Min-Kyu Song

Reservoir computing (RC) offers a promising route to real‐time temporal information processing and is widely regarded as an energy‐efficient alternative to conventional von Neumann architectures. Paradoxically, many hardware RC systems have yet to demonstrate clear advantages in both array‐level energy efficiency and processing speed, despite these being among the field's central motivations. Furthermore, their practical viability is limited by structural complexity and insufficient robustness. Here, we report a reconfigurable, dynamic random‐access memory (DRAM)‐like physical RC platform based on an oxide channel two‐transistor (2T) cell, providing a practical route toward low‐energy, fast, robust, and structurally simple physical RC. Within a single device footprint and fabrication flow, the same 2T cell can be reconfigured to volatile reservoir dynamics or quasi‐non‐volatile readout memory. Its operating mode is selected by the hold bias, enabling electrical reconfiguration with tunable relaxation dynamics. This compact strategy enables spatiotemporal programmability while preserving process simplicity, as validated by motion‐detection simulations. Owing to multi‐V _hold spatiotemporal encoding, the proposed RC architecture achieves a 3.98‐fold reduction in device‐array‐level read energy and requires 8.00‐fold fewer trainable synaptic weights than the ANN baseline. These results position oxide 2T memory as a strong candidate for next‐generation non‐von‐Neumann artificial intelligence (AI) hardware.

1. Ion intercalation is the insertion of mobile ions into a solid host, accompanied by structural, electronic, and chemical reorganisation of the material [1,2]. What makes this process physically ...

Ferroelectric capacitive memory (FeCAP) holds enormous potential for low-power, high-density in-memory computing. While hafnia-based FeCAPs have attracted attention for their silicon compatibility, they suffer from limited performance, such as a narrow memory window and relatively high switching fields. In this work, an FeCAP device is developed on the basis of a single-crystalline barium titanate (BTO) membrane, a perovskite oxide thin film that can be epitaxially lifted off and transferred onto a silicon platform. By engineering the device structure and epitaxy process, polarization asymmetry is introduced in capacitance-voltage characteristics. The resulting BTO-based FeCAP exhibits superior memory behavior, including a wide memory window of 308 picofarads and a low switching field of 0.005 megavolts per centimeter, outperforming conventional hafnia-based FeCAPs. Furthermore, these properties are preserved after active layer transfer onto a silicon platform. This approach provides a viable pathway for high-quality BTO to integrate into industry-compatible processes and to drive progress in future logic/memory applications.

The increasing demand for faster, energy-efficient, and higher bandwidth semiconductor devices has pushed conventional Si-based scaling to its fundamental limits, including mobility degradation, short-channel effects, and high power consumption. To overcome these challenges, three-dimensional integration has emerged as a promising strategy, but wafer-based approaches like through-Si-via face critical limitations in stacking density, mechanical stress, and fabrication complexity. Two-dimensional materials provide a compelling alternative due to their atomically thin structure, superior electrical and mechanical properties, and ability to sustain performance at the atomic scale. Moreover, their van der Waals integration enables heterogeneous, high-density, and efficient assembly of functional layers. This review summarizes recent advances in the preparation and van der Waals integration of 2D materials, including growth, transfer, and direct integration. Their applications in intelligent computing that range from logic to sensor devices and their potential as next-generation electronics are discussed.

The substantial power consumption of traditional computing architectures, arising from the physical separation of memory and processing units, has motivated the exploration of neuromorphic systems that emulate the remarkable energy efficiency of the human brain. However, artificial neural networks implemented on neuromorphic devices still require billions of weight updates, and conventional devices typically consume power in the milliwatt range per switching event, posing a significant challenge even for neuromorphic systems. In this study, a synapse‐like memristive device utilizing a tyrosine‐rich peptide is presented as the resistive switching layer. By emulating proton‐mediated signaling of biological synapses, the device leverages proton–electron dual‐carrier transport enabled by the redox‐active properties of tyrosine to realize ultralow‐power resistive switching. Proton modulation is implemented through two methods: i) exposure to external humidity and ii) electrically driven injection using a PdH _x proton reservoir layer. Comparative analysis reveals that the electrically driven approach achieves an ultralow switching power of 215 pW—≈2500 times lower than that of the intrinsic device—primarily owing to the sustained low off‐current during proton injection. These results demonstrate a promising strategy for developing highly energy‐efficient, synapse‐like memristive devices through precise control of proton dynamics.

Recent breakthroughs in ultrathin, single-crystalline, freestanding complex oxide systems have sparked industry interest in their potential for next-generation commercial devices^1,2. However, the mass production of these ultrathin complex oxide membranes has been hindered by the challenging requirement of inserting an artificial release layer between the epilayers and substrates^3,4. Here we introduce a technique that achieves atomic precision lift-off of ultrathin membranes without artificial release layers to facilitate the high-throughput production of scalable, ultrathin, freestanding perovskite systems. Leveraging both theoretical insights and empirical evidence, we have identified the pivotal role of lead in weakening the interface. This insight has led to the creation of a universal exfoliation strategy that enables the production of diverse ultrathin perovskite membranes less than 10 nm. Our pyroelectric membranes demonstrate a record-high pyroelectric coefficient of 1.76 × 10⁻² C m⁻² K⁻¹, attributed to their exceptionally low thickness and freestanding nature. Moreover, this method offers an approach to manufacturing cooling-free detectors that can cover the full far-infrared spectrum, marking a notable advancement in detector technology⁵.

Atomic layer deposition-grown beryllium oxide (BeO) is gaining attention as a dielectric material that can minimize device power consumption because of its high dielectric constant, high thermal conductivity, and low leakage current enabled by its wide bandgap energy. In this study, the impact of BeO dielectrics on InSnZnO (ITZO) thin-film transistors (TFTs) was investigated, revealing that adding a hafnium dioxide (HfO₂) layer can enhance electrical performance and bias stress reliability. Time-of-flight secondary-ion mass spectrometry and X-ray photoelectron spectroscopy confirmed that the single-BeO dielectric-based ITZO TFTs exhibited a low mobility of 27.6 cm²/V·s due to Be migration and demonstrated abnormal threshold voltage (V_TH) shifts under bias stress. Conversely, the HfO₂ 20 nm/BeO hetero-dielectric ITZO TFTs exhibited a high mobility of 76.6 cm²/V·s and enhanced abnormal V_TH shift characteristics. Therefore, these results demonstrate that our high-performance HfO₂/BeO hetero-dielectric-based ITZO TFTs could be utilized in back-end-of-line devices for monolithic three-dimensional memory technologies.

The demand for the three-dimensional (3D) integration of electronic components is steadily increasing. Despite substantial processing challenges, the through-silicon-via (TSV) technique emerges as the only viable method for integrating single-crystalline device components in a 3D format^1,2. Although monolithic 3D (M3D) integration schemes show promise³, the seamless connection of single-crystalline semiconductors without intervening wafers has yet to be demonstrated. This challenge arises from the inherent difficulty of growing single crystals on amorphous or polycrystalline surfaces after the back-end-of-the-line process at low temperatures to preserve the underlying circuitry. Consequently, a practical growth-based solution for M3D of single crystals remains unknown. Here we present a method for growing single-crystalline channel materials, specifically composed of transition metal dichalcogenides, on amorphous and polycrystalline surfaces at temperatures low enough to preserve the underlying electronic components. Building on this developed technique, we demonstrate the seamless monolithic integration of vertical single-crystalline logic transistor arrays. This accomplishment leads to the development of unprecedented vertical complementary metal oxide semiconductor (CMOS) arrays composed of grown single-crystalline channels. Ultimately, this achievement provides opportunities for M3D integration of various electronic hardware in the form of single crystals.

Artificial intelligence has become an essential part of the daily lives and has revolutionized various sectors, including healthcare, finance, transportation, and entertainment. With a substantial increase in processed data, neuromorphic devices that replicate the operation of the human brain have been emphasized owing to their superior efficiency. Typical neuromorphic devices focus on constructing synapse‐like structures. However, biological synapses have more complex mechanisms for efficient data processing. One of the most prominent mechanisms is proton activation, which forms an ion concentration gradient prior to the transmission of neurotransmitters and plays a key role in efficient computation. In this study, proton‐mediated signaling at biological synapses is successfully replicated by fabricating a proton‐modulated memristor device using a tyrosine‐rich peptide film. The ionic input of the memristor is controlled by applying a voltage to proton‐permeable PdH_x contacts in a hydrogen atmosphere, thus successfully adjusting the resistive switching behavior. Remarkable improvements in resistive switching and computing performance are observed through proton injection, analogous to “proton‐mediated signaling” at the actual synapse. It is believed that this study proposes a new paradigm for designing biorealistic devices and provides inspiration for precisely controllable ion‐based neuromorphic devices.

Remote epitaxy is taking center stage in creating freestanding complex oxide thin films with high crystallinity that could serve as an ideal building block for stacking artificial heterostructures with distinctive functionalities. However, there exist technical challenges, particularly in the remote epitaxy of perovskite oxides associated with their harsh growth environments, making the graphene interlayer difficult to survive. Transferred graphene, typically used for creating a remote epitaxy template, poses limitations in ensuring the yield of perovskite films, especially when pulsed laser deposition (PLD) growth is carried out, since graphene degradation can be easily observed. Here, we employ spectroscopic ellipsometry to determine the critical factors that damage the integrity of graphene during PLD by tracking the change in optical properties of graphene in situ. To mitigate the issues observed in the PLD process, we propose an alternative growth strategy based on molecular beam epitaxy to produce single-crystalline perovskite membranes.