We present a direct comparison between two types of femtosecond 2 µm sources used for seeding of an ultrafast thulium-doped fiber amplifier based on all-normal dispersion supercontinuum and soliton self-frequency shift. Both nonlinear effects were generated in microstructured silica fibers, pumped with low-power femtosecond pulses at 1.56 µm originating from an erbium-doped fiber laser. We performed a full characterization of both nonlinear processes, including their shot-to-shot stability, phase coherence, and relative intensity noise. The results revealed that the solitons show comparable performance to supercontinuum in terms of relative intensity noise and shot-to-shot stability, despite the anomalous dispersion regime. Both sources can be successfully used as seeds for Tm-doped fiber amplifiers as an alternative to Tm-doped oscillators. The results show that the sign of chromatic dispersion of the fiber is not crucial for obtaining a stable, high-quality, and low-noise spectral conversion process when pumped with sub-50 fs laser pulses.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Stable, coherent light sources that cover the near- and mid-IR range are in great demand nowadays due to their numerous potential applications in fundamental science and industry. Fiber lasers are well-established versatile solutions due to their robustness, compactness, and capability to generate ultrashort pulses using various mode-locking techniques. The most widely used include nonlinear polarization evolution [1,2], graphene [3,4], semiconductor saturable absorber mirrors (SESAMs) , carbon nanotubes , and other two-dimensional nanomaterials [7–9]. Recently, ultrafast lasers operating in the ∼2 µm range emerged as useful tools in many applications like multiphoton microscopy , absorption spectroscopy , microsurgery , mid-infrared frequency combs , optical parametric generation , and others. The typical way to address this wavelength range is to use Tm or Ho gain media combined with passive mode-locking. An alternative approach that is recently becoming more popular is to utilize spectral conversion that can occur in nonlinear optical fibers when pumped with near-infrared ultrashort pulses. For example, the phenomenon which is most widely investigated is supercontinuum (SC) generation that occurs when the narrowband source after passing through the dispersion-engineered fiber is widely broadened . Another phenomenon, the soliton self-frequency shift (SSFS), discovered in 1986 by Mitschke and Mollenauer , is a result of stimulated Raman scattering and allows for the efficient spectral conversion of the pump pulses towards longer wavelengths [17–20]. Frequency-shifted solitons find applications in various areas. The SSFS effect enables generation of non-standard wavelengths outside the gain bandwidth of typical laser media, e.g., between 1.1-1.5 and 1.6-1.7 µm, enabling biomedical applications like three-photon microscopy at 1700-1750 nm  or optical coherence tomography at 1300 nm . Frequency shifted solitons are also widely used for mid-infrared optical frequency comb generation [23–25].
The applications that involve 2 µm sources often require satisfactory noise properties, stability, and coherence of the ultrashort pulses. The incoherence or birefringence drift, usually caused by the nonlinear dynamics, limit the precision of metrological devices based on mode-locked lasers. Hence, theoretical and experimental studies concentrating on noise dynamics of nonlinear processes are conducted to make them a widely used platform for near- and mid-infrared sources [26–31]. Undoubtedly, the SC generated in an all-normal dispersion fiber (ANDi-SC) possesses excellent coherence and low-noise properties, which was confirmed by multiple studies performed by e.g., Heidt et al. [27–29]. It was also claimed that the red-shifted part of the SC generated in anomalous-dispersion fibers featured much worse coherence and shot-to-shot noise properties than ANDi, mostly due to the modulation instability [30,31]. It has been proven that the low coherence of solitons limits the application of SSFS in mid-infrared frequency comb generation . In  a comparison between ANDi and anomalous-dispersion SC has been performed, including dispersive Fourier transform (DFT) and coherence measurements, showing dramatically low stability and coherence of the anomalous-dispersion SC. However, the study was conducted using sub-picosecond pulses for pumping. It has been shown that even despite the anomalous dispersion of the nonlinear fiber, the generated solitons can maintain their coherence over a broad spectral range when pumped with sub-100 fs pulses . Both types of sources, based on SSFS and ANDi-SC, were already used for the seeding of Tm-doped fiber amplifiers (TDFAs) [35–40]. The noise dynamics of such systems are well described using the relative intensity noise (RIN) which is usually in the range of 0.2–0.7% in case of soliton seeding [35,36]. In  a high-power frequency comb seeded by Raman solitons presented by Coluccelli et al. characterized 0.3% RIN of the amplifier integrated over 1 Hz to 10 MHz which claimed to be 2 times higher than for the seed. In  Tm/Ho-doped fiber amplifier around 1.9 µm seeded by frequency-shifted Raman solitons exhibited 0.6% RIN, integrated over 1 Hz to 10 MHz, and was 4 times higher than the RIN of the input. Recently, studies on low-noise amplifying systems have led to a significant reduction of a RIN when it comes to ANDi-SC seeding. In  a low-noise Tm/Hm co-doped fiber amplifier at 2 µm seeded by ANDi SC has been presented demonstrating 0.07% RIN (10 Hz–20 MHz) and only 1.5 amplification factor. Also A. Rampur et al. studied the amplification of ANDi-SC in an all-fiber TDFA achieving 0.047% (10 Hz–10 MHz) . The amplifier provided pulses with 96 fs duration and 350 mW of average output power. Much higher powers were achieved with SSFSs as seed pulses for amplification [39,40]. In the context of being used as seed sources for amplifiers, frequency-shifted solitons seem to be more suitable, as they offer a smooth spectral shape with easy and wide tunability–the spectrum might be ideally matched to the deserved spectral range, e.g., to obtain the highest gain of the amplifier. However, the noise and shot-to-shot stability of SSFSs are still not yet thoroughly examined. The literature lacks a comprehensive, direct comparison of both types of nonlinear effects as seeds for Thulium-doped fiber amplifiers.
In this work, we present a comparison of the pulses generated via soliton self-frequency shift in anomalous dispersion and supercontinuum in normal dispersion fiber. We present the full characterization of the pulse-to-pulse coherence, shot-to-shot stability as well as relative intensity noise of the pulses emitted from both optical fibers possessing opposite signs of dispersion. Importantly, both optical fibers were manufactured using exactly the same technology and the same glass composition. The comparison was performed in the same laboratory conditions, ensuring the reliability of the measurements and comparability of the obtained results. Finally, we used both types of pulses as seeds for a Thulium-doped all-fiber amplifier. The performance of the amplifier was fully characterized in terms of the RIN and spectral/temporal phase of the output pulses using the frequency resolved optical gating (FROG) method. To reinforce the experimental investigations, we performed series of numerical simulations including the pump pulse intensity noise and analyzed the statistical properties of numerically generated spectra. The results show that both types of sources are excellent seeds for Tm-doped fiber amplifiers; however, the SSFS might outperform the ANDi-SC in terms of the RIN and shot-to-shot stability. We believe these findings are of high importance for developing broadband and tunable optical frequency comb sources (also in the mid-infrared), which often utilize nonlinear processes to extend their spectral coverage.
2. Experimental setup
2.1 Fiber design and fabrication
Both nonlinear fibers used in the experiment were germanium-doped microstructured silica fibers manufactured in the Laboratory of Optical Fiber Technology at Maria Curie-Skłodowska University in Lublin (Poland) using the stack and draw technique. The details on the design and characterization of the fibers were presented in  (SSFS) and  (ANDi). It is worth noting that exactly the same GeO2/SiO2 glass rod was used in both preforms to form the core of the fiber, with a GeO2 doping concentration of 18 mol%. The different dispersion properties were obtained by a slight modification of the filling factor of the small air holes surrounding the core. The dispersion profiles of the fibers are depicted in Fig. 1. Both fibers are highly birefringent and polarization-maintaining (PM). The SSFS fiber has a zero-dispersion wavelength (ZDW) of 1240 nm and a dispersion coefficient of approx. 26 ps/nm/km at 1560 nm wavelength. The core has an elliptical shape with major and minor axes of 3.9 and 2.53 µm, respectively. The calculated nonlinear coefficient γ equals 9.7 W−1km−1 at 1560 nm and 5.7 W−1km−1 at 2000 nm. The ANDi fiber has a less elliptical core (major axis of 3.37 µm and minor axis of 3.22 µm). The fiber maintains normal dispersion (negative chromatic dispersion coefficient) with a maximum value of −0.5 ps/km/nm at approx. 1700 nm wavelength. The γ coefficient amounts to 7.6 W−1km−1 at 1560 nm and 3.6 W−1km−1 at 2000 nm.
2.2 Experimental setup for SSFS and ANDi-SC generation
The experimental setup for the generation of SSFSs and ANDi-SC is presented in Fig. 2. An amplified Er-doped femtosecond source operating at 1.56 µm wavelength, emitting linearly polarized, 50 fs pulses at 125 MHz repetition rate was used as a pump for the nonlinear fibers. The laser contained a mode-locked oscillator followed by an Er-doped fiber amplifier, and has been already described in . The beam from the laser was coupled to the slow polarization axis of microstructured fibers via an aspheric lens. The input power was controlled by a variable attenuator made of a half-wave plate (HWP) and a polarization beam-splitter (PBS). A switchable mirror (SM) was used to change the direction of the laser beam either to SSFS or ANDi fiber. The outputs of both fibers were spliced to a short piece of PM single-mode fiber (PM-SMF) terminated with a fiber connector. It is worth noting that both fibers were placed on identical optomechanical stages and were kept next to each other on the optical table. Therefore, both experienced the same environmental perturbations (temperature fluctuations, mechanical vibrations, etc.). We believe that such an arrangement ensures the same experimental conditions for coherence, stability, and RIN measurements.
Examples of obtained SC and soliton optical spectra together with the numerical simulations at the outputs of both nonlinear fibers are depicted in Fig. 3. The supercontinuum was spanning up to 2200 nm, whereas the solitons could be tuned in the range of 1.60–2.01 µm by changing the input power from 78.0 mW–162.0 mW. The optical power stored in the long-wavelength part of the SC and the solitons was determined using a 1.75–2.25 µm bandpass filter (Thorlabs FB2000-500) which efficiently cuts out the unconverted pump spectrum. The power of the solitons was varying between 15.3 and 7.0 mW depending on the soliton wavelength, while the power of the SC was 15.0 mW at the maximum available laser power. The numerical simulations were performed using the adapted Generalized Nonlinear Schrodinger Equation solver . More details on the numerical simulations–in particular the analysis of spectra statistical properties–are given in Section 4. For both nonlinear fibers, we modeled pulse propagation over a distance specific to each fiber which was 0.5 m for SC and 1.5 m for SSFS. Both nonlinear effects were simulated assuming the same pump power as in the experiment. The obtained results are in good agreement with the experiment.
3. Experimental results
3.1 Pulse-to-pulse coherence
The coherence properties of ANDi-SC and the solitons were investigated using a fiber-based unequal-path Michelson interferometer based on a 50/50% splitter, similar to that shown in , but adapted to the repetition rate of the seed source and the wavelength of ∼2000nm. The length difference of the two interferometer arms was related to the distance between two consecutive pulses of the train. Therefore, interference was obtained only in case when the two pulses were in phase. The spectral interference pattern was recorded by an optical spectrum analyzer (Yokogawa AQ6375, OSA). To estimate the coherence, the fringe visibility parameter (V) was calculated using the following formula:
3.2 Shot-to-shot stability by dispersive Fourier transform
The dispersive Fourier transform technique allows us to investigate the spectral dependence of signal-to-noise ratio and measure the wavelengths correlations [31,46,47]. We have used a typical DFT setup similar to that presented in  with a fiber-based dispersive stretcher. In the current experiment, a 487-m-long segment of SM2000 fiber (FIBRAIN Ltd.) was used for time stretching. The time-wavelength mapping in the used piece of stretching fiber is shown in Fig. 5. The traces of subsequent pulses were recorded using a 6 GHz bandwidth oscilloscope (Agilent Infinuum DSO90604A) coupled with a 13 GHz bandwidth InGaAs photodetector (Discovery Semiconductors DSC2-50S).
Using 5000 traces for each measurement and time-wavelength mapping, we have obtained the data presented in Fig. 6 (ANDi-SC) and Fig. 7 (SSFS). We have transformed each trace to the wavelength domain (gray dots in the middle column in the figures) and compared them with spectra registered with OSA (solid blue lines). Due to the fact that the stretching fiber possess zero-dispersion wavelength around 1.3 µm and that the time-wavelength mapping is ambiguous for the spectral components near zero-dispersion wavelength, the DFT spectra are limited to ∼1.65 µm. The DFT spectra averaged over the ensembles (red lines) are almost identical to OSA-registered spectra, confirming the validity of time-wavelength mapping. It is worth noting that the spectral resolution of DFT spectra (δ) is limited by the oscilloscope bandwidth (B), chromatic dispersion of stretching fiber (DS) and stretching distance (z), δ = (BDSz)−1 . The shot-to-shot fluctuations (represented by gray dots) are very low for both sources. However, the dynamic range in the DFT measurement is clearly lower than in the optical spectrum analyzer. In consequence, the absolute value of the signal-to-noise ratio (SNR) will be limited. Nevertheless, the comparison of two nonlinear processes in terms of SNR can be made, as the experimental conditions were the same. For each wavelength, we have calculated the standard deviation of intensity, and SNR expressed as the ratio of the mean to the standard deviation. The high values of SNR correspond to all DFT traces focused in the narrow region around the mean value. For SC pulses, we obtained the maximal SNR equal to 90 around 1.85 µm, whereas for soliton pulses, the maximal SNR values exceeded 200 at central wavelength. It suggests that soliton pulses are of higher quality than SC pulses, as we expect that the noise floor level is the same in all cases due to the same experimental conditions.
Finally, we calculated the spectral correlation maps, which show the features typical for pulse wavelength jitter. When a pulse sweeps spectrally towards longer (shorter) wavelengths from its central wavelength position, then spectral intensities at different wavelengths change collectively: they increase at longer (shorter) wavelengths and decrease at shorter (longer) wavelengths. This behavior is depicted with positive and negative correlation, respectively. The contrast of the correlation map is limited by the noise since the random noise contribution dilutes the correlations. It also explains the difference between the autocorrelation map of SC spectrum and soliton spectra. The SC spectrum extends in a long-wavelength direction, which results in a long-wavelength tail on SNR [Fig. 6(c)] and a broad region of positive correlation [Fig. 6(a)]. The solitons’ spectra are narrower than the SC spectrum. Consequently, the SNR profiles (Fig. 7; right column) and regions of significant spectral correlations (Fig. 7; left column) are narrower in those cases.
3.3 Amplification of ANDi-SC and SSFS
Both types of pulses were compared as seed signals for a Tm-doped fiber amplifier. The schematic of the used all-fiber chirped pulse amplification (CPA) system is presented in Fig. 8. First, the pulses were stretched in a segment of PM dispersion compensating fiber (PM-DCF) with a dispersion coefficient of −38 ps/nm/km at 2000 nm wavelength. The stretcher was followed by a TDFA, which was bidirectionally pumped with a 1565 nm Erbium/Ytterbium-doped-fiber laser. After amplification, the pulses were compressed in a piece of standard PM-SMF. The length of the PM-SMF was optimized to each wavelength of the solitons or SC in order to obtain the shortest pulse duration at the output. The optical spectra of the output pulses were gathered using the OSA. The temporal properties of the pulses were investigated by FROG (FROG Scan Ultra2, Mesa Photonics).
The obtained results (pulse shapes, spectra, and FROG spectrograms) are summarized in Figs. 9 (ANDi-SC) and 10 (SSFS). Amplification of the ANDi-SC resulted in 92 fs pulses with 350 mW of average power which is almost identical to the result obtained by A. Rampur et al. . The FROG temporal profile indicates a relatively long tail behind the main pulse, suggesting uncompensated third-order dispersion (TOD). The optical spectrum is affected by the water vapor absorption lines present in air in the range between 1825 and 1940 nm.
It was already shown previously by A. Rampur et al.  that the long-wavelength part of the 1.55-µm-pumped ANDi-SC can serve as a low-noise seed for a Tm-doped fiber amplifier. However, the SSFS has one unquestionable advantage: broadband spectral tunability. The wavelength of the soliton can be precisely tuned to the desired value, e.g., determined by the requirements of the experiment or to match the maximum gain of the amplifier. Figure 10 shows the characterization of amplified solitons set at four different central wavelengths: 1870, 1950, 1970, and 1994 nm. The obtained output powers at these wavelengths were 365, 405, 370, 278 mW, respectively. The wavelength of 1870 nm [Figs. 10(a)–10(c)] was chosen intentionally to match the spectral range covered by the ANDi-SC, and directly compare both types of seed pulses. The pulse duration of the amplified 1870 nm soliton was slightly shorter (88 fs), and the obtained power was also higher (365 mW). The flatness of the spectral phase is comparable for both types of seeding and is mainly limited by the pulse compression method (standard optical fiber). The shortest pulse of 80 fs was obtained with the soliton centered at 1950 nm [Figs. 10(d)–10(f)]. The pulse at this wavelength is also the least affected with TOD and exhibits a nearly flat temporal and spectral phase. Further tuning of the solitons towards longer wavelengths slightly degrades the pulse duration. A 95 fs short pulse was obtained at 1970 nm [Figs. 10(g)–10(i)], while at 1994 nm the duration was equal to 107 fs [Figs. 10(j)–10(l)].
3.4 Relative intensity noise
The RIN of the investigated sources was determined following a procedure similar to that presented in . The signals were detected using a low-noise photodetector (Thorlabs PDA10D2) and recorded by an oscilloscope (Rohde-Schwarz RTA4000). The measured signal was maintained at the level of 1 V at the oscilloscope to avoid detector saturation. The recorded signals were then converted to the frequency domain using Fourier-transformation and normalized by the average value. The results determined the power spectral density (PSD) over the frequency range 10 Hz–500 kHz. To increase the amplitude resolution the oversampling method was used together with a proper analog and digital filtering (anti-aliasing filter at the input of the oscilloscope and digital filter during oversampled signal calculation). Moreover, to reduce the noise 500 FFTs spectra have been averaged. In the last step, the integrated RIN rms is calculated by integrating PSD over the frequency range 10 Hz – 500 kHz. The procedure has been performed for the femtosecond 1550 nm laser, the output of both nonlinear fibers, the Tm-doped fiber amplifier seeded by ANDi-SC and solitons, the CW 1565 nm pump laser, and the CW 980 nm laser diode used as pump for the femtosecond oscillator. The results, including intensity noise for the solitons centered at different wavelengths and the SC, are presented in Figs. 11(a) and 11(b). As we can observe, the RIN level of the femtosecond 1550 nm source is not much above the noise floor, and its integrated RIN equals 0.022%. RIN values for the solitons vary between 0.023% for the soliton at 1870 nm and 0.077% for the 1930 nm. For the SC, however, the measured RIN is at the level of 0.058%, which is higher than for the soliton centered at the same central wavelength. Clearly, as it is presented in Fig. 11 the solitons can exhibit better noise properties in terms of relative intensity noise even comparing to recently measured RIN of ANDi-SC by A. Rampur et al.  which corresponded to 0.045%. Figures 11(c) and 11(d) present the results of the RIN measurements for the Thulium amplifier and its pump (1565 nm CW fiber laser). Here, the comparison of the intensity noise has been performed for the solitons centered at two emission wavelengths: 1870 and 1950 nm and for the SC. We can easily notice the increase of the RIN for the amplified signals, which indicates that the Tm-amplifier adds additional noise to the setup. We emphasize that the Tm-amplifier was not optimized for obtaining the minimum possible RIN–the goal of the study was to compare both types of seeding in the same experimental conditions. Therefore for all measured signals the amplifier pump was set to the same power. The integrated RIN of the amplified 1870 nm solitons and the SC is very comparable, and amounts to 0.153% and 0.128%, respectively, and is only 3 and 6 times higher than the intensity fluctuations of the nonlinear processes. The highest level of the intensity noise observed in our system corresponds to the amplified soliton at 1950 nm and is equal to 0.284%, which is typical for amplification systems (0.047%–0.7% reported [30–33]). However, we underline that the minimization of the amplifier RIN is not within the scope of this work.
The noise properties of the measured signals are determined by the interplay of various effects that need to be considered when interpreting the results presented in Fig. 11. First, all traces feature noise peaks < 1 kHz which can be assigned to the free-space coupling to nonlinear fibers. In the frequency range of 100 Hz–1 kHz both SSFS and ANDi RIN are dominated by the RIN of the 1550 nm femtosecond laser. The results of the 980 nm CW pumping diode (used as pump for the oscillator in the 1550 nm femtosecond source) presented in Fig. 12 indicate the presence of several noise peaks later observable in the output of the 1550 nm femtosecond source (in the range 100 Hz–1 kHz and also from 3 kHz–100 kHz) which originates from the used laser diode driver. These noise peaks are first suppressed in the seed laser due to the mode-locking operation  and then transferred and amplified in all recorded spectra. When considering the RIN results for the nonlinear processes it is noticeable that solitons centered at different emission wavelengths are characterized by different amplitude noise, which has already been observed and reported to be caused by the complex dynamics of the soliton propagation inside the microstructured silica fiber . In case of different emission wavelength, however, additional effects may be taken into account. One can notice that the responsivity of the chosen detector varies for different signal wavelengths which would generate the variation of the RIN results for the nonlinear processes. However, the maximal difference in the detector responsivity for the maximum wavelength range stated in this analysis is about 7.8%, therefore, it should not imply the discrepancy of the signals. Another factor may originate from the wavelength-dependent behavior of the gain profile of the Tm-doped fiber in the amplifier. This would explain the highest noise peak at 300 kHz for the 1950 nm amplified soliton. Nevertheless, the obtained results indicate a low-noise performance of both processes and comparable properties at the same central wavelengths (1870 nm soliton and ANDi-SC). Therefore, we state that seeding of amplifiers with solitons may provide comparable performance to ANDi-SC seeding, with the additional property of precise spectral tuning possibility.
4. Numerical simulations
Finally, to support the conclusions taken from experimental research, we performed numerical simulations of both considered nonlinear processes. We extended the Generalized Nonlinear Schrödinger Equations solver  to include wavelength-dependent attenuation , wavelength-dependent effective mode area , and Raman response function .
In the simulations, we used the measured chromatic dispersions of both fibers (Fig. 1). The propagation distances were 0.5 m for SC and 1.5 m for SSFS (as in experiments). To perform the statistical analysis, we ran 400 simulations introducing the noise to the initial pulse profile. We used the pump pulse profile acquired with the FROG technique. Then – for each run–we assigned the random pump intensity, which was drawn from a normal distribution. The mean value of the distribution corresponds to the measured pulse power and the standard deviation corresponds to the measured pump pulse RIN [0.022% as given in the previous section and shown in Fig. 11(b)]. Finally, we added also quantum noise with a one-photon per mode approach . The spectral resolution of the frequency grid is limited by the number of discretization points in time/frequency windows. In our simulations we used 214 points what results in the spectral resolution of approximately 0.18 nm–for soliton simulation and 1.1 nm for SC simulation in the interesting spectral range. In the experiments, we used DFT to acquire spectra of single pulses. The spectral resolution in DFT measurements is wavelength dependent and related to a length of stretching fiber, a source repetition rate, and oscilloscope’s and photodetector’s bandwidths . In order to imitate the limitations of resolution in the DFT technique, we performed spectral averaging of numerically obtained output spectra. In the next step, we used the processed spectra to calculate the mean spectra, the spectral correlation maps, and signal-to-noise ratio.
The numerical results presented in Figs. 13 (ANDi-SC) and 14 (solitons) can be referred to the experimental results presented in Figs. 6 and 7. The calculated SNRs are much higher than SNRs obtained with DFT measurements. This is connected to a relatively high noise level of the detector used in the DFT setup. The dynamic range of the detector is approximately 20 dB, as can be noticed in Figs. 6 and 7, which corresponds to SNR = 100. Nevertheless, the simulations confirm the experimental observation, that maxima of SNR for solitons are significantly higher than SNR for ANDi-SC. On the other hand, the high SNR (>1000) range is spectrally limited for solitons and spectrally broad for ANDi-SC. Finally, the relatively low noise in the numerical experiment allows to obtain high contrast of autocorrelation maps. As in the experiment, they reveal wavelength jitter of the main peaks. Additionally, they show wavelength jitter for smaller peaks in the 1.7 µm–1.8 µm range, which are below the floor noise level of the DFT measurements.
To conclude, a full comparison between two nonlinear effects–supercontinuum generated in all-normal dispersion fiber and SSFS has been presented. The experiment, conducted in the same environmental conditions, tested both fibers in terms of shot-to-shot stability, coherence, and relative intensity noise. The results revealed high phase coherence, stability of the generated pulses, and low-noise properties for both nonlinear processes, yet with slightly better performance for the SSFS at each stage, despite the anomalous dispersion of the nonlinear fiber. The experimental results were supported with numerical simulations, which provided compatible conclusions. The applicability of the generated pulses as seeds for Tm-doped fiber amplifiers has also been verified by amplifying them in an all-fiber amplification setup, which confirmed that both types of sources provide low RIN and high-quality output pulses. However, we believe that the spectral tuning property of the SSFS process makes it a more flexible seed for Tm-doped fiber amplifiers, allowing for precise matching the gain peak of the used amplifier.
Fundacja na rzecz Nauki Polskiej (First TEAM/2017-4/39); Narodowe Centrum Badań i Rozwoju (POIR.04.01.01-00-0037/17).
The authors declare no conflicts of interest.
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon request.
1. P. Wang, C. Bao, B. Fu, X. Xiao, P. Grelu, and C. Yang, “Generation of wavelength-tunable soliton molecules in a 2-µm ultrafast all-fiber laser based on nonlinear polarization evolution,” Opt. Lett. 41(10), 2254–2257 (2016). [CrossRef]
2. Y. Tang, A. Chong, and F. W. Wise, “Generation of 8 nJ pulses from a normal-dispersion thulium fiber laser,” Opt. Lett. 40(10), 2361–2364 (2015). [CrossRef]
3. H. Jeong, S. Y. Choi, M. H. Kim, F. Rotermund, Y.-H. Cha, D.-Y. Jeong, S. B. Lee, K. Lee, and D.-I. Yeom, “All-fiber Tm-doped soliton laser oscillator with 6 nJ pulse energy based on evanescent field interaction with monoloayer graphene saturable absorber,” Opt. Express 24(13), 14152–14158 (2016). [CrossRef]
4. B. Fu, Y. Hua, X. Xiao, H. Zhu, Z. Sun, and C. Yang, “Broadband Graphene Saturable Absorber for Pulsed Fiber Lasers at 1, 1.5, and 2 µm,” IEEE J. Sel. Top. Quantum Electron. 20(5), 411–415 (2014). [CrossRef]
5. J. Jiang, C. Mohr, J. Bethge, A. Mills, W. Mefford, I. Hartl, M. E. Fermann, C.-C. Lee, S. Suzuki, T. R. Schibli, N. Leindecker, K. L. Vodopyanov, and P. G. Schunemann, “500 MHz, 58fs highly coherent Tm fiber soliton laser,” in 2012 Conference on Lasers and Electro-Optics (CLEO) (2012), pp. 1–2.
6. J. Wang, X. Liang, G. Hu, Z. Zheng, S. Lin, D. Ouyang, X. Wu, P. Yan, S. Ruan, Z. Sun, and T. Hasan, “152 fs nanotube-mode-locked thulium-doped all-fiber laser,” Sci. Rep. 6(1), 28885 (2016). [CrossRef]
7. Y. Song, Z. Liang, X. Jiang, Y. Chen, Z. Li, L. Lu, Y. Ge, K. Wang, J. Zheng, S. Lu, J. Ji, and H. Zhang, “Few-layer antimonene decorated microfiber: ultra-short pulse generation and all-optical thresholding with enhanced long term stability,” 2D Mater. 4(4), 045010 (2017). [CrossRef]
8. B. Guo, S.-H. Wang, Z.-X. Wu, Z.-X. Wang, D.-H. Wang, H. Huang, F. Zhang, Y.-Q. Ge, and H. Zhang, “Sub-200 fs soliton mode-locked fiber laser based on bismuthene saturable absorber,” Opt. Express 26(18), 22750–22760 (2018). [CrossRef]
9. T. Jiang, K. Yin, C. Wang, J. You, H. Ouyang, R. Miao, C. Zhang, K. Wei, H. Li, H. Chen, R. Zhang, X. Zheng, Z. Xu, X. Cheng, and H. Zhang, “Ultrafast fiber lasers mode-locked by two-dimensional materials: review and prospect,” Photonics Res. 8(1), 78–90 (2020). [CrossRef]
10. W. Liu, W. Liu, C. Li, H.-Y. Chung, H.-Y. Chung, S.-H. Chia, S.-H. Chia, Z. Zhang, F. X. Kärtner, F. X. Kärtner, F. X. Kärtner, G. Chang, and G. Chang, “Widely tunable ultrafast sources for multi-photon microscopy,” in Conference on Lasers and Electro-Optics (2016), Paper SM2I.7 (Optical Society of America, 2016), p. SM2I.7.
11. P. Kadwani, J. Chia, F. Altal, R. A. Sims, C. Willis, L. Shah, D. Killinger, and M. C. Richardson, “Atmospheric absorption spectroscopy using Tm: fiber sources around two microns,” in Atmospheric and Oceanic Propagation of Electromagnetic Waves V (International Society for Optics and Photonics, 2011), 7924, p. 79240L.
12. E. Passacantilli, G. Anichini, G. Lapadula, M. Salvati, J. Lenzi, and A. Santoro, “Assessment of the utility of the 2-µ thulium laser in surgical removal of intracranial meningiomas,” Lasers Surg. Med. 45(3), 148–154 (2013). [CrossRef]
13. F. Adler and S. A. Diddams, “High-power, hybrid Er:fiber/Tm:fiber frequency comb source in the 2 µm wavelength region,” Opt. Lett. 37(9), 1400–1402 (2012). [CrossRef]
14. N. Leindecker, A. Marandi, R. L. Byer, K. L. Vodopyanov, J. Jiang, I. Hartl, M. Fermann, and P. G. Schunemann, “Octave-spanning ultrafast OPO with 2.6-6.1 µm instantaneous bandwidth pumped by femtosecond Tm-fiber laser,” Opt. Express 20(7), 7046–7053 (2012). [CrossRef]
15. J. K. Ranka, R. S. Windeler, and A. J. Stentz, “Visible continuum generation in air–silica microstructure optical fibers with anomalous dispersion at 800 nm,” Opt. Lett. 25(1), 25–27 (2000). [CrossRef]
16. F. M. Mitschke and L. F. Mollenauer, “Discovery of the soliton self-frequency shift,” Opt. Lett. 11(10), 659–661 (1986). [CrossRef]
17. N. Nishizawa and T. Goto, “Widely wavelength-tunable ultrashort pulse generation using polarization maintaining optical fibers,” IEEE J. Sel. Top. Quantum Electron. 7(4), 518–524 (2001). [CrossRef]
18. J. Takayanagi, T. Sugiura, M. Yoshida, and N. Nishizawa, “1.0–1.7-µm Wavelength-Tunable Ultrashort-Pulse Generation Using Femtosecond Yb-Doped Fiber Laser and Photonic Crystal Fiber,” IEEE Photonics Technol. Lett. 18(21), 2284–2286 (2006). [CrossRef]
19. G. Soboń, T. Martynkien, K. Tarnowski, P. Mergo, and J. Sotor, “Generation of sub-100 fs pulses tunable from 1700 to 2100 nm from a compact frequency-shifted Er-fiber laser,” Photonics Res. 5(3), 151–155 (2017). [CrossRef]
20. Y. Tang, L. G. Wright, K. Charan, T. Wang, C. Xu, and F. W. Wise, “Generation of intense 100 fs solitons tunable from 2 to 4.3 µm in fluoride fiber,” Optica 3(9), 948–951 (2016). [CrossRef]
21. D. R. Miller, J. W. Jarrett, A. M. Hassan, and A. K. Dunn, “Deep tissue imaging with multiphoton fluorescence microscopy,” Curr. Opin. Biomed. Eng. 4, 32–39 (2017). [CrossRef]
22. K. Sumimura, Y. Genda, T. Ohta, K. Itoh, and N. Nishizawa, “Quasi-supercontinuum generation using 1.06µm ultrashort-pulse laser system for ultrahigh-resolution optical-coherence tomography,” Opt. Lett. 35(21), 3631–3633 (2010). [CrossRef]
23. K. F. Lee, C. J. Hensley, P. G. Schunemann, and M. E. Fermann, “Midinfrared frequency comb by difference frequency of erbium and thulium fiber lasers in orientation-patterned gallium phosphide,” Opt. Express 25(15), 17411–17416 (2017). [CrossRef]
24. G. Soboń, T. Martynkien, P. Mergo, L. Rutkowski, and A. Foltynowicz, “High-power frequency comb source tunable from 2.7 to 4.2 µm based on difference frequency generation pumped by an Yb-doped fiber laser,” Opt. Lett. 42(9), 1748–1751 (2017). [CrossRef]
25. K. Krzempek, D. Tomaszewska, A. Głuszek, T. Martynkien, P. Mergo, J. Sotor, A. Foltynowicz, and G. Soboń, “Stabilized all-fiber source for generation of tunable broadband fCEO-free mid-IR frequency comb in the 7–9 µm range,” Opt. Express 27(26), 37435–37445 (2019). [CrossRef]
26. J. M. Dudley, G. Genty, and S. Coen, “Supercontinuum generation in photonic crystal fiber,” Rev. Mod. Phys. 78(4), 1135–1184 (2006). [CrossRef]
27. A. M. Heidt, “Pulse preserving flat-top supercontinuum generation in all-normal dispersion photonic crystal fibers,” J. Opt. Soc. Am. B 27(3), 550–559 (2010). [CrossRef]
28. A. M. Heidt, J. S. Feehan, J. H. V. Price, and T. Feurer, “Limits of coherent supercontinuum generation in normal dispersion fibers,” J. Opt. Soc. Am. B 34(4), 764–775 (2017). [CrossRef]
29. A. M. Heidt, A. Hartung, G. W. Bosman, P. Krok, E. G. Rohwer, H. Schwoerer, and H. Bartelt, “Coherent octave spanning near-infrared and visible supercontinuum generation in all-normal dispersion photonic crystal fibers,” Opt. Express 19(4), 3775–3787 (2011). [CrossRef]
30. T. Godin, B. Wetzel, T. Sylvestre, L. Larger, A. Kudlinski, A. Mussot, A. B. Salem, M. Zghal, G. Genty, F. Dias, and J. M. Dudley, “Real time noise and wavelength correlations in octave-spanning supercontinuum generation,” Opt. Express 21(15), 18452–18460 (2013). [CrossRef]
31. B. Wetzel, A. Stefani, L. Larger, P. A. Lacourt, J. M. Merolla, T. Sylvestre, A. Kudlinski, A. Mussot, G. Genty, F. Dias, and J. M. Dudley, “Real-time full bandwidth measurement of spectral noise in supercontinuum generation,” Sci. Rep. 2(1), 882 (2012). [CrossRef]
32. T. W. Neely, T. A. Johnson, and S. A. Diddams, “High-power broadband laser source tunable from 3.0 µm to 4.4 µmbased on a femtosecond Yb:fiber oscillator,” Opt. Lett. 36(20), 4020–4022 (2011). [CrossRef]
33. M. Klimczak, G. Soboń, R. Kasztelanic, K. M. Abramski, and R. Buczyński, “Direct comparison of shot-to-shot noise performance of all normal dispersion and anomalous dispersion supercontinuum pumped with sub-picosecond pulse fiber-based laser,” Sci. Rep. 6(1), 19284 (2016). [CrossRef]
34. G. Soboń, T. Martynkien, D. Tomaszewska, K. Tarnowski, P. Mergo, and J. Sotor, “All-in-fiber amplification and compression of coherent frequency-shifted solitons tunable in the 1800-2000nm range,” Photonics Res. 6(5), 368–372 (2018). [CrossRef]
35. N. Coluccelli, M. Cassinerio, A. Gambetta, P. Laporta, and G. Galzerano, “High-power frequency comb in the range of 2 µm based on a holmium fiber amplifier seeded by wavelength-shifted Raman solitons from an erbium-fiber laser,” Opt. Lett. 39(6), 1661–1664 (2014). [CrossRef]
36. N. Coluccelli, M. Cassinerio, P. Laporta, and G. Galzerano, “Single-clad Tm–Ho:fiber amplifier for high-power sub-100-fs pulses around 1.9 µm,” Opt. Lett. 38(15), 2757–2759 (2013). [CrossRef]
37. A. M. Heidt, J. Modupeh Hodasi, A. Rampur, D.-M. Spangenberg, M. Ryser, M. Klimczak, and T. Feurer, “Low noise all-fiber amplification of a coherent supercontinuum at 2 µm and its limits imposed by polarization noise,” Sci. Rep. 10(1), 16734 (2020). [CrossRef]
38. A. Rampur, A. Rampur, Y. Stepanenko, G. Stępniewski, T. Kardaś, D. Dobrakowski, D. Dobrakowski, D.-M. Spangenberg, T. Feurer, A. Heidt, M. Klimczak, and M. Klimczak, “Ultra low-noise coherent supercontinuum amplification and compression below 100 fs in an all-fiber polarization-maintaining thulium fiber amplifier,” Opt. Express 27(24), 35041–35051 (2019). [CrossRef]
39. F. Tan, H. Shi, R. Sun, P. Wang, and P. Wang, “1 µJ, sub-300 fs pulse generation from a compact thulium-doped chirped pulse amplifier seeded by Raman shifted erbium-doped fiber laser,” Opt. Express 24(20), 22461–22468 (2016). [CrossRef]
40. R. A. Sims, P. Kadwani, A. S. L. Shah, and M. Richardson, “1 µJ, sub-500 fs chirped pulse amplification in a Tm-doped fiber system,” Opt. Lett. 38(2), 121–123 (2013). [CrossRef]
41. K. Tarnowski, T. Martynkien, P. Mergo, K. Poturaj, A. Anuszkiewicz, P. Béjot, F. Billard, O. Faucher, B. Kibler, and W. Urbanczyk, “Polarized all-normal dispersion supercontinuum reaching 2.5 µm generated in a birefringent microstructured silica fiber,” Opt. Express 25(22), 27452–27463 (2017). [CrossRef]
42. A. Głuszek, F. Senna Vieira, A. Hudzikowski, A. Wąż, J. Sotor, A. Foltynowicz, and G. Soboń, “Compact mode-locked Er-doped fiber laser for broadband cavity-enhanced spectroscopy,” Appl. Phys. B 126(8), 137 (2020). [CrossRef]
43. J. C. Travers, M. H. Frosz, and J. M. Dudley, “Nonlinear fibre optics overview,” in Supercontinuum Generation in Optical Fibers, J. M. Dudley and J. R. Taylor, eds. (Cambridge University Press, 2010), pp. 32–51.
44. A. F. J. Runge, C. Aguergaray, N. G. R. Broderick, and M. Erkintalo, “Coherence and shot-to-shot spectral fluctuations in noise-like ultrafast fiber lasers,” Opt. Lett. 38(21), 4327–4330 (2013). [CrossRef]
45. S. Kim, J. Park, S. Han, Y.-J. Kim, and S.-W. Kim, “Coherent supercontinuum generation using Er-doped fiber laser of hybrid mode-locking,” Opt. Lett. 39(10), 2986–2989 (2014). [CrossRef]
46. D. R. Solli, G. Herink, B. Jalali, and C. Ropers, “Fluctuations and correlations in modulation instability,” Nat. Photonics 6(7), 463–468 (2012). [CrossRef]
47. K. Goda and B. Jalali, “Dispersive Fourier transformation for fast continuous single-shot measurements,” Nat. Photonics 7(2), 102–112 (2013). [CrossRef]
48. A. S. Mayer, A. S. Mayer, W. Grosinger, J. Fellinger, G. Winkler, L. W. Perner, S. Droste, S. H. Salman, C. Li, C. M. Heyl, C. M. Heyl, I. Hartl, O. H. Heckl, and O. H. Heckl, “Flexible all-PM NALM Yb:fiber laser design for frequency comb applications: operation regimes and their noise properties,” Opt. Express 28(13), 18946–18968 (2020). [CrossRef]
49. H. Tünnermann, J. Neumann, D. Kracht, and P. Weßels, “Gain dynamics and refractive index changes in fiber amplifiers: a frequency domain approach,” Opt. Express 20(12), 13539–13550 (2012). [CrossRef]
50. K. L. Corwin, N. R. Newbury, J. M. Dudley, S. Coen, S. A. Diddams, B. R. Washburn, K. Weber, and R. S. Windeler, “Fundamental amplitude noise limitations to supercontinuum spectra generated in a microstructured fiber,” Appl. Phys. B 77(2-3), 269–277 (2003). [CrossRef]
51. O. Humbach, H. Fabian, U. Grzesik, U. Haken, and W. Heitmann, “Analysis of OH absorption bands in synthetic silica,” J. Non-Cryst. Solids 203, 19–26 (1996). [CrossRef]
52. J. Lægsgaard, “Mode profile dispersion in the generalized nonlinear Schrödinger equation,” Opt. Express 15(24), 16110–16123 (2007). [CrossRef]
53. Q. Lin and G. P. Agrawal, “Raman response function for silica fibers,” Opt. Lett. 31(21), 3086–3088 (2006). [CrossRef]