Photovoltaic Technology to Power Synaptic- and Neuronal-Like Devices

ICN2 | 08-13-2019

Neuromorphic engineering, which aims at developing devices that mimic the brain structures functioning, is nowadays an emerging field of research that has the potentiality to produce a change of paradigm in computing, signal processing and artificial neuroscience. Since energy efficiency and autonomy are critical aspects of the development of these systems, the implementation of self-powering through sun light is being explored.

A review of the latest progresses in photo-neuromorphic electronics by Dr Amador Pérez-Tomás of the ICN2 Oxide Nanophysics Group has been published in Advanced Materials Interfaces and portrayed on its cover. This report focuses on the application of a specific category of materials, functional oxide thin-films, for photovoltaic neuromorphic systems production.

The development of more powerful computers and artificial intelligence systems, indeed, is facing an important limitation given by their energy consumption. The diffusion of these technologies, which have great potential for a better internet-of-things (IOT) implementation, requires the devices of the future to be self-powered and able to harvest light, kinetic, electromagnetic, or thermal energy.

The inspiration for new research lines comes more and more from the human brain, which is able to process many data in parallel at very low cost. Today’s computers, which are based on the Von-Neumann architecture, are orders of magnitude less efficient than human brains. While there are about ten thousands of millions (1010) of neurons in the human cortex, powerful parallel computers can manage up to one hundred thousand (105) processors. On the other hand, these devices based on CMOS transistors are much faster than biological neurons, but they spend about ten orders of magnitude more energy than them for each operation.

Implementing the computing capability of a human brain with an artificial neural network built with silicon CMOS technology is evidently not viable, thus new neuromorphic or synaptic electronic structures are being developed, with the aim to design devices having the learning and adaptive capabilities of biological brains.

According to the review, photovoltaic technology could be employed in these systems to make them self-powered and energetically efficient. Oxide thin-films, which thanks to their properties are already becoming key to many modern energy, information and communication applications, are good candidates for neuromorphic devices self-sustainment. This paper provides a thorough overview of some of the recent functional oxide-based advances for these applications, envisioning the future design of self-powered photo-neuromorphic systems. Though, extensive research and development are needed to reach this objective.