Efficient high-power diffraction-limited fiber laser with a record low quantum defect of 1.7%

Quantum defect (QD) heating in a high-power fiber laser can lead to thermally induced nonlinear instabilities. One example is transverse mode instability (TMI), which sets a fundamental limit on power from a diffraction-limited fiber laser. A high-power diffraction-limited Yb³⁺-doped fiber laser has the lowest quantum defect (5%–10%) and the highest output power among all solid-state lasers. Lowering the quantum defect through control of pump and laser wavelengths makes the Yb³⁺ ions operate more like a three-level system, leading to severe amplified spontaneous emission (ASE) at the longer wavelengths, where it operates more like a four-level system, and associated degradation of efficiency.

In a recent study, the operation of an ytterbium fiber laser at 985.7 and 989.8 nm was demonstrated with a QD below 1%.¹ The fiber lasers were, however, core-pumped and, therefore, limited to output powers well below 1 W.

In this work, we demonstrate an approach to lower the QD potentially to ∼1% without significant compromise of efficiency at high powers (>100 W). In a first demonstration, a diffraction-limited ∼154 W laser emitting at ∼993 nm with negligible ASE, pumped at ∼976 nm, was achieved with a slope efficiency of ∼75% vs the launched pump power and ∼95% vs absorbed pump powers. The laser QD was ∼1.7%, a record low for high-power solid-state lasers to the best of the authors’ knowledge. The power was only limited by the available pump power, demonstrating the potential for a significant reduction in QD and, consequently, a significant increase in TMI threshold. This work also significantly extends the low wavelength limits of high-power ytterbium fiber lasers and thereby enables many new applications.

The four-level Yb³⁺ system has been a critical foundation for industrial high-power fiber lasers. The relatively small ytterbium absorption cross section at ∼1060 nm allows amplifiers and lasers to operate at low inversions, which enable the pump power to be mostly absorbed within the fiber. The three-level system at much shorter wavelengths has only attracted a little interest, having been limited by poor efficiencies and low powers in practical laser configurations. Conventional methods of mitigation are almost entirely based on a large core-to-cladding ratio.^2–8 This lowers the intensity of the laser relative to that of the pump, thereby allowing the required high inversion to be maintained at relatively low pump powers. This results in reduced unused pump (i.e., residual pump) but produces only limited performance improvements in practical high-power fiber lasers.

The key to the approach in this study is the use of an Yb³⁺-doped double-clad all-solid photonic bandgap fiber (ASPBF), which was engineered both to support robust single-mode operation at large cores and to provide a strong ASE suppression at longer wavelengths, critical to QD reduction. The ASE is placed outside the transmission window of the photonic bandgap while keeping the laser wavelength in the transmission window. For achieving high efficiency in the three-level system, where high inversion is required, a smaller cladding-to-core ratio is also essential.² The required inversion of slightly over 40% at the lasing wavelength can be achieved by pumping at ∼976 nm, another key for the record low QD.

The basic laser arrangement is a counter-pumped monolithic configuration (Fig. 1). Two pump diodes at ∼976 nm can be used at the 2 + 1 pump combiner. The high-reflectivity (HR) fiber Bragg grating (FBG) was written in-house in a 25/125 μm photosensitive fiber also made in-house. The FBG has a reflectivity of >99%. Initial experiments were conducted with a wavelength-stabilized 976 nm diode rated at 100 W (a 0.22 NA 105/125 μm fiber) spliced in the system.

The cross section of the Yb³⁺-doped ASPBF is given in the inset of Fig. 2, along with the measured bend loss around the wavelengths of interest in a undoped version of the same fiber with the laser and pump wavelength indicated by the vertical lines, showing the carefully engineered photonic bandgap optimized for the suppression of the ytterbium ASE peaking around 1020–1030 nm.

This fiber was designed and fabricated in-house at Clemson University and is similar to the fiber used in Refs. 9 and 10. The fiber has a core with a side-to-side distance of 21 µm and a cladding with a side-to-side distance of 124 µm. A multiple-cladding-resonance design is used for enhanced higher-order-mode suppression by resonant coupling of higher-order-mode to the guided modes in the cladding.^11,12 The fiber is coated with low index acrylate to provide a pump NA of ∼0.46, and the cladding pump absorption was measured to be ∼1.76 dB/m at 915 nm. Some bend loss can be clearly observed at the laser wavelength at ∼993 nm at the smaller coil diameters measured in Fig. 2.

The phosphosilicate core¹³ is doped with Yb³⁺ ions, and the net gains are shown in Fig. 3 for various inversions with the laser and pump wavelength indicated by the vertical lines. The maximum inversion achievable at the pump wavelength of 976 nm is ∼50%. Net gain is possible for wavelengths above ∼985 nm in the phosphosilicate glass (see the 50% inversion curve in Fig. 3).

To optimize fiber length, the active fiber was progressively cut back at a fixed coil diameter of 14 cm with an angle-cleaved output end. The efficiency vs the launched pump power was measured at low powers. The measured efficiency vs fiber length, shown in Fig. 4, peaked at ∼5.5 m. The reduction in efficiency at shorter fiber lengths is due to a reduction in pump absorption. The efficiency vs the launched pump power was then carefully characterized at low powers at various coil diameters in a 5.79 m fiber with an angle-cleaved output end (Fig. 5). The progressive decrease in efficiency at smaller coil diameters is due to the increasing bend loss at the laser wavelength, because of the proximity of the laser wavelength to the edge of the bandgap, as shown in Fig. 2.

The 5.79 m fiber with an angle-cleaved output end was then tested at high powers using the 100 W wavelength-stabilized 976 nm pump diode. The achievable maximum powers at larger coil diameters were limited by ytterbium ASE, due to the limited ASE suppression from the small bend loss at the ytterbium ASE wavelength. The highest power of ∼41 W at 993 nm was achieved at the coil diameter of 13.2 cm at the maximum launched pump power of 86.6 W (see Fig. 6). The slope efficiencies excluding the two data points at the highest powers, where some saturation was seen, are ∼55% and ∼86%, respectively, vs the launched and absorbed pump powers.

The optical spectra at the output powers of 3.2 and 41 W are shown in Fig. 7, showing the growth of ytterbium ASE with increasing power. The ASE peak power is still more than 30 dB below the laser peak power at 41 W of output power. The peak at ∼976 nm is from the pump.

Pumping at 915 nm was also tested. The optimized fiber length is then ∼12 m due to the much weaker pump absorption at ∼915 nm. Nevertheless, a slope efficiency of ∼50% vs the launched pump power was achieved. A strong ytterbium ASE at ∼976 nm was, however, also observed. Overall, pumping at ∼976 nm seems to be a better option due to the elimination of ytterbium ASE at ∼976 nm, higher efficiency, and much lower quantum defect.

In our previous work, at ∼978 nm,^9,10 an angle-cleaved output end minimized unintended laser output at the FBG end. Such an open cavity configuration does not clamp the total round trip gain to loss and inversion, and consequently, the ASE increases at high pump powers.

To further suppress ASE, a straight-cleaved output with ∼4% reflection at the output end was tested. A 5.68 m fiber coiled at 16 cm was used, and the measured output is shown in Fig. 8. At the maximum launched pump power of 86.6 W from the 100 W pump, 48.5 W at 993 nm was achieved without any sign of saturation, with slope efficiencies of ∼60% and ∼95%, respectively, vs the launched and absorbed pump powers. The output spectra at output powers of 2.3 and 48.5 W are shown in Fig. 9, showing no sign of any growth of ASE at high powers and much potential for further power scaling. Again, the peak at ∼976 nm is from the pump.

To further increase the output power with a straight-cleaved output end, two new diodes rated above 100 W were used (a 0.22 NA 105/125 μm fiber). These diodes are not wavelength stabilized, with their wavelengths approaching ∼976 nm only at maximum powers. A 5.93 m fiber was used coiled at 16 cm. A maximum output of 154.1 W was achieved at 992.8 nm (see Fig. 10), limited only by the maximum available pump power. The spectra at 14.4, 82.2, and 154.1 W are shown in Fig. 11, showing the absence of any ASE.

The first pump was increased to the maximum before the second pump was turned on. It can be clearly seen in Fig. 11 that the first pump was centered at ∼970 nm with ∼3 nm full width at half-maximum (FWHM) bandwidth at 14.4 W output power. Both pumps were turned on at 82.2 W output power with two pump peaks shown in Fig. 11. At the 154.1 W output, both pump wavelengths were at ∼976 nm. Because the pump wavelength was far from optimum at low powers, the linear fits in Fig. 10 excluded few data points at the low powers. The slope is 75% and 95% with regard to the launched and absorbed pump powers, respectively. M² fell between 1.05 at lower powers and 1.15 at the maximum power (see the inset of Fig. 11 for the mode profile at the maximum power).

In summary, an efficient cladding-pumped ytterbium fiber laser operating at ∼993 nm with a record low QD of ∼1.7% is demonstrated here for the first time. The much-reduced QD can lead to a significant increase in TMI threshold and consequently significant potential for further increase in diffraction-limited power from a fiber laser. A further decrease in laser wavelength can lead to a reduction in QD but at the cost of poorer efficiency as the laser operates more like a three-level system. Net gain can, however, be achieved above ∼985 nm while pumping at 976 nm, leading to potential QD down to ∼1%. In addition, this work significantly extends the lower wavelength limit of high-power ytterbium fiber lasers.

This work was supported by Army Research Office (Grant No. W911NF1910409).