Liquid phase exfoliation (LPE) is widely used in the fabrication of 2D materials because of its simplicity and effectiveness.^[19,43] In our experiment, LPE was used to prepare few-layer MoSe₂ nanosheets. Initially, 200 mg of the bulk MoSe₂ was mixed with 40-ml N-methyl pyrrolidone (NMP) and the mixture was treated by one-hour ultrasonication to break the bulk MoSe₂ into layered flakes. The suspension obtained was then centrifuged for 90 min at a rotation speed of 3000 rpm. After the centrifugation, the upper clear liquid from the suspension was collected as the few-layered MoSe₂ dispersion.

To verify the effectiveness of the liquid phase exfoliation of MoSe₂, the Raman spectra of the dispersion were tested. The MoSe₂ dispersion was dropped on a glass substrate and dried and the Raman spectra were measured using an excitation wavelength of 488 nm. The result obtained is shown in Fig. 1. As can be seen in the Raman spectra, the A_1g, , and Raman peaks are located at 239 cm⁻¹, 284.3 cm⁻¹, and 350.3 cm⁻¹, respectively, confirming the existence of layered MoSe₂ with 1–5 layers.^[45] The absorption spectra were recorded in a range of 500–1100 nm using a Lambda 950 UV–VIS–NIR spectrophotometer. As shown in Fig. 2, the two peaks at 695 nm and 800 nm further confirm the existence of layered MoSe₂.^[43]

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. Raman spectra of the MoSe2 nanosheet.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. Optical absorption of MoSe2 nanosheet.

The ratio of evanescent field power to total propagation power is determined by the waist diameter of the tapered fiber. The mode field distribution at the waist of the tapered fiber was simulated using finite element analysis. The wavelength and refractive index of the fiber cladding were set to be 1.55 μm and 1.45, respectively. Air, with a refractive index of 1, was taken as the medium surrounding the waist of the tapered fiber. The simulated result is shown in Fig. 3. It can be seen from the results that the ratio of the evanescent field power to the total power increases as the waist diameter of the tapered fiber decreases. To guarantee the effective interaction between the evanescent field and the material surrounding the fiber waist, the waist diameter needs to be small. Under the given experimental conditions, it is difficult to fabricate a tapered fiber with a relatively small loss and a waist diameter less than 3 μm. Therefore, the tapered fiber waist was taken to be 3 μm. The area of power of the evanescent field was about 1.2%, according to the simulation results.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. Simulation result of relationship between the ratio of evanescent field power to total propagation power and tapered fiber waist diameter.

The tapered fiber used in our experiment was prepared by stretching a segment of a standard single mode fiber (SMF-28) using an oxyhydrogen flame. The waist diameter of the tapered fiber was measured as ∼ 3 μm and the insertion loss as 0.4 dB at 1550 nm. The insertion loss was likely caused by the surface contamination of the fiber and nonuniform heating during tapering.

The optical deposition method was used to deposit the layered MoSe₂ onto the tapered fiber waist. A small volume of the dispersion was dropped on the tapered fiber waist. The tapered fiber was connected to a continuous-wave (CW) fiber laser at a wavelength of 1560 nm and a power of 26 mW. When the light of the CW laser propagated in the tapered fiber, the leakage of light field appeared in the waist region of the tapered fiber which was immersed in the dispersion droplet. Owing to the optical tweezer effect and light induced swirl and convection, the MoSe₂ nanosheets were trapped and deposited onto the tapered fiber. During the deposition, the tapered fiber was taken out of the droplet to measure the insertion loss at regular intervals. When the insertion loss rose to 4.8 dB, the laser was turned off and the deposition was completed. Using a microscope, the uneven deposition of several-micron-sized MoSe₂ nanosheets can be seen on the fiber waist with a distribution length of about 100 μm. Schematic illustrations of the optical deposition method and the tapered fiber deposited with layered MoSe₂ are shown in Figs. 4(a) and Fig. 4(b), respectively.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. Schematic illustrations of (a) the principle of optical deposition method and (b) a MoSe2 SA based on a tapered fiber.

The nonlinear transmission of the MoSe₂ SA was measured using a balanced twin-detector system. The laser source used was a commercial passively mode-locked erbium-doped fiber laser with a pulse width of ∼ 200 fs and a central wavelength of 1557 nm. The measured result is shown in Fig. 5.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. Nonlinear transmission of MoSe2 SA based on tapered fiber.

Because of the limitation of the output power of our laser source, the MoSe₂ SA was not fully saturated in the measurement. The measured modulation depth of the MoSe₂ SA was around 0.8% and the unsaturated loss around 52%. These values are comparable to those obtained in reported works. As shown in Fig. 5, the highest input intensity is about 3900 MW/cm². With the decrease of the input power, the transmittance of the SA can repeat the data recorded during the rise of the input power. This means that the MoSe₂ SA has a damage threshold higher than 3900 MW/cm².

An erbium-doped mode-locked fiber laser with MoSe₂ SA was set up as shown in Fig. 6. A laser diode operating at a wavelength of 980 nm was connected to a 980/1550 wavelength division multiplexer (WDM). A polarization-independent isolator (ISO) was used to ensure unidirectional operation. A polarization controller (PC) was spliced in the cavity to control cavity polarization. The output pulse was coupled out of the cavity using the 20% fiber of the 20/80 output coupler (OC). The whole cavity consisted of 0.65-m erbium-doped fiber (Nufern SM-ESF-7/125) and about 12-m single-mode fiber (SMF-28). The group velocity dispersions (GVDs) of the erbium-doped fiber and SMF-28 were −0.02 ps²/m and −0.022 ps²/m, respectively at 1550 nm. Using finite element analysis, the GVD of our tapered fiber waist was calculated to be −0.166 ps²/m, a value that is about 8 times higher than the one before the tapering. Despite the high GVD of the tapered waist, its short length makes the dispersion of the tapered fiber very small. Therefore, its effect on the laser operation is negligible. The net cavity dispersion was estimated as −0.278 ps².

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. Experimental setup of mode-locked fiber laser with MoSe2 SA.

When the pump power reached 30.7 mW, the mode-locking pulse could be easily obtained with proper adjustment of the PC with an output power of 0.2 mW. The laser could operate in the fundamental mode-locking state until the pump power reached 79.8 mW. Further increment in the pump power caused the mode-locked laser to operate in an unstable multi-pulse mode-locking regime. The output spectrum measured using an optical spectrum analyzer (OSA Yokogawa AQ6370C) is shown in Fig. 7.

Fig. 7.

	Figure Option ViewDownloadNew Window
	Fig. 7. Output spectrum.

When the laser is operated under the pump power of 37 mW, the output spectrum is centered at 1558.35 nm with a 3-dB bandwidth of 2.9 nm. The spectrum has typical Kelly sidebands, indicating that the output pulse is a traditional soliton. As can be seen in the output spectrum, a CW component exists at the wavelength of 1556 nm. A similar phenomenon was also observed in reported works with Bi₂Te₃,^[24] SWNT,^[46] and MoS₂.^[47] Figure 8 shows the output pulse train measured with a 1 GHz digital oscilloscope (Tektronic MDO 3102). The interval of the pulses train is 61.2 ns, corresponding to a repetition rate of 16.33 MHz and a cavity length of 12.67 m, thereby proving the single-pulse operation of the mode-locked laser. As shown in the figure, the peak intensity of the output pulses is not constant. These fluctuations were likely caused by a fraction of output laser fed back to the laser cavity.^[48] The corresponding radio frequency (RF) spectrum was measured with a resolution bandwidth (RBW) of 50 Hz, as shown in Fig. 9. The pulse repetition rate was 16.33 MHz while the signal-to-noise ratio was 52 dB.

Fig. 8.

	Figure Option ViewDownloadNew Window
	Fig. 8. Output pulse train.

Fig. 9.

	Figure Option ViewDownloadNew Window
	Fig. 9. RF spectrum.

Prior to our investigation, there have been several reports on erbium-doped mode-locked fiber lasers realized using TMD-based saturable absorbers. Details of these reports are listed below. Comparable results were obtained in our tapered-fiber MoSe₂ SA experiment as outlined at the bottom of the list.

The duration of the output pulse could not be measured directly because the output power was too low for the autocorrelator used in the experiment. However, it can be estimated using existing information. Based on the 3-dB bandwidth of the output spectrum shown in Fig. 7, the transform limited pulse duration can be calculated as 879 fs. Moreover, the output pulse obtained in the experiment is the traditional soliton, meaning that the pulse is slightly chirped. Therefore, considering the transform limited pulse duration, the slight chirp of the pulse, and the pulse width of the results with similar spectral bandwidth listed in Table 1, the pulse duration can be estimated as around 1 ps.

Table 1.

Table 1.

Table 1. Output performance of reported erbium-doped mode-locked fiber lasers incorporating with TMD-based saturable absorbers. TF is tapered fiber. . Ref. SA form Spectral width/nm Pulse duration/ps Modulation depth/% [49] MoS2-TF 2.47 – 2.82 [50] WS2-TF 6.75 0.609 7.8 [21] WSe2 film 2.1 1.25 0.5 [21] WSe2-SPF 2 1.31 0.3 [22] MoTe2 film 2 – 0.9 [22] MoTe2-SPF 2.4 1.2 1.8 [22] WTe2 film 2 – 1.5 [22] WTe2-SPF 2 – 2.8 [51] ReS2 film 2.47 1.2 0.14 [52] SnS2 film 6.09 0.623 4.6 [19] MoSe2 Film 1.75 1.45 0.63 [43] MoSe2 film 7.8 0.58 7.3 [42] MoSe2-SPF 5.4 0.798 5.4 [21] MoSe2-SPF 2.3 1.09 1.4 [21] MoSe2 film 2.2 1.18 0.4 This work MoSe2-TF 2.9 ∼ 1 0.8

Table 1. Output performance of reported erbium-doped mode-locked fiber lasers incorporating with TMD-based saturable absorbers. TF is tapered fiber. .