88 OPTICS LETTERS / Vol. 20, No. 1 / January 1, 1995

Laser spectroscopy of molecular cesium near1064 nm enhanced by a Fabry–Perot cavity

Ady Arie and Eran Inbar

Department of Physical Electronics, Faculty of Engineering, Tel Aviv University, Tel Aviv 69978, Israel

Received October 13, 1994

We made Doppler-broadened and Doppler-free absorption measurement of a cesium cell that was placed inside anexternal Fabry–Perot ring cavity, using a Nd:YAG laser. We achieved a cavity finesse of ,85, and the sensitivityof the measurement was improved by as much as 26.5 with respect to single-pass absorption. The sensitivityenhancement by the external cavity may enable one to lock the laser frequency to 133Cs2 lines at relatively lowtemperatures, in which the frequency-dependent temperature shift, as well as the collision broadening, is reducedsignificantly.

Owing to their inherently low frequency and in-tensity noise, small size, high reliability, and po-tentially high power, monolithic diode-laser-pumpedsolid-state lasers1 are very attractive for applicationsin the fields of communication and precision-lasermetrology. In these applications the laser frequencyshould be absolutely stabilized; in optical communi-cation systems this can ease the acquisition processbetween the local oscillator and receiver2 and alsomay permit cold-start communication.3 Moreover,in metrological applications—e.g., length measure-ments—the laser frequency has to be known pre-cisely and kept fixed during measurements.

Progress in the development of absolutely stabi-lized sources has been plagued by the difficulties offinding suitable absorbers to which the frequency canbe locked within the relatively narrow tuning range ofsolid-state lasers. Within the gain bandwidth of the1064-nm Nd:YAG laser there are several known ab-sorbers, including molecular cesium, 133Cs2,4 – 6 CO2,7C2H2, and C2HD.8 Unfortunately, the low cross sec-tions of these absorbers significantly complicate thepossibility of locking the laser frequency to the ab-sorption lines.

Nevertheless, molecular cesium has been stud-ied by several groups for frequency-stabilizationapplications.4 – 6 The locking of a Nd:YAG laserto 133Cs2 Doppler-free lines was demonstrated byMak et al.6 Their cesium cell was operated ata temperature of 220 ±C, and the best stability of6 parts in 1011 was reached at an integration timeof 1 s. At this temperature, the frequency stabilityis limited by temperature-induced frequency shifts.Furthermore, the high operating temperature maylead to pressure broadening, corrosive attack of cellwindows,5 and high electrical energy consumption.At lower cell temperatures these problems are re-duced, but unfortunately the absorption significantlydecreases as well, thereby preventing a sufficientsignal-to-noise ratio for stable locking.

To overcome the problems associated with lock-ing the Nd:YAG frequency to absorption lines near1064 nm, an alternative method of doubling the fre-quency of the Nd:YAG laser and locking the sec-ond harmonic to iodine transitions near 532 nm wasinvestigated.2,9 – 13 This technique yielded fairly high

0146-9592/95/010088-03$6.00/0

stability; the root Allan variance reached a level of2.5 parts in 1013 in a measurement time of 32 s.12

Because low-power cw sources often are used, thesecond-harmonic conversion should be efficient. Inaddition, reaching and locking to a specific line re-quires frequency tunability. However, efficient andtunable nonlinear frequency conversion is a fairlycomplicated and expensive process that usually re-quires resonant enhancement11,12 or a guided-wavestructure.14 Hence it still is desirable to lock directlyto absorbers whose fundamental tuning range fallswithin that of the laser.

Locking to the weak lines of Cs2 at lower pres-sures and temperatures requires higher sensitivitiesthan those attainable by conventional Doppler-freespectroscopy techniques. In this Letter we showthat the sensitivity can be improved by insertinga cesium cell into a Fabry–Perot ring cavity anddetecting the transmitted signal at the resonance fre-quency of the cavity. Similar techniques have beenused before with several different absorbers,8,15,16

but, as far as we know, this technique has not yetbeen applied to the detection of Cs2 lines. The ex-perimental setup is shown in Fig. 1. A monolithicdiode-pumped Nd:YAG laser (Lightwave Electron-ics 122) that emits 300 mW of power at 1064 nmis split into pump and probe beams and is injectedfrom opposite ports into the ring Fabry–Perot cav-ity. The cavity itself consists of four mirrors, oneof which can be translated by means of a piezo-electric transducer (pzt). The input and outputmirrors are flat mirrors with intensity reflectiv-ities of R1 ­ 0.985 and R2 ­ 0.99, whereas thetwo other mirrors are concave high reflectors (ra-dius of curvature 75 cm; reflectivity .0.997). Thetotal round-trip length of the cavity is ,3 m. With-out the Cs2 cell, we have measured a finesse of 220,which agrees well with the reflectivities of the mir-rors. When the Cs2 (L ­ 15 cm) was inserted, thefinesse of the cavity,

F ­p

pr1r2tcell

1 2 r1r2tcell

, (1)

dropped to 85, which indicated a total cell amplitudetransmission tcell ­ 0.976. r1 and r2 are the am-

1995 Optical Society of America

anuary 1, 1995 / Vol. 20, No. 1 / OPTICS LETTERS 89

J

Fig. 1. Experimental setup for cesium spectroscopy witha Fabry–Perot cavity: HR’s, high reflectors.

plitude reflectivities of the input and output mirrors.Because room-temperature Cs2 absorption is negli-gible, the loss in passage through the cell is duemainly to deviation of the windows from the exactBrewster angle. The cell is placed inside a smalloven whose cold-point temperature is monitored witha thermocouple. The cell windows are heated to atemperature that is slightly higher than that of thecell body to prevent condensation on the windows.The oven and the cell are mounted on a three-axistilt platform for optimizing the alignment.

To scan the laser frequency over the cesiumlines, we apply a ramp voltage to the tempera-ture–frequency actuator of the laser, therebycovering a laser temperature range free of modehops—between 31 and 35 ±C. During this relativelyslow scan (period ,60 s), the piezoelectric mountedmirror is dithered at a high rate (period 20 ms)and at an amplitude of approximately 0.7 mm (i.e.,slightly more than half of the laser wavelength,therefore passing at least once through the cavityresonance). The interferometer output is detectedby a photodetector followed by a digital scope operat-ing in peak-detect mode. This configuration permitsthe detection of the cavity output at resonance, as thelaser frequency is scanned over a wide range, withoutlocking the cavity resonance to the laser frequency.

When the laser frequency does not coincide witha cesium absorption line the intensity-transmissionfactor at resonance is

I0 ­

Ét1t2tcell

1 2 r1r2tcell

É 2

, (2)

and where t1 and t2 are the amplitude-transmissioncoefficients of the mirrors. For low-loss mirrors,ti ­

ps1 2 Rid, i ­ 1, 2. When the laser frequency

coincides with a cesium absorption line, with anamplitude absorption of a cm21, the output inten-sity-transmission factor becomes

Ia ­

Ét1t2tcell exps2aLd

1 2 r1r2tcell exps2aLd

É 2

. (3)

The improvement in sensitivity therefore is thenormalized cavity loss sI0 2 IadyI0 divided by thesingle-pass absorption f1 2 exps22aLdg. If the ce-sium amplitude absorption ,aL is much lower thanthe cavity loss 1 2 r1r2tcell, a simple algebraic calcu-lation yields

Ia ø I0s1 2 2aLFypd ; (4)

hence the sensitivity improvement is Fyp. (For astanding wave cavity, the improvement is 2Fyp, be-cause the light travels twice through the cell in eachround trip.8) As the cesium absorption increases,the sensitivity improvement gradually decreases.Without the pump beam, a is simply the linear lossof Cs2 sustained by the probe, but when the pumpbeam is used the loss coefficient of a frequency rangecentered at the Cs2 resonance frequency, whose widthis the homogenous linewidth, is reduced by the pumpsaturation.

Figure 2 shows the transmission of the probethrough the Fabry–Perot cavity at resonance, whilethe laser frequency is scanned over several cesiumlines. In this measurement the pump beam isblocked. The circulating intensity is 0.2 W cm22,and the beam diameter is 2.4 mm. Attenua-tion of the intensity by a factor of 2 results ina negligible effect on the shape of the transmit-ted signal, which indicates that saturation effectsare fairly insignificant. For comparison, single-pass transmission without the Fabry–Perot cavityalso is shown. The line numbers and the corre-sponding wave numbers are taken from Ref. 5.The sensitivity improvement is evident; for ex-ample, the single-pass loss of line #8 at 9394.06 cm21

is 0.08 6 0.01, whereas at the output of theFabry–Perot cavity we measure a maximum loss of0.70 6 0.02. Thus the absorption signal is improvedby 8.8 61.3, which is in good agreement with Eqs. (2)and (3), which predict for single-pass loss level of0.08 6 0.01 a cavity output loss 10.1 6 0.8 higher.

As the single-pass loss is reduced, e.g., by lower-ing the cell temperature, the improvement in ab-sorption signal should increase. The normalizedcavity output loss at the center of line #8 is 0.40 and0.21 at 200 and 185 ±C, respectively. The single-passloss at these temperatures is too low to be measureddirectly; however, we have determined, by measure-ments at higher temperatures (215–260 ±C), that theabsorption cross section of this line is 5 3 10216 cm2.

Fig. 2. Doppler-broadened scan of 133Cs2 (pump beamblocked). Upper trace, single-pass transmission throughthe 15-cm cell; lower trace, transmission through theFabry–Perot cavity. Line numbers are taken fromRef. 5.

90 OPTICS LETTERS / Vol. 20, No. 1 / January 1, 1995

Fig. 3. Scan of cesium line #8 by the probe beamduring saturation of the absorption with a counter-propagating pump beam.

Hence, because the molecular vapor pressures at 200and 185 ±C are 0.0212 and 0.0067 Pa, respectively,17

the calculated single-pass losses for a 15-cm cell are0.0241 and 0.0079. The absorption signal is im-proved therefore by a factor of 16.5 at 200 ±C and bya factor of 26.5 at 185 ±C, again in good agreementwith the predictions of Eqs. (2) and (3). As men-tioned above, the highest attainable improvement insensitivity is Fyp , 27.

By unblocking the pump signal, we obtain theDoppler-free linewidth of line #8 (Fig. 3). The pumpand probe intensities are 2 and 0.4 W cm22, re-spectively. The contrast of the Doppler-free linewith respect to the Doppler envelope is 14%. TheDoppler envelope is asymmetric because of theneighboring line at 9394.1 cm21 (Fig. 2). The con-tribution of this neighboring line also may ex-plain why the measured Doppler linewidth of 4506 25 MHz is broader than the expected linewidthof 275 MHz. The measured Doppler-free linewidthof 35 6 5 MHz is in agreement with the results ofRef. 4. It should be mentioned that not all the lineshave the same width. For example, the Doppler-freelinewidth of line #9 is 45 6 5 MHz.

With several modifications, the setup of Fig. 1 maybe used to lock the laser frequency to cesium Doppler-free lines. The pump can be phase modulated by anelectro-optic modulator, and the reflected signal thencan be used to lock the cavity resonance frequency tothe laser frequency. Because the laser is monolithic,this will reduce the sensitivity of the enhancementcavity to acoustic vibrations. We use a ring cavity,which makes it possible to eliminate the Dopplerenvelope; one way is to chop the pump and de-tect the probe signal at the chopping frequency witha lock-in amplifier,15 as in conventional saturationspectroscopy. With additional phase modulation ofthe probe it is possible to obtain an error signalfor locking the laser frequency to the cesium line.Under the locked condition the laser frequency, thecavity resonance frequency, and the Cs2 transitionwill coincide.

According to Mak et al.,6 the stability of lock-ing Nd:YAG to cesium at 220 ±C was limited bytemperature-induced frequency shifts whose mainorigins are collisions between cesium atoms and

molecules. The frequency shift is proportional toPaty

pT , where Pat is the atomic gas pressure and T

is the temperature. Lowering the temperature willlead to a significant reduction in the atomic-vaporpressure and to its variation with temperature.17

As a result, the temperature-induced frequency shiftwill be reduced as well. An improvement in sen-sitivity by a factor of 25, which is attainable withour setup, enables us to lower the temperature from220 to 160 ±C, while we maintain the signal-to-noiseratio. At 220 ±C the temperature-induced frequencyshift is ,200 kHzyK,6 but at 160 ±C it is approxi-mately 60 kHzyK.6,17 An additional benefit is thelinewidth reduction by approximately a factor of 2.These effects should enable Nd:YAG lasers to be sta-blized absolutely to 133Cs2, with improved accuracyand stability.

The technique of signal enhancement by aFabry–Perot cavity becomes even more efficientwhen an absorber that does not attack the mirrorscorrosively is used. In this case the absorber doesnot have to be maintained inside a sealed cell withBrewster windows, and very high mirror-loss-limitedfinesse can be achieved.8 This may enable one tostudy in detail the lines of CO2,7 C2H2, and C2HD(Ref. 8) and may provide additional candidates forNd:YAG frequency-stabilization at 1064 nm.

References

1. T. J. Kane and R. L. Byer, Opt. Lett. 10, 65 (1985).2. R. Heilmann and J. Kuschel, Electron. Lett. 29, 810

(1993).3. Y. C. Chung, J. Lightwave Technol. 8, 869 (1990).4. O. A. Orlov and V. I. Ustyugov, Sov. Tech. Phys. Lett.

12, 120 (1986).5. K. Wallmeroth and R. Letterer, Opt. Lett. 15, 812

(1990).6. A. A. Mak, S. G. Muravitsky, O. A. Orlov, and

V. I. Ustyugov, Proc. Soc. Photo-Opt Instrum. Eng.1121,478 (1989).

7. P. F. Fritschel and R. Weiss, Appl. Opt. 31, 1910(1992).

8. K. Nakagawa, T. Katsuda, A. S. Shelkovnikov, M. deLabachelerie, and M. Ohtsu, Opt. Commun. 107, 369(1994).

9. S. V. Kruzhalov, V. A. Parfenov, L. N. Pakhomov,and V. Yu. Petrunkin, Sov. Tech. Phys. Lett. 11, 111(1985).

10. P. Esherick and A. Owyoung, J. Opt. Soc. Am. B 4,41 (1987).

11. A. Arie, S. Schiller, E. K. Gustafson, and R. L. Byer,Opt. Lett. 17, 1204 (1992).

12. A. Arie and R. L. Byer, J. Opt. Soc. Am. B 10, 1990(1993).

13. J. A. Harison, M. Zahedi, and J. W. Nibler, Opt. Lett.18, 149 (1993).

14. A. Arie, M. L. Bortz, M. M. Fejer, and R. L. Byer, Opt.Lett. 18, 1757 (1993).

15. L. S. Ma and J. L. Hall, IEEE J. Quantum Electron.26, 2006 (1990).

16. M. de Labachelerie, K. Nakagawa, and M. Ohtsu, Opt.Lett. 19, 840 (1994).

17. A. N. Nesmeyanov, The Vapor Pressure of the Chem-ical Elements (USSR Academy of Sciences, Moscow,1961), p. 445.