Effects of experimental errors in dynamic light scattering data on the results from regularized inversions

Abstract

Three of the selection methods used with regularized data inversion were tested with dynamic light scattering data from a binary mixture of polystyrene latex particles. The three methods were the F-test (the standard method of CONTIN), the stability plot (one of the methods used with ORT) and the L-curve method. Two experimental situations differing in the scattered intensity were studied. The first was that of low mean count rates, where the random photon detection noise was dominant; the second was that of high mean count rates, where the non-random intensity fluctuation noise prevailed. The F-test proved to be very sensitive for non-random statistical and systematic errors, and then yielded too weakly regularized solutions associated with complex and poorly reproducible size distributions. The stability plot and L-curve methods, on the other hand, proved to be largely insensitive to these kinds of errors in the data and yielded reliable results, as concluded from a comparison with the size characteristics of the particles determined with electron microscopy.