Source-linked AI summary
Truly concomitant and independently expressed short- and long-term plasticity in Bi2O2Se-based three-terminal memristor
Ziyang Zhang, Tianran Li, Yujie Wu, Yinjun Jia, Congwei Tan, Xintong Xu, Guanrui Wang, Juan Lv, Wei Zhang, Yuhan He, Luping Shi, Hailin Peng, Huanglong Li
TL;DR
Existing memristors have demonstrated STP and LTP selectively rather than concomitantly, although their orchestration is associated with complex cognitive functions. This work demonstrates both in a three-terminal Bi2O2Se memristor using distinct physical mechanisms, with transient synaptic efficacy modulated from depression to facilitation. The authors also use a recurrent neural circuitry model to simulate sleep-wake cycle autoregulation.
Problem
Existing memristors have demonstrated STP and LTP selectively rather than concomitantly, despite the role of their orchestration in complex cognitive functions.
Source: pp. 3–4Method
A three-terminal Bi2O2Se memristor uses volatile channel-carrier effects for STP and persistent charge trapping or related gate-dielectric/interface changes for LTP.
Source: p. 4, S2. Mimicry of long-term potentiation… · p. 33Results
The device demonstrates concomitant STP and LTP, with their concerted action modulating transient synaptic efficacy from depression to facilitation.
Source: p. 21Takeaways & Limitations
The combined plasticity provides a versatile device platform for neuromorphic function implementation.
Source: p. 21The STP timescale extraction assumes that STP is negligible at a 1 s pulse interval.
Source: S2. Mimicry of long-term potentiation… · p. 37
Abstract
from arXiv · showhide
Orchestration of diverse synaptic plasticity mechanisms across different timescales produces complex cognitive processes. To achieve comparable cognitive complexity in memristive neuromorphic systems, devices that are capable to emulate short- and long-term plasticity (STP and LTP, respectively) concomitantly are essential. However, this fundamental bionic trait has not been reported in any existing memristors where STP and LTP can only be induced selectively because of the inability to be decoupled using different loci and mechanisms. In this work, we report the first demonstration of truly concomitant STP and LTP in a three-terminal memristor that uses independent physical phenomena to represent each form of plasticity. The emerging layered material Bi2O2Se is used in memristor for the first time, opening up the prospects for ultra-thin, high-speed and low-power neuromorphic devices. The concerted action of STP and LTP in our memristor allows full-range modulation of the transient synaptic efficacy, from depression to facilitation, by stimulus frequency or intensity, providing a versatile device platform for neuromorphic function implementation. A recurrent neural circuitry model is developed to simulate the intricate "sleep-wake cycle autoregulation" process, in which the concomitance of STP and LTP is posited as a key factor in enabling this neural homeostasis. This work sheds new light on the highly sophisticated computational capabilities of memristors and their prospects for realization of advanced neuromorphic functions.
Experimental Section
The Experimental Section describes fabrication and characterization of Bi2O2Se three-terminal memristors, hybrid-functional interface calculations, and a leaky integrate-and-fire model for simulating synaptic plasticity. The model represents short- and long-term effects separately and uses parameter values listed in Table 1.
Experimental Section
- Bi2O2Se memristors were fabricated from ~5 nm CVD samples with Pd/Au source-drain contacts and a 20 nm HfO2 gate dielectric.The devices have ~500 nm channel lengths; HfO2 was deposited by thermal atomic layer deposition at 70°C.Source: Experimental Section · pp. 21–22
- Electrical characterization used a Keysight B1500A analyzer and B1530A waveform generator at room temperature in ambient atmosphere.Source: Experimental Section · p. 22
- Interface band structures were calculated with sX-LDA on a relaxed 86-atom interface supercell after geometry optimization using the generalized gradient approximation.The calculations used CASTEP and a 680 eV plane-wave cutoff.Source: Experimental Section · p. 22
- The simulation uses a leaky integrate-and-fire model in which Heaviside and Dirac functions represent LTP and STP effects, respectively.After short-term modulation, conductance decays toward a baseline determined by LTP; simulation parameter values are listed in Table 1.Source: Experimental Section · pp. 23–24, Table 1 · p. 24
S1. The endurance, retention and operation speed of Bi2O2Se synaptic device
The supplementary results describe device cycling, retention, and switching measurements, then examine how pulse intensity and interval shape coexisting short- and long-term plasticity. The proposed mechanisms distinguish transient residual-carrier effects from lasting charge trapping.
S1. The endurance, retention and operation speed of Bi2O2Se synaptic device
- Figure S1 tracks lower- and higher-resistance states over 1000 sweep cycles, 24-hour retention, and switching speed in newly fabricated devices.Panels (a–d) use device 2 and panel (e) device 3; the measured switching timescale is described as comparable to fast charge injection in MOS-stack BTI measurements.Source: Figure S1 · p. 31
S2. Mimicry of long-term potentiation (LTp) using negative voltage pulse
- With 50 ms pulse intervals, strong pulses can leave residual electrons that favor short-term facilitation, while border-trap charge trapping reduces steady-state electrons and produces long-term depression.For weak pulses, the short-term facilitation may be too weak to outweigh the long-term depression.Source: S2. Mimicry of long-term potentiation… · p. 33
- Alternating-polarity pulse trains gradually increase postsynaptic current with negative pulses from -1 V to -9 V and reduce it with positive pulses from +2 V to +10 V.The pulses are 50 ms wide and separated by 5 s.Source: Figure S4 · p. 35, S2. Mimicry of long-term potentiation… · p. 35
- Figure S5 depicts concomitant short-term facilitation and long-term depression under positive pulses, and short-term depression with long-term potentiation under negative pulses.The reported low pulse intensities are 10 mV and 100 mV, respectively.Source: Figure S5 · p. 36
- The extracted short-term-plasticity time constants are 148 ms and 494 ms, based on the assumption that its effect is negligible at 1 s pulse intervals.The authors extract the effect by subtracting current values measured at 1 s intervals.Source: S2. Mimicry of long-term potentiation… · p. 37
- At short pulse intervals, residual electrons may raise conductivity at the next pulse, while border-trap trapping produces lasting steady-state depression; at long intervals, depression may predominate.The proposed account treats these as the electronic bases of short-term facilitation and long-term depression, respectively.Source: S2. Mimicry of long-term potentiation… · p. 38
- The supplementary sleep-wake simulation depicts enemy invasion during sleep triggering immediate wake-up and locomotion through Sy1 short-term facilitation.The caption identifies the invasion as higher-frequency stimulation than the background and notes sudden Ne1 spiking at midnight.Source: Figure S8 · p. 39, S2. Mimicry of long-term potentiation… · p. 39