1. Introduction

In the past few years, numerous ytterbium (Yb)-doped fiber (YDF) lasers and amplifiers with various output characteristics have already been successfully transitioned from laboratory prototypes to commercially available devices [1–5]. When the output power is ≤2 kilowatt (kW) at 1-1.1μm, the commercial fiber lasers can be catalogued as “low- and mid-power fiber lasers” [1,2,4,5]. These YDF lasers have been leading the innovation in the various processing industries and substitute for conventional processing equipments [1,2,4,5]. However, multi-kW (≥3kW) level high-power fiber lasers for industrial application had encountered numerous technology challenges because it typically requires high launched pump power and consequently high level of excited Yb³⁺ ions [1–5]. One of the important set of challenges for the multi-kW fiber laser technology is associated with maintaining the long-term high power reliability [6]. Long-term laser power reliability is related to a so-called photodarkening (PD) in YDF gain medium and nonlinear (NL) effects of laser setup systems such as stimulated Raman scattering (SRS), stimulated Brillouin scattering (SBS) [1–8].

To reduce NL effects, short fiber length is required. The common ways to achieve this are enlarging effective mode area or doping high Yb³⁺ ions concentration so as to increase pump absorption [1–8]. The major drawback of enlarging effective mode area is that the number of modes supported by the large mode area (LMA) fiber increases. In this case, the limit for fiber laser systems stably working at multi-kW level is mode instability (MI) phenomena [9–11]. Otherwise, V. Agrež et.al [14] showed recently that MI threshold improvement of YDF was mainly limited by PD effect as the dominant heat-source during high-power operation. In addition to this, highly Yb³⁺ ions doping is very effective to suppress NL effects [12], but easy to result in ion-clusters, which in turn increases PD possibility [13]. So, to develop high quality YDF with almost PD-free is urgently needed for high power fiber laser development.

PD, i.e. pumped induced colored centers formation, is a time dependent limiting factor of YDF laser, because it causes broad-band excess loss which depletes a pumping light (900nm-1000nm) and a signal light (1000nm-1100nm) [7,8,13–18]. Mechanisms behind PD effect have been widely investigated for many years [7,8,13–21], but still under controversy. Mechanisms of induced color center formation were mainly regarded as be correlated to Yb³⁺ ions clustering and charge transfer [7,8,13–18], greatly depending on Yb content, inversion level, the concentration of Yb-Yb pairs, and defect species neighboring Yb³⁺ ions [13,17]. Based on this, a number of PD-suppression methods including large mode fiber structure design (such as reduced mode overlap design [22] and D- and P- shaped design [23]), thermal bleaching [24], and 916nm pumping [25], etc., were proposed recently and proven to be instrumental to alleviate PD of YDF. Compared with these methods, a core composition with equal content of Al and P is more efficient to suppress PD in fiber core. In order to develop multi-kW laser fibers free from PD phenomenon, aluminophosphosilicate (Al₂O₃-P₂O₅-SiO₂, ternary APS) with equal content of Al and P attracts more and more attention in recent years, and has been shown to be more suitable as core host material compared to other binary glass materials [19,20], as it is capable of achieving almost completely PD-free, a low-core background loss, and a low numerical aperture (N.A.) due to [AlPO₄] formation in ternary APS glass matrix [3,13,16,21].

YDF fabrication technology and condition, especially rare-earths (REs) doping techniques [26–32], are also of great importance for PD resistant. Recently, by modified chemical vapor deposition (MCVD) system in conjunction with solution doping technique (SDT), Yb-doped APS fiber with equal content of Al and P has been shown strong PD-resistivity [16,26]. Compared with SDT, RE-gas-phase doping technique (GPDT) has some obvious advantages in terms of RE-doping uniformity and homogeneity, higher doping concentration, accurate control of dopant ratio, and easy modulation of refractive index profiler (RIP) [32–38]. Thus, GPDT can be expected to be very promising for low PD and high quality YDF development. Aluminosilicate (AS) glass material with high Al/Yb ratio [27,28], fabricated through GPDT, presented lower PD induced loss as compared to that obtained from SDT [33]. By GPDT, an Yb-doped APS fiber with 10 mol% P₂O₅ and 9.5 mol% Al₂O₃ was fabricated and showed strong PD-resistivity [32]. Recently, our research group fabricated a homogeneous 20/400 (core/cladding diameter, μm) Yb-APS fiber by all-gas-phase chelate precursor doping technique (CPDT). The fiber presented 3.03kW laser output at 1080nm, and showed long-term running stability of 500 minutes at 2.1kW [39]. However, SRS component occurs at 2kW-level and restricts further output power scaling at 1080nm. To directly collect high-power pump energy from commercial laser diodes (LDs), a large-scale Yb-APS fiber with 30μm core and 900μm inner-clad in diameter was also proposed and fabricated by our group. The fiber showed 6.85kW laser output at 1079.79nm with an optical-to-optical efficiency of 67.03% [40]. However, further power scaling is not straightforward but limited by low optical-to-optical efficiency and obvious nonlinear effects. Small core to clad ratio can contribute to low thermal load but will result in weak cladding absorption of pump light. As such, the 30/900 large-scale fiber brings less advantage to improve its optical-to-optical efficiency, suppress NL effects due to long fiber length in use, miniaturize fiber laser integration for real industry application. It also requires the complex procedures to cut, splice and taper [40]. Moreover, the whole pump system in our previous report [40] is difficult to stabilize for a long time. This is due to the accumulating heat at the quartz block holder (QBH) of bar pumps. Thus, no added experiment was done in [40] to address PD inhibition of Yb-APS fiber.

To scale laser power while suppressing NL effects, we designed and fabricated a 30/600 Yb-APS fiber by using novel all-gas-phase CPDT in combination with traditional MCVD system here [36–39]. An excellent laser performance could be demonstrated for such a fiber with an output power of more than 5kW and a slope efficiency of 85.2%. Long-term laser stability at 5kW-level was carried out at a master oscillator power amplifier (MOPA) laser system for PD effect examination.

2. Design and experiments

2.1. Fiber design and calculation

In contrast to AS and phosphosilicate (PS) host materials, APS with Al/P molar ratio of 1 has peculiar properties such as refractive index, density thermal expansion, etc., which are attributed to the formation of [AlPO₄] units. Thus, nearly equal content of Al and P was designed here for high suppression of PD and low N.A. of fiber core [3,13,16,21]. On this basis, excess Al content (i.e., Al/P molar ratio of a little over 1) was designed for the reduction of background loss and the central dip of P element [3,21]. Refractive index difference Δn of the fiber core relative to undoped SiO₂ is determined by Yb³⁺ ions content and concentration of the codopants. In most case, an additivity rule is valid with respect to the molar composition with constant increments, described in detail in [3]. However, for APS host materials, the additivity rule is no more valid, and the Δn is influenced by the molar ratio of Al and P, Al and P content [21]. Here, as total numbers of Al is more than that of P, the Δn can be expressed as [21]:

2.2. Fiber preform fabrication and measurements

To reach nearly “PD-free” state, an Yb-APS fiber preform, with core diameter of 2.2 ± 0.04mm and core composition of Yb₂O₃-P₂O₅-SiO₂, was fabricated by a conventional MCVD system in conjunction with CPDT, as reported in [36–39]. In order to prevent impurity and achieve 5kW-level fiber preform, ultra-high-purity (6N, 99.9999%) raw material Yb(thd)₃ was fully synthesized and purified in-house. The purity of other raw materials and process gas is above 5N. Organometallic chelate tris (2, 2, 6, 6,-tetramethyl-3, 5-heptanedonato)-Ytterbium (Yb(tmhd)₃, 99.9999%, precursor for Yb₂O₃), Aluminum(III) acetylacetonate (Al(acac)₃, 99.999%, precursor for Al₂O₃), POCl₃ (99.9999%, precursor for P₂O₅), and SiCl₄ (99.9999%, precursor for SiO₄) were used as starting raw materials. The raw materials were put in different sublimators heated at the corresponding sublimating temperature of each chemical. Compared with metal halides (i.e., YbCl₃), organic chelate precursors (i.e., Yb(tmhd)₃ and Al(acac)₃) have much lower sublimation temperature (≤200°C) and much higher saturated vapor pressure. Therefore, high-level Yb³⁺ ions are easy to be doped into fiber core without any observed concentration quenching effects. To realize mixture uniformity of reaction gas and uniform deposition during MCVD process, the original substrate tube used here was F300 silica tube with a large diameter of 24/28mm. The tube was etched by HF acid for 6 hours so as to make tube wall thin. Before cladding deposition, substrate tube wall continued to be thinned by SF₆ etch procedures with a burner at 2050°C. As a result, the tube wall decreases from 2mm to 0.9 ± 0.1mm. The ultra-thin tube is instrumental to obtain a lower collapse temperature and less number of collapse passes. Thus, it can suppress evaporation of deposited P₂O₅ and simultaneously keep Al₂O₃ and Yb₂O₃ in central deposition layers. Experimental processes and procedures to make Yb-APS fiber preform were further optimized for achieving both a high PD resistivity and low background loss. During fiber core deposition process, increasing P pressure in the last two passes contributes to getting a homogeneous glass composition in the core and as a result, a low attenuation loss was expected. Besides, gas flow, deposition temperature, and collapsing temperature were modulated and optimized to decrease elements evaporation and defects formation. Considering easy evaporation feature of P₂O₅ content, the molar ratio of Al/P was controlled to be 1.1 ± 0.05 to minimize the central dip of the refractive index profile (RIP). Furthermore, during soot deposition and collapsing processes, excess O₂ gas environment was kept to reduce possible precursors for PD. After elongated to a certain diameter, the made original preform was overcladded by suitable F300 fused silica tube to adjust the required core/clad ratio. For the characterization and PD experiment, the preform was shaped and finally drawn into 30/600 Yb-APS fiber. RIP of the fiber was measured by the refracted near field method [36–39]. The cladding absorption spectrum was measured with an optical spectrum analyzer (OSA) [36–39].

2.3. Laser performance measurements

Continuous wave (CW) laser performances of the home-made 30/600 Yb-APS fiber were characterized in a MOPA system, as shown schematically in Fig. 1. The first oscillator stage is in accordance with that reported in [37]. Before injecting pump power into the amplifier stage, traditional cut-back method was used to measure signal loss at 1064nm of Yb-APS fiber core. The propagation loss was measured as 32 ± 2dB/km at 1064nm, comparable to that (28dB/km) of Nufern-20/400-M [41]. A home-made 30/400 cladding light stripper (CLS) was used as a “bridge” to connect the oscillator with the amplifier stage. Seven groups of 150 ( ± 10)W 976nm LDs (976nm-LDs) were integrated through a 7 × 1 fiber combiner and supported 1020 ( ± 20) W pump power output. Thus, different from our previous report [40] using bar pumps as pump source, the pump source in this work can operate for a long time at no risk of burning pump system. So, long-term laser stability examination of MOPA system can carry out for PD-resistivity measure. Feeding pump power is 6.14kW from six groups of integrated 1020 ( ± 20) W 976nm-LDs via a (6 + 1) × 1 fiber combiner. This combiner was specially-designed with pump pigtail fiber of 220/242μm and signal pigtail fiber of 30/600μm. With 10°C water-cooling, 30/600 Yb-APS fiber was deployed with 200-350mm coiling diameter on an aluminum plate. Fiber length of 18m was used to suppress NL effects while ensuring enough pump absorption. High-order modes (HOMs) might leak into pump cladding due to possible TMI occurrence in fiber core, and were easily stripped by the CLS. In the case, power degradation was responsible for both PD and TMI. Thus, no the CLS was used in the amplifier stage to eliminate the effect TMI on laser stability test. One homemade QBH without beam collimation system was spliced to deliver the output signal light detected by a power meter (PM) at the end.

 

Fig. 1 Schematic diagram of MOPA laser set-up. (LDs: laser diodes; FBG: fiber Bragg grating; CLS: cladding light stripper; QBH: quartz block holder; PM: power meter)

Download Full Size | PDF

3. Results and discussion

3.1. Cross section and refractive index profile

Figure 2 presented the cross-section picture and RIP of the home-made 30/600 Yb-APS fiber. The fiber core diameter is 30.8μm, and the F300 octagonal cladding (flat to flat) was 603.2μm in diameter. Compared with 20/400 Yb-APS fiber, 30/600 Yb-APS fiber contributes to aggregating more pump power with lower thermal load distribution along the fiber body [40]. The effective N.A. of the fiber core was calculated to be 0.062, corresponding to the ∆n of 0.0013 between core and F300 pump cladding. It is important to note the well-known central dip, which comes from the natural evaporation of P₂O₅ [3,13,16,21] during MCVD processes, was disappeared due to our special composition design. In addition to this, central dip elimination might be also attributed to the high temperature (2050 ± 50°C) and low-speed (4.5 ± 0.5m/min) condition of fiber drawing [42] for which dopants diffusion behavior in radial direction takes place. This is essential for achieving good beam quality with 30.8μm core dimension. The fluctuation of the core refractive index is estimated to be 2 × 10⁻⁴ comparable to that of Yb/Ce-AS fiber [43]. Thus, CPDT is able to adjust perfectly RIP of Yb-APS fiber although P₂O₅ content is extremely easy to evaporate.

 

Fig. 2 (a) Cross-section and (b) measured RIP of 30/600 Yb-APS fiber.

Download Full Size | PDF

3.2. Elemental distribution

To examine the reason of the central dip-free, the elemental distribution and molar ratio were analyzed by electronic probe micro analyzer (EPMA). The results were given in Fig. 3(a) and 3(b). The molar percent of Yb³⁺, Al³⁺ and P⁵⁺ is measured to be 0.15mol%, 1.7mol% and 1.4mol% respectively, less than the design values due to elemental evaporation. To further demonstrate element distribution, Al/P and (Al + P)/Yb as a function of radius were calculated and presented in Fig. 3(c). It can be found Al/P and (Al + P)/Yb ratio in the fiber core is close to 1 and 20 respectively, while that of central region in the fiber core was up to 31 and 1.7 due to evaporation of P₂O₅. Thanks to large molar ratio (Al + P)/Yb of over 20 (see the Fig. 3(c)), Yb³⁺ ions are homogeneously distributed and effectively prevented from clustering. Thus, it is helpful for PD suppression. With Al³⁺/P⁵⁺molar ratio of 1.1:1, low refractive index difference and suitable N.A. of 0.062 were obtained because of [AlPO₄] formation. The usually detected central dip of Yb element in APS fiber was not only originating from the sublimation of P₂O₅ [38] but also Yb evaporation itself [36] during MCVD process. Yb evaporation evidently leads to the decrease of refractive index, especially in the central region of core. At the same time, evaporation of P₂O₅ isn’t beneficial for [AlPO₄] formation and makes the ratio of Al/P much larger than 1 in the central region of fiber core, resulting in increase in refractive index. As depicted in Fig. 3(c), the molar ratio of Al³⁺/P⁵⁺ is 1.15 ± 0.05 for outer core region, while that of the central region of core is up to 1.72 ± 0.02. Thus, Yb evaporation almost offsets P₂O₅ evaporation for case of refractive index in the central region of fiber core. As such, the central dip was free. Based on the data of radial molar dopant concentration, the RIP was calculated by Eq. (1) and shown in Fig. 3(d). Indeed, no central deep can be found. However, owing to measure error of dopant concentration by EPMA, the difference between the calculated and the measured RIP can be observed. Furthermore, the difference is also responsible for stresses and thermal treatment history influencing on refractive index in APS fiber [44]. Refractive index calculation in APS fiber is further improved by improving measure accurate and revising molar refractivity coefficient, intended as a future work.

 

Fig. 3 Radial dopant concentration: (a) Yb³⁺ and (b) Al³⁺ and P⁵⁺; (c) Al/P and (Al + P)/Yb ratio as a function of radius; (d) the calculated RIP of 30/600 Yb-APS fiber.

Download Full Size | PDF

To assess the homogeneity of the Yb-APS core, the element area distribution of Yb₂O₃, P₂O₅, Al₂O₃, and SiO₂ was characterized by EPMA with area analysis mode. The results were illustrated in Fig. 4. It can be seen that all the elements are homogeneously distributed in outer region of the fiber core, which is helpful for reduction of background loss and PD precursors. However, compared with outer core region, central core region has a lower dopant concentration and heterogeneity due to elemental evaporation, which is the main drawback for Yb-APS fibers [26,28–41]. Considering impurity and water or OH^- groups could penetrate to core layer during fiber preform fabrication and hence cause PD effect, SiO₂-P₂O₅ cladding layers were deposited next to fiber core and acted as waterproof layers, as shown Fig. 4(b).

 

Fig. 4 EPMA area analysis of Al (a), P (b), Si (c) and Yb (d) elements.

Download Full Size | PDF

shows the cladding absorption spectrum range from 600nm to 1700nm of 30/600 Yb-APS fiber measured by the cut-back method. Absorption peak coefficient α was tested to be 0.42dB/m at 915nm and 1.29dB/m at 976nm, respectively. Owing to Yb³⁺ ions evaporation, the absorption coefficients are lower than that (1.5dB/m at 976nm and 0.48dB/m at 915nm) of Nufern-20/400-M fiber [41]. In the case of MOPA configuration, fiber length of 18m is used to ensure enough pump absorption. No water-removal measures lead to high OH^- absorption coefficient of 0.25dB/m at 1383nm.

 

Fig. 5 Absorption spectrum of 30/600 Yb-APS fiber.

Download Full Size | PDF

3.3. Laser performances

Laser output power was measured experimentally and linearly fitted as a function of the input pump power in Fig. 6(a). An 18m-long 30/600 Yb-APS fiber was pumped up to an output power of 5.19kW at 1064nm, limited by the available pump power. The linearly-fitted slope efficiency reached 85.2%, much higher than that (67.3%) of [40]. The slight roll-off at the maximum power is related to a non-corrected degradation of brightness of the pump diode at the power level. The easily-fused splicing point between the combiner and the 30/600 fiber was 20 ± 2°C, much lower than the usual thermal damage threshold of 5kW-level fiber amplifier. It is potential to reach higher laser output power if with more pump power. The laser output spectrum is centered at 1064.39nm, and the full width at half maximum (FWHM) is 1.67 ± 0.01nm (see Fig. 6(b)). Owing to no CLS in the amplifier stage, residual pump light centered at 976nm can be observed, while the ratio of signal light to residual pump light is up to 29.4dB, as shown in the inset of Fig. 6(b). That is to say, residual pump intensity is 1/870 of signal light intensity. So, residual pump light can be neglected in this study. Benefiting from short fiber in use, low seed-laser power, and large mode area of 30/600 Yb-APS fiber, neither SRS component nor amplified spontaneous emission (ASE) in the spectrum was observed at 5kW-level. The beam quality factor M² in our experiment can’t be measured because no corresponding commercial QBH was used to splice with the output fiber.

 

Fig. 6 Fiber laser experimental results: (a) output power and slope efficiency, (b) laser output spectrum. The inset is laser output spectrum (in dBm).

Download Full Size | PDF

To directly characterize the PD effect, power fluctuation at 5.16kW continuous-wave output was presented at Fig. 7. This MOPA laser setup was continuously stabilized at 5.16kW for over 600 minutes. One can see almost no power degradation was found, directly justifying excellent long-term laser stability and nearly PD-free. This high PD resistivity should be originating from equal content of Al and P, the uniform distribution of Yb³⁺ ions, proper Yb³⁺-doping concentration, the low PD precursors and defects induced by the optimized fabrication process, and an oxygen-rich environment originating from P₂O₅ addition and O₂-excess atmosphere. A 30/520 Yb-APS fiber, drawn from the original preform in this study, was used as signal fiber of pump-gain integrated functional laser fiber, presenting as high as 10.45kW laser output at 1080nm [45]. This result indirectly justifies strong PD-resistivity character of the fabricated Yb-APS fiber in this work. PD was regarded as the most dominant heat-generation mechanism in active fibers [14]. Consequently, 30/600 Yb-APS fiber studied here can be excepted with high MI threshold and even nearly MI-free. Unfortunately, we canʼt examine MI here for the whole laser setup because no high-speed camera and photodiode can be used in our laboratory.

 

Fig. 7 Power stability test of laser systems setup at 5.16kW.

Download Full Size | PDF

4. Conclusion

In conclusion, we reported on theoretical design and experimental fabrication of an Yb-doped aluminophosphosilicate fiber with 30μm core and 600μm inner-cladding in diameter. The original fiber preform was fabricated by modified chemical vapor deposition system combining with chelate precursor doping technique. 0.15mol% Yb₂O₃ was dissolved into the fiber core plus with 1.7mol% Al₂O₃ and 1.4mol% P₂O₅. Thanks to Al³⁺/P⁵⁺ molar ratio of a little over 1 and dopants diffusion behavior at high drawing-temperature, central dip of refractive index profile was successfully removed. With a master oscillator power amplifier configuration directly forward-pumped by 976nm laser diodes, 18m-long 30/600 Yb-APS fiber presented 5.19kW laser output at 1064.4nm with slope efficiency of 85.2% and intense nonlinearity suppression. The Yb-APS fiber-based MOPA system was kept at 5.16kW output for over 600 minutes, and no sign of power degradation was obtained. It implies excellent laser stability and low photodarkening effect. The results justified that aluminophosphosilicate fiber is a strong photodarkening-resistivity host material even induced by over 6kW pump light. Future research work will be transferred to prevent P vaporization losses and increase P pressure during MCVD process so as to improve element uniformity of the fiber core. We also shall focus on mode instability phenomena of 5kW-level MOPA system based on nearly photodarkening-free 30/600 Yb-APS fiber.