This schematic shows the human visual-visual perception process. (Image Credit: Kumar, D., Joharji, L., Li, H. et al.)
Sensory nodes that generate huge amounts of analog data that converts into digital data before being sent to other devices to perform tasks have greatly expanded thanks to AI and IoT. The von Neumann architecture containing discrete devices leads to data access and analysis delays consuming a large amount of power. That poses a problem for applications with rigid delay and power requirements, including robotics and autonomous vehicles. Scientists at King Abdullah University of Science and Technology (KAUST) and Khalifa University developed a sensing-storage-processing node using a two-terminal solution processable MoS2-based MOS device.
Their device implants a light-sensitive 2D material-based charge-trapping layer that replicates the human vision system. It also has storage, processing, and optical data-sensing capabilities. After exposure to light, the device stores the light's wavelength and intensity. This is a different approach compared to standard devices using a photosensor that detects the intensity/wavelength and converts it into a digital domain via an analog-to-digital converter before sending that data to memory for storage.
The fabrication process of the team's light-sensitive MOS device. (Image Credit: Kumar, D., Joharji, L., Li, H. et al.)
According to the team, their device demonstrated a "memory window of approximately 2.8 V with an operating voltage of +6/-6 V, high-temperature retention (100°C) for 10 years, and excellent endurance (106 cycles) without any deterioration." The device underwent a large memory window shift from 2.8 V to over 6 V when the optical light of varying wavelengths stimulated for two seconds as the program ran. That basically means the device can detect light and store it in the same node.
"a) Schematic of the device with optical light illumination, b) C–V curves of the device using different optical light wavelengths from 600 to 400 nm with an interval of 50 nm, c) wavelength-dependent threshold voltage of the device, d) repeatability of C–V curves for 50 continuous cycles with optical programming (+6/−6) and electrically erasing (−8/+8), e) optically programmed and electrically erased endurance of the device with illumination at a wavelength of 400 nm and f) high-temperature retention stability of the device." (Image Credit: Kumar, D., Joharji, L., Li, H. et al.)
"The effect of the number and duration of electrical and optical pulses on the memory window suggested that these devices can also mimic the perceptual learning of the human visual system. To confirm this, a convolutional neural network (CNN) was used to measure the device's optical sensing, storing and processing capabilities. The array simulation received optical images transmitted over the blue light wavelength and performed inference computation to process and recognize the images. The results shows that our devices are able to recognize the objects in the images with 91% accuracy," the scientists stated.
"The determined approach is promising for the development of future artificial retina networks for artificial visual perception and in-memory light sensing applications. It should be noted that the demonstrated MOS memory devices use a similar structure as the Nobel Prize-winning charge-coupled devices (CCD) in CCD cameras, which makes this study a significant step towards the development of smart CCD cameras with artificial visual perception capabilities," the scientists said.
Have a story tip? Message me at: http://twitter.com/Cabe_Atwell