Document Type


Journal/Book Title/Conference

Physical Review B




American Physical Society

Publication Date



A numerical modeling approach was developed to predict the dielectric properties of heterogeneous particulate materials with arbitrary microstructures. To test the method, simulation and experimental data were acquired for the effective permittivities of various glass sphere suspensions. Both ordered lattices and random microstructures of up to 3600 spheres were modeled for volume fractions of 0.025–0.60. The electric fields in the suspensions were computed using an iterative multipole method that included multiple-scattering effects. The effective permittivities were obtained by averaging the electric field and electric displacement over a representative volume. Frequency spectra, electric field images, and single-scattering results (i.e., no particle-particle interactions) were additionally generated. The results were compared to experimental data for random close-packed microstructures, to effective-medium approximations, to exact lattice models, and to a perturbation expansion model. The comparisons showed that the iterative models agreed with the exact lattice models to within 3.31% for crystalline suspensions. Results for random suspensions agreed with the perturbation expansion model to within 1.76% for volume fractions up to 0.50. Single-scattering models additionally predicted permittivities for the microstructures as well as or better than the Maxwell Garnett approximation [Philos. Trans. R. Soc. London, Ser. A 203, 385 (1904)], suggesting that microstructural effects and multipole moments higher than the dipole are required for more accurate statistical prediction of effective permittivities. The effective permittivities, convergence behavior, and dispersion behavior of the simulations were sensitive to both microstructure and the extent of multiple scattering included in the models, illustrating how the macroscopic properties depend significantly on the microscopic details of the interactions. In contrast to other approaches, the iterative multipole method can model both the frequency and spatial dependencies of the electromagnetic properties of particulate materials, as well as a wide variety of microstructures, including polydisperse and hierarchical systems.