Nonvolatile memories using two-dimensional materials and high-k oxides have gained attention for their potential to achieve robust analog switching, easy memristive device integration, and low-energy consumption. In this study, we fabricated Pt/TiN/HfO 2 /WS 2 /Pt memristive devices. To implement these devices, a WS 2 film was thermally evaporated under high vacuum conditions followed by HfO 2 growth using atomic layer deposition at 400 °C. Detailed analysis using high-resolution transmission electron microscopy and X-ray photoelectron spectroscopy revealed diffusion of W and S atoms within the HfO 2 layer and extraction of oxygen by W atoms, thus resulting in a multilayer structure (HfWO y S x, W x –1 O y S x, and W 1– x O y S x ) with varying ratios of oxygen, tungsten, and sulfur atoms ( x and y ). The fabricated devices demonstrated consistent and stable analogue switching over numerous cycles, with exceptional endurance (2000 cycles) and retention (10 3 s). They exhibited high cycle-to-cycle consistency, as evidenced by the low-coefficient of variation (3.5% and 4.0% for the set and reset voltages, respectively). By modulating the reset stop voltage, we achieved five-level resistance states, thus making these devices capable of being used in artificial synapses. Furthermore, we observed analog switching with gradual resistance changes under different current compliance conditions by incrementally adjusting the reset–stop voltage. The memristor-based artificial synapses exhibited fundamental synaptic functions, such as long-term potentiation, long-term depression, paired-pulse facilitation, paired-pulse depression, and spike-timing-dependent plasticity for long-term and short-term plasticity. Moreover, we employed a three-layer artificial neural network for image recognition, achieving 94% accuracy using identical pulse amplitudes. These findings highlight the potential of HfO 2 /WS 2 bilayer films, enable controllable analogue switching, and simulate synaptic functions. They hold promise for future data storage memory and neuromorphic computing systems.

Abstract Memristors, owing to their uncomplicated structure and resemblance to biological synapses, are predicted to see increased usage in the domain of artificial intelligence. Additionally, to augment the capacity for multilayer data storage in high-density memory applications, meticulous regulation of quantized conduction with an extremely low transition energy is required. In this work, an a-HfSiO x -based memristor was grown through atomic layer deposition (ALD) and investigated for its electrical and biological properties for use in multilevel switching memory and neuromorphic computing systems. The crystal structure and chemical distribution of the HfSiOx/TaN layers were analyzed using X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS), respectively. The Pt/a-HfSiO x /TaN memristor was confirmed by transmission electron microscopy (TEM) and showed analog bipolar switching behavior with high endurance stability (1000 cycles), long data retention performance (10 4 s), and uniform voltage distribution. Its multilevel capability was demonstrated by restricting current compliance (CC) and stopping the reset voltage. The memristor exhibited synaptic properties, such as short-term plasticity, excitatory postsynaptic current (EPSC), spiking-rate-dependent plasticity (SRDP), post-tetanic potentiation (PTP), and paired-pulse facilitation (PPF). Furthermore, it demonstrated 94.6% pattern accuracy in neural network simulations. Thus, a-HfSiO x -based memristors have great potential for use in multilevel memory and neuromorphic computing systems. Graphical Abstract

Nonvolatile memories using two-dimensional materials and high-k oxides have gained attention for their potential to achieve robust analog switching, easy memristive device integration, and low-energy consumption. In this study, we fabricated Pt/TiN/HfO 2 /WS 2 /Pt memristive devices. To implement these devices, a WS 2 film was thermally evaporated under high vacuum conditions followed by HfO 2 growth using atomic layer deposition at 400 °C. Detailed analysis using high-resolution transmission electron microscopy and X-ray photoelectron spectroscopy revealed diffusion of W and S atoms within the HfO 2 layer and extraction of oxygen by W atoms, thus resulting in a multilayer structure (HfWO y S x, W x –1 O y S x, and W 1– x O y S x ) with varying ratios of oxygen, tungsten, and sulfur atoms ( x and y ). The fabricated devices demonstrated consistent and stable analogue switching over numerous cycles, with exceptional endurance (2000 cycles) and retention (10 3 s). They exhibited high cycle-to-cycle consistency, as evidenced by the low-coefficient of variation (3.5% and 4.0% for the set and reset voltages, respectively). By modulating the reset stop voltage, we achieved five-level resistance states, thus making these devices capable of being used in artificial synapses. Furthermore, we observed analog switching with gradual resistance changes under different current compliance conditions by incrementally adjusting the reset–stop voltage. The memristor-based artificial synapses exhibited fundamental synaptic functions, such as long-term potentiation, long-term depression, paired-pulse facilitation, paired-pulse depression, and spike-timing-dependent plasticity for long-term and short-term plasticity. Moreover, we employed a three-layer artificial neural network for image recognition, achieving 94% accuracy using identical pulse amplitudes. These findings highlight the potential of HfO 2 /WS 2 bilayer films, enable controllable analogue switching, and simulate synaptic functions. They hold promise for future data storage memory and neuromorphic computing systems.

Abstract Memristors, owing to their uncomplicated structure and resemblance to biological synapses, are predicted to see increased usage in the domain of artificial intelligence. Additionally, to augment the capacity for multilayer data storage in high-density memory applications, meticulous regulation of quantized conduction with an extremely low transition energy is required. In this work, an a-HfSiO x -based memristor was grown through atomic layer deposition (ALD) and investigated for its electrical and biological properties for use in multilevel switching memory and neuromorphic computing systems. The crystal structure and chemical distribution of the HfSiOx/TaN layers were analyzed using X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS), respectively. The Pt/a-HfSiO x /TaN memristor was confirmed by transmission electron microscopy (TEM) and showed analog bipolar switching behavior with high endurance stability (1000 cycles), long data retention performance (10 4 s), and uniform voltage distribution. Its multilevel capability was demonstrated by restricting current compliance (CC) and stopping the reset voltage. The memristor exhibited synaptic properties, such as short-term plasticity, excitatory postsynaptic current (EPSC), spiking-rate-dependent plasticity (SRDP), post-tetanic potentiation (PTP), and paired-pulse facilitation (PPF). Furthermore, it demonstrated 94.6% pattern accuracy in neural network simulations. Thus, a-HfSiO x -based memristors have great potential for use in multilevel memory and neuromorphic computing systems. Graphical Abstract

Here, we have identified the volatile and non-volatile traits of Pt/TaOx/TiN device through the various electrical experiments by controlling the compliance current (CC) and determined the availability of this device as a memristor. First of all, the configuration analysis is performed to ensure the structure and thickness of the device. We monitored the volatile resistive switching, which can be facilitated as short-term memory (STM) at a low current level. But in the higher CC, the forming process at the negative bias needs to be accompanied and shows the bipolar resistive switching (BRS), which can be used as non-volatile memory (NVM) within the neuromorphic system. Also, we investigated the conduction mechanism and proved that Space-Charge-Limited-Current (SCLC) dominates the high resistive state in the reset mechanism. The distinct difference between volatile and non-volatile characteristics that appear depending on the CC is shown in the retention test. Finally, we conducted the pattern recognition of the Modified National Institute of Standards and Technology (MNIST) and obtained two pattern recognition accuracy using the potentiation/depression data measured from different BRS types.

ABSTRACT This study proposes a strategy to simultaneously improve conductance uniformity and data retention characteristics by introducing the incremental step pulse with verify algorithm (ISPVA) technique and hydrogen (H 2 ) annealing into a non‐filamentary TiN/Ti/HfO 2 /TiO x /TiN resistive switching memory device. The high Schottky barrier formed at the Ti/HfO 2 interface induces asymmetric electron injection and limits reverse current flow, resulting in a rectifying ratio of approximately 1442. This self‐rectifying characteristic provides an intrinsic advantage in suppressing sneak currents in crossbar arrays. The ISPVA technique improves the linearity and uniformity of conductance modulation, enabling the implementation of up to 6‐bit multilevel states within a few‐µA current range. In addition, H 2 annealing stabilized conduction by forming hydrogen bonds with oxygen vacancies in the oxide layer and suppressing oxygen ion–vacancy recombination. As a result, data retention over 10 4 s and endurance exceeding 10 4 cycles were achieved even under a low energy consumption of 36.3 pJ. Furthermore, the experimentally obtained long‐term potentiation and depression characteristics were implemented in a Transformer‐based keyword spotting (KWS) model, achieving a recognition accuracy of 92.5%. These results suggest that the proposed device enables controlled analog conductance modulation with improved stability, showing its potential for Transformer‐based neuromorphic computing applications.

This article investigates the impact of band-to-band tunneling (BTBT) occurring in the charge trap layer (CTL) of vertical NAND (V−NAND) flash memory under excessive erasure conditions and aggressive program operation. Strong local electric fields induce BTBT within the CTL, generating electron–hole pairs that cause trap charge loss, inter-cell interference, and early threshold voltage (V${}_{\mathrm{th}}$) reduction. The generated holes accelerate the loss of trapped electrons near the programmed target cell, degrading incremental step pulse programming (ISPP) characteristics, whereas laterally redistributed electrons induce Vth shifts in adjacent cells. In addition, the time-dependent accumulation of BTBT-generated carriers contributes to initial threshold voltage shift (IVS) during retention. These results show that BTBT in the CTL provides reliability issues in V−NAND flash memory, such as short-term Vth instability and inter-cell interference.

This work proposes a distance-dependent trap model to overcome the limitations of conventional Technology Computer-Aided Design (TCAD) models, which fail to accurately represent the non-uniform charge distribution resulting from electron energy relaxation under strong tunneling conditions in 3D NAND flash memories. By incorporating a spatially varying capture cross-section within the charge trap layer (CTL), the model effectively captures trap centroid shifts under realistic programming conditions. Implemented through the physical model interface (PMI) in Sentaurus TCAD, this model shows significantly improved alignment with experimental data for incremental step pulse programming (ISPP) characteristics when compared to the conventional Shockley-Read-Hall (SRH) model. Additionally, a systematic analysis of trap distribution effects on programming speed and Z-interference is conducted. The results indicate a clear trade-off: traps located near the tunneling oxide (TOX) enhance programming speed but increase Z-interference, whereas traps near the blocking oxide (BOX) reduce Z-interference, but with slower programming speeds. These findings demonstrate the importance of precise control over trap distribution to optimize the performance and reliability of 3D NAND Flash memories.