Separation of polydisperse particle mixtures by deterministic lateral displacement. the impact of particle diffusivity on separation efficiency

Deterministic lateral displacement (DLD) has been recently proposed as a simple and efficient method to separate a polydisperse mixture of particles based on particle size. The separation device consists of a shallow rectangular channel filled with a periodic lattice of micrometer-sized obstacles, whose principal direction forms an angle with the channel walls. Particles are dragged by a carrier flow stream through the device. Experiments have shown that particles larger than a critical size depart from the average direction of the carrier flow, as they are systematically deflected by the obstacles while being dragged downstream. Theoretical models based on the geometric structure of the Stokes flow through the obstacle lattice have been proposed to predict the average direction of particle current flux. Besides, little is known about the dispersion of diffusing particles about the average particle current. In this article, we show that the interaction between the deterministic and stochastic components of particle motion results in a large-scale, possibly anisotropic, convection-enhanced dispersion process, which may hinder separation far beyond what could be predicted from the value of the bare particle diffusivity. The prediction of dispersion regimes results therefore essential for an optimal design of DLD devices. Copyright © 2012 Curtin University of Technology and John Wiley & Sons, Ltd.