A transversely chirped volume Bragg grating (TCVBG) is used for flexible wavelength-tuning of a high-power (>100 W) tunable Yb-fiber laser oscillator. Continuous tuning over 2.5 THz of the narrow-band (13 GHz) signal was achieved by transversely translating the TCVBG during high-power operation without cavity realignment. The laser operated in a single polarization with a beam propagation factor (M2) of 1.2. Since the cavity losses remained constant, the high gain fiber laser exhibited excellent power stability (<0.6% relative deviation) over the investigated tuning range. The possibility of considerably increasing the output power and extending the tuning range while maintaining the power stability is discussed.
© 2013 OSA
High-power continuous-wave tunable laser sources with excellent spectral qualities are not only readily used in spectroscopy and photochemistry applications, but also frequently used as pump sources for nonlinear frequency converters, which enables access to even broader spectral regions. In that regard, Yb-doped fiber based sources with their broad fluorescence spectrum are uniquely suitable for continuously tunable lasers in the 1 µm-region. However, in order to benefit from this flexibility an efficient method for spectral control has to be established that features excellent power scaling and thermal handling capabilities.
The most common approach to realize such sources is to use multiple amplification stages to scale the output power of low power tunable seed lasers . Since each amplification stage requires numerous and sometimes hard-to-obtain components (i.e. high-power isolators), introduces additional signal noise through ASE and adds to the overall complexity, a compact power-scalable fiber oscillator solution promises significant advantages over such solutions [2,3]. Although, fiber Bragg gratings (FBG) are the most attractive choice for spectral narrowing of high power fiber oscillators , since they feature alignment-free operation in an all-fiber configuration with low insertion loss, wavelength-tuning of FBGs by introducing thermal or mechanical strain  during high-power operation is limited to narrow tuning ranges before irreversibly damaging the component.
In contrast to that, the use of volume Bragg gratings for external wavelength-locking of high-power fiber lasers permits flexible wavelength-tuning [6–9]. Even though a tuning range of an Yb-fiber laser over 120 nm has been demonstrated (using four separate VBGs) , the applied angle-tuning method is ultimately limited by the decreasing diffraction efficiency at larger tuning angles . However, a tuning method based on transversely chirped volume Bragg gratings (TCVBG), has successfully been employed in solid-state lasers  and optical parametric oscillators . The fan-shaped structure of the used gratings amounts in a variation of the reflected wavelength when transversely translating the grating. As this method permits to maintain normal incidence of the laser beam on the grating, the entire emission spectrum of the gain medium could be addressed within the tuning range of the grating by employing one properly designed TCVBG. In this work we propose the use of a TCVBG for wavelength-tuning and -locking of a high-power Yb-fiber oscillator. To the best of our knowledge, we demonstrate the first high-power (>100 W) narrow-band fiber oscillator operating in a single-polarization, which is tunable over 2.5 THz. In particular, the presented high-power oscillator features excellent power stability (<0.6% deviation) across the investigated tuning range (1064-1073 nm), high polarization-extinction in the output (>18 dB PER) and excellent spectral (FWHM <13 GHz) and spatial (M2 = 1.2) brightness.
2. Experimental setup
The fiber laser setup is depicted in Fig. 1 . A commercially available Yb-doped PM fiber (Nufern PLMA-YDF-20/400-VIII) was used in the experiments. The conventional step-index geometry with 20 µm (19 µm) core (mode field) and 400 µm and cladding diameter with corresponding numerical apertures of 0.06 and 0.46 permitted efficient single-mode operation through fiber coiling (14 cm diameter). Hence, the suppression of higher order modes together with a limited fiber length of 8 m, which still provided 12 dB of pump absorption, prevented any detrimental self-pulsing effects.
Pump power was delivered by one 976 nm multibar pump diode (LIMO180-F200-DL976-EX1224) with sufficient brightness (delivery fiber: diameter 200 µm, NA 0.22 NA) to reach 96% launch efficiency, only reduced by the 3.6% Fresnel reflection of the perpendicular cleaved fiber end facet which served as outcoupling mirror in the laser cavity. The other fiber end was angle-cleaved at 8°, and the cavity was further containing an aspheric collimating lens, a thin film polarizer and the TCVBG. The focal length of the lens was chosen to be 12 mm, in order to minimize the lasers signal bandwidth. This will be discussed in more detail in section 2.1. The grating possessed a linear chirp of 0.35 nm/mm amounted in a tuning range from 1064 to 1073 nm within the available grating aperture of 5 mm x 27 mm.
In order to avoid additional optical elements for polarization control in the cavity, the slow axis of the polarization-maintaining fiber was oriented parallel to the transmitting axis of an intra-cavity thin film-polarizer (>20 dB of polarization extinction), by imaging the fiber end facet with a red alignment laser. Lasing operation on the slow axis was chosen since it suffers less coiling losses.
2.1. TCVBG spectral performance versus beam size on grating
As mentioned earlier, the focal length of the used lens inside the cavity and thus the collimated beam size in the cavity will define the gratings spectral response. According to Hellstrom et al , the total power reflectivity of a finite beam in a volume Bragg grating can be calculated by treating the beam as a plane wave as long as the beam waist w0 fulfills the conditionFigure 2 shows the excellent agreement of experimental and calculated data for the grating reflectivity for two different beam diameters 1.24 and 1.84 mm. Moreover, Fig. 2b) illustrated how the power reflectivity is altered for a range of different beam diameters.
Apparently, the diffraction efficiency will increase with decreasing beam sizes. However, as previous works [6–10] indicated the linewidth of lasers using VBGs as cavity delimiters will be determined by how much stronger the losses for longitudinal modes lying off-center in the gratings reflection band are compared to losses of those modes in the middle of the reflection band. Even small differences will prevent off-center modes from oscillating due to gain competition. In the case of the used TCVBG this line-narrowing effect will therefore work less efficient for small beam sizes for which the reflection profile approaches a more flat top profile characteristic for unchirped gratings. The choice of beam size on the grating was therefore a trade-off between maximizing the diffraction efficiency and minimizing the signal bandwidth. For our experiments we chose a lens with a focal length of 12 mm resulting in a measured collimated beam diameter on the grating of 1.2 mm, which still provided a high diffraction efficiency of 93% for the beam as well as the sharpest reflection peak. It should be noted that the diffraction efficiency slightly varied over the tuning range with an average value of 91.6% (compare Fig. 4).
2.2. Expected performance of the setup
To evaluate the performance of the TCVBG in the cavity, we used a numerical rate equation model described elsewhere . With the help of the model we simulated the laser output power for different diffraction efficiencies and oscillating wavelengths at the maximum available pump power of 140 W. The general result was that the power will be essentially constant over the tuning range. To visualize this we averaged the calculated output powers and plotted the corresponding relative deviation of the results from this mean value in Fig. 3(a) ). It is noteworthy, though rather obvious due to the high gain system the fiber laser represents, that using the grating with 93% diffraction efficiency instead of close to a 100% will only marginally lower the maximum output power by approximately 1%. In addition to that the expected power stability for a constant diffraction efficiency is better than 0.3%, which is unexpectedly low considering that the change in quantum efficiency λpump / λsignal across the tuning range is >0.4%, compare Fig. 3(b)).
However, considering the fact that at longer wavelengths the cavity losses are compensated at lower population inversion levels in the gain fiber, it becomes clear that the resulting increased pump absorption partially counteracts the drop in slope efficiency caused by the increasing quantum efficiency.
3. Results and discussion
To compare the numerical result with the experimental data the wavelength dependency of the laser output power is plotted at three different pump levels in Fig. 4 . The displayed results show a relative output power variation of less than 1.2% at the lowest power and less than 0.6% at the highest measured power over the whole tuning range.
Although marginally varying in output power, the overall slope efficiency at all oscillating wavelengths is above 78% with a pump threshold of just over 4 W. A typical graph for output power versus launched pump power, recorded at 1068 nm, is shown in Fig. 5(a) ) where no evidence of rollover is present.
As explained in section 2.1 special care was given to the choice of beam diameter on the grating, which optimized the spectral selectivity of the grating. This amounted in a measured signal bandwidth (see Fig. 5(b))) of 50 pm, which is significantly narrower than the grating's FWHM reflection bandwidth of 450 pm (see Fig. 2). Being especially important for the application as tunable pump source for nonlinear frequency conversion schemes, the line narrowing is maintained across the tuning range as displayed in Fig. 5(c)). In addition to that, the noise background is suppressed with more than 50 dB.
Furthermore, the polarization properties of the laser output were measured with a polarizer and a λ/2-plate. Within the investigated wavelength range the laser operated in a single polarization with more than 18 dB suppression of the opposite polarization (see Fig. 5(d))). Moreover, the long-term power stability of the laser and the spatial beam quality was evaluated at the maximum output power. The stability measurement was conducted with a fast photo diode (10 MHz) over a period of 3 hours, during which the relative standard deviation of the signal was less than 0.2% which was attributed to fluctuations in pump power. Using the knife-edge method for beam quality measurements a M2 value of less than 1.2 was determined, suggesting excellent spatial properties and focusibility.
3.1. Potential power scaling
In terms of possible power scaling the results from Xiao et al , who were using a similar fiber geometry, suggest that the output power should at least be scalable into the kW-regime. Although, the damage threshold of the photo-thermo-refractive fused silica used for the grating is lowered by approximately 30% , power-scaling beyond the kW-level will obviously not be limited by the gratings damage threshold, since the beam size on the grating is increased by a factor of approximately 4000 with respect to the mode field in the fused silica fiber.
However, earlier works [7, 15] raised concerns that thermal effects could potentially degrade the spectral performance of VBGs at higher power levels. Jelger et al.  observed a slight shift in the operating wavelength of a high-power VBG-locked laser, which was attributed to the non-uniform heating of the grating due to partial absorption of the incident laser beam.
Although measured close to the resolution limit (10 pm) of our optical spectrum analyzer, a similar shift in the lasers operating wavelength of about 20 pm between low power and high power operation could also be observed in our experiments. Coefficients (from ) for thermal expansion α = 8.5 ppm/K and thermal dispersion dn/dT = 0.5 ppm/K of photo-thermo-refractive fused silica give rise to the temperature dependence of the Bragg wavelength of roughly 9.5 pm/K around 1068 nm, suggesting a moderate temperature increase within the grating from low to high power operation of approximately 2 K. By devising a 3-dimensional FEM heat transfer model to study the temperature distribution within the grating at different power levels, the absorption of the laser beam in the grating could be estimated to be 1.5*10−3 cm−1. Here, according to results from the earlier mentioned rate equation model, the power incident on the grating was modeled to be a fifth of the eventual laser output power.
Furthermore, the experimental cooling conditions were modeled by assuming convective cooling from the surrounding air with a heat transfer coefficient of 10 W/m2K.
Results of the transversal temperature distribution on the gratings entrance facet as well as the distribution along the beam axis are given in Fig. 6(a) ), where a beam diameter of 1.2 mm and an incident power on the grating of 20 W were assumed. It is evident that the temperature distribution and thus the chirp of the Bragg wavelength (<5 pm) within the beam area is negligible compared to the predesigned grating chirp of 0.35 nm/mm. Therefore no degradation of the gratings spectral performance is expected at the investigated power levels. However, targeting higher output powers will require reduced absorption in the grating to lower the induced temperature gradients. Since Lumeau et al.  have demonstrated an absorption coefficient as low as 1*10−4 cm−1 in high-efficiency reflecting Bragg gratings, this technological obstacle is already overcome. Assuming this lower absorption coefficient, even smaller beam diameters on gratings with increased chirp are manageable. To that end Fig. 6(b)) displays temperature distributions identical to 6a) for a grating with low absorption, 1*10−4 cm−1, a ten-fold decreased beam diameter 120µm and a 30 times increased laser output power.
The temperature gradients within the beam volume are still minimal, even though only passive cooling was assumed. It is therefore conceivable that potential power scaling of the high power fiber oscillator will rather be limited by the gain fiber than the VBG used for wavelength-locking.
Referring to an equally simple continuous wave oscillator setup, Jeong et al.  devised a criterion for avoiding fiber damage which requires that the core temperature should not exceed 200 °C, when the fiber jacket is actively kept at room temperature (e.g. by water-cooling). Using this criterion and a FEM heat transfer model we estimated the maximum acceptable thermal load to 250 W/m along the fiber. Assuming an optical-to-optical conversion efficiency of 80%, this suggests that successful thermal management of up to 2 kW output power should be possible in our 8 m long fiber. Since both threshold powers for Raman and stimulated Brillouin scattering in the used fiber (gR = 10−13 m/W, gB = 5*10−11 m/W, ΔνB = 35.6 MHz) at the measured signal bandwidth of 50 pm are roughly 6 kW , a lowered doping concentration and extension of the fiber length to 14 m, could match both the nonlinear and the thermal limit for further power scaling to 3.5 kW. However, it has to be pointed out that increased thermally induced mode coupling to higher order transversal modes  and the subsequent losses, might degrade the laser efficiency as well as lower the suggested thermal limit for power scaling.
3.2. Extension of tuning range
Evidently, the design of the grating can be adapted in order to cover wider tuning ranges. In the case of the fiber used in the foregoing experiments, efficient lasing from 1030 to 1120 nm can be supported using the described cavity configuration. This extension in tuning range by one order of magnitude could either be achieved by increasing the employed grating aperture or increasing the chirp rate. According to section 2.1 an increased chirp rate would require to use a reduced beam size on the grating. In particular, increasing the chirp rate from 0.35 to 3.5 nm/mm would require the beam diameter to be decreased from 1.2 mm to 120 µm in order to maintain diffraction efficiencies over 90%. This still corresponds to a 40:1 relay-image from the fiber mode on to the VBG, which excludes the VBG as a limiting factor with regard to power scaling.
Although the extension of the tuning range can be easily realized, the preservation of the oscillator’s prominent power stability would require dealing with a significantly larger gain variation. One possible approach to address this can be directly derived from Fig. 6, and builds on the wavelength dependent tailoring of the grating’s diffraction efficiency in order to keep constant output power levels under tuning. For the two plotted pump powers 140 W and 1400 W, steady output powers can be achieved at the most for 104 W and 1050 W, respectively. However, the comparison of Fig. 7(a) and 7(b) illustrates that such a solution will only work perfectly for one specific pump level, as the gain maximum shifts to shorter wavelengths with increasing pump powers. Nonetheless, the comparison of the aforementioned figures also shows that the discussed deviations from perfectly steady output power will be less than 1% even for a change in pump level of one order of magnitude.
We have demonstrated a polarized Yb-fiber laser oscillator with an output power exceeding 100 W, whose central wavelength is tunable over 2.5 THz from 1064 to 1073 nm without significant change in output power (<0.6%). Wavelength-locking was performed with a transversely chirped volume-Bragg grating which allowed simple alignment and tuning as well as a limited footprint in a very compact setup. The relay-imaging conditions for the fiber mode on the grating were optimized in order to maximize the spectral selectivity of the grating whilst ensuring high-reflective operation, with the result of a spectrally narrowed laser output signal with a 50 pm linewidth. Moreover the laser exhibited a high slope efficiency of 78% and broadband background suppression of more than 50 dB.
Evidently, the presented method offers a high degree of flexibility and is easily extendable to higher powers or to encompass wider or other tuning ranges, within the gain media's emission band. In particular, we have shown that a 10-fold extension of the presented tuning range could be achieved without compromising any other laser properties. The proposed design is comprised of an increased chirp rate and a specifically tailored reflectivity profile of the grating.
The authors thank the Linneus Centre ADOPT, the Swedish Research Council (VR), and the Acreo Optic Fiber Center (AFOC) for the received financial support.
References and links
1. D. Richardson, J. Nilsson, and W. Clarkson, “High power fiber lasers: current status and future perspectives [Invited],” J. Opt. Soc. Am. B 27(11), B63–B92 (2010). [CrossRef]
2. Y. Jeong, A. Boyland, J. Sahu, S. Chung, J. Nilsson, and D. Payne, “Multi-kilowatt Single-mode Ytterbium-doped Large-core Fiber Laser,” J. Opt. Soc. Korea 13(4), 416–422 (2009). [CrossRef]
4. Y. Xiao, F. Brunet, M. Kanskar, M. Faucher, A. Wetter, and N. Holehouse, “1-kilowatt CW all-fiber laser oscillator pumped with wavelength-beam-combined diode stacks,” Opt. Express 20(3), 3296–3301 (2012). [CrossRef]
5. C. Goh, S. Set, K. Kikuchi, M. Mokhtar, S. Butler, and M. Ibsen, “Greater than 90 nm continuously wavelength-tunable fibre Bragg gratings,” Optical Fiber Communications Conference,2003. OFC 2003, vol., no., pp. 643- 644 vol.2, 23–28 March 2003 URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=1248468&isnumber=27940
6. J. W. Kim, P. Jelger, J. K. Sahu, F. Laurell, and W. A. Clarkson, “High-power and wavelength-tunable operation of an Er,Yb fiber laser using a volume Bragg grating,” Opt. Lett. 33(11), 1204–1206 (2008). [CrossRef]
7. P. Jelger, P. Wang, J. K. Sahu, F. Laurell, and W. A. Clarkson, “High-power linearly-polarized operation of a cladding-pumped Yb fibre laser using a volume Bragg grating for wavelength selection,” Opt. Express 16(13), 9507–9512 (2008). [CrossRef]
8. T. S. McComb, R. A. Sims, C. C. Willis, P. Kadwani, V. Sudesh, L. Shah, and M. Richardson, “High-power widely tunable thulium fiber lasers,” Appl. Opt. 49(32), 6236–6242 (2010). [CrossRef]
11. J. E. Hellstrom, B. Jacobsson, V. Pasiskevicius, and F. Laurell, “Finite Beams in Reflective Volume Bragg Gratings: Theory and Experiments,” IEEE J. Quantum Electron. 44(1), 81–89 (2008). [CrossRef]
13. B. Jacobsson, V. Pasiskevicius, F. Laurell, E. Rotari, V. Smirnov, and L. Glebov, “Tunable narrowband optical parametric oscillator using a transversely chirped Bragg grating,” Opt. Lett. 34(4), 449–451 (2009). [CrossRef]
14. O. M. Efimov, L. B. Glebov, S. Papernov, and A. W. Schmid, “Laser-induced damage of photo-thermo-refractive glasses for optical holographic element writing,” Proc. SPIE 3578, Laser-Induced Damage in Optical Materials 1998, 564 (1999).
16. O. Andrusyak, V. Smirnov, G. Venus, V. Rotar, and L. Glebov, “Spectral Combining and Coherent Coupling of Lasers by Volume Bragg Gratings,” IEEE J. Sel. Top. Quantum Electron. 15(2), 344–353 (2009). [CrossRef]
17. J. Lumeau, L. Glebova, and L. B. Glebov, “Near-IR absorption in high-purity photothermorefractive glass and holographic optical elements: measurement and application for high-energy lasers,” Appl. Opt. 50(30), 5905–5911 (2011). [CrossRef]
18. G. P. Agrawal, Nonlinear Fiber Optics, 3rd ed. (Academic Press, 2001).