Classification of single-particle two-dimensional angular optical scattering patterns and heuristic scatterer reconstruction

Two-dimensional angular optical scattering (TAOS) is an experimental technique by which the azimuth- and polar-angle-dependent intensity patterns of monochromatic light scattered by isolated material particles are collected within a limited aperture. TAOS patterns are analyzed with the ultimate goal of estimating the class of shapes and the set of complex refractive indices to which the scatterer may belong. The implemented heuristic reconstruction procedure integrates experimental techniques and methods of data analysis. It consists of three stages: (1) TAOS patterns are classified by cluster analysis and classification is validated by principal components analysis, (2) prior information about the shape and refractive index of the particles is acquired from analytical electron microscopy and drives the computation of TAOS patterns by a T-matrix code, and (3) cluster analysis is repeated by including one computed pattern at a time and ranking it according to distance to a cluster. As a result, some simulated scatterers are found, the computed patterns of which merge with clusters of experimental patterns. Whereas the reconstruction of each individual scatterer from the available TAOS pattern would not have been feasible, possible scatterer classes are identified. The investigated material is tire dust from laboratory abrasion tests of aerodynamic size < 10 μm.