Quantification of polystyrene microplastics in water, milk, and coffee using thermogravimetry coupled with Fourier transform infrared spectroscopy (TGA-FTIR)

Rapid quantification of plastic contaminants, particularly microplastics (MPs), in foods is a challenge. This study introduces a novel method using Fourier transform infrared spectroscopy coupled with thermogravimetric (TGA-FTIR) and chemometric analysis for the quantification of MPs in foods. A model study was performed using polystyrene (PS) MPs (1 μm) added to various foods, namely, water, milk, and coffee without any pretreatment. Foods were spiked with PS microbeads at different concentrations, heated in a TGA, and FTIR spectra of the gases evolved from the TGA were collected over time. The FTIR spectral data were used to construct a Gram-Schmidt profile and identify the characteristic PS peak. The spectrum corresponding to the peak maxima was extracted to represent the specific PS concentration. A dataset of selected spectra and their associated PS concentrations was preprocessed prior to calibration and cross-validation using PLS regression models, for each food matrix studied. The results showed that the PLS models reliably predicted the PS content in water, milk, and coffee with R 2 CV above 0.96, and RMSECV between 0.045 and 0.07 mg. Multivariate detection limit intervals (LOD min /LOD max ) were 0.041/0.085 mg for water, 0.061/0.128 mg for milk and 0.06/0.101 mg for coffee. This method is simple to operate, relatively rapid, and most importantly, does not require sample pretreatment. This research also suggests that the analysis is applicable to a broad range of food samples, and it is suitable for quantifying MPs and nanoplastics regardless of size and shape. The chemometric approach also shows its potential for automation in daily detection and quantification of MPs in food safety control. • Novel TGA-FTIR method quantifies microplastics in food without pretreatment. • High prediction accuracy for PS in water, milk, and coffee with R2CV > 0.96. • PLS regression models tailored for each food matrix, ensuring reliable quantification. • Detection limits: 0.041 mg for water, 0.061 mg for milk, and 0.06 mg for coffee. • Potential for automation in daily detection of microplastics in various food samples.