1. Introduction

Ytterbium-doped fiber laser (YDFL) suits for powerful laser light with high beam quality. Particularly, YDFLs outputting around 1018 nm attracted rapidly increasing interests for their lower quantum defect in high-power output from 1050 to 1080 nm via tandem pumping configuration [1], comparing with directly pumping by 976 nm diode lasers. Moreover, 1018-nm YDFLs recently prove efficient in obtaining versatile wavelengths, such as 1556 nm [2] and 1178 nm [3] (then 589 nm). They qualify candidates for, i.e., femtosecond pulse generation [4], optical parametric oscillator [5], super-fluorescent source [6] and Raman amplifications [7] in many circumstances. These features make them powerful springboards towards physical research, manufacturing and biomedical [3] applications. Generally, high output power, high optical efficiency and high beam quality are all required features of a 1018 nm YDFL for application. However, these features rarely concur in traditional YDFLs. For acquiring high output power and efficiency, Ytterbium-doped fibers (YDFs) of larger core diameters are favored. The larger core-diameters offer better overlap between pump light and active dopants. For example, Jiang et al [8] and Chen et al [9] achieved 107.5 W and 307 W, respectively in 2015 and in 2017, using 15/130 large-mode-area (LMA) YDFs of nominal absorption coefficient up to 11 dB/m. Xiao et al [10] and our group [11] even realized 476 W and 805 W, respectively in 2015 and 2017, using 30/250 LMA YDFs. In 2018, with technically challenging bidirectional-pump design, our group realized 1150 W [12] further, which is the highest record at 1018 nm to date. However, the large-core-diameter YDFs being in cavity, the output beam qualities were usually worse than 1.8-M²-factor [10–12]. For obtaining high-power single-mode 1018-nm light, single-mode YDFs became a widely adopted choice. In 2015, Ottenhues et al [13] broke 200 W output with 53% optical-optical efficiency (absorbed-signal efficiency as 75%) using 10/130 YDFs. But no information of beam quality was disclosed. In 2016, Glick et al [14] pushed the benchmark to 230 W (stably at 220 W) and 75% with 1.17-M²-factors using 10/125 YDFs, being the highest record to date that uses double-cladding YDF. Additionally, it is noteworthy that very recently, an oscillator using specially-designed 50/400 phosphosilicate photonic bandgap YDF suppressed M²-factor to 1.35 (single-mode) while outputting 240 W [15]. The photonic-bandgap fibers, on which great hopes were placed, promise high-power output with controlled ASE. Yet, it required complex auxiliary devices, such as special fiber Bragg gratings (FBG) and direct coupling systems.

In obtaining single-mode high-power high-efficiency 1018-nm light, difficult choices must be made. We note that it was very hard to use the well-known fiber-coil method to obtain high-efficiency and high-beam-quality simultaneously from multimode YDFs. As can be seen, even if the 30/250 YDFs were coiled with small radius of 5 cm [12], it could only produce 1.96-M². As this paper will show later, coiling even 20/130 YDF with such a small radius will neither be effective. Coiling the YDFs with further smaller radii would, while improving higher-mode suppression, cause considerable leakage of the large-NA pump light. Since the YDFs of smaller cladding diameters had worse potential of heat dissipation and suffer from higher density of intensity of the leaked pump light on their coating surfaces, temperature increases would be very significant and dangerous. It could lead to catastrophic chain destruction of the whole system, known as fiber fuse effect [14,16]. Coiling YDFs of larger cladding-diameter would be sometimes applicable for controlling beam quality while having less troubles of heat. For example, a 20/400 YDF could be coiled with 5-cm radius and produce M² around 1.3 in experiment [17]. However, these YDFs could neither be an acceptable choice for high-power high-efficiency 1018-nm fiber laser, as they were severely limited by strong ASE. In fact, ASE was particularly a major factor of limitation in short-wavelength fiber lasers including 1018-nm ones. Generally, longer YDFs will cause more gain for ASE. Meanwhile, YDFs of smaller ratios of core/cladding areas and higher Yb-dopant concentrations will suffer from more ASE [18]. The YDFs of larger cladding-diameters, such as 20/400 ones, have both the attributes (their core diameters cannot be too large, otherwise the beam quality cannot reach single-mode via coiling without safety problems; therefore, they often had smaller ratios of core/cladding areas. In this case, however, the dopant concentrations have to be high for creating usable absorption coefficient, as the smaller ratios of core/cladding areas means worse overlap between pump light and active cores). That is, the limitation of the lengths of the YDFs in a 1018-nm setup will be extremely severe, although they can be coiled to control beam quality. The severe limitation of length will then result in very low efficiency of the laser. For example, the latest 1018-nm setup using 20/400 YDF could only have a 39.4% efficiency [19]. Moreover, the limitation of length of YDFs also prevent the use of main oscillation power amplification (MOPA) setup to scale the output power of 1018-nm fiber laser. As can be seen, all the high-power 1018-nm fiber lasers mentioned [8–15,18] were actually of oscillator setups; MOPA structure was intentionally avoided. Instead, for suppressing the ASE, Glick [14] has intricately shortened their YDF by using high nominal absorption coefficient calculated as 5.18 dB/m at 975 nm and technically challenging bidirectional pump. As can be seen, it was very difficult to scale power further in current setup of 1018-nm fiber laser while preserving both high-efficiency and single-mode characteristics. To do that will require new experimental structures of fiber lasers.

In this study, we validate an efficient hybrid-structure monolithic setup that adopts two YDFs simultaneously to shoulder different duties for lasing at 1018 nm. By combining the merits of a high-beam-quality YDF and a high-absorption one together, this setup allows significant power scaling without degrading beam quality. It realizes single-mode output from multimode fibers without intentionally bending the fibers to select modes. We validated the advantages of this setup over traditional oscillator by control experiment that also surpassed the highest record performance. This setup may open significantly higher regions of output power unavailable by traditional oscillator, while maintaining high beam quality and efficiency. It has unique advantage for many short-wavelength single-mode fiber lasers where strong ASE and following parasitic oscillation deter the application of MOPA structures.

2. Principle and experimental design

2.1 Concept of the hybrid structure and comparison

In our perspective, gain fibers in oscillators undertake two major functions: controlling beam characteristics (indicated by beam quality) and absorbing pump power (thus scaling the final output power). As we struggle in selecting the most suitable gain fiber, an ignored possibility is that these functions can be shouldered by two different fibers concurrently, each of which is suitable for one job while not interfering the other. The essential concept of this hybrid-structure setup is illustrated in Fig. 1a, while that of typical oscillator adopted by previous studies and that of amplifier are respectively shown in Fig. 1b and c. In the hybrid-structure setup, the high-beam-quality gain fiber has the capability of producing high-beam-quality signal solely in a traditional oscillator. Likewise, the high-absorption gain fiber has high absorption of pump power solely therein. By properly shortening the high-beam-quality fiber, the cavity only produces a weak, yet of high-beam-quality, signal with a large amount of residual pump power. Thereafter, the residual pump is absorbed by the high-absorption gain fiber that is also properly short. Here, the beauty is to manipulate carefully the beam quality and the efficiency. High-absorption gain fibers of larger core diameters and NAs are known to produce inferior beam quality in traditional oscillators and amplifiers. We attribute this to that they are in most cases fed by light combined via tapered-fiber combiners or couplers. This configuration induces persistent degradation of beam quality of the light entering the fibers, which activates lasing in all areas inside the cores. This caused huge difficulties in obtaining usable beam qualities of high-power 1018-nm light from, i.e., traditional MOPA configurations. Moreover, for short-wavelength fiber lasers such as 1018-nm ones, strong ASE and following parasitic oscillation also forbade amplification by long gain fibers. While one may attempt to shorten the gain fiber in MOPA to reduce the ASE, the efficiency of the laser would drop sharply. Contrarily in the hybrid-structure setup, the high-absorption gain fiber was directly fusion-spliced to the high-beam-quality gain fiber. This localized sudden change would degrade the beam quality of the entering signal at first, yet create mostly non-guided modes. However, in the amplification process, more signal will be generated and then strictly confined in the path where signal is strong. That is, the non-guided modes will quickly dissipate without significant higher-mode amplification. It is expected that the beam quality of the signal would deteriorate only mildly in this self-sustaining process. That is, the high-absorption gain fiber may inherit an output of undiminished beam quality as the high-beam-quality gain fiber solely does in a traditional oscillator. As beam quality is ensured, the high-absorption gain fiber, supporting lower power density, better alleviates nonlinearities dependent on fiber length and power density. In this way, this setup is expected to produce stably higher signal power and efficiencies, while traditional single-mode oscillators were severely limited by strong nonlinear effects when the gain fibers became too long.

 

Fig. 1. The essential concept of (a) the hybrid-structure setup, (b) typical fiber oscillator and (c) typical fiber amplifier.

Download Full Size | PDF

2.2 Design of the 1018-nm fiber laser and control experiments

For demonstrating the technical feasibility, we designed a 1018 nm YDFL based on the hybrid-structure setup, of which the scheme read Fig. 2a. Six 105/125 (NA = 0.22) fiber-pigtailed 976-nm diode lasers were combined together by home-made 7×1 tapered fiber combiner, serving as the pump. The output powers of the combiner were measured as launched pump powers. A 1.3-meter 10/130 LMA double-cladding YDF (single-mode, NA = 0.06/0.46) with nominal absorption coefficient 5 dB/m at 976 nm served as the high-beam-quality gain fiber. The cavity was assembled by the high-beam-quality YDF, a 99.9% high-reflection (HR) FBG and a 10% output-coupling (OC) FBG at 1018 nm. The FBGs were inscribed on matched 10/130 passive fibers. The HR was fusion-spliced to output of the combiner; the OC was fusion-spliced to the high-absorption gain fiber, which was a 1.3-meter 20/130 LMA double-cladding YDF (multimode, NA = 0.06/0.46) with nominal absorption coefficient of 13 dB/m at 976 nm. The end of the high-absorption YDF was fusion-spliced to a cladding light stripper (CLS), which was of 20-dB pump loss and based on 20/130 passive fiber, to remove the unabsorbed pump power. The output end of the fiber was cleaved to be 8˚ beveled for suppressing inner-end-face reflection. Finally, all the YDFs were coiled with a 5-cm radius. Note that this radius was large for the YDFs of 130-µm-cladding-diameter that it would not cause significant mode selection (as shown later in the control experiment Setup C).

 

Fig. 2. The experimental setups of the 1018 nm YDFLs using (a) Setup A the hybrid-structure, (b) Setup B the traditional oscillator structure with solely the high-beam-quality 10/130 YDF and (c) Setup C that with solely the high-absorption 20/130 YDF.

Download Full Size | PDF

In producing single-mode high-power light, it is essential not to introduce any beam-quality degradation during the lasing processes. As the hybrid structure avoided using any intermediate combiners that were involved in between gain fibers in MOPA structures, a fusion point between the two gain fibers still existed in Setup A. Therefore, before experimental demonstration, numerical simulation was performed to anticipate the beam-profile evolution across the fusion point in Setup A. For approaching the problem, we assume that the signal produced by the cavity of Setup A was single-mode, as was reasonable. Parameters of the 10/130 YDF were used to generate a 152 W base-mode-profile 1018-nm light, which simulated the signal at the end of the cavity in Setup A. The signal propagated into the high-absorption gain fiber with a 226 W plain-intensity-profile 976-nm light simulating the residual pump. The above power parameters came from preliminary calculations based on rate equations. The evolution of beam profile in the high-absorption YDF was solved by comprising beam propagation, rate-equation-solved amplification and thermal effects that vary refractive indices. The beam quality M²-factor of the signal profile in each cross-section of the fiber was calculated. As is shown in Fig. 3, the simulated beam quality of signal only degraded at the very first of the fiber; it then converged to its initial value, suggesting that most light amplified were base-mode component. Therefore, it was expected that Setup A could produce similar beam quality with that of traditional oscillators, as Setup B was. Note that this result does not prevent other 20/130 fiber devices from causing beam-quality degradations. As can be seen later, the multimode gain structure of Setup C, comprising of 20/130 YDF and fiber devices, resulted in worse beam quality than those of Setup A and B.

 

Fig. 3. Simulation results of the evolution of beam quality of signal light in Setup A. Right-sided is enlarged part. Pseudocolor maps show intensity of signal light in the longitudinal section of the high-absorption YDF. The region before zero points of the horizontal axes is the 10/130 passive fiber (with the OC being inscribed on) that was fusion-spliced before the high-absorption YDF.

Download Full Size | PDF

In the experimental demonstration, the output signal powers of Setup A were measured by power meter, as shown in Fig. 4. The optical-optical efficiencies were calculated with the launched pump powers. The optical spectra were characterized at several output powers by an optical spectrum analyzer (Yokogawa 70D) using the highest resolution 0.020-nm, as accordingly shown in Fig. 5. The beam qualities were measured at the highest output power by a laser quality monitor (PRIMES GmbH) using standard 2^nd-moment algorithm, as is shown in Fig. 6. Afterwards, two control experiments, namely setup B and C, were made to compare the characteristic of this hybrid-structure setup with traditional fiber oscillators. Setup B, of which the scheme is shown in Fig. 2b, was built by replacing the 10/130 YDF in Fig. 2a with the same kind of YDF but of 2.6-meter-length and removing the 20/130 YDF therein. The other devices and the fiber-coil configuration in Setup A were still used in Setup B. The performances of setup B were thereby characterized and also shown in Fig. 4, 5 and 6. After that, setup C was built, of which the scheme is shown in Fig. 2c, by replacing the 10/130 YDF in Fig. 2b with 2.6-meter-length 20/130 YDF that was of the same kind as the 20/130 YDF in Setup A. Meanwhile, the FBGs were also replaced by those inscribed in 20/130 matched passive fibers. The performances of setup C were shown accordingly in Fig. 4, 5 and 6.

 

Fig. 4. Experimental output signal powers versus launched pump powers and their optical-optical efficiencies obtained from the three setups. The k values are slope efficiencies versus launched pump powers at high-signal-power regions (made to compare [15]).

Download Full Size | PDF

 

Fig. 5. Experimental optical spectra obtained respectively from the three setups. In each setup, the spectra were characterized at five different output signal powers obtained under pumping current 2, 4, 6, 8 and 10 A respectively. Insets show the linewidths of the spectra peak around 1018 nm varying with the output signal powers.

Download Full Size | PDF

 

Fig. 6. Experimental beam qualities obtained respectively from the three setups at their respective highest output powers. The hyperbolic fittings are determined by least square linear fittings of the squares of the radii of beam spots and those of the positions of Z-planes. The relative positions of Z-planes are zeroed at the minima of the hyperbolic fitting curves. Insets show the beam profiles of the smallest radii characterized in all Z-planes in the respective setups.

Download Full Size | PDF

3. Results and discussion

As is shown in Fig. 4, Setup A outputted 300.1 W with an optical-optical efficiency of 79.3% under the largest pump condition 378.5 W. Both the signal power and the efficiency were higher than those of Setup B, which were 240.3 W and 63.5%, but slightly lower than those of Setup C, which were 317.6 W and 83.9%. The slope efficiency of Setup A reached 86.3% at its high-output-power region, being higher than Setup B (70.7%) but slightly lower than Setup C (91.1%). As is shown in Fig. 5, pump spectra around 976 nm appear in Setup B but are absent in Setup A and C. The estimated pump losses of Setup A, B and C, using the nominal parameters of the YDFs and the CLS, can be respectively 43.4 dB, 33 dB and 53.8 dB for reference. We note that these estimated values should be relative to the original intensities of the pump light, which was not shown in the figure, instead of the shown intensities of signal light. Nevertheless, it was still sensed that Setup B could produce a little more residual pump light than such estimated values. We attribute this deviation mainly to the CLS in Setup B that was working in different circumstances from it in Setup A and C. As discussed before, all the Setups used the same CLS that was made on 20/130 fiber. In setup B, the CLS was input with pump light from structurally discontinuous 10/130 fiber, while it in Setup A and C continuous 20/130 fiber. More pump light entering the core of the CLS in the case of Setup B could have impaired its effect of pump stripping. Moreover, Setup B, while using the same FBGs as Setup A, produced much wider linewidths at 1018 nm in high-power regions than Setup A. Since the linewidths were still mainly determined by the configuration of the oscillation in our cases (not reaching thresholds for the known dangerous nonlinear effects), it agreed with that shortening gain fibers inside the oscillating cavity is generally effective for increasing spectral brightness. It is noted that slight signs of ASE appeared between 1030 and 1040 nm in all three Setups. In experiment, Setup B frequently produced spiky ASE peaks at its highest output condition (around 240 W). In contrast, the ASE spectra in both Setup A and C (both over 300 W) were stable without stochastically rising peaks along time, being 45 dB below that of the lasing wavelength. Moreover, it could be seen from the perspective of the integrations of ASE spectra that Setup C presented lower ASE intensity than Setup A did in Fig. 5. These results were actually expected in theory [18]. Using the nominal parameters of the YDFs, it can be estimated that the high-absorption 20/130 YDF had smaller gain of ASE than the high-beam-quality 10/130 YDF did in the same length of use. Therefore, Setup C should have the lowest ASE gain in our experimental conditions; Setup A should be also superior over Setup B in ASE suppression. Since Setup A and C were stable, further power scaling may be realized in both Setups. In previously highest report, rising ASE limited the stable output power under 220 W [14]. Here, the stable output power was 36% higher and with a 4% higher optical-optical efficiency. The slope efficiency (against launched pump power) was also slightly higher than [14], or higher than the newest [15] (75%). These suggest that Setup A fully explored the potential of the gain fibers. Furthermore, the hybrid-structure setup could offer higher brightness. Albeit no specific information of spectral linewidths was discussed in the previous studies, it can be seen from figures that the lasing spectrum of Setup A, being narrower than 200 pm at the highest output, should be narrower than the previous studies (i.e., Fig. 7 in [14] and Fig. 4 in [15]), both of which belong to the type of Setup B (over 800 pm) in both theory and result.

So far, Setup C adopting double-cladding YDF of thicker core had the highest output signal power and efficiency with favored ASE suppression. Nevertheless, Setup A also presented comparable performances in all those aspects. In this case, however, the advantage of Setup A fell on beam quality. As can be seen from Fig. 6, Setup A and B produced similar beam qualities, measuring1.255 and 1.210 M²-factors respectively at their highest output powers. Both the values reached high-level performance of single-mode fiber lasers like [15]. Contrarily, the beam quality produced by Setup C degraded to 1.534 M²-factor at its highest output power. It is noted that all the three Setups used the same one CLS on 20/130 passive fiber. This usage of CLS in Setup B itself also proved that using multimode fiber to transmit single-mode laser alone would not significantly degrade beam quality. Moreover, all the three Setups used the same fiber-coil configuration, which suggested that the good beam quality of Setup A did not result from higher-mode suppression of the fiber-coil. That is, the differences in beam quality resulted from the nature of the gain structures of the Setups. The Setup C, having multimode gain structures comprising of the 20/130 YDF, the 20/130 HR and the 20/130 OC, produced the worst beam qualities. Moreover, considering that both Setup A and Setup B had one fusion point between their 10/130 fiber and their 20/130 fiber (for Setup A, the 20/130 high-absorption YDF, and for Setup B, the 20/130 CLS), the differences between their beam qualities suggest that the gain fiber would easily amplify beam-quality degradations. It is noted that some studies, including [14], used another algorithm to calculate M²-factor, which would produce different values. Using that algorithm, the beam qualities of Setup A, B and C respectively measured 1.19, 1.12 and 1.42. That is, the control experiment Setup B also pushed the record of traditional oscillator setups in both output power and beam quality. Even in this challenging comparison, the hybrid structure Setup A still sustained uncompromised beam quality while producing much higher power. The result consists with the design purpose of the hybrid-structure. It rigorously shows that the hybrid-structure can produce high-beam-quality signal, despite the signal is amplified by multimode YDFs. This attribute is hard to be realized in traditional setups where fiber-coil with very small radii will cause safety problems and thus not be permitted.

4. Conclusion

In conclusion, this study developed a new trial towards powerful 1018-nm fiber lasers with high beam quality. The hybrid-structure setup was detailly demonstrated, which combined the advantages of high-beam-quality gain fibers and those of high-absorption ones. Using this setup, we detailly presented a high-power high-efficiency 1018-nm monolithic fiber laser of high beam quality. We also performed control experiments to make thorough comparison of the conceptual differences and experimental performances between this setup and the traditional fiber oscillators. It is noteworthy that the performance of the control experiment also reached the record level of its type to date. In consequence, this hybrid-structure setup registers a significantly higher stable signal power than literature with even higher optical-optical efficiency and single-mode beam quality at 1018 nm. Moreover, the stable optical spectrum suggests that these performances can be further increased by applying more pump power without being limited by ASE. This study reveals huge advantages of the hybrid-structure setup. It can be incorporated in almost all fiber laser systems to provide much higher seeding power while offering the same brightness and beam quality. Particularly, it offers unique advantages for many short-wavelength fiber lasers wherein power scaling was limited by strong ASE (thus amplifications by MOPA was unfavored). Therefore, it has very wide range of application in many fields using laser light beyond enumeration. For example, we are applying multiple1018-nm fiber lasers of this hybrid-structure as tandem pumping sources for fiber laser of higher output power with favorable heat management (currently writing). Preliminary results have validated that this setup supports over-5.5-kW amplification without any stability problems. As this study only provides a preliminary demonstration yet to reach the full potential of the hybrid-structure setup, more experiments and theoretical works are to be expected in days.