Brain-inspired neuromorphic computing is expected to overcome the von Neumann bottleneck by eliminating the memory wall between processing and memory units. Nevertheless, critical challenges persist in synaptic device implementation, particularly regarding nonlinear/asymmetric conductance modulation and multilevel conductance states, which substantially impede the realization of high-performance neuromorphic hardware. This study demonstrates a novel advancement in photonic–electronic modulated synaptic devices through the development of an amorphous indium–gallium–zinc oxide (a-IGZO) synaptic transistor. The device demonstrates biological synaptic functionalities, including excitatory/inhibitory post-synaptic currents (EPSCs/IPSCs) and spike-timing-dependent plasticity, while achieving excellent conductance modulation characteristics (nonlinearity of 0.0095/−0.0115 and asymmetric ratio of 0.247) and successfully implementing Pavlovian associative learning paradigms. Notably, systematic neural network simulations employing the experimental parameters reveal a 93.8% recognition accuracy on the MNIST handwritten digit dataset. The a-IGZO synaptic transistor with photonic–electronic co-modulation serves as a potential critical building block for constructing neuromorphic architectures with human-brain efficiency.

Brain-inspired neuromorphic computing has been considered as a promising novel paradigm and expected to replace traditional von Neumann architecture. Artificial synapses are recognized as the hardware foundation of neuromorphic computing, while the electrolyte-gated synapse has been proposed for its advantages in emulating biological synapses. Here, we propose a coplanar all-solid-state electrolyte-gated amorphous indium–gallium–zinc–oxide synapse that demonstrates multiple forms of both short- and long-term synaptic plasticity, including inhibitory and excitatory post-synaptic currents, paired-pulse facilitation, and relearning behavior. This artificial synapse not only efficiently executes Boolean operations but also switches between ‘AND’ and ‘OR’ logic types within a single device. Furthermore, it exhibits excellent performance in long-term potentiation and depression, with coefficients of nonlinearity of 0.046 and −0.073, achieving a notable accuracy of 93.2% in simulated handwritten digit recognition. This work emphasizes the significant potential of high-performance all-solid-state electrolyte-gated synapses in driving the development of next-generation neuromorphic computing systems.

The detection and monitoring of ions are essential for a broad range of applications, including industrial process control and biomedical diagnostics. Traditional ion-sensitive field-effect transistors require bulky and expensive reference electrodes, which face several limitations, including device miniaturization, high fabrication costs, and incompatibility with semiconductor manufacturing processes. Here, we introduce a reference-less semiconductor ion sensor (RELESIS) that utilizes anodic aluminum oxide film as both the sensitive and dielectric layer. The RELESIS is composed of a metal-oxide-semiconductor field-effect transistor and an interdigital electrode, which fundamentally eliminates the need for a reference electrode, thereby enabling device miniaturization. During fabrication, the anodic oxidation process is employed in place of the expensive atomic layer deposition method, significantly reducing manufacturing costs while maintaining high surface quality. In practical measurements, the RELESIS device demonstrated an excellent pH sensitivity of 57.8 mV/pH with a low hysteresis of 7 mV. As a proof-of-concept application, the RELESIS device was employed for real-time, non-destructive monitoring of milk freshness, accurately detecting pH changes from fresh to spoiled in milk samples. The combination of reference-less structure, low-cost fabrication, and superior sensing performance positions this technology as a promising platform for next-generation portable ion sensing systems in food safety, environmental monitoring, and point-of-care diagnostics.

Artificial synaptic devices are fundamental components for neuromorphic computing, yet achieving reliable, low-power, and linear analog weight modulation remains a critical challenge. Here, we present an a-IGZO (amorphous Indium-Gallium-Zinc Oxide) synaptic device that realizes ultra-linear voltage-based weight encoding through a shared floating-node architecture, where synaptic weights are directly represented by a continuously tunable node voltage, reducing the need for frequent refresh operations.The proposed device enables stable and low-power synaptic operation by decoupling weight update and readout through a shared floating-node architecture. By operating the read transistor in its linear regime, the synaptic weight exhibits an intrinsically linear and symmetric update behavior, achieving up to 120 discrete conductance states with minimal nonlinearity (ν p = 0.0058, ν d = -0.059). The device further demonstrates a high on/off ratio (~10⁷), ultralow offstate current (~100 fA), and consistent emulation of synaptic functions including excitatory/inhibitory postsynaptic currents and paired-pulse facilitation.The device further demonstrates two complementary neuromorphic functionalities: in-sensor visual perception with biomimetic photoadaptive behavior under optical stimulation, enabled by the intrinsic persistent photoconductivity of a-IGZO; and device-aware image classification on the CIFAR-10 dataset with accuracies exceeding 85%, closely approaching the ideal model performance.These results demonstrate that the proposed shared floating-node architecture enables ultralinear voltage-based weight encoding for precise and low-power analog synaptic modulation, establishing the device as a promising hardware platform for low-power neuromorphic computing and artificial vision systems.