Open Access
Add to BibSonomyAbstract
We demonstrate 1250 nm pulses generated in dual-zero dispersion photonic crystal fiber capable of three-photon excitation fluorescence microscopy. The total power conversion efficiency from the 28 fs seed pulse centered at 1075 nm to pulses at 1250 nm, including coupling losses from the nonlinear fiber, is 35%, with up to 67% power conversion efficiency of the fiber coupled light. Frequency-resolved optical gating measurements characterize 1250 nm pulses at 0.6 nJ and 2 nJ, illustrating the change in nonlinear spectral phase accumulation with pulse energy even for nonlinear fiber lengths < 50 mm. The 0.6 nJ pulse has a 26 fs duration and is the shortest nonlinear fiber derived 1250 nm pulse yet reported (to the best of our knowledge). The short pulse durations and energies make these pulses a viable route to producing light at 1250 nm for multiphoton microscopy, which we we demonstrate here, via a three-photon excitation fluorescence microscope.
© 2014 Optical Society of America
1. Introduction
Two-photon excitation fluorescence microscopy captures high resolution three dimensional images through the interaction of ballistic photons and fluorophores within the focal volume of an objective, and it is the exponential loss of the ballistic excitation photons, due to scattering, which limits the depth of penetration in the near-infrared [1]. The reduced optical scattering conferred to longer wavelengths allows pulses centered at 1280 nm to generate images up to 1 mm deep in adult murine brains [2, 3]. In addition to reduced scattering, the lower energy pump photons excite less autofluorescence, decreasing the background light and further extending imaging depth. Additionally, the signal light of second- and third-harmonic generation microscopy from ∼ 1250 nm pulses also benefit from a reduction in scattering as the harmonics scale to longer wavelengths too. Moving to higher-order multiphoton absorption processes (> 2 photon absorption) aggressively decreases fluorescent excitation and emission away from the focal point; further increasing image depths [4].
The prevailing sources of ultrafast pulses around 1250 nm are Ti:sapphire pumped optical parametric oscillators and Cr:forsterite modelocked lasers. Multiphoton microscope systems employing both sources have been successfully demonstrated [2, 3, 5]. To reduce system cost and complexity and to leverage the widely available ultrafast pulses from modelocked Yb-doped fiber lasers, we are pursuing nonlinear fiber propagation to re-distribute > 1 nJ of energy from pulses centered at 1075 nm into the 1250 nm spectral region. The three types of fiber that have the necessary nonlinear and dispersion characteristics are hollow-core photonic bandgap fibers, photonic crystal fibers (PCF) with a single zero-dispersion wavelength (ZDW), and fibers with two ZDWs straddling the seed pulse frequency and an approximately parabolic (concave down) dispersion coefficient curve. The low nonlinearity of air-core photonic bandgap fibers require many 10’s to 100’s of nJ pulses to launch solitons [6]. In this study, we examine processes utilizing oscillator level energies, which are far below the requisite energies of the photonic bandgap fibers.
The novel guiding structure of PCF allows researchers to tune the ZDW to a higher frequency than that of bulk silica [7]. Several commercial PCFs exist with a ZDW shorter than 1060 nm (e.g. NKT Photonics’ SC-3.7-975 and NL-PM-750). The anomalous dispersion and high nonlinearity of these fibers supports solitons at low pulse energies (< 0.1–1 nJ). A soliton propagating in PCF will undergo soliton self-frequency shifting (SSFS) induced by intrapulse Raman scattering. Sub-ps pulses will undergo significant frequency shifting in PCF lengths of < 1 m [8]. On the surface, SSFS seems like an attractive method of converting 1 micron pulses into 1250 nm pulses, but the high fiber nonlinearity also prevents efficient nonlinear power conversion to red-shifted pulse energies over 1 nJ.
A seed pulse with more energy than required to support a single soliton, becomes a higher-ordered soliton inside the PCF [9]. The low constituent soliton number required to maintain coherence and conversion efficiency sets a threshold on the seed pulse energy as a function of pulse duration, fiber nonlinearity, and fiber dispersion [9]. The constituent soliton number is given by N 2 = LD/LNL. The characteristic dispersion and nonlinear lengths are and LNL = 1/γP 0, where β 2 is the group velocity dispersion and P 0 is the maximum pulse peak power. γ = ω 0 n 2(ω 0)/cA eff(ω 0) is the fiber nonlinear coefficient, where ω 0 is the central frequency of the pulse, n 2 is the intensity dependent index of refraction, c is the speed of light, and A eff(ω 0) is the effective mode area at the central frequency. T 0 is defined for an electric field with a hyperbolic secant temporal envelope, , where t is a time frame moving with the envelope group velocity of the pulse. As the pulses generated in our fiber laser are not from a soliton modelocked laser and more Gaussian than hyperbolic secant, we numerically model Gaussian pulses, A(z = 0, t) = exp(−2ln(2)(t/Δτ)2), where the fiber is assumed to start at z = 0. The equivalent T 0 for a hyperbolic-secant having the same full-width-half-max (FWHM) pulse intensity, Δτ, as the Gaussian seed pulse is .
A common interpretation of the soliton order limitation is that by decreasing the pulse duration SSFS is once again a viable option for frequency conversion out to the 1250 nm spectral region. To examine this, we numerically model pulse propagation in a PCF with a ZDW at 890 nm (Thorlabs/Blaze Photonics NL-3.3-890-02). The model consists of a split-step fiber propagator integrating the linear and nonlinear components of the generalized nonlinear Schrödinger equation (GNLSE) separately [9]. The GNLSE used here to model the evolution of the field A(z, t) is
The βk’s are the dispersion coefficients of a taylor expansion of the propagation constant β(ω). The third-order response function is R(t) = (1 − fR)δ(t) + fRhR(t), where δ(t) is the Dirac delta function and hR is an analytic representation of the Raman cross section [10]. Following Ref [9] fR = 0.18 and τ shock = 1/ω 0. Table 1 contains the energies of seed pulses and the first ejected soliton for several pulse durations and soliton numbers for pulses centered at 1060 nm.From the model, SSFS scaling with energy seems promising: N = 6 and 25 fs FWHM seed pulses generate shifted solitons with > 1 nJ of energy. Table 1, however, does not account for the power spectra of the shifted soliton. Figure 1(a) plots the evolution of the power spectra with fiber propagation of the N = 6 and 25 fs FWHM Gaussian pulse. While the shifted soliton contains over 1 nJ of energy, it is beyond target wavelength region. The 25 fs seed pulse generates a spectacular bandwidth during the first quarter of a soliton period (∼ 1 mm), which remains broad up to soliton fission (∼ 5 mm). Supercontinuum (SC) generation is a more accurate characterization of the nonlinear propagation shown in Fig. 1(a) than SSFS, even though SSFS is the dominating effect for continued fiber propagation well beyond soliton fission. The power spectrum of the first ejected soliton, from the point of soliton fission onward to where it is spectrally isolated at 9 mm of propagation, is well beyond the desired 1200–1300 nm range. To circumvent the spectral over-shoot, one would add chirp to the seed pulse. Effectively exchanging the 25 fs and N = 6 seed pulse for a pulse with a larger soliton number and longer duration, which reduces the efficiency of coupling into the highest energy ejected soliton; leading directly to < 1 nJ of energy in the shifted soliton. The SSFS conversion route may still be a viable method of converting ∼ 1060 nm pulses out into the infrared spectral region beyond 1300 nm, but as a means of producing > 1 nJ pulses ∼ 1250 nm it falls short (in this case by overshooting).
Table 1. SSFS conversion efficiency by seed pulse duration and soliton number.
Fig. 1 (a) The modeled SC power spectra for the fiber lengths listed on the right from an N=6, 25 fs FWHM Gaussian seed pulse in PCF with ZDW at 890 nm. (b) A theoretical comparison of the effects of Raman scattering on the dual-band spectra formation in dual-ZDW PCF. Power spectra from 1 nJ, 30 fs FWHM Gaussian seed pulses at the fiber lengths shown to the far right, with (black-solid) and without (blue-dashed) Raman scattering.
The final conversion process is the one we will explore in the remainder of this article: generating a dual-band SC in a PCF which has two ZDW at 1020 and 1076 nm (per the manufacturer), NKT Photonics NL-1050-ZERO-2. Optical coherence tomography (OCT) was the original application of the dual-band spectra from the dual-ZDW PCF [11,12]. The smoothness of the generated SC power spectra was of the utmost importance to minimize the OCT point spread function pedestal, and so the PCF lengths were quite long (∼ 1 m); introducing significant higher-order dispersion effects. Another study pursuing ∼ 1250 nm pulses with > 1 nJ of energy used a higher-order mode fiber with a similar dispersion curve containing two ZDWs but ∼ 10× lower nonlinearity [13]. Coupling 1.39 nJ and 200 fs pulses at 1064 nm into higher-order mode fiber, van Howe et al. produced 0.8 nJ pulses around 1150 and a pulse duration of 49 fs. In related work, the wavelength of SC from higher-order mode fibers was extended to 1350 nm by C̆erenkov radiation; producing 0.66 nJ and 106 fs pulses [14].
Dual-ZDW PCF fiber generates SC with a dual-band power spectrum through the unique dispersion profile and low order nonlinearities, predominantly self phase modulation. Unlike in SSFS, the Raman effect does not play a role in the generation of the red-shifted light. We demonstrate this here by modeling the propagation of a 1 nJ and 30 fs FWHM Gaussian seed pulse at 1060 nm in NL-1050-ZERO-2 PCF with and without the Raman effect. We do this by “turning off” the non-instantaneous component of the response function in equation 1 by setting fR = 0. The results in Fig. 1(b) clearly show that the splitting of the power spectra happens in the absence of Raman scattering. Critically, that the frequency shift is not a SSFS process (or for that matter another soliton-driven process like soliton fission or a superposition of higher-order solitons) indicates dual-ZDW PCF avoids the SSFS energy restriction.
Using 29 mm of NL-1050-Zero-2 PCF, we report here 1250 nm pulses having up to 2nJ of energy and for less energetic pulses (0.6 nJ) durations as short as 26 fs FWHM. To extract fully separated power spectra from the short length of PCF, we use a 28 fs FWHM seed pulses centered at 1075 nm. To the best of our knowledge, the 26 fs FWHM pulse is the shortest reported nonlinear fiber derived pulse in the 1250 nm wavelength range.
2. Setup and application
2.1. Seed pulse source for the 1250 nm nonlinear fiber
The source of the ultrafast seed pulses is the home-built system shown in Fig. 2. An all normal dispersion fiber oscillator and a narrow-band, nonlinear fiber amplifier produce 140 fs FWHM pulses with over 25 nJ of energy at 61 MHz [15]. The 140 fs amplifier pulses seed a fiber broadening stage composed of 142 mm of Leikki Passive-10/125-PM fiber, with 84% total transmission efficiency through the APC connectors and aspheric collimators. A custom folded 4-F pulse shaper compresses the spectrally broadened pulse. A half waveplate and polarizer attenuate the seed pulse down to between 2–6 nJ and a telescope sizes the beam appropriately for fiber coupling.
Fig. 2 Setup for the 1075 nm pulse generation and subsequent 1250 nm pulse generation. ANDi: all normal dispersion laser, NFA: nonlinear fiber amplifier; TG1 and TG2: transmission gratings 1600 l/mm and 1000 l/mm respectively; AL: achromatic lens, MC: Martinez compressor; WP: half-waveplate; C1 and C2: aspheric fiber collimators with APC receptacles; NLF1: nonlinear fiber (10/125-PM), L1, L2, L3, and L4: lenses; PS: pulse shaper; PL: Plössl lens; SLM: spatial light modulator; AWP1, AWP2, and AWP3: achromatic half-waveplates, PBS: polarizing beam splitter; OAPC: off-axis parabolic fiber collimator with APC receptacle; NLF2: nonlinear fiber (NL-1050-ZERO-2); AS: aspheric lens.
A novel pulse shaper design using a Plössl lens dramatically reduces field curvature effects at the spatial light modulator, allowing the utilization of high efficiency transmission gratings and a compact pulse shaper footprint < 1ft2 [16]. The total transmission efficiency through the pulse shaper is 80%. The pulse shaper compensates for the group delay dispersion, β′ 2, from the short fiber broadening stage, the pulse shaper itself, and the telescope before the 1250 nm nonlinear fiber; eliminating β′ 2 ≈ 9, 000fs2 (the spectral phase is given by exp(−iβ′ 2(ω − ω 0)2/2)). The power spectrum of our seed pulse, the post-compensation spectral phase, and the intensity profile are shown in Fig. 3, as measured by second-harmonic generation frequency-resolved optical gating (SHG-FROG) [17]. As seen in Fig. 3(c), the shaped seed pulse is 28 fs FWHM and very close to the bandwidth supported transform-limit. After the pulse shaper, the seed pulse contains 11.5 nJ of energy (370 kW of peak power), although the pulse energy is attenuated for seeding the 1250 nm nonlinear fiber.
Fig. 3 1075 nm nonlinear fiber seed pulse. (a) Phantom-FROG: comparison of the measured (left) and reconstructed (right) SHG-FROG traces, (b) spectral phase (blue) and power spectrum (black), and (c) the temporal profile of the bandwidth supported transform-limited pulse (blue) and the reconstructed pulse (black).
2.2. 1250 nm nonlinear fiber and pulses
Since multiphoton excitation improves with shorter pulse duration [1], one of our primary concerns was minimizing the SC pulse duration. To minimize the effects of higher-order dispersion and de-polarization during propagation [18,19], we constructed a short 1250 nm nonlinear fiber cocktail, or segemented fiber, from 8 mm of HI1060 spliced to 29 mm of NL-1050-ZERO-2, referred to here as the 1250 nm nonlinear fiber. Epoxying and polishing the 1250 nm nonlinear fiber into an angled physical contact (APC) fiber connector makes a robust entrance coupling platform, where the larger mode-field diameter of the HI1060 entrance fiber acts as a bridge fiber to increase the coupling efficiency into the smaller core PCF. The splice loss at the fiber intersection is −0.67 dB. An off-axis parabolic collimator (Thorlabs RC02APC-P01) couples the seed pulses into the fiber cocktail. Pulses exit the fiber cocktail via a cleaved face of the PCF and an anti-reflection coated aspheric lens (4.5 mm effective focal length, Newport 5723-H-C) collimates the output beam. The total power transmission efficiency of pulses incident on the off-axis parabolic collimator through the output aspheric lens is 59%. An achromatic half waveplate before the entrance fiber collimator matches the seed pulse polarization orientation to the weak birefringence induced slow axis of the PCF [18, 19].
Generating pulses at 1250 nm containing 0.6 nJ and 2 nJ of energy, we are able to demonstrate both power scaling and the effects of increased energy on the pulse intensity profile. The two 1250 nm pulses are shown in Fig. 4. We rely on SHG-FROG reconstructions to determine the pulse structure and power spectrum. At 2 nJ, the nonlinear spectral phase causes a broadening in the main temporal peak and a reduction in peak power; limiting the pulse to 59 fs FWHM where the transform-limited duration is 20 fs FWHM. The main temporal peak in the 2 nJ pulse contains 81% of the total energy. At 0.6 nJ, the main temporal peak is exceptionally short at 26 fs FWHM out of a transform-limited duration of 23 fs FWHM and contains 69% of the total pulse energy.
Fig. 4 1250 nm pulses at 2 nJ pulse (top) and 0.6 nJ (bottom). (a,d) Phantom-FROG: comparison of the measured (left) and reconstructed (right) SHG-FROG traces, (b,e) spectral phase (blue) and power spectrum (black), and (c,f) the temporal profile of the bandwidth supported transform-limited pulse (blue) and the reconstructed pulse (black).
Defining the depolarization as I ⊥/(I ⊥ + I ‖), where I ‖ and I ⊥ are the intensities in the slow and fast axes, respectively, and the seed pulse is oriented along the slow axis. The 1250 nm SC is only 10% and 5% depolarized for the 2 nJ and 0.6 nJ pulses, respectively; meaning both are largely free of polarization instabilities. Of the light coupled through the dual-ZDW PCF, the conversion efficiency into short pulses near 1250 nm, including depolarization, is 60%; meaning the greater depolarization in the higher energy pulse is actually offset by a larger power conversion efficiency into the 1250 nm spectral region. The total conversion efficiency from the 1075 nm seed pulse to 1250 nm pulses then, including coupling losses and depolarization, is 35%.
The presence of the satellite pulses at both energies is unfortunate, as they are caused solely by nonlinear spectral phase and fiber dispersion. It seems to us, metrics like the FWHM do not fully capture the complexity of the pulses in Fig. 4(c) and (f). To try and estimate the effectiveness of a structured temporal profile, e.g. as an excitation source for multiphoton microscopy, we think there is some merit to comparing them to a square pulse. The temporal duration of a square pulse is quite simple: Δτ square = E/P 0, where E is the energy in the pulse and P 0 is still the maximum peak power. For comparison, a true Gaussian pulse with an intensity FWHM duration of 50 fs has a square pulse duration of 53 fs, only a factor of 1.06× longer. The square pulse comparison is a straightforward means of quantifying pulse structure into the common language of pulse duration. Applying this to our 2 nJ and 0.6 nJ pulses, the effective square pulse durations are 72 fs and 44 fs, respectively.
There are a few improvements, not only our conversion system, but also to the PCF which would enable scaling to still higher 1250 nm pulse energies. We were not able to exceed 2 nJ pulses due to an increased feedback into the broadening fiber stage and amplifier. A Faraday-isolator would extinguish this feedback but broadband Faraday-isolators at 1075 nm are not yet a common commercial component. Alternatively, angle cleaving the exit face of the PCF fiber would also diminish the back reflected parasitic light but is outside of our current fabrication capabilities. Another improvement to the system would be to hermetically enclose the PCF exit face, as the system did experience a degradation in output beam quality and transmission over time. Potentially the most dramatic improvement however, would involve having the NL-1050-Zero-2 PCF re-designed incorporating polarization maintaining structures. As studies have shown, polarization maintenance within nonlinear fibers is of the utmost importance when scaling to higher pulse energies [18, 19].
2.3. Demonstration of 3-photon microscopy at 1250nm
We built a stage-scanning microscope, illustrated in Fig. 5(a), as opposed to a laser scanning microscope, in an effort to preserve pulse energy. Since the stages move, the incident beam is always co-axially directed into the focusing objective. We can therefore use a highly transmissive, aspheric lenses and still produce a diffraction-limited focus at each pixel in the two-dimensional (2D) section scanned by the stages. Epi-fluorescent light is detected in the traditional manner: with a longpass dichroic mirror (Semrock FF775-Di01-25x36) and filters isolating the fluorescence light from residual fundamental and harmonic frequencies (Thorlabs FGS900-A and Thorlabs MF525-39).
Fig. 5 (a) Setup of the 3-photon epi-detected stage-scanning microscope (top) and an example of the position and velocity curves of both x (black) and y (blue) stages used to acquire sectioned 2D images (bottom). The shaded areas indicate regions of constant velocity where pixel data is captured. PMT: photomultiplier tube, F: filters, and DM: dichroic mirror. (b) An integrated image stack of fluorescein-dyed lens tissue (left) and two 2D sections separated in the z-dimension by 30 micron (right).
To minimize acquisition time, one axis of the 2D scan is run continuously at constant velocity for each one-dimensional row scan. The continuously acquired voltage samples of each row are separated into discrete pixels based on sample averaging and the sample acquisition rate. The bottom of Fig. 5(a) shows an example of the velocity and position curves for the x- and y-axis stages, where the gray-shaded areas indicate the regions of active pixel acquisition.
The image of fluorescein-dyed lens tissue, shown in Fig. 5(b), is from 21 2D planes of 495×495 pixels each separated by 3 microns in the z dimension. The 0.6 nJ 1250 nm pulses were the excitation source for the image. Two 2D planes are shown on the right side of Fig. 5(b), where the two planes are 30 microns apart and show well resolved dyed fibers at different layers. The data acquisition card (National Instruments NI USB-6363) acquired analog data from the photomultiplier tube via a current preamplifier (Stanford Research Systems SR570) at 2 MSa/s; averaging every 4000 samples to compose the value of a pixel. The constant x-axis velocity during data acquisition was 0.5 mm/s, and the acquisition time per 2D section was 9.5 minutes. The focusing objective is an aspheric lens (Newport 5721-H-C) with a 1000–1600 nm anti-reflection coating, 2.8 mm effective focal length, and 0.54 numerical aperture (NA). In future work, increasing the NA of the focusing objective should result in higher signal levels, while adding additional color filters and PMT’s will extend the microscope into the second-and third-harmonic generation modalities.
3. Conclusion
In this article, we outlined the problems inherent in generating 1250 nm pulses from ∼ 1060 nm seed pulses by SSFS in highly nonlinear PCF. The problem is two-fold, one is the low energies required (< 1 nJ) of ∼ 100 fs seed pulses to form small numbered higher-order solitons. The second is that even for seed pulses at 25 fs which can generate shifted solitons with > 1 nJ of energy, the shifted solitons over-shoot the target spectral region. As an alternative approach, we have built a custom ultrafast laser system to seed a 1250 nm nonlinear fiber cocktail. The dual-ZDW nonlinear fiber does not have the same energy constraints as a single ZDW PCF, nor does it have the same spectral over-shoot problem as it is not a soliton or Raman driven nonlinear conversion process.
In a 37 mm 1250 nm nonlinear fiber cocktail, we generate short pulses at 1250 nm with 0.6 and 2 nJ of energy. The 1250 nm pulse duration out of the fiber depends tightly on the accrued nonlinear spectral phase, with the 0.6 nJ pulse having a FWHM of 26 fs and the shortest reported pulse duration in this spectral band from a nonlinear fiber derived source (to the best of our knowledge). The more energetic pulse has a diminished pulse duration, at 59 fs FWHM. With some improvements to the system, we should be able to continue increasing the pulse energy at 1250 nm to potentially 4.2 nJ with the current seed source.
To demonstrate the utility of our 1250 nm pulse source we took images of fluorescein-dyed lens tissue using a purpose built three-photon excitation fluorescence microscope. The microscope employed a rapid stage-scanning system to raster a ∼ 0.25 mm2 area containing 495×495 pixels. As a preliminary result, with room to optimize the microscope side of the system, the three-photon excitation fluorescence image shows the potential of nonlinear fiber derived 1250 nm pulses to play an active role in the continuing development of multiphoton microscopy.
Acknowledgments
S.R. Domingue and R.A. Bartels are grateful for support from Colorado State University and a US Department of Energy phase 1 STTR grant number DE-SC0011332.
References and links
1. M. Oheim, E. Beaurepaire, E. Chaigneau, J. Mertz, and S. Charpak, “Two-photon microscopy in brain tissue: parameters influencing the imaging depth,” J. Neurosci. Methods 111, 29–37 (2001). [CrossRef] [PubMed]
2. M. Balu, T. Baldacchini, J. Carter, T. B. Krasieva, R. Zadoyan, and B. J. Tromberg, “Effect of excitation wavelength on penetration depth in nonlinear optical microscopy of turbid media,” J. Biomed. Opt. 14, 010508 (2009). [CrossRef] [PubMed]
3. D. Kobat, M. E. Durst, N. Nishimura, A. W. Wong, C. B. Schaffer, and C. Xu, “Deep tissue multiphoton microscopy using longer wavelength excitation,” Opt. Express 17, 13354–13364 (2009). [CrossRef] [PubMed]
4. N. G. Horton, K. Wang, D. Kobat, C. G. Clark, F. W. Wise, C. B. Schaffer, and C. Xu, “In vivo three-photon microscopy of subcortical structures within an intact mouse brain,” Nat. Photonics 7, 1–5 (2013). [CrossRef]
5. S. W. Chu, I. H. Chen, T. M. Liu, P. C. Chen, C. K. Sun, and B. L. Lin, “Multimodal nonlinear spectral microscopy based on a femtosecond Cr:forsterite laser,” Opt. Lett. 26, 1909–1911 (2001). [CrossRef]
6. D. G. Ouzounov, F. R. Ahmad, D. Müller, N. Venkataraman, M. T. Gallagher, M. G. Thomas, J. Silcox, K. W. Koch, and A. L. Gaeta, “Generation of megawatt optical solitons in hollow-core photonic band-gap fibers,” Science 301, 1702–1704 (2003). [CrossRef] [PubMed]
7. J. C. Knight, J. Arriaga, T. A. Birks, A. Ortigosa-Blanch, W. J. Wadsworth, and P. S. J. Russell, “Anomalous dispersion in photonic crystal fiber,” IEEE Photon. Technol. Lett. 12, 807–809 (2000). [CrossRef]
8. J. P. Gordon, “Theory of the soliton self-frequency shift,” Opt. Lett. 11, 662–664 (1986). [CrossRef] [PubMed]
9. J. M. Dudley and S. Coen, “Supercontinuum generation in photonic crystal fiber,” Rev. Mod. Phys. 78, 1135–1184 (2006). [CrossRef]
10. K. Blow and D. Wood, “Theoretical description of transient stimulated Raman scattering in optical fibers,” IEEE J. Quantum Electron. 25, 2665–2673 (1989). [CrossRef]
11. A. Aguirre, N. Nishizawa, J. Fujimoto, W. Seitz, M. Lederer, and D. Kopf, “Continuum generation in a novel photonic crystal fiber for ultrahigh resolution optical coherence tomography at 800 nm and 1300 nm,” Opt. Express 14, 1145–1160 (2006). [CrossRef] [PubMed]
12. H. Wang and A. M. Rollins, “Optimization of dual-band continuum light source for ultrahigh-resolution optical coherence tomography,” Appl. Opt. 46, 1787–1794 (2007). [CrossRef] [PubMed]
13. J. van Howe, J. H. Lee, S. Zhou, F. Wise, C. Xu, S. Ramachandran, S. Ghalmi, and M. F. Yan, “Demonstration of soliton self-frequency shift below 1300 nm in higher-order mode, solid silica-based fiber,” Opt. Lett. 32, 340–342 (2007). [CrossRef] [PubMed]
14. J. H. Lee, J. van Howe, C. Xu, S. Ramachandran, S. Ghalmi, and M. F. Yan, “Generation of femtosecond pulses at 1350 nm by Cerenkov radiation in higher-order-mode fiber,” Opt. Lett. 32, 1053–1055 (2007). [CrossRef] [PubMed]
15. S. R. Domingue and R. A. Bartels, “Nonlinear fiber amplifier with tunable transform limited pulse duration from a few 100 to sub-100-fs at watt-level powers,” Opt. Lett. 39, 359–362 (2014). [CrossRef] [PubMed]
16. S. R. Domingue and R. A. Bartels, “High-energy, sub-20 fs, nearly transform-limited pulse at 1065 nm enabled by a flat field ultrafast pulse shaper,” publication pending.
17. D. J. Kane, “Principal components generalized projections: a review,” J. Opt. Soc. Am. B. 25, A120–A132 (2008). [CrossRef]
18. S. R. Domingue and R. A. Bartels, “Overcoming temporal polarization instabilities from the latent birefringence in all-normal dispersion, wave-breaking-extended nonlinear fiber supercontinuum generation,” Opt. Express 21, 13305–13321 (2013). [CrossRef] [PubMed]
19. H. Tu, Y. Liu, X. Liu, D. Turchinovich, J. Lægsgaard, and S. A. Boppart, “Nonlinear polarization dynamics in a weakly birefringent all-normal dispersion photonic crystal fiber: toward a practical coherent fiber supercontinuum laser,” Opt. Express 20, 1113–1128 (2012). [CrossRef] [PubMed]
