1. Introduction

Ever since the initial demonstration of Tm³⁺-doped fiber lasers [1], they have been recognized as a convenient and versatile source capable of providing a broad spectrum spanning from the visible to the short-wave infrared region [2]. The recent surge in interest surrounding Tm³⁺-doped fiber lasers operating at a wavelength of 2.3 µm can be attributed to their broad range of potential applications, including optical metrology of combustion processes and non-invasive glucose blood measurement [3,4]. Furthermore, it is worth noting the significant practical applications of 2.3-µm laser in the context of expanding the mid-infrared wavelength range [5].

Various laser structures have been investigated for exciting 2.3-µm Tm³⁺-doped fiber lasers, resulting in the demonstration of different laser efficiencies and power scaling capability potentials. The first Tm³⁺-doped fiber oscillator with a wavelength of 2.3 µm was demonstrated in 1989, yielding an output power of approximately 1 mW [6]. Subsequent advancements in power scaling have been reported for a 2.3-µm laser possessing a similar structure, resulting in a notable improvement of output power from milliwatts to watts [7,8]. The presence of a 2.3-µm laser source with a high level of spectral purity holds significant importance in certain specialized applications, such as greenhouse gas detection and utilization as a pump source for gas lasers [9,10]. The Tm fiber oscillator, which incorporates fiber grating or blazed grating, has successfully achieved a narrow linewidth output at levels of 100 MHz [9,11,12]. Nevertheless, the utilization of long fibers is unfeasible within this particular configuration, thereby restricting the power output to a mere few milliwatts. This limitation poses a significant obstacle to the widespread implementation of 2.3-µm fiber lasers. Therefore, the importance of 2.3-µm fiber amplifier is highlightened. It uses a feature laser source as a seed to obtain higher feature laser outputs. In 2018, Muravyev et al. reported an ultra-broadband amplification covering 1.9- and 2.3-µm bands seeded by a supercontinuum source [13]. A maximum gain of 30 dB and 7 dB was measured at 1.9 µm and 2.3 µm, respectively. Subsequently, Kamynin et al. validated a 2.27-µm picosecond pulse amplifier using Tm³⁺-doped tellurite fiber [14]. Recently, we reported a single-frequency Tm³⁺-doped fiber amplifier at 2.33 µm, and the seed was a distributed feedback (DFB) diode laser [15]. A high gain up to ∼ 24 dB was generated with the output of 246 mW. This completely illustrates the benefits of utilizing the 2.3-µm Tm fiber amplifier. Further power scaling was limited by the competitive ³F₄ → ³H₆ laser transition at approximately 2 µm in such direct pumping system, as shown in Fig. 1. The mentioned phenomenon can be effectively avoided through the use of upconversion pumping at ∼1.05 µm to populate the higher energy level [8]. This approach has shown a significant improvement in efficiency. Regrettably, the upconversion pump necessitates a high level of pump brightness to initiate the photon avalanche effect, i.e., the fiber requires a core pump [16]. Therefore, the challenges associated with heat management and the potential for damage to the fiber tips. Furthermore, the photodarkening in the upconversion pump scheme with 1050 nm remains a potential threat, as it has a detrimental impact on the long-term stability and reduces the operational lifespan of the fiber. Despite extensive research on photon bleaching, the full solution to this issue remains obscure and its long-term effects need to be further investigated [17]. From this point of view, the direct pumping scheme depending on the ground state absorption (GSA) of ³H₆→ ³H₄ has potential for power scaling.

 

Fig. 1. Energy level diagram of Tm: ZBLAN fiber and two pumping schemes [8]. The direct pumping scheme induces a co-lasering phenomenon with a wavelength of 2 µm. The utilization of the up-conversion pump scheme, in conjunction with the process of excited-state absorption (ESA), results in the generation of a blue laser (0.48 µm) and induces the occurrence of photodarkening. NR, non-radiative relaxation; CR, cross-relaxation.

Download Full Size | PDF

It should be noted that none of the above-mentioned work was based on silica fiber, although Tm-doped silica fiber has achieved a kilowatt output in the 2-µm band. The limited transmission region and high phonon energy (1100 cm^-1) of silica fibers make it difficult for laser emission to exceed 2.2 µm. Therefore, it is necessary to select a soft glass fiber with low phonon energy (such as fluoride, tellurite and chalcogenide glasses) to achieve efficient operation of 2.3 µm. Among different soft-glass fibers, the transparent region of ZBLAN (ZrF₄-BaF₂-LaF₃-AlF₃-NaF) fiber with relatively low-phonon energy (550 cm^-1) can reach 4 µm. ZBLAN fiber has been proven to have excellent performance in mid-infrared band. In addition, the wide application of 2.3-µm laser source using commercially available Tm: ZBLAN fiber is of great significance.

In this study, we aimed to demonstrate the power-scaling potential of 2.3-µm Tm: ZBLAN fiber amplifiers directly pumped by 793-nm laser diodes (LD). The amplifier consists of two stages of amplification. The scheme produced a 1.41 W of narrow-linewidth output (less than 10 MHz), which is the highest laser output power among the ∼2.3 µm Tm³⁺-doped fiber lasers.

2. Experimental setup

The experimental setup employed a double-clad Tm: ZBLAN fiber with core and clad diameters measuring 7.5 µm (NA = 0.14) and 120 µm (NA = 0.5), respectively. The cutoff wavelength was approximately 1.9 µm. To minimize the amount of back-reflected light, the fiber was cleaved at an angle of ∼ 9° at both ends. The pump power was provided by a commercial 793-nm LD with a fiber pigtail of 105-µm core diameter (NA = 0.22). The seed laser utilized in the experiment was a DFB diode laser with a power output of 5 mW and a linewidth of less than 2 MHz. It was equipped with a fiber pigtail that had a core diameter of 7 µm and a NA of 0.2. The operating wavelength of the seed laser was measured to be 2331.9 nm.

A schematic diagram of direct-pumped Tm fiber lasers for forward and backward amplification operations is shown in Fig. 2. The pump light was coupled into the fiber by means of two uncoated lenses with a focal length of 13.5 mm (L1 and L2) and a 45° inclined dichroic mirror (DM1) that exhibited high transmission at 793 nm and high reflectivity within the range of 1.8 ∼ 2.4 µm. The DM1 was employed for the purpose of reflecting lasers with wavelengths of ∼ 2 and 2.3 µm. The estimated efficiency of the pump coupling was determined to be 70%. The pumping end of fiber was mounted in a water-cooled sink at 18°C and high refractive index glue was covered on the opposite end to prevent thermal damage and filter the residual pump light, respectively.

 

Fig. 2. Experimental setup of the Tm: ZBLAN fiber amplifier. (a) Forward amplification structure; (b) backward amplification structure; (c) two-stage amplification structure. “+” represents co-propagation, and “-” represents counter-propagation.

Download Full Size | PDF

In the depicted configuration of the forward amplification structure, as shown in Fig. 2(a), the seed laser was effectively coupled into the fiber through the utilization of L2, L3, and DM1. The focal length of L3, which has been coated with an anti-reflection coating at a wavelength of 2.3 µm, was measured to be 13.5 mm. A DM2 with high transmission at 2.3 µm and high reflectivity at ∼2 µm was intentionally tilted at an angle of 8° between L2 and DM1. This tilt was implemented with the purpose of isolating the backward laser within the 2-µm wavelength range. In the backward amplification structure depicted in Fig. 2(b), the seed laser was introduced from the opposing end. The lenses employed were identical to those utilized in the forward amplification structure, with the exception of the exclusion of DM1. The power of the signal that was coupled into the core of the gain fiber was ∼ 2 mW. The extraction of laser output at a wavelength of 2.3 µm from the 2-µm output was achieved through the utilization of the DM2 in both amplification structures.

In the experiment, two methods were used to ensure the coupling efficiency with different structures. Firstly, before coupling the laser into the Tm: ZBLAN fiber, a passive fiber with the same parameters as the gain fiber was used for coupling testing, and the experimental setup was reconstructed multiple times to ensure high coupling efficiency and repeatability. Secondly, a silica fiber was fused with Tm: ZBLAN fiber and the fused loss can be measured accurately. The parameters of Tm: ZBLAN fiber used for welding were identical to those used in the investigation of 2.3-µm amplifiers. By comparing the power with the fused fiber, the coupling efficiency can be inferred and kept consistent.

3. Experimental result and discussion

To ascertain the impact of doping concentration on the performance of laser output, a comparative investigation was conducted on two types of gain fibers with varying levels of doping concentration. One had a concentration of 2 mol. %, while the other had a concentration of 3 mol. %. The peak cladding absorption efficiencies of the two fibers at 793 nm were measured by the manufacturer to be 1 dB/m and 1.5 dB/m, respectively. In order to maintain consistent pumping absorption, the lengths of the two fibers were 6 m and 4 m, respectively. The optimization of fiber length has not been studied in this work due to the lack of longer lengths of fiber. The 2.3-µm output power of the two fibers as a function of pump power with forward amplification structure is shown in Fig. 3(a). The power evolution curve clearly demonstrates a small signal amplification, with the low-doped fiber exhibiting significantly greater amplification power compared to the high-doped fiber. In the case of a 2 mol. % fiber, the maximum output power of 101 mW was achieved at the launched pump power of approximately 10 W. The maximum output power achieved under the same pump power for a fiber concentration of 3 mol. % was only 21 mW. Similar to previous Tm fiber laser configurations operating at 2.3 µm and utilizing a pump source of ∼ 0.79 µm, this scheme also exhibits strong co-lasing behavior at 2-µm band. The combined output power at 2 µm (including both co- and counter-propagation outputs) for the two fibers is depicted in Fig. 3(b). It is evident that the highly doped fiber exhibits a higher level of efficiency in terms of 2-µm output, despite its lower output at 2.3 µm. The fiber with low concentration exhibited an output efficiency of 31.4% and a maximum power of 2.9 W at 2-µm band. In contrast, the other fiber demonstrated an increased output efficiency of 36.0% and a maximum power of 3.4 W within the same band. The observed phenomenon can be attributed to the CR process (³H₄ + ³H₆→³F₄ + ³F₄), as shown in Fig. 1. This process is influenced by the concentration of Tm³⁺ doping. The CR rate of the 3 mol. % fiber is 2.25 times that of the 2 mol. % fiber, which will significantly reduce the 2.3-µm laser efficiency [15]. The efficient CR process populates the upper energy level of 2-µm transition at the expense of the population densities of 2.3-µm energy level, which makes the laser transition between 2-µm and 2.3-µm competitive to some extent. A decrease in concentration results in a diminished effectiveness of the CR process [15,18]. Moreover, the increase of doping concentration also results in the concentration quenching of the upper level of 2.3 µm. Excessive concentration causes the signal light to be unable to be amplified [19]. Hence, a lower doping concentration proves advantageous for 2.3-µm lasers, while concurrently diminishing the output within the 2-µm band.

 

Fig. 3. The effect of doping concentration on the laser output performance. (a) Output power in 2.3 µm; (b) total output power in the 2-µm band.

Download Full Size | PDF

Subsequently, an investigation was conducted on the disparities between forward and backward amplification structures in low-doped fiber. As shown in Fig. 4(a), the power at 2.3 µm in the backward-pumped structure is marginally greater than that in the forward-pumped structure. This observation aligns with the behavior observed in fiber amplifiers operating in different wavebands, such as the Tm fiber amplifier operating at the 2-µm band. In the context of forward-pumped configuration, the laser operating at 2.3 µm exhibited a power output of 393 mW when subjected to the highest pump power of 16.4 W. The backward-pumped structure yielded a higher output power of 463 mW at 2.3 µm, while maintaining the same pump power. Nevertheless, when considering the 2-µm laser transition, the alteration of the amplification structure did not yield a notable disparity between the co-propagating and counter-propagating output, as shown in Fig. 4(b). The highest power output achieved for the 2-µm was approximately 2.2 W for co-propagating, and approximately 2.7 W for counter-propagating. The corresponding slope efficiencies were approximately 14% and 17% respectively. This observation suggests that when the injected power of the 2.3-µm seed is low, the 2.3-µm Tm fiber amplifier, which is pumped in the ground state, can be considered as an amplifier that operates without the injection of a seed [15].

 

Fig. 4. Output power evolution curve with forward and backward pumping structures. (a) 2.3-µm output; (b) 2-µm output.

Download Full Size | PDF

In order to enhance laser output in the 2.3-µm band, the second-stage backward-pumped amplifier was investigated. The experimental configuration is depicted in Fig. 2(c). The first-stage amplifier adopted backward pumping structure to obtain high seed power. The seed laser with a wavelength of 2.3 µm was coupled into the gain fiber by means of DM2 and L2, after being emitted in a collimated manner from the first stage amplifier. The coupling efficiency of core was about 36.6%, i.e., approximately 169 mW. The gain fiber utilized in the second stage amplifier was a ZBLAN fiber measured 6 m in length, with a doping concentration of Tm³⁺ at 2 mol. %.

Figure 5(a) illustrates the evolution of the output power and gain at 2.3 µm in the context of the backward amplification scheme. The laser reached the maximum output power of 1.41 W for the launched pump power of approximately 19.1 W, and the corresponding gain was 9.2 dB. Compared to the first stage amplification, an order of magnitude increase in seed power caused the curve deviate from the small signal amplification trend observed in the first stage amplifier. The linear relationship between the 2.3-µm power and the incident pump power, with a slope efficiency of 8.1%, is evident in Fig. 5(a) for pump power exceeding ∼ 7 W. However, the co-lasing of 2-µm band in the system, which was decreased but still significant, suggests there is no scope for further power scaling by merely increasing the pump power. Figure 5(b) shows the 2-µm output power versus varied pump power. The total power of the 2-µm band was measured to be 5.1 W under the highest pump power conditions, while the slope efficiency was found to be 27.7%. The slope efficiency for co-propagation was 12.2%, which was lower than 15.5% for counter-propagation, and the highest power was 2.2 W and 2.9 W respectively.

 

Fig. 5. Performance of the second stage fiber amplifier. (a) Gain (left y axis, square data points) and output power (right y axis, round data points) at 2.3 µm; (b) output power at 2 µm; (c) output spectra at 2.3 µm. Inset shows an enlarged view of the frame; (d) Fabry-Perot (F-P) spectrum of the 2.3-µm laser output.

Download Full Size | PDF

The amplified spectrum after DM2 for a maximum power were recorded by an optical spectrum analyzer (OSA), as shown in Fig. 5(c). A slight 2.3-µm band amplified spontaneous emission (ASE) was observed with a resolution of 0.5 nm. The central wavelength was approximately 2300 nm, which coincides with the center of Tm: ZBLAN emission spectrum at 2.3-µm band [8]. This indicates that the reabsorption of Tm: ZBLAN fiber at the 2.3-µm band is not obvious owing to the quasi-four-level nature of the ³H₄→ ³H₅ transition. The ASE was not discernible upon augmenting the resolution to 0.05 nm. Therefore, the absence of 2.3-µm isolator will not affect the overall system. The measured linewidth was approximately 0.06 nm, which was constrained by the resolution capabilities of the OSA. The spectral characteristics were measured in greater detail by using a scanning F-P interferometer with a minimum resolution of 7.5 MHz, as shown in Fig. 5(d). The output linewidth of 2.3-µm Tm fiber amplifier was approximately 10 MHz at each power level (including the DFB laser for direct measurements), which indicates that there was no obvious broadening phenomenon during the amplification process. In order to obtain more comprehensive linewidth information, it is necessary to employ alternative high-resolution measurement techniques, such as heterodyne or delayed self-heterodyne detection.

For the purpose of exploring further power scaling potential of the 2.3-µm Tm fiber amplifier directly pumped by 793-nm LD, the effect of seed power on the output of 2-µm band was investigated, because increasing seed power is usually an effective method to suppress ASE. Figure 6(a) illustrates the total output power within the 2-µm range as a function of the injected seed power, ranging from 10 to 169 mW, while maintaining a constant pump power of 19.1 W. The curve demonstrates a noticeable decline in 2-µm output power as the seed power increases. Additionally, the slope efficiency is calculated to be -226.1%, indicating that for every 1 W increase in seed power, the 2-µm power experiences a reduction of approximately 2.3 W. From this point of view, seeded by up-conversion pumped 2.3-µm Tm fiber laser or 2.3-µm solid-state laser will be easier to achieve power scaling for the scheme can directly obtain a higher 2.3-µm output without the influence of co-lasing at 2 µm [8,20]. In addition, it also shows that there is a competitive relationship between the 2-µm transition and the 2.3-µm transition rather than mutual promotion [21]. We attribute this competitive relationship to the CR process. As shown in Fig. 1, the ³H₄ level will be rapidly depopulated when the seeded 2.3-µm power is increased. This phenomenon results in a decrease in the efficiency of the CR process, leading to a decline in population densities at the ³F₄ level.

 

Fig. 6. Dependence of the output power on the 2.3-µm seed power. (a) Effect on 2-µm output; (b) stability of 2.3-µm laser power.

Download Full Size | PDF

The characterization of power stability is crucial in applications such as gas detection, where a reliable detection source is imperative. The impact of the competitive laser transition occurring at ∼ 2 µm on stability has received limited attention in previous investigations of direct pumped 2.3-µm laser systems. Figure 6(b) shows the 2.3-µm power stability at different seed power over a time span of 100 s, with measurements taken at intervals of 300 ms. The pump power was constant at 15 W. The investigation of stability at extended time scales and higher power levels has been deliberately avoided in order to prevent any potential damage to the fiber. As shown in Fig. 6(b), the lack of a substantial impact on the power stability of the system is evident. The normalized root-mean-square (RMS) deviation of the 2.3-µm power with different conditions remain at the level of approximately 2%. The primary cause of the power fluctuation can be attributed to the fluctuating coupling efficiency between the fibers in the space optical path. In order to mitigate this impact, future investigations necessitate the utilization of an all-fiber structure. Additionally, the exploration of the stability implications of the 2-µm transition necessitates the utilization of higher power seed sources.

The emission spectral characteristics with varying seed power were also measured. Figure 7(a) illustrates the spectrum of counter-propagation, while maintaining a constant pump power of 15 W. Comparable spectra were detected within the 2-µm wavelength range at varying seed levels, with both ASE and parasitic lasing exhibiting peak wavelengths near 1920 nm. The broad spectrum has a range of approximately 100 nm with a 20-dB bandwidth from ∼ 1870 to 1970 nm, and the obvious water vapor absorption can be observed. The spectral characteristics of co-propagation demonstrate a comparable pattern, with the exception of a limited spectral range that arises due to the re-absorption of the ground-state in the shorter wavelength region [22].

 

Fig. 7. Dependence of the output performance on the 2.3-µm seed power. (a) Output spectra at different seed power; (b) the relationship between parasitic lasing threshold and seed power.

Download Full Size | PDF

The probability of fiber damage is notably increased by the presence of parasitic lasing associated with intense giant pulses, whereas the occurrence of ASE does not have a similar effect (e.g., fiber-based superfluorescence source with output power of kilowatts has been achieved in silica fiber source [23]). In order to examine the impact of 2.3-µm seed injection on the threshold of parasitic lasing, an investigation was conducted, as shown in Fig. 7(b). The threshold power was measured by the pump power when the spectral peak appeared. The threshold of parasitic lasing experienced a slight increase as the seed power was elevated. The observed slight change can be attributed to a decrease in the gain of the 2-µm transition, resulting from the competitive interactions between transitions occurring at different energy levels. Further research on parasitic lasing needs to be judged through the temporal behaviors of the signal.

4. Conclusions

In summary, we demonstrated the first 2.3-µm Tm³⁺-doped fiber laser with continuous output power over 1 W. A maximum output power of 1.41 W was achieved using a two-stage amplification structure, and relying on the direct pump by 793-nm LDs. The output linewidth < 10 MHz (limited by the minimum resolution of the F-P interferometer) indicates that the system is promising for applications in field of atmospheric detection and gas lasers. The study investigated the impact of 2.3-µm seeds on the output properties, including concurrent competitive ³F₄ → ³H₆ laser transition, power stability and spectra. The results indicate that the augmentation of input signal power exerts a notable suppressive impact on the 2-µm transition, while it does not yield a significant influence on other aspects.