Previous work has demonstrated the simultaneous concentration and separation of proteins via a stable ion concentration gradient established within a nanochannel (Inglis et al. Angew. Chem., Int. Ed. 2001, 50, 7546-7550). To gain a better understanding of how this novel technique works, we here examine experimentally and numerically how the underlying electric potential controlled ion concentration gradients can be formed and controlled. Four nanochannel geometries are considered. Measured fluorescence profiles, a direct indicator of ion concentrations within the Tris-fluorescein buffer solution, closely match depth-averaged fluorescence profiles calculated from the simulations. The simulations include multiple reacting species within the fluid bulk and surface wall charge regulation whereby the deprotonation of silica-bound silanol groups is governed by the local pH. The three-dimensional system is simulated in two dimensions by averaging the governing equations across the (varying) nanochannel width, allowing accurate numerical results to be generated for the computationally challenging high aspect ratio nanochannel geometries. An electrokinetic circuit analysis is incorporated to directly relate the potential drop across the (simulated) nanochannel to that applied across the experimental chip device (which includes serially connected microchannels). The merit of the thick double layer, potential-controlled concentration gradient as a particle focusing and separation tool is discussed, linking this work to the previously presented protein trapping experiments. We explain why stable traps are formed when the flow is in the opposite direction to the concentration gradient, allowing particle separation near the low concentration end of the nanochannel. We predict that tapered, rather than straight nanochannels are better at separating particles of different electrophoretic mobilities.