A 5 kW Near-Single-Mode Oscillating–Amplifying Fiber Laser Employing a Broadband Output Coupler with Simultaneous Raman Suppression and Spectral Narrowing

Abstract

In this work, we propose and demonstrate a novel approach to suppressing stimulated Raman scattering in an oscillating–amplifying integrated fiber laser (OAIFL) by changing the spectral bandwidth of the output-coupler fiber Bragg gratings (OC-FBGs). The reflectance bandwidth of the fiber Bragg grating (FBG) in the oscillating section was systematically investigated as a critical parameter for SRS mitigation. Three types of long-period FBGs with distinct reflectance bandwidths (1.2 nm, 1.3 nm, and 2 nm) were comparatively studied as output couplers. The experimental results demonstrated a direct correlation between FBG bandwidth and SRS suppression efficiency, with the configuration of the OC-FBG with a 2 nm bandwidth achieving optimal suppression performance. Concurrently, the output power was enhanced to 5.02 kW with improved power scalability. And excellent beam quality was obtained with M² < 1.3. Remarkably, in the architecture of this laser, increasing the bandwidth of the output couplers in the oscillating section had a relatively minor effect on the optical-to-optical (O-O) efficiency, which reached up to 78%. Additionally, this modification also reduced the 3 dB bandwidth of the laser output, thereby achieving a beam output with enhanced monochromaticity.

1. Introduction

High-power fiber lasers offer advantages such as excellent beam quality, high power density, superior optical-to-optical (O-O) efficiency, effective thermal dissipation, compact structure, and strong environmental adaptability, making them widely applicable in industrial manufacturing, medical fields, and national defense. Recent breakthroughs in fiber oscillators/amplifiers and device upgrades have significantly enhanced laser performance (output power and beam quality), expanding their application scope.

The key factors limiting power scaling are stimulated Raman scattering (SRS) and transverse-mode instability (TMI) in optical fibers. The primary SRS suppression methods include using large-mode-area (LMA) fibers [1,2], specialty-structured fibers [3], broadband OC-FBGs [4,5], chirped tilted fiber Bragg gratings (CTFBGs) [6,7,8,9], and optimizing pump configurations [10,11]. Studies confirm that broadening the FBG bandwidth or changing the spectrum shape of FBG effectively raises the SRS threshold and reduces the Raman light proportion [12,13]. With the wide application of LMA fibers and the continuous increase in output power, TMI was first discovered in 2010 [14]. TMI mitigation strategies involve enhancing seed power, optimizing pump wavelength or structure, reducing gain fiber photodarkening, and novel fiber designs. In 2018, it was found that SRS could induce a decrease in the TMI threshold in fiber optic oscillators [15]. Recently, photodarkening has emerged as a new power-limiting factor, causing output power degradation during long-term operation (e.g., experiments showing a ~20% power drop after 15,000 h [16]). Suppression approaches include increasing the oscillator section power [17], lowering the Yb³⁺ concentration [18], gas or optical bleaching [19,20], and specialized pumping sources [21,22]. Conducting burn-in tests on the laser and suppressing its photodarkening effect is also important to ensure that the laser can maintain its original working performance under long-term working conditions.

While traditional master-oscillator power amplifier (MOPA) structures achieve a 10 kW output [23], they suffer from complexity, high cost, and challenging thermal management. A pump-sharing MOPA structure [24] fiber laser achieved a power output of 4.8 kW and obtained good beam quality and slope efficiency. The oscillating–amplifying integrated fiber laser (OAIFL) proposed by Hejaz [25] shares pump sources and fiber between sections, eliminating isolators or cladding light strippers (CLS). This combines MOPA’s high-power capability with an oscillator’s compactness. Tsinghua University achieved a 2.19 kW narrow-linewidth output in 2019 [26], and NUDT reached 5 kW in 2021 and 6 kW in 2023 with this structure [27,28] (SRS suppression ratio of ~15 dB, no TMI observed).

However, commercially fabricated 5 kW OAIFLs showed limited SRS suppression (~15 dB). Conventional CTFBG-based suppression typically sacrifices output beam bandwidth, and the O-O efficiency at the 5 kW level is only 74% [29]. Furthermore, photodarkening in OAIFLs remains unstudied. This work suppresses SRS in OAIFLs by broadening the OC-FBG reflection bandwidth, achieving a 5.02 kW output with ~35 dB SRS suppression and M² ≈ 1.29. Broadening the oscillator FBG bandwidth enables a narrower output bandwidth with minimal impact on O-O efficiency.

2. Experimental Setup

The experimental setup of the OAIFL is shown in Figure 1. The oscillating section of the laser is composed of 8 m double-clad ytterbium-doped fiber (DCYDF1) and a pair of FBGs with a central wavelength of 1070 nm. The core/cladding diameter of the oscillating section is 22/400 μm. The pump absorption coefficient of DCYDF1 is 0.3 dB/m at 915 nm. The gain fiber is fixed to the T2 copper alloy fiber winding plate and tightly wound. The winding diameter ranges from 60 mm to 76 mm. Reducing the winding diameter promotes leakage of higher-order modes, thereby improving the beam quality from the oscillating section. The 3 dB bandwidth of the commercial high-reflectivity fiber Bragg grating (HRFBG) is 3 nm, with a central-wavelength reflectivity of 99.5%. The reflectivity of the OC-FBG is 10%. The 3 dB bandwidths of the three types of fiber Bragg gratings used are 1.2 nm, 1.3 nm, and 2 nm, respectively. The reflectance curves are shown in Figure 2. (Due to the different production batches and data processing methods of the three types of gratings, the reflectance curves of the three types of gratings only reflect the corresponding bandwidths of the reflectance on their spectra and do not represent the reflectance levels.) The double-cladding ytterbium-doped fiber (DCYDF2) used for the amplifying section has a larger core size, with a core/cladding diameter of 25/400 μm. The pump absorption coefficient of DCYDF2 is 0.47 dB/m at 915 nm. Using gain fibers with a low absorption coefficient can reduce the influence of photodarkening and increase the TMI threshold of the optical path. Winding the 15 m DCYDF2 around an optical fiber spool with a minimum diameter of 100 mm and a maximum diameter of 110 mm, and with the height of the optical fiber slot being 1 mm and the width 1 mm, helps to minimize the size of the optical fiber spool as much as possible without affecting the heat dissipation of the gain optical fiber. The OC-FBG of the oscillating section is directly spliced with the gain fiber of the amplifying section without splicing ISO or CLS in the middle. This enabled the pump light of the oscillating section and the pump light of the amplifying section to be freely transmitted between the two sections, which can improve the absorption of the pump light and enhance the O-O efficiency of the laser. At the front end of the HRFBG in the oscillating section, an (8 + 1) × 1 forward pump/signal combiner (FPSC) was fused, and at the rear end of the DCYDF2, a (10 + 1) × 1 backward pump/signal combiner (BPSC) was fused. The 8 pump ports of FPSC and the 10 pump ports of BPSC were occupied by fiber pigtail laser diodes (LDs). According to previous studies, shifting the pump wavelength from the absorption peak of the YDF can increase the TMI threshold of the fiber laser, reduce the quantum loss, and also reduce the impact of the photodarkening effect. Here, a specially designed wavelength-stabilized pump source with a central wavelength of 981 nm was adopted for forward and backward pumping. Each LD can provide a maximum power of 500 W in a multimode fiber with a core/cladding diameter of 200/220 μm and a core NA of 0.22. The laser signal generated by the oscillation part increases when passing through the amplifying section. The customized end-cap output tail fiber has a length of 2 m. The signal laser generated by the oscillating section increases as it passes through the amplifying section and is output from the customized end cap. After being output, it passes through the collimation system and is then guided to the measurement system for analyzing power, spectrum, and beam quality. Among them, the reflectivity of the high-reflectivity flat mirror is 99.9%.

3. Experimental Suppression of SRS

3.1. Output Power, SRS, and Slope Efficiency

First, a preliminary experiment was conducted using an OC-FBG with a reflectance bandwidth of 1.2 nm. Under the conditions of a forward pump of 1560 W and a backward pump of 4231 W, a maximum output power of 4.51 kW was obtained. The O-O efficiency was approximately 77.88%, and the signal-to-Raman noise ratio (SRNR) was less than 20 dB, as shown in Figure 3a. To ensure the safety of the fiber laser, we stopped the pump power from increasing further. Subsequently, we attempted to adjust the pump power ratio. By reducing the pump power of the oscillating section to suppress SRS, under the conditions of a forward pump power of 1215 W and a backward pump power of 4806 W, an output power of 4.7 kW was obtained. The SRNR was approximately 20.5 dB, and the O-O efficiency was 78.06%. Then, by increasing the pump power of the amplifying section, the output power decreased, indicating the occurrence of TMI. Subsequently, we replaced the oscillating section with an OC-FBG with a reflectance bandwidth of 1.3 nm. Under the conditions of a forward pumping power of 1560 W and a backward pumping power of 4622 W, a power output of 4.85 kW was obtained, with an O-O efficiency of approximately 78.45%. At this power, the SRNR was 21 dB, and TMI occurred. By comparing Figure 3a and Figure 3b, it can be found that the variation in SRNR with a bandwidth of 1.3 nm OC-FBG is slower than that with a bandwidth of 1.2 nm OC-FBG with an increase in output power. Then, this OC-FBG is replaced with an OC-FBG with a reflectance bandwidth of 2 nm for the oscillating section. Under the conditions of a forward pumping power of 2080 W and a backward pumping power of 4372 W, an output power of 5.02 kW is obtained. The SRNR of the output spectrum is 35 dB, and the O-O efficiency is 77.96%. At this power, we also measured its beam quality, as shown in Figure 4, and the measured beam quality factor M² ≈ 1.29. A good near-single-mode beam output was obtained.

In the traditional MOPA structure, increasing the bandwidth of OC-FBG leads to an increase in longitudinal-mode competition, resulting in a decrease in pump light absorption efficiency. The oscillating spectra that cause enhanced and widened pump light leakage in the oscillating section and amplifying section may cause backlight leakage at the high-reflectivity grating, thereby reducing the O-O efficiency. The power growth curves under different gratings are shown in Figure 5. The 1.2 nm grating has the maximum O-O efficiency at an output power of 3.5 kW. At an output power of 4.5 kW, due to the increase in Raman proportion and output bandwidth, the O-O efficiency is 78.2%. By reducing the power of the oscillating section and decreasing the Raman proportion, the O-O efficiency is 78.05% at 4.7 kW. Thus, it can be seen in Figure 5c,d that the O-O efficiency of the 1.3 nm bandwidth grating is 77.82% at an output power of 4.85 kW, and that of the 2 nm bandwidth grating is 77.73% at an output power of 5.02 kW. Different reflectivity bandwidth-coupled output grating conditions were found to have a relatively small impact on the O-O efficiency, which is attributed to the sharing of pump light in the oscillating section and amplifying section of the OAIFL structure. The pump light absorption efficiency was improved, and the broadband grating reduced the Raman proportion in the output power, ensuring high O-O efficiency. As can be seen from Figure 6, the proportion of Raman in the output power is directly related to the bandwidth of the OC-FBGs. Increasing the bandwidth of the OC-FBG can significantly reduce the proportion of Raman power, and can enhance the Raman threshold. Our results show that the slope of the OC-FBG1 gradually increases at 3.5 kW. It is speculated that the narrower seed spectrum may have fluctuations in the time domain, resulting in a rapid increase in the proportion of Raman power, which also corresponds to the beginning of the decrease in the optical-to-optical efficiency at the output power of 3.5 kW. As shown in Figure 3d, it was found that the SRNR changed uniformly and slowly, and no high-order Raman peaks appeared. However, at this power level (5.02 kW), TMI was observed.

3.2. Spectrum Bandwidth

Under different OC-FBG conditions, when only forward pumping is applied, the output power is 1 kW, and the output spectrum is shown in Figure 7a. The 3 dB bandwidths are 3.36 nm, 3.12 nm, and 2.6 nm, respectively. The output spectrum when the output power is 3 kW is shown in Figure 7b. The 3 dB bandwidths are 4.8 nm, 4.7 nm, and 3.6 nm, respectively. From this, it can be seen that under the condition of a 3 kW output power, after the beam passes through the oscillating section and the amplifying section, the bandwidth of the output beam is actually narrower under the grating condition with larger reflectance bandwidth. Furthermore, as the output power reaches approximately 4.7 kW, the bandwidth of the output beam is not inversely proportional to the reflectance bandwidth of the OC-FBG. As shown in Figure 7c, the 3 dB bandwidths are 5.48 nm, 5.7 nm, 5.3 nm. Although the OC-FBG3 still maintains the narrowest 3 dB bandwidth, the OC-FBG2 has a wider output 3 dB bandwidth compared to the OC-FBG1. This indicates that there are complex nonlinear effects affecting the output bandwidth at high power, such as self-phase modulation (SPM), cross-phase modulation (XPM), and four-wave mixing (FWM).

While SPM, XPM, and FWM are typically more pronounced in pulsed lasers, these nonlinear effects also manifest in high-power continuous-wave (CW) lasers. This occurs because high-power operation induces significant intensity fluctuations, dynamically modulating the refractive index of the silica medium and generating transient phase and frequency shifts in the propagating light. In this study, both the seed laser and pump source operate at a single wavelength. Consequently, the XPM and FWM contributions to spectral broadening are negligible, and we focus exclusively on SPM.

Given that power fluctuations in CW lasers primarily arise from pump intensity noise and laser relaxation oscillations—inherently stochastic processes—SPM-induced spectral broadening exhibits symmetric characteristics. Our experiments demonstrate that employing an oscillating section output with a broader 3 dB bandwidth effectively reduces the optical power density in the amplifying section, thereby suppressing SPM-induced spectral broadening during amplification. Furthermore, our experimental results indicate that there exists an optimal 3 dB bandwidth for the OC-FBG, which minimizes the 3 dB bandwidth of the output beam. Moreover, it can be observed from the above figure that the 3 dB bandwidths corresponding to the 1.2 nm and 1.3 nm gratings have intersection points, indicating that under different output power conditions, the influence of SPM on seed light of different bandwidths is not linear. However, the precise mechanisms governing SPM-induced spectral broadening in the oscillating and amplifying sections in CW lasers remain incompletely elucidated and warrant further investigation.

4. Burn-In Test

In addition, TMI also restricts the output power of high-power fiber lasers. Current research shows that in high-power fiber lasers, the photodarkening effect can significantly reduce the TMI threshold. As shown in Figure 8a, under the condition of OC-FBG2 and with an output power of 4.85 kW, the laser was subjected to a 30 min burn-in test. During the burn-in test, the proportion of Raman power gradually increased, and there were no significant changes in the output power or the 3 dB bandwidth of the output beam. Since the Raman power ratio was reduced to a dangerous level, we paused the burn-in test at 30 min. The subsequently conducted burn-in test further confirmed the repeatability of the phenomenon between them. As shown in Figure 8b, a 5 h burn-in test was carried out when the output power was 4.7 kW. During the burn-in test, the proportion of Raman power also increased and remained stable at about 30 min, with no significant changes in the output power and the output beam bandwidth. Notably, during the burn-in test, there was no significant change in the TMI threshold. However, when the laser was retested 10 h after the burn-in test, it was found that the TMI threshold had decreased. Under the same pump power, the laser achieved TMI at 4.76 kW, with the TMI threshold dropping by nearly 90 W. Moreover, after conducting the same burn-in test with the same laser, this phenomenon was also discovered, indicating that it did not occur by chance. This phenomenon is different from those described in previous experiments and studies, and its causes still need to be further explored. As shown in Figure 8c, a 52 min burn-in test was conducted at an output power of 4.76 kW after photodarkening, and the proportion of Raman power increased within 30 min as well. Subsequently, we replaced the OC-FBG2 of the oscillating section with OC-FBG3 and conducted a 28 min burn-in test on it, as shown in Figure 8d. Since the Raman inhibition ratio exceeds 35 dB and the variation is relatively small, no specific data records were made for the proportion of Raman power. Furthermore, an interesting phenomenon is observed in this test. When the pump power increases to 2080 W in the FPSC and 4370 W in the BPSC, and the output power is 4.96 kW, TMI occurs. However, during the subsequent burn-in test process, the output power gradually increases to 5.02 kW within 10 min and tends to be stable, and the 3 dB bandwidth gradually decreases, from 5.3 nm to about 3 nm, and the reduction process shows a stepwise trend. In future experimental research, we will pay special attention to this phenomenon and attempt to conduct further studies.

5. Discussion

We conducted research on the performance of oscillating sections with different OC-FBGs based on an OAIFL-structure laser with a 5 kW power level, as well as research on SRS suppression, output bandwidth, optical-to-optical efficiency, and photodarkening. In conclusion, in an oscillating–amplifying integrated fiber laser, increasing the bandwidth of OC-FBG in the oscillation section can effectively enhance the SRS threshold. SRS is closely related to the output laser power spectral density. The oscillator cavity with a wider OC-FBG output spectrum has a wider 3 dB bandwidth, supporting more longitudinal-mode outputs. Under the same output power, it reduces the peak spectral density, thereby effectively suppressing SRS. Moreover, at a high output power, the required Raman suppression ratio can be obtained with wide-bandwidth OC-FBG. When we reduce the proportion of Raman optical power, the optical efficiency is almost unaffected. Additionally, when we raise the SRS threshold, increasing the bandwidth of the OC-FBG can suppress the spectral broadening of the output beam caused by SPM. During the burn-in test process, the wide-bandwidth OC-FBG increases the TMI threshold, and it demonstrates excellent performance in terms of output power and output bandwidth, which has important reference value for the subsequent design of oscillating–amplifying integrated fiber lasers.