Floating in Space: How to Treat the Weak Interaction between CO Molecules in Interstellar Ices

[Image: see text] In the interstellar medium, six molecules have been conclusively detected in the solid state in interstellar ices, and a few dozen have been hypothesized and modeled to be present in the solid state as well. The icy mantles covering micrometer-sized dust grains are, in fact, thought to be at the core of complex molecule formation as a consequence of the local high density of molecules that are simultaneously adsorbed. From a structural perspective, the icy mantle is considered to be layered, with an amorphous water-rich inner layer surrounding the dust grain, covered by an amorphous CO-rich outer layer. Moreover, recent studies have suggested that the CO-rich layer might be crystalline and possibly even be segregated as a single crystal atop the ice mantle. If so, there are far-reaching consequences for the formation of more complex organic molecules, such as methanol and sugars, that use CO as a backbone. Validation of these claims requires further investigation, in particular on acquiring atomistic insight into surface processes, such as adsorption, diffusion, and reactivity on CO ices. Here, we present the first detailed computational study toward treating the weak interaction of (pure) CO ices. We provide a benchmark of the performance of various density functional theory methods in treating the binding of pure CO ices. Furthermore, we perform an atomistic and in-depth study of the binding energy of CO on amorphous and crystalline CO ices using a pair-potential-based force field. We find that CO adsorption is represented by a large distribution of binding energies (200–1600 K) on amorphous CO, including a significant amount of weak binding sites (<350 K). Increasing both the cluster size and the number of neighbors increases the mean of the observed binding energy distribution. Finally, we find that CO binding energies are dominated by dispersion and, as such, exchange-correlation functionals need to include a treatment of dispersion to accurately simulate surface processes on CO ices. In particular, we find the ωB97M-V functional to be a strong candidate for such simulations.