Low Rank Approximation Method for Efficient Green's Function Calculation of Dissipative Quantum Transport

Lang Zeng, Institute of Microelectronics, Peking University
Yu He, Network for Computational Nanotechnology, Purdue University, West Lafayette, IN
Michael Povolotskyi, Network for Computational Nanotechnology, Purdue University, West Lafayette, IN
Xiao Yan Liu, Institute of Microelectronics, Peking University
Gerhard Klimeck, Network for Computational Nanotechnology, Purdue University, West Lafayette, IN
Tillmann Kubis, Network for Complutational Nanotechnology, Purdue University, West Lafayette, IN

Date of this Version

6-7-2013

Citation

Journal of Applied Physics: Volume 113, Issue 21

Comments

Copyright (2013) American Institute of Physics. This article may be downloaded for personal use only. Any other use requires prior permission of the author and the American Institute of Physics. The following article appeared in Journal of Applied Physics 113, 213707 and may be found at http://dx.doi.org/10.1063/1.4809638. The following article has been submitted to/accepted by Journal of Applied Physics. Copyright (2013) Lang Zeng, Yu He, Michael Povolotskyi, XiaoYan Liu, Gerhard Klimeck, and Tillmann Kubis. This article is distributed under a Creative Commons Attribution 3.0 Unported License.

Abstract

In this work, the low rank approximation concept is extended to the non-equilibrium Green’s function (NEGF) method to achieve a very efficient approximated algorithm for coherent and incoherent electron transport. This new method is applied to inelastic transport in various semiconductor nanodevices. Detailed benchmarks with exact NEGF solutions show (1) a very good agreement between approximated and exact NEGF results, (2) a significant reduction of the required memory, and (3) a large reduction of the computational time (a factor of speed up as high as 150 times is observed). A non-recursive solution of the inelastic NEGF transport equations of a 1000 nm long resistor on standard hardware illustrates nicely the capability of this new method.

Discipline(s)

Other Electrical and Computer Engineering

 

Share