Abstract
Recent experiments on the two-dimensional electron gas in various semiconductor devices have shown an unexpected metal-insulator transition. These experimental results cannot be understood from the conventional perspective of weak interactions. We propose that: the low-density insulating state is the Wigner glass, a phase with quasi-long-range structural order and competing ferromagnetic and antiferromagnetic spin-exchange interactions; the transition is the melting of this Wigner glass, the disorder allowing the transition to be second order; within the Wigner glass phase there are at least two distinct magnetic ground states—a ferromagnetic state at very low electron density and a spin-liquid state with a spin pseudogap at higher densities; and the metallic side of the transition is a non-Fermi liquid. We present a proposed phase diagram as a function of disorder strength and density and suggest experimental signatures of the various phases and transitions.