Michigan and the First Atomic Clock
Randall Laboratory of Physics, University of Michigan, Ann Arbor, MI 48109-1040
The first atomic clock has its origins with the research done at the University of Michigan and by its graduates over the years 1912-1948.
Although the current (2012) standard of time/frequency is based on a 9 GHz microwave transition in cesium, the first device that one could really call an atomic clock was designed and built in 1948 by Harold Lyons (UM physics PhD 1939) at the National Bureau of Standards. This clock, which predates the first cesium clock by four years, was based on the 24 GHz microwave transition in the ammonia (NH3) molecule. The existence of this transition was predicted by David Dennison in 1932; it was discovered in 1934 when the Michigan experimentalists Neil Williams and Claud Cleeton developed the magnetron tubes to produce the necessary radiation to perform the first-ever microwave spectroscopy.
But the story begins even earlier.
The theoretical understanding of the ammonia molecule rests on the experimental work started by Harrison Randall when he returned to Michigan in 1911 after working for a year on the infrared spectroscopy of atoms in Paschen’s laboratory in Tübingen. Randall built his own infrared spectrometer and initially studied some unresolved questions on the barium atom.
Then in 1913 from Berlin, Eva von Bahr published spectra arising from vibrational transitions in HCl, spectra with indications of a fine structure that was conjectured to arise from molecular rotation. Randall, whose spectrometer had a higher resolution than von Bahr’s, realized that he could resolve the question of whether molecular rotation was quantized in the same way that Bohr had assumed for electrons in his 1913 theory of the atom. Randall then directed the Michigan infrared effort toward molecules, first to diatomic molecules and then to small polyatomic molecules including NH3. The issue of quantized rotation was soon settled by Elmer Imes whose thesis experiments on HCl proved definitive , but it took longer to understand ammonia. By 1929, Barker had finally established the dimensions of the NH3 pyramid, and subsequent measurements of its spectra by Dennison and Hardy, and later by Wright and Randall, determined values for the doublet splitting of the vibrational states.
A theoretical calculation by Dennison and Uhlenbeck (1932) confirmed the dimensions and energies in this double minimum problem, however their paper does not explicitly mention the possibility of observing a microwave transition in NH3. Rather it appears that Dennison was the one who made the suggestion in 1932 that UM Professor Neil Williams and his students look for the absorption of 24 GHz radiation.
Photo of Dennison explaining to Ray Van Ausdal the focusing and state selecting action of a quadrupole field for focusing a beam of NH3 molecules. Dennison’s theory was the basis for the first Maser as built by Gordon, Zeiger and Townes in 1954.
In the early 1930s, no one had been able to produce controllable frequencies higher than 8 GHz with vacuum tubes, but Neil Williams had been engaged with the physics and engineering of vacuum tubes since 1924-5 during his time on leave at the General Electric laboratories in Schenectady. There Williams had worked under the guidance of A. W Hull, a pioneer in the development of the magnetron as a vacuum tube for generating very high frequencies. After his return from Schenectady, Williams did research at Michigan on aspects of electron emission but he did not particularly concentrate on high frequency generation.
Fortunately, Williams responded with enthusiasm to Dennison’s suggestion: He set Claud Cleeton, his graduate student who was already well-versed in research, to the task of building a magnetron to generate radiation of appropriately high frequency. Cleeton succeeded; he and Williams then built spectrometer to look at the absorption of microwaves by a gas of ammonia molecules. The successful results of this experiment, the first microwave spectroscopy ever done, were reported in 1934.
(Cleeton and Williams continued their development of magnetrons and by 1936 had pushed the limit for high frequency generation by vacuum tubes to 45 GHz, a new record; this led to Cleeton getting a wartime assignment to do radar development at the Naval Research Laboratory in Washington, DC )
Harold Lyons was a graduate student in the Michigan Physics Department over the years 1934-39 doing a PhD thesis under the direction of Professor Ora Duffendack (who had also been in Schenectady in 1924-5). Lyons graduated in 1939 and went from Michigan to work at the Naval Research Laboratory for two years; he then joined the National Bureau of Standards and was soon chosen as chief of the Microwave Standards Section. Lyons’s familiarity with frequency standards together with the post-war availability of microwave hardware led him to build a clock based on the 24 GHz transition in ammonia. This clock consisted of a quartz crystal oscillator that was stabilized to the absorption frequency of a 8-meter-long tube filled with ammonia gas. The signal from the oscillator was then divided down to a frequency appropriate to drive an ordinary electric clock that was mounted on the front of the apparatus. This ammonia clock, the first atomic clock, was built by Lyons and his staff in the NBS Laboratories; it became operational in August of 1948,
It turns out that pressure broadening and power dependence of the 24 GHz transition limited the 1948 ammonia clock’s precision to parts in 10E+8, this not really improving on carefully-built quartz oscillator frequency standards. For increased precision it seemed promising (as suggested by Rabi and others) to base a clock on transitions observed in atomic beam spectrometer. Several laboratories (including a collaboration between Lyons of NBS and Kusch of Columbia) began working on a cesium beam clocks, and some of these became operational in early 1953. By 1967 their improved precision (parts in 10E+12) and their reliability led to the adoption of the 9 GHz hyperfine transition in cesium as the international standard of frequency. At this writing (2012), the best cesium clocks approach the 10E+16 level of precision, but it appears that optical transitions in trapped beryllium ions may offer almost two orders of magnitude improvement for the standard of time/frequency.
Given the relatively close interactions among the Michigan graduate students and faculty in the mid-1930’s, it is quite likely that Lyons knew much about Cleeton’s thesis already in 1934; moreover, after leaving Michigan both men had overlapping time in the Naval Research Laboratory in Washington, DC. Both worked on radar systems during WWII and in the immediate postwar years. (I am now looking for correspondence or other evidence that they might have had discussions of the NH3 absorption during their time in the Washington area — evidence that would link Cleeton more directly than he already is to the development of the atomic clock.)
Comment: The term “atomic clock” is here regarded as describing a timekeeper or frequency standard that uses a resonance in either an atom or a molecule, either neutral or ionized, as a reference pendulum
It is interesting to note that as ammonia was being eclipsed as a frequency standard in 1952, its very large electric dipole moment made it the molecule of choice in 1954 when Gordon, Zeiger and Townes used it as the working molecule in the first maser