Filling the Gap of the Heats of Formation for Organophosphorus Species in the NIST Chemistry WebBook with Reliable Quantum-Chemistry Computations.

We have used high-level quantum-chemistry methods to obtain the heat of formation (HOF) for a set of 539 organophosphorus species in the NIST database, for which reported values are not available. Machine learning is then used to divide the HOFs into their atomic contributions. We find that the σ-electron-withdrawing CF3 group stabilizes the bonded phosphorus, while triple-bonded substituents tend to be thermochemically destabilizing. The atomic contributions are also shown to be useful for rationalizing reactivity, notably when thermochemical stability does not reflect chemical inertness. Furthermore, we have revised the optimal atomic values used for determining the HOFs by atomization with more economical methods and parametrized them for additional low-cost protocols. In particular, the combination of ωB97M-V and a revised vDZP basis set emerges as a good balance between cost and performance, making it suitable for obtaining an extensive set of HOF data with good statistical accuracy.