We present an ultrafast fiber laser system delivering 4.6 W average power at 258 nm based on two-stage fourth-harmonic generation in beta barium borate (BBO). The beam quality is close to being diffraction limited with an value of . The pulse duration is 150 fs, which, potentially, is compressible down to 40 fs. A plain BBO and a sapphire-BBO compound are compared with respect to the achievable beam quality in the conversion process. This laser is applicable in scientific and industrial fields. Further scaling to higher average power is discussed.
© 2017 Optical Society of America
Today there is an increasing interest in ultrafast ultraviolet (UV) lasers with high beam quality for scientific and industrial applications. For example, precision machining benefits from the short wavelength, combined with the short pulse duration, as it enables smaller feature sizes and small heat affected zones . Moreover, the increased linear absorption simplifies the machining of large bandgap materials .
Scientific applications include UV-induced breakdown spectroscopy , ultrafast spectroscopy , and the generation of extreme ultraviolet (XUV) light via high-harmonic generation (HHG). For HHG, the short driving wavelength leads to a significant increase of the conversion efficiency compared to state-of-the-art infrared (IR) driving wavelength [5–8]. The ultrafast UV lasers required for this process are usually made by frequency doubling or tripling of femtosecond IR lasers in nonlinear crystals, e.g., beta barium borate (BBO). The crystals have to be very thin to allow a suitable phase-matching bandwidth. Following this approach, the second harmonic (SH, 515 nm) of a ytterbium-based ultrafast fiber laser has been used to generate coherent XUV light via HHG at an average power of 800 μW and at a photon energy of 21.7 eV with a green-to-XUV conversion efficiency of . A further increase in conversion efficiency and XUV power is desired by “photon hungry” applications such as nanoscale imaging [10,11], studies on molecular dynamics , and spectroscopy of the energy levels of highly charged ions .
This increase in photon flux will be possible by using even shorter driving wavelengths and even more input power. However, generating high-power ultrashort pulses with high beam quality in the UV is challenging, as absorption in the nonlinear crystals leads to beam quality degradation and even crystal fracture. The state-of-the-art in frequency-converted IR lasers is 2.6 W of average power with 35 fs pulse duration, but without a measurement of the beam quality  and 2.7 W average power with an value of 2.5 at 4.2 ps pulse duration . The frequency conversion can also be accomplished in gas-filled waveguides with fairly low conversion efficiency . Furthermore, there are KrF excimer  and Ce:LiCAF  lasers delivering femtosecond UV pulses with up to 50 W average power, but at very low repetition rates and unspecified beam quality.
Following the approach of frequency conversion, a 100 W average power third-harmonic femtosecond laser at 343 nm with diffraction-limited beam quality has been demonstrated recently . In this system, a sandwich structure consisting of a BBO bonded between highly thermoconductive sapphire plates is used to mitigate the detrimental thermal issue in the BBO.
In this Letter, we extend this work towards a shorter wavelength and present a nearly diffraction-limited high-power ultrafast laser at 258 nm based on the fourth harmonic (FH) of an Yb-doped ultrafast fiber laser system. We compare the power and the beam quality of the UV beam generated in a plain BBO and in a sapphire-BBO-sapphire sandwich structure. The challenge is to overcome one order of magnitude larger linear and almost two orders of magnitude larger nonlinear absorption in both BBO and sapphire at 258 nm compared to the existing 343 nm laser system [19,20].
The experimental setup for the two-stage fourth-harmonic generation (FHG) is depicted in Fig. 1. First, the output of an ultrafast fiber chirped-pulse amplification system  is directed through a half-wave plate and a thin-film polarizer to control the input power to the experiment, while the laser is operated at maximum output power. The beam is collimated to 1.2 mm -diameter by means of a Galilean telescope (lenses and ). The beam is sent through a 0.5 mm thick BBO crystal cut at 23.4° for type I phase-matched second-harmonic generation (SHG). The crystal is anti-reflective coated for both the fundamental and the SH and is mounted in a tip, tilt, and rotation mount to optimize phase-matching conditions. The two wavelengths are separated by means of two 45° dichroic mirrors, and the green light is used to generate the FH in another type-I phase-matched BBO cut at 50.0°. Two types of BBO crystals were used to compare their performance in the experiment. First, an uncoated 0.1 mm thick BBO was implemented. Then a sapphire-BBO-sapphire sandwich structure consisting of two 1 mm thick sapphire plates and a 0.1 mm thick BBO was used. The sandwich structure was fabricated via direct bonding, and the outer surfaces are anti-reflective coated for both the SH and FH. This compound crystal was held in a water-cooled mount with only tilt adjustment. Thus, a half-wave plate to rotate the polarization of the green driving pulse is used additionally to achieve phase-matching. After the FH is generated, it is separated from the remaining SH using two dichroic mirrors at a 45° angle of incidence. A sequence of two sapphire wedges at a Brewster angle is used to take an s-polarized sample of the UV beam to measure the pulse duration, the beam quality, and the spectrum. A second-order cross-correlation based on difference frequency generation of the FH with the fundamental wave is used to measure the pulse duration. Thus, a beam sample of the IR driving field is taken prior to the frequency conversion. An optical delay stage is used to match the propagation time of the IR and the UV pulse to the cross-correlator.
At the beginning of the experiment, the IR beam is characterized. The fiber laser is operated at 80 W average power and at a repetition rate of 796 kHz. The optical spectrum, as shown in Fig. 2, is centered at 1032 nm with a full width at half-maximum (FWHM) of 10.2 nm. The high-frequency ripples are due to a spatial light modulator in the fiber laser system, and the slow modulations are due to a spectral amplitude shaper. The corresponding second-order auto-correlation is shown in Fig. 3 and features a FWHM of 301 fs, which corresponds to a pulse duration of 213 fs assuming a Gaussian pulse. Furthermore, the value is measured (-method) and is found to be less than 1.1 on both axes of the beam, which is consistent with previously reported values .
Now the IR beam is directed into the frequency conversion stages. The crystal orientations of the SHG and FHG stages are optimized to generate the highest power in the UV, which is 7.5 W. The value of the SH is measured to be less than 1.1 on both axes at 40 W average power. Then, the measurement is done for the FH as well. The results for the plain BBO are shown in Fig. 4 for increasing UV power. The initial value is 1.4 on both axes and remains at this level up to approximately 4 W of UV. Then, the heating of the crystal due to linear and nonlinear UV absorption leads to a significant thermal lens, distorting the beam, such that at 7.5 W UV, the value increases up to 3.4.
The experiment is set to 4.6 W of UV, and both the UV output and the green driver (prior FHG) are analyzed in more detail. The optical spectra were measured with 0.1 nm resolution from the stray light of the power meters or beam dumps. The SH spectrum [Fig. 5(a)] has a FWHM of 2.6 nm, supporting a 130 fs pulse. This is a realistic result; as in SHG, the pulse duration reduces by a factor of for a Gaussian input pulse, leading to such an optical bandwidth. The FHG spectrum has an increased bandwidth with a FWHM of 3.1 nm and is depicted in Fig. 5(b). A possible reason for the bandwidth increase is nonlinear phase accumulation in the BBO crystals, leading to a small pulse chirp that can result in spectral broadening during the conversion.
Next, the cross-correlation of the UV pulse and the IR driver is measured, which is shown in Fig. 6. The FWHM duration of the UV pulse is 150 fs, as calculated from the 264 fs FWHM of the cross-correlation and from the initially determined 213 fs FWHM pulse duration of the IR driver. In the future, compression to the transform limit of the UV spectrum with 40 fs pulse duration might be feasible, e.g., by using a prism pair or chirped mirrors.
After these investigations, the plain BBO is replaced by the sapphire-BBO-sapphire compound in the FHG stage. The compound should offer an improved heat dissipation and allow us to further increase the UV average power without the loss of beam quality. The experimentally determined values, as depicted in Fig. 7, are disproving this assumption. The value at low average power already is in the range of 1.6 to 1.8 on both axes. At 3 W UV power, the value increases to almost 2. At even higher UV power, the beam quality degrades rapidly, such that the measurement yields no satisfactory fit to the beam caustic any more. Hence, in terms of beam quality, the sapphire-BBO-sapphire sandwich structure does not perform better than a plain piece of BBO. This contradicts the experience made in third-harmonic generation (THG)  and is investigated in the following.
First, the surface temperature of the sandwich structure is monitored using a thermographic camera, which is corrected for the emissivity of the sapphire plate ( ). The measurement reveals a temperature gradient of 3°C from the beam axis to the substrate’s edge at 4 W UV average power. Then, the sapphire-BBO compound is replaced with the plain BBO and, again, 4 W of UV are generated. A sapphire plate, which is identical to the compound’s heat spreaders, is placed prior and after the BBO crystal. The surface temperatures of both BBO and sapphire are monitored and corrected for their emissivity (, ). No temperature increase in the sapphire plate is observed when it is placed in the green beam before the FHG-BBO; hence, absorption of green light expectedly is not an issue. However, when the plate is placed in the UV beam, its temperature rises to 43°C on the beam axis and decreases to 39°C on the substrate edge—the same gradient as observed on the sandwich structure. In this setting, the value of the UV beam increases by 0.1–0.2 on both axes due to the thermal lens in the sapphire plate. The origin of this thermal load is both linear and two-photon absorption (TPA). The TPA absorption is , which yields a TPA absorption coefficient of at an estimated peak intensity of for 1 mm beam diameter. This contribution is about equal to the linear absorption coefficient of at 258 nm . Thus, the heat input in the sapphire heat spreaders is significantly increased compared to the cases of SHG and THG and explains the observed temperature gradient.
At the same time, the surface temperature of the plain BBO is 165°C on axis and 45°C at the substrate edge. The large temperature gradient observed is due to the almost one order of magnitude stronger linear  and nonlinear absorption (UV TPA:  and UV-green absorption ), plus the approximately one order of magnitude lower thermal conductivity compared to sapphire. Thus, the main contribution to the beam quality degradation at high UV power still is the BBO itself. The sapphire plates will still improve on the heat dissipation of the BBO and reduce its thermal lensing, but this effect does not overcome the additional thermal lens in the sapphire.
The significant increase of both linear and nonlinear absorption impedes power scaling using the sandwich structure compared to the previous SHG and THG experiments, where the compound outperformed the plain crystal [18,23]. However, the concept is still viable in FHG for nanosecond-pulsed and continuous-wave lasers, as the nonlinear absorption is much weaker in that case. Then, crystal stacks with several layers have to be manufactured to achieve sufficient conversion length, which is not possible in the femtosecond regime due to temporal walk-off and dispersion.
In summary, we presented a high-power ultrafast UV laser at 258 nm based on two-stage FH generation of a 1 μm ultrafast fiber laser. The achieved UV output parameters are 4.6 W average power at 258 nm with 150 fs pulse duration. The beam quality is close to being diffraction limited with . These performance figures overcome the state-of-the-art average power by a factor of 2 . We compared a plain BBO to a sapphire-BBO-sapphire sandwich structure for the UV generation. The advantage of the sapphire heat spreaders as in THG is not given for FH generation, due to significantly increased linear and nonlinear absorption of the generated UV in the sapphire heat-spreader plates.
Future work will focus on increasing the UV average power by using caesium lithium borate (CLBO) or periodically poled lithium triborate (LBO) crystals, which have a higher UV absorption edge. Moreover, spatially inhomogeneous cooling will be applied , and the compression of the UV pulses will be pursued, enabling 10 W-class femtosecond UV lasers with diffraction-limited beam quality for a broad range of industrial and scientific applications.
Bundesministerium für Bildung und Forschung (BMBF) (05P15SJFFA); H2020 European Research Council (ERC) (670557 “MIMAS”).
The authors acknowledge funding by the German Ministry of Education and Research (Bundesministerium für Bildung und Forschung, BMBF). M. Müller acknowledges financial support by the Carl-Zeiss-Stiftung. This work has been partly supported by the European Research Council (ERC) grant 670557 “MIMAS.”
1. J. H. Klein-Wiele, J. Bekesi, and P. Simon, Appl. Phys. A 79, 775 (2004). [CrossRef]
2. A. Saliminia, A. Proulx, and R. Vallée, Opt. Commun. 333, 133 (2014). [CrossRef]
3. S. P. Banerjee, Z. Chen, and R. Fedosejevs, Opt. Lasers Eng. 68, 1 (2015). [CrossRef]
4. T. Kobayashi, A. Yabushita, and Y. Kida, Photonics 3, 64 (2016). [CrossRef]
5. A. D. Shiner, C. Trallero-Herrero, N. Kajumba, H. C. Bandulet, D. Comtois, F. Légaré, M. Giguère, J. C. Kieffer, P. B. Corkum, and D. M. Villeneuve, Phys. Rev. Lett. 103, 1 (2009). [CrossRef]
6. D. Popmintchev, C. Hernández-García, F. Dollar, C. Mancuso, J. A. Pérez-Hernández, M.-C. Chen, A. Hankla, X. Gao, B. Shim, A. L. Gaeta, M. Tarazkar, D. A. Romanov, R. J. Levis, J. A. Gaffney, M. Foord, S. B. Libby, A. Jaron-Becker, A. Becker, L. Plaja, M. M. Murnane, H. C. Kapteyn, and T. Popmintchev, Science 350, 1225 (2015). [CrossRef]
7. H. Wang, Y. Xu, S. Ulonska, J. S. Robinson, P. Ranitovic, and R. A. Kaindl, Nat. Commun. 6, 7459 (2015). [CrossRef]
8. D. Popmintchev, M.-C. Chen, C. H. García, J. A. P. Hernández, J. P. Siqueira, S. Brown, F. Dollar, B. C. Walker, P. Grychtol, L. Plaja, M. M. Murnane, H. Kapteyn, and T. Popmintchev, Conference on Lasers and Electro-Optics (CLEO): 2013 (Optical Society of America, 2013), paper QW1A.5.
9. R. Klas, S. Demmler, M. Tschernajew, S. Hädrich, Y. Shamir, A. Tünnermann, J. Rothhardt, and J. Limpert, Optica 3, 1167 (2016). [CrossRef]
10. G. K. Tadesse, R. Klas, S. Demmler, S. Hädrich, I. Wahyutama, M. Steinert, C. Spielmann, M. Zürch, T. Pertsch, A. Tünnermann, J. Limpert, and J. Rothhardt, Opt. Lett. 41, 5170 (2016). [CrossRef]
11. A. Ravasio, D. Gauthier, F. R. N. C. Maia, M. Billon, J. P. Caumes, D. Garzella, M. Géléoc, O. Gobert, J. F. Hergott, A. M. Pena, H. Perez, B. Carré, E. Bourhis, J. Gierak, A. Madouri, D. Mailly, B. Schiedt, M. Fajardo, J. Gautier, P. Zeitoun, P. H. Bucksbaum, J. Hajdu, and H. Merdji, Phys. Rev. Lett. 103, 1 (2009). [CrossRef]
12. D. Davydova, A. De La Cadena, S. Demmler, J. Rothhardt, J. Limpert, T. Pascher, D. Akimov, and B. Dietzek, Chem. Phys. 464, 69 (2016). [CrossRef]
13. J. Rothhardt, S. Hädrich, S. Demmler, M. Krebs, D. F. A. Winters, T. Kühl, T. Stöhlker, J. Limpert, and A. Tünnermann, Phys. Scripta T166, 14030 (2015). [CrossRef]
14. C. Chang, P. Krogen, H. Liang, G. J. Stein, J. Moses, C. Lai, J. P. Siqueira, L. E. Zapata, F. X. Kärtner, and K. Hong, Opt. Lett. 40, 665 (2015). [CrossRef]
15. L. Misoguti, S. Backus, C. G. Durfee, R. Bartels, M. M. Murnane, and H. C. Kapteyn, Phys. Rev. Lett. 87, 013601 (2001). [CrossRef]
16. Y. Nabekawa, D. Yoshitomi, T. Sekikawa, and S. Watanabe, IEEE J. Sel. Top. Quantum Electron. 7, 551 (2001). [CrossRef]
17. Z. Liu, T. Kozeki, Y. Suzuki, N. Sarukura, K. Shimamura, T. Fukuda, M. Hirano, and H. Hosono, Opt. Lett. 26, 301 (2001). [CrossRef]
18. J. Rothhardt, C. Rothhardt, M. Müller, A. Klenke, M. Kienel, S. Demmler, T. Elsman, M. Rothhardt, J. Limpert, and A. Tünnermann, Opt. Lett. 41, 1885 (2016). [CrossRef]
19. A. Dragonmir, J. G. McInerney, and D. N. Nikogosyan, Appl. Opt. 41, 4365 (2002). [CrossRef]
20. R. DeSalvo, A. A. Said, D. J. Hagan, E. W. Van Stryland, and M. Sheik-Bahae, IEEE J. Quantum Electron. 32, 1324 (1996). [CrossRef]
21. M. Müller, M. Kienel, A. Klenke, T. Gottschall, E. Shestaev, M. Plötner, J. Limpert, and A. Tünnermann, Opt. Lett. 41, 3439 (2016). [CrossRef]
22. S. G. Kaplan and L. M. Hanssen, Proc. SPIE 3425, 120 (1998). [CrossRef]
23. C. Rothhardt, J. Rothhardt, A. Klenke, T. Peschel, R. Eberhardt, J. Limpert, and A. Tünnermann, Opt. Mater. Express 4, 1092 (2014). [CrossRef]
24. N. A. Kulagin and L. A. Litvinov, Cryst. Res. Technol. 20, 1667 (1985). [CrossRef]
25. J. Liebertz and S. Stähr, Z. Krist. 165, 91 (1983). [CrossRef]
26. S. Wu, G. A. Blake, S. Sun, and H. Yu, Proc. SPIE 3928, 221 (2000). [CrossRef]
27. Y. K. Yap, K. Deki, N. Kitatochi, Y. Mori, and T. Sasaki, Opt. Lett. 23, 1016 (1998). [CrossRef]