In-sensor computing using Ti3C2Tx MXene memristor crossbar arrays for wearable electronics DOI
Jeny Gosai, Mansi Patel,

Anjalee Gosai

et al.

Flexible and Printed Electronics, Journal Year: 2024, Volume and Issue: 9(4), P. 045013 - 045013

Published: Dec. 1, 2024

Abstract The potential of memristor systems in sensing, storing, and processing signals make them highly efficient ideal for power-efficient, comfortable wearable in-sensor computing applications. In this work, we demonstrate a 3 × crossbar array based on Ti C 2 T x MXene with non-volatile characteristics, exhibiting an ON/OFF ratio ∼10 . This -based also showcases remarkable synaptic properties. Additionally, achieve near perfect accuracy pattern training after just 9 epochs as well retaining ability even 24 h. A notable feature these arrays is their to integrate storage, capabilities, demonstrated real-time muscle monitoring healthcare device. multi-channel surface electromyography data was recorded using the MXene-based track forearm movements during series distinct hand gestures. These findings open up exciting possibilities development adaptable flexible memristive arrays, which hold great promise advanced neuromorphic computing,

Language: Английский

Control-Etched Ti3C2Tx MXene Nanosheets for a Low-Voltage-Operating Flexible Memristor for Efficient Neuromorphic Computation DOI
Jeny Gosai, Mansi Patel, Lingli Liu

et al.

ACS Applied Materials & Interfaces, Journal Year: 2024, Volume and Issue: 16(14), P. 17821 - 17831

Published: March 27, 2024

Hardware neural networks with mechanical flexibility are promising next-generation computing systems for smart wearable electronics. Overcoming the challenge of developing a fully synaptic plastic network, we demonstrate low-operating-voltage PET/ITO/p-MXene/Ag flexible memristor device by controlling etching aluminum metal ions in Ti3C2Tx MXene. The presence small fraction Al partially etched MXene (p-Ti3C2Tx) significantly suppresses operating voltage to 1 V compared 7 from (f-Ti3C2Tx)-based devices. Former devices exhibit excellent non-volatile data storage properties, robust ∼103 ON/OFF ratio, high endurance ∼104 cycles, multilevel resistance states, and long retention measured up ∼106 s. High stability ∼73° bending angle environmental robustness confirmed consistent switching characteristics under increasing temperature humid conditions. Furthermore, p-Ti3C2Tx is employed mimic biological synapse measuring learning–forgetting pattern cycles as potentiation depression. Spike time-dependent plasticity (STDP) based on Hebb's Learning rules also successfully demonstrated. Moreover, remarkable accuracy ∼95% recognizing modified patterns National Institute Standards Technology (MNIST) set just 29 training epochs achieved simulation. Ultimately, our findings underscore potential MXene-based versatile components neuromorphic computing.

Language: Английский

Citations

21
Robust hybrid perovskite self-rectifying memristor for brain-inspired computing and data storage DOI Creative Commons

Manish Khemnani,

Muskan Jain,

Denish Hirpara

et al.

Journal of Applied Physics, Journal Year: 2025, Volume and Issue: 137(4)

Published: Jan. 23, 2025

Conventional computing architectures are not suited to meet the unique workload requirements of artificial intelligence and deep learning, which has sparked a growing interest in memory-centric computing. One primary challenge this field is sneak path current memory devices, degrades data storage reliability. Another critical issue ensuring device performance stability over time under varying environmental conditions. To overcome these challenges, work, we introduce Dion–Jacobson perovskite-based self-rectifying cell that only reduces but also demonstrates remarkable electrical parameters. The fabricated maintains consistent performance, including rectification ratio (∼103), on/off set voltage (∼0.52 V), for 200+ days within temperature range 25–70 °C relative humidity conditions up 70%RH. Importantly, our work represents an innovative step forward observation self-rectification stable showing way their widespread application architectures. Furthermore, understand behavior across its different states, i.e., high resistance state low state, electrochemical impedance spectroscopy performed, gives insight into individual contribution resistance, capacitance, inductance.

Language: Английский

Citations

1
Synapse and resistance switching behavior of La:HfO2/ZrO2/La:HfO2 memristors DOI

Y.K. Su,

Yan-Ping Jiang,

Jiayu Tang

et al.

International Journal of Modern Physics B, Journal Year: 2025, Volume and Issue: unknown

Published: March 8, 2025

The von Neumann bottleneck in traditional computers has hindered the rapid development of artificial intelligence. To improve computational efficiency, memristors have become a preferred device to mimic synaptic behavior and achieve neuromorphic computing, thus attracting widespread attention. In this work, La:HfO 2 /ZrO /La:HfO thin films were prepared via sol–gel deposition. When Zr was inserted as an interlayer into 6% La-doped HfO , significant resistance switching (RS) detected through voltage scanning over 100 consecutive cycles, its electrical performance enhanced compared case when there no interlayer. presence Analog switch enabled effectively simulate properties such long-term potentiation/inhibition, short-term paired-pulse facilitation, spike-timing-dependent plasticity learning rules. Moreover, exhibited good linearity weight updates excellent conductance modulation performance. Utilizing convolutional neural network architecture, information [Formula: see text] pixel array classified processed, thereby improving recognition accuracy Mixed National Institute Standards & Technology (MNIST) dataset 97.5% that Fashion-MNIST 87.0%. These advancements provided viable solution for successful construction systems future.

Language: Английский

Citations

0
BDAPbI4 Dion Jacobson hybrid perovskite-based artificial nociceptors on biodegradable substrate DOI

Manish Khemnani,

Parth Thakkar,

Aziz Lokhandvala

et al.

Sensors and Actuators A Physical, Journal Year: 2024, Volume and Issue: 373, P. 115382 - 115382

Published: April 16, 2024

Language: Английский

Citations

3
Memristor-Based Neuromorphic Computing and Artificial Neural Networks for Computer Vison and AI—Applications DOI

Prince Patel,

Mansi Patel, Ankur Solanki

et al.

Biological and medical physics series, Journal Year: 2024, Volume and Issue: unknown, P. 307 - 322

Published: Jan. 1, 2024

Language: Английский

Citations

1
Insights of BDAPbI4-Based Flexible Memristor for Artificial Synapses and In-Memory Computing DOI Creative Commons
Mansi Patel, Jeny Gosai,

Prince Patel

et al.

ACS Omega, Journal Year: 2024, Volume and Issue: 9(47), P. 46841 - 46850

Published: Nov. 16, 2024

Inspired by brain-like spiking computational frameworks, neuromorphic computing-brain-inspired computing for machine intelligence promises to realize artificial (AI) while reducing the energy requirements of platforms. In this work, we show potential advanced learnings butane-1,4-diammonium based low-dimensional Dion-Jacobson hybrid perovskite (BDAPbI4) memristor devices in realm synapses and computing. Memristors validate Hebbian learning rules with various spike-dependent plasticity within a 10 ± 2 ms time frame, reminiscent human brains under flat bending conditions (∼5 mm radium). A high recognition accuracy ∼94% handwritten images from MNIST database via an neural network (ANN) is achieved only 50 epochs. An efficient demonstration second-order memristors Pavlovian dog experiment exhibit significant promise expediting memory consolidation. To showcase in-memory potential, flexible 4 × crossbar array designed measured data retention up ∼103 s along 26 multilevel resistance states. The successfully programmed facile configurability image "Z". conclusion, integration supervised, unsupervised, associative holds great across spectrum future technologies, ranging networks computing, brain-machine interfaces, adaptive control systems.

Language: Английский

Citations

1
Nano Granular Metallic Thin Films: Unravelling Non-Linear Electrical Conduction and Resistive Switching for Neuromorphic Applications DOI

P. B. Khatkale,

Amit Khatri, P. M. Yawalkar

et al.

Journal of Nano- and Electronic Physics, Journal Year: 2024, Volume and Issue: 16(3), P. 03015 - 5

Published: Jan. 1, 2024

The arbitrarily formed golden cluster systems were created in the gas state which has strong Resistive Switching (RS) behavior around ambient temperatures and makes these attractive candidates for creation of electronics geared toward neuron categorization along with information analysis.The cluster-assembled nanotechnology coatings that are fully linked have an irregular shape includes neuromorphic crystallographic flaws, interactions frontiers grains, highlighting complex interaction among electromagnetic mechanical elements.In this analysis, we conduct a thorough investigation electroforming procedure is utilized film was assembled.The present research sheds light regarding procedure's substantial influence on relations nanopores mesoscale layer formations underlying neurological properties resistance switches activities ensure.The findings provide insight into methodical oversight operation reveal its function building distinct patterns at various sizes films discovery not only improves our understanding intricate relationships architectural electrical parts but it provides opportunities designing structures randomly constructed customized over multiple information-handling applications.

Language: Английский

Citations

0
In-sensor computing using Ti3C2Tx MXene memristor crossbar arrays for wearable electronics DOI
Jeny Gosai, Mansi Patel,

Anjalee Gosai

et al.

Flexible and Printed Electronics, Journal Year: 2024, Volume and Issue: 9(4), P. 045013 - 045013

Published: Dec. 1, 2024

Abstract The potential of memristor systems in sensing, storing, and processing signals make them highly efficient ideal for power-efficient, comfortable wearable in-sensor computing applications. In this work, we demonstrate a 3 × crossbar array based on Ti C 2 T x MXene with non-volatile characteristics, exhibiting an ON/OFF ratio ∼10 . This -based also showcases remarkable synaptic properties. Additionally, achieve near perfect accuracy pattern training after just 9 epochs as well retaining ability even 24 h. A notable feature these arrays is their to integrate storage, capabilities, demonstrated real-time muscle monitoring healthcare device. multi-channel surface electromyography data was recorded using the MXene-based track forearm movements during series distinct hand gestures. These findings open up exciting possibilities development adaptable flexible memristive arrays, which hold great promise advanced neuromorphic computing,

Language: Английский

Citations

0