Analysis and modeling of filamentary conduction in hfo2-based structures

Resumen

We are currently facing a revolution in the fields of microelectronics and information technologies that will surely affect our way of life in the years to come. In this regard, the proposal of a memory device based on the combined action of ions and electrons is considered to be a breakthrough of our time. The basic idea is that an electrical stimulus applied to the device can modulate its resistance state and that this state remains unaltered even when power is turned off. This non-volatile effect can not only be used for storing information but also as a synaptic weight in neuromorphic circuits. In particular, this Thesis deals with filamentary-based resistive switching cells which rely on the creation and partial dissolution of nanoscale conductive pathways spanning an insulator film. They are often called ReRAMs. The physical principles behind the resistive switching mechanism can be manifold including ion migration and diffusion, chemical reactions, local temperature increase due to Joule heating, and many others. The material investigated in this Thesis is basically HfO2 but multilayer dielectrics using HfO2/Al2O3 grown by Atomic-Layer deposition are also considered. The structures analyzed are two terminal metal-insulator-semiconductor and metal-insulator-metal devices, which have been electrically characterized and modeled. Even though an extense part of this work deals with devices that exhibit the resistive switching phenomenon, multilayer stacks were explored as well because of the connection of multifilamentary conduction with one-time programmable memories. In general terms, the goals of this Thesis have been to increase our knowledge and understanding on the generation of conductive filaments in thin insulating layers and to condense all this information in a compact model able to represent the electrical behavior of the devices.