Your Next Fridge Might Smell Better Than You Do

When we reach for something in the fridge with a questionable expiration date, the best option we have to check if it’s still good to eat is a quick sniff test. It’ll do the trick when something is far gone, but food can be spoiled long before our sense of smell will pick it up. We humans are good at a lot of things, but detecting the signature of trace chemicals with our noses isn’t one of them. That is why the sniff test steers so many people wrong, resulting in countless sicknesses from food-borne pathogens each year.

Researchers at the University of California, Berkeley think they may have a better solution. They have developed an electronic nose capable of identifying food spoilage and common food allergens by analyzing the gases those foods release into the air. The system combines an array of miniature gas sensors with machine learning software, allowing it to recognize scent patterns that would be impossible for a person to distinguish reliably.

The device, called ML-SCENT, is built around a 16-element sensor array. Each sensor is coated with a different gas-sensitive material, causing it to react differently when exposed to volatile organic compounds. Instead of trying to detect a single chemical, the system looks at the collective response of all 16 sensors and treats the resulting pattern as a unique fingerprint. A machine learning model then determines which food produced that signature.

To train the system, lead researcher Carla Bassil exposed the sensors to a variety of foods, including strawberries, blueberries, bananas, walnuts, hazelnuts, cashews, peanuts, chicken, milk, and eggs. The researchers also tested dairy, meat, and eggs at different stages of spoilage after being left at room temperature for up to 48 hours. The model learned to identify both the type of food and, in many cases, whether it was still fresh.

During testing, it successfully identified as few as 0.05 grams of walnut material, roughly one hundredth of an average shelled walnut. That level of sensitivity could eventually make the technology useful for people with severe food allergies.

The team reported an overall classification accuracy of 92.6% across 16 different food-related targets. Those targets included fresh and spoiled foods as well as several nut varieties, making this one of the first demonstrations of a gas sensor array being used to identify specific nut allergens through scent alone.

A key part of the design is the use of carbon nanotube field-effect transistors rather than traditional metal-oxide gas sensors. Because the sensors operate at room temperature, the researchers were able to incorporate a wider range of sensing materials and manufacture the array using a relatively simple single-step deposition process.

The researchers have already built a portable version of the electronic nose that can be controlled from an iPhone. Future testing will focus on determining how well the technology performs in more realistic environments, such as refrigerators filled with multiple foods. If those tests are successful, your next appliance may have a far better nose than you do.