Trade-off between Gradual Set and On/Off Ratio in HfO_x-Based Analog Memory with a Thin SiO_x Barrier Layer

Abstract

HfO x -based synapses are widely accepted as a viable candidate for both in-memory and neuromorphic computing. Resistance change in oxide-based synapses is caused by the motion of oxygen vacancies. HfO x -based synapses typically demonstrate an abrupt nonlinear resistance change under positive bias application (set), limiting their viability as analog memory. In this work, a thin barrier layer of AlO x or SiO x is added to the bottom electrode/ oxide interface to slow the migration of oxygen vacancies. Electrical results show that the resistance change in HfO x /SiO x devices is more controlled than the HfO x devices during the set. While the on/off ratio for the HfO x /SiO x devices is still large (∼10), it is shown to be smaller than that of HfO x /AlO x and HfO x devices. Finite element modeling suggests that the slower oxygen vacancy migration in HfO x /SiO x devices during reset results in a narrower rupture region in the conductive filament. The narrower rupture region causes a lower high resistance state and, thus, a smaller on/off ratio for the HfO x /SiO x devices. Overall, the results show that slowing the motion of oxygen vacancies in the barrier layer devices improves the resistance change during the set but lowers the on/ off ratio.

MAX Phase Ti2AlN for HfO2 Memristors with Ultra‐Low Reset Current Density and Large On/Off Ratio

“…A detailed explanation of the functions used to define the filament shape is available in. [78] The resistive switching process of the devices was dictated by three primary factors: the oxygen vacancy (V o ) concentration gradient, the local electric field, and the local thermal field influenced by Joule heating. A previously published model proposed by Ielmini et al [79] posits that the switching process could be understood by resolving three key governing equations, specifically the drift and diffusion continuity equation, the current conservation equation, and the Fourier Joule heating equation, denoted as Equations 1-3, respectively.…”

SeizFt: Interpretable Machine Learning for Seizure Detection Using Wearables

“…Machine learning (ML) and artificial intelligence (AI) have the potential to revolutionize the field of seizure detection by automating the analysis of EEG recordings, thereby drastically reducing the associated time and cost [ 15 , 16 ]. By leveraging advanced algorithms capable of identifying complex patterns in large datasets [ 17 , 18 , 19 , 20 , 21 ], ML-based systems can achieve high levels of accuracy. In fact, in some instances, ML surpasses human performance.…”