Thermal behavior of ferulic acid employing isoconversional models and artificial neural network

Ferulic acid (FA) is a phenolic acid of plant kingdom presenting antioxidant activity, a fundamental pharmaceutical property. This property suggests FA can be used in cosmetic skin formulations as a photoprotective and anti-aging agent. The purpose of this work is to investigate the kinetics of FA thermal decomposition process in non-isothermic conditions applying Friedman, Flynn–Wall–Ozawa and Kissinger–Akahira–Sunose methodologies and in isothermal conditions using a neural network. All these isoconversional results showed coherent values of apparent activation energy under nitrogen atmosphere. For the isothermal analysis, R2 model presented best performance to individually describe the data. However, the neural network assumed the decomposition as a combined event, in which ten models have contributions to describe experimental data. The DRX results showed the sample is not at steadier configuration and require a pre-treatment before the analysis by the non-isothermal experiments. From this, the sample was prepared with heat treatment up to 130 °C, and the determined activation energy showed reduction of 5 kJ mol⁻¹. The isothermal analysis endorses activation energy about 40 kJ mol⁻¹. These results proved the FA thermal decomposition is strongly influenced by experimental conditions.