A Monte Carlo method for solving the NEGF equations for electron transport
Lars Musland, Espen Flage−Larsen, Joakim Bergli
Abstract
Open-access reader
Lars Musland, Espen Flage−Larsen, Joakim Bergli
Abstract
Open-access reader
We derive, by introducing restrictions to the lesser self energy, a Monte Carlo scheme that solves the NEGF equations for electron transport. In doing so we formally prove that the Monte Carlo estimator has an expectation value equal to the lead current of the NEGF solution, and we provide a simple test of the Monte Carlo scheme by calculating conductivity in nanowires within Buttiker's approximation of scattering. Good agreement between the Monte Carlo simulations and the alternative approaches are obtained, and we also demonstrate the existence of a regime where the Monte Carlo method is the fastest method. In our tests this regime is to extreme to be of practical use. We discuss various ways in which to speed up our prototype and how it can be extended to include more physics. Although the full applicability range of our assumptions about the lesser self energy remains to be better understood, we argue that they should apply as long as the transport process can be considered stationary.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
We derive, by introducing restrictions to the lesser self energy, a Monte Carlo scheme that solves the NEGF equations for electron transport. In doing so we formally prove that the Monte Carlo estimator has an expectation value equal to the lead current of the NEGF solution, and we provide a simple test of the Monte Carlo scheme by calculating conductivity in nanowires within Buttiker's approximation of scattering. Good agreement between the Monte Carlo simulations and the alternative approaches are obtained, and we also demonstrate the existence of a regime where the Monte Carlo method is the fastest method. In our tests this regime is to extreme to be of practical use. We discuss various ways in which to speed up our prototype and how it can be extended to include more physics. Although the full applicability range of our assumptions about the lesser self energy remains to be better understood, we argue that they should apply as long as the transport process can be considered stationary.
Key concepts: Monte Carlo method, Statistical physics, Monte Carlo molecular modeling, Dynamic Monte Carlo method, Quantum Monte Carlo, Monte Carlo method in statistical physics, Physics, Estimator