1. Introduction

Over the last 20 years, there has been remarkable progress in power scaling of ytterbium (Yb)-doped fiber (YDF) lasers due to the development of high-power high-brightness laser diodes (LDs) and high quality YDFs, now yielding several kilowatts of a diffraction-limited continuous-wave (CW) laser output [1]. However, further power scaling is suffering from limited brightness of LDs, thermally induced problems [2], and nonlinear optical effects [3]. A number of techniques have been developed so far, but most of them should compromise with these limits for optimum performance due to their contradictory requirements on fiber design and laser performance. For example, increasing the fiber core size can reduce the onset of nonlinear optical phenomena, but at the expense of exacerbated thermal problems and beam quality degradation. Tandem pumping was proposed to overcome these limits [4–6]. In the tandem-pumped laser configuration, we can couple a very high pump power to a conventional double-clad fiber due to the high brightness of fiber lasers [7]. A low quantum defect in the tandem-pumped laser configuration (for example, ∼5% for Yb fiber lasers tandem-pumped at ∼1020 nm) significantly alleviates thermal problems, offering a much higher output power without thermal degradation [6–9]. Moreover, a low gain of a tandem-pumped fiber amplifier enables generation of the laser output with a better beam quality and less self-pulsing [10]. Owing to these advantages, the tandem-pumped Yb fiber laser has already demonstrated tens of kilowatts of output in a beam with diffraction-limited beam quality [6].

In addition to these advantages, there have been considerable interests on PD of tandem–pumped Yb fibers [11,12]. It is widely believed that PD is caused by the multi-pump-photon absorption (MPPA) process followed by the formation of color center defects [13–20]. Without doubt, many researchers have intensively studied the type of defects and the mechanism of color center formation in the YDF. It is already known that color center generation is mainly promoted by excitation of Yb³⁺ ions, composition of core doping ions, and fiber preparation conditions [14,15,19–21]. H. Li et al. reported the pump wavelength dependence of PD on Yb-doped silica fibers, showing that the 976 nm pumping leads to a more significant PD than the 916 nm pumping [14]. S. Jetschke et al. [20,21] proposed that PD could be caused by Tm³⁺ impurities and the associated excited state absorption. In order to describe the PD kinetics, the time-dependent PD loss $\alpha (t )$ can be fitted by the following stretched exponential function,

In this paper, we report experimental investigation of PD in the tandem-pumped YDF. The PD-induced loss in the tandem-pumped YDF was compared with the LD-pumped YDF by monitoring the transmitted probe beam at 633 nm and the amplified laser output over time, proving a dramatic reduction of the PD-induced loss in the tandem-pumped YDF. Furthermore, we present fluorescence spectra of the tandem-pumped YDF to investigate its relation with the PD mechanism and confirm that PD is directly related to the density of excited Yb³⁺ ions.

2. Experiments and results

In order to investigate the PD-induced loss in the YDF, we measured a transmitted power of a single-mode LD probe beam at 633 nm while pumping the YDF, as shown in Fig. 1. Since light at 633 nm has a ∼70 times higher sensitivity to the background loss than ∼1 mm [29,30], it can be used as a probe signal for monitoring the PD-induced loss in the YDF. The 633 nm probe signal with ∼600 µW in power was coupled to a wavelength division multiplexer (WDM) in conjunction with a pump light. The pump source used in the experiment was a single-mode fiber-coupled LD at 976 nm or a Yb fiber laser at 1018 nm. The latter Yb fiber pump laser at 1018 nm was constructed in-house. The YDF as a gain material had a core diameter of 10 µm (NA 0.08) and a cladding diameter of 125 µm (LMA-YDF-10/125-9M, Nufern Inc.). We used a rather short length of fiber, ∼1.2 m, for robust laser operation at 1018 nm. Feedback for lasing was provided by a pair of fiber Bragg gratings (FBGs), i.e. high reflectance (HR, R∼99%) and low reflectance (LR, R∼25%) FBGs at 1018 nm. Both FBGs were spliced to the opposite ends of the YDF. As a result, the fiber laser yielded 8.5 W of output at 1018 nm for the absorbed pump power of 10.7 W, corresponding to a slope efficiency of 79.5%.

 

Fig. 1. Schematic diagram for the pump-probe transmittance measurement of the YDF. The probe source is a single-mode diode laser at 633 nm. DM: dichroic mirror.

Download Full Size | PDF

The fiber used for the PD measurement was a Yb-doped aluminosilicate double-cladding fiber (YADF), fabricated by the vapor axial deposition technique (Taihan Fiberoptics Inc.). The core of the fiber was co-doped with Yb and aluminium (Al) by the solution doping technique and was measured to have the concentrations of 0.15 mol% Yb and 1.09 mol% Al by an electron probe X-ray microprobe analyzer. The YADF had a core with a 20 µm diameter and a 0.065 NA, surrounded by an octagonal pure silica inner-cladding of a 400 µm diameter. The inner-cladding was coated with a low index polymer outer-cladding (PC-373, Luvantix), giving a calculated NA of >0.45 for the inner-cladding pump guide. We applied ∼440 mW of the pump power for both pumping wavelengths and the lengths of the YADF were experimentally selected to be ∼38 mm and ∼195 mm for 976 nm and 1018 nm, respectively, leading to the same absorbed pump power of ∼76 mW. The YADF was connected to the output port of the WDM via a mode field adapter (MFA). Under this experimental condition, the YADF had the uniform population inversion densities along the fiber at both pumping wavelengths of 976 nm and 1018 nm, which were calculated to be 48% and 10%, respectively, by RP Fiber Power (RP Photonics Consulting GmbH). In order to remove the residual pump light in the transmitted output beam, we placed a dichroic mirror with high reflection at a pump wavelength (∼ 1 µm) and high transmission at 633 nm, followed by an additional band-pass filter at 633 nm before the Si photodetector. For comparison, we replaced the YADF with the commercial Yb-doped double-cladding fiber (CYDF) with a 20 µm-diameter core and a 400 µm-diameter inner-cladding (LMA-YDF-20/400-VIII, Nufern Inc.). The Yb doping concentration of the CYDF (informed by Nufern) was 10∼15% lower than that of the YADF. The absorption coefficient of the CYDF was ∼1.2 dB at 976 nm, so the lengths of the CYDF we used were ∼27 mm and ∼177 mm for pumping wavelengths of 976 nm and 1018 nm, respectively, having the same absorbed pump power of ∼76 mW. While doing the experiment, we monitored the spectrum of the transmitted beam after the fiber with the aid of the optical spectrum analyzer and confirmed a negligible amount of amplified spontaneous emission without lasing.

Figure 2 shows transmitted powers of the 633 nm probe beams as a function of time for the YADF and the CYDF while pumping at 976 nm or 1018 nm. For the YADF, the transmitted signal powers were reduced by 33.3% and 2.4% for 2 hours at the pump wavelengths of 976 nm and 1018 nm, respectively. A similar result was obtained when we replaced the YADF with the CYDF, showing a 18.5% reduction in transmitted power for 2 hours while pumping at 976 nm, but only a 1.9% reduction at 1018 nm. As expected, the decrease of the transmitted signal power over time can be attributed to the PD-induced propagation loss, so we can conclude that the PD-induced loss was dramatically reduced in the tandem pumping configuration. This result is in good agreement with the previous work [20], which reported a linear relationship of the PD loss to the density of the excited Yb³⁺ ions. We tried to fit the PD loss by the stretched exponential function, Eq. (1), to figure out PD parameters, but found that the measurement time was not long enough. A more detailed work including the fitting analysis of the PD loss is the subject of ongoing research.

 

Fig. 2. Transmitted probe signal powers at 633 nm as a function of time for the YADF and the CYDF pumped at 976 nm and 1018 nm.

Download Full Size | PDF

Further confirmation was obtained by the Yb fiber laser amplifier system pumped at 976 nm or 1018 nm. We constructed an all-fiberized Yb fiber master oscillator power amplifier (MOPA), as shown in Fig. 3. The seed laser comprised a ∼3.5 m Yb-doped single mode fiber (Nufern LMA-YDF-10/125-9M) with a 10 µm diameter core and a 125 µm diameter cladding, a pair of HR (R∼99%) and LR (R∼10%) FBGs centered at 1080 nm, a (2 + 1) by 1 pump-signal combiner, and a wavelength-stabilized pump LD at 976 nm. The Yb fiber seed laser yielded 8.0 W of output at 1080 nm for the absorbed pump power of 10.9 W. The power amplifier employed the YADF of ∼1.4 m in length as a gain medium, which had the cladding absorption efficiency of ∼43% at 976 nm. The output end of the fiber was angle-cleaved at ∼12 degrees to suppress parasitic lasing and broadband feedback. Pump light at 976 nm was applied to the YADF with the aid of a (6 + 1) by 1 pump-signal combiner. In order to remove the residual pump power in the fiber output, we placed a dichroic mirror with high transmission at pump wavelengths (900 ∼ 1030 nm) and high reflection at the lasing wavelength (∼1080 nm).

 

Fig. 3. Schematic diagram of the Yb fiber MOPA. The YADF with high PD was used in the amplifier stage.

Download Full Size | PDF

Under this configuration, the YADF MOPA reached 34.5 W of output at 1080 nm for the absorbed pump power of 43.8 W at 976 nm, corresponding to a slope efficiency of 77.6%. However, the laser output power began to drop immediately due to strong PD of the YADF. As shown in Fig. 4, the output power decreased from 34.5 W to 32.6 W, which was a 6.1% reduction over 8 hours. For the tandem pumping configuration at 1018 nm, we used a ∼16 m long YADF for the amplifier, which had the similar absorption efficiency with the LD pumping at 976 nm. As a result, the MOPA yielded 34.2 W of output for the absorbed pump power of 43.0 W at 1018 nm, corresponding to a slope efficiency of 79.9%. The tandem-pumped Yb fiber MOPA also showed reduction in the output power over time, but it was only 1.3% from 34.2 W to 33.8 W over 8 hours, which was much smaller than the LD-pumped MOPA. These experimental results proved that the PD-induced loss was remarkably reduced in the tandem-pumped Yb fiber amplifier. In the amplifier, the maximum population inversion densities along the cladding-pumped YADF were calculated to be 14.5% and 1.87% for pumping wavelengths of 976 nm and 1018 nm, respectively, by RP Fiber Power. Thus, these MOPA results are consistent with the previous report [20], claiming that PD was proportional to the Yb³⁺ population inversion density. In our previous experiment, the 633 nm signal beam allowed us to measure directly the PD-induced background loss along the fiber since the sensitivity of the propagation loss at 633 nm was ∼70 times higher than ∼1 µm. In addition, this MOPA experiment confirmed the direct impact of PD on the lasing performance. Therefore, both experimental results, i.e. transmittance of the 633 nm probe signal and the laser amplifier using the YADF, have clearly verified that the tandem pumping configuration can significantly alleviate PD in Yb silica fiber, compared to the conventional LD pumping.

 

Fig. 4. Output powers of the YADF MOPA as a function of time at the pump wavelengths of 976 nm and 1018 nm.

Download Full Size | PDF

In addition to investigation on the PD-induced loss, we measured fluorescence spectra of the YADF while pumping at 976 nm and 1018 nm (Fig. 3). For the LD pumping at 976 nm, we could observe strong blue fluorescence, which was measured to be centered at ∼470 nm by the spectrometer (STS-VIS, Ocean Insight Inc.), as seen in Fig. 5. For comparison, we also measured the fluorescence spectrum of the CYDF with negligible PD, showing greenish fluorescence with the broad spectrum from ∼480 nm to ∼550 nm (the inset picture and the green-dot line in Fig. 5(b)). This surprising difference in fluorescence suggests that blue fluorescence is closely related to strong PD of the YADF. There are two explanations for the blue fluorescence of Yb fibers at ∼470 nm. One is the color center formation due to intrinsic ODC II defects [22–26] and the other is the Tm-assisted color center formation due to Tm³⁺ impurities [31,32]. Both mechanisms are known to be responsible for the blue fluorescence of the Al/Yb-codoped silica fiber and, hence, its PD. In our experiments, we thought that PD of the YADF was more likely to be attributed to ODC II since we could not find any trace of Tm³⁺ impurities by the x-ray measurement and chemical analysis. More detailed investigation is required in order to clarify the cause of the blue fluorescence.

 

Fig. 5. (a) Blue fluorescence of the YADF for LD pumping at 976 nm, and (b) its measured spectrum (blue solid line). The inset picture and the green dashed line are the fluorescence picture and the spectrum of the CYDF pumped at 976 nm.

Download Full Size | PDF

The fluorescence behavior for the tandem-pumped YADF is shown in Fig. 6. The bottom part of the fiber, close to the fiber end where the pump beam was incident, showed strong blue fluorescence, like the LD pumping at 976 nm. However, surprisingly, this blue fluorescence turned to the green at a certain part of the fiber and moved to the output end as the pump power increased, as clearly seen in Fig. 6. To the best of our knowledge, this is the first experimental observation showing both blue and green fluorescence simultaneously along the Yb fiber. Since strong blue fluorescence was the manifestation of PD, we could realize that PD was stronger at the beginning part of the fiber and was exacerbated at a higher pump power. The color changing parts in the YADF were ∼5.1 m, ∼9.8 m, and ∼12.1 m for absorbed pump powers of 12 W, 24 W, and 30 W, respectively. Considering that the number of excited Yb³⁺ ions is proportional to the pump power, we calculated its density along the fiber using RP Fiber Power. The calculated results in Fig. 6(b) showed that the fluorescent color of the YADF was changed near the positions where the ratio of the excited Yb³⁺ ion density was over ∼1.1% for all three pump powers. That is, the blue fluorescence of the YADF, i.e. PD, became noticeable when the Yb³⁺ excitation density was higher than ∼1.1% of total Yb³⁺ ions regardless of the pump power. Our results confirm that PD in the Yb silica fiber is proportional to the Yb³⁺ excitation density [20] and becomes prominent when the density exceeds a certain value, which should be dependent on the fiber compositions, defects, and so on. However, the responsible mechanism for the fluorescence color change is not clear. S. Jetschke et al. [21,31] proposed the model that Tm³⁺ impurities could accelerate a Tm-assisted excited state absorption process, which became stronger in the longer pump wavelength. H. Li et al. reported that the dominant PD mechanism was determined by the pump wavelength, ODC II for 976 nm and CT for 915 nm [14]. Among these mechanisms, none of them can clearly explain the fluorescence color change, blue to green, along the fiber. A more detailed investigation, e.g. fitting the PD loss by Eq. (1), is required to clarify this phenomenon. Therefore, it should be highlighted that the tandem pumping at 1018 nm can dramatically mitigate PD of the Yb silica fiber due to a very low density of excited Yb³⁺ ions, preventing generation of ODC II-induced color centers by the MPPA process.

 

Fig. 6. (a) Fiber fluorescence pictures of the YADF amplifier, and (b) numerical calculation of excited Yb³⁺ ions density ratios as a function of the fiber position for 12 W, 24 W, and 30 W of absorbed pump powers at 1018 nm. The vertical green lines are the color changing positions from blue to green.

Download Full Size | PDF

3. Conclusion

In summary, we have studied PD of the YDF tandem-pumped at 1018 nm. Both the pump-probe transmittance and the laser amplifier using YADFs demonstrated that the tandem pumping at 1018 nm could dramatically reduce the PD-induced background loss compared to the LD pumping at 976 nm. We also investigated fluorescence spectra of the YADF at both pumping wavelengths, 976 nm and 1018 nm, showing that PD was closely related to strong blue fluorescence at ∼470 nm. It is seen that the strong blue fluorescence can be attributed to the existence of ODC II or Tm³⁺ impurities, requiring more detailed investigation to clarify the mechanism. Moreover, it is confirmed that PD is proportional to the density of excited Yb³⁺ ions and can be remarkably mitigated in the tandem pumping configuration at 1018 nm due to the low excited Yb³⁺ density. Therefore, our results not only support the advantages of the tandem-pumped YDF laser system for power scaling and its lifetime due to a significant reduction of PD, but can also present the impacts of ODC II and the Yb³⁺ excitation density on PD of the YDF.