High-frequency-stability diode-pumped Nd:YAG laserswith the FM sidebands method and Doppler-freeiodine lines at 532 nm

Gianluca Galzerano, Cesare Svelto, Elio Bava, and Fabrizio Bertinetto

The FM spectroscopy technique has been applied to two frequency-doubled Nd:YAG lasers to achieveabsolute frequency stabilization against the hyperfine structure components of the rovibronic P~54! 32–0iodine line at 532 nm. A fractional frequency stability of 2 3 10213 t21y2 has been obtained forintegration times in the range of 1 ms , t , 10 s. For longer integration times the stability level remainsbelow 10213, reaching a minimum value of 4.6 3 10214 at 100 s. This high stability level is, to ourknowledge, the best value achieved against iodine lines by this locking method and for a fully transport-able system. © 1999 Optical Society of America

OCIS codes: 120.3930, 120.3940, 120.4800, 300.6320, 020.2930.

aptWttt

atbtrfindosb

1. Introduction

Because of their inherently low level of amplitudeand frequency noise, diode-pumped nonplanar ringoscillator Nd:YAG lasers1,2 are attractive lightsources for different scientific applications such asmetrology,3 spectroscopy,4 space communications,5and high-resolution detectors ~gravitational wavesensors, lidar, and optical radar!.6 Although theshort-term frequency stability of these sources can beimproved at high levels by means of locking the laserfrequency to high-finesse Fabry–Perot cavities,7 formany applications the long-term stability and therepeatability of the stabilized frequency are also ofgreat importance.3 With regard to this final goal,frequency-doubled Nd:YAG lasers are well suited tothis purpose because of the good coincidence withstrong absorptions of relatively narrow B–X transi-tions of molecular iodine at 532 nm.3,4

G. Galzerano, C. Svelto, and E. Bava are with the Dipartimentodi Elettronica e Informazione, Politecnico di Milano, Istituto Na-zionale di Fisica della Materia and Consiglio Nazionale delleRicerche, Centro Studi per le Telecomunicazioni Spaziali, PiazzaLeonardo da Vinci 32, 20133 Milano, Italy. F. Bertinetto is withthe Consiglio Nazionale delle Ricerche, Istituto di MetrologiaGustavo Colonnetti, Strada delle Cacce 73, 10135 Torino, Italy.The e-mail address for G. Galzerano is gianluca.galzerano@polimi.it.

Received 17 May 1999; revised manuscript received 11 August1999.

0003-6935y99y336962-05$15.00y0© 1999 Optical Society of America

6962 APPLIED OPTICS y Vol. 38, No. 33 y 20 November 1999

To lock the laser frequency against saturated ab-sorption lines, several FM techniques can be usedsuch as the first-, third-, and fifth-harmonic meth-ods8; the FM sidebands technique9,10; and the modu-lation transfer method.11 In particular, for the532-nm wavelength and an external iodine cell, thebest performance in terms of accuracy and stability ofthe optical standard have been obtained with themodulation transfer method.12 The modulationtransfer technique, however, requires a modulationfrequency lower than the resonance linewidth, whichis of the order of 1 MHz for the saturated iodine lines~at a typical iodine pressure of a few pascals!. Tochieve the quantum-limited condition, the laser am-litude noise should be below the quantum level inhe Fourier region near the modulation frequency.

hen the laser amplitude noise is still significant athese frequencies, the proper solution is to increasehe modulation frequency to several megahertz ando use the FM sidebands method.

For the FM technique and saturation spectroscopyn unmodulated pump beam saturates the molecularransition whereas a counterpropagating probeeam, phase modulated at a frequency higher thanhe resonance linewidth, is used to detect the satu-ated absorption. This absorption profile is, to therst order, Doppler free and therefore homoge-eously broadened. Because of the absorption andispersion profiles of the saturated resonance, theriginal phase modulation is converted into a corre-ponding amplitude modulation that can be detectedy a photodiode. The in-phase component of the

ttntf

Pis

ptPw

synchronously demodulated photocurrent ~related tohe dispersion profile! is an odd function of the de-uning between the laser frequency and the reso-ance frequency ~discriminating signal! and canherefore be used as an error signal to lock the laserrequency to the resonance center.9,10

Here we report the frequency stabilization, withthe FM sidebands method, of two commercialfrequency-doubled Nd:YAG lasers against a Doppler-free P~54!32–0 iodine absorption line. Further-more, high-resolution hyperfine spectroscopy of theP~54! 32–0 iodine line was performed.

2. Experimental Setup and Results

In this experimental study, since an excess amplitudenoise in the Fourier frequency range from 100 Hz to1 MHz was observed for both employed laser sources~because of technical noise and the residual relax-ation oscillation!, the FM method was evaluated asbeing more convenient than the modulation transfertechnique. The adopted experimental setup isshown in Fig. 1. Two frequency-doubled Nd:YAGlasers ~Lightwave Model 142! emitting single-frequency 532-nm radiation with 32- and 18-mW out-put powers, respectively, have been used.13 Withthese laser sources the fundamental wavelength at1064 nm is available as well, with an output power of;28 mW. Within the thermal frequency tuningrange of the green radiation ~3 GHzy°C!, the

~54!32–0 and the R~57!32–0 rovibronic lines of theodine molecule can be observed. To reduce the ob-erved laser-frequency drift induced by power dissi-

Fig. 1. Experimental setup for the stabilization of one frequencybalanced mixer; EOM, electro-optic modulator; I, integral; L, lens;PI, proportional and integral; SM, spherical mirror; TEC, temperatplate; (, adder.

2

ation, the main bodies of both lasers wereemperature controlled at 30.0 6 0.1 °C by means ofeltier modules. A fast and fine frequency control,ith a signal bandwidth of 1 kHz and a rate of ;200

MHzyV ~depending on the particular laser device!, isalso available with use of a piezoelectric actuatorglued to the laser crystal. In our experiment, how-ever, an external acousto-optic modulator ~AOM-1!,driven by a voltage-controlled oscillator ~VCO!, wasused instead of the piezoelectric actuator for the fastand fine tuning of the laser frequency. This solutionwas used because the piezoelectric actuator could notprovide for a satisfactory behavior as a frequencyactuator. For the AOM-1 a cat’s eye double-passconfiguration was adopted. When a quarter-waveplate is used in conjunction with a polarizing beamsplitter, only the reflected and doubly frequency-shifted beam can be selected. In this way it is pos-sible to span the laser frequency by as much as 6 50MHz, with a signal bandwidth of 150 kHz, by meansof tuning the 250-MHz VCO central frequency. Theprobe beam and the pump beam counterpropagateinside a 50-cm-long iodine cell with orthogonal polar-izations and with collimated beam diameters of ;1.5mm. We kept the iodine vapor inside the cell at afixed pressure by keeping the cold-finger temperaturewithin 0.01 °C. Phase modulation of the probe beamis produced by an electro-optic modulator resonant atthe driving rf signal of 10 MHz. This frequency is agood compromise value, because, at the same time, itis higher than the iodine resonance linewidth, it issufficiently far from the major laser amplitude noise

bled Nd:YAG laser: A, amplifier; BS, beam splitter; DBM, doubleirror; P, polarizer; PBS, polarized BS; PD, photodiode; ph, pinhole;ontroller; F, phase shifter; ly2, half-wave plate; ly4, quarter-wave

-douM, mure c

0 November 1999 y Vol. 38, No. 33 y APPLIED OPTICS 6963

rpcspt

cstawcs

cl

d5l

s

t

6

contributions, and it enables iodine hyperfine struc-ture ~hfs! spectrum measurements with high resolu-tion. By means of the AOM-2 the saturating beam isfrequency shifted by 80 MHz and amplitude choppedat a rate of 95 kHz, thus avoiding interference pat-terns that are due to optical feedback on the photo-detector and allowing, with a lock-in detectionscheme, for the suppression of the linear absorption~Doppler! background profile. Two similar and in-dependent frequency-locking systems ~A and B! wereconstructed ~one of which is fully transportable! toestimate the achieved absolute frequency stability bybeat note measurements. In the A and the B stabi-lization systems the different AOM-2’s were alignedto select the iodine molecules with opposite velocityprofiles. In this way the locking frequencies wereshifted from the resonance line centers by 140 MHz~system A! and 240 MHz ~system B! so that the laserbeat frequency could be measured even when thelasers were locked against the same hfs component ofthe P~54!32–0 line.

With the usual probe and pump power conditions of0.45 and 1.4 mW, respectively, a phase-modulationfrequency of 10 MHz, with a modulation index of b >1, and an iodine cold-finger temperature of 215 60.01 °C ~corresponding to a vapor pressure of 1.36Pa!, a linewidth of 960 kHz, and a slope at line centerof 30 VyMHz have been obtained, respectively, fromthe absorption and the dispersion line shapes for themost isolated a1 hfs component of the P~54!32–0line.14 The homogeneous linewidth is a contributionof different broadening mechanisms induced by mo-lecular collisions ~pressure broadening!, by the satu-ation effect that results from the saturating beamower, and by the limited transit time of the mole-ules through the laser beams. In our case the ob-erved linewidth value is determined mainly byressure and power broadening, being the transitime contribution equal to ;10 kHz.

To lock the laser frequency with respect to the hfsomponent, the in-phase signal is simultaneouslyent to the VCO and to the laser temperature con-roller by means of proportional and integral controls,s shown in Fig. 1. The obtained closed-loop band-idth, evaluated by means of a frequency-to-voltage

onverter to resolve the frequency noise spectral den-ity of the beat signal, is equal to ;10 kHz. We

evaluated the frequency stability of the two lasers bymeasuring the Allan standard deviation of the beatfrequencies between both the fundamental and thegreen radiations, at different integration times t.From typical measurement results, shown in Figs. 2and 3, different frequency noise processes were high-lighted. In the free-running condition, indicated inthe diagrams by squares, a flicker frequency noise ata level of sy 5 10211 ~2 3 107 HzyHz for the 532-nmwavelength! was observed for integration times lowerthan 10 ms, whereas for longer integration times alinear frequency drift with sy~t! 5 3 3 10210t ~corre-sponding to ;150 kHzys for the 532-nm wavelength!an be observed for 50 ms , t , 5 s. When bothasers are locked against the hfs component a1 of the

964 APPLIED OPTICS y Vol. 38, No. 33 y 20 November 1999

P~54!32–0 line ~see the circles in Figs. 2 and 3! theominant frequency noise contribution, at the32-nm wavelength, is a white noise yielding an Al-an standard deviation sy > 2 3 10213 t 21y2 ~25 3 103

Hz2yHz!, for integration times between 1 ms and 10 s.For longer integration times the stability remains

Fig. 2. Allan standard deviation of the beat signal between thegreen radiations as a function of the integration time t. f, free-running lasers. F, both lasers locked against the hfs componenta1 of the P~54!32–0 line. Solid and dotted lines represent, respec-tively, the relations sy > 2 3 10213t21y2 for 1 ms , t , 10 s and

y~t! 5 3 3 10210t for 50 ms , t , 5 s. Error bars, data dispersionat a 1 2 s level.

Fig. 3. Allan standard deviation of the beat signal between theinfrared radiations as a function of the integration time t. f,free-running lasers. F, both lasers locked against the hfs compo-nent a1 of the P~54!32–0 line. Solid and dotted lines represent,respectively, the relations sy~t! 5 5 3 1029t21y2 for integrationimes t . 10 s and sy~t! 5 3 3 10210 t for 50 ms , t , 5 s. Error

bars, data dispersion at a 1 2 s level.

t

lp

btwi1

qo

onrlqi

Table 1. Hyperfine Frequency Splittings Measurement and Fit for

below the 10213 level, reaching its minimum value of4.6 3 10214 at 100 s.

However, it is necessary to observe that the stabil-ity of the infrared radiation does not follow the be-havior of the green radiation stability, because theexternal frequency actuator acts only on the latter.For the fundamental wavelength, in fact, only thefrequency drifts were partially corrected by the slowthermal actuator, which acts for integration times t. 10 s. This phenomenon is apparent from the di-agram shown in Fig. 3, where there is no significantdifference between the sy~t! of the infrared radiationin free-running and under the locking condition, pro-vided that the integration times are below 10 s. Forlonger integration times the relative stability at thefundamental frequency becomes sy~t! 5 5 3 1029

t21y2 under frequency-stabilized conditions.High-accuracy recording of the hfs spectrum of the

P~54!32–0 iodine transition was achieved with thefrequency-offset locking scheme. By means of aphase-locked loop and a synthesizer the frequency oflaser A ~slave! is offset locked to laser B, which isstabilized with respect to hfs component a1 ~referencesource!. In this way, by changing the synthesizerfrequency and at the same time recording the lock-inamplifier output of system A, we obtained the wholespectrum of the P~54!32–0 line, as is depicted in Fig.4. This recording shows the fifteen principal com-ponents ~labeled from a1 to a15! of the P~54!32–0 lineand other weak components that are due to transi-tions from excited vibrational ground-level states,whereas no crossover resonances are evident. Ascan be observed in Fig. 4, the baseline stability ~in-fluenced by residual amplitude modulation and lin-ear absorption! is greatly increased with the use ofhe chopped pump beam setup.

By means of beat frequency measurements and byocking of the laser frequencies to different hfs com-onents, the hfs splittings of the P~54!32–0 iodine

line were precisely evaluated. From these measure-ments the quadratic matrix of the frequency separa-tions was performed. Each matrix element consistsof the mean of five frequency measurements obtainedover an integration time of 10 s. In this way, since

Fig. 4. Hyperfine structure of the P~54!32–0 iodine line. Thelock-in amplifier is set to a sensitivity of 200 mV and an integrationtime of 10 ms.

2

the laser-frequency stability in 10 s is below 10213,the measured frequency separations, reported in Ta-ble 1, have a standard deviation lower than 100 Hz.By fitting the experimental data with a four-termhyperfine Hamiltonian, we also determined the hfscoupling-level constants. The obtained differencesbetween the upper- and the lower-level hfs con-stants15 included in Table 1 are in good agreementwith those given in Ref. 16.

3. Conclusions

A high-stability optical frequency standard at 532 nmhas been demonstrated with frequency-doubled Nd:YAG lasers locked to the hfs components of iodinerovibrational transitions with the Doppler-free FMsidebands method. In addition, high-resolution hy-perfine spectroscopy of the P~54!32–0 iodine line haseen performed. From a metrological point of viewhe 532-nm wavelength is of great interest, since itas recently recommended as an absolute reference

n the Comite Consultatif des Longueurs meeting in997 by means of the frequency value of the a10 hy-

perfine component of the iodine transition R~56!-32–0, with an uncertainty of 40 kHz. With regard tothe hyperfine component a1 of the P~54!32–0, its fre-uency is known, with an additional uncertainty ofnly 2 kHz, from the difference with the a10 compo-

nent of the R~56!32–0 transition. For these reasonsthe reproducibility of these '532-nm wavelengths isf great interest for the whole metrological commu-ity and is subject to deep investigation in our labo-atories. Since the realized frequency-stabilizedaser system is fully transportable, the obtained fre-uency standard will also be estimated by means ofnternational frequency comparisons.

P(54)32-0

Comp.

MeasuredFrequency

~MHz!

Calculateda

Frequency~MHz!

DifferenceCalculated Versus

Measured~kHz!

a1 0 0 0.00a2 260.99385 260.99377 20.08a3 —b 285.010 —a4 — 286.727 —a5 310.07174 310.07132 20.42a6 402.24920 402.25122 2.02a7 417.67155 417.66865 22.90a8 438.91882 438.92047 1.65a9 454.56720 454.56665 20.55a10 571.53694 571.53688 20.06a11 — 698.616 —a12 — 702.936 —a13 — 725.836 —a14 — 731.629 —a15 857.96357 857.96370 0.13

aCalculated by means of an iterative four-term Hamiltonianfitting. The fitting parameters are DeQq 5 1908.432~5! MHz, dC5 86.113~8! kHz, Dd 5 243.76~30! kHz, and Dd 5 210.79~30! kHz.

bComponent not resolved.

0 November 1999 y Vol. 38, No. 33 y APPLIED OPTICS 6965

heterodyne saturation spectroscopy,” Appl. Phys. Lett. 39,

1

6

References1. T. J. Kane and R. L. Byer, “Monolithic, unidirectional single-

mode Nd:YAG ring laser,” Opt. Lett. 10, 65–67 ~1985!.2. T. J. Kane and E. A. Cheng, “Fast frequency tuning and phase

locking of diode-pumped Nd:YAG ring lasers,” Opt. Lett. 13,970–972 ~1988!.

3. M. L. Eickhoff and J. L. Hall, “Optical frequency standard at532 nm,” IEEE Trans. Instrum. Meas. 44, 155–158 ~1995!.

4. A. Arie and R. L. Byer, “Laser heterodyne spectroscopy of 127I2

hyperfine structure near 532 nm,” J. Opt. Soc. Am. B 10,1990–1997 ~1993!.

5. L. G. Kazovsky, “Performance analysis and laser linewidthrequirements for optical PKS heterodyne communications sys-tems,” J. Lightwave Technol. 4, 415–425 ~1986!.

6. A. Abramovici, W. E. Althouse, R. W. P. Drever, Y. Gursel, S.Kawamura, F. J. Raab, D. Shoemaker, L. Sievers, R. E. Spero,K. S. Thorne, R. E. Vogt, R. Weiss, S. E. Whitcomb, and M. E.Zucker, “LIGO: the laser interferometer gravitational-waveobservatory,” Science 256, 325–333 ~1992!.

7. T. Day, E. K. Gustafson, and R. L. Byer, “Sub-hertz relativefrequency stabilization of two-diode laser-pumped Nd:YAG la-sers locked to a Fabry–Perot interferometer,” IEEE J. Quan-tum Electron. 28, 1106–1116 ~1992!.

8. A. J. Wallard, “The frequency stabilization of gas lasers,” J.Phys. E 6, 793–803 ~1973!.

9. J. L. Hall, L. Hollberg, T. Baer, and H. G. Robinson, “Optical

966 APPLIED OPTICS y Vol. 38, No. 33 y 20 November 1999

680–682 ~1981!.10. G. C. Bjorklund, M. D. Levenson, W. Lenth, and C. Ortiz,

“Frequency modulation ~FM! spectroscopy-theory of line-shapes and signal to noise analysis,” Appl. Phys. B 32, 145–152~1983!.

11. J. H. Shirley, “Modulation transfer processes in optical hetero-dyne saturation spectroscopy,” Opt. Lett. 7, 537–539 ~1982!.

12. J. L. Hall, L. S. Ma, M. Taubman, B. Tiemann, F. Hong, O.Pfister, and J. Ye, “Stabilization and frequency measurementof I2-stabilized Nd:YAG laser,” IEEE Trans. Instrum. Meas.48, 583–586 ~1999!.

13. D. C. Gerstenberger, G. E. Tye, and R. W. Wallace, “Efficientsecond-harmonic conversion of cw single frequency Nd:YAGlaser light by frequency locking to a monolithic ring frequencydoubler,” Opt. Lett. 16, 992–994 ~1991!.

14. G. Galzerano, E. Bava, M. Bisi, F. Bertinetto, and C. Svelto,“Frequency stabilization of frequency-doubled Nd:YAG laserat 532 nm by frequency modulation spectroscopy technique,”IEEE Trans. Instrum. Meas. 48, 540–543 ~1999!.

15. A. Yokozeki and J. S. Muenter, “Laser fluorescence state se-lected and detected molecular beam magnetic resonance in I2,”J. Chem. Phys. 72, 3796–3804 ~1980!.

6. J. Ye, L. Robertsson, S. Picard, and J. L. Hall, “Absolute fre-quency atlas of 9 molecular I2 lines at 532 nm,” IEEE Trans.Instrum. Meas. 48, 544–549 ~1999!.