# A Fully ab Initio Kinetic Monte Carlo Approach for Modeling Adsorption and Diffusion in Interstellar Icy Grain Mantles: The Case of H2S

## TL;DR

This paper presents a new computational method to model how molecules like H2S move and react on ice in space, showing that diffusion is very slow at low temperatures.

## Contribution

A fully ab initio kinetic Monte Carlo framework for modeling diffusion and adsorption on interstellar ices with detailed energy barriers.

## Key findings

- H2S diffusion is negligible below 20 K, with coefficients as low as 10–48 cm² s⁻¹ at 10 K.
- Diffusion has minimal impact on TPD peak positions under submonolayer conditions.
- A universal scaling factor for diffusion barriers based on binding energy does not apply due to variability.

## Abstract

Understanding diffusion on interstellar ices is key to
modeling
the chemical evolution of cold molecular clouds, where low temperatures
severely limit molecular mobility. In this study, we introduce a robust
and fully automated multiscale computational framework to quantify
diffusion processes of adsorbates at the surface of amorphous solid
water (ASW). Using H2S as a test case, whose binding sites
were previously studied at the ab initio level, we constructed a detailed
network of 141 adsorption sites connected by over 270 transition states.
All density functional energetics were benchmarked against DLPNO–CCSD­(T),
achieving chemical accuracy in the description of diffusion barriers,
which span from 0.1 to 27 kJ mol–1 with a median
value of 5.4 kJ mol–1. An off-lattice kinetic Monte
Carlo (kMC) model adopting both the ab initio diffusion barriers and
binding energies for the desorption processes was carried out to compute
temperature-dependent diffusion coefficients and to reconstruct the
temperature-programmed desorption (TPD) curve. Our simulations reveal
that thermal diffusion of H2S is negligible below 20 K,
with diffusion coefficients as low as 10–48 cm2 s–1 at 10 K, thus excluding Langmuir–Hinshelwood
surface encounters under typical dense cloud conditions. Moreover,
under submonolayer conditions, diffusion was found to have negligible
influence on the reconstructed TPD peak position. Furthermore, our
results demonstrate that a universal scaling factor f to guess the diffusion barriers (ΔE
diff) from the sole knowledge of BE: f = ΔE
diff/BE does not apply as it exhibits wide variability
across the sampled configurations. These findings highlight the need
for incorporating statistically meaningful distributions of binding
energies and diffusion barriers in astrochemical models to more accurately
capture diffusion and surface reactivity on interstellar ices.