Efficient high-energy pulse-train generation using a 2 n ...

1 Efficient high - energy pulse -traingeneration using a 2n- pulse Michelson interferometerCraig W. Siders, Jennifer L. W. Siders, Antoinette J. Taylor, Sang-Gyu Park, andAndrew M. WeinerWe demonstrate a novel, Michelson-based, ultrafast multiplexer with a throughput approaching 100% fora polarization-multiplexed train and 50% for a linearly polarized train, which is compatible with ahigh- energy pulse train and shaped- pulse generation . The interpulse spacings in the resultant 2n-pulsetrain can be adjusted continuously from multinanoseconds through zero.

2 using this interferometer, wealso demonstrate generation of a 16- pulse train of terahertz pulses. 1998 Optical Society of AmericaOCIS , , , , , IntroductionNumerous applications currently exist that require atrain of high - energy ultrafast pulses. Such applica-tions include multiple-laser- pulse excitation schemesfor high -gradient plasma accelerators,1photocathodeinjectors for conventional accelerators, multiple- pulse excitation of atoms, molecules, and solids,2andhigh-fluence terahertz wave-train generation for ra-dar and microwave op-timization of the efficiency of high -harmonicgeneration when short pulses are used may requirethe use of fast-rise-time flat-top pulses.

3 Such super-Gaussian pulses can be generated when several shortpulses are stacked together with an interpulse spac-ing of the order of the initial short- pulse techniques in which phase and am-plitude masks are used in the focal plane of a zero-dispersion pulse stretcher are typically used forproducing low- energy ~ , oscillator-level!ultrafastshaped pulses and pulse such tech-niques to produce trains of high - energy ~ , ampli-fied!pulses is limited because of damage to the maskresiding in the focal plane of the stretcher.

4 Pulseshaping of 200-mJ pulses by using an acousto-opticmodulator has recently been demonstrated;6how-ever, the setup exhibits 2 orders of magnitude loss,resulting in only a 2-mJ output. A further increasein the diffraction efficiency causes significant acousticnonlinearity in the acousto-optic modulator. Pulseshaping before amplification has been demonstrat-ed,7but care must be taken to avoid nonlinear effects,such as nonlinear temporal diffraction,8that occurwith amplitude masking. Even in the case of phase-only masking, amplifier output fluences are typicallyreduced compared with that from an unshaped.

5 The experimental necessity of a single shortpulse synchronized to the pulse train for cross-correlation measurements ultimately dictates thatpulse shaping occur after schemes have appeared in the literaturethat are variations on a Michelson interferometerusing multiple beam splitters and optical delay linesto produce pulse trains of 2npulses wherenis apositive ,10 These schemes do not propa-gate the beam through a focus and therefore are com-patible with pulse shaping high - energy pulses.

6 Then-fold application of a single Michelson interferome-ter would appear at first glance to reduce the totalfluence of the generated train to 22nbecause eachunit has a single beam on the input and two beams onthe output, one of which is not used. Similar con-clusions on generation efficiency are reached forpulse-train generation when a series of beam split-ters is both beams from each beamsplitter were used, the generation of two outputbeams that contain sequences of 2npulses has , interpulse spacings inthis device are limited by optical elements ps,C.

7 W. Siders, J. L. W. Siders, and A. J. Taylor are with theMaterials Science and Technology Division, Los Alamos NationalLaboratory, MS D429, Los Alamos, New Mexico 87545. and A. M. Weiner are with the School of Electrical andComputer Engineering, Purdue University, West Lafayette, Indi-ana 15 December 1997; revised manuscript received 26 March $ 1998 Optical Society of America5302 APPLIED OPTICSyVol. 37, No. 22y1 August 1998and therefore it is unsuitable for Michelson InterferometerWe present here ann-fold application of a Michelsoninterferometer, which, by using both beams fromeach beam splitter and then by polarization multi-plexing the two output beams, results in a singlepulse train of 2npulses with nearly 100% the frequency domain this interferometer repre-sents, for each polarization, a highly oscillatory am-plitude and phase filter.

8 Our device design for thegeneration of a train of 16 pulses is shown in Fig. the delay arms can be easily positioned forequal lengths, interpulse spacings from zero to mul-tinanoseconds are achievable. The first eight pulsesare orthogonally polarized to the final eight; however,one can easily polarize the output along a single axisby using a high -damage-threshold thin-film polar-izer, resulting in a 50% throughput. In some cases,for example, a laser-drivene1ye2collider, the twolinearly polarized output trains can be individuallyused.

9 In general, for a 2npulse train the designshown in Fig. 1. requiresnbeam splitters,~4n12!mirrors, 2nlinear translators~to set and vary theinterpulse spacing!, one half-wave plate, and onethin-film polarizer, leading to a cost that scales asnand an overall area that scales characterize this device, we use the 800-nm,150-fs, 1-mJ output of a chirped- pulse amplified Ti:sapphire laser system operating at 1 kHz. For a16- pulse interferometer 18 mirrors and 4 beam split-ters are needed.

10 For reduced cost, protected gold-coated mirrors and dielectric beam splitters are used,resulting in a measured throughput of 73%, as ex-pected. If we were to use high -damage-thresholddielectric mirrors, a throughput of 98% would be ex-pected. The interferometer is initially aligned sothat the pulses overlap temporally two at a temporal overlap is determined first with spa-tial fringes and is optimized with second-harmonicgeneration in a doubling crystal. The desired pulsesequence is then set.