Kinetic analysis and prediction of thermal process contaminants formation in macadamias and walnuts under different...

Roasting nuts enhances flavor and color but triggers the formation of harmful compounds, including acrylamide, 5-hydroxymethylfurfural, N^ε-carboxymethyllysine and N^ε-carboxyethyllysine. However, systematic studies on the concurrent formation patterns of these thermal process contaminants in nuts remain limited. Additionally, existing detection methods for these contaminants are complicated and time-consuming. This study aimed to investigate the formation kinetics of these four contaminants during macadamia and walnut roasting and to establish rapid prediction models.

The investigation revealed that the formation of these four contaminants followed first-order kinetics, with apparent activation energies ranging from 17.29 to 44.19 kJ mol^-1. Prediction models were established using color parameters (L*, a* and b*) and moisture content as predictors. Both partial least-squares regression (PLS-R) and support vector regression (SVR) models achieved high prediction accuracy, with the SVR model outperforming PLS-R in terms of robustness, showing higher R² and lower root mean square deviation values. These results demonstrate a clear correlation between physical parameters and chemical contaminant levels.

This study provides the evidence of concurrent formation patterns of multiple thermal process contaminants in nuts and proposes a rapid non-destructive prediction tool, offering practical strategies for controlling hazardous compounds during industrial roasting processes. © 2026 Society of Chemical Industry.