Abstract

Typical photon correlation experiments estimate temporal correlation functions of scattered light intensity as time averages. This procedure yields the desired ensemble averages as long as the detected light constitutes an ergodic stochastic process. Fortunately, this condition is met for most scattering experiments, particularly for measurements on fluid samples. But there are two notable exceptions, glasses and gels. Such systems have been recognized as nonergodic samples. Their most characteristic feature are large fluctuations in measured intercepts produced by minor changes of detector position or sample orientation. Three different schemes were proposed to construct valid and reproducible dynamic structure factor data for nonergodic samples [1-3]. This paper discusses their relative merits in terms of statistical accuracy and applicability.

© 1992 Optical Society of America