Supercontinuum spanning over an octave from 900 – 2300 nm is reported in an all-normal dispersion, soft glass photonic crystal fiber. The all-solid microstructured fiber was engineered to achieve a normal dispersion profile flattened to within −50 to −30 ps/nm/km in the wavelength range of 1100 – 2700 nm. Under pumping with 75 fs pulses centered at 1550 nm, the recorded spectral flatness is 7 dB in the 930 – 2170 nm range, and significantly less if cladding modes present in the uncoated photonic crystal fiber are removed. To the best of our knowledge, this is the first report of an octave-spanning, all-normal dispersion supercontinuum generation in a non-silica microstructured fiber, where the spectrum long-wavelength edge is red-shifted to as far as 2300 nm. This is also an important step in moving the concept of ultrafast coherent supercontinuum generation in all-normal dispersion fibers further towards the mid-infrared spectral region.
© 2014 Optical Society of America
Supercontinuum (SC) generation is a widely investigated phenomenon, which can occur in an optical medium due to combined effects of the medium’s dispersion and nonlinearity . While the visible and near-infrared range of wavelengths are covered by commercial supercontinuum sources, the mid-infrared region is still under intense investigation [2,3]. In most of the studied cases, spectral broadening takes place with pump pulses at a wavelength close to the zero-dispersion wavelength of the fiber in the anomalous dispersion range. The broadening mechanisms are then soliton fission or modulation instability (MI) on the anomalous side of the dispersion profile, accompanied by emission of dispersive waves in the normal dispersion. Spectra based on soliton dynamics are, however, susceptible to laser shot noise, and the output pulse displays complex temporal structure making recompression difficult if not impossible . An alternative approach is supercontinuum generation in a fiber that exhibits only normal dispersion, where soliton formation and broadband noise amplification by MI is prohibited. Historically, the first ever demonstration of supercontinuum was obtained in the normal dispersion region of a bulk borosilicate glass sample and covered the visible spectrum . First demonstrations of supercontinuum generation in optical fibers also involved operation in the normal dispersion range of wavelengths, although pump pulse lengths in the order of 10-20 ns enabled Stimulated Raman Scattering, which was detrimental to the recorded spectral characteristics [5,6].
More recently, in silica photonic crystal fibers (PCF) with engineered normal dispersion, telecommunication bandwidth supercontinuum (1450-1700 nm) was demonstrated in  and octave spanning bandwidths were obtained in [8,9] covering roughly 500-1500 nm and 1100-2100 nm. The broadening mechanism in these cases was identified to be an interplay between self-phase modulation (SPM), optical wave-breaking (OWB) and degenerate four-wave mixing (d-FWM) between the SPM and OWB spectral components . Since this broadening mechanism maintains a deterministic phase relation between the newly generated wavelengths and the pump pulse, the resulting supercontinuum spectrum is characterized by a high degree of pulse-to-pulse coherence [8–10]. The reported spectra were also flat to less than 10 dB across most of their bandwidths. A source of flat, broad spectrum can enable e.g. constant signal-to-noise ratio (and constant dynamic range) in spectral measurements, in communication systems based on dense wavelength division multiplexing (DWDM), or in optical frequency comb generation.
A feasible method for extending the spectral bandwidth is the use of a nonlinear all-solid soft glass PCF. The idea is to use a holey fiber design, in which the air-holes are filled with solid rods made of glass with thermal properties matched to the core and lattice glass, but with a different value of the refractive index. Such fibers were first demonstrated by Feng et al . Their advantage over the air-hole PCFs is the additional flexibility of dispersion engineering, which is achieved not only through manipulation of waveguide dispersion, but is also influenced by the material dispersions of the pair of glasses. We recently demonstrated dispersion engineering for flattened, all-normal dispersion (ANDi) profiles in the infrared range up to 2550 nm . Using dispersion engineering approach described in that work, we developed two test structures of all-solid, hexagonal lattice PCFs, similar to the structure shown in Fig. 1. One of these PCFs, with lattice parameters (defined in Fig. 1) of d = 2.11 µm – glass inclusion size, d/Λ = 0.91 – relative inclusion size), core diameter of 2.37 µm and lattice diagonal of 34.65 µm has been successfully applied for coherent SC generation in 950-1850 nm range under pumping with a nonstandard wavelength of 1360 nm (pulse duration of 120 fs) . Based on the obtained results, we fabricated a series of all-solid PCFs with the same structure, but with different geometrical dimensions of the photonic lattice. In this work we demonstrate how these changes influence the normal dispersion characteristics and the SC spectra recorded experimentally for each of the developed PCFs. In the SC generation experiments, pump pulses with duration of 75 fs at the standard telecommunications wavelength of 1550 nm and with peak power of up to 200 kW were used to generate almost perfectly flat (apart from a cladding mode peak) spectra in the range of up to 900-2300 nm, which to our best knowledge is the broadest ANDi supercontinuum spectrum reported to date. Specifically, this result exceeds, in terms of spectral width and the location of long-wavelength edge, the prior reports on ANDi supercontinuum generation in silica air-hole photonic crystal fibers reported in [8,9].
2. Experimental conditions
The all-solid photonic crystal fibers used in this work were fabricated from an in-house synthesized boron-silicate glass (labelled NC21A) and a commercial N-F2 silicate glass (Schott). A standard stack-and-draw technique was used. A N-F2 glass rod and capillaries were used for the core and lattice, which was filled with NC21A glass rods. In each PCF, the entire photonic structure is surrounded with a NC21A glass tube. The refractive indices of the glasses, measured at 1550 nm are nN-F2 = 1.594874 and nNC21A = 1.511304. The NC21A glass composition (mol.%: SiO2 - 56.84, B2O3 - 23.19, Al2O3 - 0.61, Li2O - 6.23, Na2O - 9.51, K2O - 3.63) was selected such that its thermal and rheological properties were matched to N-F2, while at the same time keeping the refractive indices of the two glasses different . Glass transition temperatures are Tg N-F2 = 569C and Tg NC21A = 492 C. Nonlinear refractive indices for these two glasses, measured at 1240 nm were reported previously  and are 2.9 × 10−20 m2/W (N-F2) and 1.1 × 10−20 m2/W (NC21A). The circular glass elements of the photonic structure preforms were transformed into hexagonal shape during the drawing process. All of the five fabricated PCFs (with designations B1-B5) had a regular honeycomb lattice with geometric parameters summarized in Table 1. Fiber designated B3 had geometric parameters identical to the one used in . Parameters of the fibers, including the outer diameter, the diagonal of photonic cladding and the widths of inclusions and the core, decrease with increasing designation digit from B1 to B5. Value of relative inclusion size d/Λ did not change from fiber to fiber. Exemplary SEM images of the PCF designated B1, for which the widest SC spectrum has been observed, are shown in Fig. 1.
Attenuation of the PCFs, measured in the range of 600-1700 nm was comparable for all five fibers and similar to one presented in . The background near-infrared attenuation is contained between 2 - 3 dB/m, peaking up to about 5 dB/m at 1400-1500 nm due to the presence of OH ions. Transmittance measurements of bulk NC21A and N-F2 glass confirmed that this pair of glasses has a transmission window extending to about 2600 nm . Dispersion characteristics of all five PCFs, measured using a Mach-Zehnder interferometer setup, are shown in Fig. 2. With decreasing of the geometric parameters of the developed fibers, the curve of the measured dispersion profiles decreases and the location of its maximum shifts towards the shorter wavelengths. Fiber B1 had the lowest absolute value of normal dispersion with the maximum located at −20 ps/nm/km, 1550 nm. With decreasing dimensions of the photonic structure, in the fiber B5, the local maximum of dispersion dropped to −47 ps/nm/km at 1450 nm.
The linear characteristics, including the dispersion profile and effective mode area, were calculated using the finite element method for the fiber B1 based on its real geometry, given by the SEM image of photonic structure, and were used later in modeling. Figure 3 shows the effective mode area and the dispersion profile of the B1 fiber obtained this way. The calculated dispersion profile is in excellent agreement with the measured dispersion. A small discrepancy is attributed to the diffusion of the glasses at the interfaces in the photonic structure, which was not taken into account in the modelling. Discrepancy can be also assigned to uncertainty of the material dispersion of the two glasses, which in the modelling is represented approximately by the Sellmeier’s formula. Liner simulations also enabled to estimate, that the fibers begin to be multimoded at wavelengths shorter than 1400 nm. This is in agreement with measured numerical aperture for the presented fibers, which was 0.21 at 1550 nm and 0.32 at 1310 nm.
It is worth noting, that similar S-shaped and flattened dispersion profiles were recently reported in ZBLAN step-index fibers . In that case, flattening of the dispersion profile occurred in the anomalous range of dispersion values for most of the demonstrated fibers. The only reported normal dispersion characteristic in the flattened part reached low absolute values (−5 to −10 ps/nm/km), which may contribute to increased bandwidth, but at the cost of decreased spectral flatness, as shown earlier for ANDi silica fibers .
The SC measurement setup is presented in Fig. 4. Pulses from an in-house built Ti:Sapphire femtosecond oscillator (785 nm central wavelength) were amplified using chirped pulse amplification (CPA) technique in a Coherent RegA 9000 regenerative amplifier with external stretcher and compressor. The amplified pulses were compressed to 75 fs duration (measured with an autocorrelator) and had maximum energy of 3.7 µJ at 100 kHz repetition rate. The laser pulses were then fed into a Coherent OPA 9800 optical parametric amplifier (OPA) to generate collinear tunable signal and idler beams. The central wavelength of the signal beam was tuned to 1550 nm, as this is a standard telecommunications wavelength, which matches the local dispersion maximum of the designed fibers. The signal beam was isolated by a polarization-sensitive dichroic mirror and coupled into the measured fibers with a × 20 microscope objective. We estimate the coupling efficiency to be 40%. The supercontinuum generated in the fiber was collected by an apochromatic objective and directed to one of two available spectrometers. For wavelength range up to 1735 nm we used an Ocean Optics NIR512 InGaAs spectrometer. For longer wavelengths we used a Shamrock SR-303i spectrograph with installed 300 l/mm grating. An in-house built camera sensitive in the wavelength range of 1.7 to 10 µm was attached to the spectrograph as a detection element. The camera, operating in a differential mode, was based on a customized two-dimensional HgCdTe MARS detector manufactured by Sofradir. High-quality spectra were obtained after performing several numerical operations: focal-plane-array nonuniformity correction, bad pixel correction , vertical binning and weighted stitching. Both spectrometers were carefully calibrated with the same 10 W Bentham CL2 spectral irradiance standard. The overlapping spectral region of 1700 – 1735 nm was used to merge the amplitude-corrected spectra acquired from the different spectrometers. The maximum peak power available from the pump source was estimated to be at the level of 200 kW. The desired power level was achieved by electronically controlling the extraction efficiency from the regenerative amplifier.
3. Results and discussion
The SC spectra, measured in the developed all-solid PCFs for coupled pump pulse energy of 30 nJ, are shown in Fig. 5. Initial numerical simulations of the SC generation process indicated that one should expect rapid spectral broadening under our experimental conditions, and the spectrum should reach full bandwidth in less than 5 cm. The length of the samples was selected as a compromise between attenuation of the fiber and handling convenience and was: 57 mm for fiber B1, 71 mm (B2), 47 mm (B3), 60 mm (B4) and 67 mm (B5). The spectra were very stable and repeatable. The peak around the pump wavelength is attributed to unconverted pump light propagating in the fiber’s cladding, as at this stage of development, the fiber was not coated with a high-index polymer coating.
Short wavelength edge of all spectra is located roughly at the same wavelength of around 900 nm. For this wavelength, measured dispersion profiles of the fibers become almost indistinguishable with a steep increase of absolute value of normal dispersion. At the red-shifted edge, the spectra reach different wavelengths: 2100 nm in case of fiber B5, up to 2300 nm in fiber B1. This difference is assigned to the wavelength of the local dispersion maximum of fiber, which for the case of fiber B1 and the broadest supercontinuum spectrum coincided with the pump wavelength (1550 nm). Dispersion profile of fiber B1 is also the flattest among the measured characteristics at wavelengths longer than 1550 nm and linear simulations allow to expect that this flatness is maintained up to about 2500-2600 nm (Fig. 3). The difference in the lengths of individual fiber samples does not influence the spectral width of the recorded spectra, which was verified numerically. The width of the spectra is limited by the pump power coupled into the fibers. Even in the case of B1 fiber, for which the long-wavelength edge was red-shifted the most, there is still potential for extension of spectrum, since the glass absorption edge is located at about 2600 nm.
Numerical analysis of supercontinuum generation for the B1 fiber is presented in Fig. 6. The model was based on the numerical solution to the scalar nonlinear propagation equation  and included the frequency-dependent effective mode area and fiber loss, as well as pump laser shot noise . Numerical and experimental results are in a very good agreement, except for the cladding mode peak, which was not reproduced by simulations. The fraction of energy, calculated as an integral under the “peak” feature riding on the flat-top spectrum in Fig. 6, is 60% of the total energy, taken as an integral over the entire experimentally recorded spectrum. This leaves the remainder of 40% under the flat-top spectrum as a rough estimate of the coupling efficiency into the core of the fiber, with the 60% of energy coupled to the cladding modes.
The presence of cladding mode radiation in the fiber was also confirmed by monitoring of the HgCdTe detector output in camera mode during coupling optimization. The detector did pick up some residual signal of the pump source. While the supercontinuum signal was always observable at the center of the detector stripe, which imaged the fiber core, the edges of the detector stripe, imaging the photonic cladding, were always recording the residual pump signal.
About 50 nm displacement between the measured and numerical spectra is also observable in Fig. 6, which can be attributed to limited precision of calculation of the effective mode area profile using the SEM image of the fiber.
Based on the obtained simulation accuracy, we calculated the temporal supercontinuum output pulse, shown in Fig. 7. A single pulse is maintained with regular shape and without fine structure, often present in soliton-based spectra. The small temporal modulation noticeable at the envelope of the pulse originates from numerical one-photon-per-mode noise. It is to be noted that the experimentally recorded spectra and the numerically generated temporal supercontinuum pulse profile are not influenced by Stimulated Raman Scattering (SRS) due to short pump pulse length used in the experiment. This is in contrast with results reported e.g. in , where 10 ps length of pump pulse was enough for SRS to become a significant contribution to the nonlinear broadening process. Dependence of contribution of SRS on pump pulse length to shaping of the ANDi spectrum has not been studied in the fibers investigated herein, but it can performed in a future work using approach demonstrated e.g. in .
The pulse in Fig. 7 is also asymmetric, with extended trailing edge, which is due to steepening of the dispersion profile for the shorter wavelengths (see Fig. 3). The numerically reconstructed dynamics of the broadening process are shown as a spectral evolution over the propagation distance and a series of spectrograms in Fig. 8. The process in the demonstrated study is typical for normal dispersion broadening, as described earlier , beginning with SPM, followed by OWB and parametric mixing between spectral components of the two processes, extending the spectrum almost symmetrically towards longer and shorter wavelengths. In general, the shape of the spectrogram resembles the shape of the dispersion profile . From the shape of B1 dispersion in Fig. 3 and the supercontinuum output spectrogram in Fig. 8d, it can be concluded, that the increasing slope of short-wavelength dispersion of the fiber limits the extent of the blue-shifted broadening. This is manifested by the curved shape of the trailing part of spectrogram. The flattened, long-wavelength part of dispersion corresponds to a “linear” shape of the leading edge of the spectrogram. Therefore it covers a longer range of red-shifted wavelengths, than it would have for a sloped long-wavelength dispersion profile – which would correspond to a bending of the leading spectrogram towards larger negative values of delay and less extent of red-shifted broadening. The lack of observed bending of spectrogram at the long-wavelength edge of spectrum also supports the earlier claim, that the broadening is pump-power limited.
The spectrum becomes flat and smooth in just less than 5 cm of propagation. Further spectral broadening was limited by the available coupled pump power before the long-wavelength edge of the fiber’s transmission band was reached. A further spectral extension towards mid-IR wavelengths under maintained pumping conditions would be possible in a fiber made from glasses with higher nonlinearity. Such a fiber is currently under study.
We have demonstrated what we believe is the first experimental report of ANDi supercontinuum generation in a non-silica glass fiber with a spectral bandwidth covering over an octave and a long-wavelength edge extending as far as 2300 nm. The flattened normal dispersion profiles of the demonstrated PCFs were engineered in an all-solid photonic structure made of two thermally matched soft glasses. The spectral bandwidth was limited by the amount of coupled pump power, even though the transmission of the bulk glasses should allow for extension of its long-wavelength edge up to around 2600 nm. Such a SC source could be applied as the first stage of a two-stage cascade with a chalcogenide nonlinear fiber for ultra-stable, ultra-broad-band SC generation, where it could replace the fluoride fiber in a setup proposed in . As an alternative, the use of other, thermally matched glasses with higher Kerr nonlinearity for the core and lattice could be investigated for a broad spectrum under maintained pump power conditions. This could open up new applications for the proposed ANDi supercontinuum source in e.g. hyperspectral microscopy, which would benefit from improved signal to noise ratio over already demonstrated setups with soliton-based supercontinuum light source . Our results demonstrate that the concept of ultrafast coherent SC generation in ANDi fibers can be successfully transferred to non-silica glasses, and therefore represent an important first step to extend this approach to mid-IR wavelengths in the near future.
This work was supported the project TEAM/2012-9/1 operated within the Foundation for Polish Science Team Programme co-financed by the European Regional Development Fund, Operational Program Innovative Economy 2007-2013 and by Harmonia project UMO-2012/06/M/ST2/00479, funded by the National Science Centre in Poland.
References and links
1. J. M. Dudley, G. Genty, and S. Coen, “Supercontinuum generation in photonic crystal fiber,” Rev. Mod. Phys. 78(4), 1135–1184 (2006). [CrossRef]
2. P. Domachuk, N. A. Wolchover, M. Cronin-Golomb, A. Wang, A. K. George, C. M. B. Cordeiro, J. C. Knight, and F. G. Omenetto, “Over 4000 nm bandwidth of mid-IR supercontinuum generation in sub-centimeter segments of highly nonlinear tellurite PCFs,” Opt. Express 16(10), 7161–7168 (2008). [CrossRef] [PubMed]
3. A. M. Heidt, J. H. V. Price, C. Baskiotis, J. S. Feehan, Z. Li, S. U. Alam, and D. J. Richardson, “Mid-infrared ZBLAN fiber supercontinuum source using picosecond diode-pumping at 2 µm,” Opt. Express 21(20), 24281–24287 (2013). [CrossRef] [PubMed]
4. R. R. Alfano and S. L. Shapiro, “Emission in the region 4000 to 7000 Å via four-photon coupling in glass,” Phys. Rev. Lett. 24(11), 584–587 (1970). [CrossRef]
5. C. Lin and R. H. Stolen, “New nanosecond continuum for excited-state spectroscopy,” Appl. Phys. Lett. 28(4), 216–218 (1976). [CrossRef]
6. C. Lin and W. G. French, “Wideband near-I.R. continuum (0.7-2.1 μm) generated in low-loss optical fibres,” Electron. Lett. 14(25), 822–823 (1978). [CrossRef]
7. K. K. Chow, Y. Takushima, C. Lin, C. Shu, and A. Bjarklev, “Flat super-continuum generation based on normal dispersion nonlinear photonic crystal fibre,” Electron. Lett. 42(17), 989–991 (2006). [CrossRef]
8. A. M. Heidt, A. Hartung, G. W. Bosman, P. Krok, E. G. Rohwer, H. Schwoerer, and H. Bartelt, “Coherent octave spanning near-infrared and visible supercontinuum generation in all-normal dispersion photonic crystal fibers,” Opt. Express 19(4), 3775–3787 (2011). [CrossRef] [PubMed]
9. N. Nishizawa and J. Takayanagi, “Octave spanning high-quality supercontinuum generation in all-fiber system,” J. Opt. Soc. Am. B 24(8), 1786–1792 (2007). [CrossRef]
10. C. Finot, B. Kibler, L. Provost, and S. Wabnitz, “Beneficial impact of wave-breaking or coherent continuum formation in normally dispersive nonlinear fibers,” J. Opt. Soc. Am. B 25(11), 1938–1948 (2008). [CrossRef]
13. G. Stepniewski, M. Klimczak, H. Bookey, B. Siwicki, D. Pysz, R. Stepien, A. K. Kar, A. J. Waddie, M. R. Taghizadeh, and R. Buczynski, “Broadband supercontinuum generation in normal dispersion all-solid photonic crystal fiber pumped near 1300 nm,” Laser Phys. Lett. 11(5), 055103 (2014). [CrossRef]
14. R. Stepien, J. Cimek, D. Pysz, I. Kujawa, M. Klimczak, and R. Buczynski, “Soft glasses for photonic crystal fibers and microstructured optical components,” Opt. Eng. 53(7), 071815 (2014). [CrossRef]
15. I. Kubat, C. S. Agger, P. M. Moselund, and O. Bang, “Mid-infrared supercontinuum generation to 4.5 um in uniform and tapered ZBLAN step-index fibers by direct pumping at 1064 or 1550 nm,” J. Opt. Soc. Am. B 30(10), 2743–2757 (2013). [CrossRef]
16. A. M. Heidt, “Pulse preserving flat-top supercontinuum generation in all-normal dispersion photonic crystal fibers,” J. Opt. Soc. Am. B 27(3), 550–559 (2010). [CrossRef]
17. P. Skibiński and C. Radzewicz, “Bad pixel correction method for locally analytic images: application to infrared spectroscopy,” J. Electron. Imaging 22(4), 043020 (2013). [CrossRef]
18. J. C. Travers, M. H. Frosz, and J. M. Dudley, Nonlinear Fiber Optics Overview, Chap. 3 in Supercontinuum Generation in Optical Fibers, J. M. Dudley and R. Taylor (Cambridge University Press 2010)
20. U. Møller and O. Bang, “Intensity noise in normal-pumped picoseconds supercontinuum generation, where higher-order Raman lines cross into the anomalous dispersion regime,” Electron. Lett. 49(1), 63–65 (2013). [CrossRef]
21. U. Møller, S. T. Sørensen, C. Jakobsen, J. Johansen, P. M. Moselund, C. L. Thomsen, and O. Bang, “Power dependence of supercontinuum noise in uniform and tapered PCFs,” Opt. Express 20(3), 2851–2857 (2012). [CrossRef] [PubMed]
22. S. Dupont, P. M. Moselund, L. Leick, J. Ramsay, and S. R. Keiding, “Up-conversion of a megahertz mid-IR supercontinuum,” J. Opt. Soc. Am. B 30(10), 2570–2575 (2013). [CrossRef]
23. I. Kubat, C. R. Petersen, U. V. Møller, A. Seddon, T. Benson, L. Brilland, D. Méchin, P. M. Moselund, and O. Bang, “Thulium pumped mid-infrared 0.9-9μm supercontinuum generation in concatenated fluoride and chalcogenide glass fibers,” Opt. Express 22(4), 3959–3967 (2014). [CrossRef] [PubMed]
24. S. Dupont, C. Petersen, J. Thøgersen, C. Agger, O. Bang, and S. R. Keiding, “IR Microscopy utilizing intense supercontinuum light source,” Opt. Express 20(5), 4887–4892 (2012). [CrossRef] [PubMed]