We report a more than 150 W spectrally-clean continuous wave Raman fiber laser at 1120 nm with an optical efficiency of 85%. A ~30 m standard single mode silica fiber is used as Raman gain fiber to avoid second Stokes emission. A spectrally asymmetric resonator (in the sense of mirror reflection bandwidth) with usual fiber Bragg gratings is designed to minimize the laser power lost into the unwanted direction, even when the effective reflectivity of the rear fiber Bragg grating becomes as low as 81.5%.
©2009 Optical Society of America
Raman fiber lasers have found various applications due to their broad gain spectrum and wavelength versatility. Fiber Bragg gratings (FBGs) are usually used as reflecting mirrors to construct Raman fiber lasers for all-fiber laser systems. But FBGs typically have narrow reflection bandwidth, which becomes a problem when scaling the Raman fiber laser to high power [1–3]. Because of spectral broadening, the oscillating light inside the cavity can have much larger bandwidth than the FBGs. Thus, the effective reflectivity of the FBGs can differ significantly from the designed values. This effect not only makes the prediction with usual numerical models not useful, but also reduces the laser efficiency. A large fraction of the laser power may leak to the unwanted direction through the rear, supposedly highly reflective, FBG.
Four-wave-mixing between numerous longitudinal modes associated with a long fiber laser cavity is thought to be responsible for this spectral broadening. Many numerical and theoretical studies have been carried out to explain the experimental observations [1–5]. Naturally, an asymmetric cavity with a broadband FBG as high reflector and narrowband FBG as output coupler was proposed to overcome the laser spectral leakage problem [1,2]. But broadband chirped FBGs typically have significant cladding-mode losses.
There are a number of reports on Raman fiber lasers with more than 10 W output. The maximum output power reported, to the best of our knowledge, is 41 W at 1480 nm band by Y. Emori et al .
In this paper, we report a high power, 153 W, highly efficient (85%), and spectrally-clean Raman fiber laser at 1120 nm. A ~30 m standard single mode silica fiber is used as Raman gain fiber to avoid second Stokes emission. Commonly available FBGs are used to construct a spectrally asymmetric cavity (in the sense of FBG bandwidth), so as to minimize the laser power lost into the unwanted direction. The laser is built as a pump source for high power narrow linewidth Raman fiber amplifier at 1178 nm, which can be frequency doubled to 589 nm for laser guide star adaptive optics application in astronomy [7,8].
2. Experimental Setup
Figure 1 shows a diagram of the experimental setup, which is a standard Raman fiber laser configuration. Special attention is made to protect the pump laser from possible backward signal light, using two wavelength division multiplexers WDM1 and WDM2. The pump source is a 200 W class CW Yb fiber laser at 1070 nm from IPG Photonics Corp. After splicing the delivery cable of the IPG fiber laser to the WDM pair, we get a maximum of about 180 W 1070 nm throughput, which is used to pump the Raman fiber laser. The power loss is due to splicing loss and imperfect WDMs. FBG1 is a high reflector with a peak reflection of 99.7% at 1120 nm and a bandwidth of 1.05 nm. FBG2 is partial reflector with a peak reflection of 8.9% at 1120 nm and a bandwidth of 0.55 nm. WDM3 is added to the laser output end to separate the signal light from any unused pump light. All WDM and FBG components are heat-sinked to metallic baseplates for stable high power operation. Laser output, unused pump light, and laser leak through the high reflector FBG1 as a function of pump power are measured. Spectra of the laser output and the leak through the high reflector are measured with an Ando spectrum analyzer.
A piece of standard single mode silica fiber (HI1060 from Corning, Inc.) is used as the Raman gain fiber. A fiber length of 30 m is determined empirically from previous work [7,8], where a 75 W 1120 nm laser was built using a 50 m long fiber pumped by a 100 W level Yb fiber laser, without observation of second Stokes lasing. Also from that previous work, we know what effective reflectivities of the FBGs to expect at high power, allowing us to estimate the fiber length needed for sufficient Raman conversion efficiency while, similarly, keeping it as short as possible to avoid second Stokes lasing.
3. Results and analysis
Figure 2 (a) shows the 1120 nm output as a function of the 1070 nm pump power. The laser has a threshold of ~30 W. Maximum 153 W of 1120nm laser is obtained with a 180 W pump input, which corresponds to an 85% optical conversion efficiency. Interesting to note that due to the nonlinear nature of the stimulated Raman scattering, the slope efficiency of the laser is more than 140% near threshold, and drops to 85% at full power.
Because of the imperfect 1070/1120 nm WDM, the signal and unused pump light are not separated completely. Near the laser threshold there is significant amount of 1070 nm light at the 1120 nm output port, because the unused pump power is high. At full power, however, the unused pump laser is as low as 1W, and the 1120 nm laser output is clean, as shown in Fig. 2 (b). We do observe both 1070 nm pump light and 1180 nm second Stokes spontaneous emission, but these are more than 35 dB and 40 dB less than the 1120nm light, respectively.
Figure 3 (a) shows the FWHM linewidth of the 1120 nm laser as a function of the laser power. Just above the threshold, the linewidth is determined by the spectral properties of the outcoupling FBG. At higher power, the linewidth broadens up to ~3.9 nm. The linewidth broadening effect in Raman fiber laser has been explained by four-wave-mixing between numerous longitudinal modes associated with long fiber laser cavity [4,5].
Figures 3 (b) and (c) show the spectra of the laser output and laser leak through FBG1 at different power levels, 1 W, 27 W, 63 W, and 153 W (black, red, green, and blue curves), respectively. In each round trip, the outcoupling FBG2 samples a small fraction of laser output within its reflection band and feeds it back into the cavity, wherein it is amplified in the direction towards FBG1. The highly reflective FBG1 in turn reflects nominally all the light that is within its reflection band, which is then amplified (in the forward direction) to generate the output through the outcoupling FBG2. The significant central dip in the laser leak spectra (behind FBG1) is due to the reflection of the FBG1. The spectral ripples originate from the side lobes of the outcoupling FBG2 spectrum.
The effective reflectivities of two FBGs are expected to be much smaller than the specified peak values, due to the spectral broadening of the circulating light. We do not simulate the linewidth broadening here. Instead, we use numerical simulations to derive the effective reflectivities by fitting the data of 1120 nm laser output, 1120 nm laser leakage through the rear FBG1, and unused pump with a classical Raman fiber laser model (Eq. (1), which is a system of first-order coupled ordinary differential equations [9,10].9], which is very close to 1 in this case. The contribution of the spontaneous terms is important only when the laser is below and near threshold. The boundary conditions (Eq. (2) are:
To perform accurate simulations, the light through the two output ports of WDM3 is spectrally separated to obtain the exact laser powers. It is found that at full power, 4.5W of 1120 nm laser goes to the 1070 nm port, which is about 2.9% percent of the total power. Therefore, the actual maximum 1120 nm laser power is 157.5 W, corresponding to a conversion efficiency of 87.5%, close to the quantum limit of 95.5%. The unused 1070 nm pump laser is only 1W, which indicates highly efficient use of pump laser. The large deviation from the specification of the WDM might result from the broad linewidth of the laser and heat dissipation inside the component.
In the simulation, fiber length L is set to 32 m to account also for the fiber pigtails of the FBGs. Since the pump and signal wavelength are close, αP is set to equal to αR. The Raman gain coefficient gr is set to 0.0012 m−1W−1, which is determined in earlier work . At each pump power, by fitting the data of 1120 nm laser output, the 1120 nm laser leakage through FBG1, and the unused pump with the model, we can derive R A, R B and αR. Figure 4 (a) shows the numerical fitting to the data, while Fig. 4 (b) shows the resulting effective reflectivities of the two FBGs. We find the effective reflectivity of the nominally highly reflective FBG1 drops from 99.7% to 81.5% at full power, while the effective reflectivity of FBG2 drops from 8.9% to 0.97%. In spite of the low reflectivity of the rear FBG (81.5%), the leaked laser power is only 3.2 W, so the directionality of the laser is still as high as 49:1.
The directionality of the laser outputs from a cavity is determined by the reflectivities of two end mirrors .
where P A and P B are output powers from end A and B, respectively; R A and R B are reflectivities of two end mirrors. Equation (3) is valid if the forward and backward trips have the same gain factor, which is the case for all continuous wave lasers. Therefore, as long as the ratio of the reflectivities of the rear mirror to the outcoupling mirror remains high, good directionality can be guaranteed.
The signal and pump power distributions along the fiber are calculated for understanding the laser power evolution inside the laser cavity. As shown in Fig. 5 , at full power, only 1.54 W of laser power is fed back to cavity at the outcoupling FBG2, which is amplified to 17.3 W at FBG1; 14.1 W of which is reflected, and amplified to 158.7 W at the output end of the laser. The majority of the laser power is generated in the forward direction. For this reason, the loss through the rear FBG1 is low even when the effective reflectivity is reduced to 81.5%.
The reflective spectrum of FBG1 is almost rectangular with a width of 1.05 nm. Assuming the laser spectrum is Gaussian, an 81.5% effective reflectivity of FBG1 suggests that the laser linewidth is about 0.84 nm when approaching FBG1. This implies that the line broadening in backward trip (towards FBG1) is about 1.5 times, from an initial 0.55 nm defined by the linewidth of FBG2 spectrum. The line broadening in the forward direction is about 4.6 times (3.9 nm/0.84 nm), and we attribute the difference to the fact that laser power is much higher in the forward direction.
The results indicate that commonly available FBGs are sufficient to overcome the laser leakage problem, provided the ratio of spectral width of the rear FBG to that of output FBG is larger than the linewidth broadening in the backward trip. The laser performance, in terms of efficiency and directionality, may be further improved by using narrower output FBG, for example, 0.25 nm instead of 0.55 nm.
In summary, we report a high power (153 W), highly efficient (85%), and spectrally-clean Raman fiber laser emitting at 1120nm. To the best of our knowledge, this is the highest power reported from a fiber Raman laser system at any wavelengths. A short (~30m) piece of standard single mode silica fiber is used to avoid second Stokes emission. Commonly available FBGs are used to construct a spectrally asymmetric (in the sense of FBG bandwidth) and extremely low-Q (essentially a double pass amplifier) cavity. All these measures are combined to minimize the laser power lost into the unwanted direction and maintain a high directionality even when the actual reflectivity of the rear mirror becomes as low as 81.5%.
References and links
1. S. A. Babin, D. V. Churkin, and E. V. Podivilov, “Intensity interactions in cascades of a two-stage Raman fiber laser,” Opt. Commun. 226(1-6), 329–335 (2003). [CrossRef]
2. R. Vallée, E. Bélanger, B. Déry, M. Bernier, and D. Faucher, “Highly Efficient and High-Power Raman Fiber Laser Based on Broadband Chirped Fiber Bragg Gratings,” J. Lightwave Technol. 24(12), 5039–5043 (2006). [CrossRef]
3. P. Suret and S. Randoux, “Influence of spectral broadening on steady characteristics of Raman fiber lasers: from experiments to questions about validity of usual models,” Opt. Commun. 237(1-3), 201–212 (2004). [CrossRef]
4. J.-C. Bouteiller, “Spectral modeling of Raman fiber lasers,” IEEE Photon. Technol. Lett. 15(12), 1698–1700 (2003). [CrossRef]
5. S. A. Babin, D. V. Churkin, A. E. Ismagulov, S. I. Kablukov, and E. V. Podivilov, “Turbulence-induced square-root broadening of the Raman fiber laser output spectrum,” Opt. Lett. 33(6), 633–635 (2008). [CrossRef] [PubMed]
6. Y. Emori, K. Tanaka, C. Headley, and A. Fujisaki, High-Power Cascaded Raman Fiber Laser with 41-W Output Power at 1480-nm Band,” in Conference on Lasers and Electro-Optics/Quantum Electronics and Laser Science Conference and Photonic Applications Systems Technologies, OSA Technical Digest Series (CD) (Optical Society of America, 2007), paper CFI2.
8. Y. Feng, L. Taylor, and D. Bonaccini Calia, “25 W Raman-fiber-amplifier-based 589 nm laser for laser guide star,” Opt. Express 17(21), 19021–19026 (2009). [CrossRef]
12. D. Y. Shen, L. Pearson, P. Wang, J. K. Sahu, and W. A. Clarkson, “Broadband Tm-doped superfluorescent fiber source with 11 W single-ended output power,” Opt. Express 16(15), 11021–11026 (2008). [CrossRef] [PubMed]