Model for multi-filamentary conduction in graphene/hexagonal-boron-nitride/graphene based resistive switching devices

Abstract

Despite the enormous interest raised by graphene and related materials, recent global concern about their real usefulness in industry has raised, as there is a preoccupying lack of 2D materials based electronic devices in the market. Moreover, analytical tools capable of describing and predicting the behavior of the devices (which are necessary before facing mass production) are very scarce. In this work we synthesize a resistive random access memory (RRAM) using graphene/hexagonal-boronnitride/graphene (G/h-BN/G) van der Waals structures, and we develop a compact model that accurately describes its functioning. The devices were fabricated using scalable methods (i.e. CVD for material growth and shadow mask for electrode patterning), and they show reproducible resistive switching (RS). The measured characteristics during the forming, set and reset processes were fitted using the model developed. The model is based on the nonlinear Landauer approach for mesoscopic conductors, in this case atomic-sized filaments formed within the 2D materials system. Besides providing excellent overall fitting results (which have been corroborated in log-log, log-linear and linear-linear plots), the model is able to explain the dispersion of the data obtained from cycle-tocycle in terms of the particular features of the filamentary paths, mainly their confinement potential barrier height.

Recommended Methods to Study Resistive Switching Devices

“…While wrinkle‐free and polymer‐free 2D materials are preferred, it should be noted that such locations are more resistive, meaning that the CFs will not form there when the electrical field is applied (see Figure d). Therefore, the presence of wrinkles and polymer residue in RS devices based on 2D materials can be understood as a reduction of the effective area of the device . It should be also noted that one method to avoid the formation of wrinkles in 2D materials is to enhance the roughness of the substrate where it is transferred .…”

“…The development of circuits and systems based on RS devices requires compact RS device model running in SPICE‐like simulation environment . These compact models rely on conceptual simplifications (e.g., the idea of a conductive filament with a given shape, e.g., cylindrical or conical) and physical assumptions inferred from empirical measurements . A variety of semi‐empirical models have been proposed, which assume that the CF/s in RS devices behaves as: i) an ohmic conductor, ii) an hourglass‐shaped quantum‐point contact, iii) a space‐charge region, iv) a semi‐conductive region that may create a Schottky junction with the electrodes, v) a highly defective region in the dielectric in which either hopping conduction or delocalized transport may occur.…”

Engineering Field Effect Transistors with 2D Semiconducting Channels: Status and Prospects

“…This is because these 2D insulators ideally also have no dangling bonds and can adhere to the 2D semiconductors by van der Waals attraction, resulting in a minimized amount of interface states. Moreover, h‐BN dielectric stacks have shown a very high reliability when exposed to electrical fields, and have been already implemented in FETs, resistive switching based nonvolatile memories, and electronic synapses . Other wide bandgap 2D layered materials that may adhere to the 2D semiconducting channel by van der Waals attraction may also work well as gate dielectric in 2D‐FETs.…”

Yield, variability, reliability, and stability of two-dimensional materials based solid-state electronic devices

“…Lattice distortions and dopants normally act as trapping sites, facilitating out-of-plane charge transfer and generation of additional defects, which also tend to promote RS and reduce the forming-and switching voltages 18 . On the contrary, wrinkles and polymer residues from the transfer are insulating and increase the out-of-plane resistance, meaning that RS will never take place at those sites 19 ; this merely represents a reduction of the effective area of the memristors and it has no remarkable effect in their variability if the samples are relatively clean (i.e. <50 nm 2 of contaminants per µm 2 ).…”