1. Introduction

Even though Yb-doped fiber lasers are known to be the most powerful and most efficient among all fiber lasers, recent successes in the eye-safe ~1.5-um Yb-Er co-doped fiber lasers (where Er³⁺ ions are excited through the quasi-resonant Yb³⁺ ⇒ Er³⁺ energy transfer, while Yb³⁺ ions are used as a convenient ‘absorber’-agent pumped with well developed and the most efficient 9XX-nm laser diodes) are quite impressive. CW output power of Yb-Er co-doped fiber laser, the highest reported to-date, reached ~297 W level at 1567 nm [1] and optical-to-optical efficiency of up to 43% was also reported (though for much lower output powers of <120W) [2]. Nevertheless, as far as real eye safety requirement goes, multi-hundred Watt Yb-Er co-doped fiber lasers always have a significant fraction of a competing 1-μm Yb emission in their output (either narrowband or ASE) [2], which seriously compromises any “eye-safe” application. This circumstance, combined with the fact that optical-to-optical conversion efficiency of the Yb-Er co-doped laser is fundamentally limited by “never 100%” efficiency of Yb-Er energy transfer added to a hefty ~40% quantum defect of any Er-doped fiber laser pumped at ~9XX-nm, provides significant motivation for further exploration of the scaling potential of resonantly-pumped Yb-free Er-doped fiber lasers. With the resonant pumping approach heat deposition can potentially be limited by low quantum defect only, which opens up significant space for fiber laser power scaling with very minor thermal management complications. While achieving high optical-to-optical efficiency (up to 67%) with resonant core pumping at 1480 nm is possible [3], high-brightness Raman fiber laser pump source is the only pumping available in order to implement this Yb-free Er-doped fiber laser architecture, which is a major power scaling limiter [3]. Resonant cladding pumping should help accommodating lower-brightness highly multimode fiber-coupled 14XX-15XX-nm InGaAsP/InP fiber coupled diode modules, which presents the only truly practical scalable scheme to-date. This architecture is supported by to-date significantly more mature ~1.5–μm pump source technology and availability of 15XX nm fiber coupled modules [4]. One of the latest successes in laser efficiency, which illustrates the efficiency potential of the resonant pumping approach as it pertains to eye-safe Er fiber lasers, was the ~85%-efficient operation of a single mode Yb-free Er-doped fiber laser under resonant pumping [5]. This essentially implies that resonantly pumped Er-doped fiber lasers can be designed to be nearly as efficient as much more mature Yb-doped ones.

Early efforts reporting on resonantly cladding-pumped Yb-free Er fiber lasers are presented in [6–8]. In [6,7] output power of ~1W was achieved for the first time. The [7] happened to be the first effort actually exploiting the most scalable LMA fiber approach. The [8] was the first successful attempt to scale the Yb-free Er fiber laser significantly beyond the ~1 W power level. Single-frequency output power of 9.3 W was obtained in MOPA configuration from resonantly cladding-pumped EDFA with the slope efficiency of ~46% versus absorbed pump power [8]. Recently we reported much more scalable version of a resonantly cladding-pumped Yb-free Er-doped fiber laser [9], where ~48 W of single transverse mode output power were obtained with the ~57% optical-to-optical efficiency based on commercial off-the-shelf (COTS) laser fibers. Here we report the next step in optimization of low quantum defect resonantly cladding-pumped Yb-free Er-doped LMA fiber laser for power and efficiency scaling. Fiber laser operation with ~88 W of 1590-nm output was achieved still based on a COTS Yb-free Er-doped fiber. Maximum obtained optical-to-optical efficiency was measured to be ~69%. This is, to the best of our knowledge, the highest power reported from resonantly cladding-pumped Yb-free Er-doped LMA fiber laser, as well as the highest efficiency ever reported for any Er-doped laser, either Yb-co-doped or Yb-free.

2. Experimental details and results

Figure 1 shows the simplified optical layout of the Er fiber laser based on the latest version of commercial off-the-shelf nLight/Liekki Er60-20/125 double clad (DC) fiber. The laser is conceived to be a free-space version of a simple power oscillator pumped through the rear, high reflectance (HR) at 1590-1600 nm, mirror. The purpose of this free space arrangement was to enable us to optimize between the three options of a rear laser mirror: (i) nonselective dichroic (highly transmissive (HT) at the pump wavelength, 1532.5 nm, corresponding to the peak of the ⁴I_15/2 ⇒ ⁴I_13/2 Er³⁺ ion absorption in silica-based Er-doped fiber, and HR in the wide 1585-1615 nm wavelength range), (ii) volume Bragg grating (VBG) in a function of a wavelength selective dichroic mirror (HR in the narrow range of 1590+/−1 nm, and HT at 1532.5 nm), and (iii) fiber Bragg grating (FBG) matching the 20/125 µm DC fiber format, with ~95% reflectivity centered at ~1590 nm and with the ~99% transmission for the pump wavelength at 1532.5 nm. The FBG was written into an undoped matching fiber from the same manufacturer and one end of it was fusion-spliced to a doped fiber for this part of the experiment. The option with the nonselective dichroic end mirror was meant to provide the fiber laser with full freedom to “self-choose” the operation wavelength, thus pointing us to the right choice of a wavelength selector at the next stage of laser optimization, in which we are likely to chose some alternative custom made narrowband end reflector.

 

Fig. 1 Simplified optical layout of the cladding-pumped Yb-free Er fiber laser based on a COTS nLight/Liekki Er60-20/125 DC fiber resonantly pumped at 1532.5 nm. Laser cavity consists of the high reflectivity (one, of the choice of three, swappable, shown in Figure) dichroic mirror and the straight cleave at the output fiber end. Re-imaging optic at the pump end is not shown.

Download Full Size | PDF

For the sake of fair comparison of laser performance with the three above mentioned choices of rear mirror, the laser was free-space cladding pumped by six fiber-coupled laser diode modules (locked at 1532.5 nm and spectrally narrowed to 0.3 nm) power-combined in a standard (6+1)x1 pump combiner into a single 125 µm, 0.22 NA fiber. The power from the combiner was then free-space coupled, with the proper lens pair (not shown in Fig. 1), into the Liekki Er 60-20/125 DC LMA fiber and we have used three different optical layouts presented in Fig. 1, where the end mirrors shown as swappable (Fig. 1, segment in the middle framed into a rectangle). The optical layout in Fig. 1 is also simplified in that it does not show an AR coated (for 1500-1600 nm) 5.6-mm focal length collimating lens inserted between the laser fiber end and the nonselective dichroic mirror or VBG for pump options (i) and (ii), respectively. The Fresnel reflection of the straight-cleaved output fiber end was used as an outcoupling mirror.

The fiber laser efficiencies obtained for the three described above pump mirror options were found to be relatively close to each other. Indeed, the comparative relative slope efficiency figures achieved for a 15-m long Er-fiber for the three configurations under investigation were 1.0, 0.98 and 0.94 for the FBG, VBG and nonselective dichroic end mirror, respectively. The laser threshold was found to be nearly the same for all three cases: ~0.45 W of absorbed pump power. The observed differences in efficiency can be explained by higher insertion losses associated with, perhaps, marginal quality of the AR coating on the used 5.6-mm focal length collimating lens in cases (i) and (ii) with respect to a minor FBG splice loss present in case (iii). Due to the highest efficiency obtained with the FBG-based version of the laser further laser optimization was conducted with the FBG end mirror for which fiber length optimization was performed. We used a fiber laser simulation tool (Liekki Application Designer (LAD) software) to provide a starting point in the selection of proper gain fiber length. The optimum length was ultimately determined by an experimental cut-back procedure, the results of which are presented in Fig. 2 for the laser configuration shown in Fig. 1 with the FBG as a pump mirror. Output power scaling was not attempted in this experiment pursuing strictly fiber length optimization. The experimental results presented in Fig. 2 for different fiber lengths are: blue solid pentagons and the line across them – fiber length 21 m; red downwards solid triangles and the line – 16 m; wine upwards solid triangles and the line – 15 m; black solid squares and the line - 14 m. As seen from Fig. 2, maximum optical-to-optical efficiency of 69% has been achieved with the fiber length of ~15 m. To the best of our knowledge, this is the highest optical-to-optical efficiency ever reported for any cladding-pumped Er-doped laser, either Yb-co-doped or Yb-free.

 

Fig. 2 Fiber length optimization of the resonantly cladding-pumped Yb-free Er fiber laser based on a COTS nLight/Liekki Er60-20/125 DC fiber. CW laser output at 1590 nm versus absorbed pump power at 1532.5 nm is presented for the configuration shown in Fig. 1 with the FBG used as a pump mirror. Color coded data are: blue (solid pentagons and the line) – fiber length 21 m; red (downwards solid triangles and the line) – 16 m; wine (upwards solid triangles and the line) – 15 m; black (solid squares and the line) - 14 m. The highest slope efficiency of ~69% is achieved with the 15-m long fiber. The lowest slope efficiency of 63% was observed with the 21-m long fiber.

Download Full Size | PDF

Figure 3 presents the result of power scaling experiment for the optimal fiber length of 15 m with the FBG in a function of rear laser mirror (wine solid squares) overlaid with the results of our previous power scaling experiment [9] (blue solid squares). The output versus absorbed pump power dependence was obtained for the resonantly cladding-pumped Er-doped fiber laser with the configuration laid out in Fig. 1. The LMA fiber was coiled to about 10 cm diameter which was sufficient to achieve nearly single transverse mode output. Indeed, the beam propagation factor M² estimated at the power point of ~88 W and was found to be ~1.25. The IR beam profiler sensitized for a 1.5-1.6 μm spectral range (Spiricon, model LW 230) was used to measure cross-sectional beam profiles near the focal plane of the 125 mm focal length lens. Spectral output of the laser with FBG reflector (having a ~2 nm wide high reflectivity peak) was reasonably narrowband, but unstable in time. Using an optical spectrum analyzer (Yokogawa AQ6370) set to a 0.05 nm resolution we found that there were typically 2 - 3 not quite fully resolved competing emission peaks observed in the laser output around 1590 nm, but they were all confined within a 0.3 nm spectral width. One of the snapshots of the spectral distribution for the cladding-pumped Er-doped fiber laser power taken at 88 W power point is depicted in Fig. 4 .

 

Fig. 3 Output laser power versus absorbed pump power dependence (wine squares) obtained for the length-optimized resonantly cladding-pumped Er-doped FBG-laser shown in Fig. 1. The results of our previous power scaling experiment [9] are presented (blue squares) for the sake of comparison.

Download Full Size | PDF

 

Fig. 4 A snapshot of the cladding-pumped Er-doped fiber laser power spectral distribution obtained using an optical spectrum analyzer set to a 0.05 nm resolution.

Download Full Size | PDF

The maximum of ~148 W of absorbed power at 1532.5 nm (out of 170 W of launched power) resulted in a maximum of ~88 W of output power at 1590 nm out of ~15 m long Yb-free Er-doped fiber. To the best of our knowledge, this is the highest power ever reported from Yb-free Er-doped LMA fiber laser.

Analysis of our experimental results with the spectrally nonselective pump mirror (mostly laser output spectral behavior without optimization), indicates that the laser spectral output in this case is extremely unstable in time with spectral features peaking anywhere between 1603 and 1608 nm. It clearly indicates that further efficiency optimization requires some alternative custom made narrowband reflector, e.g., FBG, with the peak reflectivity around 1605 nm. This replacement alone may lead to an improvement in optical-to-optical fiber laser efficiency to a level well beyond 70%.

3. Conclusions

Highly scalable, efficient, low quantum defect laser operation of the resonantly cladding-pumped Yb-free Er-doped laser based on COTS LMA fiber has been demonstrated. Our experiments were carried out as the next step in power and efficiency scaling of this type of laser. Single transverse mode operation with ~88 W of 1590-nm output was achieved. Maximum obtained optical-to-optical efficiency was 69%. This result presents, to the best of our knowledge, the highest power ever reported from resonantly-pumped Yb-free Er-doped LMA fiber laser, as well as the highest efficiency ever reported for any cladding-pumped Er-doped laser (Yb-co-doped, or Yb-free). Further efficiency scaling is expected via optimization of the peak reflectivity wavelength of the implemented selective end mirror.