搜索

x
中国物理学会期刊

低气压容性放电流体模拟的数值稳定和可靠性研究

Numerical Stability and Reliability Study of Fluid Simulations for Low-Pressure Capacitively Coupled Discharges

PDF
导出引用
  • 流体模型是一类广泛用于低温气体放电宏观模拟的连续介质方法,能够以较低计算成本描述射频容性耦合等离子体(RF-CCP)中的密度分布、鞘层运动、电子加热和电离过程。然而,作为在时间和空间上离散求解的数值模型,其稳定性和物理准确性依赖于空间分辨率和界面通量构造方式。本文以一维 13.56 MHz 氩气 RF-CCP 放电为对象,选取 1 Torr / 40 V 和 2 Torr / 200 V两组工况,探讨了空间网格和通量构造方式对电子通量、鞘层结构、功率吸收、电离源项、电子密度以及电流密度的影响。结果表明,电子通量在粗网格条件下首先受到界面通量构造方式的影响,并表现出不同的空间输运强度。电子通量差异会进一步改变鞘层厚度、电子吸收功率,并通过平均电子温度和电离速率系数反馈到电离源项,最终影响电子密度分布,此外,总功率密度和总电流密度的幅值也受到界面通量构造方式的影响,因此,在本文所考察的一维低气压氩气 RF-CCP流体模拟中,数值收敛并不意味着所得物理量已经准确可靠,应结合电子通量、鞘层演化、电子吸收功率、电离源项以及电流密度评估收敛解的可信性。

     

    Fluid models are widely used in macroscopic simulations of low-temperature gas discharges because they can describe plasma density, sheath motion, electron heating, ionization, and current response at a relatively low computational cost. However, in radio-frequency capacitively coupled plasmas (RF-CCPs), steep gradients in electron density and electric field, together with strong variations in charged-particle fluxes, occur near the oscillating sheath edges. Therefore, a numerically converged periodic solution does not necessarily guarantee physically reliable predictions. This work investigates how spatial resolution and interfacial flux discretization affect the predicted discharge behavior in a one-dimensional 13.56 MHz argon RF-CCP. A drift-diffusion fluid model based on the finite-volume method and a staggered grid is employed. The same governing equations, boundary conditions, transport and reaction data, and time-advancement strategy are used in all simulations so that the observed differences can be attributed to the spatial discretization of charged-particle transport. Two representative discharge conditions are considered: 1 Torr/40 V and 2 Torr/200 V. Four commonly used interfacial flux treatments are compared, namely the Scharfetter–Gummel exponential flux, the first-order upwind flux, the central-difference flux, and the van Leer total-variation-diminishing flux-limiter scheme. The analysis focuses on their physical consequences for electron flux, sheath thickness, electron power absorption, mean electron temperature, ionization source term, electron density, and current density. The results show that, under coarse-grid conditions, the influence of the interfacial flux treatment is first manifested in the electron-flux distribution. Through the coupling among the electron continuity equation, the Poisson equation, and the electron energy equation, flux-discretization errors produce correlated deviations in electron transport, space charge and electric field, sheath structure, electron power absorption, mean electron temperature, ionization source term, and electron density.Different flux treatments introduce distinct systematic biases into the predicted physical quantities. The first-order upwind flux introduces strong numerical dissipation, which weakens electron transport near the sheath edges and leads to underestimation of electron power absorption, ionization source term, and electron density. The central-difference flux is prone to nonphysical oscillations in steep-gradient sheath regions; in the 1 Torr/40 V case, it produces artificial peaks in the mean electron temperature and ionization source term, whereas in the 2 Torr/200 V case, the simulation fails to reach a stable periodic state. The Scharfetter–Gummel flux performs reasonably well on the coarse grid in the 1 Torr/40 V case, but in the 2 Torr/200 V case, it overestimates the electron flux and electron density, shifts the sheath boundary toward the electrode, and reduces the cycle-averaged sheath thickness by approximately 40.1\% relative to the reference solution. This sheath compression further enhances near-electrode electron power absorption and local ionization. The van Leer TVD flux exhibits comparatively robust behavior by suppressing central-difference oscillations and avoiding the strong overestimation produced by the Scharfetter–Gummel flux, although limiter-induced dissipation may still lead to underestimation of the electron density and ionization source term in steep-gradient regions. These findings demonstrate that spatial grid resolution and interfacial flux discretization should not be regarded as minor implementation details in RF-CCP fluid simulations. For coarse-grid simulations, multidimensional modeling, and large-scale parameter scans, the reliability of a fluid solution should be assessed by jointly examining electron flux, sheath evolution, electron power absorption, mean electron temperature, ionization source term, electron density, and current density.

     

    目录

    /

    返回文章
    返回