Descripción del proyecto
Neuromorphic computing is expected to have a transformative impact by exploiting the brain like computations. During the last two decades, Neuromorphic computing has been implemented on electronics, with the most important circuit element capable of best mimicking neurons and synapses that is the memristor. Yet, constructing a highly interconnected networks of memristors to resemble biological architecture requires metal wires with extremely scaled area that results in an immense heat generation and power consumption, hindering the neuromorphic electronics development. Recently the new field of neuromorphic photonics has attracted a lot of interest as a valid alternative. In neuromorphic photonics the network connections are made of optical waveguide that provides inherent parallelism, ultra low latency, and almost zero power dissipation. However neuromorphic photonic will only meet its expectation as a groundbreaking technology when an Ideal photonic memristor is available. The ideal photonic memristor (currently missing) should adjust the amplitude of transmitted light in a non-volatile and analog manner, not via light absorption but through light phase change, to minimize the synapsis insertion losses. In this project I will demonstrate the low loss photonic memristor by innovative exploitation of energy band alignment and refractive index modulation of 2-Dimensional (2D) material heterostructures. I will embed the 2D heterostructure within a photonic integrated Mach- Zehnder Interferometer to showcase the non-volatile optical phase shifting with 1-10 fJ energy consumption per bit. This project will conduct interdisciplinary research across light-matter interaction, that requires contributions from materials science, physics, photonics and electronics. Positive outcome of this action will lead to a breakthrough in the development of neuromorphic photonics, thus a revolution in various aspects of smart human societies that rely on high throughput data computation.