Researchers have shown that newly designed pixel-scale metasurface lenses -; flat surfaces that use nanostructures to manipulate light -; can be used to make imaging sensors that are roughly three times more sensitive than those used today. The new sensor architecture could enable digital cameras that can image faster or in conditions with less light.
"Traditional imaging sensors such as the ones used in smartphones, wearable devices and autonomous vehicles have a limited sensitivity because they rely on color filters placed over each pixel," said research team leader Masashi Miyata from NTT Device Technology Labs in Japan. "Our new metalenses are made from a highly-engineered surface that can collect light while simultaneously separating primary colors without any color filters, opening a pathway to dramatically improve sensitivity."
In Optica, Optica Publishing Group's journal for high-impact research, Miyata and colleagues report that filter-free color sensors made with the new metalenses significantly enhanced signal levels without sacrificing color image quality or spatial resolution. And because the new metalenses are made using a CMOS-compatible process, they could easily be integrated onto current sensors to create filter-free imaging devices.
"We envision our metalenses playing an important role in the development of filter-free color image sensors that exceed current sensitivity limits," said Miyata. "These new sensors could one day let people more easily capture night views with smartphones or enable new cameras that accurately capture high-speed objects, which will be useful in security and autonomous driving."
In a conventional sensor, color information is acquired by using color filters that absorb a portion of the light. A red filter, for example, lets through only red wavelenghts while absorbing all the other wavelenghts. This means that only about 30% of the light is actually detected.
To boost sensitivity, the NTT researchers designed a metalens array that acquires color information without optical loss through a process known as color sorting. This involves splitting the light into red, green and blue and then focusing each color onto different pixels. The pixel-scale metalens array was created by etching nanoposts into a 1250-nm-thick layer of silicon nitride.
Although other pixel-scale color splitters have been experimentally demonstrated, they haven't been practical for consumer devices because they were either inefficient, affected by the light's polarization or sensitive to light that might hit the sensor from an oblique angle. The new metalenses, however, are based on a dispersion-enriched metasurface platform that makes them polarization-insensitive and suppresses spectral crosstalk for all the color pixels. Because the metalenses are so efficient at focusing light, their color sorting performance isn't affected by oblique light.
Evaluating Sensor Performance
The researchers used an optical microscope to mimic the way that light would travel through a metalens array before reaching a sensor. This experiment showed that, compared with a filter-based sensor, the metalens-based sensor generates color images with 2.83-fold enhanced signal levels without sacrificing color quality.
Optical simulation studies also showed that the metalens-based sensor architecture exhibited less image degradation due to sensor noise, which is often the limiting factor in dark-scene or ultra-fast imaging. Now that they have demonstrated the new sensor concept, the researchers plan to create and test an integrated device by directly mounting a metalens array onto an image sensor.
"We hope our work will further boost the development of practical optical devices and systems based on metasurfaces," said Miyata. "With their ability to flatten and shrink optical components while drastically enhancing performance, we believe that optical metasurfaces can be applied not only to image sensors but also to various optoelectronics devices such as those used in displays, projectors and augmented or virtual reality devices."