Zhongming Zeng

Strong near-field analog sensing requires magnetic sensors to accommodate large field excursions without saturation while retaining sufficient resolution for small low-frequency field variations. In tunneling magnetoresistance (TMR) sensors, extending the linear range through increased effective magnetic anisotropy reduces the field sensitivity and consequently penalizes the noise-equivalent field resolution. Here, we investigate this range–sensitivity–resolution coupling in perpendicular synthetic antiferromagnetic (p-SAF) TMR sensors. The CoFeB free-layer thickness is systematically varied from 1.6 to 2.2 nm to tune the effective anisotropy and establish the range–sensitivity relationship, yielding a hysteresis-free linear response over the experimentally accessible range of at least ±1000 Oe. We subsequently vary the number of series-connected magnetic tunnel junctions under otherwise comparable magnetic and electrical conditions to determine whether array scaling can recover the low-frequency detectivity penalty associated with wide-range operation. Increasing N from 200 to 2000 improves detectivity DB from 2.27 to 1.04μT/Hz at 10 Hz, while all arrays retain a verified linear-range lower bound of ±1000 Oe. The optimized array is finally evaluated in a magnetic-switch prototype as a representative strong near-field analog sensing platform, where continuous non-saturating response is maintained under a strongly varying magnetic field. Building on low-current sensor operation, a proposed ADC-free time-over-threshold readout architecture is further explored as a prospective system-level strategy for reducing data-conversion overhead. These results identify free-layer anisotropy and array size as complementary design variables for jointly controlling wide-range response and low-frequency field resolution, while highlighting their potential for low-power strong near-field analog sensing systems.

Orbital current in light metals has attracted considerable attention as an alternative tool to switch the magnetization of ferromagnets. However, the manipulation of orbital torque efficiency, which is a key aspect of novel spin-orbitronic devices, remains a challenge. Here, the reversible modulation of orbital torque at room temperature is demonstrated in perpendicularly magnetized Ta/Pt/Co/Cu–CuOx/MoOx Hall bar devices. By applying a negative electric field, the orbital torque efficiency of a device with a Cu–CuOx thickness of 3 nm can be significantly enhanced by a factor of ∼4.64. This outstanding result is attributed to the applied electric fields driving the migration of oxygen ions, which controls the distance between the Cu/CuOx interface and the Co layer, as well as the formation or reduction of an antiferromagnetic CoOx layer at the Co surface. Consequently, the injection efficiency of the orbital current is controlled. In addition, the critical magnetization switching current density is reduced by ∼36% by applying an electric field. These findings provide a pathway for implementing electrically driven magnetization switching in potential applications while also opening new avenues for the development of spin-orbitronic devices.

With the rapid deployment of next-generation communication technologies, the demand for highly sensitive and miniaturized microwave detectors has become extremely urgent. Conventional semiconductor diodes suffer from intrinsic drawbacks when detecting submicrowatt signals. Spintronics provides an optimal alternative via the spin rectification effect, which enables widely tunable, low-power microwave rectification at room temperature. This Review examines recent advances in spintronic microwave rectifiers and their future prospects. We discuss diverse physical platforms, including devices with uniform magnetization and special nonuniform topological spin textures. Furthermore, we highlight multiorder-of-magnitude performance improvements in such devices, focusing on strategies to boost sensitivity, reduce noise equivalent power (NEP), and extend broadband as well as multiband detection capacity. Apart from core performance metrics, we discuss emerging applications: ambient radio frequency (RF) energy harvesting, high-resolution vector microwave-field imaging, monolithically integrated spintronic microwave systems, and microwave neural networks that can directly process high-frequency signals for neuromorphic computing. By sorting out the complete technical development route of this field, this Review aims to provide a systematic theoretical framework and technical references to promote the development of spintronic devices for next-generation wireless communication and edge artificial intelligence.

Layered hexagonal boron nitride (h-BN) enables the release and transfer of III-nitride epilayers toward flexible electronics, but its insufficient interfacial adhesion to the substrate would limit its compatibility with conventional fabrication processes. Here, based on a sapphire substrate coated with an AlN layer that has the same hexagonal structure as h-BN, the bonding strength of the h-BN/AlN/sapphire structure is significantly enhanced to withstand standard fabrication processes. Meanwhile, epitaxial h-BN on physical vapor deposition-AlN/sapphire sustains its layered structure with releasable properties for device peeling-off and transferring. As such, the h-BN/AlN/sapphire template enables the subsequent growth, fabrication, and wafer-scale transfer of GaN photodetectors (PDs) to flexible substrates. The obtained flexible PDs feature responsivity over 20 A/W with reliable photo-to-dark current ratios under a weak incident light intensity of 23 nW/cm2, even with large bending deformation. This work provides an effective strategy to achieve process compatibility and transferability of III-nitride heterostructures toward large-scale high-performance GaN-based flexible electronics.

Conventional vision systems for autonomous driving are hindered by energy‐inefficient and latency‐prone architectures due to physically segregated sensing, memory, and processing modules. Here, we report a bio‐inspired vision sensor that emulates the spectral adaptation mechanism of the Pacific salmon retina leveraging a van der Waals heterojunction of NbNiTe ₅ and black phosphorus (BP). Operating at zero bias, the device exhibits intrinsic wavelength‐dependent antagonistic photoresponses: negative photoconductance under 365 nm ultraviolet illumination (mimicking visual suppression in bright environments) and positive photoconductance under 820 nm near‐infrared light (emulating visual enhancement in dim conditions). This built‐in adaptability enables robust environmental perception across extreme illumination scenarios, from high‐glare daylight to low‐light nights. Furthermore, we integrated this sensor into a full functional system that unifies image perception, non‐volatile storage, and in‐sensor processing. When deployed for traffic scenario analysis, a convolutional neural network trained on features extracted by the sensor achieved 96% classification accuracy, with performance scaling proportionally to the system's noise suppression capability. Our work establishes a practical strategy for high‐contrast bidirectional photonic synapses and highlights the potential of biomimetic systems in neuromorphic vision — particularly for autonomous driving and intelligent surveillance, where reliable operation across diverse spectral environments is critical.

Stiff thin films deposited on compliant substrates inevitably suffer from thermal mismatch strain, which severely degrades the performance of strain-sensitive functional materials. A key challenge in thermal stress engineering is to quantify dissipated thermal strain and mitigate thermal stress via interlayer design. Here, we systematically investigate the thermal strain relaxation mechanism governed by structural stiffness (thickness) and intrinsic material properties (Young's modulus) by tuning the thicknesses of the HfO2 interlayer and polyimide (PI). Using a geometry-confined configuration that converts multiaxial strain into a quasi-uniaxial state, we generate one-dimensional wrinkles for quantitative strain analysis. The results show that out-of-plane buckling acts as an effective channel for thermal strain relaxation. Notably, residual strain retention is strongly modulated by the synergy between interlayer thickness and modulus. These findings provide practical mechanical guidelines for flexible heterogeneous integration, highlighting the necessity to balance interfacial rigidity with thermal strain dissipation through tailored material selection and geometric optimization.

Spin-torque diodes based on magnetic tunnel junctions have emerged as a promising paradigm for microwave detection and neuromorphic computing due to their intrinsic nonlinear electrical characteristics. Traditional artificial intelligence image processing technology based on the von Neumann architecture faces the bottlenecks of power consumption and storage wall. Therefore, we propose and demonstrate hardware-inspired artificial neurons featuring a rectified linear unit activation function by leveraging injection-locked spin-torque diodes. Taking full advantage of the above properties, we construct a supervised autoencoder architecture for image denoising. The proposed architecture is validated on the MNIST dataset with additive Gaussian noise. Experimental results show that its denoising performance is on par with traditional software-implemented neural networks, featuring similar convergence performance and stable high-quality image reconstruction. Notably, even under heavy noise intensity of 0.3, the system consistently maintains a peak signal-to-noise ratio above 20 dB and a structural similarity index exceeding 0.85 across all scenarios. This study provides a viable hardware-native technical pathway toward edge artificial intelligence applications.

We demonstrate a visible‐blind near‐ultraviolet (near‐UV) photodetector based on a hydrogen‐plasma‐treated two‐terminal p‐GaN‐gated hybrid‐anode diode. In the dark, the p‐GaN optical gate depletes the AlGaN/GaN two‐dimensional electron gas (2DEG), yielding a dark current of 6.82 × 10 ⁻⁹ mA/mm (normalized to gate width). Under 365‐nm illumination, 2DEG recovery increases the photocurrent to 12.99 mA/mm and yields a 365‐nm responsivity of 7.7 × 10 ³ A/W. Enabled by an etch‐free gate‐definition HPT process that mitigates trap‐assisted effects, the device exhibits rise and fall times of 22.96 and 167.87 μs, respectively. These results indicate that the HPT‐enabled HAD combines low dark current, high responsivity, and microsecond‐scale response for GaN‐based near‐UV photodetection.