This paper presents an all fiber high power picosecond laser at 1016 nm in master oscillator power amplifier (MOPA) configuration. A direct amplification of this seed source encounters obvious gain competition with amplified spontaneous emission (ASE) at ~1030 nm, leading to a seriously reduced amplification efficiency. To suppress the ASE and improve the amplification efficiency, we experimentally investigate the influence of the gain fiber length and the residual ASE on the perforemance of the 1016 nm amplifier. The optimized 1016 nm MOPA laser exhibits an average power of 50 W and an optical conversion efficiency of 53%.
© 2016 Optical Society of America
In recent years, the short-wavelength continuous-wave (CW) and pulsed Yb-doped fiber lasers (YDFLs) at 1010-1020 nm have achieved a great progress. The CW short-wavelength YDFLs have been successfully used in a variety of interesting applications. For example, IPG reported a 10 kW single mode YDFL pumped by 1018 nm fiber lasers in 2009 , which is called tandem pumping. Besides, laser cooling of Yb-doped solids by means of anti-Stokes fluorescence requires lasers at 1010~1020 nm [2–4]. Furthermore, laser cooling and trapping of neutral mercury atoms requires lasers at 253.7 nm, which can be generated by frequency quadrupling 1014.8 nm lasers [5–8]. However, it is challenging to make the fiber laser operating at the wavelength range from 1010 nm to 1020 nm, because this range is far away from the peak of the regular emission spectrum of Yb-doped silica fibers. Typically, the highest gain of Yb-doped silica fiber is at 1031 nm when pumped at 975 nm. Owing to the much smaller emission cross-section at 1010-1020 nm band than 1030 nm and the increasing absorption cross-section towards shorter wavelengths , the net gain of Yb-doped silica fibers at 1010-1020 nm is relatively small . Therefore, for a short-wavelength YDFL, gain competition between the laser and amplified spontaneous emission (ASE) may cause the decrease of laser efficiency, or even prevent the oscillation of the target laser . To raise the output power and efficiency of the short-wavelength CW lasers and amplifiers, some methods have been exploited. By optimization of the Yb-doped fiber (YDF) length and geometry, a record output power of 476 W 1018 nm CW fiber laser and a single-frequency 1014.8 nm CW fiber laser with 19.3 W are obtained respectively in  and . Based on the YDF cooled to liquid nitrogen temperature, a single-frequency CW fiber laser at 1015 nm with 10 W output power is presented in . Employing a piece of short Yb-doped phosphate fiber or a piece of Yb-doped phosphosilicate fiber, a more than 1.06 W MOPA fiber laser at 1014 nm and a 22.8 W CW fiber laser at 1018 nm are achieved in  and  respectively.
Especially, deep space optical communication uplinks need high power and high energy short-wavelength pulsed laser beacons [15–19]. For improved signal-noise ratio (SNR) in pointing/tracking and reduced uplink laser power requirement, it is desirable to operate at shorter wavelengths near 1000 nm, where near-IR silicon avalanche photo-detectors (Si-APD) have improved (≥3X) spectral responsivity compared to 1064 nm. In contrast to CW short-wavelength fiber lasers, pulsed short-wavelength fiber lasers are more difficult to realize because the continuous ASE consumes the upper-level population all along and accumulates along the MOPA chain whereas the signal pulses only consume the upper-level population within its duration time. The short-wavelength pulsed YDFLs at 1010-1020 nm are much more difficult to be obtained. The Fibertek Inc. has achieved a 1024 nm pulsed fiber laser with 300 W average power and 5 kW peak power and a 1030 nm pulsed fiber laser with 500 W average power and 9 kW peak power in MOPA configuration, and is projecting an efficient pulsed MOPA operation at 1018~1020 nm in future . Besides, as the seed of a high energy cryogenically cooled Yb:YLF laser,  presents a 450 mW femtosecond fiber laser at 1018 nm and  presents a 87 W pulse laser operating at 1018 nm with 1 ns pulse width and 4.9 μJ pulse energy. However, the laser in  is not an all fiber structure and a stiff rod type Yb-doped large-pitch fiber is used for power amplification. In , reseachers have attained a 1011 nm pulsed YDFL with 10.5 mW output power. Therefore, all fiber pulsed fiber lasers at shorter wavelengths below 1020 nm with high power and high energy have not been demonstrated so far.
In this paper, we investigate the influence of YDF length and the residual ASE on the performance of the amplifier. By optimizing above parameters, an all fiber picosecond pulsed 1016 nm laser in a four-stage MOPA configuration exhibits an average power of 50 W, which is much higher than previous results [20, 22] by 2~3 orders of magnitude, with an optical conversion efficiency of 53%. The pulse width is 120 ps and the repetition rate is 27.03 MHz. This is to our knowledge the highest power in pulsed operation of all fiber YDFL at the wavelength below 1020 nm so far.
2. Experimental setup
The emergence of new saturable absorbers (SAs) makes the variety of mode-locked lasers [23–25]. Especially, the semiconductor saturable absorber mirrors (SESAMs) as the maturest SAs are widely used in commercial laser system. Figure 1 illustrates the experimental setup, which is constructed in a four-stage MOPA configuration comprising a master oscillator and three cascaded amplifiers. The master oscillator is a SESAM mode-locked laser, as shown in Fig. 1(a), consisting of a fiber Bragg grating (FBG), a SESAM, an optical isolator and a piece of 0.23 m long 4/125 μm single mode YDF (Lekki, YDF1200-4/125). The YDF has a nominal absorption coefficient of 250 dB/m at 975 nm. The SESAM is utilized to mode-lock the laser and also acts as the total reflection cavity mirror. The modulation depth, saturation fluence and relaxation time constant of the SESAM are 13%, 55 μJ/cm2 and 9 ps, respectively. The FBG exhibits reflectivity of 10% with a spectral bandwidth of 0.048 nm at the center wavelength of 1016 nm and acts as the output cavity mirror. A 1:99 optical coupler is inserted before the subsequent amplifier chain to monitor the seed wavelength and power.
The first-stage amplifier AMP1 is constructed with a piece of 0.23 m long single mode YDF (Lekki, YDF1200-4/125), same as the master oscillator, pumped by a 500 mW fiber pigtailed 976 nm laser diode (LD). The second-stage amplifier AMP2 is a double-clad Yb-doped fiber amplifier (YDFA), which is constructed with a piece of 2 m long YDF (Nufern, LMA-YDF-10/125, the fiber core/inner cladding 10/125 μm; NA 0.08/0.46; peak cladding absorption 5 dB/m @ 975 nm), pumped by a 25 W fiber pigtailed 976 nm LD via a (2 + 1)*1 fiber combiner. The third-stage amplifier AMP3 is a large-mode-area (LMA) double-clad YDFA, employing two 50 W fiber pigtailed 976 nm LDs, a (2 + 1)*1 fiber combiner, a 2 m long YDF (Nufern, LMA-YDF-25/250, the fiber core/inner cladding 25/250 μm; NA 0.08/0.46; peak cladding absorption 5 dB/m @ 975 nm), and a section of matched passive double-clad fiber (Nufern, LMA-GDF-25/250). High index gel is laid on the outer cladding stripped GDF at the splicing point of the YDF and the GDF to dump out not only the residual pump light, but also the signal light propagating in the fiber cladding. The LMA YDF is coiled onto an aluminum slab and water-cooled to 20 °C to avoid thermal damage. The output end of the GDF is angle cleaved at 8° to avoid optical feedback (~4% Fresnel reflection). All the amplifier stages in this MOPA configuration are forward pumped to avoid LDs destruction by insufficient isolation of high peak power signal lasers.
Moreover, because of the broad gain bandwidth of Yb-doped silica fibers, if the incident signal of each amplifier includes residual ASE, the ASE amplification will lead to energy-shifting to longer wavelengths and eventually suppress the 1016 nm signal amplification. It may also cause parasitic lasing . To remove the residual ASE in the incident signal, an ASE filter is positioned before each amplifier stage. Figure 1(c) shows the transmission spectrum of the ASE filters. The ASE filters are short-wave-pass edge filters and the cut-on wavelength is 1020 nm. An optical isolator (ISO) is also inserted before each amplifier stage to block possible counter-propagating light from the subsequent amplifier.The pulse width is measured by a high speed photodetector (45 GHz bandwidth) and a 60 GHz digital serial analyzer sampling oscilloscope.
3. Results and discussion
3.1 Performance of the SESAM mode-locked laser
The SESAM mode-locked laser achieves self-starting mode-locked operation by increasing the pump power above a threshold without any polarization controll. As the pump power increases, the operation state of the laser changes from cw oscillation to Q-switching, then to Q-switched mode locking at above 75 mW pump power, later to stable mode locking at above 87 mW pump power, and finally to unstable pulse splits at above 130 mW pump power. Figure 2 shows the stable mode-locking state at the pump power of 130 mW. Figure 2(a) shows the pulse waveform. The full width at half maximum (FWHM) of the pulse is measured to be 120 ps. The pulse train with a span of 9 μs is shown in the inset of Fig. 2(a). The radio frequency (RF) spectrum observed in the radio frequency spectrum analyzer is shown in Fig. 2(b) with a span of 60 kHz (1 Hz bandwidth) and a span of 1 GHz (1 KHz bandwidth) in the inset, which reaveals that the repetition rate is 27.03 MHz and the signal-to-noise ratio (SNR) is suppressed better than 55 dB. The high SNR and the stable pulse train with a span of 9 μs (shown in the inset of Fig. 2(a)) indicate a high temporal stability. The optical spectrum of the pulse train is shown in Fig. 2(c). The 3dB spectral band width is 0.15 nm and the ASE suppression ratio is 45 dB. The term ‘ASE suppression’ refers to the ratio of the spectral peak intensity at 1016 nm to the ASE spectral peak intensity around 1030 nm as measured by the optical spectrum analyzer. In the stable mode locking range, the laser output power increases almost linearly with the pump power with a slope efficiency of 23%, which is shown in Fig. 2(d). At the pump power of 130 mW, the maximum output power reaches 21.2 mW.
3.2 Performance of the first-stage YDFA (AMP1)
Due to core-pumping and the relatively short YDF (0.23m) of AMP1, the 1016 nm seed source is effectively amplified by AMP1. Figure 3 shows the direct output and ASE-filtered output of AMP1. Figure 3(a) shows the output power versus the pump power. The output power is boosted up to 210 mW at the pump power of 350 mW with a slope efficiency of 59%, and then attenuated to 197 mW with a slope efficiency of 54.7% by the ASE filter. Figure 3(b) shows the optical spectrum of AMP1. The ASE suppression ratio of the direct output spectrum of AMP1 is 43 dB, and then improves to 70 dB by the ASE filter. According to the spectrum integral, the fraction of ASE in the direct output is only 0.1% of the total light, which is so small that the residual ASE appears to be negligible and the ASE filter could be removed. We investigate the influence of the slight residual ASE on the performance of the next amplifier stage AMP2 in Sec 3.3.
3.3 AMP2 performance dependence on residual ASE
Figure 4 shows the influence of residual ASE in the incident signal on the performance of AMP2, with the pump power of 3.65 W and 2 m gain fiber. Figure 4(a) shows the optical spectra of the incident signal including 0.1% residual ASE and the amplifier output of AMP2. Due to the high gain of ASE, the residual ASE consumes lots of upper-level population and increases rapidly along with the increasing pump power. Especially, when the pump power increases to 3.65 W, parasitic oscillation occurs at the wavelength of 1037 nm. The pump power is not increased any more to avoid damage of the optical devices. The experimental result indicates that the residual ASE in the incident signal must be filtered to suppress the ASE amplification and prevent the appearance of parasitic oscillation. Figure 4(b) shows the optical spectrum of the incident ASE-filtered signal and the amplifier output of AMP2. In this circumstance, the 1016 nm signal can be amplified effectively and the ASE suppression is 41 dB at the pump power of 3.65 W. In fact, because the ASE filter is a short-wave-pass edge filter and has a steep edge at 1020 nm, the incident ASE-filtered signal still has slight residual ASE within 1016~1020 nm range which results in the residual ASE at 1016~1020 nm range in output of AMP2. However, the residual ASE at 1016~1020 nm range does not deteriorate the amplification of 1016 nm signal, because of the much smaller emission cross-section at 1010-1020 nm band than 1030 nm and the increasing absorption cross-section towards 1010-1020 nm band.
3.4 AMP2 performance dependence on fiber length
To suppress ASE amplification, the performance of amplifiers with different gain fiber length are tested. The gain fiber is 10/125 um double-clad YDF. The ASE-filtered output of AMP1 is used as the incident signal of AMP2. A conventional 1064 nm pulsed fiber amplifier usually employs a piece of 4~5 m long YDF for high efficiency and output power . As a comparison, the short-wavelength fiber amplifier (AMP2) firstly employs a piece of 4.5 m long YDF. In contrast to the conventional 1064 nm pulsed fiber amplifier, as the pump power increases, ASE around 1030 nm is rapidly increased other than the 1016 nm signal as a result of two factors. The first factor is the increasing absorption cross-section towards shorter wavelengths than 1030 nm. The longer the YDF length is, the more re-absorption of 1016 nm light occurs. The second factor is that ASE has higher gain and is continuous, and thus consumes most of the upper-level population all along. Therefore, the short-wavelength fiber amplifier requires shorter YDF than 1064 nm pulsed fiber amplifier for high efficiency and output power. Figure 5 shows the influence of YDF length on the performance of AMP2, at the same pump power of 23.7 W. Figure 5(a) and Fig. 5(b) show the output spectra and the output power versus the pump power of AMP2 with different YDF lengths, respectively. It is obvious that the shorter YDF length results in the higher ASE suppression ratio. But too short YDF of 1.2 m will depress the pump absorption and reduce the efficiency of the amplifier, corresponding to a low slope efficiency of 26% (shown by black line in Fig. 5(b)). Moreover, a too long YDF of 2.5 m also exhibits a low slope efficiency of 31% (shown by blue line in Fig. 5(b)) due to the re-absorption at the laser wavelength of 1016 nm. The optimum YDF length of AMP2 is about 2 m. By optimization, the final output of AMP2 has a high ASE suppression ratio of 40 dB and 11.6 W output power with a slope efficiency of 51% at the pump power of 23.7 W (shown by red line in Fig. 5(a) and Fig. 5(b)).
3.5. Performance of the boost amplifier (AMP3)
To further raise the laser power, a boost amplifier with LMA YDF is used after the two amplifiers described above. A fiber pigtailed filter is inserted after AMP2 in order to eliminate the influence of the residual ASE on AMP3. By optimization, the length of gain fiber is 2.5 m rather than 5~6 m in conventional 1064 nm LMA amplifiers. Figure 6(a) shows the output spectra at different output power. Due to the pulse laser’s high peak power, as the output power increases, the spectrum broadens progressively due to nonlinear effects including self-phase modulation (SPM), four-wave mixing (FWM) and stimulated Raman scattering (SRS) . Distinct spectral sidebands on both sides of the narrowband signal light due to FWM effects and a broad Raman Stokes peak centered at the wavelength of 1064 nm have been generated. Because the 1st Raman Stokes locates at 1064 nm where the gain is high in Yb-doped silica fiber, the Raman Stokes component is amplified rapidly. Figure 6(b) shows the spectrum at the maximum output power on a linear scale, indicating an overwhelming predominance of the 1016 nm laser. The double-peak spectral shape shown in the inset of Fig. 6(b) is due to SPM and the double peaks situate at 1015.54 nm and 1016.22 nm, respectively. The power of the 1016 nm (−20 dB) laser is calculated by spectrum integral and occupies 69.5% of the total power at the maximum output power.
Figure 6(c) shows the total output power and 1016 nm light power versus the pump power. The proportion of 1016 nm light power is calculated through spectrum integral. The total output power increases monotonously with increasing pump power with a slope efficiency of 71.7%. However, as the pump power increases, the output spectrum broadens more and more seriously as a result of the increasing pulse peak power. Therefore, the 1016 nm light power rolls over with increasing pump power and finally achieves 49 W under the pump power of 93 W, corresponding to an optical conversion efficiency of 53%. Figure 6(d) shows the output pulse waveform at the maximum output power. The FWHM of the pulse is measured to be ~120 ps, which is comparable with that of the seed laser. At last, a high power MOPA laser operating at 1016 nm with average power of 49 W, pulse energy of 1.8 μJ and peak power of 15 kW is demonstrated.
This paper presents an all fiber high power picosecond pulsed laser at 1016 nm in four-stage MOPA configuration. The seed source is an Yb-doped single mode fiber laser passively mode-locked by a SESAM. The pulse width is 120 ps and the repetition rate is 27.03 MHz. By three stages of amplifiers: single-mode fiber amplifier, 10/125 DCF fiber amplifier and 25/250 DCF fiber amplifier, the output power of the 1016 nm picosecond pulse laser reaches 50 W with an optical conversion efficiency of 53%. This is to our knowledge the highest power in all fiber pulsed YDFL at this wavelength so far.
Suppression of ASE plays a vital role in the amplification of the short-wavelength pulsed fiber laser because of the much smaller emission cross-section at 1010-1020 nm band than 1030 nm, and the increasing absorption cross-section towards shorter wavelengths. Moreover, the continuous ASE consumes the upper-level population all along and accumulates along the MOPA chain, but the incontinuous pulse only consumes the upper-level population in its duration time. For high efficiency and output power of the 1016 nm pulse fiber laser, the influence of the gain fiber length and the residual ASE in the signal on the performance of the 1016 nm fiber amplifier is investigated experimentally. The experimental results demonstrate the possibility of increasing the output power of 1016 nm fiber amplifiers by employing an optimized fiber length and an appropriate ASE-filtered incident signal.
Moreover, as the output power of 1016 nm pulsed fiber laser increases, the output spectrum broadens more and more seriously by nonlinear effects, resulting in a relatively low optical conversion efficiency. To suppress the nonlinear effects, we can use a repetition rate increasing system  to decrease the peak power in the future work. We believe that the obtained 1016 nm high power picosecond pulsed fiber laser are of great interest and may open up new prospects of many application areas.
National High Technology Research and Development Program of China (863 Program) (2015AA021101); State Key Program of National Natural Science of China (61235008).
The authors are grateful for Tong Liu, He Chen, Ke Yin, Zhi-Yuan Dou and Sen Guo for helpful discussion on the experiment.
References and links
1. M. O’Connor, V. Gapontsev, V. Fomin, M. Abramov, and A. Ferin, “Power scaling of SM fiber lasers toward 10kW,” in Conference on Lasers and Electro-Optics (Optical Society of America, 2009), paper CThA3. [CrossRef]
2. R. I. Epstein, M. I. Buchwald, B. C. Edwards, T. R. Gosnell, and C. E. Mungan, “Observation of laser-induced fluorescent cooling of a solid,” Nature 377(6549), 500–503 (1995). [CrossRef]
3. C. Mungan, M. Buchwald, B. Edwards, R. Epstein, and T. Gosnell, “Laser cooling of a solid by 16 K starting from room temperature,” Phys. Rev. Lett. 78(6), 1030–1033 (1997). [CrossRef]
4. S. D. Melgaard, D. V. Seletskiy, A. Di Lieto, M. Tonelli, and M. Sheik-Bahae, “Optical refrigeration to 119 K, below National Institute of Standards and Technology cryogenic temperature,” Opt. Lett. 38(9), 1588–1590 (2013). [CrossRef] [PubMed]
5. J. Hu, L. Zhang, H. Liu, K. Liu, Z. Xu, and Y. Feng, “High-power single-frequency 1014.8 nm Yb-doped fiber amplifier working at room temperature,” Appl. Opt. 53(22), 4972–4977 (2014). [CrossRef] [PubMed]
6. P. Villwock, S. Siol, and T. Walther, “Magneto-optical trapping of neutral mercury,” Eur. Phys. J. D 65(1–2), 251–255 (2011). [CrossRef]
7. R. Steinborn, A. Koglbauer, P. Bachor, T. Diehl, D. Kolbe, M. Stappel, and J. Walz, “A continuous wave 10 W cryogenic fiber amplifier at 1015 nm and frequency quadrupling to 254 nm,” Opt. Express 21(19), 22693–22698 (2013). [CrossRef] [PubMed]
8. L. Yi, S. Mejri, J. J. McFerran, Y. Le Coq, and S. Bize, “Optical lattice trapping of Hg199 and determination of the magic wavelength for the ultraviolet S01↔P03 clock transition,” Phys. Rev. Lett. 106(7), 073005 (2011). [CrossRef] [PubMed]
9. R. Paschotta, J. Nilsson, A. C. Tropper, and D. C. Hanna, “Ytterbium-doped fibre amplifiers,” IEEE J. Quantum Electron. 33(7), 1049–1056 (1997). [CrossRef]
10. A. Kurkov, “Oscillation spectral range of Yb-doped fiber lasers,” Laser Phys. Lett. 4(2), 93–102 (2007). [CrossRef]
11. Y. Wang, W. Ke, Y. Sun, Y. Ma, T. Li, Y. Feng, and J. Wu, “Research of high brightness 1018 nm ytterbium doped fiber laser,” in XX International Symposium on High Power Laser Systems and Applications (International Society for Optics and Photonics, 2015), pp. 92551–92558.
12. H. Xiao, J. Leng, H. Zhang, L. Huang, J. Xu, and P. Zhou, “High-power 1018 nm ytterbium-doped fiber laser and its application in tandem pump,” Appl. Opt. 54(27), 8166–8169 (2015). [CrossRef] [PubMed]
13. C. Yang, S. Xu, Q. Yang, W. Lin, S. Mo, C. Li, Z. Feng, D. Chen, Z. Yang, and Z. Jiang, “High-efficiency watt-level 1014 nm single-frequency laser based on short Yb-doped phosphate fiber amplifiers,” Appl. Phys. Express 7(6), 062702 (2014). [CrossRef]
14. J. Wang, G. Chen, L. Zhang, J. Hu, J. Li, B. He, J. Chen, X. Gu, J. Zhou, and Y. Feng, “High-efficiency fiber laser at 1018 nm using Yb-doped phosphosilicate fiber,” Appl. Opt. 51(29), 7130–7133 (2012). [CrossRef] [PubMed]
15. D. M. Boroson, A. Biswas, and B. L. Edwards, “MLCD: overview of NASA’s Mars laser communications demonstration system,” in Lasers and Applications in Science and Engineering (International Society for Optics and Photonics, 2004), pp. 16–28.
16. H. Hemmati, Deep Space Optical Communications (John Wiley & Sons, 2006).
17. D. Engin, W. Lu, F. Kimpel, and S. Gupta, “Short-wavelength (1010-1030nm) Yb-fiber-MOPA based multi-aperture high-power uplink laser beacons for space communication,” in SPIE LASE (International Society for Optics and Photonics, 2012), pp. 82460K–82469.
18. D. Engin, J. Burton, I. Darab, F. Kimpel, and S. Gupta, “Yb-fiber-MOPA based high energy and average power uplink laser beacon for deep space communication operating under Nested PPM format,” in SPIE Defense + Security (International Society for Optics and Photonics, 2015), pp. 94660V–94612.
19. W. Farr, “Technology development for high efficiency optical communications,” in Aerospace Conference, 2012 IEEE (IEEE, 2012), pp. 1–8. [CrossRef]
20. H. Yang, “High energy femtosecond fiber laser at 1018 nm and high power Cherenkov radiation generation,” (Massachusetts Institute of Technology, 2014).
21. Y. Hua, W. Liu, M. Hemmer, L. E. Zapata, G. Zhou, D. N. Schimpf, T. Eidam, J. Limpert, A. Tünnermann, and F. X. Kaertner, “87-W, 1018-nm Yb-fiber ultrafast seeding source for cryogenic Yb: YLF amplifier,” in CLEO: Science and Innovations (Optical Society of America, 2016), pp. SM4Q. 5.
23. P. Yan, R. Lin, S. Ruan, A. Liu, and H. Chen, “A 2.95 GHz, femtosecond passive harmonic mode-locked fiber laser based on evanescent field interaction with topological insulator film,” Opt. Express 23(1), 154–164 (2015). [CrossRef] [PubMed]
24. Z. Qin, G. Xie, H. Zhang, C. Zhao, P. Yuan, S. Wen, and L. Qian, “Black phosphorus as saturable absorber for the Q-switched Er:ZBLAN fiber laser at 2.8 μm,” Opt. Express 23(19), 24713–24718 (2015). [CrossRef] [PubMed]
25. P. Yan, R. Lin, S. Ruan, A. Liu, H. Chen, Y. Zheng, S. Chen, C. Guo, and J. Hu, “A practical topological insulator saturable absorber for mode-locked fiber laser,” Sci. Rep. 5, 8690 (2015). [CrossRef] [PubMed]
26. H. W. Chen, Y. Lei, S. P. Chen, J. Hou, and Q. S. Lu, “High efficiency, high repetition rate, all-fiber picoseconds pulse MOPA source with 125 W output in 15 m fiber core,” Appl. Phys. B 109(2), 233–238 (2012). [CrossRef]
27. G. P. Agrawal, Nonlinear Fiber Optics (Academic, 2010).