Over the past few decades, ultrafast lasers are becoming ubiquitous tools in a wide variety of applications, ranging from optical communications to biomedical imaging and industrial materials processing^[1,2]. To achieve ultrashort pulse operation, the majority of lasers use a mode-locking technique, whereby a nonlinear optical element, called a saturable absorber (SA), turns the continuous-wave (CW) output into a train of optical pulses. To be a proper SA, the key requirements for materials are fast response time, broad wavelength range, strong nonlinearity, low optical loss, high power handling, low power consumption, low-cost, and ease of integration into a laser system^[3]. So far, many SAs, including the organic dyes, color filter glasses, ion-doped crystals, semiconductor SA mirror (SESAM), and single-wall carbon nanotubes (SWCNTs), have been developed^[4]. Currently, the dominant mode-locking technology is based on SESAMs. They are complex quantum well devices, typically fabricated by molecular beam epitaxy on distributed Bragg reflectors. Nevertheless, SESAMs suffer from several drawbacks, e.g., slow recovery from saturation (∼ps level), narrowband operation, complex fabrication and integration methods, and low damage threshold. A simple, cost-effective alternative SA material is SWCNTs, in which the diameter controls the energy gap and thus the operation wavelength. Broadband operation (1–2 μm) is also possible using SWCNTs with a wide diameter distribution. However, when operating at a particular wavelength, SWCNTs not in resonance are not used and contribute unwanted losses. These limitations motivate research on novel nonlinear optical materials for SAs.

Since the discovery of mechanically exfoliated graphene in 2004^[5], research on two-dimensional (2D) materials has grown exponentially in the fields of condensed matter physics, material science, chemistry, and nanotechnology due to its extraordinary properties. Inspired by studies on graphene, other 2D nanomaterials that possess similar layered structure features but versatile properties^[6–23], such as topological insulators (TIs)^[8–15], hexagonal boron nitride (h-BN), transition metal dichalcogenides (TMDCs)^[16], black phosphorus (BP), MXenes^[17], graphitic carbon nitride (g-C3N4), metal-organic frameworks (MOFs), covalent organic frameworks (COFs), polymers, metals, silicene, layered metal oxides, and layered double hydroxides (LDHs), have also been explored in recent years, greatly enriching the family of 2D materials, as illustrated in Fig. 1(a). As an example, Fig. 1(b) provides the stability analysis and semiconducting properties of 44 different TMDs (e.g., MoS2, WS2, MoSe2, and WSe2).

Interestingly, 2D materials also have profoundly promoted the fields of ultrafast photonics due to their broadband saturable absorption and nonlinear optical property, which is undoubtedly a key advantage of 2D materials over SESAMs^[24–32]. For example, study found that, BP possesses a thickness-dependent direct bandgap that could be widely tunable within a broad region from 0.3 to 1.5 eV, which can bridge the gap between its 2D counterparts graphene and TMDs (bandgap: 1.5–2.5 eV), while BP exhibits broadband third-order nonlinear optical responses ranging from the visible to mid-IR^[29–31]. On a whole, 2D materials offer several key advantages, such as ultrafast response time, small size, low loss, the ability to operate both in transmission and reflection, and compatibility to optical fibers. Since the first demonstrations of a graphene SA in 2009, research into 2D materials has been progressing at a fast pace^[33–35]. So far, various 2D materials-based SAs have been demonstrated for pulse generation in the lasers operating at 0.6 to 3 μm^[36–44].

Here, we review the current state of 2D noncarbon materials-based ultrafast photonics with a particular focus on fabrication techniques, nonlinear optical properties, integration strategies, and laser applications. We catalog and review the nonlinear optical properties of 2D noncarbon materials and their application to ultrafast lasers. We believe that 2D noncarbon materials could play a significant role in future optoelectronic and photonic technologies, particularly as a kind of wide-band SA for ultrafast lasers.

Few-layer nanomaterials fabrication techniques can be broadly separated into two approaches: top-down exfoliation (including mechanical cleavage and solution-processing techniques) and bottom-up growth [such as chemical vapor deposition (CVD) and pulsed laser deposition (PLD)]^[45], as shown in Fig. 2. Mechanical exfoliation involves repeatedly cleaving layers from bulk layered crystal materials, often using adhesive tape, leaving few-layer and a small number of monolayer flakes. This technique can be used to produce high quality, single-crystal flakes. However, despite wide-spread usage of the mechanical cleavage technique for fundamental studies of 2D materials, poor scalability and low yield render it unsuitable for realistic large-scale applications. CVD offers a scalable method for the production of single- and few-layer 2D materials. The film growth is limited by the low nucleation rate on bare substrates, and pre-treatment of the substrate is often necessary to seed the 2D materials growth. Another growth technique is hydrothermal synthesis, where crystallization is achieved at a high vapor pressure reaction and elevated temperatures. However, the mechanism for the formation of the nanosheets from the reaction, which typically produces a one-dimensional (1D) structure, is not explained. PLD produces films of material following ablation from a target. The target is placed in a chamber (typically under vacuum) and irradiated, producing a plume of ejection, which can be deposited onto a substrate. In particular, the technique allows control over the ratio of different compounds in the film due to the different evaporation rates of the two ions. Solution processing of bulk layered-materials is a widely used technique that produces a high yield of mono- and few-layer flakes dispersed in liquid, carried out under ambient conditions with a high throughput.

Bulk layered-materials can either be chemically exfoliated (e.g., via lithium intercalation) or dispersed into select solvents via liquid phase exfoliation (LPE)^[46], as shown in Fig. 3. Chemical exfoliation of bulk layered-materials is typically achieved via lithium intercalation followed by hydrothermal exfoliation. However, exfoliation of bulk layered-materials by this method can lead to structural alterations in the material, producing the other phase and requiring annealing. Ultra-centrifugation also enables sorting of 2D material flakes by thickness, providing a route to engineering 2D material dispersions with desired flake sizes. Finally, we note that other few-layer 2D materials fabrication techniques exist, such as physical vapor deposition and gas phase growth, which have been used to produce SA devices.

Optical nonlinearity plays an important role in many photonic and optoelectronic applications^[47–72]. In this part, we describe two well-known nonlinear effects in 2D materials: they are the Kerr effect and saturable absorption. A total polarization is used to model linear and nonlinear responses under electrical field, given by P=ε0(χ(1)·E+χ(2):E+χ(3)⋮E+…),(1)where ε0 is the vacuum permittivity. First-order susceptibility represents χ(1), which is the linear response of the material. Second-order susceptibility χ(2) represents two-frequency effects, such as second harmonic generation and sum-frequency generation. It is non-zero only when there is no inversion symmetry in the material molecules.

Third-order susceptibility χ(3) represents three-frequency effects, such as four-wave mixing and third harmonic generation. The real part of third-order susceptibility χ(3) represents the Kerr effect. In fiber optics, the refractive index n under high optical intensity I is given by n=n0+n2I,(2)where n0 is the linear refractive index. The nonlinear refractive index coefficient n2 is related to the real part of χ(3).

The imaginary part of χ(3) represents the nonlinear absorption of the materials under the incident light field. Saturable absorption is one kind of nonlinear absorption and is related to the imaginary part of χ(3). Two-photon absorption (TPA) is also a nonlinear absorption effect in the materials, which means that the electrons absorb two photons simultaneously to be excited to the conduction band. When the light intensity is very high, TPA becomes significant enough and should be taken into consideration. The saturable absorption expression including the TPA effect is given by α=αns+αS1+I/Isat+βI,(3)where β is the TPA coefficient, αns is the non-saturable absorption coefficient representing the intensity-independent loss, αS is saturable absorption coefficient, I is the incident optical intensity, and Isat is the saturation intensity. When the light intensity is low, the absorption is equal to αns+αS, and, when the intensity is high, the absorption is equal to αns. The maximum change of the absorption is αS.

In addition, a modified model is also widely used^[42], T=T0−ΔTexp(−I/Isat)−T0βI,(4)where T is the transmission, αS and βI are much smaller than 1 and low-intensity transmission T0=exp(−αns).

For 2D materials, two measurement methods are generally used to characterize the Kerr effect and saturable absorption, and they are the Z-scan measurement and two-arm measurement^[73–75]. For the Z-scan measurement, the sample is mounted on a Z direction translation stage, as shown in Fig. 4. The light from a pulsed source is first split into two paths. One path with low optical power is for reference and is measured by a slow detector. The other path with high optical power is for material characterization. The light beam in the measurement path is then focused to the sample by a lens. The transmitted light beam after the sample is collected by a second detector. If there is an aperture before the second detector, the Z-scan measurement is called closed-aperture. Otherwise, it is called open-aperture. For open-aperture Z-scan measurement, the intensity-dependent transmission of the sample can be obtained by comparing the readings of two detectors. Open-aperture Z-scan measurement is therefore able to measure the saturable absorption of the sample. For closed-aperture Z-scan measurement, the reading from the second detector is not only dependent on the transmission of the sample but also the beam diameter after the sample. Because the beam diameter after the sample is related to the beam induced refractive index change of the sample, closed-aperture Z-scan measurement is able to investigate the Kerr effect of the sample and obtain the nonlinear index.

For the two-arm measurement, all of the light propagation is confined in an optical fiber. Different from the open-aperture Z-scan setup, the sample is incorporated to the fiber, and there is no translation stage. Because all of the light beams are confined in the fiber, no spatial information can be collected like in a closed-aperture Z-scan. Therefore, the two-arm measurement can only measure saturable absorption of the materials.

By using the Z-scan technology, we can provide the nonlinear optical parameters of 2D materials, as shown in Table 1. For 2D materials, the origin of saturable absorption is usually believed to be Pauli blocking^[76–80]. As an example, we provide the optical and microwave saturable absorption in TI Bi2Te3, as shown in Fig. 5. Notably, the saturable absorption of 2D materials-based SAs beyond their direct bandgap belonging to sub-bandgap absorption can also happen, where the reason can be attributed to the defects, TPA^[81–84], and absorption of edge mode^[85,86]. In the experiment, chemical or physical exfoliated methods may not produce very complete nanosheets, and, thus, the absorption of defects should be weak; however, it is very difficult to obtain 2D nanosheets with a uniform shape and size. Therefore, there can be many edges in the 2D materials-based photonic device, leading to a relatively high absorption. Further investigation on the SA of 2D materials is still required in future work. Interestingly, third harmonic generation^[87–90] and four-wave mixing^[91] were also studied recently.

So far, various 2D materials-based SAs integration methods for fiber devices are developed^[34]. For example, they can be sandwiched between two fiber connectors [Fig. 6(a)], injected into in-fiber microfluidic channels [Fig. 6(b)] and photonic-crystal fibers (PCFs) [Fig. 6(c)], and coated on the surfaces of D-shaped [Fig. 6(d)] or tapered fibers [Fig. 6(e)]. Specially, several common methods have been used to obtain the 2D materials-based SAs, such as optical deposition, PLD, PCF filling, side-polished fiber or microfiber-based evanescent field interaction, and polymer film. In these methods, the fiber ferrule-type SAs are simple and low-cost but have an inherently short nonlinear interaction length; PCF-filled SAs show the strong light–matter interaction but exhibit relatively larger insertion loss and distorted guiding mode in the PCF region; the microfiber or side-polished fiber-based SAs are attractive for high power tolerance and long interaction length, but not easy to obtain uniform material profile; the 2D material polymer films could maintain the thermal stability of SAs, but they are vulnerable to destruction by high power operation; PLD shows many benefits, such as good security component, high deposition rate, and very uniform film. These strategies are particularly advantageous for lasers because they can be fully integrated into various fiber configurations while preserving an alignment-free, all-fiber format, which cannot be realized using SESAMs^[33–44].

Generally, the linear and ring cavity is used to realize the mode-locking or Q-switching operation in the fiber lasers^[85–91]. As an example, the experimental scheme of the ring cavity fiber laser is shown in Fig. 7. The pump source is a fiber-pigtailed laser diode (LD) and transferred the energy to a piece of gain fiber (DF) via a wavelength-division multiplexer (WDM). An optical coupler (OC) is used to extract the laser output. A polarization-independent isolator (ISO) is used for unidirectional operation of the fiber laser, and a polarization controller (PC) is used to adjust the polarization state in the cavity, respectively. The pulse information and laser power obtained are measured by an optical spectrum analyzer, a mixed oscilloscope combined with a high-speed photo-detector, and an optical power meter, respectively.

In order to fully understand the experiment, we adopt the propagation model of a passively mode-locked fiber laser. Numerical simulations are based on a Ginzburg–Landau equation (GLE), which considers the most important physical effects of group velocity dispersion (GVD), self-phase modulation, and saturation gain with a finite bandwidth.

The GLE can be given by ∂U∂z+iβ22∂2U∂τ2−β36∂3U∂τ3=g2U+iγ|U|2U+iγTR∂|U|2∂τU.(5)

Here, U=U(z,τ) is the slowly varying amplitude of the pulse envelope, z is the propagation coordinate, and τ is the time delay parameter. β2 and β3 are the second-order (GVD) and third-order dispersion (TOD) parameters, respectively. γ is the nonlinearity parameter given by γ=n2ω0/cAeff, where n2 is the Kerr coefficient, ω0 is the central angular frequency, c is the velocity of light in vacuum, and Aeff is the effective mode area. TR=5fs is related to the slope of the Raman gain spectrum, which is assumed to vary linearly with frequency around the central frequency. The gain is given by g=gSS1+W/W0+(ω−ω0)2/Δω2,(6)where gSS is the small-signal gain, Δω is the gain bandwidth, which is chosen to correspond to 50 nm, and W(z)=∫|U|2dτ is the pulse energy. The gain is assumed to saturate over a large number of pulses with a response time much longer than the cavity roundtrip time. As such, the saturated values of the gain along the Er fiber are assumed to depend on average power only. W0 presents an effective gain saturation energy corresponding to the saturation power (determined by pump power) for a given repetition rate. The SA is modeled by a transfer function that describes its transmittance as T(τ)=1−q01+P(τ)/P0,(7)where q0 is the unsaturated loss, P(z,τ)=|U(z,τ)|2 is the instantaneous pulse power, and P0 is the saturation power. The specific shape of the transmittance function is found not to be important.

The numerical model can be solved with a standard split-step beam propagation algorithm, and the initial field is white noise. The stable solutions are reached from different initial noise fields.

So far, various mode-locked lasers have been developed by using the 2D noncarbon materials. Then, we will briefly review these lasers as follows.

Since 2009, besides graphene and its derivatives^[92–196], the 2D noncarbon materials have been used as excellent SAs to realize Q-switching^[197–244] and mode-locking^[245–321] in the fiber lasers. The performance of Q-switched fiber lasers based on 2D noncarbon materials is classified and summarized, as shown in Table 2. Among them, the maximum pulse energy of 1.525 μJ^[204] and minimum pulse width of 154.9 ns^[229] were demonstrated, respectively. As an example, Sun et al. obtained the stable Q-switched operation with pulse energy of 39.8 nJ and pulse repetition rate from 6.2 to 40.1 kHz in an Er-doped fiber laser (EDFL) based on a few-layer Bi2Se3 SA^[200], as shown in Fig. 8. It can be seen that the tunable operation of these fiber lasers based on 2D noncarbon materials was also reported^{[198,204,210,212,214,217,221,240,241]}. For the communication band, the widest tunable range of about 78.2 nm was achieved. For example, Chen et al. obtained the Q-switched pulses with per-pulse energy up to 1.5 μJ and tunable range from 1510.9 to 1589.1 nm in an EDFL Q-switched by a Bi2Te3 SA^[204]. Meanwhhile, besides near-IR Q-switching, visible^{[202,225,226,239]} and mid-IR operations were also demonstrated^{[199,209,213,238,241,243,244]}. These results clearly show that 2D materials possess saturable absorption, which can be exploited for the Q-switched fiber lasers applications.

Since 2009, the mode-locked fiber lasers have been developed by using the 2D materials, including graphene and its derivatives^[92–168], Bi2Se3^[245–249], Bi2Te3^[250–261], Sb2Te3^[262–267], MoS2^[279–288], WS2^[290–301], ReS2^[304], MoSe2^[305,306], SnS2^[307], and BP^[313–319], as shown in Table 3. These results indicate that, after the characteristics of graphene, TIs, and TMDCs, BP is another highly potential 2D material for ultrafast lasers. Here, we review them as follows.

For the TI SAs, the maximum average power of 45.3 mW, maximum repetition rate of 2.95 GHz^[253], and minimum pulse width of 70 fs^[267], are achieved in the mode-locked fiber lasers, respectively. For example, Zhao et al. demonstrated the tunable soliton operation with a 1.57 ps pulse in an EDFL by inserting a TI Bi2Se3 SA^[245]. They also obtained the mode-locked pulses with a pulse width of 1.21 ps in an EDFL with a Bi2Te3 SA^[250], as shown in Fig. 9. The fundamental mode-locking pulse had the large full width at half-maximum of 2.33 nm with the repetition rate of ∼1.11MHz. Liu et al. also obtained the soliton pulse with ∼660fs in a fiber ring laser by using a polyvinyl alcohol (PVA)-based Bi2Se3 SA^[246]. Gao et al. also demonstrated the conventional soliton with 908 fs duration and dissipative soliton (DS) with a 7.564 ps duration in an EDFL mode-locked by Bi2Se3 nanosheets interacting with a PCF^[248]. In addition, Xu et al. obtained the mode-locking pulses in an EDFL with pulse duration of 579 fs and a repetition rate of 12.54 MHz by using a bi-layer Bi2Se3 SA^[249]. Duan et al. also demonstrated the passively harmonic mode-locking operation with a tunable repetition rate from 232 to 390 MHz and pulse duration of 1.32 ps in an EDFL with a microfiber-based Bi2Te3 SA^[254]. Meanwhile, Lee et al. obtained the soliton pulses with a temporal width of ∼600fs in an EDFL by using a bulk-structured Bi2Te3 mode-locker^[255]. Yan et al. also demonstrated that the mode-locking operation in an Yb-doped fiber laser employing the Bi2Te3 solution filled in PCF as an effective SA^[258]. Recently, Wang et al. demonstrated passive mode-locking with pulse duration of 448 fs in an EDFL by using a TI Bi2Te3 SA^[259]. Mao et al. also proposed a soliton fiber laser passively mode-locked with two different types of film-like SAs, one of which is fabricated by mixing Bi2Te3 with de-ionized water, as well as PVA, and then evaporating them in a Petri dish, and the other of which is prepared by directly dropping Bi2Te3 solution on the PVA film^[260]. Interestingly, Lin et al. demonstrated a soliton mode-locking operation with pulse widths of 400 and 385 fs in an EDFL employing n- and p-type Bi2Te3 nanosheets, respectively^[261]. Sotor et al. also demonstrated the usage of a Sb2Te3 SA for efficient mode-locking of an EDFL, which was capable of generating soliton pulses with a pulse width of 1.8 ps and pulse energy of 105 pJ^[262]. They also demonstrated the soliton operation with a pulse width of 270 fs in the EDFL incorporating Sb2Te3 SA^[264]. From the point of view, TI SAs can act as promising modulators for ultrafast lasers.

For the TMDC SAs, the minimum pulse width of 67 fs^[295], maximum average power of 110 mW^[296], and maximum repetition rate of 3.27 GHz^[306] were demonstrated in the mode-locked fiber lasers, respectively. For example, Zhang et al. demonstrated the generation of a stable mode-locked laser pulse with a 3 dB spectral bandwidth of 2.7 nm and a pulse duration of 800 ps in an Yb-doped fiber laser with a few-layer MoS2-based mode-locker^[279]. Xia et al. also achieved an EDFL mode-locked by a multilayer MoS2 SA. The soliton pulses have a central wavelength, spectral width, pulse duration, and repetition rate of 1568.9 nm, 2.6 nm, 1.28 ps, and 8.288 MHz, respectively^[280]. In addition, Liu et al. obtained the femtosecond pulse in an EDFL by using a MoS2 SA^[282]. A 710 fs pulse centered at 1569.5 nm wavelength with a repetition rate of 12.09 MHz has been achieved with proper cavity dispersion. Meanwhile, Luo et al. demonstrated the mode-locking regime with pulse duration of 1.4 ps in an EDFL based on a few-layer MoS2 SA^[287]. Ahmed et al. also demonstrated the self-started soliton pulse from an Er-doped fiber ring laser incorporating MoS2 SA^[288]. The pulse energy, pulse width, and repetition rate of the laser were 74 nJ, 0.83 ps, and 17.1 MHz, respectively. These findings suggest that MoS2 is a promising SA for ultrafast laser operation.

Besides the MoS2 SA, WS2 SAs were also developed. For example, Yan et al. demonstrated the fundamental mode-locking with pulse duration of 675 fs and harmonic mode-locking with pulse duration of 452 fs and 1 GHz repetition rate in an EDFL by incorporating a WS2 film SA fabricated by PLD^[291], as shown in Fig. 10. Wu et al. also demonstrated the mode-locking operation with a pulse width of 595 fs in a fiber laser incorporated with a WS2 SA^[290]. In addition, Khazaeinezhad et al. demonstrated an all-fiber mode-locked laser with pulse duration of 369 fs based on a tapered fiber SA enclosed in WS2 nanosheets^[292]. Liu et al. also demonstrated the soliton pulse with the 246 fs duration and 57 nm bandwidth in an EDFL with a fiber-taper WS2 SA, which was fabricated by using the PLD method^[294]. Meanwhile, Mao et al. achieved the soliton mode-locking operations in an EDFL using two types of WS2-based SAs, one of which is fabricated by depositing WS2 nanosheets on a D-shaped fiber, while the other is synthesized by mixing WS2 solution with PVA and then evaporating it on a substrate^[296]. Li et al. also demonstrated the femtosecond fundamental mode-locking with a repetition rate of 10.2 MHz and pulse duration of 660 fs and harmonic mode-locking with a repetition rate of 460.7 MHz and pulse duration of 710 fs in an EDFL with a D-shaped fiber-based WS2 solution SA^[297]. Interestingly, Chen et al. demonstrated the passive mode-locking with pulse duration of 395 fs and 1 ns in a ring-cavity EDFL based on the microfiber-based WS2 SA device by utilizing the PLD method and fiber tip integrated WS2 SA mirror (SAM) by utilizing the magnetron sputtering technique, respectively^[299]. They also demonstrated the soliton pulses with a pulse duration of 1.49 ps and large average power of 62.5 mW in the EDFL by transferring a few-layered WS2 with a large area on the facet of the pigtail, and it acted as an SA^[301]. These works demonstrated that WS2 was a competitive material as SAs in mode-locked fiber lasers.

Followed by these works, other TMDCs were developed. For example, Mao et al. demonstrated that few-layer ReS2 displays saturable absorption properties at 1.55 μm and mode-locking of EDFLs by incorporating a film-type ReS2-PVA SA^[304]. Luo et al. also achieved the soliton mode-locking with pulse duration of 1.45 ps in an EDFL by exploiting a few-layer MoSe2 SA^[305]. By optimizing the polarization state, the soliton mode-locking was also obtained, emitting a train of pulses with the duration of 1.6 ps and a fundamental repetition rate of 5.48 MHz. In addition, Koo et al. demonstrated the use of a bulk-like, MoSe2-based SA as a passive harmonic mode-locker for the production of femtosecond pulses from a fiber laser at a repetition rate of 3.27 GHz and pulse width of 737 fs^[306]. Yang et al. also demonstrated the mode-locking operation at 1.03, 1.56, and 1.91 μm in Er-, Yb-, and Tm-doped fiber lasers by using SnS2 covered on a D-shaped fiber^[307]. Li et al. also achieved a passively mode-locked Yb-doped fiber laser based on the SnS2-PVA film SA^[308]. Its 3 dB bandwidth, pulse width, pulse repetition rate, and maximum average output power were 8.63 nm, 656 ps, 39.33 MHz, and 2.23 mW, respectively. These results suggest that TMDCs are promising 2D materials for ultrafast laser applications.

For the BP SAs, the maximum average power of 80 mW^[296], maximum repetition rate of 60.5 MHz^[314], and minimum pulse width of 242 fs^[316] were demonstrated in the mode-locked fiber lasers, respectively. Considering the direct and flexible bandgap for different layers of BP, these works may provide a possible method for fabricating BP SAM to achieve ultrafast lasers. For example, by incorporating the BP-based SA device into the EDFLs, Chen et al. obtained either the passive Q-switching (with maximum pulse energy of 94.3 nJ) or the passive mode-locking operation (with pulse duration down to 946 fs). Sotor et al. also demonstrated the soliton pulse with the 272 fs duration in an EDFL by using a BP SA^[316]. Li et al. also demonstrated the soliton pulse with the 786 fs duration in an EDFL by using a BP SA^[315]. Other operation states, including bound-soliton pulses and noise-like states, have also been observed^[317]. In addition, Xu et al. demonstrated the ultrashort pulse with pulse duration of ∼635fs in an EDFL by using smaller sized BP as an SA^[318]. These works reveal that BP is a class of promising and reliable SAs for ultrafast lasers.

Meanwhile, wavelength-tunable operation in the mode-locked lasers based on 2D materials also was developed rapidly. Although various approaches, including an unbalanced Mach–Zehnder interferometer, bandpass filter, or Sagnac fiber filter have been used as the wavelength selective element, the 2D materials-based SAs exhibit more advantages. The maximum tunable range of ∼40nm was demonstrated^[300]. For example, Zhang et al. obtained the picosecond pulses, tunable from 1535 to 1565 nm in a mode-locked fiber laser with a few-layer MoS2 SA^[286]. Yan et al. also demonstrated the self-starting mode-locking with a pulse width of 1 ns in a linear EDFL^[300]. In addition, Luo et al. also demonstrated the mode-locked pulse with a duration of 940 fs and wavelength tunable from 1532 to 1570 nm in an EDFL by using a microfiber-based BP SA^[313]. Chen et al. also demonstrated that the stable passive mode-locking operation and, eventually, a record 280 fs transmission limited soliton pulse in an all-anomalous dispersion fiber laser cavity could be ensured^[314]. These findings suggest that 2D materials could be promising SAs for broadband wavelength-tunable applications.

As an important technique to increase the pulse repetition rate, the harmonic mode-locking lasers based on 2D materials have been developed. The maximum repetition rate of 3.27 GHz was demonstrated^[306]. For example, Luo et al. demonstrated the passive harmonic mode-locking with a repetition rate of 2.04 GHz in an EDFL by using a microfiber-based Bi2Te3 SA^[251]. Yan et al. also obtained the stable harmonic soliton pulse generation with repetition rate of 2.95 GHz and output power of 45.3 mW in an EDFL by using a microfiber-based Bi2Te3 SA, which was fabricated by the PLD method^[253]. In addition, Lee et al. demonstrated femtosecond harmonic mode-locking with temporal widths of 630–700  fs and harmonic frequency of 773.85 MHz in an Er-doped fiber ring cavity using a bulk-structured Bi2Te3 deposited on a side-polished fiber as a mode-locker^[256]. Sotor et al. also demonstrated the harmonic mode-locking with a repetition rate of 304 MHz and 2.2 ps duration in an EDFL by using a microfiber-based Sb2Te3 SA^[262]. Meanwhile, Liu et al. obtained the generation of high-order harmonic mode-locking with a 2.5 GHz repetition rate in a fiber laser using a microfiber-based MoS2 SA^[282]. Wu et al. also demonstrated the passive-mode-locking operation of a fiber laser with a fundamental repetition rate of 463 MHz based on MoS2 SA^[285]. These works will open up promising optoelectronic applications of 2D materials for the HML lasers.

High-energy optical pulses have attracted much interest due to their wide applications in optical sensing, optical frequency metrology, and data inquiring process. Among them, DSs in passively mode-locked lasers were extensively developed in recent years, because they allow for an increase in pulse energy of lasers by several orders of magnitude in comparison to the conventional soliton pulses. The study found that DSs are generated from the lasers in the normal dispersion regime through the mutual nonlinear interaction among the cavity dispersion, Kerr nonlinearity, spectral filtering effect, and cavity loss. Its dynamics is governed by the cubic–quintic GLE. Up to date, DSs were demonstrated in the lasers by using various schemes.

Since 2010, besides graphene and its derivatives^[322–331], the DS fiber lasers have been demonstrated by using the 2D noncarbon materials, including Bi2Se3^[332], Bi2Te3^[333,334], Sb2Te3^[335–337], MoS2^[280,338], MoSe2^[339], WS2^[340,341], and BP^[342], as shown in Table 4. Among them, the maximum pulse energy of 22.4 nJ^[334] and minimum pulse width of 128 fs^[335] were demonstrated, respectively. As an example, Dou et al. obtained an all-normal-dispersion Yb-doped fiber laser based on a few-layer Bi2Se3 SA delivering with pulse energy of 0.756 nJ, pulse width of 46 ps, and repetition rate of 44.6 MHz^[322]. In addition, Wang et al. obtained the stable DS with a 3 dB spectral bandwidth of 51.62 nm, a pulse width of 6.2 ps, and tunable range of 21.9 nm in an EDFL incorporating a polymethylmethacrylate (PMMA)-TI-PMMA SA, respectively^[333]. Boguslawski et al. also obtained the DSs with a 34 nm full width at half-maximum and a pulse width of 167 fs in an EDFL mode-locked by Sb2Te3 in the near-zero dispersion regime^[336]. By using the few-layer WS2, Mao et al. demonstrated the DS fiber lasers at 1.55 μm based on evanescent field interaction with WS2 nanosheets^[341]. In the EDFL, the pulse width and pulse energy of the DS is 21.1 ps and 2.2 nJ, respectively, as shown in Fig. 11. Notably, the rectangular pulse based on the DS resonance was also developed^[343].

In the past decade, ultrafast laser sources covering from 2 to 20 μm have received wide interest due to their potential applications in environmental monitoring, pollution control, food quality control, laser surgery, spectroscopy, nonlinear optics, and defense. Passive mode-locking, due to its simple configuration, has been a more preferred choice to obtain ultrashort mid-IR pulses. Although the SESAMs have been developed as a mode-locker or Q-switcher, they have some drawbacks, such as a complex design for improving damage threshold and narrow operation range. Fortunately, the rise of 2D materials including graphene, TIs, TMDCs, and BP provide an excellent opportunity for lasers in the mid-IR band due to their broadband saturable absorption, ultrafast carrier dynamics, and planar characteristic.

Since 2012, besides graphene and its derivatives^[344–357], the mode-locked fiber lasers at 2 μm have been developed by using the 2D noncarbon materials, including Bi2Te3^[358,359], MoS2^[360], WS2^[361], WTe2^[362], and BP^[363,364], as shown in Table 5. Among them, the maximum pulse energy of 15.5 nJ^[360], maximum repetition rate of 36.8 MHz, and minimum pulse width of 739 fs^[363] were demonstrated, respectively. As an example, Sotor et al. demonstrated a mode-locked, Tm-doped fiber laser with a pulse width of ∼739fs by using a few-layer BP SA^[363]. The SA was based on mechanically exfoliated BP deposited on a fiber connector tip, as shown in Fig. 12. Jung et al. also demonstrated a femtosecond mode-locked, Tm/Ho co-doped fiber laser with a pulse width of ∼795fs at 1935 nm by using a bulk-structured Bi2Te3 SA^[358]. The SA was prepared by depositing a mechanically exfoliated, ∼30-μm-thick Bi2Te3 layer on a side-polished fiber. Interestingly, they also demonstrated the Tm/Ho co-doped fiber laser mode-locked with a WS2-deposited side-polished fiber^[361]. Inserting the SA into the cavity, stable mode-locked pulses at 1941 nm with a temporal width of ∼1.3ps at a repetition rate of 34.8 MHz were obtained. In addition, Tian et al. demonstrated the linear-cavity Tm3+-doped fiber laser mode-locked with a multilayer MoS2 SAM^[360]. The mode-locking operation at 1905 nm is realized with maximum output power of 150 mW, pulse width of 843 ps, and repetition rate of 9.67 MHz, respectively. These results indicate that 2D materials are potential SAs at the mid-IR band.

Recently, the pulsed fiber lasers at 3 μm based on 2D materials, including graphene^[369,370], Bi2Te3^[365], BP^[366,367], and Cu2−xS nanocrystal^[368], were also developed. Among them, the minimum pulse width of 6 ps^[365], maximum pulse energy of 25.5 nJ, and maximum repetition rate of 24 MHz^[366] were demonstrated, respectively. For example, Yin et al. demonstrated the mode-locking operation of the mid-IR mode-locked HoPr-ZrF4-BaF2-LaF3-AlF3-NaF (ZBLAN) fiber laser at 2830 nm with a TI Bi2Te3 SA^[361]. The mode-locking pulse has a pulse width of ∼6ps, maximum pulse energy of 8.6 nJ, and pulse repetition rate of 10.4 MHz. In addition, Qin et al. successfully fabricated a mid-IR SAM by transferring the mechanically exfoliated BP onto the gold-coated mirror^[366]. With the as-prepared BP SAM, a passively mode-locked Er:ZBLAN fiber laser is demonstrated at 2783 nm, which delivers a maximum average output power of 613 mW, a repetition rate of 24 MHz, and a pulse duration of 42 ps, as shown in Fig. 13. Interestingly, Li et al. also demonstrated the Q-switched Ho3+-doped and mode-locked Ho3+/Pr3+-doped ZBLAN fiber laser incorporating the same BP SAM^[367]. The Q-switching pulses at ∼2970.3nm with the pulse energy of 4.93 μJ and mode-locking with the pulse duration of 8.6 ps at ∼2866.7nm were obtained, respectively. These results verify that BP is a novel kind of highly promising optical material for generating ultrafast lasers in the mid-IR region after the previous 2D SAs, such as graphene, TIs, and TMDCs.

Multiwavelength pulsed lasers are of great importance due to their versatile applications in optical communication, biomedical research, and radar systems. So far, several actively/passively mode-locked techniques have been exploited for achieving multiwavelength mode-locked pulses in fiber lasers. Compared with active schemes, passive schemes share more benefits, such as very simple, compact, low-cost without a modulator required, and easily achieved mode-locked pulses with ultrahigh peak power.

To date, several passive schemes have been used to generate the multiwavelength mode-locked pulses, i.e., nonlinear polarization rotation (NPR), nonlinear optical loop mirror (NOLM), and real SAs (e.g., SESAM, carbon nanotubes). Interestingly, the multiwavelength mode-locked lasers based on 2D materials, including graphene^[371–375], Bi2Se3^{[343,376–378]}, Bi2Te3^[379], TMDCs^[380–382], BP^[383–385], and graphene nanocomposite^[386], were also developed. Their performances are summarized in Table 6. Among them, the maximum wavelength number of 4^[376], maximum repetition rate of 388 MHz^[379], minimum pulse width of 585 fs^[380], and maximum pulse energy of 6.64 nJ^[381] were demonstrated, respectively. These results suggest that 2D materials are promising candidates for realizing multiwavelength pulsed lasers.

For example, Guo et al. demonstrated the four-wavelength mode-locked EDFL incorporating a Bi2Se3 SA, which can be both an excellent SA for mode-locking and a high-nonlinearity medium to induce a giant third-order optical nonlinear effect for mitigating the mode competition and stabilizing the multiwavelength oscillation^[376]. For the four-wavelength operation, they obtained its pulse width of ∼22ps and pulse energy of 1.1 nJ at the pump power of 155 mW. The group also obtained the dual-wavelength soliton pulses in an EDFL with a WS2-based fiber taper, which was fabricated by depositing the few-layer WS2 onto a fiber taper under the femtosecond-level pulsed laser^[380,381], as shown in Fig. 14. The pulse width is 585 and 605 fs for the dual-wavelength soliton, respectively^[380]. The group also demonstrated the dual-wavelength rectangular pulse EDFL with a Bi2Se3 SA^[377]. The rectangular pulse could be stably initiated with pulse width from 13.62 to 25.16 ns and fundamental repetition rate of 3.54 MHz. Interestingly, the group also demonstrated a switchable dual-wavelength soliton fiber laser based on a graphene ternary composite^[386]. The dual-wavelength soliton is stably initiated with a minimum pulse width of 1.25 ps and pulse energy of 1.51 nJ. Further investigations in this respect are on the way.

Interestingly, Liu et al. demonstrated the dual-wavelength passively harmonic mode-locking operation with repetition rates of 388 and 239 MHz in an EDFL with a microfiber-based Bi2Te3 SA^[379]. They also achieved the dual-wavelength pulse cluster phenomena in an EDFL, incorporating the microfiber-based BP quantum dot (QD) device^[385]. In the case of the dual-wavelength pulse cluster, each wavelength corresponds to one sequence of the pulse cluster, which has different amplitudes and time intervals. Zhao et al. also obtained a picosecond-level BP mode-locked EDFL with a synchronous tri-wavelength of 1557.2, 1557.7, and 1558.2 nm^[383]. Yun also reported the usage of few-layer BP as SAs to achieve dual-wavelength vector soliton mode-locking in an EDFL^[384].

Solid-state lasers typically consist of a free-space cavity, which is formed by mirrors using doped glass or crystalline host materials as a solid-state gain media, and are the most commonly used in industry, research, and military applications. Since 2010, the Q-switched solid-state lasers based on 2D materials, including graphene and its derivatives^[387–412]Bi2Se3^[413–417], Bi2Te3^[418–420], MoS2^{[78,421–429]}, WS2^[430–438], WSe2^[439], ReS2^[440], BP^{[243,441–451]}, and g-C3N4^[452,453], were developed. Their performance is summarized as shown in Table 7. Among them, a variety of gain media have been coupled with 2D materials SAs in pulsed solid-state lasers. These include Nd:GdVO4^{[413,420,424,454]}, Nd:Lu2O3^[414], Nd:YVO4^{[415,427,434,450,455]}, Nd:LiYF4^[416], Yb:GdAl3(BO3)4^[418], Tm:LuAG^[419,431], Tm:YAlO3 (Tm:YAP)^{[420,442,443]}, Er:Y3Sc2Ga3O12 (Er:YSGG)^[420,440], Nd:Y2.8Ga5O12 (Nd:YGG)^[78], Tm, Ho:YGG^[78], Nd:YAlO3^[421], Yb:(LuxGd1−x)3Ga5O12(x=0.062) (Yb:LGGG)^[422], Tm-doped calcium lithium niobium gallium garnet (Tm:CLNGG)^[423], Tm:GdVO4^[424], Er:Lu2O3^[425,452], Tm:Ho:YAP^[426], Er:Y3Al5O12 (Er:YAG)^[428,449], Nd:Gd0.93Y2.04Sc2Ga3O12 (Nd:GYSGG)^[430], YVO4/Nd:YVO4^[432], Tm, Ho:LuLiF4 (Tm, Ho:LLF)^[433], Nd:Lu0.15Y0.85VO4^[435], Nd:YAG^[436], Er:SrF2^[441], Tm:CaYAlO4^[444], Er:Y2O3^[444], Tm:YAG^[445], Cr:ZnSe^[243], Yb:ScBO3^[446], Ho/Pr:LiLuF4^[447], Yb:CaYAlO4 (Yb:CYA)^[448], Nd:LLF^[453], Yb:YAG^[456,457], Pr:GdLiF4^[458], Nd:GdTaO4^[459], and Yb, Lu:CaGdAlO4 (Yb, Lu:CALGO)^[460]. The output spectra of these solid-state lasers have covered 0.6–3 μm. These results open a way to use the 2D materials in pulsed solid-state lasers.

For the TI SAs, the minimum pulse width of 850 ps^[420] and maximum pulse energy of 21.7 μJ^[436] are obtained in the Q-switched solid-state lasers, respectively. For example, Yu et al. demonstrated a bulk solid-state laser by using few-layer Bi2Se3 as the pulse modulator^[413]. Xu et al. also demonstrated a passively Q-switched Nd:LiYF4 laser at 1.3 μm based on a Bi2Se3 SA^[416]. The laser has a maximum average power of 0.2 W, corresponding to a pulse repetition rate of 161.3 kHz, shortest pulse width of 433 ns, and pulse energy of 1.23 μJ, respectively. In addition, Liu et al. demonstrated a Q-switched Tm:LuAG laser at 2027 nm using a Bi2Te3 SA^[419]. Q-switched pulses with an average power of 2.03 W are generated, corresponding to the maximum pulse energy of 18.4 μJ and peak power of 23 W. The pulse duration and repetition rate of the laser is 620 ns and 118 kHz, respectively. These results indicate the promising potential of TIs as nonlinear optical switches for achieving efficient pulsed solid lasers.

For the TMDC SAs, the minimum pulse width of 850 ps^[427] and maximum pulse energy of 2.63 mJ^[436] are achieved in the Q-switched solid-state lasers, respectively. For example, Wang et al. demonstrated Nd:GdVO4, Nd:YGG, Tm/Ho:YGG lasers at 1.06, 1.42, and 2.1 μm with a broadband MoS2 SA, respectively^[78]. Xu et al. also demonstrated an Nd:YAlO3 laser at 1079.5 nm using a MoS2 SA^[421]. The laser has a maximum average power of 0.26 W corresponding to a pulse repetition rate of 232.5 kHz, pulse width of 227 ns, and pulse energy of 1.11 μJ. In addition, Tang et al. demonstrated a YVO4/Nd:YVO4 laser by using WS2 as an SA^[433]. Under the pump power of 6.28 W, a maximum output power of 1.36 W and pulse duration of 56 ns with a repetition rate of 1.03 MHz are obtained, resulting in a peak power as high as 23.6 W. Interestingly, Ling et al. demonstrated a Tm, Ho:LuLiF4 laser by using a WS2 SA^[434]. The laser has a maximum power of 0.088 W, corresponding to a pulse repetition rate of 16.89 kHz, pulse width of 4 μs, and pulse energy of 5.21 μJ. Recently, Su et al. demonstrated an Er:YSGG laser at 2.8 μm by using a ReS2 SA^[440]. Under a pump power of 920 mW, an average power of 104 mW was obtained with a pulse width of 324 ns and a repetition rate of 126 kHz. These results indicate that TMDCs have a prosperous application future in solid-state lasers.

For the BP SAs, the minimum pulse width of 2.86 ns and maximum pulse energy of 2.4 mJ are demonstrated, respectively^[450]. For example, Chu et al. demonstrated a Tm:YAP laser at 1988 nm by using a BP SA^[443]. Under the pump power of 3.38 W, an average power of 151 mW was generated with a minimum pulse width of 1.78 μs and repetition rate of 19.25 kHz, corresponding to pulse energy of 7.84 μJ. Kong et al. also demonstrated the Q-switching operation of Yb:LuYAG, Tm:CaYAlO4, and Er:Y2O3 lasers at 1.03, 1.93, and 2.72 μm by using a broadband BP SAM, respectively^[444]. In addition, Wang et al. demonstrated the mid-IR pulse generation from a BP Q-switched Cr:ZnSe laser^[243]. A pulse as short as 189 ns with an average power of 36 mW was realized at 2.4 μm, corresponding to a repetition rate of 176 kHz and single-pulse energy of 205 nJ, respectively. Nie et al. also demonstrated a Ho, Pr:LLF laser operating at 2.95 μm using a BP SAM^[447]. Under the absorbed pump power of 7.36 W, the shortest pulse width of 194.3 ns was obtained. Ma et al. also obtained the stable Q-switched pulses with a pulse width of 620 ns at 1046 nm in a Yb:CaYAlO4 laser with the BP SAM^[448]. The generated pulse train has a repetition rate of 113.6 kHz and an average power of 37 mW. Interestingly, Liu et al. demonstrated the vortex pulses at 1645 nm from an Er:YAG laser based on the BP SA^[449]. Laser guided (LG0,−1) mode pulses with maximum pulse energy of 2.15 μJ and LG0,+1 mode pulses with maximum 2.4 μJ pulse energy were obtained individually, as well as pulsed LG0,±2 laser beams. Recently, Liu et al. demonstrated an Nd:YVO4 laser by using BP, WS2, and MoS2 solutions as SAs^[450]. The pulse durations of the lasers were 2.86, 3.99, and 5.40 ns with the pulse energy of 0.166, 0.150, and 0.365 mJ, respectively. Li et al. also demonstrated a Q-switched Yb3+-doped ScBO3 laser using a BP SA^[451]. Its maximum output energy is 1.4 μJ. These results show that BP might be considered as a universal broadband SA that could compete with other 2D nanomaterials.

Besides single-wavelength operation, multiwavelength Q-switching was also reported. The minimum pulse width of 181 ns and maximum single-pulse energy of 39.5 μJ were demonstrated, respectively^[442]. For example, Wang et al. first demonstrated a dual-wavelength Q-switched Nd:Lu2O3 laser at 1066.6 and 1066.8 nm with a TI:Bi2Se3 SA^[414]. Jia et al. also demonstrated a dual-wavelength Q-switched c-cut Nd:YVO4 laser with a TI:Bi2Se3 SA^[415]. Its repetition rate is tunable in a wide range of 1–135 kHz, where the pulse width is 250 ns, and maximum pulse energy is 0.56 μJ. In addition, Sun et al. demonstrated a triple-wavelength Q-switched Yb3+:GdAl3(BO3)4 laser at 1043.7, 1045.3, and 1046.2 nm based on a TI:Bi2Te3 SA^[418]. The shortest pulse width was 370 ns with 110 kHz pulse repetition rate. The maximum average power was 57 mW with the pulse energy of 511.7 nJ. Lou et al. also obtained a dual-wavelength Q-switching at 1025.2 and 1028.1 nm in an Yb:LGGG laser based on the MoS2 SAM^[422]. The maximum average power of 0.6 W with a pulse width of 182 ns was obtained, corresponding to single-pulse energy of 1.8 μJ. Gao et al. also demonstrated the stable dual-wavelength Q-switched laser operation with a maximum average power of 367 mW, pulse repetition rate of 70.7 kHz, pulse width of 591 ns, and pulse energy of 1.05 μJ in an Nd:GYSGG laser using a WS2 SAM^[431]. Interestingly, multiwavelengh Q-switching in the mid-IR band was studied. For example, Liu et al. demonstrated a dual-wavelength Q-switched Er:SrF2 laser at 2.79 μm by using a BP SA^[441]. The maximum average output power of 180 mW with pulse duration of 702 ns and repetition rate of 77.03 kHz was achieved at a pump power of 2.47 W. Zhang et al. also achieved a dual-wavelength Q-switched Tm:YAP laser at 1969 and 1979 nm with a BP SAM^[442]. The pulses with a duration of 181 ns, average power of 3.1 W, and pulse energy of 39.5 μJ were generated at a pulse repetition rate of 81 kHz. These results well prove that 2D materials are reliable optical modulators for multiwavelength solid-state lasers.

Notably, besides the single scheme, hybrid Q-switching was also demonstrated. For example, Qiao et al. proposed a diode-pumped passively Q-switched laser with a MoS2 SA^[427]. At an incident pump of 6.54 W, a maximum output power of 1.15 W with a pulse duration of 70.6 ns was obtained. Then, by using a hybrid Q-switched laser with a MoS2 SA and an acousto-optic modulator (AOM) as a pumping fundamental laser, a sub-nanosecond KTiOPO4 (KTP) -based intracavity optical parametric oscillation was realized. With an incident pump power of 10.2 W and AOM repetition rate of 10 kHz, the maximum output power of 183 mW with minimum pulse duration of 850 ps was obtained. In addition, Tang et al. demonstrated a dual-loss modulation Q-switched and mode-locked Nd:Lu0.15Y0.85VO4 laser with the pulse duration of 467 ps and peak power of 731 kW by simultaneously employing the WS2 and electro-optic modulator (EOM) as a modulator^[437]. Sun et al. also demonstrated a passively Q-switched Nd:YAG laser at 946 nm by using a WSe2 SA and an EOM^[439]. The laser has an average power of 0.1 W, corresponding to a pulse repetition rate of 0.5 Hz, pulse width of 10.8 ns, and pulse energy of 2.63 mJ. These findings indicate that the hybrid Q-switched laser may exhibit more advantages and a need for further study.

Other 2D materials such as g-C3N4 were also proposed. For example, Fan et al. demonstrated a stable Q-switched Er:Lu2O3 laser at 2.84 μm by using a g-C3N4 SA, generating the pulse duration of 351 ns and an average output power of 1.09 W. For example, Fan et al. demonstrated a stable Q-switched Er:Lu2O3 repetition rate of 99 kHz, corresponding to the pulse energy of 11.1 μJ^[452]. They also used the g-C3N4 nanosheets as an SA in a passively Q-switched Nd:LLF laser at 1.3 μm^[453]. Under an incident pump power of 9.97 W, the pulse duration of 275 ns was acquired with output power of 0.96 W and pulse repetition rate of 154 kHz, resulting in the pulse energy of 6.2 μJ. These results indicate a great potential for g-C3N4 as a new SA in lasers.

The study found that, compared with the Kerr-lens mode-locking, 2D materials-based SAs share more advantages, such as self-starting and alignment-free. To make them convenient for application, 2D materials are normally coated on high reflectivity mirrors and then employed as cavity mirrors. Since 2010, the mode-locked solid-state lasers based on 2D materials, including graphene and its derivatives^[461–492]WS2^[456], MoS2^[458,459], and BP^{[455,457,460]}, were developed. Their performances are summarized in Table 8. Among them, the minimum pulse width of 272 fs and maximum pulse energy of 6.48 nJ were demonstrated, respectively. For example, Su et al. demonstrated a femtosecond solid-state laser modulated by a BP SA^[455]. Pulses as short as 272 fs were achieved with an average power of 0.82 W, corresponding to the pulse energy of 6.48 nJ and peak power of 23.8 MW. In addition, Hou et al. obtained excellent mode-locking performance by applying the samples into a solid-state Yb: YAG laser based on a few-layer WS2 SAM^[456]. The pulse repetition rate was 86.7 MHz with a pulse width of 736 fs. The maximum output power was 270 mW, and the peak power was 4.23 kW. Wang et al. also demonstrated the Q-switched mode-locking operation at 1066 nm of the Nd:GdTaO4 laser by using the MoS2 SAM^[459]. The maximum average output power of 156 mW was obtained. Meanwhile, Zhang et al. demonstrated ultrafast pulse generation at 1064.1 nm from a bulk laser by employing the BP SAM^[457], as shown in Fig. 15. Pulses as short as 6.1 ps with an average power of 460 mW were obtained.