Correlation and clustering in the optical properties of composites: A numerical study.

Abstract

We have numerically investigated the effects of correlation and clustering on the far-infrared (FIR) absorption and surface-plasmon modes of a model two-dimensional metal-insulator composite. We model the composite as a diluted resistor-inductor-capacitor (RLC) network. Two-site and nearest-neighbor-site correlations among the conducting bonds are found to enhance the FIR absorption per metallic bond by a factor of 5 relative to a network of the same concentration with a purely random distribution of metallic bonds. Ring-shaped clusters (which model insulating particles with a metallic coating), and percolation clusters are found to produce a FIR absorption enhanced by more than 2 orders of magnitude per metallic bond. The surface-plasmon absorption peak in a percolation cluster is found to be strongly broadened relative to the predictions of the Maxwell Garnett approximation. In the two-site correlation model, the surface-plasmon absorption peak is weakly split, whereas in the nearest-neighbor site-correlation model, it is weakly red shifted. Ring-shaped clusters are found to produce double and triple peaks in the surface-plasmon frequency range. Possible explanations for these novel features are briefly discussed.