November 1, 2005 / Vol. 30, No. 21 / OPTICS LETTERS
2921
Pump-noise transfer in optical parametric chirped-pulse amplification Nicolas Forget, Arnaud Cotel, Erik Brambrink, Patrick Audebert, and Catherine Le Blanc Laboratoire pour l’Utilisation des Lasers Intenses, Unité Mixte 7605, Centre National de la Recherche Scientifique, Commissariat a l’Energie Atomique, Ecole Polytechnique, Université Pierre et Marie Curie, Ecole Polytechnique, 91128 Palaiseau, France
Aurélie Jullien, Olivier Albert, and Gilles Chériaux Laboratoire d’Optique Appliquée, Ecole Nationale Supérieure des Techniques Avancées, Ecole Polytechnique, Chemin de la Hunière, 91761 Palaiseau Cedex, France Received May 31, 2005; revised manuscript received July 5, 2005; accepted July 5, 2005 We report on direct observation of temporal contrast degradation of short pulses amplified by optical parametric chirped-pulse amplification. We show that, despite injection seeding, quantum-noise-induced fast modulations 共⬍50 ps兲 of the temporal profile of the pump pulse are imprinted on the spectrum of the amplified chirped pulse and give rise to a large picosecond pedestal in the time domain. © 2005 Optical Society of America OCIS codes: 190.4970, 320.7090.
Optical parametric chirped-pulse amplification (OPCPA) has been introduced as an alternative highgain and broadband amplification technology that exhibits favorable properties for amplification of short pulses.1 Since its first demonstration in 1992,2 several groups have proposed3 and demonstrated broadband and efficient OPCPA preamplifiers at the millijoule level4,5 and power amplifiers at the joule level6,7 in both nanosecond and picosecond8 regimes, with bulk crystals [-BaB2O4 (Refs. 4 and 5), LiB3O5 (Refs. 6 and 7), KH2PO4 (Refs. 4 and 5)] and with quasi-phase-matched crystals.9 Of the many properties desired for a short-pulse preamplifier, temporal pulse contrast is of greatest importance because many experimental devices require clean pulse fronts. Because gain is directional and quasi-instantaneous in parametric amplifiers, OPCPA is expected to lead to a significant improvement of temporal pulse contrast and, in particular, to be almost free of prepulses. In optical parametric amplifiers (OPAs), the amplification noise originates from a quantum effect known as parametric fluorescence (PF), i.e., spontaneous decay of pump photons into photon pairs of lower energies. As parametric amplification and fluorescence rely on an instantaneous and directional interaction, PF occurs within the time window defined by the pump pulse. Like amplified spontaneous emission (ASE), PF is added to the chirped seed and gives rise, after compression, to a flat pedestal surrounding the amplified short pulse. Because the gain time window is usually shorter than the cavity dumping time of regenerative cavities, PF may consequently be lower than spontaneous and emission in regenerative cavities,10,11 nanosecond-range pulse contrasts greater than 108 were recently reported.12 A major consequence of time-dependent gain is that intensity fluctuations of the temporal profile of the pump pulse are transferred to the temporal profile of the chirped pulse and therefore to the spectrum 0146-9592/05/212921-3/$15.00
of the signal pulse, causing a degraded amplified pulse. To avoid fluctuations in the temporal profile of the pump pulses, mode-locked or injection-seeded Q-switched lasers are required for OPCPA. Nevertheless, residual noise in the temporal profile of the pump pulses originating from insufficient injection seeding (or from ASE) may still lead to minor spectral modulations, resulting in degradation of the recompressed pulse in a high dynamic range. In this Letter we report direct experimental observation of a pedestal of tens of picoseconds induced by residual temporal beating of a doubled, Q-switched, and injectionseeded Nd:YAG laser. Additionally, we show that pump-induced noise may be the predominant limit on high-contrast OPCPA. Our experimental setup is as follows: Seed pulses of ⬃150 fs centered at 1057 nm and with 10 nm FWHM bandwidth are generated by a mode-locked Ti:sapphire oscillator operating at 80 MHz. The oscillator produces 1 nJ pulses, which are then sent into a single-pass Öffner pulse stretcher. Stretching results in 600 pJ pulses of 1.3 ns FWHM duration, which are subsequently introduced into a 10 Hz Pockels cell pulse selector and directed into a Ti:sapphire-based regenerative amplifier. Pulses are amplified to 2 mJ, and their duration is reduced to 550 ps 共4.5 nm兲 as a consequence of spectral gain narrowing. A fraction of the amplified pulses 共200 J兲 is sent into a doublestage parametric amplifier and then to the compressor. Pulse contrast was measured both by a singleshot second-order autocorrelator and by a highdynamic-range cross correlator (Sequoïa, Amplitude Technologies). With 75 J per pulse, the highdynamic-range cross correlator was able to measure the pulse contrast over 8 orders of magnitude, which was sufficient for monitoring the effects of parametric amplification. Because the regenerative amplifier induced strong spectral gain narrowing and subsequent chirped© 2005 Optical Society of America
2922
OPTICS LETTERS / Vol. 30, No. 21 / November 1, 2005
pulse shortening, a double-pass OPCPA was designed to compensate for the poor temporal overlap between the pump 共7 ns兲 and signal 共550 ps兲 pulses. Pump and signal pulses were combined in a dichroic mirror, imaged on the first OPA, and then separated by another dichroic mirror. The 4 mm top-hat pump beam was relay imaged and then recombined with the signal beam on the second OPA stage. A time delay of 750 ps was inserted between the pump and signal pulses between OPA I and OPA II to improve both temporal overlap and stability. Small external noncollinear angles of 2.3° and 3° allowed the signal and idler beams to be separated after each OPA. OPA I was made from two 3 mm -BaB2O4 crystals cut at 22.8°, and OPA II was made from one 10 mm and one 15 mm LiB3O5 crystal cut at = 0° and = 11.6°, respectively. All crystals were cut for type I phase matching, 1057 nm+ 1071 nm= 532 nm, and individually phase matched. Pump pulses were generated by a commercial injection-seeded, frequencydoubled Nd:YAG laser (Quantaray, Spectra-Physics) that produces 532 nm pulses with 8.2 ns FWHM pulse duration in the pulse center with the spatiotemporal shape that is typical of unstable resonators. Pump pulses were synchronized with the selected stretched pulses with a characteristic pulse-topulse jitter of 1.5 ns rms. When the pump energy was varied from 0 to 350 mJ, the energy of amplified pulses at the exit of the OPCPA stage ranged from 150 J to 5 mJ, yielding compressed pulses of 75 J to 2.5 mJ. Spectra with and without parametric amplification are shown in Fig. 1. Even for a small gain 共⬃3.3兲, some short 共⬍0.5 nm兲 modulations appear on the amplified spectrum. These modulations vary from shot to shot in amplitude, frequency, and phase and tend to grow in amplitude with gain. High-dynamic-range cross correlations of recompressed pulses are shown in Fig. 2, with and without parametric amplification. When the pump beam is blocked, the cross correlation shows an ASE background at 10−7, with a clean pulse. For a parametric gain of ⬃30, an ⬃20 ps pedestal with a relative intensity of ⬃10−3 appears under the main 450 fs pulse. The ASE level remains, how-
Fig. 1. Dashed curve, spectrum of the chirped pulses seeded into the OPCPA 共200 J兲. Solid curve, spectrum of an amplified pulse 共5 mJ兲.
Fig. 2. (a) Cross-correlation signal of nonamplified pulses (lighter curve) and amplified pulses (darker curve). (b) Cross-correlation signal of amplified pulses (lighter curve) and cross correlation computed from experimental spectra (darker curve).
ever, at 10−7 when parametric amplification is turned on. To correlate the observed spectral modulations and the 15 ps pedestal, a theoretical cross-correlation signal was computed from the experimental spectra, assuming only a residual chirp. The average theoretical cross-correlation signal of ten successive pulses is plotted in Fig. 2 and shows good agreement with the measured cross-correlation signal. Both curves exhibit a large picosecond pedestal at 10−3, showing that the measured spectral modulations are fast enough to generate a pedestal larger than ⬃10 ps. Because of the finite resolution of the spectrometer 共0.4 nm兲, no information could be extrapolated beyond time delays of 10 ps. As can be noted, the spectral modulations seen in the seed pulses in Fig. 1 do not produce temporal replicas of the main pulse at ±7 ps at 6 ⫻ 10−5 on the cross-correlation signal, as could have been expected. These spectral modulations were identified as artifacts from the camera used in the spectrometer. An ultrafast streak camera (photocathode S1, streak camera C7700-11, 12-bit CCD, Hamamatsu) with a time resolution of ⬃10 ps was then used to monitor the temporal profile of the pump pulse (Fig. 3) on a subnanosecond scale. Despite injection seeding, temporal modulations as short as 10 ps were measured on the pump pulse, together with specklelike spatiotemporal structures, indicating that the pi-
November 1, 2005 / Vol. 30, No. 21 / OPTICS LETTERS
2923
ficiency point. In the third-order cross-correlation signal the relative intensity of the replica is given by the square of the temporal amplitude, ⑀2 / 16. As Nd:YAG has a linewidth of ⬃0.45 nm, the width of the pedestal at ⑀2 / 16 is expected to be less than 35 ps. We measured temporal modulation in the pump wave with ⑀ ⬃ 2.5⫻ 10−2, so the relative intensity of the pedestal in the cross correlation at ±17.5 ps would be, according to this model, ⬃4 ⫻ 10−5, which is of the order of the experimental result, given the uncertainty in the linewidth of the pump laser. In conclusion, we have shown that residual temporal beating in the temporal profile of the pump pulse leads to major degradation of the temporal contrast of recompressed pulses in the form of a temporal pedestal of a few tens of picoseconds. The width of this pedestal is proportional to the spectral linewidth of the pump laser, dilated by the chirp ratio of the signal pulses. Given the ultimate noise-to-signal ratios achievable in high-energy nanosecond pulsed lasers 共⬃10−4兲, the degradation in pulse contrast induced by residual noise may be the preponderant source of degradation in pulse contrast in a high-dynamic range on a scale of tens of picoseconds. Fig. 3. Pump pulse temporal profile on (a) a 20 ns scale and (b) a 1 ns scale. Dashed curve, 8.2 ns FWHM Gaussian fit.
cosecond pedestal originates from residual temporal beating on the pump pulse. The dynamic range of the measurement was 150, and the noise figure (reduced standard deviation) was 0.4%, sufficient for noise to be distinguished from fast modulation in Fig. 3 and to yield a reasonable estimation of the amplitude of fluctuations. If we ignore any mode structure, the spectrum of pulses emitted by an injection-seeded pump laser may be described as the sum of the injected frequency and a residual noise distribution that corresponds to the gain bandwidth of the laser medium and to the optical properties of the intracavity elements. If this remaining noise distribution is Gaussian, with relative intensity ⑀ Ⰶ 1 and spectral width ⌬, then, if the second-harmonic stage is taken into account, the order of magnitude of the fastest temporal mode beatings in the pump wave is given by the beatings between injected frequency 0 and frequency 0 + ⌬ / 共2冑2兲 of amplitude ⑀ / 2. If is the chirp ratio of the signal pulse, the corresponding spectral modulation imprinted on the amplified signal is of spectral period 2冑2 / ⌬. Such modulation gives, in the temporal domain, two replicas of the main pulse with time delay ±⌬ / 共2冑2兲 and relative amplitude ⑀ / 4 for operation close to the maximum ef-
N. Forget’s e-mail
[email protected]. References
address
is
1. R. Butkus, R. Danielius, A. Dubietis, A. Piskarskas, and A. Stabinis, Appl. Phys. B 79, 693 (2004). 2. A. Dubietis, G. Jonusauskas, and A. Piskarskas, Opt. Commun. 88, 437 (1992). 3. I. N. Ross, P. Matousek, M. Towrie, A. J. Langley, and J. L. Collier, Opt. Commun. 144, 125 (1997). 4. I. N. Ross, J. L. Collier, P. Matousek, C. N. Danson, D. Neely, R. M. Allott, D. A. Pepler, C. Hernandez-Gomez, and K. Osvay, Appl. Opt. 39, 2422 (2000). 5. I. Jovanovic, B. J. Comaskey, C. A. Ebbers, R. A. Bonner, D. M. Pennington, and E. C. Morse, Appl. Opt. 41, 2923 (2002). 6. X. Yang, Z. Xu, Y. Leng, H. Lu, L. Lin, Z. Zhang, R. Li, W. Zhang, D. Yin, and B. Tang, Opt. Lett. 27, 1135 (2002). 7. V. Bagnoud, I. Begishev, M. Guardalben, J. Puth, and J. Zuegel, Opt. Lett. 30, 1843 (2005). 8. N. Ishii, L. Turi, V. S. Yakovlev, T. Fuji, F. Krausz, A. Baltuska, R. Butkus, G. Veitas, V. Smilgevicius, R. Danielius, A. Piskarskas, Opt. Lett. 30, 567 (2005). 9. I. Jovanovic, J. R. Schmidt, and C. A. Ebbers, Appl. Phys. Lett. 83, 4125 (2003). 10. H. Yoshida, E. Ishii, R. Kodama, H. Fujita, Y. Kitagawa, Y. Izawa, and T. Yamanaka, Opt. Lett. 28, 1245 (2003). 11. I. Jovanovic, C. G. Brown, C. A. Ebbers, C. P. J. Barty, N. Forget, and C. Le Blanc, Opt. Lett. 30, 1036 (2005). 12. V. Bagnoud, J. Puth, I. Begishev, M. Guardalben, J. D. Zuegel, N. Forget, C. Le Blanc, and J. Bromage, in Conference on Lasers and Electro-Optics (CLEO) (Optical Society of America, 2005).
