Hyperspectral imaging was used for qualitative and quantitative monitoring of the distribution of a flame retardant on polyester fabrics. NIR reflection spectra show a specific band related to the flame retardant, which rises with increasing application weight. Multivariate data analysis tools based on the partial least squares (PLS) algorithm were applied for quantification of the spectra. Gravimetry was used as a reference method for the characterization of the calibration samples. The calibration method was optimized by the application of several spectral pretreatments and variation in the spectral range considered in the various models, which finally resulted in a prediction error of about 1.3 g/m2. The prediction performance of the developed calibration model was proven in external validations using independent samples with application weights between about 5 and 25 g/m2. Apart from the quantification, the homogeneity of the distribution of the flame retardant was investigated. It was shown that non-uniform distributions (e.g., gradients, droplets, irregular) can be detected by hyperspectral imaging. Some fabric samples were finished using a special ink jet printing technology for application to the polyester fabric. The spectral images of printed samples based on the previous calibration model achieved for samples made by impregnation do not only clearly show the different degrees of functionalization, but also the outstanding homogeneity of the distribution of the flame retardant. Moreover, printed samples finished with two different agents were analyzed.
Daikos et al. (2026) studied this question.