Estimation of bivariate probability distributions of nanoparticle characteristics, based on univariate measurements

Abstract

ABSTRACT The properties of complex particle systems typically depend on multivariate distributions of particle properties, like size and shape characteristics. Multidimensional particle property distributions can be a powerful tool to describe these systems. However, only few techniques exist which are able to simultaneously measure more than one property of individual particles in fast and efficient ways. It is shown how two-dimensional property spaces can be constructed by the combination of two univariate measurements to obtain bivariate particle size distributions. The proposed method is a general approach, which can be applied to a wide spectrum of particle systems and measurement devices. In this paper, the results of a case study are presented, which allow the estimation of bivariate distributions of length and diameter of nanorods, solely using univariate distributions of their particle mass and extinction-weighted sedimentation coefficient distributions. These quantities contain joint information about the particle lengths and diameters, which is used for the reconstruction. The method is validated in a simulation study in which the bivariate distribution to be reconstructed and the reconstruction parameters are varied. In addition, regularization techniques are introduced to reduce methodical errors. This approach can be transferred to other particle systems and measurement techniques, for which functional relationships between particle properties and measured quantities are well described.