Experimental demonstration of highly reliable dynamic memristor for artificial neuron and neuromorphic computing

Park, See-On; Jeong, Hakcheon; Park, Jongyong; Bae, Jongmin; Choi, Sung Hi

doi:10.1038/s41467-022-30539-6

Cited by 135 publications

(103 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The conductance of a filamentary RRAM depends upon the formation and rupture of the localized conductive filament (CF) consisting of a chain of oxygen vacancies (V O •• ) or metallic atoms, as shown in the left panel of Figure a. − While these devices exhibit a large memory window, fast switching speed, and good retention, they suffer from unavoidable variability as a result of the probabilistic nature of filament dynamics, − which would inevitably degrade the system performance in most cases. Currently, additional transistors are often required to not only regulate the analog conductance but also eliminate the sneak current in crossbar arrays, leading to a large circuit footprint and high power consumption. − On the other hand, interfacial RRAM operates on the basis of adjusting the Schottky or tunneling barrier at the active electrode–oxide or oxide–oxide interfaces, via charge injection or oxygen vacancy migration (middle panel of Figure a). , In general, the interface-type RRAM demonstrates analog switching behavior and substantially higher device-to-device (D2D) and cycle-to-cycle (C2C) uniformity than that of filamentary RRAMs as a result of homogeneous changes through the oxides. , Particularly, the highly nonlinear current–voltage ( I – V ) characteristics may allow for operating a passive crossbar array without the need for transistors or selector devices. , However, the low V O •• mobility inside an oxide switching layer and the high depolarization field after charge redistribution present a significant challenge for operation speed and data retention of interfacial RRAMs. ,, Therefore, a RRAM with high uniformity and large nonlinearity in I – V behavior while maintaining fast operation speed and good retention remains elusive.…”

Section: Introductionmentioning

confidence: 64%

“…The blue and green lines represent experimental data reported in representative works of filamentary RRAMs and interfacial RRAMs, respectively. It can be concluded that the PM-ATM (represented by the red line), which uses active metal nanocrystals to modulate the tunneling gap, demonstrates optimal performance as resistive memories in various 35 Ta/Ta 2 O 5 :Ag/Ru, 28 Au/multilayer h-BN/Au, 36 and Ag/SiGe/p-Si 7 ) and interfacial RRAMs (Pt/TiO x /Ti, 17 Pt/ TiO 2 :Na/Pt, 19 Pt/Ta 2 O 5 /HfO 2−x /TiN, 24 and Pt/Nb:SrTiO 3 /Al 37 ). Both C2C and D2D variations in the graph are set voltage variation.…”

Section: ■ Results and Discussionmentioning

confidence: 99%

“…Radar comparison graph of the device merits of the PM-ATM with those state-of-the-art RRAMs, including filamentary RRAMs (Ag/nanopatterned SiO 2 /Pt, Ta/Ta 2 O 5 :Ag/Ru, Au/multilayer h-BN/Au, and Ag/SiGe/p-Si) and interfacial RRAMs (Pt/TiO x /Ti, Pt/TiO 2 :Na/Pt, Pt/Ta 2 O 5 /HfO 2– x /TiN, and Pt/Nb:SrTiO 3 /Al). Both C2C and D2D variations in the graph are set voltage variation.…”

Section: Resultsmentioning

confidence: 99%

“…14,15 In general, the interface-type RRAM demonstrates analog switching behavior and substantially higher device-to-device (D2D) and cycle-to- cycle (C2C) uniformity than that of filamentary RRAMs as a result of homogeneous changes through the oxides. 16,17 Particularly, the highly nonlinear current−voltage (I−V) characteristics may allow for operating a passive crossbar array without the need for transistors or selector devices. 18,19 However, the low V O •• mobility inside an oxide switching layer and the high depolarization field after charge redistribution present a significant challenge for operation speed and data retention of interfacial RRAMs.…”

Section: ■ Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Analog Tunnel Memory Based on Programmable Metallization for Passive Neuromorphic Circuits

Chen

et al. 2022

ACS Appl. Mater. Interfaces

View full text Add to dashboard Cite

Although experimental implementations of memristive crossbar arrays have indicated the potential of these networks for in-memory computing, their performance is generally limited by an intrinsic variability on the device level as a result of the stochastic formation of conducting filaments. A tunnel-type memristive device typically exhibits small switching variations, owing to the relatively uniform interface effect. However, the low mobility of oxygen ions and large depolarization field result in slow operation speed and poor retention. Here, we demonstrate a quantum-tunneling memory with Ag-doped percolating systems, which possesses desired characteristics for large-scale artificial neural networks. The percolating layer suppresses the random formation of conductive filaments, and the nonvolatile modulation of the Fowler−Nordheim tunneling current is enabled by the collective movement of active Ag nanocrystals with high mobility and a minimal depolarization field. Such devices simultaneously possess electroforming-free characteristics, record low switching variabilities (temporal and spatial variation down to 1.6 and 2.1%, respectively), nanosecond operation speed, and long data retention (>10 4 s at 85 °C). Simulations prove that passive arrays with our analog memory of large current−voltage nonlinearity achieve a high write and recognition accuracy. Thus, our discovery of the unique tunnel memory contributes to an important step toward realizing neuromorphic circuits.

show abstract

Section: Introductionmentioning

confidence: 64%

Section: ■ Results and Discussionmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: ■ Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Analog Tunnel Memory Based on Programmable Metallization for Passive Neuromorphic Circuits

Chen

et al. 2022

ACS Appl. Mater. Interfaces

View full text Add to dashboard Cite

show abstract

“…Their operation also requires the presence of magnetic fields. Other approaches for spiking or oscillatory devices and circuits-including Mott-transition-based memristive devices 20,21 , ferroelectrics 22 , photonics 23 and two-dimensional materials 24 -have been developed, but all of them encounter similar problems. By omitting various aspects of actual biological wetware, artificial neurons based on electronics are insufficiently capable of emulating/handling the biosignal diversity and thus of operating in situ in biological environments.…”

mentioning

confidence: 99%

An organic artificial spiking neuron for in situ neuromorphic sensing and biointerfacing

et al. 2022

View full text Add to dashboard Cite

The effective mimicry of neurons is key to the development of neuromorphic electronics. However, artificial neurons are not typically capable of operating in biological environments, which limits their ability to interface with biological components and to offer realistic neuronal emulation. Organic artificial neurons based on conventional circuit oscillators have been created, but they require many elements for their implementation. Here we report an organic artificial neuron that is based on a compact nonlinear electrochemical element. The artificial neuron can operate in a liquid and is sensitive to the concentration of biological species (such as dopamine or ions) in its surroundings. The system offers in situ operation and spiking behaviour in biologically relevant environments—including typical physiological and pathological concentration ranges (5–150 mM)—and with ion specificity. Small-amplitude (1–150 mV) electrochemical oscillations and noise in the electrolytic medium shape the neuronal dynamics, whereas changes in ionic (≥2% over the physiological baseline) and biomolecular (≥ 0.1 mM dopamine) concentrations modulate the neuronal excitability. We also create biohybrid interfaces in which an artificial neuron functions synergistically and in real time with epithelial cell biological membranes.

show abstract

Functional Materials for Memristor‐Based Reservoir Computing: Dynamics and Applications

Zhang

Qin

Zhang

et al. 2023

Adv Funct Materials

View full text Add to dashboard Cite

The booming development of artificial intelligence (AI) requires faster physical processing units as well as more efficient algorithms. Recently, reservoir computing (RC) has emerged as an alternative brain‐inspired framework for fast learning with low training cost, since only the weights associated with the output layers should be trained. Physical RC becomes one of the leading paradigms for computation using high‐dimensional, nonlinear, dynamic substrates. Among them, memristor appears to be a simple, adaptable, and efficient framework for constructing physical RC since they exhibit nonlinear features and memory behavior, while memristor‐implemented artificial neural networks display increasing popularity towards neuromorphic computing. In this review, the memristor‐implemented RC systems from the following aspects: architectures, materials, and applications are summarized. It starts with an introduction to the RC structures that can be simulated with memristor blocks. Specific interest then focuses on the dynamic memory behaviors of memristors based on various material systems, optimizing the understanding of the relationship between the relaxation behaviors and materials, which provides guidance and references for building RC systems coped with on‐demand application scenarios. Furthermore, recent advances in the application of memristor‐based physical RC systems are surveyed. In the end, the further prospects of memristor‐implemented RC system in a material view are envisaged.

show abstract

Experimental demonstration of highly reliable dynamic memristor for artificial neuron and neuromorphic computing

Cited by 135 publications

References 47 publications

Analog Tunnel Memory Based on Programmable Metallization for Passive Neuromorphic Circuits

Analog Tunnel Memory Based on Programmable Metallization for Passive Neuromorphic Circuits

An organic artificial spiking neuron for in situ neuromorphic sensing and biointerfacing

Functional Materials for Memristor‐Based Reservoir Computing: Dynamics and Applications

Contact Info

Product

Resources

About