MEMS and Atomic Clocks

By Kathleen Kocks | November 1, 2004
Send Feedback

This past August, scientists at the National Institute of Standards and Technology (NIST) unveiled a development that could deliver a thousand-fold performance improvement over the quartz crystal clocks used in many communication products. Priority applications for avionics include GPS, secure communications and radar. Avionics test equipment also would benefit.

The development is a chip-scale atomic clock (CSAC), a timepiece that is small enough to fit on a computer chip. The U.S. Defense Department's Defense Advanced Research Projects Agency (DARPA) is developing the CSAC in a three-phase program, now in its second phase. Program participants include DARPA, NIST, various universities and private companies, among them Honeywell and Rockwell Collins.

Such a clock would deliver extremely stable timekeeping and could vastly improve all forms of radio. All radios -- in fact, anything that uses the radio spectrum -- require some form of internal clock to work, and the more stable the clock, the better the performance. The quest for a stable clock for radios has evolved from using simple oscillators to employing quartz crystals.

But the ultimate stable clock is the atomic clock, which is used by NASA for satellites and by telephone companies and television stations for network synchronization. Atomic clocks use the natural, regular oscillations of an atom or molecule to deliver accurate timekeeping. Materials most commonly used include hydrogen atoms, cesium atoms, and rubidium gas.

The problem is, the smallest atomic clocks are power-hungry and as big as a tabletop, limiting their use. DARPA's goal is to change that.

Scientific Advances

"What we are trying to do is take the atomic clock off tabletops and put it into the hands of users, commercial and military," states Clark Nguyen, program manager for DARPA's Microsystems Technology Office. "We want to miniaturize an atomic clock to the point where it is portable."

"Right now, the most accurate atomic clock--the NIST F-1 cesium fountain atomic clock in Boulder, Colo. -- has an accuracy of 3.8 x 10-15 [which means it does not gain or lose a second in 30 million years]. But its volume is 130 cubic feet [3.7 m3] and its power requirement is 500 watts. It's definitely not portable. Over time, developers have shrunk atomic clocks to tabletop-sized units, consuming 60 watts of power.

"But if you want a portable clock, you need much less power," says Nguyen. The CSAC program's goal is to build an atomic clock that is 0.061 cubic inch (1 cm3) and consumes only 30 milliwatts of power. "We also want a stability of 1 x 10-11 integrated over one hour," he says. "This is equivalent to an error of one microsecond per day. This performance isn't as good as the NIST F-1, but that level of accuracy is not needed for many applications."

To accomplish their goals, CSAC scientists are applying traditional atomic physics to micro electromechanical systems (MEMS) technology. MEMS create micron- and nano-scale mechanical devices that can be integrated onto silicon and work with electronics to create not just intelligent chips, but intelligent systems on a chip.

MEMS have been around for 25 years, and products are now being mass-produced (similar to computer chips) at relatively low costs. MEMS microsensors and microactuators are used in many practical applications. Two common ones are accelerometers for automobile air bags and the micromirrors in projection TVs.

"MEMS have come a long way, and DARPA is trying to take advantage of the technology," Nguyen says. "All my programs at DARPA are based on MEMS technology, and through different applications of the technology, we're finding that the smaller you make devices, the less power they consume and the better they seem to work."

What NIST unveiled in August was a CSAC physics package which is just part of the atomic clock. It measures 1 cm3, consumes 75 milliwatts of power, and its short-term stability equates to gaining or losing one second every 300 years.

"It is many times smaller than a dime," Nguyen explains. "They have solved the size problem while retaining short-term stability in the clock's performance. We now need to further improve the long-term stability.

"By mid-next year, at the end of the program's second phase, we will hopefully have atomic clocks that use chip-scale physics, are maybe 10 cm3 [0.61 cubic inch] or so in size, and use power on the order of 200 milliwatts. In the third phase of the program, we will try to further reduce the size and power requirement, and by the program's end in late 2006, hopefully we will have our cubic centimeter, 30-milliwatt atomic clock. Doing that depends a lot on having control circuits to get the long-term stability we are seeking."

Practical Applications

Nguyen says a CSAC would benefit any electronic system that is based on a timing or frequency reference. Among the priority applications attracting the Pentagon's interest are GPS, high-security communications, high-confidence identification friend/foe systems, and ultrasensitive radar that can locate small radio emitters.

According to Greg Olson, business development lead for Honeywell Labs' Advanced Sensing Technologies Group, the chip-scale atomic clock would greatly improve upon the quartz clocks used in many applications. "Quartz crystal clocks are great and work very well, but if you need something that performs better -- like delivering a synchronous time reference that is 1,000 times better with an atomic clock -- then go with it." A lot of people working with GPS, radar and secure communications see the need for a very small, low-powered, stable time source, Olson adds.

GPS has civil and military codes. Military GPS receivers require very stable time references to correlate the codes they get from GPS satellites. However, most military receivers use quartz crystal clocks, and they can't keep the required stability over days of operation. In many cases, military systems employ civil codes to get their initial lock on the satellites and then correlate the military codes, an unsatisfactory arrangement because it doesn't work if civil GPS is jammed. A battery-powered chip-scale atomic clock would solve this problem.

"If you put a stable time source in a GPS receiver," Olson explains, "you can correlate the codes more accurately and efficiently, leading to more reliability and better performance." And a GPS system with an onboard atomic clock would have enhanced anti-jam capabilities, he says.

Addressing communications, Olson says using an atomic clock would result in lower phase noise, a cleaner signal and increased channel selectivity. Another benefit is time synchronization. This is required for secure communications, such as spread spectrum, that use time stamping to encode and encrypt messages and data streams. A stable time source allows you to quickly synchronize to acquire the signal and decode it.

Improving Radar

Aircraft radars, on the other hand, are particularly vulnerable to phase noise caused by vibration and other acceleration-induced effects. This limits detection performance, especially if the radar is looking for slow-moving targets. "Atomic clocks are not so susceptible to vibration, and they should enable radars to get better stability and performance," Olson adds.

It is very early in the development stage to predict other applications for atomic clocks. But if the clock is successfully developed, Honeywell probably will explore what benefits could be achieved from integrating the atomic clock with the GPS in the company's inertial measurement units, Olson says.

Spectrum Space

Looking at atomic clock benefits from a broader perspective is Roy Berquist, principal engineer at Rockwell's Advanced Technology Center.

"The big problem is bandwidth," Berquist declares. "Everybody wants more bandwidth, and the most desirable bandwidth is in the lower frequencies, where you don't have line of sight limitations. To fit more people into the desired frequencies, we are seeing tighter and tighter channel spacing within a given bandwidth."

The more narrow the channel spacing, the more accurate a radio's clock has to be to enable the radio to transmit or receive on the intended channel, Berquist explains. "Accuracy also prevents the transmission from drifting over the channel and interfering with other channels." Better clocks would allow "guard bands" between some channels to be made narrower and narrower, he says.

"Additionally, we are transmitting more and more digital data," Berquist says. All this is pushing a need for more accurate clocks, like an atomic clock. "We're also pretty much at the limit of the physics of quartz crystals. We've used every trick we can to get every bit of frequency stability out of them, and now we need to use a different technology. An atomic clock is one answer."

Development Road

Practical applications of a CSAC are clearly several years down the road. When the DARPA program ends in late 2006, a prototype should exist, but then it will be up to private industry to take the prototype to production. Equally impressive is that this program already has produced new discoveries in atomic physics.

"It's a beautiful program because we have people from diverse communities working together to solve problems," Nguyen begins. "It brings together several academic institutions, the MEMS community and atomic physicists. There have been several instances when expert atomic physicists thought something was impossible, but then another community comes in with its developments, everyone rethinks the situation, and new developments are emerging."

Honeywell's Olson is equally upbeat. "The exciting part about DARPA's chip-scale atomic clock program is that no one has ever successfully developed such a complex system in a MEMS configuration. We've got a MEMS fabrication capability that has enabled us to build a clock on a chip. The gas cell [containing a vapor of cesium atoms] that gives you the oscillation you are measuring is just a few millimeters in size, and it's oscillating in a stable fashion, with the electronics that go along with it. Integration of the MEMS cell with the stable and miniaturized electronics is something no one has done before."

Collins' Berquist also is optimistic. "We think the size and power aspects will be OK, but the bigger question is the cost. However, these clocks will be using MEMS technology and the price for products that use this technology seems to erode very quickly to affordable levels."

Will it become economical to put atomic clocks into all kinds of products? "Now we are talking about putting GPS receivers in cell phones, whereas a short time ago we were barely fitting them in manpacks," Berquist says. "Only time will tell."

Receive the latest avionics news right to your inbox