We generate 1.24 mJ, 390 fs pulses at 1035 nm in a CPA laser system featuring a 2.8 mJ Yb:CaF2 regenerative amplifier, stretcher/compressor based on a single chirped volume Bragg grating and a compact, low-dispersion grating compressor. The auxiliary compressor is used to effectively pre-compensate the intra-cavity dispersion of the amplifier.
© 2014 Optical Society of America
The development of the chirped pulse amplification (CPA) technique  in the mid 80’s left an enduring legacy in the field of high intensity lasers, by allowing a dramatic downscaling of high peak power systems and the subsequent dissemination of those devices to the scientific community. Nowadays, CPA lasers are the omnipresent choice for generating high-intensity, sub-picosecond laser pulses.
The temporal stretching and compression operated in the CPA technique relies on the broadband nature of short pulses, by using e.g. material or angular dispersion to establish a wavelength-dependent delay line. Traditional stretchers and compressors consist in optical arrangements with one or more diffraction gratings [2,3], providing quadratic dispersion of opposite signs. However, such systems are often bulky, complex, extremely demanding in terms of alignment precision and with limited long-term stability.
Beyond the intense research in compact high-energy amplification concepts such as fiber lasers , thin disk amplifiers  and slab amplifiers , alternative stretcher/compressor devices have been explored in order to make such temporal manipulations simpler, more robust and compact [7–9]. One of the most promising candidates for use as both a stretcher and a compressor of multi-mJ energy pulses is the chirped volume Bragg grating (CVBG) in photo-thermal-reflective glass (PTR). This optical device, available in sizes of a few centimeters long, reflects different wavelengths in holographically recorded planes located at different depths inside a transparent medium effectively imposing a linear chirp. The reflection for different wavelengths occurs when the Bragg condition is met [10,11]:
These properties make CVBGs extremely attractive for pulse stretching and compression in a single device, in an extremely simple, elegant and compact setup . Their most important advantage, in comparison with e.g. competing fiber-based devices [7,8], is the fact that CVBGs in PTR glass may exhibit apertures up to 10 mm, allowing the compression of pulse energies up to 20 mJ to ~ps-level durations . However, the absence of a mechanism for tuning the amount of dispersion makes their applicability limited in some situations, such as when there is a need to compensate additional material dispersion acquired during propagation through optical elements and gain media in an amplification chain. This is the case for e.g. a regenerative amplifier, where hundreds of round-trips through the cavity optics (corresponding to up to a few meters of different dispersive media) are not uncommon. In this situation, the compression is only partial, as the net group delay dispersion (GDD) of the output pulse is positive. Nevertheless, this GDD is typically orders of magnitude smaller than that imparted by the CVBG, meaning that it may be cancelled by the use of an auxiliary dispersive line. For instance, this approach was used to generate 2 mJ, 600 fs pulses at 1047 nm by using transmission holographic gratings in a hybrid stretcher configuration .
In this work we describe the development and performance of a sub 400 fs CPA laser system based on a diode-pumped amplifier, and a single CVBG for both stretching and compression, assisted by a compact low dispersion diffraction grating compressor. To our knowledge, this is the shortest pulse duration obtained in a mJ-level, CVBG-based CPA system. This value is ~67% shorter than that reported in Ref , which is relevant for nonlinear phenomena where the shortest pulse duration is crucial such as e.g. femtosecond filamention and white-light continuum generation. For a question of maximizing the energy efficiency of the system, the grating compressor is used at the stretcher stage, i.e. before amplification. The system consists in a 2.8 mJ Yb:CaF2 regenerative amplifier operating at 1035 nm, seeded by a commercial femtosecond oscillator. The pulses are first stretched in a short CVBG in a double-pass configuration and then sent to a fine-tuning, quadratic Treacy compressor that pre-compensates the global dispersion taking into consideration the GDD later generated by the amplifier. After being amplified, the pulses are compressed in the same CVBG. The mechanism described allows the removal of the remaining GDD in a tunable fashion, while keeping a compact footprint.
2. Experimental setup
The experimental setup is shown in Fig. 1. The regenerative amplifier is seeded by a commercial femtosecond Ti:sapphire, Kerr-lens modelocked oscillator (Coherent Mira 900F, pumped by a Verdi X) delivering 1030 nm pulses with a duration of ~150 fs, an average power of 260 mW and a repetition rate of 76 MHz. The pulse train is isolated by means of a Faraday isolator (FI1), followed by a 10 Hz pulse picker consisting of a pair of crossed Glan laser polarizers with an extinction ratio of 5 × 105:1 (CVI Laser Optics) and a Pockels cell in-between with a contrast ratio of 5 × 103:1 (QX1020, Cleveland Crystals).
The 35.2 mm long, 5 × 8 mm2 aperture CVBG (OptiGrate Corp) has a chirp rate of 26 ps/nm, limited to a 13 nm bandwidth between 1028 and 1041 nm, being transparent to wavelengths outside this range. This causes a hard spectral clipping and reduces the pulse bandwidth to ~6.4 nm FWHM (Fig. 2). The pulse train is injected into the CVBG at an incident angle of 5° and a retro-reflecting mirror (M3) provides the double-pass. Apart from doubling the stretching factor, this step should also cancel the spatial chirp that emerges after a single pass. In reality, this is not entirely cancelled, and we were able to observe a residual spatial chirp by probing along the beam horizontal profile with a 400 µm diameter optical fiber connected to a spectrometer. This effect has been attributed to limitations in the CVBG writing process that lead to inhomogeneities, causing the reflection of different frequencies at slightly different angles [12,15].
For this configuration the stretched pulse duration is therefore ~330 ps, corresponding to a GDD ~30 ps2. The grating orientation is specifically chosen so that the imparted GDD is positive. In order to separate the incoming and outgoing beams a thin film polarizer (TFP1) and quarter-wave plate (λ/4) are used.
Before the amplification process, the stretched pulses are sent to a compact, low-dispersion Treacy compressor. This consists of two parallel, 25 × 25 mm2, 600 lines/mm reflective gold ruled gratings (DG1-2), separated by ~19 cm, introducing a negative GDD ~–0.3 ps2. This two-stage stretching effectively allows pre-cancelling the otherwise residual positive material dispersion generated during the amplification process. In the case of a regenerative amplifier this can be significant, due to the thickness of the intra-cavity optics and the high number of passes.
The stretched pulses are then sent to the regenerative amplifier where they are injected by passing through the thin film polarizer TFP3 (Layertec GmBH), whose contrast ratio is 5 × 102:1. Although this value is smaller than the input one, it is not very critical for a mJ level amplifier, and it can be improved if necessary e.g. by using a double Pockels cell and polarizer cascade .
The amplifier consists in a 1.35 m linear cavity delimited by two concave mirrors with radii of curvature 500 mm (R500) and 1000 mm (R1000). The crystal is placed at the cavity waist (Gaussian radius ~250 μm) and surrounded by two dichroic mirrors (DM1-2), transparent to the pump radiation. The pump consists of a 75 W fiber-coupled diode laser (Amtron GmBH) at 940 nm, set at a pulse width of 3 ms. The beam is focused by achromatic lenses (AL), in a double-pass configuration, into the 5 mm thick Yb:CaF2 crystal, whose doping concentration is 3at%, placed in a water-cooled mount. An intra-cavity Pockels cell (PC) operating in passive mode as quarter-wave retarder acts as an optical switch, allowing the pulses to stay in the cavity when switched on and extracting them when switched off. As beam injection and extraction are done along the same direction through TFP3, we have installed a Faraday rotator (FR), half-wave plate (λ/2) and another polarizer (TFP2) before the cavity to separate these two beams. The output beam is collimated by a 75 cm focal length plano-convex lens (SL) and crosses a Faraday isolator (FI2) placed to avoid optical feedback.
Finally the pulses are sent again to the CVBG in a direction opposite to the previous one, with a similar incident angle. At this stage the 1/e2 beam diameter is ~1.8 mm. Mirror M12 provides the second pass through the CVBG, exactly compensating the positive dispersion generated previously by itself. As in the stretching stage, the incoming and outgoing beams are separated by a polarizer TFP4 and a λ/4. Their energy is measured with a pyroelectric energy detector (J9LP, Coherent) connected to an oscilloscope. Apart from the main pulse we have verified the existence of a post pulse containing 10% of the energy and a low ASE pedestal containing some tens of µJ.
3. Measurement of uncompensated GDD
The optimization of the compression was done in two stages. In the first one we simply measured the remaining uncompensated dispersion when using the CVBG alone as stretcher and compressor (i.e. the grating compressor was bypassed). In the second stage, the grating compressor was introduced and we optimized the pulse compression.
For the GDD measurement the regenerative amplifier was operated with similar parameters as for the full CPA configuration, providing >2 mJ pulses after reaching saturation. The imperfectly compressed pulses were characterized by a homemade scanning second harmonic frequency-resolved optical gating (SHG-FROG) diagnostic. The retrieved results for spectrum, spectral phase, temporal profile and temporal phase are presented in Fig. 3. Due to the symmetry of the trace provided by this diagnostic, the orientation of the spectral and temporal phase curves was defined a priori as corresponding to positive dispersion, since inside the amplifier cavity the pulses cross a few meters of dispersive media, taking into account the number of round trips through the gain media and Pockels cell crystal. This was confirmed later by monitoring the pulse duration after the introduction of the auxiliary grating compressor.
Using a fourth-order polynomial fit to the spectral phase, we have calculated the remaining GDD to be 2.93 × 105 fs2. This agrees with the calculated GDD taking into account the intracavity optical media, consisting of 5 mm of Yb:CaF2, 24 mm of KD*P and 6 mm of fused silica, for 200 roundtrips. Although much lower than the dispersion imparted by the CVBG, this is large enough to prevent pulse compression to values below ~1.7 ps, 4.9 times larger than the Fourier transform-limited value (FTL). Concerning the third-order dispersion (TOD), the remaining value is −4.35 × 107 fs3, two orders of magnitude higher than what would be expected from the intracavity dispersion alone (4.65 × 105 fs3) and with the opposite sign. One can therefore conclude that this must be due to the CVBG, since no other source of such TOD is present. The reason for the uncompensated TOD is most certainly linked to imperfection in the photosensitive material where the grating is recorded and to the holographic recording process, which currently is still technologically limited .
We calculated that it is possible to compensate this residual dispersion with a compact (15-20 cm in a double-pass configuration), low dispersion (600 l/mm) compressor based on small-aperture reflective gold ruled gratings. Given the typical low efficiency of double-pass grating compressors (~60%), ours was installed before the regenerative amplifier in order to preserve the output pulse energy. However this choice leads to a somewhat higher number of round-trips for attaining the same energy, since a weaker signal is injected in the cavity, leading to a higher residual dispersion. This could however be suitably cancelled by performing a fine-tuning of the distance between the gratings. Table 1 summarizes the measured and calculated values for the GDD and TOD of all the components in the full setup.
4. Results and discussion
In the final configuration of the CPA system, we characterized the amplification performance of the 10 Hz diode-pumped Yb:CaF2 regenerative amplifier by measuring the output energy of 5000 pulses. For a 55 W pump power, saturation was reached after ~225 round-trips, providing 2.81 mJ pulses with excellent energy stability (RMS < 1.8%). During the amplification process the spectrum redshifts from 1031 nm to 1035 nm (Fig. 4). This effect can be explained by the fact that we used a low pump power around 940 nm , instead of at the main absorption peak at 980 nm. Being a quasi-three-state gain material, in these conditions Yb:CaF2 favors operation at longer wavelengths where the signal absorption is lower. In the future this effect can also be mitigated by decreasing the pump focal spot in crystal.
Figure 5 (left) shows the amplified beam intensity profile, which is approximately Gaussian (M2 = 1.28). This means that the pump focal spot is large enough to allow the fundamental Gaussian mode as well as higher transversal modes.
In order to optimize the compression, the incidence angle into the CVBG and the distance between the diffraction gratings in the auxiliary compressor was adjusted to maximize the second harmonic signal in the position of zero delay between the arms of an SHG scanning autocorrelator. The minimum value obtained for the autocorrelation is ~560 fs FWHM (Fig. 6, black curve). We calculated a theoretical fit to this curve by taking into account (i) the spectral clipping introduced by the CVBG transmission window (1028-1041 nm) and (ii) a residual third-order dispersion, which gives rise to a low pedestal and a small increase in the pulse duration. The residual GDD is sufficiently low so that it does not change the pulse duration at this level. We also calculated the B-integral of the amplifier, assuming an output efficiency of 80%, obtaining a low enough value (~1.1) to play a significant role. The resulting autocorrelation function (Fig. 6, gray curve) adjusts well to the measured autocorrelation when the pulse duration is 390 fs and a TOD ≈-1.65 × 107 fs3 is introduced, which is comparable to the calculated one (Table 1). Since the minimum width pulse allowed by the spectral clipping is 350 fs, the obtained duration is just 1.11 × FTL.
The output energy of the compressed pulse is 1.24 mJ with an RMS <1.8%, in line with the output parameters of the regenerative amplifier. The total efficiency of the compressor setup (CVBG and TFP4), considering the 7.5% loss introduced by the Faraday isolator (FI2), is 47.8% with most of the energy loss due to the relatively low transmission of the CVBG in a double-pass configuration (typ. 70-90% per pass, depending on the angle of incidence). The final beam profile changes during compression (Fig. 5, right) presenting a near-field elliptical shape and a mean M2 value of 1.81 (1.77 and 1.84 for the x and y axis respectively). This is most likely due to the input M2 being greater than unity and to the uncompensated spatial chirp induced by the CVBG  and effectively we measured a residual, irregular vertical chirp after a double pass, especially when the CVBG is used along the negative dispersion direction.
In summary we have developed a compact CPA laser system with a double stage stretcher, consisting of a single 3.5 cm long chirped volume Bragg grating and a compact, low dispersion Treacy compressor installed before a 10 Hz, 2.81 mJ Yb:CaF2 regenerative amplifier. The system delivers 1.24 mJ, sub-400 fs pulses at 1035 nm with excellent energy stability. Using a single CVBG as a matched stretcher/compressor device had proved to be unpractical for reaching sub-picosecond durations in CPA regenerative amplifiers with a high number of round-trips or including high dispersion media. In the present CPA system, using the matched CVBG-based stretcher and compressor alone the pulse duration was limited to 1.7 ps (4.9 × FTL). With the implementation of a Treacy grating pair at the stretcher, we have successfully improved the output pulse length by a factor of 4.4, achieving a value of ~390 fs (1.1 × FTL).
By using different gratings, with a larger groove density (e.g 900 lines/mm) it is possible to reduce the size of auxiliary compressor even further, down to a few centimeters, without compromising the simplicity of alignment. Additionally, this approach preserves the flexibility when compared e.g. to systems based in two different CVBGs, with different dispersion factors to stretch and compress the pulses. Future work includes testing CVBGs with larger bandwidths in order to avoid spectral clipping, thereby allowing full optimization of compression using an auxiliary dispersive line.
This work is partially supported by Fundação para a Ciência e a Tecnologia (grant SFRH/BD/68865/2010), Laserlab-Europe (EC’s FP7, grant agreement no. 284464), and Association EURATOM/IST.
References and links
1. D. Strickland and G. Mourou, “Compression of amplified chirped optical pulses,” Opt. Commun. 56(3), 219–221 (1985). [CrossRef]
2. E. B. Treacy, “Optical pulse compression with diffraction gratings,” IEEE J. Quantum Electron. 5(9), 454–458 (1969). [CrossRef]
3. O. E. Martinez, T. Prabhuram, M. C. Marconi, and J. J. Rocca, “Magnified expansion and compression of subpicosecond pulses from a frequency-doubled Nd:YLF laser,” IEEE J. Quantum Electron. 25(10), 2124–2128 (1989). [CrossRef]
4. T. Eidam, J. Rothhardt, F. Stutzki, F. Jansen, S. Hädrich, H. Carstens, C. Jauregui, J. Limpert, and A. Tünnermann, “Fiber chirped-pulse amplification system emitting 3.8 GW peak power,” Opt. Express 19(1), 255–260 (2011). [CrossRef] [PubMed]
5. J. Tümmler, R. Jung, H. Stiel, P. V. Nickles, and W. Sandner, “High-repetition-rate chirped-pulse-amplification thin-disk laser system with joule-level pulse energy,” Opt. Lett. 34(9), 1378–1380 (2009). [CrossRef] [PubMed]
6. G. H. Kim, J. Yang, S. A. Chizhov, E. G. Sall, A. V. Kulik, V. E. Yashin, D. S. Lee, and U. Kang, “High average-power ultrafast CPA Yb:KYW laser system with dual-slab amplifier,” Opt. Express 20(4), 3434–3442 (2012). [CrossRef] [PubMed]
7. A. Galvanauskas, M. E. Fermann, D. Harter, K. Sugden, and I. Bennion, “All-fiber femtosecond pulse amplification circuit using chirped Bragg gratings,” Appl. Phys. Lett. 66(9), 1053–1055 (1995). [CrossRef]
8. A. Galvanauskas, D. Harter, M. A. Arbore, M. H. Chou, and M. M. Fejer, “Chirped-pulse-amplification circuits for fiber amplifiers, based on chirped-period quasi-phase-matching gratings,” Opt. Lett. 23(21), 1695–1697 (1998). [CrossRef] [PubMed]
9. A. Galvanauskas, A. Heaney, T. Erdogan, and D. Harter, “Use of volume chirped Bragg gratings for compact high-energy chirped pulse amplification circuits,” in Conference on Lasers and Electro-Optics, Vol. 6, Technical Digest Series (Optical Society of America) (1998), pp. 362. [CrossRef]
10. O. V. Belai, E. V. Podivilov, and D. A. Shapiro, “Group delay in Bragg grating with linear chirp,” Opt. Commun. 266(2), 512–520 (2006). [CrossRef]
11. R. Kashyap, Fiber Bragg Gratings (Academic, 2010).
12. G. Chang, M. Rever, V. Smirnov, L. Glebov, and A. Galvanauskas, “Femtosecond Yb-fiber chirped-pulse-amplification system based on chirped-volume Bragg gratings,” Opt. Lett. 34(19), 2952–2954 (2009). [CrossRef] [PubMed]
13. L. Glebov, V. Smirnov, E. Rotari, I. Cohanoschi, L. Glebova, O. Smolski, J. Lumeau, C. Lantigua, and A. Glebov, “Volume-chirped Bragg gratings: monolithic components for stretching and compression of ultrashort laser pulses,” Opt. Eng. 53(5), 051514 (2014). [CrossRef]
14. E. Slobodtchikov and P. F. Moulton, “Compact femtosecond laser system with 2 mJ output,” in Conference on Lasers and Electro-Optics (CLEO) and Quantum Electronics and Laser Science Conference (QELS), Bedford, MA, USA, May 2010. [CrossRef]
15. K.-H. Hong, A. Siddiqui, J. Moses, J. Gopinath, J. Hybl, F. Ö. Ilday, T. Y. Fan, and F. X. Kärtner, “Generation of 287 W, 5.5 ps pulses at 78 MHz repetition rate from a cryogenically cooled Yb:YAG amplifier seeded by a fiber chirped-pulse amplification system,” Opt. Lett. 33(21), 2473–2475 (2008). [CrossRef] [PubMed]
16. S. Keppler, R. Bödefeld, M. Hornung, A. Sävert, J. Hein, and M. C. Kaluza, “Prepulse suppression in a multi-10-TW diode-pumped Yb:glass laser,” Appl. Phys. B 104(1), 11–16 (2011). [CrossRef]
17. C. P. João, J. Wemans, and G. Figueira, “Numerical simulation of high-energy, ytterbium-doped amplifier tunability,” Appl. Sci. 3(1), 288–298 (2013). [CrossRef]