Tuning resistive switching properties of WO_3−x-memristors by oxygen vacancy engineering for neuromorphic and memory storage applications

Abstract

High density memory storage capacity, in-memory computation and meuromorphic computing utilizing memristors are expected to solve the limitation of von-Neumann computing architecture. Controlling oxygen vacancy (VO) defects in metal oxide thin film based memristors holds the potential of designing resistive switching (RS) properties for memroy storage and neuromorphic applications. Herein, we report on RS characteristics of CMOS compatible WO3-x based memristors modulated by precisely controlled oxygen non-stoichiometry. Switchability of the resistance states has been found to depend strongly on the VOs concentration in the WO3-x layer. Depending on x, the memristors exhibited forming-free bipolar, forming-required bipolar, and non-formable characteristics. Devices with moderate VOs concentration (~5.8×1020 cm-3) exhibited a large Roff/Ron ratio of ~6500, and reset voltage-controlled multi-level resistance states. A forming-free, stable multi-level RS has been realized for a memristor possessing VOs concentration of ~6.2×1020 cm-3. WO3-x-based memristors with higher VOs concentrations (~8.9×1020 cm-3-1×1021 cm-3) exhibited lower initial resistance, low Roff/Ron ratios (~15-63) and forming-free synaptic functions with reasonable conduction modulation linearity. Investigation of the conduction mechanism suggests that tailoring VOs concentration modifies the formation and dimension of the conducting filaments and the Schottky barrier height at the WO3-x/Pt interface, which paves the way for designing WO3-x -based memristors for memory storage and neuromorphic applications.

Internal Resistor Effect of Multilayer-Structured Synaptic Device for Low-Power Operation

“…However, resistive switching was only obtained when the ratio of Ar to O 2 was 20:5. This result can be explained in terms of the oxygen vacancy density in the WO X layer [39] (Figure 3b).…”

“…However, resistive switching was only obtained when the ratio of Ar to O was 20:5. This result can be explained in terms of the oxygen vacancy density in the WO layer [ 39 ] ( Figure 3 b). When the Ar:O ratio changed to 20:1, more oxygen vacancies were present in the WO layer, resulting in an electrically short state.…”

Brain-inspired computing: can 2D materials bridge the gap between biological and artificial neural networks?

“…Defects: as most of the RS mechanisms depend upon the defects present in the material, engineering and controlling the defect chemistry of a material become very essential for repeatable and reliable memristive devices. 18,105 Moreover, Janus structures of TMDCs, referring to transition metal layers having different surfaces, have garnered intensive research interest in recent times in view of their exceptional features that originate because of the symmetry breaking in such structures. 106 Hence, these could play a pivotal role in tailoring the memristive characteristics of neuromorphic devices.…”

“…62 The electric field causes anions or cations to migrate when a voltage is applied across the device, progressively forming filaments between the electrodes that alter the resistance. Oxygen vacancies are thought of as the migratory species in anion-based memristors, and oxides are typically utilized in their fabrication, 62 such as TiO 2 , 63 WO 3 , 18 etc. On the other hand, the metal cations from the electrodes form the filaments in cation-based memristors.…”

“…Devices that exhibit resistive switching (RS) or memristive behavior have garnered the most attention due to their ability to be utilized as potential synaptic hardware systems while consuming very low energy, fulfilling the criteria of brain-inspired or neuromorphic computing. [14][15][16][17][18] A memristor is a device proposed by Prof. Chua in 1971, considered as the 'fourth fundamental non-linear circuit element' relating charge with magnetic flux, and is characterized by a pinched hysteresis in its I-V curve. 19 The term ''neuromorphic'' was coined in the late 1980s by Carver Mead, 20 who pioneered the development of neuromorphic silicon neuron circuits.…”