Inverse Laplace Transform for Dynamic Light Scattering: Impact of Regularization

Abstract

Dynamic Light Scattering (DLS) analyzes particle dynamics from the autocorrelation functions of scattered light intensity, yet extracting accurate relaxation time distributions from noisy data is challenging. We develop an inverse problem approach to recover this distribution by inverting the Laplace transform with physics-based regularization, called the CONTIN method. We improve it to use it on noisy data, across a wide range of time scales, with a selection of the regularization strength through a data-driven L-curve criterion. Our approach enhances robustness under high noise and reveals multi-scale dynamics in complex systems. Validation is performed on simulated data, compared to the Cramér-Rao bound and to parametric methods, and on experimental data from Carbopol microgels. It demonstrates superior accuracy over parametric methods, especially for broad time distributions. The algorithm's logarithmic discretization and variance-reduced correlation estimation enhance performance, offering a powerful tool for non-parametric DLS analysis and deeper insights into soft matter dynamics.