Stanford University News Service



CONTACT: Stanford University News Service (650) 723-2558

Stanford's chip technology creates sensor systems for smart appliances

STANFORD - Life is too short to eat burnt toast. Now you may never have to, provided you can shell out enough bread - up to $80 at Macy's - for a toaster that monitors the state of your breakfast with sensors and microchips.

"Smart" appliances are just one result of engineers' developing new ways to connect detectors to computer chips to create sensor systems. Enabling machines to better detect and react to their environments, sensor systems have spawned a new generation of sophisticated medical devices, automobiles, aircraft and even toys.

And in the future, they could be implanted in everything from auto engines to human bodies, warning of breakdowns before they occur.

Engineers at Stanford University's Center for Integrated Systems (CIS) are building sensor systems with an innovative technology called biCMOS, for bipolar complementary metal oxide semiconductor.

Sensor systems combine detectors on silicon wafers the same way circuits are integrated on computer chips, according to project organizer Gregory Kovacs, an assistant professor in electrical engineering.

Typically, one end of a sensor system talks to the world, while the other end talks to a computer, Kovacs said. The result is that sensor systems transmit signals that can be manipulated and analyzed as data.

Arrays of sensors on the silicon chips respond to the environment with greater sensitivity than do single sensors, Kovacs said. He explained this phenomenon with a hypothetical array of pressure sensors made of progressively larger bendable beams:

"You have an array of pressure sensors in graduated sizes. When one bends to near the limit of its useful range, the next one starts to bend. Then the next one starts to bend. So you can make an array that has an enormous dynamic range by just scaling and repeating one design.

"That concept of having little flimsy sensitive ones going to bigger and stronger ones is called recruitment, and it's used in nature. For example, in muscles you have little tiny fibers that are very sensitive and very big, strong fibers. As you increase your grip, your neuromuscular system is recruiting bigger and bigger fibers."

Arrays can enable one small chip to detect many physical properties at once.

"You can measure light intensity, temperature, pressure, vibration or a whole number of other things in one little, tiny spot," Kovacs said. "Currently that cannot be done very well."

Today's individual sensors are ubiquitous but taken for granted. Engine sensors note when a car's fuel mixture is too lean and adjust it accordingly. Optical sensors in at least one type of computer mouse relay its position on the mousepad. A photosensor in a television set turns on the appliance in response to an infrared beam sent from the remote control. Light sensors in a clock radio tell the display to dim while the lights in a room are off.

"Hair dryers are now coming out with sensors that sense leakage current," Kovacs said. "If you have your foot in a puddle of water and they sense a microamp of current, before you can get electrocuted, they trip a little breaker in the hairdryer and stop the current. . . . That's going to save lives."

In the future, sensor systems may bring safer automobiles and airplanes, more efficient ways to test the environment for toxic contaminants, and less invasive diagnostic tools for critical care medicine.

"The air bag sensor that Analog Devices brought out last year is going to be - if it's successful - one of the first mass- produced integrated sensor systems to be employed in consumer goods," Kovacs said. "It's a complete chip - like what we're trying to build - with circuits and the accelerometer all in one chip. It's cheap. It's the most elegant design you've ever seen, and it's designed to sit in dashboards and withstand all the temperatures and battering cars take, and 10 years later still be working to deploy an air bag, if needed."

Kovacs also suggested using sensor systems to monitor the internal milieu of cars.

"There are a lot of things we don't sense that would be good to know about," he said. "For example, you could put a system on a chip that could look at the motor oil in your car and tell you if something's wearing by looking at the metal present, or tell you if the oil's viscosity is breaking down. . . We don't measure anything but oil pressure and temperature now."

Easy-to-detect chemicals could be built into parts such as washers and bearings, he said. As parts wear, sensors would detect a rise in the concentration of chemicals in automotive fluids. Parts could be replaced before they break down.

Currently, Kovacs' lab is in the beginning stages of developing a sensor for such water contaminants as heavy metal ions of lead and cadmium.

"Chemical sensing is still in its infancy," he said. "People are still trying to figure out how to do it."

In the medical arena, sensor systems could provide physicians with important tools for critical care.

"Ultimately, they may even build sensors into people," Kovacs said. "There's no reason why your pacemaker couldn't also measure your blood chemistry." Information gathered by sensors could be sent to physicians by radio telemetry, a method currently used to monitor heart rhythms in pacemaker patients.

The aerospace industry, "needing zillions of sensors to run jet engines and the like," is currently one of the largest buyers of sensor systems, Kovacs said.

"If that industry declines, there'll be plenty of uses for sensors elsewhere," he said. "I think we're going to see some of the best and lowest cost sensor technology come out in toys. Go down to Toys R Us and you can see some great technology."

Mattel markets a toy called Power Glove that is essentially a crude virtual reality device powered by such a chip.

Under the supervision of electrical engineering Prof. Jim Plummer, biCMOS wafers are made three or four times a year in the Stanford center's "clean room," Kovacs said. "A large cast of supporting geniuses" - including graduate students Todd Whitehurst, Ion Opris and Rich Reay; senior research associate John Shott; and staff member Chris Storment - carry out more than a dozen delicate steps to produce the chips, he said.

Kovacs' group uses a novel approach to design and manufacture standard building blocks of compatible electronic components. Its process is cheaper and more efficient than existing methods, Kovacs said.

"The approach is to build a library of circuits that are tuned in to sensor applications," he said. "The person that wants to build a sensor system says, 'OK, I'm going to need these building blocks. Let's take them from the library and just put them down, like tiles or Legos, and build it up on a screen.' Then they submit it, and it appears on a chip a few months later."

Kovacs' work is funded through the Center for Integrated Systems, a consortium of industry, academia and government, according to CIS Executive Director Rick Reis. General Electric is the most involved of the project's 10 industrial sponsors, Reis said.



This is an archived release.

This release is not available in any other form. Images mentioned in this release are not available online.
Stanford News Service has an extensive library of images, some of which may be available to you online. Direct your request by EMail to

© Stanford University. All Rights Reserved. Stanford, CA 94305. (650) 723-2300.