Hierarchical Molecular Structure–Based Method to Estimate the Normal Boiling Point of Aliphatic Oxygen-Containing Organic Compounds

In this work, a hierarchical molecular structure method (HMSM) was proposed to extract molecular descriptors. In this method, the molecular structure is divided into three hierarchies: the first hierarchy involves the contributions of the number of vertices (vertex number m and the sum of vertex number effect, S_VNE); the second hierarchy involves the contributions of the vertex skeleton (odd-even index, OEI, and intramolecular polarization effect index, IMPEI); and the third hierarchy involves the contributions of the functional group (polarization effect index, PEI, and group influencing factor, G_N). Using these six molecular descriptors, the normal boiling points (NBP) of aliphatic oxygen-containing organic compounds (AOCOCs), including alcohols, ethers, aldehydes, ketones, carboxylic acids, and esters, can be quantitatively correlated well. The results show that the average absolute percentage error (APPE) between the experimental and calculated NBP values for each series of AOCOCs is less than 1%. Based on the obtained QSPR models, the NBPs of some AOCOCs that have not been experimentally measured were predicted, including 14 alcohols, 25 ethers, 106 aldehydes, 72 ketones, 103 carboxylic acids, and 57 esters. Additionally, a QSPR model for all 518 AOCOCs was also derived, yielding an APPE value of only 1.35%, indicating that the HMSM-based QSPR model performs exceptionally well in the field of QSPR modeling. The specific contributions of functional groups to NBP were discussed and their order is as follows: COOH > OH > COO > C(O) > CHO > O.