Structured light, also known as tailored light, shaped light, sculpted light, or custom light, refers to a series of special light beams with spatially variant amplitudes and phases, polarization distributions, or more general spatiotemporal profiles. In the past decades, structured light featuring distinct properties and unique spatial or spatiotemporal structures has grown into a significant research field and given rise to many developments from fundamentals to applications. Very recently, integrated structured light manipulation has become an important trend in the frontier of light field manipulation and attracted increasing interest as a highly promising technique for shaping structured light in an integrated, compact, and miniaturized manner. In this article, we give a comprehensive overview of recent advances in integrated structured light manipulation (generation, processing, detection, and application). After briefly introducing the basic concept and development history of structured light, we present representative works in four important aspects of integrated structured light manipulation, including multiple types of integrated structured light generation, many sorts of integrated structured light processing, diverse forms of integrated structured light detection, and various kinds of integrated structured light applications. We focus on summarizing the progress of integrated structured light manipulation from basic theories to cutting-edge technologies, to key devices, and to a wide variety of applications, from orbital angular momentum carrying light beams to more general structured light beams, from passive to active integration platforms, from micro-nano structures and metasurfaces to 2D photonic integrated circuits and 3D photonic chips, from in-plane to out-of-plane, from multiplexing to transformation, from linear to nonlinear, from classical to quantum, from optical communications to optical holography, imaging, microscopy, trapping, tweezers, metrology, etc. Finally, we also discuss in detail the future trends, opportunities, challenges, and solutions, and give a vision for integrated structured light manipulation. |
1.IntroductionStructured light beams refer to a group of unique light fields[1–53], such as vortex beams[10–19], Bessel beams[20–27], Airy beams[28–34], vector beams[35–42], and spatiotemporal beams[43–54], that possess specialized spatial amplitude, spatial phase, spatial polarization, and even more general spatiotemporal structures. Compared to traditional plane-wave light and Gaussian light with uniform field distribution, structured light with non-uniform field distribution offers richer degrees of freedom or more physical dimensions to be explored, providing more opportunities in many advanced application scenarios[55–155]. The concept of structured light can be traced back to Thomas Young’s double-slit interference experiment over 200 years ago in the early 19th century, where the interference fringes of a plane wave passing through a double slit could be regarded as a type of structured light with a one-dimensional intensity distribution[156]. Structured light can, in principle, extend to all space and time degrees of freedom of light waves, such as spatial amplitude/phase/polarization distribution and spatiotemporal structure, giving rise to many types of structured light beams[1–53]. With the rapid development in recent years, a variety of structured light beams have been reported, including vortex beams carrying orbital angular momentum (OAM)[10–19], high-order linearly polarized (LP) beams[157–167], Laguerre-Gaussian (LG) beams[10,168–177], Hermite-Gaussian (HG) beams[176–185], Bessel beams[20–27,72,114,136,137,186–190], Mathieu beams[191–198], Ince-Gaussian (IG) beams[199–209], Airy beams[2829.30.31.32.33.–34,210–216], bottle beams[217–225], needle beams[226–233], pin beams[74,234–237], array beams[238–243], spatiotemporal beams[43–54,244–249], knotted beams[250–256], in-plane waveguide modes[257–292], vector beams[35–42,293–304], skyrmion beams[305–317], Hopfion beams[117,318–324], and more general arbitrary structured light beams, as illustrated in Fig. 1. One can clearly see inhomogeneous spatial amplitude/phase/polarization distribution and spatiotemporal structure in various structured light beams. Remarkably, the distinct properties of structured light, such as infinite orthogonal OAM values in principle, as well as the unique light field structures, such as doughnut intensity profiles due to phase singularity of OAM-carrying vortex beams and polarization singularity of vector beams, enable various emerging advanced applications. Recently, structured light has shown significant advantages in many fields[55–155,325–366]. For instance, terabit free-space data transmission employing OAM multiplexing was demonstrated in 2012, showing that optical communications using OAM-carrying vortex beams can efficiently increase the transmission capacity and spectral efficiency[60]. Since then, more and more structured light communications (modulation, multiplexing, multicasting) have been proposed and demonstrated with favorable performance[21–23,61–69,71,76–78,334,335]. OAM holography for high-security encryption was demonstrated in 2019, allowing the multiplexing of a wide range of OAM-dependent holographic images with a helical mode index spanning from to 50, and leading to a 10-bit OAM-encoded hologram for high-capacity and high-security optical encryption[361]. Stimulated emission depletion (STED)-based super-resolution imaging/microscopy, in which the excitation beam is overlapped with a doughnut-shaped beam that is capable of de-exciting fluorophores by stimulated emission, was proposed in 1994[362] and demonstrated in 2006[363] to reduce the focal spot area by about one order of magnitude below the diffraction limit for resolving individual vesicles in the synapse. OAM-carrying light beams have been widely exploited as the essential roles of the interference reference, multiplexing channel, spatial filter, and excitation source to facilitate advanced optical imaging techniques of interferometry, holography, and microscopy[112]. A spiral interferometry assisted by OAM-carrying light beams was demonstrated in 2005, which could reconstruct a complete sample profile from a single exposure and distinguish elevations and depressions[364]. The photo-induced force microscopy (PiFM), combined with structured light illumination, such as an azimuthally polarized beam (APB), forming the so-called structured-light-induced force microscopy (SLiFM), was proposed and demonstrated to characterize conventionally elusive material properties with fidelity[111]. Controlled optical trapping and rotation of objects were demonstrated in 2001 using a spiral interference pattern generated by interfering an annular shaped laser beam with a reference beam[365]. The objects were trapped in the spiral arms of the pattern, and changing the optical path length caused the rotation of the pattern and the trapped objects. An interferometric pattern between two annular laser beams was also used to construct 3D trapped structures within an optical tweezer configuration in 2002[366]. Optical tweezers, optical spanners, rotational control with shaped beams, trapping with annular beams, and spinning and orbiting in optical tweezers have been widely studied using the spin angular momentum (SAM), OAM, and unusual spatial amplitude, phase, and polarization structures, promoting the development of structured light in the field of optical manipulation[119,127]. Detection of a spinning object was demonstrated in 2013 based on the rotational Doppler effect using OAM of light[129], which could be an important supplement of the traditional linear Doppler effect. Moreover, using vectorially structured light (vector beams) with spatially variant polarization, vectorial Doppler metrology was proposed and demonstrated in 2021[131] based on the vectorial Doppler effect, allowing both the velocity and motion direction of a moving particle to be fully determined. Using the spatial modes defining an infinitely dimensional discrete Hilbert space, entanglement of the OAM states of photons was demonstrated in 2001[146], which provides a practical route to entanglement involving many orthogonal quantum states. Later, various quantum information processing applications have been reported using structured light[149–151], such as multi-dimensional entanglement transport through single-mode fiber (SMF)[149], high-dimensional quantum cryptography with hybrid OAM states through ring-core fiber (RCF)[150], and ultrahigh-fidelity spatial mode quantum gates in high-dimensional space by diffractive deep neural networks[151]. These typical examples demonstrate that structured light has become an important research focus in the wide field of optics and photonics. Remarkably, the rich application scenarios and significant development potential of structured light have also promoted the research of structured light manipulation technology, including the generation, processing, and detection of structured light. Traditionally, the manipulation of structured light mainly relies on large-sized, discrete, and bulky devices, such as spatial light modulators (SLMs)[60,71,105,293,367–372], digital micromirror devices (DMDs)[373–381], diffractive optical elements (DOEs)[382–386], optical lenses[387–390], spiral phase plates/q-plates/J-plates[391–395], bulky lasers[332,396–402], detectors/receivers[403–408], and fiber-based devices[41,409–421]. Although these relatively large-sized devices have achieved excellent operation performance in structured light manipulation, they hinder the development and application of integrated structured light systems. In the future, considering various compact photonic devices on diverse platforms already reported for the flexible manipulation of conventional degrees of freedom of light waves (e.g., complex amplitude, frequency/wavelength, time, polarization)[422–462], it is foreseen that miniaturization and integration are inevitable trends, and the same applies to the structured light manipulation accessing the space degree of freedom of light waves. In recent years, as illustrated in Fig. 2, harnessing various basic theories and principles, such as spiral phase plate theory, mode-field superposition theory, phase control theory, holographic theory, coupled mode theory, whispering gallery mode theory, non-Hermitian theory, phased array theory, multiplane light conversion theory, geometric coordinate transformation theory, nonlinear interaction theory, chiral mode switching theory, surface plasmon polariton effect, photogalvanic effect, thermoelectric response of spin-Hall effect, diffractive deep neural network, and inverse design method, various miniaturized and integrated structured light manipulation technologies, in terms of generation, processing, and detection, have attracted increasing interest and achieved rapid development[1,2,422,463]. For example, the in-plane and out-of-plane integrated structured light generation can be realized by micro-nano-sized spiral phase plates[464,465], holographic gratings[466], 3D waveguides[467,468], trench waveguides[469,470], nanoantenna waveguides[471], microrings with angular gratings[472–475], metasurfaces[1,455,456,458–460,476–480], subwavelength structures[443,481–495], phased arrays[452,496], asymmetric directional couplers[422,497], etc. The flexible integrated structured light processing can be implemented by multimode waveguides[498–500], multiple phase planes[501–504], nonlinear photonic devices[505,506], 2D materials[507,508], 3D photonic chips[470,509,510], silicon photonic integrated circuits[511–513], inverse-design micro-nano structures[514–517], etc. The diverse integrated structured light detection can be achieved by plasmonic and dielectric micro-nano structures[518–522], photocurrent detectors with U-shaped electrodes[523], thermoelectric detector with spin-Hall coupler[524], hybrid optoelectronic neural network[525], silicon-based networks-on-chip[526], inverse design subwavelength structures[527], etc. These representative works show favorable performance in shaping light in an integrated and compact way, inspiring the rapid and vigorous development of the cutting-edge research field of integrated structured light manipulation. In this article, we comprehensively review recent advances in integrated structured light manipulation, which is divided into four aspects, i.e., generation, processing, detection, and application, as illustrated in Fig. 3. For the integrated structured light generation, we begin with the introduction of basic theories and underlying principles of structured light generation, including spiral phase plate theory, mode-field superposition theory, phase control theory (resonant phase, geometric phase, propagation phase, compound phase), holographic theory, coupled mode theory, whispering gallery mode theory, and non-Hermitian theory. We then present various in-plane and out-of-plane integrated generation schemes of OAM-carrying light beams based on passive and active integrated platforms, including in-plane to in-plane OAM generation, out-of-plane to in-plane OAM generation, in-plane to out-of-plane OAM generation, and out-of-plane to out-of-plane OAM generation. Additionally, we show the integrated generation of other structured light beyond OAM, such as chiral light, high-order LP modes, LG beams, HG beams, non-diffracting beams (Bessel beams, Mathieu beams, Airy beams, needle beams), vector beams, array beams, optical vortex lattices, spatiotemporal beams, knotted beams, in-plane waveguide modes, and reconfigurable structured light beams. For the integrated structured light processing, we introduce theories and principles for handling structured light, including multiplane light conversion theory, geometric coordinate transformation theory, nonlinear interaction theory, and chiral mode switching theory. Following this, we present several integrated structured light processing functionalities, such as out-of-plane/in-plane OAM (de)multiplexing, in-plane waveguide mode multiplexing, OAM mode transformation (multiplication, division), LP mode switching, array beam transformation, in-plane waveguide mode processing, switching of LP mode and OAM mode, and transformation of LP mode and in-plane waveguide mode. For the integrated structured light detection, we start by explaining theories and principles of structured light detection, including the surface plasmon polariton effect, photogalvanic effect, thermoelectric response of spin-Hall effect, diffractive deep neural network, and inverse design method. After that, we present several typical integrated structured light detection schemes, such as metal and dielectric micro-nano structures, plasmonic metasurfaces, photocurrent detector with horseshoe-shaped (U-shaped) electrodes, DMD and diffuser, hybrid optoelectronic neural network, silicon Mach-Zehnder interferometer (MZI) network, inverse design subwavelength structures, and silicon nanorod optomechanics. For the integrated structured light applications, we first introduce typical theories and operation principles for various applications with structured light, including structured light modulation communication theory, structured light multiplexing communication theory, structured light multicasting communication theory, structured light holography theory, and optical manipulation and trapping theory. Then, we show advances in a variety of integrated structured light application scenarios, such as optical communications [analog signal transmission, data-carrying digital signal transmission, high-speed spatial light modulation communication, chip-chip and chip-fiber-chip optical interconnects, direct fiber vector eigenmode multiplexing transmission, on-chip mode-division multiplexing (MDM) transmission, multi-dimensional data transmission and processing, and free-space and fiber-optic multimode communications], OAM-multiplexing holography for high-security encryption, 3D imaging using a multi-wavelength dots array, medical imaging using a needle beam and multifocal beam, PiFM using tightly focused APB, three-dimensional topography using a vortex beam, 3D optical manipulation using a 2D Airy beam, optical trapping using waveguide modes and optical phased arrays, chiral trapping using a silicon-based slot waveguide, optical tweezers, nanowires trapping and rotation, Doppler cloak by a spinning OAM metasurface, and quantum optics. Finally, we summarize the recent progress and discuss the future trend of integrated structured light manipulation in terms of multiple materials, multiple integration techniques, multiple working bands, multiple structured light types, multiple processing functions, multiple detection structures, and diverse application scenarios. We also discuss future opportunities, challenges, and solutions, and give a vision for integrated structured light manipulation. 2.Integrated Structured Light GenerationTo make full use of multiple degrees of freedom of structured light, one of the most basic technologies of structured light manipulation is to design and fabricate emitters that can generate various structured light beams. Integrated photonic devices provide a powerful platform for structured light generation with distinct advantages of light weight, small footprint, high integration, high stability, low power consumption, and low cost. In this section, the basic theories and various types of integrated structured light generation are reviewed, as outlined in Fig. 4. The first part mainly introduces basic theories and principles of structured light generation. The second part mainly presents integrated generation of OAM-carrying light beams on both passive and active platforms. The third part mainly describes an integrated generation of other structured light beams beyond OAM. 2.1.Theories and Principles of Structured Light GenerationIn this part, we introduce the basic theories and working principles of structured light generation, such as spiral phase plate theory, mode-field superposition theory, phase control theory, holographic theory, coupled mode theory, whispering gallery mode theory, non-Hermitian theory, and phased array theory. 2.1.1.Spiral phase plate theoryThe spiral phase plate, a commonly used device for generating OAM-carrying light beams, is a transmissive or reflective optical device with its thickness gradually varying around the center[528–531]. In general, the thickness () of the spiral phase plate at different positions is expressed as[532,533] where and represent the initial thickness () and the step thickness of the spiral phase plate, respectively, and is the azimuthal angle. A different thickness along the azimuthal direction introduces a different propagation phase, leading to the conversion from an input planar phase front to an output helical phase front based on Eq. (1).For the transmissive-type spiral phase plate, the relationship between the thickness and the topological charge of an OAM-carrying light beam, light wavelength in vacuum, and material refractive index is For the reflective-type spiral phase plate, the thickness is independent of the material of the spiral phase plate, and the relationship is written by Note that the spiral phase plate is mainly used for producing OAM-carrying light beams with a helical phase front. Following the similar mechanism, customized phase plates can be also designed to generate other structured light with specific spatial phase structure. 2.1.2.Mode-field superposition theoryBoth vector eigenmode and LP mode can form a set of completely orthogonal mode bases, and LP mode is basically obtained from the orthogonal transformation of vector eigenmode bases[104,534,535]. It is expected that other forms of orthogonal mode bases could be also generated by similar orthogonal transformation processes, such as the currently widely used circular polarization or linear polarization OAM mode bases. The relation for synthesizing circular polarization OAM modes by vector eigenmodes is where represents the phase difference of when two vector eigenmodes are superimposed. The superscript even (odd) of denotes even (odd) vector eigenmodes. The superscript () of mode represents the left-handed (right-handed) circular polarization states, corresponding to the spin angular momentum (SAM) of , with the simplified Plank constant. The subscript is the angular order, corresponding to the topological charge of the OAM mode, and for Eq. (4a). The positive and negative signs of the topological charge of the OAM mode represent the different rotation directions of the helical phase front. The subscript is the radial order.In addition to the circular polarization OAM mode described above, the linear polarization OAM mode is also a set of commonly used mode bases. The relationship between the linear polarization OAM mode and the vector eigenmode, LP mode, and circular polarization OAM mode is where the superscript () of linear polarization OAM modes and LP modes represents the -polarization (-polarization). The superscripts and denote two degenerate LP modes. Equation (5a) shows the synthetic relationship of OAM mode with .2.1.3.Phase control theoryA metasurface composed of subwavelength artificial structures (also known as meta-atoms) enables unprecedented wavefront manipulation and has gradually become an alternative to traditional curved optical components[536–543]. A metasurface manipulates the light field at the micro-nano scale, which is more commonly achieved through phase control[544–547]. At present, the phase control of the light field by a metasurface mainly depends on several basic principles: resonant phase, geometric phase, propagation phase, and compound phase. 1) Resonant phase The resonant phase originates from the resonance between the incident light and unit metasurface structure or periodic array. When the wavelength, phase, and direction of the incident light match the resonant mode supported by the metasurface structure, resonance occurs on the metasurface. Resonance generally brings about a significant change in amplitude, phase, and polarization. At present, common resonances can be divided into the resonance of metal antennas and Mie resonance of dielectric structures due to different materials. The resonance of metal antennas generally comes from the interaction between incident light and surface plasma at the interface between the metal antenna and the interface. Metal antennas are often able to support multiple resonance modes. Mie resonance is a widely studied and utilized scattering mechanism that typically occurs in structures with dimensions ranging from a few tenths to several wavelengths. In addition, for the multi-layer structure of metal, dielectric, and metal, an equivalent magnetic resonance mode is often supported. Moreover, there are more drastic changes in the light field via resonance, which can achieve a larger phase and amplitude modulation in smaller and thinner dimensions. It is beneficial to the integration and miniaturization of the structured light emitter. 2) Geometric phase Light has two typical angular momenta, in which the SAM of light is related to quantum spin and light with circular polarization. When a light beam is circularly polarized, each photon carries SAM, where the positive (negative) sign is corresponding to left-handed (right-handed) circular polarization. In addition, light can also rotate with its spatial phase structure to generate a specific helical phase front and carry OAM. The geometric phase, also known as the Pancharatnam-Berry (PB) phase, originates from the interaction between polarized light beams and anisotropic units. For the phase control of the light field, the geometric phase can make the SAM of the photon interact with the OAM, which is the manifestation of the SAM-OAM interaction. For easy understanding, the Jones matrix is used to simplify its internal physical processes expressed as where and are the - and -polarization components of the incident light, respectively. and are the - and -polarization components of the output light, respectively. M is the Jones matrix of the anisotropic metasurface in Cartesian coordinates written byThe following assumptions are proposed: the anisotropic structure corresponding to the coordinate system and the entire metasurface corresponding to the coordinate system have an in-plane rotation angle of with respect to the axis. Based on the above assumptions, the Jones matrix of the anisotropic metasurface can be further expressed as where and represent the complex amplitudes of the anisotropic structure along its fast and slow axes, respectively. is the rotation matrix or coordinate transformation matrix from the coordinate system to the coordinate system with an in-plane rotation angle of , which can be expressed asThen, by incorporating Eq. (9) into Eq. (8) and simplifying it, we can obtain When the incident light is a circularly polarized one, i.e., , with “+” the left-handed circular polarization (LCP) and “−” the right-handed circular polarization (RCP), the outgoing light field can be expressed as According to Eq. (11), in addition to the same polarization component, the output light also produces a cross-polarization component. In particular, the cross-polarization component carries a geometric phase of , which is twice the in-plane rotation angle. It can be concluded that the geometric phase is only related to the rotation direction of the incident circular polarization light and the rotation angle of the structure, but not to the frequency of the incident light and the material of the structure. Therefore, different from other phase modulations, the geometric phase does not increase with the accumulation of light paths, but is the manifestation of spin-orbit interaction and linked to the evolution of the geometric properties of light, such as polarization and propagation direction. 3) Propagation phase The propagation phase depends on the optical path accumulation of a certain thickness medium. Changes in the structural geometric parameters, shape, and material of the metasurface element will result in changes in its effective refractive index (). When ignoring the reflection of light waves at the interface, the phase of their transmission will change as follows: where , , and are the wavelength, the height of metasurface structure, and the phase, respectively. Therefore, phase modulation can be achieved by controlling the structural geometric parameters of the metasurface. Furthermore, by introducing anisotropic structures into the metasurface, the flexibility of propagation phase modulation can be enhanced. The effective refractive index of metasurface elements with anisotropy varies in each direction: where , , , and are the effective refractive index of , , , and directions, respectively. Based on an anisotropic unite cell, the metasurface further manipulates the polarization of the light field.4) Compound phase According to the above analyses, the propagation phase can be adjusted by changing the size of the subwavelength structure, and the geometric phase can be introduced by changing the rotation direction of the subwavelength structure. Therefore, when the rotation and size of the subwavelength structure are both changed, the geometric phase and propagation phase can be introduced to realize the compound phase modulation. For the sake of analysis, the phase delay introduced by a transparent subwavelength structure along its fast and slow axes is assumed to be . The corresponding complex amplitudes are expressed as and . If it is rotated counterclockwise by an angle, according to Eq. (11), the resulting output light field of a circular polarization light can be expressed as According to Eq. (14), the co-polarized component carries only the propagation phase , while the cross-polarized component carries the compound phase composed of the propagation phase and the geometric phase . According to the principle of propagation phase and geometric phase, the propagation phase has the characteristics of being spin-independent, frequency-dependent, and material-dependent, while the geometric phase with the opposite characteristics is spin-dependent, frequency-independent, and material-independent. The combination of the propagation phase and geometric phase uses their composite control in spin, frequency, and material to solve the problems of circular polarization multiplexing, bandwidth, and tunability of traditional phase metasurfaces in light field control. 2.1.4.Holographic theoryFor the integrated photonic platform, taking the OAM mode generation as a typical example, the holographic pattern is obtained by the interference between the target OAM mode and in-plane waveguide mode[548–554]. The OAM mode and in-plane waveguide mode can be expressed as[555,556] where and are the amplitude of the OAM mode and the in-plane waveguide mode, respectively, is the topological charge of the OAM mode, is the azimuthal angle, is the propagation constant of in-plane waveguide mode, and is the propagation direction along the in-plane waveguide in Cartesian coordinates. The azimuthal angle in polar coordinates is associated with Cartesian coordinates by . The interference between the vertically incident OAM mode and the in-plane waveguide mode forms a hologram written byThe hologram is transferred onto the upper surface of the waveguide. When the hologram on top of the waveguide is illuminated with the in-plane waveguide mode, the out-of-plane OAM mode can be generated. Similarly, other types of holograms might also, in principle, be produced by the interference between the specific structured light and the in-plane waveguide mode, and then transferred onto the upper surface of the waveguide[557–560]. Illuminating the transferred hologram with propagating in-plane waveguide mode could facilitate the flexible generation of out-of-plane specific structured light. 2.1.5.Coupled mode theoryOptical waveguides are composed of media with different dielectric constants, and the waveguide walls serve as interfaces between different media. Although most of the power usually remains inside the waveguide when light propagates, a small portion of the light always propagates near the outside of the waveguide wall. For two waveguides close to each other, when the distance between the two waveguides is small enough, power coupling occurs between them. The coupling characteristics are determined by the distance, length, and propagation mode between the two waveguides[561,562]. The basic formula and amplitude equations of two coupled waveguides can be described as where and are the light fields propagating along the -axis in waveguide 1 and waveguide 2, respectively, which are composed of varying amplitude () and propagation factor (). and are propagation constants of light fields in waveguide 1 and waveguide 2, respectively. The () is the coupling coefficient, which describes the influence of waveguide 1 (waveguide 2) on the propagation light field of waveguide 2 (waveguide 1).In general, only modes with equal or nearly equal propagation constants ( or ) can be effectively coupled. To prove this conclusion, the following assumption is made: there is no light field in waveguide 2 at [i.e., , ], and the light field in waveguide 2 at is calculated according to the coupled wave equation If the difference between two propagation constants () is large, the period of the high-speed oscillation factor is much smaller than the propagation distance , and the integral value of within the integral distance () is very small, resulting in a negligible power coupling between the two waveguides. When the propagation constants () of two waveguides are equal or almost close to each other, the integral result of may reach a certain value, leading to efficient power coupling between the guided modes in the two waveguides. Remarkably, the condition with two equal propagation constants () is called the phase matching condition. The corresponding coupling length for achieving the maximum power coupling can be derived by where . It can be seen that the coupling length depends on the coupling coefficient . The larger the coupling coefficient is, the shorter the coupling length requires, and the more compact the device size is. Taking the mode coupling between the fundamental mode in waveguide 1 and high-order mode in waveguide 2 as a typical example, the high-order mode in waveguide 2 can be effectively excited by the fundamental mode in waveguide 1 when satisfying the phase matching condition with equal propagation constants and choosing the proper coupling length according to Eq. (20).2.1.6.Whispering gallery mode theoryThe whispering gallery modes carrying high-order OAM can be generated in circular optical resonators, such as microrings and microdisks[296,472,563–567]. The periodic micro-etch structure embedded in the microresonator, such as the angular grating structure, can periodically modulate the refractive index in the azimuthal direction to extract the constrained whispering gallery mode in the resonator. Theoretical derivation shows that whispering gallery mode can be emitted from the microring resonator only when the following angle-phase matching conditions are satisfied[296,472]: where is the topological charge of the radiation OAM mode, is the order of resonant whispering gallery mode, and is the number of scattering elements of the microring. The -component propagation constant of the radiation OAM mode is given by where is the wavelength of resonant whispering gallery mode and is the microring resonator radius. Furthermore, the order of resonant whispering gallery mode can be obtained by where is the wavelength-dependent mode effective index of the waveguide. Therefore, the topological charge of emitting OAM mode can be written byBy adjusting the such as introducing a high-order mode, the microring resonance can obtain multiple OAM modes with topological charges at one wavelength. 2.1.7.Non-Hermitian theoryThe Hermitian system is a closed system that does not exchange energy with other systems and does not have gain or loss, and its physical quantities can be described by Hermitian operators[568–572]. The eigenvalues of Hermitian operators are real numbers and their eigenstates are completely orthogonal. On the contrary, the non-Hermitian is an open system that has gain or loss or open boundary conditions, and the eigenvalues of non-Hermitian operators are generally complex. In the non-Hermitian system, there are many novel physical phenomena, such as exceptional points. The exceptional point is a degenerate state of different eigenstates and eigenvalues, which brings many interesting phenomena and applications. For a general two-component system, the non-Hermitian can be described by the evolution equation where the is the system state, is the imaginary unit, and Hamiltonian is expressed as where and are the resonance frequencies, and are the gain/loss, and is the coupling coefficient. The eigenvalues of the system are where and are mean values of the resonance frequencies and gain/loss, respectively, while and are difference values of the resonance frequencies and gain/loss. When an exceptional point occurs, two eigenvalues coalesce, meaning that the square-root term of Eq. (27) is zero. Thus, the requirement for an exceptional point to occur is and .2.1.8.Phased array theoryOptical phased arrays extend the operating frequency band of radio frequency phased arrays to the optical frequency range. This technology controls the direction of a light beam by adjusting the phase of each element in an antenna array. The fundamental principle involves altering the phase difference of transmitted or received signals, causing the light waves emitted by the array elements to coherently combine in space, forming a structured light beam in a specific direction[573,574]. In an optical phased array system, the antenna array consists of multiple independent elements, each with its own transmitting and receiving modules. These modules adjust the signal phase using phase shifters. The core of the optical phased array technology lies in the precise control of the phase difference between each array element. By doing so, the transmitted or received signals interfere constructively in the desired direction, enhancing the signal strength in that direction while suppressing it in others. This precise phase control enables the formation of a well-defined light beam, which can be steered electronically without mechanical movement. Optical phased arrays may find applications in various fields[575–577], including light detection and ranging (LiDAR) systems for autonomous vehicles, optical communications, and advanced imaging systems, where they offer significant advantages in terms of speed, accuracy, and flexibility. 2.2.Integrated OAM GenerationThe OAM-carrying light beam is an optical vortex with a helical phase front, which shows a doughnut intensity profile because of the phase singularity at the beam center. Each photon of the OAM-carrying light beam carries an OAM of , corresponding to the helical phase structure. It is worth noting that the topological charge of OAM can theoretically be taken as any integer (including positive and negative numbers). Therefore, with the infinite state, OAM can, in principle, significantly improve the data transmission capacity in optical communications by OAM modulation or OAM multiplexing. However, the widespread application of photonic OAM remains challenging so far due to bulky devices. Recently, there has been increasing interest in the research of compact chip-scale integrated OAM generation devices. Photonic integration technology has achieved rapid development in recent years. The main material platforms for generating OAM modes include III–V compounds and silicon materials. Among them, due to the peculiarity of direct bandgap materials, the main advantage of III–V group materials, e.g., indium phosphide (InP), is that they can be used to make key active devices such as semiconductor lasers, modulators, and detectors. In contrast, the advantages of silicon materials are their large natural storage capacity, low cost, almost transparency in the near-infrared and even mid-infrared bands, and low material losses. The large relative refractive index difference of silicon-on-insulator (SOI) waveguides is also more conducive to high-density integration of devices. More importantly, silicon materials are compatible with existing mature electrical complementary metal oxide semiconductor (CMOS) processes, which undoubtedly play an important supporting role in the development of silicon-based optoelectronic technology. Given that silicon material is an indirect bandgap material that cannot produce efficient light sources, monolithic integration on a single material system still faces significant challenges. Currently, the main development direction is still hybrid integration technology, which integrates independently produced discrete active devices (such as lasers and detectors) onto silicon-based passive chips through bonding and other methods, thereby achieving low-cost and high-performance hybrid photonic integrated chips. 2.2.1.Integrated OAM generation on passive platformsIn this part, four types of integrated OAM emitters on passive platforms are introduced: in-plane to in-plane OAM generation, in-plane to out-of-plane OAM generation, out-of-plane to in-plane OAM generation, and out-of-plane to out-of-plane OAM generation. 1) In-plane to in-plane OAM generation For the in-plane to in-plane OAM generation, the input light is the in-plane waveguide mode, and the output light (generated OAM) is also the in-plane waveguide mode. 3D waveguide structures are usually used to implement the in-plane to in-plane OAM generation, such as a hybrid device consisting of a silica planar lightwave circuit (PLC) coupled to a 3D waveguide structure, asymmetric directional coupling vortex beam emitter embedded in a photonic chip, and the trench waveguides on a silicon/silica chip. As shown in Figs. 5(a) and 5(b), a hybrid photonic integrated circuit device composed of a silica PLC and a 3D waveguide structure is used to generate 15 OAM modes with TE and TM polarizations and relatively low loss at 1.55 µm[467]. In this work, the 2D silica PLC has a free propagation region (FPR), which is utilized to create linearly tilted wavefronts that correspond to diverse OAM states. Then, the laser inscribed 3D waveguide structure performs the geometric transformation of the linear phase tilt to azimuthal phase variation at the output apertures. In addition, the direct generation of optical vortex beams inside a photonic chip is proposed using a femtosecond laser direct writing technique. As shown in Figs. 5(c) and 5(d), the OAM generator is an asymmetric direction coupler including a standard single-mode waveguide and a doughnut-shaped waveguide in a photonic chip[578]. The Gaussian beam in the single-mode waveguide evanescently couples to the adjacent doughnut-shaped waveguide and converts it into an OAM beam. Another commonly used integrated OAM generator is based on the trench waveguide structure. Due to the asymmetric structure, the single-trench waveguide breaks the original rotation symmetry and supports two orthogonal LP-like modes whose optical axes are rotated by around 45° with respect to the horizontal and vertical directions, as depicted in the inset of Fig. 5(e)[469]. These two nondegenerate LP-like modes have different propagation constants and can be combined into two OAM modes (topological charges of or ) after propagating through different distances, corresponding to phase differences of and , respectively. Because of its simple and easy structure, the trench waveguide can be designed and fabricated not only in photolithographic silicon chips, as shown in Fig. 5(e), but also in femtosecond laser inscribed 3D silica chips, as shown in Fig. 5(f)[470]. 2) Out-of-plane to in-plane OAM generation The in-plane OAM generation can be obtained not only by directly converting the light inside the waveguide but also by coupling the light outside the waveguide into guided mode. For the out-of-plane to in-plane OAM generation, the input light is the out-of-plane light beam, while the output light (generated OAM) is the in-plane waveguide mode. In general, diffraction structures are popular to achieve the out-of-plane to in-plane OAM generation, such as the dielectric antenna waveguide and grating coupler. The universal design strategy is presented to utilize the dielectric antenna waveguide for coupling incident radiation into an integrated waveguide. Figs. 6(a) and 6(b) illustrate the conceptual view and top view of the silicon-nanoantenna-patterned silicon-nitride photonic waveguide[275]. Similar to the design principle in the trench waveguide, the dielectric antenna waveguide generates the optical vortex carrying by synthesizing two high-order and modes with phase difference, as displayed in Fig. 6(c). The left two panels of Fig. 6(d) are calculated output electric field distributions when only the mode antenna and mode antenna exist, respectively. The right two panels of Fig. 6(d) give the electric field and phase distributions when the mode antenna and mode antenna both exist, which indicates the optical vortex generation with [481]. To further verify the mode-control flexibility and configurability, higher-order (from to ) OAM modes can also be implemented by the mode mixing method. Figures 6(e) and 6(f) depict the top views of antenna waveguides and the output electric field distributions for individual antenna arrays and the OAM field inside (and outside) the meta-waveguide, where OAM topological charges are and , respectively[471]. In addition to the dielectric antenna waveguide, the grating coupler can also be used to generate modes within the photonic integrated circuit based on the mode mixing method, as shown in Fig. 6(g). Figure 6(h) gives the numerically calculated amplitude and phase distributions of the field component in the target waveguide for the grating coupler formation with OAM states of and [579]. Compared to other proposed schemes of in-plane OAM generation, the strategy has a simple structure and is compatible with standard planar nanofabrication technology. 3) In-plane to out-of-plane OAM generation For the in-plane to out-of-plane OAM generation, the input light is the in-plane waveguide mode, while the output light is the out-of-plane OAM-carrying light beam. Figure 7 shows four different structures typically used for the in-plane to out-of-plane OAM generation, including a grating array, microring, fork grating, and subwavelength structure. As shown in Fig. 7(a), the silicon photonic integrated circuit consisting of an FPR, tunable-phase arrayed waveguides, and vertical grating arrays is proposed to generate the OAM states over a topological charge range from to [580]. At the output of FPR, the wavefront acquires the tilt of the linear phase based on the different input waveguide positions. Then, after the propagation through the tunable-phase arrayed waveguides, the array gratings maintain the phase relationship and generate the corresponding OAM mode. Another structure with array gratings is shown in Fig. 7(b), including one bus waveguide, one microring, and eight download units, where each download unit includes a grating and an arc waveguide. When the resonance condition of a microring is satisfied, the whispering gallery mode in the microring can be downloaded and transformed into OAM mode by the eight download units[581]. To further reduce the footprint of the device, the silicon-based integrated optical vortex microring emitters use angular gratings to extract the OAM beams confined in whispering gallery modes. The radius of the smallest microring is 3.9 µm. As shown in Fig. 7(c), three identical microring emitters can form an OAM emitter array that is coupled to the same access waveguide[472]. Then, based on such angular-grating-assisted microring emitter, a compact vortex emitter with the ability to actively tune different OAM modes is demonstrated using a single electrically contacted thermo-optic control, as shown in Fig. 7(d)[473]. The thermo-optic tuning of the OAM modes achieves the on-off keying at rates of 10 µs and OAM mode switching at rates of 20 µs. Furthermore, in order to address the issue that only two OAM modes can be excited by clockwise and counterclockwise whispering gallery modes, the vortex emitter with a multimode microring is designed and fabricated, which can convert four in-plane waveguide modes into four free-space vector OAM beams with high mode purity. Figure 7(e) displays the schematic of the vortex emitter with a multimode microring and the and modes in the multimode microring[474]. Except for the high-order horizontal waveguide, high-order vertical modes can also be used to increase the OAM number of microring emitters, as shown in Fig. 7(f)[475]. The holography method can also be used to enable the in-plane to out-of-plane OAM generation, as explained in Fig. 7(g)[550]. In the inset of Fig. 7(g), when the target OAM mode is incident vertically to the waveguide propagating an in-plane guided mode (from left to right), there is a string with dark and bright patterns caused by the interference of the two light fields (incident OAM mode, in-plane waveguide mode). The numerically calculated results show a guiding wave propagating along the waveguide with a holographic fork grating producing the target OAM mode with high purity in the broadband visible wavelength range. The holographic fork grating can also be introduced on the top of the silicon waveguide to generate OAM mode. The experimental results show that the silicon waveguide surface holographic grating produces not only a single OAM mode but also a synthetization of two OAM modes with opposite topological charges, as shown in Fig. 7(h)[551]. In addition, the holographic fork grating is designed with the focused forked grating to reduce the feed length and with apodization to improve OAM mode fidelity, as shown in Fig. 7(i)[552]. For the high-order OAM, the multi-waveguide holographic gratings are proposed to generate the high-order OAM modes with a topological charge from to by the manipulation and configuration of the incident light phase, as shown in Fig. 7(j)[553]. Subwavelength structures with advanced algorithms are also used to enable the in-plane to out-of-plane OAM generation. A 1.2-μm-radius circular silicon device divided into 288 subpixels is proposed to emit the ultra-broad bandwidth modes, as shown in Fig. 7(k)[582]. To achieve an accurate phase modulation, the subwavelength structure is optimized via the global optimization method combining the genetic algorithm and the annealing algorithm. The experimental results show OAM emission efficiencies of and OAM mode purity of . Furthermore, studying the relationship between the input phase delays, the subwavelength structure fed by four single-mode waveguides is presented to generate the tunable OAM beam. Figure 7(l) depicts the concept of the OAM beam generation using phase delays. The relative phase delay decides the generated OAM order. The subwavelength structure tailors the transformation function with a direct binary search (DBS) algorithm to support different tunable OAM order ranges[583]. Although the above OAM generators based on the holographic fork grating show distinct performance, most of them have either single-polarization operation or narrow bandwidths, or complicated structures with a relatively large footprint. The ideal OAM generator, compatible with existing physical dimensions of light, offers impressive features of polarization diversity, broadband, and ultra-compact footprint. A multiport fork grating OAM generator with a simple two-dimensional (2D) surface structure (superposed holographic fork gratings) is presented and illustrated in Fig. 8(a). As shown in Figs. 8(b) and 8(c), two different direction fork gratings [ and ] are formed on top of the silicon waveguide by the interference between the vertically incident -polarization (-polarization) OAM mode and the -polarization (-polarization) in-plane mode. Then, the superposed holographic fork gratings are acquired by the superposition of two fork gratings [G1 (x, y) and G2 (x, y)]. Finally, the four polarization diversity OAM modes (-polarization , -polarization , -polarization , and -polarization ) are realized by different incident conditions of the silicon waveguide with superposed holographic fork gratings. As shown in Fig. 8(d), the polarization diversity OAM generator has a favorable performance in terms of purity and operating bandwidth. Figure 8(e) is the measured SEM image of the ultra-compact fork grating, which has a superposed footprint of . Figure 8(f) displays the far-field intensity profiles and interferograms for the polarization diversity OAM modes in the C-band. To evaluate the influence on each other for the OAM mode generation, Fig. 8(g) demonstrates the measured crosstalk matrix and the real accumulated crosstalk. The worst accumulated crosstalk is approximately . The obtained simulated and measured results depicted in Figs. 8(a)–8(g) show that the ultra-compact polarization diversity OAM generator with a footprint is implemented on the silicon platform with a broadband working wavelength[555]. To improve the emitting efficiency, as shown in Fig. 8(h), a backside metal mirror is placed below the device layer of the superimposed holographic fork gratings, to reflect the energy leaking down to the device substrate back to air. The layer thickness of sandwiched between the holographic fork grating layer and the mirror is optimized to ensure the higher emitting efficiency of the upward-emitting OAM beam. The measured powers of different OAM modes with/without a mirror depicted in Fig. 8(i), show that the backside metal mirror achieves an improvement of in the emitting efficiency[556]. Another subwavelength structure using the DBS algorithm is the digitized subwavelength surface structure, which enables the generation of wavelength/polarization-/charge-diverse OAM mode. Figures 9(a) and 9(b) depict the digitized nanostructure with a footprint and the concept of an OAM generator for generating the -polarization . Figure 9(c) shows simulation results of intensity profiles, interferograms, and phase distributions for -polarization . Figures 9(g) and 9(i) show the fabricated OAM generator with a footprint and measured mode crosstalk matrix. One can see that the worst crosstalk is less than . Similar design, fabrication, simulations, and measurements are carried out for generating the -polarization , as shown in Figs. 9(d)–9(f), 9(h), and 9(j). Moreover, the digitized subwavelength structure also shows the possibility of the generation of much higher-order OAM modes (e.g. -polarization ), as shown in Figs. 9(k) and 9(l). The demonstrated optical OAM generator provides the possibility of multi-dimensional optical communications with efficient capacity scaling[584]. 4) Out-of-plane to out-of-plane OAM generation For the out-of-plane to out-of-plane OAM generation, the input light is the out-of-plane light beam, while the output light is also the out-of-plane OAM-carrying light beam. There are several primary design proposals using the compact devices to implement the out-of-plane to out-of-plane OAM generation, including a metasurface, spiral phase plate, and holographic grating. The metasurface has drawn extensive attention for its ability to manipulate light flexibly (phase, amplitude, polarization) over subwavelength propagation distances. Therefore, the metasurface offers unparalleled opportunities for OAM generation in terms of reliability, flexibility, and miniaturization. Most of the metasurfaces are divided into two categories according to the material of a unit cell, including plasmonic metasurfaces and dielectric metasurfaces. For the plasmonic metasurface, there are two main operating principles: resonant phase and geometry phase. The resonant phase metasurface depends on the resonance from the unit cell or periodic array structure of the metasurface. When the wavelength, phase, and direction of incident light match the resonant mode supported by the metasurface, the resonance generally causes a more obvious change in amplitude, phase, and polarization. A V-shaped phase antenna is designed, which consists of two gold nanorods of equal length with a certain angle between them. Two resonance modes (symmetric and antisymmetric) are excited when the polarized beam is vertically incident. By adjusting the length of the antenna arm and the angle between the two arms, the amplitude and phase of the beam orthogonal to the incident polarization can be controlled. When the antenna is set to eight different dimensions as shown in Fig. 10(a), the full-wave simulation indicates that the transmission amplitude keeps unchanged and the phase changes from 0 to with increments. Thus, the eight antennas achieve all-phase manipulation. Figure 10(b) exhibits the SEM image of a representative antenna array. When the eight antennas are arranged into the array distribution as depicted in Fig. 10(c), the phase distribution of the eight antennas changes around the center. When the linearly polarized beam illuminates the array, the OAM beams with the topological charge of are generated, where its polarization state is orthogonal to the incident polarization direction[585]. The geometric phase originates from the interaction between the polarized beam and the anisotropic unit cell. At the same time, through the control of the light field, the geometric phase achieves the interaction between the SAM and OAM of the photon. Different from the traditional dynamic phase, the value of the geometric phase is determined by the rotation angle of the anisotropic unit cell and the polarization state of incident light, rather than the accumulation of light paths. An L-shaped metal metasurface is proposed, as shown in Fig. 10(d). The L-shaped structural elements, as a subwavelength birefringent device, are distributed in a circular rotation on a glass substrate, which realizes the phase shift. Figure 10(e) illustrates the SEM image of the fabricated L-shaped metasurface. The vortex beams with the topological charge of are generated by the vertically incident circularly polarized light with a wavelength of 780 nm, as shown in Fig. 10(f). The experimental results confirm the conservation of angular momentum of the photon, as well as the conversion from SAM to OAM realized through the metasurface[586]. Based on the PB phase, the phase distributions of the axicon, spiral phase plate, and Fourier transform lens are superposed to generate focused 3D broadband OAM beams, as shown in Fig. 10(g). The obtained OAM beams with different topological charges have the same beam divergence and almost constant vortex ring radius. The OAM beam is easily adjusted by the parameters of metasurface design, i.e., axicon period, operation wavelength, and lens focal length. Figure 10(h) demonstrates the interferometry patterns of OAM with topological charges at 633 nm, when the axicon periods are 8 and 4 µm, respectively. Furthermore, the superimposed metasurface is designed to produce multiple OAM beams with different topological charges and arbitrary location arrangements, as shown in Fig. 10(i)[587]. In addition, there are some different metasurfaces to produce the specific OAM-carrying light beams. The compact metal-assisted metasurface is proposed as a broadband OAM-carrying vector beam emitter, which modulates the phase and polarization of the light beam simultaneously. The designed metasurface is composed of rectangular holes etched on the gold film, as shown in Fig. 11(a). 84 rectangular holes are distributed in two rings, and the geometric phase is introduced and varied by changing the rotation direction of the rectangular holes. Each rectangular hole is used as a separate linear polarizer due to the local waveguide resonance, which converts the incident circular polarization state into a vector spatial polarization distribution, and generates an OAM-carrying vector beam with topological charges from to , as illustrated in Fig. 11(b)[588]. A reflection-enhanced plasmonic metasurface is designed and fabricated to efficiently generate a structured light with a phase helix and intensity helix at 2 µm, as shown in Fig. 11(c). A self-reference intensity helix against ambient noise is generated without the use of spatially separated light[589]. Besides, a metasurface structure on the facet of large-core fiber is proposed and designed to excite the linearly polarized and circularly polarized beams from either the meta-facet side or planar-facet side, as shown in Fig. 11(d). Figure 11(e) shows the measured phase purity of twisted light excited from the meta-facet fiber, which is higher than 93% within a broadband wavelength range from 1480 to 1640 nm with incident linearly polarized light beams[590]. Although a plasmonic metasurface, based on its ability to abruptly change the phase, allows subwavelength optical elements to generate the target OAM beams, most previously proposed metasurface designs suffer from low coupling efficiency due to the inevitable ohmic loss of metal at optical wavelengths. In contrast, a dielectric metasurface consisting of high-refractive-index dielectric nanoparticles provides one potential solution to deal with the ohmic loss, which relies on three main principles, i.e., resonant phase, geometry phase, and the combination of the geometry phase and propagation phase. The dielectric metasurface based on Mie resonances in the silicon cut-wires is designed and fabricated on an ultra-thin silver film. By adjusting the size and rotation angle of the silicon cut-wires, the phase shift of can be realized for the reflection component orthogonal to the incident polarization. Arranging these arrays of silicon columns, as shown in Fig. 12(a), produces OAM beams with topological charges of [591]. Efficient phase control in the microwave region can be rendered by the appropriate adjustment of the size and rotation of the unit resonators. Figure 12(b) depicts the microwave resonant dielectric metasurface for broadband high-efficiency OAM beam generation[592]. The all-dielectric metasurface can also rely on the PB phase (geometric phase) to generate the OAM beam, which has the advantages of broadband and robust modulation. The all-dielectric metasurface consists of a ring subwavelength grating, as shown in Fig. 12(c), which can be regarded as a spatially varying PB phase optical element[593]. The efficiency of the beam generation is as high as 90%. The generation of a fractional, quasi-focusing, and quasi-non-diffractive OAM beam is also achieved by the designed all-dielectric metasurface with width-modulated concentric ring gratings. Another metasurface is composed of nanopillars for the generation of perfect vortex beams (PVBs), as shown in Fig. 12(d)[594]. The measured OAM beams with different topological charges exhibit the same radial intensity profiles, confirming their “perfect” characteristics. The structure parameters (i.e., diameter and ellipticity) of the PVB can be simultaneously adjusted by changing the structural parameters of the metasurface. Although the geometric phase metasurface performs the generation of versatile OAM beams, the cross-polarization is only affected by the PB phase and not modulated simultaneously with the phase and amplitude. To break this restriction, a unified strategy using the combination of geometric and propagation phases is presented. By utilizing the above incorporative phase modulation approach, all the states of perfect Poincaré beams (PPBs) on the hybrid-order Poincaré sphere are implemented by an all-dielectric metasurface, as shown in Fig. 12(e)[595]. By changing the SAM of the incident light, metasurface devices can generate spin-multiplexed perfect optical vortices with arbitrary OAM. Furthermore, an all-silicon metasurface is presented to generate multiple OAM modes and arbitrary superpositions of OAM modes. The metasurface includes anisotropic unit cells and isotropic unit cells, which introduce the phase of modulation by the PB phase and propagation phase, respectively. As shown in Fig. 12(f), the left-handed and right-handed circular polarization can be used to generate multiple OAM states in three independent polarization channels[596]. In addition, the superposition states of OAM can also be generated in longitudinal polarization due to the conversions of arbitrary SAM to OAM. The longitudinally varying vector OAM beams are generated in the terahertz band using an all-dielectric metasurface. As shown in Fig. 12(g), the topological charge of the generated OAM beam evolves from to (from to ) as the propagation distance increases, under the incidence of left-handed (right-handed) circularly polarized waves. Figure 12(h) depicts the simulated intensity and phase profiles of longitudinally varying vector OAM beams[597]. In addition to metasurfaces, a spiral phase plate and holographic grating are also widely used to implement out-of-plane to out-of-plane OAM generation. As shown in Fig. 13(a), the reflected or transmitted light has a spiral-shaped wavefront, when a plane wave is incident on a spiral phase plate. An axially controllable multiple OAM generator is presented, which is based on a continuous surface profile structure with independent phase control[598]. The device depicted in Fig. 13(b) is fabricated by femtosecond laser additive manufacturing[599]. Numerical simulation and experimental measurement demonstrate that the parameters of collinear multiple OAM (i.e., the axial position, the rotation direction, and the number of topological charges) are arbitrarily controlled by designing different phase distributions. Furthermore, the spiral phase plate can be fabricated on the end face of the single-mode fiber, to efficiently produce the OAM beam, as shown in Fig. 13(c)[464]. The spiral phase plate can also be fabricated on the end face of the PbSe-doped ring-core fiber to implement the OAM generation, OAM transmission, and OAM amplification, as shown in Fig. 13(d)[465]. The obtained results indicate that the combination of spiral phase plate and fiber is beneficial to all-fiber OAM mode manipulation. The holographic fork grating obtained by interfering with the optical vortex with the plane wave can also realize the generation of out-of-plane to out-of-plane OAM light beams. When Gaussian light is incident on the fork grating hologram, the grating generates different OAM modes at different diffraction orders. Remarkably, the holographic grating can also be manufactured on the end face of the few-mode fiber. The difference is that the holographic fork grating converts the input Gaussian-like beam in the low-order diffraction directions to high-order OAM modes in fiber, which is suitable for the all-fiber OAM (de)multiplexing, as shown in Fig. 13(e)[466]. A combined subwavelength grating consisting of the holographic fork grating and blazed grating is presented to implement the superposition of two optical vortex beams with controlled topological charges, as shown in Fig. 13(f)[600]. Moiré patterns formed in twisted bilayer two-dimensional van der Waals materials have drawn great attention in the observation and exploration of emerging electronic and photonic properties. Moreover, the concept of moiré patterns has also been extended to the twist of photonic structures, e.g., twisted photonic crystal, which broadens the field of moiré photonics and paves the way for novel application of moiré physics. Very recently, twisted stacking of two-layer photonic crystals is proposed to implement robust OAM-carrying optical vortex generation through BIC, which is insensitive to the incidence angle and the illumination position of the incident light. Figure 14(a) shows the schematic of the twisted bilayer photonic crystal. It is formed by stacking two planar photonic crystal slabs with a small twist and each slab is a honeycomb array of silicon nanodisks. The BIC modes in one slab are coupled to the guided resonances in the other slab, which are coupled to the free space. The unique interlayer channel created by the moiré structure (so called “moiré channel”) allows energy transfer between BIC modes and the free space via guided resonances. To verify the existence of the moiré channel and evaluate the radiation performance, full-wave numerical simulations are conducted on a representative twisted bilayer photonic crystal with a 300-nm spacing and a 1.5° twist angle, which is excited by a pulsed Gaussian beam. As shown in Fig. 14(b), two peaks marked as “i” and “ii” are identified in the mode intensity spectrum. Figure 14(c) shows light field distributions of the optical vortex emission at peaks “i” and “ii”. The doughnut-shaped amplitude profile and helical phase distribution confirm the generation of OAM-carrying optical vortices, corresponding to topological charges of 2 for peak “i” and 1 for “ii”. In addition, the interlayer spacing and twist angle between the two layers are also varied to show the robustness of the optical vortex emission against disturbances. Figures 14(d) and 14(e) show light field distributions under different twist angles and interlayer spacings. In all cases, the topological charges of emitted optical vortices remain 1, indicating the scheme is insensitive to experimental imperfections (e.g,. fabrication errors, thermal expansion effects) and can be used as a stable platform for robust OAM generation[601]. 2.2.2.Integrated OAM generation on active platformsFor the integrated OAM generation devices on passive platforms, an external separate light source is always required. Recently, integrated OAM generation devices on active platforms have also attracted increasing interest in many emerging OAM-enabled photonic technologies that do not rely on external light sources. As one typical active integrated OAM generation device, the integrated OAM laser is an important and valuable research field for diverse OAM generation. 1) Spiral phase plate laser For the integrated OAM lasers, the straightforward method is to integrate a passive OAM generator and a compact laser on a single chip. The vertical cavity surface emitting laser (VCSEL) has numerous advantages and is suitable for integrating passive OAM generators. Figure 15(a) is the schematic of the OAM laser combining the VCSEL and the spiral phase plate array, which can emit 64 OAM modes. The topological charges of 64 OAM modes can be changed arbitrarily by adjusting the corresponding individual spiral phase plate parameter. Figure 15(b) is the spiral phase plate structure, phase distribution, and intensity profile of , , and modes[602]. Figures 15(c) and 15(d) show the top view SEM images of different spiral phase plates, intensity profiles of generated OAM modes, and simulated intensity and phase profiles of individual OAM modes and their superposition[603]. One can see that the measured intensity profiles are in good agreement with the simulated intensity profiles of OAM modes. In the experiment, the design of VCSEL integrated with a spiral phase plate can achieve low-cost wafer-scale fabrication, which is attractive for various OAM array applications. 2) Non-Hermitian-controlled laser Microrings are considered by many OAM lasers because the microring cavity supports whispering gallery modes with high-order OAM modes. However, owing to the rotation symmetry, the active microring structure can simultaneously excite clockwise and counterclockwise whispering gallery modes, i.e., two generated OAM modes carry two contrary topological charges. The result is that the carried OAMs from simultaneously excited whispering gallery modes cancel each other. Consequently, in order to obtain an individual OAM, it is imperative to introduce a mechanism to break the rotation symmetry. Non-Hermitian symmetry with gain or loss is an emerging and promising solution, which breaks the condition of rotation symmetry and lifts the degeneracy state of two spin-orbit modes, which is beneficial to the generation of an OAM beam of desirable chirality. By exploiting the exceptional point of a non-Hermitian system, an OAM microlaser with tunable topological charge is present, which gives full play to the advantages of the semiconductor microlasers and non-Hermitian system. Figure 16(a) displays the schematic of the OAM microlaser with a tunable topological charge. The OAM laser consists of a microring resonator and a bus waveguide with two control arms. By adjusting the gain (optically pumping) and loss (material loss), the OAM microlaser can operate at an exceptional point, to break the chiral symmetry and control the chirality of the OAM mode. The left-handed (right-handed) circular polarization with () is observed when only the right (left) control arm is illuminated by the control beam, as shown in Fig. 16(b)[604]. The multiple different OAM states are produced by the non-Hermitian-controlled OAM microlaser. Furthermore, the fast switch of the fractional OAM is demonstrated based on the OAM semiconductor microlaser and a control laser pulse, as shown in Fig. 16(c)[605]. In a 100-ps time window, a continuous fractional OAM sweep between the topological charge 0 and is implemented. Besides, a hyperdimensional spin-orbit microlaser consisting of two similar microrings generates the spin-orbit OAM in a four-dimensional Hilbert space, as shown in Fig. 16(d)[606]. Another scheme for adjusting the gain and loss is the modulation of the complex refractive index. The complex refractive index is achieved by the alternate Ge and Cr/Ge structures placed on top of the InGaAsP microring, as shown in Fig. 16(e)[607]. When the device works at an exceptional point, the microring laser also produces a single-mode OAM-carrying vortex laser beam. A unidirectional resonator including a microring resonator and an S-shaped waveguide is present to directly generate a tunable OAM, as shown in Fig. 16(f)[608]. An S-shaped waveguide located inside the microring offers the gain or loss to break two clockwise and counterclockwise eigenmodes. The topological charge of generated OAM modes can be precisely adjusted by temperature adjustment. 3) Micro-etching laser Some integrated micro-etching lasers are presented to generate the OAM mode. Ge micro-gears on silicon pillars are fabricated on the Ge on silicon on an insulator wafer to obtain an OAM beam, as shown in Fig. 17(a). There are some peak resonances within the Ge direct gap, as shown in Fig. 17(b)[609]. Using finite difference time domain (FDTD) simulation, the resonant peak of the micro-gear spectrum is studied and determined to be a vertical emission optical mode with a non-zero OAM beam. Additionally, a customizable chiral (i.e., clockwise or counterclockwise vortex) VCSEL OAM laser is proposed and demonstrated. This laser based on the optical breaking of time-reversal symmetry is formed by a semiconductor planar microcavity embedded with a single quantum well. Figure 17(c) shows the schematic and SEM image of a benzene-like OAM laser with the cavity etched to form hexagonal rings of coupled micropillars. Figures 17(d) and 17(e) depict phase distributions of the emission, showing a counterclockwise and clockwise phase vortex under and circularly polarized pumps, respectively[610]. The subwavelength etching metasurface placed on the top of the microring can also break the degeneracy of two counter-propagating whispering gallery modes in the active microring resonator, resulting in the direct generation of OAM mode, as shown in Fig. 17(f). A nanobeam of metal-dielectric-metal sandwiched structure is designed as the meta-atom, as shown in Fig. 17(g), which can evanescently couple with guided waves propagating in the waveguide and offers an phase shift. As shown in Fig. 17(h), the meta-atom is designed with uniform amplitude and phase shift covering the entire phase range. Figure 17(i) shows the simulated and measured results for the far-field intensity distribution and the self-interference patterns of the OAM mode, showing the impressive quality of the OAM beam generated by the metasurface laser[611]. An electrically pumped OAM laser is proposed and fabricated on an InGaAsP/InP multiple quantum well (MQW) epitaxial wafer, which is integrated by a distributed feedback semiconductor laser and an OAM emitter based on a microring, as depicted in Fig. 17(j). The micro-etching grating on the top of the microring waveguide extracts OAM beams from the MWGs in the microring resonator. This OAM laser effectively avoids the counter-propagating mode degeneracy, allowing for unidirectional OAM mode excitation from the whispering gallery mode. Figure 17(k) shows the mode purity values of the measured right-handed and left-handed circularly polarized components, which are 0.76 and 0.85, respectively[612]. 4) Topological laser The topological photonics originates from topological phases and phase transitions in condensed matter. Since Haldane and Raghu introduced band topology to the realm of photonics in 2008[613,614], topological photonics has attracted widespread research attention in integrated optics, nonlinear optics, and quantum optics, and can also be applied in the OAM microlaser to generate high-quality OAM beam. Figure 18(a) shows the schematic of the generated spin-momentum-locked edge mode from the topological OAM laser. Figures 18(b) and 18(c) are the SEM and zoom-in SEM images of the topological OAM laser fabricated on an InGaAsP MQW membrane. The x-shaped cavity geometry with multiple bends is designed to demonstrate the robust immune backscattering of the topological edge states at topological interfaces. Owing to the finite propagating cavity length, the edge modes are discrete, as shown in Fig. 18(d). Compared to the conventional cavity mode and whispering gallery mode in microrings, the edge modes have a near zero in-plane wave vector, facilitating directional surface emitting, as shown in Fig. 18(e). Similar to the microring resonator, the designed complex cavity also supports two topological edge modes with opposite spins, and the lasing spectra are shown in Fig. 18(f). When the pump of the laser is working, the spin-momentum-locked topological edge mode that is not coupled to its backpropagation wave can saturate gain to restrain other mode lasers, thus breaking the degeneracy of eigenmodes. Consequently, Figs. 18(g) and 18(h) show the off-center self-interference of the generated mode and side-mode suppression ratio (SMSR) versus the pump intensity, indicating that the topological edge laser has the feature of single-mode operation when pumped for lasing[615]. In addition, the topological photonics can be used to design the low-threshold vortex and anti-vortex nanolasers. As shown in Fig. 18(i), the important part is a symmetric optical cavity formed by a topological disclination. Figures 18(j) and 18(k) show the SEM images of the fabricated vortex and anti-vortex nanolasers, respectively. To demonstrate the unique characteristics of the two types of lasers, Figs. 18(l)–18(n) present polarization-resolved mode images[616]. The top left corner shows the mode field intensity distribution of all laser modes without a polarizer, revealing a doughnut-shaped pattern. The center null intensity is a distinctive feature of the vortex/anti-vortex modes. Furthermore, the polarization-resolved mode images for each mode are fully distinguishable. 5) Supersymmetric microlaser arrays Coupling multiple lasers to form an array is a common approach to enhance the output light of laser systems. In large-scale phase-locked laser arrays, complex nonlinear scaling often adversely affects performance, leading to instability and potentially damaging the laser cavity. To address this issue, a higher-dimensional supersymmetric form is proposed for precise mode control and nonlinear power scaling. As shown in Figs. 19(a) and 19(b), a array of identical microring lasers, composed of multiple InGaAsP quantum wells, exhibits phase-locked coherence and synchronization of all evanescently coupled microring lasers. This uniformly coupled 2D array achieves high radiance, low divergence, and single-frequency laser emission, enhancing the emission power by approximately 25 times compared to a single microlaser, as shown in Fig. 19(c). Moreover, the slope efficiency of the laser array is 26.3 times that of a single microlaser, and it also exhibits a lower lasing threshold. Furthermore, Fig. 19(d) demonstrates the superposition state of left- and right-handed circularly polarized fields, with a continuously varying phase delay between them, resulting in a vector beam with radial polarization. Using an appropriate combination of a quarter-wave plate and a linear polarizer, the two vortex beams with opposite spin-orbit relationships can be effectively separated, as shown in Figs. 19(e) and 19(f)[617]. A dark stripe is clearly observed in the figures, confirming the topological charge of the OAM light as . 2.3.Integrated Structured Light Generation beyond OAMAbove, we mainly elaborate on the generation of OAM light beams on different photonic integration platforms. Remarkably, OAM-carrying light beams are just one kind of structured light with helical phase fronts. In recent years, more general structured light has also attracted increasing interest, including the integrated structured light generation beyond OAM. 2.3.1.Integrated chiral light generationChiral light, a special light field, has drawn much attention due to its geometric properties that cannot be coincident with its mirror image by translation or rotation operation[618–621]. One important aspect of studying photonic chiral behavior is the efficient generation of the chiral light field. The chiral optical microcavity and nanocavity are promising integrated schemes. As shown in Fig. 20(a), the resonant chiral metasurface composed of a square lattice of tilted uses the chiral quasi-bound states in the continuum (BIC) to achieve the effective and controllable emission of circularly polarized light. The metasurface has two deformation (asymmetric) parameters (in-plane) and (out-of-plane), as shown in Fig. 20(b). The symmetry disruption of the combination of two parameters causes the match between the chiral filtering effect of BIC and the frequency of partially BIC radiation. Figure 20(c) depicts the emission spectra of LCP and RCP at the pumping densities of 16.5 and , respectively. When the pumping density is , the LCP mode intensity at 612.08 nm significantly increases and dominates the emission spectrum. Figure 20(d) is the integrated output intensity of the LCP and RCP components versus the pumping density. When the pump density is greater than , a significant increase in intensity can be observed. Figure 20(e) is the fitted degree of polarization (DOP), absolute DOP, and full width at half maximum (FWHM) of LCP light versus the pumping density. The absolute value of DOP rapidly increases near the transparency threshold and saturates at 0.948 to 0.989, while the linewidth of the emission peak decreases from 1.0 to 0.18 nm at the transparency threshold. The divergence angle further decreases to , as shown in Fig. 20(f). The above experimental results demonstrate a resonant metasurface, which generates a chiral light with a purity of [622]. 2.3.2.Integrated LP mode generationLP mode is one kind of special light beam in radially symmetric refractive index distribution fibers or waveguides with weakly guiding approximation, which is also the solution of complex electric field wave equations in cylindrical coordinates. In optical communication systems, different LP modes can also be used as different data channels to further improve the transmission capacity of optical communications. An optical volume element is proposed to generate different LP modes under different input angles. For a micro-scale multilayer design, an optimization method named learning tomography is adopted, which is a reconstructed method of 3D phase objects. Figures 20(g) and 20(h) are the schematic and SEM images of the three-layer volume element. When the incidence angle is (), the output light of the volume element is () mode, as shown in Figs. 20(i) and 20(j)[623]. In addition to these two LP modes, other higher LP modes can also be generated. 2.3.3.Integrated Laguerre-Gaussian/Hermite-Gaussian beam generationThe Laguerre-Gaussian (LG) beam and Hermite-Gaussian (HG) beam both play an important role in laser optics. Although the symmetries of the two types of beams are different, HG functions can be expressed as a linear superposition of LG functions, and vice versa. In particular, the LG mode characterized by the radial index and the azimuthal index has attracted more and more attention due to its ability to carry OAM. For the generation of LG/HG mode, higher-order LG/HG modes can be generated by a metal metasurface composed of a series of rectangular nanoholes with different orientation angles, as shown in Fig. 21(a). Relying on the PB phase, the metasurface has a broadband operation wavelength. Figure 21(b) is the simulated phase and intensity distribution of the high-order LG mode. Figure 21(c) shows the measured intensity distributions of different LG and HG beams ( and )[624]. Based on the PB phase, the metasurface consisting of Au unit cells is also designed to produce multiple LG modes and separate them, as shown in Fig. 21(d). The overall SEM image and zoom-in SEM image of the fabricated metasurface are shown in Fig. 21(e). The metasurface has a favorable performance over a 400-nm-wide wavelength range, as shown in Fig. 21(f). In principle, the metasurface can be designed to distribute the light energy into a greater number of modes. Figure 21(g) shows the simulated and measured diffraction patterns of the combined mode of at 1030 nm with favorable performance[625]. 2.3.4.Integrated non-diffracting beam generationA non-diffracting beam refers to a light beam with a small center spot diameter that does not vary with propagation distance[626]. Mathematically, the non-diffracting beam is the solution of a new set of Maxwell wave equations, and its light intensity and propagation distance are two irrelevant variables. This type of beam can self-recover its light field distribution after encountering obstacles, which means it has the characteristic of self-reconstruction. With the development of laser technology, researchers have found a variety of diffraction-free solutions that theoretically meet the wave equation, such as the Bessel beam, Mathieu beam, Airy beam, needle beam, pin beam, etc. 1) Integrated Bessel beam generation The Bessel function is the accurate solution of the free-space wave equation of a non-diffracting beam in cylindrical coordinates. The zeroth-order Bessel beam exhibits a circular bright spot with the highest central light intensity on the cross-section perpendicular to the propagation direction, surrounded by many rings, and the intensity of the concentric rings gradually weakens as the radius increases. For the higher-order Bessel beam, the central intensity is zero, which is a circular central dark core structure. Due to its non-diffraction and self-healing properties, Bessel beams have wide applications in particle manipulation, optical alignment and imaging, laser microfabrication, nonlinear optics, and information extraction. In order to further realize these applications, the generation of Bessel beams is particularly important. The typical generation schemes based on integrated devices include metasurfaces, phased arrays, phase plates, 3D structures, and concentric distributed grating arrays. Using metasurfaces, the operation wavelength covers the whole visible spectrum, as shown in Fig. 22(a)[627]. The metasurface also utilizes the functionalities of Dammann gratings and axicons to generate polarization-independent Bessel beams, as shown in Fig. 22(b)[628]. The measured experimental results indicate that the efficiency is 66.36% and the FWHM is about 1600 nm. As shown in Fig. 22(c), optical phased arrays integrated on a CMOS-compatible platform are presented to generate the Bessel beam with an FWHM and an Bessel length[496]. The polymer phase plate fabricated by the femtosecond laser direct writing technique is also an attractive solution to produce a high-quality Bessel beam, as shown in Fig. 22(d)[385]. The polymer-based phase plate has a miniature size of . The experimental results demonstrate the 800-mm propagation invariance of the zeroth- and higher-order Bessel beam. The femtosecond laser direct writing technique can also be used to process a 3D structure device on the end face of the fiber to generate a Bessel beam. As shown in Fig. 22(e), the 3D structure includes a helical axicon, a parabolic lens, and a mechanical holder frame[629]. After a 2-mm propagation, the pattern of the generated Bessel beam retains a nearly unchanged size. Based on an SOI standard process, an 870-μm-diameter ring structure with concentrically distributed grating arrays is present to implement a 10.24-m propagation invariance of a broadband Bessel beam, as shown in Fig. 22(f)[630]. The 10.24 m is a long-range distance, facilitating the measure of rotation speed and distance. 2) Integrated Mathieu beam generation The Mathieu beam is the solution of the Helmholtz equation for a non-diffracting beam in elliptical cylindrical coordinates. The Mathieu beam also has the typical features of the non-diffraction beam, i.e., non-diffraction and self-reconstruction. The difference is that the transverse intensity is lattice-like for the Mathieu beam. In general, the distribution pattern of the Mathieu beam is controlled by two parameters, i.e., order and ellipticity . Compared with the Bessel beam, the Mathieu beam with a complex structure is more difficult to generate. Thanks to the flexible 3D manufacturing capability of the femtosecond laser direct writing technique, a helical phase plate, as shown in Fig. 22(g), can be designed and fabricated to obtain the Mathieu beam with different orders and ellipticities . Figure 22(h) shows the intensity distribution of the even Mathieu beam with and at different distances, which indicates that the generated Mathieu beam has approximately no diffraction within the 800-mm propagation range[386]. 3) Integrated Airy beam generation The Airy beam is a type of self-accelerating optical beam and a non-diffracting solution of the paraxial Helmholtz equation. Interestingly, the Airy beam shows a parabolic trajectory during its propagation. Thus, except for the peculiarities of the non-diffraction beam, the Airy beam also shows the feature of self-bending. The nanocavity array carefully designed on a silver surface implements the particular diffraction processes of light to a plasmonic Airy beam at a visible wavelength, as shown in Fig. 23(a)[631]. The generated plasmonic Airy beam directly exhibits its properties of non-diffraction, self-reconstruction, and self-bending. Furthermore, the metasurface can also emit the Airy beam, and the all-dielectric metasurface can avoid the loss caused by metals. An ultra-thin subwavelength metasurface is proposed to produce Airy beams with high efficiency and ultra-wide bandwidth by reflecting or transmitting input light, as shown in Fig. 23(b)[632]. The versatile and multi-feature Airy beams are designed by a synthetic-phase metasurface that superposes multiple special phase profiles in a metasurface, as shown in Fig. 23(c)[633]. For example, the specially designed metasurface superposes the fine-tuning Fresnel lens phase to obtain the modifiable Airy beams with an impressive performance in terms of focal length, beam width, and non-diffractive propagation distance. The synthetic-phase metasurface produces the Airy array beams by combining the optimized Dammann grating phase. In addition, a dielectric metasurface composed of silicon posts is proposed to generate an achromatic terahertz Airy beam with autofocusing properties, as shown in Fig. 23(d)[634]. To avoid accurate alignment and reduce the difficulty of the experiment, the synthetic-phase meta-optics is introduced to the silicon waveguide to demonstrate the 1D on-chip Airy-like beams near the telecommunication wavelength. It is worth noting that the silicon waveguide with the synthetic-phase meta-structure has an ultra-compact footprint of , as shown in Fig. 23(e)[635]. In addition, a shallow-etched silicon holography grating can also be used to generate the 2D Airy beam, as shown in Fig. 23(f)[636]. 4) Integrated needle beam generation The needle beam is an optical beam with a small focal spot size and long focal line, which is very attractive for applications that utilize the spatial and temporal localization of photons. A high-intensity extended focal depth makes fundamental progress in optical system performance, especially in optical microscopes. A custom diffractive optical element (DOE) is proposed to produce the needle beam. It is demonstrated that a needle beam generated via DOE has a degree of freedom up to 28 times the Rayleigh length, and still maintains a relatively constant beam diameter, uniform axial intensity distribution, and negligible sidelobe. To form a needle beam, the designed DOEs generate several closely adjacent focal points along the axis, as shown in Figs. 24(a) and 24(b). The beam length of the needle beam depends on the number of focal points. Figure 24(c) shows the experimentally measured optical profile of the needle beam, indicating that the needle beam can maintain the beam spot size along the -axis[637]. 5) Integrated pin beam generation The needle beam mentioned above is essentially formed by superimposing multiple beams with different focal lengths, creating a needle-like beam that maintains a small beam waist over a long distance. Another similar structured light is known as the pin beam. This type of structured light beam uses the self-focusing of the annular Airy beam towards the center, resulting in a central main lobe that maintains a small beam waist over a long distance, similar to a needle beam. A single fused silica mask created through photolithography can be used to generate the pin beam. The novel mechanism is employed to produce the pin beam by eliminating the transverse wave vector components and the directed truncation of components similar to the Airy beam, as shown in Fig. 24(d). To better understand the underlying mechanism, the self-focusing pin beam formation is studied by assembling multiple phase elements with opposite transverse wave vectors. Figure 24(e) shows a side view of the pin beam propagating over 8 km, with optimal “focusing” occurring around . Figure 24(f) shows the intensity profiles captured at , 4, and 8 km[74]. 2.3.5.Integrated vector beam generationThe vector beam is a peculiar light beam with a non-uniform distribution of polarization on the beam cross-section. The vector beam is proposed relative to conventional scalar beams with the same polarization state distribution at different positions on the beam cross-section, such as linearly polarized, circularly polarized, and elliptically polarized beams. Compared to scalar beams, vector beams provide more degrees of freedom in beam manipulation. With the increasing maturity of micro/nanofabrication technology, integrated devices can be used to achieve different types of beams in the polarization state. The vector beam has attracted more attention due to the non-uniform distribution of the polarization state, among which the most typical vector beam is the cylindrical vector beam (CVB). The CVB presents an axisymmetric distribution of polarization state on the cross-section, including two typical types: a radially polarized beam (RPB) along the radius direction and an azimuthally polarized beam (APB) along the azimuthal direction. With their unique focusing and axisymmetric properties, the RPB and APB enable a variety of new functions, such as sharp focusing and optical trapping. There are several reported integrated methods for producing RPB and APB. A ring structure with concentrically distributed 64 grating arrays, as shown in Fig. 25(a), is designed on the silicon platform to generate RPB and APB using TM-polarized and TE-polarized light in the waveguides, respectively[638]. Similarly, based on the large differences between azimuthal and radial electrical distributions from different polarization waveguide modes, a silicon nitride microring resonator with shallow-etched gratings, as shown in Fig. 25(b), controls the generation of RPB and APB by configuring the TE and TM excitation waveguide modes[639]. Then, using the microring with a width-varying waveguide, the CVB with arbitrary angle emission can be achieved by superimposing a phase profile on the microring, as shown in Fig. 25(c)[640]. The all-dielectric metasurfaces composed of subwavelength silicon pillars and sapphire substrate convert horizontal (vertical) linear polarization to RPB (APB) in a mid-infrared band, as shown in Fig. 25(d)[641]. In addition to passive integrated devices, active integrated devices can also generate vector beams. The CVB laser composed of an active microring cavity and two specially designed gratings (side grating and top grating) is designed and fabricated on a fully integrated III–V semiconductor platform. Figures 26(a)–26(c) show the schematic 3D structure, top view, and SEM image of the CVB laser. The optimized triangular side grating introduced at the circumference of the microring splits the degenerate whispering gallery modes, selects the symmetrical mode with less scattering loss as the lasing mode, and obtains a single-mode laser with a high SMSR. Another second-order top grating improves the efficiency of surface emission. Figure 26(d) demonstrates the far-field intensity distribution of measured CVBs without and with a rotating linear polarizer. One can see that the radiant intensity distribution of the RPB with zeroth azimuthal order is multiple concentric rings with a dark center, rather than a single ring[642]. Furthermore, a structure consisting of two concentric microcavities is proposed, which realizes two CVB lasers in an electric pump based on the InP platform, as shown in Fig. 26(e). When only the top grating is added to the microcavity, the laser mode is an APB, as shown in Fig. 26(f). When the microcavity has a top and side grating, the laser mode is an RPB, as shown in Fig. 26(g)[643]. 2.3.6.Integrated array beam generationAn array beam is composed of multiple sub-beams arranged in a specific spatial distribution pattern. Compared with a single beam, the array beam can obtain higher laser output power. Due to the different transmission paths of each sub-beam, the sub-beams reaching the receiving system have a certain mutual compensation effect and can potentially improve the beam quality in the transmission process. There are several integration schemes for generating the array beam using a metasurface and array gratings. An ultra-thin nano-slit metasurface enables the generation of the array beam with equal power distribution, as shown in Figs. 27(a) and 27(b). The sub-beam is the OAM beam. Figure 27(c) shows the measured and simulated intensity distribution of the array beam with topological charges of and [644]. An array OAM beam with various topological charges or special arrays is obtained by designing a particular metasurface. The metasurface is applied in the laser cavity to generate a array OAM beam laser, as shown in Fig. 27(d). The charge of the vortex array is scaled up by increasing the azimuthal gradient in the metasurface design. Figure 27(e) shows the experimental near-field and far-field distributions of the array beam with a topological charge of , which has high purity[645]. In addition, as shown in Fig. 27(f), a large-scale phased array composed of gratings on a silicon chip is used to generate the array beam. Figure 27(g) shows the experimental near-field distributions of all gratings, which shows uniform emission of a large-scale phased array[646]. Optical vortex lattices are named a network of optical vortex or optical vortex arrays. In general, the optical vortex lattice is produced by bulky optical elements that superpose two specific vortex beams, which is not conducive to miniaturizing the optical vortex lattice generator. Based on the principle of three-plane-wave interference, a silicon optical vortex lattice emitter using three tilt gratings is presented, as shown in Figs. 28(a) and 28(b). The thickness of the silicon strip waveguide and the etching depth of the ridge waveguide are 340 and 220 nm, respectively. Figures 28(c)–28(e) plot the simulated results of the optical vortex lattice based on the three-plane-wave interference, including intensity, phase, and interferogram distributions. It can be clearly seen that each dark point in the intensity distribution is a phase singularity. The phase singularity with a -phase change is regarded as an optical vortex with a topological charge of or . Figures 28(f)–28(k) present the simulated and experimentally measured results of the designed and fabricated optical vortex lattice emitter, showing favorable operation performance[647]. 2.3.7.Integrated spatiotemporal beam generationThe goal of integrated spatiotemporal beam generation is to flexibly control all degrees of freedom of the light field both spatially (amplitude, phase, polarization) and temporally (pulse width and shape). The electro-optic effect of the free carrier in silicon can be used to modulate the resonance of the photonic crystal microcavity to achieve nearly complete spatiotemporal control of the narrow band light field filtered in space and time, as shown in Fig. 29(a). Reverse-designed high-precision microcavities provide a vertical coupling efficiency that is greater than 90%, as shown in Fig. 29(b). Figure 29(c) depicts the maximum phase shift and half-maximum switching interval versus CMOS trigger duration. is the maximum phase shift. Figure 29(d) shows the complex reflectivity modulation with fJ-order pulse energies. One can see that it is feasible to achieve the high-contrast phase and amplitude modulation over a 1-ns interval when the pump pulse is 5 fJ. Figure 29(e) shows the modulation bandwidth of 135 MHz in the frequency-domain response[648]. In addition, using a single-layer metasurface strongly coupled to the epsilon-near-zero (ENZ) material, the control of a synchronous spatiotemporal laser can be achieved in the fiber laser cavity. Based on the PB phase, the plasmonic metasurface for the intracavity spatial modulation converts from a Gaussian beam to an OAM beam, as shown in Figs. 29(f)–29(i). Meanwhile, the ENZ material for intracavity temporal modulation generates a pulsed laser due to the powerful nonlinear saturable absorption, as shown in Figs. 29(j)–29(m)[649]. In addition, the spatiotemporal optical vortex (STOV) can be generated by the synergistic usage of optical vortex emitters and frequency combs, which control the OAM and frequency, respectively. As shown in Fig. 30(a), the OAM comb device comprises a high-Q microring with inner sidewall gratings, fabricated on the emerging AlGaAs on insulator (AlGaAsOI) platform. This OAM comb exhibits tight mode confinement and strong optical nonlinearities, enabling low-threshold microcomb generation. Figure 30(b) displays SEM images of the fabricated AlGaAsOI device. To reduce the scattering loss of the fundamental mode in the OAM microring, a multimode waveguide with a cross-section of is employed. The grating protrusion is designed to be 30 nm to achieve notable efficiencies for vortex ejection and high Q values for comb generation. Figure 30(c) presents the measured spectra of the microcombs in soliton states with wavelength spacing of 500 GHz. Three comb lines (, 5, and 6) with an intensity variation below 1 dB are filtered to generate self-torque pulses, as depicted in the red area of Fig. 30(c). The calculated and measured spatiotemporal profiles of the targeted self-torque pulse are shown as iso-surface plots in Figs. 30(d) and 30(e), focusing on the distribution of the main lobe and demonstrating varied OAM at different times in both intensity and phase[650]. Meanwhile, an OAM comb using 800-nm-thick stoichiometric films on a silica substrate is presented and illustrated in Fig. 30(f). Figure 30(g) shows the SEM image of the OAM comb. The mean radius and width of the microresonator are set as 22 and 2 µm, leading to a free spectral range (FSR) around 1 THz, as shown in Fig. 30(h). Figures 30(i) and 30(j) show the simulated and measured 3D reconstruction of the vortex in the spatial and temporal domains, excluding points with a normalized intensity below 0.2 in each frame for better contrast[651]. Unlike conventional OAM beams with spiraling phases, the generated beam demonstrated here features a double-helical intensity profile that dynamically revolves around the optical axis. This light field belongs to the category of “light springs,” a characteristic result of the STOV correlation. The consistency between the experimental and simulation results highlights the mutual coherence of the soliton microcomb generated in this OAM comb device. 2.3.8.Integrated knotted and linked beam generationThe theory of knots is a branch of topology that is a closed line of space in math. The simplest knot is an unknotted loop. Recently, a special light field with a knot was observed and drew the attention of researchers. A metasurface is a typical strategy for the generation of knotted optical vortex lines due to its unique and powerful ability to tailor light. Based on the PB phase, a reflective-type hologram metasurface is presented to efficiently produce the broadband optical vortex knots, as illustrated in Fig. 31(a). Figure 31(b) shows the experimentally obtained Trefoil knots and Hopf links at the wavelengths of 700, 800, and 900 nm[255]. One can see that although the size and shape of optical knots vary at different wavelengths, Trefoil knots and Hopf links are always produced. The obtained optical knots are six orders of magnitude smaller than that generated by an SLM. Relying on the combination of the PB phase and propagation phase, an all-dielectric metasurface generates two ultra-small-sized switchable optical knots with different topological configurations by changing the polarization of incident light, as shown in Fig. 31(c). Figures 31(d) and 31(e) show the simulated and experimentally measured results for the Trefoil knots and Hopf links[256]. When the incident light is right-handed circular polarization, the output optical knot of the metasurface is a Trefoil knot. When the incident light is left-handed circular polarization, the output optical knot of the metasurface is a Hopf link. 2.3.9.Integrated in-plane waveguide mode generationThe in-plane waveguide mode is a light beam that is generated and propagates in the integrated photonic waveguides. With the rapid development of integrated photonics, there are many photonic platforms of various materials, including silicon photonics, lithium niobate films, silicon nitride film, silicon carbide photonics, etc. Silicon photonics is an attractive photonic integration platform to enable on-chip spatial mode manipulation due to its high-contrast index, high-density integration, low power consumption, and CMOS compatibility. The generation techniques of in-plane waveguide mode mainly include phase matching, metamaterial, beam shape, and coherent scattering. As shown in Fig. 32(a), asymmetric directional couplers (ADCs), consisting of a bus waveguide and a narrow waveguide, are the typical structure based on the principle of phase matching[497]. The ADC structure can convert the fundamental mode of the narrow waveguide to the high-order mode of the bus waveguide when their refractive indices are matched. Although the ADC is a very simple structure to obtain the high-order mode, it is difficult to satisfy the phase matching condition in a broadband wavelength range, due to the difference in dispersion slope between the fundamental mode and high-order mode. Metamaterial has been introduced into the silicon platform as the high-order mode generator, which is particularly attractive in two aspects. One of the advantages is shrinking the footprint of the device. The other one is that the metamaterial-based structure facilitates the generation of a very high-order mode, as shown in Fig. 32(b)[652]. The beam shaping method consists of three stages, i.e., power splitting, controllable phase shifting, and mode combining, as shown in Figs. 32(c)[653] and 32(d)[654]. The mode splitting and mode combining are usually achieved through symmetric Y branches. The relative phase difference can be introduced by thermo-optic controlling, electro-optic controlling, and optical path difference between adjacent branches. It is a relatively complex situation. For the coherent scattering method, the inverse design method enables efficient mode conversion in an ultra-compact size, as shown in Figs. 32(e)[655] and 32(f)[656]. Nonetheless, due to the small feature size, the resulting irregular nanostructures inevitably require high manufacturing precision. 2.3.10.Integrated reconfigurable structured light generationAlthough various integrated devices have realized the generation of diverse structured light beams with favorable performance, once integrated optical devices are designed and fabricated, their optical functions are usually fixed, which makes difficult to deal with application scenarios requiring flexibility such as beam scanning and zoom lenses. Reconfigurable structured light with active adjustment ability is proposed to improve the practical value of structured light and meet more complex application scenarios. The use of phase change materials such as (GST) is considered to be a promising option due to its non-volatility property and strong refractive index modulation. As shown in Fig. 33(a), a resonator metasurface based on the GST materials is present to produce the reconfigurable OAM mode. As shown in Fig. 33(b), when the state of GST material is amorphous GST (aGST) [crystalline GST (cGST)], the phase profile of the generated OAM demonstrates the topological charge of [657]. A compact SiN waveguide based on a phase-gradient metasurface replaces noble metals with phase change materials to produce the in-plane waveguide mode, as shown in Figs. 33(c) and 33(d). Figure 33(e) plots the simulation FDTD results of the SiN waveguide with a phase-gradient metasurface[658]. When the state of GST material is cGST, the output mode of the SiN waveguide is mode. When the state of GST material is aGST, the output mode of the SiN waveguide is mode. This device provides an efficient approach for producing reconfigurable in-plane waveguide beams and may be applied to large-scale photonic neural networks. In optical systems, BIC occurs through interference between local resonance and radiation modes and is observed in the form of quasi-BIC in many systems. In addition to ultra-high-Q-factor and low-threshold lasers, it is also predicted that BIC modes can exhibit vortex behavior with different topological charges. The laser emission at the symmetry-protected BIC can be fully optically controlled in a perovskite metasurface system, as shown in Fig. 34(a). Figure 34(b) shows the SEM image of a perovskite metasurface. The perovskite metasurface is pumped with a circular laser beam to protect symmetry, resulting in a doughnut-shaped beam, as shown in Fig. 34(c). Once the pump area shifts from a circle to an ellipse, the symmetry protection will break and two linear diffraction beams will be generated, as shown in Fig. 34(d). A similar symmetry break can also be achieved through a double-beam configuration, as shown in Fig. 34(e). When the nanostructure is only pumped by the first beam, the output is a uniform doughnut. Once the second and first beams overlap in time, the optical symmetry will be broken. The switching time from an eddy current laser to a conventional linearly polarized laser is only , as shown in Fig. 34(f). By changing the asymmetric pump beam to a circular beam, two linearly polarized beams can also be switched back to the vortex laser, with a similar transition time of , as shown in Fig. 34(g). To demonstrate the high-speed repeatability of the modified device, Fig. 34(h) shows that the doughnut-lobes-doughnut state can be combined in a single process, indicating that the transition time is not limited[659]. As a short summary, Table 1 intuitively lists the major parameters and performance of the state-of-the-art integrated structured light generation and gives a detailed comparison[74,255,256,385,386,464,466,467,469–475,481,496,497,550–553,555,556,579–598,610–612,615–617,622–625,627–660]. The compared parameters and performance include the type of structured light, number of modes, generation method, material, size, working wavelength, etc. Table 1Parameters and Performance of State-of-the-Art Integrated Structured Light Generation
3.Integrated Structured Light ProcessingIn the aforementioned section, we comprehensively introduce various integrated structured light generation devices. It is worth noting that, in addition to the generation of structured light, the processing of structured light is also a key technology to facilitate structured-light-enabled diverse applications. In this section, the basic theories and various functionalities of integrated structured light processing are reviewed, as outlined in Fig. 35. The first part introduces basic theories and principles of structured light processing. The second part presents various integrated structured light multiplexing. The third part describes various integrated structured light transformations. 3.1.Theories and Principles of Structured Light ProcessingThis part mainly introduces the theories and principles of structured light processing. Note that some operation principles similar to the structured light generation such as phase control, mode coupling, mode superposition, etc., will not be described here. The following theories of structured light processing to be introduced are multiplane light conversion theory, geometric coordinate transformation theory, nonlinear interaction theory, and chiral mode switching theory. 3.1.1.Multiplane light conversion theoryMultiplane light conversion (MPLC) employs multiple-phase planes to realize various functions of structured light processing. Specifically, MPLC utilizes materials or components to locally adjust the phase of the incident wave, thereby altering the wave’s propagation characteristics[661,662]. By modulating the phase in different regions, precise control of the wavefront can be achieved, resulting in structured light processing. Typically, MPLC methods involve changing the refractive index or thickness of the material to alter the phase of the wave as it propagates through the region. Assuming the target phase is and the refractive index of a material is , the phase change of the wave after passing through the material can be expressed as where is the thickness of the material and is the wavenumber, defined as .Typical MPLC devices include phase plates and gratings. The phase plate controls the wavefront by applying different phase delays in various regions. Assuming the phase plate is positioned at and modulates the phase of the incident wave, the transmitted wave can be expressed as where is the wave function of the incident wave at . is the phase shift introduced by the phase plate.The grating is a periodic structure that can diffract and modulate the phase of an incident wave, enabling precise manipulation of structured light. The phase modulation of the grating can be expressed as where is the maximum phase shift, and is the period of the grating. The wave after passing through the grating can be expressed asFor MPLC with multiple phase planes, the phase distribution of each phase plane can be calculated using the wavefront matching algorithm. Wavefront matching means that a forward propagating light field should match a backward propagating light field anywhere in space. When using the wavefront matching method, the inverse design MPLC has no analytic function expression. Therefore, advanced algorithms such as machine learning and neural networks can be introduced to further accelerate the design and improve the performance of MPLC. 3.1.2.Geometric coordinate transformation theoryThe geometric coordinate transformation generally requires two phase functions and located at two planes and to implement the desired light conversion[330,502,503,663,664]. After passing through plane with the phase function , the light normally incident at the point in plane will pass through to the point in plane . Then, the phase function in plane performs phase correction to make the light after the plane propagate vertically forward. The light field that is collimated onto plane is expressed as . When the light field is modulated by the phase function in plane , it can be further expressed as where the light travelling from the point in plane to the point in plane has direction cosines , , and . According to Eq. (32), it can be obtained thatFurthermore, for light propagation from the point in plane to the point in plane , and can be represented as where and is the distance between plane and plane . Due to , Eq. (33) combined with Eq. (34) can be simplified as where and describe the mapping or the desired coordinate transformation from plane to plane . By solving Eq. (35), the required phase function in plane can be obtained.After implementing the transformation at any position, it is also necessary to obtain the phase function in plane that performs phase correction to collimate the light after leaving the plane . The equation that the phase function needs to satisfy can be written by where and represent the inverse transformation from plane to plane .3.1.3.Nonlinear interaction theoryIn nonlinear optics, the nonlinear polarization intensity can be expressed by the power series of electric field intensity[665–668]: where is the dielectric constant in vacuum. represents linear polarizability, which is related to the properties of materials such as absorption, refractive index, dispersion, and birefringence. and represent the second- and third-order nonlinear polarization coefficients, respectively. is the research object of linear optics, which is related to the refractive index of the medium. The second term in Eq. (37) represents a second-order nonlinear effect, whose intensity is proportional to the square of the incident electric field. The second-order nonlinear effect is also the most common phenomenon in nonlinear optics, such as frequency doubling, sum frequency, difference frequency, and optical parametric oscillation. The third term in Eq. (37) represents the third-order nonlinear effect, which is related to the generation of third-order harmonics, optical Kerr effect, self-phase modulation, cross-phase modulation, stimulated Raman scattering, stimulated Brillouin scattering, two-photon absorption, etc.3.1.4.Chiral mode switching theoryThe energy spectrum surface forms a self-intersecting Riemann surface around exceptional points in the parameter space. Such Riemann surface topology is impossible for the Hermite system and has aroused more and more research interest. An adiabatic evolution path is constructed to achieve the transition of system states between different energy bands. The injection characteristic state around an exceptional point in an adiabatic manner can be converted into another characteristic state after one cycle but will return to itself in the Berry phase of after two cycles. However, the adiabatic theorem is no longer applicable under the dynamic evolution of non-Hermite systems. Dynamically encircling an exceptional point is particularly interesting due to its chiral response, where the final output state only depends on the surround direction and is independent of the input state. This encircling exceptional point evolution process can be depicted by discretizing the adiabatic evolution process along the propagation distance into distance intervals with many fixed Hamiltonians. For the Hamiltonian that stays constant through the distance interval , the final state can be acquired by incorporating Eq. (26) into Eq. (25)[572,667]: where the initial state is at the position , with and being arbitrary coefficients. From the expression of the final state, the variations of system state in phase and amplitude arise from the real and imaginary parts of the eigenvalues, respectively.3.2.Integrated Structured Light MultiplexingMultiplexing is one of the most important and powerful technologies for efficient capacity scaling in optical communications, where multiple collinear propagation light beams, each carrying different data information, are combined together to increase the transmission capacity. Due to the diversity of structured light fields, structured light multiplexing is of great significance to enable high-capacity optical communications in multiple application scenarios. The integrated structured light multiplexing has recently drawn great attention due to its compactness, including the out-of-plane structured light multiplexing and in-plane structured light multiplexing. In the following part, taking out-of-plane/in-plane OAM-carrying light beams and in-plane waveguide modes as typical examples, we give a brief introduction of out-of-plane/in-plane OAM (de)multiplexing and in-plane waveguide mode multiplexing. 3.2.1.Out-of-plane/in-plane OAM (de)multiplexingIn theory, the topological charge of an OAM mode is unbounded in free space, which indicates that the capacity of optical communications can be greatly improved by OAM multiplexing. Although only a limited number of OAM modes are supported in the system with a limited aperture, the scalable OAM multiplexing scheme, with almost constant levels of loss and crosstalk with an increased number of OAM modes, is highly desirable. A robust and compact device based on the MPLC is proposed to implement efficient OAM multiplexing. As shown in Fig. 36(a), five inversely designed phase planes relying on the wavefront matching algorithm convert the near-perfect Gaussian beam in each channel of the fiber array to different co-axial OAM modes. Such a simultaneous multiple-mode conversion process is equivalent to the OAM multiplexing. These phase planes are based on the nonparaxial spiral coordinate transformation and are manufactured on a silicon wafer utilizing three cycles of photolithography and etching. Figure 36(b) shows the measured OAM mode crosstalk matrix, which indicates 11 high-purity OAM modes with the mode crosstalk of . The five phase planes also enable the demultiplexing function with the mode crosstalk of [501]. In addition to the MPLC for out-of-plane OAM multiplexing, a trench waveguide structure is also proposed to achieve the on-chip in-plane OAM multiplexing. Figure 36(c) shows the trench-waveguide-based OAM mode multiplexer fabricated by the femtosecond laser direct writing technique, which consists of two waveguides for inputting two Gaussian beams, two Mach-Zehnder interferometers (MZIs), and a specially designed trench waveguide structure for converting, multiplexing and outputting two OAM modes. The planes on which the two MZIs sit maintain a 90° angle. As shown in Figs. 36(d) and 36(e), the MZI can convert the input Gaussian beam into two LP-like high-order modes with the diagonal distribution[470]. When a phase difference of two LP-like modes is achieved by tuning one of the MZI arms, the mode is generated at the output trench waveguide. Similarly, when a phase difference is introduced between two LP-like modes with proper adjustment of one of the MZI arms, the mode is achieved at the output trench waveguide. The combined two MZIs and one trench waveguide, with the 3D structure easily achievable by femtosecond laser inscription, enable the OAM multiplexing. Similarly, the OAM demultiplexing is also available when using the 3D structure in an opposite direction. The trench-waveguide-based OAM mode (de)multiplexer obtains low crosstalk of . Note that the trench waveguide structure can also enable an OAM exchange function with proper adjustment of the trench length, as shown in Figs. 36(d) and 36(e). For out-of-plane OAM demultiplexing, as illustrated in Fig. 37(a), an OAM mode sorter based on the optical coordinate transformation (spiral transformation) is proposed to implement the bidirectional conversion between seven co-axial OAM modes and seven Gaussian-like modes with linear distribution. Figure 37(b) shows the photo of the fabricated OAM sorter, which is a DOE. The OAM sorter is composed of an unwrapper, a phase corrector, and a positive lens, as shown in Figs. 37(c)–37(e)[502]. The experimental results show that the total power loss is less than 8 dB and the inter-mode crosstalk is less than for all seven OAM modes (). In addition, an ultracompact OAM sorter is proposed using metasurfaces integrated onto a CMOS camera. Based on the propagation phases, the presented OAM sorter via two metasurfaces is used to perform the unitary transformation, including the log-polar transformation, fan-out beam copying, focusing, phase correction, and Fourier transform. The metasurface is formed by nanopillars with circular cross sections. The pillar height and lattice size are set as 800 and 300 nm, respectively. Figure 37(f) shows the simulated transmittance and phase shift of eight selected nanopillars with varied diameters. One can clearly see that the phase shift can cover the range of with almost constant high transmittance. Figure 37(g) shows the schematic of the OAM sorter based on metasurfaces. Figure 37(h) shows the microscope image and tilt-view SEM image of fabricated metasurfaces. Figure 37(i) shows the measured results of the metasurface-based OAM sorter, which indicates an average crosstalk of [503]. 3.2.2.In-plane waveguide mode multiplexingLike amplitude, phase, wavelength, time, polarization, etc., in-plane waveguide mode is also an important physical dimension of light[445–454]. As a subset of space-division multiplexing (SDM), mode-division multiplexing (MDM) employs orthogonal in-plane waveguide modes as independent channels to transmit different data information, which can increase the aggregate capacity on a single wavelength and greatly improve the transmission rate of on-chip optical interconnects. Due to the orthogonal properties of different physical dimensions, the MDM is compatible with various multiplexing techniques on other physical dimensions, such as wavelength and polarization, which can further increase the total number of transmitted data channels. For example, multiple cascaded microrings are used to implement the wavelength-division multiplexing (WDM)-compatible MDM on a silicon chip. As shown in Fig. 38(a), based on the principle of mode matching, multiple in-plane waveguide modes can be multiplexed by microrings with a multimode bus waveguide[498]. As shown in Fig. 38(b), the multimode waveguide with a width of 2.37 µm can support five in-plane waveguide modes. Then, a 10-channel mode (de)multiplexer that is compatible with polarization-division multiplexing (PDM) is presented and realized by five cascaded ADCs and five polarization beam splitters (PBSs)[499]. Based on the dispersion curves of multiple modes, the parameters of ADC structure are selected to simultaneously extract high-order TE- and TM-polarization modes by fundamental modes with dual polarizations. The measured results show that 10 TM and TE mode channels all have low crosstalk of and low excess losses over a 90-nm wavelength band. As shown in Fig. 38(c), a silicon-based 11-channel single-polarization mode (de)multiplexer is demonstrated using a subwavelength grating (SWG) structure[500]. The introduced SWG structure is used to solve the problem caused by the large difference in the slope of the dispersion curve between the higher-order and fundamental modes. The 11 multiplexing modes all have favorable performance of low insertion losses () and crosstalk values (). Photonic BIC is also exploited in the on-chip mode (de)multiplexer. A four-channel TM-polarization mode (de)multiplexer based on different orders of BIC is proposed, as shown in Fig. 38(d)[670]. The integrated platform is an etchless lithium niobate platform, which simplifies the device fabrication process of the in-plane mode (de)multiplexer. 3.3.Integrated Structured Light TransformationThe target of structured light transformation is to convert one structured light into another structured light or the same type of structured light in a different order, which facilitates structured-light-enabled optical data processing. 3.3.1.Transformation of the same type of structured lightFor the transformation of the same type of structured light with different orders, this part gives a brief introduction to four types of structured light, such as OAM mode, LP mode, array beams, and in-plane waveguide mode. 1) OAM mode transformation As mentioned above, the trench waveguide obtains two co-axial transmission OAM modes by superposing two different LP-like modes. It is worth noting that the two LP-like modes possess different propagation constants ( and )[348]. When the propagation distance is , the two LP-like modes with a phase difference are combined as the mode. When the propagation distance becomes , the combined mode is changed to the mode, as shown in Figs. 36(d) and 36(e). Thus, the OAM exchanger is achieved, which facilitates flexible on-chip processing of OAM modes. OAM multiplication and division are common OAM transformations, which are also important functions of OAM processing for spatial mode coding/decoding and can potentially be applied to OAM-based optical communications and data processing as well as high-dimensional quantum information processing. The integer OAM multiplier and divider composed of two optical elements are presented to implement the flexible OAM processing. For the OAM multiplier, as shown in Figs. 39(a) and 39(b), the first element with a specially designed phase pattern performs the integer-fold multiplication by splitting and mapping the azimuthal phase gradient onto complementary circular sectors (CCSs). The second element with another phase pattern performs phase correction. The OAM divider is achieved by similar methods, as shown in Figs. 39(c) and 39(d)[504]. The difference is that the first phase pattern of the OAM divider maps the distinct CCSs of the input OAM beam into an equal number of circular phase gradients. Furthermore, a simple and efficient scheme is proposed, which uses the azimuth-scale spiral transform to perform OAM multiplication and division of arbitrary rational factors in a single stage, as shown in Fig. 40(a). Figures 40(b) and 40(c) show the simulated and measured mode purity through an OAM multiplication with a rational factor of 1.5, which shows the transformed OAM modes with high mode purity of 18 dB in the simulation and 11.5 dB in the experiment. Figure 40(d) shows the simulated and measured intensity and phase distributions of the input and output OAM beams through an OAM multiplication with a rational factor of 1.5[671]. With the development of high-power laser sources, research on nonlinear optics has made great progress. Wavelength conversion, signal amplification, pulse compression, spectral broadening, and mode locking based on nonlinear effects have been widely used in optical signal processing[672,673]. In nonlinear optics, there has been a recent growing interest in processing OAM modes by 2D materials, micro-nano structures, and lithium niobate on insulator (LNOI) films. Compared to other integrated optical material platforms such as SOI, InP, and SiN, LNOI has low light absorption, fast light modulation, and high optical nonlinearity, which has attracted much attention. The nonlinearity of LNOI is also used to facilitate OAM processing. The spirally poled nonlinear photonic device is presented to generate the OAM mode of second-harmonic (SH) waves, as shown in Fig. 41(a). In this nonlinear device, there is a phase difference between the SH waves from positive and negative domains, which can be used in the design of nonlinear Fresnel band plates for producing SH OAM states. Figure 41(b) shows the measured OAM mode of SH waves, including intensity pattern and off-axis interference distribution. The obtained results indicate that the generated SH wave is the [505]. Based on the enhanced electro-optic Kerr effect, as shown in Fig. 41(c), a periodically poled lithium niobate photonic waveguide is proposed to manipulate the state of the OAM mode. As shown in Fig. 41(d), the input mode can be converted to the mode[506]. The second-harmonic generation (SHG) and third-harmonic generation (THG) nonlinear conversion of OAM is theoretically and experimentally demonstrated from an atomically thin tungsten disulfide () monolayer, as shown in Figs. 42(a) and 42(b). Figure 42(c) shows the nonlinear response spectrum under the fundamental OAM excitation of the monolayer. One can see that the SHG and THG signals are generated at the wavelengths of 780 and 520 nm, respectively. In addition, the polarization state of the fundamental beam can precisely control and determine the intensity and polarization state of the converted nonlinear OAM beam using the symmetry properties of the crystal. From the measured results shown in Fig. 42(d), one can see that the generated SHG and THG signals are and , respectively, when the fundamental beam is mode[507]. In addition, OAM beam or structured light interaction with two-dimensional materials (strong light-matter interaction due to reduced dimensionality) is of great interest. The photoluminescence (PL) spectra of single-layer molybdenum disulfide () are studied under the excitation of the OAM beam. Figure 42(e) shows intensity profiles, wavefronts, and phase distributions of different orders of OAM beams. Figure 42(f) shows the optical microscope of the sample. Figure 42(g) shows the PL spectra curves under the excitation of , , , , and beams[508]. One can see that there is a noticeable blue shift of the PL peaks as the topological charge number of OAM beams increases. The micro-nano structures are also used to enable the SHG and THG of OAM. As illustrated in Fig. 43(a), an ultra-thin photonic metasurface is used to generate the spin-controlled OAM of light in SHG. Figure 43(b) shows the phase distribution of SHG and SEM images of the gold plasmonic metasurface. Figures 43(c) and 43(d) show the measured and simulated intensity distributions of SHG from different nonlinear metasurfaces, which are in good agreement with each other[674]. The first and second rows are the RCP and LCP components of the SHG radiation when the pump light is LCP. The vector distribution of the SHG OAM beam also depends on the topological charge of the plasmonic metasurface. The third and fourth rows are horizontal- and vertical-polarization SHG signals, when the pump light is a horizontal-polarization fundamental wave. Remarkably, for the linear polarization fundamental wave with both LCP and RCP components, the SHG radiation contains RCP and LCP components with opposite topological charges of OAM. By choosing the linear polarization SHG, one can achieve the HG beam-like interference pattern of the RCP and LCP components with opposite OAM. The mode rotation of the HG beam-like interference pattern can be flexibly controlled by rotating the polarization angles of both the fundamental wave and SHG wave. As shown in Fig. 43(e), the device consisting of a gold-fork microstructure with a nonlinear polymer thin film can also create SHG and THG of OAM. Compared to other devices for generating nonlinear OAM, the advantage of this device is its miniature footprint. Except for the production of nonlinear OAM radiation, the nonlinear processes of beam splitting are also achieved by the combined device. Figures 43(f) and 43(g) show the SHG and THG light under the front-pumping scheme. The SHG light from the device is shown in Fig. 43(f) when the polarization direction of the analyzer is set at direction. The THG light of mode at the first diffraction order from the device is shown in Fig. 43(g) when the polarization direction of the analyzer is set at direction. Figures 43(h)–43(j) show the SHG and THG light under the back-pumping scheme. The SHG light of mode from the device is shown in Fig. 43(h) when the polarization direction of the analyzer is set at direction. The THG light of mode at the first diffraction order from the device is shown in Fig. 43(i) when the polarization direction of the analyzer is set at direction. The SHG and THG light of mode are simultaneously produced after removing the analyzer in the optical path, as shown in Fig. 43(j)[675]. 2) LP mode switching To process different LP modes, a fiber-chip-fiber switching system is established using two integrated photonic chips, which are the femtosecond laser inscribed mode (de)multiplexer and silicon switch array, as shown in Fig. 44(a). This system includes three basic functions, i.e., mode/polarization demultiplexing, switching/routing, and mode/polarization multiplexing. Figures 44(b) and 44(d) show the schematic of the femtosecond laser inscribed mode demultiplexer and multiplexer. Figure 44(c) shows the schematic of the silicon switch array[511]. Compared to the traditional switch array, the topologically optimized non-blocking switch array has fewer unit switches and crossings, which indicates favorable performance in terms of loss and crosstalk. In this system, not only the advantages of fiber-optic data transmission but also the ability of integrated photonic chips to process data on spatial modes are fully utilized. The LP mode switching system uses the 3D coupler fabricated by femtosecond laser inscription to realize the efficient coupling and connection of few-mode fibers and a silicon switch array. The LP modes in the few-mode fiber are demultiplexed by the 3D coupler and then fed into the silicon switch array for flexible multi-channel mode/polarization optical switching. The data information carried by LP modes is also flexibly switched accompanied by the LP mode switching. After that, another 3D coupler is used for LP modes multiplexing and coupled back into the few-mode fiber for transmission. Remarkably, such a hybrid system (femtosecond laser inscribed 3D photonic chip, silicon-based 2D photonic integrated circuit) provides a compatible and scalable solution to implement flexible fiber-chip-fiber data transmission and processing on spatial modes, not only for LP modes but also for other diverse spatial modes. 3) Array beam transformation 3D photonic chips fabricated by the femtosecond laser can also be used to process array beams[676]. As shown in Figs. 44(e) and 44(f), a linear array beam with multiple channels is converted to a concentric circular array beam. The 3D photonic chip has impressive performance in bandwidth (covering the C + L band), insertion loss (<0.88 dB), and crosstalk (). The 3D photonic chip can be used as the important fan-in/fan-out (FIFO) device of multi-core fiber, which has seen wide applications in SDM fiber-optic communications. As shown in Figs. 44(e)-44(g), an FIFO device supporting 19 channels is fabricated by the femtosecond laser direct writing technique, which facilitates high-performance coupling and connection between the 19-channel single-mode fiber array and 19-core fiber[509]. Furthermore, the 3D photonic chip can also change the spatial distribution of each unit of the array beam. Figures 44(h) and 44(i) show the schematic and measured results of diverse array beam transformation by the 3D photonic chip[677]. Concentric circular array beams with a unit Gaussian beam or unit two-lobe few mode are produced, which are compatible with the few-mode multi-core fiber in SDM fiber-optic communications. 4) In-plane waveguide mode processing Because of the distinct advantages of light manipulation on integrated photonic chips in terms of compactness and flexibility, multiple functions of in-plane waveguide mode processing have been widely studied, such as mode exchange, mode switch, mode add/drop, and chiral switching. For the mode exchange, as shown in Fig. 45(a), based on the inverse design method, a compact and broadband silicon mode exchange device is presented, which only occupies a footprint[514]. The conversion efficiencies of and modes both exceed 71% over the whole C-band. In addition, the mode exchange device is robust to fabrication error and temperature variation. In order to meet the challenge of achieving compact mode exchange with high performance, a universal method of mode exchange is proposed by multimode excitation/interference. As shown in Fig. 45(b), the mode exchange with ultra-compact footprints implements the efficient conversion between mode and mode[678]. The mode exchange possesses a low excess loss of 0.19–0.45 dB and an extinction ratio of >10 dB in an ultrabroad band of 340–400 nm. The mode switch is also an important mode processing technology. A reconfigurable mode switch device is presented, which is composed of symmetric Y-junctions, crossing, and a thermo-optic phase shifter. Compared to conventional Y-junctions, subwavelength symmetric Y-junctions introduce a series of chirped holes to enable uniform beam splitting. As shown in Fig. 45(c), a four-port beam splitter consisting of three symmetric Y-junctions can be used as the four-port beam combiner, and they are connected to each other by the straight waveguides or crossing[679]. Through this structure, any of the four output modes can be acquired by turning the thermo-optic phase shifter, when any input mode is launched. Experimental results show that the mode switch has impressive performance of low insertion loss (), low crosstalk (), and large bandwidth (). Furthermore, the mode switch device introduces a bi-level adiabatic tapered waveguide structure to the switch of multiple modes and polarizations, as shown in Fig. 45(d)[680]. Mode add/drop is the technology that can be used to add/drop any desired modes and arbitrarily reassign them, allowing for the dynamic operation of mode carriers in multimode networks with great flexibility. Mode exchange devices are developed to form the direct-access mode add/drop multiplexers (MADMs), as illustrated in Fig. 45(e)[681]. The structure also possesses the corresponding ADC structure to implement the high-order mode add/drop. Then, multiple direct-access MADMs can be cascaded to simultaneously process multiple modes. Chiral mode switching is one of the interesting functions that could be realized by exceptional points. When the exceptional point is encircled in a clockwise or counterclockwise direction, the output modes are different. In general, the length of the device is quite long to ensure the adiabatic evolution. To cope with this problem, as shown in Fig. 46(a), a moving exceptional point is presented to encircle a smaller loop, resulting in the structure length being significantly reduced. As shown in Fig. 46(b), from left to right, the input odd mode and even mode are both converted into the output odd mode, while from right to left, the input odd mode and even mode are both converted into the output even mode[682]. Thus, the output light does not depend on the type of input light, but on the direction of the input light, corresponding to the encircling direction of the exceptional point. In a 69-μm-long structure, transmission efficiency is maintained at . A fast Hamiltonian variation on the parameter boundaries, named Hamiltonian hopping, is presented and applied to the process of encircling exceptional points to improve the efficiency of chiral mode switching, as shown in Fig. 46(c)[683]. This method avoids path-dependent loss and overcomes a long-standing challenge of non-Hermitian optical systems. As shown in Fig. 46(d), the fast parametric evolution along the parameter space boundary of the system Hamiltonian is proposed and verified to be able to shrink the length of the structure[669]. 3.3.2.Transformation of different types of structured lightThe structured light transformation may occur not only between the same type of structured light but also between different types of structured light. 1) Mapping of OAM mode and LP mode Both OAM modes and LP modes are widely used in MDM communications, and the switching of OAM mode and LP mode is of great interest to enhance the flexibility of mode management. A new type of 3D photonic device based on a deep neural network is proposed to achieve the conversion between OAM and LP modes. As shown in Fig. 47(a), the proposed optical neural network mode mapper consists of four diffractive layer structures that receive input OAM mode bases and convert them into the corresponding LP modes. A deep neural network model, based on the angular spectrum method and beam propagation method, is used to design and optimize the four-layer 3D photonic device, enabling a complex field mapping relationship between OAM and LP mode bases. To meet the requirements for compactness and miniaturization, a femtosecond laser direct writing technique based on a two-photon polymerization process is used to fabricate the 3D photonic device with dimensions of . Figures 47(b) and 47(c) show the SEM and zoom-in SEM images of the fabricated 3D photonic device, revealing high-quality and uniform pixels. Figure 47(d) shows the measured intensity distributions of the input and output light fields, with the corresponding interference pattern of the OAM mode in the upper right corner. The mode field results demonstrate that the 3D photonic device successfully achieves the conversion from multiple OAM modes to LP modes with excellent performance[510]. 2) Transformation of LP mode and in-plane waveguide mode The transformation of LP mode and in-plane waveguide mode can be used as the bridge to connect the multimode optical fiber and integrated waveguide[337,684–695]. However, it is a great challenge in the transformation of LP mode and in-plane waveguide mode, due to their mismatch in shape and size. To cope with this challenge, a mode converter consisting of a multistage tapered silicon buried in a SiN strip waveguide can achieve the conversion between LP-like mode and in-plane waveguide mode, as illustrated in Fig. 47(e)[684]. Moreover, as shown in Fig. 47(f), by introducing a specially designed tapered fiber, the multistage tapered silicon waveguide buried in a polymer strip waveguide further enables the transformation of six LP modes in a few-mode fiber and six in-plane waveguide modes[685]. To further increase the working bandwidth, the multistage tapered waveguide assisted with a slot waveguide is presented to achieve efficient mode conversion between LP modes and in-plane waveguide modes over the whole C-band, as shown in Fig. 47(g)[686]. As a short summary, Table 2 lists the major parameters and performance of the state-of-the-art integrated structured light processing and gives a detailed comparison[470,498–511,514,669–671,674,675,677–686], including the type of structured light, number of modes, processing functionality, material, size, and working wavelength. Table 2Parameters and Performance of State-of-the-Art Integrated Structured Light Processing
4.Integrated Structured Light DetectionIntegrated structured light detection means using the compact device to detect various structured light with spatially variant amplitude, phase, and polarization distribution. The main content of structured light detection is to study how different micro-nano structures on various material platforms distinguish diverse structured light. The materials involved in integrated structured light detection include metal, dielectric, semimetal, and other semiconductor materials. In this section, the basic theories and various schemes of integrated structured light detection are reviewed, as outlined in Fig. 48. The first part introduces basic theories and principles of structured light detection. The later parts present various integrated structured light detection schemes, including metal micro-nano structures, dielectric micro-nano structures, photocurrent detectors with U-shaped electrodes, thermoelectric detectors with spin-Hall couplers, diffractive planes, silicon MZI network, inverse design subwavelength structure, and silicon nanorod optomechanics. 4.1.Theories and Principles of Structured Light DetectionThis part mainly introduces some special theories and principles of structured light detection, such as the surface plasmon polariton effect, photogalvanic effect, thermoelectric response of the spin-Hall effect, diffractive deep neural network, and inverse design method. Other basic theories and operation principles similar to the aforementioned structured light generation and processing will not be introduced here. 4.1.1.Surface plasmon polariton effectThe structured light detection schemes based on metal materials use different surface plasmon polariton fields generated on the interface of the metal and dielectric to detect the properties of structured light such as SAM and OAM. Since the surface plasmon polariton wave generated by incident light on the metal-dielectric interface can be considered as the plane wave, the electric field distribution of the total surface plasmon polariton field can be calculated using the Huygens-Fresnel law. The field distribution function of different structured light may have a difference by properly designing the micro-nano structure etched on the metal surface. Then the features of incident structured light can be obtained directly using the near-field microscopy. Taking structured light carrying OAM as a typical example and considering the incident light possessing both SAM with spin quantum number and OAM with topological charge , it can be expressed in cylindrical coordinates as . The electric field component that is perpendicular to the metal-dielectric interface can excite the surface plasmon polariton. The surface plasmon polariton field generated at the metal-dielectric surface of the infinitesimal unit can be expressed as[696] where is a constant that is related to the coupling efficiency from the incident light to the surface plasmon polariton wave, is the wave vector of the surface plasmon polariton that propagates on the metal-dielectric interface, is the component of along the -axis direction, is the distance between the infinitesimal unit and the coordinate system origin point, and is the propagating distance between the detection point and the infinitesimal unit. Therefore, the surface plasmon polariton field generated by the micro-nano structure etched on the metal surface is the integral of Eq. (39). For example, for a semicircular nanoslit, the surface plasmon polariton field distribution function on the metal surface excited by a radially polarized () incident light with topological charge is expressed as[518] where is the position vector of the point on the semicircular slit. Along the direction of the opening diameter of the semicircular structure (i.e., along direction), the surface plasmon polariton field distribution approximately satisfies where represents the -order Bessel function and T is the total angular momentum. Consequently, the surface plasmon polariton distributions excited by structured light beams with distinct topological charges satisfy different Bessel functions. Moreover, by adjusting the rotation angle and spacing of the two nanoslits, light beams carrying different SAMs can have different surface plasmon polariton emission directions[697]. The same effect can be achieved by spatially lining up nanoslits of different sizes. Directional surface plasmon polariton emission using polarization-sensitive apertures etched on the metal film has been implemented[698–703]. Therefore, by designing nanoslit units that can distinguish incident light with different SAMs, and forming a specific periodic arrangement, SAM and OAM of the incident light can be detected simultaneously.4.1.2.Photogalvanic effectIn addition, since the coupling between photons and electrons needs to satisfy the law of momentum conservation, the photogalvanic effect includes SAM and OAM of the photons coupling with the electrons. The electric field of a light beam with a polarization of ( is the light helicity) and OAM () can be written by[523] where right-handed (left-handed) circularly polarized light has , and is the intensity profile. As indicated by Eq. (42), the light beam carrying OAM has a phase distribution in the direction, or equivalently, it can be described as having an azimuthal momentum besides the wave vector . Then, the direct current (DC) generated by is expressed as[523] where represents the second-order conductivity as a function of frequency and azimuthal momentum (). Vortex beams carrying different OAMs have different DC responses; therefore, OAMs can be identified based on DC current intensity.4.1.3.Thermoelectric response of spin-Hall effectSince the phase of the surface plasmon polariton far field excited by a single nanoslit is correlated with the rotation angle, two nanoslits with orthogonal rotation angles can map light beams carrying different SAMs in different directions (left or right) of the structure. Moreover, the asymmetric structure can focus the incident light carrying different OAMs at different central positions. Combining these two principles, the detection of SAM and OAM can be achieved simultaneously by designing special surface plasmon polariton metasurfaces. For example, considering the incident light with an electric field of , Eq. (40) can be rewritten as[524] By employing the paraxial approximation, the position at the peak focal point can be analytically expressed by[518] where is the first non-null zero point of the derivative of electric field distribution along the -direction. Therefore, the incident light carrying different SAM and OAM will be focused at different positions on the central region of the metasurface. However, the results of this detection method require a near-field microscope to characterize the surface plasmon polariton field, and it is not possible to read the SAM and OAM information of the incident light directly. A potential approach to directly read the SAM and OAM information of the incident light is to integrate a 2D material that is sensitive to the position and intensity of the surface plasmon polariton signal in the central region of the metasurface. For example, the 2D photothermoelectric material can cover the central region of the surface plasmon polariton metasurface. When the surface plasmon polariton focused spot undergoes a change in intensity, the intensity of the electrical response of the photohermoelectric material also undergoes a corresponding change. By properly designing the coverage area of the material, it is possible to directly detect the SAM and OAM information of the incident light signal.4.1.4.Diffractive deep neural network (D2NN)1) Wave propagation analysis in a D2NN Following the Rayleigh-Sommerfeld diffraction equation[704], each individual neuron of a given D2NN layer can be considered as a secondary source of a wave composed of the following optical mode[705]: where represents the -th layer of the network, represents the -th neuron located at (, , ) of layer , is the illumination wavelength, , and . The secondary wave’s amplitude and relative phase are governed by the product of the neuron’s input wave and transmission coefficient (), both complex-valued variables. Based on this, for the -th layer of the network, one can write the output function () of the -th neuron located at (, , ) as[705] where is defined as the input wave to -th neuron of layer , refers to the relative amplitude of the secondary wave, and refers to the additional phase delay that the secondary wave encounters due to the input wave to the neuron and its transmission coefficient. The secondary waves diffract between the layers and interfere with each other, forming a complex wave at the surface of the next layer. This complex wave then feeds the neurons of the subsequent layer. The transmission coefficient of a neuron is composed of amplitude and phase terms, . For a phase-only D2NN architecture, the amplitude is assumed to be a constant, ideally 1, ignoring the optical losses.2) Forward propagation D2NN model In order to facilitate the interpretation of the forward model in the neural network, Eq. (47) can be rewritten as[705] where refers to a neuron of the -th layer and refers to a neuron of the next layer, connected to neuron by optical diffraction. The same expressions would also apply to a reflective D2NN with a reflection coefficient per neuron: . The input pattern , which is located at the input plane (layer 0), is a complex-valued quantity and can carry information in its phase and/or amplitude channels. The resulting wave function, which is a consequence of the diffraction of the illumination plane wave as it interacts with the input, can be expressed as[705]which connects the input to the neurons of layer 1. Assuming that the D2NN design is composed of layers (excluding the input and output planes), then a detector at the output plane measures the intensity of the resulting optical field[705]:In accordance with the forward model, the results of the network output plane are compared with the targets (for which the diffractive network is training), and the resulting errors are backpropagated in order to iteratively update the layers of the diffractive network. 3) Error backpropagation To train a D2NN design, a loss function should be defined to evaluate the performance of the D2NN output with respect to the desired target, and the algorithm iteratively optimizes the diffractive neural network parameters to minimize the loss function. In general, using the mean square error between the output plane intensity and the target defines the loss function () [705]: where refers to the number of measurement points at the output plane. Different loss functions can also be used in D2NN. In light of the aforementioned error definition, the optimization problem for a D2NN design can be formulated as[705]Through a process of continuous iteration, the discrepancy between the output value and the target value is gradually reduced, thereby enabling the neural network to recognize different incident structured light. 4.1.5.Inverse design methodPhotonic devices designed through inverse engineering hold vast potential for applications across various fields of modern optics. Traditional photonic device design is typically based on known physical models, followed by numerical simulations to optimize the structures. Since the device structure heavily relies on prior models, the degree of freedom in traditional optimization design is limited. In contrast, inverse design engineering does not necessitate forward solving based on existing theoretical models and can reverse design structures based on the target functionality, which in turn leads to the design of more flexible structures[283,706,707]. Typical inverse design algorithms include gradient descent algorithms, concomitant algorithms, genetic algorithms, etc. These algorithms can be employed to design high-performance and compact structured light recognition devices through the implementation of specific optimization strategies. These methods break the design limitations of traditional approaches, allowing efficient parameter optimization across the entire parameter space, making it more likely to achieve device structures with optimal performance, including integrated structured light detection devices. 4.2.Diverse Integrated Structured Light Detection Schemes4.2.1.Structured light detection based on surface plasmon polaritonsThe structured light detection scheme based on surface plasmon polaritons relies on the influence of the special polarization and phase distribution of structured light on a surface plasmon polariton field generated on the metal-dielectric interface, which can be used to detect SAM and OAM. By designing the micro-nano structures etched on the metal surface, the information of incident structured light can be acquired with near-field microscopy. According to the complexity of micro-nano structures etched on the metal surface, structured light detection schemes can be roughly divided into four types, including structured light detection based on nanoslits, structured light detection based on metaslits, structured light detection based on plasmonic metasurfaces, and structured light detection based on other special structures. The structured light detection based on nanoslits means the micro-nano structure etched on the metal surface is a simple geometric structure, such as a rectangular nanoslit and micro-nano ring. The surface plasmon polariton field excited by nanoslits of different shapes can be seen as different integral functions of Eq. (39). As shown in Fig. 49(a), using the semi-ring plasmonic nanoslits can directly and spatially detect the vortex beam with different topological charges. The SEM image of the semi-ring nanoslit array on the metal surface is shown in Fig. 49(b). By illuminating a circularly polarized vortex beam on the nanoslits to excite a surface plasmon polariton field with a helical wavefront, these phase-modulated plasmons are focused into spatially separated subwavelength spots for different vortices due to constructive interference, as shown in Fig. 49(c)[518]. There are about 120-nm-spaced focus intervals on the metal surface between the neighboring OAM modes observed in the experiment. However, the surface plasmon polariton focus has a similar spot diameter value compared with the focus shift between the neighboring OAM modes under near-field microscopy characterization, which also puts forward higher requirements for the characterization equipment. Compared with using this focused surface plasmon polariton field distribution to detect different OAM modes, the matching mechanism between the periodic structure and the surface plasmon polariton wave vector can be used to achieve directional emission of the surface plasmon polariton field, and different OAM beams can be clearly distinguished after a certain transmission distance, as shown in Fig. 49(d). It is implemented by coupling the OAM beam into two separated surface plasmon polariton beams according to a well-designed in-plane wave vector (-vector) matching process using a nanograting etched on the metal surface. This structure can detect the OAM property directly according to the relation between the splitting angle of surface plasmon polariton beams and the topological charge of the vortex beam. From Fig. 49(e), it can be seen that on each side of the grating structure, there will be two -vector matching conditions. This indicates that two couples of separate surface plasmon polariton waves can be launched with the same splitting angle. In the experiment, different separation distances () are also detected by the charge-coupled device (CCD), when a series of OAM beams with different topological charges (, , , , , ) is incident on the asymmetric nanoslit grating. The red error bars in Fig. 49(e) show the experimental data of the separation distances , which agree well with the calculated curves[708]. To further resolve OAM beams carrying different SAMs, Fig. 49(f) proposes the symmetry-breaking metaslit grating with different periods for the upper and lower parts, which enables the unidirectional excitation of the surface plasmon polariton depending on the topological charge and polarization of the incident vortex beams. The symmetry-breaking nanograting diagram is shown in Fig. 49(f). For OAM beams with LCP/positive, RCP/negative, LCP/negative, and RCP/positive topological charges, the surface plasmon polariton field propagates to the first, second, third, and fourth quadrants, respectively, as illustrated in Fig. 49(g)[709]. However, this kind of on-chip diffraction structure often has the long lateral propagation distance to ensure that adjacent topological charges have a large longitudinal interval on the output grating, thereby ensuring the accuracy of identification. Moreover, it is noted that this scheme needs to coincide the horizontal central axis of the OAM beam with the horizontal axis of the asymmetric periodic microstructure; otherwise the accuracy of the measurement will be affected. This is difficult in practical measurements. Figures 49(h) and 49(i) propose and show that the metaslit ring can support absolute measurement of the OAM and SAM of the incident light via angle detection[710]. Compared with the OAM detectors relying on the measurement of the relative interval between the focused intensity spots of the surface plasmon polariton, this method does not depend on the measurement of any reference-order OAM beam. Similarly, using other shapes of micro-nano structures can also map the OAM information of different OAM modes on the propagation direction of the excited surface plasmon polariton field. The catenary grating can be also used to distinguish the positive and negative topological charges of the incident OAM-carrying vortex beams by employing the wave vector matching, which is shown in Figs. 49(j) and 49(k)[711]. However, it is difficult to align the vortex beam with the micro-nano structure. Therefore, we believe that it makes sense to detect the topological charges (SAM and OAM) of the structured light in the case of non-alignment. Unlike the interference effect of a surface plasmon polariton field on the metal-dielectric interface, a plasmonic metasurface can also map structured light carrying different phase and polarization information in different space locations of the transmitted light. It is similar to the structured light generation based on the metallic metasurface. Figure 50(a) shows a designed plasmonic metasurface to measure the topological charge of the OAM beam by self-interference. Many V-shaped metal nano-antennas of different lengths and angles are distributed on the surface of the medium, as shown in Fig. 50(b)[712]. When a linearly polarized OAM beam is incident to the metasurface, an identical OAM beam with 33% transmittance is refracted, and two cross-polarized OAM beams with the same value of topological charge generated by the metasurface are radiated at oblique angles. The near-field interferogram of the cross-polarized OAM beams is dependent on the topological charge. The sign and modulus of the topological charge are revealed by the symmetry and the spacing of the inner bright spots of the interferogram, respectively. In addition, the bandwidth of the metasurface is over a broad wavelength range from 1.52 to 1.6 µm. Similarly, Fig. 50(c) proposes a spin-multiplexing metasurface based on another micro-nano structure for detecting simultaneously phase and polarization singularities. The vortex light with different OAM and SAM forms focal points in different regions of the transmission surface after passing through the semicircular nanostructures etched on the metal surface. The binary phase distribution of the metal metasurface and the simulated results are shown in Fig. 50(d)[713]. Moreover, the measured experimental results agree well with the theoretical predictions, proving that the proposed method is successful in the detection of both the phase and polarization singularities. 4.2.2.Structured light detection based on dielectric micro-nano structuresThe above metal micro-nano structures and plasmonic metasurface on a metal material platform show favorable performance of structured light detection. However, due to the large ohmic loss of metal, the transmitted light usually has a relatively weak power. This means that the detection equipment should have great noise reduction ability to identify the topological charge (e.g. OAM) and polarization (e.g. SAM) information of the incident structured light. Since the inherent loss of a dielectric metasurface is much smaller than that of metal, the detection of structured light based on a dielectric metasurface is a more advantageous scheme. Figure 51(a) shows an all-dielectric metasurface for realizing high-efficiency and broadband detection of OAM modes. The unit structure of the dielectric metasurface is shown in Fig. 51(b), which is a dielectric rectangular nanopillar. Changing the rotation angle of rectangular nanopillars can realize the phase modulation of the transmitted light, which is the so-called geometric phase or PB phase. The relationship between the rotation angle and the phase of the transmitted light in the far-field region is described as Eq. (11). Figure 51(c) shows the SEM image of the dielectric metasurface[519]. Structured light carrying different OAM and SAM is detected using the single-layer all-dielectric metasurface at different wavelengths. However, the input light carrying different OAM orders has different powers of the focused spot on the output plane. At the same time, the position of the focused spot on the output plane does not vary uniformly when increasing the OAM order. The scheme shown in Fig. 51(d) can solve this problem. It proposes a single azimuthal-quadratic phase metasurface-based photonic momentum transformation (PMT). The polarization conversion ratio (PCR) of the unit cell is shown in Fig. 51(e), and the inset depicts the structure of the unit cell, which is made up of hexagonally arranged nanopillars on a substrate. The vortex beams with different OAMs are converted into focusing patterns on a transverse plane with distinct azimuthal coordinates based on this dielectric metasurface. The SEM image is shown in Fig. 51(f)[520]. It is interesting to note that this scheme may identify superposed vortices with a certain interval step as well as vectorial vortices with phase and polarization singularities. In addition, Fig. 51(g) presents the Dammann optical vortex grating (DOVG), which can also detect the angular momentum of the vortex beam based on the vector beam holographic computation. It eliminates the bottleneck of classic optical vortex grating detection devices with unequal power distribution, which is shown in Figs. 51(h) and 51(i)[714]. This scheme can increase the number of multiplexed and demultiplexed channels in OAM optical communications as well as the parallel detection range of OAM. In addition to dielectric metasurface and Dammann grating, the waveguide grating can also couple different OAM modes into different output waveguides, as shown in Fig. 51(j). The waveguide grating coupler is a phase-matching element. In order to detect different OAM modes at different output waveguides, the excited guided modes in the waveguide and the OAM modes should satisfy the phase matching condition, as shown in Fig. 51(k)[715]. 4.2.3.Structured light detection based on orbital photogalvanic effectNumerous theoretical and experimental studies have shown that the OAM of light interacts with the atomic medium to produce new selection rules and optical responses. These studies show that optical phase gradients modify the excitation process, but the results do not translate into direct photocurrent generation for OAM-sensitive photodetectors. This is because the direct photocurrent response itself does not carry phase information, and the slow variation of the vector potential associated with the OAM of light relative to the size of the Brillouin zone limits its effect on microscopic processes[716]. It is found that the OAM of light can induce a strong nonlocal interaction between electromagnetic waves and matter. Therefore, some researchers explore how the OAM generates photocurrent during the photoelectric conversion process and design a vortex light photodetector that can detect the phase directly, as shown in Fig. 52(a). This scheme designs a photodetector based on tungsten ditelluride () with an electrode geometry carefully crafted to facilitate direct characterization of the topological charge of OAM of light. This orbital photocurrent effect, driven by a helical phase gradient, is characterized by a current winding around the beam axis whose magnitude is proportional to its quantized OAM modulus. Figure 52(b) shows the electrode structure. The obtained results show that both and beams produce polarization-dependent currents, as shown in Fig. 52(c). The circular polarization current () values of the and beams have similar amplitudes but opposite polarities. The value of under different OAM modes incident condition is shown in Fig. 52(d)[523]. The mechanism of the photocurrent can be understood as light simultaneously transferring its OAM and energy to the electron. As the optical phase varies in the azimuthal direction, it induces a spatial imbalance in the excited carriers, resulting in the net current. In addition to scalar OAM beams, there are also vector OAM beams with spatially variable polarization states and spiral phase distributions. Subsequently, the response performance of OAM photodetectors based on the second type Weyl semimetal is verified. devices with specially designed electrodes in geometry can indeed achieve direct detection of OAM at 4 µm (a typical mid-infrared wavelength) by the orbital photogalvanic effect. Figures 53(a) and 53(b) show the device structure and detection scheme diagram. Figure 53(c) shows the schematic diagram of the orbital photogalvanic effect measurement. The light spot of the LG beam is focused on the center of the arcs defined by the U-shaped electrodes during the measurement process, as shown in Fig. 53(d). The experimental results show that the direction and magnitude of the orbital photogalvanic effect-driven photocurrent are proportional to the OAM quantum number of the incident light beam. Figure 53(e) shows the photocurrent response for different OAM orders at a constant excitation power of 2.5 mW[717]. The radius of the ring, which is defined by the U-shaped electrodes, is kept constant for different OAM orders in the measurement. Benefiting from the topological enhancement of the shifted current response, the device shows a significant photocurrent response at 4 µm. The experimentally observed orbital photogalvanic effect response comes from the phase gradient of the light field rather than from other photocurrent generation mechanisms. The work also performs beam size and spot position dependence measurements to confirm that electrodes with specially designed geometry are effective in collecting currents of the orbital photogalvanic effect when the spot size and position of the beam are matched to the electrode structure. Combined with the linear and circular polarization sensitivities possessed by , and the conventional light intensity response capability, such devices designed into photodetector arrays have the potential to support multi-optical parametric characterization in the mid-infrared band. 4.2.4.Thermoelectric detector with spin-Hall couplerIn contrast to the OAM detector based on the orbital photogalvanic effect, the SAM and OAM of a structured light can be simultaneously detected by combining a metasurface and a photothermoelectric material. Figure 54(a) shows an on-chip SAM and OAM photodetector based on a 2D broadband photothermoelectric material (). The photodetector consists of a surface plasmon polariton metesurface, a detection area covered by , and four electrodes for reading the electric response. The surface plasmon polariton metasurface is a well-designed SAM and OAM detector based on the spin-Hall effect, as shown in Fig. 54(b). It is divided into two semicircular regions with different whetstone-type meta-structures, which focus on the LCP and RCP lights in the center region of the rings, respectively. The 2D material located in the center region of the metasurface can convert optical signals into electrical signals. Four electrodes are connected at the boundary of the 2D material, with the purpose of detecting the photothermoelectric response of the material. As a consequence of structured light with different SAM and OAM focusing on different positions on the metasurface, the intensity of the electrical signals received at the four ports is different, as illustrated in Fig. 54(c). The photovoltage responses ( and ) from the port-pairs and for LCP and RCP vortex beams, respectively, are measured with a constant optical power (125 mW). Meanwhile, the topological charge ranges from to 4 for the LCP and RCP beams. As shown in Fig. 54(d), the experimentally determined photovoltage responses exhibit a strong correlation with the topological charge for LCP and RCP beams[524]. In addition to the intensity of the focal surface plasmon polariton spots, the photothermoelectric response in this scheme is also dominated by the position of the focal surface plasmon polariton spots. The photovoltage responses exhibit a nearly linear dependence on the OAM for a fixed SAM mode, which is consistent with the relationship between the surface plasmon polariton focal position and the OAM. 4.2.5.Structured light detection based on diffractive planesSimilar to the metasurface-based structured light detection, the phase plane can be also designed based on the scattering or diffraction principle, which maps different structured light to different regions in space. This technique combined with neural network training allows the design of powerful optical computing architectures to recognize different structured light and even complex signals[705]. Wavefront shaping with a diffuser has recently been used for OAM detection. Different OAM modes can be separated into several predetermined spatial regions through wavefront shaping by a liquid crystal spatial light modulator and a diffuser. The feedback signal is the optical intensity that is recorded by the detector. The diffuser offers more adjustable degrees of freedom than a super lens for wavefront shaping because of the multiple scattering, which allows it to surpass the traditional diffraction limit and raise the numerical aperture for focusing. Using the complex amplitude wavefront shaper based on a digital micromirror device (DMD) through a diffuser to shape the whole field (phase and amplitude) of the OAM modes, Fig. 55(a) shows a unique method for detecting multiplexed OAM modes more quickly and accurately. In order to detect multiple OAM modes, a complex amplitude mask is loaded onto the wavefront shaper and then multiple OAM modes are modulated by a diffuser. The diffuser functions like a superlens, providing different phase modulations for different OAM modes and separating them to different predetermined spatial locations. The yellow grid in Fig. 55(b) depicts that each element mask of the target amplitude field and phase distribution is made up of 5 pixel × 5 pixel. Figure 55(c) shows the appropriate DMD mask element, which is made up of 5 superpixel × 5 superpixel (amplitude and phase). Figure 55(d) shows multiple OAM modes mapped to different locations in space after wavefront shaping encoding and the diffuser plane, which enables the detection of structured light[718]. By combining the neural network training, multilayer diffractive planes can achieve complex light field manipulation, including structured light detection. Figure 56(a) shows a multilayer diffractive neural network designed using a neural network training algorithm in combination with a spatial angular spectral propagation function. The diffraction-based processor uses multiple phase planes to facilitate the measurement of the OAM spectrum. Meanwhile, this OAM mode detection scheme also includes the electronic neuron plane, which guarantees precise and direct detection of OAM modes. Specifically, optical layers extract the invisible topological charge information from the incoming light, and an electronic layer predicts the precise OAM spectrum. The hybrid optoelectronic neural network, formed by the synergy of optical and electronic neurons, does not require interference measurements and repetitive steps, which promises a compact single-shot system with high speed and energy efficiency (optical operations/electronic operations ). It also has great potential to be applied to different types of structured light detection. Using the experimental configuration shown in Fig. 56(b), various complex optical fields are reconstructed through the four-step phase shift approach for OAM spectrum detection. Note that Fig. 56(b) is used for preparing various experimental test sets, and some typical results are depicted in Fig. 56(c), which show the intensity and phase distributions of the generated single OAM modes and multiplexed OAM modes (equal/random weights). Figure 56(d) shows the blind test performance on single OAM modes by averaging the obtained results from 30 repetitions in the experiment. The dominant values on the diagonal show that the detector outputs a near-perfect OAM spectrum in the experiment with single-mode input. Figure 56(e) shows two selective results of OAM spectrum measurement for multiplexed OAM modes with equal weights and random weights, respectively[525]. The obtained results show the favorable performance of a hybrid optoelectronic neural network for structured light analysis and detection in a fast, accurate, and robust manner. Remarkably, the structured light detection based on multilayer diffractive planes (diffractive optical neural network) has a certain similarity with structured light demultiplexing based on MPLC. Remarkably, for the demultiplexing of structured light, the reverse operation of multiplexing, more attention is paid to the efficiency, crosstalk, and scalability, which are important factors to consider in structured light multiplexing communications. For the detection of structured light, the main consideration is to identify, recognize, or analyze the unique information features contained in each of the structured light, e.g., OAM spectrum measurement by the hybrid optoelectronic neural network. 4.2.6.Structured light detection based on silicon MZI networkIn addition to the above structured light detection based on the hybrid optoelectronic neural network with multilayer diffractive planes in out-of-plane 3D space, structured light detection can also be realized based on in-plane 2D chip-scale matrix operations with similar functions to linear neural networks. It has been reported that any matrix can be decomposed into the unitary matrix and diagonal matrix[719]. Using this method, one can build up the silicon-based neural network to implement vowel recognition, handwritten digit recognition, and multiple-in to multiple-out tasks with the MZI architecture. Remarkably, the structured light with spatially variant amplitude/phase/polarization can be sampled as a complex input vector when fed into the silicon MZI network with multiple input ports. By training the bias weights of the MZI network, the recognition and beam splitting of different structured light can be realized. As shown in Fig. 57(a), the integrated structured light detector consists of a mesh of tunable beam splitters, which are realized by tunable MZIs. The topology structure of the detector includes rows of cascaded MZIs. Arbitrarily overlapping orthogonal free-space light beams with the same wavelength and polarization can be separated using this integrated structured light detector. Figure 57(b) shows a top-view micrograph of the structured light detector, which has a footprint of . The detailed structures are shown in Figs. 57(c)–57(f). As shown in Fig. 57(c), the input 2D optical antenna array consists of grating couplers, which are all aligned in the same direction in a square configuration. The center-to-center spacing between the grating couplers of the 2D optical antenna array is 49 µm, resulting in multiple diffraction orders in the far-field radiation mode. For example, Fig. 57(d) shows the collimated far-field intensity field radiation from a uniform grating array measured with a near-infrared camera. All 15 tunable MZIs in the grid (eight MZIs in the first row and seven MZIs in the second row) are identical and controlled by a thermal tuner. Details of the MZI thermal tuner structure are shown in Fig. 57(e). Modulation of the intensity and phase of the MZI output is achieved by heating one of the interference arms of the MZI. Transparent photodetectors are used to monitor the switching state of each MZI to implement automatic tuning and stabilization procedures. The photonic chip is mounted on an electronic printed circuit board (PCB), as shown in Fig. 57(f), which contains the electronic ASIC for readout of the on-chip detector, as well as connects to a field programmable gate array (FPGA)-based board for real-time data processing and driving of the thermal tuner[526]. When the chip is used for structured light detection, the programmable photonic processor, acting as the mode-diversity receiver, simultaneously recognizes and separates pairs of orthogonal light beams at different output ports, i.e., the WG1 and WG2 as marked in Figs. 57(a) and 57(b). Favorable operation performance is achieved with low crosstalk even with propagation-perturbation-induced mode mixing. It is worth noting that, in principle, coupling, separation, and detection of free-space light beams can be realized by a programmable photonic processor on any set of orthogonal structured light. 4.2.7.Structured light detection based on inverse design subwavelength structureThe on-chip photonic device based on an inverse design using an optimized algorithm can also be used to distinguish or detect different structured light. This kind of device is fixed without tunability when fabricated, which differs from the above tunable MZI network but always allows for greatly reduced chip size. A typical technique for compact on-chip detection of various planar beams employing the subwavelength structure of the micrometric footprint and nanometric thickness is shown in Fig. 58(a). The subwavelength structure is designed and optimized using a linkage-tree-based genetic algorithm. It is demonstrated in the experiment on a surface plasmon polariton platform for the detection of HG beams. Figure 58(b) shows the experimental setup with the detector manufactured on a silver layer. In the device fabrication, a 70-nm silver layer is evaporated on a BK7 glass substrate, followed by spin-coated 50-nm poly (methyl methacrylate) (PMMA) resist. Electron lithography is used to define the device structure, and the focused ion beam (FIB) milling is used to add a grating coupler within the silver layer for the excitation of the desired HG plasmonic beams at the silver-air interface. The plasmonic beams propagating through the device are observed with a near-field scanning optical microscopy (NSOM) system by collecting the plasmon near-field intensity distribution. The structured light detection device is contained within a region. In the experiment, the target is to distinguish or detect the zeroth-order (HG0) and first-order (HG1) Hermite-Gaussian beams by focusing energy at the end of the device onto two separate spots. As shown in Fig. 58(c), the propagation through the detector of HG0 on the left and HG1 on the right is simulated by the beam propagation technique (BPM) method and demonstrated in the experiment. The simulated and experimental results shown in Fig. 58(c) both verify the successful routing and detection of two different light beams into the two output desired spots. Figure 58(d) depicts corresponding simulated and experimentally measured intensity profiles at the end of the device, which are marked by the red cross-section dashed lines in Fig. 58(c). The crosstalk between the two detected channels is less than 0.12. Figure 58(e) shows the intensity measured by the NSOM integrated along the transverse direction when propagating inside the detector device[527]. The propagation losses of the detector can be evaluated with the power transmission of 74% for the HG0 mode and 67% for the HG1 mode. Furthermore, another structured light detector with a size of is also implemented for the detection of HG0, HG1, and HG2 plasmonic beams. The proposed inverse design subwavelength structure can be readily applied to other types of structured light detection with an ultra-compact footprint. 4.2.8.Structured light detection based on silicon nanorod optomechanicsTraditionally, structured light beams have a special phase or polarization distribution in the cross-section perpendicular to the propagation direction. In recent years, various methods have been used to generate transverse vortex light in free space, also called transverse orbital angular momentum (TOAM) beams. The spiral phase plane of a TOAM beam is parallel to the direction of propagation. Unless examining polychromatic fields or extrinsic momentum when a beam propagates away from the coordinate origin, TOAM is seldom ever detected. Theoretically, intrinsic TOAM in monochromatic fields has been described in the near field of the diffraction by the acute barriers or slits and in superimposed co-propagating beams with differing beam-widths and/or amplitudes. Intrinsic TOAM manifests as complicated transverse phase singularity vortex lines. Using a levitated optomechanical sensor as a probe, Fig. 59(a) carries out a novel analysis of this exotic optical OAM. It also proposes a simple approach for synthesizing the optical vortices containing inherent TOAM. Therefore, a powerful new tool for the optical manipulation of matter is made possible by the ability to apply strong optical forces transverse to the propagation direction of free-space beams without the requirement for localized interference patterns, polychromatic light, or crucial optical alignment. In order to create an optical trap, two linearly polarized, counter-propagating laser beams are concentrated within a vacuum container with the capacity to introduce an - and -direction separation, which is shown in Fig. 59(b). The motion of silicon nanorods can be traced in five degrees of freedom: three translational (, , ) and two angular (). The SEM image of the silicon nanorod array is shown in Fig. 59(c). The nanorod receives the optical OAM, which causes it to rotate. The equilibrium between the optically produced torque and the damping brought on by the presence of gas determines the rotation rate of the nanorod. Figure 59(d) shows the theoretically computed torque and the measured torque when is altered. The predicted torque is marked by the solid line in Fig. 59(d). The highest rotation rate would occur when the optical intensity is at its maximum (), if the rotation is merely caused by the transfer of momentum from a single beam (as in the case of circularly polarized light). On the contrary, for a specific , a separation along the direction results in a single maximum at , which is shown in Fig. 59(e)[720]. The presented work uses levitated particles (e.g., silicon nanorods) as probes of structured light fields. The fact that the light field structure can be sensitively detected by the particle probe opens the opportunity for diverse structured light detection using fundamental physics in light-matter interactions. As a short summary, Table 3 lists the main recent representative research works on integrated structured light detection. We summarize the key parameters such as types of structured light, detection principles, material platforms, and working wavelengths for comparison[518–520,523–527,708–715,717,718,720–722]. Table 3Parameters and Performance of State-of-the-Art Integrated Structured Light Detection
5.Integrated Structured Light ApplicationsAfter significant progress made in the aforementioned techniques of integrated structured light generation, integrated structured light processing, and integrated structured light detection, various integrated structured light applications have also achieved rapid development in recent years. The unique spatial structure, the added spatial degree of freedom, and the rich spatial amplitude, spatial phase, and spatial polarization information carried by the structured light make it widely used in various advanced applications. In this section, the basic theories supporting structured light applications and diverse integrated structured light application scenarios are reviewed, as outlined in Fig. 60. The first part introduces some basic theories of structured-light-enabled applications, especially for structured light communications. The later parts discuss several typical examples of integrated structured light applications, such as optical communications, imaging, holography, interferometry, microscopy, optical manipulation, optical trapping, optical tweezers, optical metrology, and quantum optics. 5.1.Theories and Principles of Structured Light ApplicationsThis part mainly introduces some typical theories and operation principles of structured light applications, for example, data-carrying structured light communications, structured light holography, and optical manipulation and trapping. Some other common theories with the structured light generation, processing, and detection will not be elaborated specifically. 5.1.1.Data-carrying structured light communications1) Structured light modulation communications Structured light modulation communication means to produce time-varying structured light for data information transfer[63,64,66,68]. Since the structured light, accessing the space degree of freedom of light waves, has, in principle, infinite available orthogonal states, different structured light can be used to encode and decode data information, which is very similar to traditional modulation formats on complex amplitude degree of freedom of light waves, such as on-off keying (OOK), phase-shift keying (PSK), and quadrature amplitude modulation (QAM). The electric field of the structured light used for modulation communication can be expressed by where represents the spatial structure of the structured light, is the wavenumber, and is the angular frequency. The time-dependent spatial structure of the structured light, which can be regarded as structured light shift keying, enables the data information transfer. For instance, using OAM-carrying optical vortices, optical vortex modulation communications by OAM shift keying have been widely studied.2) Structured light multiplexing communications Remarkably, different orthogonal structured light beams can not only encode and decode data information but also serve as carriers to carry independent data information, the so called structured light multiplexing communication, which is similar to the well-established WDM communication exploiting the frequency/wavelength degree of freedom of light waves[60,63,64,66,68,69]. For the multiplexing of data information-carrying structured light beams, the multiplexed electric field can be expressed as where represents the data information carried by different structured light , . Note that is encoded on other degrees of freedom of light waves, e.g., complex amplitude. Although structured light beams are superposed on one another, each structured light beam carrying its own data information is, in principle, distinguishable from the others due to its unique spatial structure and the intrinsic orthogonality of different structured light beams.Remarkably, structured light multiplexing is fully compatible with other existing multiplexing techniques, such as WDM, time-division multiplexing (TDM), and PDM. Hence, multiple multiplexing techniques can be combined together to efficiently increase the transmission capacity of multi-dimensional optical communications. For example, using data-information-carrying optical vortices, high-capacity multi-dimensional optical communications, incorporating the OAM multiplexing technique, have been widely studied. 3) Structured light multicasting communications In addition to structured light modulation and multiplexing techniques for communications, structured light multicasting communication has also attracted growing interest in recent years. Structured light multicasting communication is a kind of one-to-many communications, which means to deliver and broadcast the common data information carried by the source to multiple users[64,68]. One distinct feature of structured light multicasting communication is that multiple structured light beams, corresponding to multiple users, share the same data information. For the structured light multicasting communication, the electric field of the multicasted structured light can be expressed as where is the common data information broadcasted by the source and shared by multiple structured light users. By comparing Eqs. (54) and (55), one can clearly see the difference between structured light multiplexing communication (multiple data information channels each carried by a distinct structured light) and structured light multicasting communication (the same data information shared by multiple structured light beams). Note that the structured light multicasting technique can be also extended and combined with other degrees of freedom of light waves, enabling multi-dimensional multicasting communications with a large number of multicasted channels. For instance, using OAM-carrying optical vortices as multicasting users, OAM multicasting communications and OAM-assisted multi-dimensional multicasting communications have been widely studied.5.1.2.Structured light holographyFor structured light holography, the complex amplitude distribution of the electric field can be decomposed into an infinite number of plane-wave spatial frequency components according to the Fourier integral theorem, which can be expressed as[554] where and denote the orthogonal coordinates in the image plane and hologram plane, respectively. Equation (56) can be written as a Fourier transform of the field distribution in the hologram plane:It is noted that, in the Fourier optics, the electric fields in the hologram plane and image plane form a pair of Fourier transform. Hence, in the Fourier transform holography, the spatial frequency distribution of a hologram physically equals the complex amplitude distribution in the image plane. Taking the OAM-carrying light beam as a typical example, when an incident OAM beam is used for the holographic reconstruction, the spatial frequency components of the hologram are superposed with a helical phase front expressed as where is the topological charge of the OAM beam and is the azimuthal angle in the polar coordinate system. By substituting Eq. (58) into Eq. (57), we can obtain the spatial frequency distribution of the hologram based on an incident OAM beam, which can be written by where denotes the convolution operation. We then acquire the relationship between the electric fields in the image plane and hologram plane based on an OAM incident beam as follows: where represents the complex amplitude distribution in the image plane. It indicates that the OAM-reconstructed electric field in the image plane is a convolution between the electric field of a holographic image and the Fourier transform of a helical phase front. Mathematically, the Fourier transform of a helical phase front, serving as the kernel function of the convolution, is simply copied onto each pixel of the holographic image. Remarkably, in addition to OAM-carrying light beams, other kinds of structured light could be also, in principle, applied to structured light holography.5.1.3.Optical manipulation and trappingFor optical manipulation and trapping applications, tightly focused light generates a special force on the particles near the focal point. This force can be divided into two types: one is along the direction of light propagation, which can push the particle away, called scattering force; the other is to pull the particle to the location of the maximum intensity of light, called gradient force. The strongly focused light field forms a gradient light field in the horizontal and vertical directions, and the light intensity at the focal point is the largest, so the particles can be trapped at the focal point, just like an object falling into a bowl, no matter where the initial position is. It will eventually stabilize at the bottom of the bowl, which is called “potential trap” or “light trap.” This “three-dimensional light trap” can bind particles in , , and dimensions. In contrast, a light trap with a gradient field (gradient force) in the transverse direction of the beam and only pushes away particles (scattering force) in the longitudinal direction can be called a “two-dimensional light trap.” For example, Gaussian beams and evanescent fields have the ability to capture particles in two dimensions. When the scale of the captured object is similar to the wavelength, we could use the electromagnetic model to calculate the magnitude of the light trapping force on the particle. When the electromagnetic field acts on a particle of volume , the light trapping force can be expressed as It presents that the force acting on a particle of volume can be converted into a force acting on its closed surface . represents the Maxwell stress tensor, and each component of can be written by[723] where . and are the dielectric constant and magnetic permeability, respectively. denotes the component of the trapping force acting on the unit area perpendicular to the -axis on the -axis. In the Cartesian coordinate system, the Maxwell stress tensor can be expressed as where the numbers 1, 2, and 3 correspond to the , , and components of the Cartesian coordinate system, respectively.5.2.Diverse Integrated Structured Light Application Scenarios5.2.1.Structured light applications in optical communicationsIn this part, several typical examples of structured light applications in optical communications are introduced: analog signal transmission using OAM modes, data-carrying digital signal transmission using OAM modes, silicon-chip-assisted high-speed spatial light modulation communication, femtosecond laser inscription enabling chip-chip and chip-fiber-chip optical interconnects, integrated optical vortex emitter enabling direct fiber vector eigenmode multiplexing transmission, data-carrying on-chip MDM transmission using in-plane waveguide modes, 3D/2D integrated photonic chips enabling multi-dimensional data transmission and processing, and a silicon photonic processor empowering free-space and fiber-optic multimode communications. 1) Analog signal transmission using OAM modes We propose and evaluate the performance of an analog signal transmission system with a photonic integrated vortex emitter and 3.6-km FMF link using OAM modes. Figure 61(a) shows the experimental setup. The output of a tunable laser is modulated by an intensity modulator driven by the 3-GHz radio frequency (RF) on the transmitter side to generate an analog signal, and an erbium-doped fiber amplifier (EDFA) is used to amplify the signal. Before the signal is coupled into the input waveguide of the photonic integrated vortex emitter, a polarization controller (PC) adjusts the light in the quasi-TE mode, and a power meter is put at the output port of the device to measure the power. In the experiment, and modes are generated by the photonic integrated vortex emitter to carry analog signals. In order to characterize the performance of the analog signal transmission, the RF carrier output power and distortion as a function of RF input power are measured with at a wavelength of 1531.91 nm and at a wavelength of 1556.56 nm. The measurement results are shown in Figs. 61(b) and 61(c), respectively[724]. Spurious free dynamic range (SFDR) of the second-order harmonic distortion (SHD) is an important factor used to estimate the analog link performance. One can observe analog signal transmission penalty ( degradation of SHD SFDR) for both OAM modes after transmission through the combined photonic integrated vortex emitter and 3.6-km FMF link, which could be ascribed to the optical transmission loss (emission loss of the vortex emitter, coupling loss from free space to FMF, coupling loss from free space to SMF, propagation loss of FMF, loss of optical components) and notch filtering effect of the photonic integrated vortex emitter. 2) Data-carrying digital signal transmission using OAM modes In addition to analog signal transmission, we also propose and demonstrate OAM modes emission from the photonic integrated vortex emitter for 20-Gbit/s quadrature phase-shift keying (QPSK) signal transmission in FMF. A microring resonator with a grating patterned along the inner wall can generate different OAM modes. The SEM image of the fabricated device with a 7.5 µm ring radius and the schematic of the 3D structure of the photonic integrated vortex emitter on an SOI platform are shown in Fig. 62(a). In particular, an aluminum mirror is introduced below the etched microring resonator via wafer bonding to effectively improve the emission efficiency. The rotating whispering gallery mode in the etched microring resonator is coupled to a vertically propagating cylindrical vector beam in free space. The vector beam emitted into free space is then decomposed by a quarter-wave plate (QWP) and a polarizer to generate a linearly polarized OAM beam, which is coupled into the FMF for data-carrying OAM mode transmission. By properly matching the light wavelength with the microring resonance and detuning from the grating Bragg wavelength, the desired topological charge can be achieved. Figure 62(b) illustrates OAM modes emission from the device for transmission in FMF. As shown in Fig. 62(c), the intensity profiles of mode at a wavelength of 1529.02 nm and mode at a wavelength of 1552.32 nm as well as their demodulation after transmission through a 3.6-km three-mode FMF are measured. Figure 62(d) shows the measured bit-error rate (BER) curves of OAM modes transmission through the 2-km two-mode FMF link and 3.6-km three-mode FMF link based on the high-emission-efficiency photonic integrated vortex emitter. 20-Gbit/s QPSK signals are employed and favorable transmission performance is achieved in the experiment. The observed optical signal-to-noise ratio (OSNR) penalties are less than 4 dB at a BER of . The insets in Fig. 62(d) depict typical constellations of QPSK signals carried by OAM modes after FMF transmission[725]. 3) Silicon-chip-assisted high-speed spatial light modulation communication For the aforementioned structured light modulation communication, the traditional method is based on the liquid crystal SLM with a slow response time, which greatly limits the speed of communications. We propose an integrated approach to overcome the speed limitation of SLM for structured light modulation communications. We implement high-speed silicon-chip-assisted OAM encoding data information transfer by mapping the traditional amplitude modulation to spatial mode modulation and using an integrated OAM mode multiplexer. The integrated OAM mode multiplexer consists of two bus waveguides with various widths and a multimode microring resonator with the grating integrated in the inner wall. Figure 63(a) schematically illustrates the concept and principle of high-speed photon dimension mapping with the integrated OAM mode multiplexer. Two light beams with the same wavelength are modulated by two conventional high-speed intensity modulators driven by two opposite OOK data sequences from a bit-pattern generator (BPG). Such amplitude modulation technique is mature and can be very fast. The two light beams are then fed into access bus waveguides I and II of the integrated OAM mode multiplexer, respectively. From the integrated OAM mode multiplexer, two coaxially propagating OAM beams are generated. We may convert the time-varying amplitude-modulated OOK data sequence to a time-varying spatial-mode-modulated OAM beam sequence by appropriately altering the length of two light pathways, where and are encoded to indicate 1 and 0, respectively. As a result, the integrated OAM mode multiplexer facilitates high-speed amplitude modulation to OAM mode modulation mapping. This approach cleverly bypasses the speed limitation of conventional OAM encoding using SLM. The recorded temporal waveforms of two channels (CH1 and CH2) and their subtraction (CH1-CH2) of back-to-back (B2B) and after amplitude-to-OAM modulation mapping are shown in Figs. 63(b) and 63(c), respectively. The CH1 and CH2 in Fig. 63(b) are both Gaussian beams, but they have inverted amplitude modulated data sequences. The CH1 and CH2 in Fig. 63(c) are mapped to and modes, respectively. For the multiplexed (data information “1”) and (data information “0”) modes, the overall intensity keeps a “1” level in the time domain, while different time-varying OAM modes (OAM encoding) represent the data sequence mapped from the amplitude modulation. We measure the BER performance of high-speed amplitude-to-OAM modulation mapping, as shown in Fig. 63(d)[726]. By detecting and modes, the OSNR penalties at a BER of for 15-Gbit/s OAM mode modulation are measured to be about 1.0 dB. Moreover, subtracting the two opposite signals can further improve the OSNR penalty by about 1.4 dB, showing impressive operation performance. 4) Femtosecond laser inscription enabling chip-chip and chip-fiber-chip optical interconnects The above microring resonators with etched gratings are fabricated by electron beam lithography (EBL) and inductively coupled plasma (ICP) etching processes on a silicon platform. As an important supplement, a femtosecond laser direct writing technique can flexibly inscribe 3D photonic integrated devices for integrated structured light applications. We propose and fabricate femtosecond laser-inscribed 3D photonic chips to tailor OAM modes, providing a platform for versatile OAM mode optical interconnects. Using the designed and fabricated trench-waveguide-based OAM mode generator in Fig. 5(f) and OAM mode (de)multiplexer in Fig. 36(c), we further demonstrate chip-chip and chip-fiber-chip short-distance OAM-multiplexing-enabled optical interconnects. For chip-chip optical interconnects, as shown in Fig. 64(a), the fabricated 3D photonic chips are packaged with SMFs at the input and output ports. Figure 64(b) shows the experimental setup for chip-chip optical interconnects with OAM multiplexing (, ). 30-Gbit/s 8-ary quadrature amplitude modulation (8-QAM) signals are generated at the transmitter. Two-channel optical signals are coupled from SMFs into the on-chip OAM mode multiplexer. After on-chip OAM modes multiplexing transmission, two OAM modes are separated by another on-chip OAM mode demultiplexer and coupled into the SMFs for coherent detection at the receiver. Figure 64(c) shows the measured crosstalk matrix of the two-input two-output chip-chip optical interconnect system, and the crosstalk is less than . Figure 64(d) shows the measured BER performance for chip-chip OAM modes multiplexing transmission. The measured OSNR penalties at a BER of , the 7% overhead hard-decision forward-error correction (HD-FEC) threshold, are less than 3.5 and 1.3 dB for multiplexing transmission of and modes, respectively. The typical constellations are depicted in the insets of Fig. 64(d)[470]. For chip-fiber-chip optical interconnects, as shown in Fig. 65(a), a 2-km OAM fiber supporting OAM modes is employed in the experiment, with one end connected to an on-chip OAM mode multiplexer and the other end connected to an on-chip OAM mode demultiplexer. The OAM mode (de)multiplexers are packaged with SMFs and OAM fiber. Figure 65(b) shows the experimental setup for chip-fiber-chip optical interconnects with OAM multiplexing (, ), which is similar to Fig. 64(b). The differences are the extra 2-km OAM fiber transmission link and 20-Gbit/s QPSK signals employed in the experiment. Figure 65(c) depicts the measured intensity profiles of the generated mode, mode, and their multiplexing transmission after the 2-km OAM fiber. Figure 65(d) shows the measured crosstalk matrix of the two-input two-output chip-fiber-chip optical interconnect system, and the crosstalk is less than . Figure 65(e) shows the measured BER performance for chip-fiber-chip OAM modes multiplexing transmission. The measured OSNR penalties at a BER of (7% overhead HDFEC threshold) are less than 5.7 and 4.6 dB for multiplexing transmission of and modes, respectively. The insets in Fig. 65(e) are typical constellations of QPSK signals[470]. 5) Integrated optical vortex emitter enabling direct fiber vector eigenmode multiplexing transmission In addition to the spatial phase degree of freedom such as OAM-carrying light beams having helical phase fronts, spatial polarization distribution is also considered to be an important degree of freedom of photons, which can also enhance the transmission capacity of optical communications. Vector beams with diverse nonuniform spatial polarization distribution are promising candidates for multiplexing communications. The direct fiber vector eigenmode multiplexing transmission seeded by integrated optical vortex emitters is proposed and demonstrated, as shown in Fig. 66(a). Data-carrying fiber vector eigenmode multiplexing transmission is reported through a 2-km large-core fiber (LCF) using vector vortex modes (radially and azimuthally polarized beams) generated from silicon microring resonators etched with angular gratings, showing low-level mode crosstalk and favorable link performance. The high-order fiber vector eigenmodes and supported by the LCF are considered. Instead of bulky commercial SLMs, two small integrated optical vortex emitters are used to produce the and modes. By multiplexing two orthogonal fiber vector eigenmodes, the transmission capacity of the fiber link can be doubled with each fiber vector eigenmode carrying an independent data information channel. At the output port, the two fiber vector eigenmodes can be separated with low crosstalk. Figure 66(b) shows the observed intensity distributions of the mode, mode, and the multiplexing of the and modes after fiber transmission. The intensity distributions of and modes are also observed by a rotating polarizer. The measured intensity distributions indicate good quality of the and modes radiated from both chips and transmitted over the km-scale LCF. Slight skewing of the intensity distributions observed after rotating the polarizers could be due to undesirable components radiated from the chips or excited during fiber-optic transmission. To further investigate the system performance of data-carrying fiber vector eigenmode (, ) multiplexing transmission over a 2-km LCF, a system experiment is constructed. Emitted by Chips 1 and 2, the and modes carry 10-Gbit/s QPSK or 20-Gbit/s 16-QAM signals. Figures 66(c) and 66(d) depict the measured BER as a function of the received OSNR for the QPSK and 16-QAM-carrying fiber vector eigenmode multiplexing transmission, respectively. The insets in Figs. 66(c) and 66(d) show the QPSK and 16-QAM constellations, respectively, both showing favorable performance of vector beams multiplexing communications[296]. 6) Data-carrying on-chip in-plane waveguide modes MDM transmission As one important chip-scale implementation of SDM, the MDM technology uses orthogonal waveguide modes as independent channels to transmit data information, which can greatly increase the data transmission capacity of on-chip optical interconnects. However, due to the issues such as intermodal dispersion and phase mismatch caused by the high refractive index of silicon, current on-chip MDM technologies are limited to lower-order modes. Innovatively, metasurface structures have been introduced into mode multiplexers. Silicon metamaterials achieve a gradient refractive index distribution by altering the subwavelength-scale ratio of silicon to the upper cladding within a period, thereby controlling the effective refractive index of the waveguide modes over a large scale. This resolves the refractive index mismatch between different modes, enabling ultra-compact adiabatic mode coupling. This effective waveguide is known as a gradient index metamaterial (GIM) waveguide. Figures 67(a) and 67(b) show the schematic and SEM images of the GIM-based coupler designed for mode (de)multiplexing, respectively. Based on this principle, a silicon-based on-chip MDM with 16 modes is realized, more than tripling the channel count compared to traditional MDM devices. The 16 modes are transmitted in parallel with inter-channel crosstalk below . Utilizing self-developed chips, a 16-channel MDM transmission system is constructed, as shown in Fig. 67(c). Figure 67(d) shows the BERs of 40-GBaud 16-QAM signals for the 16 modes, all of which are below the 15% FEC threshold. Figure 67(e) displays the recovered 16-QAM constellations of all 16 modes, showing good signal quality[727]. Consequently, this transmission system can carry high-speed signals to achieve a net data rate of 2.162 Tbit/s per wavelength per polarization, with a spectral efficiency of 43.24 bit/s/Hz. Furthermore, a general design framework based on metamaterials is proposed. This framework can directly define the topological arrangement of metamaterial module devices through simple mathematical formulas, quickly and efficiently achieving on-chip high-order mode control. When designing metamaterial-assisted waveguide mode converters, dielectric perturbations are typically introduced at the peaks or valleys of the electric field distribution of high-order mode to maximize the overlap between high-order and fundamental modes, resulting in ultra-compact device sizes. Both theoretical and experimental results show that this mode converter can achieve high-order mode control up to the 20th order. To further demonstrate the potential applications of on-chip signal transmission systems based on dielectric metamaterial module mode control, an eight-channel MDM on-chip signal transmission system is constructed, as shown in Fig. 67(f). Figures 67(g) and 67(h) display the microscope and SEM images of the fabricated MDM chip. Figure 67(i) shows the measured summed mode crosstalk, ranging from to at 1540 nm for all mode channels. By transmitting 28-Gbaud 16-QAM signals, a single-wavelength data transmission rate of 813 Gbit/s is achieved, as shown in Figs. 67(j) and 67(k)[652]. Moreover, as the number of modes controlled by dielectric metamaterial modules increases, the net data rate can be further enhanced. 7) 3D/2D integrated photonic chips enabling multi-dimensional data transmission and processing Apart from data transmission, data processing is also one of the key technologies of optical communications. Recent studies have shown that integrated photonic devices have great potential in data processing and have also achieved sufficient development[728]. Among them, flexible mode/polarization switching is crucial for the advancement of optical communications. Figure 68(a) shows the experimental setup of FMF-chip-FMF communication incorporating mode and polarization switching functions. In the experiment, the 2D silicon photonic switch and the 3D silica mode (de)multiplexer are employed, as shown in Figs. 68(g)–68(i). The 2D optical switches, topologically optimized to have fewer switch elements and crossings, result in lower loss and crosstalk. The 3D silica mode (de)multiplexer effectively multiplexes and demultiplexes LP modes in fibers, providing a robust connection between FMFs and 2D silicon switches. Figure 68(b) shows the measured BER performance of mode/polarization switching using 30-Gbaud QPSK signals[511]. Due to its broadband operation, the system supports 16 channels of 200G-spacing WDM, achieving a total system capacity of 5.38 Tbit/s. In addition to optical switching, reconfigurable optical add/drop multiplexing (ROADM) is also of great importance in flexible multi-dimensional optical communications. Using 3D/2D integrated photonic chips, a hybrid multi-dimensional fiber-chip system is developed, as shown in Fig. 68(c), which enables add/drop operations on multi-dimensional signals in the FMF. This add/drop functionality is realized through microring resonators, as shown in Fig. 68(j). Figure 68(d) displays the measured BER performance for multi-dimensional add/drop, showing that the BER of the selected channels is below (7% FEC threshold)[512]. With 16-Gbaud QPSK signals, this multi-dimensional optical processing system achieves a system capacity exceeding 3 Tbit/s. Furthermore, the 3D silica mode (de)multiplexer can be also combined with another kind of ROADM silicon chip to implement multi-dimensional (mode, polarization, wavelength) FMF-chip-FMF data transmission and processing, as shown in Fig. 68(e). The ROADM silicon chip, composed of arrayed waveguide gratings (AWGs) and MZIs, also enables wavelength add/drop operations, as shown in Fig. 68(k). Notably, compared to microring resonators, the AWGs have a relatively larger 3-dB bandwidth, allowing for higher communication rates. Figure 68(f) shows the measured system performance for add/drop in the multi-dimensional FMF-chip-FMF system, where each channel carries a 21-Gbaud QPSK signal[513]. Consequently, this FMF-chip-FMF system achieves a throughput of 4.032 Tbit/s. 8) Silicon photonic processor empowering free-space and fiber-optic multimode communications For high-dimensional optical communication systems exploiting spatial modes (structured light), it is generally necessary for the transmitter to emit different orthogonal modes into the transmission medium and for the receiver to perform multi-input multi-output (MIMO) digital signal processing (DSP) to eliminate arbitrary mode scrambling introduced by mode coupling and transmission in the medium. In MDM communication systems, the major challenge is the significant power consumption and time delay required for electronic DSP to unscramble mixed optical signals. The silicon photonic processor consisting of MZI meshes is a promising candidate to address this challenge for free-space and fiber-optic multimode communications. Figure 69 presents a photonic processor scheme for free-space multimode communications. The schematic and micrograph of the fabricated silicon photonic processor are shown in Figs. 57(a) and 57(b), respectively. As shown in Fig. 69(a), two free-space beams (labeled as Mode 1 and Mode 2), simultaneously impinge on the 2D optical antenna array. Mode 1 and Mode 2 refer to the fundamental Hermite-Gaussian mode and a higher-order -like mode, respectively. The coupling structure between free-space beams and the guided modes of the single-mode optical waveguides array is a 2D array with optical gratings. There is an MZI neural network on the chip that can recognize and route different input modes, as shown in Fig. 57. By changing the weight configuration of the MZI neural network, different modes can be separated at different output ports. The photonic processor is configured to separate two free-space beams at the two output ports WG1 and WG2. For example, and a higher-order -like mode can be transmitted to WG1 and WG2, respectively. The bar chart in Fig. 69(b) shows the relative transmission of the two modes measured at the two output ports after the photonic processor configuration. The same isolation (over 30 dB) is obtained if the mode output ports exchange, i.e., Mode 2 is coupled to WG1 and Mode 1 is coupled to WG2. By reversing the propagation direction of the light, the far-field profiles of the beams generated by the 2D optical antenna array can be observed under different configurations of the photonic processor. Figure 69(c) shows all the possible radiation fields, Mode 1 to WG1, Mode 2 to WG2, Mode 2 to WG1, and Mode 1 to WG2. High-speed data transmission performance with the MZI neural-network-based photonic processor is evaluated in the experiments. Two intensity-modulated 10-Gbit/s OOK data streams are transmitted over spatially and directionally overlapping modes ( and -like). Meanwhile, and -like modes have the same carrier wavelength (1550 nm) and polarization state. Figure 69(d) shows eye diagrams of the received signal after the photonic processor separation for all configurations in Fig. 69(c). The BER curves for the received data channels encoded into the modes (Mode 1) and -like (Mode 2) are plotted in Fig. 69(e). All configurations are taken into consideration (Mode 1 to WG1, Mode 2 to WG2, Mode 2 to WG1, and Mode 1 to WG2), and there is no significant OSNR penalty because of the crosstalk mitigation between the separated modes. Remarkably, even in the case when two orthogonal spatially overlapping free-space beams pass through mixing obstacles or disturbances, the photonic processor retains the ability to separate them. We first consider the mapping of an input pair of orthogonal modes to an output pair of orthogonal modes. As shown in Fig. 69(f), a phase mask transforms the fundamental mode into a 45° rotated -like mode (Mode 3), while the higher-order -like mode ( rotated) is turned into a 45° rotated -like mode (Mode 4). Figure 69(g) shows the experiment results when the photonic processor is configured to extract Mode 3 at output port WG1 and Mode 4 at output port WG2. The intensity ratios of the derived modes on each output port show an isolation of more than 30 dB. The good mode separation ability of the photonic processor is verified in a data transmission link, with each mode carrying a 10-Gbit/s OOK signal. The received eye diagrams show negligible distortion or inter-symbol interference effects. Figure 69(h) depicts the BER curves versus OSNR of data channels separated by the photonic processor. Furthermore, we then show that the photonic processor can also separate beams even with a generalized mode conversion process (mode mixing). As shown in Fig. 69(i), the phase mask, which is responsible for mode mixing, is rotated at an arbitrary angle. If it is assumed that the phase mask does not introduce any relevant loss, it produces an arbitrary mixing of the incoming modes, resulting in the generation of two beams, namely, Beam A and Beam B, that are still orthogonal but no longer resemble any of the modes of the HG family. In order to ascertain the shape of Beam A and Beam B, it is possible to reverse the propagation direction by injecting the light at ports WG1 and WG2 and then observe the far field radiated by the 2D optical antenna array with the camera. The field profiles of Beam A and Beam B are shown in Fig. 69(j). As anticipated, they exhibit an arbitrary shape that does not match any of the free-space optical modes. Nevertheless, they are still orthogonal and can be separated with extremely low crosstalk, as confirmed by the BER measurements shown in Fig. 69(k)[526]. Overall, the silicon photonic processor has demonstrated strong separation capability for free-space light beams, which are beneficial to free-space multimode optical communications. Moreover, the silicon photonic processor can also be used in fiber-optic multimode communication systems, as shown in Fig. 70(a). The mode field profiles of the eigenmodes and LP modes in FMF are depicted in Fig. 70(b). The system receiver is a reconfigurable integrated photonic processor equipped with multimode gratings. This multimode grating serves as an interface between the FMF and the silicon-based chip, as depicted in Fig. 70(c). The photonic processor performs the inverse matrix transformation on the fiber, selectively launching and separating orthogonal fiber modes, and functioning as a fully optical MIMO unscrambler. Furthermore, a multimode communication system based on an integrated photonic processor and multimode gratings is constructed, as shown in Fig. 70(d). Figure 70(e) shows the photograph of the receiver, which includes few-mode optics, a fiber array, and the silicon photonic chip. Figure 70(f) shows the eye diagrams for multimode communications, clearly demonstrating that when the photonic processor is inactive, the eye diagram is closed, whereas it opens when the photonic processor is active. Figure 70(g) shows slightly increased BER with more modes on for multimode communications, implying a minor degradation in the performance of the communication system[729]. In addition to the robust receiver for separating different modes, a transmitter generating arbitrary high-purity orthogonal modes is also of great importance for multimode communication systems. As shown in Fig. 71(a), a schematic configuration of an arbitrary optical system exhibits input and output optical apertures in both transmitters and receivers. Their transmitters and receivers are identical, composed of optical interfaces with nine gratings and an MZI network. The silicon photonic chip is shown in Fig. 71(b). Figures 71(c) and 71(d) display the mode fields generated by the nine gratings, which can be used for multimode communications. Figure 71(e) shows the BER of an arbitrary multimode optical system[730]. When scattering media are introduced, initially open eye diagrams close, but reconfiguring the MZI network reopens them, showcasing good communication performance in the experiment. 5.2.2.Structured light application in OAM holography encryptionHolography is regarded as an important platform for various applications, such as imaging, microscopy, encryption, three-dimensional displays, etc. Very recently, structured light accessing the spatial structure degree of freedom of light waves has attracted increasing interest in holography-based applications. For example, the OAM of light with a helical phase front can be used as an information carrier for holography. Hence, optical data encryption with high security can be realized by multi-bit OAM modes holograms. As shown in Fig. 72(a), the 10-bit OAM multiplexed hologram is created by encoding the 10 digits (Arabic numerals 0–9) using 10 higher-order OAM modes with topological charges ranging from to 50. Figure 72(b) shows an OAM ciphertext generated by the 10-bit OAM multiplexed hologram. A large amount of 3D information can be reconstructed using 210 unique holograms based on different OAM modes. The OAM code chart in the experiment consists of the OAM modes utilized to reconstruct the 10 OAM-dependent digits, as shown in Fig. 72(c). Five alphabet letters can be encrypted using the 10-bit OAM ciphertext. A mode-selective aperture array, which permits the transmission of a Gaussian mode, is used to increase the signal-to-noise ratio (SNR) of the OAM-multiplexing holography. Two examples are used to demonstrate the operation of a high-security holographic encryption technique based on the OAM ciphertext. Users 1 and 2 translate the plaintext messages “PSERX” and “TUXYL” into the distinct number combinations “1619051824” and “2021242512,” respectively, as shown in Fig. 72(d)[361]. The aforementioned number combinations are represented by the bespoke OAM keys and , correspondingly, according to the OAM coding chart in Fig. 72(c). Therefore, incident OAM beams with a helical mode index after the customized OAM keys should be consecutively incident on the OAM ciphertext to rebuild OAM-dependent digits in order to decipher the messages transmitted by these users. Furthermore, extending OAM-holography-enabled optical encryption into the quantum realm is interesting. This work provides a high level of information security for data sharing in quantum communications. In addition to the OAM-multiplexing holography in Fig. 72, another metasurface OAM holography with strong OAM selectivity is demonstrated, as shown in Fig. 73(a). It is realized by meta-holograms consisting of GaN nanopillars with discrete spatial frequency distributions. The unique holographic pictures (alphabet letters A, B, C, and D) can be independently reconstructed from a single OAM-multiplexing meta-hologram using incident OAM beams with topological charges of , 1, 1, and 2, respectively. However, these OAM-beam-based holography techniques do not extend to the SAM of photon. The scheme shown in Fig. 73(b) further considers SAM multiplexing in the metasurface design process, which increases the tunable dimensionality of structured light holography. Two non-interleaved metasurfaces in Fig. 73(b) are capable of creating a number of SAM- and OAM-dependent holographic channels. In this work, the angular momentum holography is classified as either spin-orbital locked holography (SOLH) or spin-overlaid holography (SSH). The SOLH, which adds the SAM to the already-existing OAM-multiplexed holography, is the result of the superposition of the SAM and OAM states. The SAM and OAM may be arbitrarily coupled because of their mutual orthogonality, which results in exponentially improved multiplexing channels as compared to OAM-multiplexing holography. Only when an incident beam carrying particular SAM and OAM eigenstates (represented as , where and represent the values of SAM and OAM) illuminates on the meta-hologram, four target images of capital letters “L, R, X, Y” can be accurately reconstructed, where each pixel appears as a Gaussian spot. The SSH, also known as vectorial holography, is created by arbitrarily superimposing the two SAM eigenstates in each operation channel and is capable of spatially manipulating the polarization vectors. The top and side views of the SEM images of the fabricated sample are shown in Figs. 73(c) and 73(d), respectively. The eight unique SOL holographic pictures that are mathematically and experimentally recreated are displayed in Fig. 73(e)[554]. With incident light bearing the proper , the eight target pictures may be accurately recreated; otherwise, the meta-hologram will produce a number of false holographic images. 5.2.3.Structured light application in 3D imaging using multi-wavelength dots arrayStructured light projection is a potential technique for quick and non-contact 3D imaging. Figure 74 depicts the design and fabrication of a multi-wavelength dots array to increase 3D imaging speed and resolution. As shown in Fig. 74(a), the dielectric metasurface can generate a dense dots array at three wavelengths at the same time. Moreover, in order to increase the transmittance efficiency, the cells on the metasurface are made up of titanium dioxide nanofins on a silica substrate with near-zero absorption. The PB phase () of each cell is only determined by the nanofin’s rotation angle (), i.e., , as depicted in Fig. 74(b). Based on the phase distribution, the Rayleigh-Sommerfeld diffraction integral formula is used to simulate the diffraction light field at three wavelengths, as shown in Fig. 74(c). Figure 74(d) shows the SEM image of the metasurface. The measurement sample with grooves of varying widths is made using 3D printing, as shown in Fig. 74(e). The grooves’ width is lowered from 2.4 to 0.8 mm in 0.2-mm steps. Three areas of the surface are measured using multi-wavelength metasurface. Figure 74(f) shows the measurement results of the three single-wavelength metasurfaces and multi-wavelength metasurface[731]. The lateral resolution exhibits noticeable improvements with the multi-wavelength method. 5.2.4.Structured light application in medical imaging using needle beam and multifocal beamIn addition, structured light also plays an important role in medical imaging and pathology detection. Wavelength-dependent optical absorption at the cellular level can be observed using optical-resolution photoacoustic microscopy. Due to tight focus of the optical excitation beam, this approach has a restricted depth of field, making it difficult to get high-resolution photographs of objects with uneven surfaces or high-quality volumetric images without -scanning. The scheme shown in Fig. 75(a) proposes the needle-shaped beam photoacoustic microscopy (NB-PAM) to overcome this limitation. This technique uses specialized diffractive optical elements (DOEs) to increase the depth of field to around a 28-fold Rayleigh length. The needle beam (NB) produced by these DOEs has a well-maintained beam diameter, a uniform axial intensity distribution, and few sidelobes. The experimental setup is shown in Fig. 75(b). The phase-modulated beam can be converted to an NB around the original focal plane of the Gaussian beam using a bespoke objective lens composed of an achromatic doublet and a corrective lens. Meanwhile, using DOEs can obtain an efficiency of up to 30% of the input beam energy. The NB may be tuned to have insignificant sidelobes in comparison to a Bessel beam with a sidelobe-to-main-lobe ratio of up to 20%, eliminating the requirement for intricate image processing in NB-PAM. The simulation results shown in Fig. 75(c) indicate that, in comparison to optical-resolution Gaussian beam photoacoustic microscopy (GB-PAM), NB-PAM offers superior 2D pictures of uneven surfaces or volumetric imaging of thick specimens[637]. Similarly, conventional optical coherence tomography (OCT) technology requires large numerical aperture (NA) lenses in order to achieve a tight focus and narrow depth of field, which requires dynamic focusing and limits the imaging speed. Figure 75(d) designs a metasurface platform to generate multiple axial foci, which increases the speed of volumetric OCT imaging by providing several focal planes. This metasurface provides precise and adaptable control over the quantity, locations, and intensities of produced axial foci. The simulated multifocal beams are shown in Fig. 75(e). The incoming light energy in Fig. 75(e) is distributed evenly across the foci, and the focal intensity is inversely correlated to the square of the number of foci. Four multifocal metasurfaces that match the four multifocal beams in Fig. 75(d) are created and evaluated using the OCT technology. The -stacked volumetric images of the normal human skin sample are shown in Fig. 75(f)[732]. 5.2.5.Structured light application in photo-induced force microscopy using tightly focused azimuthally polarized beamRecently, there has been considerable interest in materials interacting with the magnetic field of light rather than its electric field. Manipulation of magnetic transitions in nanostructures at optical frequencies would provide significant increase in memory capacity and read/write speed. Nonetheless, the lack of symmetry between electric and magnetic effects in nature means that magnetism in materials at optical frequencies is inherently much lower than the electric response. As a result, the development of effective optomagnetic devices is a critical task. The first step is to create a mechanism for directly probing the optomagnetic field at the nanoscale. A method for directly obtaining the near-magnetic field distribution in the probe region is proposed using a structured light field. A tightly focused azimuthally polarized beam (APB) excites the magnetic resonance of a unique micro-nano probe while suppressing its electric response, allowing for the detection of pure magneto-optical forces. As shown in Fig. 76, a structured light field with a tangentially polarized distribution is used as an excitation source. The electric field near the optical axis of this source is suppressed to almost zero. Based on the electromagnetic induction principle, the longitudinal magnetic field in the vicinity of the optical axis has an extremely large value. Therefore, tangentially polarized light has only a longitudinal magnetic field near the optical axis and without an electric field, making it an ideal magnetic light source. The incident APB is tightly focused by the bottom oil-immersion objective lens onto the Si truncated cone (i.e., the magnetic nanoprobe) via a glass coverslip, as shown in Fig. 76(a). The incident APB stimulates the magnetic mode in the Si probe, resulting in a magnetic dipole. When the nanoprobe approaches the glass slip (or substrate) within a few nanometers, the interaction between the nanoprobe and the substrate under APB illumination is represented by two interacting magnetic dipoles: the induced magnetic dipole in the nanoprobe and its image, which accounts for the substrate effect, as shown in Fig. 76(b). Figure 76(c) illustrates a system of two closely spaced magnetic dipoles that produces a magnetic dipole-dipole interaction force. The importance of azimuthal symmetry of the probe is illustrated in Figs. 76(d)-76(k)[733]. Figures 76(g) and 76(k) show that a typical sharp probe does not mismeasure a photo-induced force contrast. The symmetry of the force maps in Figs. 76(h) and 76(i) taken with the needle tip in Figs. 76(d) and 76(e) agrees well with the prediction of the APB magnetic field. In the case of the probe in Fig. 76(f), a doughnut-shaped force map in Fig. 76(j) is obtained. This is typical of an electric dipole–dipole map and does not measure a significant magnetic component. Using the truncated probe, a signal-to-noise ratio of about 15 is obtained. It is worth noting that the resolution of these images is low, likely equal to the size of the nanoprobe apex at 200 nm. However, the system could be improved by designing smaller probes and using rare earth metals that have exclusive magnetic responses to any optical beam rather than APB. 5.2.6.Structured light application in three-dimensional topography using vortex beamSince its discovery, optical interference technology has been employed extensively in the field of precise optical metrology, particularly in the context of three-dimensional sensing and phase information characterization. Recently, optical vortex interferometers have been proposed as a novel approach that employs vortex beams as reference beams to interrogate a 3D object. We propose a high-mode-purity self-referenced optical vortex interferometer based on a broadband geometric phase element array. To measure the surface profile of a transparent sample, the schematic of the self-reference spiral interferometer is built, as shown in Fig. 77(a). A multi-tasking geometric phase element array is employed to form the vortex phase filter. The geometric phase is derived from the polarization conversion and in-plane rotation of the anisotropic unit. By regulating the distribution of the optical axis of the anisotropic unit, it is possible to control the phase of the transmission light with high efficiency and broadband coverage. The vortex contrast filter is capable of realizing two tasks simultaneously. Firstly, it functions as a vortex filter, comprising a central hole and a vortex phase. Secondly, it deflects the modulated light, thereby achieving high mode purity in broadband. As a result, the low- and high-spatial-frequency components overlap in the image, creating an interference pattern. Figures 77(b) and 77(g) show the experimental results for the spiral interference image. Figure 77(b) shows an interference pattern with a double counterclockwise spiral, indicating an elevation in the object. In contrast, the interference pattern in Fig. 77(g) is a clockwise spiral pattern, indicating a depression in the object. To improve the representativeness of the results, Figs. 77(c) and 77(h) show the interference pattern with plane waves as a reference at the same position. This pattern features a closed ring contour. Once the spiral interference pattern is obtained, the surface topography of the object can be reconstructed, and the results are shown in Figs. 77(d) and 77(i). Meanwhile, the height of the surface is reconstructed from the sampling points using the phase recovery algorithm, as shown in Figs. 77(e) and 77(j), respectively. Finally, the recovered 3D height profile of the glass substrate is plotted in Figs. 77(f) and 77(k), respectively[734]. 5.2.7.Structured light application in 3D optical manipulation using 2D Airy beamStructured light has also been widely used in 3D optical manipulation. Traditionally, bulky optical devices are always employed to enable optical manipulation. Very recently, there has been an increasing interest in developing integrated structured light generation approaches for optical manipulation in a compact manner. Figure 78 proposes a method of using the metasurface to generate a 2D Airy beam, which enables the 3D manipulation of particles. The polarization-independent vertically accelerated 2D Airy beam is produced in the visible range using a cubic-phase dielectric metasurface made of GaN circular nanopillars, as shown in Fig. 78(a). To validate the propagation characteristics of the Airy beam generated by the cubic phase, the intensity distribution along the optical axis at different propagation distances is simulated with the results shown in Figs. 78(b) and 78(c). Figure 78(d) shows the measured intensity distribution of the vertically accelerated 2D Airy beam at three different depths inside the water[735]. It is found that the microspheres are trapped in the cross-section optical field. In addition, the experimental evidence is shown for the unique propagation properties of the 2D Airy beam that is vertically accelerated, including non-diffraction, self-acceleration, and self-healing. 5.2.8.Structured light application in optical trapping using waveguide mode and phased arrayRemarkably, the special on-chip waveguide mode allows the construction of a strong optical potential well. Therefore, particles can be captured and manipulated on a chip-scale device. The scheme shown in Fig. 79(a) uses the optical forces induced by the Bloch mode propagating along a silicon SWG waveguide to trap the particles. The Bloch mode in the silicon waveguide is simulated in Fig. 79(b)[736]. As a periodical structure, an SWG waveguide supports periodical light field distribution along the waveguide. This makes it possible to trap many nano-particles stably. The separation of the trapped nano-particles can be designed easily by modifying the grating period of an SWG waveguide. Similarly, 1D optical lattices produced by near-field mode along a few-mode silicon nanophotonic waveguide are used to illustrate the optical trapping of dielectric particles and bacteria on a chip. Figure 79(c) shows different waveguide modes in the silicon waveguide. The coupling method is illustrated in Fig. 79(d). A polarization-maintaining optical fiber with a lensed tip is used to couple the laser beam into the photonic waveguides. Three pair combinations can be obtained from these guided modes, leading to three distinct near-field mode distributions, as shown in Fig. 79(e)[737]. However, this type of capture is achieved using the near-field distribution of mixed modes on the waveguide surface to trap the particles and cannot be dynamically adjusted. In addition, another particle capture scheme is proposed based on integrated optical phased arrays (OPAs), as illustrated in Fig. 79(f). For dynamic capture and manipulation of particles on the integrated platform, the scheme enables the focusing and manipulation of the far-field radiation direction by adjusting the phase shift of the waveguide array, as shown in Fig. 79(g), which in turn achieves particle capture. The measurement of optical intensity distribution radiation from the OPAs is shown in Fig. 79(h). The right insets of Fig. 79(h) show the optical intensity distribution at the cross-section with different propagation distances. The motion of the microsphere located at the focal spot is significantly reduced compared to its neighbors, indicating successful trapping, as shown in Fig. 79(i)[738]. 5.2.9.Structured light application in chiral trapping using silicon-based slot waveguideAdditionally, Fig. 80(a) proposes an optical chiral separation scheme that enables the capture and separation of chiral nanoparticles in oppositely transmitted silicon-based slot waveguides. This silicon-based chiral separation platform has several advantages over the previously proposed optical chiral separation platform. On one hand, the gradient force generated by the slot waveguide mode is utilized for chiral separation, as illustrated in Figs. 80(b) and 80(c). This provides a more significant optical force as a conservative force compared with the non-conservative light scattering force or radiation force. On the other hand, the optical field strongly bounded by the subwavelength slot waveguide is capable of trapping and separating nanometer-sized chiral particles, which can effectively circumvent the diffraction limit imposed by free-space light focusing. The resulting gradient forces can shift the equilibrium positions of the optical trap, which manifests as a chiral-dependent separation, as shown in Figs. 80(d)–80(f)[739]. More importantly, the fabrication process of this silicon-based dielectric slit waveguide is fully compatible with the current mature silicon-based photonic integration technology and CMOS process, which facilitates the subsequent development of optical flow control technology to realize large-scale chiral separation. 5.2.10.Structured light application in optical tweezersAs a key technology to study the motion of objects and their interactions at the micro and nanoscale, optical tweezer technology has important applications and is widely used in the fields of physics, chemistry, micromechanics, and biomolecular interactions because of its advantages of non-contact, non-damage, and high accuracy. The manipulation of an object by light relies on the momentum transfer between light and the object. The transfer of linear momentum enables the capture and translation of the object, while the transfer of angular momentum leads to the rotation of the object. In addition, focused beam arrays can capture multiple particles in space at the same time. Figure 81(a) shows the scheme of using the DOE device for particle array capturing. Figure 81(b) shows successive tiling of a hologram that generates different square arrays of optical tweezers[740]. Using this method, triangular and square tweezer arrays can be created to trap up to multiple particles. 5.2.11.Structured light application in nanowire trapping and rotationThe interaction between the photon OAM and the particles can also be used to capture particles and even realize rotational manipulation. The positive or negative OAM affects the direction of rotation of the particle, and its magnitude is related to the rotation speed of the particle. The exploration of manipulating one or more silicon nanowires can be done using the LG beams. Figure 82 shows that the silicon nanowires can orbit around the optical vortex that is parallel to the beam propagation axis. Figure 82(a) shows the SEM image of the silicon nanowires. Figure 82(b) shows the position of the particles as a function of time, evincing a clear orbital motion. The right inset shows a schematic representation of the trapped nanowire overlapped with the intensity pattern of the trapping beam. Additionally, Fig. 82(c) shows how the transfer of SAM may be utilized to create light-driven nanomotors. It is noted that while the size of the particle is comparable to the waist size of the incident structured beam, the particle is predicted to be trapped at the center of the beam with the appropriate choice of beam parameters such as vortex charge and polarization, as shown in Fig. 82(d)[741]. 5.2.12.Structured light application in Doppler cloak by spinning OAM metasurfaceTheoretically, it has been shown that spatiotemporally modulated metamaterial can be utilized to build a Doppler cloak, which hides the motion of moving objects from the observer by correcting for the Doppler shift. The OAM of light dispersed by a spinning object can be used to detect the rotational Doppler effect, which is the angular equivalent to the linear Doppler effect. The spatiotemporal property can be achieved by mechanical modulation of the reflection phase of the metamaterial to generate the rotational Doppler effect. The physical spinning of a polarization-independent OAM metasurface as a 2D spatiotemporally modulated metamaterial can be employed to exploit the space and time dimensions in order to produce the rotational Doppler effect. This approach offers a straightforward method for modifying the reflection phase, which can be utilized to implement the Doppler cloak. Figure 83(a) shows the experimental setup of the Doppler cloak by a spinning OAM metasurface. In order to demonstrate the Doppler cloak concept, this work measures the spectrogram of an OAM metasurface by a 5.8-GHz Doppler radar in three cases: moving, spinning, and both moving and spinning, as shown in Fig. 83(b)[742]. The measurements confirm that the spatiotemporal metasurface has the ability to cloak a moving object by generating an opposite rotational Doppler shift to the one from linear movement. Remarkably, it is expected that such Doppler cloak can be also realized by a spinning OAM metasurface when extending the radio frequency to the light wave band, i.e., structured light Doppler cloak. 5.2.13.Structured light application in quantum opticsMetamaterials built from deep subwavelength units have shown significant values in some specific application scenes, including cloaking, emulation of general relativity, super-resolution imaging, negative refractive index, and near-zero refractive index. Moreover, metamaterials, also the simplified structures of metasurfaces, have recently been proposed as a promising quantum optics platform. Figure 84(a) demonstrates the creation of entanglement between the spin and orbital angular momentum of photons using a dielectric metasurface. The SEM image of the metasurface is shown in Fig. 84(b). A single photon in the state is produced using the experimental setup shown in Fig. 84(c)[743]. A single photon changes to an entangled state in the interaction with the metasurface. Moreover, the metasurface is capable of generating entangled biphoton states. The experimental conversion efficiency of the metasurface is about 72%. The demonstration of generating entangled photon states with metamaterials/metasurfaces paves the way for nanophotonic quantum information applications. It is anticipated that metasurfaces may become a powerful tool in future quantum optics and could be used extensively in photonic quantum information systems. The orthogonal transverse mode represents a significant degree of freedom in photonic integrated circuits. It offers a potential technique to enhance communication capability, both in classical and quantum information processing. A transverse-mode-encoded controlled-NOT (CNOT) gate is required to build large-scale on-chip multimode multi-degrees of freedom quantum systems. Figures 84(d)–84(g) show the multimode implementation of a 2-qubit CNOT gate with transverse mode encoding, which consists of a developed multimode directional coupler and attenuator[744]. The two lowest-order transverse modes of TE polarization, i.e., and , are supported by the designed multimode waveguide, which has a cross-section of , as shown in Fig. 84(d). These modes are orthogonal and do not interact with each other, allowing them to be employed directly for the encoding of quantum information. There are two created mode-dependent photonic devices, transverse-mode-dependent directional coupler (TMDDC) and multimode attenuator (MMA), which can produce mode-dependent coupling loss, as illustrated in Figs. 84(e) and 84(f), respectively. These devices can be used to create a transverse mode-encoded CNOT gate. A multimode 2-qubit CNOT gate is created by cascading one TMDDC and two MMAs, as shown in Fig. 84(g). In both classical and quantum regimes, high-capacity optical information processing calls for on-chip photon sources carrying OAM. However, the classical regime has been the main focus of currently utilized integrated OAM sources. The on-chip integrated OAM source shown in Fig. 85 emits well-collimated single photons with a purity of at room temperature. These single photons carry entangled spin and OAM states and form two spatially separated entangled radiation channels with various polarization characteristics. The OAM-encoded single photons are generated by an efficient outcoupling of diverging surface plasmon polaritons, which are excited by a deterministically positioned quantum emitter via Archimedean spiral gratings, as shown in Fig. 85(a). The introduced spiral gratings produce a spatial separation between the RCP and LCP components, with an intensity overlap of only 23% in the considered one-arm spiral grating, as shown in Fig. 85(b). Figure 85(c) displays the schematic of the spiral gratings and their corresponding intensity fields of the far-field LCP and RCP states. The analytical phase distributions, indicating -fold phase variation along the azimuthal direction, are shown in Fig. 85(d). The SEM image of the spiral nanorings with different structure parameters is shown in Fig. 85(e)[745]. This compact platform is also suitable for the development of OAM and time-bin hyperentangled photon pairs through the use of semiconductor quantum dots, opening up new avenues for the creation of high-dimensional, large-scale, and integrated photonic quantum systems. As a short summary, Table 4 lists the main representative research works on state-of-the-art integrated structured light applications[296,361,470,511–513,526,554,637,652,724–727,729–746]. The key parameters such as type of structured light, application scenario, material, and working wavelength are summarized for comparison. Table 4Parameters and Performance of State-of-the-Art Integrated Structured Light Applications
6.Conclusions and Perspectives6.1.Progress and Trend of Integrated Structured Light ManipulationAlthough structured light manipulation is an ancient topic, dating back to the laser designs[747,748], research on integrated structured light manipulation has exploded in the past few decades[1–5] thanks to the rapid development of photonic integration platforms, providing exciting opportunities to many advanced fields such as optical communications, holography, imaging, microscopy, manipulation, trapping, tweezers, metrology, and quantum information processing[3–5,11,14,55,56,60–69]. In a broad sense, structured light is a general concept of shaping light that spans from classical light to quantum light, from fundamental science to practical applications, and has formed an important branch and key part of modern optics. In a narrow sense, structured light has a unique and inhomogeneous amplitude, phase, and polarization distribution in space and time, and has many interesting properties, such as phase and polarization singularities and extraordinary dynamic characteristics[11,12,14,57–59,749,750]. Thus, structured light offers a variety of new possibilities, leading to remarkable technical progress. Moreover, integrated structured light manipulation can avoid the limitations of high cost and large volume of traditional bulky optical devices, and provide a more compact solution for the further development of structured light. In this review article, we have presented a comprehensive overview of various integrated structured light manipulation techniques (generation, processing, detection) and their diverse applications on different photonic integration platforms. Figure 86 summarizes the recent progress and future trend of integrated structured light manipulation, mainly in the following aspects of multiple materials, multiple working bands, multiple structured light types, multiple processing functions, multiple detection structures, and diverse application scenarios. 6.1.1.Multiple materialsThe photonic integration platforms on silicon[422–428,751–757], [758–761], Ge[762–764], [429,765–772], and III–V materials[773,774] are used to manufacture photonic chips based on photolithography and etching processes. Among them, silicon is particularly promising for integrated structured light manipulation due to its high-contrast index, small footprint, and CMOS compatibility for large-scale integration and mass manufacturing[422,775,776], plays an important role in ultra-low optical loss waveguides[777,778] and chip-scale nonlinear photonic devices[779,780], Ge is applied to high-speed photodetectors[763], crystal and thin film with excellent electro-optic effect are widely used in electro-optic modulators[442,776], and III–V (e.g., InGaAsP, InP) is one of the most popular materials for fabricating integrated laser sources[773] as well as modulators[781,782] and photodetectors[783,784]. Remarkably, as an important supplement, femtosecond laser direct writing techniques on silica and polymer material platforms are adopted to fabricate 3D photonic chips, which provides enhanced flexibility of on-chip structured light manipulation[385,386,450,470,510,785–789]. In addition, metal-assisted photonic devices can excite the surface plasmon polariton mode, which is advantageous for improving the integration density and reducing the footprint of photonic devices[453,457]. Photonic chips based on phase change materials (e.g., )[790–796] with the non-volatility property facilitate reconfigurable structured light manipulation with low power or near-zero power consumption. Low-dimensional materials (e.g., graphene, , )[507,508,797,798] are attractive for nonlinear structured light processing. Very recently, more and more emerging materials have been used in advanced optoelectronic devices, including Weyl semimetal (e.g., ) for structured light detection[717], perovskite[659,799,800] for lasing and light-emitting diodes (LEDs), as well as various piezoelectric and ferroelectric thin films for tunable devices and electro-optic modulators, such as lead zirconated titanite (PZT)[801,802], aluminum nitride (AlN)[803,804], hafnium dioxide ()[805,806], silicon carbide (SiC)[807–809], barium titanite (BTO)[810–812], lithium tantalate (LT)[813], etc. One expected trend is that these emerging materials with unique material properties would be also utilized for diverse integrated structured light manipulation. Meanwhile, hybrid integration platforms for multi-material systems[772,774,781–783] (e.g., III–V+silicon, , etc.) would be another important trend of integrated structured light manipulation. 6.1.2.Multiple integration techniquesIntegrated structured light manipulation aims to generate, process, and detect structured light with compact devices for a variety of applications. Various integration techniques are of great importance to achieve this goal. In general, typical integration techniques include monolithic integration, heterogeneous integration, hybrid integration, chiplet integration, 3D integration, and integration of photonics and electronics, with some overlap and similarity among these techniques[468,814–817]. Monolithic integration, usually on a single material platform, is applied not only to a single functional device but also to multiple devices. Heterogeneous integration, usually for different materials, is implemented either by die/wafer bonding and micro-transfer printing, or blanket heteroepitaxy and selective heteroepitaxy, with increased fabrication technique complexity. The wafer-level heterogeneous integration is also used for not only single functional devices but also multiple devices. At the same time, the chip with multiple devices also puts forward higher requirements for the device yield. If a single functional device in the chip fails, there is a high possibility that the entire chip is not working properly, which increases the cost and risk of failure. Hybrid integration and chiplet integration are proposed to realize the combined use of chips with different functions[818–823]. These two integration techniques only require selecting qualified functional chips, without considering the failure of individual functional components. In addition, hybrid integration and chiplet integration can also give full play to the advantages of different chip platforms. In particular, the chiplet technology, combined with the advanced packaging technologies, does not rely on the advanced semiconductor manufacturing process and can maximize the chip performance, improve the chip yield, save the chip cost, and shorten the chip manufacturing cycle. 3D integration, such as femtosecond laser direct writing[450,468,470,824,825], 3D stacking[826,827], and direct bonding/flip-chip/co-packaged optics (CPO)[814,828,829], is also another important high-density integration technique. Furthermore, the integration of photonics and electronics on a single chip (e.g., CMOS+photonics, BiCMOS+photonics) is also a trend but with a high requirement on the foundries[426,428,830–833]. The introduced integrated structured light manipulation works in this review mainly use the monolithic integration, hybrid integration, and femtosecond-laser-direct-writing-based 3D integration techniques. It is predicted that, with the continuous progress and evolution of various integration techniques, all the above mentioned integration techniques might be employed in integrated structured light manipulation, providing increased density, reduced cost, improved performance, enhanced functionality, and extended applications. 6.1.3.Multiple working bandsThe working wavelengths of the presented works on integrated structured light manipulation in this review are mainly in the visible (e.g., red, green, and blue) and near-infrared ranges (e.g., 1550 nm in the telecommunication band). Remarkably, in the telecommunication band, in addition to the C-band (1530–1565 nm) structured light manipulation, O-band (1260–1360 nm), E-band (1360–1460 nm), S-band (1565–1625 nm), and U-band (1625–1675 nm) are also of great interest. Integrated structured light manipulation in the full telecommunication band (O + E + S + C + L + U) over 415 nm is an important trend, but also full of great challenges[572,834–838]. In addition, similar to visible and near-infrared bands, other electromagnetic frequency bands[70,839–848] (e.g., radio, microwave, millimeter wave, terahertz, mid-infrared, ultraviolet, extreme ultraviolet, X-ray, and gamma ray) and even electron beams[598,849–853] and neutron beams[854,855] can be structured. In this scenario, integrated structured electromagnetic wave manipulation would be of great significance. The theories, principles, and techniques of integrated structured light generation, processing, detection, and applications might be borrowed and applied to a wider range of electromagnetic frequency bands. 6.1.4.Multiple structured light typesVarious integrated generation schemes of multiple types of structured light beams are introduced in this review, such as OAM-carrying light beams[464–467,469–475,481,550–553,555,556,578–612,615–617], chiral light[622], LP modes[623], LG beams[624,625], HG beams[624], Bessel beams[385,496,627–630], Mathieu beams[386], Airy beams[631–636], needle beams[637], pin beams[74], vector beams[638–643], array beams[644–646], optical vortex lattice[647], spatiotemporal beams[648–651], knotted and linked beams[255,256], in-plane waveguide modes[497,652–656], and reconfigurable structured light[657–659]. For the passive integrated structured light generation, the in-plane to in-plane, out-of-plane to in-plane, in-plane to out-of-plane, and out-of-plane to out-of-plane generation of OAM-carrying light beams are focused. For the active integrated structured light generation, OAM beam lasers and cylindrical vector beam lasers are focused. The passive and active generation of the above-mentioned structured light beams beyond OAM is still limited. Moreover, some other structured light beams shown in Fig. 1, such as Ince-Gaussian beams[199–209], bottle beams[217–225], skyrmion beams[305–317], hopfion beams[117,318–324], and more general structured light beams with arbitrary spatial amplitude, spatial phase, spatial polarization, and temporal distributions, have been demonstrated using traditional discrete and bulky optical devices, but have not yet been implemented on photonic integration platforms. It is believed that the integrated generation of these more general structured light beams will be the future trend to meet the requirement of high compactness and small footprint. Additionally, the quality and efficiency of integrated structured light generation are also important factors to be further improved for practical applications. The presented works on integrated structured light generation are far inferior to traditional methods (e.g., SLM-based structured light generation). The mechanisms, structure designs, and fabrication techniques of integrated structured light generation are expected to be further explored and optimized. 6.1.5.Multiple processing functionsVarious integrated structured light processing functions are presented in this review, such as out-of-plane and in-plane OAM modes (de)multiplexing[470,501–503], OAM modes exchange[470], in-plane waveguide modes multiplexing[498–500,670], OAM mode transformation (integer/fraction OAM multiplication and division[504,671], nonlinear OAM conversion[505–508,674,675]), LP mode switching[511], array beam transformation[509,677], in-plane waveguide mode processing (mode exchange[514,678], mode switch[679,680], mode add/drop[681], chiral mode switching[669,682,683]), mapping of OAM mode and LP mode[510], and transformation of LP mode and in-plane waveguide mode[684–686]. These processing functions are mainly divided into two categories, i.e., multiplexing and transformation. The multiplexing and demultiplexing of structured light beams are essential techniques of SDM-based optical communications, capable of effectively increasing the transmission capacity. The transformation functions (switching, exchange, add/drop, multiplication, division) of structured light beams are of great importance to enhance the flexibility of structured light manipulation. Despite the implementation of many multiplexing and transformation functions, developing more advanced integrated structured light processing functions is one future trend. Moreover, the reported integrated structured light processing functions are mainly for OAM-carrying light beams, LP modes, and in-plane waveguide modes. Another trend of integrated structured light processing would be to develop grooming processing functions for all the above mentioned structured light beams, especially for those that have not yet been applied in the structured light processing. 6.1.6.Multiple detection structuresVarious integrated structured light detection schemes are described in this review, such as metal micro-nano structures[518,708–711], plasmonic metasurfaces[712,713], dielectric micro-nano structures[519,520,714,715], photocurrent detector with U-shaped electrodes based on the orbital photogalvanic effect[523,717], thermoelectric detector with a spin-Hall coupler[524], DMD and diffuser[718], hybrid optoelectronic neural network[525], silicon MZI network[526], inverse design subwavelength structure[527], and silicon nanorod optomechanics[720]. These already demonstrated detection structures show favorable performance, but mainly for OAM-carrying light beams, very little for transverse OAM beams, LP modes, and HG beams, and none for other more general structured light beams. Hence, one important trend of integrated structured light detection would be to develop appropriate detection schemes applicable to more structured light beams, which requires exploring new theories, mechanisms, and device structures for more general integrated structured light detection. In addition, the reported works of integrated structured light detection mainly focus on whether the detection function is achieved, while there is little attention paid to the performance metrics of the detector. Thus, another trend of integrated structured light detection would be to focus more on the performance of detectors and improve the detection efficiency, detection sensitivity, and detection range. 6.1.7.Diverse application scenariosVarious integrated structured light applications are discussed in this review, such as optical communications and optical interconnects[296,470,511–513,526,652,724–727,729,730], OAM holography encryption[361,554], 3D imaging using multi-wavelength dots array[731], medical imaging using needle beams and multifocal beams[637,732], photo-induced force microscopy using tightly focused azimuthally polarized beams[733], three-dimensional topography using vortex beams[734], 3D optical manipulation using 2D Airy beams[735], optical trapping using waveguide modes and optical phased arrays[736–738], chiral trapping using silicon-based slot waveguides[739], optical tweezers[740], nanowires trapping and rotation[741], Doppler cloak by spinning OAM metasurfaces[742], and quantum optics[743–745]. In particular, for integrated structured light communications, diverse communication schemes are introduced, including analog signal transmission using OAM modes[724], data-carrying digital signal transmission using OAM modes[725], silicon-chip-assisted high-speed spatial light modulation communication[726], femtosecond laser inscription enabling chip-chip and chip-fiber-chip optical interconnects[470], integrated optical vortex emitters enabling direct fiber vector eigenmode multiplexing transmission[296], data-carrying on-chip MDM transmission using in-plane waveguide modes[652,727], 3D/2D integrated photonic chips enabling multi-dimensional data transmission and processing[511–513], and silicon photonic processors empowering free-space and fiber-optic multimode communications[526,729,730]. Remarkably, the involved structured light beams in these applications mainly include OAM-carrying light beams, LP modes, needle beams, multifocal beams, Airy beams, array beams, in-plane waveguide modes, and vector beams. One future trend of integrated structured light applications is to find more applications using more general structured light beams. Meanwhile, improving the performance of diverse integrated structured light applications is another trend of great importance for the promotion of practical applications. This requires more systematic and in-depth optimization of integrated photonic devices, and different specific application scenarios may have different requirements. 6.2.Opportunities, Challenges, and Solutions for Integrated Structured Light ManipulationBenefiting from the advanced light field manipulation technologies and mature semiconductor manufacturing processes, integrated structured light manipulation has been flourishing in recent years. Throughout the existing technologies and future developments, integrated structured light manipulation has many opportunities to be further explored in terms of generation, processing, detection, and application. Meanwhile, more extensive and in-depth researches on integrated structured light manipulation also put forward considerable challenges for enhanced functionality and performance. Possible solutions to superior integrated structured light manipulation are highly desired. Figure 87 briefly discusses the opportunities, challenges, and possible solutions for integrated structured light manipulation (generation, processing, detection, application). For the integrated structured light generation, from the current status, most of the reported generation schemes lack tunability or reconfigurability, i.e., the generated structured light is fixed when the integrated photonic devices are fabricated. This greatly limits their flexibility in practical use. Inspired by the commercially available SLMs and DMDs, the opportunities are tunable and a reconfigurable integrated generation of arbitrary structured light. Meanwhile, most of the reported works focus on a single spatial degree of freedom, e.g., spatial amplitude or spatial phase or spatial polarization, and the challenges of powerful integrated structured light generation are full-dimensional spatial light meta-modulation, i.e., arbitrary, independent, and simultaneous manipulation of spatial amplitude, spatial phase, spatial polarization and even spatiotemporal structure. Some possible solutions are parallel-task metasurfaces (e.g., full-dimensional meta-modulation geometric-phase metasurfaces[856], reconfigurable metasurfaces based on phase change materials) and tunable photonic integrated devices (e.g., thermal-optic tuning and electro-optic tuning). For the integrated structured light processing, viewed from the current situation, most of the demonstrated approaches are for a single function or very few functions, i.e., the ability to handle various types of structured light beams is still relatively limited. Taking into account the increasing demand for enhanced processing capabilities, the opportunities are reconfigurable multiple functions. Meanwhile, most of the demonstrated works focus on very few numbers of low-order spatial modes, which limits the full exploration and utilization of the space degree of freedom of light. The challenges are integrated structured light processing of a large number of spatial modes, including high-order ones. Some possible solutions are large-scale photonic integrated devices (e.g., large-scale programmable silicon photonic integrated circuits, large-size multiplane light conversion, or a neural network) and multi-chiplet integration (multiple small chips or chiplets each for a specific processing function, connected to each other on an interposer assisted by the advanced packaging technology). For the integrated structured light detection, looking at the current status, most of the introduced schemes require precise center alignment when detecting the structured light beams. This increases the difficulty of structured light detection. Also, the detection speed is relatively low. For further improvement, the opportunities are non-aligned and high-speed integrated structured light detection. Meanwhile, most of the introduced works focus on limited types and numbers of structured light, which is also another limitation of structured light detection. The challenges are integrated structured light detection of more general structured light beams, including those with arbitrary spatially variant amplitude, phase, and polarization distributions, and even spatiotemporal light beams. Some possible solutions are macropixel metasurface devices (e.g., metasurface-assisted detectors for spatial amplitude, phase, and polarization) and high-speed array detectors (e.g., integrated array detectors for more general structured light, including spatiotemporal light beams). For the integrated structured light applications, in the present situation, most of the conducted experiments are about one specific application for a single integrated photonic device. Single-device-based diverse and intelligent applications are limited. Although many works have been demonstrated in the laboratory, there is still a long way to go for practical applications. Future opportunities are diversity, intelligence, and practicality. Meanwhile, most of the demonstrated works are in simple application scenarios, while practical applications may encounter many complex scenes and environments, even extreme conditions, e.g., optical communications in complex media (atmosphere, rain, cloud, frog, haze, smoke, dust, obstruction) with beam divergence, beam scintillation, beam wander, pointing error, and strong scattering. Hence, the challenges are robust integrated structured light applications in complex scenes and environments. A promising approach to address these challenges is to explore the unique intrinsic properties of specially structured light beams to resist the disturbance effects of complex media. Some possible solutions are optimized integrated devices for optimized structured light beams with anti-disturbance characteristics. 6.3.A Vision of Integrated Structured Light ManipulationWith future improvement, integrated structured light manipulation can be continuously promoted from basic theories to key technologies to important applications. Figure 88 draws a vision of integrated structured light manipulation. In order for the “integrated structured light manipulation” tree to grow more vigorously, it is necessary to strengthen the innovation of basic theories and mechanisms for structured light meeting structured matter. Diverse photonic integration platforms for different materials will continue to develop and mature, supporting high-performance optoelectronic devices and integration for chip-scale structured light manipulation. Remarkably, although this review of integrated structured light manipulation mainly focuses on the space degree of freedom of photons, other degrees of freedom, such as wavelength, polarization, time, and complex amplitude, can be also flexibly manipulated using photonic integrated devices. In addition to the generation, processing, detection, and application mainly discussed in this review, there are also many other aspects of integrated structured light manipulation to be further developed, such as display, transmission, computing, storage, etc. In this review, we discuss the generation, processing, detection, and application separately. In the future, the integration of generation, processing, detection, and application is highly expected to facilitate an ultra-compact integrated structured light manipulation system. Moreover, the space degree of freedom of photos is compatible with other degrees of freedom. In this scenario, the integration of space, wavelength, polarization, time, and complex amplitude is also highly desired to enable advanced multi-dimensional integrated structured light manipulation. AcknowledgmentsThis work was supported by the National Natural Science Foundation of China (Nos. 62125503 and 62261160388) and the Natural Science Foundation of Hubei Province of China (No. 2023AFA028). ReferencesJ. Wang,
“Metasurfaces enabling structured light manipulation: advances and perspectives [Invited],”
Chin. Opt. Lett., 16 050006 https://doi.org/10.3788/COL201816.050006 CJOEE3 1671-7694
(2018).
Google Scholar
J. Wang and Y. Liang,
“Generation and detection of structured light: a review,”
Front. Phys., 9 688284 https://doi.org/10.3389/fphy.2021.688284
(2021).
Google Scholar
H. Rubinsztein-Dunlop et al.,
“Roadmap on structured light,”
J. Opt., 19 013001 https://doi.org/10.1088/2040-8978/19/1/013001
(2017).
Google Scholar
O. V. Angelsky et al.,
“Structured light: ideas and concepts,”
Front. Phys., 8 114 https://doi.org/10.3389/fphy.2020.00114
(2020).
Google Scholar
A. Forbes, M. de Oliveira, and M. R. Dennis,
“Structured light,”
Nat. Photonics, 15 253 https://doi.org/10.1038/s41566-021-00780-4 NPAHBY 1749-4885
(2021).
Google Scholar
C. He, Y. J. Shen, and A. Forbes,
“Towards higher-dimensional structured light,”
Light Sci. Appl., 11 205 https://doi.org/10.1038/s41377-022-00897-3
(2022).
Google Scholar
W. T. Buono and A. Forbes,
“Nonlinear optics with structured light,”
Opto-Electron. Adv., 5 210174 https://doi.org/10.29026/oea.2022.210174
(2022).
Google Scholar
Y. Liang et al.,
“Reconfigurable structured light generation and its coupling to air–core fiber,”
Adv. Photonics Nexus, 2 036015 https://doi.org/10.1117/1.APN.2.3.036015
(2023).
Google Scholar
Z. Wan et al.,
“Lagging propagation phase of spatially structured beams,”
Opt. Express, 31 42667 https://doi.org/10.1364/OE.510341 OPEXFF 1094-4087
(2023).
Google Scholar
L. Allen et al.,
“Orbital angular momentum of light and the transformation of Laguerre-Gaussian laser modes,”
Phys. Rev. A, 45 8185 https://doi.org/10.1103/PhysRevA.45.8185
(1992).
Google Scholar
M. J. Padgett,
“Orbital angular momentum 25 years on,”
Opt. Express, 25 11265 https://doi.org/10.1364/OE.25.011265 OPEXFF 1094-4087
(2017).
Google Scholar
X. Wang et al.,
“Recent advances on optical vortex generation,”
Nanophotonics, 7 1533 https://doi.org/10.1515/nanoph-2018-0072
(2018).
Google Scholar
L. Zhu and J. Wang,
“A review of multiple optical vortices generation: methods and applications,”
Front. Optoelectron., 12 52 https://doi.org/10.1007/s12200-019-0910-9
(2019).
Google Scholar
Y. Shen et al.,
“Optical vortices 30 years on: OAM manipulation from topological charge to multiple singularities,”
Light Sci. Appl., 8 90 https://doi.org/10.1038/s41377-019-0194-2
(2019).
Google Scholar
H. Wang et al.,
“Vortex beam generation with variable topological charge based on a spiral slit,”
Nanophotonics, 8 317 https://doi.org/10.1515/nanoph-2018-0214
(2019).
Google Scholar
J. Wang,
“High-dimensional orbital angular momentum comb,”
Adv. Photonics, 4 050501 https://doi.org/10.1117/1.AP.4.5.050501 AOPAC7 1943-8206
(2022).
Google Scholar
Y. H. Bai et al.,
“Vortex beam: generation and detection of orbital angular momentum [Invited],”
Chin. Opt. Lett., 20 012601 https://doi.org/10.3788/COL202220.012601 CJOEE3 1671-7694
(2022).
Google Scholar
H. Zhang et al.,
“Review on fractional vortex beam,”
Nanophotonics, 11 241 https://doi.org/10.1515/nanoph-2021-0616
(2022).
Google Scholar
Y. K. Li et al.,
“Theoretical study of vortex beam generation based on geometric coordinate transformation,”
Front. Phys., 10 898638 https://doi.org/10.3389/fphy.2022.898638
(2022).
Google Scholar
D. Lorenser et al.,
“Energy-efficient low-Fresnel-number Bessel beams and their application in optical coherence tomography,”
Opt. Lett., 39 548 https://doi.org/10.1364/OL.39.000548 OPLEDP 0146-9592
(2014).
Google Scholar
L. Zhu and J. Wang,
“Demonstration of obstruction-free data-carrying N-fold Bessel modes multicasting from a single Gaussian mode,”
Opt. Lett., 40 5463 https://doi.org/10.1364/OL.40.005463 OPLEDP 0146-9592
(2015).
Google Scholar
S. Chen et al.,
“Demonstration of 20-Gbit/s high-speed Bessel beam encoding/decoding link with adaptive turbulence compensation,”
Opt. Lett., 41 4680 https://doi.org/10.1364/OL.41.004680 OPLEDP 0146-9592
(2016).
Google Scholar
S. Li and J. Wang,
“Adaptive free-space optical communications through turbulence using self-healing Bessel beams,”
Sci. Rep., 7 43233 https://doi.org/10.1038/srep43233 SRCEC3 2045-2322
(2017).
Google Scholar
J. Baltrukonis et al.,
“High-order vector Bessel-Gauss beams for laser micromachining of transparent materials,”
Phys. Rev. Appl., 16 034001 https://doi.org/10.1103/PhysRevApplied.16.034001 PRAHB2 2331-7019
(2021).
Google Scholar
N. Chattrapiban et al.,
“Generation of nondiffracting Bessel beams by use of a spatial light modulator,”
Opt. Lett., 28 2183 https://doi.org/10.1364/OL.28.002183 OPLEDP 0146-9592
(2003).
Google Scholar
Y. Y. Choporova et al.,
“High-power Bessel beams with orbital angular momentum in the terahertz range,”
Phys. Rev. A, 96 023846 https://doi.org/10.1103/PhysRevA.96.023846
(2017).
Google Scholar
R. Feng et al.,
“Flexible manipulation of Bessel-like beams with a reconfigurable metasurface,”
Adv. Opt. Mater., 8 2001084 https://doi.org/10.1002/adom.202001084 2195-1071
(2020).
Google Scholar
N. K. Efremidis et al.,
“Airy beams and accelerating waves: an overview of recent advances,”
Optica, 6 686 https://doi.org/10.1364/OPTICA.6.000686
(2019).
Google Scholar
L. Chen et al.,
“Self-healing property of focused circular Airy beams,”
Opt. Express, 28 36516 https://doi.org/10.1364/OE.405867 OPEXFF 1094-4087
(2020).
Google Scholar
X. X. Chu, G. Q. Zhou, and R. P. Chen,
“Analytical study of the self-healing property of Airy beams,”
Phys. Rev. A, 85 13815 https://doi.org/10.1103/PhysRevA.85.013815
(2012).
Google Scholar
I. Dolev and A. Arie,
“Three wave mixing of airy beams in a quadratic nonlinear photonic crystals,”
Appl. Phys. Lett., 97 171102 https://doi.org/10.1063/1.3504247 APPLAB 0003-6951
(2010).
Google Scholar
I. Gris-Sánchez, D. Van Ras, and T. A. Birks,
“The Airy fiber: an optical fiber that guides light diffracted by a circular aperture,”
Optica, 3 270 https://doi.org/10.1364/OPTICA.3.000270
(2016).
Google Scholar
W. M. Hao et al.,
“High-efficiency generation of airy beams with Huygens’ metasurface,”
Phys. Rev. Appl., 11 054012 https://doi.org/10.1103/PhysRevApplied.11.054012 PRAHB2 2331-7019
(2019).
Google Scholar
J. Y. Liu et al.,
“Propagation dynamics of cosh-Airy beams in Kerr nonlinear media,”
J. Nonlinear Opt. Phys., 28 1950030 https://doi.org/10.1142/S0218863519500309 JNOMFV 0218-8635
(2019).
Google Scholar
C. Maurer et al.,
“Tailoring of arbitrary optical vector beams,”
New J. Phys., 9 78 https://doi.org/10.1088/1367-2630/9/3/078 NJOPFM 1367-2630
(2007).
Google Scholar
W. Cheng, J. W. Haus, and Q. W. Zhan,
“Propagation of vector vortex beams through a turbulent atmosphere,”
Opt. Express, 17 17829 https://doi.org/10.1364/OE.17.017829 OPEXFF 1094-4087
(2009).
Google Scholar
S. Vyas, Y. Kozawa, and S. Sato,
“Polarization singularities in superposition of vector beams,”
Opt. Express, 21 8972 https://doi.org/10.1364/OE.21.008972 OPEXFF 1094-4087
(2013).
Google Scholar
K. J. Mitchell et al.,
“High-speed spatial control of the intensity, phase and polarisation of vector beams using a digital micro-mirror device,”
Opt. Express, 24 29269 https://doi.org/10.1364/OE.24.029269 OPEXFF 1094-4087
(2016).
Google Scholar
W. X. Shu et al.,
“Propagation model for vector beams generated by metasurfaces,”
Opt. Express, 24 21177 https://doi.org/10.1364/OE.24.021177 OPEXFF 1094-4087
(2016).
Google Scholar
L. Fang and J. Wang,
“All-fiber polarization beam splitting and rotating based on vector-mode-assisted coupling,”
Opt. Express, 26 15124 https://doi.org/10.1364/OE.26.015124 OPEXFF 1094-4087
(2018).
Google Scholar
L. Fang and J. Wang,
“Full-vectorial mode coupling in optical fibers,”
IEEE J. Quantum Electron., 54 1 https://doi.org/10.1109/JQE.2018.2810326 IEJQA7 0018-9197
(2018).
Google Scholar
D. Y. Wang et al.,
“Efficient generation of complex vectorial optical fields with metasurfaces,”
Light Sci. Appl., 10 67 https://doi.org/10.1038/s41377-021-00504-x
(2021).
Google Scholar
F. Baronio et al.,
“Spatiotemporal optical dark X solitary waves,”
Opt. Lett., 41 5571 https://doi.org/10.1364/OL.41.005571 OPLEDP 0146-9592
(2016).
Google Scholar
Q. Cao et al.,
“Non-spreading Bessel spatiotemporal optical vortices,”
Sci. Bull., 67 133 https://doi.org/10.1016/j.scib.2021.07.031
(2022).
Google Scholar
L. Chen et al.,
“Synthesizing ultrafast optical pulses with arbitrary spatiotemporal control,”
Sci. Adv., 8 eabq8314 https://doi.org/10.1126/sciadv.abq8314 STAMCV 1468-6996
(2022).
Google Scholar
P. S. Claveria et al.,
“Spatiotemporal dynamics of ultrarelativistic beam-plasma instabilities,”
Phys. Rev. Res., 4 023085 https://doi.org/10.1103/PhysRevResearch.4.023085 PRSTCR 1554-9178
(2022).
Google Scholar
N. Jhajj et al.,
“Spatiotemporal optical vortices,”
Phys. Rev. X, 6 031037 https://doi.org/10.1103/PhysRevX.6.031037 PRXHAE 2160-3308
(2016).
Google Scholar
W. H. Jia et al.,
“Intracavity spatiotemporal metasurfaces,”
Adv. Photonics, 5 026002 https://doi.org/10.1117/1.AP.5.2.026002 AOPAC7 1943-8206
(2023).
Google Scholar
D. J. McCabe et al.,
“Spatio-temporal focusing of an ultrafast pulse through a multiply scattering medium,”
Nat. Commun., 2 447 https://doi.org/10.1038/ncomms1434 NCAOBW 2041-1723
(2011).
Google Scholar
A. Chong et al.,
“Generation of spatiotemporal optical vortices with controllable transverse orbital angular momentum,”
Nat. Photonics, 14 350 https://doi.org/10.1038/s41566-020-0587-z NPAHBY 1749-4885
(2020).
Google Scholar
M. Yessenov et al.,
“Space-time wave packets localized in all dimensions,”
Nat. Commun., 13 4573 https://doi.org/10.1038/s41467-022-32240-0 NCAOBW 2041-1723
(2022).
Google Scholar
D. Cruz-Delgado et al.,
“Synthesis of ultrafast wavepackets with tailored spatiotemporal properties,”
Nat. Photonics, 16 686 https://doi.org/10.1038/s41566-022-01055-2 NPAHBY 1749-4885
(2022).
Google Scholar
C. Wan et al.,
“Toroidal vortices of light,”
Nat. Photonics, 16 519 https://doi.org/10.1038/s41566-022-01013-y NPAHBY 1749-4885
(2022).
Google Scholar
X. Liu et al.,
“Spatiotemporal optical vortices with controllable radial and azimuthal quantum numbers,”
Nat. Commun., 15 5435 https://doi.org/10.1038/s41467-024-49819-4 NCAOBW 2041-1723
(2024).
Google Scholar
S. Franke-Arnold et al.,
“Advances in optical angular momentum,”
Laser Photonics Rev., 2 299 https://doi.org/10.1002/lpor.200810007
(2008).
Google Scholar
A. M. Yao et al.,
“Orbital angular momentum: origins, behavior and applications,”
Adv. Opt. Photonics, 3 161 https://doi.org/10.1364/AOP.3.000161 AOPAC7 1943-8206
(2011).
Google Scholar
|