Sub-25 fs pulses from a multipass amplifier system have been spectrally broadened in a hollow fiber up to 250 nm. Using a combination of a prism compressor and an improved acousto-optic programmable dispersive filter (AOPDF), we were able to compress the pulses close to their transform limit. Under optimized conditions we achieved pulses with a duration of 8 fs and a peak power up to 9 GW.
© 2003 Optical Society of America
Modifying the structure of matter in a controlled way without additional chemical substances is now possible by using laser pulses with a well-controlled spectral amplitude and phase. The selective breaking or formation of molecular bonds with the help of laser pulses is introduced in the literature as coherent control [1, 2]. In the last years, we have also witnessed a tremendous progress in producing soft-x-ray radiation via high harmonic generation of visible laser pulses. It has been demonstrated, theoretically and experimentally, that both the conversion efficiency and the highest obtainable photon energy can be significantly increased by using intense laser pulses as short as possible. Moreover, high harmonic generation opened the way to attosecond resolution time-resolved spectroscopy. However, to generate a single attosecond xuv-pulse a visible few-cycle laser pulse with a well-controlled electric field is indispensable [3–5]. The above mentioned experiments rely on nonlinear transition and therefore call for pulses with a fairly high energy only obtainable from amplifier systems.
The key towards the realization of the described experiments is the precise control of the spectral phase and/or amplitude of ultrashort powerful laser pulses. The phase of laser pulses can be fairly well manipulated with static elements such as pairs of prisms [6–8] gratings  and chirped mirrors [4,9,10]. Pairs of prisms and gratings are widely used for stretching and compressing pulses. The magnitude of the lowest order dispersion contribution, the group-delay dispersion (GDD) scales linearly with the separation of the two elements. The lengthening or shortening of the pulses is roughly proportional to the magnitude of the GDD, thus the pulse duration can be easily controlled. Unfortunately, not only GDD but also socalled high order dispersion contributions are simultaneously introduced. They can not adjusted independently, because they are linked to material or design parameters. Only the combination of gratings, prisms, and adding different materials offers enough flexibility to control the overall high order dispersion. To minimize unwanted high order dispersion contributions in compression schemes it is often necessary to use low dispersive gratings and prisms, resulting in huge dimensions of the whole setup. More degrees of freedom in the design are offered by chirped dielectric mirrors. The dispersion curve including high order contributions can be tailored offering a much more precise compensation of phase errors. Drawback of chirped mirrors are their fixed dispersion characteristic and the rather small magnitude of the maximum negative GDD, which is typically in the order of -50 fs2 . To tune the dispersion, the numbers of reflections upon chirped mirrors can be changed; however, the number of reflections is limited either by the size of the mirrors or by the reflection losses. The latter is especially true for compressing pulses after a substantial stretching.
All the above mentioned methods can only be implemented if the dispersion to be compensated is not varying on a daily time scale. Only in this case, it makes sense to carefully design and setup a compressor consisting of a combination of gratings, prisms and chirped mirrors. With the above-mentioned static devices also phase errors caused by nonlinear processes can only be eliminated in first order, because of the different origin of the phase changes. Further, the generation of arbitrarily shaped pulses necessary for coherent control is out of reach. To overcome these limitations several methods have been successfully introduced. Nearly all of them are based on programmable phase and/or amplitude masks put into a spectrally dispersed beam. In the most common realization, the mask is located in the focal plane of symmetric telescope surrounded by two dispersing elements such as gratings or prisms. The programmable mask can be a liquid crystal display (LCD) [12–16], an acoustooptic modulators (AOM) , an electro-optic modulator  or a deformable mirror [5,18]. The LCD is nowadays the most popular device. By putting more than one LCD into the pulse shaper the phase, the amplitude and the polarization can be shaped . The major drawbacks are the limited number of pixels, the sensitivity to the alignment and the fairly low update rate. According to the Fourier theory, the pixelation causes pre- or post-pulses in the time domain – containing sometime a substantial fraction of the total pulse energy. This problem is not present if a deformable mirror is used. This advantage is outweighed by the possibility of phase only modulation. All these drawbacks can be bypassed by a completely different realization of a pulse shaper. The need for dispersive elements and the telescope is omitted by employing an Acousto Optic Programmable Dispersive Filters (AOPDF) . A transmission diffraction grating is formed by an acoustic wave in a crystal. The transmitted pulse represents a convolution of the phase and amplitude transmission characteristic of the grating and the input pulse. This method features ease of implementation and alignment because light is collinearly propagating with the acoustic wave in the crystal. It allows to program accurate quantitative amounts of dispersion. When programming polynomial phase contributions, there is no interference between the different order terms. Finally, update rates up to tens of kHz’s are possible, only limited by transit time of the sound wave in the crystal. With the first generation AOPDF (DAZZLER Fastlite), pulse compression and multiple pulse formation have been successfully demonstrated .
This paper reports on the characterization of a newly designed broadband AOPDF. We measured the integrated and spectrally resolved amplitude transmission characteristic of the AOPDF to estimate the maximum controllable bandwidth. Based on this knowledge we designed an experiment to shorten amplified pulses below 10 fs. Self-phase modulation in a gas filled hollow fiber broadened the spectrum of the output pulses from a multipass Ti:sapphire based CPA system. The phase distortions have been corrected by a compressor consisting of an AOPDF and a prism pair emerging in 8fs pulses with a peak power of several GW.
2. Experimental setup
The amplifier systems consists of a mirror dispersion controlled femtosecond Ti:sapphire master oscillator (FEMTOLASERS Production GmbH) delivering 12 fs pulses with a single pulse energy of 5 nJ, at a repetition rate of 74 MHz. Long-term stability of the oscillator is ensured by pumping it with a frequency-doubled diode pumped solid state laser (Millenia Vs Spectra-Physics GmbH). The oscillator pulses have been stretched up to 10 ps by the material dispersion of an inserted 15-cm long SF57 glass block (approx. 33600 fs2) and the broadband isolator (Gsänger Optoelektronik GmbH, 6200 fs2). After 72 reflections upon chirped mirrors (introducing mainly third and forth order dispersion - TOD and FOD - with a magnitude 1000 fs3 and 2800 fs4 per reflection, respectively) we fed the pulses into the amplifier. Amplification took place in a 9- pass amplifier (Fig. 1) pumped by an intracavity-frequency-doubled Q-switched Nd:YLF-Laser (Model 621D, Thomson CSF Laser) with an energy of 10 mJ at 1 kHz repetition rate. After the amplification, we compressed the pulses with a LAK16A double prism compressor. The two pairs of prisms were separated by 2.1 m. The excessive negative TOD and FOD introduced by the prism pair has been pre-compensated by the chirped mirrors in front of the amplifier. In this way, we were able to obtain 20 fs pulses with energy of 1 mJ. The pulses after the prism compressor carry high order chirp. Further shortening will be possible by carefully adjusting the number of reflections upon the chirped mirrors in the pre-compensation stage. However, we found that a small, residual, high order chirp has a negligible influence on the broadening. The amplified and partly compressed pulses were launched into a hollow fiber. The inner diameter of the fiber was 200 µm and the broadest spectra have been produced when filled with Argon at a pressure of 1.5 atm. Under the optimized conditions, we observed at the output of the fiber a spectrum with a width up to 200 nm FWHM. Having the possibility for additional amplitude shaping FWHM is of limited meaning to predict the minimum achievable pulse duration. Much more interesting is the availability of spectral components with non-negligible amplitude in a very broad range. In our experiment we observed signal with reasonable amplitude in a range of 360 nm (580 – 940 nm). With the current focusing optic, we were only able to couple a smaller fraction of the amplifier output pulse energy into the fiber than observed previously . Nevertheless, the measured output pulse energy of 300 µJ opens the way to many interesting experiments. The broadened pulses are then send to the compressor.
To compress pulses reliably below 10 fs, it is imperative to slightly adjust the phase settings on a daily basis. Small changes in the beam path through the compressor will not lengthen unbroadened pulses substantially, but after broadening the spectrum the pulses are much more sensitive to such phase changes. Fluctuations of the energy coupled into the fiber will result in a different nonlinear phase shift to be compensated. To have the possibility to easily set the phase and keep the energy of the pulses as high as possible we opted for a scheme relying on a recently improved AOPDF from Fastlite. Redesigning the transducer and upgrading the driver electronics emerged in enhanced diffraction efficiency over a range of more than 300 nm.
Our compressor consisted of an AOPDF and a prism compressor. With a telescope consisting of two silver-coated mirrors, the beam out of the fiber is matched to the clear aperture of the AOPDF, which is 4 mm. The beam diameter is now fairly small. To avoid nonlinear effects and/or laser induced damage in the acousto optic crystal we put a 5 cm long SF57 glass block in front of it. In this way we reduced the peak intensity below 100 MW/cm2, which is low enough to avoid nonlinear distortions. The huge amount of GDD introduced by the additional glass block and the acoustooptic crystal itself could not be compensated by the AOPDF itself. With the current AOPDF a maximum delay of 3.5 ps can be introduced between the shortest and longest diffracted wavelength. The maximum introduced delay within a wavelength range of 300 nm will be much larger; thus it is necessary to add another dispersive element for compensation. We opted for a prism compressor, because of the high throughput. The diffracted beam from the AOPDF is launched into the prism compressor. The main task of the prism compressor is to compensate the introduced GDD without adding an excessive amount of high order dispersion. Thus, we built a prism compressor containing two sets of fused silica triple prisms with a clear aperture of 10 and 15 cm respectively. We opted for the triple prism configuration because it allows to keep the setup rather compact (prism separation=1.9 m), even though a low dispersive prism material is used. It is well known, fused silica introduces a minimum amount of high order dispersion when used around 800 nm. At the output of the compressor the pulse energy was 80 µJ. The duration of the compressed pulses have been characterized with a SPIDER . The sum frequency of the pair of pulses and the stretched pulse have been generated in a BBO crystal with a thickness of 60 µm cut at Θ=42.6° and φ=30°. The blue spectrum has been recorded with a 1024 element linear diode array resulting in a resolution of 1 nm. The red spectrum has been measured with a second spectrograph. With the spectral phase retrieved from the SPIDER setup and the simultaneously recorded spectral amplitude, we were able to immediately calculate the pulse shape for each set of shaping parameters.
3. Diffraction efficiency measurements
In a first step, we characterized the diffraction efficiency of the new broadband DAZZLER. The full output from the hollow fiber has been directed into the AOPDF. The transmitted radiation has been measured with a spectrograph switching the RF driver signal on and off. Both spectra have been measured with the same settings to quantitatively estimate the spectrally resolved transmission. The normalized transmission T is shown in Fig. 2. Due to the absence of other losses the diffraction efficiency is simply (1-T). The diffracted intensity depends on the magnitude of the launched RF power forming the transmission grating and the effective length of the grating. If the Bragg condition is fulfilled, diffraction occurs with rotated polarization [20,21]. The different group velocities of the two polarization directions introduce a delay between wavelength components depending on the position of the diffraction. If no additional chirp should be introduced all wavelength components must be diffracted at the same position within the crystal. In this “transform limited” case, i.e. no additional phase modulation is impressed, the grating length is very short and subsequently the diffraction efficiency is small.
Permitting that to the output pulses will carry a chirp of a few thousands of fs2 the grating is chirped and its length becomes comparable to the length of the crystal. In this case each wavelength component will see a much longer effective grating length resulting in substantially increased diffraction efficiency.
We measured the diffraction efficiency for a narrow and a broad diffracted spectrum. For the narrow bandwidth measurement, we have chosen a transfer function as a super Gaussian (order 3) shape with a wing to wing width of 50 nm. Launching the maximum RF power (10W) into the crystal, we obtained a diffraction efficiency of up to 90 % in a range from 750 to 770 nm. The width of the diffracted spectrum agrees very well with the programmed filter bandwidth. The maximum efficiency was observed by applying a chirp of -12000 fs2 ensuring a diffraction grating length comparable to the crystal length. Setting the chirp to zero resulted in a drastic reduction of the efficiency. In a next step, we measured the diffraction efficiency for setting the bandwidth of the diffraction window to 300 nm. For this setting we observed efficiency of about 70% for short wavelengths and higher than 30% in the full wavelength range of 640 nm to 890 nm. The prototype AOPDF was not calibrated to insure perfect flat response over a wide spectral range and we made no attempts to correct for the inherent wavelength dependent diffraction efficiency. Our measurements indicate that this broadband device can diffract light with high efficiency over a spectral range of more than 250 nm (FWHM).
4. Pulse compression results
The input pulses into the AOPDF have a spectrum (Fig. 3) broad enough to be compressed down to 5.5 fs in the transform limited case. Due to the limited diffraction bandwidth of the AOPDF, a small fraction of the spectrum is cut at 650 nm. An additional reduction of the spectral width is introduced by the prism compressor at 670 nm. However, the reduction of the spectral width is not a very severe limitation. Transform limited pulses are still possible with a duration of 6.6 fs and 7.2 fs, directly after the AOPDF and the prism compressor, respectively.
Before we started the compression experiment, we estimated the dispersion contribution of the system components. Major contributions stem from the additional stretching glass block, the AOPDF itself, the prism compressor and the pre-compensating TOD mirrors. Dispersion contributions from air, mirrors etc. are small and have thus neglected. A summary can be found in Table 1.
In the next step we estimated the minimum pulse duration for having no programmable element to correct for high order phase errors. To make a comparison we operated the AOPDF introducing only linear acoustic chirp and the width for diffraction window was set to 300 nm. As mentioned above, to have reasonably high diffraction efficiency it is necessary to apply a linearly chirped sound wave (GDD only). The highest integrated transmission was measured, if the nominal GDD was preset to -2000 fs2. To make a comparison with our calculations we subtracted the negative GDD added by the AOPDF from before calculating the prism separation for making the overall GDD zero. After adjusting the prism separation for the shortest pulses we measured them with the SPIDER apparatus. The retrieved pulses carried mainly high order dispersion and were 27 fs FWHM. The main pulses were accompanied by pre and post pulses with amplitudes less than 30% of that of the main pulse. The magnitude of dispersion estimated with the SPIDER was in good agreement with the calculation based on the parameters summarized in Table 1.
To reach the expected sub-10 fs pulse duration we introduced additional cubic and fourth order dispersions with the AOPDF. The quadratic dispersion contribution was set to -2000 fs2 as before. This number is below the GDD applied to achieve the maximum diffraction efficiency, but a further increase (to be compensated by a reduction of the prism separation) would reduce the ability to introduce TOD and FOD. By carefully adjusting the magnitude of the TOD and FOD parameters, we succeeded in flattening the total spectral phase curve (Fig. 4) resulting in 8 fs pulses. The large excessive negative TOD introduced by the prism compressor is nearly compensated by the rather high positive TOD of the SF57 glass block and the TeO2 crystal itself. This combination of materials requires only introducing a TOD of +3000 fs3 by the AOPDF to eliminate its overall pulse lengthening contribution. This calculated number agrees well with our experimental found value. Experimentally we found that applying an additional FOD of +3000 fs4 was necessary to reach the shortest pulses. Our calculations suggested a somewhat higher value (Table 1) for the FOD. We refrained from increasing it, because it could be only applied at the expense of a reduction of the diffraction efficiency at the wings, leading also to longer pulses due to a reduced spectral width.
In Fig. 5 the intensity and phase of the retrieved shortest pulses are shown on a linear and logarithmic scale. The residual phase in time domain is almost flat beside some π -phase jumps originating from a change of the sign of the electric field envelope.
Compensating the high order phase error enabled not only a reduction of the pulse duration by more than a factor of three, but more importantly than reducing the half width is the almost complete elimination of the pre and post pulses. The energy is now much better confined in a single pulse as shown in Fig. 5, resulting in an increase of the peak intensity by a factor of 6, compared to the case without high order phase compensation.
The new broadband AOPDF with improved driver electronics and transducer is suitable to shape the phase and amplitude of spectrally broadened high-energy laser pulses over a wavelength range of 300 nm. By the appropriate optimization of the chirp parameters we succeeded in the generation of nearly transform limited 8 fs pulse with 9 GW peak power.
With the system, as it is described here, our peak power is limited to about ten GW. However, improving the system by adding an additional amplifier stage, will make it possible to increase the peak power by 2 or 3 orders of magnitude without a substantial lengthening of the pulses by gain narrowing or other limiting effects. Having clean sub-10 fs pulses with a peak power of up to 1 TW will offer new possibilities such as the study of the interaction of intense laser light with matter with solid density or the generation of single attosecond XUV pulses in the water window.
This work has been sponsored by the Austrian Science Fund grant F016 and by the BMBF project 13N7922, J. Seres acknowledges support from a Lise Meitner fellowship M645. Ch. Spielmann acknowledges support from the DFG grant SP 687/1-1.
References and links
1. C. J. Bardeen, V. V. Yakovlev, K. R. Wilson, S. D. Carpenter, P. M. Weber, and W. S. Warren, “Feedback quantum control of molecular electronic population transfer,” Chem. Phys. Lett. 280, 151–158 (1997). [CrossRef]
2. T. C. Weinacht, J. Ahn, and P.H. Buckbaum, “Controlling the shape of a quantum wavefunction,” Nature 397, 233–235 (1999). [CrossRef]
3. M. Drescher, M. Hentschel, R. Kienberger, G. Tempea, Ch. Spielmann, G. A. Reider, P. B. Corkum, and F. Krausz, “X-ray Pulses Approaching the Attosecond Frontier,” Science 291, 1923–1927 (2001). [CrossRef] [PubMed]
4. Ch. Spielmann, N. H. Burnett, S. Sartania, R. Koppitsch, M. Schnürer, C. Kan, M. Lenzner, P. Wobrauschek, and F. Krausz, “Generation of Coherent X-rays in the Water Window Using 5-Femtosecond Laser Pulses,” Science 278, 661–664 (1997). [CrossRef]
5. R. Bartels, Sterling Backus, Ivan Christov, Henry Kapteyn, and M. Murnane, “Attosecond time-scale feedback control of coherent X-ray generation,” Chem. Phys. 267, 277–289 (2001). [CrossRef]
7. B. A. Richmann, S. E. Bisson, R. Trebino, E. Sidick, and A. Jacobson, “All-prism achromatic phase matching for tunabel second-harmonic generation,” Appl. Opt. 38, 3316–3323 (1999). [CrossRef]
8. C.H. Brito Cruz, P. C. Becker, R. L. Fork, and C. V. Shank, “Phase correction of femtosecond optical pulses using a combination of prisms and gratings,” Opt. Lett. 13, 123–125 (1988). [CrossRef]
9. M. Hentschel, S. Uemura, Z. Cheng, S. Sartania, Ch. Spielmann, and F. Krausz, “High-dynamic-range pulse-front steepening of amplified femtosecond pulses by third-order dispersion,” Appl. Phys. B 68, 145–148 (1999). [CrossRef]
10. S. Sartania, Z. Cheng, M. Lenzner, G. Tempea, Ch. Spielmann, F. Krausz, and K. Ferencz, “Generation of 0.1-TW 5-fs optical pulses at a 1-kHz rate,” Opt. Lett. 22, 1562–1564 (1997). [CrossRef]
12. T. Witte, D. Zeidler, D. Proch, K. L. Kompa, and M. Motzkus, “Programmable amplitude and phase-modulated femtosecond laser pulses in the mid-infrared,” Opt. Lett. 15, 131–133 (1988).
14. A. Efimov and D. H. Reitze, “Programmable dispersion compensation and pulse shaping in a 26-fs chirped-pulse amplifier” Opt. Lett. 23, 1612–1614 (1998). [CrossRef]
15. L. Xu, N. Nakagawa, R. Morita, H. Shigekawa, and M. Yamashita, “Programmable Chirp Compensation for 6-fs Pulse Generation with a Prism-Pair-Formed Pulse Shaper,” IEEE J. Quantum Electron. 36, 893–899 (2000). [CrossRef]
16. V. V. Lozovoy and M. Dantus, “Photon echo pulse sequences with femtosecond shaped laser pulses as a vehicle for molecule-based quantum computation,” Chem. Phys. Lett. 351, 213–221 (2002). [CrossRef]
17. M. D. Skeldon, “Optical pulse-shaping system based on an electro-optic modulator driven by an aperture-coupled stripline electrical-waveform generator,” J. Opt. Soc. Am. B 19, 2423–2426 (2002). [CrossRef]
18. E. Zeek, K. Bartels, M. Murnane, H. Kapteyn, S. Backus, and G. Vdovin, “Adaptive pulse compression for transform-limited 15-fs high-energy pulse generation,” Opt. Lett. 25, 587–589 (2000). [CrossRef]
19. T. Brixner and G. Gerber, “Femtosecond polarization pulse shaping,” Opt. Lett. 26, 557–559 (2001) [CrossRef]
20. D. Kaplan and P. Tournois, “Theory and performance of the acousto optic programmable dispersive filter used for femtosecond laser pulse shaping,” J. Phys. IV France 12, 69–75 (2002). [CrossRef]
21. F. Verluise, V. Laude, Z. Cheng, Ch. Spielmann, and P. Tournois, “Amplitude and phase control of ultrashort pulses by use of acousto-optic programmable dispersive filter: pulse compression and shaping,” Opt. Lett. 25, 575–577 (2000). [CrossRef]
22. C. Iaconis and I. A. Walmsey, “Self-Referencing Spectral Interferometry for Measuring Ultrashort Optical Pulses,” IEEE J. Quantum Electron. 35, 501–509 (1999). [CrossRef]