New k-space superoscillation method breaks the Abbe diffraction limit for far-field, label-free imaging

Tsinghua University researchers use a nonlocal metalens and k-space superoscillation to double the imaging resolution of far-field, label-free systems.

By: AXL Media

Published: Mar 3, 2026, 5:17 AM EST

Source: The information in this article was sourced from Light Publishing Center, Changchun Institute of Optics, Fine Mechanics and Physics, CAS

Overcoming the 150-year-old diffraction limit

Since 1873, optical systems have been governed by Ernst Abbe’s diffraction limit, which dictates that resolution is fundamentally restricted by the light's wavelength and the physical size of the lens aperture. While specific techniques like fluorescence microscopy have previously broken this barrier, they typically require chemical labels and struggle with far-field, single-shot imaging. A new study published in eLight introduces a topology-optimized nonlocal metalens that disrupts classical spatially shift-invariant properties to see beyond these traditional boundaries.

The physics of k-space superoscillation

The breakthrough centers on a concept called "k-space (momentum space) superoscillation." In traditional optics, a lens focuses light based on incident angles in a predictable, linear fashion. However, the Tsinghua team designed a metalens where the rate of change in the transmitted field exceeds the theoretical limit determined by its physical size. This allows the lens to focus a plane wave to a precise spot on the focal plane while maintaining a microscopic spot size that was previously considered impossible for far-field optics.

Experimental validation in the microwave domain

In preliminary microwave experiments, the researchers compared a standard local lens with their new nonlocal metalens. Without any digital post-processing, the local lens could only resolve two points separated by 2.90$\lambda$ (wavelengths). In contrast, the nonlocal metalens successfully resolved points at a distance of 1.38$\lambda$. This indicates that the new method provides a resolution 2.10 times higher than conventional systems. Furthermore, the team reported a focusing efficiency of 2.24%, a significant energy improvement over previous superoscillatory models.