On the origin of anomalous hysteresis in graphite/boron nitride transistors

Field-effect devices constructed by stacking flakes of van der Waals (vdW) materials, with hexagonal boron nitride (hBN) playing the role of gate dielectric, often exhibit virtually no hysteresis in their characteristics. This permits exquisitely detailed studies of diverse gate-voltage-tuned phenomena in vdW devices. Recently, however, a dramatic form of gate hysteresis, sometimes called the "gate doesn't work" (GDW) or "electron ratchet" effect, has been seen in certain individual vdW devices that seem otherwise unexceptional. When it occurs, this hysteresis phenomenon is striking and robust, yet it is difficult to reliably reproduce between devices and, largely as a result, its origin remains disputed. Most devices where it has been seen have a bilayer graphene channel and nominal rotational alignment between the graphene and hBN, which has engendered explanations based on properties of bilayer graphene combined with moiré effects. Here, we report our studies of the phenomenon observed in devices that have multilayer graphene channels. We find that the effect can occur in devices with graphite channels that have many more than two graphene layers, in which case it is unambiguously associated with just one surface of the graphite. It can also survive to room temperature, occur in the absence of intentional rotational alignment with hBN, persist when a monolayer of WSe2 is inserted between the graphene and hBN, and exhibit continuous relaxation on timescales of hours or longer. These observations impose strong constraints on the origin of this puzzling phenomenon, which has exciting potential applications if it can be mastered.