Move Over, Quartz

The atomic clock gets smaller and cheaper

A standard quartz wristwatch gains or loses about a second per day, but soon there will be a tiny, cheap atomic version that will take 300 years to stray that far.

So what? Why can't we just keep on measuring everyday things with quartz and use today's clunky, expensive atomic clocks for measurements that require greater precision? The reason is obvious to an engineer: when you find a way to do something more cheaply and easily, you're not doing the same thing, but something new.

Tiny atomic clocks could make a big difference in a lot of applications. Today's Global Positioning System, for instance, can place you accurately only if you can reach four GPS satellites at once. If you and the satellites worked off perfectly synchronized clocks, however, you'd need only three of them. That's a big advantage whenever the view is restricted and it's hard to get a direct view of the satellites, as it often is in cities.

Then there's spread-spectrum communications, in which the transmitter sends data by hopping around on many different frequencies. To avoid losing data, the receiver must measure a new, incoming frequency rapidly, and there is a trade-off between the speed of transmission and the errors that occur as the receiver hunts for the frequency. With a stable clock, the receiver can lock onto the carrier frequency more quickly, so the transmission takes less time. The faster data transmission also makes it harder for eavesdroppers to listen in--which is one reason that the Defense Advanced Research Projects Agency is helping to fund the clock.

The new atomic clock is not to be confused with clocks and watches that operate using a radio signal linked to the atomic standard. Its core is the size of a grain of rice--that is, chip-scale--because it is made with standard microelectronic manufacturing techniques, says John Kitching, who invented it at the U.S. National Institute of Standards and Technology (NIST), in Boulder, Colo. "Not only can we make these devices very small," he says, "but we can also make many of them at the same time, and that's very important for low-cost production."

The chip-scale clock uses some of the same principles as the standards institute's primary clock, which is about the size of a compact automobile. In both, electromagnetic radiation causes cesium atoms in a vapor, which is enclosed in a cell, to oscillate at a stable frequency. In the primary clock, the radiation is a microwave field, while in the chip-scale clock it is pulses of infrared light from a laser, and everything but the vapor is solid state.

Here's how it works: a local oscillator circuit toggles an infrared laser on and off at a frequency near the natural vibrating frequency of cesium. When the laser shines into a cell containing cesium atoms, it stimulates them to vibrate at exactly the frequency of the local oscillator. At the other side of the cell, a photodiode measures how much of the light has gotten through. As the laser's pulse rate approaches the natural vibration rate of the cesium atom, the amount of light transmitted through the gas rises sharply. A dc voltage proportional to the difference between the natural vibration frequency of the atoms and that of the local oscillator gets fed back into the oscillator, locking the laser's pulse rate to the vibrations of the cesium atoms, which are very accurate indeed.

Right now the cesium-vapor chamber takes up a few cubic millimeters. Throw in the laser, the optics, and the electronics, and we're up to maybe a cubic centimeter.

A particular chip-scale clock may tick a little faster or slower than the 9 192 631 770 ticks that define the second at the government's standard atomic clock in Boulder. But Kitching says manufacturers could easily calibrate the tiny clocks against that standard.

Besides being cheap, the new atomic clocks will also be frugal with power, says Kitching. Today's commercial atomic clocks are about the size of a pack of cigarettes and consume several watts; the standards institute aims to make a rice-size model that needs 30 milliwatts. It would therefore be able to log data in places where the clock must run untended for months.

Kitching says NIST has been approached by an oil exploration company that wants to measure seismic vibrations on the ocean floor with millisecond precision.

The chip-scale clock is a work in progress. The researchers have built what they call the physics package [see photo, ], which contains the enclosure for the cesium atoms, the laser, the optics, and the photodiode. The parts that remain to be integrated into the system are the local oscillator, which a team at the University of Colorado is designing, and a low-power integrated circuit for the control electronics.

"The physics package is the part that hasn't been done before," Kitching says. "And the demonstration that it's a viable technology gives a lot of impetus to go forward with the other parts." He expects the tiny clocks to be available in two to three years.

Symmetricom, Honeywell, Rockwell Scientific, and Sarnoff Corp. are developing their own chip-scale clock designs in parallel with the work at NIST.

Advertisement
Advertisement