Study of the process of fabrication ofVO2devices for applicationon Oscillatory Neural Networks

For almost half a century vanadium oxide compounds have been an object of interestto the scientific community for their very fast metal to insulator transition behavior, with a special focusonV O2for its near room temperature phase transition. Devices made with this material have beenstudied for the fabrication of oscillators to be used in oscillatory neural networks that are able to solvecomplex tasks on unstructured data such as image recognition. In this thesis we offer a study of theprocess of fabrication of such a compound using a novel technique of flash anneal on different thicknessesof ALD deposited amorphous vanadium oxide. An additionnal study of this fast anneal technique onPLD deposited and Ge doped vanadium dioxide was done. A study of different device architectures hasbeen made, comparing a 2D planar architecture and a 3D crossbar architecture. Finally, we offer a fewmacroscopical models to understand better the working principle of this devices. All of this is done onSiO2-on-Si wafers to prepare a process that is compatible with the current CMOS technology