Chinese Optics Letters, Volume. 23, Issue 6, 061406(2025)

Monolithic design of a linearly polarized single-mode Tm-doped fiber laser at 1908 nm with 207 W output

Jinwen Tang, Xiaoxiao Hua, Wenhao Cheng, Minglang Wu, Disheng Wei, Junhui Li, Runming Zhang, Baoquan Yao*, Tongyu Dai, and Xiaoming Duan
Author Affiliations
  • National Key Laboratory of Laser Spatial Information, Harbin Institute of Technology, Harbin 15001, China
  • show less

    We report a high-power, all-fiber Tm-doped laser system operating at 1908 nm, based on a master-oscillator power amplifier (MOPA) configuration. The oscillator utilizes two polarization-maintaining (PM) fiber Bragg gratings (FBGs) with orthogonal principal axes to achieve single-polarization output. The system generates a linearly polarized output power of 12.6 W, with a slope efficiency of 40.6%. The power is subsequently scaled to 207.6 W through a primary amplifier, which uses a large mode area (LMA) fiber while maintaining single-mode operation. The amplifier achieves a beam quality factor (M2) of 1.36 and a polarization extinction ratio (PER) exceeding 18.3 dB.

    Keywords

    1. Introduction

    Thulium-doped fiber lasers (TDFLs) offer a broad emission range in the near-infrared spectrum (1660–2200 nm)[1-3], covering an important eye-safe spectral region suitable for diverse applications, including medical surgery[4], free-space optical communication, plastic material processing[5], gas sensing[6], LIDAR[7], and nonlinear frequency conversion to the mid-infrared band[8]. To meet the demands of these diverse applications, different fiber designs and pumping schemes have been developed to access the full emission wavelength range.

    Over the past decades, significant progress has been made in scaling the output power of TDFLs[9,10]. Emission wavelengths above 1940 nm are generally favored due to their higher effective gains in TDFLs. In contrast, operation at shorter wavelengths typically results in reduced output power and efficiency. Tm-doped fiber lasers operating near 1908 nm are primarily used for pumping solid-state and fiber lasers, as well as certain medical applications. At this wavelength, which coincides with the absorption peak of Holmium, theoretical quantum efficiencies as high as 90.8% can be achieved for the generation of 2.1 µm lasers. High-brightness, short-wavelength Tm fiber lasers operating at 1.908 µm have also been used for resonant pumping of Tm-doped fibers to generate 2 µm lasers, demonstrating significant potential for resonant tandem-pumping applications[11,12]. To date, the maximum reported output power for a 1908 nm TDFL is 308 W, achieved using an oscillator configuration with random polarization, which effectively mitigates the impact of ground-state absorption and parasitic lasing effects[13]. In contrast, achieving power scaling at this wavelength with fiber amplifiers remains extremely challenging, let alone generating linearly polarized output. Furthermore, there have been very few reports on polarization-maintaining (PM) output from TDFLs[14,15]. The highest reported output power of a master-oscillator power amplifier (MOPA) Tm-doped fiber laser is 253 W, achieved through spectral beam combining[16].

    In this paper, we report a monolithic Tm-doped all-fiber master-oscillator power amplifier (MOPA) system operating at 1908 nm, achieving 207 W of single-mode, linearly polarized output power. This high output power, achieved with 790 nm laser diode (LD) pumping, represents the highest reported for a linearly polarized Tm-doped fiber laser at this wavelength, to the best of our knowledge. The high polarization stability is notably achieved by independently temperature-tuning the fiber Bragg gratings (FBGs), eliminating the need for additional polarization-selective elements.

    2. Fiber Bragg Gratings and Polarization Control

    The FBGs used in this experiment were inscribed in double-clad panda-type PM passive fiber (IXF-2CF-PAS-10-130-0.15, IXBLUE) via UV recording. One grating had a high reflectivity (HR) of 99.0%, while the other had a low reflectivity (LR) of 10.3%. This type of fiber exhibits strong birefringence due to its asymmetric stress distribution, resulting in two distinct Bragg wavelengths[17]. The separation between these wavelengths is related to the birefringence (B) and the grating period (Λ):Δλ=2Λ·B. To measure the transmission spectra, we separately recorded the spectra of the HR-FBG and LR-FBG along the slow and fast axes at room temperature, using a polarized spontaneous emission source with the output port fused to a PM isolator. The center wavelengths and bandwidths of the HR-FBG were λHRS=1906.6nm (0.2 nm) and λHRF=1907.0nm (0.3 nm), respectively. For the LR-FBG, the center wavelengths and bandwidths were λLRS=1906.8nm (0.2 nm) and λLRF=1907.2nm (0.3 nm). Thus, for this PM passive fiber, with a grating period Λ of 660 nm, the value of B is calculated to be 3×104.

    Figure 1 illustrates the alignment and splicing configurations of the HR-FBG and LR-FBG. When the principal axes of the HR-FBG and LR-FBG are aligned, as shown in Fig. 1(a), the Bragg wavelengths for both the slow and fast axes are aligned, forming resonators along both the X- and Y-axes. This results in two laser wavelengths with orthogonal polarizations, thereby preventing a linearly polarized output.

    (a) Alignment of FBG principal axes, forming dual-polarization resonators. (b) Orthogonal splicing of FBGs to enable single-polarization oscillation by suppressing the orthogonal mode.

    Figure 1.(a) Alignment of FBG principal axes, forming dual-polarization resonators. (b) Orthogonal splicing of FBGs to enable single-polarization oscillation by suppressing the orthogonal mode.

    In contrast, as depicted in Fig. 1(b), when the principal axes of the HR-FBG and LR-FBG are orthogonally spliced, the fast-axis Bragg wavelength of one FBG matches the slow-axis Bragg wavelength of the other. This configuration enables selective oscillation in a single polarization state. For this operation, both FBGs typically need narrower bandwidths than Δλ for this operation. Otherwise, precise temperature control is necessary to achieve single polarization output.

    3. Experimental Setup

    The experimental setup for the all-fiber, linearly polarized TDFL is shown in Fig. 2, comprising an oscillator and an amplifier. The oscillator is externally pumped by a 30 W fiber-coupled LD operating at 793 nm, utilizing a PM (6+1)×1 signal and pump combiner. The resonator comprises two PM FBGs and 1.8 m of PM single-mode Tm-doped double-clad fiber (SM-TDF). The SM-TDF has a 10 µm core diameter (NA, 0.15) and a 130 µm cladding diameter (NA, 0.46), with a cladding absorption coefficient of 7.6 dB/m at 793 nm (IXF-2CF-Tm-PM-10-130-V1, IXBLUE). The LR-FBG was cross-spliced with its fast axis aligned to the slow axis of the PM-TDF. Residual cladding pump light was removed using a PM cladding pump stripper (CPS), which was also cross-spliced to ensure that the linearly polarized seed laser propagated along the slow axis. A PM optical isolator (ISO) was fusion-spliced along the axis to prevent amplifier feedback. All fibers, including the CPS and ISO, were configured in a 10/130 µm double-clad structure, mounted in a water-cooled aluminum heatsink, maintained at 15°C, with both FBGs temperature-tuned via thermoelectric coolers (TECs) to compensate for temperature-induced wavelength shifts.

    Schematic diagram of the PM Tm-doped all-fiber laser architecture, comprising an oscillator with 10/130 µm fiber and a power amplifier with 15/250 µm fiber, operating in single-mode at 1908 nm.

    Figure 2.Schematic diagram of the PM Tm-doped all-fiber laser architecture, comprising an oscillator with 10/130 µm fiber and a power amplifier with 15/250 µm fiber, operating in single-mode at 1908 nm.

    The amplifier uses fiber with a core diameter of 15 µm and a cladding diameter of 250 µm, with an NA of 0.095/0.46 for the passive fiber and 0.1/0.46 for the TDF. Therefore, a mode field adapter (MFA) is employed to match the oscillator’s output to the amplifier’s input fiber. The amplifier is pumped by five 793 nm fiber-coupled multimode LDs, each delivering over 100 W, with a 200 µm core (NA, 0.22) and a 220 µm cladding (NA, 0.46). These LDs are combined using a PM (6+1)×1 combiner, providing a total pump power of approximately 475 W with 95% coupling efficiency. A PM large-mode-field TDF, with a cladding absorption of 2.6 dB/m at 793 nm, is spliced to the combiner output and coiled to a 15 cm diameter in an aluminum heatsink maintained at 12°C. Unabsorbed pump power is dissipated by a 15/250 PM CPS. To reduce Fresnel reflections, prevent backward power, and mitigate parasitic oscillations, the fiber ends in both the oscillator and amplifier were cleaved at an 8° angle.

    4. Results and Discussion

    The oscillator’s output power and backward power after the 10/130 µm CPS were evaluated, as shown in Fig. 3(a). The laser reached threshold operation at a pump power of 4.1 W, achieving a maximum output of 11.4 W at 29.9 W pump power, with a slope efficiency of 42.8% relative to the incident pump power. The backward power measured at the input of the 10/130 combiner increased significantly with rising pump power, peaking at 1.3 W at maximum output. The oscillator experienced severe damage due to a sudden surge in backward power. Subsequent analysis attributes this phenomenon to the combined effects of thermal disturbances and stimulated Brillouin scattering (SBS).

    (a) Output power and backward power versus launched pump power. (b) Laser spectra at oscillator PERs of 18.8 dB (solid line) and 1.8 dB (dashed line).

    Figure 3.(a) Output power and backward power versus launched pump power. (b) Laser spectra at oscillator PERs of 18.8 dB (solid line) and 1.8 dB (dashed line).

    Thermal effects caused by pump light scattering reduced the overlap of peak reflectivities of the FBGs, a critical factor for stable laser operation. This reduction led to a narrower laser output linewidth, which in turn increased the resonator’s Q-factor and lowered the threshold for instabilities. The decreased overlap contributed to the rise in backward power by reducing coupling efficiency within the cavity. This misalignment of reflectivity peaks resulted in higher levels of reflected light, which could not be effectively utilized, intensifying feedback and limiting power scalability.

    Precise FBG matching is essential, especially at higher power levels, where thermal effects significantly impact the stability of the oscillator. Thermal effects shift the FBG reflection wavelengths, further reducing the overlap of reflectivities and exacerbating instabilities. As backward power increases rapidly, instabilities such as relaxation oscillations and heightened relative intensity noise (RIN) manifest. In high-gain systems, self-Q-switching effects influenced by SBS may occur, significantly increasing the potential for oscillator damage[18].

    Several strategies can be employed in the oscillator design to reduce backward power and enhance system stability. One potential approach is the use of broader-bandwidth FBGs. Broader FBG bandwidths would help maintain the overlap of reflectivities across different power levels, thereby reducing the risk of instability. Additionally, using fibers with higher birefringence enhances polarization control, preventing polarization mode coupling and reducing instability and nonlinear effects. Furthermore, reducing the gain fiber length in the oscillator could further minimize the amplification of backward power, contributing to overall system stability.

    Moreover, accurate temperature control of the FBG pairs is crucial for ensuring proper wavelength matching, which can be effectively maintained using TECs. This is further demonstrated by the polarization extinction ratio (PER) shown in Fig. 4. Under most temperature settings, the PER remains high, with an average of 16.7 dB and a peak of 18.8 dB. However, significant polarization degradation (down to 1.8 dB) occurred when the HR-FBG was set to 15°C and the LR-FBG to 32.5°C. Figure 3(b) compares the laser spectra under two PER conditions: 18.8 and 1.8 dB. At the higher PER of 18.8 dB, the oscillator exhibits single-wavelength output with a central wavelength of 1907.7 nm at maximum power and a full width at half-maximum (FWHM) of 0.19 nm. In contrast, at a PER of 1.8 dB, the spectrum shows two peaks at 1907.0 and 1907.4 nm. This result indicates reduced reflectivity overlap between the HR fast axis and the LR slow axis, allowing simultaneous oscillation along the second axis. To relax the strict temperature control requirements, selecting FBG pairs with lower thermal drift coefficients could allow for more flexible matching conditions.

    Polarization of the oscillator measured under varying HR-FBG and LR-FBG temperature settings. The temperature range for both HR-FBG and LR-FBG spans 15°C–35°C, controlled using TECs.

    Figure 4.Polarization of the oscillator measured under varying HR-FBG and LR-FBG temperature settings. The temperature range for both HR-FBG and LR-FBG spans 15°C–35°C, controlled using TECs.

    The seed oscillator was then injected into the power amplifier to facilitate effective power scaling. To achieve higher output power, the amplifier employed a large mode area (LMA) fiber with a 15 µm core diameter and a 250 µm cladding diameter, providing two primary advantages: 1) The normalized frequency V=2πaNA/λ of the fiber is calculated to be 2.3475 at 1908 nm, which indicates that the fiber supports the propagation of only the fundamental mode; 2) It enhances laser efficiency by maintaining an optimal mode field diameter (MFD)-to-core diameter ratio, ensuring effective spatial overlap between the fundamental mode and the pumped fiber core. This reduces efficiency losses caused by inadequate overlap. The MFD of the single-mode TDF, calculated using Marcuse’s formulas, is approximately 10% larger than the core diameter, offering higher efficiency compared to larger LMA fibers, such as those with a 25 µm core and 400 µm cladding[13].

    We systematically investigated fibers of varying lengths, and the results are presented in Fig. 5. The figure illustrates the average slope efficiency as a function of active fiber length, maintaining consistent values above 50% for fibers longer than 3.4 m. In most cases, the maximum output power was limited by fiber fuse initiation, which occurred 0.3 to 0.6 m from the splice at the pump end. These maximum power levels, also shown in Fig. 5, typically ranged from 100 to 200 W, with a peak value of 207 W. In some instances, amplified spontaneous emission (ASE) and parasitic oscillations further constrained the maximum power. Notably, the threshold for parasitic oscillations decreased rapidly as fiber length increased. These oscillations were primarily caused by two factors: 1) feedback at the refractive index interface between active and passive fibers, and 2) reflections at the fiber ends.

    Average slope efficiency as a function of active fiber length in the MOPA configuration, along with maximum output power levels at which fiber fuse initiated.

    Figure 5.Average slope efficiency as a function of active fiber length in the MOPA configuration, along with maximum output power levels at which fiber fuse initiated.

    The laser spectra for active fiber lengths of 3.4, 3.1, and 2.7 m were measured using an optical spectrum analyzer with a resolution of 0.75 pm (Model 771B-IR, Bristol Instruments), as shown in Fig. 6. The 1986 nm peak observed in the MOPA outputs is attributed to parasitic oscillation. ASE contribution from thulium is minimal, likely due to population inversion depletion by parasitic oscillation, which suppresses ASE generation. Alternatively, the limited sensitivity of the spectrometer may prevent the detection of ASE components.

    Optical spectra of the amplifier measured at 81 W for a 3.4 m fiber, 147 W for a 3.1 m fiber, and 205 W for a 2.7 m fiber.

    Figure 6.Optical spectra of the amplifier measured at 81 W for a 3.4 m fiber, 147 W for a 3.1 m fiber, and 205 W for a 2.7 m fiber.

    As the active fiber length increases, the threshold for parasitic oscillation decreases. For the shortest fiber length of 2.7 m, the laser spectrum remains highly pure, even at an output power of approximately 205 W. This indicates that shorter fiber lengths are advantageous for maintaining spectral purity at higher power levels by effectively raising the threshold for parasitic oscillation.

    The experimental results for a 2.7 m LMA TDF are shown in Fig. 7, which presents the output power as a function of both launched and absorbed pump powers. The left panel illustrates the relationship between output power and both launched and absorbed pump powers, while the right panel displays the slope efficiency and optical efficiency. The slope efficiency, derived from the derivative of a polynomial fit to the output power versus launched pump power, generally increased at lower pump powers and showed a slight decline as the output power approached its maximum. Similarly, the optical efficiency, defined as the ratio of output power to launched pump power, exhibited a similar trend. In this experiment, a record output power of 207.6 W was achieved with a maximum incident pump power of 472 W, corresponding to an optical efficiency of 42.6% and a slope efficiency of 45.7%. For absorbed pump power, the slope efficiency was 58.4%, indicating suboptimal absorption efficiency in the active fiber. The measured absorption efficiency, obtained before the CPS using a dichroic mirror, was less than 75%. Notably, the optical efficiency showed no signs of decline, and neither parasitic oscillations nor ASE were observed in the spectrum, highlighting the significant potential for further power scaling.

    MOPA output power as a function of launched and absorbed pump powers, with corresponding slopes and optical efficiencies shown for both cases.

    Figure 7.MOPA output power as a function of launched and absorbed pump powers, with corresponding slopes and optical efficiencies shown for both cases.

    The M2 parameter of the laser beam was measured using the 10/90 knife-edge method. At an output power of 50 W, M2 was 1.07 in the x direction and 1.09 in the y direction, increasing to 1.20 and 1.22 at 150 W, and further to 1.35 and 1.36 at powers exceeding 200 W. The M2 measurements are presented in Fig. 8. The corresponding beam profiles at each power level, measured using a beam analyzer (Pyrocam IV, OPHIR), are shown in the insets of Fig. 8. The beam profiles demonstrate changes in beam quality with increasing power, where the Gaussian-like distribution at lower powers reflects high beam quality. The gradual decline in beam quality at higher powers is attributed to mode coupling and thermal effects. Additionally, the slight corner observed in the beam profile is likely caused by the output emerging from a bare fiber end cut at an 8° angle, which may introduce asymmetry into the beam shape.

    M2 values and beam profiles (insets) at output powers of 50, 150, and ∼200 W.

    Figure 8.M2 values and beam profiles (insets) at output powers of 50, 150, and ∼200 W.

    The output from the MOPA passed through a rotating half-wave plate to adjust the linear polarization orientation, ensuring minimal power transmission through the TFP, which is designed for HR of vertically polarized light and HT of horizontally polarized light at 1.9 µm. The transmitted and reflected powers from the TFP were continuously monitored to assess the long-term stability of the laser output and the PER. The long-term output data, collected over a 20 min period at a sampling rate of 15 Hz, is presented in Fig. 9. The average output power was recorded as 204.4 W, while the average PER was measured at 18.3 dB. To evaluate the stability of the laser system, the root mean square (RMS) stability of the output power and PER were calculated as 0.23% and 1.88%, respectively. The RMS stability was calculated using the formula RMS=STD/Pavg, where STD represents standard deviation of each individual measurement, and Pavg is the average value of the output power or PER, depending on the parameter being analyzed. These results indicate a stable output of linearly polarized light from the MOPA system, with minimal performance fluctuations. The slight variations in output power are likely attributed to thermal management of the pump LDs, while fluctuations in the PER may result from additional mechanical stress, thermally induced birefringence, or polarization crosstalk within the system. Moreover, no significant nonlinear effects, such as stimulated Raman scattering or ASE, nor any signs of gain saturation, were observed at the tested power levels. This demonstrates that the amplifier system has substantial potential for further power scaling without compromising beam quality or polarization stability.

    Long-term power stability and PER measurements. The inset shows the histogram of the power data.

    Figure 9.Long-term power stability and PER measurements. The inset shows the histogram of the power data.

    To further enhance the stability of the PER, single-polarization output can be optimized through fiber bending. The difference in propagation constants within a MOPA fiber, known as modal birefringence (Sp=neno), where ne represents the effective refractive index for the electric field aligned with the fiber’s principal axis, while no corresponds to the effective refractive index for the orthogonal electric field. Modal birefringence is influenced by several factors, including the refractive index contrast among the core and cladding, the geometric structure of the fiber, material properties, environmental conditions (such as temperature and stress), and the wavelength of the light. When the fiber is bent, the effective refractive index of a specific polarization mode changes, resulting in increased loss for that mode while other modes experience relatively lower losses. By precisely controlling the bending radius and angle, selective attenuation of the desired polarization mode can be achieved, leading to single-polarization light output[19]. For the 2.7 m long 15/250 µm (NA, 0.095) TDF in this research, achieving a one-order-of-magnitude difference in loss between the two polarization axes requires a bending radius of approximately 3.3 cm. However, further reducing the bending radius leads to decreased laser efficiency, making this approach unsuitable for this study. Nevertheless, it offers a potential method for optimizing PER in other LMA fibers, which could serve as a valuable comparison for further research.

    Additionally, fiber bending also contributes to ASE suppression by attenuating higher-order LP modes, thereby enhancing the signal-to-noise ratio and enabling further scalability of output power[20]. Thus, the consideration of fiber bending in the design serves a dual purpose, both as a conceptual approach for PER optimization and in enhancing system performance by minimizing ASE, supporting stable, high-quality output across power levels.

    5. Conclusion

    In summary, a high-power, all-fiber polarized light source operating at 1908 nm was successfully developed using a MOPA configuration, achieving stable, single-mode, linearly polarized output across both moderate and high power levels. The oscillator delivered 12.6 W with a slope efficiency of 40.6%, which was scaled up to 207 W through the power amplifier while maintaining excellent beam quality (M2=1.36) and a PER exceeding 98.5%. Notably, the system sustained single-mode operation throughout, even with an LMA fiber, and exhibited a stable Gaussian-like beam profile with minimal fluctuations in both output power and PER. These results demonstrate the system’s capability for power scaling without compromising beam quality or polarization stability, making it a promising solution for high-power applications requiring high-quality, linearly polarized light.

    [5] I. Hartl, A. L. Carter, J. Neumann et al. All-fiber linearly polarized high power 2-μm single mode Tm-fiber laser for plastic processing and Ho-laser pumping applications. Fiber Lasers XV: Technology and Systems(2018).

    [9] M. Meleshkevich, N. Platonov, D. Gapontsev et al. 415 W single-mode CW thulium fiber laser in all-fiber format. CLEO/Europe and IQEC 2007 Conference Digest, CP2_3(2007).

    Tools

    Get Citation

    Copy Citation Text

    Jinwen Tang, Xiaoxiao Hua, Wenhao Cheng, Minglang Wu, Disheng Wei, Junhui Li, Runming Zhang, Baoquan Yao, Tongyu Dai, Xiaoming Duan, "Monolithic design of a linearly polarized single-mode Tm-doped fiber laser at 1908 nm with 207 W output," Chin. Opt. Lett. 23, 061406 (2025)

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Lasers, Optical Amplifiers, and Laser Optics

    Received: Dec. 2, 2024

    Accepted: Jan. 17, 2025

    Posted: Jan. 17, 2025

    Published Online: Jun. 5, 2025

    The Author Email: Baoquan Yao (yaobq08@hit.edu.cn)

    DOI:10.3788/COL202523.061406

    CSTR:32184.14.COL202523.061406

    Topics