NanoSI researchers have developed and experimentally optimized a new class of palladium-coated aluminum nitride resonators capable of detecting hydrogen with high sensitivity, advancing compact sensing technology for applications such as hydrogen leak detection and monitoring.

The research, published in the IEEE Journal of Microelectromechanical Systems (JMEMS), investigates how the design of palladium-coated aluminum nitride (AlN) laterally vibrating resonators, or LVRs, can be optimized to improve hydrogen sensing performance.

As hydrogen continues to gain attention as a potential clean-energy carrier, reliably detecting leaks is an important safety and infrastructure challenge. Hydrogen is colorless and odorless, and its small molecular size can make leaks difficult to detect. Compact sensors capable of identifying low concentrations of hydrogen could therefore play an important role in future hydrogen production, storage, transportation and energy systems.

The NanoSI team’s work explores a resonant sensing approach that takes advantage of palladium’s interaction with hydrogen.

When palladium is exposed to hydrogen, the interaction produces mechanical stress within the palladium film. In the devices studied by the researchers, that stress changes the resonant frequency aluminum nitride resonator. By precisely measuring that frequency shift, the device can detect the presence and concentration of hydrogen.

Rather than focusing on a single device configuration, the researchers fabricated palladium-coated LVRs operating from 123 to 396 MHz on a single four-inch wafer, allowing them to systematically investigate how several design variables affect sensor performance.

Figure 1

Schematic views of the Pd-coated AlN LVR. (a) Cross-sectional view; (b) top-side view; (c) back-side view.

The team examined three key factors: operating frequency, palladium thickness and residual stress within the palladium film.

Their experiments revealed that each can significantly influence hydrogen sensitivity.

Sensitivity increased consistently as operating frequency increased, rising from approximately 15 to 54 Hz/ppm across devices operating from 123 to 396 MHz with a 100-nm palladium layer. Increasing the thickness of the palladium layer from 100 to 200 nm produced an approximately 8.5-fold increase in sensitivity.

The researchers also identified residual stress as an important—and previously underreported—design parameter. For devices with the same palladium thickness, optimizing residual stress produced as much as a 2.6-fold improvement in sensitivity.

Figure 2

SEM images of released AlN LVRs at identical stage inclination: (a) low-residual-stress device (flat membrane) and (b) high-residual-stress device (buckled membrane).

 

Giubelei states that “A single high-performing device tells you that palladium-coated resonators can work well, but it doesn’t tell you why, or how to replicate that performance at a different frequency or for a different application,” said Giubilei. “By systematically varying frequency, palladium thickness and residual stress across dozens of devices on the same wafer, we could actually separate out how each parameter contributes, instead of guessing from one lucky result. That’s what turns this into a design methodology other groups can use, rather than just one good data point.”

Importantly, the improvements in sensing performance were achieved while maintaining strong electromechanical characteristics. The devices demonstrated quality factors as high as 1,625 and electromechanical coupling coefficients reaching 1.94%, with the reported quality factor representing a 2.7-fold improvement over prior ScAlN-based implementations.

Among the devices tested, a 317 MHz resonator with a 150-nanometer palladium layer emerged as the optimal design point. The device achieved a quality factor of 1,411 and demonstrated a peak hydrogen sensitivity of 772 Hz/ppm in the low-concentration regime.

The results place the palladium-coated AlN resonators among the most sensitive resonant hydrogen sensors reported to date.

Beyond the individual performance metrics, the work provides researchers with a clearer roadmap for designing future resonant hydrogen sensors. By demonstrating how frequency, palladium thickness and residual stress work together to influence performance, the study identifies practical engineering parameters that can be adjusted to improve future devices.

“Residual stress was the real surprise,” said Giubilei. “Frequency and palladium thickness are the two variables people typically optimize, but we found that for devices with identical palladium thickness, tuning the residual stress in the film alone could shift sensitivity by up to 2.6 times. That’s a parameter that’s often overlooked in the fabrication process, and it turned out to matter almost as much as the material thickness itself.”

The research reflects NanoSI’s broader work at the intersection of advanced materials, micro- and nanoelectromechanical systems, sensing and integrated technologies—translating fundamental understanding of materials and devices into technologies capable of addressing real-world challenges.

The paper, “Experimental Characterization and Optimization of Palladium-Coated Laterally Vibrating Resonators for High-Sensitivity Hydrogen Sensing,” was published in the IEEE Journal of Microelectromechanical Systems.

Authors: Gaia Giubilei, Farah Ben Ayed, Pietro Simione, Zhenyun Qian and Matteo Rinaldi.

Congratulations to:

Gaia Giubilei

Farah Ben Ayed

Pietro Simeoni

Zhenyun Qian

Matteo Rinaldi