Humidity-Gated Memristive Dynamics Enabling Near-Sensor Spiking Computation for Wind Direction and Noise-Resilient...

Area of Science:

• Materials Science
• Neuromorphic Engineering
• Sensor Technology

Background:

• Spiking in-sensor systems require discrete transitions for event-driven encoding.
• Gradual humidity signals lack the intrinsic thresholds needed for spike generation, posing a challenge for neuromorphic applications.
• Existing methods struggle to efficiently process slow, continuous environmental data.

Purpose of the Study:

• To develop a neuromorphic humidity-sensing platform capable of event-driven encoding.
• To address the mismatch between gradual humidity signals and the requirements of spiking neural networks.
• To demonstrate a hardware-level strategy for processing slow environmental dynamics.

Main Methods:

• Fabrication of a threshold-switching memristor with an asymmetric Ag/Nafion/ITO structure.
• Utilizing the Nafion layer as both a humidity transduction medium and an electrochemical matrix.
• Investigating humidity-dependent ionic migration barriers and volatile conductance changes.
• Characterizing the humidity-dependent switching threshold of the memristor.

Main Results:

• Achieved volatile conductance changes over six orders of magnitude with switching speeds down to 60 ns.
• Demonstrated a humidity-dependent switching threshold that decreases from 0.8 V to 0.2 V with increasing relative humidity (50% to 90%).
• Successfully implemented real-time classification of spatiotemporal humidity gradients for wind direction inference and noise-resilient speech recognition.

Conclusions:

• The developed platform enables event-driven encoding of slow environmental dynamics at the hardware level.
• The threshold-switching memristor provides an embedded gating mechanism for humidity-triggered spiking.
• This work offers a pathway toward efficient, low-power sensory systems for edge-intelligent applications.