Open Access
Add to BibSonomy
Abstract
Ultra-broadband supercontinuum (SC) is generated by an all-fiber system with well-defined pulses (WDPs) as a seed laser. Through properly adjusting the lengths of the fiber segments in the system, sub-pulses with high peak powers are generated through the process of optical pulse breakup. Then, the broken optical pulses are launched into a hybrid nonlinear, which consists of a highly nonlinear optical fiber and a photonic crystal fiber, generating a SC of 1.8 W covering a spectral range from 554 nm to 2.17 µm. In this study, it is demonstrated through experimental observation that optical pulse breakup is an essential process for WDPs before the pulses are launched into a nonlinear optical fiber to generate SC. However, pulse breakup must not happen before the amplification of the WDPs; otherwise, the amplification efficiency is decreased, leading to a subsequent low efficiency of SC generation. The proper breakup of the WDPs after they are efficiently amplified is accomplished by sending these pulses through an optimum length of regular optical fiber, which depends on the peak power of the amplified WDPs. An optical fiber of an insufficient length leads to insufficient pulse breakup, which does not significantly increase the peak power of the pulses, resulting in little enhancement of SC generation. An excessively long fiber leads to excessive pulse breakup, which stretches the pulsewidth and splits the pulse energy, resulting in decreasing the efficiency of SC generation.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Efficient supercontinuum (SC) generation through optical fibers has been widely investigated for its promising applications in various areas, such as optical coherence tomography (OCT), chromatic LiDAR and spectroscopy [1–5]. Depending on the parameters of the optical laser source and those of the optical fibers, the generation of an ultra-broadband SC in an all-fiber system involves many nonlinear effects, including self-phase modulation (SPM), stimulated Raman scattering (SRS), and dispersive-wave generation [6,7]. To strongly induce these nonlinear effects, a cascade of optical fiber amplifiers, such as erbium-doped fiber amplifier and erbium/ytterbium co-doped fiber amplifier, is usually utilized to scale up the peak power of the laser pulses before pumping a nonlinear optical fiber for SC generation [8,9]. For example, a broadband SC spectrum covering a wavelength range from 690 nm to 2.35 µm was generated by launching amplified optical pulses with a high average power of 714 W into a passive fiber that has a length of 20 m [10]. Moreover, a SC spectrum covering a wavelength range from 480 nm to 2.4 µm was generated by pumping amplified optical pulses with a high average power of 556 W into a photonic crystal fiber (PCF) that has a zero-dispersion wavelength (ZDW) at 1.04 µm [11].
Recently, noise-like pulses (NLPs) have been proposed as an effective laser source for a SC generation system [12–15]. Such pulses consist of femtosecond structures that have stochastically varying temporal durations and peak powers within picosecond wavepackets. This temporal feature leads to a double-scaled autocorrelation trace, having a femtosecond spike riding on a picosecond pedestal [16–18]. Previous works have demonstrated through numerical simulation that the efficiency of SC generation by using NLPs as the laser source is higher than that by using well-defined pulses (WDPs) because, for NLPs and WDPs of the same wavepacket peak power, the peak powers of the femtosecond inner structures in the NLPs are much higher than those of the WDPs. For example, simulation results of comparing the efficiency of SC generation through a 10-cm PCF by using NLPs with that using ultrashort hyperbolic secant WDPs were reported. The hyperbolic secant WDPs required a peak power of 3 kW to generate a smooth and broadband SC spectrum, which is comparable to the SC spectrum generated by using NLPs with a wavepacket peak power of only 200 W [19]. However, in many experimental studies of SC generation by WDPs laser sources, the observed differences between the SC spectra that were generated by using WDPs and NLPs were not as significant as the differences seen in the reported simulation results [10,11,20–22]. To explain this discrepancy between simulation and experiment, we demonstrated through numerical simulation in our previous research that the process of pulse breakup is essential for WDPs to generate SC [23]. The pulse breakup makes the temporal features of amplified WDPs noise-like, generating sub-pulses with peak powers that are much higher than those of unbroken WDPs. This simulation result numerically resolved the conflict between the experimental observations and the simulation results of SC generation by using WDPs in previous research [23].
In this work, we demonstrate the importance of controlling the level of pulse breakup in an optical fiber with a negative group-velocity dispersion (GVD) for a SC generation system that uses a laser source of WDPs. We present through experiment that pulse breakup is an essential process to generate SC, and the experimental result supports our simulation result in the previous research [23]. Nonetheless, the process of pulse breakup must be prohibited before the final amplification of the WDPs. If it takes place before the amplification, the efficiency of amplification is significantly reduced because of the gain saturation caused by the high-peak-power sub-pulses and the redshifted spectrum due to the intrapulse SRS. Such reduced amplification efficiency further lowers the efficiency of SC generation. In addition, the length of the total fiber segments that connect the amplifier with the nonlinear fiber should be properly adjusted to strongly, but not excessively, induce the optical pulse breakup at an optimum level. Otherwise, the excessive pulse breakup stretches the pulsewidth and splits the pulse energy over too many sub-pulses, leading to a low efficiency of SC generation.
2. Experimental setup
Figure 1 shows the schematic of our all-fiber SC generation system. The system consists of five parts: a seed fiber laser that generates WDPs at a central wavelength of 1.56 µm, two pieces of single-mode fibers (SMFs) that are used as connecting fibers, a two-stage optical fiber amplifier that consists of a preamplifier and a booster, and a hybrid highly nonlinear fiber that consists of an 8-cm highly nonlinear fiber (HNLF) and a 2-m PCF [20,23]. In the following, we refer to the SMF that connects the preamplifier with the booster as SMF1, and the fiber that connects the booster with the HNLF as SMF2, as shown in Fig. 1. In this study, a background-free intensity autocorrelator (Femtochrone FR-103XL) is used to measure the autocorrelation traces of the optical pulses, and an optical spectrum analyzer (Anritsu MS9740A) is used to analyzed the spectral features of the seed WDPs and the amplified optical pulses. The broadband SC spectra are characterized by using a monochromator (Digikrom CM110) together with a photodiode detector module (SP AD131), covering a spectral range from 300 nm to 4 µm.
3. Simulation method
To further understand the evolution of an optical pulse propagating in an optical fiber, the generalized nonlinear Schrödinger equation (GNLSE) was used to comprehensively describes the nonlinear effects in the optical fiber, written as [7,12]:
4. Results of experiment and simulation
4.1 Supercontinuum generation by broken optical pulses
Seed WDPs that have a central wavelength at 1.56 µm and a temporal duration of 8.7 ps at a repetition rate of 19 MHz were launched into the preamplifier for their average power to be amplified from 0.7 mW to 120 mW. These pulses also propagated through SMF1 after the preamplifier. Figure 2 shows the spectral and temporal characteristics of the optical pulses at the output of SMF1 of different lengths, including 5 m, 10 m, 20 m, and 30 m. As shown by the top panels in Figs. 2(b) for a SMF1 of 5 m, the single-scaled autocorrelation with a width of 12.2 ps indicates that the amplified WDPs maintained their temporal feature with a pulsewidth of 7.9 ps. When a 10-m SMF1 was used, the pulse spectrum was broadened with its long-wavelength edge extended to 1.64 µm, and the autocorrelation trace became double-scaled, as respectively shown in the second panels of Figs. 2(a) and (b) labeled with 10 m. These experimental results indicate that for a sufficiently long SMF1 the effect of intrapulse SRS and the process of soliton fission make the temporal feature of the amplified optical pulses noise-like [25,26]. Then, when the length of the SMF1 was further increased to 20 m and 30 m, the long-wavelength edge of the pulse spectrum was further extended to 1.68 µm by the continuously induced effect of intrapulse SRS and the process of soliton fission, and the autocorrelation trace remained a double-scaled feature, as shown in the lower panels of Figs. 2(a) and (b) labeled with 20 m and 30 m, respectively. Note that the minimum irreducible length of SMF1 is 5 m because it is the total length of the necessary passive fiber components following the gain fiber in the preamplifier, including the wavelength-division multiplexer and the optical isolator.
Fig. 2. Experimentally measured (a) spectra and (b) temporal autocorrelation traces of the preamplified optical pulses at the output of SMF1 of 5 m, 10 m, 20 m, and 30 m lengths.
To further understand the evolution of the optical pulses, GNLSE was used to simulate pulse propagation in the preamplifier and SMF1. In the simulation, a WDP that has a pulsewidth of 9 ps and an average power of 0.7 mW was launched into the 1.2-m EDF of the preamplifier. Then, the amplified WDP that has a pulsewidth of 8.2 ps and an average power of 134 mW was subsequently sent into the SMF1. Figure 3 shows the temporal and spectral evolution of the amplified WDP propagating through a SMF1 over four different lengths of 5 m, 10 m, 20 m, and 30 m. As can be seen in the top panels of Figs. 3(a) and (b) for 5 m, the amplified optical pulse maintained the profile of a WDP after it propagated through the preamplifier, having a peak power of 0.9 kW. Then, as the length of the SMF1 is increased to 10 m and beyond, the process of soliton fission is gradually induced, generating numerous soliton-like sub-pulses with peak powers above the kilowatt level, as shown in the panels for 10 m, 20 m, and 30 m in Fig. 3(b). Meanwhile, the temporal waveform of the pulse is stretched because the broadened spectrum increases the difference in the propagation speed between the main pulse and the Stokes waves.
Fig. 3. Stimulated (a) spectra and (b) temporal waveforms of the preamplified optical pulses at the output of SMF1 of 5 m, 10 m, 20 m, and 30 m lengths.
Fig. 4. Experimentally measured optical spectra of the supercontinua generated by pumping one-stage amplified WDPs into a 1-m HNLF through different lengths of SMF1 of 5 m (black), 10 m (red), 20 m (blue), and 30 m (green).
Subsequently, the optical pulses from the output of SMF1 of different lengths, including 5 m, 10 m, 20 m, and 30 m, were respectively launched into a HNLF with a length of 1 m. In the case that the length of SMF1 is 5 m, the almost unchanged spectrum of the pulses from its output indicates that unbroken WDPs are launched into the HNLF. Because of their low peak powers, such unbroken WDPs are not able to induce sufficient nonlinear effects in the HNLF, thus insignificant SC generation with limited spectral broadening, as shown by the black curve in Fig. 4. By comparison, when the 10-m SMF1 is used to induce sufficient optical pulse breakup prior to the process of SC generation in the HNLF, a SC spectrum with an average power of 92 mW having a spectral power density above –30 dBm/nm covering a wavelength range from 1.13 µm to 2.09 µm is generated, as presented by the red curve in Fig. 4. Further increasing the length of SMF1 to 20 m or 30 m only slightly broadens the SC spectrum, as shown by the blue and green curves in Fig. 4, because the pulses are already broken after they propagate over 10 m in SMF1. In agreement with our previous simulation research [23], these results demonstrate for the first time through experiment that pulse breakup is an essential process for efficient SC generation before amplified WDPs are launched into the HNLF.
4.2 Amplification of broken optical pulses with booster
After the first-stage amplification, the one-stage amplified pulses are launched into the booster for further amplification, and the two-stage amplified pulses are output from SMF2 for a fixed SMF2 length of 35 cm. Figure 5(a) shows the output power from the booster as a function of the length of SMF1, which connects the preamplifier and the booster. The output power decreases from 3.5 W to 2.5 W as the length of SMF1 increases from 5 m to 30 m. One reason for the decreased amplification efficiency is gain saturation caused by the high peak powers of the sub-pulses resulting from pulse breakup in SMF1. Another reason is that the spectrum of the one-stage amplified optical pulses gradually redshifts and broadens to exceed the gain bandwidth of the booster, during the propagation in SMF1. This decreases the spectral power density of the one-stage amplified pulses within the gain bandwidth of the booster. By comparing the top panel and the bottom panel of Fig. 2(a), the spectral power density at 1.56 µm of the optical pulses output from the 5-m SMF1 is 8 dBm/nm higher than that from the 30-m SMF1. In the other word, within the gain bandwidth of the booster, the spectral power density of the optical pulses from the 5-m SMF1 is 6.3 times that of the pulses from the 30-m SMF1 when these pulses enter the booster for second-stage amplification. Therefore, for the case of using the 5-m SMF1, the two-stage amplified pulses induce strong nonlinear effects, including SPM and intrapulse SRS, thus broadening the pulse optical spectrum to cover a wavelength range from 1.42 µm to 1.75 µm, which is shown by the black curve in Fig. 5(b). It can be seen from the other curves shown in Fig. 5(b) that this pulse spectral broadening effect is reduced as the length of SMF1 is increased because the lower spectral power density of the pulses from a longer SMF1 reduces the strength of the nonlinear effects. As an example for comparison, in the case of the 30-m SMF1, the spectrum of the two-stage amplified optical pulses covers a narrower wavelength range from 1.47 µm to 1.69 µm, shown by the blue curve in Fig. 5(b).
Fig. 5. (a) Average power of the two-stage amplified pulses output from SMF2 of 35 cm as a function of the length of SMF1. (b) Experimentally measured optical spectra of the two-stage amplified pulses output from SMF2 for four different lengths of SMF1 of 5 m (black), 10 m (red), 20 m (blue), and 30 m (green). Stimulated waveforms of the two-stage amplified pulses output from SMF2 when the lengths of SMF1 are (c) 5 m and (d) 30 m.
Figures 5(c) and (d) respectively show the simulated optical waveforms of the two-stage amplified pulses output from SMF2 of a fixed length of 35 cm for two different lengths of SMF1 of 5 m and 30 m. As seen in Fig. 5(c), though the pulse at the output of 5-m SMF1 doesn’t break up before it enters the booster, the two-stage amplified output pulse at the output of SMF2 has numerous sub-pulses that have peak powers higher than 80 kW. By contrast, when SMF1 of 30 m is used prior to the booster for inducing sufficient pulse breakup before the second-stage amplification, the temporally stretched waveform of the broken pulse entering the booster lowers the amplification efficiency of the booster. Consequently, most of the sub-pulses within the two-stage amplified pulse at the output of SMF2 have peak powers below 25 kW, as seen in Fig. 5(d).
Subsequently, the two-stage amplified pulses are launched into a hybrid nonlinear fiber that consists of an 8-cm HNLF and a 40-cm PCF. Figure 6 shows the SC generated by pumping the hybrid nonlinear fiber with the two-stage amplified optical pulses for different lengths of SMF1 before the booster. With 5-m SMF1, the preamplified optical pulses entering the booster for the second-stage amplification are not broken, and the SC generated at the output of the hybrid nonlinear fiber has an average power of 2.12 W with a spectral power density above –20 dBm/nm in the wavelength range from 580 nm to 2.39 µm, as shown by the black curve in Fig. 6. By comparison, when the length of SMF1 is 30 m, the preamplified optical pulses entering the booster are broken, and the SC has an average power of 1.25 W with spectral power density above –20 dBm/nm in wavelength range from 687 nm to 2.2 µm, as shown by the green curve in Fig. 6. These trimmed SC spectral edges for the case of 30-m SFM1 indicate that the efficiencies of SC generation are reduced because of the lowered amplification efficiency for the stretched temporal waveform of the broken preamplified pulses. Therefore, even though optical pulse breakup is an essential process for SC generation, it must take place after the process of amplification so as to maximize the peak powers of the sub-pulses that are generated by pulse breakup.
Fig. 6. Experimentally measured optical spectra of the supercontinua generated by launching two-stage amplified optical pulses into a hybrid nonlinear fiber consisting of an 8-cm HNLF and a 40-cm PCF, for different lengths of SMF1 of 5 m (black), 10 m (red), 20 m (blue) and 30 m (green).
4.3 Supercontinuum generation by limited broken pulses
In the preceding subsection, we demonstrated that for efficient pulse amplification and effective SC generation, breakup of the pump pulses must not happen before their final amplification. In this subsection, the length of SMF1 is fixed at 5 m for preventing pulse breakup in SMF1 to maximize the amplification efficiency of the booster. Figure 7 shows the spectral characteristics of the two-stage amplified optical pulses at the output of SMF2 of different lengths, including 35 cm, 65 cm, 1 m, and 2 m. By comparing the spectrum for the 35-cm SMF2 to that for the 65-cm SMF2, shown respectively as the black and the red curves in Fig. 7, we see that the short-wavelength edge is trimmed from 1.43 µm to 1.48 µm while the long-wavelength edge is extended from 1.73 µm to 1.89 µm, resulting in a broadening accompanied by a redshift of the pulse spectrum as the length of SMF2 is increased from 35 cm to 65 cm. These spectral changes indicate that SPM already causes significant spectral broadening as a pulse propagates through the initial 35 cm; then, intrapulse SRS and soliton fission lead to further broadening, together with redshifting, of the spectrum as the pulse continues to propagate through the remaining 30 cm in the 65-cm SMF2. As the length of SMF2 is further increased to 1 m and beyond, the optical spectrum continues to broaden and redshift toward the long-wavelength side due to further intrapulse SRS and soliton fission. It is worth mentioning that at the power level of this experiment, 65 cm of SMF2 is sufficient long for intrapulse SRS to extend the long-wavelength edge of the pulse spectrum beyond 1.89 µm.
Fig. 7. Experimentally measured optical spectra of the two-stage amplified optical pulses at the output of SMF2 of different lengths, including 35 cm (black), 65 cm (red), 1 m (blue), and 2 m (green).
Figures 8(a) and (b) respectively show the simulated optical spectra and waveforms of the two-stage amplified pulses for different SMF2 lengths of 0 cm, 35 cm, 65 cm, 1 m, and 2 m. Here, the simulated spectra in Fig. 8(a) are more fluctuating than the experimentally measured spectra because the simulation is carried out by using a single pulse, whereas an experimentally measured spectrum is an averaged spectrum of millions of pulses in a pulse train. Because it is not practical to carry out a simulation of so many pulses, the fine fluctuating features of a simulated spectrum cannot be completely smoothed to match the smoothness of an experimentally measured spectrum. As seen in the top panels of Figs. 8(a) and (b) for 0 m of SMF2, optical pulse breakup has already taken place in the booster, generating sub-pulses of peak powers higher than 100 kW. These sub-pulses of high peak powers induce large soliton self-frequency shift (SSFS) during the propagation in SMF2; therefore, only tens of centimeters to broaden the spectrum [25]. In addition, the effect of SSFS depends on their own peak powers, causing these sub-pulses to have different wavelengths and propagate at different velocities. Therefore, while these sub-pulses interact with each other, the superposition of these inner structures enhances the peak power to an ultra-high level [27], for instance, beyond 300 kW as shown by the panel for 35 cm in Fig. 8(b). However, while the optical pulse further propagates through a longer SMF2, the further broadened spectrum increases the differences of velocities between these sub-pulses. This makes these sub-pulses separated quickly, and followed by the stretching of the waveform. Therefore, as shown in the bottom panel in Fig. 8(b), when the length of SMF2 is increased to 2 m, the stretched waveform lowers the peak powers of these sub-pulses, indicating that optical pulse breakup goes excessive.
Fig. 8. Simulated optical (a) spectra and (b) waveforms of the two-stage amplified optical pulses at the output of SMF2 of different lengths, including 0 cm, 35 cm, 65 cm, 1 m, and 2 m.
Subsequently, the pulses output from SMF2 of different lengths are pumped into the hybrid nonlinear fiber with a combination of an 8-cm HNLF and a 40-cm PCF to generate SC. As seen from the black curve in Fig. 9, in the case that the length of SMF2 is only 35 cm, a SC with an average power of 2.12 W with a spectral power density above –20 dBm/nm in a wavelength range from 580 nm to 2.4 µm is generated. The wavelength of a dispersive wave depends on the soliton that the dispersive wave generated from. Therefore, when a spectrum of a soliton redshifts due to intrapulse SRS, the spectrum of the generated dispersive wave also redshifts. In our case, the dispersive waves at around 1 µm are generated first. However, the length of the HNLF is insufficient to induce intrapulse SRS so that the dispersive waves with wavelengths between 1 µm and 1.56 µm cannot be generated, forming a spectral valley at 1.3 µm. The nonlinear processes of soliton dynamics that take place in the PCF can be observed by examining the changes in the spectral valley near 1.3 µm in the SC spectrum as the length of SMF2 is increased. As can be seen in Fig. 9, the spectral valley is initially filled as the length of SMF2 is increased from 35 cm to 65 cm. However, when the length of SMF2 is further increased to 1 m, this spectral valley becomes broadly deepened, and it further broadens as the length of SMF2 is increased to 2 m. These spectral changes indicate that a 1-m length for SMF2 is excessive in our case. A similar conclusion can also be obtained by observing the variations of the spectral power density in the short-wavelength region, which reflects the generation of dispersive waves. As can be seen, the black and red curves almost overlap near the short-wavelength edge, indicating that the level of dispersive-wave generation is similar for the cases of 35 cm and 65 cm of SMF2. However, as the length of SMF2 is increased to 1 m and beyond, the short-wavelength edge is significantly cut back, indicating that generation of dispersive waves is weakened. As a result, when a 2-m SMF2 is used to excessively enhance the process of pulse breakup, the SC spectrum generated through the hybrid nonlinear fiber covers a wavelength range from 798 nm to 2.44 µm with a small spectral portion at 750 nm and a deep spectral valley at 1.3 µm, as shown by the green curve in Fig. 9.
Fig. 9. Experimentally measured optical spectra of the supercontinua generated by pumping two-stage amplified pulses into a hybrid nonlinear fiber composed of cascaded 8-cm HNLF and 40-cm PCF after the pulses propagate through SMF2 of different lengths, including 35 cm (black), 65 cm (red), 1 m (blue), and 2 m (green).
By comparison to Fig. 9, for which the length of the PCF is 40 cm, Fig. 10 shows the flat and ultra-broadband SC spectra generated by increasing the length of the PCF from 40 cm to 2 m while the length of the HNLF remains at 8 cm, and the four curves respectively represent the cases of different SMF2 lengths of 35 cm, 65 cm, 1 m, and 2 m prior to the hybrid nonlinear fiber. Figure 11 shows the simulated SC spectra for two different SMF2 lengths of 35 cm and 2 m. As shown by the black curve in Fig. 10, when the length of SMF2 is 35 cm, the SC spectrum has an average power of 1.8 W with a spectral power density above –20 dBm/nm in a wavelength range from 554 nm to 2.17 µm. By comparing the black curves for 35-cm SMF2 in Figs. 9 and 10, it is seen that by increasing the PCF length from 40 cm to 2 m, the spectral valley at 1.3 µm is filled and the power density in the short-wavelength region is increased. These improvements in the spectral features indicate that both soliton dynamics and dispersive-wave generation can be further enhanced as the optical pulses propagate through a PCF of a length up to 2 m. When the length of SMF2 is increased to 65 cm, the short-wavelength edge slightly cut back. A comparison of the black curve for 35-cm SMF2 and the red curve for 65-cm SMF2 in Fig. 10 indicates that the optimum length of SMF2 for our system is less than 65 cm. When the length of SMF2 is further increased, the efficiency of SC generation is lowered. For the case of 2-m SMF2, as seen by comparing the green curves in Figs. 9 and 10, only a small spectral structure shows up at the spectral valley and the short-wavelength edge expands from 798 nm to 690 nm as the length of the PCF is increased from 40 cm to 2 m. This indicates that as SMF2 is lengthened its optimum length, the nonlinear processes for SC generation through a hybrid fiber are still enhanced by increasing the PCF length, but this enhancement diminishes. The reason is that an unnecessarily long SMF2 leads to excessive pulse breakup before the pulses enter the hybrid fiber for SC generation, resulting in lower peak powers of the sub-pulses in the broken pulses, thus decreased the efficiency of SC generation. The experimentally observed SC features shown in Fig. 10 and discussed above are verified by the simulation results shown in Fig. 11. It is worth mentioning that 35 cm is the absolute minimum length of SMF2 in our case because the minimum required length for splice fusion is 35 cm. Because the optimum length of SMF2 is less than 65 cm, as discussed above, the flexibility to vary the length of SMF2 is less than 30 cm in our case. In the case that the optical pulses have very different parameters from those of ours, such as a shorter pulsewidth and a higher power, the length of the fiber that connects the amplifier and the nonlinear fiber must be properly designed for the specific pulse parameters.
Fig. 10. Experimentally measured optical spectra of the supercontinua generated by pumping two-stage amplified pulses into the cascaded 8-cm HNLF and 2-m PCF through different lengths of SMF2, including 35 cm (black), 65 cm (red), 1 m (blue), and 2 m (green).
Fig. 11. Simulated optical spectra of the supercontinua generated by pumping two-stage amplified pulses into the cascaded 8-cm HNLF and 2-m PCF through SMF2 of 35 cm (black), and 2 m (red).
5. Conclusion
In this work, we demonstrate the importance of controlling the level of pulse breakup to efficiently generate an ultra-broadband SC when using WDPs as a seed laser source. By pumping unbroken one-stage amplified WDPs that have an average power of 120 mW directly into a 1-m HNLF, SC can hardly be generated. However, when a 10-m SMF is used between the preamplifier and the 1-m HNLF to induce pulse breakup before the pulses enter the HNLF, a SC having an average power of 92 mW with a spectral power density above –30 dBm/nm in a wavelength range from 1.13 µm to 2.09 µm is generated. It is demonstrated for the first time through experiment that pulse breakup is essential for the system with a seed laser source of WDPs to generate SC. Together with our previous simulation research, the conflict between the experimental observations and the simulation results of SC generation by using WDPs as the seed laser source in previous research is numerically and experimentally resolved. In the case of using two-stage amplified optical pulses to generate SC, it is worth noting that pulse breakup must not happen before the process of amplification. Otherwise, the amplification efficiency is lowered because of gain saturation caused by the sub-pulses that have high peak powers and a low spectral power density within the gain bandwidth of the booster due to a redshifted spectrum. Such lowered amplification efficiency limits the peak powers of the amplified sub-pulses at a low level, resulting in a low SC generation efficiency in the nonlinear fiber. Instead, by pumping unbroken WDPs into the booster at optimum amplification efficiency, proper optical pulse breakup takes place automatically in the booster, generating sub-pulses of high peak powers. Subsequently, by pumping these broken optical pulses into a hybrid nonlinear fiber, an ultra-broadband SC having a spectral power density above –20 dBm/nm in a wavelength range from 554 nm to 2.17 µm is generated. However, when the length of the SMF that connects the booster with the hybrid nonlinear fiber is increased from 35 cm to 2 m, the level of pulse breakup become excessive, stretching the waveform and splitting the pulse energy. The decreased peak powers of the these resulting sub-pulses further decrease the efficiency of SC generation. As a result, for the SC generated by pumping the hybrid nonlinear fiber with the excessively broken optical pulses, the spectral valley at 1.3 µm is deepened and broadened, indicating that soliton dynamics in the PCF is weakened. Meanwhile, the significance of dispersive-wave generation is reduced so that the short-wavelength edge of the SC spectrum is cut back from 554 nm to 690 nm. This study clearly demonstrates that to efficiently generate an ultra-broadband SC, the level of pulse breakup has to be properly controlled depending on the parameters of the optical pulses, such as the pulsewidth and the power.
Funding
National Science and Technology Council (MOST 110-2622-8-A49-008-SB, NSTC 112-2221-E-A49-133-MY3).
Acknowledgment
The authors would like to thank National Science and Technology Council and Hon Hai Research Institute for the financial support under contract No. MOST 110-2622-8-A49-008-SB, and Ministry of Education (MOE), Taiwan, under and Higher Education Sprout Project of the National Yang Ming Chiao Tung University and Yushan Fellowship.
Disclosures
The authors declare no conflict of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. Y.-J. You, C. Wang, Y.-L. Lin, et al., “Ultrahigh-resolution optical coherence tomography at 1.3 µm central wavelength by using a supercontinuum source pumped by noise-like pulses,” Laser Phys. Lett. 13(2), 025101 (2016). [CrossRef]
2. G. Zhou, X. Zhou, Y. Song, et al., “Design of supercontinuum laser hyperspectral light detection and ranging (LiDAR) (SCLaHS LiDAR),” Int. J. Remote Sens. 42(10), 3731–3755 (2021). [CrossRef]
3. U. Minoni, G. Manili, S. Bettoni, et al., “Chromatic confocal setup for displacement measurement using a supercontinuum light source,” Opt. Laser Technol. 49, 91–94 (2013). [CrossRef]
4. I. Zorin, P. Gattinger, A. Ebner, et al., “Advances in mid-infrared spectroscopy enabled by supercontinuum laser sources,” Opt. Express 30(4), 5222–5254 (2022). [CrossRef]
5. A. D. Torre, R. Armand, M. Sinobad, et al., “Broadband and flat mid-infrared supercontinuum generation in a varying dispersion waveguide for parallel gas spectroscopy,” in Conference on Lasers and Electro-Optics (CLEO), pp. 1–2.
6. J. M. Dudley, G. Genty, and S. Coen, “Supercontinuum generation in photonic crystal fiber,” Rev. Mod. Phys. 78(4), 1135–1184 (2006). [CrossRef]
7. J.-M. Liu, Nonlinear Photonics (Cambridge University Press, 2022), Chap. 19.
8. D. Jain, R. Sidharthan, P. M. Moselund, et al., “Record power, ultra-broadband supercontinuum source based on highly GeO2 doped silica fiber,” Opt. Express 24(23), 26667–26677 (2016). [CrossRef]
9. X. Qi, S. Chen, Z. Li, et al., “High-power visible-enhanced all-fiber supercontinuum generation in a seven-core photonic crystal fiber pumped at 1016 nm,” Opt. Lett. 43(5), 1019–1022 (2018). [CrossRef]
10. L. Jiang, R. Song, J. He, et al., “714 W all-fiber supercontinuum generation from an ytterbium-doped fiber amplifier,” Opt. Laser Technol. 161, 109168 (2023). [CrossRef]
11. H. Zhang, F. Li, R. Liao, et al., “Supercontinuum generation of 314.7 W ranging from 390 to 2400 nm by tapered photonic crystal fiber,” Opt. Lett. 46(6), 1429–1432 (2021). [CrossRef]
12. A. Zaytsev, C.-H. Lin, Y.-J. You, et al., “Supercontinuum generation by noise-like pulses transmitted through normally dispersive standard single-mode fibers,” Opt. Express 21(13), 16056–16062 (2013). [CrossRef]
13. R.-Q. Xu, J.-R. Tian, and Y.-R. Song, “Noise-like pulses with a 14.5 fs spike generated in an Yb-doped fiber nonlinear amplifier,” Opt. Lett. 43(8), 1910–1913 (2018). [CrossRef]
14. C. Xu, J.-R. Tian, R. Xu, et al., “Generation of noise-like pulses with a 920 fs pedestal in a nonlinear Yb-doped fiber amplifier,” Opt. Express 27(2), 1208–1216 (2019). [CrossRef]
15. A. Turnali, S. Xu, and M. Y. Sander, “Noise-like pulse generation and amplification from soliton pulses,” Opt. Express 30(9), 13977–13984 (2022). [CrossRef]
16. M. Horowitz, Y. Barad, and Y. Silberberg, “Noiselike pulses with a broadband spectrum generated from an erbium-doped fiber laser,” Opt. Lett. 22(11), 799–801 (1997). [CrossRef]
17. S.-S. Lin, S.-K. Hwang, and J.-M. Liu, “Supercontinuum generation in highly nonlinear fibers using amplified noise-like optical pulses,” Opt. Express 22(4), 4152–4160 (2014). [CrossRef]
18. S.-S. Lin, S.-K. Hwang, and J.-M. Liu, “High-power noise-like pulse generation using a 1.56-µm all-fiber laser system,” Opt. Express 23(14), 18256–18268 (2015). [CrossRef]
19. J. P. Lauterio-Cruz, J. D. Filoteo-Razo, O. Pottiez, et al., “Numerical comparative study of supercontinuum generation in photonic crystal Fibers using noise-kike pulses and ultrashort pulses,” IEEE Photonics J. 11(5), 1–12 (2019). [CrossRef]
20. K.-Y. Chang, W.-C. Chen, S.-S. Lin, et al., “High-power, octave-spanning supercontinuum generation in highly nonlinear fibers using noise-like and well-defined pump optical pulses,” OSA Contin. 1(3), 851–863 (2018). [CrossRef]
21. K.-Y. Chang, R.-C. Wang, H.-C. Yu, et al., “Ultra-broadband supercontinuum covering a spectrum from visible to mid-infrared generated by high-power and ultrashort noise-like pulses,” Opt. Express 29(17), 26775–26786 (2021). [CrossRef]
22. S. Kobtsev, S. Kukarin, S. Smirnov, et al., “Cascaded SRS of single-and double-scale fiber laser pulses in long extra-cavity fiber,” Opt. Express 22(17), 20770–20775 (2014). [CrossRef]
23. K.-Y. Chang, G.-Y. Chen, H.-C. Yu, et al., “Generation of supercontinuum covering 520 nm to 2.25µm by noise-like laser pulses in an integrated all-fiber system,” Opt. Commun. 533, 129281 (2023). [CrossRef]
24. A. V. Bourdine, V. A. Burdin, and O. G. Morozov, “Algorithm for solving a system of coupled nonlinear Schrödinger equations by the split-step method to describe the evolution of a high-power femtosecond optical pulse in an optical polarization maintaining fiber,” Fibers 10(3), 22 (2022). [CrossRef]
25. K. Tai, A. Hasegawa, and N. Bekki, “Fission of optical solitons induced by stimulated Raman effect,” Opt. Lett. 13(5), 392–394 (1988). [CrossRef]
26. J. H. Lee, J. van. Howe, C. Xu, et al., “Soliton self-frequency shift: experimental demonstrations and applications,” IEEE J. Sel. Top. Quantum Electron. 14(3), 713–723 (2008). [CrossRef]
27. N. C. Panoiu, I. V. Mel’nikov, D. Mihalache, et al., “Soliton generation in optical fibers for a dual-frequency input,” Phys. Rev. E 60(4), 4868–4876 (1999). [CrossRef]
