Identifying Barrierless Mechanisms for Benzene Formation in the Interstellar Medium Using Permutationally Invariant Reaction Discovery

Abstract

Complex chemical reaction environments, such as those found in combustion engines, the upper atmosphere, or the interstellar medium, can contain large numbers of different reactive species participating in similarly large numbers of different chemical reactions. In such settings, identifying the most-likely multistep reaction mechanisms which lead to the production of a particular defined product species is an extremely challenging problem, requiring search and evaluation over a large number of different possible candidate mechanisms while also addressing the permutational challenges posed when considering a large number of reaction routes available to sets of identical molecular species. In this article, the problem of generating candidate reaction mechanisms which form a defined product from a diverse set of reactive molecules is cast as a discrete optimization of a permutationally invariant cost function describing similarity between the target product and the product generated by a trial reaction mechanism. This approach is demonstrated by generating 2230 candidate reaction mechanisms which form benzene from diverse sets of reactive molecules which have been experimentally identified in the interstellar medium. By screening this set of autogenerated mechanisms, using dispersion-corrected DFT to evaluate reaction energies and activation barriers, we identify several candidate barrierless reaction mechanisms (both previously proposed and new) for benzene formation which may operate in the low temperatures found in the interstellar medium and could be investigated further to supplement existing microkinetic models.

AutoMeKin2021: An open‐source program for automated reaction discovery

“…The method also benefits from working in graph space, where reactions are simply graph transformations. This has been shown to be a practical way to explore reactive events by Habershon and co‐workers 26,27,97–100 …”

“…This has been shown to be a practical way to explore reactive events by Habershon and co-workers. 26,27,[97][98][99][100] The workflow of ChemKnow is detailed as follows:…”

Graph-Driven Reaction Discovery: Progress, Challenges, and Future Opportunities

“…So, as noted above, complex chemical reaction set-ups (spanning from highly controlled, laboratory-based reaction set-ups to unobserved reactions occurring on ice grains in interstellar space) ,,− can in principle be mapped onto CRNs. However, now consider turning this viewpoint on its head; if we can directly generate a CRN describing and characterizing the full set of reactions and chemical species in some reactive system, then we would have an in silico method which is capable of predicting how complex chemical systems will evolve.…”

“…The generation of thermodynamic and kinetic data for all of the elementary chemical reactions in a CRN using methods based on ab initio quantum chemistry is itself an enormously active field of research; several relevant challenges in this area are discussed later. However, the focus of this article is instead on automatic reaction discovery (ARD) ,,,,,,,− ,− methods that can be used to “grow” a CRN (namely, the set of all reactive chemical species and the allowed chemical reactions).…”

“…Our own work in this field ,,− and the focus of this article is exclusively on methods which fall into the second CRN generation category (Figure ), being based on concepts of molecular graphs. As described below, graphs such as adjacency matrices offer a compact description of the chemical bonding in a collection of molecules; put simply, such graphs identify which atoms are bonded and which are not.…”