Atomically Thin Synaptic Devices for Optoelectronic Neuromorphic Vision

The LaAlO<sub>3</sub>/SrTiO<sub>3</sub> interface has been one of the topics studied most during the past few years due to its many intriguing properties such as the two-dimensional electron gas, transient photoconductivity (PC), persistent photoconductivity (PPC), and the coexistence of the PC and PPC. Of them, the PPC effect is the most interesting because of its potential application in exploring the photoelectric memory devices. Until now, tuning of the PPC of the LaAlO<sub>3</sub>/SrTiO<sub>3</sub> interface under the external stimuli, such as electric or magnetic field is less addressed, while the relevant knowledge is of great value for exploring the memory devices with multifunctionality. In this paper, we report on an electric field control of the persistent PPC at the LaAlO<sub>3</sub>/SrTiO<sub>3</sub> interface. Our LaAlO<sub>3</sub>/SrTiO<sub>3</sub> heterojunction is fabricated by growing the LaAlO<sub>3</sub> film on the SrTiO<sub>3</sub> substrates through using pulsed laser deposition. The substrate temperature is kept at 750 ℃ and the partial pressure of oxygen is maintained at 3.3 × 10<sup>–5</sup> Torr (1 Torr = 1.33322 × 10<sup>2</sup> Pa) during the deposition. The thickness of LaAlO<sub>3</sub> film is controlled to be about 2 nm by setting an appropriate deposition time. The X-ray diffraction experiment confirms that the LAO film is well epitaxial and of single phase. To guarantee the good electric contacts, Al electrodes are soldered at the LaAlO<sub>3</sub>/SrTiO<sub>3</sub> interface and the back side of the SrTiO<sub>3</sub> respectively by ultrasonic welding. We find that the PPC at the LaAlO<sub>3</sub>/SrTiO<sub>3</sub> interface can be significantly reinforced and modulated by the light-enhanced gating effects: that is, after a negative back gate voltage processing combined with a simultaneous light illumination, the LaAlO<sub>3</sub>/SrTiO<sub>3</sub> interface can exhibit a notable PPC effect. And the PPC effect increases as the negative gate voltage increases, and then attains a maximum at a back gate voltage of about –70 V. Further increase of the negative gate voltage can cause the PPC to decrease. Additionally, the PPC is also found to increase monotonically with increasing the gating time. The present result can be understood in terms of the migration of the oxygen vacancies under the influence of photoelectric synergetic effect. This field enhanced PPC effects at the LaAlO<sub>3</sub>/SrTiO<sub>3</sub> interface may find their applications in designing the photoelectric memory devices with electric tunability.

Emulating synaptic functionalities in optoelectronic devices is significant in developing artificial visual-perception systems and neuromorphic photonic computing. Persistent photoconductivity (PPC) in metal oxides provides a facile way to realize the optoelectronic synaptic devices, but the PPC performance is often limited due to the oxygen vacancy defects that release excess conduction electrons without external stimuli. Herein, a high-performance optoelectronic synapse based on the stoichiometry-controlled LaAlO3/SrTiO3 (LAO/STO) heterostructure is developed. By increasing La/Al ratio up to 1.057:1, the PPC is effectively enhanced but suppressed the background conductivity at the LAO/STO interface, achieving strong synaptic behaviors. The spectral noise analyses reveal that the synaptic behaviors are attributed to the cation-related point defects and their charge compensation mechanism near the LAO/STO interface. The short-term and long-term plasticity is demonstrated, including the paired-pulse facilitation, in the La-rich LAO/STO device upon exposure to UV light pulses. As proof of concepts, two essential synaptic functionalities, the pulse-number-dependent plasticity and the self-noise cancellation, are emulated using the 5×5 array of La-rich LAO/STO synapses. Beyond the typical oxygen deficiency control, the results show how harnessing the cation stoichiometry can be used to design oxide heterostructures for advanced optoelectronic synapses and neuromorphic applications.

The slow relaxation of the persistent photoconductivity (PPC) effect in sulfur-doped hydrogenated amorphous silicon (a-Si:H) has been measured as a function of temperature and illumination time. The relaxation is found to be thermally activated, with an activation energy which varies with sulfur concentration, while illuminating the film for a longer time leads to a longer relaxation time. A correlation is observed between changes of the photoconductivity during illumination and the magnitude of the PPC effect following illumination. These effects are also observed in compensated a-Si:H, suggesting that the mechanism for the PPC effect is the same in both sulfur-doped a-Si:H and compensated a-Si:H. The presence of donor and compensating acceptor states in sulfur-doped a-Si:H could arise from valence alternation pair sulfur atom defects.

The Persistent Photoconductivity (PPC) effect was studied in individual tin oxide (SnO2) nanobelts as a function of temperature, in air, helium, and vacuum atmospheres, and low temperature Photoluminescence measurements were carried out to study the optical transitions and to determine of the acceptor/donors levels and their best representation inside the band gap. Under ultraviolet (UV) illumination and at temperatures in the range of 200 to 400K we observed a fast and strong enhancement of the photoconductivity, and the maximum value of the photocurrent induced increases as the temperature or the oxygen concentration decreases. By turning off the UV illumination the induced photocurrent decays with lifetimes up to several hours. The photoconductivity and the PPC results were explained by adsorption and desorption of molecular oxygen at the surface of the SnO2 nanobelts. Based on the temperature dependence of the PPC decay an activation energy of 230 meV was found, which corresponds to the energy necessary for thermal ionization of free holes from acceptor levels to the valence band, in agreement with the photoluminescence results presented. The molecular-oxygen recombination with holes is the origin of the PPC effect in metal oxide semiconductors, so that, the PPC effect is not related to the oxygen vacancies, as commonly presented in the literature.

Neuromorphic computing systems are based on new architecture inspired by biological learning behavior and biological structure to overcome the shortcomings of the von Neumann system, such as high energy consumption and generation of excessive redundant data. However, common synaptic‐device designs are still need to carry an external power source, which is detrimental to both the miniaturization of smart devices and the further reduction of energy consumption. Therefore, self‐powered optoelectronic synapses hold great promise for achieving energy‐efficient artificial intelligence applications. Given the advantages of heterogeneous integration and ferroelectric performance, the 2D ferroelectric In2Se3 has become a typical candidate material for biological synaptic devices and neuromorphic computing. Although, various structures of ferroelectric synaptic devices are developed, most of which are controlled by optical or electrical pulses, hardly any reports discuss the operation in self‐powered mode. In this work, an In2Se3/MoS2 ferroelectric lateral heterojunction‐type optoelectronic synaptic device is fabricated to mimic biological synaptic functions in voltage mode or self‐driven mode. In addition, two simplified image pre‐processing methods based on device arrays are shown to improve the recognition accuracy of a back‐end neural network. The proposed approach suggests a new route for designing and applying high‐performance synaptic devices.

In contrast to the commonly present UV light-stimulated synaptic oxide thin-film transistors, this study demonstrates a violet light (wavelength of 405 nm) stimulated zinc–tin oxide (ZTO) photoelectric transistor for potential application in optical neuromorphic computation. Owing to the light-induced oxygen vacancy ionization and persistent photoconductivity effect in ZTO, this device well imitates prominent synaptic functions, including photonic potentiation, electric depression, and short-term memory (STM) to long-term memory (LTM) transition. A highly linear and broad dynamic range of photonic potentiation can be achieved by modulating the light power density, while electric depression is realized by gate voltage pulsing. In addition, the brain-like re-learning experience with extended forgetting time (200 s) is well mimicked by the ZTO photoelectric transistor. As a result, the ZTO photoelectric transistor provides excessive synaptic function with multi-series of synaptic weight levels (90 levels for each given light power density), which makes it prevalent in the neuromorphic computation of massive data as well as in learning-driven artificial intelligence computation.

Detection and recognition of latent fingerprints play crucial roles in identification and security. However, the separation of sensor, memory, and processor in conventional ex-situ fingerprint recognition system seriously deteriorates the latency of decision-making and inevitably increases the overall computing power. In this work, a photoelectronic reservoir computing (RC) system, consisting of DUV photo-synapses and nonvolatile memristor array, is developed to detect and recognize the latent fingerprint with in-sensor and parallel in-memory computing. Through the Ga-rich design, we achieve amorphous GaOx (a-GaOx) photo-synapses with an enhanced persistent photoconductivity (PPC) effect. The PPC effect, which induces nonlinearly tunable conductivity, renders the a-GaOx photo-synapses an ideal deep ultraviolet (DUV) photoelectronic reservoir, thus mapping the complex input vector into a dimensionality-reduced output vector. Connecting the reservoirs and a memristor array, we further construct an in-sensor RC system for latent fingerprint identification. The system maintains over 90% recognition accuracy for latent fingerprint within 15% stochastic noise level via the proposed dual-feature strategy. This work provides a subversive prototype system of DUV in-sensor RC for highly efficient recognition of latent fingerprints.

Nonvolatile memory (NVM)‐based neuromorphic computing has been attracting considerable attention from academia and the industry. Although it is not completely successful yet, remarkable achievements have been reported pertaining to synaptic devices that can leverage NVM capable of storing multiple states. The analog synaptic devices performing computation similar to biological nerve systems are crucial in energy‐efficient analog neuromorphic computing systems. To use NVM as an analog synaptic device, researchers focus on improving device characteristics. Among various characteristics, the most challenging one is linearity and symmetry of synaptic weight update that is required for on‐chip training. In this regard, this review paper discusses recent synaptic device improvements focusing on novel schemes tailored for each NVM device to improve the linearity and symmetry. In addition to device‐level studies, recent research achievements are reviewed expanded up to chip‐level studies because in realizing neuromorphic hardware systems beyond a single synaptic device, several considerations and requirements are needed to confirm for high‐level design, and accordingly, cooptimize among synaptic devices, synapse arrays, electrical circuits, neural networks, algorithms, and implementation. Also, this review introduces various circuit and algorithmic approaches to compensate for the non‐ideality of the analog synaptic device.

Muon spin relaxation $(\ensuremath{\mu}\mathrm{SR})$ measurements have been made on ${\mathrm{Cd}}_{0.85}{\mathrm{Mn}}_{0.15}\mathrm{Te}:\mathrm{In}$ and ${\mathrm{Cd}}_{0.86}{\mathrm{Mn}}_{0.14}\mathrm{Te}:\mathrm{Al}$. The spin glass transition region has been studied using $\ensuremath{\mu}\mathrm{SR}$ and low-temperature magnetic susceptibility measurements. Persistent photoconductivity (PPC) has been shown to exist only in the In-doped sample, the effect of PPC on the spin glass transition temperature using bulk magnetic susceptibility has also been studied, indicating the presence of bound magnetic polarons. $\ensuremath{\mu}\mathrm{SR}$ on the Al-doped sample clearly shows the spin glass transition, however, the presence of the DX center, which causes PPC when doping with In donors, perturbs the muon response. Particular attention is paid to the possibility of the DX center trapping muonium, preventing the detection of the spin glass transition. PPC does not induce a change in the muon response, however, we find continuous illumination of the sample allows the observation of the spin glass transition. We suggest that these results can be explained by assuming the presence of multiple DX centers. The muon acting as a local magnetic probe of the DX center also suggests the ground state is diamagnetic.

Spectral albedos of open water, nilas, nilas with frost flowers, slush, and first-year ice with both thin and thick snow cover were measured in the East Antarctic sea-ice zone during the Sea Ice Physics and Ecosystems eXperiment II (SIPEX II) from September to November 2012, near 65°S, 120°E. Albedo was measured across the ultraviolet (UV), visible and near-infrared (nIR) wavelengths, augmenting a dataset from prior Antarctic expeditions with spectral coverage extended to longer wavelengths, and with measurement of slush and frost flowers, which had not been encountered on the prior expeditions. At visible and UV wavelengths, the albedo depends on the thickness of snow or ice; in the nIR the albedo is determined by the specific surface area. The growth of frost flowers causes the nilas albedo to increase by 0.2–0.3 in the UV and visible wavelengths. The spectral albedos are integrated over wavelength to obtain broadband albedos for wavelength bands commonly used in climate models. The albedo spectrum for deep snow on first-year sea ice shows no evidence of light-absorbing particulate impurities (LAI), such as black carbon (BC) or organics, which is consistent with the extremely small quantities of LAI found by filtering snow meltwater. Estimated BC mixing ratios were in the range 0.1–0.5 ng of carbon per gram of snow.

To overcome the coming big-data era, significant efforts are needed to implement neuromorphic computing that mimics the functional and structural features of the human brain using electronic devices. Repeated updates mimicking biological synaptic plasticity degrade the endurance of synaptic devices during in situ training. This study demonstrates a two-dimensional (2D) ferroelectric α-In2Se3-based synaptic device with enhanced durability via self-curing performed from current annealing (CA) in synaptic fatigue occurring during repetitive learning. The conduction, degradation, and self-curing mechanisms of the 2D ferroelectric-based synaptic device are quantitatively elucidated by low-frequency noise (LFN) spectroscopy. The classification accuracy of the Canadian Institute for Advanced Research (CIFAR)-10 dataset with self-cured conductance is superior to that of the device with synaptic fatigue and recovers to the initial accuracy level. The simulation results of removing defect cells through self-curing in the 2D ferroelectric synaptic array can help reduce energy consumption in the long term. The experimental results emphasize adopting 2D ferroelectric materials for future neuromorphic computing.

Zero-power optoelectronic synaptic devices

Optical artificial synapses possess several advantages, including high bandwidth, strong interference immunity, and ultra-fast signal transmission, overcoming the limitations of electrically stimulated synapses. Among various functional materials, 2D materials exhibit exceptional optical and electrical properties. By utilizing van der Waals heterostructures formed by these materials through rational design, synaptic devices can mimic the information perception ability of biological systems. This lays the foundation for low-energy artificial vision systems and neuromorphic computing. This study introduces an inhibitory artificial synapse based on photoelectric co-modulation of graphene/WSe2 van der Waals heterojunctions. By synergistically applying gate voltage and light pulses, we simulate memory and logic functions observed in the brain’s visual cortex. We achieve the construction of inhibitory synapses, enabling properties such as postsynaptic current response, short-term and long-term plasticity, and paired-pulse facilitation. Additionally, we accomplish the inverse recovery of device conductivity through separate gate voltage stimulation. Through bidirectional modulation of the artificial synaptic conductance, we construct an artificial hardware neural network that achieves 92.5% accuracy in recognizing handwritten digital images from the MNIST dataset. The network also has good recognition accuracy for handwritten digital images with different standard deviation Gaussian noise applied and other datasets. Furthermore, we successfully mimic the neural behavior of aversive learning for alcohol withdrawal in alcoholic patients using the device properties. The promising capabilities of artificial synapses constructed through electrical and optical synergistic modulation make them suitable for wearable electronics and artificial vision systems.

Bulk n-type ${\mathrm{Cd}}_{1\mathrm{\ensuremath{-}}\mathit{x}}$${\mathrm{Mn}}_{\mathit{x}}$Te crystals heavily doped with Ga are known to exhibit strong persistent photoconductivity (PPC) at low temperatures (below \ensuremath{\sim}100 K). In order to identify the origin of the PPC effect, we investigated the electronic properties of deep-level defects present in this material by deep-level transient spectroscopy (DLTS) and by thermally stimulated capacitance (TSCAP) measurements. We used a ${\mathrm{Cd}}_{1\mathrm{\ensuremath{-}}\mathit{x}}$${\mathrm{Mn}}_{\mathit{x}}$Te crystal with x=0.03 and a Ga doping level of 1\ifmmode\times\else\texttimes\fi{}${10}^{19}$ ${\mathrm{cm}}^{\mathrm{\ensuremath{-}}3}$. The DLTS measurements revealed the presence of two deep levels, with thermal activation energies for electron emission of 0.25 and 0.43 eV. The electron capture process to the 0.25-eV level was found to be thermally activated, with a thermal energy barrier of 0.11 eV. DLTS measurements performed after sub-band-gap illumination at low temperatures established that, in addition to the PPC effect, the illumination causes also a strong increase in the concentration of the 0.25-eV states, which persist to temperatures as high as T=200 K. This finding indicates that---in addition to the metastability of conduction electrons with respect to the 0.25-eV states, which results in the PPC below 100 K---the 0.25-eV states are also metastable with respect to other (ground) states from which they are created by photoexcitation. Our TSCAP measurements detected also two transitions to two different deep levels, in agreement with these results. The first of these levels is donorlike, with an activation energy of 0.21 eV. The second level is acceptorlike, and the electron transition to this level requires a thermal energy of several hundreds of meV. Because the concentrations of both the 0.25- and the 0.43-eV states found by DLTS appear to be very close to each other, we propose a single defect model with two charge states, attributing the 0.43-eV level to an acceptorlike two-electron ground state of the defect, and the 0.25-eV level to the excited, donorlike one-electron state. According to this model, the defect possesses a positive Hubbard correlation energy (positive-U system). We discuss a configuration coordinate diagram of the defect in the paper.

Brain‐inspired neuromorphic computing is recognized as a promising technology for implementing human intelligence in hardware. Neuromorphic devices, including artificial synapses and neurons, are regarded as essential components for the construction of neuromorphic hardware systems. Recently, optoelectronic neuromorphic devices are increasingly highlighted due to their potential applications in next‐generation artificial visual systems, attributed to their integrated sensing, computing, and memory capabilities. In this review, recent advancements in optoelectronic synapses and neurons are examined, with an emphasis on their structural characteristics, operational principles, and the replication of neuromorphic functions. For optoelectronic synaptic devices, such as memristor‐ and transistor‐based ones, attention is given to the two primary weight update modes: the light‐electricity synergistic mode and the all‐optical mode. Optoelectronic neurons are discussed in terms of different device types, including threshold switch neurons and semiconductor laser neurons. Last, the challenges that impede the progress of optoelectronic neuromorphic devices are identified, and potential future directions are suggested.