When Hyuk-Jun Kwon thinks about smell, he sees a puzzle that humans solve effortlessly — and machines have always struggled with. Now, his team in South Korea may have cracked it.
Scientists at the Daegu Gyeongbuk Institute of Science and Technology have built an electronic nose that can detect and tell apart tens of thousands of different odors. Their secret ingredient? Tiny porous materials called MOFs — short for metal-organic frameworks. These sponge-like structures, made by combining metal ions with organic compounds, trap odor molecules in their microscopic pores, much like a real nose captures scent particles from the air.
The research, published in the journal Progress in Materials Science, draws its biggest insight from human biology. The human nose doesn't have a separate receptor for every possible smell. Instead, it uses roughly 400 types of receptors that work together: each smell triggers a unique combination of signals, like a barcode. Kwon and his team applied the same principle to their sensors, pairing arrays of MOF materials with AI that learns to read those combinatorial codes.
"MOFs provide a virtually unlimited library of materials that can be designed to exhibit different responses to different odors, much like human olfactory receptors," Kwon said.
The team sorted MOF-based electronic nose technologies into three types. Plain MOFs form the base, with pores that can be shaped for specific tasks. MOF-composites and MOF-derivatives add extra layers of sensitivity, stability, and selectivity. When combined with machine-learning and deep-learning programs, these materials can sort through complex smell signals with impressive accuracy.
The applications stretch far beyond the lab. The researchers see MOF-based electronic noses one day helping doctors diagnose diseases by sniffing out chemical markers in a patient's breath. They could monitor air quality in cities, check food for freshness, detect hazardous gases in factories, support smart farms, and even give self-driving cars and robots a sense of smell.
What makes this approach stand out is that MOF sensors can work at room temperature using very little power — a practical advantage over older materials that required heat or bulky equipment. The team hopes their work acts as a roadmap, connecting materials science and AI research so others can build specialized electronic noses for specific needs.
For Kwon, the goal isn't just detecting smells — it's bridging two fields that rarely talk to each other. "This paper is significant in that it bridges the gap between materials research and AI-based odor recognition research," he said, "while presenting a roadmap for the development of intelligent electronic noses tailored to specific applications."