Assessment of Near-Infrared (NIR) spectroscopy for segregation measurement of low content level ingredients

Abstract

Segregation in particulate systems may be caused by particle size, density and shape distributions leading to negative effects on product quality as well as the production costs. Quantifying powder segregation using a reliable and robust method is challenging, particularly for low content level ingredients. In this paper, we evaluate the application of NIR spectral analysis for detecting the extent of segregation of components in a multi-component mixture. As a model system, a typical laundry detergent formulation, comprising spray-dried powder, known as Blown Powder (BP), Tetraacetylethylenediamine (TAED) and enzyme placebo granules, is used. The effect of using different pre-processing methods on the measured component fractions is analysed. These are scatter correction using Standard Normal Variate (SNV) as well as derivative correction using first, second, Norris-Williams and Savitzky-Golay derivatives. The results from the NIR technique are compared to those obtained by image analysis. Concepts of Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) are used to evaluate the accuracy of different pre-processing methods. The second derivative of Norris-Williams method shows the best pre-processing method for the quantification of low content level enzyme placebo granules in the ternary mixture of detergent powder. Using the proposed NIR technique and the optimum pre-processing method, the segregation index of a low content level ingredient, such as enzyme placebo granules, is estimated to be 0.71 for a ternary heap of washing powders.