EP2220541B1 - Atomic clock regulated by a static field and two oscillating fields - Google Patents

Abstract

An atomic clock including a mechanism applying both a static magnetic field and two oscillating magnetic fields, all mutually perpendicular, in a magnetic shield. The amplitudes and frequencies of the oscillating magnetic fields may be chosen so as to annihilate energy variations between sub-transition levels of excited atoms and to reinforce a clock output signal, and with low sensitivity to defects in regulation.

Description

The subject of this invention is an atomic clock set or dressed by two oscillating fields and a static field which are applied in a shield.

Atomic clocks comprise an often alkaline gaseous medium, a device for exciting the atoms of this gas such as a laser, able to pass them to higher energy states, and a means for measuring a frequency signal emitted by atoms returning to the usual energy level, using the photons from the laser.

The frequency of the photons restored by the gas is defined by the formula ν = ΔE / h, where ν is the frequency, ΔE the difference between the energy levels and h the Planck constant, equal to 6, 63x10 ^-34 Js Il It is known that this frequency is very stable and can therefore serve as a unit of reference for time. However, this is no longer true when one considers the Zeeman structure of matter: the energy levels then appear as composed of sub-levels corresponding to slightly different states, which are distinguished by their magnetic quantum number m, 0 for a reference state of the energy level and -1, -2, etc. or +1, +2, etc. for the others. This is illustrated by the figure 1 in the case of element ⁸⁷ Rb, whose decomposition of the first two levels of energy (angular moments F = 1 and F = 2) has been figured.

Energy levels are sensitive to the surrounding magnetic field. This sensitivity is low (of the second order) for the sub-level to the magnetic number equal to 0, but much stronger (of the first order) for the other sub-levels: the transitions made from or up to them produce photons whose the frequency is variable and can not be used as a reference, and only the portion of the signal corresponding to the transition between the two sub-levels of zero magnetic number is used for the measurement, which affects its quality. The reference frequency given by the clock is then the frequency of the hyperfine transition considered in the gas fo = E _o / h, where E ₀ is the energy difference between the sub-levels at m = 0 of the two states ( F = 1 and F = 2 in the example of the figure 1 ).

Magnetic shielding around the clock is therefore used to reduce external disturbances, and the application of a constant magnetic field in the shield to separate the sub-levels, failing to guarantee a zero magnetic field. If the operation of the clock is made more stable, the sub-levels then being immobile and therefore well defined, the disadvantage of undergoing a dispersion of frequencies and having to be satisfied with a weakened signal is not avoided.

With the invention, efforts are made to perfect the existing atomic clocks by making them work in a zero magnetic field in order to focus the sub-levels at the same energy value and obtain a signal with a much sharper peak.

où H₀ est l'intensité du champ statique, T le temps de relaxation des atomes, ω la pulsation du champ oscillant, et γ le moment gyromagnétique. Les différences d'énergie ΔE entre les sous-niveaux d'un même niveau deviennent alors toutes nulles dans chaque niveau, les photons restitués par le gaz correspondent tous à la différence d'énergie E₀, l'état de la matière de la figure 2 étant alors obtenu : tout se passe comme si un champ résultant (fictif) nul existait.It has been proposed to involve non-zero magnetic number sub-levels in the useful signal by suppressing the dispersion of energies between sub-levels that the static field causes. The article of Harp "Modified Zeeman hyperfine spectra observed in H1 and Rb87 ground states interacting with a nonresonant RF field", Physical Review Letters, Volume 24, Number 16, April 20, 1970, pages 861-864. , reveals that the effect of the static magnetic field can be annihilated for the excited atoms by applying an oscillating field which is perpendicular to it, provided to respect the double inequality $$H_{}$$$${}_0<<\frac{1}{T\mathrm{.}\mathrm{\gamma }}<<\frac{\mathrm{\omega }}{\mathrm{\delta }}$$

where H ₀ is the intensity of the static field, T is the relaxation time of the atoms, ω the pulsation of the oscillating field, and γ is the gyromagnetic moment. The energy differences ΔE between the sub-levels of the same level then all become zero in each level, the photons restored by the gas all correspond to the energy difference E ₀ , the state of the material of the figure 2 being then obtained: everything happens as if a null (fictitious) field existed.

This implies, however, to respect specific ratios between the intensity and the frequency of the oscillating field to obtain this effect; however a great fineness of adjustment is necessary, a disturbance even weak leaving a field fictitious residual fictitious that prevents benefit from this discovery.

où H _ω et ω sont une intensité et une fréquence du second champs magnétique oscillant, et γ est ut rapport gyromagnétique, est à un extrémum.The invention is based on an improvement, according to which a second oscillating field is added to the device. It relates to an atomic clock as defined in claim 1. In a preferred variant, a Bessel function of the first kind of a report report ${{}^{\mathit{\gamma H\omega }}/}_{\omega },$

where H _ω and ω are an intensity and a frequency of the second oscillating magnetic field, and γ is a gyromagnetic ratio, is at an extremum.

The document WO-A-2005/081 794 describes an atomic clock according to the preamble of the independent claim.

The invention will now be described with reference to the figures, the figure 1 already described and the figure 2 already described illustrate two diagrams of the energy levels of a chemical element used in an atomic clock, the figure 3 is a schematic view of the clock, and the figure 4 is a graphic representation of functions illustrating the effect of the invention.

The figure 3 is discussed. The heart of the clock is a cell 1 filled with an alkaline gas. An exciter 2 transmits energy to this gas in the form of a polarized photon flux passing through a circular polarizer 3. The exciter may also be a microwave field, for example. It will then be necessary anyway to inject a light beam (for example of laser) to detect the resonances of the gas. A photodetector 4 collects the light energy restored by the gas of the cell 1 and transmits a signal to a counting device 5. A frequency separator 6 collects the signal at the output of the counting device 5 and transmits its results to an operating device 7 of the clock and a servo-control device 8 which controls the exciter 2 as well as means for applying magnetic fields 9 and 10. These emit magnetic fields at radio frequencies of pulsations noted Ω and ω, which are perpendicular to each other and of direction dependent on the polarization (for example perpendicular to the light rays emitted by the exciter 2 in the case of a circular polarization). These oscillating magnetic fields are applied in a magnetic shielding 11 which includes the cell 1 and the magnetic field application means 9 and 10.

sin α respectivement dans la direction du champ de radiofréquences et la direction perpendiculaire à ce champ, J ₀ étant une fonction de Bessel de première espèce et α étant l'angle entre le champ statique et le champ de radiofréquences. Quand les champs sont perpendiculaires entre eux, la première composante disparaît et $H_0^{ʹ}=H_0\mathrm{.}{J}_0\left({{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)\mathrm{.}$

Or la fonction de Bessel J ₀ de première espèce est comprise entre -1 et +1 et s'annule en au moins un point. Une représentation graphique en est faite à la figure 4 (courbe 12). Des choix judicieux du rapport ${{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}$

permettent donc d'annuler le champ magnétique résultant fictif H'₀=0 ; un de ces rapports est égal à 2,4. On voit toutefois que la pente de la fonction est importante, et qu'une variation de 10% du réglage produit un champ magnétique résultant dont l'intensité est d'environ 0,1H₀, ce qui est excessif. C'est pourquoi le second champ oscillant est ajouté. Il est orthogonal au premier champ de radiofréquences et au champ statique, sa pulsation est Ω et son intensité est H_Ω. La pulsation Ω satisfait aux inégalités suivantes $H_0\ll \frac{1}{T\mathrm{.}\mathrm{\gamma }}\ll \frac{\mathrm{\Omega }}{\mathrm{\gamma }}\ll \frac{\mathrm{\omega }}{\mathrm{\gamma }},$

c'est-à-dire que le second champ de radiofréquences a les mêmes effets que le premier sur le champ statique mais que sa pulsation est bien moindre que celle du premier champ de radiofréquences. De plus, il faut remarquer que les fréquences des deux champs oscillants ne doivent pas être trop grandes : il convient qu'elles ne dépassent pas (f o/4) environ, où fo déjà mentionnée est la fréquence de la transition hyperfine et correspondant au changement de niveau d'énergie des atomes dans le gaz. Le premier champ magnétique oscillant subit aussi alors des modifications qui se traduisent par une atténuation de son amplitude H_Ω par la fonction de Bessel. Le système composé par les deux champs de radiofréquences et le champ magnétique statique est donc équivalent à un champ de radiofréquences fictif $H_{\mathrm{\Omega }}\mathrm{.}{J}_0\left({{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)•\cos \left(\mathrm{\Omega }t\right)$

et un champ statique fictif $H_0^{ʹ}=H_0\mathrm{.}{J}_0\left({{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right),$

et ce système est lui-même équivalent, d'après ce qui précède, à un champ statique fictif H ₀" atténué par la contribution des deux champs de radiofréquences, d'intensité $${H_0}^{\mathrm{..}}={H_0}^{\mathrm{.}}{J}_0\left(\frac{\mathrm{\gamma }{H}_{\mathrm{\Omega }}\mathrm{.}{J}_0\left({{}^{\mathrm{\gamma }{H}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)}{\mathrm{\Omega }}\right)=H_0{J}_0\left({{}^{\mathrm{\gamma }{H}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)J_0\left(\frac{\mathrm{\gamma }{H}_{\mathrm{\Omega }}\mathrm{.}{J}_0\left({{}^{\mathrm{\gamma }{H}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)}{\mathrm{\Omega }}\right)\mathrm{.}$$

We return to the theoretical explanation of phenomena. The combination of a static magnetic field of intensity H ₀ and a field of radio frequencies H _ω intensity and pulsation ω complying with the conditions indicated above has an equivalent effect on the atoms to that of a fictitious static magnetic field of intensity H ' ₀ whose components are equal to H _{0 ·} cos α and ${\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H</figure-callout>}}_0\mathrm{.}{\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left({{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)\mathrm{.}$

sin α respectively in the direction of the radiofrequency field and the direction perpendicular to this field, J ₀ being a Bessel function of the first kind and α being the angle between the static field and the radiofrequency field. When the fields are perpendicular to each other, the first component disappears and $H_0^{\text{'}}={\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H</figure-callout>}}_0\mathrm{.}{\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left({{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)\mathrm{.}$

Now the Bessel J ₀ function of the first kind is between -1 and +1 and vanishes in at least one point. A graphic representation is made at the figure 4 (curve 12). Judicious choices in the report ${{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}$

thus allow to cancel the fictitious resulting magnetic field H ' ₀ = 0; one of these ratios is equal to 2.4. It can be seen, however, that the slope of the function is important, and that a variation of 10% in the adjustment produces a resulting magnetic field whose intensity is about 0.1H ₀ , which is excessive. This is why the second oscillating field is added. It is orthogonal to the first radio frequency field and the static field, its pulsation is Ω and its intensity is H _Ω . The Ω pulse satisfies the following inequalities $H_{}$${}_0"\frac{1}{T\mathrm{.}\mathrm{\gamma }}"\frac{\mathrm{\Omega }}{\mathrm{\gamma }}"\frac{\mathrm{\omega }}{\mathrm{\gamma }},$

that is, the second radiofrequency field has the same effects as the first on the static field but that its pulsation is much less than that of the first field of radiofrequencies. In addition, it should be noted that the frequencies of the two oscillating fields should not be too large: they should not exceed ( fo / 4) approximately, where fo already mentioned is the frequency of the transition hyperfine and corresponding to the change of energy level of the atoms in the gas. The first oscillating magnetic field then also undergoes modifications which result in an attenuation of its amplitude H _Ω by the Bessel function. The system composed of the two radiofrequency fields and the static magnetic field is therefore equivalent to a fictitious radiofrequency field $H_{\mathrm{\Omega }}\mathrm{.}{\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left({{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)•\cos \left(\mathrm{\Omega }t\right)$

and a fictional static field $H_0^{\text{'}}={\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H</figure-callout>}}_0\mathrm{.}{\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left({{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right),$

and this system is itself equivalent, from what precedes, to a fictitious static field H ₀ "attenuated by the contribution of the two radiofrequency fields, of intensity $${H_{}}^{}$$$${{}_0}^{\mathrm{..}}={H_0}^{\mathrm{.}}{\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left(\frac{\mathrm{\gamma }{H}_{\mathrm{\Omega }}\mathrm{.}{\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left({{}^{\mathrm{\gamma }{H}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)}{\mathrm{\Omega }}\right)={\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H</figure-callout>}}_0{\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left({{}^{\mathrm{\gamma }{H}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right){\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left(\frac{\mathrm{\gamma }{H}_{\mathrm{\Omega }}\mathrm{.}{\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left({{}^{\mathrm{\gamma }{H}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)}{\mathrm{\Omega }}\right)\mathrm{.}$$

en fonction de ${{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\Omega }}}/}_{\mathrm{\Omega }}$

(courbe 13) _: $H_0^{,,}$

s'annule une première fois pour un rapport ${{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\Omega }}}/}_{\mathrm{\Omega }}=6,0.$

Cette valeur dépend de celle de $J_0\left({{}^{\mathrm{\gamma }\mathit{H}\mathrm{\omega }}/}_{\mathrm{\omega }}\right),$

qui, dans le cas présent, a été choisie à 3,8, c'est-à-dire un extrémum de la fonction de Bessel de la courbe 12. En se plaçant ainsi, on supprime la sensibilité de ${\mathrm{H}}_{\mathrm{0}}^{,,}$

aux variations de $\left({{}^{\mathrm{\gamma }\mathit{H}\mathrm{\omega }}/}_{\mathrm{\omega }}\right),$

ce qui stabilise son réglage. La sensibilité de ${\mathrm{H}}_{\mathrm{0}}^{,,}$

aux variations de ${{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\Omega }}}/}_{\mathrm{\Omega }}$

reste toutefois du premier ordre, mais elle est significativement atténuée par rapport à ce que l'on obtient avec un seul champ de radiofréquences, comme la comparaison des courbes 12 et 13 le montre, puisque la pente aux croisements de l'axe des abscisses (aux ordonnées nulles) est réduite d'un facteur qu'on peut démontrer égal à ${\left[J_0\left({{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)\right]}^2\mathrm{.}$

Une variation de 10% de ${{}^{\mathrm{\gamma }\mathit{H}\mathrm{\Omega }}/}_{\mathrm{\Omega }}$

autour de la valeur de 6,0 induit un champ fictif $H_0^{,,}={\left(J_0\left(3,8\right)\right)}^2\mathit{x}0,1x{H}_0^{,,}=0,016H_0^{,,}$

au lieu de 0,1 1 H ₀ avec un seul champ de radiofréquences : la sensibilité aux défauts de réglage est réduite de 84%. Par ailleurs, $J_0\left({{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)$

étant à un extrémum, $H_{\mathrm{0}}^{,,}$

n'est pas sensible aux variations de ce rapport autour de ce point de réglage. Il serait évidemment possible de placer le rapport ${{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}$

à d'autres extrémums de la fonction de Bessel, ce qui aurait donné une sensibilité aux défauts de réglage encore plus faible.This field can be canceled by specific settings of each of the radio frequency fields. The figure 4 shows an example of the evolution of the report ${H_{}}^{}$${{}_0}^{\mathrm{.}}/{\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H</figure-callout>}}_0^{\mathrm{..}}$

in terms of ${{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\Omega }}}/}_{\mathrm{\Omega }}$

(curve 13) _: ${\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H</figure-callout>}}_0^{,,}$

cancels a first time for a report ${{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\Omega }}}/}_{\mathrm{\Omega }}=6,0.$

This value depends on that of ${\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left({{}^{\mathrm{\gamma }\mathit{H}\mathrm{\omega }}/}_{\mathrm{\omega }}\right),$

which, in this case, was chosen at 3.8, that is to say an extremum of the Bessel function of curve 12. By placing itself thus, the sensitivity of ${\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H</figure-callout>}}_{\mathrm{0}}^{,,}$

variations of $\left({{}^{\mathrm{\gamma }\mathit{H}\mathrm{\omega }}/}_{\mathrm{\omega }}\right),$

which stabilizes its setting. The sensitivity of ${\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H</figure-callout>}}_{\mathrm{0}}^{,,}$

variations of ${{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\Omega }}}/}_{\mathrm{\Omega }}$

remains, however, of the first order, but it is significantly attenuated with respect to what is obtained with a single radio frequency field, as the comparison of the curves 12 and 13 shows, since the slope at the crossings of the abscissa ( to the null ordinates) is reduced by a factor that can be shown to be equal to ${\left[{\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left({{}^{\mathrm{\gamma }{\mathit{<figure-callout\; id="2"\; label="H\; \omega \; \omega "\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H\; </figure-callout>}}_{\mathrm{\omega }}}{}_{}/}_{\mathrm{\omega }}\right)\right]}^2\mathrm{.}$

A variation of 10% of ${{}^{\mathrm{\gamma }\mathit{H}\mathrm{\Omega }}/}_{\mathrm{\Omega }}$

around the value of 6.0 induces a fictitious field $H_{}^{}$${}_0^{,,}={\left({\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left(3,8\right)\right)}^2\mathit{x}0,1x{\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H</figure-callout>}}_0^{,,}=0,016{\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H</figure-callout>}}_0^{,,}$

instead of 0.1 1 H ₀ with a single radio frequency field: the sensitivity to adjustment faults is reduced by 84%. Otherwise, ${\mathrm{<figure-callout\; id="0"\; label="J"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">J</figure-callout>}}_0\left({{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}\right)$

being at an extremum, ${\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">H</figure-callout>}}_{\mathrm{0}}^{,,}$

is not sensitive to variations of this ratio around this set point. It would obviously be possible to place the report ${{}^{\mathrm{\gamma }{\mathit{H}}_{\mathrm{\omega }}}/}_{\mathrm{\omega }}$

at other extremums of the Bessel function, which would have given a sensitivity to the defects of setting even lower.

Experimental settings may differ slightly from the theoretical settings. It is possible to perform them by exploiting information given by a low frequency sinusoidal magnetic field υ (much less than 1/2 πT) and collinear with H ₀ . This field induces disturbances on the signal delivered by the clock at frequencies fo ± υ. It will then be possible to quantify the sensitivity of the signal delivered by the atomic clock to the variations of the static magnetic field by a synchronous detection at the frequency of this disturbance. An interesting operating point can be obtained by first adjusting the amplitude H _ω of the field at the highest frequency (ω / 2π) to a maximum of sensitivity of the static field H ₀ . The other radio frequency field H _Ω will then be added and adjusted to obtain a minimum sensitivity of H ₀ .

The servo-control device 8 can be used for a continuous adjustment of the amplitude of the second radio-frequency field according to this principle of keeping a minimum of sensitivity of the signal delivered by the clock.

The single exciter may be a photon flux such as a laser flux emitted for example by a laser diode or a lamp. The gaseous element may consist of ⁸⁷ Rb, ¹³³ C _s , with optional mixing with a buffer gas. The material of cell 1 may consist of a glass such as Pyrex (trademark). The means for applying the magnetic fields 9 and 10 may consist of triaxial coils, or of three monoaxial coils concentric with each other. The photodetector 4 can be of any kind measuring a flow of photons at the output of the cell 1. These photons must be polarized for example by polarizers adjoining the exciter. Servoing is accomplished by any known hardware including a computing unit. The coils are driven by current. The excitation at the resonance frequency is accomplished by amplitude modulation of the laser diode at the frequency f _0/2 , or by a microwave cavity resonant at the frequency f ₀ . An exciter comprising two lasers, the frequency deviation is f ₀ can also be envisaged.

The shielding being then particularly effective, all the sub-levels become equivalent since the field is zero. It is then possible to use other gases than those usually employed in atomic clocks (alkaline gases), in particular gases whose hyperfine structure of their atoms does not have zero angular momentum sub-levels, such as ³ He.

The magnetic shield 11 may consist of n metal cylinders imbricated, possibly with a soft iron cylinder. In a particular case where the element ⁸⁷ Rb was used, the wavelength of the laser photons was 780nm, a quarter wave plate imposed a left circular polarization to the incident photons, the magnetic shield 11 consisted of four cylinders of μ concentric metal and a soft iron cylinder outside, the magnetic field H ₀ was 100 microgauss in the main axis, γ was equal to 670 kilohertz per gauss, and radio frequencies were 3 kilohertz and 20 kilohertz to respective magnitudes of 27 and 114 milligauss in order to impose the previously identified conditions of process validity.

Claims (7)

1. Atomic clock comprising a cell (1) filed with a gas, an exciter (2) of the gas to make its atoms jump to a higher energy level, a detector (4) to collect a light signal passing through the gas, a magnetic shield (11) around the cell, and means for applying magnetic fields (9, 10), including a static magnetic field, characterised in that the means for applying magnetic fields (9, 10) are arranged to also apply two oscillating magnetic fields, perpendicular to each other and to the static magnetic field, so that a Bessel function of the first kind of a ratio $\gamma {\mathit{H}}_{\mathrm{\Omega }}/\mathrm{\Omega },$

   

   where H _Ω and Ω are an intensity and a frequency of one of the oscillating magnetic fields, which has a frequency lower than the other, and γ is a gyromagnetic ratio, is equal to 0.
2. Atomic clock according to claim 1, characterised in that it comprises means for regulating either the intensity or the frequency of the oscillating magnetic fields.
3. Atomic clock according to claim 1, characterised in that a Bessel function of the first kind of a ratio ${{}^{\mathit{\gamma H\omega }}/}_{\omega },$

   

   where H _ω and ω are an intensity and a frequency of said other oscillating magnetic fields, and γ is a gyromagnetic ratio, is at an extremum.
4. Atomic clock according to any of claims 1 to 3, characterised in that the means for applying magnetic fields comprise at least three concentric monoaxial coils.
5. Atomic clock according to any of claims 1 to 3, characterised in that the means for applying magnetic fields comprise at least one triaxial magnetic coil.
6. Atomic clock according to any of the preceding claims, characterised in that the gas is chosen among alkali gases and helium 3.
7. Atomic clock according to any of the preceding claims, characterised in that the oscillating magnetic fields have frequencies at the most equal to the quarter of a hyperfine transition frequency measured by the clock.

Images