November 15, 2004 / Vol. 29, No. 22 / OPTICS LETTERS 2665

11-W average power Ti:sapphire amplifier system usingdownchirped pulse amplification

David M. Gaudiosi, Amy L. Lytle, Pat Kohl, Margaret M. Murnane, Henry C. Kapteyn, and Sterling Backus

JILA, University of Colorado and National Institute of Standards and Technology, Boulder, Colorado 80309-0440

Received May 3, 2004

We demonstrate a high-power laser system that employs a new scheme in which pulses with negative chirp areamplified and then recompressed by dispersion in a block of transparent material. This scheme has signifi-cant advantages for amplification of intermediate energy pulses at high average power, including insensitivityto small misalignments of the pulse compressor, elimination of compressor gratings and their thermal load-ing issues, low compressor energy and bandwidth throughput losses, and a simplified optical design. Usingthis scheme, we demonstrate what we believe is the highest-average-power single-stage Ti:sapphire amplifiersystem with 11-W compressed output. © 2004 Optical Society of America

OCIS codes: 140.3280, 320.7090.

The technology of ultrafast lasers has progressedrapidly in the past 20 years, facilitating the investi-gation of previously inaccessible phenomena. Manyapplications of ultrafast lasers such as the study ofchemical dynamics, the generation of high-orderharmonics, and ultrafast laser machining requirefemtosecond pulses with microjoule-to-millijoule-levelenergy. The technique of chirped-pulse amplif ication(CPA) is well established as a method for generatinghigh-peak-power pulses with durations from 10 fs to10 ps. In CPA, ultrashort light pulses originatingfrom a mode-locked laser are stretched in time bypassing them through a grating pulse stretcher,1 bypropagating the light through optical fiber,2 or throughother means of introducing positive (i.e., normallydispersive) chirp on the pulse. The applied chirpincreases the duration of the light pulse by factors of103 104. The lower peak power makes it possible toamplify the pulse to high energy without distortingthe pulse due to nonlinear effects or damaging thelaser amplifiers. Subsequently, a pulse compressor isused to provide a wavelength-dependent optical pathlength with negative chirp—opposite to that of thestretcher and the amplif ier medium. This negativechirp recompresses the pulse in time, resulting in anultrashort pulse with dramatically increased peakpower. Progress in CPA lasers in recent years hasbeen enabled in large part by the development of verylarge-area gratings, making it possible to generatepeak powers of up to a petawatt.3

However, CPA has significant limitations, primar-ily associated with the pulse compression process.These include a high sensitivity to pulse compressormisalignment, which can introduce high-order disper-sion and spatial chirp,4,5 as well as thermal loadingand high-energy and bandwidth loss in the gratingcompressor. In this Letter, we demonstrate our useof a novel approach to obtain what is to our knowledgethe highest-average-power single-stage Ti:sapphireamplifier system to date with 11-W compressed out-put. In this scheme, referred to as downchirped pulseamplification (DPA), we avoid many of the problemsassociated with large-area grating pulse compres-

0146-9592/04/222665-03$15.00/0 ©

sors. In DPA, pulses from a mode-locked oscillatorare passed through an optical system that applies anegative chirp to stretch the pulse. The downchirpedpulses are then amplif ied and are recompressedsimply by being passed through a block of transparent(dispersive) material.

Pulse recompression based on material dispersionhas several signif icant advantages compared withstandard grating-based pulse compressors. First, inthe DPA scheme the compressor is insensitive to smallmisalignments that can introduce high-order disper-sion and spatial chirp. The small beam size, low pulseenergy, and absence of focusing optics in the stretchermake it easier to align the stretcher accurately andto avoid spatial chirp. Second, the compressor haslow energy and spectral bandwidth throughput losses,increasing the overall system efficiency. Third, theDPA scheme avoids beam distortion caused by the ther-mal loading of the compressor gratings, which is oftenpresent in high-average-power systems. Finally,since DPA prevents any spatial–spectral dispersionof the beam in the high-power part of the system, itsuse may be preferable for laser amplif ier systems thatgenerate carrier-envelope stabilized pulses. It hasbeen discussed that grating- or prism-based dispersioncan result in coupling of a beam pointing to the carrier-envelope offset, resulting in added noise.6 Althoughrecent measurements indicate this is not a majorissue,7 because spatial–spectral dispersion is usedonly in the low-power part of the DPA system, theseoptics can be included in-loop at high repetition ratesin any carrier-envelope offset feedback stabilization.The DPA scheme also has advantages over ampli-fier designs that use prism compressors to preventgrating throughput losses.8 Prism compressors arerelatively bulky and have limited spectral apertures.Increasing the beam aperture in a large-area prismcompressor also increases the required material path,which must be counteracted by further prism sepa-ration. Also, nonlinear distortion in a prism-basedcompressor is accumulated during a time when thebeam has angular and spatial–spectral chirp, possiblyresulting in complex beam distortions. In contrast,

2004 Optical Society of America

2666 OPTICS LETTERS / Vol. 29, No. 22 / November 15, 2004

in DPA high-intensity propagation of the pulse occursexclusively during collimated propagation, resultingin relatively little distortion.

At f irst glance, pulse compression by use of ma-terial dispersion seems counterintuitive because theCPA scheme was designed to prevent the nonlinearpulse distortion that results from propagation of ahigh-power beam in the amplifier system material.However, one can prevent this distortion by ensuringthat the beam’s cross section is large enough incompression to keep the peak intensity low. Forhigh pulse-repetition-rate systems, the beam sizerequired for this is typically smaller than would berequired to prevent thermal distortions in a con-ventional grating compressor. The material pathlength in the amplif ier system itself is also moderatecompared with the f inal compression material’s pathlength to ensure that the pulse does not compresssignificantly during amplif ication. Given theseconsiderations, the DPA approach is particularlysuited to high-repetition-rate, high-average-power,10–100-kHz broadband Ti:sapphire laser amplif iersystems. Because ultrafast laser systems have onlyrecently achieved both high average power and veryshort, �20-fs, pulse duration,9 we speculate that thisis why a DPA scheme has not, to our knowledge, beenconsidered in the past. DPA would also be appropri-ate for optical parametric CPA schemes.10 It has beenrecognized that, in parametric amplification of signalpulses with positive chirp, the idler is emitted withnegative chirp and can be compressed by materialdispersion.11 However, the possibility of imposing anegative chirp on the signal pulse before amplif icationwas not to our knowledge previously considered.

Our first experimental implementation of DPA isin a single-stage Ti:sapphire amplifier system thatgenerates nearly 11 W of compressed average power.9

Figure 1 shows a schematic of the experimental setup.Pulses with a FWHM bandwidth of 85 nm, at a repeti-tion rate of 88 MHz,12 and with a pulse energy of 6 nJper pass, are stretched to 20 ps through a two-stagepulse stretcher before being injected into the amplifier.The first stage of pulse stretching consists of a pairof 600-groove�mm gratings separated by a distanceof 7.5 cm. The laser beam is incident upon thesegratings at an angle of 70±. The second stage consistsof a pair of SF18 prisms with a tip-to-tip separation of123 cm, through which the beam makes four passes.The prism pair compensates for a third-order disper-sion of 4.81 3 104 fs3 introduced by the material-basedpulse compressor. Before amplification, the laserbeam also passes through a Pockels cell pulse selector.The downchirped pulses are then injected into asingle-stage, 12-pass ring amplif ier that employs acryogenically cooled Ti:sapphire amplifier crystal.9,13

The 6-mm-thick, Brewster angle, Ti:sapphire crystal isenclosed in a vacuum cell and cooled with a closed-cyclehelium refrigerator to a temperature of 73 K. Thisreduces the thermal lens from ,1 cm for an uncooledcrystal to approximately 6 m for the cryogenicallycooled crystal at 100-W pump power. The amplifier ispumped at a repetition rate of 7–10 kHz with a 100-Waverage-power Q-switched 532-nm laser (Quantronix

Eagle). The amplif ier ring uses two 1-m radius-of-curvature mirrors and a high angle-of-incidencef lat mirror. The output pulse energy is 1.1 mJfollowing 12 passes through the Ti:sapphire amplif iercrystal, resulting in an extraction eff iciency of 13%.

Following amplif ication, the 4-mm-diameter laserbeam is then sent into the f irst of two telescopes. Thefirst telescope expands the diameter of the beam to1 cm and serves as a spatial filter. The beam thenmakes three passes through a 10-cm-long block ofSF18 material to reduce the pulse duration to 4.3 ps.The laser beam’s diameter is expanded to 2.5 cm, andpulse compression is completed by passage of the beamthrough another 20 cm of SF18 glass. The result is arecompressed pulse width of 28 fs in a 1.1-mJ energypulse. The pulse duration (Fig. 2) was measuredby second-harmonic generation frequency-resolvedoptical gating. Residual uncompensated higher-orderdispersion is the main factor that prevents full re-compression to the transform-limited pulse durationof 25 fs. Future implementations of the DPA systemcan include a pulse shaper, such as an LCD,14 anacousto-optic modulator,15 or a deformable mirrorsetup,16 that can compensate for high-order dispersion.

Figure 3 shows the average power of the compressedoutput pulses as a function of amplif ier repetition rate.For repetition rates of 7–10 kHz, the pump pulse en-ergy was maintained at 8.85 mJ. The average outputpower increased with increasing repetition rate, and

Fig. 1. Schematic diagram of the DPA scheme. Pulsestretching is achieved with a prism-grating stretcherfollowed by amplification and pulse compression in a pieceof glass.

Fig. 2. Temporal prof ile of 1.1-mJ, 28-fs amplif ied outputpulse measured by frequency-resolved optical gating.

November 15, 2004 / Vol. 29, No. 22 / OPTICS LETTERS 2667

Fig. 3. Average output power of the DPA system as a func-tion of pulse repetition rate.

Fig. 4. Beam profiles of (a) a focused beam waist as aresult of f � 104 focusing, where the beam diameter is164 mm, and (b) a 2.5-cm-diameter unfocused, compressedoutput beam.

the highest average power recorded was 10.7 W at10 kHz, which corresponds to an output pulse energyof 1.1 mJ. The compressed extraction efficiency was11–12% over the entire repetition rate range tested.Most importantly, the throughput of the pulse com-pressor was 90%, including a 5% loss in the spatialfilter and surface losses. Typical throughput ofgrating-based pulse compressors is 60%.

Limiting nonlinear effects in the compressionstage is important in this design. The accumu-lated nonlinear phase is given by the B integralB �

R�2pn2�l�I �z�dz, where I �z� is the intensity of

the pulse as a function of propagation direction, n2 isthe nonlinear index of refraction (5.0 3 10216 cm W21

for SF18), and l is the wavelength. Our calculationsshow that the B integral for the amplif ier system is0.76 rad, with 0.44 rad accumulating in the first stageof compression where the beam is 1 cm in diameter.The B integral accumulated in the amplif ier itselfis only 0.32 rad. The B integral of 0.76 does notadversely inf luence focusing of the beam, as shown bythe focused spot size and beam profile (Fig. 4). Themeasured M2 of the amplified beam was 1.57, and theexpected focused beam diameter of 164 mm was ob-tained with f � 104 focusing optics. Future research

will expand the beam to 2.5 cm before compressionto reduce the compressor’s B integral to 0.26. Thiswill allow us to achieve diffraction-limited focusingof the DPA beam. We note that in the course of thiswork we observed no beam distortions that could beattributed to energy absorption or to thermal effectsin the pulse compressor. Total energy absorption was,, 1 W, distributed over a large volume.

In conclusion, we have demonstrated a novel am-plifier design, based on negatively chirped pulses,that increases compressor simplicity, energy, andbandwidth throughput compared with those of stan-dard ultrafast laser amplif iers. This design resultedin what we believe is the highest-average-powersingle-stage Ti:sapphire amplifier system to datewith 11 W of compressed output power. This DPAapproach is particularly suited to high-repetition-rate,high-average-power, 10–100-kHz laser amplif ier sys-tems in which the beam diameter required for avoidingnonlinear distortion effects is small and where thechirps required are modest.

The authors gratefully acknowledge support for thiswork from the National Science Foundation and theU.S. Department of Energy. D. M. Gaudiosi’s e-mailaddress is gaudiosi@colorado.edu.

References

1. O. E. Martinez, IEEE J. Quantum Electron. QE-23,1385 (1987).

2. D. Strickland and G. Mourou, Opt. Commun. 56, 219(1985).

3. M. D. Perry, D. Pennington, B. C. Stuart, G. Tietbohl,J. A. Britten, C. Brown, S. Herman, B. Golick, M.Kartz, J. Miller, H. T. Powell, M. Vergino, and V.Yanovsky, Opt. Lett. 24, 160 (1999).

4. C. Fiorini, C. Sauteret, C. Rouyer, N. Blanchot,S. Seznec, and A. Migus, IEEE J. Quantum Electron.30, 1662 (1994).

5. G. Pretzler, A. Kasper, and K. J. Witte, Appl. Phys. B70, 1 (2000).

6. F. W. Helbing, G. Steinmayer, J. Stenger, H. R. Telle,and U. Keller, Appl. Phys. B 74, S35 (2002).

7. I. Thomann, E. Gagnon, R. J. Jones, A. S. Sandhu,A. Lytle, R. Anderson, J. Ye, M. Murnane, and H.Kapteyn, Opt. Lett. 12, 3493 (2004).

8. C. Spielmann, M. Lenzner, F. Krausz, and R. Szipocs,Opt. Commun. 120, 321 (1995).

9. S. Backus, R. Bartels, S. Thompson, R. Dollinger,H. C. Kapteyn, and M. M. Murnane, Opt. Lett. 26,465 (2001).

10. I. N. Ross, P. Matousek, G. H. C. New, and K. Osvay,J. Opt. Soc. Am. B 19, 2945 (2002).

11. A. Dubietis, G. Jonusauskas, and A. Piskarkas, Opt.Commun. 88, 437 (1992).

12. M. T. Asaki, C. P. Huang, D. Garvey, J. Zhou,H. C. Kapteyn, and M. M. Murnane, Opt. Lett. 18,977 (1993).

13. S. Backus, C. Durfee, M. M. Murnane, and H. C.Kapteyn, Rev. Sci. Instrum. 69, 1207 (1998).

14. A. Efimov, M. Moores, N. Beach, J. Krause, and D.Reitze, Opt. Lett. 23, 1915 (1998).

15. M. A. Dugan, J. X. Tull, and W. S. Warren, J. Opt. Soc.Am. B 14, 2348 (1997).

16. E. Zeek, R. Bartels, M. M. Murnane, H. C. Kapteyn,S. Backus, and G. Vdovin, Opt. Lett. 25, 587 (2000).