Up to 4.8 W, ∼10 MHz, 1178 nm laser is obtained by Raman amplification of a distributed feedback diode laser in standard single mode fibers pumped by an 1120 nm Yb fiber laser. More than 10% efficiency and 27 dB amplification is achieved, limited by onset of stimulated Brillouin scattering. The ratio of Raman to Brillouin gain coefficient of a fiber is defined as a figure of merit for building a narrow linewidth fiber Raman amplifier.
©2008 Optical Society of America
Fiber Raman devices are one of the remarkable advances in the area of fiber lasers and amplifiers. They have a special advantage of flexibility in wavelength, as gain is available at arbitrary wavelengths with the right pump source and laser is widely tunable for broad Raman gain spectrum in glass fiber. Therefore, fiber Raman devices are very attractive for a variety of applications, especially where the laser sources are required to be tunable and/or operate at specific wavelengths that are not easily reachable by other laser devices. One example of such applications is laser guide stars, which requires a laser at 589 nm to excite a resonant backscattering in the mesospheric Sodium D2 line [1, 2].
Efficient fiber Raman lasers have been realized [3, 4] and commercialized. Frequency doubling of near-infrared fiber Raman laser has been studied to generate yellow lasers [5, 6]. But linewidth of these lasers is too broad (86 GHz in ) for laser guide star application. A master oscillator power amplifier scheme could be the way to solve the linewidth problem.
Fiber Raman amplifiers have been used to amplify low power narrow band signal in the field of optical communications [7, 8]. However, high power (watts) narrow-band (GHz and below) laser sources utilizing fiber Raman gain have not been researched intensively up to now. Obvious difficulties in building such laser sources are stimulated Brillouin scattering (SBS), which will send the light back along the single mode fiber and limit the achievable power, and line broadening by third-order optical nonlinearity .
We show in this paper that high power narrow band fiber Raman amplifiers of several watts are feasible to an extent which allows practical uses, and still have much room for improvement. An explanation is given at the end of section 3. We had previously reported a fiber Raman amplifier system at 1178nm with linewidth of a few GHzs and frequency doubling to yellow [1, 2]. In this paper, we report our recent works on fiber Raman amplifiers with linewidth of ~10MHz. Up to 4.8 W, ~10 MHz 1178 nm laser is obtained by Raman amplification of a distributed feedback (DFB) diode laser seed in standard single mode fibers, limited by SBS. The linewidth and spectral density are improved by a few hundreds times, compared to our previous work .
2. Experimental setup
Figure 1 shows the diagram of experimental setup, which is a standard fiber Raman amplifier scheme. A commercial (Toptica Gmbh) DFB diode laser at 1178nm is used as seed, which has a linewidth < 4MHz in 5μs according to the specifications. After an optical isolator, the seed light is coupled to a single mode fiber. The coupled power is about 9mW. The Raman amplifiers are counter-propagating pumped at 1120 nm by an Yb fiber laser via a WDM wavelength combiner. The unused pump power is coupled out with WDM2 to protect the diode. Three pieces of fibers are used in the experiments, which are standard non-polarization-maintaining single mode fibers of (A) 100m Nufern 1060XP, (B) 150m Nufern 1060XP and (C) 200m Corning HI1060, respectively. We choose these standard 1 μm fibers, because they are compatible with other fiber components so that it is simple in handling. Moreover, these fibers offer less optical nonlinearity that is responsible for linewidth broadening, which we want to minimize at the same time, comparing to other Raman speciality fibers. The spectrum of output is analyzed by a Toptica FPI unit with 1GHz free spectral range, which should give a resolution of 2.5MHz. The backward propagating light from the amplifier is reflected from the two polarizers of the isolator and collected for power and spectrum measurements.
3. Results and discussion
It is expected that a narrow linewidth (few MHz) seeded fiber Raman amplifier will experience SBS at certain power level. The signal light will be frequency down-shifted and sent back to the seed. Therefore, for a given type of available fiber, the key of the experiment is to optimize the fiber length so that the amplifier reaches SBS threshold at the highest available pump power.
Figure 2 (left) shows the amplifier output power versus the pump power with fiber A, B and C. We obtain 1.1, 4.8, and 1.35W, respectively. Fig.2 (right) shows the power curves for the corresponding backward light. The backward beam is collected from the isolator between the amplifier and seed. We see a clear threshold of backward light for the fibers B and C. If we increase the pump further for fiber C, the output actually decreases as expected. The light is sent back by SBS, and the system becomes unstable. The length of fiber B is almost optimized. It reaches the threshold of backward light closed to the highest pump power.
The backward light consists of two components. One is the amplified spontaneous Raman scattering and Raman amplified signal light, which is reflected from the fiber ends and/or the splicing points or back scattering inside the fiber, which has the same wavelength as the output light. The second is SBS of the 1178 nm signal, which is amplified by Raman scattering from 1121 nm at the same time. The latter is the limiting process. These two components are easily distinguishable because they have a wavelength difference of about 15 GHz at 1178nm.
The spectrum of the backward light is measured with a Fabry-Perot spectral analyzer, which has a free spectral range larger than SBS frequency shift, fed from the isolator output. While we increase the pump power, the Raman spectral line appears earlier than the SBS line. But the SBS line increases much faster. Figure 3 (left) shows a spectrum taken at 4.25 W for amplifier with fiber B. The peak SBS shift for Nufern 1060XP fiber is found to be 14.2 GHz at 1178 nm.
To know the proportion of the SBS component in the backward light, we broaden the laser by applying a noise RF signal to the DFB laser, and check the reduction in the backward light. Figure 3 (right) shows a comparison of backward light power for amplifiers with narrow linewidth and broadened line of 200 MHz. At 200 MHz, SBS is completely suppressed at this power level. The remaining component still increases nonlinearly because it is amplified by stimulated Raman scattering.
The spectrum of the amplifier output is measured. Figure 4 (left) shows linewidth versus output power for the case of fiber B. At low power, the linewidth seems nearly constant at around 4.5 MHz. This should be due to the resolution limit of the Fabry-Perot spectral analyzer. At higher power, we can resolve a linewidth increase, which is measured up to 10MHz at 4.25W. The linewidth broadening is not very strong, because standard fibers are used in our amplifiers, which have very low optical nonlinearities. The linewidth broadening effect is so slow that SBS reaches threshold far before significant broadening of the laser linewidth. Figure 4 (right) is a spectrum taken for fiber B at 4.25 W.
Although non-polarization-maintaining fibers are used in our amplifiers, the laser output can be adjusted to have a linear polarization better than 10:1 by a λ/4 and λ/2 waveplates pair because the DFB seed laser is linearly polarized. The polarization state is stable if the system is thermalized and isolated.
A simple model for SBS limited fiber Raman amplifier is developed . SBS is taken into consideration by seeding a single photon at the end of amplifier, which is then amplified by both Brillouin and Raman scattering. To reduce the calculation time, we first solve the partial differential equations for the Raman pump and signal only. After that, we integrate the SBS equation to determine if the SBS threshold is reached. Such process is valid for our case to simulate SBS limited maximum output power. Because SBS gain coefficient is hundreds of times larger than Raman gain coefficient, SBS threshold is well defined in both simulation and experiments. It is also valid for simulating the power curves because the SBS signal is negligible compared to Raman signal before the amplifier reaches SBS threshold.
Laser linewidth is broadened when transmitting through the fiber. However, in the experiments, even at highest power the linewidth is still far below SBS bandwidth, so taking a constant SBS gain coefficient value in the simulation is valid.
In the model, there are three fitting parameters: Raman gain coefficient, SBS gain coefficient, and passive loss. By fitting to signal power curve, unused power curve, and maximum achievable power, these parameters can be determined. Figure 5 (left) shows a numerical fit to the fiber B results. We find the Raman gain coefficients for Nufern 1060XP and Corning HI1060 fibers both are about 0.0012 m−1W−1, while SBS gain coefficients are 0.36 m−1W−1 and 0.67 m−1W−1, respectively. So in term of Raman amplification of narrow linewidth laser, Nufern 1060XP fiber is better than Corning HI1060. The difference might result from different drawing process or fiber doping composition. With the parameters provided by fitting the experimental data, we can determine optimum fiber length for given pump and seed power, and predict the performance very well.
Fiber Raman amplifiers are usually not considered as a way to generate high power and narrow linewidth lasers, because it is thought that the fiber has to be very long for a fiber Raman amplifier, so that SBS will take place at rather low power and the linewidth will be broadened by Kerr nonlinearity. All these concerns are true. But the situation for SBS in an amplifying fiber is different from that in a passive fiber. Although the fiber has to be quite long for efficient Raman amplification, the power distribution of the amplified laser is very uneven. Figure 5 (right) shows calculated signal power distribution inside the amplifier fiber for the case of fiber B at full power. The power distribution is very close to an exponential growth. Most laser power is generated in a short piece of fiber at the end of amplifier. This is the key to understand why a 150 m long fiber amplifier can generate 4.8 W, ~10 MHz laser without suffering much on SBS.
So the idea is to amplify the narrow linewidth laser as fast as possible, which means to operate far below the pump saturation. That is the reason of the low efficiency. But the efficiency can be improved by using special SBS-free fiber, using multiple stage amplifiers with isolators in between, and intra-cavity pumping, etc. We are working on these approaches right now.
Our simulation shows the performance of narrow-linewidth fiber Raman amplifier is determined by the relative strength of Raman and SBS gain. The ratio of Raman to SBS gain coefficient of a fiber, gR/gSBS, can be defined as a figure of merit for building a narrow-line fiber Raman amplifier. For Nufern 1060XP and Corning HI1060 fiber, the values are 0.00333 and 0.00179, respectively. Detailed model, analysis, and simulation results will be presented in a future publication.
4. Summary and perspective
In summary, we have obtained 4.8W, ~ 10MHz, 1178nm laser by Raman amplification of a distributed feedback diode laser in standard single mode fibers pumped by an 1120 nm Yb fiber laser. More than 10% efficiency and 27 dB amplification is achieved, limited by onset of stimulated Brillouin scattering. To our knowledge this is the narrowest linewidth achieved at this power level, with a fiber Raman amplifier. Most laser power is generated in a short piece of fiber at the end of amplifier. That is the reason why we can obtain ~5 W narrow linewidth laser in a 150 m long fiber amplifier. The ratio of Raman to SBS gain coefficient of the fiber is recognized as a figure of merit for building the narrow linewidth fiber Raman amplifier.
There is still much room to improve the performance. Many ways of SBS suppression have been proposed . One direction is to broaden the laser linewidth, which is out of interest beyond 60–100 MHz, in the context of this work. Another direction is to broaden the SBS gain spectrum instead. This can be done by designing special fibers . But the SBS gain spectrum can also be broadened artificially. For example, one can apply a temperature  or stress distribution along the fiber  to effectively broaden the SBS gain bandwidth. Furthermore, optically isolated multiple stage amplifiers and intra-cavity pumping scheme can be applied to increase the efficiency.
References and links
1. D. B. Calia, W. Hackenberg, S. Chernikov, Y. Feng, and L. Taylor, “AFIRE: fiber Raman laser for laser guide star adaptive optics,” Astronomical Telescopes and Instrumentation, SPIE Orlando, 6272–55 (2006).
2. Y. Feng, L. Taylor, W. Hackenberg, D. Bonaccini Calia, and S. Chernikov, “Multi-watt 589nm laser by frequency doubling of a fibre Raman MOPA, ” EPS-QEOD Europhoton Conference 2006, Pisa, 2006, WeE6.
3. S. A. Skubchenko, M. Y. Vyatkin, and D. V. Gapontsev, “High-Power CW Linearly Polarized All-Fiber Raman Laser,” IEEE Photon. Tech. Lett. 16, 1014–1016 (2004). [CrossRef]
4. S. Huang, Y. Feng, A. Shirakawa, and K. Ueda, “Generation of 10.5 W, 1178 nm laser based on phosphosilicate Raman fiber laser,” Jpn. J. Appl. Phys. 42, L 1439–L 1441 (2003). [CrossRef]
5. Y. Feng, S. Huang, A. Shirakawa, and K. Ueda, “589nm light source based on Raman fiber laser,” Jpn. J.Appl. Phys. 43, L722–L724, (2004). [CrossRef]
6. D. Georgiev, V. P. Gapontsev, A. G. Dronov, M. Y. Vyatkin, A. B. Rulkov, S. V. Popov, and J. R. Taylor, “Watts-level frequency doubling of a narrow line linearly polarized Raman fiber laser to 589nm,” Opt. Express 13, 6772–6776 (2005), http://www.opticsinfobase.org/abstract.cfm?URI=oe-13-18-6772. [CrossRef] [PubMed]
7. H. Masuda, K.-I. Suzuki, S. Kawai, and K. Aida, “Ultra-wideband optical amplification with 3 dB bandwidth of 65 nm using a gain-equalised two-stage erbium-doped fibre amplifier and Raman amplification,” Electron. Lett. 33, 753–754 (1997). [CrossRef]
8. P. C. Reeves-Hall, D. A. Chestnut, C. J. S. d. Matos, and J. R. Taylor, “Dual wavelength pumped L- and U-band Raman amplifier,” Electron. Lett. 37, 883–884 (2001). [CrossRef]
9. G.P. Agrawal, Nonlinear Fiber Optics (Academic Press, 1995).
10. M. -J. Li, X. Chen, J. Wang, S. Gray, A. Liu, J. A. Demeritt, A. B. Ruffin, A. M. Crowley, D. T. Walton, and L. A. Zenteno, “Al/Ge co-doped large mode area fiber with high SBS threshold,“ Opt. Express 15, 8290–8299 (2007). [CrossRef] [PubMed]
11. J. Hansryd, F. Dross, M. Westlund, P. A. Andrekson, and S. N. Knudsen, “Increase of the SBS threshold in a short highly nonlinear fiber by applying a temperature distribution,” J. Lightwave Technol. 19, 1691–1697 (2001). [CrossRef]
12. J. M. Chávez Boggio, J. D. Marconi, and H. L. Fragnito, “8 dB increase of SBS thresold in an optical fiber by applying a stair ramp strain distribution,” CLEO 2004 - San Francisco USA, (2004) CTh30.