Astrochemical reaction networks: bringing order to chaos

Thesis manuscript on improving astrochemical reaction network by Lorenzo TINACCI, developed in UNIVERSITÉ GRENOBLE ALPES and UNIVERSITÀ DEGLI STUDI DI TORINO under the supervision of Cecilia CECCARELLI and Piero UGLIENGO under the EU-ITN ACO project. Abstract: How molecules are formed, survive and thrive in the Interstellar Medium are the fascinating focus of astrochemistry. Astronomical observations have detected a multitude of molecules (>270) with up to 13 atoms (not considering the PAHs). On the other side, astrochemical models attempt to simulate the processes that govern the interstellar chemical evolution, providing predictions to reproduce the observations. In between, it lies the understanding of the chemistry that occurs in the very peculiar, with respect to the terrestrial, interstellar conditions. The reaction networks are the backbone of astrochemical models and one of the most crucial elements where the chemical knowledge is concentrated. However, due to their complexity and non-linear behaviour on the astrochemical modeling, reaction networks can also be the first source of error in the simulations, if incorrect chemical data are present. During my thesis, I tried to "bring some order" in the commonly employed astrochemical reaction networks using computational chemistry as a tool. The goal of my thesis can be divided into two immediate objectives: (1) improving the gas-phase chemical reaction networks, identifying possible incorrect reactions, and (2) improving the grain-surface chemistry reaction networks, providing new insights of the interstellar thermal desorption process. Objective 1: Present day gas-phase networks (e.g., those in the KIDA and UMIST databases) contain about 8000 reactions involving more than 500 species. Of these, less than 20% have been studied in laboratories or theoretically, and not always in the range of the interstellar densities (103-1010 and temperatures (10-200 K). Thus, my first objective was to "clean" the networks from the most obvious possible error: the presence of endothermic reactions. To this end, using quantum mechanical (QM) calculations, I first created a database of structures and properties for almost all the molecules present in the commonly used gas-phase reaction networks. Then, with these data, I performed a control of the endothermicity for each reaction present in the gas-phase reaction network generated by our group over the years, which is based on the KIDA database. I found that \(\sim\)5% of the reactions are endothermic not recognised as such, of which 2/3 are reported as barrierless. The impact of the incorrect reactions removal was studied via the modeling of typical cold and warm regions, discovering that the reaction network involving Si-bearing species is the most impacted one. After the removal of the incorrect reactions, a new gas-phase reaction network, called GRETOBAPE, was created and made publicly available to the community. In the process, I also made public a reduced network to use when only the major chemical species need to be computed. Objective 2: Two major quantities govern the thermal desorption of species from interstellar ices: the species binding energy (BE) and the pre-exponential factor (\(\nu\)). Usually, astrochemical databases provide only a single or very few BEs and the \(\nu\) is computed from approximate formulae. The BEs and, in very few cases, the \(\nu\) are usually obtained via experiments or computational studies that consider only one or a few water molecules to simulate the ice grain mantle. However, the surfaces of the interstellar ices is likely irregular and a multitude of different adsorbing sites must exist. Thus, in order to better simulate the desorption of molecules from interstellar ices, I developed a new methodology for computing an accurate and unbiased distribution of adsorption sites and the corresponding BEs and \(\nu\). First, a new large ice grain mantle model was obtained via the new ACO-FROST program, developed by our group and to which I substantially contributed. Second, I obtained a distribution of adsorption sites for ammonia and water, where I computed the BE and \(\nu\). The low- and high- end of the BE distribution of water and ammonia might have a major impact in the predicted gaseous abundance of these two species. For example, it could explain the observed presence in the gas-phase of these two molecules in cold regions, such as prestellar cores and protoplanetary outer disks, where thermal desorption would not be expected to be efficient based on the previous literature single BE values. Furthermore, I studied the impact of the commonly used formulae for \(\nu\) to compute the thermal desorption rate in the Temperature-Programmed Desorption (TPD) experiments. To this end, I simulated TPD experiments using the amorphous ice used to derive the ammonia and water BE distributions, described above, and a crystalline ice model. I found that the use of the different formulae cause important differences in how the desorption process is described. Finally, my thesis only scratched the tip of the iceberg of the astrochemical data used to simulate the chemical processes occurring in the interstellar medium. I hope that my contribution has paved the way for more systematic studies of the gas-phase reactions networks and the thermal desorption rate of interstellar molecules, to make the astrochemical model predictions more reliable