Layered iron pyrite for ultrafast photonics application

1 Introduction

Low-dimensional materials have attracted increasing attention in nonlinear optics due to their remarkable optoelectronic properties. For instance, one-dimensional carbon nanotubes (CNTs) and two-dimensional (2D) graphene were found to possess ultrafast carrier dynamics and high third-order nonlinear susceptibility [1], [2], [3], [4], [5], [6], [7], [8], [9]. Over the past decade, 2D materials have been widely reported in nonlinear optics, ultrafast photonics, and other fields due to their energy band structure and splendid crystals [10], [11], [12], [13]. Up to now, a considerable number of experiments have demonstrated that the optical nonlinearity of 2D nanomaterials is enhanced without sacrificing ultrafast response. Transition metal dichalcogenides (TMDs) are a kind of highly anisotropic layered semiconductor materials which attract growing research attention due to their excellent physical/chemical properties. The chemical formula is MX₂, where M represents the transition metal (commonly Mo, W) and X represents a group VI element (S, Se, Te). Each TMD monolayer is composed of two hexagonal planes of X atoms and an intermediate hexagonal plane of M atoms. The M atoms are coordinated by covalent bonds with the X atoms in a trigonal prismatic arrangement. They offer further optoelectronic opportunities and result in the advancement of fascinating applications such as optical switches, photo detectors, and quantum well modulators due to their unique thickness-dependent band gap. In addition, they submit to a transition from an indirect band gap in bulk to a direct band gap in the monatomic layer [14]. Compared with graphene, TMDs such as molybdenum disulfide (MoS₂) [15], [16], [17], [18] and tungsten disulfide (WS₂) exhibit distinct and complementary properties. For example, the field-effect transistors made of TMDs single layer possess high mobility and large on/off ratio. Much more effort is needed to promote their practical applications. Earth abundance and non-toxicity pyrite iron disulfide (FeS₂) as one kind of TMD is widely used in basic and applied research. It is a potential candidate to be applied for wide-range nonlinear photonics because of its large optical absorption coefficient (>105cm⁻¹) and narrow band gap 0.95eV [19], [20], [21]. Alivisatos et al. have compared all the 23 kinds of inorganic semiconductor materials and obtained the following conclusion: FeS₂ is the material with the lowest cost and the highest power supplement [22]. For example, with the photovoltaic conversion efficiency up to 4%, its power supply capacity is 100 times that of a monocrystalline silicon cell. FeS₂ is considered to be an excellent electrode material candidate for energy storage devices. At present, research on FeS₂ at home and abroad mainly focuses on solar-cell materials and electronic properties [23], [24], [25], [26].

Passive mode locking is still the most prominent technique to produce all kinds of ultrashort laser pulses [27], [28], [29], [30], [31], [32]. Saturable absorber (SA) has been widely studied because of its lower cost and convenient preparation and used in the fiber laser cavity to act as an internal loss modulator [33], [34], [35]. With the development of the laser technology, high-performance fiber lasers and high-quality ultrashort laser pulse become more reliable and practical. SA made of various materials has been emerging, such as CNTs, black phosphorus, and graphene [36]. Materials with low band gap and layered structure are widely used as SA in fiber lasers especially TMDs (such as CuS₂, SnS₂, MoS₂, WS₂, etc.). Although FeS₂ has a layered structure and low band gap, its properties of ultrafast photonics have not been investigated yet.

In this paper, we demonstrate that FeS₂ exhibits SA property and can be used as SAs of ultrafast laser. The performances of FeS₂ are further tested in erbium-doped fiber lasers (EDFL), in which the width of the pulse is about 1.7 ps and the signal-to-noise ratio (SNR) is about 72 dB. It is observed that under appropriate phases and pump power, two or more solitons form in the fiber lasers and bind together to form bound states of solitons. It is demonstrated that the FeS₂ can be a potential candidate photonics material for ultrafast pulse generations.

2 Preparation and characterization of FeS₂

The synthetic procedure for the fabrication of FeS₂ is illustrated in Figure 1A. First, 15.781-g FeSO₄·7H₂O was completely dissolved in 100 mL distilled water to get aqueous solution A, and 10.482-g Na₂S₂O₃·5H₂O was dissolved in 100-mL distilled water to prepare an aqueous solution B. Solutions A and B were mixed up at room temperature in a beaker and stirred for 1 h. Then, we transferred the mixed solution into a teflon-lined stainless autoclave and heat it for 8 h with the temperature maintained at 180°C. It is then cooled to room temperature to obtain a black precipitate. Finally, we rinse the precipitate three times with deionized water and absolute ethanol and dry it in vacuum at 80°C for 8 h to get FeS₂ powder. FeSO₄ will react with Na₂S₂O₃ and form FeS₂ product under the hydrothermal conditions according to the following reaction:

Figure 1:

Synthetic procedure and characterization of the layered FeS₂ dispersion.

(A) Schematic illustration of FeS₂ preparation. (B) SEM image with 5 μm scale. (C) SEM image with 500 nm scale. (D) Atomic structure of FeS₂ with red spheres representing Fe and grey ones representing S. (E) EDS analysis of the layered FeS₂ dispersion. (F) HRXRD analysis of the layered FeS₂ dispersion. (G) Raman spectrum of the layered FeS₂ dispersion. (H) UV-Vis-NIR absorption of the layered FeS₂ dispersion.

When we use FeS₂ in the fiber laser, we should mix up the FeS₂ powder with alcohol with a ratio of 1:2 for dispersion to form a stable suspension liquid. Before dropping the FeS₂ dispersible solution, oscillate for 40 min in the ultrasonic generator to fully disperse the FeS₂ solution. The photographs of the FeS₂ powder in a sealing bag and the FeS₂ suspension in a bottle are shown in Figure 1A from which we can see that the color of FeS₂ solution is dark grey.

The scanning electron microscopy (SEM) images of the FeS₂ dispersion at 5 μm and 500 nm scales are provided in Figure 1B and C which show that the FeS₂ have a few layers. The atomic structure of FeS₂ is shown in Figure 1D in which the local coordination of the iron atoms is sixfold and that of the sulfur atoms is fourfold. Figure 1E shows the energy dispersive spectrometer (EDS) analysis results of UV-Vis-NIR absorption of the layered FeS₂ dispersion, indicating that our material is mainly composed of 27.79% Fe and 46.62% S, and Si and Al are sample placement stations of SEM mechanism. This result is consistent with the atomic structure of FeS₂. Figure 1F is the high-resolution X-ray diffraction (HRXRD) diagram of FeS₂ microfilm dispersion and gives the direction indicators and proportions of each. There are nine peaks of intensity; the highest one is caused by an orientation index of (109). The Raman spectrum of FeS₂ dispersion is shown in Figure 1G. Different bands represent different properties of FeS₂. Six Raman bonds were observed at 208 cm⁻¹ (corresponding to distance of the sulfur atom perpendicular to the S-S bond), 274 cm⁻¹ (weak bonds), 386 cm⁻¹ (due to in-phase S-S stretched vibration FeS₂), 580 cm⁻¹ (weak bonds), 1096 cm⁻¹ (weak bonds), and 1316 cm⁻¹ (weak bonds). These above peaks show the cubic structure of FeS₂. The UV–Vis-Nir absorption spectrum of FeS₂ powder is shown in Figure 1H. The absorption of FeS₂ between 200 and 2000 nm is always larger than 75% which indicates that FeS₂ can be used as SA in intermediate infrared mode-locked fiber lasers. The transmission of the FeS₂ powder at 1562 nm is 76%.

Figure 2A shows the nonlinear-transmittance measuring system of FeS₂ dispersion. The nonlinear transmission curve of FeS₂ is shown in Figure 2B. A double-balanced detection system which contains a mode-locked fiber laser (the output parameters include 1566.2 nm wavelength, 644 fs pulse duration, and 27.4 MHz repetition rate) and a home-made erbium-doped fiber (EDF) amplifier was built. With femtosecond laser pulses passing through 50/50 output coupler (OC), the output power is divided into two parts: one part is based on SA laser pulse power of FeS₂ material, and the other part is the laser pulse power. At the same time, in order to control the incident laser power, the average power of the mode-locked fiber laser is attenuated by reducing the pump power of the amplifier. The power of the pulses propagating directly to output 2 and the power of the pulses propagating through the FeS₂ to output 1 are detected by two power meters. As a result, we got the modulation depth and saturation intensity of the FeS₂ as 4.5% and 17.5 MW·cm⁻², respectively

Figure 2:

Nonlinear characteristic of FeS₂.

(A) Detecting system for the nonlinear absorbance of FeS₂ dispersion. (B) Nonlinear transmission curve.

3 Results and discussion

Figure 3 shows our experimental setup of the passively mode-locked fiber laser by using FeS₂ as SA. The pump source is a laser diode with 980 nm central wavelength and 423 mW maximum output power. By controlling the current of the pump laser (A_p), the input power of the cavity (P_p) can be manually controlled. The functional dependency of P_p and A_p is P_p=A_p*0.74−36. EDF with a length of 1 m (core absorption at 1530 nm is 110 dBm) is used in the cavity to generate 1566.2 nm laser. By applying a polarization-independent isolator, the laser can only propagate in one direction in the cavity to keep the pump laser out of damage. When the three-paddle polarization controller (PC) is not adjusted and mode is not locked, there is only a spike in the pulse and the energy is not enough to make the fiber nonlinear, so self-phase modulation cannot occur to broaden the spectrum, and the light intensity is also absorbed by the SA. When the phase is fixed and the pulse energy is superimposed, high-energy pulse laser enhances the nonlinearity. Thus the self-phase modulation makes the spectrum broaden obviously. At this time, the high-energy laser saturates the absorption of the SA, resulting in ultrashort mode-locking pulses due to pulse width compression. FeS₂ SA is prepared by deposition of FeS₂ microcrystals in the cone region (diameter and length are 13.6 μm and 6 mm, respectively) and doped on the tapered fiber as shown in Figure 3. In the production method of microfiber, the stripped fiber was placed in the internal flame of the flamethrower and slowly stretched outward until the diameter of the fiber reached 13.6 μm to ensure the interaction between the material and evanescent wave of the fiber and to control the loss of the fiber, and then the microfiber is fixed on the optical glass. Based on the optical deposition method of materials, a small amount of FeS₂ is dripped on the microfiber to make the material and fiber fully interact [37]. Compared with the non-deposited tapered fiber, about 41% of the energy is absorbed by FeS₂, which proves that the material is completely in contact with the tapered fiber. Through an optical microscope. The transmission efficiency of the tapered fiber is approximately 86.3%. A 3:7 OC is employed to transfer 70% of the output laser back into the cavity and 30% of the output laser to the detection system. The length of our cavity is approximately 28 m. By using some monitor and record instruments (optical spectrum analyzer [Anritsu MS9710C], 2 GHz photodetector, digital oscilloscope), we can get both the spectral and temporal properties of the mode-locked fiber laser.

Figure 3:

Schematic diagram of the EDF ring laser.

WDM, wavelength division multiplex; EDF, Er-doped fiber; PI-ISO, polarization-independent isolator; PC, polarization controller; OC, output coupler.

In the experiment, mode-locked fiber laser was observed when the pump power is 423 mW and the PC is fixed at an appropriate state. Figure 4A shows the typical spectrum of fiber laser with a center wavelength of 1566.5 nm in which the spectral width is about 2.5 nm. Spectral Kelly sideband is observed here which indicates that our fiber laser operates in the conventional soliton mode-locking regime [38]. Figure 4B shows the pulse train with a period of 162 ns. The inset of Figure 4B is the pulse train with 2 μs span which indicates the stable mode-locking state. The radio frequency (RF) spectrum with a repetition rate of 6.4 MHz and 10 frequency peaks is shown in Figure 4C. The SNR is about 72 dB as shown in the inset of Figure 4C, which means that the influence of interaction between the birefringence effect of the fiber and the nonlinear optical effect of the pulse is unavoidable. Figure 4D shows the autocorrelation traces. According to the fit to a squared hyperbolic secant (sech²) pulse shape, the pulse duration is estimated to be 1.7 ps. The laser cavity can maintain a stable state for more than a week without pulse splitting when the mode locking phenomenon occurs. The output characteristics of typical materials are summarized in Table 1. FeS₂ has obvious advantages as an SA compared with these materials, and this is the first time that FeS₂ is applied to ultrafast photonics. We will further optimize the output parameters of FeS₂ in future studies.

Figure 4:

The output characteristics of Er-doped mode-locked fiber laser at the fundamental repetition rate.

(A) Output spectrum; (B) output pulse train at range of 1400 ns, the inset is in the range of 2 μs; (C) RF spectrum of the mode-locked pulses, inset: the broadband RF spectrum; (D) autocorrelation trace with a sech² fitting.

The interaction of optical solitons can be affected by periodically filtering effect which leads to the changes of separation and phase difference of neighboring solitons [44]. Bound states of solitons are known as high-order soliton solutions of the nonlinear Schrodinger equation [45] and the complex Ginzburg-Landau equation [46]. Mode-locked lasers can generate such bound solitons, which have attracted significant interests. With different pulse separation and phase difference under certain situations, different bound states could be supported by different intracavity gain, loss, dispersion, and nonlinearity. Figure 5A shows the experiment results of bound state solitons in which the period of the spectral modulation fringe of the bound state solitons is about Δν=Δλν/λ in which Δλ=1.4 nm. Figure 5B shows the autocorrelation trace of the state solitons with the pump power fixed at 316 mW and the PC fixed at appropriate condition. Once the bound states of two solitons are obtained, their soliton separation always falls into one of the three solitons separation values, which indicates that the bound states of solitons have fixed discrete solitons separation. According to the autocorrelation trace shown in Figure 5B, the pulse is a second-order bound soliton. The pulse duration is 2 ps, and the time interval of two autocorrelation traces peaks is about 5.7 ps, which fits well with the calculated result according to Δf=c·ΔΔλ/λ² [38]. The height ratio of the three peaks is 1:2:1. The mechanism of the formation of the bound state solitons is explained as a result of direct soliton interaction [47].

Figure 5:

The output of bound state solitons.

(A) The pulse train; (B) the autocorrelation trace.

With the pump power increasing, the evolutions of the pulse train, spectrum, and 3-dB width of the spectrum are investigated. Figure 6A shows the spectrum evolution with the pump power of 136, 186, 236, 286, 336, 386 and 436 mW from Figure 6A (B–H, representing the specific pump power value) [43]. The conventional Kelly sidebands in the spectrum is increasing with the continuously increasing pump power [38]. It appears as continuous waves when the pump power is weak enough as shown in Figure 6A (B and C). Figure 6B shows the pulse trains corresponding to Figure 6A. The pump power is increasing from top to bottom. Mode locking state can be partially explained by the soliton energy quantization theory, due to the soliton energy relocation and the interaction between the multiple solitons per round cavity round trip [48]. When the pump power is 136 mW, the pulse train is on basic frequency. One can clearly see that the pulse trains are splitting from top to bottom with the pump power increasing. When the pump power is high enough, the multiples will form a special state in the fiber cavity. The 3-dB width of the spectra is investigated as shown in Figure 6C; with the increase of pump power, the width of 3 dBm increased from 1.93 nm to 2.46 nm. As the nonlinearity of the fiber increases with the pump power the self-phase modulation leads to further broadening of the spectrum, while the influence of pumping power on the spectral bandwidth decreases when dispersion and nonlinearity accumulation reach a new balance. The influence of pump source on laser can be minimized by choosing proper pump power. The output power increases linearly with the increase of pump current, as shown in Figure 6D in which the dotted line is the experimental result and the solid line is the fitting result.

Figure 6:

Output signals with the power of the pump increasing.

(A) The spectrum evolution versus the pump power. (B) The pulse train evolution versus the pump power. (C) The 3-dB spectral width versus the pump power. (D) Output power and linear fitting slope curve versus pump power.

4 Conclusion

In summary, we experimentally realize a passively mode-locked EDFL by using metal dichalcogenides FeS₂ as SA. The pulse duration of the mode-locked fiber laser is 1.7 ps with the central wavelength of 1563 nm. The modulation depth and saturation intensity are about 4.5% and 17.5 MW cm⁻². Both single pulse and bound state solitons can be realized as well as the evolution between them by continuously adjusting the pump power. This work demonstrates that FeS₂ dispersion can be used as a new excellent material for SA photonics and optical modulator which is cheap and convenient.