Open Access
Add to BibSonomy
Abstract
Here, we demonstrate a compact and efficient high-power mid-infrared supercontinuum (MIR-SC) laser source based on a tunable noise-like pulse (NLP) fiber laser system and a short section of single-mode germania-core fiber (GCF). The NLP all-polarization-maintaining fiber laser system can deliver the maximum output power of ∼30.6 W and a broadband spectrum (∼1.8-2.7 µm) with a compact single-stage fiber amplifier. By directly pumping only ∼6.5 cm-long GCF with a core diameter of ∼3.5 µm, a MIR-SC (spectral coverage of ∼1.5-3.3 µm) with a maximum power of ∼25.2 W and a power conversion efficiency ∼81.2% is obtained, which represent the highest power and efficiency in any single-mode GCF-based MIR-SCs, to the best of our knowledge. Our study contributes to the high-power MIR-SC laser source with compact all-fiber configuration, and will prompt its practical applications.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
High-power mid-infrared supercontinuum (MIR-SC) as an advanced broadband laser source has significant applications in laser imaging, bio-medical, environmental monitoring, infrared spectroscopy, national defense, etc [1–5]. In the past decade, many research efforts were focused on the realization of the high-power MIR-SC, in particular with the all-fiber configuration. In general, for the generation of high-power MIR-SC in fiber laser system, the appropriate nonlinear fiber and high peak power pump laser are the two key factors.
Due to the high phonon energy (∼1100 cm-1) of the silica [6], the transmission wavelength of the mature silica-fiber is seriously limited towards long-wavelength (typically <∼2.7 µm). The commonly used MIR fiber includes the fluoride fiber, chalcogenide fiber, tellurite fiber, fluorotellurite fiber, and germania-core fiber (GCF). The fluoride fiber with a low phonon energy (<∼550 cm-1) exhibits an excellent transmission property with the wavelength less than ∼5 µm, which makes it as a potential candidate for high-power MIR-SC generation [7–10]. But it suffers the issue of photo-degradation or end facet damage caused by the OH- diffusion in the fiber tip [11,12]. Moreover, the fluoride fiber has a relatively low nonlinear refractive index, indicating a long fiber length is required for the MIR-SC generation [10]. The chalcogenide fiber with a lower phonon energy (<∼400 cm-1) owns the potential ability for the MIR-SC generation towards longer wavelength (>∼6 µm) [13,14]. Nevertheless, the low glass transition temperature limits its output power, so also the tellurite fiber [15,16]. Recently, the fluorotellurite fiber was also studied for the generation of high-power MIR-SC, which obtains the maximal output power >20 W [17,18].
Comparing with the aforementioned MIR soft glass fiber, the GCF has a better solid physical property, which is similar to the silica-fiber. The melting point of the GCF (∼1116 °C) is close to the value of silica-fiber, makes it easy to fuse with silica fiber [19]. Although the relatively high phonon energy (∼820 cm-1) of GCF means a large transmission loss with wavelength over 3 µm (minimal transmission loss ∼0.9 dB/cm at ∼3.39 µm) [20]. The high nonlinear refractive index of the germania (∼2.6 × 10−17 m2/W @ 4 µm, much stronger than silica glass) endows the GCF with a strong nonlinearity [21,22]. Moreover, the large numerical aperture and small core size of the GCF decrease the mode field area, further reinforcing the nonlinearity of the GCF, which enables the generation of MIR-SC in a quite short length of GCF [23]. Till now, based on the multimode (MM) GCF, the highest output power of the MIR-SC is over 40 W, and the long-wavelength edge is over ∼3.5 µm. Furthermore, it shows a good power stability during an hour monitoring [24]. However, with the single-mode (SM) GCF, the maximal output power of the MIR-SC is only ∼6.12 W [25].
The ultra-short pulse pump laser with high-power is another key factor for the implementation of high-power MIR-SC. Generally, ultra-short pulse laser at the 2 µm spectral region is the ideal pump source. To date, different kinds of pulse fiber lasers e.g., femto-, pico- and nano-second pulse fiber lasers have been employed for MIR-SC generation. Nevertheless, the complex and costly multi-stage fiber amplifier is essential for the high-power outputting due to the relatively low power of the seed pulse laser, which will prominently aggravate the reliability and manipuility. Recently, the ultra-short noise-like pulse (NLP) fiber laser has been extensively studied, which can be easily achieved in a passively mode-locked fiber laser. The mode-locked NLP is a pulse bunch with nano- or pico-second envelope, consisted by numerous femtosecond sub-pulses. Meantime, the ultra-short NLP exhibits a wide and smooth spectrum in frequency domain. Such the properties strongly imply the potential application in high-power MIR-SC generation [26,27]. In the mode-locked oscillator, the envelope width of the NLP will be increased with the pump power to maintain a stable peak power satisfied the mode-locked operation, which indicates that the high average output power can be expected directly in an all-fiberized oscillator [28]. Benefitting by the high-power and tunable seed pulse, the high-power NLP fiber laser system can be implemented easily, which is conducive to a compact and controllable high-power MIR-SC fiber configuration.
In this paper, we present a compact and efficient high-power MIR-SC fiber laser system based on the NLP and SM-GCF. A compact and tunable high-power NLP fiber laser system with maximal output power of ∼30.6 W was constructed with a single-stage fiber amplifier, benefitting by the high-power mode-locked NLP oscillator. After pumping a ∼6.5 cm-long SM-GCF, the ∼1.5-3.3 µm MIR-SC laser source with a maximal output power of ∼25.2 W and power conversion efficiency of ∼81.2% was obtained.
2. Experimental setup
The configuration of the high-power MIR-SC fiber laser system is depicted in Fig. 1. A home-built NALM-based all-polarization-maintaining (PM) fiber mode-locked NLP oscillator is utilized as the seed pulse laser source, which is amplified by the subsequent single-stage all-PM fiber amplifier. The gain of fiber amplifier is provided by a section of ∼2 m long PM double-clad Tm-doped fiber (TDF) with an absorption of ∼9.6 dB/m at ∼793 nm. The PM-TDF has the core and inner-clad diameters of ∼10 µm and∼130 µm respectively, which is pumped by two laser diodes, centered at ∼793 nm via a PM fiber combiner. Then, a commercial stripper is used for removing the residual pump power. For the efficient splicing with the SM-GCF, a mode field adaptor (MFA) with ∼25 cm-long PM-1950 output fiber is spliced at the end of the amplifier. The total passive fiber length behind the PM-TDF is ∼1.5 m, and all the fiber amplifier is fixed on a water-cooled plate with a temperature of ∼15 °C.
The GCF used in our experiment has the GeO2 core concentration of ∼50 mol. % with a core NA of ∼0.46 and the zero-dispersion wavelength of ∼1350 nm. The optical image of the GCF facet is depicted in Fig. 2(b) inset, which has the core/cladding diameters of ∼3.5/125 µm, respectively. To clearly understand the transmission performance of the GCF, the transmission loss of the GCF in a spectral coverage of ∼1.7-3 µm was roughly measured and calculated based on a home-made broadband laser source, as given in Fig. 2. The fixed broadband laser source was separately coupled into the optical spectrum analyzer (Yokogawa, AQ6376) via the free-space and a piece of GCF (∼30 cm). Subsequently, the transmission loss of the GCF (Fig. 2(b)) is obtained by the subtraction operation between the two traces in Fig. 2(a). The transmission loss is up to ∼8 dB at ∼3 µm wavelength with only ∼30 cm long GCF, implies that the transmission loss in long-wavelength will seriously limit the broadening of the SC. The deviation of the transmission loss from ∼2000 to 2500 nm is mainly considered as the result of absorption loss of the free-space. In addition, the reason of the prominent dips around ∼1840 nm, corresponding to the peaks in GCF-based trace (Fig. 2(a)) is speculated as the dispersive wave generation in GCF, owing the negative dispersion of the GCF.
Fig. 2. The measurement of the transmission loss of the GCF, (a) the traces from the spectrum analyzer, (b) the calculated transmission loss of the ∼30 cm-long GCF and the optical image of the GCF facet (inset).
3. Results and discussion
The output properties of the mode-locked NLPs from the oscillator are exhibited in Fig. 3. The all-PM fiber oscillator can directly deliver stable NLPs with the maximal output power over 1 W. As the enhancement of output power, the envelope width of the NLP is continuously increased from ∼3.5 ns to ∼16 ns, to maintain the stable mode-locked operation with the peak power clamping effect. At the output power of ∼0.71 W, the radio-frequency (RF) spectra with different resolutions were recorded by a RF analyzer (Keysight, N9020B) together with a photodetector (EOT, ET-5000F), as plotted in Fig. 3(b). The mode-locked NLP train has a fundamental frequency of ∼1.662 MHz, and the high signal to noise ratio (SNR) of ∼80 dB indicates an excellent stability of the mode-locked operation. In the following experiment, the seed NLPs with output powers of ∼0.376 W, ∼0.545 W, and ∼0.71 W were employed individually, and the corresponding optical spectra are plotted in Fig. 3(c). The NLPs have the central wavelength of ∼1994 nm, and the 3dB bandwidth is slightly broadened from ∼13.7 nm to ∼14.8 nm with the output power. The characteristics in time-domain of the NLPs are shown in Fig. 4, which were recorded by a high-speed oscilloscope (Teledyne Lecroy, WaveRunner 9000) and an autocorrelator (APE, pulseCheck 600). The mode-locked NLPs have the rectangle shapes with envelope widths of ∼5.8 ns, ∼8.3 ns and ∼10.8 ns, respectively. Contrary to the tendency of the envelope width, the coherence spike width gets narrow with the increasing of power, which fits in the broadening of the optical spectrum (Fig. 3(c)), originated from the effect of self-phase modulation. The corresponding coherence spike widths are ∼366 fs, ∼357 fs and ∼342 fs respectively with the sech2 profile assuming. Moreover, with the envelope width of ∼5.8 ns, the interferometric autocorrelation trace was measured and shown in Fig. 5, which exhibits a peak-to-background ratio of 4:1, and further verifies the NLP mode-locking.
Fig. 3. The output properties of the mode-locked NLP oscillator, (a) the output power and envelope width of the NLP against pump power, (b) the RF spectra at output power of ∼0.71W, (c) the output spectra.
Fig. 4. Pulse characteristics of the NLPs with output power of ∼0.376 W, ∼0.545 W and ∼0.71W, respectively, (a), (d), (g) are the corresponding pulse envelopes, respectively; (b), (e), (h) are the corresponding autocorrelation traces with a range of ∼2.1 ps, respectively; and (c), (f), (i) are the corresponding autocorrelation traces with a range of 150 ps, respectively.
Subsequently, the seed NLPs were boosted by a single-stage all-PM fiber amplifier. The evolutions of the optical spectrum and output power of the NLPs with three different parameters are separately exhibited in Fig. 6, which were recorded from the end of the MFA. With the envelope widths of ∼5.8 ns, ∼8.3 ns, and ∼10.8 ns, the maximal output powers of ∼24.1 W, ∼28.2 W, and ∼30.6 W were obtained individually. Although the maximal average output power is improved with a wider NLP envelope, the peak power of the NLP is gradually decreased from ∼2.5 kW to ∼1.7 kW with an approximate calculation, which is also verified by the optical spectrum, e.g., the spectral coverages are ∼1.84-2.71 µm @ ∼24.1 W (Fig. 6(a)), ∼1.84-2.67 µm @ ∼28.2 W (Fig. 6(b)), and ∼1.84-2.62 µm @ ∼30.6 W (Fig. 6(c)). The SCs generated from the amplifier are caused by the nonlinear effects of the self-phase-modulation (SPM) and the stimulated Raman scattering (SRS)-induced Stokes shifts. Nevertheless, the long-wavelength edge is difficult to over ∼2.7 µm due to the gigantic transmission loss of silica-fiber. As the envelope width broadening from ∼5.3 ns to ∼10.8 ns, the output slope efficiency is slightly increased from ∼42.3% to ∼45.7%. This is mainly due to the better suppression of amplifier spontaneous emission with a higher seed power. We noted that in the case of envelope width of ∼10.8 ns, the output power still increases linearly (Fig. 6(f)) without saturation, which indicates that the higher output power can be expected if the pump power is further improved.
Fig. 6. The evolutions the boosted NLPs under different envelope widths of ∼5.8 ns, ∼8.3 ns, and ∼10.8 ns. (a), (b), (c) the corresponding evolutions of spectrum; (d), (e), (f) the corresponding evolutions of output power.
For the generation of MIR-SC, the NLPs with a broad spectrum were directly launched into a section of GCF. In consideration of the high nonlinearity and large transmission loss of the GCF, ∼10 cm-long GCF is employed at the beginning. The evolutions of the output MIR-SC under different NLP envelope widths of ∼5.8 ns, ∼8.3 ns, and ∼10.8 ns are plotted in the Figs. 7(a), 7(b), and 7(c), respectively. As the enhancement of the incident power, the SC is continuously broadened. The widest spectral coverage of ∼1.5-3.29 µm with output power of ∼17.3 W was obtained, when the NLP with envelope width of ∼5.8 ns was adopted. The corresponding 20dB spectral bandwidth of the SC is ∼1160 nm spanning from ∼1.82 to ∼2.98 µm. After separately setting the envelope width to ∼8.3 ns and ∼10.8 ns, the long-wavelength edges of the SCs were shortened to ∼3.27 µm and ∼3.24 µm, due to the lower peak power. But the corresponding maximal output powers were increased to ∼20.9 W and ∼24.7 W. Theoretically, owing to the anomalous dispersion of the GCF, the broadening of the SC ought to towards the long-wavelength based on the dominant nonlinear effects of soliton fission and the SRS-induced Stokes shifts. The short wavelength of the SC in our experiment is considered as the result of the dispersive wave generation, which is verified by the peaks in Fig. 2(a). By measuring the output power, under three different envelope widths, each of the obtained MIR-SCs exhibits a high conversion efficiency of ∼78.4%, ∼79%, and ∼79.9% respectively. The reasons of high conversion efficiency are considered as: 1) the high quantum efficiency from the NLP laser with a broad spectrum to the MIR-SC; 2) the relatively low transmission loss of the short GCF; 3) the efficient splicing (∼97%) between GCF and PM-1950 fiber.
Fig. 7. With ∼10 cm-long GCF, the evolutions of the output MIR-SC under different NLP envelope widths of ∼5.8 ns, ∼8.3 ns, and ∼10.8 ns. (a), (b), (c) the corresponding evolutions of spectrum; (d), (e), (f) the corresponding evolutions of output power.
To reduce the transmission loss of the GCF, the length of the utilized GCF was further shortened to ∼6.5 cm. Relying on the aforementioned NLP pump lasers, the evolutions of the MIR-SC are depicted in Figs. 8(a), 8(b) and 8(c), respectively. Although the shorter GCF also restrains the nonlinearity, the slight broadening of the SCs was observed. The spectral coverages of ∼1.5-3.31 µm, ∼1.5-3.29 µm and ∼1.5-3.27 µm, corresponding to the 20 dB bandwidths of ∼1220 nm (∼1.79-3.01µm), ∼1160 nm (∼1.8-2.96 µm), ∼1120 nm (∼1.8-2.92 µm) were obtained respectively with the three different NLPs. Comparing with the results in Fig. 7, it strongly implies that the transmission loss of the GCF seriously limits the broadening of the spectrum. Meanwhile, the broadening of the SC with wavelength over 3.3 µm is difficult as the rapidly increased transmission loss of the GCF. Moreover, as expected, the corresponding maximal output powers have been increased to ∼19.1 W, ∼22.8 W, and ∼25.2 W. Simultaneously, the higher conversion efficiencies of ∼79.4%, ∼80.2%, and ∼81.2% were obtained respectively with the shorter GCF, as shown in Figs. 8(d), 8(e), and 8(f).
Fig. 8. With ∼6.5 cm long GCF, the evolutions of the output MIR-SC under different NLP envelope widths of ∼5.8 ns, ∼8.3 ns, and ∼10.8 ns. (a), (b), (c) the corresponding evolutions of spectrum; (d), (e), (f) the corresponding evolutions of output power.
For better understanding the research status of the GCF-based high-power MIR-SC, Table 1 summaries the representative results in recent years. Compared with the previously reported works, our high-power MIR-SC fiber laser system has more compact and costless configuration with a single-stage fiber amplifier. The maximal output power of ∼25.2 W and conversion efficiency of ∼81.2% are the highest values of SM-GCF-based MIR-SC, respectively. Our experiment results provide a compact and efficient high-power MIR-SC laser source, which will accelerate the real application of the high-power MIR-SC. Furthermore, in consideration of the unsaturation of the fiber amplifier, and the potentially wider pulse envelope based on the tunable NLP oscillator, a higher output power could be expected without the limitation of pump power.
Table 1. The representative results of the high-power MIR-SC with GCF
4. Conclusion
In summary, we have presented a compact and efficient high-power MIR-SC fiber laser system based on the NLP and GCF. Benefitting from the tunable high-power mode-locked NLP fiber oscillator, the single-stage fiber amplifier can directly deliver the maximal output power of ∼30.6 W with a spectral coverage of ∼1.8-2.7 µm, which remarkably simplified the laser system. After pumping a short segment SM-GCF (∼6.5 cm), ∼25.2 W MIR-SC with a whole spectral coverage of ∼1.5-3.3 µm and slope efficiency of ∼81.2% was obtained. The limitations of the output power and spectral coverage in our experiment are mainly caused by the currently available pump power and rapidly increased transmission loss of the GCF, respectively. Such a high-power MIR-SC can find potentially significant applications in optical spectroscopy, bio-medical, etc.
Funding
Key-Area Research and Development Program of Guangdong Province (2023B0909010005); Guangdong Basic and Applied Basic Research Foundation (2023A1515111114); Fundamental research project of Department of Education of Guangdong Province (2021ZDJS106); Shenzhen Pingshan District Science and Technology Innovation Fund (KY2022QJKCZ001, PSKG202003, PSKG202007).
Disclosures
The authors declare no conflicts 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. N. M. Israelsen, C. R. Petersen, A. Barh, et al., “Real-time high-resolution mid-infrared optical coherence tomography,” Light: Sci. Appl. 8(1), 11 (2019). [CrossRef]
2. D. An, F. Sun, Y. Bian, et al., “Mid-infrared absorption spectroscopy with enhanced detection performance for biomedical applications,” Appl. Spectrosc. Rev. 58(10), 834–868 (2023). [CrossRef]
3. C. Amiot, A. Aalto, P. Ryczkowski, et al., “Cavity enhanced absorption spectroscopy in the mid-infrared using a supercontinuum source,” Appl. Phys. Lett. 111(6), 061103 (2017). [CrossRef]
4. F. Borondics, M. Jossent, C. Sandt, et al., “Supercontinuum-based Fourier transform infrared spectromicroscopy,” Optica 5(4), 378–381 (2018). [CrossRef]
5. L. Yang, B. Zhang, K. Yin, et al., “Spectrally flat supercontinuum generation in a holmium-doped ZBLAN fiber with record power ratio beyond 3 µm,” Photonics Res. 6(5), 417–421 (2018). [CrossRef]
6. X. Zou and T. Izumitani, “Spectroscopic properties and mechanisms of excited state absorption and energy transfer upconversion for er3+–doped glasses,” J. Non-Cryst. Solids 162(1-2), 68–80 (1993). [CrossRef]
7. C. Xia, Z. Xu, M. N. Islam, et al., “10.5 W Time-Averaged Power Mid-IR Supercontinuum Generation Extending Beyond 4 µm With Direct Pulse Pattern Modulation,” IEEE J. Select. Topics Quantum Electron. 15(2), 422–434 (2009). [CrossRef]
8. C. Yao, Z. Jia, Z. Li, et al., “High-power mid-infrared supercontinuum laser source using fluorotellurite fiber,” Optica 5(10), 1264–1270 (2018). [CrossRef]
9. L. Yang, Y. Li, B. Zhang, et al., “30-W supercontinuum generation based on ZBLAN fiber in an all-fiber configuration,” Photon. Res. 7(9), 1061–1065 (2019). [CrossRef]
10. L. Yang, B. Zhang, X. He, et al., “20.6 W Mid-Infrared Supercontinuum Generation in ZBLAN Fiber With Spectrum of 1.9–4.3 µm,” Photonics Res. 38(18), 5122–5127 (2020). [CrossRef]
11. M. Le Toullec, C. J. Simmons, and J. H. Simmons, “Infrared spectroscopic studies of the hydrolysis reaction during leaching of heavy-metal fluoride glasses,” J. Am. Ceram. Soc. 71(4), 219–224 (1988). [CrossRef]
12. Y. O. Aydin, V. Fortin, R. Vallée, et al., “Towards power scaling of 2.8 microns fiber lasers,” Opt. Lett. 43(18), 4542–4545 (2018). [CrossRef]
13. Z. Zhao, X. Wang, S. Dai, et al., “1.5–14 µm midinfrared supercontinuum generation in a low-loss Te-based chalcogenide step-index fiber,” Opt. Lett. 41(22), 5222–5225 (2016). [CrossRef]
14. B. Yan, T. Huang, W. Zhang, et al., “Generation of watt-level supercontinuum covering 2-6.5 µm in an all-fiber structured infrared nonlinear transmission system,” Opt. Express 29(3), 4048–4057 (2021). [CrossRef]
15. V. Shiryaev and M. Churbanov, “Trends and prospects for development of chalcogenide fibers for mid-infrared transmission,” J. Non-Cryst. Solids 377, 225–230 (2013). [CrossRef]
16. H. Shi, X. Feng, F. Tan, et al., “Multi-watt mid-infrared supercontinuum generated from a dehydrated large-core tellurite glass fiber,” Opt. Mater. Express 6(12), 3967–3976 (2016). [CrossRef]
17. F. Wang, K. K. Wang, C. F. Yao, et al., “Tapered fluorotellurite microstructured fibers for broadband supercontinuum generation,” Opt. Lett. 41(3), 634–637 (2016). [CrossRef]
18. X. Guo, Z. Jia, Y. Jiao, et al., “25.8 W All-Fiber Mid-Infrared Supercontinuum Light Sources Based on Fluorotellurite Fibers,” IEEE Photon. Technol. Lett. 34(7), 367–370 (2022). [CrossRef]
19. E. M. Dianov and V. M. Mashinsky, “Germania-Based Core Optical Fibers,” J. Lightwave Technol. 23(11), 3500–3508 (2005). [CrossRef]
20. J. Ballato, T. Hawkins, P. Foy, et al., “Silica-clad crystalline germanium core optical fibers,” Opt. Lett. 36(5), 687–688 (2011). [CrossRef]
21. N. K. Hon, R. Soref, and B. Jalali, “The third-order nonlinear optical coefficients of Si, Ge, and Si1-xGex in the midwave and longwave infrared,” J. Appl. Phys. 110(1), 011301 (2011). [CrossRef]
22. M. Ordua and S. N. Basub, “Recent progress in germanium-core optical fibers for mid-infrared optics,” Infrared Phys. Technol. 111, 103507 (2020). [CrossRef]
23. X. Wang, C. Yao, P. Li, et al., “All-fiber high-power supercontinuum laser source over 3.5 µm based on a germania-core fiber,” Opt. Lett. 46(13), 3103–3106 (2021). [CrossRef]
24. L. Yang, Y. Yang, B. Zhang, et al., “Record power and efficient mid-infrared supercontinuum generation in germania fiber with high stability,” High Power Laser Sci. Eng. 10, e36 (2022). [CrossRef]
25. K. Yin, B. Zhang, J. Yao, et al., “1.9-3.6 µm supercontinuum generation in a very short highly nonlinear germania fiber with a high mid-infrared power ratio,” Opt. Lett. 41(21), 5067–5070 (2016). [CrossRef]
26. M. Wang, H. Wu, D. Ouyang, et al., “High power noise-like pulse at 2 µm and its applications in mid-IR Raman light and flat supercontinuum,” Infrared Phys. Technol. 131, 104635 (2023). [CrossRef]
27. X. Zhu, D. Zhao, B. Zhang, et al., “Spectrally flat mid-infrared supercontinuum pumped by a high power 2µm noise-like pulse,” Opt. Express 31(8), 13182–13194 (2023). [CrossRef]
28. 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]
29. K. Yin, B. Zhang, L. Yang, et al., “30 W monolithic 2–3 µm supercontinuum laser,” Photonics Res. 6(2), 123–126 (2018). [CrossRef]
