1010 nm Directly LD-Pumped 6kW Monolithic Fiber Laser Employing Long-Tapered Yb³⁺-Doped Fiber

Abstract

Utilizing long-wavelength laser diodes (LDs) for pumping to achieve high-power fiber laser output is an effective method for attaining high quantum efficiency and excellent thermal management. In this work, we report on a Master Oscillator Power Amplifier (MOPA)-structured long-tapered Yb³⁺-doped fiber laser directly pumped by long-wavelength laser diodes. By shifting the center wavelength of the pump source to 1010 nm, the heat generation within the fiber laser is effectively controlled, thereby increasing the transverse mode instability (TMI) threshold. Additionally, the use of a long-tapered fiber enlarges the mode area and suppresses stimulated Raman scattering (SRS) effects that typically arise from increased fiber length. As a result, an output of 6030 W is achieved with an optical-to-optical (O–O) efficiency of 83.7%, a SRS suppression ratio exceeding 50 dB, and no occurrence of dynamic TMI. This approach provides a valuable reference for optimizing long-wavelength pumping to suppress nonlinear effects and also holds potential for wide-temperature operational applications.

1. Introduction

High-power fiber lasers, widely known for their outstanding beam quality, high output power, superior thermal management, and high stability, have been employed in various fields, such as industrial processing, medical applications, and telecommunications [1,2,3]. High-power ytterbium-doped fiber (YDF) lasers, which are directly pumped by laser diodes (LDs), can efficiently convert electrical energy into optical energy. Compared to traditional pump sources, they offer higher conversion efficiency and lower energy consumption. Additionally, due to the compact size of laser diodes, they can be more effectively integrated into laser systems. Currently, fiber lasers directly pumped by LDs have achieved reported output powers up to 20 kW [4,5] and single-mode (M² < 2.0) laser output has reached 8 kW [6]. However, the primary factor currently limiting the further advancement of fiber lasers is the transverse mode instability (TMI) effect, which significantly impacts beam quality and output power [7,8]. The TMI effect is a phenomenon widely observed in all lasers. Primarily, it manifests as a degradation in beam quality and low stability of the laser as soon as the pump power increases to a specific threshold. It is essentially caused by a dynamic mode interaction between the fundamental mode and higher-order mode. Currently, researchers worldwide have conducted extensive studies on the theory of TMI [9,10,11]. From the perspective of suppressing TMI, progress has been made through an external control system, the optimization of gain fiber, and improvements in structure design [12,13,14,15].

Optimizing the LD pump wavelength to suppress TMI is an effective and straightforward approach. The origin of TMI is believed to be due to mode coupling induced by temperature variations [8]. According to the thermal source model [16], it is known that once the fiber structure design is determined, the heat generated in the fiber laser is primarily influenced by quantum efficiency and the pump absorption coefficient. According to the ytterbium ion emission cross-section, the output wavelength range of ytterbium-doped fiber lasers is between 1000 and 1130 nm. As the pump wavelength shifts from the commonly used 976 nm towards longer wavelengths, gradually closer to signal lights, it leads to a reduction in the quantum defect. Simultaneously, as the wavelength stretches, ytterbium ions decrease the ability to absorb light and the pump absorption coefficient decreases. This results in a reduction in heat generation within the fiber laser and, thereby, suppresses the TMI effect.

In addition, the direct pumping using 1010 nm laser diodes offers several other advantages. Firstly, the 1010 nm LD exhibits higher quantum efficiency compared to conventional wavelengths. Compared to tandem pumping, direct pumping with a 1010 nm LD is expected to achieve the power levels supported by tandem pumping while maintaining the inherent benefits of direct LD pumping, such as high conversion efficiency, compact size, light weight, and low cost. Secondly, at room temperature (25 °C), the 1010 nm LD can operate over a broad temperature range to meet the requirements for different applications. Currently, within the normal operating range of high-power LDs, for every 1 °C increase in temperature at a fixed operating current, the central wavelength shifts approximately 0.32 nm towards the longer wavelength. At low temperatures (around −50 °C), the central wavelength shifts to approximately 985 nm, where the fiber can absorb energy effectively. At high temperatures (around 50 °C), the central wavelength of the LD shifts to approximately 1018 nm. Conventional LDs, such as commonly used commercial high-power LDs, have anti-reflection wavelength ranges primarily located between 1030 and 1200 nm. This effectively prevents the reflection of amplified spontaneous emission (ASE) light within this band from entering the LD. However, with longer wavelength LDs, the wavelength tends to approach 1030 nm at high temperatures, which increases the risk of the potential backscattering of ASEs. In September 2023, Li et al. used 940 nm (at room temperature) LDs, constructing a laser capable of wide-temperature operation from −50 °C to 50 °C, achieving a 1 kW fiber laser output over a 100 °C range [17].

In early 2024, our research team proposed the use of a long-wavelength LD pump with a wavelength longer than 1010 nm for an integrated oscillator-amplifier ytterbium-doped fiber laser, achieving a 2 kW laser output [18]. However, to date, there have been relatively few detailed reports on ytterbium-doped fiber lasers pumped by long-wavelength LDs (>1000 nm). Additionally, the absorption cross-section of ytterbium ions at wavelengths above 1000 nm is relatively low, leading to a low pump absorption compared with LDs with conventional wavelengths. To achieve enough absorption, increasing the fiber length is important for a gain of about 20 dB. However, extending the fiber length reduces the threshold for stimulated Raman scattering (SRS), thereby introducing SRS effects [19], which hinders the further increase of output power, highlighting the inherent contradiction in fiber lasers between suppressing TMI and SRS effects [20]. Tapered fibers can effectively balance the relationship between these two effects. The part with a small core of tapered fiber reduces the number of guided modes in the core, enabling effective mode control. Meanwhile, the larger core diameter at the back end helps reduce the core power density, thereby increasing the SRS threshold [21,22,23]. Additionally, the increased diameter of the cladding at the back allows for the use of beam combiners with higher combining capability, which enhances the pump power injection.

In this paper, we demonstrate a direct LD-pumped fiber amplifier employing non-stabilized-wavelength 1010 nm LDs and a homemade long-tapered fiber. The amplification stage of the fiber laser employs a bidirectional pump configuration using non-stabilized-wavelength 1010 nm LDs, which not only enhances the pump power of the individual laser diodes but also suppresses the TMI effect. Additionally, the long-tapered YDF effectively suppresses the SRS effect and increases the number of laser diodes capable of backward pumping, thereby enhancing the achievable pump power, ultimately achieving a 6 kW laser output. During the experiment, no significant nonlinear effects, such as SRS or TMI, were observed. Our results reveal the potential of long-wavelength LDs for achieving higher-power fiber lasers.

2. Experimental Setup

Figure 1 depicts the experimental setup of a MOPA configuration. The seed source is counter-pumped by two 976 nm LDs utilizing a fiber with a core/cladding diameter of 20/400 μm; the core numerical aperture is 0.065. The central wavelength of the seed laser is 1080 nm, with a full width at half maximum (FWHM) of 1.1 nm, and the output power is approximately 90 W without the SBS effect. The seed light passes through a mode field adapter first to meet the different diameters of TYDF. To avoid an unnecessary pump light into the forward fiber, a cladding light stripper is needed. The amplifier stage employs a bidirectional pumping configuration. Forward pumping uses six 1010 nm laser diodes, injected into a (6 + 1) × 1 forward pump/signal combiner, with power of 1758 W. Backward pumping uses eighteen 1010 nm laser diodes, injected into a (18 + 1) × 1 backward pump/signal combiner, with power of 5341 W. The fiber used in the amplifier stage is a long-tapered ytterbium-doped fiber, which can be divided into 3 parts. The forward end has a small core/cladding diameter of 30/250 μm with a length of 15 m. The backward end has a large core/cladding diameter of 48/400 μm with a length of 15 m. A shorter front section or an overly long tail section will reduce the performance of TMI suppression. A taper region between both ends connects them smoothly, with a length of 30 m. Its core NA is 0.065 and the cladding absorption coefficient is 0.36 dB/m. The TYDF substrate used is composed of high-purity silicon dioxide, primarily doped with phosphorus, aluminum, and ytterbium ions, with molar ratios of approximately 3%, 2.4%, and 0.16%, respectively. After removing the pump light again with CLS, amplified signal light outputs from a quartz beam hand are followed by collimation and splitting. The output power, spectrum, time trace signal, beam profile, beam quality, and time trace signal are then recorded using a power meter, optical spectrum, M² analyzer, and photodetector.

3. Results and Discussion

3.1. 4.6 kW Output by Counter-Pumped 1010 nm LD

Firstly, to ensure the 1010 nm LDs can achieve high power output, the relevant performance, including output power, electro-optical (E-O) conversion efficiency, and spectral characteristics, was evaluated, as shown in Figure 2. As the operating current increased, the output power of the 1010 nm LD increased linearly, ultimately reaching approximately 290 W. The E-O efficiency slightly decreased with the increase in operating current after exceeding the threshold current and eventually stabilized at around 47%. Figure 2b illustrates the optical spectrum variation as operating current increases. The center wavelength of the LD shifted from 1000 nm to 1011 nm, with a 3 dB linewidth of approximately 4 nm. Due to mechanisms such as carrier absorption, a portion of the injected current contributed to heat generation. As the current increased, the heat generated by the semiconductor laser also increased, leading to an expansion of the laser cavity dimensions and a narrowing of the bandgap. Consequently, this resulted in a “red shift” of the wavelength with increasing operating current [24].

Next, 18 LDs were utilized in a counter-pump configuration based on the MOPA structure to test and record the laser’s output characteristics. The curves of output power and O–O conversion efficiency changes with pump power are shown in Figure 3a. The output power exhibited a linear connection with pump power, reaching a peak of 4660 W when the pump power was 5127 W. Neglecting initial instability in the power output, the O–O conversion efficiency remained generally stable and high, approximately 89.1% when the pump power reached the maximum. The spectrum of the signal light at the maximum power of 4660 W is shown in Figure 3b. The signal light is characterized by a central wavelength near 1080 nm, accompanied by a 3 dB linewidth of 2.4 nm. A characteristic peak of stimulated Raman scattering was expected to be around 50 nm longer than the center wavelength; however, no significant peak was observed, indicating that there was no noticeable stimulated Raman scattering effect.

3.2. 6 kW Output by the Bidirectional Pumped 1010 nm LD

It is obvious that the pump power is the main limitation for further output power; therefore, six forward-pumping 1010 nm LDs were added to develop a bidirectional pumping MOPA structure amplifier, successfully achieving a high-power laser output of 6 kW. Figure 1 shows the experimental schematic. Due to the limited core diameter at the front end of the tapered fiber, only a (6 + 1) × 1 forward pump/signal combiner was feasible. However, the larger core diameter allowed for the use of an (18 + 1) × 1 backward pump/signal combiner, enhancing the system’s pump light injection capability. Figure 4a illustrates the variation curves of output power and optical-to-optical conversion efficiency in relation to pump power. The output power exhibited a linear connection with pump power, reaching a maximum of 6030 W at a total pump power of 7099 W. The O–O conversion efficiency was 83.7%, with a slope efficiency of 84.6%. Because of the lower coupling capability of the FPSC, as the forward pump was added, the efficiency decreased compared with the result in Figure 3a. The time-domain and frequency-domain curves of the signals are shown in Figure 4b when output power reaches 6030 W, indicating stable time-domain characteristics with no significant pulsing observed. The inset displays the calculated frequency-domain signals at 6030 W, where no noticeable high-intensity peaks are observed, indicating the absence of a significant TMI effect. This result confirms that using long-wavelength LD pumping effectively reduces heat generation in the laser and raises the TMI threshold. Figure 4c shows the spectrum of seed light and maximum signal light. The 3 dB bandwidth of the signal light is 3.3 nm with a central wavelength of 1080 nm. After the introduction of the forward pump light, the spectrum of the signal light broadens obviously while the central wavelength remains unchanged. The characteristic peak of stimulated Raman scattering still does not exhibit a prominent peak (with a Raman suppression ratio of approximately 50 dB), confirming the effectiveness of the long-tapered fiber in suppressing stimulated Raman scattering. Additionally, no other side peaks or noticeable four-wave mixing phenomena are present in the spectrum. Finally, Figure 4d illustrates the variation of the M² factor of signal light with output power. Without a pump source, the seed beam quality M² factors are ${\mathrm{M}}_{\mathrm{X}}^2$: 3.29 and ${\mathrm{M}}_{\mathrm{Y}}^2$: 3.13. At an output power of 6030 W, the beam quality M² factors slightly deteriorate to ${\mathrm{M}}_{\mathrm{X}}^2$: 3.31 and ${\mathrm{M}}_{\mathrm{Y}}^2$: 3.18. This indicates that while there is a slight degradation in beam quality with the addition of the pump light, the overall stability remains around 3.2.

To address the issue of beam quality degradation, tests are conducted to assess the influence of several components on the beam quality, including the seed, combiner, and gain fiber. Measurements are taken to observe the changes in the beam quality (M² factor) of the signal light when only the seed light is input as it passes through various components.

Table 1. Beam quality degradation of each component when the seed laser is directly injected.

Component	Output M2	M2 Deterioration
SEED	1.273	0
MFA + CLS	1.545	0.272
MFA + CLS + FPSC	2.066	0.521
MFA + CLS + FPSC + TYDF	2.567	0.501
MFA + CLS + FPSC + TYDF + BPSC	3.210	0.643

records the contributions of each component to the M² increment relative to the baseline M² of the seed. The table quantifies how much each part contributes to the overall beam quality degradation. According to Table 1, the FPSC, BPSC, and TYDF significantly impact beam quality. In the following work, we plan to achieve higher power and brightness laser output by optimizing the components and refining the TYDF design.

4. Conclusions

This paper presents a direct pumping scheme using 1010 nm LDs as pump sources in a bidirectional pumping configuration for a long-tapered fiber laser based on an all-fiber MOPA structure. Initially, we achieved a fiber laser output of 4660 W using 18 LDs in a backward pumping configuration, with no observed nonlinear effects, such as TMI or SRS. Subsequently, by adding six LDs for forward pumping, we reached a maximum output power of 6030 W, with an O–O conversion efficiency of 83.7%, again, without detecting TMI or SRS effects. To address the issue of beam quality degradation, we conducted a detailed investigation of the various components and we plan to further enhance the beam quality of the output signal light through optimized component selection. This paper integrates methods for suppressing nonlinear effects in fiber lasers, utilizing long-wavelength LDs and long-tapered fibers. This approach partially mitigates the conflict between suppressing both SRS and TMI effects, providing a reference for the comprehensive suppression of nonlinear effects in future work. Additionally, future research will involve further experimental validation of the 1010 nm LDs for wide-temperature operation applications.