,for example, NVT simulation,but what we can monitor is temperature,kinetic energy,total energy.I confuse that we can obtain information about I-V from these statistical quantities.
Thank you again.

[Anders Blom]

The idea is to take the different configurations obtained from the MD (snapshots, along the trajectory) and compute the I-V for each one, and then do a statistical analysis of that.

[frank]

I don't understand what you say completely.I know that I can obtain a file describing the trajectory of the system in phase space,which includes  a series of snapshots.What is a snapshot? I think it is the configuration  at a specific time derived from initial configuration.I have two questions.
1.How to obtain different configuration from a "trajectory" file.
2.Is it necessary for MD simulation to use the bulk configuration?

[Anders Blom]

It's possible to do MD with devices, but you have to constrain the electrode copies (the atoms corresponding to the electrodes, but which are inside the central region). This also means the degrees of freedom are slightly wrong for evaluating the temperature, but that's a minor concern only.

To get image 15 in a trajectory, in a script, just use

image = trajectory.image(15)

[bubble]

Thank you for your idea.
However, each configuration probably has different weigh factor [For example, Boltzmann distribution, exp(-E/kT) ]. How to consider the distribution in the "average value" of the transport properties?  and is it meaningful?
 

[Anders Blom]

Not sure that's necessary (but if it is, you can always evaluate the energy, I guess).
The approach was used e.g. in NANO LETTERS 8, 1120-1126 (2008).

[Julian Schneider]

Ideally, the snapshots obtained from an equilibrium MD simulation will be sampled from canonical distribution automatically (ergodicity theorem).  Therefore, no additional weighting is necessary. The configurations with high energy and low exp(-E/kT) weight will be correspondingly rare in the MD simulation anyway. Of course, for shorter MD simulations this holds only approximately.