November 15, 2008 / Vol. 33, No. 22 / OPTICS LETTERS 2641

C2H2 absolutely optical frequency-stabilized and40 GHz repetition-rate-stabilized, regeneratively

mode-locked picosecond erbium fiber laserat 1.53 �m

Masataka Nakazawa,* Keisuke Kasai, and Masato YoshidaResearch Institute of Electrical Communication, Tohoku University, 2-1-1 Katahira, Aoba-ku,

Sendai-shi, 980-8577 Japan*Corresponding author: nakazawa@riec.tohoku.ac.jp

Received July 23, 2008; revised September 20, 2008; accepted September 22, 2008;posted October 8, 2008 (Doc. ID 99195); published November 12, 2008

We have succeeded in the simultaneous stabilization of the optical frequency and repetition rate of a regen-eratively mode-locked picosecond erbium-doped fiber ring laser. The optical frequency was locked to the mo-lecular absorption of C2H2 in the 1.5 �m band, and the repetition rate was stabilized to a 40 GHz synthe-sizer by using a microwave phase-locked loop. The optical frequency stability of the pulse train reached2�10−11 for �=10–100 s. The key to success is the independent control of the repetition rate without dis-turbing the optical cavity condition. © 2008 Optical Society of America

OCIS codes: 140.4050 , 140.3425, 140.3510, 140.7090, 120.3940.

The optical frequency stabilization of lasers is a veryimportant research subject if we wish to apply stablefrequencies to optical metrology, including absolutefrequency or wavelength standards, coherent com-munication, and precise optical measurements [1–3].Frequency stabilization is usually achieved in cw la-sers through the use of molecules such as CH4, I2,and C2H2, which are well-known wavelength stan-dards for 3.39, 0.63, and 1.53 �m, respectively [4].However, no one has succeeded in stabilizing the op-tical frequency of an optical pulse train to a molecu-lar reference.

In contrast, the repetition rate of a pulse laser canbe easily stabilized to a synthesizer by using a phase-locked loop (PLL) circuit. For example, we have dem-onstrated 10–40 GHz harmonically and regenera-tively mode-locked erbium-doped fiber lasers in the1.5 �m band with stabilized repetition rates [5,6]. Wealso recently reported an interesting technique forcontrolling the repetition rate without changing theoptical frequency that employs regenerative modelocking [7]. With conventional lasers, it is difficult tochange the repetition rate without changing the cav-ity itself. However, with regenerative mode locking,the repetition rate can also be controlled by changingthe microwave phase in the electrical feedback loop,which is required for regenerative mode locking. Inthis case, since we do not touch the optical cavity, theoptical frequency is not affected directly. Therefore,even though the repetition rate is being controlled,the frequency fluctuation of one of the optical longi-tudinal modes of the output pulse train can be inde-pendently detected by using a molecular reference.Therefore, the error signal thus obtained can benegatively fed back to the laser cavity, resulting inthe simultaneous stabilization of both the absoluteoptical frequency and the repetition rate of the mode-locked pulse. This technique is different from the con-ventional approach for stabilizing the frequency of

the optical comb of a mode-locked pulse laser, since

0146-9592/08/222641-3/$15.00 ©

we do not employ a carrier envelope offset detectionscheme using the one-octave method [8,9].

This Letter describes a method for simultaneouslystabilizing the optical frequency and repetition rateby using a 40 GHz harmonically and regenerativelymode-locked fiber laser. Our previous paper [7] de-scribed a new possibility for the frequency stabiliza-tion of an optical pulse train. In this Letter we actu-ally demonstrate simultaneous stabilization. Theoptical frequency is locked to the linear absorption ofC2H2 molecules, while the repetition rate is locked tothat of a 40 GHz synthesizer. The achieved opticalfrequency stability is almost the same as that of a cwfrequency-stabilized laser [10,11], and the stability ofthe repetition rate is also the same as that of the syn-thesizer.

Our method for the simultaneous stabilization ofthe optical frequency and the repetition rate is shownin Fig. 1, in which thick lines are optical paths andthin lines are electrical paths. The laser configura-tion is based on a 40 GHz harmonically and regen-eratively mode-locked fiber ring laser [12]. The rep-etition rate of the pulse train is stabilized by using aPLL control placed in the regenerative feedback loop[7]. The error signal for the repetition rate stabiliza-tion is obtained with a double balanced mixer (DBM)and is fed back to a microwave phase shifter consist-ing of a piezoelectric transducer (PZT) and an opticalfiber. The optical delay of the pulse in the feedbackloop causes a microwave phase delay for the repeti-tion rate. To stabilize the optical frequency of thepulse train, one of the longitudinal modes is ex-tracted by using a fiber Bragg grating (FBG) with apassband of 1.5 GHz. The FBG is mounted on amultilayer piezo (MLP) actuator, which enables us totune the optical filter frequency. To match the ex-tracted longitudinal mode with the resonance ofC2H2 molecules, an optical frequency componentnear the P(10) line, which has an absolute resonance

frequency around 194821826.42 MHz, was extracted

2008 Optical Society of America

2642 OPTICS LETTERS / Vol. 33, No. 22 / November 15, 2008

through the FBG and coupled to a C2H2 gas cell afterphase modulation for phase-sensitive detection of thelaser frequency fluctuation from the center of theC2H2 reference. Here, we stabilized the amplitude(power) of the extracted longitudinal component toavoid the inclusion of a false error signal caused byamplitude fluctuation in the longitudinal mode. Thiscan be achieved by employing a differential controlfrom a reference voltage as shown at the bottom ofFig. 1, where the longitudinal mode was set at theshoulder of the FBG filter characteristics so that theoutput power became constant when we tuned theMLP with the feedback signal.

The phase-sensitive detection with a C2H2 cellused here has already been reported [10,11]. That is,a frequency modulation with a width of a few hun-dred megahertz is applied to the extracted longitudi-nal mode by using a phase modulator. Then, thebeam is passed through a C2H2 cell, and the laser fre-quency deviation from the resonance center of theC2H2 molecules (the first derivative signal of the ab-sorption line) can be converted into a voltage signalusing phase-sensitive detection. With this approach aphase modulation frequency of 10 MHz is used tomaintain a sufficient frequency modulation width.This fast modulation was achieved with a LiNbO3(LN) phase modulator. A common way of stabilizingthe laser frequency is to feed the error signal back tothe pump power of the laser itself. The refractive in-dex of the gain medium changes slightly, dependingon the pump power level, and this feature can beused for optical frequency control. It is important tonote that we can stabilize the optical frequency with

Fig. 1. Configuration of absolutely frequency-stabilizedand repetition-rate-controlled, 40 GHz regenerativelymode-locked erbium fiber laser at 1.53 �m. The left-handside shows the regeneratively mode-locked laser, and theright-hand side shows the frequency stabilization part in-cluding a feedback circuit to the laser. (PZT, piezoelectrictransducer; DBM, double balanced mixer; FBG, fiber Bragggrating; MLP, multilayer piezo; LPF, low-pass filter; EDF,erbium-doped fiber; LN, lithium niobate; PI, proportionaland integral; LD, laser diode).

this method, but it inevitably causes a change in the

laser output power. Therefore, to realize coarse tun-ing when stabilizing the optical frequency, we newlyintroduce laser cavity temperature control, becausemost frequency fluctuations result from cavity fluc-tuations caused by temperature, pressure, and vibra-tions. Small changes in the pump power were used tofine-tune the laser frequency, so that the outputpower became almost constant. The contribution ofany cavity length change to the repetition ratechange is completely absorbed by the PLL, and therepetition rate is controlled with the same stabilityas with a synthesizer.

The 40 GHz fiber ring laser shown in Fig. 1 con-sists of a 4 m long erbium-doped fiber and an LN in-tensity modulator for active mode-locking. We in-stalled a PZT controller to provide constant initiallaser frequency tuning for optical frequency stabiliza-tion, but no feedback signal was employed this time.The laser cavity was similar to that reported in [12];however, the laser performance was slightly im-proved in the present experiment. The output lasercharacteristics are shown in Fig. 2, where the outputpower, pulse width, −3 dB spectral width, and signal-to-noise ratio of the extracted clock were 7.7 mW[Fig. 2(a)], 3.0 ps [Fig. 2(b)], 1.26 nm [Fig. 2(c)], and−80 dB [Fig. 2(d)], respectively. As described in [12],we installed a fiber-type etalon filter to eliminate op-tical frequency mode hopping from the harmonicmode locking. To extract one of the longitudinalmodes, we selected a longitudinal mode near1538.80 nm as shown by the arrow near the bottomof Fig. 2(c).

Figure 3 shows the change in the optical frequencywhen the frequency stabilization circuits for the opti-cal frequency and repetition rate were switched on.The optical frequency fluctuation was detected bymeasuring the beat frequency between the presentlaser and a single-frequency cw C2H2-stabilized laser

Fig. 2. Laser output characteristics. (a) Output power ver-sus pump power, (b) autocorrelation waveform, (c) opticalspectrum of the mode-locked laser, and (d) an extracted

40 GHz clock signal in the regenerative mode locking.

November 15, 2008 / Vol. 33, No. 22 / OPTICS LETTERS 2643

that we fabricated [11]. The frequency stability of oursingle-frequency laser was approximately 1.3�10−11

for �=1 s and 2.0�10−11 for �=100 s. In Fig. 3(a), it isimportant to note that even after 3600 s there was nolaser frequency drift, which indicates that the fre-quency of the 40 GHz optical pulse had been success-fully stabilized. However, one drawback of the systemis the change in the laser output power as shown inFig. 3(a). A pump power change of approximately6 mW occurred for an initial pump power of 100 mW.This power change was caused by full control of theoptical frequency by only adopting the pump powerchange. Less control of the pump power provides bet-ter amplitude stability as regards the laser outputpower. Therefore, we also fed back the error signal tothe temperature control of the laser cavity itself,which worked very well to reduce the change in thepump power. A small feedback signal was fed back tothe pump power to fine tune the optical frequency,but the frequency drift component was eliminated bythe cavity temperature control. With this dual feed-back control, the change in the pump power was re-duced to less than 0.1 mW, which means that theoutput power level is constant. Under this conditionwe successfully achieved a constant output laserpower while simultaneously stabilizing the frequencyand repetition rate.

The frequency stability of the present laser can befound in Fig. 4, where we show the square root of theAllan variance versus integration time �. The fre-quency stability between the present mode-locked fi-

Fig. 4. Optical-frequency stability of the present mode-locked fiber laser. Square root of Allan variance versus in-

Fig. 3. Changes in the optical frequency and the laser outa pump power change was used for the optical-frequency staBoth pump power and cavity temperature controls are ado

tegration time �.

ber laser (MLFL) and frequency-stabilized cw fiberlaser (CWFL) is shown by a filled circle, and that be-tween the two CWFLs we fabricated is shown by afilled triangle. Figure 4 clearly shows that the opticalfrequency stability of the MLFL in the present case isalmost the same as that of the CWFL and reached2�10−11 for �=10–100 s. Simultaneously, the stabil-ity of the repetition rate is the same as that of thesynthesizer, which is 10−10–10−11.

In conclusion, we have successfully constructed anew absolutely frequency-stabilized, mode-locked la-ser at 40 GHz. The output power was 7–8 mW for a100 mW pump and the output pulse width was 3 ps.The laser is a regeneratively and harmonically mode-locked erbium-doped fiber ring laser operating at1.53 �m, in which the optical frequency is locked to aC2H2 linear absorption [P(10) line], and the repeti-tion rate is locked to a synthesizer by using the PLLtechnique. The obtained optical frequency stabilityof the pulse train was as high as 2�10−11 for�=10–100 s.

References

1. M. Poulin, C. Latrasse, N. Cyr, and M. Têtu, IEEEPhotonics Technol. Lett. 9, 1631 (1997).

2. M. Nakazawa, M. Yoshida, K. Kasai, and J. Hongou,Electron. Lett. 42, 710 (2006).

3. A. Onae, K. Okumura, J. Yoda, K. Nakagawa, A.Yamaguchi, M. Koutogi, K. Imai, and B. Eidiyatomoko,IEEE Trans. Instrum. Meas. 48, 563 (1999).

4. T. J. Quinn, Metrologia 40, 103 (2003).5. M. Nakazawa, E. Yoshida, and K. Tamura, Electron.

Lett. 33, 1318 (1997).6. M. Nakazawa and E. Yoshida, IEEE Photonics Technol.

Lett. 12, 1613 (2000).7. M. Nakazawa and M. Yoshida, Opt. Lett. 33, 1059

(2008).8. D. J. Jones, S. A. Diddams, J. K. Ranka, A. Stentz, R.

S. Windeler, J. L. Hall, and S. T. Cundiff, Science 288,635 (2000).

9. B. R. Washburn, S. A. Diddams, N. R. Newbury, J. W.Nicholson, M. F. Yan, and C. G. Jørgensen, Opt. Lett.29, 250 (2004).

10. K. Kasai, M. Yoshida, and M. Nakazawa, in 30thECOC 2004 (IEEE, 2004), paper Th1.3.5.

11. K. Kasai, A. Suzuki, M. Yoshida, and M. Nakazawa,IEICE Electron. Express 3, 487 (2006).

12. M. Yoshida, K. Kasai, and M. Nakazawa, IEEE J.Quantum Electron. 43, 704 (2007).

power when feedback stabilization was operating. (a) Onlyation, and the laser power fluctuation was not removed. (b)to obtain a constant laser output power.

putbilizpted