A research team led by Ulsan National Institute of Science and Technology (UNIST) has developed a new Meniscus Pixel Printing (MPP) technique that enables the direct, mask-free patterning of perovskite photodetectors onto contact lenses - paving the way for ultralight, eye-mounted extended reality (XR) systems and hands-free robotic interfaces.
(a) Schematic of the MPP. (b) Optical images of the MPP process with a 100 µm nozzle on the substrate. The scale bar is 200 µm. (c) Conceptual illustration of dwell time-dependent dot sizes control. (d) Schematic of the Solution-mediated perovskite crystallization pathway following MPP. (e) The optical images show the crystallization during the annealing process. The scale bar is 5 mm. (f) SEM image of the resulting perovskite layer. The scale bar is 10 µm. Image from: Advanced Functional Materials
Integrating light sensors into a contact lens remains a challenge. Traditional lithographic and inkjet methods struggle to conform to the steep curvature of a lens surface and demand costly, multi-step processing. The team’s MPP approach addresses these obstacles by harnessing a self-confined liquid meniscus formed at the tip of a micro-pipette. In this configuration, the pipette briefly touches the substrate, forming a stable ink bridge that deposits a methylammonium lead iodide (MAPbI₃) perovskite dot with precise size control governed by dwell time and retraction speed.
This “touch-and-release” process enables the formation of 200 µm pixels within approximately one second - eliminating the need for masks, vacuum steps, or high-voltage fields. The researchers demonstrated reproducible pixel sizes between 200 and 700 µm and successfully printed onto surfaces curved up to 63.3°, with radii down to 8.6 mm - comparable to the human cornea. Unlike previous extrusion or electrohydrodynamic methods, MPP maintained sharp feature boundaries without solvent spreading or positional drift.
Using MPP, the team fabricated a 10 × 10 perovskite photodetector array (100 total pixels) on an 8 mm indium tin oxide (ITO) substrate - a configuration suitable for integration into a contact lens. To ensure durability, the device was encapsulated with a photocurable resin, achieving stable operation under visible light and retaining 92% of its initial photocurrent after two months of storage. Lead-leakage tests under repeated mechanical stress detected no release above 2 parts per billion, an essential safety benchmark for ocular electronics.
To resolve the resolution bottleneck imposed by limited pixel count, the team integrated an AI-driven super-resolution (SR) framework. A custom deep learning model based on a Super-Resolution Generative Adversarial Network (SRGAN) reconstructed 80 × 80 optical data from 10 × 10 sensor inputs, achieving 97.2% classification accuracy with a processing latency of just 0.03 seconds. Training was performed using 35,000 augmented samples representing basic geometric patterns, allowing the network to infer fine edges and spatial details beyond the physical sensor grid.
In a practical demonstration, the AI-enhanced photodetector array was combined with an eye-tracking system that recognized nine distinct eye gestures (eight directional movements plus blinking) with 99.3% accuracy. The gaze data were used to control a robotic arm in real time, enabling object manipulation through simple visual commands. Even when input resolution was reduced to a sparse 5 × 5 array, the SR algorithm restored performance close to the original 100-pixel configuration.
The MPP strategy thus offers both fabrication simplicity and computational intelligence - a scalable pathway toward miniaturized optoelectronic vision systems. With further refinement in readout electronics and long-term biocompatibility testing, this perovskite-based contact lens approach could form the foundation of next-generation wearable XR displays, vision-assisted robotics, and adaptive human–machine interfaces.
