In Situ Analysis of Filaments in Oxides Thin Film Reram, SrTiO3, NiO/TiO2, by TEM

Abstract

Resistive switching memory (ReRAM) or Memristor attracted considerable attention due to its technological potential for non-volatile memory and various applications. However, the mechanism is still not resolved clearly because of its difficulties of identifying a conducting channel particularly at oxide systems. Objective of this paper is elucidation of resistive switching mechanism by identifying and analyzing behavior of filaments in-situ in SrTiO3 (STO) and NiO/TiO2 oxides systems. ReRAM can be classified according to different criteria; it can be categorized into unipolar and bipolar systems according to current-voltage (I-V) characteristics, or Electrochemical metallization (ECM), Valence change memory (VCM), and Thermochemical memory (TCM) in terms of different aspects in participation of electrodes in operation. Most of the models that have been presented to explain resistive switching phenomena are based on the formation of filaments, no matter which category their system belongs to. Thus, in order to unravel the mechanism of resistive switching, identifying structure of the filament has its great significance. It has been readily observed in case of the ECM because the filaments, injected metal ion from electrode, are distinguished from electrolyte or insulator. On the contrary, filaments of VCM and TCM have yet to be understood thoroughly. In this paper, our focus will be on the system that could be classified into VCM/TCM. Though grain boundary, defect percolation were suggested as promising candidates of filament at these systems, they have not been clearly verified yet due to their unknown structure, limited size of nanometer scale, and geometry covered by matrix. As the structure and behavior of filament is still controversial, substantial amount of papers reported to unveil the mechanism of resistive switching employed many hypotheses. Here, we identified and analyzed the filament structure at oxide resistive switching system of Pt/SrTiO3/Pt and Pt/NiO/TiO2/Pt, employing in-situ scanning tunneling microscope (STM) / TEM holder. STM tip was in contact with oxide thin film in the TEM and scanned over the region of interest to probe filaments. The filaments could have been detected because of their high conductivity. Consequently, new conducting crystal structure, 2nd phase, is observed in the STO, NiO/TiO2 thin film. Similar as we reported before that 2nd phases of metallic crystals, Magneli TinO2n-1 (n>=4), are generated in the TiO2 thin film as conducting channels. After probing filaments, high-resolution (HR) images and electron-energy loss spectroscopy (EELS) were acquired at corresponding area for determining its crystal and electronic structure respectively. HR images were further analyzed to investigate orientation relationship between the filaments and matrix. EEL spectrum were compared with a simulation data obtained from first principle calculation using Vienna Ab initio simulation package. Furthermore, the sample was turned ‘on’ and ‘off’ repeatedly in-situ and behavior of filament is tracked. The resistance state of the sample was correlated to the structure transformation of filaments. SrTiO3 thin film of 60 nm was deposited by pulsed laser deposition process under oxygen atmosphere and NiO/TiO2 thin film of 50nm was deposited by atomic layer deposition. Platinum pad was exploited for both top and bottom electrodes. Cross-section TEM sample was prepared by focused ion beam. Gatan image filter equipped TEM, F20, was used for in-situ STM/TEM, HR images and EELS.