We present a method for an efficient spectral shift and compression of pulses from a femtosecond laser system. The method enables generation of broadly tunable (615 – 985 nm) narrow bandwidth (≈10 cm−1) pulses from the femtosecond pulses at 1030 nm. It employs a direct parametric amplification – without spectral filtering – of highly chirped white light by a narrow bandwidth (<5 cm−1) 515 nm pump pulse. The system, when pumped with just 200 μJ of the fundamental femtosecond pulse energy, generates pulses with energies of 3-9 μJ and an excellent beam quality in the entire tuning range.
© 2012 OSA
More and more ultrafast spectroscopy experiments require tunable narrow bandwidth pulses that are synchronous with independently tunable femtosecond pulses. Examples of such experiments are: broadband surface sum frequency generation spectroscopy , femtosecond stimulated Raman spectroscopy (FSRS)  or coherent anti-Stokes Raman scattering (CARS) [3,4]. To generate synchronous, independently tunable pico and femtosecond pulses one can either use sophisticated multi amplification laser chains  or start with a single femtosecond source and transform femtosecond pulses into picosecond ones. The main task here is therefore an efficient transfer of energy from a broad femtosecond laser pulse to a narrow bandwidth picosecond one. This goal becomes even more difficult to achieve if one wants to generate tunable narrow bandwidth pulses. When considered in the spectral domain the process of generating tunable picosecond pulses from a femtosecond laser source can be viewed as a superposition of two phenomena: spectral compression leading to long pulses that are close to the Fourier limit and an adjustable spectral shift resulting in the tunability.
Many different configurations allowing for such pulse transformation have been proposed [6–17]. They can be divided into two groups: linear and nonlinear techniques. In linear techniques the superfluous spectral components of the broadband femtosecond pulse are simply discarded by either using narrow bandwidth spectral filters or by placing a slit in the Fourier plane of the 4f pulse shaper [6,7]. This method, however, results in a high power loss and allows for only a limited tunability. Most of the efforts are therefore concentrated on nonlinear methods which provide superior performance.
The simplest nonlinear method available relies on the spectral narrowing of pulses during the second harmonic generation (SGH) in long crystals in the presence of a strong group velocity mismatch (GVM). The idea itself has been known for a long time [18,19] and recently it has been used by Marangoni and associates to transform tunable femtosecond pulses into picosecond ones [8,10]. The method requires only the use of a suitable nonlinear medium and a tunable source of femtosecond pulses. It has some drawbacks, though, as it strongly affects the temporal pulse profile of the processed pulses [8,9] and requires long crystals (>2.5 cm) with different cuts for different central wavelengths in order to meet phase matching conditions . The best efficiencies reported were 10-20% and bandwidths ≈10 cm−1 or less. More complex methods employ sum [11,12] and difference [13–15] frequency generation of chirped pulses. Especially, sum frequency generation of two copies of a femtosecond pulse with opposite chirps is widely used. This method can be very efficient as we will show below and, potentially, can be scaled to the very small bandwidths.
Yet another group of approaches focuses on manipulation of the white light supercontinuum which is the source of the seed pulses for tunable narrow bandwidth pulse generation. A linear spectral filtering of the white light have been tested so far by two groups [16,17]. The resulting pulses were 25-30 cm−1, even with multi optical parametric amplification stages and multiple filtering steps .
In this paper, we propose a method in which the white light is stretched and directly amplified in an optical parametric amplifier (OPA) pumped by a narrow bandwidth pulse. Imposing a high chirp rate on white light is relatively easily. In the time domain picture the principle of our method can be seen as making the white light quasi monochromatic within the time window of the picosecond pump pulse. In this approach no extra filtering of spectral components is required.
Our experimental approach can divided into 3 steps. In the first step, we generate an energetic chirp-free picosecond pump pulse starting from the femtosecond source. The method of choice here is the sum frequency mixing of the two oppositely chirped copies of the same femtosecond pulse . This method provides maximum flexibility, but other techniques can be used as well. In the second step, we generate the white light which is then stretched in a suitable white light stretcher unit. The final step is the parametric amplification of the white light by a picosecond pump. In order to increase conversion efficiency and output pulse stability the amplification process is divided into two stages.
We use Yb-based femtosecond laser system which typically produces light pulses with the time duration of order of hundreds of femtoseconds. The initial pulse duration, however, has little influence on the performance of proposed method. It can be easily applied to other sources including Ti:sapphire systems. Since Yb based sources have much more to offer in effectiveness and compactness they are an interesting alternative to the Ti:sapphire based ones. To best of our knowledge our work is the first demonstration of the spectral compression on Yb based femtosecond laser source.
2. Experiment and result
The experimental configuration is depicted in Fig. 1 . A part of the output beam of a commercial femtosecond Yb:KGW amplifier (Pharos, Light Conversion) is used to drive the whole setup. The amplifier delivers 180 fs pulses at 1030 nm central wavelength. Our system operates at a constant energy of 200 μJ of 1030 nm pulses with the repetition rate of 1 kHz. A small fraction (~5 μJ) of the fundamental beam is used for white light supercontinuum generation in a 3 mm sapphire plate. The remaining 1030 nm beam is split in half by a beam splitter. One part goes to the negative dispersion unit, the other to the positive dispersion unit. Both units stretch pulses to about 5 ps, imposing the same chirp rate but with opposite signs. The time duration of the stretched pulses were calculated and then verified experimentally by measuring crosscorelation of the pulse with a fraction of fundamental 180 fs pulse. The negative dispersion unit consists of a holographic transmission diffraction grating (40x40x2 mm, Wasatch Photonics) with 900 grooves/mm and two Porro (right angle) prisms. The beam enters the negative dispersion unit through the grating at Litrrow angle (27.6 degrees AOI). After a distance of about 25 mm the first prism reflects the beam back with a parallel shift in the horizontal plane. The second prism rotated by 90° reflects the beam back and shifts it in the vertical plane. The output beam reverses the input beam route but at a different height until it is intercepted by the pick-up mirror set below the input beam. This unit is compact and extremely easy to align. It has an efficiency of nearly 80%. It can be further increased by using better quality optics.
The positive dispersion unit is a folded Martinez stretcher configuration . It utilizes the same grating type as the negative dispersion unit, a positive cylindrical lens of focal length 200 mm, a Porro prism and a flat back mirror. The beam enters the unit pointing slightly down and exits at a different height so it could be intercepted by a lowered mirror. The beams from both dispersion units are subjected to proper delays, focused by 500 mm lenses and crossed at an angle of 3 degrees in the 1.5 mm-thick BBO crystal (type I, 23.4°, AR coated for 1030nm and 515 nm on both sides). Frequency doubling of the individual beams is weak because phase matching conditions are not fulfilled – the critical phase matching angle is in the plane of incoming beams. The position of the crystal was optimized with respect to the conversion efficiency and found to be about 100 mm before the focal plane. The upconverted beam of 515 nm generated in the BBO crystal is easily separated from the fundamental beams due to the noncolinear configuration used. The combined energy of two pulses impinging on the BBO crystal was 147 μJ and energy of the 515 nm pulse was more than 58 μJ. This indicates 40% efficiency of the sum frequency generation process and 30% overall efficiency with respect to the fundamental femtosecond pulse. The measured M2 values for the 515 nm beam were less than 1.2 in both directions. The spectral width of the upconverted signal was minimized by varying the distance between the prism and the grating in the positive dispersion unit, resulting in 4.8 cm−1 (0.127 nm) bandwidth FWHM of the output pulse.
The supercontinuum beam is directed to the white light stretcher, which consists of a reflective diffraction grating (2000 grooves/mm), a folding mirror, a corner cube retroreflector (trihedral prism) and a back flat mirror. The total distance between the grating and the retroreflector is 45 cm. The white light stretcher works at constant deviation angle (≈5 degrees). By rotating the diffraction grating around a vertical axis coinciding with the incident point a small fraction of white light spectrum is selected. The retroreflector serves two purposes: (1) it keeps handedness of the image unchanged, and (2) it shifts vertically the beam, so the beam follows the same path in the horizontal plane regardless the grating rotation. The last feature is very important because wavelength tuning does not alter the optical path in the stretcher. The group delay dispersion (GDD) realized by this setup for 600 nm and 985 nm are −2⋅106 fs2 and −1,4⋅108 fs2, respectively. Transmission of the white light stretcher is restricted to the window of 610 – 990 nm by the limit of the antireflection coating of optical elements used (shorter wavelengths) and by the diffraction grating (longer wavelengths). To reach even longer wavelengths (up to the central wavelength of the fundamental beam) a grating with a reduced density of lines should be used.
The stretched white light is amplified in a two stage BBO-based OPA. Approximately 20% of the picosecond light is used to pump the first stage. Because of a low peak intensity of the pump light, the 4 mm-thick BBO crystal (type I, 24.5°, AR coated for 515 nm on one side and AR coated for 700-1000 nm on the other side) is placed at the focus of 500 mm focal length lens which focuses the pump beam. The signal beam and the pump beam are collinear in order to achieve the best possible spatial overlap. However, due to the walk-off of the extraordinary pump beam in the birefringent BBO crystal, the interaction length is restricted to about 3 mm (1/e2 beam diameter of the pump beam is 170 μm and the walk-off angle 53-57 mrad). Optical parametric gain of up 50 is observed on a single pass. In order to achieve a higher gain, a mirror is placed right after the crystal. Both the pump and the seed beams retrace their path in the crystal, so another gain of 50 can be obtained at this stage. As a result, the signal pulse energies after the first stage were 0.1-0.2 μJ. Due to the tight focusing and the walk-off, the amplified signal beam shape is elliptical with ratio of major to minor axis roughly 2:1. This is compensated for with a cylindrical telescope. The beam is then telescoped to the 1/e2 beam diameter of 0.7 mm. The pump beam is also set to the similar spot size. The signal beam is amplified in a 6 mm-thick BBO crystal (type I, 24.5°, the same AR coating as in the BBO used in the I stage) in a single pass configuration. A small angle between the beams is introduced in order to easily separate the signal, idler and the pump beam.
The output signal pulse energies, bandwidths and a typical beam profile are shown in Fig. 2 . The signal pulse energy is higher than 3 μJ for the entire spectral range 615 – 985. About 9 μJ of energy is reached at the peak (685 nm) which indicates 21% energy conversion efficiency and almost 28% quantum efficiency in the second stage of the OPA. In the spectral range 685-950 nm the output pulse energy almost perfectly follows the theoretical curve if one assumes a constant quantum efficiency of the amplifier. Beyond this range the efficiency drops probably due the increased absorption of the idler wave in the BBO crystal (shorter wavelengths) and the decreased quality of the seed light caused by the strong modulations close to the fundamental wavelength (longer wavelengths). M2 values of the signal at maximum conversion efficiency point were measured to be less than 1.3.
The spectra shown in Fig. 3 were taken with a grating spectrometer (Andor Shamrock 500 + Newton CCD camera) utilizing 600 grooves/mm grating. Closer examination was performed on an arbitrarily chosen spectrum centered at 680.2 nm and its results are shown in Fig. 4 . The spectrum was measured with a 2400 grooves/mm grating (with the resolution of 0.6 cm−1). The corresponding temporal profile measured by crosscorrelation with a fraction of fundamental 180 fs pulse is also shown.
The experimental spectral profile is well fitted with a Gaussian profile, but the temporal profile is not. This can explained by the presence of a similar structure experimentally observed in the temporal profile of the picosecond pump pulse. Assuming flat spectral phase and the given Gaussian spectral profile, we should expect a pulse duration of 1.45 ps. This would indicate that the output pulse with its 2.2 ps duration corresponds to approximately 1.5 of the Fourier limit. However, if we assume that the OPA output pulse acquires the temporal profile identical with that of the pump pulse and, in addition, a flat temporal phase then the calculated spectrum is only slightly narrower (about 5%) than the experimental spectrum. This indicates that, indeed, the signal pulses are close to transform limited, although their temporal profile has some nontrivial structure due to a time-structured pump pulse.
Compared to other solutions reported so far, our setup provides a significant improvement in efficiency, simplicity and a scaling potential. Since different methods for spectral compression use different routes from femtosecond to picosecond regime one has to be careful when comparing relative efficiencies. Here we use the overall efficiency for the conversion of the fundamental femtosecond pump pulse energy to the narrow bandwidth picosecond pulse at the peak of the tuning curve. Typical efficiencies reported by Cerullo group were in range of 0.5-1% when employing the concept of spectral compression by SHG [8–10]. In the approach of Co and associates the overall efficiency was 0.5% . The best efficiency of nearly 3% was recently reported by Ernsting group, however at the cost of high complexity of the setup (four amplification stages, two 4f-spectral filters) and modest bandwidths 25-30 cm−1 . In our setup we obtained 4.5% overall efficiency with only two amplification stages and one of the narrowest bandwidths reported so far (≈10 cm−1) without using 4f-spectral filters at all.
We have proposed and experimentally demonstrated a method for generating broadly tunable narrow bandwidth picosecond pulses starting from a femtosecond laser source. The method employs a direct parametric amplification of a highly chirped white light by a narrow bandwidth pump pulse generated by the sum frequency mixing of two copies of femtosecond pulses with opposite chirps. Conversion efficiency of 30% from the fundamental femtosecond pulse to picosecond pump was reached. Using only a fraction (200 μJ) of the available fundamental pump pulse we obtained multi-μJ narrow bandwidth picoseconds pulses, close to Fourier limit and with an excellent beam quality. The other fraction of the fundamental light can be used to generate tunable broad bandwidth femtosecond pulses which are perfectly synchronized with the picoseconds ones. The synchronized picosecond and femtosecond tunable pulses can be used in experiments where simultaneous high temporal and spectral resolution can be obtained.
The presented setup is scalable to even narrower spectra. To achieve that one only needs to impose larger chirp rates on the fundamental beams in the sum frequency generation unit and on the white light pulse. Both can be achieved using simple geometrical means. Additional preamplifier of the white light by the portion of femtosecond light may be considered to avoid difficulties in first stage due to reduced peak intensity. In our experiment the Yb based femtosecond source was used but the method can be easily applied to Ti:sapphire based systems, which are widely used in the research laboratories.
References and links
1. T. Ishibashi and H. Onishi, “A multiplex infrared-visible sum-frequency spectrometer with wavelength tunability of the visible probe,” Appl. Phys. Lett. 81(7), 1338–1340 (2002). [CrossRef]
2. R. R. Frontiera and R. A. Mathies, “Femtosecond Stimulated Raman Spectroscopy,” Laser Photonics Rev. 5(1), 102–113 (2011). [CrossRef]
3. B. D. Prince, A. Chakraborty, B. M. Prince, and H. U. Stauffer, “Development of simultaneous frequency- and time-resolved coherent anti-Stokes Raman scattering for ultrafast detection of molecular Raman spectra,” J. Chem. Phys. 125(4), 044502 (2006). [CrossRef] [PubMed]
4. J. D. Miller, M. N. Slipchenko, T. R. Meyer, H. U. Stauffer, and J. R. Gord, “Hybrid femtosecond/picosecond coherent anti-Stokes Raman scattering for high-speed gas-phase thermometry,” Opt. Lett. 35(14), 2430–2432 (2010). [CrossRef] [PubMed]
5. G. M. Greetham, P. Burgos, Q. Cao, I. P. Clark, P. S. Codd, R. C. Farrow, M. W. George, M. Kogimtzis, P. Matousek, A. W. Parker, M. R. Pollard, D. A. Robinson, Z.-J. Xin, and M. Towrie, “ULTRA: A Unique Instrument for Time-Resolved Spectroscopy,” Appl. Spectrosc. 64(12), 1311–1319 (2010). [CrossRef] [PubMed]
6. S. Shim and R. A. Mathies, “Generation of narrow-bandwidth picosecond visible pulses from broadband femtosecond pulses for femtosecond stimulated Raman,” Appl. Phys. Lett. 89(12), 121124 (2006). [CrossRef]
7. A. Lakshmanna, B. Mallick, and S. Umapathy, “Ultrafast Raman loss spectroscopy: a new approach to vibrational structure determination,” Curr. Sci. 97(2), 210–217 (2009).
8. M. A. Marangoni, D. Brida, M. Quintavalle, G. Cirmi, F. M. Pigozzo, C. Manzoni, F. Baronio, A. D. Capobianco, and G. Cerullo, “Narrow-bandwidth picosecond pulses by spectral compression of femtosecond pulses in second-order nonlinear crystals,” Opt. Express 15(14), 8884–8891 (2007), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-15-14-8884. [CrossRef] [PubMed]
9. M. Marangoni, D. Brida, M. Conforti, A. D. Capobianco, C. Manzoni, F. Baronio, G. F. Nalesso, C. De Angelis, R. Ramponi, and G. Cerullo, “Synthesis of picosecond pulses by spectral compression and shaping of femtosecond pulses in engineered quadratic nonlinear media,” Opt. Lett. 34(3), 241–243 (2009). [CrossRef] [PubMed]
10. E. Pontecorvo, S. M. Kapetanaki, M. Badioli, D. Brida, M. Marangoni, G. Cerullo, and T. Scopigno, “Femtosecond stimulated Raman spectrometer in the 320-520nm range,” Opt. Express 19(2), 1107–1112 (2011), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-19-2-1107. [CrossRef] [PubMed]
11. F. Raoult, A. C. L. Boscheron, D. Husson, C. Sauteret, A. Modena, V. Malka, F. Dorchies, and A. Migus, “Efficient generation of narrow-bandwidth picosecond pulses by frequency doubling of femtosecond chirped pulses,” Opt. Lett. 23(14), 1117–1119 (1998). [CrossRef] [PubMed]
12. S. Laimgruber, H. Schachenmayr, B. Schmidt, W. Zinth, and P. Gilch, “A femtosecond stimulated Raman spectrograph for the near ultraviolet,” Appl. Phys. B 85(4), 557–564 (2006). [CrossRef]
13. G. Xu, L. Qian, T. Wang, H. Zhu, C. Zhu, and D. Fan, “Spectral narrowing and temporal expanding of femtosecond pulses by use of quadratic nonlinear processes,” IEEE J. Sel. Top. Quantum Electron. 10(1), 174–180 (2004). [CrossRef]
14. G. Veitas and R. Danielius, “Generation of narrow-bandwidth tunable picosecond pulses by difference-frequency mixing of stretched pulses,” J. Opt. Soc. Am. B 16(9), 1561–1565 (1999). [CrossRef]
15. H. Luo, L. Qian, P. Yuan, and H. Zhu, “Generation of tunable narrowband pulses initiating from a femtosecond optical parametric amplifier,” Opt. Express 14(22), 10631–10635 (2006), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-14-22-10631. [CrossRef] [PubMed]
16. D. T. Co, J. V. Lockard, D. W. McCamant, and M. R. Wasielewski, “Narrow-bandwidth tunable picosecond pulses in the visible produced by noncollinear optical parametric amplification with a chirped blue pump,” Appl. Opt. 49(10), 1880–1885 (2010). [CrossRef] [PubMed]
18. D. H. Auston, “Nonlinear spectroscopy of picosecond pulses,” Opt. Commun. 3(4), 272–276 (1971). [CrossRef]
19. C. Radzewicz, P. Wasylczyk, and J. S. Krasiński, “A Poor Man’s FROG,” Opt. Commun. 186(4-6), 329–333 (2000). [CrossRef]
20. O. Martinez, “3000 times grating compressor with positive group velocity dispersion: Application to fiber compensation in 1.3-1.6 µm region,” IEEE J. Quantum Electron. 23(1), 59–64 (1987). [CrossRef]