1. Introduction

The fiber laser industry has seen very rapid growth in recent years thanks largely to the simple and robust fiber laser designs available for achieving high powers. Single-mode powers up to 10kW are available commercially [1]. There is still significant demand for higher single-mode powers, mainly driven by many emerging applications. The average power of directly-diode-pumped fiber lasers is limited to ~4kW by the recently discovered transverse mode instability (TMI) [2–5]. TMI arises from the heat generated by quantum defects and other loss mechanisms in a fiber laser [6–8]. It can cause significant waveguide mode coupling at high powers when strong heating, in combination with mode beating, creates thermal gratings in the fiber. Since fiber loss is typically very low for carefully engineered fibers with low photo-darkening, quantum defect heating often dominates in a fiber laser. One way to mitigate TMI is to operate near the single-mode regime to weaken higher-order-mode (HOM) guidance [2–5]. This has led to a reduction of core diameters in recent high-power demonstrations, typically to ~20μm. Since large cores are still desired in most applications for the mitigation of nonlinear effects, ultra-low NA fibers have also been used in some recent demonstrations to allow a slightly larger core while still operating near the single-mode regime [3,5]. One significant drawback with the ultra-low NA approach is greatly weakened fundamental mode guidance and, consequently, much larger fiber coil diameters need to be used. Another method of mitigating TMI is the use of more advanced fiber designs to suppress HOM while maintaining a larger core diameter. The all-solid photonic bandgap fiber is an example [9–13]. Recently ~1kW single-mode output power was achieved in such a fiber with a core diameter of ~50μm [13].

A more fundamental method of mitigating TMI is to reduce heating by minimizing the quantum defect. This method can be used in combination with the methods based on waveguide designs discussed earlier. The record 10kW single-mode power was achieved by pumping the last-stage fiber amplifier using fiber lasers at 1018nm instead of multimode diodes at ~976nm, thus lowering quantum defect heating [1]. This tandem pumping scheme also leads to a significant increase in complexity, since 47 270W fiber lasers at 1018nm were required for the 10kW single-mode fiber laser. Recently, there has been strong interest in developing multimode fiber lasers at 1018nm for use in the tandem pumping scheme with 805W demonstrated in 2017 [14–18]. Further power scaling of single-mode1018nm fiber lasers will allow increased brightness of the combined pump powers in a tandem pumping scheme and a simpler architecture. If further power scaling is possible, such single-mode 1018nm fiber lasers can also be a potential power scaling path on its own.

Ytterbium fiber lasers operate like a three-level system at 1018nm and a higher inversion is required to achieve the necessary gain for laser operation compared to operating at longer wavelengths, where they are more like a four-level system. This means that a higher pump power is required throughout the fiber resulting in a large amount of pump power leaving the fiber. A given inversion is achieved by balancing signal and pump intensities, not powers. To lower the pump power required to maintain a given inversion, a larger core and smaller pump cladding, i.e. a larger core-to-cladding ratio, are necessary to lower the signal intensity and to raise the pump intensity. A large core-to-cladding ratio is therefore necessary to increase 1018nm ytterbium fiber laser efficiency with respect to the launched pump power. In addition, it is also important to suppress emission at the ASE peak near 1030nm. This can be achieved by a combination of short fiber length, ensuring adequately high inversion throughout the fiber, and narrow fiber Bragg grating reflectors to minimize reflection at the ASE peak.

In summary, it is relatively easy to achieve lasing at 1018nm or even shorter wavelengths in ytterbium fiber lasers, it is harder to achieve a combination of lasing at shorter wavelengths and high efficiency. Large core-to-cladding ratio is required for this. It is even harder to achieve a combination of high efficiency, single mode and a shorter lasing wavelength.

Recently, there have also been some demonstrations of single-mode ytterbium fiber lasers at 1018nm [19,20]. 200W single-mode power at 1018nm was achieved with a double-clad ytterbium-doped 10μm/130μm fiber in [19]. 230W single-mode power at 1018nm was achieved with a double-clad ytterbium-doped 10μm/125μm fiber in [20]. Both demonstrations used two fiber Bragg gratings, one as a high reflector and one as an output coupler, to suppress the ASE. The single-mode 10μm core ensures the single-mode beam quality. To maintain the necessary large core-to-cladding ratio, small pump guides were used in both cases. This can potentially limit the output power due to the limited available pump power that can be launched into this small pump guide. The quantum defect in these demonstrations was merely ~4.1%.

There has also been a demonstration of 146W single-mode power at 1009nm [21]. A complex MOPA arrangement with an oscillator formed by two fiber Bragg gratings and two amplifiers was used. As we have shown in an earlier work, it is much harder to achieve high efficiency at shorter laser wavelengths due to the ytterbium fiber laser behaving more like a three-level system [22]. A very large core-to-cladding ratio had to be used in the last-stage amplifier to ensure adequate efficiency in [21]. A large pitch fiber with a mode field diameter of 90μm and an air-clad pump guide of 283μm was used in this case. This work further demonstrates the importance of special fiber designs for achieving the combination of high efficiency, single mode and a shorter lasing wavelength.

We have also shown in [22] that high efficiency can be achieved in multimode ytterbium fiber lasers at 1018nm using a large core-to-cladding ratio and a phosphosilicate host. This demonstration used an ytterbium-doped phosphosilicate leakage channel fiber with a core diameter of ~50μm and a cladding diameter of ~400μm. A phosphosilicate host provides higher 1018nm gain at a given inversion compared to the more conventional aluminosilicate host, further lowering the required pump power and consequently increasing efficiency. We have also been developing 50μm-core single-mode ytterbium-doped all-solid photonic bandgap fibers [10,12,13]. These fibers have a large core-to-cladding ratio and a phosphosilicate host, making them ideal for efficient ytterbium fiber lasers at 1018nm. These fibers are robustly single mode, making them suitable for efficient directly-diode-pumped high-power single-mode fiber lasers at 1018nm. In addition, the large cores of these fibers allow for a much larger pump guide for a given core-to-cladding ratio, allowing the use of much higher pump powers, producing much higher single-mode fiber laser powers.

In this work, we have demonstrated 240W single-mode power at 1018nm from a directly-diode-pumped ytterbium fiber laser using a 50μm/400μm ytterbium-doped all-solid photonic bandgap fiber with a phosphosilicate host. At high powers, the efficiencies with respect to the launched and absorbed pump power are 75% and 86% respectively. The key for this demonstration is the ytterbium all-solid photonic bandgap fibers, which is potentially a significant enabling technology for achieving the combination of single mode, high power and high efficiency at 1018nm. Such efficient high-power single-mode 1018nm ytterbium fiber lasers can provide higher brightness pump and simpler overall architecture for tandem pumped ytterbium fiber lasers. If further power scaling is possible, they can also provide a power scaling path on its own. It is worth noting that 270W single-mode 1018nm ytterbium fiber lasers were claimed to be made by IPG [23], no technical details were ever given.

2. Experiments

The cross section of the ytterbium-doped all solid photonic bandgap fiber is shown in Fig. 1. The all-solid photonic bandgap fiber has multiple small cores introduced in the cladding. These small cladding cores are designed to be in resonance to couple out the higher-order modes in the main core for enhanced HOM suppression [11,13]. The fiber has a core diameter of ~50μm and cladding diameter of ~400μm. It is coated with low-index acrylic coating to give a pump NA of ~0.4. The pump absorption was measured to be ~2.3dB/m at 976nm. A matching photosensitive fiber was specially fabricated for this work at our in-house optical fiber fabrication facility. The photosensitive fiber also has a core diameter of ~50μm and cladding diameter of ~400μm, and is coated with a low-index acrylic coating. The photosensitive fiber was loaded with hydrogen and a strong fiber Bragg grating was written in it at 1018nm, before recoating. The grating reflectivity is hard to measure with certainty due to the multimode nature of the photosensitive fiber and is estimated to be significantly higher than 99%. The grating was then spliced to the all-solid photonic bandgap fiber. The fiber was continually cut until stable 1018nm operation was obtained. The optimized fiber length was 3.4m and it was coiled to 70cm in diameter. The laser setup is shown in Fig. 2. The 976nm pump from a diode bar was injected through a dichroic mirror (high reflectivity above 1µm and low reflectivity below 1µm) into the all-solid photonic bandgap fiber. The laser cavity was formed by the 4% reflection at the perpendicularly cleaved pump end and the fiber Bragg grating. The output at 1018nm was reflected off the dichroic mirror on to a detector or camera.

 

Fig. 1 Cross section of the ytterbium-doped all-solid photonic bandgap fiber.

Download Full Size | PDF

 

Fig. 2 The setup for the 1018 fiber laser.

Download Full Size | PDF

The output power with respect to the launched and absorbed pump powers is given in Fig. 3. The pump is not wavelength stabilized and is known to change its wavelength as its power is changed. This explains most of the nonlinear behavior in Fig. 3. The slope efficiencies at the high powers are ~75% and 86% respectively with regard to the launched and absorbed pump powers. The spectrum at the output was measured at several powers and plotted in Fig. 4. The laser peak wavelength was at 1017.8nm at 240W. Co-lasing at the ASE peak at ~1025.3nm was seen only at 240W. Even at 240W, the peak power at 1025.3nm was below 20dB of that at 1017.8nm.

 

Fig. 3 Output power at 1018nm versus launched and absorbed pump powers.

Download Full Size | PDF

 

Fig. 4 Spectra of the output at 80W, 180W and 240W.

Download Full Size | PDF

The output mode patterns were closely monitored at various output powers (see Fig. 5). There is no significant change in mode pattern throughout the power range, except for a small amount of light trapped in the small cladding cores. The M² was measured at an output power of 8W, giving M² = 1.25 and 1.35 for x and y directions respectively (see Fig. 6).

 

Fig. 5 Output mode patterns at various powers.

Download Full Size | PDF

 

Fig. 6 Measured M² at output power of 8W.

Download Full Size | PDF

The signal wavelength at 1018nm was very close to the ASE peak at 1025nm in this case. The spectrally dependent transmission of the all-solid photonic bandgap fiber is, consequently, expected to only play a minor role in suppressing the gain at the ASE peak. Since we used a perpendicularly cleaved fiber end as the output coupler, much better suppression of the ASE peak is expected if a fiber Bragg grating is used as the output coupler as in [19,20]. This could significantly increase the round trip loss for the ASE and result in a much higher output power at 1018nm without the co-lasing at 1025nm.

3. Conclusions

We have demonstrated a single-mode fiber laser at 1018nm with an output power of 240W. This fiber laser was directly diode pumped at 976nm by multimode diodes and had a quantum defect of 4.1%. At high powers, the fiber laser has high efficiencies of ~76% and ~86% with regard to launched and absorbed pump powers respectively.

The key to this demonstration is the ytterbium all-solid photonic bandgap fiber which enables single-mode operation in a large core and much higher available pump power due to the larger pump guide while maintaining a core-to-cladding ratio sufficiently large for the suppression of lasing at ytterbium ASE peak. This work demonstrates the potential of this approach for further power scaling of single-mode 1018nm ytterbium fiber lasers. The low quantum defect mitigates TMI in the 1018nm fiber laser. Available pump powers in the current fiber can certainly support multiple kilowatt ytterbium fiber lasers at 1018nm. Currently, 1018nm power in this work is limited by co-lasing at ~1025nm. This can be further improved by using a FBG as the output coupler and further efforts to suppress spurious reflections. Further power scaling of fiber lasers requires lower quantum defect and pumps of higher brightness, efficient high-power single-mode 1018nm fiber lasers can provide enhanced pump brightness and simpler overall architecture with further power scaling in a tandem-pumping scheme. If significant power scaling can be achieved in single-mode 1018nm fiber lasers, they can also be a significant path for power scaling on their own.