† Corresponding author. E-mail:
Project supported by the National Natural Science Foundation of China (Grant Nos. 61378025 and 61427812), the Shuangchuang Team Program of Jiangsu Province, China, the National Key Basic Research Program of China (Grant No. 2014CB921101), and the State Key Laboratory of Advanced Optical Communication Systems Networks, China.
As the fundamental optical properties and novel photophysics of graphene and related two-dimensional (2D) crystals are being extensively investigated and revealed, a range of potential applications in optical and optoelectronic devices have been proposed and demonstrated. Of the many possibilities, the use of 2D materials as broadband, cost-effective and versatile ultrafast optical switches (or saturable absorbers) for short-pulsed lasers constitutes a rapidly developing field with not only a good number of publications, but also a promising prospect for commercial exploitation. This review primarily focuses on the recent development of pulsed lasers based on several representative 2D materials. The comparative advantages of these materials are discussed, and challenges to practical exploitation, which represent good future directions of research, are laid out.
Ultrafast science is a highly interdisciplinary research area that encompasses laser physics, materials science, nanotechnology and biomedical research.[1–5] Today, ultrafast lasers can routinely generate pulses on the femtosecond scale, and advanced systems employing nonlinear optical techniques can produce durations down to several tens of an attosecond.[6,7] These ultrafast systems are playing increasingly important roles across a broad range of science and engineering subjects, enabling not only ever more profound knowledge about ultrafast physical, chemical and biological phenomena, but also orders of magnitude improvement in our manufacturing and metrological capabilities.[8–10] While the theoretical foundation of ultrafast pulse generation in mode-locked lasers was laid several decades ago,[11] the quest for ultrafast lasers with continuously improved performance, i.e., shorter pulse duration, wider spectral access, higher output power, and better noise characteristics, has never stopped.[12–14]
Of the many available techniques for ultrashort-pulse generation, passive mode-locking represents the state-of-the-art for enabling extremely short pulses.[3] The key operation principle of passive mode-locking involves the use of a nonlinear optical material or device, known as a saturable absorber (SA), which can transit between different absorption states on an ultrafast timescale, upon excitation by a high intensity light field.[15] The passive saturable absorber induces a periodic modulation of the circulating light field in the laser cavity, causing a large number of longitudinal modes to oscillate in phase, thereby forming a regular train of short pulses in the time domain.[11] The underlying mechanism for saturable absorption can be remarkably simple, typically based on Pauli exclusion principles,[16] i.e., the absorption of a material would decrease if a proportionally large number of electrons are found in the upper excited states, although other physical mechanisms can also lead to a similar overall effect.
Throughout the past few decades, a number of materials have been identified and developed as saturable absorbers, including organic dyes, colour centers, as well as transition metal and rare earth ions.[17–19] However, these early materials are not ideal. For example, organic dyes are usually difficult to handle in a laser cavity and are susceptible to optical damage. Ion-based saturable absorbers are typically employed in the form of doped glass/crystal films with relatively high losses due to unavoidable bulk optical coupling, which prevents wide use in solid-state and semiconductor lasers. An artificial saturable absorber effect known as the Kerr lens was discovered in 1991 by Sibbett and co-workers,[20] and subsequently led to the successful commercialization of the solid-state Ti: sapphire lasers. However, the Kerr lens is an extremely fast SA with a response time around a few femtoseconds. Such a transient characteristics is difficult to tune as it arises from the intrinsic dielectric response of the crystal. This was known to cause difficulty for the mode-locked laser to self-start, presenting an undesirable feature for practical systems.[3] The advent of the semiconductor saturable absorber mirror (SESAM), invented in the early 1990s by Bell Labs researchers, successfully worked around such a technical problem and since its introduction, SESAMs have been routinely used to initiate mode-locking in Kerr lens mode-locked Ti:sapphire lasers.[21] Before long, with advances in multiple quantum-well design and growth techniques, many key optical parameters of SESAMs were found capable of being flexibly customized. In particular, through low-temperature growth or post-growth processing, the carrier relaxation times, which not only critically govern the initial pulse formation processes, but also directly affect the saturation fluence of the saturable absorbers, could be broadly tailored within a range from a few picoseconds to nanoseconds.[22,23] The saturation fluence or intensity of the device can also be tuned by customizing the top reflective mirror or by precise engineering of the composition of the absorber layer.[15] Thanks to these technical advances, SESAM has established itself as the most prevalent absorber technology for solid-state, fiber and semiconductor lasers, enabling significant optimization of near-infrared lasers and making them the work horses for today's ultrafast science exploration.[3] Despite the technical prowess, SESAMs have a known drawback. Due to the use of conventional semiconductor materials and optical resonant structures, the operation range of SESAMs is limited to the near-infrared range, typically below 2 μm (while the nonlinear optical bandwidth is ∼ 100 nm about the design wavelength). Scaling the operation wavelength of SESAMs further into the mid-infrared, i.e. beyond 3 μm, remains elusive. The quest for an alternative saturable absorber technology that can address these limitations while retaining the key benefits of SESAMs remains an important research topic for the ultrafast laser community.
Single- and few-layer 2D crystals have received significant attention since the isolation of graphene in 2004.[24] Due to the unique low-dimensional physics, a range of emerging 2D materials have been used in the development of next-generation photonics and optoelectronic technologies. Of the many possibilities, the use of 2D materials as broadband, cost-effective and versatile ultrafast optical switches (or saturable absorbers) for short-pulsed lasers constitutes a rapidly developing field with not only a good number of publications, but also a promising prospect for commercial exploitation. In the following, we briefly review recent results of short-pulsed lasers enabled by a few representative two-dimensional materials, covering graphene, transition metal dichalcogenides, and black phosphorus in turn. It should be noted that due to the large number of reports in the literature, this review is not intended to be exhaustive, but rather efforts are made to highlight the most significant recent results, to discuss the comparative advantages of different materials (especially those of broad relevance to optical and optoelectronic applications), and to point out remaining technical challenges to overcome for practical exploitation.
Graphene is a flat monolayer of carbon atoms tightly arranged into a two-dimensional honeycomb lattice, and is a basic building block for graphite. Since its successful isolation in 2004, graphene has quickly emerged as the most promising material for nanoelectronics and nanophotonics. The electronic and optical properties of graphene as well as the roadmap for its technological development are summarized in a few recent review papers.[25–28] Graphene was first employed for the mode-locking of lasers independently by two groups.[29–31] Compared with carbon nanotubes, which have similar atomic arrangement and have been identified earlier as a broadband saturable absorber material,[31–33] graphene exhibits additional advantages in terms of operating bandwidth, uniform nonlinear optical response over broad spectral region, as well as ease of integration into photonic components and systems. While carbon nanotubes are most often used in the form of polymer composites, the fact that large-scale, high-quality crystalline graphene can be grown on copper foil using chemical vapor deposition (CVD) method and can subsequently be flexibly transferred to arbitrary optical substrates makes graphene highly adaptable for various laser formats. Thus far, graphene has been used to mode-lock fiber,[29–31,34–39] solid-state,[40–44] waveguide,[45] and semiconductor lasers.[46,47] A few comprehensive reviews of graphene-based saturable absorber devices are available.[48–50] Graphene derivatives, including graphene oxide and reduced graphene oxide, also exhibit ultrafast optical properties and can enable pulsed laser mode-locking. However, the necessity of additional chemical/physical processing renders the fabrication less well controlled and in the meantime unavoidably introduces a considerable amount of defects and gap-states.[51] Among various 2D materials, graphene has thus far demonstrated performance levels most comparable to or even better than conventional semiconductor devices and it has also shown the best adaptability for fulfilling requirements of a wide range of laser formats.
Similar to broadband operations of SWNTs,[52–54] graphene absorbers have been shown to enable broadly tunable Q-switched and mode-locked operations.[34,35] Yang and co-workers used a CVD-grown multilayer graphene SA (> 5 layer) to mode-lock three different lasers, including ytterbium-, erbium-, and thulium-doped fiber cavities, convincingly validating the intrinsic broadband feature of graphene flakes.[36] Sotor et al. demonstrated a passively synchronized thulium- and erbium-doped fiber laser by a common graphene SA, where soliton operations were simultaneously achieved at 1559 and 1938 nm, respectively.[37] Such synchronous lasers could be extremely useful for different-frequency generation (DFG) for mid-infrared operation beyond 3 μm. In terms of repetition rate scaling, the highest repetition was achieved based on a short-cavity fiber laser by Martinez and co-workers, where a repetition rate of 20 GHz is achieved through careful optimization of the gain and loss of a miniaturized FP cavity.[38] Zhang and co-workers demonstrated for the first time mode-locked operation other than conventional soliton regime, i.e., dissipative soliton operation, validating the versatility of graphene SA for various pulsation regimes.[39]
While early studies are mainly focused on the conventional near-infrared band, recent efforts have shifted towards the use of graphene for long-wavelength lasers, especially those around 2 μm and beyond.[55–63] The exploration of graphene-based mid-infrared fiber lasers started in 2012, with several groups reporting pulsed laser operations around 2 μm. For fiber lasers, liquid phase exfoliated (LPE) graphene-based composite films were first used to demonstrate a Q-switched laser[55] and a mode-locked thulium fiber laser,[56] respectively. Wei and co-workers demonstrated the first graphene-based pulsed laser operation at a wavelength close to 3 μm, where graphene was deposited onto a fiber ferrule mirror using the optically-driven deposition method.[57] The laser emitted 2.9 μs pulses with a pulse energy up to 1.67 μJ. By optimizing the fiber laser cavity and using a multi-layer graphene-covered gold mirror, mode-locked operation at a wavelength of 2.8 μm with a pulse duration ∼ 42 ps was demonstrated (see Fig.
Flexibility in controlling the optical parameters, such as linear absorption, relaxation times, modulation depth, etc., is an important aspect for engineering practical SA devices and the availability of a large parameter space would allow for adequate optimization of SA-enabled, pulsed laser regimes. Indeed, SESAM has been the most capable of SA devices in terms of parameter customizations. The gate-tunable feature of graphene makes it an excellent choice for an electrically-tunable SA, which may provide multiple benefits for laser operations, including actively controlling the operating regime of a pulsed laser.[64] Rotermund, Yeom, and co-workers demonstrated an electrically manipulable in-line graphene SA device by integrating graphene-based field effect transistors on a side-polished fiber (SPF) (see Fig.
Novel approaches to tune the relaxation time of graphene are to be developed and would further expand the versatility of graphene-based optical devices, pertinent not only to saturable absorbers but also to photodetectors, solar cells, as well as light-emitting devices. Apart from the active approaches, passive ways of adjusting the graphene SA’s parameters have also been demonstrated through fine tuning the thickness of a spacer between a single-layer graphene and a high-reflection mirror (see Fig.
Transition metal dichalcogenides – layered crystals comprising X−M−X sheets (where
For pulsed laser applications, TMD-enabled pulsed laser was first demonstrated by Zhang and co-workers in early 2014, where few-layer MoS2 obtained by hydrothermal method was used to mode-lock a Yb-doped fiber laser operating at 1054.3 nm, though a relatively long pulse duration of ∼ 800 ps was obtained.[89] Soon after the first report, Luo and Xu et al. demonstrated femtosecond pulse generation (∼ 700 fs) using a MoS2-PVA composite absorber in an Er-doped ring cavity.[90] A number of reports followed, unambiguously verifying the capability of TMDs as effective saturable absorbers for fiber lasers operating across the visible to the near infrared range ∼ 2 μm.[91–103] In particular, broadband operation is manifested in the demonstration of a wavelength-tunable Q-switched regime,[91] a wavelength-tunable mode-locked regime[92] and a single MoS2 device-enabled, Q-switched, Yb-, Er-, and Tm-doped fiber laser across 1–2 μm.[93] While the initial studies primarily employed MoS2 as the SA material, other TMD materials such as WS2[98–101] and MoSe2[102,103] have also been demonstrated. The earliest study on the ultrafast nonlinear optical properties of TMDs revealed strong nonlinear absorption of MoS2 in the visible and infrared range (see Fig.
Most of the existing TMD-based pulsed lasers employ LPE for obtaining single- or few-layer TMD flakes. Li and co-workers demonstrated the first mode-locked laser based on a CVD-grown multilayer MoS2 SA. The authors showed robust mode-locking in an erbium-doped fiber laser, illustrating the potential of using large and uniform, polymer-free material for pulsed laser operation, although the multilayer still has to be transferred onto the fiber ferrule end.[108] Following similar CVD growth and transfer methods, Khazaeizhad and co-workers demonstrated an evanescent field interaction device, where both anomalous and normal dispersion cavities were successfully mode-locked for 1.55 μm operation.[109] One advantage of TMD compared with graphene is the compatibility of large-area deposition directly on dielectric substrates, such as quartz and silica. Single crystal material with lateral sizes larger than several tens of microns can now be routinely grown. Devices that are free from transfer processes would be highly desirable, i.e., MoS2 multilayers directly grown on optical substrate or fiber end, and will greatly facilitate highly repeatable device fabrication.
Since graphene was identified as a promising SA material, topological insulators (Tis), another mainstream category of Dirac materials that have drawn broad attention in the field of condensed-matter physics have been identified as potential saturable absorber materials (such as Bi2Te3,Bi2Se3, Sb2Te3, etc.).[110–119] This review however will not summarize those results in detail as similar results were demonstrated following the same vein as with graphene absorbers, though the first report of TI-based SA preceded those of TMDs.[91] In the following we focus on the recently emerging black phosphorus which has become the new focus of the field.
Black phosphorus (BP), the most thermodynamically stable allotrope of the element phosphorus, was recently rediscovered as an intriguing two-dimensional material as motivated by the demonstration of a BP-based field-effect transistor by Zhang and co-workers.[120] For photonics applications, BP’s unique strength lies in its layer-dependent direct bandgap, i.e., from ∼ 0.3 eV (bulk form) to ∼ 2.0 eV (monolayer), which provides an effective bridge between the energy gap of the semi-metallic graphene and the relatively large band gap TMDs.[121] In addition to its tunable bandgap and high carrier mobility, the structural anisotropy of BP leads to highly anisotropic mechanical, electrical, and optical properties.[122,123] A number of electronic and optoelectronic applications based on BP and its heterostructures have been developed thus far.[124,125] For example, a BP-based photodetector that can respond to mid-infrared light ***∼3.39 μm with a high responsivity of ∼ 82 A/W was recently demonstrated,[126] showing great potential for chip-scale mid-infrared sensing and imaging applications based on the inherent advantages offered by this emerging 2D material.
Due to its relatively narrow direct bandgap, multilayer BP was considered a promising saturable absorber for the long wavelength range, where light–matter interactions in graphene and TMDs are not as strong. BP nanoplatelets dispersed in different solvents obtained by liquid phase exfoliation method[127] were first investigated by Z-scan technique across a broad spectral range, spanning 400–1930 nm.[128] Regardless of the characterization wavelength and pulse duration (femtosecond or picosecond), the BP-based saturable absorbers are found to exhibit quite large modulation depth (> 10%), though the saturation intensity deduced from the femtosecond Z-scan measurement is quite high, about several hundreds of GW/cm2. On the other hand, a few studies on the photocarrier relaxation dynamics in BP have been reported.[129–131] While angle-sensitive transient effects were identified in all three pump–probe studies, the present results do not converge to a unified physical picture by which photocarriers’ dynamics can be well accounted for. Further investigation of BP’s photocarrier excitation and relaxation would be important for understanding their detailed nonlinear optical behavior.
For short-pulsed laser applications, mechanically exfoliated BP was first used as a saturable absorber to enable both Q-switched and mode-locked operations for a telecom band Er-doped fiber laser ∼ 1.55 μm.[132] Inspiringly, the nonlinear optical characteristics of the exfoliated sample can be effectively tuned through layer thickness control, i.e., the modulation depth in transmittance can be changed from 8% (thin sample) to 18.5% (thick sample). The operation wavelength of BP-enabled pulsed laser was soon extended to the thulium fiber band ∼ 1.9–2.0 μm, where a mechanically exfoliated thin film of BP (about 300 nm in thickness) was deposited directly onto the fiber ferrule end.[133] The thulium laser generates 739 fs pulses, with a ∼ 36.8 MHz repetition rate and 1.5 mW output power, which corresponds to a pulse energy of 80 pJ. No damage or degradation was observed for the BP saturable absorber at this operation condition. To mitigate the potential damage caused by direct laser irradiation, evanescent field interaction was employed, where the BP is deposited either on micro-fiber or D-shape fibers. A number of Q-switched or mode-locked lasers operating in the 1–1.55 μm range were demonstrated using this scheme.[134–137] By fabricating a BP saturable absorber mirror, Zhang and co-workers demonstrated the first mode-locked bulk laser operating at a wavelength of 1064 nm. The pulse duration is relatively long ∼ 6.1 ps, with an average output power of 460 mW.[138] Through intra-cavity dispersion control, the pulse duration and output power were significantly optimized, with 272 fs pulses and a peak power approaching 24 MW achieved.[139] Thus far, the longest wavelength bulk laser enabled by a BP absorber is a passively Q-switched Cr:ZnSe laser operating at a wavelength around 2.4 μm, where pulses as short as 189 ns were achieved, corresponding to a pulse energy of 205 nJ.[140]
Qin and co-workers were the first to achieve 3 μm operation by using LPE processed BP-NMP solution. A reflective device is fabricated by dropping the solution onto a gold-coated mirror and drying it in a cabinet (80% reflectance at 2.9 μm). An Er:ZBLAN fiber laser operating at 2.78 μm was successfully Q-switched with pulse width around 1.2 μs, and the BP SAM device was used as a cavity mirror (free-space coupled).[141] By transferring a mechanically exfoliated BP onto the gold-coated mirror, mode-locked operation at 2.78 μm was subsequently obtained with a pulse duration of 42 ps, a repetition rate of 24 MHz, which are similar to those achieved by graphene-based absorbers.[142] Through thickness characterization, the BP flake was found to be ∼143 nm, corresponding to ∼238 layers. According to the bandgap equation, a bandgap of 0.33 eV is inferred, which means an operating range up to 3.8 μm is possible.[142] It should be noted that the measured modulation depth and saturation intensity is about 19% and 9 μJ/cm2, respectively (Fig.
The unique set of optical properties of emerging two-dimensional materials have established them as a technologically important category of materials that hold great potential to be exploited for cost-effective and broadband saturable absorbers. The question of whether one particular material would stand out from the relatively broad range of materials is still open to debate. Nevertheless, to be used as a versatile optical switch that will be amendable for different laser formats, some generic requirements apply. These include the reliability and repeatability of material growth; whether large area and uniform materials can be grown on arbitrary optical substrates (preferably without the need of transfer); whether the nonlinear optical features could be precisely and flexibly controlled; and if the material can tolerate high laser intensities at a variety of laser wavelengths and pulse durations. As it currently stands, while some of the requirements are adequately met by emerging 2D materials, not all criteria have been fulfilled in one specific material yet. For example, graphene possesses advantages in large-area and reliable growth based on CVD method. However, the difficulty with which to precisely control the layer number and the relatively weak light–matter interaction, limit its potential in scenarios where both a large absolute absorption and strong modulation depth are needed. Employing a capacitor-like device or optical resonance designs have shown flexible and precise control of nonlinear optical characteristics of graphene, including both modulation depth and linear absorption, but this is normally at the expense of more complex device geometry and transient characteristics of graphene are hard to customize over a large temporal range. The inherent excitonic features of TMDs dictate they can act as effective visible or UV saturable absorber with strong light–matter interaction. The fact that TMDs can be directly deposited on common dielectric substrates means they have a great potential for scalable fabrication, which has definitive advantages than devices made with alternative methods, such as mechanical exfoliation. However, the control of the nonlinear optical response of TMDs is not fully understood and requires further investigation. Using them to enable mode-locked visible laser, especially visible diode lasers widely found in the UV and visible band is of potential broad interests. In the case of BP, it should be mentioned that pulsed laser results are fast approaching those obtained by graphene and TMDs, although the first BP-enabled laser demonstrator is reported only in 2015. Enhancement of the stability of BP is still a prerequisite for the material to significantly expand its prospect for technological applications. Long-term stability optimization would remain a key aspect for BP-enabled devices. Compared to graphene and TMDs, one unique opportunity for BP arises from the fact that a relatively large number of layers is usually synergetic with functionality design. The thickness of a few hundred nanometers would comfortably allow chemical composition engineering to play a more active role in BP-based devices, compared with other few-layer 2D materials. Finally, for the emerging 2D materials to be reliably applied in optical or optoelectronic devices, the effects from adjacent layered materials or substrates (including passivation layers) have to be systematically studied.
In summary, ultrafast lasers enabled by emerging 2D materials have become a highly active field of research, where significant efforts have been dedicated to the demonstration of pulsed lasers with improved output characteristics including duration, repetition rates, average power and peak power. Revealing the underlying photo-physics together with the engineering of highly robust and tunable materials will help set the stage for these novel materials to be further exploited in practical applications, including not only ultrafast saturable absorbers, but also photodetectors, modulators and light-emitting devices.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] | |
[28] | |
[29] | |
[30] | |
[31] | |
[32] | |
[33] | |
[34] | |
[35] | |
[36] | |
[37] | |
[38] | |
[39] | |
[40] | |
[41] | |
[42] | |
[43] | |
[44] | |
[45] | |
[46] | |
[47] | |
[48] | |
[49] | |
[50] | |
[51] | |
[52] | |
[53] | |
[54] | |
[55] | |
[56] | |
[57] | |
[58] | |
[59] | |
[60] | |
[61] | |
[62] | |
[63] | |
[64] | |
[65] | |
[66] | |
[67] | |
[68] | |
[69] | |
[70] | |
[71] | |
[72] | |
[73] | |
[74] | |
[75] | |
[76] | |
[77] | |
[78] | |
[79] | |
[80] | |
[81] | |
[82] | |
[83] | |
[84] | |
[85] | |
[86] | |
[87] | |
[88] | |
[89] | |
[90] | |
[91] | |
[92] | |
[93] | |
[94] | |
[95] | |
[96] | |
[97] | |
[98] | |
[99] | |
[100] | |
[101] | |
[102] | |
[103] | |
[104] | |
[105] | |
[106] | |
[107] | |
[108] | |
[109] | |
[110] | |
[111] | |
[112] | |
[113] | |
[114] | |
[115] | |
[116] | |
[117] | |
[118] | |
[119] | |
[120] | |
[121] | |
[122] | |
[123] | |
[124] | |
[125] | |
[126] | |
[127] | |
[128] | |
[129] | |
[130] | |
[131] | |
[132] | |
[133] | |
[134] | |
[135] | |
[136] | |
[137] | |
[138] | |
[139] | |
[140] | |
[141] | |
[142] | |
[143] | |
[144] | |
[145] | |
[146] | |
[147] |