Opto-electronic memristors: Prospects and challenges in neuromorphic computing

Emboras, Alexandros; Alabastri, Alessandro; Lehmann, Paul; Portner, Kevin; Weilenmann, Christoph; Ma, Ping; Cheng, Bojun; Lewerenz, Mila; Passerini, Elias; Koch, Ueli; Aeschlimann, Jan; Ducry, Fabian; Leuthold, Juerg; Luisier, Mathieu

doi:10.1063/5.0028539

Cited by 45 publications

(29 citation statements)

References 62 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, we presented an approach to scale up the proposed technology to crossbar arrays including several electro-optical synapses stimulated by a global light beam. This could allow high-density neuromorphic computing chips where the presence of a third modulation channel, the optical signal, is leveraged to implement neo-Hebbian, three-factor learning rules in spiking neural networks. , …”

Section: Discussionmentioning

confidence: 77%

Analog Nanoscale Electro-Optical Synapses for Neuromorphic Computing Applications

et al. 2021

Self Cite

View full text Add to dashboard Cite

The typically nonlinear and asymmetric response of synaptic memristors to positive and negative electrical pulses makes the realization of accurate deep neural networks very challenging. Here, we integrate a two-terminal valence change memory (VCM) into a photonic/plasmonic circuit and show that the switching properties of this memristor become more gradual and symmetric under light irradiation. The added optical input acts on the VCM as a third, independent modulation channel. It locally heats the active area of the device, which enhances the generation of oxygen vacancies and broadens the resulting nanoscale conductive filaments. The measured conductance modulation of the VCM is then inserted into a neural network simulator. Using the MNIST data set of handwritten digits as an application, a light-enhanced recognition accuracy of 93.53% is demonstrated, similar to ideally performing memristors (94.86%) and much higher than those without light (67.37%). Notably, the optical signal does not increase the overall energy consumption by more than 3.2%. Finally, an approach to scale up our electro-optical technology is proposed, which could allow high-density, energy-efficient neuromorphic computing chips.

show abstract

Section: Discussionmentioning

confidence: 77%

Analog Nanoscale Electro-Optical Synapses for Neuromorphic Computing Applications

et al. 2021

Self Cite

View full text Add to dashboard Cite

show abstract

“…While conventional memristors are activated by purely electrical stimuli, photo-stimulation of memristive responses provides new opportunities for a variety of applications such as light-gated and electro-photo-sensitive memristors for optoneuromorphic and arithmetic computing, (Maier et al, 2016;Tan et al, 2017;John et al, 2018b;Emboras et al, 2020;John et al, 2020;Ma et al, 2021) integrated photonic neural networks, (Stark et al, 2020;Shastri et al, 2021), and preprocessing of image data in artificial vision before the transfer to the computing unit (Yang et al, 2020;Gong et al, 2021;Kim et al, 2021). Direct processing of visual information with such optoelectronic memristors may enable simpler architectures, mitigating the need for additional electro-optical converters for signal transduction and communication.…”

Section: Introductionmentioning

confidence: 99%

Inductive and Capacitive Hysteresis of Halide Perovskite Solar Cells and Memristors Under Illumination

et al. 2022

View full text Add to dashboard Cite

The current–voltage curves of memristors exhibit significant hysteresis effects of use for information storage and computing. Here, we provide a comparison of different devices based on MAPbI3 perovskite with different contact configurations, from a 15% efficient solar cell to a pure memristor that lacks directional photocurrent. Current–voltage curves and impedance spectroscopy give insights into the different types of hysteresis, photocapacitance, and inductance present in halide perovskites. It is shown that both halide perovskite memristors and solar cells show a large inverted hysteresis effect at the forward bias that is related to the presence of a chemical inductor component in the equivalent circuit. Based on the results, we classify the observed response according to recombination current in devices with selective contacts, to voltage-activated single-carrier device conduction in devices with symmetric contacts. These findings serve to gain an understanding of the mechanism of memristor currents in mixed ionic-electronic conductors such as halide perovskites. We establish the link in the electrical response between solar cells and memristors.

show abstract

“…Apart from the application in traditional multi-level memristive devices, the demonstrated programmable tunnel junctions would be integrated with plasmonic or silicon waveguides to form artificial optoelectronic or photonic synapses for nonspiking artificial neural networks. [124,125]…”

Section: Hot Electron Tunnel Junctionsmentioning

confidence: 99%

Quantum Artificial Synapses

et al. 2021

View full text Add to dashboard Cite

Neuromorphic in-memory computing systems, comprising artificial synapses and neurons, can overcome the energy inefficiency and throughput limitation of today's von Neumann computing architecture. Recently, powered by the unique properties of quantum materials, for example, high mobility, outstanding sensitivity, and strong quantum effect, researchers have built quantum artificial synapses to mimic the biological ones. These quantum electronic/photonic synapses can precisely define their conductance state (or synaptic weight) for emulating synaptic behaviors, which shows bionic performance unreachable by other conventional materials. In this review, the significant achievements in quantum artificial synapses are summarized. First, potential quantum materials used in artificial synapses are discussed with particular attention to quantum dots, nanowires, layered materials, and quasi-2DEG interfaces. Then, the major quantum effects that are utilized in quantum artificial synapses, for example, Josephson effect, quantum tunneling, and spin memory, are reviewed. In addition to the discussion on a single synaptic device, the macroscale integration into artificial visual systems and artificial nerve networks are also highlighted. Finally, the associated future research trends and target applications are also discussed.

show abstract

Opto-electronic memristors: Prospects and challenges in neuromorphic computing

Cited by 45 publications

References 62 publications

Analog Nanoscale Electro-Optical Synapses for Neuromorphic Computing Applications

Analog Nanoscale Electro-Optical Synapses for Neuromorphic Computing Applications

Inductive and Capacitive Hysteresis of Halide Perovskite Solar Cells and Memristors Under Illumination

Quantum Artificial Synapses

Contact Info

Product

Resources

About