Steerable terahertz beams using surface waves on an active metasurface Authors: Yaseman Shiri, Jeffrey Lei, Yasith Amarasinghe, Hou-Tong Chen, Chun-Chieh Chang, Michael P Lilly, Sadhvikas Addamane, Daniel M Mittleman Published: 2025 ABSTRACT AbstractThe development of dynamic components for controlling wave fronts in the sub-terahertz region of the electromagnetic spectrum has emerged as a frontier research topic for many applications in sensing and communications. One approach which has attracted much attention involves the use of active metasurfaces, tiled arrays of sub-wavelength elements with properties that can be reconfigured via external actuation. In nearly all cases, these metasurfaces are employed as either transmissive or reflective elements, taking advantage of their strong and tunable interaction with free-space electromagnetic waves. These interactions can be significantly enhanced through the use of surface waves propagating parallel to the metasurface array, although very few studies have exploited this option. Here, we integrate a metasurface into the interior of a parallel-plate waveguide in a configuration explicitly designed to exploit this surface-wave geometry. We show that varying the electrical properties of the active metasurface changes the wave vector of the guided mode, and thereby alters the emission angle of radiation out-coupled through a leaky-wave slot aperture. These results, which are consistent with numerical simulations, represent a new approach to broadband beam steering suitable for the sub-terahertz spectral range.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-33364-1.Subject terms: Engineering, Optics and photonics, Physics Introduction The ongoing exponential growth in the volume of global wireless traffic continues to drive research to explore the use of frequencies above 100 GHz for future wireless networks and systems1–3. Such sub-THz wireless links will offer many advantages, including copious bandwidth to support emerging applications requiring high data rates above 1 Gbps4. A key challenge in exploiting these high frequencies lies in the need for high-gain antennas which produce directional beams. These beams must be aimed accurately, and in the case of a mobile receiver, must be actively steered to maintain high link quality5. A variety of strategies have been pursued to implement active beam steering at these frequencies, including phased array emitters6, tunable gratings7, and liquid crystal devices8, as well as planar array devices with active scattering elements such as metasurfaces9,10 and reflectarrays11,12, typically employed in a free-space coupling geometry. At lower frequencies, optically-controlled programmable microwave metasurfaces have been developed for data transmission and dual-channel communications13,14. Another approach uses leaky-wave antennas (LWAs), which can offer the advantage of relying on well-defined and controlled guided waves rather than free-space waves, as well as agile wide-angle steering through tuning of the carrier frequency15–19. Yet, in many cases systems will require wide-angle steering at fixed carrier frequency, which poses a challenge for these LWA methods. Here, we demonstrate electrically actuated beam steering by integrating an active metasurface into the interior of an LWA. This approach preserves the advantages of the LWA while implementing an angular steering strategy that does not rely on frequency tuning of the incident wave. The response of this device is sufficiently broadband to support high data rates without introducing detrimental effects such as beam squint, which can play a key role in limiting the performance of terahertz components20. This initial proof-of-concept result manifests a steering range of a few degrees, and the approach can readily be optimized for a wider angular range, offering a unique new capability for future systems operating in the sub-THz range. Device design Leaky-wave antennas operating in the millimeter-wave and terahertz regimes have attracted considerable attention, as they can provide broad functionality for sensing and communications21–23. The simplest conceptual description of a leaky waveguide is a parallel-plate waveguide (PPWG) consisting of a pair of flat metal plates with a slot opened in one to permit some radiation to leak out into free space (or, in the inverse process, for free-space radiation to be received through the slot and converted to a guided wave). The leakage occurs only if the wave vector of the guided wave is equal to the component of the free-space wave vector that is parallel to the waveguide axis. This results in a phase-matching requirement for coupling between the waveguide mode and free-space mode. For the lowest-order transverse-electric (TE1) mode of a PPWG, the frequency-dependent propagation constant is: where c0 is the vacuum light velocity, f is the carrier frequency, b is the plate separation, and k0 is the wave vector for free-space, k0 = 2πf/c0. With a slot in one plate, radiation can couple from the guided mode to free space with the phase matching constraint that k0 cosϕ = kPPWG. Here, ϕ is the propagation angle of the free-space mode relative to the waveguide propagation axis. Because of the frequency dependence of kPPWG (Eq. 1), this condition results in an angle-dependent emission frequency: This frequency-angle coupling results in a strong angular dispersion, which is different from the behavior of other commonly used antennas. This device can therefore allow for frequency selective beam steering by enforcing a one-to-one relationship between the carrier frequency and the angle of emission of the leaked wave, enabling beam steering through frequency tuning15–19. To use a LWA for beam steering at a fixed carrier frequency, one must incorporate an active element into the (otherwise passive) waveguide device. For this purpose, we employ an active metasurface. Such devices have become increasingly popular in the terahertz range as versatile components for a variety of tasks including wave front manipulation9,24 and enhanced security25. The metasurface is inserted into the interior of the PPWG, between the two metal plates, lying in a groove milled into the lower plate (see Fig. 1). The metamaterial used in this work (illustrated in Fig. 2) is similar in concept to previously demonstrated active metasurface devices26–28. It consists of a planar array of sub-wavelength sized electric split-ring resonators (SRRs). The SRRs are square-shaped meta-elements, with a capacitive gap at each of the four corners. The SRR array is fabricated via lithography on a 2-µm-thick, n-doped gallium arsenide (GaAs) epi-layer (carrier density ~ 0.5 × 1016 cm-3) grown on a semi-insulating GaAs substrate. All of the SRRs are interconnected with metal wires, forming a Schottky contact to the underlying doped GaAs layer. The switching mechanism of this device involves the application of a DC reverse bias between this Schottky contact and the Ohmic ground, which depletes the free carrier population underneath and adjacent to the metal SRRs. In the absence of a bias voltage, the split gaps at the corners of the square SRRs are shorted, effectively creating the “gaps closed” (blue) structure shown in Fig. 2d. In contrast, an applied voltage depletes the charge carriers in this region, effectively creating the “gaps open” (red) structure in Fig. 2d (with gaps at the four corners of each square)26. This latter structure exhibits both a capacitive and inductive response, and therefore manifests an LC resonance at a frequency determined by the size of the SRRs. Because of the connecting wires and Ohmic lines, the frequency of this resonance when measured in a normal-incidence transmission configuration depends on the axis of linear polarization (see Supplementary Material for details). The switching speed of this metasurface is determined by the RC time constant of the Schottky diode and the metal interconnecting lines, and influences the response time of the metasurface to changes in bias voltage, which is important in applications requiring fast modulation, such as real-time beam steering. Schematic of a parallel-plate leaky waveguide with active metasurface for electrical steering of THz beams. The parallel-plate waveguide used in this study has a length of 50 mm, width of 21 mm and plate separation of b = 0.9 mm. The slot in the top waveguide plate has a width of 1 mm and a length of 30 mm. The active metasurface integrated into the bottom plate has a length of 30 mm. Changing the DC bias applied to the metasurface in the interior of the waveguide steers the leaky-wave output beam. Diagram of active terahertz metasurface. (a) Overview of the active metasurface used in this study. The elements of the device are grouped into column-shaped pixels, each with an independent bias pad, similar to ref 27, although in all the measurements reported here the bias is the same on all pads. (b) Detailed view indicating columns of electrically connected resonators, connected to a Schottky pad for electrical control. (c) Diagram of a single resonator showing the capacitive gaps at the corners. (d) Schematic geometry of a split-ring resonator in the two idealized cases, when the capacitive gaps are fully closed (corresponding to zero voltage applied to the Schottky contact), and fully open (maximum reverse bias voltage of −12 V applied). The resonators have a periodicity of 125 μm, a SRR size of 100 μm, a gap width of 2 μm, and a metal line width of 6 μm. Measurements As shown in Fig. 1, our prototype LWA consists of two parallel stainless-steel plates, with a rectangular slot in the upper plate through which a portion of the guided electromagnetic wave can leak. A groove is milled into the lower plate of the waveguide to accommodate the metasurface, which is fabricated on a 0.625-mm-thick substrate. This ensures that there is no height discontinuity between the top of the lower plate and the metasurface array. Thus, a guided wave coupled into the waveguide first propagates through a short region in which both the upper and lower boundaries are metallic, and then crosses a smooth (no height discontinuity) boundary to a second region with the same air gap, where the lower boundary is the metasurface array. In this region, the boundary conditions experienced by the guided wave change, and thus its wave vector is altered (i.e., modifying Eq. 1). We use this device to demonstrate the feasibility of this approach to electronic beam steering of the radiation emitted through the rectangular slot. The experiments are performed using a fiber-coupled terahertz time-domain spectroscopy system with photoconductive antennas as the source and detector for broadband characterization29. Two polyethylene lenses are placed after the transmitter to couple the radiation into the waveguide, which is further optimized using a tapered input flare30. The input beam is polarized parallel to the surfaces of the metal plates, in order to excite TE waveguide modes. By careful mode matching, we predominantly excite the TE1 waveguide mode at the input end of the waveguide31. However, when the propagating TE1 mode reaches the region containing the metasurface, the abrupt change in structure and asymmetry in the waveguide disrupts this mode, leading to coupling to numerous hybrid modes. The slot aperture in the upper plate begins 10 mm beyond the transition from the metal-metal waveguide to the metasurface region. Our experiments consist of exciting the waveguide with a broadband impulse, and measuring the spectrum of the output signal. We first measure the signal transiting through the waveguide without a slot aperture. These results (see Supplementary Material) indicate that, in addition to several narrow resonances, the phase of the signal at all frequencies varies with applied bias. Next, we characterize the signal emitted through the slot aperture, via far-field measurements with a receiver on a rotating rail. We characterize this signal as a function of emission angle and reverse bias voltage applied to the metasurface, ranging from 0 V (in which the capacitive gaps are effectively fully shorted) to the maximum reverse bias of − 12 V (in which these gaps are effectively open circuits). Figure 3 shows typical results for two cases at 250 GHz and 350 GHz, which are representative of the behavior at all measured frequencies. The blue curves correspond to the case of zero voltage applied to the metasurface (gaps closed), while the red curves correspond to − 12 V (gaps open). The results indicate a small but notable angular shift, verifying the feasibility of our approach. Angular dependence of the measured THz signal. The data points show the magnitude of the THz field emitted through the slot aperture, measured in the far field, as a function of emission angle, in the case of zero volts applied to the metasurface (blue points, capacitive gaps effectively closed) and −12 volts applied (red points, gaps effectively open). The left panel shows the results for 250 GHz, while the right panel shows the results for 350 GHz. These two frequencies are representative of the behavior at all measured frequencies. The vertical dashed black lines show the angles that would be predicted for a simple metal-metal leaky-wave aperture, as given by Eq. (2). The inset in the left panel depicts a schematic of the measurement, in which the transmitter is fixed while the receiver is mounted on a rail to allow angular positioning. Numerical simulations To assist in the interpretation of our results, we perform extensive computational simulations using the Finite Element Method (FEM). These simulations incorporate the metasurface realistically, by modeling the same SRR geometry as used in the experimental device. In these calculations, we consider two limiting cases, in which the capacitive gaps at the corners of the SRR are either fully closed (shorted) or fully open. Evidently, the corresponding experimental limits (zero and −12 volts applied to the Schottky contact, respectively) do not correspond precisely to these idealized conditions; however, these are reasonable approximations of the realistic device characteristics which are valuable for understanding the performance. Using this idealization, we first compute the dispersion diagram for the guided modes, for the section of waveguide containing the metasurface. These calculations ignore the slot aperture and employ a solid metal slab (perfect electric conductor) for the upper plate, with an air gap of 0.9 mm (as in the experiments), and also include a 1-mm-thick GaAs substrate. These results, shown in Fig. 4a, indicate a complex dispersion diagram, exhibiting both resonant (non-dispersive) modes and dispersive modes which resemble conventional waveguide TE modes (see Supplementary Material). As a result, a complex superposition of modes can be formed in this region of the waveguide, containing some elements with strong coupling to the metasurface (and therefore a notable dependence on the applied bias). Due to the large number and density of modes, this mixing can occur at any frequency. Although each frequency will exhibit its own unique mode mixture, it is clear from the results shown in Fig. 3 that, at all frequencies, the superposition contains a strong characteristic of the fundamental TE1 mode from which the mixture is excited (due to mode matching), as well as significant contribution from modes with strong surface wave characteristic which are therefore sensitive to the bias applied to the metasurface. Computed dispersion diagrams and surface modes. (a) Simulated waveguide modes are shown for the case in which the capacitive gaps in the SRR elements are closed (blue points) and open (red points), corresponding approximately to the experimental situations of zero DC bias and −12 V reverse bias applied to the metasurface, respectively. A large number of guided modes are observed in the region where the metasurface wafer is the lower boundary of the waveguide. The TE modes of the original metal-metal structure are evident, and are most strongly excited due to mode matching. Other modes include a series of TE-like modes originating from the air + substrate gap, as well as mixed modes with significant surface wave characteristics. (b) Images of the surface wave portion of the mode mixture at a representative frequency of 350 GHz, showing the strong modulation of this part of the mode upon application of a bias to the metasurface array. These images, showing the electric field pattern in the plane of the metasurface, were extracted from the full 3D numerical simulations described in the text. A second set of FEM simulations are used to model the far-field radiation emitted from the slot of the waveguide. As above, we use these simulations to compare the two situations in which zero bias and −12 V bias are applied to the Schottky contact. We construct a 3D model of the waveguide device, including a full array of meta-elements on the GaAs substrate. The SRR geometries are adjusted to either contain capacitive gaps (corresponding to reverse bias case) or not (corresponding to zero bias). To account for effects of DC bias outside of the gap regions, we also adjust the average epi-layer conductivity in the simulation to either 400 S/m (when zero bias is applied) or 50 S/m (when −12 V is applied); these values are obtained from previous measurements which characterized the average surface impedance of a similar device32. The total length of the simulated waveguide is 50 mm, with a 21 mm region in the center that contains meta-elements. Due to the electrically large nature of the structure, adaptive mesh refinement was used, with the final structure having ~ 5 million mesh elements. We run this simulation at 30 different frequencies, ranging from 150 GHz to 450 GHz for both the zero-bias and reverse-bias cases. From these simulations, we can examine the nature of the superposed mode at any frequency. An example showing the surface wave character at one typical frequency is shown in Fig. 4b. We can also extract the angle of maximum emission intensity at each frequency. These simulated results are compared to the equivalent experimental data in Fig. 5. The experimental data and simulation results are in good agreement with each other, showing that the effective surface impedance of this waveguide boundary is sensitive to the DC bias over a wide frequency range, and is not a resonant effect associated with the structure of the SRR elements. The small shift between the measured and predicted curves may be due to minor fabrication tolerances such as dimensional variations in the depth of the groove, metasurface thickness, and transition between metasurface and metal boundaries within the waveguide. While these factors can lead to minor differences between experimental and simulated results, they do not qualitatively impact the overall performance of the device. We note that, although these simulations only consider the two limiting cases corresponding to minimum and maximum applied voltage, intermediate values would result in intermediate tuning angles, as the surface impedance varies smoothly and monotonically with applied DC bias. This is consistent with the measured results of the spectrum transmitted through the waveguide at intermediate bias values (see supplementary materials). Simulated and measured results of far-field terahertz radiation. The dotted points represent measured maximum emission angle at each frequency. The solid lines represent simulations using a 3D model to extract the radiation pattern emitted through the slot aperture. The dashed black line represents the prediction of Eq. (2), which ignores the effect of the metasurface. The inset shows the difference between the blue (“gaps closed”) and red (“gaps open”) cases, both in simulation (solid curve) and measurements (points). Since these simulations account for the full structure of the metasurface, we can analyze insertion loss as well. For example, at a frequency of 350 GHz, the power of the main lobe decreases by ~ 2 dB when the metasurface is used in the waveguide versus when the waveguide is simply two flat metal plates. The inclusion of the GaAs substrate introduces insertion loss primarily due to material (free carrier) absorption in the doped epi-layer, and the waveguide asymmetry which also gives rise to mode conversion. This moderate insertion loss predicted by simulation is expected in doped substrates and is consistent with our experimental results. Discussion and conclusions We have presented a proof-of-concept design of an active metasurface integrated into a leaky waveguide to create an electrically switchable THz emitter. This provides a new approach to beam steering functionality. Since the DC voltage is applied to the metasurface in a reverse bias, the switching mechanism draws almost no current, and is therefore power-efficient. Further, the spatial pattern of the leaked wave is entirely determined by the dimensions of the aperture in the metal plate, which is independent of the switching mechanism, and can be optimized for high antenna gain. As a result, this device has the advantage that beam steering is independent of beam shaping (i.e., antenna gain), which is often not the case for other approaches to beam steering. Clearly, the relatively small range of steering angles limits the use cases for this present version of the device. However, we foresee a great deal of optimization of the metasurface and waveguide, which could dramatically increase this steering range. For example, we can further optimize the interaction between the guided wave and the switchable surface, without introducing unacceptable propagation losses, by varying the device geometry or incorporating a metasurface with different resonator geometries. A smaller plate separation in the waveguide would be advantageous because it supports a single TE mode at higher frequencies, and the propagation remains single-mode even when the wave is in the asymmetric region of the waveguide, with a metasurface on one boundary. Decreasing the waveguide gap leads to additional challenges in coupling, which can be managed through careful design of the input port. We also note that in metasurface devices such as the one used here, both the carrier density and the epi-layer thickness can be chosen to optimize the modulation range for a given frequency. This procedure is typically implemented for the case of normal-incidence transmission, which is relevant for free-space coupling. However, in our work, we consider a different geometry in which the metasurface forms one boundary of a waveguide, such that the wave vector of the terahertz wave is parallel to the plane of the metasurface. This configuration can be further optimized by modifying the physical structure of the device so that the incident beam is polarized perpendicular to the connecting wires of the split ring resonators, which may lead to stronger modulation and hence substantial improvements in device performance. The experiments and simulations described here indicate the potential for beam steering over a broad bandwidth with low power consumption, making this an intriguing new option for dynamic wavefront engineering in terahertz systems in both terrestrial and non-terrestrial settings. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Author contributions Y.S. contributed to the project ideas, data collection and data analyzation. J.L. and Y.A. contributed to the project ideas and data analysis. H.T.C., C.C.C., M.P.L., and S.A. designed and fabricated the metasurface. D.M.M. conceived and supervised the project. Y.S. and D.M.M. led the writing of the manuscript with contributions from all co-authors. Funding This work was supported in part by the NASA Rhode Island Space Grant Consortium, the U.S Navy Naval Innovative Science and Engineering Program project 219-BAR-028, the Department of Defense SMART Scholars Program, the U.S Air Force Office of Scientific Research grant FA9550-022-1-0412, and the U.S National Science Foundation grants NSF-1954780 and NSF-2224210. This work was performed, in part, at the Center for Integrated Nanotechnologies, an Office of Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science. Los Alamos National Laboratory, an affirmative action equal opportunity employer, is managed by Triad National Security, LLC for the U.S. Department of Energy’s NNSA, under contract 89233218CNA000001. Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International, Inc., for the U.S. DOE’s National Nuclear Security Administration under contract DE-NA-0003525. The views expressed in the article do not necessarily represent the views of the U.S. DOE or the United States Government. Data availability Data reported in this manuscript are available from the corresponding author upon reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Yaseman Shiri, Email: yaseman_shiri@brown.edu. Daniel M. Mittleman, Email: daniel_mittleman@brown.edu References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement Data reported in this manuscript are available from the corresponding author upon reasonable request.