Particle sizing by dynamic light scattering: non-linear cumulant analysis

We revisit the method of cumulants for analysing dynamic light scattering data in particle sizing applications. Here the data, in the form of the time correlation function of scattered light, is written as a series involving the first few cumulants (or moments) of the distribution of particle diffusion constants. Frisken (2001 Appl. Opt. 40 4087) has pointed out that, despite greater computational complexity, a non-linear, iterative, analysis of the data has advantages over the linear least-squares analysis used originally. In order to explore further the potential and limitations of cumulant methods we analyse, by both linear and non-linear methods, computer-generated data with realistic 'noise', where the parameters of the distribution can be set explicitly. We find that, with modern computers, non-linear analysis is straightforward and robust. The mean and variance of the distribution of diffusion constants can be obtained quite accurately for distributions of width (standard deviation/mean) up to about 0.6, but there appears to be little prospect of obtaining meaningful higher moments.