A. M. Aguilera, M. C. Aguilera-Morillo, J. B. Roldán, F. Jiménez-Molinos
Non-volatile RRAMs have many interesting advantages in comparison with flash memories. Because of this modelling the variability of the involved processes is of crucial importance. The switching mechanism consists of an initial high resistance state (HRS) of the device that converts into a low resistance state (LRS) by applying voltage in the “Forming” and “Set” processes. Subsequent application of voltage returns the HRS in a “Reset” process. This leads to growth and rupture of a conductive filament (CF). In this paper functional PCA is applied for modelling the current-voltage curves associated with the reset cycles simulated from a set of devices whose features were generated randomly by assuming truncated cone shapes for the CFs. Once the reset voltage is achieved, the current usually drops off in an abrupt manner so that the sample curves are not defined on the same domain. To solve this problem synchronization of curves is performed previous to polynomial smoothing.
Palabras clave: Resistive Random Access Memories, Current-Voltage Reset Curves, Functional data, Functional PCA f
Programado
X09.2 Grupo de Análisis de Datos Funcionales: últimos avances y aplicaciones
7 de septiembre de 2016 17:30
0.09 - Aula de proyectos 2
