We report on visibly white supercontinuum generation in photonic crystal fibers using a sub ns pump source at 1064nm. The spectra extend from below 400nm to 2450nm, some 50nm further into the blue than previously reported spectra. The extra bandwidth which is achieved by a simple modification to the fiber structure gives a higher apparent color temperature and a truly “white” visual appearance. The mechanism for the generation of the deeper blue to ultraviolet frequencies is outlined and our modified fiber is compared with fibers which have been conventionally used for supercontinuum generation.
©2008 Optical Society of America
Broadband supercontinuum generation [1, 2, 3] is one of the most successful applications of photonic crystal fibers (PCF) [4, 5], with compact supercontinuum sources becoming an indispensible piece of equipment in many optical laboratories. Further applications of supercontinua would be enabled if the spectra could be extended further into the infrared or especially into the ultraviolet. Compact laboratory sources consisting of a sub-nanosecond microchip pump laser and an endlessly single mode (ESM) [6, 7] fiber with a zero dispersion wavelength around 1064nm to enable efficient spectral broadening , as well as similar commercially available sources built around picosecond pump sources (again in the 1060nm band) are usually limited on the short wavelength edge to around 450-500nm [3, 8, 9]. The wavelength range immediately below this, 400-450nm is of interest as the spectra would then span the entire visible spectrum and include a useful spectral band for fluorescence imaging applications. So far this band has remained difficult to cover: reported approaches include fiber tapering  and post processing to form fiber devices  and the use of phase-matched higher-order fiber modes .
In this paper we describe how to generate blue-enhanced continua in the fundamental mode of a uniform PCF. The fiber is similar to the high-Δ fibers in  but has a relatively large core.
2. Group index matching
The generation of new optical frequencies in supercontinua arises from several non-linear processes happening simultaneously: four-wave mixing, Raman shifting solitons, self phase modulation and cross phase modulation have all be identified . The specific processes are strongly influenced by the variation of the group index across the transparency window of the fiber. The variation of the group index with wavelength generally takes the form of a skewed “U”, with rapidly decreasing group index (normal group-velocity dispersion) moving from shorter wavelengths, a zero-crossing in the group-velocity dispersion typically around the pump wavelength at 1.06µm, and increasing group index (anomalous dispersion) towards longer wavelengths. Importantly for this work, the “U” rises more steeply on the short-wavelength side than at longer wavelengths, because of the strong material dispersion of silica at these wavelengths. A specific mechanism which frequently dominates the “blue” edge was recently identified by Gorbach et al. [14, 15]. A self frequency-shifting soliton propagating in the anomalous-dispersion (infrared) regime effectively traps blue radiation propagating with the same group index on the other arm of the “U” in a potential well and scatters the blue radiation to shorter wavelengths in a cascaded four-wave mixing process. The long- and short-wavelength edges of the supercontinuum are thus intimately related. On the long wavelength side, the extent of the supercontinuum is limited by the very rapidly rising absorption of the material as one moves beyond 2.5µm, which is typically exacerbated in PCFs by the presence of OH- ions which have an absorption band at around 2.4µm . It follows from this that to generate wavelengths further into the blue or ultraviolet requires that one modify the group index in the infrared to rise more steeply with increasing wavelength in order to match the limiting infrared wavelength to a deeper blue.
Comparing the group index of bulk silica (as shown in Fig. 1) with the computed group index for an ESM PCF (computed using the using the empirical method outlined by Saitoh et al. ) and for a strand of silica surrounded by air, it is seen that although at short wavelengths the three curves are almost indistinguishable the behavior at long wavelengths is very different. The waveguide dispersion causes a steeper increase in the group index (increasing the anomalous dispersion, as is well known) and so matching the index at a specific infrared wavelength to significantly shorter wavelengths in the ultraviolet. As this group-index matching is what is required for blue light generation, it is apparent that a strand of silica surrounded by air would generate shorter wavelengths than an ESM PCF. Practically, this structure can be approximated by a fiber with a large pitch and core size with respect to the fibers reported in  and a high air filling fraction in the cladding.
A large-core high-Δ fiber was fabricated (d/Λ=0.77, Λ=3.7µm Fig 2(b)) from fused silica (Heraeus F300) using the stack and draw method . An identical length of ESM (d/Λ=0.43, Λ=3.0µm Fig 2(a)) was used for comparison. The core sizes for the two fibers were both ~4.7µm. Sub ns (600ps) pulses from a microchip laser at 1064nm, ~60mW output power and 7kHz repetition rate were launched into 10 meter sections of the fibers and the input power was varied up to by means of neutral density filters placed before the fiber input. We measured the output power using a power meter for direct comparison between the two fibers. The visible component of the dispersed output spectra of the two fibers under identical pump conditions as recorded using a digital camera is shown in Fig. 2(c). The short wavelength output (350-1750nm) was collected by a multimode fiber and recorded on an optical spectrum analyzer (OSA) (Ando AO-63158). The long wavelength edge of the continuum (900-2550nm) was collected in a short, straight length of single mode fiber and recorded on an IR spectrometer (Ocean Optics NIR-256). The pump peaks were filtered out using a long pass filter (cut-off wavelength 1600nm) to prevent multi-order interference in the measurement, The complete spectrum is shown in Fig. 3 for an output power of 12.4mW. The short wavelength edges of the ESM and high-Δ fiber are compared in Fig. 4 (insert), while the long wavelength edges were virtually identical to one another and aren’t presented.
The fiber losses were measured at 400nm wavelength to be 250dB/km and 200dB/km for the ESM and high-Δ fibers respectively. This variation is believed not to have a significant effect on the blue to ultraviolet continuum generation in the 10m fiber lengths used, as a similar high-Δ fiber with losses in excess of 0.5dB/m at 400nm has been observed to generate very similar spectra.
Spectra were recorded for four different output powers using each fiber. The short and long wavelength edges identified by identifying a point at a fixed value (either 10, 15 or 20dB) below a feature which appeared in all the spectra for either the long or short wavelength edges. The group index curves for the two fibers (modeled as in ) are plotted as functions of wavelength in Fig. 4. The continuum edges are also marked on the group index curves for each fiber for a series of different output powers, the points on either edge of the total spectra for a given power are joined by straight lines.
The agreement is good (that is, the lines joining the short and long wavelength edges are almost horizontal on the plot) This gives strong support to the concept of group-index matching between the longest and shortest wavelengths being a limiting factor in blue and ultraviolet supercontinuum generation.
The zero dispersion wavelength of our high-Δ fiber has shifted away from the pump at 1064nm and lies around 980nm. Modeling of a similar fiber with a zero dispersion at ~1060nm shows that in such a fiber the group-index-matched short-wavelength edge (matched to 2500nm) lies around 480nm, a comparable value to that in our ESM fiber. It is important to note here that the required group index profile for enhanced blue/ultraviolet radiation cannot be obtained if the zero-GVD wavelength is near 1060nm.
The fiber group index curve can be modified to push the continuum further into the near ultraviolet, by reducing the core size. However this also shifts the zero dispersion wavelength further away from the pump as shown in Fig. 5. A series of three high-Δ fibers were drawn, the fiber above and two similar fibers with different outer diameters and hence core sizes, measured to be 4.4µm and 4.2µm, (the outer diameters of the three fibers are 100µm, 95µm and 90µm). These fibers were pumped in an identical manner and the spectra were recorded directly from the fiber outputs on an OSA. The modeled group index curves and the short wavelength edges of the spectra are shown in Fig. 5(a) and (b).
The shapes of the individual spectra do indeed extend to shorter wavelengths for the smaller cores. However there is an approximately 5dB decrease in the visible spectrum for the smaller fibers, while no decrease was observed on the long wavelength side of the pump. The primary mechanism for supercontinuum generation on the long wavelength side of the pump is Raman shifting solitons, which are relatively unaffected by the proximity of the pump wavelength to the fiber zero GVD point. On the short wavelength side of the pump however, new frequencies are initially generated by other processes such as four-wave mixing  (prior to being shifted to deeper blue frequencies by group index matched solitons), which require the pump to be in close proximity to the zero dispersion wavelength of the fiber. As the fiber pitch is decreased to steepen the infrared edge of the group index curve we also shift the zero GVD point away from the pump reducing the blue-shifted power.
Similar results were obtained using an approximately 7ps pump source at 1.06µm (Fianium Femtopower 1060-1µJ-pp, 0.5MHz repetition rate) using free space coupling. The results are shown in Fig 5(c) and do not show the 5dB decrease between progressively smaller fibers as with the sub-ns source. One possible explanation for this could be the 7ps pulses are inherently broader and also experience more self-phase modulation than the longer pulses, thus providing power closer to the zero GVD point. The spectra shown correspond to an output power of 27mW.
We have demonstrated how changing fibers used for supercontinuum generation allows creation of a truly (ie, visibly) “white” light source by including the wavelength region 400-450nm in the generated spectrum. The entire spectrum is generated in the fundamental fiber-guided mode. The fiber requires transparency in the both the ultraviolet and infrared, similar high-Δ fibers fabricated from ultraviolet grade silica (Heraeus F100) have proved too lossy in the infrared to generate a continuum with a similar blue edge to those demonstrated here. We have used a fiber design with large air holes to modify the group index profile of the fiber, so as to group-index-match long-wavelength-edge (infrared) radiation to shorter wavelengths in the blue/ultraviolet than before. The wavelength band now incorporated will allow new applications to be opened up using compact supercontinuum sources.
The authors acknowledge Chunle Xiong for the endlessly single mode fiber and Fianium ltd. for the picosecond source and useful discussions. This work was funded by the UK Engineering and Physical Sciences Research Council, the Körber Foundation, and the Technology Strategy Board-led Technology Programme.
References and links
1. J. K. Ranka, R. S. Windeler, and A. J. Stentz, “Visible continuum generation in air-silica microstructure optical fibers with anomalous dispersion at 800 nm,” Opt. Lett. 25, 25–27 (2000) [CrossRef]
2. W. Wadsworth, N. Joly, J. Knight, T. Birks, F. Biancalana, and P. Russell, “Supercontinuum and four-wave mixing with Q-switched pulses in endlessly single-mode photonic crystal fibres,” Opt. Express 12, 299–309 (2004) [CrossRef] [PubMed]
3. J. M. Dudley, G. Genty, and S. Coen, “Supercontinuum generation in photonic crystal fiber,” Rev. Mod. Phys. 781135 (2006) [CrossRef]
7. Crystal Fibre A/S: http://www.crystal-fibre.com
8. Fianium: http://www.fianium.com
9. Koheras: http://www.koheras.com
10. A. Kudlinski, A. K. George, J. C. Knight, J. C. Travers, A. B. Rulkov, S. V. Popov, and J. R. Taylor, “Zero-dispersion wavelength decreasing photonic crystal fibers for ultraviolet-extended supercontinuum generation,” Opt. Express 14, 5715–5722 (2006) http://www.opticsinfobase.org/abstract.cfm?URI=oe-14-12-5715 [CrossRef] [PubMed]
11. C. Xiong, A. Witkowska, S. G. Leon-Saval, T. A. Birks, and W. J. Wadsworth, “Enhanced visible continuum generation from a microchip 1064nm laser,” Opt. Express 14, 6188–6193 (2006) http://www.opticsinfobase.org/abstract.cfm?URI=oe-14-13–6188 [CrossRef] [PubMed]
12. C. Lesvigne, V. Couderc, A. Tonello, P. Leproux, A. Barthélémy, S. Lacroix, F. Druon, P. Blandin, M. Hanna, and P. Georges, “Visible supercontinuum generation controlled by intermodal four-wave mixing in microstructured fiber,” Opt. Lett. 32, 2173–2175 (2007) [CrossRef] [PubMed]
13. G. P. Agrawal, “Nonlinear fiber optics,” 4th Edition, Academic Press (2006)
14. A. V. Gorbach, D. V. Skryabin, J. M. Stone, and J. C. Knight, “Four-wave mixing of solitons with radiation and quasi-nondispersive wave packets at the short-wavelength edge of a supercontinuum,” Opt. Express 14, 9854–9863 (2006) http://www.opticsinfobase.org/abstract.cfm?URI=oe-14-21-9854 [CrossRef] [PubMed]
15. A. V. Gorbach and D. V. Skryabin, “Light trapping in gravity-like potentials and expansion of supercontinuum spectra in photonic-crystal fibres,” Nature Photonics 1, 653–657 (2007) [CrossRef]
16. O. Humbach, H. Fabian, U. Grzesik, U. Haken, and W. Heitman, “Analysis of OH absorption bands in synthetic silica,” J. Non-Cryst. Solids 20319–26 (1996) [CrossRef]
17. K. Saitoh and M. Koshiba, “Empirical relations for simple design of photonic crystal fibers,” Opt. Express 13, 267–274 (2005) http://www.opticsinfobase.org/abstract.cfm?URI=oe-13-1-267 [CrossRef] [PubMed]