Search

Physical reservoir computing is a promising way to develop efficient artificial intelligence using physical devices exhibiting nonlinear dynamics. Although magnetic materials have advantages in miniaturization, the need for a magnetic field and large electric current results in high electric power consumption and a complex device structure. To resolve these issues, we propose a redox-based physical reservoir utilizing the planar Hall effect and anisotropic magnetoresistance, which are phenomena described by different nonlinear functions of the magnetization vector that do not need a magnetic field to be applied. The expressive power of this reservoir based on a compact all-solid-state redox transistor is higher than the previous physical reservoir. The normalized mean square error of the reservoir on a second-order nonlinear equation task was 1.69 × 10 -3 , which is lower than that of a memristor array (3.13 × 10 -3 ) even though the number of reservoir nodes was fewer than half that of the memristor array.

Artificial synapses with ideal functionalities are essential in hardware neural networks to allow for energy-efficient analog computing. However, the realization of linear and symmetric weight updates in real synaptic devices has proven challenging and ultimately limits the online training capabilities of neural network systems. Herein, we investigate the effect of Mg doping on a LiCoO 2 (LCO) channel in a Li ion-gated synaptic transistor, so as to improve long-term and short-term plasticity. Two transistor structures, based on a lithium phosphorus oxynitride electrolyte, were examined by using undoped LCO and Mg-doped LCO as the channel material between the source and drain electrodes. It was found that Mg doping increased the initial channel conductance by 3 orders of magnitude, which is probably due to the substitution of Co 3+ by Mg 2+ and the compensation of hole creation. It was further found that the doped channel transistor showed good retention characteristics and better linearity of long-term potentiation and depression when voltage pulses were applied to the gate electrode. The improved retention and linearity are attributed to an extended range of the insulator-to-conductor transition by Mg doping and Li-ion extraction/insertion cooperated in the LCO channel. Using the obtained synaptic weight update, artificial neural network simulations demonstrated that the doped channel transistor shows an image recognition accuracy of ∼80% for handwritten digits, which is higher than ∼65% exhibited by the undoped channel transistor. Mg doping also improved short-term plasticity such as paired-pulse facilitation/depression and Hebbian spike timing-dependent plasticity. These results indicate that elemental doping to the channel of Li ion-gated synaptic transistors could be a useful procedure for realizing robust neuromorphic systems based on analog computing.

Neuromorphic processors using artificial neural networks are the center of attention for energy-efficient analog computing. Artificial synapses act as building blocks in such neural networks for parallel information processing and data storage. Herein we describe the fabrication of a proton-gated synaptic transistor using a Nafion electrolyte thin film, which is patterned by electron-beam lithography (EBL). The device has an active channel of indium-zinc-oxide (IZO) between the source and drain electrodes, which shows Ohmic behavior with a conductance level on the order of 100 μS. Under voltage applications to the gate electrode, the channel conductance is changed due to the injection and extraction of protons between the IZO channel and the Nafion electrolyte, emulating various synaptic functions with short-term and long-term plasticity. When positive (negative) gate voltage pulses are consecutively applied, the device exhibits long-term potentiation (depression) at the same number of steps as the number of input pulses. Based on these characteristics, an artificial neural network using this transistor shows ∼84% image recognition accuracy for handwritten digits. The subject transistor also successfully mimics paired-pulse facilitation and depression, Hebbian spike-timing-dependent plasticity, and Pavlovian associative learning followed by extinction activities. Finally, dynamical pattern image memorization is demonstrated in a 5 × 5 array of these synaptic transistors. The results indicate that EBL patternable Nafion electrolytes have great potential for use in the fabrication and circuit-level integration of synaptic devices for neuromorphic computing applications.

Center-surround antagonism, a key mechanism in the retina, contributes to the encoding of edge contrast rather than of the overall information on a visual image. Here, a neuromorphic system consisting of multiple ionic devices is built, where each device has a lithium cobalt oxide channel arranged on a common lithium phosphorus oxynitride electrolyte. Because of the migration of Li ions between the channels through the electrolyte, the devices are highly interactive, as is seen with retinal neurons. On the basis of the excitation of single devices and device-to-device inhibition, the system successfully emulates the antagonistic center-surround receptive field and the Mach band effect in which perceived contrast is enhanced at the edges between dark and bright regions. Furthermore, a two-dimensional array system is simulated to implement edge detection for real images. This scheme enables computer vision tasks with simple and effective operations, owing to the intrinsic properties of the materials employed.

An all-solid-state redox device, composed of magnetite (Fe 3 O 4 ) thin film and Li + conducting electrolyte thin film, was fabricated for the manipulation of a magnetization angle at room temperature (RT). This is a key technology for the creation of efficient spintronics devices, but has not yet been achieved at RT by other carrier doping methods. Variations in magnetization angle and magnetic stability were precisely tracked through the use of planar Hall measurements at RT. The magnetization angle was reversibly manipulated at 10° by maintaining magnetic stability. Meanwhile, the manipulatable angle reached 56°, although the manipulation became irreversible when the magnetic stability was reduced. This large manipulation of magnetic angle was achieved through tuning of the 3d electron number and modulation of the internal strain in the Fe 3 O 4 due to the insertion of high-density Li + (approximately 10 21 cm -3 ). This RT manipulation is applicable to highly integrated spintronics devices due to its simple structure and low electric power consumption.

An all-solid-state redox device composed of Fe3O4 thin film and Li(+) ion conducting solid electrolyte was fabricated for use in tuning magnetization and magnetoresistance (MR), which are key factors in the creation of high-density magnetic storage devices. Electrical conductivity, magnetization, and MR were reversibly tuned by Li(+) insertion and removal. Tuning of the various Fe3O4 thin film properties was achieved by donation of an electron to the Fe(3+) ions. This technique should lead to the development of spintronics devices based on the reversible switching of magnetization and spin polarization (P). It should also improve the performance of conventional magnetic random access memory (MRAM) devices in which the ON/OFF ratio has been limited to a small value due to a decrease in P near the tunnel barrier.

Nanoscaled ionic conductors are crucial for future nanodevices. A well-known ionic conductor, AgI, exhibited conductivity greater than 1 Ω(-1) cm(-1) in α-phase and transformed into poorly conducting β-/γ-phase below 147 °C, thereby limiting applications. Here, we report that transition temperatures both from the β-/γ- to α-phase (Tc↑) and the α- to β-/γ-phase (Tc↓) are tuned by AgI/Ag heteronanowires embedded in anodic aluminum oxide (AAO) membranes with 10-30 nm pores. Tc↑ and Tc↓ shift to correspondingly higher and lower temperature as pore size decreases, generating a progressively enlarged thermal hysteresis. Tc↑ and Tc↓ specifically achieve 185 and 52 °C in 10 nm pores, and the final survived conductivity reaches ∼8.3 × 10(-3) Ω(-1) cm(-1) at room temperature. Moreover, the low-temperature stabilizing α-phase (down to 21 °C, the lowest in state of the art temperatures) is reproducible and survives further thermal cycling. The low-temperature phase stabilization and enhancement conductivity reported here suggest promising applications in silver-ion-based future nanodevices.

An all-solid-state electric-double-layer transistor (EDLT) with a Y-stabilized ZrO₂ (YSZ) proton conductor/SrTiO₃ (STO) single crystal has been fabricated to investigate ionic conductivity effect on the response speed, which should be a key parameter for development of next-generation EDLTs. The drain current exhibited a 4-order-of-magnitude increment by electrostatic carrier doping at the YSZ/STO interface due to ion migration, and the behavior strongly depended on the operation temperature. An Arrhenius-type plot of the ionic conductivity (σ(i)) in the YSZ and t(c)⁻¹, which is a current-rise time needed for charge accumulation at the YSZ/STO interface, shows a synchronized variation, indicating a proportional relationship between the two parameters. Analysis of the σ(i)-t(c) diagram shows that, in contrast to conventional EDLTs, the response speed should reach picosecond order at room temperature by using extreme miniaturization and superionic conductors. Furthermore, the diagram indicates that plenty of solid electrolytes, which have not been used due to the lack of criteria for evaluation, can be a candidate for all-solid-state EDLTs exceeding the carrier density of conventional EDLTs, even though the response speed becomes comparably lower than those of FETs.

In situ and nonvolatile tuning of photoluminescence (PL) has been achieved based on graphene oxide (GO), the PL of which is receiving much attention because of various potential applications of the oxide (e.g., display, lighting, and nano-biosensor). The technique is based on in situ and nonvolatile tuning of the sp(2) domain fraction to the sp(3) domain fraction (sp(2)/sp(3) fraction) in GO through an electrochemical redox reaction achieved by solid electrolyte thin films. The all-solid-state variable PL device was fabricated by GO and proton-conducting mesoporous SiO2 thin films, which showed an extremely low PL background. The device successfully tuned the PL peak wavelength in a very wide range from 393 to 712 nm, covering that for chemically tuned GO, by adjusting the applied DC voltage within several hundred seconds. We also demonstrate the sp(2)/sp(3) fraction tuning using a conductive atomic force microscope. The device achieved not only writing, but also erasing of the sp(2)/sp(3)-fraction-tuned nanodomain (both directions operation). The combination of these techniques is applicable to a wide range of nano-optoelectronic devices including nonvolatile PL memory devices and on-demand rewritable biosensors that can be integrated into nano- and microtips which are transparent, ultrathin, flexible, and inexpensive.

A potential route to extend Moore's law beyond the physical limits of existing materials and device architectures is to achieve nanotechnology breakthroughs in materials and device concepts. Here, we discuss an on-demand WO(3-x)-based nanoionic device where electrical and neuromorphic multifunctions are realized through externally induced local migration of oxygen ions. The device is found to possess a wide range of time scales of memorization, resistance switching, and rectification varying from volatile to permanent in a single device, and these can furthermore be realizable in both two- or three-terminal systems. The gradually changing volatile and nonvolatile resistance states are experimentally demonstrated to mimic the human brain's forgetting process for short-term memory and long-term memory.We propose this nanoionic device with its on-demand electrical and neuromorphic multifunction has a unique paradigm shifting potential for the fabrication of configurable circuits, analog memories, digital-neural fused networks, and more in one device architecture.