Table of Content
When we take a photo with our phone, put on VR glasses, or see LiDAR scanning in vehicles, we are actually dealing with "lenses." Over the past century, lenses have evolved from single-element spherical lenses to multi-element lenses, thinner Fresnel lenses, and finally, diffractive optical elements, all in an effort to correct aberrations. In recent years, nanotechnology has ushered optics into the "metalens era": metalens manipulate phase using subwavelength structures, allowing lenses to be designed like ICs and fabricated like thin films. Today, let's learn about this product hailed as the next generation of optical technology—the "metalens"!
The evolution of lenses: from refraction to diffraction.
The earliest humans used spherical lenses: glass was ground into a rounded shape, and light was focused by refraction. Its advantages were that it was easy to make and reliable; however, once higher image quality and larger apertures were pursued, aberrations (spherical aberration, coma, field curvature) emerged, and they could only be canceled out by "stacked multiple elements". As a result, lenses became thicker and heavier, and the cost and size could not meet the needs of mobile devices.
So engineers began to think: since thickness is a burden, is there a way to thin the same "curved surface"? This led to the development of Fresnel lenses. They cut continuous curved surfaces into concentric steps, much like cutting a "thick glass ramp" into "thin steps." Thickness and weight were reduced dramatically, benefiting projectors, car headlights, and thin magnifying glasses. However, the scattering and ghosting caused by the edges of the steps limited high-quality imaging and contrast in strong backlighting—thinner, but not "clean" enough.
Next, optical designers took it a step further: since we've already worked on microstructures, could we move beyond relying on curved glass and directly use tiny structures to control how light travels? This is where diffractive optical elements (DOEs) come in. DOEs use fine patterns to distribute light to the desired phase distribution, achieving unprecedented thinness. Simultaneously, utilizing the dispersive properties of diffraction, it's possible to design a "refractive + diffraction" achromatic combination, allowing chromatic aberration to be better canceled in certain wavelengths. However, DOEs are mostly single-layered, discrete-phase, and still sensitive to white light; once the field of view and numerical aperture (NA) increase, efficiency and off-axis image quality become new bottlenecks. Thinner and better able to "compensate chromatic aberration," but broadband, high field of view, and low stray dispersion haven't been achieved all at once.
Finally, semiconductor nanotechnology has propelled optics into the metasurface era: if each tiny unit is considered a "nanoant," its phase (including geometric and resonant phase) can be continuously controlled by its size, shape, and rotation angle. This is the metalens. It inherits the DOE's approach of "manipulating light through structure," but elevates phase control from a "stepped" to "continuous and high-resolution." Therefore, it is not only thin but also integrates functions such as focusing, polarization control, achromatic aberration, and aberration compensation onto a single chip. At the same time, because it is a planar structure, it can be arrayed on a wafer like a chip and directly integrated with sensors or waveguides, solving the old problems of size, assembly, and alignment in traditional lens arrays.
Traditional lens > Reduced to non-Nier lens > Diffractive lens
What is a metalens: a nano-antenna that controls the phase of light
Traditional lenses rely on material refraction, while meta-lenses rely on geometric and resonant phase. Imagine an ultra-thin "optical circuit board" covered with tiny antennas (approximately 50–300 nm in size). When light encounters each antenna, it is "delayed" by a different phase due to differences in shape, size, and rotation angle, much like giving the light a slight "lateness" on each pixel. By designing the entire board with a specific phase distribution, it can focus light like a traditional lens.
Common materials: TiO₂ and SiN are commonly used for visible light; a-Si and GaN are commonly used for near-infrared light; the substrate can be glass, quartz or plastic film.
Applications of super-sensitive lenses
- Mobile phones/wearables: ultra-thin sensing (near-infrared/ToF/3D face recognition), micro-projection, color filtering + focusing integration.
- AR/VR/MR: Coupled with pancake or waveguide, shortening lens stacking, reducing front-end weight and thickness, and improving wearing comfort.
- Automotive/Industrial LiDAR: Beam shaping, homogenization, DOE replacement; efficiency is relatively mature in the near-infrared band.
- Biomedicine and Microscopy: Ultra-thin endoscopes, portable microscopes, chip-scale microscopes, which are suitable for disposable or single-use devices.
- Silicon Photonics/Data Communications: On-chip coupling, beam splitting, and wavefront shaping provide a more compact coupling interface for CPO/short-distance optical interconnects.
- Laser processing/projection: beam focusing, shaping, and homogenization to improve energy utilization.
How it's made: Microfilm and Nanoimprinting
Lithography—High Precision, Design Freedom
- Master pattern/direct writing: writing nano-patterns on the resist using an electron beam (E-beam) (highest resolution, slowest speed); or using a DUV stepper/multiple exposure to increase production capacity.
- Development: Developing nano-patterns.
- Thin film deposition/dry etching: Dielectric layers such as TiO₂ and a-Si are deposited, and the pattern is transferred to the functional layer using ICP-RIE.
- Encapsulation/coating: Adding anti-reflective and protective layers, cutting into individual pieces or leaving them on the wafer.
Advantages: High resolution, precise control of shape parameters, suitable for R&D and initial mass production.
Limitations: High cost and limited throughput; stringent requirements for layout and uniformity when using large apertures and high field of view.
Nanoimprinting—High throughput, low cost
- Making "rigid master mold": High-precision master molds are usually made on quartz/silicon substrates using E-beam.
- Replicating the "working mold": using high-hardness PDMS (h-PDMS) or composite materials to create a flexible mold that can be repeatedly imprinted, balancing detail and durability.
- UV-curable resin coating: spin-coating or coating onto glass/plastic substrates.
- Imprinting + UV curing: The working mold is pressed into the resin and fixed by UV exposure.
- Demolding + Post-etching: After demolding, remove the residual layer with a light RIE, and transfer it to a high-refractive-index material if necessary.
Advantages: Wafer-level/large-area mass production, cost-friendly; compatible with existing WLO (wafer-level optics).
Limitations: mold life and contamination control, residual layer uniformity, warpage/shrinkage error, alignment and large-diameter flatness; mass production requires strict quality assurance and embossing equipment.
| Items | Microlithography | Nano Imprinting (UV-NIL) |
|---|---|---|
| accuracy | High, relatively stable | Low, affected by "demolding, air bubbles, residual layer, and mold life". |
| Cost per unit | 高 | Low cost, low material consumption |
| speed | Slow, many process stations | Fast, high scalability for mass production |
| roughness | Excellent, lowest scattering | Resin/transfer etching leaves rounded corners and RLT residue, making the edges more prone to scattering. |
| Material selection | Directly in low-loss, high-refractive-index dielectrics such as TiO₂/SiN/a-Si | Less common, mostly polymers |
| upper limit of efficiency | High potential (high breakage + low loss + straight wall) | Pure resin has low n and low absorption, but its phase efficiency is limited; adding a transfer etching process can improve this. |
| Design iteration speed | The modified version directly re-explodes/re-re-enlarges | Changes require rebuilding the master mold, resulting in slow iteration speed. |
| Typical defects | Over-etching, bias etching | Column pull-out, collapse, bubbles, residual layer thickness, mold contamination/wear |
▲ Comparison Table
Technical barriers
- Broadband achromatic: Single-layer structures are sensitive to wavelength, and white light imaging is prone to dispersion; multi-layer/multi-material structures can improve this but the manufacturing process is more complex.
- Efficiency and off-axis performance: Diffraction efficiency and aberration control under large NA and wide field of view remain challenging.
- Large-diameter yield: Uniformity, stress, pattern defects, and particle control affect yield and consistency.
- Materials and weather resistance: Encapsulation issues such as low-loss visible light materials, environmental stability (temperature, humidity, UV) and scratch resistance.
- Cost structure and ecosystem: Compared with the mature glass-plastic mirror industry, injection molding and coating supply chain, the scale, standardization and design toolchain are still under construction.
Meta-lenses aren't magic; they bring optics into semiconductors, digitally allocating phase using nanostructures. Their value lies in their thinness, lightness, integrability, and programmability, condensing the functionality of multiple lenses into a single device, making them particularly suitable for space-constrained and highly integrated applications. In the short term, the most pragmatic path is hybrid and specific wavelength/application (such as near-infrared sensing, AR/VR, on-chip coupling, etc.). In the medium to long term, with breakthroughs in nanoimprint lithography and materials, and the maturation of design automation, "on-chip optics + metasurfaces" will push imaging, display, communication, and sensing towards entirely new form factors and cost curves. The next time you see news about "mass production of meta-lenses," "NIL adoption," or "achromatic metalens," it basically signals that next-generation camera modules, AR glasses, automotive sensing, or CPOs are moving towards thinner, faster, and more energy-efficient designs.
