We will develop and employ novel automated methods to investigate systematically which reaction pathways lead to the formation of complex organic molecules. Initial focus will be on the 4 identified key reactions. Since such multi-step reaction pathways unfold with exponentially many combinatorial possibilities it is generally impossible to apply brute force mapping in this context. We will formulate strategies that employ graph convolutional methods and reinforcement learning to tackle this problem. Barriers for reactions at the catalytic sites identified under theme 1 will subsequently be evaluated and recurring intermediates will be compared to observations and astrochemical models.
Using in operando STM, AFM, XPS, mass spectrometry and temperature programmed desorption (TPD) measurements on realistic dust grain structures synthesized under Theme 1 and under conditions mimicking those found in diffuse interstellar clouds we will benchmark reaction pathways identified under WP2-1. The measured catalytic yields and branching ratios will be correlated to catalytically active sites available on the surface. The resulting molecular yields will serve as input for astrochemical models where derived branching ratios allow direct comparisons with JWST and ALMA observations.
Initial key reactions: Caproic acid formation Cytosine formation