Black phosphorus (BP), a typical 2D material, recently has gained significant attention in the communities of electronics and optoelectronics ^[1–3]. It features a unique layer-dependent direct bandgap ^[4], which can be adjusted from ∼0.3eV (bulk) to ∼2eV (monolayer), right filling up the bandgap interval between 0 eV of graphene and large values of transition metal dichalcogenides (TMDCs). The thickness of sensitive direct bandgap is favorable for its optoelectronic applications covering from the near- to mid-infrared spectrum. Recently, a series of investigations revealed that the saturable absorption of BP can span from 0.4 to 2.8 μm, indicating its potential for pulse generation ^[5]. Lately, BP has also been developed as an optical saturable absorber (SA) for both passively Q-switched and mode-locked fiber lasers at 1.55 μm with excellent performance ^[6–8].

Despite these favorable properties of BP, a main drawback is its long-term stability ^[9]. BP is sensitive to water and oxygen under ambient conditions because of the high activity of phosphorus atoms and the fragile atomic binding ^[10], resulting in compositional changes and consequent degradation of the electronic and optical performance ^[11–15]. This severely hinders its applications in flexible electronics, photoelectrons, and electrochemical applications.

To address this drawback, several protection methods have been reported, including covalent bonding with aryl diazonium ^[16] and carbon free radicals ^[17], introduction of fluorine adatoms ^[18], intercalation of alkali metal hydrides ^[19], metal-ion modification ^[20], surface coordination of titanium sulfonate ligand ^[17], and encapsulation by graphene shell ^[14], hexagonal boron nitride (BN) ^[11,21], graphene oxide ^[22], AlOx ^[15,23], organic monolayers, etc. ^[24]. Although the high stability of BP has been reached, those strategies bring either the advanced synthesis of functional groups or post-treatment of prepared BP nanomaterials, which result in the technical complexity for fabricating BP ^[25].

A demand for highly stable 2D nanomaterials drives researchers to search for other high-quality BP analog materials with enhanced long-term stability. Fortunately, a group of layered group IV monochalcogenides, such as SnSe ^[26], GeSe ^[27], GeS ^[28], and SnS ^[29] have puckered structures similar to BP. Bulk SnSe has shown great potential in thermoelectric applications ^[30,31]. Recently, inspired by the unique property of anisotropic structure in phosphorene, 2D monochalcogenides have attracted great research interest ^[32–35], especially for their thermoelectric properties ^[31,36]. Furthermore, compared with the single element in phosphorene, these monochalcogenides are composed of two different types of elements and thus possess different electronegativity ^[37]. These features can lead to more complicated physics phenomena than that of phosphorene. Indeed, some unique properties, such as giant piezoelectricity ^[38] and multiferroic with coupled ferroelectricity and ferroelasticity ^[33,34] have been theoretically proposed. At the same time, experimental investigations on 2D monochalcogenides are in progress ^[26,39]. SnS exhibits strong anisotropic mechanical and optoelectronic features ^[37,39–44]. The presence of weak van der Waals force between SnS layers shows no dangling bonds, giving a chemically inert surface ^[45]. As a result, SnS shows chemical stability even in acidic solution ^[46], a high melting point of 880°C, and a boiling point of 1230°C. In view of the environment, the Sn and S elements are relatively abundant, and they are safe elements ^[47]. These favorable properties make SnS a potential material for various applications, including photovoltaics ^[48,49], solid-state batteries ^[50,51], thermoelectricity ^[52], humidity sensors ^[53], supercapacitor electrodes ^[54], solar hydrogen production ^[55], etc.

SnS possesses unique dual direct and indirect bandgaps, with the direct bandgap varying from 1.30 to 1.39 eV and the indirect bandgap from 1.07 to 1.25 eV ^[42]. In another report, the bandgap of SnS can be adjusted from 1.03 to 1.65 eV ^[56]. Moreover, it possesses large absorption coefficients (>104cm−1), spanning the ultraviolet to near-infrared (NIR) regions ^[29,56]. Its carrier mobility was predicted to be tens of thousands of cm2·V−1·s−1 ^[29], which is even faster than that of BP. Owing to the suitable and tunable bandgap and fast carrier mobility of SnS, excellent nonlinear optical properties can be anticipated. However, investigation of such properties and optoelectronic applications in NIR regions is still in its infancy stage.

In this work, the BP-analog SnS was thus investigated as a new type of effective SA. The high stability of as-prepared SnS nanosheets (NSs) was confirmed through systematic characterizations. Nonlinear absorption was characterized at 800 and 1550 nm, respectively, both exhibiting excellent SA properties. High and tunable modulation depth makes it flexible as a Q switcher and mode locker. The carrier dynamics, presented by two time components, namely, electron thermalization within the band (873 fs) and recombination of the carriers (96.9 ps), indicates the promising potential of SnS for mode locking. The ultrafast (656 fs) pulse generation in fiber lasers based on SnS NSs was experimentally demonstrated at 1560 nm with high stability. This primary study opens a new avenue for advanced electrophotonics by using BP-analog 2D nanomaterials, suggesting the great promise of group IV monochalcogenides as the substitutions.

SnS NSs were prepared with the liquid-phase exfoliation (LPE) method. N-methyl-2-pyrrolidone (NMP) was normally chosen to exfoliate SnS ^[56]. The solvent of isopropyl alcohol (IPA) was chosen to exfoliate SnS in this work because of its lower boiling point, leading to easy removal. In Fig. 1(a), the transmission electron microscopy (TEM) image shows SnS NSs with a main 2D size of ∼50nm and a thickness of less than ∼6nm, as shown by the atomic force microscopy (AFM) image [Fig. 1(b)]. The (210) and (111) planes with interatomic space of 0.32 and 0.30 nm are determined by the observation of high-resolution TEM (HRTEM) [Fig. 1(c)] ^[57]. Fast Fourier transformation (FFT) of the crystal lattice shows the expected crystallographic lattice reflections [inset of Fig. 1(c)]. The selected-area electron diffraction (SAED) image in Fig. 1(d) suggests that the crystal phase of the SnS NSs remains after the LPE procedure, corresponding to the FFT calculation of the crystal lattice.

The crystal phases of SnS NSs and bulk SnS were compared by employing X-ray diffraction (XRD) and Raman techniques. In Fig. 1(e), both the SnS NSs and bulk SnS exhibited a similar typical XRD pattern. For Raman spectroscopy of bulk SnS, three clear Raman peaks were shown from 100 to 500cm−1. The B3g mode locates at 161.8cm−1, and the A_g mode is responsible for the peaks at 185.5 and 219.9cm−1. The peak at ∼300cm−1 comes from SnS2 owing to the surface oxidation ^[57,58]. For SnS NSs, the peaks at 163.7cm−1 (B3g) and 190.9cm−1 (A_g) indicate a Raman shift compared with the bulk SnS, especially for the A_g mode (from 185.5 to 190.9cm−1). It can be concluded that the A_g mode shows the more sensitive layer-dependent behavior in the two vibrational modes because it is along the armchair direction. The energy-dispersive X-ray spectroscopy (EDX) mapping in scanning transmission electron microscopy (STEM) mode presented the composition of Sn and S elements in SnS NSs [Fig. 1(g)]. All measurements determine the composition and crystal phase of the exfoliated SnS NSs.

Concerning the optical applications, the as-exfoliated SnS NSs must possess sufficiently high stability. The stability of SnS NSs was systematically characterized by absorbance spectrum, XRD pattern, Raman spectroscopy, and X-ray photoelectron spectroscopy (XPS). The absorbance of SnS NSs remains stable in 30 days, as shown in Fig. 2(a). Both the XRD pattern and Raman spectrum in 30 days present similar peaks [Figs. 2(b) and 2(c)]. For the XPS characterization [Fig. 2(d)], Sn 3d 5/2 and 3d 3/2 orbitals are responsible for the strong peaks at 485.6 and 494.1 eV, respectively. After 30 days, there was no obvious energy shift. All these characterizations indicate the high stability of SnS NSs in ambient conditions.

The steady-state optical response is determined by ultraviolet-visible-near-infrared (UV-vis-NIR) absorption from 350 to 1680 nm, which is measured in IPA solution when the influence of solution has been removed (Fig. 3). From 350 to ∼800nm, the absorbance decreased quasi-linearly, and the absorbance remained almost stable from ∼800 to ∼1100nm. It was interesting to find that there were two absorbance peaks centralized at 1180 and 1480 nm, corresponding to the bandgap of 1.05 and 0.84 eV [Fig. 3(a)]. According to the Tauc plot [Fig. 3(b)], the bandgap of SnS NSs was fitted to be 1.2 eV, similar to the previous report ^[56]. Therefore, the two above-mentioned absorbance peaks belonged to the subbandgap absorption, which may be caused by crystal defect. These two absorbance peaks strongly suggested that the SnS NSs might be a promising optical material in the telecommunication band.

Broadband nonlinear optical absorption of SnS NSs is then characterized by the well-developed Z-scan technique ^[5,59]. Under high optical intensity, the total nonlinear absorption could be attributed to various nonlinear absorption mechanisms, including ground state bleaching (GSB), excited state absorption (ESA), two-photon absorption (TPA), and multiphoton absorption (MPA), etc. To characterize nonlinear absorption of few-layer SnS NSs, the transmittance versus the z axis was recorded. Strong saturable absorption was observed at both 800 and 1550 nm (Fig. 4). Materials with such significant properties can be used as an efficient optical modulator for Q switching or mode-locking operations.

The absorption coefficient of the SnS NSs can be written as α=α0+αNL,(1)where α0 is the linear absorption part, αNL=βI is the nonlinear part (β is the nonlinear absorption coefficient and I is the incident optical intensity). The linear absorption coefficient is independent of the optical intensity, which can be obtained at low intensity using the equation α0=ln(LT)L, where LT is the linear transmittance (shown in Table 1), and L is the length of light path traversing SnS dispersion. αNL is strongly dependent on intensity and β can be deduced by fitting normalized transmittance Tnorm(z) versus z axis using the equation ^[60] Tnorm(z)=ln[1+q0(z)]/q0(z),(2)where q0(z)=βI0Leff(1+z2/z02), I0 is the on-axis optical intensity, Leff=[1−exp(−α0L)]/α0 is the effective propagation length, and z0 is the Rayleigh length. Normalized transmittance versus the pulse energy and the corresponding fitting curves at 800 nm are shown in Fig. 4(a); β is fitted to be −(50.5±3.4)×10−3cm/GW at the maximum pulse energy of 1.30 μJ. Considering the contribution of third-order nonlinear susceptibility to saturable absorption, the imaginary part of third-order susceptibility (Im χ(3)) has a positive correlation with β ^[61], Imχ(3)=2ε0c2n023ωβ, where c is the light speed in vacuum, ε0 is the vacuum permittivity, n0 is the linear refractive index (n0≈1.373 for IPA solvent ^[62]), and ω is the angular frequency of the light. Im χ(3) is calculated to be 4.25×10−21m2/V2 at the pulse energy of 1.30 μJ. For 1550 nm, the Z-scan results and fitting curves are shown in Fig. 4(c), and the corresponding β and Im χ(3) are summarized in Table 1.

To further characterize the saturable absorption properties of SnS NSs, the relation between the transmittance and optical intensity [Fig. 4(b)] can be obtained from the Z-scan results using the equation I=I0/(1+z2/z02).(3)The transmittance versus the optical intensity can be fitted by a simple saturable absorption model ^[63] T=1−ΔT×exp(−I/Is)−Tns,(4)where ΔT, Is, Tns are modulation depth, saturation optical intensity, and nonsaturable loss, respectively. As a BP-analog material, parameters of the SnS are compared with the widely investigated BP ^[5]. For an SA, higher ΔT means the stronger ability to modulate the intensity of the light. A higher value of ΔT was calculated to be 36.4% for SnS NSs, compared to other single-elemental 2D materials, including bismuthene (2.03%) ^[64], antimonene (6.4%) ^[65], newly found 2D Se (2.13%) ^[66], as well as BP (12.4%). Additionally, the value of Is is 1 order smaller than that of the BP, which can help to reduce the threshold of the SA-based mode-locked or Q-switched lasers. Moreover, the nonlinear coefficient β of SnS NSs was found to be larger than that of BP, illustrating the stronger interaction between light and SnS NSs (Table 1). These excellent saturable absorption properties at the communication band mean SnS can be a quite promising candidate as an optical modulator.

The relaxation time of the charge carriers is a fundamental parameter for semiconductors and significantly affects the performance of that based optoelectronic devices, such as photodetectors, solar cells, and pulsed lasers. The photo-induced carrier dynamics of SnS NSs was investigated with the assistance of a femtosecond-resolved transient absorption (TA) spectrometer. A nondegenerate setup with broad NIR probe band (1000–1500 nm) was employed. Pumped at 400 nm, clearly photon-induced absorption of SnS NSs was observed within the probe band. Figures 5(a) and 5(b) show the TA spectra at the delay line of 0-5.0 ps, which suggests the cooling speed is fast at the beginning, and then becomes slow when the delay line surpasses 5.0 ps. By global time analysis ^[67], one principal spectrum and kinetic of the charge carrier are figured out, which can well recall the relaxation dynamics [Fig. 5(c)]. ΔA is the absorption difference for the probe light with/without the pump light. The electrons are excited to high conduction band in the first 400 fs through the Franck–Condon transition, leaving over the holes in the valence band. The cooling dynamics for SnS NSs can be divided into two processes. Initially, the electrons undergo thermal equilibrium in the conduction band within 873 fs, and then the hot electrons cool down to the valence band and recombine with the holes at 96.9 ps. The probe wavelength-dependent ΔA can be fitted using a dual-exponential decay model ΔA=A1exp(−t/τ1)+A2exp(−t/τ2),(5)where τ1 is short decay time and τ2 is long decay time. In Fig. 5(d), it was observed the short decay time τ1 increased from 620.2±19.8 to 912.1±12.58fs as the probe wavelength increased from 1000 to 1500 nm, while the long decay time τ2 decreased from 153.8±9.5 to 68.9±11.0ps. The trend of τ1 and τ2 means intraband scattering dynamic constant (τ2) has a negative correlation with probe wavelength and the interband scattering constant (τ1) has a positive correlation with the wavelength.

Mode locking based on slow SA, that is to say, the pulse width can be smaller than the recovery time, has been proved to be an effective passive technique to generate ultrafast lasers other than the fast SA-like Kerr mode locking for the highly restrictive cavity design. Semiconductor saturable absorber mirrors (SESAMs) ^[68], and the SA based on recently developed 2D materials, like graphene, TMDs, and BP, enrich the diversity of the slow SA ^[5,69–71]. For an SESAM, response time associated with the photoinduced electrons and holes relaxation dynamics significantly dominates the performance of the optical devices. It is believed that the short time constant (τ1) is the intrinsic character of one kind of material, and the order of hundreds of femtoseconds for SnS materials is suitable for starting the mode locking ^[72]. While the long time constant (τ2), usually in the time scale from several picoseconds to hundreds of picoseconds, plays an important role in sustaining the mode locking, SnS semiconductor possesses relatively long τ1 than graphene and BP ^[5,69]. Therefore, both mode-locked and Q-switched operations can be readily realized using the SnS-SA. Moreover, the better chemical stability of SnS over BP shows its potential in pulse laser generation. It is worthwhile to note the contribution of the long decay time of the dynamic process is less than 10%; thus an optical modulation speed as high as terahertz is promised, e.g., gigahertz mode-locking operations.

In general, passive Q switching and mode locking are effective techniques to generate short pulses in bulk lasers. The high-quality SnS-SA facilitates the generation of short and ultrashort pulses by the two techniques. Taking the advantage of the large modulation depth and moderate response time, a passive Q-switching operation is initially generated from the fiber laser. As described previously ^[73], there is a critical parameter: E2=EGESAΔT, where ESA and ΔT represent the saturation energy and modulation depth of the SnS-SA, and EG is the gain saturation energy of the laser. In the cavity, when the single pulse energy Ep matches the relation of Ep<E2, the Q switching can be obtained. Otherwise, the continuous wave (CW) mode-locked operation may be achieved. It is worth noting that changing the ESA and ΔT with the assistance of adjusting the deposition surface of SnS NSs on microfiber is the convenient way to modify the value of E, as EG is determined by the gain fiber and is hardly modified.

Q-switched pulses are established when the pump power is increased to 150 mW. However, the pulses cannot be stabilized until the pump power is increased to 275 mW. The pulse repetition rate was 36.36 kHz for 275 mW and increased with the pump power, as Figs. 6(a)–6(e) show. The large scanning range of Q-switched pulse train in Fig. 6(f) indicates the long-term stability of the Q-switched state. Besides, once the pump power exceeded 500 mW, the Q-switched state turned to be unstable again, and the Q-switched state disappeared. If the pump power exceeded 600 mW, a CW state would be obtained. The evolution of averaged output power and repetition rate as the function of pump power is shown in Fig. 7(a). Both of them increased almost linearly versus the pump power up to 375 mW. The radio-frequency (RF) spectrum of the established Q-switched pulses at a pump power of 325 mW, giving a high signal-to-noise ratio up to ∼50dB, indicated the extremely stable Q-switching state.

Passive CW mode locking is a convenient and effective way to generate ultrashort pulses, demanding the appropriate modulation depth, saturation recovery time, and fine cavity design. Ultrashort and high repetition-rate mode-locked pulses are self-started after carefully adjusting the pump power and optimizing the cavity design. A typical mode-locked state is shown in Fig. 8. The generated pulses are recorded by a 1 GHz-high-speed oscilloscope, showing the fundamental repetition rate of 8.37 MHz, which matches well with the cavity length. A large scanning range of pulse trains at a scale bar of 20 μs shows no fluctuation, proving the stable operation of the mode-locked fiber laser. The mode-locked spectrum of the pulses is centered at ∼1560nm and the full width at half-maximum (FWHM) bandwidth is ∼5.28nm. Clear and distinct Kelly sidebands in the spectrum [Fig. 8(a)], a typical characteristic of the soliton fiber lasers with net anomalous dispersion, indicate that the mode-locked pulses can be shaped into optical soliton pulses. The pulse width was measured to be 656 fs by an intensity autocorrelator, assuming the sech2 intensity profile [see Fig. 8(c)]. The calculated time-bandwidth product is ∼0.384, indicating a slight pulse chirp.

The mode-locked laser operates at a highly stable state, for the measured signal-to-noise ratio more than 60 dB from the RF spectrum. To further reveal the long-term stability of mode locking, the mode-locked spectrum at the pump power of 135 mW is recorded over 6 h. No obvious optical spectrum drift was observed [Fig. 9(a)], proving the stable operation.

SnS, as a rising two-elemental and 2D BP-analog group IV monochalcogenide, removes the main obstacle of BP in long-term stability. High-performance 2D SnS was successfully obtained using the LPE method, and its high stability for 1 month has been confirmed by standard and comprehensive characterization techniques. The excellent saturable absorption properties of SnS were found with large and tunable optical modulation depth (maximum value of 36.4%). The saturation optical intensity (∼34.8GW/cm2) is 1 order smaller than that of the BP (∼334.6GW/cm2), which significantly reduces the threshold of the SA-based pulse lasers. The response time was characterized to be τ1=873fs and τ2=96.9ps with the assistance of femtosecond-resolved TA spectra, indicating SnS can be used as a high-speed modulator in the range of gigahertz and even terahertz. Using the SnS as an efficient SA, passive Q switching and CW mode locking were successfully demonstrated in an identical resonator with high stability. An ultrashort laser pulse sustained for 656 fs was readily obtained. Our systematic study shows that the two-elemental BP-analog SnS is definitely a new type of promising SA, with high stability and robust optical performance compared to single-elemental BP. This could compensate for the main drawbacks of BP and boost the ultrafast photonics of BP-analog materials, such as the group IV monochalcogenides.

Acknowledgment. The graduate school at Shenzhen and Tsinghua University Materials and Devices Testing Centre are also gratefully acknowledged.