Stable Synaptic Transistor-Based Humidity Sensor with Broad Detection Window

Human visual recognition is profoundly affected by ambient relative humidity, yet current bionic and neuromorphic systems lack the ability to adapt to environmental variability, resulting in mismatches between human and robotic perception. In this work, a stable humidity-sensitive synaptic transistor featuring a broad detection window is designed and fabricated to bridge the gap between human and robotic sensory capabilities. The proposed humidity sensory neuron integrates a humidity sensing unit with a synaptic transistor in a separation device structure, enabling independent optimization of sensing and neuromorphic functions. This ensures excellent operational stability with negligible transfer characteristics degradation over 90 days. More importantly, the device exhibits robust humidity-dependent synaptic behaviors, including tunable excitatory postsynaptic currents, paired-pulse facilitation index, pulse-number dependent plasticity and high-pass filter coefficient under various relative humidity. Additionally, an artificial neural network is further constructed, which can accurately simulate human visual recognition performance under varying humidity conditions, highlighting its potential for applications in next-generation neuromorphic robotics, advanced sensing platforms, and cyborg technologies.