Question about mobility

[yemeng]

I am very happy to see that ATK 2015 can calculate the mobility of bulk. The script that online manual given to calculate the mobility of graphene seems cannot run (some function seems unavailable). So I build the script myself. I know that it need many k-points and q-kpoints to calculate the electron-phonon couplin. But i just get a one-core trial license, so I have to use quite fewer k-points and q-points (5*5*1). I can get electron-phonon coupling result (although it is not correct), but the mobility I get is zero. It that correct? I think although the accuracy isvery low, the mobility should not be zero. If any one use this feature before? Waitting for your advice, thank you.

[Anders Blom]

1. It will be zero with such k-point sampling because you are not capturing the Fermi level, which is at K, and hence not included in your k-point set. Try 9x9 or something similar.

2. We generally never give single-core trial licenses - check your license file, you should have the ATKSlave feature,  and if so you can run fully in parallel, which of course is really helpful for mobility and other heavy calculations.

[yemeng]

Thanks for reply. I get this license from VNL-ATK workshop in Wuhan, they told me that I can only use one core with this license.

I try to increase the k-points to 9*9*1, this calculation last about one day on my PC. However, the mobility is still zero. I put my script here. Could you help to check it.

1. It will be zero with such k-point sampling because you are not capturing the Fermi level, which is at K, and hence not included in your k-point set. Try 9x9 or something similar.

2. We generally never give single-core trial licenses - check your license file, you should have the ATKSlave feature,  and if so you can run fully in parallel, which of course is really helpful for mobility and other heavy calculations.

[Anders Blom]

If your license file doesn't contain "ATKSlave" (it's just a text file, so you can read it) then that's true - and if it's the case, then you can get a better demo license file via http://quantumwise.com/free-trial/trial-license-form.

Yes, the mobility will be zero in your case. MonkhorstPack sampling will not be suitable, what you want is to sample closely the range around the K point. See http://docs.quantumwise.com/tutorials/mobility.html for a lot more details.

[yemeng]

It solve my problem! Thanks a lot.

If your license file doesn't contain "ATKSlave" (it's just a text file, so you can read it) then that's true - and if it's the case, then you can get a better demo license file via http://quantumwise.com/free-trial/trial-license-form.

Yes, the mobility will be zero in your case. MonkhorstPack sampling will not be suitable, what you want is to sample closely the range around the K point. See http://docs.quantumwise.com/tutorials/mobility.html for a lot more details.