Synthesis of WS1.76Te0.24 alloy through chemical vapor transport and its high-performance saturable absorption.

Layered transitional metal dichalcogenides (TMDs) are drawing significant attentions for the applications of optics and optoelectronics. To achieve optimal performances of functional devices, precisely controlled doping engineering of 2D TMDs alloys has provided a reasonable approach to tailor their physical and chemical properties. By the chemical vapor transport (CVT) method and liquid phase exfoliation technique, in this work, we synthesized WS1.76Te0.24 saturable absorber (SA) which exhibited high-performance of nonlinear optics. The nonlinear saturable absorption of the WS1.76Te0.24 SA was also measured by the open aperture Z-scan technique. Compared to that of the binary component WS2 and WTe2, WS1.76Te0.24 SA has shown 4 times deeper modulation depth, 28% lower saturable intensity and a much faster recovery time of 3.8 ps. The passively Q-switched laser based on WS1.76Te0.24 was more efficient, with pulse duration narrowed to 18%, threshold decreased to 28% and output power enlarged by 200%. The promising findings can provide a method to optimize performances of functional devices by doping engineering.

Introduction

2D transitional metal dichalcogenides (TMDs) materials with formula MX2 (M = Mo, W; X = S, Se, Te) have attracted intense attentions thanks to their unique layered structures, attractive carrier mobility, adjustable bandgap, strong light-matter interaction and stable chemical properties[1-9]. The abundant physical and chemical properties of TMDs are widely applied to photodetectors, logic devices, memories, catalysis, transistors and lasers. The discrepancy of strong covalent bonds intra layers and weak van der Waals inter layers allows down-top and top-down methods to obtain the 2D TMDs nanosheets. For example, chemical vapor deposition (CVD) and wet chemical synthesis are used to obtain the 2D TMDs nanosheets by the methods of down-top[10,11]. It is also desirable to combine two different mature methods of chemical vapor transport (CVT) and mechanical stripping to achieve the nanomaterials (top-down)[12,13]. TMDs can potentially serve as SAs, such as molybdenum disulfide and tungsten disulfide, which already exhibited promising applications with much larger modulation depth than graphene[14-18]. The band-gaps of MoS2 and WS2 are about 1.6~1.9 eV in the visible range, indicating the natural MoS2 and WS2 are far from serving as saturable absorbers in highly developed infrared lasers. In recent years, the defects have been introduced into TMDs in order to extend the bandwidth of TMDs SAs. It has been proved by theories and experiments that M or X vacancy defects in TMDs can reduce their band-gaps[19-22]. S. Wang reduced the band-gap of MoS2 by introducing defects[19]. MoS2 with S defects can be used as a broadband saturable absorber. The passively Q-switched lasers in the range of 1.06, 1.42 and 2.1 μm were realized based on the MoS2 SA with S defects, corresponding pulse duration of 970, 729 and 410 ns, respectively[19]. X. Guan et al. experimentally realized 2.7 μm self-Q-switching laser of Er:Y2O3 based on WS2 SA with pulse width of 1.36 µs[20]. The passively Q-switched lasers at the wavelengths of 1.5, 2.0, 2.7 μm are based on the binary TMDs with defects[14-22]. However, M or X vacancy defects in MX2 make its structure unstable, so M or X vacancy defects in TMDs are difficult to repeatedly synthesized. Furthermore it is less feasible to regulate intentional defects, for example, the 2D MX2 often appears to be both triangle and hexagon due to the local condition. Therefore, there is an urgent need to tune the band-gaps of the TMDs so that they can perform stably in the near-infrared and mid-infrared range.

Inspired by the history of Si semiconductor, the doping engineering appears to be the key to tailor physical and chemical properties of TMDs. Mixed chalcogenides or mixed metal elements of two different TMDs can control the band gap, such as WS2×Se2(1−x)[23] and MoxW1−xS2[24]. However, it has been proven that only a few hundred milli-electron volts (meV) could be realized, i.e. 300 meV for WS2xSe2(1−x) and 170 meV for MoxW1−xS2 solid solutions respectively. P. Yu et al. experimentally showed that 2H-WSe2 and Td-WTe2 can form stable layered WSe2×Te2(1−x) alloys[25], with a phase transition from 2H-to-Td (x = 1 − 0.6 for 2 H structure; x = 0.5 and 0.4 for 2 H and 1 Td structures; x = 0 − 0.3 for 1 Td structure) controlled by the complete composition. The electronic structures changing from semiconducting to metallic enable wide tunability of the optical and electronic properties. Extraordinary physical properties of alloys are needed for in-depth study where the alloys showed some unique advantages compared to 2D binary TMDs, making them fundamentally and technically important in applications of optics and optoelectronics. One of the impressive physical properties of mono- and few-layer alloy is that TMDs display surprisingly excellent nonlinear optical properties, Y. Wang et al. studied the nonlinear optics properties of alloys of Bi2TeSe with lower saturable intensity, deeper modulation depth[26].

TMDs alloying still remains challenging resulted from the lattice mismatch of their parent counterparts. Here we synthesized WS1.76Te0.24 alloy by doping Te2− ions in WS2 (2 H) structure. The nonlinear optics properties of WS1.76Te0.24 SA were 4 times deeper modulation depth, 28% lower saturable intensity and a much faster recovery time of 3.8 ps compared to those of WS2 and WTe2. To find out whether pulsed laser’s performance can be promoted by alloying, passively Q-switched lasers were investigated based on WS2, WS1.76Te0.24 and WTe2 SAs at the wavelength of 1060 nm. We found that the passively Q-switched laser based on WS1.76Te0.24 was more efficient, with pulse duration narrowed to 18%, threshold decreased to 28% and output power enlarged by 200%. The promising findings can provide a method to optimize performances of functional devices by doping engineering.

Methods

Synthesis and characterization of WSxTe2−x SAs

The WS1.76Te0.24 monocrystalline was prepared by the chemical vapor transport (CVT) method with well-controlled temperature. There were two steps to obtain WS1.76Te0.24 mono-crystalline. Firstly, WS1.76Te0.24 polycrystalline was synthesized by heating a mixture of sulphur (Strem Chemicals 99.9%), tungsten (Strem Chemicals 99.9%) and tellurium (Strem Chemicals 99.9%) with stoichiometric amounts at 750 °C for 48 hours in an evacuated and sealed quartz ampoule (8 mm ID, 10 mm OD, 300 mm length). Considering the powerful exothermicity of the reaction, the mixture was slowly preheated to 750 °C for 12 hours to avoid explosion. Secondly, WS1.76Te0.24 was grown by CVT method in a double zone furnace with as-prepared grinded polycrystalline powder and the transport agent was bromine (Sigma-Aldrich, 99.8%) at about 5 mg/mL. The procedure of growth was 72 hours in an evacuated and sealed quartz ampoule (8 mm ID, 10 mm OD, 300 mm length). Figure 1a showed a two-temperature zone tube furnace with well-controlled temperature. Throughout the growth process of WS1.76Te0.24, the raw material (T2) and crystal growth zones (T1) were kept at 1030 °C and 1010 °C, respectively. The parent components of WS2 and WTe2 were synthesized by the same method for the following contrast experiments. Figure 1b–d show the monocrystalline photographs and atomic structures of WS1.76Te0.24, WS2 and WTe2.

Figure 1

(a) Scheme of chemical vapor transport (CVT) for crystallization of WSxTe2-x in a temperature gradient. Monocrystalline photograph and atomic structure of (b) WS1.76Te0.24. (c) WS2. (d) WTe2. TEM and SEAD characterizations of few-layer nanosheets of (e) WS2, (f) WS1.76Te0.24 and (g) WTe2. Corresponding full EDS scanning of (h) WS2, (i) WS1.76Te0.24 and (j) WTe2. AFM images and thickness measurement of typical nanosheets and height measurement recorded at different sections of (k) WS2, (l) WS1.76Te0.24 and (m) WTe2. (n) EDS mapping of WS1.76Te0.24. The clear morphology implies the uniformity of component distribution.

We prepared WS2, WS1.76Te0.24 and WTe2 SAs by liquid-phase exfoliation and spin-coating technique with the same parameters (sonication time, speed of centrifugation and spin-coating) for further comparison. First, the mixture of grinded 0.2 mg WS1.76Te0.24 monocrystalline in 4 ml acetone solvent was sonicated in high power for 40 min. Only pure acetone was employed as the solvent to avoid introduction of extra impurities. Then, we collected the one-third top of the dispersions after the centrifugation at 2500 rpm for 10 min to remove the large sedimentations. Finally, we span and coated the dispersion on SiO2 plate to obtain WS1.76Te0.24 SA after the acetone was easily removed by volatilization in the air.

Material characterization

Transmission electron microscopy (TEM) and energy dispersive spectroscopy (EDS) were adopted to learn the morphology of the WS1.76Te0.24 nanoplates. TEM images in Fig. 1e–g showed the layered structures of WS2, WS1.76Te0.24 and WTe2 respectively, where the gray scale was directly proportional to the thickness. The observed well exfoliated nanoflakes with layered structure implied rigid mechanical property. The insets were the Selected Area Electron Diffraction (SAED) of WS2, WS1.76Te0.24 and WTe2 nanoflakes. The SAED showed that WS2, WS1.76Te0.24 and WTe2 were monocrystalline with 2 H, 2 H and Td phases, respectively. The diffraction of six-fold symmetry spots displayed the hexagonal lattice of WS1.76Te0.24 nanoflake. EDS measurement was also produced to determinate the element ratio of the three samples as shown in Fig. 1h–j. It can be seen that the ratio of S to Te in WS1.76Te0.24 is 1.76 to 0.24, indicating an efficient doping of Te ions in WS2 framework. As shown in Fig. 1k–m, the atomic force microscopy (AFM) was carried out to measure the three SAs thicknesses. The thicknesses of WS2, WS1.76Te0.24 and WTe2 nanoflakes were about 15.6, 14.9, and 16.3 nm, corresponding to 23, 19 and 27 layers, respectively. To learn more about the WS1.76Te0.24 alloy, EDS mapping was adopted shown in Fig. 1n. The green, red and yellow parts were the distribution of tungsten, sulphur, and tellurium element, respectively. The uniform doping of Te in WS2 of WS1.76Te0.24 was obtained. Furthermore, the WS1.76Te0.24 SA with no additional dangling bonds is stable in the air. The high chemical stability was due to the substitution of atoms in the alloy TMDs.

Raman spectroscopy was employed to learn the detailed lattice vibration modes of WS1.76Te0.24 affected by doping engineering where Te2− replaced the S2− in the WS2 structure. The characterization was carried out by using a Jobin Yvon LabRam 1B Raman spectrometer with laser source at 532 nm. The comparison of the Raman fingerprints among the three samples in the range of 200–450 cm−1 is shown in Fig. 2a. In Fig. 2a, the characteristic peaks at 353.2 and 422.7 cm−1 were assigned as the in-plane (E2g) and out-of-plane (A1g) vibrational modes corresponding to WS2 nanoflakes. For the Td-WTe2, the spectrum just showed the A1 Raman mode at 217.8 cm−1. The characteristic bands of WS1.76Te0.24 showed the “two mode behavior” as the coexistence of vibrations of WS2 and WTe2. The E2g and A1g peaks of WS1.76Te0.24 at 352.5 and 411.5 cm−1 are resulted from the corresponding modes in WS2. The red-shift of A1g (0.7 cm−1) and E2g (11.2 cm−1) mode is attributed to the significant changes in the electron-phonon coupling soften by the doping of Te. The frequency broadened at A1 (217.8 cm−1) was originated from the reduced structural symmetry arising from lattice distortion due to the different atomic radius of Te and S. In earlier studies, TMDs alloys, e.g. MoxW1−xS2[24] and WS2×Se2(1−x)[23], have the similar frequency shift. The Raman shifts further confirm the expected structural and compositional evolution in the WS1.76Te0.24 alloy.

Figure 2

(a) Raman spectra and vibration modes of few-layer WS1.76Te0.24 nanosheets. (b) Comparison of recorded transmittance spectra and corresponding fitted line of WS2, WS1.76Te0.24 and WTe2. (c) The refractive index of WS2, WS1.76Te0.24 and WTe2 based on the relationship of Kramers-Kronig. (d)The open-aperture Z-scan measurements of WS2, WS1.76Te0.24 and WTe2 flakes at 1060 nm. (e) An ultrafast signal of WS1.76Te0.24 proved that the relaxation time was ~3.8 ps.

Figure 2b showed linear transmittance spectra of WS1.76Te0.24 SA with WS2 and WTe2 as contrasts. The typical absorption peaks of WS1.76Te 0.24 corresponded to the trait of TMD 2 H phase. The peaks located at the wavelength of 906 (A) and 859 nm (B) coincided with inter-band transitions. A, B are resulted from spin-orbit splitting of transitions[16], while C and D (685 and 592 nm) are due to the higher density transition[16]. The transmittance spectrum of WS2 showed the TMD absorption with absorption peaks at 814 (E) and 426 (F) nm. WTe2 displayed semi-metallic phase absorption characteristic with broadband absorption. According to the Tauc plot by extrapolating the linear absorption versus photon energy curve, optical bandgaps of WS1.76Te0.24, WS2 SAs were calculated to be 1.2 eV and 1.9 eV, respectively. The electronic structure of WS1.76Te0.24 strongly depends on the coordination environment of W and its d-electron counts. The Te doping in WS2 shorten band gap by the change coordination environment of W corresponding to the ab initio[23]. Thus physics properties of WS1.76Te0.24 were tailored by ion doping.

The ellipsometer is a conventional method to measure the film’s refractive index. However, ellipsometer has strict requirement on the samples for uniform surface, large size, and thin thickness. Due to its low spatial resolution, it is difficult to obtain refractive index of nanomaterials in nanometer size. Benefitting from the Lorentz-Drude model and Kramers-Kronig (K-K) relationship of the dielectric function[27], we calculated the corresponding refractive index from the transmittance spectrum. Figure 2c shows the fitting curve of the reflectance spectra, where the refractive index parameters were obtained by the K-K relationship.

In order to understand the incorporation mechanism by Te doping into WS2 and the corresponding effect on optical nonlinear properties of 2D WS1.76Te0.24, we performed Z-scan measurement with a femtosecond laser (1060 nm, 175 fs) as excited source. The results of WS2, WS1.76Te0.24 and WTe2 SAs are showed in Fig. 2d. The increase of transmittance was easily observed with the increase of laser intensity, resulted from the nonlinear saturable absorption effect. The mechanism of saturable absorption can be explained as Pauli blocking principle in the conduction band. However, significant differences in saturable absorption efficiency and sensitivity among the three samples can be clearly distinguished in Fig. 2d. Based on the nonlinear optical theory, the transmittance is expressed in the form of[14]where As is the modulation depth, A is the non-saturable components, Isat is the saturable intensity, and I is the incident light intensity. The data was fitted with the Eq. 1, the modulation depth and the saturable intensities were obtained as presented in Table 1. The WS1.76Te0.24 saturation intensity and modulation depth were 0.88 GW/cm2 and 4.47% respectively. When WS2 became saturated above threshold of 3.15 GW/cm2, the modulation depth was 1.14%. The saturation intensity and modulation depth of WTe2 were 3.7 GW/cm2 and 1.6%, respectively. Compared to binary component of WS2 and WTe2 SA, WS1.76Te0.24 SA performed 4 times deeper modulation depth and 28% lower saturable intensity, resulting from bandgap evolution of WS1.76Te0.24.

Table 1

SA Results for Different Two-dimensional Materials.

Materials	n	α0 (104 cm−1)	αNL (104 cm/GW)	Imχ(3) (10−7 esu)	FOM (10−14 cm•esu)	As (%)	IS (GW/cm2)
WS2	6.19	0.90	−0.41	−4.98	5.53	1.15	3.15
WS1.76Te0.24	10.2	2.25	−0.53	−17.5	7.78	4.47	0.89
WTe2	8.29	1.08	−0.38	−8.28	7.66	1.60	3.70

The bandgap evolution of WS1.76Te0.24 is used to interpret the lower saturable intensity. As shown in the Fig. 3, the bandgap of WS2 SA is 1.9 eV larger than the photon energy () of 1060 nm laser. The saturable absorption of WS2 is resulted from the defect states. As shown in Fig. 3, under the excited light, the electrons in valence band of WS2 are transferred to the defect states. The electrons in the defect states jump to the conduction band with one more photon each. Therefore, there are two platforms that the transition of one electron from valence band to conduction band of WS2 and the transition requires two photons. Compared to that of WS2, the optical bandgap of WS1.76Te0.24 is 1.2 eV covering 1.0 µm, so the electrons directly transfer from valence band to the conduction band with one absorbed photon. In WTe2, the semi-metal characteristic makes it possess higher electron concentration in the conduction band. The higher concentration of electrons has stronger reflection on the excitation light as shown in Fig. 3. Therefore, compared to the binary component of WS2 and WTe2, WS1.76Te0.24 has stronger photon absorption at 1.0 µm. The stronger photon absorption of WS1.76Te0.24 can increase the number of absorbed photons to produce more electrons at the same laser intensity. Eventually, the saturable intensity of WS1.76Te0.24 is lowered.

Figure 3

Schematic diagram of linear absorption in WS2, WS1.76Te0.24 and WTe2 SAs.

Suppose at a certain photon frequency, optical absorption satisfies . Here the absorption coefficient α(I) is expressed as . dz′ is the propagation distance in the sample. The third-order nonlinear optics susceptibility Imχ(3) can be expressed ashere c is the speed of light, λ is the laser wavelength, n is the refractive index, the discrepancy caused by the linear absorption, namely figure of merit (FOM): . Based on the model 2, we can obtain WS1.76Te0.24 with αNL ~ −104 cm/GW, Imχ(3) ~10−7 esu, FOM ~10−14 cm• esu. Compared to previous works, the FOM of WS1.76Te0.24 perform one order of magnitude larger than that of grapheme, graphene oxide, MoS2/NMP dispersions ~10−15esu cm[16]. That suggests a promising potential to achieve efficient nonlinear performance by alloying TMDs. However, one should note that FOM varies with the different experiment conditions such as the wavelength, pulse width and so on. For convincing comparison, we carried out the Z-scan on the same condition and the nanosheets were prepared by the same parameters of liquid-phase exfoliation and spin-coating technique. The results are shown in Table 1. It is unambiguous that the FOM value of WS1.76Te0.24 SA was larger than those of WS2 and WTe2 SAs, which indicated the enhanced nonlinear performance of WS1.76Te0.24.

The pump-probe system was adopted to study the carrier relaxation that reflects the optical response of materials. The undegenerated pump probe system is easy to align and the relaxation time is corresponded to the carrier-phonon coupling[28]. The ultrafast signal was measured using a Ti: Sapphire laser with pulse duration of 120 fs, repetition rate of 76 MHz, fluence of 200 µJ/cm2 at 395 nm as the pump and the probe beam was at 790 nm with much lower fluence. The probe reflection was a function of the delay time that was detected by a Si photodetector and amplified by a lock-in amplifier. As shown in Fig. 2e, the ultrafast signal of WS1.76Te0.24 flakes with absorption bleaching was obtained. The signal amplitude was as large as ~ 250%, implying excellent nonlinear optics property. Notably, the decay time was 3.8 ps fitted by a single exponential function. The decay time of WS1.76Te0.24 was significantly shorter than 13 ps of WS2[29] and 5 ps of WTe2[6], as a result of higher density of trapping states induced by Te doping in the nanoflake[30].

Investigation of the WS1.76Te0.24 saturable absorption in passively Q switched laser at 1.0 μm

To identify whether the boosted saturable absorption effect of WS1.76Te0.24 did favor in Q-switched laser, we set up a passively Q-switched Yb: Gd2SrAl2O7 (Yb: GSAO) laser to investigate the performance of WS2, WS1.76Te0.24 and WTe2 SAs. In Fig. 4a, the schematic of experiment setup was shown. A laser diode of 976 nm was served as pump source, which was coupled in a fiber of a core diameter of 105μm and the numerical aperture of 0.22. A doublet lens was employed to focus the beam at 105 μm within the Yb:GSAO crystal. In a cooled down system, the Yb: GSAO gain medium was wrapped with indium foil and mounted in a copper holder with water-cooled at 21 °C. The 11 mm linear cavity composed of 1060 nm high reflectivity M1(R = −200 mm) and 18% transmittance plano M2. The as-prepared three samples on SiO2 were inserted into the cavity serving as the saturable absorber.

Figure 4

(a) Experimental setup of the LD pumped passively Q-switched Yb: GSAO laser at 1060 nm. (b)The output powers with increasing incident pump powers. (c) Pulse repetition rates. (d) Variations of pulse duration with increasing pump powers. The Q-switched lasers performance by the WS2, WS1.76Te0.24 and WTe2 SAs: (e) the oscilloscope tracings. (f) the single pulse tracings (g) the optical spectra.

Figure 4b–d show the characteristics of average output power, repetition rates and pulse durations of the lasers on absorbed pump power variation. The absorbed pump power thresholds of WS2, WTe2, and WS1.76Te0.24 were 1.02, 1.21 and 0.34 W, respectively. The threshold of the WS1.76Te0.24 SA decreased to 28% due to the lower saturable intensity, as shown in Table 1. The output power was measured and calculated to be linearly correlated with the pump power. The maximum output powers of WS2, WTe2 and WS1.76Te0.24were 247.5, 152.8 and 350 mW, respectively. It is worth noting that WS1.76Te0.24 SA achieved twice larger output power of WTe2. The repetition rate continuously increased from 108.2 to 195.2 kHz for WS2 SA, and the repetition rate range of WS1.76Te0.24 SA was 120.8 to 271.1 kHz, scope of WTe2 varied from 112.2 to 170.8 kHz. As shown in Fig. 4d–f, pulse widths of 1.285 µs, 230 ns and 550 ns were obtained in 1.06 µm Q-switched lasers based on WS2, WS1.76Te0.24 and WTe2 SAs, respectively. As shown in Fig. 4e, the typical Q-switched pulse trains of WS2, WS1.76Te0.24 and WTe2 were recorded by a 500 MHz bandwidth oscilloscope (Tektronix, DPO7054) through a high-speed detector (Thorlabs, DET10C/M), which confirms the stability of Q-switched operation. The narrowest pulse duration was obtained by WS1.76Te0.24 SA that should be attributed to the much larger modulation depth than WS2 and WTe2 (Fig. 2d and Table 1). The passively Q-switched laser based on WS1.76Te0.24 narrowed pulse duration to 18%.The optical spectra of the Q-switched lasers were measured with an infrared optical spectrum analyzer (Yokogawa, AQ-6315A) with a resolution of 0.05 nm. The wavelength of WS2, WS1.76Te0.24 and WTe2 were centered at 1061.13 nm with 0.12 nm full width at half maximum (FWHM), 1065.93 nm with 0.45 nm FWHM, and 1058.01 nm with 0.16 nm FWHM, respectively. Q-switched laser based on the WS1.76Te0.24 SA can improve the key parameters as pulse width, slope efficiency and the average output power as listed in Table 2.

Table 2

Experiment results of Q-switched lasers based on three SAs.

Sample	WS2	WS1.76Te0.24	WTe2
Wavelength (nm)	1061.1	1065.9	1058.0
Output power (mW)	247.5	350	152.8
Pulse Duration	1.285 µs	230 ns	550 ns
Repetition rate (kHz)	141.6	271.1	170.8

Conclusion

In this work, we have experimentally demonstrated the enhanced nonlinear optical properties of WS1.76Te0.24 by alloying WTe2 and WS2. We synthesized ternary WS1.76Te0.24 by CVT method. The SAED, EDS and Raman spectra showed good quality of the alloy WS1.76Te0.24 nanosheets. The saturable absorption of WS1.76Te0.24 at 1.06 µm was significantly more efficient than binary parents WTe2 and WS2 as evidenced by Z-scan and pump-probe results, where WS1.76Te0.24 SA showed 4 times deeper modulation depth, 28% lower saturable intensity and a much faster recovery time of 3.8 ps. The passively Q-switched laser based on WS1.76Te0.24 was found more efficient, with pulse duration narrowed to 18%, threshold decreased to 28% and output power enlarged twice. The doping engineering SAs can improve the Q-switched lasers performance with lower energy consumption, narrower pulse width, and larger average output power. The promising findings can provide a method to optimize performances of functional devices by doping engineering.