A team of SYRTE (Systèmes de ----
Référence Temps Espace, UMR-CNRS 8630) from Paris Observatory has just established the best upper limit, for a laboratory experiment, on a possible variation with time of the fine structure constant alpha = qe2/hc and gyromagnetic factor of the proton.

This limit, which improves by a factor 20 previous measurements, is based on the determination of the oscillation frequency of cesium and rubidium, with a relative precision of about 10-15 in atomic fountains.

Variation of fundamental constants

The idea of the variation of fundamental constants over cosmological times dates back to Dirac in 1937 : he noticed a curious coincidence between very large numbers, the relationship between the age of the Universe and the atomic time scale on the one hand, and the ratio between the electromagnetic and gravity forces on

the other hand. A simple way to harmonize these forces would be to vary the constant of gravitation by 10-10 per year (gravity would then have been stronger in the past, and comparable to the electromagnetic force). If this idea has failed against observations, many other works were developped, which predict weak variations of the fundamental constants, temporal and spatial as well, allowing to unify gravity

with other forces (theory of Kaluza-Klein, theory of the superstrings, etc.). Constraints already exist on the variation of fundamental constants, and in particular on the fine structure constant alpha = qe2/hc. These constraints can be

geological, as those of the natural nuclear reaction which occurred in Oklo (Gabon) 2 billion years ago (the isotopic ratios of the reaction products

measured today show that the constants were the same within 10-7, see Fujii, 2002), or astronomical (observation of absorption lines in front of very

remote quasars, to test the values of the constants when the Universe had only 20% of its age, i.e. 12 billion years ago, cf Murphy et al 2001, Ivanchik et al, 2002 ; or observations of radioactive decay in meteorites of the solar system, Olive et al, 2002), and very precise experiments in laboratory, like

those presented here. Until now, only Webb et al, 2001, 2002 found a relative variation of alpha, about 10 billion years ago, of (-0.57 +/- 0.10) x 10-5. To learn more, see the review article from Jean-Philippe Uzan.

Atomic fountains

The atomic fountain clocks have made important progress in measurement accuracy of times and frequencies, thanks to the technique of "cooling"

of the atoms. Usually, the atoms are animated at ambient temperature by a motion at random velocity, which disturbs the measurement of their frequency (Doppler effect,relativistic time dilation). The atoms can be stopped and cooled, by the interaction with light beam, provided by a laser. Moreover, the atoms thus

trapped can remain a long time in the measuring instrument, which improves the precision (the longer the measurement, the more precise it is). In an atomic fountain, one launches vertically a cloud of atoms, cooled by lasers. The atoms then pass in a microwave cavity which excites their hyperfine transition (see more details). The factor limiting the parking time of the atoms in the measuring instrument, is the gravity, which brings back the atoms to the bottom (the

movement of the atoms is similar to a water fountain). The following step is to make the experiment in space, without gravity, cf the pharao project. By comparing the fundamental frequencies (hyperfine structure) of the cesium 133Cs and rubidium 87Rb, during 5 years, the team of Paris Observatory was able to place an upper limit on the relative variation of alpha of 7 10-16 per year. For that, three atomic fountains were compared between them,

via an hydrogen maser (cf Figure 1). vara-f2.jpg Figure 1 - Device used by the SYRTE for the experiment : a single signal of 100MHz is sent by the hydrogen maser, and is distributed to the three microwave synthetizers of cesium (FO1, FOM, DFcs), and to the rubidium fountain (Rb clock). Click on the figure to enlarge vara-f3.gif Figure 2 - Relative measurements of the frequencies of rubidium 87Rb and cesium 133Cs spanning over 57 months. The 1999 measurement of the frequency of rubidium (6 834 682 610.904 333 Hz) is by convention taken for reference. A linear fit of the data gives a relative variation of the ratio of the frequencies of (0.2 +/-7.0) X 10 -16 per year. The dotted lines correspond to an uncertainty of one sigma. Click on the figure to enlarge Today this upper limit is the best which can be reached in laboratory (cf Figure 2), and constrains the variations of alpha at the current epoch. In the near future, this technique will allow to gain an order of magnitude, especially with the ESA experiment ACES which will be embarked on the international space station in 2006. A recent result also based on ultra-precise atomic frequency measurement (comparison between an optical transition of mercury ion 199Hg+ and the hyperfine frequency of 133Cs) constrains another combination of fondamental constants (Phys. Rev. Lett. 90, 150802). To learn more : visit the site of SYRTE Atomic Fountain