Perovskite artificial retina can read handwritten numbers
Researchers have built an artificial retina out of perovskite materials that can detect light in a similar fashion to the human eye. In tests, the device was even able to recognize handwritten numbers.
The eyes of humans and other mammals work thanks to photoreceptor cells – the rods and cones – in the retina, which absorb incoming photons and send electrical signals to the brain. For the new study, researchers at KAUST set out to mimic that process in an artificial device.
The key ingredient is perovskite, a material that’s very efficient at absorbing light, and as such is emerging as a frontrunner in the next generation of solar cells. In this case, the team harnessed that light-absorbing prowess to make a sensor instead.
Perovskite nanocrystals were embedded onto a polymer, then that layer was sandwiched between two electrodes – aluminum on the bottom and indium tin oxide on the top. This upper electrode was etched to let light through to the perovskite layer, creating an array of photoreceptors. These are made on a polyimide substrate that lets the device flex and bend into whatever shape is needed, such as that of the human retina.
To process the light input, this photoreceptor array was attached to a CMOS sensor and a neural network with 100 output neurons. In tests with a 4 x 4 array, the device was illuminated using LEDs of various colors and it was found that the optical response was very similar to that of the human eye, with the system particularly sensitive to green light. In other tests, the system was even able to recognize handwritten numbers with an accuracy of 72 percent. It was also very stable, with no change in the response to light after 129 weeks.
Despite its biological inspiration, this artificial retina isn’t likely to be destined for human transplants like some others. The team says that with more development, this device would be used to make more advanced vision systems for cameras and robots.
"The ultimate goal of our research in this area is to develop efficient neuromorphic vision sensors to build efficient cameras for computer vision applications," said Khaled Nabil Salama, corresponding author of the study. "Existing systems use photodetectors that require power for their operation and thus consume a lot of energy, even on standby. In contrast, our proposed photoreceptors are capacitive devices that don't consume static power for their operation.”
The research was published in the journal Light: Science & Applications.