NbO2-based locally active memristors: from physical mechanisms to performance optimization

Chen, Pei; Zhang, Xumeng; Liu, Qi; Liu, Ming

doi:10.1007/s00339-022-06258-6

Cited by 7 publications

(3 citation statements)

References 83 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Once the applied voltage is less than the hold voltage ( V Hold ), the device spontaneously switches back to the HRS. Due to the electron-related switching mechanism, the device exhibits a fast switching speed (~1 ns) 22 and high endurance (>10 10 cycles) 23 , which is decent for neuronal applications. According to the spike numbers within one cluster, the spiking and bursting features are two basic features that enable the neurons’ various firing behaviors.…”

Section: Resultsmentioning

confidence: 99%

“…Nevertheless, it is expensive to implement such a computation principle in conventional hardware platforms by simulating high-fidelity neurons, especially those with diverse firing behaviors that need rich dynamics [15][16][17][18] . Locally active memristors (LAMs), with rich dynamics [19][20][21] , low power consumption 22 and good scalability [23][24][25] , show intriguing potential to build Hodgkin-Huxley (H-H) neurons with excellent bio-plausibility [26][27][28][29] . However, current works typically focus on the emulation of neurons' firing behaviors, and the exhibition of firing features' computational capability remains an open question (see Supplementary Table 1).…”

mentioning

confidence: 99%

See 1 more Smart Citation

Firing feature-driven neural circuits with scalable memristive neurons for robotic obstacle avoidance

Yang,

Zhu,

Zhang

et al. 2024

Nat Commun

Self Cite

View full text Add to dashboard Cite

Neural circuits with specific structures and diverse neuronal firing features are the foundation for supporting intelligent tasks in biology and are regarded as the driver for catalyzing next-generation artificial intelligence. Emulating neural circuits in hardware underpins engineering highly efficient neuromorphic chips, however, implementing a firing features-driven functional neural circuit is still an open question. In this work, inspired by avoidance neural circuits of crickets, we construct a spiking feature-driven sensorimotor control neural circuit consisting of three memristive Hodgkin-Huxley neurons. The ascending neurons exhibit mixed tonic spiking and bursting features, which are used for encoding sensing input. Additionally, we innovatively introduce a selective communication scheme in biology to decode mixed firing features using two descending neurons. We proceed to integrate such a neural circuit with a robot for avoidance control and achieve lower latency than conventional platforms. These results provide a foundation for implementing real brain-like systems driven by firing features with memristive neurons and put constructing high-order intelligent machines on the agenda.

show abstract

Section: Resultsmentioning

confidence: 99%

mentioning

confidence: 99%

Firing feature-driven neural circuits with scalable memristive neurons for robotic obstacle avoidance

Yang,

Zhu,

Zhang

et al. 2024

Nat Commun

Self Cite

View full text Add to dashboard Cite

show abstract

“…With increasing current flow, the Joule heating effect will be more significant and hence resulting in the broadening of the metallic filament [76]. There have been several reports on the Poole-Frenkel formula to describe the thermal runaway mechanism occurring in IMT devices, with the dependence being electric field and thermally driven [77][78][79]. The increase in the electric field will cause the filament temperature to increase.…”

Section: Negative Differential Resistance (Ndr) Phenomenonmentioning

confidence: 99%

Investigation of the dynamics of NbOx-based threshold switching devices for next-generation computing and wearable electronics

Ang

View full text Add to dashboard Cite

As the current CMOS-based memory technologies face challenges to scale down to 10 nm node and beyond, scalable memory devices are highly sought after. The emergence of Resistive Random Access Memory (ReRAM) as a highly promising research area for next-generation non-volatile memory has garnered significant attention due to its exceptional memory performance, including high density, low power consumption, and scalability. However, the prevailing issue of sneak paths in ReRAM arrays leads to current leakage and data errors. To address this issue, the incorporation of a selector device is crucial for controlling the current flow in individual memory cells. NbOx-based Insulator-Metal Transition (IMT) devices have shown advantages as selector devices, offering fast-switching speed, high scalability, and excellent thermal stability. Additionally, their inherent Negative Differential Resistance (NDR) characteristic enables them to function as oscillatory neurons in Oscillatory Neural Networks (ONNs), mimicking biological neural networks and providing robustness and efficient information processing. This thesis focuses on studying the dynamics and characteristics of NbOx-based IMT devices for both relaxation oscillator and selector applications, and exploring their potential and performance reliability under different conditions of temperature and mechanical strain. The frequency tunability of NbOx-based oscillatory devices was investigated through various tuning methods, achieving a wide tunable range from 2 MHz to 33 MHz. The performance enhancements achieved through bulk and interfacial treatment, coupled with high endurance and operation at high oscillation frequencies, make these devices promising candidates for ONN applications, enabling enhanced information processing and efficient computing paradigms. Further investigating into the tunability of NbOx-based relaxation oscillators, the influence of oxygen stoichiometry reveals significant improvements in device performance. The findings provide insight into the factors affecting NbOx-based oscillator performance and offer opportunities for further optimization, enhancing their suitability for advanced computing architectures. The functionality of NbOx-based selector devices on flexible substrates was demonstrated, highlighting their mechanical endurance and recovery capabilities. The findings suggest the UV Ultraviolet VCM Valence Change Memory VC-NDR Voltage-Controlled Negative Differential Resistance VCO Voltage-Controlled Oscillator VRH Mott Variable Range Hopping WL Word Line XPS X-ray Photoelectron Spectroscopy

show abstract

Memristive Hodgkin–Huxley Neurons with Diverse Firing Patterns for High‐Order Neuromorphic Computing

Yang,

Zhang,

Chen

et al. 2024

Advanced Intelligent Systems

View full text Add to dashboard Cite

The rich firing behaviors of biological neurons enable the nervous system to execute complex computations, emulating which in hardware is advantageous for constructing advanced intelligent machines. Hodgkin–Huxley (H–H) neurons based on memristors feature great merits of high bio‐plausibility and low hardware cost. However, a universal design rule of memristive H–H neurons is still lacking, hindering its development and applications. Herein, a universal H–H neuron circuit structure is proposed and its feasibility based on NbOx memristors is demonstrated. The constructed neuron achieves 23 types of firing behaviors observed in biological neurons, simplifying the communication between neurons. To better understand the correlation between circuit parameters and firing patterns, the firing patterns into three classes according to the switching cycle ratio of two memristors are categorized. The circuit design rules of each category of firing patterns are deeply elucidated and universal regularities for tuning circuit parameters to implement the switch between different firing behaviors are presented. Finally, the potential applications of different firing behaviors in neuromorphic intelligence systems are discussed. This work provides theoretical guidance for engineering memristive H–H neuron circuits, assisting in building high‐order neuromorphic systems based on firing patterns.

show abstract

NbO2-based locally active memristors: from physical mechanisms to performance optimization

Cited by 7 publications

References 83 publications

Firing feature-driven neural circuits with scalable memristive neurons for robotic obstacle avoidance

Firing feature-driven neural circuits with scalable memristive neurons for robotic obstacle avoidance

Investigation of the dynamics of NbOx-based threshold switching devices for next-generation computing and wearable electronics

Memristive Hodgkin–Huxley Neurons with Diverse Firing Patterns for High‐Order Neuromorphic Computing

Contact Info

Product

Resources

About