Physics-informed bayesian MCMC analysis of Krypton/Argon sputter yields

Electric propulsion spacecraft increasingly require alternative propellants such as krypton and argon due to xenon’s high cost, yet sputter yield data for these propellants remain limited for many spacecraft-relevant materials. This data scarcity makes empirical parameter fitting difficult, particularly near the low-energy threshold region that is important for plume-surface interaction modeling. We develop an LSS-informed Bayesian Markov Chain Monte Carlo framework for calibrating Eckstein sputter-yield parameters for krypton and argon systems. The approach uses Lindhard-Scharff-Schiøtt (LSS) collision theory to construct LSS-informed prior information from well-characterized xenon reference systems while retaining the available experimental data in the likelihood. Predictive behavior is evaluated through comparison with data-only MCMC baselines and retrospective withheld-data validation. The method is applied to 16 krypton and 6 argon target materials spanning metallic and covalent systems, providing a consistent set of LSS-informed Eckstein parameter fits for electric-propulsion sputter-yield modeling.