Asymptotic Preserving Discontinuous Galerkin Methods for a Linear Boltzmann Semiconductor Model

Victor DeCaria, Cory D. Hauck, Stefan SchnakearXiv 2022 · DOI · arXiv

Abstract

A key property of the linear Boltzmann semiconductor model is that as the collision frequency tends to infinity, the phase space density $f = f(x,v,t)$ converges to an isotropic function $M(v)\rho(x,t)$, known as the drift-diffusion limit. In this limit, $M$ is a Maxwellian and the physical density $\rho$ satisfies a second-order parabolic PDE called the drift-diffusion equation. Numerical approximations that mirror this transition are termed "asymptotic preserving" (AP). This paper develops two discontinuous Galerkin (DG) methods for the semiconductor model: one utilizing a standard upwinding flux and the other an $\varepsilon$-scaled Lax-Friedrichs flux, where $\varepsilon$ represents the inverse of the collision frequency. We demonstrate that these schemes are uniformly stable in $\varepsilon$ and satisfy the AP property. Furthermore, we identify the necessary properties for the discrete Maxwellian to ensure convergence to an accurate $h$-approximation of the drift-diffusion limit, providing discrete versions of the drift-diffusion equation and rigorous error estimates.

Introduction

In semiconductor physics, modeling electron transport presents a significant multiscale challenge. At the microscopic level, the Linear Boltzmann Equation tracks particle distribution in a high-dimensional (6D) phase space of position and velocity. While accurate, this kinetic description is computationally expensive. At the macroscopic level, when particle collisions are frequent, the system simplifies into the Drift-Diffusion Equation, which focuses solely on spatial density.

The transition between these scales is governed by a scaling parameter $\varepsilon$. As $\varepsilon \to 0$ (high collision frequency), standard numerical methods often become "stiff," requiring prohibitively small grid sizes and time steps to maintain stability. To circumvent this, researchers develop Asymptotic Preserving (AP) schemes. An AP scheme is a numerical framework that remains stable and consistent across scales, automatically transitioning from a kinetic solver to an efficient macroscopic solver as $\varepsilon$ vanishes, without requiring the user to resolve the microscopic collision scales.

Methodology

This work employs Discontinuous Galerkin (DG) methods, which are highly regarded for their ability to handle complex geometries and accommodate discontinuities in the solution across grid cell boundaries. The authors introduce several critical technical innovations to ensure the AP property:

1. Weighted Reformulation

To improve numerical stability and simplify the collision operator, the Boltzmann equation is rewritten using a weighted distribution: $g = M^{-1/2}f$. This transformation ensures that the numerical energy estimates remain well-behaved even in the high-collision limit.

2. Discrete Maxwellian Construction

A central contribution of this research is the rigorous definition of the "numerical" Maxwellian. The authors demonstrate that simply sampling a continuous Maxwellian onto a finite grid is insufficient. For the simulation to converge to the correct physical temperature in the drift-diffusion limit, the discrete velocity space must satisfy specific symmetry and integral properties. Failure to meet these criteria leads to inaccurate macroscopic approximations.

3. Flux Variants

The study evaluates two distinct methods for handling numerical flux between grid cells:

  • Standard Upwinding Flux: A traditional approach in transport problems.
  • $\varepsilon$-scaled Lax-Friedrichs Flux: A modified flux that incorporates the scaling parameter to maintain stability across regimes.

By introducing a parameter $\beta$, the authors provide a unified mathematical framework to analyze both flux types simultaneously.

Theoretical Results and Analysis

As a theoretical numerical analysis, the validity of these schemes is established through rigorous functional analysis and mathematical proofs rather than empirical datasets.

Convergence and Stability

The paper proves Theorem 4.1, which establishes that as the collision parameter $\varepsilon \to 0$, the DG solution for the kinetic model converges to a valid DG solution for the macroscopic drift-diffusion equation. This "uniform stability" ensures that the numerical error does not explode as the system becomes more collisional.

Error Estimates

Concrete error bounds are provided in Theorem 5.1. The analysis shows that the error in the density $\rho$ is approximately $O(\sqrt{\varepsilon}/h_x + h_x^k)$, where $h_x$ is the spatial grid size and $k$ is the polynomial degree. This suggests an optimal scaling strategy: for certain configurations, setting the grid size $h_x$ proportional to $\varepsilon^{1/3}$ yields the most efficient convergence.

Avoiding Numerical Locking

The authors identify specific conditions, such as the choice of polynomial degree $k_x$, where the simulation might suffer from "locking"—a phenomenon where the numerical solution becomes artificially inaccurate in the limit. These findings provide a roadmap for selecting appropriate discretization parameters in engineering applications.

Discussion and Future Work

While this work provides a rigorous foundation for AP-DG methods in semiconductor modeling, it operates under several simplifying assumptions that leave room for future exploration:

  • Linearity: The current model assumes a fixed, given electric field. Real-world applications involve the non-linear Poisson-Boltzmann problem, where particles influence the electric field they inhabit.
  • Boundary Conditions: The analysis focuses on zero-inflow boundaries. Practical semiconductor devices involve contacts and leads that create complex boundary layers requiring further study.
  • Initial Data: The proofs assume the system begins in a relatively balanced state. Investigating the behavior of these schemes when starting far from equilibrium (non-isotropic initial data) remains a priority for future research.

Summary updated

Abstract

A key property of the linear Boltzmann semiconductor model is that as the collision frequency tends to infinity, the phase space density $f = f(x,v,t)$ converges to an isotropic function $M(v)ρ(x,t)$, called the drift-diffusion limit, where $M$ is a Maxwellian and the physical density $ρ$ satisfies a second-order parabolic PDE known as the drift-diffusion equation. Numerical approximations that mirror this property are said to be asymptotic preserving. In this paper we build two discontinuous Galerkin methods to the semiconductor model: one with the standard upwinding flux and the other with a $\varepsilon$-scaled Lax-Friedrichs flux, where 1/$\varepsilon$ is the scale of the collision frequency. We show that these schemes are uniformly stable in $\varepsilon$ and are asymptotic preserving. In particular, we discuss what properties the discrete Maxwellian must satisfy in order for the schemes to converge in $\varepsilon$ to an accurate $h$-approximation of the drift diffusion limit. Discrete versions of the drift-diffusion equation and error estimates in several norms with respect to $\varepsilon$ and the spacial resolution are also included.
BibTeX
@misc{decaria2022asymptotic,
  author = {Victor DeCaria and Cory D. Hauck and Stefan Schnake},
  title = {{Asymptotic Preserving Discontinuous Galerkin Methods for a Linear Boltzmann Semiconductor Model}},
  year = {2022},
  eprint = {2206.09805},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2206.09805},
}
Figure