POSTED 25 AUGUST 2005
Accuracy has been the goal of the clock-making game since the beginning. Back when water clocks were trendy, we guess people kvetched about the incessant "leakage" of time.
Timekeeping got a big boost with the invention of the pendulum clock in the 17th century, and again in 1928, with the quartz clock. Similar vibrating quartz crystals drive the mechanism in almost every modern wristwatch.
Although a quartz clock can stay accurate for weeks or months at a time, this does not impress scientists. These days, they build gadgets called atomic clocks, which use the principles of quantum mechanics to keep time right on the money.
Like most clocks, atomic clocks create and then count periodic movements, or oscillations. In the old pendulum clocks, a weight swung back and forth at a steady frequency, so the clockmaker only had to invent a mechanism to count the swings and drive the clock's hands. It wasn't too accurate, but it beat watching water leak from a tank.
In an atomic clock, the oscillations occur in an electromagnetic field that causes transitions between the two possible quantum-mechanical conditions of an atom. In the commonly used cesium 133 atoms, these occur about 9.19 billion times per second.
If you do it right, that's good enough to make a clock that is accurate to 1 second in 30 million years, or so they say.
Photo: National Institute of Standards and Technology
Why do we need such accuracy?
Those of us who mutter "Just a second" when we mean "In 5 or 10 minutes" may not understand the need for clocks with such accuracy. What's the big allure of hyper-accurate time?
Tight time turns out to be the basis for innumerable communication, scientific and navigation systems. Timing is critical for synchronizing signals between computers. In astronomy, fractional-second errors could sabotage long-baseline radio telescopes, which fuse distant radiotelescopes into one gargantuan receiver.
Global positioning satellites need accurate time. The U.S. GPS system can determine the three-dimensional position of a receiver anywhere on or off Earth to within a few feet. The receiver performs this trick by timing the arrival of signals from any four GPS satellites, then triangulating its position.
In 1998, Stephen Dick, the United States Naval Observatory's historian, told us that each nanosecond -- billionth of a second -- of error translates into a GPS error of one foot. A few nanoseconds of error, he points out, "may not seem like much, unless you are landing on an aircraft carrier, or targeting a missile."
Without accurate timing, GPS would stand for "generally puky system." Each of the 24 GPS satellites holds four atomic clocks, which every day get an accurate time transfusion from the Air Force, which "borrows" time from the United States Naval Observatory.
People say time is money, but when you borrow time from me, I don't lose a nickel or a second...
Quantum mechanics -- the physics of the ultra-small -- originated with the observation that sub-atomic particles can exist only in discrete states, but not in between. It's like an atomic version of a mandatory two-party system. Because only certain "states" are allowed, an electron -- once it's been measured -- can be a Democrat or a Republican, but not a Demican or a Republicrat.
Atoms can have one of two "hyperfine states," and this is the basis of the atomic clock. The magnetic field of the outermost electron must either point in the same direction as the magnetic field of the nucleus, or in the opposite direction. The laws of quantum physics forbid other orientations.
Generally, an atom remains in its hyperfine state. But when prodded by electromagnetic radiation at a specific frequency, it will go through the "hyperfine transition" and switch into the other state. The idea of building a clock around hyperfine states was proposed by physicist Isador Rabi in 1945.
Essentially, an electronic clock selects cesium atoms in one hyperfine state and exposes them to radiation that causes them to switch to the other state. The exact frequency of radiation -- 9,192,631,770 hertz -- needed to cause the transition becomes the regular beat that the clock counts to register time. Only when the atoms "hear" that exact beat will they change hyperfine states.
And after you create that beat you just count it. After every 9,192,631,770 beats, another second has passed.
Running out of time?
Having perused the previous preamble on quantum mechanics, you understand that an atomic clock:
magnetically filters atoms, selecting only those in one hyperfine state;
sprays the selected atoms through microwaves or a laser beam tuned to the hyperfine transition frequency;
measures how many changed atoms come out the other end;
tunes the microwave generator to the frequency that causes the maximum transitions (this is the exact transition frequency); and
"reads" that signal and calculates that every 9,192,631,770 oscillations (for cesium) represent one second.
Atomic clocks can be frighteningly accurate. For example, NIST-7, a cesium clock used at the National Institute of Standards and Technology (NIST) from 1993 to 1999, was accurate to five parts in 10 15. This staggering precision works out to an error of about a billionth of a second per day. To put it another way, this clock will stay within one second of true time for 6 million years.
Beat the clock
Sounds pretty decent. But it's not the watch to stop all watches. That honor belongs to NIST's hyper-accurate cesium fountain clock, which went into service in 1999. The fountain gains precision from several sources. When atoms were sprayed through earlier atomic clocks at several hundred meters per second, the resulting Doppler effect (which also changes the pitch of a speeding train whistle), requires compensation. And the measurement must take place quickly as the atoms speed past.
Diagram: NIST has a good explanation of the atomic fountain.
Instead of spraying atoms, the atomic fountain cools them with lasers to a few millionths of a degree above absolute zero. Then it will "Toss them up like a tennis ball," as Donald Sullivan, head of the time and frequency division at the National Institute of Standards and Technology, put it. The cold, slow-moving, atoms are measured in the microwave chamber on the way up and back down, using the same general technique we've already seen. The improved precision results from the reduced Doppler effect and an increase in measurement time.
The atomic fountain is accurate to one second in 30 million years.
Just curious: How do you judge one of the most accurate clocks in the world? Just stare at the dial for 30 million years, waiting for it to lose a second? It might be boring, and even then, would you use that "Rolex" you bought on the Hong Kong street for reference? No. In fact, these error rates are not rooted on observation, but in calculations based on physicists' understanding of the remaining errors. "Scientists are capable of evaluating the clocks and predicting error all by themselves, without referring it to something more accurate," Collier Smith, a public affairs specialist at NIST, told us in 1998. "By going back to first principles, they can determine what the uncertainties are."
Having forgotten to score a Rolex last time around in Hong Kong, we'll take him at his word.
An alarming clock
Surprisingly, one second lost in 30 million years may only be a steppingstone toward the mercury ion clock. Since ions are easier to contain than atoms, they stay put, eliminating problems with the Doppler effect. With the mercury ion clocks, "We're certain we can do 100 times as well as existing clocks," Sullivan said, with potential for a 1,000-fold improvement over NIST-7.
Could the mercury ion clock mark the conclusion the clockmakers' long quest for accuracy? Apparently so. Sullivan says an error of one part in 1018 is "a kind of ultimate." Had such a clock been started when the universe began, it would now have about two seconds of error.
Based on the history of progress (since 1950, reference clocks have grown 1-million times more accurate), Sullivan expects that unimaginably exactitudinous goal will be reached, but, "It's going to take a while -- there are a lot of hefty challenges in the way."
How has timekeeping changed over time?
Megan Anderson, project assistant; Terry Devitt, editor; S.V. Medaris, designer/illustrator; David Tenenbaum, feature writer; Amy Toburen, content development executive