Particle & Particle Systems Characterization, Vol.23, No.2, 175-183, 2006
Ultrasonic particle sizing of disperse systems with partly unknown properties
The determination of particle size distributions from ultrasonic attenuation spectra requires the knowledge of a variety of parameters (material properties and particle concentration) when the analysis is based on analytical models. In this paper, two strategies are tested that allow for the elicitation of size information from attenuation spectra when the relevant model parameters are only partially known. The first strategy uses the data on measured sound speed as additional information, which facilitates the fitting of the unknown model parameters parallel to the calculation of the particle size distribution. The second strategy comprises the formation of an empirical model that is based on the principal components, which are obtained by decomposing a measured attenuation spectrum into a set of orthogonal basic spectra. Both strategies are tested with colloidal emulsions: the first one for the cases of unknown particle concentration and unknown specific heat of the droplets, the second one for monitoring changes in the state of dispersion and in the temperature.
Keywords:empirical model;material properties;multiple regression;parameter fit;particle size analysis;sound speed;spectrum decomposition;ultrasonic attenuation spectroscopy