Solid-state low-loss intracavity saturable absorberfor Nd:YLF lasers: an antiresonant semiconductor
Fabry-Perot saturable absorber
U. Keller, D. A. B. Miller, G. D. Boyd, T. H. Chiu, J. F. Ferguson, and M. T. Asom
AT&T Bell Laboratories, Crawfords Corner Road, Holmdel, New Jersey 07733
Received November 26, 1991
We introduce a new low-loss fast intracavity semiconductor Fabry-Perot saturable absorber operated at anti-resonance both to start and sustain stable mode locking of a cw-pumped Nd:YLF laser. We achieved a 3.3-pspulse duration at a 220-MHz repetition rate. The average output power was 700 mW with 2 W of cw pumppower from a Ti:sapphire laser. At pump powers of less than 1.6 W the laser self-Q switches and produces 4-pspulses within a 1.4-,us Q-switched pulse at an '150-kHz repetition rate determined by the relaxation oscillationof the Nd:YLF laser. Both modes of operation are stable. In terms of coupled-cavity mode locking, the intra-cavity antiresonant Fabry-Perot saturable absorber corresponds to monolithic resonant passive mode locking.
Many solid-state lasers (i.e., rare-earth andtransition-metal lasers) exhibit a small gain crosssection and therefore require a fast and low-loss sat-urable absorber for passive mode locking. Withinan all-solid-state ultrafast laser technology, semi-conductor saturable absorbers have the advantagethat they are compact, fast, and cover band gapsfrom the visible to the infrared. Semiconductorsaturable absorbers have previously mode lockeddiode' and color-center lasers.2'3 However, in rare-earth (i.e., Nd:YLF) and transition-metal (i.e.,Ti:sapphire) solid-state lasers, intracavity semicon-ductors introduce too much loss, have a too smallsaturation intensity, and have problems withstand-ing the high intracavity peak intensities. As a re-sult, we have used semiconductors inside a low-Qcoupled cavity, referred to as resonant passive modelocking 4 -7 (RPM). However, RPM without an activecavity-length control is self-stabilized only at the ex-pense of small optical frequency fluctuations.4' 5 Anintracavity passive mode-locking technique removesthe active stabilization requirement for stable modelocking, and the overall cavity design becomes morecompact. More recently an intracavity reactivenonlinearity, such as self-focusing, has been used toproduce fast saturable absorberlike mode lockingthat is, however, not self-starting.8 9
In this Letter we introduce an intracavity satu-rable absorber element, an antiresonant Fabry-Perot saturable absorber (A-FPSA), that has arelatively fast InGaAs/GaAs semiconductor satu-rable absorber monolithically integrated betweentwo reflecting mirrors. The top reflector of theA-FPSA is a TiO2/SiO2 dielectric mirror with a 98%reflectivity. The use of the A-FPSA effectivelytransforms the semiconductor to a high-saturation-intensity, low-loss absorber as required. With thisnew A-FPSA inside a Nd:YLF laser cavity, weproduce self-starting 3.3-ps cw mode-locked pulsesat a 1.04 7-,4Lm wavelength with an average output
power of 700 mW when the laser is cw pumped with2 W of power.
At antiresonance the intensity inside the Fabry-Perot cavity is always smaller than the incident in-tensity, which increases the effective saturationintensity (observed before the top mirror) and thedamage threshold of the semiconductor saturableabsorber. We obtain no bandwidth limitations atantiresonance, because the free spectral range ismuch larger than the gain bandwidth of the Nd:YLFlaser. Additionally, the antiresonance operation isinsensitive to thermal loading and provides loose de-sign tolerances. In contrast, a FPSA operated atresonance has a higher intensity inside the Fabry-Perot cavity, has critical design tolerances, and canexhibit bistability effects, which are detrimental forour application.
The saturable absorber inside the Fabry-Perot cav-ity has a carrier lifetime that is slightly longer thanthe pulse duration but much shorter than the pulserepetition period. Therefore, under cw mode-lockedoperation, the absorption bleaching, and hence thereflectivity, is increased owing to the increasednumbers of carriers generated within one laser pulseduration. The dielectric top mirror significantly re-duces the net reflectivity change, however, protectsthe semiconductor from the high intracavity inten-sity, and significantly reduces the intracavity losses.The nonlinear reflectivity is still large enough tostrongly mode lock the Nd:YLF laser. In addition,the intracavity A-FPSA can be explained in terms ofcoupled-cavity mode locking. In fact, the A-FPSAcorresponds to monolithic RPM, which is discussedlater in this Letter.
Figure 1 shows the laser cavity setup. The laserconsists of a standard, end-pumped, linear-folded,astigmatically compensated cavity. A 5-mmNd:YLF rod has its flat end coated for high reflec-tion at the lasing wavelength of 1.047 ,um and antire-flection coated for the pump wavelength. The other
eso 950 1 too 1 oss 1t1 e 11 5s 1200Wavelength, nm
Fig. 2. Low-intensity reflectivity: curve (a), AlAs/GaAsdielectric mirror; curve (b), InGaAs/GaAs saturable ab-sorber layer (uncoated) and AlAs/GaAs dielectric mirror;curve (c), A-FPSA with a free spectral range of s138 nm.
end is cut at Brewster's angle. The A-FPSA, indiumsoldered onto a copper heat sink, forms the end mir-ror of the laser cavity. A highly reflecting mirrorwith a 100-mm radius of curvature focuses the laserbeam onto the A-FPSA to produce a calculated beamdiameter of 80 Aum. The cavity output coupler con-sists of a flat turning mirror at a 450 angle of inci-dence and 1% transmission, which results in a totaloutput coupling of 2%. The laser is pumped with acw Ti:sapphire laser at a wavelength of 798 nm. Wemeasured a slope efficiency of 35% (the sum of bothoutput beams) and a pump threshold of 43 mW. AZ-shaped cavity design with a Nd:YLF crystal cuton both sides at Brewster's angle would produce onlyone output beam and hence would be more desirable.
The A-FPSA consists of a GaAs/AlAs dielectricmirror (16 periods of 76.4-nm GaAs layers and90.5-nm AlAs layers) with a reflectivity of =:s96%,grown by molecular-beam epitaxy at normal tem-peratures (=64 0'C) on a GaAs substrate. The satu-rable absorber layer, 0.61 ,um thick with a band gapclose to the 1.047-Atm Nd:YLF lasing wavelength, isgrown at low temperatures of =:z380 "C on top of theGaAs/AlAs dielectric mirror and consists of 50 peri-ods of GaAs barriers and In.Gal-.As quantum wells(x = 0.29) with nominal thicknesses of 6 and6.2 nm, respectively. Finally, a TiO2/SiO2 dielectricmirror with 98% reflectivity is evaporated onto theabsorber layer. The low-temperature growth re-duces the carrier lifetime and produces a relativelyfast saturable absorber. We measured a 69-ps car-rier lifetime of the low-temperature InGaAs/GaAssaturable absorber using our newly developedpassively mode-locked Nd:YLF laser in a standardnoncollinear pump-probe experiment. Thus the re-sponse time of the saturable absorber is somewhatlonger than the 3.3-ps pulse duration but muchsmaller than the pulse repetition rate of =:s-4.5 ns.
Figure 2 demonstrates the FPSA operation at an-tiresonance. We measured the wavelength depen-
dence of the GaAs/AlAs dielectric mirror [curve (a)]through the GaAs substrate, which clearly showsthat the mirror is centered at -1.03 /-cm with aFWHM of ;ss120 nm. Before we deposited the topmirror, we measured [curve (b)] the reflectionthrough the saturable absorber layer; this curveshows a strong Fabry-Perot resonance peak at988 nm and the absorption edge of the InGaAs/GaAslayer at approximately 1.05 ,um. The 988-nm peakcorresponds to a Fabry-Perot resonance with a7-half-wavelength spacing (i.e., the seventh reso-nance order), if we take into account the 1.56-jimoptical penetration depth into the AlAs/GaAs dielec-tric mirror and the 1.9-Am optical thickness of theInGaAs/GaAs multiple-quantum-well layer. Thethickness of the InGaAs/GaAs layer was chosensuch that an antiresonance of the final FPSA isclose to the Nd:YLF lasing wavelength of 1.047 gtm.The final A-FPSA reflectivity [curve (c)] shows aneighth-order Fabry-Perot resonance peak at 992 nmif we take into account a 0.5-,um optical penetrationdepth into the TiO2/SiO2 dielectric mirror. The to-tal optical thickness of this Fabry-Perot structure is4 pum, which produces a free spectral range of138 nm and an antiresonance at 1.06 ,m close tothe laser wavelength of 1.047 Atm.
To monitor the laser operation fully we used a fastphotodiode on a sampling scope and on a microwavespectrum analyzer, a noncollinear real-time autocor-relator, and an optical spectrum analyzer. Figure 3shows a 3.3-ps measured autocorrelation (solid curve)and an ideal hyberbolic-secant autocorrelation(dashed curve) for comparison. We measured theautocorrelation trace while the sampling scope wassimultaneously monitored, which indicated a cleandetection-system-limited pulse width of =40 ps.The optical spectrum had a FWHM of 0.66 nm,which reveals that the pulses were 1.9 times trans-form limited.
The microwave spectrum analyzer on a 1-MHzfrequency span and a 10-kHz resolution bandwidthcentered on the pulse repetition rate of 219.4 MHzdemonstrate unambiguously that the laser is notself-Q switched [Fig. 4(a)]. A Nd:YLF laser self-Q switches at the relaxation oscillation rate, whichproduces extremely strong modulation sidebands ona microwave spectrum analyzer. At a cw pumppower of 2.2 W, Fig. 4(a) shows that the relaxationoscillations are approximately 55 dB below the firstlaser harmonic at a 10-kHz resolution bandwidth.This is typical for any active or passive mode-locked
0.
0.
0.
0.0-20 -10 0 10 20
Delay time, ps
Fig. 3. Autocorrelation of the cw mode-locked Nd:YLFlaser.
(b)Fig. 4. Microwave spectrum analyzer results: relaxa-tion oscillations appear as modulation sidebands aroundthe pulse repetition frequency of 219.4 MHz for (a) a cwmode-locked Nd:YLF laser (2.2-W pump power) and (b) acw mode-locked and Q-switched Nd:YLF laser (1.4-Wpump power). The frequency span is 1 MHz, and theresolution bandwidth is 10 kHz.
Nd:YLF lasers. Higher pump power results insmaller relaxation oscillation strength. At a pumppower of less than 1.6 W but greater than 0.8 W,the laser is stably self-Q switched [Fig. 4(b)], pro-ducing 4-ps cw mode-locked pulses within a 1.4-pusQ-switched pulse at the relaxation oscillation rate ofthe Nd:YLF laser. For both cases we measure apeak-to-peak amplitude noise of no more than 2%observed over 250 ms on an oscilloscope with agreater than 1-MHz detection bandwidth. Thehigh-frequency AM noise above a few hundred kilo-hertz is limited by the Ti:sapphire pump laser. It isimportant to note that spurious intracavity reflec-tions strongly affect the stability of the passivelymode-locked Nd:YLF laser. For example, a slightlywedged antireflection-coated Nd:YLF laser rod or afocusing lens instead of a curved highly reflectingmirror prevented mode locking.
The method of operating a Fabry-Perot structureat antiresonance is not obvious and has been moti-vated by our previous research with RPM. 7 Interms of coupled-cavity mode locking, the A-FPSAforms a monolithic coupled-cavity configuration forwhich the A-FPSA determines the end mirrors forthe two coupled cavities that spatially overlap. Thetop mirror of the A-FPSA forms the end mirror ofthe main cavity, and the dielectric mirror under-neath the saturable absorber layer forms the endmirror of the nonlinear coupled cavity. Because ofmonolithic integration of these two end mirrorswithin the same element, no relative cavity-lengthfluctuations exist. Thus there is no need for activecavity-length stabilization. In addition, theA-FPSA design minimizes the cavity-length differ-ence and hence the pulse duration. Although this
formal relation with RPM exists, it is simpler andequally correct to view the entire A-FPSA as an ef-fective intracavity saturable absorber.
In conclusion, we have demonstrated a stable, self-starting, intracavity passive mode-locked Nd:YLFlaser and produced a 3.3-ps pulse duration. In addi-tion, stable Q switching is achieved. The intracav-ity A-FPSA represents a saturable absorber forwhich we can custom design the effective saturationintensity for a given material by simply choosing theappropriate top reflector. Because the FPSA is op-erated at antiresonance, we obtain a low insertionloss (top reflector is 98%), a high damage threshold,no bandwidth limitation (free spectral range of-138 nm, which is much greater than the gainbandwidth of the laser of -0.7 nm), a high effectivesaturation intensity, insensitivity to thermal load-ing, loose design tolerances, and the possibility ofdesigning a broadband saturable absorber. Both atcw and at mode-locked operation the laser mode is aclean TEMoo mode as observed with a microwavespectrum analyzer and an infrared viewer. Thecavity alignment is optimized for maximum power,and no intracavity aperture or slit has been used.However, some small self-focusing mode-lockingcontributions might affect the steady-state pulse du-ration owing to the effective gain aperture intro-duced by longitudinal laser pumping,8-1' which isunder further investigation.
We thank Dave G. Coult for the dielectric coatingand Lightwave Electronics Corporation for theNd:YLF laser crystal.
M. T. Asom is with AT&T Bell Laboratories, 9999Hamilton Boulevard, Breinigsville, Pennsylvania18031-9359.
References
1. P. W Smith, Y. Silberberg, and D. A. B. Miller, J. Opt.Soc. Am. B 2, 1228 (1985).
2. M. N. Islam, E. R. Sunderman, I. Bar-Joseph,N. Sauer, and T. Y. Chang, Appl. Phys. Lett. 54, 1203(1989).
3. P. N. Kean, X. Zhu, D. W Crust, R. S. Grant, N.Langford, and W Sibbett, Opt. Lett. 14, 48 (1989).
4. U. Keller, W H. Knox, and H. Roskos, Opt. Lett. 15,1377 (1990).
5. H. A. Haus, U. Keller, and W H. Knox, J. Opt. Soc.Am. B 8, 1252 (1991).
6. U. Keller, T. K. Woodward, D. L. Sivco, and A. Y. Cho,Opt. Lett. 16, 390 (1991).
7. U. Keller and T. H. Chiu, "Resonant passive mode-locked NdtYLF laser," IEEE J. Quantum Electron. (tobe published).
8. D. E. Spence, P. N. Kean, and W Sibbett, Opt. Lett.16, 42 (1991).
9. U. Keller, G. W 'tHooft, W H. Knox, and J. E.Cunningham, Opt. Lett. 16, 1022 (1991).
10. D. K. Negus, L. Spinelli, N. Goldblatt, and G. Feugnet,in Digest of Conference on Advanced Solid-StateLasers (Optical Society of America, Washington, D.C.,1991), p. 120.
11. F Salin, J. Squier, and M. Pich6, Opt. Lett. 16, 1674(1991).
