Computing Binding Energies of Interstellar Molecules by Semiempirical Quantum Methods: Comparison Between DFT and GFN2 on Crystalline Ice

Interstellar Grains (IGs) spread in the Interstellar Medium (ISM) host a multitude of chemical reactions that could lead to the production of interstellar Complex Organic Molecules (iCOMs), relevant in the context of prebiotic chemistry. These IGs are composed of a silicate-based core covered by several layers of amorphous water ice, known as a grain mantle. Molecules from the ISM gas-phase can be adsorbed at the grain surfaces, diffuse and react to give iCOMs and ultimately desorbed back to the gas phase. Thus, the study of the Binding Energy (BE) of these molecules at the water ice grain surface is important to understand the molecular composition of the ISM and its evolution in time. In this paper, we propose to use a recently developed semiempirical quantum approach, named GFN-xTB, and more precisely the GFN2 method, to compute the BE of several molecular species at the crystalline water ice slab model. This method is very cheap in term of computing power and time and was already showed in a previous work to be very accurate with small water clusters. To support our proposition, we decided to use, as a benchmark, the recent work published by some of us in which a crystalline model of proton-ordered water ice (P-ice) was adopted to predict the BEs of 21 molecules relevant in the ISM. The relatively good results obtained confirm GFN2 as the method of choice to model adsorption processes occurring at the icy grains in the ISM. The only notable exception was for the CO molecule, in which both structure and BE are badly predicted by GFN2, a real pity due to the relevance of CO in astrochemistry.