1 Introduction

Over last few years, two-dimensional (2D) layered materials such as graphene (Geim and Novoselov 2017; Bonaccorso et al. 2010; Yamashita 2012), topological insulator (TI) (Hasan and Kane 2010; Culcer 2011) and transition metal dichalcogenides (TMDs) (Bonaccorso and Sun 2014; Woodward et al. 2015) have attracted increasing attention for saturable absorbers (SAs) due to their ultrafast recovery time (Popa et al. 2010), broadband absorption (Sun et al. 2010; Luo et al. 2015), controllable modulation depth (Mao et al. 2015). Graphene, the first discovered 2D material, possessing a Dirac-like electronic band structure (Geim and Novoselov 2017), has been identified a promising SA for broadband ultrafast pulse generation in various laser systems (Sun et al. 2009, 2016). However, the weak optical absorption in graphene [2.3% of incident white light for monolayer (Nair et al. 2008)] limits the modulation depth and potential applications. TIs have insulating bulky states with indirect bandgap of 0.3 eV (Chen et al. 2009) and gapless surface states, which have been experimentally demonstrated for possessing broadband saturable absorption properties (Zhao et al. 2012; Chen et al. 2013). Nevertheless, TIs also suffer the deficiency of complicated preparation process, which limits their application in optoelectronic devices.

In addition, TMDs [e.g. MoS2 (Zhang et al. 2014), MoSe2 (Mao et al. 2015), WS2 (Mao et al. 2015)] have also been investigated as SAs. The intralayer transition metal atoms and chalcogen atoms are bonded covalently, and the interlayers are coupled via weak van der Waals force (Xu et al. 2013). On account of the interlayer coupling, bulk TMDs have an indirect band-gap while mono-layer or few-layer TMDs turn out to be a direct band-gap semiconductor (Mak et al. 2010). Due to these features, TMDs are regarded as a promising candidate for optoelectronic devices. However, the TMDs, possessing 2H lattice structure, are in-plane isotropic. Most recently, black phosphorus (BP), an in-plane anisotropic layered crystal, has drawn much attention for the broadband saturable-absorption characteristics (Lu et al. 2015; Chen et al. 2015). The direct bang gap of BP materials, sensitive to the thickness, can be modulated from 0.3 to 1.5 eV (Low et al. 2014; Das et al. 2014). However, the unsatisfactory environmental stability of BP materials restricts its applications in electronics and photonics. However, it is still unknown that if there are another other two-dimensional materials, in which have the same atomic structures and similar optical properties as the above mentioned materials. If it can be done, the territory of the SAs will be expanded.

Here, we present a member of the TMDs family, rhenium disulphide (ReS2) which possesses strong in-plane anisotropic. (Tongay et al. 2014; Ho et al. 1999; Chenet et al. 2015; Wolverson et al. 2014). Compared to group VI TMDs of MoS2, WS2, ReS2 belongs to VII TMDs, in which rhenium atom contains an extra electron in the d-orbital (Yang et al. 2015; Fujita et al. 2014). ReS2 crystals with triclinic symmetry are optical biaxial (Friemelt et al. 1993; Dumcenco et al. 2008), however, other hexagonal TMDs are optically uniaxial and the optical axes of these TMDs are perpendicular to the atomic plane (Eriksson 2009). ReS2 possesses large difference with group VI TMDs. One of the key features of traditional TMDs (e.g., MoS2, WS2 and MoSe2) is that they go through crossover from indirect bandgap in bulk to direct bandgap in monolayer (Tongay et al. 2014; Lee et al. 2012). As a result, TMDs in monolayer can be employed as SAs. Exceptionally, ReS2 remains a direct bandgap semiconductor independent with the number of layers, and the Raman spectra of different layers are almost the same (Tongay et al. 2014) due to the weak interlayer interaction. Particularly, modulating the distance between adjacent layers also does not change the optical absorption and Raman spectra (Tongay et al. 2014).

To date, most of the research about ReS2 focus on its electronic properties, as for the optical performance and applied studies were mainly at wavelengths less than 830 nm (Cui et al. 2017), and the study of its nonlinear optical response was barely touched. Very recently, the nonlinear saturable absorption of ReS2 was demonstrated at infrared communication wavelength, they transferred the ReS2 made by chemical vapor deposition (CVD)onto the D-shaped fiber (Cui et al. 2017). However, this method requires complex and expensive fabrication. In this work, mechanically exfoliated few layer of ReS2 flakes were transferred on fiber end to form a sandwiched SA device, then we employed the ReS2 SA device to realize a Q-switched fiber laser at 1.5 μm. Compared with BP based SAs in few layers, ReS2 is advantageous for good environmental stability. The generated pulse presents the shortest pulse duration of 2.1 μs and highest average output power of 2.48 mW at 1532 nm.

2 Material preparation and characterization

ReS2 flakes were mechanically exfoliated from high-purity (> 99.995%) bulk ReS2 (purchased from HQ Graphene) with scotch tape. Raman spectroscopy (Renisha win Via mirco-Raman system, 633 nm) was used to characterize the ReS2 flakes, the result is presented in Fig. 1a. Owing to the reduced crystal symmetry, ReS2 shows a more complex Raman spectrum than TMDs. 10 modes can be observed in the frequency range from 100 to 400 cm−1, and the two main Raman peaks (at 163 and 213 cm−1) are in agreement with previous investigations (Tongay et al. 2014).The peak around 521 cm−1 is a fingerprint of the silicon substrate (Lee et al. 2012). Noteworthy that the intensities and frequencies of the Raman modes do not change in a repeated measurement after several months, indentifying the good environmental stability of ReS2. The exfoliated flakes were then transferred onto the fiber connector tip, covering the fiber core completely, as showed in Fig. 1b. The fiber connector with the transferred ReS2 films was connected with a fresh fiber adapter to form an all-fiber integrated saturable absorber (SA) device. Nonlinear absorption of the ReS2 SA was carried out with a home-made ultrafast source (447 fs pulse duration, 11 MHz repetition rate). From the fitting curve as shown in Fig. 1c, we notice that, before input intensity reaches ~ 150 GW/cm2, the red curve presents linear variation, and after ~ 150 GW/cm2, absorption coverts to nonlinearity.

Fig. 1
figure 1

a Raman spectrum of ReS2. b Image of transferred ReS2 layers on the fiber tip. c Nonlinear power-dependent absorption of ReS2, the vertical axis means normalized transmittance

Full size image

3 Experimental results and discussion

Figure 2 shows a schematic configuration of the fiber laser setup. A 0.8 m Erbium doped fiber (EDF, SM-ESF-7/125) was used as gain medium, which was pumped by a 980 nm laser diode (LD) via a 980/1550 nm wavelength division multiplexer (WDM). A polarization independent optical isolator (ISO)was incorporated to force the unidirectional light propagation. The laser was coupled out through the 10% port of a 10:90 output coupler (OC). An optical spectrum analyzer (OSA) was used to measure the output spectrum. An ultrafast photodetector (UPD-35-IR2-FR) connected with a broadband (1 GHz bandwidth) digital oscilloscope (Agilent DSO9104A) was employed to characterize the pulse train. A polarization controller (PC) was used to control the polarization state of the laser cavity. All the pigtail fibers and passive fibers of the laser cavity is standard telecom single mode fiber (SMF-28e). The total length of the cavity length is approximately 9.15 m.

Fig. 2
figure 2

Experimental setup of the fiber laser Q-switched with ReS2SA. LD laser diode, WDM wavelength division multiplexer, EDF erbium-doped fiber, ISO polarization-independent isolator, SMF single-mode fiber, PC polarization controller

Full size image

In the fiber cavity with ReS2 SA, continuous wave laser was emitted at the pump power of 34 mW, which was subsequently switched to Q-switched pulses when we increased the pump power to 70 mW. Stable pulse train was obtained at the pump power range between 110 and 330 mW. At the pump power over 330 mW, strong signal jitter dominates and destroys the Q-switched pulse, the Q-switched pulses could be attained again just by decreasing the pump power. This is because due to the over-saturation of ReS2 SA rather than the thermal damage (Yu et al. 2014; Li et al. 2015b). Figure 3 illustrates the performances of the Q-switched laser. Output spectrum was measured at the pump power of 260 mW, showing a central wavelength of 1532 nm, with the full width at half maximum (FWHM) of 5 nm, as depicted in Fig. 3a. The output repetition rate and pulse duration are pump power dependent, a typical fingerprint of Q-switching. The reason of that is while the pump power increases, larger gain is provided to saturate the SA, therefore with the increase of the repetition rate, the pulse duration reduces (Li et al. 2015a). In this experiment, the repetition rate increased from 43 to 64 kHz, and the pulse duration declines from 7.4 to 2.1 μs, when the pump power increased from 110 to 330 mW, as show in Fig. 3b. To confirm the stable Q-switching operation, a photodetector and an oscilloscope were used to observe the pulse train in time domain, the pulse duration is 2.57 μs and the pulse period is 17.5 μs (corresponding to a repetition rate of 57.1 kHz), as can be seen in Fig. 3c. The relation between the laser output power and pump power is depicted in Fig. 3d, the slope efficiency of the Q-switched laser is 8%. And with the increasing of the pump power, the output power monotonously increases to a maximum value of 2.48 mW (at 330 mW pump power). Figure 3e shows the measured radio frequency spectrum of the Q-switched pulse. 52.5 dB signal to noise ratio indicates a good stability of the pulse train. For Q-switched lasers, one of the key parameters is pulse energy, which is also linearly dependent on the pump power, as shown in Fig. 3f. The maximum pulse energy is 38 nJ. This is similar with the typically Q-switched with other 2D nanomaterials, such as graphene and MoS2. To ensure the Q-switching is caused by ReS2, we then removed the SA from the cavity. Q-switched pulse disappears, what we can observe from the oscilloscope is a smooth line by adjusting the pump power and polarization controller, as shown in Fig. 4. This means our ReS2 SA is the crucial device for Q-switching operation.

Fig. 3
figure 3

ReS2 Q-switched fiber laser results: a output spectrum (With YOKOGAWA AQ6370C Optical spectrum analyzer). b Pulse duration and repetition rate as a function of pump power. c Output pulse train. d Average output power versus the pump power, the slope efficiency of the Q-switched laser is 8%. e Radio-frequency spectrum (With Keysight N9000A CXA Signal Analyzer). f Pulse energy as a function of pump power

Full size image
Fig. 4
figure 4

The oscilloscope trace after removing the ReS2 films

Full size image

4 Conclusion

In conclusion, a simple and easy operation ReS2 SA is fabricated, which performs excellent optical property. Our experimental results prove that it is feasible and suitable for Q-switched lasers. The experimental results show that the pulse energy is over 38 nJ and the pulse duration of 2.1 μs at the central wavelength of 1532 nm. ReS2 SAs show promising potential to be Q-switchers for fiber laser at 1.5 μm wavelength. In the further studies, we will try to improve the ReS2 and the laser structure to obtain narrower pulse duration.