Utilizing the Insulator-to-Metal Transition of Vanadium Dioxide for Neuromorphic Computing
Xing Gao is a PhD student in the department Interfaces and Correlated Electron Systems. Promotors are prof.dr.ir. J.W.M. Hilgenkamp and prof.dr.ir. A. Brinkman from the faculty of Science & Technology.
An 'intelligent machine' with a level of intelligence comparable to the human brain has been searched throughout human history. Although current computers have benefited from Moore's law by scaling down transistors, fundamental limits fundamental limits lead to constraints on speed and energy efficiency in data-intensive applications. Neuromorphic computing systems, which are inspired by the human brain, are highly intriguing. To imitate the functions of the human brain, novel circuit elements are required.
VO2 is an attractive candidate material, since it exhibits a near-room-temperature, hysteretic insulator-to-metal transition (IMT) with resistivity changes of several orders of magnitude. Utilizing the electrically induced IMT of VO2, we propose novel switching devices with highly nonlinear behavior and tunable multi-states, which are assembled by single switching elements in a two-terminal configuration.
The resistive switching behavior of the VO2 bridge devices is not only influenced by the intrinsic material properties but can also be tuned on a device level. In this dissertation, VO2-based thin film structures with unique switching characteristics have been investigated. The study spans from thin film fabrication to the characterization and understanding of various device configurations. A higher degree of control for the multistate can be achieved by assembling single bridges in parallel configuration, starting with double bridges. The addition of extra bridges will introduce the potential for more switching events and resistive states. The spacing between bridges affects the number of switches and the potential switching bridge. Due to the complexity of the switching dynamics, nanoscale thermal mapping of in-operando devices has been carried out using scanning thermal microscopy (SThM), which gives us a straightforward indication of the current distribution among the bridges.
The ability to control the novel switching behaviors at a device level, through variation of multi-bridge configurations, provides a route to the development of novel circuit elements for neuromorphic computing.
