Abstract
Cloth nappies (or "diapers") are one of the most soiled items of laundry that a household will likely encounter, appropriate laundering practices are important to ensure a consistent and thorough clean. One important parameter is appropriate selection of detergent. The "Clean Cloth Nappies" (CCN) community has put considerable time and effort into developing a database of wash performances of laundry detergents based on user experience. We explore the potential for vibrational spectroscopic methods to rapidly predict wash performance for the end user. Raman and infrared spectra were collected from 41 liquid and 37 powder detergents sourced from Australasia, the UK and North America. The spectroscopic variation across the different formulations was assessed using principal component analysis (PCA). Support vector machine (SVM) classification models were developed for predicting wash performance using two thirds of the samples and the CCN performance database was used for reference wash performance classifications. The remaining third of samples were used to assess the accuracy of predictions of wash performance. Reasonable accuracy for distinguishing between poor and adequate performers was obtained with a 90 % test set accuracy for powder samples and an 82 % test set accuracy for the liquid samples. The ability to distinguish between the need for warm or hot wash conditions was less well classified.