Abstract The optically stimulated synaptic device incorporates gold nanoparticles (Au‐NPs) embedded within an atomic‐layer‐deposited HfTiO x /Al 2 O 3 bilayer, which enables the photonic modulation of synaptic plasticity. The HfTiO x /Au‐NP interface enhances visible light absorption and utilization efficiency by generating ionized oxygen vacancies through the neutral oxygen vacancy sites, thereby modulating the device conductance states. This architecture effectively mimics key biological synaptic functions, including paired‐pulse facilitation (PPF), memory transitions from short‐term to long‐term memory (STM to LTM), and spike‐rate‐dependent plasticity (SRDP). The device further demonstrates optical logic operations and Pavlovian associative learning under dual‐wavelength light stimulation (405 and 450 nm). The intensity‐dependent generation of photocarriers and their nonlinear interaction with oxygen vacancies enable robust synaptic behavior, facilitating the emulation of human visual perception in a 4 × 4 optoelectronic synapse array with short‐term memory capabilities. Moreover, the wavelength‐ and sequence‐dependent synaptic responses can be finely controlled for the design of light‐programmable reservoir computing systems. These results demonstrate the potential of the HfTiO x /Au‐NP/Al 2 O 3 switching layer as a promising platform for efficient neuromorphic computing and vision‐based information processing using integrated optoelectronic synapses.

Abstract Data processing through artificial synapses is gaining attention owing to the emergence of neuromorphic computing. Analog processing via these synapses can simultaneously handle large volumes of data; however, it is susceptible to interference from various environmental factors. Specifically, temperature changes can significantly affect overall signal characteristics, leading to substantial errors. Herein, an organic heterojunction‐based artificial synapse is presented that is capable of light–temperature antagonistic operations. The properly aligned band structure and trap sites, which are facilitated by oxygen penetration, enable the implementation of controlled synaptic characteristics, depending on temperature and light conditions. An increase in temperature resulted in a thermally enhanced synaptic current, while light irradiation reduced the synaptic current, with the reduction degree being dependent on the light intensity. Finally, a biomimetic analog processor system capable of signal stabilization under drastic temperature changes is implemented. The artificial synapse, which operates using a light–temperature antagonistic operation, can significantly expand the potential applications of artificial intelligence hardware.

Abstract In the realm of neuromorphic computing, integrating Binary Neural Networks (BNN) with non‐volatile memory based on emerging materials can be a promising avenue for introducing novel functionalities. This study underscores the viability of lead‐free, air‐stable Cs 2 SnI 6 (CSI) based resistive random access memory (RRAM) devices as synaptic weights in neuromorphic architectures, specifically for BNNs applications. Herein, hydrothermally synthesized CSI perovskites are explored as a resistive layer in RRAM devices either on the rigid or flexible substrate, highlighting reproducible multibit switching with self‐compliance, low‐ resistance‐state (LRS) variations, a decent On/Off ratio(or retention) of ≈10 3 (or 10 4 s), and endurance exceeding 300 cycles. Moreover, a comprehensive evaluation with the 32 × 32 × 3 RGB CIFAR‐10 dataset reveals that binary convolutional neural networks (BCNN) trained solely on binary weight values can achieve competitive rates of accuracy comparable to those of their analog weight counterparts. These findings highlight the dominance of the LRS for CSI RRAM with self‐compliance in a weighted configuration and minimal influence of the high resistance state despite substantial fluctuations for flexible CSI RRAM under varying bending radii. With its unique electrical switching capabilities, the CSI RRAM is highly anticipated to emerge as a promising candidate for embedded AI systems, especially in IoT devices and wearables.

Abstract The imperfect interfaces between 2D transition metal dichalcogenides (TMDs) are suitable for boosting the hydrogen evolution reaction (HER) during water electrolysis. Here, the improved catalytic activity at the spatial heterojunction between 1T’ Re x Mo 1− x S 2 and 2H MoS 2 is reported. Atomic‐scale electron microscopy confirms that the heterojunction is constructed by an in‐situ two‐step growth process through chemical vapor deposition. Electrochemical microcell measurements demonstrate that the 1T’ Re x Mo 1− x S 2 –2H MoS 2 lateral heterojunction exhibits the best HER catalytic performance among all TMD catalysts with an overpotential of ≈84 mV at 10 mA cm −2 current density and 58 mV dec −1 Tafel slope. Kelvin probe force microscopy shows ≈40 meV as the work function difference between 2H MoS 2 and 1T’ Re x Mo 1− x S 2 , facilitating the electron transfer from 2H MoS 2 to 1T’ Re x Mo 1− x S 2 at the heterojunction. First‐principles calculations reveal that Mo‐rich heterojunctions with high structural stability are formed, and the HER performance is improved with the combination of increased density of states near the Fermi level and optimal Δ G H* as low as 0.07 eV. Those synergetic effects with many electrons and active sites with optimal Δ G H* improve HER performance at the heterojunction. These results provide new insights into understanding the role of the heterojunction for HER.

Abstract The optically stimulated synaptic device incorporates gold nanoparticles (Au‐NPs) embedded within an atomic‐layer‐deposited HfTiO x /Al 2 O 3 bilayer, which enables the photonic modulation of synaptic plasticity. The HfTiO x /Au‐NP interface enhances visible light absorption and utilization efficiency by generating ionized oxygen vacancies through the neutral oxygen vacancy sites, thereby modulating the device conductance states. This architecture effectively mimics key biological synaptic functions, including paired‐pulse facilitation (PPF), memory transitions from short‐term to long‐term memory (STM to LTM), and spike‐rate‐dependent plasticity (SRDP). The device further demonstrates optical logic operations and Pavlovian associative learning under dual‐wavelength light stimulation (405 and 450 nm). The intensity‐dependent generation of photocarriers and their nonlinear interaction with oxygen vacancies enable robust synaptic behavior, facilitating the emulation of human visual perception in a 4 × 4 optoelectronic synapse array with short‐term memory capabilities. Moreover, the wavelength‐ and sequence‐dependent synaptic responses can be finely controlled for the design of light‐programmable reservoir computing systems. These results demonstrate the potential of the HfTiO x /Au‐NP/Al 2 O 3 switching layer as a promising platform for efficient neuromorphic computing and vision‐based information processing using integrated optoelectronic synapses.

Abstract Data processing through artificial synapses is gaining attention owing to the emergence of neuromorphic computing. Analog processing via these synapses can simultaneously handle large volumes of data; however, it is susceptible to interference from various environmental factors. Specifically, temperature changes can significantly affect overall signal characteristics, leading to substantial errors. Herein, an organic heterojunction‐based artificial synapse is presented that is capable of light–temperature antagonistic operations. The properly aligned band structure and trap sites, which are facilitated by oxygen penetration, enable the implementation of controlled synaptic characteristics, depending on temperature and light conditions. An increase in temperature resulted in a thermally enhanced synaptic current, while light irradiation reduced the synaptic current, with the reduction degree being dependent on the light intensity. Finally, a biomimetic analog processor system capable of signal stabilization under drastic temperature changes is implemented. The artificial synapse, which operates using a light–temperature antagonistic operation, can significantly expand the potential applications of artificial intelligence hardware.

Abstract In the realm of neuromorphic computing, integrating Binary Neural Networks (BNN) with non‐volatile memory based on emerging materials can be a promising avenue for introducing novel functionalities. This study underscores the viability of lead‐free, air‐stable Cs 2 SnI 6 (CSI) based resistive random access memory (RRAM) devices as synaptic weights in neuromorphic architectures, specifically for BNNs applications. Herein, hydrothermally synthesized CSI perovskites are explored as a resistive layer in RRAM devices either on the rigid or flexible substrate, highlighting reproducible multibit switching with self‐compliance, low‐ resistance‐state (LRS) variations, a decent On/Off ratio(or retention) of ≈10 3 (or 10 4 s), and endurance exceeding 300 cycles. Moreover, a comprehensive evaluation with the 32 × 32 × 3 RGB CIFAR‐10 dataset reveals that binary convolutional neural networks (BCNN) trained solely on binary weight values can achieve competitive rates of accuracy comparable to those of their analog weight counterparts. These findings highlight the dominance of the LRS for CSI RRAM with self‐compliance in a weighted configuration and minimal influence of the high resistance state despite substantial fluctuations for flexible CSI RRAM under varying bending radii. With its unique electrical switching capabilities, the CSI RRAM is highly anticipated to emerge as a promising candidate for embedded AI systems, especially in IoT devices and wearables.

Abstract The imperfect interfaces between 2D transition metal dichalcogenides (TMDs) are suitable for boosting the hydrogen evolution reaction (HER) during water electrolysis. Here, the improved catalytic activity at the spatial heterojunction between 1T’ Re x Mo 1− x S 2 and 2H MoS 2 is reported. Atomic‐scale electron microscopy confirms that the heterojunction is constructed by an in‐situ two‐step growth process through chemical vapor deposition. Electrochemical microcell measurements demonstrate that the 1T’ Re x Mo 1− x S 2 –2H MoS 2 lateral heterojunction exhibits the best HER catalytic performance among all TMD catalysts with an overpotential of ≈84 mV at 10 mA cm −2 current density and 58 mV dec −1 Tafel slope. Kelvin probe force microscopy shows ≈40 meV as the work function difference between 2H MoS 2 and 1T’ Re x Mo 1− x S 2 , facilitating the electron transfer from 2H MoS 2 to 1T’ Re x Mo 1− x S 2 at the heterojunction. First‐principles calculations reveal that Mo‐rich heterojunctions with high structural stability are formed, and the HER performance is improved with the combination of increased density of states near the Fermi level and optimal Δ G H* as low as 0.07 eV. Those synergetic effects with many electrons and active sites with optimal Δ G H* improve HER performance at the heterojunction. These results provide new insights into understanding the role of the heterojunction for HER.

Abstract Various doped materials have been investigated to improve the structural stability of layered transition metal oxides for lithium‐ion batteries. Most doped materials are obtained through solid state methods, in which the doping of cations is not strictly site selective. This paper demonstrates, for the first time, an in situ electrochemical site‐selective doping process that selectively substitutes Li + at Li sites in Mn‐rich layered oxides with Mg 2+ . Mg 2+ cations are electrochemically intercalated into Li sites in delithiated Mn‐rich layered oxides, resulting in the formation of [Li 1− x Mg y ][Mn 1− z M z ]O 2 (M = Co and Ni). This Mg 2+ intercalation is irreversible, leading to the favorable doping of Mg 2+ at the Li sites. More interestingly, the amount of intercalated Mg 2+ dopants increases with the increasing amount of Mn in Li 1− x [Mn 1− z M z ]O 2 , which is attributed to the fact that the Mn‐to‐O electron transfer enhances the attractive interaction between Mg 2+ dopants and electronegative O δ − atoms. Moreover, Mg 2+ at the Li sites in layered oxides suppresses cation mixing during cycling, resulting in markedly improved capacity retention over 200 cycles. The first‐principle calculations further clarify the role of Mg 2+ in reduced cation mixing during cycling. The new concept of in situ electrochemical doping provides a new avenue for the development of various selectively doped materials.

Doping engineering has been actively investigated for two-dimensional (2D) transition metal dichalcogenides (TMDs) to enhance their electrical behavior, particularly for use in field-effect transistors (FETs). Here, we propose unprecedented redox-active n-type and p-type dopants, naphthalene and WCl₆, respectively, for large-scale monolayer MoS₂ films synthesized via low-pressure chemical vapor deposition using a Na₂S promoter. These molecular dopants were selected based on their high redox potentials versus the reference ferrocene, which facilitated the ionization of the dopants via charge transfer. Along with the suppression effect of sulfur vacancies in the monolayer, the electronic transport behavior exhibits an ultrahigh electron mobility of 331.7 cm² V^-1 s^-1 for the n-doped MoS₂ FET and an excellent hole mobility of 31.8 cm² V^-1 s^-1 with a high on/off ratio of ∼10⁷ for the p-type FET, all of which are record-setting values among those reported for large-scale monolayer MoS₂ and chemically doped TMD-based FETs. The modulation in the dopant concentration and its correlation with the transistor performance are mainly demonstrated, along with the adjusted band structures as the potential origin of the exceptional outcomes. The extended exploration of multiple FET devices within a single large-scale monolayer film demonstrated uniform electrical characteristics.

Question How can a multimodal large language model be adapted to interpret chest X-rays and generate radiologic reports? Findings The developed CXR-LLaVA model effectively detects major pathological findings in chest X-rays and generates radiologic reports with a higher accuracy compared to general-purpose models. Clinical relevance This study demonstrates the potential of multimodal large language models to support radiologists by autonomously generating chest X-ray reports, potentially reducing diagnostic workloads and improving radiologist efficiency.

Ferroelectric memories have undergone a transformative evolution from conventional perovskite-based materials to modern fluorite-structured ferroelectrics, driven by the pursuit of scalable, low-power, and CMOS-compatible non-volatile memory solutions. The observation of ferroelectricity in nanoscale HfO₂-based films has enabled integration with CMOS-compatible processes, providing advantages such as potential scalability, low power consumption, and non-volatility, while facilitating continued scaling and high-density integration. Leveraging established materials infrastructure in the semiconductor industry, hafnia-based ferroelectrics have been incorporated in various memory architectures, including ferroelectric random-access memory (FeRAM), ferroelectric tunnel junctions (FTJs), ferroelectric field-effect transistors (FeFETs), and ferroelectric memcapacitors (FeCAPs). Beyond conventional non-volatile storage, these devices have also emerged as promising building blocks for in-memory computing applications, including neuromorphic systems, hardware security primitives, and associative memory. In this review, we explore the historical development of ferroelectric memories from a materials-device co-design perspective, examine recent advances in device architectures and in-memory computing applications, and discuss the remaining challenges in endurance, retention, variability, and scaling. Finally, we propose future research directions that integrating material innovation, interface engineering, and circuit-level optimization to realize the full potential of ferroelectric memories in next-generation computing platforms.