QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: Om Prakash Upadhyay on February 7, 2017, 15:44

Title: Electron temperature
Post by: Om Prakash Upadhyay on February 7, 2017, 15:44

Does the electron temperature given in ATK-VNL refers the normal surrounding temperature or any other else?

Title: Re: Electron temperature
Post by: Petr Khomyakov on February 7, 2017, 16:10

The electron temperature in the ATK reflects a degree of broadening of the Fermi-Dirac distribution (or its derivative), which is otherwise a discontinuous step function (or Delta function for the derivative) at T=0 K that is hard to handle in the numerical calculations such as that done with the ATK.   

Title: Re: Electron temperature
Post by: Om Prakash Upadhyay on February 7, 2017, 17:19

If we increase the electron temperature then can we say that it is temperature dependent calculations?

Title: Re: Electron temperature
Post by: Petr Khomyakov on February 7, 2017, 22:02

If the smearing (temperature) values adopted in the DFT calculations significantly affect your results, it means that you have to decrease the  artificial temperature value. It also means that you should first decide on what is the target accuracy for the physical quantities you want to calculate. This is rather similar to choosing any other computational settings for numerical calculations, in general, e.g., choosing mesh density, k-point sampling and so on.

Title: Re: Electron temperature
Post by: Anders Blom on February 7, 2017, 22:57

So, to be clear, the answer is no, a higher electron temperature in DFT is just a numerical trick to ensure the convergence is smoother. Formally, DFT is a T=0 theory unless you add extra effects like electron-phonon scattering, which is possible in ATK (see tutorials).

Title: Re: Electron temperature
Post by: Om Prakash Upadhyay on February 10, 2017, 11:46

Thanks alot for reply.
  But actually what I found that, when I increase the electron temperature in Cu2O nanowie, it is observed that the nanowire which shows hal fmetallic nature at 300K ,  shows matallic nature above 2000K (depending on diameter) rather half metallic. What is the main reason behind it???

Title: Re: Electron temperature
Post by: Jess Wellendorff on February 13, 2017, 09:18

You are probably broadening the electronic occupations around the Fermi level too much, such that states above the T=0 K Fermi level becomes occupied.

Title: Re: Electron temperature
Post by: zh on February 14, 2017, 01:01

Quote from: Om Prakash Upadhyay on February 10, 2017, 11:46

Thanks alot for reply.
  But actually what I found that, when I increase the electron temperature in Cu2O nanowie, it is observed that the nanowire which shows hal fmetallic nature at 300K ,  shows matallic nature above 2000K (depending on diameter) rather half metallic. What is the main reason behind it???

Please refer to the literature survey for the different types of computational methods on  Cu2O
www.iaea.org/inis/collection/NCLCollectionStore/_Public/43/041/43041487.pdf