1. Introduction

Since graphene and other two-dimensional materials were isolated, several demonstrations of ultrashort pulses generation have been reported. Some nanomaterials have already been used as saturable absorber (SA) through its nonlinear properties, such as graphene and carbon nanotubes [1–6]. The molybdenum disulfide (MoS₂) is a transition-metal dichalcogenide semiconductor with crystal layers consisting of three alternating hexagonal planes of Mo (molybdenum) and S (sulfur). In the bulk form, MoS₂ crystal is an indirect semiconductor with bandgap of 1.29 eV (961 nm). However, by reducing the number of layers, the energy gap has a direct bandgap of 1.8 eV (689 nm) [7] for monolayer structure. A recent work demonstrated that the MoS₂ shown better saturable absorption response than graphene using an open-aperture Z-scan technique for ultrafast nonlinear optical properties. In this work, MoS₂ dispersions (34.4%) exhibit much stronger saturable absorption response than the graphene (16.5%) dispersions [8]. Also, the introduction of suitable defects can reduce the MoS₂ bandgap from 1.8 to 0.8 eV, achieving the absorption wavelength of 1550 nm [9]. Such nonlinear optical absorption exhibited by MoS₂ was attributed to the presence of edge states that generate sub-bandgap within the bandgap [10], showing its broadband saturable absorption from visible to infrared spectrum regions.

Usually, MoS₂ samples were fabricated via liquid phase exfoliation (LPE) [11] or chemical vapor deposition (CVD) growth [12]. In one of the LPE methods, few MoS₂ layers could be obtained via lithium intercalation by placing the MoS₂ crystal in a solution of butyl-lithium with hexane and then exfoliated by ultrasonication via deionized water. In the CVD method, sulfur is reacted with molybdenum forming thin films of MoS₂ on the SiO₂ wafer within a quartz tube with N₂ flow. In these two methods, different sizes, areas and MoS₂ layers can be obtained, but also with many crystal defects.

Recently, results about ultrashort pulse generation were demonstrated using MoS₂ as SA in Ytterbium-doped fiber lasers (YDFL) [11] and Erbium-doped fiber lasers (EDFL) [13].The property of SA and cavity dispersion could affect the pulse shape, such as stability and duration. A recent review shown the progress of few-layered MoS₂ based SA for ultrashort pulse generation [14, 15]. In YDFL, pulses of 656 [13] and 800 ps [11] were obtained. In both works, the MoS₂ samples were prepared by using LPE and deposited onto the fiber connector facet. In EDFL cavity, mode-locking results were reported by using MoS₂ samples prepared via CVD [12] and LPE [13, 16] and also transferred to fiber connector facet. Via evanescent field interaction, a high-order passively harmonic mode-locked EDFL was demonstrated by using a microfiber-based MoS₂ SA obtained via LPE, which generated pulses of 3 ps [17]. At the same line, pulses as short as 637 [18] (CVD) and 521 fs [19] (LPE) were demonstrated by using MoS₂ SA deposited onto the side polished of a D-shaped optical fiber.

In this work, we obtained mechanically exfoliated MoS₂ samples through to scotch tape technique [20]. The mechanical exfoliation of crystal materials provides purest 2D-materials with less defects from other methods such as lithium (LPE) and nucleation points (CVD). In addition, this technique can be used to produce high quality, single-crystal flakes and possibly works well as saturable absorber. The MoS₂ flakes were transferred onto the side-polished surface of a D-shaped optical fiber. By incorporating this sample as SA in an EDFL cavity, it was generated pulses as short as 200 fs and spectral bandwidth of 20.5 nm. This is the first demonstration of ultrashort pulses generation by using mechanically exfoliated MoS₂ SA in EDFL and the shortest pulse duration ever reported in literature.

2. MoS₂ sample preparation

For the preparation of mechanically exfoliated MoS₂ sample, we used the method as described in [20]. In Fig. 1, thin layers of water soluble polymer PVA (polyvinyl alcohol, 1.2 μm thickness) was deposited on glass substrate (1) and then PMMA (polymethyl methacrylate, 300 nm thickness) were applied (2) to form the stacked substrate (3). In sequence, the mechanically exfoliated MoS₂ flakes (4) were deposited onto this substrate (5) and used an adhesive tape as a support (6). The stacked substrate with the sample was immersed in deionized water bath for approximately 24 hours to remove the PVA layer and unstuck the MoS₂/PMMA film from the glass substrate.

 

Fig. 1 MoS₂ sample preparation setup. The stacked substrate contains a PVA (1.2 μm) and PMMA (300 nm) layers. The mechanically exfoliated MoS₂ onto this substrate and an adhesive tape used as support to the MoS₂/PMMA film.

Download Full Size | PDF

The MoS₂ flakes were transferred onto the polished surface of a D-shaped optical fiber with no distance from the fiber core to the polished surface and polishing length of 17 mm, and it was placed on a hot plate at 140 °C for one hour to improve the adhesion of MoS₂/PMMA film.

The Fig. 2(a) shows the MoS₂ flakes deposited onto the polished surface of the D-shaped optical fiber (10X objective lens). The characterization of this sample was made by using a confocal microscopy Witec Alpha 300R with a 532 nm laser wavelength and output power of 1.6 mW. The Raman spectroscopy mapping was performed through the analysis of frequency difference between the vibrational modes $E_{2g}^1$ and $A_{1g}$ [21] on the D-shaped optical fiber with the polished surface side up, as shown in Fig. 2(b). The measured Raman spectrum shown in Fig. 2(c) was used to confirm the presence of MoS₂ of the deposited flakes through the difference between $E_{2g}^1$and $A_{1g}$ modes, which in this case, it corresponded to bulk MoS₂ flakes.

 

Fig. 2 (a) Optical image of MoS₂ flakes deposited onto the polished surface of the D-shaped optical fiber using 10X objective lens, (b) Raman mapping of the marked area of 20 x 120 µm in Fig. 2(a) showing the MoS₂ in red color and (c) Raman spectrum of the MoS₂ with the difference between the vibrational modes $E_{2g}^1$and$A_{1g}$of 25 cm⁻¹.

Download Full Size | PDF

3. Polarization and EDFL Mode-locking performance results

An experimental setup as described in [22] was used to check the performance of the fabricated MoS₂ sample as a polarizer. A 1550 nm continuous wave laser was collimated through a 20X objective lens. Then, the beam was vertically polarized by a 50:50 polarization beam splitter (PBS) and its polarization direction was controlled by the half-wave plate. The polarized beam was coupled to another 20X objective lens. Before placing the sample in the experimental setup, we used a polarization controller to optimize the beam power.

The polarization relative extinction ratio (P) analysis of the fabricated sample was performed by rotating the angle of the half-wave plate. For polarization measurements, steps of 10° from 0 to 360° polarization were used. The polarization measurements were made with the D-shaped optical fiber without and with MoS₂/PMMA, as shown in Fig. 3. The polarization loss depends on the polarization state and varies according to the mode polarization component perpendicular (TM-like mode) or parallel (TE-like mode) to the MoS₂ plane. The polarization relative extinction ratio depends on the linear attenuation difference between the two perpendicular and parallel modes and can be calculated as follows [22],

 

Fig. 3 Transmitted power as function of the beam polarization angles through the D-shaped optical fiber: Without MoS₂/PMMA (black square line) and sample with MoS₂/PMMA film (blue circle line).

Download Full Size | PDF

In order to analyze the polarization performance of MoS₂ exfoliated flakes in the D-shaped optical fiber, we measured its performance only with the PMMA layer, and it showed a polarization relative extinction ratio value of 37%. As showed in [22], the PMMA layer thickness is a determining parameter to the polarization relative extinction ratio value because of PMMA high refractive index (n = 1.49), which shifts the optical mode direction and increases the light-material interaction. This can explain the polarization relative extinction ratio value of 37% of the sample with only the PMMA layer.

The Fig. 4(a) shows the ring cavity with the MoS₂/D-shaped optical fiber saturable absorber. The EDFL consists of a 2 meters Erbium doped fiber with average dispersion of −57 ps/km/nm pumped by a 980 nm laser diode (LD), 980/1550 nm wavelength division multiplexing and isolator (WDM/Isolator) to ensure unidirectional operation, a polarization controller to adjust the intracavity polarization for mode-locking optimization and a 15% output coupler. The total cavity length is 16.7 meters with accumulated dispersion of + 136 fs/nm and average dispersion + 8.17 ps/km/nm, appropriated to generate soliton-like pulses. The Fig. 4(b) shows that the laser continuous wave (CW) regime is initiated at threshold pumping power of 20 mW (blue range). By increasing the pumping power to 40 mW, the mode-locking regime at single pulse operation is reached, and can be extended up to 120 mW (orange range). Above 120 mW pumping power, the laser enters into mode-locking regime at multiple pulses operation, presenting instability and power fluctuations (outside the orange range).

 

Fig. 4 (a) Erbium-doped fiber laser experimental setup using MoS₂/PMMA sample onto a D-shaped optical fiber. (b) Output power as a function of pump power with CW and mode-locking threshold.

Download Full Size | PDF

The Fig. 5 shows the measured output pulse duration and bandwidth as a function of cavity length and accumulated dispersion. The high MoS₂ SA performance in the EDFL was verified by managing the intracavity accumulated dispersion of the laser reducing the cavity length from 16.7 to 13.7 m. As we can see in Fig. 5, for 16.7 m cavity length and + 136 fs/nm accumulated dispersion, pulse duration of 265 fs and bandwidth of 10 nm were obtained. By reducing the cavity length to 14.7 m, corresponding to 105 fs/nm accumulated dispersion, the pulse duration decreased to 250 fs and the bandwidth increased to 12 nm.

 

Fig. 5 Output pulse duration as a function of accumulated intracavity dispersion and cavity length in mode-locking regime (red line). Bandwidth as a function of cavity length and accumulated intracavity dispersion (blue line).

Download Full Size | PDF

The best performance of EDFL in the mode-locking regime was achieved at pump power of 110 mW, which generated the spectrum centered at 1560 nm with spectral bandwidth of 20.5 nm [Fig. 6(a)], pulse duration of 200 fs [Fig. 6(b)] and fundamental cavity frequency of 14.53 MHz [Fig. 6(c)]. The laser output power was 1 mW, resulting in intracavity peak power of 2.30 kW and time-bandwidth product of 0.505. Considering the theoretical value of a transform-limited (0.315 for sech² pulse), it is possible to further reduce the pulse duration to approximately 125 fs. The best mode-locking results were achieved with 13.7 m of cavity length that corresponded of 85 fs/nm accumulated dispersion. We also measured the corresponding RF spectrum, which indicated a high mode-locking stability of the laser at fundamental cavity frequency with a high signal-to-noise ratio (SNR) up to 84 dB [Fig. 6(d)] and a high quality of MoS₂ sample. By comparing the measurements with previously works, these are the best EDFL mode-locking results ever reported in literature, using MoS₂ SA in all-fiber configuration. For future works, it is possible to reduce the laser cavity length and consequently the accumulated dispersion for further decreasing the pulse width.

 

Fig. 6 Mode-locking pulses measurements (a) Laser spectrum (inset – log scale). (b) Autocorrelation trace of output pulses. (c) Oscilloscope trace and (d) RF spectrum measure around the fundamental repetition rate of 14.53 MHz.

Download Full Size | PDF

4. Conclusion

We presented for the first time the ultrashort pulse generation of a mode-locked EDFL by using mechanically exfoliated MoS₂ deposited onto the side polished surface of a D-shaped optical fiber. With a sample of 97.5% polarization relative extinction ratio and incorporated as SA in an EDFL cavity, the laser generated pulses as shorter as 200 fs, being the shortest pulse duration ever reported in literature and the best EDFL mode-locking performance achieved with MoS₂ all-fiber based saturable absorber.