Characterization of Nanoparticle Aggregates Using Bayesian Inference via Light Scattering Experiments

Controlling the properties of the nanostructures requires use of proper characterization methods. There are many techniques available such as microscopy, time-resolved laser-induced-incandescence or light scattering methods that can be used for characterization of nanostructures. While direct observation such as microscopy is possible for some applications, the indirect characterization is carried out through measurements of structures’ emission or scattering behavior as in the laser-induced-incandescence and light scattering methods. Characterization of soot aggregates is well studied and is often classified as an inverse problem that can be formulated as a parameter estimation problem, where parameters defining the aggregate such as average size and number of nanoparticles are estimated. Estimations based on the scattering behavior can be questionable unless these parameters are directly observed by other methods. Instead of presenting a single estimate, Bayesian credible intervals would be a better option when indirect methods such as light scattering are used. The objective of this work is to investigate the strengths and weaknesses of the Bayesian approach based on numerical light scattering experiments considering soot aggregates as an example. Light scattering behavior of the soot aggregates is obtained with discrete dipole approximation, considering unpolarized light.