Comparison of the predictions of Langevin Dynamics-based diffusion charging collision kernel models with canonical experiments

Based on the prior work of Chahl and Gopalakrishnan (2019) to infer particle-ion collision time distributions using a Langevin Dynamics (LD) approach, we develop a model for the non-dimensional particle-ion diffusion charging collision kernel beta(i) or H that is applicable for 0 <= Psi(E) <= 60, 0 < Psi(1)/Psi(E) < 1, Kn(D) 2000 (defined in the main text). The developed model for beta(i) for attractive Coulomb and image potential interactions, along with the model for beta(i) for repulsive Coulomb and image potential interactions from Gopalakrishnan, Thajudeen, Ouyang, and Hogan (2013b), is tested against published diffusion charging experimental data. Current state of the art charging models, Fuchs (1963) and Wiedensohler (1988) regression for bipolar charging, are also evaluated and discussed. Comparisons reveal that the LD-based model accurately describes unipolar fractions for 10 - 100 nm particles measured in air (Adachi, Kousaka, & Okuyama, 1985), nitrogen and argon but not in helium (Adachi, Okuyama, Kousaka, Kozuru, & Pui, 1987). Fuchs model and the LD-based model yield similar predictions in the experimental conditions considered, except in helium. In the case of bipolar charging, the LD-based model captures the experimental trends quantitatively (within +/- 20%) across the entire size range of 4 - 40 nm producing superior agreement than Wiedensohler's regression. The latter systematically underpredicts charge fraction below (similar to)20 nm in air (by up to 40%) for the data presented in Adachi et al. (1985). Comparison with the data of Gopalakrishnan, McMurry, and Hogan (2015), obtained in UHP air along with measurements of the entire ion mass-mobility distribution, shows excellent agreement with the predictions of the LD-based model. This demonstrates the capability to accommodate arbitrary ion populations in any background gas, when such data is available. Wiedensohler's regression, derived for bipolar charging in air using average ion mass-mobility, also describes the data reasonably well in the conditions examined. However, both models failed to capture the fraction of singly and doubly charged particles in carbon dioxide warranting further investigation.