We report the first resonance characterization of β-Ga 2 O 3 -based doubly-clamped-beam resonators elongated along [100] and fabricated directly on a (001) β-Ga 2 O 3 substrate by crystallographic gas etching. The beams exhibited clear fundamental-mode resonances with quality factors ( Q ) on the order of 10 4 . Analysis of the beam-length dependence of the resonant frequency ( f 0 ) yielded an effective Young’s modulus of 143.2 (+16.4/−14.0) GPa within the Euler–Bernoulli beam approximation, while axial-stress analysis indicated only a small residual tensile stress. The f 0 Q values exceeded the highest reported values for transferred-flake-based doubly clamped beams, highlighting the potential advantage of crystallographic etching for fabricating β-Ga 2 O 3 microelectromechanical systems.
We report a record-high product of the current-gain cutoff frequency and gate length ( f T L g ) in an InAs quantum-well (QW) high-electron-mobility transistor (HEMT). The fabricated device with a physical gate length ( L g ) of 240 nm exhibits a peak transconductance of 2.53 mS μm −1 and achieves f T and f max of 265 GHz and 470 GHz, respectively, yielding a record-high f T L g product of 63.6 GHz⋅μm. To elucidate the physical origin of this exceptional f T L g , a delay-time analysis is performed over devices with different gate lengths. The extracted carrier transit time ( τ T ) is used to determine the average carrier velocity under the gate as v avg = L g / τ T . The resulting very high v avg provides a direct transport-based interpretation of the record-high f T L g , highlighting the superior electron transport capability of the InAs QW channel. These results highlight the potential of InAs QW HEMTs for future ultrahigh-speed transistor technologies.
We report a record-high product of the current-gain cutoff frequency and gate length ( f _T L _g ) in an InAs quantum-well (QW) high-electron-mobility transistor (HEMT). The fabricated device with a physical gate length ( L _g ) of 240 nm exhibits a peak transconductance of 2.53 mS μm ^−1 and achieves f _T and f _max of 265 GHz and 470 GHz, respectively, yielding a record-high f _T L _g product of 63.6 GHz⋅μm. To elucidate the physical origin of this exceptional f _T L _g , a delay-time analysis is performed over devices with different gate lengths. The extracted carrier transit time ( τ _T ) is used to determine the average carrier velocity under the gate as v _avg = L _g / τ _T . The resulting very high v _avg provides a direct transport-based interpretation of the record-high f _T L _g , highlighting the superior electron transport capability of the InAs QW channel. These results highlight the potential of InAs QW HEMTs for future ultrahigh-speed transistor technologies.
Superconducting beta-Bi 2Pd has attracted considerable interest as a candidate topological superconductor and a platform for investigating unconventional superconductivity. However, previously reported growth procedures commonly employ elevated substrate temperatures or post-growth annealing, which can restrict integration with temperature-sensitive materials and pre-patterned structures. Here, we demonstrate room-temperature magnetron co-sputtering of phase-pure, (001)-textured beta-Bi 2Pd films on amorphous and crystalline substrates. Epitaxial films are further realized by depositing beta-Bi 2Pd at room temperature on a thin epitaxial seed layer formed at elevated temperature. These results establish a low-thermal-budget route for integrating superconducting beta-Bi 2Pd into superconducting heterostructures and hybrid quantum devices.
Advanced scanning probe microscopy (SPM) measurements have traditionally required highly specialized skills, but the integration of artificial intelligence (AI) and machine learning has begun to transform how SPM is operated. This review surveys AI methods applied to SPM, organized by technique rather than chronology: unsupervised learning for feature extraction, support vector machines for real-time classification, convolutional neural networks for image analysis, Bayesian optimization for experiment planning, reinforcement learning for atomic and molecular manipulation, and object detection for feature localization. The emerging use of large language models as natural-language interfaces for microscope operation is discussed as a distinct topic. For each method, the operating principles and representative SPM applications are presented. This method-centric organization enables direct comparison of AI approaches applied to similar SPM tasks and clarifies the complementary roles—perception, decision-making, control, and orchestration—that each technique plays in integrated autonomous SPM systems.
