Last week Pres. Donald Trump issued an executive order to pull the nation back from climate change action. One target was methane emission regulations set by the U.S. Environmental Protection Agency and Bureau of Land Management under the Obama administration. The rules require the oil and gas industry to control methane emissions, a powerful greenhouse gas, at its operations. Trump’s order directs agencies to reconsider those rules, and possibly rewrite or get rid of them altogether.

Despite Trump’s declaration, cities, states and companies still want to stop methane leaks across industry, because the gas worsens global warming and leaks lower revenues. That’s why researchers are developing a technology they think may help: a tiny chip to continuously monitor methane emissions.

Although the U.S. releases much less methane than CO2, the former has 25 times more warming potential per pound across a 100-year period. Oil and gas producers emit more of this gas than any other industrial activity in the U.S., and a significant portion of it comes from leaks in equipment at locations like well pads, where drilling and extraction happens. Today the industry uses an expensive, inefficient method to monitor for leaks: infrared cameras, which cost $85,000 to $125,000 and require a human operator. That means companies must send a person to manually scan equipment for escaping gas. “These cameras are expensive, and it takes one to two hours to do a well pad, four to six hours for a compressor station, and it could be a full day for a processing plant,” explains Doug Jordan, corporate environmental program director at Southwestern Energy. “It’s very resource-intensive.” Companies typically do these checks only once a quarter to once a year. If, say, a piece of equipment starts leaking right after a survey, it might go undetected for months. Furthermore, a camera might only be sensitive enough to catch larger leaks, potentially missing smaller ones.

IBM scientists and engineers, working with researchers at Harvard and Princeton universities, have devised a miniature sensor chip—about five by five millimeters in size—that continuously watches for methane. The sensor, housed on a small silicon wafer, contains a laser and a silicon optical cable that channels the laser light through it.* Some of that light shines outside the cable into the air; when methane molecules waft above the sensor, they absorb a specific wavelength of the light, creating a unique signature. “Our method is designed to detect very small changes in light absorption,” explains William Green, manager of the Silicon Integrated Nanophotonics Group at IBM. That information gets fed back to a photodetector on the chip, which converts the signature into an electrical signal. “We map out the absorption and figure out how much methane is there,” Green says.

Oil and gas companies would embed a handful of these sensors at various locations around a well pad or compressor station, anywhere from 10 to 100 feet apart. The whole system is wireless, which is critical because oil and gas sites are often remote. When the chips, which continuously monitor the environment, identify a leak, they automatically send that data to IBM’s cloud-based computers. The computers rely on physical models that IBM is currently working on, which combine complex dynamics like wind turbulence, humidity, elevation and temperature to determine the methane’s origin. The system will also eventually incorporate machine learning to help improve the modeling.

Once IBM’s models determine the leak source, companies can immediately send people out to fix it. “One of the advantages is that you’re getting alerted real time and it’s being time-stamped and geographically identified,” explains Norma Sosa, manager of Systems and Technologies for Cognitive IoT at IBM. To make the system even more autonomous and maintenance-free, the sensor system can work off solar power. The research is part of a program funded by the U.S. Department of Energy’s Advanced Research Projects Agency–Energy (ARPA–E).

Some commercial tools already exist for methane detection. These include the infrared cameras used by the oil and gas industry as well as individual sensors that rely on the same laser light absorption technique that IBM uses—although the design and make-up of IBM’s chip and its smart system are unique. Plus, Green says, methane sensors on the market today are very expensive, large and require a lot of power. “It is impossible to envision deploying a network of [those] sensors, let alone even one such instrument at each well pad for continuous monitoring,” he explains. “The cost is just too high.” IBM has designed its new methane sensor to be low cost—it is aiming for $200 per chip. That means oil and gas companies could set up a large number of these wherever they need them. “This new breed of chip-scale sensors is really intriguing,” says Daniel Zimmerle, director of Colorado State University’s Electric Power Systems Laboratory and principal investigator of its Methane Emissions Test and Evaluation Center. “It’s untested, but it moves the needle on price.”

IBM is currently testing its sensor in-house, and plans to test its data models and related components this summer in Colorado. It is also in discussions to run a pilot test for at least one of Southwestern Energy’s well pad sites. IBM envisions other possibilities for this technology as well. The company says its sensors may one day monitor vehicle exhaust pollutants or water contaminants, detect chemical changes around volcanoes or even analyze someone’s breath for disease biomarkers. “That’s a vision for the more distant future,” Green says. For now, the researchers will settle for tackling climate change.

*Editor's Note (4/7/17): This sentence was edited after posting. The original incorrectly identified the cable as glass rather than silicon optical.