Estimation of the particle size distribution of a dilute latex from combined elastic and dynamic light scattering measurements: A method based on neural networks

摘要

A method for estimating the particle size distribution (PSD) of a dilute latex from light scattering measurements is presented. The method utilizes a general regression neural network (GRNN), that estimates the PSD from 2 independent sets of measurements carried out at several angles: (i) light intensity measurements, by elastic light scattering (ELS); and (ii) average diameters measurements, by dynamic light scattering. The GRNN was trained with measurements simulated on the basis of typical asymmetric PSDs (unimodal normal-logarithmic distributions of variable mean diameters and variances). First, the ability of the method was tested on the basis of two synthetic examples. Then, the obtained GRNN was used for estimating the PSD of a narrow polystyrene (PS) latex standard of nominal diameter 111 nm. The standard was also characterized by 2 independent techniques: capillary hydrodynamic fractionation, and transmission electron microscopy (TEM). The PSD predicted by the GRNN resulted close to that obtained by TEM. The estimated PSDs were better than those obtained through standard numerical techniques for 'ill-conditioned' inverse problems.