Widely tunable mid-infrared fiber laser source based ... - OSA Publishing

4094

Letter

Vol. 40, No. 17 / September 1 2015 / Optics Letters

Widely tunable mid-infrared fiber laser source based on soliton self-frequency shift in microstructured tellurite fiber M. YU. KOPTEV,1 E. A. ANASHKINA,1,2,* A. V. ANDRIANOV,1 V. V. DOROFEEV,3 A. F. KOSOLAPOV,4 S. V. MURAVYEV,1 AND A. V KIM1,2 1

Institute of Applied Physics of the Russian Academy of Sciences, 603950 Nizhny Novgorod, Russia University of Nizhny Novgorod, 603950 Nizhny Novgorod, Russia 3 Institute of Chemistry of High-Purity Substances of the Russian Academy of Sciences, 603950 Nizhny Novgorod, Russia 4 Fiber Optics Research Center of the Russian Academy of Sciences, 119333 Moscow, Russia *Corresponding author: elena.anashkina@gmail.com 2

Received 1 July 2015; revised 5 August 2015; accepted 6 August 2015; posted 7 August 2015 (Doc. ID 241420); published 27 August 2015

A turnkey fiber laser source generating high-quality pulses with a spectral sech shape and Fourier transform-limited duration of order 100 fs widely tunable in the 1.6–2.65 μm range is presented. It is based on Raman soliton selffrequency shifting in the suspended-core microstructured TeO2 -WO3 -La2 O3 glass fiber pumped by a hybrid Er/Tm fiber system. Detailed experimental and theoretical studies, which are in a very good agreement, of nonlinear pulse dynamics in the tellurite fiber with carefully measured and calculated parameters are reported. A quantitatively verified numerical model is used to show Raman soliton shift in the range well beyond 3 μm for increased pump energy. © 2015 Optical Society of America OCIS codes: (060.5530) Pulse propagation and temporal solitons; (190.5650)

Raman

effect;

(060.4005)

Microstructured

fibers;

(060.2290) Fiber materials. http://dx.doi.org/10.1364/OL.40.004094

Ultrabroadband coherent light sources in the mid-infrared (IR) are interesting for many applications: molecular spectroscopy, chemical sensing, biomedicine, ecology monitoring, defense, and security. Considerable potential for efficient mid-IR fiber laser systems is represented by several soft-glasses, including chalcogenide, fluoride, and tellurite ones. All of these materials can be used for fiber production for ultrabroadband wavelength conversion due to different mechanisms [1]. So, supercontinuums with spectral width of several octaves have been successfully generated in chalcogenide [2], fluoride [3], and tellurite [4] fibers. However, some applications require widely tunable high-quality ultrashort pulses. In particular, tunable pulses can be obtained by using the soliton self-frequency shifting (SSFS) effect in different types of soft-glass fibers, which are transparent in the mid-IR [5–9]. Here, to get high-quality femtosecond pulses with ultrabroad wavelength tuning, we focus on the SSFS in a self-made 0146-9592/15/174094-04$15/0$15.00 © 2015 Optical Society of America

microstructured tellurite fiber. Tellurite glasses possess the merits of sufficient chemical stability, wide transparency range up to ∼5 μm [10], and large linear and nonlinear refractive indices. Among the majority of tellurite glass compositions the tungstate-tellurite system has the advantages of higher stability against crystallization, which can be significantly improved due to application of higher purity starting chemicals [11] and modifying the glass composition by La2 O3 [12]. Some of TeO2 -WO3 -La2 O3 glasses are extremely stable against crystallization inside a wide range of La2 O3 concentrations [13]. High mid-IR fiber optics applicability of the glass is confirmed by production of low-absorption-loss samples with ultralow hydroxyl groups content, quality step index optical fibers with several techniques like double crucible [13] and monolithic preform stretching. Even though zero dispersion wavelength (ZDW) for tellurite glass is typically beyond 2 μm, microstructured fiber geometry for dispersion management and shifting of ZDW to the range shorter than 1.5 μm can be used. This enables operation in the anomalous dispersion range, which is required SSFS with conventional Er: fiber and Tm: fiber pump sources. Although the soliton-like spectral peaks in tellurite fibers have been experimentally demonstrated only in the range up 3 μm [5–8], we believe that these fibers are potentially suitable for SSFS in the range well beyond 3 μm. In this Letter, an all-fiber hybrid Er/Tm laser system is used to demonstrate a fiber laser source producing highquality pulses tunable in the 1.6–2.65 μm range with Fourier transform-limited duration of the order of 100 fs. This is confirmed by frequency-resolved optical gating (FROG) measurements from the Er-pumped tellurite fiber. To the best of our knowledge, solitons from tellurite fiber have been measured for the first time. The prospect of SSFS up to 4 μm in the tellurite fiber pumped at 2 μm is also discussed. We have carried out an experimental study with a setup, the layout of which is shown in Fig. 1. It is based on a passive mode-locked Er-doped fiber laser, fiber amplifiers with standard telecom components, and a highly nonlinear tellurite fiber

Letter


4095

Fig. 2. Losses and SEM image of TWL fiber cross section. Fig. 1. Optical layout of the experimental setup.

of 72TeO2 -24WO3 -4La2 O3 (mol. %) (TWL) glass. The laser with passive mode locking via nonlinear polarization rotation delivers 200-fs pulses centered at 1.57 μm at a fundamental frequency of 49 MHz. Then the pulses are amplified in the Er-doped fiber and propagate through a piece of standard telecom fiber SMF-28e. Pulses at 1.6 μm have energy of about 2 nJ and duration of 150 fs [14]. Next we consider two cases (“I” and “II” in Fig. 1). In the first one, pulses are directly free-space coupled to the TWL fiber. In the second one, before the TWL fiber, we append an additional amplifying cascade at 2 μm. We use a silica fiber to convert the signal from 1.6 μm up to 2 μm due to SSFS. Then the pulses propagate through a stretcher and are amplified in a Tm/Yb codoped GTWave fiber. At the Tmdoped fiber output, we have 150-fs pulses with energy of about 3 nJ [15]. A suspended-core microstructured fiber has been produced from TWL glass with reduced content of 3d-transition metals and hydroxyl groups impurities by drilled preforms stretching. The loss of TWL glass has been calculated from transmission spectrum of the sample of L 9.35 cm in length by the formula 10∕L · lg1 − R∕T , where T is transmission, R is Fresnel reflection. The loss of TWL fiber with initial length of L1 9.5 m has been measured by the cut-back technique using monochromator-based spectral installation with spectral range of 1–2.3 μm. The optical loss has been calculated as 10∕L1 − L2 · lgI 1 ∕I 2 , where intensity I 1 corresponds to a piece of the fiber of L1 in length, and I 2 is registered for a final piece of the fiber of L2 1 m in length. The results of loss measurements are shown in Fig. 2. The effective core diameter of the fiber is about 3.2 μm. The scanning electron microscope (SEM) image of the cross section is shown in the inset in Fig. 2. The air channels are axially asymmetric but the core is symmetric. We have not found the fiber birefringence because the fundamental mode is localized near the core. To evaluate dispersion and fundamental mode field distribution of the TWL fiber, we approximate it by axially symmetric tellurite wire waveguides located in air. We use the Sellmeier-type dispersion formula to obtain the refractive index of the wire material as for 80TeO2 -20WO3 (mol.%) glass [16]. This composition is not exactly similar to ours. However, due to a large refractive index difference Δn between the fiber core and air (n 1), small variation of glass composition does not lead to dramatic fiber dispersion changes. We used a standard procedure for finding electric and magnetic field components and wavenumber β of the mode HE11 as the exact solution of Maxwell’s equation for step index profiles [17].

The calculated effective mode area changes from 4.1 μm2 for a signal at 1 μm up to 4.4 μm2 for a signal at 2.2 μm. The estimated nonlinear coefficient γ is 900 W km−1 and 400 W km−1 for these wavelengths, assuming n2 5.9 × 10−19 m2 W −1 . The calculated dispersion (β2 ∂2 β∕∂ω2 , where ω is the angular frequency) is shown in Fig. 3(a). We have measured dispersion of the 63-cm TWL fiber and verified the value of its nonlinearity by using the broadband spectral interferometry technique. We use our previously developed light sources based on nonlinear fibers [14,18], which provide sufficiently broad and stable over time spectra in the range of 1.4–1.7 μm and 1.8–2.2 μm, respectively. Figure 3(a) also demonstrates the experimental β2 in these ranges. To estimate nonlinearity, we use high-quality 110 fs linearly polarized Raman-shifted solitons at 1.63 μm produced in a 20-m long piece of SMF-28e fiber. Before entering the interferometer, the pulse energy is attenuated in order to excite a soliton with N ≈ 1 in the TWL fiber. Spectral amplitude and phase of the pulse at the output of the TWL fiber are obtained after

Fig. 3. (a) Experimental spectrum at 63-cm piece of TWL fiber output pumped at 1.6 μm and TWL fiber dispersion. The dotted circle contains unconverted signal part. (b) Simulated spectrum at TWL fiber output. (c) Simulated spectral evolution of the pulse propagating at the fundamental mode in TWL fiber.

4096


processing the interferometric data. Then, we simulate propagation of 110 fs sech-shaped pulse and vary nonlinearity so as to obtain the best match of the simulated and measured pulses. Indeed, we have found γ ∼ 540 W km−1 at 1.63 μm, which agrees well with n2 5.9 × 10−19 m2 W −1 . In the first experiment on ultrabroadband signal transformation, 150-fs pulses at the central wavelength of 1.6 μm are coupled into a 63-cm piece of TWL fiber located back-to-back. We use a relatively long piece of fiber due to low launched energy. The input pulses are located in the anomalous dispersion range of the TWL fiber near the ZDW, which is about 1.55 μm. Due to the large soliton number N ≈ 9, the nonlinear dynamics are associated with higher-order soliton compression at the initial stage accompanying pulse fission into fundamental solitons in the anomalous dispersion range and dispersive waves (DW) in the normal dispersion range, their interaction during further propagation and SSFS [19,20]. The measured spectrum at the TWL fiber output is shown in Fig. 3(a). One can see the separated soliton peak at 2.15 μm with Fourier transform-limited duration of 140 fs. To model pulse propagation in the TWL fiber, where ultrabroad spectra transformation in the range exceeding an octave is observed, we have employed the one-way wave equation dealing with the full electric field of light. The specific feature of this study is that the effective mode area dependence on frequency should be taken into account [21]. We use the same numerical model as in our previous paper [18]. Actual dispersion and effective mode area profiles, a model loss function shown in Fig. 2, the fraction of power inside the core, and Raman contribution as in [22] have been taken into account. The best agreement is achieved for the input energy of 90 pJ [see Fig. 3(b)]. Numerical modeling confirms the higher-order soliton scenario with subsequent pulse fission into four Raman solitons and the same numbers of DW [see Figs. 3(b) and 3(c)]. Solitons at 1.85 and 2.15 μm are clearly seen in Fig. 3(b), but solitons at 1.65 and 1.7 μm are identified and filtered in the time domain and defined by inverse Fourier transform in the frequency domain, although their spectra interfere with each other. The theoretical analysis allows us to associate the components in the measured spectra with the most redshifted soliton, the wing of the second soliton, and DW [see Figs. 3(a) and 3(b)]. So, the numerical model has been verified quantitatively and can be used for further analysis and prediction. Further, we have investigated SSFS in a 50-cm TWL fiber by varying the input energy at the fundamental mode. We have first tuned the coordinates of the laser source output and TWL fiber input, located back-to-back, to maximized SSFS. After that, we have slowly changed their relative positions to lose part of the energy. This way of energy decreasing at the fundamental is very simple and allows preserving the temporal and spectral structure of the input signal. We obtained Raman solitons up to 2.25 and 2.65 μm with the pump at 1.6 and 2 μm, respectively. One can see the experimentally measured spectra in Fig. 4. The spectral widths of the most redshifted solitons correspond to Fourier transform-limited durations of order 100 fs. The measured spectral components near pump wavelengths may correspond to a signal propagating not only at the fundamental mode but also at the cladding or at the higher modes. In addition, beam divergence depends on wavelength, so the measured spectrum indicates only the presence or absence of various peaks but not their relative intensities. The numerical

Letter

Fig. 4. Measured spectra at the output of 50-cm piece of TWL fiber pumped at 1.6 and 2 μm; corresponding calculated spectra; filtered Raman solitons in the time domain and phases with presented soliton energies and durations.

simulation confirms that the most redshifted tunable pulses are highly coherent sech-shape solitons. A very good agreement between the experimental and theoretical spectra for varying input energies is achieved (with taking into account the above comments about relative intensities). The calculated pulses are Fourier transform-limited ones with plane phases having multi-picojoule energy and duration of the order of 100 fs (see Fig. 4). We have also made FROG measurements of pulses from the Er-pumped TWL fiber. The harmonics from two replicas with different delays registered with a CCD camera and their spectra are reconstructed. A conventional spectrometer

Letter

Fig. 5. (a) Experimentally measured FROG trace, (b) the retrieved spectrum, and (c) the pulse intensity distribution with phase in the time domain. FROG error is 0.018.


4097

We experimentally obtained tunable high-quality Raman solitons up to 2.25 and 2.65 μm with the pump at 1.6 and 2 μm, respectively. Their spectral widths correspond to the Fourier transform-limited duration of about 100 fs, which agrees with the calculations. The quantitatively verified numerical code is used to show SSFS in the range up to 4 μm for increased pump energy. So, the demonstrated turnkey fiber-based laser source can be used for applications requiring high-quality ultrabroadband femtosecond optical pulses. Funding. Ministry of Education and Science (02.V.49.21.0003); Russian Foundation for Basic Research (RFBR) (14-02-00537, 14-22-02076, 15-03-08324, 1543-02185). REFERENCES

Fig. 6. Calculated (a) central wavelengths and (b) energies of the most redshifted pulses versus propagation distance. The input pump-pulse energies are indicated.

has insufficient sensitivity to measure the second harmonics from several-tens picojoules pulses. Figure 5 shows the FROG trace and the retrieved sech-shape pulse with 85-fs duration and time-bandwidth product of 0.38. To the best of our knowledge, the solitons from tellurite fiber have been measured for the first time. So we have demonstrated that a simple laser source can be used for applications requiring high-quality broadband-tunable femtosecond optical pulses. The use of spectral filters allows all “parasitic” spectral components, except Raman solitons, to be removed easily if necessary. Despite the SSFS up to 2.65 μm reached in the experiment in the TWL fiber with low input energy of order 100 pJ, we believe that the designed laser system can be used for obtaining highly coherent Raman solitons in the spectral range well beyond 3 μm. As one can see from Fig. 4, the higher the input energy, the longer the Raman soliton wavelength is. However, for extending spectral range, it is necessary to operate with few-centimeter long fiber pieces because of mid-IR loss. We have simulated propagation of 150-fs pump pulses at 2 μm with increased energy. Figure 6 demonstrates the central wavelength and energy of the most redshifted solitons dependence on propagation distance. The SSFS up to 4 μm for input energy of 2 nJ is limited by high mid-IR loss and large jβ2 j. For higher pump energies, pulse evolution associated with the modulation instability regime and energy in the first Raman soliton does not practically increase at the stage of its formation. In conclusion, a detailed study of soliton-like pulses based on the SSFS effect in the suspended-core microstructured tellurite fiber pumped by an all-fiber femtosecond Er/Tm laser system is presented. We have carefully measured and calculated tellurite fiber parameters used for numerical modeling. A very good agreement between the experiments on nonlinear pulse evolution and the corresponding simulation is demonstrated.

1. G. Tao, H. Ebendorff-Heidepriem, A. M. Stolyarov, S. Danto, J. V. Badding, Y. Fink, J. Ballato, and A. F. Abouraddy, Adv. Opt. Photon. 7, 379 (2015). 2. C. R. Petersen, U. Moller, I. Kubat, B. Zhou, S. Dupont, J. Ramsay, T. Benson, S. Sujecki, N. Abdel-Moneim, Z. Tang, D. Furniss, A. Seddon, and O. Bang, Nat. Photonics 8, 830 (2014). 3. X. Jiang, N. Y. Joly, M. A. Finger, F. Babic, G. K. L. Wong, J. C. Travers, and P. S. J. Russell, Nat. Photonics 9, 133 (2015). 4. P. Domachuk, N. A. Wolchover, M. Cronin-Golomb, A. Wang, A. K. George, C. M. B. Cordeiro, J. C. Knight, and F. G. Omenetto, Opt. Express 16, 7161 (2008). 5. T. Cheng, L. Zhang, X. Xue, D. Deng, T. Suzuki, and Y. Ohishi, Opt. Express 23, 4125 (2015). 6. G. S. Qin, X. Yan, C. Kito, M. Liao, T. Suzuki, A. Mori, and Y. Ohishi, Laser Phys. 21, 1404 (2011). 7. L. Zhang, T. Cheng, D. Deng, D. Sega, L. Liu, X. Xue, T. Suzuki, and Y. Ohishi, IEEE Photon. Technol. Lett. 27, 1547 (2015). 8. M. Belal, L. Xu, P. Horak, L. Shen, X. Feng, M. Ettabib, D. J. Richardson, P. Petropoulos, and J. H. V. Price, Opt. Lett. 40, 2237 (2015). 9. T. Cheng, Y. Kanou, K. Asano, D. Deng, M. Liao, M. Matsumoto, T. Misumi, T. Suzuki, and Y. Ohishi, Appl. Phys. Lett. 104, 121911 (2014). 10. R. Thapa, D. Rhonehouse, D. Nguyen, K. Wiersma, C. Smith, J. Zong, and A. Chavez-Pirson, Proc. SPIE 8898, 889808 (2013). 11. A. N. Moiseev, V. V. Dorofeev, A. V. Chilyasov, V. G. Pimenov, T. V. Kotereva, I. A. Kraev, L. A. Ketkova, A. F. Kosolapov, V. G. Plotnichenko, and V. V. Koltashev, Inorg. Mater. 47, 665 (2011). 12. X. Feng, Ch. Qi, F. Lin, and H. Hu, J. Non-Cryst. Solids 256–257, 372 (1999). 13. V. V. Dorofeev, A. N. Moiseev, M. F. Churbanov, G. E. Snopatin, A. V. Chilyasov, I. A. Kraev, A. S. Lobanov, T. V. Kotereva, L. A. Ketkova, A. A. Pushkin, V. V. Gerasimenko, V. G. Plotnichenko, A. F. Kosolapov, and E. M. Dianov, Opt. Mater. 33, 1911 (2011). 14. E. A. Anashkina, A. V. Andrianov, S. V. Muravyev, and A. V. Kim, Opt. Express 19, 20141 (2011). 15. M. Y. Koptev, E. A. Anashkina, A. V. Andrianov, S. V. Muravyev, and A. V. Kim, Opt. Lett. 39, 2008 (2014). 16. G. Ghosh, J. Am. Ceram. Soc. 78, 2828 (1995). 17. A. W. Snyder and J. Love, Optical Waveguide Theory (Springer, 1983). 18. E. A. Anashkina, A. V. Andrianov, M. Y. Koptev, S. V. Muravyev, and A. V. Kim, IEEE J. Sel. Top. Quantum Electron. 20, 7600608 (2014). 19. G. P. Agrawal, Nonlinear Fiber Optics (Academic, 2013). 20. D. V. Skryabin and A. V. Gorbach, Rev. Mod. Phys. 82, 1287 (2010). 21. J. Laegsgaard, Opt. Express 15, 16110 (2007). 22. X. Yan, G. Qin, M. Liao, T. Suzuki, and Y. Ohishi, J. Opt. Soc. Am. B 28, 1831 (2011).