Hydrogen could become an important part of clean energy, but its leaks are dangerous. Reliable sensors are therefore needed wherever hydrogen is produced, stored, transported or used. A persistent problem is humidity: water molecules settle on the active surface of sensors and prevent them from responding accurately, especially in hot and humid air.
An international team of scientists from the WPI Advanced Institute for Materials Research (WPI-AIMR) at Tohoku University (Japan), the University of Utah and Pacific Northwest National Laboratory (USA) has proposed a catalytic hydrogen sensor that operates in such conditions without external heating. The results were published in the journal Advanced Materials.
The device is based on a network of palladium nanoparticles with many grain boundaries. This is the name for tiny boundaries between neighboring crystals. These, it turned out, help the sensor resist interference from moisture.
The sensor detects hydrogen at temperatures from 22 to 80 degrees Celsius and relative humidity from 0 to 90%. It can detect a concentration of just 100 parts per million at 80 degrees and 90% humidity, when water vapor in the air is many times greater than hydrogen. The device also showed high selectivity: it did not respond to 21 potentially interfering gases. Its operation remained reproducible over 100 consecutive measurement cycles and stable in tests lasting 125 days. The sensor meets the requirements of the ISO 26142 standard for response to humidity in challenging conditions.
To explain the results, the researchers combined experiments tracking structural changes and water removal with molecular dynamics modeling and first-principles calculations. "We found that grain boundaries help hydrogen penetrate into palladium and rapidly form palladium hydride, even when the surface is wet," Hao Li, a distinguished professor at WPI-AIMR, was quoted as saying in a press release.
This hydride binds water weakly, so the catalytic sites remain accessible for the reaction of hydrogen with oxygen, which produces the measurable signal. The combination of crystal boundaries, hydride formation and water adsorption explains why the material retains sensitivity where conventional catalysts falter.
In addition, the team integrated a machine learning model into a portable sensor prototype. Taking into account changes in temperature and humidity, the system estimated hydrogen concentration with a deviation within 5% in prototype tests under randomly changing conditions. Besides demonstrating a more robust sensor, the work points to grain-boundary engineering as a promising approach for safe hydrogen monitoring at industrial facilities and in future hydrogen energy.