A Comparison of Vibrational Spectroscopic Methods for Analyzing Wool/Polyester Textile Blends

Abstract

Vibrational spectroscopic techniques of FT-Raman, mid-infrared attenuated total reflec tance, mid-infrared diffuse reflectance, and near infrared diffuse reflectance are investigated for the quantitative analysis of wool/polyester textile blends. Data sets based on undyed fabric samples are analyzed using both classical least squares and partial least squares approaches. The PLS calibrations based on the ATR and first derivative NIR data are superior with standard errors of cross validation less than 1.5. A validation set consisting of samples extremely different from those in the training set gives standard errors of prediction of 5.2 and 2.4, respectively. These models are used to analyze blended batts and washed fabric samples. Useful results can be obtained independent of such parameters as sample structure and coloration. The complementary nature of these techniques makes their combination valuable as an industrial quality control, monitoring, and trouble shooting tool.