+ All Categories
Home > Documents > Self-stabilization of a mode-locked femtosecond fiber laser using a photonic bandgap fiber

Self-stabilization of a mode-locked femtosecond fiber laser using a photonic bandgap fiber

Date post: 08-Oct-2016
Category:
Upload: dmitry
View: 212 times
Download: 0 times
Share this document with a friend
3
Self-stabilization of a mode-locked femtosecond fiber laser using a photonic bandgap fiber Xiaomin Liu, Jesper Lægsgaard, and Dmitry Turchinovich* DTU Fotonik–Department of Photonics Engineering, Technical University of Denmark, DK-2800 Kgs. Lyngby, Denmark * Corresponding author: [email protected] Received October 27, 2009; revised January 18, 2010; accepted February 2, 2010; posted February 19, 2010 (Doc. ID 119146); published March 18, 2010 We demonstrate a self-stabilization mechanism of a semiconductor saturable absorber mode-locked linear- cavity Yb-doped fiber laser using an intracavity photonic bandgap fiber. This mechanism relies on the spec- tral shift of the laser pulses to a spectral range of higher anomalous dispersion and higher loss of the pho- tonic bandgap fiber, as a reaction to the intracavity power buildup. This, in particular, results in a smaller cavity loss for the stably mode-locked laser, as opposed to the Q-switched mode-locking scenario. The laser provides stable 39–49 pJ pulses of around 230 fs duration at 29 MHz repetition rate. © 2010 Optical Society of America OCIS codes: 060.3510, 060.5295, 140.7090, 140.3425. Mode-locked (ML) ultrafast fiber lasers nowadays successfully compete with their solid-state counter- parts, but their operational and environmental sta- bility is often a critical issue. Different stabilization schemes were demonstrated recently, for example, the introduction of a spectral filter in nonlinear po- larization evolution mode-locked (NPE-ML) ring- cavity lasers [1] or the incorporation of a fiber Bragg grating (FBG) in semiconductor saturable absorber mode-locked (SESAM-ML) linear cavity lasers [24]. The stabilization of a ML fiber laser aims at reduc- ing the environmental (i.e., mechanical and thermal) sensitivity of the laser, and at providing favorable op- erational conditions for stable mode-locking, as op- posed to chaotic Q-switched mode locking. To achieve a good environmental stability, free-space optical cou- pling in the cavity should preferably be avoided. This, however, has not yet been demonstrated in the NPE-ML femtosecond lasers, even though the NPE-ML method allows for support of high optical bandwidth and providing dechirped sub-100 fs pulses [1,5]. On the other hand, a fully monolithic SESAM-ML fiber laser with an exceptional environ- mental and operational stability has been demon- strated [4], where a narrowband FBG acted as an in- tracavity stabilizing element. However, the oscillator bandwidth in this case was sacrificed for the sake of operational stability, leading to the generation of transform-limited pulses of around 5 ps. In this Letter, we demonstrate a fully monolithic, i.e., an all-fiber SESAM-ML laser, delivering near- transform-limited pulses of around 230 fs with an ex- ceptional environmental and operational stability. The outline of our laser is shown in Fig. 1(a). It has a linear cavity, confined between a broadband fiber- pigtailed highly reflective mirror on one end and a fiber-pigtailed semiconductor saturable absorber (SESAM) by BATOP GmbH with 24% modulation depth and 500 fs recovery time on the other end. A polarization-maintaining (PM) Yb-doped single-mode fiber (SMF) was used as gain medium, and a PM pho- tonic bandgap fiber (PBGF) was used for intracavity dispersion management and stabilization. This spliced-in PM PBGF was a hybrid all-solid total in- ternal reflection/bandgap fiber from Crystal Fibre A/S [6]. A similar approach to intracavity dispersion management, but without the PM feature, was ear- lier shown in [7,8]. The rest of the cavity consisted of the standard PM SMF. The total cavity length was 3.55 m, of which 1.21 m had anomalous dispersion at the laser operating wavelength. The chosen fiber combination provided small net anomalous cavity dispersion for the laser pulses, and the laser was op- erating in a weakly stretched pulse regime [9]. The loss of all fibers and other elements of the cavity, ex- cept for the PM PBGF, was wavelength independent. The laser repetition rate at the fundamental ML op- eration was 28.77 MHz [see Fig. 1(b)]. The laser output power as a function of increasing pump power is shown in Fig. 2, with different laser ML regimes indicated. The lasing begins at a pump power of 47 mW in the Q-switched ML regime. At around 65 mW, the laser enters a stable self-starting fundamental ML operation, which is maintained un- til approximately 75 mW of pump power, resulting in stable pulses with the energies in the range 39–49 pJ. At further increase in the pump power the laser again enters a phase of Q-switched mode locking, Fig. 1. (Color online) (a) Layout of the laser: HR, high- reflectivity broadband pigtailed mirror; WDM, 980/1030 wavelength division multiplexer; LD, single-mode pumping diode at 974 nm; PM PBGF, PM all-solid photonic bandgap fiber; SESAM, semiconductor saturable absorber mirror; PFC, 80/20 polarization filter coupler. (b) Laser pulse train at stable single-pulse ML operation. April 1, 2010 / Vol. 35, No. 7 / OPTICS LETTERS 913 0146-9592/10/070913-3/$15.00 © 2010 Optical Society of America
Transcript

April 1, 2010 / Vol. 35, No. 7 / OPTICS LETTERS 913

Self-stabilization of a mode-locked femtosecondfiber laser using a photonic bandgap fiber

Xiaomin Liu, Jesper Lægsgaard, and Dmitry Turchinovich*DTU Fotonik–Department of Photonics Engineering, Technical University of Denmark,

DK-2800 Kgs. Lyngby, Denmark*Corresponding author: [email protected]

Received October 27, 2009; revised January 18, 2010; accepted February 2, 2010;posted February 19, 2010 (Doc. ID 119146); published March 18, 2010

We demonstrate a self-stabilization mechanism of a semiconductor saturable absorber mode-locked linear-cavity Yb-doped fiber laser using an intracavity photonic bandgap fiber. This mechanism relies on the spec-tral shift of the laser pulses to a spectral range of higher anomalous dispersion and higher loss of the pho-tonic bandgap fiber, as a reaction to the intracavity power buildup. This, in particular, results in a smallercavity loss for the stably mode-locked laser, as opposed to the Q-switched mode-locking scenario. The laserprovides stable 39–49 pJ pulses of around 230 fs duration at 29 MHz repetition rate. © 2010 Optical Societyof America

OCIS codes: 060.3510, 060.5295, 140.7090, 140.3425.

Mode-locked (ML) ultrafast fiber lasers nowadayssuccessfully compete with their solid-state counter-parts, but their operational and environmental sta-bility is often a critical issue. Different stabilizationschemes were demonstrated recently, for example,the introduction of a spectral filter in nonlinear po-larization evolution mode-locked (NPE-ML) ring-cavity lasers [1] or the incorporation of a fiber Bragggrating (FBG) in semiconductor saturable absorbermode-locked (SESAM-ML) linear cavity lasers [2–4].

The stabilization of a ML fiber laser aims at reduc-ing the environmental (i.e., mechanical and thermal)sensitivity of the laser, and at providing favorable op-erational conditions for stable mode-locking, as op-posed to chaotic Q-switched mode locking. To achievea good environmental stability, free-space optical cou-pling in the cavity should preferably be avoided.This, however, has not yet been demonstrated in theNPE-ML femtosecond lasers, even though theNPE-ML method allows for support of high opticalbandwidth and providing dechirped sub-100 fs pulses[1,5]. On the other hand, a fully monolithicSESAM-ML fiber laser with an exceptional environ-mental and operational stability has been demon-strated [4], where a narrowband FBG acted as an in-tracavity stabilizing element. However, the oscillatorbandwidth in this case was sacrificed for the sake ofoperational stability, leading to the generation oftransform-limited pulses of around 5 ps.

In this Letter, we demonstrate a fully monolithic,i.e., an all-fiber SESAM-ML laser, delivering near-transform-limited pulses of around 230 fs with an ex-ceptional environmental and operational stability.The outline of our laser is shown in Fig. 1(a). It has alinear cavity, confined between a broadband fiber-pigtailed highly reflective mirror on one end and afiber-pigtailed semiconductor saturable absorber(SESAM) by BATOP GmbH with 24% modulationdepth and 500 fs recovery time on the other end. Apolarization-maintaining (PM) Yb-doped single-modefiber (SMF) was used as gain medium, and a PM pho-

tonic bandgap fiber (PBGF) was used for intracavity

0146-9592/10/070913-3/$15.00 ©

dispersion management and stabilization. Thisspliced-in PM PBGF was a hybrid all-solid total in-ternal reflection/bandgap fiber from Crystal FibreA/S [6]. A similar approach to intracavity dispersionmanagement, but without the PM feature, was ear-lier shown in [7,8]. The rest of the cavity consisted ofthe standard PM SMF. The total cavity length was3.55 m, of which 1.21 m had anomalous dispersion atthe laser operating wavelength. The chosen fibercombination provided small net anomalous cavitydispersion for the laser pulses, and the laser was op-erating in a weakly stretched pulse regime [9]. Theloss of all fibers and other elements of the cavity, ex-cept for the PM PBGF, was wavelength independent.The laser repetition rate at the fundamental ML op-eration was 28.77 MHz [see Fig. 1(b)].

The laser output power as a function of increasingpump power is shown in Fig. 2, with different laserML regimes indicated. The lasing begins at a pumppower of 47 mW in the Q-switched ML regime. Ataround 65 mW, the laser enters a stable self-startingfundamental ML operation, which is maintained un-til approximately 75 mW of pump power, resulting instable pulses with the energies in the range 39–49pJ. At further increase in the pump power the laseragain enters a phase of Q-switched mode locking,

Fig. 1. (Color online) (a) Layout of the laser: HR, high-reflectivity broadband pigtailed mirror; WDM, 980/1030wavelength division multiplexer; LD, single-mode pumpingdiode at 974 nm; PM PBGF, PM all-solid photonic bandgapfiber; SESAM, semiconductor saturable absorber mirror;PFC, 80/20 polarization filter coupler. (b) Laser pulse train

at stable single-pulse ML operation.

2010 Optical Society of America

914 OPTICS LETTERS / Vol. 35, No. 7 / April 1, 2010

upon which a stable harmonic mode locking phase oc-curs at 87 mW: double-pulsing until 104 mW andtriple-pulsing at higher pump power. The same MLstates were observed when the pump power was bothincreasing and decreasing. However, the stable MLpump power range was extended by 2–3.5 mW whenthe pump power was decreasing, as compared to thepump power increase shown in Fig. 2. The slope effi-ciency of the laser in all ML regimes was about 4%,typical for this type of oscillators. The clearly visibledrops in the output power at the Q-switched ML op-eration suggest the existence of a self-stabilizationmechanism that favors a stable ML operation by pro-viding a smaller cavity loss for this regime (see Fig.2). The results of long-term and variable temperaturetests, showing excellent environmental and opera-tional stability of the laser in the stable ML regime:less than 0.38% output power fluctuations duringmore than 6 h of operation and stable ML at tem-perature sweeps of 10°C–40°C, also while movingthe fibers, will be reported elsewhere.

In Fig. 3(a) the output spectra of the laser areshown for the stable fundamental ML regime at dif-ferent pump powers. The spectra broaden from 6 to 7nm bandwidth at FWHM as the pump power grows.At the same time, more spectral weight is being builtup on the longer-wavelength side of the spectrum. InFig. 3(b) a transmittivity of the PM PBGF and thenet cavity dispersion (NCD) of the laser on one roundtrip are shown. To calculate the NCD, the dispersionof the PBGF was estimated by spectral white light in-terferometry [10]. The NCD on one round trip is zeroat 1024 nm, and at the laser central wavelength of1033 nm it is 0.089 ps/nm.

With increasing pump power, the stronger pulsesin the laser cavity undergo stronger self-phase modu-lation, which leads to the observed spectral broaden-ing. To preserve the operation in the regime ofanomalous NCD, the spectrally broadened pulseneeds to expand further away from the dispersionzero, i.e., to the long-wavelength side. Both spectral

Fig. 2. (Color online) Laser output power as a function ofincreasing pump power, indicating different mode-lockingregimes. Dashed lines are linear fits to laser efficiency instable and Q-switched mode-locking regimes.

broadening and redshift will lead to the situation,

where more and more spectral bandwidth will bebuilt up in the longer-wavelength region where thetransmission loss of the PBGF is higher, which willin turn lead to optical limiting of the laser signal.Such a self-stabilization mechanism, provided by theunique combination of a low dispersion, a high rela-tive dispersion slope, and a growing loss of the PBGFat higher wavelengths, prevents the early onset ofQ-switched mode locking. The saturation of the out-put power at the stable fundamental ML regime withgrowth in the pump power in the range 71–75 mW,preceding the Q-switched ML phase, is a particularresult of such an optical limiting (see Fig. 2). It iswell known that gain filtering or other types of spec-tral filtering can stabilize against Q-switched modelocking [11]. The current implementation is some-what similar to the one demonstrated by us in the pi-cosecond lasers with narrowband FBGs [3,4]. How-ever, when a PBGF is used, pulses of much higherbandwidth can be stabilized, resulting in direct fem-tosecond output of the oscillator. We note that thesmaller the relative dispersion slope of the PBGF is,the higher the bandwidth that can in principle bemode locked.

We performed a full numerical modeling of our la-ser by solving the nonlinear Schrödinger equation forfiber propagation and rate equations for the gain andsaturable absorber [3]. In Fig. 4(a) the output pulseenergy is shown as a function of the cavity round-tripnumber. When the wavelength-dependent cavity losscorresponding to the measured transmittivity of thePBGF is used, the pulse energy is stable. Once such awavelength-dependent cavity loss is replaced in ourmodel with a constant wavelength-independent lossafter 5000 round trips, the stable laser pulses quickly

Fig. 3. (Color online) (a) Laser output spectra in stablefundamental ML regime at variable pump power. (b) Trans-mittivity of PM PBGF and net cavity round-trip dispersion.

break down. The breakdown behavior differs depend-

April 1, 2010 / Vol. 35, No. 7 / OPTICS LETTERS 915

ing on the value of the constant loss used. However,for all loss values shown here (0.71–0.74 dB), thestable ML operation disrupts within less than 100cavity round trips, i.e., in less than 3.5 �s for this la-ser. The value of 0.74 dB corresponds to the averageloss of our PBGF. The calculated cavity loss as a func-tion of the output pulse energy is shown in Fig. 4(b),taking into account a combined action of the SESAM,gain filtering, and optical limiting by the PBGF. Wenote here that the influence of gain filtering alone onthe cavity loss dynamics is insignificant, and the cav-ity loss minimum needed for a stable pulse formationis provided by the combined action of the signal-enhancing SESAM and optical-limiting PBGF. Themodeling predicts the onsets of stable and ofQ-switched mode locking at the output pulse energiesof around 28 and 31.5 pJ, respectively, and a stableML pulse energy range of 3.5 pJ. This is in reason-able agreement with experimental values of 39, 49,

Fig. 4. (Color online) (a) Calculated output pulse energydepending on a cavity round-trip number. Solid line,wavelength-dependent cavity loss, corresponding to PBGFtransmittivity. Dashed, dashed-dotted, and dotted curves,constant cavity losses of 0.71, 0.72, and 0.74 dB, respec-tively, introduced after 5000 round trips with PBGF cavityloss. (b) Calculated cavity loss as a function of output pulseenergy—see text for details. (c) AC of the laser pulse, mea-sured after 1 m PM SMF pigtail of the outcoupler (circles),and its Gaussian fit (solid curve).

and 10 pJ, respectively. The modeled spectral dynam-

ics of the stable ML pulses featured both spectralbroadening and redshift of the laser signals with in-creasing pulse energy. This is in full agreement withour observations and our qualitative considerationsabout the nature of the self-stabilization mechanismof the laser. Finally, in Fig. 4(c) we show the experi-mental autocorrelation (AC) of the 45 pJ pulse mea-sured after 1-m-long PM SMF pigtail following thecavity outcoupler. This AC is a near-perfect Gaussianand has an FWHM of 328 fs, corresponding to thepulse duration of 232 fs. Our modeling results indi-cate that the pulses are in fact recompressed in thenormal-dispersion fiber pigtail, since the cavity out-coupling follows shortly after the passage throughthe anomalous-dispersion PBGF.

In conclusion, we demonstrated a self-stabilizationmechanism of a ML femtosecond fiber laser, based ondistributed wavelength-dependent nonlinear opticallimiting in a PBG fiber, featuring both higher anoma-lous dispersion and higher loss at the longer-wavelength side of its transmission window. This inparticular leads to a delayed onset of Q-switchedmode locking. Giant Q-switch ML pulses with largerbandwidths will experience (i) more optical limitingon the red side and (ii) breakdown into a dispersivewave on the blue (normal-dispersion) side of thePBGF transmission window, both leading to highercavity losses. Our experimental observations are con-firmed by the results of numerical modeling.

We acknowledge Danish Advanced TechnologyFoundation (HTF) for financial support; L. Leick, T.V. Andersen, and D. Noordegraaf for valuable assis-tance; and NKT Photonics A/S for providing the all-solid hybrid PBGF.

References

1. J. Buckley, A. Chong, S. Zhou, W. Renninger, and F. W.Wise, J. Opt. Soc. Am. B 24, 1803 (2007).

2. I. Hartl, G. Imeshev, L. Dong, G. C. Cho, and M. E. Fer-mann, in CLEO/QELS and PhAST (Optical Society ofAmerica, 2005), paper CThG1.

3. J. Lægsgaard, J. Phys. B 41, 095401 (2008).4. D. Turchinovich, X. Liu, and J. Lægsgaard, Opt. Ex-

press 16, 14004 (2008).5. F. Ö. Ilday, J. Buckley, L. Kuznetsova, and F. Wise,

Opt. Express 11, 3550 (2003).6. J. K. Lyngsø, B. J. Mangan, and P. J. Roberts, in

CLEO/QELS and PhAST (Optical Society of America,2008), paper CThV1.

7. C. K. Nielsen, K. G. Jespersen, and S. R. Keiding, Opt.Express 14, 6063 (2006).

8. A. Isomäki and O. G. Okhotnikov, Opt. Express 14,4368 (2006).

9. H. A. Haus, K. Tamura, L. E. Nelson, and E. P. Ippen,IEEE J. Quantum Electron. 31, 591 (1995).

10. J. Y. Lee and D. Y. Kim, Opt. Express 14, 11608 (2006).11. S. Namiki, E. P. Ippen, H. A. Haus, and C. X. Yu, J.

Opt. Soc. Am. B 14, 2099 (1997).


Recommended