Wireless Sensors That Live Forever

Energy harvesters and radioisotopes fuel tiny transmitters

Soon enough, say some engineers, miniature wireless sensors will be located in spots where it would be inconvenient, to say the least, to change their batteries—inside your body, within the steel and concrete of buildings, in the dangerous innards of chemical plants. But today, even the most robust nodes can be counted on to last only a few years. Ideally, engineers need a sensor that can last forever without external power sources or battery changes. According to research presented in December at the International Electron Devices Meeting, in Baltimore, that dream is within reach.

Two research teams tackled the problem of sensor longevity in two very different ways. Both methods rely on piezoelectric power generation, in which a microelectromechanical systems (MEMS) cantilever converts mechanical motions into electrical power. However, the cantilever's movements are propelled by very different mechanisms—one by a radioisotope and the other by vibrations harvested from the environment. In a big step forward, both methods fully powered autonomous wireless systems.

All self-powered communication nodes must be able to retain their memory state and periodically transmit that state. These tasks require an average power budget of between about 0.1 microwatt and 1 milliwatt, a budget that can be met by either of the two methods of power generation, as the recent research showed.

Cornell University researchers Amit Lal and Steven Tin created a piezoelectric generator using a small amount of nickel-63 (Ni-63)—a mildly radioactive isotope with a few extra neutrons in its nucleus. When it decays, Ni-63 emits relatively harmless beta particles (energetic electrons that penetrate only about 21 micrometers into a surface). The device was able to create sufficient output energy to achieve a 5-mW RF pulse every three minutes. Most important, because the half-life of Ni-63 is just over 100 years, the device could function autonomously, according to Lal and Tin, for about that long.

At the Holst Centre, part of nanoelectronics research center Imec, in the Netherlands, Rene Elfrink, a research engineer, and his colleagues took a different approach, creating a wireless, autonomous temperature sensor powered by an aluminum nitride vibration harvester. Vibration energy harvesting typically requires a small (approximately 1 µm) vibration at a specific frequency. Powered by the harvester, the device measured the temperature and transmitted the data to a base station up to 15 meters away every 15 seconds.

Existing energy harvesters tend to be fine machined, a costly method that can produce devices that generate power at levels from microwatts to watts. Such large, fine-machined devices can generate more power than smaller, MEMS devices like Imec's. But MEMS devices are cheaper to produce, easier to integrate with existing sensors, and now they can generate enough power to run a wireless sensor node.

Unpackaged, the Imec device produced 85 µW, a record-setting amount for a MEMS harvester. However, when packaged at atmospheric pressure (as most devices are), air damping caused the output to fall below 10 µW—not enough to keep the sensor transmitting data every 15 seconds. Elfrink's team solved the problem with a vacuum package.

Peter Hartwell, a senior researcher at HP Labs, in Palo Alto, Calif., says Imec's technology is ”definitely a step forward.” Energy harvesting research is important to HP Labs, which is developing sensors for its Central Nervous System for the Earth project, a vision of peppering the world with minuscule sensors. Power is one of the remaining obstacles in making the vision a reality; HP Labs' accelerometers require about 50 mW.

However, says Hartwell, vibration harvesters so far are tuned too narrowly to specific frequencies to be useful everywhere.