Robust binding energy distribution sampling on amorphous solid water models

Context. The astrochemically efficient icy mantles surrounding dust grains in molecular clouds have been shown to be of an amorphous water-rich nature. This therefore implies a distribution of binding energies (BEs) per species instead of a single value. Methods proposed so far for inferring BEs and their distributions on amorphous ices rely on different approaches and approximations, leading to disparate results or BE dispersions with partially overlapping ranges. Aims. This work aims to develop a method based on a structurally reliable ice model and a statistically and physicochemically robust approach to BE distribution inference, with the aim of being applicable to various relevant interstellar species. Methods. A multiscale computational approach is presented, with a molecular dynamics heat and quench protocol for the amorphous water ice model, and an ONIOM(B3LYP-D3(BJ)/6-311+G(d,p):GFN2-xtb) scheme for the BE inference, with a prime emphasis onto the convergence of the computed BEs with the real system size. The sampling of the binding configurations is twofold, exploring both regularly spaced binding sites as well as various adsorbate-to-substrate orientations on each locally distinct site. This second source of BE diversity accounts for the local roughness of the potential energy landscape of the substrate. Three different adsorbate test cases are considered, NH3, CO, and CH4, owing to their significance in icy dust mantles, and their distinct binding behavior with water ices. Results. The BE distributions for NH3, CO, and CH4 have been inferred with converged statistics. The distribution for NH3 is better represented by a double Gaussian component profile. Three starting adsorbate orientations per site are required to reach convergence for both Gaussian components of NH3, while two orientations are sufficient for CO , and one unique one for CH4 (symmetric). Further geometrical and molecular surrounding insights have been provided. These results encompass previously reported results.