Enthalpies of formation and reaction are important thermodynamic properties that have a crucial impact on the outcome of chemical transformations. Here we implement the calculation of enthalpies of formation with a general-purpose ANI-1ccx neural network atomistic potential. We demonstrate on a wide range of benchmark sets that both ANI-1ccx and our other general-purpose data-driven method AIQM1 approach the coveted chemical accuracy of 1 kcal/mol with the speed of semiempirical quantum mechanical methods (AIQM1) or faster (ANI-1ccx). It is remarkably achieved without specifically training the machine learning parts of ANI-1ccx or AIQM1 on formation enthalpies. Importantly, we show that these data-driven methods provide statistical means for uncertainty quantification of their predictions, which we use to detect and eliminate outliers and revise reference experimental data. Uncertainty quantification may also help in the systematic improvement of such datadriven methods.

Deriving diverse compound libraries from a single substrate in high yields remains to be a challenge in cycloparaphenylene chemistry. In here, a strategy for the late-stage functionalization of shape-persistent alkyne-containing cycloparaphenylene has been explored using readily available azides. The copper-free [3 + 2]azide-alkyne cycloaddition provided high yields (> 90 %) in a single reaction step. Systematic variation of the azides from electron-rich to -deficient shines light on how peripheral substitution influences the characteristics of the resulting adducts. We find that among the most affected properties are the molecular shape, the oxidation potential, excited state features, and affinities towards different fullerenes. Joint experimental and theoretical results are presented including calculations with the state-of-the-art, artificial intelligence-enhanced quantum mechanical method 1 (AIQM1).

No abstract