First-hand information for everyone
Phased-array antennas steer beams by adjusting the phase of many radiating elements. But physics places a hard ceiling on how narrowly and how far a beam can be steered: diffraction limits the angular range over which a fixed-size aperture can produce a highly directive (narrow) beam. A new theoretical and experimental study on arXiv (arXiv:2609.30060v1) maps those limits for systems that pair a planar phased array with a passive dielectric lens — and shows that a class of “nonlocal” metalenses can operate near those bounds.
The authors derive what they call angular-channel bounds: geometry-independent limits that quantify how much a passive, linear phased-array–lens system can improve beam directivity or effective aperture width without shrinking the array’s scan range. Importantly, their analysis separates two distinct operating regimes. In one regime a lens can enhance scan resolution (narrow the beam) over some angles; in the other it can boost peak directivity while keeping the original scan range unchanged. These regimes are set by diffraction physics and the role of evanescent (near-field) waves in carrying subwavelength information across the aperture.
A central conceptual point of the work is the use of nonlocal dielectric metalenses. Conventional metalenses are typically designed to suppress mutual coupling between elements and to apply a locally specified phase shift to each ray. By contrast, the nonlocal approach intentionally exploits electromagnetic interactions across the lens to laterally redistribute energy on the aperture. That redistribution expands the spatial width over which the aperture fields are phase-coherent, effectively creating a larger coherent aperture from the same physical source when the array is driven with conventional progressive phasing.
The paper analyzes how diffraction and evanescent components limit the maximum coherent-aperture expansion, and from that derives compact, geometry-independent design equations. To translate the bounds into realizable devices the authors developed an adjoint-based inverse-design framework for passive linear metalenses and used it to produce both planar and cylindrical lens designs that approach the theoretical limits.
The team also reports experimental validation of a compact cylindrical metalens. In the experiment the metalens preserved a 50° scan range and the scan resolution of a 3.5 λ0–wide phased array, while increasing peak directivity by about 2.5–3.5 dB. Those measured gains are modest but meaningful: a few decibels of directivity can translate to appreciable improvements in link budget for wireless links or radar detection sensitivity, without requiring changes to the driving electronics or array layout.
The study is cautious about what the limits mean in practice. The bounds assume passive, linear components and uniform array excitation with linear progressive phasing; active or nonlinear schemes, or nonuniform feeds, could change the trade-offs. Likewise, fabrication complexity, bandwidth, loss, and sensitivity to manufacturing tolerances are engineering challenges the paper does not eliminate, though the experimental demonstration shows the concept can be practical at least in a cylindrical implementation.
Why this matters: many applications — from phased-array radar and satellite communication to emerging wireless systems and lidar — seek wider scan ranges and higher directivity without increasing array size or complexity. The angular-channel framework gives designers a principled way to assess what a passive lens can and cannot do, and the nonlocal metalens concept points to a practical route for squeezing more performance out of existing arrays.
The work contributes both fundamental limits and a design methodology. If future studies extend the approach to different frequency bands, nonuniform excitations, or active metasurface elements, engineers may be able to further tailor the trade-offs between scan range, resolution, and directivity. For now, the combination of theoretical bounds, inverse design, and experimental proof-of-concept makes a strong case that nonlocal dielectric metalenses are a promising tool for diffraction-limited wavefront engineering in phased-array systems.