Y.G. Gladush, S.V. Korneev, V.B. Lopatko, A.G. Novoselov, E.B. Perminov

The algorithm which improve the size determination accuracy for multimodal narrow distributions of spherical particles by the dynamic light scattering method was suggested. Our algorithm, based on a non-linear regression and static light scattering methods, use the following a priori information: the form of the distribution and the dielectric constants of the particles. Levenberg-Marquardt (LM) nonlinear regression is used to fit an experimental autocorrelation function and allows to find a local minimum in the space of free parameters, which depend on the characteristics of the particles, a incident radiation and the number of scatters. When the number of free parameters is large a process of the fitting becomes incorrect. Using the static light scattering data, we suggest the modified method of nonlinear regression to improve the problem. By the methods of the static light scattering we calculate the ratio between intensities at the different angles for each local minimum. Using theoretical and observed mean intensities we set the weight that determine whether this is the best approximation to the experimental data in comparison with other equivalent local minimums. For two size particles distribution we show that our method improves the problem of size determination.