Photonics Research, Volume. 13, Issue 8, 2130(2025)

Polarization-multiplexing metasurfaces for tunable wavefront configurations via Moiré engineering

Wenhui Xu1, Hui Li1, Chenghui Zhao2, Jie Li3, Qi Tan1, Yufei Liu1, Hang Xu1,5、*, Yun Shen2,6、*, and Jianquan Yao1,4,7、*
Author Affiliations
  • 1Key Laboratory of Opto-Electronics Information Technology (Tianjin University), Ministry of Education, School of Precision Instruments and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China
  • 2Department of Physics, School of Physics and Materials Science, Nanchang University, Nanchang 330031, China
  • 3Sichuan Meteorological Optoelectronic Sensor Technology and Application Engineering Research Center, Chengdu University of Information Technology, Chengdu 610225, China
  • 4Key Laboratory of Photonic-Electronic Integration and Communication-Sensing Convergence (Southwest Jiaotong University), Ministry of Education, Southwest Jiaotong University, Chengdu 610097, China
  • 5e-mail: xh_931119@tju.edu.cn
  • 6e-mail: shenyun@ncu.edu.cn
  • 7e-mail: jqyao@tju.edu.cn
  • show less

    Traditional tunable metasurfaces have evolved through mechanisms relying on external stimuli, such as electrical, thermal, or optical excitation, to dynamically control electromagnetic (EM) wavefronts. While these approaches enable functionalities like focal varying and polarization modulation, they suffer from inherent limitations, including energy inefficiency, structural complexity, and limited adaptability. Here, cascaded all-dielectric Moiré metasurfaces are introduced, which are capable of simultaneous polarization multiplexing and focal-length control for terahertz (THz) beams without external stimuli. Moiré device 1 combines polarization-insensitive (Layer 1) and polarization-sensitive (Layer 2) meta-atoms to independently tailor orthogonal circular polarization channels, including left-handed circular polarization (LCP) and right-handed circular polarization (RCP). Under circularly polarized illumination, it generates focused beams with distinct topological charges (l=0 for LCPRCP and l=1 for RCPLCP), while relative layer rotation enables continuous focal-length adjustment from 9.28 mm to 3.22 mm, accompanied by a numerical aperture (NA) increase from 0.54 to 0.88. Moiré device 2 extends this paradigm to orthogonal linear polarization (LP) channels, producing l=1 and l=0 beams under x-LP and y-LP illumination, with a zoom range of 8.42–3.11 mm and NA up to 0.88. Experimental results validate polarization-selective focusing with efficiency exceeding 15% and robust agreement with simulation results, and the calculated absolute percentage errors (APEs) are below 5.9% for focal length and 3% for NA. These values are consistent with the expected theoretical trends, demonstrating that the experimental results align well with the predicted performance. This reconfigurable system introduces additional control dimensions through mechanical adjustments to cascaded metasurfaces, paving the way for adaptive wavefront control and opening new avenues for next-generation optical technologies.

    1. INTRODUCTION

    Metasurfaces, as planar arrays of subwavelength nanostructures, have revolutionized photonics and electromagnetics by enabling unprecedented control over light and wavefronts [18]. Their ultra-compact, lightweight, and multifunctional properties have unlocked transformative applications across diverse fields such as holography [912], beam steering [13,14], biosensing [15,16], and communications [17,18]. THz technology has emerged as a pivotal frontier in modern photonics, offering transformative potential across applications such as imaging [1922], sensing [2325], and wireless communications [26,27]. A critical challenge in this domain lies in achieving dynamic and efficient control over THz wavefronts, particularly with regard to polarization-multiplexed tunable manipulation. The evolution from static to dynamic control in THz metasurfaces has been driven by stimuli-responsive materials and mechanisms. Optically tuned metasurfaces leverage photoexcitation in phase transitions in materials like Ge2Sb2Te5 (GST) to modulate THz waves via carrier density or refractive index changes, achieving ultrafast responses but suffering from thermal damage and limited modulation depth [2830]. Thermally tuned designs, such as VO2-based metasurfaces, exploit insulator-to-metal phase transitions under heating, offering high modulation contrast but facing slow response and high energy consumption [3133]. Electrically controlled metasurfaces, utilizing graphene or liquid crystals, enable tuning via carrier injection or molecular alignment under electric fields, yet require high voltages and complex electrode integration [3436].

    A promising alternative lies in Moiré metasurfaces, which leverage rotational misalignment between cascaded metasurface layers to achieve dynamic phase modulation without external stimuli. By rotating cascaded metasurface layers, the effective phase profile can be continuously modulated, enabling reconfigurable optical responses. Zhang et al. demonstrated a Moiré meta-device using cascaded dielectric metasurfaces to dynamically tune Bessel beam order and non-diffraction length via mutual rotation; experimental results validate high-quality Bessel beam generation with reconfigurable parameters [37]. Xiao et al. introduced adaptive aberration correction during focal scanning using two rotating cascaded silicon metasurfaces. Optimized phase functions enabled dynamic focus control on custom planar and conical surfaces in the terahertz regime without external components [38]. Chen et al. proposed twisted dielectric metasurfaces that achieve on-demand subwavelength focusing localization through rotational interlayer coupling, enabling dynamic control of focal position and intensity without mechanical adjustments. Experimental validation demonstrates spatially reconfigurable focusing with diffraction-limited precision [39]. Despite these advances, existing Moiré-based designs face two critical limitations. On the one hand, they predominantly focus on single-polarization-state manipulation, neglecting the independent control of orthogonal polarization channels. On the other hand, their zoom ranges and NAs remain constrained, limiting their utility in applications requiring high-resolution and polarization-multiplexed functionalities.

    In this work, we address these challenges by introducing a cascaded all-dielectric Moiré metasurface platform that achieves simultaneous focal-length tuning and independent polarization control in the THz regime. Our design combines polarization-insensitive and polarization-sensitive sub-metasurface layers, enabling decoupled manipulation of orthogonal circular (Moiré device 1) or linear (Moiré device 2) polarization states. Here, Moiré device 1 generates focused beams carrying topological charges l=0 and l=1 in the LCP→RCP and RCP→LCP channels, respectively, under orthogonal circularly polarized illumination. In contrast, Moiré device 2 produces focused beams with l=1 and l=0 in the co-polarized channels under x-LP and y-LP illumination, achieving polarization-dependent phase control. As an example, the feasibility of the proposed design strategy is further verified using a THz near-field imaging system with the metasurface under linearly polarized illumination. By rotating the layers relative to one another, the effective phase profile is reconfigured, allowing continuous focal-length adjustment from 8.42 mm to 3.11 mm and a zoom ratio exceeding 2.7 times, while simultaneously tailoring polarization responses. This dual functionality is achieved without external stimuli, relying solely on the phase and interlayer rotational coupling. These findings demonstrate twisted metasurfaces’ capacity to minimize conventional optical dependencies, advancing cost-effective and sustainable communication technologies.

    2. DESIGN AND METHOD

    Figure 1 presents a schematic illustration of the proposed Moiré device 1 for on-demand wavefront shaping. This meta-device comprises two coaxially arranged all-dielectric metasurfaces with precise alignment: the upper metasurface, referred to as Layer 1, and the lower metasurface, referred to as Layer 2. This configuration enables the generation of a focused vortex beam with a topological charge of l=1 under RCP illumination, while simultaneously producing a focused Gaussian beam with l=0 under LCP illumination, as depicted in Fig. 1(a). The described wavefront shaping process can be mathematically expressed as [40,41] {OMS|σL=exp(il1ζ)|σR,OMS|σR=exp(il2ζ)|σL.

    Schematic representation of the designed Moiré device 1 for on-demand wavefront shaping. (a) Conceptual illustration of the proposed Moiré device 1, comprising two layers of mechanically rotated all-dielectric metasurfaces (Layer 1 and Layer 2). (b) Phase distribution of Layer 1 at a mutual twisted angle (α) of 90°, along with the corresponding phase distributions of Layer 2 in the LCP and RCP channels, and the resultant joint phase profiles. (c) Electric field distributions monitored in the orthogonal circular polarization channels on the xoy and xoz planes. (d) Focal length and NA as a function of the mutual rotation angle (α).

    Figure 1.Schematic representation of the designed Moiré device 1 for on-demand wavefront shaping. (a) Conceptual illustration of the proposed Moiré device 1, comprising two layers of mechanically rotated all-dielectric metasurfaces (Layer 1 and Layer 2). (b) Phase distribution of Layer 1 at a mutual twisted angle (α) of 90°, along with the corresponding phase distributions of Layer 2 in the LCP and RCP channels, and the resultant joint phase profiles. (c) Electric field distributions monitored in the orthogonal circular polarization channels on the xoy and xoz planes. (d) Focal length and NA as a function of the mutual rotation angle (α).

    Here, σL=[1i] denotes LCP and σR=[1i] denotes RCP; l1 and l2 present the topological charges carried by the LCP and RCP channels, respectively. In this context, we set l1=0, while l2=1, and ζ denotes the azimuthal angle. The focal length of the generated beam can be dynamically modulated by rotating Layer 1 relative to Layer 2, introducing a controlled twist angle between the two metasurfaces. This tunable configuration enables the derivation of a functional relationship between the relative twisted angle and the resulting focal length. Such a design underscores the versatility and robustness of the proposed strategy, allowing for precise tailoring of beam properties through polarization control and spatial phase manipulation.

    The functionality of the proposed Moiré device 1 relies on meticulous phase engineering. Based on the Moiré effect, the phases of the two metasurfaces constituting the composite lens are complex conjugate to each other. In polar coordinates, their original phases can be expressed as ϕ1(r,α0)=round(r2λf)(α0),ϕ2(r,α0)=round(r2λf)(α0),where r and α0 are the polar coordinate parameters, λ is the operating wavelength, and f represents the predefined focal length. The round (•) function converts the value of the operand to the nearest integer to avoid sectorization effects. For Layer 1, the design must enable dynamic focusing as the layer rotates, whereas for Layer 2, the phase distribution is tailored to achieve polarization multiplexing. According to Eq. (2), when Layer 1 is rotated around the axis by an angle θ relative to Layer 2, the phase profile of Layer 1 can be written as [42,43] ϕ1(r,α0;α)=round(r2λf)(α0α),where α denotes the mutual twisted angle. For the phase design of Layer 2, its primary function is to independently encode the orthogonal circularly polarized channels and to enable generation of vortex beams for orthogonal circularly polarized channels, where helical phase lζ is introduced. Based on Eq. (3), the encoded phase distribution at Layer 2 can be mathematically expressed as ϕ2LR(r,α0)=round(r2/λf)α0+l1ζ and ϕ2RL(r,α0)=round(r2/λf)α0+l2ζ, where the parameters l1 and l2 represent the topological charges assigned to the LCPRCP and RCPLCP channels, respectively. The variable ζ=arctan(y/x) denotes the polar angle used to encode the vortex phase. In this design, the topological charges are set as l1=0 and l2=1. Based on Eq. (4), the total phase of the Moiré device 1 for RCP/LCP can be expressed as ϕJointLR(r,α0;α)=round(r2λf)α+l1ζ,ϕJointRL(r,α0;α)=round(r2λf)α+l2ζ.

    Based on the total phase distribution, it can be seen that by superimposing the phase profiles of the two metasurfaces with a relative rotation (angle α), the interferences in the phase distribution generate dynamic effective phase gradients, which modulate the wavefront characteristics. Under RCP illumination, Moiré device 1 generates a vortex beam carrying a topological charge of l=1, whereas LCP excitation produces a focused Gaussian beam (l=0), demonstrating polarization-dependent phase modulation for selective structured field generation. To clarify the operational mechanism of the proposed design, the encoded phase distributions of Layer 1 and Layer 2 at a relative twisted angle of 90° are illustrated in Fig. 1(b). These phase profiles demonstrate the independent encoding of orthogonal circular polarization channels and the corresponding joint phase distributions. The orthogonal circularly polarized channels exhibit phase profiles consistent with the assigned topological charges, facilitating the selective generation of distinct beam types. Moreover, the electric field distributions in the xoy- and xoz-planes for the corresponding polarization channels are evaluated at the designed focal plane. As shown in Fig. 1(c), within the LCPRCP channel, the device generates a focused Gaussian beam, while in the RCPLCP channel, a vortex beam with a characteristic spiral phase profile is observed. According to Eqs. (5) and (6), the relationship between the focal length and the relative twisted angle α can be derived, enabling precise control over the focal properties. This functional relationship, which forms the foundation of the device’s tunable focusing capabilities, is expressed as F(α)=πα,where C is a constant and can be adopted as 1/λf; λ is considered a constant value. Therefore, it can be concluded that the focal length F varies with twisted angle α. Consequently, the NA can be expressed as NA=sin(arctanL2·F(α)).

    Here, L denotes the diameter of the Moiré device 1, which plays a critical role in determining the transmitted performance. As shown in Fig. 1(d), an inverse relationship is observed between the focal length and the twisted angle α (depicted by the green curve). Specifically, as the angle α increases, the focal length gradually decreases, indicating a compression of the beam’s focal region. Conversely, a direct proportional relationship is evident between the NA and the twisted angle, as illustrated by the red curve in Fig. 1(d). The NA parameter expands steadily with increasing twisted angle, which suggests an enhanced focusing capability and a reduction in the focal spot size. This behavior underscores the tunable property of the proposed meta-device, as both focal length and NA can be dynamically adjusted through precise control of the relative twisted angle.

    Given that the primary function of Layer 1 is to achieve dynamic focal adjustment (zoom), polarization-insensitive cylindrical pillars were deliberately chosen for the design of the meta-atoms. Figure 2(a) illustrates the side and top views of the cylindrical dielectric pillar, characterized by a height H=200  μm and a variable radius r, fabricated on a silicon substrate with a periodicity P=150  μm. To achieve full 2π phase modulation at the target frequency of 0.9 THz, a comprehensive parameter sweep for r was performed using the CST Microwave Studio software. From the simulation results, eight distinct meta-atoms were selected to provide a complete phase coverage ranging [π,π]. As shown in Fig. 2(b), the selected meta-atoms exhibit consistent transmission amplitude and phase response under linearly polarized illumination, ensuring uniform performance for both x-LP and y-LP incidences. Figure 2(c) presents the normalized magnetic field distribution of the cylindrical dielectric pillars monitored under periodic boundary conditions, by utilizing the time-domain solver. Apparently, the EM energy carried by the incident THz wave is localized mainly inside the dielectric pillars with high refractive indices, suggesting that the coupling between neighboring meta-atoms is negligible.

    Transmission properties of the selected meta-atoms used to assemble the Moiré device 1. (a) Side and top views of the meta-atoms employed to construct Layer 1 with polarization maintaining properties. (b) Transmission amplitudes and phase delays of cylindrical dielectric pillars labeled from 1 to 8 that satisfy the 2π phase coverage. (c) Normalized magnetic field distributions for cylindrical dielectric pillar with periodic boundary conditions. (d) Side and top views of the meta-atoms employed to construct Layer 2 with polarization converting properties. Transmission amplitudes and phase delays of rectangular dielectric pillars labeled from 1 to 15 that satisfy the 2π propagation phase coverage under (e) x-LP and (f) y-LP incidence. (g) In-plane rotation angles corresponding to the selected 15 rectangular dielectric pillars were used to generate geometric phase profiles. (h) Normalized magnetic field distributions for rectangular dielectric pillars with periodic boundary conditions.

    Figure 2.Transmission properties of the selected meta-atoms used to assemble the Moiré device 1. (a) Side and top views of the meta-atoms employed to construct Layer 1 with polarization maintaining properties. (b) Transmission amplitudes and phase delays of cylindrical dielectric pillars labeled from 1 to 8 that satisfy the 2π phase coverage. (c) Normalized magnetic field distributions for cylindrical dielectric pillar with periodic boundary conditions. (d) Side and top views of the meta-atoms employed to construct Layer 2 with polarization converting properties. Transmission amplitudes and phase delays of rectangular dielectric pillars labeled from 1 to 15 that satisfy the 2π propagation phase coverage under (e) x-LP and (f) y-LP incidence. (g) In-plane rotation angles corresponding to the selected 15 rectangular dielectric pillars were used to generate geometric phase profiles. (h) Normalized magnetic field distributions for rectangular dielectric pillars with periodic boundary conditions.

    Layer 2, on the other hand, is specifically designed for polarization conversion. Anisotropic rectangular pillars were introduced to achieve polarization-multiplexing functions. This configuration employs meta-atoms composed of rectangular nano-pillars with a height H=200  μm, fabricated on a silicon substrate with a periodicity P=150  μm. The characteristic parameters Lx and Ly are predefined for manipulating the orthogonal polarization channels and the desired meta-atoms are obtained in the time-domain solver by utilizing a parametric scanning process. The x- and y-directions along the dielectric pillars are set as periodic boundary conditions, while the z-direction is set as an open boundary condition. To enable independent encoding for orthogonal circular polarization channels, the design leverages the principle of a spin-decoupling mechanism. This approach ensures that the meta-atoms independently manipulate the phase profiles of LCP and RCP waves, expressed as [40,4446] {φxx=(φLCP+φRCP)/2,φyy=(φLCP+φRCP)/2π,θ=(φRCPφLCP)/4.

    Here, φxx (φyy) represents the propagation phase at x(y)-LP incidence, φLCP (φRCP) represents the phase distribution of the LCP (RCP), and θ is the rotation angle of the meta-atom. The details of the calculations corresponding to the spin-decoupled phase engineering are presented in Appendix A. Using CST software for parameter scanning (varying the Lx and Ly of the rectangular pillar structure), we simulated the transmission amplitude and phase shift of the meta-atom. Based on the requirements set forth by Eqs. (9) and (A5), meta-atoms were selected to enable independent phase modulation for incident LCP and RCP waves. The transmission amplitudes and phase shifts of these selected meta-atoms are illustrated in Figs. 2(e) and 2(f). Notably, the corresponding phase distributions φxx and φyy exhibit complete 2π coverage within the range of π to π, ensuring comprehensive phase modulation capability. Additionally, the transmission amplitudes remain within the range of 0.6 to 0.8 and display a relatively smooth variation across the selected meta-atoms. As depicted in Fig. 2(g), the rotation angle θ demonstrates a well-defined function relationship with both the phase distribution and the meta-atom orientation. To gain deeper insights into the electromagnetic response, the normalized magnetic field distribution of structural units with periodic boundary conditions was analyzed. The results, shown in Fig. 2(h), reveal that the energy of the incident electromagnetic wave is strongly confined within the high-refractive-index medium columns. This indicates minimal interlayer coupling, thereby preserving the independent functionality of the metasurfaces and ensuring high design robustness.

    3. RESULTS AND DISCUSSION

    Figure 3(a) presents the phase distribution of Layer 2 within the LCP channel, where each discrete pixel represents a precisely engineered anisotropic rectangular silicon pillar. These meta-atoms were optimized to encode the required phase profile, ensuring accurate polarization control and phase manipulation. Figure 3(b) systematically demonstrates the phase distribution of Layer 1 under various twisted angles. As the twisted angle α increases from 90° to 240° in 30° increments, the phase profile exhibits a counterclockwise rotational shift around the central axis of the image, highlighting the tunable characteristic of the metasurface and its ability to dynamically reconfigure the phase distribution through simple mechanical rotation. The joint phase distribution of Layer 1 and Layer 2, corresponding to the LCP channel, is shown in Fig. 3(c). This joint phase profile is computed by superimposing the phase contributions from both Layer 1 and Layer 2 under the respective rotation angles. The resultant combined phase pattern demonstrates precise control over the spatial phase distribution. Under LCP illumination, the incident wave traverses the meta-device, enabling electric field distributions to be analyzed in terms of their RCP components using a field monitor in transmission mode [Fig. 3(d)]. Notably, as the rotation angle α increases from 90° to 240°, the focal length of the generated focused beams decreases from 9.28 mm to 3.22 mm. In parallel, Fig. 3(e) shows the phase distribution of Layer 2 within the RCPLCP channel, while the corresponding phase distribution of Layer 1 [identical to Fig. 3(b)] is shown in Fig. 3(f). The joint phase distribution for the RCP channel, obtained as the Fourier sum of the phases from Layer 1 and Layer 2, is presented in Fig. 3(g). This phase distribution adjusts dynamically with changes in α, creating distinct interference patterns that enable the emitted waves to generate vortex beams focused at specific points on the focal plane [Fig. 3(h)]. The electric field distributions in the xoz-plane monitored with the time-domain solver are presented in Appendix B, with increasing relative rotation angles from 90° to 240°. Figure 3(i) demonstrates the inverse correlation between focal length and rotation angle α, coupled with a direct proportionality of NA to α, aligning qualitatively with theoretical predictions. Specifically, as α increases from 90° to 240°, the NA rises from 0.54 to 0.88. In addition, we further quantified the operating performance of the Moiré device 1 by introducing the APE parameter, the details of which are demonstrated in Appendix C. Apparently, the average APEs of focal length and NA between the theoretically predicted and numerically simulated values are 25.2% and 15%, respectively, which indicates the effectiveness of the proposed design strategy. The reason for the higher APE values is the complex near-field coupling between the layers and the sensitivity of the phase control to structural tolerances, which are reflected in the simulations but less directly in the theoretical model. Nonetheless, the device shows a consistent trend in focal length adjustment, validating the design principle of rotating Moiré phase engineering. Figure 3(j) quantifies the focusing efficiency of the Moiré device 1, defined as [47] ηf=E˜f/E˜sum=Ef(Ef(Emax/e))/Ef, where E˜f denotes the integral of the focused field distribution over a range greater than Emax/e, and E˜sum denotes the integral of the field distribution at the entire focal plane. With the gradual increase of parameter α, the focusing efficiency of the focused beam carrying topological charge l=1 generated within the LCP component exhibits a monotonic efficiency reduction, while the trend of the focusing efficiency of the focused beam carrying topological charge l=0 generated within the RCP component maintains near-constant. Further calculated results indicate an average focusing efficiency of approximately 35%, primarily affected by diffraction losses, discrete phase control defects, and material losses. To further evaluate the polarization isolation capability of the Moiré device 1, we quantified the crosstalk between LCP and RCP through numerical simulations, as detailed in Appendix D.

    Encoded phase profiles and generated electric field distributions of Moiré device 1. (a) Encoded phase distributions of Layer 2 within the LCP→RCP channel. (b) Encoded phase distributions of Layer 1 under different twisted angles from 90° to 240°. (c) Joint encoding phase distributions and (d) the produced focusing field distributions carrying the topological charge l=0 within the LCP→RCP channel. (e) Encoded phase distributions of Layer 2 within the RCP→LCP channel. (f) Encoded phase distributions of Layer 1 under different rotation angles from 90° to 240°. (g) Joint encoding phase distributions and (h) the produced focusing field distributions carrying the topological charge l=1 within the RCP→LCP channel. (i) Focal length and NA as a function of relative twisted angle α, including theoretical predictions and simulation results. (j) Focusing efficiencies calculated in different polarization channels.

    Figure 3.Encoded phase profiles and generated electric field distributions of Moiré device 1. (a) Encoded phase distributions of Layer 2 within the LCPRCP channel. (b) Encoded phase distributions of Layer 1 under different twisted angles from 90° to 240°. (c) Joint encoding phase distributions and (d) the produced focusing field distributions carrying the topological charge l=0 within the LCPRCP channel. (e) Encoded phase distributions of Layer 2 within the RCPLCP channel. (f) Encoded phase distributions of Layer 1 under different rotation angles from 90° to 240°. (g) Joint encoding phase distributions and (h) the produced focusing field distributions carrying the topological charge l=1 within the RCPLCP channel. (i) Focal length and NA as a function of relative twisted angle α, including theoretical predictions and simulation results. (j) Focusing efficiencies calculated in different polarization channels.

    While the above results demonstrate successful dynamic control of orthogonal circular polarization channels, the practical implementation of polarization-selective meta-devices often requires compatibility with both circularly polarized (CP) and linearly polarized (LP) systems depending on application scenarios. In order to validate the generality of the cascaded metasurface design across different polarization bases, as well as to satisfy different application requirements, we further designed Moiré device 2 with the aim of verifying the versatility and flexibility of this design framework. The schematic of the designed Moiré device 2 is illustrated in Fig. 4(a). Layer 1 of the metasurface is identical to the design presented in Fig. 1(a), while Layer 2 consists of rectangular pillar units engineered based on propagation phase principles, distinct from Layer 2 in Moiré device 1. This configuration facilitates the generation of a focused beam with a topological charge of l=0 in the y-LP channel under y-LP illumination. Simultaneously, a focused vortex beam carrying a topological charge of l=1 is produced in the x-LP channel under x-LP illumination. The described wavefront shaping process can be mathematically expressed as {OMD2|σx=exp(il1ϕ)|σx,OMD2|σy=exp(il2ϕ)|σy.

    Simulation results of the designed Moiré device 2 under x-LP and y-LP incidence. (a) Conceptual illustration of the proposed Moiré device 2, comprising two layers of mechanically rotated all-dielectric metasurfaces (Layer 1 and Layer 2). (b) Electric field and phase distributions under (b) x-LP illumination and (c) y-LP illumination, as the relative rotation angle α gradually increase from 90° to 240°.

    Figure 4.Simulation results of the designed Moiré device 2 under x-LP and y-LP incidence. (a) Conceptual illustration of the proposed Moiré device 2, comprising two layers of mechanically rotated all-dielectric metasurfaces (Layer 1 and Layer 2). (b) Electric field and phase distributions under (b) x-LP illumination and (c) y-LP illumination, as the relative rotation angle α gradually increase from 90° to 240°.

    Here, σx=[10] denotes x-LP incidence and σy=[01] denotes y-LP incidence; l1 and l2 present the topological charges associated with the x-LP and y-LP incidence, respectively. In this context, we set l1=1, while l2=0, and ϕ denotes the azimuthal angle. Furthermore, the ability to dynamically adjust the focal length of the generated beams is achieved by rotating Layer 1. The range of rotation angles is defined as (0°, 360°).

    The simulation results for x-LP incidence are shown in Fig. 4(b), where a focused vortex beam is generated within the x-LP channel. This beam exhibits a characteristic intensity profile with zero intensity at the phase singularity, surrounded by a ring-shaped intensity distribution. As the mutual twisted angle α between the two metasurfaces (Layer 1 and Layer 2) increases, the phase distribution region corresponding to the focal area has diminished, and the phase transition has become more localized. Similarly, Fig. 4(c) demonstrates the response under y-LP incidence, where a conventional focused beam is observed within the y-LP channel. Unlike the vortex beam, this beam features a Gaussian-like intensity profile at the focal point. Further, the electric field distribution (xoz-plane) generated by the Moiré device 2 under orthogonal linearly polarized wave illumination is exhibited in Appendix E. Apparently, the focal length of the transmitted beam is inversely related to the relative rotation angle. Due to the device’s ability to manipulate phase fronts through relative rotational alignment, both the focusing range and spatial phase distribution are systematically reduced, enabling dynamic tuning of the focal properties.

    Figure 5(a) presents a schematic diagram of the THz near-field scanning spectroscopy system, which was employed to measure the proposed Moiré meta-devices. The employed experimental system is equipped with microprobes that can record the complex amplitude information at the focal plane, pixel by pixel, under LP illumination. Sample fabrication was performed using ultraviolet (UV) lithography and inductively coupled plasma (ICP) etching [48,49]. High-resistivity silicon with a thickness of 500 μm was used as the substrate to optimize the device’s optical performance and minimize signal loss. The scanning electron microscope (SEM) images of the fabricated metasurfaces (Layer 1 and Layer 2) are shown in Figs. 5(b) and 5(c). The dimensions of the samples are 1.4  cm×1.4  cm, with a central circular region composed of 80×80 silicon pillars. For Layer 1, the magnified detail images highlight that its basic unit consists of cylindrical silicon pillars, which are characterized by sharp boundaries and smooth surfaces, reflecting a high level of fabrication precision. In contrast, Layer 2 is composed of rectangular silicon pillars, which also demonstrate excellent fabrication quality, as evidenced by the magnified detail images presented herein. To ensure sub-wavelength precision in aligning cascaded metasurface architectures, the experimental setup employed a GCT-090101 multi-axis rotational control system with integrated phase-sensitive feedback mechanisms [37]. The dual-layer configuration features independent rotational stages—Layers 1 and 2 are mounted on aerostatic bearing platforms incorporating high-resolution radial gratings (360° circumferential division). Details of the cascaded metasurfaces constructed with the assistance of the GCT-090101 components are shown in Appendix F. Interlayer spacing optimization follows an inverse proportionality principle derived from coupled-mode theory simulations. The parametric study reveals that the critical coupling distance dc between adjacent metasurfaces scales inversely with their respective phase gradients (ΔΦ), obeying dcλ2/(4πΔΦ), when operating near the Rayleigh-Sommerfeld diffraction regime [37,38,50]. This counterintuitive relationship stems from the competition between near-field evanescent coupling and far-field propagation effects—steeper phase gradients effectively compress the Fresnel zone, necessitating reduced spacing to maintain phase continuity. The synergy between rotational alignment precision and gradient-dependent spacing adjustment enables unprecedented control over the proposed Moiré meta-devices.

    Characterization of THz near-field scanning spectroscopy system and fabricated samples. (a) Schematic diagram of the THz time-domain scanning spectroscopy system. SEM photographs of the fabricated samples, including (b) Layer 1 and (c) Layer 2 of Moiré device 2.

    Figure 5.Characterization of THz near-field scanning spectroscopy system and fabricated samples. (a) Schematic diagram of the THz time-domain scanning spectroscopy system. SEM photographs of the fabricated samples, including (b) Layer 1 and (c) Layer 2 of Moiré device 2.

    To assess the simultaneous control of polarization and focal length in our Moiré device 2 for THz waves, experimental characterization was performed at discrete twisted angles between the metasurface Layer 1 and Layer 2, specifically at 90°, 120°, 150°, 180°, 210°, and 240°. Amplitude and phase distributions were systematically recorded at 0.9 THz (Fig. 6). Under x-LP incidence, Fig. 6(a) presents a characteristic doughnut-shaped pattern, with a dark core at the center, indicative of a phase singularity where the intensity is zero. The phase distribution around this singularity evolves along a closed, helical path, corresponding to a complete 2π phase winding, a hallmark of an optical vortex with topological charge l=1. As the relative twisted angle between the two metasurface layers increases from 90° to 240°, with a step size of 30°, the doughnut-shaped spots in the electric field distributions exhibit progressive radial contraction. Concurrently, the 2π phase winding in the phase distribution narrows, indicating a gradual reduction in the focal length. This experimental result aligns well with the simulation data shown in Fig. 4(b). Subsequently, the electric field distribution and phase profile of the Moiré device 2 under y-LP illumination were further measured under the same experimental conditions, as shown in Fig. 6(b). In the electric field distribution, a well-defined focused spot is observed, where the intensity reaches its maximum at the central point. The phase distribution at the focal region is nearly uniform, indicating coherent focusing with minimal phase variation. As the relative twisted angle α increases from 90° to 240°, the central focused intensity distribution gradually diminishes, which is indicative of a reduction in the effective focal length and a change in the beam’s NA. This experimental result aligns well with the simulation data shown in Fig. 4(c).

    Experimental results of Moiré device 2. (a) The obtained electric fields and phase distributions under the x-LP incidence and (b) y-LP incidence, as the twisted angle α increases. (c) Focal length and (d) NA as a function of the relative twisted angle α, including theoretical, simulated, and experimental results. (e) Focusing efficiency, obtained through simulations and experiments under x-LP and y-LP illuminations.

    Figure 6.Experimental results of Moiré device 2. (a) The obtained electric fields and phase distributions under the x-LP incidence and (b) y-LP incidence, as the twisted angle α increases. (c) Focal length and (d) NA as a function of the relative twisted angle α, including theoretical, simulated, and experimental results. (e) Focusing efficiency, obtained through simulations and experiments under x-LP and y-LP illuminations.

    To further illustrate the zoom functionality of the Moiré device 2, Figs. 6(c) and 6(d) show the focal length and NA as a function of the relative twisted angle α, respectively, with theoretical predictions, simulations, and experimental measurements overlaid. It is evident that as the rotation angle increases from 90° to 240°, the focal length steadily decreases from 8.42 mm to 3.11 mm, demonstrating the dynamic focusing capability of the metasurface. The experimental results closely match the simulation data, exhibiting an APE for focal length of less than 5.9% and an NA of less than 3% (the details can be found in Appendix G), confirming the feasibility of the designed strategy of the Moiré device 2. Such behavior is consistent with the expected response that manipulates phase and polarization at the THz frequency, enabling effective and tunable beam control. As shown in Fig. 6(d), the results show a clear proportional relationship between the NA and the rotation angle α. As the rotation angle increases, the NA parameters obtained in the experiments gradually increase, which is consistent with the evolutionary trend of the theoretical and simulation results. This agreement further underscores the device’s tunable focusing capability. Figure 6(e) presents the focusing efficiency under x-LP and y-LP incidence. The measured focusing efficiencies in both orthogonal polarization channels exhibit a gradual decline with increasing rotational angle while consistently maintaining values above 15% across all tested configurations, which is constrained by fabrication tolerances and suboptimal rotational alignment between layers. This systematic characterization confirms the dual-polarization operational capability of Moiré device 2, validating its capacity for simultaneous polarization-multiplexed focal length tunability and independent polarization channel modulation. In conjunction with the previous discussion, device 1 simulations yielded average focusing efficiencies in excess of 35%, and experimentally measured focusing efficiencies in excess of 15% for device 2, a preliminary exploration for polarization-selective focusing metasurfaces. Further optimization of geometry and material selection and more precise fabrication techniques could lead to even higher efficiencies for both devices. More importantly, the results fundamentally establish that our phase-engineering strategy enables focal length customization while preserving efficient polarization-selective control, which is a critical requirement for advanced polarization-manipulation systems.

    4. CONCLUSIONS

    In summary, we have proposed a tunable-focus cascaded dielectric meta-device (Moiré device 1) employing rotational Moiré phase engineering to achieve simultaneous polarization-channel selectivity and focal-length modulation in transmitted THz beams without external stimuli. Through strategic layer design, that is, utilizing polarization-insensitive dielectric pillars in Layer 1 and polarization-anisotropic pillars in Layer 2, independent manipulation of orthogonal circular polarization states is achieved. Under circularly polarized illumination, distinct topological charges (l=0 in LCPRCP channel, l=1 in RCPLCP channel) are generated through spin-decoupled phase control. The rotational degree of freedom (α) between cascaded metasurfaces enables continuous focal length tuning from 9.28 mm (α=90°, NA=0.54) to 3.22 mm (α=240°, NA=0.88), while maintaining simulation accuracy with a mean APE of 25.2% for focal length and 15% for NA (based on comparison between theoretical design and simulation), thereby demonstrating a dynamically reconfigurable focusing paradigm that exhibits consistent trends with theoretical expectations.

    To further demonstrate the versatility of rotational Moiré engineering, we extend this strategy to linear polarization control in Moiré device 2. Numerical simulations and experiments confirm its ability to generate distinct topological charges (l=1 and l=0) in co-polarized transmission channels under x-LP and y-LP illumination through propagation phase modulation. Experimental validation confirms the device’s dynamic focusing capabilities, achieving a continuous focal length tuning from 8.42 mm to 3.11 mm, which directly governs the corresponding NA enhancement from 0.54 to 0.88, while preserving measurement accuracy with APE of less than 5.9% and 3% for focal length and NA, respectively. Measured focal efficiencies exceeding 15% underscore the adaptability of Moiré phase engineering across polarization bases.

    Collectively, these devices exemplify a unified design philosophy, leveraging interlayer rotation to reconfigure wavefronts without external stimuli. Device 1 prioritizes polarization-selective functionality for chiral light-matter interactions, while device 2 emphasizes compatibility with conventional linear polarization systems. This dual-device approach validates the broad applicability of Moiré engineering, laying the groundwork for future applications such as adaptive aberration correction, biomedical imaging, and multi-dimensional particle manipulation.

    APPENDIX A: DERIVATION PROCEDURE FOR PHASE DECOUPLING IN ORTHOGONAL CIRCULARLY POLARIZED CHANNELS

    For independent control of orthogonal circularly polarized channels, anisotropic rectangular pillars with birefringent effects are used as meta-atoms for the construction of the metasurface, of which the Jones matrix in the global coordinate system can be expressed as the following equation: T=[txx00tyy].

    Here, txx and tyy represent the complex amplitudes derived from the co-polarization channel under the linearly polarized illumination. Subsequently, the structural rotation angle is introduced to apply geometric phase modulation to the LCP/RCP to independently manipulate the orthogonal circularly polarized channels. Thus, the Jones matrix of the meta-atoms can be expressed as J(θ)=R(θ)TR(θ)=[cosθsinθsinθcosθ][txx00tyy][cosθsinθsinθcosθ]=[txxcos2θ+tyysin2θtxxsinθcosθtyysinθcosθtxxsinθcosθtyysinθcosθtxxsin2θ+tyycos2θ]=[12(txx+tyy)+12(txxtyy)cos2θ12(txxtyy)sin2θ12(txxtyy)sin2θ12(txx+tyy)12(txxtyy)cos2θ]=12(txx+tyy)[1001]+12(txxtyy)[cos2θsin2θsin2θcos2θ]=J1+J2(θ).

    txx and tyy reflect the amplitude and phase changes during transmission through the structure, which can be expressed as txx=|txx|eiφxx, tyy=|tyy|eiφyy, and here we set |txx|=|tyy|=t, φyy=φxx+π. The above equation can be further simplified as J1=12(|txx|eiφxx+|tyy|eiφyy)[1001]=12t(1+ei(φyyφxx))eiφxx=0,J2(θ)=12t(1ei(φyyφxx))eiφxx[cos2θsin2θsin2θcos2θ]=14t(1ei(φyyφxx))[ei(φxx+2θ)[1ii1]+ei(φxx2θ)[1ii1]]=12teiφRCP[1ii1]+12teiφLCP[1ii1].

    Here, φRCP=φxx+2θ and φLCP=φxx2θ denote the phase retardation within the orthogonal circularly polarized channel; the 2D Jones vectors corresponding to RCP and LCP are given by σL=[1i] and σR=[1i]. Upon illuminating the LCP beam vertically, the Jones matrix of the transmitted electric field can be calculated as Eout=J·Ein=12t(eiφRCP[1ii1]+eiφLCP[1ii1])·[1i]=teiφRCP[1i].

    For the RCP incidence mode, the Jones matrix of the transmitted field can be described as Eout=J·Ein=12t(eiφRCP[1ii1]+eiφLCP[1ii1])·[1i]=teiφLCP[1i].

    Apparently, phase decoupling operations in orthogonal circularly polarized channels can be achieved by selecting HWP meta-atoms.

    APPENDIX B: SIMULATION RESULTS OF THE FOCUSED FIELD GENERATED BY THE DESIGNED MOIRé DEVICE 1 UNDER CIRCULARLY POLARIZED ILLUMINATION, AS THE MUTUAL ROTATION ANGLE GRADUALLY INCREASES FROM 90° TO 240°

    Figure 7 further presents the simulated performance of Moiré device 1. In the LCP→RCP conversion channel [Fig. 7(a)], the first column depicts the phase distribution profiles on the xoy focal plane, revealing near-uniform phase characteristics with a topological charge number l=0. The phase variation range at the focal center progressively narrows as the relative rotation angle between Layer 1 and Layer 2 increases from 90° to 240°. The second column shows cross-sectional electric field distributions in the xoz-plane, demonstrating a monotonic reduction in focal length from 9.28 mm to 3.22 mm with increasing rotation angles over the tested angular range (90°–240°). For the RCPLCP channel [Fig. 7(b)], the first column reveals a distinct vortex phase profile in the xoy focal plane, characterized by a central phase singularity, indicating a topological charge of l=1. The spatial extent of the vortex structure decreases systematically with increasing rotation angles. The second column displays the corresponding xoz-plane electric field distributions, exhibiting an intensity pattern with pronounced vortical features. Concurrently, the focal length decreases with larger rotation angles, mirroring the trend observed in the LCPRCP channel. These above findings are consistent with the conclusions drawn from Fig. 3, confirming the device’s robust modulation capability through mechanical rotation.

    Simulation results of Moiré device 1. (a) The obtained phase distributions in the xoy-plane and electric fields in the xoz-plane as well as the change in focal length as the rotation angle α increases from 90° to 240° at 30° intervals under LCP incidence and (b) RCP incidence.

    Figure 7.Simulation results of Moiré device 1. (a) The obtained phase distributions in the xoy-plane and electric fields in the xoz-plane as well as the change in focal length as the rotation angle α increases from 90° to 240° at 30° intervals under LCP incidence and (b) RCP incidence.

    APPENDIX C: ERROR ANALYSIS OF SIMULATION RESULTS VERSUS THEORETICAL RESULTS FOR MOIRé DEVICE 1

    To systematically assess the operational precision of the designed Moiré device 1, we implemented the APE metric to quantify deviations between simulated and theoretical values of focal length and NA. The APE is defined as APE=((|DSimDTheo|)/DSim)×100%.

    Here, DSim represents simulated data and DTheo denotes theoretically calculated parameters, including both focal length and NA. Figure 8 illustrates the APEs with respect to focal length and NA obtained at different rotation angles (including 90°, 120°, 150°, 180°, 210°, and 240°). It can be seen that the focal length simulations exhibit a maximum APE of 30% with a mean value of 25.2%, whereas the NA parameters demonstrate improved consistency with a peak error of 22% and a significant mean APE of 15%. These higher APE values between simulated data and theoretically calculated parameters for device 1 arise due to the complex near-field coupling between layers and the sensitivity of phase control to structural tolerances, which are captured in the simulations but are less directly represented in the theoretical model. Nonetheless, the device exhibits consistent trends and demonstrates the feasibility of the Moiré phase engineering approach for dynamic focusing and polarization control at THz frequencies.

    Angular-dependent APEs for focal length and NA in Moiré device 1, computed from simulation-theory comparisons across six rotational configurations (90°, 120°, 150°, 180°, 210°, 240°).

    Figure 8.Angular-dependent APEs for focal length and NA in Moiré device 1, computed from simulation-theory comparisons across six rotational configurations (90°, 120°, 150°, 180°, 210°, 240°).

    APPENDIX D: LCP/RCP CROSSTALK ANALYSIS IN MOIRé DEVICE 1 IN NUMERICAL SIMULATION

    To evaluate the polarization isolation capability of Moiré device 1, we quantitatively analyze the crosstalk between orthogonal circular polarization channels (LCP/RCP) through numerical simulations. For Moiré device 1, where the primary functionality involves spin-selective phase modulation (e.g., converting incident RCP wave to LCP and vice versa), the crosstalk is defined as the unintended retention or leakage of the input polarization state in the output channel, expressed as Cross_talk(dB)=10log10(PcoPcross),where Pcross denotes the power in the cross-polarized transmission channel (LCPRCP or RCPLCP), and Pco represents the power leaking into the co-polarized state (LCPLCP or RCPRCP). As demonstrated in Fig. 9, the simulated crosstalk values for the orthogonal circular polarization incidence (LCP and RCP) exhibit distinct trends across the rotational tuning range (α varies from 90° to 240°). Under LCP incidence (red bars), which generates a focused Gaussian beam (topological charge l=0), the crosstalk remains consistently below 20.98  dB (0.8% leakage), indicating robust polarization isolation and phase fidelity. Conversely, RCP incidence (blue bars), designed to produce a vortex beam (l=1), shows higher crosstalk levels (up to 11.63  dB, 6.9% leakage), which may result from topological charge dependence and vortex beam sensitivity. Future designs could mitigate this imbalance by optimizing layer-specific phase compensation for vortex generation.

    Crosstalk values obtained at orthogonal circular polarization incidence for different rotation angles (90°–240°); red bars for RCP incidence, blue bars for LCP incidence.

    Figure 9.Crosstalk values obtained at orthogonal circular polarization incidence for different rotation angles (90°–240°); red bars for RCP incidence, blue bars for LCP incidence.

    APPENDIX E: SIMULATION RESULTS OF THE FOCUSED FIELD ON XOZ-PLANE GENERATED BY THE DESIGNED MOIRé DEVICE 2 UNDER LINEARLY POLARIZED ILLUMINATION, AS THE ROTATION ANGLE GRADUALLY INCREASES FROM 90° TO 2

    Figure 10 systematically characterizes the field distributions of Moiré device 2 in the xoz-plane under orthogonal linear polarization states (x-LP and y-LP). Under x-LP illumination [Fig. 10(a)], the simulated field patterns reveal well-defined vortex characteristics accompanied by controlled focal length modulation through rotational manipulation. Notably, progressive angular rotation from 90° to 240° induces a systematic reduction in focal length from 8.62 mm to 3.17 mm. Quantitative analysis demonstrates focal length (F) values of 8.62 mm (α=90°), 6.68 mm (120°), 5.11 mm (150°), 4.14 mm (180°), 3.62 mm (210°), and 3.17 mm (240°), establishing an inverse relationship between rotation angle and focal length. For y-LP incidence [Fig. 10(b)], the device generates focused beams carrying topological charge l=0, while maintaining an analogous focal length dependency on angular rotation.

    Simulation results of Moiré device 2. (a) The obtained electric fields in the xoz-plane as well as the change in focal length as the rotation angle α increases from 90° to 240° at 30° intervals under x-LP incidence and (b) y-LP incidence.

    Figure 10.Simulation results of Moiré device 2. (a) The obtained electric fields in the xoz-plane as well as the change in focal length as the rotation angle α increases from 90° to 240° at 30° intervals under x-LP incidence and (b) y-LP incidence.

    APPENDIX F: INTRODUCTION OF THE OPTICAL ELEMENT GCT-090101 AND THE EXPERIMENTAL DETAILS

    The GCT-090101 component is designed for direct coaxial integration with mounting rods, utilizing four 6 mm through-holes to achieve sub-millimeter alignment precision between dual metasurface layers. The rotational stage enables full 360° angular displacement with 2° angular resolution, while maintaining repeatable positioning at 30° step intervals. In our experimental implementation, metasurface Layers 1 and 2 were mounted in two distinct GCT-090101 modules [Fig. 11(a)]. Precise coaxial alignment between the modules was subsequently established through the rod-based alignment system, ensuring optimal metasurface layer registration as demonstrated in Fig. 11(b).

    Experimental details of Moiré device 2. (a) The GCT-090101 modules embedded in metasurface sample Layer 1 and Layer 2 and (b) coaxially aligned GCT-090101 modules.

    Figure 11.Experimental details of Moiré device 2. (a) The GCT-090101 modules embedded in metasurface sample Layer 1 and Layer 2 and (b) coaxially aligned GCT-090101 modules.

    APPENDIX G: ERROR ANALYSIS OF SIMULATION RESULTS VERSUS EXPERIMENTAL RESULTS FOR MOIRé DEVICE 2

    To quantitatively evaluate the performance of Moiré device 2, we employed the mean APE metric to characterize the discrepancies between simulated and experimental results in both focal length and NA. As illustrated in Fig. 12, the focal length measurements demonstrate a maximum APE of 5.9% with a mean value of 3.4%, while the NA parameters exhibit superior agreement showing a peak APE below 3% and a mean APE of merely 1.4%.

    The APEs between simulated and experimental values for focal length and NA across angular displacements spanning 90° to 240° (30° increments).

    Figure 12.The APEs between simulated and experimental values for focal length and NA across angular displacements spanning 90° to 240° (30° increments).

    Tools

    Get Citation

    Copy Citation Text

    Wenhui Xu, Hui Li, Chenghui Zhao, Jie Li, Qi Tan, Yufei Liu, Hang Xu, Yun Shen, Jianquan Yao, "Polarization-multiplexing metasurfaces for tunable wavefront configurations via Moiré engineering," Photonics Res. 13, 2130 (2025)

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Surface Optics and Plasmonics

    Received: Mar. 5, 2025

    Accepted: May. 5, 2025

    Published Online: Jul. 25, 2025

    The Author Email: Hang Xu (xh_931119@tju.edu.cn), Yun Shen (shenyun@ncu.edu.cn), Jianquan Yao (jqyao@tju.edu.cn)

    DOI:10.1364/PRJ.561412

    CSTR:32188.14.PRJ.561412

    Topics