An L band superfluorescent fiber source (SFS) with output power of 0.94W is presented, under 4.4W 976nm pump power. The optical conversion efficiency is about 21%. The spectrum covers the broad wavelength range from 1560nm to 1615nm. The high power L band SFS is constructed by a low power L band amplified spontaneous emission (ASE) seed source and a high power erbium-ytterbium co-doped fiber (EYDF) amplifier in double pass forward pumping configuration.
©2005 Optical Society of America
Superfluorescent fiber source (SFS) located in the low loss communication window has attracted much attention in the last decade [1–13], for its usefulness as an incoherent broadband light source in various areas, such as low cost spectrum sliced wavelength division multiplexing (WDM) local access networks  and fiber optical sensors especially fiber optic gyroscopes [3,4]. Many works have focused on the C band in the first few years [3–7] to enhance its output power, spectral bandwidth and wavelength stability as well. Recent demands for the immediate expansion of the fiber optical communication window have led to the development of L band sources. SFS in the L band then became an interesting topic [8–13]. Although the output power of the C band SFS exceeded 1W several years ago , up to now, the power level of the L band SFS is limited to be less than 100mW.
In an ultra-high power SFS, Rayleigh backscattering should be carefully considered because it can cause lasing [14,15]. Minelly et al.  have demonstrated that lasing can be avoided by seeding a high power amplifier with a saturating signal from a low power amplified spontaneous emission (ASE) source. By this method, it is possible to extract almost as much power as can be obtained from a fiber laser under similar excitation. Based on this concept, Gray et al  constructed a tri-stage erbium-ytterbium co-doped fiber (EYDF) SFS with an output power of larger than 1W near 1535nm and a bandwidth of 4nm. Chen et al  employed a backward pumped high power erbium-ytterbium co-doped fiber amplifier (EYDFA) to amplify lights from a low power erbium-doped SFS. The dual stage SFS provided output power of larger than 1W near 1561nm with a bandwidth of 8nm. However, all the reported ones are located in the C band with a sharp structure in the spectrum.
In this paper, an L band SFS with an output power of nearly 1W and a spectrum covering the broad wavelength range from 1560nm to 1615nm is presented. The high power L band SFS is constructed by a low power L band ASE seed source and a high power EYDFA in double pass forward pumping configuration. Compared with the reported watt-level SFS in the C band [6,7], the L band SFS presented here risks less danger in resonant lasing. Furthermore, it possesses potential possibility of achieving ultra-flattened broadband spectrum and high output power simultaneously. To the best of our knowledge, this kind of SFS has not been reported up to now.
2. Experimental setup
The proposed L band SFS is schematically shown in Fig. 1. Left of the cross is an L band ASE seed source which is actually an erbium doped SFS constructed in double pass forward pumping configuration, yet with an extra section of unpumped erbium doped fiber (EDF) between the WDM and the fiber mirror (R1) to obtain L band output with relatively shorter EDF. Right of the cross is a double pass forward pumped EYDFA that acts as an L band power amplifier. An optical circulator is utilized to realize the double pass amplification, which makes it possible for us to realize the L band amplification with only 15m EYDF. The EYDF is double cladding pumped by high power laser diode array (LDA) through a tapered fiber bundle (TFB).
The side-pumping scheme through TFB provides flexibility of constructing laser sources and amplifiers in various kinds of configuration, while the LDA pumping scheme ensures the possibility of achieving high power operation with several low power laser diodes that need simpler fabrication technology as compared with high power ones. The LDA pumping scheme through inner cladding also partially releases the danger of permanent destruction caused by thermal effect [17–19], due to the intensity decrease by dividing the pumping lights into several parts and by enlarging the interaction area.
The EYDF has a star shaped inner cladding with numerical aperture of ~0.45 and a single mode circle core with numerical aperture of ~0.15. The diameter of the outer cladding is 250µm. The single mode core and the proper outer cladding diameter allow the EYDF to be directly fusion spliced with standard single mode fiber. The peak erbium absorption is 39dB/m at 1535nm and the peak ytterbium absorption is 389dB/m at 915nm. The length of the EYDF is 15m. The TFB has six multimode ports for launching the pumping lights and one single-mode port to guide the signal light at one side, and a double cladding port to match the EYDF at the other side. The insertion loss of the TFB is 0.2dB at the signal wavelength and 0.4dB at the pumping wavelength. The numerical apertures of the fibers of the LDA, the TFB and the EYDF are carefully chosen to avoid additional loss from mismatch. The insertion loss and isolation of the optical circulator from port 1 to port 2 are ~0.5dB and ~60dB, respectively. The values from port 2 to port 3 are ~0.6dB and ~60dB, respectively. All the components of the SFS are polarization insensitive.
3. Experimental results
After optimizing the EDF lengths, a maximum output power of about 2.8mW is obtained from the seed source, with optical spectra showing in Fig. 2. The spectral shape of the seed source changes with increasing power within the L band.
The output power versus pump power (PLDA) of the high power L band SFS under Pseed=2.8mW is shown in Fig. 3. As can be seen, the output power of the SFS increases almost linearly with PLDA, and reaches 0.94W under the pump power of 4.4W, giving an optical conversion efficiency of about 21%. Figure 4 shows the output spectra of the SFS measured at different pump power levels. As is apparent from the figure, the spectral shape changes with an increasing pump power and rises more rapidly in the short wavelength range than in the long wavelength range. The output spectra at high pump power levels cover the broad wavelength range from 1560nm to 1615nm. Note that the spectral shape of the high power SFS does not agree with that of the seed source, which will be discussed in Section 4.
When Pseed decreases from 2.8mW to 0mW, the output power of the SFS decreases about 15%, and lasing light emerges when PLDA goes above 1.5W, hence only 180mW superfluorescence can be obtained then. That is to say, the seed source plays an important role in the proposed SFS, although the power of itself is only 2.8mW. Figure 5 illustrates the variation of the output spectra of the SFS with an increasing Pseed under PLDA=4.3W. Curve 1 to 4 are measured under Pseed=0, 0.3, 1.2 and 2.8mW, respectively. Curve 1 actually corresponds to the ASE spectrum of the pumped EYDF. It is clearly seen from curve 1 and curve 2 that laser light emerges around 1565nm when Pseed is low, while curve 3 and curve 4 indicate that the lasing effect is suppressed when Pseed is relatively high (>1mW). In a word, a high power seed source has the function of preventing the SFS from lasing. Also, as Pseed increases, the spectrum of the SFS shifts toward longer wavelengths. The spectral intensity decreases in short wavelengths and increases in long wavelengths, which is helpful to realize the L band output.
The Pseed=0mW curve in Fig. 5 shows that, in addition to the lasing spikes near 1565nm, there are spikes near 1578nm and 1614nm. In the experiment, these additional spikes do not always exist. They appear and disappear randomly. Also, the operating wavelengths of these additional spikes always change. These spikes are probably caused by mode competition effect. The phenomenon that laser lights emerge from different wavelength ranges indicates the inhomogenous broadening property of the EYDF, which is consistent with our previous studies .
The utilization of the seed source is very important for the high power performance of the SFS. It not only improves the output power to a certain degree, but also has the function of preventing the SFS from lasing. The latter is particularly important for a watt-level SFS. The SFS acts as a double pass forward pumped SFS without the seed source. The output power of such an SFS can not be very high, because of the danger of resonant lasing. As described in the above, the output power of the SFS can not exceed 200mW without the seed source, otherwise laser light will emerge. The gain saturation effect in the EYDFA can be utilized to explain the lasing eliminating function of the seed source.
The laser feedback in the proposed SFS is provided by the fiber mirror (R2) at one side and by the Rayleigh backscattering at the other side. The laser threshold condition GαL(NA)2/(4n 2)=1 must be satisfied to switch on the laser. Here, G means the effective mean gain, α is the fiber loss coefficient, L, NA, and n represent the length, the numerical aperture and the core refractive index of the doped fiber (see Ref.  for more details). When PLDA is below 1.5W, the gain of the EYDF does not meet the request of the equation. While PLDA increases, the gain G increases too. At a certain level of PLDA (several watts), the threshold condition will be satisfied, and laser light emerges then. However, when lights from the seed source are injected into the EYDFA, the gain will decrease due to the gain saturation effect. More power from the seed source results in a lower gain. When the gain decreases to a level not satisfy the threshold condition, the lasing will be suppressed then.
Experimental results also show that, the higher PLDA is, the higher Pseed will be needed to prevent the SFS from lasing. However, it is well known that the laser cavity defined by Rayleigh backscattering does not exhibit a definite cavity length. There is no beating between cavity modes when laser light emerges. RF spectrum analyzer can not be used to justify the existence of laser light. It can be only judged by observing if there are sharp features in the output spectra. What is more, the lasing threshold of this SFS always changes in a small range, probably because of random disturbance that affects the intensity of the Rayleigh backscattering lights. It is hard to measure an exact threshold value.
Compared with the dual stage SFS in the C band, the SFS in the L band proposed here risks less danger in resonant lasing. The power of the seed source needed to prevent the lasing effect is only about 1mW in the watt-level L band SFS. However, the corresponding value needed in the C band SFS to prevent the lasing effect is at least 12mW . The higher gain of the EYDF in the C band than in the L band is one of the reasons. Another important reason is the relatively more flattened spectrum of the L band SFS, which means relatively lower gain values that cover a broader wavelength range. The power distributes on the wavelength range rather than concentrates on a point like the C band SFS in  and . So the risk of resonant lasing is less in this L band SFS than in the reported C band ones, although there is a fiber mirror (R2) in this L band SFS at one side to provide the feedback.
The spectral shape of the high power SFS does not agree with that of the seed source, mainly due to the directly guided out ASE component of the EYDFA and the non-uniform amplification coefficient at different wavelengths of the EYDFA. The total output of the SFS includes two components: the amplified seed source and the residual ASE of the EYDF. The proportion of the two components decides the spectral shape of the SFS. The spectrum of the SFS shifts toward longer wavelengths with an increasing Pseed, probably due to the in band amplification of the EYDFA which decreases the ASE component. The spectra of the seed source are within the L band as shown in Fig. 2. When lights from the seed source pass through the EYDFA, they are amplified and at the same time consume lots of reversed population in the EYDF. Thus the fraction of the ASE component will decrease and that of the amplified seed source component will increase. The former covers the whole wavelength range, while the latter only covers the L band. Therefore the spectral shape of the SFS will change as showing in Fig. 5. A larger Pseed means a higher fraction of the amplified seed source component in the output lights. However, it is difficult to measure the exact proportion of the two components in the output lights. A simulation work will help us to estimate the value, which is under consideration.
It should be noted that the power of the SFS could be further enhanced to a higher level of about several watts with larger pump power. Yet, the output power of the SFS was limited artificially less than 1W in the experiment to protect the optical circulator. A proper circulator allowing larger power is needed if higher output power is expected. An optical filter should be added to the output port if a more flattened spectrum is desired. Or a reflected type filter should be used instead of the reflector (R2) to flatten the output spectrum with less power penalty.
In conclusion, a dual stage L band SFS covering the broad wavelength range from 1560nm to 1615nm is presented in this paper. Output power of 0.94W has been achieved under 4.4W 976nm pump power. The optical conversion efficiency reaches 21%. Although the power of the seed source is only 2.8mW, it plays an important role in the proposed SFS. It improves the output power to a certain degree and has the function of preventing the SFS from lasing. The latter is particularly important for a watt-level SFS. The output power of the SFS can be further enhanced with a proper optical circulator and higher pump power. A reflected type filter is recommended to take the place of the reflector (R2) for flattening the output spectrum with less power penalty.
This work was supported by the High Technology Research and Development Project of China under Grant No 2003AA312100, the Key Project of the China National Natural Science Foundation under Grant No 60137010, the Tianjin Key Lab of Photonics Material and Technology for Imformation Science, and the Research Funding for the undergraduate students of the Nankai University. The authors would like to thank Prof. Zhai Hong-Chen for his assistance in English writing.
References and links
1. Jung-Hee Han, Sun-Jong Kim, and Jea-Seung Lee, “Transmission of 4×2.5-Gbs spectrum-sliced incoherent light channels over 240 km of dispersion-shifted fiber with 200-GHz channel spacing,” IEEE Photon. Technol. Lett. 11, 901–903(1999). [CrossRef]
2. Sheng-Ping Chen, Ke-Cheng Lu, Yi-Gang Li, Ming Feng, and Jia-Fang Li, “Characteristics comparison of backward and forward pumped spectrum pre-sliced multi-wavelength fibre sources,” J. Opt. A: Pure Appl. Opt. 6, 971–976 (2004). [CrossRef]
3. Lon A. Wang and C. D. Su, “Modeling of a double-pass backward Er-doped superfluorescent fiber source for fiber-optic gyroscope applications,” J. Lightwave Technol. 17, 2307–2315 (1999). [CrossRef]
4. Paul F. Wysocki, M. J. F. Digonnet, B. Y. Kim, and H. J. Shaw, “Characteristics of Erbium-Doped Superfluorescent Fiber Sources for Interferometric Sensor Applications”, J. Lightwave Technol. 12, 550–567 (1994). [CrossRef]
5. L. A. Wang and C. D. Chen, “Characteristics Comparison of Er-Doped Double-Pass Superfluorescent Fiber Sources Pumped Near 980 nm,” IEEE Photon. Technol. Lett. 9, 446–448 (1997). [CrossRef]
6. S. Gray, J. D. Minelly, A. B. Grudinin, and J. E. Caplen, “1 Watt Er/Yb singlemode superfluorescnet optical fibre source,” Electron. Lett. 33, 1382–1383 (1997). [CrossRef]
7. Sheng-Ping Chen et al, Institute of Physics, Nankai University, Tianjin 300071, P. R. China, are preparing a manuscript to be called “A dual stage superfluorescent fiber source with 1.16W output power centered at 1561nm.”
8. Ju Han Lee, Uh-Chan Ryu, and Namkyoo Park, “Passive erbium-dopedfiber seed photon generator for high-powerEr3+-doped fiberfluorescent sources with an 80-nmbandwidth,” Opt. Lett. 24, 279–281 (1999). [CrossRef]
9. W. C. Huang, P. K. A. Wai, H. Y. Tam, X. Y. Dong, Hai Ming, and J. P. Xie, “One-stage erbium ASE source with 80 nm bandwidth and low ripples,” Electron. Lett. 38, 956–957 (2002). [CrossRef]
10. H. Chen and G.W. Schinn, “Hybrid broadband superfluorescent fiber source consisting of both thulium-doped fiber and erbium-doped fiber,” Opt. Commun. 229, 141–146 (2004). [CrossRef]
11. R. P. Espindola, G. Ales, J. Park, and T. A. Strasser, “80 nm spectrally flattened, high power erbium amplified spontaneous emission fibre source,” Electron. Lett. 36, 1263–1265 (2000). [CrossRef]
12. S. C. Tsai, T. C. Tsar, P. C. Law, and Y. K. Chen, “High-power flat L-band erbium-doped fiber ASE source using dual forward-pumping scheme,” Opt. Quantum Electron. 35, 161–167 (2003). [CrossRef]
13. S. C. Tsar, T. C. Tsai, P. C. Law, and Y. K. Chen, “High Pumping-Efficiency L-Band Erbium-Doped Fiber ASE Source Using Double-Pass Bidirectional-Pumping Configuration,” IEEE Photon. Technol. Lett. 15, 197–199 (2003). [CrossRef]
14. I. N. Duling, R. P. Moeller, William K. Burns, C. A. Villarruel, L. Goldberg, Elias Snitzer, and H. Po, “Output characteristics of diode pumped fiber ASE sources,” IEEE J. Quantum Electron. 27, 995–1003 (1991). [CrossRef]
15. Amos A. Hardy and Ram Oron, “Amplified spontaneous emission and Rayleigh backscattering in strongly pumped fiber amplifiers,” J. Lightwave Technol. , 16, 1865–1873 (1998). [CrossRef]
16. J. D. Minelly, P. R. Morkel, K. P. Jedrzejewski, E. R. Taylor, J. Wang, and D. N. Payne, “Nd3+-doped singlemode fibre superfluorescent fource with 320mW output power,” Electron. Lett. 29, 1613–1614 (1993). [CrossRef]
17. Y. Jeong, J. K. Sahu, D. N. Payne, and J. Nilsson, “Ytterbium-doped large-core fibre laser with 1 kW of continuous-wave output power,” Electron. Lett. 40, 470–472 (2004) [CrossRef]
19. E. M. Dianov, I. A. Bufetov, A. A. Frolov, Y. K. Chamorovsky, G. A. Ivanov, and I. L. Vorobjev, “Fiber Fuse Effect in Microstructured Fibers” IEEE Photon. Technol. Lett. 16, 180–181 (2004) [CrossRef]
20. Sheng-Ping Chen et al, Institute of Physics, Nankai University, Tianjin 300071, P. R. China, are preparing a manuscript to be called “All fibre ring cavity watt-level erbium-ytterbium co-doped fibre laser.”