Adaptive Tunnel-Barrier Synapse for Bi-Mode Nociception in Neuromorphic Systems

There has been a growing interest in neuromorphic devices—electronic systems designed to mimic how the human nervous system works. As artificial intelligence becomes increasingly integrated into robotics, sensory platforms, and human–machine interfaces, researchers are no longer just focused on replicating cognitive functions like memory and learning. They’re also looking to emulate more fundamental biological responses, including something as critical—and complex—as pain perception. Developing an artificial nociceptor, or a device that can simulate the sensation of harmful stimuli, has become a particularly intriguing goal in this space. Pain, after all, plays a crucial role in protecting living organisms. It acts as an intelligent alarm system, activating only when a certain threshold is crossed to prevent further injury. The challenge, however, lies in how to translate this kind of adaptive, threshold-based response into a reliable and efficient piece of hardware.

One of the main difficulties stems from the need for a device that not only reacts to varying levels of stimulation but also does so in a way that mirrors the biological specificity of nociceptors. In living systems, these sensory neurons remain dormant under harmless conditions and only fire when a stimulus becomes strong enough to be considered potentially damaging. Creating that same level of selectivity in electronics demands materials and device architectures that can dynamically adjust their behavior based on input intensity—something that most current neuromorphic systems struggle with. Technologies like memristors and electrolyte-gated transistors have made some headway, but they often fall short when it comes to flexibility, power efficiency, or simplicity of design. Many still require multiple interconnected components to simulate what a single biological neuron does naturally.

To address these limitations, New research paper published in Photonics Research and led by Associate Professor Chengdong Yang, Yilong Liu, Dr. Linlin Su, Xinwei Li, Lihua Xu, and Qimei Cheng from the Jiangsu Province Engineering Research Center of Integrated Circuit Reliability Technology and Testing System at Wuxi University,  introduces a novel design built around a tunnel silicon nitride (Si₃N₄) layer integrated into a back-to-back Schottky junction (B-B SJ) structure. This configuration isn’t just compact and energy-efficient—it also enables a dynamic switching mechanism between two distinct modes of operation. In effect, the device can mimic both non-painful (non-nociceptive) and painful (nociceptive) responses depending on the strength and frequency of the stimulus it receives. That adaptability marks a significant step forward in designing artificial systems that respond more like living ones.

To bring their idea of an artificial nociceptor to life, the research team built a device that, on the surface, looks quite simple—but its behavior tells a more complex story. The device uses a layered structure, where a high-mobility organic semiconductor (C8-BTBT) is laid over a stack of silicon nitride (Si₃N₄) and silicon dioxide (SiO₂). This combination forms what’s called a back-to-back Schottky junction, which turned out to be key to achieving the kind of dynamic, responsive behavior the team was aiming for. They carefully grew bilayer C8-BTBT crystals using a floating coffee ring technique—a method that allowed them to control the thickness and alignment with high precision. They confirmed the quality of these crystals using AFM and polarized optical microscopy, which showed smooth surfaces and well-ordered molecular stacking. With the device ready, the researchers tested how it responded to light—treating light as a stand-in for sensory stimuli. Under weak light pulses, the device produced small, short-lived electrical signals similar to excitatory postsynaptic currents (EPSCs), like those found in biological neurons during non-painful interactions. But things changed dramatically when they increased the light intensity or repeated the pulses in quick succession. The device suddenly jumped to a much stronger electrical response—mimicking the abrupt way biological nociceptors fire when a harmful threshold is crossed. The authors also applied pairs of pulses with different time gaps and found when the pulses were close together, the second response was amplified—just like in biological systems that show paired-pulse facilitation. As the delay between pulses increased, the amplification effect faded, following a decay pattern that closely resembles how real neurons behave. Even more impressively, the device showed memory-like behavior. Repeated stimulation caused the response to linger longer, mimicking how the brain reinforces synaptic connections with experience. To explain what was going on inside the device, the team made simplified capacitor versions and measured how current flowed under increasing voltage. At first, tunneling current was low, but once a certain field was reached, it jumped—revealing a clear switch between low and high conduction states. This matched the “threshold” effect seen in their synaptic tests. They also found that by tweaking the thickness of the Si₃N₄ layer, they could control how easily the device flipped into its high-response mode—essentially tuning its pain sensitivity. Even better, the device worked reliably in air and used very little energy, pointing to real potential for use in future wearable electronics or intelligent robotics.

In conclusion, the study of Wuxi University scientists stands out for achieving something that’s long been difficult in both neuroscience and electronics: capturing the essence of pain perception in a compact, low-power device. Pain isn’t just another sensory response—it’s a biologically vital, highly selective function that activates only when the body senses real danger. What’s remarkable about the work by Chengdong Yang and colleagues is that their device doesn’t simply register input; it responds in a threshold-dependent, adaptive way that mirrors how real nociceptors behave. The system doesn’t fire with every small signal. Instead, it “decides” when a stimulus is significant enough to trigger a stronger, alarm-like response. This brings a new level of intelligence to artificial synapses, making them not only reactive but also context-aware. The broader implications are particularly exciting. In robotics and prosthetics, sensory feedback is critical—but up to now, most systems have lacked the subtlety of real human sensation. With a built-in nociceptive threshold, future machines could respond more appropriately to potential harm. Imagine a robotic hand that can recognize when it’s gripping something too tightly and ease off automatically, or a prosthetic limb that can alert the user to excessive pressure or temperature, all without external commands. This type of embedded sensory intelligence could significantly improve both safety and user experience. From a bioelectronic standpoint, this technology gets us one step closer to bridging the gap between artificial systems and the human body. Our nervous system doesn’t treat all stimuli equally—pain is reserved for high-priority events. Mimicking that kind of tiered response in hardware could lead to more natural and intuitive neural interfaces, such as advanced e-skin or implants that operate within the nervous system’s own language of graded, thresholded signaling. Equally compelling is the practicality of the design. The use of tunnel silicon nitride, a standard CMOS-compatible material, means the device can be manufactured using existing semiconductor infrastructure. There’s no need for exotic materials or expensive fabrication processes. Combined with the device’s stability in ambient conditions and its extremely low power demands, this positions the technology not just as an academic breakthrough, but as a realistic option for scalable, real-world applications.