Under-display facial recognition possible with quantum-inspired computational chemistry

Under-display cameras and facial recognition are in high demand as consumers yearn for smartphones with a bezel-free display. Through quantum-inspired computing, OTI Lumionics has developed a material that has made under-display facial recognition a reality. But phones that are currently on the market with under-display cameras have shown varying degrees of success. Biometric Update spoke with VP and Co-founder Jacky Qiu on how its ConductTorr cathode patterning material (CPM) enables under-display face biometrics, as well as areas of improvement for smartphones.

The goal of CPM is to pattern windows in the OLED to allow IR rays to pass through the cathodes, allowing for facial recognition. Cathodes in traditional OLED stacks are “typically highly absorbent in the IR range and semi-absorbent in the optical range,” says Qiu.

Traditional ways of patterning for OLEDs don’t work well for enabling under-display cameras and facial recognition. The process is sensitive to moisture and air, and OLEDs could easily become damaged. Instead, with CPM patterning, a cathode electrode material will be put through a self-assembly process in a high vacuum, forming a film only in the areas without the CPM material.

Qiu says the material functions “a bit like teflon,” which is used to make pans non-stick. “The area where your CPM material is, there is no deposition of metal material,” he says.

Currently, OTI Lumionics’s ConductTorr CPM is the only mass production-ready solution for cathode patterning under-display cameras that would allow facial recognition to be performed. The material was found using the quantum-inspired Materials Discovery Platform.

Chemical synthesis, which uses a traditional trial-and-error approach to discovery, “is incredibly expensive and takes a lot of time,” says Qiu. Using quantum chemistry simulations and AI, the Materials Discovery Platform allows OTI Lumionics to screen candidates quickly. “If we can reduce the number of candidates and increase the accuracy of the materials to the performances that we want, we can do testing a lot faster.” Instead of completing a few thousand simulations in a week, researchers can complete tens or hundreds of thousands of simulations.

While it uses quantum chemistry simulations, the platform isn’t using true quantum computing. “While the algorithms used are designed to be operated on a quantum computer, quantum computing hardware is still not powerful enough to solve the problems that we’re solving. The molecules that we’re dealing with, typically are pretty big molecules that have a lot of electrons. And for each electron, you need a qubit to simulate all this,” explains Qiu.

The early quantum computers that are commercially available don’t have enough qubits to accommodate, he explains. Researchers at OTI “developed a number of algorithms that can optimize the amount of qubits needed and developed novel ways to represent the problems from a quantum computer” which they then input into a supercomputer. They were able to identify the best candidates for materials like CPM.

While the company won’t comment on where their CPM is deployed, the material is currently available to OEMs as more smartphones are being released with under-display facial recognition capabilities.

It’s rumored that Apple will use under-display Face ID after the release of the iPhone 15 models, leaving only a small punch hole for the front-facing camera. And in 2022, OTI Lumionics received $55 million in Series B funding from display manufacturers like LG, Samsung (the world’s top Android phone maker), and United Display Corporation, among other investors.

There are still areas for improvement when it comes to under-display facial recognition. Putting the camera behind the display of a phone minimizes the amount of light that can pass through the screen. While CPM patterning allows for IR rays to pass through, “there will potentially be reduced fidelity.” Users of some devices with the under-display features have also found that face authentication isn’t able to work in low-light settings due to the low transparency of the display panel.

“At minimum, the amount of IR light that hits the sensor would be reduced compared to just a piece of glass in front. But as I said, there is basically significant improvement coming, in our view, for the newer processors,” he says.

In the future, smartphone OEMs may increase the size of the neural engine or the MPU built into the phones, which would improve signal processing for features like face biometrics. This in turn, could mean under-display cameras with facial recognition could yield similar performances to their traditional counterparts, Qiu says.

“The neural engine of the processor increases each year, and especially now that with the advent and popularization of generative AI, that also leverages that same neural engine.” A more powerful processor would also make for better post-image processing for photos, resulting in images that appear more true to life than what current models are producing.

When it comes to IR cameras, increasing transparency is “actually a much more significant issue” compared to blocking diffraction, says Qiu. One could evaluate optical signal performance in under-display cameras through zero-order of transparency, which is a figure determined by taking transparency and diffraction into account, Qiu explains.

OTI Lumionics will present a paper on zero-order of transparency at this year’s Society of Information Display Symposium at San Jose during Display Week in May.

Article Topics

biometrics  |  cathode patterning material  |  facial recognition  |  OLED  |  OTI Lumionics  |  smartphones  |  under display