cryogenic fountain development at nist and inrim: preliminary...

0885–3010/$25.00 © 2010 IEEE

600 IEEE TransacTIons on UlTrasonIcs, FErroElEcTrIcs, and FrEqUEncy conTrol, vol. 57, no. 3, March 2010

Abstract—This paper describes the new twin laser-cooled Cs fountain primary frequency standards NIST-F2 and IT-CsF2, and presents some of their design features. Most sig-nificant is a cryogenic microwave interrogation region which dramatically reduces the blackbody radiation shift. We also present a preliminary accuracy evaluation of IT-CsF2.

I. Introduction

since the first calibration of International atomic Time (TaI) performed with a laser-cooled cs fountain in 1996, the accuracy of this type of standard has improved by nearly a factor of 10 [1], [2]. InrIM and nIsT have both contributed to TaI with their fountain IT-csF1 and nIsT-F1. Both fountains have achieved a fractional fre-quency accuracy in the 10−16 range.

The development of the second generation of laser-cooled fountains aims to realize a primary frequency standard with a fractional frequency accuracy approaching 1 × 10−16. In a typical accuracy evaluation of these standards, 5 biases are corrected and contribute to the total uncertainty budget: Zeeman effect, atom density shift, gravitational red shift, microwave leakage, and blackbody radiation shift (BBrs). Uncertainty caused by other shift sources are calculated to be negligible at the current accuracy level. To reach the target accuracy of 1 × 10−16, all the shifts mentioned above should be measured and corrected with a fractional frequen-cy uncertainty in the 10−17 range. although this accuracy level is already demonstrated for some of them, the evalu-ation of the BBrs remains a leading uncertainty source in most room-temperature frequency standards. Given the difficulty observed in the high-precision experimental deter-mination of its magnitude we decided to reduce its absolute value by cooling the fountain interaction region to liquid nitrogen temperature.

a seminal work by Itano [3] explains that the ac stark shift caused by Blackbody radiation can be approximated by an equivalent dc stark shift and, from this approxi-mation, calculates the BBrs as a function of the radia-tion temperature seen by the atoms during the interaction

time. Itano’s work [3] introduces the well-known theoreti-cal relation that connects the BBrs to the radiation tem-perature

Dnn

b e0 0

4

0

2

1=æèççç

öø÷÷÷ +

æèççç

öø÷÷÷

é

ëêê

ù

ûúú

TT

TT

, (1)

showing its strong dependence upon the temperature T as expected from the Planck radiation theory. higher-order terms are neglected because their magnitude is completely negligible at room temperature.

over the past few years, several scientific works [4]–[6] addressed the value and the related uncertainty of the BBrs bias in cs primary frequency standards, question-ing the level of uncertainty that can be eventually as-signed to the shift.

as a matter of fact, some discrepancies in the experi-mental value of β are reported in literature [4], [5], and only recent theoretical works [7], [8] have presented sat-isfactory agreement with the experimental value reported in [4] by simon et al. using dc stark shift measurements. on the other hand, also ignoring these discrepancies and using for β the value of 1.711 × 10−14 to determine the BBrs at the level of 5 × 10−17 at room temperature, the uncertainty required in the knowledge of the radiation temperature is 0.2K for the whole extension of the ram-sey interaction region: a challenging task.

The realization of a liquid nitrogen-cooled fountain, to-gether with some care in the design that will be presented in the next section, allows the reduction of the total BBrs to 1 × 10−16, with a similar large reduction in its uncer-tainty. Two such fountains developed at nIsT and InrIM time and frequency laboratories are undergoing prelimi-nary extensive tests.

II. The cryogenic Fountain structure

The design of the physics package presented here, along with preliminary experimental results, has been described previously in [9] and [10]. The design of the cryogenic fountain is derived from the original design of nIsT-F1 [11], which started operation in 1998, and now exhibits a fractional frequency accuracy of 4 × 10−16. In Fig. 1, a schematic drawing of the cryogenic fountain is presented. The fountain system is divided into 2 parts: an ultra-high-vacuum (UhV) region and a high-vacuum (hV) region.

Cryogenic Fountain Development at NIST and INRIM: Preliminary Characterization

Filippo levi, claudio calosso, davide calonico, luca lorini, Elio K. Bertacco, aldo Godone, Giovanni a. costanzo, Barbara Mongino, steven r. Jefferts, Thomas P. heavner,

and Elizabeth a. donley

Manuscript received May 20, 2009; accepted november 10, 2009. F. levi, c. calosso, d. calonico, l. lorini, E. K. Bertacco, and a.

Godone are with the Istituto nazionale di ricerca Metrologica, Torino, Italy (e-mail: [email protected]).

G. a. costanzo and B. Mongino are with the Politecnico di Torino, Torino, Italy.

s. r. Jefferts, T. P. heavner, and E. a. donley are with the Time and Frequency division, national Institute of standards and Technol-ogy, Boulder, co.

digital object Identifier 10.1109/TUFFc.2010.1453

The UhV region includes the trapping and detection vacuum chamber (at room temperature) and the ram-sey interaction zone (at cryogenic temperature), this zone is composed of 2 microwave cavities (state selection and ramsey interaction) and a ~1-m-long copper drift tube where the atoms fly over a ballistic trajectory: the vacuum level in this region is in the 10−8 Pa and makes the scat-tering with background gas negligible for the accuracy demand of the standard. In the trapping region, we have placed an auxiliary microwave cavity that can be used to perform the state selection as an alternative option to the use of the state-selection cavity placed in the nitrogen-cooled region. This option will be used to avoid the need of a vertical blasting laser beam; this will be the case if we implement the multiple launching technique [12]. The hV part contains the c-field solenoid, 2 compensation coils, heaters for the temperature stabilization of the interaction region, 3 thermistors, 2 inner magnetic shields, the cry-ostat, and a third shield. This last component is the only one at room temperature. despite the number of parts and cables present in this region, cryopumping yields a fairly good vacuum in the 10−6 Pa range.

A. The Ramsey Interaction Zone

The ramsey interaction region, that includes 2 micro-wave cavity and the drift tube, is not in a liquid nitrogen bath: the nitrogen is contained in a ~20-l cryostat and the cooling process takes place by conductive and radia-tive processes. To help this cooling process, 2 endcaps with good thermal contact are fixed at both end of the cryostat, and the cavity system sits on top of the lower one. a time constant of more than one week was measured when the cryostat was filled for the first time with liquid nitrogen. In absence of active control, the temperature fluctuates freely according to the amount of liquid nitro-gen present in the cryostat. In Fig. 2, the temperature profile of the ramsey interaction zone is shown for one week of operation.

601levi et al.: cryogenic fountain development

Fig. 1. drawing of the cryogenic fountain.

Fig. 2. Temperature profile of IT-csF1 interaction region during one week of operation. The peaks correspond to 5 n2 fillings of the cryostat during the working days.

at room temperature, the ramsey cavity frequency is calculated to be 31 Mhz lower than cs resonance. In the project stage, the final temperature of the cavity was ex-pected to be several degrees below the achieved value, so the cavity was manufactured to be resonant with cs at 80K, whereas the actual temperature of the cavity is around 89K. This means that the ramsey cavity is de-tuned almost by 1 Mhz with respect to the atomic reso-nance. The cavity is thus currently operated out of reso-nance, but despite the need for higher microwave power, this detuning strongly reduces the sensitivity to phase and power changes caused by the temperature oscillations.

Upon cooling the cavity, the linewidth changes from 550 khz to 250 khz, in fair agreement with the Gruneisen relation that connects the surface resistivity of the metals to the temperature (predicting a 2.5 ratio);

r µ ×æèçç

öø÷÷÷

- - -òTT s ds

e es sT

Q

Q4 5

0 1 1( )( ),

/ (2)

where Θ is the cu debye temperature and T is the metal surface temperature in kelvin.

The cryogenic part of the fountain contains also the c-field solenoid and 2 magnetic shields. The magnetic field is generated by a 1.5-m-long solenoid that generates a nomi-nal field of 2.5 × 10−4 T/a. special care was taken during the fountain design to avoid the development of thermal currents. however 2 cylindrical shields are placed in the cryogenic environment. The third magnetic shield instead is still under vacuum but at room temperature. Thermal insulation between the cryostat and the outer shield is provided by several layers of aluminized Mylar insulator.

B. The Optical Bench

The fountain is operated in the (1, 1, 1) geometry to avoid the need of a vertical trapping laser beam. six in-dependent laser beams are generated by a single high power (~500 mW) laser system, frequency locked to the F = 4, F ′ = X54 cs transition, see Fig. 3 for schemat-ics. The 6 beams are amplitude and frequency fine tuned through double pass acousto-optic-modulators (aoM) [13], to achieve the right frequency detuning, and then fiber coupled. The double pass aoM efficiency is steadily above 60%, measured from the aoM input to the fiber output. Each molasses beam deliver to the atoms ap-proximately 15 mW, the diameter of the beams is ap-proximately 25 mm. Two additional beams are generated independently: one for the detection region and one for the vertical beam used to blast the F = 4 atoms left after the state selection. an independent laser frequency locked on the F = 3 F ′ = 4 cs transition is used as a repumper. Two beams are generated from this laser; one is delivered to the trap and one to the detection region. The control of the aoM frequency and amplitude is done by means of a home-built dedicated electronic system [14].

To reduce the laser amplitude noise and to compensate polarization to amplitude noise conversion, active ampli-

tude control of each laser beam is implemented using the rF power in the aoM as actuator. For this purpose, a beam sampler is placed inside each collimator after a po-larization filter. In Fig. 4, the amplitude noise spectrum at the collimator output is represented in the open- and closed-loop configuration, together with the noise spectral density of the voltage reference and that of the detector; a noise reduction of more than 40 dB can be observed in the region of interest. The post-cooling amplitude ramp is also generated by means of this control loop.

C. Atomic Signal and Stability

In nIsT-F2 and IT-csF2, direct molasses loading in a lin ⊥ lin configuration is implemented. The optical molas-


Fig. 3. schematics of the optical bench of IT-csF2. aoM stands for double-pass acousto-optic system, Fc stands for fiber coupler. The opti-cal bench of nIsT-F2 is conceptually very similar, even if different laser sources are employed.

Fig. 4. laser amplitude noise spectrum of IT-csF2 molasses beams. Black (a) laser noise in open loop condition, red (b) laser noise in closed loop condition, yellow (c) noise of the voltage reference, blue (d) noise of the photodiode and transimpedence amplifier. In the final operative configuration, the reference is filtered at 10 Khz to further improove the laser noise (not shown in the picture). references to color refer to the online version of this image.

http://dx.doi.org/10.1109/TUFFC.2010.1453/mm1

ses is directly loaded from cs vapor. no magneto-optical trap loading or beam slowing is provided yet; however, enough atoms are collected to achieve a shot noise com-patible with our local oscillator (BVa oscillator flicker floor < 1 × 10−13 from 1 to 100 s).

although a fine tuning of the post-cooling process is still required, preliminary results for IT-csF2 have shown already a s/n of 400 in the detected transition proba-bility, allowing for a short-term stability σy(τ) = 2.5 × 10−13τ−1/2. The molasses temperature achieved so far is around 1.5 µK, leaving us with an average density shift and a dead time in the duty cycle time a bit higher than our target value. a temperature approaching 0.5µK (al-ready demonstrated in nIsT-F1) should be reached to reduce the dead time and the average density of the launched cloud. In Fig. 5, the short-term stability of IT-csF2 is shown. The fountain was operated with the atoms launched at 1 m height above the trapping region, a con-figuration that allows observation of the ramsey fringes (Fig. 6) with a 0.9-hz linewidth.

III. Preliminary accuracy Evaluation

In the following section, we report a preliminary ac-curacy budget for IT-csF2. The main biases have been evaluated ,even if the ultimate accuracy and stability of the fountain has not yet been achieved.

A. Zeeman Effect

In a similar way to that used in IT-csF1 and nIsT-F1, the c-field is evaluated through the excitation of the low frequency, magnetic sensitive, Majorana transitions be-tween the F = 3, mF = 0 and F = 3, mF = ± 1 states [15]. To this purpose, a long transversal coil is placed along the whole ramsey interaction region. When atoms reach their apogee, a 100-ms-long pulse is applied, giving as a result the local value of the magnetic field (Fig. 7). The results

of the map are used to obtain the time integral value of Zeeman shift. This calculation predicts the position of the central ramsey fringe for the magnetic sensitive transi-tion F = 3, mF = 1 F = 4, mF = 1 (Fig. 8). Because the sampling atomic cloud is not a point-like source, we don’t expect a perfect matching of the map with the actual field, because a residual spatial/time integration of the field is always present. The discrepancy between the predicted value and the measured one tells us the quality of the map and allows us to determine the uncertainty in the quadratic Zeeman correction we apply to the frequency standard.

In our case, the maximum discrepancy observed be-tween the measured value of the central fringe and the cal-culated one is 100 mhz. In Fig. 8, the numerical prediction of the value of the magnetic sensitive transition central fringe is reported, together with the frequency position of the ramsey fringe measured for certain given launching


Fig. 5. short-term stability of IT-csF2.Fig. 6. ramsey fringes of IT-csF2. The signal is obtained with a ramsey time of 0.55 s, corresponding to a launching height of approximately 1 m.

Fig. 7. Map of the c-field along the interaction region.

values. The prediction provided by the local map allows unambiguous identification of the central fringe of the magnetic sensitive line.

The second source of uncertainty in the Zeeman shift is related to the time stability of the field. We have mea-sured a relative flicker level of the field of 10−4.

Thus, the Zeeman shift and its uncertainty can be writ-ten as:

Dnn

Z = ±( )´ -633 0 3 10 16. .

B. Blackbody Radiation Shift

a fundamental problem in the evaluation of the BBrs is that it is extremely difficult to measure with high pre-cision if the radiation temperature of the blackbody cor-responds to the temperature measured, especially if the system we are using is not designed to be a blackbody radiator, typically a large spherical cavity with a small hole. In almost all other geometries, we have to deal to some extent with reflections at low incidence angle. In F2, we designed the ramsey interaction region to be as close as possible to a perfect blackbody.

There is only one aperture to allow the atoms to go through the cavity (no opening on the top), and the walls of this long channel are threaded to avoid low-angle re-flection. as shown in Fig. 2, the temperature is measured with 3 platinum resistors placed along the interaction re-gion. Given the observed temperature gradient and using the value of β measured in [4], we calculate the BBrs at 90 ± 2K:

Dnn

BB = - ±( )´ -1 4 0 1 10 16. . .

C. Microwave Leakage

according to the theory developed in [16], the micro-wave leakage in F2 (that by design consideration can hap-

pen only after the second ramsey interaction) can be es-timated by means of differential measurements, where the ramsey excitation amplitude is varied from a π/2 pulse to a (2n − 1)π/2 pulse. In these preliminary measurements, we have changed the excitation pulse from π/2 to 3π/2. The result of this procedure extrapolated to π/2 opera-tion is

Dn

nLeak = ±( )´ -7 8 10 16.

D. Atomic Density Shift

The spin-exchange shift is typically one of the most relevant biases in cs fountains. several techniques have been studied in the last few years to reduce or better control this shift [17]–[19]. For the moment, none of these techniques are implemented in F2, although the physical structure of the fountain is designed to be modified to implement multiple launching. however, nIsT-F1 [20] has demonstrated that it is also possible in standard molas-ses operation to achieve a relative uncertainty in the shift measurement well below 1 × 10−16.

For the moment, however, the post-cooling process in F2 is not yet optimized and the return-atom fraction is still lower than our target, forcing us to operate at a relatively higher molasses density. a differential measure-ment, where the atom density is changed by a factor of 3 (Fig. 9) yields a shift estimation of the fountain operated in the low-density regime of

Dnn

SE = - ±( )´ -39 21 10 16.

E. Gravitational Red Shift

The elevation of InrIM on the Geoid was measured recently with high accuracy [21]. The IT-csF2 elevation is 238.9 ± 0.1 m. The resulting gravitational red shift is


Fig. 8. Time integral of the c-field, and measurement of the central fringe of the Δm = 0, m = −1 transition, with respect to the clock transition frequency.

Fig. 9. relative frequency versus atom density variation. atom density is normalized to the configuration of low density.

calculated, taking into account the value of Δν/ν = 1.09 × 10−16/m:

Dnn

G = ±( )´ -260 4 0 1 10 16. . .

IV. conclusion

We have presented the physical structure of nIsT-F2 and IT-csF2, two new cryogenically cooled cs fountain primary frequency standards, developed at nIsT (Boul-der, co) and at InrIM (Torino, Italy). We reported as well a preliminary accuracy evaluation of IT-csF2, show-ing that no unexpected biases are present at the 10−15 level. Further work and investigation will be needed to achieve the target accuracy for these 2 fountains of ~1 × 10−16.

references

[1] Circular T (June 13, 1996) no. 101 [online] available ftp://ftp2.bipm.org/pub/tai/publication/cirt.101

[2] Circular T (april 9, 2009) no. 255 [online] available ftp://ftp2.bipm.org/pub/tai/publication/cirt.255

[3] W. M. Itano, l. l. lewis, and d. J. Wineland, “shift of 2s1/2 hyper-fine splitting due to blackbody radiation,” Phys. Rev. A, vol. 25, no. 2, pp. 1233–1235, 1982.

[4] E. simon, P. laurent, and a. clairon, “Measurement of the stark shift of the cs hyperfine splitting in an atomic fountain,” Phys. Rev. A, vol. 57, no. 1, pp. 436–439, 1998.

[5] F. levi, d. calonico, l. lorini, s. Micalizio, and a. Godone, “Mea-surement of the blackbody radiation shift of the 133cs hyperfine transition in an atomic fountain,” Phys. Rev. A, vol. 70, no. 5B, art. no. 033412, 2004.

[6] r. Wynands and s. Weyers, “atomic fountain clocks,” Metrologia, vol. 42, no. 3, pp. s64–s79, 2005.

[7] E. J. angstmann, V. a. dzuba, and V. V. Flambaum, “Frequency shift of hyperfine transitions due to blackbody radiation,” Phys. Rev. A, vol. 74, no. 2, art. no. 023405, 2006.

[8] K. Beloy, U. I. safronova, and a. derevianko, “high-accuracy cal-culation of the blackbody radiation shift in the 133cs primary fre-quency standard,” Phys. Rev. Lett., vol. 97, no. 4, art. no. 040801, 2006.

[9] T. P. heavner, T. E. Parker, J. h. shirley, and s. r. Jefferts, “nIsT-F1 and F2,” in Proc. 7th Symp. Frequency Standard Metrology, 2008, pp. 299 –307.

[10] s. r. Jefferts, T. P. heavner, T. E. Parker, and J. h. shirley, “nIsT cesium fountains—current status and future prospects,” in Proc. 2007 SPIE Conf., art. no. 667309.

[11] s. r. Jefferts, J. shirley, T. E. Parker, T. P. heavner, d. M. Meek-hof, c. nelson, F. levi, G. costanzo, a. de Marchi, r. drullinger, l. hollberg, W. d. lee, and F. l. Walls, “accuracy evaluation of nIsT-F1,” Metrologia, vol. 39, pp. 321–336, 2002.

[12] F. levi, l. lorini, d. calonico, and a. Godone, “reduction of the cold collisions frequency shift in a multiple velocity fountain: a new proposal,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 48, no. 3 pp. 847–850, 2001.

[13] E. a. donley, T. P. heavner, F. levi, M. o. Tataw, and s. r. Jefferts, “double-pass acousto-optic modulator system,” Rev. Sci. Instrum., vol. 76, art. no. 063112, 2005.

[14] c. calosso, G. a. costanzo, and a. de Marchi, “a syncronous multiple frequency synthesizer for atomic fountain frequency stan-dards,” in Proc. 15th European Frequency Time Forum, neuchatel, switzerland, 2001, pp. 412–415.

[15] F. levi, l. lorini, d. calonico, and a. Godone, “IEn-csF1 accu-racy evaluation and two-way frequency comparison,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 51, no. 10, pp. 1216–1224, 2004.

[16] J. h. shirley, F. levi, T. P. heavner, d. calonico, d.-h. yu, and s. r. Jefferts, “Microwave leakage-induced frequency shifts in the pri-mary frequency standards nIsT-F1 and IEn-csF1,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 53, no. 12, pp. 2376–2385, 2006.

[17] F. levi, d. calonico, l. lorini, and a. Godone, “IEn-csF1 primary frequency standardat InrIM: accuracy evaluation and TaI calibra-tions,” Metrologia, vol. 43, pp. 545–555, 2006.

[18] K. szymaniec, W. chałupczak, E. Tiesinga, c. J. Williams, s. Wey-ers, and r. Wynands, “cancellation of the collisional frequency shift in caesium fountain clocks,” Phys. Rev. Lett., vol. 98, art. no. 153002, 2007.

[19] F. Pereira dos santos, h. Marion, s. Bize, y. sortais, a. clai-ron, and c. salomon, “controlling the cold collision shift in high precision atomic interferometry,” Phys. Rev. Lett., vol. 89, art. no. 233004, 2002.

[20] T. P. heavner, s. r. Jefferts, E. a. donley, J. h. shirley, and T. E. Parker, “nIsT-F1: recent improvements and accuracy evaluations,” Metrologia, vol. 42, pp. 411–422, 2005.

[21] d. calonico, a. cina, I. h. Bendea, F. levi, l. lorini, and a. Godone, “Gravitational redshift at InrIM,” Metrologia, vol. 44, pp. l44–l48, 2007.

authors’ photographs and biographies not available at time of publica-tion.


cryogenic fountain development at nist and inrim: preliminary...

Documents