A comprehensive collisional-radiative model is developed and coupled with a one-dimensional two-temperature fluid model to study capacitively coupled radio-frequency argon discharges at pressures ranging from 0.5 to 5 Torr and electrode peak-to-peak voltages of 150 and 300 V. The fluid model is formulated using the drift-diffusion approximation. The CR model accounts for the argon ground state, 30 excited states in the 4 s, 4p, 3d, and 5 s manifolds, electrons, atomic (Ar+) and molecular (Ar2+) ions, and argon excimer molecules (Ar2∗). The processes included in the model are electron-impact excitation and ionization, radiation emission, heavy-particle collisions, radiative recombination, and processes involving Ar2+ and Ar2∗ molecules. Non-Maxwellian electron energy distribution functions, calculated using Bolsig+, are considered, with fine-structure collisional cross-sections acquired from the LXCat database. A detailed comparison is presented between numerical predictions and experimental measurements from a glow discharge device developed at UT Austin, focusing on populations of excited states in the 4p and 4 s manifolds, and electron number density. Additional comparisons are made with experimental data and PIC/Monte Carlo simulations from the study by (Donkó et al 2023 Plasma Sources Sci. Technol. 32 065002). While agreement with the Donkó et al data is somewhat improved—particularly with the PIC/Monte Carlo results—the CR-fluid model systematically overpredicts the experimental values. Possible contributing factors to these discrepancies and improvements to the model are discussed. In addition, the sensitivity of the results to the EEDF modeling is assessed by comparing simulations using Bolsig+-derived, Druyvesteyn, and Maxwellian distributions. The results show that predictions of electron temperature and plasma potential are strongly affected by the assumed EEDF shape. The inclusion of higher-lying excited states beyond the 4p manifold was found to have a minimal impact on discharge dynamics under the conditions studied. The code has been made publicly available on GitHub as open-source software.
Tsagkaridis et al. (Wed,) studied this question.