Artificial intelligence (AI) edge devices prefer employing high-capacity nonvolatile compute-in-memory (CIM) to achieve high energy efficiency and rapid wakeup-to-response with sufficient accuracy. Most previous works are based on either memristor-based CIMs, which suffer from accuracy loss and do not support training as a result of limited endurance, or digital static random-access memory (SRAM)–based CIMs, which suffer from large area requirements and volatile storage. We report an AI edge processor that uses a memristor-SRAM CIM-fusion scheme to simultaneously exploit the high accuracy of the digital SRAM CIM and the high energy-efficiency and storage density of the resistive random-access memory memristor CIM. This also enables adaptive local training to accommodate personalized characterization and user environment. The fusion processor achieved high CIM capacity, short wakeup-to-response latency (392 microseconds), high peak energy efficiency (77.64 teraoperations per second per watt), and robust accuracy (<0.5% accuracy loss). This work demonstrates that memristor technology has moved beyond in-lab development stages and now has manufacturability for AI edge processors.

One of the promising nonvolatile memories of the next generation is resistive random-access memory (ReRAM). It has vast benefits in comparison to other emerging nonvolatile memories. Among different materials, dielectric films have been extensively studied by the scientific research community as a nonvolatile switching material over several decades and have reported many advantages and downsides. However, less attention has been given to low-dimensional materials for resistive memory compared to dielectric films. Particularly, β-Ga₂O₃ is one of the promising materials for high-power electronics and exhibits the resistive switching phenomenon. However, low-dimensional β-Ga₂O₃ nanowires have not been explored in resistive memory applications, which hinders further developments. In this article, we studied the resistance switching phenomenon using controlled electron flow in the 1D nanowires and proposed possible resistive switching and electron conduction mechanisms. High-density β-Ga₂O₃ 1D-nanowires on Si (100) substrates were produced via the VLS growth technique using Au nanoparticles as a catalyst. Structural characteristics were analyzed via SEM, TEM, and XRD. Besides, EDS, CL, and XPS binding feature analyses confirmed the composition of individual elements, the possible intermediate absorption sites in the bandgap, and the bonding characteristics, along with the presence of various oxygen species, which is crucial for the ReRAM performances. The forming-free bipolar resistance switching of a single β-Ga₂O₃ nanowire ReRAM device and performance are discussed in detail. The switching mechanism based on the formation and annihilation of conductive filaments through the oxygen vacancies is proposed, and the possible electron conduction mechanisms in HRS and LRS states are discussed.

The non-enzymatic glucose sensing response of pure and Ag-decorated vertically aligned ZnO nanorods grown on Si substrates was investigated. The simple low-temperature hydrothermal method was employed to synthesize the ZnO NRs on the Si substrates, and then Ag decoration was achieved by sputtering. The crystal structure and surface morphologies were characterized by X-ray diffraction, field-emission scanning electron microscopy (FESEM), and transmission electron microscopy (TEM). The Ag incorporation on the ZnO NR surfaces was confirmed using EDS mapping and spectra. Furthermore, the chemical states, the variation in oxygen vacancies, and the surface modifications of Ag@ZnO were investigated by XPS analysis. Both the glucose/ZnO/Si and glucose/Ag@ZnO/Si device structures were investigated for their non-enzymatic glucose sensing performances with different glucose concentrations. Based on EIS measurements and amperometric analysis, the Ag@ZnO-NR-based glucose sensor device exhibited a better sensing ability with excellent stability over time than pure ZnO NRs. The Ag@ZnO NR glucose sensor device recorded 2792 µA/(mM·cm²) sensitivity with a lowest detection limit of 1.29 µM.

In this work, we have inspected the theoretical resistive switching properties of two ReRAM models based on heterojunction structures of Cu/SiO _x nanoparticles (NPs)/Si and Si/SiO _x NPs/Si, in which dielectric layers of the silica nanoparticles present dislocations at bicrystal interfaces. To validate the theoretical model, a charge storage device with the structure Cu/SiO _x /Si was fabricated and its ReRAM properties were studied. Our examinations on the electrical, thermal and structural aspects of resistive switching uncovered the switching behavior relies upon the material properties and electrical characteristics of the switching layers, as well as the metal electrodes and the interfacial structure of grains within the dielectric materials. We also determined that the application of an external electric field at Grain Boundaries (GB) is crucial to resistive switching behavior. Moreover, we have demonstrated that the switching behavior is influenced by variations in the atomic structure and electronic properties, at the atomic length scale and picosecond timescale. Our findings furnish a useful reference for the future development and optimization of materials used in this technology.