A New Optical Technique Breaks the Diffraction Limit Using Momentum Space - Without Post-Processing

In 1873, Ernst Abbe formulated one of optics' most durable constraints: the resolution of any imaging system is limited by the wavelength of light and the numerical aperture of the lens. This diffraction limit has shaped the design of microscopes and telescopes ever since, encoding the principle that finer spatial detail requires either shorter wavelengths or larger collecting apertures - and that there is a hard floor below which no conventional lens can see.

Breaking that floor has been a long-standing challenge. The 2014 Nobel Prize in Chemistry recognized fluorescence-based superresolution microscopy techniques that circumvent the Abbe limit - but those methods require labeling the sample with fluorescent markers, work only in particular biological contexts, and rely on computational reconstruction from multiple exposures. They do not represent a general solution for far-field, label-free imaging of arbitrary samples.

A team led by Associate Professor Yuanmu Yang at Tsinghua University's Department of Precision Instrument proposes a different approach, publishing their results in eLight. They call it k-space superoscillation.

The Physics of Nonlocal Metalenses

Classical imaging systems are spatially shift-invariant: a lens treats each point of incoming light the same way, regardless of angle. This property is mathematically convenient but it is also what produces the diffraction limit. Yang's group designed a metalens - a flat optical element with engineered nanostructures - that deliberately breaks this shift-invariance.

Their lens has topology-optimized responses in both real space and momentum space (k-space). When a plane wave arrives at a given incident angle, it focuses to a spot at twice the distance from the optical axis that a conventional lens would produce - while maintaining the same focal spot size. This effectively doubles the angular resolution without requiring a larger physical aperture.

The concept of superoscillation - in which a function oscillates locally faster than its highest Fourier component - has been explored in optics before, but previous implementations in real space generated image-plane sidebands that complicated practical use. The k-space version, the Tsinghua team argues, avoids this problem, producing "advantages in field of view, energy efficiency, and robustness against disturbances."

Experimental Results in the Microwave Domain

The researchers validated their approach experimentally in the microwave domain, where the nanostructures required for optical wavelengths can be fabricated more readily at larger scales. The standard local lens resolved two points separated by 2.90 wavelengths. The nonlocal metalens resolved two points separated by 1.38 wavelengths - a resolution improvement of 2.10 times. The experimentally measured focusing efficiency was 2.24%, which the team characterizes as "significantly higher than real-space superoscillatory systems with comparable superresolution ratios and fields of view."

These results were achieved without any post-processing or computational reconstruction - a single-shot deterministic measurement. That is a meaningful practical distinction from fluorescence-based methods.

The current prototype operates in the microwave domain, with obvious applications in direction-of-arrival estimation, millimeter-wave imaging, and astronomical surveys. Translating the approach to optical wavelengths would require fabricating the metalens structures at nanometer scales - technically demanding, but the researchers characterize it as achievable with continuing advances in nanofabrication.

Several open questions will determine how broadly this technique can be applied. The current experiments use relatively simple test targets; how the technique performs on complex, realistic samples has not yet been demonstrated. The efficiency figure of 2.24% means most of the incoming light energy is not being used, which could limit performance in low-signal applications. Independent experimental replication at optical wavelengths will be the key test of whether the approach delivers on its promise beyond the microwave demonstration.