Abstract
There is a high charge density in the well of a double barrier structure when it is at resonance. This alters the potential profile through the diode and may cause it to have two stable states for a given applied bias. The authors have studied this effect both analytically and numerically. The analytical work extends the earlier calculations of Sheard and Toombs (1988). Their main result is a condition that must be satisfied for bistability to occur: the potential due to the trapped charge must shift the centre of the resonance by more than its width. This requires the resonance to be strong, which can be achieved by designing the barriers to be asymmetric at zero bias, but symmetric when the bias is near that required for resonance. The build-up of charge and the range of bistability can be increased by making the exit barrier more opaque, but the current through the diode decreases. These results are supported by numerical calculations, in which the Poisson and Schrodinger equations are solved self-consistently. The analytical and numerical calculations agree quantitatively, but they are only able to explain the experiments qualitatively. This is because their numerical model has a severe flaw, which plagues all simple quantum-mechanical simulations: it omits scattering which is vital in treating the contacts, particularly the accumulation layer from which charge is injected into the barrier.