Vibrational spectroscopic methods for the overall quality analysis of washing powders

Abstract The aim of this study was to compare and evaluate the ability of near infrared- (NIR), Raman- and attenuated-total-reflection infrared (ATR-IR) spectroscopy as tools for the identification of washing powder brands as well as for an overall quantitative analysis of all ingredients of the analyzed laundry detergents. The laundry detergents used in this work were composed of 22 different ingredients. For this purpose, principal component analysis (PCA) cluster models and partial least-squares (PLS) regression models were developed and different data pre-processing algorithms such as standard-normal-variate (SNV), multiplicative scatter correction (MSC), first derivative BCAP (db1), second derivative smoothing (ds2), smoothing Savitzky Golay 9 points (sg9) as well as different normalization procedures such as normalization between 0 and 1 (n01), normalization unit length (nle) or normalization by closure (ncl) were applied to reduce the influence of systematic disturbances. The performance of the methods was evaluated by comparison of the number of principal components (PCs), regression coefficient ( r ), Bias, Standard error of prediction (SEP), ratio performance deviation (RPD) and range error ratio (RER) for each calibration model. For each of the 22 ingredients separate calibration models were developed. Raman spectroscopy was suitable for the analysis of only two ingredients (dye transfer inhibitor 1 and surfactant 6) and it was not possible to record all Raman spectra due to high fluorescence. NIR and ATR-IR are powerful methods to analyze washing detergents with low numbers of PCs being necessary, regression coefficients of only little below 1, small Biases and SEPs compared to the range and high RPDs and RERs