Taking inspiration from the human brain to create a new class of intelligent, light-powered devices

At INL, our researchers are pushing the boundaries of brain-inspired computing, developing neuromorphic systems that process sensory data the way nature does – fast, efficiently, and with minimal energy.

In a recently published paper in Scientific Reports, INL researchers developed a tiny micropillar quantum resonant tunnelling diode, or RTD, that behaves like a sensory neuron. This ‘neuron’ is capable of detecting light, processing information, and converting it into electrical signals, all within a single nanoscale device. Researcher Bruno Romeira explains, “This is possible because we are using quantum phenomena.”

At its core, this system is a III–V semiconductor structure (materials widely used in photonics and high-speed electronics) designed to respond directly to incoming near-infrared light. When the light intensity reaches a certain threshold, the device enters a state known as negative differential resistance, triggering large-amplitude voltage oscillations. In other words, the incoming light signal is transformed into rhythmic electrical bursts, just like the firing patterns seen in biological neurons.

Traditional neuromorphic hardware often relies on complex circuits, combining separate memory components and oscillators to mimic the way biological neurons process information. This increases the size, power consumption, and complexity of the system. The new device developed by Bejoys Jacob and his colleagues brings these capabilities together into a single, integrated component.

This ability to encode optical information into electrical oscillations makes the device much more than a simple light sensor. It behaves like a sensory neuron, capable of not only detecting light but also amplifying the signal and pre-processing it – a crucial step towards creating in-sensor intelligent edge systems that process data directly where it’s collected, without needing large external processors.

What makes this even more exciting is that the device’s behaviour mimics processes found in living organisms. Insects like dragonflies rely on rhythmic bursts of neuronal activity to track moving prey. In mammals, oscillatory bursts are essential for processing sensory data and coordinating complex brain activities. By reproducing these natural burst firing patterns in hardware, INL researchers are opening the door to bioinspired artificial vision systems, all using miniaturised and energy-efficient technology.

With its compact design and compatibility with existing III–V semiconductor platforms, this neuromorphic photonic neuron is well-suited for integration into future optical sensors and systems. Whether in autonomous vehicles, next-generation LiDAR, or ultra-fast visual processing for robotics, this research developed under the European-funded project InsectNeuroNano and FCT-funded project META-LED brings us closer to hardware that not only senses the world but also interprets it, much like natural systems do.

Text by Catarina Moura
Photography by Catarina Moura & Gina Palha