MULTIPLE SCATTERING AND SIZE DISTRIBUTION EFFECTS ON THE PERFORMANCE OF A LASER DIFFRACTION PARTICLE SIZER

Abstract

The effect of multiple scattering on the performance of the Malvern particle sizer was studied experimentally. Two particle samples with the same volume median diameter but with different size distributions were used in the experiments. One sample could be described by a log-normal distribution function, the other by a Rosin-Rammler distribution function. The particle samples were specifically designed to simulate drop size distributions in liquid sprays. The model-independent, log-normal, and Rosin-Rammler models were used in the data analysis, and the results were compared. It was found that even when the sample concentration was low and multiple scattering did not occur, the results obtained with the log-normal and Rosin-Rammler models could be misleading, particularly in the high end of the distribution (large particles). When the log-normal test sample was used, the data analysis with the model-independent and log-normal models gave good results, but the data analysis with the Rosin-Rammler model seriously underestimated the fraction of large particles. When the Rosin-Rammler test sample was used, the data analysis with the model-independent and Rosin-Rammler models gave good results, but the data analysis with the log-normal model seriously overestimated the fraction of large particles. When the sample concentration is high and multiple scattering occurs, data analysis with the model-independent model is not feasible. Correction equations are available in literature for data analysis with the log-normal and Rosin-Rammler models. It was found that when the correction equations were used, both the log-normal and Rosin-Rammler models provided the mean particle size reasonably well. However, there were considerable difficulties in the high end of the distribution. This was not surprising, since the experiments with a low particle concentration had already shown that the data analysis with the log-normal and Rosin-Rammler models can produce misleading results.