Non-equivalence of different evaluation algorithms to derive mean particle size from dynamic light scattering data

Different manufacturers of dynamic light scattering (DLS) instruments implement various evaluation algorithms. Users often implicitly assume that the various algorithms yield the same mean value and deliver results of the same precision. This assumption was tested using results obtained by cumulants analysis, CONTIN and Non-Negative Least Squares fitting (NNLS) of distributions, and frequency analysis on near-spherical silica and polystyrene latex particles obtained in one laboratory and results on near-spherical silica obtained in other laboratories. Particles have been measured in the size range of 10 to 200 nm. The results show that the average particle diameters from different algorithms differ significantly and that these differences increase with an increasing polydispersity of the material. Also, precision, both within-laboratory (repeatability) as well as between-laboratory standard deviations (reproducibility), differs between the algorithms investigated, especially for the more polydisperse silica materials. The results obtained from the cumulants method usually show the best repeatability and the lowest between-laboratory standard deviation. The results also show that the conversion of intensity-weighted results to volume-weighted results increase the between-laboratory standard deviation, confirming the theoretical expectations that conversion increases the variation significantly.