Significance Since the seminal work of Arthur Ashkin in 1986, optical tweezers have been extensively researched and used invarious fields, including physics, chemistry, and biomedicine.The radiation force induced by momentum exchange in light scattering and absorption is the fundamental mechanism of optical tweezers, allowing non-contact trapping and micro and nanoparticles manipulation.Continuous waves are primarily used in traditional optical trapping approaches for small particles.As an alternative, femtosecond laser pulses with high repetition rates have recently been applied to optical tweezers. The ultrashort pulse duration of femtosecond laser pulses produces minimal thermal effect during light-matter interactions, which lays the foundation for applications in biological sciences. Furthermore, the ultra-high peak power of femtosecond laser pulses excites the nonlinear response of trapped objects. In the presence of nonlinearity, some intrinsic properties of the sample (e.g., permittivity) will be changed, thereby breaking the force balance formed under a linear condition, and generating some novel phenomena with potential applications. To provide an overview and perspective on its developments, we review the research progress of femtosecond optical tweezers and discuss the nonlinear effects and applications in detail.Progress Over the past decades,femtosecond optical tweezers have made significant progress and facilitated researches in various disciplines,particularly in the field of biological sciences.In 2005, Mao et al. reported the stable trapping of red blood cells with femtosecond optical tweezers ( Fig. 1), demonstrating of practical applications of this technology in biological research. In 2008, Zhou et al. investigated the manipulation of red blood cells and their states under different laser powers using a similar experimental setup. In addition to the spatial manipulation of cells, femtosecond optical tweezers have been used in complex operations, such as cell fusion ( Fig. 2) and cell transfection ( Fig. 3). These applications of femtosecond optical tweezers are of great significance for the analysis of gene expression and immunotherapy. However, conventional optical tweezers are limited by the far-field diffraction and therefore are inefficient for trapping nanoscale objects (e.g., atoms, molecules, and quantum dots). Although the stiffness of optical tweezers can be enhanced by increasing the incident power, the robust optical intensity damages the sample. To improve trap precision, researchers have developed a new type of optical tweezers based on the principle of near-field optics. Among the different branches of near-field optics, surface plasmons hold the greatest potential for manipulation of objects at the nanoscale. In 2012, Roxworthy et al. reported enhanced particle trapping in a femtosecond plasmonic optical tweezers system (Fig. 4). The stiffness of the optical trap is enhanced by two times compared to the use of continuous-wave. In 2013, Shoji et al. demonstrated reversible trapping and release of λ-DNA molecules by switching femtosecond pulses in a plasmonic optical tweezers system (Fig. 5). In 2016, Kotsifaki et al.investigated the trapping efficiency of the femtosecond plasmonic optical tweezers based on gold-coated black silicon arrays. In 2021, Zhang et al. numerically calculated the force distribution for the quantum dots in a metallic bowtie structure illuminated by focused femtosecond laser pulses (Fig. 6).The interaction of femtosecond laser pulses and materials can excite several significant nonlinear effects, such as two-photon absorption, second harmonics, and the Kerr effect. Among them, the Kerr effect can proactively modulate the physical properties of samples, thereby breaking the force balance established under linear conditions and exhibiting some novel phenomena distinct from linear optical traps.Nonlinear optical trapping is a new area of study in the field of femtosecond optical tweezers. Over the past decade, fruitful efforts to facilitate its development have been made, including both experimental and theoretical research. In 2010, Jiang et al. reported the phenomenon of “trap split” when trapping gold nanoparticles viafemtosecond laser pulses (Fig. 7).In 2018, Gong et al. presented an analytical model for calculating the nonlinear optical force in femtosecond optical tweezers (Fig. 8); Zhang et al. achieved the multiplexed trapping of gold nanoparticles using femtosecond cylindrical vector beams (Fig. 9). In 2019, Gong et al. demonstrated the generation of optical pulling force by regulating the nonlinearity of surroundings (Fig. 10). In 2020, Huang et al. demonstrated that the Kerr effect of gold nanoparticles can induce a three-dimensional shell-like potential well (Fig. 11).Conclusions and Prospects Compared with continuous optical tweezers, femtosecond optical tweezers can significantly reduce the thermal effect in light-matter interactions because of the ultrashort pulse duration and temperature dissipation between pulses. At the same time, the ultra-high peak power of femtosecond laser pulse provides the basis for the nonlinear modulation of optical traps.These features of femtosecond optical tweezers of ferseveral distinct advantages in the manipulation of microscopic particles. However, the study of femtosecond optical tweezers are still in their early stages, particularly nonlinear optical trapping,which has received significant attention in recent years. There are still challenges to overcome in extending the capabilities and applicability of the technology. Even though we are confident that the uses of femtosecond optical tweezers will continue to expand in the near future, and more potential applications will be developed.

Significance With the continuous development of large equipment manufacturing technology, the increase of measurement scale and the improvement of accuracy requirements make the existing large size precision measurement technology is severely challenged. Length is the most basic and core observation in the field of mechanical geometry measurement. From space-scale measurement and positioning systems, to three dimensional coordinate measurement systems in industrial manufacturing, and even micro-nano scale observation systems, high precision distance measurement plays a fundamental and crucial role. Frequency scanning interferometry distance measurement, with its good comprehensive performance in accuracy, efficiency, field adaptability, traceability and other aspects, is especially suitable for large size absolute distance measurement tasks in the current industrial measurement environment.Progress Since the 1990s, extensive research on the error factors of frequency scanning interferometry (FSI) ranging has been conducted, resulting in a series of key technologies. When the scanning of optical frequency is linear, the FSI distance measurement accuracy is mainly determined by the phase measurement accuracy of interference signal and the measurement accuracy of optical frequency change.The phase extraction of FSI interference signals is generally achieved through the principle of orthogonal phase discrimination, which can achieve high accuracy, but requires high signal-to-noise ratio of interference signals. The spectrum analysis method can extract the frequency characteristics of the target under the condition of low signal-to-noise ratio. Since the femtosecond optical frequency comb is a series of frequency combs with equal optical frequency interval in the frequency domain and the optical frequency corresponding to each comb can be directly traced to the international standard, the scanning optical frequency range can be monitored by using the frequency characteristics of the optical frequency comb with high accuracy (Fig. 2). More intensive optical frequency monitoring information can be provided by the auxiliary interferometer, and the signals of the measured optical path are sampled at equal optical frequency interval. Thus, the scanning optical frequency range can be measured with high precision, and the nonlinear error of scanning frequency can be effectively corrected (Fig. 3). In addition to the above idea of realizing nonlinear correction through optical frequency monitoring, another idea is to rely on the phase-locked loop technology to actively correct the scanning speed of the tunable laser in real time (Fig. 4). In the industrial measurement environment, the optical path difference of the auxiliary interferometer as the measurement reference will change, so the accuracy of the optical path of the auxiliary interferometer needs to be strictly guaranteed. The optical path difference of the auxiliary interferometer is usually recalibrated regularly with a more accurate measurement reference, such as a gas absorption tank with hydrogen cyanide encapsulation (Fig. 6). For tasks requiring large scanning bandwidth or long distance measurement, the dispersion mismatch of the long fiber auxiliary interferometer will cause accuracy loss of several hundred microns, which needs to be compensated by efficient and fast numerical algorithms.Industrial field environmentwill cause fluctuation of measurement optical path difference. When two lasers are simultaneously tuned upward and downward in frequency, the distance measurement errors can be compensated because the dynamic errors induced by optical path fluctuation are equal but opposite (Fig. 9). In addition, the dynamic error can be compensated by constructing a common path velocimeter to monitor the optical path fluctuations and through a single frequency laser (Fig. 10 and 11).Conclusions and Prospects FSI distance measurement has formed a set of relatively mature theory and technology. FSI distance measurement can not only provide a high-precision, efficient and on-site traceable absolute distance measurement scheme to achieve spatial multi-path absolute distance measurement and multi-degree of freedom measurement, but also build a large-size spatial measurement and positioning system, such as the multilateral spatial coordinate positioning system and the spherical coordinate positioning system. Measurement instruments based on the principle of FSI distance measurement have been widely used in large advanced industrial equipment and scientific equipment manufacturing and monitoring process in the industrial measurement tasks. At present, the laser radar, which adopts photoelectric phase-locked loop and dual-laser optical frequency opposite scanning technology, has been widely used in industrial measurement. The absolute multiline technology, which based on the phase comparison method of auxiliary interferometer, has the highest FSI distance measurement uncertainty. In recent years, with the improving of the tunable laser and key optoelectronic devices, many new FSI distance measurement techniques, such as optical frequency resampling based on auxiliary interferometer and optical frequency comb, have been extensively and deeply studied. However there are some shortcomings in efficiency of extracting interferometry signal phase and frequency spectrum, stability of measurement datum, reliability of the dynamic error suppression, etc. It is hoped that with the continuous breakthrough of laser source and key technology, researchers can continuously improve the measurement performance index and further develop new large-size space measurement equipment, which will bring new breakthroughs for FSI distance measurement technology and its industrial measurement application.

Significance Light is an electromagnetic wave that carries information of multiple dimensions, such as intensity, phase, frequency, and polarization. The change in the intensity, spectrum, polarization, and other information as a result of the interaction between objects and the light field can reflect the material, morphology, and other characteristics of the objects. However, traditional photodetectors can only detect two-dimensional (2D) light intensity information. To perceive more dimensional optical field information, additional optical components and mechanical devices are required, which will result in issues such as the system’s large volume and complex structure. Many applications today, such as three-dimensional (3D) face recognition, automatic driving, and remote sensing, have an urgent need for miniaturization and lightweight optical systems, presenting a significant opportunity for the development of integrated optical field sensing systems.A spectrum is a valuable tool for object characterization and analysis, and it is widely used in food safety, environmental monitoring, biological imaging, archaeological exploration, and other fields. Traditional dispersive and interference spectrometers can provide ultra-fine spectral resolution as well as an ultra-wide spectral detection range. Traditional spectrometers, however, have limitations in situations where real-time spectral detection is required due to the presence of optical and mechanical moving parts with large volumes and weight. People anticipate that in the future, spectral sensing devices will be reduced to centimeters or even millimeters in size and will be integrated into smartphones, drones, and other microsystems.The measurement of light’s polarization state is important in fields, such as remote sensing, medical treatment, and optical communication. Traditional polarization measurement methods are divided into two types: division-of-time and division-of-amplitude. Division-of-time polarimeters measure the intensity of various polarization components by positioning a set of rotating waveplates and polarizers in front of the detector. This method often relies on mechanical rotating structures, resulting in slow measurement speed and reliability. Polarizing beam splitters are used in division-of-amplitude polarimeters to separate different polarization components into different detectors. Both methods have problems, such as large volume and complex measurement system structure.Many emerging technologies, such as autonomous vehicles, face recognition, robotics, and augmented reality, rely on 3D imaging techniques. There are two types of 3D imaging techniques: active and passive. Active methods typically necessitate structured illumination or scanning, which adds complexity, cost, and power consumption. Passive methods, which typically use multiple views, have limited accuracy and a high computation cost. 3D imaging techniques based on conventional optical elements are limited by high cost, large size, high power consumption, and complex systems for applications requiring compactness, integration, and portability.Metasurfaces are novel planar optical elements that can control the light field by deploying subwavelength artificial antennas on the surface. Subwavelength structures of metasurfaces, unlike traditional optical elements, can interact with the incident electromagnetic field, causing abrupt changes in optical parameters on the surface and breaking traditional optical elements' dependence on the propagation optical path. Because of this property, metasurfaces can modulate the amplitude, phase, polarization, and other properties of the light field within the subwavelength thickness in a very flexible and powerful way. As a result, metasurfaces open up new avenues for the miniaturization and integration of spectrometers, polarimeters, and depth information perception (Fig. 1). We review recent research on spectral, polarization, and depth information sensing based on metasurfaces in this paper.Progress The ability of metasurfaces to flexibly regulate the spectrum opens up a new avenue for the realization of integrated spectrum sensing systems. Metasurface-based spectrometers are classified into two types based on their operating principles: narrowband filtering and computational spectrometers. Narrowband filtering spectrometers use a single tunable narrowband filter or narrowband filter array to achieve spectral sampling (Figs. 2--4). Computational spectrometers do not require narrowband filters. The spectral response of the filters can be wide and random, which makes designing narrowband filter metasurfaces much easier. Computational spectrometers can extract the original spectrum from obtained signals using algorithms (Fig. 5).In recent years, researchers have proposed several types of metasurface-based polarimeters, including division-of-amplitude (Figs. 6--8), division-of-time (Fig. 9), detector-integrated (Fig. 10), and others (Fig. 11). Metasurface-based division-of-amplitude polarimeters use metal gratings, scatters, or metalenses to distinguish the light with different polarization components in space and measure the intensity of each polarization component with detectors. Division-of-time polarimeters are based on tunable metasurfaces, which replace the waveplates and polarizers in conventional polarimeter systems and can modulate the polarization state of the incident light. Detector-integrated polarimeters are built around metasurfaces that can convert different polarization components' light into different electrical signals. The polarization state of the incident light can be determined by measuring the intensity of photocurrents. In addition, recently proposed polarimeters based on metasurface polarizers, holograms, and other technologies are discussed.Metasurfaces’ flexible wavefront manipulation enables them to realize 3D imaging systems with a miniaturized form factor and improved performance, for both active and passive methods. Typical active 3D imaging techniques include the structured light method and the beam steering method. Structured illumination achieved by metasurfaces has a simplified optical system and a much larger field of view (Fig. 12). Beam steering realized by metasurfaces is flexible, has low power consumption and high steering speed, and can reduce size and weight of metasurfaces (Fig. 13). The use of metasurfaces in passive methods has the advantages of high compactness, semiconductor process compatibility, high accuracy, and the ability to detect more dimensions of the light field with the help of algorithms (Figs. 14--16).Conclusion and Prospects This paper introduces the sensing of multidimensional light fields, such as spectrum, polarization, and depth information using metasurfaces. Future research into the flexibility of metasurfaces, such as combining metasurface design and reconstruction algorithms with inverse design, end-to-end optimization, deep learning, and other computer technologies, is expected to result in a simultaneous perception of more dimensional light field information. With the in-depth understanding of metamaterial surface, the exploration of new metamaterial surface design, and the improvement of large-scale micro-nano processing technology, metasurface will have a bright application prospect in the field of lightweight integrated multidimensional light field perception.

Significance With the advent of the period of big data, deep learning is playing a very important role in daily life and scientific research, and it has been widely applied in image processing, speech recognition, autonomous driving, and other fields. Deep learning in the optical field, as a data-driven algorithm, can effectively improve computational efficiency and imaging quality, approaching or even breaking through physical limits. An artificial neural network is a basic form of imitating biological neurons and the working principle of the human brain to complete learning process of internal principle or to extract target features. Different kinds of complicated neural networks are put forward and geared to the demands of different application scenarios for the target that often requiring different network architectures. Furthermore, an optical neural network using photons as the medium can break through the limitations of a traditional electronic neural network and provide high speed and low loss advantages.Progress In this paper, we analyze the applications of deep learning in micro-nano structure design and spectral response prediction, holographic imaging application, optical sensing and imaging technology, new photon-driven neural network, and other directions in detail through examples. We also list the challenges existing in the combination of deep learning and optics, and the development directions of this field have prospected. First, predicting spectral response based on existing micro-nano structures is an important step in micro-nano optics. The neural network can predict the near-field electromagnetic response, far-field scattering mode, Poynting vector, and other physical quantities without time-consuming calculations, which indicates that data-driven deep learning can simulate the propagation and distribution of electromagnetic fields ( Fig. 1). The use of neural networks can significantly improve prediction speed by several orders of magnitude. The process of designing micro-nano structures to get the ideal resonance response is called inverse design. Most early applications are designed for fixed micro-nano structures with some structural parameters as variables ( Fig. 2). There are non-unique solutions in the inverse design of micro-nano structures, which cause that the neural network is hard to converge. To solve the problem, a cascaded neural network is adopted, and deep learning is used in other complex response predictions for micro-nano structure ( Fig. 3). In addition, a generative adversarial network (GAN) can be used to improve the generalization and diversity of inverse design ( Fig. 4), but its shortcoming is poor interpretability. GAN can be upgraded to a multi-layer GAN system in the following work, which can predict more structural parameters and have various functions ( Fig. 5). Optimization algorithms commonly used in the inverse design include gradient optimization algorithms, genetic algorithms, etc. As a powerful optimization method, deep learning combined with topology is a good development prospect in micro-nano inverse design ( Fig. 7). Second, deep learning can learn rules or detailed features from data. The advantages of deep learning can replace conventional optimization algorithms to improve image quality and work efficiently, especially in holographic imaging. Deep learning is widely used in phase recovery and image reconstruction of holograms (Fig. 8). Compared with the conventional way of generating computer-generated hologram (CGH), deep learning can improve the quality and computing speed of CGH, and ultra-lightweight real-time 3D holography can be implemented (Fig. 9).Moreover, optical sensing and computational imaging techniques based on deep learning are a hot topic. Computational imaging technique such as single-pixel imaging based on traditional algorithms often requires scanning acquisition and calculation, which consumes a lot of computing time and power. In the case of a low sampling rate, its imaging quality cannot be guaranteed. With the advantages of data-driven deep learning, the neural network adds new vitality to single-pixel imaging technology (Figs. 11--13). Unlike computational imaging, optical sensing imaging technology employs laser-active lighting to collect optical signals via optical sensors or receivers and reproduce object information via target signal processing. Combining deep learning with non-visual imaging and replacing a single optical model with a data-driven method frequently results in high speed and adaptability, as illustrated (Figs. 15--16). Deep learning also has good applications in other optical sensing imaging fields, such as microscopic imaging, three-dimensional imaging, fusion imaging, etc.As a kind of electromagnetic wave, light is different from an electrical signal in the propagation process and is not interfered by an external electromagnetic field. Meanwhile, as a carrier of information transmission, light has high bandwidth, high transmission speed, and low loss. Researchers began to consider using light to replace electrons to establish a neural network in the optical field. A diffractive all-optical neural network can be used in image classification, optical logic operation, spectral aiming to classification, and other tasks (Figs. 18--23). Deep learning has also been applied to other optical fields, such as optical cloaking, optical anti-counterfeiting, multichannel modeling, etc.Conclusions and Prospects In conclusion, there are still some challenges in the combination of optics and deep learning algorithms. Here we put forward several prospects. First, physical models are incorporated into the deep learning computing process to increase the interpretability of the models to reduce the network’s dependence on data. Second, the combination of experimental and simulation data is necessary. Expanding the training set by using multiple algorithms to generate new data from real data can reduce the difficulty of training. Third, the combination of deep learning and traditional optimization algorithms can complement each other to a certain extent, improving the performance of the network model and greatly shorten the computing time. It is also beneficial to enhance the generalization and robustness of the network. Finally, an optical diffractive neural network breaks through the limitations of traditional electronic neural networks. But currently, most of the all-optical neural network is only for a single task, and it has low stability on-chip photonic neural network and other problems. More attempts and studies are needed, and there is still plenty of room for improvement.

Significance Optical imaging of biological tissue has become a powerful tool for biologists and clinicians. Photoacoustic imaging (PAI) is an acoustic-mediated optical imaging method that uses the photoacoustic effect to offer a highly sensitive and abundant optical contrast over a wide spatial and spectral range. PAI has achieved remarkable success in blood oxygen saturation imaging, brain vasculature functional imaging, histology-like tissue imaging, and so on. In most PAI systems, piezoelectric transducers (PZT) and coupled water are essential. However, these contact-configurations limit the application fields of PAI. Furthermore, PZT is bulky for system design, and it can degrade imaging performance.Non-contact PAI is one of the most important directions of PAI, having the potential to realize better imaging performance and a wider range of applications without water coupling. Non-contact PAI is especially suitable for PA ophthalmic imaging, intraoperative margin diagnosis, and burn diagnostics, which can avoid infection and discomfort. Meanwhile, the all-optical non-contact PA detection method can obtain a wider bandwidth and angular coverage. Minute-sized optical detection configuration makes for unblocked excitation and the multimodal design. Hence, it is of great significance to introduce and summarize the recent research in non-contact PAI and get an insight onto the characteristic application of the non-contact PAI technique in biomedical imaging.Progress Various non-contact methods on interference and noninterference have been developed to implement a non-contact PAI. Air-coupled PAI based on a special low-frequency transducer has played a significant role in non-contact PAI. Low frequency contributes to less attenuation of PA waves. Deán-Ben et al. from Technical University Munich proposed an air-coupled PAI system using a homemade transducer with 800 kHz center frequency, which realized acoustic-resolution imaging [Fig.2 (a)]. Ma et al. from South China Normal University proposed an optical-resolution imaging system. Burned rabbit’s skin imaging [Fig. 2(b)] has been implemented to show that the non-contact PAI technique can be valuable in the adjuvant diagnosis and observation of burns. A frequency-domain PA detection based on a PA cell can realize non-contact imaging and spectral measurement [Fig.2 (c)].All-optical methods were also used to achieve a higher sensitivity for non-contact PAI, including interferometric and noninterferometric method. Interferometric methods can be divided into homodyne, heterodyne, and speckle modes. We use the interferometer to detect the phase difference, which is the result of the pressure. In homodyne mode, Wang et al. from the University of Washington reported a non-contact PAM (PA microscopy) system in which a low-coherence interferometer was utilized. Yang Sihua’s team from South China Normal University proposed a PA-optical coherence tomography (PA-OCT) dual-modal system, which could provide complementary anatomical and functional information for imaging of biological tissues [Fig. 3(b)]. Based on the all-optical PA doppler effect, blood flow imaging was proposed to significantly broaden the scope of applications for obtaining the blood flow velocity of the microvasculature in biomedicine [Fig. 3(c)]. To adapt to the uneven surface, Hu et al. reported an extended depth-of-field non-contact PAM system using a dual nondiffracting Bessel beam. This system could image nonfat tissues with a high resolution (2.4 μm) and large depth-of-field (635 μm) [Fig. 3(e)], cooperated with a long coherence source. Furthermore, a 3×3 fiber coupler was used to eliminate phase noise and maintain phase stabilization. Wang et al. from Northeastern University demonstrated a 3×3 coupler-based fiber-optic interferometric system to detect local initial PA pressure. This method is fully non-contact and convenient for in vivo imaging [Fig. 4(d)]. In heterodyne mode, an acousto-optic modulator and IQ demodulator are needed for PA detection. Moreover, Eom et al. presented three-dimensional in vivo PA images of the blood vasculature of a chicken chorioallantois membrane obtained using a fiber-based non-contact PA tomography system [Fig. 5(a)]. Tian et al. proposed a non-contact PAI system using circular scan geometry, suggesting that the heterodyne interferometer could be potentially used in biomedical imaging. In speckle mode, the system configuration is simple, and it does not even need the reference light. Buj et al. developed a fast innovative holographic off-axis non-contact detection method for PAI, successfully imaging tissue phantoms with an embedded complex absorber structure, which imitated vascular networks [Fig. 6(a)]. Some methods based on correlation analysis of speckle were also utilized for imaging.For the noninterferometric method, photoacoustic remote sensing (PARS) was first introduced by Hajireza et al. in 2017. In PARS, elasto-optical refractive index changes due to the transients of the PA initial pressure producing a significant time-varying reflection of a probe beam. Based on this mechanism, PA images can be obtained without any coherence noise (Fig. 7). The reason is that PARS has better sensitivity than traditional PAM in all-optical nature. PARS has been utilized for blood oxygen saturation imaging, virtual histology imaging, and ophthalmic imaging.Furthermore, non-contact PAI has shown broad applications in the life sciences, especially in intraoperative margin diagnosis, ophthalmic imaging,and optical biopsy of cancer cells. Various tissue blocks were imaged by PARS with 266 nm excitation, and the results were highly consistent with H&E stained images (Fig. 8). A non-contact ophthalmic PA-OCT imaging could provide complementary structural and functional information of the eye (Fig. 9). Moreover, Zhou et al. demonstrated a preclinical device, all-optically integrated PA and OCT (AOPA/OCT), which can simultaneously provide label-free biomarkers of vascular patterns, temporal and spatial heterogeneity of blood flow, and tissue micro-structure changes during tumor growth with pathophysiological correlations in mice models (Fig. 10). However, these methods and systems have their advantages and disadvantages (Table 1).Conclusions and Prospects Air-coupled, interferometric, and noninterferometric methods provide new ideas and technical strategies for non-contact PAI. Therefore, a clinical transformation that uses different technical characteristics is the focus of the exploration. As technology continues to evolve, non-contact methods are expected to replace traditional contact methods, making PAI a more attractive tool for biomedical applications.

Objective Recently, mimicking the quantum electromagnetically induced transparency (EIT) effect using metamaterials in a classical way has attracted continuous attention. Achieving an active EIT effect is one of the important research directions owing to its great potential in many practical applications, such as active light switching and high-speed slow light modulation. So far, a variety of new working schemes have been proposed by integrating functional materials into the metamaterial structures, such as nonlinear media and photoactive and electroactive semiconductors. Graphene, composed of single-layer carbon atoms, exhibits excellent electrical and optical properties and the dynamic tuning of its optical conductivity is achieved by tuning its Fermi level (EF) and carrier scattering time (τ). Based on this, graphene-based active metamaterials have successfully exhibited their potentials in light modulation in which high modulation speed is shown owing to the picosecond-level relaxation time of graphene. However, the previous studies mostly rely on tuning EF to achieve the active control and the studies by tuning τ are relatively scarce. In this work, we theoretically proposed an active EIT device in the terahertz regime using graphene-metal hybrid metamaterials, in which we tune τ using the nonlinear effect of graphene under strong-field terahertz incidence. Owing to the ultrafast relaxation time of the carriers in graphene, such a nonlinear modulation route paves the way towards ultra-fast active devices.Methods The proposed active EIT metamaterial is composed of graphene-metal hybrid structures on the silicon substrate. The metal structure part is composed of meanderline resonator (MLR) and split ring resonators (SRRs), as shown in Fig. 1. The two SRRs are placed vertically and symmetrically inside the MLR. The geometric parameters of the metal structure are: L1=85 μm, L2=75 μm, d =12.5 μm, l1=29 μm, l2=25 μm, w=6 μm, s=7 μm, g =0.5 μm, and D =53.5 μm, respectively. The period is P=100 μm, and the thicknesses of the metal and silicon substrate layers are 200 nm and 640 μm, respectively. The graphene structures here are designed to locate only in the gaps of the two SRRs that connect the gap end. In order to study the active EIT effect, the finite-difference time-domain (FDTD) method is applied to simulate the transmission spectra. In the simulation, the excitation source is a plane wave propagating along the z direction, the boundary conditions along the x and y directions are periodic while those along the z direction are open boundary conditions, the substrate is set as lossless silicon (ε =11.78), and the metal is set as aluminum with a conductivity of 3.72×10 7S·m -1. Results and Discussions When the graphene structures are not presented, the metal structure can exhibit a strong EIT effect under y-polarized incidence, as shown in Fig. 2, where the MLR and SRRs function as bright mode and dark mode, respectively. In order to study the nonlinear EIT modulation effect when the graphene structures are presented, the condition of strong-field terahertz incidence (~300 kV·cm -1) is mimicked by changing the relevant graphene parameter τ in simulation according to the previously reported nonlinear behavior of graphene. The EF of graphene is fixed to be 0.15 eV, and the τ is increased from 1 fs to 13 fs (corresponding to gradually decreased terahertz field). Figure 3(a) shows the corresponding simulated transmission spectra. It can be seen that as τ increases, the overall resonance behavior gradually changes to the situation when there is only the MLR. In order to reveal the physical mechanism of the active EIT modulation, a coupled-mode theory is used to quantitatively describe the changing behavior. Figure 3(b) shows the fitted transmission spectra, which are in good agreement with the simulated results. Figure 5 shows the corresponding fitting parameters as a function of τ. It can be seen that the parameters γ1, δ, and κ basically remain unchanged, while the damping rate γ2 of the dark mode resonator obviously increases. This can be attributed to the enhanced shorting effect of the graphene structures and to the resonance of the SRRs due to the increased graphene conductivity. In addition, the influence of the gap size g of SRRs on the active EIT effect is also studied, as shown in Figs. 6(a) and (b). Figure 6(c) shows the field enhancement factor at the center of the SRR gap and the corresponding nonlinear modulation depth at different g, which both decrease as g increases. When g is small, under strong terahertz field incidence, the large field enhancement effect can greatly reduce the graphene conductivity and contributes to a strong EIT effect, while under weak terahertz field incidence, the opposite carriers at the two gap ends can easily recombine and contribute to the disappearance of the EIT effect. Thus, the modulation depth becomes larger as g decreases. Here, at g=0.5 μm, the field enhancement factor reaches 360.7 and the nonlinear modulation depth reaches 49.3%. Conclusions In summary, a nonlinear EIT modulation effect in a composite metamaterial composed of graphene-metal hybrid structures has been studied in the terahertz regime. The inner mechanism lies in the combination of the field enhancement effect and the nonlinear effect of the graphene conductivity under a strong terahertz field. Owing to the ultrafast carrier relaxation time of graphene, the nonlinear modulation speed here is thus determined by the relaxation time of the structure resonances which is in the dozens of picoseconds level. The proposed metamaterials may have potential applications in high-speed slow light modulation and optical switching. And the proposed nonlinear modulation method provides a new way towards high-speed active devices.

Objective Functional near-infrared spectroscopy (fNIRS) has several advantages, such as noninvasiveness, free radiation, and reasonable temporal/spatial resolution. This enables fNIRS-based technologies to be used as an alternative to conventional technologies, such as functional magnetic response imaging (fMRI) and electroencephalogram (EEG), and the technologies are increasingly used in clinical practice to complete neuroimaging. However, because of the reflection geometry used in fNIRS, light travels from a source, through the scalp-skull layer, into the brain, and back out through the scalp-skull layer to be measured by a detector, which decays significantly as depth increases. Therefore, the reconstructed activation using fNIRS is usually contaminated by superficial physiological signals (cardiac pulsation, respiration, and low-frequency oscillations, etc.). Besides, the random interferences induced by the photon-shot and instrumental noises, etc., also have blurring effects on the activation reconstruction because of faint activated hemodynamics in the brain. Thus, suppressing the irritating physiological interferences and random noises has been a critical task in fNIRS-based neuroimaging. In this work, we propose a long-short-term-memory (LSTM) based recurrent neural network (RNN), including a prediction and a classification layer, to suppress physiological interferences and random noises, respectively, to improve reconstruction performance with less repetitive or even individual stimulation. This has some advantages, including shorter measurement time, more subjects, and the ability to examine responses to single stimulation.Methods The proposed LSTM-based RNN, which is purely data-driven without an auxiliary measurement process, comprises two layers: First, the prediction layer is used to estimate the absorption perturbation induced by the physiological interferences during task stimulation. Then, the estimated time series is used as the reference to adaptively filter the reconstructed absorption perturbation for the removal of the interferences from the physiological signals. Second, the classification layer is applied to reduce the remaining artifacts induced by the random noises in measurements for acquiring a better space-localized solution, converting the filtering procedure to a binary classification problem. Notably, the combination of the space-time filtered results from the predication layer is used as the input to the classification layer, ensuring the robustness and efficiency of the proposed method.Results and Discussions The numerical simulations and in-vivo experiments are implemented based on fNIRS--DOT (diffuse optical tomography) to describe the network design, training, and filtering process in detail , and the effectiveness of the proposed method is compared with the reference filtering and cycle averaging method (RFCA). The results show that the proposed LSTM-based model improves reconstruction performance for the numerical simulations (Fig. 6) by effectively suppressing the physiological interferences and random noises rather than using more measurement cycles. Furthermore, we examine the effectiveness of the proposed method to deal with the potential time lags of superficial interferences compared with those in the cerebral cortex layers, and the results show that the proposed method performs better under the mentioned condition (Fig. 10). As for the in-vivo experiments, the results from the predication layer show comparable performance as the RFCA, whereas the results from the classification layer show a more concentrated activated region (Fig. 9). Because other modality imaging techniques have not been used to cross verify the activated region, determining whether the proposed model is over-optimized for the activated region is difficult. Thus, training the filtering model to avoid this problem will be an important direction for future work.Conclusions In this paper, we propose a two-layer LSTM-based RNN that utilizes the prediction and classification of the RNN model to reduce the image artifacts induced by the physiological interferences and random noises in fNIRS-based neuroimaging. The proposed method has a clear physical explanation and needs no additional hardware cost. To evaluate the proposed method, a series of preliminary numerical simulations and in-vivo experiments were implemented, and the results show that it has a promising future for achieving reasonable enhancements, providing a practical approach for the fNIRS-based brain-computer interface application.

Significance Optical computing has been proposed for a few decades, although not yet widely applied in practice. This is partially because large-scale electronic circuits have been successfully developed as universal computing platforms. In recent years, it has been witnessed that the Moore’s law faces a bottleneck and photonic chips exhibit increasingly larger integrated arrays of tiny devices. Meanwhile, the emerging artificial intelligence has been inspiring a ubiquitous interest, which is featured by large amounts of matrix computation. This particularly triggers a renewed interest in optical neural networks for voice/image recognition, channel equalization in communications, and other data processing applications.Compared with electronic neural networks, photonic neural networks potentially have the advantages of high speed and low power consumption. As a result, it has gradually attracted people's research interests in recent years. The photonic reservoir neural network is a kind of photonic recurrent neural network. Reservoir computing is expected to be suitable for processing sequence signals, and the training process is relatively simple. This could be greatly useful for optical fiber communications and wireless mobile communications.This paper first introduces in detail the system configurations and technological characteristics of reservoir computing and presents the essential conditions to realize a reservoir. Then, the research progress in photonic reservoir computing is introduced through two different hardware implementations, which are called parallel and delay-based structures. Finally, the bottlenecks and corresponding solutions are discussed.Progress Parallel photonic reservoirs are composed of optical node arrays. They have the potential to perform large-scale parallel computing. In 2008, researchers proposed a parallel photonic reservoir using semiconductor optical amplifiers. It outperformed traditional reservoirs on signal classification tasks at that time. This research team also reported a power-efficient experimental prototype in 2014, which only consists of passive waveguides, splitters, and combiners (Fig. 5). Microring resonators and photonic crystal cavities can also be used as nodes in parallel photonic reservoirs. Another kind of parallel optical reservoir is based on space optics. Maktoobi et al. demonstrated a reservoir with diffractively coupled nodes (Fig. 7), Rafayelyan et al. reported a reservoir based on multiple light scattering (Fig. 8), and Paudel et al. demonstrated a reservoir using speckles generated by mode interference in a multimode waveguide.A delay-based photonic reservoir, also called a serial photonic reservoir, contains a single optical node with time-delayed feedback. Delay-based photonic reservoirs are easier to manufacture than parallel photonic reservoirs, but their parallel-computing capability is slightly poor. In 2012, Larger et al. demonstrated an optoelectronic delay-based photonic reservoir using a Mach-Zehnder modulator as the nonlinear node. Duport et al. reported the implementation of a photonic reservoir based on a semiconductor optical amplifier in the same year and it is the first delay-based all-optical reservoir (Fig. 11). In 2013, Brunner et al. used a semiconductor laser as the nonlinear node in a delay-based reservoir. Semiconductor lasers are power efficient, high-bandwidth, and widely used in modern fiber communications. In 2014, Dejonckheere et al. used a semiconductor saturable absorber mirror as a nonlinear node. It is the first photonic reservoir using fully passive nonlinearity.There are schemes to improve the performance of reservoirs through a so-called hybrid configuration. In 2021, Nakajima et al. reported a photonic reservoir consisting of several delay-based reservoirs connected in parallel. The nodes are on-chip passive coherent cavities. This experiment realized the first image classification using an on-chip passive photonic reservoir.By comprehensively comparing the recently proposed photonic reservoir computing schemes, we show a few features and evaluators, which can be used to estimate the capacity of new reservoir computing systems, including node type, nonlinearity mechanism, optical delay, array size, and input optical power. By organizing three tables (Tables 1--3), we clearly show the technical advantages and disadvantages of different reservoir neural network configurations, with an emphasis on the practical characteristics of nonlinear actuation functions. The nonlinear functional devices are based on semiconductor optical amplifiers, nonlinear microresonators, optical lasers with feedback, semiconductor saturable absorbers, and Mach-Zehnder modulators with a nonlinear transfer function.Conclusions and Prospects Photonic reservoir neural networks can overcome some limitations of electronic neural networks, and their training processes are very simple. Photonic reservoirs have a broad development prospect. They have been used to implement speech recognition, chaotic time serie prediction, channel equalization, header recognition, and other functions.We share some high-level perspectives on the future directions of photonic reservoir computing systems, by pointing out the potential technical issues and problems competing with an electronic version of reservoir computing. The array size, speed and accuracy of computation, and all-optical processing capability have been identified as three major tasks to advance the future development of photonic reservoir computing. According to these, some representative works recently published have been discussed, and the hybrid configuration of photonic reservoirs is particularly analyzed. We believe that all-optical input and output, hybrid configuration, on-chip implementation, and large-scale reservoirs are the future development directions of photonic reservoirs.We believe this review would be of interest to the community of optical computing and neural networks as well as the community of integrated photonics.

Significance An optical fibre surface plasmon resonance (SPR) sensor is one of the new type optical fibre sensors developed on the basis of optical fibre and sensing technologies. It has the inherent advantages of optical fibre and SPR sensors, such as high detection sensitivity and free labelling. Currently, it is used to measure physical parameters and detect chemical substances and biomass. Optical fibre SPR sensors make up for the shortcomings of conventional electrochemical sensors for substance detection and have a wide range of applications in the fields of chemistry, biology, medicine, and food safety.With the rapid development of optical fibre SPR sensing technology, researchers have reviewed the current state of development in areas such as fibre SPR sensor structures, fibre substrates, and film materials. However, with the development and increased applications of fibre optic SPR sensors, sensitivity enhancement has gradually become one of the important parameters that limit the sensor’s performance. Up to date, there is no review report on this important issue. Based on the relevant literature, this paper reviewed the sensitivity-enhancement technology of fibre SPR sensing, summarised the sensitivity-enhancement methods from the perspectives of fibre substrate structure, film structure, and film material, and prospected the future development direction of these sensors.Progress The intensity of evanescent waves can be effectively enhanced by designing different fibre substrate structures. The most commonly used evanescent-wave enhanced fibre substrate structures are the D-type, U-type, tapered-type, and fibre core mismatch structures. The D-type structure is the side polishing structure, which polishes one side of the optical fibre to a plane, making it easier for the sensing film to be deposited on the surface of the core. The U-shaped structure fibre SPR sensor has a bent sensing area, which reduces the incident angle of the optical signal transmitted in the sensing area and enhances the penetration depth, leading to an enhanced evanescent field and improved sensing sensitivity. The sensing area of an optical fibre SPR sensor with a tapered-type structure is sandwiched between two tapered fibre regions. Core mismatched fibre SPR sensors mainly include multimode fibre (MMF)-single mode fibre (SMF)-MMF, SMF-MMF-SMF, SMF-hollow-fibre-SMF, SMF-photonic-crystal-SMF, and other types of structures. In addition, the sensitivity of SPR sensors can be improved using new types of optical fibres such as photonic crystal or micro-structured fibres as sensing substrates.The types of metal films commonly used in optical fibre SPR sensors include gold, silver, copper, and aluminium. The thickness of these metal films is generally about dozens of nanometres. Single-layer silver film SPR sensors have high resolution and sensitivity. However, the gold film is generally used in SPR sensors owing to the chemical activity of silver. The fibre optic SPR sensor with a silver-gold bimetallic film structure has a layer of gold film on the surface of the silver film, which makes the sensor have the advantages of high sensitivity and high resolution and enables the sensor to make full use of the advantages of stable chemical characteristics of the gold film. Because metal nanoparticles, particularly gold and silver nanoparticles, have strong plasma and catalytic effects, they are often used in the fabrication of optical devices. For some low molecular weight substances to be measured, metal nanoparticles can be combined with such substances, which amplifies the SPR signal and further improves the sensing sensitivity. Compared with other nanoparticles, gold nanoparticles have better biocompatibility and unique optical and electrical properties and are easy to prepare.The surface of a common single metal film SPR sensor is relatively active, and it is easy for the sensor to react with the solution to be tested during the detection process, resulting in a reduction of the sensitivity. This problem can be solved by adding additional film materials on the surface of the metal film, which will improve the performance of the SPR sensor. As an important sensitive material, metal oxides are widely used. When the metal oxide film is used as the auxiliary film layer, the electric field distribution in the sensing film layer changes and the electric field intensity is enhanced. Thus, the sensitivity of the optical fibre SPR sensor can be effectively improved. Common auxiliary metal oxide films mainly include zinc oxide, indium tin oxide, and titanium dioxide. Transition metal dichalcogenides (TMDCs) have attracted considerable attention owing to their unique optical, electrical, and electrochemical properties. By chemical stripping, high-quality single or multilayer TMDC films can be prepared, which can be applied to optical fibre SPR sensors to improve their sensing performance.Conclusions and Prospects Currently, the sensitivity-enhancement technologies have been used for the optical fibre SPR sensors. Furthermore, the optical fibre structure, film structure, and film material are focused on, which helps in effectively improving the fibre SPR sensor refractive index sensing sensitivity and providing practical and effective technical support for future development. Further breakthroughs are expected in the following areas. First, the sensitivity of the fibre SPR sensor can be improved using different sensitivity-enhancement methods, such as using optical fibre structure, film structure, and film material. Second, according to the target characteristics of the required sensor, the dielectric constant of the film material can be inversely calculated using reverse design and the composition and lattice structure design of the new thin-film material can be realized. Finally, by increasing the SPR coupling efficiency, the two-photon 3D micro-nano printing technology can be used to further improve the sensitivity.

Objective Spatial, oceanic and geological exploration technologies have increasingly become important directions in our country development. In order to meet the demand of accurate detections, it is necessary to obtain basic physical information (e.g. temperature and strain). However, the requirement for sensors in these complex environments is extremely stringent. Compared with traditional electromagnetic sensors, optical fiber sensors have the characteristics of small size, light weight, corrosion resistance, anti-electromagnetic interference, etc. For optical fiber mechanical and thermal sensing, there are two common methods. One is using the fiber-Bragg-grating-based sensing structure which obtains the mechanical and thermal parameters through the spectral information of the reflected light. The other is using a combination of different sensing structures such as the combination of long-period fiber gratings (LPFG) with photonic crystal fiber (PCF). The above two methods both have shortcomings such as small measurement range, large measurement error, complex structure, and high cost. In response to the above problems, a fiber Bragg grating for temperature and strain multi-parameter sensing system based on a tunable laser is theoretically proposed and environmentally tested.Methods In this study, the system consists of a tunable laser, a Fabry-Perot etalon, a driving circuit, a beam splitter, fiber Bragg grating temperature and strain sensors, photodetectors and a data acquisition card. After passing through the coupler, the light output from the tunable laser is divided into two paths, which respectively enter the multi-channel sensor and the etalon. The light entering the multi-channel sensor passes through the beam splitter and enters the 16 sensing channels. The light reflected by the sensor and the light passed through F-P etalon are transmitted to the photodetector and converted into electrical signals. The driving circuit generates a square wave signal synchronized with the triangular wave, which is used as the trigger signal of data acquisition. It controls the acquisition card and the field programmable gate array (FPGA) data processing module to process the electrical signal and demodulate the temperature and strain value according to the relationship between the wavelength and the mechanical and thermal parameters. In the aspect of sensor, we utilize a special packaging structure. The fiber optic temperature sensor is packaged with a ceramic tube. The grating coated with a high temperature resistant polyimide is used as the sensing element. The fiber grating is bonded with the outer layer of an alumina ceramic tube by using low melting point glass. This method avoids the problem of adhesive aging at low temperatures, so the temperature sensor can be applied to a wider temperature range, and the stability of the sensor is improved. The fiber optic strain sensor is composed of a metal substrate, a fiber Bragg grating and a sleeve. A desensitized substrate is used to protect the strain sensing grating. In addition, the 316L stainless steel is used as the base material of the strain sensor. The material has corrosion resistance. Besides, the thermal expansion coefficient of the 316L stainless steel is close to that of the measured structure in engineering applications.These characteristics can further improve the accuracy of strain measurements. In the manufacturing process, the metal substrate should be polished and wiped with ethanol to remove foreign matters on the surface. The fiber Bragg grating is welded with the metal substrate by using low melting point glass in the pre-stretched state. After the substrate is cooled, the epoxy adhesive with good temperature adaptability is used for further fixation. This method improves the temperature adaptability of the sensor while ensuring the accuracy of strain measurements.Results and Discussions It can be seen that in the range of 0-200 ℃, the relationship between temperature and wavelength has a good linear relationship (Fig. 6). The sensitivity of the temperature sensor is about 11.60 pm/℃. The wavelength resolution of the optical fiber mechanical and thermal sensing instrument is 1 pm, so the demodulation sensitivity of the corresponding temperature sensing system is about 0.09 ℃. In the range of temperature below zero, temperature and wavelength quadratic polynomial fitting can improve the demodulation accuracy of the temperature sensing system. At each temperature node, the measurement error of the temperature sensing system is less than ±0.8 ℃ (Fig. 8). For the strain sensing, the FBG strain sensor can still work normally in the whole temperature range of -252.75--200.94 ℃. The sensitivity of the strain sensor is 1.66 pm/με, and the average measurement error is less than 2.9 με (Fig. 7). The experimental results show that the proposed sensor system has good accuracy and stability.Conclusions This study describes a wide range and high-precision optical fiber temperature and strain sensing system based on a tunable laser. Besides, the fabrication method of an optical fiber temperature and strain sensor is improved to enhance the temperature adaptation range. Finally, the instrument development is carried out, and the performances in high and low temperature environments are tested and analyzed. The experimental results show that the system can realize accurate temperature and strain measurements in the temperature range of -252.75--200.94 ℃. The temperature measurement accuracy is less than ±0.80 ℃ and the strain measurement accuracy is less than ±2.90 με. In addition, the system can realize a multi-channel and multi-parameter measurement at the same time, which is suitable for an engineering application in special environments with high and low temperatures.

Objective This paper proposes a novel trench and crosses airhole-assisted multicore few-mode microstructured optical fiber (TCAH-MC-FM-MOF) to meet the demand for space-division multiplexing system and mode-division multiplexing system for large-capacity, multichannel communication fibers, the structural parameters of which are optimized using the finite element method (FEM). After optimization, the designed fiber can support the stable transmission of LP01, LP11, LP21, LP02, and LP31 modes at the operating wavelength of 1550 nm, and the effective mode fields are 113.14, 159.70, 174.43, 104.91 and 192.74 μm 2, respectively. The intercore crosstalk of these five modes is less than -40 dB, and the relative core multiplexing factor is 62.722. Compared with its existing counterparts, this fiber has lower crosstalk and a larger mode effective field area. It is expected to meet the needs of large-capacity and multichannel transmission of the communication systems. Methods This paper proposes TCAH-MC-FM-MOF as a good candidate for large-capacity, multichannel communication fibers. The cross-section and refractive index profile are shown in Fig. 1. The FEM optimizes the fiber structure to achieve the best performance based on the mode and power coupling theory. The intercore crosstalk formula of the four-core optical fiber is derived via theoretical analysis for a more accurate crosstalk calculation. The relationship between multiple structural parameters and fiber performance is exploited to achieve low intercore crosstalk and large field area. The initial fiber parameters are continuously optimized, and a set of satisfactory structural parameters is listed in Table 2. To demonstrate the advantages of TCAH-MC-FM-MOF designed in this paper, the performance of four types of multicore and few-mode fibers with different structures is compared by evaluating the intercore crosstalk LP31 at the transmission distance of 100 km at 1550 nm. The results demonstrate that TACH-FM-MCF-MOF has the lowest crosstalk value and the best performance, as shown in Fig. 7.Results and Discussions Achieving low crosstalk and a large mode field area in multicore and the few-mode microstructured optical fiber is critical for improving transmission capacity and overcoming nonlinear effects. The core size, core spacing, and doping concentration are adjusted to achieve the best performance under the premise of ensuring 5-LP mode transmission. Low refractive index grooves are added around the core to prevent beam leakage, and the width of the grooves is optimized to prevent crosstalk between the cores, as shown in Fig. 3 (c). As shown in Figs. 4 (a) and 4 (d), the core size and core doping concentration are appropriately selected to achieve a large mode field area. After optimizing the structural parameters, the simulation demonstrates that the designed TCAH-MC-FM-MOF has low crosstalk, a large mode area, and good bending resistance, with a relative core reuse factor of 62.722.Conclusions TCAH-MC-FM-MOF proposed in this paper exhibits the characteristics of low crosstalk, large mode field area, and good bending resistance. When transmitting 10 km at 1550 nm, the designed TCAH-MC-FM-MOF has its intercore crosstalk of all modes suppressed less than 40 dB, and the effective mode field area greater than 100 μm 2. The effective refractive index difference of all the 5-LP modes meets the weak coupling condition, the crosstalk between modes can be ignored, and the relative core reuse factor is 62.722. Compared with other kinds of multicore few-mode fiber structures also highlights TCAH-MC-FM-MOF’s advantages in suppressing interphase crosstalk and alleviating the restrictive relationship between low crosstalk and large mode field area. Combined with SDM and MDM technology, the proposed TCAH-MC-FM-MOF is expected to meet the urgent demand for large-capacity, multichannel transmission systems.

Significance Since the first demonstration of the Kerr-lens-mode-locked Ti:sapphire laser, femtosecond laser technology has attracted tremendous research interest and evolved very rapidly. Thanks to the properties of short pulse duration, broadband spectrum, and high peak power, femtosecond laser pulses can probe the high-resolution dynamics in both time and spatial dimensions, and explore new regimes of light-matter interaction. Contributing to these advantages, femtosecond laser systems could serve as powerful and reliable platforms for many cutting-edge applications, such as material processing, frequency comb generation, metrology, microscopy, spectroscopy, and nanooptics. Apart from many application fields, femtosecond laser technology has led to many breakthroughs in fundamental research fields, including attoscience, femtochemistry, and nonlinear optics. Developments in pump diodes, gain media, and saturable absorber mechanisms advance the frontiers of pulse duration and output power. Up to now, extremely short duration of pulses down to a few-optical-cycles can be achieved both directly from the oscillator and nonlinear processes outside the cavity. On the other hand, the output power level of the femtosecond laser system can reach several hundred watts. In recognition of the role of the femtosecond laser technique, Mourou and Strickland won the Nobel Prize in 2018 for chirped-pulse amplification. Apart from advancement to shorter pulse duration and higher output power, more and more research focuses are placed on ongoing efforts to expand the frequency coverage to promote femtosecond laser systems into more widespread practical applications. However, the mode-locked spectral width of femtosecond laser output is limited by the effective laser gain bandwidth due to the relatively fixed energy levels of the gain medium, which hinders its large-scale application.Nonlinear frequency conversion techniques can provide the possibility to achieve effectively tunable laser sources in a wide spectral region. Up to date, the optical parametric oscillator (OPO) has emerged as a compelling alternative to generate broadband tunable radiation, which can expand the spectral region from the UV to infrared. Among them, OPOs pumped by femtosecond fiber lasers have been recognized as ideal platforms providing tunable ultrafast pulses with formidable performance, such as high repetition rate, high output power, and broad wavelength coverage. To this end, femtosecond OPOs are appealing for numerous applications, including quantum information, laser processing, optical frequency comb generation, and biophotonics. Recent power scaling of the Yb-fiber lasers and the development of new nonlinear crystals advance the frontiers of femtosecond OPOs.Progress To fulfill more widespread applications, there remains a strong motivation to expand the spectral tuning possibilities of OPOs. The development of birefringent crystals such as BIBO, BBO, and LBO, combined with a powerful femtosecond fiber laser source, enables the generation of tunable UV radiation on an ultrafast time scale (Fig. 1). Alternatively, thanks to the unique material properties of mid-infrared materials, i.e., CSP, OP-GAP, and ZGP, the operation of femtosecond OPOs in the far-infrared at 8 μm can be realized (Fig. 3).Kerr-lens-mode-locked Ti: sapphire lasers are the most commonly used pump sources for OPOs; however, these systems suffer a limitation in terms of power scaling mainly owing to unavoidable heat load in the laser crystal. In recent years, the rapid development of a high-power Yb-laser system allows a new power scaling potential for OPOs, and W level signal output can be achieved. However, it is rather difficult for OPOs to achieve a few-cycle pulse duration directly from a femtosecond fiber laser owing to the gain bandwidth limitation and complex nonlinear control. To access even shorter pulses from OPOs pumped by a fiber laser system, chirped-pulse optical parametric oscillators and self-compressed MIR OPOs have been demonstrated by researchers in Huazhong University of Science & Technology and Tianjin University, respectively (Fig. 7 and Fig. 9).For high-speed electrooptic sampling or future optical communication applications, moving operation regime of OPOs into the gigahertz pulse repetition rate regime has advantages. OPOs operating at GHz repetition rates have been reported using both synchronous and harmonic pumping schemes.Light emission with space-variant polarization and phase distribution has become a popular topic for the research community. The development of methods to create wavelength-tunable, space-variant polarization light beams will be a very interesting topic. Hu’s research group in Tianjin University has demonstrated novel femtosecond OPOs that deliver high-order Poincaré sphere beams, cylindrical vector beams, and vortex beams (Figs. 12--14).Conclusions and Prospect In this paper, we start with the progress in Yb-doped fiber laser-pumped femtosecond OPOs in recent years. Then, we present a variety of advanced designs of fiber laser-pumped OPOs, which are categorized into widely tunable OPOs, GHz repetition rate OPOs, few-cycle optical pulse OPOs, and structured beam OPOs. Finally, the applications of femtosecond OPOs in the fields of nanophotonics and Raman spectroscopy are introduced. With further development in nonlinear materials, combined with advances in pump laser technology, as well as new design concepts, femtosecond OPOs with wider spectral coverage, higher power, higher repetition rate, and shorter pulse duration are achievable in the near future. With the growth of novel femtosecond OPOs, completely new areas in application fields will arise.

Significance Mid-infrared (2--20 μm wavelength) photonics has extensive applications in spectroscopic analysis, environmental monitoring, medical diagnosis, free-space optical communication, and ranging, due to the distinguishable fundamental vibrational transitions of molecules and the atmospheric transmission windows (e.g. 2--2.5 μm, 3--5 μm, and 8--13 μm wavelengths) in the mid-infrared spectral region. Previously, mid-infrared applications have been mainly developed based on benchtop free-space optical instruments (e.g. Fourier-transform infrared spectrometers), which inevitably suffer from expensive, heavy, and bulky setups. To overcome this limitation, mid-infrared integrated optics has been proposed and quickly developed in the past few decades. By using the nanofabrication technology, on-chip mid-infrared devices not only significantly reduce footprints, weights, and costs of mid-infrared photonic systems, but also open an avenue to explore the light-matter interaction at the nanoscale level.Nowadays, numerous optical materials have been investigated to develop mid-infrared integrated optics, namely, noble metals, low-dimensional semiconductors, chalcogenide glasses, and group-IV semiconductors. As for noble metals and low-dimensional semiconductors, high optical losses of the developed waveguides hinder the potential large-scale integration of on-chip systems. While chalcogenide-glass-waveguides have attracted a great attention in many mid-infrared applications due to their ultra-low optical losses. However, the fabrication of the chalcogenide-glass-waveguides is not fully compatible with the complementary metal-oxide-semiconductor (CMOS) technology. On the other hand, photonic devices based on group-IV semiconductors, namely, silicon, germanium, tin, have the notable advantages of low optical loss, excellent physiochemical stability, and full CMOS compatibility, which are critical for practical applications with low-cost and high-volume production requirements. Consequently, mid-infrared group-IV photonics has been a hot topic in the past few years.As for the most commonly used group-IV semiconductors, silicon is first used to explore mid-infrared photonic integrated circuits. As early as 2006, Soref et al. published a paper to discuss the prospects of mid-infrared silicon photonics. Compared with the near-infrared band, silicon dioxide has huge optical absorption to the mid-infrared light, thus the silicon photonic devices utilized for the near-infrared band cannot be directly used in the mid-infrared band. Numerous novel silicon waveguide configurations, namely, suspended membrane waveguides, subwavelength-cladding waveguides, and silicon-on-sapphire waveguides, have been demonstrated. However, due to the strong multi-phonon absorption of silicon, the low-optical-loss spectral region of silicon photonic devices can only reach the functional group region (wavelengths below 8.0 μm). For silicon-germanium alloys, the photonic devices can be operated up to at least 8.5 μm wavelength. In contrast, for undoped crystal germanium material, optical absorption can be as low as 1 dB/cm within a spectral range from 1.9 μm to 16.7 μm at room temperature. Therefore, it is extremely promising to develop mid-infrared waveguides for long wavelengths based on a germanium platform.Progress Germanium possesses advantages of wide transparency window (2--14 μm wavelength), high refractive index (~4.0), an excellent thermal optic coefficient (>10-4 K-1), large third-order nonlinear susceptibility (~10-18 m2·V-2), and low cost for high-quality and high-density device fabrication. Therefore, germanium devices could be an excellent candidate to develop mid-infrared applications, especially in the fingerprint region. Since the first germanium waveguide was developed in 2012, mid-infrared germanium photonics has been attracting increasing research attention. Currently, germanium waveguides are mainly demonstrated based on four types of integration platforms, namely, germanium-on-silicon wafer, germanium-on-silicon-on-insulator wafer, germanium-on-insulator wafer, and germanium-on-silicon nitride wafer. Based on the above germanium platforms, researchers have not only developed state-of-the-art passive optical components on a chip, such as low optical loss waveguides, grating couplers, high quality-factor microring resonators, and photonic crystal nanocavities, but also demonstrated mid-infrared waveguide-integrated lasers and electro-optical modulators. Moreover, to extend the spectral range of on-chip sensing applications to the fingerprint region, researchers have developed diverse chip-integrated gas and protein sensors by using the germanium waveguide devices. Besides, nonlinear optical phenomena, namely Kerr frequency combs and supercontinuum generation, have also been theoretically explored in the germanium devices to overcome the spectral bandwidth limitation of mid-infrared on-chip lasers.Conclusion and Prospect In this paper, we briefly review the historical progress of mid-infrared group-IV photonics, and comprehensively summarize the development of recently emerging germanium photonics integrated circuits and their applications. In addition, the prospect of mid-infrared integrated optics is discussed. We hope this paper can not only serve as a reference for researchers specialized in mid-infrared photonics, silicon photonics, germanium photonics, optoelectronic materials, optical sensing, and spectroscopy, but also arouse attentions of researchers to mid-infrared integrated optoelectronics.

Objective High-performance Q-switched mode-locked lasers can achieve high pulse-repetition-frequency (PRF) ultrashort pulse sequences with nanosecond-pulse envelopes, which are of great importance in applications such as laser remote sensing, adaptive optics, and inertial confinement fusion. Because of the clean-up effect, stimulated Raman scattering (SRS) has been regarded as a potential technical approach to achieve high-performance laser output. In particular, with the discovery of the SRS self-mode-locking phenomenon in recent years, Q-switched self-mode-locked Raman lasers with compact structure, high peak power, and high beam quality have gradually been favored by researchers. However, the mechanism of the SRS self-mode-locking phenomenon is relatively complicated, and there are few theoretical studies at present. Although some experiments have reported the SRS self-mode-locking phenomenon with corresponding explanations, most of them have yet to be verified and improved. In addition, the conversion efficiency of self-mode-locked Raman lasers needs to be further improved. Therefore, to solve the above problems, an elaborate folding coupled cavity design was employed. Taking advantage of the folding coupled cavity, the fundamental and Raman cavities can be adjusted independently. Hence, the SRS self-mode-locking effect can be verified clearly according to the experimental results, and the mode matching between the fundamental and Raman waves can also be optimized by adjusting the length of the Raman cavity to improve the conversion efficiency of SRS.Methods A schematic diagram of the Nd∶YVO4-YVO4-Cr 4+∶YAG passively Q-switched intracavity self-mode-locked Raman laser based on a folding coupled cavity is shown in Fig. 2. The fundamental resonator consisted of M1, M2, a Nd∶YVO4 crystal, and a Cr 4+∶YAG crystal. A common L-shaped Raman cavity (designated by mirror path M2-M3-M4) was adopted for the mode matching between the fundamental and Stokes waves. The radius of curvature of M1 was 150 mm, and the output coupler (OC) M2 was a flat mirror. The pump source was a fiber-coupled LD emitting at 808.2 nm with a maximum output power of 50 W. A 1∶1 multilens coupler was used to focus the pump light into an a-cut 0.3% Nd∶YVO4 crystal with a radius of ?200 μm near the incident facet of the laser gain medium, and the dimensions of the Nd∶YVO4 crystal were 3 mm×3 mm×20 mm. A 4 mm×4 mm×3 mm Cr 4+∶YAG crystal with 80% initial transmittance at 1064 nm was employed and placed as closely as possible to the OC. An a-cut 4 mm×4 mm×30 mm YVO4 crystal was used as the Raman crystal, which was 1° wedged on both facets. All the components were coated according to our requirements. The length of the fundamental cavity composed of M1 and M2 was fixed at 110 mm. By adjusting the ROC of M4 and the length of the L-shaped Raman cavity, optimization of mode matching between the fundamental and Stokes waves can be achieved effectively. Results and Discussions The linear cavity Nd∶YVO4-YVO4-Cr 4+∶YAG passively Q-switched intracavity self-mode-locked Raman laser was first studied. When the transmittance of the OC was 5%, a maximum output power of 0.81 W was obtained at 1176 nm under a pump power of 17.15 W, with an optical-optical efficiency of 4.72% (Fig. 2). The corresponding PRF and pulse width were 885.4 MHz and ?219.16 ps, respectively (Fig. 3). After that, a 45° dichroic mirror M3 was inserted into the cavity to construct an L-shaped folded Raman cavity with M4 and M2 (OC). When the radius of curvature (ROC) of M4 was 100 mm and the length of the Raman cavity was 120 mm, a maximum power of 1.23 W with 1176 nm Q-switched mode-locked output was obtained under the pump power of 17.15 W, which was an improvement of over 50% compared with the linear cavity [Fig. 4(a)]. The PRF and pulse width of the mode-locked output were 942.9 MHz and ?125.8 ps, respectively (Fig. 5). The linewidth was 0.2 nm, and the beam quality factors Mx2 and My2 were 1.39 and 1.42, respectively (Figs. 7 and 8). Replacing the ROC of M4 with 150 mm and increasing the length of the Raman cavity to 180 mm, a maximum power of 1.19 W at 1176 nm Q-switched mode-locked output was obtained at the pump power of 17.15 W, with a conversion efficiency of 6.94% [Fig. 4(b)], and the PRF was reduced to 675.6 MHz (Fig. 7). Conclusions A passively Q-switched Nd∶YVO4-YVO4-Cr 4+∶YAG self-mode-locked Raman laser based on a composite cavity was demonstrated. Taking advantage of the folded-coupled cavity, the length and mirrors of the fundamental and Raman cavities can both be adjusted independently. Hence, the SRS self-mode-locking effect has been clearly obtained in a simple manner according to the experimental results, and the mode matching between the fundamental and Raman waves can be optimized by adjusting the length of the Raman cavity. In this way, the output power and conversion efficiency of the Q-switched mode-locked Raman output can be greatly improved. In addition, the folding coupled cavity structure was proved to control the PRF of the 1176 nm mode-locked output actively by adjusting the length of the Raman cavity together with the ROC of the mirror, without a reduction of output power and efficiency.

Objective Various pulse shaping processes, including convention soliton, stretched pulse, similarity, and dissipative soliton, are formed in present passively mode-locked fiber lasers based on the diverse distribution positions of dispersion in the cavity. The development of soliton pulses has raised the single pulse energy to a new level, making fiber lasers cater to the needs of fields, such as optical metrology, biomedicine, and laser micromachining. Dissipative solitons are usually generated from lasers with a large net normal dispersion owing to the effects of dispersion, nonlinearity, gain, and loss. The spectral amplitude modulation introduced by the spectral filter plays a key role in forming the dissipative soliton. Therefore, various filters are used in the lasers. The birefringent filter has been widely used owing to its flexible filtering bandwidth and good fiber compatibility. In addition, noise-like pulses have also been extensively studied in normal dispersion lasers.Both dissipative solitons and noise-like pulses can be generated in Ytterbium (Yb)-doped fiber lasers by reasonably adjusting the cavity parameters such as the pump power. Although pulse state switching has been verified in many experiments, few reports on the multiple switching of dissipative soliton and noise-like pulses in Yb-doped fiber lasers are available. In this study, we design a Lyot filter with a stable and powerful comb filtering using a pair of polarization-maintaining 45° tilted fiber gratings as polarizers and section of polarization-maintaining fiber as the birefringent medium. Therefore, an all-normal-dispersion Yb-doped fiber laser can achieve stable dissipative soliton mode-locking. By increasing the pump power unidirectionally in the dissipative soliton mode-locking state, the laser realizes multiple switching of dissipative soliton and noise-like pulse.Methods Two polarization-maintaining 45° tilted fiber gratings are separated by a length of polarization-maintaining fiber with a particular splicing angle in the Lyot filter used in the experiment. It can be used as a comb filter in the laser cavity and a fiber-type polarizer because of its unique structure. To generate linear polarization light, the first grating couples the TE polarization component out of the fiber core and causes the TM polarization component to propagate in the fiber core. Linear polarization light accumulates linear phase shift in the polarization-maintaining fiber owing to the particular splicing angle between the grating and the polarization-maintaining fiber. The linear phase shift is transferred to the amplitude modulation in the second grating, resulting in comb filtering. The specific splicing angle is designed to be 45° for the filter to have the maximum-filtering modulation depth.Results and Discussions The laser realizes stable dissipative soliton mode-locking with a pump power of 177 mW by finely adjusting the polarization controller, and its spectrum is shown in Fig. 4(a). The switching of the mode-locking pulse state can be observed while keeping the polarization controller and only increasing the pump power. When the pump power is increased to 323 mW, the spectrum gradually broadens under the influence of enhanced self-phase modulation. At this time, the pulse generated by the laser is still a dissipative soliton. Then, the pump power is continuously increased up to 455 mW in Fig. 4(e).The sharp edges of the spectrum gradually disappear and become smooth. The autocorrelation trace in Fig. 4(f) has a wide base with a narrow peak, typical of noise-like pulses, indicating that the laser is working in a noise-like pulse regime. When the increasing pump power reaches 691 mW, the laser generates a dissipative soliton pulse again. The pulsed state switching of the laser can be attributed to the switching of the feedback mechanism of nonlinear polarization rotation. Fig. 5 shows the relationship between the instantaneous power and nonlinear polarization rotation transmittance. The sinusoidal transmission spectrum indicates that the nonlinear polarization rotation can occur in positive and negative states. The critical power of positive and negative feedback is named critical saturation power. After achieving stable mode-locking, the intracavity instantaneous power increases owing to the continuously increasing pump power. When the instantaneous power in the cavity exceeds the critical saturation power, the feedback mechanism of nonlinear polarization rotation switches from positive to negative feedback state, which causes the pulse state of the laser to switch from dissipative soliton to noise-like pulse. When the instantaneous power reaches the critical saturation power again, the feedback mechanism switches from the negative feedback state to the positive feedback state, so that the mode-locked pulse state of the laser also switches back to the dissipative soliton.Conclusions We integrated a compact Lyot filter with a pair of polarization-maintaining 45° inclined fiber gratings in an all-normal-dispersion Yb-doped fiber laser. The laser realizes stable dissipative soliton mode-locking at a pump power of 177 mW. Furthermore, the pulse state of the laser recognizes switching from dissipative soliton to noise-like pulse and then to dissipative soliton by only continuously increasing the pump power from 177 mW to 691 mW. As adjusting the state of the polarization controller during switching is not necessary, it has higher controllability and accuracy, and the laser can be designed as a compact multifunctional light source.

Objective Featuring wide bandgap tunability, high quantum efficiency, and cost-efficient solution processible fabrication methods, colloidal quantum dots (QDs) have been studied and applied in various optoelectronic devices including photo detectors, light-emitting diodes (LEDs), and solar cells. In addition to applications based on the absorption and spontaneous emission of colloidal QDs, their stimulated emission potential has attracted extensive research interests, aiming toward a landmark target: the realization of the colloidal QD laser diodes. In the study of colloidal QD lasers, different laser architectures have been demonstrated, including Fabry-Perot cavity, distributed feedback laser cavity, whispering gallery mode cavity, and photonic crystal microcavity. The optical gain has been successfully realized in colloidal QDs under direct current pumping, demonstrating a major progress toward electrically pumped colloidal QD lasers. Furthermore, a dual function device based on specially engineered QDs that can function as an optically pumped laser and an LED is fabricated and characterized, revealing a promising pathway for realizing colloidal QD laser diodes. Different from edge-emitting lasers, vertical-cavity surface-emitting lasers exhibiting surface-emitting properties, wafer-level fabrication & characterization capability, and array integration ability have been widely used in optical fiber communication, laser printers, computer mouse, and three-dimensional facial recognition fields, etc. Here, we propose and design a colloidal quantum dots vertical cavity surface emitting laser, combining with a quantum dots light-emitting diode like current injection structure to realize the electroluminescence ability.Methods As shown in Fig. 1, the QLED-like structure containing the QD gain medium is sandwiched by two high-reflective distributed feedback reflectors to form a vertical-cavity surface-emitting laser(VCSEL)-like device. The device is designed to work under optical or electrical pumping. The DBR parameters and cavity lengths, are determined by numerical simulations with optimal performance. A DBR mirror is formed by periodically arranging two materials with different refractive indices. The reflectance spectrum is determined by both the DBR materials and DBR periods. Herein, we designed and calculated two types of DBRs with different periods (Fig. 3): (a) SiNx/SiO2 DBR and (b) TiO2/SiO2 DBR. It is found that 10 periods of the designed dielectric DBR can realize a peak reflectance of greater than 99%. The cavity length is a crucial parameter of the VCSEL device. After determining the DBR parameters, the permitted longitude modes inside the cavity can be tuned using the cavity length. Here, we use the FDTD method to build the designed QD-VCSEL device model and sweep the cavity length parameter. The current injection structure along the vertical direction includes the QD gain materials, electron and hole transmission layers, and electrodes. To tune the effective cavity length while retaining the optimized current injection capability, transparent ITO electrodes are selected and designed according to a suitable thickness. By theory, the smallest cavity length of a VCSEL device is λ/2. Thus, based on this length, the current injection structure and the thickness of the gain medium are fixed, while the thickness of the transparent ITO electrode is used to change the cavity length and then tune the resonant mode (Fig. 5). In addition to the λ/2 cavity length device, a 3λ/2 cavity length device is designed and simulated to theoretically optimize the optical parameters.Results and Discussions Under optical excitation, the designed λ/2 cavity length QD-VCSEL device can support single-mode lasing at 629.5 nm with a cavity length of 172 nm. The calculated quality factor is 259632. Alternatively, the 3λ/2 cavity length device can be optimized with a 520-nm cavity length. The lasing mode is realized at 632 nm, and the quality factor is 148291. Compared to the cavity with the smallest cavity length, a longer cavity suffers further optical loss while facilitating a thicker gain region. However, a considerably longer cavity length is not favored because of the difficulty in the formation of a very thick QD layer with a high concentration. The simulated far-field pattern reveals that the designed devices achieve a low output beam divergence, comparable to conventional VCSEL devices, which is an intrinsic advantage of this type of semiconductor laser. This work proposes a new scheme for realizing QD laser diodes, providing a theoretical basis and a parameter reference for future experimental verification.Conclusions In this work, a CdSe QD vertical-cavity surface-emitting laser is designed. The QD-VCSEL device is simulated with a QLED-like structure sandwiched by two dielectric DBR mirrors. The DBR parameters and cavity lengths are determined by numerical simulations with optimal performance. Single-longitude mode lasing can be supported by two designed cavities with different lengths with a maximum quality factor Q over 250000. The new solution toward electrically pumped colloidal QD lasers is revealed with our design, along with the theoretical model and key factors, which can be helpful in subsequent experimental work.

Objective Because of their high application potential in long-distance optical communications and ultrafast laser physics, optical solitons, which are localized structures in nonlinear systems, have piqued the interest of researchers. During long-distance propagation, optical solitons can interact with each other, resulting in a variety of bound-soliton states known as “soliton molecules”. Therefore, mode-locked lasers that can support the long-distance propagation of multiple solitons within their cavities are widely regarded as ideal platforms for studying soliton interactions and dynamics. However, the studies of the complex interactions of several optical solitons are difficult in traditional passive mode-locked lasers because fast drifts and the frequent collisions of solitons caused due to intense soliton interactions can degrade the stability of the laser mode-locking operation. In this paper, we use a high-repetition-rate optomechanically mode-locked fiber laser to successfully study the complex interactions of many optical solitons. The strong optomechanical effect in a short length of solid-core photonic crystal fiber (PCF) allows forming a robust optomechanical lattice in the laser cavity, and multiple solitons can be stably trapped within each cycle of the optomechanical lattice. Experimental results reveal that complex soliton interactions can be observed and partially controlled in this optomechanically mode-locked fiber laser, highlighting the significant potential of this unique optomechanical fiber laser system for studying complex soliton dynamics.Methods To investigate the complex phenomena of multi-pulse interactions, we created an optomechanically mode-locked fiber laser with a short solid-core PCF length as the harmonically mode-locking element. Because of the strong coupling between optical and acoustic waves in the PCF, a robust optomechanical lattice formed in the laser cavity, dividing the laser cavity into two halves. Multiple solitonic pulses can be trapped within each of these time-slots, working as an optical-soliton “reactor.” The stability of the optomechanical lattice is largely enhanced by the strong optomechanical interactions in the PCF core. In contrast, the multiple solitons trapped in each lattice cycle were observed to interact intensely with each other.In the experiments (Fig. 1), an erbium-doped fiber amplifier was used with two 980 nm laser diodes as the pump sources. Two polarization controllers working together with an optical polarizer acted as an artificial saturable absorber through nonlinear polarization rotation (NPR). A time-stretch dispersion Fourier transform (TS-DFT) setup was built using a 5-km-long single-mode fiber (SMF) as the stretching element to demonstrate the detailed information of the multi-soliton interaction processes. Two 45-GHz bandwidth detectors and a 33-GHz bandwidth oscilloscope were used to record the laser output pulses' time-domain trace and DFT signal. Furthermore, the laser output spectrum was recorded using an optical spectrum analyzer with a resolution of 0.02 nm. Since most of the laser cavity was made from SMF, the cavity dispersion was strongly anomalous with a calculated average value of -23.8 ps2/km, leading to soliton operation of the laser with hyperbolic pulse shape.Results and Discussions By carefully adjusting the intra-cavity polarizer controllers, stable harmonic mode-locking at 1.89 GHz resonance frequency of the acoustic core resonance in the PCF could be realized when both of the two pump diodes have pump powers of approximately 380 mW at 980 nm. The laser output spectrum, as well as the time-domain pulse sequence, were captured. When only one soliton is trapped in each cycle of the optomechanical lattice, the stable acousto-optic mode-locking state could be obtained with a 3 dB spectral bandwidth of 2.46 nm [Fig. 2(a) and Fig. 2(d)]. At this state, a strong Kelly-sideband observed on the pulse spectrum indicates that the laser was operating in the soliton regime. Through the experiments, we found that by adjusting the intra-cavity polarizer controllers, the laser output spectrum could be varied gradually from 2.46 nm to less than 0.6 nm [Figs. 2(b)--(f)], while the optomechanical lattice remained to be stable and the average output power of the laser was kept almost constantly at approximately 70 mW, giving rise to a series of quasi-stable acousto-optic mode-locking states. In those states, multiple soliton pulses were trapped within each cycle of the optomechanical lattice. The DFT signal unveiled that intense and complex interactions between the trapped solitons occurred within each cycle of the lattice (Figs. 3--5). As an entire, this optomechanically mode-locked fiber laser system permits several solitons to coexist in its cavity, and complex interactions between these solitons in one cycle could be studied in the future using this unique platform.Conclusions In the experiments, we obtained a large number of quasi-stable states in an optomechanically mode-locked fiber laser. We discovered that each isolated cycle of the optomechanical lattice could function as robust optical-soliton “reactors,” allowing us to study complex and intense soliton interactions. We could partially adjust the number of pulses trapped in each cycle of the optomechanical lattice and, thus, the total number of solitons generated in the laser cavity by adjusting the working point of the NPR effect in the laser cavity. In this way, we could control to some extent the multi-soliton interaction processes. Compared with a traditional passive mode-locked laser, this system's stability, flexibility, and high-repetition-rate features make it an ideal experimental platform for studying complex multi-soliton interactions, providing some useful insights on soliton dynamics.

Significance Terahertz (THz) wave, whose frequency range lying between the infrared wave and microwave, is a section of the electromagnetic spectrum with unique features. Broadly, the THz frequency range covers the spectral region from 0.1 to 30 THz, but researchers universally define 0.1--10 THz as the THz band. Owing to the lack of suitable THz radiation sources and detectors, THz wave, once known as "THz Gap" for ages, became the last segment of the entire electromagnetic spectrum to be fully explored. Nevertheless, the ultrafast optoelectronics technology and microscale semiconductor technology, which received rapid development in the 1980s, made emerging from the previous dilemma possible for researchers in this field. Since then, THz technology has been widely used in scientific research and practical application, such as medical treatment, nondestructive testing, national defense, safety inspection, and communication.Among various branches of THz technologies, THz photonics undoubtedly earns a place in the current hotspots of the field. However, recently, studies on THz photonics focused more on the linear response of material rather than the nonlinear response, and the absence of suitable THz radiation sources with strong field strength might be to blame for this study status. In China, the 973 Program and several other projects associated with fundamental study supported the study of THz photonics well, unfortunately, the related study work is still focused on linear THz systems. Thus, exploiting the technology for building a more suitable THz radiation source system with stronger THz field strength, along with its optimization, is an essential prerequisite for further expanding the practical area of THz waves.As the most common liquid in life, water plays an important role in academic research. However, water has not been considered an appropriate THz radiation source for a long time owing to its strong absorption in the THz frequency range. While recently, some groups have experimentally confirmed the feasibility of THz wave generation from water under the excitation of a femtosecond laser, and several theoretical models are proposed for the mechanism of the generation processes. Thus, summarizing the current research progress in this field for the study of THz wave generation from water and other liquids is significant.Progress Under the efforts of professor Zhang Xicheng and his team, a thin water film (~170 μm) under femtosecond laser’s excitation was historically used for THz wave generation in 2017 (Fig. 1). This water film, which has a gravity-driven free-flowing structure, can be better for mitigating the inevitable absorption of THz waves caused by water. Under the identical experimental condition, the electric field of THz wave generated from the thin water film is 1.8 times stronger than the case of the air plasma. Besides, compared with air plasma, a great difference in THz generation is that liquid water prefers a laser pulse with a longer pulse duration (~600 fs) rather than a traditional femtosecond pulse (~50 fs) (Fig. 3). Further, the improvement of the THz electric field with a dual-color laser scheme in water is not able to be as high as that in air. The different photoionization mechanisms between water and air might be responsible for these discrepancies.To weaken the total internal reflection at the flat water-air interface in water film, a water line was used for solving this problem. In 2018, strong THz radiation generated from a water line of 200-μm diameter under the femtosecond laser’s excitation was first reported in the experiment (Fig. 8). Strong THz radiation can be detected only when the pump laser propagation axis deviates from the center of the waterline owing to the ponderomotive force-induced photocurrent with the symmetry broken at the air-water interface. As for its preference, just like the results of water film, subpicosecond laser pulses behave better in water lines when compared with a short temporal laser pulse. Notably, the overall distribution of THz radiation generated by water lines with different diameters may not be different in space, so the optimal radiation angle for THz generation from a water line is all about 60°(Fig. 10).Although the phenomenon is similar to air, the explanation for the mechanism of THz generation from water is still under study. While other kinds of nonlinear effects existed during the photoionization process in water and the different ionization mechanisms of these two matters may partially account for this mystery. Currently, the theoretical models to explain the phenomenon of THz generation from water include the dipole array model, ponderomotive-force-induced photocurrent model, unidirectional pulse propagation equation model, and radiation field dynamics model.Conclusions and Prospect The study on THz generation from liquid water would be beneficial for researchers to better understand the interaction between water and intense lasers, and it serves a significant role in the further study of potential THz radiation sources. Based on these reasons, research progresses on THz generation from liquid water recently, are reviewed in this article, which includes experimental schemes for two liquid water sources, the factors related to the improvement of the generated THz energy, as well as the design ideas for relevant theoretical models. Finally, the prospect of this field based on our understanding and the current study achievements is proposed.

Significance Terahertz wave generation technology has rapidly become efficient recently and is the key, foundational technology to realize a wide application of terahertz band in comprehensive research. The terahertz band, located in the transition region from traditional electronics to photonics, has low energy, water sensitivity, special penetrability and many other unique properties. With respect to these characteristics, terahertz wave has widely used applications in biomedical diagnosis, safety inspection, nondestructive testing, terahertz communication and radar. Terahertz parametric and difference frequency radiation sources, based on optical nonlinear frequency conversion technology, can generate wideband tunable, monochromatic terahertz waves. Terahertz parametric and difference frequency radiation sources also have the advantage of compact structure and utilization at room temperatures(18--30 ℃).With the improvement of laser technology and crystal growth technology, terahertz parametric radiation source and terahertz difference frequency radiation sources are developing rapidly to expand frequency tuning range, improve output energy, narrow terahertz wave linewidth resulting in a series of new technologies, such as ring cavity, circulating pump, and pulse-seeded injection.Progress Compared with congruent MgO-doped lithium niobite(MgO∶CLN) crystal, the upper tuning frequency limit of terahertz parametric oscillator based on near-stoichiometric MgO-doped lithium niobite(MgO∶SLN) crystal can be increased from 3 THz to 4.64 THz (Fig. 4). Terahertz parametric oscillator based on KTiOPO4(KTP) and KTiOAsO4(KTA) crystal can further improve the frequency tuning upper limit of terahertz wave; even though there are gaps in the frequency tuning range, continuous tuning cannot be achieved (Fig. 6). Terahertz parametric oscillator based on ring cavity can broaden the frequency tuning range of terahertz wave and improve the output energy (Fig. 7). Terahertz parametric oscillator based on pump recycling technology can improve the pump efficiency and greatly increase the output energy of terahertz wave in the entire frequency tuning range (Fig. 9). Comparing the pump source with ns pulse width, the terahertz parametric radiation source based on sub-ns pump laser can not only improve the pump peak energy but also effectively suppress the stimulated Brillouin scattering in nonlinear crystal, considerably improving the output energy. Pulse-seeded injection technology not only further improves the frequency tuning upper limit of the terahertz parametric oscillator based on LiNbO3 crystal to 5.15 THz but also addresses the disadvantage of low output energy in high frequency band to maintain high output energy in a wider range (Fig. 11).Based on dual KTP-optical parametric oscillation(KTP-OPO) technology, terahertz difference frequency radiation source based on inorganic crystal such as GaSe can achieve high repetition rate terahertz wave output (Fig. 12), which can be used in near-field microscopy, rapid scanning THz spectroscopy and other occasions, requiring high repetition rate of terahertz wave. Terahertz difference frequency radiation source based on 4’-dimethylamino-N-methyl-4-stilbazolium tosylate(DAST), 4-N,N-dimethylamino-4’-N’-methyl-stilbazolium 2,4,6-trimethylbenzenesulfonate(DSTMS), N-benzyl-2-methyl-4-nitroaniline(BNA), and other organic crystals can achieve ultra-wideband terahertz wave output in the range of 1 THz to 30 THz.Owing to the characteristics of terahertz wave, i.e., low energy, water sensitivity, and fingerprint spectrum, terahertz technology has a good application potential in the field of traumatic brain injury detection. For example, multi depth slice terahertz imaging technology can accurately identify the severity of traumatic brain injury (Fig. 15). Terahertz imaging technology based on machine learning can automatically recognize and classify different degrees of traumatic brain injury samples.Conclusions and Prospects Terahertz parametric and difference frequency sources based on optical nonlinear frequency conversion technology can generate terahertz wave with high output energy, wide frequency tuning range, and narrow linewidth. With the improvement of terahertz radiation source performance, terahertz technology will have greater applications in biomedical detection, nondestructive detection, safety inspection, terahertz radar, and so on.

Significance Recently, terahertz technology has made rapid progress in the fields of label-free analysis, cellular level imaging, chemical and biological sensing, security screening, and wireless communications. Furthermore, major advances have been made in terahertz sources, detectors, and modulators. Developing efficient modulators with natural materials is challenging due to the relatively weak interactions between terahertz waves and natural materials. Although functional modulators operating in the visible and infrared bands are mature for commercial applications, efficient, functional, and high-speed modulators are severely lacking in the technologically important terahertz band. Therefore, researchers are trying to obtain solutions that can improve light-matter interactions for terahertz applications. Metamaterials have extraordinary electromagnetic properties that show great potential to enhance local field strength significantly and improve light-matter interactions in practical terahertz modulators. The integration of metamaterials and certain active materials or techniques leads to the revolution of conventional modulators, which are named “metadevices” in this review. Metamaterials enable abundant functionalities of these devices, and integrated materials offer active responses to external stimuli. These types of hybrid metadevices lead to lower energy consumption, larger modulation depth, faster modulation speed, and more abundant functionalities due to the substantial local electric field enhancement of metamaterials. In this review, the current progress of active metadevices for terahertz applications is summarized with different approaches. In addition, the working mechanism, typical device configurations, major performance, and drawbacks are discussed.Progress This review summarizes several typical configurations of active terahertz metadevices integrated with liquid crystals (Section 2.1), micro-electromechanical systems (MEMS, Section 2.2), semiconductors (Section 2.3), graphene (Section 2.4), phase change materials (Section 2.5), superconductors (Section 2.5), nonlinear materials (Section 2.5), and chemical reactions (Section 2.5). Electrically triggered liquid crystals integrated with metamaterials exhibit excellent terahertz modulation performance, operating in both transmission and reflection modes. Due to the great flexibility of electrical actuation, this type of metadevice can realize programmable control and operate in a complex configuration for wave deflection ( Fig. 1). The modulation speed of liquid crystal metadevices is limited by the intrinsic on-off speed of liquid crystals and can be optimized to 1 kHz. Moreover, the pixel density of spatial light modulators in this type of configuration is still low, severely decreasing imaging resolution. MEMS metadevices have similar electrical actuation modes and can realize programmable control for terahertz polarization control, wavefront deflection, and dynamic hologram. MEMS cantilevers are reconfigured by an electrostatic force, which is different from the refraction index change of liquid crystals with an external stimulus, and spatial deformation of cantilevers leads to modulation of terahertz resonance frequencies. This review mainly focuses on the polarization effects induced by MEMS metadevices for strong optical activity and artificial chirality ( Fig. 2). In addition to conventional electrical actuation, MEMS cantilevers can be controlled via external forces. The first order vibration frequency of the mechanical vibration of cantilevers determines the modulation speed of MEMS metadevices that depends on the geometric parameters of cantilevers and is in the order of kHz for terahertz metadevices. In addition, scalability, reliability, and uniformity of large-area MEMS metadevice arrays need to be improved for applications in the terahertz band and shorter wavelengths. The major problem of limited modulation speed can be addressed by excluding electrical stimuli and applying an optical pump. All-optical metadevices have no theoretical limitations of modulation speed determined by the relaxation dynamics of the active materials. By integrating semiconductors (e.g., Si, Ge, GaAs, and WSe2) with metamaterials, all-optical hybrid metadevices demonstrate excellent performance for ultrafast terahertz modulation (Fig. 3). Recently reported dielectric metadevices also reveal active, low-loss, and functional operations without integrating extra active media (Fig. 3) by directly pumping the resonators. Graphene can be controlled by electrical and optical stimuli and is a promising material for terahertz applications. By directly depositing a graphene layer on metamaterials, the hybrid metadevices enable fast and efficient terahertz modulation with easy fabrication and high efficiency. Phase change materials provide nonvolatile modulation and memory effects. Superconductors can induce high-quality factor resonance modes and provide low-threshold and ultrafast terahertz modulation but must operate in cryogenic temperatures. The nonlinear effects are theoretically instantaneous and can enable femtosecond or shorter switching time. Metadevices integrated with diodes or photodiodes are also very interesting; materials whose properties change under different chemical environments are also a possible solution for active metadevices (Fig. 4).Conclusions and Prospects Different techniques have been discussed for hybrid metadevices with stimuli of electricity (e.g., liquid crystals, semiconductors, graphene, MEMS, and diodes/transistors), optics (e.g., semiconductors, graphene, phase change materials, and superconductors), heat (e.g., phase change materials and superconductors), forces (e.g., MEMS), and chemical reaction (e.g., Mg). Although there are certain limitations for different combinations, metadevices have made major progress toward realizing efficient terahertz modulators. With the maturity of the semiconductor industry, active metadevices with semiconductors are very attractive for programmable, fast, and efficient terahertz applications. Metadevices integrated with graphene are also attractive with easy fabrication and high efficiency that can be actuated by electrical or optical stimuli. All-optical metadevices are the solution to access faster modulation speed, and an appropriate combination of nonlinear materials and metamaterials would push modulation speeds to the GHz or even THz regime. All the approaches have pros and cons and should be utilized where most applicable.

Significance As the cornerstone of terahertz technologies, terahertz radiation sources and detectors have attracted lots of attention. The research on terahertz sources and detectors has run through the entire development of terahertz technologies. In recent years, terahertz technologies have become more closely integrated with other disciplines, which puts forward higher requirements on the sources and the detectors. As the most typical terahertz transmitter and receiver, an photoconductive antennas (PCA) is widely used in laboratories and commercial terahertz systems due to its low request on pump power and compatibility with fiber technologies. The current commercial photoconductive antenna can meet the needs of spectral analysis of thin and low-absorption samples. However, compared with those of solid-state terahertz sources, the power of a PCA is at least two orders of lower. While compared with those of low-temperature thermal detectors, the sensitivity of a detection antennas is also insufficient, not to mention its low-pass filtering effect that is adverse to the detection of high-frequency terahertz waves. Thus, developing a high-performance PCA is an important issue to promote terahertz technologies.Surprisingly, the emergence of metamaterials sheds light on the development of advanced PCAs. With the extremely high degree of design freedom, metamaterials have become one of the mostly concerned artificial electromagnetic media platforms. In the past 20 years, metamaterials have attracted many outstanding researchers in electromagnetics, optics, acoustics, and thermodynamics. Numerous novel functions have been realized one after another, such as negative refractive index, electromagnetic cloak, bounded state in the continuum, perfect absorption, meta?lens, plasmon induced transparency, optical orbital angular momentum coupling, and optical topological insulator. The unique properties of metamaterials can also be used to improve the performance of photoconductive antennas. This review systematically introduces the research works of efficient photoconductive antennas based on metal and dielectric metamaterials. The germination, development and broad application prospects of metamaterial?assisted photoconductive antennas are elaborated. The novel methods based on metamaterials have greatly promoted the development of photoconductive antennas and we hope this review could bring in more researchers in this new direction.Progress We have classified the reported metamaterial-assisted photoconductive antennas into two categories. The first one is to enhance the interaction strength between the femtosecond laser pump and the substrate by using nano-scale metal or dielectric metamaterials. Metamaterials can manipulate the amplitude, phase and polarization of an electromagnetic wave with a high degree of freedom. In general, surface plasmons excited on metallic metamaterials have excellent field localization, which can tightly confine the light field near the structure, thereby enhancing the interaction between the photoconductive substrate and the incident light field. But so far, most metallic metamaterials have been severely affected by high ohmic losses, especially in the visible and near-infrared domains. In recent years, it has been found that dielectric metamaterials can also form field localization and have low absorption loss. Such metamaterials are generally composed of dielectric microcavities with high refractive index, which support rich electric field resonance modes. This review summarized the works which enhance the absorption of the pump light by the metallic or dielectric nanostructures, thus greatly improve the light-terahertz conversion efficiency and the detection sensitivity (as shown in Figs. 2, 6 and 7). In addition to being integrated in the antenna gap, the metallic nano-metamaterials can also be directly used as the electrode of the antenna, which changes the electrode shape. By this method, the transport time for the carrier to the electrode is reduced, and the performance of the terahertz photoconductive antenna is improved (as shown in Fig. 3). Furthermore, as shown in Fig. 4, constructing an antenna array integrated with nano-electrodes can increase the upper limit of pump power, thereby avoiding the saturation effect of the femtosecond laser pump. This method can increase the terahertz power by more than one order of magnitude. Besides, we also mention that metamaterials can also improve the performance of photomixers, which are more important in the fields of terahertz communication and imaging (as shown in Fig. 5). The second category is to design micron-scale metallic or dielectric metamaterials for directly manipulating terahertz waves. Designing a micron structure near the antenna electrode can effectively manipulate the spectral characteristics of the radiated terahertz wave, as shown in Figs. 8 and 9. Another way to manipulate the radiated terahertz pulse is to construct an all-dielectric meta-lens sticked to the photoconductive substrate for replacing the hyper-semispherical silicon lens and collimating the terahertz wave to a parallel beam (as shown in Fig. 10).Conclusion and Prospect This review introduces a series of terahertz photoconductive antennas integrated with metamaterials. Their excellent properties and novel functions may bring a huge application potential, which have greatly promoted the development of photoconductive antennas.The methodology reviewed here is still in its early stage. We believe that more metamaterials with novel functions will be applied in the future. Besides, the introduction of new materials, for example two-dimensional materials such as graphene and black phosphorus, may cause fundamental changes in the design of photoconductive antennas. Another possible research direction is to integrate phase gradient metamaterials based on the generalized Snell’s law for manipulating the wavefront of photoconductive antenna emitted terahertz wave. Besides, considering the significance of photomixers to terahertz communications, using new metamaterials to realize an active control of photomixers becomes more and more important.In short, the novel antennas shown in this review have a great application potential in terahertz technologies, which are expected to fundamentally promote the development of photoconductive antennas and significantly advance the performance of terahertz photoconductive devices.

Objective Plasmon-induced transparency (PIT) is a quantum interference effect occurring in three-energy-level systems. Under an external laser pump, a substance that is originally opaque becomes transparent within a specific frequency range, and a wide and continuous resonance dip is replaced by a sharp transparency window in the transmission spectrum. Due to its strong dispersion, the PIT effect occurring during electromagnetic wave transmission has great potential for applications such as slow-light devices, optical dynamic storage devices, and high-sensitivity sensors. Subwavelength periodic metasurfaces are one of the most widely used methods to achieve the PIT effect for electromagnetic waves. In previous reports, the external electric field component was usually used as the excitation source, and the magnetic field component,which is an important component of the external field,was rarely used. The aim of current research is to effectively manipulate the PIT effect arising from the interaction between a subwavelength periodic metasurface and the external field. In this study, terahertz (THz) time-domain spectroscopy is adopted to perform a systematic study of PIT metasurfaces placed in a parallel-plate waveguide (PPWG). Under external TE mode excitation, effective modulation of the PIT effect based on a PPWG metasurface system is realized by varying the structural parameters of the metasurfaces both theoretically and experimentally. Our design may provide a new strategy for the design of tunable electromagnetic devices based on the PIT effect.Methods Based on our previous work, we designed and fabricated a series of aluminum split-ring resonator pairs(SRRPs) with different gap locations on a quartz substrate. There were eight periods of SRRPs along the wave propagation direction, and the gap in split-ring 2 was moved along the x and y axes [Figs. 1(b) and 2]. Then, the sandwich structures (quartz substrate-aluminum structures-quartz substrates) were inserted into a copper PPWG for characterization. The input and output ends of the waveguide were tapered to increase the conversion efficiency from free-space THz radiation to waveguide modes.To verify the experimental results, SRRPs with the same geometric parameters were simulated using CST Microwave Studio, and the coupled mode theory was used to fit the simulated results. The underlying mechanism of the modulation of the PIT effect was understood through simulations of the surface current and absolute electric field distributions at the frequencies of interest. In simulations, the TE mode of the PPWG was applied to excite the SRRPs; that is, the electric field was parallel to the waveguide plates, and the magnetic field was perpendicular to the waveguide plates. The time-domain solver was adopted, and the background was set as a perfect electric conductor. The electric boundary condition was assigned to all directions, and waveguide ports were set to the input and output ends of the waveguide. In simulations, only one row with eight SRRPs was modeled along the direction of the guided wave, and the aluminum and the quartz substrate were treated as the Drude model and lossless dielectric, respectively. The tapered parts of the waveguide were not considered in the simulations.Results and Discussions The transmission spectra for the gap in split-ring 2 moving along the x and y axes were measured, simulated, and fitted. The results (Figs. 3 and 5) indicate that the experimental, simulated,and fitted results agreed well. When the gap in split-ring 2 was moved along the x axis, the PIT effect occurred due to the interaction between the subwavelength periodic metasurface and the external field.However, the PIT effect could not be manipulated effectively by changing the gap position. When the gap in split-ring 2 was displaced along the z axis, the transmission spectra demonstrated that the PIT effect gradually disappeared.To understand the underlying mechanism of the PIT effect, the surface currents in the SRRPs at the resonance dips were simulated. The insets of Figs. 3(b) and 5(b) demonstrate that an antisymmetric mode appeared at the lower resonance frequency, where as a symmetric mode appeared at the higher frequency. When δz=25 μm, the PIT effect vanished, and the single resonance mode was a symmetric mode. The above simulated results are consistent with the plasmon hybridization picture.The underlying mechanism of the PIT effect was also analyzed by simulating the absolute electric field distributions at the PIT window frequencies. As depicted in Fig.4, the absolute electric field was mainly confined in the gap of split-ring 2, and there was a weak electric field distribution in the gap of split-ring 1 under the same excitation conditions. It is interesting that the intensity of the electric field in the gap of split-ring 1 became stronger when the gap in split-ring 2 was moved (Fig. 6). The absolute electric field distributions in the SRRPs can be explained by the bright-dark mode coupling theory, where split rings 1 and 2 can be treated as the bright and dark modes, respectively. Though the two rings can be excited by the external field, the bright mode can be suppressed by the dark mode through the deep-subwavelength distance between their adjacent arms.Conclusions Under external TE mode excitation, effective modulation of the PIT effect based on a PPWG metasurface system is realized by varying the structural parameters of the metasurfaces experimentally. The underlying mechanism of the PIT effect is analyzed through simulations of the surface current and absolute electric field distributions at the frequencies of interest. The simulated results demonstrate that the electric field in one ring is suppressed by another ring under the same excitation source at the PIT window frequencies, and the suppression effect can be controlled by varying the structural parameters of the SRRPs. Our work may open new avenues for slow-light devices, optical dynamic storage devices, and high-sensitivity sensors.