1276 OPTICS LETTERS / Vol. 29, No. 11 / June 1, 2004

Directly diode-pumped Kerr-lens mode-locked Cr4+:YAG laser

S. Naumov, E. Sorokin, and I. T. Sorokina

Institut für Photonik, Technische Universität Wien, Gusshausstrasse 27/387, A1040 Vienna, Austria

Received November 3, 2003

A directly diode-pumped Kerr-lens mode-locked Cr41:YAG laser is demonstrated for what is to our knowledgethe first time. Pulses as short as 65 fs with up to 30 mW of average output power, at a central wavelengthof 1569 nm, were obtained at a repetition rate of 100 MHz. Low-loss chirped mirrors have been used fordispersion compensation up to the third order. Comparison with an Yb-fiber-pumped configuration showsgood prospects for improvement. © 2004 Optical Society of America

OCIS codes: 140.3480, 140.7090.

Kerr-lens mode-locked (KLM) lasers, realized for thefirst time in 1991,1 have been intensively developedover the past decade and now are capable of producingalmost single-optical-cycle pulses. To make femtosec-ond technology more cost-effective and accessibleoutside the laboratory, diode pumping has been ex-tensively explored.2 Diode-pumped KLM lasers haveso far been realized only with Cr31-doped colquiriitecrystals (see, e.g., Ref. 3) in the 0.8-mm wavelengthrange and Yb31-doped media4 around 1 mm. In the1.5-mm range, diode-pumped femtosecond oscilla-tors have so far been represented by only Er-dopedfiber lasers, providing 300-fs pulses at 60 mW ofoutput power.5 To produce significantly shorterpulses, it is important to use a transition-metal-dopedmedium with broader bandwidth. One such crystalis Cr41:YAG, which has attracted a lot of attentionsince the late 1980s because of its ability to operate inthe water absorption window around 1.5 mm and toproduce pulses with durations as short as a few opticalcycles6,7 and up to a 4-GHz repetition rate.8 Usinglow-M2 pumping diodes to make a KLM laser is achallenging task. However, it is feasible, as witnessedby the successes in the 0.8-mm wavelength region,where 10- and 12-fs pulses have been demonstratedin directly diode-pumped KLM Cr:LiCAF (Ref. 9) andCr:LiSaF (Ref. 10), respectively. Until recently, thesesystems remained the only diode-pumped few-cycleKLM lasers.

With the progress in diode manufacturing tech-nology over the past few years, high-brightnesslaser diodes at 970 nm delivering up to 5 W froma 100-mm stripe have become available, makingdirect diode pumping of Cr41-doped materials pos-sible.11 – 13 Among these materials, so far onlydiode-pumped Cr41:YAG has demonstrated broadtunability and output power levels in excess of200 mW.12,13 Using a semiconductor saturable ab-sorber mirror (SESAM), we recently obtained thefirst, to our knowledge, diode-pumped passivelymode-locked operation, producing pulses of 62-fsduration at 15 mW of output power.14 But KLM tech-nology is ultimately the way to produce the shortestpulses and broadest spectra. In combination withdiode pumping, an ultrashort-pulsed KLM Cr:YAGlaser represents a significant advance in the field,because it offers a compact and cost-effective solution

0146-9592/04/111276-03$15.00/0

and may be used in such applications as opticalcoherence tomography, frequency comb generators,time-resolved spectroscopy, and dense wavelengthdivision multiplexing.

In this Letter we report what is to our knowl-edge the first KLM directly diode-pumped chirped-mirror-controlled Cr41:YAG laser, generating highlystable near-transform-limited pulses at pulse dura-tions as short as 65 fs at 10–30 mW of output power.

The laser setup follows the design of an astigmati-cally compensated X-fold cavity as shown in Fig. 1. A20-mm-long, 5-mm-diameter, Brewster-cut Cr41:YAGcrystal was mounted on a water-cooled brass block andkept at �16 ±C. The crystal had a peak small sig-nal absorption of kabs � 1.1 cm21 and passive lossesof �1.7%�pass. The cavity consisted of two dichroicconcave mirrors, M1 and M2, with a 100-mm radiusof curvature; a highly ref lective end mirror (HR); anoutput coupler (OC); and five chirped mirrors. TheHR and chirped mirrors were centered at 1500 nm andwere �350 nm broad. Each concave mirror transmit-ted approximately 95% at the pump wavelength. TheOC transmission was 0.2% and is plotted in Fig. 3 be-low. Both resonator arms were equal to 70 cm with1-cm accuracy.

In the present experiment one 4-W and two 3-Wlaser diodes operating at 970 and 1060 nm wereavailable for pumping the Cr41:YAG crystal from bothsides, all having the 100-mm broad emitting stripe butwith different divergence. The pump radiation wascollimated with a high-N.A. 4.5-mm aspheric collima-tor and a 1:20 cylindrical telescope and focused intothe active crystal by a 100-mm achromatic lens. Thecollimated beams of the two 3-W diodes were

Fig. 1. Experimental setup of the KLM diode-pumpedCr41:YAG laser design. Mirrors M1 and M2 are 100-mmradius-of-curvature HR mirrors, CM1 and CM2 are chirpedmirrors, and OC is the output coupler.

© 2004 Optical Society of America

June 1, 2004 / Vol. 29, No. 11 / OPTICS LETTERS 1277

coupled with the wavelength coupler, which transmit-ted 1060-nm radiation and ref lected 970-nm radiation.It is important to note that precise matching of thesetwo beams is required for good overlapping with thecavity mode in the active crystal. Another problemmay arise from accidental imaging of one pump diodeonto the opposite one. When the diodes are alignedin this way, it causes power instabilities or evenleads to destruction of one of the diodes. To avoidthis, one could use diodes with different wavelengthsor crystals with higher absorption. Instead of the1-cm-long crystal with kabs � 1.5 cm21 used in ourearlier works,12,14 here we used a 2-cm-long crystalthat absorbed �70% of the pump radiation at thresh-old as compared with only �55% in the case of theshorter crystal. In addition, the optimal pump spotposition of each diode is not in the center of the longercrystal but is shifted toward the corresponding en-trance face along the longitudinal axis. This preventsdirect imaging of one diode onto the opposite one.With the 2-cm-long crystal the output power was15–20% less than with the 1-cm crystal in a similarconfiguration because of the lower absorption coeff i-cient. Because of the poor beam quality of the laserdiodes, it is generally favorable to use crystals withhigher absorption. However, the highly concentratedCr:YAG samples (kabs . 2 cm21) are characterized bya slow bleaching effect that adversely affects the laserperformance.15

For dispersion compensation we used two differentsets of chirped mirrors (CM1 and CM2, Fig. 1). Wedesigned them to have second-order dispersion ofapproximately 280 fs2 and third-order dispersion of2600 fs3 per bounce at 1500 nm to compensate for40 mm of YAG crystal on a round trip (Fig. 2). Thedispersion oscillations were set to partially compen-sate each other. Chirped mirrors do not allow forfine dispersion tuning like prisms, but they havetwo major advantages: compensation of the third-order dispersion and very low insertion losses (Fig. 2).The latter is especially important for the low-gaindiode-pumped lasers because of the limited power andbeam quality of the diodes.

The laser was first optimized in cw operation,where it demonstrated output power of 110 mW at8 W of power incident on the crystal, with an OCtransmission of 0.2% and a threshold pump powerof 1 W. With the birefringent quartz plate the cwlaser wavelength was tuned from 1412 to 1569 nm[Fig. 3(a)]. Then the longitudinal position of thecrystal and the position of one of the concave mirrorswere adjusted to obtain strong modulation in thelaser output. When adjusted for KLM operation, thelaser yielded 60 mW in the cw regime (multimode)and 30 mW in the mode-locked regime. After thepulsed operation was started by moving one of the endmirrors, the laser remained in the mode-locked regimefor at least 10 min without any purging or dust protec-tion. The pulse duration was 65 fs, and the spectralwidth was 39 nm (Fig. 3). In Fig. 3(a) one can seethat the pulse spectrum maximum is shifted by 70 nmto the red relative to the cw tuning curve maximum.Similar to the case of a Yb-fiber-pumped Cr:YAG laser,

this shift can be attributed to the joint action of stimu-lated Raman scattering, third-order dispersion, andwavelength-dependent losses in the resonator.16

To analyze the soft-aperture17 KLM mechanism, wemeasured the pump beam waists with the knife-edgetechnique. The waist diameters (FWHM) of differentdiodes inside the crystal ranged from 24 mm 3 185 mm(saggital y 3 tangential x) to 49 mm 3 215 mm,whereas the computed waist diameter (FWHM) of thecavity mode was 22 mm 3 40 mm. Stronger focusingin the x (Brewster) plane was useless, because the fo-cal depth becomes shorter than the absorption length.Comparing the pump and resonator beam sizes in bothplanes, we conclude that soft aperturing takes place inthe y (vertical) plane, where the beam sizes are compa-rable. Indirect support of this hypothesis is that theoptimum position of the curved mirror was found to be�0.5 mm inside the second geometric stability region,unlike the results of Ref. 9, where KLM took placeat the edge of the stability region. The calculationof the astigmatically compensated nearly symmetricresonator predicts that the stability gap in the yplane is indeed shifted back to 0.5 mm with respect tothat in the x plane. It is known, however, that theKLM mechanism is stronger in the tangential plane,18

and therefore using the pump diodes with a smalleremitting aperture should be advantageous, because itwould provide better overlap in the tangential planeand thus the possibility of using stronger Kerr-lensmode locking, as well as improved threshold andefficiency.

Fig. 2. Dispersion compensation by chirped mirrors, in-cluding the third-order dispersion. Dashed line, disper-sion of 40 mm of YAG crystal; thick solid curve, round-tripoverall dispersion; thin solid curve, measured ref lection ofthe CMs (ten bounces, corresponds to a full round trip).

Fig. 3. (a) Spectrum and (b) autocorrelation of 65-fspulses from a diode-pumped KLM laser. The tuningcurve of the cw laser is presented by the dotted curve.Output coupler transmission is shown in gray (0.2% at1450 nm and 0.5% at 1650 nm).

1278 OPTICS LETTERS / Vol. 29, No. 11 / June 1, 2004

The mode-locking threshold was 6.5 W of the inci-dent pump power. The laser remained in the single-pulse regime up to 8.3 W of pump power, and theaverage output power was 10 mW. Above this valuethe pulse splits into several pulses with equal ampli-tude arbitrarily spread over a time span of �10 ps.The average output power in this regime was 30 mWwith a pulse duration of 68 fs. According to ourtheoretical analysis, it is necessary to increase themodulation depth and output coupling to widen thesingle-pulse operation region.19

We compared the mode-locking performance of thediode-pumped KLM laser with the same cavity pumpedby an Yb-fiber laser. At a pump power of 8 W we couldobtain 58-fs pulses with a spectral width of 43 nm and120 mW of average output power. The reason for thelower efficiency of the diode-pumped setup is the pooroverlap of the cavity mode with the pump beam, es-pecially in the tangential plane. Shortening the reso-nator arms to 50 cm allowed us to increase the outputpower (135 mW in cw), but it was diff icult to start themode locking because the mode discrimination in thesaggital plane got worse.

The Yb-fiber-pumped system allowed the follow-ing improvements: (i) The output power could beincreased with a higher OC, and (ii) the pulse couldbe shortened by adjusting (reducing) the dispersion.With optimum parameters, sub-30-fs operation canbe realized.6,7 In the diode-pumped system, usingthe output coupler with 0.5% transmission did notprovide significant improvement of the output powerbecause of the high threshold, but resulted in lowerintracavity power, making it impossible to startKerr-lens mode locking. To achieve shorter pulseduration the intracavity dispersion was optimized witha different number of ref lections from the chirpedmirrors. Under Yb-fiber laser pumping, the optimumamount of dispersion was found to be eight ref lectionsfrom chirped mirrors (2250 fs2 at 1530 nm). In thisregime the laser is close to the mode-locking stabilityregion,9 and it requires relatively high pump ratesfor stable KLM pulses. Under diode pumping, themode locking with this dispersion could be initiated,but it was stable for only a few seconds. A higheramount of second-order dispersion increases themode-locking stability at the cost of a longer pulseduration. We have found experimentally that ten re-f lections from chirped mirrors (2400 fs2 at 1530 nm)provide the best stability in the diode-pumpedregime.

It is interesting to note that the results obtained forthe diode-pumped system are comparable with thosedemonstrated in the SESAM mode-locked system[62 fs, 15 mW (Ref. 14)]. However, the relatively lowoutput power was caused by the losses in the SESAM,rather than by the pump overlap problems, and theYb-fiber pumping in the prismless configuration couldnot signif icantly improve the output power.14 Morecomparison details will be provided in a separatefull-length paper. If low-loss broadband SESAMs areavailable, the corresponding diode-pumped systemsmay seriously compete with the KLM systems also atsub-50-fs pulse durations.

In summary, we have demonstrated the f irst, to ourknowledge, directly diode-pumped KLM Cr41:YAGlaser generating stable 65-fs pulses at up to 30 mW ofoutput power in a prismless resonator setup. Theseresults are comparable with those obtained in adiode-pumped mode-locked laser by a SESAM.14 TheKLM technique has good potential for achievingeven-shorter pulses and higher average powerswhen pump diodes around 970 nm with higher facetbrightness become widely available, as indicated bycomparison with the Yb-fiber-laser-pumped setupwith optimized dispersion. The femtosecond pulsesfrom the compact directly diode-pumped Cr41:YAGlaser can be used for spectral broadening in nonlinearfibers, thus making this laser source attractive forsuch applications as optical coherence tomography,spectroscopy, and frequency standards.

It is our pleasure to thank A. V. Shestakov andE. L. S. Polyus for supplying the Cr:YAG crystals.This work was performed under Austrian Fonds zurFörderung der Wissenschaftlichen Forschung grantsT64, F-016, P14704-TPH, and BMBWK B631. E.Sorokin’s e-mail address is e.sorokin@tuwien.ac.at.

References

1. D. E. Spence, P. N. Kean, and W. Sibbett, Opt. Lett.16, 42 (1991).

2. G. Steinmeyer, D. H. Sutter, L. Gallmann, N.Matuschek, and U. Keller, Science 286, 1507 (1999).

3. V. P. Yanovsky, F. W. Wise, A. Cassanho, and H. P.Jenssen, Opt. Lett. 20, 1304 (1995).

4. H. Liu, J. Nees, and G. Mourou, Opt. Lett. 26, 1723(2001).

5. See, e.g., the data sheet on the Femtolite B-60 laser byIMRA, Inc., available at http://www.imra.com.

6. D. J. Ripin, C. Chudoba, J. T. Gopinath, J. G. Fujimoto,E. P. Ippen, U. Morgner, F. X. Kärtner, V. Scheuer, G.Angelow, and T. Tschudi, Opt. Lett. 27, 61 (2002).

7. S. Naumov, E. Sorokin, V. L. Kalashnikov, G. Tempea,and I. T. Sorokina, Appl. Phys. B 76, 1 (2003).

8. C. G. Leburn, A. A. Lagatsky, C. T. A. Brown, and W.Sibbett, in Advanced Solid-State Photonics (Optical So-ciety of America, Washington, D.C., 2004), paper WE4.

9. P. Wagenblast, R. Ell, U. Morgner, F. Grawert, and F.Kärtner, Opt. Lett. 28, 1713 (2003).

10. S. Uemura and K. Torizuka, Opt. Lett. 24, 780 (1999).11. J. M. Evans, V. Petricevic, A. B. Bykov, A. Delgano,

and R. R. Alfano, Opt. Lett. 27, 1171 (1997).12. I. T. Sorokina, S. Naumov, E. Sorokin, and E. Wintner,

Opt. Lett. 24, 1578 (1999).13. A. J. Alcock, P. Scorah, and K. Hnatovsky, Opt. Com-

mun. 215, 153 (2003).14. S. Naumov, E. Sorokin, and I. T. Sorokina, in Advanced

Solid-State Photonics, Vol. 83 of OSA Trends in Op-tics and Photonics Series (Optical Society of America,Washington, D.C., 2003), pp. 163–166.

15. I. T. Sorokina, S. Naumov, E. Sorokin, and A. G.Okhrimchuk, Proc. SPIE 4350, 99–105 (2000).

16. V. L. Kalashnikov, E. Sorokin, S. Naumov, and I. T.Sorokina, J. Opt. Soc. Am. B 20, 2084 (2003).

17. N. Piché and F. Salin, Opt. Lett. 18, 1041 (1993).18. V. Magni, G. Cerullo, S. De Silvestri, and A. Monguzzi,

J. Opt. Soc. Am. B 12, 476 (1995).19. V. L. Kalashnikov, E. Sorokin, and I. T. Sorokina, IEEE

J. Quantum Electron. 39, 323 (2003).