1. Introduction

High-power, high-energy 2-2.3 µm wavelength-tunable ultrashort pulsed fiber laser sources have essential applications in the fields of multiphoton spectroscopy [1], gas detection [2], micro- and nano-fabrication [3], and front-end pump sources for mid-infrared nonlinear lasers [4]. Nowadays, the efficient method to generate the high-power wavelength-tunable source is to perform soliton self-frequency-shifting operation directly in a thulium-doped single-mode fiber amplifier, whose purpose is to boost the power of the Raman soliton by utilizing the broad gain spectrum of the thulium-doped fiber. In 2013, Vladislav et al. used a compact thulium-doped fiber MOPA to generate tunable Raman solitons from 2 µm to 2.2 µm with an energy of 38 nJ and an average power of 3 W [5]. In 2017, Wang et al. used a short-wavelength dispersion-managed mode-locked Tm oscillator as a seed source and utilized the long-wavelength gain of a thulium-doped fiber amplifier to generate femtosecond solitons with high conversion efficiency in a wide tunable range (1.9∼2.36 µm), the energy of the soliton at 2.29 µm was 34 nJ, the maximum average output power of the soliton pulse reached 1.16 W [6]. Recently, Wu et al. obtained the current highest energy as well as the output power of tunable Raman solitons from 1.96 to 2.39 µm in single-mode silica fibers by further increasing the soliton order in the soliton self-frequency shifting process, the maximum soliton pulse energy and power at 2.39 µm was about 65.1 nJ and 3.1 W, respectively [7]. However, limited by the small effective mode field area (MFA) of single-mode fibers, none of these studies achieved a pulse energy exceeding 100 nJ.

Increasing the effective MFA of silica fibers is an effective method to enhance the energy of long-wave Raman solitons [8–13]. For LMA fibers, coupling high-energy ultrashort pulse to undoped LMA passive fibers for nonlinear conversion is currently the dominant method to obtain high-energy Raman solitons. In the wavelength range of 2-2.3 µm, the highest Raman soliton energy currently is 252 nJ by pumping an all-silicon Bragg fiber with an oversized MFA using a CPA system [14], but its tuning range is limited and the output power is only about 100 mW. Similar fibers with very large mode-field area are used in other wavelength bands, such as rod photonic crystal fibers [15], which can generate very large energy Raman solitons. However, their propagation lengths are intrinsically limited to a few tens of centimeters due to the stiffness of the waveguide and the extreme bending sensitivity of the guiding mode. On the contrary, the SiO₂ LMA fibers allow the output pulse to have a good beam quality, which itself possesses a good resistance to mechanical and laser damage. Reference [13] reported that an erbium-doped fiber laser emitting 412 fs, 0.66 MHz, 1 µJ pulses at 1550 nm was coupled to a passive fiber with a core diameter of 40 µm, which ultimately generated a soliton that could be tuned from 1580 nm to 2520 nm, the soliton energy is up to 73 nJ, and the output power is about 48 mW. In another Ref. [16], a commercial 2 µm low repetition frequency (1 MHz) laser source was used to pump a 30/250 µm fiber, the initial Raman soliton could be tuned to 2390 nm, which has a maximum energy of 120 nJ and a Raman soliton power of 120 mW at 2270 nm.

These results show that the Raman soliton with the highest energy greater than 100 nJ can be generated in LMA passive fiber currently, but all the high-energy pump sources adopt the method of reducing the repetition frequency (less than or equal to 1 MHz). That means it is easy to generate strong nonlinearities in the fiber by virtue of the higher peak power at low pump power (<1 W), which makes the output power of the Raman soliton be restricted to the order of a hundred mW. More importantly, reducing repetition frequency avoids a series of thermal effects in the high-power coupling system [17,18], such as high-power feedback light and the thermal lens effect [19,20]. Especially if the accurate coupling can't be achieved, the high peak power feedback light will damage the amplifier's pump laser diode through the combiner's fiber cladding or directly damage some isolation devices [21,22].

In summary, the pulse energy of Raman solitons has been effectively enhanced by increasing the effective MFA of silica fibers. However, the performance of using LMA passive fibers to generate high-power Raman solitons has not been effectively developed due to the harmful thermal effects in the high-power coupling system, which is not conducive to the applications requiring a laser with high-power and high-energy performance, such as long-range infrared countermeasure [23], high power terahertz generation [24], and fast materials processing [25], etc. Therefore, it is necessary to explore the area of LMA passive fibers pumped by high repetition frequency and high-power pulses to achieve tunable Raman solitons of 1.9-2.3 µm with both high-power and high-energy.

In this paper, we have systematically optimized the fundamental-mode matching technique for coupling LMA passive fibers, and theoretically as well as experimentally investigated a method of obtaining high single-mode light coupling efficiency by placing two same aspherical lenses out of focal length. Then the method is used to efficiently couple a high repetition frequency, high-power pump source at 1960nm into LMA fibers with a single-mode light coupling efficiency of >75% for collimated light. Finally, we obtained tunable Raman solitons with pulse energy of >100 nJ and output power of >5 W in both 32 µm and 48 µm core diameter fibers. Compared to previous work with tunable Raman soliton sources based on LMA passive fibers, the power level is improved by more than one order of magnitude with a guaranteed energy greater than 100 nJ.

2. Dual-lens out of focal-length placement method for improving the single-mode light coupling efficiency in LMA passive fibers

High peak power pump sources are typically required to generate high power and high energy tunable Raman solitons in LMA passive fibers [26,27]. These sources can be obtained by using the chirped pulse amplification (CPA) technique with grating pairs at the end for compression. As a result, spatial lens coupling techniques are normally used to couple the compressed laser into the LMA fiber.

A good coupling system is necessary to obtain higher power Raman solitons. To achieve precise fundamental-mode matching, it is necessary to match the profile, position, and orientation of the fundamental-mode of the focused incident beam with the profile, position, and orientation of the fundamental-mode of the LMA fiber responsible for reception. The position and orientation of the excitation beam need to be precisely aligned to match the transverse position of the LP01 mode and should be strictly co-linear with the core axis of the LMA fiber [28].

We use the precise fundamental-mode matching technique to achieve two goals. The first is to achieve a high beam-quality single-mode laser. According to the single-mode excitation technique described in Ref. [29], the mode-power coupling coefficient N∝d⁸/b⁶λ⁴ (λ is the excitation wavelength, d and b are the core diameter and the outer cladding diameter, respectively), so we subsequently choose smaller core or larger outer-cladding size to reduce mode coupling when selecting the receiving LMA fiber. What’s more, it is easier to eliminate the interference between different modes when coupling multimode fibers using ultrashort pulses, so this goal is not primary in our study (we use a femtosecond light source). Our main goal is to ensure that more light from the high peak-power pump source is coupled to the fundamental-mode of the LMA fiber to participate in the nonlinear conversion, and minimize the feedback light into the cladding of the fiber which will reduce the damage to the laser devices as well as the pump source.

In general, the mode profile matching requires not only a pure diffraction-limited input beam but also a careful selection of the optical elements’ parameters used for coupling. The situation with ideal coupling lens is shown in Fig. 1(a), where the emitting and receiving LMA fibers with a diameter of 25/250 µm are both placed in the focal length of the two aspherical lenses. The ratio of the focal length of the front and back lens is perfectly equal to the ratio of mode-field diameters of the emitting and receiving fibers, which would achieve the effect shown in Fig. 1(d). Unfortunately, when forming the coupling system, it is generally more difficult to find two aspherical lenses with suitable focal lengths according to the different mode field diameters of the emitting and receiving fibers. Once the focal length ratio of the two aspherical lenses deviates from the ratio of the fundamental mode field diameters of the two LMA passive fibers, it will result in the degradation of the single-mode light coupling efficiency. As shown in Fig. 1(b), the coupling system of 25/250-48/400 with the same two aspherical lenses will result in the situation shown in Fig. 1(e).

 

Fig. 1. Schematic of coupling (a) a 25/250-25/250 system (both emitting fiber and receiving fiber at focal length), (b) a 25/250-48/400 system (both emitting fiber and receiving fiber at focal length), and (c) a 25/250-48/400 system (adjusting the position of the emitting fiber and receiving fiber) using a pair of identical aspherical lenses; (d) the case where the diameter of the incident spot is equal to the diameter of the mode field of the receiving fiber; (e) the case where the diameter of the incident spot is less than the diameter of the mode field of the receiving fiber. F1: the emitting fiber; F2: the receiving fiber; AL1: the first aspherical lens; AL2: the second aspherical lens.

Download Full Size | PDF

For this, we would like to improve the coupling efficiency of the 25/250-48/400 system by using only two aspherical lenses with the same focal length so that the selection of the focal length of the aspherical lens does not need to be taken into account too much in the coupling of fibers with larger core diameter. The above implies that it is crucial to study the relationship between the positions of two fibers placed in a coupled system. One predictable method is to keep the distance between the two lenses unchanged and change only F1-AF1 and AF2-F2 (F1-AL1 denotes the distance between AL1 and F1, and AL2-F2 denotes the distance between AL2 and F2), as shown in Fig. 1(c).

Next, we use Zemax to determine the position between the two fibers and the lens. The two aspherical lenses (Thorlab-AL2550) used for coupling in our simulations and experiments have a focal length of 50 mm and an NA of 0.23. As shown in Fig. 2, we then calculated the fundamental modes’ MFA of the two receiving fibers (a 32/250 µm fiber and a 48/400 µm fiber) at 1.96 µm are 28.4 µm and 40 µm, respectively. The distance between AL1 and AL2 is set to 1 m (determined by the placement distance after adding the compression grating and other devices in the actual optical path), and the position of the two lenses does not change in the simulation and experiment. Subsequently, the coupling process was simulated using physical optical propagation (POP) on 48/400 µm LMA passive fibers with larger core, and the results of the simulation are shown in Fig. 3.

 

Fig. 2. Mode field area and dispersion between 1.9-2.4 µm for (a) 32/250 µm fiber as well as (b) 48/400 µm fiber

Download Full Size | PDF

 

Fig. 3. The variation of (a) AM2-F2, (b) receiving efficiency and coupling efficiency, (c) focused spot size, and (d) divergence angle and NA with F1-Al1 distance using a pair of 50 mm aspherical lens coupled to 25/250-48/400 system which simulated in Zemax.

Download Full Size | PDF

After calculating in Zemax, the focal length of the two aspherical lenses for 1.96 µm laser is about 47.62 mm. It is worth noting that the highest coupling efficiency could be achieved by placing the emitting and receiving fibers at the focused spot of the two aspherical lenses. From Fig. 3(a), the distance of AL2-F2 will be reduced as F1-AL1 increases, which means that the spot focused by AL2 will be moved forward. Therefore, when increasing F1-AL1 in the experiment, we should also reduce AL2-F2 to obtain the best coupling efficiency.

In the lens coupling simulation using Zemax, three main efficiencies are considered, which are the system efficiency, the receiving efficiency, and the coupling efficiency. Among them, the coupling efficiency is the product of the system efficiency and the receiving efficiency. The system efficiency is mainly affected by the Fresnel reflection of the lens as well as the end face of the receiving fiber, which is a constant value and is calculated to be about 75.92%. The receiving efficiency is mainly determined by the NA of the receiving fiber and the aberration. Figure 3(b) shows that the total coupling efficiency as well as the receiving efficiency tend to increase and then decrease as F1-AL1 increases, and the coupling efficiency is about 56.5% when both the emitting and receiving fibers are located at the focal length of the two aspherical lens (47.62 mm). Figure 3(c) shows that the size of the focused spot behind AL2 rises with the increase of F1-AL1. The radius of the focused spot reaches 20 µm when F1-AL1 is equal to 48.935 mm, which exactly matches the radius of the fundamental-mode's mode field of the 48/400 µm fibers. At this time, we can obtain the highest coupling efficiency of about 75.9%, this number is very close to the system efficiency. Furthermore, it is worth noting that the reduction of the beam divergence angle and NA is also one of the reasons for the improvement of the single-mode light coupling efficiency, as shown in Fig. 3(d).

We also performed the same simulations for the 32/250 fiber, for which the precise fundamental mode matching technique is also applicable.

3. Experimental setup

Figure 4 shows a schematic setup of the high-power, high-energy tunable Raman soliton system, including a seed source, a 2 µm CPA system, a high-energy & high-power Raman frequency shifting system, and associated test equipment. The seed source is a Raman soliton at 1.95 µm obtained by using a 1.55 µm femtosecond laser pumping a 1.6 m highly germanium-doped nonlinear fiber (98 mol%.), which was time-domain broadened using a ∼96 m ultra-high numerical aperture fiber (Nufern Inc. UHNA4) with a dispersion of ∼48.5 ps/nm/km at 1.95 µm. Then the broadened pulse was coupled into a two-stage TDFA. The gain fiber in the preamplifier is a 2.5 m long double-clad thulium-doped fiber (Nufern Inc. DC-TDF-10/130) with core/cladding diameters of 10/130 µm and a cladding absorption coefficient of 9 dB/m at 793 nm. The gain fiber in the main amplifier is a 2-meter-long double-clad thulium-doped fiber (Nufern Inc. LMA-25/250) with a cladding absorption coefficient of 11.5 dB/m at 793 nm.

 

Fig. 4. Scheme of the experimental setup for high-power and high-energy tunable Raman soliton system. ISO: isolator; TDF: thulium-doped fiber; LMA-TDF: large mode area thulium-doped fiber; AL: aspherical lens; GM: golden mirror; OSA: optical spectral analyzer.

Download Full Size | PDF

It should be noted that the main amplifier fiber is meticulously coiled in a water cooling panel to filter out the higher-order transverse modes to ensure that the output laser is basically single-mode, and the tail is cut at an 8-degree angle to prevent Fresnel reflection at the end face. After the main amplifier, the remaining cladding pump light is filtered by a long-pass filter (LPF), then the clean signal light is compressed by a transmission diffraction grating pair (Gitterwerk) with a groove density of 800 lines/mm, the diffraction efficiency of each grating can reach more than 95%.

The compressed pulse passes through another aspherical lens with a focal length of 50 mm into a LMA passive fiber (receiving fiber) which was coiled on an aluminum plate. Corresponding to the previous simulations, two types of LMA passive fibers are chosen to explore Raman soliton self-frequency shift (SSFS). The first one is a 32/250 µm fiber with an NA of 0.075 and a length of 3.4 m, the second is a 48/400 µm fiber with an NA of 0.065 and a length of 8 m. The dispersion (D) of the two fibers at 1.96 µm is 37.31 ps/nm/km and 41.26 ps/nm/km, respectively, as shown in Fig. 2. After experimentally coupling two types of fibers, we hope to verify the high adaptability of our precision fundamental mode matching technique and also to validate the potential for power and energy enhancement of the larger core diameter fiber.

4. Experimental results and discussion

4.1 High power 2 micron chirped pulse amplifier

The Raman soliton seed source has an output power of ∼10 mW and a repetition frequency of 57.8 MHz. As shown in Fig. 5(a), its spectrum has a central wavelength of 1.954 µm and a 3 dB bandwidth of ∼48 nm. Before this Raman soliton enters the thulium-doped fiber preamplifier, it is broadened in the time domain to ∼180 ps using a 96 m UHNA4 fiber to avoid nonlinearities introduced by high peak power pulses inside the gain fiber. Due to the mode-field mismatch between the fibers and the Fresnel reflection loss at the cross-section, the power of the broadened pulse drops to about 6.5 mW. Then the broadened pulse is amplified in the two stages of the thulium-doped fiber amplifier.

 

Fig. 5. (a) The spectrum of the seed source (black line), main amplifier (red line), and compressed pulse (blue line); (b) The output power of the main amplifier pulse and compressed pulse.; (c) The pulse width and (d) Beam quality of the pulse compressed by gratings.

Download Full Size | PDF

As shown in Fig. 5(b), the maximum average output power is about 70 W (after filtering the pump light) and the amplifier slope efficiency is close to 50%. In the next stage, the amplified pulse was compressed by a folded Treacy compressor with a compression efficiency close to 50%, and the corresponding average and peak powers were 34.54 W and 0.91 MW, respectively. In our experiments, the high peak power would play a crucial role for the frequency shift of Raman solitons. The compressed spectrum of the gratings measured at the highest output power is shown as the blue line in Fig. 5(a), which has a center wavelength of 1960nm and a FWHM of 28.4 nm.

As shown in Fig. 5(c), the pulse duration obtained after compression is 658 fs, a few pedestals can be seen in the autocorrelation trace. The formation of pedestals can be attributed to uncompensated higher-order dispersion in the spreading fiber and nonlinearities accumulated in the LMA fiber amplifiers under high-power operation. In order to ensure the efficiency of lens coupling in subsequent LMA passive fibers, it is necessary to characterize the beam quality of the compressed beam. We tested the beam quality (1.187) of the output light using the knife-edge method, as shown in Fig. 5(d). Predictably, the pulse with good beam quality is conducive to coupling the fundamental-mode light to the LMA fibers with high efficiency.

4.2 High energy and wide-tunable range soliton frequency shift in LMA passive fibers with different core diameters

According to the simulation of the two types of receiving fibers in Section 2, we used two aspherical lenses with 50 mm focal lengths as a coupling system in our experiment to verify the simulation results in Zemax at low power (pump power of 10 W). Firstly, we coupled the LMA passive fiber with a large core diameter of 48 µm, and the first step of coupling is to collimate AL1 and emitting fiber which are assembled in a three-dimensional adjustment frame. The F1-AL1 is approximately 47.5 mm after the collimation operation, which is not much different from that calculated in the simulation.

A long-pass filter to filter out the pump light and a grating compressor is then added. Before installing AL2, the distance from AL1 to AL2 in the system is carefully measured and calculated to ensure that AL2 is located at the simulated distance of 1 m from AL1. Subsequently, the compressed pulses are then coupled into the receiving fiber. Since grating pairs, gold mirrors, and other devices are added in the experimental aspherical lens coupling system, we only focus on the coupling efficiency of AL2 and the receiving fiber here for the convenience of comparison. According to the 90% transmittance of the first aspherical lens, we know that the total coupling efficiency divided by 90% can be converted into the coupling efficiency of AL2 and the receiving fibers, as shown in Fig. 6 (red dots).

 

Fig. 6. Comparison of experimental and simulated coupling efficiencies for (a) 25/250-48/400 system and (b) 25/250-32/250 system.

Download Full Size | PDF

According to the simulations in Section 2, the highest coupling efficiency can only be obtained by placing both types of receiving fibers at the focused spot of two aspherical lenses for each F1-AL1 distance. In our experiments, we first tested the case of placing both the emitting and receiving fibers at the AL1 and AL2 focal distances. Provided that the emitting fiber (25/250 µm) has been placed at the focal length of AL1, by adjusting the distance of the receiver fiber from AL2, there exists a maximum coupling efficiency so that the receiving fiber is located exactly at the focal length of AL2. After finely adjusting the 3D adjustment frame of the receiving fiber at each F1-AL1, we obtain the same coupling power trend as the simulation. The highest coupling efficiency is about 54%, which is slightly different from the simulated single-mode light coupling efficiency (61.2%) of the system when F1-AL1 is equal to the focal length. There are two main reasons for this gap. Firstly, the laser of the emitting fiber itself is not a pure single-mode beam, which is defaulted to single-mode beam in the simulation. Secondly, due to the gratings in the optical path, the small deviation of the parallelism between the two gratings will lead to the spot not being a strictly circular Gaussian beam, we obtain a lower coupling efficiency than the simulation.

After ensuring that both the emitting and receiving fibers are located at the focal lengths of the two aspherical lenses, we accurately adjust the F1-AL1 by the three-dimensional adjusting frame. In order to facilitate the comparison with the simulated results, we set an experimental interval from 45.4 mm to 50.2 mm with a step size of 0.2 mm. Similarly, after each adjustment of the F1-AL1 in our experiments, it is also necessary to adjust the AL2-F2 distance to obtain the maximum coupling efficiency. The maximum coupling efficiency at each F1-AL1 is shown in Fig. 6(a). By finely adjusting AL2-F2 when F1-AL1 is adjusted away by about 1.5 mm, we can get the highest coupling efficiency of about 77.4%. The difference from the 85.1% efficiency in the simulation is also caused by the mode of the output laser from the emitting fiber and the small spot deformation caused by the gratings. Similarly, single-mode light coupling operation was performed for the 32/250 µm fiber. The highest coupling efficiency of 75.2% is obtained at about 48.2 mm, and its trend is consistent with the simulation, as shown in Fig. 6(b).

Through the above experiments, we verified the feasibility of the dual-lens out of focal-length placement method to improve the coupling efficiency. Next, we increase the pump power to observe the change of the coupling efficiency with the increase of the pump power and test the quality of the output beam to figure out whether our fundamental-mode matching technique still works well at high power. Figure 7(a) and (b) show the variation of output power and coupling efficiency with increasing pump power in two types of receiving fibers. It can be seen that the output power continues to rise as the pump power rises, but the coupling efficiency of the two fibers has a gradual downward trend. This implies that the compressed high peak power pulse is likely to have excited Raman soliton with different frequency shifting distances. As the Raman soliton moves to the long wavelength, its loss at the long wavelength increases so that the coupling efficiency will be decreased, this thought will be verified later when we test the spectral properties. The beam qualities of 3.4 m 32/250 µm as well as 8 m 48/400 µm fibers tested at the pump power of 140 W, as shown in Fig. 7 (c) and Fig. 7(d), are 1.3196 and 1.275, respectively. Both of them possess beam qualities close to the diffraction limit, which indicates that we have successfully carried out the precise fundamental-mode matching technique. In order to further explore the reason why the coupling efficiency of the two receiving fibers decreases at different rates in different pumping intervals, we tested the spectral characteristics of the laser from the two receiving fibers, as shown in Fig. 8 below.

 

Fig. 7. (a) The variation of output power and coupling efficiency with increasing pump power in 3.4 m 32/250 µm and (b) 8 m 48/400 µm LMA passive fibers; (c) beam quality of output pulses from 3.4 m 32/250 µm and (d) 8 m 48/400 µm LMA fibers.

Download Full Size | PDF

 

Fig. 8. The experimental frequency shift spectrum of soliton and autocorrelation curves at different pump powers after (a) 3.4 m 32/250 µm LMA passive fiber and (b) 8 m 48/400 µm LMA passive fiber.

Download Full Size | PDF

Figure 8 shows the experimental frequency shift spectra of the soliton after 3.4 m 32/250 µm and 8 m 48/400 µm LMA passive fibers. For the 3.4 m 32/250 µm fiber, the furthest Raman soliton located at 2350 nm is observed at a pump power of 140 W. The right plot of Fig. 8(a) demonstrates the pulse widths of Raman solitons at each pump power. As the pump power is increased, the pulse width of the Raman soliton decreases and then increases. The reason for the decrease at the beginning is that the pulse undergoes self-phase modulation which results in spectral broadening and compressed pulse width. The soliton would have the narrowest pulse width when it is just split out of the complete soliton. As the laser wavelength is shifted to the long wavelength, it can be seen in Fig. 2 that the corresponding dispersion of the fiber increases with the shifting wavelength, which causes the pulse width of the fundamental-order Raman soliton to gradually increase as well. In addition, the loss at long wavelengths also leads to a decrease in the coupling efficiency, as shown in Fig. 7(b). In the 0-30 W pump stage, the coupling efficiency decreases slowly, which is due to the fact that the Raman soliton is just split with a slight wavelength-dependent loss. In the 30-100 W pump stage, the Raman soliton rapidly shifts to 2300 nm, which leads to a steep rise in its loss at the long-wavelength region. In the last 100-140 W pumping stage, due to the increase of the Raman soliton pulse width at long wavelength and the decrease of the nonlinear coefficient caused by the rise of the fiber mode field diameter, the first-order Raman soliton's frequency shift speed becomes very slow. At this time, the loss begins to increase gently, and the coupling efficiency begins to fall steadily.

Similarly, for the 8 m 48/400 µm LMA fiber, as shown in Fig. 8(b), its farthest Raman soliton comes to 2270 nm at the pump power of 140 W. The Raman soliton's pulse width also shows a tendency to increase and then decrease. However, due to the longer length of the 48/400 fiber used, the pulse accumulates an amount of dispersion and will be broadened faster as it transmits within the fiber so that the pulse width has an overall higher level. In addition, the Raman soliton's splitting speed is much slower than that in the 32/250 µm fiber due to the smaller nonlinear coefficient of this fiber. In the 0-40 W pump stage, when the Raman soliton is just split, the coupling efficiency decreases slowly. In the 40-80 W pump stage, the Raman soliton rapidly shifts to 2180 nm, which leads to a steep rise in the loss at its long wavelength. In the last 80-140 W pump stage, although there is no apparent tendency to split out of the second-order Raman soliton, the frequency shift of the first-order Raman soliton becomes slow very early due to the broader Raman soliton pulse width and the decreasing nonlinearity. Consequently, the loss of the long-wavelength region starts to increase smoothly, and the coupling efficiency also starts to decrease smoothly. However, since the wavelength range of the frequency shift is less than that of the 32/350 µm fiber, the Raman soliton loss is lower and its overall coupling efficiency is higher than the 32/250 µm fiber.

The characteristics of the tunable Raman soliton output from both receiving fibers are tested next. For the 32/250 µm fiber, it can be seen from Fig. 9(a) that the output power of the Raman soliton increases from 3.59 W to 5.97 W as the pump power increases from 40 W to 110 W, and the pulse energy subsequently increases from 62.2 nJ to 103.38 nJ. Besides, it's worth noting that the output pulse power and pulse energy decreased from 110 W to 140 W, which was due to the high loss at the long wavelength region of the silica fiber. The variation of the peak power of the Raman soliton is given in Fig. 9(b), where the peak power of the Raman soliton is above 0.5 MW from 40 W to 140 W typically. The peak power of the Raman soliton is close to 1 MW at a pump power of 80 W. After 80 W, the Raman soliton pulse width rises faster, leading to a gradual decrease in its peak power. In the test, the narrowest Raman soliton of 92 fs can be obtained at a pump power of 50 W. In this case, it is easy to obtain Raman solitons with high power and large energy in a wide tuning range, which verifies a better tunability performance. Furthermore, the output Raman solitons have a narrower pulse width due to the use of a shorter fiber.

 

Fig. 9. (a)The soliton power and soliton energy; (b) soliton pulse width and soliton peak power versus frequency shift wavelength for the output pulse of 3.4 m 32/250 µm LMA passive fiber; (c) the soliton power and soliton energy (d) soliton pulse width and soliton peak power versus frequency shift wavelength for 8 m 48/400 µm LMA passive fiber.

Download Full Size | PDF

For the 48/400 µm fiber, as shown in Fig. 9(c), the output power of Raman soliton increases from 4.61 W to 7.23 W and the pulse energy from 79.85 nJ to 125.2 nJ as the pump power increases from 40 W to 140 W. Typically, the tunable Raman soliton of >100 nJ is achieved above the pump power of 70 W-140 W, which differs from that of the 3.4 m 32/250 µm fiber. The output power and pulse energy show a sustained increasing trend, which is because the Raman soliton can only be shifted up to 2270 nm with the 8 m 48/400 fiber without experiencing greater loss at longer wavelengths. However, considering the longer fiber length used, the increase in power and pulse energy gradually level off at longer wavelengths and continued increase in pump power may lead to a decrease in both. The peak power variation of the Raman soliton is given in Fig. 9(d), where the peak power of the Raman soliton is above 0.5 MW from 60 W to 140 W. Typically, the peak power of Raman soliton reaches the highest level (0.588 MW) when the pump power is 110 W. After 110 W, the peak power gradually decreases due to the continuous increase of the Raman soliton pulse width and the gradually flattened increase of the pulse energy. The narrowest Raman soliton pulse width obtained with 8 m 48/400 fiber was measured to be 176 fs at a pump power of 60 W.

Compared to 32/250 fiber, the power and energy are significantly boosted by 1.43 W and 21 nJ, respectively. It can also be seen from Fig. 9(c) that there is no significant decrease in the growth trend of soliton energy and power before the pump power of 110 W (2240 nm) even though a longer fiber length is used, which shows its energy and power enhancement potential. What can be foreseeable is that if the pulse width of our light source can be further reduced, we believe that higher energy and power Raman solitons can be obtained in 48/400 fiber.

5. Conclusion

In conclusion, we have developed a dual-lens out of focal-length placement method to couple small-core LMA fibers and large-core LMA fibers. The effect of the placement of aspherical lenses with the same focal lengths on the single-mode light coupling efficiency of LMA fibers, which can be improved to the system efficiency of single-mode light coupling, has been systematically investigated theoretically and verified experimentally. This method largely simplifies one's consideration of choosing the focal length of the lens in similar situations and guarantees the safety and efficiency of the system. In addition, we have built a CPA laser source with higher repetition frequency and investigated the high-power and high-energy Raman soliton generated by single-mode excitation of LMA passive fiber. Finally, in the 25/250- 32/250 system, we achieved Raman solitons with a tuning range of 1.96-2.35 µm and a maximum output power of 5.8 W, the energy of >100 nJ has a tunable range of >80 nm. In the 48/400 µm LMA fibers, we achieved Raman soliton pulses with a tuning range of 1.96-2.27 µm and a maximum soliton output power of 7.3 W, the energy of >100 nJ has a tunable range of >170 nm. The output power of 7.3 W is the highest Raman soliton power currently available in silica fibers, which can also be used as a high-power pump source of 3-5 µm mid-infrared Raman solitons.