Search

A self-rectifying ferroelectric tunnel junction that employs a HfO2/ZrO2/HfO2superlattice (HZH SL) combined with Al2O3and TiO2layers is proposed. The 6 nm-thick HZH SL effectively suppresses the formation of non-ferroelectric phases while increasing remnant polarization (Pr). This enlarged Prmodulates the energy barrier configuration, consequently achieving a large on/off ratio of 1273 by altering the conduction mechanism from off-state thermal injection to on-state Fowler–Nordheim tunneling. Moreover, the asymmetric Schottky barriers at the top TiN/TiO2and bottom HfO2/Pt interfaces enable a self-rectifying property with a rectifying ratio of 1550. Through calculations and simulations it is found that the device demonstrates potential for achieving an integrated array size exceeding 7k while maintaining a 10% read margin, and shows potential for application in artificial synapses for neuromorphic computing with an image recognition accuracy above 92%. Finally, the self-rectifying behavior and device-to-device variation reliability are confirmed in a 9 × 9 crossbar array structure.

Inspired by the functions of biological neural networks, volatile memristors are essential for implementing neuromorphic computing. These devices enable large-scale and energy-efficient data processing by emulating neural functionalities through dynamic resistance changes. The threshold switching characteristics of volatile memristors, which are driven by various mechanisms in materials ranging from oxides to chalcogenides, make them versatile and suitable for neuromorphic computing systems. Understanding these mechanisms and selecting appropriate devices for specific applications are crucial for optimizing the performance. However, the existing literature lacks a comprehensive review of switching mechanisms, their compatibility with different applications, and a deeper exploration of the spatiotemporal processing capabilities and inherent stochasticity of volatile memristors. This review begins with a detailed analysis of the operational principles and material characteristics of volatile memristors. Their diverse applications are then explored, emphasizing their role in crossbar arrays, artificial receptors, and neurons. Furthermore, the potential of volatile memristors in artificial inference systems and reservoir computing is discussed, due to their spatiotemporal processing capabilities. Hardware security applications and probabilistic computing are also examined, where the inherent stochasticity of the devices can improve the system robustness and adaptability. To conclude, the suitability of different switching mechanisms for various applications is evaluated, and future perspectives for the development and implementation of volatile memristors are presented. This review aims to fill the gaps in existing research and highlight the potential of volatile memristors to drive innovation in neuromorphic computing, paving the way for more efficient and powerful computational paradigms.

Vertical two-terminal synaptic devices based on resistive switching have shown great potential for emulating biological signal processing and implementing artificial intelligence learning circuitries. To mimic heterosynaptic behaviors in vertical two-terminal synaptic devices, an additional terminal is required for neuromodulator activity. However, adding an extra terminal, such as a gate of the field-effect transistor, may lead to low scalability. In this study, a vertical two-terminal Pt/bilayer Sr1.8Ag0.2Nb3O10(SANO) nanosheet/Nb:SrTiO3(Nb:STO) device emulates heterosynaptic plasticity by controlling the number of trap sites in the SANO nanosheet via modulation of the tunneling current. Similar to biological neuromodulation, we modulated the synaptic plasticity, pulsed pair facilitation, and cutoff frequency of a simple two-terminal device. Therefore, our synaptic device can add high-level learning such as associative learning to a neuromorphic system with a simple cross-bar array structure.

Active cation-based diffusive memristors featuring essentially volatile threshold switching have been proposed for novel applications, such as a selector in a one-selector-and-one-resistor structure and signal generators in neuromorphic computing. However, the high variability of the switching behavior, which results from the high electroforming voltage, external environmental conditions, and transition to the non-volatile switching mode in a high-current range, is considered a major impediment to such applications. Herein, for the first time, we developed a highly reliable threshold switching device immune to atmospheric changes based on an ultraviolet-ozone (UVO)-treated diffusive memristor consisting of Ag and SiO2nanorods (NRs). UVO treatment forms a stable water reservoir on the surface of SiO2NRs, facilitating the redox reaction and ion migration of Ag. Consequently, diffusive memristors possess reliable switching characteristics, including electroforming-free, repeatable, and consistent switching with resistance to changes in ambient conditions and compliance levels during operation. We demonstrated that our approach is suitable for various metal oxides and can be used in numerous applications.

The switching characteristics and performance of oxide-based memristors are predominately determined by oxygen- or oxygen-vacancy-mediated redox reactions and the consequent formation of conducting filaments (CFs). Devices using oxide thin films as the switching layer usually require an electroforming process for subsequent switching operations, which induces large device-to-device variations. In addition, the hard-to-control redox reaction during repeated switching causes random fluctuations or degradation of each resistance state, hindering reliable switching operations. In this study, an HfO2nanorod (NR)-based memristor is proposed for simultaneously achieving highly uniform, electroforming-free, fast, and reliable analogue switching properties. The well-controlled redox reaction due to the easy gas exchange with the environment at the surface of the NRs enhances the generation of oxygen or oxygen vacancies during the switching operation, resulting in electroforming-free and reliable switching behavior. In addition, the one-dimensional surface growth of CFs facilitates highly linear conductance modulation with smaller conductance changes compared with the two-dimensional volume growth in thin-film-based memristors, resulting in a high accuracy of >92% in the Modified National Institute of Standards and Technology pattern-recognition test and desirable spike-timing-dependent plasticity.

Devices that simulate sensory systems go beyond simple stimulus processing. They comprehensively interpret daily responses, indicating an expansion into new domains. Artificial sensory systems based on memristive devices simulate perception, cognition and interaction by emulating human sensory systems. This enables robots and AI to efficiently exhibit more human‐like responses and behaviors in real‐world situations.

In the biological nervous system, the integration and cooperation of parallel system of receptors, neurons, and synapses allow efficient detection and processing of intricate and disordered external information. Such systems acquire and process environmental data in real‐time, efficiently handling complex tasks with minimal energy consumption. Memristors can mimic typical biological receptors, neurons, and synapses by implementing key features of neuronal signal‐processing functions such as selective adaption in receptors, leaky integrate‐and‐fire in neurons, and synaptic plasticity in synapses. External stimuli are sensitively detected and filtered by “artificial receptors,” encoded into spike signals via “artificial neurons,” and integrated and stored through “artificial synapses.” The high operational speed, low power consumption, and superior scalability of memristive devices make their integration with high‐performance sensors a promising approach for creating integrated artificial sensory systems. These integrated systems can extract useful data from a large volume of raw data, facilitating real‐time detection and processing of environmental information. This review explores the recent advances in memristor‐based artificial sensory systems. The authors begin with the requirements of artificial sensory elements and then present an in‐depth review of such elements demonstrated by memristive devices. Finally, the major challenges and opportunities in the development of memristor‐based artificial sensory systems are discussed. The memristor‐based artificial sensory systems composed of integrated and cooperative parallel networks of artificial sensory receptors, neurons, and synapses are reviewed. Specific mechanisms and critical functions required by the memristive device are reviewed. The reported studies that functionally implement artificial sensory systems with memristive devices are introduced, discussing the limitations and future research directions in developing memristor‐based artificial sensory systems.

Abstract

The growth characteristics of Ta2O5thin films by atomic layer deposition (ALD) were examined using Ta(NtBu)(NEt2)3(TBTDET) and Ta(NtBu)(NEt2)2Cp (TBDETCp) as Ta-precursors, where tBu, Et, and Cp represent tert-butyl, ethyl, and cyclopentadienyl groups, respectively, along with water vapor as oxygen source. The grown Ta2O5films were amorphous with very smooth surface morphology for both the Ta-precursors. The saturated ALD growth rates of Ta2O5films were 0.77 Å cycle–1at 250 °C and 0.67 Å cycle–1at 300 °C using TBTDET and TBDETCp precursors, respectively. The thermal decomposition of the amido ligand (NEt2) limited the ALD process temperature below 275 °C for TBTDET precursor. However, the ALD temperature window could be extended up to 325 °C due to a strong Ta–Cp bond for the TBDETCp precursor. Because of the improved thermal stability of TBDETCp precursor, excellent nonuniformity of ∼2% in 200 mm wafer could be achieved with a step coverage of ∼90% in a deep hole structure (aspect ratio 5:1) which is promising for 3-dimensional architecture to form high density memories. Nonetheless, a rather high concentration (∼7 at. %) of carbon impurities was incorporated into the Ta2O5film using TBDETCp, which was possibly due to readsorption of dissociated ligands as small organic molecules in the growth of Ta2O5film by ALD. Despite the presence of high carbon concentration which might be an origin of large leakage current under electric fields, the Ta2O5film using TBDETCp showed a promising resistive switching performance with an endurance cycle as high as ∼17 500 for resistance switching random access memory application. The optical refractive index of the deposited Ta2O5films was 2.1–2.2 at 632.8 nm using both the Ta-precursors, and indirect optical band gap was estimated to be ∼4.1 eV for both the cases.

To replace or succeed the present NAND flash memory, resistive switching random access memory (ReRAM) should be implemented in the vertical-type crossbar array configuration. The ReRAM cell must have a highly reproducible resistive switching (RS) performance and an electroforming-free, self-rectifying, low-power-consumption, multilevel-switching, and easy fabrication process with a deep sub-μm2cell area. In this work, a Pt/Ta2O5/HfO2–x/TiN RS memory cell fabricated in the form of a vertical-type structure was presented as a feasible contender to meet the above requirements. While the fundamental RS characteristics of this material based on the electron trapping/detrapping mechanisms have been reported elsewhere, the influence of the cell scaling size to 0.34 μm2on the RS performance by adopting the vertical integration scheme was carefully examined in this work. The smaller cell area provided much better switching uniformity while all the other benefits of this specific material system were preserved. Using the overstressing technique, the nature of RS through the localized conducting path was further examined, which elucidated the fundamental difference between the present material system and the general ionic-motion-related bipolar RS mechanism.

The electrical performances of Pt/TiO2/Ti/Pt stacked Schottky-type diode (SD) was systematically examined, and this performance is dependent on the chemical structures of the each layer and their interfaces. The Ti layers containing a tolerable amount of oxygen showed metallic electrical conduction characteristics, which was confirmed by sheet resistance measurement with elevating the temperature, transmission line measurement (TLM), and Auger electron spectroscopy (AES) analysis. However, the chemical structure of SD stack and resulting electrical properties were crucially affected by the dissolved oxygen concentration in the Ti layers. The lower oxidation potential of the Ti layer with initially higher oxygen concentration suppressed the oxygen deficiency of the overlying TiO2layer induced by consumption of the oxygen from TiO2layer. This structure results in the lower reverse current of SDs without significant degradation of forward-state current. Conductive atomic force microscopy (CAFM) analysis showed the current conduction through the local conduction paths in the presented SDs, which guarantees a sufficient forward-current density as a selection device for highly integrated crossbar array resistive memory.

Rutile structured Al-doped TiO2(ATO) and TiO2films were grown on bimetal electrodes (thin Ru/thick TiN, Pt, and Ir) for high-performance capacitors. The work function of the top Ru layer decreased on TiN and increased on Pt and Ir when it was thinner than ∼2 nm, suggesting that the lower metal within the electrodes influences the work function of the very thin Ru layer. The use of the lower electrode with a high work function for bottom electrode eventually improves the leakage current properties of the capacitor at a very thin Ru top layer (≤2 nm) because of the increased Schottky barrier height at the interface between the dielectric and the bottom electrode. The thin Ru layer was necessary to achieve the rutile structured ATO and TiO2dielectric films.

Filamentary resistance switching (RS) is one of the more obvious and useful phenomena in the family of RS mechanisms. In filamentary RS, the long reset switching time and substantially large power consumption are the critical obstacles for microelectronic applications. In this study, an innovative solution to overcome this reset problem is suggested by stacking n‐type TiO2and p‐type NiO films. Interestingly, in this stacked structure, the region where filament rupture and rejuvenation occurs could be arbitrarily controlled to be at any location between the interface with the metal electrode and the TiO2/NiO interface by using an appropriate switching sequence. This collective motion behavior of conducting filaments can be practically used to reduce reset switching time from ∼100 μs to ∼150 ns, with an extremely high off/on resistance ratio of ∼106.