Hexagonal boron nitride (h-BN) memristor arrays for analog-based machine learning hardware

Abstract

Recent studies of resistive switching devices with hexagonal boron nitride (h-BN) as the switching layer have shown the potential of two-dimensional (2D) materials for memory and neuromorphic computing applications. The use of 2D materials allows scaling the resistive switching layer thickness to sub-nanometer dimensions enabling devices to operate with low switching voltages and high programming speeds, offering large improvements in efficiency and performance as well as ultra-dense integration. These characteristics are of interest for the implementation of neuromorphic computing and machine learning hardware based on memristor crossbars. However, existing demonstrations of h-BN memristors focus on single isolated device switching properties and lack attention to fundamental machine learning functions. This paper demonstrates the hardware implementation of dot product operations, a basic analog function ubiquitous in machine learning, using h-BN memristor arrays. Moreover, we demonstrate the hardware implementation of a linear regression algorithm on h-BN memristor arrays.

Quantum Conductance in Vertical Hexagonal Boron Nitride Memristors with Graphene-Edge Contacts

“…As shown, HRS resistance increased with reducing the active area ( W ), but LRS resistance is mostly unaffected, indicative of filamentary NVRS Figure f compares HRS and LRS conductance ( G on and G off for LRS and HRS, respectively) from vertical h-BN memristors reported in this work against previous reports from planar h-BN memristors. ,,,,,,, Clearly, our vertical h-BN memristors with graphene edge contacts achieve the smallest G off because of the ultrasmall active area. It also achieves one of the smallest G on (at or slightly below G 0 ), as conductive paths are better isolated to a single or few atomic-scale CNFs.…”

“…Among the various physical mechanisms responsible for nonvolatile resistive switching (NVRS), formation and dissolution of filamentary conductive pathways is one of the most commonly employed . Recently, 2DMs have attracted significant interest for RRAM due to promising NVRS characteristics, − even with atomically thin active layers (e.g., 0.3 nm h-BN monolayers). − Specifically, 2DM-based RRAM is sought not just for their ultimate scalability but also for their experimentally demonstrated ultralow write currents (fA) and programming energies (zJ), , high thermal reliability and long-term retention, ultrafast switching speeds (ps), − reduced temporal and spatial variation, ,, etc.…”

“…For example, a 100 nm graphene edge contact (thickness of 0.3 nm) can achieve an ultrasmall active area of 30 nm 2 , which is orders of magnitude smaller than that of previous demonstrations. Whereas numerous studies have focused on NVRS behavior of planar h-BN memristors, ,,,,, no reports have discussed NVRS and scaling properties of vertical h-BN memristors. In this work, combining the benefits of CVD-grown 2DM and graphene edge contacts, we report vertical h-BN memristors with ultrasmall active areas, low operating currents (high resistance), and large R HRS / R LRS ratio.…”

Research progress in architecture and application of RRAM with computing-in-memory

“…The chip can realize dot-product operation, which is the first application of functional 3D CMOL hybrid circuits. Xie et al 45 used h-BN RRAM across-arrays to demonstrate the hardware implementation of dot product operation and the linear regression algorithm, as shown in Fig. 7(b)–(d).…”

Synapse-Mimetic Hardware-Implemented Resistive Random-Access Memory for Artificial Neural Network

“…Under an applied electric field, ions migrate inside the host materials to form a filament, and when this filament grows and creates an ohmic contact between the TE and BE, the electrical resistance rapidly decreases (set process) [ 74 , 75 ]. Two types of mobile ionic species are primarily used in memristors: metal cations, such as those of Ag, Cu, and Ti, and anions, such as oxygen (or its vacancy) [ 76 , 77 , 78 , 79 ]. In the case of a metal cation film memory, metal ions are typically introduced into the host material from the electrode through an electroforming process, or the active layer itself is mixed in a stoichiometric ratio [ 80 , 81 ].…”