Open Access
High-Power, Few-Cycle, Angular Dispersion Compensated Mid-Infrared Pulses From a Noncollinear Optical Parametric Amplifier Volume 9, Number 3, June 2017 Mark Mero Valentin Petrov
DOI: 10.1109/JPHOT.2017.2697498 1943-0655 © 2017 IEEE
IEEE Photonics Journal
High-Power, Few-Cycle, Angular Dispersion Compensated
High-Power, Few-Cycle, Angular Dispersion Compensated Mid-Infrared Pulses From a Noncollinear Optical Parametric Amplifier Mark Mero and Valentin Petrov Max Born Institute for Nonlinear and Short Pulse Spectroscopy, Berlin 12489, Germany DOI:10.1109/JPHOT.2017.2697498 C 2017 IEEE. Translations and content mining are permitted for academic research only. 1943-0655 Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Manuscript received February 23, 2017; revised April 17, 2017; accepted April 19, 2017. Date of publication April 26, 2017; date of current version May 5, 2017. Corresponding author: Mark Mero (e-mail:
[email protected]).
Abstract: We demonstrate a scheme for correcting the angular dispersion of the 3.1-μm idler beam from a high repetition rate, KTiOAsO4 (KTA) based noncollinear optical parametric amplifier pumped at 1.03 μm and seeded by broadband signal pulses at 1.55 μm. Following angular dispersion correction, successful temporal chirp compensation was possible, and a close to transform-limited pulse duration of 70 fs and a pulse energy of 30 μJ at a repetition rate of 100 kHz were obtained. The angular chirp compensated KTA-based scheme is scalable to higher pulse energies and average powers and has various advantages compared to periodically poled and angle-tuned LiNbO3 -based amplifiers, which exhibit low resistance to photorefractive damage. Index Terms: Mid-wave infrared (MWIR) devices, ultrafast technology.
1. Introduction Strong-field physics and attoscience can tremendously benefit from the development of few-cycle mid-infrared (MIR, 3-8 μm) sources providing high peak and average power simultaneously. Operating at the blue edge of the MIR (i.e., 3-4 μm) provides a good trade-off between shorter soft-X-ray wavelengths in high-order harmonic generation (HHG) and the unfavorable scaling of the single atom response in HHG at increasing driver laser wavelength. A proven technique to reach the required repetition rates (i.e., 10 kHz) and pulse energies (i.e., ≥ 10’s of μJ) at the blue edge of the MIR is optical parametric chirped pulse amplification (OPCPA) driven by picosecond, diodepumped Nd- or Yb-doped pump lasers [1]–[6]. Using Yb-doped materials (bulk or fiber) in the pump laser/amplifier chain enables shorter pulse durations around 1.03 μm, down to the sub-ps regime, which is advantageous for the parametric amplification process [2], [4]–[6]. Yb-laser systems are gaining wide popularity due to their lower cost, higher robustness, and greater average power scalability compared to the more traditional Ti:sapphire pump lasers. Typically, periodically poled lithium niobate (PPLN) is employed as the amplifying medium in the target 3-4 μm spectral range due to its high nonlinear optical coefficient, mature growth technology, and wide availability both in single-grating and in fan-out or chirped-grating form up to an aperture size of a few millimeters [7]. However, lithium niobate is known to exhibit photochromic and photorefractive damage [8]–[12], limiting its use in optical frequency conversion devices, where
Vol. 9, No. 3, June 2017
3200408
IEEE Photonics Journal
High-Power, Few-Cycle, Angular Dispersion Compensated
Fig. 1. General scheme for correcting the angular dispersion of the output of a parametric amplifier.
simultaneously high average and peak power is the requirement [4], [12]. While the development of alternative materials allowing periodic poling with homogenous domain structure at large apertures is ongoing [13], bulk (i.e., non-poled, angle-tuned) crystals exploiting noncollinear amplifying geometry can often provide the large gain bandwidths able to support few-cycle pulse durations at a sufficiently high gain. Unfortunately, the selection of bulk nonlinear materials at the blue edge of the MIR with clear transparency is small in the above demanding pulse parameter range. Although materials such as LiNbO3 , KNbO3 , LiIO3 , and KTiOPO4 are also widely available, the requirement of resistance to both high peak and average intensity is limiting the choice to KTiOAsO4 (KTA). The combination of a sufficiently high nonlinear coefficient, large band gap energy, high transmission up to 3.5 μm, reduced OH absorption at 2.8 μm, and high resistance to laser-induced damage, photodarkening and photorefractive effects makes KTA a superior material for low-gain booster amplifier stages [2]. The shortcoming of KTA when pumped near 1 μm is that broadband amplification in the traditional noncollinear arrangement, where the signal is angular dispersion free, is prohibited for MIR seeding. Here, as usual, the signal beam is by definition identical to the amplified seed beam. Similar limitations with respect to the lack of possibility of broadband noncollinear phase matching exist for LiNbO3 , KNbO3 , and LiIO3 , when seeded below the degenerate wavelength at 2 μm [14], [15]. Regarding KTA, one needs to choose either i) collinear geometry using thin crystals or (ii) noncollinear geometry using thicker crystals, where seeding is confined to the short-wave infrared (SWIR) at 1.4-1.7 μm. Case i) makes it possible to obtain both the signal and the idler beam without angular dispersion at the cost of a reduced interaction length, which leads to high required pump intensities potentially exceeding the limits set by higher order nonlinear effects or optical damage. The increased crystal thickness in Case ii) allows a reduction in pump intensity at the cost of generating an angularly dispersed MIR idler beam. Considering that the scaling of parametric gain is stronger with crystal thickness than with pump intensity [16], it is worth exploring the feasibility of angular dispersion (AD) compensation schemes at high MIR average powers [i.e., Case ii)]. The general layout for AD compensation is shown in Fig. 1. An optical system with a suitably chosen magnification images the plane of the nonlinear crystal onto an angularly dispersive optical element or elements. If the imaging system is properly designed, the angular dispersion of the beam can be matched in magnitude, but with opposite sign, to that of the angularly dispersive optical system and a collimated beam without AD and spatial chirp is obtained. Imaging also makes sure that propagating the angularly dispersed beam does not lead to spatial or temporal chirp. Various schemes have been proposed to compensate the angular dispersion of idler beams in noncollinear parametric amplifiers with an angular-dispersion-free signal beam, where the idler pulse energy, average power, and the red edge of its spectrum were limited to 3 μJ, 3 mW, and 1.6 μm, respectively [17]–[19]. The demonstrated arrangements were based on reflection gratings combined with reflective-type telescopes. A crucial aspect of the setup is its overall throughput mainly limited by the angularly dispersive element. In this paper, we demonstrate an AD correction scheme for the 3.1-μm idler output of the noncollinear booster KTA amplifier stage of the few-cycle OPCPA front-end described in [2]. The angular-chirp-corrected MIR pulses could be successfully compressed close to their few-cycle transform limited duration, at an output pulse energy and average power of up to 30 μJ and 3 W,
Vol. 9, No. 3, June 2017
3200408
IEEE Photonics Journal
High-Power, Few-Cycle, Angular Dispersion Compensated
Fig. 2. Scheme of the dual-beam OPCPA front-end. S: bulk stretcher, C: bulk compressor, ACC: angular chirp compensation, CFBG/circ.: chirped fiber Bragg grating stretcher with circulator, HNLF: highly nonlinear fiber.
respectively. The resulting dual-beam system provides an unprecedented combination of optically synchronized few-cycle pulses at 1.55 and 3.1 μm, which can already be considered a stand-alone system capable of driving a wide range of strong-field experiments. The AD compensation scheme offers a promising, alternative way for further scaling the average power of few-cycle MIR pulses that leads to a reduced pump load on the amplifier crystal compared to schemes based on the collinear amplifier geometry.
2. Experimental Setup and Results The scheme of the dual-beam OPCPA front-end is shown in Fig. 2. With the exception of the AD correction part, the source was described in detail in [2]. Here, only a short summary is given. The optically synchronized seed pulses for the OPCPA signal beamline at 1.55 μm and the 1.03-μm pump amplifier system are provided by a commercial multi-branch Er-fiber laser system. The pump amplifier chain relies on the chirped pulse amplification (CPA) concept and is based on Yb-fiber pre-amplifiers and Yb:YAG booster amplifiers with an output pulse duration of ∼1 ps. Compared to the conditions in [2], the OPCPA pump pulse energies and intensities have been changed to keep the pump-to-signal energy conversion efficiency only slightly above 15% to avoid the deterioration of the Strehl-ratio of the output of the noncollinear second stage predicted by numerical simulations [20]. At the same time, the OPCPA output pulse energies were kept the same as in [2]. The total compressed pulse energy available from the Yb-amplifier system for the present experiment was 2.25 mJ. The OPCPA consists of a MgO-doped PPLN based highgain first stage and a noncollinear, KTA-based booster amplifier stage. Both stages are seeded at 1.55 μm. Due to the noncollinear geometry of the second stage, the generated 3.1-μm idler beam is angularly dispersed and propagates in a direction different from the direction of the signal beam. The measured pulse energy at 3.1 μm after the KTA crystal is 45 μJ (i.e., 4.5 W). Using the Sellmeier equation of KTA, the predicted phase-matching angle and noncollinear angle inside the crystal for broadband, type II interaction with the 1.55-μm beam polarized in the XZ plane are θ = 50◦ and α = 3.8◦ , respectively. Momentum conservation in the three-wave parametric process leads to a constant 69 μrad/nm of angular dispersion inside the crystal in the relevant 2.9-3.35-μm spectral range of the idler. This corresponds to ∼120 μrad/nm in air and an angular spread of 3.1◦ in the full 450-nm range. The variation of angular dispersion in air as a function of wavelength (due to the wavelength dependent index of refraction of KTA) is below ±1 μrad/nm in the full idler spectral range, and is therefore neglected. The angularly dispersive element and the magnification of the imaging system have to be chosen to simultaneously allow angular dispersion correction, resistance to high average power, and high throughput. In general, diffraction grating based elements are preferred over prism based elements, the latter of which suffer from astigmatism and may induce significant laterally varying group delay dispersion (GDD) across the beam. A suitably chosen magnification factor M > 1 reduces i) the
Vol. 9, No. 3, June 2017
3200408
IEEE Photonics Journal
High-Power, Few-Cycle, Angular Dispersion Compensated
Fig. 3. Angular dispersion as a function of angle of incidence and wavelength induced by a silicon prism with an apex angle of 32.08◦ . The dashed line corresponds to our experimentally implemented condition.
intensity and peak thermal load on the dispersive element to acceptable levels and ii) the angular dispersion and angular spread of the idler beam incident on the dispersive element. Due to its availability in our lab and the low expected susceptibility to thermal issues, we employed a silicon prism for our feasibility study. The prism had an apex angle of 32.08◦ and was anti-reflection (AR) coated for the 3-5-μm range at normal incidence. Fig. 3 shows the induced angular dispersion as a function of angle of incidence on the prism and wavelength. A magnification M = 3.9 and an angle of incidence of 65.9◦ for the center ray at 3.07 μm were chosen, represented by the dashed line in Fig. 3. An optimal solution along one of the contour lines was not possible. The imaging setup consisted of an f = 100-mm spherical reflector placed ∼125 mm away from the KTA crystal. The calculated angular dispersion in the incident beam was 31 μrad/nm, while the induced angular dispersion as a function of wavelength was estimated to be in the range of (−24.5)-(−34.5) μrad/nm resulting in a residual angular dispersion in the corrected beam of (−3.5)−6.5 μrad/nm right after the prism. An upcollimating telescope with a magnification of ×2 reduced the residual angular dispersion by another factor of 2. We note that the angular dispersion and the corresponding pulse front tilt introduced by the prism are intimately related and the removal of angular dispersion is inseparable from the removal of the pulse front tilt provided that no further aberrations are introduced during the procedure. We also note that low line density reflection-type diffraction gratings can provide a superior performance to prisms provided that the thermal load is handled properly. Such a grating was not available during the time of this study. Advantages of such reflection gratings include i) a much weaker scaling of angular dispersion as a function of wavelength and angle of incidence than in the case of prisms, ii) a lack of added astigmatism, and iii) a lack of a spatially varying GDD across the beam. Fig. 4 shows the near-field beam profile of the AD-corrected idler beam measured along a ∼1-m path after the telescope which was meters behind the KTA crystal. The small residual astigmatism can be improved by appropriate optics (e.g., focusing optics tilted at an appropriate angle about an appropriate rotation axis) in a straightforward way, if needed. The large, 80-μm pixel size of our MIR camera prevented an M 2 measurement so far. After angular dispersion had been corrected, temporal compression of the idler beam became possible. As explained in [2], the GDD of the generated idler pulses was designed to be negative. This allowed chirp compensation using off-the-shelf AR-coated Si windows. The optimum Si thickness was 10 mm. Successful compression was achieved to near the transform limited duration at a pulse energy of up to 30 μJ. The overall throughput of the angular chirp compensator and the compressor was 66%, limited by the AR-coating of the prism. A typical SHG-FROG trace and retrieval results are shown in Fig. 5 and 6, respectively. We note that for complex pulses, such as the pulse
Vol. 9, No. 3, June 2017
3200408
IEEE Photonics Journal
High-Power, Few-Cycle, Angular Dispersion Compensated
Fig. 4. Beam profiles of the angular-chirp-corrected 3.1-μm beam measured 7 cm (a), 69 cm (b), and 98 cm (c) behind a telescope, which was 2 m behind the KTA crystal. The camera chip size was 12.8 ×12.8 mm, and the ISO 4σ diameter of the beam in panel (b) is 2.9 and 3.9 mm in the horizontal and vertical plane, respectively.
Fig. 5. (a) Measured SHG-FROG trace and (b) retrieved SHG-FROG trace for the compressed 3.1-μm pulses at the 30-μJ level.
Fig. 6. (a) Retrieved temporal intensity and phase and (b) retrieved spectral intensity and phase for the compressed 3.1-μm pulses at the 30-μJ level (corresponding to the trace shown in Fig. 5). The retrieved pulse duration at FWHM is 70 fs. The dashed line in (b) shows the measured spectrum. The difference between the measured and reconstructed spectrum is attributed to the low resolution of the MIR spectrometer.
Vol. 9, No. 3, June 2017
3200408
IEEE Photonics Journal
High-Power, Few-Cycle, Angular Dispersion Compensated TABLE 1
High Peak and Average Power OPCPA Systems Operating Near 3 μm
Pulse
Pulse
Repetition
Average/Peak
MIR
Compressed SWIR
duration
energy
rate
power
wavelength
beam available
Ref.
[fs]
[μJ]
[kHz]
[W/GW]
[μm]
44
22
50
1.10/0.50
3.4
-
[1]
55
20
160
3.20/0.37
3.1
+
[3]
72
10
125
1.25/0.14
3.1
+
[4]
38
40
100
4.00/1.05
max. energy at 3.1
-
[5]
85
8
100
0.80/0.09
3.0
-
[6]
70
30
100
3.00/0.22∗
3.1
+
this work
∗
The actual temporal pedestal was taken into account in the peak power calculation.
Fig. 7. (a) Measured SHG-FROG trace and (b) reconstructed SHG-FROG trace for the compressed 1.55-μm signal pulses at the 90-μJ level.
shown in Fig. 6, the temporal FWHM value, which is the classical way to characterize a pulse can be a misleading measure of how close the pulse is to the transform limited case. Even though the pulse duration was very close to the transform limit, the root-mean-square time-bandwidth product was not. The fact that the measured FROG traces are highly symmetrical confirms the lack of significant spatial chirp in the corrected MIR pulses. The recompressed pulse duration is typically in the 69-72-fs range. Even though the pulse duration at FWHM is only 70 fs, there is a significant, several-100-fs-long pedestal similar to the case of the 1.55-μm signal beam line. As a comparison, Table 1 summarizes the essential parameters of existing state-of-the-art sources near 3 μm. The peak power value listed for the present work takes into account the temporal pedestal (i.e., 50% of energy is contained in main pulse), although the cited papers often lack a long enough measurement range to confirm the lack of a pedestal. Figs. 7 and 8 show the SHG-FROG traces and reconstructed temporal and spectral properties of the chirp compensated output at 1.55 μm. The origin of the large temporal pedestal was found to be the PPLN-based first stage possibly due to an interplay of parasitic frequency conversion processes with the strongly structured spectrum and spectral phase of the seed pulses, which were provided by the nonlinear-fiber-broadened Er-fiber laser. The long, complex temporal tail did not
Vol. 9, No. 3, June 2017
3200408
IEEE Photonics Journal
High-Power, Few-Cycle, Angular Dispersion Compensated
Fig. 8. (a) Retrieved temporal intensity and phase and (b) retrieved spectral intensity and phase for the compressed 1.55-μm signal pulses at the 90-μJ level (corresponding to the trace shown in Fig. 7). The retrieved pulse duration (FWHM) is 52 fs. (Inset in (a)) Gaussian spatial beam profile with w 1/e2 (horizontal) = 2.1 mm and w 1/e2 (vertical) = 1.8 mm. The dashed line in (b) shows the measured spectrum.
Fig. 9. (a) Average power stability of the compressed idler beam. (b) Blue edge of the raw spectrum of a multi-octave supercontinuum generated in sapphire by compressed 3.1-μm pulses, recorded using a CCD spectrometer (i.e., wavelength dependent sensitivity was not corrected for).
deteriorate further by amplification in the KTA booster stage. The removal of the temporal pedestal was not possible even using the 640-pixel spatial light modulator based pulse shaper installed in the signal chain before the OPCPA stages. The average power of the compressed idler pulses in a 1-hour time window is shown in Fig. 9(a) exhibiting RMS fluctuations at the 0.5% level. The excellent power stability is due to an active timing jitter (i.e., between seed and pump pulses) compensation setup based on a high-speed, piezobased, servo-controlled translation stage, which turned out to be crucial for keeping the OPCPA output stable. As a demonstration of the utility of the compressed idler beam, a fraction of the beam was focused into a 5-mm-thick sapphire window and a multi-octave supercontinuum was generated, which extended from 450 nm to the MIR. The blue edge of the supercontinuum was recorded by a CCD spectrometer [cf. Fig. 9(b)]. Similar continua produced by an existing 3-μm OPCPA system were reported in [21].
3. Conclusion We demonstrated a technique to overcome the limitation of KTA in 1-μm-pumped noncollinear optical parametric amplifiers, when generating or amplifying few-cycle pulses above the degeneracy wavelength of 2 μm and, in particular, in the MIR near 3 μm and above. The method relies on angular
Vol. 9, No. 3, June 2017
3200408
IEEE Photonics Journal
High-Power, Few-Cycle, Angular Dispersion Compensated
dispersion compensation of the idler beam from a traditional noncollinear parametric amplifier, where the 1.4-1.7-μm signal beam is free from angular dispersion. The scheme provides a way to lower the average and peak optical intensity on the amplifying medium and, thereby, may offer a possibility for scaling the average and peak MIR power beyond what is possible using collinear KTA and traditional PPLN-based amplifiers. At the same time, two optically synchronized few-cycle beams are obtained at distinct wavelengths, despite the noncollinear amplifier geometry, which leads to a more optimal use of the available pump power.
References [1] B. W. Mayer, C. R. Phillips, L. Gallmann, and U. Keller, “Mid-infrared pulse generation via achromatic quasi-phasematched OPCPA,” Opt. Exp., vol. 22, no. 17, pp. 20798–20808, 2014. [2] M. Mero, F. Noack, F. Bach, V. Petrov, and M. J. J. Vrakking, “High- average-power, 50-fs parametric amplifier front-end at 1.55 μm,” Opt. Exp., vol. 23, no. 26, pp. 33157–33163, 2015. [3] M. Baudisch, B. Wolter, M. Pullen, M. Hemmer, and J. Biegert, “High power multi-color OPCPA source with simultanous femtosecond deep-UV to mid-IR outputs,” Opt. Lett., vol. 41, no. 15, pp. 3583–3586, 2016. [4] P. Rigaud et al., “Supercontinuum seeded few-cycle mid-infrared OPCPA system,” Opt. Exp., vol. 24, no. 23, pp. 26494– 26502, 2016. [5] N. Thire´ et al., “4-W, 100-kHz, few-cycle mid-infrared source with sub-100-mrad carrier-envelope phase noise,” Opt. Exp., vol. 25, no. 2, pp. 1505–1514, 2017. [6] G. M. Archipovaite, S. Petit, J.-C. Delagnes, and E. Cormier, “100kHz Yb-fiber laser pumped 3μm optical parametric amplifier for probing solid-state systems in the strong field regime,” Opt. Lett., vol. 42, no. 5, pp. 891–894, 2017. [7] H. Ishizuki and T. Taira, “Half-joule output optical-parametric oscillation by using 10-mm-thick periodically poled Mgdoped congruent LiNbO3 ,” Opt. Exp., vol. 20, no. 18, pp. 20002–20010, 2012. [8] S. Sasamoto, J. Hirohashi, and S. Ashihara, “Polaron dynamics in lithium niobate upon femtosecond pulse irradiation: Influence of magnesium doping and stoichiometry control,” J. Appl. Phys., vol. 105, no. 8, 2009, Art. no. 083102. [9] S. Kato, S. Kurimura, H. H. Lim, and N. Mio, “Induced heating by nonlinear absorption in LiNbO3 -type crystals under continuous-wave laser radiation,” Opt. Mater., vol. 40, pp. 10–13, 2015. [10] A. Ashkin et al., “Optically-induced refractive index inhomogeneities in LiNbO3 and LiTaO3 ,” Appl. Phys. Lett., vol. 9, no. 1, pp. 72–74, 1966. [11] J. R. Schwesyg, M. Falk, C. R. Phillips, D. H. Jundt, K. Buse, and M. M. Fejer, “Pyroelectrically induced photorefractive damage in magnesium-doped lithium niobate crystals,” J. Opt. Soc. Amer. B, vol. 28, no. 8, pp. 1973–1987, 2011. [12] M. Baudisch, M. Hemmer, H. Pires, and J. Biegert, “Performance of MgO: PPLN, KTA, and KNbO3 for mid-wave infrared broadband parametric amplification at high average power,” Opt. Lett., vol. 39, no. 20, pp. 5802–5805, 2014. [13] V. Pasiskevicius, G. Strmqvist, F. Laurell, and C. Canalias, “Quasi-phase matched nonlinear media: Progress towards nonlinear optical engineering,” Opt. Mater., vol. 34, no. 3, pp. 513–523, 2012. [14] S.-W. Huang, J. Moses, and F. X. Kartner, “Broadband noncollinear optical parametric amplification without angularly ¨ dispersed idler,” Opt. Lett., vol. 37, no. 14, pp. 2796–2798, 2012. [15] G. Toth, L. Palfalvi, L. Tokodi, J. Hebling, and J. A. Ful ´ ´ ¨ op, ¨ “Scalable broadband OPCPA in lithium niobate with signal angular dispersion,” Opt. Commun., vol. 370, pp. 250–255, 2016. [16] G. Cerullo and S. De Silvestri, “Ultrafast optical parametric amplifiers,” Rev. Sci. Instrum., vol. 74, no. 1, pp. 1–18, 2003. [17] A. Shirakawa, I. Sakane, and T. Kobayashi, “Pulse-front-matched optical parametric amplification for sub-10-fs pulse generation tunable in the visible and near infrared,” Opt. Lett., vol. 23, no. 16, pp. 1292–1294, 1998. [18] K. Okamura and T. Kobayashi, “Octave-spanning carrier-envelope phase stabilized visible pulse with sub-3-fs pulse duration,” Opt. Lett., vol. 36, no. 2, pp. 226–228, 2011. [19] T.-J. Wang, Z. Major, I. Ahmad, S. Trushin, F. Krausz, and S. Karsch, “Ultrabroadband near-infrared pulse generation by noncollinear OPA with angular dispersion compensation,” Appl. Phys. B, vol. 121, pp. 229–233, 2015. [20] A. Giree, M. Mero, G. Arisholm, M. J. J. Vrakking, and F. J. Furch, “Numerical study of spatiotemporal distortions in noncollinear optical parametric chirped-pulse amplifiers,” Opt. Exp., vol. 25, no. 4, pp. 3104–3121, 2017. [21] F. Silva et al., “Multi-octave supercontinuum generation from mid-infrared filamentation in a bulk crystal,” Nature Commun., vol. 3, 2012, Art. no. 807.
Vol. 9, No. 3, June 2017
3200408
