1. Introduction

Multiphoton microscopy is considered as a powerful tool for both deep and vivo fluorescence imaging of biological scattering tissues with subcellular resolution [1–4]. Two-photon microscopy is now well established with the rapid progress in genetically engineered probes, which enables, in turn, fast development in fields such as neuroscience, immunology and biology [5]. Recent advances show that the utilization of three-photon excitation is an effective strategy to obtain deeper multiphoton imaging, namely three-photon microscopy [6–8]. In general, the achievable imaging depth of multiphoton microscopy is limited by the scattering effect in biological tissues [9,10]. A direct way to increase the imaging depth is to use a laser source with a longer excitation wavelength, because the longer wavelength excitation is able to reduce the scattering effect in the biological tissues [11,12]. In fact, the optimum excitation spectral window needs to consider both the scattering and absorption effects in tissues simultaneously. It has been demonstrated that 1.7 µm would be a critical waveband for excitation of three-photon microscopy when taking scattering and absorption effects into account [6]. Moreover, the three-photon excitation spectra by 1.7 µm waveband also match with the red fluorochromes that were widely used in fluorescence imaging.

The demand for the three-photon microscopy motivates scientists to develop high-performance pulsed excitation laser sources operating at 1.7 µm waveband. The 1.7 µm fiber-based excitation sources are good candidates for three-photon microscopy owing to several advantages over solid-state lasers, such as compactness and robustness. To date, the 1.7 µm fiber-based sources have been demonstrated with several methods, i.e., nonlinear wavelength conversion [13–15], bismuth-doped fiber laser [16], and Tm-doped fiber laser [17–22]. However, the nonlinear wavelength conversion requires specially designed fiber to achieve the emission of high peak power pulse around 1.7 µm waveband [13–15]. Meanwhile, the bismuth-doped fiber is still not commercially available and exhibits a low gain coefficient at 1.7 µm waveband [16]. In fact, obtaining 1.7 µm ultrashort pulse directly from Tm-doped fiber laser would be a promising approach, given the mature fabrication technology of Tm-doped fiber and its broad emission ranges from 1600 nm to 2100 nm [23]. To date, both the artificial saturable absorption effects i.e., nonlinear polarization rotation (NPR) and nonlinear amplifying loop mirror (NALM) [17–21] and real saturable absorber i.e., semiconductor saturable absorber mirror (SESAM) [22] were employed to achieve 1.7 µm ultrashort pulse in Tm-doped fiber lasers. However, the NPR technique suffers the environmental instability owing to the polarization-dependent operation regimes, while the NALM-based figure-8 mode-locked fiber lasers are difficult to achieve self-starting operation to some extent as it requires a large phase shift difference between two counter-propagating beams in the fiber loop. Incorporating a SESAM into the 1.7 µm laser cavity is expected to be an efficient way to obtain passive mode-locking. Nevertheless, in general the SESAM exhibits a low damage threshold, leading to inferior performance of an ultrafast fiber laser.

In fact, the self-starting ability of the NALM-based mode-locked fiber laser can be greatly improved by incorporating an additional non-reciprocal phase bias into the fiber loop [24–28], namely figure-9 fiber laser. As a non-reciprocal phase shift element is introduced, the transmission of the NALM can be tailored [26,27]. In this way, the transmission rate of low-intensity light is able to be enhanced, which is favorable for the self-starting operation of a fiber laser. In particular, because there is no need to accumulate the nonlinear phase shift difference through a long optical fiber, the cavity length of figure-9 fiber laser can be greatly reduced. Therefore, the design of figure-9 fiber laser also facilitates the generation of mode-locked pulse with higher repetition rate and lower noise. From the viewpoint of practical applications, it would be meaningful to explore phase biased NALM technique for development of 1.7 µm waveband pulsed laser sources. In this contribution, we report a 1.7 µm mode-locked Tm-doped fiber laser based on figure-9 cavity design. A 25 nm bandpass filter at 1.7 µm waveband was incorporated into the laser cavity to suppress amplified spontaneous emission (ASE) at long wavelength. After optimization of the net cavity dispersion, the mode-locked spectrum centered at 1726 nm with a 3-dB bandwidth of 20 nm could be obtained. In addition, we also achieved the flexible adjustment of the mode-locked spectral bandwidth by adjusting the phase bias dynamically with the rotation of wave plates.

2. Experimental setup

Figure 1 shows the schematic of the figure-9 Tm-doped mode-locked fiber laser operating at 1.7 µm waveband. A 2×2 fiber optical coupler (OC) with a splitting ratio of 50:50 was employed to connect the phase-biased NALM and the reflective arm. A 0.7 m long Tm-doped fiber (TDF, TmDF200, OFS) with a dispersion parameter of - 0.01 ps²/m is used as the gain medium, which is core pumped by an erbium-doped fiber laser (EDFL) operating at 1560 nm. The relatively short gain fiber was chosen to reduce the reabsorption effect. To realize the cavity dispersion compensation, a segment of 2.9 m ultrahigh numerical aperture (UHNA4) fiber with a dispersion parameter of 0.084 ps²/m was incorporated. Moreover, a non-reciprocal phase shifter composed of a Faraday rotator and two quarter-wave plates are employed to control the linear phase shift difference between the counter-propagating beams in the NALM. The bandpass filter with a 3-dB bandwidth of 25 nm and a center wavelength of 1734 nm is placed between the two collimators to suppress ASE at long wavelength. A polarization controller (PC) is incorporated therein for adjusting the polarization state of the propagating light. In addition, we incorporate a 1×2 OC₁ after the bass filter to obtain the spectrum without ASE, and the laser pulse is guided out by the 30% port of the OC₁. The reflective arm contains a gold-plated mirror with total reflection. The pigtails for all optical devices are single-mode fiber (SMF-28e). The cavity length of the fiber laser is about 8.7 m, corresponding to a fundamental repetition frequency of 23.36 MHz. Therefore, the total net cavity dispersion of the fiber laser is about 0.0089 ps², indicating that the fiber laser operates in a slightly normal dispersion region. The mode-locked spectrum and pulse-train are measured by an optical spectrum analyzer (OSA, Yokogawa AQ6375B) and an oscilloscope (Tektronix DSA71604B, 16 GHz) with a high-speed photodetector (New Focus P818-BB-35 F, 12.5 GHz). In addition, a commercial autocorrelator (Femtochrome, FR-103WS) is used to measure the pulse duration.

 

Fig. 1. Schematic of the 1.7 µm figure-9 Tm-doped fiber laser.

Download Full Size | PDF

3. Experimental results

When the pump power was increased to 850 mW, the self-started mode-locking operation of the fiber laser could be obtained under the proper rotation of the wave plates. The mode-locked threshold is high here, which can be improved by replacing the OC₂ with a higher ratio. Initially, the fiber laser was operating in multi-soliton regime owing to the overdriven nonlinear effect by high pump power. However, by virtue of the pump hysteresis effect [29], we were still able to achieve the single-soliton operation when the pump power was gradually reduced to 350 mW. The mode-locked performance of single-soliton operation is summarized in Fig. 2. The mode-locked spectrum of output 1 is plotted in Fig. 2(a), and the typical rectangular profile of dissipative soliton can be observed due to the net-normal cavity dispersion. The mode-locked spectrum is centered at 1726 nm with a 3-dB bandwidth of 20 nm. It should be noted that the core diameters of the SMF-28, TDF and UHNA4 fibers are different from each other. Thus, the interference between the core mode and the excited cladding mode may result in the modulation on the mode-locked spectrum [30,31]. For comparison, the mode-locked spectrum of output 2 is plotted in Fig. 2(b). It can be seen that the ASE appears at long wavelength, mainly because the NALM is bidirectional operation and the dissipative soliton amplified by TDF can directly reach the output 2 without passing the bandpass filter. The mode-locked pulse-train from output 1 is presented in Fig. 2(c). The pulse repetition rate, which is determined by the cavity length, is 23.36 MHz. Note that the soliton breathing behavior could be also observed at some specific pump power levels [32,33]. Correspondingly, the output power is 0.35 mW, indicating that the pulse energy is 15 pJ. The average output power of the dissipative soliton is a bit low here, which could be partially attributed to the high insertion loss of optical components used in our fiber laser. Figure 2(d) shows the measured autocorrelation trace. Here, the pulse duration is 4.08 ps under the assumption of Gaussian shape. Considering the bandwidth of the mode-locked spectrum, the mode-locked pulse is chirped and the pulse duration can be further compressed to be hundreds of femtoseconds. We also measured the radio frequency (RF) spectrum of the proposed 1.7 µm mode-locked fiber laser with a resolution bandwidth of 10 Hz, as shown in Fig. 2(e). It can be seen that the pulse repetition frequency is located at 23.36 MHz, corresponding to the fundamental repetition rate of the fiber laser. Furthermore, the signal-to-noise ratio (SNR) is about 58 dB, suggesting that the proposed fiber laser is operating in a stable single-soliton regime. In order to verify the stability of the mode-locked fiber laser, we recorded the laser output spectra at 5-minute intervals over 3 hours. As depicted in Fig. 2(f), no significant intensity fluctuation and wavelength drift occurred in the spectra, which further verifies the stability of the proposed 1.7 µm ultrafast fiber laser. In addition, we monitored the power fluctuation at room temperature over 3 hours. And the root-mean-square (RMS) of the output power is ∼0.45%, which further verifies the stability of the proposed 1.7 µm ultrafast fiber laser.

 

Fig. 2. Single pulse operation. (a) Mode-locked spectrum of output 1. (b) Spectrum of output 2. (c) Mode-locked pulse train, inset: pulse-train over 40 µs. (d) Autocorrelation trace of the pulse at output 1. (e) RF spectrum. (f) Output spectra in 3 hours.

Download Full Size | PDF

As mentioned above, the multi-soliton operation could be observed owing to the overdriven nonlinear phase shift caused by the high pump power. Therefore, we increased the pump power again and kept other cavity parameters fixed to investigate the dynamics of the multi-soliton operation in the 1.7 µm ultrafast fiber laser. When the pump power was adjusted to 365 mW, the fiber laser was operating in double-soliton regime, as can be observed in Fig. 3(c). Note that the mode-locked spectrum plotted in Fig. 3(a) is very similar to the single-soliton spectrum despite of the slightly higher intensity. This is due to the fact that the mode-locked spectrum measured by the OSA is the superposition of two solitons. Then, if the pump power was further increased to 380 mW, the fiber laser would operate in triple-soliton regime, as shown in Fig. 3(b) and Fig. 3(d). For triple-soliton regime, the mode-locked solitons are not equally distributed along the laser cavity. Nevertheless, the interval among the generated solitons could be tuned to be equal by carefully adjusting the polarization controller or pump power, namely harmonic mode-locking operation.

 

Fig. 3. (a) and (c) Spectrum and pulse-train of double pulses. (b) and (d) Spectrum and pulse-train of triple pulses.

Download Full Size | PDF

In our fiber laser, the rotation of the wave plates will lead to the variation of the linear phase shift difference between two counter-propagating beams [27]. Since the fiber laser is a non-polarization-maintaining one, the change of the linear phase shift difference will result in the variation of the transmission curve of NALM against wavelength. For better clarity, we have plotted the transmission curves with varying linear phase shift difference between two counter-propagating beams in NALM, as shown in Fig. 4. Therefore, by combining the bandpass filter and the variation of the NALM transmission curve, it is possible to adjust the 3-dB spectral bandwidth of the dissipative soliton emitted from the fiber laser. Figure 5 shows the tuning operation of the mode-locked spectral bandwidth by rotating the wave plates. It can be seen that the 3-dB bandwidth of the mode-locked spectrum varies from 7.5 nm to 20 nm, where the tuning process is reversible. Here, the center wavelength of the mode-locked spectrum changes slightly, which is partially caused by the wavelength drift corresponding to the highest transmittance during the tuning process, as indicated in Fig. 4. It should be noted that the fiber laser still maintains a stable single-soliton operation. However, as the transmission rate of NALM changes with the rotation of the wave plates, the output power of the mode-locked soliton also varies correspondingly. In addition, the maximum 3-dB bandwidth of the mode-locked spectrum is limited to ∼20 nm, which is majorly restricted by the bandpass filter.

 

Fig. 4. Transmission curves of filter and NALM at different wave plate angles.

Download Full Size | PDF

 

Fig. 5. Mode-locked spectra at different wave plate angles.

Download Full Size | PDF

4. Discussions

In the experiment, the mode-locked pulse was obtained at 1.7 µm waveband in a figure-9 Tm-doped fiber laser. Generally, the design of figure-9 fiber laser is favorable for the generation of high-repetition-rate pulse owing to the reduced cavity length [34]. However, in our fiber laser the repetition rate of the mode-locked pulse is only around 23 MHz, which could be attributed to the fact that the net optical gain of Tm-doped fiber at 1.7 µm waveband is not high. In this case, if the cavity length is further reduced for generating the mode-locked pulse with higher repetition rate, the fiber laser would be difficult to achieve the mode-locking operation. Thus, the dopant concentration and length of the Tm-doped fiber need to be further optimized, in order to achieve higher repetition rate pulse at 1.7 µm waveband. Note that the proposed 1.7 µm mode-locked fiber laser is constructed by the single mode fiber, which still suffers the environmental instability because of the polarization-dependent operation. It is also well known that the introduction of phase bias in NALM enables the fiber laser to operate in all-polarization-maintaining regime [26]. Therefore, for the purpose of robust mode-locking operation, the all-polarization-maintaining design of figure-9 fiber laser would be preferred for generation of 1.7 µm ultrashort pulse. In addition, the output pulse energy from the fiber laser is low when compared to other dissipative soliton fiber lasers operating in net-normal dispersion regime. In fact, the pulse energy is mainly limited by the pulse splitting owing to the overdriven nonlinear effect inside the laser cavity. Note that the core diameter of the UHNA4 fiber is only ∼2.2 µm. When the mode-locked pulse passes through the UHNA4 fiber, a large nonlinear phase shift can be generated. Therefore, more proper dispersion compensation component with low nonlinear coefficient is needed to enhance the output pulse energy.

5. Conclusion

In summary, we have demonstrated a 1.7 µm passively mode-locked figure-9 fiber laser. By virtue of the phase biased NALM, the passive mode-locking operation could be easily achieved in the fiber laser with proper pump power level. With an optimization of the cavity dispersion, the fiber laser delivers the mode-locked dissipative soliton with 3-dB bandwidth of 20 nm. Moreover, the spectral bandwidth of the dissipative soliton could be flexibly tuned from 7.5 nm to 20 nm by adjusting the orientation of the wave plates. The obtained results would pave the way for further development of highly reliable laser sources at 1.7 µm waveband, which might be interested to the communities dealing with the multiphoton deep-tissue imaging and ultrafast laser technology.