Forget expiration dates. Spoilage sensors could tell us when food actually goes bad

“Use by.” “Best by.” “Best if used by.” “Enjoy by.” Food companies in the US began dating their products with this patchwork of labels during the 1970s—often with little, if any, scientific basis or uniformity. In the decades prior, packaged products had become increasingly popular as fewer and fewer people grew their own food.

Naturally, concern arose that inconsistent date labels would make it difficult to tell if food was still fresh—and whether consumers were being swindled. In 1973, the US Congress considered a bill that would have required food manufacturers to list a date that their perishable foods should be sold by. It would also have kept retail distributors from selling food after that date had passed, punishable by a $5,000 fine and 1 year in prison.

But that bill failed to make it into law. So did the next one. And the next. All told, the 1970s saw no fewer than 10 food-dating bills without significant federal legislation passed. Meanwhile, states came up with their own regulatory systems, and today, they dictate how foods in the US should be labeled. But the trouble is that no two states have identical laws in place with respect to food labeling. The wording and enforcement of an expiration date on a certain product in Arkansas isn’t necessarily the same as the wording and enforcement used in New York, even though the labels in question are on similar products packaged at similar times.

All this leads to food waste. Hundreds of millions of metric tons of food go uneaten in the US each year. Globally, that number tops 1 billion metric tons, and it has a carbon footprint equivalent to 87% of all road-transportation emissions.

Consumers view expiration dates as indicators that say “This is no longer edible.” In reality, the dates are set by manufacturers, which use a variety of criteria and analytical techniques to determine a label they deem appropriate for a product. And the dates don’t reflect how people store a particular product in their homes. The various wordings can profoundly confuse consumers. In 2015, the Food Marketing Institute (now called Food Marketplace Inc.) found that 83% of US shoppers had discarded food on the basis of the sell-by date, a label that isn’t related to whether a food is still edible but rather is meant to tell retailers to pull inventory from their shelves. In the European Union, where wording on labels is more uniform, consulting firm ICF estimates that 9.5–12% of all household food waste can still be attributed to date-marking issues.

In the face of these inconsistencies and without better regulations, food and sensor scientists would like to put diagnostics in the hands of consumers. They envision user-friendly devices that could give us a definitive readout: “This food should no longer be eaten.”

Confused consumers

“So much of our food safety guidance has been ‘When in doubt, throw it out.’ ” says Emily Broad Leib, director of the Harvard Law School Food Law and Policy Clinic. “We’re just throwing away so much food, and I find that to be a really unsatisfying heuristic.” Broad Leib supports a labeling system that’s standardized and easier to understand. The latest food labeling bill in the US, H.R. 3981, was introduced last year and would require expiration labels to use only two wordings: “BEST If Used By” for quality-related dates or “USE By” for discard dates, which is similar to the EU’s system.

More straightforward labels like these could cut down on food waste, experts say. But labels still can’t tell consumers about a food’s quality in real time. When determining a label date, food companies first choose a quality or safety attribute—for example, microbe growth, loss of nutritional value due to oxidation, or texture changes due to water migrating through the food. Then they measure how it changes over time for a particular food product and calculate a date according to models they build from their data. But that assumes the food is kept under the recommended storage conditions from farm to fork. “If that food has been exposed to conditions that are not the ones that you recommended, you have no way to know it, and now that label doesn’t mean anything,” says Maria Corradini, a food scientist at the University of Guelph who formerly conducted food aging studies in industry.

Corradini is searching for new ways to gauge food quality as products move through the supply chain. One way her lab analyzes food is by looking at fluorescence fingerprints, which involves tracking changes in the fluorescence signal of food as it ages and comparing those changes with shifts in nutrient levels and other food quality parameters. Such measurements can give you information about that specific vegetable you’re debating throwing out, meaning you don’t have to rely on a static benchmark from a food company. Also, fluorescence fingerprints let you monitor several attributes of the food simultaneously.

The downside, though, is that most households don’t have equipment for collecting fluorescence fingerprints. Corradini recognizes that in the short term, the technique will be easier to incorporate into food processing plants.

Other emerging solutions that measure food aging are package based, like a sticker that changes color in stages and a milk cap that gets bumpy after a certain amount of time when they’re exposed to elevated temperatures. But these souped-up labels are just a proxy for food spoilage and don’t sense the state of the food inside the package. The question remains, Can consumers bypass the dating game while making sure they’re reliably eating fresh food?

Spoilage in the air

One of the first forays that Firat Güder made into food spoilage analysis involved some detective work. Güder started visiting grocery stores around his home in Boston and asking employees how much food they threw out each week. “Interestingly, many of them didn’t want to talk to me,” he says.

But after forming relationships with the store managers, who kicked his questions up the chain, Güder eventually got some answers: “On average, for example, they would easily throw away a couple of thousand dollars’ worth of meat every week.” In a low-margin business like food sales, he says, that money adds up.

That’s why Güder, an engineer now at Imperial College London, has since developed a low-cost cellulose-paper-based gas sensor for tracking food quality (ACS Sens. 2019, DOI: 10.1021/acssensors.9b00555). Certain microbes that grow on top of meat and fish break down the fish’s amino acids and release gases such as ammonia, other amines, and sulfur compounds as their colonies multiply. Güder’s sensor measures how well electrical current flows through the paper, and as the paper absorbs gases from inside a food container, the flow of current changes, signaling microbe growth.

Güder’s sensors are responsive: if you were to put them in a sealed container with ammonia inside, they would register a signal in mere minutes. So consumers could theoretically install them inside a product’s wrapping after purchase. Because Güder’s sensors use an electrical signal as a readout, they would offer a more precise reading than existing color-changing labels that a consumer has to interpret as being yellow, orange, or red, he contends. Users can even tap their phones to get readings from the sensor through an integrated radio-frequency tag, Güder says.

Most homes already come equipped with a gas sensor: the human nose. While we’ve all sniffed the milk or fish in our refrigerator and thought that measurement was reasonably precise, lab-made sensors can be significantly more sensitive than our noses. Thus, they can signal not just when food has exceeded a particular level of spoilage and needs to be thrown out, which isn’t particularly helpful for reducing food waste, but also when it’s almost bad and needs to be eaten. Güder’s sensors track levels of ammonia over a wide range, picking up concentrations from 0.2 ppm—about 100 times as sensitive as the nose—up to 1,000 ppm.

Ken Suslick at the University of Illinois at Urbana-Champaign also builds sensors that can track food spoilage. Suslick points out that simple sensors like Güder’s don’t have specificity. In other words, the paper sensor gives a signal for all water-soluble gases that are absorbed, but it can’t piece apart how much each gas is contributing to the sensor’s signal. Also, Suslick says, the paper sensors in their current form would be too sensitive to changes in humidity to be usable for sensing ammonia in real-world spoilage applications. Güder’s team is looking into how it can use chemical modifications of the paper to distinguish between different gases, and it claims that changes in humidity aren’t such a big concern in an atmosphere like the inside of a package of meat, where the relative humidity is constantly close to 100%.

To get around the problems that Suslick points to, his lab at Illinois has come up with what it thinks is a more fine-tuned approach. Rather than use a single measurement, like electrical conductivity through paper, Suslick’s sensors stream air across a series of different sensors to generate a complex signal that can be processed and classified. This mimics the way our noses sense and identify gases: they hold about 400 receptors that can each bind to various molecules, creating a signal pattern that our brains interpret as a particular smell. Aptly, Suslick calls his sensor an electronic nose—a term adopted across the globe for analytical instruments that use this type of detection scheme.

In the case of Suslick’s meat spoilage sensor, a disposable plastic or paper strip with an array of chemically responsive dyes printed onto it is loaded into a handheld “sniffing machine” (ACS Sens. 2016, DOI: 10.1021/acssensors.6b00492). As air is drawn into the machine across the sensor strip, the dyes change color as they react with volatile compounds in the air sample. For instance, some dyes will change color if they’re exposed to basic compounds. The device takes a photo of the color pattern generated, compares it with a library of patterns created by known compounds at defined concentrations, and statistically groups the new pattern with similar responses that correspond to different states of freshness. Suslick’s group has also shown that a similar device can sense ethylene—a gas that fruits release when they ripen and that causes nearby fruit to ripen as well (Anal. Chem. 2018, DOI: 10.1021/acs.analchem.8b04321).

Meanwhile, Aryballe, a French digital olfaction company that raised about $7 million in series B funding last year, is also making devices that mimic the nose. In France, where it’s common to buy meat from a local butcher, a baguette from a bakery, or cheese from a fromager rather than all those from a supermarket, expiration dates aren’t the norm. “You have this fresh material that’s coming from a local store, and there’s no notion of expiration dates,” says Fanny Turlure, global product manager at Aryballe. The company thinks its technology could provide a way to track spoilage of these unmarked foods.

Aryballe’s e-nose relies on pattern recognition as well, but instead of using an array of dyes, it uses an array of immobilized peptides integrated onto the branches of a photonic chip, creating a series of mini bio­sensors. The peptides grab onto volatile chemicals from an air sample pulled into the device. Every sensor binds the volatile chemicals to some degree and changes the way light travels through its branch of the chip. So the sensors’ various binding strengths collectively create a pattern that the e-nose can read and send to the cloud to be matched with patterns in a database corresponding to volatile-chemical profiles. With a database that’s large enough, Aryballe’s device might one day tell whether food is fresh, already spoiled, or starting to flip, Turlure says.

Finding a way home

For all their merits, food sensors like these still haven’t arrived in households. One obstacle to their arrival is price.

E-noses like Aryballe’s can be expensive. For instance, the company has an older handheld device, the NeOse Pro, that can detect food spoilage. Sold for $12,500, the NeOse Pro was marketed to firms carrying out R&D for new product development. Aryballe hopes to lower the price of its new technology by integrating the sensor into devices consumers already spend their money on. Turlure says the company is approaching appliance makers to integrate the sensor into, for example, the meat drawer of refrigerators.

When it comes to consumer-friendly food spoilage detectors making it to market, though, Arun Ramesh, a research manager at the consulting firm Frost & Sullivan, is betting on simple sensors like the paper ones made by Imperial College’s Güder. The sensors cost about 2 cents to make, “which is incomparably cheap compared to others in the market,” Ramesh says. Güder’s team is able to keep the price down because it makes the sensors by drawing electronics on paper with a simple ballpoint pen loaded with commercially available conductive carbon ink.

Güder has cofounded the company BlakBear, named for the animal’s keen sense of smell, to commercialize this technology. The firm is weighing its options on how to enter the market. It hopes to have scannable tags in select grocery stores by the end of this year, and it’s prototyping consumer products like scannable stickers that shoppers could apply to the inside of a food package, and “smart Tupperware” that could monitor problematic foods.

A separate problem for consumer diagnostics is data. BlakBear and the industry as a whole need to figure out when consumers tend to throw away various types of food so that they can develop devices that can slash that waste. For instance, if consumers find products that track the spoilage of chicken most useful, firms would focus on sensors for that type of spoilage. “Where the data stream stops is as soon as a food product is sold,” Güder says. “As soon as a customer buys it, we have no idea how they use the product.”

Meanwhile, companies like Aryballe need to build up their databases to boost devices’ predictive power. For instance, Aryballe wants databases that include as many spoilage odor profiles for as many food products as possible to teach its technology how to work in various situations. “We are a start-up,” Turlure says, “and we don’t have unlimited resources. So we have to rely on our customers to help us build databases.”

On that front, Tellspec, a company that specializes in rapid, affordable, and portable analysis of samples using spectroscopic sensors, has been creating databases of spectra from food samples, which can then be fed to the firm’s machine-learning framework to determine a food’s quality and nutritional value. The company sells a handheld near-infrared scanner to food regulators, food retailers, and some restaurants for spotting food fraud and contaminants, especially in fish. While Tellspec has done spoilage studies on fruit, fish, and meat, CEO Isabel Hoffmann says her small company is not aiming to market food spoilage devices to consumers. She sees an avenue for food spoilage sensors in smartphones, which may eventually incorporate near-infrared detectors and provide consumers with access to the technology. “It’s going to be the large smartphone companies that are already manufacturing and distributing smartphones,” Hoffmann says.

In the meantime, she adds, Tellspec can build databases for food spoilage, quality, and contamination data, and once smartphones make the transition, “we can help people make informed decisions on everything.” Similarly, versions of Suslick’s e-nose work though a smartphone-based platform, and the University of Guelph’s Corradini thinks phones could collect a simplified version of fluorescence fingerprint data in the future too.

Standardized food spoilage sensors will be easier to incorporate into the in-line and large-batch processes that happen in food production plants before they make it to consumers. But as long as we’re stuck with static, and often confusing, expiration labels placed on food that’s been packaged and shipped around the globe, the use case remains for sensor companies to gather data and make new tools for monitoring food in the home.

“I’m old enough to like Star Trek,” Corradini says, so a scanner like the fictional tricorder that might one day analyze the food we eat has long been on her radar. “A lot of things that seem to be sci-fi, they’re coming to realization, so why not this one?”