The Hyperfine Pendulum

Louis Essen, Jack Parry, and the First Caesium Atomic Clock

For three thousand years, every clock on Earth was judged against a single master oscillator—the daily rotation of the planet on its axis relative to the fixed stars and the Sun, defining one second as $1/86,400$th of a mean solar day. By the late 1930s, however, temperature-controlled piezoelectric quartz-crystal ring oscillators installed at the Royal Greenwich Observatory and the Physikalisch-Technische Reichsanstalt had grown stable enough to expose a startling fact: the Earth itself is a terrible precision clock. Driven by seasonal winds, melting polar ice caps, molten iron currents in the outer core, and the tidal friction of the Moon, the spinning globe speeds up and slows down unpredictably by more than a millisecond per day—a fractional frequency drift of $10^{-8}$.

To measure time beyond one part in a hundred million, physicists had to sever timekeeping from planetary mechanics and lock their oscillators to an invariant property of matter. As Isidor Isaac Rabi suggested to his Columbia University colleagues in 1945 after inventing the molecular beam magnetic resonance technique, every atom of a given isotope is identical across the universe, and the quantum energy gap $\Delta E$ between two internal hyperfine spin states of an isolated atom in zero magnetic field is fixed by fundamental constants of nature.

At the National Physical Laboratory (NPL) in Teddington, outside London, the metrologist Louis Essen—who had already built the stable Essen quartz ring clock and remeasured the speed of light to $299,792.5\text{ km/s}$ using a resonant microwave cavity—teamed up with Jack V. L. Parry in 1953 to build a working atomic frequency standard. For their pendulum, they chose the heavy alkali metal Caesium-133 ($^{133}\text{Cs}$). A neutral Caesium-133 atom has fifty-five electrons, fifty-four of which sit in closed inner shells while a single unpaired valence electron orbits outside in the $6S_{1/2}$ ground state. The magnetic dipole moment of that lone outer electron interacts with the magnetic moment of the caesium nucleus (nuclear spin $I = 7/2$), splitting the ground state into two closely spaced hyperfine energy levels ($F = 3$ and $F = 4$) separated by a microwave transition frequency in the X-band radar range:

In June 1955, Essen and Parry brought their five-foot horizontal copper vacuum cylinder, nicknamed "Caesium I," into operation on a laboratory table in Teddington. At one end of the evacuated tube, a miniature electric oven heated metallic caesium to $100^\circ\text{C}$, spraying a thin, collisionless beam of neutral caesium atoms down the axis of the chamber at 200 meters per second. Because the atoms traveled through high vacuum without bumping into gas molecules or container walls, their internal energy levels remained unperturbed by thermal collisions.

First, the atomic beam passed through the sharply inhomogeneous gap of a polarizer Stern-Gerlach magnet ("Magnet A"), which deflected atoms in the $F = 3$ and $F = 4$ states along converging paths into a central microwave chamber. Instead of exposing the atoms to a single long microwave field—which suffers from Doppler phase shifts and magnetic non-uniformities—Essen and Parry implemented Norman Ramsey's 1949 method of separated oscillatory fields: a U-shaped copper waveguide spaced forty-seven centimeters apart that hit the flying atoms with two brief, phase-coherent microwave pulses at the entrance and exit of a magnetically shielded drift region.

After leaving the second microwave cavity, the atoms passed through a second deflecting analyzer magnet ("Magnet B") arranged so that only atoms that had undergone a quantum spin flip from one hyperfine state to the other were steered onto a hot tungsten ribbon wire at the far end of the tube. On contact with the glowing tungsten, each refocused caesium atom surrendered its outer electron and boiled off as a positive ion into a mass spectrometer and micro-ammeter. When Essen tuned his external quartz crystal oscillator and frequency-multiplier chain across the microwave band, the ion current spiked sharply at resonance with a quality factor $Q$ exceeding thirty million, allowing the clock to measure frequency to one part in $10^{10}$—equivalent to losing one second in three hundred years.

Between 1955 and 1958, Essen and Parry collaborated across the Atlantic with astronomer William Markowitz at the U.S. Naval Observatory in Washington, D.C., who used a dual-rate moon camera to compare the orbital motion of the Moon (Ephemeris Time) against time radio broadcasts ticked off by the Teddington caesium resonator. In October 1967, the Thirteenth General Conference on Weights and Measures officially ended astronomy's multi-millennial stewardship of time, defining one international second as exactly 9,192,631,770 cycles of the Caesium-133 microwave transition measured at Teddington.