Scaled resistively-coupled VO2 oscillators for neuromorphic computing

Corti, Elisabetta; Gotsmann, Bernd; Moselund, Kirsten E.; Ionescu, Adrian M.; Robertson, John; Karg, Siegfried

doi:10.1016/j.sse.2019.107729

Cited by 50 publications

(38 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The resistivity vs. temperature curve (RT) of a 250 nm × 250 nm device is shown in Figure 3A and exhibits an insulator-to-metal phase-transition with roughly a two-order of magnitude in resistance change. The step-like RT implies multi-grain transitions, as already shown in previous work ( Ruzmetov et al, 2009 ; Takami et al, 2014 ; Corti et al, 2019 ). Figure 3B shows the insulator-to-metal and metal-to-insulator transition of an electrically activated device.…”

Section: Methodssupporting

confidence: 76%

“…In this work, we further report that the cross-bar geometry yields a better reliability of the devices. In fact, in a previous work we discussed how coplanar devices needed a burn-in cycle to initialize the devices, which sometimes resulted in fatal irreversible changes in morphology ( Corti et al, 2019 ). The crossbar devices do not need a burn-in cycle, improving reliability as virtually all the devices fabricated were able to produce oscillations.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Coupled VO2 Oscillators Circuit as Analog First Layer Filter in Convolutional Neural Networks

et al. 2021

Self Cite

View full text Add to dashboard Cite

In this work we present an in-memory computing platform based on coupled VO2 oscillators fabricated in a crossbar configuration on silicon. Compared to existing platforms, the crossbar configuration promises significant improvements in terms of area density and oscillation frequency. Further, the crossbar devices exhibit low variability and extended reliability, hence, enabling experiments on 4-coupled oscillator. We demonstrate the neuromorphic computing capabilities using the phase relation of the oscillators. As an application, we propose to replace digital filtering operation in a convolutional neural network with oscillating circuits. The concept is tested with a VGG13 architecture on the MNIST dataset, achieving performances of 95% in the recognition task.

show abstract

Section: Methodssupporting

confidence: 76%

Section: Methodsmentioning

confidence: 99%

Coupled VO2 Oscillators Circuit as Analog First Layer Filter in Convolutional Neural Networks

et al. 2021

Self Cite

View full text Add to dashboard Cite

show abstract

“…Device fabrication -The VO2 devices are realized on a 4'' Si wafer on top of 1 μm thermal SiO2 layer, as described in more detail in [36]. A 50 nm VO2 layer was deposited, etched into stripes with dry-etching and contacted with Ni/Au (Fig.…”

Section: Resultsmentioning

confidence: 99%

Time-Delay Encoded Image Recognition in a Network of Resistively Coupled VO₂ on Si Oscillators

Corti

Khanna

Niang

et al. 2020

IEEE Electron Device Lett.

View full text Add to dashboard Cite

Oscillatory neural networks based on insulator to metal transition of VO2 switches are implemented for image recognition. The VO2 oscillators are fabricated on silicon in a CMOS compatible process. A fully-connected network of coupled oscillators is investigated using programmable resistors as coupling elements. In this approach, input of the image information and data processing is performed in the time domain. In particular, tuning the coupling resistors allows to control the phaserelation between the oscillators. This is used to memorize and recognize patterns in an analog circuit. The concept is demonstrated experimentally on a three-VO2 oscillator network, network whereas simulations are performed on a larger 9-oscillators circuit.

show abstract

“…VO 2 is a strongly correlated electron material which undergoes a first-order IMT at 340 K from a low-temperature monoclinic insulating state to a high-temperature rutile metallic state, accompanied by structural, electrical, and optical changes 1,2 . The potential of IMT along with its sensitivity to external stimuli makes it promising for a variety of applications in resistive memories [3][4][5] , optical switches 6,7 , sensors [8][9][10] , tunable photonic devices 11,12 , brain-inspired and neuromorphic architectures [13][14][15] . We target the sensor application in the frequency range of mm-wave and THz.…”

Section: Introductionmentioning

confidence: 99%

Millimeter- and Terahertz-wave stochastic sensors based on reversible insulator-to-metal transition in vanadium dioxide

Qaderi¹,

Rosca²,

Burla³

et al. 2021

Preprint

View full text Add to dashboard Cite

Sensitivity to low-energy photons in phase change materials enables the development of efﬁcient millimeter-wave (mm-wave) and terahertz (THz) detectors. Here, we present the concept of uncooled mm-wave detection based on the sensitivity of IMT threshold voltage to the incident wave by exploiting the characteristics of reversible insulator-to-metal transition (IMT) in Vanadium dioxide (VO2) thin ﬁlm devices. The detection concept is demonstrated through actuation of biased VO2 2-terminal switches encapsulated in a pair of coupled antennas on a Si/SiO2 substrate. We also study the behavior of VO2 switches interrupting coplanar waveguide (CPW)s. Ultimately, we propose an electromagnetic wave-sensitive voltage-controlled spike generator based on the VO2 switches in an astable circuit. The fabricated sensors show record ﬁgs. of merit, such as responsivities of around 66.3 kHz/mW with a low noise equivalent power (NEP) of 20 nW at room temperature, for a footprint of 2.5×10−5 mm2, which can be easily scaled. This solution gives 3 times better responsivity with only 1/10 footprint of the state of the art. However, the footprint is capable of being scaled down to few hundreds of nanometers. The responsivity in static measurements is 76 kV/W in the same circumstances. Based on experimental statistical data measured on robust fabricated devices, we investigate and report stochastic behavior and noise limits of VO2-based spiking sensors that are expected to form a new class of energy efﬁcient transducers. The results highlight the capability of VO2 phase transition to serve for building electromagnetic power sensors, that can be triggered by low energy photons.

show abstract

Scaled resistively-coupled VO2 oscillators for neuromorphic computing

Cited by 50 publications

References 20 publications

Coupled VO2 Oscillators Circuit as Analog First Layer Filter in Convolutional Neural Networks

Coupled VO2 Oscillators Circuit as Analog First Layer Filter in Convolutional Neural Networks

Time-Delay Encoded Image Recognition in a Network of Resistively Coupled VO₂ on Si Oscillators

Millimeter- and Terahertz-wave stochastic sensors based on reversible insulator-to-metal transition in vanadium dioxide

Contact Info

Product

Resources

About