Supercontinuum generation occurs when intense laser light creates nonlinear processes in a material and produces a continuous, broad spectrum of colors. It is a well-known phenomenon in bulk media and in optical fibers, and is also an important photonic resource that enables applications in telecommunications, the life sciences, and spectroscopy. Rather than being a single process, the concept of supercontinuum generation embraces a smorgasbord of effects, commonly initiated by self-phase modulation, as first reported in the 1970 paper by Robert Alfano and Stanley Shapiro that described supercontinuum generation in glass1. Octave-spanning bandwidths can then result from the cooperative action of phenomena such as four-wave mixing, nonlinear soliton and dispersive-wave propagation, modulation instability, and noninstantaneous Raman nonlinearities. Typically these processes rely on third-order non- linearities, which is a consequence of the fact that these nonlinearities are intrinsically broadband and are available in all materials. A defining and fascinating characteristic of supercontinua is that they evolve with propagation through a medium; the spectrum changes as the pulse travels through a material. Therefore, for a supercontinuum to be achieved, the pulse intensity must be maintained for a sufficient length. In fibers, this is guaranteed by the guided-wave nature of the medium, and in some cases allows supercontinua to be developed over lengths of tens of meters. In solids, however, diffraction restricts the useful length to millimeters, with the consequence that much higher fields are needed. This characteristic typically puts supercontinua in bulk materials out of reach of simple femtosecond laser oscillators. Nonlinear effects In the early 2000s, research by Marty M. Fejer and colleagues2 showed that supercontinuum generation could also be driven by low-energy femtosecond lasers via second-order nonlinear effects in waveguides. The enabling technology behind this work was periodically poled lithium niobate (PPLN)3,4, one of a new class of quasi-phase-matched nonlinear crystals that released second- order effects from the restrictions of birefringent phase matching. And, as predicted four decades earlier by Nicolaas Bloembergen5, quasi-phase matching provided the ability to flexibly engineer the frequency-conversion response of certain nonlinear crystals. Fejer employed a waveguide format of PPLN, which maintained the incident laser field intensity over a few centimeters, and which allowed multiple three-wave nonlinear processes to cooperatively broaden the spectrum to over one octave in bandwidth. Since the first demonstration, such processes — which are now understood to involve both second- and third-order nonlinear effects — have been extensively studied and developed both experimentally6 and theoretically7, but remain limited to examples in guided-wave devices. OPGaP Recent research from Heriot-Watt University8 introduces a new paradigm for supercontinuum generation, and shows that second- and third-order effects in a bulk crystal of orientation-patterned gallium phosphide (OPGaP)9 could convert light from an IR femtosecond oscillator into a broad visible spectrum (Figure 1). The process shares some features with the earlier work of Fejer; similar to PPLN, OPGaP is also a quasi-phase-matched nonlinear crystal, meaning that its conversion properties can be tuned by modifying the periodicity of the crystal’s internal domain structure. But OPGaP also has unique material properties of its own, allowing it to perform differently than PPLN. As a semiconductor material, it possesses an extremely high nonlinear figure of merit — 3× higher than PPLN — which is combined with an exceptional transparency range spanning the visible to the far-infrared. These properties, together with its engineerable frequency-conversion response, are what allow OPGaP to produce a new kind of supercontinuum. Figure 1. A visible supercontinuum produced by a 1-mm-long crystal of OPGaP pumped by 30-nJ femtosecond pulses from a 100-MHz, 1040-nm Yb-doped fiber laser. The crystal’s characteristic yellow color of GaP can be seen. The domain structure in OPGaP is grown by vapor-phase epitaxy from a patterned GaP substrate (the bottom surface of the crystal) and terminates with a rough upper surface. Green light corresponding to the second harmonic of the pump laser wavelength dominates the supercontinuum and serves as a pump for parametric gain processes that generate the other colors. Courtesy of Derryck T. Reid, Marius Rutkauskas, and Luke Maidment. New supercontinuum process Like many other supercontinuum processes, self-phase modulation provides the initial bandwidth broadening necessary to kick-start the spectral evolution. The exceedingly high Kerr nonlinearity of OPGaP (200× that of fused silica) means that even at the relatively modest 30-nJ pulse energies from a 1040-nm femtosecond Yb-doped fiber laser, self-phase modulation sidebands develop on the optical spectrum. At this point, multiple second-order nonlinear effects come into play. First, non-phase-matched second-harmonic generation produces intense 520-nm pulses, with energies around the 100-pJ level. These pulses then act as the pump for a number of high-order phase-matched parametric amplification processes (Figure 2), which are able to amplify light at neighboring longer wavelengths in the green, yellow, orange, and red regions of the spectrum. The origin of the seed light at these wavelengths is the weak frequency-doubled long-wavelength sideband of the original pump spectrum, at a wavelength around 550 nm. As this is amplified, the wings of this spectrum are able to seed amplification at still longer wavelengths. The process repeats, extending the spectral coverage from the green to the red, with some weak blue-green light also observed from other high-order conversion processes. Figure 2. Multiple high-order parametric gain processes drive the evolution of the supercontinuum. Pumped by intense second-harmonic pump light at 520 nm, mth-order parametric gain amplifies weak seed light at wavelengths from 550 to 620 nm (bottom), resulting in the gain profile shown in green (middle), whose structure is reflected in the measured supercontinuum spectra (middle and top). At pump powers above 1 W, self-phase modulation gives rise to IR spectral sidebands, whose second-harmonic light seeds the parametric gain processes (top). Courtesy of Derryck T. Reid. A nonlinear envelope evolution simulation, similar to those routinely used to model supercontinua in optical fibers, supports our understanding of this new supercontinuum process. In a way that is uniquely possible in simulation, the second- and third-order effects can be independently switched on or off, illustrating that without the initial spectral broadening from third-order self-phase modulation, no supercontinuum is observed. Likewise, turning off the second-order nonlinearity results in no visible light being produced (Figure 3). To observe a broadband visible output, both effects must work in tandem. Figure 3. Simulated evolution of the visible and NIR spectra after propagation through a 1-mm-long OPGaP crystal fabricated with a grating period of 27 µm. The upper panels show the spectra obtained at maximum pump power (3.2 W) and the effect on these of switching off either the χ(2) or χ(3) nonlinearity. Less self-phase modulation is observed in practice due to self-defocusing associated with a cascaded χ(3) Kerr nonlinearity, which is not simulated by the model (see Reference 8). SHG: second-harmonic generation. Courtesy of Derryck T. Reid. The modeling also confirms another aspect of how OPGaP’s unique properties are critical to enabling this supercontinuum effect. The green-pumped parametric gain processes begin by amplifying nearby seed wavelengths, resulting in a difference-frequency generation signal in the longwave infrared, from 8 to 12 µm. Far from being a by-product, this IR light is vital to the amplification; it mixes with the pump light to produce more photons at the seed wavelength, which in turn generate more IR light, and so the light field at the longer wavelength builds up. OPGaP’s remarkable transparency range10 therefore appears to play an important role in the supercontinuum process. How should this research be taken forward? Currently the powers generated are in the milliwatt level, well below what can be obtained from photonic crystal fibers using similar laser energies. But OPGaP has the potential to be engineered to optimize the conversion processes involved in supercontinuum generation. If this can be accomplished, then practical power levels may be possible across the visible region from this crystal, avoiding the need for precision coupling into fiber. Additionally, it will provide a compact alternative for visible supercontinuum generation and shrink the functionality of tens of centimeters of photonic crystal fiber into a few millimeters of crystal. Meet the author Derryck T. Reid is professor of physics at Heriot-Watt University in Edinburgh, Scotland. His research concerns the development and application of broadly tunable laser sources, from visible to infrared frequency combs for astronomy, to mid-infrared sources for chemical sensing; email: d.t.reid@hw. Acknowledgment The research was funded by the U.K. Engineering and Physical Sciences Research Council (EPSRC) under grants EP/R033013/1 and EP/P005446/1. References 1. R.R. Alfano and S.L. Shapiro (1970). Observation of self-phase modulation and small-scale filaments in crystals and glasses. Phys Rev Lett, Vol. 24, Issue 11, pp. 592-594. 2. C. Langrock et al. (2007). Generation of octave-spanning spectra inside reverse- photon-exchanged periodically poled lithium niobate waveguides. Opt Lett, Vol. 32, Issue 17, pp. 2478-2480. 3. J. Webjorn et al. (1989). Fabrication of periodically domain-inverted channel waveguides in lithium niobate for second harmonic generation. J Lightwave Technol, Vol. 7, Issue 10, pp. 1597-1600. 4. E.J. Lim et al. (1989). Second-harmonic generation of green light in periodically poled planar lithium niobate waveguide. Electron Lett, Vol. 25, Issue 3, p. 174. 5. J.A. Armstrong et al. (1962). Interactions between light waves in a nonlinear dielec- tric. Phys Rev A, Vol. 127, Issue 6, pp. 1918-1939. 6. C.R. Phillips et al. (2011). Supercontinuum generation in quasi-phase-matched LiNbO3 waveguide pumped by a Tm-doped fiber laser system. Opt Lett, Vol. 36, Issue 19, p. 3912. 7. C.R. Phillips et al. (2011). Supercontinuum generation in quasi-phasematched waveguides. Opt Express, Vol. 19, Issue 20, p. 18754. 8. M. Rutkauskas et al. (2020). Supercontinuum generation in orientation-patterned gallium phosphide. Optica, Vol. 7, Issue 2, p. 172. 9. T. Matsushita et al. (2007). Epitaxial growth of spatially inverted GaP for quasi phase matched nonlinear optical devices. Jpn J Appl Phys, Vol. 46, Issue 17, pp. L408-L410. 10. L. Maidment et al. (2017). Molecular fingerprint-region spectroscopy from 5 to 12 μm using an orientation-patterned gallium phosphide optical parametric oscillator. Opt Lett, Vol. 41, Issue 18, pp. 4261-4264.