To the best of our knowledge we report for the first time on an Yb-sensitized Er-doped cladding pumped fiber amplifier which is simultaneously seeded by two single-frequency lasers operating at 1556 nm and 1064 nm, respectively. This mode of operation ensures stable amplifier operation by reducing the gain around 1 µm wavelength to the large signal gain value, while having no significant effect on the slope efficiency of the amplification process at 1556 nm when pumping at 976 nm. We were able to demonstrate stable output power of 8.7 W at 1556 nm with an amplifier gain of > 22 dB, a co-propagating pumping scheme and the power limitation only being set by the available amount of pump power.
©2009 Optical Society of America
Fiber lasers and amplifiers have made great progress with respect to output power within the last years. These advances have been enabled by rapid developments made in both fiber and high-brightness laser diode technology. For the case of single-frequency fiber amplifiers there have been presented especially impressive results in Yb- and Tm-doped fiber amplifiers [1,2]. Progress of this kind is more difficult to achieve at 1.5 µm wavelength but is at the same time of importance for example in the research and development of future gravitational wave detectors [3,4].
There are currently two approaches with respect to fiber doping for multiwatt-level cladding pumped fiber amplifiers operating at this wavelength, which are either pure Er-doping or Er:Yb-codoping. The main issue of purely Er-doped fibers are the relatively low achievable doping concentrations due to the detrimental effects of the onset of ion-clustering and the simultaneously also relatively low values of the Erbium absorption cross-sections . This leads to the necessity of long fibers – and therefore an accordingly low threshold for stimulated Brillouin scattering – and at the same time relatively small cladding diameters between 100 and 200 µm, thus creating also the need for very high brightness pump diodes. Larger cladding diameters could only be reached by an according scaling of the signal core, which is not an option, if a good beam quality and a predominant part of radiation in the TEM00-Mode is needed, as it is for example the case in gravitational wave detector applications.
An often used approach to overcome the described limitations is the utilization of Er:Yb-codoped fibers, which are most commonly pumped in the Ytterbium absorption band between 915 nm and 980 nm. The main part of the pump light absorption is then due to the codoped Yb3+ ions, which afterwards transfer the excitation energy to the Er3+ ions via a resonant process, which exists because of the Erbium absorption band around 980 nm. To achieve reasonable efficiency values, strong codoping with Phosphorus is needed to suppress the back-transfer of energy to the Yb3+ ions, thus allowing for an efficient amplification of radiation around 1550 nm by using pump light around 976 nm . In conventional step-index designs this Phosphorus-codoping leads to high core NA values around 0.2, which are not in themselves detrimental but should be kept in mind when designing amplifiers with the goal of achieving decent beam quality.
In Er:Yb-codoped continuous wave amplifiers, power levels in excess of 100 W have been reached [7,8]. The highest single-frequency output power was reached by Jeong and associates at the ORC in Southampton. This work also showed most impressively an important issue of Er:Yb-codoped amplifiers, namely the onset of massive Yb-ASE at high pump powers. In case of insufficient seed power around 1.5 µm an increasing amount of Yb-inversion is built up as the Yb pump rate is getting larger than the rate of the Er-Yb-transfer, thus leading to large unsaturated gain for signals around 1.0 µm and hence Yb-ASE. This in turn leads to an extremely low threshold for spurious lasing and self-pulsing effects inside of the fiber. Hence, significant Yb-ASE content in Er:Yb lasers and amplifiers is an indicator of a mode of operation that is highly unstable and potentially hazardous as e.g. spontaneous formation of giant pulses can easily damage the fiber end-facets. Thus the achievement of long-term stable operation of high power Er:Yb-codoped fiber amplifiers needs great care in suppression of the gain values for signals around 1 µm wavelength.
Several approaches have been made to achieve this goal. In Ref. . several fiber Bragg gratings (FBGs) where used to filter out radiation between 1060 and 1080 nm, this solution being effective and elegant but at the same time costly and potentially time-consuming, as special FBGs have to be either bought or developed. Another straightforward approach is to use a counter-propagating pumping scheme together with a substantial amount of seed power to ensure an efficient energy extraction by the signal at 1.5 µm, thus lowering the gain for signals around 1.0 µm by efficiently depleting the Yb-inversion level . The main drawback of this approach is the relatively low overall amplifier gain of about 10 dB until the onset of the Yb-ASE, which in turn can lead to the need of carefully tailoring chains of several amplifiers for the case of high power and/or high gain applications . Another potential issue of this approach is also the all-fiber implementation of the counter-propagating pumping scheme (in contrast to the co-propagating scheme used in the work presented here), where standard tapered fiber bundle couplers are not suitable for long-term operation as they have a low core-to-cladding isolation, thus typically transmitting several percent of the amplified signal through every pump port of the coupler. In turn, the lifetime of the pump diodes will most probably be significantly decreased, unless additional optical isolators or dichroic mirrors are implemented to protect each pump diode separately.
Another approach is to use photonic bandgap fibers that show only weak guidance for radiation around 1.0 µm wavelength . The main drawback of this approach is – especially in case of cw-operation – the to date very low slope efficiency of the Er:Yb-doped bandgap fibers, as the important heavy co-doping with Phosphorus is incompatible with photonic crystal fibre (PCF) structures, which is based on the fact that these must not contain a dopant induced index step from core to cladding.
In 2007 Jeong and associates  showed that an Er:Yb-codoped fiber laser could be scaled to higher power levels around 1.5 µm wavelength, when a controlled simultaneous laser oscillation at 1.0 µm rather than a suppression of such a laser oscillation was being utilized. This is connected to the fact that a stable laser operation leads to an efficient energy extraction from the Yb3+ ions, thereby preventing uncontrolled self-pulsing and setting the gain at 1.0 µm – as well as the Yb-inversion – to fixed values. Our idea was thus to investigate the possibility of a transfer of this scheme from a fiber laser to single-pass single-frequency fiber amplifier. The important difference between the two experiments was that in our case the Yb-inversion and thus the amplifier gain at 1 µm would not be clamped to a fixed specific value, so that pump power dependent changes in inversion levels and gain behavior were to be expected.
2. Experimental setup
The experimental setup is shown in Fig. 1 . To seed the fiber amplifier at 1.5 µm wavelength a commercial fiber-coupled single-frequency single-mode DFB-diode (EM4 EM253-080-069) was used, emitting 80 mW of power at 1556 nm with a linewidth of < 10 MHz, whose upper limit was being set by the utilized laser diode controller. The additional seed source at 1064 nm was a single-frequency Nd:YAG non-planar ring oscillator (Innolight Mephisto) with a linewidth of < 1 kHz. This narrow laser linewidth was obviously not necessary for the intended use of the laser, but the device was used due to its ability of providing up to 2 W of output power and availability at the time of conducting the experiments. The two signal sources were protected against backreflections by a fiber-coupled and a free-space optical isolator, respectively. The radiation emitted by the DFB-diode was additionally passed through a 3.5% tap-coupler, which served as a monitor port.
The two signals were then combined in a commercial 1060/1550 WDM-coupler (Sifam FFW-8C32B2510), followed by a 6 + 1x1 TFB-coupler (Sifam TFB-550611B70). Six fiber-coupled laser diodes (Sheaumann SP-976-303; fiber core size: 105 µm, fiber NA: 0.15) were used as pump sources for the amplifier. Each of these diodes emitted up to 6 W of radiation at 976 nm. The TFB-coupler was spliced to the active fiber (9.5 m of Nufern SM-EY-7/130; core-diameter: 7 µm, cladding-diameter: 130 µm, core NA: 0.17); a measurement showed that about 50 mW of 1.5 µm seed power actually reached the active fiber. All free fiber ends were either angle-cleaved or terminated by angle-polished fiber connectors at an angle of approximately 8° in order to suppress unwanted backreflections.
3. Results and discussion
First, we present the output characteristics of the amplifier, when no signal at 1064 nm is coupled into the system. The shown optical slope efficiency of the amplification of the 1.5 µm signal is 25%, the maximum output power of 1.72 W being limited by the onset of spurious Yb-ASE (see Figs. 2 and 3(b) ). The slope efficiency is measured with respect to emitted pump power, i.e. including the losses at splice connections and the TFB-coupler. These connections introduced losses of 10% measured after the TFB-coupler and another estimated 5-10% at the splice connection of the TFB output and the active fiber.
This choice of maximum operational point of this amplifier is agreeably a relative conservative one. Slightly higher output powers may be extracted from such a setup, but one should keep in mind that the Yb-ASE will also rise quickly to significant power levels, indicating excessive energy storage and dramatically increasing the probability and potential effect of the occurrence of instabilities.
In our experience it is mostly a matter of seconds until spurious lasing and/or giant pulses at 1.0 µm occur in such a state of operation, immediately leading to the destruction of one or even several fiber end-facets. On the other hand it has been observed that the characteristics of the spectra around 1.5 µm (see Fig. 3(a)) do not show a significant change after the pump power is increased beyond 6 W, while at this point the slope efficiency of the signal at 1.0 µm changes (see Fig. 2) and the Yb-ASE simultaneously begins to develop a pronounced maximum around 1065 nm (see Fig. 3(b)). These facts suggested that successfully extracting the excessive energy from the Yb-ions without significantly degrading the amplification efficiency at 1.5 µm was indeed feasible.
When setting the Nd:YAG non-planar ring oscillator (NPRO) to an output power (before the fiber coupling with a transfer efficiency of about 50% from free space to active fiber) of 60 mW the optical spectra in the two emission bands changed to the ones shown in Figs. 4(a) and 4(b), respectively. The amplified signal at 1064 nm showed stable operation and a peak-to-peak ASE-suppression of > 40 dB, the resolution bandwidth of the optical spectrum analyzer (OSA) being set to 0.5 nm in all described measurements. The peak-to-peak ASE-suppression at 1.5 µm wavelength was even slightly larger and showed values of > 50 dB at high as well as at lower powers. The background ASE in Fig. 4(a) is not observable as in Fig. 3(a) as at higher signal powers a larger dynamic range was needed in order to raise these signals above the noise floor of the OSA. This dynamic range would in principle have been available by coupling more light into the OSA, but this would also have meant to risk the damage of the device in case of unexpected pulsing behavior of the amplifier. However, as the ASE peak values are still about 20 dB above the noise floor, a comparison of the achieved ASE suppression values is possible without significant inaccuracies being introduced by the limited dynamic range of the measurements.
In the spectra of the Yb-emission (see Fig. 3(b) and Fig. 4(b)) it is clearly observable that both with and without an auxiliary signal at 1.0 µm a stable amplification process without parasitic lasing was achieved, but the safely reachable output powers at 1.5 µm – and also the peak-to-peak ASE-suppression – were higher in the case with the additional signal at 1.0 µm wavelength.
The slope efficiency of the 1.5 µm signal remained unaffected at the value of 25% at low pump powers, showing only a slight rollover to a value of 20% around 25 W of pump power. The output power versus emitted pump power at 1.5 µm is plotted in Fig. 5(a) , the maximum achieved output power of 8.7 W being limited by the available pump power instead of the onset of instabilities. In Fig. 5(b) the output power at 1.0 µm wavelength is plotted against emitted pump power, the optical to optical efficiency of < 10% showing that the energy transfer between the Yb3+ and Er3+ ions was almost unaffected in spite of the additional 1064 nm signal. This conclusion is based in the fact, that in case of a significant detrimental influence of the 1064 nm signal on the energy transfer process a higher optical efficiency for the amplification at 1.0 µm – and in turn a lowered efficiency at 1.5 µm – would have to be expected.
The low efficiency of the amplification at 1064 nm is probably caused by two effects. The first being the heavy Phosphorous co-doping of the utilized fiber, efficiently depleting the backtransfer of energy from Er3+ to Yb3+ ions  and thus creating a strong advantage for the Er3+ ions in the competition for the pump energy launched into the active fiber. The second reason for the preferential amplification of 1.5 µm radiation is most probably the non-zero absorption cross-section at 1064 nm  of Yb-ions in P-codoped glass, even allowing for pumping of Er:Yb amplifiers at this wavelength . Thus the 976 nm pump light depletion at the end of the active fiber in the co-propagating pumping scheme even possibly allows for the reabsorption of the 1064 nm signal and an associated additional conversion of 1.0 µm to 1.5 µm radiation.
In order to examine the influence of this last described effect, it is instructive to examine the influence of a change of input power of the 1.0 µm signal with the 1.5 µm seed and 976 nm pump power held constant. The possibility of core-pumping the amplifier with the 1.0 µm signal is shown in Fig. 6(a) with the pump power switched off. Here, the 1.5 µm signal of which roughly 50 mW actually pass the TFB-coupler is either amplified or at least less attenuated inside of the fiber because of an increased inversion of the Er3+ ions due to the absorbed 1 µm signal. When increasing the pump power at 976 nm the influence of a variation in the 1064 nm signal power changes from that of a pure pump source to that of an also partly competing seed signal. This becomes manifest in the – with rising 976 nm pump power increasingly observable – rollover of the slope efficiency of 1.0 µm versus 1.5 µm signal power near the maximum output power of the NPRO (see Figs. 6(b) to 6(d)). However, these results also pronounce once more the robustness of the amplification efficiency of the 1.5 µm signal, even in the presence of a substantially stronger 1.0 µm signal. The maximum pump power used in these measurements was 30 W as one of the six pump diodes was damaged in the course of the experiments so that at this time only the output power of five pump diodes was available.
The small fluctuations in output power which can be observed most pronouncedly in Figs. 6(b) and 6(d) can be explained by the fact that the output powers depend – via alteration of the Yb absorption cross-section  – strongly on the temperature of the active fiber. This means that tuning of the pump power leads – besides the direct pump power induced change in output power – to a slow thermal drift of the measured output power due to the large quantum defect of the laser system. These variations most probably occur in the obtained data due to a too short time interval between the increasing the pump power and recording of the according data point. However, these effects varied the output power only in the range of a few percent and after a few seconds of waiting the thermal drift ended. Afterwards stable output power values were reached, so that no sustained power fluctuations were observed. Moreover, these effects have not been limiting the achievable output powers, so that they did not impose problems in the presented experiments.
4. Conclusion and outlook
In conclusion, we have demonstrated the possibility of stabilizing an Er:Yb-codoped fiber amplifier against spurious lasing by the utilization of readily available standard components and the use of an additional seed signal at 1064 nm wavelength. We demonstrated a maximum single-frequency single-mode output power scaling from < 2 W to 8.7 W at 1.5 µm with an ASE suppression of > 50 dB. We also discussed the governing effects and showed the influence of a power variation of the signal at 1.0 µm. As the fraction of the 1.5 µm signal that reached the active fiber was only about 50 mW, we demonstrated a stable amplification of > 22 dB, which is – especially in a co-directional pumping scheme – a very high value for large signal Er:Yb-codoped fiber amplifiers and was at this point only limited by the available pump power.
The NPRO utilized as 1064 nm seed source is clearly a very sophisticated device for this task and could be replaced by a lower power and more broadband laser source in order to decrease the costs of setting up an amplifier system of this kind.
Future work will deal with the determination of the noise figure of the amplifiers in the described configuration and the scaling of this approach to higher output powers. Another interesting task is the numerical simulation of the treated laser system, but agreeably this appears to be very difficult as no less than 5 cross-sections (Yb absorption and emission at 1.0 µm, Er absorption at 1.0 µm and Er absorption and emission at 1.5 µm wavelength) and additionally the transfer process – at 976 and 1064 nm – between the Er3+ and Yb3+ ions are involved and most of the needed spectroscopic values are uncertain – especially the transfer coefficient at 1064 nm – and/or strongly depending on the specific utilized active fiber.
Nevertheless, we are very confident that a broad applicability of the presented results is given as we have not only shown the possibility of stabilizing an Er:Yb fiber amplifier at high power levels, but also the scheme's robustness to signal variations at 1064 nm, which in turn guarantees for a broad range of applicable input powers and power densities in an amplifier of the described kind.
The authors thank the German Research Foundation (DFG) for funding the Cluster of Excellence Centre for Quantum Engineering and Space-Time Research QUEST.
References and links
1. Y. Jeong, J. Nilsson, J. Sahu, D. Payne, R. Horley, L. Hickey, and P. Turner, “Power Scaling of Single-Frequency Ytterbium-Doped Fiber Master-Oscillator Power-Amplifier Sources up to 500 W,” IEEE J. Sel. Top. Quantum Electron. 13(3), 546–551 (2007). [CrossRef]
3. F. Brückner, T. Clausnitzer, O. Burmeister, D. Friedrich, E.-B. Kley, K. Danzmann, A. Tünnermann, and R. Schnabel, “Monolithic dielectric surfaces as new low-loss light-matter interfaces,” Opt. Lett. 33(3), 264–266 (2008). [CrossRef] [PubMed]
4. M. Mehmet, S. Steinlechner, T. Eberle, H. Vahlbruch, A. Thüring, K. Danzmann, and R. Schnabel, “Observation of cw squeezed light at 1550 nm,” Opt. Lett. 34(7), 1060–1062 (2009). [CrossRef] [PubMed]
5. P. Blixt, J. Nilsson, T. Carlnäs, and B. Jaskorzynska, “Concentration-dependent upconversion in Er3+-doped fiber amplifiers: Experiments and modeling,” IEEE Photon. Technol. Lett. 3(11), 996–998 (1991). [CrossRef]
6. K. Tankala, B. Samson, A. Carter, J. Farroni, D. Machewirth, N. Jacobson, U. Manyam, A. Sanchez, M.-Y. Cheng, A. Galvanauskas, W. Torruellas, and Y. Chen, “New developments in high power eye-safe LMA fibers,” Proc. SPIE 6102, 610206 (2006). [CrossRef]
7. A. Yusim, J. Barsalou, D. Gapontsev, N. S. Platonov, O. Shkurikhin, V. P. Gapontsev, Y. A. Barannikov, and F. V. Shcherbina, “100 Watt, single-mode, CW, linearly polarized all-fiber format 1.56um laser with suppression of parasitic lasing effects,” Proc. SPIE 5709, 69–77 (2005). [CrossRef]
8. Y. Jeong, J. K. Sahu, D. B. S. Soh, C. A. Codemard, and J. Nilsson, “High-power tunable single-frequency single-mode erbium:ytterbium codoped large-core fiber master-oscillator power amplifier source,” Opt. Lett. 30(22), 2997–2999 (2005). [CrossRef] [PubMed]
9. B. Morasse, S. Agger, S. Chatigny, É. Gagnon, J.-P. de Sandro, and C. Poulsen, “10W ASE-free single mode high power double cladding Er3+-Yb3+ amplifier,” Proc. SPIE 6453, 645324 (2007). [CrossRef]
10. P. Wysocki, T. Wood, A. Grant, D. Holcomb, K. Chang, M. Santo, L. Braun, and G. Johnson, “High Reliability 49 dB Gain, 13 W PM Fiber Amplifier at 1550 nm with 30 dB PER and Record Efficiency,” in Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, Technical Digest (CD) (Optical Society of America, 2006), paper PDP17.
11. A. Shirakawa, H. Suzuki, M. Tanisho, and K. Ueda, “Yb-ASE-Free Er Amplification in Short-Wavelength Filtered Er:Yb Photonic-Crystal Fiber,” in Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, OSA Technical Digest (CD) (Optical Society of America, 2008), paper OThN2.
12. Y. Jeong, S. Yoo, C. Codemard, J. Nilsson, J. Sahu, D. Payne, R. Horley, P. Turner, L. Hickey, A. Harker, M. Lovelady, and A. Piper, “Erbium:Ytterbium Codoped Large-Core Fiber Laser With 297-W Continuous-Wave Output Power,” IEEE J. Sel. Top. Quantum Electron. 13(3), 573–579 (2007). [CrossRef]
13. J. Lincoln, W. Barnes, W. Brocklesby, and J. Townsend, “Spectroscopic evaluation of the vibrational coupling of Er3+ ions in phospho-aluminosilicate fibres and an explanation of compositional variations in Er-Yb 1.5 µm amplifier performance,” J. Lumin. 60-61, 204–207 (1994). [CrossRef]
14. S. Grubb, W. Humer, R. Cannon, T. Windhorn, S. Vendetta, K. Sweeney, P. Leilabady, W. Barnes, K. Jedrzejewski, and J. Townsend, “+21 dBm erbium power amplifier pumped by a diode-pumped Nd:YAG laser,” IEEE Photon. Technol. Lett. 4(6), 553–555 (1992). [CrossRef]