Simulation of electrophoretic separations: effect of numerical and molecular diffusion on pH calculations in poorly buffered systems.

Abstract

A poorly buffered cationic isotachophoresis separation, first simulated by Reijenga and Kasica, has been revisited to demonstrate that an inconsistent description of solute and charge transport can lead to significant errors in the pH calculation. The separation is first simulated using a second-order finite difference scheme to show that omission of molecular diffusion from the charge balance results in a pH profile with spurious dips in the steady-state zone boundaries. The separation is also simulated using two first-order methods that employ numerical diffusion to stabilize solutions against spatiotemporal oscillations. Similar pH dips are generated by these first-order schemes, even when molecular diffusion is included in the charge balance, if numerical diffusion is not considered amongst the charge transport mechanisms. When numerical diffusion, inherent in the discretization of the component balances, is introduced to the charge balance, the spurious pH dips are eliminated. The results indicate that (i) pH dips originally reported by Reijenga and Kasicka are merely artifacts of their numerical model, and (ii) nonoscillatory numerical techniques, such as upwinding and flux limiters, should incorporate artificial transport mechanisms in the charge as well as the solute balances.