Coherent ultraviolet light has many uses, for example, in the study of molecular species relevant in biology and chemistry. Very few, if any, laser materials offer ultraviolet transparency along with damage-free operation at high-photon energies and laser power. Here we report efficient generation of narrowband deep and vacuum ultraviolet light using hydrogen-filled hollow-core photonic crystal fiber. Pumping above the stimulated Raman threshold at 532 nm, coherent molecular vibrations are excited in the gas, permitting thresholdless wavelength conversion in the ultraviolet with efficiencies close to 60%. The system is uniquely pressure tunable, allows spatial structuring of the out-coupled radiation, and shows excellent performance in the vacuum ultraviolet. As the underlying scattering process is effectively linear, our approach can also in principle operate at the single-photon level, when all other alternatives are extremely inefficient.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
The majority of molecular species relevant in biology, photochemistry, and medicine, such as proteins and their building blocks the amino-acids , have narrowband outer-shell electronic transitions in the vacuum (VUV) and deep ultraviolet (DUV), i.e., from to . Spectroscopy at these wavelengths requires tunable, compact, and spectrally narrow UV light sources. Excimer lasers provide direct UV lasing transitions but are fixed wavelength and inefficient and deliver poor beam quality. Although sum-frequency generation in nonlinear crystals provides a common alternative , to be efficient and tunable, it requires stringent phase matching over a broad range of wavelengths (almost impossible to realize in collinear geometries) along with high pump intensities and good spatial overlap between the interacting fields. Moreover, the wavelength tunability of such systems is in general restricted because very few, if any, nonlinear crystals provide low dispersion, high transparency, and resistance to photo-induced damage in the DUV–VUV. Although these issues have to some degree been addressed by exploiting nonlinear interactions in gas cells [3–6], these techniques often require the use of ultrashort laser pulses with high peak intensities to generate broadband spectrally uniform UV light and to compensate for the relatively short interaction lengths. Although using gas-filled wide-bore capillaries relaxes these requirements , this requires low gas pressures to achieve phase matching in parametric interactions. Furthermore, to the best of our knowledge, most approaches reported in the literature are not suitable for generation or efficient frequency conversion of spectrally narrow DUV–VUV signals with high spectral power density.
Gas-filled hollow-core photonic crystal fiber (HC-PCF), that guides light by anti-resonant reflection, has emerged as a promising alternative that is free from these restrictions. In addition to providing guidance from the VUV to the mid-infrared [8,9], these fibers offer ultralong light–matter interaction lengths in a hollow channel only few tens of micrometers wide, together with pressure-tunable dispersion . This reduces the threshold for stimulated Raman scattering (SRS) by orders of magnitude , paving the way for multi-octave Raman combs  reaching into the VUV , and broadband frequency conversion in the near infrared .
In this Letter, we report efficient thresholdless frequency conversion of arbitrary, narrowband DUV signals in hydrogen [12–14], which has the highest Raman gain and frequency shift ( for the Q(1) vibrational transition) of any gas and is transparent down to the VUV. We used a 40-cm-long kagomé-type HC-PCF (“kagomé-PCF”) with a core in diameter. When gas filled, the fiber was pumped with pulses at 532 nm, resulting via SRS in generation of a noise-seeded Stokes signal at 683 nm (note that rotational SRS can be disregarded; see Supplement 1). The beat-note created by these two optical fields drives a coherence wave () of molecular oscillations that, within their coherence lifetime, can be used for thresholdless phase-matched frequency up- or down-conversion of narrowband UV pulses of different modal content and frequency. This is possible because of the special frequency dependence of the propagation constant of the modes guided in gas-filled kagomé-PCF.
Such dispersion curves are displayed in Fig. 1(a) for the fundamental Gaussian-like mode of the kagomé-PCF filled with 4 bar and 5.3 bar hydrogen (see Supplement 1). The arrows mark the four vectors, the vertical projection being the Raman frequency shift and the horizontal projection the wavevector , where and are the propagation constants of the pump and Stokes modes. This can be used for thresholdless up-shifting of a DUV mixing pulse at 266 nm (propagation constant ) to its anti-Stokes band (239 nm, propagation constant ) and to greatly lower the threshold for down-shifting to the Stokes band (299 nm, propagation constant ), provided the dephasing rates given by and are such that the dephasing lengths and are longer than the fiber length.
Figure 1(a) shows that these dephasing rates can be made vanishingly small (i.e., ) for collinear generation of anti-Stokes and Stokes signals at 4 bar and 5.3 bar, illustrating the exquisite pressure tunability of these conversion processes. Scanning electron microscopy of the fiber cross section [Fig. 1(b)] revealed that the core-wall thickness was , resulting in a loss-inducing anti-crossing  at . Away from this loss band, the fiber has low loss and spectrally flat dispersion. The 22 μm core diameter ensures that a dispersion landscape optimal for in-fiber SRS dynamics in the UV is achieved at pressures well above 1 bar, something that is impossible to achieve in wide-bore capillaries or bulk gas cells [5–7], although other schemes involving nonlinear dynamics in noble gases pumped by ultrashort pulses might perform optimally at higher pressures .
The experimental setup is sketched in Fig. 2(a). The linearly polarized pump pulses and DUV mixing pulses (duration ) were co-launched into the fundamental mode of the fiber (see Supplement 1 for further details). The mixing pulse energy was kept well below the SRS threshold, ensuring that the dynamics were driven purely by the pre-existing molecular coherence. Since we operated in the “transient” SRS regime , the molecular coherence built up under the pump pulse envelope and featured a typical lifetime of in at low pressure. As a result, the efficiency could be optimized by tuning the mixing pulse delay; a value of was optimal. The gas pressure was regulated in fine steps () and the generated UV bands spatially separated using a prism. This facilitated background-free identification of the wavelengths of the individual Raman lines (using an intensity-calibrated silicon CCD spectrometer) as well as measurement of their individual powers with a calibrated photodiode.
Figure 2(b) shows the spectrum of the dual comb when pump pulse energy and DUV pulse energy were launched into the fiber filled with bar of . With the 532 nm light blocked, no Raman sidebands of the mixing signal were observed, as expected [see Fig. 2(b)]. The mixing beam energy was chosen so that all the bands could be detected with the equipment available. The technique should, however, work down to the single-photon limit since the underlying UV scattering process is effectively linear, i.e., energy independent, as we will discuss below.
In the following, we use to denote the energy in the -th sideband of the mixing signal at the fiber output ( for Stokes bands) and the energy in the -th sideband of the green pump at the fiber output. Figure 3(a) shows the pressure dependence of the overall conversion efficiency to the sidebands, quantified by , where is the out-coupled energy of the mixing signal with the green pump light switched off. Note that, apart from the energy-independent fiber loss, the only mechanism responsible for the depletion of the mixing signal in the presence of pump light is the generation of further frequency-shifted mixing sidebands via SRS. The launched pump energy was , and the energy of the 266 nm pulse alone, measured at the fiber output, was . Several local maxima in are apparent in Fig. 3(a), reaching peak values greater than 45%. We attribute the complex pressure dependence of , which is particularly noticeable at higher pump energies (see Supplement 1), to several factors. First, phase matching to and is satisfied at different pressures [see Fig. 1(a)]. Second, the scattering process is enhanced at higher pressures through increased Raman gain . Finally, as the Raman gain increases, the first-order sidebands become stronger, resulting in conversion to higher-order sidebands. Moreover, higher pump energies resulted in higher conversion efficiencies; in the experiment, we obtained for of green pump energy (see Fig. 4). All these results are corroborated by numerical solutions of multimode Maxwell–Bloch equations  displayed in Fig. 3(a) lower (see Supplement 1 for details).
The pressure dependence of the and signals, normalized to their peak values, is shown in Fig. 3(b). There is remarkably good agreement between theory and experiment. Figure 1(a) shows that both and peak close to their predicted phase-matching pressures (the double-humped structure around 5.3 bar in arises from the dynamics of the conversion process). When is highest, is very weak and vice-versa, demonstrating full selectivity of the direction of energy exchange between and the coherence wave. The dips in signal at for and bar for are caused by conversion to the next-order sidebands (218 nm) and (342 nm) (see Supplement 1).
The weak signals measured at low gas pressure originate from the presence of higher-order modes (HOMs), as seen in the simulations [inset in Fig. 3(b)] and confirmed by the far-field images in Fig. 5(a). These frequency-shifted HOMs result from some HOM content in the mixing signal , together with efficient phase-matched transitions to ultraviolet HOMs via intramodal coherence waves [Fig. 5(b)]. Since the UV wavelength conversion is mode selective, it can be tailored to generate specific DUV-VUV beam profiles for different applications.
To demonstrate that the conversion process is thresholdless, we recorded as a function of at 4 bar [see Fig. 6(a)], revealing that is almost independent of , and that the scattering is effectively linear. This means that the system should in theory be efficient down to the single-photon limit. In practice, the lowest values of that could be measured in the experiments were , limited by the detector sensitivity. At high DUV energies, however, the process becomes nonlinear because the mixing beam starts to generate its own molecular coherence via SRS.
As suggested above, the unique guiding properties of gas-filled kagomé-PCF means that frequency conversion will also work in the VUV. Figures 6(b) and 6(c) show the pressure dependence of the (199 nm) and (184 nm) signals. At pressure, the signal reaches [see Fig. 6(b)], corresponding to a conversion efficiency of 6.6% from the signal. It is likely that the efficiency can be further increased by optimizing the system—far from a trivial task, given the onset of the signal [lower peak in Fig. 6(c)] and the complex spatio-temporal evolution of the coherence at high pump energies. For example, we found that by reducing the launched pump energy to , the conversion efficiency to the signal rose to (see Supplement 1). At , of VUV light were generated in the sideband at 184 nm, corresponding to conversion efficiency [see Fig. 6(c)]. It is remarkable that, apart from the low-pressure region in Fig. 6(b), the modal selectivity of the system makes it possible to generate VUV radiation in a pure fundamental mode with a smooth Gaussian-like profile [see insets in Figs. 6(b) and 6(c)], something that was not possible in previous work on noise-seeded Raman combs .
In conclusion, long-lived molecular coherence excited in the gas-filled core of a HC-PCF enables highly efficient, pressure-tunable frequency conversion of arbitrary signals in the DUV and VUV. The modal content of the DUV–VUV sidebands can be controlled to a great degree, making it possible to generate both Gaussian-like and spatially structured beams. We anticipate that, with further improvements in the DUV–VUV performance of broadband HC-PCFs, a family of compact, coherent UV light sources with broad discrete spectra will emerge, and might constitute a viable complementary route to other established approaches based on femtosecond laser technology for applications in, for example, generation of arbitrary waveforms , tailored attosecond pulse trains , and UV frequency combs .
See Supplement 1 for supporting content.
1. M. T. Neves-Petersen, G. P. Gajula, and S. B. Petersen, “UV light effects on proteins: from photochemistry to nanomedicine,” in Molecular Photochemistry, S. Saha, ed. (IntechOpen, 2012).
2. H. Xuan, H. Igarashi, S. Ito, C. Qu, Z. Zhao, and Y. Kobayashi, Appl. Sci. 8, 233 (2018). [CrossRef]
3. L. Bergé and S. Skupin, Opt. Lett. 33, 750 (2008). [CrossRef]
4. F. Reiter, U. Graf, M. Schultze, W. Schweinberger, H. Schroeder, N. Karpowicz, A. M. Azzeer, R. Kienberger, F. Krausz, and E. Goulielmakis, Opt. Lett. 35, 2248 (2010). [CrossRef]
5. Y. Mori and T. Imasaka, Appl. Sci. 8, 784 (2018). [CrossRef]
6. D. Vu, T. N. Nguyen, and T. Imasaka, Opt. Laser Technol. 88, 184 (2017). [CrossRef]
7. C. G. Durfee III, S. Backus, M. M. Murnane, and H. C. Kapteyn, Opt. Lett. 22, 1565 (1997). [CrossRef]
8. C. Wei, R. J. Weiblen, C. R. Menyuk, and J. Hu, Adv. Opt. Photon. 9, 504 (2017). [CrossRef]
9. P. St. J. Russell, P. Hölzer, W. Chang, A. Abdolvand, and J. C. Travers, Nat. Photonics 8, 278 (2014). [CrossRef]
10. F. Benabid, J. C. Knight, G. Antonopoulos, and P. St. J. Russell, Science 298, 399 (2002). [CrossRef]
11. F. Couny, F. Benabid, P. J. Roberts, P. S. Light, and M. G. Raymer, Science 318, 1118 (2007). [CrossRef]
12. M. K. Mridha, D. Novoa, S. T. Bauerschmidt, A. Abdolvand, and P. St. J. Russell, Opt. Lett. 41, 2811 (2016). [CrossRef]
13. S. T. Bauerschmidt, D. Novoa, A. Abdolvand, and P. St. J. Russell, Optica 2, 536 (2015). [CrossRef]
14. K. Hakuta, M. Suzuki, M. Katsuragawa, and J. Z. Li, Phys. Rev. Lett. 79, 209 (1997). [CrossRef]
15. J. L. Archambault, R. J. Black, S. Lacroix, and J. Bures, J. Lightwave Technol. 11, 416 (1993). [CrossRef]
16. F. Reiter, U. Graf, E. E. Serebryannikov, W. Schweinberger, M. Fiess, M. Schultze, A. M. Azzeer, R. Kienberger, F. Krausz, A. M. Zheltikov, and E. Goulielmakis, Phys. Rev. Lett. 105, 243902 (2010). [CrossRef]
17. P. Hosseini, D. Novoa, A. Abdolvand, and P. St. J. Russell, Phys. Rev. Lett. 119, 253903 (2017). [CrossRef]
18. W. K. Bischel and M. J. Dyer, J. Opt. Soc. Am. B 3, 677 (1986). [CrossRef]
19. H.-S. Chan, Z.-M. Hsieh, W.-H. Liang, A. H. Kung, C.-K. Lee, C.-J. Lai, R.-P. Pan, and L.-H. Peng, Science 331, 1165 (2011). [CrossRef]
20. K. Yoshii, J. K. Anthony, and M. Katsuragawa, Light: Sci. Appl. 2, e58 (2013). [CrossRef]
21. G. Porat, C. M. Heyl, S. B. Schoun, C. Benko, N. Dörre, K. L. Corwin, and J. Ye, Nat. Photonics 12, 387 (2018). [CrossRef]