Most of us rely on a quick sniff test to decide whether slightly expired milk or a week-old takeaway is still safe to eat. While the human nose is often reliable, it isn’t foolproof. Every year, millions of people in the U.S. become ill from foodborne pathogens that grow in spoiled or improperly cooked food.
Fortunately for our stomachs, researchers at UC Berkeley have developed a new “electronic nose” that can identify odors linked to spoiled food far more precisely than humans can. It can also detect common food allergens, such as peanuts and walnuts, which can be dangerous for people with sensitivities. A recent study published in Science Advances describes the technology.
Smart Fridges That Warn You Before Food Goes Bad
Study lead author Carla Bassil, a UC Berkeley Ph.D. student in electrical engineering and computer sciences, said that “smart” refrigerators linked to smartphones could use the system. Such fridges could alert users when food is about to spoil, for example by warning that broccoli needs to be eaten soon or that chicken is nearing its final safe day.
The artificial nose itself consists of an array of 16 miniature gas sensors, each tuned to respond to different combinations of airborne chemical compounds.
“You can think of it as a set of digital taste buds, where each sensor on the chip reacts differently to various gas molecules,” Bassil explained during a UC Grad Slam presentation of her work. “Each of the 16 sensors has a distinct sensing film, and it operates by turning chemical interactions between the sensor surface and gas molecules into electrical signals.”
Using machine learning, Bassil developed a model that could identify the sensor patterns linked to seven foods: strawberry, blueberry, banana, walnut, hazelnut, cashew, and peanut. She trained it to distinguish between the smells of raw chicken, milk, and eggs when fresh and after they had been left at room temperature for 24 and 48 hours.
Highly Sensitive Detection, Still Unproven in Real-World Food Mixtures
Bassil found that the device was sensitive enough to detect as little as 0.05 grams of isolated walnut—roughly one-hundredth of a typical shelled walnut. However, she has not yet tested how well it performs in more complex environments, such as walnuts mixed into foods like salads or cakes, or spoiled items stored alongside other groceries in a refrigerator.
“The idea is that we combine the selective responses of gas sensors with machine learning’s pattern-recognition abilities to identify the unique gas fingerprint of each food,” Bassil explained. “The result is a sensor chip that is much more sensitive and far more objective than a human nose.”
Although the concept of an electronic nose has existed since the 1980s, turning it into a practical device has proven challenging. While single gas sensors—like those used in household carbon monoxide detectors—are relatively easy to produce, building an array of different sensing films onto one chip is significantly more complex.
Carbon Nanotubes Enable Ultra-Sensitive, Room-Temperature Sensing
Bassil addressed many of these difficulties by using carbon nanotubes as the conductive material instead of metal oxides. These nanotubes can form layers only a few nanometers thick—just a handful of atoms, or about one-hundredth the width of a human hair. Their extremely large surface area gives them unique properties, including high sensitivity even at room temperature.
By designing a device that operates at room temperature rather than requiring heating, Bassil was able to work with a broader range of gas-sensitive materials, including ones that would break down under high heat, such as polymers. This also allowed her to build the sensing chip using a simpler method called drop casting instead of more complex fabrication techniques.
“The most scalable aspect of my electronic nose is that we can combine many different sensing materials and deposit them all in a single step,” Bassil said.
Although the new study did not include it, Bassil has since developed a portable version of the electronic nose and controls it through an iPhone app. She now plans to evaluate the next iteration of the device across more diverse environments while further enhancing its sensitivity and reliability.
Read the original article on: Tech Xplore
Read more: Robots can mix drinks and run marathons but still struggle to multitask