Defining kerogen maturity from orbital hybridization by machine learning

Abstract Kerogen is the primary material for oil and gas. Its maturity is used to determine the potential for hydrocarbon generation. Nowadays, kerogen maturity is mainly measured experimentally and characterized by its chemical composition. The fundamental reason for the change in its chemical composition during the maturation is the breaking and recombination of chemical bonds, manifested by the transformation in atomic hybridization based on quantum mechanics. While traditional methods are time-consuming and labor-intensive, machine learning technique has been introduced to clarify the relationship between hybridization and maturity. A kerogen maturity prediction model based on hybridization is constructed. The average error of the predicted values is only 4.91%, and more than 87% of the test samples have an error of less than 10%. The results demonstrate that the model can accurately predict the maturity of kerogen. As the evolution of kerogen maturity increases the proportion of sp2 hybridized carbons, the orbital hybridization maturity index (OrbHMI) is proposed. The chemical changes in the thermal evolution and pyrolysis mechanism of kerogen can be explained and understood more essentially by OrbHMI. The results provide a basis for guiding artificial maturation and pave a promising path toward studying the kerogen structure and predicting hydrocarbon generating potential.