Experimental validation of a reaction network model for autoxidation of linoleate esters

A deterministic mathematical model is used to study complex chemistry and determine kinetic parameters of the autoxidation polymerization (drying) of fatty acid esters such as ethyl and methyl linoleate (EL and ML). The model was developed previously and is based on an algorithm that generates a reaction network and the associated kinetic equations in an automated way (automatic reaction network generation – ARNG). The ARNG model computes the concentrations of the monomeric species, including the number and type of crosslinks (peroxyl, ether, alkyl) they possess. In this study, kinetic parameters of the main reaction families were estimated by adjusting them to obtain a good fit with new experimental data of various nature. Among these were: Fourier transform infrared (FTIR) spectra, electron spray ionization mass spectrometry (ESI-MS) data of EL with TiO₂ cured at 70 °C and size exclusion chromatography (SEC) data of ML cured at 80 °C. Sensitivity analysis revealed that addition, bis-allylic hydrogen abstraction by peroxyl radical and peroxyl radical recombination are the most influential reactions in the considered set of functional groups. These sensitivity results indicated which reaction families were considered in the parameter estimation procedure. Rate coefficients of key reactions like hydrogen abstraction, O2 to radical addition, OO- and O-radical addition to conjugated double bonds and β-scission could be well be determined from EL-conversion and conjugated double bond profiles from FTIR. Some tens of dominant peaks in the ESI-MS spectrum were also found in the ARNG model, while vice versa most dominant model predicted peaks were confirmed by ESI-MS. ARNG allowed detailed chemical pathway analysis in terms of intermediate products, which was helpful in elucidating the pathway for the most important acidic monomer, hexanoic acid. Finally, the most dominant crosslink type formed turned out to be peroxyl, less ether, while alkyl is practically absent. The ARNG model of linoleate drying has thus considerably gained power to predict properties relevant to the chemical structure of EL polymer.