Millimeter-wave to near-terahertz sensors based on reversible insulator-to-metal transition in VO2

Abstract

In the quest for low power bio-inspired spiking sensors, functional oxides like vanadium dioxide are expected to enable future energy efficient sensing. Here, we report uncooled millimeter-wave spiking detectors based on the sensitivity of insulator-to-metal transition threshold voltage to the incident wave. The detection concept is demonstrated through actuation of biased VO2 switches encapsulated in a pair of coupled antennas by interrupting coplanar waveguides for broadband measurements, on silicon substrates. Ultimately, we propose an electromagnetic-wave-sensitive voltage-controlled spike generator based on VO2 switches in an astable spiking circuit. The fabricated sensors show responsivities of around 66.3 MHz.W−1 at 1 μW, with a low noise equivalent power of 5 nW.Hz−0.5 at room temperature, for a footprint of 2.5 × 10−5 mm2. The responsivity in static characterizations is 76 kV.W−1. Based on experimental statistical data measured on robust fabricated devices, we discuss stochastic behavior and noise limits of VO2 -based spiking sensors applicable for wave power sensing in mm-wave and sub-terahertz range.

Picosecond Femtojoule Resistive Switching in Nanoscale VO2 Memristors

“…We emphasize that the small device volume, granted by our special electrode arrangement, is a crucial ingredient to the improved heat transfer toward the environment which is a prerequisite to fast thermal relaxation. This finding, together with the sub-100 ps IMT enables the orders of magnitude acceleration of VO 2 -based electronics, among them THz sensors 40 and oscillator circuits exploited for neuromorphic computing purposes. 16…”

Tuning the Intrinsic Stochasticity of Resistive Switching in VO2 Microresistors

“…In other cases, e.g. in sensory VO 2 neurons where stimuli are encoded in the spike rate, − these variations should be minimized as they induce jitter on the oscillator frequency and can thus impede the sensor resolution, as shown in a previous work . However, so far, there exist no guidelines to design VO 2 microresistors with controlled stochasticity, meaning their functionality will be impeded by the microfabricated film’s intrinsic physical properties.…”

“…These could correspond to distinct percolation hot spots and resulting filament locations. To our knowledge, multimodal V IMT distributions have not been analyzed in VO 2 literature and have been solely observed once in VO 2 oscillators with no proposed explanatory hypothesis. Our optical microscopy imaging equipment resolution did not allow us to image the filament in the devices showing multimodal V IMT distributions (due to their small channel length).…”

“…Figure shows a typical VO 2 microresistor along with its hysteretic I – V characteristic and optical imaging of the metallic filament. This phenomenon is of particular interest in neuromorphic systems and has been harnessed to emulate analog spiking neurons, both computational − and sensory. − These neurons could respectively serve as functional units and entry layers for spiking neural networks, enabling sparse and energy-efficient encoding of information. , . On the other hand, memristive effects attributed to nonvolatile film alterations have been put in evidence in VO 2 microresistors, − opening the possibility to use them to emulate synapses.…”

“…So far, most works that studied stochasticity in VO 2 focused on cycle-to-cycle variations in the threshold voltage V IMT . ,,,, In this work, we also studied variations in the insulating state resistance. The importance of taking R ins variations into account has recently been highlighted in the context of VO 2 spiking sensors, showing that they contribute to spike rate stochasticity with the same proportion as V IMT variations .…”