We experimentally compared three methods of continuum generation that can provide wavelength conversion through anti-Stokes radiation (ASR). The three methods are: dispersion micro-managed (DMM) holey fiber, tapered fiber and long holey fiber with constant core diameter. We investigated the spectral shape and the amplitude fluctuations due to the broadband noise that is amplified during the nonlinear conversion process. The results show that the DMM method can shift wavelengths with up to 20 dB lower broadband noise compared with the other methods, at the same time with controllable wavelength shift and bandwidth, and without spectral substructure.
©2005 Optical Society of America
Through photonic crystal fibers (PCF) and tapered fibers, supercontinuum (SC) has been successfully generated directly using laser oscillators [1, 2]. This spectrally broad continuum source is potentially useful in many applications such as optical coherence tomography (OCT), optical frequency metrology, fluorescence microscopy, coherent anti-Stokes Raman scattering (CARS) microscopy and two photon fluorescence microscopy [3–7]. For some experiments [5, 7], however, the amplitude fluctuations of conventional continuum sources can limit the experimental accuracy. Previous studies of SC generation have shown that the SC generation process is very sensitive to the quantum noise, technical noise, and specific parameters such as the input power, input wavelength, time duration and chirp of the input laser pulses [8–12].
An anti-Stokes light source sliced from a stable continuum would be required to improve experimental results [5, 7]. In Ref. , we experimentally demonstrated the generation of highly coherent anti-Stokes radiation (ASR) with low amplitude noise by introducing dispersion micro-management (DMM) in holey fibers (one type of PCF). Motivated by these applications needs and previous investigations, in this paper, we focus on comparing the DMM technique with other methods that are being used most commonly in continuum generation, including tapered fibers and long holey fiber (HF) methods. Specifically, we compare the ASR part of the continuum spectrum and the power fluctuations induced by broadband noise. Unlike technical noise that can be suppressed by feedback techniques, broadband noise is induced by laser quantum noise and therefore sets the limit to the pulse stability [8–10]. For applications in microscopy and frequency metrology, our results should prove to be useful in generation of light at specific desired wavelengths, when starting with a fixed-wavelength source.
A common approach to wavelength conversion is to couple as much laser energy into nonlinear fibers as possible to generate a white light continuum, then spectrally slice part of the continuum and use it as the light source for microscopy setup. However, it is likely that the selected continuum contains large amplitude fluctuations (noise), which may not be suitable for some applications. In this paper, we first give brief theoretical analysis for the DMM technique, and then experimentally show that DMM holey fibers (DMM HFs) can generate low noise and high coherence ASR light with smooth spectral shapes. In comparison with the other two methods such as tapered fibers and long HFs, we find that the ASR shifted by DMM technique from the input laser wavelength to the specific wavelength demonstrates lower broadband noise. Moreover, the DMM method can be used to control the center wavelength of the ASR with scalable bandwidth, even with a fixed λinput laser source.
2. Theoretical Analysis and Fiber Preparations
Our goal is to shift the wavelength from the fixed λinput to a specific ASR region with three different methods, and to investigate the corresponding ASR noise behavior. As our first example, the λinput is chosen to be 920 nm, and the specific ASR wavelength is 580 nm. In a previous publication , we first introduced the DMM technique in continuum generation. This technique’s design flexibility comes from the fact that we can control the tapering ratio of the input and output diameter of the holey fiber, thus tuning the resonant ASR wavelength with respect to the λinput. It is known that the continuum generation can be well described as a soliton fission process (SFP) , and the ASR can be generated following the phase matching condition in equation 1 .
Here, β is the propagation constant, ωs and ω are the input soliton and the ASR frequencies, respectively; vg is the soliton group velocity; γ is the fiber’s nonlinear coefficient and Ps is the soliton peak power. Similarly, the continuum generated through a DMM HF can be described as a SFP, but with longitudinally varying group velocity, dispersion, nonlinearity and phase matching conditions. As a result of the longitudinal variation of the fiber dispersion, the ASR is generated as it experiences a “sliding” phase matching condition. Therefore, we see that the DMM technique can broaden the bandwidth of the ASR and produce a specific desired ASR center wavelength, given certain HF parameters and a fixed λinput.
There are several methods to change the fiber longitudinal profiles [15, 16]. We fabricated DMM HFs using the CO2 laser tapering technique with computer control. We carefully control the temperature, pulling speed and scanning range for the DMM tapering process without introducing any hole-collapsing. Using equation 1 (Fig. 1), we designed and fabricated a DMM HF consisting of a 10 mm tapering region with core diameter starting from 3.3 μm and ending with 2.6 μm to generate a strong 580 nm component (see schematic inset in Fig. 2(c)). The SFP is expected to happen at the end section of the DMM HF with core diameter between 2.6μm to 2.8 μm, giving a sliding ASR generation centered ~ 580 nm.
We prepared the tapered fiber also by the CO2 tapering technique. A single mode fiber with 125 μm outside diameter is tapered down to 2.7 μm outside diameter with 6 cm waist length, with overall coupling loss ~ 2.5 dB (inset in Fig. 2(b)). For the HF, we purchased NL-2.6-825 HF (2.6 ± 0.1 μm core diameter) from a series of HFs manufactured by Crystal Fibre A/S, and use a 0.8 m length of fiber in the experiment (inset in Fig. 2(a)).
3. Experimental Results and Analysis
The experimental conditions are shown in Fig. 2. Left column of Fig. 2 (a–c) shows the continuum spectra generated through these three methods: 0.8 meter HF with constant core diameter, the tapered fiber with 6 cm constant core diameter and the DMM HF with total length 2.3 cm, separately. 100 fs pulses centered at 920 nm from a Ti: Sapphire laser are coupled into the three fibers with similar coupled power ~ 125 mw. The output spectrum is measured with an Ando 6315 optical spectrum analyzer that covers from 350 to 1700 nm.
From Fig. 2, it is clear that all three continuum spectra have successfully covered the specific spectrum range that is centered ~ 580 nm. With a filter centered ~ 580 nm (Schott Veril linear variable interference filter), we can slice the three spectra to get 580 nm components and attempt to use them in applications such as microscopy and frequency metrology etc., however, detailed characterizations of the sliced spectra reveal that these three sources can have very different performance in terms of spectral shape and broadband noise induced amplitude fluctuations.
First, we will study the sliced spectra shown in Fig. 3, which correspond to the shadowed region in Fig. 2. The spectral components generated by the HF and the tapered fiber show significant substructure, while that generated by the DMM HF has a well-behaved shape which is similar to the filter band-pass window (the bandwidth of the ASR generated by this DMM HF is ~ 50nm, and the filter width is ~25 nm). This well-behaved spectrum shape is required for many applications such as OCT, fluorescence microscopy, CARS microscopy etc., since it can preclude artifacts caused by spectral substructures and greatly simplify theoretical modeling of the input source and provide more reproducible measurements.
Next, we characterize the amplitude fluctuations of the spectrally sliced sources. A theoretical and experimental investigation of the generation of broadband noise in HF is described in Ref. . Here, since we typically generated 1–10 mw of spectrally sliced light, we used a tunable attenuator to adjust the received power of these three signals to be equal. In this way, we can guarantee that all three signals have the same shot noise, and this will not affect the broadband noise measurement. About 100 μw of power is coupled into a fiber and sent to a New Focus 1GHz visible photoreceiver. The broadband noise is measured using the Agilent 8564E RF analyzer. Fig. 4 shows the broadband noise measured with a 10 kHz resolution bandwidth. The black line represents the noise background of the detector without input light. The broadband noise of the filtered ASR generated through the DMM method is ~ 2 dB above the background, while the noises of the filtered out ASR generated through the HF and tapered fiber are ~ 15 dB above the background. Using this measurement, we calculated the relative intensity noise (RIN) for the three methods (DMM, HF, tapered fiber) to be: -124 dBc/Hz, -111 dBc/Hz and -111 dBc/Hz, respectively.
Next, we designed and fabricated two more DMM HFs using the same 3.3 μm core diameter HF to produce shifts from fixed λinput 920 nm to 535 nm and 620 nm wavelengths, respectively. According to the Fig. 1 theoretical analysis, the DMM HF that is for 535 nm ASR generation consists of 10 mm tapered region from 3.3 μm to 2.4 μm, with 19 mm total length; and the DMM HF for 620 nm ASR generation consists of 10 mm tapered region from 3.3 μm to 2.8 μm, with 24 mm total length. With these two DMM HFs, we successfully produce a 535 nm ASR with ~ 45 nm bandwidth and a 620 nm ASR with ~ 57 nm bandwidth, respectively.
After spectrally slicing out the 535 nm and 620 nm spectral components, we compared the broadband noise with the ASR counterparts generated through the HF fiber and the tapered fiber. The noise comparisons are shown in Fig. 5(b) and 5(d).
From Fig. 5(b), for the 535 nm component, the broadband noise generated through the HF is 17 dB higher than the DMM HF, and the broadband noise generated through the tapered fiber is 10 dB higher than the DMM method. From Fig. 5(d), for the 620 nm component, the broadband noise generated through the HF and the tapered fiber is 20 dB and 9 dB higher than that of the DMM HF, respectively.
From the three experimental comparisons, it is clearly seen that the DMM HF method provides lower noise continuum generation with cleaner spectral components compared with the other two methods. Our explanation for this is as follows. First, for the DMM HF, the SFP happens close to the end of the DMM HF, when the visible ASR is optimally shaped both in the time and frequency domains before additional “messy fission collisions” take place. In contrast, for the tapered fiber and the long HF, the SFP will continue and the ASR is generated not only by the first fission collision process, but also by successive fissions [14, 17]. These multi-fission processes can be clearly observed in the continuum spectrum generated through the long HF in Fig. 2(a), where there are five extinguished soliton peaks contained in the infrared part of the spectrum. Therefore, the ASR part in the HF or tapered fiber experiences more four-wave mixing, cross phase modulation and modulation instability, and the resultant spectrum is affected by these successive processes, causing significant substructures. Furthermore, these successive nonlinear processes amplify the broadband noise which is seeded from the input shot noise of the laser pulses. Therefore, it is not difficult for us to understand that the DMM method, compared with the other two, suffers less from the noise amplification process.
To gain more insight into the noise property of the ASR light generated with the DMM technique, we carried out the noise measurements by selecting a 25 nm portion of the ASR. For the 535 nm-centered ASR, we spectrally sliced the components centered at 519 nm and 558 nm and measured the corresponding broadband noise. Compared with the peak component located at 535nm, these two components show ~ 2 dB higher broadband noise, and yet very smooth spectra. Similarly, for the 580 nm-centered ASR, we spectrally sliced components centered at 556 nm and 606 nm, and they exhibit ~ 4 dB higher broadband noise compared with the centered 580 nm component. The noise distribution of the broadband ASR is of interest to our future research.
In addition to the low noise ASR property, due to the DMM phase-matching sliding effects, the resultant ASR exhibits bandwidth (in the frequency domain) which can be several times larger than the input pulse bandwidth. These scalable bandwidths offer another degree of freedom for microscopy or frequency metrology applications. At last, the design flexibility of DMM offers us the freedom to shift the ASR center-wavelength even when the input wavelength is fixed, as we demonstrated here: shifting from 920 nm to 535, 580 and 620 nm separately by simply changing the DMM design scheme. This advantage means a tunable laser source is not required as the input. We expect that our results will be useful in furthering the use of femtosecond technology in many applications areas. With regard to noise characterization on the Stokes side of the input wavelength (for DMM HF), we have measured the noise of the spectral component centered ~ 1065 nm with 13nm bandwidth. The 1065 nm component is located at an octave bandwidth away from the 535 nm ASR, and exhibits low noise property with ~2 dB broadband noise above the background. The low noise properties for the ASR and the Stokes component indicate that the DMM technique can be well applied to the carrier envelop phase (CEP) stabilization. A more complete description of the application of this technique to CEP stabilization will be reported elsewhere .
In summary, three methods of continuum generation are experimentally compared. We observed that with the same coupled power, the DMM method can produce lower broadband noise ASRs with better spectral shape compared with the tapered fiber and long HF methods. At the same time the design flexibility of the DMM provides a convenient approach to generate ASRs with scalable center-wavelengths, high spectral coherence and scalable bandwidths even with a fixed input wavelength laser as a source.
This project is supported by a grant from NYSTAR. We acknowledge assistance from Yujun Deng and Parama Pal.
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 800nm,” Opt. Lett. 25, 25–27 (2000) [CrossRef]
2. T. A. Birks, W. J. Wadsworth, and P. St. J. Russell, “Supercontinuum generation in tapered fibers,” Opt. Lett. 25, 1415–1417 (2000) [CrossRef]
3. I. Hartl, X. D. Li, C. Chudoba, R. K. Ghanta, T. H. Ko, J. G. Fujimoto, J. K. Ranka, and R. S. Windeler, “Ultrahigh-resolution optical coherence tomography using continuum generation in an air silica microstructure optical fiber,” Opt. Lett , 26, 608–610 (2001) [CrossRef]
4. D. J. Jones, S. A. Diddams, J. K. Ranka, A. Stentz, R. S. Windeler, J. L. Hall, and S. T. Cundiff, “Carrier-Envelope Phase Control of Femtosecond Mode-Locked Lasers and Direct Optical Frequency Synthesis,” Science , 288, 635–639 (2000) [CrossRef] [PubMed]
5. G. McConnell, “Confocal laser scanning fluorescence microscopy with a visible continuum source,” Opt. Express , 12, 2844–2850 (2004), http://www.opticsinfobase.org/ViewMedia.cfm?id=80302&seq=0 [CrossRef] [PubMed]
6. H. N. Paulsen, K. M. Hilligse, J. Thgersen, S. R. Keiding, and J. J. Larsen, “Coherent anti-Stokes Raman scattering microscopy with a photonic crystal fiber based light source,” Opt. Lett. 28, 1123–1125 (2003) [CrossRef] [PubMed]
7. J. A. Palero, V. O. Boer, J. C. Vijverberg, and H. C. Gerritsen, “Short-wavelength two-photon excitation fluorescence microscopy of tryptophan with a photonic crystal fiber based light source,” Opt. Express , 13, 5363–5368 (2005), http://www.opticsinfobase.org/ViewMedia.cfm?id=84901&seq=0 [CrossRef] [PubMed]
8. K. L. Corwin, N. R. Newbury, J. M. Dudley, S. Coen, S. A. Diddams, K. Webber, and R. S. Windeler, “Fundamental noise limitations to supercontinuum generation in microstructure fiber,” Phys. Rev. Lett , 90, 113904-1(2003) [CrossRef] [PubMed]
9. J. M. Dudley and S. Coen, “Coherence properties of supercontinuum spectra generated in photonic crystal and tapered optical fibers,” Opt. Lett. 27, 1180–1182 (2002) [CrossRef]
10. T. M. Fortier, J. Ye, S. T. Cundiff, and R. S. Windeler, “Nonlinear phase noise generated in air-silica microstructure fiber and its effects on carrier-envelope phase,” Opt. Lett. 27, 445–447 (2002) [CrossRef]
12. A. L. Gaeta, “Nonlinear propagation and continuum generation in microstructured optical fibers,” Opt. Lett. 27, 924–926 (2002). [CrossRef]
15. G. Kakarantzas, T. E. Dimmick, T. A. Birks, R. Le Roux, and P. St. J. Russell, “Miniature all-fiber devices based on CO2 laser microstructuring of tapered fibers,” Opt. Lett. 26, 1137–1139 (2001) [CrossRef]
16. E. C. Mägi, P. Steinvurzel, and B. J. Eggleton, “Tapered photonic crystal fibers,” Opt. Express , 12, 776–784 (2004), http://www.opticsinfobase.org/ViewMedia.cfm?id=79192&seq=0 [CrossRef] [PubMed]
17. I. Cristiani, R. Tediosi, L. Tartara, and V. Degiorgio, “Dispersive wave generation by solitons in microstructured optical fibers,” Opt. Express , 12, 124–135 (2004); http://www.opticsinfobase.org/ViewMedia.cfm?id=78438&seq=0 [CrossRef] [PubMed]
18. F. Lu and W. H. Knox, J. Opt. Soc. Am. B (to be submitted).