Advances and Challenges in Pulsed Lasers Based on Low-Dimensional Material Saturable Absorbers

Abstract

Low-dimensional materials (LDMs) are often favored by researchers in the ultrafast photonics field for their low optical loss, ultrafast carrier response, broadband nonlinear absorption, and easy integration with optoelectronic systems. High-performance broadband saturable absorbers (SAs) fabricated from LDMs have become core components for achieving compact and miniaturized ultrafast laser. This paper systematically reviews the laser applications of LDM SAs in the near/mid-infrared spectral region, focusing on the pulse modulation mechanisms, material systems, and device integration approaches. It analyzes current research progress and challenges while outlining future development trends for LDM SAs in ultrafast pulsed lasers and optoelectronic devices.

1. Introduction

Laser technology is critical to the development of modern science, industry, medicine, and communications [1,2,3]. Pulsed lasers operating at nanosecond, picosecond, and even femtosecond scales emit a significant quantity of energy within an ultrashort time interval that leads to their giant pulse energy and ultra-high peak power. They serve as indispensable instruments in domains such as ultrafast process analysis, optical storage, quantum computing, and micro-nano machining [4,5,6,7]. Typical types of lasers encompass solid-state lasers [7], optical parametric amplifiers [8], semiconductor lasers [9], and fiber lasers [10]. Laser pulsed modulation techniques can be classified into active modulation (e.g., electro-optic modulation [11], acousto-optic modulation [12], Mach-Zehnder modulators [13]) and passive modulation (e.g., saturable absorbers [14] and nonlinear polarization rotation [15]). Among them, passive Q-switching (QS) and mode-locking (ML) technologies based on saturable absorbers (SAs) have attracted much attention, as theirs structural simplicity, fast response, and low cost [16,17,18]. Exploring novel saturable absorbers (SAs) with high-performance is of great interest for compact and miniaturized commercial ultrafast pulsed lasers.

Early SAs include organic dyes [19], transition-metal-ion-doped crystals [20], color-center crystals [21], and semiconductor saturable absorber mirrors (SESAMs) [22]. The poor thermal stability of organic dyes and the long lifetime of transition-metal-ion-doped crystals limit their practical usage. SESAMs consist of a Bragg mirror and a semiconductor absorption layer (composed of III-V group compounds such as AlGaAs, InGaAs, or InAlAs). The saturable absorption characteristics can be realized through the adjustment of the thickness of the semiconductor absorption material and the quantity of Bragg mirror layers [23,24,25]. SESAMs are commonly used in various mode-locked lasers for their stable operation and high reproducibility, but they still exhibit wavelength sensitivity resulting from narrow nonlinear optical bandwidths and sophisticated fabrication processes. In recent years, a series of low-dimensional nanomaterials (LDMs), including graphene [26], black phosphorus (BP) [27], transition metal dichalcogenides (TMDs) [28,29,30], MXenes [31], carbon nanotubes (CNTs) [32], transition metal oxides (TMOs) [33], topological insulators (TIs) [34], metal-organic frameworks (MOFs) [35], and covalent organic frameworks (COFs) [36], have demonstrated advantages such as ultrafast carrier response, broadband modulation, low insertion loss, and tunable layered structures, spawning a series of innovative applications in ultrafast photonics. With the development of the integration processes with laser devices, the problems of stability and damage threshold of LDMs are continuously being improved, significantly driving the enhancement of laser performance.

This paper, beginning with an overview of passive QS and ML technologies based on SAs, reviews recent progress on LDM SAs, including graphene, BP, TMDs, heterostructures, and so on. Their laser applications in the near/mid-infrared spectral region is emphatically discussed. Furthermore, the novel advances on synthesis methods of LDM SAs are explored. This review finally summarizes the challenges associated with the future development of LDM SAs, which provides an outlook on how to achieve more desirable laser performance by enhancing the stability, damage threshold, and other saturable absorption parameters.

2. Passive Modulation Mechanism of SAs

SAs exhibiting an ultrafast nonlinear saturable absorption effect have emerged as the core component for realizing passive QS and ML technologies in lasers. Under the irradiation of an intense laser, the interaction between light and matter no longer follows the Lambert-Beer law but enters the nonlinear region. The relationship between the polarization intensity P and the electric field intensity E of the material needs to be expanded using a power series:

$$P={\epsilon }_0({\chi }^{\left(1\right)}E+{\chi }^{\left(2\right)}{E}^2+{\chi }^{\left(3\right)}{E}^3+\cdots )$$

Saturable absorption falls under the category of third-order nonlinear optical effects, ${\chi }^{\left(3\right)}$, which directly relates to the variation in the imaginary part of the material’s complex refractive index (i.e., the absorption coefficient) with light intensity. Its constitutive relationship can be expressed as:

$$\alpha \left(I\right)={\displaystyle \frac{{\alpha }_0}{1+I/I_S}}$$

where α₀ is the linear absorption coefficient, I is the incident laser intensity, I_s is the saturation intensity. The value of the saturation intensity depends on the type of SAs. Upon the incidence of a weak laser pulse on the SA, it demonstrates linear optical absorption. The incident laser wavelength is chosen such that the photon energy exceeds the band-gap energy. As the laser intensity increases, the ground-state electrons are gradually depleted and the excited-state sub-bands become filled. At this point, electrons in the ground state of the materials undergo a transition to an excited state through the absorption of photon energy. Subsequently, certain hot electrons with diminished energy occupy lower energy levels and relax back to the ground state in accordance with the Fermi-Dirac distribution. Given that the transition rate of ground-state electrons is significantly higher than the relaxation rate of excited-state electrons, overall, the quantity of ground-state electrons gradually diminishes, while the quantity of excited-state electrons increases. According to the Pauli exclusion principle, the material can no longer absorb incident photons when the ground-state electrons are depleted and the excited state sub-bands are filled. Consequently, SAs exhibit absorption saturation (i.e., they are “transparent” to light) under high-intensity laser irradiation.

2.1. Passive Q-Switching

Building on active Q-switching (QS) with electro-optic Kerr cells [37], P. P. Sorokin et al. used metal phthalocyanine as a SA to achieve passive QS for the first time in a ruby laser in 1964 [19]. Such passive QS technology, which has the advantages of a simple structure, no need for external driving, cost-effectiveness, and durability, rapidly became a key research direction in the field of military and research QS lasers. The generation of QS pulses is dependent on the sudden alteration in the quality factor (Q) of the laser resonator:

$$Q=2\pi v_0{\displaystyle \frac{E}{c\delta E/nL}}={\displaystyle \frac{2\pi nL}{\delta {\lambda }_0}}$$

where E represents the intracavitary energy, δ is the optical loss per round trip, L is the cavity length, n is the refractive index of the intracavity medium, c is the speed of light, λ₀ is the central wavelength of lasers in vacuum and ν₀ is the central frequency of lasers. Owing to the high cavity loss (i.e., the low Q-value) during the initial phase of laser pumping, the intracavity gain fails to offset the laser threshold, resulting in the continuous accumulation of the population inversion density at the upper laser level. When intracavitary energy reaches a certain value, SAs become transparent to light and the resonator’s Q-value suddenly increases. Meanwhile, the upper-level population undergoes an avalanche-like transition through stimulated emission, releasing energy instantaneously and producing large-energy QS pulses with durations ranging from a few nanoseconds to multiple microseconds.

2.2. Passive Mode-Locking

Passive mode-locking (ML) represents the most prevalently employed technique for attaining ultrashort pulses on the picosecond and even femtosecond scales. Distinct from passive QS [16,17,18], ML attains a fixed phase relationship among all cavity longitudinal modes through multiple rounds of linear amplification of the gain medium and saturable absorption of the SA. The SA acts as a loss modulator that is dependent on light intensity: it exhibits high loss for weak light (such as noise floor or pulse edges), while it saturates and “bleaches” to low loss for strong short pulses. After multiple cavity roundtrips, strong pulses are preferentially amplified, while weak signals are suppressed. This self-amplitude modulation effect, combined with the fast recovery characteristics of the saturable absorber, can continuously compress the pulse front and tail, ultimately forming a stable ultrashort pulse sequence. Simultaneously, the saturable absorber possesses both fast and slow recovery components: the slow recovery component aids in the self-starting of mode-locking, while the fast recovery component is responsible for pulse shaping on the sub-picosecond scale, thereby achieving spontaneous establishment of stable mode-locking from noise and maintaining ultrashort pulse output in a steady state [38].

In 1992, the emergence of SESAMs spurred the rapid development of passive ML technology. Keller et al. utilized molecular beam epitaxy to grow a GaAs layer with saturable absorption properties on a Bragg reflector and incorporated it into a titanium-sapphire laser to obtain ML pulses with a pulse width of 2 ps [22]. In comparison with earlier organic dyes and color-center crystals [19,21], SESAMs are characterized by a simple and compact structure, long lifetime, ultrafast response, as well as self-starting and self-sustaining capabilities. Currently, they have realized large-scale commercial production and are extensively employed in various passively ML lasers [23,24,25]. Nevertheless, it is an undeniable fact that SESAMs are confronted with the issues of fixed operating wavelengths and intricate fabrication processes, which restricts their advancement in the domains of mid-infrared lasers and tunable lasers.

3. Research Progress on LDM SAs

In 2004, the successful application of one-dimensional CNTs-as saturable absorbers in fiber ML lasers drew the attention of researchers to nonlinear nanomaterials [39]. In recent years, LDMs, such as CNTs, graphene, TIs, TMDs, BP, TMOs, MXenes, MOFs and COFs as depicted in Figure 1, have emerged as research focuses for high-performance SAs in laser applications. This can be ascribed to their unique band structures, strong light-matter interactions, and ultrafast carrier dynamics [40]. The subsequent sections focus on reviewing the LDM SAs (especially one-dimensional and two-dimensional) that are of interest, along with their diverse material systems.

3.1. Graphene

Graphene is a two-dimensional (2D) honeycomb lattice material composed of a single layer of carbon atoms arranged via sp² hybridization [39,40], as demonstrated in Figure 2a. It is found to be a zero-bandgap semimetal with a linear overlap between valence and conduction bands at the K and K’ points (Dirac points) of the Brillouin zone (refer to Figure 2b,c) [40]. Thus, it exhibits a broadband photoresponse that spans almost the entire optical spectrum [47]. As depicted in Figure 2d, the ultrafast dynamics of photo-excited carriers are further verified, characterized by an initial rapid carrier-carrier intraband scattering time of 70 fs and an electron-hole interband recombination time of 1.7 ps [48]. This indicates that graphene has the potential for the generation of femtosecond ultrafast laser pulses.

In 2009, Bao et al. prepared atomic-layer graphene SA by chemical vapor deposition (CVD) and investigated its nonlinear saturable absorption properties (refer to Figure 2e,f) for the first time [26]. It exhibits a relatively larger modulation depth, accompanied by a saturation intensity that is one order of magnitude lower than that of SESAM. These intrinsic merits arising from its 2D structure are more favorable for the generation of ultrashort pulses. Upon integration into an erbium-doped fiber (EDF) laser, the duration of the acquired passive ML pulse is 756 fs, and the central wavelength is 1565 nm with a spectral linewidth of 5 nm (refer to Figure 3a–c). This study verified the application merits of graphene (e.g., broadband modulation, ultrafast relaxation times, excellent nonlinear absorption, low insertion loss and saturation intensity) in ultrafast photonics applications, thereby initiating research on graphene-based passive ML lasers.

Subsequently, the modulation wavelength of graphene-based SAs has undergone a rapid expansion into the near/mid-infrared region. In 2010, the mode-locking of a ceramic Nd:YAG solid-state laser (SSL) at 1064 nm utilizing solution-processed graphene as a SA was demonstrated [50]. In 2012, Ma et al. experimentally demonstrated a passively mode-locking laser with laser pulses as short as 729 fs at 2018 nm based on a graphene SAM [51], as shown in Figure 3d–f. Zhang et al. applied a graphene-polymer composite SA in a Tm-doped fiber (TDF) laser and achieved a passive ML pulse of 3.6 ps operating at a wavelength of 1.94 μm [52]. Later, multiform graphene SAs fabricated through mechanical exfoliation [53], ultrasonic liquid-phase exfoliation [54], and CVD [55] were utilized in ML fiber lasers operating at 2 µm, and the narrowest pulse width attained was 370 fs [53]. In 2013, a monolayer graphene SA was used in a passive ML solid-state Cr:ZnSe laser operating at ~2.5 um [56]. At the same time, a compact, graphene QS Er³⁺-doped ZBLAN fiber laser at 2.78 um was demonstrated [57]. These findings suggest that graphene holds the potential to function as a mid-infrared SA, which combines simplicity, cost-efficiency, and broadband modulation, thus promoting the development of mid-infrared ML lasers.

However, thin film nanomaterials are limited for high-power laser applications due to their low damage threshold and poor environmental stability. In 2013, Tang et al. devised a double-piece single-layer graphene plate and employed a collimating lens to augment the laser beam size on the SA, thereby significantly enhancing the damage threshold of graphene. The maximum output power of 5.2 W stable Q-switching 2 μm TDF laser is realized with pulse energy of up to 18 μJ [58]. Almost at the same time, an evanescent field-coupled graphene/PMMA SA was proposed and delivered ML pulses with a pulse energy of more than 10 nJ at 1565 nm [59], as shown in Figure 3g–i. This scheme mitigates optically induced thermal damage resulting from direct interaction and ensures a long nonlinear interaction length to enhance the pulsating capability of SA.

Since then, graphene-based passive ML lasers have been consistently dedicated to comprehensive breakthroughs in the directions of ultra-short pulse, high pulsed energy, mid-infrared modulation, and multi-functionality. Research studies of Sobon et al. in 2015 showed that the nonlinear optical parameters of SA can be controlled by scaling the number of graphene layers [60]. As shown in Figure 3j–l, 24-layer and 37-layer graphene were used in EDF and TDF lasers to achieve pulses with durations of 737 fs and 345 fs, respectively. The multilayer graphene-based SA exhibits a relatively higher modulation depth and a higher damage threshold than the single-layer one. The shorter pulses of 88 fs are generated directly from a ML EDF laser centered at 1545 nm based on a 60-layer graphene/PMMA SA (with a damage threshold of 460 µJ/cm² and a modulation depth of 11%) [61]. To date, the narrowest pulse of 29 fs has been obtained with an average power of 52 mW, which verifies that graphene serves as a simple, low-cost, and robust SA for applications demanding high temporal resolution [62].

Figure 3. The performance of passively modulated laser based on graphene. (a) Schematic of Er³⁺–doped fiber laser based on graphene–SA. (b) Spectrum. (c) Autocorrelation trace [26]. (d) Experimental setup of the mode–locked laser based on the graphene SAM. (e) Optical spectrum. (f) Autocorrelation trace of the mode–locked pulses [51]. (g) Experimental setup of the dissipative soliton Er–doped fiber laser using a graphene/PMMA saturable absorber. (h) Optical spectrum. (i) Intensity autocorrelation function [59]. (j) General schematic of the all–PM fiber laser used in the experiments. (k) Measured optical spectra of the TDFL for different number of graphene layers in the saturable absorber. (l) Pulse durations [60]. (m) Schematic setup of the passively Q–switched Er:ZBLAN fiber laser based on the GSAM. (n) Optical spectrum. (o) Pulse trains for the 0.3 W and 0.8 W pump power [63].

Figure 3. The performance of passively modulated laser based on graphene. (a) Schematic of Er³⁺–doped fiber laser based on graphene–SA. (b) Spectrum. (c) Autocorrelation trace [26]. (d) Experimental setup of the mode–locked laser based on the graphene SAM. (e) Optical spectrum. (f) Autocorrelation trace of the mode–locked pulses [51]. (g) Experimental setup of the dissipative soliton Er–doped fiber laser using a graphene/PMMA saturable absorber. (h) Optical spectrum. (i) Intensity autocorrelation function [59]. (j) General schematic of the all–PM fiber laser used in the experiments. (k) Measured optical spectra of the TDFL for different number of graphene layers in the saturable absorber. (l) Pulse durations [60]. (m) Schematic setup of the passively Q–switched Er:ZBLAN fiber laser based on the GSAM. (n) Optical spectrum. (o) Pulse trains for the 0.3 W and 0.8 W pump power [63].

In 2016, Sobon et al. first demonstrated noise-like pulse generation with an intrinsic spectral bandwidth exceeding 60 nm from a graphene SA-driven TDF laser, expanding the applications of graphene-based lasers in the supercontinuum generation area [64]. The largest single-pulse energy reaches 51.5 nJ with an average power of 1.21 W after amplification in an all-fiber system. Chen et al. used a polymer-free graphene oxide film to achieve high-power passive ML pulses of 6.62 ps with a maximum average output power of 3.89 W [65]. The highest pulse peak power reaches 4.85 kW and the pulse energy is 32.14 nJ.

In the same year, a multilayer graphene was employed in Er³⁺-doped ZBLAN fiber laser to obtain 42 ps pulses at 2.78 µm [66]. In 2018, Pawliszewska et al. used graphene SA in a Ho³⁺-doped fiber laser to achieve stretched-pulse output of 190 fs at 2060 nm by implementing dispersion management [67]. To further broaden the operating wavelength range of graphene, A. V. Pushkin et al. reported a passive ML Fe:ZnSe laser implemented using graphene SA for the first time in 2020. The laser operates at 4.4 µm with a pulse duration of about 732 fs and a repetition frequency of 100 MHz [68].

In 2021, graphene was employed in a synchronized ML fiber laser consisting of both erbium and thulium-doped fiber laser, generating 700 fs ultrashort pulses at 1563.5 nm with 24 nJ pulse energy and 1.77 ps ultrashort pulses at 1931.9 nm with 26 nJ pulse energy. The success of this work provide a better insight in generating multiple ML lasers with different wavelength [69]. In 2022, Jin et al. proposed a low-temperature plasma-enhanced CVD to directly deposit a high-quality, transfer-free graphene SA mirror, achieving stable passive QS in an Er³⁺-doped ZBLAN fiber laser at 2.8 μm with the pulse duration of 300.2 ns (refer to Figure 3m–o) and a slope efficiency of 17.4%. The average output power, pulse energy and peak power were 142 mW, 2.33 μJ and 7.76 W [63]. This approach resolved the key challenges of impurity introduction and poor modulation stability associated with conventional transfer fabrication, providing a novel and feasible approach for developing high-performance mid-infrared pulsed laser devices.

Table 1. Performance summary of passively modulated lasers based on graphene saturable absorbers. λ₀ is the center wavelength, τ_p is the pulse width, f_rep is the repetition frequency, E_p is the pulse energy, P_ave is the average output power.

SA	Gain Medium	λ0 (nm)	τp (ps)	frep (KHz)	Ep (nJ)	Pave (mW)	Thickness	References
Graphene	Nd3+: YAG	1064	4	88,000	-	-	-	[50]
Graphene	Nd3+: GdVO4	1065	16	43,000	8.4	360	few-layer	[70]
Graphene	Er3+ silica fiber	1545	0.088	21,150	0.071	1.5	20.4 nm	[61]
Graphene	Er3+ silica fiber	1555	0.029	18,670	2.8	52	10.2~11.9 nm	[62]
Graphene	Er3+ silica fiber	1558	0.605	20,800	0.013	0.2	few-layer	[71]
Graphene	Er3+ silica fiber	1560	0.98	28,500	0.0275	-	few-layer	[72]
Graphene	Er3+ silica fiber	1560	0.174	27,400	0.044	1.2	few-layer	[73]
Graphene	Er3+ silica fiber	1563	0.74	119,900	0.1119	-	-	[74]
Graphene	Er3+ silica fiber	1564.9	0.758	22,800	0.62	14.1	few-layer	[75]
Graphene	Er3+ silica fiber	1565	0.756	1790	-	2	few-layer	[26]
Graphene	Er3+ silica fiber	1565	13.8	16,990	10.2	174	few-layer	[59]
Graphene	Er3+ silica fiber	1565.3	0.148	101,320	0.025	2.5	12.6 nm	[76]
Graphene	Er3+ silica fiber	1566~1570	3.7	3.3~65.9	16.7	34	few-layer	[77]
Graphene	Er3+ silica fiber	1576.3	0.415	6840	7.3	-	few-layer	[78]
Graphene	Er3+/Tm3+ silica fiber	1563	0.7	12,905	24	46.1	0.34 nm	[69]
		1931	1.77	12,905	26	55.9
Graphene	Tm3+ silica fiber	1876	0.603	41,000	-	-	8.2 nm	[79]
Graphene	Tm3+/Ho3+ silica fiber	1879.4	4.7	7800	-	450	0.34 nm	[80]
Graphene	Tm3+ silica fiber	1900	1.9	19,700	-	1.96	0.34 nm	[81]
Graphene	Tm3+/Bi3+ silica fiber	1902	0.37	16,700	-	1.72	0.34 nm	[53]
Graphene	Tm3+ silica fiber	1910	0.773	19,310	6	115	0.34 nm	[82]
Graphene	Tm3+ silica fiber	1923.3	0.737	28,250	-	1.21	8.16 nm	[60]
Graphene	Tm3+ silica fiber	1940	3.6	6460	0.4	2	-	[52]
Graphene	Er3+/Tm3+ silica fiber	1565	0.933	20,190	-	0.5	0.68 nm	[55]
		1944	1.03	18,430	-	1.3
Graphene	Tm3+ silica fiber	1945	205	58,870	0.22	13	11.9 nm	[83]
Graphene	Tm3+ silica fiber	1950	0.255	23,500	51.5	1210	20.4 nm	[64]
Graphene	Tm3+ silica fiber	1953.3	2.1	16,937	0.08	1.41	1.36 nm	[54]
Graphene	Tm3+ silica fiber	1983	1.88	11,350	0.3	3.43	few-layer	[84]
Graphene	Tm3+ silica fiber	2005	3.2 × 105	280	18,000	5200	0.34 nm	[58]
Graphene	Tm3+: CLNGG	2018	0.729	98,700	-	60.2	0.34~0.68 nm	[51]
Graphene	Ho3+: YVO4	2052.1	2.65 × 105	131.6	16,800	2200	1.02 nm	[85]
Graphene	Ho3+ silica fiber	2060	0.19	21,130	2.55	54	-	[67]
Graphene	Cr3+: ZnSe	2500	0.226	77,000	-	80	0.34 nm	[56]
Graphene	Er3+: ZBLAN	2783	2.9 × 106	37	1670	62	few-layer	[57]
Graphene	Er3+: ZBLAN	2784.5	42	25,400	0.7	18	1.36~2.04 nm	[66]
Graphene	Er3+: ZBLAN	2795.2	3 × 105	61	-	142	few-layer	[63]
Graphene	Er3+: YAP	2918	4.6 × 105	114	5100	1170	0.34 nm	[86]
Graphene	Fe3+: ZnSe	4410	0.732	100,000	-	415	0.34 nm	[68]

summarizes the achievements of passively modulated lasers based on graphene saturable absorbers. To date, graphene-based passively modulated lasers have covered the primary laser wavelength bands ranging from 1 μm to 4.4 μm (see in Table 1), establishing a comprehensive suite of technical solutions that includes soliton mode locked, stretched pulses, noise-like pulses, QS and multi-wavelength synchronous ML, demonstrating significant application value across various fields of scientific research.

3.2. Transition Metal Dichalcogenides

TMDs constitute a category of two-dimensional layered compounds with the chemical formula MX₂, in which M denotes a transition metal element in groups 4–10 (e.g., Mo, W, V, as shown in Figure 4a) and X represents a chalcogen element (e.g., S, Se, Te). The X atoms and M atoms are strongly connected by covalent bonds within the layers, while the bonds between adjacent layers are formed through weak van der Waals force [87]. Unlike gapless graphene, TMDs exhibit layer-dependent tunable bandgaps. Figure 4b–e illustrate the band structures of molybdenum disulfide (MoS₂) with different numbers of layers [88]. It transitions from an indirect bandgap in bulk form (~1.2 eV) to a direct bandgap (~1.8 eV) in the monolayer. In 2013, Wang et al., measured the saturable absorption of few-layer MoX₂ (X = S, Se and Te) nanosheets by Z-scan system at 400 nm, 515 nm, 800 nm, and 1030 nm [28,89]. The results are presented in Figure 4f–i. At relatively short visible wavelengths (e.g., 400 nm, 515 nm, 800 nm), few-layer MoX₂ (X = S, Se) can demonstrate a stronger nonlinear optical response than graphene under specific experimental conditions. However, in the near-infrared of 1030 nm, graphene generally maintains superior or comparable performance owing to its gapless Dirac band structure and ultrafast carrier dynamics.

Research on passively modulated lasers using a TMDs-based SA began in 2014. The key breakthroughs in the initial stage were concentrated on validating the saturable absorption characteristics of classical Group VI binary TMDs, optimizing fabrication techniques and integration strategies, and expanding the multi-band application boundaries. Zhang et al. synthesized few-layered MoS₂ nanoplatelets by the hydrothermal exfoliation method and deposited them onto the end facet of the optical fiber of a YDF laser at 1054.3 nm for the generation of 800 ps ML pulses [90]. A few-layer MoS₂ polymer composite SA was prepared and used in a passive QS YDF laser with over 100 nJ pulse energy and 40 nm of tunability, tunable from 1030 to 1070 nm. This highlights the potential of few-layer MoS₂ as a wideband saturable absorber [91]. A multilayer MoS₂ SA was fabricated via the CVD method and transferred onto the end-face of a fiber connector within an EDF laser. Leveraging its outstanding saturable absorption property, stable mode-locking with a pulse duration of 1.28 ps and a repetition rate of 8.288 MHz was achieved at a central wavelength of 1568.9 nm [92]. A MoS₂ sample with over 30 layers (featuring an absolute modulation depth of 27%) was fabricated via the pulsed laser deposition (PLD) technique and employed as SAs in passive QS SSLs at wavelengths of 1.06 μm (Nd:GdVO₄), 1.42 μm (Nd:YGG), and 2.1 μm (Tm:Ho:YGG) [93]. In addition, several typical integration methods, such as direct deposition onto microfibers [94], side-polished fibers [95], and also deposition onto quartz/BK7 glass [96], have been reported and used to QS and ML both bulk and fiber lasers from 1030 to 2100 nm, enabling laser pulse generation at kHz to GHz repetition rates and with few microsecond to sub-picosecond pulse durations.

In 2015, Chen et al. systematically compared the nonlinear optical parameters and passive QS performance of four representative Group VI binary TMD materials (e.g., MoS₂, MoSe₂, WS₂, and WSe₂) [97]. The experimental results revealed that MoSe₂ exhibited the highest modulation depth (6.73%), while WS₂ demonstrated the best long-term operational stability. This study presented systematic experimental evidence for material selection and performance optimization within the TMD family.

In the same year, benefiting from their unique electronic structures, strong light–matter interactions, and excellent nonlinear optical absorption properties, TMDs became a research focus in the field of mid-infrared SAs. Tian et al. first integrated a multilayer MoS₂ SA into a 2 μm TDF laser and successfully realized stable ML pulses at a center wavelength of 1905 nm, with a pulse width of 843 ps and a signal-to-noise ratio greater than 55 dB [98]. The cavity schematic, output spectrum and autocorrelation trace are shown in Figure 5a–c. Jung et al. demonstrate the use of an all-fiberized, thulium–holmium co-doped fiber mode-locked laser based on a WS₂-deposited side-polished fiber. The stable ML pulses with a temporal width of 1.3 ps at a repetition rate of 34.8 MHz were obtained at 1941 nm [99]. Subsequently, Fan et al. realized a passive QS operation in an Er:Lu₂O₃ laser at 2.84 μm by employing a MoS₂ SA. Under the laser pump power of 7.61 W, an average output power of 1.03 W was generated with a pulse duration of 335 ns and a repetition rate of 121 kHz, resulting in a pulse energy of 8.5 μJ [100]. These works provided the experimental confirmation of the broadband modulation capability of TMD materials in the mid-infrared spectral range, laying a core experimental foundation for the application of TMDs in ultrafast mid-infrared lasers.

In the subsequent two years, the device structures based on TMDs for passive ML lasers were further optimized. In 2017, WS₂ was deposited on the surface of a tapered fiber to form the evanescent field using the pulsed laser deposition (PLD) method [101]. A fiber-taper WS₂ SA with a large modulation depth of 35.1% was fabricated to support ultrashort pulse generation. The shortest pulse duration of 67 fs was obtained, with a 114 nm spectral width and a 93 dB signal-to-noise ratio (SNR). This indicates that the fiber-taper WS₂ SA with a large modulation depth is a promising photonic device for ML fiber lasers with a wide spectrum and ultrashort pulse duration. WSe₂ films that are both large-scale and high quality are prepared by CVD method and tightly transferred onto the side wall of a microfiber to form a hybrid structure. The integrated microfiber-WSe₂ device realized strong evanescent wave interaction and showed a large modulation depth of 54.5%. Stable soliton ML pulses are generated and the pulse durations of 477 fs (at 1.5 μm) and 1.18 ps (at 2.0 μm) are demonstrated, which suggests that the large-area and highly crystalline WSe₂ films afford an excellent broadband SA for ultrafast photonic applications [102]. At the same time, a type of microfiber-based WTe₂ SA fabricated by the magnetron-sputtering deposition (MSD) method was reported, which had a large modulation depth of 34.3%. High-performance SAs with large modulation depths are beneficial for high-energy wave-breaking free pulse generation. Consequently, pulses were generated with a pulse duration, pulse energy, and average output power of 229 fs, 2.14 nJ, and 57 mW in the 1.5 μm regime, and 1.3 ps, 13.8 nJ, and 212 mW in the 2 μm regime, respectively [103,104]. In 2019, Lv et al. reported a novel type of MoS₂-doped sol–gel glass SA with a high optical damage threshold, which was fabricated using the sol–gel technique. Stable mode-locked pulse trains were successfully generated in the normal dispersion regime, with a pulse width of 13.8 ps and an average output power of 34.6 mW. It potentially provides a new approach to improving the optical damage threshold and long-term working stability of broadband SAs [105].

The studies since 2019 were not merely restricted to classical Group VI binary TMDs but rather involved a comprehensive expansion of the Group 4–10 TMDs material system (e.g., ZrSe₂ [106], ZrTe₂ [107], HfS₂ [108], HfSe₂ [109], VSe₂ [110], VTe₂ [111], TaTe₂ [112], ReSe₂ [113], PdSe₂ [114], PdTe₂ [115]) and ternary TMD alloy systems (e.g., MoWSe₂ [116,117]). The emphasis lies on the advancements in ultrashort pulse width, ultra-high repetition rate, and high signal-to-noise ratio laser output. Simultaneously, it entails accomplishing the customization of materials and devices for specific application scenarios to propel TMD-based lasers from laboratory proof-of-concept demonstrations to practical applications. Table 2 summarizes the achievements of passively modulated lasers based on TMDs SAs.

In 2019, a MoWSe₂ ternary TMD alloy SA was developed, and its broadband modulation capability across the near-infrared to mid-infrared spectral range was first verified [116]. The continuous adjustment of the band structure and nonlinear optical parameters via bimetallic alloying paves a new technological path for the performance optimization of TMDs. In 2022, Ahmad et al. used a HfSe₂ SA through evanescent-field coupling to obtain an ultrashort ML pulse of only 0.75 ps at 1.5 μm and of 1.15 ps at 2 μm, with signal-to-noise ratios exceeding 60 dB in both wavelength bands [118]. The cavity schematic, output spectrum and autocorrelation trace are shown in Figure 5d–i. This work confirmed that Group IV TMDs possess higher carrier mobilities and superior nonlinear modulation performance compared with traditional Group VI materials, substantially broadening the material research boundaries of TMDs.

In 2024, Chen et al. systematically compared the ML laser performance of MoS₂ and MoSe₂ [119]. As shown in Figure 5j–n, MoS₂ exhibits a relatively shorter carrier recovery time (164.6 ps), thereby enabling the generation of narrower ultrashort pulses with a width of merely 470 fs. This characteristic renders it more conducive to applications in ultrashort pulse lasers. MoSe₂ demonstrates lower non-saturable loss and attains higher laser slope efficiency, rendering it more suitable for high-power, high-efficiency laser applications. This study offers accurate experimental and theoretical guidance for the selection of TMDs in diverse application scenarios.

In 2025, Wang et al. demonstrated stable operation from fundamental ML to up to 86th-order harmonic ML in an EDF laser based on a MoWSe₂ ternary TMD SA [120]. An ultrahigh repetition-rate pulse reaching up to 374.8 MHz was successfully attained, which demonstrated the substantial application potential of TMDs in high-repetition-rate ultrafast lasers, high-speed optical communications, and ultra-precision optical metrology.

In the same year, Liu et al. put forward a preparation procedure for Atomic Oxygen- Passivated 2D Zr/Ta Telluride crystal SAM intended for high-power mid-infrared lasers. This methodology entails the direct in situ growth of large-area, high-quality 2D ZrTe₃ and TaTe₂ films on CaF₂ substrates via CVD technology. This method circumvents wet-coating procedures in various previously reported SAs, thereby alleviating the performance degradation induced by impurities. After oxygen plasma treatment, a dense and stable oxide protective layer was formed on the material surface, significantly enhancing the nonlinear optical performance and environmental robustness of SA [112]. The schematic diagram of the cavity, the output spectrum, and the pulse envelope are presented in Figure 5o–q. This study provides a new technical pathway for developing highly stable 2D material SAs for high-power mid-infrared pulsed lasers. Table 2 summarizes the achievements of passively modulated lasers based on TMDs SAs.

Figure 5. The performance of passively modulated laser based on TMDs. (a) Experimental setup of the mode–locked Tm³⁺ fiber laser. (b) Output of the mode–locked Tm³⁺ fiber laser. Inset shows the laser spectrum of the mode–locked Tm³⁺ fiber laser. (c) Single pulse of the MoS₂ [98]. (d) Configuration of the EDFL cavity. (e,f) The output characteristics of the mode–locked EDFL. (g) Configuration of the THDFL cavity. (h,i) The output characteristics of the mode–locked THDFL [118]. (j) Experimental setup of an Er–doped fiber laser. (k) Optical spectrum. (l) Autocorrelation trace based on the MoS₂ SA. (m) Optical spectrum. (n) Autocorrelation trace based on the MoSe₂ SA [119]. (o) Schematic setup of the passively Q–switched Er³⁺–doped ZBLAN fiber laser incorporating TMT–SAs. (p) Optical spectrum. (q) Single pulse profiles of TaTe₂–SA and ZrTe₃–SA [112].

Figure 5. The performance of passively modulated laser based on TMDs. (a) Experimental setup of the mode–locked Tm³⁺ fiber laser. (b) Output of the mode–locked Tm³⁺ fiber laser. Inset shows the laser spectrum of the mode–locked Tm³⁺ fiber laser. (c) Single pulse of the MoS₂ [98]. (d) Configuration of the EDFL cavity. (e,f) The output characteristics of the mode–locked EDFL. (g) Configuration of the THDFL cavity. (h,i) The output characteristics of the mode–locked THDFL [118]. (j) Experimental setup of an Er–doped fiber laser. (k) Optical spectrum. (l) Autocorrelation trace based on the MoS₂ SA. (m) Optical spectrum. (n) Autocorrelation trace based on the MoSe₂ SA [119]. (o) Schematic setup of the passively Q–switched Er³⁺–doped ZBLAN fiber laser incorporating TMT–SAs. (p) Optical spectrum. (q) Single pulse profiles of TaTe₂–SA and ZrTe₃–SA [112].

Table 2. Performance summary of passively modulated lasers based on TMDs absorbers. λ₀ is the center wavelength, τ_p is the pulse width, f_rep is the repetition frequency, E_p is the pulse energy, P_ave is the average output power.

SA	Gain Medium	λ0 (nm)	τp (ps)	frep (KHz)	Ep (nJ)	Pave (mW)	Thickness	References
MoS2	Yb3+ silica fiber	1029.8	13.8	22,440	-	34.6	few-layer	[105]
MoS2	Yb3+ silica fiber	1042.6	656	6740	-	25	0.6~2.1 nm	[94]
MoS2	Nd3+: GdVO4	1060	9.7 × 105	732	310	227	few-layer	[93]
	Nd3+: YGG	1420	7.29 × 105	77	670	52
	Tm3+: Ho3+: YGG	2100	4.1 × 105	149	1380	206
MoS2	Yb3+ silica fiber	1070	2.88 × 106	74	100	9.36	2~4 nm	[91]
MoS2	Er3+ silica fiber	1568.9	1.28	8288	-	-	3.42 nm	[92]
MoS2	Tm3+ silica fiber	1905	843	9670	15.5	150	4 layers	[98]
MoS2	Tm3+ silica fiber	1927	1.51	13,900	-	5	1.86 nm	[121]
MoS2	Tm3+/Ho3+ silica fiber	1979	1.97	9120	2.2	20	5~6 layers	[122]
MoS2	Er3+: ZBLAN	2754	8.06 × 105	70	2000	140	4 layers	[123]
MoS2	Er3+: Lu2O3	2840	3.35 × 105	121	8500	1030	4~10 layers	[100]
MoS2	Er3+ silica fiber	1560	1.35 × 107	41.452	184.7	0.77	1~5 layers	[97]
MoSe2		1559	6.5 × 106	66.847	369.5	2.45
WS2		1562	6.7 × 106	77.925	1179.4	6.41
WSe2		1561	9.18 × 106	85.365	484.8	3.16
MoS2	Er3+ silica fiber	1569.4	0.47	22,380	-	5.63	31 nm	[119]
MoSe2		1568	0.639	22,310	-	7.1	25 nm
MoSe2	Nd3+: GdVO4	1340	4.2 × 105	238	227	52.6	10 layers	[124]
MoSe2	Tm3+/Ho3+ silica fiber	1912	0.92	18,210	-	-	52 nm	[125]
MoSe2	Tm3+ silica fiber	1945.35	0.98	23,530	0.39	8.2	-	[126]
MoSe2	Tm3+/Ho3+ silica fiber	1982	4300/3800	2200	2.31	5.1	7 nm	[127]
VSe2	Tm3+/Ho3+ silica fiber	1912	1.4	11,600	-	0.8	-	[110]
WTe2	Tm3+/Ho3+ silica fiber	1909.8	1.77	11,540	0.304	3.51	6 layers	[128]
WTe2	Tm3+ silica fiber	1915.5	1.25	18,720	2.13	39.9	290 nm	[103]
WTe2	Tm3+ silica fiber	1951.5	1.31	13,070	8.26	108.1	166 nm	[129]
WTe2	Er3+ silica fiber	1560.5	1.77 × 106	55.56	18.09	1.01	213.6 nm	[130]
WS2	Er3+ silica fiber	1540	0.067	135,000	-	-	-	[101]
WS2	Tm3+/Ho3+ silica fiber	1910	0.825	15,490	0.0182	2.82	6 nm	[131]
WS2	Tm3+/Ho3+ silica fiber	1925	1.3	34,800	0.017	0.6	6 nm	[99]
WS2	Tm3+: LuAG	2012.9	6.6 × 105	63	17,000	1080	4 nm	[132]
WS2	Er3+: Y2O3	2716.3	7.2 × 105	29.4	7920	233.5	5.52 nm	[7]
WSe2	Er3+ silica fiber	1552.6	0.39	24,500	0.042	1.03	0.8 nm	[133]
WSe2	Er3+ silica fiber	1555	0.7	23,950	0.21	5.28	-	[134]
WSe2	Er3+ silica fiber	1556.4	0.477	14,020	-	-	1~3 layers	[102]
	Tm3+ silica fiber	1886.2	1.18	11,,360	-	32.5
WSe2	Er3+: CaF2	2756.8	8.4 × 105	46	2750	104	0.73 nm	[135]
ReSe2	Er3+ silica fiber	1559.8	0.992	28,179	0.02	0.58	-	[113]
MoWSe2	Er3+ silica fiber	1531.8	-	4360	0.109	0.47	1.8 nm	[120]
MoWSe2	Er3+ silica fiber	1554	1.9 × 106	48	11.8	1.1	few-layer	[116]
	Tm3+ silica fiber	1964	2.4 × 106	61.5	85.3	5.2
TiS2	Er3+ silica fiber	1569.5	1.04	5340	5.05	-	9 nm	[136]
PtS2	Er3+ silica fiber	1530	1.17	5120	0.368	1.86	-	[137]
PtSe2	Er3+: ZrF4	2783.2	1.04 × 106	93.1	10,020	932.7	61.9~101.2 nm	[138]
NiS2	Er3+ silica fiber	1560.2	0.524	21,100	1.86	39.2	3~6 layers	[139]
NiTe2	Er3+ silica fiber	1567.3	1.07	22,450	0.176	3.96	5 nm	[140]
	Tm3+ silica fiber	1966	3.38	20,660	0.415	8.59
NbS2	Er3+ silica fiber	1565.5	0.753	22,730	-	1.45	10 nm	[141]
	Tm3+ silica fiber	1961.4	5.77	20,230	0.275	5.56
HfS2	Tm3+/Ho3+ silica fiber	1916.3	1.509	11,100	0.05	0.56	few-layer	[108]
HfSe2	Er3+ silica fiber	1557	0.75	14,800	0.139	2.1	few-layer	[118]
	Tm3+/Ho3+ silica fiber	1912	1.15	12,500	0.108	1.36
HfSe2	Tm3+ silica fiber	1963.4	1.53	8910	0.252	2.25	65 nm	[109]
VTe2	Tm3+: YLF	1907.1	5.63 × 105	62.5	3790	237	25 nm	[111]
	Er3+: ZBLAN	2795.9	7.49 × 105	129.8	3100	402.8
MoTe2	Er3+ silica fiber	1559.5	0.229	26,601	2.14	57	11.6 nm	[104]
	Tm3+ silica fiber	1934.8	1.3	15,370	13.8	212
MoTe2	Tm3+ silica fiber	1930	0.952	14,353	2.56	36.7	6.8 nm	[142]
1T′-MoTe2	Er3+ silica fiber	1595	1.3 × 106	135	-	8.45	6.1 nm	[143]
2H-MoTe2			0.537	32,800	-	2.5
Mo0.5W0.5S2	Tm3+: YAP	1936	6.33 × 105	146.82	21,800	3255	5.3 nm	[117]
PdTe2	Er3+ silica fiber	1559.6	0.57	12,070	0.043	0.524	13.1 nm	[115]
	Tm3+ silica fiber	1896.4	1.59	14,950	0.63	9.39
PdSe2	Er3+ silica fiber	1557	1.31	12,560	-	1.8	10 nm	[114]
ZrSe2	Er3+ silica fiber	1561.8	12.5	21,220	0.54	11.37	few-layer	[106]
ZrTe2	Er3+ silica fiber	1560.7	37,500	1220	42.6	52	4.8 nm	[107]
TaTe2	Er3+: ZBLAN	2795	6.8 × 105	188.9	-	550	30 nm	[112]
ZrTe3			3.13 × 105	196.2	-	630

Table 2. Performance summary of passively modulated lasers based on TMDs absorbers. λ₀ is the center wavelength, τ_p is the pulse width, f_rep is the repetition frequency, E_p is the pulse energy, P_ave is the average output power.

SA	Gain Medium	λ0 (nm)	τp (ps)	frep (KHz)	Ep (nJ)	Pave (mW)	Thickness	References
MoS2	Yb3+ silica fiber	1029.8	13.8	22,440	-	34.6	few-layer	[105]
MoS2	Yb3+ silica fiber	1042.6	656	6740	-	25	0.6~2.1 nm	[94]
MoS2	Nd3+: GdVO4	1060	9.7 × 105	732	310	227	few-layer	[93]
	Nd3+: YGG	1420	7.29 × 105	77	670	52
	Tm3+: Ho3+: YGG	2100	4.1 × 105	149	1380	206
MoS2	Yb3+ silica fiber	1070	2.88 × 106	74	100	9.36	2~4 nm	[91]
MoS2	Er3+ silica fiber	1568.9	1.28	8288	-	-	3.42 nm	[92]
MoS2	Tm3+ silica fiber	1905	843	9670	15.5	150	4 layers	[98]
MoS2	Tm3+ silica fiber	1927	1.51	13,900	-	5	1.86 nm	[121]
MoS2	Tm3+/Ho3+ silica fiber	1979	1.97	9120	2.2	20	5~6 layers	[122]
MoS2	Er3+: ZBLAN	2754	8.06 × 105	70	2000	140	4 layers	[123]
MoS2	Er3+: Lu2O3	2840	3.35 × 105	121	8500	1030	4~10 layers	[100]
MoS2	Er3+ silica fiber	1560	1.35 × 107	41.452	184.7	0.77	1~5 layers	[97]
MoSe2		1559	6.5 × 106	66.847	369.5	2.45
WS2		1562	6.7 × 106	77.925	1179.4	6.41
WSe2		1561	9.18 × 106	85.365	484.8	3.16
MoS2	Er3+ silica fiber	1569.4	0.47	22,380	-	5.63	31 nm	[119]
MoSe2		1568	0.639	22,310	-	7.1	25 nm
MoSe2	Nd3+: GdVO4	1340	4.2 × 105	238	227	52.6	10 layers	[124]
MoSe2	Tm3+/Ho3+ silica fiber	1912	0.92	18,210	-	-	52 nm	[125]
MoSe2	Tm3+ silica fiber	1945.35	0.98	23,530	0.39	8.2	-	[126]
MoSe2	Tm3+/Ho3+ silica fiber	1982	4300/3800	2200	2.31	5.1	7 nm	[127]
VSe2	Tm3+/Ho3+ silica fiber	1912	1.4	11,600	-	0.8	-	[110]
WTe2	Tm3+/Ho3+ silica fiber	1909.8	1.77	11,540	0.304	3.51	6 layers	[128]
WTe2	Tm3+ silica fiber	1915.5	1.25	18,720	2.13	39.9	290 nm	[103]
WTe2	Tm3+ silica fiber	1951.5	1.31	13,070	8.26	108.1	166 nm	[129]
WTe2	Er3+ silica fiber	1560.5	1.77 × 106	55.56	18.09	1.01	213.6 nm	[130]
WS2	Er3+ silica fiber	1540	0.067	135,000	-	-	-	[101]
WS2	Tm3+/Ho3+ silica fiber	1910	0.825	15,490	0.0182	2.82	6 nm	[131]
WS2	Tm3+/Ho3+ silica fiber	1925	1.3	34,800	0.017	0.6	6 nm	[99]
WS2	Tm3+: LuAG	2012.9	6.6 × 105	63	17,000	1080	4 nm	[132]
WS2	Er3+: Y2O3	2716.3	7.2 × 105	29.4	7920	233.5	5.52 nm	[7]
WSe2	Er3+ silica fiber	1552.6	0.39	24,500	0.042	1.03	0.8 nm	[133]
WSe2	Er3+ silica fiber	1555	0.7	23,950	0.21	5.28	-	[134]
WSe2	Er3+ silica fiber	1556.4	0.477	14,020	-	-	1~3 layers	[102]
	Tm3+ silica fiber	1886.2	1.18	11,,360	-	32.5
WSe2	Er3+: CaF2	2756.8	8.4 × 105	46	2750	104	0.73 nm	[135]
ReSe2	Er3+ silica fiber	1559.8	0.992	28,179	0.02	0.58	-	[113]
MoWSe2	Er3+ silica fiber	1531.8	-	4360	0.109	0.47	1.8 nm	[120]
MoWSe2	Er3+ silica fiber	1554	1.9 × 106	48	11.8	1.1	few-layer	[116]
	Tm3+ silica fiber	1964	2.4 × 106	61.5	85.3	5.2
TiS2	Er3+ silica fiber	1569.5	1.04	5340	5.05	-	9 nm	[136]
PtS2	Er3+ silica fiber	1530	1.17	5120	0.368	1.86	-	[137]
PtSe2	Er3+: ZrF4	2783.2	1.04 × 106	93.1	10,020	932.7	61.9~101.2 nm	[138]
NiS2	Er3+ silica fiber	1560.2	0.524	21,100	1.86	39.2	3~6 layers	[139]
NiTe2	Er3+ silica fiber	1567.3	1.07	22,450	0.176	3.96	5 nm	[140]
	Tm3+ silica fiber	1966	3.38	20,660	0.415	8.59
NbS2	Er3+ silica fiber	1565.5	0.753	22,730	-	1.45	10 nm	[141]
	Tm3+ silica fiber	1961.4	5.77	20,230	0.275	5.56
HfS2	Tm3+/Ho3+ silica fiber	1916.3	1.509	11,100	0.05	0.56	few-layer	[108]
HfSe2	Er3+ silica fiber	1557	0.75	14,800	0.139	2.1	few-layer	[118]
	Tm3+/Ho3+ silica fiber	1912	1.15	12,500	0.108	1.36
HfSe2	Tm3+ silica fiber	1963.4	1.53	8910	0.252	2.25	65 nm	[109]
VTe2	Tm3+: YLF	1907.1	5.63 × 105	62.5	3790	237	25 nm	[111]
	Er3+: ZBLAN	2795.9	7.49 × 105	129.8	3100	402.8
MoTe2	Er3+ silica fiber	1559.5	0.229	26,601	2.14	57	11.6 nm	[104]
	Tm3+ silica fiber	1934.8	1.3	15,370	13.8	212
MoTe2	Tm3+ silica fiber	1930	0.952	14,353	2.56	36.7	6.8 nm	[142]
1T′-MoTe2	Er3+ silica fiber	1595	1.3 × 106	135	-	8.45	6.1 nm	[143]
2H-MoTe2			0.537	32,800	-	2.5
Mo0.5W0.5S2	Tm3+: YAP	1936	6.33 × 105	146.82	21,800	3255	5.3 nm	[117]
PdTe2	Er3+ silica fiber	1559.6	0.57	12,070	0.043	0.524	13.1 nm	[115]
	Tm3+ silica fiber	1896.4	1.59	14,950	0.63	9.39
PdSe2	Er3+ silica fiber	1557	1.31	12,560	-	1.8	10 nm	[114]
ZrSe2	Er3+ silica fiber	1561.8	12.5	21,220	0.54	11.37	few-layer	[106]
ZrTe2	Er3+ silica fiber	1560.7	37,500	1220	42.6	52	4.8 nm	[107]
TaTe2	Er3+: ZBLAN	2795	6.8 × 105	188.9	-	550	30 nm	[112]
ZrTe3			3.13 × 105	196.2	-	630

3.3. Black Phosphorus

2D BP is a layered crystalline material with a puckered honeycomb atomic structure, as shown in Figure 6a [40]. In 2014, the first application of few-layer BP with a thickness down to a few nanometers in field-effect transistors garnered substantial attention within the optoelectronic domain [44]. It was found that BP is a high-mobility semiconductor with a direct bandgap sensitivity depending on the number of layers from 0.3 (bulk) to 2.0 eV (single layer), as illustrated in Figure 6b. The narrow bandgap enables it to function as a SA, which can be employed in a wider wavelength operation range of 0.8~4 μm.

In 2015, Zhang et al. utilized the liquid-phase exfoliation technique to reduce the bulk BP to a thickness of roughly ten layers and experimentally verified its broadband and enhanced saturable absorption at wavelengths of 400~1930 nm for the first time [27]. As depicted in Figure 6c–f, the saturation modulation depths were experimentally determined to be 27.6% at 400 nm, 12.4% at 800 nm, 19.5% at 1563 nm, and 16.1% at 1930 nm, respectively. Additionally, the carrier dynamics in BP suspension were studied by pump-probe experiments at a wavelength of 1550 nm, and its ultrafast recovery time was observed (τs = 24 ± 2 fs) [144]. Compared with graphene, BP possesses a faster recovery time and a larger modulation depth.

In the same year, the BP SAs prepared by mechanical exfoliation or liquid-phase-exfoliated methods were applied in pulsed lasers operating in the near/mid-infrared spectral region. Chen et al. obtained either the passive QS (with maximum pulse energy of 94.3 nJ) or the passive ML operation (with pulse duration down to 946 fs) in a 1.5 μm EDF laser based on a few-layer BP [145]. The cavity schematic, output spectrum and autocorrelation trace are shown in Figure 7a–e. Sotor et al. applied a BP SA to a 2 μm TDF laser and obtained ML pulses with a width of 739 fs [146] (refer to Figure 7f–h). Qin et al. then used a multilayer BP (with a modulation depth of 15% and saturation fluence of 9 μJ/cm²) to achieve a passive QS pulse of 1.18 μs at 2.8 μm in an Er³⁺:ZBLAN fiber laser [147] (refer to Figure 7i–k), which makes it a potential broadband SA for pulsed lasers, especially in the mid-infrared spectral regime.

During this period, researchers also developed novel fabrication methods, such as the evanescent field optical deposition method (e.g., microfiber optical deposition [148] and the BP-polymer composite films method [149]), to achieve a long nonlinear interaction length and high integration encapsulation of BP with fiber devices. This systematically enhances the core nonlinear parameters of BP, including the thermal damage threshold, modulation depth, and saturation fluence. Broadband BP SAMs were successfully fabricated through deposition on a high-reflection or high-transmittance mirror and employed in passively modulated bulk lasers operating at 639 nm [150], 1 μm [151,152,153,154], 2 μm [150,155], and 2.4 μm [156]. This offers a potential approach for realizing ultrafast solid-state lasers within the visible to mid-infrared wavelength range.

Figure 7. The performance of passively modulated laser based on BP. (a) Experimental setup of the BP–SA based Q–switching Er³⁺–doped fiber laser. (b) Optical spectrum of QS. (c) Single pulse profile of QS. (d) Optical spectrum of ML. (e) Single pulse profile of ML [145]. (f) Experimental setup of the mode–locked Tm³⁺–doped fiber laser. (g) Optical spectrum. (h) Autocorrelation trace [146]. (i) The schematic of the passively Q–switched Er³⁺:ZBLAN fiber laser. (j) Optical spectrum. (k) Corresponding pulse profiles and the RF spectrum [147]. (l) Schematic diagram of the BP–SA based near zero–dispersion fiber laser. (m) Optical spectrum. (n) Autocorrelation trace [157].

Figure 7. The performance of passively modulated laser based on BP. (a) Experimental setup of the BP–SA based Q–switching Er³⁺–doped fiber laser. (b) Optical spectrum of QS. (c) Single pulse profile of QS. (d) Optical spectrum of ML. (e) Single pulse profile of ML [145]. (f) Experimental setup of the mode–locked Tm³⁺–doped fiber laser. (g) Optical spectrum. (h) Autocorrelation trace [146]. (i) The schematic of the passively Q–switched Er³⁺:ZBLAN fiber laser. (j) Optical spectrum. (k) Corresponding pulse profiles and the RF spectrum [147]. (l) Schematic diagram of the BP–SA based near zero–dispersion fiber laser. (m) Optical spectrum. (n) Autocorrelation trace [157].

Since 2016, research on SAs has focused on two fundamental directions: rectifying the inherent deficiencies of BP and improving its laser-modulated performance (e.g., ultra-short pulse, high peak power, expansion of the mid-infrared wavelength range, and multi-functionality). Su et al. demonstrated a BP-based mode-locked solid-state laser with pulses as short as 272 fs at a central wavelength of 1053.4 nm. The peak power of 23.8 MW is the highest among BP-based mode-locked lasers so far [153]. In 2018, Qin et al. demonstrated a mid-infrared BP SAM by transferring liquid-phase exfoliated BP flakes onto a gold-coated mirror and realized QS and ML operation in an Er:ZBLAN laser at 3.5 μm wavelength for the first time. The results of this research showed that BP has great potential as a mid-infrared SA beyond 3 μm wavelength [158].

Jin et al. achieved the shortest pulse duration of 102 fs by using a BP SA fabricated with a highly functional inkjet printing technology in an EDF laser ring cavity at 1555 nm [159]. Although the ultrafast laser utilizing BP as a SA has undergone substantial development due to its broadband absorption characteristics and ability to shorten pulse width, its vulnerability to environmental factors, such as humidity and temperature, has been gradually revealed through increasing research, thereby further restricting its application in high-power scenarios. In 2020, Zhang et al. encapsulated inkjet-printed BP films with a parylene-C passivation layer to achieve long-term environmental stability [157]. Through intracavity dispersion management, they obtained mode-locked pulses as short as 139 fs in a 2 μm TDF laser, fully demonstrating the key advantage of BP in ultrashort pulse generation. The cavity schematic, output spectrum, and autocorrelation trace are shown in Figure 7l–n. Furthermore, a variety of rich pulsed dynamical behaviors, including switchable dual-wavelength and harmonic mode-locking, have been sequentially demonstrated in BP-based lasers [160,161], facilitating their advancement towards multi-scenario practical applications. Table 3 summarizes the achievements of passively modulated lasers based on BP SAs. Overall, BP-based SAs have provided a new technical approach for the development of infrared ultrafast pulsed lasers due to their unique optoelectronic advantages. However, resolving their intrinsic defects through material encapsulation and modification, fabrication process optimization, and cavity design innovation remains the core research direction in this field for the future.

Table 3. Performance summary of passively modulated lasers based on BP absorbers. λ₀ is the center wavelength, τ_p is the pulse width, f_rep is the repetition frequency, E_p is the pulse energy, P_ave is the average output power.

SA	Gain Medium	λ0 (nm)	τp (ps)	frep (KHz)	Ep (nJ)	Pave (mW)	Thickness	References
BP	Yb3+: CYA	1046	6.2 × 105	113.6	-	37	3~9 nm	[151]
BP	Yb3+: Lu3+: CALGO	1053.4	0.272	23,800	6.48	23.8	5 nm	[153]
BP	Nd3+: YVO4	1064.1	6.1	140,000	3.29	460	5 nm	[154]
BP	Nd3+: GdVO4	1342	9.42	58,140	3	350	5 nm	[162]
BP	Er3+ silica fiber	1532~1570	0.94	4960	-	5.6	0.6~2 nm	[148]
BP	Er3+ silica fiber	1542.4	1 × 107	18.5	90	1.4	15~20 nm	[163]
		1543.2	1.2 × 107	9.4	135	1.35
BP	Er3+ silica fiber	1555	0.102	23,900	0.071	1.7	3.37	[159]
BP	Er3+ silica fiber	1558	2.18	15,590	-	0.077	10 nm	[164]
BP	Er3+ silica fiber	1559.5	0.67	8770	-	53	2.8~3.8 nm	[165]
BP	Er3+ silica fiber	1560.7	0.57	6880	0.74	5.1	200 nm	[166]
BP	Er3+ silica fiber	1561	2.66	1000	7.35	7.38	4~5 nm	[167]
BP	Er3+ silica fiber	1561~1564	2.98 × 106	30.098	283.91	8.545	few-layer	[168]
BP	Er3+ silica fiber	1563	1.32 × 107	10.42	58	0.6	25 layers	[145]
		1571	0.946	5960	-	-	15 layers
BP	Zr4+/Er3+ silica fiber	1602	3.46	1000	9.89	9.89	few-layer	[169]
BP	Er3+: YAG	1645	2.8 × 106	34	10,000	330	4~7 nm	[170]
BP	Tm3+ silica fiber	1859	0.139	20,950	0.97	20.4	3.4 nm	[157]
BP	Tm3+/Ho3+ silica fiber	1898	1.58	19,200	-	2.25	50 nm	[171]
BP	Tm3+ silica fiber	1910	0.739	36,800	0.04	1.5	300 nm	[146]
BP	Tm3+ silica fiber	1941	4.38 × 106	27.82	500	14	8.88 nm	[172]
BP	Tm3+ silica fiber	1948	5.6 × 106	28.1	130	3.8	-	[173]
BP	Tm3+: YAP	1969/1979	1.81 × 105	81	39,500	3100	10 nm	[160]
BP	Tm3+ silica fiber	2009	3.12 × 106	11.6	3320	38.5	-	[155]
BP	Ho3+ silica fiber	2094	1.3	29,100	0.379	11	16~25 nm	[161]
BP	Cr2+: ZnSe	2400	1.89 × 105	176	205	36	6~8 nm	[156]
BP	Er3+: ZBLAN	2779	1180	63	7600	485	5~20 nm	[147]
BP	Er3+: SrF2	2790	7.02 × 105	77.03	2340	180	3~9 nm	[174]
BP	Er3+: CaF2	2793.8	9.548 × 105	41.93	4250	-	3~9 nm	[175]

Table 3. Performance summary of passively modulated lasers based on BP absorbers. λ₀ is the center wavelength, τ_p is the pulse width, f_rep is the repetition frequency, E_p is the pulse energy, P_ave is the average output power.

SA	Gain Medium	λ0 (nm)	τp (ps)	frep (KHz)	Ep (nJ)	Pave (mW)	Thickness	References
BP	Yb3+: CYA	1046	6.2 × 105	113.6	-	37	3~9 nm	[151]
BP	Yb3+: Lu3+: CALGO	1053.4	0.272	23,800	6.48	23.8	5 nm	[153]
BP	Nd3+: YVO4	1064.1	6.1	140,000	3.29	460	5 nm	[154]
BP	Nd3+: GdVO4	1342	9.42	58,140	3	350	5 nm	[162]
BP	Er3+ silica fiber	1532~1570	0.94	4960	-	5.6	0.6~2 nm	[148]
BP	Er3+ silica fiber	1542.4	1 × 107	18.5	90	1.4	15~20 nm	[163]
		1543.2	1.2 × 107	9.4	135	1.35
BP	Er3+ silica fiber	1555	0.102	23,900	0.071	1.7	3.37	[159]
BP	Er3+ silica fiber	1558	2.18	15,590	-	0.077	10 nm	[164]
BP	Er3+ silica fiber	1559.5	0.67	8770	-	53	2.8~3.8 nm	[165]
BP	Er3+ silica fiber	1560.7	0.57	6880	0.74	5.1	200 nm	[166]
BP	Er3+ silica fiber	1561	2.66	1000	7.35	7.38	4~5 nm	[167]
BP	Er3+ silica fiber	1561~1564	2.98 × 106	30.098	283.91	8.545	few-layer	[168]
BP	Er3+ silica fiber	1563	1.32 × 107	10.42	58	0.6	25 layers	[145]
		1571	0.946	5960	-	-	15 layers
BP	Zr4+/Er3+ silica fiber	1602	3.46	1000	9.89	9.89	few-layer	[169]
BP	Er3+: YAG	1645	2.8 × 106	34	10,000	330	4~7 nm	[170]
BP	Tm3+ silica fiber	1859	0.139	20,950	0.97	20.4	3.4 nm	[157]
BP	Tm3+/Ho3+ silica fiber	1898	1.58	19,200	-	2.25	50 nm	[171]
BP	Tm3+ silica fiber	1910	0.739	36,800	0.04	1.5	300 nm	[146]
BP	Tm3+ silica fiber	1941	4.38 × 106	27.82	500	14	8.88 nm	[172]
BP	Tm3+ silica fiber	1948	5.6 × 106	28.1	130	3.8	-	[173]
BP	Tm3+: YAP	1969/1979	1.81 × 105	81	39,500	3100	10 nm	[160]
BP	Tm3+ silica fiber	2009	3.12 × 106	11.6	3320	38.5	-	[155]
BP	Ho3+ silica fiber	2094	1.3	29,100	0.379	11	16~25 nm	[161]
BP	Cr2+: ZnSe	2400	1.89 × 105	176	205	36	6~8 nm	[156]
BP	Er3+: ZBLAN	2779	1180	63	7600	485	5~20 nm	[147]
BP	Er3+: SrF2	2790	7.02 × 105	77.03	2340	180	3~9 nm	[174]
BP	Er3+: CaF2	2793.8	9.548 × 105	41.93	4250	-	3~9 nm	[175]

3.4. Other Low-Dimensional Materials

Beyond graphene, TMDs, and BP, several other LDMs have been widely applied in passively modulated pulsed lasers, including CNTs, TIs, TMOs, and MXenes. Table 4 summarizes the achievements of passive ML lasers based on other LDM SAs.

CNTs are 1D nanocrystalline graphite materials with a uniform cylindrical structure and a high aspect ratio. They can be categorized into multiple types according to the quantity of tube walls, specifically single-walled carbon nanotubes (SWCNTs), double-walled carbon nanotubes, and multi-walled carbon nanotubes (MWCNTs). As one of the earliest nanomaterials employed for passive ML, SWCNTs exhibit high third-order nonlinear polarizability and an ultrafast recovery time, which consists of a rapid intraband carrier relaxation time ranging from 0.3 to 1.2 ps and a slow recombination process lasting from 5 to 20 ps [176]. In 2004, CNT SA was first proposed by Set et al. for passively mode-locked laser [39]. In 2006, a SWCNTs–polyvinylalcohol nanocomposite was set in a micro gap between a pair of fiber end facets in EDF laser to achieve 178 fs ML pulses with a repetition rate of 22.8 MHz at 1.56 μm [177]. In 2008, a SWCNT polymer film as a SA was first employed in a 1.93 μm TDF laser to achieve ultrashort pulse of approximately 1.32 ps [178]. The cavity schematic, output spectrum and autocorrelation trace are shown in Figure 8a–c. In 2014, the shortest pulses of 90 fs, to our knowledge, were achieved in a diode-pumped Yb:CLNGG laser by applying SWCNTs SAs for the first time [179]. Moreover, the superior thermal conductivities of CNTs guarantee intrinsic high-power handling. In 2015, a high power Nd:YVO₄ ML laser operating at 1064 nm was demonstrated by using SWCNT SAs [180]. The maximum output power up to 2.7 W was obtained with a 167 MHz repetition rate and 3.1 ps pulse duration. The calculated pulse energy and peak power are 16.1 nJ and 5.2 kW. As of 2022, Wei et al. realized ML pulse output in a 3.5 μm Er-doped fluoride fiber laser, compressing the output pulse width to 1.66 ps at a repetition frequency of 25.2 MHz [181]. It conducted a comprehensive verification of the nonlinear optical modulation capability of CNTs in the mid-infrared band. The cavity schematic, output spectrum and autocorrelation trace are shown in Figure 8d–f. However, the difficulties in chirality and tube-diameter control of CNT restricts its application as SA at longer working wavelength.

TIs have emerged as another significant LDM subsequent to graphene, attributed to their distinctive Dirac surface states [182]. They act as insulators in their inner portion, but gapless conducting states appear on their surfaces. The combination of the small bandgap bulk (0.2~0.3 eV) and the gapless surface enable TIs to possess an ultra-broad bandwidth of saturable absorption operation. In addition, TIs demonstrate an ultrashort phonon-induced carrier lifetime on the order of several picoseconds, indicating their potential application in ultrafast ML lasers. TIs were used as SAs for demonstrating an ultrafast fiber laser in 2012. Zhao et al. reported a Bi₂Te₃ SA exhibiting very-high-modulation-depth (up to 95%) and used as a passive mode locker for ultrafast pulse formation at the telecommunication band [34]. In 2013, a Bi₂Se₃ SA was employed for the operation of a bulk solid-state laser [183]. In 2014, the operating wavelength of TIs was successfully extended to 2 μm. A Bi₂Te₃ layer with a thickness of approximately 30 μm on a side-polished optical fiber platform (with a modulation depth of approximately 20.6%) was employed to generate stable, ultrafast pulses with a temporal width of approximately 795 fs at a wavelength of 1935 nm from a thulium/holmium co-doped fiber ring cavity. This experimental demonstration confirms that TI SAs can readily be used as an ultrafast mode-locker for 2-μm lasers [184]. Yin et al. integrated Bi₂Te₃ nanosheets onto a micro-nanofiber through optical deposition and achieved soliton ML with a 1.26 ps ultrashort pulse in 2 μm thulium/holmium co-doped fiber laser [185]. The cavity schematic, output spectrum and autocorrelation trace are shown in Figure 8g–i. In 2016, a fiber-taper TI SA was fabricated using the PLD method for the first time, it modulated 70 fs pulses at 1542 nm by hybridizing nonlinear polarization evolution (NPE) [186]. In 2017, a Cd₃As₂ SA with a relaxation time of 0.5 ps (2% Cr doping concentration) was used for the generation of a stable ML operation at a center wavelength of 2860 nm. The study shows robust and effective tuning of the scattering channels of Dirac fermions, opening up the long-sought parameter space crucial for the development of compact and high-performance mid-infrared ultrafast sources [187]. In 2022, magnetron-sputtered GaSb thin film was used for pulse generation at 1559 nm, 1902 nm and 2783 nm [188]. These results indicate that TIs have great potential for nonlinear optical applications with a wide operating wavelength range.

In view of their substantial third-order optical nonlinearities, rapid recovery time on the picosecond scale, and adjustable band-gap energy, TMOs have emerged as a novel research avenue in the SA domain in recent years. In 2016, Fe₃O₄ nanoparticles (FONPs) exhibited a modulation depth of 8.2% and were employed in a stable passive QS EDF laser. An output pulse energy of 23.76 nJ, a repetition rate of 33.3 kHz, and a pulse width of 3.2 µs were achieved [189]. In 2017, Mohd Rusdi et al. initially utilized a titanium dioxide (TiO₂) film as a SA in a 1.97 μm Tm³⁺Ho³⁺co-doped fiber laser to achieve passive ML, and commenced research on TMOs for 2 μm ultrafast lasers [190]. In 2019, FONPs were deposited on an Au mirror and utilized for QS in a tunable mid-infrared fiber laser with a tunable range of 2812.4~3031.6 nm. This demonstration suggests that TMOs are a promising broadband saturable absorption material for mid-infrared operation [191]. In 2022, Ahmed et al. coated micrometer-scale molybdenum trioxide (MoO₃) particles compounded with a polymer onto a micro-nanofiber and utilized the evanescent field effect to augment the light–matter interaction [192]. They successfully achieved stable ML in a 1.94 μm TDF laser, resulting in an ultrashort pulse output of 1.22 ps, as shown in Figure 8j–l. The device also demonstrated a high optical damage threshold surpassing 19.34 GW/cm², offering a novel material alternative for the practical realization of high-power ultrafast lasers.

MXenes, as a kind of newly emerging category of 2D transition metal carbides/nitrides, have exhibited significant potential in ultrafast photonics owing to their metallic conductivity, adjustable surface functional groups, and broadband nonlinear optical response. The general formula for MXene is M_n+1X_nT_x, where M refers to transition metals (Ti, Sc, Hf, Zr, Nb, V, Cr, Mo, Ta, etc.), X stands for C, N, or B, and T_x is an element from the 3rd or 4th main group (O, OH, F, etc.), where x denotes the number of terminal groups. MXenes exhibit strong saturable absorption behaviors, high carrier mobility and great bandgap tunability. Furthermore, the outstanding mechanical performance, thermoelectric properties, and environmental stability under humid conditions of MXenes materials render them particularly suitable as SAs for pulse modulation applications. The first study of the mode-locked fiber laser using MXenes as the SA was reported by Jhon et al. in 2017 [193]. The uniqueness of Ti₃C₂T_x lies in its notably small bandgap of less than 0.2 eV, which can be utilized for broadband pulsed laser generation across the visible to mid-infrared wavelengths (e.g., 607, 639, 721, 1066, and 1555 nm) [194,195]. In 2020, Wang et al. conducted a systematic investigation into the broadband nonlinear optical properties of Ti₃C₂T_x and applied it to ML and QS fiber lasers at 1.55 μm and 2.8 μm [196]. In 2021, Jhon et al. successfully generated a ML pulse with a pulse width of 897 fs in a 1.9 µm infrared fiber laser using a multilayer Ti₃C₂T_x SA [197]. The cavity schematic, output spectrum and autocorrelation trace are shown in Figure 8m–o. These studies comprehensively illustrate the unique advantages of MXenes as broadband SAs. Subsequently, a Nb₂C material, a member of the MXenes family, was measured to have an ultrafast relaxation time of 37.43 fs and a slow relaxation time of 0.57 ps [198]. The Nb₂C SA was coated onto a tapered fiber to facilitate the interaction of the evanescent field, and it generated a soliton pulse with a pulse duration of 0.77 ps at the central wavelength of 1559 nm [199]. The research findings offer a potential avenue for the development of a novel device using Nb₂C or other types of MXenes for high-performance ML lasers. As shown in Table 4, ultrafast lasers incorporating SAs based on various MXenes materials have been reported, and they demonstrate remarkable laser performance.

Figure 8. The performance of passively modulated laser based on CNTs, Tis, TMOs and MXenes. (a) Scheme of the ring–cavity Tm³⁺–doped fiber laser mode locked with a carbon nanotube absorber. (b) Optical spectrum. (c) Autocorrelation trace [178]. (d) Experimental setup of the 3.5 μm mode–locked fiber laser based on SWCNT. (e) Optical spectrum. (f) Pulse train [181]. (g) Experimental setup of the passively mode–locked THDFL based on Bi₂Te₃. (h) Optical spectrum. (i) Autocorrelation trace [185]. (j) Configuration of passively mode–locked TDFL incorporating MoO₃–SA in the ring cavity. (k) Optical spectrum. (l) Autocorrelation trace [192]. (m) The schematic of the ring–cavity Tm³⁺–Ho³⁺–codoped fiber laser system based on Ti₃C₂T_x. (n) Optical spectrum. (o) Autocorrelation trace [197].

Figure 8. The performance of passively modulated laser based on CNTs, Tis, TMOs and MXenes. (a) Scheme of the ring–cavity Tm³⁺–doped fiber laser mode locked with a carbon nanotube absorber. (b) Optical spectrum. (c) Autocorrelation trace [178]. (d) Experimental setup of the 3.5 μm mode–locked fiber laser based on SWCNT. (e) Optical spectrum. (f) Pulse train [181]. (g) Experimental setup of the passively mode–locked THDFL based on Bi₂Te₃. (h) Optical spectrum. (i) Autocorrelation trace [185]. (j) Configuration of passively mode–locked TDFL incorporating MoO₃–SA in the ring cavity. (k) Optical spectrum. (l) Autocorrelation trace [192]. (m) The schematic of the ring–cavity Tm³⁺–Ho³⁺–codoped fiber laser system based on Ti₃C₂T_x. (n) Optical spectrum. (o) Autocorrelation trace [197].

Table 4. Performance summary of passively modulated lasers based on other absorbers. λ₀ is the center wavelength, τ_p is the pulse width, f_rep is the repetition frequency, E_p is the pulse energy, P_ave is the average output power.

SA	Gain Medium	λ0 (nm)	τp (ps)	frep (KHz)	Ep (nJ)	Pave (mW)	Thickness	References
CNT	Yb3+: CLNGG	1049	0.09	83,000	-	90	single-wall	[179]
CNT	Nd3+: YVO4	1064	3.1	167,000	16.1	2700	single-wall	[180]
CNT	Er3+ silica fiber	1560	0.178	22,800	0.068	1.55	35 μm	[177]
CNT	Tm3+ silica fiber	1870	0.45	45,000	0.4	18	10 μm	[200]
CNT	Tm3+ silica fiber	1895	0.56	-	1.66	99.4	50 μm	[201]
CNT	Tm3+ silica fiber	1928	0.501	-	0.5	-	100 nm	[202]
CNT	Tm3+ silica fiber	1927	0.152	25,760	0.19	-	30 μm	[203]
CNT	Tm3+ silica fiber	1932	1.32	37,000	-	3.4	single-wall	[178]
CNT	Tm3+ silica fiber	1947	2.3	6710	0.45	3	single-wall	[204]
CNT	Tm3+ silica fiber	1941.6	3.7	17,640	-	-	single-wall	[205]
CNT	Tm3+ silica fiber	1950	0.972	21,050	0.109	2.3	single-wall	[206]
CNT	Tm3+/Ho3+ silica fiber	1860~2010	2.38~2.83	18,400	-	-	20 μm	[207]
CNT	Ho3+ silica fiber	2079	0.683	54,520	0.376	20.5	400 nm	[208]
CNT	Er3+: ZBLAN	3470	1.66	25,200	0.98	25	1 μm	[181]
Bi2Te3	Tm3+/Ho3+ silica fiber	1909.5	1.26	21,500	-	2.3	20 nm	[185]
Bi2Te3	Tm3+/Ho3+ silica fiber	1935	0.795	27,900	-	20	30 μm	[184]
Bi2Se3	Tm3+/Ho3+ silica fiber	1912	0.853	18,370	-	-	15 μm	[209]
Cd3As2	Ho3+/Pr3+: ZBLAN	2864.3	6.3	14,280	-	-	400 nm	[187]
CoSb3	Tm3+/Ho3+ silica fiber	1913	0.838	16,930	-	-	6 μm	[210]
Sb2Se3	Er3+ silica fiber	1560	2.1 × 106	62.5	2.104	1.15	3~4 nm	[211]
Sb2Se3	Er3+ silica fiber	1562	16,500	3540	0.9	3.1	25 μm	[212]
Sb2Se3	Er3+ silica fiber	1562.4	0.63	22,600	0.0156	-	4 nm	[213]
Sb2Se3	Tm3+ silica fiber	1961	0.89	22,360	-	-	125 nm	[214]
Sb2Te3	Er3+ silica fiber	1542	0.07	95,400	-	63	-	[186]
Sb2Te3	Er3+ silica fiber	1558	1.9	3750	-	-	80 nm	[215]
Sb2Te3	Er3+ silica fiber	1565.5	0.452	20,100	0.091	1.82	18 nm	[216]
Sb2Te3	Tm3+: YAG	2012.6	3.82 × 105	56.67	4800	272	70 nm	[217]
Sb2Te3	Ho3+ silica fiber	2081.9	1.85	13,410	-	0.77	-	[218]
Bi2Te3	Er3+ silica fiber	1560.8	0.286	18,550	0.027	0.5	670 nm	[219]
Bi2Se3	Er3+ silica fiber	1564.6	1.57	1210	-	-	-	[220]
GaSb	Er3+ silica fiber	1559	0.308	15,280	0.98	17.82	103 nm	[187]
	Tm3+ silica fiber	1902	0.585	19,720	1.37	64.69
	Er3+: ZBLAN	2783	4.99 × 105	137	3070	420
Te	Er3+: ZBLAN	2782.3	4.57 × 105	116.98	3050	357	40.7 nm	[221]
TiO2	Tm3+/Ho3+ silica fiber	1979	10.29	9000	1.66	-	30 μm	[189]
TiO2	Tm3+ silica fiber	1925.6	1.5	11,530	11	126.8	-	[222]
Fe3O4	Er3+ silica fiber	1560	3.2 × 106	33.3	24	0.8	20 μm	[188]
Fe3O4	Dy3+: ZBLAN	2931.2	1.25 × 106	123	900	111	-	[190]
Co3O4	Tm3+ silica fiber	1958	1.39	11,360	0.023	0.26	-	[223]
ZnO	Tm3+ silica fiber	1945	1.395	11,360	0.056	0.64	45 μm	[224]
MoOx	Tm3+/Ho3+ silica fiber	1937.1	1.64	11,500	-	1.3	-	[225]
WO3	Tm3+ silica fiber	1941.4	1.22	10,840	0.68	-	-	[191]
BiVO4	Er3+ silica fiber	1563.3	2.67 × 106	76.28	330.76	25.23	-	[226]
Ti3C2Tx	Yb3+ silica fiber	1065.9	480	18,960	-	-	-	[193]
	Er3+ silica fiber	1555	0.159	7280	-	-
Ti3C2Tx	Tm3+/Ho3+ silica fiber	1914	0.897	16,770	-	-	80~800 nm	[196]
Ti3C2Tx	Tm3+ silica fiber	1891.8	2.18	5970	-	57.6	3~12 nm	[195]
	Er3+: ZBLAN	2788	6.4 × 105	122.9	1850	227
Ti2C	Tm3+/Ho3+ silica fiber	1933.8	1.655	11,560	-	50.9	-	[227]
Ti2AlC		1894.5	1.382	12,400	-	20.4	5 μm
Ti3CN	Er3+ silica fiber	1557	0.66	15,400	-	0.05	4~6 μm	[192]
Nb2C	Er3+ silica fiber	1559	0.77	14,120	0.213	3.04	-	[198]
V2C	Tm3+/Ho3+ silica fiber	1937	1.68	11,520	0.235	-	-	[228]
V2C	Tm3+/Ho3+ silica fiber	1900	0.843	18,290	-	-	11 nm	[31]
Ni-MOF	Er3+ silica fiber	1563	0.384	17,000	-	-	4.2 nm	[35]
	Tm3+ silica fiber	1882	1.3	13,900	-	-

Table 4. Performance summary of passively modulated lasers based on other absorbers. λ₀ is the center wavelength, τ_p is the pulse width, f_rep is the repetition frequency, E_p is the pulse energy, P_ave is the average output power.

SA	Gain Medium	λ0 (nm)	τp (ps)	frep (KHz)	Ep (nJ)	Pave (mW)	Thickness	References
CNT	Yb3+: CLNGG	1049	0.09	83,000	-	90	single-wall	[179]
CNT	Nd3+: YVO4	1064	3.1	167,000	16.1	2700	single-wall	[180]
CNT	Er3+ silica fiber	1560	0.178	22,800	0.068	1.55	35 μm	[177]
CNT	Tm3+ silica fiber	1870	0.45	45,000	0.4	18	10 μm	[200]
CNT	Tm3+ silica fiber	1895	0.56	-	1.66	99.4	50 μm	[201]
CNT	Tm3+ silica fiber	1928	0.501	-	0.5	-	100 nm	[202]
CNT	Tm3+ silica fiber	1927	0.152	25,760	0.19	-	30 μm	[203]
CNT	Tm3+ silica fiber	1932	1.32	37,000	-	3.4	single-wall	[178]
CNT	Tm3+ silica fiber	1947	2.3	6710	0.45	3	single-wall	[204]
CNT	Tm3+ silica fiber	1941.6	3.7	17,640	-	-	single-wall	[205]
CNT	Tm3+ silica fiber	1950	0.972	21,050	0.109	2.3	single-wall	[206]
CNT	Tm3+/Ho3+ silica fiber	1860~2010	2.38~2.83	18,400	-	-	20 μm	[207]
CNT	Ho3+ silica fiber	2079	0.683	54,520	0.376	20.5	400 nm	[208]
CNT	Er3+: ZBLAN	3470	1.66	25,200	0.98	25	1 μm	[181]
Bi2Te3	Tm3+/Ho3+ silica fiber	1909.5	1.26	21,500	-	2.3	20 nm	[185]
Bi2Te3	Tm3+/Ho3+ silica fiber	1935	0.795	27,900	-	20	30 μm	[184]
Bi2Se3	Tm3+/Ho3+ silica fiber	1912	0.853	18,370	-	-	15 μm	[209]
Cd3As2	Ho3+/Pr3+: ZBLAN	2864.3	6.3	14,280	-	-	400 nm	[187]
CoSb3	Tm3+/Ho3+ silica fiber	1913	0.838	16,930	-	-	6 μm	[210]
Sb2Se3	Er3+ silica fiber	1560	2.1 × 106	62.5	2.104	1.15	3~4 nm	[211]
Sb2Se3	Er3+ silica fiber	1562	16,500	3540	0.9	3.1	25 μm	[212]
Sb2Se3	Er3+ silica fiber	1562.4	0.63	22,600	0.0156	-	4 nm	[213]
Sb2Se3	Tm3+ silica fiber	1961	0.89	22,360	-	-	125 nm	[214]
Sb2Te3	Er3+ silica fiber	1542	0.07	95,400	-	63	-	[186]
Sb2Te3	Er3+ silica fiber	1558	1.9	3750	-	-	80 nm	[215]
Sb2Te3	Er3+ silica fiber	1565.5	0.452	20,100	0.091	1.82	18 nm	[216]
Sb2Te3	Tm3+: YAG	2012.6	3.82 × 105	56.67	4800	272	70 nm	[217]
Sb2Te3	Ho3+ silica fiber	2081.9	1.85	13,410	-	0.77	-	[218]
Bi2Te3	Er3+ silica fiber	1560.8	0.286	18,550	0.027	0.5	670 nm	[219]
Bi2Se3	Er3+ silica fiber	1564.6	1.57	1210	-	-	-	[220]
GaSb	Er3+ silica fiber	1559	0.308	15,280	0.98	17.82	103 nm	[187]
	Tm3+ silica fiber	1902	0.585	19,720	1.37	64.69
	Er3+: ZBLAN	2783	4.99 × 105	137	3070	420
Te	Er3+: ZBLAN	2782.3	4.57 × 105	116.98	3050	357	40.7 nm	[221]
TiO2	Tm3+/Ho3+ silica fiber	1979	10.29	9000	1.66	-	30 μm	[189]
TiO2	Tm3+ silica fiber	1925.6	1.5	11,530	11	126.8	-	[222]
Fe3O4	Er3+ silica fiber	1560	3.2 × 106	33.3	24	0.8	20 μm	[188]
Fe3O4	Dy3+: ZBLAN	2931.2	1.25 × 106	123	900	111	-	[190]
Co3O4	Tm3+ silica fiber	1958	1.39	11,360	0.023	0.26	-	[223]
ZnO	Tm3+ silica fiber	1945	1.395	11,360	0.056	0.64	45 μm	[224]
MoOx	Tm3+/Ho3+ silica fiber	1937.1	1.64	11,500	-	1.3	-	[225]
WO3	Tm3+ silica fiber	1941.4	1.22	10,840	0.68	-	-	[191]
BiVO4	Er3+ silica fiber	1563.3	2.67 × 106	76.28	330.76	25.23	-	[226]
Ti3C2Tx	Yb3+ silica fiber	1065.9	480	18,960	-	-	-	[193]
	Er3+ silica fiber	1555	0.159	7280	-	-
Ti3C2Tx	Tm3+/Ho3+ silica fiber	1914	0.897	16,770	-	-	80~800 nm	[196]
Ti3C2Tx	Tm3+ silica fiber	1891.8	2.18	5970	-	57.6	3~12 nm	[195]
	Er3+: ZBLAN	2788	6.4 × 105	122.9	1850	227
Ti2C	Tm3+/Ho3+ silica fiber	1933.8	1.655	11,560	-	50.9	-	[227]
Ti2AlC		1894.5	1.382	12,400	-	20.4	5 μm
Ti3CN	Er3+ silica fiber	1557	0.66	15,400	-	0.05	4~6 μm	[192]
Nb2C	Er3+ silica fiber	1559	0.77	14,120	0.213	3.04	-	[198]
V2C	Tm3+/Ho3+ silica fiber	1937	1.68	11,520	0.235	-	-	[228]
V2C	Tm3+/Ho3+ silica fiber	1900	0.843	18,290	-	-	11 nm	[31]
Ni-MOF	Er3+ silica fiber	1563	0.384	17,000	-	-	4.2 nm	[35]
	Tm3+ silica fiber	1882	1.3	13,900	-	-

3.5. Heterostructures

As the application of individual LDMs deepens, their respective limitations gradually become apparent. For instance, graphene exhibits an inadequate modulation depth, BP demonstrates poor stability, and TMDs have a relatively long recovery time. To overcome these limitations, researchers have proposed the construction of heterostructures through the vertical stacking of various 2D materials via van der Waals forces. They utilize the interlayer coupling effects of heterostructures to facilitate electron transfer and interband transitions, thereby achieving optical synergistic effects [229]. This approach offers a novel concept for material design in the development of high-performance ultrafast lasers.

Figure 9 illustrates the energy band diagrams and carrier transport schemes of several heterostructures. Figure 9a depicts the interlayer relaxation process of photogenerated carriers within a WS₂/MoS₂ Type-II heterostructure [230], suggesting that the Type-II band alignment facilitates the spatial segregation of photogenerated electrons and holes. This enables sub-picosecond ultrafast responses by modulating carrier relaxation dynamics to support femtosecond ultrashort pulse output. Meanwhile, it resolves the self-starting issue of passive ML by optimizing the modulation depth. Additionally, it suppresses non-saturable losses resulting from intralayer recombination to enhance optical conversion efficiency. Figure 9b presents the quantitatively calibrated band offset and bandgap parameters of the MoTe₂/MoS₂ heterostructure [231]. The absorption edge and nonlinear optical properties can be customized and controlled by the regulate of heterojunction. This facilitates the expansion of the working bandwidth of saturable absorption through the combination of materials with different bandgaps, which is appropriate for accommodating lasers of diverse wavelengths. Figure 9c depicts the interfacial band evolution and charge transfer process in the MoS₂/graphene contact prior to and subsequent to contact formation [232]. The semiconductor–semimetal heterostructures can address the intrinsic conflict between the “ultrafast response speed” and the “high damage threshold” in conventional SAs materials. Photogenerated carriers are capable of undergoing ultrafast transfer at the interface within a sub-hundred-femtosecond timescale. By integrating the ultrafast relaxation characteristic of graphene with the high optical damage threshold of semiconductor materials, it is possible to mitigate intracavity thermal effects, thereby significantly enhancing the output power ceiling and long-term operational stability of the laser. Figure 9d depicts the band structure and carrier transport behavior of a Bi₂Te₃/graphene heterostructure [233], indicating that the working wavelength of SAs can be further expanded to special bands. An ultra-broadband, low-loss saturable absorption SA was realized by utilizing the topological surface states of topological insulators and the Dirac-cone band matching with graphene. Consequently, it significantly enhances the wavelength tuning capability and broadens the application scenarios of passively modulated lasers.

Figure 9. Illustration of the energy band and carrier mobility diagrams of several heterostructures [234]. (a) MoS₂/WS₂ [230]. (b) MoTe₂/MoS₂ [231]. (c) MoS₂/graphene [232]. (d) Bi₂Te₃/graphene [233].

Figure 9. Illustration of the energy band and carrier mobility diagrams of several heterostructures [234]. (a) MoS₂/WS₂ [230]. (b) MoTe₂/MoS₂ [231]. (c) MoS₂/graphene [232]. (d) Bi₂Te₃/graphene [233].

During the initial exploration phase, research on heterostructure SAs primarily focused on graphene-TIs heterostructures. In 2015, Mu et al. first fabricated a graphene/Bi₂Te₃ heterostructure SA with controllable optical properties by two-step CVD method [235]. By regulating the coverage of Bi₂Te₃ nanoplatelets, it demonstrates a larger modulation depth (ranging from 14.7% to 50.1%) in comparison to pure graphene and faster carrier dynamics (spanning from 191 fs to 296 fs) in comparison to pure Bi₂Te₃. The laser cavity, output spectrum and pulse trace are shown in Figure 10a–c. The research delivers a new tunable photonics material which may find wide applications for pulse laser generation or signal processing.

Subsequently, graphene-based heterostructure SAs achieved comprehensive expansion in terms of material systems, modulation depth, operating wavelength, and stability. Gao et al. utilized a graphene/BN heterostructure SA in a 2 μm Tm³⁺:YAP passive QS laser. A minimum pulse width of 607 ns with an average output power of 3.67 W was obtained [236]. In comparison with single-layer graphene SA, the pulse width was narrower, the output power was higher, and the thermal stability was notably enhanced. This approach addressed the issues of inadequate modulation depth and severe heat accumulation intrinsic to single-layer graphene SAs. Shao et al. devised a MoS₂/BN/graphene/BN/MoS₂ SA and attained an ultrashort pulse output of 1.2 ps in a 1.5 μm all-fiber mode-locked laser [237]. The polarization tolerance of single-pulse ML was elevated from 20% (pure graphene) to 85%, significantly augmenting the operational stability of ML lasers. The schematic diagram of the cavity, the output spectrum, and the autocorrelation trace are presented in Figure 10d–f. Wang et al. fabricated a tellurium (Te)/graphene heterostructure SA and achieved stable ML (with a pulse width of 12.18 ps) in a 3 μm fiber laser [238]. This study demonstrated that heterojunctions are an effective method to improve nonlinearity, which is a promising approach to achieve high-efficiency pulsed laser.

Within non-graphene-based heterostructure systems, combinatorial innovations among TMDs, TMOs, MXenes, and ITs have emerged as research focal points. In 2018, it was confirmed that MoS₂-Sb₂Te₂-MoS₂ heterostructure materials exhibit a large modulation depth of 64.17% and can withstand high power during the generation of ultrashort pulses [239]. The new type of TMDs/TIs provides a promising solution for the generation of stable high-energy ultrashort pulses. In 2019, Liu et al. employed the magnetron sputtering deposition technique to fabricate a MoS₂/WS₂ heterostructure coated on the tapered fiber [240]. In comparison with monolayer MoS₂, this heterostructure demonstrates a smaller band gap and higher carrier mobility, which may lead to a greater modulation depth. Consequently, a generated pulse duration of 154 fs was achieved by utilizing the prepared SA in a EDF laser. Pang et al. in 2025 further developed a Ti₃C₂T_x/CuO heterostructure SA and demonstrated both conventional soliton output of 495 fs and dissipative soliton output of 22 ps in a 1.5 μm fiber laser [241], enriching the operational modes of MXene-based heterostructures. The SEM image of Ti₃C₂T_x/CuO composite, output spectrum and autocorrelation trace are shown in Figure 10g–i. Compared to single materials, heterojunctions based on TMOs exhibit significantly improved stability, further enhancing the pulse energy and peak power of lasers [242]. Later, the unique nonlinear optical properties of multifarious heterojunctions were verified, providing an optimized method for the saturable absorption properties of materials [241,242,243,244,245,246,247,248,249,250].

Figure 10. The performance of passively modulated lasers based on heterostructure. (a) The configuration of the laser cavity based on graphene/Bi₂Te₃. (b) Optical spectrum. (c) Autocorrelation trace [235]. (d) Schematic of the mode–locked all–fiber laser and measurement system. (e) Spectra of the output lasers with the bare graphene–SA and heterostructure–SA. (f) Autocorrelation traces [237]. (g) SEM image of Ti₃C₂T_x/CuO composite. (h) Optical spectrum. (i) Autocorrelation trace [241].

Figure 10. The performance of passively modulated lasers based on heterostructure. (a) The configuration of the laser cavity based on graphene/Bi₂Te₃. (b) Optical spectrum. (c) Autocorrelation trace [235]. (d) Schematic of the mode–locked all–fiber laser and measurement system. (e) Spectra of the output lasers with the bare graphene–SA and heterostructure–SA. (f) Autocorrelation traces [237]. (g) SEM image of Ti₃C₂T_x/CuO composite. (h) Optical spectrum. (i) Autocorrelation trace [241].

In general, heterostructure SAs have fundamentally overcome the performance bottlenecks of single 2D material SAs through the performance synergy of multiple materials and have emerged as the core development direction in the field of passively modulated ultrafast lasers.

Table 5. Performance summary of passively modulated lasers based on heterostructure absorbers. λ₀ is the center wavelength, τ_p is the pulse width, f_rep is the repetition frequency, E_p is the pulse energy, P_ave is the average output power.

SA	Gain Medium	λ0 (nm)	τp (ps)	frep (KHz)	Ep (nJ)	Pave (mW)	Thickness	References
Te/Se	Er3+ silica fiber	1500	0.889	18,500	-	-	25~33 nm	[245]
Graphene/BP	Er3+ silica fiber	1529.9	0.82	7430	-	-	4.4~5.4 nm/ 7.9~13.6 nm	[251]
		1531	0.148	7500	-	-
Ti3C2Tx/CuO	Er3+ silica fiber	1537	0.495	9210	-	3.87	-	[241]
MoS2/Sb2Te3	Er3+ silica fiber	1554	0.286	36,460	-	20	24 nm	[239]
Graphene/WTe2	Er3+ silica fiber	1558	1.2 × 106	56.8	3.77	3.15	-	[252]
WS2/MoS2	Yb3+ silica fiber	1035.07	0.748	3900	-	6.8	1.69 nm	[243]
	Er3+ silica fiber	1558.6	0.682	3900	-	10.32	1.86 nm
Bi2Te3/FeTe2	Er3+ silica fiber	1558.8	0.481	23,000	0.27	6.257	-	[250]
Graphene/Mo2C	Er3+ silica fiber	1559	0.723	15,320	0.713	10.93	monolayer/ 7.8 nm	[253]
WS2/MoS2	Er3+ silica fiber	1560	0.154	74,672	-	19.8	63 nm	[240]
WS2/MoS2	Er3+ silica fiber	1562.6	0.296	36,460	-	25	132.3 nm	[230]
WSe2/MoSe2	Er3+ silica fiber	1564	1100	4459	0.7	3.1	2.5 nm	[246]
Graphene/ BN/MoS2	Er3+ silica fiber	1565.8	1.2	13,200	-	-	-	[237]
Ti3C2Tx/TiO2	Er3+ silica fiber	1565.8	0.661	22,750	-	5.964	-	[244]
Graphene/MoS2	Er3+ silica fiber	1567.2	9.31 × 106	21.9	98.6	2.16	-	[254]
		1571.8	2.2	3470	-	-
Graphene/Bi2Te3	Er3+ silica fiber	1568	0.837	17,300	0.178	3.07	monolayer/ 8.5 nm	[235]
Graphene/WS2	Er3+ silica fiber	1593.5	1.41	3630	0.52	1.9	monolayer/ monolayer	[255]
		1601.9	0.66	21,780	-	-
Nb2C/MoS2	Er3+ silica fiber	1597.5	0.096	60,040	0.065	3.9	11.1 nm	[247]
	Tm3+ silica fiber	1901.7	0.551	22,410	0.89	20
BP/ReS2	Tm3+: YAP	1932	3.66 × 105	50	0.01	528	4 nm	[249]
Graphene/BN	Tm3+: YAP	1934.8	6.07 × 105	188.43	19,430	2.66	monolayer/-	[236]
Graphene/WS2	Tm3+: YAP	1980.1	1.16 × 106	90.36	-	1293	7 nm	[256]
WSe2/CuO	Tm3+: YAP	1986.8	7.53 × 105	68.68	33,780	2320	-	[242]
WSe2/BN	Tm3+: YAP	1989.4	1.28 × 106	43.51	19,170	834	-	[248]
Graphene/BN	Ho3+: YLF	2070	1.56 × 106	52	730.36	38	monolayer/-	[257]
Graphene/Te	Er3+: ZBLAN	2776.8	12.18	41,700	3.58	149.3	few-layer	[238]

summarizes the achievements of passively modulated lasers based on heterostructure SAs. The subsequent research focuses on the standardization and upscaling of heterostructure fabrication processes, the atomic-level precise regulation of interfacial properties, the reduction of non-saturable losses, and the improvement of optical damage thresholds. It will propel heterostructure SAs from basic laboratory research towards industrialized and commercial practical applications [237].

3.6. Summary

Figure 11a compiles the center wavelength and pulse duration of passive ML lasers employing different LDM SAs, including graphene, TMDs, BP, CNTs, TIs, TMOs, MXenes, and heterostructures. In accordance with the figure, graphene-based ML lasers exhibit an ultra-broadband operation from ~1 μm to 4.4 μm, with pulse widths down to tens of femtoseconds (e.g., 29 fs), benefiting from its zero-bandgap structure and ultrafast carrier dynamics (initial relaxation time ~70 fs). TMDs (e.g., MoS₂, WS₂, MoSe₂, WSe₂) cover a wide near- to mid-infrared range (~1 μm to ~2.8 μm) and achieve pulse durations as short as ~67 fs (WS₂) or ~154 fs (MoS₂/WS₂ heterostructure), owing to their layer-dependent direct bandgap and strong nonlinear absorption. Black phosphorus (BP)-based lasers demonstrate broad tunability from ~1 μm to ~3.5 μm with ultrafast pulse widths reaching ~102 fs, attributed to its direct bandgap (0.3–2.0 eV) and large modulation depth (~15–27%). CNTs, TIs (e.g., Bi₂Te₃, Bi₂Se₃, Sb₂Te₃), TMOs (e.g., MoO₃, TiO₂), and MXenes (e.g., Ti₃C₂Tₓ) also contribute to sub-picosecond and few-picosecond pulse generation across the 1.5–3.5 μm region, with CNTs showing excellent thermal stability and TIs offering ultrahigh modulation depth (up to 95%). Heterostructures (e.g., graphene/Bi₂Te₃, WS₂/MoS₂, Ti₃C₂T_x/CuO) further extend the wavelength coverage and improve pulse performance through synergistic band alignment and interfacial charge transfer, achieving pulse widths as short as ~154 fs. Overall, the data confirm that LDM SAs enable compact, broadband, and ultrafast mode-locked lasers spanning from the near-infrared to mid-infrared.

Figure 11b presents the center wavelength and pulse duration of passive QS lasers utilizing different LDM SAs. In accordance with the figure, graphene-based QS lasers operate over a broad spectral range from ~1 μm to 2.8 μm, with pulse durations typically in the sub-microsecond to microsecond regime (e.g., 300 ns at 2.8 μm), benefiting from its high damage threshold and broadband modulation capability. TMDs (e.g., MoS₂, WS₂, MoSe₂, WSe₂) and BP also enable stable QS operation in the near- to mid-infrared (e.g., 2.8 μm for MoS₂, 3.5 μm for BP), with pulse widths ranging from hundreds of nanoseconds to a few microseconds. CNTs, TIs, TMOs (e.g., Fe₃O₄, TiO₂), and MXenes (e.g., Ti₃C₂T_x) further demonstrate QS operation at wavelengths extending to ~3 μm, with pulse energies reaching tens of microjoules and average powers at the watt level. Heterostructures (e.g., Te/graphene, WSe₂/CuO) show enhanced stability and modulation depth, enabling high-energy QS pulses in the mid-infrared. The compiled results underline that LDM SAs are highly effective for passive QS, offering flexible wavelength tunability, simple cavity designs, and robust pulse generation across the 1–4 μm spectral region, which is critical for applications requiring high pulse energy and moderate pulse width.

4. Synthesis Methods for LDM SAs

The synthetic techniques for the LDM SAs are classified into top-down and bottom-up approaches. The top-down exfoliation methods comprise liquid-phase exfoliation (LPE), mechanical exfoliation (ME), laser etching, aqueous acid etching, and electrochemical exfoliation, where single-layer or new-layer nanosheets are separated from bulk materials by violating the van der Waals force between layers. Bottom-up methods include molecular beam epitaxy (MBE), magnetron sputtering, PLD and CVD, where high-quality 2D materials in atomic layer scale are effectively synthesized by explicitly adjusting the chemical reactions among solid precursors. Figure 12 shown the schematic diagram of most common synthesis methods.

The ME technique is commonly used in the manufacturing of 2D layered nanomaterials (e.g., graphene, MoS₂, WS₂, SnS₂, BP). High-quality mono- and few-layer materials can be obtained by using adhesive tape to overcome interlayer van der Waals forces. Since its first application in graphene in 2014, this method has been extensively employed for the preparation of 2D SAs materials owing to its simplicity, flexibility, and cost-effectiveness. Nevertheless, it does possess certain drawbacks. For instance, it is unable to synthesize large-area single-layer materials, and the thickness is non-uniform and uncontrollable. LPE is capable of eliminating interlayer forces through ultrasonication and exfoliating LDMs from their bulk (or aggregation) state in liquid media. This methodology can also be integrated with other approaches, such as ion exchange and ion intercalation. Consequently, this synthesis method has been extensively applied throughout the entire LDMs system (e.g., graphene, BP, TMDs, MXenes, CNTs, TMOs, TIs, MOFs). Although this approach facilitates the regulation of the size of nanomaterials and produces highly concentrated dispersions by adjusting the reducing agent, pH, or surfactant, it yields nanosheets with wide thickness distributions and may introduce solvent residues or structural defects.

In general, the introduction of impurities and defects also constitutes the common drawback of all top-down preparation methods. In addition, the uncontrollable scale and random thickness of few-layered materials acquired using the top-down techniques are counterproductive to the efficiency of an SA. By contrast, the bottom-up approach uses physical or chemical methods to deposit or build up LDMs. During this growth process, precise control over their size and morphology can be achieved by regulating the precursor and growth conditions.

Magnetron sputtering, as a physical vapor deposition (PVD) technique, achieves the deposition of materials into thin films by bombarding a target with ions produced by glow discharge. This method has currently been used to prepare 1D and 2D thin films (e.g., carbon-based, sulfide-based, nitride-based, and heterostructures) with precisely controlled thickness. However, the equipment cost associated with this technique is comparatively high, and the sputtering process may lead to stoichiometric deviations. The other two PVD growth techniques are PLD and MED. PLD is a methodology that employs high-energy-density lasers to irradiate target materials, enabling the deposition of the ejected material onto a substrate for the fabrication of thin films. MBE enables layer-by-layer growth with atomic-level precision under ultrahigh vacuum by precisely controlling the beam current and energy of the source molecules. Both preparation methods produce nano-films of extremely high purity and perfect lattice quality, which are well-suited for growing high-quality low-dimensional quantum dots, multicomponent thin films, heterojunctions, and other nanostructures. Nevertheless, their applications in the field of nanomaterial preparation are still limited due to the high costs of equipment and the difficulties in technical control.

CVD is based on the thermodynamics of surface chemical reactions. Gaseous precursors undergo decomposition, adsorption, nucleation and growth on the surface of a high-temperature substrate. In the case of 2D materials, the growth process adheres to a self-limiting mechanism. Once a single layer is formed, the active sites on the surface become passivated. Subsequently, the growth of additional layers necessitates surmounting a higher energy barrier, which enables precise regulation of the layer number. In comparison to top-down exfoliation techniques, CVD facilitates the controllable growth of large-area, high-quality thin films with accurate layer-number regulation, and has emerged as the predominant growth approach for LDMs and heterojunctions. Figure 13 shows the schematic diagram of the CVD setup for synthesis of aligned MoTe₂ nanoribbons [262].

However, the current routes (containing two steps: chemical vapor deposition and spin coating) for constructing LDM SAs are suffering from limited flexibility in substrate choice and the introduction of impurities during the transfer process or due to molecular functional groups. This affects the broadband absorption parameters of LDM SAs, such as non-saturated loss, ultrafast optical response, modulation depth, stability, and thermal tolerance.

In recent years, our research group has been investigating methods for direct CVD growth on substrates. As of 2022, we have demonstrated a high-quality graphene saturable absorber mirror (GSAM) grown directly on a calcium fluoride (CaF₂) substrate by a low-temperature plasma enhanced chemical vapor deposition (PECVD) method for mid-infrared pulse modulation [63]. Figure 14a shows the process of PECVD synthesis and material characterization. The controllable growth of a high-quality graphene film on the nickel-modified CaF₂ substrate was realized by adjusting the growth time and hydrocarbon ratio. This method circumvents wet-coating procedures in various previously reported SAs, thereby mitigating the impurity-induced performance degradation. Consequently, the GSAM exhibits excellent nonlinear optical absorption and operates a high-efficiency and high-peak power QS laser at 2.8 μm.

In 2025, our group fabricated 2D transition metal telluride (TMT) SAs using the PECVD method [112]. This approach involves the direct in situ growth of large-area, high-quality 2D ZrTe₃ and TaTe₂ films on CaF₂ substrates as SA mirrors, followed by oxygen plasma passivation. Passivation serves to effectively protect the SA material from widespread atmospheric corrosion and laser-induced damage. The images of TMT-SAs are shown in Figure 14c–f. TaTe₂ and ZrTe₃ SAs exhibit strong nonlinear optical absorption properties, with modulation depths of 5.4% and 7.2%, respectively. Remarkably, the as-synthesized 2D TMT-SAs retain excellent long-term stability, which are capable of generating stable Q-switched MIR pulses. Our strategy paves a way for developing high-quality and stable GSAMs for industrial applications of pulsed mid-infrared lasers.

5. Conclusions and Outlooks

This paper has systematically reviewed the research progress of near/mid-infrared pulsed lasers based on LDM SAs, encompassing the fundamental principles of saturable absorption and passive modulation, the nonlinear optical properties of various LDMs (graphene, TMDs, BP, CNTs, TIs, TMOs, MXenes), device fabrication methods, and their application results in Q-switched and mode-locked lasers. Overall, significant advances in pulsed lasers based on LMDs SAs have been achieved, which focus on key performance metrics, including center wavelength (from the 400 nm range to 4.4 μm), pulse width (from sub-microsecond to tens of femtoseconds), repetition rate (from kHz to hundreds of MHz), pulsed energy (from sub-nanojoule to dozens of microjoules), and output peak power (from watt to kilowatt level). They provide compact, efficient, and ultrafast laser source solutions for applications in biomedicine, atmospheric remote sensing, materials processing, and nonlinear optics. Although LDM SAs have achieved notable advancements in the domain of ultrafast laser pulse modulation, a number of core challenges persist before the complete realization of the transition from laboratory research to industrial and practical applications. In view of the current development trends within the field, future research is anticipated to attain breakthroughs in the subsequent directions.

The subsequent crucial step to be addressed is the exploration of a highly stable and batch fabricable preparation technology for LDM SAs. Environmental stability (especially for sensitive materials like BP and MXenes) and device consistency are major obstacles to practical application. Standardizing scalable fabrication processes, such as CVD and PECVD, is crucial to achieving a low-cost and reproducible manufacturing of high-performance SAs. Simultaneously, future strategies ought to encompass the in situ passivation, atomic layer deposition encapsulation, polymer composites, and inorganic/organic multilayer coating approaches. These approaches are anticipated to significantly enhance the material’s resistance to moisture and oxidation, all the while preserving its nonlinear performance.

At the laser system level, it is of great significance to promote the integration of all-fiber, all-solid-state, and photonic chip technologies. This is conducive to further compressing the pulse width to the sub-hundred femtosecond, increasing the repetition rate, and simultaneously enhancing the single-pulse energy and peak power. At present, the mid-infrared laser still suffers from a shortage of SAs and crucial fiber-optic components (e.g., isolators, couplers, and wavelength division multiplexers). The extensive utilization of free-space SAs exerts a significant influence on the stability and compactness of the system, which constitutes a fundamental device-related bottleneck that requires immediate resolution. Therefore, the mid-infrared LDM SAs need to be monolithically integrated with components such as micro/nano optical fiber, micro/nano optical cavities, waveguides, and fiber Bragg gratings to construct compact, robust, and tunable ultrafast mid-infrared laser sources.

The current maximum operation wavelength demonstrated in LDM SAs is approximately 4.4 μm [67]. Nevertheless, it does not represent the fundamental upper bound, and there is still substantial scope for improvement. The main challenges for extending the wavelength further into the mid-infrared (e.g., 5–10 μm and beyond) include: (i) the intrinsic bandgap limitations of most LDMs, (ii) increased free-carrier absorption and phonon absorption at longer wavelengths, and (iii) the lack of compatible fiber-optic components in this spectral region. The emergence of narrow-bandgap materials (e.g., graphene, TIs, MXenes) and the development of advanced fabrication strategies present promising approaches for the extension of the operating wavelength.

In summary, LDM SAs are undergoing a critical transition from being “material-driven” to “performance-driven” and “application-driven”. Through multi-dimensional advancements of LDM SAs in material innovation, heterostructure engineering, device integration, and advanced fabrication techniques, pulsed lasers are expected to achieve superior performance, enhanced stability, and a higher level of integration in the near future.