Reading time ( words)
In nature, herbal fragrances attract insects, for example. But they are also used in industry, for example in the manufacture of perfumes and flavors. In order to differentiate between the fragrances of mint reliably, quickly and objectively, researchers at the Karlsruhe Institute of Technology (KIT) developed an electronic nose with an artificial sense of smell in an interdisciplinary collaboration: it can recognize different types of mint with great precision - making it suitable for applications from pharmaceutical quality control to the observation of mint oil as an environmentally friendly bioherbicide.
"So far, research has known an estimated 100,000 different biological connections via which neighboring plants interact with one another or control other organisms such as insects," says Professor Peter Nick from the Botanical Institute at KIT. “These compounds are very similar in plants of the same genus.” A classic example in the plant world is mint, in which the different varieties are endowed with very species-specific fragrances. In particular, the industrial monitoring of mint oil is subject to strict legal regulations to avoid counterfeiting, is time-consuming and requires a lot of skill, according to the scientist. A new electronic nose based on sensors with combined materials is supposed to support this, the researchers from the Botanical Institute,
Electronic nose based on a biological model
When developing the electronic nose, the entire research team orientates itself as closely as possible on the biological model: The olfactory cells that transmit information to the human brain via electrical impulses are replaced by a total of twelve special sensors (Quartz Crystal Microbalance, QCM sensors for short ). These consist of two electrodes with a quartz crystal. Such components are also built into cell phones, for example, as they guarantee high accuracy of cell phone frequencies in a cost-effective manner. “The scents of the mint are deposited on the surface of the sensors. This changes their resonance frequency and we get a reaction to the respective scent, ”explains Professor Christof Wöll from the IFG. Fragrances consist of organic molecules in different compositions. So that the new sensors can pick up these, the researchers from the IFG used twelve special sensor materials, including the metal-organic frameworks (MOFs) developed at the IFG. "These materials are highly porous and particularly suitable for sensor applications because they can absorb many molecules like a sponge," says Wöll. "By combining the sensors with the different materials, we virtually interconnect a neural network." because they can absorb many molecules like a sponge, ”says Wöll. "By combining the sensors with the different materials, we virtually interconnect a neural network." because they can absorb many molecules like a sponge, ”says Wöll. "By combining the sensors with the different materials, we virtually interconnect a neural network."
Six types of mint training through machine learning
The scientists tested the electronic nose with six different types of mint - including classic peppermint, horse mint and catnip. "We train the sensors using different machine learning methods so that they can create the fingerprint of the respective scent from the data collected and thus distinguish the scents from one another," explains Wöll. After each fragrance sample, the nose is rinsed with carbon dioxide (CO2) for about half an hour so that the sensors can regenerate.
The results of the interdisciplinary research team have shown that the electronic nose with QCM sensors can assign mint scents to a species with high specificity. In addition, it is a user-friendly, reliable and inexpensive alternative to conventional methods such as mass spectrometry, says the scientist. The focus for further development is on sensors that regenerate faster and can then pick up odors again. Furthermore, the researchers at the IFG are concentrating on MOF materials in order to design them for other areas of application, such as for artificial odor detection in medical diagnostics.