Wireless-sensor networks can monitor factory machinery, track environmental pollution, and measure the movement of buildings and bridges. But while their uses are seemingly endless, wireless sensors have a significant limit—power. Although improvements have brought energy consumption down, batteries still need to be changed. For networks in remote locales, replacing batteries in thousands of sensors is a staggering task.

To get around the power constraint, researchers are working on harnessing electricity from low-power sources in the environment, such as vibrations from bridges, machinery, and foot traffic. They’ve developed microelectromechanical systems, or MEMS, that tap into those sources, aiming to eventually do away with batteries entirely.

Now MIT researchers have designed a MEMS device the size of a U.S. quarter that picks up a wider range of vibrations than previous designs and generates 100 times the power of similar devices. The team’s results appeared in Applied Physics Letters.

To harvest vibrational energy, researchers have typically looked to piezoelectric materials such as quartz, which naturally accumulate electric charge in response to mechanical stress. They have exploited these materials at the microscale, engineering MEMS devices that generate small amounts of power.

A common energy-harvesting design employs a microchip with a piezoelectric material known as PZT glued atop a tiny cantilever beam. As the chip vibrates, the beam moves up and down like a diving board, bending the PZT. The stressed material builds up an electric charge, which is picked up by tiny electrodes.

However, the cantilever approach is limited. The beam itself has a resonant frequency—a specific frequency at which it wobbles the most. Outside this frequency, the beam’s response drops off, and so does the amount of power generated.

Sang-Gook Kim, PhD ‘85, a professor of mechanical engineering, and Arman Hajati, PhD ‘11, came up with a design that increases the device’s frequency range, or bandwidth. They engineered a microchip with a small bridgelike structure that’s anchored to the chip at both ends. Then they deposited a layer of PZT on the bridge, placing a small weight in the middle.

In vibration tests the device responded at a wide range of low frequencies, generating 45 microwatts of power—beating current designs by two orders of magnitude.

“Our target is at least 100 microwatts, and that’s what all the electronics guys are asking us to get to,” says Hajati. “For monitoring a pipeline, if you generate 100 microwatts, you can power a network of smart sensors that can talk forever with each other.”